Universal Genetic Testing for Newly Diagnosed Invasive Breast Cancer

Key Points Question What is the prevalence of germline pathogenic variants in breast cancer susceptibility genes among women with newly diagnosed invasive breast cancer? Findings In this cross-sectional study of 729 female patients with a first diagnosis of breast cancer who participated in a universal genetic testing program, 5.3% had germline pathogenic variants in BRCA1/2 or PALB2, and 1.8% were considered eligible for poly(adenosine diphosphate–ribose) polymerase inhibitors based on their genetic testing result. Meaning Findings suggest that universal genetic testing identifies actionable germline pathogenic variants in more than 1 in 20 patients with newly diagnosed breast cancer and is associated with systemic therapy recommendations in one-third of these cases.


Introduction
The use of multigene cancer susceptibility panels has revealed that 5% to 10% of women with breast cancer tested for these genes are found to carry a germline pathogenic or likely pathogenic variant (GPV). 1,2As the cost of genetic testing has decreased, 3 the pressure to relax genetic testing criteria has increased. 4,5This pressure has only become more insistent now that effective first-line systemic therapies for BRCA1 and BRCA2 heterozygotes with early-stage or metastatic disease have become available. 6 existing cancer genetics services cannot provide counseling and genetic testing in a timely fashion in response to this demand, it has been argued that treating physicians should offer genetic testing to a broader group of women diagnosed with breast cancer, ideally at diagnosis. 7,8This process, referred to as mainstreaming, 5 generally involves limited or no pretest genetic counseling, with positive results being handled by treating physicians, and subsequent referral to genetics health care professionals who then counsel affected women and offer appropriate follow-up with cascade testing of family members. 9Universal genetic testing (ie, without the need for meeting prespecified personal or family history criteria) was first implemented for patients with nonmucinous ovarian carcinoma [10][11][12][13][14][15] and is now being extended to patients with invasive breast cancer. 16,17The results of these studies are beginning to be published 18,19 and suggest broad acceptability of this approach by patients and health care professionals.
To address feasibility and evaluate optimal selection criteria, we performed a cross-sectional, multicentered universal genetic testing study of female patients with a first primary invasive breast cancer.The primary objective of this study was to determine the prevalence of GPVs in BRCA1, BRCA2, and PALB2 (B1B2P2) as well as in other breast cancer susceptibility genes (BCSGs) within a racially and ethnically diverse cohort of women with newly diagnosed breast cancer.Mary's Hospital Cancer Centre) in Montreal, Canada, were eligible for inclusion.Patients with a personal history of breast cancer diagnosed greater than 6 months prior to study referral, in situ malignant tumor, or prior genetic testing for hereditary breast or ovarian cancer or women diagnosed at outside institutions without central pathology review were deemed ineligible (Figure).Eligible patients were referred by treating oncologists and contacted to offer an appointment for genetic counseling, including a 3-generation pedigree with self-reported racial and ethnic origins.Following pretest counseling, written informed consent was obtained from those who elected to proceed, and a blood sample was collected.Testing had 2 components; the first was an obligatory primary panel for B1B2P2, the 3 genes most likely to affect management.All patients were then offered optional testing in a secondary panel of 14 genes-ATM, BARD1, BRIP1, CDH1, CHEK2, MLH1, MSH2, MSH6, PMS2, PTEN, RAD51C, RAD51D, STK11, and TP53.

Methods
Phase 1, which ran from September 2019 to April 2021, did not have an upper age limit for eligibility.On interim review of our data at a COVID-19-related recruitment pause, we amended the protocol such that affected women older than 70 years at diagnosis, except those with triplenegative breast cancer (TNBC), were deemed ineligible.Thus, phase 2 was implemented, starting

Variant Classification
Variants were annotated according to the American College of Medical Genetics 5-tiered categorization as pathogenic, likely pathogenic, variant of uncertain significance, likely benign, or benign using Clinvar; VarSome, release 11.9; Franklin by Genoox; and the Human Gene Mutation Database (2019).Variants that were annotated as benign or likely benign were removed.Follow-up consultations with a cancer geneticist (W.D.F.) and a certified genetic counselor (Z.R. or A.A.) was offered to persons identified to have GPVs.W.D.F.reviewed all genetic testing results and decided whether to report the variant of uncertain significance or not.Each variant of uncertain significance was assessed on its own merit, with attention paid to the likely clinical utility of the variant in the context of the personal and family history of cancer.Founder variants associated with a less than 2-fold increased odds of breast cancer-that is, CHEK2 c.470C>T (p.I157T) and c.1283C>T (p.S428F)were categorized as variants of uncertain significance.

