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


     

Published online before print September 13, 2002, 10.1148/radiol.2252011592
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2252011592v1
225/2/556    most recent
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 Majós, C.
Right arrow Articles by Gili, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Majós, C.
Right arrow Articles by Gili, J.
(Radiology 2002;225:556-566.)
© RSNA, 2002

Neuroradiology

Adult Primitive Neuroectodermal Tumor: Proton MR Spectroscopic Findings with Possible Application for Differential Diagnosis1

Carles Majós, MD, Juli Alonso, PhD, Carles Aguilera, MD, Marta Serrallonga, MD, Juan J. Acebes, MD, PhD, Carles Arús, PhD and Jaume Gili, MD, PhD

1 From the Institute de Diagnostic per la Imatge (IDI), Department of Diagnostic Imaging, Hospital Duran i Reynals, CSU de Bellvitge, Autovía de Castelldefels km 2,7, 08907 L’Hospitalet de Llobregat, Barcelona, Spain (C.M., J.A., C. Aguilera, M.S., J.G.); Department of Neurosurgery, Hospital Príncipes de España, CSU de Bellvitge, L’Hospitalet de Llobregat, Barcelona, Spain (J.J.A.); and Department of Biochemistry and Molecular Biology, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Spain (C. Arús). Received October 1, 2001; revision requested December 10; revision received January 28, 2002; accepted March 14. Supported in part by the Generalitat de Catalunya (grants CIRIT XT2000 43 and SGR1999-328), the Interministerial Commission on Science and Technology (CICYT SAF1999-101), and the European Union (IST1999-10310). Address correspondence to C.M. (e-mail: cmajos@csub.scs.es).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
PURPOSE: To assess the utility of proton magnetic resonance (MR) spectroscopy in the clinical categorization of primitive neuroectodermal tumors (PNETs) in adults.

MATERIALS AND METHODS: In vivo proton MR spectroscopy was performed with an echo time of 136 msec in nine adults with PNET, and findings were retrospectively compared with spectroscopic findings of 22 meningiomas, 12 low-grade astrocytomas, eight anaplastic astrocytomas, 23 glioblastomas, and 21 metastases. Nine resonances were semiquantitatively evaluated. Statistical analysis was performed by using Kruskal-Wallis and Mann-Whitney U tests. The Hochberg correction was applied for multiple comparisons. Results were prospectively validated in 24 tumors of the six types included in the study.

RESULTS: The resonances of choice for identifying PNET were alanine (P < .001) and glutamate and glutamine (P = .004), both decreased with respect to meningioma; choline increased with respect to low-grade (P < .001) and anaplastic astrocytoma (P = .055); and lipids at 1.30 ppm decreased and choline and other trimethyl-amine-containing compounds increased with respect to glioblastoma (P < .001 and P = .004, respectively) and metastasis (P < .001 and P = .021, respectively). We developed an algorithm for bilateral differential diagnosis between PNET and other tumor types. The leave-one-out method was used to test the five possible differential situations in the retrospective data set, with the following results: PNET versus meningioma, 31/23/5/3 (number of total/correct/unclassifiable/incorrect procedures); PNET versus low-grade astrocytoma, 21/19/2/0; PNET versus anaplastic astrocytoma, 17/6/9/2; PNET versus glioblastoma, 32/28/2/2; and PNET versus metastasis, 30/27/1/2. In total, 131 consecutive procedures produced 103 (79%) correct classifications and nine (7%) misclassifications. Twenty-five (78%) of 32 possible procedures in the prospective independent test set produced correct classifications and four (13%) produced incorrect classifications.

CONCLUSION: In vivo proton MR spectroscopy provides useful information in clinical differentiation between PNETs and common brain tumors in adults.

© RSNA, 2002

Index terms: Brain neoplasms, diagnosis, 10.36 • Brain neoplasms, MR, 10.121411, 10.121413, 10.12143 • Magnetic resonance (MR) spectroscopy, 10.12145 • Primitive neuroectodermal tumor, 10.3637, 10.364


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Primitive neuroectodermal tumors (PNETs) include a heterogeneous group of tumors thought to originate from primitive or undifferentiated neuroepithelial cells that typically occur in pediatric patients. The prototype of these tumors is cerebellar medulloblastoma, which constitutes 13%–25% of all pediatric brain tumors (1–3). Only 20%–30% of PNETs occur in adults, constituting up to 1% of all brain tumors in this age group (1,2,4,5). The typical cerebellar PNET is a midline tumor of well-defined margins that homogeneously enhances after contrast material administration at both computed tomographic (CT) and magnetic resonance (MR) imaging (1–3,6,7). As opposed to pediatric tumors, medulloblastomas in adults tend to result in a higher frequency of atypical findings (4,8), to the point that some authors have suggested always considering medulloblastoma in the differential diagnosis of a posterior fossa mass in an adult (4). On the other hand, supratentorial PNETs are uncommon highly heterogeneous tumors that can mimic other high-grade tumors on MR images (9). The high frequency of atypical findings and the low incidence of PNET in adults can make PNET diagnosis difficult in some cases. In this context, additional information from new diagnostic techniques such as proton MR spectroscopy would be welcomed to reinforce the differential diagnosis with other tumors.

Proton MR spectroscopy is a noninvasive imaging technique for measuring the biochemical content of living tissue that can now be performed with most 1.5-T clinical MR imaging instruments. This biochemical information may produce additional data about tumor metabolism that may be useful in tumor diagnosis. Several studies (10–28) have evaluated proton MR spectroscopy in the most common brain tumors both in vivo and in vitro, but little interest has been shown in PNET, especially in adults. The aim of the present study was to assess the utility of proton MR spectroscopy as a noninvasive tool for clinical categorization of PNET in adults.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Patients
We retrospectively evaluated nine consecutive proton MR spectroscopic examinations in nine patients (eight men and one woman; 18–67 years of age; mean age, 30.8 years) with intracranial PNET (two supratentorial and seven infratentorial). Tumors were untreated in six patients and recurrent in the other three. All neoplasms were confirmed histopathologically. We compared the results obtained in PNETs with a data set of 86 proton MR spectra of the most frequent brain tumors in 86 adults (45 women and 41 men; 14–81 years of age; mean age, 54.7 years). This comparative data set included meningiomas (n = 22), low-grade astrocytomas (n = 12), anaplastic astrocytomas (n = 8), glioblastomas (n = 23), and metastases (n = 21). Proton MR spectroscopic findings obtained under identical conditions in the white matter of six healthy volunteers were also considered references. Nevertheless, assessment of differences between normal brain parenchyma and tumoral tissue, well established in previous work (10–14), was considered to be out of the scope of this study.

