gms | German Medical Science

Introduction: Common behaviors may affect body fat distribution, in particular visceral adipose tissue (VAT) accumulation, glycemic status, and the prevalence of diabetes. We hypothesized that the combination of a healthy diet, physical activity, not smoking, and moderate alcohol intake results in joint beneficial health effects. Therefore, we examined the relation of a composite Lifestyle Risk Factor Index (LSRI) that captures these four health behaviors with VAT levels and diabetes/glycemic status based on data from the first MRI data freeze of the German National Cohort (NAKO).

Methods: Based on questionnaires completed at entry into the NAKO, LSRI scores were assigned (one point each for consuming <1 (women)/<2 (men) alcoholic drinks/day, =1.5 hours/week physical activity, not smoking, and adhering to =3/7 dietary recommendations). A Food Frequency Questionnaire allowed estimation of dietary intakes. VAT volumes were obtained from standardized magnetic resonance images at 3T that were analyzed by deep learning-based image segmentation. Diabetes status was classified according to self-report, while abnormal glycemic status was defined as HbA1c >38 mmol/mol. We applied general linear models to compute mean BMI (kg/m²) and VAT levels (cm²) by LSRI category and logistic regression to estimate prevalence odds ratios (POR) for diabetes/glycemic status.

Results: Of 6,828 participants with complete information, 41% had a healthy weight, 39% were overweight, and 20% were obese. The respective proportions for 0/1, 2, 3, and 4 LSRI points were 8%, 29%, 48%, and 15%. HbA1c levels were abnormal (>38 mmol/mol) for 24% of participants and 4.9% reported a diabetes diagnosis. As to individual components, 86% received a point for physical activity, 82% for not smoking, 77% for moderate alcohol intake, but only 23% for diet. Adherence to dietary recommendations was highest for Fruit plus Juice (45%), Refined Grains (38%), and Vegetables (incl. Potatoes) (32%), while it was low for Unprocessed Meat (18%), Fish (17%), Processed Meat (12%), and Whole Grains (2%). In adjusted models, participants with a LSRI score of 4 vs. 0/1 had lower mean levels of BMI (25.7; 95%CI 25.4, 26.0 vs. 26.6; 95%CI 26.2, 26.9) and VAT (0.88; 95%CI 0.85, 0.92 vs. 1.13; 95%CI 1.09, 1.18). Adding BMI to the VAT model attenuated the difference across categories: 0.95 (95%CI 0.93, 0.98) vs. 1.12 (95%CI 1.09, 1.15). In models for individual LSRI components, only physical activity and diet were inversely associated with VAT levels. Although diabetes prevalence was not associated with the LSRI score, participants with LSRI scores of 3 and 4 as compared to 0/1 were less likely to have abnormal HbA1c values (POR 0.78; 95%CI 0.63, 0.97 and 0.72; 95%CI 0.55, 0.94), but this association was attenuated after adding BMI or VAT to the model.

Conclusion: These findings support the hypothesis that a combination of healthy lifestyle factors as captured by a composite LSRI is reflected by lower VAT levels independent of BMI and a lower probability for an abnormal glycemic status.

The authors declare that a positive ethics committee vote has been obtained.