ISHL10 Abstract P089

Brentuximab vedotin after autologous stem cell transplant (ASCT) yields the strongest benefit in Hodgkin lymphoma (HL) patients with ≥ 2 risk factors: results of a multivariate analysis (MVA)

Background: In the phase 3 AETHERA trial, PFS was significantly improved with brentuximab vedotin (BV) versus placebo (PB) (HR=0.57, P=0.001) in HL patients at high risk of progression post-ASCT [a]. In a previous MVA of the study population, consolidation treatment with BV was significantly associated with improved PFS compared with placebo after adjustment for clinical factors [b]. Here we present an MVA assessing interactions between established risk factors and treatment (BV versus PB) to identify patients who may benefit most from consolidation therapy. As eligibility was limited to high-risk patients, identification of risk factors for a general population was not explored. Methods: To identify qualitative treatment-subgroup interactions, Martingale residuals from a null Cox proportional hazard (CPH) model were used as a continuous outcome variable for a qualitative interaction tree (QUINT [c]); 18 covariates were used to define potential subgroups. Some subgroups were based on number of established risk factors present in a patient, including relapse <12 months or refractory to frontline therapy, best response of partial remission or stable disease to salvage therapy, extranodal disease at pre-ASCT relapse, B symptoms at pre-ASCT relapse, and ≥2 prior salvage therapies. As a sensitivity analysis, a multivariate CPH model with treatment-covariate interactions was developed to identify the most significant interaction (smallest p-value <0.05). This interaction was retained in the model, and a second-round analysis was performed to identify the next most significant interaction among the remaining covariates. Results: The most significant qualitative treatment-subgroup interaction identified in the QUINT model was in the subgroup of patients with 1 versus ≥2 risk factors, with greater treatment benefit for patients with ≥2 risk factors. The sensitivity analysis confirmed this outcome (P=0.0006) and, after adjusting for this interaction, found no other interaction among the covariates tested to be significant. Analysis details will be presented with updated PFS and safety data after ~4 years since last patient enrolled. Conclusions: Treatment with BV appears to have the strongest impact for patients with ≥2 risk factors. [a] Moskowitz CH et al. Lancet 2015. 385:1853. [b] Walewski J et al. J Clin Oncol 2015. 33(suppl; abstr 8519) [c] Dusseldorp E, Van Mechelen I. Stat Med 2014. 33:219.

Authors

  • C. Moskowitz
  • J. Sweetenham
  • A. Chen
  • E. Ayala
  • T. Masszi
  • J. Holowiecki
  • P. Stiff
  • A. Carella
  • S. Viviani
  • V. Bachanova
  • A. Sureda
  • D. Huebner
  • S.Y. Lee
  • N. Hunder
  • L. Thomas
  • M. Uttarwar
  • J. Walewski