MYC amplification segment-size mutual exclusivity analysis with other oncogenic events in a pan-cancer sequencing cohort.

Abstract

3073 Background: MYC gene alterations resulting in increased copy number are recurrent oncogenic alterations across human malignancies. Presently no clear data exists to aid in interpreting MYC amplification characteristics for identifying MYC-dependent tumors. Methods: Analysis was conducted on a center-wide cohort of 68,920 tumor-normal matched samples from pediatric and adult patients sequenced with an FDA-authorized targeted sequencing panel (MSK-IMPACT). To ascertain tumor content, ploidy and allelic specific gene copy number, tumor-normal samples were analyzed with FACETS. Samples with tumor purity <20% were excluded. MYC amplification was defined as total copy number (TCN) minus ploidy ≥ 4. All samples were queried for presence of concurrent oncogenic or likely-oncogenic alterations based on OncoKb annotation. Co-occurrence (Co) and mutual-exclusivity (Me) analysis between MYC amplification according to segment size and other oncogenic alterations was performed with DISCOVER (false discovery rate threshold of 0.01) for each tumor type. Results: After removing low-quality samples (n=19,010/68,920), 49,910 samples were included in the analysis. 3,219 (6.4%) samples harbored MYC amplification, with segment size ranging from less than 100Kb to more than 100Mb, with 38% samples displaying focal amplifications shorter than 10 Mb and 62% broad amplifications longer than 10 Mb. TCN was inversely correlated with segment size (R= -0.34, p=2.2e-16). Triple-negative breast cancer (13.3%), hepatocellular carcinoma (8.7%), neuroendocrine tumors of the rectum (8.3%), high-grade serous ovarian cancer (7.6%) and large cell neuroendocrine carcinomas of the lung (6.7%) had the highest prevalence of MYC amplification without concurrent potentially driver oncogene mutations, amplifications or fusions. CoMe analysis demonstrated mutual exclusivity of MYC gene amplification regardless of segment size with most major tumor-type specific oncogenic events. All MYC amplifications were mutually exclusive with MET splice variants and mutations (q< 0.001) and KRAS mutations (q< 0.001) in lung adenocarcinoma (LUAD), FGFR3 mutations (q< 0.001) and fusions (q< 0.001) in bladder cancer, PTEN mutations (q< 0.001) in hormone receptor-positive breast cancer, KRAS mutations (q< 0.001) in colorectal cancer and CDK4 amplifications (q< 0.001) in sarcomas. Broad MYC amplifications were mutually exclusive with RET, ROS1 and ALK fusions in LUAD, while focal ones co-occurred with RET (q=0.007) and ALK fusions (q< 0.001) or secondary mutations (q< 0.001). Conclusions: This dataset defines a subgroup of MYC amplified cancers without co-occurring drivers and further subdivided by segment size, providing information that may help better identify patients for trials of novel MYC-directed therapies.