Abstract
Objectives: Recent studies have indicated that brain atrophy is more closely associated with cognitive impairment in multiple sclerosis (MS) than are conventional MRI lesion measures. Enlargement of the third ventricle shows a particularly strong correlation with cognitive impairment, suggesting clinical relevance of damage to surrounding structures, such as the thalamus. Previous imaging and pathology studies have demonstrated thalamic involvement in MS. In this study, we tested the hypothesis that thalamic volume is lower in MS than in normal subjects, and that thalamic atrophy in MS correlates with cognitive function.
Methods: We studied 79 patients with MS and 16 normal subjects. A subgroup of 31 MS subjects underwent cognitive testing. The thalamus was segmented in whole from three-dimensional MRI scans. We also determined whole brain atrophy (brain parenchymal fraction), third ventricular width, and whole brain T2-weighted (fluid-attenuated inversion recovery) hyperintense, T1 hypointense, and gadolinium-enhanced lesion volumes.
Results: Normalized thalamic volume was 16.8% lower in the MS group (p < 0.0001) vs controls. Cognitive performance in all domains was moderately to strongly related to thalamic volume in the MS group (r = 0.506 to 0.724, p < 0.005), and thalamic volume entered and remained in all regression models predicting cognitive performance. Thalamic volume showed a weak relationship to physical disability score (r = −0.316, p = 0.005).
Conclusion: These findings suggest that thalamic atrophy is a clinically relevant biomarker of the neurodegenerative disease process in multiple sclerosis.
Glossary
- BDI-FS=
- Beck Depression Inventory–Fast Screen for Medical Patients;
- BPF=
- brain parenchymal fraction;
- BVMT-R-D=
- Brief Visuospatial Memory Test (Delayed Recall);
- BVMT-R-TR=
- Brief Visuospatial Memory Test (Total Recall);
- COWAT=
- Controlled Oral Word Association Test;
- CVLT-II=
- California Verbal Learning Test, second edition;
- CVLT-II-D=
- California Verbal Learning Test (Delayed Recall);
- CVLT-II-TR=
- California Verbal Learning Test (Total Recall);
- EDSS=
- Expanded Disability Status Scale;
- FLAIR=
- fluid-attenuated inversion recovery;
- FOV=
- field-of-view;
- JLO=
- Judgment of Line Orientation Test;
- MS=
- multiple sclerosis;
- NC=
- normal controls;
- NSA=
- number of signal averages;
- COWAT=
- Controlled Oral Word Association Test;
- PASAT=
- Paced Auditory Serial Addition Test;
- SPGR=
- spoiled gradient recall;
- SDMT=
- Symbol Digit Modalities Test;
- TE=
- echo time;
- TR=
- repetition time.
Cognitive dysfunction is a disabling manifestation of multiple sclerosis (MS), affecting approximately 50% of patients.1 MS-related cognitive impairment mainly affects attention, information processing speed, and episodic memory.2,3 The severity of these deficits may reflect the extent of lesions and the degree of tissue loss and disorganization outside lesions.4
MS is now recognized as a destructive disease in that CNS atrophy is common, occurs early in the disease course, and is typically progressive.5,6 The underlying substrate of atrophy includes axonal and neuronal loss.7,8 Recent data indicate that brain atrophy involves both gray and white matter.9–13 Both the cortical and subcortical gray matter structures develop atrophy.8,9,14–18
A growing body of data indicates that cognitive impairment in MS is related in part to damage of the subcortical gray matter areas.4 This includes studies showing strong associations between impaired information processing speed and the bicaudate ratio,19 T2 hypointensity of deep gray matter nuclei,20 and enlargement of the lateral ventricles.21 Enlargement of the third ventricle shows a particularly strong relationship to cognitive impairment even after accounting for the influence of general lesion measures and whole brain atrophy.22 The close proximity of the thalamus to the third ventricle and the correlation between thalamic volume and third ventricle width suggest a role for the thalamus in MS-associated cognitive impairment.8
The thalamus is involved in limbic circuitry and mediates or regulates many cognitive functions.23,24 Numerous studies have shown damage to the thalamus in MS, such as T2 hypointensity,25,26 hypometabolism,27 increased diffusivity,28 decreased magnetization transfer ratio,29,30 decreased neuronal integrity and loss of neurons,8 and macroscopic atrophy.8,31 Furthermore, dysfunction of the thalamus has been linked to cognitive impairment27 and fatigue32 in patients with MS.
