Variance-component-based meta-analysis of gene-environment interactions for rare variants

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Complex diseases arise from gene-environment interactions (G×E).
• Current G×E tests for rare variants in single studies have limited statistical power due to sample size constraints.
• Meta-analysis methods for synthesizing G×E effects of rare variants across studies are underdeveloped.

Purpose of the Study:

• To propose novel meta-analysis methods for testing G×E effects of rare variants.
• To address limitations in existing methods by considering homogeneous or heterogeneous effects and fixed/random genetic effects across studies.

Main Methods:

• Developed four meta-analysis methods: HOM-INT-FIX, HET-INT-FIX, HOM-INT-RAN, and HET-INT-RAN, based on variance component models.
• Simulated data to evaluate the performance of the proposed methods under the null hypothesis.
• Compared the statistical power of the new methods against pooled analysis using individual-level data.

Main Results:

• The proposed methods demonstrated accurate null distributions, aligning empirical results with theoretical expectations.
• HOM-INT-FIX and HOM-INT-RAN achieved power comparable to pooled analysis for homogeneous G×E effects.
• HET-INT-FIX and HET-INT-RAN provided superior power compared to pooled analysis for heterogeneous G×E effects.
• Validated the methods using UK Biobank whole-exome sequencing data to test 12 candidate gene-age interactions for blood pressure traits.

Conclusions:

• The developed meta-analysis methods effectively test G×E effects for rare variants.
• These methods offer increased statistical power, particularly in scenarios with heterogeneous interaction effects across studies.
• The approach advances the ability to identify genetic and environmental contributions to complex diseases.