University of Southern California Dissertations and Theses

Quantitative MRI for oxygenation imaging in cerebrovascular diseases

(USC Thesis Other)

Quantitative MRI for oxygenation imaging in cerebrovascular diseases

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Quantitative MRI for oxygenation imaging in cerebrovascular diseases by Chau Vu A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (BIOMEDICAL ENGINEERING) December 2022 Copyright 2022 Chau Vu ii Dedication To my dad, mom, Khanh, and Jared. Thank you for everything you have given me. iii Acknowledgements This thesis represents not only the culmination of my graduate work but also all the love and support I have received, without which I would not have made it here today. To my advisor, Dr. John Wood: thank you for taking a chance on an undergraduate student seven years ago. You are someone who I can always come to when things get tough; with just a short conversation, you help me realize that there are still paths ahead and things are not as hopeless as they seem. I have learned so much from you, not only in technical knowledge and professional integrity, but also in strength, resilience, and persistence to fight for what I believe in. Thank you to my qualification and defense committee members: Dr. Krishna Nayak, Dr. Thomas Coates, Dr. Natasha Lepore, and Dr. Brent Liu. Your diverse perspectives and insights have helped me step out of my worm’s eye view and objectively evaluate the clinical significance of my work. Thank you to my core research family: Dr. Xin Miao, Dr. Soyoung Choi, Dr. Yaqiong Chai, Dr. Eamon Doyle, Jian Shen, Botian Xu, Clio Gonzalez Zacarias. Thank you for mentoring me, supporting me, and caring for me both in and out of the lab. To Dr. Julie Coloigner, thank you for taking me on as a research assistant all those years ago and inspiring me to pursue a PhD of my own. To Dr. Adam Bush, you have been and still are a mentor to me long after you have graduated; I expect that will never change. I would also like to thank my extended research family: Dr. Matthew Borzage, Dr. Jon Detterich, Dr. Michael Khoo, Dr. Roberta Kato, Obdulio Carreras, Silvie Suriany, Honglei Liu, Nathan Smith, Bertin Valdez, Dr. Chris Denton, Dr. Toey Thuptimdang, Dr. John Sunwoo, Dr. Andrew Cheng, Julia Castro, Noel Arugay, Mercy Landaverde, and Lisa Villanueva. I would also iv like to thank our collaborators at Amsterdam UMC, Dr. Aart Nederveen, Dr. Bart Biemond, Koen Baas, and Liza Afzali. Thank you to all my friends who have supported and encouraged me throughout my undergraduate and graduate years: Jennifer Chern, Jennifer Enfield-Kane, Melanie Frakes, Kelly Huang, David Jang, Chang Liu, James Lu, Stephanie Lu, Jessica Thai, Kathleen Tsung, Leslie Tung, and Valerie Yu. Fight on! Special thanks to all of my extended family for being a massive support system I can lean on. Thank you to my two cousins, Duy Pham and Thien Pham, for being the brothers I never had. Thank you to all my uncles and aunts, especially Huan Dam, Bich Dam, Anh Dam, and Trieu Pham, for taking care of me when I first came to America. Thank you to my new family, Larry, Karen, and Jessie Mano, for all the trust, confidence, and encouragement. I would not be here without all of your help. Most importantly, I would like to dedicate this thesis to my family. To my parents, thank you for giving me all the love in the world. Thank you, dad, for being my north star, guiding me on my journey, shining the way, and helping me stay on the path I have chosen. Thank you, mom, for nurturing me and encouraging me to pursue my dream against all odds. You both teach me not only to be strong and independent but also to live with compassion and empathy. After growing up and starting a family of my own, I have come to understand the difficulties you have faced and appreciated the immense character and integrity you have shown in life. Thank you to my sister for always being there for me. You are my other side of the coin, my constant companion, and my best friend through it all. I have learned so much just from observing you, especially your strength and resilience in both professional and personal lives. Even if we live thousands of miles apart in the future, I know you’ll never be far away. v Thank you to my husband and the love of my life, Jared, for all the love and patience you have given me. Thank you for loving all different sides of me, even the ones that are silly and exasperating. I can’t wait to start the next chapters of our lives together. No matter where life takes us, wherever you are, that will be my home. vi Table of Contents Dedication ...................................................................................................................................... ii Acknowledgements ....................................................................................................................... iii List of Tables ................................................................................................................................ ix List of Figures ............................................................................................................................... x Abstract ........................................................................................................................................ xii Chapter 1 : Introduction ................................................................................................................ 1 1.1. Clinical motivation: Cerebrovascular disease ............................................................... 1 1.1.1. Basics of cerebrovascular accidents ..................................................................... 1 Hemorrhagic strokes ......................................................................................................... 1 Ischemic strokes ............................................................................................................... 2 1.1.2. Etiology of stroke: Hypoxia .................................................................................... 4 Vascular diseases ............................................................................................................. 5 Hematologic diseases ....................................................................................................... 7 1.1.3. Clinical need for oxygenation evaluation ............................................................. 16 1.2. Oxygenation MR imaging ............................................................................................ 18 1.2.1. Cerebral blood flow (CBF) ................................................................................... 19 Phase contrast (PC) ........................................................................................................ 21 Arterial spin labeling (ASL) ............................................................................................. 22 Dynamic susceptibility contrast (DSC) ............................................................................ 24 1.2.2. Cerebrovascular reactivity (CVR) ........................................................................ 30 1.2.3. Oxygenation extraction fraction (OEF) ................................................................ 31 Tissue-based .................................................................................................................. 32 Flow-based ..................................................................................................................... 36 1.2.4. Cerebral metabolic rate of oxygen (CMRO2) ....................................................... 42 Calibrated BOLD ............................................................................................................. 43 Multi-model approach ..................................................................................................... 46 Chapter 2 : Quantitative Perfusion Mapping with Induced Transient Hypoxia using BOLD MRI 48 2.1. Introduction ................................................................................................................. 48 vii 2.2. Methods ...................................................................................................................... 50 2.3. Results ........................................................................................................................ 58 2.4. Discussion ................................................................................................................... 64 2.5. Supplemental Information ........................................................................................... 70 Chapter 3 : Oxygen Respiratory Challenges for Deoxygenation-based Dynamic Susceptibility Contrast ................................................................................................................ 76 3.1. Introduction ................................................................................................................. 76 3.2. Methods ...................................................................................................................... 78 3.3. Results ........................................................................................................................ 86 3.4. Discussion ................................................................................................................... 97 3.5. Supplemental Information ......................................................................................... 102 Chapter 4 : Sinusoidal CO2 respiratory challenge for concurrent perfusion and cerebrovascular reactivity MRI ................................................................................................. 120 4.1. Introduction ............................................................................................................... 120 4.2. Methods .................................................................................................................... 121 4.3. Results ...................................................................................................................... 128 4.4. Discussion ................................................................................................................. 135 4.5. Supplemental Information ......................................................................................... 140 Chapter 5 : Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia using BOLD MRI and Near-Infrared Spectroscopy ................................................................... 142 5.1. Introduction ............................................................................................................... 142 5.2. Methods .................................................................................................................... 143 5.3. Results ...................................................................................................................... 150 5.4. Discussion ................................................................................................................. 161 5.5. Supplemental Information ......................................................................................... 166 Chapter 6 : Oxygenation effects of hyperoxia challenge in sickle cell disease and chronic anemia ...................................................................................................................................... 169 viii 6.1. Introduction ............................................................................................................... 169 6.2. Methods .................................................................................................................... 170 6.3. Results ...................................................................................................................... 177 6.4. Discussion ................................................................................................................. 186 Chapter 7 : Calibration of T2 oximetry MRI for subjects with sickle cell disease ....................... 190 7.1. Introduction ............................................................................................................... 190 7.2. Methods .................................................................................................................... 191 7.3. Results ...................................................................................................................... 195 7.4. Discussion ................................................................................................................. 200 7.5. Supplemental Information ......................................................................................... 207 Chapter 8 : Reduced global cerebral oxygen metabolic rate in sickle cell disease and chronic anemias ........................................................................................................................ 211 8.1. Introduction ............................................................................................................... 211 8.2. Methods .................................................................................................................... 212 8.3. Results ...................................................................................................................... 218 8.4. Discussion ................................................................................................................. 223 8.5. Supplemental Information ......................................................................................... 231 Chapter 9 Thesis Conclusion .................................................................................................... 240 References ............................................................................................................................... 243 ix List of Tables Table 2.1. Patient demographic and baseline flow data. ............................................................ 54 Table 2.2. Grey matter (GM), white matter (WM) and GM-WM ratio group average perfusion parameters. ................................................................................................................................. 60 Table 2.3. Group average dDSC perfusion parameters in different flow territories. ................... 63 Table 3.1. Quantitative perfusion values for Desaturation, Resaturation, SineO 2, DSC, ASL, and PC. ....................................................................................................................................... 90 Table 3.2. Repeated measures correlation and limits of agreement between Desaturation, Resaturation, SineO2, and reference techniques. ....................................................................... 94 Table 4.1. Whole brain perfusion estimates by SineCO 2 and 3 standards ASL, DSC and PC. 131 Table 4.2. Correlation and Bland-Altman limits of agreement between SineCO2 and reference techniques. ............................................................................................................... 131 Table 5.1. Patient demographic and hematologic data. ........................................................... 152 Table 5.2. Group average NIRS and BOLD desaturation parameters. .................................... 154 Table 5.3. Predictors of T½BOLD, TTPBOLD and ΔBOLD. ................................................................ 156 Table 5.4. Two-tailed Student’s paired t-tests were used to compare desaturation depth and timing within infarct-prone white matter and normal appearing white matter. .................... 161 Table 6.1. Patient demographic and hematologic data. ........................................................... 178 Table 6.2. Group average of whole-brain and regional BOLD and NIRS changes due to hyperoxia. ................................................................................................................................. 181 Table 6.3. Table of oxygen supply and utilization parameters under normoxia and hyperoxia. 184 Table 6.4. Absolute and relative changes in cerebral metabolism parameters across groups. 185 Table 7.1. Subject demographics and hematologic parameters. .............................................. 193 Table 7.2. Bland-Altman analysis for different population calibration models on transfused and non-transfused sickled blood. ............................................................................................ 199 Table 8.1. Subject demographics and hematologic markers. ................................................... 214 x List of Figures Figure 1.1. Silent cerebral infarcts and segmentations. ................................................................ 3 Figure 1.2. Red blood cells morphology in healthy and sickle cell disease subjects. ................. 12 Figure 1.3. Cumulative prevalence of SCIs in SCD. Figure adapted from Debaun and Kirkham, 2016, Blood. ................................................................................................................ 15 Figure 1.4. Relationship between (A) changes in CBF and changes in PaCO 2 and (B) changes in CBF and changes in PaO2. ...................................................................................... 19 Figure 1.5. (A) Gradient-echo signal from gadolinium-based DSC. (B) Gradient-echo signal from deoxygenation-based DSC. ................................................................................................ 25 Figure 2.1. Experimental setup for hypoxia challenge and concurrent SpO 2 and BOLD acquisitions. ................................................................................................................................ 49 Figure 2.2. Transient hypoxia model. ......................................................................................... 50 Figure 2.3. Localization of input functions from the difference between baseline and hypoxic images. ....................................................................................................................................... 52 Figure 2.4. Agreement between CBFdDSC and alternative flow methods. ................................... 61 Figure 2.5. Group average perfusion and fit evaluation maps. ................................................... 62 Figure 2.6. Regional agreement between grey matter dDSC and ASL flow methods. ............... 64 Figure 3.1. Respiratory challenge patterns for Desaturation, Resaturation, and SineO 2. ........... 78 Figure 3.2. Quantitative perfusion maps for respiratory challenges and conventional perfusion techniques. .................................................................................................................. 91 Figure 3.3. Regional agreement in CBF between respiratory challenges Desaturation, Resaturation, and SineO2 and reference standards DSC and ASL in a representation subject. 93 Figure 4.1. SineCO2 CBF, CBV, TD, and MTT maps for individual subjects. ........................... 129 Figure 4.2. Respiratory challenge patterns for SineCO 2. .......................................................... 129 Figure 4.3. Regional agreement between respiratory challenge SineCO 2 and reference standards DSC and ASL in a representation subject. .............................................................. 133 Figure 4.4. SineCO2 CVR maps in individual subjects. ............................................................ 135 Figure 5.1. Experimental setup for transient hypoxia gas paradigm and concurrent SpO 2, NIRS and BOLD MRI acquisitions. ........................................................................................... 145 Figure 5.2. Transient hypoxia model and curve fitting for SpO 2, BOLD and NIRS signals. ...... 153 Figure 5.3. Correlations between BOLD and NIRS hypoxic depths and hemoglobin levels. ... 155 Figure 5.4. Mean, effect size, and hemoglobin correlation ΔBOLD maps in three patient groups. ...................................................................................................................................... 157 xi Figure 5.5. Two-sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between SCD patients and controls. .................................................................................................................................... 158 Figure 5.6. Two-sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between ACTL subjects and controls. .................................................................................................................................... 159 Figure 6.1. Experimental setup. ................................................................................................ 172 Figure 6.2. Sustained hyperoxia model signals. ....................................................................... 174 Figure 6.3. Correlation between changes in BOLD, NIRS and pulse oximetry signals. ........... 180 Figure 6.4. Group ΔBOLD response to hyperoxia for sickle cell disease (SCD), non-sickle anemic (ACTL) and controls (CTL). .......................................................................................... 181 Figure 7.1. HbS-specific Li-Bush calibration for non-transfused SCD patients. ....................... 195 Figure 7.2. Individual and population sickle calibration models. ............................................... 197 Figure 7.3. Bland-Altman analyses of different calibration models on non-transfused SCD subjects and transfused subjects. ............................................................................................. 198 Figure 8.1. Boxplot and linear correlations of oxygen supply and utilization values. ................ 219 Figure 8.2. Relationship between CMRO2, age and anemia severity in historical references. . 222 Figure 8.3. Relationship between CMRO2 and anemia severity when pooling the data from this current study with historical references. ............................................................................. 223 xii Abstract The goal of this thesis is to apply magnetic resonance imaging (MRI) to investigate the risks of silent cerebral infarcts (SCI), which are commonly seen in neurovascular diseases and normal aging. Even though the etiology of SCI remains unclear, this work proposes a two-hit hypothesis in which acute-on-chronic imbalance in oxygen supply and demand results in microvascular ischemia and stroke. Since a history of SCI increases the risk of a subsequent overt stroke, patient monitoring and prevention therapies are important to alleviate the risk of cerebrovascular accidents in vulnerable populations. At the core of stroke risk monitoring and management is regular screening of oxygen delivery and utilization in the brain. Therefore, the first part of this thesis introduces non-invasive MRI techniques to measure cerebral perfusion, oxygen extraction, and metabolic rate. The second part aims to apply these techniques in a cohort of sickle cell disease and chronic anemia patients to explore clinical biomarkers of their increased vulnerability to SCI development. To quantify regional cerebral perfusion, Chapter 2 proposes a non-invasive Deoxygenation-based Dynamic Susceptibility Contrast (dDSC) technique that leverages oxygen respiratory challenges as a source of contrast alternative to traditional gadolinium contrast in perfusion-weighted imaging. Chapters 3 and 4 extend this concept to a variety of oxygen and carbon-dioxide respiratory challenges using bolus and non-bolus paradigms. The findings from this study show regional agreement between this novel deoxygenation-based method against conventional perfusion-weighted techniques, thus demonstrating its potential application in patients in whom gadolinium contrast is contraindicated. Coupled with the capability to acquire quantitative perfusion, respiratory challenges can be used to investigate hemodynamic impairments in different regions in the brain. Chapters 5 and 6 applied transient hypoxia challenge and prolonged hyperoxia challenge respectively in a cohort xiii of healthy volunteers, sickle cell disease, and non-sickle chronic anemia patients. Even though the hyperoxia challenge failed to identify areas of flow limitation in anemic subjects, the striking variations in dynamic response to hypoxia between subjects with and without SCI suggest that these infarcts are just an iceberg phenomenon with respect to microvascular damage. Further investigation into microvascular damage and its colocalization with capillary perfusion and transit time heterogeneity can shed light on the increased SCI risk in these patient populations. In addition to the investigation of tissue perfusion, oxygen extraction fraction can be measured using T2-oximetry MRI techniques. However, these techniques require appropriate and disease-specific calibration equation to convert raw relaxivity values to venous saturation. Chapter 7 addressed this need by establishing a sickle-specific calibration by pooling in vitro datasets from two independent studies. This combined calibration was based on a larger range of hematocrit and yielded unbiased estimates compared to blood-gas oximetry results. Additionally, this work also demonstrated the need to correct for transfusion in hyper-transfused sickle cell disease patients and proposes a correction method based on patient-specific hemoglobin S concentration. Once appropriate calibration equations have been established for sickle cell disease and non-sickle chronic anemia patients, Chapter 8 investigated cerebral oxygen extraction (OEF) and metabolic rate (CMRO2) in this population. This work demonstrated decreased OEF and CMRO2 in anemia subjects, proportional to the degree of their anemia severity. To address the potential for bias due to our calibration equations, this chapter performed a meta-analysis of historical CMRO2 publications and showed striking concordance with these historical datasets from patients having broad etiologies for their anemia. Additionally, this reduced cerebral metabolism is consistent with emerging data demonstrating increased non-nutritive flow, or physiological shunting, in chronically anemic patients. xiv Overall, this thesis proposes novel non-invasive methods to measure cerebral perfusion and oxygenation as well as applies these techniques in a cohort of patients with hematologic diseases to explore their stroke risk. Further work is necessary on both the technical and the physiological fronts, including additional validation in other pathologies as well as routine hemodynamic screening as part of a stroke prevention program in vulnerable populations. 1 Chapter 1 : Introduction 1.1. Clinical motivation: Cerebrovascular disease 1.1.1. Basics of cerebrovascular accidents A cerebrovascular accident, or a stroke, happens when blood flow to the brain is interrupted. Since the brain utilizes 20% of the total metabolic demand 1 but cannot survive under anaerobic conditions, an interruption in oxygen supply will cause tissue necrosis, resulting in physical and mental disabilities as well as possible fatality. Stroke is an enormous public health issue, affecting 1 in 20 adults, with a higher incidence compared to acute coronary heart disease 2 and a higher burden in low- and middle-income countries 3,4 . The correlation between stroke burden and socioeconomic indicators suggested inequitable distribution of stroke care quality and accessibility between nations as well as between different regions within the same country 5,4 , warranting more equitable, targeted, and regional prevention strategies to curb the stroke epidemic. The disease of stroke can be categorized into two main categories: hemorrhagic strokes (ruptured vessel) and ischemic strokes (occluded vessel). Hemorrhagic strokes Hemorrhagic strokes are caused by ruptured blood vessels that bleed into the brain parenchyma or subarachnoid space, compressing on cerebral tissue and increasing intracranial pressure 6 . The subsequent inflammation causes cytotoxicity of blood, oxidative stress, edema, disruption of the blood-brain barrier and cell death 6 . The incidence of hemorrhagic strokes is approximately 12-15% of cases per 1 million per year, higher in male and in older patients 7 . And 2 even though only 10% of the annual stroke cases are hemorrhagic 8 , this stroke subtype results in worse prognosis with case fatality rate of around 50%. Ischemic strokes Beside hemorrhagic stroke, the remaining 90% of acute stroke cases are ischemic, in which a blocked or occluded artery reduces blood flow and oxygen supply to the brain. The incidence of ischemic stroke is approximately 60 per 100 thousand 9 , and 2 out of 3 patients are dead or dependent within 5 years of stroke, with less favorable outcome increasing with age 10 . The risk factors include high blood pressure, diabetes mellitus and coronary heart disease 11,12 . The mechanism of injury for ischemic stroke is the deficiency of oxygen and nutrients, resulting in necrosis of brain parenchyma. In the secondary ischemia progression, feed-forward response exacerbates the neuroinflammation by activating microglial cells and releasing proinflammatory agents 9 . The overexpression of these proinflammatory cytokines promotes apoptotic signaling and activates additional cytotoxic proteins, leading to neuronal cell death and consequently a larger infarcted area 13 . Standard procedure in stroke management indicates that neurological examination should be performed as soon as possible to identify the affected vascular region and salvage viable brain tissue by artery recanalization and penumbra reperfusion 14,15 . Different reperfusion strategies include intravenous thrombolysis, which cleaves and disintegrates the occlusion in order to improve blood flow, and the more invasive endovascular thrombectomy, which surgically removes the blockage in the affected artery 14 . Even though reperfusion delivers oxygen to the affected tissue regions, this therapy poses the risk of further injury since abrupt increases in the oxygen supply leads to deleterious generation of reactive oxygen species, causing peroxidation of cell membranes and promoting further proinflammatory response that exacerbates tissue injury 16 . Such reperfusion injury is responsible for up to 70% of the final lesion size 17 . 3 Apart from overt ischemic strokes, other types of cerebrovascular events include transient ischemic attacks (TIA) and silent cerebral infarcts (SCI). v Transient ischemic attacks (TIA) Similar to overt strokes, TIAs result from vessel blockage and occlusion; however, this occlusion usually only lasts for 30 minutes, and the associated symptoms typically resolve within 24 hours 18 . Since blood flow is restored quickly and hypoxia exposure is minimal, there is no permanent cerebral infarction and tissue necrosis. However, despite the reversal of symptoms, TIAs are considered a prelude to a full-blown stroke, and the incidence of secondary ischemic stroke is 17% within the first 6 months 18 . v Silent cerebral infarcts (SCI) Another type of cerebrovascular event that is not classified as overt ischemic stroke is SCIs, which present as lesions on brain MRI in asymptomatic patients without a clinical history of stroke 19 . Most silent infarcts present as lacunae in the white matter or brainstem but can also appear in cortical grey matter and deep nuclei regions 19,20 . SCIs can be diagnosed on structural MRIs (Figure 1.1), but the definitions vary across studies 21 . Most studies identify silent infarcts as focal hyperintensities on T2-weighted MRI of 3 mm in diameter or larger; if the infarct is in the white matter, corresponding hypointensities should be observed on T1-weighted MRI 22 . Figure 1.1. Silent cerebral infarcts and segmentations. 4 Additionally, recent advances in MRI technology including higher field strength and increasing resolution likely will lead to higher sensitivity in detecting smaller lesions 23 , so this definition might need to be modified in the future. Similar to TIAs, SCIs increases the risk of subsequent ischemic stroke 24 , with a hazard ratio of 2.08 for patients with SCIs 25 , similar to the hazard ratio for recurrent stroke for patients with clinical history of ischemic stroke 26 . The incidence of SCIs is 5 times higher than overt stroke and significantly increases with age 22 , at around 1.9-3.7% per year 27 . Additionally, SCI prevalence is 1.7% in subjects in their 40s but rises to 43.8% in subjects in their 70s 28 , increasing with age 8% per year 22 . Similar risk factors are observed for SCIs compared to acute ischemic stroke, including hypertension, heart disease and older age 27 . Despite the lack of focal neurological deficit, SCIs are associated with neuropsychological dysfunction, including depression 29 , dementia 30 and global cognitive decline 31 . In pathologic conditions, patients with SCIs demonstrate deficits in full-scale IQ 32 , memory 33 , attention and executive functions 34,35 . The presence of silent infarcts is also associated with spatial heterogeneity in cerebral response to gas perturbations 36 , indicative of heightened capillary transit time heterogeneity and extensive underlying microvascular remodeling 37,38 . Overall, SCIs have been linked to various clinical conditions that result in early mortality 39,40 . 1.1.2. Etiology of stroke: Hypoxia Currently, the etiology of SCIs remains unclear. The overlapping risk factors between silent infarcts and clinical ischemic strokes 27 suggests a similar etiology, in which SCIs originate from microvascular occlusions that are not severe enough to cause a full-blown stroke. However, short of a fully occluded arteriole, development of SCIs can likely be explained by the two-hit hypothesis. Under resting conditions, global oxygen delivery is maintained at normal levels; however, under acute interruptions in oxygen delivery or sudden spikes in the brain’s oxygen 5 requirement (such as during a viral infection), oxygen supply cannot adequately increase to match the demand, resulting in tissue hypoxia and ischemia. Previous works have demonstrated that SCIs are usually located in deep white matter regions rather than cortical grey matter 41 , thus corroborating that SCIs happen in areas that are further away from the oxygen supply. Additionally, SCIs are also often observed in border zone regions between the anterior, middle, and posterior cerebral arteries 42 , which lie further away from penetrating arterioles and thus are most vulnerable to interruptions in the vascular oxygen supply chain. Similar to overt strokes, SCIs have higher prevalence in the frontal and parietal regions 43 supplied by the middle cerebral artery 44 ; even though conditions of hypoxemia are still prevalent, the posterior circulation is better protected under acute stressors and thus is less susceptible to ischemic infarcts 45 . Conditions of microvascular hypoxemia in which SCIs likely develop can manifest under two conditions: (1) microvascular diseases in which capillary blood flow is diminished, and (2) hematologic diseases in which the low oxygen carrying capacity by red blood cells prevents adequate oxygen delivery to the tissue. Vascular diseases In terms of overt ischemic strokes, most occur due to thrombosis; however, in order to explore the etiology of SCIs, only non-thrombotic vascular diseases that impair blood flow to the brain are investigated. Two typical vasculopathies that cause narrowing of the arteries and affect cerebral blood flow include (1) atherosclerosis and (2) Moyamoya disease. v Atherosclerosis Atherosclerosis is a common disease in which plaque, or fatty deposits, appear on the inner layers of the arteries, causing thickening and hardening of the arterial walls 46 . The plaques 6 start with deposition of cholesterol crystals on the smooth muscle, proliferate to the surrounding fibrous tissues and bulge inside the arteries 46 . This plaque development triggers an inflammatory cascade that contributes to development of adhesion molecules on the endothelium and secretes growth factor that promotes proliferation of smooth muscle cells, resulting in further growth of the plaque 47,48 . As a result, atherosclerosis causes narrowing of the arteries that restricts cerebral blood flow and decreases the amount of oxygen delivery to the brain. Additionally, these plaques tend to develop at regions of lower blood flow, such as branches, bends and bifurcations of the arterial trees 49 , which in turn further lower blood flow to downstream capillary beds and tissues. v Moyamoya disease (MMD) Moyamoya Disease (MMD) is a rare and progressive condition in which the arteries to the brain become narrow, thus restricting blood flow to the brain. MMD is characterized by stenosis of the terminal internal carotid arteries (ICA) as well as the initial portions of the anterior and middle cerebral arteries at the base of the brain 50 . The incidence of MMD is highest in Asia, especially Japan, at 0.35 per 100k 51 . A few works have suggested that MMD can be caused by genetic risk factors, but there is no definite evidence showing inheritance of an allosome exists in MMD 50 . Several molecular pathways that have been implicated in MMD include proliferation of smooth muscle cells, vascular remodeling, and inflammation, many of which contribute to other vascular conditions including atherosclerosis 52 . Despite the similarities to atherosclerosis, MMD pathophysiology presents as a reduction in both inner and outer diameter of the ICA whereas atherosclerosis only demonstrates outward vascular remodeling 53 . MMD is also limited to the ICA and the surrounding branches, whereas atherosclerosis is more widespread throughout the arterial tree 53 . Since the two diseases manifest in arterial narrowing, differentiating MMD and atherosclerosis can determine the correct treatment course, as atherosclerosis typically only requires lifestyle changes 54 , whereas MMD 7 might require surgical revascularization procedures to reduce the risk of future cerebrovascular accidents 55 . In patients with MMD, ischemic stroke present in 68% of cases, primarily in cerebral areas whose blood flow is supplied by the ICA and MCA 52 . In children, common everyday actions are hypothesized to induce strokes, including crying and hyperventilation. This is in line with the two- hit hypothesis as the vasculature is fully dilated under chronic tissue hypoxia; and under acute stressors such as hyperventilation, the lack of additional vasodilation capacity coupled with the steal phenomenon results in a reduction in perfusion in vulnerable territories and leads to acute ischemic stroke 52,56 . Even for asymptomatic MMD patients who do not have a clinical history of stroke, they still demonstrate reduced cerebrovascular reactivity 57 and are still at risk for SCIs, especially in the anterior circulation 56 and watershed regions 58 , similar to at-risk regions for non- MMD-related SCIs. The prevalence of SCIs in asymptomatic MMD (which typically affects younger patients 59 ) is comparable to elderly patients older than 60 60 , which suggests that MMD- induced hypoperfusion and tissue hypoxia are the basis of accelerated aging. Overall, despite their different etiologies, MMD and atherosclerosis induce narrowing of large arterial vessels, leading to hypoperfusion in the downstream capillary beds and potential tissue hypoxia, thus increasing the likelihood of silent ischemic strokes under acute stressors. Hematologic diseases Another mechanism through which tissue hypoxia can happen is hematologic diseases, which are blood disorders that can affect different components of blood, including red blood cells, white blood cells and platelets. Hematologic malignancies that affect white blood cells include leukemias, lymphomas and myelomas 61 , which cause excessive reproduction of cancerous cells and prevent normal blood function such as fighting infections or preventing serious bleeding. Even though tissue hypoxia is a possible consequence of hematologic malignancies 62,63 and patients 8 with cancer have a higher risk of ischemic stroke 64,65 , it is unclear whether both diseases arise independently from shared risk factors or if cancer directly influences the pathophysiology of stroke 66 . This work will be focusing on hematologic diseases that directly impacts the blood’s ability to transport oxygen to tissue, thus leading to tissue hypoxia and eventual ischemia. These are pathologies that affect the red blood cells, including iron-deficiency anemia, thalassemia, and sickle cell disease. v Iron-deficiency anemia (IDA) Iron-deficiency anemia is the most common form of anemia, affecting 2 billion people worldwide 67 , most of whom are women and children 67 . Despite being the most treatable form of anemias, over 800k deaths have been attributed to IDA globally, with the highest mortality rates in developing nations 68 . Even though the prevalence of IDA has decreased slightly in the past decade, the disparate distribution of anemia remains between high- and low-income countries, warranting policies designed to target more vulnerable groups with lower socioeconomic status to reduce the global burden of anemia 69,70 . In terms of mechanism of injury, IDA is characterized by 3 phases: iron depletion, iron deficiency and iron-deficiency anemia 71 . In the first phase of iron depletion, since the body does not naturally excrete iron but instead recycles iron following degradation of senescent red blood cells 72 , the primary mechanism for low iron is blood loss through menstrual periods or gastrointestinal bleeding 73 . These conditions can be compounded by iron-poor diets 74 and will lead to iron deficiency if the iron stores are not replenished. Additionally, the iron-deficiency phase also develops secondary to various other causes, including bacterial infections 75 , inflammatory bowel disease 76 and gastrectomy complications 77 . Even though non-anemic iron deficiency is common and present with non-specific symptoms 78 , most iron-deficient patients eventually advance into the final phase of iron-deficiency anemia. Iron is an essential component of 9 hemoglobin, with each gram of hemoglobin containing approximately 3.5 mg of iron 79 . Under low stores of iron or diminished iron transport, erythropoiesis is restricted, leading to low red blood cells production and thus impaired ability to deliver oxygen to tissue. Even though IDA has been previously associated with increased risk of ischemic stroke 80 , few works apart from case reports 81,82 have explored the possible relationship between these diseases. In terms of SCIs, previous work has demonstrated the severity of anemia as a possible risk factor SCI development, but in the context of congenital hemoglobinopathy instead of acquired IDA 83 . A community-based study has shown worsening SCIs in elderly patients with general anemia; even though this study did not have information on cause of anemia, given that half of anemias are IDA, it is reasonable to speculate a potential link between IDA and silent brain infarcts development 84 . Overall, a prospective study to investigate the pathophysiology of silent strokes in the context of IDA is warranted. v Thalassemia Whereas iron-deficiency is a form of acquired anemia, thalassemia is a congenital anemia condition that involves impaired hemoglobin production and consequently low counts of functional red blood cells 85 . The incidence of thalassemia is approximately 4.4 in every 10k live births, affecting male and female equally and occurring mostly in persons with Mediterranean, African, and Southeast Asian descent 86 . Since hemoglobin comprises of four protein subunits – two alpha chains and two beta chains, thalassemia is divided into alpha thalassemia and beta thalassemia based on which protein chain undergoes defective synthesis 85 . In alpha thalassemia, instead of normal hemoglobin tetramer (α2β2), impaired synthesis of alpha globin chains leads to excess beta globin chains, which can result in Hemoglobin H disease or Hemoglobin Bart’s disease. Hemoglobin H has tenfold higher oxygen affinity compared to hemoglobin A, so it holds onto oxygen tightly without the capability to deliver oxygen to the tissue 85 . Hemoglobin H is associated with moderate anemia, hepatosplenomegaly, and 10 dependency on blood transfusion. Hemoglobin Bart’s disease is lethal in utero due to hydrops fetalis unless fetal cord blood transfusions are performed until birth. On the other hand, beta thalassemia is subdivided by its etiology into reduced (β + ) or absent (β 0 ) synthesis of the beta globin chains, leading to excess alpha chains 87 . Patients with beta thalassemia are asymptomatic at birth due to the presence of hemoglobin F (α2γ2) but begin to develop symptoms at six months of age due to the deficit of beta chains. Since the excess alpha chains cannot form viable tetramers (α4), beta thalassemia is characterized by formation of insoluble alpha aggregates that promote apoptosis of red blood cells and demonstrates greater disease severity compared to alpha thalassemia 85 . Thalassemia has a wide spectrum of disease, ranging from thalassemia trait, thalassemia minor, thalassemia intermedia and thalassemia major. At one end of the spectrum, thalassemia minor or trait patients are heterozygous, asymptomatic, require no treatment but can be silent carriers and pass the disease to their offspring 88 . Thalassemia intermedia can display mild to moderate anemia, requiring episodic blood transfusions, especially if the patients develop related complications such as pulmonary hypertension or chronic ulcerations. Lastly, at the other end of the spectrum is homozygous thalassemia major, with alpha thalassemia major typically resulting in fatal hydrops fetalis 89 and beta thalassemia major requiring lifelong blood transfusion and iron chelation therapies 90 . Stroke risks in this patient population are frequently attributed to thromboembolic cerebrovascular accidents due to the hypercoagulable state of beta thalassemia 91 . These thromboembolic events are more frequently recorded in thalassemia intermedia patients with limited transfusion instead of transfused thalassemia major subjects, since the transfused healthy red blood cells can eliminate the abnormal aggregation of thalassemic red cells 91,92 . However, previous work has demonstrated a non-trivial rate of SCIs in thalassemia major patients under regular transfusion 93 , suggesting that hypercoagulability only explains a portion of the heightened stroke risk. Since thalassemia major is characterized by severe anemia, it is reasonable to 11 postulate that chronic anemia contributes to SCI under the two-hit hypothesis of chronic anemia and acute stressors, such as hyperventilation, psychological stress, or infections. Published works have been rare and unclear about the role of anemia on SCIs in thalassemia 94,95 . Our own data in thalassemia subjects indicates a trend of lower hemoglobin value in patients with SCIs compared to without SCIs (9.7±2.1 vs 11.0±2.7 g/dl), but a larger prospective study is warranted to investigate the possible link between anemia severity and stroke risks in thalassemia. Since our hypothesized etiology of stroke is anemia-induced impaired oxygen delivery coupled with acute stressors, non-thrombotic ischemic strokes should be considered in the context of tissue hypoxia and controlling for the degree of hypercoagulability in thalassemia patients. If severe anemia is left uncorrected, low oxygen delivery especially in watershed white matter regions furthest away from the arterial tree leaves these areas vulnerable to ischemic stroke. In transfused patients, iron overload can impact vascular health 96 . Even though previous work has suggested no difference between serum ferritin concentration in thalassemic patients with and without SCIs 95,97 , careful control of transfusion status and systemic iron overload is required to isolate the effects of anemia on SCI risk. v Sickle cell disease (SCD) While there are few studies on the etiologies of silent infarcts in acquired IDA and congenital thalassemia, stroke risk in sickle cell anemia has been a topic of extensive research. Sickle cell disease (SCD) is the most common genetic disorder, with an estimated global incidence of 300-400 thousand neonates per year and primarily concentrated in sub-Saharan Africa 98 . In the United States, SCD affects 1 in 500 African Americans and 1 in 16,000 Hispanic Americans 99 . With public health advancements and improving patterns of population age, fertility, life expectancy in low- and middle-income countries, the annual number of newborns with SCD is expected to exceed 400,000 by 2050 100 . 12 Sickle cell disease results from a mutation that substitutes the 6 th -position amino acid glutamic acid with valine in the beta globin chain. This mutation leads to the formation of hemoglobin S, which polymerizes under deoxygenated conditions and thus bends red blood cells from the original biconcave disk shape into a sickled, crescent shape 101 (Figure 1.2). These sickled red cells can clog the microvasculature, decreasing oxygen delivery to organs and tissues, as well as accelerates hemolysis, leading to low red cell count and chronic anemia 101 . The severity and complications of SCD are characterized by the underlying genotypes, since the polymerization process is influenced by the concentration of hemoglobin S within the red cells and the composition of hemoglobin variants in different genotypes 102 . The heterozygous sickle cell trait (Hb AS) has half hemoglobin A and approximately 20- 45% hemoglobin S in the erythrocytes 103 . Comparing a sickle cell trait subject with 20% hemoglobin S with a transfused SCD patient with 20% hemoglobin S (Figure 1.2B), the SCD patient has 20% erythrocytes that are almost filled with HbS, whereas all the erythrocytes in the sickle cell trait patient are 20% filled with HbS and remaining filled with HbA. Therefore, the HbA- Figure 1.2. Red blood cells morphology in healthy and sickle cell disease subjects. (A) Healthy control, 100% HbA. (B) Non-transfused sickle cell anemia (non-tx SCA), 80% HbS. (C) Sickle cell trait (SCT), 20% HbS. (D) Transfused sickle cell anemia (tx SCA), 20% HbS. 13 dominated red blood cells do not sickle under normal conditions, and thus sickle cell trait is not associated with chronic anemia or related comorbidities. On the other hand, the homozygous phenotype (Hb SS), also known as sickle cell anemia, is the most severe form of the disease. Hemoglobin SS disease is characterized by chronic hemolytic anemia, unpredictable pain crises and widespread organ damage 98 . Other comorbidities include splenic sequestration, acute chest syndrome, papillary necrosis, renal insufficiency, and high risk for strokes 104 . One variant of SCD is hemoglobin SC (Hb SC) disease, in which red blood cells contain approximate proportions of hemoglobin S and hemoglobin C (a mutated version of hemoglobin that substitutes the 6 th glutamic acid with lysine in the beta globin chain) 105,106 . Hemoglobin SC displays slightly milder anemia compared to Hb SS, but patients are still at risk for splenomegaly, retinopathy, vaso-occlusive events, pain crises, and silent cerebral infarcts 107,108 . Another heterozygous variant of SCD is the sickle beta thalassemia (Hb Sβ), subdivided by the reduced (Hb Sβ + ) or absent (Hb Sβ 0 ) synthesis of the normal beta globin chains. Hemoglobin Sbeta0 has almost no hemoglobin A and is often indistinguishable from Hb SS. On the other hand, hemoglobin Sβ + presents with 5-45% hemoglobin A in the red cells, which dilute the hemoglobin S concentration similar to sickle trait and inhibit polymerization-induced cellular damage, leading to a milder phenotype compared to Hb SS 109 . For the more severe phenotypes, including Hb SS and Hb Sβ 0 , HbS polymerization coupled with membrane deformation and erythrocyte dehydration results in irreversibly sickled red cells, which are highly unstable and have reduced lifespans 110,111 . These damaged red cells are destroyed by extravascular and intravascular hemolysis much faster than the erythropoiesis process 112 , leading to hemolytic anemia 113 . The reduced oxygen affinity in hemoglobin S enables the red cells to give up oxygen more easily to tissue 114 , which has been hypothesized to compensate for the hypoxemia and chronic anemia in SCD and maintain tissue oxygenation. However, previous work has shown that tissue 14 hypoxia persists 115 , and our own data shows reduced brain tissue oxygenation index (NIRS) in SCD compared to healthy controls (0.55±0.08% vs 0.60±0.06%, p=0.01), thus corroborating the hypothesis of anemia-mediated tissue hypoxia 116 . The basal condition of tissue hypoxia contributes to a higher risk of cerebrovascular accidents in SCD. Under deoxygenated conditions, deformed red blood cells sludge and congest the microvasculature, resulting in tissue ischemia and infarction in major organs throughout the body 104 , including the bone marrow 117 , lungs 118 , kidneys 119 , spleen 120 , and brain 121 . In certain cases, infarction can happen in all these major organs at the same time in acute multisystem failure. Even though the pathophysiology of acute multiorgan failure is unclear, acute stressors such as sickle pain episodes can trigger dramatic declines in hemoglobin levels, which decrease oxygen delivery further and lead to diffuse tissue ischemia in different organ systems 122 . One of the most serious manifestations of progressive multiorgan failure in SCD is central nervous system complications, as neurologic insults can happen from a very early age and can have significant impact on neurocognitive functions. Prior to routine TCD screening and chronic transfusion therapy,11% of SCD subjects had an overt stroke by the age of 20, and this figure increased to 24% by the age of 45 123 . The incidence of infarctive stroke is highest in children under 20 and adults older than 30 years of age 124 . Even though the pathophysiology of ischemic stroke in SCD is not well-understood, previous work has demonstrated that transfusion can alleviate recurrent stroke risk 125 , suggesting the role of sickled red cells in stroke progression. Possible mechanisms include the erythrocytes’ increased adhesion to the vessel walls, leading to prolonged microvascular transit time, polymerization of red cells and potential endothelial damage 126 . Additionally, anemia-mediated chronic tissue hypoxia leaves watershed cerebral regions vulnerable to perfusion failures under acute stressors, resulting in tissue ischemia and stroke 127 . This theory is corroborated by the significant association between overt stroke risk and the severity of anemia in SCD patients 124 . 15 Even though transfusion improves the risk for overt strokes by an order of magnitude, SCD patients are still at risk for SCI 128 . The prevalence of SCIs in SCD patients is 53% at 30 years of age and does not plateau into adulthood 129 (Figure 1.3). Cumulatively, the risk for SCI is 19.2% at 8 years but increases to 39.1% at 18 years of age 130 . Children with preexisting SCI are at higher risk of developing new, progressive SCI, with an incidence of 7.06 per 100 patient-years 43 . Additionally, the presence of SCI reflects a 14-fold increase in risk of overt stroke compared to patients to normal MRI 131 . Like overt strokes, the etiology of SCI remains unclear. The higher rate of SCI in SCD patients compared to non- sickle anemic patients 132 suggests contribution of the sickle red cells morphology and characteristics. Notably, the tendency of dehydrated dense red cells to occlude the blood vessels can leave tissues more susceptible to infarcts, and fat embolism as a known complication of SCD can travel to the brain and lead to ischemic stroke 133,134 . While these mechanisms usually contribute to overt strokes, silent infarcts tend to occur in distinct brain regions compared to overt strokes 24 ; therefore, SCIs have been hypothesized to have different pathophysiology compared to overt strokes. In this work, I postulate that SCIs develop as a result of microvascular dysfunction and impaired tissue oxygen delivery to the brain. Previous work has shown that low hemoglobin level in SCD is associated with the presence of SCI 83 . Under a state of chronic hypoxia, SCD patients are more at risk for SCI development during periods of acute stressors of restricted oxygen supply (e.g. during acute anemic events 135 ) or escalated oxygen demand (e.g. during pain crises 136 ). When the microvasculature fails to balance the oxygen supply and demand, downstream tissue becomes Figure 1.3. Cumulative prevalence of SCIs in SCD. Figure adapted from Debaun and Kirkham, 2016, Blood. 16 hypoxic and ischemic, resulting in microinfarcts. Since watershed white matter areas in the frontal and parietal lobes are furthest away from penetrating arteries and less prioritized compared to grey matter, these regions are most vulnerable to SCIs 43,24 . The frontal and parietal lobes are also supplied by the anterior and middle cerebral arteries, which are less protected during acute events compared to the posterior cerebral artery, thus leaving these territories more vulnerable to SCI events 137 . Despite the nomenclature ‘silent’ infarcts, SCIs have been shown to correlate with cognitive impairment in SCD. The presence of SCIs is associated with lower FSIQ 138,139 and poor academic achievements 140 in SCD patients. Even in patients free of silent strokes, cognitive dysfunctions are still present and are related to subtle brain abnormalities, both of which are accounted for by the severity of anemia and the underlying brain hypoxia 141 . This association between diffuse brain deficits and low hematocrit has been demonstrated in multiple studies 141,142,143 , thus further pinpointing the role of anemia behind the pathophysiology of SCI and neurocognitive impairments in SCD patients. 1.1.3. Clinical need for oxygenation evaluation Since the pathophysiology of ischemic stroke and SCI is characterized by acute-on- chronic imbalance between oxygen supply and demand, it is imperative to evaluate the brain’s oxygen delivery and utilization in patients at risk for cerebrovascular accidents. Cerebral oxygen supply and demand can be assessed quantitatively through a combination of different physiological parameters, including hemoglobin (Hb), arterial and venous saturations (Ya, Yv), oxygen content (CaO2, CvO2), oxygen delivery (DO2), oxygen extraction fraction (OEF), and cerebral oxygen metabolism (CMRO2): ! " = 1.34 × × ! + 0.003 × ! " (mL O2/mL blood) # " = 1.34 × × # + 0.003 × # " (mL O2/mL blood) 17 = 1− % ! & " % # & " (%) " = × ! " (mL/100g/min) " = × " (mL O2/100g/min) Since most of the oxygen transport is conducted by hemoglobin in red blood cells, patients with hematologic diseases such as anemia have abnormal hemoglobin levels 144 and decreased arterial oxygen content. To compensate for the decreased CaO2, chronic anemia typically presents with increased systemic and cerebral blood flow 145,116 , which normalizes DO2 at a global level but has been proven insufficient to maintain DO 2 in the white matter and watershed areas of highest distribution of SCIs 146 . On the other hand, vascular disease such as Moyamoya presents with normal hemoglobin but reduced CBF on the ipsilateral hemisphere 147 , leading to impaired DO2 and heightened stroke risk in the anterior and middle cerebral arteries territory. Overall, decreased tissue oxygen delivery is a precursor for potential cerebral ischemia in both vascular and hematologic conditions. Under a state of diminished DO2, CMRO2 is theorized to be maintained by modulations in OEF. Oxygen extraction has been demonstrated to increase in Moyamoya 148,149 . However, there have been multiple divergent reports with regards to OEF and CMRO2 in sickle cell anemia patients 149,150 . Using separate, more appropriate calibrations for measuring OEF in non-sickle and sickle cell subjects 151 , our group has demonstrated decreased OEF and CMRO2 in SCD and thalassemia patients 150 . Such anemia-mediated hypometabolism is consistent with a compilation of historical CMRO2 datasets in a broad range of anemia syndromes as well as when generalized to other medical conditions 150 . Despite the robust results, such divergent findings between different centers and study cohorts warrant further investigation into the hemodynamic mechanisms underlying sickle cell disease specifically and chronic anemia in general. Even in healthy control subjects, the presence of SCI is associated with reduced CBF and CMRO2 152 similar to chronically anemic patients, thus implicating the role of oxygen delivery in 18 the pathophysiology of silent infarcts. Therefore, there is an urgent need for routine monitoring of oxygenation in patients more at risk for overt and silent strokes, so as to administer preventative therapies such as surgical revascularization in Moyamoya or transfusion in SCD patients. 1.2. Oxygenation MR imaging Oxygenation parameters, such as CBF, OEF and CMRO2, can be measured using various techniques, including Kety-Schmidt dilution 153,154 , Near-Infrared Spectroscopy (NIRS) 155,156 , Positron Emission Tomography (PET) 157 , Xenon Scintigraphy 158,159,160 and Xenon Computed Tomography (CT) 161,162 , and Magnetic Resonance Imaging (MRI). In this work, I will focus on the use of MRI for oxygenation imaging. 19 1.2.1. Cerebral blood flow (CBF) Cerebral blood flow (CBF), also known as perfusion, is defined as the blood volume that flows per unit mass per unit time in brain tissue 163 , with a typical value in adults of 50mL/100g/min 164 . Under steady-state conditions, CBF is regulated in the body through a spectrum of mechanisms, including arterial blood gases (PaCO2, PaO2) and systemic blood pressure via cerebral autoregulation 165 . Classically, the concept of cerebral autoregulation refers to the brain’s ability to maintain constant perfusion within a large range of blood pressures 166,167 , with a slope of approximately zero within this range and higher slopes outside the upper and lower autoregulatory limits. However, recent works have suggested that this zero-slope plateau was only true across subjects, and within subjects there is a more passive linear relationship between CBF and blood pressure 168 of approximately 1% change in CBF per mmHg change in mean blood pressure 165 and high variation of upper and lower autoregulatory limits across individuals 169 . This linear relationship between CBF and arterial pressure was observed at isocapnic conditions 165 , whereas at resting mean arterial pressure, acute CBF response to changes in CO2 pressure is sigmoidal in nature 170 (Figure 1.4A). The linear portion of this sigmoid is centered close to resting carbon dioxide tension, which results in linear changes in CBF in response to modulations in PaCO2, at 1-2 mL/100g/min change in CBF per mmHg within a narrow range of PaCO2 171,172,153 . Above a particular threshold of PaCO2, mean arterial pressure no longer remains constant and increase linearly with CBF and CO2 tension, which represents a range in which autoregulation is exhausted 173 . Different from the acute dynamic response, chronic hypercapnia, such as in patients with lung disease, is not Figure 1.4. Relationship between (A) changes in CBF and changes in PaCO2 and (B) changes in CBF and changes in PaO2. 20 accompanied with significantly higher CBF compared to healthy subjects 174 . This adaptation to chronic hypercapnia has been attributed to altered buffer pH in the brain’s extracellular space via carbonic anhydrase to mediate capillary response 175 and has been corroborated with the normalization of CBF after 45 minutes of continuous hypercapnic exposure 176 . In addition to the CBF modulation in response to PaCO2, decreased PaO2 and blood oxygen content during hypoxia exposure leads to increases in CBF 177 . Previous studies have demonstrated an acute linear response of 0.5 – 2.5% increase in CBF per percent reduction in SpO2 178 . On the other hand, the relationship between CBF with PaO2 needs to account for the non-linearity in the hemoglobin dissociation curve and results in relatively stable CBF at PaO2 above 50 mmHg 179 . Below PaO2 of 50 mmHg, compensatory vasodilation is triggered to maintain oxygen delivery during hypoxia, thus increasing CBF (Figure 1.4B). However, even though global oxygen delivery is preserved, there is a heterogenous response throughout the brain 137 which is worse in the watershed white matter areas 146 and in cerebrovascular diseases 36 . This hemodynamic response to acute hypoxia can draw a parallel to acute anemia, in which CBF increases in proportion to reduction in hemoglobin to compensate for the decreased oxygen carrying capacity and avoid the risk of tissue hypoxia 180 . However, this parallel does not persist between chronic anemia and chronic hypoxia. Prolonged exposure to hypoxia induces hematocrit increase by 10-15% after the first week 181 , thus normalizing both the oxygen content and CBF. On the other hand, chronically anemic patients are unable to raise their hematocrit and thus needs to maintain a constant state of hyperemia 158,182,145 . Even though compensatory hyperemia is able to maintain normal levels of oxygen delivery to the brain in chronically anemic patients, since the brain requires a constant supply of oxygen and nutrients, acute hypoperfusion can still lead to life-threatening ischemic events. Therefore, regulations of CBF under baseline and under perturbed conditions have been of great interest, especially in the context of neurodegenerative diseases and cerebrovascular accidents. To 21 monitor CBF, different perfusion-weighted MRI techniques are employed, including phase contrast (PC), arterial spin labeling (ASL), and dynamic susceptibility contrast (DSC). Phase contrast (PC) PC MRI relies on phase signals to measure velocity of flowing blood, after which total CBF can be calculated from 4 main feeding vessels – internal carotid arteries and vertebral arteries. Under the influence of a pair of equal but opposite-signed bipolar gradients, protons that are stationary experience no net phase shift, whereas moving protons undergo varying degree of phase shifts due to their changing positions within the labeling plane 183 . Faster protons experience a larger difference in applied gradients and thus larger phase shifts 183 . Therefore, velocity can be calculated as: = ∆ ∆ ' where ∆ is the net phase shift, is the gyromagnetic ratio, and ∆ ' is the change in magnetic moment proportional to the field strength. Since the degree of phase shift is constrained between 0 and 360 before aliasing occurs, the maximum velocity measurable is called the velocity encoding, ()* . This ()* is inversely proportional to the gradient strength and is usually set as 100-200 cm/s for CBF measurement depending on pathology in question 184,145 . After acquiring velocities in the 4 feeding arteries, total flow can be calculated as: = ∑ + × + ) + (_×_) M where denotes the artery, is the cross-sectional area of the vessel, _ is the total brain volume in mL 3 , and _ is the brain density of 1.05 g/mL. 22 The within-session coefficient of variation for PC MRI is 4.6±3.0% 145 . In most PC acquisitions, CBF is measured in a single plane instead of 4 separate acquisitions orthogonal to each vessel since temporal variation in a vessel is higher than the variability of their sum. Arterial spin labeling (ASL) Despite yielding robust CBF measurements, PC MRI only measures total blood flow through the 4 feeding arteries and does not provide information on the regional distribution of CBF across the brain. On the other hand, arterial spin labeling (ASL) yields quantitative images of cerebral tissue perfusion and has been frequently used to evaluate focal and diffuse perfusion abnormalities in ischemic stroke 185 , brain tumors 186,187 , and neurodegenerative diseases 188,189 . ASL is built upon the concept of magnetically labeling arterial blood water protons by inverting their longitudinal magnetization and quantifying CBF as the rate these labeled protons travel from the vascular compartment to the tissue compartment 190 . To differentiate vascular signal from static tissue signal, one labeled image and one control image without labeling are acquired; subtraction of these two images yields the perfusion-weighted image of interest 190 . There is a variety of ASL sequences, including continuous ASL (CASL), pulsed ASL (PASL), and pseudo-continuous ASL (PCASL). The principal difference between these sequences lies in the labeling strategy. CASL labels blood continuously as it flows through a labeling slice just below the imaging plane using adiabatic inversion pulses (2-4s), whereas PASL uses shorter RF pulses (5-20ms) to invert a thicker slab of blood volume (10-15cm) 191 . The advantage of PASL is higher tagging efficiency 192 but at a cost of diminished SNR 193 compared to CASL. As a compromise between these two sequences, PCASL employs a train of discrete short RF pulses to mimic CASL’s continuous inversion to preserve both labeling efficiency and SNR 193,191 . CBF for PCASL is calculated as: = 6000 × × ( *,)-.,/ − /!0(/ ) × 123 4 $% ⁄ 2 × × '0 × 13 × (1− 67 4 $% ⁄ ) 23 where is the blood-brain partition coefficient, *,)-.,/ and /!0(/ are control and label images, '0 is the longitudinal relaxation of blood, is the labeling efficiency, 13 is the proton density image, and is the labeling duration 194 . Other variations of ASL include velocity-selective ASL (VSASL), which removes intravascular signal above a cutoff velocity and is more sensitive to capillary flow 195 , and time- encoded ASL (TE-ASL) which delivers both CBF and arterial transit time maps within the same scan 196 . Since there are so many different variations of ASL, the perfusion MRI community published a consensus article on the recommended implementation of ASL, consisting of PCASL labeling, background suppression, 3D readout, and one-compartment model for CBF computation 194 . However, since these guidelines are standardized for general perfusion imaging, they might not be suitable for specific pathologies. In the case of sickle cell disease, a few confounding variables need to be addressed. Firstly, the magnetic properties of sickled erythrocytes are different from healthy red cells, thus the '0 parameter should be adjusted 197,198 . Secondly, since SCD patients have higher baseline CBF 182 , the post-labeling delay is shortened to account for faster arrival in the tissue compartment 199 . Thirdly, since labeling efficiency is a function of blood flow velocity, instead of using a static value of 85% as recommended for PCASL, this value should be weighted by arterial velocity 199 . Lastly, two-compartment model should be used for SCD patients to differentiate contributions from the vasculature and tissue, since the two compartments have widely different longitudinal relaxation in SCD (850ms in tissue vs 1800ms in blood) 199,200 . Overall, ASL presents a non-invasive technique to measure perfusion but requires careful consideration of model parameters to attain accurate measurements, especially in pathologic conditions. 24 Dynamic susceptibility contrast (DSC) Even though ASL can provide CBF measurements non-invasively, ASL is plagued by poor signal-to-noise ratio since the endogenous tracer arterial blood only makes up ~1% of total brain tissue blood water 201 . If better signal is required, DSC MRI is a popular alternative technique to evaluate regional perfusion in various disease conditions. DSC relies on injection of a contrast agent to measure perfusion parameters, including CBF, cerebral blood volume (CBV), and mean transit time (MTT). The contrast agent used is usually gadolinium-based, including brand names such as Gadavist, Omniscan, Magnevist, and Multihance 202 , but recent works have employed deoxygenated hemoglobin as an endogenous source of contrast for perfusion-weighted imaging 203,204,205 . The principle behind DSC is the fast injection of a paramagnetic contrast agent to create susceptibility gradients that extend beyond the intravascular compartment, cause dephasing of extravascular spins and induce T2-weighted signal loss in the tissue compartment 206 . When the bolus of contrast agent is administered and passes through the cerebral vasculature, the MRI signal dips before returning to baseline (Figure 1.5A), and this transient signal loss is used to quantify perfusion using tracer kinetics model 207 . This model assumes that the contrast is non- diffusible across the blood-brain barrier, since extravasation of contrast into the interstitial space can lead to additional changes in T1 and requiring special strategy for leakage correction 208,209 . In terms of MRI acquisition, DSC is usually acquired using a dynamic gradient-echo or spin-echo sequence, with gradient-echo sensitive to all vessel sizes while spin-echo more sensitive to the microvasculature 210,211 . Single-echo pulse sequence is frequently used, but dual- echo DSC has become popular due to its ability to remove T1 leakage effects 212,213 . Additionally, DSC MRI requires fast dynamic acquisition, with TR < 2 seconds to account for fast vascular transit time. To standardize DSC methodology for multicenter clinical trials, consensus recommendations for DSC protocols have been published regarding pulse sequence, contrast dosing, and post-processing 214 . 25 A typical DSC signal intensity time curve () is shown in Figure 1.5A with baseline 8 ; this signal intensity is converted into changes in relaxation: ∆ " ∗ = − 1 ln W () 8 X and subsequently converted into a concentration time curve () before computation. Estimation of perfusion requires not only signal in the tissue but also input signal from the arteries, known as the arterial input function (AIF). AIF is usually determined globally from the middle cerebral artery but can also be determined locally to yield more accurate quantitation 215 . In cases where manual identification of the AIF is difficult, AIF of an average population can be used, or automatic detection of AIF can be applied to avoid observer subjectivity 216 . Additionally, pure intravascular blood signal from the middle cerebral artery is difficult to extract due to partial volume effects 217 , but the signal from the superior sagittal sinus is more robust since this vein has a large diameter, is almost perpendicular to the axial slices and reasonably parallel to the main magnetic field, thus minimizing field perturbations and partial volume effects 218,219 . Therefore, the extracted AIF signal is usually scaled by the area under the curve of the VOF to correct for partial volume 218 . Using DSC MRI, multiple perfusion parameters can be calculated, including CBF, cerebral blood volume (CBV), and mean transit time (MTT). CBV is calculated as: = ∫ -+::;( () ∫ <=> () where is the brain density 1.05 g/mL, and is the hematocrit correction factor: 220 Figure 1.5. (A) Gradient-echo signal from gadolinium-based DSC. (B) Gradient-echo signal from deoxygenation-based DSC. 26 = 1− 1−0.69 Estimation of CBF is more complex as it requires a deconvolution process: -+::;( () = <=>(-) ⨂(×()) where () is the monotonically decreasing residue function of the contrast fraction present in the vasculature with (0)=1 and (∞) =0 221 . This deconvolution can be performed using inverse Fourier transform 222 or singular value decomposition 223 . Finally, MTT is computed based on the central volume principle = %AB %A> . The applications of DSC MRI lie in its ability to detect, characterize, and monitor central nervous system diseases. DSC is most frequently used to assess brain tumors, especially for diagnosis of low- and high-grade gliomas 224,225 and differentiation between tumor progression and pseudoprogression following chemoradiation therapies 226,227 . Additionally, DSC has also been used to detect perfusion abnormalities and diagnose cognitive impairments in neurodegenerative conditions, such as Alzheimer’s disease 228,229 . Notably, DSC is frequently used in acute stroke care 230 , including localization of viable penumbra 231 and identification of patients who will benefit from reperfusion therapy 232,233 . Stroke imaging, especially in the case of steno-occlusive disease, requires correction for bolus arrival delay and dispersion 234,235,236 , which may result in up to 35% underestimation of CBF 237 . Additionally, in many diseases, the blood-brain barrier can be disrupted, requiring robust contrast leakage correction 238,239 . Therefore, careful application of post-processing techniques is required for DSC perfusion measurement, especially in pathologic conditions. v Gadolinium-based Standard DSC MRI uses gadolinium-based contrast agents. However, the risks associated with gadolinium experiments are still unclear and require further research. Free 27 gadolinium (Gd 3+ ) is intrinsically toxic, but Gd 3+ bound to a chelating agent form a stable molecular complex that can be safely administered intravenously 240,241 . Chelating agents are generally categorized as linear or macrocyclic, and macrocyclic agents tend to have higher structural stability 240 . Popular gadolinium agents with linear ligands include Omniscan and Magnevist, whereas the most well-known macrocyclic agent is Gadavist 240 . The most significant risk of gadolinium is development of nephrogenic systemic fibrosis (NSF), especially in patients with renal insufficiency and impaired clearance of gadolinium. The pathophysiology of NSF lies in the transmetallation of Gd 3+ from the chelating agent; these free Gd 3+ ions deposit into tissues, leading to increased recruitment of fibrocytes, causing tissue injury and resulting in NSF 242 . Therefore, renal impairment with glomerular filtration rate less than 30 mL/min/1.73m 2 is contraindicated against gadolinium-based DSC experiments and is recommended to use the more stable macrocyclic agents, whereas linear contrast agents are still considered safe to patients without renal insufficiency 243 . In the recent years, gadolinium deposition has also been demonstrated in the brain 244,245 , bone 246,247 , skin 248 , and liver 249 . Despite the known gadolinium deposition in tissue, there has been no proven clinical long-term effects, including cognitive decline, diminished neuropsychological or motor functions, from retention of gadolinium in healthy subjects 240 . Therefore, gadolinium- based DSC is still widely used in clinical practice, especially when perfusion-weighted imaging is necessary for diagnosis and management of neurologic pathologies. Other technical considerations for the use of gadolinium includes the fast rate of contrast injection in DSC MRI. Typically, contrast is administered at 5 mL/s using an 18-gauge catheter, since slower injection rates have been shown to underestimate perfusion 250,251 . The dosage of gadolinium also plays an important role, as too low dose yields low signal but too high dose results in MR signal hitting the noise floor; the usual injection dose is 0.1 mmol/kg 252 . Furthermore, since certain disease conditions are accompanied by impaired blood-brain barrier and contrast 28 extravasation, gadolinium preload dosing schemes (0.05 mmol/kg a few minutes prior and 0.05 mmol/kg during DSC) have been used to minimize the T1-weighted effects of contrast leakage 253 . One point of concern is that perfusion quantitation in gadolinium-based DSC is considered only ‘semi-quantitative’ due to unknown contrast relaxivity -+::;( . This relaxivity depends on the specific type of contrast, field strength, pulse sequence, flow velocity, tissue pathology, as well as distribution of gadolinium in microvessel topology 254 . In the computation process, changes in tissue relaxation rates are converted into concentration time curves -+::;( (), assuming constant and linear -+::;( across the brain: ∆ ",-+::;( ∗ ()= -+::;( × -+::;( () with -+::;( as 44 mM -1 s -1 for 1.5T and 87 mM -1 s -1 for 3T 255 . On the other hand, the relation between ∆ ",<=> ∗ () and <=> () has been shown to be quadratic and hematocrit-dependent 256 : ∆ ",<=> ∗ = ' <=> ()+ " 1.1378 (1−) " ( <=> ()) " where ' is 7.62 mM -1 s -1 and " is 0.57 mM -1 s -1 for 1.5T, and ' is 0.49 mM -1 s -1 and " is 2.62 mM -1 s -1 for 3T 257,258 . However, in practical implementation, a linear relationship is typically assumed 259 : ∆ ",<=> ∗ ()= <=> × <=> () Since absolute quantitation in DSC MRI requires knowledge of . &'(()* . +,- and these values have only been estimated globally and through simulations, CBV and CBF values computed are considered ‘semi-quantitative’. In clinical routines, however, . &'(()* . +,- uses simulated values for different tissues 260 or most of the time is assumed to be 1. 