Evaluating Causal Effects on Time-to-Event Outcomes in an RCT in Oncology With Treatment Discontinuation

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Causal Inference

Background:

• Premature treatment discontinuation in clinical trials breaks randomization and complicates intention-to-treat effect estimation.
• Intercurrent events, such as treatment discontinuation due to adverse events, require specific handling strategies as per ICH E9(R1) guidelines.

Purpose of the Study:

• To propose and implement a principal stratum strategy for analyzing survival outcomes in the presence of continuous-time treatment discontinuation.
• To decompose the overall intention-to-treat effect into principal causal effects based on potential patient discontinuation behavior.

Main Methods:

• A principal stratum strategy is employed to handle time-to-event intermediate variables and censoring.
• A flexible model-based Bayesian approach is utilized for analysis, providing interpretable results.
• The framework is applied to synthetic data from an oncology trial and validated with a simulation study.

Main Results:

• The proposed strategy effectively handles time-to-event outcomes with treatment discontinuation and censoring.
• Covariates enhance the credibility of assumptions, improve inference precision, and characterize discontinuation behavior.
• The Bayesian principal stratification framework yields interpretable results for complex clinical trial scenarios.

Conclusions:

• The principal stratum strategy offers a robust method for analyzing causal effects in randomized controlled trials with treatment discontinuation.
• This approach provides valuable insights into patient subgroups defined by their response to treatment and discontinuation patterns.
• The findings can inform clinical practice and the design of future clinical trial protocols.