To test the reproducibility of differences found between tumors and their application in diagnosis, 51 consecutive proton MR spectroscopic examinations were prospectively performed in 51 patients with brain lesions that suggested tumor. Cases in which a definitive diagnosis was available on the basis of the same criteria as the retrospective training set and in which the diagnosis was one of the six tumor types included in the study were included in the definitive test set. Twenty-seven patients (no definitive diagnosis, n = 13; diagnosis other than tumor, n = 7; diagnosis of tumor other than one of the six types included in the study, n = 7) did not match these criteria and were excluded from the test set. The definitive test set included 24 proton MR spectra corresponding to PNET (n = 2; one supratentorial and one infratentorial), meningioma (n = 7), low-grade astrocytoma (n = 3), anaplastic astrocytoma (n = 4), glioblastoma (n = 2), and metastasis (n = 6). The study was approved by our institutional review board and informed consent obtained from all patients and volunteers.

MR Imaging
In all patients, MR imaging was performed with a 1.5-T unit (ACS-NT; Philips Medical Systems, Best, the Netherlands) in the three orthogonal planes, including at least T1- (536–541/15) (repetition time [TR] msec/echo time [TE] msec), intermediate- (2,175/20) and T2- (2,175/85) weighted images. T1-weighted images were obtained in at least two planes after intravenous administration of 0.1 mmol per kilogram of body weight of gadolinium-based contrast material (Magnevist; Schering, Berlin, Germany, or Omniscan; Nycomed, Oslo, Norway). Fast fluid-attenuated inversion recovery, or FLAIR (6,706/120/2,000 [TR msec/TE msec/inversion time msec]; turbo factor, 15 msec) images were available for six PNETs.

Proton MR Spectroscopy
In all 125 cases (nine retrospective PNETs, 86 retrospective comparative data set tumors, six healthy volunteers, and 24 prospective test set tumors), proton MR spectroscopy was performed with the same MR unit by using a point-resolved spectroscopic sequence. A volume of interest between 1.5 x 1.5 x 1.5 cm3 (3.4 mL) and 2 x 2 x 2 cm3 (8.0 mL) was placed by one of three authors (C.M., C. Aguilera, or M.S.) following the meeting of consensus criteria. The volume of interest size and position were determined by examining the MR images in all three dimensions (sagittal, coronal, and transverse planes), with the aim of positioning the largest possible voxel within the solid tumoral area, as judged at MR image inspection, with avoidance of areas of cysts or necrosis and with minimum contamination from the surrounding nontumoral tissue. A standard receiver head coil was used in all cases, and spectroscopy was incorporated into the course of a conventional MR imaging examination. Automatic shimming of the linear X, Y, and Z channels was used to optimize field homogeneity. The water resonance was set on resonance, and the water suppression pulse was optimized. Proton MR spectroscopic examinations were performed by using a spin-echo pulse sequence with parameters of 2,000/136. A total of 512 data points were collected over a spectral width of 1,000 Hz. Four dummy scans and 128 or 192 signals acquired, depending on voxel size, were obtained for each spectrum. Spectral analysis was performed off-line with the use of MRUI software (available through the MRUI Project, www.carbon .uab.es/mrui) (29). Time domain data were analyzed with the variable projection method (29) after filtering the residual water signal by using the Henkel-Lanczos singular-value decomposition algorithm.

Assignment of the resonances of interest, including N-acetylaspartate and other N-acetyl–containing compounds (NACCs) at 2.02 ppm, creatine plus phosphocreatine (CR) at 3.03 ppm, choline and other trimethyl-amine–containing compounds (CHO) at 3.20 ppm, lipids (LIPs) at 0.90 ppm (LIP 0.90) and 1.30 ppm (LIP 1.30), glutamate and glutamine (GLX) at 2.35 ppm, and glycine and/or myoinositol (Gly/MI) at 3.55 ppm was based on previous studies of brain tumors (10,11,15–21,30–33) and phantoms (15,30). Two inverted doublets due to phase modulation from J coupling were defined that corresponded to lactate (LACT) centered at 1.35 ppm and to alanine (Ala) at 1.47 ppm. Each of these nine resonances (LIP 0.90, LIP 1.30, LACT, Ala, NACC, GLX, CR, CHO, and Gly/MI) was considered separately for statistical analysis. Resonances in the region of taurine above the noise level were not detectable in most cases and, accordingly, taurine resonance was not quantified. Water (4.75 ppm) and/or CR (3.03 ppm) were chosen as chemical shift reference resonances to correct possible shifting in the frequency domain. To avoid operator bias, resonance peaks were defined in the frequency domain even when there were doubts about their differentiation from noise. Our criterion was that further quantification and analysis result in differences between noise and metabolite signal without operator influence. An area of "0" was assigned only to resonances in which the software program used did not satisfactorily fit a peak in the area of interest.

For comparative purposes, the program-fitted resonance areas for LACT, Ala, NACC, GLX, CR, CHO, and Gly/MI resonances were normalized by dividing each value by the square root of the sum of the squares of the three main spectroscopic resonances (x = 100 x xi/(NACC2 + CR2 + CHO2)1/2, where xi is the original area of the resonance being normalized, in a modification of the method used by Tate et al (16). For normalization of the resonance areas of LIPs, the square value of LIP 0.90 and 1.30 was included in the sum of the denominator {x = 100 x xi/[(NACC2 + CR2 + CHO2 + LIP 0.902 + LIP 1.302)1/2]}. Data obtained were used to compare PNETs with the other tumor types to assess which findings were characteristic of this tumor at in vivo spectroscopy. We should emphasize, as have authors of previous studies (22,31), that we assigned these characteristic resonances not to a particular metabolite but to a resonance in the spectrum, taking into account that a particular resonance origin should not be directly extrapolated to a particular metabolite without further ex vivo and in vitro examination (15,32,34).

Statistics
The goal of the current study was to determine whether there were significant differences between PNETs and each of the other common brain tumors in adults (meningiomas, low-grade astrocytomas, anaplastic astrocytomas, glioblastomas, and metastases) and to design a procedure to distinguish PNETs from the rest of the tumors. To this end, we considered the nine metabolite resonances and determined which tumor type produced significant differences with respect to PNETs.

