Quantifying direct and indirect effect for longitudinal mediator and survival outcome using joint modeling approach

Area of Science:

• Biostatistics
• Translational Research
• Epidemiology

Background:

• Longitudinal biomarkers are crucial for monitoring disease progression in biomedical research.
• Existing methods for jointly modeling longitudinal and survival data lack rigorous causal investigation.
• Understanding the causal role of biomarkers in treatment effects is essential for mechanism elucidation.

Purpose of the Study:

• To propose a causal mediation analysis method for longitudinal biomarkers and survival outcomes.
• To quantify the direct and indirect effects of a treatment mediated by a longitudinal biomarker.
• To relax the restrictive "sequential ignorability" assumption in mediation analysis.

Main Methods:

• Utilizing a joint modeling approach to link longitudinal biomarker data with survival endpoints.
• Implementing causal mediation analysis to estimate direct and indirect treatment effects.
• Applying the method to real-world case studies in AIDS and liver cirrhosis research.

Main Results:

• The proposed method successfully computes direct and indirect causal effects.
• Demonstrated the utility of the joint modeling approach in relaxing standard mediation assumptions.
• Provided a framework for evaluating the causal role of longitudinally measured biomarkers.

Conclusions:

• The developed causal mediation analysis offers a robust approach for understanding biomarker roles in disease progression.
• Joint modeling provides a flexible framework for analyzing complex longitudinal and survival data.
• This method enhances the evaluation of biomarkers in clinical and translational research.