Document resource
Background Middle meningeal artery embolization (MMAE) for chronic subdural hematoma (cSDH) employs various embolic agents. Whether treatment efficacy varies by baseline SDH imaging density phenotype remains unknown. We aimed to evaluate whether embolic agent strategy interacts with SDH imaging density to influence outcomes following MMAE.Methods This retrospective analysis of the Multicenter Embolization of Subdural Hematomas (MESH) Registry included patients undergoing MMAE with documented embolic agent class and baseline imaging phenotype. Patients were classified by embolic agent into liquid embolics, particles, and particles with coils; those receiving other combinations or missing data were excluded. Baseline SDH imaging density was classified as hypodense, isodense, hyperdense, or mixed on pre-procedure CT. Propensity scores for the three embolic groups were estimated using multinomial logistic regression including age, sex, antithrombotic use, baseline SDH thickness, contralateral SDH, midline shift, membranes/septations, coronary heart disease, and premorbid modified Rankin Scale (mRS). Stabilized inverse probability of treatment weights (IPTW) were calculated. Primary outcomes included rescue surgical evacuation, SDH recurrence, and favorable functional outcome (mRS 0-2). Secondary outcomes included all-cause mortality and complete radiographic resolution. Interaction between embolic agent and density phenotype was tested using likelihood ratio tests comparing models with and without interaction terms. Stratum-specific adjusted odds ratios (aOR) with 95% CIs were estimated using IPTW-weighted logistic regression with robust standard errors.Results Among 2,306 SDH cases in the MESH Registry, 785 patients met inclusion criteria. The cohort included 34 (4.3%) hypodense, 124 (15.8%) isodense, 253 (32.2%) hyperdense, and 374 (47.6%) mixed density SDH. Liquid embolics were used in 405 (51.6%), particles in 258 (32.9%), and particles with coils in 122 (15.5%). Mean age was 73.2-74.8 years across groups; 68-75% were male. Embolic agent selection varied significantly by phenotype (p<0.001), with liquid embolics used more frequently in hyperdense (56.5%) versus mixed (51.3%) SDH. After IPTW weighting, all covariates achieved adequate balance (maximum weighted SMD 0.071). Median follow-up was 3.7 months (IQR 1.3-11.7). Overall outcome rates were rescue surgery 6.2% (44/713), SDH recurrence 4.0% (30/750), favorable mRS 74.2% (471/635), mortality 9.6% (62/647), and complete resolution 37.4% (280/749). No statistically significant interactions were observed between embolic agent strategy and 4-level density phenotype for any primary outcome: rescue surgery (LRT p=0.06), SDH recurrence (p=0.18), or favorable mRS (p=0.11). For mortality, the interaction approached significance (LRT p=0.06) and reached significance in collapsed 3-level analysis (p=0.01). Exploratory stratum-specific analyses revealed several hypothesis-generating findings in the mixed density stratum: liquid versus particles showed higher SDH recurrence (aOR 8.27, 95% CI 1.04-65.85, p=0.05), lower mortality (aOR 0.40, 95% CI 0.19-0.85, p=0.02), and lower complete resolution (aOR 0.59, 95% CI 0.35-1.00, p=0.05). In isodense SDH, liquid embolics were associated with higher favorable mRS compared to particles (aOR 3.58, 95% CI 1.10-11.66, p=0.03). Post-hoc power analysis indicated 80% power to detect interaction OR of approximately 5.0 or larger.Conclusions No significant global interactions were observed between embolic agent strategy and SDH imaging density phenotype, though the study was underpowered to detect subtle heterogeneity. Exploratory stratum-specific findings in mixed and isodense SDH warrant prospective investigation.Disclosures M. Essibayi: None. B. DeMessie: None. A. Karandish: None. D. Lakhani: None. T. MESH Collaborators: 1; C; NIH/NINDS; NIH-NINDS (RO1NS127114-01, RO1NS138765, RO3NS126804); PCORI; NSF SBIR Grant; Medtronic; MicroVention; Stryker; Zeiss; Siemens; Site-PI EMBOLISE trial; Investigator in EMBOLISE (Medtronic);. 2; C; Medtronic; J&J Cerenovus; Route 92 Medical; Certus Critical Care; Piraeus Medical; Johnson & Johnson; Balt; Siemens; Terumo Neuro; Stryker; Viz.ai; Microvention; Werfen; NTI; Reist; Penumbra; Cerenovu. 3; C; Phenox; Cerenovus; Penumbra; Bracco; Medtronic. 4; C; Proprio; Stroke Diagnostics; Apertur; Stereotaxis; Fluid Biomed; Synchron; Hyperion Surgical. 6; C; Unrestricted educational grants from Medtronic and Stryker; editorial board of JNIS Advances; board of Society of NeuroInterventional Surgery; National Football League.