Magnetic Resonance in Medicine 63:633–640 (2010)
Quantitative Magnetization Transfer and Myelin Water Imaging of the Evolution of Acute Multiple Sclerosis Lesions Ives R. Levesque,1* Paul S. Giacomini,1 Sridar Narayanan,1 Luciana T. Ribeiro,1 John G. Sled,2 Doug L. Arnold,1 and G. Bruce Pike1 Quantitative magnetization transfer imaging provides in vivo estimates of liquid and semisolid constituents of tissue, while estimates of the liquid subpopulations, including myelin water, can be obtained from multicomponent T2 analysis. Both methods have been suggested to provide improved myelin specificity compared to conventional MRI. The goal of this study was to investigate the sensitivity of each technique to the progression of acute, gadolinium-enhancing regions of multiple sclerosis. Magnetization transfer and T2 relaxometry data were acquired longitudinally over the course of 1 year in five relapsing-remitting multiple sclerosis patients and in five healthy controls. Parametric maps were analyzed in enhancing lesions and normal-appearing white matter regions. Quantitative magnetization transfer parameters in lesions were most abnormal at the time of enhancement and followed a pattern of recovery over subsequent months. Lesion myelin water fraction was abnormal but did not show a significant trend over time. Quantitative magnetization transfer was able to track the degree and timing of the partial recovery in enhancing multiple sclerosis lesions in a small group of patients, while the recovery was not detected in myelin water estimates, possibly due to their large variability. Our data suggest the recovery is characterized by quick resolution of inflammation and a slower remyelination process. Magn C 2010 Wiley-Liss, Inc. Reson Med 63:633–640, 2010. V Key words: quantitative; magnetization transfer; component T2; multiple sclerosis; acute lesions
multi-
A few MRI methods have been suggested to provide improved pathologic specificity in human white matter (WM) over conventional MRI sequences. Two techniques proposed to be more specific for myelin are quantitative magnetization transfer imaging (QMTI) and multicomponent T2 analysis of spin echo data (QT2). In this longitudinal study, both techniques were used to follow the progression of the gadolinium (Gd)-enhancing region of acute lesions of multiple sclerosis (MS). The magnetization transfer (MT) effect can be exploited to produce contrast in MRI (for a review see 1 McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Canada. 2 Hospital for Sick Children, Toronto, Canada
This work was presented in part at the 16th Scientific Meeting and Exhibition of the International Society for Magnetic Resonance in Medicine, Toronto, Canada, 2008. *Correspondence to: Ives R. Levesque, McConnell Brain Imaging Centre, Montreal Neurological Institute, 3801 University Street, WB325, Montreal, Quebec, H3A 2B4 Canada. E-mail:
[email protected] Received 3 February 2009; revised 14 September 2009; accepted 20 September 2009. DOI 10.1002/mrm.22244 Published online 9 February 2010 in Wiley InterScience (www.interscience. wiley.com). C 2010 Wiley-Liss, Inc. V
Henkelman et al. (1) and Tofts (2)) and is sensitive to the principal constituents of myelin in human WM (3). MT can be described using a two-component model, grouping spins into the free (or liquid) pool, consisting of water hydrogen nuclei with long T2 (T2f > 10 ms), and the restricted (or semisolid) pool, consisting of hydrogen nuclei of large lipids with T2 too short to be imaged directly using MRI (T2r < 100 ls). Detailed information about these spin populations can be mapped using offresonance QMTI techniques, with an appropriate twopool model of MT in tissue (4). QMTI yields the relative size of the restricted proton pool (F), the first-order forward magnetization exchange rate (kf), and most relaxation parameters of the free and restricted pools (R1f, T2f, and T2r) (5). While QMTI measurements do not provide absolute specificity to the molecular constituents of tissue, there is convergent evidence that MT changes in WM reflect changes in myelin content. Strong correlations have been observed between F and the myelin lipid content observed with the Luxol fast blue stain in fixed (6) and fresh (7) postmortem tissue. QMTI parameter variations in the WM of healthy controls are consistent with tissue myelination (8), and densely myelinated WM fiber tracts are distinguishable in maps of the restricted pool fraction (9). In MS, F is substantially reduced in chronic lesions (5,10–12), and small but significant decreases in F have also been reported in the normal-appearing WM (NAWM) of patients (11,13). The impact of MS pathology on the T2r is still a matter for discussion as both decreases (10) and increases (11) have been reported in MS lesions. Multicomponent T2 relaxation can be observed using multiecho spin echo sequences (QT2) (14–16) and is characterized in each voxel by the T2 distribution (a plot of the component amplitude as a function of T2). In brain tissue, T2 distributions present a few peaks, which have been empirically assigned to compartmentalized spin populations: a short T2 peak around 20 ms (between 10 and 50 ms) representing the water trapped between the layers of myelin, a second peak around 70–90 ms assigned to intra- and extracellular water, and a third peak with T2 > 2 sec most often assigned to cerebrospinal fluid signal (14,16). The myelin water fraction (MWF) is computed as the fraction of signal in the T2 distribution below 40 ms (17–19) or 50 ms (16). The MWF can be computed voxel by voxel to yield parametric maps (15), a technique also called myelin water imaging.
