Keywords
tuberculosis - spine - diagnosis - Pott disease - spondylodiscitis
Introduction
Tuberculosis (TB) is one of the oldest infectious diseases that affect humans. About
10.4 million new cases a year are estimated worldwide, but only 1.3 million are diagnosed,
and 2% are in the spine.[1]
[2] It is one of the top ten global causes of death.[3]
Nonspecific symptoms and insidious onset characterize the clinical picture of the
disease. The time from the onset of symptoms to the diagnosis can vary from 4 to 11
months.[4]
[5] Axial pain is the most usual symptom, but patients may experience fever, constitutional
symptoms, respiratory symptoms, and neurological deficit.[1]
[4] The diagnosis consists of the clinical picture associated with specific aspects
in the imaging exams and isolation of the microbiological agent and/or compatible
anatomopathological exam, and/or a positive therapeutic test.[6]
[7]
Tuberculosis patients have lymphocytopenia due to a decrease in CD4 T-lymphocytes.
There is no change, however, in CD8 T-lymphocytes. This clinical picture is reversible
and normalizes after recovery.[8] The patients manifest immunological abnormalities such as a failure in the proliferation
of CD4 T-lymphocytes and interferon-gamma production in response to mycobacterial
antigens.[9]
The radiographs of Pott disease initially show radiolucency, loss of definition and
erosion of end plates, destruction of the vertebral body, and loss of disc height[10] ([Fig. 1]). Computed tomography (CT) improves the evaluation of the radiographic findings
and of the extent of the lesion ([Fig. 2]). However, magnetic resonance imaging (MRI) is the most sensitive and specific imaging
exam. The signal strength decreases in vertebral bodies in the T1-weighted sequences
and increases in T2-weighted and short tau inversion recovery (STIR)-weighted images.
The use of contrast, such as gadolinium, differentiates the infected from the uninfected
areas[11]
[12] ([Fig. 3]). The anatomopathological examination reveals granuloma with giant cells and caseous
necrosis ([Fig. 4]). Isolation of the agent is possible through a specific culture for mycobacteria,
a bacilloscopy to identify acid-fast bacilli (AFB), or a rapid molecular test for
TB (RMT-TB).[2]
[6] The treatment involves multiple medications and lasts at least 12 months, and may
be extended due to resistant cases or adverse events. Surgery is indicated in patients
with instability, neurological deficit, deformity, and large abscesses.[4]
[6]
Fig. 1 Radiographic images of the thoracic (A), thoracolumbar (B), and lumbar (C) spine on lateral view. Notice the presence of segmental kyphosis (A–C) and collapse of the vertebral body (C).
Fig. 2 Lumbosacral spine CT. (A,B) Presence of collapse of the vertebral bodies, erosion of the end plates (sagittal),
and sachet injuries (axial). (C) Presence of an area of abscess in the anterior region of the vertebral bodies (white
arrowheads).
Fig. 3 Magnetic resonance imaging scans of the lumbosacral spine on T2-weighted sagittal
(A) and thoracic views, sagittal and axial (B,C) views, weighted in T2. Notice the vertebral collapse associated with segmental kyphosis,
anterior and epidural abscesses (A,B), and spinal cord compression (B,C).
Fig. 4 Photomicrograph of bone biopsy of the vertebral body showing an epithelioid granuloma
with a giant multinucleated cell (of the Langhans type) in a patient with Pott disease
(AO: 200x) Source: Geanette Pozzan - Pathological Anatomy Service, Hospital Central
da Santa Casa de Misericórdia de São Paulo.
The patients present a characteristic clinico-radiographic picture; however, delay
in establishing the diagnosis is usual. The diagnosis involves a positive therapeutic
test. In an extensive series of 967 patients,[13] isolation of the etiologic agent occurred in only 16.6% of the cases. This shows
that some patients meet the diagnostic criteria and respond to treatment until they
are cured, but do not have the etiologic agent isolated.[2]
The objectives of the present study are to describe the clinic- epidemiological, radiological,
and laboratory characteristics of tuberculous spondylodiscitis in a sample of the
Brazilian population, and to assess if there are differences among patients in whom
isolation of the etiologic agent occurred or not.
