Molecular biomarkers of cognitive impairment and neuropsychiatric symptoms after stroke in late-life-enriched cohorts: a scoping review

Post-stroke cognitive impairment (PSCI) and neuropsychiatric symptoms (NPS) are frequent in late-life stroke populations and contribute to long-term disability and care needs. Molecular biomarkers may enable earlier risk stratification and trial enrichment, but the literature remains fragmented across biomarker classes, sampling windows, and phenotype definitions. We conducted a systematic scoping review following PRISMA-ScR and JBI guidance. PubMed, Embase, Web of Science, Scopus, the Cochrane Library, and EBSCOhost were searched from inception to 29 January 2026. Eligible cohorts were defined by an explicit minimum age threshold of ≥ 60 years or a cohort mean/median age ≥ 60 years. To strengthen translational interpretability, the final synthesis additionally used an explicit phenotyping framework for PSCI and NPS, a NOS-informed methodological quality appraisal, a consistency analysis of core biomarker families (NfL, GFAP, and hs-CRP/CRP), a China-versus-non-China subgroup analysis, and a shared/specific biomarker mapping. The evidence map comprised 105 studies, predominantly observational, acute-phase, and blood-based. Age eligibility was predominantly indirect: 5/105 studies used an explicit minimum age threshold of ≥ 60 years, whereas 100/105 qualified on cohort mean/median age ≥ 60 years. The literature was geographically concentrated in China (78/105), and methodological quality was uneven, with 86 studies in the higher-quality stratum, 14 in the moderate stratum, and 5 in the lower stratum. NfL (5/5) and GFAP (2/2) showed uniformly positive adjusted associations, mainly in prognostic analyses, whereas CRP-family findings were more heterogeneous (5/7 positive, 1 mixed/attenuated, 1 null). China studies were more often acute, prospective, and PSCI-oriented, whereas non-China studies more often examined inflammation-linked NPS and later-phase phenotypes. Shared biomarker candidates across PSCI and NPS were most evident for NfL and GFAP, while CRP-family markers were shared but more confounding-sensitive. Biomarker research for PSCI and NPS in late-life-enriched stroke cohorts is rapidly expanding but remains heterogeneous in phenotyping, timing, confounder structure, and reporting quality. Harmonized outcome definitions, quality-stratified interpretation, phase-aware serial sampling, and externally validated prediction frameworks with calibration and utility evaluation are priorities for biomarker qualification and clinical implementation. The resulting evidence map is most directly applicable to late-life-enriched rather than strictly age-restricted ≥ 60-year stroke populations.