The impact of past alcohol use on treatment response rates in patients with chronic hepatitis C

Authors


Dr F. Friedenberg, Gastroenterology Section, Temple University Hospital, 8th Floor Parkinson Pavilion, 3401 N. Broad Street, Philadelphia, PA 19140, USA.
E-mail: frank.friedenberg@temple.edu

Summary

Background : Studies have shown that past alcohol consumption reduces response rates in patients with chronic hepatitis C treated with interferon monotherapy.

Aim : To clarify the importance of alcohol consumption on response rates in patients undergoing treatment with pegylated interferon and ribavirin.

Methods : In a single centre, prospective study, median daily alcohol consumption (determined by previously validated method) and quartiles of alcohol consumption were calculated. Univariate and binary logistic regression analyses were performed using treatment response status as the dependent variable.

Results : Overall, in an intention-to-treat analysis, 34 of 115 patients (30%) responded to treatment. In univariate analysis, black patients, especially those with hepatitis C virus genotype 1, high viral load and low alanine aminotransferase were significantly less likely to respond. Predictors of response by regression analysis included alcohol <30 g/day (OR = 3.02, 95% CI: 1.02–8.93; P = 0.04), non-genotype 1 status (OR = 1.98, 95% CI: 1.03–3.80; P = 0.04) and non-black race (OR = 2.79, 95% CI: 1.33–5.85; P = 0.006).

Conclusions : Median daily alcohol use >30 g/day is associated with failure to respond to pegylated interferon and ribavirin for treatment of hepatitis C. Past alcohol use should be evaluated when considering treatment for hepatitis C.

Introduction

It is estimated that nearly 4 million people in the United States are infected with hepatitis C virus (HCV). Of these, nearly 75% will remain chronic carriers leading to increased risks of both cirrhosis and hepatocellular carcinoma (HCC).1 Viral eradication following treatment for HCV has been demonstrated to reduce the rates of cirrhosis and HCC. Several studies have shown, however, that treatment success using interferon monotherapy is lower for those who consume alcohol than non-drinkers.2 For example, a study of 53 patients found that 30% of non-habitual drinkers responded to treatment, whereas only 6% of habitual drinkers responded.3 In a similar study, 53% of non-drinkers had a sustained biochemical response to treatment vs. 0% of those who consumed >70 g/day of alcohol.4 In a later study, Ohnishi et al.5 demonstrated that those with a history of heavy alcohol intake, even while abstaining from alcohol use prior to treatment, had a lower rate of viral clearance while on interferon monotherapy than non-drinkers. Their work showed that 28% of infrequent drinkers had negative RNA levels 6 months after treatment, compared with 16% in former heavy drinkers who had an average abstinence time of 39 months, and 0% in those who continued to drink heavily during treatment. A more recent study found that while 33% of non-drinkers had a sustained virological response (SVR) to interferon monotherapy, only 9% of those who drank responded to treatment, even with a pre-treatment period of abstinence of at least 6 months.6 In the only study found that evaluated combined interferon and ribavirin, no difference in response rates (37%) was noted among those who were alcohol users and controls. The focus of this study was on adherence to therapy among a variety of psychiatric risk groups, however, and its applicability to assessing the impact of past alcohol use on response rates is not clear.7

Pegylated interferon combined with ribavirin has replaced interferon monotherapy as the treatment of choice for patients with HCV.8 To date, there are no published studies comparing the efficacy of pegylated interferon combined with ribavirin that have considered alcohol consumption as a predictor variable. Therefore, the purpose of this study was to clarify the importance of alcohol consumption on SVR rates in patients undergoing treatment for HCV with a contemporary regimen, specifically pegylated interferon and ribavirin. Our hypothesis was that patients who drank would have a lower response rate to HCV treatment than non-drinkers even with this contemporary regimen.

