University of Southern California Dissertations and Theses

Longitudinal neurocognitive profiles of empirically-derived Alzheimer’s disease variants

(USC Thesis Other)

Longitudinal neurocognitive profiles of empirically-derived Alzheimer’s disease variants

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: Neurocognitive profiles of AD variants Longitudinal neurocognitive profiles of empirically-derived Alzheimer’s disease variants Anna Emilia Blanken Faculty Advisor: Daniel A. Nation, Ph.D University of Southern California Degree being conferred: Masters in Psychology Degree conferral date: December 2017 2 Neurocognitive profiles of AD variants TABLE OF CONTENTS ABSTRACT .................................................................................................................................. 3 BACKGROUND .......................................................................................................................... 5 STUDY AIMS AND HYPOTHESES .......................................................................................... 18 MATERIALS AND METHODS .................................................................................................. 21 RESULTS ..................................................................................................................................... 28 DISCUSSION ............................................................................................................................... 38 TABLES AND FIGURES ............................................................................................................ 47 REFERENCES ............................................................................................................................. 72 3 Neurocognitive profiles of AD variants Abstract Studies have identified Alzheimer’s disease (AD) variants characterized by distinct clinical and pathological features using a variety of grouping methods. A complete neurocognitive profile is important for Alzheimer’s disease diagnosis and prognosis. The present study sought to identify AD variants through cluster analysis of both in vivo AD biomarkers and neuropsychological measures, and to compare variants on longitudinal cognitive decline and brain atrophy. Alzheimer’s Disease Neuroimaging Initiative participants (N=976, non-demented subsample) were subjected to hierarchical cluster analysis using baseline cognition (Boston Naming Test, Rey’s Auditory Verbal Learning Test, the Logical Memory, Categorical Fluency, Trail Making Test A & B) and cerebrospinal fluid (amyloid β 1-42 and phosphorylated-tau) biomarkers. Analysis of covariance (ANCOVA) was used to examine group differences in cognitive performance (raw and age-normed) and biomarker status. Longitudinal (baseline, 1, 2 year) cognitive decline (n=747) and brain atrophy (hippocampal, n=497; ventricular, n=566) were compared with generalized linear mixed models (GLMM), controlling for age, gender, education, and apolipoprotein E4 status, and LSD post-hoc tests. Subgroups included one group (No AD; n=653) with normal cognition and biomarker status. Second, a typical AD group (TAD; n=191) emerged showing greater memory decline, relatively preserved language and executive function, and biomarker profile consistent with AD. Finally, an atypical AD group (AAD; n=132) was observed with impairment in language, memory, and executive functioning, and intermediate biomarker elevation. Over two years of follow-up AAD continued to show greater impairment in language and executive ability relative 4 Neurocognitive profiles of AD variants to TAD, whereas TAD had greater memory impairment relative to AAD. The TAD and AAD groups did not differ on ratings of global impairment (MMSE, CDR, ADAS-cog), yet a higher percentage of TAD was diagnosed with AD than AAD and No AD groups over the two year interval (TAD>AAD>No AD). Consistent with their cognitive profiles, the TAD group exhibited greater medial temporal atrophy than the atypical AD group, whereas the atypical group demonstrated more widespread cortical atrophy than the TAD group. The combination of neuropsychological test scores and cerebrospinal fluid biomarkers identified unique subgroups of older patients at risk for Alzheimer’s dementia. Increased understanding of patient subgroups may improve diagnostic and predictive accuracy, and help target preventative efforts. 5 Neurocognitive profiles of AD variants Background Alzheimer’s disease (AD), a complex neurodegenerative disorder, is characterized by memory decline, β-amyloid (Aβ) and tau deposition, and neuronal loss, and is the most common cause of dementia in elderly adult populations ("2015 Alzheimer's disease facts and figures," 2015). Biologists and pathologists define AD by an abnormal deposition of proteins , yet psychologists and neurologists rely on the presentation of clinical symptoms for diagnosis However, AD does not always follow classical neuropathologic staging (G. P. Morris et al., 2014; J. C. Morris et al., 2010; Nelson et al., 2012), and similarly, cognitive and functional deficits apparent in individuals diagnosed with AD can be heterogeneous (Vardy et al., 2013), making the characterization of a classical neurocognitive profile of the disease and its progression quite difficult. A better-quality understanding of the underlying mechanisms of pathophysiological change associated cognitive and behavioral symptoms of AD, as well as contributions from genetic and environmental risk factors, is of the upmost importance in light of the rapidly expanding elderly population. Improvement of criteria for early diagnosis and characterization of disease staging would have several significant clinical implications. Diagnostically, this information would be meaningful for identifying disease mechanisms at work and the expected disease progression. A major challenge in treating AD is that individuals do not become symptomatic until well after irreversible physiological changes have already taken place (Vos et al., 2013). Identification of early disease signs would greatly improve research aimed at the development of effective disease-modifying interventions, and have significant impact on the lives and prognosis of individuals affected by the disease. Another obstacle hindering the accuracy of diagnostic criteria is that the current gold standard of AD diagnosis remains post-mortem examination of brain 6 Neurocognitive profiles of AD variants tissue. That is, there is no single disease marker measurable in vivo that has a pathognomonic association with disease risk. Given these challenges, a complete biological, genetic, and neuropsychological profile may help clinicians reach diagnoses with a clearer picture of treatment and outcome. Characterization of age-related changes, and duration of symptoms, over time can greatly improve our ability to manage and treat AD. The prototypical AD case presents with progressive change in memory function, either alone or in the presence of other forms of cognitive impairment. Subjective experience of memory decline is common in aging, even among healthy older adults, but in AD patients objective memory impairment has a substantial detrimental effect on activities of daily functioning. A classic feature of AD-related decline is an episodic memory deficit in which the individual is impaired on recall and unable to improve performance through cuing (Dubois et al., 2014). This change often represents one of the earliest recognizable symptoms of the disease, and may be followed or paired with impairment in other cognitive domains such as language and executive functioning. While memory impairment is central to the typical AD profile, these other cognitive deficits may or may not be present in an individual case, and can range in severity (Weintraub, Wicklund, & Salmon, 2012). Early neuropathologic changes in AD include tauopathy (i.e. progressive development of neurofibrillary tangles) which relates to memory impairment and cognitive decline (Braak, 2011). At later disease stages, cortical atrophy in the bilateral frontal and parietal lobes, as well as further deterioration of the medial temporal lobes, can result in non-memory impairment (e.g. executive functioning, language, praxis) (Frisoni & Jack, 2011). There is strong empirical support for reduced clearance of Aβ in the brain, which leads to the development of Aβ plaques, one of the earliest observable neuropathologic stages or perhaps 7 Neurocognitive profiles of AD variants even the initiating step, of AD (C. R. Jack, Jr. et al., 2010). The hypothesis suggests that the initial deposition of Aβ in the brain is ultimately followed by a cascade of neuropathologic changes (e.g., tau aggregation, synaptic dysfunction, neuronal loss, cortical shrinkage) that characterize AD (Hardy & Higgins, 1992). Therefore, formation of Aβ plaques, neurofibrillary tangles (NFT) comprised of hyperphosphorylated tau protein, and neurodegeneration are the most widely studied biomarkers of AD pathology to date. CSF biospecimens provide us with information about levels of tau and Aβ proteins which directly associate with the levels of these proteins in the brain, and these relationships are supported by post-mortem studies. CSF Aβ1-42 levels are inversely associated with Aβ burden in the brain (Landau et al., 2013). Increased CSF tau levels are associated with accumulation of tau in the neurons and subsequent degeneration, and increased CSF levels of P-tau are also considered a marker of tangle pathology and demonstrate a direct relationship (Andreasen et al., 1999). Changes in CSF Aβ1-42 and tau occur early in disease progression and have been observed in individuals with MCI (Andreasen et al., 1999; Skillback et al., 2015), as well as linked to cognitive outcomes (Wirth et al., 2013). In a study by Nation et al. (2012), inverse correlations were found between pulse pressure and Aβ in cerebrospinal fluid (CSF) in cognitively normal older adults, providing further evidence for vascular contributions to cerebral Aβ accumulation (Nation et al., 2012). Intermediate amounts of AD-related pathology, Aβ and tau, have been seen in individuals with mild cognitive impairment (MCI) (Bennett, Schneider, Bienias, Evans, & Wilson, 2005), but their role here is still not well understood. Both of these CSF measures correlate with MRI biomarkers, with stronger associations seen among ApoE4 carriers (Apostolova et al., 2010). 8 Neurocognitive profiles of AD variants The apolipoprotein gene (APOE) is the strongest and most replicated genetic risk factor in AD, such that presence of the E4 allele (ApoE4) is associated with earlier age of onset and increased disease prevalence (J. Kim, Basak, & Holtzman, 2009; Lambert et al., 2013). APOE plays an important role in Aβ clearance (Deane et al., 2008; J. Kim et al., 2009), immune response, oxidative stress, and synaptic plasticity (Liu et al., 2015; Rosenthal & Kamboh, 2014). ApoE4 has been associated with decreased cerebrospinal fluid Aβ and increased cerebrospinal tau levels (Karch & Goate, 2015; Karch et al., 2012) due to cerebral Aβ retention and tau- mediated neurodegeneration, respectively (Johnson et al., 2013; J. Kim et al., 2009; J. C. Morris et al., 2010; Murphy et al., 2013; Reiman et al., 2009). Patients with AD also have significantly lowered levels of APOE protein in the plasma (S. Kim et al., 2011). More recently there has been a growing body of evidence showing that the complexity of AD cannot be effectively explained by the “amyloid cascade hypothesis”, which proposes a clear linear pathway beginning with Aβ deposition, and has been a central focus of AD research. Recent investigations have demonstrated that the presence of neuropathologic changes typically associated with AD, are not necessarily accompanied by cognitive impairment (R. C. Petersen et al., 2016), and cognitive decline consistent with AD symptoms is not always coupled with increased cortical Aβ burden (G. P. Morris et al., 2014). Recent studies have identified a clinical subgroup with significant neurodegeneration but without Aβ pathology or consistent profile of clinical symptoms (C. R. Jack, Jr. et al., 2016; J. C. Morris et al., 2010). Similar disparities are seen with tau pathology. Primary age-related tauopathy (PART) is used to describe the group of older adults, ranging widely in level of cognitive impairment, who exhibit tau pathology in the absence of amyloidosis (Crary et al., 2014). Even more importantly, AD clinical trials targeting Aβ have failed (G. P. Morris et al., 2014). Recent research has suggested the possibility of 9 Neurocognitive profiles of AD variants distinct neuropathologic subtypes of AD, with serious implications for biological and cognitive marker validity. For example, in one study Murray et al. (2011) report a pattern of AD pathology that spares the hippocampus, yet is associated with a significantly higher rate of cognitive decline, earlier disease onset, and shorter disease duration. Therefore, the importance of temporal sequence and presence of Aβ and tau changes are uncertain, and these markers may represent just one of several paths leading to AD. Mild cognitive impairment (MCI) has been conceptualized as an intermediary step between normal cognition and dementia in aging (Dubois & Albert, 2004; Dubois et al., 2007; Dubois et al., 2014; R. C. Petersen, 2004; R. C. Petersen et al., 2001; Winblad et al., 2004). Specifically, the subset of individuals diagnosed with MCI who display amnestic symptoms (aMCI) have an increased conversion rate to AD (R. C. Petersen et al., 2001), and thus aMCI has been identified as a possible clinical manifestation of AD during the early disease stages. In a study by Aggarwal et al. (2005), individuals diagnosed with MCI who possess the ApoE4 allele, a genetic risk factor for AD, had a 93% increase in risk of progressing to a disease state (Aggarwal et al., 2005). Characterizing MCI by amnestic and non-amnestic cognitive performance, followed by single and multi- cognitive domain impairment, has been attempted to better conceptualize the clinical concept of MCI. However, it is still unclear whether splitting research groups based on subtypes of MCI has greater value in predicting progression to dementia. In a study by Schneider et al. (2009) subtypes of MCI had similar proportions of AD and other mixed pathologies (Schneider, Arvanitakis, Leurgans, & Bennett, 2009), and less than a quarter of MCI diagnosed individuals show a “typical” AD pathology (Abner, 2017). Furthermore, although individuals with MCI are at greater risk of AD, a large percentage develop alternate forms of dementia, remain stable, or even improve on cognitive measures at 10 Neurocognitive profiles of AD variants later time points (R. C. Petersen et al., 2006). Similar to AD, MCI diagnosis relies on clinical impression and inadequate cognitive measures. Results from Bondi et al. (2014) suggest that using more in-depth neuropsychological criteria improves MCI diagnosis in several key ways, including better prediction of progression to AD and decreased diagnostic errors (Bondi et al., 2014; Jak, Urban, McCauley, et al., 2009). Even when these approaches that optimally balance the sensitivity and specificity of diagnostic criteria are applied to large groups of older adults there are a substantial number of MCI patients who “revert” to normal over follow-up. These patients are still at increased risk of dementia over subsequent years (citation from Petersen mayo clinic) relative to those who have never been diagnosed, suggesting that neurocognitive compensation mechanisms may play a role in masking the clinical impact of AD pathology during these early disease stages. Other factors that may impact MCI definition include age of diagnosis (Ward, Arrighi, Michels, & Cedarbaum, 2012) and education level (Gauthier et al., 2006). The “cognitive reserve hypothesis” suggests that individual differences exist that allow for increased ability to compensate for AD related neuropathologic changes and associated cognitive impairment (Serra, 2015). Two such differences, brain and cognitive reserves, have been proposed, the former indicating that individuals with more initial brain volume have more protection against neurodegeneration, and the latter that differences in cognitive processing (i.e. synaptic organization and connectivity) allow individuals to compensate for loss (Stern, 2009). Education, occupational attainment and brain volume are the most common measures of cognitive reserve. Because AD pathology is progressive, we can surmise that a more severe degree of AD pathology is required for these individuals to become symptomatic, and thus would not reach diagnosis until later disease stages. Furthermore, studies have shown that those with higher 11 Neurocognitive profiles of AD variants cognitive reserve tend to die closer to time of disease onset, and demonstrate more rapid cognitive decline (Stern, 2009). Older age at AD diagnosis is meaningful in that older adults are more likely to demonstrate increased clinical heterogeneity, including differences in cortical atrophy (Dickerson, 2017), and that the oldest old are more likely to be susceptible to multiple pathologies (i.e. neurodegeneration of the AD-type and vascular pathology) (Tanskanen, 2017). These differences occlude symptom presentation and hinder early disease diagnosis. More work is necessary to characterize preclinical and mild cognitive impairment states, and improve understanding of their associations with well-established AD markers. Advances in AD neuroimaging- and CSF-based biomarker research, have led to development of disease markers that are highly sensitive to diagnostic changes that may be detected in clinically normal older adults many years prior to the diagnosis of dementia. The revelation that there is such a long prodromal phase of AD has led to increased research efforts to target the disease at an early stage. Thus, there is a clear need to understand the pathophysiological changes leading up to symptomatic states. Research criteria for “preclinical AD”, proposed by the National Institute on Aging and the Alzheimer’s Association (NIA-AA), define this phase as part of the pathophysiological process of AD, but do not equate it with the clinical status of having AD. The difficulty with this definition lies in the fact that no measurable biomarker in asymptomatic individuals strongly predicts future clinical AD symptoms. In other words, there exists a group of older adults who present as cognitively normal, but who exhibit biomarker evidence consistent with AD, and who may or may not receive a future AD diagnosis (Sperling et al., 2011). This has called into question the predominant amyloid cascade hypothesis since biomarker models are largely based on an amyloid-centric view of AD, yet prediction is imperfect and clinical trials focused on clearing Aβ have failed to yield any benefit for patients. 