Introduction

Differentiation of recurrent glioma from radiation necrosis is a vexing clinical problem. This is a critical issue now that concurrent chemoradiation and stereotactic radiosurgery have been used more extensively which increases the prevalence of necrosis. Contrast-enhanced magnetic resonance imaging (Ce-MRI) is the most commonly used technique for the evaluation of recurrent glioma. Contrast enhancement with gadolinium marks the breakage of the blood–brain barrier (BBB) on MRI and can be due to recurrence. However, leakage of the BBB might also be secondary to toxic effects of therapeutic interventions, such as chemotherapy, radiation and local therapeutics, which cannot be differentiated from recurrent tumour by Ce-MRI. Another concern is the regions of active tumour that do not enhance [1, 2].

Positron emission tomography (PET) can be used to differentiate recurrence from post-radiation necrosis and postoperative changes. PET is commonly performed using glucose metabolic agent, 18F-fluorodeoxyglucose (18F-FDG). However, the sensitivity of 18F-FDG-PET is lower because of the high physiological uptake of 18F-FDG in the brain. Therefore, other PET tracers that have lower physiological brain uptake were investigated. The most frequently evaluated among these is the amino acid 11C-methionine (11C-MET) [3]. Widespread utilisation of 11C-methionine is, however, limited by the short half-life of 11C (20 min). Therefore, 18F-labelled amino acid analogues with longer half-lives (110 min) such as 18F-fluoroethyl tyrosine (18F-FET) and 18F-fluorodopa (18F-FDOPA) have recently been used to assess brain tumours [4]. Studies have shown 18F-FDOPA to have similar tumour uptake to 11C-MET [5, 6]. Overall sensitivity of 18F-FDOPA PET-CT varied from 85 % to 100 % and specificity varied from 89 % to 100 % [6, 7]. However, most of the studies of 18F-FDOPA PET-CT were on primary gliomas and its role in recurrent gliomas is still unclear [6–9]. Therefore, the present study focussed on recurrent gliomas. To our knowledge, this is the first clinical study to systematically compare the diagnostic accuracy of 18F-FDOPA PET-computed tomography (CT) and Ce-MRI in a homogeneous population of suspected recurrent glioma.

Materials and methods

This prospective analytical study was conducted after obtaining the approval of the institutional review board (ref. no. IESC/T-64/2010). Written informed consent was obtained from all patients. Patients were recruited into the study between April 2010 and January 2011.

Patients

Thirty-five patients were included in the study. Inclusion criteria were histopathologically proven glioma, previous treatment with surgery or radiotherapy or both and clinical suspicion of recurrence. Exclusion criteria were primary brain tumour other than gliomas, proven malignancy of other sites from which metastasis can occur to the brain and patients who refused to give informed consent. The criterion to determine if the patients were likely to have recurrence was based on clinical suspicion based on signs and symptoms. The decision to include the patients in this study was made by the neurosurgeon. All patients underwent 18F-FDOPA PET-CT and Ce-MRI within a 1-week span.

18F-FDOPA PET-CT imaging

18F-FDOPA was synthesised in-house using the electrophilic fluorodemetallation method via fluorodestannylation using an automated synthesis module (Raytest Isotopenmessgerate, Straubenhardt, Germany) as previously reported [10]. After fasting for at least 4 h, a dose of 3.5 MBq/kg of 18F-FDOPA was injected intravenously into patients. After a 20- to 30-min uptake period patients were taken for PET-CT. PET-CT was acquired on a dedicated system present at our institute (Biograph 2; Siemens Medical Solutions, Erlangen, Germany). In the PET-CT system, CT was performed on spiral dual slice CT. CT-based attenuation correction of the emission images was employed. PET images were reconstructed by iterative method ordered subset expectation maximisation (two iterations and eight subsets) with a filter of 5 mm.