Results
Following the initial referral of 1017 female patients with breast cancer, 805 eligible patients were offered genetic counseling, and 729 (90.6% of those eligible) underwent pretest counseling followed by testing (Figure).Of 729 patients, 659 (90.4%) opted to receive the primary and secondary panel, while 70 (9.6%)elected to receive primary panel testing for B1B2P2 only.There were no significant differences between those who accepted and those who refused the secondary panel with respect to median age, race, high-risk (Ashkenazi Jewish) ancestry, or family history of breast cancer, ovarian cancer, or other cancers (all P > .05).

Cohort Characteristics
Within
The distributions of variants identified for B1B2P2, CHEK2, and ATM are shown in eFigure 1 in Supplement 1.Of 729 patients, 98 (13.4%) had a variant of uncertain significance (eTable 4 in Supplement 1).Ethnicity and the distribution of GPVs and variants of uncertain significance by parental origin are shown in eTable 5 and eFigure 2 in Supplement 1.
Of the 659 patients who elected to receive testing in all 17 genes, 35 (5.3%) had a GPV in B1B2P2 (including 1 patient with a GPV in both BRCA1 and CHEK2), while 14 (2.1%) had a GPV in 1 of ATM, For the 659 patients who underwent secondary panel testing, no specific clinical factors were significantly associated with GPVs identified in ATM, BARD1, BRIP1, CHEK2, CDH1, MSH2, MSH6, MLH1, PTEN, RAD51C, RAD51D, or TP53 (all P > .05)(eTable 6 in Supplement 1).However, the study was not specifically powered to detect associations between demographic or clinical features and GPVs in these genes.Only 34% of patients positive for B1B2P2 but 86% of patients positive for the other BCSGs would not have been eligible for genetic testing according to traditional risk-based criteria used in the regular medical genetics service (eTable 1 in Supplement 1), so it is clear that the high-risk criteria are, as expected, biased toward identifying B1B2P2 heterozygotes and were not designed to  (and in fact are less capable of) identifying persons carrying lower-risk GPVs (eTable 6 in Supplement 1). 1,21

Eligibility for Poly(Adenosine Diphosphate-Ribose) Polymerase Inhibitors
Prior to genetic testing, 101 of 729 patients (13.9%) with early-stage and metastatic breast cancer would have been candidates for poly(adenosine diphosphate-ribose) polymerase (PARP) inhibitors based on biologic subtype, clinical stage, pathologic stage, and/or response to neoadjuvant chemotherapy (eTable 7 in Supplement 1).Overall, most PARP inhibitor candidates had TNBC (64 of 729 [8.8% of total cohort]), predominantly those with residual disease after neoadjuvant chemotherapy (54 of 729 [7.4% of total cohort]), while the remaining 37 of 729 PARP inhibitor candidates (5.1% of total cohort) had ER-positive, ERBB2-negative breast cancer.
Of the 64 PARP inhibitor candidates with TNBC, 12 (18.8%)tested positive for a BRCA1/2 GPV, whereas of 37 PARP inhibitor candidates with ER-positive, ERBB2-negative breast cancer, 1 (2.7%) tested positive for a BRCA1 GPV, and none were positive for BRCA2.Thus, in the overall cohort of 729 patients, 13 (1.8% of total cohort) had a confirmed GPV in BRCA1/2 and were eligible for PARP inhibitors.Of these women, 12 (92.3%)met traditional institutional criteria for genetic testing, while 1 (7.7%), a 67-year-old woman with high-grade TNBC, fell outside traditional testing criteria.

