WIN Symposium 2013

Strategies to address inter- and intra- patient tumour heterogeneity: PANGEA

Dr Daniel Catenacci - University of Chicago, Chicago, USA

This is issues that have been raised by a number of speakers at the conference thus far referring to inter-patient heterogeneity molecularly from one patient to the next within the context of this cancer. And when you really look into that almost patients are really an n of 1, as we say, that they’re an individual, even looking at 200 genes they almost always have a unique profile. Then intra-patient molecular heterogeneity, there are really two examples of this. One is through space within the patient, within the primary tumour and from the primary tumour to the lymph node into the metastatic lesions, that can be a snapshot in time, can be different and can evolve. Sometimes it is identical from one set to the next, in other times it can show differences. Then a second example is over time, intra-patient molecular heterogeneity over time where most often there’s a selective pressure from a treatment and you can select for resistant clones. We have demonstrated this on a number of occasions, so this is the summary of inter- and intra-patient heterogeneity.

So if we talk about the first one, inter-patient heterogeneity, what does this mean for treatment?

This is one of the major hurdles and challenges thus far in the last ten years in the era of molecular targeted therapies and two examples: clinical trials that have not used targeted therapies for targeted populations but rather targeted therapies for all of the population. Many of those trials have been negative or have been marginally positive or marginally negative. In the second instance where trials have selected a targeted therapy for a specific subset or a targeted population, an example in this tumour type would be HER2 amplification, they have been positive. So the challenge, though, is as we do understand the profound inter-patient molecular heterogeneity where really this is a subset of the disease, multiple subsets, it becomes a challenge and a strain on the classic clinical trial design which is basically testing one therapy versus another and requires a certain number of patients to get adequate power to prove a difference. So when you have small subsets of a disease, 5% of the disease having a specific molecular aberration, you have great difficulty in accruing to the trial in a classic sense. This is what our clinical trial attempts to do, PANGEA, which is Personalized Anti-Neoplastics for Gastro-Esophageal Adenocarcinoma, which attempts to address this inter-patient heterogeneity by allowing a platform of drugs based on the patient’s profile at presentation; they get allotted a specific treatment based on this and trying to address this inter-patient heterogeneity.

All these patients would have the same type of tumour in the old classification.

Yes, and the pathology will be an adenocarcinoma and it will be either from the stomach, gastric, or the GE junction, gastro-oesophageal junction, cancer. But within that, like we’ve been discussing, there are a number of subsets based on molecular profile and so our particular iteration of this trial is patients will get standard chemotherapy with a specific targeted agent added to it based on what their profile is right from the beginning, first line therapy.

At the end of the trial how will you match the genetic abnormality with the right treatment?

This is a very important question and in a sense is really what is being tested in this trial, the algorithm by which we assign treatment, this is what is being tested. So the first iteration of the trial is a pilot phase, it is a non-randomised trial, patients will be given chemotherapy with a specific targeted therapy based on their molecular profiles, we discussed, that profile is derived from next generation sequencing, targeted sequencing, and mass spectroscopy expression analysis with follow-up low throughput testing IHC FISH when necessary to allow for progression through an algorithm and treatment assignment based on what their molecular profile is.

Maybe just two words on intra-tumour heterogeneity because this is going to be a problem for therapy in the future.

This is another challenge and this is now, even if you get it right at the beginning in terms of matching a therapy to a specific aberration, this has been brought up, we have shown it, so many have shown this, that over a variable amount of time selective clones of resistance occur. And so in this trial, which we call PANGEA Biologic Beyond Progression, or BBP, we biopsy at the time of progression a progressing lesion and then we re-run all of the testing, re-run the algorithm and for the most part we feel that the genomic aberrations will still be the same, perhaps additional genomic events will have occurred providing resistance. Another scenario is we do see loss of genomic aberrations and acquisition of different ones and so in that event we would reassign treatment based on what is currently happening. So this is addressing this intra-patient heterogeneity over time, over resistance. So we allow crossover within the five arms that we’re dealing with. Ultimately when it is a randomised setting that crossover will be occurring in the treatment personalised arm whereas in the placebo controlled arm they will be going from placebo to placebo. So we’re testing a personalised therapy versus what we do now which is basically chemotherapy first, second line plus or minus third line for those who can achieve it. So it is addressing some of the concerns, it’s forcing some patients into one of five arms, sometimes not perfectly, but it is attempting to do that and it still retains some of the classic testing of clinical trials with placebo controlled, which is one of the problems that we’re faced with now with these n of 1 trials where we just have a drug matched to a therapy but we don’t have a control and we have a number of case reports with success stories. But those are only the reported ones, we don’t hear about the ones that don’t work as often as we hear about the positive ones. So in this scenario we will have all patients on trial, we will see all those that respond, all those that don’t respond, we’ll have biopsies before and after to try and see why that is that they don’t respond or when they stop responding.

All of this data coming out of these trials, it’s going to be incredibly difficult to manage.

True, and so this is why we’re starting with a small trial with 68 patients and at our centre, one centre, to show proof of concept. If this is successful then we would open that, the randomised phase II is 192 patients, that would be through a co-operative group and that would be extending out to a number of sites. This will be a challenge with data management, etc., has also come up in a number of the talks so IT and database management will be very important for this.

So the future of clinical trials is completely changing?

I think so, I think that the knowledge that we’ve gained over the last five to ten years with this revolution with high throughput data is challenging our trial designs, as we mentioned, it’s not one size fits all. Those drugs that are put in that scenario often fail, the ones that are doing well have selected a targeted agent for a targeted genomic event that is driving the tumour which provides benefit for a period of time until they develop resistance. So the classic trial designs are obsolete with targeted therapies.