AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Latent Biomarker Phenotypes in Older Adults Identify Distinct Patterns of Frailty, Cardiovascular Disease, and Mortality Risk
By Arya Nezhad S, Mangalesh S, Rossi R et al.
Background: Frailty, cardiovascular disease, and mortality in older adults arise from multiple interacting biologic processes. However, most studies have evaluated individual biomarkers in isolation, potentially overlooking clinically meaningful biologic phenotypes. We sought to identify latent biomarker phenotypes in older adults and evaluate their associations with frailty, cardiovascular disease, and mortality. Research Question: Among older adults, do routinely available laboratory biomarkers cluster into distinct, clinically meaningful phenotypes, and are these phenotypes associated with frailty, cardiovascular disease, and mortality? Methods: We analyzed adults aged ≥65 years from NHANES 2015-2018. Nine biomarkers spanning inflammatory, glycemic, renal, hematologic, lipid, and nutritional domains were included: high-sensitivity C-reactive protein (hsCRP), hemoglobin A1c (HbA1c), urine albumin-creatinine ratio (UACR), estimated glomerular filtration rate (eGFR), vitamin D, white blood cell count, hemoglobin (Hb), HDL cholesterol, and serum albumin. Frailty was defined using a modified Fried phenotype. Latent class analysis (LCA) was used to identify biomarker-defined phenotypes. Survey-weighted logistic and Cox regression models evaluated associations between latent classes, frailty, prevalent cardiovascular disease (CVD), and mortality outcomes. Results: A total of 2,248 U.S. older adults (weighted, 41.5 million) were analyzed. Among candidate latent class models, the 3-class solution was selected based on fit statistics (Figure 1). Three clinically interpretable phenotypes were observed: Class 1, an inflammatory-metabolic phenotype characterized by elevated inflammatory markers, diabetes, and low HDL-C; Class 2, a renal-proteinuric-anemia phenotype characterized by albuminuria, renal dysfunction, anemia, and hypoalbuminemia; and Class 3, a relatively preserved biomarker phenotype. Compared with Class 3, Class 2 was associated with higher odds of prevalent clinical cardiovascular disease (adjusted odds ratio 1.98 [1.22-3.20]) and frailty (2.31 [1.69-3.16]). In fully adjusted Cox regression models, Class 2 was also associated with increased all-cause mortality (adjusted hazard ratio 2.34 [1.63-3.35]) and cardiovascular mortality (2.10 [1.10-4.03]) (Figure 2). Class 1 was associated with frailty but not mortality after adjustment. Conclusions: Among older U.S. adults, latent biomarker phenotypes exhibited markedly different profiles of frailty, cardiovascular disease, and mortality. The renal-proteinuric-anemia phenotype was associated with substantially higher risks of frailty, cardiovascular disease, and mortality, suggesting that integrated biomarker patterns may help identify biologically vulnerable older adults at increased risk of adverse outcomes.