University of Southern California Dissertations and Theses

An evaluation of risk predictors for disease progression in Veteran Affairs patients with chronic hepatitis C

(USC Thesis Other)

An evaluation of risk predictors for disease progression in Veteran Affairs patients with chronic hepatitis C

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content AN EVALUATION OF RISK PREDICTORS FOR DISEASE PROGRESSION IN VETERAN AFFAIRS PATIENTS WITH CHRONIC HEPATITIS C by Tara Matsuda A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (PHARMACEUTICAL ECONOMICS AND POLICY) August 2016 i DEDICATION This dissertation is dedicated to my family, friends, and loved ones. To my mom and grandma, thank you for believing in me and supporting me, and always pushing me to be my best self. And to my little brother, thanks for keeping me grounded and always trying to be worthy of the title ‘big sis’. And special thanks to my partner in life for the past 8 years, Yi. You have been my rock throughout the years in keeping me sane and grounded, giving me the strength and determination to get through this program and finish this dissertation. You never gave up on me and were always there to lift me out of the difficult moments. ii ACKNOWLEDGEMENTS I’d like to acknowledge and give much thanks to my advisor and dissertation chair, Dr. Jeffrey McCombs. You helped me navigate through the seemingly never-ending experience of being a PhD graduate student and also as a person dealing with everything that life throws at them. You were always a source of guidance, support, and encouragement that I will always be grateful for. A special thanks to Dr. Ivy Tonnu-Mihara for being another pillar of support both academically and personally. And of course much thanks in providing access to the VA data and helping to get through all the red tape and requirements that come with government data. I would also like to thank Pat Hines; Drs. Tim Juday, Anupama Kalsekar, Gilbert L’Italien, Sammy Saab, and Yong Yuan for supporting the project and their helpful comments and ideas in these studies. I had such great access to a wide set of expertise ranging from coding to statistical analysis to clinical knowledge. Thank you as well to Dr. Mei-Hsuan Lee for her collaboration in my second paper and the access to her study data. And much thanks to Dr. Steven Fox and Justin McGinnis in collaboration with the fifth paper. And last but not least, I would like to thank the members of my proposal and dissertation committees: Drs. Jeffrey McCombs, Ivy Tonnu-Mihara, Steven Fox, Geoffrey Joyce and Jason Doctor for their input and suggestions. It was very helpful to get both the inside and outside perspectives on the weaknesses of my work, and to make sure it was sound from both a clinical and statistical viewpoint. iii TABLE OF CONTENTS DEDICATION ............................................................................................................................... I ACKNOWLEDGEMENTS ........................................................................................................ II LIST OF TABLES ...................................................................................................................... VI LIST OF FIGURES ................................................................................................................. VIII ABSTRACT ................................................................................................................................. IX CHAPTER 1 : INTRODUCTION ............................................................................................... 1 1.1. STATEMENT OF PROBLEM ...................................................................................................................1 1.2 RESEARCH OBJECTIVE .........................................................................................................................2 1.3 SIGNIFICANCE ......................................................................................................................................2 CHAPTER 2 : BACKGROUND AND LITERATURE REVIEW .......................................... 3 2.1 EPIDEMIOLOGY ....................................................................................................................................3 2.2 NATURAL HISTORY ..............................................................................................................................4 2.3 MORBIDITY AND MORTALITY .............................................................................................................4 2.4 BURDEN OF DISEASE ............................................................................................................................5 2.5 PHARMACOLOGIC TREATMENT FOR HCV ...........................................................................................6 2.6 PRIOR WORK AT DOCUMENTING PREDICTORS .....................................................................................9 CHAPTER 3 : THE RISK OF LONG-TERM MORBIDITY AND MORTALITY IN CHRONIC HEPATITIS C PATIENTS: RESULTS FROM AN ANALYSIS OF DATA FROM A DEPARTMENT OF VETERANS AFFAIRS CLINICAL REGISTRY .............. 11 3.1 ABSTRACT ..........................................................................................................................................11 3.2 BACKGROUND ....................................................................................................................................12 3.3 METHODS ...........................................................................................................................................14 Data .....................................................................................................................................................14 Sample Selection Criteria ...................................................................................................................15 Primary and Secondary Outcomes ......................................................................................................15 Statistical Methods ..............................................................................................................................16 3.4 RESULTS .............................................................................................................................................17 Descriptive Statistics ...........................................................................................................................17 Absolute Risk of Liver-Related Events and Death ..............................................................................18 Predictors of Liver-Related Events .....................................................................................................21 Sensitivity Analysis: Impact of Fibrosis Stage ...................................................................................23 3.5 DISCUSSION ........................................................................................................................................25 Limitations ...........................................................................................................................................27 3.6 REFERENCES ......................................................................................................................................30 CHAPTER 4 : EXTERNAL VALIDATION OF THE RISK-PREDICTION MODEL FOR HEPATOCELLULAR CARCINOMA [HCC] FROM THE REVEAL HCV STUDY ....... 35 4.1 ABSTRACT ..........................................................................................................................................35 4.2 INTRODUCTION ..................................................................................................................................36 4.3 DATA AND METHODS .........................................................................................................................38 Data .....................................................................................................................................................38 Statistical analysis ...............................................................................................................................39 4.4 RESULTS .............................................................................................................................................40 iv 4.5 DISCUSSION ........................................................................................................................................46 Limitations ...........................................................................................................................................47 4.6 REFERENCES ......................................................................................................................................51 CHAPTER 5 : FIVE LABORATORY TESTS PREDICT PATIENT RISK AND TREATMENT RESPONSE IN HEPATITIS C: VA DATA FROM 1999-2010 .................. 54 5.1 ABSTRACT ..........................................................................................................................................54 5.2 INTRODUCTION ..................................................................................................................................55 5.3 METHODS ...........................................................................................................................................57 Data .....................................................................................................................................................57 Sample Selection Criteria ...................................................................................................................58 Primary and Secondary Outcomes ......................................................................................................58 Statistical Methods ..............................................................................................................................59 5.4 RESULTS .............................................................................................................................................61 Descriptive Statistics ...........................................................................................................................61 Predictive Laboratory Tests ................................................................................................................63 5.5 DISCUSSION ........................................................................................................................................68 5.5 Limitations .....................................................................................................................................70 5.5 REFERENCES ......................................................................................................................................72 CHAPTER 6 : USING THE FIB-4 SCORE TO MONITOR MORBIDITY AND MORTALITY RISK IN CHRONIC HEPATITIS C PATIENTS ......................................... 75 6.1 ABSTRACT ..........................................................................................................................................75 6.2 INTRODUCTION ..................................................................................................................................76 6.3 METHODS ...........................................................................................................................................79 Data .....................................................................................................................................................79 Sample Selection Criteria ...................................................................................................................80 Patient Outcomes ................................................................................................................................81 Statistical Methods ..............................................................................................................................81 6.4 RESULTS .............................................................................................................................................82 Descriptive Statistics ...........................................................................................................................82 Predictors of All-Cause Mortality .......................................................................................................85 Predictors of Morbidity .......................................................................................................................89 Sensitivity Analyses .............................................................................................................................92 6.5 DISCUSSION ........................................................................................................................................93 Limitations ...........................................................................................................................................95 6.6 REFERENCES ......................................................................................................................................97 CHAPTER 7 : THE IMPACT OF DELAYED HEPATITIS C VIRAL LOAD SUPPRESSION ON PATIENT RISK: HISTORICAL EVIDENCE FROM THE VETERANS ADMINISTRATION ......................................................................................... 103 7.1 ABSTRACT ........................................................................................................................................103 7.2 INTRODUCTION ................................................................................................................................105 7.3 METHODS .........................................................................................................................................106 Design ...............................................................................................................................................106 Data ...................................................................................................................................................106 Sample Selection Criteria .................................................................................................................107 Patient Outcomes ..............................................................................................................................107 Statistical Methods ............................................................................................................................110 7.4 RESULTS ...........................................................................................................................................111 v Descriptive Statistics .........................................................................................................................111 Impacting of Delaying Viral Load Suppression on Morbidity [Composite Event] ..........................113 Impact of Delayed Viral Load Suppression on Mortality .................................................................115 Other Risk Factors of Interest ...........................................................................................................116 7.5 DISCUSSION ......................................................................................................................................118 Limitations .........................................................................................................................................121 7.6 REFERENCES ....................................................................................................................................124 CHAPTER 8 : INTERNAL VALIDATION OF RISK PREDICTION MODEL DEVELOPED IN VA POPULATION .................................................................................... 129 8.1 ABSTRACT ........................................................................................................................................129 8.2 INTRODUCTION ................................................................................................................................130 8.3 METHODS .........................................................................................................................................132 Data ...................................................................................................................................................132 Statistical analysis .............................................................................................................................133 8.4 RESULTS ...........................................................................................................................................134 8.5 DISCUSSION ......................................................................................................................................138 Limitations .........................................................................................................................................139 8.6 REFERENCES ....................................................................................................................................141 CHAPTER 9 : CONCLUSIONS ............................................................................................. 145 BIBLIOGRAPHY ..................................................................................................................... 148 APPENDIX 1: LIST OF DIAGNOSES FOR LIVER-RELATED HOSPITALIZATION OUTCOME ............................................................................................................................... 163 vi LIST OF TABLES Table 3-1 Patient Characteristics .................................................................................................. 18 Table 3-2 Absolute Event Risk by Risk Group ............................................................................ 19 Table 3-3 Impact of Viral Clearance and Other Risk Factors on the Risk of Late-Stage Liver Events .................................................................................................................................... 21 Table 3-4 Impact of Viral Clearance and Other Risk Factors on the Risk of Secondary Outcomes ............................................................................................................................................... 23 Table 3-5 Impact of Viral Clearance on the Risk of Late-Stage Liver Events Adjusting for Fibrosis Stage ........................................................................................................................ 24 Table 4-1 Baseline Descriptive Statistics Of REVEAL-HCV And VA Cohorts ......................... 40 Table 5-1 Patient Characteristics .................................................................................................. 62 Table 5-2 Laboratory Data ............................................................................................................ 63 Table 5-3 Impact of Drug Therapy and Abnormal Laboratory Values on the Risk of Liver- Related Adverse Events and Death ....................................................................................... 65 Table 5-4 Impact of Abnormal Laboratory Values on the Risk of Adverse Events ..................... 66 Table 6-1 Descriptive Statistics of Patient Population ................................................................. 83 Table 6-2 Impact of Risk Factors on the Risk of Death ............................................................... 86 Table 6-3 Impact of Risk Factors on Morbidity [Composite Event] ............................................ 89 Table 6-4 Sensitivity Analysis: Comparison of FIB4 and APRI .................................................. 93 Table 7-1 List of Study Related Diagnoses and Procedure Codes ............................................. 108 Table 7-2 Descriptive Statistics of the Study Population ........................................................... 112 Table 7-3 Impact of Early and Late Viral Load Suppression on Morbidity Risk [Composite Event][Adjusted Hazard Ratios and 95% Confidence Intervals] ........................................ 114 Table 7-4 Impact of Early and Late Viral Load Suppression on Risk of Death [Adjusted Hazard Ratios and 95% Confidence Intervals] ............................................................................... 117 Table 8-1 Risk Factors Included In Model ................................................................................. 133 Table 8-2 Risk Factors For Base Model (Hazard Ratio) ............................................................ 134 vii Table 8-3 Risk Factors For Base Model Plus FIB-4 Score (Hazard Ratio) ................................ 135 Table 8-4 AUROC For Base Model ........................................................................................... 136 Table 8-5 AUROC For Base Model Plus FIB-4 Score ............................................................... 137 Table 8-6 Mean AUROC Of Base Model And Base Model Plus FIB-4 .................................... 137 viii LIST OF FIGURES Figure 4-1 Distribution Of Risk Scores, By Cohort ..................................................................... 43 Figure 4-2 Cumulative Risk Of HCC, Stratified By Risk Score Categories. Low-Risk: 0-7 Risk Score; Medium-Risk: 8-15 Risk Score; High-Risk: 16-22 Risk Score ................................. 43 Figure 4-3 Cumulative Risk Of HCC, Stratified By Viral Load. High: Viral Load>20,000 IU; Low: Viral Load ≤ 20,000 IU; Und: Undetected Viral Load (<25 IU) ................................ 44 Figure 4-4 Cumulative Risk Of HCC, Stratified By Genotype (GTP) ......................................... 45 Figure 4-5 Receiver Operator Characteristic (ROC) Curve ......................................................... 45 Figure 8-1 Distribution of HCV prevalence, by age group ........................................................ 131 ix ABSTRACT Hepatitis C virus (HCV) liver complications represent a substantial public health burden. There are several challenges exist to the effective clinical management of recognized HCV infections. HCV is an asymptomatic disease that largely remains undiagnosed until relatively late in the course of the disease. While there are new effective treatments available, they come with very high price tags. Hence many insurance providers have enforced heavy restrictions on its use. However, it has been shown that treatment later in the course of symptomatic chronic HCV infections often has a limited impact on long-term patient outcomes. As only a fraction of Hepatitis C patients go on to have costly complications, there is a need for a parsimonious and predictive model to manage and aid in treatment decisions. As the largest provider of healthcare services to Hepatitis C patients, the Veteran Affairs (VA) is unique in the volume of HCV patients it deals with as well as its nation wide mandated electronic medical record keeping since 1999. As such, all of the papers in this dissertation will be using a database of HCV patients from the VA Clinical Case Registry (CCR). Paper 1 (Chapter 3), begins to explore what demographic and viral factors may be associated with morbidity and mortality in HCV patients using a Cox proportional hazards model. Paper 2 (Chapter 4) examines a published risk model for hepatocellular carcinoma (HCC) developed in a Taiwanese population and assesses its performance in a US veteran population using area under the receiving operator characteristic (AUROC) curve analysis. Paper 3 (Chapter 5) is a continuation to Paper 1, going on to explore if there are common laboratory tests associated with risk of morbidity or death. Paper 3 also looks at if there was a difference between if treatment was started before a patient's labs became abnormal or if initiated afterwards on effect on risk. While many insurance providers have put restriction on treatment x use dependent on liver biopsies, Paper 4 (Chapter 6) looks at the use of a non-invasive test for fibrosis, the FIB-4, and its effect on risk. Paper 5 (Chapter 7) combines some of objectives from Papers 3 and 4, looking at the impact of viral load suppression before and a after a critical FIB-4 score is reached. Paper 6 (Chapter 8) is an internal validation paper, looking at the models developed in Papers 1 and 4. A k-fold cross validation (k=15) method is used as well as AUROC curve analysis. 1 CHAPTER 1 : INTRODUCTION 1.1. STATEMENT OF PROBLEM Hepatitis C virus (HCV) is the cause of chronic Hepatitis C, an infection of the liver, that is the most common chronic blood-borne infection in the United States (US) (Chen & Morgan, 2006). It affects approximately 2.2-3% of the world’s population, equating to about 130-170 million people (Lavanchy, 2009). Though safer injection practices and the screening of blood for HCV antibodies have helped in the dramatic decrease in the incidence rate since its peak in 1992, demographic shifts have led to an increasing number of those with long time infections who may develop severe liver complications (Klevens, Hu, Jiles, & Holmberg, 2012). The CDC estimates that 75% of HCV infections are from persons born in the “baby boomer” generation of 1945- 1965, and as such, have issued a recommendation that all those born during those years be screened for HCV (Smith et al., 2012). Liver disease progression usually occurs over two to four decades, so it is likely that the numbers of those with HCV related liver complications such as decompensated cirrhosis or hepatocellular cancer (HCC) will increase (Seeff, 2009). However, there are also several clinical challenges to management of chronic HCV. HCV infection is generally asymptomatic until it progress further so a majority of HCV infected patients are not diagnosed until later in the course of their disease when they become symptomatic, or unless the infection is detected through unrelated tests. And if diagnosed while asymptomatic, patients may delay or refuse treatment on the basis that they do not believe they will become symptomatic. Current treatments are costly and have significant side effects, which further deters patients from starting treatment. However, treatment later in the course of the 2 disease, when the patients have become symptomatic often has a limited impact since the damage to the liver has already occurred. 1.2 RESEARCH OBJECTIVE The primary objective is to develop and validate a risk prediction model that will identify the risk of liver disease progression in HCV infected patients based on host and viral factors, using a national database of patients in the Veteran Affairs (VA) hospitals. Liver disease progression will be defined by the following endpoints: development of compensated cirrhosis, decompensated cirrhosis, HCC, liver transplantation, and death. A secondary objective would be to evaluate the use of a published risk prediction model using a VA population. 1.3 SIGNIFICANCE The significance of developing a risk prediction model for HCV infection is that it could serve as a new set of standard of practice that allows clinicians to discuss treatment options with patients based on a patient’s individualized risk profile. Patients will now be able to weight the risk of adverse events associated with treatment against their personal risk of developing liver complications. A predictive model and risk profile also gives clinicians the ability to focus on the patients who are predicted to have critical outcomes. There are many new drugs currently in the pipeline for HCV that may prove to be more effective and/or a safer adverse event profile, so it may be beneficial if patients with a low risk profile delayed HCV treatment for a few years. Predictive modeling to identify high-risk members is now becoming an important part of medical management. It gives providers and health organizations the ability to focus on this cohort and ensure they receive the appropriate care for better control over their outcomes. Retrospective data is also very useful with chronic HCV infection since the course of liver disease is such a long period. Prospective studies are costly and would take years if not decades to complete 3 CHAPTER 2 : BACKGROUND AND LITERATURE REVIEW 2.1 EPIDEMIOLOGY HCV is an RNA virus that is primarily transmitted through exposure to infected blood, with risk factors including blood transfusion, intravenous drug use, high-risk sexual activity, solid organ transplantation from an infected donor, occupational exposure, hemodialysis, household exposure, birth from an infected mother, and intranasal cocaine use (Chen & Morgan, 2006). In the past, HCV incidence experienced a large increase from the late 1960s to the early 1980s with an incidence in the US of 100-200 cases per 100,000, representing about 250,000- 500,000 new infections annually (Armstrong, Alter, McQuillan, & Margolis, 2000). However, from 1992 through 2002, incidence rates for acute hepatitis C decreased, and remained fairly constant from 2002 through 2009. The incidence rate for 2010 was 0.3 cases per 100,000 persons, representing an increase of approximately 6% since 2006. After adjusting for asymptomatic infections and underreporting, an estimated 17,000 new infections of HCV occurred in 2010 (Viral Hepatitis Surveillance – United States, 2010, 2010). The National Health and Nutrition Examination (NHANES) survey estimates chronic HCV prevalence to be 1.3% or 3.2 million people (Armstrong et al., 2006). However, the NHANES survey missed high-risk groups such as the incarcerated and the homeless; therefore, a conservative estimate of HCV prevalence in the US including these high-risk groups is 5.2 million people (Chak, Talal, Sherman, Schiff, & Saab, 2011). However, in the VA healthcare system, the prevalence of HCV is estimated to be much higher at 5.4%, due to an overrepresentation of high risk groups compared to the general population (Dominitz et al., 2005). 4 2.2 NATURAL HISTORY Approximately 15-25% of people infected with HCV spontaneously clear the infection (are only acutely infected), meaning that about 75-85% have persistent chronic HCV infection (Seeff, 2009). Those with chronic HCV are at risk for progressive liver disease, characterized by increasing degrees of fibrosis, followed by cirrhosis, first compensated then decompensated (marked by the development of ascites, upper gastrointestinal bleeding secondary to varices or portal hypertensive gastropathy, hepatorenal syndrome and hepatic encephalopathy), and HCC (Chen & Morgan, 2006; Seeff, 2009). While reports are variable, cirrhosis occurs in approximately 5-20% of infected people within 20 years (Seeff, 2009). And of those who develop cirrhosis, the annual risk of developing HCC is 1-4% (Chen & Morgan, 2006). As such, end stage liver disease caused by chronic HCV is the most common cause for liver transplantation, accounting for 25.6% of liver transplants in 2009 (Organ Proceurement and Transplantation Networn (OPTN) and Scientific Registrry of Transplant Recipients (SRTR). OPTN/SRTR 2011 Annual Data Report, 2011). This process generally occurs over two to four decades from the time of infection (Seeff, 2009). However, due to the aging of the “baby boomer” cohort, it is predicted that the prevalence of HCV complications will peak in 2030, and gradually decline by 2040 until 2060 due to aging and natural death (G. L. Davis, Albright, Cook, & Rosenberg, 2003). Compensated cirrhosis and HCC prevalence are projected to increase by over 84% from 2000 to 2030; and decompensated cirrhosis will increase over 124% (G. L. Davis, et al., 2003). The number of liver transplants is also expected to peak around 2030 at around 3,200 transplants (Rein et al., 2011). 2.3 MORBIDITY AND MORTALITY HCV is also associated with morbidity spanning a broad range of clinical conditions. 5 Extrahepatic manifestations include vasculitis, lymphomas, and insulin resistance. The most prominent manifestations are mixed cryoglobulinemia (MC) vasculitis, lymphoproliferative malignancies, sicca syndrome, rheumatoid arthritis (RA)–like polyarthritis, autoantibody production, and glucose disorders (Jacobson, Cacoub, Dal Maso, Harrison, & Younossi, 2010). A study looking at HCV mortality in the US between 1999 and 2007, found a statistically significant average annual age-adjusted mortality rate increase of 0.18 per 100,000 person per year (Ly et al., 2012). And from 2004 through 2008, the mortality rate of hepatitis C increased from 3.7 deaths per 100,000 population in 2004 to 4.7 deaths per 100,000 population in 2008 (Viral Hepatitis Surveillance – United States, 2010, 2010). In line with the projected increasing prevalence of liver complications, it is projected that liver-related deaths will increase by over 200% from 2000 to 2030 (G. L. Davis, et al., 2003). Without treatment, 379,600 or 13.1% of the approximately 2.9 million patients infected with chronic, pre-cirrhotic hepatitis C in 2005 are forecasted to have died by 2030; and 1,071,229 or 36.8% by 2060 (Rein, et al., 2011). 2.4 BURDEN OF DISEASE Direct cost associated with hepatitis C care in the US are estimated at $694–$1660 million per year. The liver complications that may follow chronic HCV infection are very costly, as is liver transplantation. HCC due to HCV are estimated to cost $140 million per year, and chronic liver diseases and cirrhosis at $1,421 million per year. The mean costs per person per year (PPPY) for HCC are estimated at $23,755–$44,200, variceal hemorrhage at $25,595, compensated cirrhosis at $585–$1,110, refractory ascites at $24,755, hepatic encephalopathy at $16,430, moderate chronic hepatitis C at $155, and mild chronic hepatitis C at $145 per year per person (in 2010 dollars). And the mean cost of liver transplantation was estimated at $201,110, with subsequent years cost about $37,535 (El Khoury, Klimack, Wallace, & Razavi, 2012). 6 A study using claims of commercially insured patients matched by propensity scores found that mean incremental cost of HCV infection was estimated at $23,406, primarily because of higher costs for ambulatory care ($6531), hospital services ($1827), and prescription drugs ($6935). They also found that the presence of HCV was associated with a significantly higher risk of hospitalization (McCombs, Yuan, Shin, & Saab, 2011). In a managed care settings, the adjusted all-cause costs have been estimated to be between $19,660-$20,961 per HCV PPPY, compared with $5,451-$9,979 per control PPPY; and so the incremental cost of HCV PPPY was between $10,000-$15,000 dollars (K. L. Davis, Mitra, Medjedovic, Beam, & Rustgi, 2011). Hospitalization was also found to be associated with HCV patients, with hospitalization occurred in 24% of HCV patients compared with 7% of controls (K. L. Davis, et al., 2011). Assuming that one-quarter of the estimated 3.5 million HCV-infected persons in the United States are diagnosed, and using the conservative incremental PPPY cost estimate of approximately $10,000 projects to an annual U.S. burden of $8-$9 billion per year (McAdam-Marx et al., 2011; Wong, McQuillan, McHutchison, & Poynard, 2000). It is also projected that of the indirect costs of chronic HCV infection, from 2010 to 2019, the societal cost of premature mortality for those younger than 65 would be $54.2 billion, and the cost of morbidity from disability related to decompensated cirrhosis and HCC would be $21.3 billion (Wong, et al., 2000). 2.5 PHARMACOLOGIC TREATMENT FOR HCV For a long time (2001-2011), the standard of care for all patients with chronic HCV was treatment is a combination of pegylated interferon alfa (PegIFN) and ribavirin (RBV). The goal of therapy regarded, as the “virological cure”, is attainment of a sustained virological response (SVR), defined as the absence of HCV RNAfrom serum by 24 weeks following discontinuation 7 of therapy. The optimal duration of treatment is based on the viral genotype, with HCV genotypes 1 and 4, a 48 week regimen is the standard, and a 24 week regimen for HCV genotypes 2 and 3 (Ghany, Strader, Thomas, & Seeff, 2009). The efficacy of HCV in terms of SVR rates, is also highly dependent on HCV genotype. In clinical trials, the average SVR rate for HCV genotype 1 and genotypes 2 or 3 were 42-46% and 76-82%, respectively (Yee, Currie, Darling, & Wright, 2006). However, in clinical settings, the SVR rates observed for HCV genotype 1 and genotypes 2 or 3, have ranged from 14-23.6% and 37-52%, respectively (Kramer, Kanwal, Richardson, Mei, & El-Serag, 2012). Side effects during therapy with PEG-IFN include: flu-like systemic symptoms; marrow suppression; psychiatric effects; autoimmune reactions; hair thinning and loss; visual disorders; fatigue; weight loss, hearing impairment; interstitial pneumonitis; pancreatitis; colitis; and exacerbation of inflammatory diseases such as psoriasis (Dienstag & McHutchison, 2006; Ghany, et al., 2009). Ribavirin contributes additional side effects, the most important of which is hemolytic anemia (Dienstag & McHutchison, 2006). Other side effects associated with ribavirin include mild lymphopenia, hyperuricemia, itching, rash, cough and nasal stuffiness (Ghany, et al., 2009). In registration trials, 10-14% of patients have discontinued therapy due to adverse events (Ghany, et al., 2009). The cost of PegIFN/RBV for HCV genotype 1 patients was about $18,507 and $12,632 (in 2001 dollars) for all other genotypes (Salomon, Weinstein, Hammitt, & Goldie, 2003). With these old treatments, the poor efficacy and the adverse event profile associated with treatment had caused low treatment rates in clinical settings. A study done by the VA found that of patients with no contraindications to HCV treatment, only 23% of HCV mono-infected patients received any treatment for HCV (Butt, McGinnis, Skanderson, & Justice, 2011). 8 Another study using data from 2000-2005 found that only16.5% of chronic HCV patients had any prescription for HCV treatment medication. And of the cohort of patients who did not receive treatment, 43% of patients had none of the contraindications to treatment (Kramer, et al., 2012). However, since 2011, and the start of this dissertation work, there have been a wave of new treatment options that have been approved. Telaprevir (Incivek®) and boceprevir (Victrelis®) were approved by the FDA in May 2011 as the first two HCV protease inhibitors, and for a while triple therapy of either telaprevir or boceprevir with PegIFN/RBV was the recommended treatment for HCV genotype 1 patients due to its improved SVR rates. Treatment duration for this triple therapy had similar lengths of treatment time, varying from 12 weeks of triple therapy with telaprevir followed by 12 or 24 weeks of PegIFN/RBV therapy. Or as with boceprevir, PegIFN/RBV is used for 4 weeks of lead in, and then triple therapy is pursued from 24 – 44 weeks (Yee et al., 2012). Clinical trials have found that average SVR rates in treatment naïve patients range between 63-75%; and in treatment experienced patients, SVR was achieved in 69 - 88 % of relapsers and 29 - 33 % of null responders (Yee, et al., 2012). And an indirect meta-analysis found no significant differences in SVR between groups on triple therapy with either treatment option (Cooper et al., 2012). However, these new drugs did come with similar side effects of anemia, nausea, diarrhea, anal-rectal discomfort, rash, pruritis, elevated uric acid and bilirubin levels (Shiffman & Esteban, 2012; Yee, et al., 2012). The cost for a 12 week course of treatment of TPV is just under $50,000; while the cost of BOC at $1,100 per week amounts to $26,000 to $48,400 for 24 to 44 weeks of treatment (Shiffman & Esteban, 2012). The real treatment game changers came in 2013, with the approval of sofosbovir (Sovaldi®) and simeprevir (Olysio®). They boasted shortened treatment duration times, the 9 removal of interferons for the majority of patients, and SVR rates averaging in the 80s to 90s. These new treatments did come with much higher price tags as well, with many regimens costing over $100,000 (Moore & Levitsky, 2014). More drugs have come out of the pipeline since then, with the approval of Ledipasvir-sofosbuvir (Harvoni®) and Ombitasvir-Paritaprevir-Ritonavir and Dasabuvir (Viekira Pak®) in 2014; and Ombitasvir-Paritaprevir-Ritonavir (Technivie®) and Daclatasvir (Daklinza®) in 2015. These drugs are all now part of the recommended treatment options approved by the American Association for the Study of Liver Diseases and the Infectious Diseases Society of America (AASLD/IDSA HCV Guidance Panel, 2015). Many patients now have the option of interferon free treatment, with shortened durations that a more tolerable, with evidence of high SVR rates (Guiterrez et al, 2015; McConachie et al, 2015). 2.6 PRIOR WORK AT DOCUMENTING PREDICTORS Several factors have been associated with chronic HCV progression and development of liver disease complications that included age, mode of transmission, duration of HCV infection, and alcohol use. A retrospective study with 460 HCV patients in the US suggested that factors increasing the likelihood of progression to stage 3 and 4 fibrosis include: age, longer duration of HCV infection, obesity, and history of diabetes mellitus, elevated AST, alpha fetoprotein, and the presence of grade 2 to 3 steatosis (Hu, Kyulo, Xia, Hillebrand, & Hu, 2009). A few prospective studies investigating predictors for HCV progression have also been done. In Taiwan, a study of 925 participants found that high HCV RNA levels, elevated serum ALT levels, and HCV genotype 1 was associated with a higher risk of HCC (Lee et al., 2010). A prospective study in the US with 483 participants found that platelet count, AST/ALT ratio, bilirubin, and albumin levels were predictive of clinical decompensation; and platelet count, AST/ALT ratio, and albumin levels were predictive of liver-related death or liver transplant 10 (Ghany et al., 2011). These studies considered unique sets of overlapping predictors, but it would be beneficial to consider a wider range of predictors that will encompass all considered past predictors. 11 CHAPTER 3 : THE RISK OF LONG-TERM MORBIDITY AND MORTALITY IN CHRONIC HEPATITIS C PATIENTS: RESULTS FROM AN ANALYSIS OF DATA FROM A DEPARTMENT OF VETERANS AFFAIRS CLINICAL REGISTRY Jeffrey McCombs, PhD 1 , Tara Matsuda, BA 1, 2 , Ivy Tonnu-Mihara, PharmD 2 , Sammy Saab, MD 3 , Patricia Hines, BA 4 , Gilbert L’Italien, PhD 4 , Timothy Juday, PhD 4 , Yong Yuan, PhD 4 1. Department of Clinical Pharmacy and Pharmaceutical Economics and Policy, School of Pharmacy, Leonard Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, CA, USA 2. Veterans Affairs Long Beach Healthcare System, Long Beach, CA, USA 3. Departments of Medicine and Surgery, David Geffen School of Medicine, University of California-Los Angeles, Los Angeles, CA. 4. Global Health Economics and Outcomes Research, Bristol-Myers Squibb, Plainsboro, NJ, USA The final version of this paper was published in JAMA Intern Med. on Feb 1, 2014. Further acknowledgement to Anupama Kalsekar, PhD from BMS for her scientific critique and Tim Morgan, M.D., from the VA in Long Beach for his support of the project. 3.1 ABSTRACT Objective: Develop risk prediction models for HCV patients treated in real world clinical practice. Method: In this observational cohort study, patients were selected from the Veterans Administration’s [VA] HCV clinical registry [CCR], which compiles electronic medical records data from 1999 to present. Selected HCV CCR patients had a detectable viral load (>25 IU/ml) and a recorded genotype in the six months prior to or following confirmation of diagnosis [index date]. Selecting patients with detectable viral load at baseline focused the study on patients who require treatment. Risk factors considered included genotype, race, age, gender, time to achieving an observed undetected viral load (VL) and time to first treatment. The primary 12 outcomes were time to death and time to a composite of liver-related clinical events. Secondary outcomes include the components of the composite clinical outcome. Outcomes were measured using a time-to-event format and were analyzed using Cox proportional hazards models. Results: 128,769 patients out of 360,857 unique patients registered in the HCV CCR database met all study inclusion criteria. Only 24.3% of study patients received treatment at any time and only 16.4% of treated patients achieved an undetectable viral load. Patients achieving an undetectable viral load reduced their risk of the composite clinical endpoint by 22.7% [p<0.0001] and the risk of death by 45% [p<0.0001]. Patients with genotype 2 are consistently at lower risk for all study outcomes relative to genotype 1. Patients with genotype 3 were consistently at higher risk relative to genotype 1 patients (p<0.01). Black patients were at lower risk for all late-stage liver events than white patients. Conclusion: Achieving an undetectable viral load requires treatment and achieving this goal significantly reduces the risk of liver-related events and death. 3.2 BACKGROUND Hepatitis C (HCV) affects approximately 130-170 million people worldwide 1, 2 and an estimated 3.2 million people in the United States (US). 3 This latter estimate may be low since high-risk groups such as the incarcerated and the homeless were not included in the data. Alternative estimates put the number of chronic HCV patients in the US at between 5.2 million and 7 million. 