Introduction

Glioma is the most common primary brain tumours, and the prognosis of patients with high-grade glioma is very poor [1]. Despite first-line treatments, including maximal safe surgical resection, irradiation, and adjuvant chemotherapy, the median survival time for most malignant glioblastoma patients is still only 12–15 months [2]. Recurrence is a critical clinical problem in glioma patients, but the diagnosis of recurrent glioma is difficult using computed tomography (CT) or magnetic resonance (MR) imaging because radiotherapy and chemotherapy are associated with a high incidence of treatment-related imaging changes termed treatment effects (TrE), which is occasionally referred to as pseudoprogression and radiation necrosis. This phenomenon is believed to be related to the destruction of the blood–brain barrier after radiation therapy and chemotherapy, leading to increased leakiness of contrast agent exhibiting enhancement in the area on MR imaging, which is similar to the imaging findings of tumour recurrence (TuR) [3]. Although advanced multimodal MR imaging plays an important role in the post-treatment follow-up of glioma patient care in recent years, the differential diagnosis in clinical image evaluation between TrE and TuR in glioma patients remains challenging because they have overlapping clinical and radiologic features [4,5,6]. The diagnosis of TuR requires a change in the therapeutic approach to a more active anti-tumour strategy, whereas a diagnosis of TrE supports the effectiveness of current treatment. Therefore, accurate and reliable imaging evaluation is very important for making clinical decisions.

Positron emission tomography/computed tomography (PET/CT) is a molecular imaging technique allowing in vivo quantitative measurement of biological processes noninvasively, which has become an integral supplemental imaging tool for differential diagnosis of brain lesions beyond MR [7]. Some PET tracers, such as 18F-fluorodeoxyglucose (18F-FDG), 13N-ammonia (13N-NH3), 11C-methylmethionine (11C-MET), 18F-fluoroethyl-L-tyrosine (18F-FET), and 18F-fluoro-L-dihydroxy-phenylalanine (18F-FDOPA), have been used for imaging gliomas [7,8,9,10]. The Response Assessment in Neuro-Oncology working group and European Association for Neuro-Oncology have also recommended the clinical use of PET/CT imaging in gliomas, and they emphasize that PET/CT exhibits increased diagnostic accuracy than MR when differentiating TuR from TrE [11]. However, no standard imaging technology is currently available to achieve a differential diagnosis, and few articles compare different metabolic types of PET/CT tracers used in this critical issue of the glioma follow-up process. This study aims to explore the clinical potential of 13N-NH3, 18F-FDOPA, and 18F-FDG PET/CT, which represent three different types of tracers in the differential diagnosis of TuR and TrE in glioma patients, to reduce clinical misdiagnosis.

Materials and methods

Patients

Patients treated at our hospital between September 2016 and December 2019 with suspicion of recurrent gliomas based on MR imaging demonstrating new enhancement lesions and the desire for PET/CT examination were included in this study. The initial diagnosis for these patients with histopathologic glioma according to revised 2016 World Health Organization (WHO) criteria, and TuR or TrE diagnosis was determined by histology if the patient received repeated surgical resection or standard clinical MR follow-up for at least 6 months at 2- to 3-month intervals after the PET/CT exam. This study was approved by the ethics committee of our cancer centre, and informed consent was obtained from all individual participants included in the study.

Forty-three patients (16 females and 27 males) were enrolled in the study. The average age was 41.74 ± 12.71 years (range, 14–65 years). Thirty-four patients received all three tracer examinations. Three patients only underwent 13N-NH3 and 18F-FDG PET/CT, and six patients only underwent 18F-FDOPA PET/CT due to the patient's desire or shortage of tracers. Detailed patient characteristics of each tracer are listed in Table 1. Except for one patient’s lesion located in the left cerebellum, the other patients’ lesions were mainly located in the temporal lobe, frontal lobe or parietal lobe, including 23 cases on the right side and 19 cases on the left side. All patients received radiotherapy before, and the mean dose was 59.23 ± 1.84 Gy (range, 54–66 Gy). Only one patient did not receive adjuvant chemotherapy and the others received temozolomide-based adjuvant chemotherapy. The median interval time from primary diagnosis to PET/CT examination was 19.10 months (range, 4.10–88.20 months). The median interval between radiotherapy and PET/CT was 15.90 months (range, 1.67–84.77 months).

