Molecular heterogeneity of ERBB2-altered cancers: A pan-cancer landscape across US and Japanese cohorts.

Therapies targeting ERBB2 (HER2) are rapidly expanding toward tissue-agnostic indications. However, the diversity of ERBB2-altered tumors across different cancer types and populations remains poorly characterized. We performed a comprehensive pan-cancer molecular analysis of ERBB2-altered tumors using two large clinical genomic datasets: AACR Project GENIE (n = 210,640) and the nationwide Japanese comprehensive genomic profiling database (C-CAT; n = 80,600). Across all cancer types, the prevalence of ERBB2 mutations was 1.4% in GENIE and 2.2% in C-CAT, whereas ERBB2 amplification occurred in 2.1% and 6.5% of tumors, respectively. Kinase domain mutations such as L755S in breast cancer and Y772_A775dup in lung cancer were enriched in a cancer type-specific manner, while R678Q in the juxtamembrane domain was common in colorectal and gastroesophageal cancers. ERBB2 compound mutations were rare (<1% in both cohorts) and did not significantly impact overall survival in the cohort treated predominantly before the advent of novel HER2-selective targeted therapies. ERBB2 mutations and amplifications also displayed divergent co-mutational landscapes: mutations frequently co-occurred with ARID1A mutations, whereas amplifications were enriched for TP53 mutations and CCNE1 amplification. In colorectal cancer, ERBB2 mutations were associated with higher tumor mutational burden (TMB) and frequent KRAS co-mutations, while in lung adenocarcinoma, ERBB2 amplifications exhibited higher TMB and a heavier co-mutation burden. A survival analysis revealed that specific co-occurring alterations, including MAP2K1 mutations in colorectal cancer and CCNE1 amplification in breast cancer, were associated with distinct prognostic outcomes among ERBB2-amplified cases. Our findings demonstrate that "ERBB2-altered cancer" is not a homogenous entity. Optimizing precision oncology for ERBB2 requires integrating histological context, specific alteration types, and co-mutation profiles to guide therapeutic selection and overcome resistance.