29 v Deoxygenation-based Due to the risks and limitations associated with gadolinium, in recent years, respiratory challenges have been proposed as a safer alternative for DSC imaging. Transient gas challenges, such as hyperoxia (high oxygen) and hypoxia (low oxygen), have been demonstrated to change the blood oxygen content, modulate the deoxyhemoglobin concentration, and cause T2-weighted change in the MRI signal, similar to the effects of gadolinium 203,204,205 (Figure 1.5B). The hypoxia perfusion technique is termed deoxygenation-based DSC (dDSC) and relies on the principle that since deoxyhemoglobin is paramagnetic, bolus passage of deoxygenated blood induces susceptibility changes on extravascular spins in the tissue and results in T2- weighted loss on the tissue MRI signal 261 . Even though oxygen is freely diffusible between the blood-brain barrier, erythrocytes are not diffusible and thus the deoxyhemoglobin-based contrast is constrained within the vasculature, enabling perfusion calculation using tracer kinetics model. Recently, novel respiratory paradigms have been proposed and demonstrated to yield reasonable perfusion measurements, including challenge patterns such as desaturation, resaturation, sine- wave hypoxia as well as sine-wave hypercapnia 204 . A few technical implementations differ between gadolinium-based and deoxygenation- based DSC. Notably, gadolinium-specific values of -+::;( , <=> or their ratio cannot be used for dDSC computation since there is no exogenous contrast agent involved. Currently, this ratio is assumed to be 1, but further work is required to assess the linearity assumption between cerebral oxygen level and ∆ " ∗ in healthy as well as disease conditions 203,262 . Population-based AIF and even some automatic AIF methods optimized to gadolinium DSC might be unsuitable for dDSC and require further investigation before being used in non-gadolinium studies. Furthermore, it is important to recognize that deoxygenation-induced MRI signal changes are much smaller compared to gadolinium-induced, decreasing the SNR and requiring more robust noise suppression for accurate quantitation. Most importantly, gadolinium contrast does not influence the perfusion parameters of interest. On the other hand, hypoxia or hyperoxia have been shown 30 to cause vasodilation and vasoconstriction effects respectively 263 , so even transient exposure of these gases might have hemodynamic influence and potentially prevents accurate measurement of steady-state perfusion. Therefore, a more in-depth investigation into stimulus-mediated perfusion changes is required to assess their contribution to our measurements. Gadolinium-based DSC implementation can be standardized using recommended guidelines from the perfusion community 214 , but dDSC is a relatively new area of research with no consensus statement yet. Previous works have performed dDSC using fixed inspired gas challenges 203,205 as well as end-tidal targeted respiratory challenges using a computer-controlled gas blender (RespirAct, Thornhill Research, Toronto, Canada) 204 . The RespirAct’s capability to precisely modulate gas concentrations is considerably advantageous; however, this gas blender is expensive and currently not approved by the FDA, limiting its widespread usage in clinical routines. Overall, even though dDSC does not yield as strong a signal drop as gadolinium-based effects, it eliminates the reliance on exogenous contrast and enables safe, non-invasive DSC experiments, especially for renal-impaired and pediatric populations. 1.2.2. Cerebrovascular reactivity (CVR) In addition to baseline CBF measurements, cerebrovascular reactivity (CVR) – the ability of vessels to dilate or constrict in response to changes in physiological demand – can be measured to complement steady-state CBF to evaluate cerebral hemodynamic impairments in different pathologies. CVR estimation requires three key steps: the vasoactive stimulus, the signal acquisition, and the processing method 264 . A vasoactive stimulus is used to induce an increase or decrease in CBF compared to baseline, commonly through injection of a carbonic anhydrase inhibitor acetazolamide or changes in arterial CO2 partial pressures 265,266 . There is a variety of techniques to induce P aCO2 changes, 31 including breath-holding 267,268 , rebreathing 170 , hyperventilation 269 , and inhalation of CO2-enriched gases, using either fixed inspired gas administrations or end-tidal CO2 targeting methods 270 . Additionally, O2-enriched gases such as hyperoxia or carbogen can also induce hemodynamic changes but are much less frequently used 264 . MRI signal acquisitions are commonly performed either dynamically (BOLD) or by acquiring perfusion measurements at baseline and under stimulus (ASL or PC). CVR is usually calculated from MRI signal using general linear model 266 as: = :-+D;/;: − 0!:(/+)( " :-+D;/;: − " 0!:(/+)( but under maximally dilated vessels using acetazolamide can also be computed with: = :-+D;/;: − 0!:(/+)( 0!:(/+)( Oftentimes, cross-correlation is used to quantify and correct for delay between and " signals 271 , but frequency-based CVR analyses have proven robust and independent of vascular delay 272,273 . CVR mapping has been used to evaluate impaired hemodynamics in a range of clinical pathologies, including steno-occlusive diseases 274,275 , brain tumors 276,277 , neurodegenerative diseases 278 , traumatic brain injuries 279 , and ischemic strokes 280 . Specially in the evaluation of stroke risk, CVR impairment is associated with increased risk of ischemic events 281,282 and co- localize with watershed regions vulnerable to SCI development 283 . Therefore, measurement of CVR complements steady-state CBF and serves as an indicator for impaired hemodynamics in a range of cerebral pathologies. 1.2.3. Oxygenation extraction fraction (OEF) In certain pathologies with impaired blood flow, oxygen extraction fraction (OEF) can increase to maintain tissue function under conditions of low oxygen availability or high metabolic 32 demand, thus avoiding potential tissue hypoxia and ischemic injury 284 . OEF is defined as the percent arteriovenous saturation difference = ! − # ! where ! and # are the arterial and venous saturations respectively. Recently, OEF has been investigated in multiple pathologies, including steno-occlusive diseases 149 , sickle cell anemia 149,285 , multiple sclerosis 286 , and congenital heart diseases 287 . Increased OEF has been demonstrated as a biomarker for risk of subsequent stroke 288,289 and used for detection of the viable penumbra through identification of areas with misery perfusion (low CBF, high OEF) 290,291 . On the other hand, diminished OEF indicates the existence of a physical arteriovenous shunt 292 or a physiological shunting effect in which blood flow is too fast for efficient capillary oxygen extraction 293 . There is a wide range of techniques to measure OEF. Systemic OEF can be obtained by measuring # using blood draw from the brachial artery or continuous sampling from the central venous catheter 294 . Cerebral OEF, on the other hand, cannot be measured in vivo with catherization and can only be approximated using saturation from the internal jugular vein, whose difference from the superior sagittal sinus is only 1-2% under resting conditions 295,296,297 . Unlike physical catherization, oxygenation imaging enables non-invasive measurement of cerebral OEF. These techniques are subdivided into tissue-based methods to measure OEF within cerebral tissues and flow-based methods to measure OEF in large veins such as the superior sagittal sinus. Tissue-based A simplified model of the brain is divided into two compartments: intravascular and extravascular. The extravascular compartment refers to water protons that are outside the blood vessels and in the cerebral tissues. The spins that are in this extravascular space have distinct 33 magnetic properties compared to intravascular spins; therefore, OEF techniques that are based on MRI signals arising from extravascular spins are sensitive to oxygen usage within brain tissues. v Asymmetric spin echo (ASE) A well-known tissue-based MRI technique to measure OEF is asymmetric spin echo (ASE), which makes use of the paramagnetic effects of intravascular deoxygenated blood whose susceptibility effects extend beyond the intravascular compartment and create local magnetic field inhomogeneities that can cause dephasing of extravascular spins. The dephased spins lead to altered " ∗ and " , thus enabling OEF measurements in the tissue. Theoretically, the resulting MRI signal comprises of contributions from two compartments – intravascular and extravascular; however, since cerebral blood volume fraction is approximately 3-5%, the intravascular contribution is negligible 298,299 . ASE is implemented as a modified spin echo sequence, in which for each echo the 180- degree RF pulse is shifted from 2 ⁄ by a time interval ; ASE acquires 28 echoes, 21 of which are asymmetric with from 10 to 20 ms, and 7 of which are true spin echoes with =0 299 . This pulse sequence can also include diffusion gradients to suppress intravascular signals 299,300 as well as background field correction as a post-processing step to remove effects of macroscopic static field inhomogeneities to obtain more accurate OEF estimates 301 . Within the static dephasing regime, the ASE signal () is a function of the susceptibility- induced frequency shift: = × 4 3 × ∆ 8 × 8 × × where is the gyromagnetic ratio, 8 is the main magnetic field strength, and ∆ 8 is the susceptibility difference between fully-oxygenated and fully-deoxygenated blood 302 . Since the expression for () cannot be solved analytically, the signal can be approximated by two asymptotic forms: 34 : ()= × (1−) × 68.FG×(IJ×-) " for short time |×|≤1.5, and : ()= × (1−) × 6K " . (|-|6- / ) for long time |×|≥1.5, where is the venous blood volume, is the spin density, and * is the critical time * = ' IJ . The relaxation parameter " M is computed from a linear fit of the natural log of : (), whereas can be estimated from: =ln ( / ( =0))−ln ( : ( =0)) with " M and , can be calculated as = K " . G , and OEF can be calculated subsequently based on the definition of . ASE has the potential in a vast range of applications, from clinical assessments of pathologies in skeletal muscle 303,304 , kidney 305 , to brain 306,307 . Specifically, ASE has been used extensively in exploring cerebral metabolic health in sickle cell disease 308,309,310 . Additionally, OEF measured by ASE has demonstrated co-localization between areas of high oxygen extraction and deep white matter regions with high SCI density in SCD 311 , thus suggesting this technique’s capability to identify patients at risk of stroke and to detect perfusion territories most vulnerable to SCI development. A side note is that the assumption of two compartments – intravascular and extravascular – in ASE might be too simple in the brain, so a novel technique called quantitative BOLD (qBOLD) divides the model into more compartments, including intravascular, extravascular grey matter, extravascular white matter, and extracellular (CSF) and assesses contributions of each compartment to the MRI signal. This qBOLD model is considerably more complex compared to ASE but is able to account for various confounding factors to yield tissue-based OEF 312 . 35 v Near-infrared spectroscopy (NIRS) Another method for measuring tissue oxygenation is near-infrared spectroscopy (NIRS), which is a non-invasive optical technique to evaluate oxygen availability and consumption in tissue 313 . Since NIRS operates in the near-infrared 700-950 nm spectrum, the light can traverse biological tissue and is attenuated due to scattering or absorption by chromophores 314 . Typical chromophores consist of hemoglobin, myoglobin, and cytochrome oxidase 315 , in which myoglobin is normally present only in skeletal and cardiac muscles 316,317 , and cytochrome oxidase is only 5% concentration compared to the rest 315,318 . Therefore, the majority of cerebral NIRS signal arises from hemoglobin within capillary networks in brain tissues, making this technique particularly sensitive to tissue oxygenation. NIRS relies on optical attenuation by these chromophores to derive the oxyhemoglobin (OxyHb) and deoxyhemoglobin (DeoxyHb) concentrations based on modified Beer-Lambert law 314,319 . From these values, the total cerebral blood volume can be calculated with: []=[]+[] Brain tissue oxygen saturation, also known as tissue oxygenation index (TOI), can be calculated. Even though the formula for TOI is simple: = [] [] different brand-name devices may yield different values due to use of proprietary algorithms and various undeclared scaling factors 314 . Despite potential confounding factors to the NIRS cerebral signals 320–322 , the validity of NIRS measurements have been proven by many studies, ranging from its agreement with blood- gas analysis 323,324 and even with MRI ASE tissue oxygenation measurements 304 . Overall, NIRS has become an extremely popular method for cerebral oxygen quantitation, especially due to its non-invasive nature and ease of use. It is frequently used for monitoring during surgeries 325,326 , for detecting oxygen impairment in ischemic stroke or traumatic brain injury 327,328 , for managing 36 of anemia syndromes 329 , and even for diagnosing different psychiatric disorders that lack precise diagnostic laboratory tests 330,331 . Flow-based Compared to tissue-based methods for measuring OEF, flow-based techniques focus on measuring venous saturation in large veins, such as the superior sagittal sinus, straight sinus, and internal jugular vein. Under normal conditions, it is reasonable to assume that the saturations in the tissue and in a large vein are similar. Previous works have demonstrated good agreement between NIRS and peripheral venous saturation by co-oximetry 324,332 as well as between tissue- based qBOLD OEF and venous OEF with blood drawn directly from the superior sagittal sinus 333 . However, such an agreement might not hold up under pathologic conditions, such as chronic anemia in which OEF is increased in the deep brain structures but not in the cerebral cortex. 334 In conditions such as cerebral microvascular dysfunction, diffuse abnormalities in perfusion 335 , reactivity 336 , response to perturbations 261,337 , and capillary transit time 38 can contribute to the decoupling of cerebral blood flow and metabolism 338 . Such heterogeneous distribution of perfusion means that some capillary beds experience normal OEF whereas others can display characteristics of physiological shunting, with abnormally high flow and consequently inefficient oxygen extraction 150 . These shunt-like beds contribute more flow to the draining veins compared to normally capillary beds, leading to low OEF in the superior sagittal sinus 37 . This difference between flow-based and tissue-based techniques is corroborated by the difference in flow-weighted OEF measured by optical techniques versus spatially averaged OEF, which diverges even more under hyperperfusion 339 . Additionally, reduced OEF has been demonstrated under hyperemia by jugular venous catherization 340,341 , as opposed to the increased tissue OEF observed by ASE in hyperemic anemia 306 . Therefore, the divergence between flow-based and 37 tissue-based OEF methodologies is not just theoretical and requires better understanding when these techniques are applied to different cerebrovascular pathologies. Overall, flow-based oximetry techniques measure saturation in cerebral veins, namely the superior sagittal sinus, straight sinus, or internal jugular vein. There are two main classes of flow- based techniques, including T2-weighted methods that rely on the magnetic transverse relaxation of venous blood and susceptibility-weighted methods that use the susceptibility difference between venous blood and tissue to quantify oxygenation. v T2-weighted T2-weighted relaxometry methods typically take advantage of the calibratable relationship between blood T2 and oxygenation, which despite its simple nature remains a controversial topic of discussion, especially for hemoglobinopathies with altered red cell morphology 342,285,343,151 . These techniques measure venous blood T2 relaxation, which is related to blood oxygenation and hematocrit through a non-linear model 344,345,346,347 . The most widely used calibration model is proposed by Lu et al. 348 : 1 " = +(1−)+(1−) " where , , coefficients are empirically determined, hematocrit-dependent and of the forms: = ' + " + F " = ' + " " = ' (1−) Temperature-controlled in vitro experiments using bovine blood, which has similar magnetic properties compared to human blood 349 , at different hematocrits and oxygenation levels yield values for ' , " , F , ' , " , and ' that are dependent on the inter-echo spacing %1NO of the T2 preparation module. For %1NO =10, the coefficient values are: ' =−13.5 6' , " = 80.2 6' , F =−75.9 6' , ' =−0.5 6' , " =3.4 6' , and ' =247.4 6' . 38 The bovine calibration model has proven robust for converting human blood T2 into venous saturation 342 . However, this non-linear calibration model was empirically determined within a narrow range of hematocrit 0.34 – 0.42. If this model needs to be applied to hematocrit levels outside of the calibrated range, such as for patients who are cyanotic (high hematocrit) or anemic (low hematocrit), the converted oxygenation values might not be accurate. This theory is proven in a cohort of anemic thalassemia patients, in which the bovine calibration yields fairly accurate saturation estimates for hematocrit > 0.30 but severely underestimates saturation for hematocrit < 0.30 350 (Figure 1.6). Therefore, for anemic subjects, a different oximetry model calibrated over the appropriate hematocrit range is required to attain accurate measurements. Based on this observation, a simplified model is proposed and calibrated over hematocrit range of 0.10 – 0.55: 1 " = ' (1−) " + " (1−) " + F + P where ' =77.5, " =27.8, F =6.95, and P =2.34 for %1NO =10 350 . This model has been validated in healthy as well as anemic blood 350,151 and has been applied to compute OEF and oxygen metabolism in chronically anemic patients 150 . Even though the similarity in magnetic properties between bovine blood and healthy human blood has been demonstrated 349 , this parallel has not been shown for hemoglobinopathies with altered red cell morphology, such as in sickle cell disease. In SCD, high percentage of hemoglobin S in erythrocytes cause these red cells to polymerize under deoxygenated conditions, resulting in Figure 1.6. Estimation of venous R2 and saturation using bovine, neonatal, and human calibration models demonstrating underestimation at hematocrits under 0.30. Figure adapted from Bush et al., MRM, 2016. 39 deformed crescent-shaped red blood cells. Even though there is no susceptibility difference between hemoglobin A and hemoglobin S 351 , sickled erythrocytes have very different cell shape and membrane permeability compared to normal human blood cells 352 . Increased membrane permeability potentially magnifies the mesoscopic magnetic gradients that diffusing water protons experience, thus affecting the calibration model. Similarly, cell shapes have been demonstrated to affect CPMG signals 353 , which potentially impact the measured blood T2 as well as the calibration coefficients. Therefore, a separate calibration is required when evaluating SCD patients, especially for patients with severe phenotypes and higher percentage of sickled-shape erythrocytes such as Hb SS and Hb Sβ 0 . Given these differences, a sickle-specific calibration is proposed: 1 " =(1−) " + where and are empirically determined coefficients with =70.0 and =5.75 for %1NO =10 285 . This calibration is derived from a relatively small range of hematocrits 0.24 – 0.40, which might have caused the lack of hematocrit dependence and can lead to errors in SCD blood at lower hematocrit. However, this calibration model has been applied by various studies in SCD patients and has demonstrated agreement with blood-gas analysis in the brachial artery 285 . Overall, choosing a suitable T2 oximetry calibration is extremely important to obtain accurate venous saturation. Therefore, in these oximetry studies, care must be taken to evaluate the red cell morphology and its magnetic properties in order to choose the most appropriate calibration model. After choosing a suitable calibration model, the remaining obstacle to # estimation is the isolation of pure venous blood signal from the surrounding tissues, which can be solved using two different approaches, spin tagging and phase contrast. 40 • T2-weighted: T2 relaxation under phase contrast (TRUPC) Phase contrast (PC) has been previously discussed as a technique to measure total brain blood flow in the internal carotid and vertebral arteries. Under the influence of bipolar gradients, stationary spins experience no net phase shift, whereas moving spins retain a phase shift proportional to their velocity along the gradient’s direction 183 . Therefore, based on the phase signals, total CBF can be computed. To measure oxygen saturation in cerebral veins, a phase contrast-based technique called T2-relaxation-under-phase-contrast (TRUPC) is invented, in which the PC bipolar gradients are applied to a draining vein to isolate pure venous blood signal, and a T2 preparation module is used provide T2 weighting and measure venous blood T2. This blood T2 value can be converted to venous saturation through a calibration model of choice 354 . Previous work has demonstrated feasibility of utilizing TRUPC to measure venous saturations under normoxia and hypoxia in multiple draining veins, including superior sagittal sinus, straight sinus, great vein, and internal cerebral vein 354 . Most interestingly, TRUPC is not only able to target major veins but also small cerebral veins by using a lower velocity encoding 355 , which can have potential significance in characterizing capillary perfusion and oxygenation heterogeneity in cerebrovascular diseases. • T2-weighted: T2 relaxation under spin tagging (TRUST) Compared to TRUPC, T2-relaxation-under-spin-tagging (TRUST) does not use phase contrast to isolate pure blood signal, instead opting for a spin labeling technique similar to ASL to tag moving protons in a draining vein. The distinction in spin tagging between TRUST and ASL is, whereas ASL places the labeling plane below the imaging plane to tag protons in the arteries, TRUST labels protons above the imaging slice on the venous side 348,356 . Even though both TRUST and ASL take advantage of the difference between label and control images for quantitation, TRUST is not SNR poor because the difference signal is 80% of the control signal, compared to 2% in ASL, thus yielding a very good coefficient of variations across trials of 2.0% 348 . 41 Similar to TRUPC, TRUST implementation includes a T2 preparation module to apply T2 weighting and measure blood T2, which can be converted to venous saturation 348,342 . TRUST has been validated against various modalities including PET 357 , TRUPC 354 , pulse oximetry 342 as well as under different gas respiratory perturbations 263 . Additionally, previous work has demonstrated comparable venous saturation measurements using TRUST across different clinical sites and different MRI vendors 358,359 . As such a robust, rapid, well-tolerated and reproducible oximetry technique, TRUST has become a popular technique to measure oxygenation in various pathologies, including multiple sclerosis 286 , steno-occlusive disease 149 , addiction 360 , and neurodegenerative disease 361 . Notably, TRUST is frequently applied to evaluate oxygen supply and demand in sickle cell disease in order to assess risk of overt and silent strokes in this population 150,285,149,343 . v Susceptibility-based Compared to T2-weighted techniques that measure venous saturation based on relaxometry, susceptibility-based oximetry (SBO) estimates oxygenation based on the magnetic susceptibility difference between venous blood and surrounding tissue. This technique has been used to quantify oxygenation in different cerebral veins, including pial veins 362 , internal jugular veins 363 , and sagittal sinus 364 . In SBO, the susceptibility shift ∆ #(+) 6-+::;( can be estimated using the phase difference ∆Φ from phase images and the magnetic field difference ∆ #(+) 6-+::;( between a candidate vein and the brain parenchyma: ∆Φ=××∆ #(+) 6-+::;( ∆ #(+) 6-+::;( = 1 6 ×4×∆ #(+) 6-+::;( ×(3 " −1)× 8 where is the gyromagnetic ratio, and is the angle between the vein and the main magnetic field 8 . 42 Subsequently, venous saturation # can be calculated from the susceptibility shift with ∆ #(+) 6-+::;( =∆ Q, ××(1− # ) where ∆ Q, is the susceptibility difference between fully deoxygenated and fully oxygenated blood 365 . This value of ∆ Q, has been a recent topic of discussion, as previous works have demonstrated two separate values 0.18 ppm 366 or 0.27 ppm 367 based on in vitro experiments. One important issue frequently overlooked in SBO studies is the large variance of phase values along the length of the cerebral vein of interest. As explored in the superior sagittal sinus, an average variation of 22% absolute saturation point was observed along the vein, which impacts the confidence level of oxygenation measured 295 . Since the magnetic field measurement within the sagittal sinus is influenced by the background field due to air-tissue interface, applying SBO on smaller pial veins away from air-tissue interface might improve precision. However, this large variation represents a disadvantage of SBO compared to other the low coefficient of variation in T2-based TRUST MRI 348 . 1.2.4. Cerebral metabolic rate of oxygen (CMRO2) In addition to blood flow and oxygen extraction, another important metrics frequently used in assessing brain health is cerebral metabolism, including the metabolic rate of glucose (CMRglu) and of oxygen (CMRO2). Amongst these, the most applicable to stroke risk is CMRO2, which represents the demand of oxygen by the brain; if this demand is unmet by the delivery (CBF) or extraction (OEF) of oxygen, tissue hypoxia and ischemia can happen, resulting in cerebrovascular infarcts. Evaluation of CMRO2 in patients with ischemic strokes and head injuries has traditionally been performed with PET imaging 368,369 ; however, for non-acute conditions that need frequent monitoring, a less invasive and non-ionizing imaging alternative is preferred. Therefore, MRI has 43 recently become a popular method for assessment of impaired hemodynamics in patients with brain pathologies. CMRO2 can be computed using the Fick’s principle: " = " ×× where " is the arterial oxygen content calculated by: " =1.34×× ! +0.003× " where ! is arterial saturation measured with pulse oximetry, and " is the partial pressure of oxygen estimated at 100 mmHg at room air. Based on the Fick’s principle, if the three parameters CaO 2, CBF, and OEF can be measured, quantitative values of CMRO2 can be computed. If spatial distribution or temporal changes of CMRO2 is of interest, then calibrated BOLD is a specialized functional MRI technique to estimate regional CMRO2 changes in response to task activation. On the other hand, if only global CMRO2 value is needed for each individual patient, then CMRO2 can be calculated using a multi-modal approach with the input parameters to Fick’s equation measured using separate imaging methods. Calibrated BOLD Calibrated fMRI is a technique to determine changes in CMRO2 associated with a task, such as finger tapping or visual stimulation 370 . The task-induced BOLD response is a complex function of underlying changes in cerebral oxygenation, blood volume, blood flow and metabolic rate. This BOLD response and its underlying physiological changes are typically simulated using the Davis biophysical model: ∆A&23 A&23 0 =s1−t %AB %AB 0 ut [QT0] [QT0] 0 u V v [eq 1] where ∆A&23 A&23 0 is the baseline-normalized changes in the BOLD signal, %AB %AB 0 and [QT0] [QT0] 0 are the normalized CBV and deoxyhemoglobin concentrations respectively 371 . 44 Another variation of equation can be written using Grubb’s law: 372 %AB %AB 0 = t %A> %A> 0 u W and modified Fick’s principle: %NK& " %NK& ",0 =t %A> %A> 0 uW %!& " ×&X> %!& ",0 ×&X> 0 X [eq 2] resulting in the BOLD signal as a function of changes in CBF and CMRO2: ∆A&23 A&23 0 =w1−t %A> %A> 0 u W6V W %NK& " %NK& ",0 X V x [eq 3] To acquire relative changes in CMRO2 in time from equation 3, changes in both BOLD signal and CBF need to be measured. Therefore, simultaneous measurements of BOLD and CBF are accomplished using a dual acquisition ASL-BOLD pulse sequence 373 , thus enabling dynamic estimation of ∆A&23 A&23 0 and %A> %A> 0 in the same imaging sequence. Both derivations of the Davis model (equations 1 and 3) define a BOLD scaling constant , which is the maximum possible BOLD signal change and is usually calibrated using either a respiratory challenge 371,374 or a relaxometry experiment 375 . Additionally, both the and constants and have also been important topic of discussions. is typically assumed as =0.38 from Grubb’s law 372 , but modern work has demonstrated a smaller value of =0.23. 376 On the other hand, is a biophysical constant related to vessel sizes, with =2 for capillaries and = 1 for larger veins. A value of =1.5 is typically used for 1.5T and =1.3 for 3T 371,377 . As seen in the above equations, calibrated BOLD provides relative changes in CMRO2 during task-induced neuronal activation; however, newer work has demonstrated the ability to estimate CMRO2 in absolute units. Based on Fick’s principle in equation 2, since is calculated dynamically with ASL, " is computed from pulse oximetry, quantitative values of " at baseline and during task-induced activation can be computed if can be measured. Values of OEF can be estimated using a variety of techniques. Gauthier et al. estimates OEF with a combination hypercapnia and hyperoxia respiratory challenges during calibrated fMRI. 45 Based on equation 1, the normalized deoxyhemoglobin concentration can be expressed as in terms of the reciprocal normalized flow: [QT0] [QT0] 0 = t %A> 0 %A> u+ ={ %!& ",0 ×&X> 0 '6 2#3 ",0 $.56[8%] ('6&X> 0 ) | t %A> 0 %A> u+{ '6 2#3 " $.56[8%] '6 2#3 " $.56[8%] ('6&X> 0 ) | [eq 4] where both and terms are functions of 8 . Therefore, the [QT0] [QT0] 0 term in equation 1 can be substituted from equation 4, thus yielding as a function of 8 : = ∆A&23 A&23 0 ⁄ '6Y 2:- 2:- 0 Z ; Y[ 2:- 0 2:- \] Z < =( 8 ) [eq 5] This relationship =( 8 ) is determined under hyperoxia and hypercapnia separately, and the intersection of these two functions yields quantitative values for and 8 378 . On the other hand, instead of performing two different respiratory challenges, Bulte et al. assumes that net extracted oxygen content (not oxygen extraction fraction) does not change under hyperoxia challenge " − " = ",8 − ",8 Normalized deoxyhemoglobin can be computed from venous saturation estimated by end- tidal pressures and Severinghaus equation 379 , yielding: [] [] 8 = # #,8 Therefore, is expressed as: = ! −{ ",8 −( " − " ) 1.34[] | where the difference in arteriovenous oxygen content is assumed as constant and calculated as: 380 " − " ={ [QT0] [QT0] 0 −1− Y [=8%] [=8%] 0 Z%!& ",0 6%!& " '.FP[T0] | t '6([QT0] [QT0] 0 ⁄ ) '.FP[T0] u 6' [eq 6] 46 Alternatively, can also be calculated using susceptibility-based oximetry (SBO) 363,364 to measure dynamic # and baseline #,8 . Dynamic measurement # within a calibrated BOLD acquisition is enabled by replacing the dual ASL-BOLD acquisition with a three-part interleaved sequence comprising of ASL, BOLD and dual-echo gradient-echo, the last of which is applied in SBO in the superior sagittal sinus to derive # at each time point 381 . All three methods yield OEF estimates and when combined with CBF measured by ASL can calculate resting CMRO2 based on Fick’s principle. Multi-model approach The advantage of using calibrated BOLD is its ability to estimate the distribution of CMRO 2 across different areas of the brain, but this method relies on separate calibration of the M parameter using a respiratory challenge, which might be different in individuals as well as under pathologic conditions. Instead of spatial mapping of metabolism, if only a global value of CMRO2 is of interest, then a multi-model approach to measure separate input parameters to the Fick’s equation is employed to yield CMRO2 estimation 382 . According to the Fick’s principle: " = " ×× The oxygen content " can be calculated as " =1.34×× ! +0.003× " , with measured from a blood draw, ! is the arterial saturation measured with fingertip pulse oximetry, and " is assumed as 100 mmHg under room air. CBF can be estimated using either phase contrast or ASL MRI, and cerebral OEF can be acquired using TRUPC or TRUST MRI using appropriate hemoglobin-specific oximetry calibration. Therefore, from these parameters, CMRO2 can be calculated for each individual patient. This multi-modal approach is easily performed, less invasive due to lack of respiratory challenge and highly precise with a coefficient 47 of variation of 3.84% 383 , thus enabling its widespread usage in multiple pathologies, including in neonates 384 , chronic anemias 150 , and addiction 360 . 48 Chapter 2 : Quantitative Perfusion Mapping with Induced Transient Hypoxia using BOLD MRI 2.1. Introduction Gadolinium-based dynamic susceptibility contrast (DSC) imaging is the most commonly used methodology to assess cerebral perfusion and characterize tumor hemodynamics in clinical studies. 385 With the injection of a gadolinium exogenous contrast, DSC utilizes a series of echo- planar gradient-echo acquisitions to monitor the first pass of this contrast bolus through the capillary beds and derives an assortment of cerebral perfusion metrics, including cerebral blood flow (CBF), cerebral blood volume (CBV) and mean transit time (MTT). 206 The strength of this technique lies in its high temporal resolution to capture the bolus passage through brain tissue on the order of a few seconds and its ability to generate hemodynamic information within the same scan. In the last few decades, DSC has proven to be a valuable tool for evaluation of ischemic and infarcted tissues in stroke, 386 differential diagnosis of intracranial masses 387 and assessment of blood-brain barrier permeability in traumatic brain injury. 388 However, despite its utility in numerous clinical applications, DSC is limited by its reliance on various gadolinium-based contrast agents, each of which has a known or hypothetical association with nephrogenic systemic fibrosis in patients with kidney disease. 389,390 Even in patients with normal renal function, gadolinium agents have been demonstrated to accumulate in tissues in the brain, bone and kidney. 391–393 In recent years, oxygen has been employed as a safer alternative to exogenous contrast 394 since the over-supply (hyperoxia) or the lack of oxygen (hypoxia) can change the blood oxygen 49 content, modulate the deoxyhemoglobin concentration and cause a T2-weighted change of the measured MR signal. MacDonald et al. have previously demonstrated the use of hyperoxia- induced signal dynamics to estimate perfusion parameters. 205 However, Losert et al. showed hyperoxia causes a relatively small increase in oxygen saturation and percent signal change in the white matter (less than 1%), 394 which prevents accurate flow measurement and examination of the role of compromised CBF in white matter disease. 395 Therefore, we conjectured that briefly increasing the concentration of paramagnetic deoxyhemoglobin via desaturation would produce a larger and more robust signal change, especially in the white matter. In this study, we explored the feasibility of using a transient hypoxia gas paradigm (Figure 2.1), in a similar manner to gadolinium injection, to deliver a bolus of paramagnetic material (deoxygenated hemoglobin) to the cerebral vasculature. The contrast passage through the microvasculature was monitored by a dynamic blood-oxygen-level-dependent (BOLD) MR sequence (Figure 2.2). Through tracer kinetics modeling, the BOLD signal change induced by this hypoxic bolus can be used to generate regional perfusion maps; we term this approach deoxygenation-based DSC (dDSC). In order to test the new technique over a wide range of CBF, Figure 2.1. Experimental setup for hypoxia challenge and concurrent SpO2 and BOLD acquisitions. During the experiment, the subjects breathed through the mouthpiece through a two-liter reservoir rebreathing circuit that included one-way valves to prevent partial gas mixtures. The subject also wore a respiratory bellows to display the breathing pattern and frequency. 50 we evaluated this technique on a total of 66 subjects, including healthy controls as well as anemic patients (sickle cell disease and non-sickle anemia) who have higher baseline blood flow. 396,397 Comparison with other popular CBF quantification methods such as arterial spin labeling (ASL) 194 and phase contrast (PC) 398 can yield further insights into the feasibility of using transient hypoxia as a contrast mechanism for dDSC perfusion studies. 2.2. Methods 2.2.1. Theory In DSC imaging, the changes in relaxation rate DR2 * are directly proportional to the gadolinium concentration. 399,400 In parallel, this dDSC approach assumes that the changes of oxygen level in the brain can be approximated by a linear function of DR2 * during the gas paradigm: -+::;( ()= D " ∗ [1] Figure 2.2. Transient hypoxia model. (A) Representative recording of SpO2 signal during 100% nitrogen paradigm. (B) SpO2 signal from the same patient prior to gas paradigm while patient was sleeping during anatomic scanning. (C) Representative time series of global BOLD-MR signal and SpO2 signals during hypoxia paradigm. (D) Representative ∆ ! ∗ time curve and its time integral (area under curve). 51 where Ctissue(t) is the oxygen concentration time curve measured in the tissue and k is a proportionality constant that depends on tissue type, contrast agent, field strength and pulse sequence. 401 In this experiment design, voxel-wise DR2 * is calculated with: D " ∗ () = − ' 4X ln ( ^(-) ^ 0 ) [2] where S(t) is the gradient-echo time series, S0 is the baseline signal computed from the first 25 dynamics prior to bolus administration and TE is the echo time. Arterial input function (AIF) and venous output function (VOF) are extracted from the middle cerebral artery (Figure 2.3A) the superior sagittal sinus (Figure 2.3B) respectively. Due to difficulty extracting a pure intravascular blood signal from the middle cerebral artery because of partial volume effect, 217 the time integral of the AIF is scaled to match that of the VOF 402 to yield an ‘effective AIF’ signal eAIF(t) such that: ∫D " ∗ (<=> ()= ∫D " ∗ B&> () [3] Regional CBV is computed using DR2 * from the tissue and effective AIF: = _ ` ∫DK " ∗ &'(()* (-) ∫DK " ∗ *+,- (-) [4] where is the brain density 1.05g/mL and is the hematocrit correction factor. The measure = '6T*- '68.bcT*- accounts for the difference between the hematocrit in the large vessels (middle cerebral artery for AIF determination) and in the microvasculature. 220 From tracer kinetics modeling, the measured Ctissue(t) is expressed as: -+::;( ()=()⊗( × ()) [5] 52 where R(t) is the residue function of the contrast fraction present in the vasculature at time t with (0)= 1 and (∞)= 0. 399 To compute CBF, a deconvolution can be performed by an inverse Fourier transform approach 403 or a singular value decomposition (SVD) method proposed by Ostergaard et al. 404 Stable solutions for inverse discrete Fourier transforms can only be obtained with effective experimental noise suppression, whereas SVD has been shown to reproduce flow and is independent of underlying vasculature and volume. 404 In this study, we used the SVD deconvolution approach, with the previous convolution discretized in the following form: -+::;( ( ' ) -+::;( ( " ) ⋮ -+::;( ( d ) =D ( ' ) 0 … 0 ( " ) ( ' ) … 0 … … … … ( d ) ( d6' ) … ( ' ) × ( ' ) × ( " ) ⋮ × ( d ) [6] Figure 2.3. Localization of input functions from the difference between baseline and hypoxic images. (A) Representative individual arterial input function (AIF) extracted from middle cerebral artery (MCA). (B) Representative individual venous output function (VOF) extracted from superior sagittal sinus (SSS). (C) Representative AIF, eAIF with venous scaling and VOF during hypoxia. (D) Motion-corrected simple difference from baseline signal at different time points during the hypoxia paradigm. The signals in the MCA and SSS were most visible at the peak of hypoxia. 53 After fitting CBF × R(t) with SVD, CBF is determined as the initial height of this tissue response function. Finally, MTT is calculated as a direct ratio between regional CBV and CBF: = %AB %A> [7] 2.2.2. Study protocol: The Committee on Clinical Investigation at Children’s Hospital Los Angeles (CHLA) approved the protocol; written informed consent and/or assent were obtained from all subjects (CCI#2011-0083). This study was performed in accordance with the Declaration of Helsinki. A total of 66 subjects were tested between April 2012 and December 2017. This cohort was divided into two patient groups: 28 healthy controls and 38 chronically anemic patients (including sickle cell anemia, thalassemia and other anemia syndromes). Exclusion criteria were prior neurologic insult, pregnancy, acute chest or pain crisis hospitalization within one month and major medical problems outside of their chronic anemia. Only subjects older than 12 years of age were included in the study. Imaging, vital signs and blood samples were obtained on the same day for each subject. Complete blood count was analyzed in our clinical laboratory. Demographic and clinical variables of each patient group are summarized in Table 2.1. 54 Table 2.1. Patient demographic and baseline flow data. CBF was assessed using phase contrast (PC) and arterial spin labeling (ASL) in the grey matter (GM) and white matter (WM). Demographics reflects the total cohort of 66 subjects in the PC comparative analysis and the reduced cohort of 41 subjects in the ASL analysis. Student’s t-test was used to compare values between two groups. Values are reported as mean ± standard deviation. Bold letterings indicate statistical significance (p<0.05). Phase contrast analysis Control (N=28) Anemia (N=38) p-value Age (Years) 26.1±10.8 21.6±8.4 0.06 Sex 7M, 21F 17M, 21F 0.10 Hemoglobin (g/dL) 13.4±1.2 10.1±1.8 <0.01 Hematocrit (%) 39.7±3.4 29.5±5.0 <0.01 CBFPC (mL/100g/min) 60.7±9.1 91.1±22.3 <0.01 ASL analysis Control (N=16) Anemia (N=25) p-value Age (Years) 23.4±9.1 22.4±8.9 0.74 Sex 5M, 11F 10M, 15F 0.58 Hemoglobin (g/dL) 13.4±1.2 10.1±2.2 <0.01 Hematocrit (%) 40.1±3.8 29.8±5.9 <0.01 CBFASL GM (mL/100g/min) 60.9±11.4 77.3±30.0 0.02 CBFASL WM (mL/100g/min) 37.5±10.9 42.1±10.9 0.20 2.2.3. Transient hypoxia: Figure 2.1 illustrates the experimental setup for the respiratory challenge MRI. At the start of the image acquisition, patients were breathing through a custom, two-liter reservoir rebreathing circuit supplied by pressurized, non-humidified room air (21% oxygen, balanced nitrogen) at 12 liters per minute. This system included one-way valves to prevent partial gas mixtures and respiratory bellows (Invivo Corporation, Gainesville, FL) to display the breathing pattern and frequency. At 50 seconds into the data acquisition, the room air gas mixture was switched to 100% nitrogen until the patient had completed 5 breaths (approximately 25 seconds, counted through the respiratory bellows data display), then the circuit was changed back to room air. This 55 previously-developed protocol 405 caused a safe and short desaturation; typical desaturations had a minimum SpO2 of 75-80%. The pulse oximetry measurements during transient hypoxia (Figure 2.2A) demonstrated similar depth and duration to spontaneous desaturations (Figure 2.2B). More details on this protocol are provided in Supporting Information Methods section. 2.2.4. Magnetic resonance imaging acquisition: Each participant underwent an MRI study using a 3T Philips Achieva using an 8-element phased-array coil. Anatomical 3D T1, BOLD, phase contrast and arterial spin labeling scans were performed in all subjects prior to transient hypoxia. For each subject, a 3D T1-weighted image was acquired with TR = 8.20ms, TE = 3.8ms, flip angle = 8°, resolution = 1×1×1mm and FOV = 256×224×160mm. BOLD-MR images were acquired with the following parameters: TR = 2000ms, TE = 50ms, flip angle = 90°, resolution = 2.5×2.5×5mm and FOV = 220×220×130m. A total of 150 volumes were collected. Phase contrast images were obtained, positioned just above the carotid bifurcation with the following parameters: TR = 12.9ms, TE = 7.7ms, flip angle = 10°, resolution = 1.3×1.3×5mm, FOV = 240×261×5mm and velocity encoding gradient of 100cm/s. Pseudo continuous ASL scans were performed using an unbalanced Hanning shaped RF pulse labeling train (mean gradient 1G/cm, interpulse interval of 1ms, pulse duration 0.5ms) and a 3D GRASE two-shot readout scheme with the following parameters: TR = 3800ms, TE = 9.8ms, resolution = 3.7×3.7×10mm, labeling duration = 2000ms, post-labeling delay = 1600ms, dynamics = 10 and EPI factor = 5. Two inversion pulses for background suppression were used, with an inversion efficiency of 95% each. All MR acquisition parameters have been detailed in prior publications. 395,406 56 2.2.5. Image pre-processing: The phase contrast images were analyzed using an in-house MATLAB program (Mathworks, Natick, MA). Stationary tissue pixels were identified in the complex difference images by simple thresholding using a mean plus two standard deviations from a remote, nonvascular region. Phase differences of stationary pixels were fit to a second-order two-dimensional polynomial and used to remove the background phase from vessels. Vessel boundaries were identified semi-automatically from the complex difference image processed using a Canny edge- detector. A single-voxel dilation of vessel boundaries was then performed, and only voxels whose complex difference was greater than the stationary tissue threshold was retained. Finally, CBF (mL/100g/min) was computed with CBF=∑ A i ×V i i , where Ai is the area and Vi is the velocity defined by PC-MRI of each voxel. Total CBF, which was the sum of the flow in the left and right internal carotid arteries and vertebral arteries, was corrected for total grey and white matter volume assuming a brain density of 1.05g/mL. The ASL images were rigidly registered to T1-weighted images and then to Montreal Neurological Institute (MNI) template using FMRIB Software Library (FSL). 407 Regional CBF quantification was performed using a two-compartment kinetic model 395 and estimates of labeling efficiency, disease-specific blood T1 and tissue-specific arterial transit times. A comprehensive discussion of these parameters was detailed in our previous publications. 395,408 The BOLD images were preprocessed with FSL using a standard spatial functional pipeline. Images were first slice-time corrected, realigned to remove physiological motion and then co-registered to the MNI template space. Finally, the registered volumes were normalized and smoothed using 8×8×8mm Gaussian kernel similar to typical pre-processing for BOLD fMRI datasets. 409,410 More details of this preprocessing pipeline are described in previous publication. 406 Percentage change of the MR signal ΔBOLD is calculated as: ∆ = A&23 ?@ABC'# 6A&23 %#(*D'E* A&23 %#(*D'E* ×100% (Figure 2.2C) 57 where BOLDhypoxia was the nadir of the hypoxic time curve and BOLDbaseline was the average of volumes [5, 25] out of 150 temporal volumes. To minimize the partial volume effects, the AIF was rescaled by the time integral of the VOF measured at the superior sagittal sinus 402,411 (refer to Methods section 2.1) since the superior sagittal sinus is a large, long and straight lumen parallel to the main magnetic field and less vulnerable to partial volume errors. The AIF and VOF were extracted from BOLD images that had been motion-corrected but neither smoothed nor registered to the MRI template. Localizations of the middle cerebral artery and the superior sagittal sinus were done using a previously published technique on a simple difference dataset of: Q+ee(.()*( = 0!:(/+)( − fgh,i+! (Figure 2.3) 412 The transient hypoxia model was approximated with a linear function during the rise time of the desaturation bolus and an exponential decay function during the signal recovery; the pre- hypoxia and post-hypoxia baselines were modeled as constants (Figure 2.2D). To avoid the effects of recirculation and hyperemia as confounders, the end-point of the exponential decay was determined by minimizing the least-square error, and the area under the curve was only computed under the linear and exponential portions using the Riemann sum integration approach. From these datasets, perfusion maps for CBF, CBV and MTT maps were computed using an in-house MATLAB program (Mathworks, Natick, MA). Grey matter and white matter perfusion values were computed as an average within tissue-specific masks derived from an average of 152 T1-weighted MRI scans in the common MNI coordinate system. 413 White matter masks were eroded by two voxels to minimize partial volume effects. Since cerebral perfusion measurements have a bimodal distribution, values were reported for grey matter and white matter separately. Masks of the anterior cerebral artery (ACA), middle cerebral artery (MCA) and posterior cerebral artery (PCA) circulations in the grey matter were based on the published templates of vascular territories in both hemispheres. 414 The territories were created from anatomic studies of cerebral vascularization and evaluated on the bicommissural plane. 58 2.2.6. Statistical analysis: Statistical analysis on global desaturation values was performed in JMP (SAS, Cary, NC). Student’s t-test was used to examine the difference in clinical variables between the two study groups. Paired t-tests were used to compare CBF computed using dDSC, ASL and phase contrast. Correlation tests and Bland-Altman analyses were performed to assess global and regional agreement between different flow measures. Adherence to the a priori transient hypoxia model of linear rise and exponential decay was calculated with the root-mean-square percentage error (RMSPE) as: = + − + + ) +j' " M ×100% where n is the total number of sample points, + is the actual signal and + is the predicted signal. 2.3. Results All 66 subjects completed T1, BOLD, phase contrast and ASL imaging. When this study first started, we were initially unable to perform ASL compatible with the White Paper 194 until October 2014, thus comparative ASL images were available in only 45 out of 66 subjects. Additionally, four additional ASL datasets were excluded due to motion, leaving 16 control and 25 anemic subjects in the ASL analysis (Table 2.1). There were no differences in the demographic and hematologic data between the complete cohort and the cohort after subject exclusion in the ASL analysis. 2.3.1. Quality of AIF and VOF vessels: Localizations of the AIF from the middle cerebral artery and the VOF from the superior sagittal sinus are shown in a representative subject in Figure 2.3. Time integral of the normalized 59 venous signal was 140% stronger in magnitude compared to the raw AIF signal in all subjects. Scaling of the AIF by the VOF yielded an average CBF of 42.4±18.6 mL/100g/min and an average CBV of 4.8±1.3 mL/100g in healthy controls within the acceptable range in literature. 2.3.2. Perfusion measured by deoxygenation-based DSC: None of the patients consciously perceived the hypoxia episode and no complications were encountered. This gas paradigm produced an average drop of 16.5±9.0% in SpO2 and of 4.3±1.7% and 7.9±2.8% in BOLD signal in the white matter and grey matter respectively compared to baseline. Table 2.2 summarizes average perfusion values in the grey matter and white matter separately for two patient groups, with anemic patients showing significantly higher CBF and CBV compared to healthy controls. Moreover, anemic patients also had slightly faster transit time compared to healthy patients. Figure 2.4 shows a trend of agreement between CBF values derived from dDSC, phase contrast and ASL. There is a systemic underestimation of 20.0 mL/100g/min by dDSC compared to phase contrast with large limits of agreement, and a smaller bias of 12.0 mL/100g/min compared to ASL in the Bland-Altman analysis. MTT was also well-correlated with CBF in hypoxia-based and phase contrast methods (r 2 =0.25, p<0.01 and r 2 =0.13, p=0.03 respectively) but only trended for ASL (r 2 =0.08, p=0.08). Higher perfusion values were negatively correlated with age (r 2 =0.11, p=0.01) but not with sex (p=0.77). Additionally, lower hemoglobin levels were associated with higher CBF and CBV (r 2 =0.29, p<0.01 and r 2 =0.17, p<0.01 respectively) and with lower MTT (r 2 =0.11, p<0.01). 60 Table 2.2. Grey matter (GM), white matter (WM) and GM-WM ratio group average perfusion parameters. Students’ t-test was used to compare values between two groups. Values are reported as mean ± standard deviation. Bold letterings indicate statistical significance (p<0.05). Control Anemia p-value CBFdDSC GM (mL/100g/min) 41.3±17.5 70.8±24.9 <0.01 CBFdDSC WM (mL/100g/min) 28.1±10.9 45.5±17.9 <0.01 CBFdDSC GM/WM Ratio 1.5±0.2 1.6±0.1 0.01 CBVdDSC GM (mL/100g) 4.6±1.5 6.2±1.4 <0.01 CBVdDSC WM (mL/100g) 3.5±1.5 4.7±1.4 <0.01 CBVdDSC GM/WM Ratio 1.4±0.2 1.4±0.2 0.74 MTTdDSC GM (seconds) 8.5±1.9 7.5±1.9 0.04 MTTdDSC WM (seconds) 8.4±2.0 7.3±1.7 0.02 MTTdDSC GM/WM Ratio 1.0±0.1 1.0±0.1 0.16 61 In addition to global analysis, voxel-wise perfusion maps were also generated for the hypoxia-based perfusion method (Figure 2.5). RMSPE maps in Figure 2.5D demonstrated strong adherence to the a priori hypoxia model of linear rise and exponential decay, demonstrating robust desaturation observable in voxels across the brain. In addition to group average perfusion maps, examples of individual perfusion maps were also illustrated for two control and two anemic Figure 2.4. Agreement between CBFdDSC and alternative flow methods. (A) Correlation and (B) Bland-Altman analyses between CBFdDSC and phase contrast flow. (C) Correlation and (D) Bland-Altman analyses between CBFdDSC and ASL blood flow in the grey matter (GM). (E) Correlation and (F) Bland-Altman analyses between CBFdDSC and ASL blood flow in the white matter (WM). 62 subjects in Supporting Information Figure 2.S1. Test-retest was performed in two healthy control subjects, with individual values shown in Supporting Information Table 2.S1 and individual maps in Supporting Information Figure 2.S2. Regional analysis of perfusion within three arterial territories is detailed in Table 2.3. Even though the middle cerebral arteries’ territory showed a trend of higher blood flow and faster transit time compared to the anterior and posterior cerebral arteries, one-way ANOVA tests demonstrated no significant difference between perfusion in the three territories (p=0.54 for CBF, p=0.07 for CBV, p=0.83 for MTT). CBF and CBV maps also showed significantly higher values in the grey matter compared to the white matter (p<0.01), with the average grey–white matter ratio of 1.4 for CBV and 1.5 for CBF – lower than the grey–white ratio of 1.9 in ASL maps. Similarly, MTT map demonstrated grey–white matter differentiation (p=0.01), but the distinction was not as strong and more confined to the deep white matter tissue. Figure 2.5. Group average perfusion and fit evaluation maps. (A) Cerebral blood flow (CBF), (B) cerebral blood volume (CBV), (C) mean transit time (MTT) and (D) root- mean-square percentage error (RMSPE) maps derived from the dDSC protocol. RMSPE maps showed adherence to the a priori hypoxia model of linear rise and exponential decay. 63 Table 2.3. Group average dDSC perfusion parameters in different flow territories. Flow territories include territories perfused by the anterior cerebral arteries (ACA), middle cerebral arteries (MCA) and posterior cerebral arteries (PCA) in the grey matter. Values are reported as mean ± standard deviation. Control Anemia CBF (mL/100g/min) CBV (mL/100g) MTT (s) CBF (mL/100g/min) CBV (mL/100g) MTT (s) ACA 46.5±19.9 4.6±1.3 8.6±2.4 75.3±24.6 6.3±1.5 6.8±1.9 MCA 47.4±20.0 5.4±1.5 8.0±2.2 82.6±29.0 7.0±1.6 6.9±2.2 PCA 42.9±20.1 4.8±1.7 8.4±2.7 77.1±26.2 6.6±1.6 7.0±2.3 Furthermore, regional correlation analysis between CBFdDSC and CBFASL (Figures 2.4C and 2.4E) demonstrated a good association within the grey matter but not in the white matter. Even though imprecise registration prevented voxel-wise comparison between the two approaches, the ACA, MCA and PCA vascular territories within the grey matter all demonstrated reasonable agreement between CBF estimates by dDSC and ASL techniques (Figures 2.6A, 2.6C and 2.6E). There is no significant bias in perfusion measurements by dDSC compared to ASL in different vascular territories of the brain in the Bland-Altman analyses (Figures 2.6B, 2.6D and 2.6F). 64 2.4. Discussion Our hypoxia protocol induced a reproducible drop in MR signal in both control and anemic patients similar to the mechanism of gadolinium-based DSC. However, unlike DSC, the signal loss in dDSC was smaller in magnitude and induced by increased paramagnetic deoxygenated Figure 2.6. Regional agreement between grey matter dDSC and ASL flow methods. Correlation and Bland-Altman analyses between grey matter CBFdDSC and ASL blood flow in the (A, B) anterior cerebral artery territory (ACA), (C, D) middle cerebral artery territory (MCA) and (E, F) posterior cerebral artery territory (PCA). 65 hemoglobin, leading to shorter T2* and lower gradient-echo signal. While other gas paradigms such as 100% oxygen had been used to investigate baseline perfusion in healthy subjects, 205 this work represented the first study to use transient hypoxia as a contrast method for CBF quantification. It is also the first to show a reasonable trend of agreement between dDSC perfusion values and independent flow techniques like ASL and phase contrast. Additionally, the physiologic associations between dDSC with age and hemoglobin were expected and provided an extra layer of reassurance for our data quality. This novel dDSC technique offers many advantages over conventional gadolinium-based DSC. In eliminating the need for an exogenous contrast medium, this method offers an alternative contrast that is well-tolerated, repeatable, does not lead to anaphylaxis and is convenient in patients whose venous access is more challenging, such as in pediatric or anemic subjects. This method is significantly cheaper compared to gadolinium as well as iron-based contrast like ferumoxytol. It is also potentially easily translatable to clinical protocols, especially for perfusion quantification in patients under anesthesia. Apart from brain imaging, since previous work has demonstrated good performance of BOLD imaging and different respiratory challenges in the liver, kidney and heart, 415–418 this dDSC technique can potentially be applied to assess in these organs where non-contrast imaging modalities like ASL have proven less effective due to low signal-to-noise ratio. 419 Most importantly, this hypoxia gas paradigm is very safe and well-tolerated; our laboratory has performed this experiment on over 200 subjects in a larger study on sickle cell anemia without any major adverse events such as stroke or transient ischemic attack. 405,420 In a review on the safety of hypoxia challenges, Bickler et al. showed that hypoxia experiments have been performed at many different research centers across the world with arterial saturations dropping as low as 45%, which is considerably lower than the typical hypoxia levels experienced in our study. 421 Overall, these brief exposures to hypoxia have been demonstrated to be well-tolerated 66 without any evidence of cardiovascular compromise, systemic acidosis or lasting cognitive impairment. 421 Absolute quantification of CBF is difficult due to the large partial volume effects in the AIF related to small vessel sizes and limited spatial resolution. 422 While many multiplicative rescaling methods for AIF correction have been explored, 423 rescaling with a VOF from the superior sagittal sinus has proven robust due to the large vessel caliber of the superior sagittal sinus as well as its position parallel to the magnetic field and orthogonal to the imaging plane. 411,424 However, even with the use of venous rescaling, our CBV measurement was still overestimated compared to normal literature values. Additionally, even though phase contrast has been shown to overestimate total blood flow, 425 the large bias between CBF measured by dDSC and phase contrast might be due to the low quality of the AIF. AIF which were contaminated with nearby tissue voxels was lower and broader in shape, leading to lower CBF values. 426 Partial volume effects were still present in the superior sagittal sinus, leading to lower area-under-curve of the VOF and AIF and consequently to overestimation of tissue CBV. Heterogenous and low quality of AIF was the potential cause for large limits of agreement of dDSC compared to phase contrast and ASL. An overestimation of CBV and underestimation of CBF led to a systematic overestimation of MTT in this study. Future work to employ alternative and more effective methods to extract the AIF and VOF will be necessary to improve the accuracy of CBV and CBF measurements in dDSC. In general, CBFdDSC showed reasonable inter-modality agreement with phase contrast and ASL using correlation and Bland-Altman analyses. The stronger correlation between CBFdDSC with phase contrast compared to ASL was understandable since phase contrast is independent of T1 and T2 relaxations and has relatively high signal-to-noise ratios. In addition to the agreement in whole-brain blood flow, our dDSC-based CBF measures also showed regional agreement with ASL perfusion maps. Grey matter as well as the three vascular territories all displayed more robust inter-modality association between CBFdDSC and CBFASL compared to deep white matter regions, 67 suggesting a divergence in the sensitivities of these two methods to microvascular perfusion characteristics in the white matter. Our observation of significantly higher blood flow and volume in anemic patients compared to healthy controls was consistent with previous works showing increased baseline perfusion in compensation for compromised oxygen carrying capacity in chronic anemia. 396,397 The faster MTT in anemic patients was also in accordance with the shorter vascular transit times measured with multi-post-labeling delay ASL. 427 Shorter MTT in anemic subjects was apparent in the grey matter as well as regions of normally perfused white matter; however, these subjects also displayed abnormally slower contrast dynamics in the deep white matter at the end of the perfusion branches. These borderzone regions coincided with typical watershed areas of flow limitation and peak oxygen extraction 395,428 and were consistent with previous studies that showed reduced flow reserve in areas vulnerable to silent strokes, 429 suggesting a connection between hemodynamic impairment and the development of strokes in anemic patients. Even though ASL is the prominent non-invasive perfusion technique and has been the focus of much research and development in the MRI community, ASL is still hampered by some disadvantages compared to dDSC. Most notably ASL has low signal-to-noise ratio in the white matter, possibly contribution to the lack of agreement between white matter CBF measured by the two techniques. Additionally, ASL is confounded by model assumptions and physical parameters including blood T1 and arterial transit time, many of which are different in anemic patients compared to healthy controls. 408 Blood T1 can be measured with T1-mapping, 430,431 but measurement of arterial T1 remains challenging and not universally available. Arterial transit time confounds can also be avoided by utilizing multiple post-labeling delay ASL, 194 but there is currently too much heterogeneity in the acquisition techniques. Additionally, simple multi-delay ASL can take significantly longer than single-delay ASL. Acceleration methods such as Hadamard encoding or Look-Locker readouts can be used to shorten acquisition time, but these are typically unavailable on routine clinical scanners. Therefore, even though ASL will undoubtedly have an 68 important place in the future of perfusion imaging, the development of this novel dDSC technique provides an alternative and generalizable method for measuring tissue blood flow and builds upon the previous work that has already been pioneered by the perfusion MRI community. An important assumption in our dDSC model was the linearity between cerebral oxygen level and the changes in MRI signal DR2 * during the hypoxia gas paradigm in equation [1]. Calibrated BOLD model typically presents a non-linear relationship between deoxygenated hemoglobin and changes in the BOLD signal with a vessel size-dependent parameter β = 1.3 at 3T. 432 However, under hypoxic conditions, in addition to capillaries and venules, larger deoxygenated arterioles also contribute to the BOLD signal, in which case β is likely decreased closer to 1 and the relationship between the BOLD signal and oxygen saturation could be approximated with a linear function. Furthermore, previous work by Rostrup et al. has demonstrated an approximately linear relationship between DR2 * and oxygen saturation during hypoxic exposure, thus confirming that the linearity assumption was reasonable. 262 Future work to validate this linear assumption in normal tissue as well as diseased tissue is warranted to confirm and improve the dDSC diagnostic capability. In addition to validation of the linear assumption, other confounds to the BOLD signal, including the effects of vessel size, orientation, water diffusivity and the influence of hematocrit on BOLD signal variations, 433 need to be explored. The practicality of assessing the impact of these parameters in vivo is limited, but in silico validation using Monte Carlo simulation can explore the relationship between these parameters and the dDSC signal in future work. This study has some notable limitations. Firstly, several confounds in the DSC model potentially affected the perfusion measurements in our dDSC technique, including delay and dispersion of bolus in the AIF. These confounds typically introduced errors in cases of severe vasculopathy, so they likely did not affect our cohort; however, application of dDSC to subjects with vascular disease will require technical improvement to address these issues, such as by 69 modeling the vascular bed, extracting local AIF or using a delay-insensitive SVD-variant for perfusion quantification. 434–436 Secondly, these perfusion measurements showed large variability compared to ASL. However, previous comparison studies between ASL and gadolinium-based DSC have also demonstrated large variability. 437–439 And since this is a proof-of-concept study, our results demonstrated that the novel dDSC technique has the potential to measure blood flow, and that these measurements do vary proportionally with other conventional flow measurements like phase contrast and ASL. Further optimization will be necessary to improve the quantitation and control for potential confounders such as breathing rate, minute ventilation and intra-subject variability in hypoxic exposure. Gaussian smoothing for noise reduction the BOLD datasets led to underestimated grey–white matter ratio in both CBF and CBV maps. Future work to optimize the hypoxic bolus depth and duration and use repeated hypoxic stimuli to improve signal-to-noise can require less spatial smoothing and increase grey–white matter perfusion differentiation. Even though repeatability tests showed reasonable performance on two healthy subjects (Supporting Information Figure 2.S2), test-retest on a larger subset of subjects needs to be conducted for this methodology to better validate the diagnostic value of dDSC. Another important limitation to our study is that a hypoxic stimulus may impact CBF in an unpredictable manner. Prolonged hypoxic exposure increases CBF, 440 but also causes hyperventilation that tends to counteract the hypoxic vasodilation. 440,441 Whether a challenge this brief raises or lowers CBF during the observation interval has not been determined since the transient duration of the hypoxia paradigm prevented measurement of CBF at steady-state hypoxia. A more in-depth and thorough investigation into stimulus-mediated changes in CBF is required to assess their contribution to our measurement uncertainty. Dual-acquisition of ASL and BOLD can be used to capture the dynamic BOLD signal as well as regional CBF changes across the brain during the hypoxia challenge; 4D-flow MRI can also be used to assess vasodilation in of the major arteries in the Circle of Willis in response to the hypoxic stimulus. Additionally, this 70 experiment was originally designed with end-tidal O2 and CO2 acquisitions (Biopac Systems, Goleta, CA). However, during the course of the study, the gas sampling sensors broke, and we were not able to retrieve the end-tidal measurements. For future gas experiments, the use of a controlled respirator with prospective targeting of pO2 and end-tidal CO2 will be helpful to eliminate hypocapnia as a potential confounder. 442 In conclusion, in this proof-of-concept study, we have demonstrated the feasibility of using transient hypoxia to generate a contrast bolus that mimics the effect of gadolinium and yields reasonable blood flow measurements. Even though our current dDSC implementation suffers from biases in perfusion estimates and requires further validation of its assumptions, dDSC still offers a novel gadolinium-free approach which will be attractive in pediatric and end-stage renal disease patients. Additionally, dDSC can be potentially beneficial for patients in whom serial DSC studies are required (such as in brain tumors), but this application awaits further investigation into the feasibility of dDSC in measuring impaired blood-brain barrier function. To improve the diagnostic utility of this method, block-design gas paradigms 443 and transfer function analysis 272 can be applied, and hypoxic stimuli can also be used in an interleaved manner with CO2 inhalation to jointly estimate resting perfusion and cerebrovascular flow reserve. Since the AIF can easily be calculated in the heart and aorta, future studies can also extend this cerebral perfusion technique to blood flow measurements in other thoracic and abdominal organs. 2.5. Supplemental Information 2.5.1. Supplemental Methods Informed consent and assent Consent and assent forms from this study can be found on our public Github respository (https://github.com/jwoodchla/DSC). Particularly, the subjects were informed about the process of the hypoxia challenge and the associated risks: 71 “You will be asked to: lay still in the MRI machine while we look inside of your body. When you are in the MRI machine you will be asked to do is a special breathing test. We will give you a mouthpiece to breathe through and a clip that gently pinches your nose so that you have to breathe through a mouthpiece. We will ask you to breathe normally like you were sleeping.” “You might feel uncomfortable when breathing through the mouthpiece because your mouth may feel dry. While you are doing the breathing tests you may feel a little dizzy or like you have to breathe a little faster. You may also feel more alert or excited when the oxygen levels are increased. These changes in the oxygen levels will not hurt you and most people can’t tell a difference.” Pre-scan protocol Arterial saturation was measured in all subjects outside of the magnet. The hypoxia protocol is only performed in subjects who had baseline arterial saturation over 95%. During pre-scan briefing, the subjects were shown the gas tanks and the breathing circuit used in the hypoxia paradigm. The experimental setup (as shown in Figure 1) was demonstrated to the subjects. The subjects were asked to try on the nose-clip and breathing mask, which comprised of a scuba mouthpiece coupled via a three-way to the breathing reservoir. Additionally, the subjects are shown how to spit out the mouthpiece in case they felt any discomfort during the process of the gas paradigm. The subjects are then positioned in the scanner with the breathing mask and nose-clip on, with room air gas flowing continuously. The technician attached the following devices to the subjects and monitored the corresponding recordings: (1) fingertip pulse oximetry, (2) respiratory bellows, (3) ECG, (4) near-infrared spectroscopy, (5) end-tidal O2 and CO2 monitor. 72 Hypoxia protocol During the gas paradigm, at least a technician and a licensed physician are required to operate the gas switches and monitor the physiological signals. The technician operates the sequence of gas switch from room air to hypoxia as following: (1) turn off room air gas, (2) wait for the flowmeter to show zero flow (approximately 1-2 seconds), (3) turn on hypoxic gas, (4) ensure the flowmeter shows positive flow and the breathing reservoir is inflated through the monitor camera. The reverse process is implemented for the switch from hypoxia to room air gas. Note that since gas is flowing at a high rate of 12 liters per minute, only one gas should be flowing at a time. Concurrently, the physician monitors the subjects’ physiological signals, including heart rate, ECG, breathing pattern and frequency, arterial saturation and tissue oxygenation. Note that the protocol of 5 breaths of nitrogen gas is flexible and subject-specific; if the physician determines that the subject’s breathing pattern is shallow, the subject will be allowed to breathe hypoxic gas for more than 5 breaths up to 25 seconds. Conversely, if the subject was taking long deep breaths, the protocol will be stopped at 20 seconds to ensure safety. In case of an emergency, room air will immediately be switched on, scanning will be halted, and a code blue will be called. Post-scan protocol The subjects were later debriefed by the study coordinator. The subjects were followed up after 1 hour, 12 hours, 24 hours and a week after the experiment for adverse events, including headache, lightheadedness, shortness of breath, stroke, cough, fever, nausea, diarrhea, and in subjects with sickle cell disease, vaso-occlusive crisis and acute chest syndrome. 