4 HCV patients are at risk of developing liver-related complications such as cirrhosis, liver failure and hepatocellular carcinoma (HCC). 1, 2, 5-7 Using data from the U.S. Veterans Administration [VA], Butt, Wang and Moore found that infection with HCV increased the risk of death by 37%. 8 HCV is also the leading indication for liver transplantation and the incidence of 13 HCC is increasing in the US. 9 Davis, Albright, Cook, and Rosenberg 10 used a simulation model to project that 13.1% of US HCV patients living in 2005 will die of liver-related causes by 2030. Rein, Wittenborn, Weinbaum, et al. extended this projection to 2060 at which time they estmate that 36.8% of the 2005 HCV cohort will have died due to liver-causes. 11 The impact of genotype and demographic factors on the clinical course of HCV may be significant. For example, genotype 1 is thought to be highly correlated with disease progression although Seeff 5 casts some doubt on this conclusion. Kallwitz, Layden-Almer, Dhamija, et al. found that BMI and Hispanic ethnicity were associate with disease progression, while African Americans had a lower rate of disease progression relative to white patients. 12 Sustained viral response (SVR) to treatment is associated with decreased liver-related morbidity and mortality. The REVEAL HCV study 13 found a significant association between undetectable VL and liver-related events in Taiwan. Van der Meer, Veldt, Feld, et al. found that the 36% of advanced hepatic fibrosis patients who achieved SVR had reduced all-cause mortality and reduced incidence of liver-related events compared to those who did not achieve SVR. 14 No previous studies have investigated a wide range of the risk factors associated with mortality and morbidity in a large real world cohort of patients at all levels of disease progression. The objective of this research is to use HCV RNA data to quantify the impact of viral load suppression on liver-related morbidity and overall mortality using a large cohort of HCV patients, including patients in the early stages of disease progression, while controlling for the impact of genotype, race, age, gender and other factors on morbidity and mortality. 14 3.3 METHODS DATA The data used in this study were taken from the Veterans Administration [VA] clinical case registry (CCR) system for HCV infected patients. The VA IRB approved the study. Potential HCV patients were identified by the presence of an HCV-related ICD-9 diagnosis code or a positive viral load assessment using the Hepatitis C Antibody Test, the Hepatitis C RIBA Test or the Qualitative Hepatitis C RNA Test. A local CCR coordinator then manually confirmed or rejected the patient for HCV/CCR inclusion. Upon this confirmation, all historical data from the patient’s electronic medical record (EMR) were pulled and added to the CCR. The VA EMR system was fully implemented in 1999 and the data period for this study covers the entire time period over which EMR data were available from all VA regions 1999-2010. 15 An intermediate patient-level analytic database was created consisting of summary variables for each month before and after the patient’s CCR enrollment [index date]. The following summary data were created: 1. Patient demographic data (age in months at baseline, gender, race, ethnicity): Race and ethnicity data were based on patient self-report. 2. The patient’s diagnostic profile was created consisting of monthly dichotomous variables reflecting the diagnoses recorded each month. 3. Monthly dichotomous variables were created for hospital admissions for any diagnosis and for liver-related diagnoses. 4. Prescription drug data were used to create monthly variables indicating when patients received HCV-related treatment [peginterferon alfa [2a or 2b], interferon alfa [2a or 2b], 15 interferon alfacon-1, boceprevir or telaprevir]. The use of ribavirin alone was not considered to be a drug therapy for HCV. 5. The objective of treatment is to suppress the patients HCV viral load to undetectable levels. A primary objective of this research was to document the impact of viral load suppression accounting for the temporal relationship between achieving an undetectable viral load and event date. To achieve this, we calculated the time to the first undetectable viral load test as our covariate of interest and defined our primary and secondary outcomes using time to event formats. This approach is less stringent than measuring time to sustained viral response [SVR], the gold standard for measuring treatment response. Time to SVR is significantly more difficult to calculate, requiring the determination of the time at which the patient maintained consistent viral load suppression for a minimum of 6 months following the termination of treatment. SAMPLE SELECTION CRITERIA In order to estimate the risk reduction achieved if patients achieved viral load suppression, all patients were screened for a detectable baseline HCV viral load (>25 IU/ml). Study patients were also screened for a recorded genotype within 6 months of their index date. Longitudinal data were then used to measure time-to-outcome events and estimate the impact of achieving an undetectable viral load and other factors on event risk. PRIMARY AND SECONDARY OUTCOMES HCV infected patients are at risk for progressive liver disease and related complications such as cirrhosis, liver failure, hepatocellular carcinoma (HCC) and death. 1, 2, 5-7 The primary outcomes specified for this analysis were all-cause mortality and a composite of newly 16 diagnosed cirrhosis [compensated and decompensated], HCC or a liver-related hospitalization. The time to the composite event was set at the earliest event date for any of the composite events. Because HCV infections commonly go undiagnosed or untreated until complications are observed, sensitivity analysis was conducted in which clinical composite variables were measured using a one-year post-index washout period during which the clinical events were not counted. The secondary outcomes included the individual elements of the clinical composite analyzed individually. Monthly dichotomous variables were created for the outcomes of the study based on recorded diagnostic codes [e.g., diagnosis of cirrhosis, etc.] and selected CPT-4 codes included in data from hospital admissions and outpatient services. Hospitalizations were defined as being liver-related if the primary diagnosis for the hospitalization was found on Appendix 1, building on the fact that all study patients have a positive HCV viral load. Cirrhosis and HCC outcomes were compiled by searching the inpatient, outpatient and problem lists for ICD-9 codes 571.5, 571.2, 571.6 and 155, 155.1, 155.2, respectively. Decompensated cirrhosis was defined as a diagnosis of cirrhosis and a diagnosis of hepatic coma [70.44, 71.71, 348.3, 348.31, 572.2], portal hypertension [572.3], hepatorenal syndrome [572.4], jaundice [782.4], ascites [789.59], or esophageal varices [ 456, 456.1, 456.2, 456.21] or an FIB-4 score > 3.25. The FIB-4 score can also be segmented into three categories which were found to correctly classify nearly 73% of liver biopsies and to have a positive predictive value to confirm the existence of significant fibrosis of 82.1% in a HCV infected cohort. 16 STATISTICAL METHODS The time-to-event variables for primary and secondary outcomes were analyzed using Cox proportional hazards models to test the correlation between potential predictors and study 17 endpoints. Time to first observed undetected viral load (VL) was included in the analyses as a time-dependent independent variable to measure the impact of viral load suppression on each primary and secondary liver-related event controlling for genotype, race, age, gender and other factors. Race and ethnicity were initially included as separate categories but the significant correlation between race and ethnicity in the VA sample resulted in only race being included into the final model specifications. The impact of a diabetes diagnosis at baseline and any hospital admission in the 6 months prior to the patient’s index date were included in our list of risk factors based on statistical significance. A sensitivity analysis was conducted using only those patients with a baseline FIB-4 score to test if the core results on the impact of viral load suppression, genotype and other factors were sensitive to controlling statistically for the patient’s baseline fibrosis level. 3.4 RESULTS DESCRIPTIVE STATISTICS The HCV/CCR data base contains 360, 857 unique patients from which a study population of 128,769 patients met all study inclusion criteria including a detectable viral load and genotype data at baseline. Only 24.3% of patients in the analytic sample received treatment at any time following HCV diagnosis while only 16.4% of treated patients achieved an undetectable viral load post-treatment [Table 3-1]. The average post-index period consisted of 6.1 years [SD=3]. The VA/HCV patients were predominately male of either white or black race [51.4% and 31.3%, respectively]. The mean age was 52 years [SD=6.9] and close to 80% of patients were genotype 1. Just over 42% of study sample patients had baseline data for their fibrosis stage at baseline [FIB-4 score] and only 19% had an FIB-4 score > 3.25 which is 18 correlated with a Metavir fibrosis stage of F3-F4 16 or an Ishak fibrosis stage of F4-F6. 17 The FIB-4 score was not used in the core analysis due to this high level of missing data. Instead, a sensitivity analysis was conducted using only those patients with baseline FIB-4 scores and the patient’s FIB-4 categories was entered as a potential risk factor. Table 3-1 Patient Characteristics N=128,769 Treatment Data % or SD Treated 31,284 24.3% Untreated 97,485 75.7% Achieved Undetectable Viral Load Overall [n=128,769] 5,180 4.0% While Under Treatment [n=31,284] 5,141 16.4% No treatment [n=97,485] 39 0 04% Patient Demographics Gender [Male] 124,980 97.06% Age [in years][mean, SD] 51.8 6.9 Post-Index Data [in years][mean, SD] 6.1 3.0 Race White 66,168 51.39% Black 40,239 31.25% Asian 168 0.13% Other 22,194 17.24% HCV Genotype 1 102191 79.36% 2 15113 11.74% 3 9851 7.65% other 1614 1.25% Pre-index Admission [6 months] 20938 16.26% Diabetes at baseline 15091 11.72% FIB-4 score [n=54420][mean, SD] 2.51 3.08 FIB-4 >3.25 10397 19.11% ABSOLUTE RISK OF LIVER-RELATED EVENTS AND DEATH Table 3-2 presents data on the absolute risk of the composite event and death across the risk factors of interest in this analysis. There were a total of 35,253 composite events and 15,458 deaths in our sample over a total of 734,829 person-years of data. Significantly higher event 19 rates and death rates were experienced by male patients, white patients, and patients with genotype 3. Treated patients and patients who achieve viral suppression had higher unadjusted composite event rates than their comparisons, but lower death rates. This may reflect the delays in therapy until patients become symptomatic. Table 3-2 Absolute Event Risk by Risk Group Cirrhosis Decomp. Cirrhosis Hospital Admission HCC Composite Death Number of Events: All Patients 17,926 [14.5%] 8,429 [6.6%] 28,730 [22.3%] 4,517 [3.5%] 35,253 [28.6%] 15,458 [12.0%] Event Incidence Rates [per 1,000 person-years] [p-values for comparisons across categories] Gender [p-value] Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p=0.0013] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] Male 27.7 [27.3-28.1] 12.0 [11.8-12.3] 46.6 [46.0-47.1] 6.4 [6.2-6.6] 62.7 [62.0-63.4] 21.4 [21.0-21.7] Female 17.8 [16.1-19.7] 5.4 [4.6-6.6] 40.5 [37.7*43.5] 1.4 [1.0-2.0 52.2 [48.9-55.6] 10.5 [9.3-12.0] Age at Diagnosis Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p<0.0001] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] Under 45 16.0 [15.3-16.9] 5.8 [5.3-6.3] 46.9 [45.5-48.4] 1.7 [1.4-1.9] 56.6 [55.0-58.2] 11.4 [10.8-12.1] 45-50 24.4 [23.7-25.1] 9.7 [9.3-10.1] 47.7 [46.7-48.7] 4.0 [3.7-4.3] 61.0 [59.8-62.2] 16.3 [15.8-16.9] 50-55 29.5 [28.8-30.3] 13.2 [12.7-13.6] 46.7 [45.8-47.7 7.1 [6.8-7.4] 63.4 [62.3-64.5] 21.9 [21.3-22.5] 55-60 34.9 [33.8-36.0] 16.3 [15.5-17.0] 46.2 [44.9-47.5] 9.7 [9.1-10.2] 67.1 [65.5-68.8] 25.9 [25.0-26.8] 60-65 39.0 [36.8-41.4] 17.2 [15.8-18.7] 42.5 [40.2-45.0] 11.8 [10.7-13.1] 67.6 [64.4-70.8] 32.4 [30.5-34.4] >65 31.9 [29.5-34.5] 14.4 [12.9-16.1] 36.2 [33.6-38.9] 12.8 [11.4-14.4] 57.7 [54.3-61.3] 57.7 [54.7-61.0] Race Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p<0.0001] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] White 33.8 [33.2-34.4] 15.6 [15.2-16.0] 54.9 [54.0-55.7] 6.9 [6.6-7.2] 73.3 [72.3-74.3] 22.2 [21.7-22.7] Black 18.9 6.9 42.7 5.3 55.6 15.8 20 [18.4-19.5] [6.6-7.3] [41.8-43.7] [5.0-5.6] [54.5-56.7] [15.3-16.4] Other 25.4 [24.5-26.3] 10.1 [9.5-10.6] 29.9 [28.9-30.9] 5.9 [5.5-6.3] 45.6 [44.4-46.9] 27.1 [26.2-28.0] Prior Hosp. Adm. Cirrhosis [p=0.0729] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p<0.0001] HCC p=0.553 Composite [p<0.0001] Death [p<0.0001] No prior admission 27.6 [27.2-28.1] 11.3 [11.1-11.6] 39.4 [38.9-40.0] 6.2 [6.0-6.4] 56.6 [56.0-57.3] 19.4 [19.0-19.7] Prior admission 26.3 [25.3-27.3] 14.3 [13.6-15.0] 89.6 [87.6-91.6] 6.1 [5.7-6.6] 97.8 [95.6- 100.0] 29.6 [28.6-30.6] HCV Genotype Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p<0.0001] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] 1 27.4 [27.0-27.9] 11.7 [11.4-12.0] 46.8 [46.2-47.4] 6.3 [6.1-6.5] 62.9 [62.2-63.7] 21.0 [20.6-21.4] 2 20.8 [19.8-21.8] 8.1 [7.5-8.7] 38.3 [36.9-39.7] 3.8 [3.4-4.2] 51.6 [49.9-53.3] 19.2 [18.3-20.2] 3 38.9 [37.1-40.7] 19.3 [18.2-20.5] 55.7 [53.6-57.9] 9.6 [8.8-10.5] 76.0 [73.4-78.8] 24.6 [23.3-26.0] other 24.3 [21.2-27.9] 10.8 [8.9-13.2] 42.7 [38.4-47.5] 4.8 [3.6-6.4] 56.5 [51.3-62.2] 20.2 [17.5-23.3] Diabetes at baseline Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p<0.0001] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] Yes 36.1 [34.8-37.5] 15.5 [14.7-16.4] 52.9 [51.3-54.7] 8.7 [8.1-9.4] 73.8 [71.7-76.0] 33.2 [32.0-34.5] No 26.3 [25.9-26.8] 11.3 [11.1-11.6] 45.5 [45.0-46.1] 5.9 [5.7-6.1] 61.0 [60.3-61.6] 19.5 [19.1-19.8] Treated Cirrhosis [p<0.0001] Decomp. Cirrhosis [p=0.0008] Hospital Admission [p<0.0001] HCC p=0.191 Composite [p<0.0001] Death [p<0.0001] Yes 36.3 [35.4-37.2] 10.7 [10.3-11.2] 45.0 [44.0-46.0] 5.1 [4.8-5.4] 73.3 [72.0-74.7] 12.0 [11.6-12.5] No 24.1[23.6- 24.5] 12.3 [12.0-12.6] 46.9 [46.3-47.5] 6.7 [6.4-6.9] 58.3 [57.6-59.1] 24.6 [24.2-25.1] Achieved VL=0 Cirrhosis [p<0.0001] Decomp. Cirrhosis [p<0.0001] Hospital Admission [p=0.1074] HCC [p<0.0001] Composite [p<0.0001] Death [p<0.0001] Yes 29.8 [28.0-31.7] 7.3 [6.5-8.2] 40.7 [38.6-43.0] 3.7 [3.2-4.4] 66.2 [63.2-69.2] 6.8 [6.0-7.7] No 27.3 [26.9-27.7] 12.1 [11.8-12.3] 46.7 [46.1-47.2] 6.4 [6.2-6.5] 62.2 [61.5-62.8] 21.8 [21.5-22.2] 21 PREDICTORS OF LIVER-RELATED EVENTS The factors associated with our primary outcomes are displayed in Table 3-3. Table 3-3 Impact of Viral Clearance and Other Risk Factors on the Risk of Late-Stage Liver Events Patient Characteristics Composite of Clinical Outcomes Death All Events One-Year Washout N=123,065 N=106,947 N=128,769 Number of Events [%] 35,253 [28.6%] 18,595 [17.4%] 15,458 [12.0%] Gender [Male] 1.11** 1.04-1.19 1.12** 1.02-1.21 1.58*** 1.38-1.80 Age [in years] 1.00** 1.00-1.00 1.00*** 1.00-1.00 1.06** 1.05-1.06 Race [vs White] Black 0.72*** 0.71-0.74 0.76*** 0.74-0.79 0.65*** 0.62-0.67 Other 0.65*** 0.63-0.68 0.61*** 0.58-0.63 1.20*** 1.16-1.25 Prior Admission [6 mo.] 1.60*** 1.56-1.65 1.55*** 1.49-1.60 1.73*** 1.66-1.79 HCV Genotype [vs 1] 2 0.77*** 0.74-0.80 0.77*** 0.73-0.80 0.80*** 0.76-0.84 3 1.11*** 1.07-1.16 1.10** 1.05-1.16 1.17*** 1.11-1.24 other 0.89* 0.80-0.98 0.91 0.81-1.03 0.96 0.83-1.11 Diabetes at baseline 1.22*** 1.18-1.27 1.25*** 1.20-1.30 1.57*** 1.50-1.63 Achieved VL=0 0.73*** 0.66-0.82 0.72*** 0.64-0.81 0.55*** 0.47-0.64 *p<0.05; **p<0.01; ***p<0.0001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. Viral Load Suppression and Genotype: Patients who achieved an undetectable viral load significantly reduced their risk of the composite clinical endpoint by 27% [H.R.= 0.73 (0.66- 0.82)] and their risk of death by 45% [H.R. = 0.55 (0.47-0.60)] relative to patients with a detectable viral load over their entire post-index period. The risk reduction associated with the 22 composite clinical endpoint measured after a one year washout period increased slightly to a reduction of 28% [H.R. = 0.72 (0.64-0.81). Patients with genotype 2 were consistently at lower marginal risk for liver-related events compared to patients with the more common genotype 1 controlling for viral load suppression and other risk factors. The risk reduction for the composite endpoint and the composite endpoint measured with one-year wash out were 23% [H.R. = 0.77 (0.74-0.80) and 23% [H.R. = 0.77 (0.73-0.80)] respectively. The risk of all-cause mortality for genotype 2 patients was reduced by 20% relative to genotype 1 patients [H.R. = 0.80 (0.76-0.84)]. Patients with genotype 3 were consistently at higher risk compare to patients with genotype 1. The estimates of marginal increased risk for those with genotype 3 ranged between 11% for the composite clinical endpoint [H.R. = 1.11 (1.07-1.16)] to a 17% increase in the risk of death [H.R. = 1.17 (1.11-1.24)]. Other genotypes were infrequent in the VA patient population and were collapsed into a single category for which the estimated hazard ratios were not generally significant. Patient Characteristics: Male gender significantly increased the risk of the composite clinical outcome by 11% [H.R. = 1.11 (1.04-1.19)] and the risk of death by 58% [H.R. = 1.58 (1.38-1.80)]. Each additional year of age increased the risk of death by 5.5%[H.R. = 1.06 (1.05- 1.06)] but only increased the risk of the composite outcome by less than 1% [H.R. = 1.0001 (1.0012-1.0016)]. Whites were consistently at higher risk for all late-stage liver events relative to blacks and other races. A diagnosis of diabetes at baseline and a hospital admission within 6 month prior to the index date were significantly associated with liver-related events. Secondary Outcomes: The risk prediction models for the individual late-staged liver events that comprise the composite clinical outcome are presented in Table 3-4. As with the 23 primary outcomes, achieving an undetectable viral load significantly reduced the risk of all clinical events. Other estimates were consistent with the results for the composite event. Table 3-4 Impact of Viral Clearance and Other Risk Factors on the Risk of Secondary Outcomes Patient Characteristics Cirrhosis Decompensated Cirrhosis Liver-Related Hospitalization Hepatocellular Carcinoma N=123,988 N=128,055 N=128,769 N=128,481 Number of Events 17,926 [14.5%] 8,429 [6.6%] 28,730 [22.3%] 4,517 [3.5%] Gender [Male] 1.35*** 1.21-1.50 1.81*** 1.50-2.19 1.09* 1.01-1.17 3.41*** 2.39-4.88 Age [in years] 1.03** 1.03-1.03 1.04** 1.04-1.04 0.99** 0.99-1.00 1.07** 1.06-1.07 Race [vs White] Black 0.54*** 0.52-0.56 0.42*** 0.40-0.45 0.74*** 0.72-0.76 0.73*** 0.68-0.78 Other 0.73*** 0.70-0.76 0.63*** 0.59-0.67 0.58*** 0.56-0.60 0.80*** 0.74-0.87 Pre-index Admission [6 mo.] 1.02 0.97-1.06 1.26*** 1.19-1.33 2.05*** 1.99-2.11 1.07 0.99-1.17 HCV Genotype [vs 1] 2 0.64*** 0.61-0.68 0.56*** 0.52-0.61 0.80* 0.76-0.83 0.52** 0.46-0.58 3 1.24*** 1.18-1.31 1.42*** 1.32-1.521 1.10*** 1.05-1.15 1.63*** 1.47-1.79 other 0.87 0.75-1.00 0.93 0.76-1.15 0.89* 0.79-0.99 0.77 0.57-1.04 Diabetes at baseline 1.38*** 1.32-1.44 1.42*** 1.33-1.51 1.19*** 1.15-1.24 1.31*** 1.21-1.42 Achieved VL=0 0.62*** 0.54-0.73 0.48*** 0.38-0.61 0.71*** 0.63-0.80 0.62** 0.42-0.81 *p<0.05; **p<0.01; ***p<0.0001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. SENSITIVITY ANALYSIS: IMPACT OF FIBROSIS STAGE Risk models for the composite endpoint and all-cause mortality were re-estimated using only patients with a baseline FIB-4 score [N=54,420] and FIB-4 stage was entered as a potential risk factor. Several results are noteworthy [Table 3-5]. First, our core estimates of the impact of 24 achieving an undetectable viral load on event risk were very robust [Table 3.3]. If anything, the estimated risk reduction associated with viral load suppression increased when the analysis took into account baseline levels of fibrosis. Second, risk of the composite event and all-cause mortality was monotonically and positively related to the patient’s baseline fibrosis level. FIB-4 stage 2 increased risk of the composite event by 47% [H.R. = 1.47 (1.42-41.54)], while stage 3 hazard ratio for the composite event was 3.44 [3.29-3.61]. The corresponding hazard ratios for the risk of death were 1.46 [1.37-1.55] for stage2 and 3.77 [3.55-4.00] for stage 3 In the FIB-4 analyses, each additional year of age was estimated to decrease event risk, but this is likely due to age appearing in the FIB-4 calculation. Genotype 3 was also associated with lower risk relative to genotype 1 in the FIB-4 sensitivity analyses, but the estimated hazard ratios were not statistically significant. Table 3-5 Impact of Viral Clearance on the Risk of Late-Stage Liver Events Adjusting for Fibrosis Stage Patient Characteristics Composite of Clinical Outcomes Death All Events One-Year Washout N=51,831 N=46,059 N=54,420 Number of Events [%] 16,291 [31.4%] 10,649 [23.1%] 7,639 [14.0%] Gender [Male] 1.03 0.93-1.15 1.06 0.93-1.20 1.34** 1.12-1.60 Age [in years] 0.99*** 0.98-0.99 0.99*** 0.98-0.99 1.04*** 1.04-1.04 Race [vs White] Black 0.73*** 0.70-0.76 0.76*** 0.72-0.79 0.70*** 0.66-0.74 Other 0.62*** 0.61-0.68 0.61*** 0.57-0.65 1.23*** 1.16-1.30 Prior Admission [6 mo.] 1.42*** 1.37-1.47 1.40*** 1.34-1.46 1.50*** 1.42-1.57 HCV Genotype [vs 1] 2 0.82*** 0.77-0.87 0.79*** 0.74-0.85 0.85*** 0.79-0.92 3 0.97 0.91-1.04 0.98 0.91-1.05 0.94 0.86-1.03 25 other 0.87 0.74-1.02 0.89 0.74-1.07 0.84 0.68-1.04 Diabetes at baseline 1.18*** 1.12-1.24 1.21*** 1.14-1.28 1.45*** 1.37-1.53 FIB-4 Stage [vs.Ishak F0- F3: <1.45] Ishak Inconclusive [1.45 to <3.25[ 1.47*** 1.42-1.54 1.50*** 1.43-1.57 1.46*** 1.37-1.55 Ichak F4-F6 [ >3.25] 3.44*** 3.29-3.61 3.39*** 3.21-3.58 3.77*** 3.55-4.00 Achieved VL=0 0.74** 0.62-0.87 0.72** 0.60-0.86 0.53*** 0.42-0.67 *p<0.05; **p<0.01; ***p<0.0001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. 3.5 DISCUSSION We found that viral load suppression was associated with decreased risk of liver-related events using data from a large cohort of real-world HCV patients at various stages of disease progression while controlling for other risk factors, including genotype. This study considered a wide range of liver-related events as study outcome variables and measured these outcomes and viral load suppression as time-dependent outcomes over as long as 10 years depending on each patient’s availability of data. Finally, the risk factors studied here were derived from electronic medical records data and included both baseline data [gender, race, genotype] and other time dependent risk factors [BMI, age] that take full advantage of the temporal relationships between events and patient risk factors. Previous research has clearly documented that patients with an undetectable viral load 18 or who achieve a sustained viral response due to treatment are at significantly lower risk of late- stage liver events and death. 14, 19 Our results are consistent with these earlier studies. Patients who achieve an undetectable viral load reduce their risk of death by 45% and our composite of liver-related events by 27% relative to those patients whose viral load was detectable over the 26 entire period following their diagnosis. More importantly, very few patients achieve an undetectable viral load without treatment [39 of 97,485 untreated patients]. While antiviral therapy can lead to viral eradication and reduced event risk, its effectiveness under real-world clinical conditions is limited by side effects and other factors. In this study, only 1 in 4 HCV patients with a detectable viral load were willing to initiate treatment. Once treated, only a fraction of patients achieved the minimum treatment response of a single undetectable viral load test. Our rate of treatment ‘success’ of 16.4% is consistent with other studies. For example, Kramer, Kanwal, Richardson, et al. 19 documented the SVR rates achieved using standard antiviral therapy in real-world clinical settings ranged between 14-24% for HCV genotype 1 and 37-52% for genotypes 2 or 3. In this study, 16.4% of treated patients achieved viral suppression. The independent role of genotype on the risk of liver-related events has been controversial, mostly because of limited number of patients with non-genotype 1 infection. 20, 21 Our results are consistent with Larsen, Bousquet, Delarocque-Astagneau, et al. 22 who demonstrated that genotype 1 and 3 may be associated with greater rates of disease progression than other hepatitis C genotypes. Our results are also consistent with prior observations demonstrating low risk of disease progression in African American patients. 23, 24 Kallwitz, et al. found that both Hispanics and non-Hispanic whites had a higher risk of cirrhosis compared to blacks (OR of 1.6 and 2.4, respectively). 12 This study found that incidence rates of HCC were also slightly lower in blacks as well. Our sensitivity analysis using baseline FIB-4 scores to define fibrosis stages found these staging variables to be highly predictive of hepatic morbidity and mortality. This supports the expanded use of non-invasive fibrosis staging methods as substitutes for liver biopsy. It is noteworthy that the inclusion of baseline FIB-4 levels into the statistical models eliminated the 27 estimated increase in risk associated with genotype 3 relative to genotype 1. The exact reason for the differential risk effects associated with genotype 3 are unclear but may be related to the higher risk of hepatic steatosis in genotype 3 patients. 25-27 The use of HCV protease inhibitors is associated with a significant increase in SVR rates relative to standard therapy, but also with increases in frequency and severity of side effects such as anemia, neutropenia, thrombocytopenia, rash, and gastrointestinal events. 28 Early reports suggest non-interferon based therapies will deliver increased SVR rates while reducing associated substantial side effects that limit tolerability. 29-33 Clearly, new therapeutic options stand to reap significant benefits to patients and the health care system if their introduction improves the willingness of patients to initiate therapy and the likelihood that the patient will achieve viral suppression leading to SVR. Natural history data and an understanding of the challenges and expectations from patients are essential to help both providers and patients make informed decisions when to initiate antiviral therapy and to motivate patient adherence. 34 LIMITATIONS There are several important limitations in our study. First the VA study population differed significantly from the U.S. population, consisting mostly of white and black men. Therefore, result for the risk associated with gender and the catch-all category of ‘other race’ should be viewed with caution. Nevertheless, most HCV patients in the U.S. are male 3,4 and VA is the largest provider of care to chronically HCV-infected patients in the US. 35 We did not measure sustained viral response [SVR] which requires that an undetectable viral load be maintained for six months following the termination of treatment. Instead, we used time to the patient’s first undetectable viral load test. It is possible that patients achieving viral 28 load suppression at one point in time can relapse. Nevertheless, our results are consistent with previous studies even sub-optimal therapy is associated with survival benefits. 8 The sensitivity of HCV viral load tests has changed over time, presented a challenge in defining an ‘undetectable’ viral load. Many older tests used prior to 2004 have a lower threshold of 600 IU/ml below which the result would be reported as ‘undetectable.’ Newer tests are sensitive down to 10 IU/mL. For patients with more sensitive tests, we chose to define reported values under 25 IU/ml as ‘undetectable’. This over-classification of patients as being undetectable excluded some patients with baseline detectable viral loads from the study sample. Mis-classification of some patients as having achieved viral suppression was much less likely as these measurements were made later in the data period. If viral suppression is based on an older lab technology, then some ‘detectable’ post-index tests would be mis-categorized as have achieved viral suppression. This created a conservative bias in our estimates of the impact of achieving viral suppression as some patients in this category would have viral loads as high as 600 IU/ml and higher risk. This study did not estimate or control for the effects of treatment on clinical endpoints and death. This was done for two reasons. First, viral suppression without treatment is exceeding rare, consisting of only 39 patients out of total of 97,485 untreated patients. Second, the parameters with which to determine if a patient completed an adequate course of therapy vary by genotype and other factors, such as allowable duration on breaks in treatment. While developing counts of continuous days of therapy have been used by this research team in the past, 36 we elected to use viral load suppression as our measure of treatment success. Finally, our study did not capture medical care outside the VA system, such as the Medicare program, which may cloud the relationship between viral load suppression and event 29 risk. For example, viral load suppression is highly correlated with treatment which is expensive and was not well covered in the Medicare program before Part D became effective in 2006. Even when covered, drug copayments and the cost of physician visits for injections constitute a significant financial burden. This suggests that treated patients use the VA system which likely rolls over into treatment for any future liver-related events. If true, this would result in an under- estimate of the impact of viral load suppression on event risk. Fortunately, the problem of missing outcome data does not apply to mortality where viral load suppression has a larger estimated effect. 30 3.6 REFERENCES 1. Lavanchy D. The global burden of hepatitis C. Liver Int 2009;29(Suppl 1):74-81. 2. Chen SL, Morgan TR. The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006;3(2):47-52. 3. Armstrong GL, Wasley A, Simard EP, et al. The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med 2006;144(10):705-714. 4. Chak E, Talal AH, Sherman KE, Schiff ER, Saab S. Hepatitis C virus infection in USA: An estimate of true prevalence. Liver Int 2011;31(8):1090-1101. 5. Seeff LB. The history of the "natural history" of hepatitis C (1968-2009). Liver Int 2009;29(Suppl 1):89-99. 6. Klevens RM, Hu DJ, Jiles R, Holmberg SD. Evolving epidemiology of hepatitis C virus in the United States. Clin Infect Dis 2012;55(Suppl 1):S3-9. 7. Smith BD, Morgan RL, Beckett GA, et al. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012;61(RR-4):1-32. 8. Butt AA, Wang X, and Moore CG. Effect of hepatitis C virus and its treatment on survival. Hepatology 2009;50(8):387-392. 9. El-Serag HB, Mason AC. Risk factors for the rising rates of primary liver cancer in the United Stats. Arch Intern Med 2000;160(21):3227-3230.. 10. Davis GL, Albright JE, Cook SF, Rosenberg DM: Projecting future complications of chronic hepatitis C in the United States. Liver Transpl 2003;9(4):331-338. 31 11. Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB. Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis 2011;43(1):66-72. 12. Kallwitz ER, Layden-Almer J, Dhamija M, et al. Ethnicity and body mass index are associated with hepatitis C presentation and progression. Clin Gastroenterol Hepatol 2010;8(1):72-78. 13. Lee MH, Yang HI, Lu SN, et al. Chronic hepatitic C virus infection increases mortality from hepatic and extrahepatic diseases: A community-based long-term prosepctive study. JID 2012(15 Aug);206:469-477. 14. van der Meer AJ, Veldt BJ, Feld JJ, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA 2012;308(24):2584-2593. 15. Backus LI, Gavrilov S, Loomis TP, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009;16(6):775-783. 16. Vallet-Pichard A, Mallet V, Nalpas B, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology 2007;46(1):32-36. 17. Sterling RK, Lissen E, Clumeck N, Sola R, Correa MC, Montaner J, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 2006;43:1317-1325. 18. Ng V, Saab S: Effects of a sustained virologic response on outcomes of patients with chronic hepatitis C. Clin Gastroenterol Hepatol 2011;9(11):923-930. 32 19. Kramer JR, Kanwal F, Richardson P, Mei M, El-Serag HB. Gaps in the achievement of effectiveness of HCV treatment in national VA practice. J Hepatol 2012;56(2):320-325. 20. Zhou S, Terrault NA, Ferrell L, et al. Severity of liver disease in liver transplantation recipients with hepatitis C virus infection: relationship to genotype and level of viremia. Hepatology 1996;24(5):1041-6. 21. Gordon FD, Poterucha JJ, Germer J, et al. Relationship between hepatitis C genotype and severity of recurrent hepatitis C after liver transplantation. Transplantation 1997;63(10):1419-23. 22. Larsen C, Bousquet V, Delarocque-Astagneau E, et al. Hepatitis C virus genotype 3 and the risk of severe liver disease in a large population of drug users in France. J Med Virol. 2010;82(10):1647-54. doi: 10.1002/jmv.21850. 23. Crosse K, Umeadi OG, Anania FA, et al. Racial differences in liver inflammation and fibrosis related to chronic hepatitis C. Clin Gastroenterol Hepatol. 2004;2(6):463-8. 24. Sterling RK, Stravitz RT, Luketic VA, et al. A comparison of the spectrum of chronic hepatitis C virus between Caucasians and African Americans. Clin Gastroenterol Hepatol 2004;2(6):469-473. 25. Castera L, Hezode C, Roudot-Thoraval F, et al. Effect of antiviral treatment on evolution of liver steatosis in patients with chronic hepatitis C: Indirect evidence of a role of hepatitis C virus genotype 3 in steatosis. Gut 2004;53:420-424. 26. Nkontchou G, Ziol M, Aout M, et al. HCV genotype 3 is associated with a higher hepatocellular carcinoma incidence in patient with ongoing viral C cirrhosis. J Viral Hepatitis 2011;18:e516-e522. 33 27. Tapper EB, and Afdhal NH. Is 3 the new 1: Perspectives on virology, natural history and treatment for hepatitis C genotype 3. J Viral Hepatitis 2013;20:669-677. 28. Gaetano JN, Reau N. Hepatitis C: management of side effects in the era of direct-acting antivirals. Curr Gastroenterol Rep. 2013;15(1):305. doi: 10.1007/s11894-012-0305-1. 29. McHutchison JG, Everson GT, Gordon SC, et al. Telaprevir with peginterferon and ribavirin for chronic HCV genotype 1 infection. NEJM 2009;360(15):1827-38. 30. Poordad F, McCone J, Bacon BR, et al. Boceprevir for untreated chronic HCV genotype 1 infection. NEJM 2011;364(13):1195-1206. 31. Bacon BR, Gordon SC, Lawitz E, et al. Boceprevir for previously treated chronic HCV genotype 1 infection. NEJM 2011;364(13):1207-1217. 32. Jacobson IM, McHutchison JG, Dusheiko G, et al. Telaprevir for previously untreated chronic hepatitis C virus infection. NEJM 2011;364(13):2405- 2416https://vpn.nacs.uci.edu/+CSCO+0h756767633A2F2F6A6A6A2E6172777A2E626574+ +/toc/nejm/364/25/ 33. Zeuzem S, Andreone P, Pol S, et al. Telaprevir for retreatment of HCV infection, NEJM 2011;364(25):2417-2428. 34. Fusfeld L, Aggarwal J, Dougher C, et al. Assessment of motivating factors associated with the initiation and completion of treatment fore chronic hepatitis C virus [HCV] infection. BMC Infectious Diseases 2013, 13:234-245. 35. Center for Quality Management in Public Health. The state of care for veterans with chronic Hepatitis C. In. Palo Alto, California: US Department of Veteran Affairs, Public Health Strategic Health Care Group, Center for Quality Management in Public Health; 2010. 34 36. McCombs JS, Shin J, Hines P, Yuan Y and Saab S. Impact of Drug Therapy Adherence in Patients with Hepatitis C. Am J Pharmacy Benefits 2012;4[Special Issue]:SP19-SP27. 35 CHAPTER 4 : EXTERNAL VALIDATION OF THE RISK-PREDICTION MODEL FOR HEPATOCELLULAR CARCINOMA [HCC] FROM THE REVEAL HCV STUDY Tara Matsuda, BA 1,3 , Jeffrey S. McCombs, PhD 1 , Mei-Hsuan Lee, PhD 2 , Ivy Tonnu-Mihara, MS, PharmD 3 , Gilbert L’Italien, PhD 4 , Yong Yuan, PhD 4 1. Department of Clinical Pharmacy and Pharmaceutical Economics and Policy, School of Pharmacy, Leonard Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, CA, USA 2. Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan 3. Veterans Affairs Long Beach Healthcare System, Long Beach, CA, USA 4. Global Health Economics and Outcomes Research, Bristol-Myers Squibb, Plainsboro, NJ, USA Will submit to Journal of Viral Hepatitis. 4.1 ABSTRACT Objectives: The REVEAL-HCV investigators in Taiwan developed a risk-prediction model for hepatocellular cancer [HCC] using a prospective cohort of 975 hepatitis C infected [HCV] patients followed for 17 years. Our aim is to validate their risk-prediction model using data for an external cohort of U.S. military veterans with HCV. Methods: A retrospective cohort of HCV patients was followed longitudinally using data from the VA Clinical Case Registry [CCR]. The date of the patient’s HCV diagnosis was set as the index date. A subset of VA patients was then selected using REVEAL-HCV section criteria and screened for baseline data on ALT, ALT/AST ratio, cirrhosis, HCV viral load and genotype. The area under ROC curve analysis was used to evaluate the performance of the REVEAL-HCV risk models. Results: 47,578 VA HCV patients met study inclusion criteria. The VA sample varied significantly from the REVEAL-HCV sample in terms of male gender and race. The overall 36 incidence of HCC following a 5 year washout period was 4.1% [n=1,000]. Average HCV risk score was 13.2 [SD=2.4] for non-HCC patients and 14.9 [SD=2.4] in patients with HCC [p<0.0001]. The incidence of HCC increased across risk score tertiles from 0% for the first tertile to 11% for the highest tertile of risk scores. The area under the ROC was 0.69 when the REVEAL-HCV model was applied to VA patients. Conclusions: The REVEAL-HCV risk prediction model is robust even when used for VA patients who differ significantly from the Taiwan sample from which the risk score model was estimated. 4.2 INTRODUCTION Risk-prediction models play an important role in medicine. Clinicians use risk-prediction models to manage chronic diseases such as cardiovascular disease, with the Framingham equation being the most well-known, and aid in treatment decisions [1, 2]. Patient risk factors are monitored in an effort to motivate patients to undertake preventive activities and determine when medical intervention is warranted. Risk-prediction models undergo a process of validation before they gain wide acceptance with clinicians. The model developers first use a variety of methods with which to validate their model using the data they have available for model estimation. While this process of internal validation is important, the use of a risk-prediction model also depends critically on validation of the model’s predictive power using data from an external population drawn from a different data set. To be widely used, a risk-prediction model must predict well outside the medical system in which it was developed. Risk prediction modeling in hepatitis C was, until recently, designed to provide clinicians with valid data with which to motivate patients to initiate therapy. Patients frequently resisted 37 initiating standard drug therapy due to the slow and uncertain progression of the infection and the significant burden of therapy. Fortunately, recent advancements in drug therapy for hepatitis C have significantly improved the willingness of patients to initiate therapy due to a shorter and less toxic course of treatment and very high rates of cure. However, these new treatment options command a very high price. The burden of HCV in the United States (US) in 2010, was estimated by the NHANES survey at 3.5 million people [3]. However, as this survey does not include high-risk groups, this may be an underestimate of the prevalence of HCV by about 1.9 million people [4]. As such, these new drug options have transformed the use of risk-prediction models in the treatment of hepatitis C from motivating patient’s initiation of treatment into a tool with which clinicians and payers can ration therapy based on patient risk. The Risk Evaluation of Viral load Elevation and Associated Liver disease/cancer (REVEAL)-HCV study in Taiwan found that viral load, HCV genotype, and alanine aminotransferase (ALT) levels significantly increased the risk of HCC [5, 6]. Using these results, the REVEAL-HCV investigators estimated a risk prediction model for HCC [7], and conducted an external validation of their model using a second Taiwanese cohort by means of receiver operating characteristic (ROC) curves methods. The REVEAL-HCV risk prediction model achieved an area under the ROC curve of 0.83 when validated using the estimation cohort, and 0.73 when using the external validation cohort for 5-year HCC risk. The purpose of this study is to provide a second external validation study for the REVEAL-HCV HCC risk prediction model using a large population of HCV infected patients drawn from the Veterans Administration Health Care System in the United States. The VA validation cohort will be largely male and non-Asian, providing a validation test far afield from the REVEAL-HCV study population. 38 4.3 DATA AND METHODS DATA This validation study employs the retrospective cohort of HCV patients from the VA clinical case registry (CCR) data system used previously to evaluate the impact of viral load suppression [8] and the impact of abnormal laboratory values [9]. The VA had established the CCR to identify cohorts of veterans with targeted conditions at the local and national level, review clinical status and medical outcomes, and provide opportunities for improving care. Potential HCV patients are identified by the presence of an HCV-related ICD-9 diagnosis code or a positive lab test. Local CCR coordinators at the 128 reporting facilities then manually confirm all historical data on the patient before they are added to the CCR [8]. Data elements of the CCR include: demographics (age, gender, race, ethnicity, and geographic region), height and weight, inpatient admission data (primary diagnosis ICD-9 codes), outpatient visits (diagnosis ICD-9 and procedure CPT-4 codes), prescriptions drugs, and the problem file. CCR from 1999 through 2010 were used in this analysis [maximum follow-up time of 11 years]. All patients were screened for an HCV viral load and HCV genotype test and only those patients with a reported quantifiable viral load or a detectable genotype were included in this study. The first occurrence of either result became the ‘index date’ for each patient, which was a proxy of the date of HCV infection. This index date served as the baseline date for calculating the REVEAL-HCV risk score. To be included in the study the patients had to have all the information required to calculate the REVEAL-HCV risk score. Exclusion criteria from the REVEAL-HCV study was used: age < 30 and > 65 at baseline; HCC diagnosis within 5 years of baseline; patients who died during the post-index period; patients with missing baseline data 39 [index data +/- 6 months] on ALT, aspartate aminotransferase (AST)/ALT ratio, HCV viral load and genotype; and less than 5 years of post-index data. STATISTICAL ANALYSIS The baseline data for each VA patient was used to calculate a REVEAL-HCV risk prediction score [11] and the 5-year and 10-year predicted risk was assigned, given each risk score. The validity of the Taiwan risk prediction model applied to our VA sample was tested in two ways. First, an area under the ROC curve analysis was performed to compare the actual incidence of HCC after 5 years to their predicted risk. The ROC curve itself is a visual tool used in assessing the accuracy of a risk prediction model, plotting the true positives of the risk model (sensitivity) against the false positives (1-specificity) [9]. The ROC curve has been used often in biomedical informatics research to evaluate models for decision support, diagnosis, and prognosis [10]. One widely used summary index of the ROC curve is the area under the ROC curve. The area under the ROC curve measures the probability of a correct ranking of a normal (N), abnormal (A) pair of cases: Pr(X A >X N ), where X A is the degree of suspicion from the abnormal case and X N for the normal case [11]. In terms of this paper, the area under the ROC curve represents the probability that the diagnostic index of patient with HCC will be greater than that of a patient with no HCC diagnosis. The practical range of the area under the ROC curve is from 0.5 to 1. An area of 0.5 represents an accuracy of chance, visualized by a 45-degree line for the ROC curve; and 1 represents perfect accuracy, visualized by an ROC curve running vertical from origin to (0,1), and then horizontal to (1,1) [10]. Second, since the VA data were censored, Nelson-Aalen estimates were used to analyze cumulative risks for HCC by follow-up years and displayed by calculated risk score categories. Differences in baseline were assessed using the t- 40 tests and chi-squared analysis, and differences in survival functions were assessed using log-rank tests. SAS v9.2 was used to construct the dataset and STATA v11 for statistical analysis. Ethics statement This research has been approved by the VA Long Beach IRB committee. The data analyzed was taken from a de-identified retrospective database, so no consent was necessary. 