Abstract Background and aims Alteplase-based intravenous thrombolysis is a standard therapy for acute ischemic stroke, but evidence indicates it may heighten post-stroke inflammation through neutrophil activation. This study investigated whether circulating neutrophils act as mediators linking IVT to haemorrhagic transformation in patients with anterior circulation large-vessel-occlusion stroke. Methods We conducted a multicentre observational study including patients with anterior circulation LVO treated with mechanical thrombectomy, with or without IVT. Inverse probability weighting was applied to balance baseline characteristics. LASSO regression models, stratified by IVT status, identified laboratory predictors of HT. Causal mediation analysis assessed whether 24-hour neutrophil count mediated the effect of IVT on HT risk. Results Our analysis included 970 patients with acute ischemic stroke due to anterior circulation LVO treated with mechanical thrombectomy. Among them, 646 (66.6%) received IVT (alteplase), while 324 (33.4%) underwent thrombectomy alone. In the IVT group, 24-h neutrophil count independently predicted risk of HT(OR 1.15;95%CI 1.08–1.24;p0.001). Causal mediation analysis demonstrated that 24-h neutrophil count significantly mediated the effect of IVT on haemorrhagic transformation (average causal mediation effect ACME: 0.24;p0.001) accounting for a substantial proportion of the total effect. No such associations were observed in non-IVT patients. Conclusions In LVO stroke, neutrophil-driven inflammation may partly mediate the increased risk of haemorrhagic transformation after alteplase. Although confirmation is needed, these findings suggest that modulating inflammatory responses following thrombolysis could offer a promising therapeutic strategy to improve outcomes in this patient population. Conflict of interest SB is a key opinion leader for RAPIDAI. TNN is Associate Editor of Stroke; advisory board for Brainomix, Aruna Bio; speaker for Genentech, Kaneka, consultant for Medtronic. All other authors have no conflicts of interests Figure 1 - belongs to Results
Prandin et al. (2026) studied this question.