Related Experiment Video
Updated: Jun 23, 2025
Author Spotlight: A Novel Approach to Cerebral Ischemia Modeling – Enhancing Reperfusion and Simplifying Procedure
Published on: May 31, 2024
External Validation of a Model for Persistent Perfusion Deficit in Patients With Incomplete Reperfusion After
Adnan Mujanovic1, Felix C Ng1, Mattia Branca1
1From the Departments of Diagnostic and Interventional Neuroradiology (A.M., E.I.P., T.D., J.G., J.K.), Neurology (T.R.M., M.A., D.J.S., U.F.), University Hospital Bern, Inselspital, Graduate School for Health Sciences (A.M.), and CTU Bern (M.B.), University of Bern, Switzerland; Department of Medicine and Neurology, Melbourne Brain Centre (F.C.N., N.Y., G.J.S., B.C.V.C.), Melbourne Medical School (L.C.), and Department of Radiology (P.J.M.), Royal Melbourne Hospital, University of Melbourne, Parkville; Department of Neurology (F.C.N.), Austin Health, Heidelberg, Australia; Division of Neuroradiology, Vascular and Interventional Radiology Department of Radiology (H.A.D., O.N., T.G., M.K.), and Department of Neurology (T.G., M.K.), Medical University of Graz, Austria; Population Health and Immunity Division (N.Y.), The Walter and Eliza Hall Institute of Medical Research, Parkville; Department of Neurology (M.W.P.), Liverpool Hospital, University of New South Wales, Sydney, Australia; Department of Radiology and Nuclear Medicine (F.C.), Amsterdam University Medical Centers, University of Amsterdam, Amsterdam Neuroscience, the Netherlands; Department of Neurology (T.J.K.), Royal Adelaide Hospital, Australia; and Department of Neurology (U.F.), University Hospital Basel, University of Basel, Switzerland.
Background And Objectives:
We recently developed a model (PROCEED) that predicts the occurrence of persistent perfusion deficit (PPD) at 24 hours in patients with incomplete angiographic reperfusion after thrombectomy. This study aims to externally validate the PROCEED model using prospectively acquired multicenter data.
Methods:
Individual patient data for external validation were obtained from the Endovascular Therapy for Ischemic Stroke with Perfusion-Imaging Selection, Tenecteplase versus Alteplase Before Endovascular Therapy for Ischemic Stroke part 1 and 2 trials, and a prospective cohort of the Medical University of Graz. The model's primary outcome was the occurrence of PPD, defined as a focal, wedge-shaped perfusion delay on 24-hour follow-up perfusion imaging that corresponds to the capillary phase deficit on last angiographic series in patients with
Results:
We included 371 patients (38% with PPD). The externally validated model had good discrimination (C-statistic 0.81, 95% CI 0.77-0.86) and adequate calibration (intercept 0.25, 95% CI 0.21-0.29 and slope 0.98, 95% CI 0.90-1.12). Across a wide range of probability thresholds (i.e., depending on the physicians' preferences on how the model should be used), the model shows net benefit on clinical decision curves, informing physicians on the likelihood of PPD. If a physician's attitude toward false-positive and false-negative ratings is equal, the model would reduce 13 in 100 unnecessary interventions by correctly predicting complete delayed reperfusion, without missing a patient with PPD.
Discussion:
The externally validated model had adequate predictive accuracy and discrimination. Depending on the acceptable threshold probability, the model accurately predicts persistent incomplete reperfusion and may advise physicians whether additional reperfusion attempts should be performed.
