Performance of ¹⁸F-FET versus ¹⁸F-FDG-PET for the diagnosis and grading of brain tumors: systematic review and meta-analysis

Abstract

Background

For the past decade ¹⁸F-fluoro-ethyl-l-tyrosine (FET) and ¹⁸F-fluoro-deoxy-glucose (FDG) positron emission tomography (PET) have been used for the assessment of patients with brain tumor. However, direct comparison studies reported only limited numbers of patients. Our purpose was to compare the diagnostic performance of FET and FDG-PET.

Methods

We examined studies published between January 1995 and January 2015 in the PubMed database. To be included the study should: (i) use FET and FDG-PET for the assessment of patients with isolated brain lesion and (ii) use histology as the gold standard. Analysis was performed on a per patient basis. Study quality was assessed with STARD and QUADAS criteria.

Results

Five studies (119 patients) were included. For the diagnosis of brain tumor, FET-PET demonstrated a pooled sensitivity of 0.94 (95% CI: 0.79–0.98) and pooled specificity of 0.88 (95% CI: 0.37–0.99), with an area under the curve of 0.96 (95% CI: 0.94–0.97), a positive likelihood ratio (LR+) of 8.1 (95% CI: 0.8–80.6), and a negative likelihood ratio (LR−) of 0.07 (95% CI: 0.02–0.30), while FDG-PET demonstrated a sensitivity of 0.38 (95% CI: 0.27–0.50) and specificity of 0.86 (95% CI: 0.31–0.99), with an area under the curve of 0.40 (95% CI: 0.36–0.44), an LR+ of 2.7 (95% CI: 0.3–27.8), and an LR– of 0.72 (95% CI: 0.47–1.11). Target-to-background ratios of either FDG or FET, however, allow distinction between low- and high-grade gliomas (P > .11).

Conclusions

For brain tumor diagnosis, FET-PET performed much better than FDG and should be preferred when assessing a new isolated brain tumor. For glioma grading, however, both tracers showed similar performances.

Primary brain tumors have an annual age-adjusted incidence rate of 28 per 100 000 in adults. Gliomas represent 28% of all tumors but 80% of malignant tumors.¹ The World Health Organization (WHO) currently divides gliomas into 4 grades. Grades I and II are considered low-grade tumors that have a prolonged clinical course. Grade III (anaplastic glioma) and grade IV (glioblastoma) tumors are considered high-grade lesions rapidly leading to death when left untreated.² Adequate tumor diagnosis and grading is thus crucial to initiate proper treatment and improve patients' outcomes.

Molecular imaging with positron emission tomography (PET) helps to identify and delineate areas of tumor with increased growth activity.³ PET with ¹⁸F-fluoro-deoxy-glucose (FDG) was first used to detect and distinguish between low- and high-grade tumors.⁴ However, FDG-PET is limited by high uptake in normal brain and unspecific uptake in inflammatory benign lesions.^{5 18}F-fluoro-ethyl-l-tyrosine (FET) is an artificial amino acid that provides well-contrasted images in both high- and low-grade tumors while decreasing effective dose compared with FDG.⁶ FET-PET demonstrated value for guiding biopsy,^7,8 for diagnosing primary brain tumor,^9,10 for directing radiotherapy,¹¹ and for distinguishing between tumor recurrence and radionecrosis after initial therapy.^12,13 Moreover, dynamic FET-PET analysis helps to differentiate low- from high-grade tumors^9,14,15 and to predict patients' outcomes.^16–18

Since FDG-PET is poorly reliable in predicting the neoplastic nature of a lesion due to uptake by inflammatory lesions, amino acid tracers such as FET have been developed in the past decades to increase specificity. However, to date, only a few studies limited to small patient populations directly compared FDG and FET diagnostic value.

The purpose of this report is first to systematically review studies of the literature and perform a meta-analysis on diagnostic performance of FDG and FET-PET in patients with brain tumors, and second to assess whether tracer uptake may allow distinction between nontumor and tumor lesions.

