Diffusion tensor imaging and cognitive speed in children with multiple sclerosis

Journal of the Neurological Sciences 309 (2011) 68–74

Contents lists available at ScienceDirect

Journal of the Neurological Sciences j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j n s

Diffusion tensor imaging and cognitive speed in children with multiple sclerosis A. Bethune a, b, V. Tipu b, J.G. Sled b, S. Narayanan c, D.L. Arnold c, D. Mabbott b, C. Rockel b, R. Ghassemi c, C. Till d, B. Banwell a, b,⁎ a

Division of Neurology, The Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 1X8 Research Institute, The Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 1X8 McConnell Brain Imaging Center, Montreal Neurological Institute, McGill University, 3801 University Street Montreal, Quebec, Canada H3A 2B4 d Department of Psychology, York University, 4700 Keele Street, Toronto, Ontario, Canada M3J 1P3 b c

a r t i c l e

i n f o

Article history: Received 5 May 2011 Received in revised form 9 July 2011 Accepted 14 July 2011 Available online 6 August 2011 Keywords: Pediatric Demyelination Multiple sclerosis Diffusion tensor imaging Cognition

a b s t r a c t Objectives: To compare white matter (WM) integrity in children with MS and healthy children using diffusion tensor imaging (DTI), and correlate DTI findings with disease activity, lesion burden, and cognitive processing speed. Methods: Fractional anisotropy (FA) and mean diffusivity (MD) in normal-appearing white matter (NAWM) were measured in four corpus callosum (CC), eight hemispheric regions, and the normal-appearing thalamus of 33 children and adolescents with MS and 30 age-matched healthy controls. Images were acquired on a GE LX 1.5 T scanner. DTI parameters used were 25 directions, b = 1000 s/mm 2, and 5 mm slice thickness. MS patients had T2 lesion volumes and Expanded Disability Status Scale (EDSS) scores were measured; all participants underwent two speeded cognitive tasks (Visual Matching and Symbol Digit Modalities Test (SDMT)). Results: MS participants displayed lower FA values in the genu (p b 0.005), splenium (p b 0.001) and in NAWM of bilateral parietal, temporal, and occipital lobes (p b 0.001) versus controls. FA and MD in the thalamus did not differ between groups. Higher lesion volumes correlated with reduced FA in CC and hemispheric NAWM. DTI metrics did not correlate with EDSS. FA values in CC regions correlated with Visual Matching (p b 0.001) and SDMT (p b 0.005) in MS participants only. Interpretation: DTI analyses indicate widespread NAWM disruption in children with MS—with the degree of abnormality correlating with impaired cognitive processing speed. These findings support an early onset tissue pathology in MS and illustrate its functional consequence. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Multiple sclerosis (MS) is a chronic autoimmune disorder, characterized by inflammatory demyelinating lesions and progressive neurodegenerative changes [1]. Quantitative neuroimaging methods such as diffusion tensor imaging (DTI) interrogate white matter (WM) microstructure, and therefore hold potential for identifying subtle or diffuse MS pathology, even early in the MS disease process prior to visible accumulation of lesions or atrophy. In addition to DTI evidence of loss of tissue integrity in MS lesions, DTI metrics also distinguish normal-appearing white matter (NAWM) from WM of healthy individuals [2–4]. DTI studies in pediatric MS patients have recently demonstrated abnormal findings in lesions and NAWM in pediatric MS patients [5–7] implicating loss of WM integrity as an early component of MS pathology. ⁎ Corresponding author at: Research Institute, The Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 1X8. Tel.: + 1 416 813 6660; fax: + 1 416 813 6334. E-mail address: [email protected] (B. Banwell). 0022-510X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2011.07.019

DTI quantifies the magnitude and direction of hindrance to water diffusion by axons, myelin membranes and surrounding tissue structures [8]. Two measures of water diffusion in neural tissue are commonly reported: fractional anisotropy (FA) and mean diffusivity (MD). FA represents the normalized standard deviations of all measured diffusivity components; reduced FA values occur with either increased perpendicular diffusivities or with reduced parallel diffusivities [9]. FA values range from 0 to 1, where values approaching 1 represent highly anisotropic diffusion, as found in highly organized tissues; those approaching zero reflect regions of more isotropic diffusion wherein diffusion is oriented equally in all directions, such as in cerebrospinal fluid. In patients with MS, reduced FA values relative to those detected in healthy individuals may reflect destruction of cell membranes, demyelination, and microscopic cell damage. The average magnitude of water diffusion is described by MD [10]. This scalar quantity is generally elevated in MS lesions due to degradation of myelin and axonal membranes [11,12]. Recent studies of adults with MS have reported relations between DTI metrics and T2-weighted lesion volume [13], clinical disability [4], and cognitive function, particularly on speeded tasks [13–15]. In early

