Application of diffusion tensor imaging in detecting microstructural alteration of the trigeminal nerve root in patients with classical trigeminal neuralgia without neurovascular compression

to detect and of surgical intervention of the


Background
Trigeminal neuralgia (TN) is a condition that can cause intense even unbearable orofacial pain localized to the sensory supply areas of the trigeminal nerve, which can dramatically interfere with the normal activities of daily living. [1] The pathological mechanism of classical TN has been recognized by the theory that it is a kind of disease results from neurovascular compression (NVC) of the trigeminal nerve root. [9] Microvascular decompression (MVD) was exactly designed based on this etiological explanation, and has been accepted as the most e cient treatment of classical TN currently. [22] However, strong evidence suggests that there are also a cohort patients of TN without NVC (woNVC) discovered by preoperative morphological magnetic resonance imaging (MRI) and proved by surgical nding. [8; 14; 24] Traditional morphological MRI has been shown to be effective for depicting the ne trigeminal neurovascular anatomy, of paramount importance for determining NVC. [9; 23; 28] However, in several previous researches, investigators have found that there were some cases underwent MVD in absence of NVC yet showed positive NVC in preoperative image. [11; 14] In contrast, some clinical articles described that a signi cant number of TN patients have no NVC at all, and internal neurolysis (IN) has been reported as an emerging e cient surgical method for TN woNVC. [12; 31] Recently, several studies have revealed the existence of microstructural changes in TN patients with NVCS (wNVC) using diffusion tensor imaging (DTI), which showed various alteration of the values diffusivity metrics: fractional anisotropy (FA), apparent diffusion coeffcient (ADC), radial diffusivity (RD), or/and axial diffusivity (AD). [10; 18; 21] These previous results indicated a water diffusion change in the microenvironment in the affected trigeminal nerves with NVCS, and the diffusivity metrics of DTI may be potential tools in evaluating the trigeminal nerves and in estimating the prognosis of TN. Additionally, one study used DTI and its diffusivity metrics to investigate pain relief in patients with TN 4 year after MVD, and suggested that FA and ADC can provide prognostic information. [13] Another study revealed that the preoperative diffusivity metrics of DTI have high predictive value for the prognosis of patients who underwent radiofrequency lesioning (RFL). [16] Also recent study reported that increased trigeminal nerves volume and FA, and a decreased ADC, AD, and RD represent the pathophysiology of TN, and decreased trigeminal nerves volume, FA, and AD compared with the normal side after a one-year follow-up of partial rhizotomy (PR) of sensory root of trigeminal nerve. [3] These studies provided us information and concepts that DTI and its diffusivity metrics may have clinical signi cance for microstructural change and prognosis evaluation.
We systematically evaluated FA and ADC values in a patient cohort that underwent IN for TN woNVC and looked for possible interactions with the clinical and anatomical parameters. We hypothesized that the microstructural status of the trigeminal nerve of the affected side is signi cantly different compared to the unaffected side in patients and controls, and these microstructural abnormality could be improved in the diffusion of the trigeminal root after IN.

Study Population
A total of 21 patients with classical TN (CTN) and 20 healthy controls were retrospectively included in our study. All patients were clinically examined by an experienced neurosurgeon (Dr. X.-F.J.) and met the criteria of the International Headache Society for TN. [1] CTN is de ned as idiopathic, episodic, lancinating pain that lasted seconds, with pain-free episodes between attacks. [6] Only patients without previous surgical treatment for TN were included in this study. Patients with secondary factors may resulting in TN, such as cranial tumors, or vertebrobasilar dolichoectasia resulting in brainstem compression, were excluded from this study. Patients with a diagnosis of multiple sclerosis (MS) or having signs or symptoms of MS were also excluded. Patients with bilateral facial pain were also excluded. Ipsilateral and contralateral nerves in these controls were based on the laterality of TN in matched patients. For example, controls matched to right TN patients will have ipsilateral nerve on the right.

MRI Examination And Analysis
MRI was performed on a 3.0-T MRI scanner (Signa HDx; 3T GE Healthcare) using a standardized protocol. [21] Images were obtained using an 8-channel head coil with sensitivity encoding parallel processing capability, with the application of 64 non-collinear directions of diffusion gradients. The sequences and their parameters are summarized in Table 1. Morphological MRI Data Analysis Three-dimensional brain volume T1-weighted postcontrast sequence (3D-T1 BRAVO), three-dimensional fast imaging employing steady-state acquisition (3D-FIESTA) and three-dimensional Time-of-Flight magnetic resonance angiography (3D-TOF MRA) images were analyzed collaboratively by 1 experienced neuroradiologists (Dr. J.Q.) and 1 experienced neurosurgeon (Dr. M.W.), who were blinded to the clinical data including the side of nerve affection. Multiplanar reconstructions were used to assess the trigeminal nerves on both sides separately using a step-by-step approach to determine the presence or absence of the NVC. Contact was declared if no layer of the CSF was discernible between the nerve and the respective vessel on the high-resolution 3D-FIESTA sequences.

