Abnormal brain white matter in patients with right trigeminal neuralgia: a diffusion tensor imaging study

Twenty-nine patients (age range 35–77 years; 20 females and 9 males) with right-sided TN and 35 healthy control subjects (age range 41–74 years; 27 females and 8 males) were selected for this study. Patients were enrolled from Department of Functional Neurosurgery of Pingjin Hospital of Logistics University of Chinese People’s Armed Police Forces (Additional file 1: Table S1) and control subjects by newspaper advertisement. The patients belonged to a consecutive series of patients who had undergone evaluation for MVD surgery between 2014 and 2016. All these patients had a long duration (more than 1 year) complaint of classical TN according to the International Classification of Headache Disorders criteria (third edition) [19] and had high-resolution imaging to exclude secondary causes of TN. Visual analogue scale (VAS) [20] was used to assess pain intensity in the TN patients. All patients were measured during a painful attack and on medications. Exclusion criteria [18] for both the patients and controls were as follows: (1) other headache disorders; (2) chronic pain elsewhere; (3) previous TN operations; (4) untreated hypertension or diabetes mellitus; (5) left-handed; (6) alcohol or illicit drug abuse, or current intake of psychoactive medications; and (7) MRI contraindications, such as claustrophobia and metallic implants or devices in the body. This study was approved by our institutional review board, and written informed consent was obtained from all patients and control subjects.

DTI data were obtained using a 3.0-T MR scanner (Verio system; Siemens, Erlangen, Germany) with a 12-channel head coil. Comfortable and tight foam padding was used to limit head movement. Diffusion weighted images were obtained using a single-shot echo planar imaging (EPI) sequence. The scanning location was in alignment with the anterior-posterior commissural plane. The integral parallel acquisition technique (iPAT) was used and the acceleration factor was 2, which can decrease image distortion from susceptibility artifacts. Diffusion sensitizing gradients were applied along 64 non-collinear directions (b = 1000 s/mm²) together with an acquisition without diffusion weighting (b = 0 s/mm²). The imaging parameters were applied as follows: 48 continuous axial slices, slice thickness of 3 mm and no gap, field of view (FOV) = 256 mm × 256 mm, repetition time/echo time (TR/TE) = 8000/95 ms, and matrix size = 128 × 128. The reconstruction matrix was 256 × 256 with a voxel dimension of 1 mm × 1 mm × 3 mm.

All diffusion weighted images were carefully checked by three radiologists to exclude apparent artifacts resulted from instrument malfunction and subject motion. DTI data was preprocessed using FMRIB’s diffusion toolbox (FDT, http://​www.​fmrib.​ox.​ac.​uk/​fsl, FSL 4.0) [21]. First, the eddy current distortions and motion artifacts were corrected by using the affine alignment of each diffusion-weighted image to the image of b = 0 s/mm² in the DTI dataset. Then, non-brain tissue from the average b0 image was removed using the FMRIB’s Brain Extraction Toolbox, BET. The brain mask was applied to the rest of the diffusion-weighted images. Finally, the diffusion tensor was estimated for each voxel using the DTIFIT function via linear regression to derive FA, MD, AD and RD maps.

The following steps were adopted to perform voxel-wise analysis of whole brain white matter measures using the TBSS package (http://​www.​fmrib.​ox.​ac.​uk/​fsl/​tbss/​index.​html) [22]. All subjects’ FA images were aligned to a template of the averaged FA images (FMRIB-58) in Montreal Neurological Institute (MNI) space using a nonlinear registration algorithm implemented in FNIRT (FMRIB’s nonlinear registration Tool) [23]. After transformation into the MNI space, a mean FA image was generated and thinned to create a mean FA skeleton of white matter tracts. Each subject’s aligned FA images were then projected onto the mean FA skeleton according to filling the mean FA skeleton by FA values, resulting in an alignment-invariant representation of the central trajectory of white matter pathways for all subjects. These FA values were obtained by searching perpendicular to the local skeleton structure for maximum value, which was from the nearest relevant tract center. During the former registration step, this second local coregistration step can alleviate the malalignment of diffusion-weighted images. Next, this process was repeated for each subject’s MD, AD and RD map using the individual registration and projection vectors obtained in the FA nonlinear registration and skeletonization.

