Predicting Discontinuation of Pegylated Interferon as a Result of Lack of Efficacy in United States Veterans With Chronic Hepatitis C on Dual Therapy

Abstract

Background: Many patients with chronic hepatitis C virus (HCV) being treated with pegylated interferon (peg-IFN) plus ribavirin (RBV) do not respond to therapy and do not clear the virus. Standard of care during the era of dual therapy was to discontinue the patient’s therapy based on insufficient decreases in viral load after 12 and/or 24 weeks on therapy. Objectives: We identified patient characteristics that were significant predictors of discontinuation as a result of lack of efficacy (LOE) in a national database of US veterans with genotypes 1 and 4. Methods: We identified US veterans who received care at Veterans Affairs medical centers in 2004-2009 and who had lab-confirmed HCV diagnoses and initiated therapy with peg-IFN plus RBV. Patients who discontinued therapy early were classified as either LOE or non-LOE discontinuers based on pharmacy refill patterns and laboratory response data. Predictors of LOE discontinuation were identified using univariate and multivariable Cox proportional hazards modeling. Results: Of 321 238 HCV patients with an ICD-9 diagnosis code, 31 215 (9.7%) initiated dual therapy with peg-IFN plus RBV, and 10 333 (3.2%) met all inclusion criteria and were included in the analysis. Overall, 13.6% of the cohort was classified as LOE. Significant predictors of LOE discontinuation included treatment for drug abuse (hazard ratio [HR] = 2.18), age >65 years (HR = 1.75), antiretroviral therapy for HIV (HR = 1.48), black race (HR = 1.47), platelet count >100/mm³ (HR = 1.46), and drug therapy for insomnia (HR = 1.40). Conclusions: We identified risk factors for discontinuation caused by LOE. Future work should focus on determining whether these characteristics are also predictive of triple-therapy LOE discontinuations.

Keywords

hepatitis C virus pegylated interferon lack of efficacy therapy discontinuation veterans predictive model

Introduction

Hepatitis C is one of the most common blood-borne viral infections in the United States, affecting approximately 3.2 million people.¹ The infection is often asymptomatic, but once established, at least 75% of infected patients develop chronic hepatitis C virus (HCV).² Until recently, the recommended first-line treatment for chronic HCV infection was pegylated interferon alfa (peg-IFN 2a or 2b) plus ribavirin (RBV) lasting 48 weeks for HCV genotypes 1 and 4 (G1/4).³ However, sustained virological response (SVR), defined as undetectable viral load 24 weeks after therapy completion and indicating eradication of the HCV infection, was achieved in only about half of HCV monoinfected patients and in even fewer HCV/HIV coinfected patients.^4-6 Many of the failures were a result of patients’ poor adherence (taking medications as directed, at prescribed frequency, and dosage strength) or persistence (taking medications for the prescribed duration)⁷ with dual therapy, which in turn were often a result of the therapy’s adverse effects.^5,8,9 However, there was also a certain percentage of patients whose virus simply did not respond to the regimen.^5,8,9

Nonresponse to peg-IFN and RBV, or lack of efficacy (LOE), has been explored from many angles, and its cause is likely multifactorial. Observational studies of historical data have identified viral factors influencing LOE, such as baseline viral load,¹⁰ viral kinetics in response to treatment,^11-13 and the virus’s genetic diversity.¹⁴ Patient, or host, factors that have been examined include general patient characteristics^14-19 and laboratory markers such as serum proteins.^20-23 Several studies have used US veteran databases to evaluate nonresponse and the likelihood of treatment discontinuation. Two studies, by Huckans et al²⁴ and Kramer et al,²⁵ have used Veterans Affairs (VA) data sets to calculate discontinuation among HCV-infected patients as a result of LOE; however, neither study identified individual factors associated with LOE. Backus et al²⁶ examined viral and host factors (ie, viral load, demographics, baseline laboratory values, and comorbidities) contributing to achievement of SVR but provide no insight into the relationship between these factors and LOE discontinuation. Beste et al¹¹ performed a robust analysis, using data from genotype 1–infected US veterans, of viral and host factors contributing to discontinuations. However, these authors excluded patients who discontinued as a result of LOE.¹¹

The studies described above report rates of LOE discontinuation, predictors of non-LOE discontinuation, and predictors of nonresponse to HCV dual therapy. No study, to date, has identified host factors as predictors of LOE discontinuation. Our study fills this gap by using a national electronic medical record (EMR) database to evaluate patient characteristics that predict discontinuation of dual therapy as a result of LOE. In the new era of triple therapy, in which clinical guidelines^27,28 recommend adding protease inhibitors to peg-IFN and RBV to improve SVR rates in genotype-1 HCV,^29-34 understanding factors related to treatment LOE discontinuation of peg-IFN may shed light on clinical decision making with regard to nonresponse stopping rules.

Materials and Methods

Design and Data Sets

This historical cohort study used data from a national cohort of United States (US) veterans obtained from the VA Informatics and Computing Infrastructure environment. Specific data sets included Corporate Data Warehouse (CDW) data for height, weight, pharmacy records, and laboratory test results and Medical SAS data sets for death dates as well as inpatient and outpatient visits, including procedures and diagnosis codes. Patient care notes for all liver-clinic patients nationwide were also obtained from CDW, and natural language processing techniques were applied to extract additional information from these narrative records. The information obtained from clinic notes for the current analysis included HCV genotype in patients with missing laboratory data. Genotype laboratory results were extracted directly into discrete fields for analysis to augment other sources of laboratory data.