Ce-MRI imaging

MRI was performed at 1.5-T (Sonata/Avanto; Siemens, Germany) and images were acquired using a standard head coil. Transaxial T1-weighted and T2-weighted images were obtained from the second cervical vertebral body to the vertex. Slice thickness was adjusted to 1 mm. Contrast-enhanced images were also obtained after intravenous administration of gadopentetate dimeglumine (Gd-DTPA) at a dose of 0.1 mmol/kg using standard procedures.

Interpretation of images

The PET-CT images were evaluated independently by two experienced nuclear medicine physicians. They were blinded to the clinical and structural imaging findings. 18F-FDOPA PET-CT images were interpreted as positive for recurrent tumour when the lesion definitely increased 18F-FDOPA accumulation taking into consideration the background and contralateral site. The area on the CT images and the fused PET-CT images always corresponded. The two reviewers’ independently reviewed the FDOPA PET-CT images. Their findings were concordant in 33/35 cases. In the remaining two discordant cases a consensus diagnosis was reached after mutual discussion. For semi-quantitative analysis, an ROI was carefully drawn around the site of the abnormal 18F-FDOPA uptake in the consequent 4–6 PET-CT slices. For tumours located adjacent to basal ganglia, the CT images of 18F-FDOPA PET-CT were used for delineation of tumour margins. The slices with a maximal 18F-FDOPA uptake in the ROI were chosen for quantitative measurement of metabolic activity of the tracer (standardised uptake value [SUV]). From these ROIs, the SUV was calculated using the standard body weight method [11]. The site of the ROIs that are drawn to calculate the SUVmax in reference brain regions are illustrated in Fig. 1 Ratios of tumour uptake to normal tissue uptake were generated by dividing the tumour SUVmax by the SUVmax of the contralateral normal hemispheric brain tissue (T/N), the normal striatum (T/S), the normal white matter (T/W) and the normal cerebellum (T/C). One experienced neuroradiologist evaluated the Ce-MRI images and was blinded to clinical and PET-CT findings. Gd-DTPA-enhancing lesions were considered positive for recurrence on Ce-MRI images.

Fig. 1
figure 1

Reference case demonstrating the region of interest (ROI) drawn for semiquantitative analysis on 18F-FDOPA PET-CT. For the tumoral uptake ROI was drawn along the entire length of tumour and the slice showing maximal uptake (a 1) was used. For the determination of tracer uptake in normal grey matter, two circular ROIs with 8-mm diameters were placed along the cortical band of the contralateral frontal (a 2) and parietal cortices (a 3). The ROIs from the frontal and parietal cortices were then averaged to generate the grey matter data. Similar ROI was drawn in the centrum semiovale (a 4) for white matter uptake. ROIs were also drawn over the contralateral striatum (b 5) and cerebellum (c 6)

Full size image

Reference standard

Combination of clinical follow-up, repeat imaging and biopsy (when available) were taken as the reference standard. Patients who had disease-related adverse events, progressive disease on imaging and/or positive biopsy were taken as positive for recurrence.

Statistical analysis

Statistical package STATA 8.0 (STATA, College Station, TX, USA) and SPSS 16.0 (SPSS, Chicago, IL, USA) were used for all the statistical analyses. Various descriptive statistics such as mean, median, range and standard deviation (SD) were used to describe the baseline profiles of all the patients. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy with 95 % confidence interval were calculated for each technique. McNemar’s test with Yates correction was used to compare the techniques.

Results

Patient characteristics

A total of 35 patients with a mean age of 38.6 (± 1.25) years and age range of 11–62 years were evaluated. Patient characteristics including sex, primary location of tumour, histology, grade and primary therapy are summarised in Table 1.

Table 1 Patients’ characteristics
Full size table

Follow-up outcome

On the basis of clinical follow-up, repeat imaging and biopsy 26 patients were positive and nine patients were negative for recurrence. All patients were followed up for a period of at least 1 year. Twelve out of the total 35 patients had a disease-related adverse event (death). Six patients were re-operated and all of them were found to be positive at histopathology (two with GBM and four with anaplastic astrocytoma). Repeat imaging at follow-up was performed in the remaining 17 patients and showed persistent or progressive disease in eight patients.

