Abstract
Purpose: Late-stage, unresectable pancreatic ductal adenocarcinoma (PDAC) is largely resistant to chemotherapy and consequently has a very poor 5-year survival rate of <5%. The ability to assess the efficacy of a treatment soon after its initiation would enable rapid switching to potentially more effective therapies if the current treatment is found to be futile. We have evaluated the ability of the PET imaging agent, 89Zr-anti-γH2AX-TAT, to monitor DNA damage in response to fluorouracil (5-FU), gemcitabine, or capecitabine treatment in a mouse model of pancreatic cancer. We have also compared the utility of this approach against the standard clinical PET radiotracer, 18F-FDG.
Experimental Design: C57BL/6 mice bearing subcutaneous pancreatic cancer (KPC; B8484) allografts were treated with 5-FU, gemcitabine, or capecitabine. Therapeutic response was monitored by PET and ex vivo biodistribution experiments using either 89Zr-anti-γH2AX-TAT or 18F-FDG as imaging agents. To further examine the effect of therapeutic response upon uptake of these imaging agents, IHC analysis of harvested tumor allograft tissue was also performed.
Results: Accumulation of 89Zr-anti-γH2AX-TAT in the tumors of mice that received chemotherapy was higher compared with vehicle-treated mice and was shown to be specifically mediated by γH2AX. In contrast, 18F-FDG did not provide useful indications of therapeutic response.
Conclusions: 89Zr-anti-γH2AX-TAT has shown a superior ability to monitor early therapeutic responses to chemotherapy by PET imaging compared with 18F-FDG in an allograft model of PDAC in mice. Clin Cancer Res; 23(21); 6498–504. ©2017 AACR.
Current chemotherapy regimens for pancreatic ductal adenocarcinoma do not appreciably prolong patient survival due to intrinsic or rapidly acquired resistance. To improve survival outcomes, it is therefore critical to determine whether a patient is responding to treatment as early as possible so that a futile treatment can be quickly replaced with one that is more effective. Monitoring response to therapy by PET is attractive, as this imaging technique can detect the early molecular indications of treatment efficacy. The standard clinical PET imaging agent is 18F-FDG, which despite its widespread use has several limitations, including a tendency to accumulate nonspecifically in areas of inflammation and an inability to differentiate focal mass-forming pancreatitis from pancreatic cancer. As most chemotherapy agents are designed to cause DNA damage, we propose that monitoring the DNA damage response with the PET imaging agent 89Zr-anti-γH2AX-TAT would provide a more direct measure of treatment efficacy.
Introduction
At present, less than 5% of patients diagnosed with pancreatic cancer will survive longer than 5 years (1). This poor prognosis is due in part to a lack of screening measures that result in most patients being diagnosed when the disease is in an advanced, metastatic state. Consequently, the opportunity for potentially curative surgical resection has usually been missed, and the patient is typically offered chemo- and/or radiotherapy, or palliative treatments (2). Standard chemotherapy regimens for advanced pancreatic cancer include gemcitabine, fluorouracil (5-FU), capecitabine, Nab-paclitaxel (Abraxane) and FOLFIRINOX. Response to single-agent chemotherapy is typically less than 10%, and response to multiagent chemotherapy (gemcitabine/Nab-paclitaxel and FOLFIRINOX) in the region of 25% to 30% (3–5). The limited efficacy of these treatments with regard to prolonging patient survival is largely due to intrinsic or acquired resistance to these agents (6, 7). The ability, therefore, to monitor the efficacy of a particular chemotherapy early on during treatment would be advantageous as a futile treatment could be rapidly switched to an alternative therapy.
The conventional methods for assessing response to therapy are based on anatomic assessment of tumor size by X-ray, CT, or MRI (8, 9). More recently, molecular imaging techniques, including PET and single-photon emission CT (SPECT), have also been recognized as valuable tools for evaluating treatment response (10–13). These imaging techniques allow visualization and quantitation of cellular and biochemical responses to treatment, which can indicate the efficacy of therapy soon after its initiation and before changes in tumor dimensions can be measured.
