1992
0730-725X/92 $5.00 + .OO Copyright 0 1992 Pergamon Press plc
l Original Contribution
IN VIVO MAGNETIC RESONANCE DIFFUSION MEASUREMENT IN THE BRAIN OF PATIENTS WITH MULTIPLE SCLEROSIS H.B.W. LARSSON, C. THOMSEN, J. FREDERIKSEN,*
M. STUBGAARD, AND 0. HENRIKSEN Danish Research Center of Magnetic Resonance at Hvidovre University Hospital, Copenhagen, and *Gentofte University Hospital, Copenhagen, Denmark
Measurement of water self-diffusion in the brain in 25 patients with multiple sclerosis was performed by magnetic resonance imaging. Quantitative diffusion measurements were obtained using single spin-echo pulse sequences with pulsed magnetic field gradients of different magnitude. Twenty-two of these patients also underwent measurement of the transverse relaxation time ( T2). Only one plaque was evaluated in each patient. Based on prior knowledge, 12 plaques were classified as being 3 mo or less in age, and 7 plaques were classified as being more than 3 mo old. In all 25 plaques, water self-diffusion was found to be higher than in apparently normal white matter. Furthermore, water self-diffusion was found to be higher in acute plaques compared with chronic plaques. Finally, a slight tendency toward a relationship between the diffusion capability and T2 was found. We believe that an increased diffusion capability signifies an increase of the extracellular water space, which probably is related to the degree of demyelination. Thus, measurement of water self-diffusion in multiple sclerosis plaques may contribute to the study of pathogenesis of demyelination. Keywords: MRI; NMR; Diffusion; Multiple sclerosis.
INTRODUCTION
In recent years, many studies have investigated a possible correlation between nuclear magnetic resonance imaging (MRI), electrophysiology, clinical, and laboratory parameters in patients with multiple sclerosis (MS). Fewer studies have measured in vivo relaxation times (T, and T2) in plaques due to MS. However, relaxation time measurements seem to provide information about the demyelination, edema, and gliosis in plaques, ‘9’ Based on serial relaxation time measurements carried out in patients with known acute plaques, a hypothesis was proposed, suggesting that the T2 was related both to the size of the extracellular space, which presumably had expanded because of demyelination, and to the extent of gliosis, caused by proliferation of astrocytes.2 Approximately 2 mo after the appearance of plaques, high T2 values were noticed in RECEIVEDl/9/91; ACCEPTED6127191. This study was assisted by the Copenhagen acute MSstudy group. The members are B. Anthonisen, P. ArlienSoborg, J. Boas, A. Dam, M. Dam, K. Ellemann, J.L. Frederiksen, A. Heltberg, K. Hyllested, J. Vive Larsen, H.B.W. Larsson, M. Lund, A. Nordenbo, J. Olesen, O.B. Paulson, M. Petersen, M. Ravensborg, and I. Zeeberg.
the center of plaques. Values of 900 msec or more for T2 were observed in several plaques. This was explained by a considerable increase in the extracellular (free) water space. Measurements of water self-diffusion, using conventional MRI equipment have been suggested by several authors.3-6 In vivo studies in human and cat brains have clearly shown restricted diffusion of water in normal white matter.5,7-g In case of myelin degradation and edema we expect that water self-diffusion capabilities are increased compared with the normal white matter. Thus, measurements of diffusion of water in MS plaques may provide new knowledge about the pathogenesis of the demyelination. In the present cross-sectional study, the water self-diffusion coefficient was measured in patients with acute and chronic MS plaques. The aims of the present study were: This investigation was supported by the Danish Multiple Sclerosis Foundation and the Warwara Larsen Foundation. Address correspondence to Henrik B.W. Larsson, MD, Danish Research Center of Magnetic Resonance, Hvidovre Hospital, DK-2650 Copenhagen, Denmark.
8
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
1. To compare the diffusion coefficient of plaques with the diffusion coefficient of adjacent, apparently normal, white matter in patients with MS 2. To compare the diffusion coefficient of acute plaques (less than 3 mo old) with the diffusion coefficient of chronic plaques 3. To correlate the diffusion coefficient of plaques with values of Tz measured in the same plaques.
entire hemisphere, cerebellum, and brainstem were imaged. A single slice containing a large plaque was then selected with the aim of measuring T2 and diffusion. T2 relaxation measurement. A Carr-Purcell-Meiboom-Gill phase alternating-phase shift multiple spinecho pulse sequence with 32 echoes was used.12 The echoes were recorded at 30-msec intervals from 30 to 960 msec. The TR was 4 set and an g-mm-thick axial slice was used.
