Abstract
Background
Central nervous system (CNS) tuberculomas often mimic tumors on conventional imaging, differentiation of which may not be possible without invasive tissue sampling. Diffusion tensor imaging (DTI), owing to its unrivalled property of characterizing molecular diffusion, may help in better lesion characterization and tractography may help understand the pattern of white matter involvement by tuberculomas.
Purpose
To estimate qualitative and quantitative diffusion tensor changes in brain tuberculomas and to evaluate patterns of white matter involvement using 3D tractography.
Material and Methods
Thirty patients with brain tuberculomas were evaluated on a 3-T magnetic resonance scanner. Diffusion tensor images were acquired along 20 non-colinear encoding directions with two b-values (b = 0, b = 1000). Regions of interest (ROIs) were drawn on quantitative fractional anisotropy (FA) and apparent diffusion coefficient (ADC) maps in the center of the tuberculoma and perilesional area. Similar ROIs were placed in contralateral hemispheres for comparison. Tractography maps were also generated.
Results
Mean FA in the center and perilesional area of tuberculomas were 0.098 ± 0.041 and 0.311 ± 0.135, respectively. ADC values in corresponding regions were 0.920 ± 0.272 ×10−3 mm2/s and 1.157 ± 0.277 ×10−3 mm2/s. These values were significantly different compared to contralateral similar brain parenchyma. Tractography revealed interruption of white fibers in the center with deviation of fibers at the periphery in the majority of tuberculomas with none showing infiltration of white matter described in tumors.
Conclusion
Significant qualitative as well as quantitative DTI changes were seen in tuberculoma and perilesional areas compared to contralateral hemisphere with tractography showing a pattern different from that described in tumors. These findings may help to differentiate tuberculomas from infiltrating tumors.
Introduction
Central nervous system (CNS) tuberculosis (TB) has shown an increased incidence due to the outbreak of HIV even in high-income nations. The diagnosis was conventionally based on clinical, biochemical, and radiological methods with definitive diagnosis being based on isolation of mycobacteria in cerebrospinal fluid (CSF) culture. However, due to overlapping clinical presentation as well as the paucibacillary nature of the disease, diagnosis may remain difficult (1). On conventional magnetic resonance imaging (MRI), the T1-weighted (T1W) and T2-weighted (T2W) signal of tuberculoma depends upon the stage of maturation of the tuberculoma: whether non-caseating, caseating with a solid center, or caseating with a liquid center. The characteristic T2 hypointense signal, which is well described in literature, is seen in solid caseating tuberculomas and is due to the presence of macrophage infiltration, release of free radicals, and fibrosis (2). However, the tuberculomas may also show a T2 hyperintense signal when liquid caseation occurs. Heterogenous signal intensity may be seen in large conglomerate tuberculomas where few lesions may appear hypointense while others are T2 hyperintense due to varying stages of maturation. Thus, due to myriad imaging appearances on conventional MRI, tuberculomas mimic other space-occupying lesions including other benign-looking infective lesions to aggressive neoplasms, posing a challenge to diagnosis. Many advanced MRI techniques, such as magnetization transfer imaging, diffusion imaging, and proton MR spectroscopy, have been used to characterize tuberculomas and provide better lesion characterization than conventional MRI; however, differentiation may not be possible in all cases (2).
Diffusion tensor imaging (DTI) is a technique that is used to evaluate white matter tracts. Most pathologies affecting white matter disrupt organized white matter structure and produce a change in diffusion anisotropy which is most commonly assessed by the scalar parameter known as fractional anisotropy (FA). The magnitude of diffusivity also changes and is assessed quantitatively by mean diffusivity, also represented by apparent diffusion coefficient (ADC). DTI has provided additional value compared to conventional MR techniques in multiple studies involving brain tumors, ischemic brain injuries, demyelination, and dementia (3). Qualitative and quantitative evaluation of DTI parameters, especially in the perilesional area, has not been attempted in any large cohort of patients so far. The aim of the present study was to assess qualitative and quantitative DTI changes in tuberculoma as well as the surrounding white matter tracts (deviation or truncation or infiltration), which may help in diagnosis in cases of dilemma.
