30-minute postload plasma glucose levels during an oral glucose tolerance test predict the risk of future type 2 diabetes: the Hisayama Study
Yoichiro Hirakawa, Jun Hata, Masahito Yoshinari, Mayu Higashioka, Daigo Yoshida, Mao Shibata, Takanori Honda, Satoko Sakata, Hiroyuki Kato, Takanori Teramoto, Hideki Maki, Shozo Nishimoto, Takanari Kitazono, Toshiharu Ninomiya
Abstract
INTRODUCTION: To investigate the associations of 30 min postload plasma glucose (30 mPG) levels during an oral glucose tolerance test (OGTT) with the risk of future diabetes in a general Japanese population. RESEARCH DESIGN AND METHODS: A total of 2957 Japanese community-dwelling residents without diabetes, aged 40-79 years, participated in the examinations in 2007 and 2008 (participation rate, 77.1%). Among them, 2162 subjects who received 75 g OGTT in a fasting state with measurements of plasma glucose level at 0, 30, and 120 min were followed up for 7 years (2007-2014). Cox's proportional hazards model was used to estimate HRs and their 95% CIs of each index for the development of type 2 diabetes using continuous variables and quartiles with adjustment for traditional risk factors. The influence of 30 mPG on the predictive ability was estimated with Harrell's C-statistics, integrated discrimination improvement (IDI), and the continuous net reclassification index (cNRI). RESULTS: During follow-up, 275 subjects experienced type 2 diabetes. Elevated 30 mPG levels were significantly associated with increased risk of developing diabetes (p<0.01 for trend): the multivariable-adjusted HR was 8.41 (95% CI 4.97 to 14.24) for the highest versus the lowest quartile, and 2.26 (2.04 to 2.52) per 1 SD increase. This association was attenuated but remained significant after further adjustment for fasting and 2-hour postload plasma glucose levels. Incorporation of 30 mPG into the model including traditional risk factors with fasting and 2-hour postload plasma glucose levels for diabetes improved the predictive ability of type 2 diabetes (improvement in Harrell's C-statistics values: from 0.828 to 0.839, p<0.01; IDI: 0.016, p<0.01; cNRI: 0.103, p=0.37). CONCLUSIONS: Elevated 30 mPG levels were associated with increased risk of diabetes, and inclusion of 30 mPG levels significantly improved the predictive ability for future diabetes, suggesting that 30 mPG may be useful for identifying high-risk populations for type 2 diabetes.