Abstract 3572: Genomic profiling of subcutaneous patient derived xenograft models of solid childhood cancer

Background: Cancer causes significant mortality and morbidity in children. Current therapies are effective but can cause long-term health problems for patients. Development of new therapies relies on faithful preclinical models. Patient-derived xenografts (PDXs) are an important tool for pre-clinical testing in childhood cancer research. It remains incompletely understood how well genomically PDXs recapitulate primary patient tumors (PTs), particularly in rare cancers. Method: To characterize the fidelity of early passage subcutaneous PDXs derived from pediatric solid tumors, we established 70 early passage PDX models from 16 cancer types. The cohort comprises some very rare cancers such as hepatoblastoma (n=13), germ cell tumor (n=10), osteosarcoma (n=13), and Wilms tumor (n=14). We performed low pass whole genome, exome, and RNA sequencing on these PDXs, their matched PTs and germline samples when materials were available. Result: Overall, we observed low somatic mutation rates in these tumors; however, prior chemotherapy was associated with higher mutation rate. Of the 25 PT/PDX pairs, 20 showed high mutation similarity. The five pairs with low mutation similarity showed evidence of clonal selection. We observed high genomic instability in osteosarcoma. Consistently, more fusions were identified in this cancer type. PTs and PDXs showed high similarity in the copy number pattern, including both broad and focal events. GISTIC analysis identified recurrently amplified or deleted genes including MYC, CCNE1, TP53, PTEN, and BCL2. On the transcriptional level, though PTs and PDXs were generally similar, their expression is more reflective of tissue of origin. We identified fusions that are characteristic of the cancer type such as BCOR-CCND3 in an Ewing like sarcoma. We also identified an NTRK fusion in an osteosarcoma. In summary, we show that PDXs generally recapitulate PTs in mutations, copy number changes, and expression. The dataset represents a valuable resource for future preclinical and mechanistic studies. Citation Format: Funan He, Abhik M. Bandyopadhyay, Laura Klesse, Anna Rogojina, Erin Butler, Taylor Hartshorne, Trevor Holland, Luz Perez Prado, Anne-Marie Langevan, Allison C. Grimes, Chatchawin Assanasen, Zhao Lai, Yi Zou, Dias Kurmashev, Lin Xu, Yang Xie, Yidong Chen, Xiaojing Wang, Gail E. Tomlinson, Stephen X. Skapek, Raushan T. Kurmasheva, Peter J. Houghton, Siyuan Zheng. Genomic profiling of subcutaneous patient derived xenograft models of s