Multimodal assessment of regional gray matter integrity in early relapsing-remitting multiple sclerosis patients with normal cognition: a voxel-based structural and perfusion approach

Abstract

Objective:

There is increasing evidence that gray matter (GM) impairment is strongly associated with clinical performance decline. We aim to perform a voxelwise analysis between regional GM (rGM) perfusion and structural abnormalities in early relapsing-remitting multiple sclerosis patients with normal cognition (RRMS-IC) and explore clinical correlate of early rGM abnormalities.

Methods and materials:

We studied 14 early RRMS-IC patients and 14 healthy age- and sex-matched controls. Brain perfusion single photon emission computed tomography (SPECT), structural MRI, and a comprehensive neuropsychological examination were acquired from all participants. Neuropsychological tests include expanded disability status scale, minimal mental status examination, short physical performance battery, Wechsler memory scale, and quick smell test. Voxel-based morphometry was used for analyzing SPECT and T1-MR images to identify rGM hypoperfusion and atrophy, respectively (RRMS-IC vs controls (group analysis), and also, each patient vs controls (individual analysis)) (p < 0.001). Then, anatomical location of impaired regions was acquired by automated anatomical labeling software.

Results:

There was no significant difference in total GM volume between RRMS-IC and healthy controls, however, rGM atrophy and hypoperfusion were detected. Individual analysis revealed more rGM impairment compared with group analysis. rGM hypoperfusion was more extensive rather than rGM atrophy in RRMS-IC. There was no spatial association between rGM atrophy and rGM hypoperfusion (p > 0.05). rGM abnormalities correlated with several relevant minimal clinical deficits.

Conclusion:

Lack of spatial correlation between rGM atrophy and hypoperfusion might suggest that independent mechanisms might underlie atrophy and hypoperfusion. Perfusion SPECT may provide supplementary information along with MRI.

Advances in knowledge:

Association between rGM atrophy and rGM hypoperfusion and their clinical significance in early RRMS-IC is not well described yet. Our study showed that there is spatial dissociation between rGM atrophy and rGM hypoperfusion, suggesting that different mechanisms might underlie these pathologies.

Introduction

Multiple sclerosis (MS) is a progressive demyelinating neurologic disease that affects the central nervous system (CNS). Although lesions are considered as the most important diagnostic MRI finding, their measurements are only moderately correlated with clinical disability.¹ Moreover, whole-brain atrophy has a stronger relationship, albeit incomplete, with clinical disability compared to the lesion size.²

A large body of data suggests that gray matter (GM) atrophy is a major determinant of clinical disability compared to the lesion burden, whole brain measures, and white matter (WM) atrophy.^1,3 Furthermore, GM matter atrophy occurs early in the disease course and is more sensitive than WM atrophy to identify neurodegenerative progression at early stages of the disease.^1,4,5 Regional atrophy in each part of GM was strongly associated with relevant clinical disability.⁶ However, pathologic processes other than neuronal loss also underlie the disease.

Increasing evidence suggests that vascular abnormalities and perfusion impairment play a role in MS.⁷ Moreover, GM hemodynamic alteration can occur in very early stages of the disease, even in patients with clinically isolated syndrome, and may precede structural abnormalities.⁸ MR perfusion-weighted imaging, positron emission tomography (PET), and single photon emission computed tomography (SPECT) are used for brain perfusion measurement. ^99mtechnetium–ethyl-cysteinate-dimer SPECT (99mTc-ECD-SPECT) is a clinically available radiotracer that usually reflects microvasculature circulation. Brain perfusion SPECT (pSPECT) has established its added value in different neurodegenerative disorders such as Alzheimer’s disease and dementia.⁹ In comparison to MR perfusion-weighted imaging, pSPECT is less influenced by hemodynamic factors, which can yield in better performance.¹⁰

Cognitive disability has a crucial clinical significance in MS patients’ daily lives as it can impair their social activity, professional performance, and personal behavior. Cognitive decline in relapsing-remitting MS (RRMS) can be better explained by cerebral atrophy than by T₂ weighted lesion measures.¹¹ However, atrophy is not the only process that underlies MS as cognition impairment may develop in the absence of brain atrophy in these patients.^12,13

Interestingly, GM perfusion impairment is the most sensitive parameter for detecting the progression of cognitive decline compared to WM atrophy, WM perfusion, and GM atrophy.^12,14 Furthermore, some studies found an association between parietal GM atrophy and fatigue, between frontal cortex hypoperfusion and cognitive impairment, and between thalamus hypoperfusion and atrophy and clinical disability.^14–17