We focused on thalamic volume and its relationship to cognitive function in MS. We examined three hypotheses: 1) thalamic volume is lower in patients compared with healthy controls; 2) thalamic atrophy is related to cognitive impairment; and 3) thalamic volume accounts for more of the variance in cognitive function than do other brain MRI measures in the MS group, including T1 hypointense and T2 hyperintense lesion load, whole brain atrophy, and third ventricle width.
METHODS
Patients.
Demographic characteristics are summarized in table 1. We studied 79 patients who met diagnostic criteria for MS33 and 16 normal subjects. Age ranges were 21 to 63 years in the MS cohort and 21 to 58 years in the normal cohort. The participants were enrolled consecutively from a community-based, university-affiliated MS clinic. The mean (± SD) Expanded Disability Status Scale (EDSS) score34 was 3.4 ± 2.0 (median 3.0, range 0 to 8), and the timed 25-ft walk35 was 7.4 ± 5.0 seconds. Disease duration was 9.7 ± 7.1 years (range up to 28 years). The disease course was relapsing–remitting in 62 patients (78.4%), secondary progressive in 16 patients (20.3%), and primary progressive in 1 patient (1.3%). Exclusion criteria were 1) a current or past major medical, neurologic, or psychiatric disorder (other than MS related); 2) previous or current substance abuse; and 3) MS relapse or corticosteroid use within 4 weeks preceding MRI or cognitive testing. All MS patients and controls underwent MRI.
Table 1 Demographic characteristics
A subgroup of 31 MS patients (aged 22 to 55 years) underwent neuropsychological evaluation within 6 months of MRI. The EDSS score for this cohort was 3.7 ± 2.2 (median 2.5, range 1 to 8), the disease duration was 11 ± 6.7 years, and the timed 25-ft walk was 8.1 ± 6.8 seconds. The disease course was relapsing–remitting in 26 patients (83.9%) and secondary progressive in 5 patients (16.1%). The patients were selected for neuropsychological testing with the following additional selection criteria: 1) willing to undergo neuropsychological testing and 2) availability of a caregiver. Thus, these patients were identified consecutively, not based on the presence of cognitive symptoms. By two-way analysis, there were no differences between the MS patients not undergoing cognitive testing and those in the cognitive MS cohort as pertains to age (p = 0.765), sex (p = 0.793), disease duration (p = 0.165), EDSS score (p = 0.275), disease course (p = 0.532), or timed 25-ft walk (p = 0.820). Patients in the cognitive MS cohort had demographics not different from the normal controls (p > 0.05) by two-way analysis (table 1), except that they had fewer years of education (14.5 ± 2.2 vs 16.5 ± 2.4 years, p < 0.01). Disease course in the MS patients reflected the expected distribution of a typical MS sample of patients with this age range and disease duration.36
Nine patients (11.4%) had not received immunomodulating therapy within 6 months of enrollment. Fifty-five patients (69.6%) were receiving IM interferon beta-1a weekly. The remaining patients were treated with either glatiramer acetate (four patients), combined interferon with oral immunosuppressant therapy (one patient), IV immunoglobulin (two patients), or monthly IV methylprednisolone (three patients) or were switched between different injectable immunomodulating agents (five patients). None of the patients received IV chemotherapeutic agents.
MRI.
Brain MRI was performed on each subject using the same scanning protocol on a General Electric 4x/Lx 1.5-T scanner (Milwaukee, WI). The MRI protocol relevant for quantitative analysis consisted of a coronal three-dimensional T1-weighted spoiled gradient recall (SPGR), axial fluid-attenuated inversion recovery (FLAIR), and axial T1-weighted conventional spin-echo pre-/postgadolinium T1. IV injection of 0.1 mmol/kg gadolinium preceded postcontrast imaging by 5 minutes. The details of the pulse sequences were as follows: for SPGR: field-of-view (FOV) 24 × 18 cm, matrix 256 × 256, 70 slices, 2.5 mm thickness, repetition time (TR) 24 msec, echo time (TE) 7 msec, flip angle 30 °, number of signal averages (NSA) 1, scan time 5:45; for FLAIR: FOV 24 × 24, matrix 192 × 256, 28 slices, 5 mm thickness, TR 8,002, TE 128, inversion time 2,000 msec, echo train length 22, NSA 1, scan time 4:16; and for T1 spin-echo: FOV 24 × 18, matrix 192 × 256, 24 slices, 5 mm thickness, TR 600, TE 9, NSA 2, scan time 2:56. There were no interslice gaps in any of the sequences.
