Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2412051345
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Appendix
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van der Voorn, J. P.
Right arrow Articles by van der Knaap, M. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by van der Voorn, J. P.
Right arrow Articles by van der Knaap, M. S.
(Radiology 2006;241:510-517.)
© RSNA, 2006

Pediatric Imaging

Childhood White Matter Disorders: Quantitative MR Imaging and Spectroscopy1

J. Patrick van der Voorn, MD, Petra J. W. Pouwels, PhD, Augustinus A. M. Hart, MSc, Judith Serrarens, MD, Michèl A. A. P. Willemsen, MD, PhD, Hubertus P. H. Kremer, MD, PhD, Frederik Barkhof, MD, PhD and Marjo S. van der Knaap, MD, PhD

1 From the Departments of Child Neurology (J.P.v.d.V., J.S., M.S.v.d.K.), Physics and Medical Technology (P.J.W.P.), and Radiology (F.B.), Vrije Universiteit Medical Center, De Boelelaan 1117, 1007 MB Amsterdam, the Netherlands; Department of Clinical Epidemiology and Biostatistics, Academic Medical Center, Amsterdam, the Netherlands (A.A.M.H.); and Departments of Child Neurology (M.A.A.P.W.) and Neurology (H.P.H.K.), University Medical Center Nijmegen, Nijmegen, the Netherlands. Received August 12, 2005; revision requested October 18; revision received November 18; final version accepted January 2, 2006. Supported by the Dutch Organization for Scientific Research (Netherlands Organization for Scientific Research Grant 903-42-097), the Dr W. M. Phelps Foundation for Spastics (grant 00026WO), and the Optimix Foundation for Scientific Research. Address correspondence to J.P.v.d.V. (e-mail: jp.vandervoorn{at}vumc.nl).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Purpose: To prospectively investigate whether quantitative magnetic resonance (MR) parameters, including magnetization transfer ratio (MTR), apparent diffusion coefficient (ADC), fractional anisotropy (FA), and MR spectroscopic metabolite concentrations, allow for discrimination between different types of pathologic conditions that underlie signal intensity abnormalities in white matter.

Materials and Methods: Institutional review board approval and informed consent were obtained. Forty-one patients (19 male, 22 female; mean age, 15.4 years) and 41 control subjects (25 male, 16 female; mean age, 11.3 years) were included. Twelve patients had a hypomyelinating disorder; 14, a demyelinating disorder; five, a disorder characterized by myelin vacuolation; and 10, a disorder characterized by cystic degeneration. Regions of interest were selected within the parietal white matter and were transferred to the corresponding sections of the generated ADC, FA, and MTR maps to extract quantitative measurements. Linear discriminant analysis and univariate analysis of covariance were used for statistical evaluation.

Results: Linear discriminant analysis showed that 95% of patients were correctly classified by using total creatine, choline-containing compounds, myo-inositol, MTR, and ADC. In the hypomyelination group, all MR parameters were close to normal, with the exception of elevated total creatine (P = .03) and myo-inositol (P < .001) levels and decreased MTR values (P < .001). In the demyelination group, the levels of choline-containing compounds (P = .02) and myo-inositol (P < .001) were highly elevated. In the myelin vacuolation and cystic degeneration groups, high ADC values (P < .001) and variable decreases in all MR spectroscopic metabolites were seen. MTR was significantly reduced (P < .001) in the cystic degeneration group.

Conclusion: Quantitative MR techniques can be used to discriminate between different types of white matter disorders and to classify white matter lesions of unknown origin with respect to underlying pathologic conditions.

Supplemental material: http://radiology.rsnajnls.org/cgi/content/full/241/2/510/DC1

© RSNA, 2006


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Magnetic resonance (MR) imaging is highly sensitive in the detection of white matter lesions (1). However, MR imaging has a limited specificity with regard to the pathologic conditions that underlie signal intensity abnormalities in the white matter. On MR images, highly variable pathologic changes may underlie white matter disorders such as hypomyelination, demyelination, gliosis, interstitial edema, myelin vacuolation with intramyelinic edema, cystic white matter degeneration, and diffuse infiltration by tumor cells (eg, in patients with gliomatosis cerebri). For all types of pathologic disorders, T1 and T2 relaxation times become longer, leading to nonspecifically increased signal intensity on T2-weighted MR images and decreased signal intensity on T1-weighted MR images (2).

