Towards in vitro oncology trials: drug testing in cultured patient derived tumour organoid cultures.

Breast cancer is a major cause of illness and death in women. In order to develop new treatment strategies we need to test therapy ideas in models of breast cancer. We use these models to test our best ideas in the laboratory but we need these to faithfully recapitulate the biology of the patient's tumour so they can help us predict what therapies will work in patients in the clinic. It is now widely recognized that models where you implant a patient's tumour into a mouse called patient-derived xenograft models (PDX) are the best in vivo models of breast cancer and far superior to long-established cancer cell lines grown in petri dishes. We, like others, have established PDX models of very aggressive forms of breast cancer called triple negative breast cancer. We did this by transplanting patient's tumour cells in vivo into immunocompromised mice. The transplanted cells develop into a tumour model that we can use to develop novel treatments and to assess biomarkers for predicting tumour behavior. We would like to be able use similar material from patients without needing to use mice by growing them in special conditions of three-dimensional culture. We call these patient-derived organoids (PDOs).

For each successful drug that enters the market, 250 have failed in animal-based preclinical testing. Moreover, of the drugs that do enter clinical testing, the majority fails due to lack of effectiveness and the presence of side effects. Clearly, there is an urgent need to improve current laboratory testing to prevent ineffective drugs from entering the costly and undesirable pre-clinical animal testing phase. This project will directly address this need by generating three-dimensional cultures of organoids (PDO) from already established tumour models (PDX) in order to use these laboratory PDO models to reduce the number of animal experiments used in early phase drug selection processes. In this project we are partnering with a biotech company, OcellO, that has lots of experience in generating 3D organoids in the laboratory. We will do this from our PDX models.

The aims of this study are:
1) To generate companion 3D cultures of these PDX models, termed "Patient-derived organoid" (PDO) models and compare the drug response in these with the current gold standard PDX models in order to show their validity as a laboratory model of the patient's breast cancer. We will do this by culturing tumour cells from PDX models in special conditions to generate an "organ in a well". By doing this reproducibly and generating many wells we can screen drug candidates for responses in these PDO models. We hope that this will not only match but also extend what we can currently learn from PDX models. The aim will then be to have data to convince researchers and the pharmaceutical industry they can refine many of the current PDX experiments using PDOs and only go to the matched PDX of the PDO for a refined set of animal experiments.

2) To understand if we can replace the need for PDX models for many experiments, we will make PDOs direct from patient tumour sample and parallel PDXs. The performance of the two parallel models will be compared and provide evidence to stimulate possible replacement of PDX completely in some situations.

3) To achieve the aims above we will use the PDO models generated to test a range of established standard of care breast cancer treatments and novel treatments that have already passed strong "proof of concept" in different types of breast cancer.

By working with our project partners we maximize the likelihood of successful determination of whether our PDO models are faithfully modeling the PDX and the original patient's breast cancer. We believe that our partnership will provide the data, the connections and pathways to impact to convince others that companion laboratory 3D cultures (PDO) can streamline the drug discovery pipeline enabling drugs to reach patients faster and with less cost.

3D cell culture and high content imaging in combination with an unbiased informatics-based analysis allows an ex vivo approach to acquire a multi-parametric quantitative description of an in vitro 3D tumour phenotype. Patient-derived xenograft (PDX) tumours represent the gold standard of making patient tumours into pre-clinical models. By incorporating cells derived from PDX tumours we can recapitulate the complexity and heterogeneity present in patients and integrate this into our project partner OcellO's established and validated 3D cell culture platform to generate patient-derived organoid (PDO) models. Such an approach has not been established so far for breast cancer and creates a unique opportunity for high throughput 3D screening platforms to reduce and replace PDX model use.