Materials and Methods

Data Sources and Search

As the first reported study about FET synthesis was published in 1999 by Wester et al,¹⁹ we performed a systematic search in the medical database PubMed for English-language publications from January 1995 to January 2015 using the following search: “(“O-(2-fluoroethyl)tyrosine” [all fields] OR “(18F)fluoroethyltyrosine” [all fields] OR “Fluorodeoxyglucose F18” [Mesh]) AND (“PET” [all fields]) AND (“Glioma” [Mesh]) AND (“Humans” [Mesh]).” Errata, reviews, preclinical, animal, and nonradiopharmaceutical studies were excluded.

Study Selection

We considered studies using FET and FDG-PET for the assessment of patients with suspected brain tumors. Inclusion criteria were: (i) FET and FDG-PET used in the same patients with a newly diagnosed brain lesion or in patients with suspicion of recurrence of a brain tumor; (ii) patients who underwent or did not undergo radiotherapy, surgery, or chemotherapy before the PET studies; (iii) use of histology as the gold standard to assess diagnostic performance. Excluded were studies in abstract form, case reports, and studies including fewer than 10 patients.

Data Extraction and Quality Assessment

For each selected publication, we extracted the following information: first author, year of publication, study population (number of patients who underwent FET and FDG for the assessment of brain tumor, sex, age, and histology), FET and FDG results (positive or negative, and target-to-background [TBR] ratio when reported). When possible, data were recorded at the patient level. To assess study quality and applicability, we used the checklists of both the Quality Assessment of Diagnostic Accuracy Studies (QUADAS, scale 0–14) and the Standards for Reporting Studies of Diagnostic Accuracy (STARD, scale 0–25).^20,21

Statistical Analysis

All analyses were performed at the patient level with Stata 13.1 software. P < .05 was considered statistically significant. Continuous variables are presented as mean ± SD or median (interquartile range). Dichotomized histological diagnosis (tumor or not, glioma or not) according to the classification of tumors of the central nervous system of WHO² and the third edition of the International Classification of Diseases for Oncology (ICD-O-3) was used as the gold standard. Gliomas were defined by ICD-O-3 codes 9380–9384, 9391–9460, and 9480. Each study had its own criteria for defining FET and FDG-PET positivity. The bivariate mixed-effects regression model was applied for data synthesis. Average sensitivity, specificity, positive and negative likelihood ratio (LR), diagnostic odds ratio (OR), and the respective 95% CIs were calculated from the maximum likelihood estimates and graphically assessed by summarized receiver operating characteristic (ROC) curves. Forest plots, χ² test, and Cochran Q were used to graphically and statistically assess heterogeneity of the results between studies. To statistically quantify inconsistency of the results between the studies we used the I² statistic, which describes the percentage of total variation across studies attributable to heterogeneity rather than chance. The funnel plot asymmetry test was used to assess publication bias. Finally, after pooling all the patients, a ROC curve comparison between FDG and FET-PET performance for the diagnostic of either brain tumor versus nontumor lesions and brain glioma vs nonglioma lesions was performed. By convention, the small letter n and the capital letter N were used in the figures and text when describing the number of studies (n) and the number of patients (N).

Secondary analyses were performed at the patient level to compare quantitative FDG and FET uptake values. Patients were classified into 3 groups according to histological diagnosis (nonglioma tumor, glioma, or nontumor lesion). We then compared, among the groups, mean TBR (mean activity of the lesion divided by mean activity of the contralateral brain) or maximum TBR (maximum activity of the lesion divided by mean activity of the contralateral brain) measured on FDG and FET-PET images by a Kruskal–Wallis test. We also compared mean TBR and maximum TBR values in glioma according to WHO grade to assess the ability of FDG and FET-PET to distinguish between low- and high-grade gliomas.

Results

Study Characteristics

In total, 253 papers were identified in the PubMed database. After exclusion of review articles (n [studies] = 16), case reports (n = 31), preclinical and animal studies (n = 25), and errata and comments (n = 5), 176 studies about the use of PET in humans with brain tumors were found. After applying the inclusion criteria, 3 studies remained, excluding reports using FDG-PET alone (n = 56), FET-PET alone (n = 45), or other tracers alone or in combination with FDG-PET (n = 72). Two additional studies were found through reference screening of the papers (Fig. 1).

