Aims/hypothesis  

The aim of this study was to investigate the association of leptin levels with incident diabetes in middle-aged adults, taking into account factors purportedly related to leptin resistance.

Subjects and methods  

We conducted a case–cohort study (570 incident diabetes cases and 530 non-cases) representing the 9-year experience of 10,275 participants of the Atherosclerosis Risk in Communities Study. Plasma leptin was measured by direct sandwich ELISA.

Results  

In proportional hazards models adjusting for age, study centre, ethnicity and sex, high leptin levels (defined by sex-specific cut-off points) predicted an increased risk of diabetes, with a hazard ratio (HR) comparing the upper with the lower quartile of 3.9 (95% CI 2.6–5.6). However, after further adjusting additionally for obesity indices, fasting insulin, inflammation score, hypertension, triglycerides and adiponectin, high leptin predicted a lower diabetes risk (HR=0.40, 95% CI 0.23–0.67). Additional inclusion of fasting glucose attenuated this protective association (HR=0.59, 95% CI 0.32–1.08, p<0.03 for linear trend across quartiles). In similar models, protective associations were generally seen across subgroups of sex, race, nutritional status and smoking, though not among those with lower inflammation scores or impaired fasting glucose (interaction p=0.03 for both).

Conclusions/interpretation  

High leptin levels, probably reflecting leptin resistance, predict an increased risk of diabetes. Adjusting for factors purportedly related to leptin resistance unveils a protective association, independent of adiponectin and consistent with some of leptin’s described protective effects against diabetes.