Hemangioblastomas of the Lower Spinal Region: Report of Four Cases with Preoperative Embolization and Review of the Literature

Diagnostic Modalities

All our patients underwent a MR examination. MR imaging findings (Figs 1 and 3) were characteristic of and compatible with those reported in the literature (31). According to the literature (4, 11, 20, 35, 36, 46–51), MR imaging allows an accurate diagnosis of spinal HBs and associate cysts, which are reported in 70% of intramedullary cases. Symptomatic small HBs have been reported to have relatively large associated syringes, whereas asymptomatic ones did not (31). MR imaging is also an excellent way to check multiple lesions and follow patients (4, 35, 37). There is no difference in the MR imaging findings between patients with or without VHL disease except for the multiplicity and higher percentage of small tumors in patients with VHL (31). Some authors (35) affirm that MR imaging is the diagnostic technique of choice and angiography should not be performed. Spetzger et al (12) state that, although MR imaging is essential for preoperative diagnosis, spinal angiography is also mandatory because it allows a detailed study of the vascular condition, which is decisive for exact planning of a surgical strategy. In some series (11, 30), angiography has been performed in almost half of cases. In a recent series, Lee et al (14) report that preoperative angiography was performed in 11 out of 14 cases. Brisman et al (15) consider angiography to be the criterion standard in the diagnosis of spinal HBs, and it is useful in some cases in which differential diagnosis with vascular malformations could be difficult. These authors suggest that the angiography should be performed by a neurointerventional team capable to perform presurgical embolization.

Surgery

In general, HBs of the spinal cord and lumbosacral nerve roots should be removed when they become symptomatic (19, 52). In case of cauda equina lesions, these neoplasms appear to originate from the nerve root and it is necessary to sacrifice the nerve root from which the HB originates to achieve complete resection (19). Symptomatic patients should undergo surgery before they develop extensive sensory-motor deficits (32). Pietila et al (53) state that, with the new diagnostic tools now available, microsurgical removal of spinal HBs has a low morbidity rate and suggest that surgical treatment should be considered even for asymptomatic HBs in certain circumstances. Thanks to improved surgical techniques, good results based on microneurosurgical centers’ experience support the concept that radical excision of spinal HBs can be achieved at low levels of surgical mortality and morbidity (11, 53–56). Malis (56) describes a unique microsurgical technique of bipolar coagulation that offers a safe and relatively easy tumor removal. A bipolar forceps is used to shrink each tumor and detach it from its feeding and draining vessels. In Malis’s report, tumor resection was successfully accomplished in all patients. No bleeding occurred during tumor removal, and no transfusions were given. Despite progress in microsurgical techniques, however, surgery of HBs remains challenging and is often complicated by massive bleeding, especially in large tumors running over more than one segment and in tumors located in the lower spinal region. In a relatively recent report, Conway et al (1) report that, in 66 procedures of CNS HBs, surgical complications occurred in 15% of patients. Cristante et al (32) report that the functional status at discharge of the patients operated for an intramedullary HB was worse in 23%, unchanged in 59%, and improved in 18% of the patients. At follow-up (6–142 months), the status of 9% of the patients was still worse, unchanged in 50%, and better than the preoperative one in 41% of patients. Wang et al (30) report rates of 5% mortality and 13% postoperative complications in 47 patients treated surgically for medullary HB. In the series of Lee et al (14), total removal was achieved in 10 of 14 spinal HBs, and statistical analysis showed that total removal produced a significantly better outcome than incomplete removal. In four of their patients with preoperative embolization, intraoperative bleeding was minimal and total removal was possible. At the last follow-up (15–161 months), eight patients had improved, three were unchanged, and three had deteriorated.

