In a sample of adolescent girls and boys of all weight strata, depressive symptoms showed a small, but significant, association with individual variation in adolescents’ fasting insulin and insulin sensitivity even after accounting for the contribution of demographic and anthropometric factors. Although anxiety was associated with fasting insulin in bivariate analyses, anxiety symptoms did not significantly contribute to either model.
Our findings parallel a series of studies in the adult literature that have found links between depressive disorders or severity of depressive symptoms and decreased insulin sensitivity in women and men [3
]. Despite meal-level sex differences in depressive symptoms and insulin resistance, the observed relationship between depression and insulin was not significantly different among adolescent boys and girls. This pattern contrasts with one prior adult study that also explicitly tested for moderation by sex and found an effect only for women [9
]. Although further investigation into potential sex differences in adolescents is required, our results preliminarily suggest that depressive symptoms and insulin measures may be related similarly in adolescent boys as well as girls.
Although the present results are cross-sectional and cannot be construed to imply causation, a number of theoretical mechanisms could account for this link. One potential explanation is that depressive symptoms may promote lower insulin sensitivity, independent of their relationship with adiposity, via a number of physiological mechanisms including altered central nervous system insulin signaling, distress-induced up-regulation of humoral systems (e.g., glucocorticoids), and/or pro-inflammatory activation [15
]. Another possibility is that depressive symptoms contribute to higher fasting insulin and decreased insulin sensitivity through their impact on lifestyle factors such as physical inactivity [36
] or eating habits [5
]. Evidence in support of this hypothesis includes that adolescent depressive symptoms are associated with greater sedentary behavior [36
], which in turn may promote a deleterious metabolic phenotype [37
]. It is noteworthy that variations in depressive symptoms, even at subclinical levels, were associated with relatively poorer insulin action. Such a pattern is congruent with an existing body of depression research recognizing the detrimental impact of subthreshold adolescent depressive symptoms on health outcomes [39
Alternatively, it is also possible that insulin resistant states may lead to a depressive mood. Yet, we are aware of no studies finding that treatment of insulin resistance (unless in concert with weight reduction) alleviates depressive symptoms. In contrast, the hyperinsulinemia found among adults with full-syndrome depressive disorders has been shown to be ameliorated after treatment for depression [3
]. Importantly, replication of this finding among lean and obese adolescents with clinically elevated depressive symptoms is now required. If depressive symptoms contribute to decreased insulin sensitivity in adolescents, the prevention of adolescent depressive symptoms might have the potential to diminish a portion of the risk, at least to some extent, for type 2 diabetes.
Several shortcomings should be noted. The effects of psychological factors on insulin measures were small relative to the effects of anthropometric variables. In particular, fat mass was highly related to fasting insulin and insulin sensitivity. Therefore, even if depressive symptoms and insulin functioning prove to have a causal relationship, interventions that lower depressive symptoms without altering body composition may have a limited impact on insulin sensitivity and vice versa. Study of the links between symptoms and insulin measures is warranted in adolescent samples with clinically elevated symptomatology and may shed greater light on the magnitude of the depression-insulin relationship. Although the QUICKI is a well-validated, reliable estimate of insulin sensitivity that is highly correlated with clamp measurements [30
], replication of the observed effects using more sophisticated and precise measurements of insulin sensitivity in larger samples is needed. Other psychological symptoms such as externalizing behavior problems ought to be examined based on prior findings that hostility may be an important risk factor for the metabolic syndrome and cardiovascular disease in adults [9
] and youth [41
]. Another limitation is that we examined only self-reports of psychological symptoms; examination of the associations between parent reports of symptoms and insulin sensitivity should also be undertaken. Finally, in future work it will be important to study potential mechanisms such as physical inactivity [36
Strengths of the present study include the examination of psychological symptoms and insulin sensitivity in a sizeable sample of adolescents that included both non-overweight and overweight children and the use of reliable and valid measures for estimating these variables. Although adolescents were not recruited in a truly population-based fashion, participants were recruited for a study measuring plasma hormones pertaining to eating behavior, understood that they would receive no treatment as a result of participation, and had no advanced knowledge of the content of questionnaires or the interview. Thus, we believe that this sample is likely to be reasonably representative of the general population of adolescents and is unlikely to be enriched for youth who were, or were not, especially concerned about eating behavior. Moreover, unlike many prior studies of psychological influences on insulin sensitivity, we examined the associations with psychological symptoms after accounting for the contribution of significant covariates. In particular, controlling for body composition is critical because depressive symptoms have been found to be more common among overweight compared to non-overweight adolescents.
The current study is a preliminary step toward understanding links between psychopathology and insulin sensitivity among adolescents whose metabolic processes are still functioning relatively normally. Longitudinal studies will help elucidate whether these psychological factors are important predictors for, or consequences of, abnormal glucose homeostasis for adolescents.