Factors associated with vascular cognitive impairment in patients with chronic cerebral hypoperfusion and development of a competing-risk prediction model

Chronic cerebral hypoperfusion may precede cognitive decline, but the incremental predictive value of arterial spin labeling (ASL) remains uncertain. We evaluated clinical vascular cognitive impairment (VCI) risk while accounting for death before VCI. This retrospective study included 328 initially cognitively unimpaired patients with atherosclerotic cerebrovascular disease and hypoperfusion. Territory-specific ASL norms were estimated in 200 independent reference subjects. VCI was determined using multidomain testing, functional assessment, and International Society for Vascular Behavioural and Cognitive Disorders criteria. Fine–Gray models combined clinical variables, structural magnetic resonance imaging, and ASL measures. Internal validation used 1,000 bootstrap samples; final model forms and reference norms remained fixed. Paired comparisons assessed incremental performance. During follow-up, 79 patients developed VCI and 31 died before VCI. The 3-year VCI cumulative incidence was 23.8%. Older age, lower baseline Montreal Cognitive Assessment scores, greater perfusion deviation, a larger hypoperfusion volume proportion, and a higher total cerebral small vessel disease burden were associated with VCI. The optimism-corrected 3-year area under the receiver operating characteristic curve (AUC) was 0.805 (95% confidence interval [CI], 0.755–0.868). Adding ASL to clinical and structural imaging variables yielded an AUC difference of 0.0287 (95% CI, − 0.0092 to 0.0665; P = 0.138). The corrected 3-year calibration slope was 0.883. Memory impairment occurred in 60.8% of VCI cases; symptom-onset sequence was not established. The model showed internal discrimination, but the incremental value of ASL remained uncertain. Clinical VCI may reflect mixed pathology. Fixed-form internal validation, uncertain sample-size adequacy, limited events, and unmeasured amyloid and tau restrict interpretation; external validation and clinical impact assessment are required. Not applicable.