The incremental predictive value of biological aging indicators for cognitive impairment in older adults: a longitudinal analysis on the Mr. OS & Ms. OS cohort

Biological aging (BA) markers such as frailty and telomere length are closely linked to cognitive impairment (CI). However, the incremental value of biochemical marker-enriched frailty index (FI) and telomere length for predicting CI remains underexplored. We aimed to assess the incremental value of BA markers beyond conventional cognitive tests for CI risk stratification. A total of 1674 community-dwelling older adults without baseline CI were obtained from the Mr. OS & Ms. OS (Hong Kong) cohort. Baseline BA measures included frailty phenotype, three FI versions (without/with 2 or 4 serum biochemical markers: creatinine, homocysteine, high-sensitivity C-reactive protein, and 25-hydroxyvitamin D), and leukocyte telomere length. CI was assessed by concurrently using the MMSE and CSI-D tests at the 7-year follow-up. Penalized logistic regression was used to evaluate the incremental value of BA indicators beyond cognitive tests for CI prediction, with the area under the precision-recall curve (AUPRC) as the primary performance metric. The mean age of the study sample was 70.7 years (SD: 4.2), and 44.1% were females. The 7-year incidence of CI was 17.0% (285/1674). Compared to baseline CSI-D score, the FI incorporating four biochemical markers demonstrated significant incremental value for CI prediction (AUPRC improvement: 0.037, P < 0.001), while other frailty indicators (frailty phenotype and other two FI versions) showed no significant added value. Telomere length provided additional predictive value across all frailty-related models (AUPRC improvements: 0.009–0.078, all P < 0.001) except the FI without biochemical markers. The optimal prediction model, which considered the FI with four biochemical markers and telomere length, achieved an AUPRC (SD) of 0.568 ± 0.092 and an AUROC (SD) of 0.826 ± 0.037. Frailty index with four biochemical markers and telomere length showed moderate incremental value over cognitive tests for CI prediction in older adults. The findings highlight the importance of multisystem biological aging assessment in CI risk stratification. Further validation is warranted in large cohorts.