Association between estimated glucose disposal rate and metabolic syndrome in older adults with sarcopenia

Sarcopenia, an age-related syndrome marked by muscle mass and function decline, is closely linked to various adverse clinical outcomes. Insulin resistance (IR) has emerged as a pivotal link between sarcopenia and metabolic syndrome (MetS), significantly influencing their pathophysiological interactions. This study aimed to investigate the potential correlation between estimated glucose disposal rate (eGDR) and MetS among patients with sarcopenia, offering new insights into stratified prevention and treatment strategies.

A total of 970 elderly individuals who visited the 921 Hospital of the Joint Logistics Support Force of the Chinese People's Liberation Army from January 2023 to December 2024 were enrolled. Participants were categorized into three groups, including the non-sarcopenia group (n = 613), the sarcopenia group (n = 307) and the sarcopenia with MetS group (n = 50). Key variables were selected using univariate screening, the support vector machine recursive feature elimination (SVM-RFE) algorithm, and the Boruta algorithm. Three logistic regression models were constructed to evaluate the association between eGDR and MetS.

The prevalence of MetS in elderly individuals with sarcopenia was 14.01% (50/357), significantly higher than that of those without sarcopenia (3.10%, 19/613), with an odds ratio (OR) of 5.09 (95% CI: 2.95-8.79). Among the 357 sarcopenia patients, the average eGDR level in the sarcopenia with MetS group was 9.60 (9.21, 10.03) mg/kg/min, which was lower than that in the sarcopenia group, 10.82 (10.55, 11.09) mg/kg/min (Z = 2.64, P < 0.01). The Boruta algorithm identified crucial variables for developing a logistic regression model, indicating an OR for eGDR of 0.54 (95% CI: 0.30-0.84). The ORs for frailty and elevated cholesterol levels were 1.47 (95% CI: 1.01-2.94) and 3.94 (95% CI: 2.47-6.69), respectively. Model performance comparison showed that the Boruta and SVM-RFE-based models had higher area under the ROC curve values (0.78 and 0.75, respectively) than the univariate screening model (P < 0.05). Furthermore, subgroup analysis confirmed the robustness of the association between eGDR and MetS in elderly patients with sarcopenia.

In the older adult group, patients with sarcopenia have a higher possibility of developing metabolic syndrome (MetS) compared with those without sarcopenia. In addition to lower eGDR levels, frailty, elevated uric acid, BMI, and MDA levels are associate with a higher prevalence of MetS in individuals with sarcopenia. Machine learning-based regression models outperform univariate screening models in predicting MetS risk.