An Intra-Individual Comparison of MRI, [18F]-FET and [18F]-FLT PET in Patients with High-Grade Gliomas

Objectives Intra-individual spatial overlap analysis of tumor volumes assessed by MRI, the amino acid PET tracer [18F]-FET and the nucleoside PET tracer [18F]-FLT in high-grade gliomas (HGG). Methods MRI, [18F]-FET and [18F]-FLT PET data sets were retrospectively analyzed in 23 HGG patients. Morphologic tumor volumes on MRI (post-contrast T1 (cT1) and T2 images) were calculated using a semi-automatic image segmentation method. Metabolic tumor volumes for [18F]-FET and [18F]-FLT PETs were determined by image segmentation using a threshold-based volume of interest analysis. After co-registration with MRI the morphologic and metabolic tumor volumes were compared on an intra-individual basis in order to estimate spatial overlaps using the Spearman's rank correlation coefficient and the Mann-Whitney U test. Results [18F]-FLT uptake was negative in tumors with no or only moderate contrast enhancement on MRI, detecting only 21 of 23 (91%) HGG. In addition, [18F]-FLT uptake was mainly restricted to cT1 tumor areas on MRI and [18F]-FLT volumes strongly correlated with cT1 volumes (r = 0.841, p<0.001). In contrast, [18F]-FET PET detected 22 of 23 (96%) HGG. [18F]-FET uptake beyond areas of cT1 was found in 61% of cases and [18F]-FET volumes showed only a moderate correlation with cT1 volumes (r = 0.573, p<0.001). Metabolic tumor volumes beyond cT1 tumor areas were significantly larger for [18F]-FET compared to [18F]-FLT tracer uptake (8.3 vs. 2.7 cm3, p<0.001). Conclusion In HGG [18F]-FET but not [18F]-FLT PET was able to detect metabolic active tumor tissue beyond contrast enhancing tumor on MRI. In contrast to [18F]-FET, blood-brain barrier breakdown seems to be a prerequisite for [18F]-FLT tracer uptake.


Introduction
Magnetic resonance imaging (MRI), as the gold standard diagnostic tool for brain tumors, offers high spatial resolution and is widely available [1]. In high-grade gliomas (HGG) the area of contrast enhancement on MRI T1-weighted sequences is generally assumed to reflect the main tumor burden [1]. In neuropathologic studies, however, invasive glioma cells can be found far beyond contrast enhancing areas [2][3][4][5]. Recently, molecular imaging studies using the amino acid tracers [ 11 C]-MET and [ 18 F]-FET revealed that in HGG patients the ''metabolic tumor volumes'' are frequently larger on PET compared to the corresponding ''morphologic contrast enhancing tumor volumes'' on MRI. This observation indicates that the main tumor burden may be substantially underestimated on standard MRI [6,7].
O-(2-[ 18 F]-fluoro-Ethyl)-L-tyrosine ([ 18 F]-FET) is an amino acid tracer frequently used in the management of glial brain tumors [8]. [ 18 F]-FET uptake correlates with tumor cell density and proliferation rate as well as with microvascular density. From clinical studies there is increasing evidence for the practical value of [ 18 F]-FET PET in addition to MRI. Complete resection guided by [ 18 F]-FET tracer uptake in HGG increased overall survival [9] and [ 18 F]-FET PET-based radiotherapy planning in HGG improved target volume definition [10,11] and improved surgery planning in low-grade gliomas for hot spot detection with static images [12,13] and dynamic acquisition methods [14]. For treatment monitoring [ 18 F]-FET PET enabled earlier detection of tumor progression after concomitant chemo-/radiotherapy [15], during adjuvant chemotherapy [16,17], and in the course of anti-angiogenic therapy [18] and local treatment strategies [19,20].
Currently, little is known about how [ 18 F]-FET and [ 18 F]-FLT directly compare to each other and to MRI in individual HGG patients. Therefore, the objective of this retrospective study was an intra-individual comparison of MRI cT1 and T2 sequences to the corresponding [ 18 F]-FET and [ 18 F]-FLT tracer uptakes in patients with HGG to better determine the spatial overlap and the practical value of these nuclear imaging modalities. We quantitatively assessed lesion-to-brain uptake ratios in PET as well as PET-based metabolic and MR-based morphologic tumor volumes and calculated territorial overlaps using a three-dimensional volumetric approach.

