Triglyceride-Glucose-Waist-to-Weight Index: Novel Biomarker for Mild Cognitive Impairment in Older People With Sarcopenic Obesity.

BACKGROUND

Sarcopenic obesity (SO) is closely associated with mild cognitive impairment (MCI). However, traditional body composition-based diagnostic criteria for SO fail to adequately incorporate insulin resistance, a key pathophysiological mechanism.

This study aimed to investigate the association between SO, defined using multiple diagnostic criteria including novel insulin resistance indices, and MCI and to identify the optimal SO criterion for MCI identification.

METHODS

This study analysed data from the 2025 Chongming Older People Screening Project, CHARLS (2015-2020) and ELSA (2012-2017). We used multivariable logistic regression to assess the link between SO and MCI.

The optimal obesity indicator for identifying MCI in sarcopenia was identified using ROC analysis and then incorporated into an XGBoost model to detect MCI risk. Subgroup analyses were performed to assess the association between SO and decline in multiple cognitive domains.

RESULTS

This study included 2326 individuals from the 2025 Chongming Older People Screening Project, 3392 from CHARLS and 1825 from ELSA.

The mean age was 72.62 ± 5.56 years in the cross-sectional study, 66.70 ± 5.39 years in CHARLS and 67.99 ± 5.55 years in ELSA. The proportion of female participants was 55.93% in the cross-sectional study, 57.59% in ELSA and 40.60% in CHARLS.

MCI was consistently identified across all assessment tools in three independent cohorts using the metabolically oriented definition integrating sarcopenia with an elevated triglyceride-glucose-waist-to-weight index (TyG-WWI) (Chongming: OR = 3.04, 95% CI: 1.83-5.07, p < 0.001; CHARLS: HR = 1.57, 95% CI: 1.01-2.42, p = 0.043; and ELSA: HR = 1.87, 95% CI: 1.01-3.46, p = 0.047). TyG-WWI was identified as the optimal obesity indicator to identify MCI (AUC = 0.71, 95% CI: 0.63-0.78).

The machine learning model based on TyG-WWI demonstrated strong discriminative performance in the cross-sectional study (AUC = 0.93, 95% CI: 0.90-0.97) and maintained robust discriminative ability in external validation (CHARLS: AUC = 0.86, 95% CI: 0.80-0.92; ELSA: AUC = 0.84, 95% CI: 0.73-0.95). Subgroup analyses revealed that SO was associated with decline in multiple cognitive domains, particularly executive function and memory.

CONCLUSIONS

This study demonstrates that a metabolically oriented SO definition integrating sarcopenia with elevated TyG-WWI is a more effective tool for identifying MCI among community-dwelling older people compared with traditional body composition-based criteria.