Association of Prethrombectomy Hemorrhagic Infarction on MRI With Further Transformation and Functional Outcomes: A Multicenter Study.

Abstract (English)

BACKGROUND AND OBJECTIVES: Mechanical thrombectomy (MT) is the standard of care for acute ischemic stroke from large vessel occlusion (LVO). Although postprocedural parenchymal hematoma (PH) predicts poor outcomes, pretreatment hemorrhagic infarction (HI), defined as petechial bleeding within infarcted tissue, is traditionally viewed as a benign radiologic finding. Its significance on sensitive preintervention MRI remains underexplored. We evaluated whether pretreatment MRI HI is associated with 90-day functional dependence, post-MT PH, and symptomatic intracerebral hemorrhage (SICH). METHODS: This retrospective cohort study included consecutive patients with anterior circulation LVO undergoing MT at 2 stroke centers (2015-2022). Patients required pre-MT T2*-weighted MRI. HI was graded (types I/II) per European Cooperative Acute Stroke Study III criteria in blinded fashion. The primary outcome was 90-day functional dependence (modified Rankin Scale 3-6). Secondary outcomes were post-MT ipsilateral PH and SICH. We used multivariable Bayesian and frequentist logistic regression to adjust for age, sex, NIH Stroke Scale, baseline disability, comorbidities, infarct volume, and procedural factors. RESULTS: Among 479 patients (median age 72 years; 46% female; 50% received IV thrombolysis), pretreatment HI was identified in 25 (5.2%), with excellent inter-rater agreement (Cohen &#x3ba; = 0.9). Functional dependence at 90 days occurred in 47% of the total cohort, ipsilateral PH in 9.4%, and SICH in 3.1%. In multivariable analysis, pre-MT HI was independently associated with 90-day functional dependence (Bayesian adjusted odds ratio [aOR] 6.66, 95% credible interval [CrI] 2.15-24.28, Pr[OR > 1] = 1.00; frequentist aOR 5.95, 95% CI 1.91-18.52, <i>p</i> = 0.002). Pre-MT HI also demonstrated a high posterior probability of association with post-MT ipsilateral PH (Bayesian aOR 2.76, 95% CrI 0.73-9.26, Pr[OR > 1] = 0.94; frequentist aOR 2.84, 95% CI 0.84-9.58, <i>p</i> = 0.091) and SICH (Bayesian aOR 5.59, 95% CrI 0.63-34.92, Pr[OR > 1] = 0.95; frequentist aOR 6.02, 95% CI 1.00-36.11, <i>p</i> = 0.049). DISCUSSION: HI detected on pretreatment MRI is an independent predictor of poor long-term functional outcomes and correlates with post-MT PH/SICH. Pre-MT HI may reflect significant blood-brain barrier disruption and microvascular fragility, serving as a marker of reperfusion vulnerability. Limitations include the retrospective design and the relatively low frequency of HI, PH, and SICH, which were addressed using Bayesian modeling to quantify uncertainty. Although not a contraindication to MT, identifying HI may aid in periprocedural risk stratification and serve as a candidate marker for future neuroprotective trials.