Relationship of Cortical Atrophy to Fatigue in Patients With Multiple SclerosisCortical Atrophy and Fatigue in Multiple Sclerosis FREE

Fatigue is a frequent symptom of patients with multiple sclerosis (MS).¹ It has a severe effect on their quality of life,² but its pathophysiologic mechanisms remain incompletely understood.

Magnetic resonance (MR) imaging is the most sensitive noninvasive technique to study MS pathologic changes in vivo. In particular, high-field MR imaging yields images with higher contrast to noise ratio and high spatial resolution. This allows a more precise differentiation of the different brain tissues, thus providing accurate measurements of their volume.

Although numerous imaging studies have investigated possible neurobiological substrates of fatigue, no correlations were found between the extent of fatigue symptoms and white matter (WM) disease.^{3 - 4} Conversely, discordant results were reported on the impact of brain atrophy on fatigue.^{3 ,5 - 6}

With respect to gray matter (GM) pathologic characteristics, using fludeoxyglucose F 18–labeled positron emission tomography, Roelcke and colleagues⁷ demonstrated reduced glucose metabolism predominantly at the level of the frontal cortex and basal ganglia. These authors, for the first time, raised the question of whether disconnections between pathways of specific corticosubcortical circuits may underlie the existence of MS fatigue. Subsequent reports supported the impact of subcortical GM structure damage on MS fatigue.^{8 - 10}

Controversy still remains on the role of cortical pathological changes in MS fatigue. Functional MR imaging studies showed abnormal activation patterns, mainly in the basal ganglia and frontal and parietal lobes, in fatigued patients while performing simple motor¹¹ or cognitive^{12 - 13} tasks. However, in a quantitative MR imaging study, no differences were found in the magnetization transfer ratios and diffusion tensor-derived metrics of the basal ganglia and cortex between fatigued and nonfatigued patients with MS.¹⁴

On the basis of these previously reported findings, in this study we speculated that atrophy of cortical and subcortical GM structures belonging to functional circuits shown to be dysfunctional in patients with fatigue¹⁵ may represent a neurobiological substrate for this symptom in MS.

Using high-resolution structural MR imaging at 3.0 T and an automated reconstruction of cortical surface and subcortical structures, we quantified the macrostructural damage (ie, volume loss) of thalamus, basal ganglia, and frontal and parietal lobes. We then investigated the relation of cortical and subcortical structure atrophy and MR imaging–derived metrics of WM in MS to fatigue severity.

This case-control study was performed at the National Institutes of Health with approval from the institutional review board. Twenty-four patients with MS¹⁶ and 24 sex- and age-matched (±3 years) healthy volunteers (HVs) were included in the study after providing informed written consent. Patients' inclusion criteria were as follows: diagnosis of relapsing-remitting MS or secondary progressive MS with superimposed relapses; age between 18 and 60 years; Expanded Disability Status Scale (EDSS) score¹⁷ between 0 and 6.5; and treatment with interferon beta (either 1a or 1b) at fully tolerated dose for at least 6 months before enrollment and with evidence of clinical efficacy (ie, reduction [or absence] of clinical relapses with respect to the previous year) at the time of enrollment. Patients' exclusion criteria included long-term therapy with any other immunomodulatory or immunosuppressive medication (excluding standard dosages of corticosteroids intravenously/intramuscularly injected and orally taken for the treatment of relapses) in addition to interferon beta within the past 6 months and other medications used for symptomatic relief that may affect cognition. None of the subjects was taking any medications used to treat cognition or fatigue or other drugs that may act as temporary stimulants or depressants of the central nervous system. The HVs were included only if they matched for sex and age with patients. Care was also taken that each HV was free of any neurologic conditions or related risk factors as well as that he or she was not taking any medication potentially affecting cognition and fatigue.

Each participant underwent a neurologic examination (ie, EDSS, for patients), diagnostic 1.5-T MR imaging,¹⁸ and 3.0-T brain MR imaging.

Fatigue was rated by means of the Modified Fatigue Impact Scale (MFIS).¹⁹ This scale contains 21 items and assesses the patients' perceived impact of fatigue on physical (9 items), cognitive (10 items), and psychosocial (2 items) functioning within the previous 4 weeks. The global score ranges from 0 to 84. A cutoff value of 38 or greater is used to distinguish patients as fatigued. This has been shown to be associated with lack of overlap of the 10th and 90th percentiles between MS groups with and without fatigue.²⁰

The level of depression was assessed by the Center for Epidemiologic Studies Depression Scale (CESD).²¹ This is a 20-item self-reported scale that measures current (1 week) depressive symptoms. A cutoff score of 16 or greater²² is interpreted as suggestive of clinically significant depression.²³ This cutoff score corresponds to the presence of 6 depressive symptoms of 20 explored for most of the previous week or to a majority of symptoms for shorter periods.

