Association between muscle mass and diabetes prevalence independent of body fat distribution in adults under 50 years old

The study addresses why individuals with similar BMI exhibit different risks for type 2 diabetes, emphasizing that adipose tissue distribution (android vs gynoid) and skeletal muscle mass may contribute to risk beyond overall adiposity. Skeletal muscle is the most insulin-sensitive tissue, responsible for the majority of post-prandial glucose disposal, suggesting low muscle mass could elevate diabetes risk. Prior NHANES analyses linked lower muscle mass metrics to insulin resistance and diabetes but generally did not control for fat distribution nor consistently perform sex-stratified analyses. Furthermore, findings in older adults may not generalize to younger adults (under 50), and race-specific associations in young U.S. adults are unclear. The authors hypothesized that lower skeletal muscle mass is associated with higher diabetes prevalence in young men and women, independent of body fat distribution, and secondarily examined potential differences by race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White).

Prior studies have shown associations between lower appendicular lean mass (ALM)/BMI and insulin resistance in older adults; lower skeletal muscle index (bioelectrical impedance) with insulin resistance and diabetes across age groups; and lower percent lean mass with higher HbA1c in both sexes, with some age-specific effects. However, these studies often lacked control for fat distribution (android vs gynoid), did not consistently stratify by sex, and included older adults, limiting applicability to younger populations. The literature also indicates sex differences in muscle and diabetes risk and suggests adipose depot-specific effects, with android fat increasing risk and gynoid fat potentially protective, supporting the need to account for fat distribution when evaluating muscle mass-diabetes associations.

Design: Cross-sectional analysis of NHANES 2005–2006, a nationally representative, stratified multistage survey of U.S. non-institutionalized adults. The analysis incorporated survey weights, strata, and primary sampling units. Population: Adults aged 20–49 years (n=1764) after exclusions: pregnancy, nursing, bilateral oophorectomy; use of testosterone, growth hormone, or glucocorticoids; height >192.5 cm or weight >136.4 kg due to DXA limits. From 10,348 NHANES participants, 1764 were eligible. Exposures/Measures: Body composition by DXA (Hologic QDR-4500; APEX software). Android region defined between lumbar midpoint and iliac crest; gynoid region between femoral head and mid-thigh. Appendicular lean mass (ALM) = sum of lean mass in both arms and legs. Primary predictor: percent ALM/weight. Covariates: Self-reported age, sex, race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, other), education (less than 12th grade; high school/GED; more than 12th grade), smoking (never, former, current), physical inactivity (no ≥10 min bouts of moderate/vigorous activity in past 30 days). Secondary coding of physical activity used AHA guidelines (≥150 min moderate-vigorous or ≥75 min vigorous/week). Height and BMI measured; height included as a covariate due to its relationship with muscle mass. Adiposity distribution captured by android/gynoid fat ratio. Outcome: Prevalent diabetes defined by any of: HbA1c ≥48 mmol/mol (6.5%), fasting plasma glucose ≥7.0 mmol/L (126 mg/dL), 2-h OGTT glucose ≥11.1 mmol/L (200 mg/dL), self-reported diabetes, or self-reported use of diabetes medications (oral agents and/or insulin). Sensitivity analysis defined diabetes by self-report only for participants missing biochemical data. Statistical analysis: Descriptive statistics (means ± SD; counts and percentages). Group comparisons by diabetes status used Wilcoxon tests (continuous), Chi-squared or Fisher’s exact (categorical). Primary analysis: multivariable logistic regression of diabetes on percent ALM/weight with (1) sex-stratified models, (2) sex-combined model including ALM/weight-by-sex interaction, and (3) sex- and race-stratified models. Adjusted for age, race, height, smoking, education, android/gynoid fat ratio, and physical inactivity (with alternative AHA-based physical activity coding). Multiple imputation (sequential multivariate imputation) addressed nonrandom missing DXA data; five imputed datasets analyzed. Survey design features applied in SAS 9.2/9.3. Two-sided p≤0.05 considered significant. Power: with n=1764, alpha=0.05, power=0.8, diabetes prevalence 5%, detectable OR≈1.15 for below- vs above-median ALM/weight.

