Objective: We tested the hypothesis that fasting serum triglycerides (TG) predict new onset diabetes (NOD) in apparently healthy men, and that physical fitness (PF) modifies this association. Secondly, we aimed to compare the predictive ability of TG with TG/HDL-cholesterol (TG/HDL) ratio for NOD. Research Design and Methods: TG was measured 1972-1975 in 1962 healthy men aged 40-59 years with fasting blood glucose (FBG) levels (< 6.1mmol/l). Data from an almost identical follow-up, performed in 1979-1982, including HDL measurements, were used to compare the predictive impact of TG with TG/HDL ratio. PF was measured using a cycle ergometer exercise test, and defined as total work divided by body weight (kJ/kg). Survival models were analyzed in TG tertiles, and relative risks were adjusted for age, BMI, FBG, family history of maternal diabetes mellitus, and PF. Results: Incidence of NOD was 202 cases (10.3%) during 35 years follow-up. The highest TG tertile was associated with increased risk of NOD compared to the lowest tertile (hazard ratio [HR] 1.90, [95% CI] 1.32-2.79) and 1.73 [1.20-2.54]) without and with adjustments for PF, respectively. HR for risk of NOD (up to 28 years follow-up) per SD increase in log (TG) and log (TG/HDL ratio) were 1.43 (1.18-1.73) and 1.49 (1.23-1.81), respectively. Conclusions: Fasting serum TG is a strong long-term predictor of incident NOD among apparently healthy men, even after adjustments for physical fitness. The triglyceride/HDL ratio may be slightly better than TG alone to predict NOD.
Key Words: Epidemiology, lipids, diabetes, prediction, fitness.
Early identification of individuals at high risk of new onset diabetes (NOD) is important to facilitate preventive intervention. There is some evidence that high levels of fasting serum triglycerides (TG) predict NOD among both women and men (1-4). These studies have, however, used different inclusion criteria and methods for assessing TG-associated risk of NOD. Perry et al studied risk factors of NOD in a general population of middle-aged men, and Tirosh et al studied the TG-NOD association in young men (2, 5). The long-term prognostic ability of TG for NOD in healthy middle aged men has not been reported earlier. Physical fitness (PF) and physical activity (PA) are only modestly correlated, but both high PF and high PA are associated with reduced risk of NOD (1, 6-9). Previous studies have only adjusted for self-reported PA when assessing the association between TG levels and the risk of NOD (1, 2, 5). High density lipoprotein cholesterol (HDL) levels are inversely associated with TG levels and directly associated with PF. HDL might therefore modify the TG-NOD risk association. To our knowledge, no study has evaluated the combined prognostic impact of TG, HDL and PF. The aims of the present study were to test whether TG levels predict NOD in healthy middle-aged men and to assess whether PF modifies any association. Finally we also compared the predictive power of TG with the TG/HDL ratio.