Abstract PO1-28-06: Development of a plasma-based real-time qPCR gene expression assay for targeted screening and diagnosis of early-stage breast cancer
Martin Keiser,Elizabeth Cormier-May,Matthew Alderdice,Joy Kavanagh,William Guesdon,Heather Healy,Nathalie Jean-Charles,Jay Harness
DOI: https://doi.org/10.1158/1538-7445.sabcs23-po1-28-06
IF: 11.2
2024-05-03
Cancer Research
Abstract:Background Over 12 million US women undergo follow-up breast imaging and exams annually due to breast cancer suspicions, incurring an 2.18 billion is attributed to false-positive breast biopsies. Recent updates published to the Mammography Quality Standards Act (MQSA) now recommend supplemental imaging for women with heterogeneously dense and extremely dense breasts, regardless of suspicious findings. Given sensitivity limitations associated with mammography in dense-breasted women, as low as 30-47%, and non-specificity associated with ultrasound and MRI, these recent recommendations are likely to further strain the healthcare system and place substantial burden on patients and providers. With an estimated 25 million dense-breasted women who regularly screen for breast cancer now considered at-risk, there is a critical need for affordable and highly accurate testing, offering better specificity without sacrificing sensitivity. To address this gap, we developed an affordable and accurate plasma-based gene expression assay using real-time qPCR for targeted screening and diagnosis of early-stage breast cancer. Methods In previous studies, 26 cross-correlated mRNA gene targets were discovered and independently validated for non-invasive breast cancer detection across five independent patient cohorts using microarray gene expression profiling, comprised of peripheral blood mononuclear cells (n=337) and saliva (n=20), demonstrating strong efficacy, reproducibility, and concordance across bio-fluids and assay platforms in all studies. A sixth independent validation cohort comprised of 203 plasma samples was subsequently obtained for real-time qPCR clinical assay development and validation in our CLIA laboratory. The CLIA validation cohort was designed to be representative of different diagnosis status, stage, and ethnicity, and included Caucasian, Black/African American, Hispanic, and Asian women. After normalizing mRNA expression and clustering analysis, the signature was refined to 8 target genes and was assessed in a cohort of 87 plasma samples [Table 1] to verify the assay's diagnostic performance for stage I breast cancer detection, under locked laboratory protocols. We leveraged machine learning methods derived from XGBoost classification, a supervised-learning algorithm that uses sequentially built shallow decision trees to provide accurate results and avoidance of overfitting. Results The XGBoost model, selecting from the 8 mRNA gene targets and patient age, achieved >99% sensitivity, 89% specificity for stage I breast cancer detection in the held-out test set, with an overall diagnostic accuracy of 94.5%. Normalized data showed significant differences in gene expression between healthy controls and stage I patients, and distinct clustering was observed for the 8 mRNA gene signature, including patient age. Conclusions Our plasma-based real-time qPCR gene expression clinical assay, with machine learning, is positioned as a highly accurate non-invasive tool for targeted screening and diagnosis of early-stage breast cancer. Furthermore, a clinically validated assay in this space addresses the need for more targeted diagnostics, while serving as a promising tool for clinicians as they make critical care decisions across the breast health continuum, accelerating time to earlier and more precise intervention and treatment, mitigating unnecessary healthcare expenditures, and reducing the mental, emotional, and financial burden on patients and their families. Citation Format: Martin Keiser, Elizabeth Cormier-May, Matthew Alderdice, Joy Kavanagh, William Guesdon, Heather Healy, Nathalie Jean-Charles, Jay Harness. Development of a plasma-based real-time qPCR gene expression assay for targeted screening and diagnosis of early-stage breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl) nr PO1-28-06.
oncology