633
634
Myelin water imaging is also a putative marker of myelin in WM and MS. The MWF and the geometric mean of the intra/extracellular peak of T2 distribution are significantly altered in lesions and the NAWM of MS (20– 22). The MWF changes are mainly due to myelin loss, as confirmed by pathology studies using Luxol fast blue (23,24), while increases in intra/extracellular T2 have been proposed as a marker of inflammation (25). A recent study of MS lesions did not reveal differences in myelin water content between T1-hypointense and T1isointense lesions (18). A study combining QMTI and QT2 in nine controls and 19 MS patients (26) reported significantly decreased restricted pool and MWFs in lesions and significantly decreased restricted pool fraction in NAWM. These authors concluded that F and the MWF both reflect demyelination to some extent, with differences attributed to lesion pathology (inflammation, axonal loss, etc.). Despite the growing number of studies employing these two quantitative techniques in MS, there has been limited work on one of the most dynamic phases of the disease, acute Gd-enhancing lesions. The MT ratio, which reduces the entire MT effect into a single parameter, decreases with acute demyelination and increases with remyelination in an animal model of demyelination (27). In acute Gd-enhancing lesions of MS patients, the MT ratio has been observed to decrease, followed by recovery over subsequent months in most cases (28–32). QMTI, while more complex to perform than MT ratio, yields a more detailed characterization of the MT effect and can help clarify the origins of the MT ratio changes. QT2 has shown sensitivity to demyelination and edema during the acute phase of individual lesions in two patients (19). The objective of this study was to characterize the progression of acute MS lesions following Gd enhancement, using QMTI and myelin water imaging. Metrics from both techniques are reported over a 1-year period in acute lesions from five patients with relapsingremitting MS and compared to values in NAMW and WM from healthy controls. The time evolution of each parameter is considered in a context of combined acute demyelination and inflammation. To our knowledge, this is the first longitudinal, comparative study of QMTI and multicomponent T2 parameters in Gd-enhancing lesions of MS patients. MATERIALS AND METHODS This study followed five patients with relapsing remitting MS, all female, who were recruited based on the presence of an acute relapse and the discovery of an active hemispheric lesion, as defined by Gd enhancement, in the patient’s routine clinical examination. Inclusion criteria also included a diagnosis of relapsing remitting MS and age greater than 18 years. Patients were excluded if they were pregnant, were unable to receive Gd, or if they had received steroids prior to the baseline scan. Active lesions were visually inspected on the clinical postcontrast scans by a neurologist (P.S.G.): to be reliably observed and studied with the quantitative
Levesque et al.
imaging techniques, each lesion had to result in one or more low-resolution voxels after resampling and thresholding (2 2 7 mm3, see below). All patients underwent clinical evaluation by a neurologist prior to the baseline scan and at subsequent intervals of 3 months. Expanded Disability Status Scale scores were evaluated for each patient and ranged from 1 to 4 at the time of the baseline scan. One patient was treated with glatiramer acetate, while the remaining four patients were not on any disease-modifying therapy at the time of the initial scan. Steroids were permitted after the baseline scan. Only one patient showed a lesion with Gd enhancement that persisted beyond the baseline scan. The enhancement in that lesion subsequently resolved the following month without treatment. MRI examinations were performed during the period of lesion enhancement and at monthly intervals up to 5 months postenhancement, with follow-up scans at 8 and 11 months. Patient age at entry and total number of examinations were, respectively, 25 years/7, 42 years/7, 56 years/7, 41 years/7, and 49 years/6, for a total of 34 datasets. Data were also acquired in five healthy adult controls (two women, three men, aged 26–38 years) during the study. Each control was scanned at least four times, at intervals ranging from 11 days to 2.5 years. The controls did not receive Gd-diethylene triamine pentaacetic acid (DTPA). Approval for the study was obtained from the Research Ethics Board of the MNI, and each participant provided written informed consent. Scanning Protocol Three-dimensional T1-weighted data were acquired in each patient, with an isotropic voxel size of 1.3 mm. Fluid-attenuated inversion recovery (FLAIR), T2-, and proton-density (PD)-weighted data were also acquired in each patient for lesion identification and localization, all with a voxel size of 1 1 3 mm3. All quantitative data (MT, amplitude of radiofrequency field, amplitude of static field, T1, and T2) were acquired in a single, oblique 7 mm section, with 2 mm 2 mm in-plane voxel size and a rectangular 96 128 matrix. The slice position for each subject was initially selected to intersect the largest enhancing lesion during the baseline examination. The slice position was carefully reproduced for each subsequent examination by manual alignment to previous examinations, using console screen captures from the initial examination. MT data were acquired as described in Sled and Pike (5), consisting of two protocols with different pulse repetition times and saturation pulse durations, for a total of 60 measurements. The observed T1 (T1obs), required for QMTI, was measured using a Look-Locker pulse sequence (33), with published acquisition parameters (5). To improve the accuracy and reproducibility of T1 and QMTI parameters, the static and transmit (amplitude of radiofrequency) fields were mapped in each subject. The relative amplitude of static field was mapped using a two-point phasedifference technique (34), with a readout time delay chosen to bring the fat and water signals in phase (4.48 ms at 1.5 T). Separate amplitude of static field maps were also acquired for each MT sequence without the
Quantitative MT and T2 in Acute MS Lesions
635
FIG. 1. a: Initial post-Gd T1-weighted scan of a single patient, with outlines of enhancing region (green) and contralateral NAWM (cyan). b: Lower-resolution version of enhancing region and NAWM outlines containing quantitative voxels with at least 80% of the specified tissue type, displayed on a T2-weighted image resampled to the lower resolution of the quantitative maps.