Analysis Of The DTI Data Sets
Postprocessing and analysis of the DTI data sets were performed in a blind fashion (for clinical data and intra-operative ndings) using the GE Healthcare AW VolumeShare™ 5 software on a commercially available workstation (AW 4.6; GE Healthcare). Accurate measurement was con rmed by the same two physicians with profound experience using this protocol. In order to limit potential partial volume effects due to the small size of the whole segment of trigeminal nerve root (TR) in posterior fossa, special attention was paid to the region of interest (ROI)-placement procedure: two physicians reviewed the ROIs on the FA and ADC colormaps and veri ed the DTI with the 3D-FIESTA anatomical images in the axial plane for a cross-reference, and recon rm the ROIs by viewing the fusion imaging of FA and ADC colormaps emerged to corresponding 3D-BRAVO slice. An example of ROI placement is illustrated in Fig. 1a-d. For statistical analysis, we used the mean, kurtosis and skewness values of the two observers. The kurtosis of a normal distribution is equal to 3, and a normal distribution will have a skewness of 0, these two factors could provide insights into the shape of the distribution. The intraclass correlation coe cient of inter-observer reliability for the average FA and ADC were less than 5%.

Internal Neurolysis
All score, which were considered comprehensively by adding the total of both scores ( Table 2). The results of follow-up at 1 years after surgery were considered as the nal outcome in this study.

Statistical Analysis
The differences in mean values, as well as the kurtosis and skewness values of each of the diffusivity metrics (FA and ADC) was compared in the affected and unaffected sides of patients 1 years after surgery. These results were also compared with those of the affected and unaffected sides of patients preoperative and compared with controls using a paired-sample two-tailed t test. Correlations between increase of FA, decrease of ADC in affected sides and surgical outcome in BNI total scores were analyzed respectively. Data was presented as mean ± standard error of the mean (SEM) or standard deviation (SD). The Pearson correlation coe cient was calculated, and a linear regression curve was plotted using GraphPad Prism® statistical analysis software. A p-value of < 0.05 was considered signi cant.

Clinical Patient Characteristics
A total of 21 patients were enrolled (10 men and 11 women) ( Table 3). The mean duration of TN symptoms was 63.6 ± 8.2 months (range 18-138 months). All patients had TN despite being on medication (carbamazepine or oxcarbazepine) for at least 6 months before surgery was scheduled. Among these, the left side was affected in 8 patients (38.1%), and the right side was affected in 13 patients (61.9%). 15 patients presented with an isolated affection in a single branch of the trigeminal nerve, which included the second branch (V2) in 9 patients (42.9%) and the third branch (V3) in 6 patients (28.6%). No patient exhibited involvement of the rst trigeminal branch (V1) alone. 5 patients (23.8%) presented with a combined affection of V2 plus V3, and 1 patient (4.8%) presented with a combination of V1 and V2. No patients, all 3 branches (V1-V3) were involved.  (Fig. 3a). Both of the kurtosis values of FA and ADC in the affected side were signi cantly higher (p < 0.05) than that of the unaffected side of patients and mean of both sides of the controls preoperatively.  (Fig. 2a). The mean FA in the affected side was signi cantly higher (p < 0.05) than the mean FA in the affected side of patients before surgery (Fig. 2b). One year after IN, The mean ADC in the affected side (3.49 ± 0.65 mm 2 /s, CI 3.19-3.78) was signi cantly higher (p < 0.05) than the mean ADC in the unaffected side of patients (2.67 ± 0.62 mm 2 /s, CI 2.38-2.95), but had become similar (p > 0.05) to the mean ADC in both sides of the controls (Fig. 3a). The mean ADC in the affected side was signi cantly lower (p < 0.05) than the mean ADC in the affected side of patients before surgery (Fig. 3b). The analysis of kurtosis and skewness of FA and ADC comparing affected side before IN and Affected side 1 year after IN respectively showed that, the kurtosis of both metrics became notably lower (p < 0.05) postoperatively.
After IN, the Pearson correlation coe cient did not show a strong correlation (r = 0.3928, p = 0.0202) between increase of FA in the affected side compared to the surgical outcome in BNI total score (Fig. 4a). However, the Pearson correlation coe cient showed a strong negative correlation (r = -0.5237, p = 0.0148) between decrease of ADC in the affected side compared to the surgical outcome in BNI total score (Fig. 4b).
Discussion DTI abnormalities in the trigeminal nerve root of TN  [2-4; 19], and utilization of 7.0 T MRI tractography also supported these opinions by a recent paper [25]. , and focused on the cisternal segment of trigeminal nerves on affected side, in symptoms of < 5 years cases were associated with decreased FA ,while in symptoms of ≧ 5 years cases were associated with increased FA post-SRS. [15] From the results of our study, not only the mean values of FA and ADC showed signi cant differences between affected and unaffected sides, affected and control's sides, but also the kurtosis of FA and ADC behaved signi cant higher differences between affected and unaffected sides, affected and control's sides. The former nding is consistent with the mainstream opinion of above studies, the latter may elucidate further understanding of TN1 woNVC. No matter the decrease of FA mean value or the increase of ADC mean value re ects the microstructural changes in the trigeminal nerve even with the absence of neurovascular compression. And the signi cantly higher kurtosis values in FA and ADC, which indicates great concentrated around the mean and had a little variance of microstructural texture of the involved nerves. The interesting results may inspire the discovery of mechanism of TN1 woNVC which is distinct from TN1 wNVC.

