MR Lesion Load and Cognitive Function in Patients with Relapsing-Remitting Multiple Sclerosis

Multiple Sclerosis (MS) is a demyelinating disease most often associated with progressive physical impairment; however, its effects are noted to extend beyond measures of physical disability. As early as 1877, it was recognized by Charcot that cognitive deterioration often accompanied MS. Cognitive impairment occurs in 40% to 70% of patients with MS (1−4). The estimate of the prevalence of cognitive dysfunction is diverse because of the widely varied populations studied and the differing methods of evaluation used. Disabilities in cognition usually take the form of difficulties with executive function, which includes working memory, problem solving, initiation and inhibition of responses, conceptual ability, strategic planning, and attention deficits. Long- and short-term memory and visuospatial perception are also often affected. Previously thought to be only a feature of advanced disease, cognitive dysfunction may actually begin during the early stages of disease and in the absence of physical disability (5). Recent studies indicate that cognitive impairment is not correlated with age, the level of physical disability, disease duration, disease type, or performance on standardized brief mental status examinations (6, 7). Instead, cognitive dysfunction seems to occur on a patient-specific basis.

Previous studies have attempted to show a correlation between MR imaging and the physical manifestations of MS as evaluated through the Kurtzke Expanded Disability Status Scale (EDSS) or the Ambulation Index (AI) (8). Nevertheless, there seems to be little or no correlation between the physical disabilities and lesion burden as reported by MR imaging (9−12). It has been suggested by Comi et al (4) that clinical status may be better correlated with spinal cord damage than with demyelinating plaques in the brain. Lesions in the brain may instead prove to correlate with neurocognitive dysfunction. Investigators have found a strong connection between the total surface area of MS plaques on MR images and cognitive abilities (13, 14). Other associations have been found between corpus callosum atrophy and ventricular dilation and cognitive dysfunction (15, 16). Nonetheless, little is known about the relationship between total lesion volume, as measured on proton density–and T2-weighted images, and neurocognitive ability. Assessing lesion burden in this manner may provide an important correlation with global disability.

This study examined the relationship between total lesion volume, measured by a highly reproducible validated semiautomated computerized technique, and neurocognitive and physical disability in a group of patients with untreated clinically definite relapsing-remitting MS. The neuropsychological tests conducted as part of this evaluation include batteries of tasks that involve working memory, attention, and information-processing speed. These tests are particularly sensitive to cognitive impairment and for predicting subsequent decline in MS. We also administered measures of visual and verbal anterograde memory and a small number of additional measures that have been found to be impaired in limited numbers of patients with MS, such as language and visuoperceptual functioning. As part of our analysis, we examined the lesion volume for each patient in our cohort and their EDSS and AI scores. We also evaluated the neurocognitive tests over time. This included repeat measurements of these tests. It must be noted that because there was no statistical correlation of these tests, no Bonferonni correction was indicated.

Methods

Results

Table 1 shows the relevant patient parameters included in this study. For each patient, we list the EDSS and AI scores at entry and exit from the study along with total T2 lesion volume of the entire brain at both points. The total T2 lesion volume was compared with the neurocognitive test results and clinical test results for each patient over the duration (Table 2). Of 12 neurocognitive parameters tested, two correlated significantly with T2 lesion volume. The Symbol-Digit Modalities test, which analyzes information-processing speed, was significant, with a P value of .0204. The correlation of the fifth trial of the Rey Auditory Verbal Learning test, which tests verbal long-term memory, was also significant, with a P value of .0348. None of the other 10 neurocognitive examinations, however, which cover other tests of executive function, information-processing speed, and verbal short-term memory, showed a significant correlation with total lesion volume.

Furthermore, the T2 lesion volume also did not correlate with EDSS (P = .68) or with AI (P = .95). Table 2 summarizes the correlations between the lesion load measurements and the neurocognitive as well as physical tests.

Discussion

Although it has been reported that secondary progressive MS causes more frequent and severe cognitive deficits than relapsing-remitting MS, patients with relapsing-remitting MS have also reported a decline in cognitive function (30, 31). It is interesting to note that our group found no significant change in neurocognitive examination data over the duration of the study. Additionally, our results reveal a significant correlation between clinically definite relapsing-remitting MS and performance on only two of 12 standardized neuropsychological tests. The tests reflect performance of executive function and learning. There may be several possible explanations for this result. The neuropsychological testing may be relatively insensitive, T2 lesion volume is nonspecific with a spectrum of pathologic abnormalities appearing similar (high signal on T2-weighted images), and lesion location, which we did not assess, may be important. Some or all of these variables may act together to influence cognitive disability.