Patient Selection

Included in the cohort were all US veterans (a) with an International Classification of Diseases, 9th Revision (ICD-9) HCV diagnosis code (ie, 070.41, 070.44, 070.51, 070.54, 070.70, 070.71, or V02.62); (b) with an inpatient/outpatient encounter between January 1, 2004, and December 31, 2009, as documented in Medical SAS data; (c) who initiated therapy with peg-IFN 2a or 2b plus RBV (each drug within 30 days of the other) during this study period, as documented in CDW pharmacy record data; and (d) who had viral genotype lab result indicating HCV genotypes 1 or 4. The date of the first prescription fill for peg-IFN during the study period was defined as the index date.

Excluded from the cohort were patients who did not demonstrate a pattern of receiving routine care in the VA for at least 6 months (defined as at least 1 outpatient visit more than 180 days prior, found in Medical SAS data), so as to ensure that patients were truly HCV treatment naïve for at least 6 months. We also excluded (a) patients whose pharmacy data contained missing or invalid National Drug Code numbers, quantities dispensed, or days supplied, such that we could not determine the true duration of therapy; (b) patients who died before the end of the target treatment duration of 44 weeks; and (c) patients with insufficient HCV-RNA viral load laboratory data to determine virological responses to therapy, which were used to classify discontinuations in patients who persisted beyond 10 weeks.

Institutional review board (IRB) approval for this study was obtained through the University of Utah’s IRB and the VA’s Office of Research and Development. This work was conducted in compliance with the requirements of the University of Utah’s IRB. All relevant ethical safeguards have been met in relation to patient or subject protection.

Independent Variables

Independent variables were those available in structured electronic medical records that had known or theoretical associations with discontinuation or nonresponse and were captured in the 6 months prior to index date.^14,35,36 They comprised demographics (age, body mass index, marital status, race, and sex); comorbid diagnoses (alcohol abuse, chronic fatigue, cirrhosis, chronic obstructive pulmonary disorder [COPD], cerebrovascular disease, dementia, depression/anxiety/psychosis, drug abuse, HIV, cancer/malignancies, myocardial infarction/congestive heart failure, osteoarthritis, plegia, renal disease, rheumatoid arthritis, sleep apnea, other sleep disorders, and ulcers); laboratory observations (albumin, alanine aminotransferase, bilirubin, creatinine, hemoglobin, low-density lipoprotein [LDL], and platelet counts); and drug exposures (alcohol addiction treatment, cancer, COPD, dementia, depression, diabetes, drug abuse, HIV, insomnia, opioids, rheumatoid arthritis, sleep somnolence, smoking cessation, and ulcer).

Outcome

The outcome in our analysis was LOE discontinuation. All patients who failed to complete at least 44 weeks of peg-IFN were classified as either LOE or non-LOE discontinuers, according to the discontinuation classification rule shown in Table 1, which is based on current HCV treatment guidelines.^27,28 Each patient’s classification depended on (a) persistence with therapy, as demonstrated by pharmacy refills (ie, persistence condition), and (b) the HCV-RNA viral load results subsequently ordered by clinicians at recommended milestones^27,28 to determine whether the patient’s virus was responding to therapy (ie, laboratory condition). Essentially, patients were considered to have discontinued for LOE reasons if they (a) persisted/received refills for at least 10 weeks and (b) had subsequent HCV-RNA viral load results indicating that the virus was not responding to therapy, after week 10 and before discontinuing. Patients who completed at least 44 weeks of peg-IFN (according to pharmacy refill records) were classified as completed. Patients who died before the end of 44 weeks of treatment and those with insufficient HCV-RNA viral load laboratory data to determine response to therapy were excluded (see Patient Selection subsection and Table 1).

Table 1.

Discontinuation Classification Rule.^a

Persistence Condition	Laboratory Condition	Classification
Patient died before the end of their target treatment duration (44 weeks)	N/A	Excluded
Patient discontinued before week 10 (before the lower limit of the 12-week stopping rule)	N/A	Non-LOE
Patient discontinued between week 10 (lower limit of the 12-week stopping rule) and week 22 (lower limit of the 24-week stopping rule)	Most recent postbaseline quantitative or qualitative HCV-RNA since week 10 was “undetectable” or 2-log lower than baseline	Non-LOE
	Most recent postbaseline quantitative HCV-RNA since week 10 was not 2-log lower than baseline	LOE^27,28
	Insufficient lab data for one of the above classifications	Excluded
Patient discontinued between week 22 (lower limit of the 24-week stopping rule) and week 44 (lower limit of “completion”)	Most recent postbaseline quantitative or qualitative HCV-RNA since week 22 was “undetectable”	Non-LOE
	Most recent postbaseline quantitative or qualitative HCV-RNA since week 22 is still detectable	LOE^27,28
	Insufficient lab data for one of the above classifications	Excluded
Patient completed 44 weeks of therapy	N/A	Completed

Abbreviations: N/A, not applicable; LOE, lack of efficacy; HCV-RNA, laboratory test for hepatitis C virus loads.

Patients were classified as “LOE,” “non-LOE,” or “Completed” depending on the set of persistence and laboratory conditions met. If the patient died before the end of their treatment period or met no complete set of conditions because of a lack of laboratory observations, they were excluded from further analysis.

Statistical Analysis

We built univariate and multivariable Cox proportional hazards regression models (using continuous time-to-discontinuation data) to identify the independent variables that were predictors of LOE discontinuation. We used a combination of forward stepwise and backward selection for the multivariable models, with stepwise elimination if P >0.1. Variables were added one at a time, and all previously included variables were retested at each iteration; those with P <0.1 remained in the model. Statistical analyses were conducted using SAS version 9.2 and Stata version 12.