The clinical PET imaging agent fluorodeoxyglucose (18F-FDG) accumulates in tumor tissue due to an increased reliance on glycolysis for energy production even under normoxic conditions compared with normal cells (i.e., the Warburg effect; refs. 14–16). Like glucose, 18F-FDG is internalized via glucose transporters (particularly GLUT-1 and GLUT-3) and is then phosphorylated by hexokinase. However, unlike glucose-6-phosphate, the newly formed 18F-FDG-6-phosphate is resistant to subsequent enzymatic metabolism and is retained within the cell, resulting in specific accumulation of the imaging agent within tumors. In addition to its roles in cancer detection and staging, 18F-FDG has also been used successfully for monitoring response to various forms of therapy in many cancer types (17–25), including pancreatic ductal adenocarcinoma (PDAC; refs. 26–28).
Although the use of 18F-FDG for monitoring therapy is showing significant promise, this imaging agent has several well-recognized limitations, particularly relating to pancreatic cancer (29–33). These limitations include the occurrence of nonspecific uptake of 18F-FDG in inflammatory lesions and regions of infection, and an inability to reliably distinguish focal mass-forming pancreatitis from pancreatic cancer.
A more direct strategy for monitoring treatment efficacy involves gauging molecular effects such as the DNA damage response (DDR), which is activated in response to most chemotherapy agents (34–36). At present, in clinical settings, this can be accomplished by performing IHC analysis of biopsied tissues (37). However, recovery of tissue biopsies is an invasive procedure that can pose risks of hemorrhaging and infection. Furthermore, biopsies prevent longitudinal assessment via repeated monitoring of the same region and offer only limited insights into tumor heterogeneity. Therefore, the ability to probe for biomarkers of DDR noninvasively via PET or SPECT imaging is an attractive prospect.
A well-established biomarker of DDR is the phosphorylated histone γH2AX, which forms foci around double-strand breaks (DSB) of DNA (37, 38). γH2AX arises in response to DNA DSBs when members of the PI3K-related protein kinase family (including ATM, ATR, and DNA-PKcs) phosphorylate the X isoform of H2A at the serine-139 position (39, 40). Thereafter, γH2AX mediates the repair of DNA DSBs by recruiting several other DNA repair proteins to the damaged site (41, 42).
Research efforts are currently under way to develop a noninvasive means of quantifying γH2AX expression levels in vivo using both PET and SPECT imaging techniques (43–45). Previously, we showed that the uptake of an antibody-based SPECT imaging agent, 111In-anti-γH2AX-TAT, in MDA-MB-468 xenograft tumors in mice was linearly dependent on the number of γH2AX foci per cell and linearly dependent on radiation deposited dose after γ-irradiation of the tumor (43). Importantly, in cases where γH2AX expression is not increased in response to treatment, uptake of this imaging agent in tumors is not significantly increased, which further supports the specificity of this probe for γH2AX (46). More recently, we have developed an analogous imaging agent containing the PET radioisotope 89Zr, which was successful in detecting elevated levels of γH2AX following radiation-induced DNA damage in breast cancer xenografts in mice (45). In the current study, we sought to investigate whether 89Zr-anti-γH2AX-TAT can be used to monitor the activation of DDR in response to three standard chemotherapy agents in a mouse allograft model of PDAC and have performed a comparison with 18F-FDG.
Materials and Methods
General methods
All reagents were purchased from Sigma-Aldrich unless otherwise stated and were used without further purification. The chelating agent p-SCN-Bn-DFO was purchased from Macrocyclics Inc. Water was deionized using a Barnstead NANOpure purification system (Thermo Fisher Scientific) and had a resistance of >18.2 MΩ/cm at 25°C. Protein concentration measurements were made on a ND-1000 spectrophotometer (NanoDrop Technologies, Inc.). Instant thin-layer chromatography (iTLC) was performed on glass microfiber chromatography paper (Agilent Technologies) and strips were analyzed with either a Bioscan AR-2000 radio-TLC scanner (Eckert & Ziegler) or a Cyclone Plus Phosphor Imager (PerkinElmer). pH was determined using pH indicator paper (Merck Millipore). Radioactivity measurements were made using a CRC-25R dose calibrator (Capintec, Inc.).
