Background: Molecular profiling from the Australian Zero childhood cancer precision medicine program (ZERO) showed that approximately 30% of all children and young adults with cancer lack actionable targets, and just 31% experienced objective clinical benefit from molecular profiling-guided therapy (1). Key challenges in personalised medicine for childhood cancers include understanding the connection between tumour molecular features and sensitivity to targeted therapies, and resistance to single targeted drugs.
Objective: We aimed to identify alternative personalised treatments and better understand the link between tumour molecular characteristics and sensitivity to drugs.
Methods: Molecularly profiled biopsies from ZERO were used for the development of model systems for ex vivo high-throughput screening (HTS) of single drugs and drug combinations using a paediatric cancer customised drug library. Integration of matching tumour molecular and drug response profiling datasets was performed to identify novel potential biomarkers for targeted therapies. Drug-drug correlations were studied for their potential to identify drug combinations.
Results: Single drug testing was successfully undertaken for 154 tumour samples from 151 ZERO patients. Drug efficacy profiles were independent of the type of model system (i.e., freshly dissociated cells or cells expanded in vitro or in vivo). Notably, >80% of the screens on expanded tumour cells could be completed while the patient is still alive. Known associations between drivers and their targeting inhibitors (e.g., BRAFV600E/dabrafenib) were confirmed and using drug-drug correlations could predict effective combinations (i.e., alisertib plus chemotherapy). Integrative data analysis identified MYCN activation as potential biomarker for adavosertib (WEE1 inhibitor) efficacy in neuroblastoma and revealed that trametinib (MEK inhibitor) efficacy in aggressive subtypes of brain cancer is potentially associated with PIK3R1 mutations. Drug combination testing was successfully undertaken for 71 tumour samples from 70 patients. Vertical suppression of the MAPK pathway by combined MEK/RAF inhibition resulted in strong synergism with high overall efficacy in clinically relevant concentrations in sarcoma samples with homozygous inactivation of NF1 and in subsets of brain tumour samples.
Conclusion: In vitro drug testing on patient-derived tumour models is a valuable tool to improve the identification of personalised treatment options in paediatric cancer and enhances our understanding of response to targeted therapy.