Predictive Value of 1 Month Postoperative MRSI and FDG-PET Evaluations of Glioblastomas

Dorra Ben Sellem; Waisse Waissi; Mojdeh Dormishian; Caroline Bund; Jean-Louis Dietemann; Marie-Pierrette Chenard; Georges Noël; Izzie- Jacques Namer

doi:10.4172/2167-7964.1000220

ISSN:2167-7964

OMICS Journal of Radiology

Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.

Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on Medical, Pharma, Engineering, Science, Technology and Business

Predictive Value of 1 Month Postoperative MRSI and FDG-PET Evaluations of Glioblastomas

Dorra Ben Sellem^1#, Waisse Waissi^2#, Mojdeh Dormishian¹, Caroline Bund¹, Jean-Louis Dietemann³, Marie-Pierrette Chenard⁴, Georges Noël² and Izzie-Jacques Namer^1,5,6*
¹Department of Biophysics and Nuclear Medicine, Hautepierre Hospital, University Hospital of Strasbourg , Strasbourg, France
²Department of Radiotherapy, Centre Paul Strauss, Strasbourg, France
³Department of Radiology, Hautepierre Hospital, University Hospital of Strasbourg, Strasbourg, France
⁴Anatomical pathology service, Hautepierre Hospital, University Hospital of Strasbourg, Strasbourg, France
⁵ICube, University of Strasbourg/CNRS UMR 7357, Strasbourg, France
⁶FMTS - Faculty of Medicine, Strasbourg, France
^#Authors contributed equally
*Corresponding Author :	Izzie-Jacques Namer Department of Biophysics and Nuclear Medicine, Hautepierre Hospital University Hospital of Strasbourg, Strasbourg, France Tel: +33 388 127 550 Fax: +33 388 128 342 E-mail: Izzie.Jacques.NAMER@chru-strasbourg.fr
Received date: February 26, 2016; Accepted date: April 04, 2016; Published date: April 08, 2016
Citation: Sellem DB, Waissi W, Dormishian M, Bund C, Dietemann JL, et al. (2016) Predictive Value of 1 Month Postoperative MRSI and FDGPET Evaluations of Glioblastomas. OMICS J Radiol 5:220. doi:10.4172/2167-7964.1000220
Copyright: © 2016 Sellem DB, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Visit for more related articles at OMICS Journal of Radiology

View PDF Download PDF Tables & Figures

Abstract

Aim: The purpose of this study was to investigate prospectively the predictive value of positron emission tomography with proton magnetic resonance spectroscopic imaging (MRSI) and 18F-fluorodeoxyglucose (FDG-PET) performed the 1st month after surgery and before radio- chemotherapy in 43 patients with glioblastoma (GBM).

Patients and methods: Metabolite concentrations were quantified using LCModel. Overall survival (OS) and progression free survival (PFS) were calculated including all 43 patients using Kaplan-Meier curves, and the Cox proportional hazard model was used to calculate the predictor of survival.

Results: At the end of the follow-up period, all patients died within a period of 1–70.2 months. In 32 patients (74.4%), increased FDG-uptake was seen around the resection cavity and abnormal metabolic profiles on MRSI, indicative of residual disease, were present in all patients. There was no significant difference between the median OS in patients with hypometabolic FDG lesions compared to patients with hypermetabolic FDG lesions. On univariate analysis, normalized choline-containing compounds/creatine (nCho/Cr) and normalized lactate/creatine (nLac/Cr) were significantly predictive of OS and nLac/Cr and normalized N-Acetylaspartylglutamate and NAcetylaspartate/ creatine (NAA/Cr) were significantly predictive of PFS.

Conclusions: nCho/Cr and nLac/Cr ratio after surgery and before radio-chemotherapy were independent metabolic predictive factors of OS times in newly diagnosed patients with GBM.

Keywords

Brain; Glioma; Glioblastoma; MR spectroscopy; FDGPET; Radiotherapy; Chemotherapy; Prognosis

Introduction

Glioblastoma (GBM) is the most common (approximately 50%) and most aggressive primary brain tumor [1]. Despite recent developments in surgical technique and the current standard-of-care radiotherapy plus concomitant and adjuvant temozolomide chemotherapy, the prognosis to date remains poor, with a median survival range of 15-18 months and a 2 year survival rate of 26.5% [2-5].