Patient population
Patients gave written informed consent to both PET and MRI investigations during routine diagnostic procedure. The ethics committee of Innsbruck Medical University approved the retrospective data evaluation of imaging and clinical data from those patients. All data were stored in the clinics' database. The ethics committee waived the need for another written informed consent from those patients to retrospectively analyze their data.
In this study 23 patients ( Histological diagnoses of the study population revealed glioblastoma multiforme WHO IV (GBM; n = 16), anaplastic astrocytoma WHO III (AA, n = 5), anaplastic oligodendroglioma WHO III (AO, n = 2). All three newly diagnosed patients had no treatment prior to imaging. In 20 HGG patients who were diagnosed with tumor recurrence (14 patients with first tumor recurrence, 5 with a second tumor recurrence and 1 patient with third tumor recurrence), treatment prior to imaging included surgery in all patients (19 macroscopic total resections, 1 subtotal tumor resections and 3 stereotactic biopsies), radiation and temozolomide chemotherapy according to the EORTC 26981/ 22981 NCIC CE.3 protocol [23] in 19 patients; one patient was treated with adjuvant PCV chemotherapy. Mean time interval from radiation until tumor progression was 1.9 years (range 0.1-8.3 years). At the time of neuroimaging 6 patients received corticosteroids (dexamethasone dose ranging between 2 mg and 12 mg daily). None of the patients had previous treatment with bevacizumab. Individual clinical data are summarized in Table 1.

MR imaging protocol
MRI studies were conducted on a 1.5 Tesla scanner (Sonata, Siemens-Erlangen, Germany) and included T1-weighted (TR = 1860 ms, TE = 4.38 ms with 1.2 mm slice thickness, 2566192 matrix), T2-weighted and fast-spin echo (6600 ms/ 100-110 ms, 2 mm slice thickness, 3206240 matrix) sequences. Post-contrast T1-weighted images were acquired 5 minutes after contrast agent injection (Omniscan, Dotarem, 0.1 mmol/kg). The scan duration was five minutes. Images were acquired in a three-dimensional mode in contiguous transaxial slices of the entire brain. An iterative reconstruction of the attenuationcorrected emission data set was obtained using the ordered subset expectation maximization (OSEM) algorithm.

MRI segmentation
MRI segmentation and morphologic tumor volume calculation of cT1 and T2 images were performed using a semi-automated active contour method (snake evolution, ITK-SNAP 2.0). This software has already demonstrated excellent reliability and efficacy of 3D segmentation [27].

PET image analysis and segmentation
To estimate [ 18 F]-FET and [ 18 F]-FLT tracer uptakes and metabolic tumor volumes, the established semi-quantitative standard uptake value (SUV) calculation analysis [28] was performed by an in-house software package developed in Matlab [29]. To measure the maximal tracer uptake activity (SUVmax) spherical region of interest (ROI) volumes covering the maximal extension of the tumor were drawn in the affected brain hemisphere. To determine the background activity (SUV background) a spherical ROI volume was placed on the contralateral hemisphere (mirror region) including white and gray matter but not the ventricle (ROI volume ranged between 1.7 cm 3 to 2 cm 3 ). Afterwards, the lesion-to-background ratios (LBRs) were calculated by dividing SUVmax/SUV background and used for further volumetric and statistical analysis. For determination of the metabolic tumor volumes, LBR cut-off values were calculated using the receiver operating characteristic (ROC) curve analysis [30]  were considered to be metabolically active tumor and were used for image segmentation and tumor volume calculation. Afterwards a comparison of the spatial relationship between the metabolic PET and morphologic MRI volumes was performed using an overlap statistic calculation for two volumes. Importantly, due to the possible influence of a partial-volume effect on small residual tumor volumes on [ 18 F]-FET and [ 18 F]-FLT PET, metabolic tumor volumes below 2 cm 3 were considered as not significant and were not taken into account for further analysis [31].

