Boceprevir dosing for late responders and null responders: The role of bridging data between treatment-naïve and -experienced subjects§

Authors


  • Potential conflict of interest: Nothing to report.

  • Data for these analyses were submitted to the United States Food and Drug Administration by Merck as part of a new drug application.

  • §

    Disclaimer: the contents of the article represent the authors' personal opinion and do not necessarily reflect any position of the Food and Drug Administration.

  • See Editorial on Page 875

On May 13th, 2011 the U.S. Food and Drug Administration (FDA) approved boceprevir (BOC) for use in combination with peginterferon alpha and ribavirin (P/R) for the treatment of genotype 1 chronic hepatitis C virus (HCV) in adults1 who are either P/R treatment-naïve or -experienced.1 BOC is an HCV NS3/4A protease inhibitor and represents a new class of small molecules that directly targets HCV replication. The pivotal Phase III trials supporting BOC approval were SPRINT-II,2 which included treatment-naïve subjects, and RESPOND-II,3 which included subjects who were prior P/R relapsers (HCV RNA-negative at end of treatment with P/R but rebound of HCV RNA off treatment) or partial responders (≥2 log10 decline in HCV RNA and HCV RNA detected at week 12 but never achieving HCV RNA undetected).

Two key questions that the FDA needed to consider for BOC labeling recommendations included: (1) Is there evidence of effectiveness for prior P/R null responders (defined as <2 log10 decline from baseline in HCV-RNA after 12 weeks of P/R treatment), a subgroup that was specifically excluded from RESPOND-II? (2) What is an appropriate dosing regimen for a subset of treatment-naïve subjects who are late responders (as defined below)?

To address these questions, analyses were performed by bridging knowledge from the sponsor's two registrational trials, which ultimately supported certain dosing recommendations that were not prospectively evaluated during Phase 3 registrational trials. These bridging analyses are based on the concept that a patient's virologic response to P/R remains relatively unchanged with subsequent treatment courses.4 This is unlike treatment with directly targeting antivirals, such as those for HCV or HIV, where treatment failure is often associated with drug resistance that can affect responsiveness to subsequent courses of treatment. Other examples where bridging analyses have impacted regulatory decision making include oxcarbazepine, topiramate, clevidipine, and levofloxacin, to name a few.5-8 Bridging knowledge to provide clinical evidence of effectiveness and to support dosing recommendations not only is acceptable from a regulatory perspective, but when scientifically supported and warranted it is also encouraged to increase the efficiency by which new drugs and optimal dosing recommendations are made available to patients in need.9 This report summarizes the rationale to support the BOC dosing recommendations in prior P/R-null responders and treatment-naïve BOC late responders.10

Abbreviations

BOC, boceprevir; FDA, U.S. Food and Drug Administration; HCV, hepatitis C virus; P/R, peginterferon alpha and ribavirin; RGT, response-guided therapy; SVR, sustained virologic response.

BOC Phase III Trial Designs

The SPRINT-II and RESPOND-II trial designs are illustrated in Fig. 1. The primary endpoint for both SPRINT-II and RESPOND-II was the proportion of subjects with SVR (sustained virologic response), defined as HCV RNA undetected at 24 weeks after the last dose of the study drug. Both trials had three treatment arms: (1) P/R for 48 weeks (Arm 1, P/R); (2) a response-guided therapy (RGT) arm with P/R lead-in for 4 weeks, followed by P/R with BOC for 24 weeks in SPRINT-II and 32 weeks in RESPOND-II (Arm 2, BOC-RGT); and (3) P/R lead-in for 4 weeks, followed by BOC with P/R for 44 weeks (BOC44) (Arm 3, BOC44). In SPRINT-II subjects randomized to the RGT arm who had HCV RNA undetected at weeks 8 through week 24 stopped all treatment at week 28 and were classified as BOC treatment-naïve early responders. The remaining subjects in the RGT arm, who had HCV RNA detected at treatment week 8 or any subsequent week, but who had HCV RNA undetected at week 24, stopped BOC at week 28 but received P/R through week 48. These subjects are referred to as BOC treatment-naïve late responders. In the RGT arm of RESPOND-II treatment-experienced early responders (subjects with HCV RNA undetected at weeks 8 through 12) stopped all treatment at week 36, whereas treatment-experienced late responders (HCV RNA detected at week 8 but HCV RNA undetected at week 12) stopped BOC at week 36 but continued P/R through week 48.