Results

An attrition summary for cohort selection is shown in Figure 1. Out of 321 238 HCV patients with an ICD-9 code for HCV infection, 31 215 (9.7%) initiated dual therapy with peg-IFN plus RBV; 10 333 (3.2%) met all inclusion criteria and were included in the Cox regression analysis. Of these 10 333, 152 patients (1.5%) were genotype 4, and 10 181 (98.5%) were genotype 1. The mean age of the Cox regression cohort was 53.4 (SD = 6.0) years, and 96.0% were male (Table 2). As summarized in Figure 2, 13.6% of patients discontinued therapy as a result of LOE. The mean (SD) duration of therapy was 33.7 (22.5) weeks.

Figure 1.

Attrition summary.

Table 2.

Patient Characteristics of the Cohort (n = 10 333).

Variable	n	Percentage
Demographics
Age (years)
<50	2331	23.7%
50-59	6718	65.2%
60-64	1095	10.6%
65+	189	1.8%
BMI (kg/m²)
<18.5	54	0.4%
18.5-24.9	2488	23.6%
25-29.9	4412	42.7%
30+	3271	32.5%
Unknown	108	0.8%
Marital status
Married	3565	36.1%
Divorced	3945	36.5%
Single	1828	18.0%
Other/Unknown	995	9.4%
Race/Ethnicity
White	2550	25.3%
Black	1666	14.2%
Hispanic	317	3.1%
Other/Unknown	5800	57.3%
Male	9916	95.6%
Comorbid diagnoses
Alcohol abuse	2803	27.1%
Chronic fatigue	22	0.2%
Cirrhosis	1150	11.1%
COPD	71	0.7%
Cerebrovascular disease	110	1.1%
Dementia	16	0.2%
Diabetes	2613	17.1%
Psychiatric disorder^a	2497	24.2%
Drug abuse	2436	23.6%
HIV	1138	11.0%
Malignancy	435	4.2%
Cardiovascular disease^b	99	1.0%
Osteoarthritis	1697	16.4%
Plegia	65	0.6%
Renal disease	162	1.6%
Rheumatoid arthritis	83	0.8%
Sleep apnea	331	3.2%
Sleep disorders (other)	539	5.2%
Ulcer	135	1.3%
Baseline lab observations
HCV RNA (2+ copies/mL × 10⁶)	2552	24.7%
Albumin (3.5+ mg/dL)	561	5.4%
ALT (90+ IU/mL)	3812	36.9%
Bilirubin (1.5+ mg/dL)	727	7.0%
Creatinine (1.1+ mg/dL)	1085	10.5%
Hemoglobin (13+ g/dL male; 12+ g/dL female)	363	3.5%
LDL (100+ mg/dL)	3331	32.2%
Platelet (100+ /mm³)	408	3.9%
Drug exposures
Alcohol addiction treatment	57	0.6%
Cancer	6	0.1%
COPD	723	7.0%
Dementia	17	0.2%
Depression	3329	32.2%
Diabetes	1686	16.3%
Drug abuse	83	0.8%
HIV	295	2.9%
Insomnia/Sleep aids	322	3.1%
Opioids	0	0.0%
Rheumatoid arthritis	1	0.0%
Sleep somnolence	14	0.1%
Smoking cessation	1514	14.7%
Ulcer	7	0.1%

Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; HCV, hepatitis C virus; ALT, alanine aminotransferase; LDL, low-density lipoprotein.

Psychiatric disorders comprising depression, anxiety, or psychosis.

Cardiovascular disease comprising congestive heart failure or history of myocardial infarction.

Figure 2.

Summary of patient discontinuation outcomes by genotype, discontinuation type, and time to discontinuation.

The results of multivariable models for predicting LOE discontinuation are summarized in Table 3. The strongest predictors of LOE discontinuation in terms of the magnitude of hazard ratio (HR) were treatment for drug abuse (HR = 2.18), age >65 years (HR = 1.75), antiretroviral therapy for HIV (HR = 1.48), black race (HR = 1.47), platelet count >100/mm³ (HR = 1.46), and drug therapy for insomnia (HR = 1.40). Additional significant risk factors included comorbidities of cirrhosis (HR = 1.24) and drug abuse (HR = 1.19), LDL ≥100 mg/dL (HR = 1.35), and medication taken for depression (HR = 1.18), diabetes (HR = 1.22), and smoking cessation (HR = 1.18). Many laboratory observations were not significant predictors of LOE discontinuation, including low albumin, high alanine aminotransferase, high bilirubin, high serum creatinine, and low hemoglobin.

Table 3.

Results of Univariate and Multivariable Cox Proportional Hazards Models for LOE Discontinuations in Genotypes 1 and 4 (n = 10 333).