73 Study notes The investigators believe that caution must be exercised when using the 100% nitrogen gas. Even though we reported no adverse events, we monitored our subjects very closely during the course of the challenge. The following recommendations are made through our experience with the hypoxia challenge: a. Real-time pulse oximetry: Fingertip pulse oximetry cannot act as the sole physiological signal for monitoring since the pulse oximetry signal lags the cerebral desaturation by 15-20 seconds. Concurrent measurement of oxygenation level, such as near-infrared spectroscopy in this study, must be used to monitor the cerebral changes. b. Alternative hypoxic gas paradigm: For investigators who did not previously have experience working with hypoxia challenges, we recommend the use of a non-lethal hypoxic gas mixture (such as 10% oxygen and balanced nitrogen). Even though the challenge will no longer be transient (it takes at least 5-10 minutes to desaturate to 90% saturation) and might lead to less accurate results, this will ensure the safety of the participants. c. Controlled gas systems: Since this paradigm is a complicated process, we recommend the use of a commercial gas system specific for MR imaging. Notably, preliminary work using a prospective end-tidal targeting system such as the RespirAct (Thornhill Medical, Toronto, Canada) has shown promising results using non-lethal hypoxic mixture. This will be manifold safer and easier to control and standardize the hypoxic dosage to individual patient. 74 Supporting Information Figure 2.S1. Examples of individual perfusion maps in two representative control subjects and two representative anemic subjects. (A) Cerebral blood flow by dDSC (CBFdDSC), (B) CBF by ASL (CBFASL) (C) cerebral blood volume (CBVdDSC) and (D) mean transit time (MTTdDSC). The ASL images were acquired at resolution = 3.7×3.7×10mm and then nonlinearly registered onto MNI template, so nonlinear warping could be seen on the sagittal plane. 75 Supporting Information Figure 2.S2. Two repeated measurements of CBF, CBV and MTT maps on two healthy controls. Supporting Information Table 2.S1. Two repetitions of whole-brain perfusion measurements in two healthy controls. CBF (mL/100g/min) CBV (mL/100g) MTT (seconds) Control 1 Repeat 1 57.7 5.8 5.1 Repeat 2 68.7 5.9 6.0 Control 2 Repeat 1 44.2 6.1 8.7 Repeat 2 37.4 4.9 8.1 76 Chapter 3 : Oxygen Respiratory Challenges for Deoxygenation-based Dynamic Susceptibility Contrast 3.1. Introduction Perfusion weighted MRI is frequently used in clinical diagnoses of various neurological conditions, including cerebral infarctions or gliomas. 206,444 Dynamic Susceptibility Contrast (DSC) MRI is a popular perfusion MRI technique due to its fast acquisition, its high signal-to-noise ratio (SNR), its capability to differentiate pathologic tissues from normal appearing tissues 445,446 as well as its value in evaluation of the treatment course. 447,448 Despite its clinical utility, DSC is contraindicated in renal-impaired patients who are at a higher risk for gadolinium-induced nephrogenic systemic fibrosis. 449 To avoid the use of gadolinium, alternative contrast-free perfusion MRI techniques have been developed, including Arterial Spin Labeling (ASL) and Phase Contrast (PC). However, ASL suffers from low signal as well as difficulty in quantitation in diseased populations, 194 whereas PC only yields total blood flow instead of a spatial distribution of cerebral perfusion. 183 To address the gap in contrast-free perfusion, previous works have proposed deoxygenation-based DSC (dDSC) which delivers transient exposure of low oxygen (hypoxia) or high oxygen (hyperoxia) through respiratory challenges to modulate the intravascular concentration of deoxygenated hemoglobin as a source of endogenous contrast. 203–205 Deoxyhemoglobin is paramagnetic, resulting in MRI signal losses similar to the effects of gadolinium. Therefore, modulation of inspired oxygen levels yields patterns of gradient-echo or spin-echo signal loss similar to traditional DSC, allowing calculation of cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT) using tracer kinetic modeling. 77 This work aims to investigate deoxygenation-based perfusion measurements using several bolus and non-bolus respiratory challenges. Bolus challenges include Desaturation paradigm, which consists of a transient exposure of hypoxia during normoxia (Figure 3.1A-D), and Resaturation, which consists of a transient exposure of hyperoxia during a prolonged hypoxic baseline (Figure 3.1E-H). Additionally, this work proposes a novel respiratory paradigm in which inspired oxygen level is modulated to yield sinusoidal fluctuation of gradient-echo MRI signal (SineO2, Figure 3.1I-L). All three respiratory challenges were acquired to estimate CBF, CBV, and transit time in healthy volunteers and were compared against existing perfusion MRI techniques, including gadolinium-based DSC, ASL, and PC. 78 3.2. Methods 3.2.1. Study protocol The Committee on Clinical Investigation at Children's Hospital Los Angeles approved the protocol; written informed consent was obtained from all subjects (CCI#20-00050). This study was performed in accordance with the Declaration of Helsinki. A total of 10 healthy volunteers participated in this study between April and May of 2021. Exclusion criteria included pregnancy, hypertension, diabetes, stroke or other known neurologic insult, seizures, known developmental delay or learning disability, at least one ‘yes’ answer to the Figure 3.1. Respiratory challenge patterns for Desaturation, Resaturation, and SineO2. End-tidal oxygen (EtO2), end-tidal carbon dioxide (EtCO2), fingertip pulse oximetry (SpO2), and changes in R2* (∆R2*) respectively for (A-D) Desaturation, (E-H) Resaturation, and (I-L) SineO2. Grey shading reflects 95% confidence interval. 79 6-question Choyke survey, 450,451 and measured glomerular filtration rate (GFR) lower than 60 mL/min/1.73mm 2 . Imaging, vital signs, and blood samples were obtained on the same day for each subject. 3.2.2. Respiratory challenges Respiratory challenges were performed using a specialized computer-controlled gas blender (RespirAct, Thornhill Research, Toronto, Canada) that prospectively targets and manipulates end-tidal O2 (EtO2) and end-tidal CO2 (EtCO2) partial pressures. 432,452 Three respiratory patterns were investigated, in which EtO2 levels were modulated and EtCO2 were clamped at subject-specific baseline values. Fingertip pulse oximetry SpO2 (Nonin, Plymouth, MN) was recorded during each challenge. The order of respiratory challenges was randomized. a. Desaturation: 2.6 minutes long for each iteration, made up of [1-minute normoxia; 35- second hypoxia; 1-minute normoxia] (Figure 3.1A). Targeted EtO2 values for normoxia and hypoxia were 120 mmHg and 35 mmHg respectively. Two iterations of Desaturation were performed back-to-back to investigate reproducibility (4.2 minutes total, with overlapping post-hypoxia in iteration #1 and pre-hypoxia in iteration #2 for time efficiency). b. Resaturation: 2.2 minutes long for each iteration, consisting of [1-minute hypoxia; 10- second normoxia; 1-minute hypoxia] (Figure 3.1E). Targeted EtO2 for hypoxia was 40 mmHg and for normoxia was 100 mmHg. Two iterations of Resaturation were performed consecutively (3.3 minutes total, with overlapping periods of post-normoxia for iteration #1 and pre-normoxia for iteration #2). 80 c. SineO2: 4.3 minutes long. To generate 2 cycles of sinusoidal fluctuation in saturation between 98% and 85%, the Hill equation was used to calculate the partial pressure of oxygen to account for the non-linearity in the hemoglobin dissociation curve: = (h& " ) E (hk8) E \ (h& " ) E , where 50 is the " at 50% saturation estimated as 26.4 mmHg and is the Hill coefficient estimated as 2.8. 453 This equation is approximated using two linear ramps and a half-wave sinusoid between EtO2 of 105 and 50 mmHg to yield the pattern in Figure 3.1L. The piecewise equation also accounted for the longer decay time from normoxia to hypoxia and faster recovery time from hypoxia to normoxia. Further details on the generation of the EtO 2 pattern for this paradigm are provided in Supplemental Methods M1. Post-study survey was performed in which subjects were asked to rate each respiratory challenge as one of the 4 options: no discomfort, mild discomfort, moderate discomfort, and extreme discomfort. Subjects also rated the discomfort of wearing the respiratory mask on the same scale. 3.2.3. MRI acquisitions All imaging was acquired on a 3T Philips Achieva (Philips Medical Systems, Best, Netherlands) with a 32-channel head-coil. Details regarding MRI acquisition and pre-processing for structural MRI, deoxygenation-based DSC MRI, and other reference perfusion MRI sequences can be found in Supplemental Methods M2. An additional segmentation step was performed to divide the brain into 300 regions-of-interest (ROI) using the USCBrain atlas 454 (Supplemental Methods M2). 81 3.2.4. Perfusion quantification All dDSC processing was performed in MATLAB (MathWorks, Natick, MA). Suppression of signal contribution from pial veins was performed by eliminating voxels with higher signal amplitude than the 98th percentile in each subject. Calculation of whole brain, GM, and WM perfusion values was performed by averaging voxels within brain and tissue-specific masks in each functional sequence’s native space. a. Gradient-echo ∆ " ∗ : Single echo ∆ " ∗ was calculated from TE = 35 ms: ∆ " ∗ ()=− ' 4X t ^(-) ^ 0 u [Eq. 1] where () is the tissue signal and 8 is the signal at baseline. For SineO2, baseline 8 is the average value across (). Dual echo ∆ " ∗ was calculated for TE1 = 35 ms and TE2 = 90 ms: 212 ∆ " ∗ = ' 4X " 64X $ sW ^ FG$ (-) ^ 0,FG$ X−W ^ FG" (-) ^ 0,FG" Xv [Eq. 2] b. Tissue concentration-time curves: The dDSC contrast concentration-time curve, (), could be estimated as the changes in oxygen saturation using formula () =∆(). To ensure () is positive for Desaturation, we set 3(:!-;.!-+,) =−1 because ∆ 3(:!-;.!-+,) <0. Likewise, K(:!-;.!-+,) =1 because ∆ K(:!-;.!-+,) >0. For SineO2, ^+)( & " =1 and () is a sinusoid centered at 0. To convert tissue ∆ " ∗ into -+::;( (), a linear relationship was used: ∆ ",-+::;( ∗ ()= ] _-+::;( × -+::;( () [Eq. 3] 82 in which ] _-+::;( =22.6. This value of ] _-+::;( was obtained using Monte Carlo simulation. The specifics of this simulation procedure are detailed in Supplemental Methods M3. Arterial ∆ " ∗ signals were converted into arterial concentration-time curves using a previously published T2* oximetry calibration. 455 Quantitative values of transverse relaxivity are quadratically related to blood oxygenation, ",0/,,Q ∗ =12.7+101.1(1−) " at hematocrit of 39%. 16 Differentiating this relationship with respect to time and substituting blood tracer concentration yield: ∆ ",0/,,Q ∗ ()=202.2 × (1−)× 0/,,Q () [Eq. 4] We approximated within the scaling term as the average between each subject’s maximum and minimum arterial SpO2 for each respiratory challenge. c. Arterial input function (AIF): Individual AIFs were obtained semi-automatically as a combination of manual and automatic AIF selection techniques using unsmoothed images. The first step included manual selection of arterial signal from the middle cerebral artery, !.-(.+!/ , to ensure the reasonable shape of the AIF. 203 The second step was automatic selection of venous output signal, #(),;: , by choosing 20 voxels with later delay compared to the global signal and the highest area-under- curve. 456 Lastly, to reduce partial volume effects on !.-(.+!/ , 217 the time integral of !.-(.+!/ was scaled to match that of #(),;: . 290 Since DSC is often performed in a semiquantitative manner in clinical practice, we also explored CBF, CBV, and transit time using a linear relationship between ∆ " ∗ and () in both blood and tissue voxels, with ] _-+::;( = ] _0/,,Q =1. 83 d. Cerebral blood volume (CBV): CBV was calculated as = _ ` <m%(% &'(()* (-)) <m%(% %DBB= (-)) , where (∙)is the function to calculate the area under the curve of each signal. Final quantitative CBV values were scaled by the brain density 1.05 g/mL and the hematocrit correction factor . In deoxygenation-based DSC, since the contrast is confined within red blood cells instead of plasma, we used =1/0.69 to account for the difference between hematocrit within the capillaries and within large blood vessels. 220,457 • Desaturation & Resaturation: Since () is positive, (()) =∫(). • SineO2: To account for the negative trough of sine wave, () =∫|()|. e. Time delay (TD): TD was calculated as the temporal lag between tissue and arterial signals. • Desaturation & Resaturation: Signals were interpolated from a TR of 1.5 seconds with zeropadding in the frequency domain by a factor of 10. Time delay of the tissue compared to the global signal ( -+::;( ) was calculated using cross-correlation method. 458 Again, voxels with the fastest 98 th percentile were excluded to suppress arterial signal. Therefore, delay of the arterial signal ( !.-(.+!/ ) was estimated as the 2 nd percentile of the time delay distribution within each subject. Finally, TD was calculated as = -+::;( − !.-(.+!/ . • SineO2: Since the signal at each voxel was a sinusoid whose phase could be estimated, TD was computed as the phase delay between tissue and AIF: = n &'(()* 6n #H&*H'#D "oe / [Eq. 5] 84 where * is the fundamental frequency of the sinusoidal stimulus and is the phase of the sine wave calculated from the Fourier transform of the BOLD signal with respect to time. 273 The arterial phase !.-(.+!/ was estimated as the 2 nd percentile of the phase distribution for each subject. f. Cerebral blood flow (CBF): • Desaturation & Resaturation: CBF was performed using singular value decomposition (SVD) deconvolution between the tissue signals and the AIF: 459 -+::;( ()= 0/,,Q ()⨂(×()) [Eq. 6] where () is the residue function representing the contrast fraction present in the vasculature at the particular time point. 460 In gadolinium DSC, the SVD threshold, ^B3 , is calculated from the signal-to- noise. 461 Since the SNR of deoxygenation-based DSC is lower, we fixed ^B3 at a more permissive threshold of 0.10 to avoid CBF underestimation. • SineO2: CBF was calculated based on the Fourier transform of the tracer kinetics model in Eq. 6 in the frequency domain: | -+::;( ()|=| 0/,,Q ()|××|()|. [Eq. 7] Magnitude spectra | 0/,,Q ()| and | -+::;( ()| were computed from the Fourier transforms of the concentration-time curves in the tissue and the AIF. 85 The residue function was modeled as a decaying exponential, () = 6-/7 with time constant . 460 Since at low frequencies the time constant of a first- order system can be approximated as the phase delay, 462 we set = calculated from Eq. 5. The magnitude spectrum of the residue function was computed as |()|= ' q(' 7 ⁄ ) " \("oe) " . In Equation 7, the only unknown was the scalar CBF. Instead of calculating CBF by division at the fundamental frequency, least-squares fitting was used to fit the entire magnitude spectra to estimate CBF at each voxel. g. Mean transit time (MTT): Using central volume theorem, mean transit time (MTT) was computed as: = %AB %A> [Eq. 8] 3.2.5. Statistical analysis Statistical analysis was performed in R statistical package. 463 To compare global CBF, CBV, MTT, and TD between two different perfusion techniques, paired t-test was performed. Between three or more techniques, one-way repeated measures ANOVA was performed; if significant, paired t-test with Bonferroni correction was performed post hoc and adjusted p-value was reported. Ordinal ratings from the post-study survey were compared using the Wilcoxon signed rank test. Reproducibility was assessed from two iterations of Desaturation and Resaturation and from two iterations of one-cycle sinusoid for SineO2. Paired t-test, test-retest coefficient of 86 variation, and intersubject coefficient of variation were reported for each perfusion parameter in each technique. Within each subject, correlation and limits of agreement analyses were performed to assess agreement between different methods. Pearson correlation coefficient was calculated from the linear fit between perfusion values from 300 regions-of-interest for pairs of perfusion techniques. 95% limits of agreement were calculated as ̅ ±1.96× Q , where ̅ is the mean difference and Q is the standard deviation of the differences between two methods in the ROI set. For the cohort-wide analysis, repeated measures correlation 464 and Bland-Altman limits of agreement for multiple observations per individual 465 were performed after controlling for between-subject variance. All limits of agreement were normalized by the average of the two methods and reported as percentages. 466 3.3. Results 3.3.1. Demographics The 10 subjects were predominantly female (60% female), 37 ± 11 years of age, and 70 ± 18 kg in weight. Average hematocrit was higher in male (44.1 ± 2.5%) compared to female (39.6 ± 1.7%, p=0.03). Average eGFR was 120 ± 32 mL/min/1.73m 2 and negatively correlated with age (r 2 =0.54, p=0.02). None of the subjects had eGFR lower than 60 mL/min/1.73m 2 , and every subject responded ‘no’ to all 6 questions on the Choyke survey. 3.3.2. Gadolinium-based DSC Gadolinium injection was successful in all subjects. One subject developed symptom of dizziness and nausea immediately following injection as well as hypertension which lasted 30 minutes. A second subject developed a diffuse maculopapular rash over the face and upper trunk in addition to hypertension. Neither developed any shortness of breath or progression of 87 symptoms. Both subjects were treated with Benadryl 50 mg by mouth and observed for one hour before discharge. 3.3.3. Respiratory challenges All 10 subjects successfully completed the 3 respiratory challenges, but one subject was excluded from the group analysis because of gas leakage from the mask caused by facial hair. The remaining 9 subjects had baseline SpO 2 of 98.4 ± 0.7%, EtO2 of 111.1 ± 6.9 mmHg, and EtCO2 of 40.4 ± 3.3 mmHg. • Desaturation: EtO2 dropped from the initial baseline to a nadir of 42.7± 5.8 mmHg (Figure 3.1A); the nadirs were not significantly different between the first and second repetitions of desaturation (p=0.75). The post-challenge EtO2 demonstrated a spike to 146.9 ± 16.3 mmHg immediately at the return to normoxia but leveled out to a stable baseline similar to the initial values (p=0.29, Figure 3.1A). This post-desaturation increase in EtO2 was not reflected in the SpO2 (Figure 3.1C), which decreased to 88.8 ± 2.9% at the nadir and was not different between the iterations (p=0.21). Gradient-echo MRI signals demonstrated an average 4.0 ± 1.3% decrease compared to baseline (∆ " ∗ of 1.14 ± 0.38 s -1 , Figure 3.1D), significantly higher in GM (1.34 ± 0.45 s -1 ) compared to WM (0.81 ± 0.28 s -1 , p<0.01) and not different between the two desaturation iterations (p=0.99). Overall temporal SNR was 6.0 ± 1.9, with higher tSNR in the GM (6.6 ± 2.3) than the WM (5.3 ± 1.5, p<0.01). • Resaturation: EtO2 started at 42.1 ± 3.6 mmHg at baseline, increasing to a peak of 125.2 ± 14.5 mmHg and overshooting the target EtO2 of 100 mmHg by 25% (Figure 3.1E). The two repetitions of 88 resaturation showed similar EtO2 changes (p=0.38) and SpO2 changes (p=0.22); pre- and post- challenge values were not significantly different for EtO2 (p=0.19) and SpO2 (p=0.30). SpO2 response was asymmetric with respect to EtO2, with more sluggish response to reach the hypoxic baseline but rapidly increasing to the resaturation peak. The gradient-echo signals exhibited a slight negative trend during the hypoxic baseline but were relatively stable compared to the large effect size of the resaturation peaks (Figure 3.1H). Gas-induced increases in the gradient-echo signals were 4.0 ± 0.5% compared to initial baseline (∆ " ∗ of 1.14 ± 0.20 s -1 ), with an average GM-WM ratio of 1.8 ± 0.2 that were similar across the two repetitions (p=0.39). Accordingly, temporal SNR was 6.4 ± 1.9, higher in GM than WM (p<0.01) and similar to the tSNR in the Desaturation challenge (p=0.63). • SineO2: For the SineO2 respiratory pattern, EtO2 fluctuated in the non-linear pattern (Supplemental Methods M1) with a peak of 103.5 ± 2.7 mmHg and nadir of 51.0 ± 0.7 mmHg (Figure 3.1I). This stimulus yielded an approximately sinusoidal pattern of SpO2 (Figure 3.1K) that fluctuated between 96.8 ± 1.0% peak and 86.7 ± 2.4% nadir. EtCO2 was well controlled at resting values throughout the challenge (Figure 3.1J). Gradient-echo signals were modulated sinusoidally, with a peak-to-peak sine-wave amplitude of 2.7 ± 0.5% (peak-to-peak ∆ " ∗ of 0.77 ± 0.13 s -1 , Figure 3.1L). Sinusoidal amplitudes were higher in GM (0.92 ± 0.15 s -1 ) compared to WM (0.53 ± 0.09 s -1 , p=0.03). Temporal SNR was 2.9 ± 0.7 on average, with higher tSNR of 3.3 ± 0.8 in the GM and 2.4 ± 0.6 in the WM (p=0.02). • Survey: At the end of the three oxygen challenges, all physiological parameters returned to pre- stimulus baselines. Post-study surveys (Supplemental Table 3.S1) showed that most subjects 89 found the Desaturation and SineO2 respiratory challenges ‘not’ uncomfortable, not significantly different from the discomfort of wearing the mask (p=0.13 and p=0.13 respectively), whereas the Resaturation challenge was rated between ‘mildly’ and ‘moderately’ uncomfortable, higher than Desaturation (p=0.02) but not significantly different from SineO2 (p=0.16). None of the subjects reported any lasting symptoms related to respiratory challenges after the experiment. 3.3.4. Quantitative single-echo perfusion Quantitative measurements of CBF, CBV, TD, and MTT in the whole brain, GM, and WM for each respiratory challenge are shown in Table 3.1. Perfusion maps are shown in a representative subject for three respiratory challenges and two reference perfusion techniques (Figure 3.2). Each individual subject’s CBF, CBV, TD, and MTT maps are shown respectively in Supplemental Figures 3.S1-S4. 90 Table 3.1. Quantitative perfusion values for Desaturation, Resaturation, SineO2, DSC, ASL, and PC. WB = whole brain, GM = grey matter, WM = white matter. All results are displayed in the format: mean ± standard deviation (coefficient of variation). Deoxygenation-based DSC techniques CBF (mL/100g/min) CBV (mL/100g) TD (seconds) MTT (seconds) Desaturation WB 44.6 ± 16.6 (0.37) 4.5 ± 1.8 (0.39) 3.9 ± 0.9 (0.23) 6.5 ± 1.0 (0.15) GM 54.1 ± 20.4 (0.38) 5.3 ± 2.0 (0.38) 3.6 ± 0.8 (0.22) 6.2 ± 1.0 (0.16) WM 28.1 ± 10.8 (0.39) 3.0 ± 1.2 (0.41) 4.4 ± 1.1 (0.25) 6.9 ± 0.9 (0.12) Resaturation WB 59.2 ± 19.1 (0.32) 5.4 ± 1.2 (0.22) 3.2 ± 0.9 (0.28) 6.1 ± 1.0 (0.17) GM 72.0 ± 22.8 (0.32) 6.4 ± 1.3 (0.20) 3.0 ± 0.9 (0.29) 5.8 ± 1.0 (0.17) WM 36.0 ± 12.2 (0.34) 3.4 ± 0.8 (0.24) 3.6 ± 1.0 (0.28) 6.5 ± 1.0 (0.15) SineO2 WB 50.0 ± 13.1 (0.26) 3.5 ± 1.0 (0.29) 7.2 ± 3.1 (0.43) 5.1 ± 0.9 (0.18) GM 61.0 ± 7.4 (0.25) 4.1 ± 1.1 (0.28) 6.7 ± 3.2 (0.49) 5.0 ± 1.0 (0.20) WM 29.3 ± 7.4 (0.25) 2.3 ± 0.7 (0.31) 8.0 ± 3.0 (0.38) 5.3 ± 1.0 (0.18) Reference perfusion techniques DSC WB 30.6 ± 5.8 (0.19) 2.9 ± 0.4 (0.13) 2.3 ± 0.2 (0.08) 6.2 ± 1.0 (0.16) GM 36.9 ± 6.4 (0.17) 3.4 ± 0.4 (0.12) 2.1 ± 0.2 (0.10) 6.0 ± 0.9 (0.15) WM 21.0 ± 4.7 (0.22) 2.0 ± 0.3 (0.13) 2.5 ± 0.2 (0.07) 6.3 ± 1.2 (0.19) ASL WB 47.4 ± 7.1 (0.15) NA NA 1.13 ± 0.08 (0.7) GM 60.4 ± 9.8 (0.16) NA NA 1.06 ± 0.08 (0.08) WM 31.3 ± 7.4 (0.24) NA NA 1.22 ± 0.08 (0.06) PC WB 66.1 ± 7.7 (0.12) NA NA NA 91 a. CBF: Among the three respiratory challenges (Table 3.1, Figure 3.2A), there were no significant differences in CBF (p=0.10). Desaturation, Resaturation, and SineO2 resulted in higher CBF measurements compared to gadolinium-based DSC (p=0.04, p<0.01, and p<0.01 respectively) but no significant difference compared to ASL (p=0.60, p=0.16, and p=0.52 respectively). Additionally, Desaturation and SineO2 yielded lower CBF compared to PC (p<0.01) but the bias did not reach statistical significance for Resaturation (p=0.23). In terms of reproducibility, there was no significant difference between CBF computed from the two repetitions of Desaturation and Resaturation (p=0.45 and p=0.30 respectively), and test-retest coefficient of variation was 20% and 23% respectively. For SineO2, CBF was not significantly different between the two cycles (p=0.36), and the coefficient of variation was 21%. The intersubject CBF coefficient of variation was lowest in phase contrast (12%), comparable in Figure 3.2. Quantitative perfusion maps for respiratory challenges and conventional perfusion techniques. (A) CBF, (B) CBV, (C) TD, and (D) MTT maps for Desaturation, Resaturation, SineO2, DSC and ASL. 92 ASL (15%) and DSC (19%). SineO2 (26%) and Resaturation (32%) had nearly double the intersubject CBF variability, and Desaturation had the highest coefficient of variation (37%). Similar biases were observed regionally in both the GM and WM between the respiratory- derived CBF values with ASL and DSC CBF. Representative CBF maps (Figure 3.2A) by the 5 different methods and the corresponding ROI-based pairwise correlation analyses and Bland- Altman plots within a representative subject are shown in Figure 3.3. These analyses were repeated for each individual subject and shown in Supplemental Figures 3.S5-S10. Both the bias and scatter vary considerably across subjects. There are a greater number of positive outliers in the Bland Altman plots, suggesting some regions with inappropriately high CBF estimates by Resaturation. Table 3.2A summarizes the Pearson correlation coefficient and Bland-Altman limits of agreement between Desaturation, Resaturation, and SineO2 compared with the reference techniques ASL and DSC. These limits of agreement across ROI were large between all combinations of techniques but were comparable to values observed between ASL and DSC techniques (Supplemental Figure 3.S11). In general, the three respiratory-based perfusion methods exhibited higher Pearson correlation and smaller limits of agreement among themselves than compared to either DSC or ASL (Table 3.2A). 93 Figure 3.3. Regional agreement in CBF between respiratory challenges Desaturation, Resaturation, and SineO2 and reference standards DSC and ASL in a representation subject. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between pairs of respiratory challenge and reference perfusion measurement. 94 Table 3.2. Repeated measures correlation and limits of agreement between Desaturation, Resaturation, SineO2, and reference techniques. Pearson correlation r Limits of Agreement (%) Pearson correlation r Limits of Agreement (%) A. CBF (mL/100g/min) B. CBV (mL/100g) DSC Desaturation 0.58 –13 ± 67 [–79, 53] 0.54 –22 ± 60 [–81, 36] DSC Resaturation 0.64 –37 ± 61 [–97, 24] 0.60 –41 ± 55 [–95, 13] DSC SineO2 0.42 –16 ± 74 [–88, 56] 0.51 –5 ± 57 [–61, 51] ASL Desaturation 0.38 29 ± 76 [–46, 103] NA NA ASL Resaturation 0.42 5 ± 76 [–69, 80] NA NA ASL SineO2 0.31 25 ± 82 [–55, 106] NA NA Desaturation Resaturation 0.81 –24 ± 45 [–68, 20] 0.72 –18 ± 46 [–63, 27] Desaturation SineO2 0.60 –3 ± 58 [–60, 53] 0.80 17 ± 39 [–20, 55] Resaturation SineO2 0.65 21 ± 57 [–35, 76] 0.85 36 ± 36 [1, 72] DSC ASL 0.58 –42 ± 62 [–103, 19] NA NA C. TD (seconds) D. MTT (seconds) DSC Desaturation 0.12 –59 ± 109 [–163, 52] 0.10 –10 ± 54 [–63, 43] DSC Resaturation 0.32 –36 ± 109 [–143, 71] 0.18 –6 ± 52 [–57, 45] DSC SineO2 0.19 –96 ± 83 [–178, –15] 0.21 11 ± 62 [–50, 72] Desaturation Resaturation 0.46 22 ± 60 [–37, 81] 0.39 4 ± 38 [–33, 41] Desaturation SineO2 0.49 –49 ± 67 [–114, 17] 0.46 21 ± 49 [–27, 69] Resaturation SineO2 0.60 –69 ± 56 [–124, –15] 0.38 17 ± 49 [–31, 65] 95 b. CBV: Measurements of CBV were higher in Resaturation compared to SineO2 (p<0.01) but not significantly different from Desaturation (p=0.27, Figure 3.2B). Resaturation and Desaturation CBV values were higher than gadolinium-based DSC CBV (p<0.01 and p=0.03), but CBV by SineO2 were not significantly different from DSC (p=0.11 respectively). The coefficients of variation for Desaturation, Resaturation, and SineO2 were 18%, 19%, and 30% respectively, with no difference between CBV calculated from each iteration or sinusoidal cycle for each challenge (p=0.78, p=0.25, and p=0.96 respectively). The intersubject variability was comparable in DSC (13%) and Resaturation (22%) but higher in SineO2 (29%) and Desaturation (39%). On a regional basis, correlation and Bland-Altman analyses between respiratory-based and gadolinium-based CBV values are shown in Table 3.2B and Supplemental Figures 3.S12- S14. CBV limits of agreement for Desaturation, Resaturation, and SineO2 with respect to the DSC CBV were comparable to CBF measurements (Table 3.2B). The three respiratory challenges had significant tighter agreement among themselves (average r-value 0.79) than with DSC (average r-value 0.55). c. TD: Average TD measured by temporal lag in Desaturation and Resaturation paradigms were shorter compared to the phase lag measured with SineO2 (p<0.01 for both). TD maps (Figure 3.2C) demonstrated robust grey-white matter differentiation, with significantly longer delay in WM compared to GM (p<0.01 for all). The test-retest coefficient of variation for TD was 11%, 10%, and 16% for Desaturation, Resaturation, and SineO2 with no significant difference between the iterations (p=0.53, p=0.08, and p=0.57 respectively). Among the susceptibility-based techniques, DSC demonstrated the 96 lowest intersubject variability (8%), followed by SineO2 (14%), Desaturation (23%) and Resaturation (28%). Regionally, DSC-derived and respiratory-derived TD maps were qualitatively similar, but the latter were more heterogenous and noisy (Figure 3.2C). Correlation and Bland-Altman analyses are shown for respiratory-derived TD compared with gadolinium-derived TD (representative subject in Supplemental Figure 3.S15). Desaturation and Resaturation have better agreement versus DSC compared to SineO2 (Table 3.2C), although overall performance was much worse than for CBF or CBV measurements. d. MTT: Global MTT calculated with the central volume theorem was lower in SineO2 compared to Desaturation (p<0.01) and Resaturation (p=0.03). MTT was significantly higher in gadolinium DSC compared to SineO2 (p=0.02) but not different from Desaturation (p=0.60) and Resaturation (p=0.63). Arterial transit time calculated by ASL was an order of magnitude shorter compared to all MTT values (p<0.01). There was no difference between MTT calculated from different iterations for Desaturation, Resaturation and SineO2 (p=0.22, p=0.06, and p=0.07 respectively), and the test- retest coefficient of variation was comparable between SineO2 (15%), Desaturation (11%) and Resaturation (18%). The intersubject variability was also comparable between the different challenges: Desaturation (15%), Resaturation (17%), SineO2 (18%) and gadolinium DSC (16%). TD and MTT maps exhibited similar spatial characteristics, although MTT maps were smoother (Figure 3.2C vs Figure 3.2D). Even though WM still demonstrated prolonged MTT compared to grey matter in Desaturation and Resaturation (p<0.01 and p<0.01 respectively), there was less contrast between GM and WM with SineO2 and no significant grey-white difference (p=0.28). Low correlation coefficients were observed between DSC MTT with the three respiratory 97 challenges (Supplemental Figure 3.S15), but high MTT correlation was still observed across Desaturation, Resaturation, and SineO2 (Table 3.2D). 3.3.5. Quantitative dual-echo perfusion Dual-echo perfusion values are shown in Supplemental Table 3.S2, demonstrating underestimation compared to single-echo perfusion in Table 3.1. Despite similar patterns of grey- white differentiation in CBF, CBV and TD, dual-echo MTT values were no longer different between GM and WM (p=0.70). Lower tSNR was observed, with 4.2 ± 1.0 for Desaturation, 4.7 ± 2.0 for Resaturation, and 1.8 ± 0.4 for SineO2 (p<0.01 for all compared to single-echo tSNR). Even with higher noise and lower quality, dual-echo perfusion (Supplemental Figure 3.S16) still showed similar spatial patterns seen on single-echo perfusion maps. 3.3.6. Semi-quantitative perfusion In assuming () =∆ " ∗ () for both tissue and blood, Supplemental Table 3.S3 summarizes semi-quantitative perfusion values. The most striking difference was almost a twofold higher CBV computed from Desaturation compared to Resaturation and SineO2 (p<0.01) as well as the significantly higher CBF in Desaturation than Resaturation (p<0.01). However, despite the bias in absolute perfusion measurements, spatial distribution of CBF and CBV were similar between the quantitative and semi-quantitative perfusion maps (Supplemental Figure 3.S17). 3.4. Discussion In this work, we performed deoxygenation-based DSC (dDSC) with oxygen respiratory challenges to modulate cerebral saturations and induce susceptibility-weighted MRI signal changes, after which regional CBF, CBV, TD and MTT were computed using tracer kinetics modeling. We investigated three different respiratory paradigms, including Desaturation, 98 Resaturation, and SineO2. Overall, dDSC perfusion values showed reasonable reproducibility and were within acceptable range of the literature, 467 and perfusion maps demonstrated expected spatial distribution, with greater CBF and CBV in the grey matter and prolonged transit time in the white matter. 468,469 These maps also demonstrated good regional agreement, giving confidence in the capability of measuring perfusion using oxygen gas challenges as an alternative to gadolinium exogenous contrast. Traditional DSC perfusion experiments utilize injection of a gadolinium-based contrast agent that washes through the vasculature and causes susceptibility-induced signal losses, which enables quantitation of perfusion parameters through tracer kinetics modeling. 470 Despite its frequent usage in clinical routines, gadolinium contrasts pose a risk of anaphylactic shock, 471 with a rate of serious adverse reactions of 1 in 40,000 and fatal reactions of 1 in 1,000,000. 472 Even though no serious gadolinium reactions were encountered in this study, two subjects developed symptoms that required medical intervention and monitoring. Additionally, gadolinium has been shown to increase the risk of nephrogenic systemic fibrosis, especially in patients with renal impairment, 473,474 and to accumulate in different tissues such as brain, bone, and liver with unknown long-term consequences. 475–479 Therefore, to eliminate the reliance on gadolinium, this work proposes the use of oxygen respiratory challenge as an alternative endogenous contrast, thus removing any risks associated with exogenous contrasts. Even though the use of hypoxia respiratory challenge carries its own set of precautions, 480–483 the risks are significantly minimized 484 compared to injection of exogenous contrasts especially when using a high-precision computer-controlled gas blender, 452 and thus this enables safer and less invasive perfusion experiments. The Desaturation and Resaturation paradigms were designed to mimic a gadolinium DSC experiment. Both approaches produced bolus susceptibility contrast in the BOLD signal, one negative and the other positive. There was significant asymmetry in the EtO2 response, with longer response to Desaturation than Resaturation, likely representing capacitive effects of 99 pulmonary dead space and the ability to use oxygen concentrations greater than room air. The area under the ∆ " ∗ curve was greater for Resaturation than Desaturation and likely accounts for its more robust estimates of CBF, CBV, and transit time. However, Resaturation was the least comfortable of the respiratory challenges and the more prolonged exposure to saturations below 80% could raise concerns in subjects with evolving cerebral ischemia. The SineO2 paradigm represents a novel oxygen respiratory challenge with several potential advantages. Firstly, it uses the entire observation interval in the measurement rather than requiring stimuli interspersed with recovery periods. Secondly, CBF can be directly calculated in the frequency domain using model-based approximations to the residue function, circumventing the need for deconvolution using Fourier-transform division 485 or SVD-based approaches in the time domain. 459 Direct Fourier transform deconvolution techniques require robust noise suppression due to amplified noise in the division between the tissue and AIF frequency spectrum. 485 On the other hand, SVD-based deconvolution induces distortion of high frequency components 461 and yields absolute CBF that is highly dependent on the threshold PSVD chosen. 486 Adaptive PSVD thresholds have been proposed for gadolinium DSC, 461 but SVD-derived CBF values are still only an approximation of the deconvolution results. The SineO2 stimulus produces a sinc function at the fundamental frequency, smoothed by the tissue residue function. This method reduces the effects of noise components, including acquisition-related noise and low-frequency physiological noise, 487,488 and yields high quality CBF maps independent of possible confounding parameters in the SVD deconvolution process. Thirdly, the SineO2 stimulus provides a simple and robust mechanism for calculating TD via the phase of the stimulus frequency. Taken together SineO2 resulted in comparable intersubject variability in perfusion measurements with the other respiratory challenges and was considered only mildly uncomfortable. All three deoxygenation-based perfusion maps showed reasonable global and regional agreement with gadolinium-based DSC and ASL. For CBF, Desaturation and Resaturation had 100 better agreement with DSC and ASL than did SineO2; they also had excellent agreement with one another. This discrepancy is understandable since DSC, Desaturation and Resaturation use the same SVD deconvolution pipeline for CBF calculation 459 whereas SineO2 CBF was calculated in the frequency domain. For CBV, all three respiratory challenges had comparable agreement with DSC. All techniques were calculated from the same method of area under the curve, 459,489 so better agreement is expected. Even though CBF computed from gas challenges were comparable to ASL, the limits of agreement were larger than when compared to gadolinium DSC. Scattergram analysis demonstrated areas of overestimation by dDSC CBF compared to ASL, possibly due to incomplete pial vein suppression. Since gadolinium and deoxygenation DSC techniques are both heavily affected by large veins close to the surface of the brain, 490 they showed better agreement with each other compared to ASL. 190,194 Compared to CBF and CBV, regional agreement in temporal dynamics between different techniques was lower. While estimates were unbiased and MTT was consistent among the three gas challenges, low correlation with DSC was observed. Since MTT was derived from CBF and CBV, any errors introduced during either CBF or CBV calculation will propagate into the MTT estimates. Poor agreement was also observed between TD from gadolinium DSC and gas challenges. Additionally, a larger bias was observed between SineO2 and gadolinium DSC TD compared to the other two challenges. This longer TD could be explained by the approximation of the residue function as a decaying exponential. 459 In this first-order system, the phase delay of transfer function – TD in SineO2 case – is related to transit time in an inverse tangent function. Therefore, at low frequency, TD is similar to MTT, 462 as observed in our SineO2 data. On the other hand, Desaturation, Resaturation, and gadolinium injection impart a much higher and broader range of input frequencies. The shorter TD in the Desaturation and Resaturation paradigms suggested that cross-correlation was dominated by low-frequency components, causing non- linear variations between TD and MTT with respect to stimulus characteristics. Despite the 101 discrepancy in agreement with standard techniques, similar spatial distribution of prolonged transit time in WM compared to GM was still observed across challenges in both MTT and TD. One important aspect of DSC MRI is the quantitative values computed for CBF and CBV. 255,256,259,491 Most clinical DSC studies only estimate relative perfusion values, such as tumor compared to healthy appearing tissue perfusion. 445,492,493 However, normality in contralateral normal appearing white matter might not be a valid assumption, stressing the need of absolute perfusion values. 402,494,495 In this work, we take advantage of previous in vitro T2* oximetry calibration 455 to relate changes in intravascular gradient-echo signals to cerebral saturation, as the equivalent to the gadolinium concentration-time curve in traditional DSC experiments. 256 For tissue signals, a simple simulation procedure was performed to estimate the linear relationship between tissue gradient-echo signal and saturation. However, this simulation was performed with limited range of parameters and requires further work to investigate the effects of confounders such as diffusivity, air-tissue interface, and contamination of signals due to pial vein proximity on the measured dDSC signals. 496 Overall, the capability to compute absolute perfusion measurements instead of relying on semi-quantitative values or values normalized to WM CBF 497 represents a step forward for deoxygenation-based DSC developments. One limitation to this work is the possible vasoactive effects of hypoxia on CBF and CBV. Desaturation and SineO2, which maintained EtO2 above 50 mmHg, were not expected to trigger compensatory vasodilation. 179 On the other hand, Resaturation had an EtO2 baseline at 40 mmHg and yielded higher CBF compared to Desaturation, as expected since prolonged exposure to hypoxia has been shown to increase blood flow by 7.0 ± 2.9% per 10% decrease in SpO2. 498 Another possible source of bias between gadolinium-based DSC and other methods is from the lack of a power injector at our research facility, so suboptimal injection rates of contrast likely contributed to the underestimation of flow compared to dDSC and compared to the standard ASL technique. 499,500 102 Despite these limitations, high correlation and tight limits of agreement were observed, especially amongst the three respiratory challenges. Overall, this work demonstrates the feasibility, safety, and sensitivity of dDSC to measure perfusion in healthy controls. Agreement between dDSC perfusion values with gadolinium DSC measurements gives confidence in this technique’s ability to accurately quantify CBF, CBV, and temporal dynamics in the brain. Looking forward, further investigation is required to validate this perfusion technique in a larger cohort of subjects in addition to assessing its clinical utility in patients with ischemic strokes or brain tumors before dDSC can be considered for clinical routines. 3.5. Supplemental Information 3.5.1. Supplemental Methods Supplemental Methods M1 v Structural MRI: All imaging was acquired on a 3T Philips Achieva (Philips Medical Systems, Best, Netherlands) with a 32-channel head-coil. Anatomical 3D T1 was acquired at the start of the protocol with the following parameters: TR = 8 ms, TE = 3.7 ms, flip angle = 8°, with 1 mm isotropic resolution. Post-gadolinium 3D T1 was also acquired after gadolinium-based DSC to assess for contrast leakage in all subjects. Pre-processing steps on structural pre-contrast T1-weighted images consist of brain extraction and tissue classification in BrainSuite. 18 Additional segmentation into 300 regions-of- interest (ROI) was performed by co-registration of T1 images into the USCBrain anatomical atlas, 19 with modifications to the deep white matter structures where white matter ROI were manually added. 103 v Deoxygenation-based DSC: Dynamic gradient-echo BOLD MRI was acquired for each respiratory challenge, with TR = 1.5 seconds, TE = 35/90 ms, flip angle = 52°, FOV = 190×190×100 mm 3 , resolution = 2.5 mm isotropic, SENSE = 1, multi-band SENSE = 4, phase-encoding direction = AP, and fat-shift direction = P. Desaturation, Resaturation and SineO2 challenges were acquired with 220, 240, and 280 dynamics respectively. Reverse-gradient BOLD was acquired for only 1 dynamic using the same imaging parameters except the opposite fat-shift direction = A along the phase encoding direction. BOLD images were corrected for EPI-induced distortion with a field map calculated from opposite phase encoding directions. Motion correction was performed in AFNI. 20 Slice timing correction was performed using FSL. 21 Registration of each BOLD dataset to T1 image was performed in BrainSuite, then the anatomical ROI in T1 native space were transformed into the BOLD functional space. Finally, BOLD images were smoothed using a 4×4×4 mm 3 Gaussian kernel. v Gadolinium-based DSC: Traditional gadolinium DSC was acquired with a dual-echo gradient-echo sequence: TR = 1.5 seconds, TE = 8/35 ms, flip angle = 30°, FOV = 190×190×100 mm 3 , resolution = 2.5×2.5×5 mm3, 160 dynamics, SENSE = 2, and no multi-band acceleration. The field-of-view was aligned with the previous dDSC acquisition at the time of scanning. Gadovist at 0.1 mmol/kg was injected using a 20 or 22 gauge IV at a rate of 4 cc per second. Due to the lack of a power injector at our research facility, contrast was injected manually by a physician, introducing some variability in the injection rate. Contrast bolus was followed immediately by 20 mL of saline flush via a three-way stopcock. Gadolinium-based DSC BOLD images were preprocessed using the previous spatial functional pipeline as for dDSC. CBF, CBV, and MTT were calculated based on previously 104 published DSC pipelines. 22 TD derived from temporal lag was also calculated. Finally, all perfusion images were rigidly registered to the deoxygenation DSC native space for regional comparison. v Arterial spin labeling (ASL): Time-encoded pseudo-continuous ASL was acquired at baseline with the following parameters: TE = 16 ms, TR = 5040 ms, Hadamard-8 matrix with seven blocks of 2000, 800, 500, 300, 250, 200, and 150 ms, PLD = 100 ms, SENSE = 2.5, resolution = 3×3×6 mm3, FOV = 240×240×114 mm 3 , 2 FOCI background suppression pulses, 2D single-shot EPI readout, and NSA = 12. M0 scans were acquired by switching off labeling and background suppression and by using the same parameters except for the TR = 2500 ms. Quantification of CBF was performed using FSL BASIL toolbox 35 with hematocrit- corrected blood T1. 36 Additional details on acquisition and processing of the time-encoded ASL sequence have been previously published. 37 Afterward, these perfusion maps were registered to dDSC native space for comparison. v Phase contrast (PC): Single-slice PC images were acquired just above the carotid bifurcation with the following parameters: TR = 17 ms, TE = 10 ms, flip angle = 10°, resolution = 0.6×0.6 mm 3 , FOV = 220×220 mm 2 , slice thickness = 5 mm, and velocity encoding gradient of 80 cm/s. Details on calculation of total blood flow from the two internal carotid arteries and vertebral arteries were published in previous works. 9,38 Finally, CBF was calculated by normalizing the total flow by the brain density 1.05 g/mL and subject-specific brain volume derived from pre-contrast T1-weighted image. 9,39 105 Supplemental Methods M2 The goal of SineO2 respiratory challenge was to generate sinusoidal fluctuations of saturation and corresponding gradient-echo signals. However, the computer-controlled gas blended (RespirAct, Thornhill Research, Toronto) manipulates EtO2 partial pressures instead of saturations. Therefore, careful designing of EtO2 pattern to account for the non-linearity in the relationship between oxygen partial pressure and saturation was required. This section assumed that EtO2 values are approximate to blood pO2 due to inability to measure pO2 in vivo continuously during respiratory challenges. To generate sinusoidal fluctuation in SpO 2 between 85% and 98% (Supplemental Figure 3.S18A), pO2 was calculated from the hemoglobin dissociation curve: " = (h& " ) E (hk8) E \ (h& " ) E , where 50 is the " at 50% saturation estimated as 26.4 mmHg and is the Hill coefficient estimated as 2.8. The computed pO2 time series is shown in Supplemental Figure 3.S18B, with sharper peaks and rounder troughs. Additionally, the slower decay time during transition from normoxia to hypoxia (Supplemental Figure 3.S18C) and faster recovery time between hypoxia and normoxia (Supplemental Figure 3.S18D) needed to be corrected for in order to yield better sinusoidal modulations in saturations. Therefore, the decay and recovery time constants ),.D,i+! →fgh,i+! and fgh,i+! →),.D,i+! were measured in a subset of 3 subjects. The final values were ),.D,i+! →fgh,i+! =30.3 and fgh,i+! →),.D,i+! =7.7 , and these time constants were used to correct for the difference in cerebral dynamic responses to transitions between oxygenation states. The final EtO2 pattern is shown in Supplemental Figure 3.S18E. However, the gas-delivery device can only provide inputs of 3 types: constant, ramp, or sine wave. Therefore, this EtO2 pattern was approximated using 2 linear ramps and 2 sine waves (Supplemental Figure 3.S18F). The final piecewise equation for EtO2 with period of 120 seconds is: 106 " ()= ⎩ ⎪ ⎪ ⎨ ⎪ ⎪ ⎧ 105−2.3∗t, h(.+,Q P u, (,) <20 −16.2∗sint "o(D,Q (-,h(.+,Q )6"8) 's8.P u+62.4, 20≤(,)<60 −11.4∗cost "o(D,Q (-,h(.+,Q )6b8) '8s.P u+62.4, 60≤(,)<90 62.4+1.5∗t, h(.+,Q P u, 90≤(,)<120 As illustrated in Figures 3.1K and 3.1L, the resulting SpO 2 and gradient-echo signals for the SineO2 were approximately sinusoidal despite non-sinusoidal modulation of EtO2. Supplemental Methods M3 Monte Carlo simulations were performed 211,300 to estimate the linear relationship between ∆ ",-+::;( ∗ and saturation as follows: ∆ ",-+::;( ∗ = -+::;( ×∆. Simulations steps include: 1. A sphere with radius : =200 was generated containing blood vessels represented by infinite and random-oriented cylinders with fixed radius, * between 1 and 50 μm. Total volume fraction was constrained at 3%. 2. Each proton’s random walk was performed to simulate diffusion of water. Each step in the x, y, and z directions followed the normal distribution ~ (0,√2∆), where is the diffusion coefficient 1 " / and ∆ is the time step 20 . 3. Phase accumulation of proton at each step was calculated by summing over the field contributions from each vessel: ∆ + =2 8 ∆ (1−) ∆ t K / . ' u " 2 + " + , where is the gyromagnetic ratio 42.58 MHz/T, is the blood oxygen saturation between 1 and 99%, is the blood hematocrit at 40%, ∆ is the susceptibility difference between fully oxygenated and fully deoxygenated blood 0.27 ppm, 367 is the angle of the vessel to 8 , is the angle with respect to the projection of 8 onto a 107 place orthogonal to the vessel and is the distance between proton and vessel. Only protons outside the vessel were simulated and the intravascular signal was neglected. 4. Signal decay curve was computed by summing up all the protons with calculated phase for a specified gradient echo sequence with TE of 35 ms. Signal changes ∆ ",-+::;( ∗ due to saturation differences and the slopes were calculated at each vessel radius * (example relationship at radius 25 μm shown in Supplemental Figure 3.S19A). Final -+::;( value was averaged for radius 10 to 50 μm (Supplemental Figure 3.S19B): -+::;( =22.6 108 3.5.2. Supplemental Tables Supplemental Table 3.S1. Frequency distribution table. Level of discomfort None Mild Moderate Extreme Desaturation 8 1 1 0 Resaturation 2 4 3 1 SineO2 6 4 0 0 Wearing Mask 3 4 2 1 109 Supplemental Table 3.S2. Quantitative dual-echo perfusion values. WB = whole brain, GM = grey matter, WM = white matter. Supplemental Table 3.S3. Semi-quantitative perfusion values. WB = whole brain, GM = grey matter, WM = white matter. CBF (mL/100g/min) CBV (mL/100g) TD (seconds) MTT (seconds) Desaturation WB 71.3 ± 17.5 (0.25) 7.7 ± 1.9 (0.24) 3.9 ± 0.9 (0.23) 6.9 ± 1.2 (0.17) GM 86.1 ± 20.9 (0.24) 9.0 ± 2.2 (0.24) 3.6 ± 0.8 (0.22) 6.6 ± 1.2 (0.18) WM 45.2 ± 12.5 (0.28) 5.1 ± 1.4 (0.27) 4.4 ± 1.1 (0.25) 7.4 ± 1.1 (0.15) Resaturation WB 48.0 ± 15.3 (0.32) 5.4 ± 1.0 (0.18) 3.2 ± 0.9 (0.28) 7.6 ± 1.5 (0.20) GM 58.5 ± 18.5 (0.32) 6.4 ± 1.1 (0.17) 3.0 ± 0.9 (0.29) 7.2 ± 1.5 (0.20) WM 29.0 ± 9.4 (0.32) 3.4 ± 0.07 (0.20) 3.6 ± 1.0 (0.28) 8.2 ± 1.5 (0.19) SineO2 WB 64.4 ± 14.9 (0.23) 4.6 ± 0.9 (0.20) 7.2 ± 3.1 (0.43) 5.5 ± 1.0 (0.19) GM 79.1 ± 18.1 (0.23) 5.5 ± 1.0 (0.19) 6.7 ± 3.2 (0.49) 5.2 ± 1.2 (0.23) WM 37.3 ± 9.4 (0.25) 3.1 ± 0.7 (0.23) 8.0 ± 3.0 (0.38) 5.9 ± 0.9 (0.15) CBF (mL/100g/min) CBV (mL/100g) TD (seconds) MTT (seconds) Desaturation WB 45.3 ± 14.0 (0.31) 4.0 ± 1.4 (0.34) 3.9 ± 0.9 (0.23) 5.9 ± 1.2 (0.20) GM 52.7 ± 16.7 (0.32) 4.6 ± 1.6 (0.34) 3.6 ± 0.8 (0.22) 6.0 ± 1.2 (0.20) WM 34.2 ± 10.8 (0.32) 3.0 ± 1.1 (0.38) 4.4 ± 1.1 (0.25) 5.8 ± 1.3 (0.22) Resaturation WB 46.1 ± 9.1 (0.20) 3.9 ± 0.6 (0.14) 3.2 ± 0.9 (0.28) 5.9 ± 1.5 (0.25) GM 55.3 ± 11.1 (0.20) 4.7 ± 0.7 (0.15) 3.0 ± 0.9 (0.29) 5.8 ± 1.5 (0.26) WM 32.5 ± 7.0 (0.21) 2.8 ± 0.4 (0.14) 3.6 ± 1.0 (0.28) 6.0 ± 1.4 (0.23) SineO2 WB 53.2 ± 7.8 (0.15) 3.2 ± 0.8 (0.25) 7.2 ± 3.1 (0.43) 5.4 ± 1.1 (0.20) GM 62.3 ± 7.1 (0.11) 3.7 ± 0.9 (0.24) 6.7 ± 3.2 (0.49) 5.4 ± 1.4 (0.26) WM 39.1 ± 9.9 (0.25) 2.4 ± 0.6 (0.25) 8.0 ± 3.0 (0.38) 5.5 ± 0.8 (0.15) 110 3.5.1. Supplemental Figures Supplemental Figure 3.S1. Individual quantitative CBF maps for Desaturation, Resaturation, and SineO2. Supplemental Figure 3.S2. Individual quantitative CBV maps for Desaturation, Resaturation, and SineO2. 111 Supplemental Figure 3.S3. Individual quantitative TD maps for Desaturation, Resaturation, and SineO2. Supplemental Figure 3.S4. Individual quantitative MTT maps for Desaturation, Resaturation, and SineO2. 112 Supplemental Figure 3.S5. Correlation and Bland-Altman limits of agreement analyses using 300 regions- of-interest between Desaturation and DSC CBF measurements. Supplemental Figure 3.S6. Correlation and Bland-Altman limits of agreement analyses using 300 regions- of-interest between Resaturation and DSC CBF measurements. 113 Supplemental Figure 3.S7. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between SineO2 and DSC CBF measurements. Supplemental Figure 3.S8. Correlation and Bland-Altman limits of agreement analyses using 300 regions- of-interest between Desaturation and ASL CBF measurements. 114 Supplemental Figure 3.S9. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between Resaturation and ASL CBF measurements. Supplemental Figure 3.S10. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between SineO2 and ASL CBF measurements. 115 Supplemental Figure 3.S11. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between DSC and ASL CBF measurements. Supplemental Figure 3.S12. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between Desaturation and DSC CBV measurements. 116 Supplemental Figure 3.S13. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between Resaturation and DSC CBV measurements. Supplemental Figure 3.S14. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between SineO2 and DSC CBV measurements. 117 Supplemental Figure 3.S15. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between respiratory challenges Desaturation, Resaturation, and SineO2 with DSC for TD and MTT measurements. 118 Supplemental Figure 3.S16. Quantitative dual-echo CBF, CBV, TD, and MTT maps for 9 subjects. Supplemental Figure 3.S17. Semi-quantitative single-echo CBF, CBV, TD, and MTT maps for 9 subjects. 119 Supplemental Figure 3.S18. Design of SineO2 respiratory paradigm. (A) Ideal sinusoid for SpO2 saturation input. (B) Computed pO2 signal from saturation and hemoglobin dissociation curve. (C) Decay and (D) recovery signals modeled as exponential between hypoxia and normoxia. (E) EtO2 approximated by blood pO2 signal. (F) EtO2 approximated by piecewise function of ramps and sine waves. Supplemental Figure 3.S19. Results of Monte Carlo simulation to derive relationship between ∆ !,$%&&'( ∗ and saturation . (A) The slope of linear relationship between ∆ !,$%&&'( ∗ and (1−) at vessel radius 25 μm. (B) Final value $%&&'( of was computed from average of slopes at radius 10 to 50 μm. 120 Chapter 4 : Sinusoidal CO 2 respiratory challenge for concurrent perfusion and cerebrovascular reactivity MRI 4.1. Introduction Perfusion MRI is a popular imaging technique for assessing hemodynamic impairments in a variety of central nervous system abnormalities such as intracranial tumors and acute strokes 501,444 . There are multiple different MRI techniques to measure cerebral perfusion, including phase contrast (PC), arterial spin labeling (ASL), and dynamic susceptibility contrast (DSC). Particularly, DSC MRI is a perfusion technique that is frequently performed in clinical routines, requiring intravenous injection of a contrast agent (gadolinium chelate) and dynamic imaging to capture the passage of the contrast bolus through the vasculature 385,502 . Based on the susceptibility-induced signal loss caused by the paramagnetic contrast, tracer kinetics models are applied to calculate multiple perfusion parameters. Despite its popular usage and clinical utility, DSC suffers from its reliance on exogenous gadolinium contrasts, which pose increased risks of anaphylaxis 472 , nephrogenic systemic fibrosis 242 , and gadolinium deposition in different tissues 503 . To address this drawback, recent works have proposed contrast-free deoxygenation- based DSC (dDSC) which take advantage of endogenous paramagnetic deoxyhemoglobin to induce susceptibility-weighted MRI signal losses, similar to the effects of gadolinium 203,204,205 . This dDSC technique delivers boluses of deoxygenated hemoglobin through transient exposure to low- oxygen (hypoxia) or high-oxygen (hyperoxia) gas inhalation 504,399,459 and has demonstrated 121 feasibility in healthy volunteers as well as chronic anemia subjects who had elevated blood flow and shortened transit time 203 . One of the obstacles to perfusion quantification in both gadolinium-based and deoxygenation-based DSC is the determination of CBF, which requires a deconvolution between the signals in the blood and in the tissue. Traditionally, this deconvolution is performed using a singular-value decomposition (SVD) approach in the time domain 459 . In this work, we propose to replace the transient contrast bolus with a sinusoidal respiratory gas paradigm and compute perfusion at the fundamental sinusoidal frequency in the Fourier domain, thereby simplifying the SVD deconvolution process. Additionally, instead of delivering boluses of deoxyhemoglobin through oxygen respiratory challenges, we also propose to raise and lower the concentration of deoxygenated hemoglobin through modulations of inspired CO2 level (Figure 4.1A). When combining the CO2 inhalation with the sine wave gas pattern (SineCO2), this respiratory challenge results in sinusoidal fluctuations in gradient-echo MRI signals. In order to assess the feasibility of this new perfusion technique, we evaluated SineCO2 on 10 healthy volunteers in comparison with perfusion measurements from standard gadolinium-based DSC, ASL and PC MRI. 4.2. Methods 4.2.1. Study protocol The Committee on Clinical Investigation at Children's Hospital Los Angeles approved the protocol; written informed consent was obtained from all subjects (CCI#20-00050). This study was performed in accordance with the Declaration of Helsinki. A total of 10 healthy volunteers participated in this study between April and May of 2021. Exclusion criteria included pregnancy, hypertension, diabetes, stroke or other known neurologic insult, seizures, known developmental delay or learning disability, at least one ‘yes’ answer to the 122 6-question Choyke survey 450,451 , and measured glomerular filtration rate (GFR) lower than 60 mL/min/1.73mm 2 . Imaging, vital signs, and blood samples were obtained on the same day for each subject. 4.2.2. Respiratory challenges Respiratory challenges were performed using a specialized computer-controlled gas blender (RespirAct, Thornhill Research, Toronto, Canada) that prospectively targets and manipulates end-tidal O2 (EtO2) and end-tidal CO2 (EtCO2) partial pressures 505,506 . This device measures the subject’s baseline EtO2 and EtCO2 during the initial preparation phase and delivers specific concentrations of oxygen and carbon-dioxide during the challenge phase to accurately target EtO2 and EtCO2 values. Fingertip pulse oximetry SpO2 (Nonin, Plymouth, MN) was recorded during each challenge. SineCO2 challenge was performed, in which EtO2 was clamped at subject-specific baseline and EtCO2 was modulated in a sine wave between 35 and 45 mmHg with a period of 60 seconds. 4.2.3. Structural MRI All imaging was acquired on a 3T Philips Achieva (Philips Medical Systems, Best, Netherlands) with a 32-channel head-coil. Anatomical 3D T1 was acquired at the start of the protocol with the following parameters: TR = 8 ms, TE = 3.7 ms, flip angle = 8°, with 1 mm isotropic resolution. Post-gadolinium 3D T1 was also acquired after gadolinium-based DSC to assess for contrast leakage in all subjects. Pre-processing steps on structural pre-contrast T1-weighted images consist of brain extraction and tissue classification in BrainSuite 507 . Additional segmentation into 300 regions-of- interest (ROI) was performed by co-registration of T1 images into the USCBrain anatomical 123 atlas 454 , with modifications to the deep white matter structures where white matter ROI were manually added. 4.2.4. Deoxygenation-based DSC MRI Dynamic gradient-echo BOLD MRI was acquired for each respiratory challenge, with TR = 1.5 seconds, TE = 35/90 ms, flip angle = 52°, FOV = 190×190×100 mm 3 , resolution = 2.5 mm isotropic, SENSE = 1, multi-band SENSE = 4, phase-encoding direction = AP, fat-shift direction = P, and 220 dynamics. Reverse-gradient BOLD was acquired for only 1 dynamic using the same imaging parameters except the opposite fat-shift direction = A along the phase encoding direction. BOLD images were corrected for EPI-induced distortion with field map calculated from opposite phase encoding directions. Motion correction was performed in AFNI 508 . Slice timing correction was performed using FSL 509 . Registration of each BOLD dataset to T1 image was performed in BrainSuite, then the anatomical ROI in T1 native space were transformed into the BOLD functional space. Finally, BOLD images were smoothed using a 4×4×4 mm 3 Gaussian kernel. All subsequent dDSC processing was performed in MATLAB (MathWorks, Natick, MA). Suppression of signal contribution from pial veins was performed by eliminating voxels with higher signal amplitude than the 98 th percentile in each subject. Calculation of whole brain, GM, and WM perfusion values was performed by averaging voxels within brain and tissue-specific masks in each functional sequence’s native space. Gradient-echo ∆ ∗ Single echo ∆ " ∗ was calculated from TE = 35 ms: ∆ " ∗ ()=− ' 4X t ^(-) ^ 0 u [Eq. 1] where () is the tissue signal and 8 is the average value across (). 124 Dual echo ∆ " ∗ was calculated for TE1 = 35 ms and TE2 = 90 ms 212 : ∆ " ∗ = ' 4X " 64X $ sW ^ FG$ (-) ^ 0,FG$ X−W ^ FG" (-) ^ 0,FG" Xv [Eq. 2] Venous output function (VOF) Individual VOFs were obtained automatically by choosing 20 voxels with later delay compared to the global signal and the highest area-under-curve 510 . To convert ∆ " ∗ to concentration-time curve () in both blood and tissue voxels, this manuscript assumed a linear relationship ∆ " ∗ ()= ] ×(), with coefficients ] _-+::;( = ] _0/,,Q =1. Cerebral blood volume (CBV) CBV was calculated as = _ ` ∫|% &'(()* (-)| ∫|% %DBB= (-)| , where is the brain density 1.05 g/mL and the is the hematocrit correction factor. In deoxygenation-based DSC, since the contrast is confined within red blood cells instead of plasma, we used =1/0.69 to account for the difference between hematocrit within the capillaries and within large blood vessels 220,457 . Time delay (TD) Since the signal at each voxel was a sinusoid whose phase could be estimated, TD was computed as the phase delay of tissue: = n &'(()* 6n #H&*H'#D "oe / [Eq. 5] where * is the fundamental frequency of the sinusoidal stimulus and is the phase of the sine wave 273 calculated from the Fourier transform of the BOLD signal with respect to time. The 125 arterial phase !.-(.+!/ was estimated as the 2 nd percentile of the phase distribution for each subject. Cerebral blood flow (CBF) For traditional DSC, CBF is usually calculated using singular value decomposition (SVD) deconvolution between the tissue signals and the AIF 459 : -+::;( ()= 0/,,Q ()⨂(×()) [Eq. 6] CBF was calculated based on the Fourier transform of the tracer kinetics model in Eq. 6 in the frequency domain: | -+::;( ()|=| 0/,,Q ()|××|()| [Eq. 7] Magnitude spectra | 0/,,Q ()| and | -+::;( ()| were computed from the Fourier transforms of the concentration-time curves in the tissue and the VOF. The residue function was modeled as a decaying exponential, ()= 6-/7 with time constant . 399 Since at low frequencies the time constant of a first-order system can be approximated as the phase delay 462 , we set = calculated from Eq. 5. The magnitude spectrum of the residue function was computed as |()|= ' q(' 7 ⁄ ) " \("oe) " . In Equation 7, the only unknown was the scalar CBF. Instead of calculating CBF by division at the fundamental frequency, least-squares fitting was used to fit the entire magnitude spectra to estimate CBF at each voxel. Mean transit time (MTT) Using central volume theorem, mean transit time (MTT) was computed as: = %AB %A> [Eq. 8] 126 4.2.5. Reference perfusion MRI Deoxygenation-based DSC perfusion was compared against reference perfusion techniques, including gadolinium-based DSC, ASL, and PC. Gadolinium-based DSC Traditional gadolinium DSC was acquired with a dual-echo gradient-echo sequence: TR = 1.5 seconds, TE = 8/35 ms, flip angle = 30°, FOV = 190×190×100 mm 3 , resolution = 2.5×2.5×5 mm 3 , 160 dynamics, SENSE = 2, and no multi-band acceleration. The FOV was aligned with the previous dDSC acquisition at the time of scanning. Gadovist at 0.1 mmol/kg was injected using a 20 or 22 gauge IV at a rate of 4 cc per second. Due to the lack of a power injector at our research facility, contrast was injected manually by a physician, introducing some variability in the injection rate. Contrast bolus was followed immediately by 20 mL of saline flush via a three-way stopcock. Gadolinium-based DSC BOLD images were preprocessed using the previous spatial functional pipeline as for dDSC. CBF, CBV, and MTT were calculated based on previously published DSC pipelines 212 . TD derived from temporal lag was also calculated. Finally, all perfusion images were rigidly registered to the deoxygenation DSC native space for regional comparison. Arterial spin labeling (ASL) Time-encoded pseudo-continuous ASL was acquired at baseline with the following parameters: TE = 16 ms, TR = 5040 ms, Hadamard-8 matrix with seven blocks of 2000, 800, 500, 300, 250, 200, and 150 ms, PLD = 100 ms, SENSE = 2.5, resolution = 3×3×6 mm 3 , FOV = 240×240×114 mm 3 , 2 FOCI background suppression pulses, 2D single-shot EPI readout, and 127 NSA = 12. M0 scans were acquired by switching off labeling and background suppression and by using the same parameters except for the TR = 2500 ms. Quantification of CBF and arterial transit time (aTT) was performed using FSL BASIL toolbox 511 with hematocrit-corrected blood T1 430 . Additional details on acquisition and processing of the time-encoded ASL sequence have been previously published 512 . Afterward, these perfusion maps were registered to dDSC native space for comparison. Phase contrast (PC) Single-slice PC images were acquired just above the carotid bifurcation with the following parameters: TR = 17 ms, TE = 10 ms, flip angle = 10°, resolution = 0.6×0.6 mm 3 , FOV = 220×220 mm 2 , slice thickness = 5 mm, and velocity encoding gradient of 80 cm/s. Details on calculation of total cerebral blood flow from the two internal carotid arteries and vertebral arteries were published in previous works 183,513 . Finally, CBF was calculated by normalizing the total flow by the brain density 1.05 g/mL and subject-specific brain volume derived from pre-contrast T1-weighted image 183,514 . 4.2.6. Statistical analysis Statistical analysis was performed in R statistical package 463 . To compare global CBF, CBV, MTT, and TD between two different perfusion techniques, paired t-test was performed. Reproducibility was assessed from two iterations of two-cycle sinusoid for SineCO2. Paired t-test, test-retest coefficient of variation, and intersubject coefficient of variation were reported for each perfusion parameter in each technique. Within each subject, correlation and limits of agreement analyses were performed to assess agreement between different methods. Pearson correlation coefficient was calculated from the linear fit between perfusion values from 300 ROI for pairs of perfusion techniques. 95% 128 limits of agreement were calculated as ̅ ±1.96× Q , where ̅ is the mean difference and Q is the standard deviation of the differences between two methods in the ROI set. For the cohort- wide analysis, repeated measures correlation 464 and Bland-Altman limits of agreement for multiple observations per individual 465 were performed after controlling for between-subject variance. All limits of agreement were normalized by the average of the two methods and reported as percentages 466 . 4.3. Results 4.3.1. Respiratory challenges All 10 subjects successfully completed the SineCO2 respiratory challenge, but one subject was excluded from the group analysis because of gas leakage from the mask caused by facial hair, and one subject was excluded since the timing of the gas challenge and the BOLD imaging was incorrectly aligned at the time of the experiment. None of the subjects reported discomfort during SineCO2 respiratory challenge. Baseline tidal volumes and respiration rates were 772 ± 292 mL and 17.2 ± 3.3 breaths/min and did not change significantly during CO2 modulations (p=0.27 and p=0.82 respectively). Initial EtCO2 and EtO2 recordings were 41.0 ± 3.5 mmHg and 110.4 ± 7.0 mmHg in the cohort. During SineCO2, continuous measurements of EtCO2 demonstrated sinusoidal amplitudes of 4.6 ± 0.8 mmHg (Figure 4.1A), whereas EtO2 was kept level at baseline (Figure 4.1B). SpO 2 remained level at 98.4 ± 0.9% during the challenge. Under CO2-induced vasodilation and vasoconstriction, single-echo gradient-echo MRI signal varied in a sinusoidal pattern with ΔS of 1.20 ± 0.44% (ΔR2 * = 0.34 ± 0.13 s -1 ) relative to baseline (Figure 4.1C), higher in the GM (ΔS = 1.52 ± 0.57%, ΔR2 * = 0.43 ± 0.16 s -1 ) compared to WM (ΔS = 0.58 ± 0.26%, ΔR2 * = 0.17 ± 0.07 s -1 , p<0.01). Temporal SNR was 1.36 ± 0.52 in the whole brain, 1.72 ± 0.66 in GM, and 0.92 ± 0.38 in WM. In the Fourier domain, global signals 129 demonstrated a peak at 0.17 Hz, corresponding to a sine wave period of 60 seconds (Figure 4.1D). Figure 4.2. Respiratory challenge patterns for SineCO2. (A) End-tidal oxygen (EtO2), (B) end-tidal carbon dioxide (EtCO2), (C) percent signal change in the time domain and (D) in the frequency domain. Figure 4.1. SineCO2 CBF, CBV, TD, and MTT maps for individual subjects. 130 4.3.2. Perfusion measurements Single-echo SineCO2 Perfusion parameters for the whole brain, GM, and WM are displayed in Table 4.1; individual CBF, CBV, TD, and MTT maps by SineCO2 are shown in Figure 4.2. Similar spatial distribution is observed in CBF and CBV maps (Figure 4.2AB), with GM-WM ratio of 2.1 ± 0.1 for CBV and 1.9 ± 0.1 for CBF. TD maps showed a trend of longer delay in deep WM compared to GM (p=0.24), but distribution is heterogeneous between subjects (Figure 4.2C). MTT maps demonstrated no grey-white matter differentiation (Figure 4.2D). 131 Table 4.1. Whole brain perfusion estimates by SineCO2 and 3 standards ASL, DSC and PC. CBF (mL/100g/min) CBV (mL/100g) TD (seconds) MTT (seconds) SineCO2 WB 52.0 ± 11.6 (0.22) 3.9 ± 0.5 (0.13) 7.0 ± 2.5 (0.35) 4.8 ± 1.0 (0.21) GM 65.2 ± 15.2 (0.23) 5.0 ± 0.6 (0.12) 6.3 ± 2.7 (0.44) 5.1 ± 1.1 (0.22) WM 33.2 ± 6.5 (0.20) 2.4 ± 0.3 (0.13) 8.0 ± 2.2 (0.28) 4.5 ± 0.8 (0.18) DSC WB 29.6 ± 5.4 (0.18) 2.9 ± 0.3 (0.10) 2.4 ± 0.2 (0.10) 6.4 ± 1.0 (0.16) GM 35.6 ± 6.2 (0.17) 3.4 ± 0.3 (0.10) 2.1 ± 0.2 (0.11) 6.3 ± 0.9 (0.14) WM 20.3 ± 4.2 (0.21) 2.0 ± 0.2 (0.12) 2.6 ± 0.2 (0.08) 6.6 ± 1.3 (0.20) ASL WB 48.0 ± 7.8 (0.16) NA NA 1.13 ± 0.08 (0.7) GM 60.5 ± 10.8 (0.18) NA NA 1.07 ± 0.09 (0.08) WM 32.3 ± 8.0 (0.25) NA NA 1.22 ± 0.09 (0.07) PC WB 65.9 ± 8.3 (0.13) NA NA NA Table 4.2. Correlation and Bland-Altman limits of agreement between SineCO2 and reference techniques. Pearson correlation r Limits of Agreement (%) Pearson correlation r Limits of Agreement (%) A. CBF (mL/100g/min) B. CBV (mL/100g) DSC SineCO2 0.45 –92 ± 66 [–157, –27] 0.52 –54 ± 68 [–121, 12] ASL SineCO2 0.35 –43 ± 88 [–129, 44] NA NA DSC ASL 0.52 –56 ± 65 [–120, 8] NA NA C. TD (seconds) D. MTT (seconds) DSC SineCO2 0.23 –107 ± 51 [–157, –57] 0.00 25 ± 54 [–28, 77] 132 • CBF: To evaluate quantitative perfusion by SineCO2, mean CBF values are shown in Table 4.1 in comparison with gadolinium-based DSC, ASL and PC. In the whole brain, SineCO2 CBF was not different from PC (p=0.17) and ASL (p=0.09) but was significantly higher compared to DSC (p<0.01). In terms of reproducibility, there was no significant difference between two repetitions (p=0.48) with a test-retest coefficient of variation of 28%. The intersubject coefficient of variation was 22%, slightly higher compared to DSC (18%), ASL (16%), and PC (13%). Regionally, SineCO2 overestimated DSC in both GM (p<0.01) and WM (p=0.01); however, compared to ASL, SineCO2 was higher in GM (p=0.06) but comparable within the WM (p=0.32). Within-subject correlations and Bland-Altman analyses are shown for a representative subject in Figure 4.3, and individual analyses are in Supplemental Figures 4.S1 and 4.S2. Intrasubject correlation and limits of agreement are shown in Table 4.2A, demonstrating similar correlation and width of the limits of agreement between SineCO2 and reference techniques compared to agreement amongst DSC and ASL references. 133 Figure 4.3. Regional agreement between respiratory challenge SineCO2 and reference standards DSC and ASL in a representation subject. Correlation and Bland-Altman limits of agreement analyses using 300 regions-of-interest between (A-B) SineCO2 and DSC and (C-D) SineCO2 and ASL. CBF maps in representative subject by three techniques. 134 • CBV: SineCO2 systematically overestimated CBV by DSC (p=0.04). However, regional correlation was high across ROIs and similar to agreement observed in CBF (Table 4.2B). Intersubject coefficient of variation was 13%, and test-retest coefficient of variation was 15%, demonstrating no significant difference between the two repetitions (p=0.47). • TD: TD values by SineCO2 were significantly longer compared to DSC (p=0.03). Compared to CBF and CBV measurements, TD maps were noisier (Figure 4.2C) with similar limits of agreement but lower Pearson correlation coefficient (Table 4.2C). TD demonstrated a test-retest coefficient of variation of 22% and intersubject coefficient of variation of 35%. • MTT: Similar MTT values were observed in SineCO2 compared to DSC (p=0.76), but no grey matter – white matter differentiation was observed (p=0.13). No intrasubject correlation was observed with DSC MTT (Table 4.2D). Test-retest and intersubject coefficients of variation were 14% and 21% respectively. Dual-echo SineCO2 Compared to the global ΔR2 * 0.34 ± 0.13 s -1 obtained at the first TE=35ms, dual-echo ΔR2 * was 0.22 ± 0.08 s -1 (p<0.01). Temporal SNR was lower in the dual-echo signal (tSNR = 0.82 ± 0.32, p=0.03). Individual CBF, CBV, TD, and MTT maps using the dual-echo approach (Supplemental Figure 4.S3) show a similar spatial distribution compared to single-echo perfusion maps (Figure 4.2). However, dual-echo maps are noisier and yield a higher bias in CBF compared to DSC and ASL (data not shown). 135 4.3.3. Cerebrovascular reactivity Individual CVR maps are shown in Figure 4.4. Mean CVR was 0.24 ± 0.06 %/mmHg in the cohort, significantly higher in the GM (0.28 ± 0.07 %/mmHg) compared to the WM (0.13 ± 0.03 %/mmHg, p<0.01). Spatial patterns of CVR maps are similar to CBF and CBV maps generated from the SineCO2 technique. 4.4. Discussion In this work, we employed a technique previously used to measure CVR with a CO2 respiratory challenge to modulate cerebral saturation and BOLD signal in a sinusoidal pattern 273 , after which tracer kinetics equations were applied in the frequency domain to compute perfusion. SineCO2 CBF and CBV values were within acceptable range of literature 515 , but MTT and TD was larger than expected 516 . Single-echo acquisition yielded better temporal SNR and better image quality compared to dual-echo approach. CBF estimates were compared with 3 standard techniques, gadolinium-based DSC, ASL and PC, and demonstrated no bias with ASL and PC but overestimation compared to DSC. The limits of agreement were large between SineCO2 and ASL, DSC, and PC but were comparable to agreement amongst the 3 standard techniques and Figure 4.4. SineCO2 CVR maps in individual subjects. 