Abstract
Introduction: While the majority of patients with classical Hodgkin lymphoma (cHL) are cured with first-line multi-agent chemotherapy, approximately one-third progress to second- or later-line therapy (Copeland A et al, Curr Opin Oncol 2012). Treatment options for patients with relapsed or refractory (R/R) cHL include autologous hematopoietic cell transplantation (auto-HCT), brentuximab vedotin (BV), and a collection of chemotherapies and supportive agents that comprise best supportive care (BSC) (Ansell SM, Mayo Clin Proc 2015). Nivolumab (nivo), an anti-programmed death-1 immune checkpoint inhibitor, showed promising results in single-arm clinical studies that included patients with R/R cHL (Armand P et al, J Clin Oncol 2018). Due to the lack of comparative trials, the relative effects of nivo compared with BV and BSC are unknown. The objective of this study was to estimate the relative survival of nivo compared with BV or BSC in Canadian patients with cHL in whom auto-HCT had failed, using an unadjusted indirect treatment comparison (ITC) and matching-adjusted indirect comparison (MAIC).
Methods: Survival estimates from the ITC and MAIC were based on the comparison of area under the survival curves (AUC) over 15 years for patients who had received nivo versus BV or BSC. Nivo data were taken from Cohort A of the CheckMate 205 study (NCT02181738) that comprised patients with cHL in whom auto-HCT had failed and who were BV naïve. Data specific to BV were obtained from pivotal clinical trial results (Chen R et al, Blood 2016). Canadian data for BSC were obtained from a population-based registry of patients with cHL and failed auto-HCT, and a published observational study (unadjusted ITC only) (Dhamko H et al, Canadian BMT 2015 [poster]). Published survival curves for BV and BSC were used to recreate patient-level data and were analyzed alongside available patient-level data from the nivo trial and BSC registry.
Parametric survival curves, selected based on statistical fit and clinical plausibility, were fitted to these datasets for each comparator. For the ITC, AUC was calculated from the parametric curves without further adjustment. Confidence intervals (CIs) of incremental survival were generated via bootstrapping. The MAIC (conducted based on best practice guidelines; Phillippo DM et al, NICE DSU Technical Support Document 18 2016) used the same approach as the ITC, but results were based on patient-level data weighted to account for differences in patient characteristics across treatment cohorts and potential impact on survival outcomes. Results were generated by estimating outcomes for nivo and BSC for a hypothetical population with baseline characteristics matching those reported for the BV trial population.
Results: In total, 63, 102, and 110 patients were identified for nivo, BV, and BSC, respectively. For nivo, median follow-up was 19.1 months; median progression-free survival (PFS) by independent radiology review committee was 18.3 months (Figure). Median PFS by investigator and median overall survival (OS) were not reached. For BV, after 5 years of follow-up, observed median PFS was 9.3 months and median OS was 40.5 months. For BSC, median PFS was 7.2 months and median OS was 30 months. Under conservative assumptions of long-term survival with nivo relative to BV, nivo was associated with an estimated OS difference of +22 months (95% CI -39, 92). Relative to BSC, nivo was associated with an estimated OS difference of +40 months (95% CI -18, 108). Across less conservative sensitivity analyses for long-term survival with nivo, greater survival benefits were estimated. Estimates of relative survival remained consistent when based on the MAIC.
Conclusions: Estimates of relative survival with nivo compared with BV or BSC across a range of plausible scenarios indicate a survival benefit associated with nivo in patients with R/R cHL in whom auto-HCT had failed. While clinical trial data for nivo are still immature, the durable responses observed thus far and their potential impact on long-term outcomes demonstrate clinical value for these patients.
Study support: Bristol-Myers Squibb (BMS). Medical writing: K Tran, Caudex, funded by BMS.
Lozano-Ortega:Broadstreet HEOR: Employment; Bristol-Myers Squibb: Other: Contracted Broadstreet to conduct the work being presented. Rogula:Broadstreet HEOR: Employment; Bristol-Myers Squibb: Other: Contracted Broadstreet to conduct the work being presented. Johnston:Broadstreet HEOR: Employment; Bristol-Myers Squibb: Other: Broadstreet HEOR was contracted by Bristol-Myers Squibb to conduct the described work. Villeneuve:Bristol-Myers Squibb: Employment. Moshyk:Bristol-Myers Squibb: Employment, Other: company stock ownership. O'Brien:Bristol-Myers Squibb: Employment. Chen:Bristol-Myers Squibb: Employment. Connors:Amgen: Research Funding; Bayer Healthcare: Research Funding; F Hoffmann-La Roche: Research Funding; Bristol Myers-Squibb: Research Funding; Seattle Genetics: Honoraria, Research Funding; Janssen: Research Funding; Cephalon: Research Funding; NanoString Technologies: Patents & Royalties: Named Inventor on a patent licensed to NanoString Technologies, Research Funding; Lilly: Research Funding; Merck: Research Funding; Roche Canada: Research Funding; Genentech: Research Funding; Takeda: Research Funding.