	Counts		Univariate			Multivariable
Covariate	n	Events	P	HR	95% CI	P	HR	95% CI
Demographics
Age (65+ vs <65 years)	189	35	0.002	1.68	1.20-2.35	0.001	1.75	1.25-2.45
BMI (30+ vs <30 kg/m²)	3271	455	0.760	1.02	0.91-1.13
Divorced	3945	516	0.963	1.00	0.89-1.11
Black race	1666	274	<0.001	1.56	1.36-1.77	<0.001	1.47	1.28-1.68
Male	9916	1356	0.077	1.29	0.97-1.71	0.058	1.33	0.99-1.77
Comorbid diagnoses
Alcohol	2803	423	<0.001	1.24	1.10-1.38
Chronic fatigue	22	3	0.584	1.16	0.68-1.94
Cirrhosis	1150	187	0.000	1.33	1.14-1.55	0.010	1.24	1.05-1.45
COPD	71	12	0.945	0.96	0.30-3.01
Cerebrovascular disease	110	15	0.373	1.30	0.73-2.28
Dementia	16	3	0.643	1.30	0.43-3.92
Diabetes	2613	1850	<0.001	1.33	1.17-1.51
Psychiatric disorder^a	2497	352	0.145	1.09	0.97-1.23
Drug abuse	2436	376	<0.001	1.30	1.15-1.46	0.006	1.19	1.05-1.34
HIV	1138	185	0.000	1.35	1.15-1.57	0.077	1.18	0.98-1.41
Malignancy	435	56	0.765	0.96	0.73-1.25
Cardiovascular disease^b	99	14	0.133	1.50	0.88-2.53
Osteoarthritis	1697	215	0.277	0.64	0.29-1.42
Plegia	65	6	0.208	0.91	0.78-1.05
Renal disease	162	21	0.981	0.99	0.51-1.91
Rheumatoid arthritis	83	10	0.154	1.38	0.88-2.13
Sleep apnea	331	50	0.383	1.13	0.85-1.50
Sleep disorders (other)	539	76	0.393	1.11	0.87-1.39
Ulcer	135	24	0.214	1.28	0.86-1.89
Baseline lab observations
Albumin (≥3.5 vs <3.5 mg/dL)	561	90	<0.001	1.48	1.19-1.82
ALT ≥90 vs <90 IU/mL)	3812	530	0.810	1.01	0.91-1.12
Bilirubin ≥1.5 vs <1.5 mg/dL)	727	105	0.066	1.20	0.98-1.46
Creatinine (≥1.1 vs <1.1 mg/dL)	1085	147	0.389	1.08	0.90-1.27
Hemoglobin (males ≥13 vs <13 g/dL) females ≥12 vs <12 g/dL)^c	363	45	0.402	1.13	0.84-1.52
LDL ≥100 vs <100 mg/dL)	3331	537	<0.001	1.42	1.27-1.58	<0.001	1.35	1.20-1.49
Platelet (≥100 vs <100/mm³)	408	68	<0.001	1.59	1.24-2.02	0.004	1.46	1.12-1.88
Drug exposures
Alcohol addiction treatment	57	6	0.364	0.69	0.31-1.53
Cancer	6	1	0.640	1.52	0.26-8.79
COPD	723	100	0.566	1.06	0.86-1.30
Dementia	17	1	0.417	0.46	0.07-2.96
Depression	3329	497	0.001	1.21	1.08-1.34	0.004	1.18	1.05-1.31
Diabetes	1686	264	<0.001	1.35	1.18-1.54	0.004	1.22	1.06-1.39
Drug abuse	83	20	<0.001	2.55	1.64-3.97	0.001	2.18	1.39-3.41
HIV	295	63	0.002	1.51	1.16-1.97	0.010	1.48	1.10-1.99
Insomnia/Sleep aids	322	56	<0.001	1.87	1.45-2.40	0.014	1.40	1.07-1.83
Opioid exposure	0	0
Rheumatoid arthritis	1	0	0.414	0.368	0.03-4.06
Sleep somnolence	1514	236	0.286	2.04	0.55-7.51
Smoking cessation	57	6	0.005	1.23	1.06-1.40	0.024	1.18	1.02-1.35
Ulcer	7	0	<0.001	0.000	0.00-0.00

Abbreviations: LOE, lack of efficacy; HR, hazard ratio; BMI, body mass index; COPD, chronic obstructive pulmonary disease; ALT, alanine aminotransferase; LDL, low-density lipoprotein.

Psychiatric disorders comprising depression, anxiety, or psychosis.

Cardiovascular disease comprising congestive heart failure or history of myocardial infarction.

Hemoglobin <12 g/dL if female or <13 g/dL if male defined as anemia.

Discussion

Whereas many studies have evaluated predictors of nonresponse to dual therapy, ours is the first to characterize LOE-related discontinuation using real-world data and to identify associated predictors for genotypes 1 and 4.

Our rate of LOE discontinuation was 13.6% of all patients who initiated therapy and had sufficient laboratory data to classify discontinuation reason, which is at the bottom of the range of documented LOE rates. It is worth noting that our LOE discontinuation rate (13.6% in genotypes 1 and 4) is comparable to that found in the study by Beste et al¹¹ (16.5% in genotype 1), particularly considering our more stringent definition of LOE discontinuation (completion of less than 44 out of 48 weeks of therapy and considering a gap length of longer than 60 days as a “discontinuation,” versus <80% of the recommended duration in Beste et al¹¹). Clinical trials have consistently shown higher nonresponse rates, ranging from 14% to 40%.^5,37,38 One potential reason for low LOE rates in our study (compared with clinical trials) was the high early non-LOE discontinuation rate that we observed; 30.3% of patients discontinued treatment in the first 10 weeks, before persisting long enough to determine if they would have a virological response. Another reason may be that the patients included in the clinical trials are those with much more severe disease than those who were treated in a real-world setting and were captured in our data. Even though the mean baseline HCV RNA level for patients in our study (3.5 copies/mL × 10⁶) was within the range of values reported in clinical trials (1.7-6.0 copies/mL × 10⁶), only 55% of the patients in our study had nonmissing values for this laboratory measure, thus making it difficult to truly compare the study populations.