4.4 RESULTS Out of a total of 360,857 unique patients registered in the VA HCV CCR database, a study population of 47,578 patients met all of the study inclusion criteria for the validation sample. The descriptive statistics comparing the VA and REVEAL-HCV samples are provided in Table 4-1. Table 4-1 Baseline Descriptive Statistics Of REVEAL-HCV And VA Cohorts Characteristics VA cohort (N=47,578) REVEAL-HCV cohort † (N=975) p-value Age, Mean±SD 50.1 ± 5.5 50.9 ± 9.3 <.0001 Age, N (%) 30-39 1,411 (3.0) 163 (16.7) <.0001 40-49 19,838 (41.7) 217 (22.3) 50-59 24,347 (51.2) 399 (40.9) 60-65 1,982 (4.2) 196 (20.1) Gender*, N (%) Female 1,550 (3.3) 550 (56.4) <.0001 Male 46,022 (96.7) 425 (43.6) Serum ALT levels, N (%) ≤15 582 (1.2) 429 (44) <.0001 16-45 14,393 (30.3) 387 (39.7) >45 32,603 (68.5) 159 (16.3) Liver cirrhosis, N (%) 41 No 43,937 (92.4) 961 (98.6) p<.0001 Yes 3,641 (7.7) 14 (1.4) Serum HCV RNA levels*, N (%) Undetectable (<25 IU) 1,249 (2.6) 298 (30.6) p<.0001 Low levels (<20,000 IU) 1,569 (3.3) 339 (34.8) High levels (>20,000 IU) 43,409 (91.2) 338 (34.7) AST/ALT ratio <1 35,104 (73.8) 340 (34.9) <.0001 ≥1 12,474 (26.2) 635 (65.1) Genotype** (among those with detectable HCV RNA) Non-1 9185 (19.3) 271 (29.5) <.0001 1 35824 (75.3) 351 (38.2) SD = standard deviation; ALT= alanine aminotransferase; HCV= Hepatitis C virus; RNA= ribonucleic acid; AST = aspartate aminotransferase † Data source: Lee MHet al: Prediction models for risk of hepatocellular carcinoma using seromarkers of hepatitis C virus infection. Poster at The International Liver Congress; 2011. *Data allowed to be missing as long as risk score could be calculated ** REVEAL-HCV cohort genotype was reported only for those with detectable HCV RNA levels The average age in the two samples both averaged around 50 years, but the VA sample was much more tightly clustered in the age strata 40-49 [41.7%] and 50-59 [51.2%] than the REVEAL-HCV sample [22.3% and 40.9%, respectively]. The VA sample was predominately male [96.7%], while the REVEAL-HCV sample was mostly comprised of females [56.4%]. Looking at lab results, a greater fraction of the VA sample had high ALT levels (>45 U ml -1 ) and HCV RNA levels (>20,000 IU ml -1 ) compared to the REVEAL-HCV sample. However, the REVEAL-HCV sample had a greater fraction of their sample with a high AST/ALT ratio (≥1). 7.7% of the VA sample were diagnosed with liver cirrhosis at baseline, compared to 1.4% of the REVEAL-HCV sample. Though race was not specified for the REVEAL-HCV cohort, as the study was done in Taiwan, a majority of the sample is assumed to be Asian. The VA cohort was 42 primarily Caucasian [51.1%] and Black [31%]. The majority in the VA cohort were genotype 1 [75.3%], compared to only 38.2% in the REVEAL-HCV cohort. All of these demographics and baseline characteristics differences between the two samples were significant at an alpha level of .0001. A total number of 1,000 HCC cases developed in the VA cohort. The median follow-up time was 7.6 years, and the cumulative risk of HCC over follow-up was 4.1%. The REVEAL-HCV cumulative risk score was calculated for each patient in the VA validation sample. The mean risk score of the VA cohort was 13.3 [standard deviation (SD) of 2.5], with a range of 2 to 22. The distribution of risk scores is shown in Figure 4-1. When stratified by HCC outcome, the mean risk score for patients who did not develop HCC was 13.2 [SD=2.44], and 14.9 [SD=2.38] for patients did develop HCC. Time to HCC was analyzed by risk score categories: high [16-22], medium [8-15], and low risk [0-7] [Figure 4-2]. 43 Figure 4-1 Distribution Of Risk Scores, By Cohort Figure 4-2 Cumulative Risk Of HCC, Stratified By Risk Score Categories. Low-Risk: 0-7 Risk Score; Medium-Risk: 8-15 Risk Score; High-Risk: 16-22 Risk Score 0 5 10 15 20 25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Percentage (%) Risk Score VA REVEAL p<0.001 0.110 0.035 0.00 0.00 0.05 0.10 Cumulative Risk 5 6 7 8 9 10 11 Years of Follow-up High risk Low-risk Medium-risk 44 None of the VA patients classified as low risk using the Taiwan risk model developed HCC, while the cumulative risk of HCC was 3.48% for those at medium risk and 11.02% for high risk patients (p<0.001). The cumulative risk of HCC is also broken down by baseline viral load [Figure 4-3] and genotype [Figure 4-4]. The cumulative risk for HCC came out to 0%, 4.32%, and 4.85% for undetected (<25 IU ml -1 ), low viral load (<20k IU ml -1 ), and high viral load (>20k IU ml -1 ), respectively [Figure 4-3] (p<0.001). And when broken down by genotype, the cumulative risk for HCC was 3.75% for non-1 genotypes and 5.28% for genotype 1 [Figure 4-4] (p=0.001). Both the five year and ten year area under the ROC were estimated at 0.69 using the REVEAL- HCV scored calculated for the VA validation cohort from the U.S [Figure 4-5]. This value compares well to the ROC of 0.83 when estimated against the estimation sample [9] and 0.73 when estimated using an external validation sample from Taiwan. Figure 4-3 Cumulative Risk Of HCC, Stratified By Viral Load. High: Viral Load>20,000 IU; Low: Viral Load ≤ 20,000 IU; Und: Undetected Viral Load (<25 IU) p<0.001 0.049 0.043 0.00 0.00 0.01 0.02 0.03 0.04 0.05 Cumulative Risk 5 6 7 8 9 10 11 Years of Follow-up High Low Und 45 Figure 4-4 Cumulative Risk Of HCC, Stratified By Genotype (GTP) Figure 4-5 Receiver Operator Characteristic (ROC) Curve p=0.001 0.038 0.00 0.01 0.02 0.03 0.04 0.05 Cumulative Risk 5 6 7 8 9 10 11 Years of Follow-up non-1 GTP GTP 1 0.053 0.00 0.25 0.50 0.75 1.00 Sensitivity 0.00 0.25 0.50 0.75 1.00 1 - Specificity Area under ROC curve = 0.6924 46 4.5 DISCUSSION Given the highly divergent patient populations between the REVEAL-HCV and VA cohorts, the REVEAL-HCV risk prediction model fit the VA data reasonably well. The 0.83 ROC estimate using the REVEAL-HCV cohort and the 0.73 ROC estimate using the Taiwanese validation cohort are not so much different from the 0.69 area under the ROC curve estimate obtained using this VA cohort. Though an area under the ROC curve of greater than 0.7 is considered a ‘fair’ estimate, the .69 area obtained in the VA cohort does give credence to the REVEAL-HCV risk prediction model and its chosen predictors, as it did a relatively good job in predicting risk for HCC in a very different cohort. Despite the similar mean ages, the majority of the VA cohort was between the ages of 40-59, while the ages had a more even distribution in the REVEAL-HCV cohort. And unsurprisingly, the VA cohort was predominately male, with the majority of the sample being of white or black race. This was in stark contrast to the predominately female REVEAL-HCV cohort of Taiwanese people. The VA and REVEAL- HCV cohorts differed not only in patient demographics, but also staging of HCV progression, if evaluated by the baseline diagnosis and labs and based on risk scores. In terms of diagnosis of cirrhosis and the lab results at baseline, the only result where the REVEAL-HCV cohort had a greater percentage of their sample (compared to the VA cohort) within the abnormal category was for AST/ALT ratio. And in contrast to the REVEAL-HCV patient cohort, whose risk scores resemble a bimodal distribution, peaking at seven and eleven [10]; the risk scores for the VA patient cohort are centered further down, peaking around thirteen. The cumulative risk analysis by risk scores confirms that the REVEAL-HCV risk scores are very predictive of HCC risk, as are baseline viral load and genotype, evidenced by the statistically significant difference in survival function. In accordance to the REVEAL-HCV model, the cumulative risk of HCC was 47 highest for those with the highest risk scores, and zero for those with the lowest. The analysis of HCC risk by baseline viral load is consistent with the recent findings using the VA data set that found achieving an undetectable viral load reduced the risk of HCC 38% [8]. LIMITATIONS There are several important limitations in our study. First, the VA study population differs significantly from the U.S. population, most notably by the overwhelming majority of men. Nevertheless, the VA is the provider of care to the largest cohort of chronically HCV-infected patients in the US, so there is still much use for this application. Secondly, the index date is probably not a good proxy for the date of diagnosis, and we are probably catching veterans at a later time in course of infection than the REVEAL-HCV cohort, as shown by the differing baseline characteristics between the REVEAL-HCV and VA cohorts. However, these differences between the two cohorts highlight how well the REVEAL-HCV risk prediction model performed, given these divergent samples. And though imperfect, the index date is representative of the majority of patients who are diagnosed well after the initial date of infection due to the asymptomatic nature of HCV infection. There are also important limitations that are associated with retrospective database studies. The sensitivity of HCV viral load tests have improved over time, with many older tests having a lower threshold of 600 while newer tests are sensitive up to 10 IU mL -1 . It is also likely that there may be missing clinical data as the VA sample as this study is also highly dependent on ICD-9 codes to identify HCC cases. This missing data issue is further augmented, as the CCR does not capture medical care outside the VA system. If patients developed complications of liver disease outside the VA, the data may not have been captured. However, we believe that because VA system is inclusive and provides significant benefits at lower costs to patients compared to private plans, most of the care was 48 continued in VA facilities. We are also limited by the follow-up time in the VA sample. HCC typically take decades to develop following the point of infection, and with a maximum follow- up time of 11 years, this time frame is probably catching only a fraction of those who will develop HCC. However, despite these limitations of the data, the risk prediction model generated by the REVEAL-HCV group has worked very well in this VA sample. It is incredibly useful to be able to give physicians another tool in which they can focus on patients who are at most risk for developing complications due to HCV to balance with the increased demand for treatment. The complications that may result from HCV are serious and costly, with risk of developing progressive liver disease and HCC [12, 13]. As such chronic HCV is the most common cause for liver transplantation [14]. Successful treatment of HCV has been shown to reduce the risk of events such as HCC and death [15]. However, delays in treatment of HCV may be detrimental for some patients. A follow-up study that used this VA CCR data found that delaying treatment until late in the course of HCV infections significantly attenuates the effectiveness of treatment (Chapter 7). This validated REVEAL-HCV model does concur with published literature that has found a relation between age, viral load, and genotypes with the development of HCC. McCombs et al. found that age, genotype and viral load were associated with risk of HCC, as well as gender, race, and diabetes [16]. Similarly, Sinn et al. reported that age as well as gender, diabetes, platelet levels, and APRI score were associated with HCV disease progression in patients not on therapy [17]. Age, race, alkaline phosphatase and platelet labs, smoking history, and presence of esophageal varices were found to be significant in a study by Lok et al. [18]. However, all of these factors have not been validated, and would be worth including in new risk score calculations, especially if used for a population that is more racially diverse than 49 the REVEAL-HCV population. Expanding the age range for which the model would be useful for would also be beneficial. The CDC estimated that nearly three quarters of HCV infections in the US were among people born between 1945-1965, so many HCV infected people are approaching or have crossed the 65 year age limit of the model [16]. Although the REVEAL- HCV risk prediction model technically allows a risk projection beyond 65 years old by assigning a higher score, this extrapolation won’t be precise and should be used with caution, as most likely the projected risk for the elderly is underestimated. It may also be apt to consider risk prediction models for earlier HCV complications, for which a greater percent of the population would be at risk for. Risk prediction models based on HCC may not be particularly useful for improving patient acceptance of treatment due to how much later it develops in the course of the HCV infection as well as its low cumulative risk. A 10% chance of developing HCC for a patient in the ‘high risk’ category may still not be enough of an incentive to consider treatment. A more useful outcome may be compensated cirrhosis, as it develops earlier, its risk is higher, and it is the progenitor to decompensated cirrhosis and HCC. The Taiwan REVEAL-HCV investigators have proved that their risk-prediction model for HCC is robust in a general Taiwanese population, but as there are several differences between a population in Taiwan and the US, this study strove to test the robustness in a US veteran population. Using a model that incorporated baseline data on ALT, ALT/AST ratio, cirrhosis, HCV viral load and genotype, risk scores for HCC were calculated. While the VA sample varied significantly from the REVEAL-HCV sample in terms of male gender and race, an area under the ROC of 0.69 was estimated when the REVEAL-HCV model was applied to VA patients. This AUROC indicated that the REVEAL-HCV risk prediction model is robust in 50 a US veteran population, a significantly different patient population from which the risk score model was estimated. 51 4.6 REFERENCES 1. Folsom AR: Classical and novel biomarkers for cardiovascular risk prediction in the United States. J Epidemiol 2013, 23(3):158-162. 2. Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB: Prediction of coronary heart disease using risk factor categories. Circulation 1998, 97(18):1837-1847. 3. Ditah I, Ditah F, Devaki P, Ewelukwa O, Ditah C, Njei B, Luma H, Charlton M: The Changing Epidemiology of Hepatitis C Virus Infection in the United States: National Health and Nutrition Examination Survey 2001 through 2010. J Hepatol 2013. 4. Chak E, Talal AH, Sherman KE, Schiff ER, Saab S: Hepatitis C virus infection in USA: an estimate of true prevalence. Liver Int 2011, 31(8):1090-1101. 5. Lee MH, Yang HI, Lu SN, Jen CL, Yeh SH, Liu CJ, Chen PJ, You SL, Wang LY, Chen WJ et al: Hepatitis C virus seromarkers and subsequent risk of hepatocellular carcinoma: long-term predictors from a community-based cohort study. J Clin Oncol 2010, 28(30):4587-4593. 6. Lee MH, Yang HI, Lu SN, Jen CL, You SL, Wang LY, L'Italien G, Chen CJ, Yuan Y: Hepatitis C virus genotype 1b increases cumulative lifetime risk of hepatocellular carcinoma. International journal of cancer 2014. 7. Lee MH, Yang HI, Lu SN, Jen CL, You SL, Wang LY, Yuan Y, Litalian G, Chen CJ, Group R-HS: Prediction models for risk of hepatocellular carcinoma using seromarkers of hepatitis C virus infection. In: International Liver Congress; Berlin, Germany. 2011 March 30-April 3. 52 8. Backus LI, Gavrilov S, Loomis TP, Halloran JP, Phillips BR, Belperio PS, Mole LA: Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009, 16(6):775-783. 9. Alemayehu D, Zou KH: Applications of ROC analysis in medical research: recent developments and future directions. Acad Radiol 2012, 19(12):1457-1464. 10. Lasko TA, Bhagwat JG, Zou KH, Ohno-Machado L: The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform 2005, 38(5):404-415. 11. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982, 143(1):29-36. 12. Seeff LB: The history of the "natural history" of hepatitis C (1968-2009). Liver Int 2009, 29 Suppl 1:89-99. 13. Chen SL, Morgan TR: The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006, 3(2):47-52. 14. Organ Proceurement and Transplantation Networn (OPTN) and Scientific Registrry of Transplant Recipients (SRTR). OPTN/SRTR 2011 Annual Data Report. Department of Health and Human Services, Health Resources and Services Administration, Healthcare Systems Bureau, Division of Transplantation. http://srtr.transplant.hrsa.gov/annual_reports/2011/. Accessed April 3, 2013. 15. Butt AA, McGinnis K, Skanderson M, Justice AC: A comparison of treatment eligibility for hepatitis C virus in HCV-monoinfected versus HCV/HIV-coinfected persons in electronically retrieved cohort of HCV-infected veterans. AIDS research and human retroviruses 2011, 27(9):973-979. 53 16. Smith BD, Morgan RL, Beckett GA, Falck-Ytter Y, Holtzman D, Teo CG, Jewett A, Baack B, Rein DB, Patel N et al: Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012, 61(RR-4):1-32. 54 CHAPTER 5 : FIVE LABORATORY TESTS PREDICT PATIENT RISK AND TREATMENT RESPONSE IN HEPATITIS C: VA DATA FROM 1999-2010 Ivy Tonnu-Mihara 1 , Tara Matsuda 1,2 , Jeff McCombs 2 , Sammy Saab 3 , Patricia Hines 4 , Gilbert L’Italien 4,5 , Timothy Juday 4,6 , Yong Yuan 4 1. Veterans Affairs Long Beach Healthcare System, Long Beach, CA, USA 2. Department of Clinical Pharmacy and Pharmaceutical Economics and Policy, School of Pharmacy, Leonard Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, CA, USA 3. Departments of Medicine and Surgery, David Geffen School of Medicine, University of California-Los Angeles, Los Angeles, CA. 4. Global Health Economics and Outcomes Research, Bristol-Myers Squibb, Plainsboro, NJ, USA 5. Biogen Idec, Boston, MA, USA 6. Abbvie Pharmaceuticals, Abbott Park, Il, USA This manuscript has been accepted pending revisions into the Universal Journal of Medical Science. Further acknowledgement to Anupama Kalsekar, MS from Bristol Myers Squibb for her scientific critique and Tim Morgan, M.D., from the VA in Long Beach for his support of the project. 5.1 ABSTRACT Background: New Hepatitis C [HCV] drugs are expensive but highly effective but have created an overwhelming ‘cash flow’ problem for payers facing a large bolus of infected patients. Health plans are developing ‘watchful waiting’ strategies to safely defer treatment for low risk patients. This study identifies five laboratory tests which predict increasing risk of liver related events and documents that delaying treatment until after an abnormal lab test diminished effectiveness of treatment with interferon alpha plus ribavirin. 55 Methods: Patients from the Veterans Administration’s clinical registry of HCV patients [1999- 2010] were screened for a detectable viral load at baseline and a recorded baseline genotype. The primary outcomes were time to death and time to time to a composite clinical event. Cox proportional hazards models were estimated with time dependent independent variables for initial treatment and first abnormal laboratory test. Results: 128,769 patients met all inclusion criteria. Abnormal values for five laboratory were associated with increased risk for the composite outcome/death: 1.35/1.84 for the AST/ALT ratio > 1; 2.35/5.01 for albumin < 3 g/Dl; 1.58/1.15 for GGT > 195 IU/L; 3.85/1.55 for platelet count < 100 k/mm 2 and 4.48/2.39 for alpha fetoprotein > 144 ng/mL. Delaying drug therapy until after an abnormal lab test significantly reduced treatment effectiveness. Conclusion: Five tests predict liver complications for HCV patients which can be used to develop “watchful waiting” protocols which monitor untreated HCV patients over time and defer access to expensive new drug regimens to those patients most at risk adverse liver-related events. 5.2 INTRODUCTION Hepatitis C (HCV) affects approximately 130-170 million people worldwide [1, 2] and an estimated 3.2 million people in the U.S.[3] HCV patients are at risk for developing progressive liver disease including cirrhosis, liver failure and hepatocellular carcinoma (HCC).[1, 2, 4-6] However, many patients with chronic HCV infections may fail to develop significant symptoms over their lifetime. Progression to cirrhosis is estimated to be about 5-20% after 20 years of infection.[7] It is estimated that 13.1% of patients with HCV in 2005 will die of liver-related causes by 2030 [8], increasing to 36.8% by 2060.[9] While HCV is considered cured if the patient achieves sustained viral response (SVR)[10], achieving SVR using older therapies [pegylated interferon-alpha and ribavirin] has 56 been daunting. Standard therapy requires 24 to 48 weeks of weekly injections and is associated with significant side effect burden. Standard therapy achieves SVR rates ranging between 14- 24% for HCV genotype 1 and 37-52% for genotypes 2 or 3.[11-13] More recent “triple” therapy, achieved by adding a protease inhibitor to pegylated interferon-alpha and ribavirin, is associated with a significant increase in SVR rates and an increase in frequency and severity of adverse side effects such as anemia, neutropenia, thrombocytopenia, rash, and gastrointestinal events.[14] Other reports suggest that ‘triple therapy” may increase SVR rates while reducing the side effects that limit tolerability.[15-21] It is not surprising that clinicians patients faced the twin problems of patient reluctance to initiate therapy and non-adherence once treatment is initiated. The FDA has approved two new therapies [Fall 2014] which have radically changed the decision by patients and physicians when to initiate treatment by radically increasing expected SVR rates and shortening the duration of therapy. However, the costs of these therapies have created a need to ration drug using ‘watchful waiting’/deferred treatment strategies under which clinicians and patients monitor the patient’s risk of disease progression over time and initiate treatment before its effectiveness is adversely impacted. The objective of this research is twofold. First, we use a large historical cohort of HCV patients from to document which specific abnormal laboratory tests predict an increased risk of liver- related liver events, including death. We then investigate whether or not a delay in initiating standard treatment until after one or more of these laboratory tests become abnormal degrades trea tment effectiveness relative to ‘early’ treatment. 57 5.3 METHODS DATA The data used in this study were derived from the Veterans Administration [VA] clinical case registry (CCR) for HCV infected patients. Potential HCV patients were identified by the presence of an HCV-related ICD-9 diagnosis code or a positive viral load lab test. Local CCR coordinators then manually confirm or reject the patient for inclusion in the CCR and all historical data from the patient’s electronic medical record (EMR) were added to the CCR. The VA EMR system was fully implemented in 1999 and provides data on patient demographic and clinical characteristics, diagnoses, laboratory test results and prescription and medical care utilization for over a decade.[22] Data period for this study cover 1999-2010. A patient-level analytic database was created consisting of summary variables for each month before and after the patient’s date of diagnosis [CCR enrollment date]. The post-index period for any individual patient could range up to 10 years depending on their index date. The following summary data were created: 1. Patient demographic data (age in months at baseline, gender, race, ethnicity, and geographic region at baseline). 2. Body mass index [BMI] in each month that these characteristics were recorded. 3. The patient’s diagnostic profile consisting of monthly dichotomous variables reflecting the diagnoses recorded each month. 4. Monthly dichotomous variables for hospital admissions for any diagnosis and for liver related diagnoses. 5. Monthly values for most common laboratory tests, including viral load [VL] and viral genotype. Missing values were assigned when no tests were recorded during the month. 58 6. Prescription drug data were used to create a time-dependent dummy indicating whether or not the patient received HCV-related treatment before each event included as a study outcome. SAMPLE SELECTION CRITERIA All patients were screened for a detectible HCV viral load [> 25 IU/ml] and a documented HCV viral genotype. The minimum value for a detectable viral load was selected based on the ‘best’ test methodology available within the VA system during the study period [1999-2010] and across multiple VA sites that reported data into the CCR. This selection criteria errored on the side of including patients whose viral load was reported as ‘undetectable’ at a higher minimum value [e.g., < 600 IU/ml] using older testing methods. PRIMARY AND SECONDARY OUTCOMES Individuals infected with HCV are at risk for progressive liver disease and related complications such as cirrhosis, liver failure and hepatocellular carcinoma (HCC) which frequently require hospitalization and are at increased the risk of death.[1, 2, 4-6] The primary outcomes for this analysis are a composite of observed liver-related, clinical complications [compensated cirrhosis, decompensated cirrhosis, HCC or a liver-related hospitalization] or all- cause mortality. To ensure the clinical complications were due to chronic HCV infection, any composite clinical outcome occurring within a one-year “washout” period following the index date were not counted. The secondary outcomes included the individual elements of the clinical composite analyzed individually. Monthly dichotomous variables were created for the outcomes of the study based on recorded diagnostic codes [e.g., diagnosis of cirrhosis, etc.] and selected CPT-4 codes included in data from hospital admission data and outpatient services. Liver-related hospitalization events 59 were defined as the first occurrence of a diagnosis for complications due to HCV (see Appendix 1) as the primary diagnosis in inpatient visits. Compensated cirrhosis and HCC outcomes were compiled by searching the inpatient, outpatient and problem lists for matching ICD-9 codes [571.5, 571.2, 571.6 and 155,155.1, 155.2, respectively]. Decompensated cirrhosis was defined as a diagnosis of cirrhosis and a diagnosis of hepatic coma [70.44, 71.71, 348.3, 348.31, 572.2], portal hypertension [572.3], hepatorenal syndrome [572.4], jaundice [782.4], ascites [789.59], or esophageal varices [456, 456.1, 456.2, 456.21] or an FIB-4 score > 3.25.23 An FIB-4 score is a function of age, aspartate transaminase [AST] level, platelet count and alanine transaminase [ALT]. An FIB-4> 3.25 has been correlated with a Metavir fibrosis stage of F3-F4 [23] or an Ishak fibrosis stage of F4-F6. [24] Vallet-Pichard, et al. also found the FIB-4 to have a positive predictive value to confirm the existence of significant fibrosis of 82.1% in a HCV infected cohort.[23] STATISTICAL METHODS The time-to-event variables for primary and secondary outcomes were analyzed using Cox proportional hazards.[25] Time-related independent variables were created for several risk factors to take advantage of the longitudinal nature of the data. The time-related laboratory test risk factors assessed whether or not the laboratory test values were abnormal immediately prior to the event date. The time-related treatment variable was included to assess whether or not treatment was initiated before or after the occurrence of each liver-related outcome variable. The time at which treatment is initiated is endogenous over time and correlated with observed and unobservable patient factors that are also correlated with patient outcomes. Therefore, the treatment effects measured in this analysis may be biased to the extent that important but unobserved patient characteristics are not included in the statistical models. However, the 60 objective of this research is to test whether or not initiating treatment after one or more of the patient’s laboratory value becomes abnormal rather than develop an unbiased estimate of treatment effects. While complicated, the methods used are at the heart of the analysis which ultimately provides information on what do clinicians watch to determine whether or not treatment is needed, and whether or not it is possible to wait too long, thus diminishing the effectiveness of treatment at reducing future risk. The answer to this question will be measured by the estimated coefficient for the interaction term between the time dependent dichotomous variables for treatment and the existence of one or more abnormal lab test. A total of 22 individual clinical laboratory tests were entered into the analyses of each adverse liver event and death in order to determine which abnormal laboratory tests were predictive of these events. Laboratory test result data were entered as time-to-abnormal test result and the specific cut-points for defining an abnormal value can be found in Table 5-2. These cut- off values were set by the authors at a level high enough to generate wide agreement that the recorded lab value was abnormal. All study patients were required to have a detectable viral load in order to focus the analysis on patients who could benefit from treatment. Separate Cox models were estimated for each primary and secondary outcome under study. Each analysis began by estimating a Cox model with all possible risk factors included in the list of independent variables, including the complete list of 22 common laboratory tests thought to be related to liver related adverse events. Since models of event risk should be parsimonious to be useful in clinical practice, the results for these all-inclusive models were reviewed and non-significant factors [p>0.20] were dropped from subsequent model estimates. Age, gender, HCV genotype, and race were included in all models regardless of their statistical significance. Race and ethnicity were initially entered as 61 separately coded categories. However, the significant correlation between race and ethnicity in this VA sample resulted in ethnicity being dropped from the final model specification. The dummy variables denoting a diabetes diagnosis at baseline and any hospital admission in the 6 months prior to the patient’s index date were included in final model specifications based on statistical significance. Once we documented which abnormal laboratory tests were predictive of future liver events, we tested if treatment effectiveness was dependent on the timing of treatment either before or after any one lab value becomes abnormal. This was accomplished by entering both the time dependent treatment variable and its interaction term with a time dependent variable denoting whether or not any abnormal laboratory test preceded treatment. This interaction variable measures the extent to which the effect of treatment initiated after an abnormal laboratory test is statistically different from the impact of ‘early’ treatment. The VA restricted access to CCR data to VA employees [Dr. Tonnu-Mihara] and employees without compensation [Matsuda]. Authors from the funding source [Yuan, Juday, Hines and L’Italien] were involved the research design and with reviewing and interpreting statistical results. All authors reviewed and approved the final manuscript. The analysis was conducted using SAS v9.2.[26] 5.4 RESULTS DESCRIPTIVE STATISTICS A total of 128,769 patients met the study inclusion criteria. Only 24.3% of patients received treatment at any time following diagnosis [Table 5-1]. Only 16.4% of treated patients achieved an undetectable viral load post-treatment [4% of all patients]. The average post-index period consisted of 6.1 years [Standard deviation (SD)=3]. The VA/HCV patients are 62 predominately male of either white or black race [51.4% and 31.3% respectively]. The mean age is 52 years [SD=6.9] and close to 80% of patients were genotype 1, followed by 12% being genotype 2. Over 16% of patients had a history of a hospital admission within 6 months prior to the index date and 12% of patients had a diabetes diagnosis at baseline. Table 5-1 Patient Characteristics N=128,769 Treatment Data Count or Mean % or SD Treated 31,284 24.3% Untreated 97,485 75.7% Achieved Undetectable Viral Load Overall 5,180 4.0% Under Treatment 5,141 16.4% Patient Demographics Gender [Male] 124,980 97.1% Age [mean in years] 51.8 6.9 Race White 66,168 51.4% Black 40,239 31.3% Asian 168 0.1% Other 22,194 17.2% Ethnicity Non-Hispanic 107,586 83.6% Hispanic 6,901 5.4% Other 14,282 11.1% HCV Genotype 1 102191 79.4% 2 15113 11.7% 3 9851 7.7% other 1614 1.23% Pre-index Admission [6 months] 20938 16.3% Diabetes at baseline 15091 11.7% 63 PREDICTIVE LABORATORY TESTS The laboratory data in Table 5-2 document the deteriorating health status of the study HCV population over time. The proportion of patients with abnormal values is higher in the post- index period for all laboratory tests with the exception of HbA1c for diabetes and for viral load. The laboratory tests found to be correlated with the risk of long-term liver complications and death are denoted by a bold font in Table 5-2. The parsimonious models for each of our primary and secondary outcomes are displayed in Table 5-3 [Primary Outcomes] and Table 5-4 [Secondary Outcomes]. The ‘side-by-side’ arrangement of the models makes it easy to evaluate the extent to which risk factors are shared across target outcomes, and the relative strength of risk factors across outcomes. Table 5-2 Laboratory Data BASELINE Definition of Abnormal POST-INDEX LABORATORY TESTS N % Abnormal % Abnormal N BMI [Body Mass Index] 125069 28.3 > 35 45.1 125069 Viral Load [IU/mL] 128769 100 > 25 95.6 128769 Liver Function Tests Direct bilirubin [mg/dL] 29536 0.9 >3.9 3.3 89039 Albumin [g/dL] 74619 7.3 <3.0 24.0 126171 Alanine transaminase (ALT) [IU/L] 81410 21.9 >120 40.2 127199 Aspartate transaminase (AST) [IU/L] 62808 15.8 >120 32.5 105794 AST/ALT Ratio 61221 32.1 >1 67.3 105680 Alkaline phosphatase (ALP) [IU/L] 80344 0.5 >345 2.8 126787 64 Gamma-glutayltransferase (GGT) [IU/L] 19514 25.2 >195 30.5 64651 Hematology Red Blood Cells(RBC) [M/mm 3 ] 56641 0.3 <2.19 3.4 95537 Hemoglobin [g/dL] 77935 1.0 <8 7.8 126417 Hematocrit [%] 75210 1.3 <25 9.7 126073 White Blood Cells(WBC)[k/mm 3 ] 77093 0.3 <2 7.5 126246 Platelets [k/mm 3 ] 77565 7.9 <100 25.8 126253 Neutrophils 53498 3.2 <30 12.5 104403 Renal Function/Panel Creatinine [mg/dL] 83066 6.3 >1.4 24.2 126691 Blood urea nitrogen [BUN] [mg/dL] 79585 0.4 >75 3.6 123719 Estimated glomerular filtration rate [eGFR] [mL/min/1.73 m 2 ] 19431 10.1 <60 25.2 106849 Ions and trace metals Sodium [mEq/L] 81325 0.5 <125 4.1 125302 Potassium [mEq/L] 80746 0.2 <2.7 2.0 124943 Total Serum Iron (Iron)[mcg/dL] 9606 26.8 >140 36.1 89629 Total Iron Building Capacity [TIBC] [mcg/dL] 7457 22.4 >400 29.1 74193 Ferritin [ng/mL] 9384 36.5 >300 41.3 88880 Other tests Glycosylated hemoglobin [HbA1c (%)] 23505 30.6 >7 26.4 87375 Alpha fetoprotein [ng/mL] 7809 1.2 >144 2.6 101623 65 Table 5-3 Impact of Drug Therapy and Abnormal Laboratory Values on the Risk of Liver- Related Adverse Events and Death Composite of Clinical Outcomes Death All Events One-Year Washout N=123,065 N=106,947 N=128,769 Number of Events [%] 35,253 [28.6%] 18,595 [17.4%] 15,458 [12.0%] Estimated Coefficients, HAZARD RATIOS and [95% Confidence Intervals] DRUG THERAPY -0.08 H.R.=0.92 [0.84-1.01] -0.08 H.R.=0.92 [0.84-1.01] -0.20 H.R.=0.82*** [0.75-0.90] LAB TESTS a Albumin [< 3 g/dL] 0.85 H.R.=2.35*** [2.08-2.66] 0.86 H.R.=2.37*** [2.09-2.69] 1.16 H.R.=5.01*** [4.58-5.48] AST/ALT Ratio [> 1] 0.30 H.R.=1.35*** [1.25-1.46] 0.30 H.R.=1.35*** [1.24-1.46] 0.61 H.R.=1.84*** [1.68-2.01] Alkaline phosphatase [> 345 IU/L] 0.87 H.R.=2.39*** [2.04-2.80] Gamma-glutayltransferase [> 195 IU/L] 0.46 H.R.=1.58*** [1.44-1.72] 0.47 H.R.=1.60*** [1.46-1.75] 0.14 H.R.=1.15** [1.06-1.25] Hematocrit [< 25%] 0.75 H.R.=2.11*** [1.89-2.35] Platelets [< 100 k/mm 3 ] 1.35 H.R.=3.85*** [3.49-4.24] 1.35 H.R.=3.85*** [3.49-4.26] 0.44 H.R.=1.55*** [1.43-1.68] Creatinine [> 1.4 mg/dL] 0.70 H.R.=2.02 *** [1.86-2.20] Estimated Glomerular Filtration Rate [< 60 mL/min/1.73 m 2 ] -0.15 H.R.=0.86* [0.76-0.99] -0.14 H.R.=0.870* [0.760-0.99] Sodium [< 125 mEq/L] 0.92 H.R.=2.51*** [2.16-2.91] 66 Serum Iron [> 140 mcg/dL] 0.12 H.R.=1.13** [1.04-1.23] 0.12 H.R.=1.14** [1.05-1.25] Ferritin [> 300 ng/mL] 0.11 H.R.=1.12** [1.04-1.20] Glycosylated hemoglobin [> 7%] -0.16 H.R.=0.85** [0.76-0.95] Alpha fetoprotein [> 144 ng/mL] 1.50 H.R.=4.48*** [3.48-5.77] 1.44 H.R.=4.24*** [3.27-5.52] 0.87 H.R.=2.39*** [2.06-2.78] a. Lab tests that were included in the initial model specifications as potential risk factors but which never achieved statistical significance in any risk prediction model include: hemoglobin, white blood cell count, potassium *p<0.05; **p<0.01; ***p<0.0001 Table 5-4 Impact of Abnormal Laboratory Values on the Risk of Adverse Events Cirrhosis Decompensated Cirrhosis Liver-Related Hospitalization Hepatocellular Carcinoma N=123,988 N=128,055 N=128,769 N=128,481 Number of Events [%] 17,926 [14.5%] 8,429 [6.6%] 28,730 [22.3%] 4,517 [3.5%] Estimated Coefficients, HAZARD RATIOS and [95% Confidence Intervals] DRUG THERAPY 0.02 H.R.=1.02 [0.90-1.15] 0.00 H.R.=1.00 [0.91-1.10] -0.21 H.R.=0.81*** [0.76-0.87] -0.01 H.R.=0.99 [0.87-1.12] LAB TESTS a Direct bilirubin [> 3.9 mg/dL] 0.41 H.R.=1.51** [1.11-2.06] Albumin [< 3 g/dL] 1.05 H.R.=2.86*** [2.46-3.34] 1.34 H.R.=3.82*** [3.48-4.20] 1.03 H.R.=2.79*** [2.58-3.03] 0.78 H.R.=2.19*** [1.90-2.51] AST/ALT Ratio [> 1] 0.37 H.R.=1.45*** [1.29-1.63] 0.96 H.R.=2.60*** [2.34-2.88] 0.45 H.R.=1.57*** [1.47-1.67] 0.64 H.R.=1.90*** [1.67-2.17] Alkaline phosphatase [> 345 IU/L] 0.39 H.R.=1.48** [1.15-1.90] 0.65 H.R.=1.92** [1.28-2.86] 67 Gamma- glutayltransferase [> 195 IU/L] 0.59 H.R.=1.81*** [1.60-2.04] 0.38 H.R.=1.46*** [1.33-1.59] 0.30 H.R.=1.35*** [1.26-1.44] 0.36 H.R.=1.44*** [1.27-1.64] Red blood cell count [< 2.19 M/mm 3 ] -1.31 H.R.=0.27* [0.08-0.94] Platelets [< 100 k/mm 3 ] 1.70 H.R.=5.45*** [4.80-6.19] 1.55 H.R.=4.71*** [4.29-5.17] 0.76 H.R.=2.14*** [2.00-2.30] 1.10 H.R.=3.01*** [2.64-3.42] Neutrophils [< 30] -0.27 H.R.=0.76* [0.58-0.99] Serum Iron [> 140 mcg/dL] 0.26 H.R.=1.30*** [1.15-1.46] Ferritin [> 300 ng/mL] 0.16 H.R.=1.17** [1.04-1.32] 0.30 H.R.=1.35*** [1.20-1.52] Total Iron Building Capacity [> 400 mcg/dL] 0.19 H.R.=1.21** [1.07-1.38] Alpha fetoprotein [> 144 ng/dL] 0.97 H.R.=2.64*** [1.84-3.79] 1.55 H.R.=4.71*** [3.99-5.57] 1.13 H.R.=3.11*** [2.64-3.67] 2.88 H.R.=17.87*** [14.94-21.36] Estimated Glomerular Filtration Rate [< 60 mL/min/1.73 m 2 ] -0.27 H.R.=0.76** [0.63-0.91] Blood Urea Nitrogen [> 75 mg/dL] -1.90 H.R.=0.15** [0.04-0.51] *p<0.05; **p<0.01; ***p<0.0001 a Indicates a significant reduction in treatment effectiveness if treatment is initiated after the patient’s first recorded abnormal laboratory test for albumin, AST/ALT ratio, platelets, GGT or alpha fetoprotein. Estimates of the impact of initiating treatment before and after the emergence of an abnormal laboratory tests are presented in [Table 5-5]. The second column of Table 5-5 repeats the average results of drug therapy effectiveness found in Table 5-3 and 5-4 to assist in interpreting results. The last two columns of Table 5-5 report our estimates of the effectiveness of drug therapy started ‘early’ [before any abnormal laboratory value is reported] or ‘late’ [following an abnormal laboratory test]. There are two tests of significance provided for the 68 estimated effectiveness of ‘late’ treatment. The first measure of statistical significance denoted by ‘*’ footnotes designates whether or not the estimated effect is different from the null hypothesis. The second statistical test, designated by ‘a’ indicates whether or not the estimated effects for early and late initiation of drug therapy are statistically different. That is, did delaying the initiation of treatment adversely impact the effectiveness of initiating treatment? There are two basic results of significance. First, the initiation of treatment prior to any laboratory test becoming abnormal is uniformly protective, but limited. Early treatment reduced the risk of the composite clinical event is 20% [H.R. = 0.80; CI=(0.70-0.90] while the reduction in the risk of death is 22% [H.R.=0.78; CI=(0.65-0.93)]. The limited impact of initiating treatment is not surprising given that only 16.4% of patients who initiate treatment achieve an undetectable viral load. Second, initiating standard HCV treatment after one or more laboratory test becomes abnormal never has a significant impact on the risk of these events. That is, delaying standard treatment until the patient exceeds any one of the lab test upper bounds listed in Table 5-2 significantly degrades the effectiveness of treatment. 5.5 DISCUSSION Tools are needed with which to manage newly approved drug therapies. Our results also suggest that delaying therapy until the patient’s lab profile becomes ‘abnormal’ could compromise patient outcomes. Our results demonstrating the limited effectiveness of standard therapy are also consistent with earlier research using VA data [11] and other ‘real world’ treatment settings.[12, 13] Treatment rates in our VA sample using standard therapy are only 24.3% and few treated patients [16.4%] achieved at least one undetectable level of viral load test [4% of the total HCV study population]. It is no surprise that asymptomatic HCV patients have 69 gambled in the past that they will not develop HCV complications in order to avoid the treatment burden associated with standard therapy. New treatment options have reversed the dynamics of the treatment decision. Patients are now demanding treatment with expensive, efficacious and less burdensome therapies while payers seek to ration treatment to high risk patients. A cost-containment strategy of reserving treatment for patients ‘abnormal’ lab tests cannot be implemented because the thresholds used here to define abnormal lab tests were set high to avoid controversy about whether or not the test was truly abnormal. More research is needed to set the screening values for each of the 5 test in this analysis at a level high enough to predict increased event risk but low enough to avoid inappropriate delays in treatment that impact the effectiveness of treatment. Previous research has found similar relationships between the risk of liver events and abnormal laboratory test results based on relatively small samples of patients with disease progression. Lee, et al. [27] found that elevated ALT levels were associated with an increased risk of HCC in Taiwan. Hu, et al.[28] found that elevated AST and alpha-feta protein [AFP] were associated with stages 3-4 fibrosis in Taiwan patients. In a Japanese study of 459 patients, Imazeki, et al. [29] found that increasing fibrotic stage was highly correlated with the risk of all- cause mortality, but baseline ALT, AST, albumin and platelet counts were not statistically significant when fibrosis stage was entered as an explanatory variable. In a second small Japanese study [n=345], Shiratori, et al. [30] found elevated albumin to be protective for HCC and elevated AST to increase the risk of death. In a small study in Spain [N=568]. Fernandez- Rodriguez, et al. [31] found that baseline albumin [<3.9 g per 100 ml] was associated with a three-fold increase in the risk of long-term liver complications. In a study of treated patients, van der Meer, et al. found that an increasing AST/ALT ratio was associated with increased risk of 70 mortality.[32] A study that used VA data found that abnormal laboratory values for albumin, AST, AST/ALT ratio, creatinine clearance, and sodium were associated with mortality in genotype 1 patients.[33] Finally, the HALT-C study found that changes from baseline for platelets, AST/ALT ratio, bilirubin and albumin were predictive of decompensated cirrhosis and liver-related death or liver transplantation.[34] 5.5 LIMITATIONS There are several additional and important limitations in our study. First, most HCV cohorts from the US would not match well with our study sample which consist almost exclusively of men. Second, treatment for HCV patients in the VA health care system may vary from treatment patterns found in the civilian health care system. However, the Veteran Administration is the largest provider of care to chronically HCV-infected patients in the US. [35] The asymptomatic nature of HCV and the retrospective design of this study make it likely that the index date identified here will not correspond to the actual date of HCV infection. However, this uncertainty regarding the actual date of infection is common in clinical practice. Most patients with hepatitis C are unaware of their infection until their exposure to HCV is detected by routine screening of blood donations or their health status deteriorates due to the infection. The changing sensitivity of HCV viral load tests also presented a challenge in defining an ‘undetectable’ viral load. While many older tests have a lower threshold of 600, newer tests are sensitive up to 10 IU/mL. We chose to apply a cutoff of 25 as the limit of ‘detectable’ for all patients to be consistent, but it is highly likely that some patients with older labs will be mis- categorized due to the low sensitivity of the tests in use at that time. 71 Finally, our study does not capture medical care outside the VA system, such as the Medicare program. If patients received treatment for complications of liver disease outside the VA, these data may not have been captured and time to event data will be accurate or missing. However, we believe that most of the care for HCV patients was provided within VA system due to the complete coverage for prescription medications and the very high cost of HCV related medications. 