U-shaped and J-shaped relationships between serum insulin and coronary heart disease in the general population: The Bruneck study

OBJECTIVE — To evaluate the relationship existing between serum insulin and coronary heart disease (CHD) in the general population. RESEARCH PESIGN AND METHODS — In a cross-sectional survey on atherosclerosis and its risk factors, 500 men and 500 women aged 40-79 years were randomly selected from the population of Bruneck, Italy. Clinical, biochemical, and behavioral risk factors of atherosclerosis were assessed in the 936 subjects who participated in the study. Serum insulin was measured at fasting (n = 888) and 2 h (n = 811, known diabetic subjects were excluded) after an oral glucose load. CHD was ascertained by an abnormal electrocardiogram and/or a history of angina or myocardial infarction. RESULTS— Subjects were stratified according to serum insulin quintiles at fasting or 2 h after glucose loading. After adjustment for sex, age, BMI, smoking, physical activity, alcohol intake, and socioeconomic status (analysis of covariance), cardiovascular risk factors clustered in subjects of the top insulin quintile. Multiple logistic regression analysis, including sex and age in model 1, sex, age, BMI, glucose tolerance, socioeconomic status, and behavioral variables in model 2, or this set of variables together with triglycerides and apoproteins Al and B, fibrinogen, and blood pressure status in model 3, revealed a significant association between high serum insulin and CHD when median insulin quintile was used as the reference class. Moreover, low serum insulin levels, such as those found in subjects of the lowest quintile, were independently related to CHD. These results were found either before (model 1) or after (models 2 and 3) adjusting for several covariates. Consistent results were found in men and women, as well as in younger and older subjects. CONCLUSIONS — Results of the present study suggest that both hyperinsulinemia and "hypoinsulinemia" are independent indicators of CHD. Furthermore, it is proposed that the relationship between CHD and fasting insulin is U-shaped, whereas that between CHD and postglucose insulin may be J-shaped