Image analysis.
Analysis was performed using the software package Jim (version 3.0, Xinapse Systems Ltd., Northants, UK, http://www.xinapse.com) by trained technicians who were blind to the clinical data. As previously described,37 hypointense lesions on T1-weighted images were segmented using an edge-finding tool. FLAIR lesion volume was determined using a thresholding procedure.37 Third ventricle width was measured from T1-weighted axial images as previously described.22 To assess whole brain atrophy, a normalized measure of whole brain volume, brain parenchymal fraction (BPF), was obtained from the axial T1-weighted spin-echo images.37,38 BPF was the ratio of brain parenchymal volume (tissue compartment) to the total intracranial volume.
The whole thalamus was traced from the coronal three-dimensional images by trained technicians with verification by an experienced observer (M.H.), none of whom were aware of clinical information. The thalamic boundaries were determined using an edge-finding tool with manual adjustments as necessary (figure 1). Raw thalamic volumes were normalized within each subject as a ratio to the intracranial volume. The resulting normalized thalamic volume was referred to as the thalamic fraction. Only three subjects had overt lesions in the thalamus (hypointensities) on the SPGR images. Only three subjects had gadolinium-enhancing lesions; thus, this measure was excluded from subsequent analysis.
Figure 1 The thalamus traced in whole from coronal three-dimensional MRI scans
A representative slice is shown with the segmented borders of the thalamus.
Reproducibility of MRI data.
Ten randomly chosen subjects (5 patients with MS and 5 normal controls) had thalamic segmentation repeated by the experienced observer (M.H.) to determine intrarater reliability (expressed as the coefficient of variation [COV]). The mean COV for the 10 subjects was 5.4% (8.9% in the MS group and 2.0% in the control group). We have already established intrarater COV for the other MRI measures (1.7% for T1 hypointense lesion volume, 1.2% for FLAIR hyperintense lesion volume, 5.2% for third ventricle width, and 0.31% for BPF).22,37
Cognitive testing.
Neuropsychological testing was according to consensus panel recommendations.3 The neuropsychological tests are reliable and valid.39,40 This battery, known as the Minimal Assessment of Cognitive Function in Multiple Sclerosis, includes the Controlled Oral Word Association Test (COWAT),41 Judgment of Line Orientation Test (JLO),41 California Verbal Learning Test, second edition (CVLT-II),42 Brief Visuospatial Memory Test–Revised (BVMT-R),43 Paced Auditory Serial Addition Test (PASAT),44 and Symbol Digit Modalities Test (SDMT).45
The COWAT was administered by the Benton method.41 In successive 1-minute trials, participants generated as many words as possible, beginning with each of three designated letters. The dependent measure was the total number of correct words over the three trials.
The JLO required participants to identify the angle defined by two stimulus lines from among those defined by a visual array of lines covering 180 °. Both oral and pointing responses were allowed. The dependent variable was the total number of correct responses over 30 items.
The CVLT-II and BVMT-R are both learning and memory tests. Both require discrete exposures to new material followed by unaided recall immediately after presentation. There is a 25-minute interval after the final learning trial, after which participants recall the information again without further exposure to the to-be-learned material. Delayed recall is followed by a yes/no, forced-choice recognition task. For the CVLT-II, there were five learning trials. Examiners read 16 words and asked participants to repeat as many words as possible. The entire word list was repeated each time. For the BVMT-R, the stimulus material was a matrix of six visual designs, held before the participant for 10 seconds. Participants were asked to render the designs using paper and pencil, taking as much time as needed. Each design received a score of 0, 1, or 2, based on accuracy and location scoring criteria. There were three free-recall trials followed by 25-minute delayed recall and yes/no recognition trials. In this study, we considered the following measures for each memory test: total recall over all learning trials (CVLT-II-TR, BVMT-R-TR) and recall after the delay interval (CVLT-II-D, BVMT-R-D).
