Risk-stratified lifestyle intervention to prevent type 2 diabetes

Background: Lifestyle intervention (LI) can successfully prevent type 2 diabetes, but response to LI strongly varies depending on risk subphenotypes. We tested if individuals with prediabetes and a high-risk phenotype benefit from an intensification of LI. Methods and findings: We conducted a risk stratified multicenter randomized controlled intervention study over 12 months with additional 2 year follow up. In eight University Hospitals in Germany, 1105 individuals (female 59%, age 58±11 years, BMI 31.1±6.0 kg/m2 (mean±SD)) with impaired fasting glucose and/or impaired glucose tolerance were included between May 2012 and May 2016 in the study. Participants were stratified into 2 groups; a high- and low-risk phenotype, based on insulin secretion, insulin sensitivity and liver fat content. Low-risk individuals were randomly assigned to conventional LI or control (1:1), high-risk individuals to conventional or intensified LI (1:1), each over one year. Intensified LI included doubling of physical exercise and time of counselling. The primary endpoint was change in post-challenge glucose levels, assessed by frequently sampled oral glucose tolerance tests. Secondary endpoints included changes in liver fat content, assessed by magnetic resonance spectroscopy. A total of 908 (82%) participants completed the study after 12 months of LI. In high-risk individuals, the mean difference estimate between conventional and intensified LI in change in post-challenge glucose levels from baseline was -0.290 mmol/l [CI: -0.544;-0.036], p=0.025. Liver fat content was more reduced by intensified LI than by conventional LI (mean difference estimate: -1.34 percentage points [CI: -2.17;-0.50], p=0.002), and cardiovascular risk decreased stronger with intensified LI than with conventional LI (mean difference estimate -1.82 [CI: -3.13-0.50], p=0.007). In low-risk individuals, conventional LI was not superior to control in reducing postprandial glucose, liver fat or cardiovascular risk. During the total observation period of 3 years, high-risk participants with intensified LI had a higher probability to normalize glucose tolerance compared to conventional LI (p=0.003). The limitations of this study include a relative short duration of LI, a non-completer rate of 18% and an underrepresentation of low risk individuals. Conclusions: In high-risk individuals with prediabetes it is possible to improve glycemic and cardiometabolic outcomes by intensification of the commonly recommended conventional LI. Our results show that individualized, risk-phenotype-based LI can be implemented for the prevention of diabetes. Registration: NCT01947595