72 5.5 REFERENCES [1] Lavanchy D: The global burden of hepatitis C. Liver Int 2009, 29(Suppl 1):74-81. [2] Chen SL, Morgan TR: The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006, 3(2):47-52. [3] Armstrong GL, Wasley A, Simard EP, McQuillan GM, Kuhnert WL, Alter MJ: The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med 2006, 144(10):705-714. [4] Seeff LB: The history of the "natural history" of hepatitis C (1968-2009). Liver Int 2009, 29(Suppl 1):89-99. [5] Klevens RM, Hu DJ, Jiles R, Holmberg SD: Evolving epidemiology of hepatitis C virus in the United States. Clin Infect Dis 2012, 55(Suppl 1):S3-9. [6] Smith BD, Morgan RL, Beckett GA, Falck-Ytter Y, Holtzman D, Teo CG, et al: Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012, 61(RR-4):1-32. [7] CDC. Hepatitis C Information for Health Professionals. Available from: http://www.cdc.gov/hepatitis/HCV/HCVfaq.htm#section1. Accessed October 31, 2014. [8] Davis GL, Albright JE, Cook SF, Rosenberg DM: Projecting future complications of chronic hepatitis C in the United States. Liver Transpl 2003, 9(4):331-338. [9] Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB: Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis 2011, 43(1):66-72. [10] Ng V, Saab S. Effects of a sustained virologic response on outcomes of patients with chronic hepatitis C. Clin. Gastroenterol. Hepatol. 2011, 9(11):923-30) [11] Kramer JR, Kanwal F, Richardson P, Mei M, El-Serag HB: Gaps in the achievement of effectiveness of HCV treatment in national VA practice. J Hepatol 2012, 56(2):320-325. [12] Shepherd J, Brodin H, Cave C, Waugh N, Price A, Gabbay J. Pegylated interferon alpha- 2a and -2b in combination with ribavirin in the treatment of chronic hepatitis C: a systematic review and economic evaluation. Technol Assess. 2004;8(39):iii-iv, 1-125. 73 [13] Borroni G, Andreoletti M, Casiraghi MA, Ceriani R, Guerzoni P, Omazzi B, et al. Effectiveness of pegylated interferon/ribavirin combination in 'real world' patients with chronic hepatitis C virus infection. Aliment Pharmacol Ther. 2008;27(9):790-7. [14] Gaetano JN, Reau N. Hepatitis C: management of side effects in the era of direct-acting antivirals. Curr Gastroenterol Rep. 2013 15(1):305. doi: 10.1007/s11894-012-0305-1. [15] McHutchison JG, Everson GT, Gordon SC, Jacobson IM, Sulkowsi M, Kauffman R, et al. Telaprevir with peginterferon and ribavirin for chronic HCV genotype 1 infection. N Engl J Med 2009;360:1827-38. [16] Poordad F, McCone J, Bacon BR, Bruno S, Manna MP, Sulkowski MS, et al. Boceprevir for Untreated Chronic HCV Genotype 1 Infection. N Eng J Med 2011;364(13):1195-206. [17] Bacon BR, Gordon SC, Lawitz E, Marcellin P, Vierling JM, Zeuzem S, et al. Boceprevir for Previously Treated Chronic HCV Genotype 1 Infection N Eng J Med 2011;364(13):1207-17. [18] Jacobson IM, McHutchison JG., Dusheiko G, Di Bisceglie AM, Reddy KR, Bzowei NH, et al. Telaprevir for Previously Untreated Chronic Hepatitis C Virus Infection. N Engl J Med 2011;364(25):2405. [19] Zeuzem S, Andreone P, Pol S, Lawitz, E, Diago, M, Roberts, S, et al. Telaprevir for Retreatment of HCV Infection. N Eng J Med 2011;364(25):2417-28 [20] Jesudian AB, de Jong YP, Jacobson IM. Emerging therapeutic targets for hepatitis C virus infection. Clin Gastroenterol Hepatol. 2013 Jun;11(6):612-9. [21] Doyle JS, Aspinall E, Liew D, Thompson AJ, Hellard ME. Current and emerging antiviral treatments for hepatitis C infection. Br J Clin Pharmacol. 2013 Apr;75(4):931-43. [22] Backus LI, Gavrilov S, Loomis TP, Halloran JP, Phillips BR, Belperio PS, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009, 16(6):775-783. [23] Vallet-Pichard A, Mallet V, Nalpas B, Verkarre V, Nalpas A, Dhalluin-Venier V, et al., FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology 2007, 46(1):32-36. [24] Sterling RK, Lissen E, Clumeck N, Sola R, Correa MC, Montaner J, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 2006;43:1317-1325. 74 [25] Cox DR. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological). 1972;34(2):187-220. [26] Allison, PD. (2010). Survival analysis using SAS: A practical guide (2nd Edition). Cary, NC, USA: SAS Institute Inc. [27] Lee MH, Yang HI, Lu SN, Jen CL, Yeh SH, Liu CJ, Chen PJ, You SL, Wang LY, Chen WJ et al: Hepatitis C virus seromarkers and subsequent risk of hepatocellular carcinoma: long- term predictors from a community-based cohort study. J Clin Oncol 2010, 28(30):4587-4593. [28] Hu SX, Kyulo NL, Xia VW, Hillebrand DJ, Hu KQ. Factors associated with hepatic fibrosis in patients with chronic hepatitis C: a retrospective study of a large cohort of US patients. J Clin Gastroenterol 2009, 43(8):758-764. [29] Imazeki F, Yokosuka O, Fukai K and Saisho H. Favorable prognosis of chronic hepatitis C after interferon therapy by long-term cohort study. Hepatology 2003;38(8):493-502. [30] Shiratori Y, Ito Y, Yokosuka O, Imazeki, F, Nakata, R, Tanaka, N, et al. Antiviral therapy for cirrhotic hepatitis C: Association with reduced hepatocellular carcinoma development and improved survival. Ann Inter Med 2005;142(2):105-114. [31] Fernandez-Rodriguez CM, Alonso S, Martinez SM, Forns X, Sanchez-Tapias JM, Rincón D, et al. Peginterferon plus ribavirin and sustained virological response in HCV-related cirrhosis: Outcomes and factors prediction response. Am J Gastroenterol 2010;105(10):2164-2172. [32] van der Meer AJ, Veldt BJ, Feld JJ, Wedemeyer H, Dufour JF, Lammert F, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA 2012, 308(24):2584-2593. [33] Backus LI, Boothroyd DB, Phillips BR, Belperio P, Halloran J, Mole LA. A sustained virologic response reduces risk of all-cause mortality in patients with hepatitis C. Clin Gastroenterol Hepatol. 2011;9(6):509-516. [34] Ghany MG, Kim HY, Stoddard A, Wright EC, Seeff LB, Lok AS. Predicting clinical outcomes using baseline and follow-up laboratory data from the hepatitis C long-term treatment against cirrhosis trial. Hepatology. 2011;54(5):1527-1537 [35] Center for Quality Management in Public Health. The state of care for veterans with chronic Hepatitis C. In. Palo Alto, California: US Department of Veteran Affairs, Public Health Strategic Health Care Group, Center for Quality Management in Public Health; 2010. 75 CHAPTER 6 : USING THE FIB-4 SCORE TO MONITOR MORBIDITY AND MORTALITY RISK IN CHRONIC HEPATITIS C PATIENTS Tara Matsuda 1, 2 , Jeffrey McCombs 1 , Ivy Tonnu-Mihara 2 , Sammy Saab 3 , Patricia Hines 4 , Gilbert L’Italien 4 , Anupama Kalsekar 4 , Yong Yuan 4, 1. Department of Clinical Pharmacy and Pharmaceutical Economics and Policy, School of Pharmacy and the Leonard Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, California, United States of America 2. Veterans Affairs Long Beach Healthcare System, Long Beach, California, United States of America 3. Departments of Medicine and Surgery, David Geffen School of Medicine, University of California-Los Angeles, Los Angeles, California, United States of America 4. World Wide Health Economics and Outcomes Research, Bristol-Myers Princeton New Jersey, United States of America The final version of this paper has been accepted into the Journal of Virology & Retro Virology. Further acknowledgement to Timothy R. Morgan, M.D., from the VA Long Beach Healthcare System for his support of the project. 6.1 ABSTRACT Introduction: In the recent past, many patients with chronic Hepatitis C infections [HCV] delayed treatment until symptoms emerged due to the significant direct and indirect costs associated with older treatments. Physicians may have acquiesced to delays in therapy as clinical research data on new treatment options began to emerge in the literature. However, delaying therapy must be coupled with valid methods for monitoring the patient’s disease progression over time to insure that mortality and morbidity risks do not become untenable. Objective: This study examined if FIB-4 values can be used to monitor the risk of liver-related events and mortality in HCV patients. 76 Methods: A cohort of patients was selected from the Veterans Administration’s Hepatitis C clinical registry of confirmed HCV patients. The primary outcomes were time to death and time to the first liver-related clinical event. Cox proportional hazards models were estimated using time to a patient’s FIB-4 exceeding 1.45 or 3.25 as the risk factor of interest. Cox models controlled for age, gender, genotype, race, ethnicity, BMI, prior hospitalizations and HIV and hepatitis B co-infection. Results: 187,860 patients met study requirements. Patients whose FIB-4 level exceeded 3.25 were at significantly higher risk of death [Hazard Ratio (HR) = 3.56 (3.47-3.65)] and an adverse liver-related clinical event [HR = 4.01 (3.92-4.10)]]. Exceeding FIB-4>1.45 was also associated with a significant but smaller increased risk of death [HR = 2.27 (2.21-2.33)] and the composite event [HR = 2.23 ([2.18-2.28)]. Conclusion: FIB-4 is a significant predictor of risk, even at the lower threshold (1.45). 6.2 INTRODUCTION Hepatitis C (HCV) is associated with significant global public health and economic burden, affecting approximately 130-170 million people worldwide and an estimated 3.2 to 7 million people in the United States [1-4]. HCV patients are at risk of developing progressive liver disease and related complications such as cirrhosis, liver failure, liver transplantation, hepatocellular carcinoma (HCC), and death [1,2,5-12]. Many patients with chronic HCV infections may fail to develop symptoms over their lifetime that would require medical treatment. The rate of progression to cirrhosis is estimated to be about 5-20% in chronic patients at 20 years of infection [13]. In the recent past, most patients with chronic HCV infections delayed treatment until symptoms emerged due to the significant direct and indirect costs associated with older 77 treatments. Physicians were thought to delay therapy as clinical research data on new treatment options began to emerge in the literature. These treatment options are now available but at very high cost per course of therapy. This welcomed change in clinical options has created a demand for information on how best to focus limited resources on HCV patients at highest risk for adverse clinical events, especially in managed care plans and European countries. The purpose of this study is to evaluate the feasibility of using a single clinical marker to monitor the progression of HCV as measured by the risk of future adverse events and death. Specifically, we investigate the use of the FIB-4 score as a non-invasive biomarker/predictor of future risk of liver related events and death. The FIB-4 score is used to estimate liver fibrosis stages, with a FIB-4 index > 3.25 having been found to have a positive predictive value of 82.1% to confirm the existence of significant fibrosis in a HCV infected cohort, while a FIB-4 index < 1.45 found to have a negative predictive value of 94.7% [14]. Recent research has further explored the use of FIB4 as well as gender as predictors of CHC disease progression [15, 16]. To better accomplish this goal, we expanded the sample of VA/HCV patients included in this study by eliminating the previous restriction that patients have a detectable viral load at baseline, which increased the number of patients available for study by 46%. Using this large VA cohort, our hypothesis is that a FIB-4 index value of greater than 1.45 and 3.25 are associated with increased risk of liver-related complications and death. Multivariate statistical models will be estimated to better clarify the utility of FIB-4 as clinical marker for HCV progression. These models will generate estimates of the effects of numerous other risk factors. Previous research has documented the impact of age, male gender, alcohol consumption, HIV co-infection and a fatty liver on the likelihood of disease progression [17]. BMI and Hispanic ethnicity have been found to be associated with disease progression 78 [18], while African Americans may have a lower rate of disease progression relative to white patients [18-20]. Results from our earlier analysis of 128,769 VA patients with detectable viral loads at baseline found that black patients were at lower risk for the composite late-stage liver event [HR=0.72 (0.71-0.72)] and death [HR=0.65 (0.62-0.67)] than white patients. But more importantly, this study documented that achieving viral suppression reduced risk of the composite clinical endpoint by 27% [HR=0.73 (0.66-0.82)] and the risk of death by 45% [HR=0.55 (0.47-0.64)][21]. The impact of viral genotype on the risk of future liver-related events and death is much less clear. Preliminary data suggested patients with genotype 1 may be at higher risk of disease progression [22]. However, follow up studies did not confirm these observations [5. 23]. More recent studies have found that genotype 3 carries an increased risk of worse clinical outcome [24-26]. Our previous study using VA patients found that patients with genotype 2 were at significantly lower risk, and patients with genotype 3 were at higher risk for all study outcomes relative to genotype 1 (p<0.01 for all estimates)[21]. Other studies have looked at the impact of laboratory tests on disease progression and mortality, include albumin, AST/ALT ratio, and platelets [27-29]. A second analysis of the VA data identified 5 laboratory tests for which abnormal values were found to be associated with increased risk. The estimated hazard ratios for the composite of liver-related complications/death were 1.35/1.84 for the AST/ALT ratio > 1; 2.35/5.01 for albumin < 3 g/Dl; 1.58/1.15 for GGT > 195 IU/L; 3.85/1.55 for platelet count < 100 k/mm 2 and 4.48/2.39 for alpha fetoprotein > 144 ng/mL. But more importantly, this analysis determined that starting drug therapy after any one of the above lab tests became abnormal significantly reduced the risk reduction associated with treatment initiation than if patients were treated before the abnormal lab. 79 6.3 METHODS DATA The data used in this study were taken from the Veterans Administration [VA] Clinical Case Registry (CCR) for HCV infected patients. The study was approved by the Institutional Review Board for the VA Long Beach Healthcare System (VALB HS). This retrospective cohort data was de-identified before being made available to the research team. Moreover, the research team members responsible for conducting the analyses using the de-identified data [Matsuda, Tonnu-Mihara] were required to conduct all analyses on site at the VA in Long Beach. Potential HCV patients were identified by the presence of an HCV-related ICD-9 diagnosis code or a positive viral load assessment using the Hepatitis C Antibody Test, the Hepatitis C recombinant immunoblot assay [RIBA] or the Qualitative Hepatitis C RNA Test. A local CCR coordinator then manually confirmed or rejected the patient for HCV/CCR inclusion. Upon this confirmation, all historical data from the patient’s electronic medical record (EMR) were pulled and added to the CCR. The VA EMR system was fully implemented in 1999 and the data period for this study covers the entire time period over which EMR data were available from all VA regions from 1999 to 2010 [30]. An intermediate patient-level analytic database was created consisting of summary variables for each month before and after the patient’s CCR enrollment [index date]. The following summary data were created: 6. Patient demographic data (age in months at baseline, gender, race, ethnicity): Race and ethnicity data were based on patient self-report as recorded in the VA EMR system and were included in the analysis based on previous research documenting race and ethnicity as risk factors for liver-related clinical events. 80 7. The patient’s diagnostic profile was created consisting of monthly dichotomous variables reflecting the diagnoses recorded each month. 8. Monthly dichotomous variables were created for hospital admissions for any diagnosis and for liver related diagnoses. 9. Prescription drug data were used to create monthly variables indicating when patients received HCV-related treatment [peg-interferon alfa [2a or 2b], interferon alfa [2a or 2b], interferon alfacon-1]. The use of ribavirin alone was not considered to be a drug therapy for HCV. 10. The objective of treatment is to suppress the patients HCV viral load to undetectable levels. Another important factor of interest of this research was to document the impact of viral load suppression, while taking into account the temporal relationship between achieving an undetectable viral load and event date. To achieve this, we specified undetectable viral load as a time dependent variable. This specification should present a more practical improvement in the real world data analysis relative to sustained viral response [SVR], the gold standard for measuring treatment response in clinical trials. Whether or not the patient has achieved an undetectable viral load will be updated in the Cox model whenever a more recent measurement is available regardless of the interval between tests. This time-dependent specification can help better capture the long-term sustainability of viral suppression beyond 6 months. SAMPLE SELECTION CRITERIA Study patients were screened for baseline data for at least 6 months prior to their index date and sufficient data to calculate one or more FIB-4 score at some point in longitudinal data. These data were then used to measure time-to-events and estimate the impact of achieving two 81 alternative FIB-4 levels: 1.45 and 3.25. Time to an undetectable viral load was calculated and included in the analysis of event risk. PATIENT OUTCOMES HCV infected patients are at risk for progressive liver disease and related complications such as cirrhosis, liver failure, hepatocellular carcinoma (HCC) and death. 1, 2, 5-7 Therefore, the patient outcomes specified for this analysis were all-cause mortality and liver-related morbidity [a composite of newly diagnosed cirrhosis (compensated or decompensated), HCC, or a liver- related hospitalization]. The time to the composite event was set at the earliest event date for any of the composite events. Monthly dichotomous variables were created for the outcomes of the study based on recorded diagnostic codes [e.g., diagnosis of cirrhosis, etc.] and selected CPT-4 codes included in data from hospital admissions and outpatient services. Hospitalizations were defined being liver-related if the primary diagnosis for the hospitalization was found in Appendix 1. Cirrhosis and HCC events were compiled by searching the inpatient, outpatient and problem lists for ICD-9 codes 571.5, 571.2, 571.6 and 155,155.1, 155.2, respectively. Decompensated cirrhosis was defined as a diagnosis of cirrhosis combined with a diagnosis of hepatic coma [70.44, 71.71, 348.3, 348.31, 572.2], portal hypertension [572.3], hepatorenal syndrome [572.4], jaundice [782.4], ascites [789.59], or esophageal varices [456, 456.1, 456.2, 456.21] or a FIB-4 score > 3.25. The FIB-4 score was also segmented into three categories which was previously found to correctly classify nearly 73% of liver biopsies in a HCV infected cohort [14]. STATISTICAL METHODS The time-to-event variables for death and the composite event [morbidity] were analyzed using time dependent Cox proportional hazards models to test the correlation between potential 82 predictors and study endpoints. Achieving FIB-4>1.45 and FIB-4>3.25 were included in separate analyses as time-dependent independent variables. The models control for genotype, race, ethnicity, age, gender, BMI and other factors such as diagnosis of diabetes, co-infections with HIV or HBV at baseline and any hospital admission in the 6 months prior to the patient’s index date. Time to an undetected viral load (VL) was also included as an independent variable. We also conducted sensitivity analyses in which we replace the FIB4 with the AST to Platelet Ratio Index (APRI) which is a serological marker that has satisfactory sensitivity and specificity together with a high predictive value of fibrosis. The APRI can be useful either in the absence of a biopsy or to reduce the frequency with which biopsies need to be carried out to monitor the evolution of chronic hepatitis C [31]. APRI score is an easy, low cost and practical alternative method for assessing structural changes in chronic hepatitis C (CHC). The performance of APRI in predicting significant fibrosis and cirrhosis has been evaluated in various studies and patient cohorts [31-34]. 6.4 Results DESCRIPTIVE STATISTICS A total of 233,424 patients met study requirements for 6 months of data before their index date. Eighty-one percent [81%] of these patients met the criteria for a minimum of one FIB-4 measurement at some point on their longitudinal data file and comprised the study sample [N=187,860]. Table 6-1 presents the descriptive statistics for the study sample and compares these patients to patients excluded due to missing FIB-4 data. Overall, 19.7% of patients in the analytic sample received treatment at any time following HCV diagnosis and treatment rates were significantly higher in the cohort with FIB-4 scores [21%] than for patients with no FIB-4 data [14%]. The VA/HCV patients were predominately male of either white or black race [51% 83 and 28%, respectively] with nearly 18% of patients with unknown race. The mean age was 53 years [SD=8] and nearly 50% of patients were documented as genotype 1 [38% missing data]. Just under 15% of the FIB-4 cohort had no baseline FIB-4 score. Of patients with a baseline score, only 18% had an FIB-4 score > 3.25 (not shown in Table 6-1), which is correlated with a Metavir fibrosis stage of F3-F4 [14] or an Ishak fibrosis stage of F4-F6 [35]. Finally, just over 3% of 233,424 patients had an undetectable viral load at baseline and 19% of patients had no reported viral load at baseline (not shown on Table 6-1). Table 6-1 Descriptive Statistics of Patient Population With FIB-4 data N=187,860 No FIB-4 data N=45,564 Demographic characteristics N % N % p-value Age [Mean ± SD] 52.32 ± 7.84 53.88 ± 8.31 <0.0001 Male (n, %) 182014 96.89 44093 96.77 0.2003 Ethnicity N % N % Hispanic 11241 5.8 2176 4.78 <0.0001 Non-Hispanic 150086 79.9 35564 78.05 Multi-ethnic 1037 0.5 55 0.12 Unknown 25496 13.7 7769 17.05 Race N % N % White 95486 50.83 23794 52.22 <0.0001 Black 54424 28.97 11252 24.69 84 With FIB-4 data N=187,860 No FIB-4 data N=45,564 Demographic characteristics N % N % p-value Mixed 3519 1.87 445 0.98 Other 2302 1.23 690 1.51 Unknown 32129 17.1 9383 20.59 Diabetes prior 31269 16.64 7840 17.21 0.004 Hospitalization prior 36925 19.66 6335 13.90 <0.0001 Viral load at baseline N % N % Missing [no baseline readings] 38193 20.33 5240 11.50 <0.0001 Detectable 144108 76.71 38240 83.93 Undetectable 5559 2.96 2084 4.57 Ever treated 39651 21.11 6345 13.93 <0.0001 Genotype N % N % <0.0001 Missing 66663 35.49 22374 49.10 <0.0001 1 96365 51.3 18251 40.06 2 13966 7.43 2875 6.31 3 9307 4.95 1811 3.97 other 1559 0.83 253 0.56 Baseline FIB-4* N % N % 85 With FIB-4 data N=187,860 No FIB-4 data N=45,564 Demographic characteristics N % N % p-value Missing 27572 14.68 45564 100 <1.45 74702 39.76 0 0 1.45-3.25 57274 30.49 0 0 >3.25 28312 15.07 0 0 *Formula used to calculate this FIB-4 value [14]: FIB-4 = !"# (&#!'() ∗ +,- (.//1) 23!4#3#4( (56 7 /1) ∗ √+1-(.//1) PREDICTORS OF ALL-CAUSE MORTALITY The estimated hazard ratios for the risk factors for all-cause mortality are displayed in Table 6-2. The models using FIB-4>1.45 and FIB-4>3.25 are displayed side-by-side to facilitate comparing results. A patient’s FIB-4 is a significant predictor of the risk of death. Patients whose FIB-4 value exceeds 1.45 experience an increased risk of death by 127% [H.R.= 2.27, 95% CI = (2.21 – 2.33)]. Mortality risk increases significantly to +2.56% if the patient’s FIB-4 exceeds [H.R. = 3.56, 95% CI = (3.47-3.65)]. Equally important, the impacts of other risk factors on mortality risk are independent of FIB-4. Patients who achieved an undetectable viral load significantly reduced their risk of the death by between 22% and 29% relative to patients with a detectable viral load over their entire post-index period. Mortality risk increased monotonically with age, 86 while minority and Hispanic patients exhibit lower risk. Genotype 2 patients are at lower risk of death than patients infected with HCV genotype 1. Genotype 3 patients may be at higher risk of death than genotype 1 patients but the evidence is marginally significant. Patients with lower than normal and higher than normal BMI are at higher risk of death. Patients with diabetes, HIV and HBV at baseline are also at higher risk of death. Table 6-2 Impact of Risk Factors on the Risk of Death FIB-4 > 1.45 FIB-4 > 3.25 N=187,860 Number of Events [%] 29,316 [15.6%] FIB-4 > Critical value 2.27*** [2.21-2.33] 3.56*** [3.47-3.65] Achieved undetectable VL 0.71*** [0.67-0.75] 0.78*** [0.72-0.83] Gender [Male] 1.63*** [1.49-1.79] 1.65*** [1.50-1.81] Age [vs. < 45] 45-65 1.49*** [1.43-1.55] 1.54*** [1.47-1.60] >65 2.53*** [2.40-2.67] 2.64*** [2.50-2.79] Race [vs White] Black 0.71*** 0.79*** 87 FIB-4 > 1.45 FIB-4 > 3.25 [0.69-0.73] [0.77-0.82] Mixed 0.73*** [0.67-0.81] 0.77*** [0.70-0.84] Other 0.83** [0.74-0.93] 0.83** [0.74-0.94] Unknown 0.88*** [0.85-0.92] 0.91*** [0.87-0.95] Ethnicity [vs non-Hispanic] Hispanic 0.97 [0.92-1.02] 0.91* [0.87-0.96] Mixed 0.65*** [0.53-0.78] 0.61*** [0.50-0.74] Other/Unknown 2.18*** [2.09-2.27] 2.07*** [1.99-2.15] Prior Admission [6 mo.] 1.63*** [1.59-1.68] 1.63*** [1.59-1.68] HCV Genotype [vs 1] 2 0.92** [0.87-0.92] 0.95* [0.90-1.00] 3 1.08** 1.02 88 FIB-4 > 1.45 FIB-4 > 3.25 [1.02-1.14] [0.97-1.08] Missing 1.51*** [1.47-1.55] 1.50*** [1.46-1.54] other 0.91 [0.78-1.05] 0.90 [0.78-1.04] Body Mass Index < 25 1.27*** [1.24-1.31] 1.28*** [1.24-1.31] >30 1.03 [1.00-1.06] 1.02 [1.00-1.05] Missing 1.16*** [1.08-1.24] 1.11*** [1.04-1.19] Diagnosis at baseline Diabetes 1.68*** [1.64-1.73] 1.66*** [1.61-1.70] HIV 1.41*** [1.20-1.66] 1.41*** [1.20-1.66] HBV 1.07 [0.99-1.16] 1.05 [0.97-1.13] *p<0.05; **p<0.01; ***p<0.0001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. 89 PREDICTORS OF MORBIDITY The estimates for the risk of a patient experiencing the composite clinical event are presented in Table 6-3. The risk of experiencing an adverse clinical event more than doubles if FIB-4 exceeds 1.45 [H>R> = 2.23, 95% CI = (2.18-2.28)]. If the patient’s FIB-4 exceeds 3.25 at some point over time, the risk of the composite event increases four-fold [H.R.= 4.01 (3.92- 4.10)]. As with the risk of death, the morbidity risk effects of other risk factors are independent of the patient’s FIB-4 level. In particular, achieving an undetectable viral load reduced mortality risk by 29% to 32%. Male gender increased mortality risk 10-13%. Age at diagnosis monotonically decreased mortality risk, possibly due to the effect of a delayed diagnosis, or possible missing comorbidities due to secondary insurance coverage to VA coverage. Black HCV patients are at lower risk for all clinical events while Hispanic patients are at higher risk for clinical events. Genotype 2 patients were at 13-15% lower risk for experiencing the composite clinical event while genotype 3 patients were at 2-7% higher risk relative to genotype 1 patients, though significance was mixed. Increased BMI monotonically increased mortality risk. Prior hospital admissions at baseline and a baseline diagnosis of diabetes were consistent predictors of increased morbidity risk while baseline diagnoses of concomitant infection with HIV or HBV had no impact on risk in HCV patients. Table 6-3 Impact of Risk Factors on Morbidity [Composite Event] FIB-4 > 1.45 FIB-4 > 3.25 Patient Characteristics N=180,789 Number of Events [%] 52,863 [29.2%] FIB-4 > Critical Value 2.23*** 4.01*** 90 FIB-4 > 1.45 FIB-4 > 3.25 [2.18-2.28] [3.92-4.10] Achieved undetectable VL 0.68*** [0.64-0.72] 0.71*** [0.67-0.76] Gender [Male] 1.10** [1.04-1.18] 1.13** [1.06-1.20] Age [vs. < 45] 45-65 0.86*** [0.83-0.89] 0.90*** [0.88-0.93] >65 0.58*** [0.55-0.62] 0.59*** [0.55-0.63] Race [vs White] Black 0.77*** [0.75-0.79] 0.83*** [0.81-0.85] Mixed 1.11*** [1.04-1.19] 1.16*** [1.08-1.24] Other 0.84** [0.76-0.92] 0.84** [0.76-0.92] Unknown 0.61*** [0.58-0.64] 0.63*** [0.60-0.66] Ethnicity [vs non-Hispanic] 91 FIB-4 > 1.45 FIB-4 > 3.25 Hispanic 1.26*** [1.21-1.31] 1.20*** [1.16-1.26] Mixed 1.42*** [1.26-1.58] 1.38*** [1.24-1.55] Other/Unknown 0.97 [0.92-1.01] 0.94** [0.89-0.98] Prior Admission [6 mo.] 1.56*** [1.52-1.60] 1.53*** [1.50-1.57] HCV Genotype [vs 1] 2 0.85*** [0.82-0.88] 0.87*** [0.83-0.90] 3 1.07** [1.03-1.12] 1.02 [0.97-1.06] Missing 0.62**** [0.60-0.63] 0.61**** [0.59-0.62] other 0.93 [0.84-1.04] 0.92 [0.83-1.02] Body Mass Index < 25 0.93*** [0.91-0.95] 0.92*** [0.90-0.94] >30 1.05** 1.05** 92 FIB-4 > 1.45 FIB-4 > 3.25 [1.02-1.07] [1.02-1.07] Missing 0.43*** [0.39-0.48] 0.42*** [0.38-0.46] Diagnosis at baseline Diabetes 1.18*** [1.15-1.21] 1.16*** [1.13-1.19] HIV 0.89 [0.76-1.04] 0.89 [0.76-1.04] HBV 0.96 [0.89-1.04] 0.95 [0.88-1.02] *p<0.05; **p<0.01; ***p<0.0001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. SENSITIVITY ANALYSES The results from the sensitivity analyses substituting APRI cutoff points for the FIB4 values used in the main analysis are presented in Table 6-4. The results for the components of the composite event have been added to this table for both the FIB4 and APRI risk factors. The results for the APRI are unstable. Patients who exceed APRI>0.70 are at lower risk than patients who have not exceeded this value prior to any of the events studied and the results are statistically significant except for the composite event. Conversely, patients who exceed APRI> 1.0 are at higher risk than patients who do not exceed this cut point. Moreover, the estimated hazard ratios are much larger for FIB4 indicating that it is a superior predictor of future risk than the APRI. 93 Table 6-4 Sensitivity Analysis: Comparison of FIB4 and APRI EVENT Number of Events FIB-4>1.45 FIB-4>3.25 APRI>0.70 APRI>1.00 Death N=29,316 [15.6%] 2.27*** [2.21-2.33] 3.56*** [3.47-3.65] 0.87*** [0.83-0.90] 2.62*** [2.56-2.68] Composite Event N=52,863 [29.2%] 2.23*** [2.18-2.28] 4.01*** [3.92-4.10] 0.97 [0.94-1.00] 3.06*** [2.99-3.12] Cirrhosis N=25,791 [14.3%] 7.42*** [7.10-7.75] 10.14*** [9.84-10.44] 0.90*** [0.86-0.94] 7.84*** [7.60-8.09] Decompensated Cirrhosis N=12,313 [6.6%] 23.74*** [21.48-26.25] 18.54*** [17.69-19.43] 0.75*** [0.70-0.80] 12.89*** [12.27-13.55] HCC N=6,837 [3.7%] 9.02*** [8.23-9.88] 8.85*** [8.38-9.34] 0.86** [0.80-0.94] 7.59*** [7.17-5.04] Liver-related Hospitalization N=43,960 [23.4%] 1.91*** [1.86-1.95] 3.24*** [3.17-3.32] 0.93*** [0.90-0.96] 2.53*** [2.48-2.59] 6.5 DISCUSSION This study used data from a large cohort of real-world HCV patients at various stages of disease progression to investigate if FIB-4 can be used to monitor changes in patient risk for liver-related events and death. Our results confirm that the risk of liver-related events and death increased significantly with elevated FIB-4, even at a FIB-4 as low as 1.45. Healthcare systems need to prioritize immediate access to the new HCV treatment and the FIB-4 index may be a viable tool to assess liver disease risk profile and treatment prioritization. While other biomarker 94 methods of assessing liver fibrosis are available to fill the need for monitoring asymptomatic patients, we used the FIB-4 as the data for this calculation was widely available in the VA data. The FIB-4 has been shown to be superior to APRI, another well-known biomarker method [36- 38], but may be not as accurate as transient elastography (Fibroscan) methods, though evidence is mixed [36, 38-39]. However, Fibroscan has its limitations as it does require elastometry equipment that would be more costly and not as readily available as the FIB-4 calculation, and has been shown to have a high rate of failure in obese patients [36]. These results are relevant to the rational use of the newest therapies emerging onto the market with very high cure rates but very high price per patient treated. For example, sofosbuvir + ribavirin treatments are expected to cost between $86,500 and $173,000 depending on genotype and actual regiment [40], yet sofosbuvir + ribavirin with or without peg-interferon may be the most overall cost-effective approach to this disease state when taking into account of the costs of hospitalization and liver-related events [41]. It is not surprising that health insurance companies, government programs and HMOs face a significant increase in demand for HCV treatment from two sources. First, more HCV patients are clinically eligible for treatment due to the improved SVR rates and more benign side effect profile of new therapies. But more important, significant pent-up demand exists for treatment from patients who could not tolerate older treatment regimens and patients who have delayed initiating treatment awaiting the approval of more effective and more tolerable treatment alternatives. Health insurance companies and HMOs are reluctant to immediately approve treatment for all patients infected with HCV, opting instead for a watchful waiting treatment strategy to avoid the staggering costs of treating all patients quickly. Some insurance companies and government programs are adopting for the first time ever criteria for approving therapies, and requiring a minimal liver 95 biopsy threshold of stage 2 or 3 fibrosis. This is in line with the recent recommendation by the American Association for the Study of Liver Diseases and the Infectious Diseases Society of America to give priority to patients at high risk for severe liver-related and extrahepatic Hepatitis C complications [42]. LIMITATIONS This study only considers the risk of developing significant liver-related events such as progressing to a diagnosis of cirrhosis or being hospitalized for a liver related event. Even access to electronic medical records data cannot measure the risk of less well-defined HCV related ‘events’ such as chronic fatigue that can impose significant quality of life costs on HCV patients. More research is needed to tease out the relationship between viral load suppression and FIB-4 levels on these HCV-related costs. There are several important technical limitations in our study. First the VA study population differs significantly from the U.S. population, consisting mostly of non-Asian men. Therefore, result for the risk associated with gender and the catch-all category of ‘other race’ should be viewed with caution. Nevertheless, most US patients with HCV are male [3-4], and the VA is the largest provider of care to chronically HCV-infected patients in the United States [43]. We do not measure sustained viral response [SVR], which has been shown to reduce risk of mortality and disease progression [25, 27, 44]. SVR requires that an undetectable viral load be maintained for six months following the termination of treatment, a requirement that is difficult to document even in an EMR environment. Instead, we used time-dependent specification of undetectable VL variable, which is a more practical measure of viral suppression in this real world data analysis, and is a proxy for treatment in the majority of the cases. 96 This study does not estimate or control for the effects of treatment on clinical endpoints and death. This was done for two reasons. First, viral suppression without treatment is exceeding rare. Second, the parameters with which to determine if a patient completed an adequate course of therapy vary by genotype and other factors, such as allowable duration or breaks in treatment. While developing counts of continuous days of therapy have been used by this research team in the past [45], we elected to use viral load suppression as our measure of treatment success. The effect of treatment and viral suppression before and after a patient has crossed these FIB-4 thresholds is also unknown and should be investigated further. Finally, our study does not capture medical care outside the VA system, such as the Medicare program, which may cloud the relationship between viral load suppression and event risk. The potential for missing Medicare data lead us to enter age as a categorical variable and the “protective” effects of age>65 for hospitalization likely reflects the availability of Medicare coverage for this age group and is consistent with our mixed results on the effect of age on the risk of events. Conclusion Health insurance companies, managed care organizations and government health care programs are struggling to develop a rational treatment protocol that manages the new, very expensive and very effective treatments for HCV while taking advantage of these cost-effective treatment alternatives. Plans are looking for a method of optimizing access over time, treating the highest risk patients first while monitoring untreated patients for emerging risk. In order to win support of physicians, any treatment protocol designed to rationalize the use of these products must be evidenced based. Our results demonstrate a patient’s FIB-4 level may be a viable tool in this quest to provide care more efficiently. 97 6.6 REFERENCES 1. Lavanchy D. The global burden of hepatitis C. Liver Int 2009;29(Suppl 1):74-81. 2. Chen SL, Morgan TR. The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006;3(2):47-52. 3. Armstrong GL, Wasley A, Simard E, McQuillan GM, Kuhnert WL, and Alte MJ. The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med 2006;144(10):705-714. 4. Chak E, Talal AH, Sherman KE, Schiff ER, Saab S. Hepatitis C virus infection in USA: An estimate of true prevalence. Liver Int 2011;31(8):1090-1101. 5. Seeff LB. The history of the "natural history" of hepatitis C (1968-2009). Liver Int 2009;29(Suppl 1):89-99. 6. Klevens RM, Hu DJ, Jiles R, Holmberg SD. Evolving epidemiology of hepatitis C virus in the United States. Clin Infect Dis 2012;55(Suppl 1):S3-9. 7. Smith BD, Morgan RL, Beckett GA, Beckett GA, Falck-Ytter Y, Holtzman D, et al. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012;61(RR-4):1-32. 8. Butt AA, Wang X, and Moore CG. Effect of hepatitis C virus and its treatment on survival. Hepatology 2009;50(8):387-392. 9. El-Serag HB, Mason AC. Risk factors for the rising rates of primary liver cancer in the United Stats. Arch Intern Med 2000;160(21):3227-3230.. 10. Davis GL, Albright JE, Cook SF, Rosenberg DM: Projecting future complications of chronic hepatitis C in the United States. Liver Transpl 2003;9(4):331-338. 98 11. Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB. Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis 2011;43(1):66-72. 12. Lee MH, Yang HI, Lu SN, Jen CL, You SL, Wang LY, et al. Chronic hepatitic C virus infection increases mortality from hepatic and extrahepatic diseases: A community-based long-term prosepctive study. JID 2012(15 Aug);206:469-477. 13. CDC. Hepatitis C Information for Health Professionals. Available from: http://www.cdc.gov/hepatitis/HCV/HCVfaq.htm#section1. Accessed October 31, 2014. 14. Vallet-Pichard A, Mallet V, Nalpas B, Verkarre V, Nalpas A, Dhalluin-Venier V, Fontaine H, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology 2007;46(1):32-36. 15. Butt AA, Yan P, Lo Re V, 3rd, et al. Liver Fibrosis Progression in Hepatitis C Virus Infection After Seroconversion. JAMA Intern Med. Dec 8 2014. 16. Xu F, Xing, J., Moorman, A., Gordon, S., Rupp, L., Lu, M., Spradling, P., Teshale, E., Boscarino, J., Vijayadeva, V., Schmidt, M., Holmberg, S. FIB4 Score and Gender Predict the Incidence of Hepatocellular Carcinoma (HCC) among Patients with Chronic Hepatitis C Virus (HCV) Infection: Chronic Hepatitis Cohort Study (CHeCS) 65th Annual Meeting of the American Association for the Study of Liver Diseases. Boston, MA2014. 17. Dienstag JL, McHutchison JG: American Gastroenterological Association medical position statement on the management of hepatitis C. Gastroenterology 2006, 130(1):225-230. 18. Kallwitz ER, Layden-Almer J, Dhamija M, et al. Ethnicity and body mass index are associated with hepatitis C presentation and progression. Clin Gastroenterol Hepatol 2010;8(1):72-78. 99 19. Crosse K, Umeadi OG, Anania FA, Laurin J, Papadimitriou J, Drachenberget C, et al. Racial differences in liver inflammation and fibrosis related to chronic hepatitis C. Clin Gastroenterol Hepatol. 2004;2(6):463-8. 20. Sterling RK, Stravitz RT, Luketic VA, Sanyal AJ, Contos MJ, Mills AS, et al. A comparison of the spectrum of chronic hepatitis C virus between Caucasians and African Americans. Clin Gastroenterol Hepatol 2004;2(6):469-473. 21. McCombs J, Matsuda T, Tonnu-Mihara I, Saab S, Hines P, L’Italien G, et al. The risk of long-term morbidity and mortality in patients with chronic hepatitis C: results from an analysis of data from a Department of Veterans Affairs Clinical Registry. JAMA Intern Med 2014;174:204-12. 22. Kobayashi M, Tanaka E, Sodeyama T, Urushihara A, Matsumoto A and Kiyosawa K. The natural course of chronic hepatitis C: a comparison between patients with genotypes 1 and 2 hepatitis C viruses. Hepatology 1996; 23:695-699. 23. Zhou S, Terrault NA, Ferrell L, Hahn JA, Lau JY, Simmonds P, et al. Severity of liver disease in liver transplantation recipients with hepatitis C virus infection: relationship to genotype and level of viremia. Hepatology 1996;24(5):1041-6. 24. Larsen C, Bousquet V, Delarocque-Astagneau E, Pioche C and Roudot-Thoraval F. Hepatitis C virus genotype 3 and the risk of severe liver disease in a large population of drug users in France. J Med Virol. 2010;82(10):1647-54. doi: 10.1002/jmv.21850. 25. van der Meer AJ, Hansen BE, Fattovich G, JJ Feld, Wedemeyer H, Dufour JF. Et al. Reliable prediction of clinical outcome in patients with chronic HCV infection and compensated advanced hepatic fibrosis: a validated model using objective and readily available clinical parameters. Gut 2015;64:2 322-331. 100 26. Kanwal F, Kramer JR, Ilyas J, Duan Z, El-Serag HB. HCV genotype 3 is associated with an increased risk of cirrhosis and hepatocellular cancer in a national sample of U.S. Veterans with HCV. Hepatology 2014;60:98-105. 27. van der Meer AJ, Veldt BJ, Feld JJ, Wedemeyer H, Dufour JF, Lammert F, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA 2012;308(24):2584-2593. 28. Ghany MG, Lok AS, Everhart JE, Everson GT, Lee WM, Curto TM, et al. Predicting clinical and histologic outcomes based on standard laboratory tests in advanced chronic hepatitis C. Gastroenterology 2010;138:136-46. 29. Backus LI, Boothroyd DB, Phillips BR, Belperio P, Halloran J, Mole LA. A sustained virologic response reduces risk of all-cause mortality in patients with hepatitis C. Clin Gastroenterol Hepatol 2011;9:509-16 e1. 30. Backus LI, Gavrilov S, Loomis TP, Halloran JP, Phillips BR, Belperio PS, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009;16(6):775-783. 31. Borsoi Viana, M. S., Takei, K., Collarile Yamaguti, D. C., Guz, B., & Strauss, E. (2009). Use of AST platelet ratio index (APRI Score) as an alternative to liver biopsy for treatment indication in chronic hepatitis C. Ann Hepatol, 8(1), 26-31. 32. Al-Mohri, H., Cooper, C., Murphy, T., & Klein, M. B. (2005). Validation of a simple model for predicting liver fibrosis in HIV/hepatitis C virus-coinfected patients. HIV Med, 6(6), 375- 378. doi: 10.1111/j.1468-1293.2005.00330.x 101 33. Snyder, N., Gajula, L., Xiao, S. Y., Grady, J., Luxon, B., Lau, et al. APRI: an easy and validated predictor of hepatic fibrosis in chronic hepatitis C. J Clin Gastroenterol, 40(6), 535- 542. 34. Wai, C. T., Greenson, J. K., Fontana, R. J., Kalbfleisch, J. D., Marrero, J. A., Conjeevaram, H. S., & Lok, A. S. (2003). A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology, 38(2), 518-526. doi: 10.1053/jhep.2003.50346 35. Sterling RK, Lissen E, Clumeck N, Sola R, Correa MC, Montaner J, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 2006;43:1317-1325. 36. Vergniol J, Foucher J, Terrebonne E, Bernard PH, le Bail B, Merrouche W, et al. Noninvasive tests for fibrosis and liver stiffness predict 5-year outcomes of patients with chronic hepatitis C. Gastroenterology 2011;140:1970-9, 9 e1-3. 37. Holmberg SD, Lu M, Rupp LB, Lamerato LE, Moorman AC, Vijayadeva V, et al. Noninvasive serum fibrosis markers for screening and staging chronic hepatitis C virus patients in a large US cohort. Clin Infect Dis 2013;57:240-6. 38. Bonnard P, Elsharkawy A, Zalata K, Delarocque-Astagneau E, Biard L, Le Fouler L, et al. Comparison of liver biopsy and noninvasive techniques for liver fibrosis assessment in patients infected with HCV-genotype 4 in Egypt. J Viral Hepat published online : 30 JUL 2014, DOI: 10.1111/jvh.12285. 39. Sanchez-Conde M, Montes-Ramirez ML, Miralles P, Castro Alvarez JM,Bellón JM, Ramírez M, et al. Comparison of transient elastography and liver biopsy for the assessment of liver 102 fibrosis in HIV/hepatitis C virus-coinfected patients and correlation with noninvasive serum markers. J Viral Hepat 2010;17:280-6. 40. Moore C, Levitsky J. The current state and future prospects of chronic hepatitis C virus infection treatment. Curr Infect Dis Rep 2014;16:413. 41. Saab S, Gordon SC, Park H, Sulkowski M, Ahmed A, Younossi Z. Cost-effectiveness analysis of sofosbuvir plus peginterferon/ribavirin in the treatment of chronic hepatitis C virus genotype 1 infection. Aliment Pharmacol Ther 2014 [Epub ahead of print]. 42. When and in whom to initiate HCV therapy. American Association for the Study of Liver Diseases and the Infectious Diseases Society of America. Available from: http://hcvguidelines.org/full-report/when-and-whom-initiate-hcv-therapy. Accessed August 14, 2014. 