This paper describes methods for the generation, drug treatment, and analysis of patient-derived explants for assessing tumor drug responses in a live, patient-relevant, preclinical model system.
An understanding of drug resistance and the development of novel strategies to sensitize highly resistant cancers rely on the availability of suitable preclinical models that can accurately predict patient responses. One of the disadvantages of existing preclinical models is the inability to contextually preserve the human tumor microenvironment (TME) and accurately represent intratumoral heterogeneity, thus limiting the clinical translation of data. By contrast, by representing the culture of live fragments of human tumors, the patient-derived explant (PDE) platform allows drug responses to be examined in a three-dimensional (3D) context that mirrors the pathological and architectural features of the original tumors as closely as possible. Previous reports with PDEs have documented the ability of the platform to distinguish chemosensitive from chemoresistant tumors, and it has been shown that this segregation is predictive of patient responses to the same chemotherapies. Simultaneously, PDEs allow the opportunity to interrogate molecular, genetic, and histological features of tumors that predict drug responses, thereby identifying biomarkers for patient stratification as well as novel interventional approaches to sensitize resistant tumors. This paper reports PDE methodology in detail, from collection of patient samples through to endpoint analysis. It provides a detailed description of explant derivation and culture methods, highlighting bespoke conditions for particular tumors, where appropriate. For endpoint analysis, there is a focus on multiplexed immunofluorescence and multispectral imaging for the spatial profiling of key biomarkers within both tumoral and stromal regions. By combining these methods, it is possible to generate quantitative and qualitative drug response data that can be related to various clinicopathological parameters and thus potentially be used for biomarker identification.
The development of effective and safe anticancer agents requires appropriate preclinical models that can also provide insight into mechanisms of action that can facilitate the identification of predictive and pharmacodynamic biomarkers. Inter- and intratumor heterogeneity1,2,3,4,5 and the TME6,7,8,9,10,11,12 are known to influence anticancer drug responses, and many existing preclinical cancer models such as cell lines, organoids, and mouse models are not able to fully accommodate these crucial features. An "ideal" model is one that can recapitulate the complex spatial interactions of malignant with non-malignant cells within tumors as well as reflect the regional differences within tumors. This article focuses on PDEs as an emerging platform that can fulfil many of these requirements13.
The first example of the use of human PDEs, also known as histocultures, dates back to the late 1980s when Hoffman et al. generated slices of freshly resected human tumors and cultured them in a collagen matrix14,15. This involved establishing a 3D culture system that preserved tissue architecture, ensuring the maintenance of stromal components and cell interactions within the TME. Without deconstructing the original tumor, Hoffman et al.16 heralded a new approach of translational research, and since this time, many groups have optimized different explant methods with the aim of preserving the tissue integrity and generating accurate drug response data17,18,19,20,21,22,23,24, although some differences between protocols are evident. Butler et al. cultured explants in gelatin sponges to help the diffusion of nutrients and drugs through the specimen20,21,25, whereas Majumder et al. created a tumor ecosystem by culturing explants on top of a matrix composed of tumor and stromal proteins in the presence of autologous serum derived from the same patient22,23.
More recently, our group set up a protocol whereby explants are generated by fragmentation of tumors into 2 – 3 mm3-sized pieces that are then placed without additional components on permeable membranes at the air-liquid interface of a culture system24. Taken together, these numerous studies have demonstrated that PDEs allow the culture of intact, live fragments of human tumors that retain the spatial architecture and regional heterogeneity of the original tumors. In original experiments, explants or histocultures were usually subjected to homogenization following drug treatment, after which various viability assays were applied to the homogenized samples such as the histoculture drug response assay20,21, the MTT (3-(6)-2,5-diphenyltetrazolium bromide) assay, the lactate dehydrogenase assay, or the resazurin-based assay26,27,28. Recent progress in endpoint analysis techniques, particularly digital pathology, have now expanded the repertoire of endpoint tests and assays that can be performed on explants29,30. To apply these new technologies, instead of homogenization, explants are fixed in formalin, embedded in paraffin (FFPE) and then analyzed using immunostaining techniques, allowing spatial profiling. Examples of this approach have been documented for non-small cell lung cancer (NSCLC), breast cancer, colorectal cancer, and mesothelioma explants whereby immunohistochemical staining for the proliferation marker, Ki67, and the apoptotic marker, cleaved poly-ADP ribose polymerase (cPARP), was used to monitor changes in cell proliferation and cell death24,31,32,33,34.