Materials and methods

This was a single-centre, prospective study of patients treated in the Section of Gastroenterology, Temple University Hospital. Our institution approved the study prior to patient enrolment (IRB protocol no. 4114). All patients who presented to our out-patient centre for treatment of HCV were evaluated for study inclusion (Figure 1). Patient enrolment ended 1 September 2003. To be included, patients needed to have documented hepatitis C by a polymerase chain reaction (PCR)-based test and negative studies for all other forms of liver disease. All patients were HIV-negative and treatment naive, and were able to provide informed consent. Prior to treatment all patients underwent liver biopsy. The majority of biopsies were done at our institution; those done at outside institutions prior to referral were subsequently reviewed by our pathologist. At our institution fibrosis is staged on a 0 (normal) to 6 (established cirrhosis) scale and inflammation graded on a 1–18 scale. All patients were treated with pegylated interferon α-2b (Peg-Intron, Schering-Plough, Kenilworth, NJ, USA) and weight-based ribavirin (Rebetol, Schering-Plough) in divided doses. The pharmaceutical company did not sponsor the study or were they informed of the results. Follow-up appointments and laboratory data were checked at monthly intervals. For patients with symptomatic anaemia or haemoglobin levels below 10 g/dL, treatment with erythropoietin was initiated and ribavirin doses reduced if necessary until symptoms resolved or haemoglobin levels returned to an acceptable level (usually approximating 10 g/dL). Therapy was continued for 24 (genotype 2 or 3) or 48 (genotype 1a and 1b) weeks depending on the HCV genotype. Evaluation for SVR was performed no <24 weeks after the cessation of treatment using quantitative transcription-mediated amplification (TMA) technology (Heptimax, Quest Diagnostics Nichols Institute, Chantilly, VA, USA). A level of <5 IU/mL, the lower limit of viral detection by this assay, was necessary to satisfy the criteria for SVR. All reasons for treatment failure were recorded. Examples included intolerable side-effects such as major depression, significant cytopenias not responding to growth factors, and virological non-response at 12 weeks (only for genotypes 1a and 1b), defined by a <2-log drop in viral load in comparison with the starting viral load.

Figure 1.

Flowchart of hepatitis C patients in study.

Alcohol consumption

Alcohol consumption was assessed using a previously validated questionnaire administered by one of the authors. The setting was a private clinical office and care was taken to provide sufficient time for question administration and response clarification. Years of consumption were estimated by the patient. The number of drinks consumed over the patient's lifetime was multiplied by the alcohol content for each type of alcohol9 (see Table 1), resulting in an estimate of lifetime alcohol consumption. Daily alcohol intake was calculated by determining the age at both the beginning and cessation of alcohol intake, and dividing the cumulative alcohol intake by that period of time. These methods have been validated previously.10 The duration of abstinence prior to starting treatment for HCV was also estimated by the patient in the same questionnaire. Our policy is to require a minimum of 6 months of abstinence prior to consideration for HCV treatment.11

Table 1.  Conversion table used to estimate alcohol intake for study
Alcohol typeEthanol (%)Volume (mL)Alcohol content (g)
  1. Taken in part from Mukamal et al.9

12 oz beer435512.8
40 oz beer (malt liquor)6118356
4 oz glass of wine1212011.2
Bottle of wine1275070.2
1 oz liquor4030–4514
Pint of liquor40473147.6
1/5 gallon of liquor40757234

Statistical analysis

Baseline variables were analysed using Pearson's chi-square test (with Yate's correction) or Fisher's Exact test for categorical data and independent samples t-test for continuous data. Alcohol consumption was analysed using the Mann–Whitney U-test and results expressed as medians as the data were not normally distributed. After the preliminary analysis, alcohol consumption per day was dichotomized using a cutoff of ≥30 g/day. We subsequently performed a binary logistic regression analysis using all variables, which were significant from univariate analysis with virological responder status as the dependent variable. Results of the logistic regression yielded odds ratios (OR) with their respective 95% confidence intervals. A P-value of ≤0.05 was considered statistically significant and all results were two-tailed. spss version 10.0 (SPSS Inc., Chicago, IL, USA) was used as the database and for statistical calculations.

Results

Table 2 shows the baseline characteristics of the patients who achieved (responder), and failed to achieve (non-responder), a SVR. Overall, in an intention-to-treat (ITT) analysis, 34 of 115 patients (29.6%) responded defined as an undetectable level of HCV-RNA 24 weeks after the completion of treatment. The mean age, gender distribution, weight and histological scores did not differ between the two groups. Black patients and those with HCV genotype 1 were significantly less likely to respond. Those patients with a low baseline viral load and greater elevation of alanine aminotransferase (ALT) were more likely to respond.

