Accommodating missingness in environmental measurements in gene-environment interaction analysis

Area of Science:

• Genetics
• Environmental Health
• Biostatistics

Background:

• Complex diseases involve genetic (G) and environmental (E) factors, with gene-environment (G-E) interactions playing a crucial role in prognosis.
• Existing G-E interaction analysis methods often assume complete data, which is unrealistic as missing environmental measurements are common.
• Failure to account for missing environmental data can lead to biased results and incorrect identification of disease markers.

Purpose of the Study:

• To develop a robust statistical method for gene-environment interaction analysis in disease prognosis that can handle missing environmental data.
• To improve the accuracy of identifying significant genetic, environmental, and gene-environment interaction effects.

Main Methods:

• Utilized an accelerated failure time (AFT) model for prognosis analysis.
• Employed a nonparametric kernel-based data augmentation approach to accommodate missing environmental measurements.
• Implemented a penalization strategy that respects the main effects and interactions hierarchy for selection and regularized estimation.

Main Results:

• The proposed method demonstrated robustness and sound statistical interpretation.
• Simulation studies showed superior performance compared to existing alternative methods.
• Analysis of TCGA lung cancer and melanoma data yielded significant findings and models with enhanced predictive power.

Conclusions:

• The developed approach effectively handles missing environmental data in G-E interaction analysis.
• This method offers a reliable tool for improving disease prognosis by accurately identifying key genetic and environmental influences.
• The findings suggest potential for improved clinical prediction models in complex diseases.