Mutations in BRCA1/2 and Other Panel Genes in Patients With Metastatic Breast Cancer —Association With Patient and Disease Characteristics and Effect on Prognosis

Abstract

PURPOSE

Among patients with metastatic breast cancer (mBC), the frequency of germline mutations in cancer susceptibility genes and the clinical relevance of these mutations are unclear. In this study, a prospective cohort of patients with mBC was used to determine mutation rates for breast cancer (BC) predisposition genes, to evaluate the clinical characteristics of patients with mutations, and to assess the influence of mutations on patient outcome.

PATIENTS AND METHODS

Germline DNA from 2,595 patients with mBC enrolled in the prospective PRAEGNANT registry was evaluated for mutations in cancer predisposition genes. The frequencies of mutations in known BC predisposition genes were compared with results from a prospective registry of patients with nonmetastatic BC sequenced using the same QIAseq method and with public reference controls. Associations between mutation status and tumor characteristics, progression-free survival, and overall survival were assessed.

RESULTS

Germline mutations in 12 established BC predisposition genes (including BRCA1 and BRCA2) were detected in 271 (10.4%) patients. A mutation in BRCA1 or BRCA2 was seen in 129 patients (5.0%). BRCA1 mutation carriers had a higher proportion of brain metastasis (27.1%) compared with nonmutation carriers (12.8%). Mutations were significantly enriched in PRAEGNANT patients with mBC compared with patients with nonmetastatic BC (10.4% v 6.6%, P < .01). Mutations did not significantly modify progression-free survival or overall survival for patients with mBC.

CONCLUSION

Multigene panel testing may be considered in all patients with mBC because of the high frequency of germline mutations in BRCA1/2 and other BC predisposition genes. Although the prognosis of mutation carriers and nonmutation carriers with mBC was similar, differences observed in tumor characteristics have implications for treatment and for future studies of targeted therapies.

INTRODUCTION

Novel therapies for breast cancer (BC) have been developed mainly for molecular subtypes on the basis of gene expression. Germline genetic markers are under consideration for specific targeted therapies in BC.¹ However, the only actionable germline mutations are in BRCA1 or BRCA2 (BRCA1/2 hereafter). In patients with human epidermal growth factor receptor 2 (HER2)–negative (HER2−) metastatic BC (mBC) and a BRCA1/2 mutation, treatment with poly (ADP-ribose) polymerase inhibitors (PARPi) can lead to clinically meaningful improvement in progression-free survival (PFS).^2,3 BRCA1/2 mutations have also been noted to be predictive of chemotherapy efficacy.^4-7

CONTEXT

• Key Objective

• Germline mutations in BRCA1 and BRCA2 (BRCA1/2) and PALB2 are associated with high response rates to chemotherapy and PARP inhibitors. However, the frequency of mutations in BRCA1/2 and other cancer risk genes in patients with metastatic breast cancer (mBC) are not well-defined. The objective of this study was to define the frequencies of mutations in these genes among patients with mBC and to assess tumor and patient characteristics associated with the presence of mutations.

• Knowledge Generated

• In 2,595 patients with genotyped mBC from a German cancer registry study, the mutation frequency was 5% for BRCA1/2 and approximately 10% for a panel of established BC risk genes. Patients with mutations in homologous recombination deficiency genes had a higher frequency of luminal B–like tumors, whereas patients with triple-negative BC with BRCA1 mutations had a high frequency of brain metastases.

• Relevance

• This work provides accurate germline mutation frequencies for BC risk genes in a population of patients with mBC. Differences in patient and disease characteristics according to mutation status may inform genetic testing and future clinical trials.

Although little is known about the role of genetic variants in other moderate- or high-penetrance susceptibility genes, germline common genetic variants with small effects of cancer risk have been described as prognostic and predictive factors in patients with BC.^8-15 However, higher penetrance–inherited mutations may have greater clinical relevance as molecular drivers because of stronger biological effects. Recently, BC risks associated with germline mutations in 21 candidate genes, including BRCA1/2, from multigene genetic testing panels, were shown in a large case-control study.¹⁶ As several of these commonly tested genes have similar biological functions, development of targeted therapies might have a foreseeable benefit for patients with BC with mutations. This is especially important for patients with mBC with limited treatment options.