Table 1 Patient characteristics of each PET/CT tracer
Full size table

PET/CT imaging protocol

Tracers were synthesized through a commercially available system for isotope generation (Cyclone-10, Ion Beam Applications S.A., Belgium; AllinOne synthesizer, Trasis, Belgium). 13N-NH3 and 18F-FDG were produced as previously mentioned [7]. 18F-FDOPA was produced using the method of Libert et al. [12]. The radio-chemical purity of 13N-NH3 was > 99%, 18F-FDOPA was > 98%, and 18F-FDG was > 95%. PET/CT examinations were performed on a Gemini GXL 16 scanner (Philips, Netherlands) in 3-dimensional acquisition mode. The reconstruction protocols of PET/CT have been published in detail previously [7]. Briefly, patients underwent a fast for at least 4 h before the 18F-FDOPA examination. Ten minutes after the intravenous injection of 2 MBq/kg of 18F-FDOPA, a dedicated CT scan of the brain (120 kV, 80 mAs, 3 mm slice collimation) was performed followed by a brain-centred static 3D PET acquisition of 10 min. Then, a clear PET image was obtained. 18F-FDG PET/CT was performed after 13N-NH3 with a minimum interval of 2 h on another day within 1 week, and patients remained untreated until the PET/CT study was completed. All patients had fasted for at least 6 h before 13N-NH3 and 18F-FDG examination. First, after intravenous injection of 7.4 MBq (0.20 mCi)/kg of 13N-NH3, patients rested in a quiet room, and PET/CT was performed 10 min later. Then, a dose of 5.18 MBq (0.14 mCi)/kg 18F-FDG was injected intravenously, and serial scanning was performed approximately 30–45 min after the injection with the patient resting with their eyes closed. Finally, PET/CT images were reconstructed using the LOR-RAMLA algorithm with low-dose CT images for attenuation correction.

PET image parameters

PET/CT images were evaluated by two experienced nuclear medicine physicians independently who were blinded to the final clinical diagnosis of the lesions. The tracers’ uptake of the lesion was evaluated by quantitative analysis. A region of interest (ROI) was placed over the entire lesion on the transverse PET/CT image by the nuclear physicians. For lesions with reduced or equal uptake, the ROI was drawn based on the anatomical information on the brain lesions presented by previous MR. The grey-matter ROI was drawn in the lobe contralateral to the lesion according to the method we reported before [13, 14]. After reaching the consensus of the target lesions and ROI, the maximum and mean standardized uptake value (SUV) of the lesion was detected as SUVmax and SUVmean. The lesion-to-normal grey-matter cortex uptake (L/G) ratio was calculated as the SUVmax of the lesion divided by the average SUV of the contralateral normal grey-matter (CNGM) cortex.

True diagnosis of lesion

TuR or TrE was determined by histopathology if repeated surgical resection was performed. When the histopathologic examination was not available, clinical follow-up, including longitudinal MR and clinical symptoms, was considered the standard of truth according to the Response Assessment in Neuro-Oncology (RANO) criteria [15] and our previous report [4, 16]. Briefly, the lesion was defined as TuR based on a progressive increase in size and/or the number of enhancing lesions with the development of neurologic symptoms. TrE was defined based on a reduced or stable lesion size on subsequent follow-up images with a stable or improved clinical condition for at least 6 months without additional steroid treatment.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics 23.0 (SPSS Inc. Chicago, IL, USA) software. Continuous variables are expressed as the mean ± standard deviation (SD) or medians with range. Student’s t-test or Chi-square tests were used to compare the differences between the TuR and TrE groups. A paired Student’s t-test was used to compare lesions and CNGM within each PET/CT tracer. The overall survival of glioma patients after PET/CT examination was evaluated using the log rank test with Kaplan–Meier analysis. The receiver operating characteristic (ROC) curve was chosen to assess the performance of PET/CT parameters in differentiating between TuR and TrE. The optimal cutoff values were determined by maximizing the Youden’s index value. P < 0.05 was considered statistically significant.