In accordance with published guidelines,3 we used Rao's adaptations1 of the PASAT and SDMT. The PASAT included 60 trials presented at interstimulus intervals of 3 and 2 seconds. The 3-second version is part of the Multiple Sclerosis Functional Composite, a clinical outcome measure composed of quantitative measures of leg, upper extremity, and cognitive function.35,46 The dependent measure was the number of correct responses from each of two trials. We used only the oral response version of the SDMT. Participants were presented with a series of nine symbols, each paired with a single digit in a key at the top of an 8.5 × 11-in sheet. The remainder of the page presented a pseudo-randomized sequence of symbols. Participants responded by voicing the digit associated with each symbol as quickly as possible.
Depression was assessed using the Beck Depression Inventory–Fast Screen for Medical Patients (BDI-FS).47 The BDI-FS emphasizes thought content (e.g., negative self-evaluation or guilt) and mood states (e.g., dysphoria) and avoids assessment of vegetative signs that can occur in medical illness without depression. This test has been validated in an MS sample.48
Statistical analysis.
Group comparisons used analysis of variance or analysis of covariance. All between-group neuropsychological comparisons controlled for the influence of education (in years). The significance of correlations was tested using the Pearson r statistic. Regression modeling was performed with a forward selection procedure with p to enter < 0.05 and p to exit > 0.10. The linear regression models controlled for the effects of age, sex, and depression as measured by BDI-FS. Specifically, all models included age and sex entered and maintained in Block 1, followed by the MRI variables in Block 2. Including depression in the models did not alter the results. Using this method, models were tested in which all candidate MRI variables (thalamic fraction, third ventricular width, BPF, FLAIR hyperintense lesion volume, and T1 hypointense lesion volume) were used to predict neuropsychological tests. Models were then repeated after controlling for the effects of disease duration. Effects sizes were the difference between means divided by the pooled SD.49 Because of the exploratory nature of the study and limitations in statistical power, the threshold for significance among univariate tests was p < 0.05. All of the dependent measures were examined for approximation to normality and found to be normal by the Kolmogorov–Smirnov test (all p values > 0.05). Two of the MRI measures, T1 hypointense and FLAIR hyperintense lesion volumes, were positively skewed. Because these variables were independent variables in the regression analyses, we elected to not transform these particular distributions. Regarding multicollinearity, the independent variables were intercorrelated but not inordinately so, i.e., no correlations between independent variables exceed 0.85. Regarding linearity, residuals were examined for lack of fit, and there were no clear patterns indicating higher order nonlinearity. Regarding homoscedasticity, we examined normal probability plots and z-residual histograms to assess the assumption of normally distributed residual error at the values of the independent variables, and there were no regression models with marked deviation from a normal distribution of residual error.
RESULTS
MRI variables.
Because the right and left thalamic fractions and raw volumes were highly correlated at r = 0.91 (p < 0.0001), we calculated a mean value for statistical analysis in order to control for Type 1 error. Absolute thalamic volumes were markedly decreased in MS patients (mean 10,324 ± 2,252.3 mm3) vs controls (mean 12,581.9 ± 1,138.7 mm3; p < 0.001). This represented a 17.8% mean difference between groups and an effect size49 (d) of 1.3. This relationship persisted after adjusting thalamic volume for intracranial volume in each subject: thalamic fractions were lower in MS patients (mean 0.0079 ± 0.0017) vs healthy subjects (mean 0.0095 ± 0.00068; p < 0.0001; figure 2 and table 2). This represented a 16.8% mean difference between groups and an effect size of 1.2. This relationship persisted after adjusting for age and sex, and remained significant when right and left thalami were analyzed separately. Thalamic fractions were larger in women than in men in the MS group (women: 0.0081 ± 0.0016; men: 0.0070 ± 0.0019; p = 0.01). This sex difference was not present in normal subjects (p = 0.12). There were no other significant sex-related differences in MRI measures. Age was not related to thalamic fractions in the MS (p = 0.44) and control groups (p = 0.28). The difference in mean BPF between the patient and control groups was 2.6% (p < 0.001, effect size 0.6; table 2).