Quantitative MR techniques, such as diffusion-tensor MR imaging, magnetization transfer imaging, and hydrogen (H1) MR spectroscopy, may provide more insight into underlying pathologic changes in the white matter (2–5). The purpose of our study was to prospectively investigate whether quantitative MR parameters, including magnetization transfer ratio (MTR), apparent diffusion coefficient (ADC), fractional anisotropy (FA), and MR spectroscopic metabolite concentrations, allow for discrimination between different types of pathologic conditions that underlie signal intensity abnormalities in white matter.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
Patients and Control Subjects
This study was performed with the informed consent of the patients and control subjects or of their parents and with the approval of the institutional ethics review board. Characteristics of patients and control subjects are listed in the Table.


View this table:
[in this window]
[in a new window]

  		Characteristics of Patients and Control Subjects

 
In a prospective study, 41 consecutive patients (19 male, 22 female; mean age, 15.4 years; age range, 8 months to 34 years) with a white matter disorder of known cause underwent quantitative MR imaging and MR spectroscopy at the Vrije Universiteit Medical Center between January 2000 and January 2005. Twelve patients had a hypomyelinating disorder. Initial diagnosis was based on MR imaging criteria (6), which comprised the results of two MR imaging examinations performed at least 12 months apart that showed an unchanging pattern of seriously deficient myelination in patients older than 18 months. In two of these patients, Pelizaeus-Merzbacher disease was diagnosed by demonstrating a mutation in the proteolipid protein gene. In two patients who died during the study, autopsy results confirmed serious hypomyelination.

Fourteen patients had a demyelinating disorder, including metachromatic leukodystrophy in 10 patients and globoid cell leukodystrophy (also called Krabbe disease) in four patients. In all patients with metachromatic leukodystrophy and globoid cell leukodystrophy, the diagnosis was proved by demonstrating deficient activity in the respective lysosomal enzymes. Five patients had megalencephalic leukoencephalopathy with subcortical cysts, which is a disorder characterized by myelin vacuolation. Ten patients had vanishing white matter disease, which is a disorder characterized by rarefaction and cystic degeneration. In patients with megalencephalic leukoencephalopathy and vanishing white matter disease, the diagnosis was genetically confirmed.

Forty-one control subjects (25 male, 16 female; mean age, 11.3 years; age range, 7 months to 36 years) were included in the study on the basis of normal MR imaging results. Seventeen of these subjects were healthy adolescent or adult volunteers (eight male, nine female) whose age range was similar to that of the patients. The health status of the control subjects was determined by evaluating medical history. Twenty-four pediatric subjects (17 male, seven female) consecutively underwent MR imaging during the study period, and no abnormalities were found. All pediatric subjects had normal results at neurologic examination, and most of them underwent MR imaging because of seizures.

MR Imaging and H1 MR Spectroscopy
All examinations were performed by using a 1.5-T MR imager (Vision; Siemens, Erlangen, Germany). The imaging protocol included sagittal T1-weighted MR images obtained by using a three-dimensional magnetization-prepared rapid acquisition gradient-echo sequence (15/4 [repetition time msec/echo time msec] and one signal acquired), transverse T2-weighted spin-echo MR images (3000/22, 3000/60, 3000/120, and one signal acquired), coronal or transverse fluid-attenuated inversion recovery images (9000/105/2200 [repetition time msec/echo time msec/inversion time msec] and one signal acquired), and transverse diffusion-weighted MR images obtained by using an echo-planar sequence with b values of 0, 500, and 1000 sec/mm2 (5100/137, 20 sections acquired, 5-mm section thickness, and 96 x 128 matrix). Automatically generated ADC maps were also obtained.