Materials and Methods
The present is a cross-sectional study involving a non-probabilistic sample of all
patients who underwent outpatient follow-up between 2009 and 2019 in a quaternary
hospital. The data were extracted from medical records, laboratory tests, and imaging
exams. Patients diagnosed with TB of the spine, regardless of the age group and gender,
were included. We excluded patients with incomplete data, any other type of infection,
or previous spinal surgery. The institutional Ethics Committee approved the study
(under CAAE 92596718.2.0000.5479).
The diagnosis of TB of the spine was established through the typical clinical characteristics
and specific imaging tests, as well as the presence of at least one of the following
criteria: isolation of the etiological agent, compatible anatomopathological examination,
or positive therapeutic test.[6] The clinical criteria considered the presence of at least two of the following symptoms:
back pain, pain on local palpation, night sweats, weight loss (at least 10% of body
weight in the previous 6 months), fatigue, malaise, or fever (temperature greater
than or equal to 37.8° Celsius).[6] The radiological criteria were established considering radiographs of the spine
on the posteroanterior and lateral views, with the presence of at least four of the
following findings: radiolucency, loss of definition and/or erosion of the vertebral
plateaus, loss of height of the intervertebral disc space, vertebral geodes, bone
sequestration, sclerosis, anterior wedging, and destruction of the vertebral body.[11] The criteria also included CT scans with at least one of the following findings:
paravertebral masses and abscesses, calcifications, and involvement of posterior elements.[11] The MRI criteria were hyposignal of the affected vertebral body on T1-weighted images
and hypersignal on T2- and STIR-weighted images, decreased intervertebral disc space,
and loss of definition of disc boundaries, and hypersignal of the intervertebral disc
on T2-weighted images.[11]
[14]
Magnetic resonance images were obtained using a 1.5-Tesla device (Magnetom Symphony,
Siemens Healthineers AG, Erlangen, Germany) with T1-, T2-, STIR-weighted cuts and
with contrast added. The images were analyzed by an orthopedist and a radiologist
who was a spine specialist with at least five years of experience.
The isolation of the etiological agent involved the presence of AFB by the Ziehl-Neelsen
method, the RMT-TB, and mycobacteria culture (Löwenstein- Jensen or Ogawa-Kudoh).[6]
The anatomopathological examination was considered compatible in the case of the proliferation
of epithelioid cells with the formation of confluent granulomas and Langhans giant
cells with central foci of caseous necrosis around lymphocytic infiltrate upon staining
with hematoxylin and eosin (H&E).[15] The therapeutic test was considered positive in cases in which the clinical and
radiological findings were suggestive of TB, but the anatomopathological examination
was not compatible with the disease and isolation of the etiological agent was not
possible.[15] The cured cases were considered positive, and were defined through clinical improvement
after the treatment for TB (absence or significant improvement in pain, sweating,
and fever), and through the normalization of the inflammatory tests, of the erythrocyte
sedimentation rate (ESR), and of the levels of C-reactive protein (CRP). All of the
cured cases should remain stable for at least one month of observation.[16]
Variables Studied
The demographic and epidemiological variables analyzed were age, gender, presence
of diabetes mellitus, chronic kidney disease or neoplasia, chronic use of corticosteroids
for at least six months at an immunosuppressive dose in the last twelve months, HIV
infection, pulmonary and/or extrapulmonary TB (except for spine TB) and type of household
(urban, rural or free area).
The clinical characteristics from symptom onset to diagnosis were pain, score on the
Visual Analogue Scale for Pain (VAS, from 0 to 10), respiratory symptoms (cough and
dyspnea), constitutional symptoms (weight loss of at least 10% in the last 6 months,
fatigue, malaise), fever (temperature greater than or equal to 37.8° Celsius), and
neurological status at baseline and after 6 months (measured through the American
Spinal Cord Injury Association [ASIA] impairment scale).
The laboratory variables at the time of diagnosis were hemoglobin level, leukocyte
count, lymphocyte count, monocyte count, monocyte/lymphocyte ratio, ESR, and the levels
of CRP, lactic dehydrogenase (DHL), creatinine, total proteins, and albumin. The biopsy
was percutaneous, open, or guided by CT. The microbiological aspects studied were
reactive tuberculin skin test (induration greater than or equal to 5 mm with the Mantoux
technique),[17] bacilloscopy (presence of AFB using the Ziehl-Neelsen stain), compatible anatomopathological
examination, culture (Lowenstein-Jensen medium), and RMT-TB with an evaluation of
resistance to rifampcin (GeneXpert MTB/RIF, Cepheid, Sunnyvale, CA, United States).
Through MRI scans, we established the location of the Pott disease (cervical, thoracic,
thoracolumbar transition, and lumbosacral), the number of levels affected, the presence
of disease in multiple levels (3 or more), involvement of more than 2 vertebrae at
adjacent levels, presence of collection within a vertebra or in adjacent paravertebral
tissues, subligamentous spread, destruction of the vertebral body with a decrease
in height greater than 50%, involvement of non-contiguous levels in different regions
of the spine, hypointensity on T1 and hyperintensity on T2, irregularity of the margins
of the end plates of the vertebral bodies on T2- and STIR-weighted sequences, and
invasion of the epidural space by caseation or granulation tissue.[12]
The treatment variables included type (non-surgical or surgical), duration (months),
drug and period of use (months), resistance to any drug in the treatment regimen for
TB, presence of monoresistance or multidrug resistance (resistance to at least isoniazid
and rifampcin), and the presence of complications (surgical site infection, neurological
deficit, chronic pain after cure, and death).[6]
The patients were divided into two groups: those in whom the etiologic agent was successfully
isolated (SI) or those with unsuccessful isolation (UI) through bacilloscopy and/or
negative culture and/or RMT-TB. All cases in the study met the cure criteria.
Statistical Analyses
The categorical variables were expressed as frequencies (numbers and percentages),
and the quantitative variables, as summary measures (mean, standard deviation, minimum
and maximum values). The receiver operating characteristic (ROC) curve was used for
the variables with a statistically significant difference regarding the assessment
of the sensitivity and specificity, using as the gold standard the distribution of
the values of that variable in the UI group. The data obtained were submitted to a
statistical evaluation, using the Statistical Package for the Social Sciences (SPSS
for Windows, SPSS Inc., Chicago, IL, United States) software, version 13.0. The level
of statistical significance adopted was values of p lower than or equal to 0.05.
Results
The study included a total of 26 patients with TB of the spine, with 21 (80.7%) male
subjects, and a mean age of 40 ± 22.5 years ([Table 1]). Axial pain was the most prevalent symptom (84.6%), with a mean VAS score of 6.6 ± 2.6.
The mean time from symptom onset to diagnosis was of 23.8 ± 24.8 weeks. The mean score
on the ASIA scale ranged from 172.0 in the first assessment to 202.0 after 6 months.0.
Biopsy was performed in all patients: 11 were open, 10 were percutaneous, and 5 were
guided by CT.
Table 1
|
N
|
SI
|
%
|
UI
|
%
|
p-value
|
|
N
|
26
|
9
|
34.6
|
17
|
65.4
|
|
|
Age (mean and standard deviation in years)
|
40 ± 22.5
|
42
|
−
|
40
|
−
|
0.578**
|
|
Gender
|
26
|
9
|
34.6
|
17
|
65.4
|
1.000*
|
|
Male
|
21
|
8
|
38.1
|
13
|
61.9
|
|
|
Female
|
5
|
1
|
20.0
|
4
|
80.0
|
|
|
Diabetes mellitus
|
4
|
1
|
25.0
|
3
|
75.0
|
1.000*
|
|
Chronic kidney disease
|
1
|
0
|
0.0
|
1
|
100.0
|
1.000*
|
|
Neoplasia
|
2
|
0
|
0.0
|
2
|
100.0
|
1.000*
|
|
Duration of the corticotherapy (months)
|
1
|
0
|
0.0
|
1
|
100.0
|
1.000*
|
|
Type of Household
|
26
|
9
|
34.6
|
17
|
65.4
|
0.216**
|
|
Urban
|
12
|
2
|
16.7
|
10
|
83.3
|
|
|
Rural
|
14
|
6
|
42,9
|
8
|
57.1
|
|
|
Treatment
|
26
|
9
|
34.62
|
17
|
65.38
|
0.078*
|
|
Non-surgical
|
17
|
4
|
44.44
|
13
|
76.47
|
|
|
Surgical
|
9
|
5
|
55.56
|
4
|
23.53
|
|
|
Time until cure (months)
|
14.2
|
11.25
|
|
14.24
|
|
0.613**
|
|
Location
|
|
|
|
|
|
1.000*
|
|
Thoracic
|
13
|
5
|
38.5
|
8
|
61.5
|
|
|
Lumbar
|
13
|
4
|
30.8
|
9
|
69.2
|
|
|
Levels
|
|
|
|
|
|
0.316***
|
|
1
|
21
|
7
|
33.3
|
14
|
66.7
|
|
|
2
|
2
|
1
|
50.0
|
1
|
50.0
|
|
|
3
|
2
|
−
|
−
|
2
|
100.0
|
|
|
4
|
1
|
1
|
−
|
−
|
−
|
|
|
Multiple-level disease
|
3
|
1
|
33.3
|
2
|
66.7
|
1.000*
|
|
Complications
|
10
|
5
|
50
|
5
|
50
|
0.16***
|
|
Death
|
1
|
0
|
0
|
1
|
100
|
|
|
Neurological deficit
|
8
|
4
|
50
|
4
|
50
|
|
|
Chronic pain
|
1
|
1
|
100
|
0
|
0
|
|
|
Postoperative infection
|
0
|
0
|
0
|
0
|
0
|
|
|
None
|
16
|
4
|
25
|
12
|
75
|
|
The etiologic agent was isolated (through culture and/or sputum-smear bacilloscopy
and/or positive RMT-TB) in 9 cases. In the remaining 17 patients, isolation was not
possible, and the diagnosis was established through a positive therapeutic test in
11 cases and through an anatomopathological examination compatible with TB in 6 cases
([Table 2]).
Table 2
|
n
|
%
|
|
Isolation of the etiologic agent
|
9
|
34.62
|
|
Pathology
|
6
|
23.08
|
|
Therapeutic testing
|
11
|
42.31
|
Regarding the tests that led to the diagnosis, 5 (19.23%) were bacilloscopies, 11
(42.31%) were anatomopathological tests, 5 (19.23%) were cultures, and 7 (26.92%)
were RMT-TBs. It is important to note that, in the cases in which the agent was isolated,
, if one of these tests were positive, one of the other two or both were also positive.
We identified the exclusive form of TB of the spine (Pott disease) in 14 patients,
and it was associated with pulmonary TB in 7 cases. The disseminated form (2 or more
extrapulmonary sites affected) was identified in 5 patients: compromise of the spine
and lymph node (2), of the spine and knee (2), and of the spine and pleura (1).[18]
Moreover, the lumbar region was affected in half of the cases, and the thoracic region
was affected in the other half ([Table 1]). The disease occurred in 1 level in 21 (80.8%) cases, and in 2, 3 and 4 levels
in 2, 2, and 1 cases respectively. Only 9 (34.62%) cases required surgery ([Table 1]). The of average time from the beginning of the treatment until the cure was 14.2
months. All patients underwent the basic TB regimen, which consists of the administration
of rifampicin, isoniazid, pyrazinamide, and ethambutol (RIPE), and only 1 patient
presented TB that was resistant to isoniazid and ethambutol, which were replaced by
levofloxacin and streptomycin. A smal part of the cases evolved with complications:
1 death and 8 cases of persistent neurological deficit ([Table 1]).
Results of the Groups
Significant differences were observed regarding the relative lymphocyte count (p = 0.025) and the monocyte/lymphocyte ratio (p = 0.009). The lymphocyte count was higher in the UI group (25.35 ± 13.08) compared
to the SI (14.18 ± 7.48) group, while the monocyte/lymphocyte ratio presented a lower
mean value in the UI (0,39 ± 0.22) group compared to the SI group (0.89 ± 0.65; [Table 3]). We analyzed the results of the ROC curve, using as a reference for the relative
lymphocyte count the value of 16.7, that is, values greater than or equal to 16.7%
in relation to the total leukocytes. We found a sensitivity of 76.9% and a specificity
of 62.5%, and the area under the ROC curve was of 0.798, and the 95% confidence interval
(95%CI) was of 0.60 to 0.98. As for the monocyte/lymphocyte ratio, values greater
than or equal to 0.58 presented a sensitivity of 84.6% and a specificity of 75%. The
area under the curve was of 0.846, and the 95%CI was of 0.67 to 1.01 ([Figs. 5] and [6]).
Fig. 5 Receiver operating characteristic curve for lymphocyte counts higher than or equal
to those just below. The reference is the group without isolation of the etiological
agent.
Fig. 6 Receiver operating characteristic curve values for monocyte/lymphocyte ratios higher
than or equal to those just below. The reference is the group without isolation of
the etiological agent.
Table 3
|
N
|
SD/%
|
SI
|
SD/%
|
UI
|
SD/%
|
p-value
|
|
Hemoglobin (g/dL)
|
12.1
|
2.1
|
12.1
|
2.71
|
12.15
|
1.61
|
0.678**
|
|
Leukocytes (mil/μL)
|
8.1
|
2.3
|
7.45
|
2.47
|
8.6
|
2.08
|
0.148**
|
|
Lymphocytes (%)
|
21.1
|
12.4
|
14.18
|
7.48
|
25.35
|
13.08
|
0.025**
|
|
Monocytes (%)
|
8.5
|
2.4
|
9.65
|
3.04
|
7.86
|
1.79
|
0.110**
|
|
Monocyte/Lymphocyte Ratio
|
0.6
|
0.5
|
0.89
|
0.65
|
0.39
|
0.22
|
0.009**
|
|
CRP (mg/dL)
|
7.0
|
6.7
|
7.25
|
3.18
|
6.83
|
8.51
|
0.195**
|
|
ESR (mm)
|
83.3
|
27.0
|
97.5
|
18.81
|
77.6
|
28.48
|
0.179**
|
|
LDH (mg/dL)
|
550.8
|
437.5
|
775.3
|
735.77
|
438.5
|
210.6
|
0.606**
|
|
Creatinine (mg/dL)
|
0.8
|
0.8
|
0.6
|
0.23
|
0.95
|
0.98
|
0.683**
|
|
Total proteins (g/dL)
|
6.8
|
0.6
|
7.23
|
0.68
|
6.5
|
0.47
|
0.131**
|
|
Albumin (g/dL)
|
3.2
|
0.5
|
2.9
|
0.53
|
3.37
|
0.51
|
0.157**
|
|
Positive reactive tuberculin skin test
|
4
|
15.38
|
3
|
33.33
|
1
|
5.88
|
1.000*
|
|
Bacilloscopy
|
5
|
19.23
|
5
|
55.56
|
0
|
0.00
|
0.001*
|
|
Pathology
|
11
|
42.31
|
5
|
55.56
|
6
|
35.29
|
0.637*
|
|
Culture
|
5
|
19.23
|
5
|
55.56
|
0
|
0.00
|
0.001*
|
|
Rapid molecular test
|
7
|
26.92
|
7
|
77.78
|
0
|
0.00
|
0.0001*
|
When studying only the 11 cases in which the anatomopathological examination was compatible
with the disease and dividing these cases according to the bacilloscopy (AFB positive = bacillus
observed in the granuloma), the percentage of lymphocytes among patients with positive
AFB (n: 5; mean: 12.54) and negative AFB (n: 6; mean: 33.18) presented a significant
difference (p = 0.004; [Table 4]).
Table 4
|
n
|
Mean
|
p-value**
|
|
Positive anatomopathological examination
|
11
|
|
|
|
Positive sputum-smear
|
5
|
12.54
|
0.004
|
|
Negative sputum-smear
|
6
|
33.18
|
We observed that the presence of collection within a vertebra or in adjacent paravertebral
tissues, subligamentous spread, and irregularity of the edges of the end plates of
vertebral bodies on T2- and STIR-weighted sequences presented 100% of sensitivity
for the diagnosis of patients in the SI group ([Table 5]). These values refer to the variables in which the etiologic agent was isolated.
Table 5
|
Sensitivity (%)
|
Specificity (%)
|
PPV (%)
|
NVP (%)
|
Accuracy (%)
|
|
Involvement of more than 2 vertebrae in adjacent levels
|
50.0
|
66.7
|
57.1
|
66.7
|
58.8
|
|
Presence of collection within a vertebra or in adjacent paravertebral tissues
|
100.0
|
10.0
|
47.1
|
10.0
|
50.0
|
|
Subligamentous spread
|
100.0
|
20.0
|
42.9
|
20.0
|
50.0
|
|
Destruction of the vertebral body with a decrease in the height of more than 50%
|
85.7
|
30.0
|
46.2
|
30.0
|
52.9
|
|
Involvement of non-contiguous levels in different regions of the spine
|
0.0
|
90.0
|
0.0
|
90.0
|
52.9
|
|
Alteration of the disk morphology with CSF-like isosignal, hypointense on T1, and
hyperintense on T2
|
71.4
|
30.0
|
41.7
|
30.0
|
47.1
|
|
Irregularity of the end plates' margins of the vertebral bodies in T2- and STIR-weighted
sequences
|
100.0
|
0.0
|
41.2
|
0.0
|
41.2
|
|
Invasion of the epidural space by bone or granulation tissue
|
71.4
|
50.0
|
50.0
|
50.0
|
58.8
|
Discussion
Patients in the UI group had a higher mean lymphocyte count and a lower monocyte/lymphocyte
ratio. One possible explanation could be the depletion of T lymphocytes in the SI
group due to the higher bacillary load. O'Shea et al.[19] (2018) compared patients with active TB, latent TB, and controls regarding the bacillary
load and the lymphocyte count. Patients with a higher bacillary load (active form)
had lower a lymphocyte count compared to those with a lower bacillary load, that is,
those with the latent disease (p = 0.017) and the controls (p = 0.003) respectively.
To date, studies on lymphocyte count and the monocyte/lymphocyte ratio have only been
conducted in groups with extravertebral TB.[20]
[21]
[22] Bashir et al.[20] (2014) and Wang et al.[23] (2015) observed that lymphocyte count is lower in patients with TB compared to the
general population. La Manna et al.[24] (2017) found that a ratio higher than 0.28 could predict the incidence of TB with
a sensitivity of 91.04% and a specificity of 93.55% (p < 0.001). Moreover, Liana et al.[21] (2019) observed that a monocyte/lymphocyte ratio higher than 0.476 had a sensitivity
of 95.1% and a specificity of 70% for the diagnosis of TB (p < 0.001). The differences among the studies are probably related to the characteristics
of the study subjects. La Manna et al.[24] compared healthy patients and TB patients, while Liana et al.[21] compared patients with chronic cough and positive and negative bacterial cultures.
Elevated lymphocyte counts are controversial in patients without mycobacterial isolation,
and they indicate a better immune response against infection, resulting in larger
granulomas with fewer bacilli, thus affecting mycobacterial isolation.[25]
[26]
[27]
[28] Another possible complication involved in mycobacterial isolation are the limitations
of the diagnostic tests. Although the culture growth requires up to 8 weeks, the RMT
has a detection limit of at least 116 colony-forming units per milliliter (CFU/mL),
while the culture requires 1 to 10 CFU/mL.
Assessing the possible limitations of this study, we could observe that some variables
were not significantly different between the groups, probably due to the small sample
size.
The diagnosis of Pott disease must involve a combination of criteria, because unsuccessful
isolation of the etiological agent does not exclude the presence of the disease. Therefore,
unsuccessful mycobacterial isolation in the presence of other diagnostic criteria
should not delay the beginning of the treatment. T-cell response should always be
considered.
Conclusion
We did not observe differences regarding the clinico-epidemiological and radiological
characteristics between the groups in whom the etiological agent was isolated or not.
Higher lymphocyte count and lower monocyte/lymphocyte ratio were observed in the UI
group.
