Fingolimod‐improved axonal and myelin integrity of white matter tracts associated with multiple sclerosis‐related functional impairments

Summary

Aims

Fingolimod hydrochloride is an effective immunomodulatory drug in improving relapsing‐remitting multiple sclerosis (RRMS). However, data on the neuroradiologic effects on white matter (WM) have not been demonstrated. In this study, we aimed elucidating the impact of 1‐year fingolimod treatment on WM integrity in patients with RRMS.

Methods

Diffusion tensor imaging (DTI) was applied to assess axonal and myelin integrity in specific WM tracts of patients with RRMS prior to and 1 year postfingolimod treatment (n = 30). The fractional anisotropy (FA), radial diffusivity (RD), axial diffusivity, and mean diffusivity were analyzed using tract‐based spatial statistics on specific regions of interest associated with impaired Expanded Disability Status Scale functional scores before treatment.

Results

In patients with impaired pyramidal function at baseline (average score 2.3 ± 0.2, n = 25), fingolimod induced a significant increase in FA (P = 0.002) and decrease in RD (P = 0.03) in the corticospinal tract. In patients with impaired cerebellar function at baseline (average score 2.0 ± 0.1, n = 19), significant increases in FA and decreases in RD were observed in the superior (P = 0.02, P = 0.01, respectively) and inferior (P = 0.03, P = 0.05, respectively) cerebellar peduncles.

Conclusion

The observed results suggest increased microstructural integrity and decreased demyelination of damaged WM tracts and support the possible direct mechanism of fingolimod action.

1. INTRODUCTION

A majority of available treatments for multiple sclerosis (MS) act as immunomodulators, inhibiting the inflammatory process behind the disease without direct involvement of the central nervous system (CNS). Fingolimod hydrochloride (Gilenya), a recently approved orally administered drug, has shown promising results with an approximate 50% reduction in relapse rates.1 One of its mechanisms of action is due to its properties as a sphingosine‐1‐phosphate (S1P) receptor modulator, inhibiting lymphocyte migration from lymph nodes into circulation2 and preventing its eventual breach of the blood‐brain barrier. Additionally, fingolimod is thought to act through the arachidonic acid pathway, inhibiting the function of CD8+ cytotoxic T cells directly.3 However, findings also show a direct effect on the CNS by modulation of S1P receptors on most neural cell lineages, suggesting a direct anti‐inflammatory and neuroprotective effect.2, 4, 5, 6, 7, 8, 9

In the current literature, there are few studies investigating the anatomical changes to the white matter (WM) of the brain during fingolimod therapy. Kappos et al10 have shown the efficacy of the drug during clinical trials in halting the onset of new lesions seen on magnetic resonance (MR) imaging. However, while anatomical changes in WM develop in MS, they are more difficult to detect with conventional MR modalities. Diffusion tensor imaging (DTI) is a method used to assess axonal and myelin integrity in specific WM tracts by measuring water diffusivity in 3D space. The 3 eigenvectors obtained by DTI can be used to calculate several biometrics such as fractional anisotropy (FA), radial diffusivity (RD), axial diffusivity (AD), and mean diffusivity (MD).11 Although the conclusions regarding the utility of AD in describing neurologic integrity or pathology may be directly proportional to axonal damage,12 FA values describing the degree of isotropic diffusion have been proposed to be directly proportional to axonal integrity,13 while RD values describing the degree of orthogonal diffusion have been proposed to be inversely proportional to the degree of myelination.14

Until now, one study demonstrated effects of fingolimod in experimental autoimmune encephalitis mouse models using DTI of spinal cord WM. The study showed a significantly increased AD in the treated group compared to the untreated controls, as well as a strong correlation between increasing signs of clinical impairment and decreased AD and increased RD, suggesting decreased axonal and myelin integrity, respectively.15 In a study by Senda et al16 based on 4 patients with relapsing‐remitting multiple sclerosis (RRMS), they reported no improvements by DTI metrics and progressively increasing MD after 4 months of fingolimod treatment.

