Associations of Body Composition and Physical Function With Incident Diabetes in Older Adults: A 14-Year Prospective...

Longitudinal associations between age-related changes in body composition, physical performance and incident diabetes mellitus (DM) in older adults and the mediating role of branched-chain amino acids (BCAAs) remain understudied.

This prospective cohort study included 4000 community-dwelling older adults at baseline. Longitudinal analysis of incident DM involved 3421 participants without baseline DM. Incident DM was identified through self-reported diagnosis, medication use, hospital records or fasting glucose ≥ 7.0 mmol/L. Body composition (PBF [percent body fat], FMI [fat mass index], ASM [appendicular skeletal muscle mass]) was measured using DXA. Associations of baseline body composition with incident DM were assessed using Cox regression in the full cohort, with sensitivity analysis accounting for competing risk of death. Associations in long-term subgroups and analysis of 14-year and prediabetes changes were evaluated using multivariable logistic regression. Serum BCAAs were quantified by LC-MS for mediation analysis.

Among 3421 participants (median age 72 years [IQR 68-76]; 50.1% female) without baseline DM, higher baseline adiposity (PBF: HR 1.613; FMI: HR 1.684; waist: 1.587) and ASM/ht² (HR 1.561) were consistently associated with increased DM risk in adjusted Cox models (all p < 0.001). Conversely, higher relative muscle mass (ASM/weight: HR 0.603; ASM/BMI: HR 0.631; both p < 0.001) was associated with a lower risk. Faster gait speed was also linked to lower DM risk (HR 0.748, p < 0.001). Sensitivity analysis accounting for competing mortality demonstrated consistent results. In the complete 14-year follow-up subgroup (n = 937, 130 incident DM cases), key baseline associations were confirmed using logistic regression (e.g., ASM/weight: OR 0.633, p = 0.005). Analysis of 14-year changes showed that attenuated loss of relative muscle mass (ΔASM/weight: OR 0.657; p = 0.005) was correlated with lower DM risk, whereas increased adiposity was correlated with higher risk (∆PBF: OR 1.509, p = 0.019). Pre-DM changes over 0-4 years also revealed that early increases in waist (OR 1.428, p = 0.005) and declines in ASM/weight (OR 0.745, p = 0.013) were associated with subsequent DM. Serum BCAAs mediated 18.6%-38.2% of associations between baseline body composition and incident DM (all p < 0.01). No significant correlation was found between dietary protein intake and serum BCAAs after adjustment.

In older adults, higher baseline adiposity and lower relative muscle mass were associated with long-term DM risk, while preserved muscle mass and faster gait speed were linked to a lower risk. These associations, partially mediated by BCAAs, advance the understanding of DM risk pathways and highlight the potential clinical relevance of weight- or BMI-adjusted relative muscle mass assessment in this population.