Figure 1.

Study designs and treatment arms for SPRINT-II2 (top) and REPSOND-II3 (bottom).

Regulatory Decisions

Predicting Treatment Effect in Prior Null Responders.

Although prior P/R null responders were excluded from RESPOND-II, the sponsor proposed that data from treatment-naïve subjects could be used to estimate the treatment effect in prior null responders because included among treatment-naïve subjects are a subset of subjects who are intrinsically poor responders to interferon. The sponsor proposed a surrogate definition for categorizing a subject as a P/R null responder, not based on prior treatment history, but based on week 4 HCV RNA measurements during the PR-only lead-in period from SPRINT-II. Based on retrospective analyses of the IDEAL trial,11 the sponsor proposed that an HCV-RNA decline ≤1.0 log10 at week 4 was predictive of the more traditionally defined null response on P/R (i.e., <2 log10 HCV RNA decline at week 12). SVR rates for SPRINT-II subjects administered BOC who had <1.0 log10 decline at week 4 were 28% in Arm 2 and 38% in Arm 3, compared to 4% in the P/R control arm.

The FDA's SVR analysis of treatment-naïve subjects in SPRINT-II demonstrated that 31% (26/83) of subjects with <1.0 log10 decline at week 4 in the P/R control arm (Arm 1) would be incorrectly classified as null responders10 (the remaining subjects discontinued treatment, had a partial response, or relapsed). To obtain a more conservative estimate of the SVR rate in null responders, an alternative surrogate definition of <0.5 log10 HCV RNA decline at week 4 was investigated. Based on the end of study outcomes (i.e., null responder, partial responder, relapser, or responder achieving SVR) for such subjects in SPRINT-II treated with P/R (n = 25), 22 subjects were null responders (88%), one was a partial responder, and two discontinued treatment.

The SPRINT-II study design allowed us to analyze outcomes of treatment-naïve subjects who were treated with BOC (Arms 2 and 3) and had similar interferon responsiveness during the lead-in period. The SVR rates in these poorly interferon responsive subjects (defined here as <0.5 log10 decline in HCV RNA at week 4) who received BOC treatment was 28% (Arm 2: n/N = 13/47) and 30% (Arm 3: n/N = 11/37). By comparison, the observed SVR rate for poorly interferon responsive subjects treated with P/R (i.e., those with ≤0.5 log10 HCV RNA decline at week 4) was 0%. Whether a <1.0 or <0.5 log10 decline in HCV RNA at week 4 was used to categorize poorly interferon responsive subjects from SPRINT-II, a treatment benefit was demonstrated in the BOC regimens over treatment with P/R alone.

Further, it is important to consider the reasons behind equating prior null responders and treatment-naïve subjects with poor interferon responsiveness. The previous analysis demonstrated that BOC provided meaningful benefit in treatment-naïve subjects with interferon response characteristics similar to prior null responders. However, to consider using data from treatment-naïve patients to predict response in P/R-experienced patients, one needs to demonstrate that P/R response remains similar after a second course of P/R treatment. To address this question we assessed the relationship between virologic responsiveness through week 4 and treatment outcome for P/R arm subjects in SPRINT-II (i.e., log10 decline in HCV RNA at week 4 grouped according to end of study outcome) (Fig. 2). Similarly, subjects from RESPOND-II were grouped according to previous treatment outcome (relapser or partial responders) (Fig. 2).

Figure 2.