Postoperative DTI Changes Of TN And Underlying Mechanism
Few reports has been published on long-term investigation of DTI abnormalities after surgical treatment for TN. In Dr. Sindou's series of studies, all patients underwent MVD with a complete exploration of the TR and detection of direct neurovascular contact in all cases, after removal of the compression, the loss of FA remained, but ADC normalized in the affected nerves, suggesting improvement in the diffusion of the trigeminal root. [13] DeSouza et al. studied DTI parameters and brain gray matter (GM) analyses of TN patients who underwent GKRS (15 cases) or MVD (10 cases) and 14 controls, and found that the FA abnormality in the affected side resolved such that FA increased and was no longer signi cantly different from the unaffected side or controls in the effective treatment group, however, FA remained signi cantly lower in the affected side compared to the unaffected side and controls in the ineffective treatment group. Then the authors concluded that surgical treatment can effectively resolve pain by normalizing trigeminal root abnormalities. following MVD in short term. [32] In our study, a strong negative correlation was found between decrease of ADC in the affected side compared to the surgical outcome in BNI total score. The ADC maps has been acknowledged to used to characterize myelination and to detect abnormalities in the developing brain. [27] We hypothesize that nerve pathological alteration revealed by ADC values could be related to irreversible lesions in TR myelin. Little information is available about the long-term completeness of remyelination in the cranial nerves. Pathological myelination seems to be the most likely explanation for TN woNVC, even that the nerve roots appear entirely normal. In our study, the kurtosis values of FA and ADC became signi cantly lower (p < 0.05) postoperatively also demonstrate that recovery of the ultra little variation may contribute to pain relief.

Correlation between changes of diffusion metric values of the trigeminal root and outcome of IN
Diffusive abnormalities after IN, especially the decrease of ADC, con rmed a strong negative correlation with the 1 year surgical outcome in BNI total score in our study. As far as we know, no prior studies have reported the relevance between the diffusivity metrics and the prognosis of TN after IN. Our results attempted to provide potential noninvasive, brain-based biomarkers to support an optimal treatment selection for individual patients with TN woNVC.

Technical Limitations
In regard to limitations of the study, rstly, the small sample size of our cases is also one of the limitations in this study. Secondly, the follow up period may not be long enough for investigate the full view of alteration of diffusive metrics.

Conclusions
The DTI technique was used to investigate TR microstructure in patients with TN1 woNVC. The changes of diffusive property of TR could explain the e ciency of IN. The pretreatment DTI metrics could be an effective prognostic factor for providing potential noninvasive, brain-based biomarkers to support an optimal treatment selection for individual patients with TN woNVC.

Competing interest
The authors declare that they have no competing interests.

Availability of data and material
The data and material of this study will be shared with reviewers or editors if necessary.   Linear regression plot derived from the relationship between the changes in the fraction of anisotropy (FA) and the apparent diffusion coe cient (ADC), in affected nerves vs surgical outcome in BNI total score respectively. a the Pearson correlation coe cient did not show strong correlation between increase of FA in the affected side compared to the surgical outcome in BNI total score. b the Pearson correlation coe cient showed a strong negative correlation between decrease of ADC in the affected side compared to the surgical outcome in BNI total score. * statistical signi cance