Case-only gene-environment interaction studies: when does association imply mechanistic interaction?

Area of Science:

• Epidemiology
• Genetic Epidemiology
• Statistical Genetics

Background:

• Case-only studies are frequently employed to detect gene-environment interactions, assuming independence between genetic and environmental factors in the general population.
• Statistical associations observed between genetic and environmental factors within a case group do not automatically confirm a mechanistic gene-environment interaction.

Purpose of the Study:

• To investigate the conditions under which statistical associations observed in case-only studies accurately reflect mechanistic gene-environment interactions.
• To differentiate between spurious associations and true mechanistic interactions using a sufficient cause framework.

Main Methods:

• Application of the sufficient cause framework to model gene-environment interactions.
• Analysis of statistical associations among cases under various assumptions regarding factor independence and preventive effects.
• Extension of methods to scenarios with negative associations between genetic and environmental factors.

Main Results:

• Observed associations among cases can occur without mechanistic gene-environment interaction.
• A positive association among cases implies mechanistic interaction if neither factor can prevent the outcome.
• A multiplicative interaction exceeding two is required to infer mechanistic interaction without the preventive effect assumption.

Conclusions:

• The interpretation of case-only study findings requires careful consideration of underlying assumptions.
• The sufficient cause framework provides a robust method for assessing mechanistic gene-environment interactions.
• The developed methods are applicable even when genetic and environmental factors exhibit population-level dependencies.