|Home | About | Journals | Submit | Contact Us | Français|
Spinal angiolipomas are benign uncommon neoplasm composed of mature lipocytes admixed with abnormal blood vessels. They account for only 0.04–1.2% of all spinal tumors. We report two cases of lumbar extradural angiolipoma and review previously reported cases. We found 118 cases of spinal epidural angiolipoma (70 females and 48 males; age range 1.5–85 years, mean 44.03) spanning from 1890 to 2006. Prior to diagnosis 40.6% of the patients had weakness of the lower limbs. The interval between the initial symptoms and tumor diagnosis ranged from 1 day to 17 years (mean 20.2 months). Except for four cases diagnosed at autopsy, 109 patients underwent surgery and gross-total resection was performed in 79 cases (72.4%). Spinal angiolipomas are tumors containing angiomatous and lipomatous tissue, predominantly located in the mid-thoracic region. All angiolipomas show iso- or hyperintensity on T1-weighted images and hyperintensity on T2-weighted images and most lesions enhance with gadolinium administration. The treatment for spinal extradural angiolipomas is total surgical resection and no adjuvant therapy should be administered.
Spinal angiolipomas are benign tumors composed of both mature fatty tissue and abnormal vascular elements that represent a distinct clinical and pathological entity. These lesions are most commonly found in the subcutaneous tissue of the trunk and extremities, but other sites have been reported as well. They account for 0.04–1.2% of all spinal axis tumors, and are predominantly found in the epidural space, where they represent 2–3% of spinal tumors. Intracranial angiolipomas are extremely rare (4). It is not clear from the literature if the lesions originate in the epidural space and “infiltrate” the bone, if they originate in the bone and spread to the epidural space, or if they arise in both compartments simultaneously .
A review of the literature since 1892 to the present day revealed 123 cases of spinal angiolipomas (118 extradural and 5 intradural). The purpose of this paper is to report our two cases of SAL and review the pertinent literature.
A 16-year-old male with a 6-month history of low back pain, and worsening of the symptoms in the weeks prior to diagnosis presented to the study. The pain radiated down the posterior aspects of both legs, and increased with physical movement and walking. A neurological examination revealed spasm of the lumbar paravertebral muscles, bilateral Lasègue signs at 60°, and absence of the Achilles reflexes.
Noncontrast and gadolinium-enhanced MR studies of the lumbar spine showed an extradural lesion at L5–S1 level. On T1-weighted images the lesion was slightly inhomogeneous (Fig. 1). The T2 signal was heterogeneously hypointense.
The patient was submitted to a L5–S1 laminectomy with exposure of the dura and bilateral sacral roots. A 3 × 2 cm high vascularized mass was found pushing the duramater, and was totally excised with preservation of the roots. The patient was asymptomatic at 12-month follow-up.
Microscopic examination revealed mature adipose tissue and abnormal vascular channels with blood vessels of variable caliber and wall thickness which mostly filled with erythrocytes.
A 45-year-old woman with numbness in both feet, leg weakness of 6 months which commenced during the fifth month of gestation complained of a sudden increased numbness and weakness in the right leg and aggravated low-back pain on the day before admission.
Examination revealed that she was obese, bilateral Lasègue sign at 30°, bilateral absence of the Achilles tendon reflex and perineal hypoesthesia.
MRI showed a posterior epidural mass extending from L5 to S1, predominantly on the right side (3 × 1.5 cm). The lesion was slightly inhomogeneous epidural mass, which is moderately hiperintense relative to spinal cord in T1 and T2 weighted images. Disc hernia was initially diagnosed, hence no contrast was administered. The L5-S1 disc showed a mild protrusion (Fig. 2). The patient was submitted to a L5–S1 foraminotomies, a slightly protruded L5–S1 disc and a soft brown extradural mass was observed compressing the L5 and S1 roots; the operative field was enlarged, and the mass was excised from the dorsal dura without difficulty. The postoperative course was uneventful and at 6-month follow-up examination the patient had perineal hypoesthesia.
Histological examination of the tumor showed mature adipose tissue interlaced with numerous blood vessels ranging from distended capillaries, venules and blood-filled lakes to small arteries, no mitotic figures were observed, and angiolipoma was diagnosed (Fig. 3).
All reported cases of spinal angiolipoma reported between 1890 and 2006 were reviewed; only cases in which authors reported details about the relevant topics were included for analysis (see Table 1). For example, a review of 48 angiolipoma cases, 8 of which were surprisingly neither described in other series nor included in the bibliography, were omitted .
Age and sex. Most angiolipomas occurred in adults with an age range of 1.5 to 85 years (mean 44.03 ± 5.9), and peak incidence was 46 (7 cases). A female predominance (70 females, 48 males) was observed, and mean age was 43.6 ± 5.1 for males, and 45.1 ± 6.1 for females.