12 Neurocognitive profiles of AD variants In order to improve our understanding of the early stages of AD, and to help clarify which biomarkers are most proximal to clinical progression, the NIA-AA criteria for preclinical AD suggested that “subtle cognitive decline” may occur during the preclinical phase of disease and that this could be an area for future research. Early studies attempting to operationalize subtle cognitive decline using neuropsychological criteria indicate that small changes in cognitive performance may be detectable in this preclinical AD stage (Edmonds, Delano-Wood, Galasko, Salmon, & Bondi, 2015). In one study of 570 cognitively normal subjects, researchers classified groups by preclinical AD stage based on NIA-AA criteria using the number of abnormal biomarkers (e.g., neurodegeneration and amyloidosis) and cognitive markers. A high number of individuals were identified demonstrating neurodegeneration in absence of amyloidosis, in direct contrast to the temporal pathway predicted by the amyloid cascade hypothesis, and that a group of patients with subtle cognitive decline and high likelihood of progression to AD showed normal neuroimaging markers. Results from this study add to the body of growing evidence suggesting that AD is heterogenous in regard to sequence and presentation of both cognitive and biomarker profiles. Particularly during early disease stages, changes in cognition and biomarker status may surface, and are not well predicted by published models of biomarker sequence. Rather than presence of certain disease biomarkers (e.g., CSF Aβ) fitting within a defined temporal sequence (i.e., expectation that specific markers are indicative of early vs. late disease stages) (Sperling et al., 2011), this research group found that cumulative abnormal markers were the best indicator of dementia risk. Furthermore, combining both biological and cognitive disease markers resulted in better identification of individuals with greater risk for dementia. Although more research is necessary to examine the usefulness of subtle cognitive decline in AD prediction, these findings highlight the need for a combination of 13 Neurocognitive profiles of AD variants sensitive biological and cognitive AD markers to understand neuropathologic changes and predict AD diagnosis from variant disease presentations. Biomarkers have the most clinical utility when they can be measured in vivo, are present early in disease progression, and are indicative of changes specific to the disease of interest. The diagnostic utility of biomarkers, such as cerebrospinal fluid (CSF) Aβ and tau protein levels (Gupta et al., 2011), and magnetic resonance imaging (MRI) of the brain has been long established (Albert et al., 2011; McKhann et al., 1984). MRI has been used extensively in neuroimaging studies to investigate associations of structural brain differences with cognitive deficits in MCI and AD. The hippocampus is one of the most vulnerable regions first targeted by AD pathology, with both AD and MCI patients showing reduced hippocampal volumes (Callen, Black, Gao, Caldwell, & Szalai, 2001; Du et al., 2001). Tau pathology, on the other hand, has been observed, beginning in early adulthood, in select subcortical nuclei of the lower brainstem, namely the noradrenic projection neurons of the locus coeruleus, before spreading to transentorhinal regions of the cortex, the hippocampus, and then to further widespread cortical areas (Arnold, Hyman, Flory, Damasio, & Van Hoesen, 1991; Bobinski et al., 1995; Braak & Del Tredici, 2011; Braak, Thal, Ghebremedhin, & Del Tredici, 2011; Schonheit, Zarski, & Ohm, 2004). Earlier stages of AD are associated with atrophy in medial temporal regions, and later stages see spread of atrophy to prefrontal, parietal, posterior temporal, and cingulate cortex regions (McDonald et al., 2009). Smaller hippocampi are observed in individuals with AD, and are a conversion risk factor for patients with MCI (Apostolova et al., 2007). Entorhinal cortex volume has also been shown to predict conversion to AD and disease progression (Devanand et al., 2007). These structural differences have been established as reliable biomarkers of AD and 14 Neurocognitive profiles of AD variants MCI, and can distinguish individuals with cognitive impairment and neurodegenerative disease from healthy older adults. In order to further explore the pathway between MCI and AD, research has begun to focus on alternate explanations of cognitive impairment. For example, there appears to be a significant number of vascular changes that contribute to cognitive impairment and neurodegeneration in AD. Individuals with vascular risk factors such as hypertension have been shown to have more white matter lesions and greater hippocampal atrophy, greater ventricular volume, reduced gray matter, and reduced overall brain volume (Gianaros, Greer, Ryan, & Jennings, 2006; Raz & Rodrigue, 2006; Strassburger et al., 1997; Wiseman et al., 2004). Increased number of vascular risk factors has also been shown to associate with poorer cognitive performance and increased risk of dementia. (Kress et al., 2014; Weller, Subash, Preston, Mazanti, & Carare, 2008; Zlokovic, 2011). It is thought that stiffening of the cerebrovasculature, an age-related process, may decrease the capacity for drainage of cerebral Aß (Kress, 2014). Aß then accumulates in cortical tissue, leading to plaque formation and possibly increased white- matter lesions (Bell et al., 2009; Qiu, Winblad, & Fratiglioni, 2005; Weller, Boche, & Nicoll, 2009). Therefore, it is possible that vascular dysfunction plays a role in Aβ accumulation. Cerebrovascular etiology has also been presented as an explanation for the heterogeneity of MCI, as individuals with cognitive impairment in multiple domains are more likely to experience vascular risk factors (R. C. Petersen & Morris, 2005). Vascular risk factors have been previously associated with executive, attentional, and visuospatial impairment in the absence of memory dysfunction (Lo & Jagust, 2012; Muller et al., 2011; Reijmer et al., 2012). MRI markers of vascular pathology (e.g., white matter lesion) have been shown to associate with cognitive impairment, particularly on tasks of processing speed (de Groot, 2000). Furthermore, 15 Neurocognitive profiles of AD variants accumulation of vascular risk factors is associated with greater prevalence of cerebrovascular change in AD patients, and significant differences in MRI markers of cerebrovascular insult in MCI patients (Kennedy & Raz, 2009), as well as at autopsy (Bangen et al., 2015). Furthermore, the AD group with evidence of cerebrovascular changes had lower AD-related neuropathologic burden, yet the same degree of global cognitive impairment, suggesting that the combination of AD and vascular pathologies catalyze disease-associated cognitive symptoms. Previous studies have used cluster analysis in order to parse out the heterogeneity of AD and/or MCI presentation based on either biological or cognitive markers of AD. Cluster analysis is an exploratory data analysis approach that groups individuals together based on their similarity on certain measures. Studies have typically derived clusters of patients by scores on neuropsychological testing, by neuropathologic differences, or by alternate biomarker measures that may have relevance or predictive value for diagnosis, prognosis, and conversion from MCI to AD. Altogether, a number of studies using variable of classification techniques have explored heterogeneity of AD across all symptoms and disease markers, with little to no consensus. Using cluster analysis Noh et al (2014) derived AD groups with distinct patterns of cortical atrophy (Noh et al., 2014). One group with greater atrophy in bilateral parietal precuneus/dorsolateral frontal areas, yet relatively spared medial temporal lobes, performed substantially worse in multiple cognitive domains, including attention, visuospatial, and executive functioning. Noh et al. (2014) also suggested that because this group consisted of less ApoE4 carriers, it is possible that the neurodegenerative effects of ApoE4 are regionally specific to the medial temporal lobe, making alternate pathologic groups less susceptible to AD risk typically associated with the ApoE4 allele (Hirono et al., 2002; Noh et al., 2014). Others still have derived distinct clinicopathologic AD profiles based off neurofibrillary pathology (Murray 16 Neurocognitive profiles of AD variants et al., 2011), finding evidence for groups of typical and atypical (e.g., hippocampal-sparing and limbic-predominant) distributions of tau pathology in diagnosed AD cases. Using classification by neural networks, Warkentin et al. (2004) found evidence of at least three subgroups of AD patients with distinct cerebral blood flow pathology in the frontal, central and occipital regions, but a common deficit of cerebral blood flow in the bilateral temporoparietal regions. Still other studies have looked at differing AD phenotypes in the context of inflammatory response (Wilcock, 2014) Several studies demonstrate importance of comprehensive diagnostic criteria in the identification of clinically relevant cognitive impairment, and how the inclusion of subtypes may reduce inaccurate diagnoses. Libon et al. (2010) and Delano-Wood et al. (2009) provide evidence for distinct clusters of MCI patients, derived from cognitive scores, with amnestic and non-amnestic impairment (Delano-Wood et al., 2009; Libon et al., 2010). Using cluster analysis, Clark et al. (2013) report that a large proportion of conventionally-diagnosed MCI individuals do not show impairment across a wide range of neuropsychological tests (Clark et al., 2013) and do not show biomarker abnormalities (i.e., CSF A  and tau) or genetic factors (i.e., ApoE4) consistent with AD risk. Furthermore, using either conventional or more comprehensive cognitive criteria led to the identification of MCI subtypes that deviated from the typical amnestic presentation associated with AD. Vardy et al. (2013) report 3 distinct clusters of patients impaired in attention and language, but differing in level of memory impairment, after adjustment for both age and disease duration. This result has been echoed in similar studies, with evidence that a large proportion of patients exhibit cognitive impairment in domains other than memory, and that this difference appears to be independent of disease severity (Stopford, Snowden, Thompson, & Neary, 2008). Several studies have found evidence for a specifically 17 Neurocognitive profiles of AD variants dysexecutive group, with impairment in attention, executive functioning, and visuospatial abilities, as well as a mixed impairment group with both memory deficits and language impairment (Clark et al., 2013; Delano-Wood et al., 2009; Libon et al., 2010). Thus far, results have been inconclusive as to whether these identified patient groups are representative of truly distinct disease subtypes, or different stages of AD. The majority of subgroup identification studies are limited to cross-sectional data, making it difficult to examine the stability of subtypes over time. Furthermore, these distinct subtypes differ not only in neuropsychological deficits, but also in AD-related biomarkers. Upon further examination, individuals belonging to derived neuropsychological phenotypes have presented with distinct patterns of cortical atrophy corresponding to cognitive deficits (Edmonds et al., 2016). Clark et al. (2013) reported increased cortical atrophy in medial temporal lobe for the amnestic and mixed subtypes, when compared to both a group of normal controls, and the cluster derived normal group in their study. Delano- Wood et al. (2009) reported higher levels of white matter injury in the dysexecutive subtype. These studies highlight the importance of heterogeneity seen in both AD and MCI, as the currently established biomarkers, such as hippocampal atrophy or CSF Aβ or tau, may not hold associations with disease progression across these groups. In fact, cognitive impairment and AD pathology in older adults may originate from a number of different risk factors and biological mechanisms. Increased understanding of various AD presentations and their corresponding timelines, inclusive of biomarker, cognitive, and genetic status, could improve diagnostic criteria and better inform the development of early interventions. 18 Neurocognitive profiles of AD variants Study Aims and Hypotheses Previous studies have established separate biomarker and cognitive signatures of preclinical AD, consistent with the key distinction between the etiology of AD and its manifestation as a clinical syndrome. However, research has demonstrated heterogeneity in both etiology and clinical syndrome in AD, as reviewed above, and improvement in our understanding of how etiological markers relate to clinical syndromes is a key research goal. Additionally, the separate focus on etiological factors or clinical syndromes in prior studies is at odds with the NIA-AA diagnostic criteria for preclinical AD and MCI due to AD, which takes both factors into account prior to diagnosis, as is common practice in clinical settings. Thus, research aiming to identify AD subtypes may benefit from an approach that uses a combination of biological and cognitive markers to reveal distinct phenotypes of aging that could provide a more clinically useful heuristic for patient classification and prediction of aging trajectories. Previously, researchers have used cluster analysis to identify distinct subgroups of AD using either cognitive performance or biomarker signature (Cappa, Ciccarelli, Baldonero, Martelli, & Silveri, 2014; Clark et al., 2013; Noh et al., 2014; Peter et al., 2014; Scheltens et al., 2015; Vardy et al., 2013), but to date no studies have investigated empirically-derived neurocognitive profiles of preclinical AD using a combination of CSF and cognitive markers. Such an approach may provide more complete information regarding aging phenotypes and that it may be more readily applied to future clinical practice where diagnostic and prognostic decision-making are likely to involve simultaneous examination of both cognitive and CSF markers. Furthermore, by considering group trajectories over multiple time points, we may be able to expand on our ability to identify these individuals at different levels of symptom presentation and disease severity. Many studies which have contributed to the definition of MCI and AD are conducted from cross 19 Neurocognitive profiles of AD variants sectional data. Yet, as a progressive disease characterized by decline and degeneration, examination of longitudinal outcomes has a great advantage in conceptualization of clinical presentation and identification of true subtypes. Thus, the current study aimed to classify participants based on both biological and cognitive markers of AD using cluster analysis, and to compare the resulting groups in terms of demographic profiles and trajectories of brain atrophy and clinical progression. Findings will better inform clinical practices in terms of identifying those at greatest risk for future decline, but may also shed light on potential factors beyond AD biomarkers that may impact the clinical expression of the diseaseSeveral groups, or subtypes, of preclinical AD were expected to emerge from the cluster analysis. At least one group was expected to present with a pattern of cognitive impairment andbiomarker signature that differs from both the cognitively normal and classic AD. Another aim was to determine whether groups differ rates of progression to AD diagnosis. It is hypothesized that using a combination of biological and cognitive markers to characterize preclinical aging profiles will improve prediction of progression to AD, and identify true disease subtypes. Therefore, the study will include further investigation of associations on functional, cognitive, and behavioral measures for clinical utility (including: Clinical Dementia Rating, Functional Activities Questionnaire, Mini Mental Status Exam, ADAS-cog). Group differences in important disease-related characteristics and risk factors will be explored. It is expected that the typical AD group will show patterns of cortical thinning in brain regions already associated with AD and memory (medial temporal lobe, including the hippocampus), and the normal group to have relatively preserved brain volume. Should atypical 20 Neurocognitive profiles of AD variants groups result from the analysis, they may show a different pattern of cortical atrophy that will be consistent with their