Response assessment of bevacizumab in patients with recurrent malignant glioma using [18F]Fluoroethyl-l-tyrosine PET in comparison to MRI

  • Norbert Galldiks
  • Marion Rapp
  • Gabriele Stoffels
  • Gereon R. Fink
  • Nadim J. Shah
  • Heinz H. Coenen
  • Michael Sabel
  • Karl-Josef Langen
Original Article

DOI: 10.1007/s00259-012-2251-4

Cite this article as:
Galldiks, N., Rapp, M., Stoffels, G. et al. Eur J Nucl Med Mol Imaging (2013) 40: 22. doi:10.1007/s00259-012-2251-4

Abstract

Purpose

To investigate prospectively the potential of O-(2-[18F]fluoroethyl)-l-tyrosine (18F-FET) PET in comparison to MRI for the assessment of the response of patients with recurrent high-grade glioma (rHGG) to antiangiogenic treatment.

Methods

Ten patients with rHGG were treated biweekly with bevacizumab/irinotecan (BEV/IR). MR images and dynamic 18F-FET PET scans were obtained at baseline and at follow-up after the start of treatment (median 4.9 weeks). Using MRI treatment response was evaluated according to RANO (Response Assessment in Neuro-Oncology) criteria. For 18F-FET PET evaluation, a reduction >45 % of the metabolically active tumour volume was considered as a treatment response, with the metabolically active tumour being defined as a tumour-to-brain ratio (TBR) of ≥1.6. The results of the treatment assessments were related to progression-free survival (PFS) and overall survival (OS). For further evaluation of PET data, maximum and mean TBR were calculated using region-of-interest analysis at baseline and at follow-up. Additionally, 18F-FET uptake kinetic studies were performed at baseline and at follow-up in all patients. Time–activity curves were generated and the times to peak (TTP) uptake (in minutes from the beginning of the dynamic acquisition to the maximum uptake) were calculated.

Results

At follow-up, MRI showed a complete response according to RANO criteria in one of the ten patients (10 %), a partial response in five patients (50 %), and stable disease in four patients (40 %). Thus, MRI did not detect tumour progression. In contrast, 18F-FET PET revealed six metabolic responders (60 %) and four nonresponders (40 %). In the univariate survival analyses, a response detected by 18F-FET PET predicted a significantly longer PFS (median PFS, 9 vs. 3 months; P = 0.001) and OS (median OS 23.0 months vs. 3.5 months; P = 0.001). Furthermore, in four patients (40 %), diagnosis according to RANO criteria and by 18F-FET PET was discordant. In these patients, PET was able to detect tumour progression earlier than MRI (median time benefit 10.5 weeks; range 6–12 weeks). At baseline and at follow-up, in nonresponders TTP was significantly shorter than in responders (baseline TTP 10 ± 8 min vs. 35 ± 9 min; P = 0.002; follow-up TTP 23 ± 9 min vs. 39 ± 8 min; P = 0.02). Additionally, at baseline a kinetic pattern characterized by an early peak of 18F-FET uptake followed by a constant descent was more frequently observed in the nonresponders (P = 0.018).

Conclusion

Both standard and kinetic imaging parameters derived from18F-FET PET seem to predict BEV/IR treatment failure and thus contribute important additional information for clinical management over and above the information obtained by MRI response assessment based on RANO criteria.

Keywords

Kinetic analysis RANO criteria Anti-VEGF treatment Amino acid PET Metabolic response 

Introduction

Bevacizumab (Avastin®; Genentech/Roche) is a humanized monoclonal antibody inhibiting the biologic activity of vascular endothelial growth factor (VEGF). It is increasingly used as a single antiangiogenic agent or in combination with chemotherapy (e.g. irinotecan) in patients with recurrent high-grade glioma [1]. Bevacizumab is thought to normalize tumour vasculature and restore the blood–brain barrier (BBB), changes resulting in decreased contrast enhancement and reduced peritumoral oedema. The conventional approach to measuring tumour response is based on the criteria of Macdonald et al. [2] which reflect the two-dimensional extent of contrast-enhancing tumour (the product of the maximal cross-sectional enhancing diameters in the same plane) as seen on CT or MRI images. However, after bevacizumab treatment, in 20–30 % of all patients, high-grade gliomas progress as nonenhancing tumours with a growth pattern similar to that of gliomatosis cerebri [3]. Therefore, the assessment of tumour progression during antiangiogenic therapy using bevacizumab may be difficult when using the standard MRI criteria of Macdonald et al.