In this study, we have used DTI metrics to evaluate regional changes in WM and axonal integrity of functionally impaired WM tracts in patients with RRMS under fingolimod treatment. The specificity of fingolimod‐induced changes was tested by comparison with DTI metrics of interferon beta‐1b (Betaferon)‐treated patients.

2. MATERIALS AND METHODS

2.1. Subjects

Patients with relapsing‐remitting MS were recruited before fingolimod or interferon beta‐1b treatment from the Multiple Sclerosis Center at Tel‐HaShomer Medical Center, Israel, according to the following criteria: (i) age of disease onset 18‐60 years; (ii) diagnosis of RRMS according to the 2010 McDonald Criteria17; (iii) relapsing‐remitting disease time course; (iv) no treatment with corticosteroids at least 3 months prior to study date. Patients in the midst of significant cognitive decline, diagnosed with primary or secondary progressive MS at any point during the study, or experiencing an acute febrile illness at the time of scans, were excluded from the study. All subjects gave written informed consent prior to participation. The study was approved by the Sheba IRB committee.

All patients were subjected to MRI/DTI acquisition at baseline and after 1 year of fingolimod hydrochloride (Gilenya, 0.5 mg PO once daily) or interferon beta‐1b (Betaferon, 250 μg/mL SC every other day).

2.2. MRI acquisition protocol

All subjects underwent the brain MR imaging protocol, performed on 3.0‐T MR scanner (Signa, GE) using a high‐resolution 8‐channel head coil. Data were obtained using the following sequences: (i) 3D‐FSPGR (1 × 1 × 1 mm voxel, TE = 2 ms, TR = 6 ms), (ii) T2‐FSE (slice thickness 2.6 mm, TE = 102 ms, TR = 3500 ms), (iii) FLAIR (slice thickness 2.6 mm, TE = 122 ms, TR = 9502 ms, TI = 2375 ms), (iv) 2D‐FSPGR with contrast GD‐DTPA (slice thickness 2.6 mm, TE = 2 ms, TR = 250 ms). Axial DTI data were acquired along 31 independent directions using a single‐shot echo‐planar imaging sequence (TE = 76 ms, TR = 14 000 ms, b = 1000 s/mm², FOV 256 × 256 mm, matrix 128 × 128). Two additional images without diffusion weighting (b = 0 s/mm²) were acquired. Axial images were acquired by contiguous slices with 2.6 mm thickness. The slices were positioned to run parallel to a line that joins the anterior commissure‐posterior commissure plane. During image acquisition, the same image resolution and the same localizer were used for T2‐FLAIR and DTI series, to obtain similar axial slice positions. MS lesions were identified on T1, T2, FLAIR, and gadolinium (GD)‐enhancing modalities, measuring lesion number and volume using semiautomated segmentation software as described.18 Paired t test analysis was performed comparing lesion number and volume on GD‐enhancing, T1, T2, and FLAIR images between baseline and 1 year on fingolimod treatment.

2.3. Image analysis and postprocessing

All image manipulation tools used in this study are part of the FMRIB Software Library (FSL, http://www.fmrib.ox.ac.uk/fsl). In brief, DICOM files of the DTI acquisitions were converted into a single multivolume NIFTI file (Neuroimaging Informatics Technology Initiative file) and transferred to a Linux‐based workstation. Analysis with FMRIB's Diffusion Toolbox (FDT) was initiated with eddy‐current correction followed by automatic brain extraction, which then allowed fitting a diffusion tensor model for each voxel of the diffusion images. Subsequently, FA, AD, RD, and MD values were generated.