Cell culture
KPC cells (B8484) were derived from KrasLSL-G12D/+; Trp53LSL-R172H/+; Pdx1Cre (KPC) tumors. Cells were maintained in DMEM, supplemented with 10% FBS, 2 mmol/L l-glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin. Cells were grown in a 37°C environment containing 5% CO2 and were harvested and passaged as required using Trypsin-EDTA solution. The cumulative length of culture was less than 6 months following retrieval from liquid nitrogen storage.
Preparation of 89Zr-anti-γH2AX-TAT
Anti-γH2AX (DR1017, Merck Millipore) was modified with the cell-penetrating peptide TAT (GRKKRRQRRRPPQGYG) by a previously described method (43). Modification of anti-γH2AX-TAT with p-SCN-Bn-DFO and subsequent radiolabeling with zirconium-89 were conducted using previously reported methods (47). In brief, to a solution of γH2AX-TAT (200 μg) or rabbit IgG (200 μg, I5006, Sigma Aldrich) in 0.1 mol/L NaHCO3 (pH 8.9, 125 μL, Chelex treated) was added 10 molar equivalents of p-SCN-Bn-DFO (6.64 mmol/L) in anhydrous DMSO. The reaction mixture was incubated at 37°C for 60 minutes with gentle shaking (450 rpm), and the excess p-SCN-Bn-DFO was removed by Sephadex-G50 size exclusion chromatography.
Zirconium-89 in 1 mol/L oxalic acid (sourced from VU Amsterdam) was adjusted to pH 7 by the addition of 1 mol/L sodium carbonate. The resulting solution was added to a 2 mg/mL solution of the DFO-modified antibody to achieve a ratio of 0.1 MBq to 1 μg. The reaction mixture was incubated at room temperature for 1 hour, and the radiolabeling efficiency was determined by iTLC using an eluent of 50 mmol/L ethylenediaminetetraacetic acid (EDTA) buffer (pH 6). The crude reaction mixture was purified by Sephadex-G50 size exclusion chromatography, eluting with 100 μL fractions of PBS (pH 7.4).
89Zr-labeled TAT-modified rabbit IgG (RIgG-TAT) was used as a negative control and synthesized from rabbit IgG as described above for 89Zr-anti-γH2AX-TAT.
In vivo studies
All animal procedures were performed in accordance with the UK Animals (Scientific Procedures) Act 1986 and with local ethical committee approval. Allograft tumors were established in the right hind flank of female C57BL/6 mice by subcutaneous injection of 1 × 106 KPC cells in PBS (100 μL). Tumor volumes (V) were calculated after caliper measurement using the following equation: V = (a2 × b)/2, where a is the width of the tumor and b is the length (small and large diameters, respectively). The individual relative tumor volume (RTV) was defined as Vt/V0, where Vt is the volume at a given time and V0 at the start of treatment.
PET/CT imaging experiments
Ex vivo biodistribution experiments
Mice were euthanized by cervical dislocation and selected organs, tissues, and blood were removed. The samples were immediately rinsed with water, dried, and transferred into a preweighed counting tube. After weighing the filled counting tubes, the amount of radioactivity in each was measured using a 2480 WIZARD2 or 1470 WIZARD gamma counter (PerkinElmer). Counts per minute were converted into radioactivity units (MBq) using calibration curves generated from known standards. These values were decay-corrected to the time of injection, and the percentage of the injected dose per gram (%ID/g) of each sample was calculated.
IHC staining for GLUT-1 and γH2AX
Sections of tumor allograft tissue were obtained at 7-μm thickness using a cryostat (OTF5000, Bright Instruments). Tissue sections were fixed in 4% paraformaldehyde for 10 minutes at room temperature, washed twice in PBS, and then permeabilized in 1% Triton/PBS for 10 minutes at room temperature. After rinsing in PBS, tissue sections were incubated with blocking buffer (2% BSA/0.1% Triton in PBS) for 1 hour at room temperature and then incubated with the relevant primary antibody [anti-GLUT-1 rabbit pAb (1:250 dilution, ab652, Abcam) or anti-γH2AX rabbit pAb (1:800 dilution, DR1017 (gemcitabine and capecitabine experiments) or 07-164 (5-FU experiments), Merck Millipore)] overnight at 4°C. After washing with PBS, tissue sections were then incubated with fluorescently-labeled secondary goat-rabbit IgG-Dylite 488 (1:250 dilution in blocking buffer) for 1 hour at room temperature. After washing with PBS, slides were mounted with VECTASHIELD mounting medium with DAPI (Vector Laboratories). Confocal microscopy images were acquired using a Zeiss 530 microscope (Zeiss). Standardized relative quantification of GLUT-1 and γH2AX immunofluorescence was performed by normalizing the Dylite 488 signal intensity to the DAPI signal.
