Introduction
Activating mutations in the
PIK3CA are found in approximately 30–40% of patients with cancer and induce hyperactivation of the alpha isoform (p110α) of the phosphatidylinositol 3-kinase (PI3K) [
1‐
3]. In patients with HR+/HER2− BC, mTOR/mTOR pathway has been associated with endocrine therapy resistance [
4]. In addition, the role of this pathway is becoming increasingly important in HER2+ and TNBC [
5‐
7]. Thus, inhibition of PI3K in
PIK3CA-mutated BC has been a major focus in the last decade [
3].
Alpelisib is an orally bioavailable, small-molecule, α-specific PI3K inhibitor that inhibits p110α approximately 50 times as strongly as other isoforms [
8]. Following successful preclinical and phase 1 data [
4,
9], the SOLAR-1 phase III randomized trial evaluated the efficacy of alpelisib plus fulvestrant in 572 patients with HR+/HER2− advanced BC who had received prior endocrine therapy [
10]. A clinically relevant treatment benefit was only observed in the cohort of patients with
PIK3CAmut disease. In May 2019, the FDA approved alpelisib for the treatment of patients with advanced
PIK3CAmut HR+/HER2− BC.
Together with alpelisib, the FDA also approved the companion diagnostic therascreen®
PIK3CA test (QIAGEN Manchester, Ltd.) used in SOLAR-1 to select patients who had
PIK3CA mutations in tumor tissue specimens and/or in circulating tumor DNA (ctDNA) isolated from plasma specimens [
11]. Therascreen
PIK3CA detects 11
PIK3CA hotspot mutations, mostly found in exons 9 and 20 [
11]. In SOLAR-1, the type of
PIK3CA mutation did not seem to impact the main results [
10].
In this context, patients and physicians might choose not to use the therascreen PIK3CA test and use other available tests, which provide a more comprehensive mutational analysis of PIK3CA as well as other genes. This might lead to the clinical situation where PIK3CA mutations not detected by the therascreen PIK3CA assay, and thus not evaluated in SOLAR-1, are used to indicate alpelisib. To define the potential frequency of this clinical situation, here we aimed to evaluate the distribution of PIK3CA mutations in BC in relation to the therascreen PIK3CA panel.
Discussion
PIK3CA mutations have recently reached level 1 evidence for predicting benefit from alpelisib, an alpha-specific PI3K inhibitor, in combination with fulvestrant in patients with advanced HR+/HER2− BC previously treated with endocrine therapy [
10]. In addition, several trials are now evaluating alpelisib and other alpha-specific PI3K inhibitors in other BC subtypes harboring
PIK3CA mutations [
27]. Thus, there is a need to better understand the heterogeneity of the mutational landscape of
PIK3CA and, at the same time, relate this heterogeneity with the recently introduced therascreen
PIK3CA companion diagnostic assay approved to indicate alpelisib.
To address this topic, we performed a comprehensive evaluation of the distribution of
PIK3CA mutations in BC and made the following observations. First, although
PIK3CA mutations are highly heterogeneous, 5 mutations (H1047R, E545K, E542K, N345K, and H1047L) represented ~ 70% of all known types of
PIK3CA mutations in the dataset. Second, the therascreen
PIK3CA mutational panel would represent 72% of all the known
PIK3CA mutations and 80% of all patients with a known
PIK3CA mutation. Third, 83% of patients with 2 or more
PIK3CA known mutations would have mutations found in the therascreen panel; however, in 78% of the cases, only 1 single
PIK3CA mutation would be represented in the therascreen assay. Finally, the proportion of
PIK3CA mutations differed by BC subtype with HR+/HER2− disease having the highest proportion, followed by HER2+ disease and TNBC. Although less frequent in the HER2+ and TNBC, the proportion is not negligible and several studies, including pivotal or registrational clinical trials, are focusing on these two populations [
5‐
7]. To our knowledge, this is the first report to perform a comprehensive analysis of
PIK3CA mutations in BC and to relate these findings with the type of mutations captured by the therascreen
PIK3CA assay across the three main subtypes of BC.
The SOLAR-1 phase III trial that led to the approval of alpelisib used the therascreen
PIK3CA 11-mutation assay in tumor tissue to identify
PIK3CA mutations [
10]. From a total of 1173 patients tested for
PIK3CA mutation status that had interpretable results, 341 (29%) patients had
PIK3CAmut disease [
10], a proportion which is very similar (28%) to our predicted results if the assay would have been performed in our combined dataset. More importantly, mutations in exon 9 versus exon 20 predicted similarly the degree of benefit to alpelisib in SOLAR-1 [
10]. Thus, based on these results, the FDA approved the use of this assay in tumor and plasma samples as a companion diagnostic to indicate alpelisib. The approval of therascreen in plasma samples is based on a subanalysis of the SOLAR-1 trial which showed that
PIK3CA mutations identified in plasma samples were also associated with treatment benefit [
28].
