A Combined Study with 18F-FDG and 11C-Methionine Dynamic PET for the Grading of Brain Gliomas.

Purpose: Conventional MRI based on contrast enhancement and T2/FLAIR is often not sucient in differentiating grade II from grade III and grade III from grade IV diffuse gliomas. Here we assessed advanced metabolic imaging using two well characterized PET tracers, namely 18 F-FDG and 11 C-Methionine. Methods: In this prospective study, 39 patients were enrolled with diffuse gliomas of grades II, III or IV underwent dynamic [ 18 F]-FDF-PET and [ 11 C]-Methionine. The rst minutes were taken into account Results: The use of 11 C- Methionine provided signicant differences between the different histologic subgroups with a higher number of parameters than did the use of 18 F-FDG. The most informative parameter is T/Np (T/N at the peak of the rst maximum) with 11 C- Methionine. Conclusion: The study of the rst minute passage of 18 F-FDG and/or 11 C- Methionine through the tumor and healthy tissues in brain gliomas could not only allow improving the identication of the different glioma grades, but also to shorten the time spent by the patients under the camera. In case of using one tracer, methionine still would be the best choice. Otherwise, the use of 18 F-FDG and SUVр (SUV at the peak of the rst maximum) would provide results likely comparable to methionine T/N index. mainly


Introduction
The attempt to grading brain gliomas by positron emission tomography (PET) started by the early 1980s. Since that time, the most commonly used radiopharmaceuticals were 18 FDG (18F-uorodeoxyglucose, a glucose analog), and 11 C-MET (L-methyl-L-methionine, an essential amino acid). The uptake of the former radiopharmaceutical is linked with the tissue demand for glucose [1]. The latter one has several metabolic pathways [2], but its uptake would be mainly related to membrane synthesis and cells growth and proliferation. Comparing the accuracy of both tracers for predicting the histological grade and aggressiveness of the different types of brain gliomas is still rather controversial [3], despite the great number of publications. Most publications relied upon data collected retrospectively. The standardized uptake value (SUV) for FDG, and T/N ratio (tumor to normal tissue ratio) for methionine, were considered as routine clinical quanti cation methods. The result of each PET investigation usually consisted in one value, measuring the nal averaged uptake of the tracer in the region of interest (ROI), over a time period starting several minutes post injection. The aim of our study was to investigate the pharmacokinetics of the rst minute of the radiopharmaceutical passing through both the tumor and the healthy tissue, and to compare the results with the histological relevant parameters for grading brain gliomas.

Materials And Methods
This prospective study included patients in whom an MRI examination (T1, T1 with contrast enhancement, T2, and T2-FLAIR) led to suspect the presence of a malignant neoplasm in the brain, before any treatment. Later, the diagnosis was con rmed by stereotaxic biopsy or tumor tissue obtained during open surgery. A total of 39 patients were recruited (20 men, 19 women, age 49 ± 13 years). The study was approved by the institutional review board. All patients provided written informed consent and had agreed on the use of their (anonymized) data for scienti c purposes The histological characterization and grading of the tumors are following: glioblastoma n = 14, anaplastic oligodendroglioma n = 5, anaplastic astrocytoma n = 8, oligodendroglioma n = 5, diffuse astrocytoma n = 7. After MRI, each patient underwent two PET examinations (Siemens Truepoint, Siemens Medical Solutions. Knoxville, USA), with 18 F-FDG and 11 C-MET, lasting 40 and 20 minutes, respectively. Data acquisition started simultaneously with the intravenous administration of the radiotracer. 34 frames were reconstructed for 18 F-FDG and 26 frames for 11 C-methionine, with a similar duration of 6 * 10 sec, 6 * 20 sec, 6 * 30 sec, 4 * 60 sec and then 12 * 150 sec for 18 F-FDG, and 4 * 150 sec for 11 C-MET. The OSEM 3D algorithm, with 5 iterations and 8 subsets was used for image reconstruction. No arterial blood sampling was performed.
The images were processed using PMOD software (PMOD v.4.0, Zurich, Switzerland). Dynamic PET series were co-registered with a reference T1 contrast enhanced MR-image. Motion artifacts were compensated. The tumor (T) volume of interest (VOI) was selected on averaged PET images. It included 1 cubic cm of the tissue with the highest tracer uptake. VOI had an arbitrary shape, excluding any internal cavity. The reference (N) volume of interest, a sphere with a diameter of 16 mm, was placed in the healthy part of the brain in the hemisphere contralateral to the tumor, in the frontal cortex if possible. Generated time activity curves (TAC) provided the following parameters: SUV (standardized uptake value) and T/N ratio. Additional parameters have been introduced, which characterize the early tracer uptake and its passage through the tumor during the rst minute (see Fig. 1): T/Nb (b for bolus) -peak of the rst pass (PFP) -is the ratio at the moment while tracer is entering the brain, 23 ± 7 sec after the tracer injection for 18 F-FDG, and 24 ± 7 sec for 11 C-MET (corresponding to point 1, Fig. 1). T/Np -peak of the rst maximum (PFM), is the ratio at the moment of local maximum accumulation, that occurs within the rst minute, 38 ± 18 sec for 18 F-FDG and 37 ± 24 sec for 11 C-methionine (point 2, Fig. 1). SUVp is the standardized uptake value in the tumor at the PFM, P/(T/N) is the ratio of the T/N at the rst maximum divided by the ratio at the end of the acquisition, grad is the slope of the 18 F-FDG T/N curve (min − 1 ).
Comparisons between patient groups were performed using Student's t test and Statistica software (Statsoft,f USA).

