Abstract 3455: Transcriptional and phenotypic heterogeneity underpinning venetoclax resistance

Abstract Venetoclax (VEN), an inhibitor of anti-apoptotic protein BCL2 is an important therapeutic intervention in acute myeloid leukemia (AML). However, development of resistance to VEN and its combinations is common, which poses a significant clinical challenge to the management of AML. Heterogeneity in resistance can exist not only across different tumors but also within the same tumor. While some mechanisms of resistance have been revealed, others remain elusive. Here, we derive a VEN resistance signature from AML cell-lines in CCLE, and utilize non-negative matrix factorization to decompose the signature and systematically define four transcriptionally distinct clusters of VEN resistant AML patient using the BeatAML cohort, consisting of 186 patients. By integrating gene expression and drug response profiles from these patients, we elucidate distinct expression programs underlying VEN resistance in these clusters and their associated vulnerabilities. Samples in Cluster 1 are characterized by high frequency of DNMT3A and RAS (KRAS and NRAS) mutations, strong monocytic phenotype and high PI3K-mTOR signaling, oxidative phosphorylation (OXPHOS) and fatty acid metabolism, indicating improved metabolic fitness. These samples also show remarkable sensitivity to panobinostat, an HDAC inhibitor, which suppresses mTORC1 signaling and OXPHOS. Cluster 3, in contrast, is characterized by TP53 and SRSF2 mutations, high CTL infiltration and activation of JAK-STAT and MAPK signaling, which regulate pro-survival pathways and proliferation. They also show selective sensitivity to JAK family inhibitors. Cluster 3-like AML cell-lines also show remarkable sensitivity to the JAK inhibitor, ruxolitinib. Cluster 2 samples exhibit high rates of NRAS mutations and suppression of HOX genes. Finally, Cluster 4 samples are transcriptionally similar to VEN sensitive patients, but show higher interferon signaling and higher rates of DNMT3A and IDH1 mutations. We validated these clusters using the TCGA AML (n=173) and a second BeatAML cohort (n=182).We further confirmed the existence and therapeutic value of these clusters by projecting them onto single-cell RNA-seq data obtained from a patient before and after VEN+azacitidine treatment. In pre-treatment cells, both VEN sensitive (41%) and resistance (59%) signatures were present in the patient tumor, however, post-treatment we observed that Cluster 1 (5%) and Cluster 2 (93%) like resistant cells accounted for most of the tumor cells. While Cluster 1 like cells were characterized by activation of metabolic pathways, the more proliferative Cluster 2 like cells dominated the tumor after relapse. Our study systematically identified and characterized mechanistic heterogeneity in venetoclax resistant AMLs and can aid therapeutic decision making and development of effective combination therapies with venetoclax. Citation Format: Vakul Mohanty, Natalia Baran, Yuefan Huang, Shan He, Ramiz Iqbal, CTD2 consortium, Jeffrey Tyner, Gordon Mills, Marina Konopleva, Ken Chen. Transcriptional and phenotypic heterogeneity underpinning venetoclax resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3455.