The demographic and clinical data were compared between the two groups using independent-sample t-test for age and Chi-square test for sex distribution, which was conducted with Statistical Package for the Social Sciences version 22.0 (SPSS, Chicago, Ill, USA). Differences were considered significant when P was less than 0.05.

In our study, 35 patients with right-sided TN were enrolled. Due to the abnormal data quality and no surgery, 6 patients were excluded in this study. Therefore, 29 patients (age range 35–77 years; 20 females and 9 males) and 35 healthy control subjects (age range 41–74 years; 27 females and 8 males) were selected for this study. Demographic and clinical characteristics of each group are summarized in Additional file 1: Table S2. The self-reported duration in TN patients was 10.2 ± 9.6 years (range: 1–30 years) and the VAS was 5.9 ± 3.1 (range: 2–10). There were no significant differences (P > 0.05) between TN patients and healthy controls in age and gender.

Compared with the control group, the TN group showed significantly lower FA in the bilateral superior corona radiata, bilateral anterior corona radiata, body of corpus callosum, splenium of corpus callosum, genu of corpus callosum, left cingulum, left superior fronto-occipital fasciculus, bilateral anterior limb of internal capsule, left posterior limb of internal capsule, left external capsule, left fornix cerebri, internal sagittal stratum and left cerebral peduncle (P < 0.05, FWE corrected) (Fig. 1; Additional file 1: Table S3). Moreover, the TN group demonstrated higher RD in the bilateral superior corona radiata, bilateral anterior corona radiata, body of corpus callosum, splenium of corpus callosum, left cingulum, left superior fronto-occipital fasciculus, bilateral anterior limb of internal capsule, bilateral posterior limb of internal capsule, bilateral external capsule, left retrolenticular portion, left fornix cerebri, pontine crossing tract, corticospinal tract and left cerebral peduncle (P < 0.05, FWE corrected) (Additional file 1: Figure S1, Table S3). However, no significant difference was found in MD and AD between the TN group and control group.

In the TN patients, negative correlations were observed between disease duration and the FA values of left anterior corona radiata (r = 0.211, P = 0.012), genu of corpus callosum (r = 0.166, P = 0.028), left external capsule (r = 0.190, P = 0.018), left cerebral peduncle (r = 0.192, P = 0.017), and between VAS and the FA values of left anterior corona radiata (r = 0.221, P = 0.010), left external capsule (r = 0.218, P = 0.011), left cerebral peduncle (r = 0.168, P = 0.027) (Fig. 2). Positive correlations were observed for disease duration and the RD values of left anterior corona radiata (r = 0.190, P = 0.018), right external capsule (r = 0.170, P = 0.026), left fornix cerebri (r = 0.168, P = 0.027), left cerebral peduncle (r = 0.156, P = 0.034), and for VAS and the RD values of left anterior corona radiata (r = 0.174, P = 0.025), left external capsule (r = 0.191, P = 0.018) (Additional file 1: Figure S2). However, once Bonferroni corrections were applied, these correlations were not statistically significant.

Compared with voxel-based analysis (VBA), the TBSS method is applied more and more popularly to reveal microstructural alterations of white matter fibers between groups [24‐27]. VBA has two severe limitations about different subjects’ alignment of FA images and the self-willed choice of smoothing kernels without any principle or standard [28]. TBSS solve these issues using carefully tuned non-linear registration, and then projecting onto the “mean FA skeleton”(an alignment-invariant tract representation). Moreover, TBSS doesn’t need a smoothing process [22]. Therefore, TBSS can avoid the two limitations and provides us with more diffusion metrics [29]. In many disease studies, TBSS has a wide application to research microstructural white matter alterations, such as adiposity, parkinsonism, Alzheimer’s disease (AD), genetic disease, type2 diabetes mellitus and so on [30‐35]. In this study, we have been showed well results of the diffusion metrics (FA, MD, AD and RD) in TBSS methods between groups.