43. Center for Quality Management in Public Health. The state of care for veterans with chronic Hepatitis C. In. Palo Alto, California: US Department of Veteran Affairs, Public Health Strategic Health Care Group, Center for Quality Management in Public Health; 2010. 44. Ng V, Saab S: Effects of a sustained virologic response on outcomes of patients with chronic hepatitis C. Clin Gastroenterol Hepatol 2011;9(11):923-930. 45. McCombs JS, Shin J, Hines P, Yuan Y and Saab S. Impact of Drug Therapy Adherence in Patients with Hepatitis C. Am J Pharmacy Benefits 2012;4[Special Issue]:SP19-SP27. 103 CHAPTER 7 : THE IMPACT OF DELAYED HEPATITIS C VIRAL LOAD SUPPRESSION ON PATIENT RISK: HISTORICAL EVIDENCE FROM THE VETERANS ADMINISTRATION Tara Matsuda 1, 2 , Jeffrey McCombs 1 , Ivy Tonnu-Mihara 2 , Justin McGinnis 1,2 , Steven Fox 1 1. Department of Clinical Pharmacy and Pharmaceutical Economics and Policy, School of Pharmacy and the Leonard Schaeffer Center for Health Policy and Economics, University of Southern California, Los Angeles, California, United States of America 2. Veterans Affairs Long Beach Healthcare System, Long Beach, California, United States of America The final version of this manuscript has been submitted to Forum for Health Economics & Policy. 7.1 ABSTRACT Background: The high cost of new HCV treatments has resulted in ‘Watchful waiting’ strategies being developed to safely delay treatment and viral load suppression for low risk patients. Objective: To document if delayed viral load suppression adversely impacted patient risk for adverse events and death. Methods: 187,860 patients were selected from the Veterans Administration’s [VA] clinical registry [CCR], a longitudinal compilation of records data for 1999-2010. Inclusion criteria required at least 6 months of CCR/EMR data prior to their HCV diagnosis and sufficient data post-diagnosis to calculate one or more FIB-4 scores. Primary outcome measures were time-to- death and time-to-a composite of liver-related clinical events. Cox proportional hazards models were estimated separately using three critical FIB-4 levels. 104 Results: Achieving an undetectable viral load before the patient’s FIB-4 level exceed pre- specified critical values [1.00, 1.45 and 3.25] effectively reduced the risk of an adverse clinical event by 33-35% and death by 21-26%. However, achieving viral load suppression after FIB-4 exceeds 3.25 significantly reduced the benefit of viral response. Conclusions: Delaying viral load suppression until FIB-4 > 3.25 reduces the benefits of viral load suppression in reducing patient risk. 105 7.2 INTRODUCTION Hepatitis C (HCV) imposes a significant global public health and economic burden, affecting approximately 130-170 million people worldwide and an estimated 3.2 to 7 million people in the U.S. HCV patients are at risk of developing progressive liver disease and liver- related complications such as cirrhosis, liver failure, hepatocellular carcinoma (HCC), and death Risk factors for progression in HCV patients include body mass index, ethnicity, race, genotype and viral load suppression [Lavanchy, 2009; Chen and Morgan, 2006; Armstrong, et al., 2006; Chak, et al., 2011; Seeff, 2009; Klevens, et al., 2012; Smith, et al., 2012; Butt, Wang and Moore, 2009; El-Serag and Mason, 2000; Davis, et al., 2003; Rein, et al., 2011; Lee, et al., 2012]. Newer drug therapies show promise for very high rates of sustained viral response [SVR] and shorter, more benign treatment courses. However, these new treatment options command very high prices that exceed the ability of most payers to immediately treat all infected patients. As a result, payers are seeking evidence to guide “watchful waiting” strategies that smooth out treatment costs over time, without increasing patient risk. The purpose of this study is to use historical data from 1999-2010 to evaluate whether delaying viral load suppression reduced the impact of that viral load suppression on morbidity and mortality risk. While the newer oral HCV regimens appear more effective than the interferon–based regimens used during that period in achieving viral load suppression, all treatments likely exert their beneficial effects through suppressing further viral replication. Therefore, this study examined patients who did achieve viral load suppression using the medications available before 2010, and then tested the association between clinical benefits and when (during the patient’s clinical course) viral suppression was achieved. The patient’s FIB-4 score was used to define “delayed” versus early viral load suppression. These results may help 106 provide guidance regarding whether or not health plans and government payers can used the minimally invasive FIB-4 score to monitor HCV patients and indicate, safely, when to initiate treatment with the new, very costly treatment alternatives now available. The FIB-4 is a minimally invasive biomarker based index used to estimate liver fibrosis stages. An FIB-4 index > 3.25 was found to have a positive predictive value of 82.1% for significant fibrosis, while a FIB-4 index < 1.45 was found to have a negative predictive value of 94.7% [Vallet-Pichard, et al., 2007]. FIB-4 and gender are independent predictors of chronic hepatitis C disease progression [Butt, et al., 2015; Xu, et al., 2014]. 7.3 METHODS DESIGN Our research conducted three separate retrospective cohort analyses using electronic medical records data from 1999-2010. Each repeated analyses employed progressively higher definitions of ‘elevated’ FIB-4: >1.0; >1.45 and >3.25. The analysis was designed to determine if delaying viral load suppression until after a patient’s FIB-4 became ‘elevated’ diminish the effectiveness of viral response in preventing disease progression and reducing the patient’s risk of death. We used time-to-undetectable viral load as a proxy for sustained viral response [SVR]. While this approach ignores the possibility of viral rebound and incomplete therapy, the focus of the study was to investigate the timing of treatment initiation on effectiveness as measured by risk reduction. DATA The data used in this study were taken from the Veterans Administration [VA] Clinical Case Registry (CCR) for HCV infected patients. The study was approved by the Institutional Review Board for the VA Long Beach Healthcare System [VALB HS; MIRB #1059, September 107 23, 2011]. HCV patients included in the CCR were initially identified using routine computer- based scans of the EMR data for the presence of an HCV-related ICD-9 diagnosis code [see Table 1] or a positive HCV exposure assessment using the Hepatitis C Antibody Test, the Hepatitis C recombinant immunoblot assay [RIBA] or the Qualitative Hepatitis C RNA Test. A local CCR coordinator is provided a list of all newly identified HCV patients. The local CCR coordinator can remove a patient from the CCR manually if they determine that the patient has been included in the HCV/CCR erroneously. Upon this confirmation, all historical data from the patient’s electronic medical record (EMR) were pulled and added to the CCR. The VA EMR system was fully implemented in 1999 and the data period for this study covers the entire time period over which EMR data were available from all VA regions from 1999 to 2010 [Backus, et al, 2009]. A patient-level analytic database was developed consisting of monthly summary variables created for all months before and after the patient’s CCR enrollment [index date]. These data included demographic characteristics, diagnostic profile, prescription drug profile, prior hospital admissions, time-to-treatment and time-to-viral load suppression. SAMPLE SELECTION CRITERIA Study patients records were screened for baseline data covering at least 6 months prior to their index date, and sufficient data to calculate at least one FIB-4 score at some point post- diagnosis. These data were then used to calculate time-to-elevated FIB-4 [3 levels] - which captures the patient’s changing clinical severity over time. PATIENT OUTCOMES HCV infected patients are at elevated risk for both death and progressive liver disease with related complications [Lavanchy, et al., 2006; Chen and Morgan, 2006; Seeff, 2009; 108 Klevens, et al., 2012; Smith, et al., 2012]. Therefore, this analysis specified two primary patient outcome measures: time-to-all-cause mortality, and time-to-liver-related morbidity [a composite of: newly diagnosed cirrhosis (compensated or decompensated), HCC, or a liver-related hospitalization]. Hospitalizations were defined as being liver-related if the primary diagnosis code for the hospitalization was one of those found in Table 7-1. The time to the composite event was set as the earliest event date for any of the composite events. Patients with cirrhosis or a liver-related hospitalization prior to their index date were excluded from the analysis of future clinical events. Table 7-1 List of Study Related Diagnoses and Procedure Codes Hepatitis C Diagnostic Codes and Laboratory Tests The ICD-9 codes used to identify pending hepatitis C patients are: V02.62, 070.41, 070.44, 070.51, 070.54, 070.70, and 070.71. The LOINC codes for the lab tests used to identify pending hepatitis C patients are: Hepatitis C Antibody Test 11259, 13955-0, 16128-1, 16129-9, 16936-7, 22327- 1, 33462-3, 34162-8, 39008-8, 40762-2, 5198-7, 5199-5 Hepatitis C RIBA Test 24011-9 Qualitative Hepatitis C RNA Test 11259-9, 5010-4, 5011-2, 5012-0, 6422-0 Liver-related Diagnoses for Hospital Admissions Acute or unspecified hepatitis C with hepatic coma Chronic hepatitis C with hepatic coma Other specified viral hepatitis with hepatic coma Other specified viral hepatitis without mention of hepatic coma code range Unspecified viral hepatitis with hepatic coma Unspecified viral hepatitis C code range Unspecified viral hepatitis without mention of hepatic coma Toxoplasma hepatitis Malignant neoplasm of liver, primary Malignant neoplasm of intrahepatic bile ducts Malignant neoplasm of liver, not specified as primary or secondary 109 Esophageal varices with bleeding Spontaneous bacterial peritonitis Alcoholic fatty liver Acute alcoholic hepatitis, Alcoholic cirrhosis of liver Alcoholic liver damage, unspecified Chronic hepatitis, unspecified Chronic persistent hepatitis Chronic active hepatitis Cirrhosis of liver without mention of alcohol Biliary cirrhosis (chronic nonsuppurative destructive cholangitis) Other chronic non-alcoholic liver disease Unspecified chronic liver disease without mention of alcohol Portial pyemia Hepatic coma Portal hypertension Hepatorenal syndrome Other sequelae of chronic liver disease Hepatitis in viral diseases classified elsewhere Hepatitis in other infectious diseases classified elsewhere. Hepatitis, unspecified (trauma and toxic reactions) Other specified disorders of liver Unspecified disorder of liver Jaundice Hepatomegaly Ascites Hepatitis C carrier, unspecified Liver transplant status 110 STATISTICAL METHODS The objective of this study was to determine if delayed viral load suppression was associated with diminished impact of viral response on reducing patient morbidity and mortality during the period 1999-2010. Three separate sets of analyses were conducted: These varied the definition used for ‘delayed’ viral load suppression relative to a patient’s FIB-4 score exceeding a pre-specified level (1.00, 1.45 and 3.25, respectively) at the time of initial viral response. In each analysis, HCV patients were partitioned into three groups: patients with viral load suppression before the cutoff (early suppression), those with delayed viral load suppression (after the FIB-4 cutoff), and patients who never achieved virologic suppression. The analysis was performed using interaction terms between time-to-viral load suppression and time-to-elevated FIB-4. This specification tests whether or not a significant difference exists in effectiveness of viral response between the early and delayed viral response patient as defined in each analysis. That is, patients who achieved viral load suppression are divided into early and late responders based on the temporal relationship between their time-to-response and time-to-the critical FIB-4 value specified for each analysis. The comparison group was patients who never achieved viral load suppression. Each of the time-to-event patient outcome variables were analyzed using time dependent Cox proportional hazards models. All models controlled for genotype, race, ethnicity, age, gender, BMI and other co-morbidity factors such as diagnosis of diabetes, co-infections with human immunodeficiency virus [HIV] or hepatitis B virus [HBV] at baseline, or any hospital admission in the 6 months prior to the patient’s index date. 111 7.4 RESULTS DESCRIPTIVE STATISTICS A total of 187,860 patients met study requirements. The average age of VA/HCV patients in the study was 52 years [Table 7-2]. Patients were predominately male [97%], of either white or black race [51% and 29%, respectively, with just over 17% of patients with unreported/unknown race]. Only 3% of patients had an undetectable viral load at baseline, with another 20% missing a baseline assessment. Similarly, 35% of patients did not have a documented genotype at any time during the study period, while over 51% of all patients were documented as genotype 1 [nearly 80% of patients with genotype data]. For patients with a FIB-4 score at baseline, only 15% had an FIB-4 score > 3.25 – that level correlates with a Metavir fibrosis stage of F3-F4[Vallet-Pichard, et al., 2007] or an Ishak fibrosis stage of F4-F6 [Sterling, et al., 2006]. 112 Table 7-2 Descriptive Statistics of the Study Population Demographic Characteristics N=187,860 Age [Mean ± SD] 52.32 ± 7.84 N % Male 182014 96.89 Ethnicity N % Hispanic 11241 5.8 Non-Hispanic 150086 79.9 Multi-ethnic 1037 0.5 Unknown 25496 13.7 Race N % White 95486 50.83 Black 54424 28.97 Mixed 3519 1.87 Other 2302 1.23 Unknown 32129 17.1 Diabetes diagnosis 6 months prior to index date 31269 16.64 Hospitalization 6 months prior to index date 36925 19.66 Viral load at baseline N % Missing [no baseline readings] 38193 20.33 Undetectable 5559 2.96 Detectable 144108 76.71 Ever treated for HCV post-index date 39651 21.11 Genotype N % Missing 66663 35.49 1 96365 51.3 2 13966 7.43 3 9307 4.95 other 1559 0.83 Baseline FIB-4 a N % Missing 27572 14.68 <1.45 74702 39.76 1.45-3.25 57274 30.49 >3.25 28312 15.07 113 a Formula used to calculate this FIB-4 value 25 : FIB-4 = !"# (&#!'() ∗ +,- (.//1) 23!4#3#4( (56 7 /1) ∗ √+1-(.//1) Only 21% of patients in the analytic sample received treatment at any time following their HCV diagnosis. Equally important, only 16% of treated patients with a detectable viral load achieved viral load suppression which translates into 4% of the total HCV population [McCombs, et al., 2014]. IMPACTING OF DELAYING VIRAL LOAD SUPPRESSION ON MORBIDITY [COMPOSITE EVENT] A patient’s FIB-4 history over time is highly correlated with the risk of experiencing the composite clinical event [Table 7-3]. Patients whose FIB-4 value exceeded 1.00 over time experienced a 85% increase in morbidity risk [H.R.= 1.85, 95% CI(1.80-1.90)]. Patients whose FIB-4 exceeded 1.45 over time experienced a 123% increase in morbidity risk [H.R.= 2.23, 95% CI(2.18-2.28)]. Exceeding an FIB-4 of 3.25 increased morbidity risk 226% [H.R.= 3.26, 95% CI(3.20-3.33)]. Achieving viral load suppression before the patient’s FIB-4 exceeds any of these 3 critical FIB-4 values significantly reduced the risk of the composite event by 33% to 35% versus having never achieved viral response. However, patients who achieve viral load suppression “late” experienced less benefit: the risk of experiencing the composite clinical event was reduced 36% if suppression was achieved after FIB-4>1.00 [H.R.=0.64, 95% CI(0.60-0.69)] but was only reduced by 12% if viral response is delayed until after FIB-4>3.25 [H.R.= 0.88, 95% CI (0.78- 1.00)]. That is, achieving an undetectable viral load late adversely impacted the morbidity benefit associated with achieving that viral load suppression. 114 Table 7-3 Impact of Early and Late Viral Load Suppression on Morbidity Risk [Composite Event][Adjusted Hazard Ratios and 95% Confidence Intervals] N=180,789 e Number of Events [%] 52,863 [29.2%] FIB-4 Level Used to Define Early and Delayed Treatment Initiation Patient Characteristics FIB-4 > 1 FIB-4 > 1.45 FIB-4 > 3.25 FIB-4 > Critical Value 1.85 c [1.80-1.90] 2.23 c [2.18-2.28] 3.26 c [3.20-3.33] Viral Load Suppression [vs. no treatment response] Before FIB-4 > Critical Value 0.67 c [0.59-0.75] 0.65 c [0.60-0.72] 0.67 c [0.62-0.71] After FIB-4 > Critical Value 0.64 c [0.60-0.69] 0.70 c [0.65-0.77] 0.88 a,d [0.78-1.00] Gender [Male] 1.13 c [1.06-1.21] 1.10 b [1.03-1.18] 1.13 c [1.06-1.21] Age [vs. < 45] 45-65 0.90 c [0.88-0.93] 0.86 c [0.83-0.89] 0.92 c [0.89-0.95] >65 0.70 c [0.66-0.75] 0.58 c [0.55-0.62] 0.61 c [0.57-0.65] Race [vs White] Black 0.77 c [0.75-0.79] 0.77 c [0.75-0.79] 0.82 c [0.80-0.84] Mixed 1.11 b [1.04-1.19] 1.11 b [1.04-1.19] 1.16 c [1.08-1.23] Other 0.84 c [0.76-0.93] 0.84 b [0.76-0.92] 0.84 b [0.77-0.93] Unknown 0.61 c [0.58-0.64] 0.61 c [0.58-0.64] 0.62 c [0.59-0.65] Ethnicity [vs non-Hispanic] Hispanic 1.28 c [1.23-1.34] 1.26 c [1.21-1.31] 1.23 c [1.18-1.28] Mixed 1.45 c [1.30-1.62] 1.42 c [1.27-1.58] 1.41 c [1.26-1.58] Other/Unknown 0.98 [0.93-1.02] 0.97 [0.92-1.01] 0.92 b [0.88-0.96] Prior Admission [6 mo.] 1.56 c [1.52-1.60] 1.56 c [1.52-1.60] 1.50 c [1.47-1.54] HCV Genotype [vs. type 1] 2 0.83 c [0.80-0.87] 0.85 c [0.82-0.88] 0.86 c [0.82-0.89] 115 3 1.11 c [1.06-1.16] 1.07 b [1.03-1.12] 1.04 [0.99-1.08] Missing 0.60 c [0.59-0.62] 0.62 c [0.60-0.63] 0.59 c [0.57-0.60] other 0.93 [0.83-1.03] 0.93 [0.84-1.04] 0.92 [0.83-1.02] Body Mass Index [vs. 25-30] < 25 0.94 c [0.92-0.96] 0.93 c [0.91-0.95] 0.91 c [0.89-0.94] >30 1.05 c [1.02-1.08] 1.05 b [1.02-1.07] 1.04 b [1.02-1.07] Missing 0.45 c [0.40-0.50] 0.43 c [0.39-0.48] 0.38 c [0.34-0.42] Diagnosis at baseline Diabetes 1.20 c [1.17-1.23] 1.18 c [1.15-1.21] 1.18 c [1.14-1.21] HIV 0.92 [0.78-1.07] 0.89 [0.76-1.04] 0.90 [0.77-1.05] Hepatitis B 0.97 [0.90-1.05] 0.96 [0.89-1.03] 0.97 [0.90-1.04] a p<0.05; b p<0.01; c p<0.001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. d The estimated impact of delay viral load suppression mortality was significantly reduced relative to early viral load suppression. e Sample for this analysis excluded patients with an event reported prior to their index date. IMPACT OF DELAYED VIRAL LOAD SUPPRESSION ON MORTALITY A patient’s FIB-4 history over time was highly correlated with the risk of the death [Table 7-4]. Patient’s whose FIB-4 value exceeded 1.00 over time experienced a 68% increase in morbidity risk [H.R.= 1.68, 95% CI(1.63-1.74)]. Patients whose FIB-4 exceeded 1.45 over time experienced an increased mortality risk of 129% [H.R.= 2.29, 95% CI(2.22-2.35)] while an FIB-4 > 3.25 increased risk by 202% [H.R.= 3.02. 95% CI(2.96-3.09)]. As with morbidity risk, achieving viral load suppression ‘early’ before the patient’s FIB- 4 exceeds any of these 3 critical FIB-4 values significantly reduced their risk of death, with risk reductions ranging between by 21% to 26% versus having never achieved viral response. 116 However, patients who achieve viral load suppression “late” experienced a decrease in the reduction in the risk of death associated with viral load suppression, dropping from a 38% reduction if viral response was achieved after FIB-4>1.00 [H.R.=0.62, 95% CI(0.58-0.66)] to only a 20% reduction if viral response is delayed until after FIB-4>3.25 [H.R.= 0.80, 95% CI (0.72-0.79]. That is, delaying the achievement of an undetectable viral load adversely impacted the mortality benefit associated with suppressing viral load. OTHER RISK FACTORS OF INTEREST The analyses reported in Tables 7-3 and 7-4 also provide useful information concerning the impact of other patient characteristics on morbidity and mortality risk. Male patients are consistently at higher risk than female patients. Risk of death increased with age, while the risk of clinical events decreased with age, possibly due to missing utilization data for patients over 65 covered by Medicare. Black patients were at consistently lower risk than white patients, while the impact of other ethnicity is mixed [Crosse, et al., 2004; Kallwitz, et al., 2010; sterling, et al., 2004]. Genotype 2 patients displayed lower risk relative to genotype 1 patients, while genotype 3 patients were generally at higher risk [Kanwal, et al., 2014; kobayashi, et al., 1996; Larsen, et al., 2010; Zhou, et al., 1996]. Co-infection with HIV has no impact on morbidity risk but consistently increased the risk of death. The impact of co-infection with hepatitis B [HBV] on risk was generally not significant. 117 Table 7-4 Impact of Early and Late Viral Load Suppression on Risk of Death [Adjusted Hazard Ratios and 95% Confidence Intervals] N=187,860 Number of Events [%] 29,316 [15.6%] FIB-4 Level Used to Define Early and Delayed Treatment Initiation Patient Characteristics FIB-4 > 1 FIB-4 > 1.45 FIB-4 > 3.25 FIB-4 > Critical Value 1.68 c [1.63-1.74] 2.29 c [2.22-2.35] 3.02 c [2.96-3.09] Treatment Initiation [vs. no treatment] Before FIB-4 > Critical Value 0.79 c [0.71-0.89] 0.78 c [0.71-0.85] 0.74 c [0.69-0.80] After FIB-4 > Critical Value 0.62 c, d [0.58-0.66] 0.66 c, d [0.61-0.72] 0.80 c [0.72-0.90] Gender [Male] 1.69 c [1.54-1.86] 1.63 c [1.49-1.79] 1.64 c [1.49-1.80] Age [vs. < 45] 45-65 1.60 c [1.53-1.67] 1.49 c [1.43-1.55] 1.55 c [1.49-1.62] >65 3.09 c [2.93-3.27] 2.53 c [2.40-2.68] 2.72 c [2.57-2.87] Race [vs White] Black 0.71 c [0.69-0.73] 0.71 c [0.69-0.73] 0.78 c [0.76-0.81] Mixed 0.74 c [0.67-0.81] 0.73 c [0.67-0.81] 0.76 c [0.69-0.84] Other 0.83 b [0.74-0.93] 0.83 b [0.74-0.93] 0.84 c [0.75-0.94] Unknown 0.88 c [0.85-0.92] 0.88 c [0.85-0.92] 0.90 c [0.86-0.93] Ethnicity [vs non-Hispanic] Hispanic 0.99 [0.94-1.05] 0.97 [0.92-1.02] 0.93 b [0.87-0.98] Mixed 0.67 c [0.55-0.81] 0.65 c [0.53-0.78] 0.62 c [0.51-0.75] Other/Unknown 2.21 c [2.12-2.30] 2.18 c [2.09-2.27] 2.04 c [1.96-2.13] Prior Admission [6 mo.] 1.73 c [1.69-1.78] 1.71 c [1.66-1.75] 1.62 c [1.58-1.66] HCV Genotype [vs type 1] 2 0.89 c [0.85-0.94] 0.92 b [0.87-0.97] 0.94 a [0.89-0.99] 118 3 1.12 c [1.06-1.18] 1.08 b [1.02-1.14] 1.04 a [0.99-1.10] Missing 1.46 c [1.42-1.50] 1.51 c [1.47-1.55] 1.45 c [1.42-1.49] other 0.89 [0.77-1.03] 0.91 [0.78-1.05] 0.90 [0.77-1.04] Body Mass Index < 25 1.28 c [1.25-1.32] 1.27 c [1.24-1.31] 1.27 c [1.23-1.31] >30 1.03 [1.00-1.06] 1.03 [1.00-1.06] 1.02 a [0.99-1.05] Missing 1.19 c [1.11-1.27] 1.16 c [1.08-1.24] 1.05 [0.98-1.12] Diagnosis at baseline Diabetes 1.72 c [1.67-1.77] 1.68 c [1.64-1.73] 1.67 c [1.63-1.72] HIV 1.44 c [1.22-1.70] 1.41 c [1.20-1.66] 1.42 c [1.21-1.68] Hepatitis B 1.08 a [1.00-1.17] 1.07 [1.00-1.16] 1.06 [0.98-1.15] a p<0.05; b p<0.01; c p<0.001. Estimates adjusted for the risk factors in the table for which individual results are presented and for the patient’s baseline diagnostic profile. d The estimated impact of delay viral load suppression mortality was significantly reduced relative to early viral load suppression. 7.5 DISCUSSION This study used data from a large cohort of real-world HCV patients treated before 2010 to investigate whether delaying the suppression of viral load until later stages of disease progression [based on FIB-4 level] impacted patient risk. These analyses controlled for other risk factors, including genotype. We included a wide range of liver-related events in the composite clinical morbidity event. Viral load suppression and FIB-4 levels entered into the analyses as time-dependent variables and were measured over as long as 10 years, depending on each patient’s available data. While the suppression of viral load appears to be beneficial regardless of its timing relative to a patient’s FIB-4 level, we also found that delaying viral 119 suppression until after the patient’s FIB-4 increased to 3.25 was associated with a significant reduction in the risk benefit from suppressing the patient’s viral load. It is also worth noting that a ‘watchful waiting’ treatment algorithm that sets the threshold for initiating treatment at the point when FIB-4 exceeds 1.00 would trigger immediate treatment for nearly 90% of the VA HCV patient population [see Table 7-2]. Most payers will be unable to afford immediate treatment for 90% of their HCV patients. Conversely, restricting treatment initiation until FIB-4 > 3.25 delays treatment, and therefore the benefits of viral suppression, for 65% of all patients in this sample - but at some increased risk [Table 7-2]. While we are left to speculate how the new, highly effective therapies available today may be impacted by the patient’s FIB-4 level at the initiation of therapy, our results clearly support the use of the FIB-4 to monitor untreated patients if ‘watchful waiting’ strategies are to be used to smooth out the resource demand crisis created by the cost of new treatments. Our results also document that patients with an FIB-4 under 1.45 [47% of patients with baseline measurements] do not suffer a significant increase in either morbidity or mortality risk if viral load suppression was delayed. Other minimally invasive biomarker-based methods to assess liver fibrosis are also available to monitor asymptomatic patients. FIB-4 was selected since the necessary input data for was generally available in the VA records. FIB-4 has also been shown to be superior to the aspartate aminotransferase [AST] to Platelet Ratio Index [APRI], another well-known biomarker method [Vergniol, et al., 2011; Holmberg, et al., 2013; Bonnard, et al., 2015; Ghany, et al., 2010]. Conversely, FIB-4 may be not as accurate as transient elastography (Fibroscan) methods, although evidence is mixed [Vergniol, et al., 2011; Bonnard, et al., 2015; Sanchez- Conde, et al., 2010]. However, Fibroscan has its limitations as it requires costly elastometry 120 equipment and is therefore not as readily available as the more easily calculable FIB-4. Fibroscan has also been shown to have a high rate of failure in obese patients [Vergniol, et al., 2011]. Therefore, our results appear highly relevant to the ongoing policy debate regarding how best to use the new oral HCV therapies now entering the market. It is no surprise that payers such as health insurance companies, government programs and health maintenance organizations [HMOs] face a significant and immediate increase in demand for HCV treatment. Significant pent-up demand exists for newer treatments from patients who could not tolerate older treatment regimens or did not achieve viral load suppression with treatment [McCombs, et al. 2014]. In addition, there are many patients who have delayed initiating treatment awaiting the approval of more effective and more tolerable treatment alternatives, often on advice of their physicians [Loftus, 2013]. Health insurance companies and HMOs are reluctant to immediately approve treatment for all patients infected with HCV. At least 3 million patients in the U.S. are infected by HCV, and immediate treatment for all patients could easily cost hundreds of billions of Dollars. Thus, some insurance companies and government programs are opting instead to adopt specific criteria for approving therapies in order to avoid this treatment cost bolus. For example, some health insurance companies may require a minimal liver biopsy threshold of stage 2 or 3 fibrosis before treatment is authorized. This is in line with the recent recommendations by the American Association for the Study of Liver Diseases and the Infectious Diseases Society of America, which give priority to patients at high risk for severe liver-related and extra-hepatic Hepatitis C complications American Association for the Study of Liver Diseases and the Infectious Diseases Society of America, 2014]. 121 Our results document that the FIB-4 index can predict clinical events without incurring the cost and risks of a more invasive liver biopsy. Liver-related events and death increased significantly with elevated FIB-4, even at a FIB-4 as low as 1.00. Given the need for healthcare systems to prioritize who requires immediate access to the new HCV treatment, the FIB-4 index should be considered as an additional tool to assess liver disease risk profile and treatment priority. Patients with elevated FIB-4 values should be monitored closely for decompensation and be strongly considered for antiviral therapy. This study only considered the risk of developing significant liver-related events, such as progressing to a diagnosis of cirrhosis or being hospitalized for a liver related illness. Even access to electronic medical records data cannot measure the risk of less well-defined HCV related ‘events’ such as chronic fatigue - which can impose significant quality of life costs on HCV patients. More research is needed to tease out the relationship between viral load suppression and FIB-4 levels on these other HCV-related burdens. LIMITATIONS There are also several important technical limitations to our study. First the VA study population differs significantly from the HCV affected U.S. population, consisting mostly of non-Asian men. Therefore, the results quantifying the risks associated with gender and the catch-all category of ‘other race’ should be viewed with caution. Nevertheless, most US patients with HCV are male [Armstrong, et al., 2006; Chak, et al.,2011], and the VA is the largest provider of care to chronically HCV-infected patients in the United States [Center for Quality Management in Public Health, 2010]. We also did not measure sustained viral response [SVR], which has previously been shown to reduce risk of mortality and morbidity [van der Meer, et al., 2015; van der Meer et al, 122 2012; Ng and Saab, 2011; Backus, et al., 2011]. Achieving SVR requires that an undetectable viral load be maintained for six months following the termination of treatment, a requirement that is difficult to document even in an EMR environment. Instead, we used time-dependent specification of at least one undetectable viral load, which is a more practical measure of viral suppression using real world data. Our study also does not capture medical care outside the VA system, such as the Medicare program, which may cloud the relationship between viral load suppression and event risk. The missing Medicare data prompted us to enter age as a categorical variable. This re- captured some of that influence as an apparent “protective” effect of age>65 – accounting for missing Medicare data on clinical services related to HCV this age group. This factor is consistent with our mixed results for the effect of age on the risk of events. This was a retrospective cohort study, using regression analysis to assess the influence of achieving viral load suppression on the risk of morbidity and mortality. We did control for a number of patient demographic and clinical characteristics. However possible treatment selection bias still cannot be excluded. There may remain unmeasured differences between patients who received treatment (and therefore achieved viral suppression) and those that did not. Nevertheless, the key finding appears well supported – achieving viral load suppression late in the clinical course is strongly associated with reduced morbidity and mortality benefits. This is likely to be true regardless of the treatment regimen used to achieve viral suppression. CONCLUSIONS Health insurance companies, managed care organizations and government health care programs are struggling to develop a rational treatment protocol that manages the new, highly effective, but very expensive treatments for HCV. Payers are seeking a method to optimizing 123 access over time: treating the highest risk patients first, while monitoring untreated patients for emerging risk. Our results demonstrate that delaying viral load suppression increased the risk of poor outcomes and can decrease the effectiveness of treatment – even when it suppresses viral replication. Monitoring a patient’s FIB-4 level may offer a viable tool in the quest to deliver patient centered, but also rational and efficient HCV care. 124 7.6 REFERENCES 1. American Association for the Study of Liver Diseases and the Infectious Diseases Society of America. When and in whom to initiate HCV therapy. <http://hcvguidelines.org/full- report/when-and-whom-initiate-hcv-therapy> Accessed August 14, 2014. 2. Armstrong GL, Wasley A, Simard EP, et al. The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med. 2006;144(10):705-714. 3. Backus LI, Boothroyd DB, Phillips BR, et al. A sustained virologic response reduces risk of all-cause mortality in patients with hepatitis C. Clin Gastroenterol Hepatol 2011;9:509- 516.e1. 4. Backus LI, Gavrilov S, Loomis TP, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc. 2009;16(6):775-783. 5. Bonnard P, Elsharkawy A, Zalata K, et al. Comparison of liver biopsy and noninvasive techniques for liver fibrosis assessment in patients infected with HCV-genotype 4 in Egypt. J Viral Hepat. 2015;22:245-253. 6. Butt AA, Yan P, Lo Re V, 3rd, et al. Liver Fibrosis Progression in Hepatitis C Virus Infection After Seroconversion. JAMA Intern Med. 2015;172(2):178-185. 7. Butt AA, Wang X, and Moore CG. Effect of hepatitis C virus and its treatment on survival. Hepatology. 2009;50(8):387-392. 8. Center for Quality Management in Public Health. The state of care for veterans with chronic Hepatitis C. In. Palo Alto, California: US Department of Veteran Affairs, Public Health Strategic Health Care Group, Center for Quality Management in Public Health; 2010. 125 9. Chak E, Talal AH, Sherman KE, et al Hepatitis C virus infection in USA: An estimate of true prevalence. Liver Int. 2011;31(8):1090-1101. 10. Chen SL, Morgan TR. The natural history of hepatitis C virus (HCV) infection. Int J Med Sci. 2006;3(2):47-52. 11. Crosse K, Umeadi OG, Anania FA, et al. Racial differences in liver inflammation and fibrosis related to chronic hepatitis C. Clin Gastroenterol Hepatol. 2004;2(6):463-8. 12. Davis GL, Albright JE, Cook SF, et al. Projecting future complications of chronic hepatitis C in the United States. Liver Transpl. 2003;9(4):331-338. 13. El-Serag HB, Mason AC. Risk factors for the rising rates of primary liver cancer in the United Stats. Arch Intern Med. 2000;160(21):3227-3230. 14. Ghany MG, Lok AS, Everhart JE, et al. Predicting clinical and histologic outcomes based on standard laboratory tests in advanced chronic hepatitis C. Gastroenterology. 2010;138:136- 46. 15. Holmberg SD, Lu M, Rupp LB, Lamerato LE, Moorman AC, Vijayadeva V, et al. Noninvasive serum fibrosis markers for screening and staging chronic hepatitis C virus patients in a large US cohort. Clin Infect Dis 2013;57:240-6. 16. Kallwitz ER, Layden-Almer J, Dhamija M, et al. Ethnicity and body mass index are associated with hepatitis C presentation and progression. Clin Gastroenterol Hepatol. 2010;8(1):72-78. 17. Kanwal F, Kramer JR, Ilyas J, et al. HCV genotype 3 is associated with an increased risk of cirrhosis and hepatocellular cancer in a national sample of U.S. Veterans with HCV. Hepatology. 2014;60:98-105. 126 18. Klevens RM, Hu DJ, Jiles R, et al. Evolving epidemiology of hepatitis C virus in the United States. Clin Infect Dis. 2012;55(Suppl 1):S3-9. 19. Kobayashi M, Tanaka E, Sodeyama T, et al. The natural course of chronic hepatitis C: a comparison between patients with genotypes 1 and 2 hepatitis C viruses. Hepatology. 1996; 23:695-699. 20. Larsen C, Bousquet V, Delarocque-Astagneau E, et al. Hepatitis C virus genotype 3 and the risk of severe liver disease in a large population of drug users in France. J Med Virol. 2010;82(10):1647-54. 21. Lavanchy D. The global burden of hepatitis C. Liver Int. 2009;29(Suppl 1):74-81. 22. Lee MH, Yang HI, Lu SN, et al. Chronic hepatitic C virus infection increases mortality from hepatic and extrahepatic diseases: A community-based long-term prosepctive study. JID. 2012;206:469-477. 23. Loftus P. Patient dilemma: Treat hepatitic C now or hold out? Wall Street Journal, March 4, 2013. http://www.wsj.com/articles/SB10001424127887323293704578330712442353712. Accessed May 30, 2015. 24. McCombs J, Matsuda T, Tonnu-Mihara I, et al. The risk of long-term morbidity and mortality in patients with chronic hepatitis C: results from an analysis of data from a Department of Veterans Affairs Clinical Registry. JAMA Intern Med. 2014;174:204-12. 25. Ng V, Saab S: Effects of a sustained virologic response on outcomes of patients with chronic hepatitis C. Clin Gastroenterol Hepatol. 2011;9(11):923-930. 26. Rein DB, Wittenborn JS, Weinbaum CM, et al. Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis. 2011;43(1):66-72. 127 27. Sanchez-Conde M, Montes-Ramirez ML, Miralles P, et al. Comparison of transient elastography and liver biopsy for the assessment of liver fibrosis in HIV/hepatitis C virus- coinfected patients and correlation with noninvasive serum markers. J Viral Hepat. 2010;17:280-286. 28. Seeff LB. The history of the "natural history" of hepatitis C (1968-2009). Liver Int. 2009;29(Suppl 1):89-99. 29. Smith BD, Morgan RL, Beckett GA, et al. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep. 2012;61(RR-4):1-32. 30. Sterling RK, Lissen E, Clumeck N, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology. 2006;43:1317-1325. 31. Sterling RK, Stravitz RT, Luketic VA, et al. A comparison of the spectrum of chronic hepatitis C virus between Caucasians and African Americans. Clin Gastroenterol Hepatol. 2004;2(6):469-473. 32. Vallet-Pichard A, Mallet V, Nalpas B, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology. 2007;46(1):32-36. 33. van der Meer AJ, Hansen BE, Fattovich G, et al. Reliable prediction of clinical outcome in patients with chronic HCV infection and compensated advanced hepatic fibrosis: a validated model using objective and readily available clinical parameters. Gut. 2015;64:2 322-331. 128 34. van der Meer AJ, Veldt BJ, Feld JJ, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA. 2012;308(24):2584-2593. 35. Vergniol J, Foucher J, Terrebonne E, et al. Noninvasive tests for fibrosis and liver stiffness predict 5-year outcomes of patients with chronic hepatitis C. Gastroenterology. 2011;140:1970-1979.e3. 36. Xu F, Xing, J., Moorman, A., et al. FIB-4 Score and Gender Predict the Incidence of Hepatocellular Carcinoma (HCC) among Patients with Chronic Hepatitis C Virus (HCV) Infection: Chronic Hepatitis Cohort Study (CHeCS) 65th Annual Meeting of the American Association for the Study of Liver Diseases. Boston, MA. 2014. 37. Zhou S, Terrault NA, Ferrell L, et al. Severity of liver disease in liver transplantation recipients with hepatitis C virus infection: relationship to genotype and level of viremia. Hepatology. 1996;24(5):1041-6. 129 CHAPTER 8 : INTERNAL VALIDATION OF RISK PREDICTION MODEL DEVELOPED IN VA POPULATION 8.1 ABSTRACT Background: Expensive new drug treatments for Hepatitis C may lead to the use of risk- prediction models as tools with which clinicians and payers can ration therapy based on patient risk. Two previously developed models have looked at risk for morbidity and mortality in HCV patients treated at Veteran Affairs hospitals, based on select patient demographics and HCV viral characteristic, as well as the impact of a non-invasive fibrosis index, the FIB-4 score. Objective: The purpose of this study is to perform an internal validation of the model previously developed with the addition of the FIB04 score, and evaluate its discriminatory performance. Methods: A retrospective cohort of HCV patients was followed longitudinally using data from the VA Clinical Case Registry. K-fold cross validation was used (k=15), and area under ROC curve (AUROC) analysis was used to evaluate the performance of the REVEAL-HCV risk models. Results: 128,769 patients met all study inclusion criteria. Little variation in the AUROC indicates a relative stable model. The mean AUROC of the base models were generally low, ranging from a low of 0.53 in the cirrhosis model to a high of 0.68 in the hospitalization model. Adding the baseline FIB-4 score greatly increased the predictive accuracy of the model, leading to a range from 0.68 in the mortality model to 0.88 in the decompensated cirrhosis model. Conclusion: This study highlights that while demographic and the viral factors of genotype and viral load are important, they do not give a complete picture of a patient’s risk for mortality or 130 morbidity complications of HCV. The addition of a non-invasive index such as the FIB-4 score, or possibly liver lab tests are necessary additions to a predictive risk model. 8.2 INTRODUCTION Based on the 2010 National Health and Nutrition Examination (NHANES) survey, an estimated 3.5 million people in the United States (US) were Hepatitis C virus (HCV) positive in 2010, with 2.3 million having evidence of a chronic HCV infection [1, 2]. However, HCV infection has been estimated to be as high as 5.2 million as the NHANES survey does not include high-risk groups such as the incarcerated and the homeless [3]. Hepatitis C is an asymptomatic and slow progressing disease, where decades may pass before serious liver damage is developed, if at all. About 20% of those with chronic infections will go on to develop cirrhosis, but the time frame in which this happens is highly variable and dependent on host, viral and environmental characteristics [4-6]. Of these patients who develop cirrhosis, some will advance to hepatocellular cancer or end stage liver disease. As such, HCV is the most common cause of chronic liver disease and cirrhosis, and is the most common indication for liver transplantation in the US [7, 8]. However, the majority of HCV infected persons will go throughout life without experiencing these serious liver-related consequences. The incidence of HCV has remained relatively stable over the years after peaking in the 1980s following the routine testing of the blood supply for HCV [9]. Reflecting this peak, the majority of those infected were born in the years 1945 to 1965, the ‘baby boomer’ cohort, with a prevalence of about 3.5%, as reflected in Figure 8-1 [1, 10]. 131 Figure 8-1 Distribution of HCV prevalence, by age group As this skewed cohort of HCV infected people age, HCV morbidity and mortality are estimated to increase and peak around 2030 [11]. Fortunately, there have been recent advancements in drug therapy for hepatitis C which have significantly improved the willingness of patients to initiate therapy due to a shorter and less toxic course of treatment and very high rates of sustained viralogic response (SVR), the ‘cure’ proxy. However, these new treatment options command a very high price, with typical 12 week treatments for treatment-naïve patients ranging from $83,319 to $150,000 [12]. The high price tag of these cures and the slow progression of acute disease has caused some payers to begin triaging treatment to patients at risk for adverse events and imposing treatment guidelines [13]. Risk prediction modeling in hepatitis C was, until recently, designed to provide clinicians with valid data with which to motivate patients to initiate therapy. However, these new drug options have transformed the use of risk-prediction models into tools with which clinicians and payers can triage therapy based on patient risk, thus spreading the enormous cost required to treat 0 0 0.5 0.3 0.5 1.4 3.5 0.5 0.2 0 1 2 3 4 <18 19 to 24 25 to 29 30 to 24 35 to 39 40 to 44 45 to 65 66 to 74 75+ HCV prevalence (%) Age groups (years) HCV prevalence, by age group NHANES 2009-2010 from Ditah et al. 2014 132 all HCV patients over multiple years. We previously investigated the risk for morbidity and mortality in HCV patients treated at Veteran Affairs (VA) hospitals, based on select patient demographics and HCV viral characteristics [14]. The primary results from these earlier studies documented that achieving viral load suppression significantly reduced patient risk and that a non-invasive fibrosis index, the FIB-4 score, was a significant predictor of patient risk for HCV related morbidity and mortality. The purpose of this study is to use these earlier results to guide the estimation of risk prediction models in HCV and perform internal validations of these models to evaluate their discriminatory performance. 