cognitive presentation. Vascular risk variables that may contribute to cognitive impairment and neurodegeneration will be explored. Should a subgroup Atypical subgroups are expected to present with a more widespread (i.e., not purely amnestic) pattern of cognitive impairment exist, and it may be expected that such groups will have greater vascular risk burden than the amnestic or normal cognition groups.It is possible that an atypical group would exhibit differences in risk profile, including key factors such as age, sex, ApoE4, white matter lesion burden, sex and eductation, that may contribute to differences in clinical and biological presentation.. The majority of preclinical AD studies are limited by cross-sectional design. The aim of this study was to examine longitudinal trajectories of patterns of cognitive impairment and rates of change. Another aim was to examine atrophic changes of cortical matter for each group over time. All three groups were expected to show variation in baseline neuropsychological testing scores in all domains. No AD group was expected to show minimal decline in cognitive skills over time, and the typical AD group to deteriorate further in memory and other cognitive domains over time. All three groups were expected to show variation in baseline measures of hippocampal and ventricular volume. Finally, all groups were expected to show reduction of hippocampal volume and increased ventricular volume over time. 21 Neurocognitive profiles of AD variants Materials and Methods Participants The current study population is composed of 367 normal control (NC) and 609 mild cognitive impairment (MCI) subjects participating in the Alzheimer’s Disease Neuroimaging Initiative (ADNI). ADNI is a longitudinal study with approximately 50 sites across the United States and Canada (www.adni.loni.usc.edu). ADNI was designed to track the progression of Alzheimer’s disease using clinical and cognitive tests, magnetic resonance imaging (MRI), fluorodeoxyglucose (FDG), and Aβ PET, as well as cerebrospinal fluid and blood biomarkers. The study is supported by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies and nonprofit organizations. Michael W. Weiner, MD, VA Medical Center and University of California–San Francisco, is the Principal Investigator for ADNI. ADNI represents the efforts of many co- investigators from a broad range of academic institutions and private corporations. The initial goal of ADNI was to recruit 800 adults, ages 55 to 90, to participate in the research— approximately 200 cognitively normal older individuals to be followed for 3 years, 400 individuals with MCI to be followed for 3 years, and 200 individuals with early AD to be followed for 2 years. The clinical description of the ADNI cohort has been previously published (R. C. Petersen, 2007; R. C. Petersen et al., 2010; R. C. S. Petersen, G.E.; Waring, S.C.; Ivnik, R.J.; Tangalos, E.G.; Kokmen, E., 1999). Diagnosis of AD was based on the National Institute of Neurological and Communicative Disorders and Stroke and the AD and Related Disorders Association (NINCDS-ADRDA) criteria (Dubois et al., 2007; C. R. J. A. Jack, M.S.; Knopman, D.S.; McKhann, G.M.; Sperling, R.A.; Carrillo, M.C.; Thies, B.; Phelps, C.H., 2011; McKhann 22 Neurocognitive profiles of AD variants et al., 1984). MCI subjects had memory complaints but no significant functional impairment, scored between 24 and 30 on the MMSE, had a global CDR score of 0.5, a CDR memory score of 0.5 or greater, and objective memory impairment on Wechsler Memory Scale – Logical Memory II test (D., 1987). NC subjects had MMSE scores between 24 and 30, a global CDR of 0 and did not meet criteria for MCI and AD. Subjects were excluded if they refused or were unable to undergo MRI, had other neurological disorders, active depression, or history of psychiatric diagnosis, alcohol, or substance dependence within the past 2 years, less than 6 years of education, or were not fluent in English or Spanish. The full list of inclusion/exclusion criteria may be accessed on pages 23–29 of the online ADNI protocol (see http://www.adni- info.org/Scientists/ADNIScientistsHome.aspx). Written informed consent was obtained from all participants. The original ADNI-1 protocol has been followed by ADNI-GO and ADNI-2. For up-to-date information see www.adni-info.org Neuropsychological Battery All ADNI subjects completed an annual battery of neuropsychological tests. Baseline, month 12, and month 24 cognitive assessment scores (N=747) were available from the Laboratory of NeuroImaging (LONI) ADNI repository (https://ida.loni.usc.edu). This battery includes several clinical ratings, including the Clinical Dementia Rating (CDR), and Mini Mental Status Exam (MMSE). The Alzheimers Disease Assesment Scale (ADAS-cog) is a popular test of global cognition used by many researchers, and is also administered as part of this battery. Six neuropsychological assessments from three cognitive domains were chosen for classification of subjects. The raw scores for each neuropsychological measure were converted to z-scores prior to grouping analyses. Normed scores were also calculated using UDS Z-scores normed by age, 23 Neurocognitive profiles of AD variants sex, and education (TMT A&B, Category Fluency, BNT, MMSE) or age-normed Z-scores (RAVLT). Language Language was assessed using a Category Fluency test and the Boston Naming Test (BNT). During Category Fluency, participants are asked to name as many animals as they can in 60 seconds (Kaplan, Goodglass, & Weintraub, 1983). The Boston Naming Test is a test of word retrieval during which participants are asked to name a series of line drawings (Kaplan et al., 1983). Executive Functioning Trail-Making Test (TMT) A and B were used to assess executive functioning. In Trails A participants are presented with an array of 25 circles numbered from 1 to 25. The participant is instructed to draw lines connecting circles in ascending numerical order as quickly as possible (e.g. from 1 to 2; 2 to 3; 3 to 4, and so on). A maximum of 150 seconds is allowed before the trial is ended. Trails B also includes 25 circles, with 13 numbered from 1 to 13, and 13 labelled with letters A through L. Participants are instructed to connect circles, alternating between numbers and letters, in ascending numerical and alphabetical orders respectively (e.g. from A to 1; 1 to B; B to 2; 2 to C, and so on). A maximum of 500 seconds for completion is allowed. For both trials, scoring is based on number of seconds taken to complete each task (Spreen & Strauss, 1998). Memory Rey’s Auditory Verbal Learning Test (RAVLT) delayed recall and recognition subtests were used to evaluate memory. The RAVLT involves the oral presentation of a list of 15 words over 5 learning trials. After each trial participants recall as many words as possible. A distractor 24 Neurocognitive profiles of AD variants list of 15 words is next introduced, and participants recall as many of these new unrelated words as possible. The participants are asked to recall words from the first list after the distractor list trial, and again after a 30 minute delay. Finally, a recognition test is administered (Rey, 1941). Physiological and Clinical Measures Subjects completed annual blood draws and physiological assessments including seated blood pressure, weight, height, and temperature measurements. At screening, participants completed an extensive medical history inventory. Physiological data used in the present study included seated brachial artery systolic and diastolic blood pressures, weight, and height. Pulse pressure (PP) was calculated as systolic minus diastolic pressure. Body mass index (BMI) was calculated as weight (kg) divided by height (meters) squared. Vascular risk factors were identified from information collected during the clinical interview and physical examination, including evidence of hypertension, hypercholesterolemia, smoker status, and type 2 diabetes. The standard ADNI protocol for blood plasma collection is outlined online (http://adni.loni.usc.edu/methods/biomarker-analysis/). Fasting plasma samples were obtained and analyzed according to the xMAP Luminex platform and Innogenetics/Fujirebio AlzBio3 immunoassay kits. Levels of cholesterol and triglycerides were derived from blood samples. Cerebral Spinal Fluid Biomarkers Subjects participated in biennial lumbar punctures for CSF collection. Aliquots of all ADNI GO plus ADNI 2 CSF samples were analyzed using the xMAP Luminex platform and Innogenetics/Fujirebio AlzBio3 immunoassay kits. Using cutoff values reported by Shaw and colleagues (2009) for the ADNI sample, subject’s biomarker values for Aβ and p-tau were separately coded “positive” or “negative” for AD-type profile. In a separate variable, individuals were also coded positive for AD-type profile on both biomarkers. 25 Neurocognitive profiles of AD variants Genotyping All ADNI 1 participants were genotyped according to the Illumina Human610-Quad BeadChip (Illumina, Inc., San Diego, CA) protocol, and all ADNI-2/GO participants were genotyped according to the Illumina HumanOmniExpress BeadChip (Illumina, Inc., San Diego, CA) manufacturer’s protocol. Magnetic Resonance Imaging Data Volumetric MRI data available was processed by the Center from Imaging of Neurodegenerative Diseases at University of California, San Francisco. Cortical reconstruction and volumetric segmentation wass performed with the FreeSurfer image analysis suite. FreeSurfer analysis was completed using Version 4.3 for ADNI1 cross-sectional data[UCSFFSX], Version 4.4 for ADNI1 longitudinal data[UCSFFSL], and Version 5.1 for ADNI GO and 2 data[UCSFFSX51]. Semi-automated hippocampal volumetry is performed using Medtronic Surgical Navigation Technologies (SNT). Data were available for 768 subjects at baseline. White matter hyperintensity volumes were derived from available subject MRI scans by the Department of Neurology and Center for Neuroscience at the University of California, Davis. ADNI 1 subjects white matter lesion burden was detected using an automated segmentation method taking a previously published Bayesian Markov-Random Field (MRF) approach (Schwarz, Fletcher, DeCarli, & Carmichael, 2009). For ADNI 2/GO subjects, white matter hyperintensities were quantified using an updated four-tissue segmentation pipeline (adni.loni.usc.edu). Thus, separate analyses were run for each of the two segmentation methods. Data were available for 292 ADNI 1 subjects at baseline. Data were available for 559 26 Neurocognitive profiles of AD variants ADNIGO/2 subjects at baseline. Log-transformation was applied to correct for observed kurtosis in the distribution of the WMH volumes. Statistical Analyses Cluster analysis is a technique used to empirically group individuals based on their similarities within a set of characteristics. The advantage of this technique is that it does not rely on predetermined diagnostic criteria to form groups. We applied cluster analysis to all participants who underwent neuropsychological testing and lumbar puncture and were identified as either cognitively normal or MCI, specifying that the groups be clustered on a set of both cognitive and biomarker variables regardless of diagnostic status (Table 1). The cognitive variables included baseline neuropsychological testing scores across several cognitive domains, and the biomarker variables included baseline CSF measures of Aβ and p-tau. Discriminant function analyses (DFA) selected the optimal number of clusters and to quantitatively assess the ability of our selected cognitive and biomarker measures to correctly predict cluster membership. Chi-square analyses and one-way ANOVA were used to examine cluster-derived group demographic differences (e.g. age, gender, education, ApoE4, etc.). Analaysis of covariance (ANCOVA) and logistic regression were used as follow-up analyses to include covariates (e.g., age, gender, education, ApoE4) in group comparisons. For cross-sectional analyses, analysis of covariance (ANCOVA) was conducted with least significant difference (LSD) post-hoc analysis to compare the derived groups on cognitive performance, biomarker status, and other disease risk factors. Age, sex, education, and ApoE4 carrier status were included as covariates. General linear mixed models (GLMM) with unstructured covariance matrix and maximum likelihood estimation were used for longitudinal examination of group membership as a predictor of a subset of the outcome variables. Group x time, age, sex education and ApoE4 carrier status were 27 Neurocognitive profiles of AD variants also included in the model as fixed factors. The intercept and time were entered as a random effects. Three time points (e.g., baseline, month 12, and month 24 visit) were included in the model and coded as 0, 1, and 2. LSD post-hoc analysis was used to compare group differences over time. Cox’s regression was employed to examine the rate of progression to dementia for the three cluster-derived groups. Analyses were two-tailed with α set at p < .05. All analyses were performed with SPSS for Windows OS version 20.0.0 (SPSS, 263 Armonk, NY: IBM Corp). 28 Neurocognitive profiles of AD variants Results Hierarchical cluster analysis Participants were sorted into three distinct subgroups resulting from the cluster analysis. One group was labeled No AD (n=653), which was characterized by normal cognitive performance and least amount of biomarker elevation. Another group, labeled typical AD (n=191), was characterized by substantial memory impairment, with relatively well preserved executive and language abilities, and elevated biomarker signature consistent with AD. Finally, a third group, labeled atypical AD (n=132), was characterized by more mildly impaired scores on memory testing, but also similar impairments in executive functioning and language domains, and biomarker signature that fell mid-way between the normal group and the typical AD group. Means and standard deviations for baseline measures used in cluster analysis are provided in Table 2. Discriminant function analysis Two discriminant functions (Figure 1) were obtained accounting for 69.2% and 30.8% of variance among the three subgroups. The combination of neuropsychological and biomarker measures correctly classified 89.2% of original grouped cases. To validate these results, a leave- one-out cross-validation technique was used after which the percentage of correctly classified cases was 88.9%. Group comparisons on neuropsychological assessment Cross-sectional Resulting groups from the cluster analysis were characterized by pattern of neuropsychological performance and biomarker status. Mean normed scores on 29 Neurocognitive profiles of AD variants neuropsychological tests for each group are shown in Figure 2. For each group, mean raw scores for each neuropsychological test and standard deviations are provided in Table 2. All three groups differed from one another on raw baseline memory scores such that the atypical AD group had greater score than the typical AD group, and no AD had greater score than both AD groups, [AVLT delayed recall: F(2, 968)=148.13, p<0.001; AVLT recognition: F(2, 968)=247.14, p<0.001]. These differences were also significant on normed Z-scores, [AVLT delayed recall: F(2, 968)=139.04, p<0.001; AVLT recognition: F(2,968)=180.49, p<0.001]. Post-hoc pairwise comparisons revealed the typical AD group to be significantly more impaired in memory than the atypical AD group (AVLT delayed recall: p<0.001; AVLT recognition: p<0.001) and the No AD group (AVLT delayed recall: p<0.001; AVLT recognition: p<0.001). The atypical AD showed significantly lower memory scores than the No AD group, (AVLT delayed recall: p<0.001; AVLT recognition: p<0.001). In the language domain, all three groups differed onbaseline raw scores such that the atypical AD group score was significantly lower than the typical AD group, and both AD groups score was lower than the No AD group [Category Fluency: F(2, 968)=86.40, p<0.001; BNT: F(2, 968)=132.95, p<0.001], as well as normed Z-scores [Category Fluency: F(2, 968)=86.35, p<0.001; BNT: F(2, 968)=132.91, p<0.001]. Post-hoc pairwise comparisons revealed the atypical AD group to be more impaired on both measures than the typical AD group (Category Fluency: p<0.001; BNT: p<0.001) and the No AD group (Category Fluency: p<0.001; BNT: p<0.001), and the typical AD had lower score than the No AD group (Category Fluency: p<0.001; BNT: p<0.001). All three groups differed on both raw and age-adjusted Z-scores scores for baseline executive functioning measures such that the atypical AD group was significantly worse than the 30 Neurocognitive profiles of AD variants typical AD group, and both AD groups was worse than the No AD group [TMT A: F(2, 968)=152.04, p<0.001; TMT B: F(2, 968)=382.73, p<0.001], as well as normed Z-scores [TMT A: F(2, 968)=151.92, p<0.001; TMT B: F(2, 968)=381.94, p<0.001]. Post-hoc pairwise comparisons revealed the atypical AD group to show greater impairment on both measures compared to the typical AD group (TMT A: p<0.001; TMT B: p<0.001) and No AD group (TMT A: p<0.001; TMT B: p<0.001). The typical AD group also performed significantly worse on tests of executive function than the No AD group (TMT A: p<0.001; TMT B: p<0.001). Longitudinal Figures 3–5 describe group differences on neuropsychological tests over time. At both month 12 and month 24 follow up visits, the atypical AD group performed worse on language and executive functioning tests than both the typical AD (language: p<0.001; executive functioning: p<0.001) and No AD (language: p<0.001; executive functioning: p<0.001) groups. The atypical AD group also performed worse than No AD on tests of memory (p<0.001). The typical AD group performed worse on memory tests than the atypical (p<0.001) and No AD groups, and performed worse on tests of language and executive functioning than the No AD group (language: p<0.001; executive functioning: p<0.001). Observed differences are expected as these tests were used to characterize groups in the cluster analysis. Group comparisons on biomarker status Significant group differences were observed for biomarker status [Aβ: F(2, 968)=102.01, p<0.001; p-tau: F(2, 968)=58.17, p<0.001]. After post-hoc pairwise comparisons, the typical AD group demonstrated the highest level of biomarker elevation on both CSF measures, followed by the atypical AD group, and the No AD group (p<0.001) (Figure 6). 