Multiplexed immunofluorescence is particularly amenable for spatial profiling of drug responses in explants at endpoint35. For example, it is possible to measure the relocalization and spatial distribution of specific classes of immune cells, such as macrophages or T cells, within the TME upon drug treatment13,36,37,38, and investigate whether a therapeutic agent can favor the transition from "cold tumor" to "hot tumor"39. In recent years, this group has focused on the derivation of PDEs from different tumor types (NSCLC, renal cancer, breast cancer, colorectal cancer, melanoma) and the testing of a range of anticancer agents including chemotherapies, small-molecule inhibitors, and immune checkpoint inhibitors (ICIs). Endpoint analysis methods have been optimized to include multiplexed immunofluorescence to allow spatial profiling of biomarkers for viability as well as biomarkers for different constituents of the TME.
1. Tissue collection
2. Explant preparation
3. Drug treatment
4. Histological processing
5. Hematoxylin and eosin (H&E) staining
6. Immunostaining
NOTE: The following steps should be carried out at RT unless stated otherwise.
7. Scanning
NOTE: Slide scanning was performed using a multispectral automated imaging system (see the Table of Materials).
8. Analysis
NOTE: The protocol below illustrates the method for phenotype analysis.
Multispectral imaging of mIF-stained histological sections permits identification and phenotyping of individual cell populations and identification of tumor and stromal components in the explant TME (Figure 2). Multispectral imaging is particularly useful for the analysis of tissues with high intrinsic autofluorescence, such as tissue with a high collagen content, as it allows the autofluorescence signal to be deconvoluted from other signals and excluded from subsequent analysis. Subsequent in silico tissue and cell segmentation allows the quantification of drug response with a multitude of outputs including, but not limited to, raw cell numbers (e.g., percent positivity), signal intensity, cell size, tissue area, and spatial location of individual cells.
Tissue and cell segmentation are therefore extremely useful for the quantitative analysis of drug treatment outcomes and the identification of tumor- and stromal-specific responses. For example, staining for Ki67 and cPARP alongside a tumor marker allows for the quantitation of proliferation and cell death levels, respectively, in explants (Figure 3). In the example given, Ki67 and cPARP levels are quantitated for stromal and tumor regions in PDEs in response to the anti-programmed cell death protein 1 (PD-1) immune checkpoint inhibitor (ICI), nivolumab. In addition, the ability to extract spatial information from stained sections allows the calculation of intercell distances as well as distances to tumor borders and other structures. Therefore, changes in cellular distribution following drug treatment can be quantitatively analyzed (Figure 4). The example shown represents PDE treatment with nivolumab.
Figure 1: Workflow for the generation and analysis of PDEs. (A) Freshly resected human tumor specimens are processed and arranged on a permeable culture insert disc floating in culture medium. (B) Following drug treatment, the explants are harvested, subjected to FFPE, and then sectioned for histological analysis, e.g., for H&E staining. (C) mIF staining and scanning of the tissue sections are performed to generate multispectral images from which individual signals can be deconvoluted and analyzed separately. (D) Analysis of composite images allows separation of tumor and stromal areas and assessment of the phenotype of different cell types. Abbreviations: PDE = patient-derived explant; FFPE = fixed in formalin, embedded in paraffin; H&E = hematoxylin and eosin; mIF = multiplex immunofluorescence; HRP = horseradish peroxidase; H2O2 = hydrogen peroxide; Ab = antibody; TSA = tyramide signal amplification. Please click here to view a larger version of this figure.
