Intracranial Vessel Perforation During Mechanical Thrombectomy for Acute Ischemic Stroke: Incidence, Management, and Outcomes — A Systematic Review and Meta-Analysis

BACKGROUND AND PURPOSE: Intracranial vessel perforation — DSA-confirmed contrast extravasation during mechanical thrombectomy — carries approximately 50% ninety-day mortality. Its incidence, procedural characteristics, hemostatic management, and clinical outcomes have not been systematically quantified. MATERIALS AND METHODS: A systematic review and meta-analysis (PROSPERO CRD420261385744; PRISMA 2020) searched seven databases up to 15 May 2026. Incidence was pooled using REML random-effects meta-analysis of logit-transformed proportions with Hartung-Knapp-Sidik-Jonkman confidence intervals; procedural characteristics and management were synthesized narratively. Ninety-day mortality and mRS 0-2 were pooled (k=2, N=475). Risk of bias (Risk of Bias In Non-randomized Studies of Interventions) and certainty of evidence (Grading of Recommendations Assessment Development and Evaluation) were assessed. RESULTS: Fourteen studies were included (primary incidence pool: k=11, N=40,656, 655 vessel perforation events). Vessel perforation incidence was 2.05% (95% CI 1.43%–2.93%; 95% PI 0.64%–6.34%; I²=92.6%). Medium-vessel occlusion thrombectomy showed a higher observed incidence than large-vessel occlusion thrombectomy (3.68% vs 1.52%), but this comparison was based on three medium-vessel cohorts with substantial heterogeneity and should be interpreted as hypothesis-generating. Quantitative pooling of procedural data was not feasible; narrative synthesis identified device navigation beyond the occlusion and stent retriever retraction as principal precipitating steps, while wireless microcatheter navigation was the only technique evaluated through a direct within-study comparison of angiographically detected hemorrhagic complications (single-center; Serious risk-of-bias assessment). Five hemostatic strategy categories were identified; in one study, successful reperfusion was associated with functional independence in 60% versus 12.5% (p=0.006). Pooled 90-day mortality was 52.2% (95% CI 47.7%–56.7%; I²=0.0%) and mRS 0-2 was 16.7% (95% CI 13.6%–20.3%); both were pooled from two independent datasets but were statistically dominated by one large cohort. Certainty of evidence was Low for incidence and mortality and Very Low for MeVO-specific incidence, mRS 0-2, procedural characteristics, and management. CONCLUSIONS: Intracranial vessel perforation complicates approximately 1 in 49 thrombectomy procedures on average, with wide between-study variation, and carries approximately 52% ninety-day mortality. Medium-vessel occlusion thrombectomy showed a higher observed incidence than large-vessel occlusion thrombectomy, a hypothesis-generating signal based on three studies. Navigation technique emerged as a potential preventive target, but the supporting evidence is limited to a preliminary, single-center observation requiring prospective validation; comparative hemostatic-management data are also needed.