136 previously reported agreement between DSC and PET 515 . Despite the systematic biases, perfusion maps showed regional agreement between the techniques, indicating that SineCO2 has the potential to differentiate diseased and normal-appearing tissue in cerebral pathologies such as ischemic strokes or brain tumors. The use of CO2 vasoactive stimulus represents a divergence from previous deoxygenation-based DSC studies, which utilize hypoxia or hyperoxia respiratory challenges to directly deliver boluses of deoxygenated hemoglobin 203,204,205 . Capnic challenges raise and lower cerebral saturation through vasodilation and vasoconstriction within the capillary beds, so the sinusoidal modulations are not present on the arterial side but instead only in vessels undergoing oxygen exchange and large veins. Therefore, this source of contrast results in an anti-causal system where the VOF is used in lieu of an AIF. Conceptually, this is analogous to playing a cine- angiogram in reverse. Even though the anti-causality is not compatible with traditional tracer kinetics model 504,399 , computation of CBF in the frequency domain ignores the phase in favor of the magnitude, which is independent of the relative delay between tissue and VOF signals. CBV estimates in typical DSC experiments are corrected with the area-under-curve of a VOF signal, which is usually less vulnerable to partial volume effects compared to AIF 218 ; therefore, CBV measures are also independent of the use of VOF. Additionally, TD is calculated as the delay between the phase of the tissue signal and the 2 nd percentile of the phase distribution, which is an approximation for arterial phase and is thus unaffected by the lack of an AIF. The SineCO2 approach has some interesting properties. Overall, since the endogenous contrast is generated by oxygen exchange, it cannot detect actual or effective shunt flow, potentially underestimating true perfusion. The contrast change results from a cascaded transport system, in which the CO2 stimulus passes through an initial cerebrovascular response transfer function followed by a secondary residue function that governs the propagation of deoxyhemoglobin. The complexity of this higher-order system is simplified by the capability to extract a VOF signal, which relates to the signal only through the residue transfer function. The 137 indirect mode of contrast generation also requires some cerebrovascular reactivity to generate a signal suitable for CBF estimation, hence it is not surprising that CVR, CBF, and CBV maps resemble one another. In brain regions where resting flow is preserved but CVR is abnormal, signal-to-noise of the CBF and CBV estimates will be poor. CO2 modulations also have complex cerebrovascular and peripheral hemodynamic effects, including changes in respiratory rates, tidal volumes, heart rates, blood pressures 517 and perfusion values 153,175 , proportionally to the extent and duration of CO2 inhalation. The upward swing of the CO2 sinusoid is a hypercapnic stimulus which results in increase in CBF 153 , whereas the trough of the sinusoid represents a hypocapnic stimulus with a decrease in flow 175 . Assuming ±5 mmHg fluctuations in EtCO2 remain within the autoregulatory range 173 , 1 mmHg change in EtCO2 typically induces 1–2 mL/100g/min change in CBF 171,172,153 . By oscillating between hypercapnia and hypocapnia, this work assumes that the value measured is the average perfusion and is comparable to baseline blood flow. However, this assumption requires validation with a dynamic acquisition of ASL or PC with high enough temporal resolution to quantify fluctuations in CBF in response to CO2 respiratory challenge. Additionally, the vasoactive effects of CO 2 challenge can potentially explain the divergence in regional agreement between SineCO2 and ASL, in which CBF is overestimated compared to ASL in GM but underestimated in WM. During sustained hyperemia, cortical GM regions are prioritized compared to deep WM 146 . Even during CO2 stimulation, a steal phenomenon occurs in which blood flow preferentially increases in GM at the expense of WM 518,519,520 , causing higher sinusoid amplitudes in GM and thus overestimation of CBF in cortical regions. On the other hand, since flow changes are lower within WM, CBF measurements are underestimated in WM compared to ASL values. Other limitations include the study design of clamping EtCO2 sinusoid at 40 mmHg regardless of the subject’s initial EtCO2 levels; in subjects of high baseline CO2 partial pressure, this paradigm induced hypocapnia and hyperventilation response that lengthened transit time 521 138 and potentially explained the heterogeneous distribution in several TD maps 37,522,523 . Additionally, despite targeting a single fundamental frequency fc, in practice only a perfect sinusoid can be accomplished on positive cycles. The shape of the negative cycle depends on the subject’s hyperpneic response from the previous positive cycle, thus introducing a small nonlinearity and frequencies outside the target range. Lastly, since the signal is venous-weighted, SpO2 values could not be used to convert ∆ " ∗ to arterial saturation in concentration-time curves; therefore, the values presented here are only considered semi-quantitative. However, since relative perfusion is frequently used in clinical routines, semi-quantitative measurements may still offer insight into diseased tissue relative to contralateral normal-appearing tissue. Despite the shortcomings, the most significant advantage to SineCO2 perfusion imaging is that sinusoidal CO2 respiratory challenge is a robust mechanism to measure CVR 273,524 . Previous works have demonstrated that 32% of the variation in GM CVR is explained by variation in baseline CBF 525 , so these two parameters are tightly coupled together. However, measurement of CVR can still yield additional information, as illustrated by the existence of negative CVR values in deep WM (Figure 4.4) unseen on CBF maps. Divergence in CBF and CVR typically happens in areas of low flow and long delay, in which CBF can increase in response to CO2 but requires sufficient time to reach the hypercapnic ceiling and can potentially be classified as negative CVR 526 . In this current technique, SineCO2 CBF measurements are calculated purely from the magnitude spectrum and are independent of phase delay, but CVR estimates computed from traditional general linear model approach 266 are influenced by vascular delay 526,266,520 . Therefore, SineCO2 capability to acquire both perfusion and reactivity simultaneously in one imaging sequence is of high interest in cerebrovascular diseases and gives it an edge over other conventional perfusion MRI techniques. Most of SineCO2 potential diagnostic power lies in perfusion imaging of strokes or gliomas, especially in more vulnerable populations in whom gadolinium injection is undesirable, such as renal-impaired or pediatric patients 473,474,527,528 . However, the fundamental difference between 139 gadolinium contrast and deoxyhemoglobin contrast may allow them to play complementary roles in perfusion imaging for these pathologies. For ischemic strokes in which the penumbra is under low oxygen delivery, CO2-induced modulations in CBF can lead to reperfusion of the damaged regions 529,530 , which can yield a completely different perfusion distribution compared to gadolinium DSC. In brain tumors, gadolinium-based contrast extravasation through the disrupted blood-brain barrier can result in altered CBV measurements 531,532,533 ; on the other hand, deoxygenation-based contrast remains purely intravascular. Therefore, CBV measured using gadolinium-based DSC within gliomas might differ compared to SineCO2 CBV. These potential divergences in the two techniques require additional work to evaluate the diagnostic role of SineCO2 in different cerebrovascular pathologies. In conclusion, this validation study established feasibility of using SineCO2 to measure perfusion and demonstrated agreement between SineCO2 against 3 reference perfusion techniques, DSC, ASL, and PC. Despite the systematic bias, in clinical routines, neuroradiologists typically rely on relative perfusion differences between diseased and normal-appearing tissue rather than absolute perfusion, so SineCO2 relative perfusion maps may still be useful clinically independent of VOF selection. Additionally, SineCO2 also represents an easy approach to generate CBF maps independent of confounding parameters in SVD deconvolution and minimize MRI time by simultaneous acquisition of perfusion and reactivity in one imaging sequence. 140 4.5. Supplemental Information Supplemental Figures Supplemental Figure 4.S1. Correlation and Bland-Altman limits of agreement analyses using 300 regions- of-interest between SineCO2 and DSC CBF measurements. 141 Supplemental Figure 4.S2. Correlation and Bland-Altman limits of agreement analyses using 300 regions- of-interest between SineCO2 and ASL CBF measurements. Supplemental Figure 4.S3. SineCO2 dual-echo CBF, CBV, TD, and MTT maps for individual subjects. 142 Chapter 5 : Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia using BOLD MRI and Near-Infrared Spectroscopy 5.1. Introduction Anemia is a common blood disorder characterized by decreased red blood cells, leading to low oxygen-carrying capacity and reduced delivery of oxygen to the tissue 534 . Anemia is the hallmark of several common genetic diseases, including sickle cell disease (SCD) and thalassemia. Since the brain is exquisitely sensitive to interruptions in oxygen supply, previous studies have shown that both acquired and congenital anemias, including thalassemia and sickle cell anemia, are often associated with higher risk of stroke and silent cerebral infarcts (SCI) 535,536 . Among many stressors connected to the etiology of SCI, obstructive sleep apnea and nocturnal desaturation have been linked to cerebrovascular accidents in SCD and anemic patients 537,538 . Even though apnea-related low oxygenation levels and frequent hypoxia episodes are likely contributing factors to the increased stroke risks in chronically anemic subjects, the dynamics of these desaturations has not been well studied. This present work explored the brain’s dynamic response to hypoxia using two methods: near-infrared spectroscopy (NIRS) and blood-oxygen-level-dependent (BOLD) MRI. NIRS is an optical imaging technique that measures tissue oxygen saturation based on the difference in transmitted and received photons at near-infrared wavelengths 539 . On the other hand, functional BOLD MRI measures subtle changes in regional cerebral deoxyhemoglobin concentration 143 consequent to the modulation of neural metabolism; BOLD MRI is frequently used as a biomarker for disease or evaluation of clinical therapies 540 . Thus, the goal of this study was to compare cerebrovascular responses to a transient hypoxia exposure in a patient population (sickle cell disease) enriched for both SCI and obstructive sleep apnea, using patients with non-sickle anemia to control for the effects of decreased hemoglobin alone as well as normal control subjects for reference. 5.2. Methods 5.2.1. Study population The Committee on Clinical Investigation at Children’s Hospital Los Angeles approved the protocol; written informed consent and/or assent were obtained from all subjects. This study was performed in accordance with the Declaration of Helsinki. A total of 72 subjects were enrolled between January 2012 and May 2017. Exclusion criteria were the following: 1) pregnancy; 2) hypertension; 3) diabetes; 4) overt stroke or other known neurologic insult; 5) seizures; and 6) known developmental delay or learning disability. Patients with previously identified SCI were allowed to participate. As defined by the Silent Cerebral Infarct Multi-Center Clinical Trial (SIT), an SCI is an MRI signal abnormality that was equal or greater than 3mm in diameter and visible on two orthogonal planes on T2-FLAIR images and no associated neurological symptoms 541 . Only subjects older than 12 years of age were included in the study. Imaging, vital signs and blood samples were obtained on the same day for each subject. Complete blood count and quantitative hemoglobin electrophoresis were analyzed in our clinical laboratory. All chronically transfused patients were studied on the same day of their blood transfusion, prior to receiving blood, when their hemoglobin levels were at the nadir. 144 The first patient group consisted of 26 SCD subjects: 19 subjects with homozygous hemoglobin S (HbSS), 7 subjects with heterozygous combination of hemoglobin S and hemoglobin C (HbSC) and 5 subjects with heterozygous hemoglobin S and a β0 thalassemia mutation (HbSβ0) which behaves similarly to the HbSS mutations. The second group consisted of 15 patients with non-sickle chronic anemia disorders who did not suffer from any other known medical conditions apart from their chronic anemia (referred to as anemic controls, ACTL). The third group consisted of 31 African and Hispanic American healthy controls (CTL). Most controls were first or second-degree relatives of the SCD patients in order to ensure a similar economic and social background compared to the patient group. Therefore, sickle cell trait was common among control subjects; these were subjects who had approximately 40% of hemoglobin S (HbS) in their blood but no cells containing exclusively HbS. 5.2.2. Hypoxia gas challenge The experimental setup is illustrated in Figure 5.1. At the start of the image acquisitions, patients were breathing through a two-liter reservoir rebreathing circuit supplied by pressurized, non-humidified room air at 12 liters per minute. This system included one-way valves to prevent partial gas mixtures and respiratory bellows (Invivo Corporation, Gainesville, FL) to display the breathing pattern and frequency. At 50 seconds into the data acquisition, the room air gas mixture was switched to 100% nitrogen until the patient had completed 5 breaths (approximately 25 seconds), then the circuit was changed back to room air. Peripheral arterial oxygen saturation (SpO2) was acquired by fingertip pulse oximetry concurrently with forehead NIRS and BOLD MRI during the gas challenge. 145 Figure 5.1. Experimental setup for transient hypoxia gas paradigm and concurrent SpO2, NIRS and BOLD MRI acquisitions. 5.2.3. Near-infrared spectroscopy (NIRS) A NIRS system (NIRO-200 system, Hamamatsu Photonics, Hamamatsu City, Japan) was used to measure relative changes of oxygenated hemoglobin (OxyHb) and deoxygenated hemoglobin (DeoxyHb) in the frontal cerebral circulation continuously throughout the hypoxia challenge 539 . The NIRS system operated with three wavelengths of light (775, 810 and 850nm). The probes were placed on the side of the subject’s forehead just above the eyebrow. These NIRS signals were acquired at 1kHz and synchronized to the BOLD acquisitions via a Biopac MP150 data acquisition system (BioPac, Goleta, CA). A modified Beer-Lambert law was used to calculate the relative change in oxygenated hemoglobin (OxyHb), deoxygenation hemoglobin (DeoxyHb) and total hemoglobin (TotalHb, equivalent to the sum of OxyHb and DeoxyHb). 5.2.4. Magnetic resonance imaging Each participant underwent an MRI study using a 3T Philips Achieva with an 8-element phased-array coil. A 3D T1-weighted image was acquired covering the whole brain (160 sagittal slices) with TR = 8ms, TE = 4ms, flip angle = 8°, in-plane resolution = 1mm × 1mm, FOV = 256mm Figure 5.1. Experimental setup for transient hypoxia gas paradigm and concurrent SpO2, NIRS and BOLD MRI acquisitions. 146 × 256mm and slice thickness = 1.0mm. T2-weighted fluid-attenuated inversion recovery (T2- FLAIR) 3D image was acquired with the following parameters: TR = 4800ms, TE = 257.9ms, TI = 1650ms, in-plane resolution = 1mm × 1mm, FOV = 256mm × 256mm and slice thickness = 1.3mm. BOLD images were acquired using gradient-echo echo-planar-imaging sequence: TR = 2000ms, TE = 50ms, in-plane resolution = 2.3mm × 2.3mm, FOV = 220mm × 220mm, 26 axial slices, slice thickness = 5mm, SENSE factor of 2. A total of 150 volumes were collected in five minutes. To measure global cerebral blood flow, a phase contrast MRI was also obtained, positioned just above the carotid bifurcation: TR = 1286ms, TE = 77ms, FOV = 220mm × 220mm, resolution = 1.2mm × 1.2mm, velocity encoding gradient of 200mm/s and slice thickness = 5mm. MR angiography image was acquired: TR = 23ms, TE = 3.45ms, flip angle = 18°, FOV = 220mm × 220mm, resolution = 0.38mm × 0.38mm and slice thickness = 1.4mm. 5.2.5. Definitions of NIRS and BOLD dynamic hypoxia parameters Different parameters for hypoxia dynamics were defined from the NIRS and BOLD responses in hypoxia, which were modeled by a piecewise function: an affine function during desaturation and an exponential function during recovery. The following definitions of hypoxic parameters were provided for DeoxyHb but without loss of generality were used to compute response dynamics for OxyHb and BOLD time series: (1) DeoxyHbhypoxia measure was defined as the peak absolute value of DeoxyHb measured in the entire challenge duration. (2) DeoxyHbpre-hypoxia was the mean DeoxyHb signal before hypoxic gas, calculated from the start of the sequence to the start of nitrogen gas administration at 50 second. (3) DeoxyHbpost-hypoxia was the mean signal after the signal has returned to baseline; because typical hypoxia effects do not last past 100 seconds following the start of gas administration, the post-hypoxia period was defined from 120 seconds to the end of the sequence. 147 (4) ∆DeoxyHb (the depth of the hypoxic change, µmol) was calculated as ∆DeoxyHb = |DeoxyHbpre-hypoxia – DeoxyHbhypoxia|. (5) TTPDeoxyHb (time-to-peak) was defined as the duration between the start of desaturation response to the peak DeoxyHbhypoxia signal. (6) T½DeoxyHb (half-life of the return to baseline) was the time for the hypoxia signal to recover 50% of the previous signal change. The transition time points (time at onset of hypoxia and end of hypoxia) were determined by iterating through and selecting the time points with the lowest least-squares error compared to the piecewise model. (7) RDeoxyHb was the signal recovery level computed as RDeoxyHb = DeoxyHbpost-hypoxia – DeoxyHbpre-hypoxia. This signal was not computed for the BOLD time series due to confounds from low-frequency drifting during the five-minute acquisition. Regarding the BOLD signal, the same measures were calculated for each voxel of the BOLD time series, normalized by baseline BOLDpre-hypoxia. Pictorial illustrations of these definitions can be found in Figure 5.2D. 5.2.6. Image processing Structural T1 images were registered to the Montreal Neurological Institute (MNI) template with the FMRIB Software Library (FSL); BOLD images were preprocessed with FSL and registered to MNI template space using a standard spatial functional pipeline 406 . Global hypoxic parameters were computed as an average within a whole-brain mask in the MNI template. MNI-based grey matter mask, an eroded white matter mask and three vascular territories masks were used to calculate average BOLD hypoxic responses for these different brain regions. The white and grey matter masks were generated out of the Colin 27 Average Brain white matter and grey matter templates 542 . The white matter mask was eroded using a 2mm Gaussian kernel to avoid the partial volume effect. Masks of the anterior cerebral artery, middle 148 cerebral artery and posterior cerebral artery circulations were based on the published templates of vascular territories in both hemispheres 414 . 5.2.7. Baseline cerebral blood flow Whole-brain cerebral blood flow was measured using phase contrast MRI 543 . A single, two-dimensional phase contrast slice was positioned approximately 1 to 5cm superior to the carotid bifurcation with the aim of optimal orthogonality to the carotid and vertebral arteries. Vessel boundaries were semi-automatically segmented from the complex difference images using Canny edge detection and mapped to the phase difference image for blood flow calculation. 5.2.8. Clinical MR Imaging MR angiography for identification of vasculopathy and T2-FLAIR images for identification of SCI were reviewed by three observers blinded to disease status: BT (12 years of experience), JCW (10 years of experience) and SC (10 years of experience). Discrepancies were resolved by simultaneous viewing with consensus. 5.2.9. Silent cerebral infarcts (SCI) analysis Semi-manual segmentation of individual SCI was performed on T2-FLAIR in a previous study on the same patient cohort 395 . Each SCI lesion volume and centroid on the MNI template were saved for statistical analysis. The segmentations were then fused to create a SCI density map that localized SCIs across all the SCD subjects. In order to explore the effect of silent strokes, the SCD group was split into two subgroups: subjects without infarcts (SCI–) and subjects with abnormal infarcts (SCI+). The ACTL group was not subdivided because the group size was too small. 149 To determine spatial co-localization between BOLD parameters and silent infarct distribution, we performed permutation analysis of SCI on desaturation depth and timing two- sample t-maps 395 . To minimize the bias caused by the BOLD signal within the dead tissue of an infarction, this permutation analysis was not performed on individual subject’s maps, but the group maps in which areas with SCI presence constituted lesion-prone regions – or regions more vulnerable to SCI development. The reference signal was computed as the mean values within the original infarct positions on the map. To generate a robust background signal for the null statistic, 100 permutation operations were performed, each placing an infarct centroid in a random location within an eroded white matter mask and calculating the mean values in the displaced volume. Two-tailed two-sample t-tests were performed between BOLD parameters derived from normal appearing regions and regions vulnerable to silent infarcts. 5.2.10. Statistical analysis Statistical analysis on global desaturation values was performed in JMP (SAS, Cary, NC). One-way ANOVA with Dunnett’s post hoc correction was used to examine the difference in clinical and hypoxic variables between SCD and control, and between ACTL and control; independent samples t-test was used to compare SCD and ACTL patient groups. Univariate and stepwise multivariate regressions were performed against predictors of anemia and SCD severity, including age, sex, hemoglobin, cerebral blood flow, presence of SCI, transfusion status, white blood cell count, platelet count, reticulocytes, fraction of S-cells, fetal hemoglobin and cell-free hemoglobin level. These predictors have been demonstrated to be characteristic of the pathophysiology of SCD and chronic anemia disorders 536,544–548 . All variables having univariate p<0.10 were included as candidates for stepwise multivariate analysis. Variables were retained in the final model for p<0.05. 150 Pearson correlation tests were performed on ΔBOLD, T½, TTP and R against hemoglobin level and SpO2. The resulting p-values were corrected for multiple comparisons using False Discovery Rate Benjamini-Hochberg method, with q=0.05. Voxel-wise effect size tests were performed to generate a Cohen’s d-map in each patient group. Cohen’s d was calculated by: d = Δ BOLD s baseline £ with Δ BOLD as the group mean desaturation depth and s baseline as the standard deviation of the normalized baseline. Two-sample t-maps were generated with Analysis Functional NeuroImages (AFNI) to compare desaturation parameters between controls and SCD or ACTL patients; t-statistic was calculated by: t = (x SCD/ACTL – x CTL ) (s p ¤ 1 n SCD/ACTL + 1 n CTL ) M with n SCD/ACTL and n CTL as group sizes, ^%3/<%42 and %42 as parameter means, s SCD/ACTL and s CTL as parameter standard deviations and s p as the pooled standard deviation: s p =¥ (n SCD/ACTL –1) s SCD/ACTL 2 +(n CTL –1) s CTL 2 n SCD/ACTL +n CTL -2 . 5.3. Results 5.3.1. Demographics Table 5.1 summarizes the demographics of the three participating groups. The SCD and ACTL groups were matched for age and sex but the healthy controls were older (p<0.01) and had a higher proportion of females (p=0.03). Therefore, age and sex were included as independent variables in all regression analyses. No significant difference was observed in the NIRS and global BOLD dynamics between control subjects with sickle cell trait and non-trait control subjects 151 (p=0.43 for ΔBOLD, p=0.77 for ΔOxyHb and p=0.40 for ΔDeoxyHb). None of the patients consciously perceived the desaturation episode and no complications were encountered. The BOLD imaging and fingertip SpO2 were acquired on all three subject groups. Due to sensor malfunction, the NIRS analysis only retained 19 SCD patients, 24 controls and 12 ACTL patients out of 72 subjects in the study cohort. 5.3.2. Hematologic markers Compared to anemic patients, control subjects had higher hemoglobin concentration (p<0.01), higher hematocrit (p<0.01), lower cell-free hemoglobin level (p<0.01) and lower cerebral blood flow (p<0.01). The SCD and ACTL groups were well matched for hemoglobin (p=0.78) and cell-free hemoglobin levels (p=0.62), but the SCD patients had higher reticulocytes (p=0.03), white blood cells count (p<0.01), fetal hemoglobin (p=0.02) and fraction of S-cells (p<0.01). Out of 26 SCD patients, 46.2% were transfused compared to 80.0% in the ACTL patients. Within the SCD group, the median HbS percentage was 67.2% in the non-transfused group and reduced to 15.3% in the transfused group. 152 Table 5.1. Patient demographic and hematologic data. Bold letterings indicate statistical significance (p<0.05). CTL (N=31) SCD (N=26) ACTL (N=15) p-value (SCD vs. CTL) p-value (ACTL vs. CTL) p-value (SCD vs. ACTL) Age (Years) 28±12.3 21±8.2 22±5.8 0.02 0.13 0.65 Sex 7M, 24F 14M, 12F 5M, 10F 0.03 0.70 0.18 Hemoglobin Electrophoresis 15AA, 16AS 19SS, 5SC, 2Sβ0 13AA, 2AE Cerebral Blood Flow (mL/100g/min) 62±10.9 95±22.6 85±12.7 <0.01 <0.01 0.07 Transfusion 0/31 12/26 12/15 <0.01 <0.01 <0.01 SCI presence 3/31 9/26 2/15 0.03 0.94 0.09 Hemoglobin (g/dL) 13±1.4 10±1.9 10±1.5 <0.01 <0.01 0.78 Hematocrit (%) 39±3.6 29±4.3 31±3.9 <0.01 <0.01 0.10 White Blood Cell (x10 3 ) 6±2.0 10±4.7 6±1.9 <0.01 0.91 <0.01 Platelet Count (x10 3 /µL) 245±53.8 292±112.6 255±111.8 0.10 0.93 0.21 Reticulocytes (%) 1±0.6 9±5.3 2±2.8 0.11 0.99 0.03 S-cells Fraction (%) 0 44±30.1 0 <0.01 1.00 <0.01 Fetal Hemoglobin (%) 1±2.1 6±7.8 2±2.3 <0.01 0.66 0.02 Cell-free Hemoglobin (%) 6±5.0 18±18.1 20±24.0 <0.01 <0.01 0.62 5.3.3. Clinical MR imaging Based on MR angiography, only 1 SCD patient had severe bilateral anterior cerebral artery stenosis. Although no subject had a history of stroke, T2- FLAIR imaging demonstrated SCIs in 3 in control patients (9.7%), 9 in SCD patients (34.6%, p=0.05) and 2 in ACTL patients (13.3%, p=1.00), as described in Table 5.1. 153 5.3.4. Pulse oximetry and NIRS dynamics Figure 5.2A shows a typical SpO2 measurements during the challenge, demonstrating similar depth and duration to spontaneous desaturations during anatomic scanning when the subject was sleeping (Figure 5.2B). The nitrogen challenge produced minimum SpO2 values of 78.8 ± 9.8%. The ΔSpO2 measure was strongly correlated with hemoglobin levels (r 2 =0.37, p<0.01). Group average NIRS parameters are summarized in Table 5.2. Both ΔOxyHb and ΔDeoxyHb were correlated with ΔSpO2 (r 2 =0.21, p<0.01 for ΔOxyHb; r 2 =0.30, p<0.01 for ΔDeoxyHb). The depth ΔDeoxyHb was associated with hemoglobin level (r 2 =0.11, p=0.04), but ΔOxyHb was not. The T½ for both oxyhemoglobin and deoxyhemoglobin were shorter in anemic subjects (p=0.04 and p<0.01 respectively) and correlated with hemoglobin (r 2 =0.11, p=0.02 for OxyHb; r 2 =0.18, p<0.01 for Figure 5.2. Transient hypoxia model and curve fitting for SpO2, BOLD and NIRS signals. (A) Representative recording of SpO2 signal during the hypoxia challenge. (B) SpO2 signal recording from the same patient prior to gas challenge while patient was sleeping during anatomic scanning. (C) Representative recordings of SpO2, global BOLD, NIRS OxyHb and DeoxyHb changes during the hypoxia challenge. (D) Example of curve fitting for a typical DeoxyHb signal. 154 DeoxyHb), but TTP was not different in anemic subjects. Compared to pre-hypoxia baseline, all three groups demonstrated a decrease in ROxyHb (p<0.01) and a rise in both RDeoxyHb (p<0.01) and RTotalHb (p<0.01). The RTotalHb level was correlated with resting hemoglobin concentration (Figure 5.3A), with patients with significantly lower hemoglobin exhibiting a deficit in the capability to increase total hemoglobin in response to hypoxia. Table 5.2. Group average NIRS and BOLD desaturation parameters. Bold letterings indicate statistical significance (p<0.05). CTL (N=31) SCD (N=26) ACTL (N=15) p-value (SCD vs. CTL) p-value (ACTL vs. CTL) p-value (SCD vs. ACTL) ΔDeoxyHb (µmol) 1±0.4 2±0.6 1±0.6 0.03 0.21 0.69 ΔOxyHb (µmol) 1±0.6 2±0.5 2±0.9 0.22 0.08 0.37 T½DeoxyHb (s) 24±8.0 18±8.5 17±5.5 0.06 0.03 0.59 T½OxyHb (s) 19±10.2 12±5.9 16±8.9 0.03 0.53 0.32 TTPDeoxyHb (s) 25±4.1 21±5.3 21±6.3 0.07 0.13 0.89 TTPOxyHb (s) 26±9.5 20±3.7 25±6.7 0.02 0.89 0.09 RTotalHb (µmol) 0.08±0.074 0.05±0.061 0.03±0.059 0.28 0.09 0.38 ΔBOLD (%) 6±2.1 8±2.6 8±2.6 <0.01 0.03 0.94 T½BOLD (s) 16±3.5 13±3.0 15±3.6 <0.01 0.35 0.03 TTPBOLD (s) 23±3.7 21±4.5 23±3.7 0.27 0.96 0.37 5.3.5. Global BOLD dynamics Similar to the NIRS analysis, a larger drop in SpO2 correlated with higher BOLD changes (r 2 =0.50, p<0.01) and shorter timing dynamics (r 2 =0.42, p<0.01). The ΔBOLD was negatively correlated with resting hemoglobin level (r 2 =0.09, p=0.04) as illustrated in Figure 5.3B. Group average BOLD parameters are shown in Table 5.2, and hypoxic parameters within different tissue 155 types and arterial territories are shown in Supplemental Table 5.S1.The SCD and ACTL groups displayed a deeper desaturation in the white matter (p<0.01) and grey matter (p<0.01) compared to healthy controls. T½ for white matter (p<0.01), grey matter (p<0.01) and arterial perfusion territories (p<0.01 for all three territories) were shorter for SCD group in comparison with controls. Strong associations were also observed between the ΔBOLD, T½BOLD, TTPBOLD and corresponding DeoxyHb and OxyHb signals (Figures 5.3C and 5.3D). T½BOLD demonstrated a strong association with hemoglobin (r 2 =0.19, p<0.01) but TTPBOLD had no relationship with hemoglobin level. Figure 5.3. Correlations between BOLD and NIRS hypoxic depths and hemoglobin levels. Correlations between hemoglobin levels with (A) baseline recovery RTotalHb and (B) ΔBOLD demonstrated the role of anemia in compromised hypoxic and vasodilatory response in chronic anemia. Strong inter- modality correlations between BOLD desaturation depth ΔBOLD and NIRS (C) ΔDeoxyHb and (D) ΔOxyHb further affirmed the use of these signals as estimates for global cerebral tissue oxygenation. 156 5.3.6. Classic risk factors as predictors of global BOLD dynamics Table 5.3 summarizes the univariate and multivariate predictors for the global values of T½BOLD, TTPBOLD and ΔBOLD. Patient sex (p=0.03) and cell-free hemoglobin levels (p=0.03) were the only significant predictors of TTP, and both were retained on multivariate analysis (combined r 2 =0.18). Out of 12 regressors listed, 9 predictors were correlated with T½BOLD on univariate analysis, but only cerebral blood flow and white blood cell count remained significant on multivariate analysis (combined r 2 =0.37). Univariate predictors of ΔBOLD were age, cerebral blood flow, hemoglobin, transfusion status, tissue oxygenation and white blood cell count, but only age and white blood cell count were retained after multivariate analysis (combined r 2 =0.25). Table 5.3. Predictors of T½BOLD, TTPBOLD and ΔBOLD. Bold letterings indicate retention on multivariate analysis. β is the standardized regression coefficient for predictors retained on the multivariate model, and r 2 is the coefficient of determination for the univariate regression. Predictor T½BOLD TTPBOLD ΔBOLD Cerebral Blood Flow (mL/100g/min) r 2 =0.22, p<0.01, β = –0.47 – r 2 =0.16, p<0.01 Age (years) r 2 =0.09, p=0.02 – r 2 =0.14, p<0.01 β = –0.37 Sex (male = 1) – r 2 =0.08, p=0.03, β = –0.28 – Hemoglobin (g/dL) r 2 =0.17, p<0.01 – r 2 =0.09, p=0.02 Transfusion Status r 2 =0.09, p=0.02 – r 2 =0.08, p=0.03 White Blood Cells (x10 3 ) r 2 =0.21, p<0.01, β = –0.46 – r 2 =0.15, p<0.01, β = 0.39 Reticulocytes (%) r 2 =0.05, p=0.08 – – S-cells fraction (%) r 2 =0.09, p=0.02 – – Fetal Hemoglobin (%) r 2 =0.07, p=0.04 – – Cell-free Hemoglobin (mg/dL) r 2 =0.09, p=0.02 r 2 =0.07, p=0.03, β = –0.27 – 157 5.3.7. Effects of hemoglobin on regional variations in BOLD signal Figure 5.4A displays the group average ΔBOLD maps in representative axial and sagittal views in the three patient groups. Similar to global ΔBOLD, regional ΔBOLD dynamics also had significantly larger depth in SCD and ACTL compared to control subjects, with greater between- group differences observed in the grey matter compared to the white matter. With similar baseline BOLD variability between subject groups, SCD and ACTL patients showed larger Cohen’s d- values compared to healthy controls (Figure 5.4B). Figure 5.4. Mean, effect size, and hemoglobin correlation ΔBOLD maps in three patient groups. (A) Average desaturation depth ΔBOLD showed globally higher desaturation depth in anemic patients compared to healthy controls, more so in grey matter than white matter. Representative silent cerebral infarcts are illustrated as white labels on these maps. (B) Cohen’s d-maps of ΔBOLD demonstrated larger hypoxic effect size in anemic patients compared to controls. 158 Figure 5.5. Two-sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between SCD patients and controls. (A) Two-sample t-maps of desaturation depth ΔBOLD, time constants T½BOLD and TTPBOLD between SCD patients and healthy controls with correction for age and sex showed areas of high variability in hypoxic response co-localized with regions of high silent-stroke risk. (B) Parameter t-maps corrected for hemoglobin levels in addition to age and sex showed that anemia accounted for a large degree of group discrepancy. The disparity between parameter t-maps corrected for hemoglobin level and subdivided into (C) SCI– subgroup and (D) SCI+ subgroup compared to healthy controls demonstrated that patients with silent infarcts have a higher degree of spatial heterogeneity in their regional BOLD hypoxic responses. Representative silent cerebral infarcts are illustrated as white labels on these maps. 159 In examining the voxel-wise hypoxic dynamics between anemic and healthy subjects, two- sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between SCD patients and healthy controls were shown in Figure 5.5A, and between ACTL patients and healthy controls in Figure 5.6. Generally, t-scores for T½BOLD and TTPBOLD were less than zero, indicating shorter regional timing response parameters in SCD patients compared to controls. After adjusting for hemoglobin, age and sex, both T½BOLD and TTPBOLD maps had many more normal t-scores but spatial variability remained higher than expected (Figure 5.5B). Figure 5.6. Two-sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between ACTL subjects and controls. Age-sex-corrected (A) and age-sex-hemoglobin-corrected (B) two-sample t-maps of ΔBOLD, T½BOLD and TTPBOLD between non-sickle anemia patients and healthy controls showed that correcting for hemoglobin in this patient population produced a modest effect on reducing the variations between anemic patients and controls. Representative silent cerebral infarcts are illustrated as white labels on these maps. 160 Insight into this observation was provided by plotting hemoglobin-corrected two-sample t- maps separately for SCI– (Figure 5.5C) and SCI+ patients (Figure 5.5D) compared with healthy controls. The ΔBOLD, T½BOLD, and TTPBOLD t-maps for SCI– subjects were spatially homogenous and had relatively few abnormal t-scores (0.54%, 1.30% and 0.12% of voxels with |t|>2 respectively). In contrast, t-score maps from patients who were SCI+ demonstrated marked spatial heterogeneity and more severe t-score extremes in both positive and negative directions (13.1%, 10.6% and 4.5% of voxels with |t|>2 respectively). SCI+ patients had markedly prolonged T½BOLD in the right anterior and middle cerebral arteries distribution and shorter in the left anterior cerebral artery and middle cerebral artery distribution (Figure 5.5D and Supplemental Figure 5.S2). The TTPBOLD measures remained shortened in the frontal regions and were abnormally increased in the posterior region distribution in SCD patients (Supplemental Figure 5.S2). Voxel- wise two-sample t-maps are shown for ACTL group in Figure 5.6, with a distribution of abnormal t-scores in their ΔBOLD, T½BOLD, and TTPBOLD t-maps (10.4%, 2.5% and 2.4% of voxels with |t|>2 respectively). However, the hypoxic timing responses in ACTL were less spatially heterogenous compared to the SCD group (Figure 5.6 and Supplemental Figure 5.S2). Table 5.4 summarizes the permutation results computed on the hemoglobin-corrected t- maps for SCI– and SCI+ sickle-cell patients. SCI+ patients had slightly shorter T½BOLD, markedly shorter TTPBOLD and deeper ΔBOLD in both white matter regions at risk and normal appearing white matter than SCI– patients. In SCI+ patients, regions at risk for stroke had shorter TTPBOLD and smaller ΔBOLD than normal appearing white matter, while the converse was true in SCI– patients. T½BOLD was longer in regions at risk for stroke for patients in both subgroups. 161 Table 5.4. Two-tailed Student’s paired t-tests were used to compare desaturation depth and timing within infarct-prone white matter and normal appearing white matter. Bold letterings indicate statistical significance (p<0.05). Hb-corrected t-maps Infarct-prone WM Normal Appearing WM p-value T½BOLD SCI– –0.3 ± 0.47 –0.4 ± 0.55 0.14 SCI+ –0.4 ± 0.81 –0.5 ± 1.02 <0.01 TTPBOLD SCI– 0.4 ± 0.42 0.4 ± 0.46 0.03 SCI+ –0.6 ± 0.67 –0.4 ± 0.75 <0.01 ΔBOLD SCI– 0.6 ± 0.49 0.6 ± 0.49 <0.01 SCI+ 1.2 ± 0.86 1.4 ± 0.88 <0.01 5.4. Discussion Using concurrent measurement of NIRS and BOLD MRI during a desaturation gas challenge, the present work investigated the human dynamic brain response to transient hypoxia. This state of induced hypoxia mimics nocturnal desaturation – a common disorder in SCD and chronic anemia disorders linked to increased risks for strokes and cardiovascular diseases 537,538 . Our results revealed that anemic patients experienced deeper global oxygen desaturation and shorter half-time of recovery consistent with their higher blood flow. We demonstrated striking differences in the timing and depth of the desaturation between anemic patients and controls, especially between SCI+ and SCI– sickle-cell patients. The patients with SCI exhibited pronounced spatial heterogeneity in these comparative maps, notably anterior-posterior variations in TTP maps and left-right variations in T½ maps. In these patients, regions at high risks for infarcts were associated with slower saturation recovery despite a shorter time to peak desaturation suggestive of local micro-vascular obstruction. Our short and well-tolerated hypoxia protocol 405 behaved similarly to dynamic susceptibility contrast MRI except the signal loss was caused by increased concentration of paramagnetic deoxygenated hemoglobin rather than an exogenous contrast agent. Transient hyperoxia has been used as a magnetic tracer 205 , but use of transient hypoxia has not been 162 described in humans. Comparison of this transient hypoxia model between NIRS and BOLD time courses showed strong inter-modality correlation 549 ; this was reassuring since NIRS predated BOLD and was a well-validated tool in tissue oxygenation measurement 539 . The strongest correlation was observed between DBOLD and DDeoxyHb, rather than RTotalHb 549 , suggestive of hemoglobin deoxygenation as the initial hypoxic response before blood flow changes; this observation was in line with previous work which showed more rapid deoxyhemoglobin-sensitive R2* changes during hypoxia compared to cerebral blood flow 550 . The depth and time course of NIRS and BOLD response paralleled the SpO2 signal, suggesting that cerebral metabolic rate and blood flow were relatively stable over the period of transient hypoxia despite the small rise in RTotalHb. To place the predicted flow changes into perspective, step changes of inhaled oxygen from room air to pO2 of 50 torr lowered oxygen saturation to 80% (comparable to the desaturations in this study) and increased brain blood flow by 10% 551 . However, the time constant of response was 79.6 ± 29.2 seconds, so approximately 25 seconds of exposure is likely to trigger a much smaller response. Using these empiric parameters, we demonstrated abnormal blood flow, resistance and compliance of the vascular bed in the anemic patients. We contend that the higher ΔBOLD, shorter TTPBOLD and shorter T½BOLD in anemic patients reflect lower cerebrovascular resistance and serve as a compensatory mechanism for higher blood flow. These changes were greatest in the grey matter, probably as a result of increased capillary density in the cerebral cortex in chronically hypoxic patients 552 . Neovascularization is more hindered in the white matter, potentially explaining the more muted changes there 552 . While TTPBOLD and T½BOLD reflect many factors, increased RTotalHb is specific for compensatory vasodilation which is triggered when partial pressure pO2 falls below 50mmHg 179 . The RTotalHb was positive in most healthy subjects but blunted in anemic subjects, potentially reflecting exhaustion of vasodilatory reserve 553 . Hypoxic challenge, however, has been demonstrated to cause greater increases in minute ventilation in anemic subjects 554 , and the 163 corresponding decrease in end-tidal CO2 could partially counter hypoxic vasodilation. Subsequent studies with controlled end-tidal CO2 concentration are warranted to eliminate contributions from exaggerated respiratory compensation to RTotalHb changes. In multivariate regression analyses of global desaturation parameters, the dependence of DBOLD on age is physiologically understandable, as vascular resistance tends to increase monotonically with age in many disease states including SCD 536,546,555 . Similarly, the association between shorter timing dynamics and higher cerebral blood flow was also more likely a physiologic than pathologic statistical association. The association between microvascular flow dynamics and white blood cell count had been previously demonstrated in the Cooperative Study of Sickle Cell Disease 545 . The association between increased cell-free hemoglobin and male sex with global BOLD timing was consistent with disease severity as cell-free hemoglobin had been shown to correlate with systemic endothelial dysfunction in SCD 556 and male sex with higher blood pressure and higher stroke rate 557 . The striking spatial heterogeneity in the timing and depth of the desaturation, particularly the SCI+ subgroup is likely a marker of heightened capillary transit time heterogeneity 37 , which is frequently observed in progressive microvascular diseases 38 . The increased baseline blood flow in chronic anemias tends to worsen the transit time heterogeneity, leading to impaired regional oxygen extraction and exposing these patients to ischemic injury during periods of restricted oxygen delivery 37 . This microvascular impairment and heterogeneity was consistent with previous works that demonstrated diffuse abnormalities in regional cerebral blood flow in sickle-cell patients lacking overt stroke or abnormal angiography 395,558 . T½ maps were particularly prolonged in infarct prone white matter regions in deep watershed areas, consistent with reports of increased oxygen extraction fraction in these regions in SCD patients 428 . Previous work using hypercapnia challenges in the elderly has also suggested that blunted cerebrovascular responsiveness is ominous even in non-anemic subjects 429 , progressing to SCI even in otherwise normal appearing white matter 559 . 164 If regional CBF were completely constant throughout the hypoxic stimulus changes, one would expect variations in T½BOLD and TTPBOLD to be highly correlated. However, hypoxia could potentially produce blood flow redistribution to favor important structures. PET data suggests that phylogenetically older brain regions, preferentially supplied by the posterior circulation, are better protected and show higher increase in CBF in response to hypoxia than middle and anterior brain regions 560 . Anemia could be exaggerating this physiologic redistribution, explaining the shorter T½BOLD and prolonged TTPBOLD in the posterior circulation of anemic subjects compared to controls. Selective reduction of vascular resistance in the posterior circulation could create a physiological steal phenomenon 520 which would leave the anterior and middle cerebral artery territories with lower flow and longer T½BOLD post-hypoxia. Physiological steal could also explain the higher stroke rate observed the anterior circulation compared to posterior territory 561 . A large fraction of the NIRS and BOLD response to hypoxia could be explained by differences in pulse oximetry response to the nitrogen challenge. Since chronically anemic patients tended to have higher cardiac output and pulmonary blood flow to compensate for their decreased oxygen carrying capacity, oxygen washout during nitrogen inhalation in the lungs was increased, causing greater hemoglobin desaturation. The larger hemoglobin desaturation led to a larger effect size observed in SpO2 and BOLD hypoxic signal changes. This observation was corroborated by the positive association between baseline blood flow and DBOLD in our work and the correlation between the oximetry desaturation depth and hemoglobin in previous animal work 554 . Limitations A limitation to our study was the uncorrectable variations in the patient tidal volume and minute ventilation that influenced individual desaturation depth. Intra-subject variability in exposure made it difficult to interpret individual parameter maps. End-tidal CO2 sensors were built 165 into the breathing circuit, but technical difficulties prevented us from using these data. Previous work has shown that hematocrit affects the BOLD signal 433 , thus potentially influencing the parameter estimation in SCD and anemic subjects. The desaturation signal could have also been confounded by the lower signal-to-noise ratio within the vicinity of the SCI, which confounded the difference in hypoxic parameters observed between SCI-prone regions and normal appearing white matter. Additionally, even though multivariate regression analyses of global desaturation parameters yielded several classic risk factors for anemia and sickle cell disease severity, our study was underpowered to detect these correlations within individual subject groups. Furthermore, we could not standardize the hypoxic dosage with our current equipment since the nadir of the pulse oximetry measurement is reached 15-20 seconds after the patient is switched back to room air, so we could not use the pulse oximeter to titrate the hypoxia to consistent depths. It is likely that the use of a controlled respirator such as the RespirAct system (Toronto, ON, Canada) could achieve more consistent hypoxia exposures. In addition, the bolus timing parameters (T½, TTP) were primitive and indirect metrics of cerebral blood flow; the addition of post-processing tools designed for dynamic susceptibility contrast to our controlled hypoxia design could yield contrast-free voxel-wise mappings of relative cerebral blood volume, blood flow and mean transit time in the brain. Conclusion Our data suggested that SCI represented an iceberg phenomenon with respect to microvascular damage, namely that the presence of infarcts reflected much more extensive underlying microvascular remodeling. This observation was concordant with the 14-fold increase in relative stroke risk associated with SCI and raised the question of whether routine screening for SCI should be performed for sickle-cell patients 544 . Looking forward, the striking regional variations in hypoxic dynamics observed in this study warrants further imaging work to examine cerebral capillary transit time heterogeneity 37 in patients with SCD and other chronic anemias. 166 5.5. Supplemental Information 5.5.1. Supplemental Tables Supplemental Table 5.S1. Group average NIRS and BOLD desaturation parameters in the GM, WM and vascular territories (ACA, MCA and PCA). Bold letterings indicate statistical significance (p<0.05). CTL (N=31) SCD (N=26) ACTL (N=15) p- value (SCD vs. CTL) p-value (ACTL vs. CTL) p-value (SCD vs. ACTL) ΔBOLD GM (%) 7±2.3 9±2.9 9±2.7 <0.01 0.02 0.76 T½BOLD GM (s) 16±3.4 12±3.0 15±3.6 <0.01 0.36 0.06 TTPBOLD GM (s) 23±3.6 21±4.6 23±3.1 0.25 0.99 0.20 ΔBOLD WM (%) 3±1.2 5±1.7 5±1.7 <0.01 0.05 0.85 T½BOLD WM (s) 17±2.9 13±3.0 15±3.9 <0.01 0.31 0.04 TTPBOLD WM (s) 23±3.5 22±4.8 23±3.6 0.58 0.93 0.30 ΔBOLD ACA (%) 6±2.5 8±2.8 8±2.7 0.03 0.14 0.79 T½BOLD ACA (s) 16±4.3 12±2.8 15±3.9 <0.01 0.33 0.08 TTPBOLD ACA (s) 23±3.5 22±4.5 23±3.4 0.32 0.98 0.21 ΔBOLD MCA (%) 6±2.3 9±3.0 9±2.9 <0.01 0.03 0.84 T½BOLD MCA (s) 16±3.3 12±3.3 14±3.9 <0.01 0.17 0.24 TTPBOLD MCA (s) 24±3.7 21±4.5 23±3.6 0.24 0.99 0.18 ΔBOLD PCA (%) 6±2.1 8±2.7 8±2.6 <0.01 <0.01 0.80 T½BOLD PCA (s) 17±4.4 12±3.5 14±4.1 <0.01 0.09 0.16 TTPBOLD PCA (s) 23±3.8 21±5.4 23±4.3 0.27 0.88 0.46 167 5.5.2. Supplemental Figures Supplemental Figure 5.S1. Mean and standard deviation of BOLD hypoxia response in different subgroups. (A) Mean BOLD time series. (B) Standard deviation of BOLD hypoxic response. BOLD time series and range of one standard deviation in (C) CTL, (D) ACTL, (E) SCD SCI– and (F) SCI+ subgroups. 168 Supplemental Figure 5.S2. Group average maps of ΔBOLD, T½BOLD and TTPBOLD and root-mean-squared percent error (RMSPE) of the parameter estimation in healthy controls, ACTL, SCI– and SCI+ SCD subgroups. Supplemental Figure 5.S3. Example BOLD times series within two silent cerebral infarcts. 169 Chapter 6 : Oxygenation effects of hyperoxia challenge in sickle cell disease and chronic anemia 6.1. Introduction Congenital anemias, including sickle cell anemia and thalassemia, are some of the most common monogenic disorders in the world and lead to an array of clinical manifestations in many organ systems. 562 In the brain, chronic anemia is associated with inefficient oxygen delivery, tissue hypoxia and increased stroke risks. 535,536,563,564 Recent work using MRI tissue oximetry and arterial spin labeling suggest that deep white matter structures in watershed areas have inadequate oxygen delivery, 395 increased oxygen extraction fraction (OEF) 428 and impaired vasodilatory reserve. 565 In sickle cell disease (SCD) mouse models, the use of 100% oxygen gas challenge has been shown to identify chronically hypoxic brain regions. 566 The hyperoxic ventilation increases arterial oxygen content by approximately 1.1 mL O2/mL blood through increases in dissolved oxygen concentration. Although this is a modest effect in healthy non-anemic subjects, this addition would increase oxygen content by approximately 10% in anemic patients of hemoglobin of 7 g/dL, which could be beneficial in hypoxic brain regions of severe flow limitations. Therefore, to investigate for possible regional cerebral oxygen delivery impairment, we performed a hyperoxia challenge and used non-invasive near-infrared spectroscopy (NIRS) and Blood-Oxygen-Level-Dependent (BOLD) MRI acquisitions to measure hyperoxic responses in a cohort of healthy volunteers, patients with sickle cell anemia, and patients with non-sickle anemia. We hypothesized that hyperoxia would reveal brain regions with pathologic flow limitations and elevated peak extraction at baseline conditions by way of greater increase in BOLD signal 170 compared to adequately perfused, healthy tissue. We augmented this analysis with measurements of global oxygen delivery and metabolism under baseline and hyperoxic conditions to further shed light on the supply-demand mismatch and the brain’s compensation mechanism for the decreased oxygen carrying capacity in SCD and other anemias. 6.2. Methods 6.2.1. Study population All studies were performed at Children’s Hospital Los Angeles and approved by the Committee on Clinical Investigation (CCI#2011-0083). Informed consent or assent was obtained from all study participants. This study was performed in accordance with the Declaration of Helsinki of 1975. A total of 94 subjects were tested between April 2012 and February 2018. Only subjects older than 12 years of age were included in the study because of the challenge of complying with the respiratory apparatus. Imaging, vital signs and blood samples were obtained on the same day for each subject. The first group consisted of 38 SCD subjects: 34 subjects with SS hemoglobin and four subjects with SC hemoglobin. Exclusion criteria were prior overt stroke, pregnancy, acute chest or pain crisis hospitalization within one month and major medical problems outside of their chronic anemia. Asymptomatic iron overload was permitted. Out of 38 SCD subjects, 15 patients were receiving blood transfusions every three to four weeks to lower their hemoglobin S percentage to below 30%. The second group of anemic control (ACTL) subjects consisted of 25 individuals with non- sickle chronic anemia. This group was not race-matched to the SCD group, with eleven subjects with β thalassemia major, two subjects with thalassemia intermedia, three subjects with Eβ thalassemia, one subject with autoimmune hemolytic anemia, one subject with aplastic anemia, three subjects with hemoglobin H constant spring, one subject with congenital dyserythropoietic 171 anemia and three subjects with hereditary spherocytosis. Patients were excluded if they had any chronic organ dysfunction such as diabetes, abnormal renal function, or chronic active hepatitis. Asymptomatic iron overload was permitted. The three patients with spherocytosis had been splenectomized and had supranormal hemoglobin levels. Fifteen out of 25 ACTL subjects were receiving blood transfusions every three to four weeks. All chronically transfused patients (SCD and ACTL) were studied on the same day of their regularly scheduled blood transfusion, prior to receiving blood, when their hemoglobin levels were at the lowest. The third group of healthy volunteers consisted of 31 African and Hispanic American controls (CTL). Subjects were excluded if they had any poorly controlled chronic illnesses such as diabetes, hypertension or hepatitis. Since most subjects in this group were first or second- degree relatives of the SCD patients, sickle cell trait was common, occurring in half of the controls; these are subjects who had approximately 40% HbS in their blood but no cells containing exclusively HbS. Imaging and blood samples were obtained on the same day for each subject. All imaging was performed on a Philips Achieva 3T MR system with an eight-channel, receive-only head coil. Complete blood count, reticulocytes, quantitative hemoglobin electrophoresis and cell-free hemoglobin levels were analyzed in our clinical laboratory. Demographic and clinical variables of each subject group are summarized in Table 6.1. 6.2.2. Hyperoxia gas challenge The gas challenge setup has been previously reported 203 and is illustrated in Figure 6.1. At the start of the image acquisitions, patients were breathing through a two-liter reservoir rebreathing circuit supplied by pressurized, non-humidified room air at 12 liters per minute. This system included one-way valves to prevent partial gas mixtures and respiratory bellows (Invivo Corporation, Gainesville, FL) to display the breathing pattern and frequency. At 50 seconds into the data acquisition, the room air gas mixture was switched to 100% oxygen until the end of the 172 acquisition. Peripheral arterial oxygen saturation (SpO2) was acquired by fingertip pulse oximetry concurrently with forehead NIRS and BOLD MRI during the gas challenge. 6.2.3. Near-infrared spectroscopy (NIRS) A NIRS system (NIRO-200 system, Hamamatsu Photonics, Hamamatsu City, Japan) was used to measure relative changes in oxygenated hemoglobin (OxyHb), deoxygenated hemoglobin (DeoxyHb) and tissue oxygen index (TOI) of the mixed venous-weighted blood in the cerebral circulation continuously throughout the anatomical and challenge imaging exam (Figure 6.1). 539 The probes were placed on the subject’s forehead to record oxygenation in the watershed area between the anterior and medial cerebral arteries. 567 NIRS signals were acquired and synchronized via a Biopac MP150 data acquisition system (BioPac, Goleta, CA) oversampled at 1kHz. Figure 6.1. Experimental setup. Anatomical 3D T1 and T2-FLAIR, BOLD, T2-Relaxation-Under-Spin-Tagging (TRUST) and phase contrast scans were performed in all subjects under normoxia and hyperoxia conditions. BOLD imaging was performed under 50 seconds of normoxia and 4 minutes of hyperoxia to capture the cerebral hemodynamic response to 100% oxygen inspiration. Near-infrared spectroscopy (NIRS) and peripheral SpO2 were measured simultaneously with the BOLD acquisition during hyperoxia. 173 NIRS measurements were recorded concurrently during the hyperoxia challenge (Figure 6.1). A modified Beer-Lambert law was used to calculate the relative change in OxyHb, DeoxyHb, TOI, and total hemoglobin (TotalHb, equivalent to the sum of OxyHb and DeoxyHb). 6.2.4. Magnetic resonance imaging Each participant underwent an MRI study using a 3T Philips Achieva with an 8-element phased-array coil. Anatomical 3D T1, T2-Relaxation-Under-Spin-Tagging (TRUST) 568 to measure venous saturation (Yv) and phase contrast 543 to measure global cerebral blood flow (CBF) were performed in all subjects under normoxic condition. Following the last baseline imaging sequence, the hyperoxic ventilation commenced after 50 seconds into the five-minute BOLD acquisition. After BOLD imaging, repeated TRUST and phase contrast scans under hyperoxia were performed. The sequence of all MR acquisitions is illustrated in Figure 6.1. All MR acquisition parameters are listed below and were identical between normoxic and hyperoxic imaging. A 3D T1-weighted image was acquired covering the whole brain (160 sagittal slices) with TR = 8ms, TE = 4ms, flip angle = 8°, in-plane resolution = 1mm × 1mm, FOV = 256mm × 256mm and slice thickness = 1.0mm. To measure CBF, phase contrast MRI was obtained, positioned just above the carotid bifurcation: FOV = 260x260mm, TE = 7.5ms, slice thickness = 5mm, encoding velocity = 150cm/s, and 10 signal averages. To measure T2 relaxation of blood, TRUST was acquired at the superior sagittal sinus with the following parameters: CPMG T2 preparation with effective echo times of 0, 40, 80 and 160ms, TR = 1978ms, delay time = 1022ms, resolution = 3.4×3.4×5mm and FOV = 220×220×5mm. BOLD images were acquired using gradient-echo echo-planar-imaging sequence: TR = 2000ms, TE = 50ms, in-plane resolution = 2.3×2.3mm, FOV = 220×220mm, 26 axial slices, slice thickness = 5mm, SENSE factor of 2. A total of 150 volumes were collected in five minutes. All MR acquisition parameters have been detailed in prior publications. 285,395,406 174 6.2.5. Definitions of NIRS and BOLD changes in response to hyperoxia The following definitions of hypoxic parameters were provided for DeoxyHb but without loss of generality were used to compute response dynamics for OxyHb, TotalHb, TOI and BOLD time series: (1) DeoxyHbbaseline was the mean DeoxyHb signal before hyperoxic gas, calculated from the start of the BOLD acquisition to the start of oxygen gas administration at 50 seconds. (2) DeoxyHbhyperoxia was the mean signal after the signal has stabilized during hyperoxia, calculated from 120 seconds to the end of the BOLD acquisition. (3) ∆DeoxyHb (the depth of the hyperoxic change, µmol for NIRS and % for BOLD changes) was calculated as ∆DeoxyHb = |DeoxyHbbaseline – DeoxyHbhyperoxia|. Pictorial illustrations of these definitions can be found in Figure 6.2. 6.2.6. Image processing Structural T1 images were registered to the Montreal Neurological Institute (MNI) template with the FMRIB Software Library (FSL); BOLD images were preprocessed with FSL and Figure 6.2. Sustained hyperoxia model signals. Representative recordings of (A) global BOLD, SpO2, tissue oxygen TOI, (B) deoxygenated hemoglobin DeoxyHb, oxygenated hemoglobin OxyHb and total hemoglobin TotalHb changes during the hyperoxia challenge. 175 registered to MNI template space using a standard spatial functional pipeline. 406 Global ΔBOLD values were computed as an average within a whole-brain mask derived from an average of 152 T1-weighted MRI scans in the common MNI coordinate system. 413 MNI-based grey matter mask, an eroded white matter mask and three vascular territories masks were used to calculate average ΔBOLD for these different brain regions. Masks of the anterior cerebral artery, middle cerebral artery and posterior cerebral artery circulations were based on the published templates of vascular territories in both hemispheres. 414 The territories were created from anatomic studies of cerebral vascularization and evaluated on the bicommissural plane. Whole-brain CBF was measured using phase contrast MRI, 543 and phase contrast images were analyzed using an in-house MATLAB program (Mathworks, Natick, MA). Stationary tissue pixels were identified in the complex difference images by simple thresholding using a mean plus two standard deviations from a remote, nonvascular region. Vessel boundaries were semi- automatically segmented from complex difference images using Canny edge detection and mapped to the phase difference image for flow calculation. Total CBF, which was the sum of the flow in left and right interior carotid arteries and vertebral arteries, was corrected for brain size assuming a brain density of 1.05 g/mL. For the TRUST acquisition, the difference signals between tag and control images of the superior sagittal sinus cross sections were fitted to a mono-exponential decay equation ∆S=∆S 0 e eTE×(1 T 1b ⁄ –1/T 2b ) , where eTE = 0, 40, 80 and 160ms, T1b and T2b are the relaxation parameters of blood. T1b was estimated from the hematocrit in patients with normal hemoglobin or as a value of 1818ms for patients with sickle. 199 T2b was derived from nonlinear least squares fit to the signal decay equation. Afterward, T2b and hematocrit were used to estimate venous saturation from a previously-published calibration curve for non-sickle 350 and SCD blood. 569 176 6.2.7. Physiological parameters Several physiological parameters were derived using the following equations: ! " = 1.34 × × ! + 0.003 × ! " (mL O2/mL blood) [1] # " = 1.34 × × # + 0.003 × # " (mL O2/mL blood) [2] = 1− % ! & " % # & " [3] " = × ! " (mL/100g/min) [4] " = × ( ! " − # " ) (mL O2/100g/min) [5] where PaO2 is the arterial partial pressure of oxygen estimated to be 100 mmHg at normoxia and 400 mmHg at hyperoxia, PvO2 is the venous partial pressure of oxygen estimated to be 40 mmHg at normoxia and 52 mmHg at hyperoxia, 570 CaO2 is the arterial oxygen content, CvO2 is the venous oxygen content, Hb is the hemoglobin level of each subject, Yv is the venous saturation measured by TRUST, Ya is the arterial saturation measured by pulse oximetry and OEF is the cerebral oxygen extraction fraction. 6.2.8. Statistical analysis Statistical analysis was performed in JMP (SAS, Cary, NC). One-way ANOVA with Dunnett’s post hoc correction was used to examine the difference in clinical and oxygenation variables between SCD and control, and between ACTL and control; independent samples t-test was used to compare SCD and ACTL groups. Univariate and stepwise multivariate regressions were performed against hemoglobin, HbS, transfusion status, fetal hemoglobin, white blood cells, platelets, reticulocytes and cell-free hemoglobin; multivariate predictors were retained in the final model for p<0.05. Two-sample t-maps were generated with Analysis Functional NeuroImages (AFNI) 508 to compare desaturation parameters between controls and SCD or ACTL patients; each voxel’s t-statistic was calculated by 177 = ( ^%3/<%42 – %42 ) ( h ¤ 1 ^%3/<%42 + 1 %42 ) M with ^%3/<%42 and %42 as group sizes, ^%3/<%42 and %42 as parameter means, ^%3/<%42 and %42 as parameter standard deviations and h as the pooled standard deviation h =¤ ^%3/<%42 –1 ^%3/<%42 " +( %42 –1) %42 " ^%3/<%42 + %42 −2 Age and sex regressions were performed in JMP, and the average was added to the regressed values to ensure final values were within physiologically meaningful range. 6.3. Results 6.3.1. Demographics The demographics for this study population is summarized in Table 6.1. Since the SCD group trended younger and had more male subjects compared to controls, all subsequent analyses were corrected for age and sex. There was no significant difference between sickle-cell trait and non-trait controls in NIRS (p=0.74 for ΔOxyHb, p=0.95 for ΔDeoxyHb, p=0.54 for ΔTOI), global ΔBOLD (p=0.52), CBF (p=0.76), or CMRO2 (p=0.28). Thus, trait and non-trait control data was pooled. Pre- and post-hyperoxia CBF, arterial and cerebral venous saturations were recorded in all patients in this study. All patients were included in the oxygen delivery and metabolism analysis. Of the 94 participating subjects, only 76 subjects were included in the BOLD analysis; the rest were excluded due to severe subject motion or incomplete BOLD acquisition. The NIRS analysis included 21 SCD subjects, 13 ACTL subjects and 20 controls, and the rest were excluded due to sensor and data acquisition system malfunctions. There were no differences in the demographic and hematologic data between the complete cohort presented in Table 6.1 and the cohort after subject exclusion in the BOLD and NIRS analyses. 178 Table 6.1. Patient demographic and hematologic data. Bold letterings indicate statistical significance (p<0.05). CTL (N=31) SCD (N=38) ACTL (N=25) p-value (SCD vs. CTL) p-value (ACTL vs. CTL) p-value (SCD vs. ACTL) Age (Years) 25.2±10.1 22.4±7.6 25.4±10.7 0.37 0.99 0.22 Sex 10M, 21F 23M, 15F 12M, 13F 0.04 0.39 0.33 Transfusion 0/31 15/38 15/25 <0.01 <0.01 0.05 Arterial Saturation 99±1 98±2 99±1 <0.01 0.15 0.03 Hemoglobin Electrophoresis 15AA, 16AS 34SS, 4SC 21AA, 3AE, 1AS Hemoglobin (g/dL) 13.6±1.2 9.7±1.8 10.6±2.7 <0.01 <0.01 0.05 Hematocrit (%) 40.2±3.3 27.5±4.6 32.3±6.4 <0.01 <0.01 <0.01 White Blood Cell Count (x10 3 ) 5.7±1.9 9.7±4.4 7.0±2.3 <0.01 0.23 <0.01 Platelet Count (x10 3 /µL) 244±52 307±113 260±119 0.02 0.75 0.07 Reticulocytes (%) 1.5±0.6 9.6±5.0 2.7±2.9 <0.01 0.26 <0.01 S-cells Fraction (%) 0 52.0±31.0 0 <0.01 1.00 <0.01 Fetal Hemoglobin (%) 0.5±2.1 8.4±8.8 2.2±4.2 <0.01 0.50 <0.01 Cell-free Hemoglobin (%) 6.5±5.1 22.6±21.4 17.5±18.6 <0.01 0.04 0.24 6.3.2. Hematologic markers The SCD and ACTL groups had significantly lower hemoglobin levels (p<0.01), lower hematocrit (p<0.01) and higher cell-free hemoglobin (p<0.01) compared to healthy controls. Within the two groups of anemic subjects, SCD patients had higher reticulocytes (p<0.01), white blood cells count (p<0.01), fetal hemoglobin (p<0.01) and fraction of S-cells (p<0.01). Of all the participating ACTL subjects, 60% subjects were transfused whereas only 40% of SCD patients received regular transfusion; the transfused SCD subjects had a median HbS percentage of 179 19.7%, significantly lower than 76.8% in the non-transfused subgroup. Pulse oximetry values were lower in the SCD patients than control subjects at baseline (p<0.01). 6.3.3. NIRS and BOLD response to hyperoxia None of the patients consciously perceived the hyperoxia episode, and no complications were encountered. The gas challenge produced a maximum S pO2 values of 100%. Compared to ΔSpO2 in healthy controls (1.1±0.7%), ΔSpO2 was greater in SCD (2.4±1.6%, p<0.01) and in ACTL subjects (1.8±0.7%, p=0.06). This translated to increases in O2 content of 1.1, 1.2, and 1.2 ml O2/ml blood for control, ACTL, and SCD respectively. Hyperoxia significantly increased the OxyHb concentration (p<0.01) and tissue oxygenation index (p<0.01) in the entire cohort, while DeoxyHb and total hemoglobin level decreased (both p<0.01). A summary of NIRS signal changes in each subject group is shown in Table 6.2. SCD subjects showed larger magnitude of change in OxyHb (p=0.01) and TOI (p<0.01) compared to controls, but there was no significant difference in the hyperoxic effect on TotalHb between the three groups (p=0.95). All NIRS changes were correlated with ΔSpO2 (r 2 =0.14, p<0.01 for OxyHb; r 2 =0.09, p=0.03 for OxyHb; r 2 =0.12, p<0.01 for TotalHb; r 2 =0.07, p=0.06 for TOI). In addition to the NIRS signal, global BOLD signal was significantly increased during hyperoxic ventilation (p<0.01) and correlated with ΔSpO2 (Figure 6.3A). Global and regional ΔBOLD trended higher in SCD subjects compared to healthy controls, but the difference was not statistically significant (Table 6.2). Additionally, ΔBOLD was correlated with changes in tissue oxygenation, with 26% of the variation in ΔBOLD explained by TOI (Figure 6.3B). The BOLD signal varied with both ΔOxyHb and ΔDeoxyHb (Figure 6.3C and 6.3D). Even though global ΔBOLD was not correlated with total hemoglobin level (p=0.38), grey matter ΔBOLD had a significant association with TotalHb (r 2 =0.10, p=0.02), whereas white matter ΔBOLD did not (p=0.07). 180 Overall, grey matter displayed a change of 3.5±1.5% in the BOLD signal in response to hyperoxic ventilation, significantly larger compared to 1.1±0.9% in the white matter (p<0.01). This pronounced grey-white matter differentiation is demonstrated by the group average Δ BOLD maps in Figure 6.4. Deep white matter structures did not demonstrate a different hyperoxic effect compared to gyral or more superficial white matter, and there was no significant difference in ΔBOLD within the anterior, middle and posterior cerebral artery perfusion territories (p=0.14). Voxelwise comparison of ΔBOLD between SCD and ACTL subjects and control subjects failed to identify any systematic differences after correction for multiple comparisons. Figure 6.3. Correlation between changes in BOLD, NIRS and pulse oximetry signals. Correlation between ΔBOLD and (A) pulse oximetry ΔSpO2, (B) tissue oxygenation ΔTOI, (C) oxygenated hemoglobin ΔOxyHb, and (D) deoxygenated hemoglobin ΔDeoxyHb. 181 Table 6.2. Group average of whole-brain and regional BOLD and NIRS changes due to hyperoxia. Bold letterings indicate statistical significance (p<0.05). GM = grey matter; WM = white matter; ACA = anterior cerebral artery; MCA = middle cerebral artery; PCA = posterior cerebral artery. CTL SCD ACTL p-value (SCD vs. CTL) p-value (ACTL vs. CTL) p-value (SCD vs. ACTL) ΔDeoxyHb (µmol) –0.83±0.27 –1.10±0.62 –0.64±0.25 0.10 0.41 <0.01 ΔOxyHb (µmol) 0.67±0.44 1.08±0.56 0.48±0.21 0.01 0.40 <0.01 ΔTotalHb (µmol) –0.15±0.33 –0.19±0.44 –0.16±0.15 0.93 0.99 0.84 ΔTOI (%) 2.7±1.1 4.3±2.2 2.3±1.0 <0.01 0.70 <0.01 Global ΔBOLD (%) 3.0±1.0 3.6±1.7 3.1±1.2 0.22 0.96 0.26 GM ΔBOLD (%) 3.3±1.1 3.7±1.7 3.6±1.5 0.58 0.77 0.87 WM ΔBOLD (%) 1.0±0.6 1.2±1.0 1.1±1.0 0.53 0.85 0.71 ACA ΔBOLD (%) 3.9±1.4 3.5±2.3 3.7±1.8 0.57 0.90 0.68 MCA ΔBOLD (%) 3.7±1.3 4.1±1.9 3.7±1.3 0.62 0.99 0.45 PCA ΔBOLD (%) 4.0±1.5 4.3±1.9 4.8±3.8 0.77 0.39 0.50 Figure 6.4. Group ΔBOLD response to hyperoxia for sickle cell disease (SCD), non-sickle anemic (ACTL) and controls (CTL). T-scores between SCD and ACTL groups versus control subjects failed to identify any systematically different regional behavior (not shown). 182 6.3.4. Oxygenation parameters in response to hyperoxia Table 6.3 shows the effect of the hyperoxia challenge on the cerebral oxygen supply and demand in the three subject groups. Anemic subjects had lower arterial and venous oxygen content during both normoxic and hyperoxic ventilation compared to healthy controls (p<0.01 for both conditions). Even though CBF was higher in both anemic cohorts (p<0.01 for both normoxia and hyperoxia), there was no significant difference in O2 delivery in the three groups during normoxia (p=0.20) and hyperoxia (p=0.58). Resting OEF varied negatively with resting CBF (r 2 =0.31, p<0.01) and positively with hemoglobin (r 2 =0.20, p<0.01), resulting in significant baseline differences across the three groups (lowest in SCD). OEF in the two anemic cohorts became closer to values in CTL subjects with hyperoxia, although OEF remained significantly lower in SCD subjects. CMRO2 results roughly paralleled OEF results (r 2 =0.40, p<0.01). ACTL subjects (p<0.01) and SCD subjects (p<0.01) both exhibited lower baseline CMRO2 compared to controls; this disparity was less pronounced under hyperoxia (p=0.06 for ACTL and p<0.01 for SCD). Table 6.4 contrasts the absolute and relative changes across groups for all the brain metabolism parameters. Although the absolute increase in arterial oxygen content was identical across groups, the percentage increase was 43% – 67% higher in anemic subjects (p<0.01). In contrast, both the absolute increase (p=0.38) and percentage increase (p=0.63) in venous oxygen content was similar across groups. The absolute decline in CBF was highly correlated with initial CBF (r 2 =0.43, p<0.01) and was greatest in SCD patients (p<0.01 compared to controls). However, the percentage change in CBF was independent of disease state (p=0.08), averaging 22.1%. Since hyperoxia triggered a larger percentage reduction in CBF compare with the increase in oxygen content (5.6% – 9.4%), cerebral oxygen delivery fell in all three cohorts (p<0.01). Neither the absolute decrease (p=0.31) nor the percentage decrease (p=0.48) in oxygen delivery were different across groups. The change in OEF varied inversely with initial OEF (r 2 =0.11, p<0.01). However, while both the absolute and relative change in OEF trended higher in anemic subjects, the differences did not reach statistical significance (p=0.09 for absolute and p=0.09 for relative 183 changes). Interestingly, if transfusion status was taken into account, transfused SCD subjects showed an OEF increase in response to hyperoxia (p<0.01) but non-transfused SCD did not (p=0.64); this effect was not observed between transfused and non-transfused ACTL patients. The change in CMRO2 was negatively correlated with baseline CMRO 2 (r 2 =0.36, p<0.01). However, both the absolute changes (p=0.28) and the percentage changes (p=0.57) in CMRO2 were similar across groups. Resting O2 delivery was the strongest predictor of the brain’s metabolic rate, explaining approximately 36% of the variation in CMRO2. After regressing out the differences in O2 delivery, CMRO2 remained directly proportional to hemoglobin (p<0.01) and fetal hemoglobin (p<0.01) with a combined r 2 =0.61. Under hyperoxic ventilation, CMRO2 was independently predicted by hyperoxic O2 delivery and fetal hemoglobin with a combined r 2 =0.40. The change in metabolic rate between normoxia and hyperoxia was not different across the three patient groups (p=0.58) and was most closely associated by the change in oxygen delivery followed by hemoglobin and platelet count (combined r 2 =0.38). 184 Table 6.3. Table of oxygen supply and utilization parameters under normoxia and hyperoxia. Paired t-test was used to compare values in room air and hyperoxia conditions. Values are expressed as mean ± standard deviation. Normoxia Hyperoxia p-value CTL Arterial O2 Content (mL O2/mL blood) 18.4±1.7 19.4±1.7 <0.01 Venous O2 Content (mL O2/mL blood) 11.6±1.3 12.4±1.8 <0.01 CBF (mL/100g/min) 62.9±9.0 50.0±8.1 <0.01 O2 Delivery (mL/100g/min) 11.5±1.8 9.7±1.6 <0.01 Oxygen Extraction Fraction (%) 36.8±3.9 36.1±5.8 0.51 CMRO2 (mL O2/100g/min) 4.2±0.6 3.5±0.6 <0.01 SCD Arterial O2 Content (mL O2/mL blood) 13.0±2.5 14.2±2.4 <0.01 Venous O2 Content (mL O2/mL blood) 9.4±2.0 10.0±2.1 <0.01 CBF (mL/100g/min) 98.3±26.8 72.2±22.3 <0.01 O2 Delivery (mL/100g/min) 12.4±2.9 10.0±2.7 <0.01 Oxygen Extraction Fraction (%) 27.7±5.2 29.5±5.7 0.02 CMRO2 (mL O2/100g/min) 3.4±0.9 2.8±0.6 <0.01 ACTL Arterial O2 Content (mL O2/mL blood) 14.4±3.6 15.5±3.6 <0.01 Venous O2 Content (mL O2/mL blood) 9.8±2.2 10.3±2.4 <0.01 CBF (mL/100g/min) 83.0±17.1 62.9±14.5 <0.01 O2 Delivery (mL/100g/min) 11.4±1.9 9.4±1.8 <0.01 Oxygen Extraction Fraction (%) 31.6±5.1 33.5±6.8 0.10 CMRO2 (mL O2/100g/min) 3.6±0.6 3.1±0.6 <0.01 185 Table 6.4. Absolute and relative changes in cerebral metabolism parameters across groups. Values are expressed as mean ± standard deviation. Bold values are significant relative to control subjects at p<0.05. Parameter CTL SCD ACTL Absolute Percentage Absolute Percentage Absolute Percentage Arterial O2 Content (mLO2/mL blood) 1.1±0.2 5.6±1.1 1.2±0.2 9.4 ± 2.9 1.1 ± 0.2 8.0 ± 2.5 Venous O2 Content (mLO2/mL blood) 0.8±1.1 6.9±9.6 0.6±0.7 6.8±7.6 0.5±0.8 5.0±8.0 CBF (mL/100g/min) –12.9±7.6 –20.0±10.3 –26.1±15.5 26.2±12.5 –20.1±9.7 -24.0±9.4 O2 Delivery (mL/100g/min) –1.9±1.4 –15.6±10.7 –2.4±1.8 –19.2±14.2 –2.1±1.3 –17.9±10.4 OEF (%) –0.6±5.7 –1.3±15.8 1.8±4.6 9.3±23.2 1.9±5.6 6.9±20.4 CMRO2 (mLO2/100g/min) –0.7±0.7 –16.9±16.8 –0.5±0.6 –13±17.8 –0.5±0.7 –12.2±20.9 186 6.4. Discussion The purpose of the present study was to determine whether we could use 100% oxygen as a contrast agent to identify brain regions having increased oxygen extraction secondary to flow limitation, as has successfully been performed in sickle cell mouse models. 