3D genomics identifies previously unreported yet clinically actionable gene fusions and complex rearrangements in a cohort of driver-negative colorectal carcinoma tumors.

Abstract

e15608 Background: Molecular characterization by next-generation sequencing (NGS) has been unable to detect pathogenic driver mutations in a significant number of tumors. For example, 40% of colorectal carcinomas (CRC) are RAS and BRAF wild-type and do not have therapeutic targets after comprehensive molecular characterization using FISH, IHC and NGS. Analysis of 3D genomics data produced by DNA-based chromosome conformation capture (Hi-C) assay has been shown to detect gene fusions and other rearrangements in tumor biopsies. We applied Hi-C in a cohort of CRC tumors previously characterized using FISH, IHC, whole exome sequencing (WES) and whole transcriptome sequencing (WTS) by a large commercial reference laboratory, where no targets had been reported, to evaluate the potential for Hi-C to identify novel therapeutic targets. Methods: FFPE tissue from 14 CRC tumors were subjected to Hi-C using the Arima HiC+ for FFPE kit. Hi-C libraries were sequencing to ~100M raw read-pairs on an Illumina NovaSeq, and fusions were called using the Arima-SV pipeline. To attribute clinical significance, we characterized fusions based on whether the fusion event involved a Tier 1 therapeutic level biomarker or a Tier 3 diagnostic or prognostic biomarker according to NCCN and OncoKb, or a Tier 2 biomarker if the biomarker was the target of a therapy in an ongoing clinical trial. Results: Hi-C detected fusions implicating a Tier 1 biomarker in 50% (7/14) of tumors, a Tier 2 level biomarker in 14% (2/14), and a Tier 3 level biomarker in 14% (2/14). Homologous recombination repair (HRR) gene rearrangements were detected in 29% (n = 4/14) of CRC tumors, disrupting HRR genes ( BRIP1, BRCA1, BRCA2, RAD51D). NRG1 gene fusions were detected in 14% (2/14), and amplified complex rearrangement carrying KRAS was detected in one CRC case. A complex rearrangement involving extrachromosomal DNA (ecDNA) amplification carrying EGFR was detected in two cases that were both previously found to have EGFR amplifications, but the ecDNA nature was not previously known and was confirmed by FISH. Lastly, Hi-C detected gene rearrangements involving POLE or APC in two CRC cases, alterations in which are known to increase risk for colonic polyposis. Conclusions: Fusion detection by Hi-C in CRC may uncover numerous therapeutic biomarkers not currently detected by molecular profiling. Our data suggests that homologous recombination deficiency (HRD) may frequently result from HRR gene rearrangements readily detectable by Hi-C, and this alternate mechanism for HRD may serve as a therapeutic biomarker for HRD directed therapies. Our data also suggests that therapeutically targetable gene fusions and complex rearrangements involving therapeutic level biomarkers are also readily detectable by Hi-C, detection of which may increase the number of CRC patients treatable with targeted therapies.