The Pan-Tumor Landscape of Gene Amplifications and Copy Number Amplification Ratio for Established and Emerging Clinical Targets.

PURPOSE: Gene copy number (CN) amplifications and protein overexpression are common drug targets, and detection relies on various methodologies, including next-generation sequencing-based CN, immunohistochemistry (IHC), and in situ hybridization (ISH). We investigated the pan-tumor landscape of amplifications and developed AmpRatio, a novel method of CN quantitation.

EXPERIMENTAL

DESIGN

Pan-tumor tissue (N = 486,340) and liquid (N = 85,635) samples underwent hybrid capture-based comprehensive genomic profiling. A genome-wide CN model for each sample was generated to estimate the purity, ploidy, and segment-level CN. AmpRatio was calculated by dividing gene CN/sample ploidy. A US-based deidentified clinicogenomic database was utilized to assess the relationship between ERBB2 AmpRatio and HER2 IHC/FISH and outcomes on anti-HER2 therapies.

RESULTS

Amplifications with varying degrees of gain were reported in 38.6% of pan-tumor tissue samples, most frequently MYC (5.6%), 11q13 (5.2%), ERBB2 (5.2%), and CCNE1 (3.2%). ERBB2 AmpRatio was associated with HER2 positivity by IHC/FISH in gastroesophageal [overall percent agreement (OPA) 90%] and breast (OPA 95%) cancers. Among patients treated with anti-HER2 therapies, ERBB2 AmpRatio significantly stratified outcomes within the ERBB2-amplified and IHC-defined HER2+ and HER2-low/ultralow populations. High concordance (sensitivity 88%) of amplification detection in liquid biopsy versus tissue was associated with higher AmpRatio and ctDNA tumor fraction ≥20%.

CONCLUSIONS

CN amplifications are prevalent and diverse biomarkers, and AmpRatio is variable across genes and tumor types. ERBB2 AmpRatio is associated with outcomes to HER2-directed therapies and may have utility alongside IHC for clinical decision-making. With the increasing number of therapies targeting amplifications/overexpression, it will be important to define harmonized methods for CN quantification for optimal patient selection.