Effect of hepatitis C virus and its treatment on survival

Abstract

The effect of hepatitis C virus (HCV) and its treatment on survival is not well defined. We undertook this study to determine the effect of HCV and its treatment on survival in a national cohort of HCV-infected veterans and uninfected controls. We used a national sample of HCV-infected persons and HCV-uninfected controls from the Electronically Retrieved Cohort of HCV Infected Veterans (ERCHIVES) to compare survival between the two groups. We also compared the effect of treatment and treatment duration on survival in the HCV-infected group. We used matched Cox proportional hazards model to determine the predictors of mortality. Kaplan-Meier survival plots were generated to determine and compare survival among HCV-infected and HCV-uninfected persons, and among treated and untreated HCV-infected persons.We identified 34,480 matched pairs of HCV-infected subjects and controls. HCV infection was independently associated with a higher risk of mortality (hazards ratio, 1.37; 95% confidence interval, 1.31-1.47). Subjects treated for 48 weeks or longer had the lowest mortality among HCV-infected subjects (hazards ratio, 0.41; 95% confidence interval, 0.27-0.64), whereas those who received less than 48 week of treatment had intermediate mortality (hazards ratio, 0.71 and 0.60 for 0-23 weeks and 24-47 weeks of treatment, respectively) compared with untreated subjects. Conclusion: HCV infection is associated with a substantial increase in mortality. Subjects who are initiated on treatment, and particularly those who proceed to finish a full course of treatment, have significantly reduced risk of mortality. Further studies are warranted to determine the effect of virological control on survival. (HEPATOLOGY 2009.)

It is estimated that more than 170 million persons are infected with the hepatitis C virus (HCV) worldwide.1 HCV infection is a leading cause of liver cirrhosis, end-stage liver disease, hepatocellular carcinoma, and liver transplantation.2, 3 Treatment for HCV is associated with a reduced risk of liver disease progression and a lower incidence of hepatocellular carcinoma, even when the treatment does not achieve viral eradication.4–6 The impact of HCV infection on survival in the general population is controversial. Some studies have shown a significant increase in mortality in HCV-infected persons,7, 8 whereas others have shown relatively low mortality rate and liver disease progression in otherwise healthy persons.9, 10 In subsets of patients who have undergone liver transplantation, who have human immunodeficiency virus coinfection, or those on hemodialysis, HCV is associated with significantly shortened survival.11–16 Treatment for chronic HCV has been found to be cost-effective17 and is associated with sustained viral clearance in approximately 54% to 63% of patients overall.18–21 However, the effect of HCV infection itself and its treatment on long-term survival is not well known. We used a large national electronically retrieved cohort of HCV infected veterans (ERCHIVES) to compare survival between HCV-infected and HCV-uninfected subjects and to determine the effect of HCV treatment on survival within the HCV-infected subjects. Our study does not attempt to define or determine the appropriateness of treatment initiation, nor the appropriate duration of treatment for any of the subjects. When assessing the relationship between duration of therapy and survival, we used major treatment decision points (12 weeks for early virological response, 24 weeks, which is the recommended treatment duration for genotypes 2/3, and 48 weeks, which is the recommended treatment duration for genotypes 1/4) as the cutoffs for treatment effect.

DSS, Decisions Support System; ERCHIVES, Electronically Retrieved Cohort of HCV-Infected Veterans; HCV, hepatitis C virus; ICD-9, International Classification of Diseases, Ninth Revision; VA, Veterans Administration.