Fig. 1.

Flowchart of study selection.

Overall, 5 studies including 190 patients (Table 1) respected the inclusion criteria and were included.^22–26 In one study,²⁶ all patients did not have both FDG and FET-PET for evaluation, and only patients who underwent both imaging modalities (n = 23) were thus included in the analysis. In one study,²⁴ the histological diagnosis could not be established in 3 patients, and therefore only the remaining 18 patients were included in the analysis. Finally, in the study by Floeth et al,²² we included 11 of 14 reported patients who had both FDG and FET-PET examinations. Thus 119 patients remained (median age: 45 y [37–57], mean age: 46 ± 14 y, sex ratio: 2.2 M:F). Of these patients, 90 had a brain tumor, of whom 43 had a low-grade glioma and 39 a high-grade glioma. Low-grade gliomas included pilocytic astrocytoma (n = 1), ganglioglioma (n = 1), astrocytomas (n = 20), oligoastrocytomas (n = 7), oligodendrogliomas (n = 10), and 4 unspecified low-grade gliomas. High-grade gliomas included anaplastic astrocytomas (n = 14), anaplastic oligoastrocytomas (n = 5), anaplastic oligodendroglioma (n = 1), and glioblastomas (n = 19). Eight patients had a nonglioma brain tumor: metastasis (n = 3), lymphoma (n = 2), invasive adenoma (n = 1), ganglioneuroblastoma (n = 1), and meningioma (n = 1). Twenty-nine patients had nontumoral lesions, including 9 abscesses or empyemas, 4 hemorrhages, 2 encephalitis, 1 cortical dysplasia, and 13 unspecified lesions.

Table 1.

Study characteristics

Study#	Reference	Year	# Patients	# Patients Excluded	Age, y	Sex ratio, F:M	# Patients With Tumors	# Patients With Gliomas	# High-grade Gliomas
1*	Floeth et al	2006	14	3 (21)	54 ± 12	0:11	4 (29)	4 (29)	4 (29)
2*	Pauleit et al	2009	52	0 (0)	46 ± 14	16:36	45 (87)	43 (83)	21 (40)
3*	Lau et al	2010	21	3 (14)	42 ± 16	5:13	12 (57)	10 (48)	3 (14)
4	Plotkin et al	2010	15	0 (0)	44 ± 11	9:6	15 (100)	15 (100)	6 (40)
5*	Pichler et al	2010	88	65 (74)	–	–	14 (16)	10 (11)	5 (6)
Meta-analysis subtotal			175	71 (41)	46 ± 14	21:60	75 (43)	67 (38)	33 (19)
Overall total			190	71 (37)	46 ± 14	30:66	90 (47)	82 (43)	39 (21)

*Studies included in the meta-analysis. Data in parentheses are percents and age is mean ± SD.

Performances of FDG and FET-PET

From the 5 selected studies, 4 with a total of 104 patients were used in the bivariate mixed-effects regression model. The fifth one²⁵ could not be included because it did not report any true-negative or false-positive case to compute specificity. However, the pooled results of the 5 studies (n = 119 patients) were used to compare the area under the curve (AUC) of FDG and FET-PET. Criteria for FET and FDG-PET positivity varied among studies. Positivity definition was based on qualitative visual analysis compared with nontumor brain background in 4 studies^23–26 or on quantitative assessment of TBR using defined threshold in 1 study.²²

Including 4 of the 5 selected studies, FDG-PET demonstrated an overall sensitivity of 0.38 (95% CI: 0.27–0.50) and specificity of 0.86 (95% CI: 0.31–0.99), with an AUC of 0.40 (95% CI: 0.36–0.44), positive LR of 2.7 (95% CI: 0.3–27.8), negative LR of 0.72 (95% CI: 0.47–1.11), and diagnostic OR of 4 (95% CI: 0–58) for the diagnosis of brain tumor versus nontumor lesions. FET-PET demonstrated a sensitivity of 0.94 (95% CI: 0.79–0.98) and specificity of 0.88 (95% CI: 0.37–0.99), with an AUC of 0.96 (95% CI: 0.94–0.97), positive LR of 8.1 (95% CI: 0.8–80.6), negative LR of 0.07 (95% CI: 0.02–0.30), and diagnostic OR of 113 (95% CI: 4–2975).

