Does variability in automated perfusion software outputs for acute ischemic stroke matter? Reanalysis of EXTEND perfusion imaging

Aims: We reprocessed the Extending the time for Thrombolysis in Emergency Neurological Deficits (EXTEND) perfusion imaging with a different automated software with the aim of comparing mismatch eligibility and outcomes. Methods: EXTEND baseline perfusion imaging data were reprocessed using autoMIStar software to identify patients who were eligible based on the same target mismatch criteria as per the original trial. Results: From the 225 patients fulfilling RAPID-based mismatch criteria randomized in the EXTEND study, 196 (87%) patients met the revised mismatch criteria. Most common reasons for not meeting revised criteria were core >70 ml (n = 9), and no perfusion lesion/lack of penumbral tissue (n = 20). The revised perfusion lesion volumes were significantly smaller compared to the original RAPID volumes (median 68 ml IQR 34–102 ml vs. 42 ml 16–92 ml, p = 0.036). Of the patients who met the revised mismatch criteria, 40% receiving alteplase had modified Rankin Scale (mRS) 0–1 at 3-month compared to 28% with placebo (Adjusted Odds Ratio (OR) = 2.23, CI 1.08–4.58, p = 0.028). In contrast, in the original mismatch cohort, 35% receiving alteplase had mRS 0–1 at 3-month compared to 30% with placebo (adjusted OR = 1.88, p = 0.056). Conclusions: These data reinforce the benefit of alteplase in the later time window, and suggest that differences in automated perfusion imaging software outputs may be clinically relevant.