Discussion
Over the past decade, restrictive genetic testing criteria have gradually loosened as improved technology 3 and legal rulings 22 have driven testing costs down.Moreover, the development of PARP inhibitor therapies specifically targeting tumors with homologous recombination repair deficiency 6,23 has motivated a wholesale review of the process by which women affected by breast cancer receive genetic testing. 24Thus, new models of genetic testing are emerging in breast cancer genetics, 25 and mainstreaming is becoming increasingly popular as the demand for testing cannot be met by existing genetics services. 8The question, however, of whether all women with invasive breast cancer should be offered publicly funded early genetic testing for BCSGs 4 is unresolved, and there is a paucity of data on which to make decisions in this sphere.
Here, we show in this large cross-sectional study that from a cohort of 729 women with incident breast cancer, 53 (7.3%) carried a GPV in a BCSG, including 5.3% with a GPV in B1B2P2.7][28] As expected, on multivariable analysis, clinical factors significantly associated with GPVs in the 3 major BCSGs (BRCA1, BRCA2, and PALB2) included being younger than 40 years at diagnosis, women with TNBC, and those with a family history of ovarian cancer.If we wished to identify 95% of all GPVs in these 3 genes, then testing all women younger than 50 years, all patients with TNBC, and all those with a family history of breast or ovarian cancer would achieve this goal, with 95% of patients with B1B2P2 identified and 543 tests performed.Simplifying criteria by testing all women with TNBC or breast cancer at younger than 65 years of age would result in identification of 92.3% of patients with B1B2P2 and 634 tests performed, whereas a single age cutoff of breast cancer diagnosed at or younger than 70 years of age would yield identical identification rates but with the highest number of tests (n = 673) required.
Recent consensus panel guidelines for germline testing for patients with breast cancer suggest that women who fall outside age-based criteria should be offered BRCA1/2 testing if they are candidates for PARP inhibitor therapy for early-stage or metastatic disease. 29Candidacy is based on eligibility criteria derived from clinical trials for BRCA-associated metastatic breast cancer, 30,31 as well as eligibility criteria from OlympiA for early-stage disease. 6In our study, 18.8% of PARP inhibitor candidates with TNBC and 2.7% of PARP inhibitor candidates with ER-positive, ERBB2-negative breast cancer tested positive for a GPV in BRCA1/2.Alternatively, one-third of BRCA1/2 carriers were eligible for PARP inhibitors, of whom 4 (30.8%) were older than 50 years and 1 (7.7%) was older than 65 years with TNBC.Thus, testing criteria that include all patients with TNBC as well as these

Study
Design, Setting, and Patient Selection All women aged 18 years or older with a first diagnosis of pathologically confirmed, stage I to stage IV breast cancer between September 2019 and April 2022 at 1 of 3 McGill University-affiliated institutions (McGill University Cedars Cancer Centre; Jewish General Hospital Segal Cancer Centre; St October 2021, until study completion in April 2022.The study was approved by the research ethics boards at McGill University Health Centre and integrated university health and social services centres (ie, CIUSSS West-Central hospitals).The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

