Amin Sonesh D, Chen Huanwen, Rewinkel Scott E, Lockwood David A, Kim Daniel, Priest Ryan A, Nesbit Gary M, Liu Jesse J, Horikawa Masahiro, Clark Wayne M, Laursen Rachel K, Jindal Gaurav, Chaturvedi Seemant, Colasurdo Marco
Department of Interventional Radiology, Oregon Health and Science University, Portland, OR 97239, USA; The Oregon Stroke Center at Oregon Health and Science University, Portland, OR 97239, USA.
Department of Neurology, MedStar Georgetown University Hospital, Washington DC 20007, USA.
Clin Neurol Neurosurg. 2025 Feb;249:108722. doi: 10.1016/j.clineuro.2025.108722. Epub 2025 Jan 6.
While endovascular thrombectomy (EVT) has become standard of care for patients' acute ischemic stroke (AIS) due to large vessel occlusion (LVO), many patients still suffer profound neurological disability, also termed futile recanalization (FR). The BAND score, which incorporates baseline disability, age, stroke severity, and treatment time window, is derived as a simple tool for upfront prediction of FR prior to EVT. This study aims to externally validate the BAND score and to incorporate upfront imaging biomarkers into the prediction tool.
Consecutive stroke thrombectomy patients with anterior circulation LVO who achieved successful recanalization (mTICI 2b or greater) were retrospectively identified at a single institution from 2019 to 2023. Clinical information, procedural details, and 90-day outcomes were recorded. The performance of the BAND score in predicting FR (90-day modified Rankin scale [mRS] >3) and loss of complete independence (LCID, 90-day mRS>2) was assessed. Then, Alberta stroke programme early CT score (ASPECTS) was added to create the PANDA score (pre-stroke disability, age, NIH stroke scale, delay from last known normal, and ASPECTS). The performance of PANDA to predict FR was assessed and compared with the original BAND score and also the widely validated THRIVE score.
296 patients were included; 36.5 % experienced FR. BAND had areas under the receiver-operating curve (AUCs) of 0.72 and 0.74 for predicting FR and LCID, respectively (both p < 0.001). The new PANDA score had AUCs of 0.76 and 0.78 for predicting FR and LCID, respectively (both p < 0.001), and it outperformed both BAND and THRIVE (all p < 0.05). Of the 30 patients (11.2 %) with high PANDA scores (≥7), 24 patients (80.0 %) suffered FR and 26 (86.7 %) suffered LCID.
This external validation study confirmed the adequate performance of BAND in predicting FR. The improved PANDA score performed better than the original BAND score and the widely validated THRIVE score.
