Abstract PS4-02-10: Agreement of cell-free DNA methylation-based molecular breast subtyping and tissue subtyping in hormone receptor positive metastatic breast cancer: a clinical cohort analysis

Abstract Background: Hormone receptor (HR) and HER2 status, typically assessed on tissue biopsies, are essential for breast cancer (BC) management. As multiple therapeutic options are available for receptors positive metastatic BC (mBC), ongoing subtype assessment may be recommended to monitor changes under the selective pressure and/or evolution of disease. Cell-free DNA (cfDNA) showed reliability for biomarker evaluation and high concordance with tissue-based methods and may be an opportunity for non-invasive, real -time subtype assessment. This study aimed to evaluate agreement between cfDNA methylation-based molecular breast subtyping (MBS) and tissue status in a clinical cohort who underwent cfDNA testing as part of routine clinical care. Methods: Blood samples drawn from mBC patients with HR+ disease on the most recent tissue biopsy and tested with Guardant360 at Weill Cornell Medicine from Nov 2024-June 2025 were queried. A novel cfDNA MBS signature was applied to test results through bioinformatics and/or residual sample testing. MBS confidence scores for HR+/HER2+ where tumor fraction (TF) >0.5% were applied. In this research setting, MBS predicted overall subtype was generated to compare it to the most recent tissue biopsy available. Clinical and genomic factors were annotated for correlation; Fisher’s exact and t-tests were used for statistical analysis. Results: 73 samples were screened of which 42 (57.5%) were evaluable for MBS (TF>0.5%). Of those not evaluable, 16/31 (51.6%) had maximum variant allele frequency <1% and two had no circulating tumor DNA detected. Notably, 24/31 (77%) non-evaluable samples had >1 genomic alteration annotated as clonal hematopoiesis. Two had undetermined results with TF>0.5% but confidence below MBS thresholds, thus 40 samples were included in agreement analysis. A total of 31/40 (77.5%) samples had initial MBS agreement with tissue of whom 27 (87.1%) were HR+/HER2+. The median confidence score from MBS for HR+ prediction was 80% (range: 30%->90%). Of 9/40 (22.5%) samples with initial MBS disagreement with tissue, one (11.1%) differed in HR status (MBS HR-/tissue HR+) and 8 (88.9%) differed in HER2 status (3 MBS HER2+/tissue HER2-; 5 MBS HER2-/tissue HER2+). For the HR case, concomitant genomic analysis did not identify any alterations typically associated with HR+ (ESR1, PIK3CA, PTEN, AKT1). When exploring clinical factors, among samples that had MBS/tissue agreement vs. disagreement, there was no significant difference in: average time from tissue to ctDNA testing (521 days vs. 494 days), number of metastatic sites (2.84 vs. 2.78), number of prior therapies (2.78 vs. 2.35) or prior exposure to CDK4/6 inhibitors (64.2% vs. 77.8%); all p>0.05. Although TF was higher in cases with agreement (mean: 23.4% vs. 20.5%; median: 8.06% vs. 5.26%) this difference was not significant (p>0.05). Upon clinical review of cases with disagreement,1 had MBS HR+/HER2- yet had HR+/HER2- on previous tissue biopsies, concordant with the MBS subtype. A second case had MBS HR+/HER2+ yet HR+/HER2- on the most recent tissue biopsy done in 2023; this case had previous HR+/HER2+ biopsies in 2017 (primary diagnosis) and 2020 (metastatic relapse) and received anti-HER2 therapies, suggesting MBS may capture heterogeneity even after therapeutic intervention. Thus, adjusted agreement of MBS with additional clinical information was 82.5% (33/40). Conclusions: In this mBC cohort, cfDNA MBS showed reliable agreement to tissue biopsy and replicated previously reported accuracy. Subtype switching, heterogeneity and/or lower levels of MBS confidence may contribute to disagreement with tissue. MBS may be a feasible non-invasive option for continuous subtype assessment for HR+ mBC. Citation Format: L. Pontolillo, L. Bucheit, L. Tung, E. Nicolò, M. Serafini, N. Bayou, B. Pasto, L. Gerratana, E. Andreopolou, E. Bria, C. Reduzzi, M. Cristofanilli. Agreement of cell-free DNA methylation-based molecular breast subtyping and tissue subtyping in hormone receptor positive metastatic breast cancer: a clinical cohort analysis [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS4-02-10.