For results that have...

But don't include...


Click to add/remove this article to your list of 'My Favorites' Insulin Secretion, Weight Loss, and the Risk of Developing Diabetes in Participants Treated with Metformin vs. Placebo in the Diabetes Prevention Program (DPP)

Year: 2005

Abstract Number: 1004-P


Institutions: Rockville, MD

Results: In the DPP, metformin (MET) treatment reduced the risk of diabetes onset by 30% compared with placebo (PLA). Proportional Hazard Models assessed the association of adiposity and metabolic factors at baseline, and their changes over time, with risk of diabetes. Time dependent variables included weight, HbA1c, fasting insulin (FI) and proinsulin, an insulin secretory index (ISI) [(I30min - I0)/(Glucose30 - G0)]. Models also included baseline fasting glucose (FPG).

In the MET group multivariate model, the strongest predictors of diabetes risk were ISI at baseline and during follow-up (table). Change in body weight, baseline FGP, and FI at baseline and during follow-up were less significant predictors. Conversely, in the PLA group multivariate model, the dominant predictors of diabetes risk were baseline FPG followed by change in body weight. ISI at baseline and follow-up and FI at baseline but not follow-up were significant but contributed little to the model.

Effect of weight and metabolic factors on diabetes risk in multivariate models
MeasurementR2Relative hazardR2Relative hazard
FPG at baseline (per 10 mg/dl)1.0%1.327.9%2.05
Weight change over time (per kg weight gain)1.1%1.073.9%1.10
ISI at baseline (per 10 units lower)3.6%1.101.6%1.03
ISI over time (per 10 unit decrease)2.7%1.160.59%1.01
FI at baseline (per uU/ml higher)1.0%1.020.53%1.01
FI over time (per uU/ml increase)0.64%1.020.24%1.01
The effect of changes in ISI, but not in body weight or FI, on risk differed significantly between groups. The greater mean weight change in the MET vs. PLA group (-1.82 vs. +0.12 kg) did not wholly explain the reduction in diabetes risk with metformin. Risk reduction was further explained by increases in insulin secretion relative to fasting insulin in the MET group.

Category: Epidemiology