Table 2.  Characteristics of 115 study patients with respect to sustained virological response status
VariableResponders (n = 34)Non-responders (n = 81)P-value
  1. For continuous variables values represent the mean ± S.E.M.; for discrete variable numbers in parentheses are rounded percentages.

  2. P-value calculated using Pearson's chi-square or Fisher's Exact tests for discrete variables, independent sample t-test for continuous variable.

  3. * Fibrosis staged 0–6 scale,

  4. † Activity graded 1–18 scale.

  5. ALT, alanine aminotransferase.

Age (years)46.4 ± 1.547.5 ± 0.90.53
Male:female18:1654:270.17
Race (n)
 White20 (59)18 (22)0.001
 Black12 (35)50 (62)
 Other2 (6)10 (12)
Weight (kg)89.0 ± 4.286.0 ± 2.20.50
Genotype (n)
 120 (59)75 (93)<0.001
 210 (29)1 (1)
 32 (6)3 (4)
 Other2 (6)2 (3)
Initial viral load (IU/mL × 105)5.1 ± 0.828 ± 8.10.006
ALT (IU/mL)86.0 ± 9.767.2 ± 4.50.05
Fibrosis stage*2.5 ± 0.23.0 ± 0.20.07
Activity grade†7.4 ± 0.57.7 ± 0.30.56

Alcohol consumption and sustained virological response

Alcohol data were available for 98 patients entered into the study. Seventeen patients could not provide a history with sufficient detail or reliability to estimate intake. The median daily alcohol consumption for those patients who had no response was significantly higher than those who responded (62 g/day vs. 10 g/day; P = 0.02). Median alcohol consumption by black patients was substantially higher than white patients (54 g/day vs. 31 g/day; P = 0.003). We subsequently dichotomized alcohol intake at <30 g/day (roughly equivalent to the USDA recommended limit of alcohol consumption for the USA population) vs. ≥30 g/day.12 Sustained responders were significantly more likely to have consumed <30 g/day vs. those consuming higher levels of alcohol (OR = 3.2, 95% CI: 1.3–8.1; P = 0.013). We then performed a binary logistic regression analysis to determine the relationship between the independent variables associated with responder status by univariate analysis (namely, ALT, initial viral load, genotype, race and alcohol consumption dichotomized at 30 g/day) and responder status as the dependent variable. Predictors of response by regression analysis were alcohol <30 g/day (OR = 3.02, 95% CI: 1.02–8.93; P = 0.04), non-genotype 1 status (OR = 1.98, 95% CI: 1.03–3.80; P = 0.04) and non-black race (OR = 2.79, 95% CI: 1.33–5.85; P = 0.006). The predictive model satisfied the Hosmer-Lemeshow test for goodness-of-fit (P = 0.61) and was therefore well calibrated.

We explored the relationship between quartiles of alcohol consumed per day and rates of virological response. As can be seen from Figure 2, those patients in the lowest quartile of alcohol consumption had the highest rate of response at 48.4%. The relationship was not linear through all four quartiles of alcohol consumption. The median duration of abstinence before starting treatment in responders was greater than those who were non-responders, however this did not achieve statistical significance (5.0 vs. 3.0 years; P = 0.15). Median daily alcohol consumption was not associated with median duration of abstinence prior to starting HCV treatment (P = 0.71).

Figure 2.

Rates of hepatitis C virus (HCV) virological response compared with quartile of alcohol consumption. * Quartiles based on 25, 50 and 75% median values. Interquartile ranges are as follows: 1, 0–9.8; 2, 9.9–45.0; 3, 45.1–156.8; 4, 156.9–987.0.

Adverse events, discontinuation of treatment, per-protocol analysis

A total of 53 patients did not complete a full course of treatment for HCV. For genotype 1 patients, the majority (n = 25) were discontinued on therapy because there was a failure to achieve a virological response when measured at 12 weeks. The remainder of patients did not complete therapy because of depression (n = 10), non-compliance (n = 8), anaemia or leucopoenia (n = 6) and alcohol relapse (n = 1). For genotypes 2 and 3, only three patients prematurely discontinued therapy (depression n = 1; non-compliance n = 1; severe skin rash n = 1). For genotype 1 patients we repeated an analysis of the relationship between alcohol consumption, duration of abstinence and responder status for patients who completed all 48 weeks of treatment. For this subgroup of genotype 1 patients (n = 45) there appeared to be no relationship between median alcohol consumption or duration of abstinence with responder status (P = 0.20 and P = 0.73 respectively). These relationships should be viewed with caution as only patients who responded after 12 weeks of therapy are included.

