Age-Stratified Serum Biomarkers for Discriminating Clinically Defined Cognitive Impairment in Neurological Inpatients: A Cross-Sectional Study

Background: With the aging of the global population and the increasing prevalence of cognitive dysfunction worldwide, early identification of cognitive impairment has become an important public health issue. This study aims to investigate the correlation between peripheral serum neurodegenerative biomarkers and age, as well as the discriminative efficacy of cognitive impairment across different age cohorts. Notably, subjects were grouped by clinical cognitive assessment without pathological confirmation via amyloid/tau PET or cerebrospinal fluid testing to define Alzheimer’s disease pathology. Methods: A cross-sectional study enrolled 425 inpatients from the neurology department. Nonparametric tests were used to explore the correlation between age and peripheral serum biomarkers, as well as intergroup differences in biomarker levels among different cognitive statuses. Age-stratified logistic regression models were constructed to evaluate the association between biomarkers and cognitive impairment, with odds ratios (OR) and 95% confidence intervals (CI) calculated. Receiver operating characteristic (ROC) curves and corresponding area under the curve (AUC) were adopted to assess the discriminative performance of each indicator. In this study, cognitive impairment was defined clinically based on MoCA and clinical assessment, not biomarker-confirmed Alzheimer’s disease pathology. Results: Except for glial fibrillary acidic protein (GFAP) (r = 0.062, p = 0.207), all detected biomarkers were significantly correlated with age. Among them, p-Tau181 (r = 0.288, p < 0.001), p-Tau217 (r = 0.275, p < 0.001), Aβ1-42 (r = 0.132, p = 0.007), Aβ1-40 (r = 0.254, p < 0.001), and NfL (r = 0.200, p < 0.001) were positively associated with age, while Aβ42/40 ratio was negatively correlated with age (r = −0.253, p < 0.001). After age-based grouping, phosphorylated tau protein (p-Tau)181 showed relatively better discriminative performance in the <65 years group (AUC = 0.6212, 95% CI: 0.537–0.705); GFAP (AUC = 0.7776, 95% CI: 0.718–0.837) and NfL (AUC = 0.7329, 95% CI: 0.670–0.796) showed higher AUC values among participants aged ≥65 years. Conclusions: The correlation and discriminative efficacy of peripheral serum biomarkers for clinically defined cognitive impairment showed differences across age groups in this exploratory analysis. Within this inpatient cohort, p-Tau181 showed relatively better discriminative performance in the <65 years group, while GFAP and NfL showed numerically higher AUC values in the ≥65 years group. These exploratory findings suggest age-related differences in biomarker performance patterns that warrant further investigation in independent cohorts. The modest AUC (0.6212) for p-Tau181 in the younger group suggests that while it offers a statistical advantage, its clinical utility as a standalone screening tool remains limited, reinforcing the need for comprehensive clinical evaluation. Our results provide exploratory evidence for age-stratified minimally invasive serum-based assessment of clinically defined cognitive impairment in neurological inpatients, and cannot be directly generalized to community populations or used for definite Alzheimer’s disease pathological screening; rather, our results serve as a reference for age-stratified screening in clinical inpatient settings rather than standalone diagnostic criteria.