ITPA genotype protects against anemia during peginterferon and ribavirin therapy but does not influence virological response

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

On-treatment anemia is associated with higher sustained virological response (SVR) rates during peginterferon plus ribavirin (RBV) therapy. Inosine triphosphatase (ITPA) variants causing ITPase deficiency have been shown to protect against RBV-induced anemia. However, ITPase activity has not been associated with SVR. To study this discrepancy, we examined the relationships between ITPase activity, on-treatment anemia, SVR, and RBV levels in hepatitis C virus genotype 1 (HCV-1) patients from the CHARIOT study. ITPA genotype (rs7270101, rs1127354) was used to define ITPase activity in 546 patients. Plasma RBV levels were measured using high-performance liquid chromatography (HPLC). Relationships between ITPase activity, on-treatment hemoglobin (Hb) levels, RBV levels, and SVR were tested using regression modeling, survival analysis, and locally weighted scatterplot smoothing (LOWESS) plot analysis. Hb decline was independently associated with SVR (P < 0.0001). ITPase deficiency was present in 35%. ITPase deficiency strongly protected against Hb decline (P < 0.0001), but was not associated with SVR (P = 0.28). The probability of SVR increased with lower nadir Hb for both wild-type and deficient ITPase activity, but the association curve shifted to describe a parallel relationship at higher Hb levels in patients with ITPase deficiency. In a subset (n = 203), we tested the hypothesis that the association between Hb decline and SVR reflected RBV levels rather than actual Hb level. RBV levels were associated with on-treatment Hb decline and SVR, but not ITPase activity. In regression models, adjustment for RBV levels attenuated the association between Hb decline and SVR. Conclusion: ITPase deficiency protects against RBV-induced anemia, but is not associated with SVR. Our data suggest that the relationship between Hb decline and SVR is not mechanistic, but is linked to RBV levels. (Hepatology 2014;59:2152–2160)

Abbreviations
DAA

direct-acting antiviral

Hb

hemoglobin

HCV

hepatitis C virus

HCV-1

hepatitis C virus genotype 1

HWE

Hardy-Weinberg equilibrium

ITPA

inosine triphosphatase

LOWESS

locally weighted scatterplot smoothing

peg-IFN

peginterferon

RBV

ribavirin

SVR

sustained virological response

Genetic variation in the inosine triphosphatase (ITPA) gene on chromosome 20 has recently been strongly associated with protection from hemolytic anemia in patients with hepatitis C virus (HCV) infection treated with peginterferon-α (peg-IFN) and ribavirin (RBV).[1-3] Two functional variants that cause ITPase deficiency, rs1127354 and rs7270101, reduce RBV-induced hemolysis by preventing depletion of erythrocyte adenosine triphosphate (ATP) and oxidative stress.[4] The protective ITPA genotypes are associated with lower rates of RBV dose reduction and higher cumulative RBV doses.[2, 3] However, despite the well-documented association between RBV exposure and response to peg-IFN and RBV therapy, ITPA genotype has not been found to predict virological outcome.

The relationship between RBV therapy, on-treatment hemoglobin (Hb) decline, and virological response to peg-IFN and RBV treatment is complex. RBV is essential for maximizing antiviral potency and reducing relapse rates during dual therapy with peg-IFN and RBV, and continues to play an important role in the setting of direct-acting antiviral (DAA) therapy.[5-8] Anemia is a common side effect of peg-IFN plus RBV therapy, affecting up to 30% of patients. The most important cause of treatment-related anemia is RBV-induced hemolysis, although peg-IFN-related bone marrow suppression also contributes. While RBV exposure has been shown to be important for optimizing virological response,[9, 10] and anemia frequently requires RBV dose reduction, we and others have previously shown that on-treatment anemia itself is an important and independent predictor of sustained virological response (SVR) following dual therapy with peg-IFN and RBV.[11, 12] More recently, anemia has also been shown to predict the outcome of a triple therapy regimen including the protease inhibitor, boceprevir.[13] However, the mechanism for the association between Hb reduction and SVR remains poorly defined. The most popular hypothesis is that Hb levels can be considered a surrogate for RBV exposure, but this has been difficult to prove. Interpretation of clinical trial results has been limited by frequent use of erythropoietin, a paucity of plasma RBV level data, and now potential confounding by ITPA genotype.[11, 13] Currently, RBV levels are not part of routine monitoring during peg-IFN plus RBV therapy for HCV and there are only limited data in small cohorts. Week 4 and 8 plasma RBV levels have been associated with treatment response, and also with on-treatment anemia.[14, 15] Target plasma RBV levels ranging between 2.0-2.5 mg/L have been reported to be optimal from previous threshold analyses.[16]

