Validation of Prognostic Marker Tests: Statistical Lessons Learned From Regulatory Experience

Area of Science:

• Biomarkers and diagnostics
• Clinical trial design
• Statistical methodology

Background:

• Prognostic biomarkers are vital for personalized medicine but face challenges in clinical validation and adoption.
• Few prognostic marker tests achieve widespread clinical acceptance or health plan coverage.

Purpose of the Study:

• To identify key statistical considerations for designing, conducting, and analyzing clinical validation studies of prognostic tests.
• To provide guidance based on the experiences of US Food and Drug Administration-approved prognostic marker tests.

Main Methods:

• Review of pivotal clinical studies for 6 US Food and Drug Administration-cleared prognostic marker tests.
• Analysis of diverse study designs, statistical methods, and validation approaches.
• Emphasis on distinguishing prognostic biomarkers from the tests measuring them.

Main Results:

• Pivotal studies for approved prognostic marker tests utilized varied designs and statistical analyses.
• Key themes identified include prospective vs. retrospective designs, sample size determination, performance evaluation, and missing data handling.
• The importance of test measurement validation is highlighted.

Conclusions:

• Successful clinical validation of prognostic tests requires careful statistical planning and execution.
• Clear guidelines are needed to improve the design and analysis of studies supporting prognostic test development and adoption.
• Distinguishing biomarker validation from test measurement validation is critical for reliable prognostic tools.