saturation pulse to account for eddy currents due to the large crushing gradients in the MT sequence. A twopoint method with nonselective preparation (flip angles 33 and 66 ) was used to map the relative amplitude of radiofrequency field (35). Multiecho spin echo data were acquired using a 32echo spin-echo sequence (15) with nonselective 90x – 180y –90x composite refocusing pulses and crusher gradient scheme to spoil stimulated echoes (36). A pulse repetition time of 2 sec and echo spacing of 10 ms were used. These data were acquired with the same matrix and voxel size as all other quantitative data. A postcontrast T1-weighted scan was acquired following injection of a 0.2-mL/kg dose of Gd-DTPA (Magnevist; Bayer HealthCare, Toronto, ON) at the end of the session to avoid any influence of the contrast agent on the quantitative parameter estimates. Total scan time for the entire protocol was 64 min (28 min for conventional T1/ T2/PD-weighted, FLAIR, and post-Gd scans, and 36 min for all quantitative imaging). All data were acquired on a 1.5-T Siemens Sonata (Erlangen, Germany), using a quadrature head coil. Processing of Quantitative Parameter Maps Maps of QMTI model parameters were produced from MT-weighted data and T1obs, using a rectangular-pulse approximation of pulsed saturation and including corrections for amplitude of static field and amplitude of radiofrequency field inhomogeneities (5). The solid pool was modeled using a super-Lorentzian line shape (4). Fitting was performed using a Levenberg-Marquardt nonlinear least-squares algorithm implemented in MATLAB (The Mathworks, Natick, MA). Multicomponent analysis of the spin echo data was carried out using regularized nonnegative least squares to estimate the T2 distribution (37) at each voxel. A distribution of 120 T2 values was used, spanning a range of 10 ms to 4 sec in logarithmic steps. Regularization was performed by penalizing the energy of the T2 distribution and caused an increase of 2–2.5% in the resulting v2 value (38). The MWF is usually computed by summing
the T2 distribution within a predefined range of short T2 values. Ranges of 10–40 ms (17–19) or 10–50 ms (16) have been used previously. We selected 10–40 ms (10– 40.9 ms to be exact) based on visual inspection of peak separation in control data. The geometric mean of the entire T2 distribution, hT2i, was also computed at each voxel (21). Regional Analysis A regional analysis of the parameters was performed on enhancing regions and homologous NAWM regions. Enhancing regions of interest (ROIs) were labeled on the post-Gd T1-weighted scan from the first time point, illustrated by a green outline in Fig. 1a. NAWM ROIs were defined on the high-resolution scans contralateral to the lesions in regions free of any visible T2-hyperintense pathology, by the same neurologist (P.S.G.). These NAWM ROIs were selected to be as homologous as possible within the constraints of the single-slice quantitative data, avoiding partial-volume effects with other tissues and any lesions in the slice of interest or in neighboring slices. An example is illustrated in Fig. 1a by the blue outline. Labels were propagated to subsequent examination time points using software developed at the Montreal Neurologic Institute. The label maps were then resampled to the lower resolution of the QMTI scans, retaining only those voxels containing greater than 80% of each label, to limit partial-volume contamination by perilesional tissue. An example of the low-resolution ROI labels at the first time point is illustrated in Fig. 1b. RESULTS Examples of the quantitative MT and T2 maps are displayed in Fig. 2. The ROIs included in the analysis are also displayed, green for the enhancing region and blue for NAWM. In total, six enhancing lesions were identified (one per patient and two in patient 1). The mean parameter values were computed at each time point for the initially enhancing portion of the lesion and for the NAWM ROIs. Each parameter was analyzed over time
636
Levesque et al.