In order to further investigate the relationships between ITPA genotype, on-treatment Hb levels, and SVR, we performed a retrospective analysis of a large cohort of well-characterized patients from the CHARIOT study. CHARIOT was a phase IV multicenter international study that evaluated the use of induction-dosing peg-IFN in treatment-naïve patients with genotype 1 HCV infection (HCV-1). Induction-dosing with 360 μg peg-IFN subcutaneously weekly was well tolerated but did not improve SVR rates.[17] The rate of hemopoietic growth factor use was less than 2%. Our group have previously confirmed an association between decline in Hb level and virological response in this cohort.[12] The aim of this further analysis was to examine the complex associations between ITPA genotype, on-treatment hemoglobin decline, and treatment response in a large cohort of HCV-1 participants from the CHARIOT study where the use of hemopoietic growth factors was not routinely available, and a subset of patients, to examine the additional role of RBV pharmacokinetics.

Materials and Methods

Patients

Patients included in this study were participants in the CHARIOT study.[17] In brief, 871 patients were randomized to induction-dose peg-IFNα-2a (360 μg subcutaneously weekly) versus standard-dose peg-IFNα-2a (180 μg subcutaneously weekly), plus weight-based RBV at 1,000-1,200 mg per day for 12 weeks, followed by 36 weeks of peg-IFNα-2a 180 μg subcutaneously weekly plus weight-based RBV at 1,000-1,200 mg per day. Induction-dosing was well tolerated, but was not associated with any SVR benefit.[17] Liver histology was not required for inclusion, but was available in a subset (n = 625/871). Of the 871 CHARIOT participants, 564 consented to the storage of clinical material for biomarker studies. Specific dose-adjustment guidelines were included in the study protocol.[12, 17] In brief, RBV dose reduction was indicated for an Hb value less than 100 g/L, with stepwise reduction of 200 mg decrements. RBV was withheld if the Hb was less than 85 g/L. Hemopoietic growth factor support for patients in this study was infrequent (<2%). Four patients eligible for inclusion in the current substudy received erythropoietin, and these patients were excluded from the analysis. The protocol specified that peg-IFN dose reduction was mandatory if neutropenia occurred.[17] Physician-prescribed changes to study medication dose throughout the treatment period and patient discontinuation from treatment were recorded during the original study. The reasons for alteration in the prescribed medication doses were recorded, allowing for calculation of percentage of the expected peg-IFN and RBV doses received, cumulative peg-IFN and RBV dose, and proportion of total number of RBV dose reductions required for the indication of anemia.

ITPA Genotyping

Stored serum was used for genotyping of the ITPA polymorphisms rs1127354 and rs7270101. DNA was extracted from serum samples and eluted in 50 μL of AE buffer using the QIAmp DNA mini kit (Qiagen, Valencia, CA) as per the manufacturer's instructions. ITPA genotyping was performed using the TaqMan allelic discrimination kit using real-time polymerase chain reaction (PCR) (Applied Biosciences, Foster City, CA), as previously described.[1] Cycle threshold values of less than 36.5 were accepted. Where the cycle threshold values were greater than or equal to 36.5, or if there was difficulty determining positivity and negativity for the two alleles, the extracted DNA was precipitated, concentrated, and redissolved, or DNA was reextracted from the remaining serum. ITPA genotyping was then repeated, and again, only samples with cycle values of less than 36.5 were accepted.

Predicted ITPase Activity and Definition of an ITPase Activity Variable According to rs1127354 and rs7270101 ITPA Genotypes

Predicted ITPase activity, and hence severity of ITPase deficiency, was defined by combining both ITPA genotypes as described in previous studies.[1, 18-21] For the purpose of this analysis, predicted ITPase activity was divided into a binary variable: normal (or wild-type) ITPase activity versus ITPase deficiency (combined mild, moderate, and severe).

Plasma Ribavirin Levels

Plasma ribavirin level data were available for 203 patients from a previous substudy examining the relationship between RBV levels and fibrosis stage. Data were available for weeks 4, 8, and 12. Plasma samples were separated from whole blood within 2 hours of collection and aliquots were stored at −20°C until analysis was performed. Ribavirin levels were determined using a modified high-performance liquid chromatography (HPLC) with ultraviolet detection, as previously described.[22] In brief, 1.5 mL of 5-bromouracil was added to 250 μL of plasma. After centrifugation, the supernatant was removed and evaporated at 50°C. After resuspension of the dried supernatant at mobile phase, 30 μL was added for chromatography. The Atlantis T3 column (Waters, Milan, Italy) was used for separation, with detection at 235 nm. The lower limit of quantification was 0.25 mg/L, and linear calibration curves were observed over 0.25-5.0 mg/L.

Statistical Analysis

For descriptive statistics, continuous variables are summarized as median (25th to 75th centile). Categorical variables were described as frequency and percentage. Comparisons between groups for demographic, clinical, and virological data were performed using a Wilcoxon test for continuous data or the chi-squared test / Fisher's exact test for categorical data. Significance was defined at P < 0.05.

We considered the following anemia outcomes at week 4: Hb reduction (continuous variable), Hb reduction >30 g/L (categorical variable), and anemia (Hb <100 g/L). We also considered patterns of Hb decline over the entire treatment course, including nadir Hb level, Hb reduction >30 g/L, and anemia (Hb <100 g/L) at any time during treatment.

The association between predicted ITPase activity and anemia phenotypes was tested using single-marker genotype trend tests of association in linear or logistic regression models. Multivariate regression models with backward selection were used to identify independent predictors of anemia. A significance level of 0.05 was used for removal from the model. Covariates included age, gender, body mass index (kg/m2), baseline Hb level, baseline serum creatinine level, and RBV starting dose (mg/kg). Similarly, multivariate logistic regression modeling was used to explore the relationships between ITPase activity, Hb decline, RBV levels, and SVR. We used the locally weighted scatter plot smoothing (LOWESS) method to explore the relationship between SVR and on-treatment Hb according to the predicted ITPase activity. The association between predicted ITPase activity and the timepoint of first RBV dose reduction was also tested using survival analysis.

All statistical analyses were performed in SAS v. 9.2 (SAS Institute, Cary, NC) or Stata v. 10.1 (StataCorp LP, College Station, TX).

Results

Patient and Treatment Characteristics

In all, 564 of 871 subjects (64.8%) consented to storage of samples for biomarker testing and 546 (96.8%) of these patients were successfully genotyped for the causal ITPA polymorphisms, rs1127354 and rs7270101. The clinical characteristics of the study population are described in Table 1. There were no significant differences in patient characteristics or baseline demographics between the patients randomized to either standard-dosing or induction-dosing peg-IFN (Supporting Table 1) or according to ITPase activity (Table 1), and the population for this analysis was similar to the overall CHARIOT study population.[17] The observed SVR rate did not differ significantly from that observed in the parent study (57% versus 52%). There was no significant difference in the percentage of patients who received standard versus induction-dose peg-IFN according to ITPase activity (Table 1).