Results

Tumour recurrence (TuR) and treatment effects (TrE)

Finally, thirty-four patients were diagnosed with TuR, whereas the remaining 9 were diagnosed with TrE. Note that seven patients were diagnosed from histopathology after reoperation, and the remaining patients were diagnosed based on clinical and radiological follow-up for more than 6 months. TrE occurred more frequently in females (7/16, 43.75%) compared with males (2/27, 7.41%) (P = 0.02). Among the 37 patients with IDH1 mutation status results, the probability of IDH1 mutation in the TrE group was 100% (8/8), whereas that in the TuR group was 55.17% (16/29) (P = 0.03). The two groups exhibited no significant differences in terms of patients’ age (P = 0.54), WHO grade (P = 0.09), radiotherapy dose (P = 0.37) or the time interval between radiotherapy and diagnosis surgery (P = 0.36 and 0.39) (Table 2). The median follow-up time from PET/CT performed was 10.23 months (95% CI, 9.26–14.63 months). Kaplan–Meier analysis showed that patients diagnosed with TrE exhibited improved survival compared with those diagnosed with TuR (log-rank test, x2 = 5.524, P = 0.019) (Fig. 1).

Table 2 Patient characteristics of TuR and TrE
Full size table
Fig. 1
figure 1

The overall survival curves of glioma patients after PET/CT examination according to final diagnosis with TrE and TuR

Full size image

PET/CT findings

The SUVmax and SUVmean of lesions were significantly increased compared with the SUVmean of contralateral normal grey matter (CNGM) in 13N-NH3 and 18F-FDOPA PET (13N-NH3 0.696 ± 0.558, 0.625 ± 0.507 vs 0.486 ± 0.413; 18F-FDOPA 0.455 ± 0.518, 0.415 ± 0.477 vs 0.194 ± 0.203; both P < 0.01), but there was no significant difference in 18F-FDG (6.918 ± 3.190, 6.016 ± 2.807 vs 6.356 ± 3.104, P = 0.290 and 0.493) (Fig. 2a–c).

Fig. 2
figure 2

The SUVmax and SUVmean of the lesions compared with SUVmean of CNGM and with groups of TuR and TrE in different PET tracers (a 18F-FDG; b 13N-NH3; c 18F-FDOPA). The L/G ratios of 18F-FDG, 13N-NH3 and 18F-FDOPA were compared in the TuR and TrE groups (d). *P < 0.05, **P < 0.01, and ***P < 0.001

Full size image

The lesions’ SUVmax and SUVmean of 13N-NH3 were significantly different in patients with TuR compared to patients with TrE (Fig. 2b), whereas TuR exhibited significantly increased L/G ratios than TrE both in 13N-NH3 and 18F-FDOPA PET/CT (13N-NH3, 1.573 ± 0.099 vs 1.025 ± 0.128, P = 0.008; 18F-FDOPA, 2.729 ± 0.131 vs 1.514 ± 0.141, P < 0.001) but not in 18F-FDG (Fig. 2d). Figures 3 and 4 report different PET/CT performances in two typical cases of TuR and TrE.

Fig. 3
figure 3

A representative patient with glioma TuR. Contrast MR image (a) reveals an enhanced lesion with obviously increased uptake of 18F-FDOPA (b), slightly increased uptake of 13N-NH3 (c), and decreased uptake of 18F-FDG (d). Two months later, contrast MR showed that the lesions continued to increase (e), and the patient died 8 months later

Full size image
Fig. 4
figure 4

A representative patient with TrE. Contrast MR image (a) presents an enhanced lesion with slightly increased uptake of 18F-FDOPA (b) and no increased uptake of 13N-NH3 (c) and 18F-FDG (d). Follow-up MR revealed that the enhanced lesions gradually decreased and disappeared (e, f)

