Diffusion tensor imaging of lesions and normal-appearing white matter in multiple sclerosis

Patients.

Six patients (mean age, 34.2 ± 4.2 years; one man, five women) with clinically definite MS (ascertained using the criteria of Poser et al.15) attending the National Hospital for Neurology and Neurosurgery, London, were studied. A full neurologic history and examination was performed, and disability was assessed using Kurtzke’s Expanded Disability Status Scale (EDSS).16 Five patients had relapsing–remitting MS and one patient had secondary progressive MS. Six normal, healthy matched control subjects (mean age, 34.3 ± 8.9 years; one man, five women) were also studied.

MRI methods.

All scans were performed on a 1.5-T Signa Horizon Echospeed scanner (Signa, General Electric Medical Systems, Milwaukee, WI) equipped with shielded magnetic field gradients of up to 22 mT m^-1 and a standard quadrature head coil for radio frequency transmission and reception of the NMR signal. Head motion was minimized with foam pad immobilization as provided by the manufacturer. The field of view was 24 × 24 cm. High-resolution mildly and heavily T2-weighted images were acquired at 26 contiguous 5-mm axial slices covering the whole brain (repetition time [TR], 2,000 msec; echo time [TE], 30 msec, 120 msec; matrix, 256 × 256), followed by diffusion-weighted echo planar images with no repositioning (TE, 78 msec; matrix, 96 × 96). Diffusion gradients at 4 b values increasing from approximately 0 to 700 sec/mm² were applied in each of seven noncolinear directions) at 10 slice locations centered on the lateral ventricles and matched to the T2-weighted image slices. Five acquisitions of each set of diffusion data were performed and co-added following magnitude reconstruction to improve the signal-to-noise ratio (SNR). Cardiac gating was used; image acquisition was triggered from every second R wave monitored using a pulse oximeter. The diffusion tensor, MD, and FA were calculated on a pixel-by-pixel basis as described by Basser et al.9,10 (FA was chosen because it provided excellent contrast between white matter tracts and cortical gray matter. Although FA is affected by the amount of noise in diffusion data, particularly at a low SNR, the high SNR achieved in our study by multiple averaging provided FA maps with similar gray/white matter contrast to that obtained using the volume ratio, which is less affected by noise at low SNR.) After DTI, T1-weighted images were acquired (TR, 540 msec; TE, 20 msec; matrix, 256 × 256) 5 minutes after the administration of 0.1 mmol/kg gadolinium-diethylenetriamine pentaacetic acid (DTPA).

All images were displayed on a Sun workstation (Sun Microsystems, Mountain View, CA). Lesions were outlined on the T2-weighted images using a semi-automated contouring technique (DispImage version 4.7; D. Plummer, UCL Hospitals Trust, London, UK). Each lesion was first classified as enhancing or nonenhancing and then the nonenhancing lesions were classified (using the postcontrast T1-weighted images) as T1 isointense or hypointense. T1 hypointensity was defined as reduced lesion signal intensity with respect to the surrounding NAWM. Rectangular NAWM regions of interest (ROIs) of uniform size were defined in similar anatomic areas to the lesions (frontal, parietal, temporal, and occipital white matter) on the T2-weighted images. To allow the accurate transfer of ROIs between the T2-weighted images and the quantitative DTI maps, the latter were interpolated up to the same matrix size as the T2-weighted images. For each ROI, MD and FA were measured. The accuracy of mapping was confirmed by correct registration of the outlined brain and ventricles between the two sets of images (T2-weighted and DTI) for each patient.

Statistical analysis.

The MD and FA values for each set of regions (control NAWM, patient NAWM, nonenhancing lesions, enhancing lesions, T1 hypointensities) were compared using the Mann–Whitney U test. The relationship between MD and FA was assessed by determining Spearman’s correlation coefficient.

Findings in NAWM.

In this study patient NAWM showed significantly higher MD and lower FA than control subject white matter. Structural abnormalities in NAWM may play an important role in the development of irreversible disability in MS. Previous work14,17 has demonstrated that diffusion is higher in NAWM of MS patients than in control subjects, in agreement with our data. At present we can only speculate on the structural substrate of our findings. Studies of NAA concentrations using NMR spectroscopy suggest that axonal loss occurs in the NAWM of MS patients,4 and that this may underlie persistent disability.18 The expected diffusion changes due to axonal loss have already been mentioned; namely, increased MD and reduced FA. Our findings are therefore consistent with the major diffuse NAWM pathology being axonal loss. An advantage of DTI over spectroscopy is that it is possible to map abnormalities with greater spatial resolution; the relatively smaller range of values in the healthy brain for MD may make this the preferred measurement in such studies. In one of the few histopathologic studies of MS NAWM, a high prevalence of gliosis was reported,19 but this would be expected to reduce both anisotropy (assuming that glial proliferation is a rather structurally disorganized process) and diffusivity.