The aim of this study was to evaluate the prevalence of mutations in cancer susceptibility genes among patients with mBC, identify clinical features associated with these mutations, and assess the effect of mutations on prognosis.

PATIENTS AND METHODS

Patients

PRAEGNANT (ClinicalTrials.gov identifier: NCT02338167; Data Supplement [online only])¹⁷ is an ongoing, prospective, multicentric BC registry in hospitals and practices across Germany. Patients are eligible at any time point of their disease. Data are documented in electronic case report forms¹⁷ and monitored using plausibility checks and on-site monitoring. Patient and disease data are collected at study entry and at follow-up visits (Data Supplement). All patients provided informed consent at enrollment. The study was approved by the respective German ethics committees and by the Mayo Clinic institutional review board. A total of 2,647 patients with mBC were registered between July 2014 and March 2018. Of these, 2,595 patients yielded high-quality germline sequencing data and were included in the correlative analyses. Among these, 1,516 had complete information on molecular subtype, line of therapy at enrollment, and follow-up duration and were included in survival analyses (Fig 1).

FIG 1.

Patient flowchart. ^aPatient population for correlative analysis, ^bpatient population for survival analysis, and ^cprospective means patients had to be included not more than 90 days after the start of the respective therapy line. HER2, human epidermal growth factor receptor 2; HR, hormone receptor; TNBC, triple-negative breast cancer.

Ascertainment of Clinical and Histopathologic Data

Definitions of hormone receptor (HR) status, HER2 status, and grade have been described previously.¹⁸ Briefly, estrogen receptor status, progesterone receptor status, HER2 status, and tumor grade were requested from all available biopsies (primary BC and mBC). If biomarker assessment of a metastatic site was available, these assessments were used. Otherwise, the latest biomarker results from the primary tumor were used. All patients treated with endocrine therapy up to the study inclusion were assumed to be HR-positive (HR+), and all patients treated with an anti-HER2 therapy were assumed to be HER2+. There was no central review of biomarkers. Estrogen receptor and progesterone receptor were considered positive if ≥ 1% of tumor cells were stained. A positive HER2 status required an IHC score of 3+ or a positive fluorescence in situ hybridization or chromogenic in situ hybridization. Tumors were classified into mutually exclusive groups as luminal A–like (HR+ and grade 1/2), luminal B–like (HR+ and grade 3), triple-negative breast cancer (TNBC), or HER2+. The site of metastasis was classified into four discrete categories in the following hierarchical order—brain (additional locations allowed), visceral (additional locations except brain allowed), bone only, and other—on the basis of the presence or absence of tumors at these locations. The site of metastasis, body mass index, and family history were used from the documentation at study entry.

Germline Genetic Testing, Bioinformatics Analysis, and Classification of Variants

Germline genetic testing and bioinformatics analysis for 37 cancer predisposition genes were performed using a custom amplicon-based QIAseq panel (QIAGEN, Hilden, Germany) as previously described¹⁹ (Data Supplement). Annotation of mutations was conducted using the American College of Medical Genetics and the Association for Molecular Pathology guidelines.²⁰ Low-penetrance missense variants in CHEK2 were excluded from analyses. For correlative analyses, genes were grouped into nonoverlapping functional categories: BRCA1/2, homologous recombination defect repair ([HRD] ATM, BARD1, FANCC, PALB2, RAD51C, RAD51D, SLX4, and XRCC2), other DNA repair (BLM, BRIP1, CHEK2, ERCC2, ERCC3, FANCM, MLH1, MRE11A, MSH2, MSH6, MUTYH, NBN, PMS2, RAD50, RECQL, and RINT1), and other (APC, CDH1, CDKN2A, EPCAM, KRAS, MEN1, NF1, PPM1D, PRSS1, PTEN, and TP53). For the comparison with healthy controls, patients were classified according to 12 established BC predisposition genes (ATM, BARD1, BRCA1, BRCA2, CDH1, CHEK2, NF1, PALB2, PTEN, RAD51C, RAD51D, and TP53). All gene categorizations are shown in the Data Supplement.