8.3 METHODS DATA This validation study employs the retrospective cohort of HCV patients from the VA clinical case registry (CCR) data system used previously to evaluate the impact of viral load suppression and the impact of FIB-4 scores. The VA had established the CCR to identify cohorts of veterans with targeted conditions at the local and national level, review clinical status and medical outcomes, and provide opportunities for improving care, identifying potential HCV patients by the presence of an HCV-related ICD-9 diagnosis code or a positive lab test. [15]. Data elements of the CCR include: demographics (age, gender, race, ethnicity, and geographic region), height and weight, inpatient admission data (primary diagnosis ICD-9 codes), outpatient visits (diagnosis ICD-9 and procedure CPT-4 codes), prescriptions drugs, and the problem file. CCR from 1999 through 2010 were used in this analysis [maximum follow-up time of 11 years]. All patients were screened for an HCV viral load and HCV genotype test and only those patients with a reported quantifiable viral load or a detectable genotype were included in this study. The 133 first occurrence of either result became the ‘index date’ for each patient, which was a proxy of the date of HCV infection. STATISTICAL ANALYSIS The risk factors used in the models are taken from earlier research and outlined in Table 8.1. The primary outcomes specified were all-cause mortality and a composite of newly diagnosed cirrhosis (compensated and decompensated), HCC, or a liver-related hospitalization. The secondary outcomes included the individual morbidity elements of the composite. Cox proportional hazard regression analysis was used to assess the association between risk factors and time to event, and a time-dependent variable was used to measure impact of viral load suppression. Table 8-1 Risk Factors Included In Model • Age (in months) • Gender • Race - White, Black, other • Pre-index hospital admission (within 6 months) • Genotype (GT) - GT 1, GT2, GT3, Other GT • Diabetes (at baseline) • Undetected VL (time-dependent, undetected measured as <45 IU/mL) • FIB-4 score - <1.45, 1.45-3.25, >3.25 GT, Genotype; VL, Viral Load; IU, international units; mL, milliliter; FIB-4, Fibrosis-4 A k-fold cross validation process was employed [k=15] in which the study sample was randomly partitioned into 15 sub-samples. In k-fold cross validation, all sub-samples except one are used for the ‘training set’ on which the model is developed, with the left out sample used as 134 the ‘test set’ on which the model is validated. This process is repeated k-1 times, each time leaving out a different sub-sample as the test set until each one is used. From each model, the probability of an event by time t is calculated by P(t|X) = 1- S 0 (t) exp[Σ(Bi*Xi)-Σ(Bi*Xbari)] , where So(t) is the average survival in training set [16]. The 5 year and 10 years risks were used to evaluate the discriminatory ability of the model and calculate the sensitivity and specificity, and summarized by the area under the receiver-operating characteristic (AUROC) curve. The ROC curve has been used often in biomedical informatics research to evaluate models for decision support, diagnosis, and prognosis [17]. The area under the ROC curve is a widely used summary index that measures the probability of a correct ranking of a normal (N), abnormal (A) pair of cases: Pr(X A >X N ), where X A is the degree of suspicion from the abnormal case and X N for the normal case [18]. In terms of this paper, the area under the ROC curve represents the probability that the diagnostic index of patient with a specified outcome e.g. cirrhosis, will be greater than that of a patient with no cirrhosis diagnosis. This process was done using the base model and then repeated to include the baseline FIB-4 score. 8.4 RESULTS 128,769 patients met all study inclusion criteria, including a detectable viral load and data on genotype at baseline. This patient population has been described previously [14]. Briefly, the mean age of the sample was 52 years and the population was majority men (97%) of white (51%) or black (31%) race, and a genotype 1 HCV infection (79%). The predictive models are provided in Tables 8-2 and 8-3. Table 8-2 Risk Factors For Base Model (Hazard Ratio) Risk Factor Composite Death Cirrhosis Decompensated Cirrhosis Hospitalization HCC 135 Age 1.00 1.06 1.02 1.04 0.99 1.07 Male 1.11 1.58 1.35 1.81 1.09 3.41 Black race 0.72 0.65 0.54 0.42 0.74 0.73 Other race 0.65 1.20 0.73 0.63 0.58 0.80 Hospital admission 1.60 1.73 1.02 1.26 2.05 1.07 GT2 0.77 0.80 0.64 0.56 0.80 0.52 GT3 1.11 1.17 1.24 1.42 1.10 1.63 GT Other 0.89 0.96 0.87 0.93 0.89 0.77 Diabetes 1.22 1.57 1.38 1.42 1.19 1.31 Undetected VL 0.73 0.55 0.62 0.48 0.71 0.62 Table 8-3 Risk Factors For Base Model Plus FIB-4 Score (Hazard Ratio) Risk Factor Death Composite HCC Cirrhosis Decompensated Cirrhosis Hospitalization Age 1.003 0.999 1.003 0.999 1 0.998 Male 1.431 1.072 3.724 1.171 1.527 1.058 Black race 0.705 0.73 0.825 0.542 0.495 0.786 Other race 1.237 0.65 0.865 0.749 0.673 0.576 Hospital admission 1.61 1.558 1.008 0.952 1.127 1.931 GT2 0.845 0.829 0.586 0.701 0.704 0.874 GT3 0.998 0.992 1.252 0.975 1.047 1.002 GT Other 0.916 0.911 0.808 0.919 0.975 0.892 Diabetes 1.502 1.184 1.223 1.295 1.289 1.159 Undetected VL 0.552 0.721 0.65 0.656 0.501 0.686 FIB-4 >1.45 1.386 1.491 3.594 2.934 5.121 1.28 FIB-4 >3.25 3.967 4.037 16.432 13.164 37.376 3.039 136 The predictive accuracy of each training set for base model and the FIB-4 model, as measured by the AUROC for their 5 and 10 year risks, are presented in Tables 8-4 and 8-5. There was not much variation in the AUROC, with the greatest range seen in the 5-year base HCC model (0.51 to 0.58), whose AUROC differed at the 0.10 level. The 5-year composite FIB- 4 model also had AUROC that were significantly different at the 0.10 level, though the AUROC ranged only from 0.73 to 0.76. The AUROC were not significantly different at the 0.05 level for all models. The mean AUROC of the base models were generally low (Table 8-6), with the cirrhosis model have the lowest AUROC at 0.53 for both the 5 year and 10 year risks. The highest AUROC were seen in the hospitalization model at 0.66 and 0.68. Adding the baseline FIB-4 score greatly increased the predictive accuracy of the model. The least accurate model was the mortality model with an average AUROC of around 0.68 for both time points. The model with the greatest predictive accuracy was the decompensated cirrhosis model, with AUROC of around 0.88 for both time points. Table 8-4 AUROC For Base Model Death Composite HCC Cirrhosis Decompensated Cirrhosis Hospitalization k 5 year 10 year 5 year 10 year 5 year* 10 year 5 year 10 year 5 year 10 year 5 year 10 year 0 0.55 0.56 0.58 0.59 0.57 0.55 0.54 0.54 0.58 0.58 0.66 0.68 1 0.53 0.55 0.56 0.57 0.51 0.51 0.51 0.51 0.55 0.55 0.65 0.66 2 0.54 0.56 0.59 0.61 0.56 0.54 0.55 0.55 0.58 0.57 0.67 0.70 3 0.55 0.57 0.58 0.59 0.55 0.53 0.53 0.52 0.57 0.55 0.67 0.69 4 0.56 0.58 0.58 0.58 0.54 0.52 0.54 0.53 0.57 0.56 0.67 0.69 5 0.54 0.56 0.58 0.60 0.53 0.52 0.53 0.55 0.57 0.57 0.66 0.68 6 0.55 0.57 0.57 0.59 0.55 0.54 0.53 0.53 0.57 0.57 0.66 0.68 7 0.53 0.56 0.58 0.61 0.55 0.54 0.53 0.55 0.58 0.58 0.67 0.71 8 0.54 0.55 0.59 0.61 0.55 0.52 0.54 0.52 0.60 0.58 0.67 0.69 9 0.55 0.56 0.57 0.56 0.58 0.56 0.53 0.51 0.58 0.55 0.66 0.66 10 0.55 0.56 0.58 0.57 0.56 0.55 0.55 0.53 0.60 0.58 0.67 0.67 11 0.53 0.55 0.58 0.57 0.53 0.53 0.53 0.55 0.58 0.58 0.66 0.66 12 0.56 0.57 0.58 0.58 0.53 0.50 0.53 0.51 0.58 0.55 0.66 0.67 137 13 0.52 0.55 0.56 0.59 0.55 0.52 0.52 0.50 0.57 0.55 0.65 0.69 14 0.55 0.56 0.57 0.59 0.58 0.57 0.53 0.52 0.59 0.59 0.66 0.69 p 0.44 0.53 0.14 0.27 0.09 0.49 0.21 0.47 0.31 0.74 0.53 0.22 *Significant difference in AUROC at the 0.10 level Table 8-5 AUROC For Base Model Plus FIB-4 Score Death Composite HCC Cirrhosis Decompensated Cirrhosis Hospitalization k 5 year 10 year 5 year* 10 year 5 year 10 year 5 year 10 year 5 year 10 year 5 year 10 year 0 0.69 0.67 0.74 0.76 0.80 0.79 0.80 0.82 0.88 0.88 0.74 0.76 1 0.68 0.70 0.75 0.77 0.79 0.82 0.81 0.82 0.87 0.90 0.75 0.77 2 0.67 0.66 0.76 0.79 0.82 0.82 0.80 0.82 0.88 0.89 0.75 0.79 3 0.69 0.69 0.74 0.76 0.80 0.80 0.81 0.82 0.87 0.88 0.75 0.77 4 0.69 0.68 0.74 0.75 0.82 0.83 0.80 0.81 0.89 0.90 0.74 0.75 5 0.69 0.69 0.74 0.77 0.80 0.82 0.81 0.83 0.87 0.89 0.73 0.76 6 0.69 0.68 0.74 0.77 0.80 0.80 0.82 0.85 0.88 0.89 0.73 0.76 7 0.67 0.68 0.75 0.79 0.78 0.80 0.81 0.83 0.88 0.89 0.74 0.78 8 0.68 0.66 0.75 0.78 0.79 0.78 0.81 0.81 0.88 0.88 0.75 0.77 9 0.67 0.66 0.73 0.75 0.80 0.80 0.80 0.81 0.87 0.87 0.73 0.75 10 0.68 0.68 0.75 0.77 0.79 0.81 0.81 0.82 0.88 0.90 0.76 0.78 11 0.66 0.65 0.73 0.73 0.78 0.79 0.78 0.80 0.87 0.89 0.73 0.74 12 0.69 0.66 0.75 0.76 0.80 0.80 0.82 0.82 0.88 0.89 0.75 0.77 13 0.68 0.68 0.73 0.79 0.77 0.79 0.79 0.80 0.86 0.88 0.73 0.79 14 0.67 0.68 0.73 0.78 0.78 0.80 0.82 0.84 0.87 0.89 0.74 0.78 P 0.66 0.68 0.096 0.33 0.52 0.90 0.19 0.31 0.84 0.93 0.14 0.23 *Significant difference in AUROC at the 0.10 level Table 8-6 Mean AUROC Of Base Model And Base Model Plus FIB-4 5 year 10 year Event Base model Base + FIB-4 Base model Base + FIB-4 Death 0.54 0.68 0.56 0.67 Composite 0.58 0.74 0.59 0.77 HCC 0.55 0.79 0.53 0.78 Cirrhosis 0.53 0.81 0.53 0.82 Decompensated cirrhosis 0.58 0.88 0.57 0.89 Hospitalizations 0.56 0.74 0.66 0.77 138 8.5 DISCUSSION This study used data from a large cohort of real-world HCV patients at various stages of disease progression to internally validate several models that have looked at different outcomes of HCV infection. The k-fold cross validation methods gave evidence of relatively stable estimates across the 15 sub-samples that the models were trained and tested in. The discriminatory ability of the models was evaluated using AUROC techniques. The practical range of the area under the ROC curve is from 0.5 to 1, with an area of 0.5 representing an accuracy of 50/50 chance, and 1 representing perfect accuracy [17]. The arbitrary guidelines on AUROC are that values less than 0.70 are considered sub-optimal, 0.7 to 0.8 are considered good, and values greater than 0.8 considered excellent [16]. According to these guidelines, it was found that while the base model of demographics and viral suppression turned out sub- optimal results, the addition of the baseline FIB-4 score greatly improved the performance. While the mortality model remained sub-optimal, the composite, HCC, and hospitalization model had ‘good’ predictive accuracy; and the cirrhosis models had ‘excellent’ predictive accuracy. These validation studies tell us that while demographic factors such as gender and race and the viral factors of genotype and viral load are important, they do not give a complete picture of a patient’s risk for mortality or morbidity complications of HCV over time. This study also highlights the usefulness of a non-invasive index of fibrosis level such as the FIB-4 score. Using the common liver lab test of ALT, platelets, and AST, the FIB-4 score gives the status of the liver [19, 20], and in conjunction with demographics and viral load, may help predict other morbidity events. 139 Risk-prediction models play an important role in medicine, with many clinicians already employing the use of risk-prediction models to manage chronic diseases such as cardiovascular disease, and aid in treatment decisions [21-26]. Current AASLD guidelines give ‘highest’ priority to patients with METAVIR fibrosis scores of F3-F4 [27], but some insurance programs have made this mandatory as well as putting other restrictions and contraindications [13]. This method treats the sickest of HCV patients, but it would be useful to also treat those who have a high risk of HCV morbidity events before they have progressed to that point. This would in turn save both the patients and payers money in the long run. There have been other models that have looked at mortality and various endpoints in HCV clinical disease progression, but these have all been relatively smaller studies [23-26, 28, 29]. LIMITATIONS There are several important limitations in our study. First, the VA study population differs significantly from the U.S. population, most notably by the overwhelming majority of men. Nevertheless, the VA is the provider of care to the largest cohort of chronically HCV- infected patients in the US, so there is still much use for this application. Secondly, the index date may not be a good proxy for the date of diagnosis, and we are probably catching veterans at a later time in course of infection. However, the index date is representative of the majority of patients who are diagnosed well after the initial date of infection due to the asymptomatic nature of HCV infection. There are also important limitations that are associated with retrospective database studies. The sensitivity of HCV viral load tests have improved over time, with many older tests having a lower threshold of 600 while newer tests are sensitive up to 10 IU mL -1 . It is also likely that there may be missing clinical data as the VA sample as this study is also highly dependent on ICD-9 codes to identify outcomes. This missing data issue is further augmented, 140 as the CCR does not capture medical care outside the VA system. If patients developed complications of liver disease outside the VA, the data may not have been captured. However, we believe that because VA system is inclusive and provides significant benefits at lower costs to patients compared to private plans, most of the care was continued in VA facilities. We are also limited by the follow-up time in the VA sample. Outcomes such as decompensated cirrhosis and HCC typically take decades to develop following the point of infection, and with a maximum follow-up time of 11 years, this time frame is probably catching only a fraction. Future work While this process of internal validation is important, the use of a risk-prediction model also depends critically on validation of the model’s predictive power using data from an external population outside of the VA. To be widely used, a risk-prediction model must predict well outside the medical system in which it was developed. This external validation would address concerns about the specialized population in the VA and be able provide evidence of the model’s true ability to generalize. 141 8.6 REFERENCES 1. Ditah I, Ditah F, Devaki P, Ewelukwa O, Ditah C, Njei B, Luma H, Charlton M: The Changing Epidemiology of Hepatitis C Virus Infection in the United States: National Health and Nutrition Examination Survey 2001 through 2010. J Hepatol 2013. 2. Denniston MM, Jiles RB, Drobeniuc J, Klevens RM, Ward JW, McQuillan GM, Holmberg SD: Chronic hepatitis C virus infection in the United States, National Health and Nutrition Examination Survey 2003 to 2010. Ann Intern Med 2014, 160(5):293-300. 3. Chak E, Talal AH, Sherman KE, Schiff ER, Saab S: Hepatitis C virus infection in USA: an estimate of true prevalence. Liver Int 2011, 31(8):1090-1101. 4. Seeff LB: Natural history of chronic hepatitis C. Hepatology 2002, 36(5 Suppl 1):S35-46. 5. Lauer GM, Walker BD: Hepatitis C virus infection. N Engl J Med 2001, 345(1):41-52. 6. Poynard T, Bedossa P, Opolon P: Natural history of liver fibrosis progression in patients with chronic hepatitis C. The OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups. Lancet 1997, 349(9055):825-832. 7. Chen SL, Morgan TR: The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006, 3(2):47-52. 8. Kim WR, Lake JR, Smith JM, Skeans MA, Schladt DP, Edwards EB, Harper AM, Wainright JL, Snyder JJ, Israni AK et al: OPTN/SRTR 2013 Annual Data Report: liver. Am J Transplant 2015, 15 Suppl 2:1-28. 9. Armstrong GL, Alter MJ, McQuillan GM, Margolis HS: The past incidence of hepatitis C virus infection: implications for the future burden of chronic liver disease in the United States. Hepatology 2000, 31(3):777-782. 142 10. Smith BD, Morgan RL, Beckett GA, Falck-Ytter Y, Holtzman D, Teo CG, Jewett A, Baack B, Rein DB, Patel N et al: Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012, 61(RR-4):1- 32. 11. Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB: Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis 2011, 43(1):66-72. 12. Red Book Online [database online]. Greenwood Village, CO: Truven Health Analytics, Inc. Available at: http://www.micromedexsolutions.com.libproxy1.usc.edu/. Accessed July 21, 2015. 13. Canary LA, Klevens RM, Holmberg SD: Limited Access to New Hepatitis C Virus Treatment Under State Medicaid Programs. Ann Intern Med 2015. 14. McCombs J, Matsuda T, Tonnu-Mihara I, Saab S, Hines P, L'Italien G, Juday T, Yuan Y: The risk of long-term morbidity and mortality in patients with chronic hepatitis C: results from an analysis of data from a Department of Veterans Affairs Clinical Registry. JAMA Intern Med 2014, 174(2):204-212. 15. Backus LI, Gavrilov S, Loomis TP, Halloran JP, Phillips BR, Belperio PS, Mole LA: Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009, 16(6):775-783. 16. D'Agostino RB, Sr., Pencina MJ, Massaro JM, Coady S: Cardiovascular Disease Risk Assessment: Insights from Framingham. Glob Heart 2013, 8(1):11-23. 17. Lasko TA, Bhagwat JG, Zou KH, Ohno-Machado L: The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform 2005, 38(5):404-415. 143 18. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982, 143(1):29-36. 19. Sterling RK, Lissen E, Clumeck N, Sola R, Correa MC, Montaner J, M SS, Torriani FJ, Dieterich DT, Thomas DL et al: Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 2006, 43(6):1317-1325. 20. Vallet-Pichard A, Mallet V, Nalpas B, Verkarre V, Nalpas A, Dhalluin-Venier V, Fontaine H, Pol S: FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. comparison with liver biopsy and fibrotest. Hepatology 2007, 46(1):32-36. 21. Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB: Prediction of coronary heart disease using risk factor categories. Circulation 1998, 97(18):1837- 1847. 22. Folsom AR: Classical and novel biomarkers for cardiovascular risk prediction in the United States. J Epidemiol 2013, 23(3):158-162. 23. Ghany MG, Kim HY, Stoddard A, Wright EC, Seeff LB, Lok AS: Predicting clinical outcomes using baseline and follow-up laboratory data from the hepatitis C long-term treatment against cirrhosis trial. Hepatology 2011, 54(5):1527-1537. 24. Ghany MG, Lok AS, Everhart JE, Everson GT, Lee WM, Curto TM, Wright EC, Stoddard AM, Sterling RK, Di Bisceglie AM et al: Predicting clinical and histologic outcomes based on standard laboratory tests in advanced chronic hepatitis C. Gastroenterology 2010, 138(1):136-146. 25. Lok AS, Ghany MG, Goodman ZD, Wright EC, Everson GT, Sterling RK, Everhart JE, Lindsay KL, Bonkovsky HL, Di Bisceglie AM et al: Predicting cirrhosis in patients with 144 hepatitis C based on standard laboratory tests: results of the HALT-C cohort. Hepatology 2005, 42(2):282-292. 26. Lok AS, Seeff LB, Morgan TR, di Bisceglie AM, Sterling RK, Curto TM, Everson GT, Lindsay KL, Lee WM, Bonkovsky HL et al: Incidence of hepatocellular carcinoma and associated risk factors in hepatitis C-related advanced liver disease. Gastroenterology 2009, 136(1):138-148. 27. American Association for the Study of Liver Diseases and the Infectious Diseases Society of America. When and in whom to initiate HCV therapy? http://www.hcvguidelines.org/full-report/when-and-whom-initiate-hcv-therapy. Accessed July 15, 2015. 28. van der Meer AJ, Hansen BE, Fattovich G, Feld JJ, Wedemeyer H, Dufour JF, Lammert F, Duarte-Rojo A, Manns MP, Ieluzzi D et al: Reliable prediction of clinical outcome in patients with chronic HCV infection and compensated advanced hepatic fibrosis: a validated model using objective and readily available clinical parameters. Gut 2014. 29. Lee MH, Lu SN, Yuan Y, Yang HI, Jen CL, You SL, Wang LY, L'Italien G, Chen CJ: Development and validation of a clinical scoring system for predicting risk of HCC in asymptomatic individuals seropositive for anti-HCV antibodies. PLoS One 2014, 9(5):e94760. 145 CHAPTER 9 : CONCLUSIONS The treatment landscape for Hepatitis C has drastically changed in the past few years, going from options that included poorly tolerated interferons and low SVR ‘cure’ rates to treatments that include all-oral options (meaning no interferons) and SVR rates in the ninety percent range. The problem of how to encourage treatment uptake in patients has now switched to how to ration treatment and target the high risk HCV patients first. With around 2.7 million people in the US chronically infected with HCV (Denniston et al, 2014), providers and payers are looking for ways to determine who to give priority to and treat first. Criteria that some payers are using now include level of fibrosis, substance use, co-infection with HIV, and prescriber limitations (Canary, Klevens, Holberg, 2015). These methods are targeting the sickest of patients, but that means that they have to progress very far in the course of the disease and incur the health problems and costs that go along with this. As only a small percentage of patients with Hepatitis C actually progress to liver fibrosis and worse (Chen & Morgan, 2006; Seeff, 2009), a more efficient way of prioritizing treatment would be to try and target this subset. Risk-prediction models have played an important role in medicine, being used to manage chronic diseases and aid in treatment decisions (Folsom 2013; Wilson et al, 1998). All of the papers in this dissertation have investigated different risk factors for HCV morbidity and mortality models spanning demographics, viral host factors, and laboratory tests. These studies used data from a large cohort of real-world HCV patients at various stages of disease progression to document risk factors for liver-related events and death while controlling for other risk factors, including genotype. Paper 1 found that viral load suppression had a significant impact on risk of morbidity and death. It also found that factors of genotype and race 146 were significant, in that genotype 2 patients experienced lower risk compared to genotype 1. Paper 2 found that the factors of ALT, AST/ALT ratio, genotype, viral load, and presence of cirrhosis were still predictive for HCC in a US veteran population. Paper 3 was able to narrow down 22 common laboratory tests down to the five ones (albumin, AST/ALT ratio, GGT, platelet, alpha fetoprotein) with the greatest significance on risk of morbidity and death. It also found that 'early’ treatment reduces risk, but delaying drug therapy until after an abnormal lab test significantly reduced the treatment effectiveness. This leads to the conclusion that untreated HCV patients should be closely monitored over time and treatment decisions revisited before patients exhibit significant changes in mentioned labs. Paper 4 looked at the impact of the non- invasive fibrosis test, FIB-4, using a a combination of risk factors of age, platelets, AST, and ALT labs. It found that not only was there a significant effect on morbidity and mortality at the specified level (>3.25) for positive prediction of fibrosis; but an effect at the lower threshold (>1.45) as well, though to a lesser extent. The sensitivity analysis looking at comparisons to another population non-invasive fibrosis tests, the APRI, showed that the estimates were much more stable in the FIB-4 model. Paper 5 found that there was also a reduction in the benefits of viral suppression if treated after a patient had progressed past the 3.25 threshold. This opens the possible use of monitoring a patient’s FIB-4 score as a way to deliver patient centered care. Paper 6 was an internal validation study that concluded that while demographics and viral characteristics from Paper 1 were inadequate in predicting morbidity and mortality, the addition of the FIB-4 turned it into an acceptable model for the morbidity outcomes. The cross validation also found that the estimates for most models were stable except for the 5-year composite morbidity model. 147 All of these studies share common limitations, one of them being as with all retrospective database studies, we are restricted to the data that is available. And as these studies were very dependent on ICD-9 codes and lab tests, it is very likely that there are some clinical data missing for any care received outside the VA. This is also a population that is not representative of the US general population, so questions of external validity must be considered. However, these studies have given a base of certain factors that are associated with increased risk of morbidity and mortality in HCV patients. Future work may include doing validation studies on a general population to examine the external validity, as well as customizing the models to each outcome directly. 148 BIBLIOGRAPHY AASLD/IDSA HCV Guidance Panel. Hepatitis C guidance: AASLD-IDSA recommendations for testing, managing, and treating adults infected with hepatitis C virus. Hepatology. 2015 Sep;62(3):932-54. Al-Mohri, H., Cooper, C., Murphy, T., & Klein, M. B. (2005). Validation of a simple model for predicting liver fibrosis in HIV/hepatitis C virus-coinfected patients. HIV Med, 6(6), 375- 378. doi: 10.1111/j.1468-1293.2005.00330.x Alemayehu D, Zou KH: Applications of ROC analysis in medical research: recent developments and future directions. Acad Radiol 2012, 19(12):1457-1464. Allison, PD. (2010). Survival analysis using SAS: A practical guide (2nd Edition). Cary, NC, USA: SAS Institute Inc. American Association for the Study of Liver Diseases and the Infectious Diseases Society of America. “When and in whom to initiate HCV therapy.” Available from: http://hcvguidelines.org/full-report/when-and-whom-initiate-hcv-therapy. Accessed August 14, 2014. Armstrong GL, Alter MJ, McQuillan GM, Margolis HS: The past incidence of hepatitis C virus infection: implications for the future burden of chronic liver disease in the United States. Hepatology 2000, 31(3):777-782. Armstrong GL, Wasley A, Simard EP, et al. The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med 2006;144(10):705-714. 149 Backus LI, Boothroyd DB, Phillips BR, Belperio P, Halloran J, Mole LA. A sustained virologic response reduces risk of all-cause mortality in patients with hepatitis C. Clin Gastroenterol Hepatol. 2011;9(6):509-516. Backus LI, Gavrilov S, Loomis TP, et al. Clinical Case Registries: simultaneous local and national disease registries for population quality management. J Am Med Inform Assoc 2009;16(6):775-783. Bacon BR, Gordon SC, Lawitz E, et al. Boceprevir for previously treated chronic HCV genotype 1 infection. NEJM 2011;364(13):1207-1217. Bonnard P, Elsharkawy A, Zalata K, Delarocque-Astagneau E, Biard L, Le Fouler L, et al. Comparison of liver biopsy and noninvasive techniques for liver fibrosis assessment in patients infected with HCV-genotype 4 in Egypt. J Viral Hepat published online : 30 JUL 2014, DOI: 10.1111/jvh.12285. Borroni G, Andreoletti M, Casiraghi MA, Ceriani R, Guerzoni P, Omazzi B, et al. Effectiveness of pegylated interferon/ribavirin combination in 'real world' patients with chronic hepatitis C virus infection. Aliment Pharmacol Ther. 2008;27(9):790-7. Borsoi Viana, M. S., Takei, K., Collarile Yamaguti, D. C., Guz, B., & Strauss, E. (2009). Use of AST platelet ratio index (APRI Score) as an alternative to liver biopsy for treatment indication in chronic hepatitis C. Ann Hepatol, 8(1), 26-31. Butt AA, McGinnis K, Skanderson M, Justice AC: A comparison of treatment eligibility for hepatitis C virus in HCV-monoinfected versus HCV/HIV-coinfected persons in 150 electronically retrieved cohort of HCV-infected veterans. AIDS research and human retroviruses 2011, 27(9):973-979. Butt AA, McGinnis K, Skanderson M, Justice AC: A comparison of treatment eligibility for hepatitis C virus in HCV-monoinfected versus HCV/HIV-coinfected persons in electronically retrieved cohort of HCV-infected veterans. AIDS Res Hum Retroviruses 2011, 27(9):973-979. Butt AA, Wang X, and Moore CG. Effect of hepatitis C virus and its treatment on survival. Hepatology 2009;50(8):387-392. Butt AA, Yan P, Lo Re V, 3rd, et al. Liver Fibrosis Progression in Hepatitis C Virus Infection After Seroconversion. JAMA Intern Med. Dec 8 2014. Canary LA, Klevens RM, Holmberg SD: Limited Access to New Hepatitis C Virus Treatment Under State Medicaid Programs. Ann Intern Med 2015. Castera L, Hezode C, Roudot-Thoraval F, et al. Effect of antiviral treatment on evolution of liver steatosis in patients with chronic hepatitis C: Indirect evidence of a role of hepatitis C virus genotype 3 in steatosis. Gut 2004;53:420-424. CDC. Hepatitis C Information for Health Professionals. Available from: http://www.cdc.gov/hepatitis/HCV/HCVfaq.htm#section1. Accessed October 31, 2014. Center for Quality Management in Public Health. The state of care for veterans with chronic Hepatitis C. In. Palo Alto, California: US Department of Veteran Affairs, Public Health Strategic Health Care Group, Center for Quality Management in Public Health; 2010. 151 Chak E, Talal AH, Sherman KE, Schiff ER, Saab S. Hepatitis C virus infection in USA: An estimate of true prevalence. Liver Int 2011;31(8):1090-1101. Chen SL, Morgan TR. The natural history of hepatitis C virus (HCV) infection. Int J Med Sci 2006;3(2):47-52. Cooper CL, Druyts E, Thorlund K, Nachega JB, El Khoury AC, O'Regan C, Mills EJ: Boceprevir and telaprevir for the treatment of chronic hepatitis C genotype 1 infection: an indirect comparison meta-analysis. Ther Clin Risk Manag 2012, 8:105-130. Cox DR. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological). 1972;34(2):187-220. Crosse K, Umeadi OG, Anania FA, et al. Racial differences in liver inflammation and fibrosis related to chronic hepatitis C. Clin Gastroenterol Hepatol. 2004;2(6):463-8. D'Agostino RB, Sr., Pencina MJ, Massaro JM, Coady S: Cardiovascular Disease Risk Assessment: Insights from Framingham. Glob Heart 2013, 8(1):11-23. Davis GL, Albright JE, Cook SF, Rosenberg DM: Projecting future complications of chronic hepatitis C in the United States. Liver Transpl 2003;9(4):331-338. Denniston MM, Jiles RB, Drobeniuc J, Klevens RM, Ward JW, McQuillan GM, Holmberg SD: Chronic hepatitis C virus infection in the United States, National Health and Nutrition Examination Survey 2003 to 2010. Ann Intern Med 2014, 160(5):293-300. Dienstag JL, McHutchison JG: American Gastroenterological Association medical position statement on the management of hepatitis C. Gastroenterology 2006, 130(1):225-230. 152 Ditah I, Ditah F, Devaki P, Ewelukwa O, Ditah C, Njei B, Luma H, Charlton M: The Changing Epidemiology of Hepatitis C Virus Infection in the United States: National Health and Nutrition Examination Survey 2001 through 2010. J Hepatol 2013. Dominitz JA, Boyko EJ, Koepsell TD, Heagerty PJ, Maynard C, Sporleder JL, Stenhouse A, Kling MA, Hrushesky W, Zeilman C et al: Elevated prevalence of hepatitis C infection in users of United States veterans medical centers. Hepatology 2005, 41(1):88-96. Doyle JS, Aspinall E, Liew D, Thompson AJ, Hellard ME. Current and emerging antiviral treatments for hepatitis C infection. Br J Clin Pharmacol. 2013 Apr;75(4):931-43. El Khoury AC, Klimack WK, Wallace C, Razavi H: Economic burden of hepatitis C-associated diseases in the United States. J Viral Hepat 2012, 19(3):153-160. El-Serag HB, Mason AC. Risk factors for the rising rates of primary liver cancer in the United Stats. Arch Intern Med 2000;160(21):3227-3230.. Fernandez-Rodriguez CM, Alonso S, Martinez SM, Forns X, Sanchez-Tapias JM, Rincón D, et al. Peginterferon plus ribavirin and sustained virological response in HCV-related cirrhosis: Outcomes and factors prediction response. Am J Gastroenterol 2010;105(10):2164-2172. Folsom AR: Classical and novel biomarkers for cardiovascular risk prediction in the United States. J Epidemiol 2013, 23(3):158-162. Fusfeld L, Aggarwal J, Dougher C, et al. Assessment of motivating factors associated with the initiation and completion of treatment fore chronic hepatitis C virus [HCV] infection. BMC Infectious Diseases 2013, 13:234-245. 153 Gaetano JN, Reau N. Hepatitis C: management of side effects in the era of direct-acting antivirals. Curr Gastroenterol Rep. 2013;15(1):305. doi: 10.1007/s11894-012-0305-1. Ghany MG, Kim HY, Stoddard A, Wright EC, Seeff LB, Lok AS. Predicting clinical outcomes using baseline and follow-up laboratory data from the hepatitis C long-term treatment against cirrhosis trial. Hepatology. 2011;54(5):1527-1537 Ghany MG, Lok AS, Everhart JE, Everson GT, Lee WM, Curto TM, et al. Predicting clinical and histologic outcomes based on standard laboratory tests in advanced chronic hepatitis C. Gastroenterology 2010;138:136-46. Gordon FD, Poterucha JJ, Germer J, et al. Relationship between hepatitis C genotype and severity of recurrent hepatitis C after liver transplantation. Transplantation 1997;63(10):1419-23. Gutierrez JA, Lawitz EJ, Poordad F. Interferon-free, direct-acting antiviral therapy for chronic hepatitis C. J Viral Hepat. 2015 Nov;22(11):861-70. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982, 143(1):29-36. Holmberg SD, Lu M, Rupp LB, Lamerato LE, Moorman AC, Vijayadeva V, et al. Noninvasive serum fibrosis markers for screening and staging chronic hepatitis C virus patients in a large US cohort. Clin Infect Dis 2013;57:240-6. Hu SX, Kyulo NL, Xia VW, Hillebrand DJ, Hu KQ. Factors associated with hepatic fibrosis in patients with chronic hepatitis C: a retrospective study of a large cohort of US patients. J Clin Gastroenterol 2009, 43(8):758-764. 154 Imazeki F, Yokosuka O, Fukai K and Saisho H. Favorable prognosis of chronic hepatitis C after interferon therapy by long-term cohort study. Hepatology 2003;38(8):493-502. Jacobson IM, Cacoub P, Dal Maso L, Harrison SA, Younossi ZM: Manifestations of chronic hepatitis C virus infection beyond the liver. Clin Gastroenterol Hepatol 2010, 8(12):1017-1029. Jacobson IM, McHutchison JG, Dusheiko G, et al. Telaprevir for previously untreated chronic hepatitis C virus infection. NEJM 2011;364(13):2405-2416 Jesudian AB, de Jong YP, Jacobson IM. Emerging therapeutic targets for hepatitis C virus infection. Clin Gastroenterol Hepatol. 2013 Jun;11(6):612-9. Kallwitz ER, Layden-Almer J, Dhamija M, et al. Ethnicity and body mass index are associated with hepatitis C presentation and progression. Clin Gastroenterol Hepatol 2010;8(1):72- 78. Kanwal F, Kramer JR, Ilyas J, Duan Z, El-Serag HB. HCV genotype 3 is associated with an increased risk of cirrhosis and hepatocellular cancer in a national sample of U.S. Veterans with HCV. Hepatology 2014;60:98-105. Kim WR, Lake JR, Smith JM, Skeans MA, Schladt DP, Edwards EB, Harper AM, Wainright JL, Snyder JJ, Israni AK et al: OPTN/SRTR 2013 Annual Data Report: liver. Am J Transplant 2015, 15 Suppl 2:1-28. Klevens RM, Hu DJ, Jiles R, Holmberg SD. Evolving epidemiology of hepatitis C virus in the United States. Clin Infect Dis 2012;55(Suppl 1):S3-9. 155 Kobayashi M, Tanaka E, Sodeyama T, Urushihara A, Matsumoto A and Kiyosawa K. The natural course of chronic hepatitis C: a comparison between patients with genotypes 1 and 2 hepatitis C viruses. Hepatology 1996; 23:695-699. Kramer JR, Kanwal F, Richardson P, Mei M, El-Serag HB. Gaps in the achievement of effectiveness of HCV treatment in national VA practice. J Hepatol 2012;56(2):320-325. Larsen C, Bousquet V, Delarocque-Astagneau E, et al. Hepatitis C virus genotype 3 and the risk of severe liver disease in a large population of drug users in France. J Med Virol. 2010;82(10):1647-54. doi: 10.1002/jmv.21850. Lasko TA, Bhagwat JG, Zou KH, Ohno-Machado L: The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform 2005, 38(5):404-415. Lauer GM, Walker BD: Hepatitis C virus infection. N Engl J Med 2001, 345(1):41-52. Lavanchy D. The global burden of hepatitis C. Liver Int 2009;29(Suppl 1):74-81. Lee MH, Yang HI, Lu SN, et al. Chronic hepatitic C virus infection increases mortality from hepatic and extrahepatic diseases: A community-based long-term prosepctive study. JID 2012(15 Aug);206:469-477. Lee MH, Yang HI, Lu SN, Jen CL, Yeh SH, Liu CJ, Chen PJ, You SL, Wang LY, Chen WJ et al: Hepatitis C virus seromarkers and subsequent risk of hepatocellular carcinoma: long- term predictors from a community-based cohort study. J Clin Oncol 2010, 28(30):4587- 4593. 156 Lee MH, Yang HI, Lu SN, Jen CL, You SL, Wang LY, L'Italien G, Chen CJ, Yuan Y: Hepatitis C virus genotype 1b increases cumulative lifetime risk of hepatocellular carcinoma. International journal of cancer 2014. Lee MH, Yang HI, Lu SN, Jen CL, You SL, Wang LY, Yuan Y, Litalian G, Chen CJ, Group R- HS: Prediction models for risk of hepatocellular carcinoma using seromarkers of hepatitis C virus infection. In: International Liver Congress; Berlin, Germany. 2011 March 30- April 3. Loftus P. Patient dilemma: Treat hepatitic C now or hold out? Wall Street Journal, March 4, 2013. http://www.wsj.com/articles/SB10001424127887323293704578330712442353712. Accessed May 30, 2015. Lok AS, Ghany MG, Goodman ZD, Wright EC, Everson GT, Sterling RK, Everhart JE, Lindsay KL, Bonkovsky HL, Di Bisceglie AM et al: Predicting cirrhosis in patients with hepatitis C based on standard laboratory tests: results of the HALT-C cohort. Hepatology 2005, 42(2):282-292. Lok AS, Seeff LB, Morgan TR, di Bisceglie AM, Sterling RK, Curto TM, Everson GT, Lindsay KL, Lee WM, Bonkovsky HL et al: Incidence of hepatocellular carcinoma and associated risk factors in hepatitis C-related advanced liver disease. Gastroenterology 2009, 136(1):138-148. Ly KN, Xing J, Klevens RM, Jiles RB, Ward JW, Holmberg SD: The increasing burden of mortality from viral hepatitis in the United States between 1999 and 2007. Ann Intern Med 2012, 156(4):271-278. 157 McAdam-Marx C, McGarry LJ, Hane CA, Biskupiak J, Deniz B, Brixner DI: All-cause and incremental per patient per year cost associated with chronic hepatitis C virus and associated liver complications in the United States: a managed care perspective. J Manag Care Pharm 2011, 17(7):531-546. McCombs J, Matsuda T, Tonnu-Mihara I, Saab S, Hines P, L’Italien G, et al. The risk of long- term morbidity and mortality in patients with chronic hepatitis C: results from an analysis of data from a Department of Veterans Affairs Clinical Registry. JAMA Intern Med 2014;174:204-12. McCombs JS, Shin J, Hines P, Yuan Y and Saab S. Impact of Drug Therapy Adherence in Patients with Hepatitis C. Am J Pharmacy Benefits 2012;4[Special Issue]:SP19-SP27. McConachie S, Wilhelm SM, Kale-Pradhan PB. New Direct-Acting Antivirals in Hepatitis C Therapy: A Review of Sofosbuvir, Ledipasvir, Daclatasvir, Simeprevir, Paritaprevir, Ombitasvir and Dasabuvir. Expert Rev Clin Pharmacol. 2015 Dec 10. [Epub ahead of print] McHutchison JG, Everson GT, Gordon SC, et al. Telaprevir with peginterferon and ribavirin for chronic HCV genotype 1 infection. NEJM 2009;360(15):1827-38. Moore C, Levitsky J. The current state and future prospects of chronic hepatitis C virus infection treatment. Curr Infect Dis Rep 2014;16:413. Ng V, Saab S: Effects of a sustained virologic response on outcomes of patients with chronic hepatitis C. Clin Gastroenterol Hepatol 2011;9(11):923-930. 158 Nkontchou G, Ziol M, Aout M, et al. HCV genotype 3 is associated with a higher hepatocellular carcinoma incidence in patient with ongoing viral C cirrhosis. J Viral Hepatitis 2011;18:e516-e522. Organ Proceurement and Transplantation Networn (OPTN) and Scientific Registrry of Transplant Recipients (SRTR). OPTN/SRTR 2011 Annual Data Report. Department of Health and Human Services, Health Resources and Services Administration, Healthcare Systems Bureau, Division of Transplantation. http://srtr.transplant.hrsa.gov/annual_reports/2011/. Accessed April 3, 2013. Poordad F, McCone J, Bacon BR, et al. Boceprevir for untreated chronic HCV genotype 1 infection. NEJM 2011;364(13):1195-1206. Poynard T, Bedossa P, Opolon P: Natural history of liver fibrosis progression in patients with chronic hepatitis C. The OBSVIRC, METAVIR, CLINIVIR, and DOSVIRC groups. Lancet 1997, 349(9055):825-832. Red Book Online [database online]. Greenwood Village, CO: Truven Health Analytics, Inc. Available at: http://www.micromedexsolutions.com.libproxy1.usc.edu/. Accessed July 21, 2015. Rein DB, Wittenborn JS, Weinbaum CM, Sabin M, Smith BD, Lesesne SB. Forecasting the morbidity and mortality associated with prevalent cases of pre-cirrhotic chronic hepatitis C in the United States. Dig Liver Dis 2011;43(1):66-72. 159 Saab S, Gordon SC, Park H, Sulkowski M, Ahmed A, Younossi Z. Cost-effectiveness analysis of sofosbuvir plus peginterferon/ribavirin in the treatment of chronic hepatitis C virus genotype 1 infection. Aliment Pharmacol Ther 2014 [Epub ahead of print]. Salomon JA, Weinstein MC, Hammitt JK, Goldie SJ: Cost-effectiveness of treatment for chronic hepatitis C infection in an evolving patient population. Jama 2003, 290(2):228-237. Sanchez-Conde M, Montes-Ramirez ML, Miralles P, Castro Alvarez JM,Bellón JM, Ramírez M, et al. Comparison of transient elastography and liver biopsy for the assessment of liver fibrosis in HIV/hepatitis C virus-coinfected patients and correlation with noninvasive serum markers. J Viral Hepat 2010;17:280-6. Seeff LB. The history of the "natural history" of hepatitis C (1968-2009). Liver Int 2009;29(Suppl 1):89-99. Shepherd J, Brodin H, Cave C, Waugh N, Price A, Gabbay J. Pegylated interferon alpha-2a and -2b in combination with ribavirin in the treatment of chronic hepatitis C: a systematic review and economic evaluation. Technol Assess. 2004;8(39):iii-iv, 1-125. Shiffman ML, Esteban R: Triple therapy for HCV genotype 1 infection: telaprevir or boceprevir? Liver Int 2012, 32 Suppl 1:54-60. Shiratori Y, Ito Y, Yokosuka O, Imazeki, F, Nakata, R, Tanaka, N, et al. Antiviral therapy for cirrhotic hepatitis C: Association with reduced hepatocellular carcinoma development and improved survival. Ann Inter Med 2005;142(2):105-114. 160 Smith BD, Morgan RL, Beckett GA, et al. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945-1965. MMWR Recomm Rep 2012;61(RR-4):1-32. Snyder, N., Gajula, L., Xiao, S. Y., Grady, J., Luxon, B., Lau, et al. APRI: an easy and validated predictor of hepatic fibrosis in chronic hepatitis C. J Clin Gastroenterol, 40(6), 535-542. Sterling RK, Lissen E, Clumeck N, Sola R, Correa MC, Montaner J, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 2006;43:1317-1325. Sterling RK, Stravitz RT, Luketic VA, et al. A comparison of the spectrum of chronic hepatitis C virus between Caucasians and African Americans. Clin Gastroenterol Hepatol 2004;2(6):469-473. Tapper EB, and Afdhal NH. Is 3 the new 1: Perspectives on virology, natural history and treatment for hepatitis C genotype 3. J Viral Hepatitis 2013;20:669-677. Vallet-Pichard A, Mallet V, Nalpas B, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and fibrotest. Hepatology 2007;46(1):32-36. van der Meer AJ, Hansen BE, Fattovich G, JJ Feld, Wedemeyer H, Dufour JF. Et al. Reliable prediction of clinical outcome in patients with chronic HCV infection and compensated advanced hepatic fibrosis: a validated model using objective and readily available clinical parameters. Gut 2015;64:2 322-331. 161 van der Meer AJ, Veldt BJ, Feld JJ, et al. Association between sustained virological response and all-cause mortality among patients with chronic hepatitis C and advanced hepatic fibrosis. JAMA 2012;308(24):2584-2593. Vergniol J, Foucher J, Terrebonne E, Bernard PH, le Bail B, Merrouche W, et al. Noninvasive tests for fibrosis and liver stiffness predict 5-year outcomes of patients with chronic hepatitis C. Gastroenterology 2011;140:1970-9, 9 e1-3. Wai, C. T., Greenson, J. K., Fontana, R. J., Kalbfleisch, J. D., Marrero, J. A., Conjeevaram, H. S., & Lok, A. S. (2003). A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology, 38(2), 518-526. doi: 10.1053/jhep.2003.50346 Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB: Prediction of coronary heart disease using risk factor categories. Circulation 1998, 97(18):1837-1847. Wong JB, McQuillan GM, McHutchison JG, Poynard T: Estimating future hepatitis C morbidity, mortality, and costs in the United States. Am J Public Health 2000, 90(10):1562-1569. Xu F, Xing, J., Moorman, A., Gordon, S., Rupp, L., Lu, M., Spradling, P., Teshale, E., Boscarino, J., Vijayadeva, V., Schmidt, M., Holmberg, S. FIB4 Score and Gender Predict the Incidence of Hepatocellular Carcinoma (HCC) among Patients with Chronic Hepatitis C Virus (HCV) Infection: Chronic Hepatitis Cohort Study (CHeCS) 65th Annual Meeting of the American Association for the Study of Liver Diseases. Boston, MA2014. Yee HS, Chang MF, Pocha C, Lim J, Ross D, Morgan TR, Monto A: Update on the management and treatment of hepatitis C virus infection: recommendations from the Department of 162 Veterans Affairs Hepatitis C Resource Center Program and the National Hepatitis C Program Office. Am J Gastroenterol 2012, 107(5):669-689; quiz 690. Yee HS, Currie SL, Darling JM, Wright TL: Management and treatment of hepatitis C viral infection: recommendations from the Department of Veterans Affairs Hepatitis C Resource Center program and the National Hepatitis C Program office. Am J Gastroenterol 2006, 101(10):2360-2378. Zeuzem S, Andreone P, Pol S, et al. Telaprevir for retreatment of HCV infection, NEJM 2011;364(25):2417-2428. Zhou S, Terrault NA, Ferrell L, et al. Severity of liver disease in liver transplantation recipients with hepatitis C virus infection: relationship to genotype and level of viremia. Hepatology 1996;24(5):1041-6. 163 APPENDIX 1: LIST OF DIAGNOSES FOR LIVER-RELATED HOSPITALIZATION OUTCOME Liver-related diagnoses Acute or unspecified hepatitis C with hepatic coma Chronic hepatitis C with hepatic coma Other specified viral hepatitis with hepatic coma Other specified viral hepatitis without mention of hepatic coma code range Unspecified viral hepatitis with hepatic coma Unspecified viral hepatitis C code range Unspecified viral hepatitis without mention of hepatic coma Toxoplasma hepatitis Malignant neoplasm of liver, primary Malignant neoplasm of intrahepatic bile ducts Malignant neoplasm of liver, not specified as primary or secondary Esophageal varices with bleeding Spontaneous bacterial peritonitis Alcoholic fatty liver Acute alcoholic hepatitis, Alcoholic cirrhosis of liver Alcoholic liver damage, unspecified Chronic hepatitis, unspecified Chronic persistent hepatitis Chronic active hepatitis Cirrhosis of liver without mention of alcohol Biliary cirrhosis (chronic nonsuppurative destructive cholangitis) Other chronic non-alcoholic liver disease Unspecified chronic liver disease without mention of alcohol Portial pyemia 164 Hepatic coma Portal hypertension Hepatorenal syndrome Other sequelae of chronic liver disease Hepatitis in viral diseases classified elsewhere Hepatitis in other infectious diseases classified elsewhere. Hepatitis, unspecified (trauma and toxic reactions) Other specified disorders of liver Unspecified disorder of liver Jaundice Hepatomegaly Ascites Hepatitis C carrier, unspecified Liver transplant status

