high-grade cerebral glioma characterization: … · 57 the neuroradiology journal 25: 57-66, 2012...

57

TheNeuroradiologyJournal25:57-66,2012 www.centauro.it

SUMMARY – The aim of our study was to evaluate if both spectroscopy and perfusion magnetic resonance (MR) imaging are necessary to differentiate high grade gliomas from low grade tumour, or if only one of these techniques is sufficient. Sixty-five patients with cerebral glioma were retro-spectively evaluated. All patients were studied both with spectroscopy and perfusion imaging. In 43 cases histological examination showed a high grade glioma while a low grade glioma was found in 22 patients. For every patient spectroscopic maximum Cho/NAA ratio and lactate presence was established maximum relative CBV value was evaluated by perfusion MR. Both for Cho/NAA and rCBV threshold values were obtained by means of ROC curves. Then diagnostic sensitivity and spe-cificity for high grade gliomas identification was evaluated for spectroscopic data only (Cho/NAA and lactate presence that was considered a high grade glioma marker), for perfusional data only (rCBV) and finally for both spectroscopic and perfusional data together. Sensitivity was signifi-cantly highest evaluating both spectroscopic and perfusional data together (89.7%) in comparison with spectroscopy (74.4%) or perfusion (79.4%) alone. Instead specificity was slightly lower with all data (91.7%) in comparison with spectroscopy (95.8%) and perfusion (95.8%) alone. In conclusion, to characterize high grade gliomas it is more useful to evaluate spectroscopic and perfusional data together with respect only one of these techniques alone.

High-GradeCerebralGliomaCharacterization:UsefulnessofMRSpectroscopyandPerfusionImagingAssociatedEvaluation

I.APRILE1,C.TORNI1,P.FIASCHINI1,M.MUTI2

1 Department of Neuroradiology, 2 Department of Oncologic Radiotherapy, S. Maria General Hospital; Terni, Italy

Key words: brain tumours, gliomas characterization, magnetic resonance perfusion imaging, magnetic resonance spectroscopy, magnetic resonance comparative studies

Introduction

Even when Magnetic Resonance (MR) is car-ried out using conventional morphological se-quences, also after contrast media administra-tion, there is no certainty regarding the precise detection of high grade gliomas, with respect to low grade tumours. In many cases the diagno-sis can be reasonably certain, especially with regard to glioblastomas, even if morphological reports are often uncertain 1,2.

Both MR perfusion and spectroscopy pro-vide enough quantitative data to detect the malignancy grade of cerebral gliomas 3-6. Even if perfusion is a minimally invasive applica-tion because it uses a double dose of contrast media, spectroscopy involves a long study that requires high levels of patient compliance and considerable availability of time. Therefore, it

is important to establish whether both meth-ods are necessary for a correct diagnosis.

The purpose of this study was to evaluate whether both techniques - perfusion and spec-troscopy - are essential or if the quantitative data of either method give a precise enough de-tection of high grade gliomas compared to low grade ones.

Materials and Methods

Patients

We performed a retrospective study on 65 patients with past clinical records of cerebral gliomas. They had all undergone a biopsy or a resection operation with a subsequent histo-logical diagnosis of the neoplastic type (Table

58

High-Grade Cerebral Glioma Characterization: Usefulness of MR Spectroscopy and Perfusion Imaging Associated Evaluation I. Aprile

and 110 ms, 221×256 matrix, 230 mm FOV, 2 NEX, 5 mm scan thickness, 16 ETL).– Fluid attenuated inversion recovery (FLAIR) axial T2-weighted images (TR 6000 ms, TE 100 ms, 205×256 matrix, 230 mm FOV, 2 NEX, 5 mm scan thickness, 21 ETL).– PRESS sequences spectroscopic CSI with wa-ter suppression and acquisition of images with 20×20 matrix, TR 2000 ms, TE 288 ms, Hz 1000 spectral bandwidth. Single voxel sequences were acquired with the same technical parameters.– Post contrast 3D PRESTO axial images (GRE-EPI T2*) for the study of cerebral per-fusion (TR 20 ms, TE 29 ms, FA 8°, 64×128 matrix, 230 mm FOV, 1 NEX) with 40 dynamic phases and a time resolution of 1.67 s.– Images acquired post-contrast with axial, coro-nal and sagittal scans: SE T1-weighted image technique (TR 560 ms, TE 14 ms, 205×256 ma-trix, 230 mm FOV, 2 NEX, 5 mm scan thickness).

