Memory Dysfunction in Multiple Sclerosis Corresponds to Juxtacortical Lesion Load on Fast Fluid-Attenuated Inversion-Recovery MR Images

MR imaging provides a sensitive diagnostic tool and paraclinical marker of disease activity and prognosis in multiple sclerosis (MS) (1, 2). The role of MR imaging in the investigation and management of MS is nonetheless controversial. MR imaging is used to confirm a suspected diagnosis of MS or to exclude other possible causes of a patient's symptoms. In these circumstances, conventional spin-echo (CSE) sequences provide adequate information, revealing established plaques of demyelination with high conspicuity. CSE sequences have also been widely used to document lesion load in clinical research and therapeutic trials (3, 4). The increased speed of modern scanners allows CSE techniques to be combined with the collection of fast spin-echo (FSE) images to allow higher in-plane spatial resolution and thinner sections, which further increase the lesion load detected.

Despite the high sensitivity of CSE and FSE sequences to MS lesions, attempts to correlate patient disability, symptom expression, and disease progression have been disappointing (5). This has led many groups to investigate other MR techniques that might detect lesions with stronger symptomatic and functional correlations (6).

De Coene et al (7) described the use of a fluid-attenuated inversion-recovery (FLAIR) MR imaging sequence to increase the sensitivity of spin-echo sequences to MS lesions. The FLAIR sequence uses an initial 180° inversion pulse followed by a normal spin-echo sequence that is performed at the null point of water. The resulting suppression of CSF signal not only improves the detection of lesions at CSF surfaces but also allows the use of far higher levels of T2 weighting, with resulting increases in contrast between lesions and normal tissue. The combination of FLAIR and FSE sequences offers improvements in scan time and resolution, allowing the routine use of high-resolution fast FLAIR imaging in diagnostic protocols (8). Rydberg et al (9) described a further modification of the FLAIR sequence designed to optimize the contrast between MS lesions and normal white matter. This optimized fast FLAIR sequence depicts lesions in the prosencephalon that are not apparent on CSE images (10, 11), but it has been shown to be inferior in detecting lesions in the brain stem and cerebellum (12, 13). This observation has cast doubt on the value of the fast FLAIR sequence, and it has been argued that brain stem lesions seen on CSE sequences are likely to be more strongly related to dysfunction in MS patients (14). In practice, the correlation between lesion load on CSE images and clinical disability as measured by Kurtzke's Expanded Disability Status Scale (EDSS) or ambulatory index (AI) remains poor (6, 15–17). One possible reason for this poor correlation is that these methods of assessment are rather blunt clinical instruments, designed to document disability that is likely to interfere with independent daily living (18). Consequently, both the EDSS and AI are disproportionately weighted toward motor disability.

Cognitive problems are characterized by subcortical dysfunction and are a recognized feature of MS, providing an additional indicator of disease severity (19). Previous workers have found associations between lesion load and cognitive deficit (20–23) and have correlated specific lesion sites with specific deficits in cognitive function (24–29).

The aim of this study was to compare cognitive dysfunction with the extent and distribution of lesion load exhibited on CSE and fast FLAIR imaging sequences. The study was conducted in a cohort of 18 patients with relapsing-remitting MS. Clinical assessment included EDSS and a comprehensive battery of tests of general psychology, memory, and frontal lobe executive function.

Methods

Results

Eighteen of the 20 patients completed the MR imaging and neuropsychological components of the study. The mean EDSS score in women was 3 (range, 1.5–4.5); in men, the mean score was 3.5 (range, 2.5–5.0).

The results of the MR study are shown in Table 1. A comparison of the total number of lesions on CSE and fast FLAIR images revealed a significant increase in small (P < .01) and medium-sized (P < .02) cortical lesions on the fast FLAIR images. This difference was reflected in comparisons of the supratentorial and whole-brain lesion loads, which also revealed significantly (P < .01) more lesions on the fast FLAIR images. Juxtacortical lesions were seen in all 18 patients, but lesions not seen on CSE images were identified on fast FLAIR images in 15 cases. The majority of juxtacortical lesions were at the gray/white matter interface. Most were very small and were typically seen in groups around the base of a sulcus. In many cases, the lesions were ovoid in shape, lying parallel and perpendicular to the gray/white matter interface (Fig 1).

• Download figure
• Open in new tab
• Download powerpoint

fig 1.