31 Neurocognitive profiles of AD variants The typical AD group was composed of a greater percentage of Aβ positive individuals (97.9%) than the atypical AD group (77.9%) (χ 2 [1]=33.94 p<0.001) and the No AD group (40.2%) (χ 2 [1]=197.26 p<0.001). The atypical AD group had a greater percentage of Aβ positive individuals than the No AD group (40.2%) (χ 2 [1]=62.02 p<0.001). The typical AD group was composed of a greater percentage of p-tau positive individuals (93.2%) than the atypical AD group (84.8%) (χ 2 [1]=5.93 p<0.013) and the No AD group (68.9%) (χ 2 [1]=45.75 p<0.001). The atypical AD group had a greater percentage of p-tau positive individuals than the No AD group (χ 2 [1]=13.71 p<0.001). 91.6% of the typical AD group was positive for both Aβ and p-tau markers, which was a greater percentage than the atypical AD group (68.7%) (χ 2 [1]=29.11, p<0.001) and No AD group (33.5%) (χ 2 [1]= 201.43, p<0.001). The atypical AD group had a greater percentage of individuals positive for both biomarkers than the No AD group (χ 2 [1]=55.77, p<0.001). Demographic differences Subject demographics in each cluster-derived group are described in Table 3. The three groups did not significantly differ from one another in sex (χ 2 [2]=3.22, p=0.199). The groups significantly differed in education (F[2,974]=7.43 p=0.001) and age (F[2,974]=12.30 p<0.001). The atypical AD group had less years of education compared to both typical AD (p=0.022) and No AD groups (p<0.001). The atypical AD group was older than both the typical AD (p=0.044) and normal groups (p<0.001). Groups significantly differed from one another on baseline diagnosis (χ 2 [2]=132.60, p<0.001). Specifically, the typical AD group was composed of significantly more MCI diagnosed individuals than the No AD (χ 2 [1]=2.61 p<0.001) and atypical group (χ 2 [1=0.097] p=0.039), with 32 Neurocognitive profiles of AD variants 90.6% of the typical AD group, 49.9% of the no AD group, and 83.3% of the atypical AD group being diagnosed (Figure 7). Finally, all three groups significantly differed from one another on number of ApoE4 carriers (typical AD > atypical AD > No AD: χ 2 [2]= p<0.001; typical vs. atypical χ 2 [1]=27.55, p<0.001). The No AD group had 30.8% carriers, the typical AD group had 74.9% carriers, and the atypical AD group had 46.2% carriers (Figure 8). Vascular risk No significant differences between any of the cluster-derived groups were observed in seated diastolic blood pressure at baseline visit, smoker status, history of diabetes, history of hypercholesterolemia, or total blood levels of triglycerides or cholesterol (Table 3). In the ADNI 1 sample at baseline, white matter lesion burden was no different for the typical AD group than the atypical (p=0.709) or No AD (p=0.735) after including covariates in the model. White matter lesion burden was not significantly different in the ADNI 1 sample for the atypical AD group compared to the No AD group (p=0.426). In the ADNIGO/2 sample at baseline, white matter lesion burden was not significantly increased for the atypical AD group compared to both the typical AD group (p=0.188). White matter lesion burden was increased for the atypical AD group compared to the No AD (p=0.004) group. Using one-way ANOVA, the atypical AD group had significantly higher average systolic blood pressure than the No AD group (p=0.039), however this difference was no longer seen after including the specified covariates in the univariate general linear model (p=0.203). Neither the No AD group (β=0.097, p=0.591) or the atypical AD group (β=0.282, p=0.231) had a greater proportion of individuals with history of hypertension than the typical AD group. There was a significant difference among groups in BMI [F(2, 33 Neurocognitive profiles of AD variants 964)=6.546, p=0.002]. Post-hoc pairwise comparisons revealed both atypical (p<0.001) and typical (p<0.001) AD groups to have lower BMI when compared with the No AD group. Progression to dementia The three groups differed from one another significantly on proportion of individuals exhibiting diagnosis of either AD or MCI at month 24 follow up (atypical vs. typical: χ 2 [1]=0.097 p=0.008), but there was no significant difference between AD groups on dementia diagnosis alone (χ 2 [1]=0.985 p=0.195). The typical AD group had a higher number of diagnosed individuals (i.e. AD or MCI) (92.9%) than the atypical AD group (82.1%). At month 24 follow up visit, 41% of the typical AD group had been diagnosed with AD, 34.7% of the atypical AD group, and 4.5% of the No AD group (atypical vs. typical χ 2 [1]=7.07 p=0.029) (Figure 9). Cox’s regression revealed a significantly greater risk of AD diagnosis at month 24 follow up visit between the typical and No AD groups (p<0.001), but not between typical and atypical AD groups (p=0.804). The No AD group had 1.99 times lesser risk of AD diagnosis than the typical group. Cox’s regression revealed no significantly different risk of MCI diagnosis at month 24 follow up visit between the typical and No AD groups (p=0.246), nor between typical and atypical AD groups (p=0.726). Cox’s regression revealed a significantly greater risk of either MCI or AD diagnosis at month 24 follow up visit between the typical and No AD groups (p<0.001), but not between typical and atypical AD groups (p=0.804). The No AD group had 0.590 times lesser risk of AD diagnosis than the typical group. Using Cox’s regression (Figure 10), there was evidence of a significantly greater risk of eventual AD diagnosis (i.e., over variable follow up time points) between the typical AD group and the No AD group (p<0.001), but no significant difference between the typical AD group and the atypical AD group (p=0.730). When compared to the No AD group, belonging to the typical 34 Neurocognitive profiles of AD variants AD group had 4.398 times greater risk of AD diagnosis, and belonging to the atypical AD group had 4.120 times greater risk of AD diagnosis. Clinical outcomes There was a non-significant trend toward decreased MMSE score (i.e., worse cognition) for the atypical group compared with the typical group (p=0.075). Both the typical (p<0.001) and atypical (p<0.001) significantly differed from the No AD group (Figure 11). Using GLMM, there was a significant group by time interaction for MMSE score such that the No AD group exhibited a slower rate of decline than the typical AD group, (Mean Difference ± SE=1.88±0.21, t(2, 849.8)=9.16, p<0.001), but the rate at which AD groups declined was not different, (Mean Difference ± SE=-0.13±0.29, t(2, 881.1)=-0.458, p=0.647). There was a significant effect of group on MMSE score such that the No AD group performed better than the typical group, (Mean Difference ± SE=2.22±0.15, t(2, 993.2)=14.9, p<0.001), but the atypical AD group did not differ from the typical AD group in MMSE score, (Mean Difference ± SE=0.26±0.20, t(2, 2638)=-1.29, p=0.199). Post-hoc pairwise comparison revealed that the change in MMSE score differed between the atypical and No AD groups (p<0.001). There was an effect of time, as average MMSE score decreased for all individuals, (Mean Difference ± SE=-2.09±0.18, t(2, 853.5)=-11.529, p<0.001). There was a significant group by time interaction in CDR score from baseline to month 24. The atypical AD group and typical AD group showed the same rate of decline, (Mean Difference ± SE=-0.091±0.17, t(2, 889.0)=-0.547, p=0.584). The No AD group had less decline than the typical group over time, (Mean Difference ± SE=-1.44±0.12, t(2, 865.5)=-12.2, p<0.001). There was a significant effect of group on CDR score such that the No AD group performed better than the typical group, (Mean Difference ± SE=-1.62±0.01, t(2, 1012.0)=-16.3, 35 Neurocognitive profiles of AD variants p<0.001) but the atypical AD group did not differ from the typical AD group in CDR score (Mean Difference ± SE=-0.16±0.13, t(2, 981.5)=-1.20, p=0.230) (Figure 12). Post-hoc pairwise comparison revealed that the No AD group showed less change in CDR score than the atypical AD group, (Mean Difference ± SE=1.64±0.10, t(2,868.6)=15.7, p<0.001). There was an effect of time, as average CDR score decreased for all individuals (p<0.001). Similarly, there was a significant group by time interaction on ADAS-cog score from baseline to month 24. The atypical AD group exhibited the same amount change in ADAS-cog score over follow-up as the typical AD group, (Mean Difference ± SE=0.36±0.76, t(2, 909.8)=0.471, p=0.638). The typical group exhibited greater rate of change on the ADAS-cog than the No AD group over follow-up, (Mean Difference ± SE=-3.43±0.54, t(2, 891.5)=-6.36, p<0.001). There was a significant effect of group on ADAS-cog score such that the No AD group performed better than the typical group, (Mean Difference ± SE=-9.55±0.53, t(2, 1000.8)=-18.2, p<0.001) but the atypical AD group did not score differently than the typical AD group (Mean Difference ± SE=-0.57±0.70, t(2, 990.5)=-0.823, p=0.411). Post-hoc pairwise comparison revealed that the change in ADAS-cog was greater for the atypical than the No AD group (p<0.001). There was an effect of time, as ADAS-cog score increased for all individuals, (Mean Difference ± SE=2.98±0.48, t(2, 894.7)=6.261, p<0.001). On the FAQ, there was a significant group by time interaction such that the No AD group, again, declined less rapidly than the typical AD group, (Mean Difference ± SE=-4.11±0.37, t(2, 864.6)=-11.0, p<0.001), but the atypical AD group did not differ in rate of change of FAQ score from the typical AD group, (Mean Difference ± SE=0.34±0.53, t(2, 891.9)=-0.651, p=0.515). There was an effect of group on FAQ score such that the atypical AD group scored lower on the FAQ (i.e., more preserved functioning) than the typical AD group, (Mean Difference ± SE=- 36 Neurocognitive profiles of AD variants 1.21±0.47, t(2, 990.4)=-2.58, p=0.010). Additionally, the No AD group scored lower than the typical group, (Mean Difference ± SE=-5.05±0.35, t(2, 1020.8)=-14.5, p<0.001) (Figure 13). Post-hoc pairwise comparison revealed that the change in FAQ score was greater for the atypical than the No AD group (p<0.001). There was a significant effect of time, as average FAQ score decreased for all individuals, (Mean Difference ± SE=4.78±0.33, t(2,867.09)=14.6, p<0.001). Group comparisons on MRI volumetrics Cross-sectional The typical AD group showed smaller brain volume within the medial temporal lobe. Specifically, groups significantly differed on baseline measures of bilateral hippocampal volume, [F(2, 853)=65.72, p<0.001]. The typical AD group had smaller bilateral hippocampal volumes than the atypical AD group (p=0.020), and both AD groups had smaller volumes than the No AD group, (typical vs. No AD: p<0.001; atypical vs. No AD: p<0.001). The atypical AD group showed greater preservation of the right entorhinal cortex compared to the typical (p=0.023) and No AD groups (p=0.039). The typical AD group also showed smaller average cortical volume than the No AD (p=0.010) group in the left entorhinal region. In the left entorhinal region, there was a trend towards significance when comparing the typical and atypical groups (p=0.085). The atypical AD group exhibited smaller brain volume in widespread cortical regions when compared to the typical AD group. Groups significantly differed on baseline measures of bilateral ventricular volume, [F(2, 929)=18.69, p<0.001]. The atypical AD group demonstrated greater bilateral ventricular enlargement than the typical AD group (p=0.049), and both AD groups demonstrated greater bilateral enlargement than the No AD group (typical vs. No AD: p<0.001; atypical vs. No AD: p<0.001). The atypical AD group showed greater signs of cortical atrophy than both the typical and No AD groups in the right supramarginal (p=0.007), right 37 Neurocognitive profiles of AD variants medial orbitofrontal (p=0.002), left superior temporal (p=0.040), and left lingual (p=0.011) regions. Both AD groups showed smaller cortical volumes than the No AD group across a wide range of cortical and subcortical areas. Specifically, the atypical group had smaller average volumes than the No AD group in the right parahippocampal (p<0.001), right posterior cingulate (p=0.003), right putamen (p=0.017), right rostral anterior cingulate (p=0.049), right superior frontal (p<0.001), right isthumus cingulate (p=0.017), left amygdala (p=0.031), and left caudate (p=0.008). Figures 14 & 15 provide visualizations of differing pattern of atrophy between the atypical with the typical AD group. Longitudinal Lack of follow-up data hindered the power and validity of longitudinal analyses in all regions of interest, and so models were run only with ventricular and hippocampal volumes over multiple time points. Using GLMM, there was a significant group by time interaction such that the No AD group had less profound reduction in hippocampal volume than the typical AD group over follow-up, (Mean Difference ± SE=-224.22±32.74, t(2, 733.9)=6.85, p<0.001), no difference in reduction of hippocampal volume over time existed between the two AD groups, (Mean Difference ± SE=80.40±47.57, t(2, 766.7)=1.69, p=0.091). Similarly, there was a significant group by time interaction such that the No AD group had a significantly less pronounced ventricular enlargement than the typical AD group over follow-up (Mean Difference ± SE=-4134.6±377.75, t(2, 810.0)=-11.0, p<0.001), the two AD groups did not differ in rate of ventricular enlargement (Mean Difference ± SE=176.07±530.25, t(2, 834.8)=0.332, p=0.740) (see Figures 16 & 17). 38 Neurocognitive profiles of AD variants Discussion The clinical manifestation of AD during its lengthy prodromal phase varies greatly among individuals. This variability also extends to disease biomarkers, which can differ in temporal sequence, number of present markers, and degree of abnormality across individuals. This muddles both characterization of the initial presentation and symptom development throughout the span of the disease, including the prodromal transition period. Poor characterization often leads to poor diagnostic decisions, lack of ability to make an accurate prognosis, and development of ineffective treatment options. In the present study, three subgroups were identified using a combination of baseline neurocognitive and biomarker features. One, the No AD group, represented healthy older adults at the least risk of developing AD, with normal biomarker signature, more preserved brain volume and with preserved cognitive functioning. Another group, typical AD, had a memory impairment and elevated biomarker status, with relatively preserved language and executive functioning, and greatest medial temporal atrophy. The last group, atypical AD, was greatly dysexecutive and dysnomic, as well as amnestic, but did not show as severe memory impairment as the typical AD group. This atypical group had intermediate biomarker elevation, suggesting CSF levels of Aβ and tau to be less salient disease biomarkers for this group. Despite these differences both AD groups seemed to progress to AD diagnosis at similar rates, and were rated and scored similarly on clinical measures of global cognition over time. At month 24 follow up visit a smaller proportion of members of the atypical AD group were identified as cognitively impaired (i.e. MCI or AD) though they demonstrated impairment across multiple cognitive domains. It was notable that the atypical AD group was composed of a smaller percentage of ApoE4 carriers. It is well documented that ApoE4 carrier status greatly increases risk of AD 39 Neurocognitive profiles of AD variants development. The ApoE4 allele plays a role in amyloid accumulation and clearance that influences AD disease progression (J. Kim et al., 2009). ApoE4 has also been shown to influence the complex pathophysiological processes in AD. In other words, an interaction between the APOE-dependent non-pathological changes and age-related pathological changes may provide some explanation for the mechanism leading to AD progression and resulting neurodegeneration (Reinvang, Espeseth, & Westlye, 2013). ApoE4 has been shown to increase the effects that vascular risk factors have on cognitive performance, and researchers have proposed that this relationship is driven by high systolic blood pressure (Zade et al., 2010). Interestingly, the atypical group in this study showed some evidence of vascular risk compared to the No AD group, including some evidence of higher systolic blood pressure, however ApoE4 carrier status was largely absent in this subgroup. Although a strong association exists between ApoE4 genotype and CSF levels of Aβ, studies suggest that the relationship between Aβ and AD exist independent of ApoE4 (Lautner et al., 2014). In sum, the atypical AD group was composed of fewer ApoE4 carriers, exhibited less biomarker elevation, and no difference in vascular risk, compared to the typical AD group. Thus, it is likely that some ApoE4 independent mechanism may be driving the neurodegenerative and cognitive changes seen in this subgroup. Investigation of potential differences in regional atrophy between the AD subgroups was conducted for several important