Discussion

Several studies have shown that high levels of alcohol intake have a negative effect on treatment of HCV with interferon monotherapy.2–6 To our knowledge, this is the first study evaluating the effects of alcohol intake on treatment response using the current standard of care, pegylated interferon and ribavirin. The present study demonstrates that there is a relationship between median daily alcohol intake and response to treatment for chronic hepatitis C using these medications. There is also a trend towards response to therapy and duration of abstinence, although this was not statistically significant.

Our study group appeared to be representative of others which have been used to evaluate treatment response for HCV. For instance, in univariate analysis, we found that black race, genotype 1 and a high viral load were independently associated with a decreased response rate. Multiple other studies have yielded similar findings.8, 13, 14 There was a trend towards worsening liver histology and non-response, but this was not statistically significant.15 An exception for our group was the finding that higher ALT levels were associated with treatment response. Interestingly, higher ALT levels approached significance by univariate analysis in a very recent study of patients with genotypes 2 and 3 (P = 0.06).16 As in the present study, it was no longer significant on multivariate analysis. This phenomenon has been observed in patients undergoing treatment for chronic hepatitis B, but we could find no other evidence of, or an explanation for, improved response rates in hepatitis C patients with higher ALT levels. In our regression model we found that consumption of <30 g of alcohol (equivalent to 2–4 oz drinks of liquor) per day (OR = 3.02, P = 0.04), non-genotype 1 infection (OR = 1.98, P = 0.04) and non-black race (OR = 2.79, P = 0.006) were associated with response to pegylated interferon α-2b and ribavirin combination treatment. We found no association between the duration of alcohol abstinence prior to treatment and responder status.

There is no established mechanism for a decreased response rate among those who have previously consumed a large amount of alcohol. Proposed mechanisms include: impaired cellular immunity; higher HCV-RNA levels; interference by lipid peroxidation products; emergence of quasi-species; increased fibrosis and increased hepatic iron stores.17, 18 Recent research has confirmed that heavy alcohol intake (>50 g/day) in patients with chronic hepatitis C is associated with an increase in mean fibrosis scores.9 While the differences in fibrosis among responders and non-responders was not statistically significant in our study, the similarity in the median amount of alcohol intake among non-responders in our study (62 g/day) and the amount cited above is notable.

There are limitations to the present study. First, patients’ assessments of their past alcohol use and period of alcohol cessation are inherently suspect.17 Secondly, complete data for all patients was not available, including alcohol intake and abstinence periods for 17 of 115 patients. We believe the relatively high number of patients included mitigates these concerns. Thirdly, we did not evaluate the influences of hepatic fat content or stainable iron on response rates. While this information may have been insightful, our study protocol did not require quantitative scores for fat content and stainable iron. Of note, a recent paper did assess the impact of steatosis on response rate in patients with genotypes 2 and 3, and found it to be insignificant (P = 0.94).16 Finally, the definition of non-response included all patients who failed to achieve a sustained response, including those who stopped treatment for haematological or psychiatric concerns, intolerance of therapy, non-compliance with treatment or follow-up visits and lack of early virological response (i.e. an ITT analysis was performed). A comparison of those who actually completed treatment (per-protocol analysis) revealed that past alcohol intake was not statistically different among responders and non-responders. However, because past alcohol use may be directly or indirectly related to the non-virological reasons patients fail treatment (e.g. depression), the ITT analysis was appropriate as the primary analysis.

In conclusion, median daily alcohol use >30 g/day is associated with failure to respond to pegylated interferon and ribavirin for the treatment of chronic hepatitis C. Therefore, past alcohol use should be evaluated along with other pre-treatment variables (e.g. viral load, genotype, biopsy results) when considering treatment for hepatitis C. Further studies using populations of patients with different characteristics (in particular race and genotype) should be performed to determine if our findings warrant broad applicability.

Acknowledgement

No external funding was received for this study.

Ancillary