Abstract PO-037: Machine learning-enabled transomics identifies three therapeutic targets for MYC-driven diffuse large B cell lymphoma

Abstract MYC activation and dysregulation is a powerful oncogenic driver in multiple cancers, including diffuse large B cell lymphoma (DLBCL). There is a known correlation between MYC expression and poor prognosis. While the 5-year overall survival for DLBCL patients on first-line therapy, R-CHOP, is approximately 75%, the 5-year overall survival drops to a staggering 30% for patients co-expressing MYC and BCL2 (double expressor lymphoma). However, it is challenging to target MYC directly. Therefore, we identified a novel approach to circumvent targeting MYC directly by leveraging Pepper’s proprietary machine learning (ML) transomic analysis platform, COMPASS, to identify novel targets associated with high MYC activity that, when inhibited, are predicted to recapitulate the effect of MYC inactivation. Specifically, COMPASS unlocks functional drivers of disease to identify novel drug targets. To study the role of MYC in lymphoma biology, we utilize a MYC-conditional T cell acute lymphoblastic leukemia/lymphoma (T-ALL) cell line (4188) with tunable MYC expression. We collect genomic, transcriptomic, proteomic, and phosphoproteomic data from samples with MYC transgene expression “on” and “off” and compare these biological cell states via COMPASS to identify novel targets that reproduce MYC inactivation. This provides new hope for patients suffering from MYC driven DLBCL. The MYC-conditional cell line allows the regulation of MYC expression via the tetracycline regulatory (Tet-Off) system. Four omic datasets were collected from each sample: genomics (next-generation sequencing, NGS), transcriptomics (NGS), proteomics (mass spectrometry), and phosphoproteomics (mass spectrometry). We employed the COMPASS target prioritization algorithm to identify and rank novel targets that mimic “turning MYC off”. Targets were then filtered to select those with an available pharmacological tool compound (PTC). The PTCs were used to evaluate the targets in the SU-DHL-06 xenograft model of DLBCL. The PTCs inhibited tumor growth from 8-91%. A total of 20 targets were tested of which 13/20 (65%) were validated as evidenced by significant inhibition of tumor growth (p0. 05, two-way ANOVA). All targets were kinases or related to kinase activity and pathways. Three of the PTCs resulted in maximum tumor growth inhibition of 86%, 89% and 91% at end of treatment, resulting in stasis of tumor growth and increased survival compared with control vehicle-treated mice. We have previously shown how the COMPASS transomics analysis approach can identify novel drug targets for drug-resistant EGFRm NSCLC and for MYC-driven hepatocellular carcinoma (HCC). The success rate of target validation across several indications including EGFRm NSCLC, HCC and lymphoma is 73%. The data presented here on MYC-driven DLBCL further validates this approach. Further validation studies will investigate novel targets using gene silencing, as PTCs were unavailable for many of the novel high-ranked targets identified in this study. Citation Format: Simon P Fricker, Christopher J Nicholson, Samuel J Roth, Arudhir Singh, Caitlin Brown, Jon Hu, Petronela Buiga, Vishnu P Kanakaveti, Anja Deutzmann, Dean Felsher, Samantha D Strasser. Machine learning-enabled transomics identifies three therapeutic targets for MYC-driven diffuse large B cell lymphoma abstract. In: Proceedings of the Fourth AACR International Meeting on Advances in Malignant Lymphoma: Maximizing the Basic-Translational Interface for Clinical Application; 2024 Jun 19-22; Philadelphia, PA. Philadelphia (PA): AACR; Blood Cancer Discov 2024;5 (3Suppl): Abstract nr PO-037.