Table 1. Baseline Demographics, Patient Characteristics, and Anemia Endpoints According to ITPase Activity
 ITPase wild-type activity (n = 356)ITPase deficiency (n = 190)P
Age, years (median, 25th – 75th centile)45 (39-50)44 (37-50)0.2383
Male gender (n, %) 236 (66%)133 (70%)0.3890
Ethnicity (n, %)Caucasian312 (88%)169 (89%)0.5360
Asian35 (10%)19 (10%) 
Other9 (2%)2 (1%) 
Body Mass Index (kg/m2) (median, 25th – 75th centile)26.1 (23.1-29.1)26.7 (22.8-29.0)0.8556
HCV genotype (n, %)1142 (40%)68 (36%)0.8590
1a90 (25%)53 (28%) 
1b99 (28%)57 (30%) 
1a/b25 (7%)12 (6%) 
Baseline HCV RNA (log10IU/mL)(median, 25th – 75th centile)6.2 (5.7-6.7)6.3 (5.8-6.7)0.7342
Baseline HCV RNA>800,000 IU/mL251 (71%)142 (75%)0.5000
Baseline ALT (IU/mL)(median, 25th – 75th centile)50.6 (31.5-75.8)51.9 (33.6-90.4)0.3340
METAVIR Fibrosis stage (n, %)F0-2176/217 (81%)90/111 (81%)1.0000
F3-441/217 (19%)21/111 (19%) 
Missing139 (39%)79 (42%) 
Treatment arm (n, %)Standard-dosing179 (50%)90 (47%)0.5300
 Induction-dosing177 (50%)100 (53%) 
Baseline Hb (g/L)(median, 25th – 75th centile)150 (141-159)152 (141-160)0.6317
Week 4 Hb (g/L)Median (median, 25th – 75th centile)123 (114-132)139 (127-148)<0.0001
<100g/L (n, %)19 (5%)3 (2%)0.0390
>30g/L reduction (n, %)149 (42%)17 (9%)<0.0001
Overall Hb (g/L)Nadir Hb (median, 25th – 75th centile)111 (101-119)118 (107-127)<0.0001
Hb <100g/L (n, %)79 (22%)26 (14%)0.0380
>30g/L reduction (n, %)291 (82%)120 (66%)<0.0001

Anemia endpoints were considered at week 4, as well as over the entire course of treatment. Week 4 was chosen to immunize the impact of RBV dose reduction on the analysis of the relationship between ITPase deficiency and Hb reduction. We also considered anemia and Hb reduction at any point during treatment given the previously documented association with SVR. The overall frequency of anemia endpoints is summarized in Table 1, and was not significantly different for patients who received standard versus induction dosing of peg-IFN, either at week 4 or overall (Supporting Table 1).

ITPA Genotype Distribution

The genotype frequencies of the polymorphisms rs1127354 and rs7270101 are summarized in Table 2. The minor allele (A) frequency for rs1127354 was 0.08 and the population distribution was in Hardy-Weinberg equilibrium (HWE) (P = 0.13). The minor allele frequency (C) for rs7270101 was 0.10, with a minor deviation from HWE (P = 0.01). The two ITPA genotypes were then combined to define the predicted ITPase activity (Table 2). Thirty-five percent of the patients were predicted to have ITPase deficiency (Table 2), with 18.9% demonstrating mild ITPase deficiency and 15.9% predicted to have moderate-severe ITPase deficiency. ITPase deficiency was present in 50% of patients receiving standard and 50% of patients randomized to induction-dose peg-IFN. As the patient demographics, virological response, anemia outcomes, and ITPA genotype frequencies were comparable between the two treatment arms, all patients were pooled together for subsequent analyses.

Table 2. Definitions and Frequencies of Predicted ITPase Activity With Corresponding rs1127354 and rs7270101 ITPA Genotypes
rs1127354 genotype (Minor allele A)rs7270101 genotype (Minor allele C)Combined ITPA genotypePredicted IT Pase activityPredicted IT Pase deficiencyn (%)
C/CA/ACCAA100%None [Wild-type]356 (65.2%)
C/CA/CCCAC60%Mild [+]103 (18.9%)
C/AA/ACAAA30%Moderate [++]81 (14.8%)
C/CC/CCCCC30%Moderate [++]0 (0%)
C/AA/CCAAC10%Severe [+++]6 (1.1%)
A/AA/AAAAA<5%Severe [+++]0 (0%)