GBMs have a propensity for infiltration, especially along white matter tracks, and surgical removal of the entire lesion is often not achievable because of the tumor’s large size or difficult surgical location of the neoplasm. It is inevitable that a large percentage of patients will have residual disease after surgical resection and the tumor progression in the vicinity of the original GBM site remains the most prevalent form of failure after treatment [6]. It is important to define the full extent of residual tumor in order to adapt adjuvant therapies, mainly radiotherapy target volumes. However, the ability of contrastenhanced conventional anatomic imaging (MRI, CT) to differentiate between surgery-induced changes and residual tumor is limited [6-8].

In recent years, integrating metabolic imaging such as FDG-PET and MRSI into imaging-guided surgery or physical imaging for radiation treatment planning has been suggested [6,8-13]. FDG-PET is the most widely used imaging technique that provides glucose metabolism information in brain tumours. FDG, a glucose analog, is actively transported across the blood–brain barrier into the cell, and therefore a disturbance of the blood–brain barrier does not influence FDG-uptake in the tumor. MRSI is another non-invasive imaging technique that is able to characterize the spatial distribution of pathologic biochemical changes in the tumor on the basis of the levels of tissue metabolites. Metabolic imaging has revealed a clear discordance between the GBM target volumes that result from metabolic imaging and those derived from anatomic imaging.

The aim of this study was to determine, in a prospective longitudinal study of a homogenous group of patients with GBM, the predictive values of FDG-PET and MRSI performed the 1^st month after surgery and before radio chemotherapy on clinical outcome.

Patients and Methods

Patient population

Forty-three consecutive patients with a histologically confirmed GBM (grade IV astrocytoma, according to the histological criteria of the World Health Organization classification) were included in this prospective study, from March 2004 to October 2007. All patients underwent a gross surgical resection and the 1^st month after surgery (before radio chemotherapy) MRI/MRSI and FDG-PET were performed within 72 h. All patients received the same standard radiation protocol (60 Gy) plus concomitant and adjuvant temozolomide treatment [14,15]. The follow-up consisted of a clinical and MRI evaluation.

MRI and MRSI

All MRI and MRSI investigations were performed on a 1.5 Tesla MR unit (AVANTO, Siemens Medical Systems, Erlangen, Germany). MRI sequences included conventional sagittal spin-echo T₁-weighted, axial turbo spin-echo proton density, T₂-weighted and axial fluid-attenuated inversion recovery (FLAIR) images. Multiplanar spin-echo T₁- weighted and 3D magnetization-prepared rapid-acquisition gradientecho (MPRAGE) images with intravenous injection of 0.2 mmol/kg of gadolinium contrast agent were obtained following MRSI. The spatial resolution was 1 × 1 × 1 mm for the 3D MPRAGE sequence and 1 × 1 × 5 mm for all 2D sequences.

MRSI spectra were obtained using a double-spin echo pointresolved spectroscopy sequence (PRESS) with 2D or 3D 16 × 16 phaseencodings depending on the resection cavity. Measurement parameters were as follows: TR, 1500 ms; TE, 135 ms; excitations, 2; slice thickness, 15 mm; FOV, 240 mm; data points, 1024; and acquisition time, 7 min and 14 min for 2D and 3D sequences, respectively. The 2D or 3D of acquisition volume was positioned to cover the edges of the resection cavity and the majority of abnormalities and contralateral normal-appearing tissue.

All 2D and 3D spectra chemical-shift imaging (CSI) data were then analyzed using LCModel version 6.3 k-1 K [16]. For each case a subset of the voxels in the slice were selected and analyzed by LCModel in one multivoxel run according to LCModel and LCMgui user’s manual. A file containing the basis set of model metabolite spectra for PRESS with TE=135 ms freely available with LCModel was used for analyzing our data. Different metablolites concentrations including alanine, Cr, glutamine, glutamate, glutamine, Cho, glutathione, inositol, taurine, Lac, NAA were calculated with LCModel. For each metabolite concentration ratios to Cr were calculated. Only metabolites with estimated standard deviations (%SD) equal or less than 15% (Cho, NAA and Lac) were used for statistical analysis.