while T2f was slightly longer in NAWM. The percentage difference relative to NAWM was computed for each patient, and the average of the changes for all parameters is plotted in Fig. 4. The overall trends observed in the patient group were not observed in any individual patients. F, kf, and R1f were substantially reduced in enhancing regions and followed a general recovery pattern after resolution of enhancement, stabilizing around 2 to 3 months postenhancement. T2f was dramatically increased in enhancing regions and quickly decreased in the month following enhancement. T2r was stable in lesions following enhancement, with a maximum change of 6% (Fig. 4). The ANOVA revealed significant effects (P < 0.05) of subject and ROI for all but one QMTI parameter. The effect of subject accounted for 13–15% of the variation in F, R1f, and T2f and 53% of the variation in T2r; the effect of ROI accounted for 47–62% of the variation in F, R1f, and T2f but only 5% for T2r. kf was the notable exception, showing significant effects for ROI only, which explained 78% of the variation. Significant ROI-by-time interactions were observed for F and R1f (6– 7% of the variation explained). None of the abnormal QMTI parameters returned to NAWM levels during the course of this study, but significant recoveries of all parameters were observed in enhancing regions (P < 0.05) at 8 months postenhancement. Multicomponent T2 Analysis Results
FIG. 2. QMTI (a) and QT2 (b) parameter maps from the initial scan of one patient. The green outline is the enhancing region and the blue outline is NAWM. Units are seconds1 for kf and R1f and seconds for T2f and hT2i. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
using a three-way analysis of variance (ANOVA) with factors ROI (enhancing and NAWM), subject, and time; in addition, parameter values in enhancing regions at baseline and in the chronic phase, approximately 8 months later (time-point 8), were compared using a t test. Parameter values acquired in healthy control subjects, using the same protocol and scanner, were analyzed in WM ROIs (minor forceps, genu and splenium of the corpus callosum, and major forceps). Quantitative MT Results Key QMTI parameter values from enhancing regions and NAWM are plotted vs time postenhancement in Fig. 3, where the error bars indicate the standard error of the mean. WM values acquired in healthy control subjects are also plotted to enable comparison. Values of F, kf, and R1f were lower in MS NAWM than in control WM,
The MWF and hT2i of enhancing regions and NAWM ROIs are plotted vs time postenhancement in Fig. 5. WM values acquired in healthy control subjects are plotted to enable comparison. The MWF was lower in NAWM than in control WM, while hT2i was higher in NAWM. The MWF appeared to be reduced in enhancing regions compared to NAWM, but the difference was not statistically significant according to our ANOVA results; on the other hand, the ANOVA revealed a significant effect of subject for the MWF, explaining 58% of the variation. No clear time evolution was observed following enhancement. hT2i values were greater in enhancing regions than in NAWM, a difference that was significant in the ANOVA (P < 0.05) and largest at initial enhancement. Partial recovery of hT2i was observed postenhancement, and this recovery was significant in the ANOVA (as effects of time and ROI-by-time interactions; P 0.008) and when comparing baseline and month 8 (P < 0.05). The hT2i in enhancing regions did not return to baseline during the course of this study. Since both the MWF and F have been proposed as markers of myelin content, we computed the Spearman’s rank correlation coefficient (qs) between them. This was also evaluated for hT2i and T2f to evaluate the extent to which the T2f estimated with QMTI reflects the average T2 of the observable water protons. F and MWF were not significantly correlated in Gd-enhancing regions or NAWM ROIs (P 0.15) but were very weakly correlated when Gd-enhancing and NAWM regions were pooled (qs ¼ 0.25; P ¼ 0.018). The correlation between hT2i and T2f was modest in enhancing regions (qs ¼ 0.61, P < 0.0001) and weaker in NAWM regions (qs ¼ 0.33, P ¼ 0.037);
Quantitative MT and T2 in Acute MS Lesions
637
FIG. 3. Plots of QMTI parameters, averaged across subjects, for each month, starting at initial Gd enhancement. Error bars indicate the standard error of the mean across subjects. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]
moreover, the strength of the correlation between hT2i and T2f increased when Gd-enhancing and NAWM ROIs were pooled (qs ¼ 0.76; P < 0.0001). DISCUSSION Despite the high sensitivity of MRI to the detection of WM lesions in MS, a robust and specific method of mapping demyelination and remyelination remains elusive. Potential candidates such as myelin water imaging and QMTI have their strengths and limitations. The MT effect can only provide partial specificity to the macromolecular components of myelin as it is also sensitive to other nonmyelin semisolids. Myelin water imaging, on the other hand, probes myelin indirectly by measuring the water trapped within it and relies on the assumption that variations in myelin water content equate to variations in myelin content. Our experience shows that myelin water imaging suffers from relatively high variability, which stems mainly from the ill-posed nature of the constrained nonnegative least squares analysis and the limited number of samples of the decay curve. While threedimensional implementations of QMTI (39) and myelin water imaging (40) have recently been reported, singleslice acquisitions remain the norm. QMTI parameters F, kf, and R1f were lower in NAWM than in control WM, while T2f was higher in NAWM. The MWF and hT2i were also abnormal in NAWM when compared to control WM. The observations for F, MWF,
and hT2i were consistent with previous reports (11,13,21,22) and are likely indicative of diffuse disease. This interpretation is also consistent with previous observations of diffuse MT ratio abnormalities in NAWM (41).
