Predictors for hemorrhagic transformation and cerebral edema in stroke patients with first-pass complete recanalization

Abstract

Background: Predictors of radiological complications attributable to reperfusion injury remain unknown when baseline setting is optimal for endovascular treatment and procedural setting is the best in stroke patients with large vessel occlusion (LVO). Aims: To identify clinical and radiological/procedural predictors for hemorrhagic transformation (HT) and cerebral edema (CED) at 24 hr in patients obtaining complete recanalization in one pass of thrombectomy for ischemic stroke ⩽ 6 h from symptom onset with intra-cranial anterior circulation LVO and ASPECTS ⩾ 6. Methods: We conducted a cohort study on prospectively collected data from 1400 patients enrolled in the Italian Registry of Endovascular Treatment in Acute Stroke. Results: HT was reported in 248 (18%) patients and early CED was reported in 260 (19.2%) patients. In the logistic regression model including predictors from a first model with clinical variables and from a second model with radiological/procedural variables, diabetes mellitus (odds ratio (OR) = 1.832, 95% confidence interval (CI) = 1.201–2.795), higher National Institutes of Health Stroke Scale (NIHSS) (OR = 1.076, 95% CI = 1.044–1.110), lower Alberta Stroke Program Early CT (ASPECTS) (OR = 0.815, 95% CI = 0.694–0.957), and longer onset-to-groin time (OR = 1.005, 95% CI = 1.002–1.007) were predictors of HT, whereas general anesthesia was inversely associated with HT (OR = 0.540, 95% CI = 0.355–0.820). Higher NIHSS (OR = 1.049, 95% CI = 1.021–1.077), lower ASPECTS (OR = 0.700, 95% CI = 0.613–0.801), intravenous thrombolysis (OR = 1.464, 95% CI = 1.061–2.020), longer onset-to-groin time (OR = 1.002, 95% CI = 1.001–1.005), and longer procedure time (OR = 1.009, 95% CI = 1.004–1.015) were predictors of early CED. After repeating a fourth logistic regression model including also good collaterals, the same variables remained predictors for HT and/or early CED, except diabetes mellitus and thrombolysis, while good collaterals were inversely associated with early CED (OR = 0.385, 95% CI = 0.248–0.599). Conclusions: Higher NIHSS, lower ASPECTS, and longer onset-to-groin time were predictors for both HT and early CED. General anesthesia and good collaterals were inversely associated with HT and early CED, respectively. Longer procedure time was predictor of early CED.