FDG-PET

FDG-PET imaging was performed using the Discovery ST combined PET/CT (GE Medical System, Milwaukee, USA). ¹⁸F-FDG was injected intravenously at 2.5 MBq/kg after a fast of at least 6 h (with the exception of ad libitum water intake) and capillary glycemia lower than 6.6 mmol/l. Image acquisition was initiated 30 min and 4 h after FDG injection, including low-dose non-contrast transmission CT scan followed by a 3D PET scan with an acquisition time of 15 min. Fusioned PET/CT images were obtained after attenuation correction using CT scans with the Ordered Subset Expectation Maximization (OSEM) algorithm with two full iterations of 21 subsets. The image reconstruction matrix was 128 × 128 with a transverse field of view of 60 × 60 cm. The PET component of the scanner had an in-plane spatial resolution of 4.7 mm.

For semi quantitative analysis of FDG-uptake, we used the maximum standardized uptake value (SUV_max) per focus defined as: SUV_max = [maximum value of radioactivity concentration (kBq/ml)] / [injected dose (MBq) / patient weight (kg)].

A metabolic index, normalized SUV_max (nSUV_max), was calculated, at 30 min and 4 h, as the ratio of maximum residual tumor FDGuptake obtained using the region of interest (ROI) covering the entire residual tumor on trans axial slices to contralateral normal brain ¹⁸FFDG- uptake. An nSUV_max value greater than 1 is considered hypermetabolic; less than or equal to 1 is considered hypometabolic.

A retention index was calculated as follow: RI = [SUV (at 4 h)] – SUV (at 30 min)] / SUV (at 30 min).

Statistical analysis

Overall survival (OS) time was defined as the interval from the date of diagnosis to the date of death. Progression-free survival (PFS) was defined as the interval from the date of surgery to the date of the first clinical and/or MRI evidence of disease progression. Data were analysed using the SPSS system. Receiver Operating Characteristic (ROC) curve analyses were performed to estimate the sensitivity and the specificity at the optimal nSUV threshold and all MRSI parameters for distinguishing the two extreme survival groups.

For univariate analysis, survival curves were constructed using the Kaplan-Meier method. A log rank test was used to delineate which covariates independently influenced the patient’s outcome. All variables significant at p<0.05 in the univariate analysis were included in Cox regression model.

Results

Clinical outcome

The median age of the patients at the time of surgery was 64 years (range, 18-79 years). There were 15 women (35%) and 28 men (65%). At the end of the follow-up period, the 43 patients had died within a period of 1-70.2 months (median, 13 months). The median PFS was 6.7 months. According to OS time, patients were divided into three subgroups: group 1 with a low survival rate (<9 months) included 14 patients, group 2 with a high survival rate (>18 months) included 15 patients, and the last with intermediary survival included 14 patients.

FDG-PET analysis

In 32 patients (74.4%), higher levels of FDG-uptake (nSUV_max at 30 min) were seen around the resection cavity showing residual disease (Figure 1) and in 11 (25%) patients, was less than or equal to 1 (Figure 2).

There was no significant difference between the median OS in patients with hypometabolic FDG lesions compared to patients with hypermetabolic FDG lesions. The Kaplan-Meier plot showed that the median OS was not significantly different between group 1 and group 2 (p=0.138, log-rank) (Figure 3). There was also no significant relationship between nSUV_max (4 h), RI and OS.

MRSI analysis

An optimal n(Cho/Cr) cut-off level (2.56) clearly differentiating the two prognostic groups, with good sensitivity and specificity of 71.40%, and 73.30% was obtained from the ROC curves. The Kaplan-Meier curves (Figure 4) showed that the OS of patients with n(Cho/Cr) greater than or equal to 2.56 was significantly less than the survival of patients with n(Cho/Cr) less than 2.56 (p=0.02, log-rank). The Kaplan- Meier curves of PFS for n(Cho/Cr) showed no significant relationship (p=0.14, log-rank).

An optimal n(Lac/Cr) cut-off level (1.64) separating the two prognostic groups, with good sensitivity and specificity of 85.7%, and 71.4% was obtained from the ROC curves. The Kaplan-Meier curves (Figure 5) showed that the OS of patients with n(Lac/Cr) greater than or equal to 1.64 was significantly lower compared to the survival of patients with n(Lac/Cr) less than 1.64 (p=0.02, log-rank). The Kaplan- Meier curves of PFS for n(Lac/Cr) showed that the PFS of patients with n(Lac/Cr) greater than or equal to 1.64 was significantly lower than the PFS of patients with n(Lac/Cr) less than 1.64 (p=0.03, log-rank).

