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. 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. 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 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[21–23]. 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 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 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. Bashir et al found that a 20-minute 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 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 minutes, whereas that of 11C-MET is only 20 minutes. Humbert O. et al 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.
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 mutations in patients with relatively late recurrence, and IDH1 wild type patients are more likely to identify recurrence because of rapid disease progression. 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. 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.