Determining cut-points for Alzheimer's disease biomarkers: statistical issues, methods and challenges

Area of Science:

• Neurology
• Biostatistics
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) diagnosis increasingly relies on biomarkers.
• Biomarkers are often continuous but require dichotomization for clinical criteria.
• Selecting an appropriate cut-point is essential for operationalizing diagnostic criteria.

Purpose of the Study:

• To review statistical principles for selecting biomarker cut-points in AD diagnosis.
• To describe common statistical methods for estimating cut-points.
• To highlight potential pitfalls and the optimal sense of estimated cut-points.

Main Methods:

• Review of statistical principles for cut-point selection.
• Description of commonly adopted statistical approaches.
• Analysis of potential pitfalls and optimality of methods.

Main Results:

• Continuous biomarkers necessitate dichotomization using cut-points for AD diagnosis.
• Various statistical methods exist for estimating cut-points, each with strengths and weaknesses.
• The choice of cut-point must align with the intended use and consequences of test results.

Conclusions:

• Statistical rigor in cut-point selection is vital for reliable Alzheimer's disease biomarker interpretation.
• Understanding different statistical approaches and their limitations is key.
• The clinical utility of a dichotomized biomarker depends on the cut-point's relevance to subsequent actions.