We performed Kruskal-Wallis nonparametric analysis of variance for each metabolite resonance to test for significant differences among the six tumor types. Then, to compare the group of interest (PNETs) with the other tumor types, Mann-Whitney U tests were performed. Because we considered five statistical tests for every metabolite resonance, we corrected the obtained P values by using the Hochberg method (35). Then the significant differences were defined by using the corrected P values (P*) instead of the original P value. Differences of P* less than .05 were considered statistically significant. Statistics were computed with SPSS software (SPSS, Chicago, Ill).

Empirical Method for Bilateral Differential Diagnosis of PNET
With the aim of testing the usefulness of differences between PNETs and the rest of the tumors in PNET identification, we elaborated an empiric path for discrimination between PNETs and the rest of the tumors in pairwise comparisons. This path used the 90th percentiles of the resonances that showed the strongest discriminative performance (one or two for every bilateral comparison) in the tumors under consideration. A scheme of this path is shown in Figure 1. In the first step, the case was included in one of the two tumor groups being compared if the value for the first discriminative resonance was in the 90th percentile of one of the tumor types. Values in the overlapping area of the 90th percentile or between the boundaries of these percentiles were considered transitorily unclassifiable. A second step was performed, when possible, in the transitorily unclassifiable group by using a second discriminative resonance. This path was developed by using the retrospective data set. Its performance was tested in the retrospective training set by using the leave-one-out method and in a prospective independent test set.



View larger version (35K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Flow chart shows the empiric path used for bilateral differential diagnosis of PNET. DR = discriminative resonance, 1H MRS = proton MR spectroscopy, 90% P = 90th percentile. * = First and second discriminative resonances may vary depending on the tumor type to be differentiated from PNET. ** = Non-PNET tumors were meningioma, low-grade astrocytoma, anaplastic astrocytoma, glioblastoma, and metastasis. TE = echo time.

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
MR Imaging
The imaging findings in the nine PNETs are summarized in Table 1. Only one (14%) of the seven posterior fossa PNETs showed the typical aspect of medulloblastomas reported in childhood as a homogeneous lesion of the inferior vermis with well-defined margins and homogeneous enhancement with contrast material administration (3,6–8,36). The rest were heterogeneous tumors. Five (71%) of seven infratentorial PNETs were located in the cerebellar hemisphere, while only two (29%) were in their typical location in the vermis. The signal intensity of the tumors was diverse with the sequences performed. In the current study, PNETs were most commonly found to be heterogeneous lesions with poorly defined margins located in the cerebellar hemisphere (four [44%] of nine tumors). Three (43%) of seven posterior fossa PNETs were heterogeneous poorly defined tumors located in the cerebellar hemispheres with areas of cysts and/or necrosis.


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

  		TABLE 1. MR Imaging Findings in PNETs in Nine Adults

 
Proton MR Spectroscopy
The means and SDs of resonances in the nine PNETs, the 86 other tumors in the retrospective data set, and the six healthy volunteers are shown in Table 2. The metabolite resonances found to differ significantly between PNETs and other tumor groups are indicated, and relevant P* values are shown.


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

  		TABLE 2. Normalized Area Values of Nine Metabolite Resonances from Database of 95 Brain Tumors Analyzed with in Vivo Proton MR Spectroscopy

 
PNETs were characterized by high CHO resonances and low Ala, NACC, CR, and LIP resonances (Table 2). Scatter diagrams of the resonances that showed greater differences between tumors are shown in Figure 2. CHO and NACC were the strongest discriminative resonances in the pairwise comparison between PNETs and the rest of the tumors and showed significant differences at comparison with low-grade astrocytoma (P* < .001 and P* = .013 for CHO and NACC, respectively), glioblastoma (P* = .004 and P* = .013), and metastasis (P* = .021 and P* = .010) (Fig 2, Table 2). CR also showed significant differences at comparison with low-grade astrocytoma (P* < .001) and glioblastoma (P* = .012). Ala and GLX were significantly decreased with respect to meningiomas (P* < .001 and P* = .004, respectively) (Table 2). Ala resonance was detectable in four PNETs in amounts lower than those in meningiomas. Broad resonances at 1.30 ppm that were attributable to LIPs were found in only one PNET, with a lower signal than those in glioblastomas and metastases (Fig 2c). In the rest of the cases, no LIP signals were detectable (P* < .001 for LIP 1.30 between PNET and glioblastoma; P* < .001 for LIP 1.30 between PNET and metastasis). In this context, CHO, NACC, and CR were the resonances of choice for discriminating PNET from low-grade astrocytomas; LIP 1.30, for distinguishing PNET from glioblastoma and metastasis; and Ala, for discriminating between PNET and meningioma. Anaplastic astrocytomas showed the greatest similarity to PNET, with overlapping of most resonance values (Fig 2; Table 2) and a lack of significant differences. In this context, the resonances that showed the greatest tendency to differ between this tumor group and PNETs were CHO (P* = .055) and CR (P* = .062).



View larger version (17K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a. Scatterplots show the distribution of normalized area values (ordinates, arbitrary units) of the resonances that allowed better differentiation of PNET (*) and other tumor types. The mean value for every tumor group is also labeled (-) and its numeric value given. (a) Scatterplot shows the distribution of the resonance that corresponds to CHO in PNET, low-grade astrocytoma (LGA) ({square}), anaplastic astrocytoma (AA) ({triangleup}), and glioblastoma (GBM) ({diamond}). Note the higher values for PNET (*). The tumor group showing the most overlap with PNET is anaplastic astrocytoma. Glioblastoma also shows overlap that is of low diagnostic effect, as the main differences between PNET and glioblastoma are found in LIP 1.30 (Figs 2c and 7). (b) Scatterplot shows the distribution of Ala in meningioma (MEN) ({circ}) and PNET. The mean value is significantly higher in meningioma. (c) Scatterplot shows the distribution of LIP 1.30 in glioblastoma (GBM) ({diamond}), metastasis (MET) (x), and PNET. The mean value is significantly higher in glioblastoma and metastasis than in PNET.

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b. Scatterplots show the distribution of normalized area values (ordinates, arbitrary units) of the resonances that allowed better differentiation of PNET (*) and other tumor types. The mean value for every tumor group is also labeled (-) and its numeric value given. (a) Scatterplot shows the distribution of the resonance that corresponds to CHO in PNET, low-grade astrocytoma (LGA) ({square}), anaplastic astrocytoma (AA) ({triangleup}), and glioblastoma (GBM) ({diamond}). Note the higher values for PNET (*). The tumor group showing the most overlap with PNET is anaplastic astrocytoma. Glioblastoma also shows overlap that is of low diagnostic effect, as the main differences between PNET and glioblastoma are found in LIP 1.30 (Figs 2c and 7). (b) Scatterplot shows the distribution of Ala in meningioma (MEN) ({circ}) and PNET. The mean value is significantly higher in meningioma. (c) Scatterplot shows the distribution of LIP 1.30 in glioblastoma (GBM) ({diamond}), metastasis (MET) (x), and PNET. The mean value is significantly higher in glioblastoma and metastasis than in PNET.

