Internal Construct Assessment of the Perfusion Collateral Impairment Score: Component Independence and Outcomes Discrimination in LVO Stroke

BACKGROUND: The Perfusion Collateral Impairment Score (PCIS), an ordinal composite (range 0-3) of hypoperfusion intensity ratio (HIR ≥ 0.4), cerebral blood volume index (CBVI < 0.8), and prolonged venous transit (PVT+), was associated with poor 90-day outcomes in large vessel occlusion (LVO) stroke. Whether it outperforms its components remains untested. METHODS: We analyzed 283 anterior circulation LVO patients from two centers. Ordinal PCIS was compared with each component using DeLong testing for unfavorable outcome (modified Rankin Scale [mRS] 3-6) at 90 days, with Bonferroni correction (k = 3). Equivalence was tested by TOST (Δ = 0.10; sensitivity Δ = 0.05). Calibration, reclassification, decision curve analysis, multivariable adjustment, and subgroup analyses were performed. RESULTS: Unfavorable outcome occurred in 148 (52.3%). Ordinal PCIS (AUC, 0.61) was numerically higher than HIR ≥ 0.4, CBVI < 0.8, and PVT+ (AUC 0.56-0.57); ΔAUC vs HIR and CBVI was 0.05 (P = 0.04 uncorrected; P = 0.13 Bonferroni). Components were equivalent at Δ = 0.10 but not Δ = 0.05; VIFs 1.31-1.72. Calibration was good (slope 1.00; Hosmer-Lemeshow P = 0.83). PCIS retained independent association on multivariable adjustment (aOR, 1.41; 95% CI, 1.03-1.92; P = 0.03). CONCLUSIONS: Ordinal PCIS showed marginal AUC gain over its components and good calibration. Absolute discrimination was modest; external validation is needed.