Full size image

ROC analysis for differentiation between TuR and TrE yielded an optimal L/G ratio of 2.165 for 18F-FDOPA (sensitivity, 84.6%; specificity, 100%; AUC, 0.938; 95% CI [0.859–1.000];), 1.260 for 18F-FDG (sensitivity, 46.2%; specificity, 100%; AUC, 0.726; 95% CI [0.547–0.905];), and 1.515 for 13N-NH3 (sensitivity, 57.7%; specificity, 100%; AUC, 0.726; 95% CI [0.547–0.905]) (Table 3). The L/G ratio of 18F-FDOPA exhibited the best PET/CT parameters compared to 18F-FDG and 13N-NH3. The area under the ROC curves of the L/G ratio evaluation of 13N-NH3 and 18F-FDOPA was 0.803 and 0.938, respectively (Fig. 5). Combining the L/G ratio evaluation of 18F-FDOPA and 13N-NH3, the diagnostic performance did not show further improvement, yielding a sensitivity of 80.8%, a specificity of 100%, and an AUC of 0.952.

Table 3 ROC curve analyses of the L/G ratio of FDG, NH3, DOPA, and the combination of NH3 and DOPA in differentiating TuR and TrE
Full size table
Fig. 5
figure 5

ROC curve analysis of the L/G ratio to differentiate TuR and TrE. The area under the curve (AUC) was 0.726 for 18F-FDG, 0.803 for 13N-NH3, 0.938 for 18F-FDOPA and 0.952 for the combination of 18F-FDOPA and 13N-NH3

Full size image

Discussion

The prognosis of patients with gliomas is very poor once recurrence occurs after comprehensive treatment. Some studies have demonstrated that the median survival time of glioma patients is only 9–10 months after the first tumour recurrence [17, 18]. Generally, the follow-up of glioma patients after or during comprehensive treatment mainly depends on the MR examination. However, TrE may lead to mimicking of tumour recurrence because they exhibit similar features on contrast-enhanced MR, which are also known as early effects (i.e., pseudoprogression) or late effects, such as radiation necrosis mainly based on timing. Although pseudoprogression and radiation necrosis are thought to represent distinct clinical and pathophysiologic mechanisms, they share many histologic similarities, such as inflammatory infiltrates and necrosis, which translate into similar imaging characteristics [19]. This similarity makes it difficult to identify TuR and TrE with the most commonly used MR images, especially contrast-enhanced MR. Although several advanced MR techniques, such as amide proton transfer, diffusion, perfusion, and spectroscopy imaging, can improve the differential diagnosis and are widely used in clinical applications [4, 20], their accuracy and efficacy remain limited. The final diagnosis requires longitudinal MR observation for several months or repeat surgery. This process is resource-intensive as well as burdensome for the patients, and the longitudinal MR observation also delays definitive treatment. Then, experienced doctors will use PET/CT, which can reflect the metabolism of lesions, to assist MR for optimal diagnosis.

In this study, we investigated the diagnostic performance of multitracer PET/CT techniques, including 13N-NH3, 18F-FDOPA, and 18F-FDG, in distinguishing TuR from TrE in a cohort of patients with suspected recurrent gliomas. 18F-FDG is the most widely used tracer in PET/CT imaging and reflects the glucose metabolism level of tumours, especially malignant tumours. However, 18F-FDG may have some limitations in differentiating intracranial tumours with low metabolic levels from normal brain tissues, inflammation, and benign tumours because it accumulates largely in normal brain tissue, leading to small differences between glioma lesions and normal brain tissue [21]. Therefore, 18F-FDG is not very good for glioma display, especially for low-grade glioma and lesions close to grey matter. This limitation was also confirmed in this study, in which 18F-FDG metabolism was not significantly different between lesions and contralateral grey cortical tissue. The low L/G ratio also limits its role in the differential diagnosis of TuR and TrE, and no significant difference was found in this study.