Figure 2: Example of an mIF-stained tissue section. NSCLC explant tissue was stained for markers of cell viability, namely, cPARP (red) and Ki67 (green) as well as a pan-cytokeratin marker (yellow) and DAPI (blue) to mark nuclei. Multispectral imaging was performed to deconvolute the fluorescent signals and remove autofluorescence. Images show 20x magnification of the selected field. Scale bars = 50 µm. Abbreviations: mIF = multiplex immunofluorescence; NSCLC = non-small cell lung cancer; cPARP = cleaved poly-ADP ribose polymerase; CK = cytokeratin; DAPI = 4'-6-diamidino-2-phenylindole; AF = autofluorescence. Please click here to view a larger version of this figure.
Figure 3: Quantification of cell death and proliferation in PDEs. Example of box and whisker plots showing apoptosis induction in NSCLC PDEs after 24 h of treatment with 5 µg/mL nivolumab compared to medium control. The percentage of proliferation (Ki67) and percentage of apoptotic cell death (cPARP) events is displayed in the tumor area and stroma, respectively. Each point represents a single explant, the central line of the box represents the median of the distribution, and the sides of the box (bottom and top) represent the first and third quartiles, respectively. Error bars represent the interquartile range (further than 1.5 x IQR). Abbreviations: PDE = patient-derived explant; NSCLC = non-small cell lung cancer; cPARP = cleaved poly-ADP ribose polymerase; IQR = interquartile range. Please click here to view a larger version of this figure.
Figure 4: Spatial organization following drug treatment. (A) Representative images showing mIF staining of T cell markers-CD4, FOXP3, CD8 (left panel)-performed on a melanoma explant, and corresponding phenotype analysis (right) identifying intercell distances between CD8+ cells and Treg cells (CD4+/FOXP3+). Scale bars = 200 µm. (B) Histogram plot showing increased distance between cytotoxic T cells (CD8+) and Treg cells (CD4+/FOXP3+) after treatment of melanoma PDEs with 5 µg/mL of nivolumab, confirming on-target effects of the IO drug. Density unit is expressed as the number of cells divided by sum of all cells per bandwidth. Abbreviations: mIF = multiplex immunofluorescence; CD = cluster of differentiation; FOXP3 = Forkhead box P3; PDE = patient-derived explant; NSCLC = non-small cell lung cancer; IO = immunooncology. Please click here to view a larger version of this figure.
This paper describes the methods for generation, drug treatment, and analysis of PDEs and highlights the advantages of the platform as a preclinical model system. Ex vivo culturing of a freshly resected tumor, which does not involve its deconstruction, allows for the retention of the tumor architecture13,24 and thus, the spatial interactions of cellular components in the TME as well as intratumoral heterogeneity. This method demonstrates how, by using a tumor-specific marker, it is possible to identify areas of tumor tissue versus areas of stroma and therefore separate drug responses within these compartments (Figure 3). Additionally, multiple biomarkers can be profiled simultaneously to assess, for example, the on-target movement of immune cells within the TME (Figure 4).
A previous publication documented the application of this PDE platform to the stratification of NSCLC PDEs for the standard of care chemotherapy, cisplatin, showing that it is possible to segregate chemoresistant from chemosensitive populations24. These PDE-related findings mirror patient responses. By following the methods described here, it is possible to undertake similar approaches for other tumor types and with different types of drugs, chemotherapeutics or otherwise. The separation of PDEs into drug-sensitive and drug-resistant populations creates an invaluable resource for generating further mechanistic insight into drug resistance. For example, because PDEs can be processed for RNA, DNA, protein, or metabolite isolation, this can allow the implementation of "omic" technologies to identify key biomarkers that are predictive of response. Alternatively, FFPE sections generated from treated PDEs can be used for extensive spatial profiling to understand how different cell types in the PDE contribute to drug resistance.