Material and Methods
Patients were recruited from individuals undergoing brain MRI for clinical suspicion of CNS TB for this cross-sectional study after obtaining written informed consent. Thirty patients aged >12 years who were found to have parenchymal tubercular lesions on MRI and subsequently confirmed to have TB were included after obtaining informed written consent. The diagnosis of CNS TB was based on characteristic CSF findings (n = 27). In cases where CSF was non-confirmatory, the presence of clinical features consistent with CNS TB with diagnosed TB of another body part (n = 3) was taken as a criterion for diagnosis. Relevant clinical details and CSF biochemical findings were also recorded. The study was approved by Maulana Azad Medical College Institutional Ethical Committee.
MR acquisition
MRI of the brain was performed on a 3-T MR scanner (Magnetom Skyra, Siemens, Erlangen, Germany), using a 20-channel head coil. Spin echo sequences were used to obtain T1W, T2W, and fluid-attenuated inversion recovery (FLAIR) images. Multi-directional diffusion-weighted images (MDDWI) were acquired using EPI sequence with two b-values (b = 0 and 1000 s/mm2) along 20 non-colinear encoding directions. The following parameters were used for MDDWI: repetition time = 3700 ms (concatenation 1); echo time = 92 ms; number of excitations [NEX] = 13 and 1, respectively, for b = 0 images and b = 1000 images; matrix size = 128 × 128; and field of view (FOV) = 220 mm. This was followed by a 3D Fourier transformed MPRAGE sequence for postcontrast imaging after intravenous injection of gadolinium contrast agent at a dose of 0.1 mmol/kg followed by 20 ml saline chase.
Postprocessing and image analysis
Postprocessing of the DTI data was performed on a Syngovia workstation (vB30A_HF06) that auto-generated gray-scale FA and ADC maps as well as color-coded FA maps (based on standard color coding where blue represented craniocaudal, green represented anteroposterior, and red represented mediolateral directions, respectively). Qualitative analysis of color FA maps was carried out for all parenchymal lesions and changes in the color intensity (i.e. absent color, reduced color intensity, normal color intensity) and directional color hues (change in the color of white fiber compared to contralateral similar white matter fiber) in the lesion center and perilesional area were recorded and compared to the contralateral hemisphere. Quantitative assessment of DTI scalar parameters such as FA and mean diffusivity (MD) was also done by drawing circular ROIs on gray-scale FA and ADC maps in the center and perilesional edema area of the tuberculoma (Fig. 1). Care was taken to avoid inclusion of gray matter in the ROI. Similar ROIs were also placed in the corresponding normal-appearing contralateral hemisphere for comparison.
(a) Multiple conglomerate tuberculomas in right parieto-temporal region on postcontrast image. (b) Quantitative FA image and (c) ADC image showing placement of ROI in tuberculoma, perilesional edema, and contralateral parenchyma. ADC, apparent diffusion coefficient; FA, fractional anisotropy; ROI, region of interest.
For tractography, color FA maps were fused with MPRAGE images and elliptical volume of interest (VOI) were placed on fused images enclosing the cross-section of tuberculoma and adjacent brain parenchyma. Similar VOI was also drawn in a contralateral similar location. Relevant construction algorithms (AND, OR, NOT operations) with FA threshold of 0.2 and angle threshold of 45° were used to generate tractography images of the desired region. The patterns of tract alteration as described by Jellison et al. (4) were used. A tract was considered “deviated” if fiber continuity was maintained but it was abnormally located or had changed direction (as indicated by change in directional color hues) as a result of the lesion mass effect. It was considered “interrupted” if any portion of the tract was discontinuous or appeared truncated. Tract “infiltration” would be suggested if it showed significantly reduced anisotropy and abnormal color of the fibers due to disorganization of fibers but no deviation/displacement of fibers was visibly observed. A combination of these patterns was also looked for. Tract characterization in the present study for each tuberculoma was independently assessed by two radiologists and discrepancy was resolved by mutual agreement.
Statistical analysis
Qualitative data were expressed as percentages. Quantitative data were expressed by mean and standard deviation and percentages. Mann–Whitney U-test was used for the statistical analysis of quantitative DTI data.
Results
The present study included 30 adult patients (age range = 13–55 years; mean age = 34 years; female : male ratio = 1.3 : 1). The most common presenting complaints were fever (96.67%) and headache (93.3%). On examination, neck rigidity suggesting meningitis was seen in 50% of patients. Two patients (6.7%) had cranial nerve palsy involving the third and fourth cranial nerves. Focal neurological deficit was observed in 7/30 (23.3%) patients. Evidence of extracranial TB was found in 10/30 (33.33%) patients, all of whom had pulmonary involvement.
