Estimating the probability for major gene Alzheimer disease

Area of Science:

• Neurogenetics
• Computational Biology
• Epidemiology

Background:

• Alzheimer disease (AD) prediction in families is complex due to late onset, diagnostic issues, and limited data.
• Identifying genetic forms of AD (MGAD) is crucial for understanding disease etiology.
• Existing methods face challenges with censoring bias and familial heterogeneity.

Purpose of the Study:

• To develop and evaluate a Bayesian probability method for ranking families with potential major gene Alzheimer disease (MGAD).
• To compare the performance of the Bayesian approach against a maximum-likelihood method.
• To assess the utility of these methods in genetic and epidemiological studies of AD.

Main Methods:

• Developed a Bayesian probability model to compute a continuous variable for ranking AD families.
• Incorporated sex- and age-adjusted risk estimates, accounting for phenocopies and age at onset clustering.
• Compared the Bayesian method with a maximum-likelihood approach using Spearman rank correlation.

Main Results:

• The Bayesian method effectively ranks families for major gene Alzheimer disease (MGAD).
• High agreement (Spearman rank [r] = .92) was observed between the Bayesian and maximum-likelihood approaches.
• Numerical outcomes are sensitive to population gene frequency and disease incidence assumptions.

Conclusions:

• The developed Bayesian method provides a robust tool for identifying families with potential MGAD.
• Both Bayesian and maximum-likelihood methods show high concordance in ranking families.
• Caution is advised for direct risk counseling due to sensitivity to population parameters, but applications in research are numerous.