Results
The observed uptake patterns within the outlined tumor VOI (Fig. 2) shows obvious differences between methionine and glucose. Note that the nal uptake of 18 F-FDG in the tumor does not differ from the uptake in the intact tissue, whereas within the rst minute the tumor is clearly identi ed. T/Np is 1.6-2 times higher than the nal T/N for FDG. For methionine this is not the case as P/(T/N) is 0.8-1.0 for tumors of any degree of malignancy.
The quantitative values of the parameters and statistical differences between the groups of patients with tumors of various histological types are presented in Tables 1 and 2. All histological diagnoses show a speci c pattern, in particular, oligo tumors have higher values for all parameters with both tracers [4], and therefore malignancy groups (II, III and IV) which combine several tumor subtypes do not look optimal. However, for the sake of consistency and for comparison with the results of earlier publications, such results are also presented. These results should be treated with caution as the values of the parameters may vary depending on the proportions of histological types within each group. Table 1 Values (with standard deviations) of parameters calculated from PET time activity curves. GB -glioblastoma, AOD -anaplastic oligodendroglioma, AA -ana astrocytoma, OD -oligodendroglioma, DA -diffuse astrocytoma. II, III, IV -grades of malignancy. SUVt -standardized uptake value averaged over the last 20 m scanning with 18 F-FDG (10 minutes for 11 C-MET), SUVp -tumor SUV at the peak of the rst maximum (PFM), T/N, T/Nb, T/Np -nal uptake index, T/N at the rst pass and at the peak of the rst maximum respectively, P/(T/N) is the ratio of T/N in the PFM to the nal T/N, grad (for FDG) is the increase in T/N per T/N parameter averaged over the rst 60 seconds post injection was also considered. The results were similar but still inferior to SUVp, therefore T/N60 data were not included in the results.
The mean dynamic SUV and T / N curves shown in Figs. 3 and 4 also demonstrate differences between histological tumor subtypes.
From Table 2, it is clear also that the use of 11 C-MET provided signi cant differences between the different histologic subgroups with a higher number of parameters than did the use of 18 F-FDG.

