Molecularly guided precision medicine has revolutionised childhood cancer therapy and significantly improved patient outcomes. However, several patients are still missing out on receiving personalised treatment recommendations. Preclinical in vivo and in vitro drug testing using patient-derived models may identify the treatment options for such patients and provide evidence to support molecular-based therapeutics recommendations. However, the feasibility of using patient-derived xenograft (PDX) drug efficacy data to provide treatment recommendations in a clinically relevant timeframe remains unclear.
Herein, we report on the establishment and clinical utility of an extensive paediatric PDX panel established within the ZERO Childhood Cancer Precision Medicine Program (ZERO). PDX establishment was attempted using 342 patient specimens from 328 high-risk paediatric patients. A panel of 203 unique PDX models has been established and is available for research through the Paediatric Preclinical Biobank, including leukaemia, sarcoma, neuroblastoma, brain tumour and other rare paediatric cancer models. Ten patients have matched PDX models established at various disease stages. The engraftment success rates vary across tumour types (sarcoma 68%, leukaemia 77%, brain tumour 28% (biopsy 54% and culture 36%), neuroblastoma 56%, other tumours 46%), and engraftment success correlates with patient prognosis.
Drug efficacy studies were conducted in 38 PDX models using drugs and drug combinations selected based on molecular profiling, in vitro high-throughput drug screening and empirical consideration and assessed using stringent objective response criteria developed from clinical trials. Effective therapeutic options were identified in 21 models (55%) when patients remained alive. From 21 evaluable patients who received drugs matching the PDX-tested drugs, 86% showed concordant responses.
In summary, we have established a new and comprehensive paediatric PDX panel to facilitate paediatric cancer research worldwide. We also demonstrate the utility of such models within a precision medicine study, highlighting how additional therapeutic recommendations can be provided within a clinically relevant timeframe.