FIG. 4. Percentage change in each parameter relative to NAWM. Note the relative stability of T2r. Error bars indicate the standard error of the mean across subjects, normalized to the mean across subjects and expressed in percentage. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]
638
Levesque et al.
FIG. 5. QT2 distribution metrics (left: MWF; right: hT2i), averaged across subjects, for each month starting at initial Gd enhancement. Error bars indicate the standard error of the mean, computed across subjects. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
The QMTI parameter changes that we observed in acute lesions all exceeded the range of longitudinal variability reported in WM of healthy controls (42). When compared with previous reports from our group (12,13), the F, kf, and T2r of the enhancing lesions in this study are slightly closer to normal than in chronic lesions; this can likely be attributed to less severe tissue damage in the acute lesions than in older chronic lesions. Enhancing regions had significantly lower F and significantly longer T2f than in homologous NAWM. These lesions presented with more heterogeneous parameter estimates compared to NAWM, as would be expected from pathologic heterogeneity. Deviation from NAWM values in the QMTI parameters was greatest in these lesions at the time of enhancement, consistent with acute demyelination, inflammation, and edema, that resolved over 2 to 3 months. T2r was significantly altered in these acute lesions, as reported in chronic lesions (10), but by a smaller margin (a decrease of 6% here and approximately 13% in chronic lesions). This suggests that the structural integrity of the myelin is also affected in acute, enhancing lesions, lending support to the proposed pathologic specificity of T2r (10). Perhaps the initial inflammatory tissue injury of acute lesions produces only mild changes in semisolid constituents that result in small changes in T2r, in contrast to more advanced lesions with more severe destruction of myelin. The apparent differences in the timing of recovery of the various QMTI parameters are also of interest. We suggest that the quick drop in T2f between the first and the second month postenhancement primarily reflects the resolution of inflammation, while the slower recovery in F and kf reflects the combination of remyelination and further resolution of inflammation. Since changes in F can result from a decrease in the restricted proton population or an increase in the free proton population, we evaluated the water increase in each lesion voxel relative to the contralateral homologous NAWM ROI. We used a method similar to that proposed in Vavasour et al. (17,18), in which the total spin echo signal was extrapolated to echo time ¼ 0 ms, corrected for incomplete T1 recovery, and normalized to the
corresponding value from the NAWM ROI. We observed an average increase of 9% in water content in enhancing regions when compared to NAWM. From this, we can derive that the overall decrease in F observed here (about 60%) would require a 50% decrease in macromolecular content relative to NAWM. Alternatively, if our observed changes in F were driven solely by edema, the water content would have to increase by about 125%, which is physically impossible. In short, water increases alone cannot explain the change in the F. The hT2i was sensitive to changes in the enhancing lesions and to the postenhancement partial recovery. We believe that this change in the hT2i is mostly indicative of the resolution of inflammation as the increase is much too large to be driven only by myelin water fluctuations. Furthermore, hT2i was significantly correlated with T2f. The MWF was not significantly different between lesions and NAWM and was unable to detect changes during the partial recovery of enhancing regions. A recent study examining three acute enhancing lesions (19) at 1 month and 6 months also found MWF unchanged over this period. Differences between the MWF of lesion and NAWM have been reported (20,22), suggesting the present study was underpowered, due to the small patient group, to detect this difference. In this study, the MWF was the most variable metric considered, and we believe that this is the reason why no clear conclusions can be drawn as to the comparison of MWF between NAWM and acute lesions and why no clear trend was observed over time following resolution of enhancement. Based on an ROI analysis of our control data, we evaluated the coefficient of variation of the MWF over time in individual subjects to be 13% (range, 5–25%), which is comparable to what has been reported in 5 controls by Vavasour et al. (17). Using the results published in that study, we computed the average coefficient of variation over time (averaged across subjects) to be 19%, with a range of 7–40%, over WM ROIs. Possible causes for the variability of the myelin water imaging results include (i) a signal-to-noise ratio (SNR) that is too low to allow reliable multicomponent analysis of the spin-echo data; (ii) the voxel size of our MWF images
Quantitative MT and T2 in Acute MS Lesions
