Diffusion-weighted MR Imaging of Multiple Sclerosis: Added Clinical Value or “Just Another Pretty Face?”

In this issue of the AJNR, Castriota-Scanderberg et al (page 862) describe the diffusion-weighted imaging (DWI) findings from 10 patients with relapsing-remitting multiple sclerosis (MS), 10 patients with secondary-progressive MS, and 11 control subjects. Orientationally averaged apparent diffusion coefficient (ADC) values (“<D>”, equal to one-third the trace of the diffusion tensor) were computed for selected white matter regions of interest within T2-hyperintense plaques for each of the MS patients, as well as for selected normal white matter regions in the control subjects. Mean <D> values were found to be significantly higher in the secondary-progressive group (1.45 × 10⁻³ mm²/sec) than in the relapsing-remitting (0.95 × 10⁻³ mm²/sec) or control groups (0.73 × 10⁻³ mm²/sec). Also, these values correlated highly with disease duration and disability. This is not surprising, given the two clinical subgroups of MS that were studied. Additionally, for a smaller number of regions selected within MS plaques characterized by both T2-hyperintense and visually evident T1-hypointense signal intensity, a strong inverse correlation was found between the <D> values and the T1 signal intensities. The authors conclude that <D> values can be used “to distinguish between MS lesions of different severity, which are associated with a different degree of clinical disability.”

These results are certainly intriguing, but they beg the following question. Does DWI contribute any new, clinically relevant data about MS that cannot already be determined using conventional T2- and T1-weighted MR imaging? For a new imaging technology, such as a novel pulse sequence, to displace older, more established techniques, it must either do a better job at detection (sensitivity) or diagnosis (specificity) of disease, or do an equally accurate job, but more quickly or less expensively. In the case of MS, DWI may satisfy the requirements of specificity, speed, and, possibly, of sensitivity, by providing visually evident “functional” data not otherwise easily obtainable. Clearly, further work is required to establish fully the role of DWI in the clinical evaluation of MS, but based on the preliminary results of Castriota-Scanderberg et al and others, we can answer “yes” to the question, “Does DWI offer added clinical value?”

MR imaging is the most sensitive test for detecting MS lesions of the craniospinal axis, and has become essential in the evaluation of this disease. The definitive diagnosis of MS, however, continues to be based on a spectrum of findings, the most notable being the occurrence of focal neurologic deficits that vary with time in both degree and location. The caveat that the diagnosis of MS remains primarily clinical cannot be over-emphasized (1). The characteristic MR appearance of MS plaques is that of multiple ovoid, well-circumscribed, T2-hyperintense foci, which may show halos of T2-hyperintense signal probably caused by inflammatory edema. Rarely, such lesions can be quite large, with a pseudotumor-type appearance. Approximately 10–20% of T2-hyperintense MS plaques are also hypointense on T1-weighted images (2). In the acute phase, this probably reflects vasogenic edema without underlying tissue destruction, and may be reversible as inflammation wanes; in the chronic phase, this “black hole” appearance more likely reflects severe, irreversible tissue damage (2). The presence of contrast enhancement suggests blood brain–barrier disruption in acutely inflammatory lesions; without steroid treatment, enhancement may persist for 2 to 6 weeks (2). The extent of the T2 signal abnormalities at initial presentation, together with a history suggestive of demyelination, is strongly predictive of the risk of developing clinically definite MS within the next few years. In established MS, however, the correlation between the extent of the T2 signal abnormalities and disability is modest (1).

Although conventional T2-weighted MR imaging is highly sensitive in the detection of the white matter lesions of MS, it is limited, as is pointed out by Castriota-Scanderberg et al, by its lack of histopathologic specificity. Demyelination, inflammation, edema, gliosis, and axonal loss all may appear as foci of T2-hyperintense signal. These different pathologic entities not only reflect different stages of the disease, but are associated with different prognoses.

MS can manifest clinically in one of two major forms. Relapsing-remitting disease is characterized by repeated, acute bouts of exacerbations or relapses, separated by weeks or months of partial or complete clinical remission (3). The underlying histopathologic process of this form of disease appears to be remodeling of the demyelinated axonal membranes, such that they acquire a higher-than-normal sodium channel density, permitting increased action potential conduction velocity despite their loss of myelin (3). Progressive forms of MS, however, are characterized by an unrelenting downhill course, either beginning with the first clinical presentation (primary-progressive), or after a period of relapsing-remitting disease (secondary-progressive).

The pathologic substrate underlying progressive forms of MS has been elucidated only recently. Suprisingly, although every medical student “knows” MS to be the poster child for demyelinating disease (indeed, the demonstration of slowed nerve-conduction velocities as measured by evoked potential studies—a hallmark of demyelination—remains an important diagnostic feature of MS), MS recently has been proven to have components of both demyelination and axonal transection (3, 4). In a 1998 study using confocal microscopy and computer-based three-dimensional reconstruction techniques, axonal transection was shown to occur commonly in active MS plaques (both acute and chronic), and was postulated to be the pathologic correlate of the irreversible neurologic impairment found in this disease (4).

The idea that progressive axonal loss, in addition to demyelination, may be a feature of MS, is supported by radiologic studies. Early MR spectroscopy investigations using N-acetyl aspartate (NAA) as a neuronal marker showed reductions in cerebellar NAA that correlated with persistent disease progression (5). In a more recent proton MR spectroscopy study, T1-hypointense MS plaques were found to correlate with axonal loss at autopsy and biopsy (6). In this work, NAA concentration correlated highly with the degree of T1–relaxation time prolongation within the spectroscopic voxels. Visually evident T1-hypointense lesions showed a lower concentration of both NAA and creatine compared with deranged but normal-appearing white matter, which showed less severe reductions in NAA only. These findings provide in vivo evidence of axonal damage in severely T1-hypointense MS lesions, and underscore the point that T1 relaxation, in itself, could be an important parameter in monitoring disease progression in MS (6). Thus, the presence of visually detectable, persistent, T1-hypointense signal within an MS plaque appears to have greater specificity than T2-hyperintense signal alone in identifying lesions associated with axonal loss, and therefore, could potentially aid in identifying MS patients with a more severe, progressive clinical course (6).