Interferon responsiveness (change from baseline in HCV RNA at week 4) for treatment-naïve (SPRINT-II – Arm 1 P/R: gray) and previously treated (RESPOND-II – all: black) subjects divided according to end of study status (i.e., null responder, partial responder, relapser, or responder achieving SVR) for treatment-naïve and baseline status for treated subjects. Data are plotted as counts (upper panel) and density (lower panel) as a function of interferon responsiveness. This figure demonstrates similar change from baseline in HCV RNA at week 4 for patients who had an end of study status of partial responder from SPRINT-II and baseline status of prior partial responder from RESPOND-II. A similar agreement was seen between relapsers from SPRINT-II and patients with baseline status of prior relapser from RESPOND-II.

In SPRINT-II, subjects treated with P/R who achieved an SVR (responders) had a median 3.4 log10 IU/mL decrease in HCV RNA at week 4, compared to 2.1 log10 IU/mL in relapsers, 1.6 log10 IU/mL in partial responders, and 0.7 log10 IU/mL in null responders. Likewise, known previous partial responders and relapsers from RESPOND-II had a similar week 4 HCV RNA decline as that observed in subgroups from SPRINT-II (prior relapsers: 2.2 log10 IU/mL [n = 253], prior partial responders: 1.2 log10 IU/mL [n = 140]) (Fig. 2). These results support using data from the treatment-naïve SPRINT-II trial in treatment-naïve subjects to predict a beneficial BOC treatment effect in prior P/R null responders.

A subsequent FDA analysis showed that interferon responsiveness remains relatively unchanged with a second round of P/R.6 Therefore, FDA agreed to bridge the data from “future” null responders (i.e., poorly interferon responsive subjects at week 4 of P/R treatment) in SPRINT-II to establish evidence of effectiveness in prior null responders. The FDA review concluded that BOC is expected to provide a treatment benefit in prior null responders.

Bridging Previous P/R Treatment Subjects and Treatment-Naïve Subjects to Derive Dosing Recommendations for Late Responders.

The second key question concerned dosing recommendations for late responders in the treatment-naïve population. SVR rates were similar between the RGT and BOC44 arms for the following: (1) treatment-naïve early responders (≈97% SVR rate) in SPRINT-II; (2) P/R-experienced subjects (study limited to prior relapsers and prior partial responders) who were early responders (≈91% SVR rate) in RESPOND-II; and (3) P/R-experienced late responders (≈79% SVR rate) in RESPOND-II. However, treatment-naïve late responders had a numerically lower SVR rate in the RGT arm (66%) compared to subjects in the BOC44 (75%) treatment arm in SPRINT-II.

The observed numerical difference was further explored by investigating the percentage of subjects with HCV RNA undetected over time for early and late responders in Arm 2 (RGT) and Arm 3 (BOC44) of SPRINT-II. There was an increase in the percentage of subjects with HCV RNA detected beyond week 28 in the RGT arm compared to the BOC44 arm (Fig. 3), reflecting an increase in virologic breakthrough in the RGT arm after subjects stopped BOC and continued on P/R only. In other words, for at least a subset of RGT arm subjects BOC continued to provide an antiviral treatment effect through week 28, which presumably would have continued with extended BOC exposure. Importantly, this difference in virologic breakthrough rates between the BOC44 and RGT arms was not observed among late responders in RESPOND-II, who remained on BOC for an additional 8 weeks through week 36.

Figure 3.

Percentage of treatment-naïve subjects with HCV RNA undetected at different treatment timepoints for late responders from SPRINT-II. A percentage of late responders in Arm 2 had HCV RNA detected at the subsequent treatment visit after treatment with boceprevir was halted at week 28.