 


View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2c. Scatterplots show the distribution of normalized area values (ordinates, arbitrary units) of the resonances that allowed better differentiation of PNET (*) and other tumor types. The mean value for every tumor group is also labeled (-) and its numeric value given. (a) Scatterplot shows the distribution of the resonance that corresponds to CHO in PNET, low-grade astrocytoma (LGA) ({square}), anaplastic astrocytoma (AA) ({triangleup}), and glioblastoma (GBM) ({diamond}). Note the higher values for PNET (*). The tumor group showing the most overlap with PNET is anaplastic astrocytoma. Glioblastoma also shows overlap that is of low diagnostic effect, as the main differences between PNET and glioblastoma are found in LIP 1.30 (Figs 2c and 7). (b) Scatterplot shows the distribution of Ala in meningioma (MEN) ({circ}) and PNET. The mean value is significantly higher in meningioma. (c) Scatterplot shows the distribution of LIP 1.30 in glioblastoma (GBM) ({diamond}), metastasis (MET) (x), and PNET. The mean value is significantly higher in glioblastoma and metastasis than in PNET.

 
Comparison of supra- and infratentorial PNETs, as well as of treated and untreated tumors, showed no major differences at visual evaluation and after quantification. Statistical analysis was still considered inappropriate because of the small number of cases in some uncommon subgroups (ie, supratentorial PNET).

Empirical Method for Bilateral Differential Diagnosis of PNET
The resonances that showed the strongest discriminative performance and thus were considered "first discriminative resonances" with PNET were Ala for meningioma, CHO for low-grade and anaplastic astrocytoma, and LIP 1.30 for glioblastoma and metastasis. Additional discriminative utility was found in GLX for meningioma and in CHO for metastasis and glioblastoma; accordingly, these resonances were chosen as second discriminative resonances for these tumors. This second step was considered inappropriate for low-grade and anaplastic astrocytomas, since the values of the alternative resonances (NACC and CR) had a possible correlation with the elective one (CHO) because of the normalization method used. Table 3 shows the resonances of choice for pairwise comparison in this study, the boundaries used for application of the path shown in Figure 1, and the results obtained at categorization of the retrospective data set, as assessed by using the leave-one-out method (16,22). A total of 131 differential procedures were used to test the system in the retrospective training set by using the leave-one-out method: (a) five testing procedures were performed with every PNET (n = 9) to test the algorithm versus every one of the other five tumor types (total, 45 procedures); (b) only one testing procedure was performed in every non–PNET tumor (n = 86) to test the procedure that differentiated this tumor type from PNET (total, 86 procedures). One hundred three (79%) of 131 total bilateral procedures produced a correct classification, while 19 (14%) were unclassifiable, and nine (7%) were incorrectly classified. We point out that these global rates are not a direct absolute estimate of the global probability of correct classification but an orientational evaluation of the performance of the method. Table 3 shows the range of results in the five bilateral comparison algorithms considered.


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

  		TABLE 3. Criteria for Bilateral Tumor Discrimination of PNET and Assessment of Results in 95 Retrospective Cases

 
Examples of the potential use of proton MR spectroscopy in the differential diagnosis of PNET are shown in Figures 3–7. Figures 3–5 include examples of midline posterior fossa tumors with similar homogeneous MR imaging signal patterns, in which a definitive diagnosis could not be established with MR imaging alone. These figures include PNET (Fig 3), metastasis (Fig 4), and meningioma (Fig 5). Prominent resonances from LIP at 1.30 ppm in Figure 4 and Ala at 1.47 ppm in Figure 5 give additional support to a presumptive diagnosis of metastasis (or glioblastoma) and meningioma. Figures 6 and 7 illustrate the differential diagnosis between PNET and glioblastoma in heterogeneous hemispheric tumors. LIP resonance in Figure 7 oriented the case as glioblastoma (or metastasis). All diagnoses were histopathologically proved.



View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a. PNET. (a) Transverse T2-weighted MR image (2,175/85) shows a relatively homogeneous tumor (arrows) in the posterior fossa, adjacent to the posterior wall of the fourth ventricle. The voxel position for proton MR spectroscopy (box) is also shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 acquisitions) of the tumor shows prominent resonances from CHO and low CR and NACC resonance. There is a small amount of LACT (Lact) and a resonance at 3.55 ppm that is attributable to Gly/MI. Some peaks (*) around 3.30 and 3.40 ppm suggest the presence of taurine in this particular case. The histologic diagnosis was PNET. The empiric algorithm satisfactorily classified this tumor as PNET at bilateral comparisons with the main tumors included in the differential diagnosis (metastasis and meningioma).

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b. PNET. (a) Transverse T2-weighted MR image (2,175/85) shows a relatively homogeneous tumor (arrows) in the posterior fossa, adjacent to the posterior wall of the fourth ventricle. The voxel position for proton MR spectroscopy (box) is also shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 acquisitions) of the tumor shows prominent resonances from CHO and low CR and NACC resonance. There is a small amount of LACT (Lact) and a resonance at 3.55 ppm that is attributable to Gly/MI. Some peaks (*) around 3.30 and 3.40 ppm suggest the presence of taurine in this particular case. The histologic diagnosis was PNET. The empiric algorithm satisfactorily classified this tumor as PNET at bilateral comparisons with the main tumors included in the differential diagnosis (metastasis and meningioma).

 


View larger version (145K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. Metastasis. (a) Transverse T2-weighted MR image (2,175/85) shows a well-defined midline tumor (arrows) adjacent to the aqueduct and fourth ventricle. There is some heterogeneity with central low signal intensity and peritumoral edema. The voxel position for proton MR spectroscopy (box) is shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 signals acquired) of the tumor shows LIP resonance at 1.30 ppm and a small amount at 0.90 ppm. This finding is highly suggestive of metastasis or glioblastoma. The diagnosis after tumor removal was metastasis. Bilateral comparison with PNET satisfactorily suggested metastasis.

 


View larger version (18K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. Metastasis. (a) Transverse T2-weighted MR image (2,175/85) shows a well-defined midline tumor (arrows) adjacent to the aqueduct and fourth ventricle. There is some heterogeneity with central low signal intensity and peritumoral edema. The voxel position for proton MR spectroscopy (box) is shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 signals acquired) of the tumor shows LIP resonance at 1.30 ppm and a small amount at 0.90 ppm. This finding is highly suggestive of metastasis or glioblastoma. The diagnosis after tumor removal was metastasis. Bilateral comparison with PNET satisfactorily suggested metastasis.