To achieve a more reliable assessment of both tumour response and progression by MRI, the RANO (Response Assessment in Neuro-Oncology) criteria were recently introduced. The RANO criteria recommend hyperintensity on fluid-attenuated inversion recovery (FLAIR) or T2-weighted MR images as a surrogate for nonenhancing tumour with an intact BBB [4]. It is, however, difficult to differentiate nonenhancing tumour from other disorders that cause FLAIR/T2 hyperintensity (e.g. radiation-induced gliosis, peritumoral oedema, ischaemia and demyelination) [5]. Furthermore, RANO criteria do not quantify the degree of FLAIR or T2 change necessary to define progression.

PET studies using the radiolabelled amino acid O-(2-[18F]fluoroethyl)-l-tyrosine (18F-FET) offer improved detection of the “true” extent of glioblastoma, i.e. the extent of metabolically active tumour, which is not limited to the area of BBB disruption and is more specific than the information provided by FLAIR hyperintensity on MRI alone [6].

Based on recent promising experience with radiolabelled amino acids concerning assessment of treatment response [7, 8, 9, 10, 11, 12], we hypothesized that 18F-FET PET could be used to gather valuable information about high-grade gliomas that cannot be derived from MRI alone. Furthermore, a recent retrospective study has indicated that 18F-FET PET may indeed be more reliable for monitoring the effects of antiangiogenic treatment than MRI-based RANO criteria [13]. In that study, 18F-FET PET detected changes in tumour volumes in 4 of 11 patients, indicating no response in these patients, earlier than MRI. The PET results reported in that study, however, were based solely on the calculation of the metabolically active tumour volume. We additionally explored tumour-to-brain ratios (TBRs) and 18F-FET PET kinetics since a number of studies have shown that the evaluation of 18F-FET kinetics may provide relevant diagnostic information, especially for noninvasive tumour grading and the differentiation of recurrent high-grade glioma from radiation necrosis [14, 15, 16, 17, 18].

To conclude, the key purpose of this prospective study was to further investigate the predictive value of 18F-FET PET-derived parameters in patients with recurrent high-grade glioma during treatment with bevacizumab/irinotecan (BEV/IR).

Materials and methods

Patients and treatments

Ten patients were included in this prospective study. Inclusion criteria were as follows: (1) histologically proven progressive high-grade glioma who had failed first-line or second-line treatment and were not eligible for further surgery, re-irradiation, or conventional chemotherapy; (2) measurable disease on MRI; (3) Karnofsky index of ≥60 %; (4) a minimum time of 6 weeks since last treatment with surgery, radiation therapy, or chemotherapeutic agents; and (5) recovery from prior treatment. Exclusion criteria were as follows: (1) Karnofsky index of <60 %; (2) evidence of haemorrhage on baseline MRI; (3) prior malignancy; (4) previous treatment with bevacizumab; (5) pregnancy or breast feeding in female patients; and (6) any other condition that could influence treatment (e.g. bone marrow depression). The study was approved by the local ethics committee and federal authorities and all patients gave written informed consent prior to each MRI and 18F-FET PET investigation. At the time of disease progression, patients had the following histological diagnoses: five primary glioblastomas (GBM), four secondary GBMs (sGBMs), and one oligoastrocytoma, WHO grade 3 (all of these diagnoses were confirmed histologically; Table 1).
Table 1

Characteristics of the patient population

Patient no.

Sex

Age before BEV/IR therapy (years)

Tumour location

Initial diagnosis

No. of recurrences before BEV/IR therapy

Treatment course before BEV/IR therapy

Tumour type before BEV/IR therapy

BEV/IR PFS (months)

BEV/IR overall survival (months)

Total overall survival (months)

Surgery

Radiotherapy (Gy)

Chemotherapy

Experimental

1

M

42

Right temporal

Glioblastoma grade IV

3

Primary tumour, first recurrence, partial resection; second recurrence, R + BCNU

Primary tumour, 60

Primary tumour, TMZ; first recurrence, rTMZ; third recurrence, CCNU

 

Glioblastoma grade IV

4

6

65

2

M

66

Right temporal

Astrocytoma grade III

2

Primary tumour complete resection; first recurrence, second recurrence, partial resection

Primary tumour, 54; first recurrence, 60

Primary tumour, –; first recurrence, TMZ

Second recurrence PDT

Secondary glioblastoma grade IV

12

38

68

3

M

53

Right frontal

Astrocytoma grade II

1

Primary tumour, biopsy; first recurrence, partial resection

Primary tumour, –; first recurrence, 60

Primary tumour, –; first recurrence TMZ

 

Secondary glioblastoma grade IV

No progress

Alive at time of reporting

Alive at time of reporting

4

F

49

Right frontotemporal

Glioneuronal tumour grade II

2

Primary tumour, first recurrence, partial resection

Primary tumour, –; first recurrence, 60

Primary tumour, –; first recurrence, TMZ; second recurrence diTMZ

 

