| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Neuroradiology |
1 From the Departments of Radiology (K.V., L.V., N.L.), Oncology (L.K., A.M.H., H.J.), Child Neurology, Hospital for Children and Adolescents (E.G., M.L.G.), Neurosurgery (G.B.), and Pathology (A.P.), Haartman Institute, Helsinki University Central Hospital, Haartmaninkatu 4, PO Box 180, 00029 Helsinki, Finland. Received December 31, 2002; revision requested March 10, 2003; final revision received July 10; accepted August 6. Supported by a HUCH Special Federal Grant TYH0227. K.V. supported by grants from Nylands Nation and the Kurt and Doris Palander Foundation. Address correspondence to K.V. (e-mail: kim.vuori@oriola.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Eighteen patients with seizures and a cortical brain lesion on MR images were studied with proton MR spectroscopy. A metabolite ratio analysis was performed, and the metabolite signals in the lesion core were compared with those in the contralateral centrum semiovale and in the corresponding brain sites in 18 control subjects to separately obtain the changes in N-acetylaspartate (NAA), choline-containing compounds (Cho), and creatine-phosphocreatine (Cr). Ten patients had a low-grade glioma (three, oligodendrogliomas; three, oligoastrocytomas; three, astrocytomas; and one, pilocytic astrocytoma), and eight had FCDM (five, focal cortical dysplasias and three, dysembryoplastic neuroepithelial tumors). Linear discriminant analysis and Student t test were used for statistical comparisons.
RESULTS: Loss of NAA and increase of Cho were more pronounced in low-grade gliomas than in FCDMs (NAA, -72% ± 15 [± SD] vs -29% ± 22, P < .001; Cho, 117% ± 56 vs 21% ± 66, P < .01). Changes in NAA and Cho helped differentiate low-grade gliomas from FCDMs, and changes in Cho and Cr helped differentiate astrocytomas from oligodendrogliomas and oligoastrocytomas. Metabolite NAA/Cho and NAA/Cr ratios helped differentiate low-grade gliomas from FCDMs but did not differentiate glioma subtypes.
CONCLUSION: MR spectroscopy allows distinction between low-grade gliomas and FCDMs and between low-grade glioma subtypes. Metabolite changes are more informative than are metabolite ratios.
© RSNA, 2004
Index terms: Brain neoplasms, diagnosis, 13.1559, 13.363, 13.3635, 13.3639 • Brain neoplasms, MR, 13.121413, 13.121415, 13.12143 • Brain neoplasms, MR spectroscopy, 13.121411, 13.12145 • Magnetic resonance (MR), spectroscopy, 13.121411, 13.12145
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MR spectroscopy has been used to differentiate neurofibromatosis type 1–associated hamartomas from gliomas (6–9). In long echo time studies of asymptomatic focal lesions neurofibromatosis type 1, either modest metabolic changes or marked Cho increase with NAA decrease have been found (6,7). Neurofibromatosis type 1–associated hamartomas did not differ significantly from the normal brain, while gliomas had lower NAA/Cr, Cr/Cho, and NAA/Cho ratios (7). In short echo time studies, gliomas differed from benign lesions neurofibromatosis type 1 based on differences in Cho/Cr ratio (8) and on increased Cho and myo-inositol and decreased glutamate/glutamine (9). In none of these reports was histopathologic confirmation for the lesions available. Developmental malformations of the cerebral cortex at MR spectroscopy have shown low levels of NAA and high levels of Cho, with extensive variation in metabolite signal intensities (10). Variably low levels of NAA and high levels of Cho have been found in frontal lobe epileptic foci (11).
Patients with seizures and a focal lesion on MR images often pose a differential diagnostic problem with regard to the nature of their lesion. To our knowledge, there is no study in which MR spectroscopic imaging has been investigated in making a differential diagnosis between gliomas and cortical malformations. Thus, the purpose of our study was to assess proton MR spectroscopic imaging in differentiating low-grade gliomas from focal cortical dysplasias (FCDs) and dysembryoplastic neuroepithelial tumors (DNTs).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patients
Between August 1998 and February 2002, we studied 18 consecutive patients who were evaluated at the Helsinki University Central Hospital for either tumor surgery or epilepsy surgery because of a cortical brain lesion. All 18 patients understood the nature of the study. The mean age of the patients was 30 years ± 13 (SD) (range, 9–50 years). Nine were female (mean age, 30 years ± 12; range, 12–49 years) and nine were male (mean age, 29 years ± 15; range, 9–50 years) patients. All patients first underwent our routine MR imaging, findings of which were evaluated by an experienced neuroradiologist (L.V.) blinded to MR spectroscopic data. The MR spectroscopic data did not influence the subsequent clinical treatment of the patients, as we were gathering data to determine if MR spectroscopy would be at all helpful.
