Abstract Patient-derived organoid models provide a highly amenable experimental platform to study human tumor biology. Over the past seven years, our group has generated a living biobank of 82 organoid lines established from urothelial carcinomas. These organoids have been generated from all stages, are capable of unlimited expansion in vitro, and recapitulate histopathological and molecular features of their corresponding primary tumors. In addition, we have characterized the genomic, transcriptomic, and chromatin accessibility profiles of these organoid lines. Using these lines, we have identified a novel form of lineage plasticity in which a subset of organoids derived from luminal NMIBC tumors can acquire basal features in culture, but reverse back to a luminal phenotype following orthotopic xenografting in immunodeficient mice. We have focused our studies on this luminal-to-basal (L-B) plasticity since the acquisition of basal phenotypes is linked to secondary MIBC. In ongoing studies, we have examined the genomic correlates of L-B plasticity and found that loss-of-function (LOF) mutations in the epigenetic regulator KMT2D are enriched in plastic lines derived from NMIBC and in stable basal lines from basal/squamous MIBC. We also found that tumors with squamous histology in the TCGA cohort are enriched for KMT2D LOF mutations. KMT2D encodes a histone methyltransferase that adds H3K4me1 active histone marks to enhancer regions and is frequently mutated in urothelial cancers. To investigate its functional role, we performed CRISPR knock-out of KMT2D in a stable luminal organoid line and found that isogenic knock-out organoids acquired basal phenotypes. Using a barcoded lineage-tracing approach known as CellTag, we have found that acquisition of basal phenotypes occurs in a clonal fashion at the single-cell level. In parallel, we have performed a small molecule screen to identify compounds that can reverse L-B plasticity, using 17 compounds that target a range of epigenetic regulators. We found that GSK-LSD1, an inhibitor of KDM1A, can restore luminal phenotypes to plastic organoid lines in culture and increase the genomic distribution of the H3K4me1 mark, as detected by CUT&Tag. Interestingly, KDM1A is a histone demethylase that removes H3K4me1/2 marks, and thus can antagonize KMT2D. Using CellTag, we found that reversal of the basal phenotype occurs in a non-clonal fashion and represents a cell state transition at the level of single cells. In further studies, we have shown that KDM1A inhibitor treatment of fresh patient samples can prevent the L-B transition. We have also demonstrated in an organoid outgrowth assay that KDM1A inhibition can prevent invasive behavior in vitro. Moreover, an orally available KDM1A inhibitor, ORY-1001, can also reverse phenotypic plasticity, and we are currently testing its activity in xenografts to determine whether it can reverse invasive phenotypes. Overall, these results demonstrate the utility of patient-derived organoid models to investigate novel features of bladder cancer and develop therapeutic approaches. Citation Format: John R. Christin, Alana Nguyen, Talal Syed, Kwanghee Kim, Clementine Le Coz, Caroline J. Laplaca, Luis A. Pina, Reuben Akabas, Hanina Hibshoosh, Andrew T. Lenis, Guarionex J. Decastro, Christopher B. Anderson, James M. McKiernan, Chao Lu, Hikmat A. Al-Ahmadie, David B. Solit, Michael M. Shen. A living biobank of patient-derived organoids reveals a role for KMT2D in maintaining luminal identity in urothelial carcinoma [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2024 May 17-20; Charlotte, NC. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(10_Suppl):Abstract nr A028.