Background: Personalised medicine approaches may improve the poor survival rate of high-risk neuroblastoma, however the relatively low mutation rate and paucity of recurrent actionable mutations render therapy selection based on sequencing techniques challenging. We are currently using patient-derived xenograft (PDX) models to investigate whether unbiased, ex vivo drug screening is feasible as an alternative strategy.
Methods: PDX models of high-risk neuroblastoma were obtained from Children’s Oncology Group and passaged in NSG mice. Tumours were harvested from five models, immediately processed using a semi-automated tissue dissociator and plated in 384-well tissue culture plates. Ex vivo screening was conducted using 136 established adult and paediatric anti-cancer drugs with cells treated at five doses (1–10,000 nM) to generate dose-response using an end-point viability assay.
Results: PDX models proved amenable to ex vivo manipulation, short-term culture and drug screening using a viability assay. For each of the five models, we were able to identify at least one drug for which that model was greater than 10-fold more responsive ex vivo compared to the average response of the other models based on IC50 values. In some cases this window was greater than 100-fold. Importantly, these included drugs approved for paediatric use, such as cytarabine, tretinoin (ATRA), gemcitabine and imatinib.
Conclusions: Drugs identified as effective in ex vivo screening will now need to be confirmed using these PDX models in vivo. Our results to date suggest that ex vivo drug screening can identify unexpected drug sensitivities that may aid in identifying therapy options for patients with high-risk neuroblastoma.