Fourteen of the 18 lesions were verified histologically. Ten patients had low-grade gliomas: three were oligodendrogliomas; three, oligoastrocytomas; three, astrocytomas; and one, a pilocytic astrocytoma. Eight patients had focal cortical developmental malformations (FCDMs), two of which were verified histologically as DNTs and two as FCDs. In the remaining four patients for whom brain tissue was not available for histologic review, the diagnosis was either DNT (n = 1) or FCD (n = 3) based on repeated MR studies (L.V.). These four patients had been clinically followed up (E.G., M.L.G.) for a minimum of 6 years.
Controls
An age- and sex-matched control subject was selected for each patient. The mean age of the 18 controls was 30 years ± 14 (range, 10–56 years). Nine of the controls were female (mean age, 30 years ± 12; range, 12–50 years) and nine were male (mean age, 30 years ± 16; range, 10–56 years). The controls were healthy volunteers recruited from the community who did not have a diagnosis of a central nervous system disease and who had no history of seizures on the basis of the results of the health questionnaire. All control subjects had normal brain MR images.
MR Imaging and Proton MR Spectroscopic Data Collection
MR imaging and MR spectroscopy were performed in separate sessions with a 1.5-T imager (Magnetom Vision; Siemens, Erlangen, Germany) by using a standard circularly polarized head coil. Our routine MR imaging consisted of T2-weighted (3,500/90 [repetition time msec/echo time msec], 24-cm field of view), fluid-attenuated inversion recovery (9,999/105/2,500 [repetition time msec/echo time msec/inversion time msec], 28-cm field of view), T1-weighted (750/14, 23-cm field of view), diffusion-weighted (4,700/118, 23-cm field of view), and gadolinium-enhanced (gadopentetate dimeglumine [0.2 mL/kg], Magnevist; Schering, Berlin, Germany) T1-weighted (750/14, 23-cm field of view) images obtained in three orthogonal planes with a 256 x 256 matrix and a 5-mm section thickness. Proton MR spectroscopy consisted of a double spin-echo sequence (2,600/270) with 16 x 16 phase-encoding steps, two signals acquired, and a 16-cm field of view. Volume preselection was 60/80/100 x 80/100 x 15 mm3, with a 1.5-cm3 nominal voxel volume. The data were collected from one section in the region that was selected (N.L.) on the basis of the abnormalities on MR images. The angulation of the head and the dimensions of the spectroscopic volume of interest were similar in the patient and in the control. When the lesion was close to the scalp, extra care was taken in the positioning of the volume of interest to minimize fat contamination.
Data Analysis
The lesion was evident on the metabolite signal intensity maps and as abnormal spectra and comprised one to six voxels. Depending on the size of the lesion, one or two of the most abnormal voxels, with respect to NAA and Cho, were selected to represent the lesion core. A volume of interest consisting of three to five voxels in the contralateral centrum semiovale was selected for normalization. The mean metabolite signal of the lesion core (LC) was then divided by the mean metabolite signal of the contralateral centrum semiovale (CSO) to yield metabolite indices NAALC/NAACSO, ChoLC/ChoCSO, and CrLC/CrCSO. The same anatomic regions (lesion core and contralateral centrum semiovale) were sampled in the patient and in the matched control. The metabolite index of the patient was then divided by the index of the control subject to yield the lesional metabolite change. Metabolite changes were plotted as a function of one another, that is, NAA change versus Cho change and Cho change versus Cr change. We also calculated the lesional metabolite ratios NAA/Cho, NAA/Cr, and Cho/Cr in the lesion core and in the corresponding volume of interest in the controls. The spectroscopic data analysis was performed by two of us (K.V. and N.L.). An experienced neuropathologist (A.P.) estimated the cell proliferative activity with immunohistochemistry by using the monoclonal antibody MIB-1 for the Ki-67 antigen (12).