ITPase Deficiency and Hb Reduction and On-treatment Anemia

ITPase activity was strongly associated with quantitative week 4 Hb reduction in multivariate linear regression analysis, where ITPase deficiency was protective. Fewer patients with ITPase deficiency compared to wild-type ITPase activity experienced an Hb reduction >30 g/L at week 4 (9.0% versus 42.1%, P < 0.0001) or Hb <100 g/L at week 4 (1.6% versus 5.4%, P = 0.039) (Table 1). Median Hb levels were higher at all timepoints during treatment in patients with ITPase deficiency, with the differences in median Hb most marked during the first 12 weeks of treatment (Fig. 1A). ITPase activity was also strongly associated with anemia outcomes at any point during treatment. Specifically, ITPase deficiency was associated with nadir Hb (median 111 versus 118 g/L in patients with wild-type ITPase activity versus ITPase deficiency, P < 0.0001) and Hb <100 g/L (22.2% versus 14.4% in patients with wild-type ITPase activity versus ITPase deficiency, P = 0.038). There was no association between the ITPA variants, or the ITPase deficiency variable, and baseline Hb level (Table 1). We also considered the relationship between ITPase deficiency and time to key anemia outcomes. ITPase deficiency was associated with a significant delay to time of first Hb decline >30 g/L, as well as delay to first Hb <100 g/L (Fig. 1B,C).

Figure 1.

(A) Survival analysis of time to Hb reduction >30 g/L according to ITPase activity (wild-type versus deficiency). (B) Survival analysis of time to Hb <100 g/L according to ITPase activity. (C) Median Hb during treatment according to ITPase activity. *P < 0.0001. (D) Survival analysis of RBV dose reduction in the first 170 days of treatment according to ITPase activity.

ITPase Deficiency and RBV Dose Reduction

A total of 259 patients were documented to have at least one RBV dose reduction during the course of treatment. Listed indications for RBV dose reduction in the study database were: anemia (Hb <100 g/L) in 72 patients (27.8%), “other” RBV-related side effects in 56 patients (21.6%), and one or more missed doses in 181 patients (69.9%). The median time to first RBV dose reduction for anemia was 88.5 days (44.8-195.5). We considered the cumulative probability of RBV dose reduction for anemia within the first 24 weeks of therapy, as nonresponders were required to stop treatment at week 24 for futility. ITPase deficiency was significantly associated with fewer RBV dose reductions (Fig. 1D, P = 0.024). This translated into a higher cumulative RBV dose (mg/day) in those with ITPase deficiency compared to wild-type activity (1,072 versus 1,054 mg/day, P = 0.026).

Associations Between ITPase Activity, SVR, and Hb Decline

ITPase deficiency was not associated with virological response to peg-IFN and RBV therapy. There was no association with responses at week 4, week 12, or end-of-treatment, nor was there an association with SVR (Tables 3, 4,{TBL3-4} Fig. 2A; Supporting Table 2). In addition, when considering the stricter phenotype of ITPase deficiency (moderate to severe ITPase deficiency), there was still no association with SVR (59.8% versus 55.1% for moderate-severe ITPase deficiency and ITPase WT activity respectively, P = 0.471). However, patients who developed anemia (Hb <100 g/L) at any time during therapy have previously been shown to achieve higher SVR rates in this cohort.[12] We confirmed that the reduction of Hb to a level <100 g/L at any time during treatment was independently associated with SVR in the current ITPA subset (SVR 65.0% versus 54.5%, P = 0.046 for Hb <100 g/L and Hb ≥100 g/L, respectively, Fig. 2B, Table 3). The nadir Hb level on-treatment was also independently associated with SVR (P = 0.001, Table 4). This was true in the overall ITPA cohort, as well as the subset of patients in whom liver histology could be considered in the model (n = 319, Table 4).

Table 3. Multivariate Logistic Regression Models for Sustained Virological Response (SVR) and Hb <100g/L Any Time During Treatment
Multivariate model for SVR including Hb <100g/L at any time during treatment
  1. Other covariates added to the models included age, sex, ethnicity, body mass index (BMI), baseline HCV RNA level (log10IU/mL), baseline ALT, and ITPase activity. Fibrosis stage (F0-1 vs. F3-4) was added to model B) and week 8 ribavirin (RBV) levels were added to model C). Backwards selection was used to remove variables with P ≥ 0.05 from the model.

