Direct epidural ethanol injection in aggressive vertebral hemangiomas to decompress the central canal: a multistep percutaneous treatment strategy

Abstract

Background

Vertebral hemangiomas are incidental and typically, asymptomatic lesions of the spine, present in 10–12% of the population. However, aggressive vertebral hemangiomas (AVHs) can compromise the spinal canal, leading to spinal cord or nerve root compression, and require timely treatment to prevent permanent neurological deficits. Surgical management is challenging owing to the high vascularity of AVHs, and carries a significant risk of perioperative blood loss. Intraosseous ethanol injection is commonly used for sclerotization, but may not adequately deal with epidural components.

Objective

To carry out a staged treatment with an image-guided puncture and ethanol injection of the epidural component in 12 patients.

Methods

We retrospectively analyzed 12 patients with symptomatic AVHs who underwent targeted epidural ethanol injection followed by vertebral body cement augmentation, between 2017 and 2024, at three tertiary hospitals. Data collection included pre- and post-treatment imaging and clinical outcomes.

Results

Among 12 patients (mean age 50, women 50%), all had extensive epidural involvement and were symptomatic, including spinal cord compression and pain. Reduction in size of more than 75% of the epidural hemangioma was achieved in 8 cases, with 11 patients experiencing complete symptom resolution. Laminectomy was performed in 3 cases, while corpectomy was avoided in all cases. Two patients had neurological worsening, with one achieving complete resolution and the other having mild residual impairment after rehabilitation due to a small spinal cord ischemic lesion. No other major complications occurred.

Conclusion

Direct epidural ethanol injection provides a minimally invasive alternative to surgery, such as corpectomy, including rapid size reduction of the compressive epidural component, and potentially, prevents retrograde flow into arterial collaterals. Adding vertebroplasty enhances vertebral stability.

Introduction

Vertebral hemangiomas are common incidental vascular malformations of the spine, that can be found in 10–12% of the general population.^{1 2} Hemangiomas are post-capillary vascular lesions and therefore are composed of thin-walled, endothelial-lined capillary and venous channels, surrounded by edematous, fibrous stroma, and bony trabeculae that can cause compartmentalization. They are predominantly fed by low-pressure venous systems. This histological architecture makes them more prone to localized expansion, but owing to their post-capillary nature, hemangiomas progress relatively slowly.³ Hemangiomas can be classified according to the Enneking classification into (1) latent or fatty (stage 1): these lesions exhibit well-demarcated borders with a high content of fat and a low vascularity, and are generally asymptomatic; (2) active (stage 2): characterized by indistinct borders, higher vascularity, but without extraosseous involvement. These hemangiomas can cause symptoms; (3) aggressive (stage 3): these tumors also present with indistinct borders but demonstrate rapid growth, a higher vascularization, and extraosseous growth.⁴

Although usually asymptomatic, the more aggressive variants can show posterior wall destruction and infiltration of the spinal canal with accompanying symptoms due to radicular or spinal cord compression.³ Extensive involvement of the vertebral body increases the risk of pathological fracture.^{5 6} In symptomatic aggressive vertebral hemangiomas (AVHs), especially with neurological deficit or poorly controlled pain, timely treatment is indicated.⁷

Surgical management of AVHs presents significant challenges due to the tumor’s high vascularity, which increases the risk of substantial intraoperative blood loss.⁸ Laminectomy or hemilaminectomy can be performed for emergency decompression, but relevant blood loss is possible if the hemangioma extends into the posterior elements. This also applies to open surgical treatment, such as corpectomy and dorsal instrumentation, which is a very invasive treatment, especially in the case of thoracic involvement.⁹ Owing to the post-capillary nature of this vascular malformation, transarterial embolization is not curative and is mostly performed for temporary devascularization before surgery.² Even with preoperative embolization, the estimated blood loss during corpectomy in patients with AVHs can range from 1100 to 2600 mL, leading to increased morbidity and the need for blood transfusion.⁸

