Generalizability of the EMBOLISE, MAGIC-MT, STEM, EMPROTECT, and MEMBRANE trials of middle meningeal artery embolization for chronic subdural hematoma: A single-center cohort analysis.

OBJECTIVE: The pivotal EMBOLISE, MAGIC-MT, and STEM randomized trials (RCTs) established middle meningeal artery embolization (MMAE) for chronic subdural hematoma (cSDH) but used differing criteria of uncertain real-world representativeness, and two further RCTs (EMPROTECT, MEMBRANE) have since reported. We quantified how many recipients met each trial's criteria, which criteria drove ineligibility, and whether outcomes differed between eligible and ineligible patients.

METHODS: We retrospectively analyzed 214 consecutive single-center MMAE patients, classifying each as eligible or ineligible for all five trials by published criteria (bilateral hematomas judged by worst side). Outcomes were recurrence, rescue surgery, complications, radiologic resolution, and functional outcome (modified Rankin Scale, mRS); eligible-versus-ineligible outcomes were compared directly with Fisher exact tests (Benjamini-Hochberg/Holm correction), favorable outcome was modeled with multivariable logistic regression, and agreement used Cohen's kappa.

RESULTS: Among 214 patients (mean age 73.4 years; 29.0% bilateral; 25.7% anticoagulated; 69.2% standalone MMAE), eligibility was 13.5% for EMBOLISE, 24.7% for MAGIC-MT, 17.7% for STEM, 6.1% for EMPROTECT, and 78.9% for MEMBRANE; the two newer trials bracketed rather than resolved the generalizability gap, EMPROTECT being the most restrictive and MEMBRANE the most permissive. Across the three original trials, 142 (66.4%) met none; across all five, 60 (28.0%) met none and only 1 met all five. Pairwise agreement was minimal except MAGIC-MT-versus-STEM (kappa 0.57). Recurrence (16.8%), rescue surgery (16.8%), complications (5.6%), and radiologic resolution did not differ by eligibility for any trial, and no eligible-versus-ineligible difference survived multiple-comparison correction. Favorable functional outcome (mRS 0-2; available in 95.8%) was nominally higher among eligible patients for MAGIC-MT, STEM, and MEMBRANE, but this advantage vanished after adjustment for pre-morbid mRS and age (all adjusted odds ratios non-significant); pre-morbid mRS and age-not trial eligibility-were the only independent predictors of functional outcome, and no baseline characteristic independently predicted recurrence, rescue surgery, or resolution.

CONCLUSIONS: Across five pivotal RCTs, a majority of real-world recipients fell outside any single trial population, and the two newest trials widened rather than closed this gap. Comparable outcomes across every eligibility stratum suggest MMAE effectiveness may extend beyond trial-defined populations, pending confirmation in dedicated trials.