Imaging features and atypical signs of symptomatic vertebral haemangioma: a retrospective single-centre analysis of 118 patients

Abstract

Objectives:

Symptomatic vertebral haemangioma (SVH) can present with atypical imaging features. Thus, this study analysed the imaging features of SVH using CT and MRI to improve SVH awareness.

Methods:

We retrospectively analysed CT and MRI characteristics of 118 patients with clinically and pathologically confirmed SVH.

Results:

Overall, 118 patients were diagnosed with SVH, including 79 females and 39 males (mean age, 45.76 ± 16.36 years). The thoracic spine (n = 86) was the most common location of SVH, followed by the lumbar spine (n = 17). Involvement of multiple spinal segments was observed in 15 patients (12.71%). A total of 101 lesions (85.59%) were centred in the vertebral body, 15 lesions (12.71%) were centred in the posterior attachment, and two lesions (2%) were centred in the paraspinal region. CT showed 39 lesions (33.05%) without a typical honeycomb or polka-dot pattern. Compression fracture was observed in 23 patients (19.49%). Extraosseous extension was present in 111 patients (94.1%), and 17 lesions (14.41%) presented with foraminal extension. Epidural bony compression was observed in 46 patients (38.98%). 20 lesions (16.95%) had atypical T₂ weighted MRI signals, and 8 lesions (10.26%) showed atypical enhancement.

Conclusion:

SVH was predominantly located in the thoracic spine. Involvement of multiple segments, posterior attachment localisation, absence of honeycomb or polka-dot signs, compression fracture, and atypical T₂ weighted imaging signals and enhancement were uncommon. Epidural bony compression was not uncommon and has important clinical significance.

Advances in knowledge:

The imaging features of SVHs are not fully understood. We examined the largest series of SVH cases reported to date.

Introduction

Vertebral haemangiomas (VHs) are the most common benign angiomatous spinal lesions with an estimated incidence of 10–12%.¹ VHs originate from endothelial cells and they can be classified into capillary, cavernous, arteriovenous, and venous malformations based on the predominant type of vascular channel.² Most of spinal epidural haemangiomas are the cavernous type. VHs can be classified into four categories based on patients’ lesions and symptoms: mild bony destruction with no symptoms (Type I); bony destruction with pain (Type II); aggressive lesions with epidural and/or soft tissue extension (Type III); and aggressive, neurological deficit with epidural and/or soft-tissue extension (Type IV).³ Based on MRI characteristics, VHs can also be classified as typical, atypical, and aggressive. This classification is correlated with the ratio of fatty to vascular components and interstitial oedema.⁴

Most VHs are incidentally discovered and are asymptomatic; only 1% is estimated to become symptomatic. Symptomatic vertebral haemangiomas (SVHs) are more frequently atypical and aggressive with active behaviours. Patients can present with back pain, radiculopathy, or myelopathy, which are observed in approximately 0.9–1.2% of spinal haemangiomas.^5,6 Approximately, 55% of SVHs present with pain as the only symptom. The other 45% of SVHs are aggressive, with possible invasion of the spinal canal or paravertebral space, which leads to neurological deficit.⁷ SVHs can be fast-growing with extension beyond the vertebral body invading the paravertebral or epidural space, especially in pregnant females.⁸ It is very important to make an early imaging diagnosis to select reasonable treatment.

Due to their low incidence, SVHs have not been fully understood. In this study, we retrospectively analysed the imaging and clinical presentation of 118 patients with SVH who were diagnosed at our department between 2007 and 2019. This is the largest series of symptomatic VH cases reported to date; thus, this study will improve the awareness of this relatively rare disease.

Methods and materials

Subjects

This study was approved by the institutional review board. Clinical manifestation and radiologic information of 118 patients (39 male and 79 female) with an age range of 11–71 years and a mean age of 45.76 ± 16.36 years (median age, 47.5 years) with clinically and/or pathologically confirmed SVH was collected between 2007 and 2019. SVH manifested clinically as pain and discomfort in the neck and shoulder (30 cases), radiculopathy (15 cases), or myelopathy (73 cases) with an onset ranging from 2 months to 2 years.