Multivariate model A): without fibrosis stage (n = 546)
VariablesOdds Ratio (95% CI)P
Age (per year)0.95 (0.93-0.97)<0.0001
Male gender0.59 (0.39-0.89)0.0125
Hb <100g/L any time1.68 (1.02-2.76)0.0433
HCV RNA (log10IU/mL)0.69 (0.53-0.89)0.0051
Multivariate model B): including fibrosis stage (n = 319)
VariablesOdds RatioP
Age (years)0.953 (0.92-0.98)0.0017
Hb <100g/L any time2.67 (1.39-5.14)0.0032
Fibrosis stage (F0-1 vs. F3-4)0.24 (0.13-0.46)<0.0001
HCV RNA (log10IU/mL)0.69 (0.49-0.96)0.0297
Multivariate model C): including fibrosis stage and week 8 RBV levels (n = 203)
VariablesOdds RatioP
Age (years)0.94 (0.903-0.979)0.0026
Hb <100g/L any time3.039 (1.271-7.268)0.0125
Fibrosis stage (F0-1 vs. F3-4)0.178 (0.072-0.436)0.0002
Figure 2.

(A) ITPase activity (wild-type versus deficiency) and SVR. (B) Hemoglobin <100 g/L or ≥100 g/L any time during therapy and SVR.

Table 4. Multivariate Logistic Regression Models for Sustained Virological Response (SVR) and Nadir Hb During Treatment
Multivariate model for SVR including nadir Hb during treatment
  1. Other covariates added to the models included age, sex, ethnicity, body mass index (BMI), baseline HCV RNA level (log10IU/mL), baseline ALT, and ITPase activity. Fibrosis stage (F0-1 vs. F3-4) was added to models (B) and (C), and week 8 RBV levels to model (C). Backwards selection was used to remove variables with P ≥ 0.05 from the model.

Multivariate model A): all patients (not including fibrosis stage, n = 546)
VariablesOdds RatioP
Age (years)0.949 (0.928-0.970)<0.0001
Nadir Hb (g/L)0.976 (0.963-0.990)0.0005
HCV RNA (log10IU/mL)0.678 (0.523-0.880)0.0035
Multivariate model B): including fibrosis stage (n = 319)
VariablesOdds RatioP
Age (years)0.936 (0.899-0.976)0.0017
Fibrosis Stage (F0-1 vs. F3-4)0.148 (0.059-0.371)0.0036
Nadir Hb (g/L)0.963 (0.940-0.988)<0.0001
Multivariate model C): including fibrosis stage and week 8 RBV levels (n = 203)
VariablesOdds RatioP
Age (years)0.933 (0.895-0.974)0.0013
Fibrosis Stage (F0-1 vs. F3-4)0.149 (0.059-0.381)<0.0001
Week 8 RBV level (mg/L)1.79 (1.079-2.971)0.0242
Nadir Hb (g/L)0.974 (0.949-1.010)0.0556

We then performed a stratified analysis to explore the relationship between ITPase activity, nadir Hb and SVR in greater detail. We used the LOWESS method to plot the estimated local probabilities of SVR against nadir Hb levels for patients with wild-type ITPase activity versus ITPase deficiency (Fig. 3). The plots demonstrated that the probability of SVR increased with lower nadir Hb levels in both groups, to a maximum benefit at nadir levels of Hb <100 g/L. In patients with ITPase deficiency the LOWESS plot was shifted to the right, describing a parallel curve at higher nadir Hb levels (Fig. 3). We interpreted the data as demonstrating that the relationship between SVR and nadir Hb was similar irrespective of ITPase activity, but the set-point at which this relationship occurs was higher in patients with ITPase deficiency. The data therefore strongly suggested that the SVR benefit associated with Hb reduction was not a direct effect of Hb level itself. We hypothesized that the relationship between Hb decline and SVR might be explained by an association between RBV pharmacokinetics and SVR.

Figure 3.

