Insulin Resistance and Cognitive Impairment: Title and subTitle BreakA View Through the Prism of Epidemiology

The relationship between insulin resistance and neurodegenerative disease is tantalizing in its potential to offer an integrated architecture for aging of the body and the brain. While there are contradictory findings in this area, the prevalence of insulin resistance is reportedly increased in many age-related neurodegenerative disorders, including Alzheimer disease, vascular dementia, and Parkinson disease.¹ Insulin resistance may contribute to the expression of these disorders through shared as well as distinct mechanisms. Identification of these mechanisms could aid the development of effective preventative and therapeutic strategies.

In this issue, Geroldi et al² report an ambitious study that views the relationship between insulin resistance and cognitive impairment through the prism of epidemiology. While this approach can identify important trends in large populations, it requires careful delineation of criteria to define a sample and its subgroups. Each criterion may refract the investigators’ perspective, which may in turn modify the likelihood of obtaining a specific set of relationships. Geroldi et al describe their defining criteria with commendable clarity, thus enabling the reader to interpret the results in perspective. These authors also describe clearly their method of operationalizing difficult constructs including the retrospective identification of insulin resistance. The selection of metabolic measures that define this condition is controversial. Geroldi et al used criteria developed by the Adult Treatment Panel III³ (ATP III), which include hypertension, elevated triglycerides, low levels of high-density lipoprotein, large waist circumference, and high fasting glucose values. Despite their widespread use, these criteria may not accomplish their intended goal. Some studies suggest insufficient sensitivity of these guidelines for assessing insulin resistance.^{4 - 5} Against such “gold standards” as the hyperinsulinemic-euglycemic clamp, the ATP III criteria appear to have sensitivity of only 50%.^{6 - 7} This shortcoming is not surprising because these criteria were intended to identify adults at risk for cardiovascular disease, and not insulin resistance per se.⁴ The finding from the Geroldi et al study that the participant subgroup with subcortical vascular features has higher rates of ATP III criteria is therefore a logical extension of the known association between cardiovascular disease and cerebrovascular disease.⁸

Insulin resistance is nearly always accompanied by peripheral hyperinsulinemia, a metabolic condition with effects both related to and independent of impaired glucose uptake into muscle (the standard index of insulin resistance). Geroldi et al improve on earlier work by adding fasting hyperinsulinemia to the ATP III criteria for identification of insulin resistance. While this approach is preferable to consideration of only cardiovascular risk factors, the authors note that fasting insulin estimates may also be insensitive. For example, fasting insulin cutoffs at the highest tertile of a population identify insulin resistance in only about 66% of those meeting the “gold standard” for the syndrome.⁹ The present study is likely to have an even lower sensitivity, given that the authors used the highest insulin quintile as a cut-off point. Insulin measures from oral glucose tolerance testing (either 120-minute values or integrated area under the curve values) offer better detection of hyperinsulinemia,¹⁰ but may be impractical for use in large cohort studies. Improved methods for identification of hyperinsulinemia are clearly needed, especially with the growing understanding of its negative consequences.

The previous discussion is not intended to detract from the valuable contributions of the article by Geroldi et al, but rather to illustrate the many considerations needed to interpret its results and those of other retrospective studies designed originally to address other scientific questions. The authors’ conclusion that insulin resistance might contribute to cognitive impairment through a vascular mechanism is well justified. The authors make an important observation concerning the likely contributory effects of microvascular damage and attribute this damage to aspects of the insulin resistance syndrome such as dyslipidemia. It is notable, however, that direct effects of insulin on vascular function have been documented in both animal and human work.¹¹ For example, chronic hyperinsulinemia increases cytokine release and a related endothelial inflammatory response.¹¹ In this way, hyperinsulinemia might be a proximal cause of vascular dysfunction, rather than a correlate.

As the authors note, hyperinsulinemia has also been implicated in a variety of nonvascular mechanisms in the pathogenesis of Alzheimer disease and other dementias. Chronic peripheral hyperinsulinemia typically reduces insulin transport into the brain,¹² thereby potentially creating an insulin-deficient state that deprives the central nervous system of the beneficial effects of insulin. Such effects may include inhibition of τ phosphorylation,¹³ promotion of Aβ release from intracellular to extracellular compartments in which clearance can more readily occur,¹⁴ or expression of insulin degrading enzyme, a major protease responsible for Aβ clearance.¹⁵ Chronic peripheral hyperinsulinemia may also interfere with peripheral clearance of Aβ. Although the significance of plasma Aβ elevations that have been described in patients with Alzheimer disease is uncertain,¹⁶ it is possible that such elevations interfere with Aβ transport between brain to periphery, and increase the likelihood of central nervous system Aβ accumulation. Thus, there are several pathways through which insulin resistance can effect the risk for Alzheimer disease and other neurodegenerative disorders.

How then can we best study the complex relationship between insulin resistance and neurodegenerative disease, thereby extending the findings of Geroldi et al? Prospective epidemiologic studies using sensitive measures of insulin resistance and hyperinsulinemia would certainly be helpful. In the event that oral glucose tolerance testing is not practical for such studies, better standardization of insulin assays and age-adjusted criteria for abnormal levels may enhance the sensitivity of fasting indices. A less obvious requirement for advancement in this field is the need for an integrated, interdisciplinary research paradigm that reconciles the peripheral and central effects of insulin resistance. It may be that regardless of its etiology, insulin resistance constitutes a metabolic stressor that interacts with a preexisting neurobiological template to induce a given disorder. Yet, greater consideration should also be given to the etiologic heterogeneity of the insulin resistance. Different causes of the syndrome may vary in their implications for the risk of a given neurodegenerative disease. Endocrinologic research has made impressive progress in identifying and characterizing a number of genetic and environmental factors that promote peripheral insulin resistance, and we have much to learn from these studies. Considerably less effort has been focused on the impact of insulin resistance on the central nervous system. Further advances in understanding this complex topic will require the convergence of new findings from basic science, clinical research, and epidemiology. That is the work that lies before us now.

Correspondence: Dr Craft, VAPSHCS, S-182-GRECC, 1660 S Columbian Way, Seattle, WA 98108 (scraft@u.washington.edu).

Funding/Support: This study was supported by the Department of Veterans Affairs.

Acknowledgment: I thank John C. S. Breitner, MD, MPH, for his helpful comments.