Fig. 1

Description of the study cohorts. A Consort diagram of the HER2-low breast cancer (HLBC)-FPO cohort. A total of 99 patients were included, characterized by an IHC score 1+ (n = 34) or 2+ and FISH non-amplified (n = 65). HLBC-2 comprised lesions harboring a ERBB2 copy number (CN) < 4 and a ratio ERBB2/CEP17 ≤ 2 (FISH-negative, HLBC-2N, n = 15) and tumors with ERBB2 FISH results in the equivocal range (ratio ERBB2/CEP17 < 2, CN ≥ 4 and < 6, HLBC-2E, n = 50). B Consort diagram of the public Memorial Sloan-Kettering Cancer Center (MSKCC) BC cohort, which was exploited for comparison of mutational landscapes. The MSKCC BC cohort contains (i) discrete HER2 IHC scores, (ii) gene mutations identified with a DNA-based targeted panel, and (iii) histopathological data (ER status, histological grade). From 1756 samples, we derived two populations: (i) the unmatched cohort including 1317 annotated samples and (ii) the matched cohort, which contains 545/1317 samples matched by ER status and histological grade of the HLBC-FPO cohort. The detailed inclusion criteria are described in the “Methods” section and Additional file 1: Supplementary Methods section; a summary of the criteria is illustrated in the figure. C Consort diagram of the control cohort for RNA-based analyses. A control cohort of BC was retrieved from our archives (n = 45), composed of HER2-negative (score 0, n = 30) and HER2-positive (score 3+, n = 15). These cases were profiled by the NanoString BC360 panel, similarly to the HLBC-FPO cohort. Legend: *public control cohort for DNA data comparison; ^§cases score 2+ ERBB2 not amplified; red colored text for HLBC-MSKCC cases; **internal control cohort for RNA data comparison