Establishment and validation of a scoring model for predicting the recurrence risk after endovascular embolization of intracranial aneurysms / 中国脑血管病杂志

ABSTRACT

Objective To establish a comprehensive,simple,and effective scoring model for predicting the recurrence risk after endovascular embolization of intracranial aneurysms in order to assess the possibility of recurrence and to provide guidance for the selection of surgical protocols and postoperative management.Methods From May 2012 to May 2014,434 patients (441 aneurysms) with intracranial aneurysm treated with endovascular embolization at the Department of Neurosurgery,Changhai Hospital,the Second Military Medical University were enrolled retrospectively,and they were used as a modeling group.After modeling,109 patients (109 aneurysms) were used as a validation group.In the modeling cohort,a predictive scoring model of recurrence risk was established according to the results of multivariate logistic regression analysis;the model was validated in the validation cohort.According to the scoring model of the modeling group,the scoring table of best cut-off value of the receiver operating characteristic (ROC) curves was divided into a low-risk and a high-risk of recurrence.The recurrence risk score model was compared with the North America aneurysm recanalization stratification scale (ARSS) model,and Raymond grade.Results Multivariate logistic regression analysis showed that the 3 factors included in the scores and finally,a established scoring model of recurrence risk prediction were non-stent assisted embolization (1 point),Raymond grade ≥Ⅱ (1 point),and the size of aneurysm (aneurysm >25 mm[3 points)],aneurysm 10-25 mm[1 point],and aneurysm <10 mm[0 point]).The validation indicated that the scoring system had higher predictive value (AUC=0.738,95%CI 0.641-0.834,P<0.05) and goodness of fit (Hosmer-Lemeshow χ2=2.109,P=0.146).The scoring table was further divided into the low-risk recurrence (0-1 point) and high-risk recurrence (2-5 points),its sensitivity was 72.73% (48/66) and specificity was 68.80% (258/375).The predictive ability of the aneurysm recurrence risk score model was similar to that of the ARSS score (χ2=0.54,P=0.462),and it was better than the Raymond grade (χ2=15.10,P<0.01).Conclusion The established simple aneurysm recurrence risk predicting score model in this study may accurately predict the recurrence of aneurysms,however,a multicenter,large sample prospective study is needed for further validation.