Abstract (if available)

Abstract Hepatitis C virus (HCV) liver complications represent a substantial public health burden. There are several challenges exist to the effective clinical management of recognized HCV infections. HCV is an asymptomatic disease that largely remains undiagnosed until relatively late in the course of the disease. While there are new effective treatments available, they come with very high price tags. Hence many insurance providers have enforced heavy restrictions on its use. However, it has been shown that treatment later in the course of symptomatic chronic HCV infections often has a limited impact on long-term patient outcomes. As only a fraction of Hepatitis C patients go on to have costly complications, there is a need for a parsimonious and predictive model to manage and aid in treatment decisions. ❧ As the largest provider of healthcare services to Hepatitis C patients, the Veteran Affairs (VA) is unique in the volume of HCV patients it deals with as well as its nation wide mandated electronic medical record keeping since 1999. As such, all of the papers in this dissertation will be using a database of HCV patients from the VA Clinical Case Registry (CCR). ❧ Paper 1 (Chapter 3), begins to explore what demographic and viral factors may be associated with morbidity and mortality in HCV patients using a Cox proportional hazards model. Paper 2 (Chapter 4) examines a published risk model for hepatocellular carcinoma (HCC) developed in a Taiwanese population and assesses its performance in a US veteran population using area under the receiving operator characteristic (AUROC) curve analysis. Paper 3 (Chapter 5) is a continuation to Paper 1, going on to explore if there are common laboratory tests associated with risk of morbidity or death. Paper 3 also looks at if there was a difference between if treatment was started before a patient's labs became abnormal or if initiated afterwards on effect on risk. While many insurance providers have put restriction on treatment use dependent on liver biopsies, Paper 4 (Chapter 6) looks at the use of a non-invasive test for fibrosis, the FIB-4, and its effect on risk. Paper 5 (Chapter 7) combines some of objectives from Papers 3 and 4, looking at the impact of viral load suppression before and after a critical FIB-4 score is reached. Paper 6 (Chapter 8) is an internal validation paper, looking at the models developed in Papers 1 and 4. A k-fold cross validation (k=15) method is used as well as AUROC curve analysis.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding primary nonadherence to medications and its associated healthcare outcomes: a retrospective analysis of electronic medical records in an integrated healthcare setting