Statistical analyses
All statistical analyses and nonlinear regression were performed using GraphPad Prism (GraphPad Software). Data were tested for normality and analyzed either by the unpaired, two-tailed Student t test where appropriate, or one-way ANOVA for multiple comparisons, with Dunnet posttests to calculate significance of differences between groups. All data were obtained at least in triplicate and results reported as mean ± SD, unless stated otherwise.
Results
Zirconium-89 radiolabeling of DFO-anti-γH2AX-TAT
89Zr-anti-γH2AX-TAT and 89Zr-RIgG-TAT were routinely synthesized in high radiochemical yields (87.7% ± 13.5% and 85.2% ± 14.5%, respectively) and obtained in excellent purity (>99%) following G50 size exclusion chromatography. See Supplementary Fig. S1 for representative iTLC chromatograms.
Tumor uptake and radiotracer distribution
5-FU therapy.
PET/CT images acquired 3 days after a single dose of 5-FU revealed higher uptake of 89Zr-anti-γH2AX-TAT in tumors of treated mice compared with experimental controls (Fig. 1A). Within this timeframe, the growth rate of KPC allografts in mice treated with a single dose of 5-FU was not significantly impeded compared with vehicle-treated mice (Fig. 1B). Values obtained from ex vivo biodistribution experiments at 24 hours p.i. revealed that uptake of 89Zr-anti-γH2AX-TAT in the tumors of treated mice reached a value of 8.5 ± 1.1 %ID/g (Supplementary Table S1), which was higher than the vehicle-treated mice and mice administered an imaging agent lacking γH2AX specificity, 89Zr-RIgG-TAT (P < 0.01; Supplementary Table S1). These observations correlate well with IHC analysis of the harvested tumor allograft tissues (Fig. 1D and E), as significantly higher expression levels of γH2AX were measured in the tumors of 5-FU–treated mice compared with tumors of vehicle-treated mice (67 ± 19 and 49 ± 21 a.u., respectively; P < 0.05).
Monitoring 5-FU therapy with 89Zr-anti-γH2AX-TAT. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from ex vivo biodistribution experiments. *, P < 0.05. D, Representative confocal microscopy images, 63× (blue, DAPI; green, γH2AX). E, Quantification of γH2AX signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
Monitoring 5-FU therapy with 89Zr-anti-γH2AX-TAT. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from ex vivo biodistribution experiments. *, P < 0.05. D, Representative confocal microscopy images, 63× (blue, DAPI; green, γH2AX). E, Quantification of γH2AX signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
Similar PET/CT imaging experiments with 18F-FDG did not reveal any differences in tumor uptake after initiation of 5-FU treatment either at an early stage (day 3) before any effect upon RTV was observed, or on day 9 when the mean RTV of treated mice was significantly smaller than vehicle-treated mice (Fig. 2A–C). Ex vivo biodistribution analysis performed immediately after the final imaging session on day 9 (Supplementary Table S2) and IHC staining of harvested tumors confirmed these results and showed no significant difference in GLUT-1 expression levels (Fig. 2D and E).
Monitoring 5-FU therapy with 18F-FDG. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from VOI analysis. D, Representative confocal microscopy images, 20× (blue, DAPI; green, γH2AX). E, Quantification of GLUT-1 signal on confocal microscopy images normalized to DAPI signal. Error bars, SEM.
Monitoring 5-FU therapy with 18F-FDG. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from VOI analysis. D, Representative confocal microscopy images, 20× (blue, DAPI; green, γH2AX). E, Quantification of GLUT-1 signal on confocal microscopy images normalized to DAPI signal. Error bars, SEM.
