Methodology for generating chorioallantoic membrane patient-derived xenograft (CAM-PDX) models of pleural mesothelioma and performing preclinical imaging for the translation of cancer studies and drug screening.

Patient biopsy tissue processing and cryopreservation

Sample collection

Fresh surplus biopsy tissue samples were donated under written informed consent from treatment-naïve patients undergoing diagnostic percutaneous biopsies at NHS Liverpool Heart and Chest Hospital Teaching Hospitals between March 2021 and November 2024. All specimen collection was conducted under the approved ethics (East of England – Cambridge Central Research Ethics Committee; approved 9^th August 2019, 18/EE/0161; renewed 30^th August 2023, 23/EE/0139) for the mesothelioma biobank, Mesobank UK (Rintoul et al., 2016). Samples were immediately placed into a 50 mL centrifuge tube containing 25 mL RPMI 1640 Glutamax (Thermo Fisher Scientific, MA, USA) supplemented with 1% penicillin-streptomycin (Fisher Scientific, NH, USA) and kept on ice. Samples were assigned a unique Mesobank sample identifier and transferred on ice to the research laboratory (University of Liverpool) within 30 minutes. All experiments were conducted according to the Declaration of Helsinki and in compliance with all local policies and standard operating procedures for working with human material.

Preparation of tissue samples for implantation, freezing or fixation for histology

All work with human tissue was carried out in a ESCO Airstream class II biosafety cabinet (ESCO Lifesciences Group, Singapore) and all equipment (10 cm dishes, tweezers, scalpel) were sterilized via ultraviolet (UV) light (germicidal irradiation) prior to collecting the sample. Immediately upon arrival at the lab, the contents of the 50 ml tube were transferred into a 10 cm plate (Starlab, UK). The sample was then immediately cut into 1 mm³ fragments with a scalpel (Swann Morton, UK) according to the diagram in Figure 1. Tissue pieces were kept in transport media to prevent the tissue drying out.

Figure 1. Preparation of patient tissue for parallel histological assessment and cryopreservation.

A, Workflow diagram of how tissue was processed. Figure created in Adobe Illustrator 2025 (Version 29.4 (64-bit), RRID:SCR_010279). B, Series of photographs showing tissue preparation. Pleural biopsy tissue (Bi) cut into different regions (Bii). A single region (Biii) cut into three equally sized sections (Biv). The middle section (purple box, “fresh-fixed”) was fixed in 10% NBF and the outer sections were cut into 1 mm³ fragments (Bv) and slow-frozen in either high FBS or low FBS cryopreservation media ( Table 2). One piece from each section (blue box) served as a control for the freezing process.

Cryopreservation of tissue

Two milliliter cryovials (Corning, NY, USA) were prepared with 1 mL freezing media ( Table 2) immediately prior to collection of samples, or ahead of time and frozen at -20°C. Using tweezers, a maximum of up to five 1 mm³ fragments were transferred into one cryovial ( Figure 1). Vials were placed into a Mr. Frosty freezing container (Thermo Scientific; MA, USA) and transferred to a -80°C freezer (Sanyo Electric Company Ltd., Japan) as quickly as possible. Samples were then transferred to liquid nitrogen (vapor phase) for long-term storage within a week of processing.

Reanimation of cryopreserved tissue

Frozen vials were transported on dry ice and placed in a water bath (Grant Instruments Ltd, UK) at 37°C for 30-60 seconds. Vials were agitated until the contents had thawed, ensuring all ice crystals had melted. The contents of the cryovial were poured into a 10 cm plate, ensuring no fragments were stuck in the tube. The tissue fragments were thoroughly rinsed to remove all cryoprotectant solvent by adding 5 ml of sterile Dulbecco’s phosphate-buffered saline (PBS; Thermo Scientific) to the dish and aspirating carefully. This was repeated two times prior to proceeding with CAM implantation or fixation for histology.

Chorioallantoic membrane (CAM) assay

Our methods for generating cell-line derived xenografts have briefly been described previously (Barnett et al., 2022), some stages of the process are very similar but here we provide a more detailed overview.

