Methyl-l-¹¹C-Methionine PET as a Diagnostic Marker for Malignant Progression in Patients with Glioma

Patients

This retrospective study included 24 patients with primary supratentorial cerebral gliomas (Table 1). We investigated 14 male patients and 10 female patients (mean age, 40 y; SD, 11.6 y). Each patient gave written informed consent for additional PET imaging, which is not a standard diagnostic tool for patients with gliomas in Germany. All patients who underwent repeated ¹¹C-MET PET and had a corresponding neuropathologic diagnosis during the period from 1993 to 2006 were included in the study (Fig. 1). The ¹¹C-MET PET investigations were part of the routine preoperative diagnostic procedure and were used to guide the biopsy so that the tumor portion with the highest ¹¹C-MET uptake would be obtained. Repeated PET investigations with corresponding biopsy were performed when there were findings suggestive of tumor progression on MRI or CT or suggestive clinical symptoms. The interval between PET measurements varied from 1 mo to 6 y. In total, data from 57 PET scans that had a corresponding histologic diagnosis within the following 3 weeks were available for 24 patients. ¹¹C-MET PET was performed after combined radio- and chemotherapy in 9 of 57 cases, after radiotherapy alone in 8 of 57 cases, and without preceding radio- or chemotherapy in 40 of 57 cases.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

¹¹C-MET PET of 39-y-old man with malignant progression of recurrent glioma. (A) Newly diagnosed grade II astrocytoma with average ¹¹C-MET uptake of 1.3 to contralateral gray matter, with no enhancement on contrast-enhanced CT and no immunohistochemical VEGF expression. (B) One year later, patient presented with malignant progression to grade III astrocytoma associated with significant increase in ¹¹C-MET uptake (to 2.1-fold) and only slight contrast enhancement outside metabolically active tumor. Histologic analysis from resection showed increase in cellularity, numerous pleomorphic nuclei, and low VEGF expression. (C) In following year, resection of tumor again confirmed malignant progression to glioblastoma multiforme, showing markedly increased uptake of ¹¹C-MET (to 2.8-fold), marginal contrast enhancement on MRI, and ∼35% of tumor cells expressing VEGF (original magnification, ×400). PET-guided biopsies were taken from region with highest ¹¹C-MET uptake, which was in different locations within same tumor over time.

All 24 patients underwent baseline ¹¹C-MET PET, 16 of 24 underwent follow-up ¹¹C-MET once, 7 of 24 underwent follow-up ¹¹C-MET twice, and 1 of 24 underwent follow-up ¹¹C-MET 3 times. Information on the patients included age, sex, presence of contrast enhancement on MRI or CT, and extent of surgical resection (13 stereotactic biopsies, 23 subtotal resections, and 21 macroscopic total resections). The tumors were graded according to the World Health Organization (WHO) classification for neuroepithelial tumors (2).

In the initial histologic diagnosis, tumor types were distributed as follows: WHO grade II astrocytoma (n = 10), WHO grade III anaplastic astrocytoma (n = 3), WHO grade II oligoastrocytoma (n = 7), WHO grade III anaplastic oligoastrocytoma (n = 2), WHO grade II oligodendroglioma (n = 1), and WHO grade III anaplastic oligodendroglioma (n = 1) (Table 1).

PET Studies

For PET imaging, we used an ECAT EXACT scanner (CTI/Siemens; in-plane full width at half maximum, 6 mm; slice thickness, 3.375 mm; axial field of view, 162 mm) and an ECAT EXACT HR scanner (CTI/Siemens; in-plane full width at half maximum, 3.6 mm; slice thickness, 3.125 mm; axial field of view, 150 mm) (15,16). Subsequent imaging was performed on the same scanner. All patients fasted for at least 4 h before undergoing PET. Images were acquired with the patient supine with eyes closed. Before tracer application, a 10-min transmission scan with 3 rotating ⁶⁸Ga/⁶⁸Ge sources was obtained. ¹¹C-MET was synthesized according the method of Berger et al. (17) and injected intravenously as a bolus injection of 740 MBq (20 mCi). Accumulation of the tracer was recorded 0–60 min after tracer injection. For assessment of ¹¹C-MET PET images, frames recorded 20–60 min after tracer application in 47 transaxial slices of the entire brain were used. The spatial resolution was 6 mm or better in all dimensions.

