Abstract Body: Introduction: As more patients with stroke, including those with large cores, are treated with endovascular thrombectomy (EVT), understanding the pathophysiology of hemorrhagic transformation (HT) is becoming increasingly important. Pre-EVT infarct topography may have implications for treatment decisions acutely (e.g. stenting) and for post EVT care (e.g. antithrombotics and blood pressure goals). We sought to identify associations between HT and brain regions involved in ischemic lesions.

Methods: Consecutive patients with LVO treated with EVT who underwent pre-EVT MRI were identified from two tertiary referral centers (2011-2019). Acute ischemic lesions were extracted through a deep learning enabled pipeline from DWI and spatially normalized. Individual lesions were parcellated (atlas-defined 94 cortical regions, 14 subcortical nuclei, 20 white matter tracts) and reduced to ten essential lesion patterns using unsupervised dimensionality reduction techniques. HT, defined as ECASS PH1 or PH2, was modeled via Bayesian regression, taking the ten lesion patterns as inputs, and controlling for total lesion volume, age, sex, initial NIH Stroke Scale (NIHSS), thrombolysis treatment, good reperfusion (TICI2b-3), acute stenting, last known well-to-puncture time, diabetes mellitus, hypertension, coronary artery disease, smoking, atrial fibrillation, and site of enrollment.

Results: A total of 567 (mean age 69 ±15 years; 45% female) patients had pre-EVT DWI without significant artifacts that could undergo lesion segmentation and registration. The median NIHSS was 16 (IQR 11-20) and mean total infarct volume was 22.5 ±36.7ml. Thrombolysis was administered in 51% and good reperfusion was achieved in 83%. HT occurred in 10%. Lesion locations predictive of HT (Figure) involved bilateral caudate, putamen, pallidum, and anterior thalamic radiation; and, right more than left thalamus, corticospinal tract, and inferior fronto-occipital fasciculus (area under the curve: 0.73).

Conclusions: These data from a large, multicenter cohort with precise MRI-defined infarcts underscore the significance for HT of specific brain regions involved in ischemic lesions before EVT. An understanding of this pathophysiology can inform not only current clinical practice but also the development of future novel therapeutic strategies to prevent HT and reperfusion injury as more patients with large infarct cores are treated with EVT.