 


View larger version (128K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5a. Meningioma. (a) Transverse T2-weighted MR image (2,175/85) shows a well-defined midline tumor (arrows). The voxel position for proton MR spectroscopy (box) is depicted. "A" marks correspond to unremovable marks made by the spectroscopic software package used. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 signals acquired) of the tumor shows an inverted Ala doublet centered at 1.47 ppm that is highly suggestive of meningioma. Note also a clear resonance of GLX. The histologic diagnosis was meningioma. Bilateral comparison satisfactorily resulted in differentiation of this tumor from PNET.

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5b. Meningioma. (a) Transverse T2-weighted MR image (2,175/85) shows a well-defined midline tumor (arrows). The voxel position for proton MR spectroscopy (box) is depicted. "A" marks correspond to unremovable marks made by the spectroscopic software package used. (b) Localized spin-echo proton MR spectrum (2,000/136, 128 signals acquired) of the tumor shows an inverted Ala doublet centered at 1.47 ppm that is highly suggestive of meningioma. Note also a clear resonance of GLX. The histologic diagnosis was meningioma. Bilateral comparison satisfactorily resulted in differentiation of this tumor from PNET.

 


View larger version (136K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6a. PNET. (a) Transverse fast FLAIR MR image (6,706/120/2,000; turbo factor, 15) shows a heterogeneous tumor (arrows) with a cystic and/or necrotic area. The voxel position for proton MR spectroscopy (box) is shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 192 signals acquired) of the tumor shows prominent resonance from CHO and low CR and NACC resonances. A certain amount of LACT is probably overlapping with Ala. Note also a small peak (*) at 3.4 ppm that is difficult to differentiate from noise and could suggest resonance from taurine. The definitive diagnosis after partial tumor removal was PNET. Bilateral differentiation from glioblastoma satisfactorily suggested PNET.

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6b. PNET. (a) Transverse fast FLAIR MR image (6,706/120/2,000; turbo factor, 15) shows a heterogeneous tumor (arrows) with a cystic and/or necrotic area. The voxel position for proton MR spectroscopy (box) is shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 192 signals acquired) of the tumor shows prominent resonance from CHO and low CR and NACC resonances. A certain amount of LACT is probably overlapping with Ala. Note also a small peak (*) at 3.4 ppm that is difficult to differentiate from noise and could suggest resonance from taurine. The definitive diagnosis after partial tumor removal was PNET. Bilateral differentiation from glioblastoma satisfactorily suggested PNET.

 


View larger version (133K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7a. Glioblastoma. (a) Transverse intermediate-weighted MR image (2,175/20) shows a heterogeneous tumor (arrows) with a necrotic cystlike area in the left frontal lobe. The voxel position for proton MR spectroscopy (box) is also shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 192 signals acquired) of the tumor shows LIP resonance at 1.30 ppm that is highly suggestive of metastasis or glioblastoma. The histologic diagnosis after partial tumor removal was glioblastoma. Bilateral discrimination with PNET satisfactorily suggested glioblastoma.

 


View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7b. Glioblastoma. (a) Transverse intermediate-weighted MR image (2,175/20) shows a heterogeneous tumor (arrows) with a necrotic cystlike area in the left frontal lobe. The voxel position for proton MR spectroscopy (box) is also shown. (b) Localized spin-echo proton MR spectrum (2,000/136, 192 signals acquired) of the tumor shows LIP resonance at 1.30 ppm that is highly suggestive of metastasis or glioblastoma. The histologic diagnosis after partial tumor removal was glioblastoma. Bilateral discrimination with PNET satisfactorily suggested glioblastoma.

 
This empiric classification approach was tested in a prospective test set of 24 tumors, two of which were PNETs. Because both PNETs were used to test the algorithms for differentiation from all five of the other groups (10 testing procedures for two PNETs), a total of 32 procedures were performed to prospectively test the algorithms for bilateral differential diagnosis between PNETs and other tumors, as described previously. Of the 32 total procedures, 25 (78%) yielded correct classifications, three (9%) yielded unclassifiable tumors, and four (13%) yielded misclassifications (Table 4).


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

  		TABLE 4. Results of 32 Differential Diagnostic Procedures in a Prospective Test Set of 24 Tumors to Test the Empirical Method for Bilateral Differential Diagnosis of PNET

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
 
PNETs are undifferentiated round cell tumors that typically occur in childhood. Only 20%–30% of these tumors occur in adulthood, at which time, as opposed to the pediatric tumors, they tend to show a higher frequency of atypical features (4,8). This is the reason that some authors have suggested that medulloblastoma always be considered in the differential diagnosis of a mass in the posterior fossa of an adult (4). Our results agree with this atypical aspect of medulloblastoma in adulthood. In the current study participants, the most common pattern for posterior fossa PNET in 43% of medulloblastomas was a heterogeneous poorly defined tumor located in the cerebellar hemisphere with areas of cystic or necrotic degeneration into the tumor. On the other hand, supratentorial PNETs, similar to other supratentorial tumors, tend to show a highly heterogeneous pattern with areas of necrosis (9). Thus, additional information for diagnosis of these tumors would be of interest, especially when considering that diagnosis of PNET prior to surgery has important repercussions in therapeutic decisions and prognosis (ie, suspicion of PNET could indicate spinal staging MR imaging to discard drop metastases throughout the cerebrospinal fluid before surgery [1,2]; insertion of a ventriculoperitoneal shunt should be considered only if strictly necessary because it may be a channel for neoplastic spread to extraneural sites [37], and the high sensitivity of these tumors to chemotherapy and radiation therapy could have a major effect on patient care [37]).

The biochemical information sampled with proton MR spectroscopy brings additional data about the metabolism and pathophysiologic status of brain tumors, provides diagnostic orientation on the basis of the use of molecular or metabolic profiles, and complements the information gathered at anatomic imaging. Nevertheless, before applying proton MR spectroscopy for tumor classification in daily clinical practice, it is necessary to know the characteristics of every tumor group and the most useful findings in differentiating between tumor types. Authors of many studies (10–20,22–25) have evaluated proton MR spectroscopy in the most common brain tumors both in vivo and in vitro, and their spectral characteristics are being progressively established, with promising results in differentiating between them. Evaluation of less common tumors constitutes an additional step in the introduction of proton MR spectroscopy in clinical practice, which is essential for completing the study of the clinical application of proton MR spectroscopy to diagnose brain tumors. In this respect, little interest has been shown in proton MR spectroscopy of PNETs, especially in adults. Most PNETs described in the spectroscopic literature correspond to a few cases included in large series of brain tumors.

