Application of diffusion tensor imaging in brain lesions: A comparative study of neoplastic and non-neoplastic brain lesions
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.246270
Source of Support: None, Conflict of Interest: None
Keywords: Diffusion tensor imaging, fractional anisotropy, mean diffusivity, magnetic resonance imanging, neoplasm
Different brain lesions are encountered in daily clinical practice and it is important to differentiate these lesions to provide appropriate treatment. On conventional magnetic resonance imaging (MRI) sequences, sometimes it is difficult to distinguish benign from malignant brain lesions, and a biopsy is needed to confirm the diagnosis. Diffusion-tensor imaging (DTI) is an advanced MRI sequence based on diffusion weighted imaging (DWI) that non-invasively displays the pathophysiological changes of altered brain structures. DTI provides information on tissue microstructure mainly by employing two parameters: mean diffusivity (MD), which characterizes the presence of obstacles to diffusion; and, fractional anisotropy (FA), which is related to the presence of oriented white matter (WM) structures. Highly compact WM fiber tracts exhibit a higher degree of anisotropy, and DTI provides a sensitive means to detect alterations in the integrity of WM structures. This helps in the preoperative neurosurgical planning by depicting the effects of the tumor on surrounding WM tracts. When a tumor has only displaced WM tracts, it has normal FA values compared with the contralateral side; whereas, FA is increased when the tumor has infiltrated the WM tract. When the tumor has completely disrupted a white matter tract, it is not identifiable on either FA or directionally encoded color maps.,,
The perilesional edema that occurs in neoplastic and non-neoplastic brain lesions differs in composition. The perilesional edema of glioma contains the affected brain tissue and infiltrating tumor cells with glial cell alterations and increased aquaporin-4 expression. Capillaries related to a metastatic brain lesion resemble the organ of origin and do not possess the unique blood brain barrier function, which makes them highly leaky resulting only in perilesional vasogenic edema. The DTI assessment in the presence of peritumoural edema surrounding HGG and metastases has shown an increase in the MD and a decrease in the FA values compared to the contralateral white matter (CWM). DTI has been helpful in the differentiation of various infected brain lesions also. In the most commonly encountered clinical setting, intracranial tuberculomas are sometimes difficult to differentiate from neoplastic brain lesions. MR spectroscopy and DWI have enabled a better differentiation of tuberculomas from other brain lesions. This differentiation is of immense importance as both have different therapeutic approaches and prognosis. NCC is also common in developing countries and difficult to differentiate from a tuberculoma. DWI and magnetic resonance spectroscopy (MRS) studies have shown good results, in conjunction with the conventional MRI, in the differentiation of these two entities.,,, The aim of the present study was to evaluate the DTI derived FA and MD diffusion parameters in brain lesions and compare the utility of these parameters in the differentiation of neoplastic from non-neoplastic lesions.
Fifty newly diagnosed and untreated patients (their mean age being 39 years with an age range of 18-70 years and their male: female ratio being 27:23) of brain lesions with non-enhancing perilesional white matter edema were included in the study. These patients were referred for MRI to plan their clinical management, based upon the routine clinical practice. They were included in the study after approval of study by the institute ethics committee. None of the patients had previously undergone any form of specific medical or surgical treatment. The final diagnosis of these patients was obtained based on their clinical and biochemical parameters, their imaging information, the organism cultured, the surgical findings, the histopathology of the lesion and the follow-up course. Based upon these parameters, these patients were divided into the neoplastic and the non-neoplastic group. After taking informed consent and fulfilling the inclusion and exclusion criteria, the patients underwent an MRI including the DTI sequence. All patients underwent an MR imaging on a 3 Tesla General Electric scanner (Signa Hdxt, General Electric, Milwaukee, USA), using a 12 channel head coil. Th following MR sequences were planned - Axial T2, T1/T2 fluid attenuated inversion recovery (FLAIR), diffusion weighted imaging (b value = 1000mm2/sec) and diffusion tensor imaging [DTI] (frequency- 128, phase- 128, phase field of view (FOV)-1.00, repetition time (TR)-12400, echo time (TE)-86.4, slice thickness-3.0mm, spacing-0, FOV-24, axial plane, number of directions- 30, number of excitations [NEX]=1).