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

stages of pediatric MS, cognitive deficits are reported to occur in at least 30% of children [16,17] and are particularly notable across the domains of processing speed, attention, and executive function—arguably functions closely related to white matter integrity. Reduced corpus callosum (CC) area and reduced thalamic volume measured using structural MRI is negatively correlated with cognitive performance on measures of processing speed, language, and global intelligence [18]. Correlations of DTI metrics and cognitive performance in children with MS have yet to be reported. We evaluate DTI features of hemispheric, thalamic and regional CC normal-appearing white matter (NAWM) integrity in children and adolescents with MS relative to normal WM in age- and sex-matched healthy controls. We explore the association between DTI metrics and total T2-weighted lesion volume, hypothesizing that increased brain lesion burden will compromise the integrity of hemispheric, thalamic and callosal integrity. Finally, we evaluate the functional consequences of loss of WM and thalamic integrity by correlating DTI metrics with neuropsychological measures of cognitive processing speed. 2. Methods 2.1. Subjects Patients were recruited consecutively from the Pediatric Demyelinating Disease Clinic at the Hospital for Sick Children in Toronto and were evaluated more than 4 weeks from any recent relapse or corticosteroid therapy. All MS patients met current criteria for a diagnosis of relapsing–remitting MS [19]. Healthy volunteers were recruited through local advertisement. All participants were less than 19 years of age at enrolment. Participants were excluded if they had a current or past diagnosis of a neurological disorder (other than MS) or a psychiatric or learning disorder, alcohol or illicit drug abuse, or a prior history of brain injury or concussion. Institutional Research Ethics Board approval and written informed consent from each subject and/or their guardian were obtained. 2.2. Magnetic Resonance Imaging acquisition All brain imaging data were acquired on the same GE Signa Excite MRI scanner (General Electric Healthcare, Milwaukee, WI) using a single channel quadrature headcoil. A standardized imaging protocol was employed and a rigorous quality control performed. A sagittal T1-weighted 3D spoiled gradient-recalled echo (SPGR) sequence (TR/TE = 22/8 ms, 1 signal average, 250 mm FOV, 1.5 mm slice thickness, 30° flip angle) was acquired for region of interest (ROI) definition, along with an axial proton-density/T2 weighted dual-echo sequence (TR = 3500 ms, TE= 15 ms and 63 ms, 1 signal average, 90 axial slices, 250 mm FOV, 2 mm thick contiguous slices, 256 × 256 matrix, 90° flip angle). For diffusion tensor calculation, a single shot spin-echo sequence with an EPI readout, TR= 8300 ms, TE= 79 ms, 32 contiguous axial slices, 5 mm thick, and 128 × 128 matrix was used to acquire one non-diffusion-weighted (b= 0) and 25 diffusion-weighted (b= 1000 s/mm2) images with different diffusion-encoding directions. The acquisition of 25 diffusion directions used here affords more robust tensor estimation as compared to prior studies that relied on 8 directions [6,7]. 2.3. Diffusion tensor computation Eddy-current correction was performed using the Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB) software library (FSL) [20] eddy_correct utility [21] on raw diffusionweighted images to remove spatial distortion associated with diffusion gradient onset and offset. The skull was then removed using FSL Brain Extraction Tool [22]; for the sake of robustness, each

69

axial slice from each diffusion-weighted volume was registered to the corresponding b0 slice using the flirt tool [21]; the resulting slices were reassembled as 3D volumes using Analysis of Functional Neuroimages (AFNI) [23] tools, and were then linearly registered to the b0 volume using flirt[21]. The diffusion tensor, as well as FA and MD were then computed using Camino's implementation of the RESTORE algorithm for robust estimation in the presence of outliers [24]. 2.4. ROI definition Non-linear registration of each subject's T1-weighted image to a hemispheric brain atlas [25] was performed to allow automatic template-based segmentation of bilateral frontal, temporal, parietal and occipital hemispheric regions [26]. This template-based approach was extended to automatically segment the CC, as well as four previously described [27] callosal regions of interest (ROIs), the genu, anterior body, posterior body and splenium, using the atlas for the initial definition of these ROIs. Using the Display tool (http://packages. bic.mni.mcgill.ca/), the CC objects were traced on the mid-sagittal slice of the brain atlas and, to ensure complete coverage once transformed into the native space of each subject, on 10 slices bilaterally. The thalamic ROI was identified using a template based approach, with manual correction where necessary, as previously described [18]. T2-weighted and T1-weighted images were linearly registered using minctracc[28]. Non-linear alignment between the subject's T2weighted and b0 diffusion images was performed using the Advanced Normalization Tools (ANTs) toolset; [29] this removed any additional susceptibility-related distortions present in echo-planar DTI images. Alignment of the CC and hemisphere ROIs with each subject's native DTI image space was performed using the appropriately combined transformations, allowing MD and FA values to be generated for each ROI. 2.5. Lesion segmentation Total brain (supra and infratentorial) T2-weighted lesion segmentation was performed in a semi-automated manner. An initial segmentation was performed using MNI developed automated software [30] employing a Bayesian algorithm, using multi-modal tissue intensity class-conditional posterior probability, and an anatomical and tissue class spatial probability to generate a probability map for each tissue class. The tissue classification image was then used to extract the primary lesion maps. Using the Display (http://packages.bic.mni.mcgill.ca/) visualization software, the T2weighted lesions were superimposed on the registered T1-, T2-, and PD-weighted images, carefully reviewed, and manually corrected, if necessary. The lesions were removed from the hemispheric and CC WM, as well as the thalamus, this enabled DTI analysis of NAWM and normal-appearing thalamic tissue. 2.6. Cognitive outcomes and clinical disability Two standard measures of information processing speed were completed by 26 MS patients and 29 healthy controls within 90 days of the MRI. The oral rather than motoric form of the Symbol Digit Modalities Test (SDMT) [31] was administered to avoid confounding by motor impairment. The SDMT is a speeded verbal transcription task requiring the Visual Matching of symbol and number pairings. This task was included in the study because it has been shown to be one of the most sensitive measures of information processing deficits in pediatric MS [32], and to have moderate correlations with diffusionweighted imaging techniques in adult MS [15]. The score represents the number of items completed correctly within 90 s. The Visual Matching subtest from the Woodcock–Johnson III (WJ-III) Test of

