1.

Introduction

Multiple sclerosis (MS) is a chronic inflammatory neurological disease that results in loss of both myelin and axons.¹ In particular, the corpus callosum is known to undergo considerable damage during the progression of MS.^2,3 Such damage could greatly reduce the functional communication between the hemispheres. Indeed, studies using functional magnetic resonance imaging (fMRI) have demonstrated disrupted inter-regional communication throughout a number of brain networks in MS patients.^4,5

Reduction in functional communication may, thus, be a quantifiable indicator of MS progression and response to treatments aimed at reducing demyelination, stimulating remyelination, and restoring brain function in MS. However, fMRI does not lend itself to a quick and easy bedside assessment of inter-regional communication, which is desirable for clinical application and for repeated assessments of treatment response, especially over a short period of observation.

Brain activity increases oxyhemoglobin concentration (HbO) and total hemoglobin concentration (HbT), and reduces deoxyhemoglobin concentration (HbR) due to increased perfusion.^6,7 Changes in these parameters are easily measurable using functional near-infrared spectroscopy (fNIRS), which is a noninvasive and portable method that monitors brain oxygenation by detecting the attenuation of light by HbO and HbR at two separate wavelengths.⁸

We and others have recently demonstrated the successful use of fNIRS in the detection of interhemispheric communication between the motor cortices of healthy control subjects by examining the coherence of brain signals in the frequency domain.^9,10 Coherence quantifies the similarity between signals originating from different brain regions (and, hence, the strength of the functional connection) by comparing the magnitude of specific frequency components.¹¹

In this study, we used fNIRS to determine the coherence between the left and right motor cortices during task execution and rest, in MS patients and healthy controls. We hypothesized that coherence would be reduced in MS patients compared to healthy participants due to reduced interhemispheric communication. In addition, MS patients recruit more brain centers during task activation.^12,13 This additional recruitment may also reduce coherence, so we measured during resting state as well as during task activation. If this hypothesis was proven, fNIRS measures of coherence could potentially be a noninvasive indicator of axonal damage and demyelination.

2.

Methods

3.

Results

ANOVAs revealed no effect of the task block (i.e., there was no difference between the six rest/tapping tasks) on the change in HbT or HbO concentration [$\mathrm{F}(1,93)=1.38$; $p=0.243$]. There was, however, an effect of subject group, with MS subjects exhibiting a smaller change in concentration of HbT [$F(1,93)=9.43$;$p=0.003$] and HbO [$F(1,93)=4.40$; $p=0.039$] during finger tapping, compared to controls (Fig. 1).

Fig. 1

Changes in HbO and HbT concentrations in the left motor cortex during finger tapping. Control subjects exhibit significantly greater changes in HbO and HbT concentrations compared to multiple sclerosis (MS) patients ($\text{mean}\pm \mathrm{SEM}$; $n=7$ controls, 8 patients; **$p<0.01$).

Figure 2(a) shows a representative HbT coherence map calculated from data collected during finger tapping in a healthy control subject, demonstrating high coherence in both the left and right motor cortices. Figure 2(b) shows a representative HbT coherence map during finger tapping for an MS patient, demonstrating reduced coherence relative to the control subject.

Fig. 2

HbT coherence maps during finger tapping for a representative (a) control subject and (b) MS patient. White arrows indicate the location of the source-detector pair that was selected as the seed signal.

Finger tapping resulted in a significant increase in HbT coherence in controls in both the ipsilateral, as well as the contralateral side [$F(1,22)=8.46$; $p=0.009$], but not in patients with MS. There was a significant difference in HbT coherence during finger tapping between patients and control subjects [$F(1,22)=8.23$; $p=0.009$]. Follow-up $t$ tests revealed that HbT coherence in the contralateral hemisphere was significantly reduced in MS patients compared to controls [$t(12)=-4.00$, $p=0.006$)] (Fig. 3). There is a significant difference in coherence between the ipsilateral and contralateral hemispheres relative to the seed in both controls and MS patients [$F(1,22)=12.45$; $p=0.002$] (Fig. 3). There were no interactions between the subject group and the hemisphere [$F(1,22)=2.08$; $p=0.16$]. Interestingly, there were no significant differences in resting-state coherence between MS and controls. There were also no significant differences in HbO coherence between groups or hemispheres, for finger tapping and the resting state.

