Le Infezioni in Medicina, n. 1, 62-69, 2023

doi: 10.53854/liim-3101-9

Original articles

Differentiating tubercular from pyogenic causes of spine involvement on Magnetic Resonance Imaging

Nitin Gupta1,2, Rajagopal Kadavigere3, Sundeep Malla4, Shyamasunder N Bhat5, Kavitha Saravu1,2

1Department of Infectious Diseases, Kasturba Medical College, Manipal, Manipal Academy of Higher Education, Manipal, Karnataka, India;

2Manipal Center for Infectious Diseases, Prasanna School of Public Health, Manipal Academy of Higher Education, Manipal, Karnataka, India;

3Department of Radiodiagnosis, Kasturba Medical College and Hospital, Manipal, Manipal Academy of Higher Education, Manipal, Karnataka, India;

4Department of Radiodiagnosis, Indraprastha Apollo Hospitals, New Delhi, India;

5Department of Orthopaedics, Kasturba Medical College and Hospital, Manipal, Manipal Academy of Higher Education, Manipal, Karnataka, India

Article received 29 June 2022, accepted 30 December 2022

Corresponding author

Kavitha Saravu

E-mail: kavithasaravu@gmail.com

SummaRY

Introduction: Based on clinical parameters alone, it is difficult to differentiate between tubercular (TB) and pyogenic spondylodiscitis. Therefore, this study aimed to identify imaging features that can help to differentiate these two entities.

Methodology: This retrospective observational study included microbiologically confirmed cases of TB and pyogenic spondylitis over three years. Magnetic resonance imaging of included cases was reviewed for a predefined set of imaging findings. A multivariable logistic regression analysis was done to compare the radiological features of both groups. A scoring system was created using the features that were significant in univariate and multivariable analysis.

Results: Of the 90 included cases, 44 were TB and 46 were pyogenic. Involvement of the thoracic vertebra, paravertebral abscess, epidural abscess, posterior element involvement, loss of height, spine deformity, nerve compression, destruction of the disc and larger abscess size were more common in patients with TB. Involvement of the thoracic spine, posterior element involvement and spine deformity were independent predictors of diagnosing TB spine. At a score of 7 (out of 11) or above, the scoring system had a sensitivity and specificity of 86% and 94%, respectively, in diagnosing the TB spine.

Conclusion: In the presence of relevant clinical findings, specific MRI findings can help to differentiate TB from pyogenic spondylitis.

Keywords: Spondylodiscitis; vertebral osteomyelitis; tubercular; brucellosis.

INTRODUCTION

Infectious causes of spine involvement include tuberculosis and other bacterial infections such as Brucella spp, Staphylococcus spp., Streptococcus spp. and Enterobacteriaceae [1]. It is essential to make a microbiological diagnosis as treating tubercular (TB) vs pyogenic causes are starkly different [1]. Since a definitive diagnosis with a biopsy of the affected spine is often not possible, especially in resource-constrained settings, there is a heavy reliance on clinical presentation and radiological findings to make a provisional diagnosis [2]. However, this is challenging as all infections significantly overlap in their radiological features. As a result, the final impression of a radiologist’s report in these cases is often subjective. Amongst the various radiological imaging modalities, Magnetic resonance imaging (MRI) is considered the gold standard imaging technique in these cases [3]. Therefore, this study aimed to identify objective MRI features that can help to differentiate tubercular from pyogenic spine involvement.

PATIENTS AND METHODS

This is a retrospective study in which all patients admitted between 2018 and 2021 to the Infectious diseases or Orthopaedics department of Kasturba Hospital, Manipal, were screened for evidence of infectious spondylodiscitis. Those patients with a microbiologically confirmed diagnosis and an MRI image taken at the time of diagnosis were included in the study. Post-traumatic or post-operative spine infections were excluded from the study. All cases of infectious spondylitis were grouped as tubercular or pyogenic spondylitis. A case of infectious spondylitis was deemed as tubercular if the biopsy of the affected vertebra was positive by mycobacterial culture or cartridge-based nucleic acid assay (GeneXpert). A case was deemed as pyogenic infection if the biopsy of the affected site or blood culture were positive for a known pathogenic bacteria.

The anonymised MRI images of the included patients were reviewed by experienced radiologists, and a predefined set of objective MRI findings were recorded for each case. The total number of vertebrae was counted for each patient. The presence of paravertebral soft tissue enhancement and the paravertebral abscess was noted separately. The largest dimension of the largest abscess in any plane in centimetres was noted. The presence of epidural abscess or spine deformity (kyphosis or scoliosis) was noted. Involvement of posterior elements (pedicle or spinous process) or disc was also noted. More than a 25% decrease in the height of the vertebral body was considered a significant loss of height. Compression of the nerve root or spinal cord was also noted. Significant destruction of the disc was defined when more than 50% of destruction was present. Lower lumbar vertebrae involvement was defined as involvement of L3-L5 vertebrae with or without sacral involvement. Those cases where L3-L5 involvement was accompanied by simultaneous involvement of L1, L2, thoracic or cervical vertebrae were not considered as lower lumbar involvement.

Data analysis

Qualitative variables were summarised as frequency and percentage, while quantitative variables were summarised as mean + standard deviation or median (Interquartile range). A univariate analysis was done to compare radiological features of TB vs pyogenic causes of spine infection.

The Chi-square test was used for qualitative variables, while the Mann-Whitney U test was used for quantitative variables. Those factors that were significant in univariate analysis were analysed using multivariable logistic regression analysis to find independent predictors for making a diagnosis of TB spine. Based on the variables that were significant in univariate and multivariable analysis, a scoring system was generated. A Receiver Operating Characteristic (ROC) curve was plotted to calculate the cut-off for the scoring system. The sensitivity and specificity of the selected cut-off were calculated. The Area under the curve (AUC) for the ROC curve was also calculated along with the 95% Confidence Interval (CI). A p-value of less than 0.05 was considered significant.

RESULTS

A total of 90 microbiologically proven cases of spondylodiscitis where an MRI was done were included in this study. Tubercular spondylitis was diagnosed in 44 patients, while pyogenic spondylitis was diagnosed in 46 patients. Pyogenic infections included brucellosis (n=22), Staphylococcus aureus (n=12), Enterobacteriaceae (n=9), Streptococcus spp. (n=2), and Burkholderia pseudomallei (n=1). The features of infectious spine involvement in patients with tuberculosis, brucellosis and other pyogenic infections have been compiled in Table 1. Involvement of the thoracic vertebra, paravertebral soft tissue involvement, paravertebral abscess, epidural abscess, posterior element involvement, loss of height, spine deformity, nerve compression, destruction of the disc and larger abscess size were more common in patients with TB (Figures 1-6) (Table 1). On multivariable analysis, thoracic vertebra involvement, posterior element involvement and deformity of the spine were found to be independent predictors of making a diagnosis of tubercular spondylitis (Table 2).

Figure 1 - Contrast-enhanced MRI images of a patient diagnosed with tuberculosis. (A) T2W axial images show bilateral paravertebral abscesses with well-defined walls and central liquefaction (black asterisk). (B) T1W post-contrast-enhanced axial images show peripherally enhancing abscesses (white asterisk) with paravertebral soft tissue involvement.

Figure 2 - Contrast-enhanced MRI images of a patient with tuberculosis. (A) T1W post-contrast-enhanced sagittal images show spondylodiscitis at L1-L2 1 level with a reduction in the vertebral >25% decrease in vertebral body height of L2 vertebra compared to L3. (B) T2 STIR sagittal images showing multifocal spondylodiscitis in the dorso-lumbar spine with paravertebral soft tissue inflammation.

Figure 3 - Contrast-enhanced MRI images of a patient with tuberculosis. (A) T2 W axial images show epidural abscesses (white arrows) causing narrowing of bilateral neural foramina with associated nerve root compression. (B) T1W post-contrast-enhanced axial images show peripherally enhancing epidural abscesses (white arrow) with paravertebral soft tissue involvement.

Figure 4 - Contrast-enhanced MRI sagittal images of a patient with tubercular spondylodiscitis with the destruction of the intervertebral disc (>50%) at the lower dorsal level.

Figure 5 - Contrast-enhanced MRI images of a patient with tuberculosis. (A) T1W post-contrast-enhanced axial images show enhancement in the pedicles and transverse processes (posterior elements) apart from the body. (B) T2 W axial images showing marrow oedema and inflammation in pedicles and the transverse processes.

Figure 6 - T2 weight and post-contrast T1 weighted sagittal images of a patient with tuberculosis showing a large paravertebral abscess with a subligamentous spread up to the sacral spine (black asterisk). The kyphotic deformity is noted at the L2-L3 level (white arrows) with multifocal areas of spondylodiscitis.

A scoring system was developed in which two points each were assigned for thoracic spine involvement, posterior element involvement and spine deformity. One point each was assigned for paravertebral soft tissue involvement, epidural abscess, >25% loss of vertebral body height, nerve compression and >50% disc destruction (Table 3). Of a maximum of 11 points, a score of 7 or above had a sensitivity and specificity of 86% and 94%, respectively. The ROC curve had an AUC of 0.92 (95% CI: 0.86-0.99) and a p-value of <0.001 (Figure 7).

Figure 7 - Receiver operating characteristic curve for the scoring system to diagnose tubercular spondylitis.

DISCUSSION

TB spine accounts for 60% of all musculoskeletal TB cases [4]. In a previous study, we identified that a longer duration of illness was significantly more common in patients with TB spine, while male gender, the presence of diabetes mellitus, leucocytosis, and increased C-reactive protein were more common in pyogenic spondylitis [2]. It must be noted here that the clinical findings in isolation are only suggestive. In the absence of a confirmed microbiological diagnosis, radiological findings supplemented by clinical history can further help in narrowing the list of differentials [5]. The present study was focused on devising an objective set of parameters that can help differentiate TB and pyogenic causes of spine infection.

Classically, TB is known to affect the thoracolumbar junction [6]. Similar to previous studies, thoracic involvement was more common in TB, while lumbar involvement was more common in pyogenic infection in this study [7, 8]. In fact, thoracic involvement was found to be an independent predictor in making a diagnosis of TB spondylodiscitis. Three distinct patterns of TB spine have been described, with paradiscal involvement being the most classical pattern [4]. Since the intervertebral disc is avascular and paradiscal arteries supply the endplates on either side of the disc, paradiscal involvement is the commonest site [3]. In those with central vertebral involvement, only a single vertebra can be involved. In such a case, the disc may be preserved [9]. In those with anterior vertebrae involvement, contiguous vertebrae involvement due to subligamentous spread has been classically described [9].

Like some previous studies, more vertebral levels were involved in TB compared to pyogenic [7]. The median number of vertebral involvement was 3 in patients with TB spine in our study. In a study by Jain et al., the mean number of vertebral involvement in the TB spine was 3.2. Primary involvement of the disc was believed to be uncommon with TB due to a lack of proteolytic enzymes [6, 10]. However, with the increasing use of MRI, it has been noticed that subtle disc involvement may have been missed on X-rays in some cases [11]. Also, disc destruction may be more evident in those who present late [11]. The median duration of illness in patients who presented with TB spondylodiscitis in a previous study of ours was 105 days [2]. Disc involvement is reported to be present in 25-72% of the patients [6]. It should be noted here that most studies do not differentiate between disc involvement due to a decrease in intervertebral space and disc destruction. In our study, disc destruction was seen in 43% of TB patients, and this was significantly higher than in those with pyogenic spondylodiscitis.

Paravertebral soft tissue involvement and formation of abscesses are more common in those with TB in our study. These abscesses are seen in 55-96% of TB spine cases [6]. In our study, paravertebral soft tissue involvement was seen in 97% of the cases. Similar to other studies, the size of the paravertebral abscess was also significantly larger in those with TB [5, 7]. Paraspinal abscess tracks through the iliopsoas compartment and can reach the retroperitoneum, pelvis or thigh [9]. In previous studies, it has been shown that there is no relationship between the abscess size and the severity of spondylodiscitis [9].

The epidural abscess was also significantly more common in TB spine patients [5, 7]. In a study by Jain et al., nerve compression was seen in 61% of TB spine cases [12]. In another study, spinal cord compression was seen in 50% of TB cases [6]. Compared to these studies, nerve compression (93%) and spinal cord (80%) compression was seen in a relatively higher number of cases with TB spine in our study. This might be reflective of the late presentation of our cases. Vertebral loss of height was more common in TB patients in our study. In a study by Kanna et al., vertebral collapse was found to be a strong predictor of TB spine [13]. Vertebral loss of height and, consequently, spine deformity are generally seen in the later stages of TB [5, 6]. Since anterior vertebrae are more commonly involved, the collapse of anterior spinal elements can lead to kyphotic deformity [14]. These can lead to potentially irreversible sequelae, and therefore, early diagnosis and treatment are paramount in the TB spine [6]. Spine deformity was found to be an independent predictor of TB diagnosis.

Although isolated involvement of posterior elements is rare in TB, concomitant posterior element involvement is common in TB [4, 5]. In a study by Andronikou et al., 50% of the patients with TB spine have posterior element involvement [6]. In another study by Kilborn et al., pedicle involvement was seen in 68% of adult patients [11]. Posterior element involvement was found to be a significant predictor of TB diagnosis. Pedicular involvement is found to be lower in cases of pyogenic spondylitis [11, 15]. It is important to identify posterior spine involvement as its presence is associated with neurological involvement and indicates a possible requirement for laminectomy and spine stabilisation [6].

The scoring system that we developed had excellent sensitivity and specificity in making a diagnosis of TB spine. Since there could be subjectivity in an impression made by radiologists, an objective scoring system such as ours can help in reaching a better provisional diagnosis. This will be particularly helpful in resource-limited settings, where an invasive sampling and microbiological diagnosis may not be possible. It must also be emphasised here that this scoring system needs to be prospectively validated in a study with a larger sample size.

In conclusion, involvement of the thoracic spine, posterior element involvement and spine deformity are independent predictors of making a diagnosis of TB spine. In the presence of relevant clinical findings, an objective scoring system can help differentiate TB from pyogenic spondylitis, especially in cases where a definite microbiological diagnosis cannot be made.

Conflict of interest

None

Funding received

None

REFERENCES

[1] Berbari EF, Kanj SS, Kowalski TJ, et al. Infectious Diseases Society of America (IDSA) Clinical Practice Guidelines for the diagnosis and treatment of native vertebral osteomyelitis in adults. Clin Infect Dis. 2015; 6 (6), e26-46.

[2] Gupta, Bhat SN, Reddysetti S, et al. Clinical profile, diagnosis, treatment, and outcome of patients with tubercular versus nontubercular causes of spine involvement: A retrospective cohort study from India. Int J Mycobacteriology. 2022; 11 (1), 75-82.

[3] Rajasekaran S, Soundararajan DCR, Shetty AP, Kanna RM. Spinal tuberculosis: current concepts. Glob Spine J. 2018; 8 (Suppl. 4), 96S-108S.

[4] Boruah DK, Gogoi BB, Prakash A, Lal NR, Hazarika K, Borah KK. Magnetic resonance imaging evaluation of posterior spinal tuberculosis: a cross-sectional study. Acta Radiologica. 2021; 62 (8), 1035-1044.

[5] Teo ELHJ, Peh WCG. Imaging of tuberculosis of the spine. Singapore Med J. 2004; 45 (9), 439-44; quiz 445.

[6] Andronikou S, Jadwat S, Douis H. Patterns of disease on MRI in 53 children with tuberculous spondylitis and the role of gadolinium. Pediatr Radiol. 2002; 32 (11), 798-805.

[7] Chang MC, Wu HTH, Lee CH, Liu CL, Chen TH. Tuberculous spondylitis and pyogenic spondylitis: comparative magnetic resonance imaging features. Spine. 2006; 31 (7), 782-788.

[8] Jung NY, Jee WH, Ha KY, Park CK, Byun JY. Discrimination of tuberculous spondylitis from pyogenic spondylitis on MRI. AJR Am J Roentgenol. 2004; 182 (6), 1405-1410.

[9] Rivas-Garcia A, Sarria-Estrada S, Torrents-Odin C, Casas-Gomila L, Franquet E. Imaging findings of Pott’s disease. Eur Spine J. 2013; 22 (Suppl. 4), 567-578.

[10] Miyamoto H, Akagi M. Usefulness of dynamic contrast-enhanced magnetic resonance images for distinguishing between pyogenic spondylitis and tuberculous spondylitis. Eur Spine. 2019; 28 (12), 3011-3017.

[11] Kilborn T, Janse van Rensburg P, Candy S. Pediatric and adult spinal tuberculosis: imaging and pathophysiology. Neuroimaging Clin N Am. 2015; 25 (2), 209-231.

[12] Jain AK, Sreenivasan R, Saini NS, Kumar S, Jain S, Dhammi IK. Magnetic resonance evaluation of tubercular lesion in spine. Int Orthop. 2012; 36 (2), 261-269.

[13] Kanna RM, Babu N, Kannan M, Shetty AP, Rajasekaran S. Diagnostic accuracy of whole spine magnetic resonance imaging in spinal tuberculosis validated through tissue studies. Eur Spine J. 2019; 28 (12), 3003-3010.

[14] Leonard MK, Blumberg HM. Musculoskeletal Tuberculosis. Microbiol Spectr. 2017; 5 (2).

[15] Shikhare SN, Singh DR, Shimpi TR, Peh WCG. Tuberculous osteomyelitis and spondylodiscitis. Semin Musculoskelet Radiol. 2011; 15 (5), 446-458.

[16] Thammaroj J, Kitkhuandee A, Sawanyawisuth K, Chowchuan P, Promon K. MR findings in spinal tuberculosis in an endemic country. J Med Imaging Radiat Oncol. 2014; 58 (3), 267-276.