Further analyses between subjects were carried out using tract‐based spatial statistics (TBSS).19, 20 First, individual FA images were transformed into 1 × 1 × 1 mm³ MNI‐152 common space included in the FSL template library by means of nonlinear registration. After this, the transformed FA images were averaged to generate a mean FA image which was subsequently skeletonized, representing a common tract for all subjects. To prevent inclusion of non‐skeleton voxels in further analysis, each subject's aligned FA maps were mapped onto the “mean FA skeleton” using a lower threshold of FA of 0.2 to exclude gray matter voxels. The approach of carefully tuned nonlinear registration, followed by a creation of a mean FA skeleton, intends to face the cross‐subject spatial variability effect.21 Individual registration and vectors obtained in the aforementioned process were also used to obtain the mean, radial, and AD data.

2.4. Selecting functionally impaired tracts

Our study aims to observe the effects fingolimod has on WM tracts that were already damaged clinically and anatomically. To that end, region‐of‐interest (ROI) analysis was performed to evaluate the magnitude of diffusivity changes of each subject at baseline and following 1 year of treatment within predetermined regions. The FA, MD, RD, and AD values for each ROI were calculated by splitting the 4D postthreshold skeletonized volume of each subject's WM tracts at baseline and after 1 year of treatment and repackaging them into group 4D files, where WM tracts were used as filters according to the ICBM‐DTI‐81 WM labels atlas, which includes labels created by hand segmentation of a standard‐space average of diffusion MRI tensor maps from 81 healthy subjects.22 Regions from the atlas were selected by focusing diffusivity analysis on subjects grouped by functional impairments as determined by baseline functional scores of the Expanded Disability Status Scale (EDSS). Specifically, ROI analysis of the corticospinal tract as demarcated by the ICBM atlas was performed on patients with impaired baseline pyramidal function. Furthermore, ROI analysis of the superior, middle, and inferior cerebellar peduncles, containing bundles of the dentato‐rubro‐thalamic tract, ponto‐cerebellar tract, and olivo‐cerebellar tracts, respectively, was performed on patients with impaired baseline cerebellar functions.

2.5. Selecting impaired side

In addition, for each patient, we analyzed the most affected side at baseline in regard to FA, taking into consideration that a low FA value suggests impairment of WM and axonal integrity. The least affected side at baseline by FA was analyzed as a control measurement for specific unilateral changes of impaired sites.

2.6. Statistical analysis

Paired t tests were used to compare each patient's baseline DTI parameters to its own data at 1 year posttreatment, thus isolating the effect of treatment over time and excluding possible confounding factors. Spearman correlation was used to analyze the correlation between EDSS and DTI measurements. We considered a P value <0.05 after false discovery rate (FDR) correction to be statistically significant.

3. RESULTS

Thirty patients with RRMS, 20 females, mean ± SE age 37.4 ± 1.9 years, disease duration (DD) 12.7 ± 1.2 years, baseline EDSS 3.0 ± 0.3, and annualized relapse rate (ARR) 0.88 ± 0.1 relapses per year were included in the study before fingolimod treatment. After 1 year of fingolimod treatment, the mean ARR halved to 0.43 ± 0.2 (P = 0.02), EDSS score remained stable (2.9 ± 0.3), and no significant changes were found in any of the MRI sequences. The demographic and clinical information of the patients is presented in Table 1.

Table 1.

Demographic and clinical information of the fingolimod group

Number of patients (n)	30
Gender F(M)	20 (10)
Age, mean ± SE (y)	37.4 ± 1.9
Age at MS onset, mean ± SE (y)	24.8 ± 1.6
Disease duration, mean ± SE (y)	12.7 ± 1.2
	Baseline	1 y posttreatment
Annualized RR, mean ± SE (relapses/y)	0.88 ± 0.1	0.43 ± 0.2a
EDSS score, mean ± SE	3.1 ± 0.3	2.9 ± 0.3
# of GD‐enhancing lesions, mean ± SE	1.4 ± 0.5	0.36 ± 0.12
# of T1 lesions, mean ± SE	23 ± 3.4	25.3 ± 3.2
# of T2 lesions, mean ± SE	32 ± 3.7	30.1 ± 3.3
# of FLAIR lesions, mean ± SE	30.2 ± 3.8	30.8 ± 3.5
GD lesion V, mean ± SE (mm3)	0.2 ± 0.07	0.1 ± 0.06
T1 lesion V, mean ± SE (mm3)	6.6 ± 1.5	8.4 ± 2.0
T2 lesion V, mean ± SE (mm3)	7.5 ± 1.5	7.3 ± 1.5
FLAIR lesion V, mean ± SE (mm3)	9.1 ± 1.8	9.7 ± 2.1

RR, relapse rate; GD, gadolinium; V, volume; EDSS, Expanded Disability Status Scale; MS, multiple sclerosis.