A combination of CSF leukocytosis with lymphocyte predominance, decreased sugar level, and raised proteins was seen in only 50% of patients. All patients had raised CSF protein, 76.67% showed CSF pleocytosis with lymphocyte predominance while low levels of CSF-glucose were observed in 53.3% of patients. Levels of CSF-ADA were raised in 59.2% of patients (cutoff taken as 10) while nucleic acid amplification tests (NAAT) positivity was observed in only 24% of patients.
On MRI, a total of 60 tuberculomas were identified in 30 patients. Of them, 71.7% showed a hypointense signal on T2W/FLAIR images while 23.3% showed a T2 hyperintense signal. The remaining tuberculomas showed mixed signal intensity with some areas appearing hypointense and some hyperintense. Such a heterogenous signal was seen in large conglomerate lesions. Perilesional edema was observed in 85% of tuberculomas, the diameter of which were in the range of 10 mm in small lesions to 45 mm in large conglomerate lesions (measured from one edge of the edema to the other, including the tuberculoma). The median diameter of edema was 17 mm seen with a tuberculoma of size 7 mm. Most tuberculomas had edema that was approximately double the size of the lesion, while a higher degree of edema in the range of thrice the size of tuberculoma was seen in large conglomerate tuberculomas.
Of the 60 tuberculomas identified on T2/FLAIR images, 10 were entirely located in gray matter and were not visualized well on the FA map. These tuberculomas were small in size and had no or very minimal perilesional edema. Thus, 50 tuberculomas were included for analysis of DTI changes. Of these, 31 tuberculomas (51.67%) were located in the cerebral hemisphere, 11 (18.33%) in the cerebellum, and 4 (6.67%) each in the thalamus and brainstem.
Assessment of qualitative color FA maps revealed complete absence of color in the center of the tuberculoma in 90%. In the remaining 10% of tuberculomas, which were conglomerate lesions, there were areas of both decreased color intensity and absent color.
In the area of perilesional edema, a combination of reduced color intensity and changed color hues was seen in 29/50 (58%) tuberculomas. In two of these lesions, an area of complete loss of color was also noted besides the aforementioned changes, suggesting complete loss of anisotropy. In 8/50 lesions (16%), preserved color hues but only decreased color intensity in the area of white matter edema was noted. Only deviation with changed color hues but no reduction in color intensity was seen in 8/50 (16%) lesions. In 5/50 (10%) lesions, no perilesional edema was identified on conventional MRI and no abnormality was identified in the perilesional white matter on color FA maps.
Tractography maps generated by fusion of color FA maps and MPRAGE images showed three patterns: interruption and deviation (30/50, 60%) (Fig. 2); only interruption (12/50, 24%) (Fig. 3); and only deviation (8/50, 16%) (Fig. 4).
(a) Multiple conglomerate tuberculomas in right frontoparietal region in a 20-year-old woman complaining of altered sensorium and left hemiplegia. (b) On color FA map, there is non-visualization of subcortical red fibers in the region of tuberculomas (arrowhead). In the area of the perilesional edema, decreased color intensity and changed color hues from blue to violet suggest deviation noted in right internal capsule explaining patient's symptoms (yellow arrow). (c, d) Tractography images show interruption of subcortical red fibers compared to opposite sides with deviation of fibers in the perilesional area indicated by changed color and orientation of fibers (white arrows). FA, fractional anisotropy.
(a) Ring enhancing tuberculoma in left thalamus (arrow) and a nodular enhancing tuberculoma in right parieto-occipital region (arrowhead) in an 18-year-old man who presented with fever, headache, and vomiting associated with seizures. (b) On color FA map, there is an absence of color in the lesion with non-visualization of green anteroposterior fibers compared to the contralateral side (arrow). (C) Tractography image showing interruption of green antero-posterior oriented fibres and red media-lateral oriented fibres in the region of both tuberculomas (arrow). No changed colour hues or abnormally oriented fibres are seen in periphery to suggest deviation in the perilesional area.
A 22-year-old woman presented with fever, headache, and neck rigidity. (a–c) Fused tractography images showing tuberculoma in the pons causing interruption of few fibers of the blue corticospinal tract while causing posterolateral deviation of the majority of fibers (arrow in (a)). (b, c) Sequential coronal images show tuberculoma in the pons causing anterior deviation of pontine crossing fibers as indicated by the change in the orientation/shape but maintained hues (arrows in (c)). No associated neurological deficit was seen in this patient.
Quantitative DTI in tuberculoma
Mean FA values in the center and perilesional white matter of all 50 tuberculomas were 0.098 ± 0.041 and 0.311 ± 0.135, respectively. Mean ADC values in the respective areas were 0.92 ± 0.272 ×10−3 mm2/s and 1.157 ± 0.277 ×10−3 mm2/s.
As anisotropy and diffusivity values vary in the different regions of the normal brain with high anisotropy values seen in compactly arranged fiber bundles in the brainstem compared to the cerebellum, where it is low (5,6), tuberculomas were categorized based on their location as cerebral, cerebellar, thalamic, and brainstem tuberculomas. Mean FA and MD values were obtained for each of these regions separately. When compared to contralateral parenchyma, FA values were low and ADC values were high in the tuberculoma center and perilesional white matter for tuberculomas in all locations; however, the difference was statistically significant for cerebral and cerebellar lesions (P < 0.01). Statistical significance could not be assessed for thalamic and brainstem tuberculomas due to the small sample size in these groups. Relative FA decrease in the tuberculoma centers compared to contralateral parenchyma was 81% for cerebral lesions and 70% for cerebellar lesions. Relative MD increase in the tuberculoma centers compared to contralateral parenchyma was 26% in cerebral lesions and 11% in cerebellar lesions (Tables 1 and 2).
Comparison of mean FA in tuberculomas with contralateral parenchyma (n = 50).
Values are given as mean ± SD.
FA, fractional anisotropy.
Comparison of MD/ADC values in tuberculoma with contralateral hemisphere (n = 50).
Values are given as mean ± SD.
ADC, apparent diffusion coefficient; MD, mean diffusivity.
Discussion
CNS TB is a condition associated with considerable morbidity and even mortality. Clinical features are often non-specific, with fever and headache being the most common presenting complaints. Focal neurological deficit was observed in 23.33% of patients. Of these patients, 57.1% had large conglomerate tuberculomas in eloquent cortex, while in the remaining patients there were infarcts in the right middle cerebral artery territory and ganglio-capsular region.
Typical CSF features of CNS TB, including a combination of CSF leukocytosis with lymphocyte predominance, decreased sugar level, and raised proteins, was seen in only 50% of patients, consistent with the previous literature (1). The level of CSF-ADA, which has a sensitivity in the range of 44%–100% (1), was raised in 59.2% of patients, while the most definitive test, NAAT, depicted mycobacterial DNA in CSF in 24% of patients, consistent with the literature (7). This low sensitivity is due to the presence of inhibitors in CSF that precludes mycobacterial detection. This implies that biochemical features alone cannot be relied upon for diagnosis in a number of cases and thus radiology has an immense value in diagnosis.
On conventional MRI, the majority of tuberculomas showed a T2 hypointense signal. However, this characteristic T2 hypointense signal also overlaps with that described in lymphoma, glioblastoma, and fungal granulomas. In addition, the remaining tuberculomas showed a hyperintense or heterogenous signal, especially the large conglomerate lesions. The majority of tuberculomas had surrounding perilesional edema that was quite pronounced (almost double or triple the size of tuberculoma) in larger lesions resembling high-grade tumors. Thus, due to myriad imaging appearances on conventional MRI, diagnosis may be difficult in some cases.
DTI is a sensitive tool that has been extensively studied in evaluation of brain neoplasms. It allows differentiation between infiltrated white matter tracts from deviated ones in the perilesional edema area of tumors, thus allowing safer tumor-free margins (8). Scalar DTI parameters—FA and MD—have also been used in the differentiation of glioblastoma from metastasis where these parameters were significantly different within the lesion and in the perilesional area of the two lesions (9). Infections have been evaluated to a much lesser extent on DTI. A few studies with relatively smaller sample sizes have reported significantly different DTI parameters in pyogenic abscesses compared to other cystic intracranial lesions (10,11). Since it is known that tuberculoma is a great mimicker and frequently figures in differential diagnosis of brain tumors, it is important to be aware of DTI changes in them. In addition, in surgical candidates where either a resection, drainage, or biopsy is indicated for a large tubercular granuloma/abscess, DTI may help in neuro-navigation; it has been reported in studies that DTI-based neuro-navigation allows maximal safe resection of tumors with minimal damage to vital white matter fibers (8).
