Effect of body mass index on insulin secretion or sensitivity and diabetes.

Abstract

Although the association between obesity and diabetes is well known, the factors predisposing to diabetes in non-obese Asians are less clearly characterized. To investigate the effects of impaired insulin secretion (IIS) and insulin resistance (IR) according to BMI on the incidence of diabetes in the Saku Study. This 4-year cohort study involved 3,083 participants aged 30-69 years without diabetes at baseline (2006-2007). Participants were stratified by BMI (<23.0, 23.0-24.9, and ≥25). Based on insulinogenic index and homeostasis model of IR values, participants were classified into four categories: normal; isolated IIS (i-IIS); isolated IR (i-IR); and IIS plus IR. All data were collected in 2006-2011 and analyzed in 2013-2014. For participants with BMI <23.0, the risk of developing diabetes was higher in the i-IIS (adjusted hazard ratio=6.6; 95% CI=3.7, 11.6) and IIS plus IR groups (9.4; 3.1, 28.4) than in the normal group. For participants with BMI 23.0-24.9, risk was higher in the i-IIS (9.9; 4.4, 22.1); i-IR (3.4; 1.2, 9.5); and IIS plus IR (23.4; 9.3, 58.9) groups. Among participants with BMI ≥25, risk was higher in the i-IIS (16.9; 6.0, 47.7); i-IR (7.9; 2.8, 22.6); and IIS plus IR (26.9; 9.0, 80.8) groups. BMI was negatively associated with incidence of IIS but positively associated with IR incidence. Individuals with normal BMI may develop diabetes mainly through IIS, whereas individuals with high BMI may develop diabetes primarily through IR.