Adiponectin, change in adiponectin, and progression to diabetes in the Diabetes Prevention Program

Abstract

Objective: To determine whether baseline adiponectin levels or intervention-associated change in adiponectin levels were independently associated with progression to diabetes in the Diabetes Prevention Program (DPP).

Research design and methods: Cox proportional hazards analysis was used to evaluate the contribution of adiponectin and treatment-related change in adiponectin to risk of progression to diabetes.

Results: Baseline adiponectin was a strong independent predictor of incident diabetes in all treatment groups (hazard ratio per approximately 3 microg/ml higher level; 0.61 in the lifestyle, 0.76 in the metformin, and the 0.79 in placebo groups; all P < 0.001, P = 0.13 comparing groups). Baseline differences in adiponectin between sexes and race/ethnicity groups were not reflected in differences in diabetes risk. DPP interventions increased adiponectin levels ([means +/- SE] 0.83 +/- 0.05 microg/ml in the lifestyle group, 0.23 +/- 0.05 microg/ml in the metformin group, and 0.10 +/- 0.05 microg/ml in the placebo group; P < 0.001 for increases versus baseline, P < 0.01 comparing groups). These increases were associated with reductions in diabetes incidence independent of baseline adiponectin levels in the lifestyle and placebo groups but not in the metformin subjects (hazard ratio 0.72 in the lifestyle group (P < 0.001), 0.92 in the metformin group (P = 0.18), and 0.89 in the placebo group; P = 0.02 per approximately 1 microg/ml increase, P = 0.02 comparing groups). In the lifestyle group, adjusting for change in weight reduced, but did not remove, the effect of increased adiponectin.

Conclusions: Adiponectin is a powerful marker of diabetes risk in subjects at high risk for diabetes, even after adjustment for weight. An increase in adiponectin in the lifestyle and placebo groups was associated with a reduction in diabetes risk. However, these changes in adiponectin were comparatively small and less strongly related to diabetes outcome than baseline adiponectin levels.

FIG. 1

Box-and-whisker plot showing the distribution of adiponectin at baseline by sex and self-described race/ethnicity. The box indicates the 25th through 75th percentile of the distribution, the whiskers show the 10th to 90th percentiles, the large dot shows the mean, and the line dissecting the box shows the median. Women had significantly higher levels of baseline adiponectin within each race/ethnic group (P < 0.00001), and white participants had significantly higher baseline adiponectin than all other race/ethnicity groups (P < 0.00001).

FIG. 2

Diabetes hazard rates and baseline adiponectin adjusted for sex, age, race/ethnicity, and baseline weight (Table 2, model 2). Cox proportional hazards models were used to estimate the risk of developing diabetes. Estimates of the absolute risk gradient associated with a given value of baseline adiponectin and adiponectin change using the range of values (5th to 95th percentiles) were used to describe the hazard rate for a participant with a value equal to the group mean. The point on each line indicates the estimated hazard rate for a subject with a value equal to the mean value for the group as estimated in the life table analysis.

FIG. 3

Change in adiponectin by treatment group. Unadjusted values are presented. Bars represent means ± SE for change in adiponectin over 1 year of treatment. ■, men; □, women.

FIG. 4

Diabetes hazard rates and change in adiponectin (Table 2, model 4). Cox proportional hazards models were used to estimate the risk of developing diabetes. Estimates of the absolute risk gradient associated with a given value of baseline adiponectin and adiponectin change using the range of values (5th to 95th percentiles) were used to describe the hazard rate for a participant with a value equal to the group mean. The point on each line indicates the estimated hazard rate for a subject with a value equal to the mean value for the group as estimated in the life table analysis.