Revealing symptoms. Initial symptoms were progressive or sudden weakness in 48 patients (40.6%), thoracic or lumbar pain in 27 cases (22.8%), alterations in sensation (numbness) in 24 (20.3%), sciatica uni or bilateral in 6 (5%), gait disturbance in 2 (1.7%), 1 bleeding to lumbar puncture , and 1 congenital . In nine cases initial symptoms were not reported. Eight patients with a relapsing clinical course presented paraparesis during pregnancy or puerperyum . In five cases the patients were overweight and five patients were administered steroids.
The interval between the initial symptoms and tumor diagnosis ranged from 1 day to 17 years (mean interval 20.2 ± 9.6 months); in 16 cases (13.5%) the duration was not mentioned or not specified. The duration of symptoms was longer in patients with lumbar pain (53.1 ± weeks), whereas the shortest duration of symptoms was observed in four cases involving sudden motor deficit (<1 week) [5, 22, 26].
Location. Tumors were located as follows: two in the cervical region [16, 36], 93 in the thoracic region, 12 in the lumbar, 1 in the sacral , 5 in the cervico-thoracic region, 3 in thoracic-lumbar area, and 3 in the lumbo-sacral region. Of the 108 cases where the location was specified, 14 comprised one vertebral segment; 28 two segments, 22 three segments, 19 four segments, and 25 more than four segments.
Size. As for tumor size, data were provided for 21 cases, i.e., 10 cases described three dimensions, 5 two dimensions were reported, and only one dimension was specified in 6.
Biochemical studies. Cerebrospinal fluid samples were analyzed in 11 cases with slight protein concentrations (range 0.70–1.41 mg/dl), mean: 1.0 mg/dl, being observed in 5 cases.
Neurophysiological study. Only two neurophysiological studies were reported: the case reported by Bailey et al.  where the lower limb somatosensory evoked responses were normal, and the case published by Nishiura et al.  where the electromyogram showed a neurogenic pattern in the paravertebral muscles from T8 to T10 bilaterally.
Radiology. X-rays of the spine essentially show no bone abnormalities. The most frequent alterations were erosion of the walls of the vertebras (bodies and pedicles) with enlargement of the interpeduncular distances [1, 57, 75] primarily in infiltrated tumors. Other cases were associated to scoliosis [10, 56, 90], osteoporosis  or Klippel-Feil syndrome .
Myelography was performed in 43 cases, and all showed a characteristic posterior extradural spinal mass with partial or complete obstruction. In 20 cases the myelography was assisted with TC.
Angiography. Selective spinal angiography was performed in only two cases with prominent vascularity in the lateral aspect of the T9 vertebral body in one case , and in the T6 in the other (4). In both cases the tumors were supplied by a branch of the intercostals artery.
Computed tomography was performed in 34 cases (21 cases CT-myelography). CT scan demonstrated the presence of a space-occupying lesion in 21 cases as well as 6 CTs were normal, 7 were unspecified, and in a further 2 cases disc hernia was observed [6, 56]. Moreover, CT revealed tumor calcification in two patients [1, 73], trabeculation of the vertebral body in two cases [90, 93], erosion of the vertebral body and pedicle in another , a paravertebral thoracic component in two cases [1, 90] and association with vertebral hemangioma in other  The tumor was hypo dense with the cord in 16 patients (between −20 to −72 HU), isodense in three patients [41, 52, 90], slight hyperdensity in three cases [63, 74, 77], and heterogeneous in only one case  where a reference to density was available. The epidural tumor did not show contrast enhancement in one case , demonstrated minimal contrast enhancement in two patients [52, 93] and sharp contrast gain in three [57, 63, 76].
MRI was performed in 47 cases (since 1988 onwards), but sufficient data were provided in only 31 cases. In the T1-weighted sequence, the tumor was as isointense as the medular conus in 7 cases, 18 were hyperintense, 5 were hypointense, and one was heterogeneous . In the T2-weigthed sequence, the tumor was hyperintense compared to CSF in 25 cases; two were isointense [75, 93], two hypointense [1, 17] and two mixed signal [74, 95]. After Gd injection, 13 cases presented remarkably homogeneous enhancement, 8 were heterogeneous, and in the remaining 8 cases low enhancement. Disc hernia was observed in one case , infiltration of the vertebral body and/or extending to the pedicle in three cases [4, 40, 75], and a paravertebral thoracic component in one case .
Treatment. Except for 4 cases diagnosed at autopsy, 109 patients underwent surgery and gross-total resection was carried out in 79 cases (72.4%), subtotal resection in 6 (5.5%), biopsy sampling in 1 case , but no mention of the extent of removal in 22 (20.1%), and no mention of surgery in 5 cases. Rabin et al.  performed preoperative embolization prior to T9 corporectomy and resection. Finally, radiotherapy was undertaken in only three cases [5, 15, 28].
