Multiple sclerosis is a chronic inflammatory demyelinating disease of the central nervous system [18]. Relapsing-remitting MS is the most common disease subtype and consists of relapses separated by periods of remission of variable length. After these alternating episodes of neurological disability and recovery, around 70% of the patients develop a secondary-progressive disease course which is characterized by progressive neurological decline [13].
In spite of the belief that MS is primarily a white matter disease, as early as 1962 pathologic studies had reported focal areas of cortical demyelination in MS patients and even reported cortical demyelinating lesions up to 26% of the total number of cerebral plaques [19].
Despite initial focus on white matter demyelination, there has been increasing focus on the gray matter pathology occurring in MS. This shift in focus was encouraged by new histochemical techniques which markedly increased the visibility of cortical lesions ex vivo [20].
New neuroimaging techniques like Double inversion recovery (DIR) improve the performance of FLAIR and T2 imaging; although DIR shows more sensitivity than FLAIR and T2 imaging, it still misses most of cortical lesions [21].
Measurement of cortical gray matter thickness has several advantages over detection of cortical lesions. Gray matter lesions are difficult to visualize even with advanced sequences, and there is significant variation between readers. Cortical Gray matter thickness measurements, on the other hand, are more sensitive and reliable with reproducible results among research institutions [22].
Different methods of cortical thickness measurements are now available, manual and automated methods, usually manual segmentation is limited to a certain area of the cortex rather than the fully automated methods which can do whole cortical segmentation. Different techniques are now used in the automated method as surface or voxel based or both [23].
Automated surface-based measures of cortical thickness have become the dominant standard in the brain imaging community [24].
FreeSurfer is an open-source software package developed by Harvard University that measures the thickness of the cortex in different brain regions using surface-based analysis techniques [25]. FreeSurfer’s distance measure is to take the closest vertex on the oppo2site surface, then find that vertex’ closest point, and average the two distances. Then, it uses Surface-based alignment, a tool to improved alignment of cortical thickness maps to increase the accuracy of localizing cortical thickness measures across subjects [13].
The general linear model (GLM) is the optimal statistical model used in morphometric brain studies. It has been used successfully in the analysis of the structures of the brain as it is flexible to analyze both continuous and categorical variables [26, 27].
Our study included 30 patients diagnosed with RRMS: 22 females and 8 males with ages ranging from 20 to 35 years and a mean age 27.54 years, and 30 control individuals with a mean age 26.42 years.
The studied patients group had variable degrees of clinical disability represented by different EDSS scores which ranged from 1 to 6.5, with a mean score of 3.64 and a standard deviation of 1.39.
Each patient in this study undertook a high-resolution thin cut 3D T1W SPGR sequence. Then, cortical thickness was measured using FreeSurfer software package. Next, we did whole brain vertex-based analysis is a point-by-point group comparison of thickness across the cortical surface, without any prior hypothesis, starting with the average images of each group.
After cortical thickness measurement, we found that there is a significant decrease in the cortical thickness in the RRMS group compared to the control group in the mean cortical thickness of the left and right hemispheres, and found focal thickness reduction in precentral, paracentral, postcentral and posterior cingulate cortices in both hemispheres.
We subdivided our studied group of patients into three subgroups based on their EDSS scores; the first subgroup represented 23.3% of the total patient’s group, they had EDSS scores ranging from 1 to 2.5 and they showed significant but mild decreased thickness of the precentral, paracentral, postcentral and posterior cingulate cortices in both hemispheres. The second subgroup represented 53.3% of the total patient’s group, they had EDSS scores ranging from 3 to 4.5 and showed significant decreased thickness of the precentral, paracentral, postcentral, posterior cingulate cortices in both hemispheres and mean cortical thickness of both hemispheres. The third subgroup represented 23.3% of the total patient’s group, they had EDSS scores ranging from 5 to 6.5 and showed the most significant and most reduction of thickness of the precentral, paracentral, postcentral, posterior cingulate cortices in both hemispheres and mean cortical thickness of both hemispheres with atrophy patterns starting to affect other areas of the cortex as visual cortex and superior frontal cortex.
Based on the results of these subdivisions and calculating the correlation coefficients, we found that the mean cortical thickness of both hemispheres, precentral, paracentral, postcentral, and posterior cingulate cortices in both hemispheres were negatively correlated with the EDSS scores with correlation coefficients ranging from − 0.9878 to − 0.7977. These findings indicate that physical dysfunction, as measured by higher EDSS scores, was predicted by atrophy of the bilateral sensorimotor cortex, PCC, and global cortical thickness reduction.
The reduction of the mean cortical thickness observed in our study is a key finding in understanding the disease burden and pathology, as it shows that cortical gray matter is affected in MS. Studies used 7 T scanners reported that there is a huge disease burden in the cortical gray matter of MS patients. The possibility to detect this pathology in 1.5 T scanners is very crucial in the clinical setting as it is the most available scanners in the clinical arena [28].