Mean diffusivity findings in lesions.

Our finding of raised MD in lesions compared with NAWM is consistent with previous studies showing an increased lesion ADC,13,14 and suggests a net loss of structural barriers to water molecular motion in plaques. We also observed higher MD in acute (enhancing) versus chronic (nonenhancing) lesions. Postmortem studies have confirmed that lesion enhancement reflects inflammation, blood–brain barrier leakage, and edema.20,21 There is also evidence that axonal loss can occur in acute lesions.22 Our findings could thus be explained by an expansion of the extracellular space due to a combination of vasogenic edema and axonal loss in such lesions. In the population of acute lesions studied, the effects of these processes outweigh those of others that could potentially restrict water molecular motion, such as the accumulation of inflammatory cells including macrophages and the presence of myelin breakdown products.23,24 There is evidence from proton NMR spectroscopy that acute MS lesions contain lipid products of myelin breakdown.25 Further work combining diffusion MRI with proton NMR spectroscopy may help to clarify the effect of these products on water diffusion.

We found a larger range of MD values in chronic (nonenhancing) plaques, suggesting considerable pathologic heterogeneity. Possible contributory processes include gliosis, which has been suggested to reduce diffusion,14 and persisting demyelination, or axonal loss, both of which would be expected to increase diffusion by increasing the extracellular space.26 We observed the highest MD values in T1 hypointense plaques, which probably represent destructive lesions containing extensive axonal loss.5 Our data are consistent with those obtained from a navigated spin-echo diffusion study27 that also reported that T1 hypointense lesions exhibit higher diffusion values than T1 isointense lesions. In this regard, MD appears to give complementary information to T1-weighted images, with the advantage that it is inherently quantitative.

Anisotropy findings in lesions.

Quantitative data on the anisotropy properties of lesions in MS have not been described previously. This study shows that lesions (considered together) have significantly lower FA than NAWM, and that acute (enhancing) lesions have significantly lower FA than chronic (nonenhancing) lesions. Moreover, FA proved more sensitive in discriminating between acute and chronic lesions than MD, as judged by the magnitude and significance of the observed differences. A study of the animal model experimental allergic encephalomyelitis (EAE) using DWI acquired with diffusion sensitization in three directions demonstrated lower anisotropy in acute compared with chronic lesions,28 in keeping with our data. The FA differences in enhancing and nonenhancing lesions could at least be accounted for partly by acute vasogenic edema.

T1 hypointense lesions had lower FA than T1 isointense lesions, as anticipated from the greater degree of presumed axonal loss. The lowest FA was seen in acute enhancing lesions, suggesting that extracellular edema too has a marked effect on tissue FA. It will be of interest to measure FA as lesions evolve to determine how much of the structural loss is permanent. Changes due to edema should be reversible, whereas axonal loss may cause a persistent reduction in FA.

Future questions for DTI in MS.

It is likely that a complex interaction between the pathologic features of MS plaques—edema, demyelination, axonal loss, and gliosis—determines diffusion measurements. The data presented suggest that of these, edema and axonal loss contribute most to water diffusion changes because enhancement and T1 hypointensity are associated with the most significant diffusion changes in FA and MD respectively. Ideally, experimental and postmortem studies correlating the DTI changes and histopathology in MS are needed to establish the pathologic basis of the diffusion MRI findings.

DTI is likely to illuminate further the pathophysiologic changes occurring within MS lesions, particularly when employed in longitudinal studies of lesion evolution together with other MRI techniques including NAA spectroscopy and magnetization transfer imaging. That the correlation observed between FA and MD for all regions studied was modest (see figure 3) implies that they may provide partially independent and complementary data for evaluating brain pathology. Measurement of both diffusivity and anisotropy may improve the ability to detect specific pathologic changes, and warrants further investigation.

The relatively small number of patients in this study is offset by the large number of lesions and NAWM regions examined, allowing a systematic characterization of their structural properties. However, our sample size does not permit an investigation of the relationship of DTI changes with the clinical course or level of disability. Future studies will examine the relationship of DTI changes to clinical deficit, and the natural history of diffusion changes within lesions.