The primary diffusion metrics (FA and MD) reflect overall white matter health, organization and maturation [36]. In addition, AD reflects axon integrity and RD reflects myelin sheath integrity [31]. Both AD and RD are of great significance in understanding the underlying physiological mechanism [26]. Decreased FA values maybe based on predominantly increase of RD or both RD and AD [30]. The study of DeSouza et al. showed the right-sided TN patients had significantly decreased FA and increased RD, MD, and AD in the brain white matter including the corpus callosum, posterior corona radiata, cingulum, and superior longitudinal fasciculus. Moreover, MD and RD changes of brain white matter in TN patients maybe have relation to central nervous system plasticity, neuroinflammation and edema [18]. In our study, we found the similar results that reduced FA and elevated RD in the corona radiata (mainly concentrating on bilateral superior corona radiate and anterior corona radiata), corpus callosum and left cingulum. Additionally, we found reduced FA and elevated RD in the fronto-occipital fasciculus, internal capsule, external capsule, fornix cerebri and cerebral peduncle in the TN patients. Moreover, altered FA and RD was mainly located in white matter of left hemisphere. However, the differences of MD and AD were not statistically significant. As we all know, TN is involved in trigeminal nerve functional disorders, but other theories of central nervous system pathology is not clear [18, 37]. Due to peripheral trigeminal nerve injury, central nervous system plasticity will most probably occur [18, 38, 39], including fiber organization changes, astrocyte morphology and angiogenesis [18, 40‐42]. In our study, compared with healthy controls, TN patients showed decreased FA. The decreased FA may correspond to less fiber organization, such as more axonal sprouting/branching, larger axons, or more crossing fibers [18, 42]. Besides, neuropathic pain is usually related to chronic painful influence on central nervous system. This leads to central sensitization, a process involving demyelination and neuroinflammation processes [18]. The decreased FA is presumably caused by a significant increase of RD, inferring that demyelination and neuroinflammation processes may lead to the impairment of white matter integrity in the TN patients. The RD abnormalities maybe result from these mechanisms in the white matter of TN patients [8].

Many studies have reported cortical and subcortical gray matter impairment in the cognitive-affective, sensory, modulation of pain, attention and motor regions of TN patients [43‐45]. In anatomy and/or function respect, these brain areas are connected [13, 46‐48]. In our study, we reported decreased FA and increased RD in the corona radiata, corpus callosum, cingulum, fronto-occipital fasciculus, internal capsule, external capsule, fornix cerebri and cerebral peduncle in the TN patients. These fiber connection of brain regions is related to rapid transmission of pain, attention and motor function [49], and maybe lead to the unique sensory symptoms of TN. Our finding also revealed that altered FA and RD was mainly located in white matter of left hemisphere, suggesting contralateral white matter lesions of TN patients.

Correlation analyses found negative correlations between the disease duration and the FA values of left anterior corona radiata, genu of corpus callosum, left external capsule, left cerebral peduncle, indicating the white matter impairment is more and more severe as the disease progressed. With impairing progression of left anterior corona radiata, left external capsule and left cerebral peduncle, painful sensation is more serious. What is more, we infer these regions are probably related to transmission of pain [50]. The RD and the disease duration also reveal positive correlations in the regions of left anterior corona radiata, right external capsule, left fornix cerebri, left cerebral peduncle, demonstrating the white matter demyelination and neuroinflammation of these regions is aggravated in the disease progression. And as demyelination and neuroinflammation of left anterior corona radiate and left external capsule progresses, painful sensation is also more serious. Regrettably, the results of these correlation analyses had not been able to withstand multiple comparison correction.

Several limitations should be considered when interpreting our results. First, the sample size of the present study was not much, which might cause correlation analyses to fail to withstand multiple comparison correction. Second, non-isotropic voxels were used for DTI data acquisition in this study. In terms of the tensor evaluation, isotropic voxels are more accurate than non-isotropic voxels. Finally, all patients in our study were on medications for TN pain and the anticonvulsant carbamazepine is the most common. The influences of antiepileptics on brain structure are not clear. Future studies are needed to collect more sample sizes, adopt more optimized DTI parameters and avoid the effects of drugs.