LOWESS plot of estimated local probability of SVR and nadir Hb levels according to ITPase activity.

Plasma RBV Levels and Association Between Hb Decline and SVR

Ribavirin exposure is directly linked to reduction in Hb levels by way of erythrocyte retention of RBV metabolites causing oxidative stress and hemolysis. Ribavirin exposure has previously been associated with response to peg-FN and RBV treatment,[9, 10, 23] and therapeutic RBV levels have been reported in the range of 2.0-2.5 mg/L.[16]

We therefore tested the hypothesis that the relationship between Hb decline and SVR is a reflection of RBV exposure. Plasma RBV level data were available for 203 patients at weeks 4, 8, and 12 from a previous substudy that examined if RBV levels differ by fibrosis stage.[24] Patients were selected if liver histology was available and if adequate plasma was available at weeks 4, 8, and 12. Patient demographics and baseline characteristics were similar between the RBV substudy cohort and the original ITPA cohort (Supporting Table 3).

Ribavirin is recognized to have complicated pharmacokinetics, with delayed achievement of steady-state concentrations and considerable interindividual variation.[25, 26] In the current cohort, plasma RBV levels were observed to increase between week 4 and week 8, before plateauing between weeks 8 and 12, suggesting steady state was reached between weeks 4 and 8 (median plasma RBV concentration at week 4 was 1.74 mg/L [1.32-2.31], at week 8 was 1.91 mg/L [1.54-2.43] and at week 12 was 2.00 mg/L [1.52-2.51], P < 0.0001 for week 4 versus 8 and P = 0.4374 for week 8 versus 12, respectively). We therefore chose to focus on week 8 plasma RBV levels for all subsequent analyses. Week 8 RBV levels were strongly associated with nadir Hb level (P < 0.0001). In a linear regression model, both ITPase deficiency and week 8 RBV levels were independent predictors of nadir Hb level. There was no statistically significant association between ITPase deficiency and plasma RBV levels (P = 0.11).

Greater Hb decline between weeks 4 and 12 have previously been associated with SVR in the parent cohort.[12] We observed a trend for first Hb decline <100 g/L between weeks 5 and 12 to be associated with higher rates of SVR in the current pharmacogenetics cohort (SVR 67% versus 56% for Hb <100 g/L and ≥100 g/L, respectively, P = 0.15). Similar findings were also seen between weeks 13 and 24, but not during the first 4 weeks of therapy. When a stratified analysis was performed according to ITPase activity, Hb decline <100 g/L was significantly associated with SVR among patients with wild-type ITPase activity (n = 354, SVR 69% versus 53% for Hb <100 g/L and ≥100 g/L, respectively, P = 0.047), but not among patients with ITPase deficiency (n = 188, SVR 64% versus 60% for Hb <100 g/L and ≥100 g/L, respectively, P = 0.71).

We subsequently evaluated the relationships between nadir Hb level, plasma RBV levels, and treatment outcome in this subset of patients. As was observed in the overall ITPA cohort, nadir Hb level was independently associated with SVR (P < 0.0001, Table 4). Plasma RBV levels at week 8 were also associated with SVR by univariate analysis (P = 0.011). In multivariate analyses adjustment for plasma RBV levels attenuated the association between nadir Hb and SVR (Table 4), suggesting that RBV levels explained the association signal. Exploratory analyses for critical RBV level thresholds that might be used to inform clinical dosing and maximize SVR rates were examined; however, we were not able to identify any clinically useful thresholds, although we note that only 23.9% of patients had levels in the reported therapeutic range of 2.0-2.5 mg/L at week 8, with the majority (55.1%) having subtherapeutic levels.

Discussion

ITPA variants causing ITPase deficiency have been shown to be strongly associated with protection from RBV-induced hemolytic anemia during peg-IFN and RBV dual therapy.[1, 2] In separate studies, Hb decline during treatment has been shown to predict the likelihood of SVR. However, ITPA variants have not been linked to virological response, and the reason for this discrepancy has not previously been explained. We performed a detailed study of the relationships between ITPase deficiency, on-treatment Hb reduction, and treatment response in a large, well-characterized cohort of patients from the CHARIOT study in which confounding by erythropoietin use was absent.

