The differential effects of metronomic gemcitabine and antiangiogenic treatment in patient-derived xenografts of pancreatic cancer: treatment effects on metabolism, vascular function, cell proliferation, and tumor growth

Fresh pancreatic ductal adenocarcinoma (PDAC) tissues were obtained from consented patients undergoing Whipple resection at Vancouver General Hospital in 2006–2008. Resected, non-diagnostic specimens were collected as previously described [9, 12]. Briefly, viable tumor tissues were implanted subcutaneously into male C.B-17 SCID mice (Taconic, Germantown, NY, USA). When tumor volumes reached 600–800 mm³, the subcutaneous tumors were excised, cut into small pieces, and surgically implanted on the pancreas of additional mice. Two primary pancreatic ductal adenocarcinoma xenograft lines, PaCa8 and PaCa13, established from two patients were used in this study. All animal studies were done in accordance with the guidelines from the Canadian Council for Animal Care and approved by the University of British Columbia’s animal care committee.

Gemcitabine hydrochloride (Gemzar^®; Eli Lilly Canada Inc., Toronto, ON, Canada) was obtained from the British Columbia Cancer Agency (BCCA) pharmacy. Mouse monoclonal VEGFR-2 antibody, DC101, was provided by Imclone Systems (New York, NY, USA) via a materials transfer agreement. When the orthotopic primary xenografts reached a palpable size of 200–400 mm³, tumor-bearing mice were randomly assigned to three groups (n = 8/group) and treated with vehicle control (Veh-ctrl; 0.9 % saline, i.p.), DC101 (800 μg, q3d, i.p.), or metronomic gemcitabine (Met-Gem; 30 mg/kg, q3d, i.p.). Gemcitabine doses were chosen to emulate typical serum concentrations in the clinic as reported previously [9], and DC101 doses were based on previous studies that showed efficacy [13, 14]. On days 3, 7, and 21 following initiation of treatment, mice from each group were scanned with dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) or ¹⁸F-fluorodeoxyglucose positron emission tomography (FDG-PET). Once imaging procedures were complete, mice were euthanized and tumors harvested for immunostaining or molecular interrogation (See Scheme 1 for experimental flow).

At each time point, tumor-bearing mice were injected with FDG (¹⁸F-radiolabeled deoxyglucose; 100 ± 5 μCi; BCCA) via the lateral tail vein. One hour after injection, the mice were anesthetized with isoflurane and placed on the scanning platform. The Inveon multi-modality small animal PET-CT scanner (Siemens, Knoxville, TN, USA) was used for imaging studies. PET data were collected for 15 min in list mode and subsequently histogrammed into a single frame. Images were reconstructed in three dimensions using OSEM-MAP3D algorithms following CT-based attenuation scans to correct for the animal’s body mass. Three-dimensional regions of interest were placed on parts of the tumor that were actively taking up FDG in the reconstructed animal images to quantify FDG activity present per volume of tumor tissue. FDG-PET imaging is based on detecting the uptake of FDG by cancer cells. The avidity of the cells for FDG, and by extension glucose, is a gauge of the tumor’s proliferative ability and viability [15]. Generally, the more FDG is taken up by a tumor, the more viable and proliferative the tumor is. In the clinic, FDG-PET images are used to stage tumors and diagnose metastatic disease [15, 16]. CT images, based on tissue densities and injected contrast agents, provide information on the physical volume of the tumor. The FDG signal is used to corroborate the CT image of the tumor. Changes in the FDG signal occur more rapidly than changes in tumor volume, so this information can also be used to infer early response to a treatment—i.e., if the FDG signal is reduced by a treatment (compared to a baseline scan before treatment), it indicates that the cells are less viable and proliferative. We have used PET data reported here in a similar fashion to provide additional information on the effects of metronomic gemcitabine and DC101 on the viability and proliferative ability of the tumor before and after treatment.

Following DCE-MRI and FDG-PET, all tumor-bearing mice were injected with the hypoxia marker EF5 (30 mg/kg, i.v.; Dr. Cameron Koch, University of Pennsylvania, PA, USA) and the perfusion marker Hoechst 33342 (40 mg/kg, i.v.; Sigma-Aldrich, Oakville, ON, Canada) at 3 h and 5 min, respectively, before euthanasia and immediate tumor excision. Perpendicular tumor diameters were measured with a caliper, and tumor volumes were calculated (π/6 × a × b², where ‘a’ is the longest dimension of the tumor, and ‘b’ is the width). Tumors were preserved in OCT (Sakura Finetek, Torrance, CA, USA) and snap-frozen in liquid nitrogen vapor. A known value of 10-μm cryosections were cut from each tumor, air-dried, and imaged first for native Hoechst 33362 fluorescence as an indicator of tumor tissue perfusion [17]. Subsequently, the cryosections were fixed in 50 % (v/v) acetone/methanol and blocked in a buffer containing 1 % BSA, 1 % goat serum, 1 % donkey serum, and 0.1 % Tween-20. Endothelial cells were stained with a CD31/PECAM-1 primary antibody (1:1000; BD PharMingen, San Diego, CA, USA) followed by an Alexa 647-conjugated secondary antibody (1:20,000; Invitrogen, Burlington, ON, Canada). Reduced EF5 adducts in viable hypoxic cells were stained using a Cy3-conjugated monoclonal antibody to ELK3-51 (1:400; Dr. C. Koch). To visualize proliferating cells, cryosections were stained with a Ki67 (1:100; Invitrogen) primary antibody followed by an Alexa 488 conjugated secondary antibody (1:200; Invitrogen). The in situ Cell Death Detection Kit (Roche Indianapolis, IN, USA), which involves TdT-mediated dUTP nick end labeling (TUNEL) reaction, was used to evaluate cell apoptosis. Finally, cryosections were counterstained with Hoechst 33362 (10 μg/mL) and mounted. Fluorescent images of whole tumor sections were captured (×10 objective) using a microscope system (Leica Microsystems Inc., Richmond Hill, ON, Canada) equipped with a cooled 350FX monochrome CCD camera, an automated scanning stage (DM6000B), and Surveyor software (Objective Imaging Kansasville, WI, USA). All parameters stained on the same cryosection were imaged separately and then overlaid and aligned. With NIH ImageJ (http://​rsb.​info.​nih.​gov/​ij/​) software and user-supplied algorithms, images were cropped to tumor tissue boundaries with necrosis and artifacts removed before quantification. The extent of hypoxia, proliferation, and apoptosis was represented by the percentage of EF5+, Ki67+, and TUNEL+ pixels, respectively, of the total number of viable tumor tissue pixels. The percentage of perfused vessels was reported as the percentage of CD31 + Hoechst + pixels divided by the total number of CD31 + pixels. To measure microvessel density, all image pixels were sorted to determine their distance to the nearest CD31 + pixel, and the average of distances to the nearest vessel (i.e., CD31 + object) was calculated. A short average distance to the nearest vessel indicates high microvessel density [9, 18]. The degree of necrosis present in the tumors was calculated as a percentage of necrotic pixels divided by total tumor tissue pixels; necrotic areas were identified based on side-by-side comparisons with H&E images of the section.

