Association of Host Pharmacodynamic Effects with Virologic Response to Peginterferon Alfa-2a/Ribavirin in Chronic Hepatitis C

Abstract

Patients receiving therapy for chronic hepatitis C virus (HCV) infection frequently experience cytopenias and weight loss. We retrospectively assessed the pharmacodynamic effects of peginterferon alfa-2a and ribavirin by evaluating the relationship between changes in hematologic parameters, body weight, and virologic response. Patients with HCV genotypes 1, 4, 5, or 6 receiving 24 or 48 weeks of peginterferon alfa-2a and ribavirin therapy were pooled from four phase III/IV trials. Maximum decreases in hemoglobin, neutrophils, platelets, and weight during therapy were assessed by virologic response category (sustained virologic response [SVR], relapse, breakthrough, and nonresponder) and by race/ethnicity. Of 1778 patients analyzed, more than half were male, non-Hispanic Caucasians, infected with HCV genotype 1, with baseline HCV RNA >800,000, and alanine aminotransferase ≤3 × ULN. Virologic responders (SVR, relapse, and breakthrough) experienced greater maximum decreases from baseline in neutrophils, platelets, hemoglobin, and weight compared with nonresponders; however, no clear trend was observed between SVR, relapse, and breakthrough. After adjusting for drug exposure and treatment duration, only decreases in neutrophils remained associated with virologic response. Significantly greater declines in neutrophils (P<0.0001) and platelets (P<0.005) were observed at weeks 4, 12, and 24 of therapy in virologic responders compared with nonresponders. This difference between responders and nonresponders was also observed among race/ethnic groups although statistical significance was not consistent across all groups.

Conclusion

This post hoc analysis of HCV patients treated with peginterferon alfa-2a and ribavirin shows that maximum decreases from baseline in hematologic parameters and weight loss were associated with virologic response. However, after adjusting for drug exposure and accounting for duration of therapy, only neutropenia was independently associated with virologic response.

Approximately 40%-50% of patients with hepatitis C virus (HCV) genotype 1 treated with peginterferon and ribavirin in clinical trials achieve a sustained virologic response (SVR).(1-4) Both pre- and on-treatment factors can significantly impact response rates (e.g., viral load, age, presence of fibrosis, steatosis, race/ethnicity, presence of insulin resistance, and time to first HCV RNA undetectability).(1-9) During therapy, peginterferon and ribavirin are known to elicit a pharmacodynamic response in both the virus and the host. The viral response can be measured by the number of patients achieving undetectable HCV RNA levels while the host response commonly manifests as systemic effects such as influenza-like symptoms, weight loss, depression, and myelosuppression (e.g., anemia, neutropenia, and thrombocytopenia).(10-13) Both the rapidity of viral clearance (e.g., rapid virologic response and complete early virologic response) and the magnitude of cytopenias and weight loss have been shown to correlate with viral response.(4,14-16)

The association of cytopenias and weight loss with viral response raises the potential dilemma of trying to maintain patients on therapy despite the occurrence of adverse events. Anemia is the most significant of the cytopenias as substantial reductions in hemoglobin can profoundly impact a patient’s functional status and quality of life.(17) In many cases, the anemia warrants a reduction in the dose of ribavirin.(17,18) However, response rates may be significantly lower among patients who have required ribavirin or peginterferon dose reductions,(19-21) suggesting that drug exposure is an important predictor of response.

Finding the optimal balance between managing therapy-related adverse effects and optimizing the chance of SVR is more complex for African Americans and Latinos because both groups experience significantly lower SVR rates than Caucasians.(6-8,22) African Americans may also have lower baseline leukocyte, neutrophil, and hemoglobin counts compared with Caucasians,(6) potentially decreasing the therapeutic window before dose modification is required. Latinos are more likely to experience significant anemia, neutropenia, and thrombocytopenia during therapy.(8) Whether any correlation between viral response and host pharmacodynamic effects holds true for African Americans and Latinos has yet to be clearly demonstrated.