This may be facilitated by further developments in multispectral imaging and mass cytometry approaches35,40,41,42,43, which would allow hundreds of biomarkers to be profiled simultaneously. Preclinical models that evaluate immunotherapeutic efficacy are much sought after, and this paper demonstrates that PDEs can fill this critical niche (Figure 3 and Figure 4). Monoclonal antibodies targeting immune checkpoints, such as cytotoxic T-lymphocyte antigen 4 and PD-1/programmed death ligand-1 (PD-L1), have been developed44,45,46,47 with some remarkable improvements in overall patient survival in a large number of solid tumors compared to standard chemotherapy. However, ICIs are effective in a limited number of patients for reasons that are not clear, necessitating the identification of predictive biomarkers48. The outstanding feature of PDEs-preserving the 3D architecture of tumor tissue-facilitates the evaluation of ICI efficacy (Figure 3) and the monitoring of immune cells in response to ICI treatment (Figure 4). Thus, PDEs are an ideal platform for distinguishing ICI-sensitive versus ICI-resistant cases and for investigating the mechanisms underpinning this distinction.
PDE technology does suffer from a few disadvantages. The generation of accurate, experimental results from PDEs relies on tumor integrity, and occasionally, tumor samples after surgery are too necrotic to process for PDEs. Furthermore, despite some specific examples of retention of tissue integrity after prolonged culture25, for most reported cases, integrity and viability has been lost after 72 h in culture, and tissue disintegration occurs. The window of time to perform drug experiments is therefore relatively limited, prohibiting the use of this model for the study of the mechanisms of acquired drug resistance or the study of invasion and metastasis13. Extending the viability of explants may become possible in the future through the development of new technologies, such as scaffolds and perfused channels, which may facilitate diffusion and uptake of nutrients, allowing more extended culture.
However, until these improvements can be made, the PDE platform should be regarded as a short-term culture method that can provide immediate drug response data. It should be utilized alongside other model systems, such as organoids and PDX models, that can provide longer term drug response data. Overall, the PDE platform is a proof-of-concept preclinical model system that is of use in determining the sensitivity of a patient's tumor to a given anticancer agent, for deriving insight into mechanisms of drug action in a real tumor, and for developing predictive and pharmacodynamic biomarkers.
The authors have nothing to disclose.
We thank the surgeons and pathologists at University Hospitals of Leicester NHS Trust for providing surgical resected tumor tissue. We also thank the Histology facility within Core Biotechnology Services for help with tissue processing and sectioning of FFPE tissue blocks and Kees Straatman for support with use of the Vectra Polaris. This research was supported and funded by the Explant Consortium comprising four partners: The University of Leicester, The MRC Toxicology Unit, Cancer Research UK Therapeutic Discovery Laboratories, and LifeArc. Additional support was provided by the CRUK-NIHR Leicester Experimental Cancer Medicine Centre (C10604/A25151). Funding for GM, CD, and NA was provided by Breast Cancer Now's Catalyst Programme (2017NOVPCC1066), which is supported by funding from Pfizer.