Secondary glioblastoma grade IV

12

23

53

5

M

45

Right parietal

Astrocytoma grade II

3

Primary tumour, first recurrence, partial resection; second recurrence, R + BCNU

Primary tumour, –; first recurrence, 60

Primary tumour, –; first recurrence, TMZ; third recurrence, diTMZ

 

Secondary glioblastoma grade IV

6

12

34

6

F

26

Bifrontal

Glioblastoma grade IV

2

Primary tumour, partial resection; first recurrence, R + BCNU

Primary tumour, 60

Primary tumour, TMZ; second recurrence, diTMZ

 

Glioblastoma grade IV

2.5

2.5

10

7

M

21

Right frontal

Oligoastrocytoma grade III

5

Primary tumour, first recurrence, fourth recurrence partial resection; third recurrence R + BCNU

Primary tumour, 54; first recurrence 60

Primary tumour, HIT-GBM-C; second recurrence TMZ

Fifth recurrence, DC

Oligoastrocytoma grade III

3

3.5

103

8

M

45

Right frontal

Glioblastoma grade IV

1

Primary tumour, biopsy

Primary tumour, 60

Primary tumour, TMZ; first recurrence, diTMZ

 

Glioblastoma grade IV

3.5

4.5

23

9

M

69

Right frontal

Glioblastoma grade IV

2

Primary tumour, first recurrence, complete resection; second recurrence, R + BCNU

Primary tumour, 60

Primary tumour, TMZ

 

Glioblastoma grade IV

6

24.5

34

10

F

53

Right temporal

Glioblastoma grade IV

1

Primary tumour, first recurrence, complete resection

Primary tumour, 60

Primary tumour, TMZ

First recurrence, PDT

GBM grade IV

9

13

18

R + BCNU resection with carmustine wafer implantation, TMZ temozolomide, rTMZ temozolomide rechallenge, diTMZ dose-intensified temozolomide, CCNU lomustine, PDT photodynamic therapy, HIT-GBM-C treatment according to the HIT-GBM-C protocol [20], DC immune therapy with dendritic cell vaccination.

At first GBM occurrence, nine of ten patients were treated according to the EORTC 26981 study [19]. Accordingly, these nine patients underwent surgery (two macroscopic total resections, six partial tumour resections, one stereotactic biopsy), radiation therapy (extended tumour field, cumulative maximum dose of 60 Gy), and concomitant and adjuvant temozolomide chemotherapy (dosage according to the EORTC 26981 protocol [19]). One patient (patient 7) was treated according to the HIT-GBM-C protocol as initial therapy [20]. During the disease course and before BEV/IR therapy, the patient cohort developed between one and five recurrences, which were treated with further surgery alone or surgery with carmustine wafer implantation [21] (Gliadel®; Esai), re-irradiation, lomustine chemotherapy (Cecenu®; Medac), temozolomide chemotherapy (temozolomide re-challenge [22] or dose-intensified regimen [23]), photodynamic therapy [24], and immune therapy with dendritic cell vaccination (Table 1).

Patients were treated biweekly with bevacizumab (10 mg/kg body weight) and irinotecan (125 mg/m2 body surface area; BEV/IR). Three of the ten patients received corticosteroids (dexamethasone) at the time of baseline imaging before BEV/IR treatment (dose range 4–12 mg daily). In all patients steroid administration was discontinued during antiangiogenic therapy.

Progression-free survival (PFS) was defined as time in months from the beginning of BEV/IR therapy to radiologic progression according to RANO criteria [4]. Median PFS was 6 months and varied from 2.5 to 12 months (mean PFS 6 ± 4 months). One patient (patient 3) showed no progression at the 6-month follow-up and was still alive and progression-free at the time of study evaluation (PFS at study evaluation 34 months). Overall survival (OS) was defined as time in months from the beginning of BEV/IR therapy until death. Median OS was 12 months and ranged from 2.5 to 38 months (mean OS 14 ± 12 months). Median total OS (tOS), specified as time from first tumour occurrence until death, was 34 months and ranged between 10 to 103 months (mean tOS 45 ± 30 months).

Patient evaluations

Within 14 days of initiating therapy, a full medical history was obtained in all patients, and the patients underwent full physical and neurologic examinations, evaluation of vital sign signs, Karnofsky status assessment, routine blood tests, and, if necessary, a pregnancy test for any woman of child-bearing potential. 18F-FET PET/MRI assessment was then performed within 7 days and before beginning the treatment. 18F-FET PET/MRI was additionally performed at follow-up after treatment onset of BEV/IR (6–8 days after the second administration of BEV/IR). At baseline and follow-up, the median time between 18F-FET PET and MR imaging was 24 h. Afterwards, a follow-up MRI scan was performed during therapy at 6- to 8-week intervals. A full medical history and physical examination, including a full neurologic examination, were completed every 6–8 weeks.

