Investigating treatment-effect modification by a continuous covariate in IPD meta-analysis: an approach using...

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Translational Oncology

Background:

• Investigating treatment effect heterogeneity is crucial in clinical trials.
• Continuous predictors are often categorized, losing valuable information.
• Existing methods like Subpopulation Treatment Effect Pattern Plot (STEPP) and multivariable fractional polynomial interaction (MFPI) address this, with Meta-STEPP recently proposed for meta-analyses.

Purpose of the Study:

• To derive a treatment effect function (TEF) on the continuous covariate scale within individual studies.
• To conduct a meta-analysis of these continuous TEFs using pointwise averaging.
• To introduce MethProf-MA for improved reporting of data and analysis steps in individual patient data meta-analyses.

Main Methods:

• Utilized data from eight randomized controlled trials in breast cancer.
• Applied the novel metaTEF method to analyze treatment heterogeneity.
• Employed pointwise averaging for meta-analysis of continuous TEFs.

Main Results:

• Demonstrated clear evidence of an interaction between estrogen receptors and chemotherapy in breast cancer patients.
• Results were robust irrespective of fractional polynomial function choice or random/fixed effect models.
• Acknowledged considerable differences across studies in patient populations and trial characteristics.

Conclusions:

• The metaTEF approach retains full information from continuous covariates, unlike cutpoint-based analyses.
• It effectively avoids critical issues in individual patient data meta-analyses of continuous effect modifiers.
• MetaTEF shows promise as an advanced method for analyzing treatment heterogeneity in randomized trials.