Egg and incubator information

Fertilized brown chicken eggs were purchased from MedEggs Ltd (UK) and stored for a maximum of 2 weeks at 14°C in a temperature-controlled chiller (Haier, China) until starting the experiments. Egg incubation was carried out in OvaEasy 380 Advance EX Series II Automatic egg incubators fitted with humidity pumps (Brinsea, UK) ( Figure 2A). Temperature and humidity were set at 37.8°C and 50%, respectively and continually monitored via dual humidity and temperature probes (T-Scan Solutions, UK).

Figure 2. Technical considerations for egg experiments.

A, Brinsea egg incubator modifications. Dual humidity and temperature T-Scan probes are fitted inside the egg incubators via the front vent (Ai-ii). 5L plastic containers are used as water reservoirs to increase the water capacity (Aiii). B, Eggs are incubated horizontally and placed to ensure they are touching neighboring eggs and unable to roll.

Initiating embryo development

Embryonic development was initiated by transferring eggs to 37°C. Egg trays and dividers were cleaned with 1X Brinsea disinfectant prior to adding the eggs. Eggs were cleaned with 1X Brinsea disinfectant by spraying the solution onto 2-ply blue roll and then thoroughly wiping each egg, to ensure all contaminants and foreign bodies were removed. The eggs were placed on their side in a row (max 5 per row, maximum 30 per tray), starting at one end of the tray ensuring that each egg was butted up against the next one to prevent eggs rotating forwards ( Figure 2B). Each egg was labelled with pencil to indicate start date and incubated with rocking (shelf rotation set to 45°, rotating every 45 minutes) from embryonic day 0 (E0) until E3.

Windowing

On E3, eggs were taken one at a time from the incubator and a small window was cut into the eggshell to provide access to the CAM (see Extended data, Supplementary video 1 (Schulze et al., 2025)). To allow for the opening to be cut, a second air pocket was made by pricking a hole in the blunt end of the egg (location of the natural air cell) with an egg piercer (Amazon, WA, USA). Next, a 5 mL syringe (Scientific Lab Supplies Ltd, UK) equipped with a 20 g 1 inch needle (Fisher Scientific) was inserted through the hole at a 45° angle to the bottom of the shell and 3-5 mL of albumin was extracted. The hole was sealed with a small square piece of orange tape (Scientific Lab Supplies Ltd). Next a three-sided rectangular window (approximately 2 cm × 1 cm) was cut in the top of the egg (the side facing up during E0-E3 incubation). Firstly, a hole where the scissors were to be inserted to cut the window was created with the egg piercer, then a piece of Scotch Magic Tape (Banner, UK) was placed over the region to be cut to prevent eggshell falling into the egg. Next, the point of the scissors was inserted into the pre-made hole and the rectangle cut, ensuring scissors were kept horizontal to avoid damage to the embryo. The egg was visually inspected for a viable embryo (determined via observation of a beating heart) and then the window sealed with a second piece of Scotch tape. Eggs were then incubated as before but without rocking until the end of the experiment.

Implantation

In preparation for implantation: lens tissue (Fisher Scientific) was cut into 1 cm × 2 cm strips and UV sterilized for 1 hour; 1-2 mm thick rings for implanting were cut from Portex tubing (1 mm wall thickness; 4 mm inner diameter) and disinfected overnight in 70% ethanol; and tweezers were cleaned (1 for lens tissue, 1 for opening egg window, 1 for handling tissue fragments). On E7 eggs were inspected to ascertain how many viable eggs were available for implant, and any non-viable eggs were discarded. In case of any leakage from the egg, a new tray lined with 2-ply blue roll was prepared. The eggs were returned to 37°C until ready to implant. The required number of vials of tissue were thawed (see section on reanimation of cryopreserved tissue fragments). One egg was removed from the incubator at a time for implantation to reduce temperature fluctuations. The piece of eggshell covering the window was removed completely, transferred to a new piece of scotch tape and left facing upwards in the hood. The egg was inspected to locate the CAM and identify an appropriate region for implantation, ideally at the bifurcation of a blood vessel. Then, using sterile tweezers, a strip of folded lens tissue was used to touch the CAM to dry/traumatize the area to be implanted (see Extended data, Supplementary video 2 (Schulze et al., 2025)). This was repeated until the area was sufficiently dry (indicated by the tissue sticking to the CAM) or a small bleed was visible. One sterilized ring was then placed on the area and one tissue fragment placed inside it. The window was resealed with the tape holding the piece of eggshell, with a second piece added perpendicularly to form a cross and ensure the window was fully sealed to reduce evaporation. The egg was given a unique identifier, labelled with pen on the colored tape on the blunt end or on the eggshell in pencil, and carefully transferred back to the incubator.