Embolization

The goals of preoperative embolization are control of inaccessible arterial supply and reduction of tumor vascularity. Reduction of intraoperative bleeding is important because excessive bleeding might prevent complete resection of the tumor (14, 24). Conway et al (1) recommend embolization in patients with HBs fed by vessels that are located away from the surgical approach and are not expected to be encountered immediately during resection. In literature, there are few reports concerning preoperative embolization of HBs. Regarding location, most of the embolized lesions were in the cerebellum (1, 23–27) and five lesions in the brain stem, in three anatomic areas: pontomedullary, medullary, and cervicomedullary (1, 24, 30). In spinal locations, eight cases have been reported: five in the cervical spinal cord (23, 28, 29), usually located in the upper cervical cord or at the cervicomedullary junction, and three in the thoracic region (23, 26). Few other cases of embolized spinal HBs have been reported in other articles not specifically dealing with preoperative embolization, and results are not detailed: Lee et al (14) mention preoperative embolization in three cervical HBs and one low thoracic lesion. Conway et al (1) report that, in one sacral HB, embolization arrested symptoms progression 1 month after treatment and surgery was not necessary. After preoperative embolization, facilitated operative excision and complete surgical resection of tumors are usually reported, as observed in our cases (14, 23–26, 29). To the best of our knowledge, there are no articles reporting methods and results of embolization in HBs located in the lower spinal region.

In cerebellar, medullary, and cervical HBs, preoperative embolization is usually performed through the posterior inferior cerebellar artery, more rarely through the anterior inferior and superior cerebellar arteries, the dural branch of the vertebral arteries, the cervical arteries, or the external carotid artery branches (23, 26, 29, 30). In thoracic location, tumors are usually located in the posterior half of spinal cord, and in these cases preoperative embolization of the posterior spinal arteries is reported useful in reducing perioperative hemorrhage (36). Tampieri et al (26) embolized a patient with an intramedullary hemangioblastoma at T4. Eskridge et al (23) performed embolization in two thoracic tumors through the intercostal arteries. Presurgical devascularization of HBs through the ASA has not been reported.

HBs located in the lower spinal region have a predominant vascular supply from the anterior spinal axis, as observed in all our cases. In the thoraco-lumbar region, the artery of the lumbar enlargement, the so-called artery of Adamkiewicz, is a radiculomedullary artery that in 80% of cases originates from T8 to L1 and is more common on the left side. When the origin is above T8 or below L1, an additional supply to the anterior spinal axis exists caudally or cranially, as observed in one of our cases (Fig 2). Vascular supply to the filum terminale is assured only from the anterior spinal axis, and this fact explains the predominant ASA vascular supply in tumors of this region. In our four HBs occurring in the conus medullaris, cauda equina, and filum terminale, devascularization was obtained by embolization performed through the ASA. Embolization of the anterior spinal axis, however, requires caution and can produce severe and irreversible neurologic damage. The size of particles must be chosen in relation to the size of vessels and tumoral structures to occlude. Very small particles can occlude normal small vessels, in particular sulco-commissural arteries, and overly large particles can produce ineffective proximal occlusion and damage of the ASA. Evaluation of hemodynamic changes during embolization is also mandatory. Criteria for ending the endovascular session were the same reported in particles embolization of the spinal arteriovenous malformation (57). In addition, it must be considered that tumors can displace normal structures and, in some cases, the ASA can be displaced off the midline and therefore can be confused with a posterior spinal artery. In these cases, lateral view angiograms are very useful in the exact identification of the spinal arteries.

In our cases, the embolization caused no complications. One patient, however, presented a transient and moderate worsening probably due to edematous changes. The patient rapidly recovered under corticoid therapy. Eskridge et al (23), in their series including seven cerebellar and two spinal HBs embolized preoperatively, describe one patient with a posterior fossa HB and hydrocephalus who worsened clinically within 12 hours of embolization. This event was thought to be caused by obstructive hydrocephalus resulting from tumor swelling. Takeuchi et al (25) report no permanent neurologic deficits after embolization in eight cases of cerebellar HBs, but cerebellar infarction occurred in one patient. Conway et al, in their series—including one cerebellar, one medullary, and two spinal embolized HBs—reported one Wallenberg syndrome deficit in one case of medullary HB. The risk of embolization should be compared with the surgical risk in large and high vascular lesions treated surgically without preoperative devascularization.