reasons. It is unclear why the atypical group exhibits different neurocognitive features from the typical group, and previous work has identified neuropathologic subgroups with hippocampal sparing pattern of atrophy. Furthermore, MRI scans are commonly used by clinicians in the event of a suspected neurodegenerative disorder. As the atypical AD group in our sample presented with less clinically relevant CSF biomarker signature, and less clear biomarkers of AD-related neurodegeneration, characterization of a distinct pattern of 40 Neurocognitive profiles of AD variants atrophy may help to prevent this group from escaping neurologist notice. For this reason, clinicians would greatly benefit from comprehensive neurocognitive profiles to diagnose patients. The atypical AD group differed from the typical AD group on pattern of cortical atrophy, in several key regions reported to play a role in AD. The typical AD group tended to show increased atrophy of the medial temporal lobe (e.g., hippocampus and entorhinal cortex), whereas the average hippocampal volume of atypical AD group fell somewhere between the typical and No AD groups. The atypical AD group exhibited more widespread cortical atrophy than the typical group across frontal and temporal regions (e.g., right supramarginal, right medial orbitofrontal left superior temporal, and left lingual). Regions of atrophy were consistent with observed cognitive profile. Again, relative to the typical AD group, the atypical AD group showed relative preservation of the medial temporal lobe. Therefore, as evidenced by this anatomical difference, along with described cognitive and CSF biomarker differences between the atypical and typical AD groups, the atypical AD group appears to represent a distinct subgroup. As predicted, the amnestic, or typical AD, group exhibited greatest hippocampal atrophy of all three groups, along with greater ventricular enlargement compared to the cognitively normal group. Because the atypical group presented with greatest impairment in language and executive functioning, in addition to memory impairment, it was predicted that a different pattern of atrophy localized to the parietal, precuneus, and dorsolateral frontal lobes, similar to the dysexecutive group described by Noh and colleagues (2012), would arise. The observed differences between the AD groups were centered around frontal, temporal, and subcortical regions, with the atypical AD group showing increased signs of atrophy in these areas. These differences are at least partly consistent with previous work, and coincide with presentation of 41 Neurocognitive profiles of AD variants cognitive deficits. The typical group continued to fit into the expected profile of AD, with early signs of atrophy in the entorhinal cortex. The percentage of AD biomarker positive, in both Aβ and p-tau, individuals for the atypical AD group was smaller than that of the typical AD group. The typical group showed biomarker signature well beyond clinical cutoffs for AD in the ADNI study (Shaw et al., 2009). Evidence, in the framework of the amyloid cascade hypothesis, suggests that biomarkers become abnormal prior to clinical presentation of symptoms or neurodegeneration (C. R. Jack, Jr. et al., 2010). Because the atypical AD group exhibited more severe cognitive deficits, and greater cortical atrophy in several regions, yet less biomarker elevation, it does not appear that this group represents a more intermediate disease stage than the typical AD group. A cognitive profile characterized by executive impairment and relative preservation of memory ability has been associated with vascular disease impacting frontal-subcortical networks. Thus, it was hypothesized that the atypical AD group may show increased evidence of vascular risk or cerebrovascular disease. No significant differences in vascular risk factors, or other key demographic features, were observed between AD groups, making it difficult to categorize individuals without more detailed neuropsychological and biomarker information. Differences in vascular risk or white matter lesion burden between groups were inconclusive in this study, as white matter lesion burden was significantly increased for the atypical group compared to No AD but not compared to typical AD. Similarly, the atypical AD group showed some evidence of increased systolic blood pressure, however this difference was no longer significant after inclusion of covariates. Finally, the atypical AD group had increased percentage of individuals with history of hypertension compared to the No AD group. It is likely that age differences between the atypical and No AD group contributed to these group differences in 42 Neurocognitive profiles of AD variants white matter lesion burden, systolic blood pressure, and history of hypertension. This suggests that more work is necessary to investigate the pathophysiological differences that may be leading to clinical and biomarker differences. Of note, the atypical group was somewhat older than individuals belonging to No AD, the group least likely to progress towards a dementia diagnosis. The atypical group was older, yet increased age was not indicative of greater disease severity nor biomarker elevation in this group. Therefore, it is unlikely that the differences between groups are driven by increasing age alone. Older adults with AD often exhibit co-occurring neuropathologic changes (i.e., vascular brain injury, lewy body disease) (Brenowitz et al., 2017; Kawas et al., 2015; Schneider, Arvanitakis, Bang, & Bennett, 2007), particularly vascular pathology (Rahimi & Kovacs, 2014). As no evidence of increased vascular injury was observed in this group, and subjects were free from stroke (i.e., Hachinski Ischemic Score of less than or equal to 4), it is unlikely that the differences between groups are driven by increasing mixed pathologies. Evidence has pointed to high prevalence of mixed vascular and AD pathology, particularly in older adult samples (Mungas, Reed, Ellis, & Jagust, 2001). This does not seem to apply as although the atypical AD group was significantly older, there were no major group differences in vascular risk. The lack of distinction in vascular risk or markers of vascular injury between the two cluster-derived AD groups is suggestive that differentiable patterns of AD-related pathology, or other age-related non-pathological changes, in the brain may be the driving force behind observed differences in CSF biomarker levels and neuropsychological impairment. In addition to exhibiting older age, the atypical group also tended to have less education than the typical and No AD groups. Evidence suggests that individuals with higher educational attainment may show increased resilience to both age-related non-pathological and AD-related 43 Neurocognitive profiles of AD variants brain changes (Stern, 2012). High tau and Aβ burden, measured using Pittsburgh compound (PiB) positron emission tomography (PET), have previously been associated with lower-level education (Shimada et al., 2017). The severity and span of cognitive impairment seen in the atypical AD group, despite less salient biomarker elevation, may be partially explained by increased susceptibility to brain changes due to age and education level. Group differences in global cognitive performance seen at baseline remained stable over multiple time points, and there was an overall decline in scores on the MMSE, CDR, and ADAS- cog for each group. Stability of presentation over time provides evidence that identified groups are representative of distinct disease subtypes. Domain specific characterization of the groups (e.g., in memory, language, executive functioning) revealed that despite presenting with less severe memory impairment and greater language and executive functioning impairment relative to the typical group, both groups progressed to dementia at the same rate. The typical AD group was more likely to be identified as cognitively impaired than the atypical group (i.e., MCI at baseline, MCI or AD at month 24) by a modest (7.3%), yet statistically significant percentage. Clinical ratings of global impairment (CDR) were not different between AD groups over multiple time points. On informant-report of daily functioning the atypical group tended to score better than the typical group across multiple timepoints (FAQ). These differences have several important implications. The AD groups did not differ in rate of decline on these measures of cognition, and this stability over time suggests they are indicative of true disease subtypes. It is possible that the atypical group is harder to diagnose earlier on due to unusual and subthreshold presentation of cognitive and biological markers, different genotypic expression, and lower informant report of functional impairment. Approximately 16.7% of atypical AD subjects were not identified as MCI. These “false negative” cases are indicative of the importance of a 44 Neurocognitive profiles of AD variants comprehensive profile of disease, as their atypical neurocognitive profile is likely a large factor in their mis-diagnosis as cognitively normal. In relation to this, our finding that more than two thirds of subjects were classified as No AD is also important, as the study sample was enriched for MCI using conventional diagnostic criteria. Previous work using cluster analysis has identified a “false positive” problem in ADNI and other studies when using standard MCI diagnosis criteria (Clark et al., 2013; Edmonds, Delano-Wood, Clark, et al., 2015; Jak, Urban, Mc, et al., 2009). In this case, it is possible that diagnostic criteria for clinical AD should be expanded to include alternate profiles of disease in order to avoid diagnostic error. We may also gain insight from these results about the characterization of non- pathological aging and which particular cognitive or functional changes individuals find distressing. It is interesting that though the atypical AD group presented on neuropsychological testing with more severe and widespread cognitive deficits, they were rated similarly by clinicians on the CDR, and scored similarly on tests of global cognition such as the MMSE and ADAS-cog. Additionally, informants rated those in the typical AD group as more functionally impaired than those in the atypical AD group. The presence of memory impairment may be more debilitating in daily life, or more noticeable to others, and more likely to drive individuals to visit a specialist, than language or executive function deficits. Memory deficits have historically been central to the definition of MCI (R. C. Petersen, 2004, 2007; R. C. S. Petersen, G.E.; Waring, S.C.; Ivnik, R.J.; Tangalos, E.G.; Kokmen, E., 1999). Informant rating of functional impairment has been shown to be predictive of progression to AD, rather than self-assessment of ability (Tabert et al., 2002). Some research has investigated whether specific subjective cognitive complaints in preclinical stages of AD have more utility for clinicians in prediction of AD. In one study, forgetfulness of immediate information (i.e., immediate memory or naming failures), 45 Neurocognitive profiles of AD variants executive functioning (i.e., distractibility), and prospective memory (i.e., monitoring of information one has to recall in the future) complaints were best at discriminating between healthy and MCI individuals (Avila-Villanueva et al., 2016). Thus, the type of impairment is not only important to bringing the individual into the office, it is also possible that clinicians may weight different concerns are more or less important in the diagnostic decision making process. The present study is subject to several limitations. The ADNI study recruits and collects data from subjects from over 50 centers across the United States and Canada, and is not representative of the general population, due to exclusion criteria and oversampling of memory impairment. The study is currently in its third phase and not all measures have been collected for all participants, and a high percentage of individuals have discontinued from all or parts (e.g., lumbar puncture) of data collection. A limitation to the validity of CSF measures is that although CSF comes into contact with the extracellular space around the brain, the sample is taken from the spine and is therefore an indirect measure that does not provide visual or spatial information regarding the location of amyloid or tau deposition in the brain and vulnerable structures. Another limitation is the diminished follow-up data, particularly for neuroimaging and CSF measures, which threatens the power of longitudinal analyses. A unique contribution of this study is the use of the combination of biological and cognitive disease markers to characterize preclinical aging. This combination of markers may provide superior prognostic and diagnostic value at early disease stages. Previous work has suggested that such a “Bioprofile” (Escudero, Ifeachor, & Zajicek, 2012; Escudero, Zajicek, & Ifeachor, 2011). However, previous work aimed at identifying a neurocognitive profile of AD have used suboptimal instruments of cognitive assessment to classify individuals. One advantage of the present study is that only two distinct subgroups were identified within a mixed group 46 Neurocognitive profiles of AD variants previously identified as cognitively normal or MCI. This cluster solution is somewhat simpler and perhaps easier to identify and apply to the current diagnostic conceptualization than previous studies where 4+ groups were identified. Another strength of the study is the large community- based sample which is provided by the ADNI study, and wide availability of different measures over multiple timepoints, such as a comprehensive neuropsychological testing battery. Longitudinal assessment allows for examination of subgroups over time, and stability of group features. This provides greater support that groups truly represent clinical subtypes rather than disease stages. The differences in biomarker status between two groups equally at risk for AD suggest that the Aβ and tau mechanisms typically attributed to AD pathology and progression are not an adequate explanation for cognitive impairment and neurodegeneration seen in the atypical AD group. This is especially highlighted by the fact that the atypical AD group is significantly older and less educated, yet did not show the same level of memory impairment or pathology. These results have clinical significance in terms of treatment, early diagnosis, and prognosis. For example, while memory impairment and CSF Aβ and tau levels may be good indicators of disease progression for individuals in the typical AD group, the atypical AD group do not meet the same criteria. Differences in pathologic presentation may also indicate that these groups may respond differently to treatment options. Future work will aim at explaining differences between these groups. The information provided by this study hopefully will help research move towards a better and more comprehensive understanding of how different factors associated with AD may interact and influence presentation of biological and cognitive disease markers. 47 Neurocognitive profiles of AD variants Tables and Figures Domain Measure Memory AVLT 30 Min Delay Free Recall AVLT Delayed Recognition Executive Function Trails A Trails B Language Boston Naming Test Category Fluency CSF Biomarker Aβ1-42 P-Tau181P Table 1. Features used for classification in hierarchical cluster analysis Abbreviations: AVLT = Rey Auditory Verbal Learning Test; P-Tau = Phosphorylated Tau; CSF = Cerebrospinal Fluid 48 Neurocognitive profiles of AD variants No AD Typical AD Atypical AD AVLT Delayed Recall (words) 6.80 (3.83) 1.30 (1.78) 3.64 (3.54) AVLT Recognition (words) 12.80 (0.10) 7.89 (0.20) 10.27 (0.23) Category Fluency (words) 20.02 (5.06) 18.55 (5.54) 13.18 (4.24) Boston Naming Test (names) 27.88 (2.32) 26.76 (2.60) 23.08 (5.06) Trails A (seconds) 33.59 (10.22) 39.86 (13.10) 59.22 (27.28) Trails B (seconds) 80.67 (27.38) 114.17 (56.90) 200.20 (69.79) CSF Aβ1-42 (pg/mL) 201.23 (50.91) 131.76 (24.05) 160.87 (46.12) CSF P-Tau181P (pg/mL) 31.64 (18.00) 54.08 (26.62) 36.93 (15.74) Table 2. Summarized Mean (Standard Deviation) raw baseline neuropsychological and biomarker values. Each pairwise comparison between groups (e.g., Atypical vs. Typical, No AD vs. Typical, No AD vs. Atypical) was statistically significant with p<0.001 for each comparison. Abbreviations: AVLT = Rey Auditory Verbal Learning Test; CSF = Cerebrospinal Fluid 49 Neurocognitive profiles of AD variants No AD (N=653) TAD (N=191) AAD (N=132) Group comparison p- value Age (years) M(SD) 72.2 (6.8) 73.4 (7.0) 76.1 (7.0) No AD<TAD<AAD <0.001 Baseline Diagnosis (MCI) % 49.9 90.6 83.3 No AD<AAD<TAD <0.001 Gender (M:F) % 52.8:47.2 58.1:41.9 59.8:40.2 0.199 Education (years) M(SD) 16.4 (2.6) 16.1 (2.7) 15.4 (15.4) AAD<TAD=No AD <0.001 ApoE4 (Carriers:Non-carriers) % 30.8 74.9 46.2 No AD<AAD<TAD <0.001 Hippocampal Volume (mm^3) M(SD) WML Volume (cm^3) ADNI 1 M(SD) WML Volume (cm^3) ADNI Go/2 M(SD) 7380.9 (985.8) -0.66 (0.75) 0.46 (0.52) 6357.1 (1011.6) -0.62 (0.68) 0.63 (0.47) 6547.3 (1011.1) -0.51 (0.65) 0.79 (0.54) TAD<AAD<No AD AAD>No AD <0.001 0.724 0.004 50 Neurocognitive profiles of AD variants Ventricular Volume (mm^3) M(SD) Smoker Status % Hypertension % Hypercholesterolemia % Cholesterol M(SD) Triglycerides M(SD) Diabetes % 34024.1 (18989.4) 40.0 46.1 50.4 193.6 (38.4) 143.9 (83.9) 9.0 41104.8 (22669.9) 40.3 46.6 51.5 194.1 (39.5) 144.5 (98.7) 7.9 47850.6 (25041.0) 36.4 56.1 57.1 193.3 (45.7) 155.7 (194.8) 8.3 AAD<TAD<No AD AAD>No AD <0.001 0.721 0.023 0.266 0.985 0.550 0.866 51 Neurocognitive profiles of AD variants Pulse Pressure M(SD) 59.3 (14.5) 60.8 (15.3) 61.9 (15.0) 0.127 Systolic BP M(SD) Diastolic BP M(SD) BMI M(SD) 133.9 (16.5) 74.6 (9.4) 27.7 135.3 (16.6) 74.6 (9.6) 26.2 137.1 (15.7) 75.3 (9.4) 26.1 No AD<AAD TAD=AAD<No AD 0.039 0.765 <0.001 Table 3. Demographic comparisons Data are summarized as either Mean(Standard Deviation) or percentage as indicated. Significant differences (p < .05) among groups are indicated in bold. White matter lesion values were log-transformed for analyses. Abbreviations: ApoE4 = apolipoprotein E; BMI = body mass index; BP = blood pressure; MCI = mild cognitive impairment; WML = white matter lesion; TAD = typical AD; AAD = atypical AD 52 Neurocognitive profiles of AD variants Subjects with Month 24 Follow-up DX N=809 Subject without Month 24 Follow-up DX N=167 p-value Age (years) M(SD) 72.8 (7.1) 73.6 (6.5) 0.196 Baseline Diagnosis (Normal:MCI) % 38.4:61.6 33.5:66.5 0.134 Gender (M:F) % 54.9:45.1 54.5:45.5 0.496 Education (years) M(SD) 16.2 (2.7) 16.0 (2.8) 0.296 ApoE4 (Carriers:Non-carriers) % 59.7:40.3 52.7:47.3 0.058 Hippocampal Volume (mm^3) M(SD) WML Volume (cm^3) a M(SD) Ventricular Volume (mm^3) M(SD) 7106.0 (1084.4) 0.098 (0.83) 36466.3 (20324.9) 6929.0 (1103.6) 0.22 (0.74) 6929.0 (1103.6) 0.073 0.105 0.009 53 Neurocognitive profiles of AD variants Smoker Status % Hypertension % Hypercholesterolemia % Cholesterol M(SD) Triglycerides M(SD) Diabetes % 38.3 46.7 52.5 192.8 (38.4) 143.9 (88.1) 9.1 39.8 47.7 48.5 198.2 (45.3) 154.5 (181.0) 6.6 0.395 0.440 0.194 0.139 0.288 0.180 Pulse Pressure M(SD) 59.9 (14.5) 59.9 (15.7) 0.997 Systolic BP M(SD) Diastolic BP M(SD) 134.3 (16.1) 74.3 (9.3) 136.1 (17.7) 76.2 (10.0) 0.185 0.021 54 Neurocognitive profiles of AD variants BMI M(SD) 27.2 (4.8) 27.2 (5.0) 0.909 Supplementary Table 1. Data are summarized as either Mean(Standard Deviation) or percentage as indicated. Significant differences (p < .05) among groups are indicated in bold. a Log transformed values Abbreviations: ApoE4 = apolipoprotein E; BMI = body mass index; BP = blood pressure; MCI = mild cognitive impairment; WML = white matter lesion; DX = diagnosis 55 Neurocognitive profiles of AD variants Figure 1. Discriminant function analysis indicating three cluster groups. Individual scores are plotted on two statistically significant discriminant dimensions (i.e., function 1 and function 2) in order to visualize predicted group membership classifications. 