Patients and Methods

The creation of ERCHIVES has been described in previous publications.22–24 Briefly, we used the Department of Veterans Affairs (VA) National Patient Care Database to retrieve demographic and clinical information, the Pharmacy Benefits Management database to retrieve all treatment information, the Decisions Support System (DSS) database to retrieve pertinent laboratory data, and the Beneficiary Identification Records Locator Subsystem to retrieve mortality data. HCV-infected subjects were initially identified for ERCHIVES for years 1998 through 2006 based on the presence of at least one inpatient or two outpatient International Classification of Diseases, Ninth Revision (ICD-9) codes for HCV. HCV-uninfected controls were identified based on absence of any ICD-9 code for HCV at the time of matching and were individually matched by age (5-year blocks), race, sex, and geographic location (based on VA Integrated Service Network location), and whether they had a visit to the VA facility the same year as a visit with HCV diagnosis for the HCV-infected case. For the current study, subjects were deemed to have HCV infection if they had at least one positive HCV antibody test or detectable HCV RNA during the years 2001 through 2006. (DSS laboratory results are available from year 2001 onward.) Individually matched controls from ERCHIVES, who had not been diagnosed with HCV at baseline, were retained as controls. Subjects with human immunodeficiency virus coinfection were excluded. We use scrambled social security numbers to link individual subjects from one database to another. This is a well-validated method of linking various databases and addresses the privacy concerns of the VA healthcare regulatory agencies and the local institutional review board. The scrambled social security number is generated using an internal VA algorithm (which is unknown to the investigators, so subjects cannot be traced back), and an honest broker then performed the linkages and retrieved the data.

Treatment data was retrieved from the Pharmacy Benefits Management database, which is housed at the VA Information Resource Center in Hines, IL, and included prescriptions of interferon alfa, pegylated interferon alfa, ribavirin, and combinations of either type of interferon with ribavirin. Dates of prescription, as well as cumulative duration of prescription, were obtained. The doses of interferon generally used in the treatment of HCV were retrieved, and high-dose interferon alfa therapy (>5 million units per dose given ≥ 3 times per week) was excluded because such higher doses of interferon are not the standard doses used to treat HCV infection. The utility of these definitions has been established in previous publications.22, 24 The laboratory data was retrieved from the DSS database and included hemoglobin, alanine and aspartate aminotransferase, HCV antibody, HCV RNA, serum creatinine, glucose, total cholesterol, low-density lipoprotein, high-density lipoprotein, and triglycerides.

Potential major comorbidities affecting survival were identified based on the following definitions if they were present at baseline: anemia was defined as hemoglobin level less than 13 g/dL for men and less than 12 g/dL for women. Chronic kidney disease was defined based on estimated glomerular filtration rate as determined by the Modification of Diet in Renal Disease equation. Diabetes was defined by the presence of any one of the following criteria: (1) glucose ≥ 200 mg/dL on two separate occasions; (2) ICD-9 codes (two outpatient OR one inpatient) PLUS treatment with an oral hypoglycemic or insulin for 30 days or longer; (3) ICD-9 codes (two outpatient OR one inpatient) PLUS glucose ≥126 mg/dL on two separate occasions; or (4) glucose ≥ 200 mg/dL on one occasion PLUS treatment with an oral hypoglycemic or insulin for 30 days or longer. Dyslipidemia was defined if any of the following were present at baseline: (1) total cholesterol >200 mg/dL; (2) low-density lipoprotein >130 mg/dL; (3) triglycerides greater than 150 mg/dL; (4) high-density lipoprotein <40 mg/dL; or (5) use of a lipid-lowering agent for more than 30 days. Presence of coronary artery disease, chronic obstructive pulmonary disease, hypertension, alcohol and drug abuse, or dependence were defined by the presence of at least one inpatient or two outpatient ICD-9 codes.

Time at risk was calculated from the date of first visit with an HCV diagnosis for the HCV infected and the match date for the HCV-uninfected subjects. Subjects were followed until death or the last observation date in the cohort. Mortality data were obtained from the VA Beneficiary Identification Records Locator System and the National Death Index. Subjects who were not recorded in either Beneficiary Identification Records Locator Subsystem or National Death Index and had at least one follow-up encounter in the VA were considered alive until the last observation data and censored at that time. Subjects with no follow-up visits after the baseline visit were excluded from analysis.