Statistics
Data were analyzed using SPSS 20.0 statistical software. Testing on normal distribution was performed with the Kolmogorov-Smirnov-Test. Data not distributed normally were further analyzed by the Spearman correlation coefficient in order to detect correlations between tumor volumes and LBRs. The Mann-Whitney U test (MWU) was performed to detect differences in tumor volumes and WHO grading. Data are expressed in mean 6 (standard error). Probability values ,0.05 were considered as significant.

Tracer uptake beyond the borders of cT1 and T2 on MRI
In order to explain this limited metabolic trapping of [ 18 F]-FLT even into proliferative tumor areas of HGG [34], it is necessary to be aware of its mechanisms regulating transport, accumulation, and retention in tissues [35]. On a molecular level [ 18 F]-FLT uptake was shown to be associated with the activity of thymidine kinase-1 (TK1), an intracellular enzyme expressed during DNA synthesis of the cell cycle [36]. In line with this mechanism [ 18 F]-FLT uptake correlates with the Ki-67 proliferation rate in patients with newly diagnosed HGG [21]. [ 18 F]-FLT, however, has to cross the plasma membrane of cells before it can be trapped by TK1. This cellular uptake is mediated by the bidirectional equilibrative nucleoside transporter 1 (ENT1) [35,37], a widely distributed plasma membrane nucleoside transporter molecule in the central nervous system [37]. Previous studies investigating the kinetic properties of the [ 18 F]-FLT tracer in HGG also showed that [ 18 F]-FLT uptake predominantly occurs in tumor regions with disrupted BBB [38,39]. This was supported by a strong correlation between [ 18 F]-FLT LBR and the passive tracer influx rate constant K1 [39], but not with ENT1 expression or the CD34 vascular density score [35].
In the light of the increasing use of anti-angiogenic treatment strategies, such an association of [ 18 F]-FLT uptake in tumor tissue dependent on BBB permeability is clinically relevant. Bevacizumab, a monoclonal antibody against VEGF, may lead to a pseudonormalization of tumor vessels and a functional restoration of the BBB permeability [40]. Recently, [ 18 F]-FLT PET was proposed to assess treatment response to anti-angiogenic treatment in recurrent HGG [41,42]. On the basis of our results, response assessment of anti-angiogenic treatment in HGG solely based on [ [43]. More importantly, in 61% of the patients [ 18 F]-FET tracer uptake was found beyond the borders of contrast enhancing tumor on T1-weighted MRI and in 35% even beyond the borders of T2. Previous studies including patients with primary and recurrent GBM [6,44] as well as gliomas of various malignancy grades and brain metastases [7] showed that both amino acid tracers [ 18 F]-FET and [ 11 C]-MET PET delineate tumor tissue outside of MRI cT1 and T2 changes.
A [ 18 F]-FET PET signal results from specific tracer uptake into tumor and endothelial cells depending on tumor cell density and microvascular density [45] mediated by LAT amino acid transporters [46]. In addition to this predominant specific uptake, to a lesser part amino acid tracer uptake also results from nonspecific uptake due to various pharmacokinetic processes associated with a raised BBB permeability (e.g. VEGF-mediated, reactive astrocytosis [47], radiation induced necrosis [48]), variable tumor blood volume and perfusion (''blood pooling effect'' [46]). An association between [ 18 F]-FET LBR and contrast enhancement on MRI has also been shown recently [49] by our study group.
Limitations of the current analysis are the relatively small patient cohort, the retrospective study design and the lack of a histologic correlation analysis of the recurrent brain tumors. A lack of histologic correlation studies always raises concern on the presence of possible pseudoprogression. However, image analysis was performed in awareness of this radiographic phenomenon and clinical parameters (median time from radiation until tumor progression was 1.9 years) as well as radiation field analysis was highly suggestive for true tumor progression.
In conclusion, BBB breakdown seems to be a prerequisite for [ 18