Final results of the HERACLES trial in HER2-amplified colorectal cancer

Share :
Published: 10 Apr 2017
Views: 3521
Rating:
Save
Dr Silvia Marsoni - Candiolo Cancer Institute, Turin, Italy

Dr Silvia Marsoni presents trial data at AACR 2017 from the HERACLES trial, in which patient-derived xenografts were able to guide treatment choices for cetuximab-resistant colorectal cancer.

To view the slides from this presentation, click here.

Dr Marsoni discussed these results further with ecancer here.

I’m going to try to address some of your scepticism with my talk although… pardon me, I have a squeaky voice and terrible cold so bear with me through the presentation. These are my disclosures.

Just a little bit of background on colorectal cancer. We all know it’s a big killer and 50% of the patients that get colon cancer eventually come up with metastases so it’s still a big unmet clinical need. The funny thing about treatment with colon cancer is that we had chemotherapy, only chemotherapy, up to 2004 and then it was one of the diseases in which we started using targeted therapy in all the population. The funny thing is that we really didn’t have a target because the two drugs, cetuximab and panitumumab, that are monoclonal antibodies against the EGFR, the epithelial growth factor receptor, that oncogene is neither mutated nor amplified nor translocated in 99.9% of the population. So here we were giving targeted therapy against a target that was not aberrant. This is one of the things that makes colon cancer targeted therapy kind of different from everywhere else.

We then discovered, and our group also gave a big contribution to that, that however in patients that had KRAS mutation or NRAS mutation and even BRAF mutation cetuximab or panitumumab, either in monotherapy or in combination with chemotherapy, really were not effective. Therefore the low response rate that was 10% in the unselected population, yes, went up to 20% if you considered only wildtype patients for KRAS, NRAS and BRAF but that’s just a mathematical swift because 50% of the colon cancer patients are, indeed, RAS or RAF mutated.

So that was the situation that was therefore to be exploited. What we did, our group is a large group of about eighty people with all kinds of expertise including basic and translational ones, decided to harness the possibility of  molecularly looking into the tumour by setting up patient avatars using the old trick of the patient derived xenograft but implanting not lines of tumour but real patient metastases. We now have a very large colony, and by very large I mean very large, it’s about 1,000 patients in mice now. The reason why we wanted to do and we had this tremendous effort done by a colleague who is actually sitting here, Livio Trusolino and his colleague Andrea Bertotti, it’s because we wanted to make sure to capture also those mutations or anyhow aberrations that were in a low frequency in the population so we needed big numbers.

Then what we did, we treated a subset of this population with cetuximab which was the standard targeted treatment for the population and then did a molecular analysis of the landscape of the non-responding patients. Lo and behold, we figured out that these patients, these avatars, that did not respond were indeed for the majority KRAS, BRAF and NRAS mutated but also had other aberrations. Among those we found HER2 amplification. We now have seven, or more perhaps now, but at the time seven avatars that were HER2 amplified. Those avatars we treated in a preclinical randomised clinical trial and, as you can see from this slide, we used at the beginning… there’s a lot of armamentarium there, five or ten drugs that you can use for anti-HER2, small molecule inhibitors of tyrosine kinase, monoclonal antibodies, immuno-conjugates. These are the two standards or were at the time we started this treatment. Lapatinib is a small molecule tyrosine kinase inhibitor and trastuzumab is the standard monoclonal antibody. That is somehow surprising – as you can see, trastuzumab was completely negative and even lapatinib, which is dual HER1 and HER2, only induced tumour growth inhibition. So this combination really had a frank impact on the tumour. So I wonder if we had used, if my colleagues before have gone into the clinic with trastuzumab or lapatinib alone we probably would have missed this combination. Actually, I can tell you that neratinib, because we then found the mutation, is really behaving very well in colon if you couple it with trastuzumab or pertuzumab. So the problem is not that these are not drivers, the problem is that we tend to go from the discovery to the clinic without an experimental [iato - hiatus] which will make us understand what is the complexity of the interaction between the tumour cell and its lineage, in this case breast versus colon.

So that prompted us to go into the clinic again. We now started some seven trials, two diagnostic, three therapeutic and three translational to really understand what is the landscape of response and resistance to HER2 in colon cancer. By the way, I forgot to tell you that this is a tiny proportion, it’s only 3% overall or 6% of the population which is non-KRAS wildtype, but it’s still a bold number. There are a lot of colon cancers so 3% of 40,000 patients every year in Italy or 2.4 million in the world, as there will be in 2035, it’s a large number.