Statistical Analysis
The statistical analysis was performed by one author (K.V.). Linear discriminant analysis (KyPlot; Kyence, Tokyo, Japan) was used to differentiate low-grade gliomas from FCDMs and astrocytomas from oligodendrogliomas and oligoastrocytomas. Metabolite ratios (NAA/Cho, NAA/Cr, and Cho/Cr), cell densities, and cell proliferation rates (the MIB-1 indices) between groups were compared by using the Student t test.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
.5, data not shown). There were no differences in cellularity between astrocytomas and oligodendrogliomas and oligoastrocytomas (3,900 cells/mm2 ± 1,700 vs 4,600 cells/mm2 ± 1,200, P > .5) or in MIB-1 index (9.8% ± 4.6 vs 6.7% ± 4.4, P > .5, Fig 3). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Measurement of metabolite ratios or use of linear discriminant analysis of the spectral profiles have served well in the characterization of lesions that arise from and contain nonneural elements such as gliomas, metastases, and meningiomas (16). Our goal was to examine lesions that differ little from the normal brain and are even more challenging to diagnose at MR imaging. To estimate neuronal dysfunction or loss, membrane turnover, and tissue energetics, we measured changes in the individual compounds NAA, Cho, and Cr instead of merely using their ratios, and we took into account as many of the caveats of the method as was possible without having access to absolute quantification. To achieve this, we determined each metabolite by normalizing the lesional metabolite signal to that of the contralateral centrum semiovale and used the same brain sites in the controls. Contralateral centrum semiovale is homogenous both magnetically and in tissue composition, unlike cortical areas, and is therefore a suitable region for normalization. Further, the present approach corrects for imperfections in the pulse profile (17).
Our study, as any MR spectroscopic study, is limited by the spatial resolution of the system, and, therefore, the measured metabolic changes might be underestimated in very small lesions. Further, the effective voxel volume is approximately 20% larger than the nominal voxel volume, which causes signal contamination from adjacent voxels. We focused on the lesion core and therefore included only one or two lesion voxels in the data analysis. These voxels were likely to best represent the metabolite change in the lesion, with minimal contamination from the normal brain or the sulcal cerebrospinal fluid at the periphery of the lesion. Our study did not include segmentation, which, however, would make the analysis more laborious and demanding.
In general, Cho may be regarded as a marker of membrane phosholipid turnover (18). In neoplasia, a high Cho signal has been connected with rapid proliferation of the malignant cells (anabolism) and possibly with cell membrane breakdown (catabolism) when large numbers of mobile water-soluble MR-detectable Cho membrane precursors or breakdown products are present. Malignant glioma cells infiltrate diffusely into the normal brain tissue, displacing and killing neurons. As a consequence, NAA that is synthesized in neurons will decrease (19,20). Reduction of NAA is considered to reflect neuronal loss or dysfunction; although, oligodendrocytes are also involved in the metabolism of NAA (21).
Our results suggest that the changes in NAA and Cho can help differentiate low-grade gliomas from FCDMs. In both types of lesions, however, the Cho signal was moderately to substantially increased. In gliomas, the high Cho signal was present concomitantly with a low level of NAA. In contrast, in the FCDs and DNTs, a high Cho signal appeared with a nearly normal NAA level, and normal or low Cho signal was present only in association with a much reduced NAA level in the current patient series. Within the subgroup of low-grade gliomas, changes in Cho and Cr appeared to be useful in differentiating low-grade astrocytomas from oligodendrogliomas and oligoastrocytomas, although the small number of patients studied precludes making firm conclusions. In oligodendrogliomas and oligoastrocytomas, both Cho and Cr were elevated, whereas in astrocytomas, Cho was modestly increased and Cr was decreased. Oligodendrogliomas generally have a higher cell density than do astrocytomas (22), although this difference was not significant in our series (Fig 3a). However, a higher cell density might account for the higher Cho of oligodendrogliomas and oligoastrocytomas. Negendank et al (2) found higher Cho levels in four oligodendrogliomas than in 75 astrocytomas by comparing the lesional Cho level with the contralateral Cr level. Similarly in line with the present results (Fig 1b), Urenjak et al (19), using in vitro cultured cells, found that oligodendrocytes have a higher Cho and Cr content than do glial cells derived from astrocytoma. Lactate appeared in all low-grade glioma subtypes and was, therefore, not valuable in the differentiation.
In FCDMs, NAA and Cho changes were on average milder than they were in low-grade gliomas. However, in FCDMs, the variation in both NAA and Cho levels was large, ranging from reduced to elevated Cho levels, and NAA/Cr ratio was on average decreased, although to a lesser degree than in the study of Li et al (10). Highly elevated Cho (41%–155% increase) was seen in FCDs (patients 2, 5, and 6) together with a minimally decreased NAA. In two of these patients (patients 5 and 6), both the cell density and the MIB-1-index were low. Thus, neither the overall cell density nor the proportion of proliferating cells seem to explain the high Cho level. Instead, it may be a result of abundant intrinsic epileptic ictal activity in the region of the FCD (23), which could lead to increased membrane turnover.
