Tumor model
All experimental procedures were approved by the Danish Animal Welfare Council, the Danish Ministry of Justice. Dulbecco’s Modified Eagle Medium were supplemented with 10 % Fetal Calf Serum (FCS) and 1 % penicillin-streptomycin for growth of FaDu cells in culture flaks until confluency, retained in 5 % CO2 incubator at 37 ° C. Six weeks old female NMRI nude (Naval Medical Research Institute) mice were purchased from Taconic Europe (Borup, Denmark). After one week of adaptation, 34 animals were inoculated with FaDu tumor cells subcutaneously on both left and right flank. Half of the mice (n = 17) were injected with a suspension for each tumor consisting of 2.5 × 106 FaDu cells in 50 μL Dulbecco’s Modified Eagle Medium (Life Technologies, Carlsbad, CA, USA) (−MG group; n = 17). For +MG group (n = 17) 50 μL Matrixgel™ Basement Membrane Matrix (BD Biosciences, San Jose, CA, USA) was added and a total volume of 100 μL containing 2.5 × 106 FaDu cells injected. Tumor size and weight were measured continuously from day 5 post implant to follow the development of tumors and monitor the health of the mice. Mice were housed in IVC rack from Techniplast in Type III SPF cages with 8 mice in each cage. Purified water and chow food was available ad libitum for mice unless anything else is described.
Group determination
Of the 34 mice included in study; 14 mice (+MG, n = 7; −MG, n = 7) were randomized to sub study I where all mice were 18F- fluorodeoxyglucose (18F-FDG) PET/CT scanned on Day 5, 8, 12, 15, and 19. In sub study II that included ten mice from the +MG group (n = 10) and ten mice from the -MG group (n = 10), tumors were collected when reaching predetermined sizes (200–400 mm3, 500–700 mm3, 800–1100 mm3) for immunohistochemistry (IHC) analysis of tumor characteristics.
Volume determination
Tumor volume determination with external caliper was made by measuring the greatest longitudinal diameter and transverse diameter. Tumor volume was then calculated by following ellipsoid equation [15, 16]:
$$ Tumor\kern0.5em Volume=\pi \kern0.5em \cdot \kern0.5em {\left(\frac{longitudinal\kern0.5em diameter+ transverse\kern0.5em diameter}{2}\right)}^3\kern0.5em \cdot \kern0.5em \frac{1}{6} $$
Tumor volumes determined from 18F-FDG PET/CT were generated by manually drawing regions of interest (ROIs) to cover the entire tumor by numerous tomographic voxels, and summation of these defined the 3D tumor volume.
18F-FDG PET/CT imaging – sub study I
Mice were injected via the tail vein with a mean activity of 8.90 ± 1.55 MBq (Mean ± SD) 18F-FDG in 0.2 mL 0.9 % isotonic saline solution. Prior to injection all mice were fasted for approximately 12 h to minimize the variation in 18F-FDG uptake [17]. For injection, distribution, and scanning, all mice were kept anaesthetized with 3 % sevoflurane (Abbott Scandinavia AB, Solna, Sweden) mixed with 35 % O2 and 65 % N2. Body temperature was kept stable by external heating device when anaesthetized, and positioned on a heating pad during scan. 18F-FDG PET/CT imaging was performed on Siemens Inveon® Small Animal Scanner (Siemens Medical Systems, PA, USA). The protocol included a five minute PET scan followed by a CT scan with attenuation correction to be used for reconstruction. Reconstruction of PET scans were performed using maximum a posteriori (MAP) reconstruction algorithm (voxel size: 0.815 × 0.815 × 0.796 mm; resolution (FWHM) 1.2 mm). Reconstructed images were analyzed with Inveon Research Workspace software (Siemens Medical Systems, PA, USA). Tracer uptake was determined as mean and maximum % injected dose pr. gram of tumor (%ID/g) (1 gram per cm3), and mean and maximum standardized uptake value (SUV), corrected for decay.