Results of Ce-MRI

Ce-MRI was positive for recurrence in 29 patients and negative in six patients. Table 2 shows the true-positive, true-negative, false-positive and false-negative results on Ce-MRI. Table 3 shows the overall and grade-wise sensitivity, specificity, PPV, NPV and accuracy of Ce-MRI. Mean lesion size obtained from the contrast-enhanced area in positive cases was 3.60 cm (± 0.37 cm) and the median size was 3.4 cm (range 1.1–6.6 cm).

Table 2 18F-FDOPA PET-CT and MRI findings in all patients with different primary grades
Full size table
Table 3 Sensitivity, specificity, PPV, NPV and accuracy of 18F-FDOPA PET-CT and MRI in detection of recurrent glioma; overall and in patients with different histopathological grades (with 95 % CI)
Full size table

Results of 18F-FDOPA PET-CT

18F-FDOPA PET-CT was positive for recurrence in 27 and negative in eight patients. Table 2 shows the true-positive, true-negative, false-positive and false-negative results on 18F-FDOPA PET-CT. Table 3 shows the overall and grade-wise sensitivity, specificity, PPV, NPV and accuracy of 18F-FDOPA PET-CT.

Comparison of Ce-MRI and 18F-FDOPA PET-CT

Overall, FDOPA PET-CT was highly sensitive (100 %) and had higher specificity (88.89 %) for the detection of recurrence. Ce-MRI had high sensitivity of 92.3 % with poor specificity (44.44 %). The comparative diagnostic accuracy of 18F-FDOPA PET-CT and Ce-MRI are given in Table 3.

18F-FDOPA PET-CT and Ce-MRI findings were concordant in 29 (74.3 %) patients (Figs. 2, 3, and 4). There was discordance between the findings of Ce-MRI and 18F-FDOPA PET-CT in the remaining six patients (17.1 %; Fig. 5). On McNemar analysis the difference was not statistically significant (P = 0.687). The difference was not significant either for high-grade tumours (P = 0.5) or for low-grade tumours (P = 1.0). Ce-MRI was false negative in two patients (Grade I-1; Grade II-1). There were no false-negative reports on 18F-FDOPA PET-CT. There were five false-positive cases on Ce-MRI (GBM-1; Grade III-1; Grade II-3) but only one false-positive case (Grade II-1) on 18F-FDOPA PET-CT (Table 2). 18F-FDOPA PET-CT was more specific than Ce-MRI overall (P = 0.0002) as well as for high-grade tumours (P = 0.0066) and low-grade tumours (P = 0.004).

Fig. 2
figure 2

A 32-year-old man with left parieto-occipital grade III oligoastrocytoma, primarily treated with surgery and radiotherapy. Transaxial T1-weighted Ce-MRI (a) showed a postoperative gliotic cavity in the left parieto-occipital region without any contrast enhancement and was interpreted as negative for recurrence. Transaxial 18F-FDOPA PET (b) and PET-CT (c) were also negative. This patient was followed up for 16 months and was fine without any further treatment. Both Ce-MRI and 18F-FDOPA PET-CT were true negative in this patient

Full size image
Fig. 3
figure 3

A 56-year-old woman with glioblastoma multiforme in the left thalamic region primarily treated with surgery, radiotherapy and temozolamide presented with right-sided weakness. Transaxial T1-weighted Ce-MRI (a) images reveals a contrast enhanced lesion in the left thalamic region (arrow). Transaxial 18F-FDOPA PET (b) and PET-CT (c) images reveal intense uptake in the left thalamic region (arrow). Findings are positive for recurrence but with obscure margins owing to physiological uptake in the striatum. The patient died of progressive disease within 3 weeks of PET-CT. Both Ce-MRI and 18F-FDOPA PET-CT were true positive but MRI showed distinct margins

