Predictors of Acute Neurological Worsening after Endovascular Thrombectomy

Abstract

Background: Successful reperfusion after endovascular thrombectomy (EVT) correlates with good outcome. However, radiographic reperfusion does not always translate into good clinical outcomes even if the reperfusion occurs early after the stroke onset. Reasons for neurological worsening (NW) are thought to be many, such as progression of the stroke, hemorrhagic conversion post tissue plasminogen activator and/or EVT, and procedural complications such as vessel dissection or perforation, distal emboli, and re-occlusion. Data on patients worsening in the acute phase after EVT are limited. Objective: We studied the factors associated with acute NW and also identified the predictors of NW after EVT and its association with poor outcome at discharge. Methods: A retrospective cohort from a single comprehensive stroke center includes patients with acute ischemic stroke and large vessel occlusion in anterior and posterior circulation who presented between December 2014 and May 2017 and received EVT were reviewed. Primary outcome was defined as acute NW defined as change in NIHSS ≥4 from baseline in the first 24 h after EVT. Secondary outcome were modified Rankin scale (mRS) 0–2 at discharge and final infarct volume. Univariate and multivariate analyses were performed to evaluate clinical and radiographic variables independently correlating with NW after EVT. Receiver operating curve analysis was also performed to identify predictors. Results: 178 patients were included in the analysis, 26 (14.7%) met the criteria for acute NW. For these 178 patients, the median age was 63 (IQR 53–74, range 26–89), baseline median NIHSS was 19 (IQR 14–24, range 5–37), ASPECTS was 8 (IQR 7–9, range 4–10), admission median systolic blood pressure (SBP) was 150 (IQR 131–170, range 94–287), and initial median blood glucose (BG) was 123 (IQR 106–157, range 69–433). The most common reasons for worsening were progression of the stroke (42.3%) and reperfusion injury PH-2 (26.9%) (p < 0.0001). Univariate logistic analysis showed that race, ASPECTS, collateral score, diabetes mellitus, admission SBP, and admission BG were associated with acute NW. In multivariate analysis, only admission BG (OR 1.00, CI 1.00–1.01, p = 0.04) was found to have a significant association with acute NW. We ran a prediction analysis for variables and found the area under the curve to be 0.75. Finally, there was strong association between NW and poor outcome at discharge (MRS 3–6, p < 0.01) by Fisher’s exact test. About 46.1% in the NW group died during hospitalization compared to 10% in the non-NW group (p < 0.0001). Conclusion: Our single-center retrospective cohort result is limited by small sample size. It showed that high admission BG is an independent predictor of NW after EVT and ultimately leads to poor outcome.