In vivo proton MR spectroscopy in children has focused on differentiation between the three most common posterior fossa tumors in this age group: PNET, ependymoma, and astrocytoma, with good results in most series (21,26). Authors of previous studies (21,26) found low ratios of NACC/CHO, CR/CHO, and LACT/CHO to be characteristic of medulloblastoma in children and resulted in improvement in discriminating among PNETs, low-grade astrocytomas, and ependymomas in children when proton MR spectroscopic findings were considered (26). Furthermore, most in vitro studies (23,24,27,28) have reported an increase in total CHO, taurine, Gly, and inositol levels in PNET. The differential diagnosis in adults is significantly different from that in children and should include the most common tumors in adults: meningioma, metastasis, and astrocytic tumor (low-grade astrocytoma, anaplastic astrocytoma, and glioblastoma). The first two may simulate medulloblastoma in homogeneous posterior fossa midline tumors or in tumors located peripherally in the hemispheres. Metastasis and astrocytic tumors can be confused with PNETs in cases of heterogeneous tumors located outside the vermis or in the cerebral hemispheres (Figs 6, 7). On the other hand, an identical spectroscopic pattern of medulloblastoma in children and adults cannot be taken for granted. PNET has shown different radiologic behavior in adults (4,8); for example, a desmoplastic histologic variant has been described to be more frequent in this age group (4,5,8), and some authors have suggested a different origin of PNETs in adults with respect to children (38). Accordingly, our study focuses on the characteristics of PNET in adults and its differential diagnosis in this age group.

Our results were similar to those previously reported in pediatric PNETs. As in previous work in children (27,28), we found high relative CHO values in PNET with respect to the rest of the tumors, showing significant differences with all tumors except meningioma and anaplastic astrocytoma. These high CHO values would correlate with the high cellularity of PNET observed at histologic examination, with densely packed cells and scant cytoplasm (36). Another characteristic of PNET found in the current study was the absence of or low amounts of LIP signal, in contrast with that in other high-grade tumors (glioblastoma and metastases). LIP signals have been correlated with necrosis in tumors (39–42). The low intensity of LIPs in PNET, considered grade 4 malignancy by the World Health Organization (43), may be a result of their high cellularity and a low amount of necrosis. There were also differences in Ala and GLX between PNET and meningioma. Ala was significantly increased in meningiomas, a well-known finding with respect to other tumors (10,13,17,23,24) that was also found with respect to PNET in the current study. We also evaluated GLX resonance, although it should be taken into account that it could have contributions from LIP resonances when prominent LIP signals are present at 0.90 and 1.30 ppm (39,40,44). Previous phantom studies (30,33) have shown that GLX is detectable with a echo time of 136 msec. We found this resonance to be increased in meningioma at comparison with that in PNET.

The ideal echo time for tumor classification with proton MR spectroscopy is under discussion. In the current study, we preferred to use a relatively long echo time (136 msec) sequence, since it allows good differentiation between LACT and Ala and LIPs because of J modulation, produces less baseline distortion, and is easier to quantify; we thus believed that it would give more reproducible results. Another point under discussion is the choice between single- or multivoxel techniques (45). Multivoxel techniques more completely represent the tumor heterogeneity. Nevertheless, single-voxel techniques may have major advantages for brain tumor discrimination in clinical practice: It takes less time to perform the examination, it is easier and quicker to process data to obtain quantitative results, and it is possible to obtain better magnetic field homogeneity in the volume of interest. Nevertheless, although we used a single-voxel technique, we hope that the spectroscopic pattern knowledge gained will be of interest in interpreting multivoxel data in the future.

The most accurate method of clinical proton MR spectroscopic interpretation also remains an open question. While a robust classification method is unanimously accepted, some preliminary methods could be useful to exploit the quantitative potential of proton MR spectroscopy. An additional use for such preliminary methods would be prospective testing of the usefulness and reproducibility of spectroscopic findings in differentiating between tumors. In this respect, we obtained satisfactory results in pairwise differentiation between PNET and the five more common brain tumors in adults in a retrospective training set, with correct classification of 79% (103 of 131) of cases, in the range seen in previous research on brain tumors (12,16,21,26). Of high interest for clinical application is that in 7% (nine of 131) of cases, an incorrect diagnosis was suggested as a result of our method. Differentiation of PNET in our study was carried out by using the 90th percentiles of some specific discriminative resonances that included Ala and GLX for meningioma, CHO for low-grade and anaplastic astrocytomas, and LIP 1.30 and CHO for glioblastoma and metastasis. We confirmed the consistency and reproducibility of our findings by means of a prospective test set of 24 tumors (correctly classified, 78% [25 of 32 procedures]; unclassifiable, 9% [three of 32 procedures]; and misclassified, 13% [four of 32 procedures]). Only a few studies of brain tumor classification with proton MR spectroscopy (16,46) have included prospective test sets to validate the performance produced by their discrimination strategies. In the present study, we preferred to use such an independent test set because it avoids the possible bias of using the same cases for training and testing (16,47). In this way, we expected to gain better insight into the utility of our results for bilateral tumor discrimination in real new cases.

Our aim in elaborating a bilateral discriminative path was not to definitively establish a method for tumor classification but to test the findings reported in the study, its reproducibility, and its usefulness when applied in tumor discrimination. This method could be used as a guide for applying proton MR spectroscopic findings in tumor categorization. Nevertheless, boundaries (the 90th percentile in our case) should be recalculated and tested for every clinical setting because of differences in the acquisition protocols and MR imaging systems used. Nonetheless, our findings confirm that in vivo proton MR spectroscopy provides additional information for identifying PNETs in adults on the basis of the tumors’ biochemical characteristics, which are reflected in their spectral pattern.


   ACKNOWLEDGMENTS
 
We thank the Statistic Support Facility of Universitat Autònoma de Barcelona for technical advice on statistical data processing.


