Background: Breast cancer is detected after stage I for many women, particularly women with dense breast tissue and those under 50, due to screening mammography performance and program participation. This report provides data for a new CE-marked blood test currently available in Canada. The test uses a non-fractionated blood sample and involves analysis of a gene expression panel (12 targets) using custom reagents and machine-learning informed software for identification of an active breast cancer signature that has been validated through a retrospective clinical study.
Methodology: Blood samples (2.5 ml) were collected from women aged 25 to 80 near the time of a screening mammogram (BI-RADS 1–2) or negative physical exam, or with a BI-RADs 3–5 score in secondary care but pre-biopsy, as part of the ongoing Identify Breast Cancer (IDBC) prospective study (NCT04495244). Participants’ (n = 1107: 240 asymptomatic breast cancer, 867 normal) blood was analyzed in a facility with ISO 13485:2016 certification. Machine learning-based model development was conducted on 132 cancer and 251 normal samples, and blind independent testing was performed on 724 samples (103 cancer, 621 normal). Analytical performance characteristics were evaluated, and clinical performance metrics reported with 99.5% confidence intervals (CI) computed through an exact binomial test.
Interim clinical analysis yielded an inferred accuracy of 92.2% (CI: 88.9%–94.6%) with a specificity of 94.3% (CI: 91.0%–96.4%) and sensitivity of 79.2% (CI: 65.5%–88.4%). For women under 50 (Table 1
), a specificity of 99.0% and a sensitivity of 91.7% were achieved. This analysis showed most (84%) cancers were stage 1 or 2 (59% Stage 1), lymph node negative (73.5%), and hormone receptor positive (75%; 10% HER2+, 5% TN). The median tumor size was 18.0 mm. Importantly, results also showed that small tumors (<10 mm; n = 19) and participants with negative lymph nodes were detected by the test. Linearity and reportable range of targets, analytical specificity, interference, sample and reagent stability, repeatability, and reproducibility of the assay were defined, evaluated, and established.
Table 1:Clinical Performance Metrics
Conclusions: Interim data shows strong analytical and clinical performance of the test for detection of active breast cancer as well as identification of those negative for breast cancer particularly women under 50 and those with small tumors and disease-free lymph nodes, supporting a potential role of the test as a screening option to supplement imaging approaches.
Conflict of interest:
Ownership: Kenneth Fuh, Robert Shepherd and Kristina Rinker are cofounders and partly own Syantra Inc.
Board of Directors: Robert Shepherd and Kristina Rinker are members of Syantra’s Board of Directors.
© 2022 Elsevier Ltd. All rights reserved. Published by Elsevier Inc. All rights reserved.