Both scales were administered in the same order to all the subjects, always in the morning and on the same day as the 3.0-T MR imaging.

The MR images were obtained on a 3.0-T scanner (Signa 6 Excite; General Electric GE Healthcare, Waukesha, Wisconsin) using an 8-channel receive-only coil array (MRI Devices Corp, Pewaukee, Wisconsin).The MR imaging protocol revelant to this study (Table 1) included 3-dimensional magnetization-prepared rapid-acquisition gradient-echo (MPRAGE), 2-dimensional fluid-attenuated inversion recovery, and T2-weighted fast spin-echo pulse sequences.

Lesion volume was calculated from T2-weighted fast spin-echo and fluid-attenuated inversion recovery images by a single observer blinded to the patients' clinical characteristics. Lesions were first identified on T2-weighted hard-copy images and then outlined on the electronic ones by means of a semi-automatic thresholding segmentation technique (Medical Image Processing, Analysis, and Visualization application, MIPAV version 3.1.6; http://mipav.cit.nih.gov).

Brain cortical thickness (CTh) and volume of deep GM structures were measured from the MPRAGE images using FreeSurfer software, version 3.0.5 (http://surfer.nmr.mgh.harvard.edu). This process consists of several stages, detailed elsewhere.²⁴ Appropriate manual editing was performed.

We considered CTh measures of each subregion of the frontal and parietal lobes. To obtain the whole frontal and parietal lobe CTh, an average CTh weighted for the surface area of each region was computed for both lobes.

Among the deep GM structures, the volumes of caudate, putamen, globus pallidus, and thalamus were considered. All volumes were normalized by the intracranial volume automatically generated by FreeSurfer.

Because left and right CTh and deep GM structure volumes were found to be highly correlated, average values combining both left and right measures were used in the statistical analysis.

The volume of normal-appearing WM (NAWM) was obtained by using the mask of WM generated from the MPRAGE images by FreeSurfer. An affine transformation was then calculated to register the T2-weighted fast spin-echo images to the MPRAGE ones by means of MINC tools.²⁵ The same transformation matrix was used to transfer the T2 lesion mask previously generated in MIPAV onto the WM mask. The T2 lesion mask was subtracted from the WM mask, resulting in an NAWM mask. The latter was then used to compute the NAWM volume. The volume of the NAWM was normalized for brain size by dividing it by the intracranial volume.

A paired t test assessed differences in clinical and MR imaging metrics between patients (n = 24) and HVs (n = 24). A t test between 2 independent samples was used to compare clinical and MR imaging metrics between fatigued and 15 nonfatigued patients. For the comparison of the EDSS scores, the Wilcoxon rank sum test was used because of observed variable skewness. Group differences (ie, patients vs HVs) in the MFIS score were investigated also by the means of an analysis of covariance using the CESD score as a covariate. This correction was done because it remains unresolved whether fatigue is a symptom in the spectrum of depressive episodes or whether depression occurs as a psychological reaction to the daily limitations caused by fatigue.^{26 - 27}

In the patients, Spearman rank correlation analysis assessed correlations between MFIS and CESD scores, MFIS and EDSS scores, MFIS score and years of disease, and EDSS score and each of the 3 components of the MFIS score. In the same group, Spearman rank correlation analysis also examined correlation between each MR imaging metric in explaining the variability in MFIS score.

All reported P values were based on 2-tailed statistical tests, with a significance level of .05. Given the exploratory nature of the study, no corrections for multiple comparisons were used. The statistical analyses were performed with SPSS version 12.0 (SPSS Inc, Chicago, Illinois).

Demographic, clinical, and MR imaging characteristics of the study cohort are presented in Table 2 and Table 3. Patients were more depressed (P = .003) than HVs. Seven patients (29%) and 1 HV had a CESD score of 16 or greater.

Only 1 patient had 2 active lesions on the 1.5-T MR images. Patients had an average reduction of 11.6% in the volume of thalamus, 8.4% in the volume of caudate, and 3.7% in CTh of the frontal and parietal lobes. The differences in regional CTh of parietal and frontal lobe areas between patients and HVs are presented in Table 4 and Table 5.

Patients were more fatigued than HVs (Table 2). Eight patients (33%) and 1 HV had an MFIS score of 38 or greater.

When the 3 components of the MFIS score (ie, physical, cognitive, and psychosocial) were analyzed (Table 2), all 3 were significantly higher in patients than in HVs.