Comparison With Mayo Clinic Breast Cancer Study

To determine whether mutations in specific genes were enriched in patients with mBC compared with a non-mBC population, gene-specific mutation frequencies in known BC predisposition genes from the present study were compared with frequencies among primary (non-mBC) invasive BC cases in Mayo Clinic Breast Cancer Study (MCBCS)²¹ that were identified using the same QIAseq panel. The MCBCS is a clinic-based prospective registry of consecutive patients with BC, unselected for age at diagnosis or family history, seen at the Mayo Clinic between 2000 and 2016 (Data Supplement).

Statistical Considerations

Patient and tumor characteristics of patients in PRAEGNANT and MCBCS are shown as counts and frequencies.

For each patient, PFS was only considered for the earliest line of therapy during which the patient entered the study. PFS was defined as the time from the date of initiation of therapy to the earliest date of disease progression or death from any cause. Observation time was left-truncated for the time at which the patient entered the study if study entry was later than the start of treatment. Overall survival (OS) was defined similarly. For PFS and OS ascertainment, see the Data Supplement. A multivariable Cox regression model was fitted with PFS as outcome and the following predictors: age at study entry, HR status (positive or negative), HER2 status (positive or negative), tumor grade, selected therapy line, ECOG status, metastasis pattern, and number of concomitant diseases and mutation status (no mutation, gBRCA1/2 mutation, HRD mutation, DNA repair gene mutation, and other).

Similar analyses were performed for OS. Unadjusted survival rates were estimated using the Kaplan-Meier product-limit method. Two sensitivity analyses were done. First, survival analyses were repeated for patients included in PRAEGNANT at the time of first therapy line and second, survival analyses were repeated excluding patients with left-truncated survival information. Missing predictor values were imputed.²² The proportional hazards assumptions were checked using the method of Grambsch and Therneau.²³ All the tests were two-sided, and a P value < .05 was regarded as statistically significant. Statistical analyses were performed using the R software (v3.4.0).

RESULTS

Patient and Tumor Characteristics According to Mutation Status

Patient characteristics according to the mutually exclusive mutation groups are shown in Table 1. In the total cohort, the mean age of BC metastasis diagnosis was 57.8 years, and 41.4% of patients were enrolled while receiving the first line of therapy. Of those with available tumor pathology information, 11.1% (n = 254) were TNBC, 27.6% (n = 632) HER2+, and 61.3% (n = 1,401) HER2−/HR+. Visceral metastases were seen in 50.9% of patients, 17.8% had bone-only metastases, and 14.3% had brain metastases.

TABLE 1.

Patient and Tumor Characteristics Based on Mutation Status

Frequencies of Germline Mutations

Among 2,595 patients, 425 mutations in any of the 37 analyzed genes were found in 396 (15.3%) patients (Table 2). BRCA2 (n = 76; 2.9%) was the most frequently mutated gene, followed by CHEK2 (n = 57; 2.2%), BRCA1 (n = 53; 2.0%), PALB2 (n = 28; 1.1%), and ATM (n = 23; 0.9%). Combining the 12 established BC risk genes (including BRCA1/2), 271 patients (10.4%) had a mutation in at least one gene. A total of 25 patients had two mutations, and two patients had three separate mutations in these genes.

TABLE 2.

Frequency of Germline Mutations Among 2,595 Patients

When considering groups of genes, 129 (5.0%) patients had a mutation in BRCA1/2, 76 (2.9%) had a mutation in one of the other HRD genes, 150 (5.8%) in other DNA repair genes, and 41 (1.6%) in other cancer risk genes. Only patients with BRCA1/2 mutations had an earlier BC diagnosis (45.0 ± 11.2 years) than patients without mutations. BRCA1/2 carriers more frequently had a TNBC (29.1%), whereas patients with other HRD mutations more frequently had luminal B–like tumors (29.9%). Both mutation groups had a high proportion of grade 3 tumors and a more frequent family history of BC than in other patient groups. Interestingly, there were more brain metastases in patients with BRCA1/2, HRD gene, and other DNA repair gene mutations than in patients without mutations (Table 1).