The arrival of HCV protease inhibitors in the next 2 years is anticipated to bring significant improvements in SVR, but will likely compound the adverse events and costs that are associated with HCV therapy since they will be added to a backbone of peginterferon and ribavirin treatment. Identifying the host effects that correlate with viral response may be important in order to optimize the risk to benefit ratio of triple combination therapy. We therefore conducted a large, pooled, post hoc analysis of patients with HCV genotypes 1, 4, 5, or 6 from four trials of peginterferon alfa-2a and ribavirin therapy to better understand the association between virologic response and pharmacodynamic effects as reflected by changes in hematologic parameters and body weight.

Methods

Patients

Patients with HCV genotypes 1, 4, 5, or 6 receiving 24 or 48 weeks of combination therapy with peginterferon alfa-2a (Pegasys; Roche, Nutley, NJ; 180 μg/week) and ribavirin (Copegus; Roche, Nutley, NJ; 1000 or 1200 mg/day) were pooled from two registration trials(1,2) and two phase IV trials.(7,8) The registration trials were randomized, multicenter, phase III studies in interferon-naïve patients with chronic hepatitis C; the first trial compared the efficacy of peginterferon alfa-2a and ribavirin therapy with interferon alfa-2b and ribavirin therapy for 48 weeks,(1) and the second trial of peginterferon alfa-2a and ribavirin therapy compared different treatment duration and ribavirin dose combinations.(2) The phase IV studies were noncomparative, open-label studies of peginterferon alfa-2a and ribavirin for 48 weeks in treatment-naïve patients with HCV genotype 1; the majority (> 73%) of patients in the first study were non-Hispanic black patients,(7) and the second study was conducted in Latino and non-Latino white patients (ClinicalTrials.gov Identifier: NCT00087607).(8) All studies included stopping rules for nonresponse except for the trial in African American patients.(7) Patients who received peginterferon monotherapy or interferon alfa-2b and ribavirin combination therapy (Rebetron), and HCV/HIV coinfected patients were excluded from this analysis.

Study Design

The objectives of this analysis were: (1) to explore the association between pharmacodynamic parameters and virologic response category (SVR, relapse, breakthrough, and nonresponders); (2) to explore the association between pharmacodynamic parameters and race/ethnicity (African American, Latino Caucasians, non-Latino Caucasians, and other races); and (3) to evaluate the effects of clinically significant hemoglobin decline (> 3 g/dL versus ≤ 3 g/dL) on SVR. The pharmacodynamic effects of interest in this analysis were hematologic parameters (hemoglobin levels, neutrophil counts, and platelet counts) and weight loss. Maximum decrease (baseline value for the hematologic test minus the lowest value for that test while on therapy) was used to assess the change in hematologic parameters. To better adjust for the impact of baseline difference, percent of change from baseline was used to analyze race/ethnic group differences and body weight changes. For patients without the specified hematologic test or body weight measurement during treatment, the corresponding maximum decrease was set as missing.

The four mutually exclusive viral response categories were: (1) SVR, undetectable HCV RNA at 24-weeks post-end of treatment; (2) relapse, undetectable HCV RNA at end of treatment but detectable or missing test result at 24 weeks post-end of treatment; (3) breakthrough, undetectable HCV RNA during treatment and detectable or missing test result at end of treatment; and (4) nonresponder, no negative test result during treatment or at end of treatment, or no post-baseline HCV RNA test. Due to the sample sizes and similar results found in initial analyses, the SVR, relapse, and breakthrough categories were combined to form the responder group in some analyses. Undetectable HCV RNA levels were defined as HCV RNA < 28 IU/mL in one study (Roche High Pure System/COBAS TaqMan HCV Monitor Test)(8) and < 50 IU/mL in the remaining three studies (Roche Amplicor polymerase chain reaction assay).(1,2,7)