566 We postulated that chronically hypoxic brain regions would demonstrate greater rise in BOLD signal than better perfused neighboring tissue. In this study, we evaluated the oxygen supply and utilization in the brain in response to brief hyperoxic ventilation in chronically anemic patients and healthy controls. We observed higher BOLD signal and NIRS tissue oxygenation, corresponding with the increased oxygen content during hyperoxia. BOLD signal changes were more than 3-fold higher in grey matter than white matter, consistent with known relative differences in cerebral blood volume. 394 However, we failed to identify any systemic regional differences in BOLD signal changes across groups, suggesting that 100% oxygen could not be used to identify chronically ischemic brain. The failure to replicate the results from the mouse model most likely stems from the disease severity observed in sickle cell anemia mouse models, 566 in which oxygen regulation was so heavily impaired that hyperoxia elicited a paradoxical increase in CBF. 571 In contrast, our anemic and SCD subjects were relatively healthy and demonstrated the expected vasoconstrictive response to hyperoxia on the same magnitude as healthy controls. Since our patients did not demonstrate vasculopathy and large vessel disease, they likely had less severe flow limitation and pathologically high OEF in the watershed regions compared to other SCD cohorts, 428 thus hyperoxic response was similar between deep and more superficial white matter. Furthermore, it is possible that white matter BOLD changes follow the grey matter response because white matter tissues are downstream from grey matter in the cerebral perfusion chain. Previous work has shown that tissue oxygenation improvement due to hyperoxia is small in hypoperfused areas such as the white matter compared to cortical regions, 572 so it is possible that the improvement to OEF in flow-limited was too small to observe in our study. Additionally, 187 differential effects of flow could potentially confound our regional BOLD signal, 573 obscuring the saturation-driven changes in hypoxic tissue during the hyperoxic ventilation. The effect of supplemental oxygen therapy to increase cerebral partial pressure of oxygen was clearly observed in our NIRS measurements, with an average of 3% boost in tissue oxygenation index in the entire cohort. The larger TOI effect size observed in anemic patients reflects the proportionally larger increase in oxygen content in anemic subjects. 570 NIRS and BOLD changes with hyperoxia were correlated, 549 providing inter-modality corroboration of our results. However, in contrast to the hypoxia gas challenge in which the strongest correlation was observed between ΔBOLD and ΔDeoxyHb, 574 the hyperoxic BOLD changes was best correlated with ΔOxyHb demonstrating that hemoglobin saturation was the strongest influence of the BOLD signal rather than blood flow changes. In fact, ΔTotalHb was inversely proportional to the BOLD signal. This observation supports the use of hyperoxia in calibrated fMRI, in which the increase arterial saturation rather than hyperemia is the main driving force of the changes in capillary and venous saturation observed on quantitative BOLD imaging. 575 MRI and NIRS evidence of improved oxygenation was accompanied by powerful cerebral vasoconstriction. The vasoconstriction effect of hyperoxic ventilation was reflected by both decreased NIRS TotalHb signal and phase contrast CBF. The 20-25% decrease we observed was concordant with previous works 571,576 from independent groups. Hyperoxia with 100% oxygen administration has previously been shown to produce a ~3-7% decrease in end-tidal CO2, but hypocapnia of this magnitude cannot explain a 20-25% reduction in CBF. 571,576 Therefore, the vasoconstriction in our results could be attributed mostly to hyperoxia, 577 which attenuates the effective nitric-oxide concentration and leads to an increase of reactive oxygen species in the brain. 578 Improved oxygen content with hyperoxia was more than counterbalanced by the reduction in CBF. As a result, cerebral oxygen delivery declined an average of 17.7%. The decreased delivery contradicted several previous studies which found no significant change in oxygen 188 delivery. 570,579,580 However, these studies had much smaller cohort sizes compared to the 94 subjects in this current work. In contrast, recent work by Bodetoft et al. has reported a significant decrease in systemic and coronary oxygen delivery of 4-11%, independent of arterial carbon dioxide pressure. 581 Decreased oxygen delivery was associated with a parallel reduction in CMRO 2 during hyperoxia. This observation is concordant with other reports of lower metabolic rate in both the systemic 29,47 and cerebral circulations with hyperoxia. 48 The changes in oxygen delivery and metabolic rate were well-correlated, but it would be difficult to decipher whether the CMRO2 reduction was a primary physiological response, triggering a subsequent reduction in O2 delivery, or a secondary response to reduced oxygen delivery. However, given the rise in BOLD signal and NIRS tissue oxygenation index with hyperoxia, the former possibility seems more likely. Hyperoxia, often in the form of 100% oxygen therapy, has traditionally been a transport and emergency room treatment for conditions related to poor oxygen delivery to vital organs, such as during stroke or myocardial infarction. However, subsequent studies demonstrating neutral or negative benefits have led to limiting oxygen therapy to the correction of hypoxia. 584–586 The reduction in oxygen delivery and CMRO2 observed in this study support that change in practice, although it is unclear whether similar responses would be observed in at risk tissues. Our study had some important limitations. Since we were interested in a potentially clinically useful diagnostic, we did not attempt to “clamp” exhaled CO2 levels. End-tidal CO2 sensors were built into the breathing circuit, but technical difficulties prevented us from using these data. This study also assumed the blood pO2 levels 570 instead of measuring it in individual patient using a radial arterial catheter. However, a 10% variation in pO2 only leads to less than 1% change in oxygen delivery and CMRO2 without changing our results. Additionally, our work only evaluated oxygen metabolism after very brief administration of 100% oxygen. Since vasoconstriction effects in hyperoxia has been shown to be reversible, 570,582 and oxygen toxicity does not manifest until several days under sustained oxygen therapy, further assessment of 189 oxygen utilization is required to determine whether adaptation and normalization of CMRO2 occurs under prolonged hyperoxic ventilation. In summary, this work evaluated the oxygen delivery and consumption in the brain during normoxic and hyperoxic ventilation. From our results, we were unable to identify brain regions of flow limitation in chronically anemic patients. However, we were able to demonstrate higher tissue oxygenation as well as arterial and venous oxygen content during hyperoxia. Even though chronic anemia is associated with hyperemia at baseline, this degree of vasodilation in SCD and anemic subjects was not enough to compensate for the hyperoxia-induced vasoconstriction in response to abnormally high pO2, leading to diminished oxygen delivery and metabolic rate. Both controls and chronically anemic subjects showed proportional CMRO2 decline compared to baseline, but the hyperoxic CMRO2 values in SCD and non-sickle anemic subjects were abnormally low in our studies. Therefore, further work is needed to evaluate the risk of prolonged oxygen therapy, especially in anemic populations at risk for oxygen toxicity and cerebral dysfunction. 190 Chapter 7 : Calibration of T 2 oximetry MRI for subjects with sickle cell disease 7.1. Introduction Cerebral MRI oximetry is an approach to measure venous oxygenation based on the principle that oxyhemoglobin experiences a shift in magnetic susceptibility when transitioning to deoxyhemoglobin. Amongst different MRI oximetry techniques, T2-relaxation-under-spin-tagging (TRUST) uses arterial spin labeling to isolate venous blood, applies T2 preparation and measures blood T2 signal in a large cerebral vein. 568 This technique has recently been applied by different groups to estimate oxygen extraction fraction and cerebral metabolic rate of oxygen in sickle cell disease (SCD) subjects. 149,285,343,587,588 However, since T2 oximetry relies on a calibration model to convert blood T2 into oxygen saturation, the estimates are heavily dependent on the choice of empirical calibration model, whether derived from bovine blood, 568 healthy human blood 350 or human blood from patients with SCD. 285,343 In particular, the non-sickle calibration curves yield much higher estimates of oxygen extraction fraction and cerebral metabolic rate than calibration curves derived from patients with SCD. Compared to calibrations derived from bovine and sickle blood, Li et al. demonstrated better accuracy in measuring cerebral oxygenation in SCD patients using each subject’s individual calibration. 343 While performing individual calibrations would be ideal, such an approach is impractical for clinical use or even many research applications. Since the calibration methodologies are challenging and time-consuming, prior sickle cell blood calibration studies suffered from small sample size and limited power. The current work combines raw calibration data from two independent reports by Bush et al. 285 and Li et al. 343 to establish a unified, robust 191 T2 calibration for SCD patients. It also presents a method to compensate for the admixture of normal and sickle hemoglobin observed in transfused patients with SCD, who make up approximately 10-20% of the SCD population. 7.2. Methods The Committee on Clinical Investigation at Children’s Hospital Los Angeles (CHLA) approved the protocol; written informed consent and/or assent were obtained from all subjects (CCI#2011-0083). This study was performed in accordance with the Declaration of Helsinki. 7.2.1. T2 oximetry calibration model The sequence to measure venous T2 has been explained in prior publications. 285,348,350 Briefly, a transverse, single-slice T2 preparation sequence that uses Carr-Purcell-Meiboom-Gill (CPMG) T2 weighting was used to measured blood T 2 in the sagittal sinus. The number of refocusing pulses was varied across four acquisitions (0, 4, 8 and 16 pulses) to acquire images at effective echo times (eTE) of 0ms, 40ms, 80ms and 160ms. Unlike in vivo TRUST acquisitions, since ex vivo imaging did not involve flowing blood, the magnetic tagging module is turned off. The signal at four eTE were fit to the mono-exponential decay equation: S(eTE) = S 8 × e 6 IJK J " [1] where S is the signal measured, S0 is a constant across echo times and T2 is the transverse relaxation times of blood. Equation 1 did not include the blood longitudinal relaxation time T1 since the magnetic spin labeling is off. The relationship between venous blood relaxation and oxygenation is illustrated with the empirical model: " = ' 4 " = × (1 − ) " + [2] 192 where R2 is the measured T2 relaxation rate, Y is saturation, A and B are empirically determined coefficients from least-square fitting. We previously observed that both A and B coefficients varied linearly with hematocrit: = ' × + " [3] = ' × + " [4] where a1, a2, b1 and b2 are coefficients for linear dependence on hematocrit. Equations 2-4 are consistent with our prior calibration studies. 285,343,350 When formulating this model, care was taken to convert apparent T2 (T2,app) from the Bush study 285 to T2 corrected for pulse width (T2,corr) with ",*,.. = ",!hh × where =1−_ℎ (2 × ℎ_) ⁄ =0.913 with pulse width of 1.74ms and CPMG interecho spacing of 10ms. 343,589,590 7.2.2. Raw calibration data Raw T2 oximetry calibration data and individual hematologic measurements were extracted from Bush et al. 285 (N=11) and Li et al. 343 (N=11). Each individual dataset consisted of the subject’s hematocrit, percentage of hemoglobin S (HbS), a set of venous saturation measurements by blood gas co-oximetry and corresponding measured venous R2. Briefly, this set of saturation values and venous R2 was determined from a blood sample drawn from the antecubital vein from each participant and was used to construct individual T2 oximetry calibration at the native hematocrit. For each point on the R2–Y empirical model in Equation 2, the blood sample underwent oxygenation adjustment by exposure to room air or nitrogen gas and was equilibrated at 37˚C; each sample’s saturation was measured with co- oximeter and corresponding blood R2 was measured with TRUST MRI acquisition. This process 193 was repeated 5-7 times in each sample to provide the R2–Y relationship, and least squares fitting of this relationship with Equation 2 yielded A and B coefficients for each subject. 7.2.3. Sickle calibration for non-transfused blood To have the most robust SCD calibration, only non-transfused patients with hemoglobin SS disease were extracted from the Bush study (N=5) and pooled with similar patients from the Li study (N=11), resulting in a cohort with high HbS (80±12%, N=16, Table 7.1). Patients with hemoglobin SC and S+ were eliminated from analysis. A population calibration model based on Equation 2 was derived for this non-transfused SCD group. Table 7.1. Subject demographics and hematologic parameters. Subject Hematocrit (%) HbS (%) A B Non-transfused Li #1 18.0 91.0 51.7 7.8 Li #2 34.0 79.0 98.1 7.2 Li #3 27.0 84.8 54.4 4.7 Li #4 25.0 86.5 46.2 7.6 Li #5 25.0 68.4 72.7 9.2 Li #6 26.0 78.2 63.9 6.9 Li #7 28.0 N/A 68.0 6.4 Li #8 28.0 85.2 102.7 6.8 Li #9 23.0 80.4 78.5 6.2 Li #10 19.0 86.8 64.7 7.7 Li #11 23.0 86.4 67.7 7.0 Bush #1 35.2 64.1 84.2 5.7 Bush #2 33.4 69.3 80.2 5.4 Bush #3 36.0 95.3 76.2 7.0 Bush #4 24.0 85.0 69.4 5.8 Bush #5 24.0 93.2 70.1 5.8 Transfused Bush #6 30.3 24.5 73.7 6.6 Bush #7 28.9 39.4 68.7 5.5 Bush #8 37.0 45.9 75.5 5.4 7.2.4. Sickle calibration for transfused blood In the case of hyper-transfused patients with lower HbS, we hypothesized that a different HbS-based mixture calibration model, representing a linear mixture of HbA and HbS calibrations, 194 would be more appropriate for SCD patients who receive chronic transfusion. The accuracy of this transfusion-specific calibration was assessed on three SCD patients (N=3, including 2 hemoglobin SS and 1 hemoglobin Sβ0) from Bush et al. 285 who were undergoing chronic transfusion to maintain HbS less than 30%. The mixture model provided a weighted calibration between the sickle calibration (current Li-Bush calibration) and normal hemoglobin calibration, 350 namely the model coefficients would be weighted by individual HbS percentage: ' =t1− T0^% u8% u× ',T0< + t T0^% u8% u × ',T0^ [5] " =t1− T0^% u8% u× ",T0< + t T0^% u8% u × ",T0^ [6] ' =t1− T0^% u8% u× ',T0< + t T0^% u8% u × ',T0^ [7] " =t1− T0^% u8% u× ",T0< + t T0^% u8% u × ",T0^ [8] where HbS% of each subject is the model weight, a1,HbA, a2,HbA, b1,HbA and b2,HbA are coefficients from calibration 350 for normal blood and a1,HbS, a2,HbS, b1,HbS and b2,HbS are coefficients from Li-Bush calibration for sickle blood. Since the pure sickle Li-Bush model is derived from non-transfused blood with an average HbS% of 80%, the HbS model weights are normalized by 80% instead of 100%. Summary of calibration models in this manuscript is shown in Supporting Information Table 7.S1. 7.2.5. Statistical analysis Statistical analysis was performed in JMP (SAS, Cary, NC). Bland-Altman analyses were performed to assess agreement between different calibration models and blood-gas co-oximetry. 195 7.3. Results 7.3.1. Non-transfused population SCD calibration Figure 7.1 demonstrates the linear dependence of the A and B coefficients on hematocrit levels; open circles represent data from the Bush study and closed circles are from the Li study. Hematocrit explains 44% of the variance in A and 10% of the variance in B, whereas neither A nor B demonstrate a significant dependence on hemoglobin S concentrations (p=0.22 and p=0.91 respectively). Confidence limits and standard error of the parameter estimates are illustrated in Supporting Information Figures 7.S1 and 7.S2. The combined data have double the number of points at hematocrits greater than 0.3, stabilizing the linear estimation. From this bilinear dependence of blood R2 on both saturation and hematocrit, the Li-Bush population calibration model for non-transfused S hemoglobin patients was formulated: ",*,.. =(196.8 × +16.7) × (1 –) " +(–6.6 × +8.6) [9] A and B coefficients calculated from the individual model and the population model are shown in Table 7.1 and Supporting Information Table 7.S2. Figure 7.1. HbS-specific Li-Bush calibration for non-transfused SCD patients. Linear dependence of (A) A and (B) B coefficients on hematocrit using pooled datasets from Bush et al. 285 and Li et al. 343 196 Graphical comparison between sickle-specific calibrations, bovine calibration 568 and HbA calibration 350 is shown in Figure 7.2 for a subset of patients and in Supporting Information Figure 7.S3 for the full cohort. Subject-based blood calibrations yielded no significant bias and small variance compared to blood-gas co-oximetry (Figure 7.3A). Bovine 568 and non-sickle hemoglobin A 350 models demonstrated systemic underestimation of 16.6% and 12.0% saturation units compared to co-oximetry in SCD patients (Table 7.2, Figure 7.3BC). On the other hand, the sickle- specific Bush calibration, Li calibration and Li-Bush combined calibration (Figure 7.3D-F) were unbiased, had comparable variance to one another and smaller variability compared to bovine and HbA models. Between the Li-Bush and patient-specific blood calibrations, there is no significant bias and 95% limits of agreement of 10%. 7.3.2. HbS-based SCD calibration for hyper-transfused patients Comparisons between individual calibration, Li-Bush calibration uncorrected for transfusion, HbS-based mixture model for transfused blood is demonstrated in Table 7.2 and Figure 7.3G-I. When T2 oximetry measurements in hyper-transfused subjects were assessed with uncorrected sickle calibration, venous saturation showed a slight overestimation compared to blood-gas analysis. However, when the HbS-based mixture model was used to correct for transfusion, the bias between T2 oximetry and blood-gas co-oximetry was reduced. 197 Figure 7.2. Individual and population sickle calibration models. Individual and population sickle calibration models in a subset of 2 subjects from Li et al. 343 and 2 non- transfused HbSS subjects from Bush et al. 285 Individual calibration data (black filled circles) were fitted with individual calibration (dotted line), Li population calibration (red solid line), Bush population calibration (blue), Li-Bush combined calibration (green), bovine calibration (purple) and HbA calibration (orange). In Li Subject 2, Bush Subjects 1 and 2, the blue Bush calibration overlaps and is plotted underneath the red Li-Bush combined calibration. 198 Figure 7.3. Bland-Altman analyses of different calibration models on non-transfused SCD subjects and transfused subjects. Non-transfused SCD subjects include 5 from Bush et al. 285 and 11 from Li et al. 343 Transfused subjects include 3 from Bush et al. 285 Each data point represents a set of in vitro blood saturation measurement by cerebral T2 oximetry and blood-gas co-oximetry. (A) Subject-specific calibration gave the highest accuracy and lowest bias but is infeasible for routine clinical use. (B) Bovine and (C) healthy human blood HbA models exhibited large bias and variance when used in subjects with sickle cell disease. Sickle-specific population models by (D) Bush et al. 285 and (E) Li et al. 343 displayed comparable results. Therefore, (F) a combined consensus Li-Bush HbS model was reported and recommended for use in non-transfused SCD subjects. (G) In hyper-transfused patients, subject-specific calibration yielded low bias and variance, whereas (H) the Li-Bush consensus sickle calibration displayed a bias in transfused subjects. To correct for this bias, (I) a mixture HbS-weighted model was developed that demonstrated a smaller bias in transfused SCD patients. 199 Table 7.2. Bland-Altman analysis for different population calibration models on transfused and non- transfused sickled blood. Even though Li et al. only reported patient-specific sickle calibration, a population model was derived from individual datasets. 343 * denotes the bias is significantly different from zero. Non-transfused subjects Transfused subjects Calibration Bias Standard Deviation 95% Confidence Interval Bias Standard Deviation 95% Confidence Interval Individual 343 0.5 3.0 (–5.5, 6.5) –0.1 2.3 (–4.7, 4.5) Bovine 568 16.6* 11.5 (–6.4, 39.7) 7.3* 6.2 (–5.1, 19.7) HbA 350 12.0* 6.7 (–1.5, 25.4) 7.0* 3.5 (0.0, 13.9) HbS Bush 285 0.3 5.2 (–10.2, 10.7) –2.7* 2.9 (–8.4, 3.1) HbS Li 343 –0.6 6.1 (–12.8, 11.7) –3.9* 2.9 (–9.6, 1.9) HbS Li-Bush 0.8 5.4 (–10.1, 11.6) –2.1* 2.5 (–7.0, 2.8) HbS- weighted Li-Bush 0.6 6.1 (–11.7, 12.8) 1.1 3.6 (–6.1, 8.3) 200 7.4. Discussion In this study, we established a T2 oximetry calibration for non-transfused sickle blood by combining calibration data from two independent studies. 285,343 The combined Li-Bush calibration was unbiased compared to blood-gas co-oximetry and yielded limits of agreement of (–10.1%, 11.6%). Although the combined calibration did not demonstrate statistical superiority compared to previous Li or Bush models, by increasing the number of patients and broadening the hematocrit range, we strengthened the robustness and generalizability of our calibration. This larger working range of hematocrit also revealed the dependency of blood T2 on hematocrit, which was not previously observed by the Bush sickle calibration. 285 Additionally, this work demonstrated the need to correct for transfusion in T2 oximetry in hyper-transfused SCD patients and explored a correction method based on HbS percentage; this correction demonstrated no bias and lower variance than uncorrected Li-Bush calibration in transfused blood. 7.4.1. Sickle-specific Li-Bush population model A and B coefficients derived from both the individual calibration and the sickle population calibration demonstrated large inter-subject variations; this high variance in measured A and B coefficients can be explained by the variation in hematocrit in the patient cohort. Additional variations in the ex vivo blood calibration procedure such as precise temperature control and minimization of red cell aggregation, as well as inter-subject differences in red cell damage, could potentially contribute to the remaining variation in blood T2 measurements. Similar to previous studies, 285,343 our results did not demonstrate a statistically significant dependence between A, B and hemoglobin S percentage. The lack of HbS dependence could be due to its smaller effect size compared to the larger inter-subject variability in subjects with similar HbS. 343 Additionally, in subjects with lower hemoglobin S percentage, HbA-containing red blood 201 cells could be damaged by the inflamed vascular endothelium, 591 altering the density and membrane permeability and leading to similar magnetic characteristics as sickle red blood cells. Additionally, since our Li-Bush calibration model was constructed from hemoglobin SS subjects, we could not predict how well the calibration will perform on other sickle genotypes, including SC, S0, and S+. Patients with S0 have very similar clinical and hematological characteristics as SS disease, 592,593 so we anticipate that the SS calibration would translate well to S0 genotype. S+ has varying mixtures of hemoglobin A and S within individual red blood cells depending on the type of mutations. 594 Mild forms will behave like sickle cell trait (best described by the normal hemoglobin calibration) 285 while more severe forms will mimic non- transfused SS disease. Red blood cells in hemoglobin SC disease suffer from significant dehydration and membrane abnormalities, 595,596 similar to cells with SS hemoglobin, but it is possible that the calibration coefficients would be subtly different. In the absence of future disease-specific calibration data, we advocate using the SS calibration for SC patients based upon these a priori considerations. 7.4.2. Importance of a unified SCD calibration Currently, there is instability in the field of cerebral oxygenation with respect to patients with SCD. Oxygen extraction fraction has been reported as increased, 149,587 decreased, 285 and unchanged 343 in SCD by different, well-regarded groups. The choice of T2 oximetry calibration is at the center of the controversy, and the recent addition of two, potentially-competing sickle cell calibrations complicates the issue. 285,343 Since calibration experiments are challenging to perform, sample size constraints and intrinsic inter-subject variability limits the accuracy of both prior reports. Furthermore, institutional differences in treatment of SCD and patient compliance with medications limit the generalizability of results from any single center. The combined Li-Bush calibration was derived from wider intrinsic hematocrit range, lowering variability, increasing robustness and increasing the confidence of individual parameter estimates. It also demonstrated 202 favorable Bland-Altman agreement with patient-specific calibrations. Most importantly, it represents a “consensus” calibration between the two groups responsible for the prior SCD calibrations and should serve as the single reference for future T2 oximetry studies in SCD, even if statistical superiority could not be demonstrated with respect to previous Li and Bush calibrations. 7.4.3. Importance of a calibration for transfused SCD patients Regular transfusion therapy is the treatment of choice for stroke prophylaxis for SCD patients with abnormal transcranial Doppler, pulmonary hypertension, and acute chest syndrome. 597 Chronically transfused SCD patients typically maintain pre-transfusion HbS of 30%. Roughly 10-20% children and young adults with SCD receive chronic transfusion, so it is important to be able to apply cerebral oximetry in this patient cohort. Both the hemoglobin A model and all of the hemoglobin S models were biased with respect to transfused SCD subjects. We demonstrated that a simple linear admixture model using HbS percentage as weighting coefficient produced unbiased estimate with lower error than any of the individual models. While there is no a priori reason to believe that the calibration coefficients should vary linearly with HbS, this simple approach provides smooth behavior between well-calibrated boundary points, similar to Taylor’s series expansion, and is unlikely to introduce wild deviations from expected behaviors. Although the validation sample was small, additional validation work in this cohort is unlikely, so it is important to establish a practical and logical approximation. 7.4.4. Is it fair to pool data between Li and Bush studies? In vitro studies of blood transverse relaxivity are challenging because blood oxygenation must be altered while controlling for temperature, red-cell aggregation, susceptibility artifacts, and pulse sequence differences. 285,350,568 Therefore, it is reasonable to question whether data from the 203 two studies can legitimately be pooled. To address this possibility, Li et al. compared the observed blood R2 from 9 healthy volunteers against the Bush HbA calibration (Figure 7.S2 in Li et al.) 343 . Bland Altman analysis between observed and predicted oximetry values were –0.8±2.8% (Figure 7.S4 in Li et al.), 343 representing phenomenal concordance between two studies. While this does not guarantee that sickle blood would yield as robust a result between the two centers, it does exclude significant systematic biases from instrumentation and experimental differences. Furthermore, the open and filled symbols in Figure 7.1 are plausibly from the same distribution, although much larger sample sizes would be needed to prove that point. 7.4.5. Importance of a sickle-specific calibration The bovine calibration is used by many studies in subjects with normal hemoglobin. 568 Even though bovine red cells do not form Rouleaux formations and are smaller than human red blood cells, previous work has shown that the bias between bovine model and healthy human blood model was small over the hematocrit range of 35-55% in which this model was originally calibrated. 350 Outside this range, the bovine model severely underestimated venous oxygenation; this large systematic bias introduced by extrapolating the bovine model outside of its useful calibration range has been reported in anemic subjects. 285 This bias was further worsened by the failure to account for the presence of sickle hemoglobin. This manuscript demonstrates that whether one uses the Li model, Bush model, the Li-Bush combined model, or individual calibration, the R2 changes produced by sickle red cells are poorly described by either the bovine or HbA calibrations (Figure 7.3). 7.4.6. Relationship between T2 and saturation in sickle blood Red blood cell R2 relaxation is primarily determined by their intrinsic magnetic susceptibility, their membrane integrity and degree of hemoglobin S aggregation. 598 Sickle cell 204 trait patients, who have normal red cell membrane characteristics and no hemoglobin polymerization under most physiological conditions despite 40% hemoglobin S concentrations, have identical R2 characteristics as controls with exclusively hemoglobin A, suggesting that the fundamental magnetic susceptibility of hemoglobin S is not different from hemoglobin A. 350 However, simple visual inspection of blood smear collected from a patient with SCD highlighted the tremendous spectrum of membrane surface and cytoskeletal damage experienced by these patients. 599 This aberrant shape and increased red cell–red cell interactions potentially magnify the contributions of two relaxation mechanisms to the sickle blood transverse relaxation, with relative importance of each mechanism depending on field strength. 600 The first mechanism is simple movement of water protons (by bulk flow and diffusion) in nonhomogeneous magnetic field which causes phase accrual that is incompletely reversed by the CPMG pulses in T2 preparation. 601 Since sickle red blood cells are relatively dehydrated compared to normal blood cells (they have a high mean corpuscular hemoglobin concentration, or MCHC), 602 it is not surprising that the A and B coefficients might have stronger hematocrit weighting. 603 In addition, the A coefficient depends on the relative length-scale between the magnetic inhomogeneities and the proton mobility, 603 which will be impacted by abnormal cell shape and red cell aggregation in sickle blood. The second relaxation mechanism involves water exchange across red cell membranes. In the chemical exchange model, 604,605 the coefficient A depends upon susceptibility difference between plasma, deoxygenated red blood cells and exchange time across the membrane. As in the diffusion model, the susceptibility difference between plasma and red cells depends upon hematocrit, MCHC, and saturation. The red cell exchange time varies with cell shape and intrinsic water permeability, both of which are abnormal in SCD. At 3T, diffusion-mediated loss dominate in non-sickle blood 606 but the relative contributions of the two mechanisms could differ in SCD blood because red cell membrane shape and membrane properties differ so dramatically. 599 205 Regardless, both the diffusion and chemical exchange model predict that sickle blood should have a unique calibration compared with non-sickle blood. 7.4.7. Limitations A limitation to T2 oximetry imaging is its dependence on hemoglobin-specific calibration curves. The most accurate solution is the use of individual calibrations for each subject. 343 Unfortunately, constructing patient-specific calibration curve for each patient may be infeasible for routine clinical or research use because it requires significant time and resources (blood draw, oxygenation adjustments, MRI time). And even though we did not observe bias in oxygenation estimates in the Li-Bush combined calibration compared to co-oximetry, further validation on a separate cohort of hemoglobin SS subjects is important. Additionally, validation against independent methods to measure venous saturation, such as central or peripheral venous catherization, or MRI susceptometry methods 607,608 is required to affirm the clinical utility of this T2 oximetry model. We acknowledge that our HbS-weighted mixture model for hyper-transfused subjects is empiric and simplistic. While bias was not observed in our approximation, we were only able to test this model in three transfused patients. Since all three subjects had very similar hemoglobin S (32.1 ± 4.3%), the mixture model could not demonstrate superiority compared to the uncorrected calibration. However, in the clinical setting, transfused SCD subjects can have large variability in hemoglobin S concentrations (range of 0 to 60% after 36 months of continuous transfusion). 541 Therefore, the proposed HbS-weighted model that exhibits smooth transition near the HbS extrema is a reasonable correction option. Future work to evaluate the accuracy of this HbS-based correction method on a larger cohort of transfused subjects is desirable, however, it is not a high priority. 206 7.4.8. Conclusion In summary, this study established a blood T2 oximetry calibration derived from pooled data 285,343 to provide a single reference standard in patients with SCD. The revised sickle calibration is unbiased, and the derived oximetry exhibits limits of agreement of (–10.1%, 11.6%), despite residual inter-subject variability. Our results call into question previous MR oximetry studies 149,587,588 in SCD that used calibration curves derived from bovine blood or human blood with normal hemoglobin. Additionally, this study proposes a correction method for patients undergoing chronic transfusion to correct for the lower HbS percentage in transfusion. 207 7.5. Supplemental Information 7.5.1. Supplemental Tables Supplemental Table 7.S1. All group calibration formulas used in this manuscript. Note that formulas are shown in similar format to original form found in references. Care was taken to covert R2 apparent values to R2 corrected for pulse width (R2,corr) with 1 !,)*++ ⁄ = !,)*++ = !,,-- × =1 !,,-- ⁄ ×, where =1− _ℎ (2 × ℎ_) ⁄ =0.913 with pulse width of 1.74ms and CPMG module interecho spacing (t) of 10ms. Calibration Derived from References Bovine Bovine blood Lu et al. 568 ",!hh = +×(1 − )+×(1 – ) " , where =−13.5+80.2×−75.9× " , = −0.5×+3.4× " , =247.4××(1−) Human HbA Healthy human blood Bush et al. 350 ",!hh = 77.5××(1 – ) " +27.8×(1 – ) " +6.95× +2.34 Bush HbS Sickled blood in Bush et al. 285 Bush et al. 285 ",!hh = 70.0×(1 − ) " +5.75 Li HbS Sickled blood data recalculated from Li et al. 343 Li et al. 343 ",*,.. = 278××(1 − ) " −5×(1 − ) " −8.4×+9.3 Li-Bush HbS Sickled blood, for non-transfused patients Current study ",*,.. = 196.8××(1 − ) " +16.7×(1 − ) " −6.6× +8.6 Mixture HbS- weighted Sickled blood, for transfused patients Current study ",*,.. = ' ××(1 − ) " + " ×(1 − ) " + F ×+ P where =t1− T0^% u8% u× T0< + t T0^% u8% u × T0^ with T0< as coefficients from Human HbA model and T0^ as coefficients from Li-Bush HbS model 208 Supplemental Table 7.S2. Subject hematologic and calibration parameters. A and B coefficients were calculated using the individual calibration model and the sickle-specific population calibration model. Li et al. did not measure HbA, pH and pCO2, but the reader can impute pH and pCO2 from our data and approximate ≈100%−−. Subject Hct (%) HbS (%) HbA (%) HbF (%) pH pCO2 Individual Li-Bush Population A B A B Non- transfused Li #1 18.0 91.0 2.7 51.7 7.8 51.1 7.4 Li #2 34.0 79.0 14.2 98.1 7.2 83.6 6.4 Li #3 27.0 84.8 6.5 54.4 4.7 69.8 6.8 Li #4 25.0 86.5 6.5 46.2 7.6 65.9 7.0 Li #5 25.0 68.4 4.8 72.7 9.2 65.9 7.0 Li #6 26.0 78.2 16.4 63.9 6.9 67.9 6.9 Li #7 28.0 N/A N/A 68.0 6.4 71.8 6.8 Li #8 28.0 85.2 9.1 102. 7 6.8 71.8 6.8 Li #9 23.0 80.4 11.9 78.5 6.2 62.0 7.1 Li #10 19.0 86.8 7.8 64.7 7.7 54.1 7.3 Li #11 23.0 86.4 7.5 67.7 7.0 62.0 7.1 Bush #1 35.2 64.1 0 30.9 7.34 26.5 84.2 5.7 83.6 7.4 Bush #2 33.4 69.3 18.5 7.3 7.32 26.5 80.2 5.4 69.9 6.4 Bush #3 36.0 95.3 0 1.5 7.47 40.6 76.2 7.0 65.9 6.8 Bush #4 24.0 85.0 0 2.3 7.39 53.8 69.4 5.8 65.9 7.0 Bush #5 24.0 93.2 0 12.1 N/A N/A 70.1 5.8 67.9 7.0 Transfused Bush #6 30.3 24.5 77 1.3 7.36 30.8 73.7 6.6 62.3 5.4 Bush #7 28.9 39.4 56.1 1.7 7.41 54.3 68.7 5.5 64.1 5.7 Bush #8 37.0 45.9 39.1 5.2 7.32 24.9 75.5 5.4 77.7 5.8 209 7.5.2. Supplemental Figures Supplemental Figure 7.S1. Confidence limits of the hematocrit dependence of (A) A and (B) B coefficients using pooled datasets of sickle cell disease subjects from Bush et al. 285 and Li et al. 343 Individual A and B error bars are calculated from the standard error of parameter estimates. Individual hematocrit error bars are calculated from the hematocrit coefficient-of-variation 3%. Supplemental Figure 7.S2. Confidence limits of the hematocrit dependence of (A) A and (B) B coefficients using five healthy control subjects from Bush et al. 285 Individual A and B error bars are calculated from the standard error of parameter estimates. Individual hematocrit error bars are calculated from the hematocrit coefficient-of-variation 3%. 210 Supplemental Figure 7.S3. Individual and group sickle calibration models based on 11 subjects from Li et al. 343 and 5 non-transfused HbSS subjects from Bush et al. 285 Individual calibration data (black filled circles) were fitted with individual calibration (dotted line), Li population calibration (red solid line), Bush population calibration (blue), Li-Bush combined calibration (green), bovine calibration (purple) and HbA calibration (orange). 211 Chapter 8 : Reduced global cerebral oxygen metabolic rate in sickle cell disease and chronic anemias 8.1. Introduction Although the brain only accounts for 2% of the body mass, it accounts for 20% of the body’s resting energy expenditures. 609 Since the brain is unable to survive under anaerobic conditions, cerebral oxygen delivery is tightly regulated. Over 98% of the oxygen delivered to the brain is transported by hemoglobin. 610 Low hemoglobin level, or anemia, is common in the general population. 534 Anemia is a hallmark of many genetic diseases worldwide, including sickle cell disease (SCD) and thalassemia. Hemoglobin level is inversely correlated with cerebral blood flow (CBF) among healthy subjects. 433,611,612 To compensate for chronically decreased oxygen capacity, resting CBF rises to preserve global oxygen delivery. 396,397 Despite having adequate resting oxygen delivery, chronically anemic subjects demonstrate diffuse brain volume loss and silent strokes, suggesting chronic or acute-on-chronic supply-demand imbalance. 613–617 Early prior works in SCD T2- relaxation-under-spin-tagging (TRUST) 348 MRI with a calibration curve derived from bovine blood argued that brain oxygen extraction fraction (OEF) in SCD patients is increased proportional to the degree of anemia 149,587 to maintain normal cerebral metabolic rate of oxygen (CMRO2). 343,618 Recent in vitro calibration work demonstrated unbiased and improved limits of agreement in OEF measurement using sickle-specific calibration compared to calibration derived from bovine blood. 568,569 Using these more appropriate calibration curves for human hemoglobins A and S, these works have demonstrated decreased OEF and CMRO2 in SCD patients compared to healthy subjects. 285,350 Therefore, in light of these divergent and controversial findings, this study 212 evaluated CMRO2 using a recently published consensus calibration derived from sickle blood 569 and compared venous oximetry, oxygen extraction and metabolic rate in a large cohort of control subjects, SCD and non-sickle anemic patients. We also systemically reviewed CMRO2 in anemic patients performed using invasive, gold standard, MRI-independent assays (Positron Emission Tomography and Kety-Schmidt dilution technique 153 ). 8.2. Methods 8.2.1. Study population All studies were performed at Children’s Hospital Los Angeles and approved by the Committee on Clinical Investigation (CCI#2011-0083). A total of 118 participants between 12 and 63 years of age were separated into three groups: healthy control subjects (CTL, N=44), subjects with sickle cell disease (SCD, N=47), and subjects with non-sickle chronic anemia (ACTL, N=27). Out of 118 subjects, 10 controls, 18 SCD and 7 ACTL subjects were between 12-18 years of age. Controls were drawn from families and friends of patients with SCD, so sickle cell trait was common, occurring in 23/44 control subjects. The ACTL group consisted of patients with thalassemia major (N=14), thalassemia intermedia (N=3), aplastic anemia (N=1), autoimmune hemolytic anemia (N=1), hereditary spherocytosis (N=3), hemoglobin H constant spring (N=4) and congenital dyserythropoietic anemia (N=1). The SCD group consisted of 39 subjects with SS hemoglobin, six subjects with SC hemoglobin, one with Sβ 0 hemoglobin and one with Sβ + hemoglobin. Chronic transfusion therapy was administered to 16/47 SCD patients and 16/27 ACTL patients every 3-4 weeks. Transfused SCD and ACTL patients were studied immediately prior to transfusion. Eleven non-transfused SCD subjects were undergoing hydroxyurea treatment at the time of the study. Exclusion criteria included: 1) pregnancy; 2) hypertension; 3) diabetes; 4) stroke or other known neurologic insult; 5) seizures; and 6) known developmental delay or learning disability. Group demographic and clinical variables are summarized in Table 8.1. 213 Imaging and blood samples were obtained on the same day for each subject. Blood gas analysis, complete blood count and quantitative hemoglobin electrophoresis were analyzed in our clinical laboratory. 214 Table 8.1. Subject demographics and hematologic markers. Bold lettering denotes statistically significant p-values (p < 0.05). CTL (N = 44) ACTL (N = 27) SCD (N = 47) p-value (ACTL– CTL) p-value (SCD– CTL) p-value (SCD– ACTL) Age (Years) 26.4±10.6 25.0±10.4 21.1±7.1 0.81 0.07 0.39 Sex 14M, 30F 13M, 14F 26M, 21F 0.37 0.06 0.82 Body Mass Index 25.7±7.0 23.1±2.5 22.3±4.4 0.11 <0.01 0.79 Transfused 0/44 16/27 16/47 <0.01 <0.01 0.02 Arterial Saturation (%) 99±1 99±1 98±2 0.23 <0.01 0.01 Systolic Blood Pressure (mmHg) 118±13 113±9 112±11 0.16 0.04 0.97 Diastolic Blood Pressure (mmHg) 68±10 62±7 60±7 <0.01 <0.01 0.55 Mean Blood Pressure (mmHg) 85±13 80±7 78±9 0.09 <0.01 0.83 Hemoglobin Electrophoresis 21AA, 23AS 23AA, 3AE, 1AS 39SS, 6SC, 1Sβ0, 1Sβ+ Hemoglobin (g/dL) 13.5±1.2 10.6±2.6 9.7±1.8 <0.01 <0.01 <0.01 Hematocrit (%) 40.0±3.2 32.5±6.2 27.6±4.7 <0.01 <0.01 0.07 White Blood Cell Count (x10 3 ) 5.7±1.8 6.9±2.3 9.7±4.3 0.26 <0.01 <0.01 Platelet Count (x10 3 /µL) 251±56 268±118 321±128 0.79 <0.01 0.09 Reticulocytes (%) 1.3±0.5 2.6±2.8 9.2±5.0 0.30 <0.01 <0.01 HbS (%) 20±19 0 55±29 <0.01 <0.01 <0.01 Fetal Hemoglobin (%) 0.4±1.8 2.2±4.0 8.4±8.7 0.46 <0.01 <0.01 Cell-free Hemoglobin (mg/dL) 6.3±5.3 16.9±18.0 21.3±20.0 0.02 <0.01 0.49 Red Blood Cell Distribution Width 13.6±1.7 17.9±5.4 18.6±3.8 0.01 <0.01 0.76 Silent Cerebral Infarct Presence 9/44 8/27 20/47 0.63 0.04 0.25 CBF (mL/100g/min) 61.8±9.8 83.0±16.4 97.8±24.9 <0.01 <0.01 0.01 O2 Content (mL O2/mL blood) 18.2±1.5 14.4±3.4 13.0±2.4 <0.01 <0.01 0.05 O2 Delivery (mL/100g/min) 11.2±1.9 11.5±1.9 12.3±2.3 0.75 0.07 0.44 OEF (%) 36.1±5.8 30.9±5.6 27.5±4.1 <0.01 <0.01 0.03 CMRO2 (mL O2/100g/min) 4.01±0.63 3.53±0.65 3.39±0.71 0.01 <0.01 0.22 215 8.2.2. MRI All imaging was performed on a Philips Achieva 3T MR system with an eight-channel, receive-only head coil. A 3D T1-weighted, T2-weighted fluid-attenuated inversion recovery (T2- FLAIR) 3D image and MR angiography image were acquired; details can be found in previous work. 547,574 Grey matter volume, white matter volume, cortical thickness and cortical surface area were calculated from T1 images based on previously-published methods. 613 Silent cerebral infarcts equal or greater than 3 mm in diameter in two orthogonal planes 541,619 were documented on T2-FLAIR images by the consensus of a neuroradiologist and neuroanatomist. Subjects with more than one silent cerebral infarction per decade of life were considered to have an abnormal burden of silent stroke. 620 3D MRI angiography of the Circle of Willis and distal internal carotid artery was performed using standard techniques and assessed by a neuroradiologist for vasculopathy. Whole-brain CBF was measured using phase contrast MRI; 543 details of phase contrast MRI acquisition has been explained in previous work. 203,396 Cerebral venous oxygenation Yv was measured using TRUST MRI. The TRUST sequence used in this study has been explained in prior publications. 285,348,350 Briefly, a transverse, single-slice TRUST sequence with Carr-Purcell- Meiboom-Gill (CPMG) T2 weighting was used to measured blood T 2 in the sagittal sinus. A sickle- specific calibration model was used to convert blood T2 to venous saturation in SCD group; 569 it reflected a consensus calibration derived from two independent studies. 285,343 A calibration model from human AA blood 350 was used for control and non-sickle anemic patients. This calibration has been independently validated. 343 Chronically transfused patients were characterized using a mixture model 569 between the hemoglobin A and hemoglobin S calibration curves based upon their hemoglobin electrophoresis results at the time of the study. Although patients with sickle cell trait have up to 45% sickle hemoglobin, they have 0% sickle cells so their blood oximetry is 216 accurately described by the hemoglobin A model. 350 A summary of calibration models and their usage are shown in Supplemental Table 8.S1. 8.2.3. Physiological parameters Several physiological parameters were derived using the following equations: O 2 content = 1.34 × Hb × Y a + 0.003 × pO 2 (mL O2/mL blood) [1] OEF = Y a – Y v Y a [2] O 2 delivery = CBF × O 2 content (mL/100g/min) [3] CMRO 2 = CBF × OEF × O 2 content (mL O2/100g/min) [4] where pO2 is the partial pressure of oxygen estimated to be 100mmHg, Hb is the hemoglobin level, Yv is the venous saturation measured by TRUST and Ya is the arterial saturation measured by pulse oximetry. 8.2.4. Peripheral laboratory and venous blood gas measurements To probe for evidence of impaired oxygen extraction outside the brain, we performed blood-gas analysis and co-oximetry measurements of oxygen saturation from the brachial circulation. These measurements also have the advantage of being independent of MRI calibration and underlying disease state. A 22-gauge intravenous catheter was placed in the antecubital vein, and the tourniquet was released. After the intravenous line was secured (taking approximately 3-4 minutes), 15 mL of blood was drawn for a complete blood count, hemoglobin electrophoresis, and blood rheology. Following this, 0.5 ml was drawn slowly into a 1 ml blood gas syringe for analysis of venous blood gas (Alere Inc., EPOC Blood Analysis System, Waltham, MA) and oxygen saturation (Radiometer, OSM-3, Copenhagen, Denmark). 217 8.2.5. Reference data In addition to MRI data acquired in this study, we performed a literature review of studies that measured CMRO2 using Kety-Schmidt 153 or Positron Emission Tomography (PET) in anemic patients and healthy subjects. Since we wanted to assess these reference datasets in comparison with our results, we limited our search to publications of various types of anemia that provided data of individual patients. Required parameters included hemoglobin, age and CMRO2 values, or a complete set of measurements that could be used to calculate these parameters. This literature search was performed on PubMed, limited to papers published after 1900 and excluded studies that were related to animal work, infants, neonates and any other pathophysiology other than anemia. 8.2.6. Statistical analysis Statistical analysis was performed in JMP (SAS, Cary, NC). One-way ANOVA with Dunnett’s post hoc correction was used to examine the difference in parameters between SCD and control, and between ACTL and control; independent samples t-test was used to compare SCD and ACTL groups. Univariate and stepwise multivariate regressions were performed against hemoglobin, CBF, HbS, transfusion status, fetal hemoglobin, white blood cells, platelets, reticulocytes, cell-free hemoglobin and red blood cell distribution width and other imaging biomarkers including white matter volume, cortical thickness, cortical surface area and presence of silent cerebral infarcts; multivariate predictors were retained in the final model for p<0.05. Correction for age and sex were performed by removing the linear relationship with the measurement and then adding the measurement average back into the residuals to ensure final values were within physiologically meaningful range. 218 8.3. Results 8.3.1. Current study Table 8.1 summarizes the demographics for three groups. Since the control group trended slightly older and had more females compared to anemic subjects, subsequent analyses of OEF and CMRO2 measures were corrected for age and sex. There were no significant differences in OEF and CMRO2 between patients with hemoglobin SS, SC, Sβ + and Sβ 0 (p=0.92 for OEF, p=0.57 for CMRO2) so all genotypes were pooled together. Control subjects with and without sickle cell trait had indistinguishable physiologic and laboratory values except for hemoglobin electrophoresis results, so data for hemoglobin AS and AA were pooled for all comparisons. Anemic patients had lower hemoglobin, hematocrit and blood pressure compared to controls. Patients with SCD had higher white blood cells, platelets, reticulocytes, and fetal hemoglobin compared to ACTL subjects at the same hemoglobin level. Transfused SCD subjects had HbS percentages of 23±15%, whereas non-transfused patients had HbS percentages of 71±20%. Although no subject had a clinical history of stroke, T2-FLAIR MRI showed silent cerebral infarct presence in nine control subjects (20%), 20 SCD subjects (43%, p=0.04) and eight ACTL subjects (30%, p=0.63). MRI angiography was completely normal in all subjects with no evidence of intracranial or extracranial vasculopathy. Figure 8.1 compares oxygenation measures between three groups, and average oxygen parameters are summarized in Table 8.1. CBF was significantly higher in chronic anemias than normal controls (Figure 8.1A) but cerebral oxygen delivery was identical across groups (Figure 8.1B). Cerebral OEF was decreased in anemic subjects (SCD to a greater extent than ACTL) compared with control subjects (Figure 8.1C). CMRO2 values were also decreased in SCD and ACTL groups compared with healthy controls (Figure 8.1D). OEF and CMRO 2 values were independent of transfusion status in SCD and ACTL subjects. 219 8.3.2. Effects of anemia on cerebral oxygenation On univariate analysis, cerebral OEF was positively correlated with hemoglobin (r 2 =0.24, p<0.01) and negatively correlated with CBF (r 2 =0.26, p<0.01) and HbF (r 2 =0.11, p<0.01). Upon Figure 8.1. Boxplot and linear correlations of oxygen supply and utilization values. Chronically anemic subjects demonstrated (A) increased CBF, (B) similar oxygen delivery but (C) lower OEF and (D) lower CMRO2 compared to healthy controls. Linear correlation (E) between CMRO2 and hemoglobin and (F) between cerebral OEF and brachial OEF in the cohort in this study. * denoted statistical significance p<0.05; NS denoted no significant difference. 220 multivariate analysis, hemoglobin and HbF explained 24% and 5% of the variability respectively. Univariate and multivariate results are summarized in Supplemental Table 8.S2. Figure 8.1E demonstrate that CMRO2 was moderately correlated with hemoglobin (r 2 =0.17, p<0.01). When the impact of hemoglobin was removed by linear regression, CMRO 2 was no longer significantly different between three groups (p=0.35), nor were any hematologic parameters retained as significant predictors. Neither OEF nor CMRO2 were significantly correlated with imaging biomarkers of damage including white matter volume, cortical thickness, cortical surface area and presence of silent cerebral infarcts (Supplemental Table 8.S2). 8.3.3. Peripheral oxygenation Brachial OEF was more variable than cerebral OEF, with an interquartile range of 18– 61%, but was correlated with TRUST-based cerebral OEF (r 2 =0.17, p<0.01, Figure 8.1F). SCD and ACTL patients had brachial OEF of 29±22% and 37±23% respectively, significantly lower compared to control subjects (60±23%, p<0.01). Brachial OEF also demonstrated a positive correlation with hemoglobin (r 2 =0.14, p<0.01). After correcting for hemoglobin, brachial OEF was still significantly lower in transfused subjects but no longer different between the anemic and control groups (p=0.57, Supplemental Table 8.S2). 8.3.4. Reference data Literature search yielded a total of 37 reports on brain oxygen metabolism of various pathophysiologies that included individual patient data, 28 of which reported both hemoglobin and CMRO2. Of these publications, only eight studies focused on subjects with anemia and healthy controls. Details on these historical references can be found in Supplemental Table 8.S4. The eight studies included Herold et al. 621 who studied SCD and control subjects using PET to demonstrate decreased CMRO2 in SCD. Similarly, Frackowiak et al. 622 also measured 221 oxygen metabolism using PET in healthy volunteers. Using a nitrous oxide Kety-Schmidt technique, 153 Heyman et al. 623 studied sickle cell anemia, iron-deficiency anemia, aplastic anemia and anemia due to blood loss. With the same method, Scheinberg et al. studied oxygen utilization in patients with pernicious anemia 624 and healthy controls. 625 Stewart et al. 626 also studied oxygen circulation in pernicious anemia. Fazekas et al., 627 Mangold et al. 628 and Kety et al. 629 published on cerebral hemodynamics on healthy controls under different observation conditions; only control datasets under baseline conditions were included to compare with CMRO2 in anemia. Details on values extracted from the 8 references are shown in Supplemental Table 8.S3. The historical data are summarized in Figure 8.2. Figure 8.2A reveals a clear decrease in CMRO2 with age (r 2 =0.20, p=0.01), making it necessary to perform age adjustment for all comparisons. Figure 8.2B demonstrates that age-adjusted CMRO2 is directly proportional to hemoglobin level (r 2 =0.31), mirroring our present observations using MRI (Figure 8.1E). Figure 8.2C demonstrates that CMRO2 was depressed in all anemia subjects. After controlling for hemoglobin level, no differences were observed among controls and all anemia subtypes. The same relationship with anemia severity remained even when the effects of varying arterial CO2 tensions were accounted for (Supplemental Figure 8.S1). We further broadened our analysis of historical CMRO2 data to 20 additional studies focused on specific diseases or conditions, where fluctuations in hemoglobin were a nuisance variable (Supplemental Table 8.S4). An almost identical linear relationship between CMRO 2 and hemoglobin was observed (illustrated in Supplemental Figure 8.S2). After controlling for variations in hemoglobin, the differences in CMRO2 across studies were readily explainable by the primary disease process (such as hyperthyroidism, uremia, anesthesia, etc.). 222 In order to directly compare the results from our study with the historical references, we compared our MRI results to Figure 8.2A data in Figure 8.3. The scattergrams between CMRO2 and hemoglobin were superimposed (Figure 8.3A). Figure 8.3B reveals that the MRI-based CMRO2 measurements in both SCD and control groups corresponded quite well with results from Kety-Schmidt and PET. After controlling for hemoglobin, all differences among the subgroups were eliminated. Figure 8.2. Relationship between CMRO2, age and anemia severity in historical references. (A) Linear correlation between CMRO2 and age. (B) Linear correlation between age-adjusted CMRO2 and hemoglobin. (C) Lower CMRO2 in different anemia types compared to controls. 223 8.4. Discussion In this manuscript, we examined 118 subjects and identified a reduction in OEF and CMRO2 in patients with chronic anemia compared to controls. Decreased CMRO 2 has been previously reported in a subset of this cohort. 396,630 In this current work, we demonstrate that decreased CMRO2 is correlated with anemia severity, independent of transfusion status and anemia subtype. We further corroborated our observations by comparing our results to reference data from eight independent studies in 146 healthy controls and anemic patients using gold- standard nitrous-oxide method and PET. The blunted cerebral OEF observed in anemic subjects was mirrored by cooximetry-assayed peripheral venous oxygen extraction, suggesting that decreased extraction was neither confined to the cerebral circulation nor an unanticipated artifact of the TRUST cerebral oximetry technique. Importantly, this work observed decreased OEF and CMRO2 over a wide range of anemic etiologies in otherwise healthy anemic subjects. This decreased cerebral oxygen consumption despite preserved global oxygen delivery 395 illustrates poor oxygen supply-demand matching and can potentially lead to cerebral hypoxia and Figure 8.3. Relationship between CMRO2 and anemia severity when pooling the data from this current study with historical references. (A) Linear correlation between CMRO2 and hemoglobin. (B) Lower CMRO2 in different anemia types compared to controls. 224 ischemia. 631,632 Therefore, given the high incidence of neurovascular disease and stroke in anemic patients, mismatches in oxygen delivery and consumption need to be monitored and may provide meaningful clues into stroke etiology in chronic anemia physiology. Multivariate regression analysis demonstrated univariate correlations between hematologic parameters and OEF and CMRO2 – with hemoglobin as a consistent and strong predictor of both cerebral and brachial oxygenation values. After controlling for the effects of varying anemia severity, most hematologic variables were no longer significant predictors, demonstrating that anemia is the underlying and unifying biomarker of decreased oxygen extraction and metabolism in our cohort. The significant association between both fetal hemoglobin and transfusion with impaired OEF is consistent with a left shift of the hemoglobin dissociation curve at the same pO2 operating point. 610,633 Such correlations suggested that even though induction of high HbF (through hydroxyurea use) and chronic transfusion are popular SCD treatments and offered protection against disease severity, 634 these treatments should be accompanied by frequent monitoring of cerebral oxygenation and stroke risk. Additionally, the lack of correlation between silent infarct presence and reduced CMRO2 could potentially be explained by the regional characteristics of infarcts. The majority of brain metabolism occurs in the grey matter, 635,636 whereas silent infarcts tend to happen in the deep white matter regions. 283,574 Therefore, there is no a priori reason to expect correlation between white matter silent infarcts and global CMRO2 values, except for their respective associations of anemia severity. 546 Classically, CMRO2 has been thought to be preserved via modulations in blood flow and oxygen extraction. In humans, studies of hemodilution show that flow reserves respond initially and most robustly to ischemia. 637 As flow reserves become exhausted, extraction reserves are accessed and symptomatic impairment occurs, as seen in carotid stentosis. 284,638 In more extreme animal models of hemodilution, oxygen extraction and metabolic rate typically remain in the normal range until oxygen content is lowered below 40%. 639–641 In this current cross-sectional study, the oxygen content in our anemic subjects was 72±14% of the control cohort, much higher 225 than hemoglobin levels associated with increased OEF in animal studies. More importantly, our subjects have experienced anemia for decades, allowing possible chronic compensation such as vascular remodeling 642,643 and alterations in oxygen diffusional paths 644 to affect oxygen unloading in the microvasculature. Though a few animal studies have observed increased OEF in severe subacute (2-3 weeks) anemia, none have systemically explored longer, more mild exposure to anemia. 645–647 Lastly, acute hemodilution is only a surrogate for chronic anemia and does not mirror the physiological complexity of sickle cell disease, thalassemia or any of the anemia syndromes we studied. Even though preservation of global oxygen delivery via elevated blood flow in patients with chronic anemias is well established, 397,648 our data in combination with historical data compiled in this manuscript demonstrate concretely that CMRO2 is diminished in chronically anemic populations. This observation is seemingly contradictory to recent reports of elevated OEF in SCD subjects suggested by Asymmetric Spin Echo (ASE) MRI techniques 649 and Near Infrared Spectrometry (NIRS). 539 However, these techniques measure spatial averages of oxygenated and deoxygenated hemoglobin concentration to estimate oxygen saturation within the tissue, independent of blood flow. As evidence of the difference between flow-weighted OEF and tissue saturation, a study using optical measurements of blood flow and oxygenation in mouse cerebral cortex demonstrated that flow-weighted oxygen extraction is lower than spatially-averaged oxygen saturation and that this difference increases under hyperemic conditions. 339 Laser Doppler and jugular venous catherization studies also demonstrated reduced arteriovenous difference during hyperemia, contrary to the increased tissue OEF observed in hyperemic SCD subjects by ASE. 340,341,650 In addition, ASE weighs grey matter and white matter oxygen consumption equally through spatial averaging, while the whole brain CMRO2 measured by flow-weighted TRUST is dominated by grey matter oxygen consumption. As a result, neither ASE nor NIRS can be used to infer whole brain oxygen consumption, and OEF differences between ASE and TRUST are to 226 be expected. Thus, flow-weighted and spatially-weighted OEF techniques complement one another and provide a window into brain capillary flow dynamics. Our patient cohort also has physiological differences that might account for lower OEF values relative to some other published works. Firstly, we had no evidence of cerebral vasculopathy, so patients were able to adequately increase their CBF to maintain resting oxygen delivery, 396 unlike carotid stenosis. This absence of clinical vasculopathy could also explain the lack of significant correlation between OEF and other clinical manifestation of disease, such as the presence of silent cerebral infarcts. Secondly, 1/3 of our patients were on chronic transfusion therapy. Transfused blood causes a left shift in the hemoglobin dissociation curve because of the 2,3-DPG depletion during blood storage, 633 leading to a lower OEF for any brain pO2. Our non- transfused SS and S0 patients were taking hydroxyurea with good F response, which also promotes lower cerebral oxygen extraction. In fact, hemoglobin F concentration was retained as a significant predictor of OEF. Thirdly, 7/47 SCD patients had a milder genotype (SC or S + ). Older, sicker SCD patients in other studies have more circulating dense red blood cells, causing greater increases in their p50 and a larger right shift in the hemoglobin dissociation curve. 651 With this right shift, one would expect larger OEF for the same brain pO2 operating point 652 compared to the younger, healthier cohort with normal resting oxygen saturation in this current study. Lastly, our study used TRUST oximetry calibration curves specifically derived from normal human subjects and patients with sickle cell disease, rather than relying on calibrations derived from cow blood. 350,568,569 In fact, the SCD calibration used in this study represents a consensus from two independent investigations characterizing the relationships between hematocrit, blood T 2, and oxygen saturation in SCD patients. 569 Our cerebral oxygenation results were in agreement with several other recent publications as well as an all-inclusive historical sample of PET and Kety Schmidt oximetry data. Vaclavu et al. demonstrated decreased metabolic rate in SCD subjects relative to controls using TRUST oximetry and appropriate calibration curves. 653 Using MRI susceptometry (which is robust to 227 hemoglobin subtype), 351 Croal et al. demonstrated that OEF decreased proportionally with anemia severity in SCD patients. 654 Lastly, our observed dependence of CMRO2 mirrored decades of observations in a broad range of anemia syndromes. After corrections for age and hemoglobin level, CMRO2 was identical across disease type, study, measurement technique, and era. The relationship between CMRO2 and hemoglobin was maintained when the analysis was generalized to other medical conditions in which anemia was a confounding variable. The decrease in CMRO2 in our results could have several physiological explanations. Previously we have shown diffuse brain volume loss, 514,614 higher than normal rates of silent cerebral infarcts, 544,617,655 impaired resting state functional neural activity 406,656 and impaired neurocognitive function in chronically anemic subjects compared to healthy controls. 657–659 It is plausible to speculate that these neurovascular biomarkers could be associated with decreased CMRO2, although we were not able to demonstrate such an association in this paper. In SCD, low hemoglobin and acute anemic events are the strongest predictors of new silent stroke. 660 Additionally, stroke is most common in children (~6 years old) when CMRO 2 is also the highest. 660 Therefore, understanding the mechanism and clinical significance of diminished OEF and CMRO2 is not merely an academic enterprise. Low CMRO2 could also be explained by a protective downregulation of metabolic activity in anemic patients. Downregulation of metabolism could be induced by chronically impaired oxygen carrying capacity and long-term exposure to hypoxia. Hypoxic hypometabolism has been demonstrated in animal models to conserve oxygen and protect against hypoxic ischemic injuries. 661 Even though humans are believed not to exhibit hypoxic hypometabolism, 661 a previous study has demonstrated a reduction in metabolic heat and body temperature in response to acute hypoxia in humans. 662 Additionally, a study on native dwellers of high-altitude regions has shown that humans exposed to chronic hypoxia demonstrate a reduction in brain glucose metabolism especially in cerebral regions responsible for higher cortical functions, 663 thus indicating that protective hypometabolism could be a contributor to decreased CMRO2 in our anemic subjects. 228 However, we believe that the reduction in OEF and CMRO2 observed in anemic subjects represents a predictable, physiologic consequence of compensatory hyperemia. 396,397 Hyperemia reduces capillary transit time 203,427,664 and favors shorter mean cerebral capillary path lengths. 339 Because microvascular oxygen unloading requires sufficient residence time in the capillary network for efficient oxygen extraction, elevated CBF limits oxygen extraction. The resulting physiology is analogous to the disruption of microvascular oxygen exchange in conditions such as arteriovenous fistula. Even though our subjects do not have anatomical arteriovenous malformations, microvascular cerebral shunting has been suggested in SCD patients 285 and is associated with lower OEF. 618 Importantly, this physiology does not appear to be limited to the cerebral circulation. Our venous brachial oxygenation data, as well as prior reports of arterialization of peripheral blood flow in SCD 665 suggests a common underlying cerebral and peripheral vascular response to chronic anemia, strengthening confidence in our methods and conclusions. Furthermore, mathematical models of capillary oxygen transport provide insights into the reduced OEF with hyperemia. 666 Increased CBF causes higher heterogeneity of blood flow and transit time in microvascular networks, leading to nonlinear oxygen-metabolism coupling, decreased OEF and lower oxygen consumption. 666,667 Whereas some microvascular beds have normal capillary transit, other beds can demonstrate a phenomenon similar to physiological shunting with vasodilated vessels, abnormally high flow and inefficient oxygen unloading (Supplemental Figure 8.S3). These shunt-like beds contribute more flow to draining veins, leading to low overall OEF. Chronically, anemia and hyperemia damage the microvasculature potentially exacerbating physiological increases in capillary transit time heterogeneity and leading to a state of lower microvascular oxygen availability. 37,668–670 229 Limitations Our study was cross-sectional, only reflected steady state conditions, and was biased toward young adults who have preserved resting oxygen delivery. Microvascular disease and impaired vascular reactivity increase with age in the general population 671 and may be accelerated in hemoglobinopathy patients. 672 SCD patients at our institution have low stroke rates compared to the SCD populations in other studies. 618,673 Thus, lower OEF might not have been evident in an older cohort. On the other hand, since our study did not include children younger than 12 years of age whose CMRO2 values are higher compared to young adults, our results could not be fairly compared with other work on pediatric SCD subjects. 428 Additionally, our oximetry measurements assumed saturation values acquired in the sagittal sinus was similar to whole-brain venous saturation. Although this is an accurate assumption in healthy controls, 354 SCD patients may preserve cortical perfusion at the expense of deeper structures. 395 The study would also have been strengthened by studying CMRO2 after hemoglobin manipulations, such as transfusions or phlebotomy, to build inference for causality between hemoglobin levels and CMRO2. Our historical reference data was drawn from only eight studies because there was a limited number of papers that included data for individual patients. We mitigated this shortcoming in Supplemental Figure 8.S2 by expanding our selection criteria to include disorders in which CMRO2 was likely to be abnormal and were able to replicate the relationship between Hb and CMRO2. Since newer studies tended to only report group means instead of individual values, our references in Figure 8.2 had to be drawn from older studies (1930s-1980s). Since these references were from ~50 years ago, differences in OEF and CMRO2 values between historical data and data in the current study could still be expected, even within control subjects. Even though these references were acquired with gold-standard methods, it would be useful to obtain more recent datasets from different acquisitions methods and test sites to further assess the relationship between OEF, CMRO2 and anemia severity. 230 Conclusion In summary, using a hemoglobin specific T2 oximetry calibration, 569 we demonstrated that patients with SCD and other chronic anemias have lower OEF and CMRO2 than control subjects, proportional to their anemia severity. CMRO2 data agreed well with previously published study cohorts. The attenuated oxygen extraction at lower oxygen delivery in chronically anemic patients can best be explained by relative increases in non-nutritive flow, however this hypothesis awaits further experimental validation. 231 8.5. Supplemental Information 8.5.1. Supplemental Tables Supplemental Table 8.S1. Summary of different calibration models and their usage. Human HbA 350 For healthy controls, sickle cell trait and non-sickle anemic subjects ",!hh = 77.5××(1 – ) " +27.8×(1 – ) " +6.95×+2.34 Li-Bush HbS 569 For non-transfused SCD patients ",*,.. = 196.8××(1 − ) " +16.7×(1 − ) " −6.6×+8.6 Mixture HbS- weighted 569 For transfused SCD patients ",*,.. = ' ××(1 − ) " + " ×(1 − ) " + F ×+ P where =t1− T0^% u8% u× T0< + t T0^% u8% u × T0^ with T0< as coefficients from Human HbA model and T0^ as coefficients from Li-Bush HbS model 232 Supplemental Table 8.S2. Univariate and multivariate predictors of cerebral OEF, CMRO2 and brachial OEF. Arrows, r 2 and p are the direction, coefficient of determination and p-value of the regression respectively. Correlation direction is not available for categorical variables. Univariate Predictor Cerebral OEF CMRO2 Brachial OEF Hemoglobin r 2 =0.24, p<0.01, ↑ r 2 =0.17, p<0.01, ↑ r 2 =0.14, p<0.01, ↑ Group r 2 =0.30, p<0.01, – r 2 =0.15, p<0.01, – r 2 =0.22, p<0.01, – CBF r 2 =0.26, p<0.01, ↓ – r 2 =0.05, p=0.04, ↓ HbS r 2 =0.12, p<0.01, ↓ r 2 =0.05, p=0.03, ↓ – Transfusion Status r 2 =0.11, p<0.01, ↓ – r 2 =0.30, p<0.01, ↓ Fetal Hemoglobin r 2 =0.11, p<0.01, ↓ r 2 =0.06, p<0.01, ↓ – White Blood Cells – – r 2 =0.07, p=0.01, ↓ Platelets – – – Reticulocytes r 2 =0.13, p<0.01, ↓ – r 2 =0.06, p=0.02, ↓ Cell-free Hemoglobin r 2 =0.05, p=0.02, ↓ – – RBC Distribution Width r 2 =0.18, p<0.01, ↓ r 2 =0.09, p<0.01, ↓ – Hydroxyurea Status – – – Grey Matter Volume – – – White Matter Volume – – – Cortical Thickness – – – Cortical Surface Area – – – Silent Infarct Presence – – – Multivariate Predictor r 2 =0.29, p<0.01 r 2 =0.17, p<0.01 r 2 =0.31, p<0.01 Hemoglobin r 2 =0.24 r 2 =0.17 r 2 =0.14 HbF r 2 =0.05 – – Transfusion Status – – r 2 =0.17 233 Supplemental Table 8.S3. Mean demographic, clinical and oxygenation parameters extracted from 8 historical datasets. Age (Years) PaCO2 (mmHg) Hb (g/dL) CBF OEF (%) CMRO2 Herold, 621 27.5±8.7 – 8.1±1.1 65.0±12.1 43.5±6.7 3.0±0.4 Frackowiak, 622 47.6±14.8 – 14.1±1.8 44.2±11.1 48.4±7.6 4.0±0.9 Heyman, 623 33.0±17.1 38.4±3.8 7.2±1.9 68.2±14.1 41.8±9.9 2.6±0.7 Scheinberg, 624,625 48.0±18.8 43.2±1.6 10.9±3.6 58.6±28.4 36.3±7.1 2.9±1.1 Stewart, 626 50.8±9.7 – 8.1±2.5 46.4±28.2 53.7±7.8 2.3±0.3 Fazekas, 627 41.9±14.1 43.2±3.0 10.9±1.1 60.6±13.3 39.8±7.9 3.5±0.5 Mangold, 628 23.8±2.9 42.5±4.8 14.1±0.9 63.1±16.3 30.9±3.8 3.6±0.7 Kety, 629 25.3±2.9 44.9±4.8 13.1±0.6 68.5±10.5 36.7±4.5 4.5±0.7 234 Supplemental Table 8.S4. Inclusion and exclusion notes for historical references. Reasons for exclusion were provided for references that provided individual data for CMRO2. * denotes the eight studies in anemia and healthy controls. Historical References: Inclusions Reference Disease condition Inclusion Note Exclusion Note Fazekas, 627 * Hypertension, vascular insufficiency, normal Normal and hypertensive subjects without vascular disease, normotensive subjects with vascular insufficiency. Hypertensive patients with vascular insufficiency excluded. Data under hyperventilation also excluded. Frackowiak, 622 * Normal Weighted average of PET measurements in white and grey matter. Griffiths, 674 Cerebral aneurysm Pre-repair cerebral aneurysm. Cerebral aneurysm under induced hypotension. Hafkenschiel, 675 Hypertension Pre- and post- sympathectomy. Data under head-up tilt excluded. Harmel, 676 Hypertension, normotension, Parkinson’s Post-treatment with bilateral stellate ganglion block excluded. Herold, 621 * SCD Normal subjects excluded due to lack of hemoglobin values. Heyman, 623 * SCD, blood loss, iron-deficiency, aplastic anemia Heyman, 677 Uremia Ibaraki, 611 Unilateral carotid stenosis, normal Patient age was regression imputed from hemoglobin correlation reported. Kennedy, 678 Children Normal children. Children with intellectual disability. Kety, 679 Hypertension Kety, 680 Increased intracranial pressure Brain tumor patients. Kety, 629 * Normal Data under hyperventilation excluded. Kety, 681 Schizophrenia Pre-treatment schizophrenia. Post-treatment schizophrenia. 235 Kety, 682 Severe diabetes Diabetic acidosis and diabetic coma. Kuwabara, 683 Renal failure Patient age was mean imputed in subgroups. Mangold, 628 * Normal Normal subjects with regular sleep. Normal subjects with lack of sleep. Scheinberg, 624 * Pernicious anemia Scheinberg, 625 Normal Normal subjects under baseline. Subjects under vigorous exercise excluded. Scheinberg, 684 Hypothyroidism Sensenbach, 685 Hyperthyroidism, hypothyroidism, euthyroidism Euthyroidism values were post-treatment values for hyperthyroidism patients. Shenkin, 686 Carotid ligation Values averaged between ligated and patent sides of carotid ligation. Shenkin, 687 Hypertension Pre- and post- sympathectomy. Shenkin, 688 Post-operation brain tumor Data under caffeine influence excluded. Sokoloff, 689 Hyperthyroidism Pre-treatment hyperthyroidism. Post-treatment hyperthyroidism. Stewart, 626 * Pernicious anemia Wechsler, 690 Anesthesia Values averaged between right and left hemispheres. Wechsler, 691 Hospitalized adults Adults hospitalized for various conditions at baseline. Post-treatment with aminophylline in adults. Historical references: Exclusions Reference Disease condition Exclusion Note Bentinck, 692 Endocrine disease Hemoglobin unreported and unable to calculate. Himwich, 693 Anesthesia Hemoglobin unreported and unable to calculate. Patterson, 694 Neurosyphilis Hemoglobin unreported and unable to calculate. Scheinberg, 695 Heart failure Hemoglobin unreported and unable to calculate. Scheinberg, 696 Hyperthyroidism Hemoglobin unreported and unable to calculate. Scheinberg, 697 Nicotinic acid Hemoglobin unreported and unable to calculate. Scheinberg, 698 Normal Hemoglobin unreported and unable to calculate. Scheinberg, 699 Vascular disease Hemoglobin unreported and unable to calculate. Shenkin, 700 Arteriovenous shunt Consisting of only two subjects. 