For the diagnosis of glioma versus nonglioma lesions, FDG-PET demonstrated an overall sensitivity of 0.35 (95% CI: 0.11–0.71) and specificity of 0.65 (95% CI: 0.48–0.79), with an AUC of 0.60 (95% CI: 0.56–0.65), positive LR of 1.0 (95% CI: 0.4–2.7), negative LR of 1.0 (95% CI: 0.58–1.73), and diagnostic OR of 1.0 (95% CI: 0–5), while FET-PET demonstrated an overall sensitivity of 0.92 (95% CI: 0.75–0.98) and specificity of 0.62 (95% CI: 0.43–0.79), with an AUC of 0.89 (95% CI: 0.86–0.91), positive LR of 2.4 (95% CI: 1.4–4.1), negative LR of 0.13 (95% CI: 0.04–0.48), and diagnostic OR of 18 (95% CI: 4–92).

By pooling patients' results of the 5 selected studies (n = 119), the FET-PET AUC (0.85 [95% CI: 0.77–0.93]) was significantly higher than that of FDG-PET (0.56 [95% CI: 0.47–0.66], P < .0001) for the diagnosis of brain tumor (Fig. 2). For the diagnosis of glioma, the FET-PET AUC (0.76 [95% CI: 0.67–0.84]) was also significantly higher than that of FDG-PET (0.49 [95% CI: 0.40–0.58], P < .0001).

Fig. 2.

ROC curves for discrimination between brain tumoral and nontumoral lesion for FDG-PET and FET-PET (n = 119 patients). Dashed line indicates FDG-PET; solid line indicates FET-PET; fine dashed line indicates chance.

Assessment of Heterogeneity, Inconsistency, and Quality Studies

For the differentiation between brain tumoral and nontumoral lesions, a forest plot did not show any significant performance heterogeneity (Cochran Q = 3.4, P = .092) but mild inconsistency between studies (I² 41% attributable to heterogeneity rather than chance) for FDG-PET. There was neither performance heterogeneity (Cochran Q = 1.3, P = .27) nor inconsistency (I² 0%) between studies for FET-PET. For the diagnosis of brain glioma vs nonglioma lesions, a forest plot showed major inconsistency between studies for FDG-PET (I² 100%) but not for FET-PET (I² 0%). This was mainly due to heterogeneity and inconsistency of sensitivity (Cochran Q = 9.10, P = .03 and I² 67%) due to the high sensitivity value of FDG-PET in the study by Floeth et al²² that includes only high-grade gliomas with no false-negative case (Fig. 3). Funnel plots did not demonstrate publication bias for FDG (P > .051) or FET (P > .18) PET analysis. QUADAS and STARD scores for the assessment of study quality are reported in Fig. 4.

Fig. 3.

Forest plot of studies included in the meta-analysis for discrimination between glioma vs nonglioma lesions with FDG-PET.

Fig. 4.

Study quality grading using QUADAS scores (range 0–14) and STARD scores (range 0–25). *Studies included in the meta-analysis. Dashed line indicates maximal score for QUADAS.

Quantitative Analysis

Among the 5 studies selected, only 2 (n = 63) reported mean and maximum TBR values of the lesions for both FDG and FET-PET. Among these 63 cases, 47 gliomas, 2 nonglioma tumors, and 14 nontumoral lesions were included. Of the 47 gliomas, 22 were low-grade and 25 were high-grade lesions. Neither mean TBR (1.3 ± 0.5 vs 1.1 ± 0.5, P = .14) nor maximum TBR (2.0 ± 1.0 vs 1.8 ± 0.9, P = .32) on FDG-PET were significantly different between tumoral and nontumoral lesions. On FET-PET images, both mean TBR (2.1 ± 0.8 vs 1.4 ± 0.3, P = .0015) and maximum TBR (2.9 ± 1.2 vs 1.9 ± 0.5, P = .0007) were significantly higher in tumoral than in nontumoral lesions.

