A targeted simulation-extrapolation method for evaluating biomarkers based on new technologies in precision medicine.

Area of Science:

• Biostatistics
• Biomarker Discovery
• Precision Medicine

Background:

• Novel biomarkers are crucial for precision medicine, but technologies like next-generation sequencing have inherent misclassification rates.
• These misclassification rates pose challenges in applications such as liquid biopsy for early tumor detection.
• Current biomarker development is often limited to specific technologies and settings, hindering broader application planning.

Purpose of the Study:

• To develop a statistical method for projecting biomarker performance metrics under varying misclassification rates.
• To enable informed planning of biomarker development and clinical trials by simulating different technological or application scenarios.

Main Methods:

• An extended simulation extrapolation (SIMEX) approach was developed to project biomarker performance.
• Simulation studies were conducted using logistic regression and proportional hazards models.
• The method was validated using a lung cancer dataset with two gene panel biomarkers.

Main Results:

• The proposed SIMEX-based method accurately projects biomarker performance when switching between different technology or application settings.
• The method demonstrated good precision in simulations for logistic regression and proportional hazards models.
• Analysis of a lung cancer dataset confirmed the feasibility of inferring implications across various scenarios with limited data.

Conclusions:

• The extended SIMEX method provides a robust tool for projecting biomarker performance across diverse misclassification rates.
• This approach facilitates strategic planning for biomarker development and clinical trials in precision medicine.
• It allows researchers to assess the potential impact of different technologies and applications even with limited experimental data.