The data confirm the strong protective effect of ITPase deficiency against Hb decline during peg-IFN and RBV dual therapy. Furthermore, ITPase deficiency was associated with a higher cumulative RBV dosage, and yet was not associated with SVR. This was observed when considering any degree of ITPase deficiency compared to wild-type (WT) ITPase activity, and also when comparing a more severe ITPase deficient phenotype (moderate to severe deficiency) and WT ITPase activity. As previously described, hemoglobin decline was, however, associated independently with SVR. Our analysis provides further insight into this paradox by showing for the first time that the relationship between Hb reduction and SVR is independent of ITPase activity. LOWESS plot analyses clearly demonstrated parallel curves for the relationships between nadir Hb and SVR in patients with WT versus deficient ITPase activity, where in patients with WT ITPase activity the curve was described at lower Hb levels. ITPase activity therefore modulates the association between RBV and hemolysis, but not the association between Hb decline and treatment response. The models suggest that ITPase activity has no direct influence on treatment response.

The relationship between Hb decline and treatment response is likely to reflect RBV exposure. This hypothesis has been put forward previously, but has been difficult to validate. This assay is not widely available, and interindividual variation in RBV pharmacokinetics may confound small cohorts. One of the strengths of the current study is the analysis of plasma RBV levels in a large subset of patients. Plasma RBV levels were strongly associated with both nadir Hb decline and treatment response in the current dataset. Moreover, after adjustment for plasma RBV levels, Hb decline was no longer significantly associated with SVR. Therefore, the relationship between Hb decline and SVR is not mechanistic, but is rather explained by RBV pharmacokinetics. The data support the importance of adequate RBV exposure for maximizing virological responses during dual therapy with peg-IFN and RBV. Although we did not observe any association with overall SVR, it is possible that a small but clinically relevant effect might be observed in certain subpopulations, particularly patients at higher risk for anemia such as the elderly.

Peginterferon and RBV dual therapy is no longer standard-of-care for genotype 1 HCV. First-line treatment is now the combination of a protease inhibitor (PI), telaprevir or boceprevir, in combination with peg-IFN and RBV as triple therapy.[27-29] Anemia is more problematic, due to the incremental Hb reduction associated with PI treatment, and ITPA genotyping may be more useful for individualizing the risk of anemia-related morbidity. RBV still remains critical for preventing the emergence of PI-resistant variants. Furthermore, on-treatment anemia is still associated with improved treatment outcome.[13] Although RBV may be more liberally dose-reduced in the context of PI triple therapy, we believe our data support the need for detailed studies of the clinical utility of ITPA genotyping and plasma RBV level measurement for minimizing the risk of anemia-related morbidity, while maximizing virological responses in patients receiving direct-acting antiviral therapy.

In conclusion, the current analysis confirms that ITPA genotype protects against RBV-hemolysis, but is not associated with SVR. We have shown that the relationship between Hb decline and SVR is independent of ITPA genotype. Furthermore, analysis of plasma RBV levels in a subset of patients suggests that the relationship between Hb reduction and SVR is not mechanistic, but is explained by RBV pharmacokinetics. The data emphasize the importance of adequate RBV exposure during dual therapy for HCV. Studies evaluating the role of ITPA genotyping and monitoring of plasma RBV levels in the context of PI-based triple therapy are warranted.

Acknowledgment

AJT and GM are supported by Career Development Fellowships from the National Health and Medical Research Council of Australia (NHMRC). The study was supported by a research grant from Roche Products, Pty, Limited (Australia).

Author Contributions

Study concept and design: JAH, AJT, GVM; Acquisition of data: JAH, AJT, GVM, SKR; Analysis and interpretation of data: JAH, AJT, GVM; Drafting the article: JAH, AJT, GVM; Critical revision of the article; All listed authors; Statistical analysis: JAH, AJT, VS; Funding: AJT; ITPA testing and supervision: JAH, DSB; RBV level testing and supervision: RA, GJD, GVM.

Ancillary

Advertisement