Factors significantly associated with LOE discontinuation were demographics, such as age >65 years and black race; comorbidities, such as cirrhosis and drug abuse; lab values of platelets >100/mm³ and LDL ≥100 mg/dL; and medications used to treat depression, diabetes, drug withdrawal, HIV, insomnia, and smoking cessation. In contrast, we had expected that most factors associated with LOE discontinuation would be biological/pathological factors, such as baseline laboratory abnormalities.^39-42 Although certain demographics such as older age and black race appear to have a biological connection to viral response,^43-45 we also identified many behavioral factors as independent predictors, such as drug abuse, depression, and smoking, which are traditionally associated with disease-state independent adherence and persistence behaviors.^46-49

Several of the factors found to be significantly associated with LOE discontinuation deserve additional comment. First, advanced age significantly increased the risk of ineffective therapy. This is especially concerning because the number of older HCV patients is projected to increase at least through the next decade.^50,51 Triple therapy will help in this regard because several studies have shown that adding a protease inhibitor to peg-IFN and RBV should increase SVR across all age brackets.^29-34 Second, platelet count >100 mg/dL was found to increase the risk of LOE discontinuation. High platelet count may be an indicator for other conditions such as inflammatory bowel disease or serious infection.^52-54

The high correlation between behavioral-type factors and LOE discontinuation suggests several potential hypotheses. The first is that patients with these characteristics were less adherent in the weeks prior to their LOE discontinuation, diminishing viral response, and the consequence was LOE discontinuation. In fact, we calculated an average medication possession ratio, a commonly used method for measuring adherence that has been used in several published studies dealing with HCV,^55-57 in the first 10 weeks for LOE discontinuers of 96%, compared with 97% among completers; the difference was minimal and is thus unlikely to be an explanation for our findings. A second potential hypothesis is that clinicians’ LOE discontinuation decisions may not be based on laboratory response data alone, but may also be based on their perceptions about patient behaviors and characteristics. Clinicians may simply be more likely to recommend discontinuation in patients with an underwhelming response to therapy depending on these characteristics.

Although ours is the first to predict LOE discontinuation in G1/4, numerous multivariable models attempting to predict nonresponse (independently of discontinuation behavior) have preceded ours.^{10,12,14,16,17,36,58,59} The small scale and relative diversity of candidate predictors (some of which are rather obscure) identified in most of these models make them minimally useful in a clinical setting. Although viral genotype⁶⁰ and baseline viral load¹⁰ are constituent data in the clinical management of HCV, clinicians have little reason to routinely test for other potential predictors such as viral genetic diversity,¹⁴ gene expression,^59,61,62 or obscure laboratory or histological markers such as interferon-γ inducible protein 10⁶³ or anti-NS4a antibodies.⁶⁴ Testing for these factors is currently time-consuming, expensive, or both, largely without substantial support that they consistently predict response to treatment. An exception to this is IL28B genotype testing, which is becoming a frequently used, cost-effective predictor of response to treatment in chronic HCV patients with G1/4.⁶⁵ This test was not available during the time period of our study and was thus not included in this analysis.

Because peg-IFN will remain a component of triple therapy, along with RBV and a protease inhibitor, we expect that our prediction model will remain useful in the current era of triple therapy, despite the improved SVR rates documented in clinical trials. ^29-34 It is worth noting that interferon-free therapies—comprising various polymerase inhibitors designed to directly target the virus rather than the patient’s immune system—are being developed, with SVRs projected to be 90%.^66-69 These therapies are expected to be available in 2014 or 2015. The near certainty of SVR may be enough to overcome some of the seemingly behavioral or adherence-pattern predictors of LOE found in our study.

Limitations

This analysis is subject to some limitations. First, we only captured patient information within the VA health care system. Although we made efforts to ensure that patients were receiving routine care in the VA by looking for encounters at least 6 months prior to the index date, some patients could have received prescriptions and follow-up outside the VA, although the very low copays in the VA act as an incentive for veterans to fill their interferon prescriptions in the VA. In addition, whereas the VA setting was chosen for this study in part because of the richness of the available EMR data and the wide geographical representation, veterans with HCV may not be representative of the general US population of HCV patients.

A major limitation, perhaps not of the study methodology but of the data sources, was the necessity of excluding more than 5000 patients because of our inability to find data on treatment response. Some patients may have had laboratory results that were not in the structured data and were never incorporated into clinic notes that we mined using natural language processing. If these patients had been nonresponders, then the proportion of LOE discontinuers we observed could well have been much higher. However, a plausible mechanism for the missing data is that many patients whose laboratory response data were never entered into the medical record were, in fact, never tested for a response, suggesting that a majority of them might have been non-LOE discontinuers.

In using historical EMR data in each aspect of our study, we explicitly assumed that patients took the medications that were in the prescriptions they filled; we identified discontinuation dates based on fill dates and days supplied for prescriptions in the patients’ pharmacy records. An implication of this is that we were unable to adjust our discontinuation outcome for missed doses. Our data source also required us to use ICD-9 codes to determine the presence of comorbid conditions. This may be inaccurate because some patients’ recorded diagnoses may not be completely accurate in their health records. Although these assumptions likely led to misclassification of some patients (as being discontinuers in the case of pharmacy records and as having or not having certain comorbid conditions in the case of ICD-9 codes), these are common concerns with any observational studies using historical electronic records. We feel that with the ability to conduct this study in a cohort that is national in scope, to include medication exposures and laboratory results as additional independent variables, and to supplement genotype information found in structured lab data with that embedded in unstructured clinical narrative, the benefits of using electronic data from the VA far outweigh the costs.

Conclusions

We found that 13.6% of patients in our cohort of treated HCV patients discontinued because of nonresponse, which was much lower than the proportion of patients who discontinued for other reasons (30%). Factors significantly associated with LOE discontinuation were demographics, such as age >65 years and black race; comorbidities, such as cirrhosis and drug abuse; lab values of platelets >100/mm³ and LDL ≥100 mg/dL; and medication used to treat depression, diabetes, drug withdrawal, HIV, insomnia, and smoking cessation. Our results suggest that many host factors may contribute to LOE discontinuation rates in veterans. Future work should focus on determining whether these characteristics are also predictive of triple-therapy LOE discontinuations.