In 20 of 24 patients, 48 of 57 PET scans had corresponding contrast-enhanced MRI (n = 36) or CT (n = 12) scans. In 12 of 24 patients, MRI and PET were performed, in 6 of 24 patients MRI and CT and PET were performed, in 2 of 24 patients CT and PET were performed, and in 4 of 24 patients only PET was performed. Regions of ¹¹C-MET uptake were compared with areas of contrast enhancement on MRI or CT.

PET data were evaluated using a region-of-interest analysis. Because ¹¹C-MET uptake at later imaging times more specifically reflects transport activity, we used summed images covering the time frame 20–60 min after injection for data analysis. As described by Herholz et al., a circular region of interest 7 mm in diameter was placed on the area of highest ¹¹C-MET uptake to determine the maximal tracer uptake (18). A mirrored, contralateral region of interest of the same diameter was placed as a reference. The relative index of ¹¹C-MET uptake was calculated as the ratio of tumor area to reference area. The change in ¹¹C-MET uptake was defined as the relative percentage change in ¹¹C-MET uptake between 2 subsequent scans of the same patient.

MRI/CT Studies

MRI was performed on a 1.5-T system (Gyroscan ACS-NT; Philips Medical Systems) using a head coil. T1-weighted spin-echo enhanced images were acquired with a slice thickness of 2 mm and a matrix of 512 × 512 pixels. To monitor gadopentetate dimeglumine enhancement of the tumors, we used a T1-weighted 3-dimensional gradient-echo sequence. CT was performed after administration of contrast medium (Solutrast 300R; Bracco). We acquired 22–40 CT images 2 mm thick.

Histologic Analysis

Histologic analysis and immunohistochemistry of the biopsy samples of the initial tumors of all patients, as well as recurrences obtained by stereotactic biopsy or open surgery, were performed on formalin-fixed, paraffin-embedded 4-μm sections. For immunohistochemistry, we applied an automated staining system (BioGenex) using the avidin-biotin-peroxidase complex technique, with 3,3′-diaminobenzidine as chromogene and H₂O₂ as cosubstrate. In brief, classification of the tumors according to the WHO classification of neuroepithelial tumors was based on hematoxylin and eosin staining and immunohistochemistry with monoclonal antibodies against rabbit antihuman MIB-1 (DCS, Innovative Diagnostik-Systeme; clone SP6; dilution, 1:200) and mouse antihuman p53 protein (BioGenex; clone 1801; dilution, 1:200) and polyclonal rabbit antihuman antibodies against glial fibrillary acidic protein and S100 protein (Dako; dilution, glial fibrillary acidic protein, 1:1,000; S100-protein, 1:2,000). Advanced immunohistochemistry was performed with the following monoclonal mouse antihuman antibodies: epidermal growth factor receptor (Merck; clone E30; dilution, 1:20), platelet-derived growth factor receptor (BD Biosciences; clone 28; dilution, 1:200), retinoblastoma protein (pRb) (Zymed Laboratories; clone Rb1; dilution, 1:50), pentaerythritol tetranitrate (PTEN) (Biogenex; clone 28H6; dilution, 1:10), and vascular endothelial growth factor (VEGF) (DCS; clone VG1; dilution, 1:50). Histologic evaluation was performed by 2 independent neuropathologists. The number of immunoreactive nuclei was determined, comprising 3 areas (F1, F2, and F3) of 3 high-power fields each, with maximal frequency, moderate frequency, and minimal frequency, respectively, of immunoreactive nuclei. The number of positive nuclei was determined as F1 plus F2 plus F3, divided by 3 (%).