An optimal n(NAA/Cr) cut-off level (0.6) separating the two prognostic groups with sensitivity and specificity of 64.3% and 78.6%, respectively was obtained from the ROC curves. The Kaplan-Meier curves (Figure 6) showed that the OS of patients with n(NAA/Cr) less than 0.6 was lower than the survival of patients with n(NAA/Cr) greater than or equal to 0.6 but the difference was not significant (p=0.06, log-rank). The Kaplan-Meier curves of PFS for n(NAA/Cr) showed that the PFS of patients with n(NAA/Cr) less than 0.6 was significantly lower than the PFS of patients with n(NAA / Cr) greater than or equal to 0.6 (p=0.01, log-rank).

Cox univariate analysis confirmed the association between OS and the n(Cho/Cr) and n(Lac/Cr) levels. According to PFS, n(Lac/Cr) and n(NAA/Cr) were significant prognosis factor in univariate analysis (Table 1).

Discussion

Several prognostic factors have been identified in patients with GBM: age, Karnofsky performance status, pathologic grade, and tumor location, the extent of surgery, neurological status, genetic and molecular biomarker status, and concomitant chemotherapy [17-22].

Major efforts have been undertaken within recent years to apply metabolic imaging such as FDG-PET and MRSI to GBM that will add an understanding of the metabolic and cellular picture of the tumor to the anatomic or structural information derived from conventional imaging techniques such as MRI and CT [6,9,21-23].

In this current study, we investigated the prognostic values of two functional imaging modalities, FDG-PET and MRS, in patients with GBM after surgery and before radio chemotherapy. FDG-PET, which reflects glucose metabolism, has been widely used in patients with malignant brain tumours for more than a decade. However, its utility remains controversial: it depends on the purpose to achieve. It has been used before treatment [21,24,25] to guide stereotactic biopsies of intracerebral tumours [26,27]. It has been correlated with glioma grade [28,29], gadolinium enhancement on MR images [30], and MRSI findings [31]. FDG-PET imaging was predictive of the progression of low-grade gliomas to a higher grade [32]. FDG-PET was also used to define the volume for radiation dose escalation of GBMs [33]. And to clearly distinguish between GBM recurrent tumors and radiation necrosis [34,35].

The prognostic utility of FDG-PET remains controversial. Many studies have demonstrated an inverse statistical correlation between deoxyglucose trapping by the tumor and survival. Tralins et al. showed that the volume of FDG-PET avid tumor is predictive of survival and time to tumor progression [9]. More recently, Colavolpe et al. [36] demonstrated that FDG uptake before surgery is an independent prognostic factor in high-grade glioma. In a prospective pilot study, De Witte et al. compared MRI and FDG-PET in a group of ten patients, as a postoperative assessment of the extent of resection. They emphasized that, even after complete surgical resection, signs of residual tumor on either MRI or FDG-PET were associated with a high probability of recurrence within 6 months [21]. Disagreement regarding the value of FDG-PET studies is often related to heterogeneity in the patient cohort.

In the present study, we investigated only patients with a histologically confirmed GBM, imaged 1 month after surgery, before radio chemotherapy. We found that residual tumor was hypermetabolic in 74.4% of cases and hypometabolic in 25% of cases. The median OS was 11 months in patients with hypometabolic lesions and 15 months in patients with hypermetabolic lesions. In fact, GBMs are characterized histologically by the presence of necrosis, which derives from the mismatch between rapid cell growth and insufficient endothelial development. Kubota et al. have shown that prenecrotic cells are hypermetabolic [37]. Intensity of necrosis may differ among GBMs: in tumours where macroscopic necrosis largely predominates, metabolism may be reduced. In such patients, we found that there was a non-significant trend toward shorter OS compared to those with hypermetabolic lesions.

However, due to the FDG uptake in normal brain, the sensitivity of detecting recurrent or residual tumor is low. In the last decade, studies have focused on other radio markers or on the coregistration of PET and MRI, which has increased sensitivity over using either modality alone [35]. Therefore, the specificity of distinguishing gliomas from normal tissue could be increased from 68% with the use of MRI alone to 97% with the use of MRI in conjunction with ¹⁸Ffluoroethyltyrosine (FET-PET) and MRSI [38].