(larger than many other publications), perhaps resulting in too much partial-volume contamination; (iii) the low number of patients/lesions studied, leading to a study power too low for the detection of MWF changes; and finally, (iv) the ill-posed nature of the multiexponential analysis of spin echo data (43). With respect to SNR, care was taken to maintain the SNR to the same level as other investigators when designing the study: our protocol yielded an SNR in the range of approximately 200 in the first echo, which is similar to the value we compute for the 32-echo imaging protocol used by MacKay and colleagues (15,16). Concerning points 2 and 3, similar conditions were imposed on the QMTI acquisition, yet these measurements were sensitive to the acute lesion progression, while the MWF was not. Finally, the difficult nature of multiexponential analysis, while attenuated by the regularization process, likely contributes to variability in the MWF estimates. It is most likely that this inherent variability of the MWF estimation method is responsible for the negative results in this case. In an attempt to improve the analysis of our multiecho spinecho data, these data were also processed after averaging the raw spin-echo signal within the ROIs (as suggested by Graham et al. (44)) in contrast to the voxelwise process; unfortunately, this did not result in any appreciable improvement in our measurements. Other models have been proposed for the analysis of such data, for instance a log-gaussian T2 distribution (25), but their application and comparison are beyond the scope of this study. Despite the variability of MWF estimates, correlation coefficients were computed for lesions and NAWM, in particular between F and MWF and between hT2i and T2f. F and the MWF were not significantly correlated when considering the enhancing regions or the NAWM ROIs. This observation contrasts with prior reports that the subject-mean restricted pool ratio and MWF were significantly correlated in the NAWM of patients (26) (no correlations were observed within individuals). This study also reported a positive correlation between F and MWF in lesions for combined between- and within-subject measurements. While the larger number of subjects in that study (N ¼ 19) was more likely to yield significant correlations, we suggest that the very weak correlation in this study is due to the high variability of MWF estimates. Alternatively, the weak correlation between F and the MWF could indicate that they provide largely independent and complementary information, as suggested by Tozer et al. (26). The hT2i (from QT2) and T2f (from QMTI), while not equal, were moderately but significantly correlated in Gd-enhancing regions but not in the NAWM ROIs, where the low dynamic range in T2f is the most likely explanation of the absence of correlation. This study revealed postenhancement recovery in portions of acute lesions with QMTI but was underpowered to lead to meaningful conclusions with the QT2 technique. Our general approach could be improved in a number of ways. Inclusion of a larger number of patients would certainly increase the statistical power of the study. Increasing the volumetric coverage of the imaging methods would enable the study of more acute lesions within the subject group. Optimization of the QMTI acquisition techniques (45), or the use of more recently
639
published alternative QMTI methods based on fast steady-state imaging (46), could result in improved measurements. Improved myelin water mapping might be obtained from using a volumetric T2-prepared imaging approach (47). In conclusion, QMTI parameters F, kf, and T2f have been shown to be sensitive to the dynamic changes in Gd-enhancing regions of acute lesions and to the partial recovery following resolution of enhancement. MWF did not detect these changes and was characterized by variability approximately double that of F. The hT2i estimated from multiecho spin-echo data behaved in the same way as the T2f estimated from the QMTI data. Both were greatly increased at enhancement, followed by a quick partial recovery at month 2. We suggest that the dynamics of F and the T2f obtained from QMTI reflect a quick resolution of inflammation over the first month following enhancement, followed by a slower partial remyelination process that continues into the second and third month. ACKNOWLEDGMENTS The authors thank the volunteers who participated in this study and the MRI technologists at the Montreal Neurologic Institute for acquiring the data. This work was supported by grants from the Canadian Institutes for Health Research and the Multiple Sclerosis Society of Canada. I.R.L. acknowledges fellowship support from the Natural Science and Engineering Council of Canada and the Carl Reinhardt Foundation. REFERENCES 1. Henkelman RM, Stanisz GJ, Graham SJ. Magnetization transfer in MRI: a review. NMR Biomed 2001;14:57–64. 2. Tofts PS. Quantitative MRI of the brain: measuring changes caused by disease. London: John Wiley & Sons, Ltd.; 2003. 