PET imaging with 18F-FET and analysis

18F-FET was produced via nucleophilic 18F-fluorination with a specific radioactivity of >200 GBq/μmol as described previously [25]. The radiochemical yield of the tracer was about 60–65 %, at a radiochemical purity >98 %. The tracer was administered as an isotonic neutral solution. All patients fasted for at least 12 h before the PET studies. Dynamic PET studies were acquired up to 50 min after intravenous injection of approximately 200 MBq 18F-FET on an ECAT EXACT HR+ scanner (Siemens Medical Systems) in three-dimensional mode (32 rings; axial field of view, 15.5 cm). The emission recording consisted of 16 time frames (time frames 1–5 1 min, 6–10 3 min, and 11–16 5 min) covering the period up to 50 min after injection. For attenuation correction, transmission was measured with three 68Ge/68 Ga rotating line sources. After correction for random and scattered coincidences as well as dead time, image data were obtained by filtered back projection in Fourier space using the ECAT 7.2 software. The reconstructed image resolution was approximately 5.5 mm.

Summed 18F-FET PET images (20–40 min after injection) and MR images (contrast-enhanced T1-weighted images and FLAIR/T2-weighted images) were automatically coregistered using the VINCI tool [26]. The fusion results were inspected and, if necessary, adapted based on anatomic landmarks. 18F-FET uptake in the unaffected brain tissue was determined using a large region of interest (ROI) placed on the contralateral hemisphere in an area of normal appearing brain tissue including white and grey matter. The tumour area on 18F-FET PET scans was determined by a three-dimensional autocontouring process using a cut-off for the TBR of 18F-FET uptake of ≥1.6. This cut-off was adapted from the results of a previous biopsy-controlled study in which the best lesion-to-brain ratio for differentiating tumour from peritumoral tissue was 1.6 [6]. Manual corrections of the tumour ROI were applied if radioactivity in blood vessels or tracer uptake in postoperative extracerebral soft tissue exceeded the cut-off value.

18F-FET uptake in the tissue was expressed as standardized uptake value (SUV) by dividing the radioactivity (in kilobecquerels per millilitre) in the tissue by the radioactivity injected per gram of body weight. The TBRs of the tumour lesion were generated as follows: the SUVmean of the lesion was divided by the SUVmean of the contralateral normal brain tissue ROI (TBRmean), and by dividing the SUVmax of the lesion by the SUVmean of the contralateral normal brain tissue ROI (TBRmax). TBR evaluation was based on the summed 18F-FET PET data from 20 to 40 min after injection. The 18F-FET-positive tumour volume at a TBR cut-off of ≥1.6 was determined using a volume-of-interest analysis [27].

Kinetic analysis of 18F-FET tracer uptake

18F-FET uptake kinetics were assessed in the tumour area with the highest tracer uptake on summed 18F-FET PET images (20–40 min after injection). Therefore, a ROI at the threshold of the 18F-FET TBR of ≥1.6 was defined by a two-dimensional autocontouring process for each individual transverse slice within the area of 18F-FET uptake. These ROIs were then applied to the corresponding slices recorded at later time points, and the time–activity curves (TACs) within the ROIs were evaluated for the entire dynamic dataset. As previously described [18], we assessed the time to peak (TTP). TTP was defined as the time (in minutes) from the beginning of the dynamic acquisition to the maximum SUV of the tumour, i.e. peak SUV. Patterns of TACs for each lesion were assigned independently to three different predefined kinetic patterns (Fig. 1) by three independent observers (three experienced consultants in nuclear medicine; G.S., K.J.L., N.G.): constantly increasing 18F-FET uptake, curve always ascending with a clear identifiable peak SUV at the end of the dynamic study (type 1); peak SUV reached at a midway point (TTP 20–45 min) followed by a plateau or a slow descent (type 2); and maximum SUV of the tumour peaking early (TTP ≤20 min) followed by a constant descent (type 3). These different 18F-FET uptake kinetic patterns were defined on the basis of previous observations in patients with low- and high-grade gliomas [14, 15, 16, 17, 18].
Fig. 1

Example of the three types of tumour TACs derived from dynamic 18F-FET PET before BEV/IR treatment. a Type 1: constantly increasing 18F-FET uptake, the curve is always ascending with a clear identifiable peak SUV at the end of the dynamic study. b Type 2: the maximum peak is reached at a midway point (TTP >20–45 min) followed by a plateau or a slow descent. c Type 3: the peak of the curve occurs at an early time point (TTP ≤20 min) followed by a steep decrease

Definition of 18F-FET PET response

18F-FET PET metabolic responders were defined as those with a tumour volume reduction of 45 % or more at the time of the follow-up scan [13]. This threshold was identified by a receiver operating characteristic (ROC) curve analysis using a PFS of 6 months as a cut-off to differentiate long-term and short-term survivors. Metabolic responders were identified with a sensitivity of 88 % and a specificity of 100 % [13].