Full size image
Fig. 4
figure 4

A 58-year-old woman with Grade II oligodendroglioma in the right frontal region, primarily treated with surgery and radiotherapy. Transaxial T1-weighted Ce-MRI (a) shows a contrast enhancing lesion in the bilateral frontal lobe (arrows) which corresponds to lesion uptake of radiotracer in 18F-FDOPA PET (b) and PET-CT (c). Follow-up MRI (d) done 6 months later revealed disappearance of the lesion and the patient remained stable at 18 months’ follow-up. Both Ce-MRI and 18F-FDOPA PET-CT were false positive in this patient

Full size image
Fig. 5
figure 5

A 48-year-old man with a right medial frontal lobe Grade I astrocytoma, primarily treated with surgery and radiotherapy. He presented with left-sided weakness. Transaxial T1-weighted Ce-MRI (a) showed postoperative gliotic cavity in the right frontal region without any contrast enhancement (arrow) and was interpreted as negative for recurrence. Transaxial PET (b) and PET-CT (c) show an irregular area of 18F-FDOPA uptake in the right medial frontal region (arrow) suggestive of recurrence. The recurrence was confirmed on post-operative histopathology which turned out to be grade III anaplastic astrocytoma. Ce-MRI was false negative and 18F-FDOPA PET-CT was true positive for recurrence in this patient

Full size image

Semi-quantitative analysis

On semi-quantitative analysis of 18F-FDOPA PET-CT parameters, the mean SUV max of the tumour lesion was 1.35 ± 0.45 (median = 1.3; range 0.4–2.4), the mean T/S ratio in all cases was 0.80 ± 0.24 (median = 0.86; range 0.27–1.25), mean T/W ratio in all cases was 2.17 ± 0.81 (median = 2.4; range 0.8–4.53), mean T/N ratio in positive cases was 1.96 ± 0.47 (median = 2; range 0.7–3.4) and mean T/C ratio in positive cases was 1.68 ± 0.58 (median = 1.74; range 0.5–2.67). No significant positive correlation was noted between tumour SUV max, T/C, T/N, T/S and T/W of 18F-FDOPA PET-CT with size of contrast enhancement on MRI. Also no significant difference was noted between tumour SUV max, T/C, T/N, T/S and T/W ratios of 18F-FDOPA PET-CT among different grades of glioma.

Discussion

Contrast enhancement of brain tumours on MRI relies on BBB damage (frequent in grades 3 and 4, usually absent in grades 1 and 2) and morphological appearance (e.g. presence of necrosis, vascularity) [12]. Although this is regarded as largely sufficient in untreated gliomas, it becomes less reliable in treated tumours. Necrosis induced by therapy or occurring spontaneously during tumour progression may also show contrast enhancement and hence cannot be distinguished reliably from a recurrent solid tumour after therapy [13, 14]. In addition, uptake of contrast agent may be substantially reduced by dexamethasone given for reducing cerebral oedema [15]. In the present study, Ce-MRI showed high sensitivity and poor specificity because of its high false-positive rate. This is in accordance with existing literature [16, 17]. This limitation of Ce-MRI necessitates imaging methods that distinguish recurrent tumour from reactive non-neoplastic tissue.