Treatment for HCV was determined by the cumulative duration of standard or pegylated interferon (with or without ribavirin) prescribed for each subject and was treated as a time-dependent variable in the time to event analysis.25 Because major treatment decision points are at 12, 24, and 48 weeks after initiation of therapy (early virological response, recommended treatment duration for HCV genotypes 2/3, and for genotype 1/4, respectively), subjects were classified into these treatment duration categories based on their cumulative treatment duration. It is important to point out that although these are recommended decision points, they may not always be followed in actual clinical setting for various reasons, including issues of adherence, tolerability, and other significant adverse events.

We compared baseline characteristics between HCV-infected subjects and controls using McNemar test for categorical variables and paired t test for continuous variables. We used a matched Cox proportional hazards model to determine the predictors of mortality. Kaplan-Meier survival plots were generated to determine and compare survival among HCV-infected and uninfected persons, and among treated and untreated HCV-infected persons.

Results

Because laboratory data in DSS are available from 2001 onward, there were 77,119 HCV-infected subjects and 66,271 HCV-uninfected subjects with complete laboratory data. After excluding HIV-infected subjects, and retaining only individually matched subject pairs with complete data, the final dataset consisted of 34,480 HCV-infected subjects and their matched controls (Fig. 1).

Figure 1. Identification of subjects and controls for the current study from within ERCHIVES (Electronically Retrieved Cohort of HCV-Infected Veterans).

The mean age of the subjects was 53 years; 97.3% were male; and 57% were white, 28.8% black, 1.7% Hispanics, and 12.4% other/unknown race (Table 1). HCV-infected subjects were less likely to have diabetes (20.7% versus 24.4%), dyslipidemia (37.6% versus 71.8%), and hypertension (48.4% versus 57.0%), whereas they were more likely to have a diagnosis of alcohol (31.8% versus 13.5%) or drug abuse and dependence (24.3% versus 7.6%) (P < 0.001 for all comparisons). In Kaplan-Meier survival analysis, HCV infection was associated with a shortened survival (Fig. 2). Among subjects with HCV infection, those who received no treatment also had a shortened survival compared with those who received 48 weeks or more of treatment for HCV. Subjects who received any treatment had survival rates intermediate between untreated subjects and those who received 48 weeks or more of treatment (Fig. 3).

Table 1. Baseline Characteristics of HCV-Infected Subjects and HCV-Uninfected Controls

	HCV+ (n = 34,480)	HCV− (n = 34,480)	P Value
* Subjects were matched on 5-year blocks for age.
Mean age, years (SD)	53.1 (7.9)	53.3 (8.0)	*
Race			*
White	57.1	57.1
Black	28.8	28.8
Hispanic	1.7	1.7
Other/unknown	12.4	12.4
Gender, percent male	97.3	97.3	*
Anemia	12.4	12.0	0.2
Chronic kidney disease (stage 3–5)	7.1	9.0	<0.001
CAD	11.8	17.8	<0.001
COPD	12.0	11.3	0.002
Diabetes	20.7	24.4	<0.001
Dyslipidemia	37.6	71.8	<0.001
Hypertension	48.4	57.0	<0.001
Alcohol abuse or dependence	31.8	13.5	<0.001
Drug abuse or dependence	24.3	7.6	<0.001

Figure 2. Kaplan-Meier curve for mortality, unadjusted, HCV-positive versus HCV-negative.

Figure 3. Survival rate by treatment group for all HCV subjects.

In a Cox proportional hazards model, HCV was associated with a higher risk of mortality in univariable analysis (hazards ratio, 1.56; 95% confidence interval, 1.48-1.65), which persisted in the multivariable model after adjusting for demographic characteristics and comorbidities (hazards ratio, 1.37; 95% confidence interval, 1.31-1.47) (Tables 2 and 3). Other significant predictors of mortality in the multivariable model included diagnoses of anemia, chronic kidney disease, coronary artery disease, chronic obstructive pulmonary disease, diabetes, and alcohol abuse or dependence (Table 3). Compared with subjects who received no treatment, any duration of treatment for HCV was associated with a lower risk of mortality in the HCV-infected subjects. Treatment decreased the risk of death in a “dose-dependent” manner, with increasing duration of treatment being associated with decreasing mortality (Table 3).

