Approval for this retrospective study of medical records, photographs, and radiologic imaging was obtained from the Institutional Review Board (No.SH9H-2019-Q-024). All patients’ consent for participation in the study was waived.

The flow chart of this study is shown in Fig. 1. bAVMs are defined as AVMs that occur in the region of the buttocks, the superior aspect of which ends at the iliac crest, and the lower part is outlined by the horizontal gluteal crease [13]. Physical examination demonstrated common symptoms of bleeding, bloody stool, erythema, elevated skin temperature (measured by the infrared measure temperature gun), pain, pigmentation, pulsating mass, swelling (the inelastic tapeline measured hip perimeter), and ulceration. A total of consecutive 32 patients with bAVMs who had received ethanol embolization treatment in our center from July 25, 2012, to August 1, 2021, were included.

Flow chart of the ethanol embolization of arteriovenous malformations in the buttock (bAVMs). IIA: Internal iliac artery; EIA: External iliac artery

After the clinical investigation, Doppler ultrasound (DUS) and contrast-enhanced computed tomography (CT) were used to evaluate the hemodynamic and anatomic features of the bAVMs. According to Schöbinger stage of AVMs, stage II or higher indicates the need for treatment [2]. Contraindications included the following: previous allergic reactions to ethanol or the contrast medium, previous major complications from ethanol embolization, and severe impairment of hepatic, renal, and cardiopulmonary function.

All ethanol embolization procedures were done under general anesthesia. Pulmonary arterial pressure (PAP) monitoring through the Swan-Ganz catheter (Edwards Lifesciences, Irvine, CA, USA) was not routinely performed. Eleven patients (11/32; 34.4%) had peri-operative PAP monitoring when the estimated volume of absolute ethanol per procedure was more than 0.5 mL/kg based on the size of the lesion [6].

A baseline angiography (contrast media: Ultravist®, Bayer, Germany) of all the AVMs was performed through the contralateral femoral artery approach to determine the hemodynamic and angioarchitectural features of the AVMs. The dilated vein with the maximum and fastest flow rate in the AVMs lesion was regarded as the dominant outflow vein (DOV) [8]. According to the Yakes classification [14] Type IIa bAVMs were treated with the direct puncture of nidus under the guidance of DUS after the baseline angiography. The DOV of the Type IIb/IIIa/IIIb bAVMs was identified by super-selective angiography (Fig. 2a). The patients were turned from the supine to the prone position before receiving the puncture procedures. Percutaneous puncture of the DOV was performed with an 18-gauge needle (Cook Medical, Bloomington, IN, USA) under the guidance of DUS. A 2.1-F microcatheter (Asahi, Seto, Japan) was then placed into the DOV through the needle (Fig. 2b). Next, three-dimensional (3D) mechanically detachable coils (Micro Therapeutics Inc., Irvine, CA, USA), followed by pushable coils with attached synthetic fiber (Cook Medical, Bloomington, IN, USA), were deployed through the microcatheter. The coils were injected towards arterial inflow. Next, the direct puncture of the nidus was performed by another 21-gauge butterfly needle or 18-gauge needle under the direction of DUS, and angiography via the needle demonstrated the expected position (Fig. 2c). The ethanol was injected via the second needle introduced into the nidus. It was directed towards the afferent side of the draining vein and delivered in small aliquots based on the amount of contrast medium filling the nidus without the opacification of normal vessels (Fig. 2d). Angiography performed 5–10 min after ethanol injection indicated that the nidus was embolized (Fig. 2e). Repeated ethanol injection was required if the nidus was still present. After the direct puncture procedures, the patients were turned back to the supine position for the final angiography, which demonstrated if residual feeding vessels, nidus, and outflow veins were completely embolized (Fig. 2f). In cases where direct puncture failed because of the deep location of the AVM nidus, trans-arterial embolization of the nidus was performed. A 1.7-F microcatheter was placed coaxially into the nidus, which was confirmed by angiography.

Ethanol embolization assisted with coils treating type IIIb arteriovenous malformations in the buttock (bAVMs). (a) Super-selective angiography indicating artery (white arrow), nidus, and dilated vein (black arrow). (b) Percutaneous puncture of outflow vein and insertion of microcatheter. (c) Configuration of coils and direct-puncture of nidus (black arrow). (d) The injection of absolute ethanol within nidus (black arrow). (e) Repeat angiography indicated the occlusion of primary nidus (black arrow). (f) Final angiography after completed embolization (red arrows: coils in situ)

Type IV bAVMs were treated with 50% diluted ethanol, which was made by mixing absolute ethanol with Ultravist® at 1:1 volume. After the baseline angiography of the AVMs (Fig. 3a and Video 1) and arterial super-selection by microcatheter (Fig. 3b), the diluted ethanol was injected trans-arterially (Fig. 3c). Because of the diffuse microfistulas, the remaining microarteries were accessed in order to complete the embolization (Fig. 3d).

Ethanol embolization treating type IV arteriovenous malformations in the buttock (bAVMs). (a) Angiography showed diffused micro-fistulas. (b) Super selective angiography by microcatheter. (c) Injection of diluted ethanol by microcatheter. (d) Super selective angiography of other microarteries. Red arrows: the tip of microcatheter

The dose of absolute ethanol for a bolus injection was < 0.1 mL/kg, and the total volume in a single embolization session was < 1 mL/kg [15]. Dexamethasone and omeprazole were routinely used to mitigate postoperative swelling and prevent gastrointestinal injury, respectively [15]. The ketorolac tromethamine was administrated routinely to ease postoperative pain.

Patients with extensive AVM lesions were treated in a multistage pattern. One month was set between two embolization procedures to avoid unexpected bleeding episodes, skin ulcer, or necrosis. The postoperative reactions to ethanol embolization include hemoglobinuria, postoperative swelling, bleb formation, and non-purulent exudation. Each patient’s follow-up was conducted as routine clinic visits. Patients were also advised to report or return if they experienced any post-treatment complications: aggravated pain, numbness, local necrosis, and infection.

The therapeutic outcomes were evaluated by noting the amelioration of clinical symptoms and devascularization of bAVMs lesions observed on angiography before and after treatment. The devascularization rate was divided into 0–25%; 26–75%; 76–100% and was evaluated by two independent interventional radiologists with 6 and 10 years of experience, respectively. The senior interventional radiologist with 20 years of experience made the final decision in case of disagreement.

Finally, the therapeutic outcomes were classified as: no response (NR): none of the clinical symptoms disappeared or 0–25% devascularization, partial response (PR): part of clinical symptoms disappeared or 26–75% devascularization, complete response (CR): no residual clinical symptoms, and 76–100% devascularization.

The normality of data was evaluated by the D’Agostino and Pearson test. Data was expressed as the median and inter-quartile range (IQR). The descriptive statistics method was used to calculate each data type’s 95% confidence interval (CI). Two-tailed student’s t-test analyzed the difference between groups. All statistical analyses were performed using SPSS 24.0 (IBM Corp., Armonk, NY, USA) and Prism GraphPad 8.3 (GraphPad Software, San Diego, CA, USA). P-value < 0.05 was considered as statistical significance.