LIFE-COURSE LONGITUDINAL STUDIES ARE NEEDED TO ADVANCE INTEGRATION OF GENOMICS AND SOCIAL EPIDEMIOLOGY

Genome science offers new opportunities for social epidemiology, including opportunities to observe molecular states and processes hypothesized to mediate effects of social-environmental exposures on morbidity, disability, and mortality (1). With these opportunities come several challenges, some new and some old. Among the new challenges are the very large volume of measurements generated by whole-genome assays, difficulties assigning biological meaning to genomic measurements, and new potential confounders, including the leukocyte composition of blood samples from which genomic measurements are made. The recent article by Levine et al. (2) provides an example of how studies may navigate these new challenges in the context of whole-genome gene-expression data. Among the old challenges is one rightfully highlighted in the recent letter by Huang and Kaufman (3), having to do with causal inference.

Our commentary suggested that the gene-expression signature studied by Levine et al., the conserved transcriptional response to adversity (CTRA), might inform understanding of biological processes through which social gradients in health come about. The CTRA comprises down-regulation of viral defenses and up-regulation of inflammation—processes that, if chronically activated, would be health damaging (4). As Huang and Kaufman (3) point out, to establish a mediating role for the CTRA in social gradients in health, life-course longitudinal studies are needed that include repeated measures of socioeconomic circumstances, gene expression, and health outcomes. We agree that this longitudinal design is essential for refining causal inference about how early-life risk factors shape outcomes in aging (5). The observations in the study by Levine et al. are valuable because they contribute evidence suggesting that the CTRA should be among the measurements investigated in life-course studies to understand biological mediators of social gradients in health.

A second issue to address is the language used to describe biology mediating social gradients in health. Huang and Kaufman’s letter uses the words “embodiment” and “programming,” which did not appear in our commentary and which suggest some confusion about our meaning. If the CTRA is a mediator of social gradients in health, the process could involve damage to organ systems arising from chronic activation of the CTRA by social-disadvantage–related stressors. Such a process does not sound like “programming” as it is often discussed in the epigenetics literature. The potential for early-life experiences to modify the epigenome in ways that affect disease risk in aging has received much attention, especially in the media. But epidemiologic observations of such phenomena in humans remain few in number. In fact, the genomic imprint of social adversity that is most readily detectable using current methods turns out to be the signature of an already well-known health risk that itself shows a strong social gradient, tobacco smoking (6, 7). The CTRA is of interest in understanding social gradients in health not because it reflects a process in which social stress causes enduring changes to the genome but because it may be part of the process through which social stress causes enduring changes to health.

Research is still needed to establish whether chronic CTRA activation causes increased morbidity and mortality. Animal studies that experimentally control chronic social stress exposure and CTRA activation can help. But there is no substitute for human longitudinal studies with repeated observations of CTRA-activating exposures, the CTRA itself, and health outcomes. Falling costs of genome sequencing are making it possible to conduct longitudinal, repeated-measures studies of the CTRA. Such studies offer value for funders and researchers interested in advancing integration of genomics and social epidemiology.

Acknowledgments

Conflict of interest: none declared.