650 p. 3. Kucharczyk W, Macdonald PM, Stanisz GJ, Henkelman RM. Relaxivity and magnetization transfer of white matter lipids at MR imaging: importance of cerebrosides and pH. Radiology 1994;192: 521–529. 4. Morrison C, Henkelman RM. A model for magnetization transfer in tissues. Magn Reson Med 1995;33:475–482. 5. Sled JG, Pike GB. Quantitative imaging of magnetization transfer exchange and relaxation properties in vivo using MRI. Magn Reson Med 2001;46:923–931. 6. Schmierer K, Tozer DJ, Scaravilli F, Altmann DR, Barker GJ, Tofts PS, Miller DH. Quantitative magnetization transfer imaging in postmortem multiple sclerosis brain. J Magn Reson Imaging 2007;26: 41–51. 7. Schmierer K, Wheeler-Kingshott CAM, Tozer DJ, Boulby PA, Parkes HG, Yousry TA, Scaravilli F, Barker GJ, Tofts PS, Miller DH. Quantitative magnetic resonance of postmortem multiple sclerosis brain before and after fixation. Magn Reson Med 2008;59:268–277. 8. Sled JG, Levesque I, Santos AC, Francis SJ, Narayanan S, Brass SD, Arnold DL, Pike GB. Regional variations in normal brain shown by quantitative magnetization transfer imaging. Magn Reson Med 2004; 51:299–303. 9. Yarnykh VL, Yuan C. Cross-relaxation imaging reveals detailed anatomy of white matter fiber tracts in the human brain. Neuroimage 2004;23:409–424. 10. Tozer DJ, Ramani A, Barker GJ, Davies GR, Miller DH, Tofts PS. Quantitative magnetization transfer mapping of bound protons in multiple sclerosis. Magn Reson Med 2003;50:83–91. 11. Davies GR, Tozer DJ, Cercignani M, Ramani A, Dalton CM, Thompson AJ, Barker GJ, Tofts PS, Miller DH. Estimation of the macromolecular proton fraction and bound pool T2 in multiple sclerosis. Mult Scler 2004;10:607–613.
640 12. Levesque I, Sled JG, Santos AC, Francis SJ, Narayanan S, Brass SD, Arnold DL, Pike GB. The role of edema and demyelination in chronic T1 black holes: a quantitative magnetization transfer study. J Magn Reson Imaging 2005;21:103–110. 13. Narayanan S, Francis SJ, Sled JG, Santos AC, Antel S, Levesque I, Brass S, Lapierre Y, Sappey-Marinier D, Pike GB, Arnold DL. Assessment of axonal injury and demyelination in the cerebral normalappearing white matter of patients with multiple sclerosis. Neuroimage 2006;29:637–642. 14. Menon RS, Allen PS. Application of continuous relaxation time distributions to the fitting of data from model systems and excised tissue. Magn Reson Med 1991;20:214–227. 15. MacKay AL, Whittall KP, Adler J, Li D, Paty D, Graeb D. In vivo visualization of myelin water in brain by magnetic resonance. Magn Reson Med 1994;31:673–677. 16. MacKay AL, Laule C, Vavasour IM, Bjarnason T, Kolind S, Ma¨dler B. Insights into brain microstructure from the T2 distribution. Magn Reson Imaging 2006;24:515–525. 17. Vavasour IM, Clark CA, Li DKB, MacKay AL. Reproducibility and reliability of MR measurements in white matter: clinical implications. Neuroimage 2006;32:637–642. 18. Vavasour IM, Li DKB, Laule C, Traboulsee AL, Moore GRW, MacKay AL. Multi-parametric MR assessment of T1 black holes in multiple sclerosis: evidence that myelin loss is not greater in hypointense versus isointense T1 lesions. J Neurol 2007;254:1653–1659. 19. Vavasour IM, Laule C, Li DKB, Oger J, Moore GRW, Traboulsee AL, MacKay AL. Longitudinal changes in myelin water fraction in two MS patients with active disease. J Neurol Sci 2009;276:49–53. 20. Vavasour IM, Whittall KP, MacKay AL, Li DKB, Vorobeychik G, Paty DW. A comparison between magnetization transfer ratios and myelin water percentages in normals and multiple sclerosis patients. Magn Reson Med 1998;40:763–768. 21. Whittall KP, MacKay AL, Li DKB, Vavasour IM, Jones CK, Paty DW. Normal-appearing white matter in multiple sclerosis has heterogeneous, diffusely prolonged T2. Magn Reson Med 2002;47:403–408. 22. Laule C, M. Vavasour I, Moore GRW, Oger J, Li DKB, Paty DW, MacKay AL. Water content and myelin water fraction in multiple sclerosis: a T2 relaxation study. J Neurol 2004;251:284–293. 23. Moore GRW, Leung E, MacKay AL, Vavasour IM, Whittall KP, Cover KS, Li DKB, Hashimoto SA, Oger J, Sprinkle TJ, Paty DW. A pathology-MRI study of the short-T2 component in formalin-fixed multiple sclerosis brain. Neurology 2000;55:1506–1510. 24. Laule C, Leung E, Li DKB, Traboulsee AL, Paty DW, MacKay AL, Moore GRW. Myelin water imaging in multiple sclerosis: quantitative correlations with histopathology. Mult Scler 2006;12:747–753. 25. Stanisz GJ, Webb S, Munro CA, Pun T, Midha R. MR properties of excised neural tissue following experimentally induced inflammation. Magn Reson Med 2004;51:473–479. 26. Tozer DJ, Davies GR, Altmann DR, Miller DH, Tofts PS. Correlation of apparent myelin measures obtained in multiple sclerosis patients and controls from magnetization transfer and multicompartmental T2 analysis. Magn Reson Med 2005;53:1415–1422. 27. Deloire-Grassin MS, Brochet B, Quesson B, Delalande C, Dousset V, Canioni P, Petry KG. In vivo evaluation of remyelination in rat brain by magnetization transfer imaging. J Neurol Sci 2000;178:10–16. 28. Lai HM, Davie CA, Barker GJ, Webb S, Tofts PS, Thompson AJ, McDonald WI, Miller DH. Serial magnetisation transfer ratios in gadolinium-enhancing lesions in multiple sclerosis. J Neurol 1997;244: 308–311.