MR imaging

All patients had a routine MRI scan using a 1.5-T MRI scanner with a standard head coil (T1-weighted, T2-weighted and FLAIR sequences) during follow-up. Axial T1-weighted images were obtained from the second cervical vertebral body to the vertex. After intravenous administration of gadolinium-diethylenetriaminepentaacetic acid (Gd-DTPA; 0.1 mmol/kg body weight), axial T1-weighted images were obtained using standard procedures.

Definition of treatment response based upon MRI criteria

Treatment responses were assessed routinely and visually by an experienced neuroradiologist before (baseline MRI scan) and after initiation of BEV/IR treatment (follow-up MRI scan) according to RANO criteria [4]. In brief, in addition to standard Macdonald criteria [2], partial and complete responses require stable and improved nonenhancing (T2/FLAIR) lesions, stable disease requires stable nonenhancing lesions, and progressive disease requires a significant increase in nonenhancing lesions [4].

Statistics

Descriptive statistics are provided as means and standard deviations or as medians and ranges. To compare two groups, Student’s t-test was used. The Mann-Whitney rank sum test was used when variables could not be assumed to be normally distributed. The distribution of dichotomized variables was determined using the chi-squared test. A p value of <0.05 was considered to be significant. Patients were stratified into treatment responders/nonresponders according to neuroradiologic (MRI; using the RANO criteria, see above) and metabolic criteria (PET; 18F-FET volume reduction >45 %, see above). The patients were also stratified into subgroups depending on time to tumour progression. Patients with a PFS of 6 months or more were designated long-term survivors, and patients with a PFS of less than 6 months short-term survivors. Survival estimates were performed using the log-rank test (univariate analysis) for PFS and OS (data presented as median values).

To differentiate treatment responders from nonresponders, the diagnostic performance of additional PET parameters (i.e. TBRmax, TBRmean, TTP at baseline, change of TTP at follow-up, and the combination of TBR values, kinetic pattern, and TTP) were assessed by ROC curve analyses using the PFS (<6 months vs. ≥ 6 months) as reference. Decision cut-off was considered optimal when the product of paired values for sensitivity and specificity reached a maximum. In addition, the area under the ROC curve (AUC), its standard error, and level of significance were determined as a measure of diagnostic quality of the test.

Statistical analyses were performed using SigmaStat software (SigmaPlot for Windows 11.0, Chicago, IL) and PASW Statistics software (release 18.0.3; SPSS, Chicago, IL).

Results

Response evaluation

At follow-up 6–8 days after second administration of BEV/IR (mean 5.7 ± 2.4 weeks, median 4.9 weeks), according to RANO criteria MR images showed complete response in one (10 %), partial response in five (50 %), and stable disease in four (40 %) of the ten patients (Table 2). Univariate survival analysis using the RANO criteria was not possible, since no patient showed tumour progression on the MRI at follow-up scan.
Table 2

Results of 18F-FET PET tumour volumetry, TBRs, and kinetic analyses

Patient no.

Baseline

Follow-up

Response assessment

PFS (months)

OS (months)

Volume (cm3)

TBRmax

TBRmean

TTP (min)

Kinetic pattern

Volume (cm3)/percent change

TBRmax/percent change

TBRmean/percent change

TTP (min)

Kinetic pattern

RANO

18F-FET PET

1

13.4

3.7

2.0

20

3

8.3 (−38)

1.9 (−49)

1.5 (−15)

25

2

Partial response

Nonresponder

4

6

2

30.4

5.5

3.8

25

2

13.8 (−55)

3.6 (−35)

2.0 (−47)

40

1

Partial response

Responder

12

38

3

6.9

2.6

1.8

35

1

0 (−100)

1.3 (−50)

1.0 (−44)

45

1

Partial response

Responder

No progress

Alive at time of reporting

4

96.3

3.2

2.1

45

1

20.9 (−78)

2.4 (−25)

1.7 (−19)

45

1

Partial response

Responder

12

23

5

30.6

2.3

1.8

45

1

0 (−100)

1.5 (−35)

1.0 (−44)

45

1

Partial response

Responder

6

12

6

45.7

3.2

1.8

4

3

44 (−4)

3.1 (−3)

2.0 (+11)

20

3

Stable disease

Nonresponder

2.5

2.5

7

69.0

4.6

2.7

13

3

110.9 (+61)

3.9 (−15)

2.8 (+4)

35

1

Stable disease

Nonresponder

3

3.5

8

61.3

3.3

2.2

4

3

36.2 (−41)

2.8 (−15)

1.9 (−14)