- Sample: 958 men and 806 women; diabetes prevalence 5.2% (men) and 5.1% (women). - Individuals with diabetes were older, had higher BMI, higher android/gynoid fat ratio, lower ALM/weight, and were more physically inactive than those without diabetes. - Primary exposure-outcome associations (OR for diabetes per 1% decrease in ALM/weight): • Men: Adjusted for age, race, height, smoking, education: OR 1.31 (95% CI 1.18–1.45), p=0.0001. Further adjusted for android/gynoid fat: OR 1.20 (1.04–1.37), p=0.01. Further adjusted for physical inactivity: OR 1.18 (1.02–1.37), p=0.02. Using AHA physical activity categorization: OR 1.19 (1.01–1.39), p=0.04. • Women: Adjusted for age, race, height, smoking, education: OR 1.24 (1.05–1.46), p=0.01. Further adjusted for android/gynoid fat: OR 1.08 (0.90–1.30), p=0.42. Further adjusted for physical inactivity: OR 1.07 (0.87–1.31), p=0.54. Using AHA PA categorization: OR 1.13 (0.99–1.31), p=0.095. • Sex-combined: Adjusted for age, race, height, smoking, education: OR 1.26 (1.14–1.39), p=0.0001. Further adjusted for android/gynoid fat: OR 1.12 (1.01–1.25), p=0.038. Further adjusted for physical inactivity: OR 1.11 (0.98–1.25), p=0.096. Using AHA PA categorization: OR 1.14 (1.05–1.24), p=0.003. No significant sex interaction. - Sensitivity analysis (diabetes by self-report only): Effect estimates were similar but often not statistically significant due to fewer diabetes cases. - Race-stratified findings: In non-Hispanic White men, OR 1.30 (0.97–1.73), p=0.08 after adjustment (including android/gynoid fat); other sex–race groups showed non-significant associations. - Example provided: For two 91 kg men differing by 1% ALM/weight (31% vs 32%), the one with lower ALM/weight had approximately 1.20-times higher odds of diabetes (adjusted for android/gynoid fat).

The findings support the hypothesis that lower skeletal muscle mass relative to body weight is associated with higher odds of prevalent diabetes in young adults, and that in men this association persists after accounting for body fat distribution (android/gynoid ratio) and physical inactivity. This underscores the potential role of skeletal muscle in glucose homeostasis beyond overall adiposity and fat distribution. In women, the association attenuated and was not statistically significant after adjusting for android/gynoid fat, suggesting sex-specific differences in how muscle and fat depots relate to diabetes risk. Possible explanations include greater absolute muscle mass and higher diabetes prevalence in men, and a potentially more protective role of gynoid fat in women. The race-stratified analyses were underpowered, showing only a non-significant trend among non-Hispanic White men. Overall, results indicate that muscle mass may contribute independently to diabetes risk in men, while in women body fat distribution may play a comparatively larger role. Further longitudinal and interventional studies are needed to delineate causality and mechanisms.

Less skeletal muscle mass was independently associated with higher diabetes prevalence in young men after accounting for android and gynoid adiposity and physical inactivity. In women, the association was not statistically significant after adjusting for body fat distribution. These findings suggest that relative muscle mass may be an important marker—or potential contributor—to diabetes risk, particularly in men. Future research should include prospective cohorts and randomized controlled trials targeting muscle mass to determine causality, explore mechanisms underlying sex differences, and assess potential race/ethnicity-specific effects with adequate sample sizes.

- Cross-sectional design precludes causal inference; reverse causality is possible. - Residual confounding by adiposity may persist because ALM/weight correlates with multiple adiposity measures despite adjustment for android/gynoid fat. - Limited power in race-stratified analyses due to small subgroup sizes, leading to wide CIs and non-significant findings. - Generalizability may be limited to adults aged 20–49 years and to individuals within DXA scanner limits; older adults and those exceeding height/weight limits were excluded. - Some participants lacked biochemical measurements; although sensitivity analyses and multiple imputation were used, misclassification or reduced power may affect estimates. - Physical activity was self-reported, which may introduce measurement error.