Sensitivity profiling reveals consistent drug responses across preclinical neuroblastoma models

Abstract Despite intensive treatment, overall survival for high-risk and relapse neuroblastoma patients remains below 50%. Even though comprehensive molecular profiling enables treatment stratification, druggable alterations have been identified for only a subset of patients. In vitro drug screening offers a complementary approach. Here, we compare the translational potential of three preclinical drug screening methods: ex vivo short-term, in vitro patient-derived organoid and in vivo patient-derived xenograft (PDX) drug testing. In total, 55 screens were performed from 38 neuroblastoma samples and five pediatric non-malignant samples, testing 77-224 drugs per screen. Ex vivo short-term drug screens achieved higher success rates than organoid screens (65% versus 23%) and shorter turnaround times (14 days versus 3-12 months). Matched samples showed consistent drug sensitivities across sample origin (patient versus PDX-derived; mean r = 0.84) and method ( ex vivo short-term versus organoids; mean r = 0.87), demonstrating that ex vivo short-term screens recapitulate drug sensitivities found in long-term organoid models. In parallel, as part of the ITCC-P4 consortium, ten compounds were tested in vivo in eight PDX models, with samples matching the ex vivo screens. For seven out of ten clinically available compounds, ex vivo drug responses were comparable with in vivo responses in matched PDX models. These results demonstrate that, while organoids and PDX models remain essential for drug discovery, ex vivo short-term drug screening provides a rapid alternative for functional precision oncology in neuroblastoma.