Title: Genomic landscape of ER-positive HER2-low early-stage breast cancers in the FLEX Study: MammaPrint, BluePrint and whole transcriptome analysis

Publication: SABCS 2023, PO4-02-02


Abirami Sivapiragasam, Adam Brufsky, Hannah Linden, Natasha Hunter, Cynthia Osborne, Joyce O’Shaughnessy, Sami Diab, Robert Maganini, Manojkumar Bupathi, Sung Ho Lee, Theodore Kim, Josien Haan, Lavanya Samraj, Katie Quinn, William Audeh, FLEX Investigators’ Group


Antibody-drug conjugates (ADCs) continue to emerge for the treatment of a new subset of patients with HER2-low breast cancer. There is limited evidence to demonstrate HER2-low tumors as a distinct biological subtype and why/if these tumors benefit from ADCs. To improve our understanding of this newly defined HER2 category of breast cancers, we evaluated clinical characteristics, MammaPrint (MP), BluePrint (BP), and the whole transcriptomic profile of HER2-low breast cancers in FLEX study.


FLEX (NCT03053193) is a prospective, observational trial that includes stage I-III breast cancer patients who undergo MammaPrint testing (with or without BluePrint) as standard of care, and consent to full transcriptome and clinical data collection. In this study, clinically ER+/HER2- tumors were analyzed. The HER2-low cohort group (n=1698) was defined as HER2 IHC 1+ (ISH positive excluded) and IHC 2+, ISH Negative, and the HER2-0 group (n=1181) was defined as HER2 IHC 0. MP classified tumors as Low Risk or High Risk (further stratified as High 1 and High 2). MP together with BP categorized tumors as Luminal A (MP Low Risk), Luminal B (MP High Risk), HER2 or Basal. Two-tailed proportional z-test was used to compare clinical features and genomic subtypes of HER2-low vs. HER2-0 and the limma R package for differential gene expression analysis (DGEA). P-values were adjusted for multiple testing by Benjamini-Hochberg; significant differentially expressed genes (DEGs) had a p-value < 0.05 and a fold change >2.


In this FLEX study, clinically ER+/HER2- tumors showed that the clinical characteristics between HER2-low and HER2-0 did not differ significantly except higher percentage of premenopausal within HER2-low (23% vs 17%, p < 0.01). Of all HER2-low patients, 46% were MP High Risk as within HER2-0 patients (44%, p = 0.27). In both groups with MP High Risk tumors, there was a higher frequency of MP High 1 (83% in HER2-low, 80% in HER2-0) compared to MP High 2 (17% in HER2-low, 20% in HER2-0). BP subtyping showed similar distribution for Luminal subtype between HER2-low and HER2-0, but the frequency of ER+, Basal subtype was lower in HER2-low (2.5%), compared to HER2-0 (4.5%) (p=0.005). Principal component analysis (PCA) of the 500 most variable genes did not reveal a separation of HER2-low and -0 tumors, but clustering was apparent when tumors were classified by BP. DGEA within Basal tumors revealed no DEGs. Within Luminal A tumors, more than 1800 DEGs were identified, and within Luminal B tumors, nearly 300 DEGs were identified. However, all DEGs were < 1.5-fold change: mean, max (1.09, 1.38) for Luminal A and (1.12, 1.44) for Luminal B. In addition, a significant difference (p<0.01) towards increased ERBB2 (HER2) expression was detected from HER2-0 to HER2-low, but there was a large overlap of expression between the 2 groups.


Our study showed that HER2-low and HER2-0 tumors are clinically and biologically similar. The differences in ERBB2 and low-level genome wide expression detected between HER2-low and HER2-0 should be further investigated to determine how these changes affect tumor biology and treatment benefit. The biological heterogeneity among IHC-defined HER2-negative tumors was better captured by MammaPrint and BluePrint than IHC. MammaPrint identified 53% of HER2-low tumors as Low Risk, a subgroup of patients known to have good outcomes without chemotherapy. Our data suggest a subset of patients with HER2-low, MP Low Risk tumors could be spared from the potential toxicities of ADCs.