A novel technique for middle meningeal artery embolization with proximal coiling for subdural hematoma: Institutional predictors of surgical requirement and longitudinal outcomes

Assessment and variable definitions

Baseline demographics, Body Mass Index (BMI), clinical symptoms (e.g., headache), and medication profiles were extracted via electronic medical record (EMR) review. Radiological metrics, including maximal thickness (mm) and hemispheric location, were determined from presentation computed tomography (CT) imaging. SDH acuity was categorized strictly by time from symptom onset to procedure: Acute (<3 days), Subacute (3–21 days), or Chronic (>21 days). Presentation with headache was coded as present (“Yes”) if documented during the initial neurological assessment.

Patient global and symptom outcomes were tracked at 1, 3, and 6 months post-procedure using a standardized institutional questionnaire incorporated into clinical Subjective, Objective, Assessment, Plan (SOAP) notes. This instrument captures nuanced SDH-specific recoveries—such as headache resolution, subtle cognitive/language changes, and focal neurological deficits—that stroke-centric scales like the modified Rankin Scale (mRS) often miss.

Outcomes across imaging resolution, symptom relief, and functional status were quantified using a 4-point ordinal scale (1=minimal improvement; 2=partial improvement; 3=substantial improvement; 4=complete resolution/recovery).

Patient selection and surgical triage

MMAE was offered to patients with symptomatic SDH or those exhibiting progressive radiological expansion on short-interval serial imaging. Incidental, small (<5 mm), non-expanding, asymptomatic SDHs were managed conservatively and excluded.

The decision for adjunct surgical drainage (burr-hole craniostomy, craniotomy, or external ventricular drain) was made independently by the on-call attending neurosurgeon based on clinical evidence of mass effect, midline shift, or progressive neurological decline. For patients undergoing both interventions, surgical drainage was performed during the index admission, either immediately preceding or within 24–48 hours of MMAE.

Endovascular technique

To eliminate inter-operator technical variability, all 100 procedures were performed by a single neurointerventionist via a transfemoral approach. Selective diagnostic angiography was performed to delineate middle meningeal artery (MMA) anatomy and exclude dangerous extracranial-to-intracranial collaterals (specifically petromandibular or ophthalmic anastomoses) [18].

Following distal microcatheter placement into the MMA main trunk, mechanical occlusion was executed using detachable platinum microcoils. A standardized proximal coiling strategy was utilized: microcoils were targeted distal to the petrosal (facial) branch—to eliminate the risk of facial nerve ischemic paralysis—but strictly proximal to the primary frontal-parietal bifurcation (Fig. 1). This targeted deployment achieves main-trunk mechanical flow stasis, devascularizing the primary dural supply while preserving physiological distal collaterals.

Fig. 1.

Radiographic imaging and anatomical localization of the middle meningeal artery (MMA). (A) Anteroposterior fluoroscopic image of bilateral proximal MMAE with coils (blue arrows). (B) Lateral fluoroscopic image showing proximal microcoil (blue arrow) deployment. (C) Schematic illustration of middle meningeal artery anatomy demonstrating the standardized proximal occlusion safe zone for microcoil deployment. (D) Axial non-contrast head CT (computed tomography) demonstrating proximal coil artifacts within the skull base framework.

Statistical analysis

Statistical analysis was conducted using Jamovi (Version 2.3). Continuous variables were tested for normality using the Shapiro-Wilk test. Because all continuous variables demonstrated non-normal distributions (p ≤0.010), non-parametric tests were utilized throughout. Continuous metrics are reported as Median [Interquartile Range, IQR], and categorical metrics as counts and percentages (n, %).

Bivariate comparisons between the MMAE Alone and Adjunct Surgery cohorts were conducted using Mann-Whitney U tests for continuous variables and Chi-Square or Fisher’s Exact tests for categorical variables. To account for potential confounding and identify true independent clinical predictors, variables were evaluated using a multivariable binomial logistic regression model (reporting adjusted odds ratios [aOR] and multivariable p-values), with model performance quantified via Receiver Operating Characteristic (ROC) curve analysis (AUC) (Fig. 2). Stratified comparisons across the three acuity cohorts (Acute, Subacute, Chronic) were analyzed using One-Way Non-Parametric ANOVA (Kruskal-Wallis test with Dwass-Steel-Critchlow-Fligner post-hoc pairwise comparisons) and contingency Chi-Square tests. Longitudinal outcomes across 1, 3, and 6-month follow-ups (anchored to t=0 at the date of index MMAE) were evaluated using Friedman tests. Statistical significance was set at α=0.05.

