Human tissue models for the replacement of mice in cancer research: Are we there yet?

I recently came across a review which asked if it’s time for peer reviewers to request ‘organ on a chip’ models instead of animal validation studies, and it got me thinking – are we there yet?

As cancer researchers, if we submit an article for publication that contains only data from cell lines, we’re often asked by peer reviewers to carry out animal studies – usually in mice. This review suggests that it might be nearly time for reviewers to ask for human tissue work instead – maybe some of our newest human tissue models are good enough to replace these types of animal studies?

Personally, I’m a big advocate for human tissue work in cancer research. Anyone who collaborates with me knows that I favour ex vivo / 3D culture of human tumours over mouse models. Of course there are ethical considerations here around reducing the number of animals used for research, but my opinion stems mostly from the science – because of the very simple fact that mice are not humans. The differences between mouse biology and human biology are too wide ranging, with far too many variables to feasibly take into account. Frankly, neither have been characterized rigorously enough to pick apart their similarities and normalize for their differences.

Of course, to date, mouse xenografts (and more recently, patient derived xenografts) are pretty much the best models we’ve got in terms of testing new cancer drugs in a better model than cell lines, without the ethical risks of testing them in living humans too early.

As such, many scientists like me around the world have been developing a huge range of human tissue models, usually removed from a cancer patient at biopsy or surgery, and donated for research. The idea being that one day we’ll get these cells or tissues to survive outside the body while changing as little of their biology as possible, and treat them with experimental drugs for research. Ultimately, replacing animal models.

Roughly speaking, these types of models fall into three categories: explant cultures, organoids/tumouroids, and ‘organ on a chip’ models.

Explant cultures involve taking a small piece of donated human tissue, and trying to keep it alive for a few days in an incubator, helped along by different nutrients and materials. One of the main benefits of explants is that the tissue stays whole, rather than the scientist isolating out particular cell types. The original architecture of the tissue, and range of different cell types within it can remain somewhat intact (this isn’t perfect, but it’s improving). I’ve been using a version of explants for the last five years, testing new drugs in prostate cancer, as part of my fellowship project ‘SCREEN’, kindly funded by Prostate Cancer UK.

Organoids, or specifically within our field of cancer research – ‘tumouroids’, represent human tumour cells that are grown in 3D outside of the human body, including multiple key cell types and environmental factors. Here the structure of the tissue does not remain intact as with explants, but key molecular signals added by scientists can induce the cells to organise themselves in the same way that the original tumour would have done in the human. These can be cultured for longer than explants generally, and offer more flexibility for the researcher to tweak particular aspects of their behaviour.

Organ on a chip models can be based on either of the above, but include additional extras like midrofluidics (a system that allows for nutrients to flow over and around the cells in the same way blood would in the body), which can encourage blood vessels to grow and feed the tumour, as they would in a human. These are getting ever closer replicating human tumours outside of humans.

But are any of these good enough to replace mouse experiments yet? My gut says no – but we really are very very close.

One of the issues with this branch of cancer research is that there are just so many different types of models being investigated. Yes, they do fall roughly within three categories, but within each of these categories, there are dozens if not hundreds of iterations being researched around the world. In my view, to properly validate them, we need a consensus – not a new model every five minutes! This consensus will be difficult to achieve, as within the structure of academic research we are encouraged to generate new intellectual property (IP), and we’re generally taught that to get a model validated and used in the clinic, we need to either commercialize it ourselves, or licence it to a company who will develop it for us. This is the approach that will get us the next grant, the next paper, the next promotion – i.e. more cred, and potentially personal financial gain. So why would we bother to further develop, independently validate and rigorously characterize someone else’s model, when we could be changing it slightly to add our own ‘unique selling point’ and branding it as our own?

My hope is to reject this way of thinking. Over the first few years of my new lab, I am to compare and contrast the leading models from around the world in a fully independent setting, where I’m not backing any horse in the race – where I have no allegiance to one human tissue model over another – and just purely try to see if the best one(s) reflect how humans actually respond to anti-cancer treatments. If we can pull this unbiased validation and rigorous characterization off, then I truly believe the peer reviewer mentioned in the paper linked above should absolutely be asking researchers to validate their research in these human models rather than animal models.

It’s worth mentioning that I also tweeted this paper and got varying responses. While one person replied a jokey ‘I wonder what reviewer 3 wrote in the report :)’, another expressed caution:

And I agree somewhat – we still don’t have strong enough validation in my mind to fully replace animal studies. But should reviewers be requesting more human work incrementally as our models get better and better? Yes, I think so. They’re certainly worth carrying out in addition to animal studies – just maybe not instead of animal studies just yet.

Dr Dania Movia from Trinity College Dublin commented on the frustration of human tissue researchers still being required to validate their findings in animals instead of humans – why do we think of mice as a gold standard for how human biology behaves? It makes no sense, and I couldn’t agree more! While mouse models bring some valuable extra data that human models don’t have perfect yet, they’re certainly imperfect in a lot of other ways, and not the right place to validate a human model.

Check out the review linked at the top of this blog if you’d like to read a more technical summary of where the field is at (though the review is not specific to cancer research). And let me know what you think! Are we ready to replace animal models with human models today? Will be there in a year, in a decade, or ever?