Image-guided percutaneous procedures, such as transpedicular sclerotization using ethanol, vertebroplasty, or a combination of both, are minimally invasive alternatives to treat AVHs and augment the affected vertebral body.10,14

Controversy exists about the safety of percutaneous ethanol embolization of AVHs in general, as cases with serious neurological and cardiovascular complications have been reported. In this context, a recently published case report advocates the use of preinterventional transarterial embolization prior to percutaneous ethanol injection to avoid retrograde flow of ethanol to spinal medullary arteries via dangerous anastomoses.¹⁵ Additionally, direct ethanol injection into the osseous part of the vertebral hemangiomas carries a risk of postprocedural vertebral body fractures due to the weakening of weight-bearing structure from sclerosis and necrosis of the tumor’s vascular components,⁵ therefore adding a stabilizing procedure like vertebroplasty should be considered.¹⁶

Despite a positive effect in the previously mentioned studies, an intraosseous injection of ethanol or polymethyl methacrylate (PMMA) does not reliably address the epidural component of the hemangioma, owing to compartmentalization of the vascular architecture of AVHs.⁵ In cases of spinal cord or nerve root compression, immediate volume reduction of the epidural part is crucial. If the injected ethanol does not reach the epidural component during intraosseous injection, a direct sclerotherapy of the epidural compartments could be considered for an immediate therapeutic effect.^{10 17}

We present a case series of 12 patients with AVH, featuring extensive epidural spinal cord compression. In all patients a staged treatment with an image-guided puncture and ethanol injection of the epidural component was performed.

Methods

The data for this study were collected retrospectively from the electronic health records of patients treated with ethanol sclerotization of AVH between 2017 and 2023 at three tertiary hospitals specializing in spine interventions. The retrospective analysis was approved by the local ethics committees. Informed consent was obtained for all procedures.

Among patients diagnosed with AVHs and treated with ethanol sclerotization, those treated with additional direct epidural ethanol injection were selected.

Preinterventional imaging

All patients underwent preinterventional imaging, including unenhanced spinal CT and MRI with intravenous gadolinium contrast injection and fat suppression techniques.

Percutaneous ethanol injection

All procedures were performed under general anesthesia, using CT guidance, equipped with fluoro-CT capability. As a first step, for intraosseous injection, 14G vertebroplasty needles were inserted through the pedicles into the posterior third of the vertebral body. Venous bleeding was observed through the needles. If injection of intraosseous contrast agent (2–3 mL at 0.5 mL/s) failed to opacify the epidural components (figure 1), a direct puncture of the epidural pouches of the AVH, responsible for central canal or foraminal compression, was considered.

Figure 1. Adult patient presented with an aggressive vertebral hemangioma at T2, accompanied by mild signs of myelopathy and localized pain. During the first treatment session, the extensive epidural components of the hemangioma, visible on preinterventional MRI (A, C), failed to opacify during intraosseously injected contrast agent in CT (B, D).

In extensive epidural hemangioma formations, a staged direct epidural ethanol injection approach was chosen to reduce the potential risk of thrombus-related swelling and worsening of spinal cord compression. In these cases, the target volume for the first sclerotization was roughly 50% of the compressive epidural component, followed by a second session a few days later to deal with the remaining epidural component (figure 2).

Figure 2. Adolescent patient with an aggressive vertebral hemangioma at T2, accompanied by an extensive epidural component. Baseline MRI images show severe spinal canal stenosis and spinal cord compression: in sagittal fat-suppressed, contrast-enhanced, T1-weighted (A) and transverse T2-weighted images (B) the typical imaging characteristics of an aggressive vertebral hemangioma are visible. CT-guided direct epidural punctures were performed in both the left (C) and right (D) epidural components, through the laminae, via coaxial needle approach, in different sessions. Visible compartmentalization of the epidural pouch is observed following contrast injection on the left (C). Follow-up MRI after left-sided (E) and right-sided (F) sclerotherapy shows thrombosis within the respective treated compartment (note that the E and F images are at slightly different levels). Final follow-up in Figure 4.