On qualitative assessment of color FA maps, a complete absence of color was noted in most tuberculomas with a small percentage showing areas of both reduced color intensity and absence of color. All these latter lesions were large conglomerate tuberculomas. The area of perilesional white matter showed a combination of reduction in color intensity along with change in color hues suggesting change in direction/deviation of white matter fibers in the majority of tuberculomas. In two of these lesions, an area of complete loss of color was also noted besides the aforementioned changes, suggesting complete loss of anisotropy due to fiber disruption. Infiltration of white fibers (suggested by changed color hues without associated deviation) was observed in none of the lesions.
White matter tractography using DTI data to analyze the pattern of white matter involvement by tuberculoma showed three patterns: interruption and deviation (60%); only interruption (24%); and only deviation (16%).
Significant interruption of tracts due to tuberculomas in an eloquent region was seen in only four patients. These patients had a neurological deficit in the form of paraplegia/hemiplegia on clinical examination. Few patients had interrupted fiber tracts on tractography; however, these were located in non-eloquent regions such as the occipital cortex. Some lesions were located in the cerebellum where, due to inherent low anisotropy, continuous fiber tracking is difficult. The tuberculomas that were located in the brainstem were small (<7 mm in size) and caused interruption of only few fibers with predominantly deviated fibers and were not associated with any neurological deficit despite being located in compactly placed white matter tracts in the pyramidal tracts. Thus, larger lesions, located in sensorimotor cortex with significant interruption of white matter fibers only resulted in motor deficit. Thus, DTI may help in disease prognostication.
Not many studies described tractography changes in tuberculomas; however, in a case of brainstem tuberculoma, there was predominant displacement of fiber tracts in the region of the tuberculoma with few truncated corticospinal tracts found adjacent to the lesion, which correlated to the patient's neurological deficit (12). None of the tuberculomas on tractography showed infiltration of white matter tracts. Such pattern of predominant infiltration and destruction of white matter tracts is a feature seen in high-grade gliomas (8).
On quantitative analysis of FA maps, mean FA in the center and perilesional area was 0.098 ± 0.313 and 0.311 ± 0.135, respectively. This low FA value could possibly be due to the disruption of the axonal membrane and myelin resulting in loss of ordered microstructure in the region of infection and edema (13). Similar mean FA values have been described in the tuberculoma center by some authors; however, reported mean FA in the perilesional area was lower than our mean observation (14,15). This could be due to differences in ROI placement as well as different MR systems and gradients being used (16). An increasing FA trend from the tuberculoma center towards the rim and perilesional white matter, as seen in the present study, has also been described. This is due to higher damage of cell membrane integrity and myelin in the tuberculoma center (13). The mean FA obtained in tuberculomas were also different and were lower than that reported in similar T2 hypointense lesions such as lymphoma (mean FA = 0.14 ± 0.024) and glioblastoma (mean FA = 0.229 ± 0.069) (17).
Mean MD values in the center of the tuberculomas and perilesional edema were 0.92 ± 0.272 ×10−3 mm2 /s and 1.157 ± 0.277 ×10−3 mm2/s, respectively. Similar ADC values have been reported in few studies (13,15) with smaller sample sizes.
When compared to contralateral similar regions, the mean FA values were lower and MD values were higher in tuberculomas in all regions with the statistically significant difference observed for cerebral and cerebellar tuberculomas. Relative FA decrease in tuberculomas center compared to contralateral parenchyma was 81% for cerebral lesions and 70% for cerebellar lesions, while the relative MD increases in the respective regions were 26% and 11%. Such analysis has not been done previously. The ratio may allow more accurate and uniform comparison of FA across scans performed at different times and on different MR scanners. In addition, larger multicentric studies in this direction would help to better understand the DTI changes in tuberculoma and aid in differentiating them from other space-occupying lesions.
In conclusion, tuberculomas often mimic tumors on conventional imaging, and since these are commonly encountered lesions on MRI, it is important to be aware of DTI changes and white matter alteration by them. In the present study, there were reduced FA values and increased MD values in tuberculoma centers as well as in the perilesional area compared to the contralateral similar region in all locations, consistent with previous studies. On tractography, there was interruption of white fibers in the center of the tuberculoma with deviation of fibers at periphery. The FA values were different from those described in glioblastoma and lymphoma in the literature. In addition, it has been stated that tumors predominantly cause infiltration of surrounding white fibers, which was not displayed by a single tuberculomas in the present series. Thus, DTI may help in differentiating tuberculomas from infiltrating gliomas and lymphoma, but probably not from all tumors. Further larger multicentric studies in this direction may further elaborate the role of DTI in differentiating tuberculomas from other space-occupying lesions.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.