MATERIAL AND METHODS Patients Nearly all patients were selected from a group of patients which had been followed by serial MRX examinations at regular intervals for more than 1 yr in our department. Thus, plaques could be classified in two groups: plaques which had persisted more than 3 mo and plaques which appeared within the last 3 mo. This separation was chosen because a previous study2 had shown a decline of T2 in plaques approximately 2 mo after occurrence, indicating the possibility of separation between acute plaques and chronic plaques based on T2 values. In total, 25 patients with definite MS” were investigated. These patients (12 females, 13 males) had a mean age of 33 yr (range l&52), a mean duration of disease of 5.2 yr (range 7 days to 15 yr) and a mean Kurtzke’s expanded disability score” of 3.8 (range 2.0-7.0). Nine patients suffered an acute attack (evaluated clinically) at the time of examination. Only one large plaque was evaluated in each patient. Twelve plaques were classified as being 3 mo or less in age, and seven plaques were classified as being more than 3 mo old. In fact, these seven plaques had persisted for more than 5 mo. In six patients, the age of the plaque could not be determined. Twenty-two patients underwent both T2 measurements and diffusion measurements, that is, 14 patients with a less-than-3-moold lesion and 2 patients with a more-than-3-mo-old lesion, and 6 patients with a lesion of unknown age. h4RI Investigation The study was carried out on a whole-body MR scanner (Siemens Magnetom H15, Siemens, AG, Erlangen, FRG), operating at 1.5 T. The brain was imaged in the axial plane in the head coil using a double spin-echo sequence with a repetition time (TR) of 1.8 set, and echo times (TE) of 30 and 90 msec. The slice thickness was 4 mm, and a matrix size of 256 x 256 was used, giving a voxel size of 1.2 x 1.2 x 4 mm3. Twelve or 13 slices with an interslice spacing of 4 mm were acquired. Subsequently, the sequence was repeated, imaging the interslice space. In this way, the
Diffusion measurement. A detailed description is given in reference 5. In short, the sequence consisted of a normal spin-echo imaging sequence modified by addition of symmetrical gradients around the 180” refocusing radiofrequency (rf) pulse. These gradients were in the same direction as the slice-selective gradients, encoding diffusion perpendicular to the slice. The sequence was run seven times with different magnitude of the symmetrical gradients: 0.00, 2.45, 3.46, 4.24, 4.90, 5.48, and 6.00 mT/m. This corresponds to a variation of the gradient factor b from 0.012.10’ sec/m2 (at 2.45 mT/m) to 0.074. lo9 sec/m2 (at 6.00 mT/m) (6 = 45 msec and A = 59.1 msec). TR was 1.O set, and TE was 122 msec. Diffusion was measured in an 8-mm axial slice using a matrix size of 128 x 128. Calculation and Statistics Based on the seven diffusion-encoded images, the signals were read out from regions of interest (ROIs) in white matter and MS plaques. The average size of
Fig. 1. A spin-echo MR image of a MS patient. TR set, TE = 30 msec.
In vivo MR diffusion measurement in brain of MS patients 0 H.B.W. LARSSONET
AL.
9
ROIs was 46 pixels (range 12-70). The calculation of the diffusion coefficient is based on the linear relationship between the logarithm of the MR signal and the square of gradient strength used. The numerical value of the slope of the line, which is proportional to the diffusion coefficient, was estimated by linear regression analysis. Measurement of the diffusion is critically sensitive to even small movements of the head. As nearly all our patients with MS had involuntary movements of the limbs and head, artifacts in the diffusion images could not be avoided. The impact of movement artifact is a smearing of the MR signal along the phase-encoding direction. This problem was reduced by normalization of the diffusion coefficient of the plaque with the diffusion coefficient of adjacent white matter. With regard to T2estimation in plaques,
Fig. 2. The corresponding calculated diffusion image of the same patient and at the same level as presented in Fig. 1. Bright area correspond to area with high diffusion coefficient.
:0&C lnt
6.0
5.9
5.4 5.3 5.2
5.1
%
b
112
Ib
22
$0
3’6
G2
Fig. 3. An example of the diffusion curves read out from a ROI in a plaque (a) and adjacent white matter (x). The lines give the least-squares fit.
Fig. 4. The distribution of the normalized diffusion coefficient of 19 plaques in which the age could be estimated. The horizontal lines represent the mean of the groups, 1.33 and 1.14, p < .05.
10
MagneticResonance Imaging 0 Volume 10, Number 1, 1992
ROIs were evaluated by monoexponential curve analysis using a least-squares fit. Comparison between groups was performed using the Mann-Whitney rank sum test for unpaired samples. I3 The correlation between the data sets was evaluated using the Spearman rank correlation. l3
more, there was a statistically significant difference between acute and chronic plaques, with a mean of the normalized diffusion coefficient of 1.33 and 1.14, respectively (p < .05) (see Fig. 4). In Fig. 5, the normalized diffusion coefficient is depicted as a function of the transverse relaxation rate R2 (1/T2). The correlation coefficient was -0.42, 0.5
.I). In all 25 plaques, the diffusion coefficient was found to be higher (median value 4.2. 10m9 m2/sec) than in the adjacent white matter (p < .OOl). Further-
Diffusion
DISCUSSION To our knowledge, this is the first study which provides evidence for an increase of water self-diffusion in plaques due to multiple sclerosis (MS). This may give new insight into the pathogenesis of the disease, and diffusion measurement may be useful for the assessment of the degree of demyelination seen in MS. Measurements of water self-diffusion in vivo are extremely sensitive to displacement during TE and movements during acquisition of the MR signal. Pa-
ratio
. 1.8-
.
