BACKGROUND: The Perfusion Collateral Impairment Score (PCIS) was linked to 90-day outcomes in patients with large vessel occlusion (LVO) stroke, but its incremental prognostic value over other standard imaging parameters has not been examined. METHODS: We evaluated 283 anterior circulation LVO patients. We compared the PCIS with Tan, COVES, ASPECTS, relative cerebral blood flow less than 30% volume, and mismatch volume using paired DeLong testing with Bonferroni correction (k = 3). Incremental value was further assessed by nested logistic regression (unadjusted and age + NIHSS-adjusted), net reclassification improvement (NRI) and integrated discrimination improvement (IDI), and decision curve analysis. The primary outcome was mRS 3–6 at 90 days. RESULTS: Among 283 LVO patients, there were 148 (52.3%) with unfavorable 90-day outcomes. Addition of ordinal PCIS to each of the base imaging models was independently associated with mRS 3–6 (Tan + PCIS aOR, 1.45; 95% CI, 1.10–1.91; P = 0.008; COVES + PCIS aOR, 1.53; 95% CI, 1.17–2.01; P = 0.002; ASPECTS + PCIS aOR, 1.49; 95% CI, 1.14–1.94; P = 0.003), and neither Tan nor COVES alone maintained significance in any of the extended models. With age + NIHSS adjustment, only the COVES + PCIS model retained an independent association with outcome (aOR, 1.37; 95% CI, 1.02–1.85; P = 0.039), with PCIS only approaching significance in the Tan + PCIS (aOR, 1.29; 95% CI, 0.96–1.74; P = 0.10) and ASPECTS + PCIS (aOR, 1.26; 95% CI, 0.94–1.70; P = 0.12) models. NRI and IDI were significant for all 5 pairs (all P ≤ 0.014). Standalone PCIS AUC did not exceed comparator AUCs after Bonferroni correction. CONCLUSIONS: The PCIS offers additional information beyond CTA collateral scores and standard imaging measures and may assess unique hemodynamic properties, though its individual discrimination was modest. Future studies can explore how to optimally incorporate the PCIS as a hemodynamic adjunctive imaging parameter with other established stroke imaging techniques.
Biswas et al. (Sat,) studied this question.