Work

Association of metabolic obesity phenotypes with cognitive decline in the ELSA ‐Brasil study

Year: 2026

Type: article

Abstract: AIMS: To investigate the association between metabolic obesity phenotypes and cognitive decline and evaluate the potential mediating role of C-reactive protein (CRP). METHODS: Longitudinal cohort study using three waves (2008-2019) of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Data were analysed from December 2024 to May 2025. Baseline sample consisted of 15 105 participants aged 35-74 years. Six phenotypes were defined by body mass index (BMI) category (normal weight, overweight, obesity) and metabolic health. Metabolic health was defined using traditional metabolic syndrome criteria and more stringent criteria: low waist-to-hip ratio (WHR), no diabetes or hypertension. Global cognition Z-scores were derived from tests of memory (immediate, delayed recall, and recognition of a word list), verbal fluency (phonemic and semantic), and Trail-Making tests (TMT-B). Mediation analysis evaluated CRP as a potential mediator. RESULTS: Among 12 795 participants (mean age 51.1 years; 55% women; 53% White) followed for a median of 8 years, metabolically unhealthy phenotypes-across all BMI categories-were associated with faster cognitive decline (β estimates ranged from -0.037 to -0.053; all p < 0.001), whereas metabolically healthy overweight (β = 0.016; 95% CI = -0.002, 0.034; p = 0.081) and metabolically healthy obesity (β = 0.000; 95% CI = -0.027, 0.026; p = 0.981) were not. No evidence of CRP mediation was identified. BMI was not associated with cognitive decline (β = -0.001; 95% CI = -0.002, 0.000; p = 0.170), whereas WHR was (β = -0.020, 95% CI = -0.026, -0.014, p < 0.001). CONCLUSIONS: Metabolic dysfunction may be a stronger predictor of subsequent cognitive decline than excess body weight. Dementia prevention strategies may benefit from early identification and management of metabolic dysfunction across all weight categories.

Source: Diabetes Obesity and Metabolism

DOI: 10.1111/dom.70466