The contrast medium (Gadoteridol, Pro-Hance, Bracco, Milan, Italy) was injected via intravenous injection with an 18 Gauge needle through an automatic injector at a 5 ml/s flow rate. In all patients the contrast media dose was 0.25 mmol/Kg (a bolus of 0.2 mmol/Kg pre-ceded by a pre-bolus of 0.05 mmol/Kg).

Both spectroscopic and perfusion images were subsequently processed and converted in coloured maps on a dedicated workstation (Philips Easy Vision).

Quantitative Analysis

Perfusion and spectroscopy imaging quanti-tative data were analysed blindly by two ex-perienced neuroradiologists using a dedicated post-processing workstation. Perfusion CBV maps were evaluated by detecting the nor-malised maximum value of the lesion in rela-tion to the contralateral normal white matter value (rCBV). With reference to spectroscopy, the maximum Cho/NAA ratio of a voxel se-lected in the solid tumour area was calculated. In addition, the presence of lactate peak was also taken into account. Lastly, we estimated diagnostic sensitivity and specificity of the techniques aiming at the identification of high grade (grade III and IV) gliomas in comparison to low malignity grade lesions (grade I and II).

Both for perfusion and spectroscopy data a threshold value was established by means of ROC (Receiver Operating Characteristic) curves (Figure 1). Quantitative rCBV analysis was carried out adopting a threshold value of 3, whereas a threshold value of 1.85 was adopted

1). They consisted of 42 males and 23 females ranging in age from 19 to 89 years (the mean age of the patients was 54.12).

Table 1 Distribution of the cases by histological diagnosis.

Gangliocytoma 3

Pilocytic astrocytoma 2

Grade I astrocytoma 4

Grade II astrocytoma 10

Grade II oligodendroglioma 3

Grade III astrocytoma 10

Grade III oligodendroglioma 1

Glioblastoma 32

Sequences

All examinations were performed on a 1.5 Tesla MR scanner with 23 mT/m gradients (Gyroscan ACS-NT PT 3000, Philips Medical System, Amsterdam, The Netherlands). All pa-tients underwent both a spectroscopic investi-gation and a cerebral perfusion evaluation. In addition, a study of the brain was performed on all patients with a conventional technique of fast spin echo (FSE) images and fluid attenu-ated inversion recovery (FLAIR) T2-weighted images, spin echo (SE) T1-weighted images with and without contrast agent. Finally, two fast sagittal and coronal FSE T2-weighted im-ages were also performed on those patients who had undergone spectroscopy for a better definition of the plane of interest.

Spectroscopic study was carried out on most patients (50 cases) using chemical shift imag-ing (CSI) and acquiring a localized scan at the lesion’s equator. The remaining 15 patients underwent investigation with single voxel ac-quisition as the lesions were localized in areas where multi-voxel acquisition had proved dif-ficult (cerebellum, brainstem or in supraten-torial cortico-subcortical site). In any case a voxel of a minimum 10×10×10 mm size was acquired, making sure that it only contained the neoplastic tissue in order to avoid the “par-tial volume” effect.

For each patient the following sequences were acquired:– Spin echo (SE) axial T1-weighted images (TR 560 ms, TE 14 ms, 205×256 matrix, 230 mm field of view (FOV), two excitations (NEX), 5 mm scan thickness).– Fast spin-echo (FSE) axial proton density and T2-weighted images (TR 2000 ms, TE 8.8

59

www.centauro.it TheNeuroradiologyJournal25:57-66,2012

sets (Cho/NAA + lactate presence + rCBV). The data evaluation was based on sensitivity and specificity terms.

Results

The perfusional and spectroscopic data of all patients are reported in Table 2. Grade I and II gliomas tended to have low rCBV and Cho/NAA values (Figures 2 and 3) while high grade ones showed higher rCBV and Cho/NAA val-ues with, in some cases, lipids peak detection (Figure 4).

Diagnostic sensitivity and specificity with re-gard to the evaluation of data referring to per-fusion (rCBV), those referring to spectroscopy (Cho/NAA in association with the evaluation of lactate presence) and lastly to the evaluation of all the data analysed together (rCBV in asso-ciation with evaluation of Cho/NAA rCBV and lactate presence) are reported in Table 3.