Fig. 3

HbT coherence during resting state and finger tapping. Coherence on the contralateral side is significantly reduced in MS patients compared to controls during finger tapping ($\text{mean}\pm \mathrm{SEM}$; **$p<0.01$).

4.

Discussion

The fNIRS data show increased HbO and HbT concentrations for all participants in response to finger tapping. However, the magnitude of oxygenation changes in response to task execution was reduced in MS patients (i.e., less increase in HbT or HbO) compared to controls. Studies using blood oxygen level dependent MRI have shown that MS patients may have a larger volume of activation compared to controls for a given task,¹⁶ and more brain centers are recruited.^12,13 Since more brain centers are needed for a particular task, it is possible that an individual region has a reduced magnitude of activation, which would explain the smaller changes in HbT and HbO.

The use of coherence analysis has been applied previously to fNIRS data to study regional connectivity.^9,10 Using a frequency band similar to that in the current study, these groups reported a mean interhemispheric coherence during the resting states of 0.6 and 0.5, respectively, which is similar to that of 0.53 reported in this paper. This repeatability supports the concept that fNIRS is a reliable method for studying functional connectivity in the cortex.

There was no significant difference in contralateral coherence between controls and MS patients during the resting state. fMRI data have been used to report both higher and lower coherence in the motor cortex during the resting state when comparing MS to controls.^5,17–19 Resolution of these differences will need additional studies.

Finger tapping significantly increased HbT coherence in both ipsilateral and contralateral hemispheres of control subjects. However, MS patients did not show significant changes in HbT coherence during finger tapping. The relative increase in coherence in controls during task activation resulted in a significant difference in coherence in the contralateral hemisphere between controls and MS patients during the task. Although we do not have a clear mechanism at this time, there is clearly a change in how coherence is regulated that relates to the presence of task activation. This is in agreement with previous fMRI studies, which found that increased functional connectivity occurred during task activation.^20,21 Reduced coherence could be the result of reduced axonal integrity,² atrophy of the corpus callosum, ^22,23 or increased demyelination.^24,25

Interhemispheric HbO coherence, on the other hand, was not significantly different between MS patients and healthy controls. One possible reason for this discrepancy could be that HbT is less susceptible to contamination signals from pial vessels²⁶ and, hence, is a better indicator of functional communication between brain regions.

In fNIRS, there is a partial volume effect in that light passes though tissue other than brain. There may also be extracerebral signals, such as pial vessels and blood in the skull that will contaminate the signal.²⁶ This would mean that brain atrophy may influence the results if there is a change in distance from the optical fibers to the brain. It is known that brain atrophy is common in MS patients.^27,28 MS patients with a long disease duration have cortical thinning,²⁹ which could increase the distance between the skull and the brain. The reduced magnitude of HbT and HbO responses in MS patients could, in part, be explained if there was an increased distance from the fibers to the brain. However, half of the MS patients included in this study were relapsing-remitting patients, who typically have less brain atrophy compared to progressive MS patients.³⁰ An advantage of coherence analysis is that these partial volume problems would not be expected to impact the frequency responses.

In conclusion, we showed that fNIRS measurements of coherence may be useful in detecting functional changes in MS patients. The overlap between the results of patients and controls would suggest that it would be difficult to identify a single MS patient. However, improvements in data collection and analysis may provide useful additional power with respect to making such a discrimination. For instance, cognitive impairments occur frequently in patients with MS, suggesting alterations in the prefrontal cortex.³¹ Therefore, choosing another region other than the motor cortex may be useful in identifying patients on an individual basis. As we hypothesize that the reduction in coherence is a result of impaired communication following white matter degeneration and axonal loss, it follows that this study paradigm may also be useful for any condition with such a degeneration. This technique could then be applied to disorders such as mild traumatic brain injury³² and schizophrenia.³³ Coherence analysis using fNIRS is a promising technique in brain research and clinical practice.