13

3

Stable disease

Nonresponder

3.5

4.5

9

2.0

2.5

1.7

35

1

1.0 (−50)

2.1 (−16)

1.3 (−24)

25

2

Complete response

Responder

6

24.5

10

16.3

2.0

1.5

25

2

1.3 (−92)

2.0 (±0)

1.5 (±0)

35

1

Stable disease

Responder

9

13

According to 18F-FET PET response criteria, six of the ten patients (60 %) showed a response of the tumour to the treatment (responders). However, four of the ten patients (40 %) showed tumour progression (nonresponders) at follow-up. Detailed data on 18F-FET PET tumour volumes at baseline and at follow-up are presented in Table 2. Mean 18F-FET PET tumour volumes decreased significantly in long-term survivors but remained almost unchanged in short-term survivors (mean change −79.2 % vs. −5.5 %; P = 0.01, t-test).

18F-FET PET responders demonstrated a significantly longer PFS than nonresponders (median PFS 9 vs. 3.25 months; P = 0.016, Mann-Whitney rank sum test). In univariate survival analysis, a response by 18F-FET PET predicted a significantly longer PFS (median PFS, 9 vs. 3 months; P = 0.001, log rank test; Fig. 2). In comparison to nonresponders, a significantly longer OS was observed in 18F-FET PET responders (median OS 23 vs. 4 months; P = 0.016, Mann-Whitney rank sum test). In univariate survival analysis, a response in 18F-FET PET predicted a significantly longer OS (median OS 23.0 vs. 3.5 months; P = 0.001, log rank test; Fig. 2).
Fig. 2

Univariate survival analyses of 18F-FET PET responders vs. nonresponders during antiangiogenic treatment. Changes in the metabolically active tumour volume on 18F-FET PET predict significantly longer PFS (left) and OS (right) in responders than in nonresponders (P = 0.001, log rank test). Censored observations are marked with dots

Comparison of MRI and 18F-FET PET response assessment

Comparison of RANO and 18F-FET PET response criteria revealed that six patients (60 %) had a response or stable disease at follow-up. In contrast, four patients (40 %) demonstrated a discrepancy in response assessment, namely a partial response (one patient) or stable disease (three patients) according to RANO criteria, but no treatment response according to 18F-FET PET criteria (Fig. 3). In these four patients, 18F-FET PET detected treatment failure earlier than MRI. In our study, a median time benefit of 10.5 weeks (range 6–12 weeks) for earlier detection of treatment failure with 18F-FET PET was calculated.
Fig. 3

MRI and 18F-FET PET imaging in patient 7. Left column Pretreatment images. The contrast-enhanced T1-weighted MR image shows a contrast-enhancing lesion in the dorsomedial part of the resection cavity and an extensive nonenhancing lesion (hyperintensity on the T2-weighted MR image), which crosses the midline. The 18F-FET PET image shows metabolically active tumour in spatial correspondence with the contrast-enhancing lesion. Centre column First follow-up images after 4 weeks of BEV/IR treatment. The MR image shows a decrease in intensity of contrast enhancement and a decrease in T2 signal hyperintensity. MRI findings are not consistent with progressive disease. In contrast, 18F-FET PET shows a clear increase in the metabolically active tumour volume (+61 %). Right column Complementary images acquired after 12 weeks of treatment for clinical purposes, the MR and 18F-FET PET images show extensive tumour progression. In this patient 18F-FET PET was able to detect tumour progression 8 weeks earlier than MRI

Evaluation of 18F-FET PET kinetic studies

Detailed data of TTP and patterns of TACs at baseline and at follow-up are presented in Table 2. In nonresponders (PFS <6 months), the mean TTP at baseline was significantly shorter than in responders (PFS ≥6 months) (TTP 10 ± 8 vs. 35 ± 9 min; P = 0.002, t-test). At follow-up, in nonresponders TTP was still significantly shorter than in responders (TTP 23 ± 9 vs. 39 ± 8 min; P = 0.02, t-test). TTP at baseline alone (threshold value 23 min) differentiated responders from nonresponders with a high diagnostic performance (AUC 1.0 ± 0.0, sensitivity 100 %, specificity 100 %; P = 0.011, ROC analysis; Table 3). Furthermore, a high diagnostic performance for identifying responders was obtained using a combined approach when two of three criteria were fulfilled: (1) an increase in TTP of ≥10 min between baseline and follow-up; (2) a TTP of ≥25 min at baseline; and (3) a type 1 or 2 kinetic pattern at follow-up (AUC 0.938 ± 0.082, sensitivity 100 %, specificity 75 %; P = 0.025, ROC analysis; Table 3). Additionally, at baseline a type 3 kinetic pattern was more frequently present in nonresponders than in responders (P = 0.018; chi-squared test; Fig. 4). Likewise, at follow-up, a type 1 kinetic pattern was more frequently present in responders (P = 0.03; chi-squared test). In contrast, changes in TTP between baseline and follow-up identified responders with a low diagnostic performance (Table 3).
Table 3