1.7-
. 1.6-
.
1.5-
.
.
.
1.4
. 1.3
. . .
1.2
. l
. l
. .
.
1.1
.
.
. . .
lCL 1.0
2.0
3.0
4.0
510
610
7:o
s!o
9:o
I 101)
.
'/T,
Fig. 5. A plot of the normalized diffusion coefficient as a function of the transverse relaxation rate (l/T’) of the plaques in the 22 patients that both underwent T2measurement and diffusion measurement, i.e., 14 patients with plaque age of less than 3 mo 2 patients with plaque age of more than 3 mo, and 6 with unknown plaque age.
In vivo MR diffusion measurement in brain of MS patients 0 H.B.W. LARSSONET AL.
tients with MS are prone to involuntary movements due to spasticity, which results in the occurrence of the well-known movement artifacts along the phaseencoding direction, and an overestimation of the diffusion coefficient. It was obvious that the measured diffusion coefficient of white matter in the patients were about twice as large as anticipated from our previous study of healthy volunteers,’ and diffusion measurements were much more difficult in the type of patients compared with healthy volunteers. As the plaque and adjacent white matter are exposed to the same degree of movement, the problem was alleviated, or at least reduced, by calculation of the ratio between the diffusion coefficient of the plaque and the diffusion coefficient of the adjacent white matter along the same phase-encoded column. Furthermore, placement of ROIs in the frontal lobes were in general avoided because pulsatile brain motion was more pronounced there,r4 which also can be seen in Fig. 2. The calculated ratio is then a relative measure of the diffusion in plaques compared with apparently normal white matter in patients. Thus a more reliable measurement of the diffusion differences was achieved. We hypothesize that an increase of the diffusion coefficient in acute plaques may be related to an increase in the extracellular space due to demyelination. It is generally known that diffusion is an anisotropic process in white matter, that is, diffusion is nearly unrestricted parallel to the nerve fibers, but restricted perpendicular to the nerve fibers.7-9 The myelin removed is replaced by water, resulting in an increase in the extracellular
(4
11
space, and the formerly restricted diffusion becomes unrestricted as illustrated in Fig. 6. Evidence for this hypothesis comes from previous histological studies as well as MR studies.2,‘5J6 The finding of a decrease of the diffusion coefficient in ischemic brain regions in cats is in accordance with this. The ischemic damage may result in intracellular swelling, during which water protons originally in the faster-diffusing extracellular space migrate into a more diffusion-restricted intracellular compartment.‘7 The relative diffusion coefficient in acute plaques was found to be higher than in chronic plaques. Gliosis dominates in these plaques and may be responsible for the lowering of the diffusion coefficient compared with the diffusion coefficient of acute plaques. We anticipated a relationship between the transverse relaxation rate and the diffusion coefficient of the plaques. However, only a slight tendency toward such a relationship was found. The anticipated relationship might have been modified by the local production of IgG and the number of inflammatory cells. In conclusion, diffusion measurements seem to provide important information concerning benign intracranial hypertension, ‘* cerebral infarction,7 and now also MS. The diffusion coefficient was found to be higher in plaques due to MS, and acute plaques had a higher diffusion coefficient than chronic plaques. A slight tendency toward a relationship between the diffusion coefficient and T2 was also found. Further refinements of diffusion measurements are warranted to minimize the effect of movements of the brain due to
@I
Fig. 6. A schematic representation of the proposed hypothesis. (A) Normal myelinated nerve fiber; the diffusion path is clearly restricted. (B) demyelinated nerve fibers; the extracellular space has increased and the diffusion path is now unrestricted.
12
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
pulsation and involuntary motion. Work along these lines is currently in progress in our laboratory. I9 Acknowledgment-Poul Ring is acknowledged for writing the computer program for the calculation of the diffusion images.