Figure 2 Thalamic atrophy in MS
Bar heights represent mean and error bars represent SD of thalamic fraction (bilateral thalamic volume normalized to intracranial volume) in the healthy controls (n = 16, open bar) and multiple sclerosis (MS) group (n = 79, hatched bar) (p < 0.0001).
Table 2 MRI data on MS patients and control subjects
The thalamic fractions and other MRI measures were moderately to strongly intercorrelated (table 3). The strongest Pearson correlation was between thalamic fraction and BPF (r = 0.718, p < 0.0001; table 3 and figure 3). Thalamic fraction was moderately to strongly correlated with lesion measures and third ventricle width (table 3). When comparing all lesion and atrophy-related MRI variables to EDSS score, thalamic atrophy showed the strongest correlation (r = 0.316, p = 0.005), albeit modest (table 3). When comparing all MRI variables, including atrophy variables, with EDSS score, third ventricle width was selected as the only variable remaining in the most parsimonious model identified by the forward selection procedure predicting EDSS score (R2 = 0.125, p = 0.008)
Table 3 Relationship between thalamic atrophy and other MRI data and clinical data in MS patients (n = 79)
Figure 3 Total thalamic fraction correlates with brain parenchymal fraction in the MS group (n = 79)
Pearson r = 0.718, p < 0.0001. MS = multiple sclerosis.
Cognitive performance.
As expected, normal controls performed better than MS patients on all neuropsychological tests, although this difference was only significant for BVMT-R-TR and SDMT (table 4). In the MS group, thalamic atrophy was associated with impairment on tests of processing speed/working memory and visuospatial memory (figure 4).
Table 4 Cognitive data in patients with MS vs controls
Figure 4 Scatter plot of thalamic fraction and neuropsychological test score in 31 patients with MS
Thalamic atrophy was associated with impaired performance on tests of processing speed/working memory (Symbol Digit Modalities Test [SDMT], black squares; Pearson r = 0.658, p < 0.0001) and visuospatial memory (Brief Visuospatial Memory Test–Total Recall [BVMT], open squares; Pearson r = 0.724, p < 0.0001). MS = multiple sclerosis.
Pearson correlation coefficients between all MRI and cognitive tests are shown in table 5. Thalamic fraction correlated strongest with all cognitive tests as compared with all other MRI variables. However, the other MRI variables also showed moderate to strong correlations with cognitive data. We could not definitively demonstrate that thalamic atrophy had better correlations than the other MRI measures; i.e., when the magnitude of the top three Pearson correlation coefficients for each cognitive test were compared by t test using the method of Blalock,50 none of the comparisons were significant.
Table 5 Correlation between MRI and cognitive variables in MS patients
Regression modeling results are shown in table 6. Thalamic fraction was the only MRI measure that entered and remained in regression models predicting COWAT (R2 = 0.266, p < 0.05), CVLT-II-TR (R2 = 0.451, p < 0.01), CVLT-II-D (R2 = 0.497, p < 0.001), BVMT-R-TR (R2 = 0.549, p < 0.001), BVMT-R-D (R2 = 0.526, p < 0.001), PASAT (R2 = 0.504, p < 0.001), and SDMT (R2 = 0.514, p < 0.001). The model predicting JLO included both thalamic fraction and third ventricle width (R2 = 0.581, p < 0.001). Repeating the analyses controlling for depression (BDI-FS score), age, sex, and disease duration did not change any of the results. Thus, thalamic fraction accounted for the most variance in all models predicting neuropsychological test performance.