70

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

Table 1 Demographic and clinical characteristics of MS patients and healthy controls. Characteristic

MS patients

Healthy controls

No. of subjects Females, no. (%) Age at MRI assessment, y No. with cognitive data SDMT-oral WJ-III Visual Matching Age at MS onset, y Disease duration, y Median EDSS score, range Total number of relapses Disease modifying therapy ⁎⁎, no. (%) Total brain T2-LV (median, SD), mm3 Log10 total brain T2-LV (mean, SD)

33 26 (83.9) 16.1 ± 2.3 26 58.4 ± 16.0 47.0 ± 8.1 12.2 ± 3.6 3.4 ± 3.5 1.0 (0–6.0) 3.5 ± 3.1 25 (75.8%)

30 24 (80.0) 15.6 (2.0) 29 65.2 ± 10.1 54.1 ± 4.7 – – – – –

4962.4 ± 11201.2 3.6 ± 0.59

– –

p Value

0.91 0.34 0.06 7 b 0.001

Within the MS group, DTI metrics within lesions were compared to the surrounding NAWM of the region. Associations between clinical variables (EDSS score, number of relapses, and total brain T2-weighted lesion volume) and the ROI-averaged DTI metrics throughout the CC, total supratentorial hemispheric WM, and thalami were examined using Spearman and Pearson correlations, respectively. To account for the different numbers of voxels within each ROI, the weighted averages of FA and MD values were used. The significance threshold was set at p b 0.01 to minimize Type I errors. All p values are reported as two-tailed. 3. Results 3.1. Sample characteristics

Mean and standard deviation are shown for each measure, except where indicated. EDSS = Expanded Disability Status Scale; SDMT = Symbol Digit Modalities Test (total score); WJ-III = Woodcock Johnson Test, Third edition (total score indicated). ⁎⁎ Disease modifying therapy: interferon: 16 patients, glatiramer acetate: 9 patients.

Cognitive Abilities [33], which is a timed paper-and-pencil visual search task, was used as a measure of visuo-perceptual speed. This test requires the individual to identify and circle two identical numbers that increase in complexity, from single to triple digit series, within a row of six numbers. The score represents the number of correct matches made within 3 min. All patients were examined at the time of testing by the same neurologist (BB) and an Expanded Disability Status Scale (EDSS) [34] score was assigned as an index of clinical disability. 2.7. Statistical analysis A Student's t-test and χ 2 test were used to examine group age and gender differences, respectively. A repeated measures analysis of variance (ANOVA) was used to compare the MS patients and the matched controls in terms of FA and MD values in normal-appearing white matter (NAWM) for the four CC regions and for the frontal, parietal, temporal, and occipital ROI bilaterally. The within-subjects factor was brain region and the between-subjects factors included age at MRI scan (covariate) and group (MS versus HC). As groups were matched for sex, sex was not included as a covariate. The dependent variables were mean FA and MD in the CC, thalamic (averaged right and left) and hemispheric ROIs. Post-hoc analyses examined group differences on DTI metrics for each ROI. For the cognitive and DTI analyses, Pearson product correlations accounting for age at MRI were conducted, separately for each group.

Thirty-six MS patients and 30 controls were enrolled. Scans from 3 MS patients were excluded due to motion or dental artifact. Demographic characteristics for each group and clinical features of the MS participants are shown in Table 1. Groups were well-matched for sex and age. The majority of MS patients were receiving some form of disease-modifying therapy. 3.2. DTI metrics: effects of brain region, group, and age Results of the repeated measures ANOVA are shown in Supplementary Table 1. The interaction of Region × Age was not significant for either DTI metric in the CC or hemispheres, indicating that FA and MD values in NAWM (lesions excluded) did not change as a function of our participant age range. A weak correlation between thalamic FA and age exists (FA: r = 0.483, p = 0.008, MD: r = 0.263, p = 0.172). For the CC, the effect of Region was significant indicating that the FA and MD values differ across the different CC segments (FA: Wilks Λ (3, 58) = 7.52, p b .001; MD; Wilks Λ (3, 58) = 3.37, p = .024). Post-hoc analysis (Least Significant Difference) revealed that the splenium and genu showed the highest FA and lowest MD values whereas the mid-body sections showed the lowest FA and highest MD values. These regional CC DTI differences were noted for both the HC and MS groups (as outlined in Table 2). With regard to hemispheric lobe differences, a significant effect of Region was revealed for both FA and MD lobar values (FA: Wilks Λ (7, 54) = 6.86. p b .001, MD: Wilks Λ (7, 54) = 3.57, p = .003). Post-hoc analyses conducted with the entire sample revealed higher FA in the left frontal and left temporal regions compared to all other regions. For both the HC and MS groups, MD values were significantly higher in the occipital lobe compared to all other regions. The NAWM (lesions removed) FA values were lower in the MS group relative to HC, in the genu and splenium (p b 0.01), although not in the mid-body segments of the CC (p N 0.05) (Table 2). For the MD values, MS patients showed higher MD in the NAWM than healthy