Statistical Analysis

Statistical analyses were performed using SPSS software (release 11.0.1, SPSS Inc.). For correlation analysis, the Pearson method was applied, with subsequent parametric tests; 2-sample ANOVA was used for comparisons between the groups with and without malignant progression. Tests were performed 2-sided at a significance level of 0.05, and the P values were understood in an explorative sense regarding the multiple-hypothesis problem. The sensitivity and specificity of changes in ¹¹C-MET uptake were calculated for several thresholds, and the optimum cut-off was determined by receiver-operating-characteristic analysis. An iterative leave-one-out approach was used to validate the receiver-operating-characteristic analysis. At each step, 1 case (i.e. 1 follow-up) was left out of the analysis; a fit of the model was produced for the remaining follow-ups, and a malignant progression prediction was made for the left-out case. Thus, each follow-up value was compared with the cohort of the remaining 32 of 33 values with regard to its individual percentage change in ¹¹C-MET uptake. Moreover, baseline ¹¹C-MET uptake was included as a covariate in a Cox regression model to investigate its relation to the time to histologic progression. To avoid an artificial reduction of variance caused by the mixing of independent and dependent observations, we used the data of only the first follow-up investigation (n = 24) for all methods of statistical inference (Pearson method and 2-sample ANOVA).

Malignant Progression as Detected by ¹¹C-MET PET

For evaluation of the correlation between ¹¹C-MET uptake and histologic progression of the tumor, we calculated the percentage change in ¹¹C-MET uptake between the prior ¹¹C-MET PET examination and the subsequent examination. The percentage change in ¹¹C-MET uptake was then compared with the progression as assessed by histology and immunohistochemistry.

For the 24 patients, 57 PET scans with corresponding histologic investigations were available. Among the 24 patients, 16 patients underwent 1 follow-up investigation, 7 patients underwent 2, and 1 patient underwent 3, leading to a total of 33 follow-up investigations. Among these 33, 20 showed histologically proven malignant progression of the tumor. Eleven tumors progressed from grade II to grade III, 6 from grade III to grade IV, and 3 from grade II to grade IV. In 13 cases, the biopsy did not indicate malignant progression.

The mean percentage increase in ¹¹C-MET uptake between PET studies in the group with histologically proven malignant progression was 54.4% (SD, 45.5%; range, 3.1%−162.2%). In contrast, the mean percentage change in ¹¹C-MET uptake in the group without a change in tumor grade was 3.9% (SD, 13.7%; range, −24.4% to 26.3%) (Fig. 2). The mean difference in change in ¹¹C-MET uptake between the group with malignant progression and the group without was highly significant (P < 0.001).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Comparison of value of ¹¹C-MET uptake ratio at time of biopsy vs. percentage change in ¹¹C-MET uptake in distinguishing malignant from nonmalignant progression.

To identify the percentage increase in ¹¹C-MET uptake that best distinguished malignant from nonmalignant progression, we performed a receiver-operating-characteristic analysis (Fig. 3) and calculated the sensitivity and specificity for each value. The percentage change in ¹¹C-MET uptake that best determined malignant progression was 14.6% (sensitivity, 90%; specificity, 92.3%). The area under the curve was 0.96. At a threshold of 14.6%, we identified in only 2 of 20 cases (10%) false-negative ¹¹C-MET uptake findings of malignant progression. In both these tumors, the percentage increase in ¹¹C-MET uptake was less than 14.6% (3.1% and 9.0%, respectively). Malignant progression in both these cases was from grade III to grade IV. We had only 1 false-positive ¹¹C-MET uptake finding; uptake increased by 26.3% whereas the tumor remained histologically stable at grade II. For the 14.6% threshold, positive predictive value was 94.7% and negative predictive value was 85.7%. In the 33-fold cross-validation using the leave-one-out strategy, we correctly classified 30 of 33 follow-up investigations; 1 observation was false-positive and 2 were false-negative, using the 14.6% threshold for change in ¹¹C-MET uptake.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Receiver operating characteristic analysis to identify change in ¹¹C-MET uptake for differentiation between malignant progression of tumor grade and no malignant progression. Percentage increase that best distinguished malignant progression from no malignant progression was at threshold of 14.6%, with sensitivity of 90% and specificity of 92.3%.