Levesque et al. 29. Dousset V, Gayou A, Brochet B, Caille JM. Early structural changes in acute MS lesions assessed by serial magnetization transfer studies. Neurology 1998;51:1150–1155. 30. Richert ND, Ostuni JL, Bash CN, Leist TP, McFarland HF, Frank JA. Interferon beta-1b and intravenous methylprednisolone promote lesion recovery in multiple sclerosis. Mult Scler 2001;7:49–58. 31. Chen JT, Kuhlmann T, Jansen GH, Collins DL, Atkins HL, Freedman MS, O’Connor PW, Arnold DL, Canadian MS/BMT Study Group. Voxel-based analysis of the evolution of magnetization transfer ratio to quantify remyelination and demyelination with histopathological validation in a multiple sclerosis lesion. Neuroimage 2007;36: 1152–1158. 32. Chen JT, Collins DL, Atkins HL, Freedman MS, Canadian MS/BMT Study Group. Magnetization transfer ratio evolution with demyelination and remyelination in multiple sclerosis lesions. Ann Neurol 2008;63:254–262. 33. Kay I, Henkelman RM. Practical implementation and optimization of one-shot T1 imaging. Magn Reson Med 1991;22:414–424. 34. Skinner TE, Glover GH. An extended two-point Dixon algorithm for calculating separate water, fat, and B0 images. Magn Reson Med 1997;37:628–630. 35. Sled JG, Pike GB. Correction for B1 and B0 variations in quantitative T2 measurements using MRI. Magn Reson Med 2000;43:589–593. 36. Poon CS, Henkelman RM. Practical T2 quantitation for clinical applications. J Magn Reson Imaging 1992;2:541–553. 37. Whittall KP, MacKay AL. Quantitative interpretation of NMR relaxation data. J Magn Reson 1989;84:134–152. 38. Jones CK, Whittall KP, MacKay AL. Robust myelin water quantification: averaging vs. spatial filtering. Magn Reson Med 2003;50: 206–209. 39. Cercignani M, Symms MR, Schmierer K, Boulby PA, Tozer DJ, Ron M, Tofts PS, Barker GJ. Three-dimensional quantitative magnetisation transfer imaging of the human brain. Neuroimage 2005;27:436–441. 40. Ma¨dler B, MacKay AL. In-vivo 3D multi-component T2-relaxation measurements for quantitative myelin imaging at 3T. In: Proceedings of the 14th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Seattle, USA, 2006. p 2112. 41. Pike GB, Stefano ND, Narayanan S, Worsley KJ, Pelletier D, Francis GS, Antel JP, Arnold DL. Multiple sclerosis: magnetization transfer MR imaging of white matter before lesion appearance on T2weighted images. Radiology 2000;215:824–830. 42. Levesque IR, Narayanan S, Ribeiro LT, Sled JG, Arnold DL, Pike GB. Reproducibility of quantitative magnetization transfer imaging parameters. In: Proceedings of the 15th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Berlin, Germany, 2007. p 2187. 43. Istratov AA, Vyvenko OF. Exponential analysis in physical phenomena. Rev Sci Instrum 1999;70:1233–1257. 44. Graham SJ, Stanchev PL, Bronskill MJ. Criteria for analysis of multicomponent tissue T2 relaxation data. Magn Reson Med 1996;35: 370–378. 45. Cercignani M, Alexander DC. Optimal acquisition schemes for in vivo quantitative magnetization transfer MRI. Magn Reson Med 2006; 56:803–810. 46. Gloor M, Schefflliger K, Bieri O. Quantitative magnetization transfer imaging using balanced SSFP. Magn Reson Med 2008;60:691–700. 47. Oh J, Han ET, Pelletier D, Nelson SJ. Measurement of in vivo multicomponent T2 relaxation times for brain tissue using multi-slice T2 prep at 1.5 and 3 T. Magn Reson Imaging 2006;24:33–43.