   FOOTNOTES
 
Abbreviations: Ala = alanine, CHO = choline and other trimethyl-amine–containing compounds, CR = creatine plus phosphocreatine, GLX = glutamate and glutamine, Gly/MI = glycine and/or myoinositol, LACT = lactate, LIP = lipid, LIP 0.90 = LIP at 0.90 ppm, LIP 1.30 = LIP at 1.30 ppm, NACC = N-acetyl–containing compounds, P* = corrected P value, PNET = primitive neuroectodermal tumor

Author contributions: Guarantors of integrity of entire study, C.M., J.A.; study concepts, C.M., J.A., C. Arús, J.G.; study design, C.M., C. Aguilera, M.S., J.J.A., C. Arús; literature research, C.M., J.A., M.S., C. Arús; clinical studies, J.J.A.; data acquisition, C.M., M.S., C. Aguilera; data analysis/interpretation, C.M., J.A., C. Arús, J.G.; statistical analysis, C.M., C. Aguilera; manuscript preparation, C.M., J.A., C. Arús; manuscript definition of intellectual content, C.M., J.A., C. Arús, J.G.; manuscript editing and revision/review, C.M., C. Aguilera; manuscript final version approval, all authors.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 

  1. Zimmerman RA, Bilaniuk LT. Brain tumors. In: Zimmerman RA, Gibby WA, Carmody RF, eds. Neuroimaging: clinical and physical principles. New York, NY: Springer-Verlag, 1997; 1004-1009.
  2. Osborn AG. Meningiomas and other nonglial neoplasms. In: Osborn AG, eds. Diagnostic neuroradiology. St Louis, Mo: Mosby, 1994; 613-620.
  3. Meyers SP, Kemp SS, Tarr RW. MR imaging features of medulloblastomas. AJR Am J Roentgenol 1992; 158:859-865.[Abstract/Free Full Text]
  4. Bourgouin PM, Tampieri D, Grahovac SZ, Léger C, Del Carpio R, Melançon D. CT and MR imaging findings in adults with cerebellar medulloblastoma: comparison with findings in children. AJR Am J Roentgenol 1992; 159:609-612.[Abstract/Free Full Text]
  5. Koci TM, Chiang F, Mehringer CM, et al. Adult cerebellar medulloblastoma: imaging features with emphasis on MR findings. AJNR Am J Neuroradiol 1993; 14:929-939.[Abstract]
  6. Nelson M, Diebler C, Forbes WS. Paediatric medulloblastoma: atypical CT features at presentation in the SIOP II trial. Neuroradiology 1991; 33:140-142.[CrossRef][Medline]
  7. Tortori-Donati P, Fondelli MP, Rossi A, et al. Medulloblastoma in children: CT and MRI findings. Neuroradiology 1996; 38:352-359.[Medline]
  8. Maleci A, Cervoni L, Delfini R. Medulloblastoma in children and in adults: a comparative study. Acta Neurochir (Wien) 1992; 119:62-67.
  9. Davis PC, Wichman RD, Takei Y, Hoffman JC, Jr. Primary cerebral neuroblastoma: CT and MR findings in 12 cases. AJNR Am J Neuroradiol 1990; 11:115-120.[Abstract]
  10. Kugel H, Heindel W, Ernestus RI, Bunke J, du Mesnil R, Friedmann G. Human brain tumors: spectral patterns detected with localized H-1 MR spectroscopy. Radiology 1992; 183:701-709.[Abstract/Free Full Text]
  11. Poptani H, Gupta RK, Roy R, Pandey R, Jain VK, Chhabra DK. Characterization of intracranial mass lesions with in vivo proton MR spectroscopy. AJNR Am J Neuroradiol 1995; 16:1593-1603.[Abstract]
  12. Usenius JP, Tuohimetsa S, Vainio P, Ala-Korpela M, Hiltunen Y, Kauppinen RA. Automated classification of human brain tumours by neural network analysis using in vivo 1H magnetic resonance spectroscopic metabolite phenotypes. Neuroreport 1996; 7:1597-1600.[Medline]
  13. Shimizu H, Kumabe T, Tominaga T, et al. Noninvasive evaluation of malignancy of brain tumors with proton MR spectroscopy. AJNR Am J Neuroradiol 1996; 17:737-747.[Abstract]
  14. Adamson AJ, Rand SD, Prost RW, Kim TA, Schultz C, Haughton VM. Focal brain lesions: effect of single-voxel proton MR spectroscopic findings on treatment decisions. Radiology 1998; 209:73-78.[Abstract/Free Full Text]
  15. Barba I, Moreno A, Martínez-Pérez I, et al. Magnetic resonance spectroscopy of brain hemangiopericytomas: high myoinositol concentrations and discrimination from meningiomas. J Neurosurg 2001; 94:55-60.[Medline]
  16. Tate AR, Griffiths JR, Martínez-Pérez I, et al. Towards a method for automated classification of 1H MRS spectra from brain tumours. NMR Biomed 1998; 11:177-191.[CrossRef][Medline]
  17. Majós C, Cucurella G, Aguilera C, Coll S, Pons LC. Intraventricular meningiomas: MR imaging and MR spectroscopic findings in two cases. AJNR Am J Neuroradiol 1999; 20:882-885.[Abstract/Free Full Text]
  18. Meyerand ME, Pipas JM, Mamourian A, Tosteson TD, Dunn JF. Classification of biopsy-confirmed brain tumors using single-voxel MR spectroscopy. AJNR Am J Neuroradiol 1999; 20:117-123.[Abstract/Free Full Text]
  19. Negendank WG, Sauter R, Brown TR, et al. Proton magnetic resonance spectroscopy in patients with glial tumors: a multicenter study. J Neurosurg 1996; 84:449-458.[Medline]
  20. Burtscher IM, Skagerberg G, Geijer B, Englund E, Stahlberg F, Holtas S. Proton MR spectroscopy and preoperative diagnostic accuracy: an evaluation of intracranial mass lesions characterized by stereotactic biopsy findings. AJNR Am J Neuroradiol 2000; 21:84-93.[Abstract/Free Full Text]
  21. Wang Z, Sutton LN, Cnaan A, et al. Proton MR spectroscopy of pediatric cerebellar tumors. AJNR Am J Neuroradiol 1995; 16:1821-1833.[Abstract]
  22. Preul MC, Caramanos Z, Collins DL, et al. Accurate, noninvasive diagnosis of human brain tumors by using proton magnetic resonance spectroscopy. Nat Med 1996; 2:323-325.[CrossRef][Medline]
  23. Kinoshita Y, Kajiwara H, Yokota A, Koga Y. Proton magnetic resonance spectroscopy of brain tumors: an in vitro study. Neurosurgery 1994; 35:606-614.[Medline]
  24. Kinoshita Y, Yokota A. Absolute concentrations of metabolites in human brain tumors using in vitro proton magnetic resonance spectroscopy. NMR Biomed 1997; 10:2-12.[CrossRef][Medline]
  25. Tien RD, Lai PH, Smith JS, Lazeyras F. Single-voxel proton brain spectroscopy exam (PROBE/SV) in patients with primary brain tumors. AJR Am J Roentgenol 1996; 167:201-209.[Abstract/Free Full Text]
  26. Arle JE, Morriss C, Wang ZJ, Zimmerman RA, Phillips PG, Sutton LN. Prediction of posterior fossa tumor type in children by means of magnetic resonance image properties, spectroscopy, and neural networks. J Neurosurg 1997; 86:755-761.[Medline]
  27. Carapella CM, Carpinelli G, Knijn A, Raus L, Caroli F, Podo F. Potential role of in vitro 1H magnetic resonance spectroscopy in the definition of malignancy grading of human neuroepithelial brain tumours. Acta Neurochirur Suppl 1997; 68:127-132.
  28. Sutton LN, Wehrli SL, Gennarelli L, et al. High-resolution 1H-magnetic resonance spectroscopy of pediatric posterior fossa tumors in vitro. J Neurosurg 1994; 81:443-448.[Medline]
  29. van den Boogaart A. Quantitative data analysis of in vivo MRS data sets. Magn Reson Chem 1997; 35:S146-S152.[CrossRef]
  30. Ernst T, Hennig J. Coupling effects in volume selective 1H spectroscopy of major brain metabolites. Magn Reson Med 1991; 21:82-96.[Medline]
  31. De Edelenyi FS, Rubin C, Estève F, et al. A new approach for analyzing proton magnetic resonance spectroscopic images of brain tumors: nosologic images. Nat Med 2000; 6:1287-1289.[CrossRef][Medline]
  32. Martínez-Pérez I, Moreno A, Alonso J, et al. Diagnosis of brain abscess by magnetic resonance spectroscopy: report of two cases. J Neurosurg 1997; 86:708-713.[Medline]
  33. Connelly A, Cross JH, Gadian DG, Hunter JV, Kirkham FJ, Leonard JV. Magnetic resonance spectroscopy shows increased brain glutamine in ornithine carbamoyl transferase deficiency. Pediatr Res 1993; 33:77-81.[Medline]
  34. Grand S, Passaro G, Ziegler A, et al. Necrotic tumor versus brain abscess: importance of amino acids detected at 1H MR spectroscopy—initial results. Radiology 1999; 213:785-793.[Abstract/Free Full Text]
  35. Hochberg Y, Tamhane AC, eds. Multiple comparison procedures New York, NY: Wiley, 1987.
  36. Becker LE. Pathology of pediatric brain tumors. Neuroimaging Clin North Am 1999; 9:671-690.
  37. Brandes AA, Palmisano V, Monfardini S. Medulloblastoma in adults: clinical characteristics and treatment. Cancer Treat Rev 1999; 25:3-12.[CrossRef][Medline]
  38. Rubinstein LJ. Embryonal central neuroepithelial tumors and their differentiating potential: a cytogenic view of a complex neuro-oncological problem. J Neurosurg 1985; 62:795-805.[Medline]
  39. Rémy C, Arés C, Ziegler A, et al. In vivo, ex vivo, and in vitro one- and two-dimensional nuclear magnetic resonance spectroscopy of an intracerebral glioma in rat brain: assignment of resonances. J Neurochem 1994; 62:166-179.[Medline]
  40. Kuesel AC, Sutherland GR, Halliday W, Smith ICP. 1H MRS of high grade astrocytomas: mobile lipid accumulation in necrotic tissue. NMR Biomed 1994; 7:149-155.[Medline]
  41. Sijens PE, Knopp MV, Brunetti A, et al. 1H MR spectroscopy in patients with metastatic brain tumors: a multicenter study. Magn Reson Med 1995; 33:818-826.[Medline]
  42. Gotsis ED, Fountas K, Kapsalaki E, Toulas P, Peristeris G, Papadakis N. In vivo proton MR spectroscopy: the diagnostic possibilities of lipid resonances in brain tumors. Anticancer Res 1996; 16:1565-1568.[Medline]
  43. Kleihues P, Burger PC, Scheithauer BW. The new WHO classification of brain tumours. Brain Pathol 1993; 3:255-268.[Medline]
  44. Bell JD, Brown JCC, Nicholson JK, Sadler PJ. Assignment of resonances for "acute-phase" glycoproteins in high resolution proton NMR spectra of human blood plasma. FEBS Lett 1987; 215:311-315.[CrossRef][Medline]
  45. Lenkinski RE. MR spectroscopy: clinical tool or research probe? (revisited). Acad Radiol 2001; 8:567-570.[CrossRef][Medline]
  46. Roser W, Hagberg G, Mader I, et al. Assignment of glial brain tumors in humans by in vivo 1H-magnetic resonance spectroscopy and multidimensional metabolic classification. MAGMA 1997; 5:179-183.
  47. Miller AJ, ed. Subset selection in regression London, England: Chapman & Hall, 1990.