Pathological features. Angiolipoma was diagnosed in all but five cases [67, 78, 93], and the most common feature was a lesion with two distinct components: mature adipocytes with peripheral, sometimes indented nuclei, and a single lipid droplet intermingled with abnormal blood vessels of varying diameter. In the case reported by Fourney et al.  mild atypia was observed in the endothelial cells. In two cases, the angiolipoma was associated with another tumor, osteochondroma , and glioblastoma . Inmunoshistochemical assay was performed in a few cases [60, 79] and a low proliferation rate was found on Ki-67 [22, 68]. Samdami et al.  observed a positive stain for CD31, Factor XIIIa, and Factor VIII, and negative for Glio Fibrilar Acid Protein.
Outcome. The postoperative course was diverse ranging from 3 months to 10 years. Overall, there were two deaths (1.8%) in patients with non-infiltrating angiolipomas [11, 29] and none in the infiltrating subgroup. Though the results of surgery are difficult to assess due to the variations in terminology in the literature, most patients were reported to have improved postoperatively.
Angiolipomas are benign tumors which usually appear as painful subcutaneous nodules, particularly in the forearm, trunk, or neck but occur rarely in the spinal canal . Spinal angiolipomas account for 0.14–1.2% of all spinal tumors, about 2–3% of extradural spinal tumors, and 16–35% of spinal lipomas [19, 26, 87]. Prior to the reports of five intramedullary angiolipomas by Palkovic et al. , Preul et al. , Maggi et al. , Klisch et al.  and Weill et al. , these were thought to occur exclusively in the epidural space. Three cases of intracranial angiolipomas have been described: two were parasellar, adherent to the dura and in intimate contact with the cavernous sinus [89, 97], and the other located in the cortico-subcortical area of the left frontal lobe .
The first case of spinal angiolipoma was reported by Berembruch in 1820 in a 16-year-old boy with an unsuccessfully, surgically treated thoracic mass which was confirmed at autopsy . In 1901 Liebscher was the first to describe a spinal angiolipoma : however, the term angiolipoma was established in 1960 by Howard and Helwig as an anatomopathological entity containing mature fat cells and proliferating vessels .
The histogenesis of angiolipomas is poorly understood and several theories have been advanced. Ehni and Love postulate that, following some undefined stimulus, angiolipomas arise from pluripotential mesenchymal stem cells by divergent differentiation along both adipose tissue and angioid lines (smooth muscle and vascular endothelium) . Willis  simply considered them to be a congenital malformation or a benign hamartoma, but most would now agree that angiolipomas and mesenchymal hamartomas arise from primitive mesenchyme [11, 42]. Hemangiomas and lipomas may represent a spectrum within which angiolipomas constitute an intermediate entity . The more invasive infiltrating type of spinal angiolipoma would then represent a shift towards the hemangioma end of the spectrum . Pagni and Canavero support the theory of abnormal developmental origin, based on their findings of spinal angiolipomas in patients with birth defects outside the CNS . Spinal lipomas differ from angiolipomas in that they are most commonly found in the lumbosacral region, rather than mid-thoracic region, and may be associated with dysraphic abnormalities [21, 46]. Some consider spinal angiolipomas to be more aggressive than spinal lipomas .
Pregnancy was an aggravating factor in 11 (15.7%) of the 70 female patients reported. Termination of pregnancy, as in the case reported by Preul et al.  may bring about regression of symptoms [8, 18]. Pregnancy may interfere with spinal venous drainage and increase epidural venous pressure from compression of pelvic and abdominal veins; the redistribution of blood, the increase in extracellular fluid, and the forceful Valsalva maneuvers associated with vaginal delivery may acutely exacerbate symptoms [8, 73]. Other possible factors reported are vascular steal phenomenon as a cause of spinal cord ischemia  and pulsatile compressive effect on the adjacent cord because of its vascularity . Hormonal influences and increased adiposity produced by pregnancy may also cause an increase in the size of spinal angiolipomas, much as obesity may lead to symptoms induced by increasing their fatty component as occurs in patients who receive steroid treatment or are overweight [31, 46]. Tumor thrombosis and/or hemorrhaging may cause a sudden deterioration in condition, as occurred in two cases reported by Labram et al.  and Anson et al. .