The study inclusion criteria were as follows: (1) complete CT and MRI examinations; (2) at least one aggressive lesion (S3 if the lesion extended into the epidural and/or paravertebral space or S2 VHs present with pain as the only symptom) if the patient had VH at multiple spinal levels); (3) a definite pathological diagnosis of VH. Patients with incomplete imaging data, a pathological diagnosis of a disease other than VH, or non-aggressive VH (S1 asymptomatic VH) were excluded.⁹

Imaging protocol

118 patients underwent CT at our hospital (Perking university third hospital) using a multidetector CT system (General Electric Light Speed 64). A tube voltage of 120 kV and a tube current of 250 mA were used. For image reconstruction, a standard method was selected. The acquisition slice thickness was 0.625 mm. All data sets were reconstructed with an effective slice thickness of 3 mm.

118 MRI scans were performed at our hospital using a 3.0 T MRI scanner (Siemens Magnetom Trio Tim). The standardised protocol included sagittal T₁ weighted fast spin echo (FSE) imaging (repetition time [TR]/echo time [TE], 618 ms/11 ms), and sagittal (TR/TE, 2852 ms/96 ms) and axial (TR/TE, 2720 ms/88 ms) T₂ weighted imaging. Sagittal proton density imaging (TR/TE, 2800 ms/34 ms) was also performed, with a field of view (FOV) of 190 × 180 mm to 280 × 250 mm. The same slice thickness (3 mm) and slice gap (0.3 mm) were used in all procedures. 78 patients underwent axial and sagittal gadolinium-enhanced (0.2 mmol/kg; injection rate, 2 ml s⁻¹) T₁ weighted fat-saturated (three-dimensional volumetric interpolated breath-hold examination) imaging (TR/TE, 700 ms/11 ms; FOV, 280 × 100 mm; slice thickness, 3 mm; slice gap, 0.3 mm).

Image analysis

Two radiologists specialising in spinal diseases reviewed all images on a picture archiving and communication system workstation monitor. These reviewers identified and characterised the abnormalities by consensus. The following features were assessed on CT and MRI images: lesion location; bone crest; bone destruction (expansive or non-expansive); compression fracture; MRI signal intensity (low, iso, or high relative to the normal spinal cord or muscle); enhancement (similar or lower than vascular enhancement).

Statistical analysis

All analyses were performed using SPSS 19.0 software, and a p-value of < 0.05 was considered statistically significant. Continuous variables are presented as x̄+s and classification variables are presented as number (percentage).

Results

Spinal location and involvement

SVHs were located in the cervical (n = 12, 10.17%), thoracic (n = 86, 72.88%), cervical–thoracic (n = 1, 0.85%), lumbar (n = 17, 14.41%), thoracic–lumbar (n = 1, 0.85%), and sacral (n = 1, 0.85%) spine. Among them, 103 patients (87.29%) had single-vertebra involvement and 15 patients (12.71%) had multiple-vertebra involvement.

CT and MRI imaging findings

CT findings

Of the 118 lesions, 101 lesions (85.59%) involved both the vertebral body and the appendages simultaneously. 15 lesions (12.71%) were located mainly in the vertebral appendages, while 2 lesions (2%) were located mainly in the paraspinal region. 35 lesions (29.66%) involved the vertebral arch unilaterally and 72 tumours (61.02%) involved the vertebral arch bilaterally. CT images showed vertical striations and a honeycomb appearance and/or a polka-dot pattern in 79 lesions (66.95%) and an atypical appearance in 39 lesions (33.05%). Osteolytic or cystic destruction was present in 21 lesions (17.80%) without bone crest formation(Figure 1a). No apparent bone destruction was seen in four patients (3.39%). A total of 101 lesions (85.59%) perforated the vertebral cortex, and 60 lesions (50.85%) showed expansive bone destruction. Pathologic vertebral fracture was observed in 23 patients (19.49%). Epidural bony compression was observed in 46 patients (38.98%) (Figure 2a, b)(Table 1).

Figure 1.

