Rita Fior uses zebrafish to do basic and translational research in cancer at the Champalimaud Foundation. A few years ago, having to deal with cancer in her family led her to design a test, based on her animal “model” – that would allow doctors to choose, among the available chemotherapeutic options, the best one for a given patient. How? Using the little zebrafish as “avatars”, as personalised “alter-egos” of the patients. Tumour cells from a patient are injected in the fish, generating the “avatars'' that will then be submitted to the treatment options available for that patient. The idea is to perform this test before administering treatment in order to choose the most effective for that particular patient, thus avoiding giving treatment by trial and error, as can happen in some cases. The results are promising, but this approach is still a long way from being able to be used in daily clinical practice.
Rita Fior’s most recent project concerns rectal cancer, where patients may be submitted to radiotherapy before surgery. In this case, the problem is that nobody knows if it is worthwhile to give this treatment to a person for several weeks – a treatment that can have serious side effects – because there is no way to predict whether the tumour will respond to the radiotherapy. Rita Fior believes that, one day, this type of predictive tests using avatars may help oncologists to achieve the ultimate treatment personalisation, in the interest of patients’ well-being and quality of life.
You work with zebrafish. What do these laboratory animals have to do with the treatment of human cancers, and in particular with colorectal cancer?
In my lab, the Cancer Development and Innate Immune Evasion Lab at Champalimaud Research, we are developing a test to help MDs choose the best available therapy for each individual patient, using zebrafish avatars. And we started this work in colorectal cancer.
What is an “avatar”?
It’s a personalised in vivo model of the patient. To create an avatar, we obtain tumoral cells from the patient, be it through a biopsy, a surgical procedure or from biological fluids that contain many tumor cells. Then, we transplant these tumoural cells into a zebrafish embryo, which constitutes the in vivo model I mentioned and that we call the patient’s “avatar”. This way, we now have the patient’s cells inside a living organism. We do this because, in biological terms, it puts the tumour cells in a more realistic environment: they are not in vitro, in a Petri dish. Lastly, we apply the different treatments to the little fish, to see which of them will be more efficient against the tumour.
Why do you need to do this?
The idea of using avatars has to do with the way cancer patients are treated today. Therapies are approved in clinical trials, which involve many patients. These therapies have an average response rate – in other words, we know the average number of patients who will respond to a given therapy. Doctors and scientists perform big clinical trials, collect all the data from the trials and define international therapeutic guidelines for each stage of each type of cancer.
Very often, those guidelines contain several equivalent therapeutic options. Therefore, oncologists have more than one option at hand to treat a patient, which can be chemotherapies, radiotherapy, combinations of both, etc.. And those options exist because it was demonstrated, during the clinical trials, that they provide a good therapeutic response.
But the thing is, when we have a therapy with a 70% rate of positive responses for the treatment of a given cancer – which by the way is rare –, we don’t know if our patient is one of the lucky ones that respond to the therapy or belongs to the 30% that do not respond. So, in my lab, we believe that there is a need for a test to previously assess the patient’s tumoural cells, and see if they actually are sensitive to those options – and also, whether one of the options is better than the others for that specific patient.
And in the case of colorectal cancer?
For colorectal cancer we have two big projects. One is in the context of so-called “neoadjuvant” therapy – that is, a radiotherapy and chemotherapy treatment administered before rectal surgery to, for instance, reduce the tumor to make it operable. This work is ongoing, here at the Champalimaud Foundation, in close collaboration with Oriol Parés, a radiation oncologist who works at the Radiotherapy Department, and also with surgeons such as Laura Fernández and gastroenterologists like Ricardo Rio-Tinto and Paulo Fidalgo, among others. We receive biopsies of the rectal cancers, and we want to determine, using our zebrafish avatars, if we can predict each patient’s response to the radiochemotherapy that will precede the surgery.
Our other project is in the so-called “adjuvant” context, and it is the most advanced of the two. Following surgery, colorectal cancer patients are also treated with chemotherapy to reduce the probability of tumour relapse. In theory, the surgery has removed all the tumoural cells, but often some circulating micro-metastases are left behind. And it was demonstrated, again through clinical trials, that doing a post-surgical chemotherapy treatment effectively reduces the probability of relapse. In this case, we take a sample of the tumour that was removed during surgery and what we do is put that sample into the zebrafish and give the animal the same treatment that was given to the patient after the surgery. That way, we can see whether the “avatar” responds or not to the treatment and whether there is a match between the “avatar” response to the treatment and the actual response of the patient who “donated” his tumoural cells for the test.