This article has been cited by other articles:


Home page
 JNMHome page
B. J. Pichler, H. F. Wehrl, and M. S. Judenhofer
Latest Advances in Molecular Imaging Instrumentation
J. Nucl. Med., June 1, 2008; 49(Suppl_2): 5S - 23S.
[Abstract] [Full Text] [PDF]

Home page
 JNMHome page
A. Stadlbauer, O. Prante, C. Nimsky, E. Salomonowitz, M. Buchfelder, T. Kuwert, R. Linke, and O. Ganslandt
Metabolic Imaging of Cerebral Gliomas: Spatial Correlation of Changes in O-(2-18F-Fluoroethyl)-L-Tyrosine PET and Proton Magnetic Resonance Spectroscopic Imaging
J. Nucl. Med., May 1, 2008; 49(5): 721 - 729.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
J. F. Faria, S. M. Goldman, J. Szejnfeld, H. Melo, C. Kater, P. Kenney, M. P. Huayllas, G. Demarchi, V. V. Francisco, C. Andreoni, et al.
Adrenal Masses: Characterization with in Vivo Proton MR Spectroscopy Initial Experience
Radiology, December 1, 2007; 245(3): 788 - 797.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
S. Chawla, S. Wang, R.L. Wolf, J.H. Woo, J. Wang, D.M. O'Rourke, K.D. Judy, M.S. Grady, E.R. Melhem, and H. Poptani
Arterial Spin-Labeling and MR Spectroscopy in the Differentiation of Gliomas
AJNR Am. J. Neuroradiol., October 1, 2007; 28(9): 1683 - 1689.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
A. Stadlbauer, C. Nimsky, S. Gruber