Diagnostic performance of 18F-FET PET parameters between baseline and follow-up

Identification of metabolic responders

Reduction in TBRmax

Reduction in TBRmean

Reduction in 18F-FET tumour volume

TTP at baseline alone

Change in TTP

Combination of TTP and kinetic pattern

Best threshold

16 %

17 %

45 %

≤ 23 min

10 min

a

Sensitivity

83 %

83 %

100 %

100 %

50 %

100 %

Specificity

75 %

100 %

100 %

100 %

50 %

75 %

AUC ± standard error

0.667 ± 0.197

0.917 ± 0.095

1.0 ± 0.0

1.0 ± 0.0

0.250 ± 0.169

0.938 ± 0.082

P value

0.394

0.033

0.011

0.011

0.201

0.025

aWhen two of three criteria are fulfilled: (1) an increase in TTP of ≥10 min between baseline and follow-up; (2) a TTP of ≥25 min at baseline; and (3) a type 1 or 2 kinetic pattern at follow-up.

Fig. 4

Kinetic patterns before initiation of BEV/IR in relation to metabolic response: type 1 pattern in responders; type 3 pattern in nonresponders

Changes in TBRs during antiangiogenic treatment

At follow-up, a change in TBRmax of more than 16 % differentiated responders (PFS ≥6 months) from nonresponders (PFS <6 months) with a sensitivity of 83 % and a specificity of 75 % (AUC 0.667 ± 0.197; P = 0.394, unfitted ROC curve analysis; Table 3). In contrast, at follow-up a reduction in TBRmean of more than 17 % differentiated responders from nonresponders with a sensitivity of 83 % and a higher specificity of 100 % (AUC 0.917 ± 0.095; P = 0.033, ROC curve analysis; Table 3).

Discussion

This study demonstrated that both 18F-FET PET-derived standard imaging parameters (i.e. TBR and metabolically active tumour volume) and 18F-FET PET kinetics provide important clinical information about tumour response in patients with recurrent high-grade glioma treated with BEV/IR which cannot be derived from MRI assessment based upon RANO criteria.

Recently it has been demonstrated that at the time of the follow-up scan (after 8–12 weeks) the best threshold for identifying responders with a PFS of ≥6 months is a tumour volume reduction of 45 % or more (sensitivity 88 %, specificity 100 %) [13]. In line with this finding, in our study a tumour volume reduction at this threshold best differentiated responders from nonresponders even at an earlier follow-up (18F-FET PET after a median time of 4.9 weeks) and with a similar high diagnostic performance. Furthermore, we observed a high predictive value of TTP at baseline. However, in the study by Hutterer et al., the predictive value of frequently used TBRs, i.e. TBRmax and TBRmean, was not assessed. In the present study, we observed that a reduction in TBRmean of ≥17 % at follow-up differentiated responders (PFS ≥6 months) from nonresponders (PFS <6 months) with an excellent sensitivity (83 %) and specificity (100 %). Additionally, a certain constellation of TTP and kinetic pattern at baseline and follow-up differentiated responders from nonresponders with a favourable diagnostic performance. In contrast, changes in TBRmax and TTP as single parameters had a low predictive value in this dataset. It has to be born in mind, however, that our patient sample was very small and the results of the ROC analyses should be considered with caution.

Our findings are in line with those of a previous study [12] which revealed that patients with a decrease in TBRmean (more than 5 %) between baseline 18F-FET PET imaging and follow-up imaging directly after completion of radiochemotherapy with temozolomide exhibited a significantly longer median PFS and OS. The higher cut-off value in the present study may have been due to the smaller number of patients or differences in the study populations.

In addition to 18F-FET PET-derived standard imaging parameters (i.e. TBR and metabolically active tumour volume), a number of studies suggest that the evaluation of 18F-FET kinetics may add relevant diagnostic information, especially for noninvasive tumour grading and a differentiation of recurrent high-grade glioma from radiation necrosis [14, 15, 16, 17, 18]. Typically, TACs of high-grade gliomas are characterized by an early peak of 18F-FET uptake followed by a constant descent while low-grade gliomas and benign brain lesions typically show steadily increasing uptake. To date, experience with 18F-FET PET kinetics in patients with recurrent malignant glioma treated with anti-VEGF therapy remain scarce, and changes in 18F-FET kinetics due to antiangiogenic treatment have not yet been assessed.