236 8.5.2. Supplemental Figures Supplemental Figure 8.S1. (A) Correlation between age-adjusted CMRO2 and PaCO2. (B) Correlation between age- and PaCO2-adjusted CMRO2 and hemoglobin values. 237 Supplemental Figure 8.S2. Age-adjusted CMRO2 values demonstrated positive linear relation with hemoglobin levels and disease conditions, including anesthesia, 690 post-operation brain tumor, 688 severe diabetes, 682 renal failure, 683 uremia, 677 hypo- and hyperthyroidism, 684,685,689 post-carotid ligation, 686 hypertension, 627,675,676,679,687 increased intracranial pressure, 680 Parkinson’s, 676 cerebral aneurysm, 674 schizophrenia, 681 chronic vascular insufficiency, 627 unilateral stenosis, 611 hospitalized adults 691 and children. 678 Data on chronic anemias, SCD and healthy controls were drawn from references in Figure 4 and 5. (A) Linear correlation between age-adjusted CMRO2 and hemoglobin in normal subjects and 238 subjects with primary anemic conditions (sickle cell anemia, pernicious anemia, iron deficiency anemia and acute blood loss) not thought to impact CMRO2. (B) Correlation between age-adjusted CMRO2 and hemoglobin in other conditions possibly associated with abnormal CMRO2 who have concomitant anemia; the fit is statistically identical to (A). (C) After correcting for age and hemoglobin, CMRO2 residuals are shown for difference disease conditions in panel (B). Subjects with unilateral carotid stenosis, chronic vascular insufficiency, post-carotid ligation, cerebral aneurysm and schizophrenia did not demonstrate a systematic change in CMRO2 relative to degree of anemia. Hypertension and post-sympathectomy hypertension also did not demonstrate a change in metabolism. In contrast, subjects with renal failure and uremia showed impaired CMRO2 as has been previously described. 701,702 Additionally, subjects with diabetic acidosis or diabetic coma showed lower than expected CMRO2 likely due to their confused or comatose states, and post-operation brain tumor and under-anesthesia patients had low CMRO2 compared to their hemoglobin levels. 703 Normal children and patients with hyperthyroidism had increased metabolic rate which has also been reported, 704–706 whereas hypothyroidism subjects did not show a change in CMRO2. Data from normal hospitalized adults also had higher CMRO2, but there was insufficient physiological data in this paper to determine whether there were mitigated clinical findings, such as fever, that would explain elevated CMRO2 in this cohort. 707 * denotes residuals that are significant different from zero value (p<0.01). 239 Supplemental Figure 8.S3. Illustration of how high capillary transit time heterogeneity affects OEF measurement. Whereas some microvascular beds have normal capillary transit under hyperemic condition, other beds have abnormally high flow and inefficient oxygen unloading, contributing more flow to the draining veins, leading to low overall measurement of OEF. 240 Chapter 9 : Thesis Conclusion The study of hemodynamic impairment and the associated stroke risk can serve as a model of accelerated neurovascular disease, not only in the case of hematologic conditions but also in normal aging. The risk of silent cerebral infarcts increases with a multitude of factors, including age, blood pressure, obesity, and history of prior cerebrovascular accidents. Notably, patients with vascular diseases and hematologic diseases are at heightened risk for SCIs due to their impaired ability to maintain adequate oxygen delivery to deep watershed brain areas. This work postulates that the development of SCIs can be explained by a two-hit hypothesis: chronic tissue hypoxia coupled with acute stressors. Specifically in chronic anemia patients, to compensate for the low oxygen carrying capacity, blood vessels are fully dilated to maintain cerebral blood flow. Under acute events such as hemorrhage, these vessels cannot vasodilate further, leading to reduced perfusion and ischemia in deep white matter areas of the brain. Since MRI methodologies to evaluate cerebral hemodynamics are at the core of stroke prevention, diagnostics, and therapies, this thesis looked into various MRI techniques to measure cerebral perfusion and oxygenation. Notably, this work explored non-contrast perfusion imaging as an alternative to gadolinium-based DSC MRI. Gadolinium experiments have been implicated in nephrogenic systemic fibrosis as well as long-term contrast accumulation in different tissues, which highlights the need for non-contrast imaging options especially for at-risk populations. To address this need, this work proposed a novel deoxygenation-based DSC MRI technique that uses respiratory challenges to induce susceptibility-weighted signal losses similar to the effects of gadolinium contrast. Multiple respiratory paradigms with varying levels of oxygen and carbon-dioxide were explored, demonstrating good agreement with clinical reference techniques in a cohort of anemic subjects and healthy volunteers. The initial work using fixed- 241 inspired gas challenge was improved with the addition of a computer-controlled gas blender, allowing more precise and standardized respiratory stimuli, independent of individual variations in breathing pattern, metabolic rate, and lung function. Overall, these deoxygenation-based respiratory challenges represent a non-contrast alternative for perfusion imaging, especially for renally-impaired and pediatric patient populations. In addition to perfusion imaging, this thesis also examined a T2 oximetry MRI technique to measure blood T2 values and convert them to venous saturation using carefully validated and disease-specific calibrations. Previous works have applied T2 oximetry to measure venous saturation in sickle cell disease. However, inappropriate bovine calibration derived from a limited range of hematocrit was used, thus preventing accurate estimates of oxygen extraction fraction in this patient population. Due to the difference in red cell morphology and permeability, sickled erythrocytes have different magnetic characteristics from normal red blood cells and require a separate calibration. Therefore, this work proposed a unified SCD calibration using pooled datasets from two independent groups to serve as a reference standard for future T2 oximetry studies in SCD. Using appropriate and disease-specific calibrations, this work examined a large cohort of sickle cell and non-sickle anemia subjects and observed decreased OEF and CMRO2 compared to healthy controls. This impaired extraction capability could be attributed to physiological shunting, in which certain vascular beds demonstrated abnormally high flow and inefficient oxygen unloading. Furthermore, the reduced CMRO2 was proportional to the degree of anemia severity and showed marked agreement with historical data acquired by MRI-independent techniques. Additional predictors of CMRO2 included fetal hemoglobin and transfusion status, which implicates the left-shifted hemoglobin dissociation curve and indicates the need for frequent monitoring of cerebral oxygenation in SCD subjects. Overall, the lower cerebral oxygen availability and utilization in this population suggests an imbalance in oxygen supply and demand, which can potentially lead to brain tissue hypoxia, silent cerebral infarcts, and ischemic strokes. 242 Future Work This thesis proposed several non-invasive MRI methodologies to measure cerebral perfusion and oxygenation. The novel perfusion-weighted deoxygenation-based DSC technique has demonstrated good agreement with other clinical references in healthy volunteers but has not been assessed in other pathologies, including overt strokes, brain tumors, and neurodegenerative diseases. Further clinical trials on these patient populations are required to evaluate this technique’s clinical value in diagnostics and therapies. On the other hand, venous saturation imaging using T2 oximetry has been widely applied in many different pathologies. However, care must be taken, even if retrospectively, to apply appropriate and disease specific T2 oximetry calibrations, especially for hemoglobinopathies in which the magnetic characteristics of red blood cells are abnormal. After acquiring OEF and CMRO2 estimates by using appropriate calibrations, biomarker analysis to identify predictors of decreased oxygenation and SCI risk is needed on a larger and more heterogenous cohort, balanced across subtypes of sickle cell disease and congenital anemias. Lastly, neuropsychological correlates of impaired hemodynamics must be explored to assess for brain- behavior relationships and informs early prevention and intervention for at-risk patients. 243 References 1. Magistretti PJ, Allaman I. A cellular perspective on brain energy metabolism and functional imaging. Neuron. 2015;86(4):883-901. doi:10.1016/J.NEURON.2015.03.035 2. Feigin VL, Lawes CM, Bennett DA, Barker-Collo SL, Parag V. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review. Lancet Neurol. 2009;8(4):355-369. doi:10.1016/S1474-4422(09)70025-0 3. Mathers CD, Lopez AD, Murray CJL. The Burden of Disease and Mortality by Condition: Data, Methods, and Results for 2001. Glob Burd Dis Risk Factors. 2006. https://www.ncbi.nlm.nih.gov/books/NBK11808/. Accessed January 18, 2022. 4. Thrift AG, Arabshahi S. Is stroke incidence in low- to middle-income countries driven by economics? Int J Stroke. 2012;7(4):307-308. doi:10.1111/J.1747-4949.2012.00819.X 5. Sposato LA, Saposnik G. Gross domestic product and health expenditure associated with incidence, 30-day fatality, and age at stroke onset: a systematic review. Stroke. 2012;43(1):170-177. doi:10.1161/STROKEAHA.111.632158 6. Aronowski J, Zhao X. Molecular pathophysiology of cerebral hemorrhage: secondary brain injury. Stroke. 2011;42(6):1781-1786. doi:10.1161/STROKEAHA.110.596718 7. Unnithan AKA, Mehta P. Hemorrhagic Stroke. StatPearls. August 2021. https://www.ncbi.nlm.nih.gov/books/NBK559173/. Accessed January 18, 2022. 8. Montaño A, Hanley DF, Hemphill JC. Hemorrhagic stroke. Handb Clin Neurol. 2021;176:229-248. doi:10.1016/B978-0-444-64034-5.00019-5 9. Maida CD, Norrito RL, Daidone M, Tuttolomondo A, Pinto A. Neuroinflammatory Mechanisms in Ischemic Stroke: Focus on Cardioembolic Stroke, Background, and Therapeutic Approaches. Int J Mol Sci. 2020;21(18):1-33. doi:10.3390/IJMS21186454 10. Sennfält S, Norrving B, Petersson J, Ullberg T. Long-Term Survival and Function After Stroke. Stroke. 2018;50(1):53-61. doi:10.1161/STROKEAHA.118.022913 11. Paul S, Candelario-Jalil E. Emerging neuroprotective strategies for the treatment of ischemic stroke: An overview of clinical and preclinical studies. Exp Neurol. 2021;335. doi:10.1016/J.EXPNEUROL.2020.113518 12. Allen CL, Bayraktutan U. Risk factors for ischaemic stroke. Int J Stroke. 2008;3(2):105- 116. doi:10.1111/J.1747-4949.2008.00187.X 13. Doll DN, Barr TL, Simpkins JW. Cytokines: their role in stroke and potential use as biomarkers and therapeutic targets. Aging Dis. 2014;5(5):294-306. doi:10.14336/AD.2014.0500294 244 14. Hurford R, Sekhar A, Hughes TAT, Muir KW. Diagnosis and management of acute ischaemic stroke. Pract Neurol. 2020;20(4):304. doi:10.1136/PRACTNEUROL-2020- 002557 15. Astrup J, Siesjö BK, Symon L. Thresholds in cerebral ischemia - the ischemic penumbra. Stroke. 1981;12(6):723-725. doi:10.1161/01.STR.12.6.723 16. Mandalaneni K, Rayi A, Jillella D V. Stroke Reperfusion Injury. StatPearls. November 2021. https://www.ncbi.nlm.nih.gov/books/NBK564350/. Accessed January 18, 2022. 17. Gauberti M, Lapergue B, De Lizarrondo SM, et al. Ischemia-Reperfusion Injury After Endovascular Thrombectomy for Ischemic Stroke. Stroke. 2018;49(12):3071-3074. doi:10.1161/STROKEAHA.118.022015 18. Ganzer CA, Barnes A, Uphold C, Jacobs AR. Transient Ischemic Attack and Cognitive Impairment: A Review. J Neurosci Nurs. 2016;48(6):322-327. doi:10.1097/JNN.0000000000000236 19. Price TR, Manolio TA, Kronmal RA, et al. Silent brain infarction on magnetic resonance imaging and neurological abnormalities in community-dwelling older adults. The Cardiovascular Health Study. CHS Collaborative Research Group. Stroke. 1997;28(6):1158-1164. doi:10.1161/01.STR.28.6.1158 20. Vermeer SE, Longstreth WT, Koudstaal PJ. Silent brain infarcts: a systematic review. Lancet Neurol. 2007;6(7):611-619. doi:10.1016/S1474-4422(07)70170-9 21. Zhu YC, Dufouil C, Tzourio C, Chabriat H. Silent brain infarcts: a review of MRI diagnostic criteria. Stroke. 2011;42(4):1140-1145. doi:10.1161/STROKEAHA.110.600114 22. Vermeer SE, Koudstaal PJ, Oudkerk M, Hofman A, Breteler MMB. Prevalence and risk factors of silent brain infarcts in the population-based Rotterdam Scan Study. Stroke. 2002;33(1):21-25. doi:10.1161/HS0102.101629 23. Van Der Land V, Zwanenburg JJM, Fijnvandraat K, et al. Cerebral lesions on 7 tesla MRI in patients with sickle cell anemia. Cerebrovasc Dis. 2015;39(3-4):181-189. doi:10.1159/000373917 24. DeBaun MR, Armstrong FD, McKinstry RC, Ware RE, Vichinsky E, Kirkham FJ. Silent cerebral infarcts: a review on a prevalent and progressive cause of neurologic injury in sickle cell anemia. Blood. 2012;119(20):4587-4596. doi:10.1182/blood-2011-02-272682 25. Gupta A, Giambrone AE, Gialdini G, et al. Silent Brain Infarction and Risk of Future Stroke: A Systematic Review and Meta-Analysis. Stroke. 2016;47(3):719. doi:10.1161/STROKEAHA.115.011889 26. Diener HC, Eikelboom J, Connolly SJ, et al. Apixaban versus aspirin in patients with atrial fibrillation and previous stroke or transient ischaemic attack: a predefined subgroup analysis from AVERROES, a randomised trial. Lancet Neurol. 2012;11(3):225-231. doi:10.1016/S1474-4422(12)70017-0 245 27. Fanning JP, Wong AA, Fraser JF. The epidemiology of silent brain infarction: a systematic review of population-based cohorts. BMC Med. 2014;12(1):119. doi:10.1186/S12916-014- 0119-0 28. Lee SC, Park SJ, Ki HK, et al. Prevalence and risk factors of silent cerebral infarction in apparently normal adults. Hypertens (Dallas, Tex 1979). 2000;36(1):73-77. doi:10.1161/01.HYP.36.1.73-A 29. Krishnan KRR, Goli V, Ellinwood EH, France RD, Blazer DG, Nemeroff CB. Leukoencephalopathy in patients diagnosed as major depressive. Biol Psychiatry. 1988;23(5):519-522. doi:10.1016/0006-3223(88)90025-X 30. Vermeer SE, Prins ND, den Heijer T, Hofman A, Koudstaal PJ, Breteler MMB. Silent brain infarcts and the risk of dementia and cognitive decline. N Engl J Med. 2003;348(13):1215- 1222. doi:10.1056/NEJMOA022066 31. De Groot JC, De Leeuw FE, Oudkerk M, et al. Periventricular cerebral white matter lesions predict rate of cognitive decline. Ann Neurol. 2002;52(3):335-341. doi:10.1002/ANA.10294 32. Wang W, Enos L, Gallagher D, et al. Neuropsychologic performance in school-aged children with sickle cell disease: a report from the Cooperative Study of Sickle Cell Disease. J Pediatr. 2001;139(3):391-397. doi:10.1067/MPD.2001.116935 33. Blum S, Luchsinger JA, Manly JJ, et al. Memory after silent stroke: Hippocampus and infarcts both matter. Neurology. 2012;78(1):38. doi:10.1212/WNL.0B013E31823ED0CC 34. Schatz J, Brown RT, Pascual JM, Hsu L, DeBaun MR. Poor school and cognitive functioning with silent cerebral infarcts and sickle cell disease. Neurology. 2001;56(8):1109-1111. doi:10.1212/WNL.56.8.1109 35. Berkelhammer LD, Williamson AL, Sanford SD, et al. Neurocognitive sequelae of pediatric sickle cell disease: a review of the literature. Child Neuropsychol. 2007;13(2):120-131. doi:10.1080/09297040600800956 36. Coloigner J, Vu C, Borzage M, et al. Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia Using BOLD MRI and Near-Infrared Spectroscopy. J Magn Reson Imaging. 2020;52(5):1400-1412. doi:10.1002/JMRI.27210 37. Jespersen SN, Østergaard L. The Roles of Cerebral Blood Flow, Capillary Transit Time Heterogeneity, and Oxygen Tension in Brain Oxygenation and Metabolism. J Cereb Blood Flow Metab. 2012;32(2):264-277. doi:10.1038/jcbfm.2011.153 38. Østergaard L, Engedal TS, Aamand R, et al. Capillary transit time heterogeneity and flow- metabolism coupling after traumatic brain injury. J Cereb Blood Flow Metab. 2014;34(10):1585-1598. doi:10.1038/jcbfm.2014.131 39. Liebetrau M, Steen B, Hamann GF, Skoog I. Silent and symptomatic infarcts on cranial computerized tomography in relation to dementia and mortality: a population-based study in 85-year-old subjects. Stroke. 2004;35(8):1816-1820. doi:10.1161/01.STR.0000131928.47478.44 246 40. Bokura H, Kobayashi S, Yamaguchi S, et al. Silent brain infarction and subcortical white matter lesions increase the risk of stroke and mortality: a prospective cohort study. J Stroke Cerebrovasc Dis. 2006;15(2):57-63. doi:10.1016/J.JSTROKECEREBROVASDIS.2005.11.001 41. Shinkawa A, Ueda K, Kiyohara Y, et al. Silent cerebral infarction in a community-based autopsy series in Japan. The Hisayama Study. Stroke. 1995;26(3):380-385. doi:10.1161/01.STR.26.3.380 42. Bergui M, Castagno D, D’Agata F, et al. Selective Vulnerability of Cortical Border Zone to Microembolic Infarct. Stroke. 2015;46(7):1864-1869. doi:10.1161/STROKEAHA.114.008194 43. Pegelow CH, Macklin EA, Moser FG, et al. Longitudinal changes in brain magnetic resonance imaging findings in children with sickle cell disease. Blood. 2002;99(8):3014- 3018. http://www.ncbi.nlm.nih.gov/pubmed/11929794. Accessed March 26, 2018. 44. Navarro-Orozco D, Sánchez-Manso JC. Neuroanatomy, Middle Cerebral Artery. StatPearls. July 2021. https://www.ncbi.nlm.nih.gov/books/NBK526002/. Accessed January 18, 2022. 45. Ropper A, Samuels M, Klein J, Prasad S. Adams and Victor’s Principles of Neurology. McGraw Hill / Medical; 2019. 46. Rafieian-Kopaei M, Setorki M, Doudi M, Baradaran A, Nasri H. Atherosclerosis: Process, Indicators, Risk Factors and New Hopes. Int J Prev Med. 2014;5(8):927. /pmc/articles/PMC4258672/. Accessed January 18, 2022. 47. Hansson GK, Robertson AKL, Söderberg-Nauclér C. Inflammation and atherosclerosis. Annu Rev Pathol. 2006;1:297-329. doi:10.1146/ANNUREV.PATHOL.1.110304.100100 48. Libby P, Ridker PM, Maseri A. Inflammation and atherosclerosis. Circulation. 2002;105(9):1135-1143. doi:10.1161/HC0902.104353 49. Warboys CM, Amini N, De Luca A, Evans PC. The role of blood flow in determining the sites of atherosclerotic plaques. F1000 Med Rep. 2011;3(1). doi:10.3410/M3-5 50. Huang S, Guo ZN, Shi M, Yang Y, Rao M. Etiology and pathogenesis of Moyamoya Disease: An update on disease prevalence. Int J Stroke. 2017;12(3):246-253. doi:10.1177/1747493017694393 51. Wakai K, Tamakoshi A, Ikezaki K, et al. Epidemiological features of moyamoya disease in Japan: findings from a nationwide survey. Clin Neurol Neurosurg. 1997;99 Suppl 2(SUPPL. 2). doi:10.1016/S0303-8467(97)00031-0 52. Fox BM, Dorschel KB, Lawton MT, Wanebo JE. Pathophysiology of Vascular Stenosis and Remodeling in Moyamoya Disease. Front Neurol. 2021;12. doi:10.3389/FNEUR.2021.661578 247 53. Kaku Y, Morioka M, Ohmori Y, et al. Outer-diameter narrowing of the internal carotid and middle cerebral arteries in moyamoya disease detected on 3D constructive interference in steady-state MR image: is arterial constrictive remodeling a major pathogenesis? Acta Neurochir (Wien). 2012;154(12):2151-2157. doi:10.1007/S00701-012-1472-4 54. Lechner K, von Schacky C, McKenzie AL, et al. Lifestyle factors and high-risk atherosclerosis: Pathways and mechanisms beyond traditional risk factors. Eur J Prev Cardiol. 2020;27(4):394-406. doi:10.1177/2047487319869400 55. Rupareliya C, Lui F. Moyamoya Disease. StatPearls. September 2021. https://www.ncbi.nlm.nih.gov/books/NBK535455/. Accessed January 18, 2022. 56. Kim JS. Moyamoya Disease: Epidemiology, Clinical Features, and Diagnosis. J Stroke. 2016;18(1):2. doi:10.5853/JOS.2015.01627 57. Kuroda S, Clinical Centers, Data Management Center, et al. Asymptomatic moyamoya disease: literature review and ongoing AMORE study. Neurol Med Chir (Tokyo). 2015;55(3):194-198. doi:10.2176/NMC.RA.2014-0305 58. Nanba R, Kuroda S, Takeda M, et al. Clinical features and outcomes of 10 asymptomatic adult patients with moyamoya disease. No Shinkei Geka. https://pubmed.ncbi.nlm.nih.gov/14719442/. Published 2003. Accessed January 18, 2022. 59. Kim SK, Seol HJ, Cho BK, et al. Moyamoya disease among young patients: its aggressive clinical course and the role of active surgical treatment. Neurosurgery. 2004;54(4):840- 846. doi:10.1227/01.NEU.0000114140.41509.14 60. Kuroda S, Hashimoto N, Yoshimoto T, Iwasaki Y. Radiological findings, clinical course, and outcome in asymptomatic moyamoya disease: results of multicenter survey in Japan. Stroke. 2007;38(5):1430-1435. doi:10.1161/STROKEAHA.106.478297 61. Li J, Smith A, Crouch S, Oliver S, Roman E. Estimating the prevalence of hematological malignancies and precursor conditions using data from Haematological Malignancy Research Network (HMRN). Cancer Causes Control. 2016;27(8):1019. doi:10.1007/S10552-016-0780-Z 62. Muz B, De La Puente P, Azab F, Luderer M, Azab AK. The role of hypoxia and exploitation of the hypoxic environment in hematologic malignancies. Mol Cancer Res. 2014;12(10):1347-1354. doi:10.1158/1541-7786.MCR-14-0028 63. Muz B, Puente P de la, Azab F, Azab AK. The role of hypoxia in cancer progression, angiogenesis, metastasis, and resistance to therapy. Hypoxia. 2015;3:83. doi:10.2147/HP.S93413 64. Quintas S, Rogado J, Gullón P, et al. Predictors of unknown cancer in patients with ischemic stroke. J Neurooncol. 2018;137(3):551-557. doi:10.1007/S11060-017-2741-0 65. Bang OY, Chung JW, Lee MJ, Seo WK, Kim GM, Ahn MJ. Cancer-Related Stroke: An Emerging Subtype of Ischemic Stroke with Unique Pathomechanisms. J stroke. 2020;22(1):1-10. doi:10.5853/JOS.2019.02278 248 66. Salazar-Camelo RA, Moreno-Vargas EA, Cardona AF, Bayona-Ortiz HF. Ischemic stroke: A paradoxical manifestation of cancer. Crit Rev Oncol Hematol. 2021;157. doi:10.1016/J.CRITREVONC.2020.103181 67. McLean E, Cogswell M, Egli I, Wojdyla D, De Benoist B. Worldwide prevalence of anaemia, WHO Vitamin and Mineral Nutrition Information System, 1993-2005. Public Health Nutr. 2009;12(4):444-454. doi:10.1017/S1368980008002401 68. Stoltzfus RJ. Iron deficiency: global prevalence and consequences. Food Nutr Bull. 2003;24(4 Suppl):S99-S103. doi:10.1177/15648265030244S206 69. Stevens GA, Finucane MM, De-Regil LM, et al. Global, regional, and national trends in haemoglobin concentration and prevalence of total and severe anaemia in children and pregnant and non-pregnant women for 1995-2011: a systematic analysis of population- representative data. Lancet Glob Heal. 2013;1(1). doi:10.1016/S2214-109X(13)70001-9 70. Yang F, Liu X, Zha P. Trends in Socioeconomic Inequalities and Prevalence of Anemia Among Children and Nonpregnant Women in Low- and Middle-Income Countries. JAMA Netw Open. 2018;1(5):e182899. doi:10.1001/JAMANETWORKOPEN.2018.2899 71. Baynes RD. Assessment of iron status. Clin Biochem. 1996;29(3):209-215. doi:10.1016/0009-9120(96)00010-K 72. Beaumont C, Canonne-Hergaux F. Erythrophagocytosis and recycling of heme iron in normal and pathological conditions; regulation by hepcidin. Transfus Clin Biol. 2005;12(2):123-130. doi:10.1016/J.TRACLI.2005.04.017 73. DeLoughery TG. Iron Deficiency Anemia. Med Clin North Am. 2017;101(2):319-332. doi:10.1016/J.MCNA.2016.09.004 74. Beck KL, Conlon CA, Kruger R, Coad J. Dietary Determinants of and Possible Solutions to Iron Deficiency for Young Women Living in Industrialized Countries: A Review. Nutrients. 2014;6(9):3747. doi:10.3390/NU6093747s 75. Hudak L, Jaraisy A, Haj S, Muhsen K. An updated systematic review and meta-analysis on the association between Helicobacter pylori infection and iron deficiency anemia. Helicobacter. 2017;22(1). doi:10.1111/HEL.12330 76. Kaitha S, Bashir M, Ali T. Iron deficiency anemia in inflammatory bowel disease. World J Gastrointest Pathophysiol. 2015;6(3):62. doi:10.4291/WJGP.V6.I3.62 77. Lim CH, Kim SW, Kim WC, et al. Anemia after gastrectomy for early gastric cancer: Long- term follow-up observational study. World J Gastroenterol. 2012;18(42):6114. doi:10.3748/WJG.V18.I42.6114 78. Balendran S, Forsyth C. Non-anaemic iron deficiency. Aust Prescr. 2021;44(6):193-196. doi:10.18773/austprescr.2021.052 79. Abbaspour N, Hurrell R, Kelishadi R. Review on iron and its importance for human health. J Res Med Sci. 2014;19(2):164. /pmc/articles/PMC3999603/. Accessed January 18, 2022. 249 80. Chang YL, Hung SH, Ling W, Lin HC, Li HC, Chung SD. Association between ischemic stroke and iron-deficiency anemia: a population-based study. PLoS One. 2013;8(12). doi:10.1371/JOURNAL.PONE.0082952 81. Awad D, Kousa O, Essa A, et al. Acute ischemic stroke as initial manifestation of undiagnosed iron deficiency anemia: case-report and literature review. AME case reports. 2020;4:23-23. doi:10.21037/ACR-20-72 82. Mehta PJ, Chapman S, Jayam-Trouth A, Kurukumbi M. Acute ischemic stroke secondary to iron deficiency anemia: a case report. Case Rep Neurol Med. 2012;2012:1-5. doi:10.1155/2012/487080 83. DeBaun MR, Sarnaik SA, Rodeghier MJ, et al. Associated risk factors for silent cerebral infarcts in sickle cell anemia: low baseline hemoglobin, sex, and relative high systolic blood pressure. Blood. 2012;119(16):3684. doi:10.1182/BLOOD-2011-05-349621 84. Inzitari M, Studenski S, Rosano C, et al. Anemia is associated with the progression of white matter disease in older adults with high blood pressure: the Cardiovascular Health Study. J Am Geriatr Soc. 2008;56(10):1867. doi:10.1111/J.1532-5415.2008.01950.X 85. Helmi N, Bashir M, Shireen A, Ahmed IM. Thalassemia review: features, dental considerations and management. Electron Physician. 2017;9(3):4003. doi:10.19082/4003 86. Alpha and beta thalassemia - PubMed. https://pubmed.ncbi.nlm.nih.gov/19678601/. Accessed July 5, 2022. 87. Origa R. β-Thalassemia. Genet Med. 2017;19(6):609-619. doi:10.1038/GIM.2016.173 88. Aristizabal A, Merino S, Catediano E, Sasieta M, Aragües P, Navajas A. [Clinical consequences of alpha-thalassemia in the Basque Country, Spain. Impact of neonatal screening]. An Pediatr (Barc). 2015;83(2):85-88. doi:10.1016/J.ANPEDI.2014.10.014 89. Joshi DD, Nickerson HJ, McManus MJ. Hydrops Fetalis Caused by Homozygous α- Thalassemia and Rh Antigen Alloimmunization: Report of a Survivor and Literature Review. Clin Med Res. 2004;2(4):228. doi:10.3121/CMR.2.4.228 90. Viprakasit V, Ekwattanakit S. Clinical Classification, Screening and Diagnosis for Thalassemia. Hematol Oncol Clin North Am. 2018;32(2):193-211. doi:10.1016/J.HOC.2017.11.006 91. Cappellini MD, Musallam KM, Marcon A, Taher AT. Coagulopathy in Beta-Thalassemia: Current Understanding and Future Perspectives. Mediterr J Hematol Infect Dis. 2009;1(1):220090. doi:10.4084/MJHID.2009.029 92. Chen S, Eldor A, Barshtein G, et al. Enhanced aggregability of red blood cells of beta- thalassemia major patients. Am J Physiol. 1996;270(6 Pt 2). doi:10.1152/AJPHEART.1996.270.6.H1951 93. Pazgal I, Inbar E, Cohen M, Shpilberg O, Stark P. High incidence of silent cerebral infarcts in adult patients with beta thalassemia major. Thromb Res. 2016;144:119-122. doi:10.1016/J.THROMRES.2016.06.010 250 94. Manfrè L, Giarratano E, Maggio A, Banco A, Vaccaro G, Lagalla R. MR imaging of the brain: findings in asymptomatic patients with thalassemia intermedia and sickle cell- thalassemia disease. AJR Am J Roentgenol. 1999;173(6):1477-1480. doi:10.2214/AJR.173.6.10584785 95. Taher AT, Musallam KM, Nasreddine W, Hourani R, Inati A, Beydoun A. Asymptomatic brain magnetic resonance imaging abnormalities in splenectomized adults with thalassemia intermedia. J Thromb Haemost. 2010;8(1):54-59. doi:10.1111/J.1538- 7836.2009.03651.X 96. Musallam KM, Cappellini MD, Wood JC, Taher AT. Iron overload in non-transfusion- dependent thalassemia: a clinical perspective. Blood Rev. 2012;26 Suppl 1(SUPPL.1). doi:10.1016/S0268-960X(12)70006-1 97. Karimi M, Haghpanah S, Bagheri MH, Bordbar MR, Pishdad P, Rachmilewitz EA. Frequency and distribution of asymptomatic brain lesions in patients with β-thalassemia intermedia. Ann Hematol. 2012;91(12):1833-1838. doi:10.1007/S00277-012-1527-X 98. Kato GJ, Piel FB, Reid CD, et al. Sickle cell disease. Nat Rev Dis Prim. 2018;4. doi:10.1038/NRDP.2018.10 99. Sedrak A, Kondamudi NP. Sickle Cell Disease. StatPearls. November 2021. https://www.ncbi.nlm.nih.gov/books/NBK482384/. Accessed January 18, 2022. 100. Piel FB, Hay SI, Gupta S, Weatherall DJ, Williams TN. Global burden of sickle cell anaemia in children under five, 2010-2050: modelling based on demographics, excess mortality, and interventions. PLoS Med. 2013;10(7). doi:10.1371/JOURNAL.PMED.1001484 101. Ashley-Koch A, Yang Q, Olney RS. Sickle hemoglobin (HbS) allele and sickle cell disease: a HuGE review. Am J Epidemiol. 2000;151(9):839-845. doi:10.1093/OXFORDJOURNALS.AJE.A010288 102. Noguchi CT, Schechter AN. Sickle hemoglobin polymerization in solution and in cells. Annu Rev Biophys Biophys Chem. 1985;14:239-263. doi:10.1146/ANNUREV.BB.14.060185.001323 103. Serjeant GR. The Natural History of Sickle Cell Disease. Cold Spring Harb Perspect Med. 2013;3(10). doi:10.1101/CSHPERSPECT.A011783 104. Lonergan GJ, Cline DB, Abbondanzo SL. Sickle cell anemia. Radiographics. 2001;21(4):971-994. doi:10.1148/RADIOGRAPHICS.21.4.G01JL23971 105. Karna B, Jha SK, Zaabi E Al. Hemoglobin C Disease. StatPearls. July 2021. https://www.ncbi.nlm.nih.gov/books/NBK559043/. Accessed January 18, 2022. 106. Nagel RL, Fabry ME, Steinberg MH. The paradox of hemoglobin SC disease. Blood Rev. 2003;17(3):167-178. doi:10.1016/S0268-960X(03)00003-1 107. Mobark NA, Alharbi M, Alhabeeb L, et al. Sickle cell disease: A distinction of two most frequent genotypes (HbSS and HbSC). PLoS One. 2020;15(1). doi:10.1371/JOURNAL.PONE.0228399 251 108. Houwing ME, Grohssteiner RL, Dremmen MHG, et al. Silent cerebral infarcts in patients with sickle cell disease: a systematic review and meta-analysis. BMC Med. 2020;18(1). doi:10.1186/S12916-020-01864-8 109. Figueiredo MS. The compound state: Hb S/beta-thalassemia. Rev Bras Hematol Hemoter. 2015;37(3):150-152. doi:10.1016/J.BJHH.2015.02.008 110. The irreversibly sickled cell: a perspective - PubMed. https://pubmed.ncbi.nlm.nih.gov/15040427/. Accessed July 5, 2022. 111. Quinn CT, Smith EP, Arbabi S, et al. Biochemical surrogate markers of hemolysis do not correlate with directly measured erythrocyte survival in sickle cell anemia. Am J Hematol. 2016;91(12):1195-1201. doi:10.1002/AJH.24562 112. El Nemer W, Godard A, El Hoss S. Ineffective erythropoiesis in sickle cell disease: new insights and future implications. Curr Opin Hematol. 2021;28(3):171-176. doi:10.1097/MOH.0000000000000642 113. Kato GJ, Steinberg MH, Gladwin MT. Intravascular hemolysis and the pathophysiology of sickle cell disease. J Clin Invest. 2017;127(3):750-760. doi:10.1172/JCI89741 114. Milner PF. Oxygen transport in sickle cell anemia. Arch Intern Med. 1974;133(4):565-572. http://www.ncbi.nlm.nih.gov/pubmed/4594394. Accessed April 1, 2019. 115. Cahill LS, Gazdzinski LM, Tsui AK, et al. Functional and anatomical evidence of cerebral tissue hypoxia in young sickle cell anemia mice. J Cereb Blood Flow Metab. 2017;37(3):994-1005. doi:10.1177/0271678X16649194 116. Kuwabara Y, Sasaki M, Hirakata H, et al. Cerebral blood flow and vasodilatory capacity in anemia secondary to chronic renal failure. Kidney Int. 2002;61(2):564-569. doi:10.1046/J.1523-1755.2002.00142.X 117. Bone marrow infarction in sickle cell anemia: correlation with hematologic profiles - PubMed. https://pubmed.ncbi.nlm.nih.gov/7139126/. Accessed July 5, 2022. 118. Miller AC, Gladwin MT. Pulmonary Complications of Sickle Cell Disease. Am J Respir Crit Care Med. 2012;185(11):1154. doi:10.1164/RCCM.201111-2082CI 119. Granfortuna J, Zamkoff K, Urrutia E. Acute renal infarction in sickle cell disease. Am J Hematol. 1986;23(1):59-64. doi:10.1002/AJH.2830230109 120. Al-Salem AH. Massive splenic infarction in children with sickle cell anemia and the role of splenectomy. Pediatr Surg Int. 2013;29(3):281-285. doi:10.1007/S00383-012-3223-2 121. Pavlakis SG, Bello J, Prohovnik I, et al. Brain infarction in sickle cell anemia: magnetic resonance imaging correlates. Ann Neurol. 1988;23(2):125-130. doi:10.1002/ANA.410230204 122. Hassell KL, Eckman JR, Lane PA. Acute multiorgan failure syndrome: a potentially catastrophic complication of severe sickle cell pain episodes. Am J Med. 1994;96(2):155- 162. doi:10.1016/0002-9343(94)90136-8 252 123. Chakravorty S, Williams TN. Sickle cell disease: a neglected chronic disease of increasing global health importance. Arch Dis Child. 2015;100(1):48. doi:10.1136/ARCHDISCHILD- 2013-303773 124. Cerebrovascular accidents in sickle cell disease: rates and risk factors - PubMed. https://pubmed.ncbi.nlm.nih.gov/9414296/. Accessed July 5, 2022. 125. Pegelow CH, Adams RJ, McKie V, et al. Risk of recurrent stroke in patients with sickle cell disease treated with erythrocyte transfusions. J Pediatr. 1995;126(6):896-899. doi:10.1016/S0022-3476(95)70204-0 126. Switzer JA, Hess DC, Nichols FT, Adams RJ. Pathophysiology and treatment of stroke in sickle-cell disease: present and future. Lancet Neurol. 2006;5(6):501-512. doi:10.1016/S1474-4422(06)70469-0 127. Debaun MR, Kirkham FJ. Central nervous system complications and management in sickle cell disease. Blood. 2016;127(7):829-838. doi:10.1182/BLOOD-2015-09-618579 128. Hulbert ML, McKinstry RC, Lacey JAL, et al. Silent cerebral infarcts occur despite regular blood transfusion therapy after first strokes in children with sickle cell disease. Blood. 2011;117(3):772. doi:10.1182/BLOOD-2010-01-261123 129. Kassim AA, Pruthi S, Day M, et al. Silent cerebral infarcts and cerebral aneurysms are prevalent in adults with sickle cell anemia. Blood. 2016;127(16):2038-2040. doi:10.1182/BLOOD-2016-01-694562 130. Bernaudin F, Verlhac S, Arnaud C, et al. Chronic and acute anemia and extracranial internal carotid stenosis are risk factors for silent cerebral infarcts in sickle cell anemia. Blood. 2015;125(10):1653-1661. doi:10.1182/BLOOD-2014-09-599852 131. Miller ST, Macklin EA, Pegelow CH, et al. Silent infarction as a risk factor for overt stroke in children with sickle cell anemia: a report from the Cooperative Study of Sickle Cell Disease. J Pediatr. 2001;139(3):385-390. doi:10.1067/MPD.2001.117580 132. Chai Y, Ji C, Coloigner J, et al. Tract-specific analysis and neurocognitive functioning in sickle cell patients without history of overt stroke. Brain Behav. 2021;11(3). doi:10.1002/BRB3.1978 133. Gendreau S, Scholer M, Cecchini J, et al. Cerebral fat embolism in sickle cell disease. Am J Hematol. 2020;95(2):E41-E45. doi:10.1002/AJH.25686 134. Adams RJ, Nichols FT, McKie V, McKie K, Milner P, Gammal TE. Cerebral infarction in sickle cell anemia: mechanism based on CT and MRI. Neurology. 1988;38(7):1012-1017. doi:10.1212/WNL.38.7.1012 135. Dowling MM, Quinn CT, Plumb P, et al. Acute silent cerebral ischemia and infarction during acute anemia in children with and without sickle cell disease. Blood. 2012;120(19):3891- 3897. doi:10.1182/BLOOD-2012-01-406314 136. Rowley CA, Ikeda AK, Seidel M, et al. Microvascular oxygen consumption during sickle cell pain crisis. Blood. 2014;123(20):3101. doi:10.1182/BLOOD-2013-11-533406 253 137. Binks AP, Cunningham VJ, Adams L, Banzett RB. Gray matter blood flow change is unevenly distributed during moderate isocapnic hypoxia in humans. J Appl Physiol. 2008;104(1):212-217. doi:10.1152/JAPPLPHYSIOL.00069.2007 138. King AA, Strouse JJ, Rodeghier MJ, et al. Parent education and biologic factors influence on cognition in sickle cell anemia. Am J Hematol. 2014;89(2):162. doi:10.1002/AJH.23604 139. Choudhury NA, DeBaun MR, Rodeghier M, King AA, Strouse JJ, McKinstry RC. Silent cerebral infarct definitions and full-scale IQ loss in children with sickle cell anemia. Neurology. 2018;90(3):e239. doi:10.1212/WNL.0000000000004832 140. King AA, Rodeghier MJ, Panepinto JA, et al. Silent cerebral infarction, income, and grade retention among students with sickle cell anemia. Am J Hematol. 2014;89(10):E188. doi:10.1002/AJH.23805 141. Steen R, Xiong X, Mulhern R, Langston J, Wang W. Subtle brain abnormalities in children with sickle cell disease: relationship to blood hematocrit. Ann Neurol. https://pubmed.ncbi.nlm.nih.gov/10072041/. Published 1999. Accessed January 18, 2022. 142. Cognitive impairment in children with hemoglobin SS sickle cell disease: relationship to MR imaging findings and hematocrit - PubMed. https://pubmed.ncbi.nlm.nih.gov/12637286/. Accessed July 5, 2022. 143. Comparison of high-performance liquid chromatography with a radiometric assay for determination of the effect of intra-articular administration of corticosteroid and saline solution on synovial fluid hyaluronate concentration in horses - PubMed. https://pubmed.ncbi.nlm.nih.gov/1789505/. Accessed July 5, 2022. 144. Turner J, Parsi M, Badireddy M. Anemia. https://www.ncbi.nlm.nih.gov/books/NBK499994/. Accessed January 18, 2022. 145. Borzage MT, Bush AM, Choi S, et al. Predictors of cerebral blood flow in patients with and without anemia. J Appl Physiol. 2016;120(8):976-981. doi:10.1152/JAPPLPHYSIOL.00994.2015 146. Chai Y, Bush AM, Coloigner J, et al. White matter has impaired resting oxygen delivery in sickle cell patients. Am J Hematol. 2019;94(4):467-474. doi:10.1002/AJH.25423 147. Blauwblomme T, Lemaitre H, Naggara O, et al. Cerebral Blood Flow Improvement after Indirect Revascularization for Pediatric Moyamoya Disease: A Statistical Analysis of Arterial Spin-Labeling MRI. AJNR Am J Neuroradiol. 2016;37(4):706-712. doi:10.3174/AJNR.A4592 148. Kuwabara Y, Ichiya Y, Sasaki M, et al. Cerebral hemodynamics and metabolism in moyamoya disease--a positron emission tomography study. Clin Neurol Neurosurg. 1997;99 Suppl 2(SUPPL. 2). doi:10.1016/S0303-8467(97)00061-9 149. Watchmaker JM, Juttukonda MR, Davis LT, et al. Hemodynamic mechanisms underlying elevated oxygen extraction fraction (OEF) in moyamoya and sickle cell anemia patients. J Cereb Blood Flow Metab. 2018;38(9):1618-1630. doi:10.1177/0271678X16682509 254 150. Vu C, Bush A, Choi S, et al. Reduced global cerebral oxygen metabolic rate in sickle cell disease and chronic anemias. Am J Hematol. 2021;96(8):901-913. doi:10.1002/AJH.26203 151. Bush A, Vu C, Choi S, et al. Calibration of T 2 oximetry MRI for subjects with sickle cell disease. Magn Reson Med. 2021;86(2):1019-1028. doi:10.1002/MRM.28757 152. Nakane H, Ibayashi S, Fujii K, et al. Cerebral blood flow and metabolism in patients with silent brain infarction: occult misery perfusion in the cerebral cortex. J Neurol Neurosurg Psychiatry. 1998;65(3):317-321. doi:10.1136/JNNP.65.3.317 153. Kety SS, Schmidt CF. The effects of altered arterial tensions of carbon dioxide and oxygen on cerebral blood flow and cerebral oxygen consumption of normal young men. J Clin Invest. 1948;27(4):484. doi:10.1172/JCI101995 154. Lee JJ, Powers WJ, Faulkner CB, Boyle PJ, Derdeyn CP. The Kety-Schmidt Technique for Quantitative Perfusion and Oxygen Metabolism Measurements in the MR Imaging Environment. AJNR Am J Neuroradiol. 2013;34(9):E100. doi:10.3174/AJNR.A3270 155. Verdecchia K, Diop M, Lee T-Y, St. Lawrence K. Quantifying the cerebral metabolic rate of oxygen by combining diffuse correlation spectroscopy and time-resolved near-infrared spectroscopy. J Biomed Opt. 2013;18(2):27007. doi:10.1117/1.JBO.18.2.027007 156. Baker WB, Balu R, He L, et al. Continuous non-invasive optical monitoring of cerebral blood flow and oxidative metabolism after acute brain injury. J Cereb Blood Flow Metab. 2019;39(8):1469-1485. doi:10.1177/0271678X19846657 157. Ibaraki M, Miura S, Shimosegawa E, et al. Quantification of cerebral blood flow and oxygen metabolism with 3-dimensional PET and 15O: validation by comparison with 2-dimensional PET. J Nucl Med. 2008;49(1):50-59. doi:10.2967/JNUMED.107.044008 158. Brown MM, Marshall J. Regulation of cerebral blood flow in response to changes in blood viscosity. Lancet (London, England). 1985;1(8429):604-609. doi:10.1016/S0140- 6736(85)92145-2 159. Prohovnik I, Hurlet-Jensen A, Adams R, De Vivo D, Pavlakis SG. Hemodynamic etiology of elevated flow velocity and stroke in sickle-cell disease. J Cereb Blood Flow Metab. 2009;29(4):803-810. doi:10.1038/JCBFM.2009.6 160. Prohovnik I, Pavlakis SG, Piomelli S, et al. Cerebral hyperemia, stroke, and transfusion in sickle cell disease. Neurology. 1989;39(3):344-348. doi:10.1212/WNL.39.3.344 161. Yonas H, Pindzola R, Johnson D. Xenon/computed tomography cerebral blood flow and its use in clinical management. Neurosurg Clin N Am. 1996;7(4):605-616. https://pubmed.ncbi.nlm.nih.gov/8905775/. Accessed January 18, 2022. 162. Masaoka H, Takasato Y, Hayakawa T, et al. Cerebral blood flow measurement of severely head-injured patients during mild hypothermia. Keio J Med. 2000. https://pubmed.ncbi.nlm.nih.gov/10750372/. Accessed January 18, 2022. 255 163. Fantini S, Sassaroli A, Tgavalekos KT, Kornbluth J. Cerebral blood flow and autoregulation: current measurement techniques and prospects for noninvasive optical methods. Neurophotonics. 2016;3(3):31411. doi:10.1117/1.NPH.3.3.031411 164. Lassen NA. Normal average value of cerebral blood flow in younger adults is 50 ml/100 g/min. J Cereb Blood Flow Metab. 1985;5(3):347-349. doi:10.1038/JCBFM.1985.48 165. Lucas SJE, Tzeng YC, Galvin SD, Thomas KN, Ogoh S, Ainslie PN. Influence of changes in blood pressure on cerebral perfusion and oxygenation. Hypertens (Dallas, Tex 1979). 2010;55(3):698-705. doi:10.1161/HYPERTENSIONAHA.109.146290 166. Cerebral autoregulation - PubMed. https://pubmed.ncbi.nlm.nih.gov/2201348/. Accessed July 5, 2022. 167. LASSEN NA. Cerebral blood flow and oxygen consumption in man. Physiol Rev. 1959;39(2):183-238. doi:10.1152/PHYSREV.1959.39.2.183 168. Numan T, Bain AR, Hoiland RL, Smirl JD, Lewis NC, Ainslie PN. Static autoregulation in humans: a review and reanalysis. Med Eng Phys. 2014;36(11):1487-1495. doi:10.1016/J.MEDENGPHY.2014.08.001 169. Drummond JC. The lower limit of autoregulation: time to revise our thinking? Anesthesiology. 1997;86(6):1431-1433. doi:10.1097/00000542-199706000-00034 170. Claassen JAHR, Zhang R, Fu Q, Witkowski S, Levine BD. Transcranial Doppler estimation of cerebral blood flow and cerebrovascular conductance during modified rebreathing. J Appl Physiol. 2007;102(3):870-877. doi:10.1152/JAPPLPHYSIOL.00906.2006 171. Moreton FC, Dani KA, Goutcher C, O’Hare K, Muir KW. Respiratory challenge MRI: Practical aspects. NeuroImage Clin. 2016;11:667-677. doi:10.1016/j.nicl.2016.05.003 172. Brian JE. Carbon dioxide and the cerebral circulation. Anesthesiology. 1998;88(5):1365- 1386. doi:10.1097/00000542-199805000-00029 173. Battisti-Charbonney A, Fisher J, Duffin J. The cerebrovascular response to carbon dioxide in humans. J Physiol. 2011;589(12):3039-3048. doi:10.1113/jphysiol.2011.206052 174. Wijnant SRA, Bos D, Brusselle G, et al. Comparison of cerebral blood flow in subjects with and without chronic obstructive pulmonary disease from the population-based Rotterdam Study. BMJ Open. 2021;11(12). doi:10.1136/BMJOPEN-2021-053671 175. Raichle ME, Plum F. Hyperventilation and cerebra blood flow. Stroke. 1972;3(5):566-575. doi:10.1161/01.STR.3.5.566 176. Stetter C, Weidner F, Lilla N, et al. Therapeutic hypercapnia for prevention of secondary ischemia after severe subarachnoid hemorrhage: physiological responses to continuous hypercapnia. Sci Rep. 2021;11(1). doi:10.1038/S41598-021-91007-7 177. González-Alonso J, Richardson RS, Saltin B. Exercising skeletal muscle blood flow in humans responds to reduction in arterial oxyhaemoglobin, but not to altered free oxygen. J Physiol. 2001;530(Pt 2):331-341. doi:10.1111/J.1469-7793.2001.0331L.X 256 178. Hoiland RL, Bain AR, Rieger MG, Bailey DM, Ainslie PN. Hypoxemia, oxygen content, and the regulation of cerebral blood flow. Am J Physiol Regul Integr Comp Physiol. 2016;310(5):R398-413. doi:10.1152/ajpregu.00270.2015 179. Borgström L, Jóhannsson H, Siesjö BK. The Relationship between Arterial P O2 and Cerebral Blood Flow in Hypoxic Hypoxia. Acta Physiol Scand. 1975;93(3):423-432. doi:10.1111/j.1748-1716.1975.tb05832.x 180. Borgström L, Jóhannsson H, Siesjö BK. The influence of acute normovolemic anemia on cerebral blood flow and oxygen consumption of anesthetized rats. Acta Physiol Scand. 1975;93(4):505-514. doi:10.1111/J.1748-1716.1975.TB05842.X 181. Lucas SJE, Burgess KR, Thomas KN, et al. Alterations in cerebral blood flow and cerebrovascular reactivity during 14 days at 5050 m. J Physiol. 2011;589(Pt 3):741-753. doi:10.1113/JPHYSIOL.2010.192534 182. Bush AM, Borzage MT, Choi S, et al. Determinants of resting cerebral blood flow in sickle cell disease. Am J Hematol. 2016;91(9):912-917. doi:10.1002/AJH.24441 183. Wymer DT, Patel KP, Burke WF, Bhatia VK. Phase-Contrast MRI: Physics, Techniques, and Clinical Applications. Radiographics. 2020;40(1):122-140. doi:10.1148/RG.2020190039 184. Vestergaard MB, Lindberg U, Aachmann-Andersen NJ, et al. Comparison of global cerebral blood flow measured by phase-contrast mapping MRI with 15 O-H 2 O positron emission tomography. J Magn Reson Imaging. 2017;45(3):692-699. doi:10.1002/JMRI.25442 185. Zaharchuk G. Arterial Spin Labeled Perfusion Imaging in Acute Ischemic Stroke. Stroke. 2014;45(4):1202. doi:10.1161/STROKEAHA.113.003612 186. Lambert EA, Holmes S. Differentiating Radiation-Induced Necrosis from Tumor Progression After Stereotactic Radiosurgery for Brain Metastases, Using Evaluation of Blood Flow with Arterial Spin Labeling (ASL): The Importance of Setting a Baseline. Acta Neurochir Suppl. 2021;128:113-119. doi:10.1007/978-3-030-69217-9_12 187. Noguchi T, Yoshiura T, Hiwatashi A, et al. Perfusion imaging of brain tumors using arterial spin-labeling: correlation with histopathologic vascular density. AJNR Am J Neuroradiol. 2008;29(4):688-693. doi:10.3174/AJNR.A0903 188. Chao LL, Buckley ST, Kornak J, et al. ASL perfusion MRI predicts cognitive decline and conversion from MCI to dementia. Alzheimer Dis Assoc Disord. 2010;24(1):19-27. doi:10.1097/WAD.0B013E3181B4F736 189. Le Heron CJ, Wright SL, Melzer TR, et al. Comparing cerebral perfusion in Alzheimer’s disease and Parkinson’s disease dementia: an ASL-MRI study. J Cereb Blood Flow Metab. 2014;34(6):964-970. doi:10.1038/JCBFM.2014.40 190. Grade M, Hernandez Tamames JA, Pizzini FB, Achten E, Golay X, Smits M. A neuroradiologist’s guide to arterial spin labeling MRI in clinical practice. Neuroradiology. 2015;57(12):1181-1202. doi:10.1007/S00234-015-1571-Z 257 191. S P, J R, M C. Arterial spin labeling in neuroimaging. World J Radiol. 2010;2(10):384. doi:10.4329/WJR.V2.I10.384 192. Wong EC. Quantifying CBF with pulsed ASL: technical and pulse sequence factors. J Magn Reson Imaging. 2005;22(6):727-731. doi:10.1002/JMRI.20459 193. Wu WC, Fernández-Seara M, Detre JA, Wehrli FW, Wang J. A theoretical and experimental investigation of the tagging efficiency of pseudocontinuous arterial spin labeling. Magn Reson Med. 2007;58(5):1020-1027. doi:10.1002/MRM.21403 194. Alsop DC, Detre JA, Golay X, et al. Recommended implementation of arterial spin-labeled perfusion MRI for clinical applications: A consensus of the ISMRM perfusion study group and the European consortium for ASL in dementia. Magn Reson Med. 2015;73(1):102-116. doi:10.1002/mrm.25197 195. Wong EC, Cronin M, Wu WC, Inglis B, Frank LR, Liu TT. Velocity-selective arterial spin labeling. Magn Reson Med. 2006;55(6):1334-1341. doi:10.1002/MRM.20906 196. Teeuwisse WM, Schmid S, Ghariq E, Veer IM, Van Osch MJP. Time-encoded pseudocontinuous arterial spin labeling: basic properties and timing strategies for human applications. Magn Reson Med. 2014;72(6):1712-1722. doi:10.1002/MRM.25083 197. Lu H, Clingman C, Golay X, Van Zijl PCM. Determining the longitudinal relaxation time (T1) of blood at 3.0 Tesla. Magn Reson Med. 2004;52(3):679-682. doi:10.1002/MRM.20178 198. Vaclavu L, Van Der Land V, Heijtel DFR, et al. In Vivo T1 of Blood Measurements in Children with Sickle Cell Disease Improve Cerebral Blood Flow Quantification from Arterial Spin-Labeling MRI. AJNR Am J Neuroradiol. 2016;37(9):1727-1732. doi:10.3174/AJNR.A4793 199. Bush A, Chai Y, Choi SY, et al. Pseudo Continuous Arterial Spin Labeling Quantification in Anemic Subjects with Hyperemic Cerebral Blood Flow. Magn Reson Imaging. 2018;47:137. doi:10.1016/J.MRI.2017.12.011 200. Parkes LM. Quantification of cerebral perfusion using arterial spin labeling: two- compartment models. J Magn Reson Imaging. 2005;22(6):732-736. doi:10.1002/JMRI.20456 201. Golay X, Hendrikse J, Lim TCC. Perfusion imaging using arterial spin labeling. Top Magn Reson Imaging. 2004;15(1):10-27. doi:10.1097/00002142-200402000-00003 202. Kanal E. Gadolinium based contrast agents (GBCA): Safety overview after 3 decades of clinical experience. Magn Reson Imaging. 2016;34(10):1341-1345. doi:10.1016/J.MRI.2016.08.017 203. Vu C, Chai Y, Coloigner J, et al. Quantitative perfusion mapping with induced transient hypoxia using BOLD MRI. Magn Reson Med. 2021;85(1):168-181. doi:10.1002/mrm.28422 204. Poublanc J, Sobczyk O, Shafi R, et al. Perfusion MRI using endogenous deoxyhemoglobin as a contrast agent: Preliminary data. Magn Reson Med. 2021;86(6):3012-3021. doi:10.1002/MRM.28974 258 205. MacDonald ME, Berman AJL, Mazerolle EL, Williams RJ, Pike GB. Modeling hyperoxia- induced BOLD signal dynamics to estimate cerebral blood flow, volume and mean transit time. Neuroimage. 2018;178:461-474. doi:10.1016/j.neuroimage.2018.05.066 206. Jahng G-H, Li K-L, Ostergaard L, Calamante F. Perfusion magnetic resonance imaging: a comprehensive update on principles and techniques. Korean J Radiol. 2014;15(5):554- 577. doi:10.3348/kjr.2014.15.5.554 207. MEIER P, ZIERLER KL. On the theory of the indicator-dilution method for measurement of blood flow and volume. J Appl Physiol. 1954;6(12):731-744. doi:10.1152/JAPPL.1954.6.12.731 208. Leu K, Boxerman JL, Cloughesy TF, et al. Improved Leakage Correction for Single-Echo Dynamic Susceptibility Contrast Perfusion MRI Estimates of Relative Cerebral Blood Volume in High-Grade Gliomas by Accounting for Bidirectional Contrast Agent Exchange. AJNR Am J Neuroradiol. 2016;37(8):1440. doi:10.3174/AJNR.A4759 209. Boxerman JL, Prah DE, Paulson ES, Machan JT, Bedekar D, Schmainda KM. The Role of Preload and Leakage Correction in Gadolinium-Based Cerebral Blood Volume Estimation Determined by Comparison with MION as a Criterion Standard. AJNR Am J Neuroradiol. 2012;33(6):1081. doi:10.3174/AJNR.A2934 210. Weisskoff R, Zuo CS, Boxerman JL, Rosen BR. Microscopic susceptibility variation and transverse relaxation: theory and experiment. Magn Reson Med. 1994;31(6):601-610. doi:10.1002/MRM.1910310605 211. Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. MR contrast due to intravascular magnetic susceptibility perturbations. Magn Reson Med. 1995;34(4):555-566. doi:10.1002/MRM.1910340412 212. Stokes AM, Bergamino M, Alhilali L, et al. Evaluation of single bolus, dual-echo dynamic susceptibility contrast MRI protocols in brain tumor patients. J Cereb Blood Flow Metab. 2021;41(12):3378-3390. doi:10.1177/0271678X211039597 213. Stokes AM, Semmineh NB, Nespodzany A, Bell LC, Quarles CC. Systematic Assessment of Multi-Echo Dynamic Susceptibility Contrast (DSC) MRI using a Digital Reference Object (DRO). Magn Reson Med. 2020;83(1):109. doi:10.1002/MRM.27914 214. Boxerman JL, Quarles CC, Hu LS, et al. Consensus recommendations for a dynamic susceptibility contrast MRI protocol for use in high-grade gliomas. Neuro Oncol. 2020;22(9):1262. doi:10.1093/NEUONC/NOAA141 215. Calamante F, Mørup M, Hansen LK. Defining a local arterial input function for perfusion MRI using independent component analysis. Magn Reson Med. 2004;52(4):789-797. doi:10.1002/MRM.20227 216. Ziayee F, Müller-Lutz A, Gross J, et al. Influence of arterial input function (AIF) on quantitative prostate dynamic contrast-enhanced (DCE) MRI and zonal prostate anatomy. Magn Reson Imaging. 2018;53:28-33. doi:10.1016/J.MRI.2018.06.004 259 217. Chen JJ, Smith MR, Frayne R. The impact of partial-volume effects in dynamic susceptibility contrast magnetic resonance perfusion imaging. J Magn Reson Imaging. 2005;22(3):390-399. doi:10.1002/jmri.20393 218. Knutsson L, van Westen D, Petersen ET, et al. Absolute quantification of cerebral blood flow: correlation between dynamic susceptibility contrast MRI and model-free arterial spin labeling. Magn Reson Imaging. 2010;28(1):1-7. doi:10.1016/J.MRI.2009.06.006 219. Knutsson L, Börjesson S, Larsson EM, et al. Absolute quantification of cerebral blood flow in normal volunteers: correlation between Xe-133 SPECT and dynamic susceptibility contrast MRI. J Magn Reson Imaging. 2007;26(4):913-920. doi:10.1002/JMRI.21093 220. Tudorica A, Fang Li H, Hospod F, et al. Cerebral blood volume measurements by rapid contrast infusion andT2*-weighted echo planar MRI. Magn Reson Med. 2002;47(6):1145- 1157. doi:10.1002/mrm.10167 221. Østergaard L. Principles of cerebral perfusion imaging by bolus tracking. J Magn Reson Imaging. 2005;22(6):710-717. doi:10.1002/JMRI.20460 222. Rempp KA, Brix G, Wenz F, Becker CR, Gückel F, Lorenz WJ. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology. 1994;193(3):637-641. doi:10.1148/RADIOLOGY.193.3.7972800 223. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted G, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med. 1996;36(5):715-725. doi:10.1002/MRM.1910360510 224. Liang J, Liu D, Gao P, et al. Diagnostic Values of DCE-MRI and DSC-MRI for Differentiation Between High-grade and Low-grade Gliomas: A Comprehensive Meta-analysis. Acad Radiol. 2018;25(3):338-348. doi:10.1016/J.ACRA.2017.10.001 225. Falk A, Fahlström M, Rostrup E, et al. Discrimination between glioma grades II and III in suspected low-grade gliomas using dynamic contrast-enhanced and dynamic susceptibility contrast perfusion MR imaging: a histogram analysis approach. Neuroradiology. 2014;56(12):1031-1038. doi:10.1007/S00234-014-1426-Z 226. Young RJ, Gupta A, Shah AD, et al. MRI perfusion in determining pseudoprogression in patients with glioblastoma. Clin Imaging. 2013;37(1):41. doi:10.1016/J.CLINIMAG.2012.02.016 227. Boxerman JL, Ellingson BM, Jeyapalan S, et al. Longitudinal DSC-MRI for Distinguishing Tumor Recurrence From Pseudoprogression in Patients With a High-grade Glioma. Am J Clin Oncol. 2017;40(3):228-234. doi:10.1097/COC.0000000000000156 228. Eskildsen SF, Gyldensted L, Nagenthiraja K, et al. Increased cortical capillary transit time heterogeneity in Alzheimer’s disease: a DSC-MRI perfusion study. Neurobiol Aging. 2017;50:107-118. doi:10.1016/J.NEUROBIOLAGING.2016.11.004 260 229. Schmidt MA, Engelhorn T, Lang S, et al. DSC Brain Perfusion Using Advanced Deconvolution Models in the Diagnostic Work-Up of Dementia and Mild Cognitive Impairment: A Semiquantitative Comparison with HMPAO-SPECT-Brain Perfusion. J Clin Med. 2020;9(6):1-14. doi:10.3390/JCM9061800 230. Copen WA, Schaefer PW, Wu O. MR Perfusion Imaging in Acute Ischemic Stroke. Neuroimaging Clin N Am. 2011;21(2):259. doi:10.1016/J.NIC.2011.02.007 231. Zaro-Weber O, Fleischer H, Reiblich L, Schuster A, Moeller-Hartmann W, Heiss WD. Penumbra detection in acute stroke with perfusion magnetic resonance imaging: Validation with 15O-positron emission tomography. Ann Neurol. 2019;85(6):875. doi:10.1002/ANA.25479 232. Albers GW, Thijs VN, Wechsler L, et al. Magnetic resonance imaging profiles predict clinical response to early reperfusion: the diffusion and perfusion imaging evaluation for understanding stroke evolution (DEFUSE) study. Ann Neurol. 2006;60(5):508-517. doi:10.1002/ANA.20976 233. Davis SM, Donnan GA, Parsons MW, et al. Effects of alteplase beyond 3 h after stroke in the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET): a placebo-controlled randomised trial. Lancet Neurol. 2008;7(4):299-309. doi:10.1016/S1474-4422(08)70044-9 234. Lythgoe DJ, Østergaard L, Williams SCR, et al. Quantitative perfusion imaging in carotid artery stenosis using dynamic susceptibility contrast-enhanced magnetic resonance imaging. Magn Reson Imaging. 2000;18(1):1-11. doi:10.1016/S0730-725X(99)00112-5 235. Wu O, Østergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. Tracer arrival timing-insensitive technique for estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magn Reson Med. 2003;50(1):164-174. doi:10.1002/MRM.10522 236. Rose SE, Janke AL, Griffin M, et al. Improving the prediction of final infarct size in acute stroke with bolus delay-corrected perfusion MRI measures. J Magn Reson Imaging. 2004;20(6):941-947. doi:10.1002/JMRI.20216 237. Delay and dispersion effects in dynamic susceptibility contrast MRI: simulations using singular value decomposition - PubMed. https://pubmed.ncbi.nlm.nih.gov/10975900/. Accessed July 6, 2022. 238. Arzanforoosh F, Croal PL, van Garderen KA, Smits M, Chappell MA, Warnert EAH. Effect of Applying Leakage Correction on rCBV Measurement Derived From DSC-MRI in Enhancing and Nonenhancing Glioma. Front Oncol. 2021;11. doi:10.3389/FONC.2021.648528 239. Stokes AM, Semmineh N, Quarles CC. Validation of a T1 and T2* leakage correction method based on multiecho dynamic susceptibility contrast MRI using MION as a reference standard. Magn Reson Med. 2016;76(2):613-625. doi:10.1002/mrm.25906 261 240. Blumfield E, Swenson DW, Iyer RS, Stanescu AL. Gadolinium-based contrast agents - review of recent literature on magnetic resonance imaging signal intensity changes and tissue deposits, with emphasis on pediatric patients. Pediatr Radiol. 2019;49(4):448-457. doi:10.1007/S00247-018-4304-8 241. Blumfield E, Moore MM, Drake MK, et al. Survey of gadolinium-based contrast agent utilization among the members of the Society for Pediatric Radiology: a Quality and Safety Committee report. Pediatr Radiol. 2017;47(6):665-673. doi:10.1007/S00247-017-3807-Z 242. Schlaudecker J, Bernheisel C. Gadolinium-associated nephrogenic systemic fibrosis. Am Fam Physician. 2009;80(7):711-714. https://pubmed.ncbi.nlm.nih.gov/19817341/. Accessed January 18, 2022. 243. Leiner T, Kucharczyk W. NSF prevention in clinical practice: summary of recommendations and guidelines in the United States, Canada, and Europe. J Magn Reson Imaging. 2009;30(6):1357-1363. doi:10.1002/JMRI.22021 244. Guo BJ, Yang ZL, Zhang LJ. Gadolinium Deposition in Brain: Current Scientific Evidence and Future Perspectives. Front Mol Neurosci. 2018;11. doi:10.3389/FNMOL.2018.00335 245. Kanda T, Ishii K, Kawaguchi H, Kitajima K, Takenaka D. High signal intensity in the dentate nucleus and globus pallidus on unenhanced T1-weighted MR images: relationship with increasing cumulative dose of a gadolinium-based contrast material. Radiology. 2014;270(3):834-841. doi:10.1148/RADIOL.13131669 246. White GW, Gibby WA, Tweedle MF. Comparison of Gd(DTPA-BMA) (Omniscan) versus Gd(HP-DO3A) (ProHance) relative to gadolinium retention in human bone tissue by inductively coupled plasma mass spectroscopy. Invest Radiol. 2006;41(3):272-278. doi:10.1097/01.RLI.0000186569.32408.95 247. Gibby WA, Gibby KA, Gibby WA. Comparison of Gd DTPA-BMA (Omniscan) versus Gd HP-DO3A (ProHance) retention in human bone tissue by inductively coupled plasma atomic emission spectroscopy. Invest Radiol. 2004;39(3):138-142. doi:10.1097/01.RLI.0000112789.57341.01 248. Christensen KN, Lee CU, Hanley MM, Leung N, Moyer TP, Pittelkow MR. Quantification of gadolinium in fresh skin and serum samples from patients with nephrogenic systemic fibrosis. J Am Acad Dermatol. 2011;64(1):91-96. doi:10.1016/J.JAAD.2009.12.044 249. Maximova N, Gregori M, Zennaro F, Sonzogni A, Simeone R, Zanon D. Hepatic Gadolinium Deposition and Reversibility after Contrast Agent-enhanced MR Imaging of Pediatric Hematopoietic Stem Cell Transplant Recipients. Radiology. 2016;281(2):418- 426. doi:10.1148/RADIOL.2016152846 250. Matsuoka S, Uchiyama K, Shima H, et al. Effect of the rate of gadolinium injection on magnetic resonance pulmonary perfusion imaging. J Magn Reson Imaging. 2002;15(1):108-113. doi:10.1002/JMRI.10038 251. Van Osch MJP, Vonken EJPA, Wu O, Viergever MA, Van der Grond J, Bakker CJG. Model of the human vasculature for studying the influence of contrast injection speed on cerebral perfusion MRI. Magn Reson Med. 2003;50(3):614-622. doi:10.1002/MRM.10567 262 252. Alger JR, Schaewe TJ, Lai TC, et al. Contrast agent dose effects in cerebral dynamic susceptibility contrast magnetic resonance perfusion imaging. J Magn Reson Imaging. 2009;29(1):52-64. doi:10.1002/JMRI.21613 253. Bell LC, Hu LS, Stokes AM, McGee SC, Baxter LC, Quarles CC. Characterizing the Influence of Preload Dosing on Percent Signal Recovery (PSR) and Cerebral Blood Volume (CBV) Measurements in a Patient Population With High-Grade Glioma Using Dynamic Susceptibility Contrast MRI. Tomography. 2017;3(2):89. doi:10.18383/J.TOM.2017.00004 254. Semmineh NB, Xu J, Boxerman JL, et al. An efficient computational approach to characterize DSC-MRI signals arising from three-dimensional heterogeneous tissue structures. PLoS One. 2014;9(1). doi:10.1371/JOURNAL.PONE.0084764 255. Kjølby BF, Østergaard L, Kiselev VG. Theoretical model of intravascular paramagnetic tracers effect on tissue relaxation. Magn Reson Med. 2006;56(1):187-197. doi:10.1002/MRM.20920 256. van Osch MJP, Vonken EPA, Viergever MA, van der Grond J, Bakker CJG. Measuring the arterial input function with gradient echo sequences. Magn Reson Med. 2003;49(6):1067- 1076. doi:10.1002/mrm.10461 257. Calamante F, Connelly A, Van Osch MJP. Nonlinear DeltaR*2 effects in perfusion quantification using bolus-tracking MRI. Magn Reson Med. 2009;61(2):486-492. doi:10.1002/MRM.21839 258. Kjølby BF, Mikkelsen IK, Pedersen M, Østergaard L, Kiselev VG. Analysis of partial volume effects on arterial input functions using gradient echo: a simulation study. Magn Reson Med. 2009;61(6):1300-1309. doi:10.1002/MRM.21849 259. Willats L, Calamante F. The 39 steps: evading error and deciphering the secrets for accurate dynamic susceptibility contrast MRI. NMR Biomed. 2013;26(8):913-931. doi:10.1002/NBM.2833 260. Newman GC, Hospod FE, Patlak CS, et al. Experimental estimates of the constants relating signal change to contrast concentration for cerebral blood volume by T2* MRI. J Cereb Blood Flow Metab. 2006;26(6):760-770. doi:10.1038/SJ.JCBFM.9600242 261. Coloigner J, Vu C, Borzage M, et al. Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia Using BOLD MRI and Near-Infrared Spectroscopy. J Magn Reson Imaging. 2020;52(5):1400-1412. doi:10.1002/JMRI.27210 262. Rostrup E, Larsson HB, Toft PB, Garde K, Henriksen O. Signal changes in gradient echo images of human brain induced by hypo- and hyperoxia. NMR Biomed. 1995;8(1):41-47. http://www.ncbi.nlm.nih.gov/pubmed/7547184. Accessed July 29, 2019. 263. Xu F, Liu P, Pascual JM, Xiao G, Lu H. Effect of hypoxia and hyperoxia on cerebral blood flow, blood oxygenation, and oxidative metabolism. J Cereb Blood Flow Metab. 2012;32(10):1909-1918. doi:10.1038/JCBFM.2012.93 263 264. Sleight E, Stringer MS, Marshall I, Wardlaw JM, Thrippleton MJ. Cerebrovascular Reactivity Measurement Using Magnetic Resonance Imaging: A Systematic Review. Front Physiol. 2021;12:643468. doi:10.3389/FPHYS.2021.643468/FULL 265. Vagal AS, Leach JL, Fernandez-Ulloa M, Zuccarello M. The acetazolamide challenge: techniques and applications in the evaluation of chronic cerebral ischemia. AJNR Am J Neuroradiol. 2009;30(5):876-884. doi:10.3174/AJNR.A1538 266. Liu P, De Vis JB, Lu H. Cerebrovascular reactivity (CVR) MRI with CO2 challenge: A technical review. Neuroimage. 2019;187:104-115. doi:10.1016/J.NEUROIMAGE.2018.03.047 267. Markus HS, Harrison MJG. Estimation of cerebrovascular reactivity using transcranial Doppler, including the use of breath-holding as the vasodilatory stimulus. Stroke. 1992;23(5):668-673. doi:10.1161/01.STR.23.5.668 268. Bruce CD, Steinback CD, Chauhan U V., et al. Quantifying cerebrovascular reactivity in anterior and posterior cerebral circulations during voluntary breath holding. Exp Physiol. 2016;101(12):1517-1527. doi:10.1113/EP085764 269. Urback AL, MacIntosh BJ, Goldstein BI. Cerebrovascular reactivity measured by functional magnetic resonance imaging during breath-hold challenge: A systematic review. Neurosci Biobehav Rev. 2017;79:27-47. doi:10.1016/J.NEUBIOREV.2017.05.003 270. Fierstra J, Sobczyk O, Battisti-Charbonney A, et al. Measuring cerebrovascular reactivity: what stimulus to use? J Physiol. 2013;591(23):5809-5821. doi:10.1113/jphysiol.2013.259150 271. Donahue MJ, Strother MK, Lindsey KP, Hocke LM, Tong Y, Frederick BDB. Time delay processing of hypercapnic fMRI allows quantitative parameterization of cerebrovascular reactivity and blood flow delays. J Cereb Blood Flow Metab. 2016;36(10):1767-1779. doi:10.1177/0271678X15608643 272. Duffin J, Sobczyk O, Crawley AP, Poublanc J, Mikulis DJ, Fisher JA. The dynamics of cerebrovascular reactivity shown with transfer function analysis. Neuroimage. 2015;114:207-216. doi:10.1016/j.neuroimage.2015.04.029 273. Blockley NP, Driver ID, Francis ST, Fisher JA, Gowland PA. An improved method for acquiring cerebrovascular reactivity maps. Magn Reson Med. 2011;65(5):1278-1286. doi:10.1002/mrm.22719 274. Donahue MJ, Ayad M, Moore R, et al. Relationships between hypercarbic reactivity, cerebral blood flow, and arterial circulation times in patients with moyamoya disease. J Magn Reson Imaging. 2013;38(5):1129-1139. doi:10.1002/JMRI.24070 275. Mandell DM, Han JS, Poublanc J, et al. Mapping cerebrovascular reactivity using blood oxygen level-dependent MRI in Patients with arterial steno-occlusive disease: comparison with arterial spin labeling MRI. Stroke. 2008;39(7):2021-2028. doi:10.1161/STROKEAHA.107.506709 264 276. Fierstra J, van Niftrik B, Piccirelli M, et al. Altered intraoperative cerebrovascular reactivity in brain areas of high-grade glioma recurrence. Magn Reson Imaging. 2016;34(6):803-808. doi:10.1016/J.MRI.2016.02.003 277. JJ P, D Z. Clinical utility of cerebrovascular reactivity mapping in patients with low grade gliomas. World J Clin Oncol. 2011;2(12):397. doi:10.5306/WJCO.V2.I12.397 278. Glodzik L, Randall C, Rusinek H, De Leon MJ. Cerebrovascular reactivity to carbon dioxide in Alzheimer’s disease. J Alzheimers Dis. 2013;35(3):427-440. doi:10.3233/JAD-122011 279. Lavinio A, Ene-Iordache B, Nodari I, et al. Cerebrovascular reactivity and autonomic drive following traumatic brain injury. Acta Neurochir Suppl. 2008;102(102):3-7. doi:10.1007/978-3-211-85578-2_1 280. Geranmayeh F, Wise RJS, Leech R, Murphy K. Measuring vascular reactivity with breath- holds after stroke: a method to aid interpretation of group-level BOLD signal changes in longitudinal fMRI studies. Hum Brain Mapp. 2015;36(5):1755-1771. doi:10.1002/HBM.22735 281. Gupta A, Chazen JL, Hartman M, et al. Cerebrovascular reserve and stroke risk in patients with carotid stenosis or occlusion: a systematic review and meta-analysis. Stroke. 2012;43(11):2884-2891. doi:10.1161/STROKEAHA.112.663716 282. Dlamini N, Slim M, Kirkham F, et al. Predicting Ischemic Risk Using Blood Oxygen Level- Dependent MRI in Children with Moyamoya. AJNR Am J Neuroradiol. 2020;41(1):160-166. doi:10.3174/AJNR.A6324 283. Ford AL, Ragan DK, Fellah S, et al. Silent infarcts in sickle cell disease occur in the border zone region and are associated with low cerebral blood flow. Blood. 2018;132(16):1714- 1723. doi:10.1182/blood-2018-04-841247 284. Derdeyn CP, Videen TO, Yundt KD, et al. Variability of cerebral blood volume and oxygen extraction: stages of cerebral haemodynamic impairment revisited. Brain. 2002;125(3):595-607. doi:10.1093/brain/awf047 285. Bush AM, Coates TD, Wood JC. Diminished cerebral oxygen extraction and metabolic rate in sickle cell disease using T2 relaxation under spin tagging MRI. Magn Reson Med. 2018;80(1):294-303. doi:10.1002/mrm.27015 286. Ge Y, Zhang Z, Lu H, et al. Characterizing brain oxygen metabolism in patients with multiple sclerosis with T2-relaxation-under-spin-tagging MRI. J Cereb Blood Flow Metab. 2012;32(3):403-412. doi:10.1038/JCBFM.2011.191 287. Jain V, Buckley EM, Licht DJ, et al. Cerebral oxygen metabolism in neonates with congenital heart disease quantified by MRI and optics. J Cereb Blood Flow Metab. 2014;34(3):380-388. doi:10.1038/JCBFM.2013.214 288. Yamauchi H, Fukuyama H, Nagahama Y, et al. Significance of increased oxygen extraction fraction in five-year prognosis of major cerebral arterial occlusive diseases. J Nucl Med. 1999;40(12):1992-1998. https://pubmed.ncbi.nlm.nih.gov/10616876/. Accessed January 26, 2022. 265 289. Gupta A, Baradaran H, Schweitzer AD, et al. Oxygen Extraction Fraction and Stroke Risk in Patients with Carotid Stenosis or Occlusion: A Systematic Review and Meta-Analysis. AJNR Am J Neuroradiol. 2014;35(2):250. doi:10.3174/AJNR.A3668 290. Leigh R, Knutsson L, Zhou J, van Zijl PCM. Imaging the physiological evolution of the ischemic penumbra in acute ischemic stroke. J Cereb Blood Flow Metab. 2018;38(9):1500. doi:10.1177/0271678X17700913 291. An H, Ford AL, Chen Y, et al. Defining the Ischemic Penumbra using Magnetic Resonance Oxygen Metabolic Index. Stroke. 2015;46(4):982. doi:10.1161/STROKEAHA.114.008154 292. Biondetti E, Rojas-Villabona A, Sokolska M, et al. Investigating the oxygenation of brain arteriovenous malformations using quantitative susceptibility mapping. Neuroimage. 2019;199:440-453. doi:10.1016/J.NEUROIMAGE.2019.05.014 293. Edelman G, Hoffman WE. Cerebral venous and tissue gases and arteriovenous shunting in the dog. Anesth Analg. 1999;89(3):679-683. doi:10.1097/00000539-199909000-00028 294. Shanmukhappa S, Lokeshwaran S. Venous Oxygen Saturation. StatPearls (Treasure Island); 2021. https://pubmed.ncbi.nlm.nih.gov/33232065/. Accessed January 26, 2022. 295. Miao X, Nayak KS, Wood JC. In vivo validation of T2- and susceptibility-based S v O 2 measurements with jugular vein catheterization under hypoxia and hypercapnia. Magn Reson Med. 2019;82(6):2188-2198. doi:10.1002/mrm.27871 296. De Vis JB, Lu H, Ravi H, Hendrikse J, Liu P. Spatial distribution of flow and oxygenation in the cerebral venous drainage system. J Magn Reson Imaging. 2018;47(4):1091. doi:10.1002/JMRI.25833 297. Jain V, Magland J, Langham M, Wehrli FW. High Temporal Resolution In Vivo Blood Oximetry Via Projection-Based T2 Measurement. Magn Reson Med. 2013;70(3):785. doi:10.1002/MRM.24519 298. Sakai F, Nakazawa K, Tazaki Y, et al. Regional cerebral blood volume and hematocrit measured in normal human volunteers by single-photon emission computed tomography. J Cereb Blood Flow Metab. 1985;5(2):207-213. doi:10.1038/JCBFM.1985.27 299. An H, Lin W. Impact of intravascular signal on quantitative measures of cerebral oxygen extraction and blood volume under normo- and hypercapnic conditions using an asymmetric spin echo approach. Magn Reson Med. 2003;50(4):708-716. doi:10.1002/MRM.10576 300. Stone AJ, Holland NC, Berman AJL, Blockley NP. Simulations of the effect of diffusion on asymmetric spin echo based quantitative BOLD: An investigation of the origin of deoxygenated blood volume overestimation. Neuroimage. 2019;201. doi:10.1016/J.NEUROIMAGE.2019.116035 301. An H, Lin W. Cerebral oxygen extraction fraction and cerebral venous blood volume measurements using MRI: effects of magnetic field variation. Magn Reson Med. 2002;47(5):958-966. doi:10.1002/MRM.10148 266 302. Yablonskiy DA, Haacke EM. Theory of NMR signal behavior in magnetically inhomogeneous tissues: the static dephasing regime. Magn Reson Med. 1994;32(6):749- 763. doi:10.1002/MRM.1910320610 303. Wang C, Zhang R, Zhang X, et al. Simultaneous dynamic R 2 R2’, and R2* measurement using periodic π pulse shifting multiecho asymmetric spin echo sequence moving estimation strategy: A feasibility study for lower extremity muscle. Magn Reson Med. 2017;77(2):766-773. doi:10.1002/MRM.26126 304. Elder CP, Cook RN, Chance MA, Copenhaver EA, Damon BM. Image Based Calculation of Perfusion and Oxyhemoglobin Saturation in Skeletal Muscle during Submaximal Isometric Contractions. Magn Reson Med. 2010;64(3):852. doi:10.1002/MRM.22475 305. Zhang X, Zhang Y, Yang X, et al. Feasibility of noninvasive quantitative measurements of intrarenal R(2) ’ in humans using an asymmetric spin echo echo planar imaging sequence. NMR Biomed. 2013;26(1):91-97. doi:10.1002/NBM.2823 306. Guilliams KP, Fields ME, Ragan DK, et al. Red cell exchange transfusions lower cerebral blood flow and oxygen extraction fraction in pediatric sickle cell anemia. Blood. 2018;131(9):1012-1021. doi:10.1182/blood-2017-06-789842 307. An H, Liu Q, Chen Y, Lin W. Evaluation of MR-derived cerebral oxygen metabolic index in experimental hyperoxic hypercapnia, hypoxia, and ischemia. Stroke. 2009;40(6):2165- 2172. doi:10.1161/STROKEAHA.108.540864 308. Fields ME, Mirro AE, Guilliams KP, et al. Functional Connectivity Decreases with Metabolic Stress in Sickle Cell Disease. Ann Neurol. 2020;88(5):995-1008. doi:10.1002/ANA.25891 309. Fields ME, Guilliams KP, Ragan D, et al. Hydroxyurea reduces cerebral metabolic stress in patients with sickle cell anemia. Blood. 2019;133(22):2436-2444. doi:10.1182/BLOOD- 2018-09-876318 310. Wang Y, Fellah S, Fields ME, et al. Cerebral Oxygen Metabolic Stress, Microstructural Injury, and Infarction in Adults With Sickle Cell Disease. Neurology. 2021;97(9):e902-e912. doi:10.1212/WNL.0000000000012404 311. Fields ME, Guilliams KP, Ragan DK, et al. Regional oxygen extraction predicts border zone vulnerability to stroke in sickle cell disease. Neurology. 2018;90(13):e1134-e1144. doi:10.1212/WNL.0000000000005194 312. He X, Yablonskiy DA. Quantitative BOLD: Mapping of Human Cerebral Deoxygenated Blood Volume and Oxygen Extraction Fraction: Default State. Magn Reson Med. 2007;57(1):115. doi:10.1002/MRM.21108 313. Simonson SG, Piantadosi CA. Near-infrared spectroscopy. Clinical applications. Crit Care Clin. 1996;12(4):1019-1029. doi:10.1016/S0749-0704(05)70290-6 314. Ghosh A, Elwell C, Smith M. Review article: cerebral near-infrared spectroscopy in adults: a work in progress. Anesth Analg. 2012;115(6):1373-1383. doi:10.1213/ANE.0B013E31826DD6A6 267 315. Barstow TJ. Understanding near infrared spectroscopy and its application to skeletal muscle research. J Appl Physiol. 2019;126(5):1360-1376. doi:10.1152/JAPPLPHYSIOL.00166.2018 316. Kagen LJ. Myoglobin: methods and diagnostic uses. CRC Crit Rev Clin Lab Sci. 1978;9(4):273-302. doi:10.3109/10408367809150922 317. Xie LK, Yang SH. Brain globins in physiology and pathology. Med Gas Res. 2016;6(3):154. doi:10.4103/2045-9912.191361 318. Balaban RS, Mootha VK, Arai A. Spectroscopic determination of cytochrome c oxidase content in tissues containing myoglobin or hemoglobin. Anal Biochem. 1996;237(2):274- 278. doi:10.1006/ABIO.1996.0239 319. Jöbsis FF. Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters. Science. 1977;198(4323):1264-1266. doi:10.1126/SCIENCE.929199 320. Sunwoo J, Shah P, Thuptimdang W, Khaleel M, Coates T, Khoo M. Estimation of cognitive brain activity in sickle cell disease using functional near-infrared spectroscopy and dynamic systems modeling. Front Hum Neurosci. 2018;12. doi:10.3389/conf.fnhum.2018.227.00096 321. Caldwell M, Scholkmann F, Wolf U, Wolf M, Elwell C, Tachtsidis I. Modelling confounding effects from extracerebral contamination and systemic factors on functional near-infrared spectroscopy. Neuroimage. 2016;143:91-105. doi:10.1016/j.neuroimage.2016.08.058 322. Kovacsova Z, Bale G, Mitra S, et al. Investigation of Confounding Factors in Measuring Tissue Saturation with NIRS Spatially Resolved Spectroscopy. Adv Exp Med Biol. 2018;1072:307-312. doi:10.1007/978-3-319-91287-5_49 323. A la C, G G, S H-S. In vivo validation of cerebral near-infrared spectroscopy: a review. Neurophotonics. 2018;5(4):1. doi:10.1117/1.NPH.5.4.040901 324. Bay-Hansen R, Elfving B, Greisen G. Use of near infrared spectroscopy for estimation of peripheral venous saturation in newborns: comparison with co-oximetry of central venous blood. Biol Neonate. 2002;82(1):1-8. doi:10.1159/000064145 325. Hogue CW, Levine A, Hudson A, Lewis C. Clinical Applications of Near-infrared Spectroscopy Monitoring in Cardiovascular Surgery. Anesthesiology. 2021;134(5):784- 791. doi:10.1097/ALN.0000000000003700 326. Casati A, Fanelli G, Pietropaoli P, et al. Monitoring cerebral oxygen saturation in elderly patients undergoing general abdominal surgery: a prospective cohort study. Eur J Anaesthesiol. 2007;24(1):59-65. doi:10.1017/S0265021506001025 327. G G, M Z, R R, et al. Time-domain near-infrared spectroscopy in acute ischemic stroke patients. Neurophotonics. 2019;6(1):1. doi:10.1117/1.NPH.6.1.015003 328. Roldán M, Kyriacou PA. Near-Infrared Spectroscopy (NIRS) in Traumatic Brain Injury (TBI). Sensors (Basel). 2021;21(5):1-30. doi:10.3390/S21051586 268 329. Crispin P, Forwood K. Near Infrared Spectroscopy in Anemia Detection and Management: A Systematic Review. Transfus Med Rev. 2021;35(1):22-28. doi:10.1016/J.TMRV.2020.07.003 330. [Near-infrared spectroscopy in psychiatry] - PubMed. https://pubmed.ncbi.nlm.nih.gov/22308262/. Accessed July 6, 2022. 331. Ehlis AC, Schneider S, Dresler T, Fallgatter AJ. Application of functional near-infrared spectroscopy in psychiatry. Neuroimage. 2014;85 Pt 1:478-488. doi:10.1016/J.NEUROIMAGE.2013.03.067 332. Sun Y, Ferguson BS, Rogatzki MJ, McDonald JR, Gladden LB. Muscle Near-Infrared Spectroscopy Signals versus Venous Blood Hemoglobin Oxygen Saturation in Skeletal Muscle. Med Sci Sports Exerc. 