Acetic acid | Sigma | 320099 | Staining reagent |
Antibody Diluent / Block, 1x | Perkin Elmer | ARD1001EA | Antibody diluent/blocking buffer |
Barnstead NANOpure Diamond | Barnstead | Ultra Pure (UP) H2O machine | |
Citric Acid Monohydrate | Sigma-Aldrich | C7129 | Reagent for citrate buffer |
Costar Multiple Well Cell Culture Plates | Corning Incorporated | 3516 | 6 multiwell plate |
DAPI Dilactate | Life Technologies | D3571 | |
100 x 17 mm Dish, Nunclon Delta | ThermoFisher Scientific | 150350 | 100 mm diameter dish for tissue culture |
DMEM (1x) Dubelcco's Modified Eagle Medium + 4.5 g/L D-Glucose + 110 mg/mL Sodium Pyruvate | Gibco (Life Technologies) | 10569-010 | Tissue culture medium (500 mL) |
DPX mountant | VWR | 360294H | Mounting medium |
DPX mountant | Merck | 6522 | Mounting medium |
Ethylenediaminetetraacetic acid (EDTA) | Sigma-Aldrich | 3609 | Reagent for TE buffer |
Eosin | CellPath | RBC-0100-00A | Staining reagent |
Foetal Bovine Serum | Gibco | 10500-064 | For use in tissue culture medium |
37% Formaldehyde | Fisher (Acros) | 119690010 | 10% Formalin |
iGenix, microwave oven IG2095 | iGenix | IG2095 | Microwave used for antigen retreival |
Industrial methylated spirit (IMS) | Genta Medical | 199050 | 99% Industrial Denatured Alcohol (IDA) |
InForm Advanced Image Analysis Software | Akoya Biosciences | InForm | |
Leica ASP3000 Tissue Processor | Leica Biosystems | Automated Vacuum Tissue Processor | |
Leica Arcadia H and C | Leica Biosystems | Embedding wax bath | |
Leica RM2125RT | Leica Biosystems | Rotary microtome | |
Leica ST4040 Linear Stainer | Leica Biosystems | H&E stainer | |
Mayer's Haematoxylin | Sigma | GHS132-1L | Staining reagent |
Millicell Cell Culture Inserts, 30 mm, hydrophilic PTFE, 0.4 µm | Merck Milipore | PICMORG50 | Organotypic culture insert disc |
Novolink Polymer Detection System | Leica Biosystems | RE7150-K | DAB staining kit |
OPAL 480 | Akoya Biosciences | FP1500001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 520 | Akoya Biosciences | FP1487001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 570 | Akoya Biosciences | FP1488001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 620 | Akoya Biosciences | FP1495001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 650 | Akoya Biosciences | FP1496001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 690 | Akoya Biosciences | FP1497001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent |
OPAL 780 / OPAL TSA-DIG Reagent | Akoya Biosciences | FP1501001KT | Fluorophore with Dimethyl Sulfoxide (DMSO) diluent and TSA-DIG reagent |
Opal Polymer HRP Ms Plus Rb, 1x | Perkin Elmer | ARH1001EA | HRP polymer |
Penicillin/streptomycin solution | Fisher Scientific | 11548876 | For use in tissue culture medium |
PhenoChart Whole Slide Contextual Viewer | Akoya Biosciences | PhenoChart | Viewer software for scanned images |
Phosphate Buffered Saline Tablets | Thermo Scientific Oxoid | BR0014G | PBS |
1x Plus Amplification Diluent | Perkin Elmer | FP1498 | Fluorophore diluent |
Prolong Diamond Antifade Mountant | Invitrogen | P36961 | Mounting medium |
Slide Carrier | Perkin Elmer | To load slides into Slide Carrier Hotel for scanning with Vectra Polaris | |
Sodium Chloride | Fisher Scientific | S/3160/63 | 10% Formalin |
Sodium Hydroxide pellets | Fisher Scientific | S/4920/53 | Reagent for citrate buffer |
Tenatex Toughened Wax – Pink (500 g) | KEMDENT | 1-601 | Dental wax surface |
Thermo Scientific Shandon Sequenza Slide Rack for Immunostaining Center | Fisher Scientific | 10098889 | Holder for slides and slide clips |
Thermo Scientific Shandon Plastic Coverplates | Fisher Scientific | 11927774 | Slide clips |
Tris(hydroxymethyl)aminomethane (Tris) | Sigma-Aldrich | 252859 | Reagent for TE buffer |
VectaShield | Vecta Laboratories | H-1000-10 | Mounting medium |
Vectra Polaris Slide Scanner | Perkin Elmer | Vectra Polaris | Slide scanner |
Xylene | Genta Medical | XYL050 | De-waxing agent |