Statistical Analysis

Analyses were performed using the intent-to-treat population who received at least one dose of study medication. A linear regression analysis was performed to test the null hypothesis that the mean maximum decrease was the same across virologic response categories. The maximum decrease was the dependent variable. Cirrhosis, an independent predictor of non-SVR, was included in the model if it was significant (P < 0.05). To account for the impact of drug exposure, total peginterferon received over the whole treatment duration and total ribavirin received per kg of baseline weight were included in the model. Per protocol, ribavirin dose was based on baseline weight and was not modified due to changes in weight during treatment. With the virologic response category forced into the model regardless of significance, the backward selection method was used to eliminate the covariates (cirrhosis, and peginterferon and ribavirin exposures) that were not significant (P > 0.05). Adjusted mean maximum decreases for SVR, relapse, breakthrough, and nonresponder were calculated using the least square means from the final models. A sensitivity analysis including only treatment completers was performed to take into consideration the duration of therapy. In addition, separate models were conducted using the same procedures with changes in pharmacodynamic parameters from baseline to weeks 4, 12, and 24 as dependent variables. In these models, the total dose received up to the corresponding time point was used in the analysis. The same procedures were also used to assess the effects of race/ethnicity on hematologic parameters and weight. The association between hemoglobin decline and SVR was assessed with and without adjusting for drug exposure using logistic regression models with SVR/non-SVR as the dependent variable.

Results

Patient Demographic and Clinical Characteristics at Baseline

Table 1 presents the baseline demographic and clinical characteristics of 1778 patients infected with HCV genotypes 1, 4, 5, or 6 from four randomized clinical trials of 24 or 48 weeks of treatment with peginterferon alfa-2a and ribavirin. A majority of patients had HCV genotype 1 infection and more than 70% were assigned to receive peginterferon alfa-2a and ribavirin 1000-1200 mg for 48 weeks.

Table 1. Patient demographic and clinical characteristics at baseline.

Variables	Total (N = 1778)
Male sex, n (%)	1183 (66.5)
Age (years), mean ± SD	44.5 ± 9.7
Range	18.0–76.0
Weight (kg), mean ± SD*	80.8 ± 17.9
Body mass index (kg/m2), mean ± SD†	27.5 ± 5.2
< 25 kg/m2, n (%)	614 (34.9)
25 - < 30 kg/m2, n (%)	708 (40.3)
≥ 30 kg/m2, n (%)	435 (24.8)
Race, n (%)
Non-Latino Caucasian	1272 (71.5)
Latino Caucasian	287 (16.1)
African American	131 (7.4)
Other	88 (4.9)
ALT quotient ≤ 3 × ULN, n (%)	1328 (74.7)
HCV genotype, n (%)
1	1711 (96.2)
4, 5, or 6	67 (3.8)
HCV RNA log10 lU/mL, mean ± SD‡	6.1 ± 0.7
> 800,000 lU/mL, n (%)	1161 (65.3)
METAVIR activity score, mean ± SD§	1.6 ± 0.6
Cirrhosis or transition to cirrhosis, n (%)	314 (17.7)
Neutrophils, 109/L, mean ± SD∥	3.6 ± 1.4
Platelets, 109/L, mean ± SD∥	216.6 ± 64.2
Hemoglobin, g/dL, mean ± SD‡	15.5 ± 1.2
Treatment arm, n (%)
48-week peginterferon alfa-2a + 1000-1200 mg ribavirin	1269 (71.4)
24-week peginterferon alfa-2a + 1000-1200 mg ribavirin	136 (7.6)
48-week peginterferon alfa-2a + 800 mg ribavirin	262 (14.7)
24-week peginterferon alfa-2a + 800 mg ribavirin	111 (6.2)
Duration of treatment, weeks, mean ± SD	38.9 ± 13.2
24-week treatment arm (n = 247)	23.4 (3.5)
48-week treatment arm (n = 1531)	41.4 (12.5)

^*n = 1769

^†n = 1757

^‡n = 1777

^§n = 1495

^∥n = 1776.

Abbreviations: ALT, alanine aminotransferase; HCV, hepatitis C virus; ULN, upper limit of normal.

In this analysis sample, 42% of patients achieved SVR, 23% relapsed, 10% had breakthrough, and 24% were nonresponders. Approximately one third of the breakthrough and nonresponder patients completed at least 44 weeks of treatment. Of the four trials included in this analysis, epoetin alfa use was only allowed in the Latino study and was received by 10.4% of patients in the Latino group and 18% of patients in the non-Latino group.(8) The mean (± standard deviation) maximum decrease from baseline was 2.5 (± 1.3) × 10⁹/L for neutrophils, 93.6 (± 44.9) × 10⁹/L for platelets, 3.9 (± 1.5) g/dL for hemoglobin, and 6.3% (± 4.7%) for weight.