18F-FDOPA is an amino acid tracer for PET that crosses the BBB via a carrier-mediated transport mechanism commonly referred to as the large neutral amino acid (LNAA) transport system [18, 19]. The 18F label in position 6 of the aromatic ring is believed to provide primary transport information as it has been shown to be removed immediately following membrane transport [20]. 18F-FDOPA PET-CT has been evaluated for imaging brain tumours and has shown high specificity across all grades. The best results for 18F-FDOPA PET-CT are seen in patients with low-grade gliomas. In the present study, the overall sensitivity and specificity of 18F-FDOPA PET-CT were 100 % and 88.8 %, whereas those for Ce-MRI were 92.3 % and 44.4 %, respectively. Also 18F-FDOPA PET-CT showed higher accuracy (97.1 %) than contrast-enhanced Ce-MRI (80 %). However, no statistically significant difference was seen between the overall results of Ce-MRI and 18F-FDOPA PET/CT (P = 0.687). In addition, the difference was significant neither for high-grade (P = 0.5) nor for low-grade tumours (P = 1.0). The possible explanation could be due to relatively small sample size, especially that of low-grade glioma. The major advantage of 18F-FDOPA PET-CT over Ce-MRI was its higher specificity. It was more specific than Ce-MRI for both high-grade (P = 0.006) and low-grade tumours (P = 0.004). Being more specific it will prevent unnecessary surgery and other forms of treatment and the complications associated with them. Though we did not assess the exact cost effectiveness, the use of 18F-FDOPA PET-CT for patients with clinical suspicion of recurrent glioma is likely to reduce the cost of overall management. Further studies are warranted to prove its exact cost effectiveness in this setting.

Ce-MRI was false negative in two patients (Grade I-1; Grade II-1), while there were no false negatives on 18F-FDOPA PET-CT. Ce-MRI showed poor sensitivity in low-grade gliomas (75 %) because of the limited BBB disruption in such tumours (Fig. 5). Ce-MRI had a relatively large number of false-positive cases (n = 5; GBM-1; Grade III-1; Grade II-3). Conversely only one patient of Grade II had a false-positive finding on 18F-FDOPA PET-CT (Fig. 4). The possible explanation for false positive 18F-FDOPA PET-CT includes high levels of amino acid transport into macrophages, which are activated after surgery. Alternatively, leakage of 18F-FDOPA due to BBB disruption could account for this finding [8].

In general, 18F-FDOPA PET-CT shows high uptake in tumour tissue and low uptake in normal brain, thereby yielding a greater tumour to normal contrast. But its physiological uptake in bilateral basal ganglia causes difficulty in delineating the margins of tumours located adjacent to striatum. In these cases correlation with Ce-MRI is important to characterise the margins (Fig. 3).

On semi-quantitative analysis, no significant positive correlation was noted among tumour SUV max, T/C, T/N, T/S and T/W ratios of 18F-FDOPA PET-CT with the grade of glioma, implying that the tumour grade did not significantly affect tracer uptake. This finding is consistent with the results of most studies with amino acid tracers [5, 6, 21]. Also, no statistically significant difference was seen in uptake ratios between tumours that were contrast enhancing on MRI and those that were unenhancing. This supports the fact that the tumour accumulation of 18F-FDOPA is mostly mediated through a specific transport system rather than requiring the breakdown of the BBB [22].

In recent years, advanced MRI sequences (spectroscopy, perfusion, diffusion-weighted, diffusion tensor imaging and tractography) and functional MRI have added important complementary information in the characterisation, therapy planning and recurrence differential diagnosis of brain tumours. They can improve the diagnostic accuracy of Ce-MRI alone for recurrent glioma. Few recent studies comparing diffusion-weighted MRI with 18F-FDOPA [23] and 18F-FDG PET [24] were conducted with newly diagnosed gliomas. In future, studies comparing advanced MRI sequences with 18F-FDOPA PET-CT for recurrent glioma are warranted. Also, the increasing availability of PET-MRI systems opens the exciting avenue of combining the two techniques in the form of 18F-FDOPA PET-MRI.

The present study has certain limitations. The sample size was relatively small, especially that of low-grade glioma. This reduced the power of the study. Hence any inference derived from this study needs to be revalidated with a larger study with a sufficient number of patients in each subgroup. Also, histopathological confirmation of recurrent tumour was available in only six patients. Although ideal, this was not ethically or technically feasible.

In conclusion, 18F-FDOPA PET shows a high but comparable diagnostic accuracy to Ce-MRI for the detection of recurrent glioma. However, it is more specific than Ce-MRI for this purpose.