Replacing liver cancer models by modeling human liver cancer in vitro

Liver cancer, classified into 2 main entities, Hepatocellularcarcinoma and Cholangiocarcinoma, is the 3rd cause of cancer death worldwide and its incidence is increasing due to co-funding factors such as obesity and life-style (e.g. alcoholism). The primary management of liver cancer is mainly restricted to surgical removal of the tumour. Unfortunately, few patients can benefit from it as, at the diagnostic time, the tumour has already spread and metastized. This is particularly the case for Cholangiocarcinomas, which are the most aggressive and with the poorest prognosis. When tumour resection is not an option, patients are left with little therapeutic options, which mainly rely on the use of clinically approved compounds such as sorafenib. However, this treatments have little efficacy as liver cancer is a very heterogenous entity, where the spectre of tumour mutations makes each patient's tumour an almost individual entity. Therefore, liver cancer would extremely benefit from personalized treatments instead of one-fits-all therapeutic approaches.
To develop personalized treatments for liver cancer, at present, both industry and academics rely on Patient derived tumour xenografts (PDXs), which require the patient's tumour to be implanted into a mouse host. While these are very good models that retain all the histological, genetic and transcriptomic features of the patient's original tumour, they are not suitable to most of the patients as usually they fail to develop and, if they do, they are not amenable to test >5 drugs, as they require many animals and resources.
A good alternative would be to be able to grow and expand the patient's primary tumour directly on a dish. To be a good and suitable model, these primary cultures should be expandable while it should retain the characteristics of the patient's tumour and be amenable for large drug screening (at least 20 coumpounds). However, up to now, these human liver cancer in vitro cultures have yet to be developed.
We have recently shown that human liver cancer cells can be expanded long-term from biospies of human donors and can be used to model human liver Alpha 1 Antitrypsin Deficiency, a rare monogenic disease of the liver. Therefore, in this proposal, we aim at translating our organoid culture system from modelling human liver biology in homeostasis to modelling human liver cancer in vitro.

Therefore, our aim is to develop, characterize and validate primary liver cancer organoid cultures and prove the relevance in terms of similarity to the original patient's tissue as well as their applicability as platform for anti-cancer drug testing for future personalized medicine approaches for liver cancer patients. Also, because we will be studying human liver cancer tissue directly we believe that our culture system will result in the future replacement of animal models, and specially PDXs, for drug screening tests for human liver cancer personalized approaches.

Liver cancer is among the most lethal cancers. This is because at present, there is only an elemental understanding of the pathogenesis and imited therapeutic options. This is in part due to the lack of good models that faithfully represent the genetic variability of human liver cancer. In fact, the majority of the existing animal models (from mice to dog) do not replicate the human disease and existing cell lines do not mimic the complex genetic variability of these tumours. Only Patient Derived Xenografts (PDXs) retain all the patient's tumour characteristics, from their histological architecture to their transcriptome, methylome and mutation landscape. Thus, they represent the only reliable model for human liver cancer up to now. However, despite their utility and predictive value, PDXs are not amenable for large drug testing (20 drugs at a time) as they would require many animals and resources.
An alternative to that, would be the development of primary liver cancer cultures that could expand long-term, be easy to manipulate and where to perform large drug screening tests (>20 compounds at a time), in a timely manner since the collection of the specimen from the patient. However, primary liver cells have proven challenging to expand in culture.
We have recently described that human liver healthy tissue can be cultured and expanded long-term in vitro, into genetically stable 3D-structures that we have termed "liver organoids". Using this technology we have successfully modelled 2 human liver monogenic diseases: Alpha-1 Anti-trypsin and Allagile Synrome. Here we aim at obtaining liver cancer organoids directly from the patient's tumour biopsy. This will facilitate assaying personalized anti-cancer therapeutic strategies and studying human liver cancer biology. As a consequence, our project will replace the use of animals in liver cancer research and drug testing thus significantly impacting on the application of the 3Rs.

Currently, we are lacking good models that mimic human liver cancer in vitro. At present the only human liver cancer models that faithfully reproduce the histological and molecular charecteristics of the human pathology rely on Patient-derived xenografts (PDXs), which are not amenable for drug testing and require many animals. In this proposal, we aim at translating our organoid culture system from modelling human liver biology in homeostasis to modelling human liver cancer. To do that, we will use our extensive knowledge on developing 3D culture models for liver, pancreas and other organs (Huch et al., Nature 2013; Huch et al., Cell 2015; Broutier et al., Nature Protocols 2016) to expand, characterize and validate primary liver cancer organoid cultures. Our aim is to prove the relevance of the models, their similarity to the original patient's tissue and their applicability as platform for anti-cancer drug testing for future personalized medicine approaches.

The outcomes of this research will have a huge impact on liver cancer patients, liver cancer researchers and also the 3Rs. Deriving organoid cultures that faithfully recapitulate the histoarchitecture, genomic and transcriptomic profiles of the original patient's tumour will allow modelling the specific disease of each patient, and thus finding new therapeutic treatments in personalized approaches. Not only patients, but also liver cancer researchers will extremely benefit from this research. Organoids could become a platform to facilitate the identification of new cancer mutations or new liver cancer biomarkers, thus increasing the odds to detect this cancer at earlier (and easy to treat) stages. Also, modelling human liver cancer in vitro will provide the researches with a platform to study the molecular mechanisms underlying this disease.
Finally, reliable human liver cancer models, that faithfully recapitulate the human disease, will prompt liver cancer researchers to replace the actual animal models used in liver cancer research (mainly rodents and dog) for human in vitro models, where to test novel hypothesis ranging from drug testing to human liver cancer biology. Consequently, it will have a tremendous impact on the application of the 3Rs in the liver field. Specifically, our plan of work will result in the replacement and reduction of animal use in liver cancer (specially PDXs for large drug testing) and, consequently, it will improve animal welfare (refinement), as less animals suffering from liver failure and/or liver cancer would be used. Briefly, according to Pubmed database, using "animal models liver cancer" as search criteria, 1,654 peer-reviewed articles (excluding reviews) have been published in the last 36 months (from 01-Jan-2014 to 01-Jan-2017). Restricting animal numbers to the very minimum, animal studies use on average 20 animals per project/paper (5 animals/group x 2 groups x 2 experiments), indicating that, in only 36 months, a minimum of 33,080 (1,654 publications x 20 animals/publication) animals have been enrolled in animal experiments, from which a minimum of 16,540 animals (1/2 of all assuming that 1 group develops tumour and the other is a control that does not) have developed tumours. Similarly, according to Pubmed database, using "Patient derived xenografts" as a search criteria, 2,460 peer-reviewed articles (excluding reviews) have been published up to now, taking into account that, as an average, each PDXs requires a minimum of ~50 mice, a minimum of 12,300 animals have been used to generated tumours.

Furthermore, our dissemination and communication plans (outlined in communications sectin) will be a key element to generate transfer the impact to this project to the public community and thus, will also have an impact on the application of the 3Rs.