REFERENCES 1. Ormerod, I.; Bronstein, A.; Rudge, P.; Johnson, G.; Macmanus, D.; Halliday, A.M.; Barratt, H.; DuBoulay, E.P.G.H.; Kendal, B.E.; Moseley, I.F.; Jones, S.J.; Kriss, A.; Peringer, E. Magnetic resonance imaging in clinically isolated lesions of the brainstem. J. Neurol. Neurosurg. Psychiatry 49:737-743; 1986. 2. Larsson, H.B.W.; Frederiksen, J.; Petersen, J.; Nordenbo, A.; Zeeberg, I.; Henriksen, 0.; Olesen, J. Assessment of demyelination, edema, and gliosis by in vivo determination of T, and Tz in the brain of patients with acute attack of multiple sclerosis. Magn. Reson. Med. 11:337-348; 1989. 3. Wesbye, G.E.; Mosely, M.E.; Ehman, R.L. Translational molecular self-diffusion in magnetic resonance imaging. Invest. Radiol. 19(6):491-498; 1984. 4. Le Bihan, D.; Breton, E.; Lallemand, D.; Grenier, P.; Cabanis, E.; Lavel-Jeantet, M. MR imaging of intravoxel incoherent motions: Application to diffusion and perfusion in neurologic disorders. Radiology 161: 401-407; 1986. 5. Thomsen, C.; Henriksen, 0.; Ring, P. In vivo measure-
ment of water self-diffusion in the human brain by magnetic resonance imaging. Acta Radiol. 28:353-361; 1987. 6. Chien, D.; Buxton, R.; Kwong, K. J.; Brady, T.; Rosen, B. MR diffusion imaging of the human brain. J. Comput. Assist. Tomogr. 14(4):514-520; 1990. 7. Mosely, M.E.; Cohen, Y.; Mintorovitch, J.; Chileuitt,
L.; Shimizu, H.; Kucharczyk, J., Wendland, M.F.; Weinstein, P.R. Early detection of regional cerebral ischemia in cats: Comparison of diffusion- and Tz-weighted MRI and spectroscopy. Magn. Reson. Med. 14:330346; 1990. 8. Larsson, H.B.W.; Stubgaard, M.; Thomsen, C.; the Scandinavian Flow Group. In vivo measurement of diffusion in CNS using a flow compensated spin echo sequence. Proceedings of the 9th Meeting of the Society of Magnetic Resonance in Medicine, New York, 1990. Abstract 389.
9. Le Bihan, D.; Turner, R.; Douek, P. Is water diffusion restricted in human brain? Proceedings of the 9th Meeting of the Society of Magnetic Resonance in Medicine, New York, 1990. Abstract 377. 10. Poser, C.M.; Paty, D.W.; Scheinberg, L.; McDonald, W.I.; Davis, S.A.; Ebers, G.C.; Johnson, K.P.; Sibley, W.A.; Silverberg, D.H.; Tourtellotte, W.W. New diagnostic criteria for multiple sclerosis: Guidelines for research protecols. Ann. Neural. 13:227-231; 1983. 11. Kurtzke, J.F. On the evaluation of disability in multiple sclerosis. Neurology 11:686-694; 1961. 12. Graumann, R.; Oppelt, A.; Stetter, E. Multiple spinecho imaging with a 2D Fourier method. Magn. Reson. Med. 3:707-721; 1986. 13. Siegel, S. Nonparametric Statistics. New York, Toronto, London: McGraw-Hill Book Company; 1956. 14. Wirestam, R.; Greitz, D.; Thomsen, C.; Stahlberg, F.; Nordell, B.; Stubgaard, M.; the Scandinavian Flow Group. Measurements of pulsatile brain motion using MR phase imaging. Proceedings of the 9th Meeting of the Society of Magnetic Resonance in Medicine, New York, 1990. Abstract 1107. 15. Hallpike, J.F., Adams, C.W.M.; Tourtellotte, W.W. Multiple sclerosis: Pathology, Diagnosis and Management. London: Chapman & Hall; 1983. 16. Tourtellotte, W.W.; Parker, J.A. Some spaces and barriers in postmortem multiple sclerosis. Prog. Brain Res. 29~493-522; 1967. 17. Moseley, M.E.; Cohen, Y.; Mintorovitch,
J.; Chileuitt, L.; Shimizu, H.; Kucharczyk, J.; Wendland, M.F.; Weinstein, P.R. Early detection of regional cerebral ischemia in cats: Comparison of diffusion- and Tzweighted MRI and spectroscopy. Magn. Reson. Med. 14:330-346;
1990.
18. Soelberg Sorensen, P.; Thomsen, C.; Gjerris, F.; Schmidt, J.; Kjrer, L.; Henriksen, 0. Increased brain water content in pseudotumour cerebri measured by magnetic resonance imaging of brain water self diffusion. Neural. Res. 11:160-165; 1989. 19. Larsson, H.B.W.; Christiansen, P.; Thomsen, C.; Stubgaard, M.; Frederiksen, J.; Henriksen, 0. In vivo measurement of water self-diffusion in patients with chronic multiple sclerosis. Proceedings of the 9th Meeting of the Society of Magnetic Resonance in Medicine, New York, 1990. Abstract 150.