A 57-year-old male with neck pain and symptomatic vertebral haemangioma. An axial CT image (a) shows osteolytic bone destruction without a typical polka-dot appearance. A C2 lesion shows a hypointense signal on sagittal T₁ weighted magnetic resonance images (b) and a hyperintense signal on T₂ weighted images (1C) due to a low fat content and abundant vasculature. The lesion shows marked and inhomogeneous enhancement on contrast-enhanced fat-saturated T₁ weighted magnetic resonance images.

Figure 2.

A 32-year-old male with back pain and T4 symptomatic vertebral haemangioma. Sagittal CT (a) shows a honeycomb appearance in T4 symptomatic (thin arrow) and T5 asymptomatic (thick arrow) vertebral haemangioma. Axial CT (b) shows a polka-dot appearance and involvement of the posterior appendages. Bone hypertrophy can be seen in the posterior appendages of T4, which caused epidural bony compression. A sagittal T₁ weighted magnetic resonance image (c) shows a low signal intensity in the entire T4 vertebral body and T5–6 hyperintensity. An epidural soft mass with a low signal intensity on T₁ weighted images and a high signal intensity on T₂ weighted images extends from T3–5 and compresses the spinal cord. On sagittal T₂ weighted magnetic resonance images (d), the entire T4–6 vertebral body and the T4 posterior appendages show a high signal intensity. On axial T2 images (e), a soft mass extends into the epidural spinal canal with intervertebral foramen enlargement. On post-contrast fat-saturated T₁ weighted magnetic resonance images(2F), T4–5 and the epidural mass extending from T3–5 demonstrate vivid contrast enhancement.

Table 1.

Imaging characteristics of SVH

Imaging characteristics	Number	Percentage
Involved spinal segment
Single vertebra	103	87.29%
Two segments	7	5.93%
Three or more segments	8	6.78%
CT characteristics
Bone destruction
Vertebral body	101	85.59%
Appendages	15	12.71%
Paraspinal	2	2%
Honeycomb or polka-dot	79	66.95%
Vertebral fracture	23	19.49%
Intervertebral foramen enlargement	17	14.41%
Ossification	46	38.98%
MRI characteristics
T2-weighted images
Hyperintense	98	83.05%
Isointense	12	10.17%
Hypointense	3	2.5%
Heterogeneous	5	4.2%
Enhancement
Significant	70	89.74%
Mild or moderate	8	10.26%

MRI findings

SVHs showed hypointense (n = 96, 81.36%), hyperintense (n = 11, 9.32%), isointense (n = 8，6.78%), or heterogeneous (n = 3, 2.54%) signals on T₁ weighted imaging. These tumours were hyperintense (n = 98, 83.05%), isointense (n = 12, 10.17%), hypointense (n = 3, 2.5%), or heterogeneous (n = 5, 4.2%) on T₂ weighted images (Figures 1b, c and 2c–e).

After gadolinium administration, 70 lesions (89.74%) showed significant enhancement, and 8 lesions (10.26%) showed mild or moderate enhancement (Figures 1d and 2f)(Table 1).Extraosseous extension was present in 111 patients (94.1%). Extension into the epidural spinal canal was present in 31 patients (27.93%), extension into the paravertebral space was present in 5 patients (4.5%), and extension into both the spinal canal and paravertebral space was present in 75 patients (67.57%). Foraminal extension was present in 17 patients (14.41%). Neurologic compression was caused by epidural soft tumour masses in 61 patients (51.69%) and by a combination of bony compression and soft-tissue lesions in 46 patients (38.98%).

Atypical radiological feature analysis

53 lesions (44.92%) had at least 1 unusual radiological feature. 15 lesions (12.71%) had multiple-vertebra involvement. 15 lesions (12.71%) were localised in the appendages, and 2 lesions (2%) were located in the paraspinal region. 18 lesions (15.25%) had osteolytic bone destruction, while 3 lesions (2.54%) did not have obvious bone destruction. 23 lesions (19.49%) presented with pathologic vertebral fracture. 46 patients (38.98%) had epidural bony compression. 20 lesions (16.95%) had atypical T₂ weighted MRI signals. Eight lesions (10.26%) showed mild or moderate enhancement. 17 (14.41%) presented with foraminal extension (Table 2).

