Introduction
Mutational profiling of advanced solid tumors is an important component of early phase clinical trials testing drugs in molecularly defined patient populations. This approach, which currently involves the application of next-generation sequencing (NGS) technologies, is used to match a specific therapy to the particular somatic molecular alteration within a patient’s tumor [
1]. Although there is a variety of approved systemic therapies for metastatic breast cancer (MBC) that prolong progression-free or overall survival, including endocrine therapy, chemotherapy, and HER2-targeted therapy, MBC is an incurable disease and a leading cause of cancer death worldwide. The identification of subsets of breast cancers with overexpression of the
HER-
2/
neu oncogene by IHC or amplification by fluorescence in situ hybridization (FISH) has led to the addition of targeted therapy for this subtype and improved survival [
2,
3]. Additional molecular alterations such as somatic
PIK3CA mutations, part of the PI3K/AKT/mTOR pathway, are commonly identified in breast cancers but are not yet used in the selection of approved therapies [
4‐
7].
Integrated Molecular Profiling in Advanced Cancers Trial (IMPACT) and our community hospital program Community Oncology Molecular Profiling in Advanced Cancers Trial (COMPACT) were studies at Princess Margaret Cancer Centre (PM) to provide molecular profiling data for advanced solid tumor patients treated at PM and local community hospitals [
8]. In this current study, we focus on molecular profiling in advanced breast cancers beyond standard estrogen receptor (ER), progesterone receptor (PR), and HER2 testing. We report on clinical characteristics, somatic mutation frequency, and therapeutic outcomes on genotype-matched and unmatched trials for MBC patients undergoing molecular sequencing of archival tumor tissues.
Discussion
In this study, we identified one or more somatic mutations in 46% of patients with metastatic breast cancer using Sequenom hotspot or small targeted NGS panels. An important feature of our study is that we included patients receiving treatment in both academic and community settings that were referred for profiling while on standard of care therapies for metastatic disease. Albeit a non-randomized comparison, we did not observe a difference in time on treatment or best responses for profiled MBC patients treated on genotype-matched versus non-matched trials.
An earlier study by Von Hoff and colleagues used IHC, FISH, and oligonucleotide microarrays for molecular profiling of tumor tissue from 66 patients (only 18 of whom had MBC) with refractory metastatic disease to match patients to specific therapies and found a longer PFS on matched therapy compared to the patient’s previous regimen [
14]. However, the majority of MBC patients on this study received chemotherapy, hormonal or HER2-targeted therapies, not targeted treatments based upon identification of somatic mutations. Tsimberidou and colleagues used standard PCR-based sequencing to detect specific somatic genomic alterations in patients with refractory advanced malignancies and patients were enrolled in clinical trials according to genotype results [
15]. Patients matched to targeted therapy had improved response rates, longer time to treatment failure (TTF), and longer survival than non-matched patients. However, only 16 patients in this study had MBC. In breast cancer-specific studies, SAFIR01 study used comparative genomic hybridization (CGH) array and Sanger sequencing to identify molecular alterations and match patients to targeted therapies [
16]. Only 9% of patients matched to targeted therapies had objective responses to treatment. In contrast to our study, where we used archival tissues for molecular testing, SAFIR required biopsy of the metastatic site if accessible. Also SAFIR limited enrollment to patients with no more than two prior lines of chemotherapy whereas our study did not have a limit on prior number of lines of chemotherapy.
More recent studies have used new technologies, such as NGS, to identify targetable molecular alterations. In the SHIVA trial, Le Tourneau and colleagues established molecular profiles for advanced solid tumors using targeted NGS, analysis of gene copy number alterations, and analysis of hormone receptor expression by IHC to match patients to targeted therapies [
17]. This study differs from ours in several important aspects. The targeted agents given to the experimental group in SHIVA were drugs approved for clinical use but outside their approved indications, while in our study we matched patients to investigational drugs available through clinical trials at our institution. SHIVA required mandatory fresh biopsies whereas we tested archival samples.
In the MOSCATO 01 Trial, patients with advanced cancers were matched to targeted therapies based on molecular alterations and 33% were shown to have improved PFS outcomes with matched therapies when compared to PFS on prior therapy. Of the entire cohort, 19% of those matched to therapies were MBC patients. Molecular alterations were identified through array comparative genomic hybridization, NGS, and RNA sequencing performed on fresh frozen tumor biopsies [
18]. MOSCATO 01 differs from our study in that they required fresh biopsies, which limited the patients enrolling in study to those willing to undergo invasive testing, those with tumors accessible for biopsy, and those willing to wait for the results of their molecular profiling for a decision on subsequent treatment. In our study patients had molecular profiling done on archival tissues while they were on systemic therapy and at the time of progression could be referred back for discussion of available targeted therapies on clinical trials based on identified genomic alterations.
In our study, the most common mutation identified among patients with genotyping results (
n = 440) was in
PIK3CA (
n = 123, 28%), which is consistent with other recent molecular profiling studies in MBC [
16,
19‐
22]. Although mutations in
PIK3CA are common in breast cancer, they have not been reliably predictive of clinical response to drugs targeting the PI3K/mTOR pathway [
4‐
7].