Footnotes

Authors’ Note

A preliminary version of this work was presented at the Liver Meeting of the American Association for the Study of Liver Diseases in San Francisco, California, in November 2011.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RH is an employee of the sponsor and owns stock in Genentech and Roche. EK is an employee of the sponsor and owns stock in Genentech and Roche. SDV has received research grants from Anolinx LLC, Genentech Inc, F Hoffmann-La Roche Ltd, Amgen Inc, Shire PLC, Mylan Specialty LP, and Merck & Co, Inc, and was a partner in 2010-2011 for Clinical Methods. CLH is an employee of the University of Utah Department of Internal Medicine. KK is an employee of the University of Utah Department of Pharmacotherapy. TM has received research funding to perform clinical trials from Roche, Merck, Bristol-Myers-Squibb, Genentech, Gilead, and Vertex. PP has been a speaker at Onyx/Bayer, Vertex, and Genentech and has received research support from Genentech, Roche, and Merck. JH is an employee of the sponsor and owns stock in Genentech and Roche. JLF has been sponsored by Agency for Healthcare Research and Quality (AHRQ; HS018582), Department of Defense Congressionally Directed Medical Research Programs, ASHP Foundation Federal Services Junior Investigator Research Grant Program, University of Utah Center on Aging Pilot Grant Program, Center for Medicare Medicaid Services Medicaid Transformation Grant, Veterans Affairs Rocky Mountain Acquisition Center, Anolinx, Amgen, Clinical Methods, Centers for Disease Control, Genentech, Norvartis, and Utah Department of Health.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was sponsored by a grant from Genentech. REN was supported in part by funding from the National Institutes of Health and the National Cancer Institute Grant 1 KM1CA156723 and the National Institutes of Health Office of the Director Grant \5TL1RR025762-03. He has also been funded by Anolinx, Genentech, Centers for Disease Control and Prevention (CDC), VA Health Services Research & Development (HSR&D), Agency for Healthcare Research and Quality (AHRQ), and VA Office of Rural Health (ORH).

References

Armstrong

Wasley

Simard

McQuillan

Kuhnert

Alter

. The prevalence of hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med. 2006;144:705-714.

Centers for Disease Control and Prevention. Hepatitis Surveilance Report No. 59. Atlanta, GA: Centers for Disease Control and Prevention; 2004.

Sulkowski

Benhamou

. Therapeutic issues in HIV/HCV-coinfected patients. J Viral Hepat. 2007;14:371-86.

Maylin

Martinot-Peignoux

Moucari

. Eradication of hepatitis C virus in patients successfully treated for chronic hepatitis C. Gastroenterology. 2008;135:821-829.

Manns

McHutchison

Gordon

. Peginterferon alfa-2b plus ribavirin compared with interferon alfa-2b plus ribavirin for initial treatment of chronic hepatitis C: a randomised trial. Lancet. 2001;358:958-965.

Torriani

Rodriguez-Torres

Rockstroh

. Peginterferon Alfa-2a plus ribavirin for chronic hepatitis C virus infection in HIV-infected patients. N Engl J Med. 2004;351:438-450.

Caetano

Lam

Morgan

. Toward a standard definition and measurement of persistence with drug therapy: examples from research on statin and antihypertensive utilization. Clin Ther. 2006;28:1411-1424, discussion 1410.

McHutchison

Manns

Patel

. Adherence to combination therapy enhances sustained response in genotype-1-infected patients with chronic hepatitis C. Gastroenterology. 2002;123:1061-1069.

Weiss

Brau

Stivala

Swan

Fishbein

. Review article. Adherence to medication for chronic hepatitis C: building on the model of human immunodeficiency virus antiretroviral adherence research. Aliment Pharmacol Ther. 2009;30:14-27.

10.

Berg

Sarrazin

Herrmann

. Prediction of treatment outcome in patients with chronic hepatitis C: significance of baseline parameters and viral dynamics during therapy. Hepatology. 2003;37:600-609.

11.

Beste

Ioannou

Larson

Chapko

Dominitz

. Predictors of early treatment discontinuation among patients with genotype 1 hepatitis C and implications for viral eradication. Clin Gastroenterol Hepatol. 2010;8:972-978.

12.

Reau

Satoskar

. Evaluation of early null response to pegylated interferon and ribavirin as a predictor of therapeutic nonresponse in patients undergoing treatment for chronic hepatitis C. Am J Gastroenterol. 2011;106:452-458.

13.

Oze

Hiramatsu

Song

. Reducing Peg-IFN doses causes later virologic response or no response in HCV genotype 1 patients treated with Peg-IFN alfa-2b plus ribavirin. J Gastroenterol. 2012;47:334-342.

14.

Saludes

Bracho

Valero

. Baseline prediction of combination therapy outcome in hepatitis C virus 1b infected patients by discriminant analysis using viral and host factors. PLoS One. 2010;5:e14132.

15.

Harrison

Brunt

Qazi

. Effect of significant histologic steatosis or steatohepatitis on response to antiviral therapy in patients with chronic hepatitis C. Clin Gastroenterol Hepatol. 2005;3:604-609.

16.

Moreno

Barcena

Garcia-Garzon

. Viral kinetics and early prediction of nonresponse to peg-IFN-alpha-2b plus ribavirin in HCV genotypes 1/4 according to HIV serostatus. J Viral Hepat. 2006;13:466-473.

17.