Diffusion-tensor MR imaging was performed by using a multisection echo-planar sequence, with optimized gradients used according to the method described by Jones et al (7) (reference b value of 0 sec/mm2 and eight noncollinear gradient vectors with a b value of 1044 sec/mm2). In the transverse orientation, 16 sections measuring 5 mm each were acquired (3600/123 and 128 x 128 matrix). Diffusion-tensor MR imaging analysis included eddy current correction and calculation of eigenvalues for the diffusion-tensor and FA maps.

Magnetization transfer imaging was performed with a three-dimensional fast low-angle shot sequence. Two sets of images were obtained with (MS) or without (M0) a magnetization transfer saturation pulse (7.68-msec Gaussian radiofrequency pulse with 1500-Hz off-resonance saturation). Imaging parameters included 23/4, a flip angle of 20°, and a three-dimensional slab consisting of 54 transverse sections measuring 3 mm each. MTR maps were created according to the equation MTR = (1 – MS)/M0.

MR spectroscopy was performed by one of two authors (J.P.v.d.V. or P.J.W.P., with 5 and 10 years of experience in MR spectroscopy of the brain, respectively) who used a short-echo-time stimulated-echo acquisition mode MR imaging sequence (6000/20, 10-msec mixing time, and 64 accumulations). Single-acquisition reference measurements without water suppression were additionally acquired to enable eddy current correction.

Spectra were acquired from a single volume of interest in the parietal white matter (4–6 mL). The location of the volume of interest was selected by the spectroscopist so as to avoid or minimize contamination by cerebrospinal fluid and gray matter. Furthermore, 10 single-acquisition stimulated-echo acquisition mode measurements without water suppression, with echo times ranging from 20 to 1500 msec and an intermeasurement delay of 10 seconds, were obtained. These measurements can be used to determine the fractional free water of each MR spectroscopic voxel by means of a two-component fit to the data (8). From these measurements, only the shorter T2-component was included as a separate parameter for statistical analysis.

Metabolite concentrations were calculated by using LCModel (Stephen Provencher, PhD, http://s-provencher.com/pages/lcmodel.shtml) (9) and were expressed as millimoles per liter. Volume of interest concentrations have been determined for a large number of metabolites (10), but in this study attention was focused on total creatine (tCr) (ie, creatine and phosphocreatine), total N-acetylaspartate (NAA)—N-acetylaspartylglutamate (total NAA [tNAA]), choline-containing compounds (Cho), myo-inositol (mI), lactate, glutamate, and glutamine.

Regions of interest that corresponded to the MR spectroscopic volumes of interest were transferred to the equivalent ADC, FA, or MTR maps by one of two authors (J.S. or J.P.v.d.V.), and mean ADC, FA, and MTR values were determined in these regions of interest.

Statistical Analysis
The mean ± standard deviation was determined for all groups. First, all 11 parameters, which included the seven metabolite concentrations and ADC, FA, MTR, and T2, were compared between the patient groups and the control group by using a univariate analysis of covariance, with a Bonferroni correction applied for multiple comparisons by multiplying the raw P values by a factor of 11, which was the total number of statistical tests performed. Age and sex were included as covariates in the statistical model if these variables were found to have a significant effect (SPSS for Windows, version 9.0; SPSS, Chicago, Ill). On the basis of a residual analysis on the results of the univariate analysis of covariance, it was decided whether to use a logarithmic transformation and, if so, whether to add a small constant to all values before transformation. P values of less than .05 were considered to indicate a statistically significant difference.

Subsequently, on the basis of the original or log-transformed variables and, if necessary, the standardized variables of age and/or sex, a Fisher polytomous linear discriminant analysis was applied by using S-Plus for Windows (version 6.2; Insightful, Seattle, Wash) (11) to optimally separate the four patient groups. In nine patients, FA was not measured. In three of these patients, MTR was not determined, and in one patient ADC and MTR were not determined. For the linear discriminant analysis that contained those parameters, these patients could not be included.

Linear discriminant analysis resulted in posterior probabilities for a patient coming from a particular group on the basis of the MR parameters for that patient. These probabilities were calculated from discriminant functions of the parameters (which could be either log-transformed or standardized), with one function per group. By using the results of the linear discriminant analysis, patients were classified as having the disorder that resulted in either the highest a posteriori probability or, equivalently, the highest discriminant score. The performance of the method was judged by the misclassification rates calculated with the leave-one-out cross-validation method (12).