Table 2. Predictors of Mortality, Matched Univariable Cox PH Model

Predictor	Overall HR (95% CI)	HCV+ HR (95% CI)	HCV− HR (95% CI)
* Matched paired analysis.
HCV	1.56 (1.48–1.65)
Age, per 5-year increase in age	*	1.28 (1.26–1.30)	1.34 (1.32–1.37)
Race	*	1.00	1.00
White		0.84 (0.78–0.90)	0.94 (0.85–1.03)
Black		1.10 (0.87–1.39)	0.64 (0.44–0.92)
Hispanic		0.86 (0.77–0.97)	0.76 (0.65–0.88)
Other/unknown
Male sex	*	2.43 (1.76–3.34)	1.69 (1.20–2.37)
Anemia	4.23 (4.05–4.54)	4.45 (4.15–4.79)	4.17 (3.81–4.56)
Chronic kidney disease (stages 3–5)	3.06 (2.86–3.27)	3.16 (2.89–3.45)	3.21 (2.90–3.55)
Coronary artery disease	2.15 (2.02–2.28)	2.26 (2.08–2.45)	2.35 (2.15–2.58)
Chronic obstructive pulmonary disease	2.18 (2.05–2.33)	1.92 (1.77–2.09)	2.62 (2.37–2.89)
Diabetes	1.89 (1.78–2.00)	1.96 (1.82–2.11)	1.90 (1.74–2.08)
Dyslipidemia	0.59 (0.56–0.63)	0.68 (0.63–0.73)	0.63 (0.58–0.69)
Hypertension	1.32 (1.39 (1.56)	1.38 (1.29–1.48)	1.38 (1.26–1.52)
Alcohol abuse or dependence	1.48 (1.39–1.56)	1.26 (1.17–1.35)	1.57 (1.40–1.74)
Drug abuse or dependence	1.05 (0.98–1.13)	0.88 (0.81–0.96)	1.08 (0.92–1.26)
Treatment for HCV
No treatment		1.00
0–23 weeks		0.54 (0.45–0.66)
24–47 weeks		0.44 (0.34–0.57)
> 48 weeks		0.32 (0.20–0.49)

Table 3. Predictors of Mortality, Matched Multivariable Cox PH Model (All Subjects)

Predictor	Overall HR (95% CI)	HCV+ HR (95% CI)	HCV− HR (95% CI)
* Matched paired analysis.
HCV	1.37 (1.31–1.47)	–	–
Age, per 5-year increase in age	*	1.15 (1.13–1.18)	1.23 (1.20–1.26)
Race	*
White		1.00	1.00
Black		0.72 (0.66–0.78)	0.77 (0.70–0.86)
Hispanic		1.19 (0.94–1.51)	0.64 (0.45–0.93)
Other/unknown		0.98 (0.87–1.10)	0.84 (0.72–0.98)
Male sex	*	1.67 (1.21–2.30)	1.11 (0.79–1.56)
Anemia	3.11 (2.93–3.31)	3.08 (2.85–3.34)	2.73 (2.47–3.01)
Chronic kidney disease (stage 3–5)	1.89 (1.76–2.04)	1.52 (1.37–1.68)	1.68 (1.50–1.88)
Coronary artery disease	1.74 (1.62–1.86)	1.47 (1.26–1.72)	1.57 (1.42–1.74)
Chronic obstructive pulmonary disease	1.82 (1.70–1.94)	1.44 (1.32–1.58)	1.86 (1.68–2.06)
Diabetes	1.47 (1.38–1.56)	1.50 (1.38–1.62)	1.43 (1.30–1.56)
Dyslipidemia	0.58 (0.54–0.61)	0.58 (0.54–0.63)	0.57 (0.52–0.63)
Hypertension	0.98 (0.92–1.04)	0.97 (0.90–1.04)	0.92 (0.83–1.01)
Alcohol abuse or dependence	1.61 (1.50–1.72)	1.64 (1.51–1.79)	1.93 (1.71–2.19)
Drug abuse or dependence	0.71 (0.65–0.77)	0.78 (0.71–0.86)	0.94 (0.78–1.12)
Treatment for HCV	–		–
No treatment		1.00
0–23 weeks		0.71 (0.58–0.86)
24–47 weeks		0.60 (0.46–0.77)
> 48 weeks		0.41 (0.27–0.64)