56 Neurocognitive profiles of AD variants Figure 2. Baseline average neuropsychological performance using normed scores for the three cluster-derived groups. Normed scores were calculated using UDS Z-scores normed by age, sex, and education (TMT A&B, Category Fluency, BNT,) or age-normed Z-scores (RAVLT). The No AD group shows little to no baseline cognitive impairment, whereas the typical group performs significantly worse than both the No AD and atypical AD groups on tests of memory, and the atypical group performs significantly worse than the typical AD and No AD groups on tests of language and executive function. The atypical AD group also performs worse than the No AD group on tests of memory. 57 Neurocognitive profiles of AD variants Figure 3. Visualization of group differences in memory performance over time. There was a significant effect of group on memory performance over time. At baseline, groups significantly differed on memory performance (typical<atypical<No AD). For visualization and descriptive purposes only, a memory score was computed using z-scores for the two memory tests used for cluster analysis. 58 Neurocognitive profiles of AD variants Figure 4. Visualization of group differences in executive function performance over time. There was a significant effect of group on executive function performance over time. At baseline, significant group differences were observed (atypical<typical<No AD). For visualization and descriptive purposes only, a executive functioning score was computed using z-scores for the two executive functioning tests used for cluster analysis. 59 Neurocognitive profiles of AD variants Figure 5. Visualization of group differences in language performance over time. At baseline, group differences in language performance were observed (atypical<typical<No AD). There was a significant effect of group on language performance over time. For visualization and descriptive purposes only, a language score was computed using z-scores for the two language tests used for cluster analysis. 60 Neurocognitive profiles of AD variants Figure 6. Baseline comparison of cluster-derived group differences on mean CSF biomarker level. Z-scores were calculated using means and standard deviations of the sample. Inverse z- scores for abeta are presented such that positive values indicate higher biomarker elevation. The typical AD group presented with significantly higher levels of AD-related biomarkers, followed by the atypical group. The No AD group showed little evidence of AD-related biomarker status. 61 Neurocognitive profiles of AD variants Figure 7. Diagnosis at baseline. All three groups significantly differ at baseline on diagnosis. The typical group exhibited highest percentage of MCI diagnosed individuals, followed by the atypical, and then the No AD group. 62 Neurocognitive profiles of AD variants Figure 8. Group percentages of ApoE4 carriers. All three groups statistically differ from one another on proportion of carriers. The typical AD group is composed of significantly more ApoE4 carriers, followed by the atypical, and then the No AD groups. 63 Neurocognitive profiles of AD variants Figure 9. Diagnosis at month 24. All three groups differ on proportion of diagnosed individuals at month 24. The typical and atypical groups did not differ on percentage of AD diagnosed individuals, however they did statistically differ on percentage of individuals identified as cognitively impaired (AD or MCI). The No AD group had highest percentage of cognitively normal individuals. 64 Neurocognitive profiles of AD variants Figure 10. Cox’s regression indicating risk of AD diagnosis for each group over multiple time points (measured in months). The No AD group had significantly less risk of developing AD over more than 5 years. The two AD groups did not differ in risk of progressing to AD diagnosis. 65 Neurocognitive profiles of AD variants Figure 11. Baseline age-adjusted MMSE z-scores. The AD groups did not differ from one another on measure of global cognitive impairment. Both AD groups were showed more global cognitive impairment than the No AD group at baseline. *p<0.05 66 Neurocognitive profiles of AD variants Figure 12. Change in CDR scores over 3 time points. Visualization of group trajectories on CDR sum of boxes score over a two-year time period. Further analyses revealed a group x time interaction, showing the No AD group performance remained stable over time, while both AD groups declined in ratings over time. 67 Neurocognitive profiles of AD variants Figure 13. Change in self-rating of functional impairment over 3 time points. Visualization of group trajectories on FAQ score over a two-year time period. Further analysis revealed that the No AD group score remained stable over time, whereas the typical AD group showed the greatest amount of decline over two years, followed by the atypical AD group. 68 Neurocognitive profiles of AD variants Figure 14. Pink areas indicate regions in which the atypical AD group showed greater cortical atrophy than the typical AD group at baseline. MRI scans are presented in radiological orientation. Regions were highlighted on the fslView standard MNI152 T1 brain using the Harvard-Oxford Cortical and Subcortical structural atlases for region denotation. A) and C) provide two coronal view of regions at different slices within the brain. B) provides sagittal view of regions. 69 Neurocognitive profiles of AD variants Figure 15. Pink regions indicate areas in which the typical AD group exhibited smaller average cortical volumes than the atypical AD group at baseline. MRI scans are presented in radiological orientation. Regions were highlighted on the fslView standard MNI152 T1 brain using the Harvard-Oxford Cortical and Subcortical structural atlases for region denotation. A) provides coronal view and B) provides sagittal view of regions. 70 Neurocognitive profiles of AD variants Figure 16. Visualization of group differences in hippocampal volume over time. Relative comparisons (typical<atypical<No AD) made at baseline remain significant at follow-up visits. Further analyses revealed that while the No AD group saw relatively little hippocampal atrophy over two years, both AD groups declined in hippocampal volume at similar rates. 71 Neurocognitive profiles of AD variants Figure 17. Visualization of group differences in memory performance over time. Further analyses revealed that while the No AD group saw relatively little ventricular enlargement over two years, both AD groups’ ventricles grew at similar rates. 72 Neurocognitive profiles of AD variants References 2015 Alzheimer's disease facts and figures. (2015). Alzheimers Dement, 11(3), 332-384. Aggarwal, N. T., Wilson, R. S., Beck, T. L., Bienias, J. L., Berry-Kravis, E., & Bennett, D. A. (2005). The apolipoprotein E epsilon4 allele and incident Alzheimer's disease in persons with mild cognitive impairment. Neurocase, 11(1), 3-7. doi:10.1080/13554790490903038 Albert, M. S., DeKosky, S. T., Dickson, D., Dubois, B., Feldman, H. H., Fox, N. C., . . . Phelps, C. H. (2011). The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7(3), 270-279. doi:10.1016/j.jalz.2011.03.008 Andreasen, N., Minthon, L., Vanmechelen, E., Vanderstichele, H., Davidsson, P., Winblad, B., & Blennow, K. (1999). Cerebrospinal fluid tau and Abeta42 as predictors of development of Alzheimer's disease in patients with mild cognitive impairment. Neurosci Lett, 273(1), 5-8. Apostolova, L. G., Hwang, K. S., Andrawis, J. P., Green, A. E., Babakchanian, S., Morra, J. H., . . . Thompson, P. M. (2010). 3D PIB and CSF biomarker associations with hippocampal atrophy in ADNI subjects. Neurobiol Aging, 31(8), 1284-1303. doi:10.1016/j.neurobiolaging.2010.05.003 Apostolova, L. G., Steiner, C. A., Akopyan, G. G., Dutton, R. A., Hayashi, K. M., Toga, A. W., . . . Thompson, P. M. (2007). Three-dimensional gray matter atrophy mapping in mild cognitive impairment and mild Alzheimer disease. Arch Neurol, 64(10), 1489-1495. doi:10.1001/archneur.64.10.1489 73 Neurocognitive profiles of AD variants Arnold, S. E., Hyman, B. T., Flory, J., Damasio, A. R., & Van Hoesen, G. W. (1991). The topographical and neuroanatomical distribution of neurofibrillary tangles and neuritic plaques in the cerebral cortex of patients with Alzheimer's disease. Cereb Cortex, 1(1), 103-116. Avila-Villanueva, M., Rebollo-Vazquez, A., Ruiz-Sanchez de Leon, J. M., Valenti, M., Medina, M., & Fernandez-Blazquez, M. A. (2016). Clinical Relevance of Specific Cognitive Complaints in Determining Mild Cognitive Impairment from Cognitively Normal States in a Study of Healthy Elderly Controls. Front Aging Neurosci, 8, 233. doi:10.3389/fnagi.2016.00233 Bell, R. D., Deane, R., Chow, N., Long, X., Sagare, A., Singh, I., . . . Zlokovic, B. V. (2009). SRF and myocardin regulate LRP-mediated amyloid-beta clearance in brain vascular cells. Nature cell biology, 11(2), 143. Bennett, D. A., Schneider, J. A., Bienias, J. L., Evans, D. A., & Wilson, R. S. (2005). Mild cognitive impairment is related to Alzheimer disease pathology and cerebral infarctions. Neurology, 64(5), 834-841. doi:10.1212/01.wnl.0000152982.47274.9e Bobinski, M., Wegiel, J., Wisniewski, H. M., Tarnawski, M., Reisberg, B., Mlodzik, B., . . . Miller, D. C. (1995). Atrophy of hippocampal formation subdivisions correlates with stage and duration of Alzheimer disease. Dementia, 6(4), 205-210. Bondi, M. W., Edmonds, E. C., Jak, A. J., Clark, L. R., Delano-Wood, L., McDonald, C. R., . . . Salmon, D. P. (2014). Neuropsychological criteria for mild cognitive impairment improves diagnostic precision, biomarker associations, and progression rates. J Alzheimers Dis, 42(1), 275-289. doi:10.3233/jad-140276 74 Neurocognitive profiles of AD variants Braak, H., & Del Tredici, K. (2011). The pathological process underlying Alzheimer's disease in individuals under thirty. Acta Neuropathol, 121(2), 171-181. doi:10.1007/s00401-010- 0789-4 Braak, H., Thal, D. R., Ghebremedhin, E., & Del Tredici, K. (2011). Stages of the pathologic process in Alzheimer disease: age categories from 1 to 100 years. J Neuropathol Exp Neurol, 70(11), 960-969. doi:10.1097/NEN.0b013e318232a379 Brenowitz, W. D., Keene, C. D., Hawes, S. E., Hubbard, R. A., Longstreth, W. T., Jr., Woltjer, R. L., . . . Kukull, W. A. (2017). Alzheimer's disease neuropathologic change, Lewy body disease, and vascular brain injury in clinic- and community-based samples. Neurobiol Aging, 53, 83-92. doi:10.1016/j.neurobiolaging.2017.01.017 Callen, D. J., Black, S. E., Gao, F., Caldwell, C. B., & Szalai, J. P. (2001). Beyond the hippocampus: MRI volumetry confirms widespread limbic atrophy in AD. Neurology, 57(9), 1669-1674. Cappa, A., Ciccarelli, N., Baldonero, E., Martelli, M., & Silveri, M. C. (2014). Posterior AD- type pathology: cognitive subtypes emerging from a cluster analysis. Behav Neurol, 2014, 259358. doi:10.1155/2014/259358 Clark, L. R., Delano-Wood, L., Libon, D. J., McDonald, C. R., Nation, D. A., Bangen, K. J., . . . Bondi, M. W. (2013). Are empirically-derived subtypes of mild cognitive impairment consistent with conventional subtypes? J Int Neuropsychol Soc, 19(6), 635-645. doi:10.1017/s1355617713000313 Crary, J. F., Trojanowski, J. Q., Schneider, J. A., Abisambra, J. F., Abner, E. L., Alafuzoff, I., . . . Nelson, P. T. (2014). Primary age-related tauopathy (PART): a common pathology 75 Neurocognitive profiles of AD variants associated with human aging. Acta Neuropathol, 128(6), 755-766. doi:10.1007/s00401- 014-1349-0 D., W. (1987). Wechsler Memory Scale – Revised. Psychological Corporation; San, Antonio, TX: 1987. Deane, R., Sagare, A., Hamm, K., Parisi, M., Lane, S., Finn, M. B., . . . Zlokovic, B. V. (2008). apoE isoform-specific disruption of amyloid beta peptide clearance from mouse brain. J Clin Invest, 118(12), 4002-4013. doi:10.1172/jci36663 Delano-Wood, L., Bondi, M. W., Sacco, J., Abeles, N., Jak, A. J., Libon, D. J., & Bozoki, A. (2009). Heterogeneity in mild cognitive impairment: differences in neuropsychological profile and associated white matter lesion pathology. J Int Neuropsychol Soc, 15(6), 906- 914. doi:10.1017/s1355617709990257 Devanand, D. P., Pradhaban, G., Liu, X., Khandji, A., De Santi, S., Segal, S., . . . de Leon, M. J. (2007). Hippocampal and entorhinal atrophy in mild cognitive impairment: prediction of Alzheimer disease. Neurology, 68(11), 828-836. doi:10.1212/01.wnl.0000256697.20968.d7 Du, A. T., Schuff, N., Amend, D., Laakso, M. P., Hsu, Y. Y., Jagust, W. J., . . . Weiner, M. W. (2001). Magnetic resonance imaging of the entorhinal cortex and hippocampus in mild cognitive impairment and Alzheimer's disease. J Neurol Neurosurg Psychiatry, 71(4), 441-447. Dubois, B., & Albert, M. L. (2004). Amnestic MCI or prodromal Alzheimer's disease? Lancet Neurol, 3(4), 246-248. doi:10.1016/s1474-4422(04)00710-0 76 Neurocognitive profiles of AD variants Dubois, B., Feldman, H. H., Jacova, C., Dekosky, S. T., Barberger-Gateau, P., Cummings, J., . . . Scheltens, P. (2007). Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol, 6(8), 734-746. Dubois, B., Feldman, H. H., Jacova, C., Hampel, H., Molinuevo, J. L., Blennow, K., . . . Cummings, J. L. (2014). Advancing research diagnostic criteria for Alzheimer's disease: the IWG-2 criteria. Lancet Neurol, 13(6), 614-629. doi:10.1016/s1474-4422(14)70090-0 Edmonds, E. C., Delano-Wood, L., Clark, L. R., Jak, A. J., Nation, D. A., McDonald, C. R., . . . Bondi, M. W. (2015). Susceptibility of the conventional criteria for mild cognitive impairment to false-positive diagnostic errors. Alzheimers Dement, 11(4), 415-424. doi:10.1016/j.jalz.2014.03.005 Edmonds, E. C., Delano-Wood, L., Galasko, D. R., Salmon, D. P., & Bondi, M. W. (2015). Subtle Cognitive Decline and Biomarker Staging in Preclinical Alzheimer's Disease. J Alzheimers Dis, 47(1), 231-242. doi:10.3233/jad-150128 Edmonds, E. C., Eppig, J., Bondi, M. W., Leyden, K. M., Goodwin, B., Delano-Wood, L., & McDonald, C. R. (2016). Heterogeneous cortical atrophy patterns in MCI not captured by conventional diagnostic criteria. Neurology, 87(20), 2108-2116. doi:10.1212/wnl.0000000000003326 Escudero, J., Ifeachor, E., & Zajicek, J. P. (2012). Bioprofile analysis: a new approach for the analysis of biomedical data in Alzheimer's disease. J Alzheimers Dis, 32(4), 997-1010. doi:10.3233/jad-2012-121024 Escudero, J., Zajicek, J. P., & Ifeachor, E. (2011). Early detection and characterization of Alzheimer's disease in clinical scenarios using Bioprofile concepts and K-means. Conf Proc IEEE Eng Med Biol Soc, 2011, 6470-6473. doi:10.1109/iembs.2011.6091597 77 Neurocognitive profiles of AD variants Frisoni, G. B., & Jack, C. R. (2011). Harmonization of magnetic resonance-based manual hippocampal segmentation: a mandatory step for wide clinical use. Alzheimers Dement, 7(2), 171-174. doi:10.1016/j.jalz.2010.06.007 Gauthier, S., Reisberg, B., Zaudig, M., Petersen, R. C., Ritchie, K., Broich, K., . . . Winblad, B. (2006). Mild cognitive impairment. Lancet, 367(9518), 1262-1270. doi:10.1016/s0140- 6736(06)68542-5 Gianaros, P. J., Greer, P. J., Ryan, C. M., & Jennings, J. R. (2006). Higher blood pressure predicts lower regional grey matter volume: Consequences on short-term information processing. NeuroImage, 31(2), 754-765. Gupta, V. B., Laws, S. M., Villemagne, V. L., Ames, D., Bush, A. I., Ellis, K. A., . . . Martins, R. N. (2011). Plasma apolipoprotein E and Alzheimer disease risk: the AIBL study of aging. Neurology, 76(12), 1091-1098. doi:10.1212/WNL.0b013e318211c352 Hardy, J. A., & Higgins, G. A. (1992). Alzheimer's disease: the amyloid cascade hypothesis. Science, 256(5054), 184-185. Hirono, N., Hashimoto, M., Yasuda, M., Ishii, K., Sakamoto, S., Kazui, H., & Mori, E. (2002). The effect of APOE epsilon4 allele on cerebral glucose metabolism in AD is a function of age at onset. Neurology, 58(5), 743-750. Jack, C. R., Jr., Knopman, D. S., Chetelat, G., Dickson, D., Fagan, A. M., Frisoni, G. B., . . . Vos, S. J. (2016). Suspected non-Alzheimer disease pathophysiology - concept and controversy. Nat Rev Neurol, 12(2), 117-124. doi:10.1038/nrneurol.2015.251 Jack, C. R., Jr., Knopman, D. S., Jagust, W. J., Shaw, L. M., Aisen, P. S., Weiner, M. W., . . . Trojanowski, J. Q. (2010). Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurol, 9(1), 119-128. doi:10.1016/s1474-4422(09)70299-6 78 Neurocognitive profiles of AD variants Jack, C. R. J. A., M.S.; Knopman, D.S.; McKhann, G.M.; Sperling, R.A.; Carrillo, M.C.; Thies, B.; Phelps, C.H. (2011). Introduction to the recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimer's Dementia, 7(3), 257-262. Jak, A. J., Urban, S., Mc, C. A., Bangen, K. J., Delano-Wood, L., Corey-Bloom, J., & Bondi, M. W. (2009). Profile of hippocampal volumes and stroke risk varies by neuropsychological definition of mild cognitive impairment. J Int Neuropsychol Soc, 15(6), 890-897. doi:10.1017/s1355617709090638 Jak, A. J., Urban, S., McCauley, A., Bangen, K. J., Delano-Wood, L., Corey-Bloom, J., & Bondi, M. W. (2009). Profile of hippocampal volumes and stroke risk varies by neuropsychological definition of mild cognitive impairment. J Int Neuropsychol Soc, 15(6), 890-897. doi:10.1017/s1355617709090638 Johnson, K. A., Sperling, R. A., Gidicsin, C. M., Carmasin, J. S., Maye, J. E., Coleman, R. E., . . . Skovronsky, D. M. (2013). Florbetapir (F18-AV-45) PET to assess amyloid burden in Alzheimer's disease dementia, mild cognitive impairment, and normal aging. Alzheimers Dement, 9(5 Suppl), S72-83. doi:10.1016/j.jalz.2012.10.007 Kaplan, E., Goodglass, H., & Weintraub, S. (1983). Boston Naming Test. Philadelphia: Lea & Febiger. Karch, C. M., & Goate, A. M. (2015). Alzheimer's disease risk genes and mechanisms of disease pathogenesis. Biol Psychiatry, 77(1), 43-51. doi:10.1016/j.biopsych.2014.05.006 Karch, C. M., Jeng, A. T., Nowotny, P., Cady, J., Cruchaga, C., & Goate, A. M. (2012). Expression of novel Alzheimer's disease risk genes in control and Alzheimer's disease brains. PLoS One, 7(11), e50976. doi:10.1371/journal.pone.0050976 79 Neurocognitive profiles of AD variants Kawas, C. H., Kim, R. C., Sonnen, J. A., Bullain, S. S., Trieu, T., & Corrada, M. M. (2015). Multiple pathologies are common and related to dementia in the oldest-old: The 90+ Study. Neurology, 85(6), 535-542. doi:10.1212/wnl.0000000000001831 Kennedy, K. M., & Raz, N. (2009). Pattern of normal age-related regional differences in white matter microstructure is modified by vascular risk. Brain Res, 1297, 41-56. doi:10.1016/j.brainres.2009.08.058 Kim, J., Basak, J. M., & Holtzman, D. M. (2009). The role of apolipoprotein E in Alzheimer's disease. Neuron, 63(3), 287-303. doi:10.1016/j.neuron.2009.06.026 Kim, S., Swaminathan, S., Shen, L., Risacher, S. L., Nho, K., Foroud, T., . . . Saykin, A. J. (2011). Genome-wide association study of CSF biomarkers Abeta1-42, t-tau, and p- tau181p in the ADNI cohort. Neurology, 76(1), 69-79. doi:10.1212/WNL.0b013e318204a397 Kress, B. T., Iliff, J. J., Xia, M., Wang, M., Wei, H. S., Zeppenfeld, D., . . . Nedergaard, M. (2014). Impairment of paravascular clearance pathways in the aging brain. Annals of Neurology, 76, 845-861. Lambert, J. C., Ibrahim-Verbaas, C. A., Harold, D., Naj, A. C., Sims, R., Bellenguez, C., . . . Amouyel, P. (2013). Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer's disease. Nat Genet, 45(12), 1452-1458. doi:10.1038/ng.2802 Landau, S. M., Lu, M., Joshi, A. D., Pontecorvo, M., Mintun, M. A., Trojanowski, J. Q., . . . Jagust, W. J. (2013). Comparing positron emission tomography imaging and cerebrospinal fluid measurements of beta-amyloid. Ann Neurol, 74(6), 826-836. doi:10.1002/ana.23908 80 Neurocognitive profiles of AD variants Lautner, R., Palmqvist, S., Mattsson, N., Andreasson, U., Wallin, A., Palsson, E., . . . Hansson, O. (2014). Apolipoprotein E genotype and the diagnostic accuracy of cerebrospinal fluid biomarkers for Alzheimer disease. JAMA Psychiatry, 71(10), 1183-1191. doi:10.1001/jamapsychiatry.2014.1060 Libon, D. J., Xie, S. X., Eppig, J., Wicas, G., Lamar, M., Lippa, C., . . . Wambach, D. M. (2010). The heterogeneity of mild cognitive impairment: a neuropsychological analysis. J Int Neuropsychol Soc, 16(1), 84-93. doi:10.1017/s1355617709990993 Liu, Y., Yu, J. T., Wang, H. F., Han, P. R., Tan, C. C., Wang, C., . . . Tan, L. (2015). APOE genotype and neuroimaging markers of Alzheimer's disease: systematic review and meta- analysis. J Neurol Neurosurg Psychiatry, 86(2), 127-134. doi:10.1136/jnnp-2014-307719 Lo, R. Y., & Jagust, W. J. (2012). Vascular burden and Alzheimer disease pathologic progression. Neurology, 79(13), 1349-1355. doi:10.1212/WNL.0b013e31826c1b9d McDonald, C. R., McEvoy, L. K., Gharapetian, L., Fennema-Notestine, C., Hagler, D. J., Jr., Holland, D., . . . Dale, A. M. (2009). Regional rates of neocortical atrophy from normal aging to early Alzheimer disease. Neurology, 73(6), 457-465. doi:10.1212/WNL.0b013e3181b16431 McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E. M. (1984). Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology, 34(7), 939-944. Morris, G. P., Clark, I. A., & Vissel, B. (2014). Inconsistencies and controversies surrounding the amyloid hypothesis of Alzheimer's disease. Acta Neuropathol Commun, 2, 135. doi:10.1186/s40478-014-0135-5 81 Neurocognitive profiles of AD variants Morris, J. C., Roe, C. M., Xiong, C., Fagan, A. M., Goate, A. M., Holtzman, D. M., & Mintun, M. A. (2010). APOE predicts amyloid-beta but not tau Alzheimer pathology in cognitively normal aging. Ann Neurol, 67(1), 122-131. doi:10.1002/ana.21843 Muller, M., Appelman, A. P., van der Graaf, Y., Vincken, K. L., Mali, W. P., & Geerlings, M. I. (2011). Brain atrophy and cognition: interaction with cerebrovascular pathology? Neurobiol Aging, 32(5), 885-893. doi:10.1016/j.neurobiolaging.2009.05.005 Mungas, D., Reed, B. R., Ellis, W. G., & Jagust, W. J. (2001). The effects of age on rate of progression of Alzheimer disease and dementia with associated cerebrovascular disease. Arch Neurol, 58(8), 1243-1247. Murphy, K. R., Landau, S. M., Choudhury, K. R., Hostage, C. A., Shpanskaya, K. S., Sair, H. I., . . . Doraiswamy, P. M. (2013). Mapping the effects of ApoE4, age and cognitive status on 18F-florbetapir PET measured regional cortical patterns of beta-amyloid density and growth. Neuroimage, 78, 474-480. doi:10.1016/j.neuroimage.2013.04.048 Murray, M. E., Graff-Radford, N. R., Ross, O. A., Petersen, R. C., Duara, R., & Dickson, D. W. (2011). Neuropathologically defined subtypes of Alzheimer's disease with distinct clinical characteristics: A retrospective study. Lancet Neurol, 10(9), 785-796. doi:10.1016/s1474-4422(11)70156-9 Nation, D. A., Delano-Wood, L., Bangen, K. J., Wierenga, C. E., Jak, A. J., Hansen, L. A., . . . Bondi, M. W. (2012). Antemortem pulse pressure elevation predicts cerebrovascular disease in autopsy-confirmed Alzheimer's disease. J Alzheimers Dis, 30(3), 595-603. doi:10.3233/jad-2012-111697 Nelson, P. T., Alafuzoff, I., Bigio, E. H., Bouras, C., Braak, H., Cairns, N. J., . . . Beach, T. G. (2012). Correlation of Alzheimer disease neuropathologic changes with cognitive status: 82 Neurocognitive profiles of AD variants a review of the literature. J Neuropathol Exp Neurol, 71(5), 362-381. doi:10.1097/NEN.0b013e31825018f7 Noh, Y., Jeon, S., Lee, J. M., Seo, S. W., Kim, G. H., Cho, H., . . . Na, D. L. (2014). Anatomical heterogeneity of Alzheimer disease: based on cortical thickness on MRIs. Neurology, 83(21), 1936-1944. doi:10.1212/wnl.0000000000001003 Peter, J., Abdulkadir, A., Kaller, C., Kummerer, D., Hull, M., Vach, W., & Kloppel, S. (2014). Subgroups of Alzheimer's disease: stability of empirical clusters over time. J Alzheimers Dis, 42(2), 651-661. doi:10.3233/jad-140261 Petersen, R. C. (2004). Mild cognitive impairment as a diagnostic entity. J Intern Med, 256(3), 183-194. doi:10.1111/j.1365-2796.2004.01388.x Petersen, R. C. (2007). Mild cognitive impairment. Continuum Lifelong Learning Neurol, 13(2), 13-36. Petersen, R. C., Aisen, P. S., Beckett, L. A., Donohue, M. C., Gamst, A. C., Harvey, D. J., . . . Weiner, M. W. (2010). Alzheimer's Disease Neuroimaging Initiative (ADNI): clinical characterization. Neurology, 74(3), 201-209. doi:10.1212/WNL.0b013e3181cb3e25 Petersen, R. C., & Morris, J. C. (2005). Mild cognitive impairment as a clinical entity and treatment target. Arch Neurol, 62(7), 1160-1163; discussion 1167. doi:10.1001/archneur.62.7.1160 Petersen, R. C., Parisi, J. E., Dickson, D. W., Johnson, K. A., Knopman, D. S., Boeve, B. F., . . . Kokmen, E. (2006). Neuropathologic features of amnestic mild cognitive impairment. Arch Neurol, 63(5), 665-672. doi:10.1001/archneur.63.5.665 Petersen, R. C., Stevens, J. C., Ganguli, M., Tangalos, E. G., Cummings, J. L., & DeKosky, S. T. (2001). Practice parameter: early detection of dementia: mild cognitive impairment (an 83 Neurocognitive profiles of AD variants evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology, 56(9), 1133-1142. Petersen, R. C., Wiste, H. J., Weigand, S. D., Rocca, W. A., Roberts, R. O., Mielke, M. M., . . . Jack, C. R., Jr. (2016). Association of Elevated Amyloid Levels With Cognition and Biomarkers in Cognitively Normal People From the Community. JAMA Neurol, 73(1), 85-92. doi:10.1001/jamaneurol.2015.3098 Petersen, R. C. S., G.E.; Waring, S.C.; Ivnik, R.J.; Tangalos, E.G.; Kokmen, E. (1999). Mild cognitive impairment: clinical characterization and outcome. Archives of Neurology, 56(3), 303-308. Qiu, C., Winblad, B., & Fratiglioni, L. (2005). The age-dependent relation of blood pressure to cognitive function and dementia. The Lancet Neurology, 4(8), 487-499. Rahimi, J., & Kovacs, G. G. (2014). Prevalence of mixed pathologies in the aging brain. In Alzheimers Res Ther (Vol. 6). Raz, N., & Rodrigue, K. M. (2006). Differential aging of the brain: patterns, cognitive correlates and modifiers. Neurosci Biobehav Rev, 30(6), 730-748. doi:10.1016/j.neubiorev.2006.07.001 Reijmer, Y. D., van den Berg, E., Dekker, J. M., Nijpels, G., Stehouwer, C. D., Kappelle, L. J., & Biessels, G. J. (2012). Development of vascular risk factors over 15 years in relation to cognition: the Hoorn Study. J Am Geriatr Soc, 60(8), 1426-1433. doi:10.1111/j.1532- 5415.2012.04081.x Reiman, E. M., Chen, K., Liu, X., Bandy, D., Yu, M., Lee, W., . . . Caselli, R. J. (2009). Fibrillar amyloid-beta burden in cognitively normal people at 3 levels of genetic risk for 84 Neurocognitive profiles of AD variants Alzheimer's disease. Proc Natl Acad Sci U S A, 106(16), 6820-6825. doi:10.1073/pnas.0900345106 Reinvang, I., Espeseth, T., & Westlye, L. T. (2013). APOE-related biomarker profiles in non- pathological aging and early phases of Alzheimer's disease. Neurosci Biobehav Rev, 37(8), 1322-1335. doi:10.1016/j.neubiorev.2013.05.006 Rey, A. (1941). L'examen psychologique dans les cas d'encéphalopathie traumatique. Archives de Psychologie, 28, 215-285. Rosenthal, S. L., & Kamboh, M. I. (2014). Late-Onset Alzheimer's Disease Genes and the Potentially Implicated Pathways. In Curr Genet Med Rep (Vol. 2, pp. 85-101). Scheltens, N. M., Galindo-Garre, F., Pijnenburg, Y. A., van der Vlies, A. E., Smits, L. L., Koene, T., . . . van der Flier, W. M. (2015). The identification of cognitive subtypes in Alzheimer's disease dementia using latent class analysis. J Neurol Neurosurg Psychiatry. doi:10.1136/jnnp-2014-309582 Schneider, J. A., Arvanitakis, Z., Bang, W., & Bennett, D. A. (2007). Mixed brain pathologies account for most dementia cases in community-dwelling older persons. Neurology, 69(24), 2197-2204. doi:10.1212/01.wnl.0000271090.28148.24 Schneider, J. A., Arvanitakis, Z., Leurgans, S. E., & Bennett, D. A. (2009). The neuropathology of probable Alzheimer disease and mild cognitive impairment. Ann Neurol, 66(2), 200- 208. doi:10.1002/ana.21706 Schonheit, B., Zarski, R., & Ohm, T. G. (2004). Spatial and temporal relationships between plaques and tangles in Alzheimer-pathology. Neurobiol Aging, 25(6), 697-711. doi:10.1016/j.neurobiolaging.2003.09.009 85 Neurocognitive profiles of AD variants Schwarz, C., Fletcher, E., DeCarli, C., & Carmichael, O. (2009). Fully-automated white matter hyperintensity detection with anatomical prior knowledge and without FLAIR. Inf Process Med Imaging, 21, 239-251. Shaw, L. M., Vanderstichele, H., Knapik-Czajka, M., Clark, C. M., Aisen, P. S., Petersen, R. C., . . . Trojanowski, J. Q. (2009). Cerebrospinal Fluid Biomarker Signature in Alzheimer’s Disease Neuroimaging Initiative Subjects. Ann Neurol, 65(4), 403-413. doi:10.1002/ana.21610 Shimada, H., Kitamura, S., Shinotoh, H., Endo, H., Niwa, F., Hirano, S., . . . Higuchi, M. (2017). Association between Abeta and tau accumulations and their influence on clinical features in aging and Alzheimer's disease spectrum brains: A [11C]PBB3-PET study. Alzheimers Dement (Amst), 6, 11-20. doi:10.1016/j.dadm.2016.12.009 Skillback, T., Farahmand, B. Y., Rosen, C., Mattsson, N., Nagga, K., Kilander, L., . . . Zetterberg, H. (2015). Cerebrospinal fluid tau and amyloid-beta1-42 in patients with dementia. Brain, 138(Pt 9), 2716-2731. doi:10.1093/brain/awv181 Sperling, R. A., Aisen, P. S., Beckett, L. A., Bennett, D. A., Craft, S., Fagan, A. M., . . . Phelps, C. H. (2011). Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7(3), 280-292. doi:10.1016/j.jalz.2011.03.003 Spreen, O., & Strauss, E. (1998). A compendium of neuropsychological tests (2nd ed.). New York: Oxford University Press. Stern, Y. (2012). Cognitive reserve in ageing and Alzheimer's disease. Lancet Neurol, 11(11), 1006-1012. doi:10.1016/s1474-4422(12)70191-6 86 Neurocognitive profiles of AD variants Stopford, C. L., Snowden, J. S., Thompson, J. C., & Neary, D. (2008). Variability in cognitive presentation of Alzheimer's disease. Cortex, 44(2), 185-195. doi:10.1016/j.cortex.2005.11.002 Strassburger, T. L., Lee, H.-C., Daly, E. M., Szczepanik, J., Krasuski, J. S., Mentis, M. J., . . . Alexander, G. E. (1997). Interactive effects of age and hypertension on volumes of brain structures. Stroke, 28(7), 1410-1417. Tabert, M. H., Albert, S. M., Borukhova-Milov, L., Camacho, Y., Pelton, G., Liu, X., . . . Devanand, D. P. (2002). Functional deficits in patients with mild cognitive impairment: prediction of AD. Neurology, 58(5), 758-764. Vardy, E. R., Ford, A. H., Gallagher, P., Watson, R., McKeith, I. G., Blamire, A., & O'Brien, J. T. (2013). Distinct cognitive phenotypes in Alzheimer's disease in older people. Int Psychogeriatr, 25(10), 1659-1666. doi:10.1017/s1041610213000914 Vos, S. J., Xiong, C., Visser, P. J., Jasielec, M. S., Hassenstab, J., Grant, E. A., . . . Fagan, A. M. (2013). Preclinical Alzheimer's disease and its outcome: a longitudinal cohort study. Lancet Neurol, 12(10), 957-965. doi:10.1016/s1474-4422(13)70194-7 Ward, A., Arrighi, H. M., Michels, S., & Cedarbaum, J. M. (2012). Mild cognitive impairment: disparity of incidence and prevalence estimates. Alzheimers Dement, 8(1), 14-21. doi:10.1016/j.jalz.2011.01.002 Warkentin, S., Ohlsson, M., Wollmer, P., Edenbrandt, L., & Minthon, L. (2004). Regional cerebral blood flow in Alzheimer's disease: classification and analysis of heterogeneity. Dement Geriatr Cogn Disord, 17(3), 207-214. doi:10.1159/000076358 87 Neurocognitive profiles of AD variants Weintraub, S., Wicklund, A. H., & Salmon, D. P. (2012). The neuropsychological profile of Alzheimer disease. Cold Spring Harb Perspect Med, 2(4), a006171. doi:10.1101/cshperspect.a006171 Weller, R. O., Boche, D., & Nicoll, J. A. R. (2009). Microvasculature changes and cerebral amyloid angiopathy in Alzheimer's disease and their potential impact on therapy. Acta Neuropathologica, 118, 87-102. Weller, R. O., Subash, M., Preston, S. D., Mazanti, I., & Carare, R. O. (2008). Perivascular drainage of amyloid-ß peptides from the brain and its failure in cerebral amyloid angiopathy and Alzheimer's disease. Brain Pathology, 18, 253-266. Wilcock, D. M. (2014). Neuroinflammatory phenotypes and their roles in Alzheimer's disease. Neurodegener Dis, 13(2-3), 183-185. doi:10.1159/000354228 Winblad, B., Palmer, K., Kivipelto, M., Jelic, V., Fratiglioni, L., Wahlund, L. O., . . . Petersen, R. C. (2004). Mild cognitive impairment--beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med, 256(3), 240-246. doi:10.1111/j.1365-2796.2004.01380.x Wirth, M., Villeneuve, S., Haase, C. M., Madison, C. M., Oh, H., Landau, S. M., . . . Jagust, W. J. (2013). Associations between Alzheimer disease biomarkers, neurodegeneration, and cognition in cognitively normal older people. JAMA Neurol, 70(12), 1512-1519. doi:10.1001/jamaneurol.2013.4013 Wiseman, R. M., Saxby, B. K., Burton, E. J., Barber, R., Ford, G. A., & O'Brien, J. T. (2004). Hippocampal atrophy, whole brain volume, and white matter lesions in older hypertensive subjects. Neurology, 63, 1892-1897. 88 Neurocognitive profiles of AD variants Zade, D., Beiser, A., McGlinchey, R., Au, R., Seshadri, S., Palumbo, C., . . . Milberg, W. (2010). Interactive effects of apolipoprotein E type 4 genotype and cerebrovascular risk on neuropsychological performance and structural brain changes. J Stroke Cerebrovasc Dis, 19(4), 261-268. doi:10.1016/j.jstrokecerebrovasdis.2009.05.001 Zlokovic, B. V. (2011). Neurovascular pathways to neurodegeneration in Alzheimer's disease and other disorders. Nature Reviews Neuroscience, 12(12), 723-738.