Table 2.

Typical and atypical imaging characteristics of SVH

Imaging characteristics	Typical	Atypical	Percentage of atypical signs
Location	101	17	14.41%
Honeycomb/polka-dot	79	29	30.5%
Pathologic vertebral fracture	95	23	19.49%
Intervertebral foramen enlargement	101	17	14.41%
T2 weighted imaging signal	98	20	16.95%
Enhancement	70	8	10.26%

SVH, symptomatic vertebral haemangioma.

Pathological results, misdiagnosis analysis, and differential diagnosis

20 patients were misdiagnosed, including 5 patients misdiagnosed with plasmacytoma owing to the hypointense signal on T₂ weighted images; 3 patients misdiagnosed with chondroma owing to the high signal on T₂ weighted images and atypical osteolytic bone destruction; 3 patients misdiagnosed with neurilemmomas owing to intervertebral foremen enlargement; 3 patients misdiagnosed with lymphoma owing to the longitudinal growth pattern; 2 patients diagnosed with metastatic tumour owing to presence of multiple lesions; 2 patients diagnosed with giant cell tumours owing to a heterogeneous signal on T₂ weighted images (Table 3); 1 patient misdiagnosed with a bone cyst owing to a hyperintense signal on T₂ weighted images and mild enhancement; and 1 patient (a 15-year-old boy) misdiagnosed with Langerhans cell histiocytosis. Pathological results showed 105 cases (88.98%) of cavernous haemangioma, 6 cases (5.08%) of mixed haemangioma, 4 cases (3.39%) of capillary haemangioma, and 3 cases (2.54%) of epithelial haemangioma.

Table 3.

SVH and differential diagnosis

	SVH	Lymphoma	Neurilemmoma	Chondroma	Plasmacytoma	Giant cell tumour	Metastatic tumour
CT findings
Location
Vertebral body	+++	++	+	++++	+++	+++	++
Posterior appendage	+		+		+	+	+++
Bone crest	+++	+	+	+	++	+	+
Intervertebral foramen enlargement	+	+	++++	+	+	+	+
Honeycomb or polka-dot	+++	+	+	+	+	+	+
MRI findings T2WI
Isointense	++	++	+	+	++	++	+
Hyperintense	+++		+	++	+	+	++
Heterogenous	+		+++	++	+	++	++

MRI, magnetic resonce image; SVH, symptomatic vertebral haemangioma; T₂WI, T₂ weighted imaging.

+: <25%; ++: 25%–50%; +++: 50%–75%; ++++: >75%

Discussion

SVHs account for approximately 1% of all VHs.⁶ The present study enrolled 79 females and 39 males, and the female to male ratio was 2, which is higher compared with the ratio of 1.5 in the literature.¹⁰ VH is more common in females compared with males. Some scholars believe this may be due to a relationship between the level of oestrogen and VH. Pregnancy-related SVH also supports the hormone-related theory.¹¹ Two cases of SVH in this group occurred during pregnancy. Some scholars believe that VH may be related to haemodynamic changes, vascular endothelial growth factors, and hormones.¹¹ SVHs are mainly found in middle-aged patients. The mean onset age in the present study group was 49.27 ± 12.36 years (median, 48 years), which is similar to that reported in previous literature.¹²

SVHs occur most frequently in the thoracic spine, followed by the lumbar spine. In this study, 72.88% of lesions were located in the thoracic spine and 14.41% of lesions were located in the lumbar spine, which is similar to previous literature.¹³ Typical haemangiomas show vertical striations and a characteristic honeycomb or polka-dot appearance on CT, regardless of the presence or absence of symptoms.¹⁴ This honeycomb/polka-dot appearance was observed in 66.95% of SVHs in this study. Atypical SVHs may show osteolytic, cystic, or mild destruction without a honeycomb or polka-dot appearance; these types of lesion accounted for 33.05% of lesions in this study. Without a typical appearance on CT, it is difficult to make a correct diagnosis; thus, CT needs to be used in combination with MRI.

SVHs may show aggressive growth patterns.¹⁵ In this study, 85.59% of vertebral body SVHs extended into the appendages and 12.71% of appendage SVHs extended into the vertebral body. Equally, SVHs can extend into the spinal canal and paravertebral region, which was observed in 67.57% of patients in this study. An aggressive growth pattern may lead to spinal cord compression or neurological deficit. Neurological deficit may be caused by soft-tissue or bone compression.¹⁶ Soft-tissue compression was observed in 51.69% of patients and resulted from SVH extension into the epidural and foraminal spaces. In 38.98% of patients, a combination of bony and soft-tissue compression was observed, which resulted from hypertrophy or ballooning of the posterior cortex of the vertebral body and/or enlargement of the laminae and facets subsequent to VH development.⁵ CT can reveal ossification in the intraspinal or paraspinal regions and bone compression; however, bony compression has not been mentioned in previous radiological reports. In this study, we analysed this sign, which should be mentioned in radiological reports and may provide detailed information for orthopaedic surgeons. Compression fracture was observed in 19.49% of VHs; compression fracture can lead to vertebral instability and spinal stenosis. Decompression surgery, such as intralesional vertebrectomy and total en-bloc spondylectomy, have demonstrated acceptable results in patients with compression fracture.^17,18

Compared with asymptomatic VHs, symptomatic VHs contain less fat, have a richer vasculature, and obvious interstitial oedema.¹⁹ SVHs showed a low signal on T₁ weighted MRI in 81.36% of cases and a high signal on T₂ weighted MRI in 83.05% of cases in this study. These signals reflect the abundant vasculature and lower composition of fat. Approximately, 10.17% of SVHs appeared isointense on T₂ weighted images, which reflects the relatively lower lipid and vascular content. Occasionally, symptomatic VHs produce a hypointense or heterogeneous signal on T₂ weighted images. This is related to bone sclerosis, internal haemorrhage, or compression fracture. In this study, 91.7% of SVHs show significant enhancement, which reflects their abundant vasculature. SVHs can extend into the extradural space, which can in turn enlarge the intervertebral foremen. In this study, 14.41% of lesions caused foremen enlargement mainly due to the relatively large intervertebral space and tumour spread.²⁰ SVHs with intervertebral foremen enlargement need to make differential diagnosis with neurilemmomas. Neurogenic tumours are prone to cystic changes or necrosis and tend to demonstrate in-homogeneous enhancement.^21,22

SVHs can also show a longitudinal growth pattern that can involve multiple spinal segments. In this study, 12.71% of SVHs affected multiple vertebras. It is important to achieve a differential diagnosis of lymphoma.²³ SVHs produce a stronger signal compared with lymphoma on T₂ weighted images. Vertical striations or bone crests can not only be seen in VHs, but also in plasmacytoma, giant cell tumour, and metastatic tumours.²⁴ Both plasmacytoma and giant cell tumour show iso- or hypointense signals on T₂ weighted images, which is different from SVHs.²⁵ Percutaneous CT-guided biopsy should be performed due to overlap in imaging features.

This study has some limitations that should be highlighted. First, this was a retrospective study performed at a single clinic. Second, the relationship between radiological and pathological appearances requires further comparative research. Third, long-term follow-up is needed to observe tumour outcomes.

Conclusion

The imaging features of SVHs can be atypical, and more than 40% of SVHs demonstrate at least one unusual radiological feature. SVHs are relatively rare in the appendages or paraspinal region. It is also rare for SVHs to involve multiple spinal segments. Atypical SVHs show osteolytic destruction without a honeycomb or polka-dot pattern on CT. More than 80% of SVHs show a low signal on T₁ weighted MRI and a high signal on T₂ weighted imaging, reflecting a richer vasculature. SVHs can extend from the bony compartment into the epidural or paravertebral space with foraminal enlargement. In addition to soft-tissue compression, intraspinal or paravertebral bone formation can cause bone compression, which is not uncommon and has important clinical significance. Surgery is required to relieve compression; thus, radiologists should pay particular attention to this sign.