Fig. 2.

Predictive model performance. Receiver Operating Characteristic (ROC) curve for the multivariable binomial logistic regression model predicting adjunct surgical requirement (AUC=0.745; Sensitivity=61.7%; Specificity=73.6%).

Use of generative artificial intelligence tools

Generative AI (Gemini Nano) was utilized solely to assist in vector rendering the initial graphical layout of Fig. 1C. All anatomical landmarks, vascular branch points, and microcoil deployment boundaries were independently reviewed, refined, and verified for clinical accuracy by the authors. No AI tools were used for data collection, statistical modeling, analysis, or manuscript writing.

Baseline demographics and acuity distribution

The total cohort (N=100) had a median age of 76.0 years (IQR 13.0) and was 76% male (n=76). The overall cohort median BMI was 26.6 kg/m² (IQR 22.5–29.8). The acuity breakdown demonstrated a predominant subacute representation: 54% subacute (n=54), 27% chronic (n=27), and 19% acute (n=19). The standardized proximal coiling technique required a median of 4.00 coils (IQR 2.00) with a median fluoroscopy time of 7.60 minutes (IQR 5.56) (Table 1).

Table 1.

Baseline clinical characteristics and bivariate comparisons stratified by treatment cohort

Predictors of surgical requirement

Bivariate analysis demonstrated that presentation with headache was significantly more frequent in patients requiring adjunct surgical drainage compared to MMAE alone (40.4% vs. 17.0%, p=0.009). In contrast, maximal SDH thickness did not differ significantly between the surgical and non-surgical groups in bivariate testing (12.9 mm vs. 11.0 mm, p=0.202).

The multivariable logistic regression model (AUC 0.745; Sensitivity 61.7%; Specificity 73.6%) identified three independent predictors of surgical requirement: BMI, headache presentation, and hemispheric location. Higher BMI was independently associated with surgical requirement (aOR 1.09 per 1-unit increase; 95% CI 1.01–1.17; p=0.028) (Fig. 3). Surgical patients had a significantly higher median BMI (28.8 kg/m², IQR 24.2–31.4) than MMAE-alone patients (26.2 kg/m², IQR 22.5–29.1). Presentation with headache carried nearly three-fold higher odds of requiring surgical drainage (aOR 2.86; 95% CI 1.03–7.99; p=0.045). Left-sided hematomas were associated with significantly lower odds of requiring surgical drainage compared to right-sided collections (aOR 0.31; 95% CI 0.10–0.97; p=0.043).

Fig. 3.

Baseline Body Mass Index (BMI) stratified by surgical requirement. Raincloud plots illustrating baseline BMI distributions for middle meningeal artery embolization (MMAE) Alone (0) vs. Adjunct Surgery (1). Median BMI was significantly higher in the surgical cohort (28.8 kg/m², IQR 24.2–31.4) compared to MMAE Alone (26.2 kg/m², IQR 22.5–29.1; p=0.028).

Maximal hematoma thickness did not achieve independent significance in the multivariable model (aOR 1.07; p=0.099).

Acuity subgroup stratification

Non-parametric ANOVA across the three acuity cohorts confirmed baseline demographic and procedural homogeneity (Table 2). Specifically, baseline demographic and radiological parameters demonstrated parity across groups, with age (p=0.265), BMI (p=0.569), and maximal SDH thickness (p=0.235) showing no statistically significant differences between acute, subacute, and chronic collections. Procedural execution was similarly standardized, as neither the number of microcoils deployed (p=0.393) nor the total fluoroscopy time (p=0.969) varied significantly by acuity. Natural history validation was confirmed by a significant variation in time from symptom onset to procedure across cohorts (median 3.00 days acute vs. 8.50 days subacute vs. 55.0 days chronic; H=79.32, p<0.001), with post-hoc pairwise DSCF comparisons demonstrating significant differences between all cohort pairs (p <0.001). Finally, although acute cases presented with a higher numerical rate of adjunct surgical drainage (68.4%, 13 out of 19) compared to subacute (48.1%, 26 out of 54) and chronic (51.9%, 14 out of 27) presentations, this variation did not reach statistical significance (χ²=2.34, p=0.311).

Table 2.

Baseline clinical characteristics and procedural outcomes stratified by SDH acuity

Safety and longitudinal outcomes

Technical success was 100% across all 100 procedures. Zero periprocedural complications (including ischemic stroke, facial nerve palsy, or groin hematoma) occurred.