13N-NH3 is another PET imaging agent that is low fat-soluble and has a small diameter, allowing it to penetrate the blood–brain barrier (BBB). We have previously reported that the uptake of 13N-NH3 is superior to 18F-FDG not only in enabling differentiation between glioma and non-neoplastic lesions but also in separating low-grade gliomas (LGG) from high-grade gliomas [13, 14]. This difference may be due to the uptake of 13N-NH3 in normal brain tissue being relatively lower than 18F-FDG, whereas the uptake of 13N-NH3 in glioma is significantly increased. Therefore, the L/G ratio of 13N-NH3 PET/CT imaging is higher in tumourous lesions, making it more beneficial to distinguish glioma from some inflammatory or benign tumours. Furthermore, our other study [9] also found that the increased 13N-NH3 uptake in recurrent glioma and the absent or lower uptake in radiation necrosis due to perfusion and glutamine synthetase activity in the recurrent tumour is higher than that in the TrE. In this study, we also found that 13N-NH3 is a promising tracer for separating TuR from TrE. In addition, SUVmax, SUVmean, and L/G ratio all showed good performance for this purpose, and their role seemed superior to 18F-FDG.

Because gliomas have upregulated amino acid transporters and increased amino acid metabolism, and labelled amino acid tracers, including 11C-MET, 18F-FET, and 18F-FDOPA, are increasingly being widely used in gliomas in recent years with better tumour-to-background contrast for differentiation glioma grade, biopsy guiding, radiotherapy planning, therapy monitoring, and differentiation between TrE and residual or recurrent glioma [21,22,23]. Amino acid PET/CT typically demonstrates high uptake in glioma and low uptake in the normal brain; thus, the L/G ratio exhibits advantages over SUVmax and SUVmean of the lesion. These amino acid tracers can be used in PET/CT for suspicion of recurrent glioma based on pathophysiological differences between the actively growing tumour, which exhibits increased transport and metabolism of the amino acid; conversely, treatment-induced brain changes result in a low level of metabolism in lesions. For example, Martinez-Amador et al. [24] applied an L/CP SUVmax index to differentiate post-therapeutic changes from tumour presence with a sensitivity of 89.3%, specificity of 90.0%, positive predictive value of 96.1%, negative predictive value of 75%, and accuracy of 82.9%. Hotta Masatoshi et al. [25] found that 11C-MET radiomics yielded excellent outcomes for differentiating recurrent brain tumours from radiation necrosis, which outperformed the T/N ratio evaluation with areas under the curve of 0.98 and 0.73. This result means that 11C-MET PET/CT is useful in differentiating glioma TuR from TrE. However, another study suggests that increased uptake of 11C-MET, such as 18F-FDG, may have limited specificity in distinguishing inflammatory lesions from tumours [26]. Bashir et al. [27] found that a 20-min 18F-FET PET scan is a powerful tool with TBRmax (sensitivity 99%, specificity 94%) to distinguish posttreatment changes from recurrent glioblastoma 6 months postradiotherapy. Bogsrud et al. [28] reported the performance of a new type of amino acid 18F-fluciclovine in PET/CT of suspected residual or recurrent glioma, but the ability of 18F-fluciclovine PET/CT to discriminate between recurrent glioma and treatment-related changes could not be determined because no patients had confirmed treatment-related changes. More recently, 18F-labelled DOPA is a more widely used amino acid tracer than 11C-MET because it has a longer half-life of up to 110 min, whereas that of 11C-MET is only 20 min. Humbert et al. [8] found that 18F-FDOPA PET has a significant impact on the management of patients with a suspicion of brain tumour recurrence, either glioblastoma or brain metastases, but a low impact when used to evaluate residual glioblastoma infiltration after first-line radiochemotherapy or second-line bevacizumab. 18F-FDOPA has been increasingly used in glioma and exhibits potential value in the identification of TrE and TuR [21].

Some other tracers such as the nucleoside analog 18F-fluorothymidine (18F-FLT) and hypoxia tracer 18F-fluoromisonidazole (18F-FMISO) also displayed reliable performances in glioma imaging. 18F-FLT can reflect tumor proliferation rate and be a marker of glioma aggressiveness due to thymidine is a nucleoside encountered in DNA [29, 30]. But it reflects proliferative indices to variable and potentially unreliable extents [31]. Given the low accumulation of 18F-FLT in low-grade gliomas, 18F-FLT PET/CT should not be used in low-grade recurrent gliomas [32]. What’s more, Enslow et al. [33] found there was no significant difference between TrE and TuR with regard to SUVmax parameter of 18F-FLT PET/CT. 18F-FMISO is the most common radiotracer for hypoxia imaging which can distinguish glioblastomas from lower-grade gliomas due to glioblastomas presents with necrosis and hypoxic environment, whereas lower-grade gliomas do not develop necrosis, and 18F-FMISO PET/CT also can predict the tumor microenvironment, including necrosis, vascularization, and permeability [34]. As ischemia and hypoxia are also important mechanisms of TrE [35], 18F-FMISO PET/CT is limited used in differentiating TuR from TrE, but it has potential value to assess treatment response for anti-angiogenic therapy [36].

