高リスク骨髄異形成症候群における後期段階試験の失敗からの学習：適応型およびバイオマーカー強化型デザインに向けて

Higher-risk myelodysplastic syndromes (HR-MDS) remain an area of high unmet need in which multiple randomized late-stage trials have failed to improve outcomes beyond hypomethylating agent monotherapy, despite encouraging early-phase signals.

Key randomized late-stage HR-MDS trials that failed to meet primary endpoints are evaluated for recurring design and interpretive challenges. Key themes include endpoint limitations, optimistic effect-size assumptions, underpowering for modest but clinically meaningful hazard ratios, and signal dilution from biological and operational heterogeneity. We discuss adaptive methodologies applicable to HR-MDS to improve trial efficiency and preserve statistical integrity. Regulatory perspectives including contemporary guidance and emerging international principles for confirmatory adaptive trials are considered. Randomized Phase 2/3 HR‑MDS trials and relevant literature were identified through searches of PubMed, ClinicalTrials.gov, and major conference proceedings (2010-2025).

Many new HR-MDS therapies are likely to deliver incremental survival gains rather than large overall survival improvements, particularly given an apparent survival plateau around 2 years. Future confirmatory programs should, therefore, plan for incremental improvements, prespecify handling of transplantation and post-protocol therapies, and integrate molecular stratification/enrichment to reduce variance. Adaptive designs should be used to stop futile programs earlier and to expand adequately powered trials when emerging data support clinically meaningful benefit.