An Institutional Review of Tuberculosis Spine Mimics on MR Imaging: Cases of Mistaken Identity
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.273630
Source of Support: None, Conflict of Interest: None
Keywords: Mimics, MRI, Pott's spine, spine, spondylodiscitis, tuberculosis
”Percival Pott” was the first person to describe spinal tuberculosis (TB) in 1779.
TB is caused by the mycobacterium TB. Spinal TB accounts to 2% of all cases of TB. The lower thoracic and lumbar vertebrae are the most common sites followed by mid-thoracic and cervical vertebrae. MRI is recommended when there is a clinical suspicion of spondylitis because early diagnosis would avoid severe spinal or neurological complications. It also provides anatomic localization of the disease in different planes, enables early detection of disk and bone destruction, delineates extension in bone and soft tissues, and determines the skip lesions in noncontiguous spinal TB. Also, the surgical choice between anterior and posterior decompression would be established on the basis of MRI. Advanced spine MR imaging techniques, including diffusion, diffusion tensor imaging, and perfusion imaging are of particular interest when conventional CT and MR imaging findings are negative/equivocal. These advanced imaging techniques show promising results in group comparison analyses. They are limited due to cost and time constraints. There is still insufficient evidence to conclude whether these advanced techniques can be used for routine clinical use and diagnosis.
MRI characteristics in spinal TB are detailed below.
Four different MRI patterns of disease are found in vertebral TB [Figure 1]a and [Figure 1]b: paradiskal, anterior, central, and posterior lesions.
A significant number of authors have proposed that the onset of the spinal infection is paradiskal. It begins in the vertebral metaphysis, eroding the cartilaginous endplate, leading to disk space narrowing due to the infection itself or to disk herniation into the endplate. Due to osseous resorption, there is endplate demineralization with the loss of cortical bone.
On MRI, it shows as low signal on T1-weighted sequences, whereas it is hyper intense on T2-weighted imaging and STIR (Short TI Inversion Recovery) sequences. This is due to the replacement of bone marrow by inflammatory exudates, cells, and hyperemia. With gadolinium, the infected disk shows enhancement, allowing differentiation of the non-affected part. This pattern can mimic many other inflammatory and infective pathologies on MR imaging.
Abscess and collections show high signal intensity on T1-weighted images with low signal intensity on T2-weighted sequences, as compared with the cerebrospinal fluid signal. However, mixed signal intensity on both T1- and T2-weighted images can be observed in phlegmons. There is a thin smooth wall contrast enhancement in cases of abscess, whereas the phlegmon shows uniform enhancement.
In the anterior pattern, the infection starts in the corner of the vertebral body and spreads to the adjacent vertebrae underneath the anterior longitudinal ligament. Subligamentous dissemination strips the periosteum and the anterior longitudinal ligament from the vertebral surface and makes the avascular vertebrae vulnerable to infection. This appearance may mimic a bone tumor.
Finally, the progression of the bone lesion produces anterior vertebral collapse, leading to kyphosis, which is typically seen in Pott's disease.
MRI findings in the anterior pattern include the presence of a subligamentous abscess with contrast enhancement, preservation of the disks, and abnormal signal involving multiple vertebral segments with heterogeneous signal intensity.
In central pattern, the infection affects one single vertebral body and if the infection progresses, the whole vertebral body collapses. This is commonly confused with malignancy. The disk remains healthy as the nutrition is provided from the adjacent vertebra. Infection progresses to the contiguous vertebra or to the paraspinal space.
MRI in this stage shows hypointense T1-weighted signal in a single vertebra and vertebral collapse with disk preservation.
The posterior pattern is the least common of all the types, as TB rarely affects the vertebral arch. When affected, the radiological pattern is frequently difficult to differentiate from metastasis, especially when disk space is preserved. This occurs in only 5% of cases, and biopsy may be necessary for the confirmation of the diagnosis.
In recent times, many neoplastic and non-neoplastic pathologies resemble TB spine radiologically with overlap of clinical manifestations ranging from back pain, fever, paraparesis, kyphosis, and sensory disturbances. These pathologies could also coexist with TB spine, adding to the diagnostic chaos. As the management of many of these “TB spine mimics” is different, the goal of this review is to explore strategies to minimize over-diagnosis of TB spine on imaging by considering clinical correlation, imaging follow-up to identify the right diagnosis and thus limit unnecessary biopsy. We present 11 different pathologies that mimic TB spine on MRI and discuss their clinical and MRI findings. To the best of our knowledge, there have been only case reports in the literature on TB spine mimics. So, this is the first comprehensive review paper on the subject.
Finally, we summarize those key MRI findings which if present should suggest a diagnosis other than TB spine. The authors not only review the established signs but also propose strategies for error reduction in MR imaging by providing a table as a checklist to minimize over diagnosis of TB spine.
After reviewing MR reports of all the MRI spine studies from picture archiving and communication systems at our institute to identify patients with spinal TB between January 2012 and May 2016, only the MRI spine reports that showed a differential diagnosis or a diagnosis of TB were identified. Out of these 306 cases, we filtered 78 cases that closely mimicked TB spine; which on careful clinical, radiological analysis and follow-up could have been interpreted as an alternate diagnosis other than the TB spine. Definitive clinical diagnosis was established in all these included cases either by histopathology, lab investigations, and follow-up MRIs.
MR image analysis
The MR images were retrospectively reviewed by the radiologists. All individual spine MR images were evaluated for the location and distribution with contrast enhancement pattern. Clinical history and laboratory parameters of these cases were reviewed. CT images with bony reconstructions of the same patients were also reviewed wherever available.
All patients were scanned on a 1.5 T MRI system using a dedicated spine coil with T1, T2 weighted imaging, and FLAIR sequences. Dedicated axial cuts were performed on identified locations of the spine. A contrast study was done wherever necessary with intravenous gadolinium (1 ml per 10 kg of body weight) in the respective patients.
There were 78 cases that closely mimicked TB on MRI in 52 males and 26 females, ranging in age from 20 to 65 years with varying clinical presentations. Of the 78 patients with TB mimics, we categorized them into 8 different etiologies on the basis of clinical history, laboratory results, MRI, and CT findings [Table 1].
There was a spectrum of pathologies that mimic TB spine, categorized under the headings of infective, inflammatory, neoplastic, trauma, metabolic, degenerative, neuropathic, and miscellaneous diseases. In our study, pyogenic spondylitis followed by degenerative disc disease was the most common clinical entity that mimicked TB spine. We also discuss the TB mimics in respective clinical scenarios highlighting the unique MRI features for each clinical entity [Table 2].
Etiologically spinal infections can be categorized as pyogenic, granulomatous, and parasitic. Most bacteria induce a pyogenic response, whereas mycobacteria, fungi, Brucella More Details, and syphilis cause granulomatous reactions. In our study, pyogenic spondylitis (most of the patients revealed positive blood culture of Staphylococcus aureus) and Brucellosis More Details were the infective pathologies that were encountered which masqueraded as TB spine.
Tuberculous spondylitis usually has an insidious onset and chronic progression compared to pyogenic spondylitis. In pyogenic spondylitis, lumbar and cervical segments are more commonly affected. The destruction of the intervertebral disk is more conspicuous than the destruction of the bone elements and there is usually no kyphotic deformity. The onset of the disease is more acute with a marked systemic involvement. Na-Young et al. in his study concluded that MRI showed a sensitivity of 100%, a specificity of 80%, and an accuracy of 90% in diagnosing TB when compared to pyogenic infection. Relative preservation/minimal destruction of disc is noted in early stages of tubercular spondylitis due to the lack of proteolytic enzymes in mycobacterium as compared to pyogenic spondylitis, where there is moderate to complete disc destruction. Finally, diagnosis should be considered in combination with corresponding clinical manifestations, radiological findings, blood and tissue cultures, and histopathological findings [Figure 2]A.
Brucellosis, a zoonotic infection that is caused by small Gram-negative bacilli from the genus Brucella and in humans, is often contracted by handling contaminated animal products or by consuming dairy products prepared from unpasteurized milk. The musculoskeletal system is commonly affected, and the usual site of bone brucellosis is the spine. In brucellosis, spondylitic changes could occur in the form of osteophytes on an anterior endplate (Parrot beak spine), intact vertebral body with disc gas or discal vacuum on MRI with para-spinal soft tissue involvement. In brucellar spondylitis, the lumbar spine is commonly affected and the discal involvement is very obvious. Osteoporotic lesions are frequently found in thoracic spine. The pedicles are normally spared but the mineral bone density is decreased. Gas in discs or the vertebral body is also a characteristic finding in brucellar spondylitis. These imaging findings in the clinical context of fever, malaise, weight loss, and myelopathy could prompt the diagnosis of brucellar spondylitis over TB [Figure 2]B.
In TB, vertebral collapse with gibbus deformity of dorsolumbar spine and psoas abscess formation occurs but gas or air within the disc, vertebra, or abscess is not a feature of tuberculous spondylitis.
Ankylosing spondylitis (AS) is a chronic inflammatory disease that primarily affects the spine and sacroiliac joints, causing pain, stiffness, and a progressive thoracolumbar kyphotic deformity. A significant complication in patients with AS is the formation of localized vertebral or disco-vertebral lesions of the spine, which was first described by Andersson in 1937. Anderssons's lesion/discitis may result from inflammation or (stress-) fractures of the ankylosed spine. This is also referred as “sterile discitis,” which on focal sagittal MRI sequences can easily mimic TB spondylodiscitis. Hence, it is important to look for other imaging findings like the “bamboo spine,” facetal arthropathy, stress fractures and syndesmophytes; it is important not the limit oneself to the finding of “discitis”. It is, therefore, mandatory to screen the entire spine including SI joints when there is slightest suspicion of AS. Identifying these findings become important as there is no indication for a diagnostic biopsy in these cases [Figure 3]A.
Rheumatoid arthritis (RA) is a systemic inflammatory disease with manifestations of a peripheral polyarthritis. Cervical spine involvement occurs late in the course of the disease. Bone erosions and atlantoaxial subluxation on radiographs are important signs of cervical spine involvement in RA. MRI can detect earlier signs of RA such as bone marrow edema and synovitis and has a higher sensitivity in detecting bone erosions compared to conventional radiography [Figure 3]B. MRI can also show the level and the degree of narrowing of spinal canal caused by dislocation with or without extradural pannus tissue that can compress cord. Pannus can be fibrotic (low signal on T1, T2) or hypervascular (low on T1 and high on T2 sequences). It can be misdiagnosed as TB if MRI is interpreted without adequate clinical history and clinical examination details.
Primary bone lymphomas constitute mainly diffuse large B-cell lymphomas. Peak prevalence is 5th–7th decades and occurs more commonly in males. Spinal involvement may manifest as paraspinal, vertebral, or epidural lesions either in isolation or combination and vertebral lesions can have a sclerotic, lytic, or mixed appearance. Bone scintigraphy shows increased radionuclide uptake in nearly all patients. The imaging appearance of an area of bone marrow replacement with soft-tissue mass without large areas of cortical bone destruction usually suggests lymphoma. Contiguous vertebral involvement is also noted. Since it has a nonspecific appearance on CT and MR imaging, it often mimics tubercular spondylitis [Figure 4]. Huang et al. reported a case of primary non-Hodgkin lymphoma originating from a lumbar vertebra that was initially misdiagnosed as tuberculous spondylitis. Lymphomas in spine usually occur as paraspinal masses with vertebral lesion but do not have extensive cortical bone destruction; discal height is usually preserved, but malignancies like lymphoma and multiple myeloma may involve the disk. Shankar et al. concluded in his study that CT perfusion technique has potential for differentiating inflammatory diseases like TB from neoplastic lesions affecting spine associated with paraspinal mass.
The common primary tumor with usual metastasis to spine is often from the lung ranging from 14%–31% and thoracic segments are the most common location for metastases compared to other vertebra. Zamzuri et al. described a case of spinal metastasis, which has the MRI appearances of an infection. He concludes that one must correlate with the clinical findings and do further investigations to confirm if it is an “infection” or “metastasis” [Figure 5]. Lee et al. reported a case of spinal metastasis involving ≥5 contiguous vertebrae seen on MRI. TB can involve ≥3 contiguous or noncontiguous vertebrae, which may not be the typical finding in spinal metastasis. Intervertebral discs and disc spaces could be spared in both spinal metastasis and tuberculous spondylodiscitis. A destructive bone lesion associated with a well-preserved disc space with sharp endplates or the involvement of only one vertebral body or posterior elements suggests neoplastic infiltration; whereas a destructive bone lesion associated with a poorly defined vertebral body endplate, with or without loss of disc height, suggests infection. However, in early stages, these changes are mild and variable and, hence, cannot be easily detected. DCE-MRI may provide additional information for differentiation between spinal TB and metastasis, when their manifestations on conventional imaging are similar.
It is very challenging on imaging to differentiate post-traumatic fracture from infectious spondylitis, mostly TB which is more prevalent in endemic locations owing to the epidemiological reasons, when MRI spine is done within a span of few days to weeks after trauma. An acute traumatic fracture may cause diffuse low signal intensity of the body on T1WI, mimicking a malignant fracture or an atypical infection. On the other hand, the involvement of a single vertebral body with/without disc involvement may lead to diagnostic chaos. In most cases of this type, the differential diagnosis includes metastatic disease and mycobacterial infection. But with detailed clinical history and follow-up imaging, these signal changes in the vertebral marrow resolve on MRI and the diagnosis of a post-traumatic fracture can be established [Figure 6]A and [Figure 6]B. In early trauma (up to 3 months), there could be signal changes and enhancement on MRI as a result of inflammation due to the microfractures secondary to trauma [Figure 6]A. But these signal changes in the vertebral marrow resolve eventually over a period of time. However, if the history of trauma is old (>6 months) and when signal changes and enhancement are encountered on MRI in the clinical background of fever, a biopsy might be necessary to rule out infection [Figure 6]B.
Degenerative disc disease
Modic type 1 degeneration can cause paraspinal soft tissue changes, and these reactive bone changes related to disc degeneration cannot always be confidently distinguished from infection by imaging. Vital et al. reported that Modic type 1 changes correspond to edema of vertebral endplates and subchondral bone. This edema corresponds to microfractures of cancellous bone and endplate cracks accompanied by an increased vascular density along with an increase in the number of nerve endings and in the levels of proinflammatory chemical mediators, and these vascular and inflammatory changes would follow the initial mechanical phenomena. As a result of this, sometimes there is presence of enhancement along with signal changes associated with degenerative disc disease, often misleading to infection [Figure 7]. Clinical and laboratory findings, such as white blood cell count, ESR, and elevated body temperature, may provide supportive but not confirmatory evidence of infectious spondylodiscitis. Accordingly, when clinical history, laboratory findings, and imaging do not allow the exclusion of infection, a biopsy should be performed.
Baastrup's disease (kissing spine syndrome) refers to approximation of adjacent spinous processes due to general degenerative changes of the spine. Baastrup's disease commonly affects the lumbar spine with L4-L5 being the usual affected level. It occurs at ages over 70 with no gender predilection. The excessive lordosis of spine with resultant mechanical pressure causes repetitive strains of the interspinous ligament resulting in degeneration and collapse. Thus, adjacent spinous processes come in contact and on further repetitive shearing movements, there is inflammation of an adventitious bursa present in the interspinous space. There may be additional architectural distortion, flattening, sclerosis, and cyst formation in the opposing surfaces. Changes in Baastrup's disease mostly occur in association with other degenerative factors such as loss of disc height, spondylolisthesis, and spondylosis with osteophyte formation. The edema at the level of interspinous ligaments, inflamed vertebra secondary to degeneration shows enhancement postintravenous gadolinium administration on MRI and this may misguide toward the diagnosis of TB spine [Figure 8]. However, it is very important to look for spinous processes in any suspected case of spondylodiscitis, as close approximation and contact of enlarged spinous processes should prompt the diagnosis of Baastrup's disease.
Acute osteoporotic fracture will show T2 hyperintensity and enhancement on postcontrast MR imaging for initial days up to few weeks [Figure 9]. The enhancement related to the inflammatory response may mislead to the diagnosis of TB spine. Acute osteoporotic fractures usually show a bandlike area of low signal intensity adjacent to the fractured endplates on T1WI with spared normal marrow signal intensity. Also, an area of low signal intensity corresponding to the fracture line or trabecular impaction can be seen on T2WI with a preserved disc. However, in presence of degenerative changes like marginal osteophytes and decreased density in bones, it is important to consider osteoporotic fracture, which can be further confirmed by radiographs where diffuse osteoporotic changes are significantly evident.
Spinal neuroarthropathy (SNA)
It is also called as Charcot spine or spinal neuropathic joint, which is an uncommon progressive osseous and ligamentous injury of the spine secondary to loss of deep sensation and proprioception due to which patients lack symptoms early in the course of disease and will eventually progress to advanced destruction before SNA is suspected clinically. Although SNA was historically described in patients with neurosyphilis, presently traumatic spinal cord injury is now the most common associated condition. Other conditions that occur in association with SNA include diabetes mellitus, congenital insensitivity to pain, syringohydromyelia, Charcot–Marie–Tooth disease, Guillain–Barre syndrome, transverse myelitis, and Friedreich ataxia. SNA most commonly involves thoracolumbar and lumbosacral junctions. Most findings occur in the late stage of the disease. MRI is the best modality in evaluation of SNA. It can affect disc space, vertebral bodies, facets, and paraspinal soft tissue. Findings include loss of disc space height, endplate erosions, and effusions with peripheral enhancement of disc space. Gas within the disc space is an important finding and is the hallmark of SNA. However, gas is easier to visualize on CT than MRI. Vertebral bodies show evidence of bone marrow edema, enhancement, and compression deformities due to microtrauma. There is also associated facetal joint involvement leading to spinal instability. Large complex fluid collections in the disc, vertebra, and paravertebral soft tissues are noted because of microtrauma, microhemorrhage, and inflammation. This spectrum of imaging findings closely mimic tubercular spondylitis [Figure 10]. When there is gas within the disk space (indicating motion), bone fragmentation, increased bone density, and malalignment in the background of appropriate clinical history, it favours SNA and not Pott's spine.
Rosai–Dorfman disease (RDD)
It is a rare histiocytic proliferative disorder of unknown etiology mainly characterized by painless bilateral cervical lymphadenopathy. Extra nodal manifestations are uncommon and spinal involvement is rare.
The common extra nodal sites include skin, upper respiratory tract, and bone. Lesion can be intradural or extradural in location, appears low or iso intense on T1- and T2-weighted MR images and shows intense homogenous enhancement with contrast. Marrow enhancement of the vertebra is rare. When there are associated vertebral marrow signal changes with enhancement, they mimic TB [Figure 11]. Since RDD has good prognosis and is self-limiting in about 70% cases, it is important to be aware of this entity when a patient presents with painless cervical lymphadenopathy. Surgical debulking is only done to relieve cord compression.
Despite the presence of definitive MRI findings indicative of spinal TB, there can still be significant overlap with other diagnoses and so a biopsy may be necessary to establish a confident diagnosis. However, many authors in the literature have attempted to find relevant differences between spinal TB and other disease entities. Since TB spine is the leading infection in developing countries, it is not only important to identify its imaging patterns, but also crucial to not overcall this diagnosis as it may lead to unnecessary patient anxiety and biopsy. It is thus critical to not label the “many other spine pathologies,” which show edema in vertebra causing signal changes on T1, T2 and enhancement on MRI as “TB spine”; as the treatment, surgical course, and prognosis also differ. Clinicians and radiologists need to be aware of the clinical and radiological aspect of the lesions that resemble and mimic TB spine on MRI, as a checklist, before they pen down with the final diagnosis of “TB spine.”
Taken from the Institutional Review Board.
Written consent was obtained from all the patients during the MRI scan.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 11]
[Table 1], [Table 2]