I’m going to just show you the first trial, the final results of the first trial, which combined lapatinib to trastuzumab. Endpoint and statistics are the same and lets go back immediately to the clinical results. We treated 33 patients, we got clinical benefit in 23 and of these 10 had an objective response rate according to RECIST with two complete responders. One of these two complete responders is still alive without evidence of disease at almost 36 months from the beginning of his treatment. We had very good duration of response and of DFS. This is just to give you an idea of the type of response, it’s the type of response that you see with targeted therapy when the tumour is truly addicted. Early responses up to eight weeks, the first is a patient with a lung disease and the second one with a liver disease, and really impressive responses which last in time, as you can see.

There is really no toxicity. We used a dose of lapatinib which is less, 20% less, of what it’s been used in breast cancer because we were very much afraid about diarrhoea, GI toxicity, since we were shutting down in a colon cancer patient two of the main epithelial growth factor receptors. As you can see, really there is no grade 3 toxicity. We had one case of diarrhoea and the other ones are fatigue and 90% of the patients were compliant with the treatment. The patient that has been on treatment for 36 months has no complaints whatsoever.

Just to put our results in perspective with standard treatment for third line and beyond because these patients were heavily pre-treated, the median number of regimens they had received was five. 75% of the population had four or more prior treatments and all of them, by the way, were cetuximab treated and cetuximab resistance. So we confirmed and we validated the PDX model. Here, as you can see, this is the proportion of response, objective response, that you have with chemotherapy in third line, it’s practically zero, it’s less than 5%. As was mentioned, anti-BRAF treatment against the V600E mutation is really not effective for questions that have to do with the crosstalk between the tumour and its ambiance. Anti-EGFR therapy in the EGFR wild population is really less, around 15%, whereas you jump to 30% with the anti-HER2 treatment and that kind of mirrors what you can see in microsatellite unstable high patients when treated with checkpoint inhibitors with immunotherapy. Coincidentally, both the microsatellite high patients and the HER2 positive patients are about 3%.

Then we were not satisfied and we had observed, if you see the cartoon on the right, that we had one avatar patient, mouse patient, that did not respond well to the lapatinib and trastuzumab therapy, he only got a growth inhibition rather than a tumour shrinkage. So we decided to tweak the combination by adding precision chemotherapy i.e. TDM1 which is an immuno-conjugate in which trastuzumab is conjugated with a cytotoxic cargo. So the idea is that the drug is targeting the cells and then the toxic cargo is interiorised into the cells and it’s a chemical bomb within the cells in addition to the fact that you are inhibiting the target. That’s just started in September last year; we’ve accrued twelve patients at the end of March, eight patients are already available for response and seven of these patients, 80%, showed a clinical benefit with two having a partial response. So obviously the numbers are too small but it certainly is comparing well with the previous study and perhaps maybe doing better.

In conclusion we can say that even if HER2 positive patients in colon cancer are only 3% of the population, they are still an important little piece of the cake that can be actually targeted with a new no chemotherapy, chemotherapy-free, combination of lapatinib and trastuzumab and in these patients, very heavily pre-treated, five median average prior treatments, you still can have a 30% response rate, objective response rate, with durable complete responders. A person that has 36 months in complete disease, I wouldn’t say it’s cured but I hope so. And with an overall benefit of 70%. Since these patients are also those that do not respond to cetuximab therapy we’ve got a new potential treatment for this population. Thank you.

Thank you so much, Silvia. So, again, if people with questions would like to gravitate to the middle microphone here. While we’re waiting for that you see where this was going, the idea that there are many ways of carving up cancers, the way we contextualise mutations and genetics in clinical cancers is going to take context into account but also the multiplex genetic factors. Here I want to ask you a question about how patients were actually screened. So use these standard molecular screening and then immunohisto, presumably? But if you could explain that?

The reverse, yes. Thirty years after we first treated our first patient with trastuzumab, breast cancer patient, we still don’t know what a HER2 positive breast cancer patient really is. So trying if not to avoid to at least limit this problem. Before going to the clinic we did a retrospective analysis on almost 300 patients to define our own diagnostic algorithm for defining HER2 positivity. We came out with something that was a hybrid between gastric cancer and breast cancer that is the patient had to be immunohistochemistry 3 or 2 but confirmed in FISH or SISH. The difference between us and the rest of the world is that we were really looking for what we call addicted patients – patients in which we were sure that the alteration was driving the disease. So we put a cut-off for positivity at 50% cellularity. So not only did the patients have to have expressing the protein of HER2 in a large amount but they have to do it in at least 50% of the cells. That’s important; if you keep in mind breast and gastric only have this cut-off at 10% so to be declared HER2 positive it’s enough to have 11 cells within 100 to be declared positive. I think that this is one other reason why there is a lot of confusion about results.

At the beginning we were screening this monothematically in a way; we screened 1,200 patients to get 69 positive patients of whom we treated 44 because there were a lot of patients with comorbidities. Now we’re doing a multiplex, as Hyman was saying before, now we are screening for several mutations, amplifications and even translocations like Trk so it’s not a problem anymore. But in any case we want the confirmation by FISH because we want to be sure that they…