External validation of the R2eD AVM scoring system to assess rupture risk in pediatric AVM patients

Abstract

OBJECTIVE The assessment of hemorrhagic risk is of central importance in the management of pediatric brain arteriovenous malformations (AVMs). A recently published scoring system, the R2eD AVM (race, exclusive deep location or not, AVM size < 3 or ≥ 3, venous drainage exclusively deep or not, and monoarterial feeding or >1 arterial feeders) considers patient race and AVM location, size, venous drainage, and arterial supply and was successfully externally validated for use in adults to predict the risk of AVM hemorrhagic presentation. In this study the authors sought to externally validate the R2eD AVM score for prediction of hemorrhagic risk in the pediatric AVM population. METHODS A pediatric database at Boston Children’s Hospital was retrospectively reviewed for all patients with a diagnosis of intracranial AVM. Exclusion criteria included age > 21 years, multiple AVMs, and incomplete clinical/radiographic data. Demographic data and R2eD AVM score calculations were collected for all patients. Univariate binary logistic regression and multivariate stepwise backward elimination models were used to determine factors associated with hemorrhagic presentation. RESULTS A total of 212 AVMs were identified in 212 patients with vascular anomalies from 1995 to 2021, and 144 patients met the inclusion criteria (74 [51.4%] male, 111 [77.1%] White), of whom 87 (60.4%) patients presented with rupture and 122 (84.7%) underwent resection. The mean age at surgery was 12 years (range 3 months to 20 years). The R2eD AVM scoring system was applied to each patient. The score components of size < 3 cm and exclusive deep venous drainage were significantly associated with rupture risk (p < 0.05). The complete model for the R2eD AVM score had an area under the curve of 0.671 (95% CI 0.586–0.757). CONCLUSIONS This study demonstrated poor external validity of the R2eD AVM score in predicting pediatric AVM rupture risk. The results suggest that future studies are warranted to determine a better scoring method to capture pediatric rupture risk, given the significant differences in clinical presentation in pediatric compared with adult AVM patients.