Impact of clinical targeted sequencing on endocrine responsiveness in estrogen receptor-positive, HER2-negative metastatic breast cancer

Clinical targeted sequencing allows for the selection of patients expected to have a better treatment response, and reveals mechanisms of resistance to molecular targeted therapies based on actionable gene mutations. We underwent comprehensive genomic testing with either our original in-house CLHURC system or with OncoPrime. Samples from 24 patients with estrogen receptor-positive, human epidermal growth factor receptor 2-negative metastatic breast cancer underwent targeted sequencing between 2016 and 2018. Germline and somatic gene alterations and patients’ prognosis were retrospectively analyzed according to the response to endocrine therapy. All of the patients had one or more germline and/or somatic gene alterations. Four patients with primary or secondary endocrine-resistant breast cancer harbored germline pathogenic variants of BRCA1, BRCA2, or PTEN. Among somatic gene alterations, TP53, PIK3CA, AKT1, ESR1, and MYC were the most frequently mutated genes. TP53 gene mutation was more frequently observed in patients with primary endocrine resistance compared to those with secondary endocrine resistance or endocrine-responsive breast cancer. Recurrent breast cancer patients carrying TP53-mutant tumors had significantly worse overall survival compared to those with TP53-wild type tumors. Our 160-gene cancer panel will be useful to identify clinically actionable gene alterations in breast cancer in clinical practice.

Germline and somatic gene alterations. Detailed information regarding germline and somatic gene alterations according to endocrine responsiveness is shown in Table 3. All of the patients had one or more germline and/or somatic gene alterations. Germline pathogenic variants were identified in four (17%) patients: BRCA1 in one patient (No. 13), BRCA2 in two patients (No. 4 and No. 15), and PTEN in one patient (No. 11). All four patients harboring germline pathogenic variants had family history of breast and/or ovarian cancers. Somatic gene alterations were not detected in one patient (No. 11) who instead has a germline PTEN pathogenic variant. Germline pathogenic variants were not detected in patients with endocrine-responsive breast cancer.
Among somatic gene alterations, missense mutation was the most frequently detected mutation type. Mutation types found within the TP53 gene were missense in four tumors (Y220C in No. 2, A161D in No. 3, and E11Q in No. 6 and No. 23), nonsense in two tumors (E339* in No. 7, and Q192* in No. 9), and frameshift in one tumor (T253Nfs*11 in No. 1). Mutation types found within the PIK3CA gene were missense in six tumors (H1047L in No. 5, H1047R in No. 7, 16, and19, C420R in No. 9, E453K in No. 18) and in-frame deletion in one tumor (V105_R108del in No. 1). All six tumors with AKT1 gene alterations had the E17K missense mutation. Mutation types found within the ESR1 gene were missense in four tumors (D538G in No. 7 and No. 8,G442R in No. 14,and E380Q in No. 21), in-frame deletion in one tumor (G415_C417del in No. 9), and amplification in one tumor (No. 3).
Copy number variations such as copy number gain and copy number loss were identified as somatic mutations in nine tumors. Copy number gain was observed in seven (29%) tumors. All MYC gene alterations in three tumors corresponded to copy number gain (No. 3, 9, and 15). Copy number loss was identified in three (13%) tumors (No. 10, 13, and 15). A patient (No. 10) whose tumor had copy number loss of the CRLE2 and SMO genes did not carry other somatic gene mutations nor germline pathogenic variants.
TP53 gene mutation either in a primary or metastatic site was more frequently observed in patients with primary endocrine resistance compared to those with secondary endocrine resistance or endocrine-responsive breast cancer (P = 0.014, Table 4). In contrast, the frequencies of alterations in the PIK3CA, AKT1, ESR1, and MYC genes were not correlated with a particular endocrine-responsive status (Table 4). In addition, the frequencies of alterations in the TP53, PIK3CA, AKT1, ESR1, and MYC genes were similar in samples taken from primary or metastatic sites (  Fig. 1a). The presence of alterations in the PIK3CA, AKT1, ESR1, and MYC genes was not correlated with overall survival (Fig. 1b-e). Recurrent breast cancer patients with endocrine-responsive disease had significantly longer overall survival compared to those with primary or secondary endocrine-resistant breast cancers (P = 0.008, Fig. 1f). Univariate analysis using clinical factors (age, family history, smoking history, initial stage of diagnosis, and endocrine responsiveness) and gene mutation status (TP53, PIK3CA, AKT1, ESR1, and MYC) demonstrated that endocrine responsiveness was the only significant factor affecting overall survival in recurrent breast cancer patients (P = 0.0059).

