We presented the results of the SOLAR-1 study which is a phase III trial comparing alpelisib, a PI3 kinase alpha inhibitor with fulvestrant compared to placebo plus fulvestrant in patients progressing on AI-based therapy. We conducted the study in two cohorts based on tissue analysis, either archival or fresh. We determined the PIK3CA status of these patients and separated them into PIK3CA mutant and PIK3CA non-mutant. The primary endpoint of the study was the median PFS in patients with PIK3CA mutation. The PIK3CA non-mutant cohort had a proof of concept statistical endpoint and in our study the benefit was clear only in the PIK3CA mutant cohort where we saw a prolongation of median PFS from 5.7 months to 11 months which is both statistically significant and clinically meaningful. The hazard ratio for this comparison was 0.65. The difference in median PFS in PIK3CA non-mutant patients did not meet the pre-specified criteria suggesting that the benefit there is lacking or it’s minimal.
We subsequently conducted several subgroup analyses including looking at the impact of line of therapy on the benefit of alpelisib. Specifically we looked at patients in the first line setting and those in the second line setting and we further subdivided patients in the first line setting to those that are endocrine sensitive and endocrine resistant. As you progress through these three subgroups you see clear improvement in the benefit and the largest magnitude of effect is seen in the second line setting where patients go from 3.7 months median PFS in the placebo arm to 10.9 months in the experimental arm with alpelisib. The hazard ratio for that comparison is 0.61.
I suspect that this change as you progress through multiple lines of therapy reflects the temporal evolution of ER positive breast cancer where sequential acquisition of additional alterations, activation of the pathway, decreased dependency on ER are all responsible for this synergistic effect that we see between alpelisib and fulvestrant in that particular setting.
We did not have a sufficient amount of patients to specifically and robustly look at patients with or without prior CDK4 treatment. As a matter of fact, the number of patients with prior CDK4 therapy was fairly small and all we can say is that the hazard ratio in that patient subset is consistent with the analysis in the overall PIK3CA mutant cohort.
A very, very interesting analysis is the one where we moved away from tissue based analysis and simply looked at PIK3CA mutation status in circulating tumour DNA which was collected at baseline, reflecting, you would say, the true nature of disease at that particular time point, as opposed to assessment that we have from archival material which can be many years before the initiation of therapy. When we look at PIK3CA mutant versus PIK3CA non-mutant patients based on the ctDNA analysis we again saw a very wide benefit, again going from 3.7 months to 10.9 months with a hazard ratio of 0.55. This clearly illustrates the importance of liquid biopsies in this setting and the potential utility of this biospecimen easily obtainable blood sample, basically plasma sample, which can be used for a very rapid detection of PIK3CA mutation status. What remains to be answered is how different biological features of disease, different disease burden, affects the shedding of circulating tumour DNA and the mutational status determined based on that sample. We hope with additional work on our sample set we’ll be able to define some of these parameters.
Either way when you take tissue or plasma it is clear that detecting and determining PIK3CA mutation is critically important because now we do have this mutation, it’s a predictive biomarker basically, for alpelisib and we have clear separation of the benefit from alpelisib seen in PIK3CA mutant versus wildtype PIK3CA. Since PIK3CA mutation can be tested with dedicated assays but it can also be detected as part of larger next generation sequencing based panels I’m hoping, and this should be good for our patients, that having one of these oncogenic dependencies targeted with, this time, PI3 kinase alpha inhibitor will be able to launch a series of other programmes focussing on other or less frequent but potentially equally important oncogenic drivers such as AKT1, HER2 mutation as opposed to amplification and others.