Change in Pharmacodynamic Parameters by Virologic Response Category

The mean maximum decreases in hematologic parameters and weight by virologic response category are shown in Fig. 1. Cirrhosis was associated with smaller declines in neutrophils and platelets and as a result was included in the final models. There appeared to be a correlation between virologic response and maximum decreases from baseline in each parameter. Patients with a virologic response (SVR, relapse, and breakthrough) experienced greater declines in neutrophils, platelets, hemoglobin, and weight than nonresponders (P < 0.01 for all parameters except for the difference in hemoglobin decline between breakthrough and nonresponders which was not significant).

Fig. 1.

(A) Mean maximum decreases from baseline in pharmacodynamic parameters by virologic response category. Fig. shows cirrhosis-adjusted least squares means for neutrophils and platelets. For haemoglobin and weight, cirrhosis is not significant and adjustment was not required. Error bars represent upper 95% confidence intervals. NS, not significant; *P < 0.01 versus nonresponders; **P < 0.0001 versus nonresponders.

Effects of Drug Exposure and Duration of Therapy on Changes in Pharmacodynamic Parameters

After adjusting for total peginterferon and ribavirin received (Fig. 2A), the differences between virologic responders and nonresponders in neutrophil and platelet declines remained statistically significant; however, the decrease in hemoglobin and weight was no longer consistently associated with virologic response. An additional sensitivity analysis which included only a subset of patients who completed at least 44 weeks of therapy (Fig. 2B) showed that after adjusting for drug exposure and taking into consideration of duration of therapy, only neutrophil decline was independently associated with virologic response.

Fig. 2.

Adjusted mean maximum decreases from baseline in pharmacodynamic parameters by virologic response status – A) All patients; B) Treatment completers (patients completing at least 44 weeks of treatment). Fig. shows least squares means adjusted for cirrhosis and drug exposure (total peginterferon and total ribavirin received per kg of weight at baseline) if significant (P < 0.05). Error bars represent upper 95% confidence intervals. NS, not significant; *P < 0.05 versus nonresponders; **P < 0.01 versus nonresponders; ***P < 0.001 versus nonresponders; ****P < 0.0001 versus nonresponders.

Change in Pharmacodynamic Parameters from Baseline to Weeks 4, 12, and 24

The mean changes in pharmacodynamic parameters after adjusting for drug exposure from baseline to weeks 4, 12, and 24 for virologic responders and nonresponders are shown in Fig. 3. Following the initiation of combination therapy, rapid decreases in neutrophils, platelets, and hemoglobin were observed, with most of the decline occurring by week 4 for both responders and nonresponders. Significantly greater declines in neutrophils (P < 0.0001) and platelets (P < 0.005) were seen at all time points in virologic responders compared with nonresponders. The mean decreases from baseline in hemoglobin were similar between virologic responders and nonresponders at all time points after adjusting for drug exposure. A gradual decline in weight was observed during treatment, with a trend towards a greater decline in virologic responders compared with nonresponders observed at weeks 12 and 24 (P < 0.05). These results are consistent with those observed in the analysis of maximum decrease.

Fig. 3.

Adjusted mean change in pharmacodynamic parameters from baseline to weeks 4, 12, and 24 for virologic responders and nonresponders. Error bars represent 95% confidence intervals. Figure shows least squares means adjusted for cirrhosis and drug exposure (total peginterferon and total ribavirin received per kg of weight up to the time of measurement) if significant (P < 0.05) at one or more time points. Responders = Patients with SVR, relapse, and breakthroughs. P values are from tests comparing responders and nonresponders. NS, not significant; *P < 0.05; **P < 0.005; ***P < 0.001; ****P < 0.0001.

Change in Pharmacodynamic Parameters by Race/Ethnicity

Mean percent decreases from baseline in pharmacodynamic parameters after adjusting for drug exposure by race/ethnicity are presented in Fig. 4. Overall, African Americans, Latinos, and other races (mostly Asian) had significantly smaller declines in neutrophils than non-Latino Caucasians (P < 0.0001). African Americans and other races also had smaller declines in platelets when compared with non-Latino Caucasians. In contrast, Latinos had greater declines in hemoglobin and African Americans had greater declines in weight than non-Latino Caucasians (P < 0.01 for both).

Fig. 4.

