Cognitive impairment occurs in 40%–65% of patients with multiple sclerosis (MS),
1 including those with clinically isolated syndromes (CIS).
2 Information-processing speed, attention, memory, and executive functions are commonly impaired.
3 Correlating magnetic resonance imaging (MRI) abnormalities with measures of cognitive functioning is important both in elucidating pathological mechanisms underpinning cognitive impairment and in predicting early in the disease those likely to develop cognitive impairment later.
4–6 Focal inflammatory lesions on conventional MRI correlate only modestly with cognitive impairment,
1 and the role of diffuse pathology in normal-appearing brain tissue is increasingly recognized. Several cross-sectional studies have reported that measures of atrophy,
7,8 changes in magnetization transfer ratio (MTR),
9,10 and in 1H-MRS (magnetic resonance spectroscopy) metabolite concentrations
11 are more closely correlated with cognitive impairment than lesion metrics. The contribution of gray-matter pathology to physical and cognitive disability
12 is increasingly recognized, and gray-matter atrophy has been described in association with cognitive impairment.
13In a previous study, using imaging data obtained shortly after the initial CIS, we reported that T1- and T2-weighted and Gad-enhancing lesion metrics and increase in myo-inositol concentration in normal-appearing white matter (NAWM) were the best predictors of cognitive impairment in patients assessed 7 years later, whereas the volume of the gray matter was not an independent predictor.
5 Using the same longitudinal cohort, we examine here the associations between a wide range of imaging parameters and neuropsychological data obtained contemporaneously, 7 years after the initial CIS, using a cross-sectional design. We aimed to explore whether the imaging parameters associated with cognitive impairment in this cross-sectional study were the same as those from baseline imaging that predicted future cognitive impairment in our earlier study.
5Results
At follow-up, median EDSS was 1 (range: 0–6); 59 patients were able to work; 58 lived independently; and 1 was in residential care; 10 patients were receiving, or had previously received, disease-modifying treatment (DMT), and 3 were on antidepressant medication.
Of the 54 patients for whom ratings of anxiety and depression were obtained, 16 (29.6%) scored above the cut-off point for anxiety, and 4 (7.4%) were over the cut-off point for depression. All patients were included in the analysis, but we controlled for HADS scores.
There were no differences between patients whose diagnosis was still CIS and those who had converted to MS in age, years of education (overall mean: 13.8 years; SD: 3.10), premorbid IQ (overall mean: 104.2; SD: 14.0) or depression ratings (overall mean: 4.87; SD: 3.83), but those with MS had higher EDSS (median [interquartile range, or IQR]: MS: 1.5 [1], CIS 1 [
1]; p <0.001) and anxiety ratings (median [IQR]: MS: 9.5 [6], CIS 5.5 [4.5]; p <0.05).
Pattern of Cognitive Impairment
Neuropsychological results are detailed in
Table 1. Fifty-seven patients had completed most or all tests. Of these, 29 (50.9%) were not impaired on any test, and 17 (29.8%) scored below the 5th percentile on at least two tests and were considered to be “cognitively impaired.” Compared with the normative data, more patients than expected had significant Verbal and Performance IQ decline and were impaired on tests of attention/speed of information-processing (PASAT and SDMT) and executive functioning (Brixton Test and a trend for Hayling Test; see
Table 1 for proportions impaired and significance levels).
Four out of 22 CIS patients (18.2%) were “cognitively impaired,” versus 13 out of 35 MS patients (37.1%); this difference failed to reach statistical significance (χ2=2.3205; NS). Cognitive scores were lower in MS than CIS patients, but these differences also failed to reach statistical significance. Older age at disease onset predicted poorer performance on the Brixton Test (p=0.018) and on the delayed Story Recall test (p=0.043).
MRI Correlates of Cognitive Performance
Only performance on the SDMT, decline in verbal IQ, and overall cognitive performance were predicted by MRI parameters independently of demographic covariates (age, gender, premorbid full-scale IQ, and level of education;
Table 2). For the SDMT, a model incorporating the number of T2 lesions, GM volume, premorbid IQ, and educational level explained 63% of the variance. For Verbal IQ decline, the best model included the number of T2 lesions and female gender; this model only explained 14% of the variance. For the overall cognitive performance index, a model including number of T2 lesions, male gender, and premorbid IQ explained 26% of the variance.
T2 lesion metrics were the most significant MRI parameters in all three models, but it was not possible to separate the independent effects of number, volume, or location of lesions, as these variables were closely related. T1 lesion metrics were not independently associated with cognitive performance after controlling for T2 lesion metrics and/or demographic variables. 1H-MRS metabolite concentrations, MTR changes, and WM volume loss were not associated with cognitive performance.
Discussion
The salient finding of this cross-sectional study of patients examined 7 years after their initial CIS is the association between cognitive deficits, in particular attention and speed of information-processing, and T2 lesion metrics and loss of gray-matter volume in contemporary imaging. Changes in white-matter volume, MTR, or 1H-MRS metabolite concentrations were not significantly associated with the presence of cognitive impairment.
Two-thirds of our patients met criteria for MS 7 years after experiencing a CIS, in keeping with the findings of several longitudinal studies,
28,29 and, therefore, our sample can be seen as representative of the natural history of the disease. Patients who had not converted to MS did not differ significantly in their cognitive performance from those who had, although they tended to perform better, and fewer were “cognitively impaired.” This finding is in keeping with our earlier report in a different patient cohort,
30 and the presence of cognitive impairment in CIS patients is the more likely explanation.
31One-third of our patients were cognitively impaired, by use of standard criteria,
2,10 and their cognitive profile was similar to that described by others,
2,4,32 with impairment in attention/speed of information-processing and executive functions. This pattern has been attributed to disturbed connectivity resulting from white-matter damage,
33 as these cognitive functions require the rapid transfer of information between different cortical regions along white-matter tracts. Decline in Verbal and Performance IQ was also present in our patients, a finding previously considered to be uncommon in MS populations,
34 but in keeping with a recent meta-analysis reporting intellectual decline of a small-to-moderate effect size.
32 High premorbid IQ and educational history predicted better cognitive performance, suggesting, as in our previous study,
5 that “cognitive reserve” influences compensatory reorganization of cognitive networks.
In our previous study of an almost-identical patient sample,
5 cognitive performance at the 7-year follow-up was predicted by T1 and T2 lesion metrics, but not by the loss of gray-matter volume, measured at the time of the initial CIS. By contrast, in the cross-sectional study presented here using imaging data obtained at the 7-year follow-up, loss of gray-matter volume was more closely, and independently, associated with cognitive performance. The reasons for the differences between these two studies are unclear. Gray-matter atrophy reflects demyelination and neuroaxonal damage due to intrinsic cortical pathology and/or secondary to white-matter damage,
12 and it has been observed to increase over time
35 and to be associated with impaired cognitive performance early in the disease
13,36–38 and to predict early conversion to MS.
39 A possible explanation for our findings is that whereas white-matter lesions are the main pathology in CIS, gray-matter pathology becomes increasingly evident over time and increasingly contributes to cognitive impairment. The relative contribution of primary and secondary gray-matter pathology to cognitive impairment remains to be determined.
1H-MRS metabolite concentrations were not associated with contemporaneous cognitive impairment, although, in our previous study,
5 increases in myo-inositol, a marker of glial and inflammatory cells, shortly after the initial CIS, predicted poor executive functioning years later. There is no obvious explanation for this discrepancy, although serial metabolite concentrations appear to be a poor indicator of progressive pathology in relapsing–remitting MS.
40The major limitation of our study is that patients in this longitudinal cohort were not tested neuropsychologically when recruited after their initial CIS, and, therefore, it has not been possible to compare baseline cross-sectional MRI–cognitive associations with those presented here. It remains possible that gray-matter volume would have been associated with contemporaneous cognitive performance at baseline. Furthermore, we did not control for levels of fatigue, a potential confounder of the association between MRI parameters and cognitive impairment. Future studies using imaging methods more able to detect intrinsic cortical pathology and subtle abnormalities in normal-appearing brain tissue will increase our understanding of the neuropathological mechanisms that determine cognitive impairment over the course of the illness.