Nachnani

Gidwani

Sadeddin

Clarkston

Fiorella

Alba

. Clinicopathological predictors to predict sustained viral response rates in patients with chronic hepatitis C infection. Indian J Gastroenterol. 2007;26:279-282.

18.

Morisco

Stroffolini

Medda

. Retrospective, observational, multicentre study on an Italian population affected by chronic hepatitis C who failed to clear HCV-RNA after the combined therapy (PEG-IFN and ribavirin): NADIR study. J Viral Hepat. 2010;17:427-434.

19.

Berenguer

von Wichmann

Quereda

. Effect of accompanying antiretroviral drugs on virological response to pegylated interferon and ribavirin in patients co-infected with HIV and hepatitis C virus. J Antimicrob Chemother. 2011;66:2843-2849.

20.

Morra

Munteanu

Bedossa

. Diagnostic value of serum protein profiling by SELDI-TOF ProteinChip compared with a biochemical marker, FibroTest, for the diagnosis of advanced fibrosis in patients with chronic hepatitis C. Aliment Pharmacol Ther. 2007;26:847-858.

21.

Darling

Aerssens

Fanning

. Quantitation of pretreatment serum interferon-gamma-inducible protein-10 improves the predictive value of an IL28B gene polymorphism for hepatitis C treatment response. Hepatology. 2011;53:14-22.

22.

Kobayashi

Kawaguchi

Mizuta

. Metabolic factors are associated with serum alanine aminotransferase levels in patients with chronic hepatitis C. J Gastroenterol. 2011;46:529-35.

23.

Sievert

Dore

McCaughan

. Virological response is associated with decline in hemoglobin concentration during pegylated interferon and ribavirin therapy in hepatitis C virus genotype 1. Hepatology. 2011;53:1109-1117.

24.

Huckans

Loftis

Blackwell

Linke

Hauser

. Interferon alpha therapy for hepatitis C: treatment completion and response rates among patients with substance use disorders. Subst Abuse Treat Prev Policy. 2007;2:4.

25.

Kramer

Kanwal

Richardson

Mei

El-Serag

. Gaps in the achievement of effectiveness of HCV treatment in national VA practice. J Hepatol. 2012;56:320-325.

26.

Backus

Boothroyd

Phillips

Belperio

Halloran

Mole

. A sustained virologic response reduces risk of all-cause mortality in patients with hepatitis C. Clin Gastroenterol Hepatol. 2011;9:509-516.e1.

27.

Ghany

Nelson

Strader

Thomas

Seeff

. An update on treatment of genotype 1 chronic hepatitis C virus infection: 2011 practice guideline by the American Association for the Study of Liver Diseases. Hepatology. 2011;54:1433-1444.

28.

Yee

Chang

Pocha

. Update on the management and treatment of hepatitis C virus infection: recommendations from the Department of Veterans Affairs Hepatitis C Resource Center Program and the National Hepatitis C Program Office. Am J Gastroenterol. 2012;107:669-689; quiz 90.

29.

Kwo

Lawitz

McCone

. Efficacy of boceprevir, an NS3 protease inhibitor, in combination with peginterferon alfa-2b and ribavirin in treatment-naive patients with genotype 1 hepatitis C infection (SPRINT-1): an open-label, randomised, multicentre phase 2 trial. Lancet. 2010;376:705-716.

30.

Bacon

Gordon

Lawitz

. Boceprevir for previously treated chronic HCV genotype 1 infection. N Engl J Med. 2011;364:1207-1217.

31.

Poordad

McCone

J Jr

Bacon

. Boceprevir for untreated chronic HCV genotype 1 infection. N Engl J Med. 2011;364:1195-1206.

32.

McHutchison

Manns

Muir

. Telaprevir for previously treated chronic HCV infection. N Engl J Med. 2010;362:1292-1303.

33.

Jacobson

McHutchison

Dusheiko

. Telaprevir for previously untreated chronic hepatitis C virus infection. N Engl J Med. 2011;364:2405-2416.

34.

Zeuzem

Andreone

Pol

. Telaprevir for retreatment of HCV infection. N Engl J Med. 2011;364:2417-2428.

35.

Tai

Chung

. Treatment failure in hepatitis C: mechanisms of non-response. J Hepatol. 2009;50:412-420.

36.

Ladero

Martin

Fernandez

. Predicting response to therapy in chronic hepatitis C: an approach combining interleukin-28B gene polymorphisms and clinical data. J Gastroenterol Hepatol. 2012;27:279-285.

37.

Fried

Shiffman

Reddy

. Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection. N Engl J Med. 2002;347:975-982.

38.

Reddy

Wright

Pockros

. Efficacy and safety of pegylated (40-kd) interferon alpha-2a compared with interferon alpha-2a in noncirrhotic patients with chronic hepatitis C. Hepatology. 2001;33:433-438.

39.

Sarwar

Butt

Khan

Alam

Ahmad

Dilshad

. Serum alanine aminotransferase level and response to interferon-ribavirin combination therapy in patients with chronic hepatitis C. J Coll Physicians Surg Pak. 2006;16:460-463.

40.

Ishigami

Hayashi

Katano

Itoh

Hirooka

Goto

. Impact of early elevation of serum bilirubin during treatment with pegylated interferon and ribavirin in patients with chronic hepatitis C. Hepatol Res. 2010;40:963-970.

41.

Aghemo

Rumi

Monico

. The pattern of pegylated interferon-alpha2b and ribavirin treatment failure in cirrhotic patients depends on hepatitis C virus genotype. Antivir Ther. 2009;14:577-584.

42.

Floreani

Baldo

Rizzotto

Carderi

Baldovin

Minola

. Pegylated interferon alpha-2b plus ribavirin for naive patients with HCV-related cirrhosis. J Clin Gastroenterol. 2008;42:734-737.