Dissection and Imaging of CAM-PDX

On E14, CAM-PDXs were imaged and dissected. PET/CT may be performed on E13 (see In ovo PET/CT imaging section), and where this was the case, the residual radioactivity was measured with a radiation monitor (Radhound; Southern Scientific, UK) to ensure the radiotracer has decayed sufficiently (below 1000 count per second) prior to dissection. Before starting, 1 ml aliquots of 10% neutral buffered formalin (NBF) were prepared in 1.5 ml microfuge tubes (1 per sample) in a fume hood. For imaging in situ, the egg was placed in a holder, such as a 35 mm dish, the window in the eggshell was enlarged (if not done at a previous stage) and brightfield images of the xenograft on the CAM were captured using Leica M165FC fluorescence stereomicroscope with 16.5:1 zoom optics, fitted with a Leica DFC425 C camera (Leica Biosystems, Germany). Then the PDX was excised. This was achieved using tweezers, such as No.5 110 mm tweezers (Dumont, Switzerland), to hold the CAM taut and cutting around the PDX with micro-spring scissors (Fine Science Tools, Germany) ensuring 0.5-1.0 cm of CAM remained around the sample. The sample was placed in a drop of PBS in a 10 cm dish and the embryo immediately terminated via decapitation in situ. The sample was washed briefly to remove excess blood by agitation in the PBS droplet and then transferred to a fresh drop of PBS for imaging. The sample was orientated using two pairs of tweezers so that the underneath of the sample could be imaged, whilst ensuring the CAM was laid out flat. The sample was then placed into fixative.

Passaging

On E14 the resulting CAM-PDX was dissected as described in the previous section. Following imaging, the PDX was trimmed using micro-spring scissors to remove the CAM and then cut in half using a sterile scalpel. One half was then implanted onto the CAM in an E7 egg, as described above, ensuring that the freshly cut side was placed onto the prepared CAM.

CAM-PDX scoring

Each CAM-PDX was evaluated macroscopically for vascularization, as an indicator for engraftment, by assessing two images. Images of the PDX in situ on the CAM and post-dissection from the underneath were scored (yes/no) for the presence of radial and feeder vessels, respectively ( Figure 3). Scoring was conducted by two independent observers. Eggs were classified as ‘not scorable’ if a decision could not be rendered by both observers with example cases including inadequate image quality, the CAM-PDX being too close the eggshell or too much blood on the CAM.

Figure 3. Scoring CAM-PDX for vascularization as a surrogate marker of engraftment.

Representative images of radial vessels (black arrowheads) surrounding the CAM-PDX in ovo and feeder vessels (white arrowheads) visible on the underneath of the CAM-PDX post-dissection. Scale bar represents 2.5 mm.

Histology

Fixation, processing, embedding and sectioning

Tissue fragments or CAM xenografts were placed in 1.5 mL tubes containing 1 mL 10% NBF and fixed for a minimum of 16 hours. The 10% NBF was removed and replaced with 1 mL 70% ethanol. Prior to processing, tissue samples were transferred to biopsy tissue cassettes (Fisher Scientific) and kept submerged in 70% ethanol until they were processed. Any small tissue fragments, such as “fresh-fixed” 1 mm³ controls, were wrapped in lens tissue prior to placing inside the cassettes to prevent loss of the small samples during processing. The cassettes containing the samples were loaded into a Leica ASP 300S tissue processor and run according to the program outlined in Table 3. Upon completion, samples were immediately transferred to a Leica EG1150H embedding station and samples embedded in Formula R paraffin wax. CAM xenograft samples were embedded in an orientation to ensure a sagittal section through the sample ( Figure 4A). The paraffin blocks were sectioned using a Thermo HM 340E Electronic Rotary Microtome fitted with MX35 ultra blades. Four-micron sections were collected on SuperFrost plus adhesion slides (according to layout in Figure 4Aiv and 4B) and placed in an oven overnight at 37°C.

