On behalf of my colleagues at Foundation Medicine we are presenting an abstract at the ASCO 2020 meeting regarding the types of mutations in BRCA1 and BRCA2 and their association with response to PARP inhibitors. This is actually a very topical theme right now because just one week ago the FDA has approved two PARP inhibitors for use in advanced metastatic prostate cancer, the first was rucaparib, which was approved on May 15th, and the second was olaparib which was approved on May 19th. So over the course of four or five days we had two PARP inhibitors approved for prostate cancer.

These drugs, we all think, work better or primarily in patients that have a type of DNA repair mutation called the homologous recombination repair mutation, also called HRR mutation. The most famous or notorious genes that are involved in HRR are the BRCA1 and BRCA2 genes. Interestingly, in prostate cancer, unlike breast cancer or ovarian cancer, BRCA2 is the predominantly mutated gene in that disease compared to BRCA1. So approximately 80-90% of the BRCA mutations in prostate cancer are BRCA2 and about 10-20% in prostate cancer are BRCA1.

One of the things that has not been explored previously, although there is reasonable hypothesis behind this, is whether the type of BRCA1 or BRCA2 alteration might be prognostic for the response to a PARP inhibitor. In particular, we’ve had a recent insight which is that one of the ways in which cancer cells become resistant or immune to PARP inhibitor treatment is that they take the mutation which made them susceptible to the therapy in the first place and they change it so that the open reading frame of the gene can be restored. These mutations are called restoring mutations or reversion mutations. Just to explain what these mutations are, just imagine that you have an abnormal version of the BRCA2 gene which causes a frameshift mutation resulting in a truncation of the protein. So let’s say that half of the protein, or 80% of the protein, is missing because of a truncating mutation. This actually becomes a liability when you use a PARP inhibitor because the cancer cell will die because of synthetic lethality. Now, of course the cancer cell wants to try to survive and one of the ways that it can do that is it can develop a second somatic mutation in the same gene, let’s say in BRCA2 in this case, and that mutation then restores the open reading frame and restores a functional full-length protein. By producing a full-length BRCA2 protein, now that cell no longer succumbs to synthetic lethality so the PARP inhibitor treatment doesn’t work, at least theoretically.

So what we did here in this abstract was we collaborated with Foundation Medicine and Flatiron Health and we looked at a combined database that involves both genomic data from Foundation Medicine and clinical data from electronic medical records from Flatiron. We were looking for patients that had either a BRCA1 or BRCA2 alteration and had received a PARP inhibitor on which we had clinical outcomes data. We found 29 such patients, so 29 patients that had either a BRCA1 or a BRCA2 mutation plus enough clinical data on their outcomes to a PARP inhibitor. Then we looked at those 29 patients and we asked, ‘Is the type of BRCA1 or BRCA2 alteration prognostic?’ We were very interested in looking at a class of mutations called homozygous deletions which means that both copies of the gene are deleted. In theory if a patient or tumour has a homozygous deletion in BRCA1 or BRCA2 this cancer is not able to develop a reversion mutation or a restoring mutation. So the hypothesis was that patients that have a genomic deletion, especially a homozygous deletion, meaning that both copies were deleted, would have better outcomes compared to patients that had other small nucleotide changes or small insertions or deletions, also called indels.

So this is, in fact, what we found. We found that about one-third to one-quarter of all of the BRCA mutations in either BRCA1 or BRCA2 were, in fact, homozygous deletions and patients that had tumours with these homozygous deletions actually had better or longer outcomes, both in terms of progression free survival, we actually used a surrogate of progression free survival which was length of time that they remained on the PARP inhibitor therapy, as well as overall survival. So when comparing the duration of therapy and overall survival in patients that had homozygous deletions in BRCA1 or BRCA2 versus all other mutations, the patients with the homozygous deletions actually had longer time on drug and longer overall survival. That was encouraging to us because it supported the preclinical biological hypothesis.
The second thing that we looked at was the zygosity of the mutation, in other words was one copy of the gene or two copies of the gene inactivated. We call that a monoallelic inactivation or a biallelic inactivation. The hypothesis there was that a biallelic pathogenic or likely pathogenic mutation would produce better outcomes to a PARP inhibitor than a monoallelic alteration. Here we were a little bit underpowered so we didn’t exactly meet the statistical significance but we showed a very strong trend towards better outcomes both in terms of prolonged time on PARP inhibitor therapy and prolonged overall survival in patients with BRCA1 or BRCA2 mutations that had biallelic mutations compared to monoallelic mutations.

So, in conclusion, our data from a small population of 29 patients appears to suggest that BRCA1 and BRCA2 homozygous deletion mutations are correlated with the best outcomes to PARP inhibitor therapy as well as potentially a second finding that biallelic mutations portend better outcomes compared to monoallelic mutations.

The PARP inhibitors that were included in this study were either olaparib or rucaparib and there were patients with both. There is no reason to believe that our fundamental hypothesis would change depending on which PARP inhibitor was used. There are at least two other PARP inhibitors that are now being tested in prostate cancer, namely niraparib as well as talazoparib, but in this particular ASCO abstract the two agents that we included were olaparib and rucaparib.

Of course this is a small study, it supports a hypothesis but it does not prove it. So we are now excited to design additional studies trying to answer the question about mutation type, for example homozygous deletion versus other, mutation zygosity, meaning monoallelic versus biallelic, and also there’s a third factor which is whether the mutation is a germline inherited mutation or whether it’s a somatic only mutation. There are two ways that we can try to move this study forward. One is to take the previously completed prospective trials and do a post-hoc analysis of those three factors – the mutation type, the mutation zygosity and the mutation origin, germline versus somatic – in the previously published studies. Or we can do something that would be more difficult which would be to design new prospective studies where the enrolment of patients is stratified by these factors. I think we will begin by doing the former first, in other words looking at previously conducted studies that may have captured this information, this genomic information, and trying to see in a post-hoc fashion if this hypothesis holds up. But what I would be most interested in doing in the future, and this is definitely more challenging, would be to design future studies employing PARP inhibitors in patients with homologous recombination mutations but then stratifying the patients into those different molecular categories.