Gemcitabine therapy.
PET/CT images acquired 3 days after a single administration of gemcitabine revealed higher uptake of 89Zr-anti-γH2AX-TAT in the tumors of gemcitabine-treated mice compared with the nonspecific controls (Fig. 3A). Within this timeframe, there was no significant difference in mean RTV compared with a vehicle-treated control cohort of mice (Fig. 3B). Analysis of the ex vivo biodistribution data (Supplementary Table S3) indicated that uptake of 89Zr-anti-γH2AX-TAT in the tumors of gemcitabine-treated mice reached a value of 6.8 ± 0.6 %ID/g and was significantly higher than experimental controls (Fig. 3C; P < 0.01, Supplementary Table S3). Confocal microscopy images obtained from tumor tissue sections revealed higher γH2AX expression levels in the tumors of gemcitabine-treated mice compared with tumors harvested from vehicle-treated mice (67 ± 30 and 40 ± 13 a.u., respectively; P < 0.05; Fig. 3D and E).
Monitoring gemcitabine therapy with 89Zr-anti-γH2AX-TAT. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from ex vivo biodistribution experiments. *, P < 0.05. D, Representative confocal microscopy images, 63× (blue, DAPI; green, γH2AX). E, Quantification of γH2AX signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
Monitoring gemcitabine therapy with 89Zr-anti-γH2AX-TAT. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. C, Tumor uptake values obtained from ex vivo biodistribution experiments. *, P < 0.05. D, Representative confocal microscopy images, 63× (blue, DAPI; green, γH2AX). E, Quantification of γH2AX signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
In a similar fashion to the 5-FU therapy experiments, PET images acquired with 18F-FDG did not reveal any significant difference in tumor uptake between gemcitabine- and vehicle-treated mice at day 3 (before any effect on RTV was observed) or day 8 when a significant difference in RTV between these groups could be measured (Fig. 4A–C). Interestingly, analysis of the ex vivo biodistribution data acquired immediately after the final imaging session on Day 8 (Supplementary Table S4) revealed significantly higher uptake of 18F-FDG in the tumors of gemcitabine-treated mice compared with vehicle-treated mice (7.6 ± 1.5 and 3.3 ± 2.0 %ID/g, respectively; P < 0.05). Conversely, analysis of confocal microscopy images (Fig. 4D and E) indicated that GLUT-1 expression levels were reduced in the tumors of gemcitabine-treated mice (96 ± 40 a.u.) compared with vehicle-treated mice (65 ± 44 a.u.; P < 0.05).
Monitoring gemcitabine therapy with 18F-FDG. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. **, P < 0.01. C, Tumor uptake values obtained from VOI analysis. D, Representative confocal microscopy images, 20× (blue, DAPI; green, γH2AX). E, Quantification of GLUT-1 signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
Monitoring gemcitabine therapy with 18F-FDG. A, PET/CT images showing coronal (top) and transaxial (bottom) sections intersecting the center of the allograft tumor (white dotted circle). B, Tumor growth curve. **, P < 0.01. C, Tumor uptake values obtained from VOI analysis. D, Representative confocal microscopy images, 20× (blue, DAPI; green, γH2AX). E, Quantification of GLUT-1 signal on confocal microscopy images normalized to DAPI signal. *, P < 0.05. Error bars, SEM.
Capecitabine therapy.
The results of the capecitabine experiments with 89Zr-anti-γH2AX-TAT and 18F-FDG, which showed similar trends but did not reach statistical significance, are provided in full in the Supplementary Information.
Discussion
89Zr-anti-γH2AX-TAT was effective in detecting upregulation of γH2AX in each of the applied chemotherapy regimens. Notably, in all cases, the detection of treatment-induced γH2AX upregulation preceded a change in relative tumor volume compared with vehicle-treated control mice. In contrast, VOI analysis of PET images indicated that uptake of 18F-FDG within tumors did not change as a result of any of the administered chemotherapies, either soon after the initiation of therapy or following completion of therapy. This observation is notable, particularly in the case of 5-FU and gemcitabine-treated mice, which experienced significant impediment of tumor growth compared with vehicle-treated mice. It would be reasonable to expect the rate of glucose metabolism to be reduced in the tumors of mice treated with chemotherapy, and indeed, this was partially reflected by a reduction in the expression levels of GLUT-1 in tumors harvested from gemcitabine-treated mice. However, in mice treated with 5-FU or capecitabine, no signification reduction in GLUT-1 expression levels was observed. This supports our contention that compared with measuring changes in glucose metabolism, it is of more value to monitor the effects of DNA-damaging treatments in a more direct manner by tracking the upregulation of key DNA repair proteins, such as γH2AX.