Adjusted mean percent decreases from baseline in pharmacodynamic parameters by race/ethnicity. Figure shows least squares means adjusted for cirrhosis and drug exposure (total peginterferon and total ribavirin received per kg of weight at baseline) if significant (P < 0.05). Error bars represent upper 95% confidence intervals. NS, not significant; *P < 0.01 versus non-Latino Caucasians; **P < 0.001 versus non-Latino Caucasians; ***P < 0.0001 versus Non-Latino Caucasians.

The mean maximum decreases in pharmacodynamic parameters after adjusting for drug exposure for virologic responders and nonresponders in each race/ethnic group are shown in Fig. 5. With the exception of African Americans, a significant difference in pharmacodynamic effects between responders and nonresponders was observed in neutrophils and platelets. Although a similar trend was observed in African Americans, the difference was not significant which was most likely due to the small sample size of this population. There was no difference between responders and nonresponders in hemoglobin and weight loss.

Fig. 5.

Adjusted mean maximum decrease in pharmacodynamic parameters in virologic responders and nonresponders by race/ethnicity. Error bars represent upper 95% confidence intervals. Figure shows least squares means adjusted for cirrhosis and drug exposure (total peginterferon and total ribavirin received per kg of weight at baseline) if significant (P < 0.05). Patients achieved undetectable HCV RNA levels on treatment (i.e., SVR, breakthrough, and relapse). P values are from tests comparing responders and nonresponders among race/ethnic groups. NS, not significant; *P < 0.05; **P < 0.01; ***P < 0.001.

Association Between Maximum Decrease in Hemoglobin and SVR Rates

Fig. 6 shows the predicted percentage of SVR among all patients with a clinically significant hemoglobin decline > 3 g/dL versus hemoglobin decline ≤ 3 g/dL before (Fig. 6A) and after (Fig. 6B) adjusting for total drug exposure. Before adjusting for drug exposure, the rate of SVR was significantly higher among patients who had > 3 g/dL decline in hemoglobin compared with patients who had ≤ 3 g/dL decline in hemoglobin (odds ratio = 1.29; P = 0.02). After adjusting for drug exposure, the difference in SVR rates between the two groups was not statistically significant (odds ratio = 0.88; P = 0.30). Similar results were seen when cutoffs of 1, 2, or 4 g/dL decrease in hemoglobin were used (data not shown).

Fig. 6.

Association between hemoglobin decline and SVR rates. Fig. shows predicted percentages of SVR from logistic regression models before and after adjusting for total exposure for peginterferon alfa-2a and total ribavirin exposure per kg. Cirrhosis was adjusted for in both models. P values are from the corresponding tests for odds ratios. Error bars represent upper 95% confidence intervals.

Discussion

Cytopenias and weight loss are commonly observed in patients with chronic hepatitis C treated with peginterferon alfa and ribavirin combination therapy. These adverse events are likely reflective of the effect of interferon on the host and specifically on the bone marrow for leukopenia and thrombocytopenia. Anemia is primarily a result of hemolysis caused by ribavirin in addition to a myelosuppressive effect of interferon. These adverse events and their negative effect on quality of life during HCV therapy have been well documented.(1-3,18,23)

Several analyses have demonstrated a relationship between virologic response and the degree of changes in weight and hematologic parameters. In a study of previous HCV nonresponders to prior therapy, patients who achieved at least a 1 log reduction in HCV RNA after 20 weeks of retreatment with peginterferon alfa-2a plus ribavirin had a greater reduction in body weight, platelets, and white blood cells than null responders (< 1 log₁₀ reduction in HCV RNA levels by 20 weeks), possibly reflecting a systemic resistance to interferon in null responders.(14) In a different analysis, treatment-naïve HCV genotype 1 patients treated with peginterferon alfa-2a and ribavirin who became HCV RNA undetectable at week 12 (i.e., complete early virologic response) experienced greater decreases in hematologic parameters compared with non-complete early virologic response patients suggesting that viral response and hematologic changes are pharmacodynamic effects of interferon and ribavirin.(15)