MS lesions are heterogeneous histopathologically, consisting of areas with gliosis, demyelination, remyelination, edema, and neuronal loss. The relationship between high-intensity abnormalities on MR images and the pathologic substrate is not fully understood. Several recent publications, however, have suggested that T2-weighted images are not adequately sensitive to predict the pathologic basis of hyperintense lesions (32−34). More sensitive imaging techniques, including MR spectroscopy, magnetization transfer, and T1-weighted hypointensity measurements may produce more robust correlations with neurocognitive data and other disability measurements (35, 36).

The ultimate effect of white matter disease on the neuron is still in question. Using MR spectroscopy, it is possible to observe a decrease in the ratio of N-acetylaspartate to creatine with severe acute or chronic demyelination. In their 1994 study, Arnold et al (33) found statistically significant changes in N-acetylaspartate:creatine ratios without statistically significant changes in total MR imaging lesion volume. This indicates that MR spectroscopy may be a more sensitive index of disease progression and may yield more significant correlations than does MR imaging. Furthermore, normal-appearing white matter on T2-weighted images may contain microscopic changes that are significant in terms of brain function. Narayanan et al (34) compared average metabolite levels and T2-weighted images, noting that diffuse axonal volume loss or dysfunction extends beyond areas of inflammation and could be caused by microscopic disease or wallerian degeneration. T1 and T2 relaxation times in normal-appearing white matter were reported to be significantly abnormal in patients with MS compared with those normal participants (37−39). Other studies using magnetization-transfer imaging have also revealed abnormalities in normal-appearing white matter (40, 41).

Lesions that appear similar on T2-weighted images may have profoundly different effects on neuronal function and neurologic performance. Comi et al (11) believe that cognitive dysfunction seems to be determined by the disruption of neuronal connections in the cortical associative areas and those areas between the cortical and subcortical structures. This interference seems to be induced by demyelination and axonal degeneration. Thus, lesions that are similarly hyperintense on T2-weighted images can have variable histopathologic substrates and potentially very different effects on cognitive performance.

Our studies indicate that T2 lesion volume in cases of MS is correlated to several specific markers of cognitive function (Symbol-Digit Modalities test and Rey Auditory Verbal Learning test), but the T2 lesion volume is not predictive of global cognitive dysfunction. In addition, we did not find a relationship between lesion load and disability as measured by EDSS or AI. This also may reflect the lack of histologic specificity of T2 lesions, the insensitivity of the EDSS system to subtle changes in neurologic function, and the reliance on intracranial lesion measurements exclusive of the spinal cord, which has a predominant role in motor disability (a significant portion of EDSS or AI) in cases of MS. The lack of better correlations may also be due in part to the relatively small number of patients examined in this particular study or to the fact that our participants were limited to patients with relapsing-remitting MS. In their 1989 article, Rao et al (13) suggested that a critical threshold of lesion load must be surpassed for cognitive dysfunction to be detected. In our series of patients, it may be that those who had lower lesion loads were experiencing some cognitive dysfunction that was missed by our test battery. It has also been suggested that lesion load may be less important than lesion location when determining cognitive disability (38). A small lesion in a critical area could do more damage than a large lesion in a less eloquent area of the brain. Because we did not segregate lesions in terms of location, our results do not reflect this possibly important differentiation.

We recognize that clinical function in patients with MS is highly variable. Therefore, as has been suggested by Beatty et al (31), correlating a unified neuropsychological profile to T2 lesion volume in patients with MS may be as difficult as correlating other clinical parameters, such as EDSS and AI (31). We think that more specific global measurements of disease may offer more precise correlation with cognitive dysfunction and other disability parameters.

Conclusion

Our results indicate that T2 lesion volume does not seem to be a robust surrogate marker of neuropsychological impairment in patients with MS. We think that global measurements of parameters that are more specific to the disease process may offer more precise correlation with cognitive dysfunction and other disability parameters.