Table 6 Regression modeling examining relationships between MRI and cognitive variables in MS patients
DISCUSSION
Our study showed a significant decrease in thalamic size in MS patients relative to healthy controls. This relationship was seen for raw (18.7% difference) and normalized thalamic volume (thalamic fraction) (16.8% difference), both of which showed large effect sizes. Thalamic size correlated strongly with brain parenchymal fraction, FLAIR and T1 lesion volume, and third ventricular width in MS patients. Modest but significant correlations were seen between thalamic volume and EDSS scores. The most interesting observation was that thalamic atrophy accounted for a large amount of variance in predicting cognitive performance in patients with MS, and entered and remained in regression models more frequently than all other MRI variables, including conventional T1 and T2 lesion measures, whole brain atrophy, and third ventricle width. However, the other MRI measures also showed moderate to strong correlations with cognitive performance. We could not definitively demonstrate that thalamic atrophy had better correlations than the other MRI measures. Nonetheless, these results highlight the significance of thalamic volume loss in MS patients.
Our findings of 16.8% decrease in thalamic volume support previously published results of 17% to 25% lower thalamic volumes in MS patients.8,51 The degree of thalamic atrophy is similar to previously reported substantial and selective atrophy of other deep gray matter structures in MS patients, such as caudate nucleus, in which we reported a 19% lower normalized bicaudate volume in MS patients compared with controls.16
Thalamic fraction correlated strongly with whole brain atrophy (BPF) in our study (r = 0.718, p < 0.0001). However, the absolute difference in BPF between the patient and the control groups was less than 3%, with only a moderate effect size, suggesting that thalamus may be disproportionately vulnerable to the destructive processes in MS. These results agree with data showing selective atrophy of the caudate nucleus in MS16 and progressive loss of gray matter, in particular deep gray structures, in patients with relapsing–remitting and secondary progressive MS.18 MRI-histologic postmortem correlation showed a 22% reduction of whole thalamic volumes and a similar reduction in mean neuronal density in MS patients compared with controls.8
There are several potential explanations for preferential loss of thalamic volume compared with whole brain volume, ranging from biologic to technical factors. The thalamus has rich reciprocal connectivity with much of the brain and might be particularly susceptible to hypometabolism and wallerian degeneration due to demyelination and axonal loss in cerebral white matter. This is supported by an observation that hypometabolism in the thalamus measured by PET showed a significant association with white matter lesion burden in patients with MS.27 In addition, reduction of N-acetylaspartate in the thalamus correlated with reduction of N-acetylaspartate in the normal-appearing frontal white matter.51 Consistent with these observations, we found a moderate relationship between thalamic atrophy and white matter lesion volume in the present study.
The thalamus might also suffer direct damage such as iron deposition or MS plaque formation. One study showed that thalamic T2 hypointensity, a proposed marker of iron deposition, predicts subsequent whole brain atrophy early in the disease course in patients with relapsing–remitting MS.26 Thus, one putative mechanism for thalamic damage is free radicals and lipid peroxidation related to high levels of iron. Demyelinating plaques may be found in the deep gray matter, including the thalamus.9,52,53 These lesions may be focal and discrete or may affect up to one third of the thalamus. Demyelinating lesions in the gray matter as opposed to white matter are thought to be relatively devoid of lymphocytic inflammation but show prominent neuronal loss,9,53,54 making them potentially difficult to detect on conventional MRI scans.55,56
One needs also to consider the potential effects of measurement error affecting our results. Our semiautomated measure of thalamic volume showed much lower reproducibility than our semiautomated measure of whole brain volume. It is likely that the relatively poor reproducibility of the thalamic segmentation was related to difficulty identifying the borders of the thalamus, such as the delineation from the capsula interna, and anterior and posterior edges. This would probably be even more problematic in the MS group, presumably because of disease-related changes in the thalamus and adjacent tissues as reflected in a higher intrarater COV than in the normal control group (leading to a segmentation bias). However, the differences in thalamic volume in MS vs controls exceeded the variability, and the effect sizes were larger than for BPF. Thus, our method likely detected truly increased sensitivity of the thalamic vs whole brain atrophy measure despite the technical limitations. Future studies using automated segmentation of the thalamus and other individual gray matter structures are warranted to confirm and extend our findings.
Our findings agree with previous work8,51 showing that thalamic volume is significantly inversely correlated with third ventricular width in MS patients. Previous studies indicate that third ventricular width is related to cognitive impairment in MS.4,22 However, in the present study, while both variables showed moderate to strong correlations with cognitive performance, regression modeling suggested that thalamic volume was even more closely related to cognitive impairment than was third ventricular width.