Figure 4. Technical considerations for histology.

A, Workflow diagram of CAM-PDX processing for histology. CAM-PDX samples are embedded in paraffin so that sectioning the block results in sagittal sections through the tumor nodule and CAM (Ai-iii). Sections (here numbered 1-48 as an example) are placed onto slides in a configuration of 3 sections per slide that provides an overview through the xenograft with minimal slide staining; numbering (black text) refers to sequentially cut sections (Aiv). Sections are collected in up to 10 batches containing 8 slides that will be stained for different markers on contiguous slides. Figure created with Adobe Illustrator 2025 (Version 29.4 (64-bit), RRID:SCR_010279). B, One slide per batch was stained for pan-Cytokeratin to identify areas within the sample with detectable tumor cells (indicated by arrowheads), to identify the most suitable batch for staining for further markers. Scale bars represent 1000 μm.

Hematoxylin and Eosin (H&E) staining

Automated H&E staining and cover slipping was carried out on a Leica ST5020-CV5030 Stainer Integrated Workstation according to program shown in Table 4.

Immunohistochemistry

Antibody staining was conducted on a Leica Bond RXm automated stainer. The staining program used ( Table 5) was the BOND Polymer Refine IHC Protocol, utilizing the BOND Polymer Refine Detection system which is supplied as ready-to-use containing a peroxide block, post primary (secondary antibodies: rabbit anti-mouse IgG and anti-rabbit Poly-HRP IgG), polymer reagent, 3,3’-Diaminobenzidine tetrahydrochloride hydrate (DAB) chromogen for visualization and hematoxylin counterstain. Details of primary antibodies are provided in Table 6. All antibodies were optimized in-house on appropriate control tissue and staining performed in parallel to negative or isotope controls (Barnett et al., 2022). At the end of the staining program, slides were removed and placed in a staining rack. The rack was placed in a slide-staining pot and rinsed continually for 5 minutes with cold tap water, ensuring that the flowing water did not run directly onto the slides. The slides were then passed through a dehydration series consisting of 3X washes in 100% ethanol and 2X washes in xylene. Slides were gently agitated during each wash step for 30 seconds. Slides were kept in the final xylene solution and removed one at a time for cover slipping. Each slide was sealed with DPX mountant and a single 22 × 50 mm coverslip, ensuring any bubbles were removed.

Slide scanning and image processing

Whole slide scans were obtained using Leica Aperio CS2 digital slide scanner (services provided by Liverpool University Biobank). Images were viewed in QuPath (RRID:SCR_018257) (Bankhead et al., 2017) and scale bars added via the built-in ImageJ extension (RRID:SCR_003070) (Schindelin et al., 2012).

In Ovo Positron-Emission-Tomography/Computed Tomography (PET/CT) imaging

Radiotracer preparation

PET/CT imaging of tumors may be performed in ovo, as optional analysis prior to dissection of the tumor nodules. Since the half-life of [¹⁸F]-FDG is approximately 110 minutes (Ido et al., 1978), samples are safe to handle the day after radiotracer injection, and PET/CT imaging can therefore be performed on E13 followed by dissection and termination of experiments on E14 ( Figure 5A). [¹⁸F]-FDG (1000 MBq; Alliance Medical Radiopharmacy, UK) was ordered for delivery on the morning of E13 for each experiment. Upon arrival total radioactivity was measured and typically ranged from 600-800 MBq. Behind lead shielding, the [¹⁸F]-FDG stock solution was diluted in a 1.5 mL microfuge tube with sterile 0.9% sodium chloride solution (B.Braun, Germany) to give approximately 60 MBq [¹⁸F]-FDG in 1.5 mL. Due to the relatively short half-life of [¹⁸F]-FDG, the tracer volume was adjusted each time the tracer was prepared for injection. The injected radioactivity per egg was 5 ± 1 MBq in a final volume of 100-150 μL, administered using a 1 mL BD Luer slip syringe (Scientific Laboratory Supplies Ltd, UK) with a 33G 12 mm hypodermic needle (Meso-relle; UKMedi, UK). Radioactivity was measured via an activimeter (Capintec CRC-15R; Southern Scientific Ltd, UK) and documented immediately prior to injection. All work was performed in line with local rules and UK regulations, the Ionizing Radiations Regulations 2017 (IRR17 plus the Approved Code of Practice) and the Environmental Permitting Regulations 2016 (EPR2016).

