Risk Stratification for Early Neurological Deterioration After Mechanical Thrombectomy and Development of a 24-h...

Following mechanical thrombectomy for acute ischemic stroke due to anterior-circulation large-vessel occlusion, early neurological deterioration (END) continues to be a clinically consequential complication. Computed tomography perfusion (CTP)-derived mismatch volume estimates the amount of hypoperfused tissue that is not part of the ischemic core, but its value for END stratification at different time points has not been clarified. We therefore examined the relationship between mismatch volume and END and constructed risk models anchored to clearly specified temporal windows.

Consecutive patients treated with mechanical thrombectomy between January 2021 and December 2024 were retrospectively included. END encompassed either a ≥4-point increase from the preprocedural National Institutes of Health Stroke Scale (NIHSS) score or death from any cause during the first 72 h after thrombectomy; deaths occurring within 72 h were analyzed as END events. Potential nonlinearity was examined using restricted cubic spline (RCS) analysis. Only information known before thrombectomy was entered into the preprocedural baseline model. For the 24-h landmark analysis, patients who remained free of END through 24 h formed the reassessment cohort; successful recanalization status and hemorrhagic transformation subtypes were added to predict delayed END arising after >24-72 h. Discrimination, calibration, clinical utility, and internal validity were evaluated by receiver operating characteristic (ROC) analysis, calibration assessment, decision curve analysis (DCA), and 1000 bootstrap resamples, respectively.

END occurred in 87 of 438 patients (19.9%): 78 experienced neurological worsening and 9 died within 72 h. The association between mismatch volume and END departed from linearity (p for nonlinearity = 0.023). Observed END frequencies were 33.0%, 18.9%, and 12.5% for mismatch volumes <75 mL, 75-105 mL, and >105 mL, respectively (p for trend <0.001). Mismatch volume, ischemic core volume, admission NIHSS score, and collateral score constituted the preprocedural model, which produced an area under the ROC curve (AUC) of 0.789 (95% confidence interval [CI], 0.735-0.843) and a bootstrap-corrected AUC of 0.774. END occurred in 8.0%, 20.7%, and 38.5% of patients assigned to the low-, intermediate-, and high-risk categories of the simplified preprocedural score. Among 387 patients eligible for 24-h reassessment, 36 developed delayed END; the corresponding model yielded an AUC of 0.838 (95% CI, 0.773-0.903). Delayed END was associated with parenchymal hematoma but not with hemorrhagic infarction.

CTP mismatch volume was related to END following mechanical thrombectomy. The preprocedural and 24-h reassessment models, each defined by its prediction time, retained acceptable discrimination after internal validation; independent external validation remains necessary.