BackgroundLifestyle 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 findingsWe 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.ConclusionsIn 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.RegistrationNCT01947595Author summaryWhy Was This Study Done?Clinical trials in individuals with prediabetes have shown that the onset of type 2 diabetes can be delayed or prevented with lifestyle intervention.Among individuals with prediabetes, there is a large variability in the response to lifestyle intervention.It is unknown whether an intensification of intervention is able to improve the beneficial response.What Did the Researchers Do and Find?The present multicenter, risk stratified randomized and controlled intervention trial in 1105 German individuals with prediabetes prospectively confirms the existence of a high-risk prediabetes phenotypeThe intensification of lifestyle intervention in high-risk individuals improves the glycemic outcome after 1 year of lifestyle intervention, and additionally results in a higher frequency of regression to normal glucose tolerance after 3 years of follow up..Intensification of lifestyle intervention results in a larger reduction of liver fat content and stronger improves cardiometabolic outcomes in high-risk individuals.What Do These Findings Mean?Strategies for the prevention of type 2 diabetes should include risk stratification and individualised interventions.Our results highlight a dose-effect relationship for lifestyle intervention and suggest that “one size fits NOT all” in the field of diabetes prevention.It remains to be clarified whether low risk individuals benefit from lifestyle intervention, as there was a low number of individuals in this risk group in the current study.