Quantifying procedural burden during stent-assisted coil embolization for unruptured intracranial aneurysms: development of the aneurysm endovascular procedural complexity score and internal validation of a multivariable complication model

Procedural complexity during stent-assisted coil embolization (SAC) is commonly assessed subjectively and is not adequately represented by conventional pre-procedural risk factors. This single-center retrospective cohort study developed the Aneurysm Endovascular Procedural Complexity Score (AEPCS) and internally validated a multivariable complication model incorporating the score in 484 patients with unruptured intracranial aneurysms treated with SAC between January 2018 and December 2025. The AEPCS integrated procedure duration, contrast volume, and the system-reported procedure-level radiation index (mGy) after natural-log transformation, Z -standardization, equal-weight averaging, and final standardization. The primary outcome was a composite of symptomatic cerebral infarction, symptomatic intracranial hemorrhage, or all-cause death within 28 days. Major periprocedural complications occurred in 71 patients (14.7%). After adjustment for age, sex, hypertension, diabetes mellitus, posterior-circulation location, aneurysm diameter, neck width, and multiple aneurysms, each 1-standard-deviation increase in the AEPCS was associated with higher odds of a major complication [adjusted odds ratio (OR), 5.25; 95% confidence interval (CI), 3.51–7.86; p < 0.001]. The AEPCS showed an area under the receiver operating characteristic curve (AUC) of 0.846 (95% CI, 0.797–0.890); its discrimination was significantly better than that of procedure duration alone but did not differ significantly from the radiation index or contrast volume. In a sensitivity analysis that reincluded 33 patients with definite intraprocedural complications requiring rescue treatment, the association remained materially unchanged (adjusted OR, 5.31; 95% CI, 3.55–7.93; p < 0.001), with an AUC of 0.842. The multivariable model had an apparent AUC of 0.853 and an optimism-corrected AUC of 0.834 (95% CI, 0.795–0.890) after 2,000 full-pipeline bootstrap resamples. The AEPCS should therefore be regarded primarily as a standardized composite measure of cumulative procedural burden rather than an outcome-optimized prediction score. External validation is required before clinical implementation.