In the present study, we demonstrated the application of three typical tracers in the detection of recurrent glioma. The overall L/G ratio of 18F-FDOPA and 13N-NH3 is better than that of 18F-FDG. Further ROC analysis showed that the L/G ratio of 18F-FDOPA appears to outperform sensitivity to 13N-NH3 in the assessment of TrE from TuR even when the specificity is 100%. 18F-FDG itself has no advantage in the diagnosis of recurrent glioma, only 13N-NH3 and 18F-FDOPA are selected for combined diagnosis analysis. Compared with 18F-DOPA alone, the diagnostic sensitivity and efficiency of the combination of 18F-DOPA and 13N-NH3 have not been improved. These findings suggest that 18F-FDOPA alone may be more acceptable to patients and that the diagnostic effect is equivalent. According to our previous research, this finding may be due to the fact that 13N-NH3 PET/CT has high specificity in the diagnosis of brain tumours compared with nontumour lesions, but the sensitivity is low when differentiating these lesions from LGG, which is consistent with the research results in this paper. There may be a certain degree of similarity in metabolic level between the process of glioma recurrence after chemoradiotherapy and the formation of LGG. Thus, 18F-DOPA itself has high specificity (100%) and sensitivity (84.6%) in distinguishing TuR and TrE. The combination of low sensitivity 13N-NH3 PET/CT does not help.

To our knowledge, this study is the first comparison among 13N-NH3, 18F-FDG and 18F-FDOPA PET/CT in patients with suspected glioma recurrence, and 18F-FDOPA exhibited good performance in the differential diagnosis of TuR from TrE. However, some limitations in this study should be noted. First, given the influence of reagents and patients' wishes, the sample size in this study is relatively small. Larger sample studies are needed to support our conclusions. Second, this cohort is biased to IDH1 mutated patients (24/37), and most patients were treated almost two years before this investigation. This finding may be due to the prevalence of IDH1 mutation in patients with relatively late recurrence, and IDH1 wild type patients are more likely to identify recurrence because of rapid disease progression. In addition, the high rate of IDH1 mutation on TrE in our data may be due to the radiosensitization and a less aggressive phenotype of IDH1 mutated gliomas [37]. Third, WHO tumour grade and other molecular profiles, such as 1p19q, may affect the process and metabolic level of glioma recurrence, which is also not further discussed in this study. Moreover, due to the lack of sufficient diagnostic gold standards at present, we can only use pathology combined with follow-up for final diagnosis, which is similar to that noted in many other studies [4, 16, 38, 39]. TrE also includes changes in different periods and TrE or recurrent tumours will often coexist. These tumours are difficult to distinguish in many cases. Finally, the L/G ratio reflects only a single voxel uptake of the lesion and does not include volume-based information; thus, this metric can yield false-negative results due to the obvious heterogeneity of glioma. This notion may be the reason why the sensitivity is relatively low compared with the specificity in this study.

Conclusion

In conclusion, PET/CT is a powerful tool to distinguish glioma TuR from TrE, and 18F-FDOPA PET/CT exhibited remarkably improved differential diagnosis efficacy than 13N-NH3 and 18F-FDG. Moreover, considering that the combination of 18F-FDOPA and 13N-NH3 shows comparably diagnostic performance with 18F-FDOPA alone, 18F-FDOPA alone is a good choice for effective diagnosis and could facilitate more effective therapeutic decision-making for patients with suspected glioma recurrence. However, this conclusion should be confirmed in more studies with larger samples.