Tumor microenvironment – sub study II
Tumors were collected when reaching a size of 200–400 mm3 (+MG; n = 6, −MG; n = 5), 500–700 mm3 (+MG; n = 7, −MG; n = 6), and 800–1100 mm3 (+MG; n = 6, −MG; n = 4). Collection of tumors for IHC staining was initiated two weeks post injection of FaDu cells and the collection periods lasted for two weeks. Two hours prior to euthanasia 0.06 mg/g pimonidazole was injected i.p.. Tumors were fixed in 4 % formaldehyde for 24 h, hereafter transferred to 70 % ethanol, and finally embedded in paraffin and cut into 4 μm slices. Each tumor was stained with the following antibodies; pimonidazole (PIMO; hypoxia) (HypoxyProbe-Omni Kit, HypoxyProbe Inc., Burlington, USA), Ki-67 (proliferation) (Dako; M7240), and CD31 (endothelial cell marker) (Abcam; ab28364). In addition haematoxylin eosin (HE) staining was performed.
Antibody concentrations were optimized on tumor samples from mice included in this study for optimal binding specificity. The following concentrations were used for analysis; PIMO 1:400, Ki-67 1:400, and CD31 1:50.
Deparaffination was performed by heating slides for 1 h at 40 ° C, increasing temperature to 60 °C and incubating for one additional hour. Slides were subsequently treated with xylene and rehydrated in descending concentrations of ethanol (99, 96, 70 %). Slides for Ki-67 antibody staining were furthermore exposed to microwave heating after rehydration to retrieve optimal binding. Endogenous peroxidase was blocked by Peroxidase Blocking Reagent (Dako, Glostrup, Denmark) for 8 min followed by Bovine Serum Albumin (BSA) blocking with 2 % BSA for 10 min to avoid unspecific binding of antibodies. Primary antibody incubated for 1 h followed by secondary biotinylated EnVision FLEX™ (Dako, Glostrup, Denmark) incubation for 40 min. Finally antibody staining was evoked by 3,3′-Diaminobenzidine (DAB) (Dako, Glostrup, Denmark) incubation for 10 min and counterstained with haematoxylin. Between all steps slides were rinsed in phosphate buffered saline (PBS, 0.2 M, pH = 7.4). After dehydration in increasing alcohol concentrations cover slides were mounted and slides scanned on an Axio scanner (Axio scan, Carl Zeiss, Germany). The Following parameters were analyzed; cell density, hypoxia percentages, micro vessel density (MVD), average vessel area, non-viable cell percentages, and proliferation percentages. Cell density and hypoxia was determined using the publicly available software Fiji (ImageJ). For nuclear density a nuclei count threshold of 50 pixels2 to infinity was used (pixel size 0.022 × 0.022 μm). The percentage of tumor hypoxia was evaluated using Color Deconvolution based on pimonidazole DAB-H staining. Based on constructed binary images (threshold between 210 and 220 RGB values of intensity) the percentage of hypoxia positive stained area in tumor slides was determined. MVD and average vessel area was determined using online image segmentation and endothelial cell analysis software CAIMAN (CAncer IMage ANalysis: http://www.caiman.org.uk) [18] in 5 selected ROIs. ROIs, excluding necrotic regions and artifacts, were manually drawn to represent entire slide (pixel size 0.088 × 0.088 μm). Non-viable cell counts were determined in Fiji using the Advanced Weka Segmentation plug-in. Regions of viable cells, non-viable cells, and background in slide were marked to train the classifier and determine final segmentation. Calculation was made from the result image constructed by classifier. Proliferation in tumors was calculated using the online automated image analysis application ImmunoRatio (http://jvsmicroscope.uta.fi/sites/default/files/software/immunoratio-plugin/index.html) [19]. Five defined ROIs representing entire slide, excluding necrosis and artifacts, were manually drawn and uploaded to define percentage of proliferating cells in total nuclei area (pixel size 0.088 × 0.088 μm).
Statistical analysis
Statistical analysis was performed in GraphPad 6 (GraphPad Software, CA, USA). Comparison between groups of data from PET/CT scan was performed using Student’s T-test. Results are presented as Mean ± SEM (Standard Error of Mean). Analysis of data from histological staining’s was performed using One-Way ANOVA variance analysis with Holm-Sidak’s post hoc test for multiple comparisons test to evaluate differences between groups of different tumor sizes. A p-value < 0.05 was considered statistically significant in all tests.