A–F, Matching lesions on proton density–weighted (A), T2-weighted CSE (C) and fast FLAIR (E) images. 1 denotes a population of lesions in the gray matter, extending to the juxtacortical boundary, which are evident on fast FLAIR but not on proton density—or T2-weighted CSE images. These lesions are further demonstrated on the magnified (×2) proton density–weighted (B) T2-weighted CSE (D), and fast FLAIR (F) images. 2 indicates a population of lesions that are less conspicuous on proton density–and T2-weighted CSE images (C). 3 relates to a medium-sized white matter lesion that is obvious on all imaging sequences.

Small cortical lesions (0–5 mm) depicted only on the fast FLAIR images correlated significantly with reduced performance in the fifth trial (P = .007) and in the delayed trial (P = .005) of the RAVLT (Table 2). Both tests assess delayed verbal memory function. Medium-sized cerebellar lesions seen on both fast FLAIR and CSE images correlated with the second trial of the RAVLT (P < .001).

Correlations between EDSS and lesion scores are shown in Table 3. Small lesions in the cerebellum, seen on both sequences, correlated with the cerebellar FS score (r = .6, P = .008). Small cerebellar lesions seen on CSE images also correlated with EDSS score (r = .63), although lesions seen on fast FLAIR images did not. Small periventricular lesions seen on CSE (r = .7, P = .001) and fast FLAIR (r = .7, P = .001) images correlated with the bladder dysfunction component of the EDSS. Confluent periventricular lesions seen on CSE (r = .64, P = .004) and fast FLAIR (r = .64, P = .004) images correlated with the brain stem disability component. Large (> 10 mm) discrete lesions depicted on CSE images correlated with the pyramidal disability component (r = .59, P = .001).

Repeated lesion analysis scored nearly 400 lesions for the comparative six-patient data sets. Analysis of agreement found any variation in lesion score to lie within 2 SD of the mean (Fig 2). The data point that can be identified in Figure 2 as having a difference of 17 and a mean of 105 is the score for small cortical lesions (0–5 mm). This reflects the fact that the second observer identified more small cortical lesions than the initial observers did (112 versus 95).

• Download figure
• Open in new tab
• Download powerpoint

fig 2.

Bland and Altman plot shows the differences against mean for interobserver variation data.

All analyses performed in the study used the lesion scores from the first scoring. Correlation analysis found no significant effect (P > .05) relating to an increase in the size of the values. Furthermore, Wilcoxon matched pairs analysis found no significant difference (P > .05) between the two sets of lesion scores.

Discussion

Cognitive dysfunction is evident in up to 65% of MS patients (38) and is known to affect patients from an early stage of the disease (24). MS sufferers consider cognitive dysfunction to be more devastating than physical disability because of its dramatic effect on relationships with friends and family (39). Since the association between T2 lesion scores and clinical disability as assessed by EDSS is limited, an alternative strategy is to relate pathologic changes evident on MR images to neuropsychological function.

The fast FLAIR sequence has received considerable attention and has been widely used in the study of MS (14). It has consistently been found to be superior to CSE or FSE in depicting lesions in the prosencephalon but inferior in the posterior fossa and cervical cord (10, 11). It has the greatest sensitivity to active lesions when unenhanced images are used (38) and is the most reliable sequence for lesion quantification when semiautomated techniques are employed (40). This study confirms the findings of previous investigators that fast FLAIR images show more lesions in the prosencephalon than do CSE sequences. The increase in the number of lesions detected is largely due to the demonstration of small lesions in the cortex at the cortical/subcortical boundary. The detection of these small lesions is difficult, with an 18% increase in identification by the second observer in this study; however, it is important, as they constitute nearly 20% of the total lesion load (41).

Lesions in the cortex and at the cortical/subcortical boundary may affect the function of U fibers, which provide interconnections between cortical areas. Miki et al (29) recently investigated the effect of U fiber lesions in MS and found that patients with multiple U fiber lesions performed significantly worse on cognitive tests of memory function than did unaffected persons. These observations were based on the demonstration of small numbers of cortical or juxtacortical lesions on CSE images in 53% of patients. In the present study, no correlation was found between memory dysfunction and juxtacortical lesion load on CSE images, possibly reflecting the smaller patient sample or a methodological difference in the lesion identification technique. The appearance of these small lesions is nonetheless distinctive. They are commonly ovoid and perpendicular to the gray/white matter interface, often at the base of major sulci. These appearances agree with the common histologic finding of perivascular inflammatory infiltration around penetrating cortical vessels (42). Use of the fast FLAIR sequence increased the number of cortical and juxtacortical lesions seen by over 100%. The number of small cortical lesions depicted on fast FLAIR images correlated closely with impairment on cognitive tests of delayed memory.