Spearman rank correlation analysis performed within patients and HVs separately showed a strong correlation between MFIS and CESD scores (r = 0.60, P < .001 in patients; r = 0.50, P = .01 in HVs). In patients, significant correlations were found between the MFIS and EDSS scores (r = 0.60, P = .004) but not between MFIS score and disease duration (r = 0.20, P = .20). Significant correlations were also found between the EDSS score and the physical (r = 0.80, P < .001) and psychosocial (r = 0.70, P < .001) components of the MFIS but not with the cognitive component (r = 0.30, P = .20). Results of the analysis of covariance performed to analyze group differences (patients with MS vs HVs) in the fatigue score after controlling for the effect of depression confirmed the group difference in fatigue score (P = .04).

Differences in all demographic, clinical, and MR imaging features between fatigued and nonfatigued patients were analyzed. We report only the significant results. Fatigued patients (n = 8) had a higher EDSS score (median [range], 3.0 [1.0-6.5]) (P = .01) and depression level on the CESD (mean [SD], 18.5 [7.4]; median [range], 19.5 [3.0-27.0]) (P < .001) than nonfatigued ones (n = 16) (EDSS: 1.5 [0.0-4.5]; CESD: 7.2 [4.9], 6.5 [1.0-16.0]). In particular, with respect to depression incidence, 6 patients scored above the normal value on both the CESD and the MFIS, 2 patients were classified as fatigued but not depressed, and 1 patient was classified as depressed but not fatigued. Trends toward significance were also found for greater age (P = .05), lower CTh of the parietal lobe (P = .05), and smaller thalamic volume (P = .07) in fatigued patients (n = 8) compared with nonfatigued patients (n = 16) by means of the 2-sample t test.

The results of the correlation analyses performed to assess the independent contribution of each MR imaging variable on fatigue in the patient group (n = 24) are shown in Table 6. Of all the variables analyzed, only CTh of the parietal lobe showed a significant correlation with the MFIS score and explained 25% of its variance. Because of the significant correlation between the parietal lobe CTh and the MFIS score, Spearman correlation analysis was performed to investigate which of the subregions of the parietal lobe mostly contributed to this effect. Only the posterior parietal cortex (PPC) (given by the supramarginal gyrus, superior parietal cortex, and inferior parietal cortex) (Table 7), and, in particular, the supramarginal gyrus and inferior parietal cortex, showed a significant association with the level of fatigue.

Despite the absence of significant correlation between CTh of the frontal lobe and the MFIS, an explorative Spearman correlation analysis investigating correlations between CTh of each frontal lobe region and the MFIS score was performed in patients. No significant correlations emerged (Table 7).

A challenging issue regarding fatigue in patients with MS is the common use of subjective ratings to assess its subjective perception. Because of the disabling impact of this symptom, identifying an in vivo biological marker of fatigue is crucial. To the best of our knowledge, this is the first study showing a positive correlation linking fatigue and cortical thinning.

In agreement with existing literature,^{1 ,28} the incidence of fatigue reached approximately 33% in our cohort of patients with relatively mild MS. The fatigue-driven impairment was visible in the physical, cognitive, and psychosocial components of the MFIS. Differences in fatigue score between patients and HVs persisted after correction for depression. Given the results from the analysis of covariance in our cohort and the uncertainties existing on the causality role of depression in determining fatigue in MS,^{26 - 27} depression was not used as a correcting factor in the subsequent correlation analyses.

As previously reported,^{29 - 30} we found a significant relationship between physical disability and fatigue severity. However, when the correlation between the EDSS score and the different components of the MFIS were analyzed, only its physical and psychosocial components were found to be correlated with the physical disability. Conversely, no significant correlation was found between the EDSS score and the cognitive component of the MFIS. The finding demonstrates that fatigue in MS encompasses several mechanisms and that the extent of physical disability affects only some of these. The EDSS was not used as a correcting factor in any of the correlation analyses subsequently performed in this report for several reasons. First, given its significant correlation with only some of the components of the MFIS scale, we reasoned that the role of the EDSS score in influencing the fatigue level was only partial. In addition, we took into account the exploratory nature of our study and, similar to the symptom of depression, the as-yet-unclear causal role of the EDSS score in determining fatigue.

We confirm the lack of correlation between fatigue and WM disease expressed by focal lesion volume.^{3 - 5} We did not find a significant correlation between WM pathological findings extending beyond visible lesions (ie, NAWM volume) and the MFIS score. The finding contradicts the previously reported correlation between diffuse WM axonal loss and fatigue.³¹ Such interstudy variability might be due to the different sensitivity of the method used to detect damage in NAWM and the different clinical features of patients included in the 2 studies.