PDF

Three essays on comparative economic and clinical outcomes of antipsychotics using retrospective claims data

PDF

Impact of medication persistence on clinical and health care cost outcomes in patients with major depressive disorder using retrospective claims data

PDF

Burden of illness in hemophilia A: taking the patient’s perspective

PDF

Ribavirin systemic and intracellular exposure and its role on viral response and anemia in chronic hepatitis C genotype 1 infection

PDF

The impact of treatment decisions and adherence on outcomes in small hereditary disease populations

PDF

Three essays on emerging issues in hemophilia care

PDF

Hepatitis C test result awareness among reported cases in Los Angeles County

PDF

Crohn’s disease: health outcomes and resource utilization in the biologic era

PDF

An assessment of impact of early local progression on subsequent risk for the treatment failure in adolescent and young adult patients with non-metastatic osteosarcoma

PDF

The value of novel antihyperlipidemic treatments in the U.S. healthcare system: Reducing the burden of cardiovascular diseases and filling the gap of low adherence in statins

PDF

Therapeutic potential of host defense peptides for the treatment of cystic fibrosis lung disease

PDF

The preclinical evaluation of moxidectin as a platform for drug development for alcohol use disorder

PDF

Age related macular degeneration in Latinos: risk factors and impact on quality of life

PDF

Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study

Asset Metadata

Creator Matsuda, Tara (author)

Core Title An evaluation of risk predictors for disease progression in Veteran Affairs patients with chronic hepatitis C

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Publication Date 08/02/2016

Defense Date 12/15/2015

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

Tag Cirrhosis,Cox model,FIB-4,HCC,HCV,hepatitis c virus,hepatocellular carcinoma,OAI-PMH Harvest,retrospective database,risk prediction,ROC curve,survival analysis,VA,validation,Veteran Affairs

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McCombs, Jeffrey (committee chair), Fox, Steven (committee member), Joyce, Geoffrey (committee member)

Creator Email tsmatsud@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-291347

Unique identifier UC11280529

Identifier etd-MatsudaTar-4713.pdf (filename),usctheses-c40-291347 (legacy record id)

Legacy Identifier etd-MatsudaTar-4713.pdf

Dmrecord 291347

Document Type Dissertation

Format application/pdf (imt)

Rights Matsuda, Tara

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