UPPER-BODY FAT DISTRIBUTION - A HYPERINSULINEMIA-INDEPENDENT PREDICTOR OF CORONARY HEART-DISEASE MORTALITY - THE PARIS PROSPECTIVE-STUDY

The Paris Prospective Study is a long-term investigation of the factors predicting coronary heart disease in a large population of middle-aged men. The first follow-up examination involved 7,152 subjects, who were natives of metropolitan France and were free of any cardiovascular history. At that time, the usual cardiovascular risk factors and plasma insulin levels were recorded. An index of body fat distribution, the iliac-to-thigh ratio, was entered into the list of predictive variables, despite the fact that it had been measured 1 year before the first follow-up examination. After 11 years of mean follow-up, 129 of the men had died of coronary heart disease. Univariate analysis showed that the iliac-to-thigh ratio (p<0.0001) and plasma insulin level (both fasting [p<0.003] and 2-hour postload [p<0.02]), as well as the four major risk factors of coronary heart disease (age, smoking, blood pressure, and plasma cholesterol level) were significantly higher in subjects who died of coronary heart disease compared with those who had died of another cause or were alive at the end of follow-up. In multivariate stepwise logistic regression, the iliac-to-thigh ratio appeared as an independent predictor of coronary heart disease death, thereby causing the removal of fasting insulin level from the list of significant independent predictors. Nevertheless, in a model that entered 2-hour postload insulin in two classes (high or low), both the insulin level and iliac-to-thigh ratio were found as significant independent predictors. These results suggest that much of the already demonstrated predictive power of plasma insulin levels could be mediated through their association with upper-body fat distribution. However, 2-hour postload insulin concentration appears to improve the prediction and may denote the existence of metabolic abnormalities that are not entirely summarized by the type of fat distribution.