Abstract (if available)

Abstract Studies have identified Alzheimer’s disease (AD) variants characterized by distinct clinical and pathological features using a variety of grouping methods. A complete neurocognitive profile is important for Alzheimer’s disease diagnosis and prognosis. The present study sought to identify AD variants through cluster analysis of both in vivo AD biomarkers and neuropsychological measures, and to compare variants on longitudinal cognitive decline and brain atrophy. ❧ Alzheimer’s Disease Neuroimaging Initiative participants (N=976, non-demented subsample) were subjected to hierarchical cluster analysis using baseline cognition (Boston Naming Test, Rey’s Auditory Verbal Learning Test, the Logical Memory, Categorical Fluency, Trail Making Test A & B) and cerebrospinal fluid (amyloid β 1-42 and phosphorylated-tau) biomarkers. Analysis of covariance (ANCOVA) was used to examine group differences in cognitive performance (raw and age-normed) and biomarker status. Longitudinal (baseline, 1, 2 year) cognitive decline (n=747) and brain atrophy (hippocampal, n=497

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Long-term blood pressure variability across the clinical and biomarker spectrum of Alzheimer’s disease

PDF

Affective neuropsychiatric symptoms and neural connectivity in the early stages of Alzheimer’s disease

PDF

Brainstem structural integrity in the progression of Alzheimer's disease

PDF

Indexing cerebrovascular resistance in cognitive decline and Alzheimer's disease

PDF

The role of the locus coeruleus in Alzheimer’s disease and cerebrovascular function: insights from neuroimaging, neuropsychology, and biofluid markers

PDF

Role of oxidative stress in age-associated mild cognitive impairment and Alzheimer's disease

PDF

Accuray of subjective cognitive complaints in a longitudinal context: the effect of depression and dementia status

PDF

Effects of AT-1 receptor blockers on cognitive decline and Alzheimer's disease

PDF

The role of blood pressure variability in cognitive decline, cerebrovascular disease and Alzheimer’s disease

PDF

Arterial stiffness and resting-state functional connectivity in older adults

PDF

Untreated hyperglycemia associated with tau pathology and worse cognitive performance in older adults

PDF

Alzheimer’s disease: dysregulated genes, ethno-racial disparities, and environmental pollution

PDF

Blood-brain barrier pathophysiology in cognitive impairment and injury

PDF

Insulin sensitivity in cognition, Alzheimer's disease and brain aging

PDF

Neuroinflammation and ApoE4 genotype in at-risk female aging: implications for Alzheimer's disease

PDF

Development of biomarker profiles for early detection of women with an at-risk for Alzheimer's disease phenotype

PDF

Using neuroinformatics to identify genomic and proteomic markers of suboptimal aging and Alzheimer's disease

PDF

Comparison of participant and study partner predictions of cognitive impairment in the Alzheimer's disease neuroimaging initiative 3 study

PDF

The contribution of periascular spaces to Parkinson's disease pathophysiology

PDF

A virtual reality exergaming system to enhance brain health in older adults at risk for Alzheimer’s disease

Asset Metadata

Creator Blanken, Anna Emilia (author)

Core Title Longitudinal neurocognitive profiles of empirically-derived Alzheimer’s disease variants

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Publication Date 09/29/2017

Defense Date 04/13/2017

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

Tag Alzheimer's disease,biomarkers,clinicopathologic,heterogeneity,mild cognitive impairment,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nation, Daniel Addison (committee chair), Mather, Mara (committee member), Saxbe, Darby (committee member)

Creator Email aeblanken1314@gmail.com,blanken@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-439799

Unique identifier UC11263859

Identifier etd-BlankenAnn-5801.pdf (filename),usctheses-c40-439799 (legacy record id)

Legacy Identifier etd-BlankenAnn-5801.pdf

Dmrecord 439799

Document Type Thesis

Rights Blanken, Anna Emilia

Type texts

Source 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 a...

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