Discussion

Our study demonstrates that HCV infection is associated with a higher risk of early death. This effect persists after adjusting for several of the common comorbidities that are associated with a higher risk of death in an aging population. Although these results are not unexpected, our study is one of the first and largest to demonstrate this in a national population. Although the prevalence of several comorbidities was significantly lower in the HCV-infected subjects, HCV increased the risk of death by approximately 37% after adjusting for demographic characteristics and common comorbidities.

Several medical comorbidities were associated with an increased risk of mortality in our study. These included presence of anemia, chronic kidney disease, coronary artery disease, chronic obstructive pulmonary disease, and diabetes. These findings are also not unexpected, and the magnitude of risk was relatively similar in the HCV-infected and HCV-uninfected subjects. We have previously demonstrated the effect of HCV and comorbidities on survival in patients on dialysis,16, 26 and the current results confirm that in the nondialysis population.

Perhaps the most important finding of our study is the demonstration that the subjects who were initiated on treatment, and especially those who proceeded to complete a full course, had a lower risk of death compared with those who were not initiated on treatment. Although a full course of treatment for HCV is 48 weeks for genotypes 1 and 4, and 24 weeks for genotypes 2 and 3 infection, any duration of treatment was associated with a lower risk of mortality compared with untreated subjects. We did not have genotype information on all subjects to determine the appropriate duration of treatment, but from previous studies, 75% to 85% of HCV-infected veterans have genotype 1 infection,27, 28 which would necessitate a 48-week course of treatment. Subjects who received 48 weeks or more of treatment in our analysis had the highest benefit, whereas those who received less than 48 weeks had an intermediate benefit. It is possible that subjects who did not complete treatment may have had more comorbidities or more severe liver disease adversely impacting their survival. Regardless, this underscores the need to identify and treat HCV-infected persons who are otherwise eligible for therapy.

The strengths of our study include a large national sample, and presence of an HCV-uninfected control group. With the exception of being predominantly male, HCV-infected veterans are quite representative of the national HCV epidemic in the United States.29 One important limitation of our study is that the data were retrieved from administrative and clinical records and were not specifically collected for this study. We did not analyze the effect of treatment success or failure on survival. Because genotype data were not uniformly available, we did not assess the appropriateness of duration of therapy, nor did we determine reasons for early discontinuation of treatment in those subjects who did not complete a full course of treatment. Because we did not have histological or radiographic data, the effect of liver disease stage on mortality was not studied. We studied overall, all-cause mortality, and not liver disease–related mortality. Hence it is unclear what proportion of deaths were actually related to HCV versus other causes. It should also be noted that many subjects who were not initiated on treatment have comorbidities that may preclude treatment. How such selection may affect survival is unclear.

In conclusion, HCV is associated with a significantly higher risk of mortality, which persists after adjusting for demographics factors and common comorbidities. This risk is lower in those persons who are initiated on treatment, especially among those who are able to complete a 48-week course of treatment. Strategies to identify appropriate candidates for treatment, and to ensure completion of treatment, may substantially reduce mortality in HCV-infected persons.