There was no statistically significant difference of mean TBR (2.1 ± 0.9 vs 2.0 ± 0.1, P = .69) and maximum TBR values (3.0 ± 1.2 vs 2.6 ± 0.1, P = .40) on FET-PET images between glial and nonglial tumors. FDG mean TBR (1.3 ± 0.5 vs 1.7 ± 1.3, P = .88) and maximum TBR values (2.0 ± 0.9 vs 2.5 ± 1.9, P = .88) were also not significantly different between glial and nonglial tumors.

Taking into account all gliomas (n = 47), while mean TBR (2.1 ± 0.9 vs 1.4 ± 0.3, P = .003) and maximum TBR values (3.0 ± 1.2 vs 1.9 ± 0.5, P = .0009) on FET-PET images were significantly higher than in nontumoral lesions, neither mean TBR (1.3 ± 0.5 vs 1.1 ± 0.6, P = .15) nor maximum TBR values (2.0 ± 0.9 vs 1.8 ± 0.9, P = .33) on FDG-PET images were significantly different. However, both mean TBR and maximum TBR on FDG and FET-PET images were significantly higher in high-grade lesion (n = 25) compared with low-grade lesion (n = 22) (Fig. 5). ROC curve analysis showed that a mean TBR of at least 1.4 and a maximum TBR of at least 1.8 had the best value to distinguish between low- and high-grade glioma, with FDG-PET reaching a sensitivity, specificity, accuracy of 0.60, 0.91, 0.74 and 0.72, 0.73, 0.72, respectively. For FET-PET we observed that a mean TBR of at least 2.0 and a maximum TBR of at least 3.0 reached a sensitivity, specificity, accuracy of 0.88, 0.73, 0.81 and 0.80, 0.82, 0.81, respectively. Performances of these thresholds for glioma grading were not different between FDG and FET-PET using mean TBR (P = .22) or maximum TBR (P = .11).

Fig. 5.

TBR comparison according to histologic WHO grading. Light gray and medium light gray indicate mean TBR and maximum TBR from FDG-PET; medium dark and dark gray indicate mean TBR and maximum TBR from FET-PET. *P = .0028 vs WHO grades I–II; **P = .0065 vs WHO grades I–II, ^||P = .0001 vs WHO grades I–II. For comparison between nontumoral lesions and WHO grades I–II gliomas, all P-values >.44.

Discussion

The main results of this meta-analysis may be summarized as follows: (i) FET-PET demonstrated significantly higher diagnostic performance for the diagnosis of brain tumor (AUC = 0.96 vs 0.40, P < .0001) and glioma (AUC = 0.89 vs 0.60, P < .0001) compared with FDG-PET; (ii) mean and maximum TBR values on FET-PET can distinguish between tumoral and nontumoral lesions in the brain, while mean and maximum TBR values on FDG-PET cannot; and (iii) both FDG and FET quantitative parameters allow distinction between low- and high-grade gliomas.

Due to the known lack of specificity of conventional MRI to noninvasively characterize brain lesions, metabolic imaging using PET tracers has been increasingly studied. FDG-PET being limited by high uptake in normal brain and unspecific uptake in inflammatory benign lesions, radiolabeled amino acid tracers such as ¹¹C-methionine (MET) and ¹⁸F-fluoro-ethyl-tyrosine have been developed to overcome these limitations. FET-PET has demonstrated its value for the diagnosis^9,10 and grading^9,14,15 of newly identified brain tumor, for the diagnosis²⁷ and grading²⁸ of tumor recurrence, for the differentiation between brain tumor recurrence and radiation necrosis,^12,13 and for the assessment of treatment response²⁹ with lower radiation burden than FDG-PET.⁶ However, only a few studies with small patient populations report direct comparison of FET and FDG-PET for the qualitative and quantitative characterization of brain lesions in humans. In the presented meta-analysis, we demonstrated the strong advantage of FET-PET over FDG-PET for the diagnosis of brain tumors (AUC = 0.96 vs 0.40, P < .0001) and gliomas (AUC = 0.89 vs 0.60, P < .0001). This is in line with a recent meta-analysis reporting the good performance of FET-PET with an AUC of 0.84 (95% CI: 0.80–0.87) for the initial assessment of patients with new isolated brain lesions.⁹ Regarding clinical applications, due to positive and negative LRs of 2.7 (95% CI: 0.3–27.8) and 0.72 (95% CI: 0.47–1.11), respectively, FDG-PET qualitative analysis has very small informational value for the differentiation of brain tumors versus nontumoral lesions. In contrast, FET-PET positive and negative LRs (8.1 [95% CI: 0.8–80.6] and 0.07 [95% CI: 0.02–0.30], respectively) indicate that FET-PET may help to exclude and to confirm the diagnosis of brain tumor. The higher accuracy for brain tumor diagnosis was also demonstrated with other radiolabeled amino acid tracers compared with FDG-PET,^30–33 especially in a recent meta-analysis by Zhao et al,³³ who argued for the excellent diagnostic performance of MET while conceding the major inconvenience of tracer supply.

