International journal of technology assessment in health care · 2026
Restricted mean survival time ratio in oncology cost-effectiveness analysis: stability and data-driven support for treatment-effect waning modeling and survival extrapolation.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Zhao Mingye, Shao Hanqiao, Shao Taihang, Jiang Yunong, Jiang Yunlin, Li Yuan, Fang Hongshu, Tang Wenxi
Аннотация
OBJECTIVES: We examined long-term treatment-effect patterns relevant to treatment effect waning (TEW) modeling in oncology health technology assessment (HTA), compared the dispersion and update stability of hazard ratios (HRs) and restricted mean survival time ratios (RMSTRs), and developed prediction models to support extrapolation model selection and TEW modeling.
METHODS: Using the FDA Hematology/Oncology Approvals Database, we identified phase II/III randomized controlled trials (2006-2025) with updates. We included overall survival (OS) and other time-to-event end points (non-OS). We conducted 2/3/4/5-year landmark analyses to estimate post-landmark HRs and RMSTRs, summarized by event-weighted medians and dispersion (coefficient of quartile relative dispersion, CQR). Update-related change was quantified using updated-to-initial ratios. We predicted longer-term RMSTR using elastic-net regression with nested cross-validation and split-conformal 90 percent prediction intervals.
RESULTS: We analyzed 168 trials (270 contrasts; 150 OS, 120 non-OS). Across landmark analyses at 2, 3, 4, and 5 years, RMSTR was consistently less dispersed than HR for both OS and non-OS end points, while post-landmark HRs generally remained < 1. Update ratios were more stable for RMSTR than HR. Initial and updated RMSTR were strongly correlated (OS r = .87; non-OS r = .82). Parsimonious regularized models accurately predicted updated RMSTR, with conformal prediction intervals achieving 91 percent and 96 percent coverage for OS and non-OS.
CONCLUSIONS: Reliance on current HR-based TEW modeling practices alone may be insufficient to fully characterize long-term treatment-effect uncertainty. RMSTR was consistently less dispersed than HR and predictable from early trial data, providing an anchor for extrapolation model selection and TEW uncertainty analysis in oncology economic evaluations.
METHODS: Using the FDA Hematology/Oncology Approvals Database, we identified phase II/III randomized controlled trials (2006-2025) with updates. We included overall survival (OS) and other time-to-event end points (non-OS). We conducted 2/3/4/5-year landmark analyses to estimate post-landmark HRs and RMSTRs, summarized by event-weighted medians and dispersion (coefficient of quartile relative dispersion, CQR). Update-related change was quantified using updated-to-initial ratios. We predicted longer-term RMSTR using elastic-net regression with nested cross-validation and split-conformal 90 percent prediction intervals.
RESULTS: We analyzed 168 trials (270 contrasts; 150 OS, 120 non-OS). Across landmark analyses at 2, 3, 4, and 5 years, RMSTR was consistently less dispersed than HR for both OS and non-OS end points, while post-landmark HRs generally remained < 1. Update ratios were more stable for RMSTR than HR. Initial and updated RMSTR were strongly correlated (OS r = .87; non-OS r = .82). Parsimonious regularized models accurately predicted updated RMSTR, with conformal prediction intervals achieving 91 percent and 96 percent coverage for OS and non-OS.
CONCLUSIONS: Reliance on current HR-based TEW modeling practices alone may be insufficient to fully characterize long-term treatment-effect uncertainty. RMSTR was consistently less dispersed than HR and predictable from early trial data, providing an anchor for extrapolation model selection and TEW uncertainty analysis in oncology economic evaluations.