Our results have important considerations for patients and physicians. In certain parts of the world, determination of PIK3CA status is commonplace using various types of sequencing-based assays. Some of these widely used assays such as Foundation One CDx or Guardant360 cover most or all exons of the PIK3CA gene. Thus, it is highly likely that mutations which are not part of the therascreen PIK3CA 11-mutation assay will be identified with other assays and treatment decisions will be made. In other parts of the world that have not yet implemented somatic genetic testing in BC, the fact that the therascreen panel misses ~ 20–30% of patients with known PIK3CA mutations might be a reason to choose more comprehensive PIK3CA panels.
Critical questions raised by our results are if patients with
PIK3CA mutations which are not part of the therascreen panel, or hotspot and non-hotspot
PIK3CA mutations identified using sequencing-based assays with higher sensitivities than therascreen, will benefit from alpelisib. For example, mutation N345K represented 5.5% of all
PIK3CA mutations in the analyzed dataset and is not captured by the therascreen assay. This mutation was the fourth most frequent
PIK3CA mutation in the BC dataset, and COSMIC [
29] and OncoKB [
30] datasets consider it pathogenic (score 0.95) and oncogenic. Moreover, N345K confers a gain of function and it has shown to increase sensitivity to PI3K inhibitors in preclinical models [
23,
24]. A similar situation exists for the sixth most frequently observed
PIK3CA mutation, E726K, although OncoKB [
30] states that there is conflicting and/or weak data describing the oncogenic function of this mutation, it has been shown that as a single mutation it is weakly activating but as a double mutation (with E545K or H1047R) it is synergistically activating [
31]. It is important to notice that the vast majority of E726K mutations are found precisely as double mutants in BC [
31]. On the other hand, some less frequent mutations, as G1049R, have demonstrated strong driver activity in a mutation assessment platform. G1049R exhibited activity levels similar to the E542K variant with 20-fold higher frequency [
24]. Thus, better functional characterization of these and other non-hotspot
PIK3CA mutations together with clinical evidence that predict benefit to alpelisib and other alpha-specific PI3K inhibitors is now of uttermost importance. At the end of the day, each type of
PIK3CA mutation should be considered a biomarker by itself.
Another interesting observation is that ~ 4% of all BC, or ~ 12% of all patients with
PIK3CAmut BC, have double
PIK3CA mutations. Preclinically, double compound
PIK3CA mutations result in increased PI3K activity and downstream signaling compared to single hotspot mutants in nontransformed cells and in HR+ BC cells [
31]. More importantly, these compound mutations seem to predict for increased sensitivity to PI3K alpha-specific inhibitors compared to single hotspot mutants in both preclinical models and also in selected patients with BC treated in early phase 1 trials [
31]. According to our results, the therascreen panel would not capture well double
PIK3CA mutations since only 5% of patients known to harbor 2 or more
PIK3CA mutations would have mutations represented in the therascreen panel. Thus, if double mutations are confirmed to be a biomarker of ultra-high sensitivity to alpelisib, the therascreen assay might not be ideal for this purpose.
Our study has limitations worth noting. First, we did not evaluate the actual analytical concordance of the therascreen assay versus other sequencing assays. In other words, we assumed that the results of the combined dataset using various sequencing-based strategies was the gold standard and that the therascreen assay would identify 100% of all the
PIK3CA-wild-type tumors as “no
PIK3CA mutation detected” and 100% of all the
PIK3CAmut tumors in the combined dataset as “
PIK3CAmut” if the type of mutation was on the therascreen mutation panel. However, the differences in the sensitive and specificity of the various sequencing assays will affect the concordance rates among them [
32,
33]. According to the FDA therascreen
PIK3CA assay specification sheet, the overall percent agreement between the therascreen assay and an NGS-based assay in SOLAR-1 was 94.7%. Second, the next-generation sequencing assays and the methods used across the 10 studies evaluated in our study are highly heterogeneous and most are not standardized. Third, the analyzed datasets were mostly from primary tumor samples and acquisition of new
PIK3CA mutations has been described in the metastatic setting in 8–10% of the cases [
34]. Whether the frequency and spectrum of
PIK3CA mutations would change if metastatic-only samples had been analyzed is currently unknown.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.