For direct epidural sclerotherapy the largest accessible epidural pouches based on MRI findings were targeted. The access was usually through the ligamenta flava using a 22G spinal needle.When a direct interlaminar or transforaminal approach was not feasible, the access was transosseous, through the laminae or the medial cortex of the pedicles of the affected vertebral body using a 14G bone access cannula (Bonopty Biopsy system, AprioMed AB, Uppsala, Sweden) and coaxial drill to perforate the inner osseous cortex, and accessing the epidural pouch with a coaxial 22G spinal needle through the biopsy cannula (figure 3). Injection of a low-volume (1–2 mL) of iodinated non-ionic contrast agent (300 mg/mL) at 0.3 mL/s, while dynamic fluoro-CT scanning, was performed to monitor intralesional distribution, including visualization of further different epidural compartments, and to rule out reflux into arteries or abundant venous outflow. Simultaneously, special attention was given to monitor contrast medium backflow into the anterior spinal artery. If correct positioning and intended contrast agent distribution was proved, manual low-pressure injection of 1–3 mL of ethanol (96%) was then slowly injected at 0.3 mL/s using an ethanol resistant syringe, with cardiovascular monitoring maintained throughout. The total amount of ethanol varied depending on size of the epidural compartment and contrast agent distribution in the prior injection.

Figure 3. Illustration of the technique. When transpedicular access cannot opacify the epidural component of the hemangioma (A), a direct epidural puncture through the lamina (B) or a direct route, if accessible, for ethanol injection is possible. Depending on the degree of postinterventional size reduction (C), a second injection into a different compartment may be necessary (D). After sufficient decompression of the spinal canal (E), cement augmentation is recommended for final stabilization (F).

Additional treatment

In the final session, high-viscosity PMMA bone cement vertebroplasty was performed in all patients to complete intraosseous AVH embolization and prevent fracture. For vertebral augmentation, a single 14G vertebroplasty needle was advanced transpedicularly into the anterior third of the vertebral body. High-viscosity bone cement was then incrementally injected until more than two-thirds of the vertebral body was adequately filled, extending from endplate to endplate. In patients with acute, severe, or progressive neurological impairment, additional dorsal decompression (laminectomy or hemilaminectomy) was performed.

Outcome

Complications were documented, and patients were followed up with MRI and CT at least once to evaluate the AVH, particularly the epidural component, with size reduction categorized into five groups (<25%, ≥25%, ≥50%, ≥75%, ≥90%). Clinical follow-up occurred 3–12 months after the last intervention.

Results

At the three participating centers, of 23 patients with AVH treated with ethanol sclerotization, 11 patients were treated by intraosseous ethanol injection and vertebroplasty, while 12 patients were treated with direct ethanol injections of an extensive epidural component. A total of 27 direct ethanol injections of an epidural hemangioma compartment were performed in these latter 12 patients. Six patients were women, six were men. Mean age for the women was 55 years (range 40–62), and for men, 45 years (range 16–60 years) (online supplemental table 1).

Clinical presentation was newly developed clinical signs of myelopathy and/or radiculopathy, with back pain in three patients, neurological deficit without back pain in five, and back pain without accompanying neurological symptoms in four patients. Two patients presented with rapidly progressive acute neurological impairment, resulting in paraparesis due to spinal cord compression. The other patients had mild myelopathy and presented with lower extremity hyper-reflexia, radicular pain, and/or numbness.

The vertebral levels treated were T2 (n=2), T4 (n=1), T5 (n=2), T8 (n=1), T12 (n=1), L3 (n=3), and L4 (n=2).

Based on imaging, clinical presentation, and a comprehensive multidisciplinary discussion, a treatment plan was devised. In two cases, immediate surgical decompressive laminectomy was performed, followed by percutaneous epidural ethanol sclerotization and vertebroplasty in two to three subsequent sessions. In one patient, decompressive laminectomy was performed after the first direct ethanol injection due to clinical deterioration occurring 1 week later, followed by a second ethanol sclerotization and vertebroplasty. For all remaining cases, a stand-alone percutaneous treatment was employed. The mean number of percutaneous interventions, including ethanol injection and vertebroplasty in the final session, was 2.3 (range 1–3). The mean volume of ethanol injected per individual site was approximately 2 mL (range 1–3 mL), with a total session volume averaging 5 mL (range 1–7 mL). The injected ethanol volume was adjusted based on the distribution pattern of the contrast agent administered prior to sclerotherapy. In all cases, ethanol intraosseous injection was also performed into the vertebral body, typically as a first-step approach, followed by direct epidural sclerotization due to failure in reaching the epidural pouch via intraosseous injection.

