In a hypertensive family cohort characterized by insulin resistance [2–4], we found a cumulative effect of MetS components on the risk of developing diabetes, which was relatively similar to that reported in the West of Scotland Coronary Prevention Study (WOSCOPS) . In a high insulin-resistance population, the effect of MetS accumulation was also observed in subjects with impaired fasting glucose . Our most important finding is that the risk of developing new-onset diabetes was associated with the number of positive MetS components, but not with central obesity. When analyses were limited to those subjects with three MetS components, neither insulin resistance (expressed by HOMA-IR) at baseline, nor diabetes incidence during follow-up was significantly different between subjects with and without central obesity. Previous reports that subjects with central obesity had a higher risk of developing diabetes might be based on the evidence that these individuals has a higher number of associated MetS components [21, 22]. In accord with our observations, results from the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) demonstrated that MetS was a better predictor of development of diabetes than BMI in hypertensive subjects . Nevertheless, Yasuda T, et al.  found that central obesity did not impose any additional influence on carotid atherosclerosis among subjects with the same number of positive MetS components in a cross-sectional study. Non-obese subjects with metabolic syndrome also had a similar risk of developing cardiovascular disease compared with obese individuals in prospective observations [25, 26]. In fact, a joint interim statement from several associated organizations has suggested that MetS criteria need to be unified and waist circumference might not be the prerequisite for the criteria .
The proportion of new-onset diabetes among the members of hypertensive families in the present study was higher than that in the WOSCOPS and also than that reported in a previous Chinese community survey [7, 28]. On the other hand, Shirakawa et al.  had reported that a family history of hypertension was associated with the presence of hypertension but not diabetes in a previous cross-sectional study. According to the high cumulative incidence of diabetes in our prospective cohort, however, a family history of hypertension does provide important information in predicting development of diabetes. These findings might be due to the hyperinsulinemia observed in hypertensive family . High fasting insulin is associated with many metabolic abnormities and can lead MetS to progress . Similarly, in the present study, insulin resistance with high fasting insulin levels had a positive association with the number of MetS components and incidence of diabetes. However, we did not further assess the possible mechanisms connecting a family history of hypertension and the incidence of diabetes. It has been reported that circulating aldosterone might associate with insulin resistance , and the autonomic imbalance is probably related to diabetes .
The components of MetS partially share the same mechanisms, and frequently coexist. Although the predictive power of MetS seems not as strong as the sum of all its components, the diagnosis of MetS is important as a superimposed risk on traditional risk factors. [20, 32, 33]. On the other hand, subjects without obesity probably have several metabolic abnormalities which can lead to an increased risk of diabetes [34, 35]. In this regard, we should not ignore the potential risk in subjects without central obesity. Furthermore, several studies showed that impaired fasting glucose is a strong predictor of development of diabetes [20, 23, 28]. In addition to glucose, fasting triglyceride was the other independent predictor of onset of diabetes in Chinese subjects . In particular, these finding is obvious in the subjects with central obesity in our data (see Table 4). In fact, lipotoxicity by deposition of triglycerides and free fatty acids may be associated with insulin resistance and beta-cell dysfunction [36, 37]. Therefore, it is important to closely monitor or treat subjects with high levels of triglycerides and glucose because these two components are clearly associated with development of diabetes, especially in those with central obesity [37, 38].
There were a few limitations in the present study. First, central obesity is one of the present MetS criteria, so subjects with central obesity wound inevitably have a higher number of positive MetS components than those without central obesity if we did not select subjects with the same number of metabolic abnormalities. We focused on subjects with only three positive MetS components in the present study. No analyses of data for subjects with four positive components, the maximal number of abnormalities in the without central obesity group, were performed because there were fewer cases in this subgroup. Secondary, as the aim of the present study was to investigate MetS, we used the criterion of waist circumference to define central obesity. However, BMI is often assessed in the study of metabolically obese, normal-weight (MONW) subjects . In addition, we did not find significant differences in predicting diabetes in subjects with normal waist circumference. However, some adipocytokines, including serum leptin, are associated with MetS and insulin resistance in non-obese as well as obese subjects [39–41]. Further studies on predisposing factors leading to diabetes are needed.