89Zr-anti-γH2AX-TAT is one of a small number of PET imaging agents that have demonstrated the ability to track DDR proteins during cancer therapy in preclinical studies. Other recent examples include radiolabeled small-molecule inhibitors of PARP-1, such as 18F-BO, which showed substantially decreased uptake in A2780 human ovarian cancer xenograft tumors in mice following treatment with olaparib (48). While PARP-1 is principally involved in the repair of single-strand breaks in DNA, γH2AX is mostly upregulated in response to DNA DSBs, which require longer repair times. In principle, targeting more slowly dissipating epitopes, such as γH2AX, may provide an extended time window during which useful PET imaging measurements of therapy response can be obtained.
Although significant differences can be observed following chemotherapy, for each of the chemotherapy regimens, it can be observed that a proportion of the overall uptake of 89Zr-anti-γH2AX-TAT is nonspecific and is consistent with uptake levels expected from the enhanced permeability and retention (EPR) effect (49, 50). This phenomenon is caused by the leaky vasculature within tumors, which causes high molecular weight species, such as 89Zr-anti-γH2AX-TAT, to passively extravasate to tumor tissue. To overcome this limitation, future investigations will be focused on amplifying the proportion of γH2AX-mediated signal by, for example, utilization of lower molecular weight targeting vectors (to diminish the EPR effect; ref. 51), and molecularly targeted CPPs (to improve tumor targeting; ref. 52).
Conclusions
It has been demonstrated that the PET imaging agent 89Zr-anti-γH2AX-TAT is capable of monitoring the induction of γH2AX that occurs as a result of chemotherapy in an allograft model of PDAC in mice. Notably, 89Zr-anti-γH2AX-TAT has shown a superior ability to provide early indications of therapy response compared with the standard clinical PET radiotracer, 18F-FDG. Despite being commonly used to evaluate response to therapy in patients with pancreatic cancer, we found that 18F-FDG did not provide any indication of therapeutic response either soon after initiation of treatment or following its completion. Furthermore, in cases where a reduction of GLUT-1 expression was observed in tumors following treatment, this was not accompanied by a reduction of 18F-FDG uptake, indicating that overall uptake of 18F-FDG was dominated by a nonspecific contribution. In contrast, 89Zr-anti-γH2AX-TAT offers a highly sensitive and more direct means of monitoring response to DNA-damaging therapies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: S. Mukherjee, B. Cornelissen
Development of methodology: J.C. Knight, M.J. Mosley, L.C. Bravo, V. Kersemans, B. Cornelissen
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.C. Knight, M.J. Mosley, L.C. Bravo, V. Kersemans, B. Cornelissen
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.C. Knight, M.J. Mosley, V. Kersemans, P.D. Allen, B. Cornelissen
Writing, review, and/or revision of the manuscript: J.C. Knight, V. Kersemans, S. Mukherjee, E. O'Neill, B. Cornelissen
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.J. Mosley, L.C. Bravo
Study supervision: E. O'Neill, B. Cornelissen
Acknowledgments
We also thank David Y. Lewis, Ferdia A. Gallagher, Deborah Sneddon, and Julia Baguña Torres for helpful discussions in the preparation of this manuscript. Finally, we wish to thank Fergus Gleeson and Paul Murphy at the Oxford University Hospitals NHS Foundation Trust for providing 18F-FDG.
Grant Support
This research was supported by Pancreatic Cancer UK, CRUK through the Oxford Institute for Radiation Oncology and the CRUK Oxford Centre, and the CRUK/EPSRC Imaging Centre in Oxford. S. Mukherjee is partly funded by NIHR Oxford Biomedical Research Centre.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.