This study was a noninterventional, retrospective medical record review of adult patients with HER2-negative ABC and germline BRCA1/2 mutations who received cytotoxic chemotherapy. The RTI International Institutional Review Board determined that this study was not human subjects research (STUDY00020382), as the data were unidentifiable. ABC was defined as locally advanced breast cancer that was not amenable to curative radiation or surgery, or as metastatic disease. The sample was further stratified into two groups consisting of patients with advanced triple-negative breast cancer (TNBC) and patients with hormone receptor–positive (HR+)/HER2-negative ABC. Data describing patient and clinical characteristics, treatment patterns, and health outcomes were abstracted directly by eligible oncologists. A purposive, quasi-random sampling method was employed through the selection of medical records for patients whose last names began with a randomly generated letter. The inclusion criteria for oncologists, inclusion criteria for medical records, and exclusion criteria for medical records are presented in Table 1 . Medical record abstraction occurred from July 2019 to May 2020, which allowed for extraction of follow-up data (when available) until May 2020.

Demographic characteristics, including primary medical specialty, primary practice setting, and geographic region of practice, were collected for the recruited oncologists. For eligible patients, demographic data, as well as baseline clinical characteristics related to their initial breast cancer diagnosis and the diagnosis of ABC, were collected from medical records. Patient demographic data included sex, race, and family history of a BRCA-related cancer. Baseline clinical characteristics included age and Eastern Cooperative Oncology Group (ECOG) performance status at ABC diagnosis.

Data measures related to the treatment of breast cancer were collected before and after diagnosis of ABC ( Figure 1 ). Treatment choices before the diagnosis of ABC included neoadjuvant treatment(s), surgery, adjuvant chemotherapy, adjuvant targeted therapy/biologics, adjuvant hormonal therapy, and radiation therapy. Types of treatment approaches, along with total number of systemic lines, collected after diagnosis of ABC included surgery, chemotherapy, targeted therapy/biologics, hormonal therapy, and radiation therapy.

Health outcomes included time-to-event estimates (i.e., real-world PFS [rwPFS], overall survival [OS]) and prespecified treatment-related toxicities. Dates of progression and death (i.e., whether the patient died during the follow-up period and whether the death was related to breast cancer) were documented. Time-to-event estimates were calculated from the date of ABC diagnosis and the start of each treatment line (received after ABC diagnosis). Several treatment-related hematological (i.e., anemia, neutropenia, thrombocytopenia) and nonhematologic toxicities (i.e., hand-foot syndrome [or palmar-plantar erythrodysesthesia], nausea, vomiting) commonly observed in clinical trials and real-world studies (10, 20) were documented for each treatment line received after diagnosis of ABC.

Measures were summarized descriptively through tabular and graphical display of mean values, medians, ranges, and standard deviations for continuous variables of interest and frequency distributions for categorical variables. Time-to-event outcomes (i.e., rwPFS, OS) were calculated using Kaplan-Meier methods. All analyses were conducted using SAS (version 9.4 or higher) statistical software. This study was descriptive, with the primary goal of retrospectively summarizing existing treatment patterns and patient outcomes.

A total of 97 oncologists provided abstraction of 305 medical records of adult patients with germline BRCA1/2-mutated, HER2-negative ABC. The majority were medical oncologists (62.9%) and practiced in a private hospital or clinic (48.5%) ( Supplementary Table 1 ). The geographic locations were approximately evenly distributed among the Southern (30.9%), Northeastern (26.8%), Midwestern (21.7%), and Western US (20.6%).

Patient demographics and clinical characteristics are presented in Table 2 . Of the 305 included medical records, 194 (63.6%) patients had advanced TNBC and 111 (36.4%) patients had HR+/HER2-negative ABC. The median age at ABC diagnosis for the overall patient population with germline BRCA1/2-mutated, HER2-negative ABC was 57.3 (range, 33.3-79.8) years. Similar median ages at ABC diagnosis were found for the TNBC (57.2 years) and HR+/HER2-negative (58.2 years) subgroups. There was one male patient in the TNBC subgroup. The overall group was mostly white (76.4%); this composition decreased for the TNBC subgroup (71.7%) and increased for the HR+/HER2-negative subgroup (84.7%). The percentages of patients of Ashkenazi Jewish ancestry were similar across the overall group (11.5%), the TNBC subgroup (12.9%), and the HR+/HER2-negative subgroup (9.0%). The percentages of patients with no known family history of a BRCA-related cancer were 24.6%, 27.8%, and 18.9% for the overall group, the TNBC subgroup, and the HR+/HER2-negative subgroup, respectively. The percentage of patients with an ECOG performance status of 0 or 1 at ABC diagnosis was 74.6% for the overall group, 70.2% for the TNBC subgroup, and 83.2% for the HR+/HER2-negative subgroup.