Discussion
In neuro-oncology, a major issue is to differentiate between malignant and fast growing tumors, on the one hand, and less aggressive ones, on the other hand.
The gold standard for such grading remains histological analysis, with its molecular biology and genetics components, the accuracy of which have considerably improved over the last 2 decades. However, morphology analysis cannot be easily repeated along the course of the disease. Sometimes it can be di cult to target the area of tissue sampling [5]. For these reasons, imaging techniques, which can be repeated as often as necessary, and provide a view of the full volume of the lesion, became more and more important for the management of the disease. MRI is the most accessible to qualitatively investigate many parameters of the tumor, through its large panel of sequences, like contrast enhancement, diffusion coe cient, perfusion and spectroscopy. However, MRI does not propose quantitative data analysis, and both the resolution and the sensitivity of spectroscopy is still far from optimal. This is why PET approach, which gives access to metabolic aspects of the tumor, with quantitative or semi-quantitative results, had an increasing role since its introduction in the late 70's. 18 FDG is the most universally used tracer [1], but 11 C-MET has been for long considered as a tracer of choice for the investigation of brain tumors [6][7][8], this is why we have used both for our study.
The usual quanti cation of 18 FDG uptake in clinical practice is referred to SUV and T/N ratio. In the literature, the results of this approach were contradictory [9,10]. 18 F-FDG quanti cation with SUV or T/N has not been very informative. Even in the case of high glucose uptake no plateau is observed (which is not the case with methionine), so the SUV values are highly dependent on the time elapsed after the tracer injection. Moreover, even malignant tumors with necrosis and BBB damage may have no visible accumulation. In our study however, T/N ratio was better than SUV for differentiating grade IV gliomas (GB) from AA and AOD. Glioblastoma demonstrated higher rates of all measured parameters. The other tumor histological subtypes had a T/N < 1. With 11 C-MET, the obtained T/N ratios in our study were in good agreement with earlier publications [11][12][13]. And both the T/N ratio and SUV looked adequate for differentiating grade II or III gliomas from grade IV, but not grade II from grade III. Among earlier publications, only in the subgroup of oligodendrogliomas a highly signi cant difference was shown between grades II and III [11], which does not appear in our shorter series.
The study of the rst minute after injection of 18 FDG showed that the T/Nb and T/Np ratios exceeded the nal T/N ratio by about 1.6-2 times (parameter P/(T/N) in Table 1) for all types of tumors. Thus, it became possible to identify FDG uptake at the rst minute post-injection even if there was no clear over uptake registered 20-40 minutes later (Fig. 2). As the early tracer uptake is partly dependent on tissue perfusion, this observation might suggest that angiogenesis and perfusion are increased early, at least in aggressive tumors, a situation shown by many other investigations, including morphology. However, the relationship between this early uptake of 18 FDG and perfusion might not be linear or even clearly established, however, since earlier investigations of perfusion with PET [14] using 15 O tracers did not indicate any correlation between tissue blood ow and the vascularization of the tumor among different grades. The most successful parameter in our dynamic 18 FDG approach was SUVp, the SUV value at the PFM. According to this parameter, groups with a high degree of malignancy (groups III + IV) were clearly separated from benign tumors (groups II, p < 0.01, see Table 2). Therefore, the analysis of the rst minute of dynamic PET acquisition with 18 FDG was more informative than the analysis of the averaged image at 20 or 40 min post injection, as it is commonly used in clinical practice. It should be noted that most glioblastomas accumulate 18 FDG faster than healthy tissue (as it was shown by the positive values of the accumulation gradient), while in other histological types the healthy tissue binds the tracer faster than the tumor tissue, which leads to a reduction of T/N over time.
The observation of 11 C-MET TACs showed, similarly, that SUVp and T/Np were the best parameters for differential diagnosis, as effective as the commonly used T/N ratio (Table 2). For instance, anaplastic oligodendrogliomas and glioblastomas had close T/N ratios but they differed signi cantly by T/Nb and T/Np. Diffuse astrocytomas differed from oligodendrogliomas by the T/Np parameter, only with 11 C-MET investigations. The ratio T/Np over nal T/N for methionine was not correlated with the degree of malignancy (lying in the range 0.8-1.0). This result suggests that the effect of BBB disruption on the total accumulation of methionine is insigni cant.
In our series, combination of two dynamic scans, with 18 F-FDG and 11 C-MET, did not signi cantly improve the accuracy of differential diagnosis, as it has already been reported in other publications [15]. Most experts consider that 11 C-MET alone would be the best choice. But, even if 18 F-FDG was inferior to 11 C-MET regarding the number of correlated parameters, most PET centers have no access to the latter tracer. In this case, using the SUVp parameter in dynamic 18 F-FDG acquisition should have a diagnostic value close to the T/N parameter obtained with a static 11 C-MET acquisition.

Conclusions
The study of the rst minute passage of 18 F-FDG and/or 11 C-MET through the tumor and healthy tissues in brain gliomas should not only allow to improve the identi cation of the different glioma grades, but also to shorten the time spent by the patients under the camera. The most informative parameters are SUVр (standardized uptake value at the peak of the rst maximum) with 18 F-FDG, and T/Np (T/N at the peak of the rst maximum) with 11 C-MET. In case of using one tracer, methionine still would be the best choice. Otherwise, the use of 18 F-FDG and SUVр would provide results likely comparable to methionine T/N index.

Declarations
Ethics approval and consent to participate The study was approved by the institutional review board. All patients provided written informed consent and had agreed on the use of their (anonymized) data for scienti c purposes.
Consent for publication The authors declare that they give consent for publication.

Availability of data and material
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests Funding This work is supported by the Russian Science Foundation under grant № 18-15-00337  T/N for FDG (A, B -rst minute) and methionine (C, D -rst minute) for tumors of various histological types. Curves represent average over groups.