The case-only test for gene-environment interaction is not uniformly powerful: an empirical example

Area of Science:

• Genetics
• Epidemiology
• Statistical genomics

Background:

• Gene-environment (G × E) interactions are crucial for understanding complex diseases.
• The case-only test is a proposed powerful method for detecting G × E interactions.
• This test assumes independence between genetic and environmental factors, but violations can impact results.

Purpose of the Study:

• To compare the performance of the case-only test with other G × E interaction detection methods.
• To investigate the impact of gene-environment correlation on the power of the case-only test.
• To identify G × E interactions in a genome-wide association study (GWAS) of esophageal squamous-cell carcinoma (ESCC) in Chinese populations.

Main Methods:

• Comparison of statistical approaches including case-only test, standard logistic regression, two-step, empirical Bayes, and cocktail methods.
• Analysis of G × E interactions in a GWAS dataset of Chinese ESCC patients.
• Evaluation of the influence of gene-environment correlation on test performance.

Main Results:

• G × E interactions were identified in two regions (4q23 and 12q24) related to alcohol metabolism and drinking status.
• The case-only test showed high significance in the 4q23 region.
• The case-only test failed to detect significant interactions in the 12q24 region, likely due to strong gene-environment correlation, while other methods succeeded.

Conclusions:

• The case-only test's power can be significantly reduced by gene-environment correlation.
• Different methods for detecting G × E interactions have varying sensitivities to violations of independence assumptions.
• Employing multiple analytical approaches is essential for comprehensive and robust identification of G × E interactions.