Validation and branch involvement-based refinement of prediction scores for incomplete aneurysm occlusion after flow diversion

Background FDSS, DIANES, 4F-FPS, and ABC are scoring models developed to predict incomplete occlusion of intracranial aneurysms after treatment with flow diverters (FDs). Methods We retrospectively included consecutive patients with intracranial aneurysms treated with FDs at our center between January 2020 and December 2025 with at least 6 months of angiographic follow-up. Predictors of incomplete occlusion were analyzed by univariable analysis followed by multivariable Firth penalized logistic regression. The predictive performance of the four scores was evaluated. Modified scores (mFDSS, mDIANES, and mABC) were generated by excluding the ophthalmic artery (OphA) and adult-type posterior communicating artery (aPComA) from the variable ‘side branch arising from the aneurysm neck/dome’ and were compared with the original scores. Results The study included 152 patients harboring 185 aneurysms, 86.49% of which were located in the internal carotid artery (ICA), with an incomplete occlusion rate of 13.51%. Male sex and branch involvement were independent predictors of incomplete occlusion (aOR, 3.67 and 3.81, respectively). Area under the curve (AUC) for FDSS, DIANES, 4F-FPS, and ABC were 0.715, 0.741, 0.539, and 0.732, respectively. Excluding the OphA and aPComA from branch involvement improved the accuracy of mFDSS, mDIANES, and mABC compared with the original scores. Conclusion FDSS, DIANES, and ABC showed fair predictive performance, whereas 4F-FPS had limited predictive value in this predominantly ICA saccular aneurysm cohort. Branch involvement was the strongest predictor of incomplete occlusion. The exploratory refinement of branch involvement by excluding the OphA and aPComA improved score accuracy. The rich collateral circulation of these branches could be the potential explanation but this theory requires further verification.