LAUSR

Background: There continues to be an active debate as to the relative contribution of health services, sociodemographic, clinical, and genomic factors in breast cancer disparities. Efforts to understand the contributions of each of these factors have historically focused on separate investigations of socioeconomic, demographic, and genomic elements. However, true multidisciplinary investigation of these factors requires the existence of datasets that combine each of these elements together into population-level, real world cohorts. To this end we created the first-in-kind linkage of SEER-Medicare data to physical tumor samples to combine clinical, health services, and genomic data into a single cohort and used it to investigate the impact of screening and socioeconomic status on early stage breast cancer biology and mortality.Methods: This retrospective study used formalin-fixed, paraffin-embedded (FFPE) breast cancer (BC) tissue collected between 1992-2006 within the Iowa and Hawaii SEER Residual Tissue Repositories (SEER-RTR). Medicare fee-for-service claims data were linked for participating patients age 65 and older. Molecular subtyping and exploratory genomic analyses were completed using the NanoString Breast Cancer 360 (BC360) gene expression panel containing the PAM50 classifier. Screening status was assessed by validated claims-based algorithms. Factors associated with overall (OS) & breast cancer-specific (BCS) survival were analyzed in women aged 66-75 with clinically high risk ER+ disease (M0, T1b+, LN+) using Cox proportional hazards models combining sociodemographic, clinical, and genomic data.Results: SEER-Medicare data were available for 1,232 women, of which 379 (30.7%) were diagnosed by screening mammogram. Screen-detected disease was associated with lower T stage, N stage, and improved OS (HR 0.72) & BCS (HR 0.68) in multivariable analysis. Molecular analysis of 130 luminal A/B cases revealed superior outcomes of luminal A and luminal B tumors compared to symptom detected tumors of the same molecular subtype (p = 0.02). In multivariable cox proportional hazards models increased overall mortality was associated with impoverished neighborhoods, PR- receptor status, upregulation of TGF-beta and p53 dysregulation, whereas protective factors included upregulation of androgen receptor (AR), stromal, and cytotoxicity signaling (all P Citation Format: Timothy J Robinson, Lauren Wilson, Paul K Marcom, Melissa Troester, Charles F Lynch, Brenda Hernandez, Edgardo P Castellar, Heather Ann Brauer, Lindsey Enewold, Michaela Dinan. Comprehensive analysis of health services, sociodemographic, clinical, and genomic factors driving locally advanced breast cancer mortality via a first-in-kind linkage of SEER-Medicare data with physical tumor samples [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr SS1-11.