Characterization of diverse targetable alterations in ERBB2 and ERBB3 in 93,465 non-small cell lung cancers (NSCLC).

Abstract

9094 Background: Recently, fam-trastuzumab deruxtecan-nxki was approved for NSCLC with selected activating ERBB2 (HER2) mutations. Patritumab deruxtecan, a novel HER3 directed antibody drug conjugate, exhibits in vitro activity against HER3 mutations. We present a comprehensive landscape of ERBB2/ ERBB3 alterations (alt) that may predict sensitivity to HER2/HER3 targeted therapies in development. Methods: We queried an institutional database (Foundation Medicine) of tissue or liquid comprehensive genomic profiling (CGP) results from 93,465 patients with NSCLC (85,704 tissue and 7,761 liquid samples). ERBB2 amp was defined as ≥ sample ploidy +3. Activating/likely activating alts were called using annotations including presence in COSMIC, functional knowledge of the gene affected, internal insights, and characterization in the literature. ctDNA tumor fraction (TF) on FoundationOneLiquid CDx was estimated using a composite algorithm. Results: Activating/likely activating alt in ERBB2 and ERBB3 were detected in 3.9% (n = 3337) and 0.8% (n = 696) of NSCLC tissue samples, respectively ( ERBB2: 2.0% amp, 1.7% mutations [mut; SNVs or indels], 0.001% rearrangements [RE], and 0.2% multiple; ERBB3: 0.5% amp, 0.3% mut, and 0.004% multiple). Age at diagnosis, sex, and predicted ancestry were similar among patients with ERBB2alt and ERBB2wt tumors. ERBB2 alt were more frequently observed in non-squamous NSCLC (4.3% vs 2.3% in SCC), largely driven by differential ERBB2 mut frequencies (2.0% vs 0.4%). ERBB2 mut were most common in the kinase domain (KD; 70%, 56% of which were exon 20 insertions [ex20ins]), followed by the extracellular domain (ECD; 20%) and transmembrane domain (TMD; 6%), while ERBB3 mut were most common in the ECD (88%, most commonly V104L/M and G284R). ERBB2 RE or splice mut affecting exon 16 were detected in 23 cases (0.03%). While ERBB2 alt were mutually exclusive with other oncogenic drivers ( KRAS, ALK, BRAF, EGFR, MET, ROS1, RET; each p < 0.05), ERBB3 alt were found to co-occur with both ERBB2 (8.5% vs 3.9% ERBB2wt, p < 0.001) and MET (8.6% vs 5.2% ERBB2wt, p < 0.001) alt. A similar distribution of ERBB2 alt (1.6% amp, 2.4% mut, 0.4% multiple) was detected in liquid samples with elevated TF (≥10%; n = 1458); across all liquid samples, ERBB2 mut frequency was comparable to tissue (1.5%) but frequency of amp was 0.5%. In our study, 26% of ERBB2 mut tissue samples had activating/likely activating mut not included in the DESTINY-Lung01 trial, most notably 100% of both TMD mut (n = 99) and exon 16 alt (n = 25), as well as 35% of ECD mut (117/331) and 14% of KD mut (162/1145, including 6% of ex20ins). Conclusions: Diverse ERBB2 and ERBB3 alt in NSCLC may predict sensitivity to anti-HER2/3 therapies, but not all classes of alts are detected/reported by all profiling assays. Utilization of comprehensive testing to guide biomarker definitions, treatment selection, and clinical trial enrollment is imperative.