^a P < 0.05.

3.1. DTI analysis

The whole brain FA negatively and RD and MD positively correlated with EDSS at baseline (r = −0.53, P = 0.003; r = 0.49, P = 0.007; r = 0.42, P = 0.02, respectively) (Figure 1). The whole brain and voxelwise DTI analyses showed no significant changes in DTI metrics after 1 year of fingolimod treatment as compared to baseline, and no DTI parameters of ROI passed the false discovery rate criteria associated with multiple parameter testing.

Figure 1.

Correlations between Expanded Disability Status Scale scores and whole brain fractional anisotropy (A), radial diffusivity (B), and mean diffusivity (C) diffusion tensor imaging metrics prior to fingolimod therapy start

3.2. Corticospinal tract

After 1 year of fingolimod treatment, patients with impaired pyramidal function at baseline (average score 2.3 ± 0.2, range 1.0‐5.0, n = 25) had a significant increase in FA (0.602 ± 0.009 vs 0.625 ± 0.010, P = 0.003, Figure 2A) and a decrease in RD (4.25E‐04 ± 0.10E‐04 vs 4.06E‐04 ± 0.11E‐04, P = 0.03, Figure 2B) in the corticospinal tract. The statistical maps in Figure 3A,B show voxels exhibiting significantly increased FA after 1 year of treatment as compared to baseline in a single representative patient and overall, respectively.

Figure 2.

Significant increase in fractional anisotropy (A) and decrease in radial diffusivity (B) after 1 y of fingolimod therapy in the corticospinal tracts of patients with impaired pyramidal function at baseline are demonstrated

Figure 3.

Statistical voxel map showing significantly different changes in fractional anisotropy (FA) between baseline and 1 y of fingolimod therapy over the mean FA skeleton. The A shows diffusion tensor imaging (DTI) images of a single representative patient at baseline, after 1 y of fingolimod treatment, and the differences between 1 y and baseline obtained by TBSS analysis of the corticospinal tracts. The green color represents the mean FA skeleton and yellow‐red the statistically significant differences in FA between 1 y of treatment and baseline. The B represents the statistical analysis of all patients with impaired pyramidal function at baseline. The green color represents the mean FA skeleton, and the yellow‐red color represents overall statistically significant differences in FA of the corticospinal tracts between 1 y of fingolimod treatment and baseline. The C shows DTI images of a single representative patient at baseline, after 1 y of fingolimod treatment, and the differences in FA between 1 y and baseline obtained by TBSS analysis of the superior, middle, and inferior cerebellar peduncles. The green color represents the mean FA skeleton and yellow‐red the statistically significant differences in FA between 1 y of treatment and baseline. The D represents the statistical analysis of all patients with impaired cerebellar function at baseline. The green color represents the mean FA skeleton, and the yellow‐red color represents overall statistically significant differences in FA of the superior, middle, and inferior cerebellar peduncles between 1 y of fingolimod treatment and baseline. The MNI‐152 standard‐space T1‐weighted MRI image was used in B and D as background for the results

3.3. Cerebellar tracts

Patients with impaired cerebellar function at baseline (average score 2.0 ± 0.2, range 1.0‐3.0, n = 19) demonstrated significant increases in FA in the superior (0.57 ± 0.01 vs 0.59 ± 0.01, P = 0.02) and inferior (0.53 ± 0.01 vs 0.54 ± 0.01, P = 0.03) cerebellar peduncles after 1 year of fingolimod treatment (Figure 4A). One year of fingolimod treatment also led to a decrease in RD in the superior (5.75E‐04 ± 0.14E‐04 vs 5.56E‐04 ± 0.14E‐04, P = 0.01) and inferior (4.88E‐04 ± 9.33E‐06 vs 4.72E‐04 ± 7.45E‐06, P = 0.05) cerebellar peduncles (Figure 4B). Figure 3C,D show the voxels in the cerebellar peduncles with significant changes in FA in a single representative patient and overall, respectively.