The evaluation of rCBV with perfusion imag-ing was highly specific: only one low grade tumour presented values higher than threshold value 3.

Even the evaluation of spectroscopy param-eters provided higher specificity values (Table 3). The evaluation of all available data in joint association (perfusion + spectroscopy) notice-ably increased the diagnostic sensitivity and caused a slight reduction in specificity (Table 3).

for Cho/NAA analysis. Above these values the lesion was considered at a high grade. Lactate peak presence was considered a marker for high grade lesions. In the evaluations of more associated parameters (Cho/NAA + lactate presence; Cho/NAA + lactate presence + rCBV) the discrimination analysis method 7 was used to evaluate the ROC curve of more parameters associated with one another.

Lastly, sensitivity and specificity values of the parameters associated in couples and tri-ples were calculated in correspondence to the optimized threshold of that ROC curve. The discrimination analysis allows the definition of the discrimination function, that is the relative weight of every variable in the combination of couples or triples. It also allows the discrimina-tion threshold, that is the discrimination func-tion value beyond which the result of each case is considered to be positive. In our case the re-sult was 3.574 for the associated evaluation of rCBV, Cho/NAA and lactate presence, whereas it was 1.756 for the associated evaluation of Cho/NAA and lactate presence.

Therefore, we first evaluated sensitivity and specificity values obtained from the analysis of perfusion imaging (rCBV) data. Secondly we evaluated those values relating to spectros-copy data (as a couple: Cho/NAA association + lactate presence) and lastly those relating to the combined perfusion and spectroscopy data

A B

Figure 1 ROC curves of rCBV (A) and Cho/NAA ratio (B).

60


Patient Histology rCBV Cho/NAA Lip 1 gangliocytoma normal 1.46 – 2 gangliocytoma normal 1.00 – 3 gangliocytoma normal 1.75 – 4 pilocytic astrocytoma 2 3.50 – 5 pilocytic astrocytoma 10 4.36 – 6 grade I astrocytoma normal 2.75 – 7 grade I astrocytoma normal 1.33 – 8 grade I astrocytoma normal 1.83 – 9 grade I astrocytoma normal 1.84 –10 grade II astrocytoma normal 1.24 –11 grade II astrocytoma normal 1.23 –12 grade II astrocytoma normal 4.40 –13 grade II astrocytoma normal 2.87 –14 grade II astrocytoma 3 1.40 –15 grade II astrocytoma normal 0.61 –16 grade II astrocytoma normal 1.00 –17 grade II astrocytoma normal 0.61 –18 grade II astrocytoma 3 1.36 –19 grade II astrocytoma normal 1.51 –20 grade II oligodendroglioma normal 0.90 –21 grade II oligodendroglioma normal 11.33 –22 grade II oligodendroglioma 2 1.25 –23 grade III astrocytoma 7 1.44 +24 grade III astrocytoma 4 14.46 –25 grade III astrocytoma 10 4.93 –26 grade III astrocytoma normal 1.56 +27 grade III astrocytoma normal 0.83 –28 grade III astrocytoma 7 0.63 –29 grade III astrocytoma normal 1.00 –30 grade III astrocytoma 8 4.20 –31 grade III astrocytoma normal 0.39 –32 grade III astrocytoma 2 1.85 –33 grade III oligodendroglioma 7 4.49 –34 glioblastoma 10 2.46 –35 glioblastoma 9 7.00 +36 glioblastoma 9 2.14 –37 glioblastoma 9 1.20 +38 glioblastoma 8 1.50 +39 glioblastoma normal 3.85 –40 glioblastoma 8 3.85 +41 glioblastoma normal 2.44 –42 glioblastoma 14 5.06 +43 glioblastoma 10 9.50 –44 glioblastoma 8 6.85 +45 glioblastoma 4 6.13 +46 glioblastoma 10 1.18 +47 glioblastoma 14 3.07 +48 glioblastoma 12 2.60 +49 glioblastoma normal 4.00 +50 glioblastoma 10 2.10 +51 glioblastoma 2 9.28 +52 glioblastoma 12 7.71 –53 glioblastoma 12 7.69 –54 glioblastoma 8 2.87 –55 glioblastoma 4 2.00 +56 glioblastoma 8 0.80 +57 glioblastoma 8 2.30 +58 glioblastoma 10 0.72 –59 glioblastoma 10 3.14 +60 glioblastoma 8 4.80 +61 glioblastoma 3 5.66 +62 glioblastoma 11 3.33 +63 glioblastoma 11 3.90 +64 glioblastoma 2 3.32 –65 glioblastoma 6 2.99 –

Table 2 All patients’ perfusional and spectroscopic data; rCBv values below 2 were considered normal.

