Using Individualized Treatment Effects to Assess Treatment Effect Heterogeneity

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Pharmacovigilance

Background:

• Assessing treatment effect heterogeneity (TEH) is vital for drug development and personalized medicine.
• Understanding patient variability in treatment response informs clinical decisions.
• Existing methods may not fully capture individualized treatment effects.

Purpose of the Study:

• To introduce novel methodologies for assessing treatment effect heterogeneity (TEH) based on individualized treatment effects.
• To develop tools for global heterogeneity testing, covariate effect modification ranking, and individualized treatment effect estimation.
• To integrate these methods into a robust framework for clinical trial analysis.

Main Methods:

• Utilized a doubly robust (DR) learner to infer a pseudo-outcome reflecting causal contrast.
• Applied the pseudo-outcome for global heterogeneity testing, covariate effect modification analysis, and individualized treatment effect estimation.
• Compared the DR-learner with alternative methods in simulations and a pooled analysis of psoriatic arthritis (PsA) trials.

Main Results:

• The DR-learner demonstrated robust performance in estimating individualized treatment effects and assessing heterogeneity.
• Simulation studies validated the proposed methods' effectiveness compared to competing approaches.
• Analysis of psoriatic arthritis (PsA) trials revealed significant insights into treatment effect heterogeneity.

Conclusions:

• The novel DR-learner-based methodologies provide a robust framework for assessing treatment effect heterogeneity (TEH).
• These methods enhance decision-making in drug development and facilitate personalized medicine strategies.
• Integration with the WATCH workflow offers comprehensive TEH analysis for clinical trial sponsors.