Table 2 Means and standard deviations for diffusion tensor metrics by callosal region, hemispheric lobe and thalamus for multiple sclerosis (MS) and healthy control (HC) groups. Region

Genu Anterior body Posterior body Splenium Frontal–L Frontal–R Parietal–L Parietal–R Temporal–L Temporal–R Occipital–L Occipital–R Thalamus–R, L average

NAWM MD (mm2/s × 10− 3)

NAWM FA MS mean FA (SD)

HC mean FA (SD)

P value

% difference

MS mean MD (SD)

HC mean MD (SD)

P value

% difference

0.494 0.321 0.281 0.539 0.297 0.293 0.281 0.286 0.288 0.282 0.202 0.209 0.241

0.551 0.352 0.318 0.608 0.311 0.306 0.307 0.310 0.315 0.309 0.226 0.232 0.244

0.002 0.106 0.068 b 0.001 0.019 0.040 b 0.001 0.001 b 0.001 b 0.001 b 0.001 b 0.001 0.265

− 10.3 − 8.8 − 11.1 − 11.4 − 4.5 − 4.2 − 9.2 − 7.8 − 8.6 − 8.7 − 10.6 − 9.9 − 1.2

1.065 1.182 1.242 1.011 0.849 0.849 0.847 0.843 0.883 0.884 0.969 0.962 0.893

0.971 1.091 1.127 0.947 0.826 0.825 0.826 0.823 0.858 0.865 0.936 0.925 0.866

0.017 0.024 0.004 0.026 0.050 0.039 0.073 0.063 0.008 0.060 0.066 0.073 0.080

9.4 7.5 9.5 6.4 2.9 2.9 2.5 2.4 2.1 2.2 3.5 4.0 3.0

(0.087) (0.071) (0.075) (0.064) (0.027) (0.029) (0.028) (0.031) (0.030) (0.028) (0.021) (0.028) (0.014)

(0.052) (0.079) (0.086) (0.052) (0.018) (0.019) (0.019) (0.022) (0.018) (0.021) (0.024) (0.023) (0.013)

(0.183) (0.189) (0.176) (0.115) (0.056) (0.054) (0.053) (0.050) (0.047) (0.046) (0.072) (0.094) (0.068)

(0.109) (0.116) (0.125) (0.107) (0.028) (0.029) (0.031) (0.029) (0.021) (0.027) (0.068) (0.056) (0.042)

FA, fractional anisotropy; MD, mean diffusivity (expressed in units of mm2/s × 10− 3); NAWM, normal-appearing white matter; mean and standard deviation are shown for each measure.

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

controls' WM in each CC region; however, this reached significance only in the posterior body (p b 0.01). Regarding the different hemispheric lobes, FA showed a main effect of Group (MS versus HC, p b 0.001) whereas MD only trended towards significance (p = 0.03). As shown in Table 2, FA values in the NAWM were lower in the MS group compared with the normal WM of controls for all hemispheric regions (p b 0.01) except the frontal lobes. For MD values, the only significant group difference was found in the left temporal lobe. The thalamic FA (F(1, 60) = 1.266, p = 0.265) and MD values (F(1, 60) = 3.180, p = 0.080) were not statistically different between MS and HC participants, as shown in Table 2. DTI results from normalappearing thalamus tissue (with lesions removed) did not differ from the total thalamus (Pearson r = 1.000, p b 0.000). Therefore we report only the normal-appearing thalamic tissue. As might be expected, lesional tissue showed FA values (DTI metrics within lesions provided in Supplementary Table 2) that were lower and MD values that were higher than the corresponding NAWM values in all MS patients. 3.3. Relation of total brain lesion volume with DTI metrics The median total brain T2 lesion volume for the MS patients was 4962.4 mm 3 (range: 255.6–53259.1) (Table 1). Illustration of the relationship between the total brain T2 lesion volume and average FA of NAWM of the CC is shown in Fig. 1a. The average CC FA decreases