A backward stepwise linear discriminant analysis (SAS for Windows, version 8.2; SAS Institute, Cary, NC) that was based on the partial R2 criterion (13) was used to examine which MR parameters contributed most to the correct classification. After each elimination, performance was again estimated by using the leave-one-out cross-validation method. However, the ordering of the parameters for exclusion was not incorporated during this validation, and therefore the performance of the reduced models may have been overestimated. Thus, this part of the analysis has to be considered as exploratory.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 
MTR, ADC, and FA Maps
The MTR, ADC, and FA maps that were obtained in the patient and control groups are shown in Figure 1. MTR was lower in the cystic degeneration group than in the hypomyelination, demyelination, and myelin vacuolation groups (P < .001) (Fig 2a). ADC was higher in the cystic degeneration group than in the hypomyelination and demyelination groups (P < .001) and was higher in the myelin vacuolation group than in the hypomyelination group (P < .001) (Fig 2b). FA was higher in the hypomyelination group than in the demyelination, myelin vacuolation, and cystic degeneration groups (P < .001) (Fig 2c).


Figure 1
View larger version (113K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1: Volume of interest localizations for MR spectroscopy (stimulated-echo acquisition mode, 6000/20, 10-msec mixing time, 64 accumulations) are shown on transverse T2-weighted MR images (3000/120, one signal acquired) (first image in A–E), with equivalent transverse ADC (second image in A–E), FA (third image in A–E), and MTR (fourth image in A–E) maps and corresponding spectra (right) in, A, healthy 5-year-old boy, B, 2-year-old boy with hypomyelination, C, 6-year-old boy with demyelination, D, 11-year-old girl with myelin vacuolation, and E, 3-year-old girl with cystic degeneration. In B, MR imaging parameters are close to normal, with the exception of increased tCr and mI and decreased MTR. In C, a decrease in tNAA and increase in Cho and mI are evident. In D and E, high ADC and variable decreases of all MR spectroscopy metabolites are seen. MTR is markedly reduced in C.

 

Figure 2
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a: Graphs show mean and standard deviation (error bars) for (a) MTR, (b) ADC, and (c) FA in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). ** = P < .01, *** = P < .001.

 

Figure 2
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b: Graphs show mean and standard deviation (error bars) for (a) MTR, (b) ADC, and (c) FA in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). ** = P < .01, *** = P < .001.

 

Figure 2
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2c: Graphs show mean and standard deviation (error bars) for (a) MTR, (b) ADC, and (c) FA in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). ** = P < .01, *** = P < .001.

 
Metabolites and T2
The spectra of patients from each group and the spectrum of a control subject are shown in Figure 1. The concentration of tCr was higher in the hypomyelination and demyelination groups than in the myelin vacuolation and cystic degeneration groups (P < .001 for all) (Fig 3a). The concentration of tNAA was higher in the hypomyelination group than in the demyelination, myelin vacuolation, and cystic degeneration groups (P < .001) (Fig 3b). The concentration of Cho was higher in the demyelination group than in the hypomyelination, myelin vacuolation, and cystic degeneration groups (P < .001) and was lower in the cystic degeneration group than in the hypomyelination group (P = .009) (Fig 3c). The concentration of mI was higher in the demyelination group than in the myelin vacuolation and cystic degeneration groups (P < .001), higher in the hypomyelination group than in the myelin vacuolation (P = .004) and cystic degeneration groups (P < .001), and higher in the myelin vacuolation group than in the cystic degeneration group (P = .003) (Fig 3d). The concentration of lactate was higher in the demyelination group than in the hypomyelination group (P = .027) (Fig 3e). No significant differences in glutamate (Fig 3f) and glutamine (Fig 3g) concentrations were found. T2 was higher in the cystic degeneration group than in the hypomyelination (P < .001), demyelination (P < .001), and myelin vacuolation (P = .004) groups, and T2 was higher in the myelin vacuolation group than in the hypomyelination (P < .001) and demyelination (P = .001) groups (Fig 3h).


Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (10K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3c: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3d: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3e: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3f: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3g: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 

Figure 3
View larger version (10K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3h: Graphs show mean and standard deviation (error bars) for (a) tCr, (b) tNAA, (c) Cho, (d) mI, (e) lactate, (f) glutamate, (g) glutamine, and (h) T2 in control subjects (Con) and patients with hypomyelination (HM), demyelination (DM), myelin vacuolation (MV), and cystic degeneration (CD). * = P < .05, ** = P < .01, *** = P < .001.

 
Linear Discriminant Analysis and Leave-One-Out Cross Validation
Linear discriminant analysis revealed four discriminant functions (one per group) that contained the observed MR parameters. These functions are given in Appendix E1 (http://radiology.rsnajnls.org/cgi/content/full/241/2/510/DC1). Filling in the seven MR parameters (ie, MTR, ADC, tCr, tNAA, Cho, mI, and lactate) for the analyzed patients in the discriminant functions resulted in the correct classification of most patients. In one patient, hypomyelination was classified as demyelination, and in another patient demyelination was classified as hypomyelination. The leave-one-out cross-validation method resulted in the same two misclassifications. When MTR and ADC were excluded from the linear discriminant analysis (leaving only the five MR spectroscopic metabolites), the leave-one-out cross-validation method resulted in the misclassification of eight patients. The combination of tCr, Cho, mI, MTR, and ADC resulted in the same two misclassifications that resulted from the use of all seven MR parameters, while further elimination of ADC and mI added one misclassification each. Elimination of one more parameter (MTR, Cho, or tCr) led to 10 or more misclassifications. Backward stepwise linear discriminant analysis listed the MR parameters in decreasing order of importance for classification of patient groups as follows: tCr, Cho, MTR, mI, ADC, lactate, and tNAA.

Adding glutamate, glutamine, and T2 to the linear discriminant analysis resulted in four misclassifications during leave-one-out cross validation. For the backward stepwise linear discriminant analysis, these variables were eliminated as the first (T2), fourth (glutamine), and fifth (glutamate) parameter.

Adding FA to the linear discriminant analysis increased the number of misclassifications during leave-one-out cross validation from two to four. Adding FA to tCr, Cho, mI, MTR, and ADC, however, increased the number of misclassifications during leave-one-out cross validation from two to three.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
ADVANCES IN KNOWLEDGE
 References
 
In our study, clear differences in these MR parameters were found between patients and control subjects and between patient groups. For the hypomyelination group, all MR parameters were close to normal, with the exception of elevated tCr and mI levels and decreased MTR values. The most striking findings for the demyelination group were the highly elevated levels of Cho and mI. Findings for the myelin vacuolation and cystic degeneration groups tended to be similar, with high ADC values and variable decreases for most MR spectroscopic metabolites, which were more pronounced in the cystic degeneration group. MTR values were more severely reduced in the cystic degeneration group than in the myelin vacuolation group.

We wanted to determine whether, by using these parameters, white matter abnormalities with different underlying pathologic characteristics could be discriminated. Linear discriminant analysis showed that the combination of tCr, Cho, mI, MTR, and ADC measurements resulted in only two misclassifications (95% of all patients were classified correctly), and further elimination of ADC and mI added one misclassification each.

The two consistently misclassified patients included one patient with demyelination that was classified as hypomyelination and one patient with hypomyelination that was classified as demyelination. The first patient was the only patient with metachromatic leukodystrophy who had stable disease. She had undergone bone marrow transplantation several years before participation in the study, and her disease has since been categorized as stable at clinical examination and MR imaging. Thus, she did not have active demyelination, as was suggested by the linear discriminant analysis model. The parameters in the patient with hypomyelination suggested progressive loss of myelin. In addition to the lack of myelination, demyelination and axonal degeneration can occur in patients with hypomyelination (14,15). This was the only patient with hypomyelination in whom MR imaging showed evidence of further loss of the little myelin that was left, as was correctly demonstrated by the MR parameters.