In this post hoc analysis of treatment-naïve patients infected with HCV genotypes 1, 4, 5, or 6 who were treated with peginterferon alfa-2a and ribavirin, four mutually exclusive subgroups of virologic response were compared. Greater declines in neutrophils, hemoglobin, platelets, and weight were demonstrated among patients with SVR, relapse, and breakthrough (i.e., responder) compared with nonresponders after adjusting for the presence of cirrhosis, a known predictor of thrombocytopenia and nonresponse.(24) However, among responders, no significant differences between patients with SVR, relapse, and breakthrough were observed. This supports the concept that achieving undetectable HCV RNA levels correlates with a similar host pharmacodynamic response to therapy irrespective of whether the patient remains HCV RNA undetectable or not. This finding also highlights the need to exercise some caution when attempting to predict SVR based on changes in hematologic parameters while on therapy due to the difficulty in differentiating sustained virologic responders from patients who breakthrough or relapse.

Drug exposure and duration of therapy can affect both virologic response and pharmacodynamic effects. Up to a point, greater cumulative exposure results in greater virologic and pharmacodynamic effects. However, treatment discontinuation for nonresponse may lead to less total exposure to therapy in studies with week-12 or -24 stopping rules. Three of the four studies (1,2,8) in our analysis included nonresponse stopping rules while the study comparing African American patients with Caucasian patients (7) continued treatment for the full duration regardless of on-treatment response. Our analysis comparing virologic and pharmacodynamic responses of therapy among treatment completers accounts for the effect of drug exposure independent of the different duration of therapy received. These results suggest that neutropenia may be the best surrogate pharmacodynamic marker for virologic response after adjusting for drug exposure and accounting for duration of therapy. In contrast, the sensitivity analysis confirms that thrombocytopenia, anemia, and weight loss were no longer associated with virologic response after adjusting for drug exposure and duration of therapy. These findings suggest that thrombocytopenia, anemia, and weight loss may be largely affected by the extent of drug exposure that is common to all patients rather than to specific differences in host effects.

Clear differences were observed between African Americans and non-African Americans for declines in neutrophils and platelets, but not for hemoglobin. This is consistent with evidence demonstrating a blunted systemic response to interferon for pharmacodynamic parameters, including virologic response, in African Americans.(14) These observed differences indicate that the blunted responses were more attributable to specific host effects. In contrast, there were no differences between the two groups in hemoglobin decline, suggesting that while host genetic factors could explain some of the observed between-group differences, genetic markers for anemia are likely different from any markers related to viral response or myelosuppression.(25,26) Similarly, the greater degree of weight loss among African Americans versus non-African Americans, and anemia among Latino and non-Latino Caucasians may be related to varying genetic profiles between the race/ethnic groups.

Despite the importance of these findings, there are several limitations in our study. It would have been of interest to know the underlying host predispositions to interferon responsiveness, such as IL28B genotype, which may have explained some of the differences observed between race/ethnic groups in this analysis. However, since DNA samples were not collected during the conduct of these trials, we did not perform any analysis of potential genetic contributions to our findings.(25,26) In addition, the original trials used in this analysis were not designed to evaluate the pharmacodynamic effect of peginterferon and ribavirin and so serum levels of peginterferon or ribavirin were not measured. Therefore, low levels of peginterferon or ribavirin in the serum may have contributed to the reduction in decline observed in some pharmacodynamic parameters.

In conclusion, this post hoc analysis in patients infected with HCV genotypes 1, 4, 5, or 6 and treated with peginterferon alfa-2a and ribavirin shows that maximum decreases from baseline in hematologic parameters and weight loss were associated with virologic response. The blunted effect observed in nonresponders, as well as African Americans, supports the hypothesis that these patients may have systemic resistance to combination therapy and that the lack of interferon response may have a genetic basis. Measuring the pharmacodynamic response taking into consideration the effect of cumulative drug exposure and treatment duration in patients receiving combination therapy may provide a better understanding of the mechanisms involved in the response and resistance to antiviral therapy. The important implication for clinical practice is that, as markers of therapeutic effectiveness, changes in pharmacodynamicparameters may help guide clinical decisions for individualized treatments involving therapy continuation, extension, or discontinuation, as well as for evaluating the impact that novel HCV therapies will have when added to a backbone of peginterferon and ribavirin therapy.

Abbreviations

HCV

hepatitis C virus

ULN

upper limit of normal

ALT

alanine aminotransferase

SVR

sustained virologic response