PS09: Challenges in Multi-Sequence Trial Design and Analysis in Oncology

Oncology drug development has been increasingly reliant on innovative study designs that increase trial efficiency and commercial success, while maintaining regulatory acceptability. This session navigates the challenges faced in sequenced multi-treatment trial designs and the associated regulatory implication. Oncology trials often involve multiple sequential treatments. In hematologic malignancies especially, trials typically involve several treatment phases in which the decision to move to a subsequent treatment stage depends on the outcomes from the previous stage. In early disease setting for certain solid tumors, trials often involve neoadjuvant treatment, surgery followed by adjuvant treatment that is contingent on the degree of resection at surgery. The time-to-event efficacy outcome measures of interest in these trials may range from overall survival to progression-free survival, or event free survival. Interpretation of some of these endpoints often become challenging due to confounding factors and intercurrent events. Addressing these challenges require carefully planned trial designs (e.g., re-randomized designs, multi-arm designs) and appropriate statistical methods that account for attrition, intercurrent events, and variability in treatment sequences. When adding a new agent to an existing standard of care, it may not be inherently clear whether the optimal strategy is concurrent administration during induction/consolidation, sequential use as maintenance, or both. To address multiple therapeutic objectives within a single trial-such as evaluating benefit during induction and consolidation as well as during maintenance-one could adopt a two-part re-randomization trial design, where patients are first randomized to induction treatment arms, and then proceed through transplant and consolidation. Eligible patients (e.g., responders) are then re-randomized to maintenance treatment options. This design requires precise enrollment strategies to account for the second randomization and specific analytic methods to obtain reliable estimates of treatment benefit while accounting for attrition. Alternatively, a multi-arm design that randomizes patients to different preset treatment combinations from the start, may allow for direct arm-to-arm comparisons that isolate the contribution of individual treatment phases, but may be operationally complex. Although this approach simplifies assessment of phase-specific effects, it may introduce interpretational and statistical challenges when comparing closely similar treatment sequences. This session brings together experts from industry, academia, and regulatory agencies to discuss practical applications of innovative designs, appropriate endpoints and analytical methods in the evolving world of oncology drug development. The session will start with three presentations (each 15 min), followed by a 30-minutes panel discussion. Reference 1. Lokhnygina, Y., & Helterbrand, J. D. (2007). Cox regression methods for two-stage randomization designs. Biometrics, 63(2), 422-428. 2. Boris Freidlin and others, Design Issues in Randomized Clinical Trials of Maintenance Therapies, JNCI: Journal of the National Cancer Institute, Volume 107, Issue 11, November 2015, djv225 https://doi.org/10.1093/jnci/djv225