A New Scoring System for Angiographic Assessment of Embolotherapy Outcome of Peripheral Arteriovenous Malformations

Purpose: Studies evaluated angiographic response after arteriovenous malformation (AVM)  embolization based on the degree of devascularization in which the percentage of response did not consistently correlate with clinical outcome, and the overall therapeutic outcome was more dependent on clinical outcome assessment than angiographic response. Additionally, AVMs show variable degrees of AV shunting. We propose a new scoring system to predict therapeutic outcome based both on degree of nidus and shunt resolution.

Materials and Methods: We retrospectively reviewed the baseline and final angiograms of 17 patients (2012–2021) for changes in nidus occlusion and shunt changes, which were scored by two independent reviewers using a 7-point scale for each of the nidus and shunt. Final categorization of therapeutic response was based on cumulative scores of both components. Patient-reported average pain before and after embolization was scored on a 0 to 10 scale (with 10 representing maximum improvement). Patients also categorized their pain as improved, no change, or worse. Interobserver reliability was calculated using Cohen’s kappa coefficient. For measures of effect, Pearson correlation coefficient (R2) and Kruskal-Wallis rank-sum test were used. Scoring System Nidus Score Shunt Score Angiographic Response Worse -1 Worse -1 -1– -2 Aggravating No change 0 No change 0 0 No Change Residual >75% 1 Occlusion < 25% 1 1–4 Poor 50–75% 2 25–49% 2 5–7 Good 25–49% 3 50–75% 3 < 25% 4 >75% 4 8–9 Excellent 100% 5 100% 5 10 Cure

Results: R2 values between nidus and shunt were significant for both reviewers (0.93, 0.86; P < 0.001). Kappa values for nidus, shunt, and final numeric scores were 0.66, 0.41, 0.40, respectively (P < 0.01). Interobserver reliability for categorical final score was 84% (P = 0.0002). There was moderate correlation between nidus, shunt, final scores, and pain scores (R2 values: 0.50, 0.54, 0.52, respectively; P = 0.10, 0.07, 0.08, respectively; n = 12). Kruskal-Wallis values between average nidus, shunt, final scores, and categorical pain scores were 1.29, 1.21 and 1.43, respectively (P =0.53, 0.54, 0.49; n = 14).

Conclusions: Our proposed scoring system implements a more meticulous assessment of AVM structural details, including changes in both nidus and shunt. Although a limited cohort, correlation between average AVM scores and pain scores was moderate and trended toward significance. Further testing and correlation with clinical outcomes are necessary to validate this system.