Figure 5. Preparation of eggs for preclinical PET/CT imaging.

A, Experimental timeline. PET/CT performed on embryonic day 13 (E13). B, Preparing the eggs for imaging. Radiotracer is injected intravenously, demonstrated here by injecting food dye at the point indicated by the arrow (Bi). Approximately 30 seconds after injection, food dye was completely distributed within the egg (Bii). Immediate compression (a few seconds after injection) should be used to stop any bleeding at the site of injection if required. Two eggs are fixed in position in the imaging bed with tape (Biii) and then loaded into the PET/CT scanner (Biv).

Intravenous injection of radiotracer

One at a time, an implanted egg was placed in a suitable holder (such as a 35 mm dish) and the window in the eggshell was enlarged using sterile stainless-steel blunt end forceps to increase CAM accessibility for injection of the radiotracer. Whilst enlarging the window, care was taken to avoid dropping eggshell debris inside the egg or removing too much shell, which might damage the CAM and cause subsequent bleeding. A large vein was located by observing the blood flow direction and blood vessel structure. A site near the shell, with the blood flow towards the center of the egg, was chosen for injection ( Figure 5Bi). Injection of air bubbles during the procedure, which may increase the risk of killing the embryo, was avoided by carefully releasing any entrapped air before injection. The pre-loaded 33G needle and 1 mL syringe were inserted into the CAM (first pressure point) next to the chosen vein and then into the vein (second pressure point), ensuring that the vein was accessed from the side to enable visualization of the needle depth and to reduce the chance of puncturing through the blood vessel. Successful insertion of the needle was assessed by carefully moving the needle side-to-side, to check that the blood vessel followed these movements. Once it was confirmed that the needle was in the vein, the radiotracer was injected slowly and continuously in the direction of the blood flow ( Figure 5Bi, see Extended data, Supplementary video 3 (Schulze et al., 2025)). Finally, the needle was slowly removed. In case of bleeding, light compression with a sterile wipe was applied for 30-60 seconds to stop any outflow ( Figure 5Bii). Bleeding caused by the injection technique did not typically cause issues if it was stopped swiftly. Two eggs were prepared for imaging at a time, with 20-minute intervals between subsequent injections. Radioactivity before and after the injection was recorded and used to calculate the ‘activity of injected dose at time of injection’ (Equation 1-2, Table 7).

Technical considerations for intravenous injection

In principle, IV injections can be performed from E8 or earlier (Easty et al., 1969). However, commercially available needles with a small enough gauge are difficult to acquire and pulling needles in house requires specialist equipment. Therefore, we recommend performing IV injection from E11 onwards since a higher injection success rate is achievable due to larger blood vessels being available to inject. Choosing a larger vessel reduces the chance of causing damage and blood loss, thus having less effect on survival. Once competent in the radiotracer injection technique, there is the potential to image at least 24 eggs per day, with the only limiting factor being the activity of the decaying radiotracer.

CT scan followed by static PET (Static PET/CT)

Eggs were incubated for at least 30 minutes post-injection to allow radiotracer distribution and uptake before scanning (Warnock et al., 2013, Smith et al., 2023). CT scans were carried out before PET to determine the optimal field of view (FOV, optimization of bed position) to ensure a high spatial resolution. Up to two eggs were placed into the rat bed, held in place with a piece of scotch tape ( Figure 5Biii) and loaded into the X-CUBE Preclinical CT imager (Molecubes NV, Belgium) together ( Figure 5Biv). CT scans were acquired using the standard protocol (‘general purpose’; eggs received 27 mGy dose). Once the CT scans were complete, the multimodal bed containing the eggs was immediately transferred to the β-CUBE Preclinical PET imager (Molecubes NV). Acquisition duration was set to 15 minutes and started in the optimal bed position (as determined via CT). The time between injection of the radiotracer and starting PET imaging was noted. Once scanning was complete, the eggs were returned to the incubator to maintain the correct temperature.