Diffusion-tensor MR imaging results and MTR values provide information about tissue microstructure. Within abnormal white matter, ADC, which measures isotropic water diffusivity, was increased and FA, which measures the degree of diffusion anisotropy, was decreased. This indicates the enhanced mobility of water molecules in all directions as a result of damage to the tissue matrix. Water mobility was highest in the white matter of patients with myelin vacuolation and cystic degeneration. This indicated the presence of rarefied white matter in patients with cystic degeneration and of spongiform white matter changes with numerous vacuoles in patients with myelin vacuolation, both of which resulted in reduced cellular density and increased water-filled spaces.

In patients with hypomyelination, diffusion anisotropy was only slightly decreased, which suggests that diffusion anisotropy does not necessarily depend on myelinated fibers. Hypomyelinated white matter contains little or no myelin but has normal axonal density, which is apparently sufficient to maintain close to normal diffusion parameters. The results of diffusion-tensor MR imaging are in concordance with the findings obtained by other researchers in several case reports (16–21).

The white matter of patients with demyelination, hypomyelination, or myelin vacuolation showed comparable decreases in MTR, thereby indicating a reduced capacity of the macromolecule-bound protons in brain tissue to exchange magnetization with the surrounding protons in free water, which probably indicates nonspecific damage to myelin and axonal membranes. The decrease in MTR was most pronounced in the white matter of patients with cystic degeneration. The results of autopsy in these patients revealed diffusely rarefied to cavitated white matter, with profound losses of oligodendrocytes, myelin sheaths, and axons, accompanied by a feeble astrogliosis and macrophage response (22). The extremely low MTR values indicate the loss of all tissue structures (23).

Patients with myelin vacuolation or demyelination had similar MTR values, but ADC values observed in patients with demyelination were lower than those observed in patients with myelin vacuolation. This finding could be explained by the dense accumulation of lipid-containing macrophages and glial cells in the white matter of patients with metachromatic leukodystrophy and globoid cell leukodystrophy, which would hinder water diffusion prior to the extensive demyelination. Low ADC values within the unaffected subcortical white matter have been observed in patients with metachromatic leukodystrophy (19).

Quantitative localized H1 MR spectroscopy of white matter in patients with myelin vacuolation revealed marked decreases in tNAA, tCr, and Cho, with close to normal values for mI that were consistent with axonal damage or loss and with astrocytic proliferation. The results of MR spectroscopy in the white matter of patients with cystic degeneration showed a decrease in all normal signals and the presence of lactate, which was compatible with the presence of cerebrospinal fluid and little brain tissue. In patients with hypomyelination, tNAA was normal, whereas mI and tCr were increased, which indicates normal axonal density and increased astrogliosis. In patients with demyelination, MR spectroscopy showed decreased tNAA that was accompanied by elevated Cho, mI, and lactate levels, which indicated axonal damage or loss, enhanced membrane turnover or accumulation of myelin breakdown products, astrogliosis, and infiltrating macrophages.

Our MR spectroscopic observations are in agreement with those obtained in previous studies that were performed on each disorder separately (14–17,24–28). Interestingly, tCr was the most important MR parameter for the classification of patient groups, which demonstrates that this metabolite should not be used as an internal reference when white matter disorders are studied.

Our study confirms the limited value of T2 measurements apart from their qualitative use at T2-weighted MR imaging to identify a white matter disorder as such. First, the difference in T2 between patients with hypomyelination and those with demyelination was not significant. Secondly, T2 did not contribute to discrimination between the four white matter disorders. Unfortunately, we did not measure T1 because we had to compromise when deciding on the MR protocol in order to limit the time for each patient in the magnetic bore.

Although our study is one of the largest reported series of patients with a white matter disorder, a limitation of our study is that the number of patients is rather small, which is related to the rarity of the diseases. This factor prevents us from taking a random sample from the population of these patients and performing a validation study in addition to leave-one-out cross validation. Another limitation of our study is that some MR parameters, such as MTR, depend on the features of the imager and the implementation of the sequence. Thus, the exact linear discriminant analysis functions that were used in our study are valid only for patients examined with our imager and not for those examined with other imagers.

In conclusion, the results of our study demonstrate that, in contrast to conventional MR techniques for which signal intensity changes are not specific, quantitative MR techniques are useful in the discrimination of different white matter disorders. As such, these parameters may help to classify unknown white matter lesions as demyelinating and hypomyelinating disorders or as disorders characterized by myelin vacuolation, rarefaction, and cystic degeneration.