71

with increasing brain lesion load. Of interest, moderate correlations between T2 lesion volume and FA were observed specifically in the genu (r = −0.469, p b 0.01) and splenium (r = −0.583, p b 0.005). MD was positively associated with the splenium lesion volume (r = 0.572, p b 0.005). The small volume of lesions within the CC itself did not alter the FA or MD values in each ROI (data not shown). Associations between total brain lesion volume and average hemispheric DTI values were observed (FA: r = −0.488, p = 0.005 and MD: r = 0.542, p = 0.002). The relation of the average hemispheric FA values and lesion volume are shown in Fig. 2. No relationship was found between thalamic FA and total brain T2 lesion volume. (FA: r = −0.195, p = 0.385; MD: r = 0.173, p = 0.440). 3.4. Relation between MRI metrics and processing speed Fig. 2 highlights the difference in performance on the two processing speed tests between participant groups. The relationship between performance on the processing speed tasks and CC DTI metrics is also evident. Relative to controls, MS patients demonstrated slower visuoperceptual speed on the Visual Matching test [Visual Matching: t (53)=−3.99, pb 0.001]. Table 3 shows the Pearson product correlations between FA and MD and the two cognitive tests, after controlling for age at scan. The relationship of DTI metrics to processing speed differed between groups. In the MS group, faster processing speed was correlated with higher FA values throughout lobar regions, (particularly in the right hemisphere, all r values N0.50, p b 0.01). In the CC, faster SDMT performance was most strongly associated with higher FA in the genu, whereas faster performance on the Visual Matching subtest was strongly associated with higher FA across all CC segments. The association between lower MD and faster processing speed was significant for 3 of 4 CC segments on both cognitive measures. The control group showed associations between DTI and cognitive speed in the occipital lobe only, with higher FA and lower MD being correlated with SDMT. FA in the thalamus is significantly correlated with SDMT (r = 0.596, p b 0.005), and approaches significance with Visual Matching (r= 0.485, p b 0.05) among MS participants (Table 4), but not controls. MD was not correlated with either processing speed measure in both groups. The association between log total brain T2-weighted lesion volume in the MS group and cognitive processing speed did not reach significance (SDMT: r = −0.19, p = 0.35; Visual Matching: r = −0.23, p = 0.27). 3.5. Relation of clinical variables of the MS group with DTI metrics Correlates of clinical variables and outcome measures were examined against DTI metrics in NAWM. In order to account for multiple comparisons, a conservative p value of 0.01 was used; however this does mask several trends that exist (see Table 4). Lower FA trended with longer disease duration, and earlier age of onset. FA and MD were not correlated with EDSS, although hemispheric MD approached significance (p = 0.02). Higher number of relapses also trended toward correlations with lower FA in the CC (r = −0.368, p = 0.035), and higher hemispheric WM MD (r = 0.326, p = 0.045). As shown in Table 4, the clinical variables did not show correlations with thalamic DTI values. 4. Discussion

Fig. 1. a: Whole brainT2 lesion volume versus average FA of corpus callosal NAWM. Increasing T2 lesion volumes appear inversely related to FA values in the NAWM of the CC. (Spearman r = − 0.445, p = 0.009). Ninety-five percent confidence intervals for the fitted line are shown in grey. b: Whole brain T2 lesion volume versus the average FA of hemispheric NAWM. Whole brain T2 lesion volumes also appear inversely proportional to lower FA values in hemispheric NAWM. (Spearman r = − 0.431, p = 0.016). Ninetyfive percent confidence intervals for the fitted line are shown in grey.

DTI evidence of disruption in NAWM clearly distinguishes children and adolescents with MS from age-matched healthy children. FA reductions and elevated MD values in the MS patients were evident throughout the CC and hemispheric regions, indicating widespread disruption in WM integrity. The degree of FA reduction or MD increase

72

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

Fig. 2. Relationship of average CC DTI values with Symbol Digit Modality Test (SDMT) and Visual Matching Performance. The degree of WM disruption in the MS participants' CC appears to be strongly linked with performance on tests of information processing speed. Lower FA values, indicative of more CC damage, lend to poorer performance on SDMT. A relationship was not observed in the control participants.

in the MS group correlated with reduced cognitive processing speed, confirming a negative functional impact of MS on white matter pathways, even in young MS patients. Our findings build on recent observations from DTI in pediatric MS patients [5–7]. Preliminary tract-based DTI studies in pediatric MS patients showed reduced FA and increased MD values in three of ten specific white matter pathways [5]. The expected trends of reduced FA and increased MD were observed in our MS participants; the degree of FA reduction reached significance in 8 regions, whereas fewer regions showed MD increases. While FA is highly sensitive to WM damage, it is less specific in differentiating axonal injury from demyelination, as Table 3 Summary of exploratory clinical correlates with FA and MD in MS patients. Region (normal- Normal-appearing tissue correlation coefficient, significance level appearing tissue) Disease Age at onseta EDSSb durationa

Number of attacksb

CC FA MD

− 0.306, 0.083 0.333, 0.058 − 0.212, 0.235 − 0.368, 0.035 0.376, 0.031 − 0.271, 0.127 0.271, 0.127 0.315, 0.075

Hemispheric FA MD

− 0.405, 0.024 0.376, 0.037 − 0.287, 0.118 − 0.346, 0.057 0.382, 0.034 − 0.234, 0.205 0.271, 0.127 0.408, 0.023

Thalamus FA MD

− 0.205, 0.305 − 0.205, 0.305 − 0.149, 0.487 − 0.128, 0.551 0.193, .334 0.193, 0.334 0.263, 0.214 0.138, 0.520

CC NAWM n = 33, hemispheric NAWM n = 31, thalamus NA tissue n = 30. a Pearson partial correlation coefficient. b Spearman correlation coefficient.