Vertebroplasty was then performed as the final step, in all patients, using one or two transpedicularly positioned cannulas, achieving filling of more than two-thirds of the vertebral body with PMMA.

As previously noted, one patient, with AVH at the T2 level, exhibited clinical deterioration between two sclerotization sessions, with worsening paraparesis occurring 1 week after the first treatment, and just before the regularly scheduled second session. In response, emergent decompression laminectomy was performed. The patient experienced no long-term sequelae (grade 3 complication according to the Cardiovascular and Interventional Radiological Society of Europe (CIRSE)).¹⁸ In a second case, with T8 AVH, initial treatment involved dorsal decompression and instrumentation due to epidural spinal cord compression and acute mild paraparesis. The patient developed epidural hematoma and worsening neurological conditions. Digital subtraction angiography (DSA) for transarterial embolization followed by intraosseous ethanol injection were performed, without improvement and without epidural AVH component volume reduction on MR. The patient was therefore further operated on with extended laminectomy followed by direct epidural ethanol sclerotization. The patient developed a new motor deficit worsening immediately postprocedure. MR performed immediately after the intervention revealed signs of a small localized spinal cord ischemia in a posterior radiculo-medullary artery territory at T7. Six months after rehabilitation, the patient exhibited residual paresthesia in the right foot, with no residual motor deficit (grade 4 complication according to CIRSE).¹⁸

No other adverse events were observed. Of the 12 patients treated, 11 experienced complete resolution of their symptoms. Follow-up imaging, performed 6–9 months after the final treatment, demonstrated a size reduction of the epidural component of the AVH of at least 75% in eight patients (figure 4,online supplemental figure 5) and a reduction of between 25% and 50% in the remaining four patients, in comparison with baseline pretreatment imaging. These four patients were asymptomatic at the outset, and further intervention was avoided to minimize risk. Additionally, two of these patients had already undergone laminectomy, resulting in sufficient decompression of the spinal canal despite the residual epidural component.

Figure 4. The previously mentioned adolescent patient underwent fat-suppressed, contrast-enhanced, T1-weighted MRI before (A, C) and after (B, D) three sessions of percutaneous ethanol injection, dorsal decompression, and vertebroplasty. A significant reduction in size (>75%) of the epidural components of the hemangioma is clearly visible, accompanied by a complete resolution of the spinal cord compression.

Discussion

In our case series of 12 consecutive patients with AVH and extensive epidural involvement, direct epidural sclerotization with ethanol resulted in significant epidural decompression in 9 out of 12 cases. Adjunctive surgical laminectomy was performed in two cases before sclerotization, in one case between two sclerotization sessions, due to secondary deterioration that was probably caused by insufficient size reduction of the extensive epidural component rather than local swelling due to thrombosis of the sclerotized epidural component, which would have manifested in the first 48 hours post-treatment. After rehabilitation the patient completely recovered from his deficit. Importantly, a corpectomy was avoided in all patients. All but one patient experienced complete symptom resolution; the exception involved an adverse event following surgical decompression and ethanol injection, probably resulting in a small localized spinal cord ischemia. However, this patient showed improvement after rehabilitation, with only residual paresthesia in the right foot. Four patients (4/12; 33%) with just minor to intermediate size reduction (25–50%) were oligo- to asymptomatic at the outset. Further intervention was withheld to minimize risk. Additionally, two of these four patients had previously undergone laminectomy, achieving sufficient decompression.

