Abstract DP337: Modification and optimization of the PANDA score with laboratory values for upfront prognostication of post-thrombectomy functional outcomes

Background: Despite its proven effectiveness, a substantial proportion of patients undergoing endovascular thrombectomy (EVT) for large vessel occlusion (LVO) stroke still suffer poor neurological outcomes despite successful reperfusion, also termed futile recanalization (FR). Prior work has established the PANDA score, which incorporates pre-stroke disability, age, NIH stroke scale, delay from time of last known well, and Alberta Stroke Programme Early Computed Tomography Score (ASPECTS), to be a reliable predictor of FR. In this study, we explored whether addition of laboratory markers can further improve the performance of PANDA. Methods: This was a retrospective analysis of EVT patients at a single Comprehensive Stroke Center. Patients with anterior circulation LVO and successful recanalization (mTICI 2b or better) were included, and the PANDA score was calculated. The primary outcome was FR, defined as a 90-day modified Rankin scale ≥4. Admission laboratory values (white blood cell [WBC], platelets, glucose, creatinine, troponin) were collected. Patients missing PANDA components were excluded. Each laboratory value was screened for association with FR in bivariate analysis; significant predictors underwent decision tree analysis to define cut-points. Point values were assigned, where each point represented approximately10% difference in FR, as in the initial PANDA derivation. Finally, scored laboratory values were added to PANDA to create PANDA+, and performance was assessed using area under the receiver operating characteristic curve (AUC) and compared with the original score. Results: A total of 465 patients were identified. Mean age was 67 years, 51.4 percent were male, mean NIH stroke scale was 15.7, and mean PANDA score was 3.5. Overall, PANDA performed well in predicting FR, with an AUC of 0.74 (95%CI 0.70-0.79). Bivariate analyses showed that WBC and Glucose were significant predictors of FR, and decision tree analysis suggested assigning 1 point for glucose 150 or higher, and 2 points for glucose 200 or higher, as well as assigning 1 point for WBC 8 or higher, and 3 points for 16 or higher. Altogether, the PANDA plus score (range -1 to 12) achieved an AUC of 0.77 (95%CI 0.72-0.82), which significantly outperformed the original PANDA score (p=0.023). Conclusions: The PANDA+ score, which adds WBC and glucose levels to the previously validated PANDA score, outperformed the original PANDA score in predicting FR using only pre-EVT information.