Candidate-gene association studies with pedigree data: controlling for environmental covariates

Area of Science:

• Epidemiology
• Statistical Genetics
• Biostatistics

Background:

• Case-control studies are vital for evaluating candidate genes.
• Multiplex case-control (MCC) designs compare allele frequencies between related cases from multiplex families and unrelated controls.
• Existing statistical methods must account for within-family genotype correlations and potential confounders.

Purpose of the Study:

• To develop and evaluate a robust statistical method for analyzing multiplex case-control (MCC) data.
• To extend a previously developed model-based approach to simultaneously control for covariates.
• To assess the validity and power of proposed score tests under various ascertainment schemes.

Main Methods:

• Utilized a model-based approach using score statistics derived from a composite likelihood.
• Proposed three distinct methods for estimating the variance of the score statistic.
• Conducted simulations to test score tests under random and nonrandom ascertainment schemes.
• Applied methods to prostate cancer data.

Main Results:

• The robust score statistic demonstrated accurate type I error rates across all simulated scenarios.
• Alternative variance estimation methods showed inflated type I error rates under nonrandom ascertainment.
• Generalized estimating equations (GEE) exhibited high failure rates (5% to over 80%) in simulations.
• The robust method proved reliable where GEE failed.

Conclusions:

• The proposed robust score statistic is a valid and powerful method for analyzing multiplex case-control data.
• This robust method maintains accurate error rates even under nonrandom ascertainment, unlike other approaches.
• The developed robust method may be the only currently available regression analysis technique for MCC data, especially when GEE fails.