Diabetes and intelligence? No, genes, intelligence and, then, diabetes

It is widely believed that having diabetes has a detrimental effect on one's cognitive abilities and, as a result, there is an established and vibrant research tradition, which aims to understand why this might be the case. Yet, from the many findings of cognitive epidemiology we know that some of the health conditions that are believed to impair cognitive functioning are, in fact, predicted by low pre-morbid cognitive ability. Such cases are often referred to as reverse causation. To find out if reverse causation also applies to the diabetes-cognitive ability association, we grouped the members of the Lothian Birth Cohort 1936 (LBC1936) based on whether or not they had diabetes at age 70 and compared mean intelligence test scores of the two groups at ages 11 and 70. Having diabetes already at age 11 was unlikely and therefore the age-11 intelligence scores of those with and without diabetes at older age were expected to be similar, unless reverse causation was in operation. The results showed that those with older-age diabetes scored lower in cognitive ability than those without diabetes roughly by the same amount at ages 11 and 70 (about d = 0.3). These results, showing that the cognitive ability differences between people with and without diabetes was in place long before the typical diabetes onset age, support the reverse causation explanation for the association (of course, there also remains the possibility of confounded association).
In a separate study, we investigated whether low cognitive ability measured in childhood interacted with genetic risk for type 2 diabetes in predicting diabetes diagnosis or a high glycated hemoglobin (HbA1C) level in older age (the cut-off of HbA1C ≥ 6.5% is currently used to diagnose diabetes). The rationale for this research question was two-fold. First, it is possible that the negative effect of low cognitive ability on diabetes is more likely to occur when the organism is already predisposed to develop the disease, whereas low genetic predisposition for the disease may provide people with immunity against the negative effect of low cognitive ability and its consequences. Second, it is possible that cognitive ability and its socioeconomic or life-style consequences may moderate the realization of genetic risk such that, for example, high ability may buffer high genetic risk.
We tested this hypothesis in LBC1936. Polygenetic risk for type 2 diabetes was estimated by applying the previously found meta-analytic associations of thousands of single-nucleotide polymorphisms (SNPs) with type 2 diabetes to each participant's genome. Several polygenetic risk scores were created by employing different combinations of SNPs based on the strength of their meta-analytic associations with diabetes. The risk scores did not interact significantly with childhood intelligence scores when predicting self-reported diabetes diagnosis, possibly due to a relatively low number of participants with the diagnosis. However, childhood cognitive ability and the polygenetic risk scores interacted significantly in the prediction of HbA1C levels. Specifically, the association between low cognitive ability and high HbA1C levels tended to be stronger for those at higher polygenetic risk and the genetic risk-HbA1C association was stronger for participants with low childhood cognitive ability.
These results, providing some evidence for gene by cognitive ability interaction in predicting diabetes symptoms, may advance our understanding of the behavioural trait-health associations and help to increase the accuracy of (early) genetic prediction of disease risk.