Discussion
Since 2019 in Japan, NGS-based panel testing has been covered by insurance for metastatic cancer patients who have not had (or do not respond to) standard treatments. In the present study, we describe our experience of clinical targeted sequencing in ER-positive metastatic breast cancer. The results demonstrated that all of the 24 patients had one or more germline and/or somatic gene alterations. Four (17%) patients with primary or secondary endocrine-resistant breast cancer harbored germline pathogenic variants of BRCA1, BRCA2, or PTEN. TP53 gene mutations either in the primary or metastatic sites were associated with primary endocrine resistance and poor overall survival. Large-scale genomic sequencing studies have revealed that ER-positive breast cancer is heterogeneous regarding gene expression, mutation, and copy number variations 4,6 . Somatic gene mutations of PIK3CA, GATA3, and MAP3K1 in luminal A subtype, and those of TP53, MDM2, PIK3CA, PTEN, and cyclin D1 in luminal B subtype were identified in early breast cancer 4 . On the other hand, gene alterations of TP53, ESR1, GATA3, KMT2C, NCOR1, AKT1, NF1, RIC8A, and RB1 were more frequently observed in metastatic breast cancer compared with those in early breast cancer in hormone receptor-positive, HER2-negative disease 7 . This latter report also showed that TP53, RB1, and NF1 gene alterations correlated with poor prognosis in hormone receptor-positive, HER2-negative metastatic disease. Furthermore, a study on endocrine-resistant advanced breast cancers revealed that mutations in the mitogen-activated protein kinase (MAPK) pathway, transcriptional regulators of ER (MYC, CTCF, FOXA1, and TBX3), and ESR1 mutations were present in post-endocrine therapy tumors 21 . Identification of the mechanisms of resistance to endocrine therapy in each patient will help inform the use of specific molecular-targeted therapies in order to optimize the therapeutic regimen.
The TP53 gene is the most frequently mutated gene in all cancers, and is also the most frequent mutation found in breast cancer 4 . A TP53 gene mutation was present in 12% of luminal A and 32% of luminal B subtypes, although the frequency of TP53 gene alterations in luminal subtypes is lower than that in basal-like (84%) or HER2-positive (75%) subtypes 4 . Approximately 90% of TP53 gene alterations encode missense mutant proteins that span ~ 190 different codons located in the DNA-binding domain 22 . Our recent analysis of TP53 gene Table 2. Characteristics, treatment, and survivals in 24 patients according to endocrine responsiveness. ER estrogen receptor (% of positive cells), DFS disease-free survival (months), OS overall survival (months), AI aromatase inhibitor, TAM tamoxifen, OFS ovarian function suppression, ILC invasive lobular carcinoma, N/A not applicable, − unknown. a Age at the time of initial diagnosis (years). b Age at the time of clinical sequencing (years). www.nature.com/scientificreports/  www.nature.com/scientificreports/ www.nature.com/scientificreports/ alterations in the primary tumors of 56 ER-positive breast cancers showed that all of the 22 mutations existed at different sites 23 . Ellis and colleagues examined somatic gene alterations in pretreatment tumors from patients who were treated with neoadjuvant aromatase inhibitor therapy 24 . They reported that mutations in the TP53 signaling pathway including TP53, ATR , APAF1, or THBS1 were present in 38% of the aromatase inhibitor-resistant tumors (11 of 29 tumors). We previously reported that p53 protein accumulation in the nucleus, a surrogate marker for TP53 missense mutation, was a prognostic marker for disease-free survival in patients treated with aromatase inhibitors as adjuvant endocrine therapy 25 . Patients treated with adjuvant endocrine therapy who had TP53 wild-type tumors also showed longer 5-year overall survival 26 . Furthermore, TP53 mutations in the primary or metastatic tumors were correlated with poor prognosis in ER-positive metastatic breast cancer 27,28 .
Similarly, we showed in the present study that TP53 mutations either in the primary or metastatic sites affected primary endocrine resistance and poor outcome in ER-positive metastatic breast cancer. The activation of PI3K/AKT/mTOR signaling pathway is involved in endocrine resistance in ER-positive breast cancer 12,13 . Mutations of the PIK3CA gene are frequently observed in luminal A (49%) and luminal B (32%) subtypes 4,29 . Approximately 95% of mutations exist in the helical domain (exon 9, commonly E542 and E545) and the kinase domain (exon 20, commonly H1047) 30 .We and others previously reported that PIK3CA mutations were a positive prognostic marker in ER-positive, HER2-negative early breast cancer [31][32][33] . In contrast, activation of the PI3K pathway is suggested to cause endocrine resistance in ER-positive advanced breast cancer, and therefore inhibitors for PI3K, AKT, and mTOR have potential therapeutic utility in patients with this tumor subtype 12,13 . Furthermore, Mosele and colleagues recently reported that patients with PIK3CA-mutated metastatic tumors presented with a poor prognosis and resistance to chemotherapy in hormone receptor-positive, HER2negative metastatic breast cancer 34 . However, the present study indicated that PIK3CA mutations were present in seven tumors (29%) either in the primary or metastatic sites, and that PIK3CA mutations were not associated with endocrine responsiveness or overall survival.
On the other hand, somatic RB1 gene alterations were identified in two tumors in the present study. Loss of RB protein is one of the key mechanisms of CDK4/6 inhibitor resistance 35 . Because a CDK4/6 inhibitor combined with endocrine therapy is recommended as first-or second-line treatments in ER-positive, HER2-negative metastatic breast cancer 36 , information regarding the presence of RB1 gene alterations was essential for selecting an appropriate treatment strategy.
