Carefully thinking about telomeres

We were initially very enthusiastic about the recent report by Mitchell et al. (1). We were particularly enthusiastic, given the consistency of their results with a series of previous studies (2–4). Although certainly there are significant strengths to this report, we feel it critical to raise several concerns.

First, previous studies have used DNA from saliva for telomere length (TL) measurement (2). Second, although the authors state that their focus on African American children is unique, published studies that focused predominately on African American youth do exist (2, 3). Our most significant concern relates to the genetic interaction analyses. Mitchell et al. (1) state that there are no differences in allele frequencies between the two extreme groups; however, a quick analysis of the distribution of genotypes from supplemental table S1 in ref. 1 indicates that significant differences in genotype frequencies do, in fact, exist for certain alleles (e.g., the 5httlpr genotype χ² = 12.54, P = 0.002; TPH2A χ² = 15.70, P = 0.004). Mitchell et al. also do not note whether their allele frequencies, for individual genes, reflect the expected population frequencies for each polymorphism within African Americans, nor whether their genotype results are consistent with the Hardy–Weinberg Equilibrium. The potential of these factors to confound their results exists and are currently untested. Relative to their interaction with the negative environment, it does not appear that the dopaminergic interaction reached statistical significance (P = 0.098). In examining the graphs, it appears that this interaction is also perhaps less suggestive of differential susceptibility than it is of a diathesis stress model. Although the serotonergic interaction presented in figure 1 of ref. 1 suggests differential susceptibility, Mitchell et al. fail to test this, although papers exist outlining an appropriate analytic approach (e.g., ref. 5).

In short, although the conceptual framework for the report by Mitchell et al. (1) has significant merit, and the consideration of a genetic interaction with TL is intriguing, the methodological issues are concerning. The relationship between TL, early life stress, and health outcomes across the life course is a significant one, with substantial implications for prevention, intervention, and public policy. Future studies can build upon this work and carefully examine tissue-specific differences in TL, across development and racial groups, as well as thoughtfully construct biologically informed genosets as potential moderators of the relationship between early life adversity, telomere dynamics, and negative health trajectories.