IMPAKT Breast Cancer Conference 2013

Biomarker analyses in emerging data from PI3K/AKT/mTOR programme clinical trials

Dr Anna Vivancos – Vall d’Hebron Institute of Oncology, Barcelona, Spain

One of the important jobs you do, Anna, is to choose, or help doctors to choose, patients for phase I studies. Obviously very crucial and in this era of genomics and proteomics and everything that can be done in quite a refined way. What have you been doing in your research, can you explain to me?

Yes. We are participating in phase I clinical trial patient enrolment strategies based on biomarkers. What we do is we have been working for three years already in the Vall d’Hebron Institute of Oncology and we have been performing analyses mainly of hotspot mutations in oncogenes and these we have been doing with a Sequenom mass array instrument and we use several panels that allow us to assess different mutations.

So, for example, supposing a particular research group wants to know about the influence of a marker like PI3K, you would help that research group to find the right patients to recruit?

Right.

How would you do that?

The way this works is that patients that are candidates to be enrolled in a phase I trial in our hospital are sent to us so that we perform this mutation analysis. Based on that they are stratified or they are derived to one or other clinical trial.

So you need to have tests that can detect the mutation in question?

Exactly, so we have panels that cover some 200-300 point mutations and now we are planning to move on to next generation sequencing techniques to perform this breast screening so that we are empowered with a higher amount of information to provide to the medical doctor so that he can decide which is the best-suited clinical trial on-going at the moment in the hospital.

So how many biomarkers are really interesting at the moment in breast cancer, then?

There are quite a few. Besides the hormone dependent, the HER2, we have the PI3K mutations, the PI3K amplification, the AKT mutation. What we are looking onto is to be able to assess a wider number of genes that have a function in the PI3K pathway. This is the reason why we are moving to next gen sequencing.

And from your position as an advisor to research groups recruiting patients for phase I studies, what’s happening? Is this a very busy area at the moment?

It is and it is an area that is evolving very quickly and sometimes technology is faster than the knowledge, the functional knowledge. So sometimes we do it the other way around: that we have patients that have been already enrolled in our clinical trials who have responded extremely well to the treatment and then we go for whole genome sequencing of those patients to try and figure out the mechanism underlying this excellent response. Then many times we find ourselves with mutations that are tough to interpret because there’s no functional data for them so that many mutations we can figure out what their involvement is and many we are yet to know what, functionally, the relevance of those mutations might be. So we are a bit ahead of the functional assay so I would say this is a very relevant topic and something that in the future has to be properly assessed and probably studying the tumour patients’ xenografts or some kind of in vivo model of the tumour of the patient will be also required.

Of course, when you’re in phase I you’ve gone beyond the in vivo, beyond xenografts, you’re doing first in man or woman. In fact, how much has this not been done in the past and how much can you improve phase I studies by being refined like this?

I think there is a lot to be done. So we have already been trying to improve not only by detecting mutations but also we are making an effort in trying to assess whether those mutations are contained in a large number of the tumour cells in the particular patient. So, for instance, if you have a PI3K mutation that is present in only a subset of her tumour cells or is present in all tumour cells, that could have an effect on the response. So we are trying to assess that additionally, not only to detect a mutation – are you mutated, yes or no? – but how are you mutated? Are you mutated homogeneously or heterogeneously in your tumour.

Indeed, we’ve been hearing a lot about heterogeneity in tumours.

Exactly, so we implemented this assessment of heterogeneity one year ago in our studies. So we are starting to see some tendencies and we’re starting to see very interesting things.

Now, you’ve got a number of potential biomarkers that are targets for therapy and you’ve also been sitting on a panel here, I know it’s not your special expertise, but a panel to help teams choose how to combine different agents. How do you go about doing that because there are now quite a few possible combinations, aren’t there?

Yes. In fact, the PI3K, so far what trials are telling us is that the best scenario for these PI3K inhibitors is a combination scenario. This probably has to do with the fact that when we detect PI3K mutations, for instance in colorectal cancer, we find them as a sub-clonal event. So it’s a later event to KRAS mutation, for instance. So it appears as if PI3K mutations are collaborating or are improving some of the hallmarks of the tumour but probably it is not an essential target, meaning if you inhibit it with a single agent you’re going to be able to stop the tumour.

So in other words you’re finding there is an indication for dual targeting or multi-targeting?

Yes, I would say there is. I would say that everything is going in that line.

Right. Can you describe the ideal phase I study to me, then, in this genomic and proteomic era?

The ideal one, yes. So in the ideal world one should be able to study the whole patient tumour burden, meaning if the patient is carrying several metastases lesions. The ideal scenario would be to biopsy them all to have an idea about how heterogeneous they are and what are the driver mutations that are present as root mutations in that particular tumour. Of course we would need to define the subtype of this breast cancer, for instance, in this case, that we would do by expression and then I would probably try and assess also the circulating DNA levels and mutations that we can find in the circulating DNA. So I would go for a complete picture of the patient at that moment of treatment.

Would you advise teams, research teams, to actually contact centres like yours who are capable of giving this sort of guidance before embarking on a phase I study?

Yes. I mean, doing all this is going to be of high cost. So, yes, before embarking on anything I would suggest contacting someone who is technically proficient and who has hands-on on these kinds of settings.

Can you give me some kind of idea how much this sort of properly targeted and planned approach could improve the performance of phase I studies?

Well, hopefully I do think that the fact of being able to more properly profile the molecular characteristics of the tumour can lead us to improve the clinical trial itself. Although we do know, for instance, that the fact that you find a mutation is not always going to be a strong correlation to response because there are many other mutations in a tumour that might affect the behaviour in front of that treatment. So I would say we need to compile a lot of data so as to be able in the future to profile the tumour at a genomic level and tell the patient what is the best therapy to receive.

So, what’s the final message you’d like doctors to take out of this, then?

I would tell them that the genomics area is a very exciting area and I think this is something that everyone would agree with me. And that they have to open themselves to it and learn about everything that genomics can offer and integrate it into their practice every day. I would say this is the future in genomics and in breast cancer.

Anna, thank you very much for joining us on ecancer.tv.

You’re welcome, thank you for inviting me.