The increase in ¹¹C-MET uptake clearly differed between progression from grade II to grade III and progression from grade III to grade IV, being 72.9% (SD, 46.2%) and 33.5% (SD, 35.4%), respectively (Table 2). The change in uptake for progression from grade II to IV was heterogeneous: 77% in 1 case but only 18.5% and 18.9% in the other 2 cases.

We further analyzed the influence of baseline ¹¹C-MET uptake on the time to tumor progression but found no interrelationship between these parameters (P = 0.96). Moreover, we compared absolute ¹¹C-MET uptake ratios at follow-up with changes in ¹¹C-MET uptake for their potential in distinguishing between malignant and nonmalignant progression: for changes in ¹¹C-MET uptake in patients with malignant progression (38.2%−83.3%) and without malignant progression (−13.2% to 6.6%), the 95% confidence intervals were separated, whereas for absolute ¹¹C-MET uptake ratios at follow-up, the 95% confidence intervals overlapped (1.5–2.7 uptake ratio for malignant progression and 2.4–3.1 uptake ratio for stable disease), suggesting that relative changes between uptake ratios more sensitively distinguish malignant progression from stable disease (Fig. 2).

Changes in Contrast Enhancement on MRI or CT During Malignant Progression

For 48 of 57 PET investigations, corresponding MRI (n = 36) or CT (n = 12) had been performed within 6 d beforehand. Using a yes/no categorization, we analyzed for the presence or absence of changes in tumor accumulation of contrast agent in 28 PET-corresponding follow-up investigations. We compared regions of contrast enhancement on the MRI or CT with regions of increased ¹¹C-MET uptake (Table 3).

The MRI or CT measurements were obtained in 17 cases of histologically proven malignant progression and in 11 cases of no progression. We did not observe coherence between newly appearing contrast enhancement on MRI and an increase in ¹¹C-MET uptake and histologic malignant progression. In 8 of 17 cases with histologically proven malignant progression, tumor progression could be detected by newly appearing contrast enhancement on MRI or CT. In 6 of 17 cases, we observed contrast enhancement at baseline as well as at follow-up. In 3 of 17 cases, we observed contrast enhancement neither at baseline nor at follow-up.

In 7 of 11 cases without histologically proven malignant progression, we observed contrast enhancement on baseline MRI or CT as well as at follow-up. In 1 of 11 cases, MRI and CT showed contrast enhancement neither in the prior investigation nor in the subsequent investigation. Three of 11 patients showed newly appearing contrast enhancement on MRI without histologic signs of malignant progression.

Correlation of Molecular Changes on Immunostaining with Metabolic Changes on ¹¹C-MET PET

Histologic analysis was performed on the 52 biopsy samples that corresponded to the PET investigation. All 52 samples were immunostained for Ki-67, VEGF, epidermal growth factor receptor, p53, PTEN, platelet-derived growth factor receptor, and pRb. To avoid an artificial reduction of variance caused by mixing independent and dependent observations, we used only the data of the first follow-up investigations in determining correlations between changes in ¹¹C-MET uptake and changes in the expression of molecular markers. Changes in the expression level of VEGF correlated to changes in ¹¹C-MET uptake (r = 0.62, P = 0.005), indicating that an increase in ¹¹C-MET uptake was related to tumor angiogenesis (Fig. 4). No significant correlation was observed between changes in ¹¹C-MET uptake and changes in the expression pattern of Ki-67, epidermal growth factor receptor, p53, PTEN, platelet-derived growth factor receptor, or pRb.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Correlation between changes in expression of VEGF and changes in ¹¹C-MET uptake during first follow-up investigation.