Background: Genomic profiling cannot solely predict the complexity of how tumor cells behave in their in vivo microenvironment and their susceptibility to therapies. The aim of the study was to establish a functional drug prediction model utilizing patient-derived GBM tumor samples for in vitro testing of drug efficacy followed by in vivo validation to overcome the disadvantages of a strict pharmacogenomics approach.
Methods: High-throughput in vitro pharmacologic testing of patient-derived GBM tumors cultured as 3D organoids offered a cost-effective, clinically and phenotypically relevant model, inclusive of tumor plasticity and stroma. RNAseq analysis supplemented this 128-compound screening to predict more efficacious and patient-specific drug combinations with additional tumor stemness evaluated using flow cytometry. In vivo PDX mouse models rapidly validated (50 days) and determined mutational influence alongside of drug efficacy. We present a representative GBM case of three tumors resected at initial presentation, at first recurrence without any treatment, and at a second recurrence following radiation and chemotherapy, all from the same patient.
Results: Molecular and in vitro screening helped identify effective drug targets against several pathways as well as synergistic drug combinations of cobimetinib and vemurafenib for this patient, supported in part by in vivo tumor growth assessment. Each tumor iteration showed significantly varying stemness and drug resistance.
Conclusions: Our integrative model utilizing molecular, in vitro, and in vivo approaches provides direct evidence of a patient's tumor response drifting with treatment and time, as demonstrated by dynamic changes in their tumor profile, which may affect how one would address that drift pharmacologically.
Keywords: GBM; cancer; high-throughput screening (HTS); organoid; phenotypic.