Abstract Heterogeneity is ubiquitous in cancer and treatment responses among patient cohorts are highly variable, leading to drug resistance and disease relapse. Yet, preclinical drug screening is usually performed in cell lines or patient cells, one at a time, without consideration of cancer heterogeneity. Consequentially, positive outcomes obtained in the preclinical stage do not translate well into subsequent clinical successes. Castration-resistant prostate cancer (CRPC) is a highly lethal form of prostate cancer (PC) exhibiting a high degree of genetic and phenotypic heterogeneity with very few therapeutic options. Therefore, we created mixed-cell models depicting heterogeneity in various clinical contexts, where multiple PC cell lines representative of distinct CRPC tumor variants were labeled and combined in co-cultures to mimic the complex molecular landscape of CRPC patient cohorts. Specifically, transcriptomic profiles of PC cell lines were clustered with patient tumor specimens to identify cell lines that are genetically similar to distinct patient clusters representing CRPC tumor subtypes. Since heterogeneity has been linked to the clinical efficacy of combinatorial therapies, we applied these models to evaluate drug combinations. A combination of docetaxel and dasatinib that was previously assessed in a PC trial, was tested in the mixed-cell model and experimental findings concurred with the published clinical results, demonstrating the utility and accuracy of the model. Additionally, a computational model, IDACombo, was employed to nominate novel potentially efficacious drug combinations using monotherapy response data of cancer cell lines, a number of which were successfully validated in the mixed-cell model. More importantly, we demonstrated that while individual cell line screens may result in erroneous conclusions, combination efficacy can be accurately measured in a heterogeneous cell pool emulating a diverse patient cohort. We further explored phenotypic heterogeneity in advanced disease states, by mixing parent and standard-of-care (SOC)-resistant CRPC clonal variants. Drugs with selective potency in the resistant (vs. the parent) clone were computational prioritized and when combined with the SOC and tested in the mixed-cell models, outperformed either monotherapy. Additionally, we investigated the effects of altering the relative proportion of the constituent clones to mimic different stages of acquired therapy resistance. Furthermore, we generated mixed-cell xenograft models by subcutaneously injecting the labeled cell-line mixtures in mice and characterized the harvested tumors to detect and quantify the individual subpopulations. Development of these preclinical models holds enormous promise not only for advancing novel therapeutic strategies for cancer, but also for providing insights into the mechanisms of resistance and recurrence. Citation Format: Sampreeti Jena, Daniel Kim, Adam M. Lee, Weijie Zhang, Scott Dehm, R. Stephanie Huang. Development of novel mixed-cell models to capture heterogeneity in castration-resistant prostate cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6625.