We are still at the study phase of both projects: testing the predictive power of the avatars – that is, the measure to which they can allow us to determine what will be, in the future, the patient’s response to the treatment. But it’s just a model, an idea that isn’t yet implemented and that needs to be validated. For now, we are studying if there is a good match between what happened to the patient and what is happening in our little fish.
How promising are the results?
I think the results are very promising and I hope to publish them before the end of the year. Currently, in terms of post-surgical chemotherapy, we have 82% matches and a positive predictive value. We are able to predict, in 82% of the cases, that if a chemotherapy produces a response in the avatar, the patient will also respond to that chemotherapy.
We already have 50 patients in our study. In 2017, we predicted the outcomes correctly in four patients out of five, an 80% positive match. Now, with 50 patients, we have a rate of 82% correct predictions, which strengthens our 2017 results. In this context, the avatar can help choose the better therapy among the existing chemotherapy options.
What about the avatars’ usefulness for predicting sensitivity to radiotherapy in rectal cancer?
In the neoadjuvant context in rectal cancer, where radiotherapy is used (sometimes with chemotherapy), there are no different options as in the previous context. In this case, the issue is that there may be tumors that are not at all sensitive to radiotherapy or chemotherapy. But, as nobody knows how to identify them, there is a risk of submitting patients to four or five months of radiochemotherapy, with toxic side effects and no benefit whatsoever for the patient. Perhaps, had we known, it would have been better to send these patients directly to surgery. So, for rectal cancer, our goal is to use the “avatar fish” test to help identify which patients should not be submitted to radiotherapy because their tumour does not respond to radiation.”
In the other context, as I already said, the avatar can help choose between the existing chemotherapy options – and here, the results are very promising.
All the patients are from the Champalimaud Foundation?
They are from the Champalimaud Foundation and the Amadora-Sintra Hospital [Hospital Prof. Doutor Fernando Fonseca, EPE]. We have a long-standing collaboration with them.
In the case of colorectal cancer, the test could prove very useful to plan, in the most personalised way possible, the protocol that will be applied to each patient. How long do you think it can take for this approach to become part of the clinical practice?
Our next step – which I’m writing up right now – is to perform a so-called randomised clinical trial. It won’t involve colorectal cancer patients, but metastatic patients, most likely with ovary and breast cancer. We are going to do this trial to compare “the doctor’s choice” with “the fish avatar choice”. The idea is to test whether it is beneficial to have an assay to help clinicians and guide the therapy. Only after that, may our test be accepted for clinical use.
Where did you get the idea, in the beginning, of creating zebrafish avatars?
It stemmed from my personal story. My mother had lung cancer. I was doing my post-doc at the iMM (Instituto de Medicina Molecular João Lobo Antunes), in Lisbon, and discussed what was going on with a doctor and the treatment my mother was taking. And she turned to me and said: “Well, at the IPO [Portuguese Oncological Institute] we give the patients [I don’t remember which drug]”. To which I replied: “But I don’t think she’s taking that, she’s taking something else”. She answered: “Well, at the IPO we use a treatment, but at other places they use other treatments”. Says I: “I don’t understand. If my mother were at the IPO she would be taking another chemotherapy? So it can go well, or it can go wrong? How does that work?” It was then that I started to understand how this works.
I learned that there were guidelines and equivalent therapeutic options – which are often very similar, but nonetheless different. And I started to ask: “But don’t you do any test, you have no way to test the tumoural cells, just as they do antibiograms for bacterial cells?” I was so ignorant… And the answer was always: no, no, we don’t do that. That’s basically the story.
At that point, I was getting irritated. And since I was working in embryology, in developmental biology, with zebrafish, I attended a summer course at Woods Hole (in the US) in the middle of all this process of my mother’s treatment.
Were zebrafish already being used for this kind of tests?
Zebrafish are very much used to assess the effectiveness of potential drugs in a living organism, but not with a human cancer inside them. I asked my teacher in Woods Hole why this wasn’t done in zebrafish – I already knew that these efficacy tests were not done in vitro with tumoural cells because they weren’t predictive.