Vertebral hematomas (VHs) are the most common type of vertebral lesion, yet only a small percentage demonstrate an aggressive growth pattern.^{6 13} In these cases, the VH may extend into the surrounding soft tissues, leading to expansion, thinning, or penetration of the periosteum.² VH can become symptomatic when the lesion extends beyond the vertebral body, resulting in progressive epidural spinal cord or nerve root compression, or due to complications such as compression fractures or epidural hemorrhage.^{11 19} The management of symptomatic VH with extensive epidural involvement includes a variety of treatment options, each with specific advantages and limitations.

Radiation therapy, often considered as an alternative for VH, presents challenges due to its delayed therapeutic effect and possible edema after radiotherapy.²⁰ This is particularly problematic in cases of acute spinal cord compression or myelopathy, where immediate decompression is required.^{21 22}

Corpectomy and spinal instrumentation, traditionally regarded as the standard treatment for AVH, pose significant risks of perioperative morbidity, primarily due to substantial blood loss. Earlier studies have documented estimated blood loss of up to 5000 mL,²³ while more recent reports have recorded estimated blood loss of approximately 2300 mL in cases without preoperative embolization.²⁴ Even when preoperative transarterial embolization is performed, intraoperative blood loss can still range between 1500 and 1945 mL.^{23 24} A recently introduced technique using preoperative direct percutaneous embolization with polymethylmethacrylate has further reduced blood loss during corpectomy. However, significant blood loss still occurs, with an average of 1005 mL (range 100–3900 mL) reported,²⁵ and it remains a very invasive procedure, requiring a 360° spine surgical approach with extended in-hospital stay.

Because AVHs are post-capillary vascular malformations, transarterial embolization, primarily addressing the feeding arteries, does not occlude the malformed venous vascular bed of the lesion.^{8 26} Therefore transarterial embolization is only temporarily effective as an adjunct, prior to surgery, as it can reduce intraoperative blood loss, but it has limited efficacy as a stand-alone treatment.,^{8 25}

Conversely, a critical advantage of direct ethanol injection is its ability to target and disrupt the capillary and post-capillary components of the hemangioma.¹⁰ Direct ethanol injection achieves immediate endothelial damage, thrombosis, and subsequent sclerotization of the VH. Rapid size reduction of the AVH, demonstrated as early as 24–48 hours after ethanol injection, is of clear interest, especially in cases where epidural AVH pouches cause compression and neurological deficits.^{3 5}

Ethanol sclerotization of AVH has been reported by many authors, although usually in small series, due to the rarity of the pathological entity.⁵ The reported techniques vary, but invariably describe an intraosseous injection of the ethanol, usually at the posterior third of the vertebral body, following a transpedicular approach, to obtain diffusion of the ethanol to the intrasomatic and ventral epidural components of the AVH.⁵ Sclerotization solely of the intraosseous component, as usually done with cement embolization and augmentation, has unpredictable effects, if any, on the compressive epidural components, which are indeed the most important target of treatment if a central canal decompression is seeked. It would be expected that an injection of a fluid inside the AVH, from the posterior third of the vertebral body would easily reflux in the epidural pouches of the AVH, just a few millimeters distant from the injection site. In reality, due to compartmentalization of the AVH angioarchitecture, the diffusion of fluids directly injected into the vascular malformation is quite unpredictable.¹¹ Since ethanol is radio-transparent, its actual diffusion cannot be directly seen under fluoroscopic and CT guidance, but its diffusion can be estimated and predicted by injecting a contrast agent at the injection site before the actual ethanol injection. Our first approach included an initial intraosseous injection of contrast agent, as described in the standard ethanol sclerotization technique.¹³ When we could not achieve satisfactory opacification of the epidural pouches of the AVH from the intraosseous injection site, we proceeded with a direct access and sclerotization of the epidural components.