Although atrophy and hypoperfusion are important pathologies in RRMS, the association between these pathologies is not studied well in early RRMS. Literature is scarce on the spatial association between GM structural and perfusion impairment at the early stage of RRMS, RRMS patients with intact cognition (RRMS-IC) per se. On the other hand, GM abnormalities is a major determinant of clinical performance in RRMS. However, the significance of early GM abnormalities on the clinical performance is poorly understood. We performed a voxel-wise analysis between regional GM (rGM) perfusion and structural abnormalities in early RRMS-IC and explored clinical correlate of early rGM abnormalities.

Methods and materials

Participants

19 patients with early RRMS (<5 years) and 14 age- and sex-matched healthy subjects were prospectively included in the study between July 2015 and September 2018. The diagnosis was made by two senior neurologist (>10 years of experience) based on the 2010 revised McDonald’s Diagnostic Criteria,¹⁸ then, early RRMS were referred from a tertiary referral university hospital. The participants’ demographic data including age, sex, disease duration, and history of medication and relapse were documented. The exclusion criteria were alcohol consumption, drug abuse, inability to read and write, history of pre-morbid MS (pre-MS), corticosteroid use or relapse within the past 3 months, psychiatric conditions, concurrent neurologic or psychiatric disorders, head injury, and imaging limitation (contraindications to MRI or SPECT). Informed consent was obtained from all participants. The study was approved by a local ethics committee. The study was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans.

Neuropsychological evaluation and neurological clinical tests

All participants underwent a comprehensive clinical history and neuropsychological evaluation within 1 week of brain MRI and perfusion scan acquisition. RRMS subjects completed the Expanded Disability Status Scale (EDSS), Minimal Mental Status Examination (MMSE),¹⁹ Wechsler Memory Scale (WMS), Short Physical Performance Battery (SPPB),²⁰ Stroop Color Word Test (SCWT), and Quick Smell Test (QST).

WMS is a neuropsychological cognitive battery that contains seven subdomains including information, orientation, mental control, logical memory, digit span (forward and backward), visual reproduction, and associative learning. The age-adjusted memory quotient score as well as the score of each subdomain (information, orientation, mental control, logic memory, digit span, visual reproduction, and associate learning) were calculated.²¹ MMSE is a widely used test for cognitive assessment in the clinical setting for which total and subdomain (orientation, registration, attention and calculation, recall, and language) scores were measured for each patient. Impaired cognition was defined as a z-score ≤2 SD of the total MMSE score or memory quotient score.

Physical performance was evaluated using the Short Physical Performance Battery (SPPB), including walking (going straight-4 meter test (GS-4)), balance (balance test (BL)), and strength tasks (chair stand test (ChS)); for which total and each subdomain score were measured. It has been demonstrated that it can independently predict mobility disability and activities of daily living disability.²² SCWT is a measure of cognitive flexibility and executive functioning that works on an individual performance scale, named interference score. The score measures an individual’s ability to suppress a usual task in support of an analogous unusual task by comparing the performance on the usual task (reading congruous colored-names) with the performance on the unusual task (reading incongruous colored-names). QST is a brief 6-item olfaction test that is culturally adjusted to minimize cross-cultural bias. According to the manufacturer’s instructions (number of correctly identified odorants) and large local normative data, a score of 4 or less was considered as olfactory dysfunction. Raw total and subdomain test scores were converted to z-scores based on age- and sex-adjusted (±education-adjusted) normative data for each test and z-scores ≤2 SD were considered impaired.

Image acquisition

MRI acquisition

MR images were acquired at axial sections using 1.5 T MRI system (GENESIS SIGNA, General Electric; USA) equipped with an 8-channel head coil receiver. T₁ weighted fast spin-echo sequence was acquired with following parameters: repetition time: 600 ms, echo time: 10 ms, flip angle: 90°, field of view: 235.5*235.5 mm, and voxel size: 0.46*0.46*3 mm).

