Patient‐derived mouse xenografts from pediatric liver cancer predict tumor recurrence and advise clinical management

Identification of new treatments for relapsing pediatric cancer is an unmet clinical need and a societal challenge. Liver cancer occurrence in infancy, 1.5 for million children per year, falls far below the threshold of interest for dedicated drug development programs, and this disease is so rare that it is very difficult to gather enough children into a phase II clinical trial. Here, we present the establishment of an unprecedented preclinical platform of 24 pediatric liver cancer patient‐derived xenografts (PLC‐PDXs) from 20 hepatoblastomas (HBs), 1 transitional liver cell tumor (TCLT), 1 hepatocellular carcinoma, and 2 malignant rhabdoid tumors. Cytogenetic array and mutational analysis of the parental tumors and the corresponding PLC‐PDXs show high conservation of the molecular features of the parental tumors. The histology of PLC‐PDXs is strikingly similar to that observed in primary tumors and recapitulates the heterogeneity of recurrent disease observed in the clinic. Tumor growth in the mouse is strongly associated with elevated circulating alpha‐fetoprotein (AFP), low rate of necrosis/fibrosis after treatment, and gain of chromosome 20, all indicators of resistance to chemotherapy and poor outcome. Accordingly, the ability of a tumor to generate PLC‐PDX is predictive of poor prognosis. Exposure of PLC‐PDXs to standards of care or therapeutic options already in use for other pediatric malignancies revealed unique response profiles in these models. Among these, the irinotecan/temozolomide combination induced strong tumor regression in the TCLT and in a model derived from an AFP‐negative relapsing HB. Conclusion: These results provide evidence that PLC‐PDX preclinical platform can strongly contribute to accelerate the identification and diversification of anticancer treatment for aggressive subtypes of pediatric liver cancer. (Hepatology 2016;64:1121‐1135)