   ADVANCES IN KNOWLEDGE
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
References
 


   ACKNOWLEDGMENTS
 
We thank the following physicians for referral of patients for this study: H. Stroink, C. E. Catsman-Berrevoets, J. W. Weber, E. A. J. Peeters, W. C. G. Overweg-Plandsoen, I. N. Snoeck-Streef, and K. P. J. Braun. We are grateful to the parents and the patients for their participation in the study.


   FOOTNOTES
 

Abbreviations: ADC = apparent diffusion coefficient • Cho = choline-containing compounds • FA = fractional anisotropy • mI = myo-inositol • MTR = magnetization transfer ratio • NAA = N-acetylaspartate • tCr = total creatine • tNAA = total NAA

Authors stated no financial relationship to disclose.

Author contributions: Guarantors of integrity of entire study, J.P.v.d.V., M.S.v.d.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, J.P.v.d.V., J.S.; clinical studies, J.P.v.d.V., P.J.W.P., J.S., M.S.v.d.K.; statistical analysis, J.P.v.d.V., P.J.W.P., A.A.M.H.; and manuscript editing, J.P.v.d.V., P.J.W.P., J.S., M.A.A.P.W., H.P.H.K., F.B., M.S.v.d.K.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 References
 

  1. Miller DH, Robb SA, Ormerod IE, et al. Magnetic resonance imaging of inflammatory and demyelinating white-matter diseases of childhood. Dev Med Child Neurol 1990;32:97–107.[Medline]
  2. van der Knaap MS. Magnetic resonance in childhood white-matter disorders. Dev Med Child Neurol 2001;43:705–712.[CrossRef][Medline]
  3. Barker PB, Horska A. Neuroimaging in leukodystrophies. J Child Neurol 2004;19:559–570.[Medline]
  4. Horsfield MA, Jones DK. Applications of diffusion-weighted and diffusion tensor MRI to white matter diseases: a review. NMR Biomed 2002;15:570–577.[CrossRef][Medline]
  5. Symms M, Jager HR, Schmierer K, Yousry TA. A review of structural magnetic resonance neuroimaging. J Neurol Neurosurg Psychiatry 2004;75(9):1235–1244.[Abstract/Free Full Text]
  6. van der Knaap MS, Naidu S, Pouwels PJ, et al. New syndrome characterized by hypomyelination with atrophy of the basal ganglia and cerebellum. AJNR Am J Neuroradiol 2002;23:1466–1474.[Abstract/Free Full Text]
  7. Jones DK, Horsfield MA, Simmons A. Optimal strategies for measuring diffusion in anisotropic systems by magnetic resonance imaging. Magn Reson Med 1999;42:515–525.[CrossRef][Medline]
  8. Ernst T, Kreis R, Ross BD. Absolute quantification of water and metabolites in the human brain. I. Compartments and water. J Magn Reson 1993;102:1–8.
  9. Provencher SW. Estimation of metabolite concentrations from localized in vivo proton MR spectra. Magn Reson Med 1993;30:672–679.[Medline]
  10. Pouwels PJ, Brockmann K, Kruse B, et al. Regional age dependence of human brain metabolites from infancy to adulthood as detected by quantitative localized proton MRS. Pediatr Res 1999;46:474–485.[Medline]
  11. Venables WN, Ripley BD. Modern applied statistics with S. 4th ed. New York, NY: Springer, 2002.
  12. McLachlan GJ. Discriminant analysis and statistical pattern recognition. New York, NY: Wiley, 1992.
  13. Jennrich RI. Stepwise discriminant analysis. In: Enslein K, Ralston A, Wilf H, eds. Statistical methods for digital computers. New York, NY: Wiley, 1977.
  14. Plecko B, Stockler-Ipsiroglu S, Gruber S, et al. Degree of hypomyelination and magnetic resonance spectroscopy findings in patients with Pelizaeus Merzbacher phenotype. Neuropediatrics 2003;34:127–136.[CrossRef][Medline]