Exploring these associations among the five most frequently mutated genes (Table 3), the location of the metastasis was mainly attributable to patients with BRCA1 mutations, with 27.1% of BRCA1 mutation carriers having brain metastases compared with 12.8% in nonmutation carriers. Although sample sizes were small, further analysis of metastatic sites (Data Supplement) indicated that the association between BRCA1 and brain metastases was mainly seen in patients with TNBC, with 10 of 24 BRCA1 mutation carriers (41.2%) having brain metastases.

TABLE 3.

Breast Cancer Characteristics Among Carriers of Five Most Frequently Mutated Genes

Mutation frequencies according to tumor and patient characteristics for each gene category and gene are provided in the Data Supplement.

Comparison of Mutation Frequencies Between Patients With mBC and Nonmetastatic BC

A higher proportion of patients in PRAEGNANT were found to carry germline mutations in known BC predisposition genes compared with patients with nonmetastatic invasive BC in MCBCS (10.4% v 6.6%, P < .01). Results of this exploratory analysis for each gene in the overall population and subgroups according to HR status are provided in the Data Supplement. Germline mutations in BRCA2, PALB2, and TP53 were enriched in the PRAEGNANT cohort.

Clinical Outcome of Patients With mBC With Predisposition Gene Mutations

Patient characteristics of the population for the survival analysis are shown in the Data Supplement. Patients with a BRCA1/2 mutation did not have significantly different PFS compared with patients with no mutations in any genes (adjusted hazard ratio [HR], 1.00; 95% CI, 0.77 to 1.65) (Table 4 and Fig 2). The analysis of OS showed similar results (adjusted HR, 0.90; 95% CI, 0.61 to 1.35). In addition, comparing patients with a mutation in any of the other categories of genes did not show any difference in PFS or OS. The sensitivity analysis of patients recruited at the first line of therapy yielded similar results (Data Supplement). In addition, the sensitivity analysis of patients with time of enrollment being prior to the initiation of the therapy found consistent estimates for the HRs (data not shown).

TABLE 4.

Associations Between Predisposition Gene Mutation Status and PFS and OS

FIG 2.

Kaplan-Meier curves for (A) PFS and (B) OS in the overall cohort according to mutation status. Estimates for the adjusted Cox regression model are provided in Table 4. HRD, homologous recombination defect repair; OS, overall survival; PFS, progression-free survival.

The genes included in the QIASeq panel along with the list of all mutations identified are provided in the Data Supplement.

DISCUSSION

In this large unselected cohort of 2,595 patients with mBC, a germline BRCA1/2 mutation was found in 5.0% and a mutation in any of the 12 BC predisposition genes was seen in 10.4% of the population. Categorizing patients by a mutation in BRCA1/2, in other HRD genes, in other DNA repair genes, or in other cancer risk genes did not reveal a group with a different prognosis compared with patients without mutations.

Concerning the prevalence of germline mutations in BC susceptibility genes, except from the subgroup of women with a grade 1 tumor and the subgroup of patients 60 years or older, there was no subgroup with a mutation frequency in BRCA1/2 under 3% and in all genes with a mutation frequency under 10%. These findings suggest that germline genetic testing may be appropriate for all patients with mBC irrespective of age or family history. Although the National Comprehensive Cancer Network (NCCN) criteria for BRCA1/2 testing acknowledge that germline genetic testing for BRCA1/2 should be considered for all patients with mBC to assess treatment eligibility for PARPi, the high frequency of mutations in other genes suggests that multigene panel testing, rather than targeted BRCA1/2 testing, should be considered for patients with mBC to assess eligibility for clinical trials with targeted therapies on the basis of germline mutation status and to inform cascade testing in family members. As most clinical genetic testing is currently performed using multigene panels, testing of patients with mBC for a gene panel such as the 12 predisposition genes used in this analysis—rather than just BRCA1/2 or a much larger panel—is feasible.