reductions in FA are attributable to both increased perpendicular diffusion and reduced parallel diffusion [9]. In contrast, changes in MD values reflect changes in diffusion restriction without reference to whether the restriction is in a particular direction. Thus, MD is inherently less sensitive to change than FA, which may explain why fewer regions displayed significant increases in MD values. A recent study of 38 pediatric relapsing–remitting (RR)MS patients, 10 children with a first demyelinating attack (clinically isolated syndrome, CIS), and 15 pediatric controls, found FA and MD in NAWM of children with RRMS altered relative to WM of the pediatric controls [6]. Interestingly, the difference in DTI metrics was not evident between the children with CIS and healthy controls, despite a similar T2 lesion burden in the CIS and RRMS cohorts; this suggests T2 lesion burden is not directly linked to the degree of DTI abnormality. Our examination of T2 lesion burden and DTI features, does, however, indicate strong correlations between whole brain T2 lesion volume and average CC FA (Fig. 1a), as well as between T2-lesion volume and average supratentorial hemispheric NAWM FA (Fig. 2). Similarly, correlations between total brain T2 lesion volume and DTI measures in adults with MS have been identified [35]. Our patients displayed markedly strong correlations of total lesion burden with FA in the genu and splenium. These are CC regions with high densities of interhemispheric WM tracts; thus the impact of focal MS lesions on WM integrity may be particularly evident here. Varying fiber volume fraction (the proportion of a voxel filled with axonal fibres) across the CC may contribute to higher FA values observed in the genu and splenium [36]. Our a priori hypothesis was that the impact of MS on WM integrity would differ across brain regions. As primary myelination proceeds in a caudal–rostral manner [37–39], we further hypothesized that ongoing myelin maturation in the frontal lobes in children and adolescents might convey a degree of protection from MS related WM insult. As shown in

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

73

Table 4 Pearson partial correlations between DTI metrics and performance on processing speed measures for the MS and healthy control participants, controlling for age. Region of interest

CC Genu Anterior body Posterior body Splenium

MS group (n = 26) Pearson correlation value

Healthy controls (n = 29) Pearson correlation value

SDMT

SDMT

Visual matching

Visual matching

FA

MD

FA

MD

FA

MD

FA

.575⁎⁎ .460 .392 .435

− .572⁎⁎ − .596⁎⁎ − .612⁎⁎ − .408

.745⁎⁎⁎ .694⁎⁎⁎ .523⁎⁎ .558⁎⁎⁎

− .652⁎⁎⁎ − .566⁎⁎ − .490 − .535⁎⁎

.195 .249 .144 .261

− .368 − .399 − .471 − .352

.033 .103 .002 − .151

.003 .083 .043 .053

.540⁎ .610⁎⁎ .595⁎⁎ .613⁎⁎

.663⁎⁎ .689⁎⁎⁎ .684⁎⁎⁎ .688⁎⁎⁎ .596⁎⁎ .661⁎⁎⁎ .722⁎⁎⁎ .766⁎⁎⁎

− .475 − .525 − .524 − .603⁎⁎ − .423 − .622⁎⁎ − .678⁎⁎⁎ − .701⁎⁎⁎

.292 .276 .344 .221 .300 .138 .586⁎⁎ .513⁎⁎

− .362 − .392 − .465 − .348 − .150 − .133 − .666⁎⁎⁎ − .493⁎⁎

− .183 − .288 − .180 − .268 − .147 − .079 .018 − .134

.009 .089 .049 .057 .132 .113 − .278 − .227

.485

− .417

.120

− .083

− .072

− .234

Hemispheric brain regions L. frontal R. frontal L. parietal R. parietal L. temporal R. temporal L. occipital R. occipital

.512 .511 .640⁎⁎ .604⁎⁎

− .406 − .450 − .447 − .517 − .481 − .650⁎⁎⁎ − .505 − .360

Thalamus

.596⁎⁎

− .257

MD

⁎ p b 0.01. ⁎⁎ p b 0.005. ⁎⁎⁎ p b 0.001.

Table 2 relative to other brain regions, the degree of FA reduction in the frontal lobes was indeed less notable than the degree of FA difference between the MS and controls groups in other hemispheric regions. We provide DTI evidence that loss of NAWM integrity in children with MS has a functional correlate. Specifically, lower FA and higher MD in almost all regional structures (CC and thalamus) and hemispheric ROIs were associated with reduced visuo-perceptual speed on the Visual Matching subtest, reflecting the dependence of efficient information transfer between and within hemispheres on intact WM pathways. Slower performance on the SDMT was also robustly correlated with lower FA and higher MD for most ROIs in the MS patients. In contrast, FA and MD in the healthy controls did not tend to correlate with processing speed measures, likely reflecting the limited range of FA and MD values in healthy WM and supporting the proficiency of such pathways for performance on speeded tasks in healthy children. In contrast to the strong relationships between DTI metrics and NAWM (hemispheric and callosal), we demonstrate only equivocal DTI relationships between thalamic tissue and processing speed in MS and HC participants. This modest relationship is in contrast to the robust correlations between thalamic volume and more global measures of cognitive performance in the same cohort [18]. Inability to detect DTI abnormalities in the thalamus of our MS group may reflect a reduced sensitivity of FA metrics in brain structures with lower white matter content. Interestingly, in a study of thalamic DTI and cognitive performance in adults with MS, increased FA values were associated with cognitive impairment [40]. Increased FA values in the thalamus, rather than the decreased FA found in CC and hemispheric NAWM, in MS patients could conceptually reflect either that (i) grey matter has lower FA than oriented WM, and thus preferential GM loss (as reflected by thalamic atrophy) could increase the proportion of WM in the thalamic ROIs and thus lead to an increase in FA, or (ii) degeneration of crossing WM fiber results in an increase of the anisotropy of diffusion in the remaining fiber, thus increasing FA. Therefore our observation of comparable thalamic FA values in MS and HC participants may reflect a cancelation among competing factors rather than preservation of healthy tissue structure. We were unable to perform an analysis of DTI metrics in cortical grey matter. Careful evaluation of the DTI resolution indicated that partial volume contamination with adjacent WM would preclude accurate estimates. We intend to refine our DTI methods in our future work to address this important point. While we feel that DTI analysis of cortical grey matter will enhance potential correlations between

