PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 31, 2026Pharmaceutical Statistics0 citations

A Constrained Hierarchical Bayesian Model Considering Latent Biomarker Subgroups for Time‐To‐Event Endpoints in Randomized Phase II Trials

View Full Paper
YHYifei HuangKTKentaro TakedaYZYongyun Zhao

Key Points

  • This research aims to improve time-to-event analysis in Phase II oncology trials by utilizing a Bayesian model that accounts for biomarker subgroups.
  • Developed a constrained hierarchical Bayesian model (CHBM-LS) for time-to-event endpoints.
  • Aggregated biomarker populations into latent subgroups to account for treatment effect heterogeneity.
  • Compared CHBM-LS with other existing trial designs to evaluate its performance.
  • Demonstrated improved accuracy of hazard ratio estimates using CHBM-LS.
  • Increased power to detect true treatment effects in Phase II trials.
  • Maintained control over the Type I error rate compared to conventional methods.

Abstract

ABSTRACT In randomized Phase III oncology trials, the long‐term time‐to‐event endpoint is the most relevant outcome for participants and regulators. However, in Phase II trials, the short‐term binary outcome of tumor response is often used as a surrogate endpoint to evaluate the treatment benefit. This may lead to a high failure rate in Phase III trials, as the tumor response may not reflect the actual survival benefit. Moreover, many oncology trials collect biomarker data, especially those that may predict clinical outcomes and identify participants who are more likely to respond to the experimental treatment. Therefore, there is a growing need for a biomarker‐based design to enrich the trial by selecting participants whose biomarker levels exceed certain thresholds. This paper proposes a constrained hierarchical Bayesian model that considers latent biomarker subgroups (CHBM‐LS) for long‐term time‐to‐event endpoints in Phase II randomized trials. CHBM‐LS aggregates the biomarker populations into latent subgroups and accounts for the heterogeneity of treatment effects across biomarker levels in each model. We compare our proposed design with other approaches and show the benefits of CHBM‐LS in improving the accuracy of hazard ratio estimates and increasing the power to detect true effects while maintaining control over the Type I error rate.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Huang et al. (2026) studied this question.

synapsesocial.com/papers/69cb6526e6a8c024954b92f3https://doi.org/10.1002/pst.70085
Ask AI
Helpful
Bookmark
Share
View Full Paper