The heterogeneity in tumor characteristics is not unexpected as germline mutations in several genes are known to be the drivers of tumor biology and therapy responsiveness.^5,6,24,25 With this, there may be opportunities for expanding targeted therapy use beyond BRCA1/2 mutation carriers.

Furthermore, the heterogeneity in metastatic sites among mutation carriers may also have to be taken into account since the location of metastasis is known to affect prognosis and efficacy of treatments.^24,26 The higher frequency of brain metastasis in BRCA1 carriers as noted in this study is also indicative of the biological differences in tumors associated with these mutations. Prior studies have demonstrated that BRCA1-associated breast tumors frequently exhibit HRD, but ATM tumors do not.²⁷ These differences need to be investigated further to identify whether carriers of mutations in different genes can be studied together in clinical trials.

One clinical study has tested the efficacy of olaparib monotherapy in patients with a mutation in HRD genes, other than BRCA1/2.²⁸ Among 11 germline PALB2 mutation carriers, nine had partial responses.²⁸ There were no responses in patients with CHEK2 mutations (n = 8), with an ATM, BARD1, or RAD50 mutation (each n = 1), or with a combined ATM/CHEK2 mutation (n = 1).²⁸ Because of the small sample size, this study does not exclude a response to a PARPi or other therapies among patients with mutations in these genes, but it gives a clear signal for PALB2, which had a mutation frequency of 1.1% in our study.

Despite being an unselected cohort, the frequencies of germline mutations in this study are similar to mutation rates from high-risk cohorts enriched for younger age at diagnosis of BC and family history of cancer.^16,29,30 In addition, a comparison of the PRAEGNANT cohort with the MCBCS showed that germline mutations in known BC predisposition genes are enriched in patients with metastatic compared with nonmetastatic BC. Some patient subgroups such as patients > 60 years had low BRCA1/2 mutation frequencies (1.7%-1.8%), consistent with other studies.³¹

The higher frequency of germline mutations in patients with mBC compared with patients with primary BC also suggests that breast tumors in germline mutation carriers may be more likely to progress to mBC than tumors in noncarriers, as a consequence of either tumor biology or treatment efficacy. Enrichment of germline mutation carriers has been reported in metastatic and more biologically aggressive forms of prostate cancer.^32,33 For BRCA1/2 mutation carriers with early BC, there are reports of worse prognosis,^34,35 unaltered prognosis,^5,36-40 and better prognosis.^7,36 However, most of these studies were performed in younger patients with TNBC, with an overrepresentation of patients treated with chemotherapy. Therefore, studies on the effect of BRCA1/2 mutations on prognosis in unselected patient populations may show different prognostic patterns for mutation carriers as indicated by our study. Similarly, there are some data indicating a worse prognosis for patients with early BC with a PALB2 mutation,⁴¹ which fits the higher frequency of PALB2 mutations in our mBC population. Other genes such as NF1, RAD51C, and RAD51D may play a role in the pathogenesis¹⁶ but not the progression of BC. However, our study is too small to draw any conclusions about these genes. The higher prevalence of TP53 among patients with mBC in this study may be due in part to clonal hematopoiesis.^42,43

Although differences in tumor characteristics were noted among mutation carriers and noncarriers, no difference in prognosis was noted in this study for BRCA1/2 or any other combination of mutation carriers. In the TNT trial, PFS was longer in BRCA1/2 mutation carriers treated with platinum-based chemotherapy compared with those without a BRCA1/2 mutation.⁶ However, a significant difference between BRCA1/2 mutation carriers and noncarriers was not noted for patients treated only with paclitaxel. Our study did not evaluate the effects of each therapy line according to clinical tumor subtype to answer questions concerning specific therapies, survival, and BRCA1/2 mutation status. Most patients in the PRAEGNANT registry are HR+ and are predominantly treated with endocrine therapy.¹⁸ Therefore, it is possible that a prognostic effect of BRCA1/2, which is usually observed under chemotherapy, did not have an influence on the prognosis in our cohort.