overall cognitive measures and DTI findings, we emphasize processing speed in our present work which is perhaps more influenced by the densely myelinated WM pathways (hemispheric and callosal) analyzed in this present study. Examination of clinical variables revealed few correlations with DTI metrics. Recent adult literature indicates correlations between increasing EDSS score and decreasing FA within the pyramidal tracts, and in the splenium of the CC [41], which was not found in our work. The use of a conservative p value in clinical investigations, along with a low level of disability (and thus low EDSS scores) in the MS participants likely limits the potential for correlations between DTI metrics and some clinical measures. Widespread disruption of WM pathways is evident in children with MS. This loss of WM integrity is measurable despite the young age of pediatric MS patients, limited disease duration and an agerelated capacity for active myelination. That this loss of WM integrity correlates with impaired cognitive processing speed highlights the deleterious impact of MS in the maturing brain, and emphasizes the need for neuroprotective strategies. Supplementary materials related to this article can be found online at doi:10.1016/j.jns.2011.07.019. Acknowledgements This study was funded by the Canadian Institutes of Health Research, the Multiple Sclerosis Society of Canada, and by the Multiple Sclerosis Society of Canada Scientific Research Foundation. The authors would also like to acknowledge the invaluable assistance of Ms. Melissa McGowan, Ms. Julie Colman, and the staff of the McConnell Brain Imaging Center at the Montreal Neurological Institute. This study would not have been possible without the cooperation of the children and their families. References [1] Trapp BD, Peterson J, Ransohoff RM, Rudick R, Mork S, Bo L. Axonal transection in the lesions of multiple sclerosis [see comments]. N Engl J Med 1998;338:278–85. [2] Roosendaal SD, Geurts JJ, Vrenken H, Hulst HE, Cover KS, Castelijns JA, et al. Regional DTI differences in multiple sclerosis patients. Neuroimage 2009;44:1397–403. [3] Hasan KM, Gupta RK, Santos RM, Wolinsky JS, Narayana PA. Diffusion tensor fractional anisotropy of the normal-appearing seven segments of the corpus callosum in healthy adults and relapsing–remitting multiple sclerosis patients. J Magn Reson Imaging 2005;21:735–43.

74

A. Bethune et al. / Journal of the Neurological Sciences 309 (2011) 68–74

[4] Sigal T, Shmuel M, Mark D, Gil H, Anat A. Diffusion tensor imaging of corpus callosum integrity in multiple sclerosis: correlation with disease variables. J Neuroimaging 2010 Dec 1 [Electronic publication ahead of print]. [5] Vishwas MS, Chitnis T, Pienaar R, Healy BC, Grant PE. Tract-based analysis of callosal, projection, and association pathways in pediatric patients with multiple sclerosis: a preliminary study. AJNR Am J Neuroradiol 2010;31:121–8. [6] Absinta M, Rocca MA, Moiola L, Ghezzi A, Milani N, Veggiotti P, et al. Brain macro- and microscopic damage in patients with paediatric MS. J Neurol Neurosurg Psychiatry 2010. [7] Tortorella P, Rocca MA, Mezzapesa DM, Ghezzi A, Lamantia L, Comi G, et al. MRI quantification of gray and white matter damage in patients with early-onset multiple sclerosis. J Neurol 2006;253:903–7. [8] Beaulieu C. The basis of anisotropic water diffusion in the nervous system—a technical review. NMR Biomed 2002;15:435–55. [9] Budde MD, Kim JH, Liang HF, Russell JH, Cross AH, Song SK. Axonal injury detected by in vivo diffusion tensor imaging correlates with neurological disability in a mouse model of multiple sclerosis. NMR Biomed 2008;21:589–97. [10] Basser PJ, Pierpaoli C. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B 1996;111:209–19. [11] Song SK, Yoshino J, Le TQ, Lin SJ, Sun SW, Cross AH, et al. Demyelination increases radial diffusivity in corpus callosum of mouse brain. Neuroimage 2005;26:132–40. [12] Werring DJ, Clark CA, Barker GJ, Thompson AJ, Miller DH. Diffusion tensor imaging of lesions and normal-appearing white matter in multiple sclerosis. Neurology 1999;52:1626–32. [13] Lin X, Tench CR, Morgan PS, Niepel G, Constantinescu CS. ‘Importance sampling’ in MS: use of diffusion tensor tractography to quantify pathology related to specific impairment. J Neurol Sci 2005;237:13–9. [14] Rovaris M, Iannucci G, Falautano M, Possa F, Martinelli V, Comi G, et al. Cognitive dysfunction in patients with mildly disabling relapsing–remitting multiple sclerosis: an exploratory study with diffusion tensor MR imaging. J Neurol Sci 2002;195:103–9. [15] Benedict RH, Bruce J, Dwyer MG, Weinstock-Guttman B, Tjoa C, Tavazzi E, et al. Diffusion-weighted imaging predicts cognitive impairment in multiple sclerosis. Mult Scler 2007;13:722–30. [16] Amato MP, Goretti B, Ghezzi A, Lori S, Zipoli V, Portaccio E, et al. Cognitive and psychosocial features of childhood and juvenile MS. Neurology 2008;70:1891–7. [17] Banwell BL, Anderson PE. The cognitive burden of multiple sclerosis in children. Neurology 2005;64:891–4. [18] Till C, Ghassemi R, Aubert-Broche B, Kerbrat A, Collins DL, Narayanan S, et al. MRI correlates of cognitive impairment in childhood-onset multiple sclerosis. Neuropsychology 2011;25:319–32. [19] Polman CH, Reingold SC, Edan G, Filippi M, Hartung HP, Kappos L, et al. Diagnostic criteria for multiple sclerosis: 2005 revisions to the “McDonald Criteria”. Ann Neurol 2005;58:840–6. [20] Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TE, Johansen-Berg H, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23(Suppl 1):S208–19. [21] Jenkinson M, Smith S. A global optimisation method for robust affine registration of brain images. Med Image Anal 2001;5:143–56. [22] Smith SM, Zhang Y, Jenkinson M, Chen J, Matthews PM, Federico A, et al. Accurate, robust, and automated longitudinal and cross-sectional brain change analysis. Neuroimage 2002;17:479–89.