Further analyses of interferon responsiveness (i.e., HCV RNA decline at week 4 on P/R treatment) demonstrated that subjects with characteristics similar to P/R treatment-experienced subjects are represented within treatment-naïve late responder subjects (Fig. 4). Subjects who were treatment-naïve late responders in SPRINT-II had a median 1.1 log10 IU/mL decrease in HCV RNA at week 4, compared to a median 3.1 log10 IU/mL decrease for treatment-naïve early responders. These results indicate that treatment-naïve late responders may benefit from following a BOC RGT treatment approach that is more comparable to what was validated for previous P/R treatment subjects in RESPOND-II (i.e., total BOC duration ≥32 weeks).

Figure 4.

Interferon responsiveness (change from baseline in HCV RNA at week 4) from SPRINT-II for subjects treated with only P/R (SPRINT-II – Arm 1 P/R: gray) or BOC and P/R (SPRINT-II – Arm 2 RGT: black). Subjects receiving BOC were divided according to early or late responder classification. Subjects receiving only P/R were divided according to end of study status. Data are plotted as counts (upper panel) and density (lower panel) as a function of interferon responsiveness. This figure demonstrates similar change from baseline in HCV RNA at week 4 for patients who were late responders from Arm 2 of SPRINT-II and patients from Arm 1 whose end of study status was relapser, partial responder, or null responder.

Based on these results, the following treatment recommendations for treatment-naïve late responders were considered:

  • Recommend the BOC44 regimen for treatment-naïve late responders and P/R 4 + BOC P/R 32 + P/R 12 for treatment-experienced late responders: This recommendation is based on empirical evidence, as the regimen was prospectively studied in SPRINT-II. However, this recommendation would result in treatment-naïve late responders receiving BOC for 44 weeks, whereas treatment-experienced late responders would receive BOC for only 32 weeks (i.e., a shorter BOC treatment duration for late responders who had previously failed a course of P/R than those patients receiving treatment for the first time).

  • Recommend P/R 4 + BOC P/R 32 + P/R 12 for treatment-naïve and treatment-experienced late responders (approved dosing regimen): This dosing recommendation was studied in treatment-experienced late responders, for whom BOC44 provided no apparent additional benefit. However, extending this dosing recommendation to treatment-naïve late responders relies on the bridging analysis between populations and the “interferon responsiveness” analysis.

  • Recommend P/R 4 + BOC P/R 32 + P/R 12 for treatment-naïve and treatment-experienced late responders AND BOC44 for “poor interferon responsive” subjects: This dosing recommendation modifies the recommendation in Option (2) to address concerns that subjects with “poor interferon responsiveness” may benefit from a longer duration of BOC exposure. However, this dosing recommendation introduces an additional decision point (log10 decline in HCV RNA at week 4), further complicating the dosing recommendations.

Option (1) was supported by empirical evidence; however, the review team recognized the inconsistency in this recommendation in that subjects with a known prior P/R treatment outcome would be treated with a shorter BOC duration than treatment-naïve subjects, regardless of similar interferon treatment responses. As such, Option (1) was considered less appropriate than Option (2). Option (3) was considered because it was anticipated that subjects with characteristics similar to prior null responders would also be more likely to meet the late responder criteria and that these subjects may benefit from a full 44 weeks of BOC treatment with P/R. However, Option (3) was rejected in favor of Option (2) because of its complexity and impracticality for use in the clinical setting.

Conclusions

The analyses summarized above provided the basis for including the following recommendations in the BOC label1:

  • Use of P/R lead-in for 4 weeks, P/R with BOC for 32 weeks, and 12 weeks of P/R in previously treatment-naïve subjects with late response (i.e., subjects with HCV RNA detected at week 8 of treatment but with HCV RNA undetected at week 24 of treatment).

  • Treatment recommendations of P/R lead-in for 4 weeks followed by P/R with BOC for 44 weeks for prior P/R null responders, with a caution that the population was not prospectively studied, and a suggestion that previously treatment-naïve subjects who are poorly interferon responsive (defined as <1-log10 decline in HCV RNA at week 4 of treatment) might benefit from longer treatment to maximize rates of SVR.

Ancillary