Our sample size is small, and the relationships found should be considered preliminary and require replication. In particular, future studies should test a larger sample of patients with MS and normal controls with MRI and cognitive testing to evaluate more completely the relationship between thalamic atrophy and cognitive dysfunction. Furthermore, although the MS patients were mildly to moderately impaired compared with normal subjects on tests of visual memory (BVMT-R) and processing speed (SDMT), the difference in performance on the remaining cognitive tests was not significant (although it showed a trend toward impairment). A larger patient sample or patients with more severe cognitive impairment would have allowed for statistical power to detect medium-size effects (d = 0.48 to 1.01)49 such as seen in our sample. Thalamic volume accounted for the main variance in predicting neuropsychological test performance, indicating a specific relationship between cognitive function and thalamic atrophy. However, partial correlation coefficients for other MRI variables, particularly third ventricular width and BPF, also indicate moderate to strong correlations with neuropsychological function. The observed significant association between thalamic atrophy and cognition explains only 50% of variance in cognitive impairment. One must consider the possibility that the degree of atrophy may not exceed an individual patient's brain reserve capacity. Adaptive mechanisms such as recruitment of secondary neural pathways would limit the association between structural damage and clinical status early in the disease course. It is likely that integrity of other circuits and structures, not explored in this study, also contribute to cognitive function among our patients. Further studies are warranted to compare the correlation with cognitive function of gray matter atrophy in individual structures such as the thalamus to diffuse occult damage in the white matter or gray matter with techniques such as magnetization transfer imaging,57 diffusion tensor imaging,58 magnetic resonance spectroscopy,59 or other new techniques.52
There are several plausible reasons for the link between thalamic atrophy and cognitive dysfunction in MS. The thalamus is an integral component of the limbic system and Papez circuit. It consists of five functional classes of nuclei that subserve memory, emotion, attention, arousal, mood, motivation, and language modulation.60 Vascular and inflammatory lesions that involve the thalamic nuclei in various combinations produce unique sensorimotor and behavioral syndromes.61 A wide range of cortical or subcortical behavioral syndromes may be mimicked by isolated strokes in various thalamic vascular territories.62 Dysexecutive syndrome and poor cognitive planning, among other phenomena, are common features of cognitive impairment associated with injury to the thalamus. A PET study showed a correlation between thalamic hypometabolism and cognitive impairment in patients with MS.27
We report mild correlations between thalamic volume and neurologic disease severity scores. The EDSS is biased heavily toward motor performance, whereas relatively little weight is given to sensory impairment or cognitive disability. A recent study showed no correlation between thalamic magnetization transfer ratio and physical disability in MS patients.30 Our previous work also failed to show a relationship between thalamic damage (as assessed by diffusion imaging) and EDSS score or disease duration.28 Anatomically, the thalamus is not involved in generating or sustaining motor function. Its role in motor control is best described as functional modulator. It is not surprising, therefore, that thalamic involvement in MS is only weakly related to physical disability.
Our study furthers our understanding of mechanisms of MS-related cognitive dysfunction and suggests that thalamic atrophy is a clinically relevant biomarker of the neurodegenerative disease process in MS. These findings should continue to fuel the growing interest in uncovering the mechanisms behind gray matter involvement in MS.9
ACKNOWLEDGMENT
The authors thank Ms. Sophie Tamm for assistance with manuscript preparation and Dr. Ashish Arora and Dr. Venkata Dandamudi for technical assistance. The authors are also grateful to Gary Cutter, PhD, and Diane L. Cookfair, PhD, for statistical consultation.
Footnotes
-
Supported in part by research grants from the NIH (NS42379-01—Dr. Bakshi) and National Multiple Sclerosis Society (RG 3574A1 and RG 3705A1—Drs. Bakshi and Guttmann) and a Clinical Investigator Training Program from Harvard/Massachusetts Institutes of Technology Health Sciences and Technology, Beth Israel Deaconess Medical Center, Pfizer, and Merck & Co. (Dr. Houtchens).
Disclosure: The authors report no conflicts of interest.
Received December 27, 2006. Accepted in final form April 16, 2007.
REFERENCES
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.