[23] Cox RW, Hyde JS. Software tools for analysis and visualization of fMRI data. NMR Biomed 1997;10:171–8. [24] Chang LC, Jones DK, Pierpaoli C. RESTORE: robust estimation of tensors by outlier rejection. Magn Reson Med 2005;53:1088–95. [25] Fonov V, Evans A, McKinstry RC, Almli CR, Collins D. Unbiased nonlinear average ageappropriate brain templates from birth to adulthood. Neuroimage 2009;47:S102. [26] Collins D, Evans A. ANIMAL: validation and applications of non-linear registrationbased segmentation. Int J Pattern Recognit Artif Intell 1997;8:1271–94. [27] Mabbott DJ, Noseworthy MD, Bouffet E, Rockel C, Laughlin S. Diffusion tensor imaging of white matter after cranial radiation in children for medulloblastoma: correlation with IQ. Neuro Oncol 2006;8:244–52. [28] Collins DL, Neelin P, Peters TM, Evans AC. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J Comput Assist Tomogr 1994;18:192–205. [29] Avants BB, Epstein CL, Grossman M, Gee JC. Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Med Image Anal 2008;12:26–41. [30] Francis S. Automatic lesion identification in MRI of multiple sclerosis patients; division of neuroscience. McGill University: Department of Neurology and Neurosurgery; 2004. [31] Smith A. Symbol digit modalities test (SDMT). In: Smith A, editor. Los Angeles: Western Psychological Services; 1991. [32] Portaccio E, Goretti B, Lori S, Zipoli V, Centorrino S, Ghezzi A, et al. The brief neuropsychological battery for children: a screening tool for cognitive impairment in childhood and juvenile multiple sclerosis. Mult Scler 2009;15:620–6. [33] Woodcock R, McGrew K, Mather N. Woodcock–Johnson III tests of achievement. Riverside Publishing: Itasca, IL; 2001. [34] Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 1983;33:1444–52. [35] Filippi M, Cercignani M, Inglese M, Horsfield MA, Comi G. Diffusion tensor magnetic resonance imaging in multiple sclerosis. Neurology 2001;56:304–11. [36] Stikov N, Perry LM, Mezer A, Rykhlevskaia E, Wandell BA, Pauly JM, et al. Bound pool fractions complement diffusion measures to describe white matter micro and macrostructure. Neuroimage 2011;54:1112–21. [37] Paus T, Collins DL, Evans AC, Leonard G, Pike B, Zijdenbos A. Maturation of white matter in the human brain: a review of magnetic resonance studies. Brain Res Bull 2001;54:255–66. [38] Klingberg T, Vaidya CJ, Gabrieli JD, Moseley ME, Hedehus M. Myelination and organization of the frontal white matter in children: a diffusion tensor MRI study. Neuroreport 1999;10:2817–21. [39] Schmithorst VJ, Wilke M, Dardzinski BJ, Holland SK. Correlation of white matter diffusivity and anisotropy with age during childhood and adolescence: a crosssectional diffusion-tensor MR imaging study. Radiology 2002;222:212–8. [40] Tovar-Moll F, Evangelou IE, Chiu AW, Richert ND, Ostuni JL, Ohayon JM, et al. Thalamic involvement and its impact on clinical disability in patients with multiple sclerosis: a diffusion tensor imaging study at 3 T. AJNR Am J Neuroradiol 2009;30:1380–6. [41] Giorgio A, Palace J, Johansen-Berg H, Smith SM, Ropele S, Fuchs S, et al. Relationships of brain white matter microstructure with clinical and MR measures in relapsing–remitting multiple sclerosis. J Magn Reson Imaging 2010;31:309–16.