Comparison of Grading Scales Regarding Perioperative Complications and Clinical Outcomes of Brain Arteriovenous Malformations After Endovascular Therapy—Multicenter Study

Several scales have been proposed for risk assessment and outcome determination in brain arteriovenous malformations treated by endovascular therapy. We aim to validate and compare the efficacy of these scales in predicting perioperative complications and clinical outcomes. We retrospectively reviewed brain arteriovenous malformations patients who underwent endovascular therapy at 4 centers in China from January 2012 to December 2015. The primary outcomes were complications, unfavorable outcome (mRS ≥ 3), and complete obliteration. Each patient was assessed using the Spetzler-Martin grading system (SM), Puerto Rico scale, Buffalo score, and arteriovenous malformation embocure score (AVMES). Correlation analysis was performed between primary outcomes incidence rate and the grades of each scale. The area under the receiver operating characteristic curve of these scales was calculated. Pairwise comparison of receiver operating characteristic curves was performed to compare the efficacy of the scales. A total of 270 patients were included. Correlation analysis demonstrated that the complication rate increased with increasing grade in SM (P = 0.002), Puerto Rico scale (P = 0.014), and Buffalo score (P = 0.001); complete obliteration rate decreased with increasing grade in AVMES (P = 0.017); unfavorable outcome rate increased with increasing grade in the Puerto Rico scale (P = 0.005). The area under the receiver operating characteristic curve analysis showed statistical differences between the Puerto Rico score and SM (P = 0.047) in predicting complications and between the Puerto Rico score and SM (P = 0.008) in predicting unfavorable outcomes. The area under the curve of the AVMES in predicting complete obliteration was 0.757. The Puerto Rico score predicts complications and unfavorable outcomes better than the SM. The AVMES scale has medium efficacy in predicting complete obliteration.