Figure 4.

Significant increase in fractional anisotropy (A) and decrease in radial diffusivity (B) after 1 y of fingolimod therapy in the superior and inferior cerebellar peduncles of patients with impaired cerebellar function at baseline are demonstrated

3.4. Verification of unilateral changes and treatment specificity

In the control experiments testing the effect of fingolimod on the least affected side at baseline by FA, we saw no significant changes in DTI parameters in neither the corticospinal nor the cerebellar tracts in patients with impaired function at baseline.

To prove specific fingolimod‐associated effects on the brain, DTI analysis was performed for 20 patients with RRMS, 11 females, before and after 1 year of interferon beta‐1b treatment. The patients’ mean ± SE age was 42.1 ± 2.2 years, age at MS onset 34.2 ± 2.4 years, disease duration 7.8 ± 1.5 years, and baseline EDSS 3.3 ± 0.4. After 1 year of interferon beta‐1b treatment, EDSS score remained stable. No changes were observed in the corticospinal tracts of treated patients with affected pyramidal function (average score 2.2 ± 0.2, range 1.0‐4.0, n = 20). There was only a significant decrease in mean RD and MD in the superior cerebellar peduncle (P = 0.01 and P = 0.05, respectively) of patients with cerebellar impairment (average score 1.9 ± 0.2, range 1.0‐3.0, n = 11). The detailed results of the patients’ DTI metric changes after interferon beta‐1b treatment are presented in Table 2.

Table 2.

Diffusion tensor imaging results of the interferon beta‐1b (Betaferon) group

	CST	Cerebellar peduncles
		Inferior	Middle	Superior
FA baseline mean	0.587	0.532	0.535	0.600
FA 1 y BF mean	0.624	0.544	0.550	0.617
RD baseline mean (×10−3)	0.462	0.503	0.456	0.566
RD 1 y BF mean (×10−3)	0.399	0.473	0.419	0.512a
MD baseline mean (×10−3)	0.720	0.746	0.689	0.905
MD 1 y BF mean (×10−3)	0.655	0.706	0.644	0.839a
AD baseline mean (×10−2)	0.124	0.123	0.116	0.158
AD 1 y BF mean (×10−2)	0.117	0.117	0.109	0.149

CST, corticospinal tract; FA, fractional anisotropy; BF, Betaferon; RD, radial diffusivity; MD, mean diffusivity; AD, axial diffusivity.

^a P < 0.05.

4. DISCUSSION

In this study, using TBSS analysis, we for the first time demonstrated changes in DTI parameters suggesting improvement in the state of the WM in impaired tracts, possibly due to a correction of brain axonal and myelin integrity or a decrease in inflammation and edema of the regions studied, in RRMS patients under fingolimod treatment. In contrast to the previously published study by Senda et al16 where no therapeutic effect by DTI measures was found after 4 months of fingolimod treatment, we investigated a longer therapeutic interval and used more restricted analytical criteria taking into consideration the patients’ pretreatment functional impairments.

Only by applying this tailored analysis of MS‐associated clinical and anatomical impairments at baseline, we identified tracts with increased impairment and measured their anatomical changes after 1 year of treatment. Importantly, the observed DTI parameter improvements were not observed in the same patients’ less impaired contralateral tracts. Additionally, the minor changes observed in functionally impaired tracts of interferon beta‐1b‐treated patients compared to the fingolimod group confirmed specificity of fingolimod‐induced DTI changes and raises question about fingolimod's direct or indirect mechanism of action.

