Gene-Environment Interactions and Depression

To the Editor: In their meta-analysis, Dr Risch and colleagues¹ concluded that the study of gene-environment interaction in mental disorders should await the identification of “robust marginal gene associations.” We believe that this conclusion extends well beyond the data, and an alternate explanation of their findings suggests other courses of action. The absence of replicable findings across studies that assessed both direct genotype-depression associations and gene-environment interactions may be explained by mismeasurement and undermeasurement of relevant environmental contexts.

Two compelling strands of evidence support this hypothesis. First, there is a sharp contrast in the consistency of success in studies that have sought genotype-phenotype associations in animals and in humans. For example, animal models of depression and anxiety disorders have consistently demonstrated genotype-phenotype associations.² By contrast, a recent genome-wide association study (GWAS) of depression found no significant associations.³ One central difference between these 2 research approaches lies in control over potentially relevant environmental exposures. These exposures are effectively randomized in animal models, but such control is absent from observational human gene-hunting studies.

Second, outside the gene-environment literature reviewed by Risch et al,¹ the evidence for environmental modification of genetic effects on human behavior is robust and increasing. The heritability of many phenotypes is modified by environmental factors such as socioeconomic status.⁴ Genotype-phenotype associations are also modified by context familiarity in animal models² and features of social environments in human studies.⁵ Unmeasured aspects of environmental context could contribute to nonreplication of gene-environment findings that at best limit the measurement of environment to life events.

Rather than conducting less research on how genotype and a range of environmental factors jointly produce mental disorders, what is needed is more and better-quality research. Unfortunately, to date GWAS of mental disorders have exclusively tested for genetic main effects, and gene-environment interaction studies have focused on candidate genes and individual-level measures of environmental exposures. Genome-wide association studies of mental disorders may produce more robust findings if populations were sampled conditional on exposure to a range of plausible environmental risk factors. Gene-environment interaction studies would benefit from moving away from focus on single candidate genes and toward considering multiple levels of environmental exposures.

Studies that integrate state-of-the-science methods of measurement of both genetic and environmental factors will provide a more comprehensive test of the role of gene-environment interaction in mental disorders than a meta-analysis of a single gene–environmental risk factor disorder association.