Radiolabeled amino acids have been of particular interest, especially for the identification of low-grade tumor extent because they show high uptake in biologically active tumor tissue and low uptake in normal brain. Their use for tumor imaging is based on the observation that amino acid transport and/or metabolism are up regulated in malignant transformation [38]. The radiolabeled amino acids commonly used in clinical studies are ¹¹C-methionine (MET-PET), ¹⁸F-dihydroxyphenylalanine (F-DOPA), and FET-PET. The rate of uptake is correlated with expression of Ki-67 proliferating cell nuclear antigen [39] and micro vessel density [40,41]. Therefore, Kim et al. found that in contrast to FDG-PET, MET-PET was found to be an independently significant prognostic factor, and MET uptake was correlated with cellular proliferation [36]. More recently, Kinoshita et al. [42] developed a novel image analysis method using matched FDGPET and MET-PET scans to calculate a decoupling score.

The second metabolic technique we have investigated is MRSI, which allows non-invasive exploration of the spatial distribution of metabolic differences in brain lesions during an MRI examination. It can characterize biochemical, metabolic, and pathologic changes in brain tissue [42,43]. It provides information on tumor activity on the basis of the levels of cellular metabolites, including Cho, Cr, NAA, and Lac/lipid [6]. Cho is a membrane component, and it is thought to be a marker of increased membrane turnover or higher cellular density. It is increased in tumours and reflects the hyper cellularity and hyperplasia of the tumor [44]. NAA is a neuronal metabolite marker, not present in other central nervous system cells. It is decreased in tumours due to neuronal loss or dysfunction [6,44]. Cr is necessary for cellular bio energetic processes and osmotic balance and might be a potential marker for cell energetics. Lactate is an end-product of modulated aerobic glycolysis in tumoral cell, characteristic by pyruvate generated from glucose metabolism is preferentially converted to lactate as opposed to entering the tricarboxylic acid cycle [45]. Lipid peaks generally correlate with necrosis and represent cellular breakdown [6].

Spectra associated with tumor are distinguished from those associated with normal tissue by the characteristic increase in the resonance peak of Cho and reduction in the NAA peak [44,46]. In fact, a significant finding from a study by Dowling et al. [42] showed that 100% of biopsy specimens obtained from regions where Cho was elevated more than two SDs above normal and the NAA was decreased at least two SDs below normal contained tumor. In our study, abnormal metabolic profiles on MRSI, indicative of residual tumor, were present in all patients. A decrease in the NAA peak was observed in all patients. An increase in the Cho peak was noted in 75.6% of cases. A decrease in the Cr peak was observed in 80% of patients. The range of Cho increase and NAA decrease is compatible with the range of tumor infiltration [46].

Before treatment, this was used to define the sites of metabolically active tumor inside and outside MRI lesions and to assess their infiltration of the brain [12,47-50]. Guillevin et al. have suggested that MRSI was a reliable tool to evaluate the proliferation activity of WHO grade II glioma and to identify potentially more aggressive clinical behaviour [48]. Laprie et al. [12] have demonstrated that areas of metabolic abnormalities inside tumours were predictive of the site of relapse compared with regions of normal Cho/NAA. This predictive value of MRSI has also been described in association with perfusion and diffusion MRI, providing additional information on T₁ contrast enhancing images [51]. It was shown that there is a clear relationship between high cerebral blood volume, a low apparent diffusion coefficient, and high Cho for GBMs.

After surgical resection and before radio chemotherapy, it was suggested that MRSI would be a valuable diagnostic tool for the assessment of residual disease, clearly demonstrating that MRSI provides additional information regarding the tumour’s spatial extent [6,49,50]. In general, the tumor volume determined by MRSI is greater than that shown by MRI [6]. Even if surgical removal of the entire lesion is often not achievable because of the lesion’s large size or a difficult surgical location of the neoplasm, incorporating MRSI data into postoperative treatment planning may improve target definition. Analysis of MRSI metabolites may help differentiate regions of aerobic from regions of anaerobic metabolism, thus detecting hypoxic areas. The regions with low Cr and high Lac will require a greater dose of radiotherapy if poor oxygenation is causing such effects. This study investigated the prognostic factor of MRSI metabolites. We found that n(Lac/Cr) and n(Cho/Cr) values were independent metabolic predictive factors of OS time in newly diagnosed patients with GBM after surgery and before radio chemotherapy.