There are some limitations to this study. Despite the large sample size, the numbers of mutations in some genes were low. Therefore, we were not able to generate conclusive evidence about the effect of mutations in several genes on prognosis or associations with clinical characteristics. Additionally, a mutation that was already clinically known to the treating physicians could have altered both chosen therapies and the course of the disease. Allowing late entry into the study and handling the survival data by left truncation could potentially introduce a bias into the survival analysis; however, we addressed this by performing a sensitivity analysis, which yielded similar survival estimates. In addition, the inherent differences in the patient population and ascertainment between PRAEGNANT and MCBCS cohorts must be acknowledged, which could potentially affect the comparison of mutation frequencies. For example, in PRAEGNANT, only 63 patients were not of White European descent. In addition, the young mean age at diagnosis (52.8 years) could point to a possible selection bias. However, because young patients have a worse prognosis when diagnosed with early BC,⁴⁴ an enrichment of young patients is expected in an mBC population. In addition, a French mBC registry with more than 22,000 patients reported a similar mean age at diagnosis of 54.0 years.⁴⁵

In conclusion, the higher frequency of germline mutations in mBC compared with nonmetastatic BC suggests that mutations in specific genes may promote progression from early to metastatic BC and that multigene panel testing should be considered in all patients with mBC to identify patients and family members who may benefit from targeted therapy and cascade testing, respectively. In addition, significant heterogeneity in tumor characteristics across genes without a significant effect on OS confirms the role of germline mutations in driving specific biological subtypes of BC, highlights the challenges in extending novel targeted therapies such as PARPi beyond BRCA1/2 carriers, and has significant implications on the design of future clinical trials for germline mutation carriers with mBC.

AUTHOR CONTRIBUTIONS

Conception and design: Peter A. Fasching, Siddhartha Yadav, Chunling Hu, Marius Wunderle, Michael P. Lux, Diana Lüftner, Markus Wallwiener, Erik Belleville, Michael Untch, Arndt Hartmann, Pauline Wimberger, Diethelm Wallwiener, Andreas Schneeweiss, Andreas D. Hartkopf, Fergus J. Couch

Administrative support: Peter A. Fasching, Matthias Rübner, Erik Belleville, Matthias W. Beckmann, Arndt Hartmann

Provision of study materials or patients: Matthias Rübner, Kun Y. Lee, Rohan D. Gnanaolivu, Hanna Hübner, Hans Tesch, Johannes Ettl, Michael P. Lux, Arif B. Ekici, Bernhard Volz, Michael Untch, Arndt Hartmann, Florin-Andrei Taran, Andreas Schneeweiss, Andreas D. Hartkopf, Fergus J. Couch

Collection and assembly of data: Peter A. Fasching, Siddhartha Yadav, Chunling Hu, Marius Wunderle, Matthias Rübner, Kun Y. Lee, Peyman Hadji, Hanna Hübner, Johannes Ettl, Michael P. Lux, Sabrina Uhrig, Diana Lüftner, Markus Wallwiener, Erik Belleville, Michael Untch, Hans-Christian Kolberg, Matthias W. Beckmann, André Reis, Arndt Hartmann, Pauline Wimberger, Tanja N. Fehm, Andreas D. Hartkopf, Fergus J. Couch

Data analysis and interpretation: Peter A. Fasching, Siddhartha Yadav, Chunling Hu, Marius Wunderle, Lothar Häberle, Steven N. Hart, Matthias Rübner, Eric C. Polley, Rohan D. Gnanaolivu, Peyman Hadji, Hanna Hübner, Hans Tesch, Johannes Ettl, Friedrich Overkamp, Michael P. Lux, Arif B. Ekici, Bernhard Volz, Markus Wallwiener, Volkmar Müller, Erik Belleville, Michael Untch, Hans-Christian Kolberg, Matthias W. Beckmann, André Reis, Arndt Hartmann, Wolfgang Janni, Pauline Wimberger, Florin-Andrei Taran, Sara Y. Brucker, Andreas D. Hartkopf, Fergus J. Couch

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Mutations in BRCA1/2 and Other Panel Genes in Patients With Metastatic Breast Cancer—Association With Patient and Disease Characteristics and Effect on Prognosis

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).