Postprocessing and data analysis
The DTI data was transferred to the workstation (ADW4.4, General Electric, USA) for post- processing and the images were analyzed using FuncTool Software with automated generation of diffusion maps. After using a motion correction algorithm for compensating for head motion and image distortion due to the presence of eddy currents, the FA and color-coded structural diffusion tensor maps were generated. Prior to evaluation of the diffusion maps, conventional MRI sequences were assessed for the lesion localization, characterization and enhancement pattern. For quantitative evaluation, we manually placed three circular ROIs (regions of interests, 5-10 mm2) within the lesion and in the region of perilesional edema (within 1 cm from the lesion) on DTI post-processed maps. Means of FA and MD of lesional (FAL, MDL) and perilesional edema (FAPE, MDPE) were taken from DTI post processed images and not normalized to cerebral white matter (CWM).
The normality of continuous variables (the DTI parameters) was tested using Kolmogorov Smirov test. For normal data, mean ± standard deviation (SD) was used as descriptive statistics. The FA and MD values between the two groups (neoplastic and non-neoplastic) were compared using Independent t test. A P < 0.05 was considered statistically significant and P < 0.001 highly significant. A statistical analysis was performed using the Statistical Package for the Social Sciences, version 23 (SPSS-23, IBM, Chicago, USA).
Of the 25 neoplastic lesions (these patients had a male:female ratio: 11:14; age range: 41-50 years), 10 were proven cases of glioblastoma multiforme (GBM), and four lesions were assigned as low grade glioma (LGG) on the basis of the imaging features. The patients harbouring these lesions were under regular follow up at the time of the study. Of the 11 metastatic brain lesions, 7 lesions were proven primary carcinoma of the breast and 4 lesions were diagnosed to be primary carcinoma the lung. Of the 25 non-neoplastic lesions (these patients had a male: female ratio: 16:9; age range-18-30 years), 13 were designated as tuberculomas, and 12 as neurocysticercosis (NCC), based on their imaging features, clinical course and treatment response. The mean FAL, FAPE, MDL and MDPE values were obtained from each subtype of the main group, as given in [Table 1]. The mean FA and MD values from the lesion and from the region of the perilesional edema of the neoplastic group were compared with that of the non-neoplastic group [Table 2]. The perilesional MD was higher for the non-neoplastic group compared to the neoplastic group, which was statistically significant (P value = 0.015); while the difference in lesional MD was statistically insignificant (P = 0.362). The mean FAPE and FAL values were higher in the neoplastic group compared to the non-neoplastic group; however the difference was statistically insignificant (P = 0.156 and P = 0.070 respectively).
After the main group comparison, the group subtypes were differentiated based on the lesional and perilesional FA and MD values. HGGs [Figure 1] were compared to metastases [Figure 2]; the perilesional FA was higher in HGGs, while the perilesional MD in metastases was higher. These results were statistically highly significant (P < 0.001). Lesional FA and MD were higher in HGGs compared to metastases; however, these results were statistically insignificant (P = 0.079; P = 0.842, respectively). Similar results were found when HGGs were compared to tuberculomas [Figure 3]. The perilesional FA was significantly higher in HGGs; and, the MDPE in tuberculomas was significantly higher (P value < 0.001). The FA and MD from the lesion were statistically insignificant (P = 0.918; P = 0.654), when compared between HGGs and tuberculomas.
The next comparison was done between two closely resembling lesions, metastases and tuberculomas. The lesional FA was statistically higher in tuberculomas compared to metastases (P = 0.046); while, the differences in FAPE, MDPE and MDL were not statistically significant between tuberculomas and metastases (P = 0.201; P = 0.710; P = 0.941, respectively). The differentiation of a NCC lesion [Figure 4] from a tuberculoma always remains a diagnostic challenge, and we found no statistically significant difference in the mean FAPE (P = 0.456), FAL (P = 0.067), MDPE (P = 0.201) and MDL (P = 0.552) values between the two lesions.
