ABSTRACT NUMBER: ESOC2026A1884 ASSOCIATION OF HIGH PLASMA TUMOUR NECROSIS FACTOR-ALPHA AT ADMISSION WITH FUNCTIONAL OUTCOME IN PATIENTS WITH POSTERIOR CIRCULATION LARGE-VESSEL OCCLUSION STROKE UNDERGOING MECHANICAL TROMBECTOMY

Abstract Background and aims Posterior circulation large-vessel occlusion (PC-LVO) is associated with high morbidity and mortality. This study aimed to determine whether admission tumour necrosis factor (TNF)-alpha levels are associated with functional outcomes in PC-LVO patients undergoing thrombectomy. Methods Admission plasma TNF-alpha concentrations were measured in 71 PC-LVO patients from a national stroke thrombectomy cohort, using enzyme-linked immunosorbent assay (ELISA). Functional outcomes at 3 months were assessed using the modified Rankin Scale (mRS) categorised as good (mRS≤2) or favourable (mRS≤3). Binary and multivariate logistic regression analyses were performed. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC-ROC). Results Patients with poor 3-month outcomes had significantly higher TNF-alpha levels than those with good and favourable outcomes (3.53 [2.32 – 5.30] vs. 2.64 [1.98 – 3.78]pg/mL, p = 0.029, and 4.30 [2.90 – 5.70] vs. 2.60 [1.99 – 3.72]pg/mL, p = 0.001). Significant variables associated with poor outcomes in univariate analysis were baseline NIHSS (aOR: 1.075 [1.018 – 1.136], p = 0.010) and TNF-alpha (aOR: 1.381 [1.008 – 1.891], p = 0.044) for mRS≤2 dichotomisation, and NIHSS (aOR: 1.109 [1.044 – 1.179], p = 0.001), baseline systolic blood pressure (aOR: 1.019 [ 1.000 – 1.037], p = 0.047) and TNF-alpha (aOR: 1.685 [1.166 – 2.435], p = 0.005) for mRS≤3 dichotomisation. Multivariable models incorporating these variables achieved AUCs of 0.757 and 0.892, with positive predictive values of 80.7% and 82.3%, respectively. Conclusions Higher admission TNF-alpha levels are independently associated with poor 3-month functional outcomes in PC-LVO patients undergoing thrombectomy, suggesting its potential as a prognostic biomarker. Conflict of interest All authors: nothing to disclose