For both the TNBC ( Figure 2 ) and HR+/HER2-negative ( Figure 3 ) subgroups, first-line treatments were similar, with chemotherapy being the most commonly used treatment, but differences in treatments between these subgroups emerged in the second-line setting and beyond. First-line treatment for the TNBC subgroup was composed of mostly nonplatinum-based chemotherapy (60.8%) and platinum-based chemotherapy (39.2%). In second-line treatment, the TNBC subgroup had nearly equivalent rates of nonplatinum-based chemotherapy (40.2%) and PARP inhibitors (42.0%). Third-line treatment in the TNBC subgroup again had similar rates for nonplatinum-based chemotherapy (39.1%) and PARP inhibitors (43.5%). In the HR+/HER2-negative subgroup, first-line treatment consisted of chemotherapy for 78.4% of patients and endocrine-based therapy (EBT) for 19.8% of patients. In second-line treatment, most of the HR+/HER2-negative subgroup received EBT (51.2%), with lower percentages of chemotherapy (31.0%) and PARP inhibitors (13.1%). In third-line treatment for the HR+/HER2-negative subgroup, chemotherapy again became the dominant treatment (50.0%), with decreased use of EBT (27.8%) and increased use of PARP inhibitors (19.4%).

During the first-, second-, and third-line treatments, rwPFS was assessed for the TNBC and HR+/HER2-negative ABC subgroups ( Table 3 ). Among patients with TNBC, the median PFS for first-line treatment (n = 190) was 11.6 months (95% confidence interval [CI], 9.9-13.6); for second-line treatment (n = 64), the median PFS was 9.0 months (95% CI, 6.2-11.4); and for third-line treatment (n = 26), the median PFS was 5.9 months (95% CI, 3.5-10.9). Among patients who were HR+/HER2 negative, the median PFS for first-line treatment (n = 110) was 12.1 months (95% CI, 9.1-14.3); for second-line treatment (n = 68), the median PFS was 14.2 months (95% CI, 10.7-29.5); and for third-line treatment (n = 29), the median PFS was 6.5 months (95% CI, 4.4-9.2). Overall survival from index chemotherapy was assessed for the TNBC and HR+/HER2-negative ABC subgroups ( Table 4 ). Patients in the TNBC subgroup had a 2-year OS rate of 73.9%, which decreased to 53.1% at 5 years. The median OS was not estimable during the study follow-up. Patients in the HR+/HER2-negative subgroup had a 2-year OS rate of 77.0%, which decreased to 44.2% at 5 years, and a median OS of 45.9 months.

The prespecified treatment-related hematological toxicities and evidence of hand-foot syndrome, nausea, and vomiting during the first-, second-, and third-line treatments were documented for the TNBC and HR+/HER2-negative ABC subgroups ( Table 5 ). For first-line treatment, the most commonly experienced toxicities for both the TNBC (n = 190) and HR+/HER2-negative (n = 110) subgroups were nausea (41.1% and 27.3%, respectively) and anemia (26.3% and 20.9%, respectively). Toxicities were not documented for 24.7% of the patients in the TNBC subgroup and 44.6% of patients in the HR+/HER2-negative subgroup in the first-line setting. For second-line treatment, the TNBC subgroup (n = 64) documented anemia (32.8%) and the HR+/HER2-negative subgroup (n = 68) documented neutropenia (17.7%) as the most commonly experienced toxicities. Toxicities were not documented for 32.8% of the TNBC subgroup and 51.5% of the HR+/HER2-negative subgroup in the second-line setting. For third-line treatment, nearly half (44.8%) of the 29 patients in the HR+/HER2-negative subgroup did not document toxicities, compared with 15.4% of patients who did not document toxicities in the TNBC subgroup (n = 26). However, the small sample sizes for third-line treatment warrant caution in interpretation of the toxicities experienced.

This study represents the first dataset to examine real-world treatment patterns and outcomes exclusively among HER2-negative patients with germline BRCA1/2-mutated ABC and provides valuable findings related to how targeted therapies, such as PARP inhibitors, are being incorporated into real-world practice in this select patient population. However, given that data abstraction occurred between July 2019 and May 2020 (maximum follow-up until May 2020), the incidence of PARP inhibitor therapy may have changed, and PARP inhibitors might be prescribed more often. A limitation of this study is that patients selected for study inclusion represented a purposive sample of medical records obtained from oncologists who were willing to participate in the study. Therefore, study findings may not be generalizable to the overall population of patients with germline BRCA1/2-mutated ABC. To minimize this limitation and improve generalizability of findings, oncologists were recruited from a variety of geographic regions and practice types. Another limitation was that data were limited to medical records to which oncologists had access. Information on health care services received outside the physician’s care setting that was not recorded in the medical record was unavailable for this study (e.g., treatment for adverse events received through care rather than the abstracting physician specialist). Accordingly, some adverse events may not have been included because they were reported elsewhere and not collected in the medical record. Also, while not all known treatment-related toxicities were evaluated in this study, toxicities commonly observed in clinical trials and real-world studies were assessed. Lastly, to homogenize the sample, patients in this study were required to have initiated at least one cytotoxic chemotherapy for ABC, which may have resulted in selection bias and could be the reason for relatively lower utilization of endocrine or endocrine-based therapies and PARP inhibitors observed in our study population.

Given the growing evidence that PARP inhibitors may improve PFS better than chemotherapy while potentially offering a more manageable toxicity profile and improved quality of life, clinicians should consider the use of targeted agents such as PARP inhibitors in earlier lines of therapy for ABC. Further prospective trials examining the sequencing of targeted agents in ABC are needed. Ultimately, the results of this study may contextualize the results of ongoing clinical trials, inform routine clinical-practice, and improve patient lives.