利用FFPE肿瘤的Hi-C测序发现并功能性表征NSCLC中的增强子劫持致癌基因重排

INTRO: Molecular profiling of solid tumors has revealed many targetable biomarkers; in NSCLC, ~50% of patients harbor such biomarkers (EGFR mutations or ALK fusions). However, limited precision therapeutic options exist for the remaining ~50%. To address this need, we aimed to identify and functionally characterize novel tumor-driving genomic mechanisms involving known oncogenes using Hi-C sequencing of FFPE NSCLC tumors. METHODS: FFPE NSCLC tumors (n=97) were retrospectively selected from Stage III/IV patients whose prior CGP showed no EGFR/RAS mutations or gene fusions (“driver-negative”). Hi-C sequencing was performed by Arima Genomics, and rearrangements, gene fusions, and CNVs were identified with Arima-SV pipeline. Oncogenes within ~1 Mb of a rearrangement (“oncogene rearrangements”) were evaluated for potential enhancer hijacking events by analyzing breakpoint-crossing 3D regulatory interactions with putative enhancers at rearrangement partner loci. Functional characterization of rearranged oncogene activation was assessed by IHC. RESULTS: 95/97 (98%) Hi-C libraries passed QC and were deeply sequenced and analyzed. 35/95 (37%) of tumors had oncogene rearrangements involving 47 oncogenes. 26/95 (27%) tumors carried 33 rearranged oncogenes linked to response to targeted therapies (Level 1/2 therapeutic response evidence; OncoKb), including BRAF (1), ERBB2 (3), FGFR1 (9), FGFR2 (2), FGFR3 (2), NRG1 (4), NTRK1 (1), NTRK2 (1), RET (5), ROS1 (2), KRAS (2), and PIK3CA (1). 3/95 (3%) tumors had PD-L1 rearrangements. 8/95 (8%) tumors had 11 rearranged oncogenes of lower-level significance, such as PIK3CB (1), CCND1 (3), BCL6 (1), and MYC (2), CLDN18 (3), and HRAS (1). 23 had sufficient tissue remaining for further functional characterization and a corresponding commercially available IHC test for protein expression. 11/23 (48%) were expressed in >10% of the tumor cells, including FGFR1 (2/9, 22%), ERBB2 (2/3, 66%), NTRK1 (0/1, 0%), NTRK2 (1/1, 100%), ROS1 (0/2, 0%), CLDN18 (0/1, 0%), BCL6 (1/1, 100%), MYC (2/2, 100%), and CCND1 (3/3, 100%). Lastly, we observed that 10/11 (91%) rearrangements with both features of enhancer hijacking ((1) strong breakpoint-crossing oncogene 3D interactions and (2) partner locus enhancers) were expressed at the protein level, versus only 1/12 (8%) lacking one or both features. CONCLUSIONS: These data demonstrate the capability of Hi-C to detect rearranged oncogenes and predict their exogenous expression. Many activated rearranged oncogenes correspond to targets of FDA-approved or standard-of-care therapies, albeit potentially activated by a mechanism different than those well-established (point mutations or gene fusions). Further studies are needed to determine whether such oncogene rearrangements confer therapeutic susceptibility or serve as other predictive biomarkers or drug targets.