Other radiologic studies have suggested that the integrated use of “functional” MR imaging techniques, such as magnetization transfer and spectroscopy, might provide a more complete description of the pathologic features of MS than conventional MR imaging alone (7, 8). In a recent AJNR-published study that compared the combined magnetization-transfer and proton-spectroscopic MR imaging results of patients with relapsing-remitting, primary-progressive, and secondary-progressive MS with those of control subjects, the magnetization-transfer ratio (MTR) of normal-appearing white matter in MS patients was found to be significantly lower than that of the control subjects. MS lesions showed a large reduction in MTR, with old lesions exhibiting lower MTR than new lesions. Average lesion MTR and relative NAA concentrations correlated positively in patients with relapsing-remitting MS, and more strongly in regions containing new lesions (8). Importantly, the results of this study, as well as those of previously discussed studies by Trapp et al and van Walderveen et al, suggest that: 1) axonal damage is not exclusively a late feature of MS, and 2) even white matter that appears normal on conventional MR images may be histopathologically deranged. Although a number of the acute imaging changes of MS are reversible, persistent reduction in MR parameters such as NAA concentration, MTR, and T1 signal intensity, suggests the presence of demyelination, irreversible axonal degeneration, or both in many chronic MS lesions (1).

Could the addition of DWI further strengthen this imaging assessment of MS? DWI already has been shown to have great clinical benefit in the radiologic evaluation of acute stroke, as well as in the differentiation of arachnoid cysts from epidermoid tumors, and in the differentiation of epidural abscesses from sterile extraaxial fluid collections. Pilot investigations assessing the role of DWI in the evaluation of MS have shown that, unlike the reduced or “restricted” ADC values found in regions of acute infarction, which reflect the presence of cytotoxic edema, the typical DWI abnormality found in MS plaques is that of truly elevated ADC values (9, 10). In early studies, this increased diffusivity of MS plaques, compared to that of normal white matter, appears to be more pronounced than corresponding T2 signal intensity changes (9).

The results reported by Castriota-Scanderberg et al present a compelling case for the specificity of DWI in distinguishing relapsing-remitting from secondary-progressive MS. A careful reading of Castriota-Scanderberg et al's method for region-of-interest selection additionally suggests the possibility that, because fewer T2-hyperintense plaques with concurrent T1 hypointensity were identified than T2-hyperintense plaques with concurrent elevated <D> values, the finding of a markedly increased diffusion coefficient within an MS plaque might also be a more sensitive predictor of axonal injury, and thus of clinical progression, than the finding of T1-hypointense signal only. Although the authors did not report sufficient data either to prove or refute this hypothesis, their observations do support the assertion that DWI probably provides added clinical value regarding MR imaging's accuracy in the clinical subtyping of MS patients.

Like all good studies, this one raises far more questions than it answers. Does the degree of elevation of diffusivity within an MS plaque truly correlate with axonal injury? Is marked elevation of diffusivity within plaques really a more specific and sensitive indicator of a clinically progressive disease subtype than the degree of T1 prolongation is? What is the correlation between <D> values and NAA concentrations? Between <D> values and MTRs? Between <D> values and enhancement? At my institution, we have observed only a poor correlation between the enhancement found in “new” MS plaques and their DWI signal changes. Only four enhancing lesions were noted in the study by Castriota-Scanderberg et al, and these were excluded from analysis. Might diffusivity changes correlate more highly with the axonal transection of “chronic” plaques than with inflammatory demyelination of “new” plaques? Under what clinical circumstances, if any, are reduced <D> values found within plaques? Might the clinical value of DWI in MS be further refined, as has been suggested by some (and successfully applied in the setting of acute stroke), by a detailed assessment of diffusion anisotropy, the “shape” of the diffusion tensor, within and around plaques (11–13)?

Finally, how can discrepancies between the results of the study by Castriota-Scanderberg et al and those of others be explained (10)? Such discrepancies might be attributed to subtle yet important differences in the criteria for patient inclusion, or in the methods used for region-of-interest selection. Future MS imaging studies must carefully distinguish “acute” from “chronic” plaques based not only on their current MR imaging characteristics and clinical presentation, but on comparison with prior studies. A well-designed study might also attempt to correlate conventional and “functional” MR imaging findings directly with those of serial follow-up MR examinations and long-term clinical outcome. The subtypes of relapsing-remitting, primary-progressive, and secondary-progressive MS would need to be defined rigorously according to a strict clinical standard of reference.

In conclusion, conventional MR imaging is a sensitive but not specific test for MS. MR imaging findings may be present in asymptomatic individuals; conversely, clinically definite MS may present occasionally with a normal T2-weighted MR examination of the brain and spinal cord (2). If the findings of Castriota-Scanderberg et al could be confirmed and expanded upon in a larger, well-controlled study, this could have important consequences with respect to MR imaging's ability not only to help one determine more accurately the clinical subtypes of MS patients, but to be predictive of prognosis or response to treatment. Conventional MR imaging, because of the poor correlation between MR signal abnormalities and clinical disability in established disease, is of only limited value as a surrogate marker of disease progression in MS clinical trials. DWI and other “functional” techniques have the potential to improve further the detection and characterization of clinically relevant lesions in MS patients, which could impact positively on patient care.