61


of biological membranes and is, therefore, a marker of replicating cells) and is inversely pro-portional to that of N-acetyl aspartate which, instead, is a marker of normal neurons 17-20. For this reason the glioma malignancy grade is di-rectly proportional to the Cho/NAA ratio.

Furthermore, spectroscopy detects the pres-ence of lactate as a necrosis marker. When the lactate peak is present it characterizes high grade gliomas, even if the report can be un-specified because necrosis can also be present in non-neoplastic lesions 17-20. Diffusion is the third advanced MR technique that can be em-ployed to characterize gliomas. With this tech-nique the MR signal depends on water mol-ecule diffusivity and it is, therefore, inversely proportional to cell density. With this reason-ing some authors tried to identify high grade gliomas with respect to those at a low grade, where high grade lesions are characterized by a lower diffusivity 21. More recent studies, how-ever, have not confirmed the initial findings and there is often an excessive overlap between the two tumour groups with consequent impos-sibility of definitite detection 22.

Therefore, the advanced MR techniques that currently provide reliable quantitative data for gliomas characterization are perfusion and spectroscopy 23-26.

Yang et al. 27 evaluated 17 cerebral gliomas with diffusion, perfusion and spectroscopy to detect their malignancy grade. They only as-sessed the average of the numeric data result-ing from the three techniques in the analysis of the two types of tumours. Important differ-ences both with reference to apparent diffusion coefficient (ADC) as well as to rCBV and Cho/NAA relationship were spotted. They are, how-ever, data of little importance due to the low number of test cases. In 2003 Law et al. 28 pub-lished a paper on the characterization of high grade gliomas. This study was accomplished with a high number of cases (160 cases) and represents the first important reference point in the evaluation of the diagnostic indications of perfusion and spectroscopy.

The authors evaluated the diagnostic sensi-tivity and specificity of both perfusion (rCBV) and spectroscopy (Cho/NAA) separately, as

Discussion

It is necessary to identify the lesion type in patients who have undergone MR for the detec-tion of suspected cerebral tumour. It would be equally important to consider the presence of tumoral tissue infiltrating beyond the enhance-ment area to be able to differentiate metastases from gliomas. As a final step, it would be neces-sary to define the malignancy grade by differen-tiating low grade gliomas (I and II grade) from those at a high grade (III and IV grade) 1-3,8,9.

In some cases the morphological study itself, with and without contrast, accomplishes differ-ential diagnoses between low and high grade tumours. Glioblastomas often have typical morphologic characteristics (corpus callosum involvement, conspicuous vasogenic edema, necrotic aspect), different from low grade as-trocytic gliomas. However, in many cases the differential diagnosis is not easy, especially among gliomas of II and III grade with low or no enhancement after contrast 1-3.

Gliomas characterisation is important firstly because it precisely assesses the risk/benefit re-lationship of the surgical operation. Secondly, it is generally known that, in some cases, the histological examination can underestimate the tumour malignancy grade in the case of bi-opsy or partial ablation. In this case both per-fusion and spectroscopy imaging can be useful to detect the lesion malignancy grade 10-13 cor-rectly, in addition to identifying accurately the most malignant part of the lesion for a proper biopsy sampling.

Perfusion imaging provides brain vascular maps that identify cerebral tissues on the ba-sis of several hemodynamic parameters. The parameter that most effectively relates to his-tological characteristics of cerebral gliomas is CBV that is proportional to the volume of micro-vessels supplying an area of interest 14-16. In this case, therefore, the malignancy grade is propor-tional to the density of the lesion’s microvessels.

Spectroscopy, on the other hand, detects the concentration of some metabolites, always in specific lesion areas. The glioma malignancy grade is directly proportional to the choline peak (which is produced during the synthesis

rCBV Cho/NAA+Lac Cho/NAA+Lac+rCBV

Sensitivity 79.4% 74.4% 89.7%

Specificity 95.8% 95.8% 91.7%

Table 3 MR perfusion and spectroscopy diagnostic sensibility and specificity in the detection of high grade gliomas.