Through the 6-month follow-up, zero patients experienced surgical recurrence requiring re-operation (0% recurrence rate). Comprehensive follow-up imaging was available for 99 patients at 6 months; all 99 underwent CT imaging demonstrating stability or complete resolution, with zero cases of delayed radiological re-expansion. Longitudinal analysis demonstrated significant step-wise improvements from 1 to 6 months in radiological resolution (p<0.001), symptom relief (p<0.001), and functional recovery (Friedman χ²=42.7, p<0.001) (Fig. 4).

Fig. 4.

Longitudinal clinical, radiological, and functional outcomes. Step-wise improvement across 1-, 3-, and 6-month post-procedural time points for imaging resolution, symptom relief, and functional recovery (Friedman test, p <0.001). Follow-up intervals were anchored to the index MMAE date (t=0). MMA, middle meningeal artery

Biological rationale across acuities

In chronic SDH, MMAE halts dural neo-angiogenesis within the hypervascular outer membrane.4) In acute and subacute settings, traumatic micro-tearing of dural border cells and fragile bridging vessels creates active, low-grade arterial oozing. We hypothesize that targeted proximal embolization reduces hydraulic pressure within the dural arterial bed, stanching microbleeding and preventing acute collections from evolving into expanding, chronic collections. Crucially, our subgroup analysis (Table 2) demonstrates that total fluoroscopy duration (median ~7–8 minutes) and coil requirements (median 3–4 coils) remain identical across acute, subacute, and chronic presentations, supporting a unified technical protocol. Prospective physiological studies are needed to validate this mechanism.

Predictor competition and clinical triage

Maximal hematoma thickness did not achieve independent statistical significance in our multivariable model (p=0.099). In clinical practice, neurosurgical decompression is rarely driven by a single static imaging parameter; rather, it reflects symptomatic mass effect. In our cohort, headache presentation served as a potent surrogate for mass effect, carrying nearly three-fold higher odds of requiring surgery (aOR 2.86). However, because headache severity heavily influences an attending neurosurgeon’s decision to operate, this finding partly reflects clinician triage bias rather than pure hematoma biology.

Pathophysiological mechanisms: BMI and hemispheric laterality

The association between elevated BMI and surgical requirement (aOR 1.09 per kg/m²) represents a novel systemic predictor. Central obesity elevates intra-abdominal and intrathoracic pressures, driving up central venous pressure [14,17]. This downstream intracranial venous hypertension impairs dural venous sinus outflow and elevates capillary hydrostatic pressure, which is hypothesized to enhance dural capillary leakage and accelerate hematoma expansion. Prospective physiological studies are needed to validate this mechanism.

Regarding anatomical laterality, left-sided hematomas exhibited 69% lower odds of requiring surgical drainage (aOR 0.31). Dominant-hemisphere lesions produce early, highly noticeable neurological symptoms (e.g., subtle paraphasias or word-finding hesitation) at smaller hematoma volumes. These early clinical indicators may prompt earlier medical presentation and MMAE triage, successfully arresting hematoma progression before mass effect reaches an emergent threshold requiring surgical rescue. In contrast, right-sided collections may remain clinically silent until significant mass effect develops. In this series, hemispheric dominance was assumed based on anatomical side; future studies should incorporate formal handedness testing.

Technique contextualization and anatomical landmarks

In contrast to diffuse distal liquid embolic penetration or distal microcoil packing [1], our technique emphasizes a precise anatomical deployment zone: distal to the petrosal branch but proximal to the frontoparietal bifurcation (Fig. 1C). This choice of embolic agent and targeted deployment aligns with broader comparative evaluations of embolic materials in MMAE [8]. This stasis-inducing proximal coil pack avoids non-target facial nerve ischemia while effectively devascularizing the main MMA trunk.

Study limitations

This study is limited by its retrospective, single-center design. Although evaluating a single operator eliminates technical confounding, it may limit generalizability. Additionally, treatment outcomes were evaluated using an institutional questionnaire rather than the modified Rankin Scale (mRS) or Extended Glasgow Outcome Scale (GOS-E). While this internal tool captures subtle SDH-specific recoveries (such as headache resolution), its lack of formal external validation limits direct meta-analytical comparison with large-scale clinical trials. Nevertheless, evaluating outcome domains alongside clinical presentation aligns with recent international consensus statements on chronic subdural hematoma management [9,12]. Finally, our definition of “0% recurrence” was strictly clinical/surgical (requiring re-operation); while 99 out of 100 patients had CT-confirmed radiological stability at 6 months, subtle asymptomatic minor fluid fluctuations managed conservatively were not classified as recurrences.