PIK3CA mutations identified in archival tissues in BOLERO-2 [
5], FERGI [
23], BELLE-2 [
24], and BELLE-3 [
25] studies were not predictive of differential benefit with inhibitors of PI3K or mTOR. Thus in our study, where the majority of patients treated on genotype-matched trials were treated with PI3K inhibitors, the lack of significant differences in best responses and time on treatment may be reflective of the limited efficacy of PI3K inhibitors observed in
PIK3CA-mutant metastatic breast cancers.
As an alternative to tissue biopsies, identifying key mutations in circulating tumor DNA (ctDNA) in peripheral blood samples may provide more actionable information about the molecular profile of metastatic tumors. For example, in the BELLE-2 [
24] and BELLE-3 [
25] studies with the pan-PI3K inhibitor buparlisib combined with endocrine therapy, there was greater magnitude of effect on progression-free survival in the
PIK3CA mutant subgroup identified through cell-free DNA but not in the
PIK3CA mutant subgroup identified through archival tissue samples. However, in the BOLERO-2 study which examined the benefit of everolimus, an inhibitor of the PI3K/AKT/mTOR pathway, patients benefited from the addition of everolimus to endocrine therapy regardless of the presence of a
PIK3CA mutation in either archival tissues [
5] or cell-free DNA [
26]. With novel isoform-selective PI3K inhibitors currently in testing clinical trials [
27,
28], it remains to be determined whether PIK3CA mutation status tested using archival tissue samples or cell-free DNA is a biomarker of treatment response.
It is important to note that FERGI, BELLE-2, and BELLE-3 all involved pan-PI3K inhibitors, while the majority of patients treated on genotype-matched clinical trials at our institution were treated with isoform-selective PI3K inhibitors. More selective, isoform-specific PI3K inhibitors may be more effective and less toxic than pan-PI3K inhibitors, potentially leading to improved therapeutic outcomes [
29]. However, the number of patients treated on isoform-selective versus pan-PI3K inhibitors is too small in our study to be able to make any conclusions regarding differential therapeutic outcomes.
TP53 was the second most common mutation identified in our breast cohort. The prevalence of
TP53 mutations is lower in our cohort compared with other reports because
TP53 mutation hotspots were not included in the Sequenom assay and the TSACP panel does not include full sequencing of the
TP53 gene.
TP53 mutations are associated with aggressive breast cancers and are identified in > 80% of basal-like breast cancer. Basho et al. analyzed archival tissues from 500 MBC patients including all subtypes using hotspot mutation testing and found that
TP53 mutations were associated with worse clinical outcomes [
22].
TP53 mutation is not currently a targetable genomic alteration, although inhibition of the protein kinase WEE1 in
TP53-mutated cancers may be a potential future therapeutic approach [
30,
31]. Additionally more rare genomic alterations such as
AKT1 and
ERBB2 mutations, each occurring in 2% of MBC patients in our cohort, are potential targets for AKT and ERBB2 inhibitors [
32].
There are several important limitations to note in our study. First, we used archival tumor tissue which in some patients was many years removed from the date of study enrollment. Second, although our institution has a broad portfolio of early phase clinical trials, not all classes of targeted drugs were available during the course of the study for patients with actionable mutations. The majority of genotype-matched clinical trials available at our institution during the study period involved a PI3K inhibitor and most trials did not involve alpha isoform-selective/specific PI3K inhibitor that may have greater activity in
PIK3CA mutant breast cancers [
33,
34]. Access to genotype-matched therapies for patients with actionable mutations was limited, and the clinical outcomes reported in our study may reflect the availability of effective targeted therapies and/or the usefulness of a sequencing-based treatment strategy. Available data do not allow for the exploration of the relative contribution of these limitations on our results. Access to a greater number of drugs targeting specific molecular alterations in basket trials, which enroll patients based on the molecular alteration of the tumor not on specific tumor types, such as the NCI-MATCH program [
35], would broaden the availability of targeted drugs for actionable mutations. Unlike some other types of solid tumors, MBC patients have access to many lines of systemic therapy outside of clinical trials and some may have preferred other treatment options when invited to participate in a genotype-matched clinical trial. Only a small number of patients were enrolled in clinical trials following receipt of molecular profiling results and there was significant heterogeneity in the type of matched drug treatments received. There were also limitations in our sequencing approach. We used small hotspot genotyping or targeted sequencing panels for frequent point mutations and insertions/deletions only. While these capture the majority of somatic alterations in breast cancer, we did not assess gene amplifications or fusions which are potential genomic driver or tumor suppressor alterations that are relevant for genotype-matched treatment selection, such
FGFR1 amplification or fusions,
BRCA1/2 mutations,
NTRK fusions,
PIK3CA fusions and
ERBB2 fusions [
36].
In conclusion, our data provide “real-world” clinical outcomes for patients with MBC who have undergone molecular profiling with hotspot genotyping or small targeted NGS testing. We found that only a small percentage of patients with MBC profiled using hotspot genotyping or small targeted NGS testing subsequently enrolled in genotype-matched clinical trials. In this non-randomized comparison, we did not observe a difference in time of treatment for patients subsequently enrolled in genotype-matched versus non-genotyped-matched clinical trials. Further studies are required to establish the clinical utility of routine multi-gene mutation testing for patients with MBC.