62


Figure 2 Cerebellar low grade astrocytoma. FLAIR image (A) and post-contrast sequence (B) without tumour enhancement. In CBV maps (C) the lesion is iso-hypoperfused in comparison to controlateral cerebellar lobe, while spectroscopy (D) shows high choline and low NAA peak’s height.

B CA

D

63


Figure 3 Pilocytic astrocytoma. T2-weighted (A) and post-contrast image (B) with depiction of tumour solid tissues. Medially there is a part of the lesion hyperperfused (arrow) in CBV maps (C) and also in spectroscopic imaging (D) a voxel with high Cho peak was showed (arrow). Both in morphologic and in perfusional and spectroscopic images this low grade tumour mimic high grade aspects.

A

D

B C

64


Figure 4 High grade (III) cerebellar astrocytoma. T2-weighted (A) and post-contrast image (B) with intense lesion’s enhancement. In CBV maps (C) the tumour is isoperfused to controlateral lobe, while spectroscopy (D) shows Cho peak increasing and NAA one decreasing with respect to creatine.

A

D

B C

65


those with a low malignancy. However, other studies have contradicted this statement and have, in fact, emphasized that there is a sta-tistically significant rCBV difference between grade II oligodendrogliomas and anaplastic oli-godendrogliomas 34. For this reason study data resulting from general range of cases including all types of gliomas could not be very specific for the presence of oligodendrogliomas. On the basis of some literature articles, oligodendro-gliomas could have vascularization character-istics, assessed with perfusion RM, different from astrocytomas.

In our study we evaluated three low grade oligodendrogliomas that in all cases presented maximum rCBV values lower than the 3 limit value and hence were correctly characterized. Therefore, in our range of cases perfusion im-aging diagnostic specificity was not reduced by the presence of II grade oligodendrogliomas, as they presented vascularization characteristics similar to those of low grade astrocytomas.

Our work analysed results from a range of cases of 65 gliomas, all being studied with both techniques. Unlike other similar studies, our study also evaluated the presence of lactate as a marker of malignant gliomas. In this way we obtained very good diagnostic sensitivity val-ues mainly concerning data evaluation of spec-troscopy and perfusion in association (89.7%). On the contrary, when we isolated perfusion data from those resulting from spectroscopy, in both cases we obtained good specificity values (95.8%), with less sensitivity (Table 2).

In our range of cases we preferred to consider all gliomas, including glioblastomas (IV grade), even though it would been more interesting to assess only gliomas of I, II and III grade. How-ever, if we had excluded IV grade gliomas, the range of cases would have been too limited (33 cases).

In conclusion, perfusion and spectroscopy are both useful to detect the malignancy grade of gliomas, and both techniques should be em-ployed. Perfusion and spectroscopy, when in-dividually evaluated, have got good specificity (Table 3), but joint evaluation of both tech-niques has clearly improved diagnostic sensi-tivity by slightly reducing specificity.

well as of both methods together. The results are similar to ours with reference to spectros-copy and the associated evaluation of perfusion and spectroscopy. With reference to perfusion evaluation we recorded high specificity (95.8%) and less sensitivity (79.4%), whereas Law et al.

28 obtained high sensitivity (95%) and low spe-cificity (57.5%) using, however, a limit value lower than ours (1.75 vs 3). Another difference from our study is that they did not evaluate lactate as a marker of high malignancy grade gliomas.

A study carried out in 2008 29 analysed 44 gliomas with a 3 Tesla device. Again, impor-tant differences between gliomas at low and high grade were detected using only spectros-copy (Cho/NAA), even if in this study did not calculate diagnostic sensitivity and specificity. Another interesting work 30 with a high number of cases (150 cases) spotted a statistically im-portant difference between gliomas at low and high grade both with perfusion (rCBV) and with spectroscopy (NAA/Cr and Cho/Cr), but also in this case diagnostic sensitivity and spe-cificity had not been considered. A recent study by Fayed et al. 31 evaluated the malignity level of 24 tumours both with spectroscopy (Cho/Cr) and with perfusion (rCBV) obtaining good re-sults in relation to both diagnostic sensitivity and specificity. In the range of cases, metas-tases were also included in addition to gliomas.