Previous researchers included juxtacortical lesions in a four-parameter model that best predicts conversion to clinically definite MS at first presentation (43), suggesting that juxtacortical lesions have prognostic implications not associated with some other, more commonly quantified lesions. This is supported by the finding that verbal memory was the only cognitive function that showed continuing decline over a 4-year follow-up period (44) and that did not improve after remission in mildly impaired MS patients (45). If juxtacortical lesions do represent a correlate of progressive neurologic deficit that does not remit between attacks, it seems reasonable to hypothesize that they are associated with permanent neuronal damage. In fact, lesions that are conspicuous on T2-weighted images are known to contain transected axons (46). In the juxtacortical area, this type of axonal injury would affect the dense interconnecting cortical U fibers, causing cortical disconnection syndromes. This is supported by the finding that cerebral atrophy can be detected in MS patients and that the rate of progression of atrophy is linked to the rate of increasing disability in MS (47).

The disruption of longer cortical association fibers may also play a part in the impairment of delayed memory demonstrated in this study. The hallmark of memory impairment in subcortical disease is inefficient retrieval of information in the presence of near normal learning ability. Memory performance in this study was consistent with this model, as we found no correlation between early verbal memory tasks and lesions, but strong links with delayed memory tasks.

Although the significance of the anatomic distribution of juxtacortical lesions was not examined in the current study, evidence exists of a direct relationship between lesion sites and memory dysfunction. Lesions of the deep white matter in the left parietal lobe have been found to be associated with impaired performance on the paired associates test, a test of learning and memory (24). This suggests disruption of the arcuate longitudinal fasciculus, which sweeps through this region in a large bundle. It connects the superior and middle frontal gyri to large sections of the frontal lobe, possibly affecting the ability of the brain to encode novel verbal associations. The combined gray/white matter lesion load in the left frontal lobe has also been shown to correlate with reduced performance on memory tasks (25). Temporal lobe lesions in the corona radiata, internal capsule, insula, and hippocampus have also been linked to memory problems (27). Poor memory retrieval is characteristic of the subcortical dysfunction evident in MS. Further dysfunction is evident by assessment of problem solving and abstract conceptualization, processing of information, and an absence of aphasia, agnosia, or apraxia, except in isolated cases (24).

In addition to the correlation between juxtacortical lesion load and memory dysfunction, correlations were also found between medium-sized cerebellar lesions and immediate verbal memory tasks. Functional studies of memory have identified hypometabolism in the cerebellum of cognitively impaired MS patients (48). Although these changes may simply represent cerebellar diaschisis, the role of the cerebellum in higher cognitive functioning remains unclear (49). The number of cerebellarlesions in the present study was small, and it is possible that this observation would not be substantiated in larger studies.

Further relationships were found between individual functional system scores and lesions identified on both sequences and on the sequences in isolation. The individual functional scores are relatively high in comparison with established work in the area, and this is most probably explained by the methodological differences, such as the differing disease durations, the different number of patients, and the differences in lesion assessment between such cross-sectional studies (21). No correlation was found between the overall EDSS score and any lesion score. This highlights the variability in results of lesion/clinical disability analysis and reinforces the need for alternative methods of assessing the effect of pathologic change.

Conclusion

We have demonstrated a relationship between small juxtacortical lesions and memory dysfunction in patients with relapsing-remitting MS. These lesions are not well seen on CSE sequences and their accurate demonstration requires the use of a fast FLAIR sequence. There is growing evidence in the literature to suggest that cognitive dysfunction of this type is an important, though underused, marker of morbidity in MS. Other studies have suggested that juxtacortical lesions represent an important prognosticator of disease progression. The insights drawn in this study must be tempered by recognition of the small patient group used. Nevertheless, on the basis of our own and previous findings, we believe that the use of the fast FLAIR sequence is potentially important, as it depicts areas of pathologic processes that have been found to relate to clinical disability scores and, in particular, to neuropsychological function.