2016;48(10):2013-2020. doi:10.1249/MSS.0000000000001001 333. He X, Zhu M, Yablonskiy DA. Validation of oxygen extraction fraction measurement by qBOLD technique. Magn Reson Med. 2008;60(4):882-888. doi:10.1002/MRM.21719 334. Shen J, Miao X, Vu C, et al. Anemia Increases Oxygen Extraction Fraction in Deep Brain Structures but Not in the Cerebral Cortex. Front Physiol. 2022;13:896006. doi:10.3389/fphys.2022.896006 335. Numaguchi Y, Haller JS, Humbert JR, et al. Cerebral blood flow mapping using stable xenon-enhanced CT in sickle cell cerebrovascular disease. Neuroradiology. 1990;32(4):289-295. doi:10.1007/BF00593047 336. Atwi S, Shao H, Crane DE, et al. BOLD-based cerebrovascular reactivity vascular transfer function isolates amplitude and timing responses to better characterize cerebral small vessel disease. NMR Biomed. 2019;32(3). doi:10.1002/nbm.4064 337. Sam K, Conklin J, Holmes KR, et al. Impaired dynamic cerebrovascular response to hypercapnia predicts development of white matter hyperintensities. NeuroImage Clin. 2016;11:796-801. doi:10.1016/J.NICL.2016.05.008 338. Buxton RB, Frank LR. A model for the coupling between cerebral blood flow and oxygen metabolism during neural stimulation. J Cereb Blood Flow Metab. 1997;17(1):64-72. doi:10.1097/00004647-199701000-00009 339. Sakadžić S, Mandeville ET, Gagnon L, et al. Large arteriolar component of oxygen delivery implies a safe margin of oxygen supply to cerebral tissue. Nat Commun. 2014;5. doi:10.1038/ncomms6734 340. Kim MB, Ward DS, Cartwright CR, Kolano J, Chlebowski S, Henson LC. Estimation of jugular venous O2 saturation from cerebral oximetry or arterial O2 saturation during isocapnic hypoxia. J Clin Monit Comput. 2000;16(3):191-199. doi:10.1023/A:1009940031063 269 341. Ter Minassian A, Poirier N, Pierrot M, et al. Correlation between cerebral oxygen saturation measured by near-infrared spectroscopy and jugular oxygen saturation in patients with severe closed head injury. Anesthesiology. 1999;91(4):985-990. doi:10.1097/00000542- 199910000-00018 342. Lu H, Xu F, Grgac K, Liu P, Qin Q, Van Zijl P. Calibration and validation of TRUST MRI for the estimation of cerebral blood oxygenation. Magn Reson Med. 2012;67(1):42-49. doi:10.1002/MRM.22970 343. Li W, Xu X, Liu P, et al. Quantification of whole-brain oxygenation extraction fraction and cerebral metabolic rate of oxygen consumption in adults with sickle cell anemia using individual T 2 -based oxygenation calibrations. Magn Reson Med. 2019;83(3):1066-1080. doi:10.1002/mrm.27972 344. Golay X, Johanna Silvennoinen M, Zhou J, et al. Measurement of tissue oxygen extraction ratios from venous blood T(2): increased precision and validation of principle. Magn Reson Med. 2001;46(2):282-291. doi:10.1002/MRM.1189 345. Wright GA, Hu BS, Macovski A. 1991 I.I. Rabi Award. Estimating oxygen saturation of blood in vivo with MR imaging at 1.5 T. J Magn Reson Imaging. 1991;1(3):275-283. doi:10.1002/JMRI.1880010303 346. Silvennoinen MJ, Clingman CS, Golay X, Kauppinen RA, Van Zijl PCM. Comparison of the dependence of blood R2 and R2* on oxygen saturation at 1.5 and 4.7 Tesla. Magn Reson Med. 2003;49(1):47-60. doi:10.1002/MRM.10355 347. Zhao JM, Clingman CS, Närväinen MJ, Kauppinen RA, Van Zijl PCM. Oxygenation and hematocrit dependence of transverse relaxation rates of blood at 3T. Magn Reson Med. 2007;58(3):592-597. doi:10.1002/MRM.21342 348. Lu H, Ge Y. Quantitative evaluation of oxygenation in venous vessels using T2-Relaxation- Under-Spin-Tagging MRI. Magn Reson Med. 2008;60(2):357-363. doi:10.1002/mrm.21627 349. Benga G, Borza T. Diffusional water permeability of mammalian red blood cells. Comp Biochem Physiol B Biochem Mol Biol. 1995;112(4):653-659. doi:10.1016/0305- 0491(95)00116-6 350. Bush A, Borzage M, Detterich J, et al. Empirical model of human blood transverse relaxation at 3 T improves MRI T 2 oximetry. Magn Reson Med. 2017;77(6):2364-2371. doi:10.1002/mrm.26311 351. Eldeniz C, Binkley MM, Fields M, et al. Bulk volume susceptibility difference between deoxyhemoglobin and oxyhemoglobin for HbA and HbS: A comparative study. Magn Reson Med. January 2021:mrm.28668. doi:10.1002/mrm.28668 352. Permeability characteristics of deoxygenated sickle cells - PubMed. https://pubmed.ncbi.nlm.nih.gov/2122921/. Accessed July 6, 2022. 353. Kiselev VG, Novikov DS. Transverse NMR relaxation as a probe of mesoscopic structure. Phys Rev Lett. 2002;89(27). doi:10.1103/PHYSREVLETT.89.278101 270 354. Krishnamurthy LC, Liu P, Ge Y, Lu H. Vessel-specific quantification of blood oxygenation with T2-relaxation-under-phase-contrast MRI. Magn Reson Med. 2014;71(3):978-989. doi:10.1002/mrm.24750 355. Krishnamurthy LC, Mao D, King KS, Lu H. Correction and optimization of a T2-based approach to map blood oxygenation in small cerebral veins. Magn Reson Med. 2016;75(3):1100-1109. doi:10.1002/MRM.25686 356. Xu F, Uh J, Liu P, Lu H. On improving the speed and reliability of T2-relaxation-under-spin- tagging (TRUST) MRI. Magn Reson Med. 2012;68(1):198-204. doi:10.1002/mrm.23207 357. Jiang D, Deng S, Franklin CG, et al. Validation of T 2 -based oxygen extraction fraction measurement with 15 O positron emission tomography. Magn Reson Med. 2021;85(1):290-297. doi:10.1002/MRM.28410 358. Liu P, Dimitrov I, Andrews T, et al. Multisite evaluations of a T2 -relaxation-under-spin- tagging (TRUST) MRI technique to measure brain oxygenation. Magn Reson Med. 2016;75(2):680-687. doi:10.1002/MRM.25627 359. Jiang D, Liu P, Li Y, Mao D, Xu C, Lu H. Cross-vendor harmonization of T 2 -relaxation- under-spin-tagging (TRUST) MRI for the assessment of cerebral venous oxygenation. Magn Reson Med. 2018;80(3):1125-1131. doi:10.1002/MRM.27080 360. Liu P, Lu H, Filbey FM, Tamminga CA, Cao Y, Adinoff B. MRI assessment of cerebral oxygen metabolism in cocaine-addicted individuals: hypoactivity and dose dependence. NMR Biomed. 2014;27(6):726-732. doi:10.1002/NBM.3114 361. Jiang D, Lin Z, Liu P, et al. Brain Oxygen Extraction Is Differentially Altered by Alzheimer’s and Vascular Diseases. J Magn Reson Imaging. 2020;52(6):1829-1837. doi:10.1002/JMRI.27264 362. In vivo measurement of blood oxygen saturation using magnetic resonance imaging: a direct validation of the blood oxygen level-dependent concept in functional brain imaging - PubMed. https://pubmed.ncbi.nlm.nih.gov/20408238/. Accessed July 6, 2022. 363. Fernández-Seara MA, Techawiboonwong A, Detre JA, Wehrli FW. MR susceptometry for measuring global brain oxygen extraction. Magn Reson Med. 2006;55(5):967-973. doi:10.1002/MRM.20892 364. Jain V, Langham MC, Wehrli FW. MRI estimation of global brain oxygen consumption rate. J Cereb Blood Flow Metab. 2010;30(9):1598-1607. doi:10.1038/JCBFM.2010.49 365. Fan AP, Benner T, Bolar DS, Rosen BR, Adalsteinsson E. Phase-based regional oxygen metabolism (PROM) using MRI. Magn Reson Med. 2012;67(3):669-678. doi:10.1002/MRM.23050 366. Weisskoff RM, Kiihne S. MRI susceptometry: image-based measurement of absolute susceptibility of MR contrast agents and human blood. Magn Reson Med. 1992;24(2):375- 383. http://www.ncbi.nlm.nih.gov/pubmed/1569876. Accessed May 31, 2019. 271 367. Spees WM, Yablonskiy DA, Oswood MC, Ackerman JJ. Water proton MR properties of human blood at 1.5 Tesla: magnetic susceptibility, T(1), T(2), T*(2), and non-Lorentzian signal behavior. Magn Reson Med. 2001;45(4):533-542. http://www.ncbi.nlm.nih.gov/pubmed/11283978. Accessed May 31, 2019. 368. Ibaraki M, Shimosegawa E, Miura S, et al. PET measurements of CBF, OEF, and CMRO2 without arterial sampling in hyperacute ischemic stroke: method and error analysis. Ann Nucl Med. 2004;18(1):35-44. doi:10.1007/BF02985612 369. Brain oxygen utilization measured with O-15 radiotracers and positron emission tomography - PubMed. https://pubmed.ncbi.nlm.nih.gov/6610032/. Accessed July 6, 2022. 370. Blockley NP, Griffeth VEM, Simon AB, Buxton RB. A review of calibrated blood oxygenation level-dependent (BOLD) methods for the measurement of task-induced changes in brain oxygen metabolism. NMR Biomed. 2013;26(8):987-1003. doi:10.1002/nbm.2847 371. Davis TL, Kwong KK, Weisskoff RM, Rosen BR. Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci U S A. 1998;95(4):1834-1839. doi:10.1073/PNAS.95.4.1834 372. Grubb RL, Raichle ME, Eichling JO, Ter-Pogossian MM. The effects of changes in PaCO2 on cerebral blood volume, blood flow, and vascular mean transit time. Stroke. 1974;5(5):630-639. doi:10.1161/01.STR.5.5.630 373. Fernández-Seara MA, Rodgers ZB, Englund EK, Wehrli FW. Calibrated bold fMRI with an optimized ASL-BOLD dual-acquisition sequence. Neuroimage. 2016;142:474-482. doi:10.1016/J.NEUROIMAGE.2016.08.007 374. Investigation of BOLD signal dependence on cerebral blood flow and oxygen consumption: the deoxyhemoglobin dilution model - PubMed. https://pubmed.ncbi.nlm.nih.gov/10542343/. Accessed July 6, 2022. 375. Blockley NP, Griffeth VEM, Simon AB, Dubowitz DJ, Buxton RB. Calibrating the BOLD response without administering gases: comparison of hypercapnia calibration with calibration using an asymmetric spin echo. Neuroimage. 2015;104:423-429. doi:10.1016/J.NEUROIMAGE.2014.09.061 376. Chen JJ, Pike GB. BOLD-specific cerebral blood volume and blood flow changes during neuronal activation in humans. NMR Biomed. 2009;22(10):1054-1062. doi:10.1002/NBM.1411 377. Mark CI, Fisher JA, Pike GB. Improved fMRI calibration: precisely controlled hyperoxic versus hypercapnic stimuli. Neuroimage. 2011;54(2):1102-1111. doi:10.1016/J.NEUROIMAGE.2010.08.070 378. Gauthier CJ, Hoge RD. Magnetic resonance imaging of resting OEF and CMRO₂ using a generalized calibration model for hypercapnia and hyperoxia. Neuroimage. 2012;60(2):1212-1225. doi:10.1016/J.NEUROIMAGE.2011.12.056 379. Severinghaus JW. Simple, accurate equations for human blood O2 dissociation computations. J Appl Physiol. 1979;46(3):599-602. doi:10.1152/JAPPL.1979.46.3.599 272 380. Bulte DP, Kelly M, Germuska M, et al. Quantitative measurement of cerebral physiology using respiratory-calibrated MRI. Neuroimage. 2012;60(1):582-591. doi:10.1016/J.NEUROIMAGE.2011.12.017 381. Englund EK, Fernández-Seara MA, Rodríguez-Soto AE, et al. Calibrated fMRI for dynamic mapping of CMRO 2 responses using MR-based measurements of whole-brain venous oxygen saturation. J Cereb Blood Flow Metab. 2020;40(7):1501-1516. doi:10.1177/0271678X19867276 382. Xu F, Ge Y, Lu H. Noninvasive quantification of whole-brain cerebral metabolic rate of oxygen (CMRO2) by MRI. Magn Reson Med. 2009;62(1):141-148. doi:10.1002/MRM.21994 383. Liu P, Xu F, Lu H. Test-retest reproducibility of a rapid method to measure brain oxygen metabolism. Magn Reson Med. 2013;69(3):675-681. doi:10.1002/MRM.24295 384. Liu P, Huang H, Rollins N, et al. Quantitative assessment of global cerebral metabolic rate of oxygen (CMRO2) in neonates using MRI. NMR Biomed. 2014;27(3):332-340. doi:10.1002/NBM.3067 385. Shiroishi MS, Castellazzi G, Boxerman JL, et al. Principles of T 2 *-weighted dynamic susceptibility contrast MRI technique in brain tumor imaging. J Magn Reson Imaging. 2015;41(2):296-313. doi:10.1002/jmri.24648 386. Yamada K, Wu O, Gonzalez RG, et al. Magnetic resonance perfusion-weighted imaging of acute cerebral infarction: effect of the calculation methods and underlying vasculopathy. Stroke. 2002;33(1):87-94. http://www.ncbi.nlm.nih.gov/pubmed/11779894. Accessed June 3, 2019. 387. Young GS. Advanced MRI of Adult Brain Tumors. Neurol Clin. 2007;25(4):947-973. doi:10.1016/j.ncl.2007.07.010 388. Beaumont A, Fatouros P, Gennarelli T, Corwin F, Marmarou A. Bolus tracer delivery measured by MRI confirms edema without blood-brain barrier permeability in diffuse traumatic brain injury. Acta Neurochir Suppl. 2006;96:171-174. http://www.ncbi.nlm.nih.gov/pubmed/16671449. Accessed June 3, 2019. 389. Agarwal R, Brunelli SM, Williams K, Mitchell MD, Feldman HI, Umscheid CA. Gadolinium- based contrast agents and nephrogenic systemic fibrosis: a systematic review and meta- analysis. Nephrol Dial Transplant. 2008;24(3):856-863. doi:10.1093/ndt/gfn593 390. Broome DR, Girguis MS, Baron PW, Cottrell AC, Kjellin I, Kirk GA. Gadodiamide- Associated Nephrogenic Systemic Fibrosis: Why Radiologists Should Be Concerned. Am J Roentgenol. 2007;188(2):586-592. doi:10.2214/AJR.06.1094 391. Murata N, Gonzalez-Cuyar LF, Murata K, et al. Macrocyclic and Other Non–Group 1 Gadolinium Contrast Agents Deposit Low Levels of Gadolinium in Brain and Bone Tissue. Invest Radiol. 2016;51(7):447-453. doi:10.1097/RLI.0000000000000252 273 392. McDonald RJ, McDonald JS, Kallmes DF, et al. Intracranial Gadolinium Deposition after Contrast-enhanced MR Imaging. Radiology. 2015;275(3):772-782. doi:10.1148/radiol.15150025 393. Wáng Y-XJ, Schroeder J, Siegmund H, et al. Total gadolinium tissue deposition and skin structural findings following the administration of structurally different gadolinium chelates in healthy and ovariectomized female rats. Quant Imaging Med Surg. 2015;5(4):534-545. doi:10.3978/j.issn.2223-4292.2015.05.03 394. Losert C, Peller M, Schneider P, Reiser M. Oxygen-enhanced MRI of the brain. Magn Reson Med. 2002;48(2):271-277. doi:10.1002/mrm.10215 395. Chai Y, Bush AM, Coloigner J, et al. White Matter Has Impaired Resting Oxygen Delivery in Sickle Cell Patients. Am J Hematol. 2019;94(4):467-474. doi:10.1002/ajh.25423 396. Bush AM, Borzage MT, Choi S, et al. Determinants of resting cerebral blood flow in sickle cell disease. Am J Hematol. 2016;91(9):912-917. doi:10.1002/ajh.24441 397. Borzage MT, Bush AM, Choi S, et al. Predictors of cerebral blood flow in patients with and without anemia. J Appl Physiol. 2016;120:976-981. doi:10.1152/japplphysiol.00994.2015 398. Peng S-L, Su P, Wang F-N, et al. Optimization of phase-contrast MRI for the quantification of whole-brain cerebral blood flow. J Magn Reson Imaging. 2015;42(4):1126-1133. doi:10.1002/jmri.24866 399. Østergaard L. Principles of cerebral perfusion imaging by bolus tracking. J Magn Reson Imaging. 2005;22(6):710-717. doi:10.1002/jmri.20460 400. Simonsen CZ, Ostergaard L, Vestergaard-Poulsen P, Røhl L, Bjørnerud A, Gyldensted C. CBF and CBV measurements by USPIO bolus tracking: reproducibility and comparison with Gd-based values. J Magn Reson Imaging. 1999;9(2):342-347. http://www.ncbi.nlm.nih.gov/pubmed/10077035. Accessed June 27, 2019. 401. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. Perfusion imaging with NMR contrast agents. Magn Reson Med. 1990;14(2):249-265. http://www.ncbi.nlm.nih.gov/pubmed/2345506. Accessed April 22, 2019. 402. Knutsson L, van Westen D, Petersen ET, et al. Absolute quantification of cerebral blood flow: correlation between dynamic susceptibility contrast MRI and model-free arterial spin labeling. Magn Reson Imaging. 2010;28(1):1-7. doi:10.1016/J.MRI.2009.06.006 403. Rempp KA, Brix G, Wenz F, Becker CR, Gückel F, Lorenz WJ. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology. 1994;193(3):637-641. doi:10.1148/radiology.193.3.7972800 404. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med. 1996;36(5):715-725. doi:10.1002/mrm.1910360510 274 405. Sangkatumvong S, Khoo MCK, Kato R, et al. Peripheral Vasoconstriction and Abnormal Parasympathetic Response to Sighs and Transient Hypoxia in Sickle Cell Disease. doi:10.1164/rccm.201103-0537OC 406. Coloigner J, Kim Y, Bush A, et al. Contrasting resting-state fMRI abnormalities from sickle and non-sickle anemia. Kassner A, ed. PLoS One. 2017;12(10):e0184860. doi:10.1371/journal.pone.0184860 407. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL. Neuroimage. 2012;62(2):782-790. doi:10.1016/J.NEUROIMAGE.2011.09.015 408. Bush A, Chai Y, Choi SY, et al. Pseudo continuous arterial spin labeling quantification in anemic subjects with hyperemic cerebral blood flow. Magn Reson Imaging. 2018;47:137- 146. doi:10.1016/J.MRI.2017.12.011 409. Soares JM, Magalhães R, Moreira PS, et al. A Hitchhiker’s Guide to Functional Magnetic Resonance Imaging. Front Neurosci. 2016;10:515. doi:10.3389/fnins.2016.00515 410. Mikl M, Mareček R, Hluštík P, et al. Effects of spatial smoothing on fMRI group inferences. Magn Reson Imaging. 2008;26(4):490-503. doi:10.1016/j.mri.2007.08.006 411. Lin W, Celik A, Derdeyn C, et al. Quantitative measurements of cerebral blood flow in patients with unilateral carotid artery occlusion: a PET and MR study. J Magn Reson Imaging. 2001;14(6):659-667. http://www.ncbi.nlm.nih.gov/pubmed/11747021. Accessed April 22, 2019. 412. Bulte D, Chiarelli P, Wise R, Jezzard P. Measurement of cerebral blood volume in humans using hyperoxic MRI contrast. J Magn Reson Imaging. 2007;26(4):894-899. doi:10.1002/jmri.21096 413. Grabner G, Janke AL, Budge MM, Smith D, Pruessner J, Collins DL. Symmetric atlasing and model based segmentation: an application to the hippocampus in older adults. Med Image Comput Comput Assist Interv. 2006;9(Pt 2):58-66. http://www.ncbi.nlm.nih.gov/pubmed/17354756. Accessed May 6, 2019. 414. Tatu L, Moulin T, Vuillier F, Bogousslavsky J. Arterial Territories of the Human Brain. In: Frontiers of Neurology and Neuroscience. Vol 30. ; 2012:99-110. doi:10.1159/000333602 415. Guensch DP, Fischer K, Yamaji K, et al. Effect of Hyperoxia on Myocardial Oxygenation and Function in Patients With Stable Multivessel Coronary Artery Disease. J Am Heart Assoc. 2020;9(5):e014739. doi:10.1161/JAHA.119.014739 416. Cox EF, Palaniyappan N, Aithal GP, Guha IN, Francis ST. Using MRI to study the alterations in liver blood flow, perfusion, and oxygenation in response to physiological stress challenges: Meal, hyperoxia, and hypercapnia. J Magn Reson Imaging. 2019;49(6):1577-1586. doi:10.1002/jmri.26341 417. Wengler K, Wang J, Serrano Sosa M, et al. Mapping hepatic blood oxygenation by quantitative BOLD (qBOLD) MRI. Magn Reson Med. 2019;81(5):3272-3282. doi:10.1002/mrm.27642 275 418. van den Boomen M, Manhard MK, Snel GJH, et al. Blood oxygen level–dependent MRI of the myocardium with multiecho gradient-echo spin-echo imaging. Radiology. 2020;294(2):538-545. doi:10.1148/radiol.2020191845 419. Epstein FH, Meyer CH. Myocardial perfusion using arterial spin labeling CMR: promise and challenges. JACC Cardiovasc Imaging. 2011;4(12):1262-1264. doi:10.1016/j.jcmg.2011.08.015 420. Sangkatumvong S, Khoo MCK, Coates TD. Abnormal cardiac autonomic control in sickle cell disease following transient hypoxia. In: 2008 30th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE; 2008:1996-1999. doi:10.1109/IEMBS.2008.4649581 421. Bickler PE, Feiner JR, Lipnick MS, Batchelder P, MacLeod DB, Severinghaus JW. Effects of Acute, Profound Hypoxia on Healthy Humans. Anesth Analg. 2017;124(1):146-153. doi:10.1213/ANE.0000000000001421 422. van Osch MJP, van der Grond J, Bakker CJG. Partial volume effects on arterial input functions: Shape and amplitude distortions and their correction. J Magn Reson Imaging. 2005;22(6):704-709. doi:10.1002/jmri.20455 423. Hansen AE, Pedersen H, Rostrup E, Larsson HBW. Partial volume effect (PVE) on the arterial input function (AIF) in T 1 -weighted perfusion imaging and limitations of the multiplicative rescaling approach. Magn Reson Med. 2009;62(4):1055-1059. doi:10.1002/mrm.22098 424. Knutsson L, Börjesson S, Larsson E-M, et al. Absolute quantification of cerebral blood flow in normal volunteers: Correlation between Xe-133 SPECT and dynamic susceptibility contrast MRI. J Magn Reson Imaging. 2007;26(4):913-920. doi:10.1002/jmri.21093 425. Dolui S, Wang Z, Wang DJJ, et al. Comparison of non-invasive MRI measurements of cerebral blood flow in a large multisite cohort. J Cereb Blood Flow Metab. 2016;36(7):1244- 1256. doi:10.1177/0271678X16646124 426. Knutsson L, Ståhlberg F, Wirestam R. Aspects on the accuracy of cerebral perfusion parameters obtained by dynamic susceptibility contrast MRI: a simulation study. Magn Reson Imaging. 2004;22(6):789-798. doi:10.1016/J.MRI.2003.12.002 427. Juttukonda MR, Jordan LC, Gindville MC, et al. Cerebral hemodynamics and pseudo- continuous arterial spin labeling considerations in adults with sickle cell anemia. NMR Biomed. 2017;30(2):e3681. doi:10.1002/nbm.3681 428. Fields ME, Guilliams KP, Ragan DK, et al. Regional oxygen extraction predicts border zone vulnerability to stroke in sickle cell disease. Neurology. March 2018:10.1212/WNL.0000000000005194. doi:10.1212/WNL.0000000000005194 429. Mandell DM, Han JS, Poublanc J, et al. Selective Reduction of Blood Flow to White Matter During Hypercapnia Corresponds With Leukoaraiosis. Stroke. 2008;39(7):1993-1998. doi:10.1161/STROKEAHA.107.501692 276 430. Lu H, Clingman C, Golay X, van Zijl PCM. Determining the longitudinal relaxation time (T1) of blood at 3.0 Tesla. Magn Reson Med. 2004;52(3):679-682. doi:10.1002/mrm.20178 431. Wu W-C, Jain V, Li C, et al. In vivo venous blood T1 measurement using inversion recovery true-FISP in children and adults. Magn Reson Med. 2010;64(4):1140-1147. doi:10.1002/mrm.22484 432. Mark CI, Fisher JA, Pike GB. Improved fMRI calibration: Precisely controlled hyperoxic versus hypercapnic stimuli. Neuroimage. 2011;54(2):1102-1111. doi:10.1016/j.neuroimage.2010.08.070 433. Xu F, Li W, Liu P, et al. Accounting for the role of hematocrit in between-subject variations of MRI-derived baseline cerebral hemodynamic parameters and functional BOLD responses. Hum Brain Mapp. 2018;39(1):344-353. doi:10.1002/hbm.23846 434. Calamante F, Mørup M, Hansen LK. Defining a local arterial input function for perfusion MRI using independent component analysis. Magn Reson Med. 2004;52(4):789-797. doi:10.1002/mrm.20227 435. Mehndiratta A, Calamante F, Macintosh BJ, Crane DE, Payne SJ, Chappell MA. Modeling and correction of bolus dispersion effects in dynamic susceptibility contrast MRI. Magn Reson Med. 2014;72(6):1762-1774. doi:10.1002/mrm.25077 436. Wu O, Østergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. Tracer arrival timing-insensitive technique for estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magn Reson Med. 2003;50(1):164-174. doi:10.1002/mrm.10522 437. Mutke MA, Madai VI, von Samson-Himmelstjerna FC, et al. Clinical evaluation of an arterial-spin-labeling product sequence in steno-occlusive disease of the brain. PLoS One. 2014;9(2):e87143. doi:10.1371/journal.pone.0087143 438. D’Ortenzio RM, Hojjat SP, Vitorino R, et al. Comparison of Quantitative Cerebral Blood Flow Measurements Performed by Bookend Dynamic Susceptibility Contrast and Arterial Spin-Labeling MRI in Relapsing-Remitting Multiple Sclerosis. AJNR Am J Neuroradiol. 2016;37(12):2265-2272. doi:10.3174/ajnr.A4939 439. Reginster P, Martin B, Denolin V. Comparative Study of Pseudo-Continuous Arterial Spin Labeling and Dynamic Susceptibility Contrast Imaging at 3.0 Tesla in Brain Tumors. 2017. https://www.semanticscholar.org/paper/Comparative-Study-of-Pseudo-Continuous- Arterial-and-Reginster-Martin/27d4956225eb94cc345f3eb3d6eeb4381123de58. Accessed October 16, 2019. 440. Xu F, Liu P, Pascual JM, Xiao G, Lu H. Effect of Hypoxia and Hyperoxia on Cerebral Blood Flow, Blood Oxygenation, and Oxidative Metabolism. J Cereb Blood Flow Metab. 2012;32(10):1909-1918. doi:10.1038/jcbfm.2012.93 441. LLOYD BB, JUKES MG, CUNNINGHAM DJ. The relation between alveolar oxygen pressure and the respiratory response to carbon dioxide in man. Q J Exp Physiol Cogn Med Sci. 1958;43(2):214-227. http://www.ncbi.nlm.nih.gov/pubmed/13542754. Accessed June 3, 2019. 277 442. Slessarev M, Han J, Mardimae A, et al. Prospective targeting and control of end-tidal CO2 and O2 concentrations. J Physiol. 2007;581(Pt 3):1207-1219. doi:10.1113/jphysiol.2007.129395 443. Dale AM, Buckner RL. Selective averaging of rapidly presented individual trials using fMRI. Hum Brain Mapp. 1997;5(5):329-340. doi:10.1002/(SICI)1097-0193(1997)5:5<329::AID- HBM1>3.0.CO;2-5 444. Essig M, Shiroishi MS, Nguyen TB, et al. Perfusion MRI: the five most frequently asked technical questions. AJR Am J Roentgenol. 2013;200(1):24-34. doi:10.2214/AJR.12.9543 445. Schmainda KM, Prah M, Connelly J, et al. Dynamic-susceptibility contrast agent MRI measures of relative cerebral blood volume predict response to bevacizumab in recurrent high-grade glioma. Neuro Oncol. 2014;16(6):880-888. doi:10.1093/NEUONC/NOT216 446. Schellinger PD, Bryan RN, Caplan LR, et al. Evidence-based guideline: The role of diffusion and perfusion MRI for the diagnosis of acute ischemic stroke: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology. 2010;75(2):177-185. doi:10.1212/WNL.0B013E3181E7C9DD 447. Quan G, Zhang K, Liu Y, et al. Role of Dynamic Susceptibility Contrast Perfusion MRI in Glioma Progression Evaluation. J Oncol. 2021;2021. doi:10.1155/2021/1696387 448. Hillis AE, Wityk RJ, Beauchamp NJ, Ulatowski JA, Jacobs MA, Barker PB. Perfusion- weighted MRI as a marker of response to treatment in acute and subacute stroke. Neuroradiology. 2004;46(1):31-39. doi:10.1007/S00234-002-0918-4 449. Reiter T, Ritter O, Prince MR, et al. Minimizing risk of nephrogenic systemic fibrosis in cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2012;14(1). doi:10.1186/1532-429X-14-31 450. Choyke P, Cady J, DePollar S, Austin H. Determination of serum creatinine prior to iodinated contrast media: is it necessary in all patients? Tech Urol. 1998;4(2):65-69. https://pubmed.ncbi.nlm.nih.gov/9623618/. Accessed July 2, 2022. 451. Too CW, Ng WY, Tan CC, Mahmood MI, Tay KH. Screening for impaired renal function in outpatients before iodinated contrast injection: Comparing the Choyke questionnaire with a rapid point-of-care-test. Eur J Radiol. 2015;84(7):1227-1231. doi:10.1016/J.EJRAD.2015.04.001 452. Slessarev M, Han J, Mardimae A, et al. Prospective targeting and control of end-tidal CO2 and O2 concentrations. J Physiol. 2007;581(Pt 3):1207-1219. doi:10.1113/JPHYSIOL.2007.129395 453. Severinghaus JW. Simple, accurate equations for human blood O2 dissociation computations. J Appl Physiol. 1979;46(3):599-602. doi:10.1152/JAPPL.1979.46.3.599 454. Joshi AA, Choi S, Liu Y, et al. A hybrid high-resolution anatomical MRI atlas with sub- parcellation of cortical gyri using resting fMRI. J Neurosci Methods. 2022;374:109566. doi:10.1016/j.jneumeth.2022.109566 278 455. Mao D, Li Y, Liu P, Peng S-L, Pillai JJ, Lu H. Three-dimensional mapping of brain venous oxygenation using R2* oximetry. Magn Reson Med. 2018;79(3):1304-1313. doi:10.1002/mrm.26763 456. Carroll TJ, Rowley HA, Haughton VM. Automatic calculation of the arterial input function for cerebral perfusion imaging with MR imaging. Radiology. 2003;227(2):593-600. doi:10.1148/RADIOL.2272020092 457. Schulman J, Sayin ES, Manalac A, et al. Perfusion Quantification in the Human Brain Using DSC MRI – Simulations and Validations at 3T. bioRxiv. 2022:2022.04.27.489686. 458. Liu P, Liu G, Pinho MC, et al. Cerebrovascular Reactivity Mapping Using Resting-State BOLD Functional MRI in Healthy Adults and Patients with Moyamoya Disease. Radiology. 2021;299(2):419-425. doi:10.1148/RADIOL.2021203568 459. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted G, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med. 1996;36(5):715-725. doi:10.1002/MRM.1910360510 460. Østergaard L. Principles of cerebral perfusion imaging by bolus tracking. J Magn Reson Imaging. 2005;22(6):710-717. doi:10.1002/JMRI.20460 461. Cerebral blood flow measurement by dynamic contrast MRI using singular value decomposition with an adaptive threshold - PubMed. https://pubmed.ncbi.nlm.nih.gov/10398963/. Accessed July 2, 2022. 462. Thompson MT. Intuitive Analog Circuit Design. 2nd ed. Oxford: Newnes; 2014. doi:10.1016/C2012-0-03027-X 463. R Core Team. R: A language and environment for statistical computing. 2021. 464. Bakdash JZ, Marusich LR. Repeated Measures Correlation. Front Psychol. 2017;8(MAR). doi:10.3389/FPSYG.2017.00456 465. Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat. 2007;17(4):571-582. doi:10.1080/10543400701329422 466. Giavarina D. Understanding Bland Altman analysis. Biochem medica. 2015;25(2):141-151. doi:10.11613/BM.2015.015 467. Grandin CB, Bol A, Smith AM, Michel C, Cosnard G. Absolute CBF and CBV measurements by MRI bolus tracking before and after acetazolamide challenge: repeatabilily and comparison with PET in humans. Neuroimage. 2005;26(2):525-535. doi:10.1016/J.NEUROIMAGE.2005.02.028 468. Ibaraki M, Ito H, Shimosegawa E, et al. Cerebral vascular mean transit time in healthy humans: a comparative study with PET and dynamic susceptibility contrast-enhanced MRI. J Cereb Blood Flow Metab. 2007;27(2):404-413. doi:10.1038/SJ.JCBFM.9600337 279 469. Paling D, Thade Petersen E, Tozer DJ, et al. Cerebral arterial bolus arrival time is prolonged in multiple sclerosis and associated with disability. J Cereb Blood Flow Metab. 2014;34(1):34-42. doi:10.1038/JCBFM.2013.161 470. MEIER P, ZIERLER KL. On the theory of the indicator-dilution method for measurement of blood flow and volume. J Appl Physiol. 1954;6(12):731-744. doi:10.1152/JAPPL.1954.6.12.731 471. Raisch DW, Garg V, Arabyat R, et al. Anaphylaxis associated with gadolinium-based contrast agents: data from the Food and Drug Administration’s Adverse Event Reporting System and review of case reports in the literature. Expert Opin Drug Saf. 2014;13(1):15- 23. doi:10.1517/14740338.2013.832752 472. Prince MR, Zhang H, Zou Z, Staron RB, Brill PW. Incidence of immediate gadolinium contrast media reactions. AJR Am J Roentgenol. 2011;196(2). doi:10.2214/AJR.10.4885 473. Hellman RN. Gadolinium-induced nephrogenic systemic fibrosis. Semin Nephrol. 2011;31(3):310-316. doi:10.1016/J.SEMNEPHROL.2011.05.010 474. Woolen SA, Shankar PR, Gagnier JJ, MacEachern MP, Singer L, Davenport MS. Risk of Nephrogenic Systemic Fibrosis in Patients With Stage 4 or 5 Chronic Kidney Disease Receiving a Group II Gadolinium-Based Contrast Agent: A Systematic Review and Meta- analysis. JAMA Intern Med. 2020;180(2):223-230. doi:10.1001/JAMAINTERNMED.2019.5284 475. Kanda T, Oba H, Toyoda K, Kitajima K, Furui S. Brain gadolinium deposition after administration of gadolinium-based contrast agents. Jpn J Radiol. 2016;34(1):3-9. doi:10.1007/S11604-015-0503-5 476. Gulani V, Calamante F, Shellock FG, Kanal E, Reeder SB. Gadolinium deposition in the brain: summary of evidence and recommendations. Lancet Neurol. 2017;16(7):564-570. doi:10.1016/S1474-4422(17)30158-8 477. Mathur M, Jones JR, Weinreb JC. Gadolinium Deposition and Nephrogenic Systemic Fibrosis: A Radiologist’s Primer. Radiographics. 2020;40(1):153-162. doi:10.1148/RG.2020190110 478. Murata N, Murata K, Gonzalez-Cuyar LF, Maravilla KR. Gadolinium tissue deposition in brain and bone. Magn Reson Imaging. 2016;34(10):1359-1365. doi:10.1016/J.MRI.2016.08.025 479. Maximova N, Gregori M, Zennaro F, Sonzogni A, Simeone R, Zanon D. Hepatic Gadolinium Deposition and Reversibility after Contrast Agent-enhanced MR Imaging of Pediatric Hematopoietic Stem Cell Transplant Recipients. Radiology. 2016;281(2):418- 426. doi:10.1148/RADIOL.2016152846 480. Netzer N, Strohl K, Faulhaber M, Gatterer H, Burtscher M. Hypoxia-related altitude illnesses. J Travel Med. 2013;20(4):247-255. doi:10.1111/JTM.12017 481. Luks AM, Swenson ER, Bärtsch P. Acute high-altitude sickness. Eur Respir Rev. 2017;26(143). doi:10.1183/16000617.0096-2016 280 482. Cameron L, Pilcher J, Weatherall M, Beasley R, Perrin K. The risk of serious adverse outcomes associated with hypoxaemia and hyperoxaemia in acute exacerbations of COPD. Postgrad Med J. 2012;88(1046):684-689. doi:10.1136/POSTGRADMEDJ-2012- 130809 483. Pighin S, Bonini N, Hadjichristidis C, Schena F, Savadori L. Decision making under stress: mild hypoxia leads to increased risk-taking. Stress. 2020;23(3):290-297. doi:10.1080/10253890.2019.1680634 484. Bickler PE, Feiner JR, Lipnick MS, Batchelder P, Macleod DB, Severinghaus JW. Effects of Acute, Profound Hypoxia on Healthy Humans: Implications for Safety of Tests Evaluating Pulse Oximetry or Tissue Oximetry Performance. Anesth Analg. 2017;124(1):146-153. doi:10.1213/ANE.0000000000001421 485. Rempp KA, Brix G, Wenz F, Becker CR, Gückel F, Lorenz WJ. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology. 1994;193(3):637-641. doi:10.1148/RADIOLOGY.193.3.7972800 486. Murase K, Shinohara M, Yamazaki Y. Accuracy of deconvolution analysis based on singular value decomposition for quantification of cerebral blood flow using dynamic susceptibility contrast-enhanced magnetic resonance imaging. Phys Med Biol. 2001;46(12):3147-3159. doi:10.1088/0031-9155/46/12/306 487. Obrig H, Neufang M, Wenzel R, et al. Spontaneous low frequency oscillations of cerebral hemodynamics and metabolism in human adults. Neuroimage. 2000;12(6):623-639. doi:10.1006/NIMG.2000.0657 488. Wise RG, Ide K, Poulin MJ, Tracey I. Resting fluctuations in arterial carbon dioxide induce significant low frequency variations in BOLD signal. Neuroimage. 2004;21(4):1652-1664. doi:10.1016/J.NEUROIMAGE.2003.11.025 489. Bell LC, Semmineh N, An H, et al. Evaluating Multisite rCBV Consistency from DSC-MRI Imaging Protocols and Postprocessing Software Across the NCI Quantitative Imaging Network Sites Using a Digital Reference Object (DRO). Tomogr (Ann Arbor, Mich). 2019;5(1):110-117. doi:10.18383/J.TOM.2018.00041 490. Chappell MA, Mehndiratta A, Calamante F. Correcting for large vessel contamination in dynamic susceptibility contrast perfusion MRI by extension to a physiological model of the vasculature. Magn Reson Med. 2015;74(1):280-290. doi:10.1002/MRM.25390 491. Calamante F. Perfusion MRI using dynamic-susceptibility contrast MRI: quantification issues in patient studies. Top Magn Reson Imaging. 2010;21(2):75-85. doi:10.1097/RMR.0B013E31821E53F5 492. Hoxworth JM, Eschbacher JM, Gonzales AC, et al. Performance of Standardized Relative CBV for Quantifying Regional Histologic Tumor Burden in Recurrent High-Grade Glioma: Comparison against Normalized Relative CBV Using Image-Localized Stereotactic Biopsies. AJNR Am J Neuroradiol. 2020;41(3):408-415. doi:10.3174/AJNR.A6486 281 493. Wetzel SG, Cha S, Johnson G, et al. Relative cerebral blood volume measurements in intracranial mass lesions: interobserver and intraobserver reproducibility study. Radiology. 2002;224(3):797-803. doi:10.1148/RADIOL.2243011014 494. WAQAR M, LEWIS D, AGUSHI E, GITTINS M, JACKSON A, COOPE D. Cerebral and tumoral blood flow in adult gliomas: a systematic review of results from magnetic resonance imaging. Br J Radiol. 2021;94(1125). doi:10.1259/BJR.20201450 495. Grandin CB, Duprez TP, Smith AM, et al. Usefulness of magnetic resonance-derived quantitative measurements of cerebral blood flow and volume in prediction of infarct growth in hyperacute stroke. Stroke. 2001;32(5):1147-1153. doi:10.1161/01.STR.32.5.1147 496. Weisskoff RM, Zuo CS, Boxerman JL, Rosen BR. Microscopic susceptibility variation and transverse relaxation: theory and experiment. Magn Reson Med. 1994;31(6):601-610. http://www.ncbi.nlm.nih.gov/pubmed/8057812. Accessed May 23, 2019. 497. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med. 1996;36(5):726-736. doi:10.1002/MRM.1910360511 498. Nöth U, Kotajima F, Deichmann R, Turner R, Corfield DR. Mapping of the cerebral vascular response to hypoxia and hypercapnia using quantitative perfusion MRI at 3 T. NMR Biomed. 2008;21(5):464-472. doi:10.1002/NBM.1210 499. Knutsson L, Ståhlberg F, Wirestam R. Aspects on the accuracy of cerebral perfusion parameters obtained by dynamic susceptibility contrast MRI: a simulation study. Magn Reson Imaging. 2004;22(6):789-798. doi:10.1016/J.MRI.2003.12.002 500. Gaudino S, Martucci M, Botto A, et al. Brain DSC MR Perfusion in Children: A Clinical Feasibility Study Using Different Technical Standards of Contrast Administration. AJNR Am J Neuroradiol. 2019;40(2):359-365. doi:10.3174/AJNR.A5954 501. Jahng G-H, Li K-L, Ostergaard L, Calamante F. Perfusion magnetic resonance imaging: a comprehensive update on principles and techniques. Korean J Radiol. 2014;15(5):554- 577. doi:10.3348/kjr.2014.15.5.554 502. Østergaard L. Principles of cerebral perfusion imaging by bolus tracking. J Magn Reson Imaging. 2005;22(6):710-717. doi:10.1002/JMRI.20460 503. Gadolinium Deposition: A Study Review - PubMed. https://pubmed.ncbi.nlm.nih.gov/33472877/. Accessed July 12, 2022. 504. MEIER P, ZIERLER KL. On the theory of the indicator-dilution method for measurement of blood flow and volume. J Appl Physiol. 1954;6(12):731-744. doi:10.1152/jappl.1954.6.12.731 505. Slessarev M, Han J, Mardimae A, et al. Prospective targeting and control of end-tidal CO2 and O2 concentrations. J Physiol. 2007;581(Pt 3):1207-1219. doi:10.1113/jphysiol.2007.129395 282 506. Fisher JA. The CO2 stimulus for cerebrovascular reactivity: Fixing inspired concentrations vs. targeting end-tidal partial pressures. J Cereb Blood Flow Metab. 2016;36(6):1004-1011. doi:10.1177/0271678X16639326 507. Shattuck DW, Leahy RM. BrainSuite: an automated cortical surface identification tool. Med Image Anal. 2002;6(2):129-142. doi:10.1016/S1361-8415(02)00054-3 508. Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res. 1996;29(3):162-173. http://www.ncbi.nlm.nih.gov/pubmed/8812068. Accessed May 18, 2018. 509. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL. Neuroimage. 2012;62(2):782-790. doi:10.1016/J.NEUROIMAGE.2011.09.015 510. Carroll TJ, Rowley HA, Haughton VM. Automatic Calculation of the Arterial Input Function for Cerebral Perfusion Imaging with MR Imaging. Radiology. 2003;227(2):593-600. doi:10.1148/radiol.2272020092 511. Chappell MA, MacIntosh BJ, Donahue MJ, Günther M, Jezzard P, Woolrich MW. Separation of macrovascular signal in multi-inversion time arterial spin labelling MRI. Magn Reson Med. 2010;63(5):1357-1365. doi:10.1002/MRM.22320 512. Afzali-Hashemi L, Baas KPA, Schrantee A, et al. Impairment of Cerebrovascular Hemodynamics in Patients With Severe and Milder Forms of Sickle Cell Disease. Front Physiol. 2021;12. doi:10.3389/FPHYS.2021.645205/FULL 513. Coloigner J, Vu C, Borzage M, et al. Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia Using BOLD MRI and Near-Infrared Spectroscopy. J Magn Reson Imaging. 2020;52(5). doi:10.1002/jmri.27210 514. Choi S, Bush AM, Borzage MT, et al. Hemoglobin and mean platelet volume predicts diffuse T1-MRI white matter volume decrease in sickle cell disease patients. NeuroImage Clin. 2017;15:239-246. doi:10.1016/j.nicl.2017.04.023 515. GRANDIN C, BOL A, SMITH A, MICHEL C, COSNARD G. Absolute CBF and CBV measurements by MRI bolus tracking before and after acetazolamide challenge: Repeatabilily and comparison with PET in humans. Neuroimage. 2005;26(2):525-535. doi:10.1016/j.neuroimage.2005.02.028 516. Ibaraki M, Ito H, Shimosegawa E, et al. Cerebral Vascular Mean Transit Time in Healthy Humans: A Comparative Study with PET and Dynamic Susceptibility Contrast-Enhanced MRI. J Cereb Blood Flow Metab. 2007;27(2):404-413. doi:10.1038/sj.jcbfm.9600337 517. Gitelman DR, Prohovnik I, Tatemichi TK. Safety of hypercapnic challenge: cardiovascular and neurologic considerations. J Cereb Blood Flow Metab. 1991;11(6):1036-1040. doi:10.1038/JCBFM.1991.172 518. Taylor CL, Selman WR, Ratcheson RA, et al. Steal affecting the central nervous system. Neurosurgery. 2002;50(4):679-689. doi:10.1097/00006123-200204000-00002 283 519. Arteaga DF, Strother MK, Faraco CC, et al. The vascular steal phenomenon is an incomplete contributor to negative cerebrovascular reactivity in patients with symptomatic intracranial stenosis. J Cereb Blood Flow Metab. 2014;34(9):1453-1462. doi:10.1038/JCBFM.2014.106 520. Poublanc J, Han JS, Mandell DM, et al. Vascular Steal Explains Early Paradoxical Blood Oxygen Level-Dependent Cerebrovascular Response in Brain Regions with Delayed Arterial Transit Times. Cerebrovasc Dis Extra. 2013;3(1):55-64. doi:10.1159/000348841 521. Ito H, Kanno I, Ibaraki M, Hatazawa J, Miura S. Changes in human cerebral blood flow and cerebral blood volume during hypercapnia and hypocapnia measured by positron emission tomography. J Cereb Blood Flow Metab. 2003;23(6):665-670. doi:10.1097/01.WCB.0000067721.64998.F5 522. Angleys H, Jespersen SN, Østergaard L. The effects of capillary transit time heterogeneity on the BOLD signal. Hum Brain Mapp. 2018;39(6):2329-2352. doi:10.1002/HBM.23991 523. Angleys H, Østergaard L, Jespersen SN. The effects of capillary transit time heterogeneity (CTH) on brain oxygenation. J Cereb Blood Flow Metab. 2015;35(5):806-817. doi:10.1038/JCBFM.2014.254 524. Blockley NP, Harkin JW, Bulte DP. Rapid cerebrovascular reactivity mapping: Enabling vascular reactivity information to be routinely acquired. Neuroimage. 2017;159:214-223. doi:10.1016/J.NEUROIMAGE.2017.07.048 525. Afzali-Hashemi L, Baas KPA, Schrantee A, et al. Impairment of Cerebrovascular Hemodynamics in Patients With Severe and Milder Forms of Sickle Cell Disease. Front Physiol. 2021;12. doi:10.3389/FPHYS.2021.645205 526. Poublanc J, Crawley AP, Sobczyk O, et al. Measuring cerebrovascular reactivity: the dynamic response to a step hypercapnic stimulus. J Cereb Blood Flow Metab. 2015;35(11):1746-1756. doi:10.1038/jcbfm.2015.114 527. Gale EM, Caravan P, Rao AG, et al. Gadolinium-based contrast agents in pediatric magnetic resonance imaging. Pediatr Radiol. 2017;47(5):507-521. doi:10.1007/S00247- 017-3806-0 528. Holowka S, Shroff M, Chavhan GB. Use and Safety of Gadolinium Based Contrast Agents in Pediatric MR Imaging. Indian J Pediatr. 2019;86(10):961-966. doi:10.1007/S12098-019- 02891-X 529. Brambrink A, Orfanakis A. “Therapeutic hypercapnia” after ischemic brain injury: is there a potential for neuroprotection? Anesthesiology. 2010;112(2):274-276. doi:10.1097/ALN.0B013E3181CA8273 530. Zhou Q, Cao B, Niu L, et al. Effects of permissive hypercapnia on transient global cerebral ischemia-reperfusion injury in rats. Anesthesiology. 2010;112(2):288-297. doi:10.1097/ALN.0B013E3181CA8257 284 531. Ho CY, Cardinal JS, Kamer AP, Lin C, Kralik S. Contrast Leakage Patterns from Dynamic Susceptibility Contrast Perfusion MRI in the Grading of Primary Pediatric Brain Tumors. Am J Neuroradiol. 2016;37(3):544-551. doi:10.3174/ajnr.A4559 532. Kluge A, Lukas M, Toth V, Pyka T, Zimmer C, Preibisch C. Analysis of three leakage- correction methods for DSC-based measurement of relative cerebral blood volume with respect to heterogeneity in human gliomas. Magn Reson Imaging. 2016;34(4):410-421. doi:10.1016/J.MRI.2015.12.015 533. Bjornerud A, Sorensen AG, Mouridsen K, Emblem KE. T1- and T2*-dominant extravasation correction in DSC-MRI: part I--theoretical considerations and implications for assessment of tumor hemodynamic properties. J Cereb Blood Flow Metab. 2011;31(10):2041-2053. doi:10.1038/JCBFM.2011.52 534. Coyer SM. Anemia: Diagnosis and Management. J Pediatr Heal Care. 2005;19(6):380- 385. doi:10.1016/j.pedhc.2005.07.014 535. Kaiafa G, Savopoulos C, Kanellos I, et al. Anemia and stroke: Where do we stand? Acta Neurol Scand. 2017;135(6):596-602. doi:10.1111/ane.12657 536. Kassim AA, Pruthi S, Day M, et al. Silent cerebral infarcts and cerebral aneurysms are prevalent in adults with sickle cell anemia. Blood. 2016;127(16):2038-2040. doi:10.1182/blood-2016-01-694562 537. Spivey JF, Uong EC, Strunk R, Boslaugh SE, DeBaun MR. Low daytime pulse oximetry reading is associated with nocturnal desaturation and obstructive sleep apnea in children with sickle cell anemia. Pediatr Blood Cancer. 2008;50(2):359-362. doi:10.1002/pbc.21054 538. Quinn CT, Sargent JW. Daytime steady-state haemoglobin desaturation is a risk factor for overt stroke in children with sickle cell anaemia. Br J Haematol. 2008;140(3):336-339. doi:10.1111/j.1365-2141.2007.06927.x 539. Jöbsis FF. Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters. Science. 1977;198(4323):1264-1267. http://www.ncbi.nlm.nih.gov/pubmed/929199. Accessed March 13, 2018. 540. Glover GH. Overview of Functional Magnetic Resonance Imaging. Neurosurg Clin N Am. 2011;22(2):133. doi:10.1016/J.NEC.2010.11.001 541. DeBaun MR, Gordon M, McKinstry RC, et al. Controlled Trial of Transfusions for Silent Cerebral Infarcts in Sickle Cell Anemia. N Engl J Med. 2014;371(8):699-710. doi:10.1056/NEJMoa1401731 542. Holmes CJ, Hoge R, Collins L, Woods R, Toga AW, Evans AC. Enhancement of MR images using registration for signal averaging. J Comput Assist Tomogr. 1998;22(2):324- 333. doi:10.1097/00004728-199803000-00032 543. Enzmann DR, Ross MR, Marks MP, Pelc NJ. Blood flow in major cerebral arteries measured by phase-contrast cine MR. AJNR Am J Neuroradiol. 1994;15(1):123-129. http://www.ncbi.nlm.nih.gov/pubmed/8141043. Accessed August 18, 2019. 285 544. DeBaun MR, Kirkham FJ. Central nervous system complications and management in sickle cell disease. Blood. 2016;127(7):829-838. doi:10.1182/blood-2015-09-618579 545. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular Accidents in Sickle Cell Disease: Rates and Risk Factors and the Cooperative Study of Sickle Cell Disease. http://www.bloodjournal.org/content/bloodjournal/91/1/288.full.pdf. Accessed March 13, 2018. 546. Kinney TR, Sleeper LA, Wang WC, et al. Silent cerebral infarcts in sickle cell anemia: a risk factor analysis. The Cooperative Study of Sickle Cell Disease. Pediatrics. 1999;103(3):640-645. http://www.ncbi.nlm.nih.gov/pubmed/10049969. Accessed March 12, 2018. 547. Choi S, O’Neil SH, Joshi AA, et al. Anemia predicts lower white matter volume and cognitive performance in sickle and non-sickle cell anemia syndrome. Am J Hematol. 2019;94(10):1055-1065. doi:10.1002/ajh.25570 548. Inwald DP, Kirkham FJ, Peters MJ, et al. Platelet and leucocyte activation in childhood sickle cell disease: association with nocturnal hypoxaemia. Br J Haematol. 2000;111(2):474-481. http://www.ncbi.nlm.nih.gov/pubmed/11122087. Accessed May 23, 2018. 549. Huppert TJ, Hoge RD, Diamond SG, Franceschini MA, Boas DA. A temporal comparison of BOLD, ASL, and NIRS hemodynamic responses to motor stimuli in adult humans. Neuroimage. 2006;29(2):368-382. doi:10.1016/j.neuroimage.2005.08.065 550. Harris AD, Murphy K, Diaz CM, et al. Cerebral blood flow response to acute hypoxic hypoxia. NMR Biomed. 2013;26(12):1844-1852. doi:10.1002/nbm.3026 551. Poulin MJ, Liang PJ, Robbins PA. Dynamics of the cerebral blood flow response to step changes in end-tidal PCO2 and PO2 in humans. J Appl Physiol. 1996;81(3):1084-1095. doi:10.1152/jappl.1996.81.3.1084 552. El Hasnaoui-Saadani R, Pichon A, Marchant D, et al. Cerebral adaptations to chronic anemia in a model of erythropoietin-deficient mice exposed to hypoxia. Am J Physiol Integr Comp Physiol. 2009;296(3):R801-R811. doi:10.1152/ajpregu.00119.2008 553. Kosinski PD, Croal PL, Leung J, et al. The severity of anaemia depletes cerebrovascular dilatory reserve in children with sickle cell disease: a quantitative magnetic resonance imaging study. Br J Haematol. 2017;176(2):280-287. doi:10.1111/bjh.14424 554. Santiago T V, Edelman NH, Fishman AP. The effect of anemia on the ventilatory response to transient and steady-state hypoxia. J Clin Invest. 1975;55(2):410-418. doi:10.1172/JCI107945 555. Miao X, Choi S, Tamrazi B, et al. Increased brain iron deposition in patients with sickle cell disease: an MRI quantitative susceptibility mapping study. Blood. 2018;132(15):1618- 1621. doi:10.1182/BLOOD-2018-04-840322 286 556. Detterich JA, Kato RM, Rabai M, Meiselman HJ, Coates TD, Wood JC. Chronic transfusion therapy improves but does not normalize systemic and pulmonary vasculopathy in sickle cell disease. Blood. 2015;126(6):703-710. doi:10.1182/blood-2014-12-614370 557. Reckelhoff JF. Gender differences in the regulation of blood pressure. Hypertens (Dallas, Tex 1979). 2001;37(5):1199-1208. http://www.ncbi.nlm.nih.gov/pubmed/11358929. Accessed July 2, 2018. 558. Numaguchi Y, Haller JS, Humbert JR, et al. Cerebral blood flow mapping using stable Xenon-enhanced CT in sickle cell cerebrovascular disease. Neuroradiology. 1990;32(4):289-295. doi:10.1007/BF00593047 559. Sam K, Conklin J, Holmes KR, et al. Impaired dynamic cerebrovascular response to hypercapnia predicts development of white matter hyperintensities. 2016. doi:10.1016/j.nicl.2016.05.008 560. Binks AP, Cunningham VJ, Adams L, Banzett RB. Gray matter blood flow change is unevenly distributed during moderate isocapnic hypoxia in humans. J Appl Physiol. 2008;104(1):212-217. doi:10.1152/japplphysiol.00069.2007 561. Von Sarnowski B, Schminke U, Grittner U, et al. Posterior versus Anterior Circulation Stroke in Young Adults: A Comparative Study of Stroke Aetiologies and Risk Factors in Stroke among Young Fabry Patients (sifap1). Cerebrovasc Dis. 2017;43(3-4):152-160. doi:10.1159/000454840 562. Mentzer WC, Kan YW. Prospects for research in hematologic disorders: Sickle cell disease and thalassemia. J Am Med Assoc. 2001;285(5):640-642. doi:10.1001/jama.285.5.640 563. Chinawa JM, Ubesie AC, Chukwu BF, Ikefuna AN, Emodi IJ. Prevalence of hypoxemia among children with sickle cell anemia during steady state and crises: A cross-sectional study. Niger J Clin Pract. 2013;16(1):91-95. doi:10.4103/1119-3077.106774 564. Elsayh KI, Zahran AM, El-Abaseri TB, Mohamed AO, El-Metwally TH. Hypoxia biomarkers, oxidative stress, and circulating microparticles in pediatric patients with thalassemia in upper Egypt. Clin Appl Thromb. 2014;20(5):536-545. doi:10.1177/1076029612472552 565. Václavů L, Meynart BN, Mutsaerts HJMM, et al. Hemodynamic provocation with acetazolamide shows impaired cerebrovascular reserve in adults with sickle cell disease. Haematologica. 2019;104(4):690-699. doi:10.3324/haematol.2018.206094 566. Kennan RP, Suzuka SM, Nagel RL, Fabry ME. Decreased cerebral perfusion correlates with increased BOLD hyperoxia response in transgenic mouse models of sickle cell disease. Magn Reson Med. 2004;51(3):525-532. doi:10.1002/mrm.20014 567. Scheeren TWL, Schober P, Schwarte LA. Monitoring tissue oxygenation by near infrared spectroscopy (NIRS): background and current applications. J Clin Monit Comput. 2012;26(4):279-287. doi:10.1007/s10877-012-9348-y 568. Lu H, Xu F, Grgac K, Liu P, Qin Q, van Zijl P. Calibration and validation of TRUST MRI for the estimation of cerebral blood oxygenation. Magn Reson Med. 2012;67(1):42-49. doi:10.1002/mrm.22970 287 569. Bush A, Vu C, Choi S, et al. Calibration of T2 oximetry MRI for subjects with sickle cell disease. Magn Reson Med. 2021;86(2):1019-1028. doi:10.1002/mrm.28757 570. Reinhart K, Bloos F, Konig F, Bredle D, Hannemann L. Reversible decrease of oxygen consumption by hyperoxia. Chest. 1991;99(3):690-694. doi:10.1378/chest.99.3.690 571. Watson NA, Beards SC, Altaf N, Kassner A, Jackson A. The effect of hyperoxia on cerebral blood flow: A study in healthy volunteers using magnetic resonance phase-contrast angiography. Eur J Anaesthesiol. 2000;17(3):152-159. doi:10.1046/j.1365- 2346.2000.00640.x 572. Hlatky R, Valadka AB, Gopinath SP, Robertson CS. Brain Tissue Oxygen Tension Response to Induced Hyperoxia Reduced in Hypoperfused Brain. J Neurosurg. 2008;108(1):53-58. doi:10.3171/JNS/2008/108/01/0053 573. Howe FA, Robinson SP, McIntyre DJO, Stubbs M, Griffiths JR. Issues in flow and oxygenation dependent contrast (FLOOD) imaging of tumours. NMR Biomed. 2001;14(7- 8):497-506. doi:10.1002/nbm.716 574. Coloigner J, Vu C, Borzage M, et al. Transient Hypoxia Model Revealed Cerebrovascular Impairment in Anemia Using BOLD MRI and Near-Infrared Spectroscopy. J Magn Reson Imaging. July 2020:jmri.27210. doi:10.1002/jmri.27210 575. Chiarelli PA, Bulte DP, Wise R, Gallichan D, Jezzard P. A calibration method for quantitative BOLD fMRI based on hyperoxia. Neuroimage. 2007;37(3):808-820. doi:10.1016/J.NEUROIMAGE.2007.05.033 576. Floyd TF, Clark JM, Gelfand R, et al. Independent cerebral vasoconstrictive effects of hyperoxia and accompanying arterial hypocapnia at 1 ATA. J Appl Physiol. 2003;95(6):2453-2461. doi:10.1152/japplphysiol.00303.2003 577. Mouren S, Souktani R, Beaussier M, et al. Mechanisms of coronary vasoconstriction induced by high arterial oxygen tension. Am J Physiol Circ Physiol. 1997;272(1):H67-H75. doi:10.1152/ajpheart.1997.272.1.H67 578. Demchenko IT, Boso AE, Bennett PB, Whorton AR, Piantadosi CA. Hyperbaric oxygen reduces cerebral blood flow by inactivating nitric oxide. Nitric Oxide - Biol Chem. 2000;4(6):597-608. doi:10.1006/niox.2000.0313 579. Karetzky MS, Keighley JF, Mithoefer JC. The effect of oxygen administration on gas exchange and cardiopulmonary function in normal subjects. Respir Physiol. 1971;12(3):361-370. doi:10.1016/0034-5687(71)90076-4 580. Smit B, Smulders YM, Eringa EC, et al. Hyperoxia does not affect oxygen delivery in healthy volunteers while causing a decrease in sublingual perfusion. Microcirculation. 2018;25(2). doi:10.1111/micc.12433 581. Bodetoft S, Carlsson M, Arheden H, Ekelund U. Effects of oxygen inhalation on cardiac output, coronary blood flow and oxygen delivery in healthy individuals, assessed with MRI. Eur J Emerg Med. 2011;18(1):25-30. doi:10.1097/MEJ.0b013e32833a295e 288 582. Lauscher P, Lauscher S, Kertscho H, Habler O, Meier J. Hyperoxia Reversibly Alters Oxygen Consumption and Metabolism. Sci World J. 2012;2012. doi:10.1100/2012/410321 583. Xu F, Liu P, Pascual JM, Xiao G, Lu H. Effect of hypoxia and hyperoxia on cerebral blood flow, blood oxygenation, and oxidative metabolism. J Cereb Blood Flow Metab. 2012;32(10):1909-1918. doi:10.1038/jcbfm.2012.93 584. Chu DK, Kim LHY, Young PJ, et al. Mortality and morbidity in acutely ill adults treated with liberal versus conservative oxygen therapy (IOTA): a systematic review and meta-analysis. Lancet. 2018;391(10131):1693-1705. doi:10.1016/S0140-6736(18)30479-3 585. Iscoe S, Beasley R, Fisher JA. Supplementary oxygen for nonhypoxemic patients: O 2much of a good thing? Crit Care. 2011;15(3):305. doi:10.1186/cc10229 586. Kallet RH, Branson RD. Should oxygen therapy be tightly regulated to minimize hyperoxia in critically ill patients? Respir Care. 2016;61(6):801-817. doi:10.4187/respcare.04933 587. Jordan LC, Gindville MC, Scott AO, et al. Non-invasive imaging of oxygen extraction fraction in adults with sickle cell anaemia. Brain. 2016;139(3):738-750. doi:10.1093/brain/awv397 588. Bush AM, Borzage M, Choi S, Coates T, Wood JC. Elevated Cerebral Metabolic Oxygen Consumption in Sickle Cell Disease. Blood. 2014;124(21). 589. Robson MD, Gore JC, Constable RT. Measurement of the point spread function in MRI using constant time imaging. Magn Reson Med. 1997;38(5):733-740. doi:10.1002/mrm.1910380509 590. Qin Q, Grgac K, Van Zijl PCM. Determination of whole-brain oxygen extraction fractions by fast measurement of blood T2 in the jugular vein. Magn Reson Med. 2011;65(2):471-479. doi:10.1002/mrm.22556 591. Blum A, Yeganeh S, Peleg A, et al. Endothelial function in patients with sickle cell anemia during and after sickle cell crises. J Thromb Thrombolysis. 2005;19(2):83-86. doi:10.1007/s11239-005-1377-7 592. Figueiredo MS. The compound state: Hb S/beta-thalassemia. Rev Bras Hematol Hemoter. 2015;37(3):150-152. doi:10.1016/j.bjhh.2015.02.008 593. Serjeant GR, Sommereux AM, Stevenson M, Mason K, Serjeant BE. Comparison of Sickle Cell-β° Thalassaemia with Homozygous Sickle Cell Disease. Br J Haematol. 1979;41(1):83-93. doi:10.1111/j.1365-2141.1979.tb03684.x 594. Gonzalez-Redondo JM, Kutlar F, Kutlar A, et al. Hb S(c)-β+ -thalassaemia: Different mutations are associated with different levels of normal Hb A. Br J Haematol. 1988;70(1):85-89. doi:10.1111/j.1365-2141.1988.tb02438.x 595. Clive Ellory J. Haemoglobin C Promotes Distinct Membrane Properties in Heterozygous HbSC red Cells. EBioMedicine. 2015;2(11):1577. doi:10.1016/j.ebiom.2015.10.026 289 596. Karna B, Jha SK, Al Zaabi E. Hemoglobin C Disease. StatPearls Publishing; 2020. http://www.ncbi.nlm.nih.gov/pubmed/32644469. Accessed October 5, 2020. 597. Aliyu ZY, Tumblin AR, Kato GJ. Current therapy of sickle cell disease. Haematologica. 2006;91(1):7-10. http://www.ncbi.nlm.nih.gov/pubmed/16434364. Accessed September 16, 2019. 598. Cottam GL, Valentine KM, Yamaoka K, Waterman MR. The gelation of deoxyhemoglobin S in erythrocytes as detected by transverse water proton relaxation measurements. Arch Biochem Biophys. 1974;162(2):487-492. doi:10.1016/0003-9861(74)90208-2 599. Eaton WA, Hofrichter J. Sickle Cell Hemoglobin Polymerization. Adv Protein Chem. 1990;40(C):63-279. doi:10.1016/S0065-3233(08)60287-9 600. Stefanovic B, Pike GB. Human whole-blood relaxometry at 1.5T: Assessment of diffusion and exchange models. Magn Reson Med. 2004;52(4):716-723. doi:10.1002/mrm.20218 601. Gillis P, Petö S, Moiny F, Mispelter J, Cuenod C -A. Proton Transverse Nuclear Magnetic Relaxation in Oxidized Blood: a Numerical Approach. Magn Reson Med. 1995;33(1):93- 100. doi:10.1002/mrm.1910330114 602. CLARK MR. Mean Corpuscular Hemoglobin Concentration and Cell Deformability. Ann N Y Acad Sci. 1989;565(1):284-294. doi:10.1111/j.1749-6632.1989.tb24176.x 603. Jensen JH, Chandra R. NMR relaxation in tissues with weak magnetic inhomogeneities. Magn Reson Med. 2000;44(1):144-156. doi:10.1002/1522-2594(200007)44:1<144::AID- MRM21>3.0.CO;2-O 604. Brooks RA, Moiny F, Gillis P. On T2-shortening by weakly magnetized particles: The chemical exchange model. Magn Reson Med. 2001;45(6):1014-1020. doi:10.1002/mrm.1135 605. Wright GA, Hu BS, Macovski A. Estimating oxygen saturation of blood in vivo with MR imaging at 1.5 T. J Magn Reson Imaging. 1991;1(3):275-283. doi:10.1002/jmri.1880010303 606. Chen JJ, Pike GB. Human whole blood T2 relaxometry at 3 tesla. Magn Reson Med. 2009;61(2):249-254. doi:10.1002/mrm.21858 607. Fan AP, Bilgic B, Gagnon L, et al. Quantitative oxygenation venography from MRI phase. Magn Reson Med. 2014;72(1):149-159. doi:10.1002/mrm.24918 608. Hsieh CY, Cheng YCN, Xie H, Haacke EM, Neelavalli J. Susceptibility and size quantification of small human veins from an MRI method. Magn Reson Imaging. 2015;33(10):1191-1204. doi:10.1016/j.mri.2015.07.008 609. Raichle ME, Gusnard DA. Appraising the brain’s energy budget. Proc Natl Acad Sci U S A. 2002;99(16):10237-10239. doi:10.1073/pnas.172399499 290 610. Kaufman DP, Dhamoon AS. Physiology, Oxyhemoglobin Dissociation Curve. StatPearls Publishing; 2019. http://www.ncbi.nlm.nih.gov/pubmed/29762993. Accessed August 19, 2019. 611. Ibaraki M, Shinohara Y, Nakamura K, Miura S, Kinoshita F, Kinoshita T. Interindividual variations of cerebral blood flow, oxygen delivery, and metabolism in relation to hemoglobin concentration measured by positron emission tomography in humans. J Cereb Blood Flow Metab. 2010;30(7):1296-1305. doi:10.1038/jcbfm.2010.13 612. Henriksen OM, Kruuse C, Olesen J, et al. Sources of variability of resting cerebral blood flow in healthy subjects: A study using 133 Xe SPECT measurements. J Cereb Blood Flow Metab. 2013;33(5):787-792. doi:10.1038/jcbfm.2013.17 613. Choi S, Bush AM, Borzage MT, et al. Hemoglobin and mean platelet volume predicts diffuse T1-MRI white matter volume decrease in sickle cell disease patients. NeuroImage Clin. 2017;15:239-246. doi:10.1016/j.nicl.2017.04.023 614. Jonassaint CR, Varma VR, Chuang YF, et al. Lower hemoglobin is associated with poorer cognitive performance and smaller brain volume in older adults. J Am Geriatr Soc. 2014;62(5):972-973. doi:10.1111/jgs.12810 615. Inzitari M, Studenski S, Rosano C, et al. Anemia is associated with the progression of white matter disease in older adults with high blood pressure: The cardiovascular health study. J Am Geriatr Soc. 2008;56(10):1867-1872. doi:10.1111/j.1532-5415.2008.01950.x 616. Choi S, O’Neil SH, Joshi AA, et al. Exploring Anemia’s Impact on Brain Microstructure, Volume, Functional Connectivity, Iron and Cognitive Performance. Blood. 2019;134(Supplement_1):3553-3553. doi:10.1182/blood-2019-129544 617. Choi S, O’Neil SH, Joshi AA, et al. Anemia predicts lower white matter volume and cognitive performance in sickle and non-sickle cell anemia syndrome. Am J Hematol. 2019;94(10):1055-1065. doi:10.1002/ajh.25570 618. Juttukonda MR, Donahue MJ, Davis LT, et al. Preliminary evidence for cerebral capillary shunting in adults with sickle cell anemia. J Cereb Blood Flow Metab. 2019;39(6):1099- 1110. doi:10.1177/0271678X17746808 619. Helton KJ, Adams RJ, Kesler KL, et al. Magnetic resonance imaging/angiography and transcranial Doppler velocities in sickle cell anemia: Results from the SWiTCH trial. Blood. 2014;124(6):891-898. doi:10.1182/blood-2013-12-545186 620. Postma IR, De Groot JC, Aukes AM, Aarnoudse JG, Zeeman GG. Cerebral white matter lesions and perceived cognitive dysfunction: The role of pregnancy. In: American Journal of Obstetrics and Gynecology. Vol 211. Mosby Inc.; 2014:257.e1-257.e5. doi:10.1016/j.ajog.2014.02.031 621. Herold S, Brozovic M, Gibbs J, et al. Measurement of regional cerebral blood flow, blood volume and oxygen metabolism in patients with sickle cell disease using positron emission tomography. Stroke. 17(4):692-698. doi:10.1161/01.str.17.4.692 291 622. Frackowiak RSJ, Lenzi G-L, Jones T, Heather JD. Quantitative Measurement of Regional Cerebral Blood Flow and Oxygen Metabolism in Man Using 15O and Positron Emission Tomography. J Comput Assist Tomogr. 1980;4(6):727-736. doi:10.1097/00004728- 198012000-00001 623. Heyman A, Patterson JL, Duke TW. Cerebral circulation and metabolism in sickle cell and other chronic anemias, with observations on the effects of oxygen inhalation. J Clin Invest. 1952;31(9):824-828. doi:10.1172/JCI102668 624. SCHEINBERG P. Cerebral blood flow and metabolism in pernicious anemia. Blood. 1951;6(3):213-227. http://www.ncbi.nlm.nih.gov/pubmed/14811911. Accessed August 18, 2019. 625. Scheinberg P, Blackburn LI, Rich M, Saslaw M. Effects of vigorous physical exercise on cerebral circulation and metabolism. Am J Med. 1954;16(4):549-554. doi:10.1016/0002- 9343(54)90371-X 626. Stewart HJ, Crane NF, Deitrick JE. Studies of the circulation in pernicious anemia. J Clin Invest. 1937;16(3):431-441. doi:10.1172/JCI100868 627. Fazekas JF, Mchenry LC, Alman RW, Sullivan JF. Cerebral Hemodynamics During Brief Hyperventilation. Arch Neurol. 1961;4(2):132-138. doi:10.1001/archneur.1961.00450080014002 628. Mangold R, Sokoloff L, Conner E, Kleinerman J, Therman P-OG, Kety SS. THE EFFECTS OF SLEEP AND LACK OF SLEEP ON THE CEREBRAL CIRCULATION AND METABOLISM OF NORMAL YOUNG MEN 1. J Clin Invest. 1955;34(7 Pt 1):1092-1100. doi:10.1172/JCI103158 629. KETY SS, SCHMIDT CF. The effects of active and passive hyperventilation on cerebral blood flow, cerebral oxygen consumption, cardiac output, and blood pressure of normal young men. J Clin Invest. 1946;25(5):107-119. doi:10.1172/JCI101680 630. Borzage MT, Bush AM, Choi S, et al. Predictors of cerebral blood flow in patients with and without anemia. J Appl Physiol. 2016;120(8):976-981. doi:10.1152/japplphysiol.00994.2015 631. Quinn CT. Silent cerebral infarction: Supply and demand. Blood. 2018;132(16):1632-1634. doi:10.1182/blood-2018-08-868075 632. Kirkman MA, Smith M. Brain Oxygenation Monitoring. Anesthesiol Clin. 2016;34(3):537- 556. doi:10.1016/j.anclin.2016.04.007 633. Sohmer PR, Dawson R Ben, Huggins C. The significance of 2, 3-DPG in red blood cell transfusions. Crit Rev Clin Lab Sci. 1979;11(2):107-174. doi:10.3109/10408367909105855 634. Meier ER. Treatment Options for Sickle Cell Disease. Pediatr Clin North Am. 2018;65(3):427-443. doi:10.1016/j.pcl.2018.01.005 292 635. Ibaraki M, Miura S, Shimosegawa E, et al. Quantification of cerebral blood flow and oxygen metabolism with 3-dimensional PET and 15O: Validation by comparison with 2-dimensional PET. J Nucl Med. 2008;49(1):50-59. doi:10.2967/jnumed.107.044008 636. Pantano P, Baron JC, Lebrun-Grandié P, Duquesnoy N, Bousser MG, Comar D. Regional cerebral blood flow and oxygen consumption in human aging. Stroke. 1984;15(4):635-641. doi:10.1161/01.STR.15.4.635 637. Hino A, Ueda S, Mizukawa N, Imahori Y, Tenjin H. Effect of hemodilution on cerebral hemodynamics and oxygen metabolism. Stroke. 1992;23(3):423-426. doi:10.1161/01.STR.23.3.423 638. Powers WJ. Cerebral hemodynamics in ischemic cerebrovascular disease. Ann Neurol. 1991;29(3):231-240. doi:10.1002/ana.410290302 639. Kei T, Mistry N, Tsui AKY, et al. Experimental assessment of oxygen homeostasis during acute hemodilution: the integrated role of hemoglobin concentration and blood pressure. Intensive Care Med Exp. 2017;5(1):12. doi:10.1186/s40635-017-0125-6 640. Kocsi S, Demeter G, Érces D, Kaszaki J, Molnár Z. Central Venous-To-Arterial CO2-Gap May Increase in Severe Isovolemic Anemia. Guo Y, ed. PLoS One. 2014;9(8):e105148. doi:10.1371/journal.pone.0105148 641. Maruyama M, Shimoji K, Ichikawa T, Hashiba M, Naito E. The effects of extreme hemodilutions on the autoregulation of cerebral blood flow, electroencephalogram and cerebral metabolic rate of oxygen in the dog. Stroke. 16(4):675-679. doi:10.1161/01.str.16.4.675 642. Belhassen L, Pelle G, Sediame S, et al. Endothelial dysfunction in patients with sickle cell disease is related to selective impairment of shear stress-mediated vasodilation. Blood. 2001;97(6):1584-1589. doi:10.1182/blood.V97.6.1584 643. Zinchuk V V., Pronko TP, Lis MA. Blood oxygen transport and endothelial dysfunction in patients with arterial hypertension. Clin Physiol Funct Imaging. 2004;24(4):205-211. doi:10.1111/j.1475-097X.2004.00549.x 644. Richardson SL, Hulikova A, Proven M, et al. Single-cell O2 exchange imaging shows that cytoplasmic diffusion is a dominant barrier to efficient gas transport in red blood cells. Proc Natl Acad Sci U S A. 2020;117(18):10067-10078. doi:10.1073/pnas.1916641117 645. Lee EJ, Hung YC, Lee MY. Anemic hypoxia in moderate intracerebral hemorrhage: the alterations of cerebral hemodynamics and brain metabolism. J Neurol Sci. 1999;164(2):117-123. doi:10.1016/s0022-510x(99)00068-4 646. Lee EJ, Hung YC. Marked anemic hypoxia deteriorates cerebral hemodynamics and brain metabolism during massive intracerebral hemorrhage. J Neurol Sci. 2001;190(1-2):3-10. doi:10.1016/s0022-510x(01)00567-6 647. Bernstein D, Teitel DF, Rudolph AM. Chronic anemia in the newborn lamb: Cardiovascular adaptations and comparison to chronic hypoxemia. Pediatr Res. 1988;23(6):621-627. doi:10.1203/00006450-198806000-00019 293 648. Brown MM, Wade JPH, Marshall J. Fundamental importance of arterial oxygen content in the regulation of cerebral blood flow in man. Brain. 1985;108(1):81-93. doi:10.1093/brain/108.1.81 649. An H, Lin W. Impact of intravascular signal on quantitative measures of cerebral oxygen extraction and blood volume under normo- and hypercapnic conditions using an asymmetric spin echo approach. Magn Reson Med. 2003;50(4):708-716. doi:10.1002/mrm.10576 650. Klein KU, Glaser M, Reisch R, Tresch A, Werner C, Engelhard K. The effects of arterial carbon dioxide partial pressure and sevoflurane on capillary venous cerebral blood flow and oxygen saturation during craniotomy. Anesth Analg. 2009;109(1):199-204. doi:10.1213/ane.0b013e3181a800e5 651. Di Liberto G, Kiger L, Marden MC, et al. Dense red blood cell and oxygen desaturation in sickle-cell disease. Am J Hematol. 2016;91(10):1008-1013. doi:10.1002/ajh.24467 652. Wood JC. Red Cells, Iron, and Erythropoiesis: Brain O2 reserve in sickle cell disease. Blood. 2019;133(22):2356. doi:10.1182/BLOOD-2019-04-901124 653. Vaclavu L, Petersen E, Mutsaerts H, et al. Cerebral oxygen metabolism measurements with MRI in adults with sickle cell disease. HemaSphere. 2019;3:324-325. doi:10.1097/01.HS9.0000561252.50212.c6 654. Croal PL, Leung J, Phillips CL, Serafin MG, Kassner A. Quantification of pathophysiological alterations in venous oxygen saturation: A comparison of global MR susceptometry techniques. Magn Reson Imaging. 2019;58:18-23. doi:10.1016/j.mri.2019.01.008 655. Metarugcheep P, Chanyawattiwongs S, Srisubat K, Pootrakul P. Clinical silent cerebral infarct (SCI) in patients with thalassemia diseases assessed by Magnetic Resonance Imaging (MRI). J Med Assoc Thail. 2008;91(6):889-894. 656. Colombatti R, Lucchetta M, Montanaro M, et al. Cognition and the default mode network in children with sickle cell disease: A resting state functional MRI study. PLoS One. 2016;11(6). doi:10.1371/journal.pone.0157090 657. Hijmans CT, Fijnvandraat K, Grootenhuis MA, et al. Neurocognitive deficits in children with sickle cell disease: A comprehensive profile. Pediatr Blood Cancer. 2011;56(5):783-788. doi:10.1002/pbc.22879 658. Armstrong FD. Thalassemia and learning: Neurocognitive functioning in children. In: Annals of the New York Academy of Sciences. Vol 1054. New York Academy of Sciences; 2005:283-289. doi:10.1196/annals.1345.036 659. Chai Y, Ji C, Coloigner J, et al. Tract-specific analysis and neurocognitive functioning in sickle cell patients without history of overt stroke. Brain Behav. 2021. doi:10.1002/brb3.1978 660. Bernaudin F, Verlhac S, Arnaud C, et al. Chronic acute anemia and extracranial internal carotid stenosis are risk factors for silent cerebral infarcts in sickle cell anemia. Blood. 2015;125(10):1653-1661. doi:10.1182/blood-2014-09-599852 294 661. Gu C, Jun JC. Does Hypoxia Decrease the Metabolic Rate? Front Endocrinol (Lausanne). 2018;9:668. doi:10.3389/fendo.2018.00668 662. DiPasquale DM, Kolkhorst FW, Buono MJ. Acute Normobaric Hypoxia Reduces Body Temperature in Humans. High Alt Med Biol. 2015;16(1):61-66. doi:10.1089/ham.2014.1098 663. Hochachka PW, Clark CM, Brown WD, et al. The Brain at High Altitude: Hypometabolism as a Defense against Chronic Hypoxia? J Cereb Blood Flow Metab. 1994;14(4):671-679. doi:10.1038/jcbfm.1994.84 664. Bush AM, Kato RM, Coates TD, Wood JC. Cerebral Tissue Transit Time in Patients with Sickle Cell Anemia. Blood. 2015;126(23). 665. Nahavandi M, Millis RM, Tavakkoli F, et al. Arterialization of peripheral venous blood in sickle cell disease. J Natl Med Assoc. 2002;94(5):320-326. http://www.ncbi.nlm.nih.gov/pubmed/12069211. Accessed April 1, 2019. 666. Buxton RB, Frank LR. A Model for the Coupling between Cerebral Blood Flow and Oxygen Metabolism during Neural Stimulation. J Cereb Blood Flow Metab. 1997;17(1):64-72. doi:10.1097/00004647-199701000-00009 667. Angleys H, Østergaard L, Jespersen SN. The effects of capillary transit time heterogeneity (CTH) on brain oxygenation. J Cereb Blood Flow Metab. 2015;35(10):806-817. doi:10.1038/jcbfm.2014.254 668. Cheung ATW, Chen PCY, Larkin EC, et al. Microvascular abnormalities in sickle cell disease: a computer-assisted intravital microscopy study. Blood. 2002;99(11):3999-4005. doi:10.1182/blood.v99.11.3999 669. Stenmark KR, Fagan KA, Frid MG. Hypoxia-Induced Pulmonary Vascular Remodeling. Circ Res. 2006;99(7):675-691. doi:10.1161/01.RES.0000243584.45145.3f 670. Metivier F, Marchais SJ, Guerin AP, Pannier B, London GM. Pathophysiology of anaemia: focus on the heart and blood vessels. Nephrol Dial Transplant. 2000;15(suppl_3):14-18. doi:10.1093/oxfordjournals.ndt.a027970 671. Kim DH, Grodstein F, Newman AB, et al. Microvascular and Macrovascular Abnormalities and Cognitive and Physical Function in Older Adults: Cardiovascular Health Study. J Am Geriatr Soc. 2015;63(9):1886-1893. doi:10.1111/jgs.13594 672. Hahalis G, Kremastinos DT, Terzis G, et al. Global vasomotor dysfunction and accelerated vascular aging in β-thalassemia major. Atherosclerosis. 2008;198(2):448-457. doi:10.1016/j.atherosclerosis.2007.09.030 673. Guilliams KP, Fields ME, Ragan DK, et al. Red cell exchange transfusions lower cerebral blood flow and oxygen extraction fraction in pediatric sickle cell anemia. Blood. 2018;131(9):1012-1021. http://www.bloodjournal.org/content/bloodjournal/131/9/1012.full.pdf?sso-checked=true. Accessed March 26, 2018. 295 674. Griffiths DP, Cummins BH, Greenbaum R, et al. Cerebral blood flow and metabolism during hypotension induced with sodium nitroprusside. Br J Anaesth. 1974;46(9):671-679. doi:10.1093/bja/46.9.671 675. HAFKENSCHIEL JH, CRUMPTON CW, SHENKIN HA, et al. The effects of twenty degree head-up tilt upon the cerebral circulation of patients with arterial hypertension before and after sympathectomy. J Clin Invest. 1951;30(8):793-798. doi:10.1172/JCI102494 676. HARMEL MH, HAFKENSCHIEL JH. The effect of bilateral stellate ganglion block on the cerebral circulation in normotensive and hypertensive patients. J Clin Invest. 1949;28(3):415-418. doi:10.1172/JCI102085 677. HEYMAN A, PATTERSON JL, JONES RW. Cerebral circulation and metabolism in uremia. Circulation. 1951;3(4):558-563. doi:10.1161/01.CIR.3.4.558 678. Kennedy C. The Cerebral Metabolic Rate in Mentally Retarded Children. Arch Neurol. 1967;16(1):55-58. doi:10.1001/archneur.1967.00470190059007 679. KETY SS, HAFKENSCHIEL JH. The blood flow, vascular resistance, and oxygen consumption of the brain in essential hypertension. J Clin Invest. 1948;27(4):511-514. http://www.ncbi.nlm.nih.gov/pubmed/18873801. Accessed April 20, 2020. 680. KETY SS, SKENKIN HA, SCHMIDT CF. The effects of increased intracranial pressure on cerebral circulatory functions in man. J Clin Invest. 1948;27(4):493-499. doi:10.1172/JCI101996 681. KETY SS, WOODFORD RB. Cerebral blood flow and metabolism in schizophrenia; the effects of barbiturate semi-narcosis, insulin coma and electroshock. Am J Psychiatry. 1948;104(12):765-770. doi:10.1176/ajp.104.12.765 682. KETY SS, POLIS BD. The blood flow and oxygen consumption of the human brain in diabetic acidosis and coma. J Clin Invest. 1948;27(4):500-510. doi:10.1172/JCI101997 683. Kuwabara Y, Sasaki M, Hirakata H, et al. Cerebral blood flow and vasodilatory capacity in anemia secondary to chronic renal failure. Kidney Int. 2002;61:564-569. doi:10.1046/j.1523-1755.2002.00142.x 684. SCHEINBERG P, STEAD EA, BRANNON ES, WARREN J V. Correlative observations on cerebral metabolism and cardiac output in myxedema. J Clin Invest. 1950;29(9):1139- 1146. doi:10.1172/JCI102351 685. SENSENBACH W, MADISON L, EISENBERG S, OCHS L. The cerebral circulation and metabolism in hyperthyroidism and myxedema. J Clin Invest. 1954;33(11):1434-1440. doi:10.1172/JCI103021 686. SHENKIN HA, CABIESES F, VAN DEN NOORDT G, SAYERS P, COPPERMAN R. The hemodynamic effect of unilateral carotid ligation on the cerebral. J Neurosurg. 1951;8(1):38-45. doi:10.3171/jns.1951.8.1.0038 296 687. SHENKIN HA, HAFKENSCHIEL JH, KETY SS. Effects of sympathectomy on the cerebral circulation of hypertensive patients. Arch Surg. 1950;61(2):319-324. doi:10.1001/archsurg.1950.01250020323014 688. SHENKIN HA. Effects of various drugs upon cerebral circulation and metabolism of. J Appl Physiol. 1951;3(8):465-471. doi:10.1152/jappl.1951.3.8.465 689. SOKOLOFF L, WECHSLER RL, MANGOLD R, BALLS K, KETY SS. Cerebral blood flow and oxygen consumption in hyperthyroidism before and after treatment. J Clin Invest. 1953;32(3):202-208. doi:10.1172/JCI102728 690. WECHSLER RL, DRIPPS RD, KETY SS. Blood flow and oxygen consumption of the human brain during anesthesia. Anesthesiology. 1951;12(3):308-314. doi:10.1097/00000542-195105000-00006 691. WECHSLER RL, KLEISS LM, KETY SS. The effects of intravenously administered aminophylline on cerebral circulation and metabolism in man. J Clin Invest. 1950;29(1):28- 30. doi:10.1172/JCI102230 692. BENTINCK RC, GORDAN GS, ADAMS JE, ARNSTEIN LH, LEAKE TB. Effect of desoxycorticosterone glucoside upon cerebral blood flow and. J Clin Invest. 1951;30(2):200-205. doi:10.1172/JCI102432 693. HIMWICH WA, HOMBURGER E. Brain metabolism in man; unanesthetized and in pentothal narcosis. Am J Psychiatry. 1947;103(5):689-696. doi:10.1176/ajp.103.5.689 694. PATTERSON JL, HEYMAN A, NICHOLS FT. Cerebral blood flow and oxygen consumption in neurosyphilis. J Clin Invest. 1950;29(10):1327-1334. doi:10.1172/JCI102369 695. SCHEINBERG P. Cerebral circulation in heart failure. Am J Med. 1950;8(2):148-152. doi:10.1016/0002-9343(50)90355-x 696. SCHEINBERG P. Cerebral circulation and metabolism in hyperthyroidism. J Clin Invest. 1950;29(8):1010-1013. doi:10.1172/JCI102331 697. SCHEINBERG P. The effect of nicotinic acid on the cerebral circulation, with observations on extra-cerebral contamination of cerebral venous blood in the nitrous oxide procedure for cerebral blood flow. Circulation. 1950;1(5):1148-1154. doi:10.1161/01.CIR.1.5.1148 698. Scheinberg P, Stead EA. The cerebral blood flow in male subjects as measured by the nitrous oxide technique. Normal values for blood flow, oxygen utilization, glucose utilization, and peripheral resistance, with observations on the effect of tilting and anxiety. J Clin Invest. 1949;28(5 Pt 2):1163-1171. doi:10.1172/JCI102150 699. Scheinberg P. Cerebral blood flow in vascular disease of the brain. With observations on the effects of stellate ganglion block. Am J Med. 1950;8(2):139-147. doi:10.1016/0002- 9343(50)90354-8 700. SHENKIN HA, SPITZ EB. Physiologic studies of arteriovenous anomalies of the brain. J Neurosurg. 1948;5(2):165-172. doi:10.3171/jns.1948.5.2.0165 297 701. Sedaghat S, Vernooij MW, Loehrer E, et al. Kidney function and cerebral blood flow: The Rotterdam study. J Am Soc Nephrol. 2016;27(3):715-721. doi:10.1681/ASN.2014111118 702. Ghoshal S, Freedman BI. Mechanisms of Stroke in Patients with Chronic Kidney Disease. Am J Nephrol. 2019;50(4):229-239. doi:10.1159/000502446 703. Oshima T, Karasawa F, Satoh T. Effects of propofol on cerebral blood flow and the metabolic rate of oxygen in humans. Acta Anaesthesiol Scand. 2002;46(7):831-835. doi:10.1034/j.1399-6576.2002.460713.x 704. Jog MA, Yan L, Kilroy E, et al. Developmental trajectories of cerebral blood flow and oxidative metabolism at baseline and during working memory tasks. Neuroimage. 2016;134:587-596. doi:10.1016/j.neuroimage.2016.04.035 705. Kuroda S, Kamiyama H, Abe H, et al. Cerebral Blood Flow in Children with Spontaneous Occlusion of the Circle of Willis (Moyamoya Disease): Comparison with Healthy Children and Evaluation of Annual Changes. Neurol Med Chir (Tokyo). 1993;33(7):434-438. doi:10.2176/nmc.33.434 706. Mullur R, Liu YY, Brent GA. Thyroid hormone regulation of metabolism. Physiol Rev. 2014;94(2):355-382. doi:10.1152/physrev.00030.2013 707. Busija DW, Leffler CW, Pourcyrous M. Hyperthermia increases cerebral metabolic rate and blood flow in neonatal pigs. Am J Physiol - Hear Circ Physiol. 1988;255(2). doi:10.1152/ajpheart.1988.255.2.h343

