Brain tumors are responsible for the largest number of cancer related deaths in children. Recent advances in genomic and epigenomic research allow for better characterization and classification of these brain tumors, which could lead to more effective targeted treatments. Pediatric brain tumors are classified into a small number of broad categories based on histology of the tissue. However, histology of brain tumors does not allow doctors to effectively compare driver mutations, response to therapeutic treatments, or outcomes. In order to test different therapies for each type of tumor, physicians need preclinical models that accurately represent each subgroup. Patient-derived xenograft (PDX) models are already established in certain types of tumors. These can be tested to measure the response of targeted therapies allowing researchers to prioritize treatments. Because pediatric brain tumors are rarer, very few PDX brain tumor models have been established. Many of these models have not been adequately characterized at the molecular level and usually do not grow when transplanted into a new model in the original location, known as orthotopic transplant. The laboratory of Dr. James Olson (Clinical Research Division) at Fred Hutch, along with collaborators in Germany, generated 30 new preclinical models consisting of 14 different subgroups of brain tumors. In a recent article published in Nature Medicine, Dr. Olson and colleagues investigated the molecular characteristics of these models and tested their response to therapeutics. They also made these models as well as seven new brain tumor cell lines readily available to other researchers through an online biobank.
Dr. Olson’s team and collaborators created patient-derived orthotopic xenograft (PDOX) models of pediatric brain tumors by transplanting the tumor cells into mice lacking T, B, and NK cells, known as NOD scid gamma (NSG) mice. These implanted tumors resembled primary tumors histologically, and in most cases, their molecular classification aligned with histological classification. The researchers then sequenced the tumor genomes of the PDOX models to identify oncogenes and tumor suppressor genes which are mutated. They found that the mutations were typical of their respective diseases. “There are over 30 categories of pediatric brain tumors and each is rare,” Dr. Olson said. “Now we have models of almost all of them (we had only one, in the world, when I started at Fred Hutch!). We really need to prioritize candidate therapies in models that represent patients with high fidelity to avoid putting kids on trials that will induce toxicity but no benefit.”
The molecular data of the PDOX models allowed the researchers to predict which therapeutics they might be vulnerable to. One of the mutations they noticed that varied among models was in the gene for epidermal growth factor receptor (EGFR), a receptor tyrosine kinase involved in cell division. Overexpression of EGFR has been shown to cause uncontrolled proliferation more common tumor types. In order to test whether therapeutics might be able to target brain tumors expressing EGFR, Olson and colleagues looked at two types of PDOX models, which both looked similar histologically and were classified as high-grade glioma. However, one contained an EGFR amplification, while the other had low EGFR expression. When the scientists administered the EGFR inhibitor erlotinib, mice with the EGFR-amplified tumors survived significantly longer compared to the control treated group. On the other hand, mice bearing tumors with low EGFR expression were better off receiving control treatment. The scientists researched several other models similarly, demonstrating the need for PDOX models representing different subgroups. “Rather than testing a candidate therapeutic molecule on 10-20 replicates of one or two models, we can now test candidates on dozens of models,” Dr. Olson explained. “This is more like human clinical trials and better captures the inter-patient heterogeneity.”
The scientists made the PDOX models readily available in an online biobank (http://www.btrl.org). In addition, they made the data on the models available in more detail in the PDX Explorer (http://www.r2platform.com/pdxexplorer). “Because this work was supported by our patient families through the Run of Hope and other fundraisers, we are able to make these models available world-wide to accelerate research for kids with brain tumors. I'm excited to see what other labs and our team come up with in the coming years.”
Funding for this research was provided by National Institutes of Health, the Seattle Run of Hope, the Seattle Children’s Brain Tumor Research Endowment, the Dutch Cancer Foundations KWF and KIKA, Deutsche Krebshilfe, BMBF, and the Helmholtz International Graduate School for Cancer Research
Fred Hutch/UW Cancer Consortium authors Sarah Leary and James Olson contributed to this work
Brabetz S, Leary SES, Gröbner SN, Nakamoto MW, Şeker-Cin H, Girard EJ, Cole B, Strand AD, Bloom KL, Hovestadt V, Mack NL, Pakiam F, Schwalm B, Korshunov A, Balasubramanian GP, Northcott PA, Pedro KD, Dey J, Hansen S, Ditzler S, Lichter P, Chavez L, Jones DTW, Koster J, Pfister SM, Kool M, Olson JM. 2018. A biobank of patient-derived pediatric brain tumor models. Nature Medicine. 24(11):1752-61. doi: 10.1038/s41591-018-0207-3.