OcellO has established this principle for cancer cell lines and some prototype in vitro PDO models. Here, we aim to develop a panel of in vitro breast PDO cultures. To systematically evaluate in vitro to in vivo correlation of our models, we will focus on the therapeutically challenging breast cancer sub-type-TNBC. In particular this focus and availability of defined PDX material as well as new drug candidates takes into account our laboratory's focus and the clinical need in analysis of our models for the reduction and partial replacement of animal testing. We will focus on PDO-derived assays for the concordance of therapeutic responses to PARP inhibitors in TNBC matched companion PDOs and PDXs. We will also assess whether drug combination strategies involving PARP inhibitors can be interrogated in such PDO systems. By observing the concordance between the results of the in vitro testing from results of "top-candidate" testing in PDX models in vivo we will assess the concordance between these results. This will establish the use of 3D high-throughput tumour phenotype testing as a triage in order to establish whether novel drugs will be taken forward to in vivo pre-clinical testing

The problem: A major rate-limiting step in cancer drug development is in the use of animals, especially mice, to assess the in vivo efficacy and tolerability of novel drugs. More often than not, promising drugs fail to inhibit tumours in mice and their development is therefore halted. In many cases this is not because the drugs are ineffective, but because they are assessed in inappropriate animal models. The end result is that; (1) for each successful drug that enters the market, 250 have failed in animal-based preclinical assessment, and that (2) countless animals are used in uninformative experiments. Furthermore, most mouse studies use long-term administration of drugs in animals carrying tumours, both of which can lead to suffering. Clearly there is a need to replace, reduce and refine these studies as far as is possible.

The proposal: We propose to develop an in vitro reagent bank of patient-derived organoid (PDO) tumour models of a particularly difficult to treat form of breast cancer, triple negative breast cancer (TNBC). In brief, PDOs act as a three-dimensional (3D) surrogate tumour that can be grown and manipulated ex vivo. These PDOs models would provide an in vitro platform for assessing the effectiveness of drugs prior to, or in the place of, any assessment in mice. The information gained from testing drugs in PDO models would mean that inappropriate animal studies are avoided altogether (replacement) and those animal studies that are carried out are better designed and more likely to be informative, leading to the use of fewer animals (reduction). Furthermore, most in vivo studies of cancer drug effectiveness measure the effect of a drug by assessing tumour volumes or mass after a long-term drug administration. The more complex 3D biological phenotypes that can be assessed in PDOs will allow us to predict how cancer drugs will work at the molecular level on more complex cancer "hallmark" phenotypes; this will mean that molecular and cellular biomarkers of tumour response, that are elicited after short term treatments, could be used in eventual in vivo studies thus minimising suffering and therefore representing a refinement in commonly used approaches.
In concordance with the principles of NC3Rs, we anticipate that this project can lead to a dramatic reduction and replacement of the number of exploratory pre-clinical PDX in vivo experiments that take place. In vivo PDX experiments take place widely to enable pre-screening and selection of novel candidates: for example, it is typical for a company to test a few lead candidates in 10-20 PDX models; in vitro PDX model pre-screening would be expected to reduce this to 3-5 models. Use of the technology could also enable in vitro PDX propagation to replace in vivo for supply of tumour material. The 3D PDO models could also inform primary drug screening and screening of drug combinations: by making use of the high throughput capacity of the screening platform, large numbers of in vivo pre-clinical experiments could be replaced.

For example in our laboratory we use 350 mice annually to maintain and generate PDX tumours for use in in vivo experiments. Using PDO models to triage compound testing we would expect to replace 25% of the mice used and using PDOs to generate new models we would expect to replace a further 25% of animals used (50% overall). We know of 5 other groups at our institution using PDX models of breast cancer. Assuming a similar usage, PDO models could replace 875 animals annually at our institution. A PubMed search shows in 2015 there were 56 primary articles published on the use of PDX models in breast cancer. If each laboratory publishing these studies has similar animal usage to us the replacement with PDO could reduce the number of animals by 50% - equating to many thousands of mice internationally that would no longer be used in exploratory experiments for breast cancer, not taking accounting the use in the pharmaceutical industry.