PET/CT data reconstruction

The PET/CT data were reconstructed (see Table 8 for settings) using the proprietary Molecubes software (Build version: 1.7.6 H 989d0998; Copyright license registered to University of Liverpool – Centre for Preclinical Imaging). Reconstructed PET/CT data were then visualized, images co-registered and PET signal from PDX tumors quantified by using Invicro VivoQuant 2020 (Copyright license registered to University of Liverpool – Centre for Preclinical Imaging; RRID:SCR_025778; Invicro LLCm, MA, USA). A three-dimensional region of interest (ROI) was manually drawn around each xenograft ( Figure 6) to quantify the concentration of [¹⁸F]-FDG at a defined timepoint (C_ROIt). This value was used to calculate the accumulated standardized uptake value (SUVacc) across the CAM-PDX (Equation 3, Table 7 (Basu et al., 2007)).

Figure 6. PET data analysis.

A, User Interface of Invicro Vivoquant software showing different sectional planes (transverse, sagittal and coronal) from a reconstructed CAM-PDX. B, Input controls and operator settings for drawing the region of interest (ROI) manually (painting tool; sphere, size set to 3 pixel). C, CAM-PDX PET signal of example shown in A overlaid with manually drawn ROI. D, Linearity between voxel size and accumulated standardized uptake value (SUVacc) PET signal for different CAM-PDX underlined consistency in manual ROI drawing. Coefficient of determination; R² = 0.90. Values used are reported in Table 9.

Experimental design and statistical evaluation

Each xenografted egg was considered as one biological unit and all eggs prepared and imaged together as one experimental unit. Sample size per experimental unit was determined by the number of eggs (24 maximum) that can be scanned in one day. Biological units were included if they underwent successful engraftment and tracer injection. Biological units were excluded if they were unfertilized, died prior to E13 or showed signs of contamination. To account for exclusions, experiments were started with twice the number of biological units required at E14. For comparison of frozen tissue, biological units were randomly divided into two groups on E7.

Statistical analyses were performed in OriginPro 2021b Academic (OriginLab Corporation, MA, USA; RRID:SCR_014212). Distribution and variance of data was assessed by Shapiro–Wilk test and Two-Sample Test for variance, respectively. Data were then analyzed by parametric or non-parametric test as appropriate. The number of independent samples, definition of error bars, and the statistical test employed are described in relevant figure legends. p values less than 0.05 were considered significant.

Controlling bias, variability and limitations of datasets

Potential sources of bias

Sample bias: No formal inclusion or exclusion criteria were applied to the sample collection and all samples available from patients that consented were collected without prior knowledge of histological subtype, or BAP1 status ( Table 10). Samples are linked anonymized but patient information, such as age and gender, has not been accessed during method development. However, we would expect a bias towards epithelioid as this represents 70% of mesothelioma cases in England and Wales, as well as samples from older males, reflecting the much higher incidence of mesothelioma in men (Husain et al., 2024).

Vascularization scoring: To reduce observer bias, each CAM-PDX was scored macroscopically for engraftment by two independent researchers.

PET/CT analysis: To quantify the PET signal for each CAM-PDX, a ROI was manually drawn in the Invicro Vivoquant software ( Figure 6). For consistency the same settings (geometry and size) for the selection tool were used each time. To check for variability in ROIs, the SUVacc for each CAM-PDX was plotted against voxel size. A coefficient of determination (R²) of 0.90 confirmed consistency between ROIs.

Potential sources of variability

Egg survival: Many factors influence egg quality and variability in survival, including the climate the chickens are exposed to before laying and sudden changes to their environment, temperature during transportation, duration between laying and initiating development, temperature fluctuations during the experiment, number of interventions during the experiment, cleanliness and decontamination procedures. In our experience survival to E14 can range from 50 to 80%.

Implantation: There could be inter-researcher differences in technique, however experiments were performed by three independent researchers according to standard operating protocols.

PET signal readout: Standardized uptake value (SUV) allows for normalization between CAM-PDXs by removing variability introduced through injected doses and time differences.

Limitations of datasets

Patient samples: Since this article is intended to demonstrate proof of principle and methodology development, here we present detailed data for samples collected from only three patients (one sarcomatoid and two epithelioid). However, vascularization of samples from five patients is also summarized. An overview of all patient samples collected is provided in Table 10.