Injection of large volumes of ethanol, above 10 mL, as usually performed during intraosseous injections, carries risks of cardiovascular complications (ie, pulmonary hypertension). As post-capillary vascular malformations the AVH should have no direct communication with the radiculo-medullary arterial system, but reflux into dangerous arterial collaterals has also been observed, potentially leading to spinal cord ischemia.²⁷ This complication has been reported in the literature, prompting some to recommend transarterial DSA and embolization of collateral vessels supplying the hemangioma-bearing vertebral body to minimize the risk of spinal cord ischemia.^{15 27} Prophylactic embolization is possible though only at levels where no radiculo-medullary arteries are present, which makes the prophylactic embolization possibly less useful. Fluoroscopy and DSA with high temporal resolution may enhance the visualization of spinal cord-supplying arteries, such as the artery of Adamkiewicz (A. radicularis magna). However, in cases of low-flow and low-pressure injections, such small arteries might not be visible on DSA, therefore it would be difficult to replicate on the DSA mapping the the intraprocedural setting. Cross-sectional imaging modalities such as CT, on the other hand, may not help visualization of the smaller intradural arteries, but may instead provide superior anatomical detail; they have proved to be very precise in guiding the puncture of the epidural pouches.To mitigate the risk of spinal cord ischemia resulting from ethanol reflux into spinal cord-supplying arteries, it is recommended to inject low volumes of ethanol, at a slow rate and low pressure Another potential advantage of direct epidural injection is the reduced volume of ethanol required, as we injected just 1–3 mL of ethanol per injection compared with 5–16.5 mL described in the literature.⁵ Since the epidural compartment is more localized and smaller than the compartment inside the vertebral body, smaller ethanol volumes and slow, low-pressure injections can achieve the desired therapeutic effect, thus minimizing the risk of complications such as ethanol reflux and spinal cord ischemia. Nevertheless, in our case series, a small spinal cord ischemia, in a localized posterior spinal artery territory, occurred, and it is not possible to know whether this complication happened because of the intraosseous injection, direct epidural sclerotization, which can both lead to a reflux, or due to other causes. A further advantage of epidural sclerotization is the potential reduction of the risk of cement leakage into the epidural space, as the epidural component of an AVH may serve as a path of least resistance for cement migration.²⁸ By inducing thrombosis and/or shrinkage of the epidural component prior to vertebroplasty, this technique could minimize the risk of cement migration into the epidural space, which might otherwise contribute to worsening spinal cord compression.

Although ethanol injections carry the risk of cardiovascular complications, such as arrhythmias and pulmonary hypertension, careful dosing, maintaining a threshold dose of 0.1 mL/kg per bolus, and proper patient monitoring can keep the risk of significant cardiovascular events low.^{10 29} In all cases of direct epidural ethanol injections, the volume required for complete opacification of the treated compartment remained well below the safety threshold, with injection volumes ranging from 1 to 3 mL of 96% ethanol directly into the epidural pouch.^{10 26}

Owing to the diminished bone density, honeycombing, vascular proliferation, and fat deposition characteristic of AVH, the increased mechanical instability should be addressed. In this regard, combining ethanol injection with vertebroplasty is generally recommended and has been previously documented.^{1 5 10 19} Compression fractures following ethanol embolization, attributed to osteonecrosis, have been reported as early as 4 weeks postprocedure, particularly in cases where the vertebral body is already structurally compromised.^{13 30} Vertebroplasty provides immediate structural reinforcement and improving long-term patient outcomes. Additionally, the thermal properties of cement may offer a therapeutic benefit by contributing to the destruction of pain fibers and hemangioma capillaries.¹

Limitations

This study has several limitations. The most significant is the small sample size of the patient cohort and its retrospective design. Symptomatic AVHs are rare, and treating them percutaneously without immediate surgical intervention or corpectomy—particularly in cases involving spinal canal stenosis—requires the collaboration of an interdisciplinary team willing to assume the risk of potential secondary deterioration rather than opting for more immediate but high-risk surgery.

Conclusion

Direct epidural ethanol injection offers several advantages in the management of AVH with extensive epidural involvement. Addressing the epidural component of the AVH ensures rapid decompression of the spinal canal, providing a less invasive measure than decompressive surgery, and when followed by cement augmentation to reinforce the weight-bearing structure of the vertebral body avoids the need of a high risk corpectomy.