Several factors may contribute to the kinetic behaviour of 18F-FET uptake and its changes after BEV/IR treatment. Weckesser et al. [14] suggested that the early peak in 18F-FET kinetics in high-grade gliomas might be due to a higher regional blood volume as a consequence of increased angiogenesis and intratumoral microvessel density in patients with malignant progression [28]. Additionally, the high initial uptake may be influenced by an upregulation of facilitated amino acid transporter in tumour vessels, which is the assumed mechanism for increased 18F-FET uptake in gliomas [29, 30]. Thus, it is not unlikely that treatment-induced effects on tumour angiogenesis may change the pattern of 18F-FET kinetics.

The putative mechanism underlying BEV/IR efficacy is the normalization of tumour vasculature [31]. Tumour angiogenesis produces abnormal vessels with increased tortuosity. The resultant high interstitial pressure produces hypoxia and a growth advantage for tumour cells [32]. Thus, an abnormal vasculature of the tumour microenvironment supports progression and resistance to treatment [32]. Furthermore, it has been demonstrated that structural and functional abnormalities of the newly formed tumour vessels increase the BBB permeability [31]. Antiangiogenic treatment is assumed to normalize the structure and function of the tumour vasculature, thereby promoting improved tumour blood perfusion and decreased permeability of the BBB. This assumption is consistent with recent findings in patients with recurrent glioblastoma treated with cediranib, a pan-VEGF receptor tyrosine kinase inhibitor, where responders showed increased tumour blood perfusion as measured by perfusion-weighted MRI, leading to a better outcome [33]. Taken together, these and our findings support the hypothesis that vascular normalization promotes tumour regression and longer patient survival. It is tempting to speculate that the high performance of 18F-FET PET in monitoring BEV/IR treatment is based on the capacity of this tracer to identify tumour parts with decreased permeability of the BBB that are difficult to detect by MRI [6, 34].

Different explanations for the better prognosis due to the vascular normalization can be suggested. First, it is conceivable that the normalized vessels permit better delivery of BEV to the glioblastoma cells, leading to a better antitumoral effect. Recently it has been demonstrated that regression of glioma cells can occur independently from vascular regression, suggesting that BEV has anticancer cell effects [35]. Second, the vascular remodelling resulting in improved tumour perfusion might facilitate an innate immune response [36]. Interestingly, a fraction of the recurrent high-grade gliomas, especially those responders with a PFS of ≥6 months, showed a curve pattern with steadily increasing 18F-FET uptake and higher TTP before initiation of BEV/IR treatment (Fig. 4), a pattern which has been observed previously mainly in patients with low-grade gliomas and benign brain lesions. Due to the various treatments each patient had undergone prior to BEV/IR, the reasons for this finding may be complex, but we suggest that this observation in responders may be explained best by a better structure and function of the tumour vasculature than in nonresponders which could enhance the effects of BEV/IR treatment. Future studies (e.g. in vitro and animal studies) are necessary to further clarify the molecular fundamentals of the differences in kinetic behaviour during antiangiogenic treatment.

Conclusion

Both standard and kinetic 18F-FET PET-derived imaging parameters seem to predict BEV/IR treatment response and may thus contribute significant information for the management of patients suffering from recurrent high-grade gliomas. Importantly, this information was derived over and above that which could be gained by assessing treatment response using MRI and RANO criteria. In clinical practice, 18F-FET PET in addition to MRI might add important information in response assessment, which could be useful for decision making, e.g. discontinuation of bevacizumab. However, it should be noted that our patient sample was very small and our results should be considered with caution. Therefore, a clinical trial with a larger patient cohort should be performed to further assess the diagnostic potential of 18F-FET uptake in assessing the effects of BEV/IR treatment.

Acknowledgments

The authors wish to thank Suzanne Schaden, Elisabeth Theelen and Kornelia Frey for assistance in the patient studies and Dr. Johannes Ermert, Silke Grafmüller, Erika Wabbals and Sascha Rehbein for radiosynthesis of 18F-FET. The Brain Imaging Center West (BICW) supported this work.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Norbert Galldiks
    • 1
    • 2
  • Marion Rapp
    • 3
  • Gabriele Stoffels
    • 1
    • 4
  • Gereon R. Fink
    • 1
    • 2
  • Nadim J. Shah
    • 1
    • 4
  • Heinz H. Coenen
    • 1
    • 4
  • Michael Sabel
    • 3
  • Karl-Josef Langen
    • 1
    • 4
  1. 1.Institute of Neuroscience and Medicine (INM-3,-4,-5), Forschungszentrum JülichJülichGermany
  2. 2.Department of NeurologyUniversity of CologneCologneGermany
  3. 3.Department of NeurosurgeryUniversity of DüsseldorfDüsseldorfGermany
  4. 4.Jülich-Aachen Research Alliance (JARA) – Section JARA-BrainAachenGermany

Personalised recommendations