Macroscopically, the tumor is an encapsulated or uncapsulated, reddish soft mass extending into the extradural space of the spinal canal. They are subdivided into two types which should be considered and treated differently: non-infiltrating and infiltrating, the latter extending into the vertebral body . In 1966, Gonzalez-Crussi  described the first case of vertebral “infiltrating” angiolipoma and stated that these tumors should be distinguished from angiolipomas described by Howard and Hellwig . Lin and Lin  established infiltrating angiolipoma as a separate entity, justifying their decision on differences in clinical course, pathological picture, and prognosis from those of “plain angiolipomas”. Histologically, they are composed of mature adipose tissue and blood vessels, features of which are described as being either normal or mimicking capillary angioma, cavernous angioma or arteriovenous malformations. The fatty tissue is of the adult type and shows no remarkable findings. The ratio of fat to vessels is variable and ranges from1:3 to 2:3. Tumors with an abundance of smooth muscle proliferation are further subclassified as angiomyolipomas. A thin capsule, often defective in many areas, may surround the lesion. Secretory activity has been described in one case with lipid-like material in perivascular granules . In two cases described by Pearson et al. , additional osteoid tissue was observed. Low proliferation rates are found on Ki-67 immunostaining . Neither atypia, pleomorphism or mitotic figures or karyotypical abnormalities were found . Sciot et al.’s  analysis of the karyotype in 20 angiolipomas revealed that all were normal; the authors speculate that a normal karyotype supports the theory of angiolipomas as a reactive or hamartomatous lesion. They also suggest that the vascular proliferation could be a primary event in the development of these tumors, and the adipose tissue, a secondary, albeit prominent, component.
From a clinical point of view, extradural angiolipomas do not differ from other benign space-occupying spinal lesions. Subjective complaints are mostly of sensory disorders and motor deficits below the level of the lesion  that often progress to weakness in the lower limbs for long periods with sphincter dysfunction in the later stages .
In most instances spinal radiographs were normal with only a few reports of bone involvement, i.e., bone erosion had a coarsened trabecular pattern more typical of vertebral hemangioma in two infiltrating angiolipomas [41, 93], but no calcifications were reported. Myelography was the most common radiologic technique for identifying the tumor prior to 1988; the findings, however, were not specific and showed partial or total blocking of the contrast column that was also characteristic of other extradural lesions compressing the spinal cord . Spinal angiography was undertaken in two cases revealing prominent vascularity in the vertebral body  or the lateral aspect of the T9 vertebra in the other , both of which were supplied by a branch of the intercostals artery.
Rubin et al.  and Weil et al. , report that CT scan revealed dense fatty tissue in most of the extradural angiolipomas. Though CT scan usually demonstrates a hypodense lesion, some tumors are isodense  or hyperdense depending on the extent of the vascular component or the presence of calcification. Pagni and Canavero  suggest that CT is not specific for spinal epidural angiolipomas and could be misleading.
MRI is the most valuable radiological technique for diagnosing spinal angiolipomas , which are typically hyperintenses on non-contrast T1-weighted images owing to their fatty content . Provenzale and Mclendon  showed that large hypointense foci observed within spinal angiolipomas on non-contrast T1-weighted images are correlated with increased vascularity and most lesions enhance with gadolinium administration whereas T2-weighted imaging can be variable, but are usually hyperintense [52, 63, 71, 90]. Non-infiltrating angiolipomas are generally located in the posterior portion of the epidural space and are well delimited from the surrounding tissue and can usually be removed easily by laminectomy.
Total removal of infiltrating angiolipomas, that often involve the body of the vertebral than the posterior arch, has been recommended using the anterolateral approach  and stabilization of the affected vertebrae is desirable. Some authors believe that the tumor-invaded vertebral body should be preserved, because analogous to vertebral hemangiomas, spinal angiolipomas may not enlarge .
In three cases, postoperative radiotherapy was administered following a partial excision owing to concerns of potential malignancy [5, 15, 28]. No adjuvant radiation should be applied to patients with this benign pathological entity, since even in the infiltrating group, the prognosis is very good. Most authors report good outcomes after surgical excision of spinal angiolipomas in spite of severe preoperative neurological deterioration . Only one case of recurrence of an angiolipoma has ever been reported with successful surgery 12 years after the first intervention  and none has been reported in all the other cases of infiltrating angiolipomas, even in cases in which a complete removal could not be attained . A total of six deaths have been reported, there of which date to the late 1800s or early 1900s with no clear indication as to the cause of death [8, 12, 36, 44]. More recently, one patient who had never undergone surgery died from a cervical glioblastoma  and the other from a respiratory infection .
Spinal angiolipomas are rare benign tumors, of angiomatous and lipomatous tissue, predominantly located in the mid-thoracic region. Although the etiology of angiolipomas remains unknown, they may result from the abnormal development of the primitive, pluripotential mesenchyme from which adipose tissue, smooth muscle, and vascular endothelium arise, or they may be hamartomatous in nature.
All angiolipomas show iso- or hyperintensity on T1-weighted images and hyperintensity on T2-weighted images and most lesions enhance with gadolinium administration. The treatment for spinal extradural angiolipomas is total surgical resection and no adjuvant radiation therapy should be administered to patients with this pathological entity since prognosis, even if infiltrating, is excellent.
The authors thank Mr. Romen Das Gupta for his assistance in the preparation of this paper.