Batra et al. 32 evaluated only gliomas with little enhancement after contrast, therefore ex-cluding glioblastomas that generally undergo strong enhancement. The number of cases was low (22 cases), but the work is interesting because it included only gliomas of II and III grade that are the most difficult to differenti-ate with conventional morphological sequences, even using a contrast agent. Also in this case they found statistically significant differences between the two types of tumour both with per-fusion (rCBV) and with spectroscopy (Cho/Cr).

Finally, we wish to comment on oligoden-drogliomas. It is well-known that low grade oligodendrogliomas tend to have rCBV values similar to III grade oligodendrogliomas 33. For this reason, unlike astrocytomas, perfusion can-not be used to detect malignant tumours from

66


21 Sugahara T, Korogi Y, Kochi M, et al. Usefulness of diffusion-weighted MRI with echo-planar technique in the evaluation of cellularity in gliomas. J Magn Reson Imaging. 1999; 9: 53-60.

22 Kono K, Inoue Y, Nakayama K, et al. The role of diffu-sion-weighted imaging in patients with brain tumors. Am J Neuroradiol. 2001; 22: 1081-1088.

23 Sadeghi N, D’Haene N, Decaestecker C, et al. Appar-ent diffusion coefficient and cerebral blood volume in brain gliomas: relation to tumor cell density and tumor microvessel density based on stereotactic biopsies. Am J Neuroradiol. 2008; 29: 476-482.

24 Arvinda HR, Kesavadas C, Sarma PS, et al. Glioma grading: sensitivity, specificity, positive and negative predictive values of diffusion and perfusion imaging. J Neurooncol. 2009; 94: 87-96.

25 Fayed N, Modrego PJ. The contribution of magnetic resonance spectroscopy and echoplanar perfusion-weighted MRI in the initial assessment of brain tu-mours. J Neurooncol. 2005; 72: 261-265.

26 Rock JP, Scarpace L, Hearshen D, et al. Associations among magnetic resonance spectroscopy, apparent dif-fusion coefficient, and image-guided histopathology with special attention to radiation necrosis. Neurosur-gery. 2004; 54: 1111-1117.

27 Yang D, Korogi Y, Sugahara T, et al. Cerebral gliomas: prospective comparison of multivoxel 2D chemical-shift imaging proton MR spectroscopy, echoplanar perfusion and diffusion-weighted MRI. Neuroradiology. 2002; 44: 656-666.

28 Law M, Yang S, Wang H, et al. Glioma grading: sen-sitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. Am J Neu-roradiol. 2003; 24: 1989-1998.

29 Di Costanzo A, Scarabino T, Trojsi F, et al. Proton MR spectroscopy of cerebral gliomas at 3T: spatial hetero-geneity, and tumor grade and extent. Eur Radiol. 2008; 18: 1727-1735.

30 Zonari P, Baraldi P and Crisi G. Multimodal MRI in the characterization of glial neoplasms: the combined role of single-voxel MR spectroscopy, diffusion imaging and echo-planar perfusion imaging. Neuroradiology. 2007; 49: 795-803.

31 Fayed N, Davila J, Medrano J, et al. Malignancy as-sessment of brain tumours with magnetic resonance spectroscopy and dynamic susceptibility contrast MRI. Eur J Radiol. 2008: 67: 427-433.

32 Batra A, Tripathi RP, Singh AK. Perfusion magnetic resonance imaging and magnetic resonance spectros-copy of cerebral gliomas showing imperceptible con-trast enhancement on conventional magnetic reso-nance imaging. Astralas Radiol. 2004; 48: 324-332.

33 Xu M, See SJ, Ng WH, et al. Comparison of magnetic resonance spectroscopy and perfusion weighted imag-ing in presurgical grading of oligodendroglial tumors. Neurosurgery. 2004; 56: 919-926.

34 Spampinato MV, Sith JK, Kwock L, et al. Cerebral bood volume measurements and proton MR spectros-copy in grading of oligodendroglial tumors. Am J Roet-genol. 2007; 188: 204-212.