In this study, we have quantitatively assessed the DTI derived FA and MD parameters from the lesion and perilesional edema in distinguishing neoplastic from non-neoplastic lesions. This was based upon the premise that brain lesions have different compositions of perilesional edema. Majority of the benign lesions only have vasogenic perilesional edema, while gliomas have infiltrating cells along with vasogenic edema. Taking advantage of this diversity in the composition of perilesional edema, subtle differences in the DTI parameters were evaluated in our study. On comparing, it was found that the mean MDPE values were significantly higher in the non-neoplastic group (P = 0.015) compared to the neoplastic group, while the differences in MDL, FAL and FAPE values were not statistically significant.
On further subgroup analysis of HGG and metastasis, the mean FAPE was found to be higher in HGG, while the MDPE was found to be higher in metastases (P value < 0.001), and the differences in FAL and MDL were statistically insignificant. A few studies comparing the FA and MD values in perilesional edema of HGG and metastases are available and the comparison with previous studies is given in the following table [Table 3]. Glioblastomas produce plenty of tumor-specific extracellular matrix and have shown a higher degree of peritumoral infiltration than metastases, leading to a higher anisotropy, which explains the higher perilesional FA in HGG.,,, MD is mainly determined by an increased extracellular water, and metastatic lesions have a high expression of vascular endothelial growth factor, which increases the vascular permeability. The later factor is responsible for a higher MDPE in metastases than in HGGs.,,,,,,,,,,,,,,,
On comparing HGGs with tuberculomas, a statistically significant higher perilesional FA and decreased MD was found in HGGs than in tuberculomas because of peritumoral infiltration and less vasogenic edema in HGGs. However the difference in lesional FA and MD were not statistically significant. Gupta et al., found a significantly decreased FA, CL (linear tensor) and Cp (planar tensor) and increased MD and Cs (spherical tensor) in brain tuberculomas compared to normal white matter. However, no DTI study that is differentiating HGGs and tuberculomas based on perilesional diffusion characteristics has been found in the literature till date.
Tuberculomas showed a statistically higher FAL compared to metastases, which points towards more WM microstructural destruction in metastases. This is because a higher FA value results from a more regular and orderly arrangement of the white matter fibers. No statistically significant difference was found in the perilesional FA and MD values between metastases and tuberculomas, as both have vasogenic edema only. Recently, a study by Chatterjee et al., using DWI parameters showed overlapping apparent diffusion coefficient values between tuberculomas and metastases. NCC lesions are common in highly endemic areas, and sometimes, it becomes difficult to differentiate them from tuberculomas on imaging alone. A few studies have shown the role of DWI and MR spectroscopy (MRS) in differentiating tuberculomas from NCC granulomas; however the search of literature did not reveal any study differentiating these entitites using DTI.,, We have found no statistically significant difference in the FA and MD values between tuberculomas and NCC granulomas.
Our study has shown promising results in the differentiation of brain lesions, depending on the two important DTI derived FA and MD parameters obtained from the lesion and from the region of perilesional edema. The number of patients recruited in the study is, however, small. A larger sample size would be helpful in further establishing the role of DTI in differentiation of brain lesions of different aetiologies.
DTI studies provide valuable information regarding the status of white matter tracts adjacent to the intra-axial brain lesions. FA depends on the overall orientation and density of WM fibres and MD depends on the degree of vasogenic edema. Giomas present with more WM fibre destruction and less vasogenic edema in comparison to non-neoplastic lesions, and a higher perilesional MD favours a non-neoplastic pathology. Thus, the DTI sequences, in addition to the routine MR sequences, may help in differentiating brain lesions of different aetiologies. This may prove to be of great significance in specifying treatment and in improving the patient's quality of life.
We would like to acknowledge Mr. Abhay Jain, Mr. Abhay Jha and Mr. Dharmendra, MR technicians, Department of Radiodiagnosis, SGPGIMS, Lucknow for their help in the image acquisition. No grant has been taken from any organization.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3]