43.

Carrion

Martin

. Viral hepatitis in the elderly. Am J Gastroenterol. 2012;107:691-697.

44.

Rauch

Kutalik

Descombes

. Genetic variation in IL28B is associated with chronic hepatitis C and treatment failure: a genome-wide association study. Gastroenterology. 2010;138:1338-1345.e1-e7.

45.

Fellay

Thompson

. Genetic variation in IL28B predicts hepatitis C treatment-induced viral clearance. Nature. 2009;461:399-401.

46.

Chodick

Shalev

Gerber

. Long-term persistence with statin treatment in a not-for-profit health maintenance organization: a population-based retrospective cohort study in Israel. Clin Ther. 2008;30:2167-2179.

47.

Yeaw

Benner

Walt

Sian

Smith

. Comparing adherence and persistence across 6 chronic medication classes. J Manag Care Pharm. 2009;15:728-740.

48.

Vieta

Azorin

Bauer

. Psychiatrists’ perceptions of potential reasons for non- and partial adherence to medication: Results of a survey in bipolar disorder from eight European countries. J Affect Disord. 2012;143:125-130.

49.

Richards

Cannon

Hayden

. Adherence with bisphosphonate therapy in US veterans with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2012;64:1864-1870.

50.

Davis

Alter

El-Serag

Poynard

Jennings

. Aging of hepatitis C virus (HCV)-infected persons in the United States: a multiple cohort model of HCV prevalence and disease progression. Gastroenterology. 2010;138:513-521.e1-e6.

51.

Biggins

Bambha

Terrault

. Projected future increase in aging hepatitis C virus-infected liver transplant candidates: a potential effect of hepatocellular carcinoma. Liver Transpl. 2012;18:1471-1478.

52.

Kulnigg-Dabsch

Evstatiev

Dejaco

Gasche

. Effect of iron therapy on platelet counts in patients with inflammatory bowel disease-associated anemia. PLoS One. 2012;7:e34520.

53.

Rose

Petersen

Gardner

Hamill

Trautner

. Etiology of thrombocytosis in a general medicine population: analysis of 801 cases with emphasis on infectious causes. J Clin Med Res. 2012;4:415-423.

54.

Mirsaeidi

Peyrani

Aliberti

. Thrombocytopenia and thrombocytosis at time of hospitalization predict mortality in patients with community-acquired pneumonia. Chest. 2010;137:416-420.

55.

Cassler

Castle

Dreitlein

Khalid

Kourlas

Barlow

. Adherence to hepatitis c therapies and the impact of comorbid conditions. In: International Conference on Viral Hepatitis, April 11-12, 2011. Baltimore, MD: International Association of Providers of AIDS Care; 2011.

56.

Mitra

Davis

Beam

Medjedovic

Rustgi

. Treatment patterns and adherence among patients with chronic hepatitis C virus in a US managed care population. Value Health. 2010;13:479-486.

57.

Backus

Boothroyd

Phillips

Mole

. Predictors of response of U.S. veterans to treatment for the hepatitis C virus. Hepatology. 2007;46:37-47.

58.

Aziz

Gil

Waheed

. Evaluation of prognostic factors for Peg Interferon alfa-2b plus ribavirin treatment on HCV infected patients in Pakistan. Infect Genet Evol. 2011;11:640-645.

59.

Dill

Duong

Vogt

. Interferon-induced gene expression is a stronger predictor of treatment response than IL28B genotype in patients with hepatitis C. Gastroenterology. 2011;140:1021-1031.

60.

Roulot

Bourcier

Grando

. Epidemiological characteristics and response to peginterferon plus ribavirin treatment of hepatitis C virus genotype 4 infection. J Viral Hepat. 2007;14:460-467.

61.

Persico

Capasso

Persico

. Suppressor of cytokine signaling 3 (SOCS3) expression and hepatitis C virus-related chronic hepatitis: Insulin resistance and response to antiviral therapy. Hepatology. 2007;46:1009-1015.

62.

Chen

Borozan

Feld

. Hepatic gene expression discriminates responders and nonresponders in treatment of chronic hepatitis C viral infection. Gastroenterology. 2005;128:1437-1444.

63.

Reiberger

Aberle

Kundi

. IP-10 correlates with hepatitis C viral load, hepatic inflammation and fibrosis and predicts hepatitis C virus relapse or non-response in HIV-HCV coinfection. Antivir Ther. 2008;13:969-976.

64.

Orlent

Desombere

Hansen

. Baseline anti-NS4a antibodies in combination with on-treatment quantitative HCV-RNA reliably identifies nonresponders to pegylated interferon-ribavirin combination therapy after 4 weeks of treatment. Eur J Gastroenterol Hepatol. 2010;22:1443-1448.

65.

Jia

Ding

Tian

Niu

Jiang

. Test of IL28B polymorphisms in chronic hepatitis C patients treated with PegIFN and ribavirin depends on HCV genotypes: results from a meta-analysis. PLoS One. 2012;7:e45698.

66.

Stedman

. Current prospects for interferon-free treatment of hepatitis C in 2012. J Gastroenterol Hepatol. 2013;28:38-45.

67.

Ferenci

. Treatment of chronic hepatitis C: are interferons really necessary? Liver Int. 2012;32(suppl 1):108-112.

68.

Trial watch: an all-oral and interferon-free future for HCV therapy? Nat Rev Drug Discov. 2011;10:891.

69.

Gane

Stedman

Hyland

. Nucleotide polymerase inhibitor sofosbuvir plus ribavirin for hepatitis C. N Engl J Med. 2013;368:34-44.