  15. Pizzini F, Fatemi AS, Barker PB. Proton MR spectroscopic imaging in Pelizaeus-Merzbacher disease. AJNR Am J Neuroradiol 2003;24:1683–1689.[Abstract/Free Full Text]
  16. Sijens PE, Boon M, Meiners LC, Brouwer OF, Oudkerk M. 1H chemical shift imaging, MRI, and diffusion-weighted imaging in vanishing white matter disease. Eur Radiol 2005;15(11):2377–2379.[CrossRef][Medline]
  17. Brockmann K, Finsterbusch J, Terwey B, Frahm J, Hanefeld F. Megalencephalic leukoencephalopathy with subcortical cysts in an adult: quantitative proton MR spectroscopy and diffusion tensor MRI. Neuroradiology 2003;45:137–142.[Medline]
  18. Engelbrecht V, Scherer A, Rassek M, Witsack HJ, Modder U. Diffusion-weighted MR imaging in the brain in children: findings in the normal brain and in the brain with white matter diseases. Radiology 2002;222:410–418.[Abstract/Free Full Text]
  19. Oguz KK, Anlar B, Senbil N, Cila A. Diffusion-weighted imaging findings in juvenile metachromatic leukodystrophy. Neuropediatrics 2004;35:279–282.[CrossRef][Medline]
  20. Guo AC, Petrella JR, Kurtzberg J, Provenzale JM. Evaluation of white matter anisotropy in Krabbe disease with diffusion tensor MR imaging: initial experience. Radiology 2001;218(3):809–815.[Abstract/Free Full Text]
  21. Ono J, Harada K, Sakurai K, et al. MR diffusion imaging in Pelizaeus-Merzbacher disease. Brain Dev 1994;16:219–223.[CrossRef][Medline]
  22. van der Knaap MS, Barth PG, Gabreels FJ, et al. A new leukoencephalopathy with vanishing white matter. Neurology 1997;48:845–855.[Abstract]
  23. van Waesberghe JH, Kamphorst W, de Groot CJ, et al. Axonal loss in multiple sclerosis lesions: magnetic resonance imaging insights into substrates of disability. Ann Neurol 1999;46:747–754.[CrossRef][Medline]
  24. Brockmann K, Dechent P, Wilken B, Rusch O, Frahm J, Hanefeld F. Proton MRS profile of cerebral metabolic abnormalities in Krabbe disease. Neurology 2003;60:819–825.[Abstract/Free Full Text]
  25. Kruse B, Hanefeld F, Christen HJ, et al. Alterations of brain metabolites in metachromatic leukodystrophy as detected by localized proton magnetic resonance spectroscopy in vivo. J Neurol 1993;241:68–74.[CrossRef][Medline]
  26. De Stefano N, Balestri P, Dotti MT, et al. Severe metabolic abnormalities in the white matter of patients with vacuolating megalencephalic leukoencephalopathy with subcortical cysts: a proton MR spectroscopic imaging study. J Neurol 2001;248:403–409.[CrossRef][Medline]
  27. Zarifi MK, Tzika AA, Astrakas LG, Poussaint TY, Anthony DC, Darras BT. Magnetic resonance spectroscopy and magnetic resonance imaging findings in Krabbe's disease. J Child Neurol 2001;16:522–526.[Medline]
  28. Spalice A, Popolizio T, Parisi P, Scarabino T, Iannetti P. Proton MR spectroscopy in connatal Pelizaeus-Merzbacher disease. Pediatr Radiol 2000;30:171–175.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 Am. J. Neuroradiol.Home page
A. Bizzi, G. Castelli, M. Bugiani, P.B. Barker, E.H. Herskovits, U. Danesi, A. Erbetta, I. Moroni, L. Farina, and G. Uziel
Classification of Childhood White Matter Disorders Using Proton MR Spectroscopic Imaging
AJNR Am. J. Neuroradiol., August 1, 2008; 29(7): 1270 - 1275.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Appendix
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van der Voorn, J. P.
Right arrow Articles by van der Knaap, M. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by van der Voorn, J. P.
Right arrow Articles by van der Knaap, M. S.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE