Visual Abstract
Abstract
Accurate differentiation between tumor progression (TP) and pseudoprogression remains a critical unmet need in neurooncology. 18F-fluciclovine is a widely available synthetic amino acid PET radiotracer. In this study, we aimed to assess the value of 18F-fluciclovine PET for differentiating pseudoprogression from TP in a prospective cohort of patients with suspected radiographic recurrence of glioblastoma. Methods: We enrolled 30 glioblastoma patients with radiographic progression after first-line chemoradiotherapy for whom surgical resection was planned. The patients underwent preoperative 18F-fluciclovine PET and MRI. The relative percentages of viable tumor and therapy-related changes observed in histopathology were quantified and categorized as TP (≥50% viable tumor), mixed TP (<50% and >10% viable tumor), or pseudoprogression (≤10% viable tumor). Results: Eighteen patients had TP, 4 had mixed TP, and 8 had pseudoprogression. Patients with TP/mixed TP had a significantly higher 40- to 50-min SUVmax (6.64 + 1.88 vs. 4.11 ± 1.52, P = 0.009) than patients with pseudoprogression. A 40- to 50-min SUVmax cutoff of 4.66 provided 90% sensitivity and 83% specificity for differentiation of TP/mixed TP from pseudoprogression (area under the curve [AUC], 0.86). A maximum relative cerebral blood volume cutoff of 3.672 provided 90% sensitivity and 71% specificity for differentiation of TP/mixed TP from pseudoprogression (AUC, 0.779). Combining a 40- to 50-min SUVmax cutoff of 4.66 and a maximum relative cerebral blood volume of 3.67 on MRI provided 100% sensitivity and 80% specificity for differentiating TP/mixed TP from pseudoprogression (AUC, 0.95). Conclusion: 18F-fluciclovine PET uptake can accurately differentiate pseudoprogression from TP in glioblastoma, with even greater accuracy when combined with multiparametric MRI. Given the wide availability of 18F-fluciclovine, larger, multicenter studies are warranted to determine whether amino acid PET with 18F-fluciclovine should be used in the routine posttreatment assessment of glioblastoma.
Glioblastoma is the most common malignant primary brain tumor in adults and remains incurable after standard temozolomide-based chemoradiotherapy (1). Accurate assessment of tumor response and progression after treatment remains a significant clinical challenge in neurooncology, complicating both routine care and the conduct of clinical trials (2). Contrast-enhanced MRI, the current standard for tumor monitoring, lacks specificity for detecting neoplastic progression in the brain. This is largely because an increase in contrast enhancement can be due to any cause of blood–brain barrier breakdown and resultant contrast extravasation, including secondary to effective chemoradiotherapy (3). This phenomenon, referred to as pseudoprogression (4), has been reported in up to 30% of patients after chemoradiotherapy as defined by the Response Assessment in Neuro-Oncology Working Group (i.e., within 12 wk after completion of radiotherapy) (5); however, late pseudoprogression may also occur beyond 12 wk (6). Radiographic changes that occur 6 mo to several years after treatment, namely radiation necrosis (7), also share common pathophysiologic features with pseudoprogression (7).
In posttreatment glioblastoma, accurate differentiation of pseudoprogression from true tumor progression (TP) represents a significant unmet clinical need, as erroneous interpretation can lead to premature discontinuation of an effective treatment or overestimation of the efficacy of subsequent salvage therapies (8). In addition, early recognition of TP offers the possibility for earlier therapeutic interventions, such as reresection or recruitment to experimental clinical trials, at a time in the disease course when patients are healthier overall and with relatively preserved performance status. Although advanced MRI techniques such as dynamic susceptibility contrast (DSC), dynamic contrast-enhanced (DCE), and diffusion-weighted imaging have improved the ability to differentiate pseudoprogression from TP, the application of these techniques in both routine practice and clinical trials has been hampered by considerable variability in acquisition and analysis approaches between institutions (9). In addition, these techniques have imperfect accuracy and are frequently affected by imaging artifacts, especially in the posttreatment setting (10).
Metabolic imaging can provide additional valuable information about tumor status. In particular, PET with amino acid tracers, including 11C-methionine (11), 18F-fluorodopa (12), and O-(2-18F-fluoroethyl)-l-tyrosine (18F-FET) (13,14), have shown value in differentiating pseudoprogression from TP, as amino acid uptake is increased in tumor tissue but low in areas of treatment-related change. Recent expert reviews by the Response Assessment in Neuro-Oncology Working Group found amino acid tracers (11C-methionine, 18F-FET, and 18F-fluorodopa) to have higher diagnostic accuracy than conventional and advanced MRI in the differentiation of glioma recurrence (15,16).
18F-fluciclovine is a synthetic amino acid PET tracer approved by the U.S. Food and Drug Administration in the setting of biochemical recurrence of prostate cancer and has an excellent safety profile (17). 18F-fluciclovine is an isoleucine analog that is transported into glial cells by both L-amino acid transporters (especially subtype 1) and alanine–serine–cysteine transporters (especially subtype 2), which are upregulated in glioma cells and demonstrate low expression in the normal brain parenchyma (18). 18F-fluciclovine has also demonstrated utility for discrimination between low-grade and high-grade glioma (19–22). In comparison with 11C-methionine in both treatment-naïve and recurrent glioma, 18F-fluciclovine demonstrated similar detection accuracy but better contrast between tumor and background uptake (19). In contrast to 11C-methionine, which has a short half-life (20 min) and is limited to PET centers with a cyclotron (19), 18F-fluciclovine has the same longer half-life (109.8 min) as 18F-FET and 18F-fluorodopa, which allows time for it to be shipped from the manufacturer to the imaging centers. In addition, unlike 18F-FET and 18F-fluorodopa, 18F-fluciclovine is widely available in the United States, given its use in the setting of prostate cancer (23). Our group previously demonstrated the feasibility of 18F-fluciclovine PET/MRI–guided biopsy in posttreatment glioblastoma to distinguish areas of highest tumor recurrence from areas of treatment-related changes in a small case series (24). The aim of this study was to assess the independent and additive value of 18F-fluciclovine PET and multiparametric MRI for differentiating pseudoprogression from TP. Importantly, we used a study cohort in which all patients had available resected tumor tissue to serve as ground truth for TP versus pseudoprogression.
MATERIALS AND METHODS
Study Design and Patient Population
We conducted a prospective cohort study (NCT03990285) of 30 adults with a previously confirmed diagnosis of glioblastoma (defined according to the World Health Organization 2021 classification) (25) who were previously treated with standard-of-care radiation and temozolomide and scheduled to have surgery based on radiographic progression (i.e., new contrast-enhancing lesions or lesions showing a ≥25% increase in the sum of the products of the perpendicular diameters on MRI) according to the criteria of the Response Assessment in Neuro-Oncology Working Group (26). The patients underwent preoperative multiparametric MRI and 60 min of dynamic brain PET/CT after intravenous administration of 18F-fluciclovine and subsequently underwent maximal safe resection of the enhancing lesion. The percentages of viable tumor and therapy-related changes comprising the specimen were quantified on histopathology by a board-certified neuropathologist. All patients provided written informed consent. This study was approved by the Institutional Review Board of the University of Pennsylvania.
Image Acquisition
PET
18F-fluciclovine (Axumin; Blue Earth Diagnostics) was produced by PETNET Solutions facilities under U.S. Pharmacopeia–compliant procedures and was administered under an investigational-new-drug application exemption. All PET studies were performed on an Ingenuity TF PET/CT device (Philips Healthcare) using a previously described method of image reconstruction (27). Patients underwent 60 min of dynamic PET imaging after injection of 191 ± 21 MBq of 18F-fluciclovine. In 2 patients, PET acquisition was performed for 40 and 45 min instead of 60 min. Patients fasted for 4 h before administration of 18F-fluciclovine. The patients were monitored for adverse events for 24 h after each injection of the 18F-fluciclovine radiotracer, and none were observed.
MRI
Brain MRI was performed using the brain tumor imaging protocol of the University of Pennsylvania on a 3-T magnet (Trio; Siemens), which included an axial T1-weighted 3-dimensional magnetization-prepared rapid gradient echo sequence before and after contrast administration, a postcontrast axial fluid-attenuated inversion recovery sequence, a diffusion tensor imaging sequence (n = 30), a DCE perfusion sequence (n = 29), and a DSC perfusion sequence (n = 27 after excluding 2 patients with degraded source data due to susceptibility effects). Representative imaging parameters are presented in Supplemental Table 1 (supplemental materials are available at http://jnm.snmjournals.org). Two contrast boluses (gadoterate meglumine; Guerbet) with a dose of 0.1 mmol/kg were sequentially administered for DCE followed by DSC imaging, with the dose administered for DCE serving as a preload dose for DSC to reduce the effect of contrast agent leakage on relative cerebral blood volume (rCBV) measurements.
Image Analysis
PET
PET measurements were performed in MIM (the versions were updated throughout the study from 6.9 to 7.1). Volumes of interest were defined in the resected tumor using PET images and coregistered to T1 postcontrast MR images and fluid-attenuated inversion recovery images, with placement confirmed by a board-certified neuroradiologist and a nuclear radiologist. Measurements were taken of tumor SUVmax, 1 cm3 SUVpeak, and 50% threshold SUVmean defined on summed images at 20–30, 30–40, 40–50, 50–60, 40–60, and 30–60 min after injection (g/mL units). Normal-tissue volumes of interest and SUVmean measurements were made in the contralateral normal brain, pituitary gland, and superior sagittal sinus. The volumes of interest in the contralateral normal brain were 20 mm in diameter, whereas those in the pituitary and superior sagittal sinus were 15 mm in diameter. Tumor SUVs and SUV ratios normalized to each normal-tissue SUVmean were calculated.
Time–activity curves of 18F-fluciclovine SUVpeak uptake in the tumor were generated by application, to the entire dynamic dataset, of a spheric volume of interest with a volume of 1 cm3 centered on maximal tumor uptake. Time–activity curves for each lesion were visually assessed by an experienced board-certified nuclear radiologist as previously described (28) and assigned to one of the following curve patterns: constantly increasing without identifiable peak uptake (pattern I), peaking at a midpoint followed by a plateau or a small descent (pattern II), and peaking early followed by a constant descent (pattern III) (28). In addition, time to peak (TTP) with a lower threshold time of at least 10 min was measured and compared among groups.
MRI
Regions of abnormal contrast enhancement, necrosis, and nonenhancing fluid-attenuated inversion recovery signal intensity were segmented using a semiautomated segmentation tool (ITK-SNAP) (29) followed by manual editing by 2 board-certified neuroradiologists. Diffusion tensor imaging was processed with FSL (30) and included removal of nonbrain tissue as well as correction for motion and eddy currents. Diffusion data were then fit to the tensor model, and maps of whole-brain apparent diffusion coefficients were used in subsequent analysis.
DCE images were corrected for motion, and nonbrain tissue was removed using image processing tools available in FSL. DCE data were then analyzed using the extended Toft model, as implemented by the ROCKETSHIP software package (31) in the MATLAB programming environment (2014a; MathWorks), to derive voxelwise maps of the volume transfer constant Ktrans, plasma volume fraction Vp, extravascular extracellular volume fraction Ve, and washout rate constant of contrast agent from the EES to the intravascular space Kep defined as Kep = Ktrans/Ve. Because of the inconsistent availability of T1 mapping among subjects, a fixed prebolus T1 value (1,000 ms) was used to transform signal intensity curves to contrast concentration curves in DCE analysis. DSC data were used to generate leakage‐corrected rCBV maps using the γ‐variate function as implemented in NordicICE software (NordicNeuroLab). For each functional modality (diffusion tensor imaging, DCE, and DSC), a reference image derived from source data was used to compute a linear transformation from the functional space to the subject’s T1 postcontrast magnetization-prepared rapid gradient echo sequence using the Advanced Normalization Tools registration tool (32). These transformations were then used to coregister all parameter maps to the anatomic space. DCE and DSC imaging metrics were normalized to the median value of a region of interest placed in normal-appearing white matter by a board-certified neuroradiologist at the same slice levels as the abnormality. Subsequent statistical analysis was based on the mean rCBV and maximum rCBV (rCBVmax) and DCE metric values as well as the mean and minimum apparent diffusion coefficients extracted from the intersection of the segmented contrast-enhanced and PET ROI.
Histopathologic Evaluation and Analysis of Molecular Markers
After resection, the surgically extracted tissue specimens were entirely fixed in 10% buffered formalin, routinely processed, and embedded in paraffin. Five-micrometer-thick sections of each specimen were cut onto glass slides, stained with hematoxylin and eosin, and assessed by a board-certified neuropathologist (who was unaware of the preoperative MRI and PET imaging data). The presence or absence of pseudopalisading necrosis and microvascular proliferation (features of recurrent glioblastoma), the presence or absence of dystrophic calcification and vascular hyalinization, and the percentage of geographic necrosis (representative of treatment-related changes) were quantified. Proliferative activity was determined by quantification of the number of mitotic figures in 10 high-power fields and semiquantitative assessment of the Ki-67 proliferative index by immunostaining (mouse monoclonal, MIB-1, IR62661; Dako). On the basis of the combined assessment of these features, the entire resected specimen was assigned a tumor percentage of 0%–100%. Patients were considered TP if viable tumor represented at least 50% of the resected specimen, mixed TP if less than 50% and more than 10%, and pseudoprogression if no more than 10%.
Statistical Methods
Given that the number of PET variables (6 summed images and normalized to normal contralateral brain, pituitary, and superior sagittal sinus) and MRI variables was higher than the number of subjects, the least absolute shrinkage and selection operator (LASSO) was used to determine the variables most predictive of viable tumor percentage on histopathology. In addition to imaging variables, the clinical variables of age, sex, O-6-methylguanine methyltransferase (MGMT) promoter methylation status, and duration from the end of first-line radiotherapy to recurrent surgery were included in the LASSO analysis. The strengths of correlations between primary outcome and selected variables were assessed by Pearson correlation or point-biserial correlation (rpb). A receiver-operating-characteristic curve was used to illustrate the diagnostic ability of a binary classifier system as its discrimination threshold was varied. The criteria for determination of the most appropriate cutoff was based on the point on the curve at a minimum distance from the left upper corner of the unit square. Furthermore, the leave-one-out cross-validation procedure was used to estimate the performance of the LASSO regression model by making predictions on test data. The differences in PET uptake between different groups (TP/mixed TP vs. pseudoprogression and TP vs. pseudoprogression) were compared using the Wilcoxon rank-sum exact test. The Kruskal–Wallis test was used to compare time to peak (TTP) between different groups. The χ2 test was performed to compare MGMT methylation status in patients with TP compared with pseudoprogression. An α value of 0.05 was used as the cutoff for significance. All the statistical analyses were computed using code written in R, version 4.1.0 (R Foundation).
RESULTS
The baseline characteristics of the study cohort are summarized in Table 1. Histopathologic analysis revealed 18 patients with TP, 4 with mixed TP, and 8 with pseudoprogression. Tumor percentage ranged from 0% to 100% (median, 57.50%; SD, 31.66), and Ki-67 ranged from 1% to 70% (median, 10%; SD, 13.38). Of patients with TP, 8 of 18 (44%) had MGMT methylated tumors. Of patients with pseudoprogression, 1 of 8 (13%) had MGMT methylated tumor. A χ2 test revealed no statistically significant difference in the rate of MGMT methylation between patients with TP and those with pseudoprogression (P = 0.29). All patients in this study (100%) had isocitrate dehydrogenase wild-type tumors. A detailed description of demographics, clinical symptoms at the time of radiographic progression, MGMT and isocitrate dehydrogenase status, and tumor percentage on histopathology is provided in Supplemental Table 2).
Demographics and Baseline Clinical Characteristics of Patients (n = 30)
Correlation of 18F-Fluciclovine PET and MRI Parameters with Histopathologic and Clinical Variables
A 50- to 60-min 50% threshold SUVmean (r = 0.54, P = 0.004), 50- to 60-min 50% threshold SUVmean/superior sagittal sinus (r = 0.55, P = 0.003), and a 40- to 50-min SUVmax/pituitary (r = 0.51, P = 0.008) correlated positively with viable tumor percentage on histopathology. Among the MRI parameters, only rCBVmax was selected by the LASSO analysis and had a positive correlation with tumor percentage (r = 0.49; P = 0.011). The 40- to 50-min SUVpeak correlated positively with Ki-67 (r = 0.38), with a trend toward significance (P = 0.056). There was no correlation between tumor percentage and age, sex, MGMT promoter methylation status, time elapsed between end of radiation and the patient’s reoperation, or prior radiotherapy dose.
Differentiation of TP/Mixed TP from Pseudoprogression
Analysis of 25 subjects who had all the PET and MRI parameters (Table 2) demonstrated that patients with histopathologically proven TP/mixed TP had a higher 40- to 50-min SUVmax (odds ratio [OR], 1.14; rpb, 0.49; P = 0.011) and 40- to 50-min SUVmax/pituitary (OR, 1.43; rpb, 0.56; P = 0.003). None of the PET variables normalized to normal brain were selected by the LASSO analysis. Among the MRI parameters, rCBVmax (OR, 1.13; rpb, 0.47, P = 0.016) was higher in the TP/mixed TP groups than in the pseudoprogression group. Other MRI parameters were not selected by the LASSO analysis.
Analysis of PET and MRI Parameters for Differentiation of TP/Mixed TP from Pseudoprogression
Analysis of 28 subjects with an available 0- to 60-min dynamic acquisition demonstrated that the 50- to 60-min 50% threshold SUVmean (OR, 1.31; rpb, 0.52, P = 0.004), 40- to 50-min SUVmax (OR, 1.20; rpb, 0.50, P = 0.005), 40- to 50-min SUVmax/pituitary (OR, 1.01; rpb, 0.47, P = 0.010), and 20- to 30-min SUVmax (OR, 1.12; rpb, 0.49, P = 0.007) were higher in the TP/mixed TP group than in the pseudoprogression group.
Patients who demonstrated TP/mixed TP had a significantly higher 40- to 50-min SUVmax (6.64 ± 1.88 vs. 4.11 ± 1.52, P = 0.009) and 20- to 30-min SUVmax (6.59 ± 2.15 vs. 3.89 ± 1.30, P = 0.002) than did patients with histologic pseudoprogression (Fig. 1). An illustrative case in which 18F-fluciclovine PET uptake correctly predicted TP, whereas rCBV on DSC perfusion MRI did not, is displayed in Figure 2.
Comparison of PET parameters in patients with TP/mixed TP vs. pseudoprogression. PsP = pseudoprogression.
Example of false-negative MRI results. In postcontrast T1 (A) and fused PET/MRI (B), a 65-y-old man with history of right occipital glioblastoma showed progressive enhancement adjacent to resection cavity. (C and D) 18F-fluciclovine PET imaging (C) demonstrated marked increased uptake (SUVmax, 5.46) compared with only mild increase (rCBV, 2.43) in rCBVmap on DSC perfusion MRI (D). Patient underwent resection, and pathology showed that 60% of specimen consisted of viable tumor and 40% consisted of therapy-related changes. The intensity scale bar represents SUV.
Although the main purpose of this work was to differentiate patients with TP/mixed TP from those with pseudoprogression, an exploratory analysis was also performed to differentiate TP from pseudoprogression (Supplemental Results; Supplemental Fig. 1; Supplemental Table 3).
Receiver-Operating-Characteristic Curve Analyses
A 20- to 30-min SUVmax cutoff of 4.457 provided 95% sensitivity and 83% specificity for differentiation of TP/mixed TP from pseudoprogression (area under the curve [AUC], 0.902). A 40- to 50-min SUVmax cutoff of 4.662 provided 90% sensitivity and 83% specificity for differentiation of TP/mixed TP from pseudoprogression (AUC, 0.856). An rCBVmax cutoff of 3.672 provided 90% sensitivity and 71% specificity for differentiation of TP/mixed TP from pseudoprogression (AUC, 0.779). Combining a 40- to 50-min SUVmax cutoff of 4.662 and an rCBVmax cutoff of 3.672 provided 100% sensitivity and 80% specificity for differentiating TP/mixed TP from pseudoprogression (AUC, 0.95; Fig. 3). A similar AUC of 0.95 was obtained after combining a 20- to 30-min SUVmax cutoff of 4.457 and an rCBVmax cutoff of 3.672. Combining a 40- to 50-min SUVmax cutoff of 4.662 and an rCBVmax cutoff of 3.672 applying leave-one-out cross-validation provided 100% sensitivity and 80% specificity for differentiation of TP/mixed TP from pseudoprogression (AUC, 0.800; Supplemental Fig. 2).
Receiver-operating-characteristic curve analysis of 20- to 30-min SUVmax (A), 40- to 50-min SUVmax (B), rCBVmax (C), and combined 40- to 50-min SUVmax–rCBVmax (D) to differentiate TP/mixed TP from pseudoprogression.
PET Tracer Kinetics
The time–activity curve demonstrated accumulation at the tumor bed that reached a steady state after 20 min (Fig. 4A). All patients except two demonstrated a type II (plateau) pattern (Fig. 4B). One patient with TP and extracranial extension of tumor to the overlying scalp demonstrated a type III pattern. One patient with TP demonstrated a type I pattern. The TTP did not differ (P = 0.830) between groups (Fig. 4C). Representative patients from the TP, TP/mixed TP, and pseudoprogression groups are demonstrated in Figure 5.
(A) Comparison of mean time–activity curves in patients with TP, mixed TP, and pseudoprogression demonstrated accumulation at tumor bed that reached steady state after 20 min. (B) Time–activity curves in individual patients. All patients except two demonstrated type II (plateau) pattern. One patient with TP and extracranial extension of tumor (black arrow) to overlying scalp demonstrated type III pattern, and 1 patient with TP demonstrated type I pattern (blue arrow). (C) Comparison of mean TTP in patients with TP, mixed TP, and pseudoprogression. PsP = pseudoprogression.
Imaging and histopathologic findings in 3 patients with true progression (A), mixed true progression–pseudoprogression (B), and pseudoprogression (C). (A) A 69-y-old woman with history of left temporoparietal glioblastoma with progressive enhancement in resection bed. 18F-fluciclovine PET imaging demonstrated marked increased uptake (SUVmax, 6.73). Patient underwent resection, and histopathology showed that most resected tissue comprised atypical glial cells, with enlarged, irregular, hyperchromatic nuclei consistent with recurrent glioma (overall, 80% tumor). (B) A 64-y-old woman with history of right temporal glioblastoma with progressive enhancement in resection bed. 18F-fluciclovine PET imaging demonstrated moderate increased uptake (SUVmax, 4.75). Histopathology demonstrated treatment-related changes and 40% infiltrating glioma throughout specimen. (C) A 65-y-old man with history of left temporal glioblastoma with progressive enhancement in resection bed. 18F-fluciclovine PET imaging demonstrated only mild uptake (SUVmax, 2.65). Histopathology demonstrated abundant treatment-related changes with no tumor in specimen. The intensity scale bar represents SUV.
DISCUSSION
In this study, we demonstrated that 18F-fluciclovine PET can accurately differentiate pseudoprogression from TP/mixed TP in glioblastoma patients after chemoradiotherapy. The accuracy of 18F-fluciclovine PET in our study is in the higher range of prior studies using 18F-FET and 18F-fluorodopa and higher than with 11C-methionine PET (13,14,16,28,33–41). In addition, we demonstrated that 18F-fluciclovine PET has higher accuracy than advanced MRI sequences and that combining 18F-fluciclovine PET with DSC perfusion MRI resulted in even better performance for differentiation between TP/mixed TP and pseudoprogression. A recent study of 21 patients with suspected recurrent high-grade glioma who received 18F-fluciclovine PET imaging (42) demonstrated a high median lesion-to-background ratio (42); however, no patients with confirmed pseudoprogression were included. To the best of our knowledge, our study is the first report that demonstrates the ability of 18F-fluciclovine PET to discriminate between TP and pseudoprogression. We also found a positive correlation between Ki-67 and PET parameters, which is consistent with a previous 18F-fluciclovine study on biopsy-proven low- and high-grade gliomas (20) and a report of 18F-fluciclovine PET/MRI–guided biopsy in a patient with treatment-naïve oligodendroglioma (21).
Although multiparametric MRI including postcontrast imaging is widely used for posttreatment surveillance of glioblastoma, differentiation of pseudoprogression from TP using MRI is challenging because both of these entities may disrupt the blood–brain barrier and lead to contrast extravasation and enhancement (43). Amino acid PET imaging enables analysis of the tumor environment beyond disruption of the blood–brain barrier because it is based on upregulation of amino acid transporters in the tumor cells (independent of blood–brain barrier disruption) (44–46). In addition, chronic blood products in the surgical bed cause susceptibility artifacts and interfere with advanced MRI techniques such as DSC perfusion imaging, frequently confounding studies in posttreatment glioblastoma patients. In our study, DSC perfusion images were nondiagnostic in 2 patients and had to be excluded. Multiple studies compared the diagnostic accuracy of 18F-FET PET and DSC perfusion MRI, and the results ranged from superior performance for PET to equal performance for both modalities (34,47–49). PET and DSC perfusion imaging were found to have added value in some studies (34,48,49), but these consisted of heterogeneous cohorts with grades 2–4 glioma, and the final diagnosis was based on histology in only a subset of patients. Our results showed that the accuracy of DSC MRI was comparable to that observed in prior studies, and multiparametric analysis of PET and MRI provided improved accuracy in our study, consistent with prior 18F-FET studies (34,47,49).
Unlike other amino acid PET agents, we found that absolute SUV measurements are more accurate for differentiating pseudoprogression from TP than are normalized tumor-to-brain ratios, secondary to very low normal-brain uptake and variability of normal-brain uptake between patients. Conversely, normalization of SUVmax to the pituitary gland showed similar accuracy to SUVmax in differentiation of TP/mixed TP from pseudoprogression. Previous studies with 18F-FET demonstrated the usefulness of a time–activity curve with curve pattern II or III and a TTP of less than 45 min for differentiating TP from treatment-related changes (34,35,50). In our study, we did not find TTP and time–activity curve patterns to be useful because almost all patients demonstrated a type II (plateau) pattern, consistent with a prior study on pretreatment glioma (17). One patient with extracranial extension of tumor to the overlying scalp demonstrated a type III pattern (uptake peaking early followed by a constant descent), similar to the described literature on prostate cancer (51).
Our study had limitations, including its single-institution nature, which warrants future multicenter prospective studies to validate the generalizability of our findings. However, the prospective design, availability of histopathologic confirmation for all patients, and short interval between 18F-fluciclovine PET and surgery are unique strengths compared with most prior studies of amino acid PET tracers, which have been largely retrospective and have relied on clinical follow-up to distinguish TP from pseudoprogression in most patients. In this study, all patients underwent dynamic PET imaging over 60 min. Given the 20-min plateau in uptake and the high accuracy of both 20- to 30-min and 40- to 50-min SUVmax, static imaging with both these imaging windows can be used in clinical practice, depending on the preference of the center. Given the significant uptake when compared with normal brain parenchyma in all patients (Fig. 5), quantitative criteria and not visual analysis alone is needed to differentiate between TP and pseudoprogression. Overall, our results suggest that 18F-fluciclovine PET imaging can accurately differentiate pseudoprogression from TP in glioblastoma patients. Given the wide availability of 18F-fluciclovine, studies that are larger and multicenter are warranted to determine whether amino acid PET imaging with 18F-fluciclovine should be used in the routine posttreatment assessment of glioblastoma.
CONCLUSION
18F-fluciclovine PET uptake can accurately differentiate pseudoprogression from TP in glioblastoma, with even greater accuracy when combined with multiparametric MRI. Given the wide availability of 18F-fluciclovine, larger, multicenter studies are warranted to determine whether amino acid PET with 18F-fluciclovine should be used in the routine posttreatment assessment of glioblastoma.
DISCLOSURE
Ali Nabavizadeh received research funding (paid to the institution) from Blue Earth Diagnostics and compensation of travel expenses to present the results of this study. No other potential conflict of interest relevant to this article was reported.
KEY POINTS
QUESTION: Can 18F-fluciclovine PET distinguish progression from pseudoprogression in glioblastoma?
PERTINENT FINDINGS: Our prospective study demonstrated that 18F-fluciclovine PET uptake correlated with tumor percentage on histology and can distinguish progression from pseudoprogression in glioblastoma.
IMPLICATIONS FOR PATIENT CARE: Given the wide availability of 18F-fluciclovine in the United States as a Food and Drug Administration–approved radiotracer, this study has immediate translational relevance.
Footnotes
Published online Dec. 22, 2022.
- © 2023 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication August 21, 2022.
- Revision received December 21, 2022.