A survey of real-world outcomes of studies that employed targeted sequencing of small panels (48 to 50 genes) in metastatic breast cancer patients revealed that few patients were admissible to clinical trials with moleculartargeted agents 37 . In contrast, clinical targeted sequencing also led to germline secondary findings. The rate of actionable secondary findings was shown to be from 2 to 5% [37][38][39][40][41] . In the present study, four (17%) out of 24 patients harbored germline pathogenic variants including BRCA1, BRCA2, or PTEN. All four patients with germline pathogenic variants had family members with breast and/or ovarian cancers, whereas germline pathogenic variants were not detected in patients with endocrine-responsive disease. Patients in whom standard treatments are not effective, or whose family members had breast cancer, are likely to undergo targeted sequencing in clinical practice. Furthermore, one patient in the present study with no germline BRCA1 pathogenic variants had a primary tumor with a somatic BRCA1 gene mutation. De novo somatic BRCA1/2 mutations in breast cancer are considered to be rare: the prevalence of somatic BRCA1 gene mutations in primary tumors is only 1.55%, and that of somatic BRCA2 gene mutations is 1.68% 42 . Although the clinical success of PARP inhibitors for women with germline BRCA1/2 pathogenic variants is clear, the utility of using PARP inhibition to treat breast cancers with somatic BRCA1/2 mutations is not known.
The present study has several limitations. First, the size of our cohort was too small for a valid statistical analysis and to provide reliable real-world evidence. Furthermore, the site of samples used for clinical sequencing was either the primary or metastatic tumor, because it is often difficult to obtain samples from metastatic sites in clinical practice. However, we do show details of the characteristics, treatments and gene alterations of each patient (Tables 2 and 3). Because 24 samples were all the cases that we experienced during the study period, it is difficult to set an independent dataset to replicate the original findings. It is necessary to construct a prediction model for endocrine therapy based on the results of gene mutations 43 . Using the deep learning method will help to predict endocrine resistance and the efficacy of targeted therapies for precision medicine in ER-positive breast cancer 44 .
Second, our clinical targeted sequencing with 160 cancer-related genes is not breast cancer-specific. Thus, some genes essential for breast cancer progression, especially germline variants, might not be covered in this system. Third, the estimation of alterations of each gene has been improving over time. Consequently, we reanalyzed the sequencing results with reference to the newest versions (the Ensembl Variant Effect Predictor (version 95) for somatic variants and ClinVar for germline variants).
In this study, we report our clinical experience of targeted sequencing in ER-positive, HER2-negative metastatic breast cancer. Germline and somatic gene alterations as well as patients' prognosis were retrospectively analyzed, and we determined whether mutation patterns were correlated with a particular response to endocrine therapy. Our study demonstrated that 17% of patients with primary or secondary endocrine-resistant breast cancer harbored germline pathogenic variants of BRCA1, BRCA2, or PTEN. TP53 gene mutations were affected by primary endocrine resistance and poor overall survival. Our 160-gene cancer panel will be useful to identify clinically actionable gene alterations in patients with endocrine-responsive and endocrine-resistant breast cancer in clinical practice. Accumulation of real-world data on clinical targeted sequencing from individual patients will help to select appropriate therapies in endocrine-resistant breast cancer.  51 with alternate allele count, variant allele frequency and read depth and normal pool dataset created before this study, followed by manual review with Integrative Genomics Viewer (Version2.3.57) 52 and BAF-LogR profiles. We have used the Ensembl Variant Effect Predictor (version 95) 53 for genomic and functional annotation of the variants. Germline variants were classified according to the American College of Medical Genetics and Genomics (ACMG) guidelines 38 , and retained as pathogenic with conflicting interpretation in ClinVar (https:// www. ncbi. nlm. nih. gov/ clinv ar/) 54 . The results of germline secondary findings were disclosed to patients only who agreed to disclose germline information.
Endocrine responsiveness in metastatic breast cancer. We defined endocrine responsiveness according to ABC 4 14 as follows: Primary endocrine resistance was defined as relapse while on the first 2 years of adjuvant endocrine therapy, or progressive disease (PD) within first 6 months of first-line endocrine therapy for advanced breast cancer, while on endocrine therapy. Secondary endocrine resistance was defined as relapse while on adjuvant endocrine therapy but after the first 2 years, or relapse within 12 months of completing adjuvant endocrine therapy, or PD within 24 months of first-line endocrine therapy for advanced breast cancer. Endocrine-responsive breast cancer was defined as relapse after 12 months of completing adjuvant endocrine therapy, or PD ≥ 24 months of first-line endocrine therapy for advanced breast cancer. Statistical analysis. The significance of differences among three groups was evaluated using the chisquared test. P values < 0.05 were considered significant. Estimation of overall survival (time from the diagnosis of primary breast cancer to death) was performed using the Kaplan-Meier method, and differences between survival curves were assessed using the log-rank test. The Cox proportional hazards model was used for prognostic analysis. Statistical analysis was performed using Excel software (Microsoft corp., Albuquerque, MX, USA). www.nature.com/scientificreports/