Figure .
Figure.Cohort Selection and Genetic Testing Results

Table 1 .
Cohort CharacteristicsThe remaining 610 patients (92.6%) had no GPV identified on the full panel (hereafter referred to as noncarriers).Notably, testing for ATM, BRCA1, BRCA2, CHEK2, and PALB2 identified 93% of all GPVs found in the 17-gene panel.Patients with a GPV in B1B2P2 on the primary panel were younger (median age, 42 years [range, 29-79 years]) than noncarriers (median age, 53 years [range, 23-91 years]) and women with GPVs in the secondary panel (median age, 58 years [range, 36-69]; P = .03).There was no difference in race and ethnicity, ancestry, laterality, histology, or clinical nodal status between women with GPV in the primary or secondary panel and those who tested negative (Table2).Women with a GPV in B1B2P2 were more 95% CI, 2.22-20.90),and compared with patients with no family history of cancer, those with a family history of ovarian cancer had a nearly 10-fold increased odds (OR, 9.75; 95% CI, 2.65-35.85).
BARD1, BRIP1, CHEK2, RAD51D, or STK11 (Table2).Clinical Factors Associated With GPVs in BRCA1/2 and PALB2 and Secondary Panel GenesOn univariate analysis, young age, family history of ovarian cancer, histology, histologic grade, biologic subtype, tumor size, and stage were significantly associated with a GPV on the primary panel of B1B2P2 (Table3).Despite the expected higher frequency of BRCA1/2 GPVs in the Ashkenazi Jewish population, we did not see this in this study as only 3 of 54 Ashkenazi Jewish patients were positive for GPVs in these 2 genes.Younger than 40 years of age, 16 of 88 patients tested (18.2%) positive for B1B2P2 compared with 10 of 196 patients (5.1%) aged 40 to 49 years and 6 of 227 patients (2.6%) aged 50 to 59 years (P < .001).Of 112 patients with TNBC, 22 (19.6%)tested positive for B1B2P2 compared with only 14 of 487 ER-positive, ERBB2-negative patients (2.9%) and 3 of 130 ERBB2positive patients (2.3%) (P < .001).Among patients with TNBC diagnosed at younger than 40 years of age, 41.7% (10 of 24) had a GPV in B1B2P2, decreasing to 29.4% (5 of 17) for those 40 to 49 years of age, 13.8% (4 of 29) for those 50 to 59 years, 7.1% (2 of 28) for those 60 to 69 years, and 7.1% (1 of 14) for those older than 70 years.On multivariable logistic regression, the factors that remained independently associated with B1B2P2 included age, family history of ovarian cancer, histologic grade, and biologic subtype.Relative to patients aged 50 to 59 years, those diagnosed with breast cancer at younger than 40 years had a 6.8-fold increased likelihood of carrying a GPV in B1B2P2 (OR, 6.83;

Table 3 .
P/LP Variants in BRCA1/2 or PALB2 Among Women With Newly Diagnosed Invasive Breast Cancer b Statistically significant at P < .05.c Enriched for patients with TNBC in phase 2 of the study (eTable 8 in Supplement 1); 3.6% of all subtypes were included in phase 1 of the study.

Table 4 .
Comparison of Universal Genetic Testing Studies

Table 4 .
27mparison of Universal Genetic Testing Studies (continued)Abbreviations: NA, not available; P/LP, pathogenic or likely pathogenic.The Mayo Clinic Breast Cancer Study was 1 of the studies contributing to the Cancer Risk Estimates Related to Susceptibility consortium.27TraditionalHereditary Breast and Ovarian Cancer Testing Criteria Used at McGill Hospitals eTable 2. Germline Pathogenic and Likely Pathogenic Variants (n=729) eTable 3A.Cases With a Pathogenic or Likely Pathogenic Variant eTable 3B.Patients With a Pathogenic/Likely Pathogenic Variants That Were Not Found To Be Disease Causing Due To a Low Allele Frequency in Blood DNA and Absence From the Breast Cancer Tumor eTable 4. Variants of Uncertain Significance (n=98) eTable 5. Distribution of Germline Pathogenic/Likely Pathogenic Variants and Variants of Uncertain Significance by Parental Origin eTable 6. Predictors of Pathogenic or Likely Pathogenic Variants in Secondary 14-Gene Panel Among Women With Newly Diagnosed Invasive Breast Cancer (n=659) eTable 7. PARP Inhibitor Eligibility eTable 8. Clinical Characteristics of Patients Stratified by Phase I and Phase II of the Study (n=729) eTable 9. Biologic Subtype by Age Group (n=729) eFigure 1A.Lollipop Plot for BRCA1 Germline Variants eFigure 1B.Lollipop Plot for BRCA2 Germline Variants eFigure 1C.Lollipop Plot for PALB2 Germline Variants eFigure 1D.Lollipop Plot for ATM Germline Variants eFigure 1E.Lollipop Plot for CHEK2 Germline Variants eFigure 2. World Map Showing Ethnicity of Patients With a Pathogenic Variant or Variant of Uncertain Significance, per Region eReferences.