Regarding quantitative analysis, only mean and maximum TBR values on FET-PET images had the ability to distinguish between tumoral and nontumoral brain lesions, mainly due to high FDG uptake in inflammatory lesions such as abscess, as previously demonstrated.⁵ Based on the small number of cases where uptake quantification of the 2 tracers was performed, respective values for the differentiation of nonglioma versus glioma tumors could not reliably be assessed in our study. However, both tracers were able to distinguish between low-grade and high-grade gliomas, which is consistent with previously published studies on FET-PET^{9,14,15,28,34} and FDG-PET.^4,35–38 Though mean and maximum TBR cutoff values were different between FDG and FET-PET, performances were similar with both tracers (P > .11) and close to those reported in the literature.^4,14,15 Similar performance for distinguishing low- and high-grade gliomas has also been reported for FDG-PET and MET-PET.^35,37 Among current amino acid tracers, however, the performance of FET-PET for glioma grading seems to be better than that of ¹⁸F-fluoro-dihydroxy-phenalalanine (FDOPA)³⁹ and MET,⁴⁰ the use of time-activity curve parameters from dynamic FET-PET acquisition^14,28,34,40 even improving tumor characterization. It is, however, important to take into account that glioma is a heterogeneous histological family. An oligodendroglial component may have a singular behavior both on FET-PET⁴¹ and FDG-PET³⁵ that may impair diagnostic accuracy for both examination types. In a recent study, Manabe et al³⁵ thus concluded that the results of PET imaging should be revised after obtaining an histology report to better classify patient recurrence risk.

Substantial data in the literature also demonstrated the value of FET-PET for guiding and evaluating response to therapy and for the prediction of patient outcome. FET-PET may help to delineate tumoral volume before radiotherapy,^11,42 to monitor the effects of radiotherapy^43,44 and chemotherapy.^45,46 The prognostic value of FET-PET has also been demonstrated for the assessment of low-grade and high-grade gliomas. Floeth et al¹⁷ first found that low-grade gliomas exhibiting a diffuse tumoral pattern with positive uptake on baseline FET-PET have a significantly lower progression-free survival. Two recent studies reported that dynamic FET-PET analysis could also help in identifying low-grade gliomas at high risk of progression.^47,48 FET-PET is also useful to evaluate patient prognosis in the preoperative, postoperative, and pre-radiative phases of high-grade glioma management.^{16,29,49–53} Untreated gliomas with high TBR on baseline static FET-PET images have a lower overall survival,⁴⁹ while grade III astrocytoma tumors with an early minimal time-to-peak on dynamic FET-PET images exhibit similar survival to glioblastoma.⁵⁰ Higher postoperative residual tumor volume on FET-PET and decreasing time-activity curve^51,52 as well as decreasing time-activity curve prior to re-irradiation of recurrent glioblastoma⁵³ were also related to impaired patient survival. In contrast, early TBR decrease on serial static FET-PET examinations^16,29,54 but not dynamic FET-PET parameter changes⁵⁴ after radiochemotherapy in glioblastoma was associated with a better patient survival. Though the prognostic value of FDG-PET has also been reported in newly diagnosed and recurrent gliomas prior to therapy^55–57 and for response assessment,⁵⁸ it seems to be lower than for amino acid tracer PET.^37,59