Actually, the gold-standard for this type of tests in oncology was the mouse. You put tumoural cells from patients into mice and you test the treatments. Only this takes months, so it’s not feasible in the clinic. It’s used in big American cancer centres, such as MD Anderson, in patients who, if they relapse, will have their mouse-avatar ready, but it is very expensive and ends up not being doable.
And I, who was working on the zebrafish, asked why not do the same kind of test with avatar fish instead of mice? My teacher recommended that I talk to someone else, and I went to see what was being done. There were already people trying to do this.
And were those tests reliable?
No. I had to design a test from scratch. I think my training in developmental biology made a difference with respect to what had been done until then in the zebrafish. I looked at the cells I had introduced in the fish to be sure that they were human tumoural cells. Under the microscope, I saw a beautiful image, just like the ones that had already been published, but I felt something was wrong.
It so happens that the zebrafish has these macrophages [immune system cells] that ingest the tumoural cells, but people thought they were seeing metastases. I realised that things had to be done in a more detailed, careful, way. Then, I discovered a way to inject the tumoural cells into the fish in a place where the cells won’t die, and of quantifying everything – doing rigorous work, in sum – to reach the optimisation we have today.
The test I designed, following all this, takes four days, plus a week to analise the results, which means it is designed for two weeks. In most cases, cancer patients have to wait for other results and it sometimes takes three weeks before they can start treatment. Our test was designed to be performed within that period of time.
And that’s it, that’s how I started optimising and developing the model. Because although several labs had already tried, things were less than perfect at that time. I now believe that at the moment, we are one of the best labs in the world doing zebrafish avatars.
How long did it take?
Four years.
How long do you think the avatars might take to enter the clinic, if everything goes well with the clinical trial you are preparing?
I don’t know. Four, five years? But maybe this is my optimism and my lack of clinical experience talking.
The test could have a great impact, namely for radiotherapy in rectal cancer.
I think it could help. Our big handicap is that we need tumoural cells for the test and in some cases they are very difficult to obtain.
Nonetheless, there are many patients who go directly to surgery. When that happens, we receive a sample of the tumour, in case the patients need to do chemotherapy afterwards.
But for the radiochemotherapy testing, we need a biopsy. For example, in rectal cancer, we always get a rectal biopsy because it is necessary to perform one to confirm the diagnosis. After that, some patients go to radiotherapy – and only after that, to surgery. If we have access to the biopsy, then we can do the test.
What, in your opinion, are the challenges that scientists and doctors face when translating scientific results such as yours into clinical practice?
We have to learn to communicate. Most doctors think the test would be useful for them. But I’ve already understood that it is not enough to show that the test is predictive. Perhaps that’s the difference between a scientist and a clinician. I thought that before going on to a randomised clinical trial, we could do some case-studies, for example in patients with highly advanced cancers for whom nobody knows very well what to do. That we could use the avatar, because we had already shown its predictive power in chemotherapy.
But this isn’t possible. We have to do the clinical trial and see what level of evidence we will get and if it is high enough to convince the medical community. Now we’re designing the trial’s protocol – it’s not a project, it’s a protocol, as the clinicians call it; I’m learning to make a protocol like doctors do [laughs] so that we can move forward. Once it is written, it will have to be approved, financed and performed. But I feel confident. Luckily, I have help from several doctors who are helping me write the protocol.
As to colorectal cancer, doctors will not want to risk doing new things right away. It’s still too early. Of course, if we show that our avatars are good predictors for radiotherapy, we will also be able to go forward in a similar way. But there you have it. I don’t know if they will be willing to do that. At the same time, I understand that these are patients we are speaking of.
Be as it may, in rectal cancer radiotherapy we still don’t have the predictability levels that we have achieved for colorectal cancer chemotherapy. I still don’t understand why this is so, but it could be linked to the sampling of the biopsy itself, to the fact that we don’t have direct access to the tumour. Pre-operative radiotherapy is only performed for rectal cancer. The biopsy they give us to do the radiosensitivity test in the zebrafish might not be representative of the rectal tumour – and could therefore result in more false negatives than it should. We still have to learn to select those biopsies to ensure that we have the right one. We're still learning.
By Ana Gerschenfeld, Health & Science Writer of the Champalimaud Foundation.