Abstract (if available)

Abstract The goal of this thesis is to apply magnetic resonance imaging (MRI) to investigate the risks of silent cerebral infarcts (SCI), which are commonly seen in neurovascular diseases and normal aging. Even though the etiology of SCI remains unclear, this work proposes a two-hit hypothesis in which acute-on-chronic imbalance in oxygen supply and demand results in microvascular ischemia and stroke. Since a history of SCI increases the risk of a subsequent overt stroke, patient monitoring and prevention therapies are important to alleviate the risk of cerebrovascular accidents in vulnerable populations.

At the core of stroke risk monitoring and management is regular screening of oxygen delivery and utilization in the brain. Therefore, the first part of this thesis introduces non-invasive MRI techniques to measure cerebral perfusion, oxygen extraction, and metabolic rate. The second part aims to apply these techniques in a cohort of sickle cell disease and chronic anemia patients to explore clinical biomarkers of their increased vulnerability to SCI development.

To quantify regional cerebral perfusion, Chapter 2 proposes a non-invasive Deoxygenation-based Dynamic Susceptibility Contrast (dDSC) technique that leverages oxygen respiratory challenges as a source of contrast alternative to traditional gadolinium contrast in perfusion-weighted imaging. Chapters 3 and 4 extend this concept to a variety of oxygen and carbon-dioxide respiratory challenges using bolus and non-bolus paradigms. The findings from this study show regional agreement between this novel deoxygenation-based method against conventional perfusion-weighted techniques, thus demonstrating its potential application in patients in whom gadolinium contrast is contraindicated.

Coupled with the capability to acquire quantitative perfusion, respiratory challenges can be used to investigate hemodynamic impairments in different regions in the brain. Chapters 5 and 6 applied transient hypoxia challenge and prolonged hyperoxia challenge respectively in a cohort of healthy volunteers, sickle cell disease, and non-sickle chronic anemia patients. Even though the hyperoxia challenge failed to identify areas of flow limitation in anemic subjects, the striking variations in dynamic response to hypoxia between subjects with and without SCI suggest that these infarcts are just an iceberg phenomenon with respect to microvascular damage. Further investigation into microvascular damage and its colocalization with capillary perfusion and transit time heterogeneity can shed light on the increased SCI risk in these patient populations.

In addition to the investigation of tissue perfusion, oxygen extraction fraction can be measured using T2-oximetry MRI techniques. However, these techniques require appropriate and disease-specific calibration equation to convert raw relaxivity values to venous saturation. Chapter 7 addressed this need by establishing a sickle-specific calibration by pooling in vitro datasets from two independent studies. This combined calibration was based on a larger range of hematocrit and yielded unbiased estimates compared to blood-gas oximetry results. Additionally, this work also demonstrated the need to correct for transfusion in hyper-transfused sickle cell disease patients and proposes a correction method based on patient-specific hemoglobin S concentration.
Once appropriate calibration equations have been established for sickle cell disease and non-sickle chronic anemia patients, Chapter 8 investigated cerebral oxygen extraction (OEF) and metabolic rate (CMRO2) in this population. This work demonstrated decreased OEF and CMRO2 in anemia subjects, proportional to the degree of their anemia severity. To address the potential for bias due to our calibration equations, this chapter performed a meta-analysis of historical CMRO2 publications and showed striking concordance with these historical datasets from patients having broad etiologies for their anemia. Additionally, this reduced cerebral metabolism is consistent with emerging data demonstrating increased non-nutritive flow, or physiological shunting, in chronically anemic patients.

Overall, this thesis proposes novel non-invasive methods to measure cerebral perfusion and oxygenation as well as applies these techniques in a cohort of patients with hematologic diseases to explore their stroke risk. Further work is necessary on both the technical and the physiological fronts, including additional validation in other pathologies as well as routine hemodynamic screening as part of a stroke prevention program in vulnerable populations.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

$Quantitative MRI for the measurement of cerebral oxygen extraction fraction in sickle cell disease$

PDF

Quantitative MRI for the measurement of cerebral oxygen extraction fraction in sickle cell disease

PDF

Susceptibility-weighted MRI for the evaluation of brain oxygenation and brain iron in sickle cell disease

PDF

Assessing cerebral oxygen supply and demand in sickle cell anemia using MRI

PDF

Cerebrovascular disease of white matter in patients with chronic anemia syndrome

PDF

Estimation of cognitive brain activity in sickle cell disease using functional near-infrared spectroscopy and dynamic systems modeling

PDF

Characterization of lenticulostriate arteries using high-resolution black blood MRI as an early imaging biomarker for vascular cognitive impairment and dementia

PDF

Estimating liver iron non-invasively with high-field MRI

PDF

Cerebral oxygen saturation and cerebrovascular instability in newborn infants with congenital heart disease compared to healthy controls

PDF

Improving sensitivity and spatial coverage of myocardial arterial spin labeling

PDF

Photoplethysmogram-based biomarker for assessing risk of vaso-occlusive crisis in sickle cell disease: machine learning approaches

PDF

Improving the sensitivity and spatial coverage of cardiac arterial spin labeling for assessment of coronary artery disease

PDF

Mapping water exchange rate across the blood-brain barrier

PDF

Calibration of iron-mediated MRI relaxation by Monte Carlo modeling

PDF

Oxygen therapy for the treatment of retinal ischemia

PDF

New theory and methods for accelerated MRI reconstruction

PDF

High-dimensional magnetic resonance imaging of microstructure

PDF

Dynamic cardiovascular magnetic resonance imaging for improved assessment of ischemic heart disease

PDF

Shift-invariant autoregressive reconstruction for MRI

PDF

Diffusion MRI of the human brain: signal modeling and quantitative analysis

PDF

Radiofrequency pulse performance for myocardial ASL

Asset Metadata

Creator Vu, Chau (author)

Core Title Quantitative MRI for oxygenation imaging in cerebrovascular diseases

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Degree Conferral Date 2022-12

Publication Date 10/17/2022

Defense Date 08/03/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag anemia,cerebrovascular,magnetic resonance imaging,OAI-PMH Harvest,oxygenation,stroke

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Wood, John (committee chair), Coates, Thomas (committee member), Nayak, Krishna (committee member)

Creator Email chauvu.q@gmail.com,chauvu@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112124638

Unique identifier UC112124638

Identifier etd-VuChau-11271.pdf (filename)

Legacy Identifier etd-VuChau-11271

Document Type Thesis

Format theses (aat)

Rights Vu, Chau

Internet Media Type application/pdf

Type texts

Source 20221019-usctheses-batch-987 (), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu