Computed tomography detection of clinically unsuspected skeletal tuberculosis

Clinical Imaging 39 (2015) 1056–1060

Contents lists available at ScienceDirect

Clinical Imaging journal homepage: http://www.clinicalimaging.org

Computed tomography detection of clinically unsuspected skeletal tuberculosis☆,☆☆,★ Pankaj Gupta a,1, Mahesh Prakash a,⁎,1, Navneet Sharma b, Rajinder Kanojia c, Niranjan Khandelwal a a b c

Department of Radiodiagnosis and Imaging, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, 160012 Department of Internal Medicine, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, 160012 Department of Orthopedics, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, 160012

a r t i c l e

i n f o

Article history: Received 25 April 2015 Received in revised form 18 July 2015 Accepted 30 July 2015 Keywords: Computed tomography Skeletal tuberculosis Spine Tuberculosis

a b s t r a c t Objective: To report the frequency of clinically unsuspected axial skeletal tuberculosis (STB) and findings on computed tomography (CT). Materials and methods: An evaluation of CT chest, abdomen, and pelvis of patients with tuberculosis was done. Bone window images were evaluated for skeletal involvement. Results: Of the 726 CT studies, 34 (4.7%) patients had skeletal involvement. Thoracic spine was the most commonly affected site with involvement of body in 58% cases. Intervertebral disc involvement, soft tissue abscess, and epidural extension were identified in 83%, 53%, and 39% of cases, respectively. Conclusion: Evaluation of bone window on CT can detect axial STB. © 2015 Elsevier Inc. All rights reserved.

1. Introduction Tuberculosis (TB) is a major health problem in developing countries [1]. Even in the developed countries, there is an upsurge in the cases of TB [2]. This is related to the pandemic of human immunodeficiency virus (HIV) infection. Skeletal tuberculosis (STB) comprises one third of all cases of extrapulmonary TB [3]. Spine is the most common site of STB [4]. The presentation of patients with STB is vague, leading to inadvertent delay in diagnosis. Imaging plays a crucial role in diagnosis of patients presenting with localising symptoms such as back pain. Magnetic resonance imaging (MRI) is the modality of choice for evaluation of patients suspected to have STB. Computed tomography (CT) has inferior soft tissue resolution and provides limited evaluation of spinal canal and, hence, is less frequently utilised. However, CT is highly sensitive in detecting soft tissue calcification, a pathognomonic feature of TB [5]. Moreover, CT is frequently performed in patients with suspected pulmonary or abdominal TB and can detect clinically unsuspected STB. Besides, CT allows planning of image-guided fine needle aspirations or biopsies to establish the diagnosis. The imaging appearance of STB including tubercular spondylitis on CT and MRI has been extensively reviewed [3,6,7]. However, in the present ☆ Conflict of interest: none declared. ☆☆ Financial disclosure: none declared. ★ Source of funding: none declared. ⁎ Corresponding author. Department of Radiodiagnosis and Imaging, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, 160012. Tel.: +91-9914209870. E-mail address: [email protected] (M. Prakash). 1 Both authors contributed equally. http://dx.doi.org/10.1016/j.clinimag.2015.07.033 0899-7071/© 2015 Elsevier Inc. All rights reserved.

study, we aimed to detect the skeletal changes in patients not clinically suspected to have STB who underwent CT studies for evaluation of TB in chest and/or abdomen [(Ab)]. The role of CT in detection of clinically unsuspected STB has not been reviewed previously.

2. Material and methods We conducted a retrospective study to evaluate clinically unsuspected skeletal involvement on CT chest, CT (Ab)/pelvis (CT AP) and CT chest, (Ab), and pelvis (CAP) in patients with pulmonary, abdominal, or disseminated TB. The study was approved by departmental ethics committee. A total of 799 patients who were referred to our department with a diagnosis of TB of chest and/or (Ab) were included. These patients underwent CT from January 2013 to June 2014. The clinical data [presence of constitutional symptoms (fever, night sweats, weight loss), neurological involvement, back pain, joint pain, sputum for acid fast bacilli (AFB), AFB culture, any fine needle aspiration cytology or biopsy, previous history of TB, intake of antitubercular therapy (ATT), duration of intake of ATT, response to ATT] and imaging files of these patients were retrieved from our database. The diagnosis of TB of chest and/or (Ab) was based on a combination of clinical (fever, weight loss), laboratory (positive Mantoux test, sputum positivity for AFB), histological (demonstration of caseating granuloma and AFB), and radiological criteria (typical radiological findings of pulmonary, nodal, or abdominal involvement). Patients who had clinical diagnosis of STB or neurological features at presentation were excluded. This led to exclusion of 73 patients. A total of 726 patients finally entered for analysis.

P. Gupta et al. / Clinical Imaging 39 (2015) 1056–1060

3. Imaging technique Imaging was performed on three multidetector CT scanners (Somatom Sensation 16; Somatom Definition Flash; Philips iCT). Patients reported fasting (8 h) on the morning of examination for contrastenhanced studies. Abdominal CT scans were performed following oral intake of 2 l of contrast [40 ml of ionic water soluble contrast (UROGRAFIN®, Schering AG, Germany) in 2 l of plain water] over 45 min. Abdominal CT [or chest plus (Ab)] and chest CT scans were performed with 85 ml and 50 ml of nonionic intravenous contrast (OMNIPAQUE™ 300, Iohexol, GE Healthcare, Princeton, NJ, USA), respectively, injected via 18-G cannula in the cubital fossa of upper limb. The intravenous injection was achieved with a pressure injector at a rate of 2.5 ml/s. Imaging was performed in the portal venous phase at 65 s from the start of contrast injection for abdominal [or chest plus (Ab)] CT scan and 40 s from the start of contrast injection for chest CT scan and included the area from domes of diaphragm to pubic symphysis for abdominal CT and from the level of thoracic inlet to domes of diaphragm for chest CT. Analysis of images was done on workstations (Syngo ®.via, Siemens; Brilliance ™ Workspace V2, Philips). Axial and multiple planar reformatted (MPR) images were reviewed in soft tissue [(Ab), mediastinum; window width (WW)/window level (WL): 45/315], lung (WW/WL: − 400/ 1500), and bone windows (WW/WL: 300/1500). 4. Analysis of CT images The CT images were analysed in consensus by two radiologists (blinded to original CT reports) with 3 years and 10 years experience in general radiology. The lung and soft tissue window images were reviewed for pattern of involvement: (a) presence of centrilobular (CL) nodules, miliary nodules, consolidation (Cons) with or without cavitation; (b) lymphadenopathy with or without necrosis or calcification; (c) ileocecal involvement (thickening with or without luminal narrowing); (d) ascites with or without peritoneal involvement (Peritn) (thickening and enhancement). The bone window images were reviewed for (a) site of involvement: divided into vertebral, joint, and appendicular skeleton; (b) pattern of involvement: vertebral involvement was classified into body (anterior, central, posterior), posterior elements, or both; pattern of involvement was classified into fragmentary, osteolytic, subperiosteal, and well-defined lytic with sclerotic margins based on the work of Jain et al. [6]; (c) intervertebral disc involvement; (d) associated soft tissue; (e) associated collection; and (f) calcification within the collection. Though the reviewers were blinded to the original CT reports, no comparison with the original CT reports was done. 5. Results A total of 726 patients were evaluated. There were 263 females (Fs) and 463 males (Ms). The mean age was 42.9 years (range, 13–76 years).

1057

A total of 304 chest CT, 316 CT AP, and 106 CT CAP were evaluated. Thirty-eight patients were HIV positive. CT findings in each group (chest, AP, CAP) were separately reported and compared to findings in the group with STB (Table 1). Thirty-four patients [who underwent CT chest (n= 5), AP (n= 4), and CAP (n=25)] had evidence of skeletal involvement (4.7%). In the rest of the subjects, the bones were either normal or showed evidence of degenerative changes (n=96). Among patients with skeletal involvement, mean age was 43.4 years (range, 14–72 years) with M predominance (n= 21). The presence of caseating granulomas in biopsy specimen [from chest/(Ab)] with or without the presence of AFB (on microscopy or culture) comprising histopathologically confirmed cases were found in 15 cases. In the rest of the patients, diagnosis was based on typical clinical presentation with positive Mantoux test (n= 8), clinical, and/or radiological response to ATT (n=11). Five patients were HIV positive. Twenty-three patients had active disease in the chest (n= 16), (Ab) (n= 4), or both sites (n= 3). In the chest, the most common findings were presence of CL nodules (n= 13; Fig. 1a), Cons (n= 10), and mediastinal/hilar lymphadenopathy (n= 10; Fig. 1b). “Tree in bud” appearance suggesting endobronchial spread was noted in 10 patients, while pleural effusion (Pl Ef) and hydropneumothorax (Hydro Pneu) were noted in 8 and 3 patients, respectively. In the (Ab), ileocecal (IC) junction and lymph nodes were the most commonly involved sites (n=5, Fig. 2a), followed by adrenals (Ads) (n=4), peritoneum (n=2, Fig. 2b), and pancreas (Panc) (n=1). Ascites and hepatic and splenic lesions were noted in four, two, and five patients, respectively. Overall nodal calcification was found in 20% and necrosis in 15% of the cases. Bone involvement: Average number of bones involved was 2.5 (range, 1–9). Spine was the most common site of involvement (n= 23). Thoracic spine was involved in 14 patients. Lower thoracic spine (T8–T12) was the most common thoracic spinal segment involved. Lumbar spine was involved in five and sacrum in four cases. Single vertebral level involvement was noted in three patients. Contiguous involvement of two or more vertebrae was seen in 16 cases (Fig. 1c). Noncontiguous involvement of two vertebral levels was present in four cases. Intervertebral disc involvement was seen in 19 patients. Overall vertebral body was the most commonly involved part of the spine (n=20). Anterior part, central part, and posterior part of the vertebral body were involved in 17, 2, and 1 cases, respectively (Fig. 2c). Both anterior and posterior parts of the vertebral body were involved in two cases. Posterior element involvement was seen in three cases. Most common pattern of involvement was fragmentary (n= 16; Fig. 3a) followed by osteolytic (n=4), subperiosteal (n=1), and welldefined lytic lesion (n= 1; Fig. 3). In one case, there was sclerosis. Paravertebral soft tissue involvement was seen in more than 50% cases (n= 18; Fig. 4a). Epidural extension was seen in nine patients. Soft tissue calcification was seen in five cases (Fig. 4a). Other sites of involvement were hip joints (n= 5), pelvic bones (n= 3), ribs (n= 2),

Table 1 Findings among patients with TB undergoing CT chest, AP and CAP, and those with STB

Total patients with TB of chest/(Ab)

CT site (n)

CT findings

M/F

n (%)

Chest (304) 187/117 AP (316) 201/115 CAP (106)

CL Nodules 210 (69%) Bowel/ICJ 163/55 (51/17%) CL Nodules 48 (45%) Bowel/ICJ 37/15 (35/14%) CL Nodules 13 (38%) Bowel/ICJ 7/5 (20/15%)

75/31 Patients with STB

34 21/13

Cons. 190 (62%) Node 110 (35%) Cons. 35 (33%) Node (Ab) 19 (18%) Cons. 10 (29%) Node (Ab) 5 (15%)

Node 144 (47%) Ascites 40 (13%) Node (c) 25 (23%) Ascites 10 (10%) Node (c) 10 (29%) Ascites 4 (12%)

Pl Ef 63 (21%) H/S 15/23 (5/7%) Pl Ef 21 (20%) H/S 4/7 (4/7%) Pl Ef 8 (23%) H/S 2/5 (6/15%)

Hydro Pneu 29 (9%) Peritn 21 (6%) HydroPneu 7 (7%) Peritn 3 (3%) Hydro Pneu 3 (9%) Peritn 2 (6%)

PE 8 (3%) Ad 8 (2%) PE 1 (1%) Ad 3 (3%) PE 0 Ad 4 (12%)

TIB/M 64/18 (21/6%) Panc 2 (1%) TIB/M 18/5 (17/5%) Panc 0 TIB/M 10/0 (29/0%) Panc 1 (3%)

PE, pericardial effusion; TIB/M, tree in bud/miliary nodules; ICJ, ileocecal junction; H/S, hepatic/splenic lesions; (c), chest; (Ab), abdomen.

Multiple findings

Necrotic nodes

Calcified nodes

121 (40%)

87 (28%)

21 (7%)

112 (35%)

53 (18%)

8 (2%)

34 (32%)

16 (15%)

5 (5%)

12 (35%)

5 (15%)

7 (20%)

1058

P. Gupta et al. / Clinical Imaging 39 (2015) 1056–1060

Fig. 2. (a–c) Axial abdominal CT shows necrotic retroperitoneal nodes (Fig. 2a, arrow) and ascites with peritoneal enhancement (Fig. 2b, arrow). Axial CT of the same patient shows destruction of anterior aspect of body of L1 (Fig. 2c, arrow) with prevertebral abscess (Fig. 2c short arrow).

Fig. 1. (a–c) Axial chest CT shows multiple bilateral CL nodules (Fig. 1a) and necrotic mediastinal (Fig. 1b, arrow) and hilar lymphadenopathy (Fig. 1b, short arrow). Sagittal reformatted image of spine of the same patient shows destruction of two adjacent lumbar vertebral bodies (Fig. 1c, arrows).

sacroiliac joint (SIJ) (n=2, Fig. 4b), and manubrium sterni (n=1). Psoas abscess was seen in four cases. Other involved muscles included gluteus maximum (n= 2), pyriformis (n= 1, Fig. 4c), and pectoralis major (n=1) (Table 2). 6. Discussion STB comprises one-third cases of extrapulmonary TB. Overall, it affects less than 3% of patients suffering from TB [3]. In this subgroup of TB patients, spine is involved in more than 50% of patients and, as such, represents the most common site of STB. Besides the high incidence of TB in endemic countries, there is resurgence of TB even in the developed countries [2]. Most of these newly diagnosed cases are

those of pulmonary TB; however, corresponding increase in the incidence of STB is also reported [8]. The primary reason for this is the current HIV pandemic. Other contributory factors are increasing migrant population, growing drug resistance, and exposure of the health care personals. A recent survey conducted by a UK agency concluded that there was a significantly greater annual incidence of TB in 2012 than 2010 [2]. A confident diagnosis of skeletal involvement in TB is often difficult as the clinical presentation is nonspecific [9]. Constitutional symptoms and nonspecific back pain are the predominant complaints. Thus a delayed diagnosis is common. This accounts for increased neurological morbidity due to compression of the spinal cord/adjacent nerves from cold abscesses, vertebral collapse, and kyphosis. In clinical practice, particularly in endemic areas, patients with long-standing cough/hemoptysis and constitution symptoms are subjected to CT chest, CT AP, or CT CAP. A presumptive diagnosis of TB in endemic areas is based on combination of clinical presentation and relatively specific imaging findings (necrotic lymphadenopathy/CL nodules with tree in bud appearance) [10,11]. Meticulous evaluation of the MPRs in bone window setting may lead to the diagnosis of unsuspected skeletal involvement, particularly that of the spine. This holds significance as STB is more resistant to treatment and mandates a longer course of ATT and early consideration of surgical treatment should neurological symptoms develop [12].

P. Gupta et al. / Clinical Imaging 39 (2015) 1056–1060

1059

Fig. 3. (a–d) Axial CT image shows fragmentary pattern of destruction of L2 vertebral body (Fig. 3a,). Axial CT image shows osteolytic pattern of vertebral destruction (Fig. 3b). Sagittal reformatted image shows subperiosteal pattern of destruction of lower thoracic vertebra (Fig. 3c). Axial CT image shows a well-defined lytic lesion involving the anterior L1 vertebral body (Fig. 3d).

Though, the CT findings of spinal TB have previously been described, the issue of detection of clinically unsuspected STB on CT chest/AP/CAP has not been addressed. Our study is one of the larger series describing the CT findings of STB and importance of careful evaluation of bone window in patients of suspected pulmonary/abdominal TB undergoing CT. The incidence of STB in our series is similar to that reported in literature [3]. We noted vertebral involvement in 67% of patients. This incidence is higher than that reported in most series [6]. This can be explained by the exclusion of most of the appendicular skeleton as the CT scans were tailored to include chest and (Ab). Tubercular spondylitis involved thoracic spine in 67% and lumbar spine in 22% cases. In most of the series, lower thoracic and lumbar spine is reported to be the most commonly involved site with the former being affected more commonly [13]. Involvement of cervical and sacral spinal segments is uncommonly reported [13]. The absence of cases of cervical spine involvement in our series is simply due to the nonevaluation of the cervical spine by the CT scans. This could also be the reason for a relatively higher incidence (17%) of the sacral spine involvement noted in our series. In most of our patients, two or more vertebral bodies were involved contiguously. This is a typical pattern of tubercular spondylitis [14]. Single vertebral level involvement is uncommon as is the skip involvement of two different vertebral levels but these patterns are well documented in literature [15]. Single vertebral involvement and skip vertebral involvement were noted in 13% and 17% of cases, respectively, in our study. Four distinct pattern of vertebral destruction have been reported [6]. These include fragmentary, osteolytic, subperiosteal, and well-defined lytic with sclerotic margins. Fragmentary type, the most common type, is characterised by

destruction of vertebral body into numerous fragments. These frequently migrate into the adjacent soft tissue collections. The incidence of this pattern of destruction has been reported to be between 47% and 81% [6,16]. We found this pattern of destruction in 70% of cases. Involvement of neural arch is an atypical presentation of tubercular spondylitis with reported incidence between 2% and 10% [17]. In our study, involvement of posterior elements was found in 13% patients. Paravertebral soft tissue abscess is an important imaging feature of tubercular spondylitis with reported incidence of 65–100% [6,7,14,18]. Although MRI scores over CT in demonstration of soft tissue abscesses, the latter is better at detecting soft tissue calcifications, one of the pathognomonic features of tubercular spondylitis [5]. This finding has been reported in 40–65% of cases [6,14]. Incidence of paravertebral soft tissue involvement of 53% in our study is in agreement with that reported in literature. We found calcification in the paravertebral soft tissue abscesses in a lower percentage of patients (28%). Intervertebral disc involvement is a relatively late feature of tubercular spondylitis owing to the lack of proteolytic enzyme production by the tubercular bacilli [19]. However, when vertebral destruction is documented, intervertebral disc involvement is always present. In our study, disc involvement was found in 83% cases. Incidence of 70–100% has been reported in the CT studies [6,16]. Although MRI is better at detecting epidural extension of disease, CT can also detect such extension. The reported incidence of epidural extension on CT scans is 50–65% [6,16]. In the present study, epidural extension was found in 39% of cases. The lower incidence of epidural extension attests to the selection of asymptomatic patients in this study. The other sites of involvement reported in this study were hip joint (15%), pelvic bones including SIJ (12%), ribs (6%), and

1060

P. Gupta et al. / Clinical Imaging 39 (2015) 1056–1060 Table 2 Salient findings of skeletal involvement on CT (total n=34) Findings

Number, n (%)

Skeletal involvement Vertebral involvement Thoracic spine Lumbar spine Sacrum Body Anterior body Central body Posterior body Posterior elements Paravertebral soft tissue involvement Psoas abscess Soft tissue calcification Other sites of involvement Hip joints Pelvic bones Ribs Manubrium sterni

34 23 (67%) 14 (61 %) 5 (22 %) 4 (17 %) 20 (58%) 17 (50 %) 2 (6%) 1 (3%) 3 (9%) 18 (53%) 4 (12%) 5 (15 %) 5 (15 %) 4 (12 %) 2 (6%) 1 (3%)

References

Fig. 4. (a–c) Axial CT image shows presacral abscess (Fig. 4a, arrow) with calcification (Fig. 4a, short arrows). Axial CT shows destruction of left SIJ (Fig. 4b, arrow). Axial CT shows abscess in the left pyriformis muscle (Fig. 4c, arrow).

manubrium sterni (3%). Reported incidence in literature of hip joint, SIJ involvement, ribs, and sternal involvement is 15%, 10%, 5%, and 1.4%, respectively [20–24]. There were few limitations in the study. Probably, the greatest limitation was that the study population was not homogeneous. CT CAP was performed in only 15% cases, thus limiting the evaluation of entire axial skeleton to this subgroup only. As expected, CT CAP was the group in which maximum number of cases of STB was detected. This could be a confounding factor in calculating the incidence of various sites of involvement. However, this bias is unavoidable considering the retrospective nature of the study and the fact that the sickest patients are likely to get CT CAP. Histopathological confirmation of STB was not available. Although, based on our selection criteria, patients had TB in chest and/or (Ab), the remote possibility of atypical mycobacterial or nonmycobacterial infection could not be excluded. Follow-up data were not available.

[1] World Health Organisation. WHO report on global tuberculosis control: epidemiology, strategy, financing. Geneva: World Health Organisation; 2010. [2] Health Protection Agency. Tuberculosis in the UK: 2012 report. London: Health Protection Agency; 2012. [3] Moore SL, Rafii M. Imaging of musculoskeletal and spinal tuberculosis. Radiol Clin North Am 2001;39(2):329–42. [4] Lindahl S, Nyman RS, Brismar J, Hugosson C, Lundstedt C. Imaging of tuberculosis. IV. Spinal manifestations in 63 patients. Acta Radiol 1996;37:506–11. [5] Tali ET. Spinal infections. Eur J Radiol 2004;50:120–3. [6] Jain R, Sawhney S, Berry M. Computed tomography of vertebral tuberculosis: patterns of bone destruction. Clin Radiol 1993;47:196–9. [7] De Vuyst D, Vanhoenacker F, Gielen J, Bernaerts A, De Schepper AM. Imaging features of musculoskeletal tuberculosis. Eur Radiol 2003;13:1809–19. [8] Prakash M, Gupta P, Sen RK, Sharma A, Khandelwal N. Magnetic resonance evaluation of tubercular arthritis of the ankle and foot. Acta Radiol 2014. http://dx.doi.org/10. 1177/0284185114552882. [9] Moorthy S, Prabhu N. Spectrum of MR imaging findings in spinal tuberculosis. AJR 2002;179:979–83. [10] Skoura E, Zumla A, Bomanji J. Imaging in tuberculosis. Int J Infect Dis 2015;32:87–93. [11] Ringe A, Jankharia B, Spinal TB. Pediatr Infect Dis 2013;5:104–7. [12] Harisinghani M, McLoud T, Shepard JA, Ko J, Shroff M, Mueller P. Tuberculosis from head to toe. Radiographics 2000;20:449–70. [13] 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:S567–78. [14] Boxer DI, Pratt C, Hine AL, McNicol M. Radiological features during and following treatment of spinal tuberculosis. Br J Radiol 1992;65:476–9. [15] Ousehal A, Gharbi A, Zamiati W, Saidi A, Kadiri R. Imagerie di Mal de Pott. Neurochirurgie 2002;48(5):409–18. [16] Sinan T, Al-Khawari H, Ismail M, Ben-Nakhi A, Sheikh M. Spinal tuberculosis: CT and MRI feature. Ann Saudi Med 2004;24(6):437–41. [17] Babhulkar SS, Tayade WB, Babhulkar SK. Atypical spinal tuberculosis. J Bone Joint Surg (Br) 1984;66-B(2):294–6. [18] Burill J, Williams CJ, Bain G, et al. Tuberculosis: a radiologic review. RadioGraphics 2007;27:1255–73. [19] Tins BJ, Di PL, Cassar-Pullicino VN. MR imaging of spinal infection. Semin Musculoskelet Radiol 2004;8:215–29. [20] Babhulkar S, Pande S. Tuberculosis of the hip. Clin Orthop Relat Res 2002;398:93–8. [21] Goldberg J, Kovarsky J. Tuberculous sacroiliitis. South Med J 1983;76:1175–6. [22] Tatelman M, Drouillard EJP. Tuberculosis of the ribs. AJR Am J Roentgenol 1953;70: 923–35. [23] Sharma S, Juneja M, Garg A. Primary tubercular osteomyelitis of the sternum. Indian J Pediatr 2005;72:709–10. [24] Martini M, Ouahes M. Bone and joint tuberculosis: a review of 652 cases. Orthopedics 1988;6:861–6.