Trends in and determinants of germline BRCA1/2 testing in patients with breast and ovarian cancer.

Abstract

10583 Background: Germline BRCA1/2 testing (GT) is instrumental in identifying patients with breast and ovarian cancer who may be eligible for biomarker-driven poly ADP ribose polymerase inhibitor (PARPi) therapy. Little is known about recent trends and determinants of GT since PARPi were approved for these patients. Methods: We performed a retrospective cohort study of patients with breast and ovarian cancer who were eligible for GT (diagnosed with breast cancer under age 45, triple negative breast cancer under age 60, male breast cancer or ovarian cancer) between 1/2011 and 3/2020 in the nationwide Flatiron Health EHR-derived deidentified database. Duration of follow-up was at least one year for each patient. Spline regressions estimated the annual prevalence of GT within one year of diagnosis. Multivariable log binomial regressions estimated adjusted relative risks (RR) of GT by patient and tumor characteristics. Multiple imputation with chained equations was conducted for missing data. Results: Among 2,982 eligible patients with breast cancer, 1,682 (56%) underwent GT within one year of diagnosis with a median time of 42 days. GT increased from 37% in 2011 to 68% in 2020, with a significantly higher RR after PARPi were approved for breast cancer in 1/2018 (RR 1.35, 95% CI 1.20-1.51). In multivariable analyses, there were no appreciable differences by sex, race or ethnicity, but there was a negative linear relationship between GT and age (RR 0.93, 95% CI 0.90-0.96 for every 5 years). After adjusting for age, the 87 breast cancer patients with Medicare were also less likely to undergo GT despite being eligible (RR 0.67, 95% CI 0.48-0.95 vs commercial insurance). Among 5,563 eligible patients with ovarian cancer, 1,968 (35%) underwent GT within one year of diagnosis with a median time of 101 days. GT increased from 23% in 2011 to 53% in 2020, with a significantly higher RR after PARPi were approved for ovarian cancer in 12/2014 (RR 2.25, 95% CI 2.01-2.52). Although insurance status was not a significant determinant of GT in patients with ovarian cancer, older age (RR 0.95, 95% CI 0.93-0.97 for every 5 years) and Black race (RR 0.80, 95% CI 0.66-0.98 vs white race) were associated with a lower likelihood of GT in multivariable analyses. All results remained similar in pre-planned sensitivity analyses restricted to the post-PARPi approval period, limiting to patients who remained alive one year after diagnosis and using the non-imputed dataset. Conclusions: GT remains underutilized in patients with breast and ovarian cancer. Although GT has increased since PARPi were approved for this population, significant disparities by age, race and insurance status persist but differ by tumor type. This study is limited by potential misclassification due to missing GT performed outside the Flatiron Health network. Multifaceted patient-, clinician- and system-level strategies are needed to ensure that all eligible patients receive GT.