Predictive value of CT imaging features on the risk of hemorrhagic transformation after mechanical thrombectomy for acute ischemic stroke with large vessel obstruction.

Abstract

Objective: To investigate the predictive value of computer tomography (CT) imaging features for the risk of hemorrhagic transformation (HT) after mechanical thrombectomy for acute ischemic stroke with large vessel obstruction (AIS-LVO).

Methods: A total of 135 patients with AIS-LVO diagnosed and treated in our hospital from August 2021 to May 2023 were selected as the research subjects. Their clinical data were retrospectively analyzed. Mechanical thrombectomy was performed in all patients. The patients were divided into the HT group (n = 27) and the non-HT group (n = 108) according to whether HT occurred within 24 h after thrombectomy. CT examination was performed after mechanical thrombectomy in the two groups, and the changes in CT imaging indexes in the two groups were observed. Logistic regression was used to analyze the influencing factors and a prediction model was constructed based on the influencing factors. The receiver operating characteristic (ROC) curve was established to analyze the predictive value. Additionally, ROC curve was used to analyze the diagnostic value of serum CT imaging features.

Results: Compared with the non-HT group, the proportion of atrial fibrillation history in the HT group was significantly increased, and the National Institute of Health Stroke Scale (NIHSS) score and galectin-3 (Gal-3) level were significantly increased before thrombectomy (P < 0.01). Compared with the non-HT group, the proportion of exudation of contrast medium and Hyperdense Middle Cerebral Artery Sign (HMCAS) in the HT group was significantly increased, time to peak (TTP) was significantly prolonged, and cerebral blood flow (CBF) was significantly decreased (P < 0.001). The history of atrial fibrillation, NIHSS score before thrombectomy, Gal-3, contrast agent exudation, HMCAS, TTP and CBF were the influencing factors of postoperative HT after mechanical thrombectomy in AIS-LVO (P < 0.05). Based on the results of multivariate logistic regression analysis, a prediction model was established as follows: Logit (P) = -3.520 + 1.529 × history of atrial fibrillation + 0.968 × NIHSS score before thrombectomy + 0.806 × Gal-3 + 1.134 × contrast agent exudation + 2.146 × HMCAS + 0.684 × TTP-0.725 × CBF. The area under the curve (AUC) of the logistic prediction model for predicting HT after AIS-LVOLVO mechanical thrombectomy was 0.873 (95% CI 0.817-0.929) with a sensitivity of 78.75% and a specificity of 83.33%, indicating that the prediction model had good prediction efficiency. The AUC of TTP and CBF alone in predicting HT after mechanical thrombectomy in AIS-LVO patients was 0.728 and 0.736, respectively. The AUC of combined detection was 0.783, and the combined detection had a high diagnostic value for HT after mechanical thrombectomy in AIS-LVO patients.

Conclusion: The combined detection of TTP and CBF of CT imaging features had certain diagnostic value for HT in AIS-LVO patients after mechanical thrombectomy. The logistic prediction model based on these influencing factors had a high diagnostic value for HT after mechanical thrombectomy.