SPECT acquisition

A brain perfusion SPECT study (pSPECT) was done about 1 h after an i.v. injection of 740 MBq (20 mCi) 99mTc-ECD (Pars isotope Company, Iran). All pSPECT images were acquired on a dual-head γ camera (Philips [ADAC] Vertex Plus) equipped with low-energy high resolution collimator. All participants had an intravenous (i.v.) access indwelled, while they had been laid down with closed eyes and unplugged ears, in a quiet and darkened room with minimal sound noise and light. Standard head position was determined based on uniform alignment of the external auditory meatus using an automated table positioning and camera-to-head-detector ratio values. The entire acquisition time was 35 min for each subject. Images were acquired in a 64*64*64 three-dimensional voxel matrix at 64 steps, 30 s each step (voxel size: 5*5*5 mm). Attenuation correction was done using the Chang method (attenuation coefficient 0.12 cm⁻¹). The projections were then processed using backprojection and Butterworth filter (Nyquist frequency cutoff = 0.5, order = 5). Images were presented in three orthogonal planes.

Image processing

All T1 images were preprocessed using Statistical Parameter Mapping 12 (SPM12; Wellcome Centre for Human Neuroimaging, UCL, UK). Briefly, perfusion SPECT images were co-registered to corresponding T1 images for each subject. In the next step, T1 images were corrected for bias field inhomogeneities and were segmented into GM, WM, and cerebrospinal fluid (CSF).²³ The segmentation step was done using “segmentation” function (part of SPM12), and then, accuracy of segmentation was manually checked for any misclassification in all images. After controlling for the accuracy of segmentation, GM and WM images were used to create a study-specific template using Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra (DARTEL) (part of SPM12) with default parameters. Following transforming each participant’s native space GM to study-specific template using a nonlinear registration, they were affine-transformed into standard Montreal Neurological Institute (MNI) space (Montreal Neurological Institute, McGill University), with a voxel size of 2*2*2. Then, they were smoothed with an 8 mm full width half maximum (FWHM) isotropic Gaussian kernel.

Following pre-processing, a mass univariate approach of SPM12 was used to perform a voxelwise two sample t-test that compares every voxel in RRMS-IC group with control group, trying to identify regions with significant GM volume reduction in the RRMS-IC cohort. Additionally, we performed an individualized voxel-wise analysis that compares each patient with the control group in order to detect individual GM atrophy across different patients.

For perfusion SPECT, co-registered SPECT images were segmented into GM, WM, and CSF using transformation matrix of corresponding T1-MRI in each subject. GM-SPECT images were normalized into MNI space by nonlinear registration and affine transformation using the same normalization steps applied to corresponding GM-MRI in each subject. Finally, they were smoothed by an 8 mm isotropic Gaussian kernel.

After the preprocessing steps, voxelwise group analysis and voxelwise individual analysis were performed for SPECT images using the same statistical design applied to GM-T1 images. Proportional scaling global normalization with the mean value was applied.

Statistical significance was defined as a voxel-based p threshold (p < 0.001, not corrected for multiple comparison) and no extent threshold of contiguous voxels was defined. The rationale for individual analysis was that we hypothesized that individual RRMS-related GM abnormalities might be masked during intergroup analysis. Anatomical locations of abnormal coordinates were acquired using automated anatomical labeling (AAL).²⁴

Statistical analysis

IBM SPSS v. 22 (IBM Corp., Armonk, NY) was used for statistical analysis. rGM hypoperfusion and rGM atrophy acquired by individual analysis are expressed as categorical variables (normal vs abnormal) for each patient. χ² was used to explore the spatial association between rGM perfusion impairment and rGM volume loss in RRMS-IC patients.

Moreover, χ² was applied to assess the correlation between individual rGM impairments (hypoperfusion and atrophy) and clinical tests (MMSE subdomains z-scores, WMS subdomains z-scores, executive functioning z-score, total SPPB and its subdomain scores, QST score). Independent sample t-test was used for continuous variables to compare the mean values between the two groups (i.e. age, years of education, mean GM volume). p-values < 0.05 were considered statistically significant.

To better visualize the spatial association of rGM atrophy and hypoperfusion as well as the correlation between imaging findings and clinical data, a heat map analysis with hierarchical clustering was applied. To create a heatmap, the imaging abnormalities was defined as T-value for different brain regions (16 regions). Clinical impairments were defined as z-score, as described earlier. Then, the correlation between imaging metrics (hypoperfusion and atrophy) of different brain regions was calculated using Spearman’s correlation coefficient. Also, the same was applied to correlate between imaging metrics and clinical scores. Finally, heat map analysis with hierarchical clustering of correlations was conducted using the Spearman’s correlation coefficient as the similarity measure. The heat map and correlation clustering were designed using the gplots package of the R software (http: //www.R-project.org).

Results

A total of 19 RRMS subjects were included in this study and cognition status was determined based on the MMSE score, while taking account of the level of education. Five patients were diagnosed with cognitive impairment and were excluded from the study and 14 RRMS-IC patients (mean age: 35.90 ± 9.22; F/M: 6/1) and 14 control subjects (mean age: 34.86 ± 9.09; F/M: 6/1) were included in the analysis. The demographic and clinical data are presented in Table 1. 92% of the patients had a disease duration of 2 years or less and 12 out of 14 patients had an EDSS score of 2.0 or less, reflecting minimal clinical impairment at the early stage of the disease. Age, sex, education, and disease duration were not associated with neither rGM atrophy and hypoperfusion nor any clinical score (p > 0.05).

Table 1.

Demographic and clinical data of participants.

	RRMS-IC	Healthy controls
Demographical data
Mean age, years (SD)	35.9 ± 9.2a	34.8 ± 9.0a
Female to male ratio	6:1a	6:1a
Mean education, year (SD)	12.3 ± 2.9a	12.6 ± 2.2a
Imaging data
Mean GM volume (ml)	565.98a	576.56a
Clinical data
Mean disease duration (SD)	22.3 ± 18.5
Mean EDSS (range)	1.8 (0–4.5)
Mean total MMSE (SD)	28.7 ± 1.2a
Mean orientation-MMSE (SD)	10.0 ± 0.0
Mean registration (SD)	3.0 ± 0.0
Mean recall (SD)	2.6 ± 0.4
Mean attention and calculation (SD)	4.4 ± 0.4
Mean language (SD)	8.7 ± 0.6
Mean memory quotient (SD)	108.9 ± 16.5a
Mean information (SD)	5.7 ± 0.0
Mean orientation-WMS (SD)	5.0 ± 0.0
Mean mental control (SD)	7.2 ± 2.2
Mean logic memory (SD)	10.6 ± 4.2
Mean digit span (SD)	9.8 ± 2.7
Mean visual reproduction (SD)	9.2 ± 3.7
Mean associative learning (SD)	17.0 ± 3.3
Mean executive functionating z-score	0.67 ± 2.5
Mean SPPB (SD)	8.1 ± 2.3
Mean BL (SD)	3.5 ± 0.8
Mean ChS (SD)	1.7 ± 1.1
Mean GS-4 (SD)	2.7 ± 1.0
Mean QST (range)	5.6 (5,6)

BL: balance test, ChS: chair-stand test, EDSS: expanded disability status scale, GM: gray matter, GS-4: gait speed-4 meter test, MMSE: minimal mental status examination, and QST: quick smell test;RRMS-IC: relapsing-remitting multiple sclerosis with intact cognition, SD: standard deviation, SPPB: short physical performance battery, WMS: Wechsler memory scale.

^athere is no statistically significant association (p > 0.05).

Voxelwise image analysis

Visual assessment of GM segments in each patient showed no evident tissue misclassification.

Group image analysis

Group analysis of structural MR and pSPECT images in RRMS-IC patients compared to control subjects was done to identify rGM impairments (Figure 1). Brain regions were segmented into 16 different areas. Group image analysis showed that rGM hypoperfusion is more widespread than rGM atrophy (hypoperfusion = 9 areas out of 16 areas, MRI = 4 areas out of 16 areas) (p < 0.001). Two regions showed (precentral and inferior frontal) concurrent hypoperfusion and atrophy. Multimodal imaging (pSPECT/MR) yielded the highest coverage for rGM abnormalities (pSPECT/MR = 11 areas (hypoperfusion and/or atrophy) out of 16 areas) (p < 0.001). The rGM abnormalities in the early RRMS-IC are summarized in Tables 2 and 3 and Figure 1. Furthermore, there was no significant reduction in the total GM volume of RRMS-IC patients compared with healthy controls (p > 0.05); however, rGM atrophy and hypoperfusion were identified (p < 0.001).

Figure 1.

Voxelwise group analysis of brain pSPECT (a, c) and MRI (b, d) in RRMS-IC patients compared with controls (p < 0.001). In multislice views (a, b), green areas are depicting rGM hypoperfusion (a) and rGM atrophy (b). In three-dimensional brain glass views (c, d), areas with significantly lower rGM perfusion (c) and volume (d) in RRMS-IC patients are depicted. These images depict spatial dissociation between rGM atrophy and rGM hypoperfusion. rGM: regional gray matter; RRMS-IC: Relapsing-Remitting Multiple Sclerosis with Intact Cognition; SPECT, single photon emission tomography.

Table 2.

Group analysis of regional GM volume and perfusion abnormalities in RRMS-IC group compared to control subjects

Brain region	X,Y,Z (mm)	Number of voxels	T - Value
MRI
Left inferior frontal gyrus	−38, 12,–16	61	5.85
Right insula	44, 2,–6	81	6.73
Left insula	-42,–10, 4	17	4.22
Left precentral gyrus	−46, 0, 10	8	3.99
pSPECT
Right parahippocampal gyrus, right hippocampus, right amygdala, right fusiform gyrus	38,2,–20	251	8.25
Left gyrus rectus, right gyrus rectus, right middle frontal gyrus, left middle frontal gyrus, left superior frontal gyrus, right inferior frontal gyrus, right superior frontal gyrus	−4, 38,–20	340	6.57
Right inferior parietal lobule, right superior parietal lobule, right postcentral gyrus, right angular gyrus	50,–50, 54	238	5.72
Left inferior parietal lobule, left postcentral gyrus, left angular gyrus	-54,–42, 54	190	5.3
Right precentral gyrus, right superior frontal gyrus, left precentral gyrus, left paracentral lobule	20,–10, 76	365	5.24
Right middle temporal gyrus	60,–26, −14	27	4.89
Left inferior temporal gyrus	-52,–30, −24	7	4.51
Left crus I of cerebellar hemisphere	-36,–76, −26	7	4.46
Right middle temporal gyrus	64,–12, −14	3	4.38
Left inferior temporal gyrus, left middle temporal gyrus	-60,–22, −20	6	4.3
Left crus I of cerebellar hemisphere	-32,–78, −26	1	3.99
Right postcentral gyrus	48,–30, 62	2	3.69
Left inferior frontal gyrus	−54, 18,–2	1	3.62
Left postcentral gyrus	-34,–32, 68	1	3.39
Right postcentral gyrus	38,–42, 62	1	3.38

GM, gray matter; RRMS-IC, Relapsing-Remitting Multiple Sclerosis with Intact Cognition; SPECT, single photon emission tomography.

Table 3.

Group and individual analysis of rGM volume and perfusion abnormalities in RRMS-IC group compared to control subjects

Integrated analysis of regional GM structural and hemodynamic impairments
			Regional GM atrophy	Cerebral cortex area	Regional GM hypoperfusion
Superior temporal cortex	temporal lobe	Superior temporal cortex
Inferior temporal cortex		Inferior temporal cortex
Middle temporal cortex		Middle temporal cortex
Precentral cortex	frontal lobe	Precentral cortex
Inferior frontal cortex		Inferior frontal cortex
Middle frontal cortex		Middle frontal cortex
Superior frontal cortex		Superior frontal cortex
Paracentral cortex		Paracentral cortex
Superior parietal lobule	parietal lobe	Superior parietal lobule
Inferior parietal lobule		Inferior parietal lobule
Postcentral cortex		Postcentral cortex
	occipital lobe
Insula cortex	insula	Insula cortex
	Limbic system
Cerebellar cortex	Cerebellum Cortex	Cerebellar cortex

GM, gray matter.

Green-colored areas indicate GM regions with significant reduction in volume or perfusion (p < 0.001). Red-colored areas indicate GM regions without significant abnormality (p > 0.05). Yellow-colored areas indicate GM regions with significant difference in volume and/or perfusion that were only identified by individual analysis (p < 0.001).

Individual image analysis

In this step, each RRMS-IC subject was compared to the control group to evaluate the individual rGM volume loss and perfusion defect of each patient (p < 0.001). Individual analysis resulted in the detection of a larger extent of the disease burden compared to group analysis, reflecting the fact that less frequent or less severe abnormalities may be masked in the process of group analysis, particularly in early RRMS-IC. The results of individual analysis and comparison with group analysis are presented in Table 3.

Spatial association between rGM volume loss and rGM perfusion impairment

Based on individual image analysis, there was no spatial association between rGM atrophy and rGM perfusion impairment in corresponded regions of the brain (p > 0.05). Furthermore, the heatmap analysis showed that no distinct cluster was identified in hierarchical clustering. This suggests lack of spatial correlation between rGM atrophy and rGM hypoperfusion (Figures 2 and 3).

Figure 2.

Heatmap analysis with hierarchical clustering of regional GM abnormalities in early RRMS-IC. According to Spearman’s correlation coefficient, there is no spatial correlation between regional GM atrophy and regional GM hypoperfusion. GM: gray matter; RRMS-IC: Relapsing-Remitting Multiple Sclerosis with Intact Cognition.

Figure 3.

rGM hypoperfusion and rGM atrophy in a patient with RRMS-IC. rGM atrophy is depicted by arrows in MRI sequences (a, axial T₁ weighted; C, axial fluid attenuation inversion recovery; and D, axial T₂ weighted). rGM hypoperfusion is shown via arrowheads in axial pSPECT (b). rGM: regional gray matter; RRMS-IC: Relapsing-Remitting Multiple Sclerosis with Intact Cognition; SPECT, single photon emission tomography.

Imaging correlation with clinical scores

In the next step, the correlation of individual rGM atrophy and rGM hypoperfusion with clinically relevant performance was explored in RRMS-IC patients. The clinical correlation of rGM volume loss and rGM perfusion defect is presented in Table 4 and Figure 4. Both imaging modalities correlated with relevant clinical scores in the RRMS-IC cohort; however, MR imaging correlated with cognitive tests (n = 10) and ambulatory tests (n = 3) better than pSPECT (p < 0.05). Meanwhile, pSPECT was associated with a larger scope of early clinical deficits, including cognitive tests (n = 5), ambulatory tests (n = 1), visual (n = 1), sensory (n = 1), bowel & bladder voiding (n = 1), and olfactory performance (n = 1) (p < 0.05).

Table 4.

Association of rGM impairment with clinical performance deficit in early RRMS-IC based on individual image analysis and neuropsychological score

Impaired test performance associateda with regional GM atrophyb	Cortical region	Impaired test performance associated with regional GM hypoperfusionb
			Functional mentation (EDSS)b	Temporal cortex	Functional sensory (EDSS)
			Mental control (WMS)	Total frontal cortex	Attention and calculation (MMSE)
Forward & backward digit span (WMS)	Precentral cortex	Functional ambulation (EDSS)b
		Total EDSS
	Paracentral cortex	Functional Bowl & bladder (EDSS)
	Superior frontal cortex
Language (MMSE)	Middle frontal cortex
Chair-stand (SPPB)
Information (WMS)	Inferior frontal cortex	Associate learning (WMS)
Functional mentation score (EDSS)
Mental control (WMS)
Functional ambulation (EDSS)
	Parietal cortex	QST
	Occipital cortex
	Insula cortex	Recall (MMSE)
		Information (WMS)
		Functional visual (EDSS)
Visual reproduction (WMS)	Limbic system	Information (WMS)
Forward & backward digit span (WMS)
Information (WMS)
Chair-stand (SPPB)	Cerebellum cortex

EDSS: Expanded Disability Status Scale,GM: Gray Matter, MMSE: Mini-Mental State Examination, QST: quick smell test; SPPB: Short Physical Performance Battery, WMS: Wechsler Memory Scale.

This table presents the association of rGM atrophy and rGM hypoperfusion with clinical performance decline in RRMS-IC patients (p < 0.05). rGM atrophy was strongly associated with cognitive (10 sub domain scores) and ambulatory tests (three sub domain scores) impairments, while, rGM hypoperfusion associated with a broader range of early clinical performance decline.

^aEach whole test (e.g., EDSS, MMSE, SPPB, or WMS) are divided to their sub domain scores in order to explore association between early GM abnormalities and minimal performance decline.

^bGreen- and blue-colored clinical tests indicate ambulatory tests and cognitive tests, respectively.

Figure 4.

Heatmap analysis with hierarchical clustering of correlation between imaging findings and clinical deficits in early RRMS-IC. RRMS-IC: Relapsing-Remitting Multiple Sclerosis with Intact Cognition.

These results may suggest that pSPECT is more sensitive than MRI for early detection of clinical significance of GM abnormalities before apparent clinical impairment. Furthermore, rGM atrophy showed a stronger correlation with early cognitive and ambulatory impairment before an obvious cognitive decline compared to rGM hypoperfusion in RRMS-IC patients. There was an association between parietal hypoperfusion and smell performance decline, which was within the normal range.