As previously reported, atrophy selectively affected the deep GM nuclei,³² whereas atrophy of cortical regions was more diffusely observed.³³ We extend these findings herein by showing another important clinical correlate of GM atrophy in MS.³⁴ Significant correlations were seen between the MFIS score and the parietal CTh. A trend toward significant correlation was seen between the MFIS score and the thalamic and caudate atrophy. We think that these findings, while paralleling the results obtained by previous investigators regarding the relation between subcortical GM abnormalities^{9 - 11} and MS-related fatigue, stress the importance of corticostriatothalamic loop damage in determining fatigue. Considering the functional role of parietostriatal connectivity in the organization of the motor planning hierarchy,³⁵ one can hypothesize that damage in these structures is part of the neurobiological substrates for fatigue.

We used a subsequent Spearman correlation analysis to explore the relationship between fatigue and atrophy of specific parietal and frontal lobes areas. No correlations were found between fatigue and the CTh of any cortical region of the frontal lobe. Among all parietal cortical areas, fatigue was significantly correlated only with reduced CTh of the PPC. The PPC, classically regarded as a large associative cortical region, is implicated in cognitive functions³⁶ and in all 3 components (shifting, updating, and inhibition) of executive functioning.³⁷ The PPC plays an important role in the higher-order aspects of motor control, such as motor planning.^{38 - 39} In patients with PPC damage due to stroke,⁴⁰ an impaired selection of action under situations of response conflict is observed. In patients with chronic fatigue syndrome, the pattern of activation in PPC and premotor areas suggests that fatigue is associated with dysfunctional movement preparation.^{41 - 42} In MS, in agreement with our findings, DeLuca and colleagues¹³ reported increased across-run fatigue activation in the parietal lobe and basal ganglia. The crucial role of PPC integrity in motor planning may subtend the strength of our correlations despite the fact that the overall decrease of the parietal CTh was on average very low (−3.7%).

Taken together, the interpretation of our results is that atrophy of the parietal lobe may be a substrate for the multifactorial manifestation of fatigue in MS. Bearing in mind the model of central fatigue proposed by Chaudhuri and Behan,¹⁵ that is, “the failure to initiate and/or sustain attention task (mental fatigue) and physical activities (physical fatigue) requiring self motivation,” our findings support the hypothesis that fatigue in MS could be due to impaired selecting and planning of actions, as well as integrating different information, rather than the extent of physical disability and motor impairment.^{12 - 13 ,29 ,43}

We recognize the relatively small sample size of our study and its commensurately low statistical power. The latter factor might mask other important associations and renders important the extension of our work to larger cohorts of patients. We therefore provide preliminary evidence on the role of regional cortical atrophy in determining fatigue in MS when other GM and WM disease variables are considered.

Correspondence: Francesca Bagnato, MD, PhD, Neuroimmunology Branch, National Institute of Neurological Disorders and Stroke, National Institutes of Health, 10 Center Dr, Bldg 10, Room 5C103, Bethesda, MD 20892-1400.

Accepted for Publication: August 18, 2009.

Author Contributions: Dr Bagnato had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pellicano, Cantor, McFarland, and Bagnato. Acquisition of data: Gallo, Ikonomidou, Evangelou, Ohayon, Stern, Ehrmantraut, Cantor, and Bagnato. Analysis and interpretation of data: Pellicano, Gallo, Li, Ikonomidou, Ehrmantraut, Cantor, and Bagnato. Drafting of the manuscript: Pellicano, Li, and Bagnato. Critical revision of the manuscript for important intellectual content: Gallo, Li, Ikonomidou, Evangelou, Ohayon, Stern, Ehrmantraut, Cantor, McFarland, and Bagnato. Statistical analysis: Li. Administrative, technical, and material support: Ehrmantraut and McFarland. Study supervision: Evangelou, McFarland, and Bagnato.

Financial Disclosure: None reported.

Funding/Support: This research was supported by the Intramural Research Program of the National Institute of Neurological Disorders and Stroke, National Institutes of Health.

Additional Contributions: Helen Griffith, RN, and Nancy Richert, MD, PhD, provided clinical assistance; Jhalak Agarwal, BS, assisted with image postprocessing; Zeena Salman, MD, critically revised the manuscript; and Devera Glazer-Schoenberg, MSc, edited the material. Sungyoung Auh, PhD, provided helpful insights with statistical analyses. We are sincerely grateful to all our patients and healthy subjects for the time and cooperation required to participate into the study.