1095 Background: MBC remains incurable due to the invariable development of resistance, with most deaths resulting from HR+ MBC that acquires resistance to endocrine therapy (ET) combinations. HR and HER2 remain the most important biomarkers in MBC but, while it is known that receptor status and expression based molecular subtypes (e.g. PAM50) can vary over an individual patient’s (pts) disease course, serial evaluation is currently limited in routine clinical care by tissue requirements. Methods: The EMBRACE cohort study collects biospecimens with matched clinical annotations from pts with MBC at Dana-Farber Cancer Institute. A subset of 510 plasma samples collected around CDK4/6i treatment (tx; pre-tx: -90 to +30 days from start; on-tx; post-tx: -30 to +90 days from end) from 323 pts with HR+ MBC were assessed with Guardant360 Liquid. MBS calls were generated using a previously described cell-free DNA methylation-based deconvolution model that classifies HR and HER2 status in cases with tumor fraction (TF) ≥0.5% with accuracy >85%, and compared with tissue-based HR and HER2 status by immunohistochemistry (IHC). Results: Of 510 samples profiled, 308 (60.4%) were evaluable for MBS with TF ≥0.5%. The evaluable rate was lowest in on-tx samples at 48.7%, in line with lower median TF in this set. Of evaluable samples, 285 (92.5%) aligned with sufficient confidence to report a subtype: 273 (88.6%) of evaluable samples were classified as HR+, 23 (7.5%) undetermined, and only 12 (4.9%) HR-, all of which were either from on- or post-tx timepoints. MBS calls on pre-tx samples showed high concordance with clinical IHC, with higher concordance at metastatic diagnosis (91%) than primary diagnosis (86%). Notably, all discordances were cases abstracted as HR+/HER2- at metastatic diagnosis by IHC where MBS yielded HR+/HER2+. Overall, samples from 28 pts were assigned a HER2+ MBS and, in 9 (32%) of these, a concordant genomic alteration in ERBB2 (copy number gain, gain-of-function mutation and/or fusion) was detected on Guardant360 Liquid. Out of 48 pts with multiple classifiable samples, 11 (22.9%) showed differences in MBS across serial samples including 9 with changes in HER2 (5 gaining and 4 losing HER2 MBS), and 2 with changes in HR (both with an on-tx sample classified as HR- but the subsequent post-tx sample HR+). Conclusions: A methylation-based liquid biopsy assay can classify MBS with high correlation to clinical pathology in pre-tx samples and facilitate identification of discordant, or perhaps heterogenous, subtype while capturing the dynamics of MBS over disease course and treatment. Future studies are indicated to investigate the clinical utility of this assay as a biomarker for therapy selection.
Abravanel et al. (Wed,) studied this question.