More efficient and inclusive time-to-event trials with covariate adjustment: a simulation study

Area of Science:

• Clinical Trial Design
• Biostatistics
• Survival Analysis

Background:

• Covariate adjustment is known to increase statistical power in randomized trials with continuous outcomes.
• Factors influencing power gains from covariate adjustment in time-to-event trials are less understood.
• Broadening eligibility criteria often decreases statistical power in clinical trials.

Purpose of the Study:

• To investigate factors influencing statistical power and sample size requirements in time-to-event clinical trials.
• To quantify the sample size reduction achievable through covariate adjustment.
• To assess the impact of covariate adjustment on maintaining power when broadening eligibility criteria.

Main Methods:

• Parametric simulations were employed to model time-to-event outcomes.
• Simulations utilized data from the Cancer Genome Atlas (TCGA) hepatocellular carcinoma (HCC) cohort.
• The prognostic performance of covariates (C-index) and cumulative incidence were key simulation parameters.

Main Results:

• Sample size reduction from covariate adjustment increases with the covariate's prognostic performance (C-index) and the event's cumulative incidence.
• For a covariate with C-index=0.65, sample size reduction ranged from 3.1% (10% cumulative incidence) to 29.1% (90% cumulative incidence).
• Covariate adjustment maintained statistical power when broadening eligibility criteria, reducing screened patients by 2.4-fold in an HCC adjuvant trial simulation.

Conclusions:

• Systematic adjustment for prognostic covariates enhances the efficiency and inclusivity of clinical trials, particularly those with high cumulative incidence.
• Covariate adjustment is crucial for maintaining statistical power in trials with broad eligibility criteria, common in advanced cancer studies.
• The Cox-Snell R-squared provides a conservative estimate of sample size reduction benefits from covariate adjustment.