More interesting, in preoperative and intraoperative electrocorticography, both patients 5 and 6 showed very frequent ictal discharges. In line with this, electroconvulsive therapy has been shown to cause a transient Cho elevation in patients (24), and tonic-clonic seizures have been shown to increase Cho in a rat brain (25). In the remaining two FCDs (patients 1 and 3), NAA was markedly diminished without a concomitant elevation in Cho, suggesting the presence of little glial proliferation or membrane breakdown. The reason for the low NAA in FCDs remains unclear, but the presence of abnormal neurons may play a role. In DNTs (patients 7 and 8), NAA, Cho, and Cr were all near normal or diminished, but NAA was better preserved than in low-grade gliomas. The nature of these metabolic changes agrees with the benign biologic behavior of DNTs, compared to regular gliomas with aggressive or invasive features that would increase Cho and lower NAA.
Because of the relatively small size of the series, larger prospective studies are needed to confirm the present findings. A long echo time was used to obtain spectra free from signals caused by macromolecules and lipids. As a result, the spectra were heavily T2 weighted, and no signal from myo-inositol and glutamate/glutamine was available. However, the changes in NAA, Cho, and Cr allowed differentiation of FCDMs from low-grade gliomas and provided useful information for differential diagnosis between histologic types of gliomas. Further studies with shorter echo times are warranted to show whether adding information on myo-inositol and glutamate/glutamine will improve the discriminatory power of the present technique.
In conclusion, both NAA/Cho and NAA/Cr metabolite ratios and NAA, Cho, and Cr changes were capable of differentiating low-grade gliomas from DNTs and FCDs. However, the metabolite changes were, in addition, able to differentiate between glioma subtypes and therefore showed to be more strongly coupled to the underlying metabolic, histologic, and biochemical features. The present results underline the importance of the neuronal marker NAA as the most distinctive feature between low-grade gliomas and FCDMs. Cho is the second most important metabolite in this differentiation. However, substantial changes in the Cho signal can be detected both in anabolism and catabolism, which calls for caution in interpreting modest increases in Cho.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, N.L.; study concepts, K.V., A.M.H., N.L.; study design, K.V., L.K., A.M.H., E.G., H.J., A.P., N.L.; literature research, K.V., A.M.H., E.G., A.P., N.L.; clinical studies, L.K., E.G., L.V., M.L.G., G.B., A.P.; data acquisition, K.V., A.M.H., E.G., M.L.G., G.B., A.P., N.L.; data analysis/interpretation, K.V., N.L., A.P., E.G., G.B.; statistical analysis, K.V.; manuscript preparation, K.V., A.M.H., H.J., N.L.; manuscript definition of intellectual content and revision/review, all authors; manuscript editing and final version approval, K.V., N.L.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Hourani, L.J. Brant, T. Rizk, J.D. Weingart, P.B. Barker, and A. Horska Can Proton MR Spectroscopic and Perfusion Imaging Differentiate Between Neoplastic and Nonneoplastic Brain Lesions in Adults? AJNR Am. J. Neuroradiol., February 1, 2008; 29(2): 366 - 372. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Castillo Spectroscopy Evidence of Diffuse Brain Abnormalities in Patients with Epileptogenic Foci AJNR Am. J. Neuroradiol., June 1, 2007; 28(6): 1076 - 1077. [Full Text] [PDF]
|
|
|
|
 |
|
 R. N. Al-Okaili, J. Krejza, J. H. Woo, R. L. Wolf, D. M. O'Rourke, K. D. Judy, H. Poptani, and E. R. Melhem Intraaxial Brain Masses: MR Imaging-based Diagnostic Strategy--Initial Experience Radiology, May 1, 2007; 243(2): 539 - 550. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Stadlbauer, S. Gruber, C. Nimsky, R. Fahlbusch, T. Hammen, R. Buslei, B. Tomandl, E. Moser, and O. Ganslandt Preoperative Grading of Gliomas by Using Metabolite Quantification with High-Spatial-Resolution Proton MR Spectroscopic Imaging Radiology, March 1, 2006; 238(3): 958 - 969. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), oligodendrogliomas and oligoastrocytomas (
), FCDs (
), and DNTs (
). Graphs depict (a) NAA change versus Cho change and (b) Cr change versus Cho change. The dotted lines are the linear discriminant functions (LDF) that differentiate (a) gliomas from FCDMs (LDF = 12.04 - 0.043 · 
Cho + 0.181 · 