It is worth mentioning that we found that there was affection of other cortical areas in the patients who had EDSS scores of 5 or more. This shows that cortical gray matter affection is increased with disability progression.
Lots of studies used cortical thickness measurements to investigate cortical affection in MS. Steenwijk et al. [29] studied cortical atrophy patterns in multiple sclerosis. They investigated whether gray matter atrophy in multiple sclerosis is a more diffuse ‘global’ process or develops, instead, according to clearly distinct anatomical patterns.
They found that several cortical thickness patterns were differently loaded in patients with multiple sclerosis compared with healthy controls. The patterns relevant for multiple sclerosis were largely symmetric and the degree to which they occurred in patients correlated with clinical dysfunction [29].
Our study is consistent with their findings as they found that higher EDSS scores, was predicted by atrophy of the bilateral sensorimotor cortex and global atrophy despite the absence of an absolute difference in white matter lesion load between patients with multiple sclerosis and controls.
These findings were in confirmation with Narayana et al. [30], who found that there is a strong association between physical dysfunction and atrophy in the sensorimotor cortex.
Schoonheim et al. [31] found that posterior cingulate cortex and bilateral temporal pole atrophy predicted Cognitive dysfunction and information processing speed.
In multiple sclerosis, these areas are known for their significant role in cognitive dysfunction, while in our work we did not find significant affection of the temporal poles.
Furthermore, a study by Louapre et al. [32] revealed that the interruption in the default mode network was strongly predicted by the posterior cingulate cortex atrophy in MS patients with cognitive impaired. Based on this, the authors even proposed that default mode network disconnection may remove the compensatory mechanism that MS patients use to adjust the widespread damage in the brain.
Van den Heuvel et al. [33] tried to explain why non-random cortical atrophy patterns were observed in multiple sclerosis. They hypothesize that, after local gray matter pathology starts, a second-order effect follows that causes atrophy in other anatomically connected gray matter areas. This would explain why the cortical atrophy patterns were mostly located in cortical regions known as ‘network hubs’ in the brain (e.g., posterior cingulate) due to their central position in the structural network these regions are particularly sensitive to atrophy and pathology elsewhere in the brain [34].
This theory is promising as it can explain also the weaker relationship between gray matter atrophy and white matter pathology that was previously observed in the patients with progressive multiple sclerosis [34].
Our study agrees with this theory by observing the decreased thickness of the posterior cingulate cortex, a known central hub in the default mode network.
In addition to cortical thickness measurement studies, it is interesting to find that functional MRI studies found that there was reduction of activity and functional connectivity of the default mode network. A finding that matches our work as we found reduction of the cortical thickness of the posterior cingulate cortex that is a dominant part of that functional network [35, 36].
There are other different methods of MRI brain segmentation applied in MS patients as whole gray and white matter segmentation, white matter lesion volume, cerebellum and brain stem volume and subcortical gray matter segmentation [37, 38].
The uniqueness of the current study is investigating cortical thickness changes in different brain areas rather than whole gray matter measurement enabling us to see if there is focal rather than generalized cortical affection.
In our study, we implemented a widely used fully automated segmentation software, limiting the human (operator) interaction and bias in imaging segmentation, make it possible to replicate the finding and compare our results relative to other studies using the same technique.
Unsupervised fully automated segmentation is not without side effects as there may be segmentation errors [39]; but in our study, we checked the segmentation in every patient; if errors were detected, we used a tool available in the Freesurfer software package to check, edit and correct these errors and reevaluate the thickness measurement.
Limitation of the study
Firstly, the relatively small sample size can produce type I error and can miss subtle differences between the patients and controls. A second limitation is the exclusion of subcortical and cerebellar gray matter from the current work. Lack of longitudinal study is also a major limitation as it will show us the dynamic correlation between disease progression, disability and brain atrophy which can be diffuse or local process; however, this could not be done in the current study.
Another limit of our study lies in the heterogeneity of the patient group in terms of clinical variables, which facilitated the detection of potential correlates of EDSS and cortical thickness, but in turn limited the current approach to the study of the common patterns of brain tissue involvement across a heterogeneous population, while correlations that may be present only in MS subgroups (e.g., early MS, benign MS) may have remained undetected.
Recommendation
Future studies on this subject can benefit from including other subtypes of MS, as PPMS and SPMS, which can aid specific patterns of the cortical gray matter affection related to each subtype, another recommendation is to use higher magnet scanners as 3 T or 7 T to investigate if there any correlation between the site of cortical lesions and the areas of cortical thickness reduction.
It is also important to use other advanced imaging techniques such as Diffusion tensor imaging (DTI), and functional magnetic resonance imaging (fMRI), either resting-state or task-based, to have another perspective in the pathology and brain affection detected by MRI.