Regarding the development of hybrid PET/MR imaging, it is furthermore worthy to mention that the respective value of combining FDG and FET-PET with MRI techniques cannot be deduced from this meta-analysis. FET-PET increases MRI accuracy^7,8 to guide biopsies and notably helps in determining the outcome of patients with low-grade glioma.¹⁷ However, only a few studies report the combination of multiparametric MRI with quantitative analysis of FDG³⁸ or FET-PET.^41,60,61 Yoon et al³⁸ concluded that in case of concordant results of multiparametric MR techniques for high-grade lesions, the additive value of FDG-PET may be limited. In contrast, the combination of dynamic FET-PET with diffusion MRI improves glioma grading⁴¹ and improves presurgical biopsy guidance⁶¹ compared with a single modality approach. Furthermore, spatial congruence of an increased FET or FDOPA uptake area and abnormal area on enhanced MRI⁶² or perfusion-weighted MRI^60,63 are different, highlighting that practical guidelines for interpreting multimodal imaging have to be developed to ensure accurate glioma classification. The diagnostic superiority of combined FET-PET/MRI over FDG-PET/MRI in a same patient population also remains to be demonstrated. Finally, although Heinzel et al⁶⁴ demonstrated that the combined use of FET-PET and conventional MRI was cost-effective in the planning of biopsies of glioma, the cost-effectiveness of multiparametric MRI associated or not with FDG or FET-PET remains to be determined.

Our systematic review of the literature found only 5 studies that directly compare FDG and FET-PET for assessing patients with suspected brain tumor. While all achieved a good quality (QUADAS scores >10 and STARD scores >18), the small number of studies resulted in substantial inconsistency between study results for FDG-PET but not for FET-PET. No publication bias was observed for both tracers. There were, however, some limitations. First, only 4 studies were included in the meta-analysis because of the absence of true-negative and false-positive cases in one study. Second, due to the small number of pooled patients, a definitive conclusion about the value of FDG and FET TBR to differentiate glioma (n = 47) versus nonglioma tumors (n = 2) cannot be reliably made. Though we did not observe patients’ characteristics overlap, the 2 studies that gave TBR values both on FDG and FET-PET came from the same institution, emphasizing the need for multicenter prospective studies to overcome limitations of single-center multiple retrospective reports. Multicenter prospective studies could also assess the comparative value of parameters extracted from dynamic PET acquisition (ie, time-activity curve for FET or cerebral metabolic rate of glucose for FDG) and from multiparametric MRI for the diagnostic and prognostic assessment of patients with brain tumors, which could not be performed hereby.

On the basis of our systematic review and meta-analysis we can recommend that FET-PET should be preferred to FDG-PET for the diagnosis of brain tumor and glioma. Moreover, FET and FDG TBR may be used indifferently to distinguish between low- and high-grade gliomas. Multicentric multitracer studies should be developed to assess the respective values of dynamic PET parameters notably to distinguish between glioma and nonglioma tumors. Regarding the emergence of hybrid PET/MR imaging, development of integrated interpretation guidelines and evaluation of diagnostic performance and cost-effectiveness of multiparametric MRI in comparison or in combination with PET is also mandatory in order to avoid wasting time and funds.

Conclusion

This systematic review and meta-analysis indicates that FET-PET has significantly higher diagnostic performance for the diagnosis of brain tumor and glioma than FDG-PET. Although both FDG and FET quantitative parameters allow distinction between low- and high-grade tumors, only TBR values on FET-PET can distinguish between tumoral and nontumoral lesions, confirming FET-PET superiority over FDG-PET for brain lesion characterization. Additive value and cost-effectiveness of the use of FDG and FET-PET in combination with multiparametric MRI in the same population have to be assessed considering the development of hybrid PET/MR imaging and should provide new insights to reduce diagnostic time and cost.

Funding

The authors declare they did not receive any funding for this study.

Conflict of interest statement. The authors do not declare any potential conflict of interest relevant to this article.