A 7.0 Tesla MR scanner (Bruker, Ettlingen, Germany), with a quadrature birdcage coil for transmission and a 1.7 × 1.4 cm rectangular surface coil for reception, was used for DCE-MRI studies. At each time point, tumor-bearing mice were anesthetized with isoflurane (Baxter Corporation, Mississauga, ON, Canada) and given an i.v. bolus injection of 0.3 mmol/kg of the MR contrast agent gadodiamide (Omniscan™; Nycomed, Oslo, Norway). 3D-FLASH was used to acquire the data for estimating the concentration of the contrast agent (FOV = 3.84 × 21.6 × 2.4 cm; voxel size = 0.3 × 0.3 × 1 mm): three scans with different flip angles were used to calculate in vivo flip angle maps [19] in order to correct the T₁ estimates (α_nom = 145°, 180°, 215°; TE/TR = 3.5/460 ms, 2 × zero-filling for 0.3 × 0.3 × 1 mm voxel); a three-scan variable flip angle method was used to calculate native T₁ in the tumor [20] (α_nom = 10°, 20°, 50°; TE/TR = 2.7/144 ms); and a rapid T₁-weighted scan series (TE/TR = 2.7/9 ms, α_nom = 25°, 15.6 s per scan) was performed before and after bolus injection of the contrast agent (20 pre-contrast scans, 150 post-contrast scans) to observe the initial uptake and subsequent washout in the tumor. Concentration was derived assuming linearity between contrast concentration and T₁ according to the equations described by Schabel and Parker [21]. The gadodiamide concentration–time curve for each pixel was fitted to the extended Kety model [22, 23] which describes the pharmacokinetics of the contrast agent using three parameters: K ^trans—volume transfer constant between vascular space and extravascular extracellular space; v _e—fractional volume of extravascular extracellular space; and v _p—fractional volume of vascular space. The arterial input function was derived from a population average in the same tumor model as previously described [24]. K ^trans, which is often a mixed measure of blood flow and vascular permeability [25], was determined using in-house software and expressed as median values of K ^trans in viable tumor tissues. To assess K ^trans in the tumor periphery versus tumor core, three-dimensional K ^trans voxel maps were processed by binary erosion [26], which segmented the K ^trans voxel distribution into three-dimensional concentric shells that were each a single voxel thick. Using this segmentation of voxels, K ^trans was averaged from the outer one-third and inner two-thirds of tumor voxels to represent the tumor periphery and the tumor core, respectively.

Magnetic resonance imaging is used extensively in the clinic for high-resolution anatomical imaging [15, 27]. MR anatomical images are based on detecting water molecules under a magnetic field and the differences between water molecules in different environments. In some cases, MRI is also used to examine perfusion or blood flow in a tumor using contrast agents and specialized imaging sequences [27]. We have taken a similar approach in our study, and imaged/quantified the effects of an injected contrast agent as it appears and exits the tumor via the blood vessels. The data are then processed to provide us with an estimate of perfusion rates in the tissue before and after treatment. In this way, we are able to associate the overall vascular function of the tumor following Met-Gem and DC101 treatment.

Harvested tumors from the three treatment groups at each time point were snap-frozen in liquid nitrogen and homogenized in a lysis buffer (50 mM HEPES, 10 % glycerol, 1 % Triton X-100, 150 mM NaCl, 1 mM EDTA, 1.5 mM MgCl₂, 100 mM NaF, 10 mM Na₄P₂O₇, 100 μg/ml PMSF, 5 μg/ml leupeptin, and 5 μg/ml aprotinin) using a Polytron PT10-35 homogenizer (Kinematica AG, Lucerne, Switzerland) and stored at −80 °C. Total protein concentration in each tumor lysate was determined using a Micro BCA proteins assay kit (Pierce. Rockford, IL, USA). The protein levels of human vascular endothelial growth factor (hVEGF), mouse vascular endothelial growth factor (mVEGF), mouse platelet-derived growth factor-BB (mPDGF-BB), mouse stromal cell-derived factor-1α (mSDF-1α), human placental growth factor (hPlGF), and human thrombospondin-1 (hTSP-1) were determined using ELISA kits (R&D Systems. Minneapolis, MN. USA).