There are several proposed mechanisms contributing to fingolimod's therapeutic effect; its phosphorylated form binds to 4 out of 5 sphingosine‐1‐phosphate (S1P) receptor subtypes, acting as a functional antagonist by permanently internalizing the receptors and down‐regulating their expression, resulting in the inhibition of lymphocyte migration from the lymph nodes into circulation.2 This sequestration indirectly leads to a neuroprotective effect by reducing the inflammatory response in the CNS and enables the endogenous mechanisms of neuronal repair to perform their tasks.

In conjunction, fingolimod can also cross the blood‐brain barrier and have direct CNS effects by modulating S1P receptors on most neural cell lineages, such as oligodendrocytes, microglia, neurons, and astrocytes.2, 4, 6, 8 It stimulates oligodendrocyte survival and differentiation, while protecting the progenitors from cell death.7 In rat models of MS, fingolimod has been shown to have important neuroprotective functions as well; it prevented decreases in axonal density and restored neuronal and motor function.2 Other studies demonstrated a neuroregenerative effect as well as a neuroprotective effect on synapses of murine MS models after fingolimod treatment.5 The drug elicits a neuronal gene response, modulating neurite growth, and axonal regeneration.9

In our study, interferon beta‐1b, well known to suppress inflammatory processes in the periphery, was unable to make significant changes in impaired WM tracts. Accordingly, we suggest that the direct mechanism of fingolimod action through S1P receptor modulation in the CNS is the major contributor to the overall effects observed. However, we cannot exclude the indirect mechanism as a contributor to local WM regeneration by decreasing inflammation in peripheral and brain microenvironments. Given that our study only observed the first year of starting therapy, the “pseudoatrophy” phenomenon, where fluid shifts due to resolution of inflammation and edema, may also be considered as a possible mechanism for the changes in DTI metrics observed coinciding with treatment onset.23 In this case, given the overall anti‐inflammatory effects of the therapy, whole brain changes in DTI would be expected; however, our study demonstrated only regional changes motivating us to suggest an alternative hypothesis. The exact mechanism of fingolimod's direct CNS effects is out of our study's scope; the recently published review by Hunter et al24 provides an in‐depth look at the possible mechanisms involved, such as neuronal survival and remyelination, astrogliosis inhibition, and reduced leakiness of the blood‐brain barrier.

While clinical and MRI parameters failed to show any significant changes after 1 year of fingolimod treatment, the DTI analysis was sensitive enough to show focal improvements of microstructural changes in WM. Earlier studies produced conflicting results investigating the association between DTI metrics and clinical disability.25, 26, 27, 28, 29, 30, 31, 32 However, more recent studies using TBSS showed significant correlations between clinical disability and FA and RD values.33, 34, 35 It has been shown that decreases in FA and relative anisotropy of the pyramidal tracts are significantly correlated with decreased pyramidal EDSS functional scores as well as decreased overall EDSS scores.32, 35 Similarly, lower FA in cerebellar WM is significantly correlated with increased disability by EDSS.36 These correlations, which corroborated with our data, motivated us to analyze patients with similar clinical disabilities associated with known WM tracts and study the microstructural changes in those tracts during treatment.

Our study was not without limitation; the association between specific EDSS functions and corresponding ROIs were tested as chosen from the atlas and did not include all the fibers of the WM tracts. The relatively small sample size was compensated by applying a paired t test analysis, allowing us to analyze small changes in DTI parameters after 1 year of treatment and to reduce patients’ individual bias.

To conclude, this study is the first to demonstrate fingolimod's protective effect on WM tracts in humans. After 1 year of treatment, fingolimod induced DTI changes in WM tracts associated with specific functional EDSS impairments that suggest better protection of axonal and myelin integrity and support a potential congruent CNS mechanism.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Gurevich M, Waknin R, Stone E, Achiron A. Fingolimod‐improved axonal and myelin integrity of white matter tracts associated with multiple sclerosis‐related functional impairments. CNS Neurosci Ther. 2018;24:412–419. 10.1111/cns.12796