The heatmap analysis (Figure 4) shows the hierarchical clustering of correlations between imaging impairments (t-values) and clinical data (z-scores) based on the Spearman’s correlation coefficient as a measure of similarity. t-Values represent the degree of imaging impairment and are expressed as positive values. Meanwhile, z-scores represent the degree of clinical impairment majority of clinical scores (z-scores) are negative values. Therefore, as a degree of metric increases the relevant z-score decreases, justifying the majority of relevant imaging-clinical correlations are reverse correlations in the Spearman’s correlation test (-1 to 0). Furthermore, poor clustering results suggest that regional assessment of GM is insufficient to completely correlate GM abnormalities with clinical performance, as neural networks are wired across different regions of the brain.

Discussion

To the best of our knowledge, this is first assessment of rGM integrity by means of quantitative structural and perfusion modalities using MRI and pSPECT in early RRMS with normal cognition. Our results indicate that rGM perfusion impairment was spatially independent of rGM atrophy in early stage of RRMS-IC. This spatial dissociation may suggest that different mechanisms underlie these pathologies. Moreover, our study demonstrated that both rGM abnormalities were of clinical significance, highlighting the role of rGM abnormalities in early stage of the disease.

GM perfusion impairment in early RRMS

Eijlers el al²⁵ conducted a study to determine factors contributing to cognitive decline in 332 MS patients and concluded that WM damage was associated with cognition impairment in the presence of atrophy. Intriguingly, two unexpected groups were observed in MS patients, one including patients with impaired cognition without GM atrophy and other including patients without cognition impairment with GM atrophy. Furthermore, Hojjat el at¹² found that rGM perfusion impairment in the absence of atrophy could lead to cognition impairment. Another study³ found that GM perfusion reduction was the most sensitive factor for the disease progression toward cognition decline in RRMS patients.

Debernard el al.²⁶ reported an association between memory decline and perfusion values in the absence of structural differences. However, Debernard el al. demonstrated that there was no reduction in global or regional analysis of GM volume in 25 early RRMS patients versus controls.²⁶ Similar to their findings,²⁶ there was not significant difference in global GM volume of RRMS-IC patients compared with healthy controls in the present study. In contrast to their findings,²⁶ there were areas with significant rGM atrophy in our RRMS-IC cohort compared with controls. In comparison with previous study,²⁶ in the present study, RRMS-IC cohort had a shorter disease duration (1.8 vs 2.4 years) and higher neurological disabilities (EDSS: 1.8 vs 1.5). This may suggest that our RRMS-IC cohort had a more progressive course of disease compared to patient population of previous study,²⁶ resulting in the development of rGM atrophy even in the early course of disease.

Consistent with these studies, clinical correlation was found between rGM hypoperfusion and minimal clinical disabilities in different neural functions, indicating the crucial role of cerebral perfusion in early RRMS.

Although atrophy is irreversible, it seems that perfusion impairment is a reversible event in MS.^27,28 The present study, in good agreement with the current literature, showed that rGM atrophy represented a pathology with a smaller extent but a higher clinical significance, particularly in terms of cognitive and ambulatory task impairment, while rGM hypoperfusion represented a more frequent, less severe pathologic process with a larger scope of association with early clinical deficits. Therefore, together with the current literature, the results of this study might indicate that rGM perfusion values could be considered a surrogate of early clinically relevant imaging biomarker better than MRI measures before apparent clinical impairment.

Debernard el al.²⁶ and Hojjat et al¹² showed that perfusion impairment could develop in the absence of structural abnormalities and might affect the cognition status. Mulholland et al²⁹ investigated the spatial correlation between atrophy and hypoperfusion in RRMS, secondary progressive, and healthy control groups. They concluded that there was no spatial association between cortical GM and WM pathologies. However, this analysis was not done in RRMS patients due to normal GM volumes in these subjects.

The results of the present study completed the puzzle by showing lack of spatial correlation between the rGM volume loss and rGM perfusion impairment in patients with early RRMS-IC. Therefore, together with previous studies, the findings suggest that atrophy and hypoperfusion are at least largely independent pathologies with possible different underlying pathophysiological mechanisms. Besides, consistent with above studies, our data lend evidence for the complementary role of perfusion imaging along with structural imaging in order to provide a larger coverage for detecting rGM impairments.

GM volume loss in early RRMS

A large body of evidence supports the key role of GM atrophy in the development and progression of clinical disabilities in RRMS patients.¹ Amato et al⁴ showed that GM atrophy was related to cognition decline in early RRMS patients with mild cognition impairment. Moreover, Tiberio et al⁵ found increasing GM atrophy in the early clinical course of RRMS. Consistent with the available evidence, the present study demonstrated that rGM atrophy was present in early RRMS patients, even in the absence of cognitive impairment. Furthermore, the results showed that rGM was strongly associated with clinical performance reduction on relevant tests.

Bergsland et al³⁰ conducted a 10-year study on the spatial pattern of the GM volume loss associated with cognitive decline in RRMS patients. They found that the pattern of GM atrophy associated with cognitive decline included regions that were mainly involved in motor functions and different cognition domains, especially memory and learning.

Moreover, “no evidence of disease activity” in a 2-year follow-up could not predict long-term disabilities in RRMS patients. The MS-EPIC team recently conducted a 10-year prospective study³¹ and concluded that long-term disabilities were a common destination for RRMS patients, but they develop largely independent of clinical relapses. They coined the term “silent progression” to describe a progressive insidious brain atrophy mechanism in RRMS that is largely independent of clinical relapses and available therapeutics and is strongly associated with long-term disability accumulation. Therefore, given these valuable findings, the results of this study indicate that GM atrophy is of clinical significance in the early disease stage. Moreover, it highlights an unmet need for early GM atrophy identification and atrophy-preventing therapeutic interventions in this stage.

Other considerations

There are several MRI techniques that measure tissue perfusion.³² These techniques include dynamic susceptibility contrast (DSC), dynamic contrast enhancement (DCE), and arterial spin labeling (ASL).³² In DSC and DCE, use extrinsic contrast and their perfusion metrics can be influenced via large vessels.³² ASL can reflect microvasculature perfusion independent of large vessel effect.³² Furthermore, ASL is more sensitive than bolus perfusion weighted imaging.³³ One study showed that ASL can identify perfusion abnormalities in half of patients with normal findings on bolus perfusion weighted imaging.³³ However, major limitation of ASL is its intrinsic low signal-to-noise ratio.³² On the other hand, MR perfusion-weighted imaging allows for a one-stop shop MRI routine evaluation. This is the main advantage of MR perfusion-weighted imaging compared to pSPECT, which might be more relevant in clinical practice.

Brain perfusion SPECT (pSPECT) has established its role in different neurodegenerative disorders such as Alzheimer’s disease and dementia.⁹ Brain pSPECT can reflect tissue perfusion at microvasculature level. In comparison to ASL, pSPECT was less influenced by hemodynamic factors such as arterial transit time, which can result in better performance.¹⁰ This advantage of pSPECT over MR perfusion-weighted imaging may be useful in early RRMS, as the amount of cerebral involvement is low. However, pSPECT has some limitations that include low spatial resolution, lack of absolute quantification, and radiation exposure. Also, pSPECT requires a two-step process of separate MR and pSPECT acquisitions, which is less feasible in clinical practice compared to MR perfusion-weighted imaging.

One study showed that ASL is less affected by hemodynamic parameters compared with pSPECT.¹⁰ One study showed that ASL is less affected by hemodynamic parameters compared with pSPECT.¹⁰ However, patient population in this study was older than our cohort (mean age: 72 vs 35.9).¹⁰ Another study showed that the effect of hemodynamic parameters such as arterial transit artifact on ASL is more frequent in older patients rather than younger cases.³³ Thus, given the age-group of MS patients, the superiority of pSPECT over ASL is questionable in RRMS patients, which is needed to be further studied in future.

The small size of the RRMS-IC group was the main limitation of the present study. This limitation can reduce generalizability of our results to larger population of RRMS patients. Therefore, our results should be interpreted with caution. In order to gain a better insight into the role of perfusion impairment in RRMS-IC population, another study with larger sample size is required to validate our findings. Also, it may reveal a more accurate association between imaging abnormalities and clinical disabilities. In the present study, hyperintense T2 lesions were not take into account during patient selection, which may produce a bias. These lesions can intersect neural tracts, leading to clinical disabilities independent of brain atrophy and hypoperfusion. Therefore, we suggest to take lesion load into account in future studies, which will allow for more comprehensive assessment of disease. In the present study, we used fast spin-echo pulse sequences with a slice thickness of 3 mm. This slice thickness might be too much thick for accurate calculation of GM volume due to the partial volume effect. Furthermore, in the absence of inversion pulse, the GM-WM contrast can be limited on spin-echo pulse sequences, compared with 3D-MPRAGE sequences.

Conclusion

In conclusion, this study suggests that structural and perfusion impairments are spatially independent pathologies that might have different underlying mechanisms and are capable of declining the clinical performance in early RRMS. Furthermore,perfusion SPECT may provide supplementary information along with MRI.