References

1 Sartor K. MR imaging of the brain: tumors. Eur Ra-diol. 1999; 9: 1047-1054.

2 Yanaka K, Kamezaki T, Kobayashi E, et al. MR im-aging of diffuse glioma. Am J Neuroradiol. 1992; 13: 1081-1086.

3 Wilms G, Demaerel P, Sunaert S. Intra-axial brain tu-mours. Eur Radiol. 2005; 15: 468-484.

4 Majos C, Alonso J, Aguilera C, et al. Proton magnetic resonance spectroscopy ((1) H MRS) of human brain tumours: assessment of differences between tumour types and its applicability in brain tumour categoriza-tion. Eur Radiol. 2003; 13: 582-591.

5 Hakyemez B, Erdogan C, Ercan I, et al. High-grade and low-grade gliomas: differentiation by using per-fusion MR imaging. Clin Radiol. 2005; 60: 493-502.

6 Stadlbauer A, Gruber S, Nimsky C, et al. Preoperative grading of gliomas by using metabolite quantification with high-spatial-resolution proton MR spectroscopic imaging. Radiology. 2006; 238: 958-969.

7 Candini G. A statistical method for the quantification of the learning index by means of multiple answer tests Graduate-Response Test. Comput Biol Med. 2004; 34: 663-678.

8 Hourani R, Brant LJ, Rizk T, et al. Can proton MR spectroscopic and perfusion imaging differentiate be-tween neoplastic and non neoplastic brain lesions in audults? Am J Neuroradiol. 2008; 29: 366-372.

9 Law M, Cha S, Knopp EA, et al. High-grade gliomas and solitary metastases: differentiation by using per-fusion and proton spectroscopic MR imaging. Radiol-ogy. 2002; 222: 715-721.

10 Di Costanzo A, Scarabino T, Trojsi F, et al. Multipara-metric 3 T MR approach to the assessment of cerebral gliomas: tumor extent and malignancy. Neuroradiol-ogy. 2006; 48: 622-631.

11 Dowling C, Bollen AW, Noworolski SM, et al. Preop-erative proton MR spectroscopic imaging of brain tu-mors: correlation with histopathologic analysis of resec-tion specimens. Am J Neuroradiol. 2001; 22: 604-612.

12 Li X, Lu Y, Pirzkall A, et al. Analysis of the spatial characteristics of metabolic abnormalities in newly di-agnosed glioma patients. J Magn Reson Imaging. 2002; 16: 229-237.

13 Fan GG, Deng QL, Wu ZH, et al. Usefulness of diffu-sion/perfusion-weighted MRI in patients with non-en-hancing supratentorial brain gliomas: a valuable tool to predict tumour grading? Br J Radiol. 2006; 79: 652-658.

14 Sugahara T, Korogi Y, Kochi M, et al. Correlation of MR imaging-determined cerebral blood volume maps with histologic and angiographic determination of vas-cularity of gliomas. Am J Roentgenol. 1998; 171: 1479-1486.

15 Maia AC Jr, Malheiros SM, da Rocha AJ, et al. MR cerebral blood volume maps correlated with vascular endothelial growth factor expression and tumor grade in nonenhancing gliomas. Am J Neuroradiol. 2005: 26: 777-783.

16 Rollin N, Guyotat J, Streichenberger N, et al. Clini-cal relevance of diffusion and perfusion magnetic reso-nance imaging in assessing intra-axial brain tumors. Neuroradiology. 2006; 48: 150-159.

17 Fulham MJ, Bizzi A, Dietz MJ, et al. Mapping of brain tumor metabolites with proton MR spectroscopic imag-ing: clinical relevance. Radiology. 1992; 185: 675-686.

18 Poptani H, Gupta RK, Roy R, et al. Characterization of intracranial mass lesions with in vivo proton MR spec-troscopy. Am J Neuroradiol. 1995; 16: 1593-1603.

19 Shimizu H, Kumabe T, Tominaga T, et al. Noninvasive evaluation of malignancy of brain tumors with proton MR spectroscopy. Am J Neuroradiol. 1996; 17: 737-747.

20 Magalhaes A, Godfrey W, Shen Y, et al. Proton mag-netic resonance spectroscopy of brain tumors corre-lated with pathology. Acad Radiol. 2005; 12: 51-57.

I. Aprile, MDDepartment Diagnostica ImmaginiNeuroradiologiaP.Le T. Di Joannuccio 105100 Terni, ItalyTel.: +39.340.5970346E-mail: [email protected]

high-grade cerebral glioma characterization: … · 57 the neuroradiology journal 25: 57-66, 2012...

Documents