Research Programs


Children with high-grade glioma, diffuse intrinsic pontine glioma, rhabdoid brain tumors, and certain other types of brain tumors experience a low likelihood of survival even with maximal therapy, including surgery, intensive chemotherapy, and radiation therapy. As an alternative to the often-toxic standard therapies, scientists have for years worked diligently to find ways to help the body’s own immune system recognize cancer cells and kill them. Recognizing that cancer cells cloak themselves in molecules that instruct the immune system to protect the cancer cells and then developing drugs to block these protective mechanisms were big steps in the right direction. These drugs, called PD-1 inhibitors, are effective in some types of cancer; however, they are not effective in pediatric brain tumors when used alone. Rather, these drugs may be more effective at killing pediatric tumor cells if they are combined with other therapies to help the immune system identify cancer cells. In addition, two other classes of drugs have been shown to cause molecular changes in cancer cells by mimicking what occurs when human cells are infected with viruses. The immune system is triggered to kill the cells following treatment with these drugs; however, sometimes the protective surface molecules described above thwart treatment.

Our approach is to combine the drugs that mimic viral infection with a drug that blocks the PD-1 protective response to see whether the combination works in mice that are growing aggressive malignant pediatric brain tumors, as well as in children who are newly diagnosed with poor-prognosis brain tumors. Selection of the most active compounds will lead directly to future human trials in the 15-member Collaborative Network of Neuro-oncology Clinical Trials (CONNECT) and will rapidly determine if the combination therapy is safe and feasible to deliver in a multi-site clinical trial.

Functional Genomics

Brain tumors are the leading cause of cancer-related death in children. Genomic studies have provided insights into molecular subgroups and oncogenic drivers of pediatric brain tumors that may lead to novel therapeutic strategies. To evaluate new treatments, we need better preclinical models that adequately reflect the biological heterogeneity. Through the Children’s Oncology Group ACNS02B3 study, we have generated and comprehensively characterized 30 patient-derived orthotopic xenograft (PDOX) models and seven cell lines representing fourteen molecular subgroups of pediatric brain tumors. We found the PDOX models to be representative of the human tumors they were derived from in terms of histology, immunohistochemistry, gene expression, DNA methylation, copy-number, and mutational profiles, with in vivo drug sensitivity of targeted therapeutics associated with distinct molecular tumor subgroups and specific genetic alterations. These models and their molecular characterization provide an unprecedented resource for the cancer community to study key oncogenic drivers and evaluate novel treatment strategies.

Diffuse Midline Gliomas

We aim to better understand the role of histone mutations in DIPG, as well as to identify translationally impactful new epigenetic and immunologic treatment strategies to improve the lives of children with DIPG and other diffuse midline gliomas. Current models of DIPG are most often created from autopsy tissue that does not exactly replicate newly arisen DIPG. At Seattle Children’s Hospital, children with DIPG are offered biopsies to genetically characterize their tumor. Additional samples are also used in the Olson Lab at Fred Hutch to create treatment-naïve DIPG models. We are then able to use a viral system that makes the tumor cells glow and to inject DIPG biopsy samples into the brains of mice to create animal models that truly mimic newly diagnosed DIPG in patients. Our first treatment-naïve animal model, PBT-09FHTC, is currently under study.

By screening next-generation epigenetic drugs against DIPG cells and in our treatment-naïve DIPG animal models, we have also identified several new drugs that may be beneficial against DIPG. We work to understand their most effective doses and how they kill DIPG cells to determine the most effective and safest clinical trials possible. Our studies incorporate the use of radiation. Though radiation is commonly used for patients with DIPG, it is infrequently incorporated in lab studies. Understanding he effect of radiation and its limitations will be critical to our understanding of how to build better treatment plans for children with DIPG.

Brain Tumor Resource Library

In the late 1990s, when Dr. Olson’s lab was just beginning, the Washington Women’s Foundation provided a grant that enabled our laboratory to serve as a National Brain Tumor Resource Laboratory (BTRL). Our BTRL generated resources from patients’ surgical samples and made panels that could be shared freely with investigators around the world. We established practices that reduced barriers to collaboration and promoted advanced molecular studies. These studies led to the incredible understanding of pediatric brain tumor biology in the subsequent decade. With continued support from Seattle Children’s Hospital Guilds, we expanded services to the generation and sharing of mouse medulloblastoma models. Our models are now used in more than 50 labs worldwide and provide the scientific basis for four national clinical trials.

In 2009, we launched a new program focused on generating patient-specific cell lines for drug screening and mouse models for drug prioritization. We challenged our team to imagine the day when a surgical sample from a child could generate cell lines and mice would rapidly generate data that could then guide clinical decisions for that child. We reasoned that the data generated with these resources could shape the next generation of national clinical trials. Since 2009, the Olson Lab has generated more than 30 new mouse models that carry human brain tumors derived from our patients’ surgical specimens. The same specimens are used to generate patient-derived cell lines that are useful for drug screening and prioritization. Because of the generosity of donors, we can now share these resources with no strings attached to any laboratory around the world that wishes to study pediatric brain tumors.

Learn more about BTRL


High-risk neuroblastoma is the most common cancer in babies and the third most common pediatric cancer after leukemia and brain cancer. It affects approximately 1 in every 7,000 children, and despite recent advances, it remains fatal for most affected children. Dinutuximab, the most recently approved treatment, has improved outcomes somewhat, but its use is limited by the extreme, opiate-resistant pain that results from the expression of the disialoganglioside-2 (GD2) target antigen in healthy nerve tissue. Recently, Glypican-2 (GPC2) has emerged as a promising target antigen, with high expression in neuroblastoma and what appears to be low expression in most healthy tissue. However, expression of any single antigen will almost certainly not be restricted to cancer, and any therapy that targets a single antigen, including CAR-T cells, antibodies, or antibody-drug conjugates, will be inherently limited by target expression in healthy tissue.

Our lab proposes a new approach, one that allows us to selectively target cancers that express two antigens while ignoring non-cancerous tissues that express only one. We accomplish this through the simultaneous administration of two bispecific molecules, each of which binds to a different cancer antigen on one end and to a T cell receptor or co-receptor on the other. Importantly, each of these two bispecific molecules is selected for very low activity as a single agent, and the T cells are only activated when both target antigens are expressed. Because GPC2 and GD2 are two well-defined target antigens, neuroblastoma represents an ideal test case for this approach. Using two bispecific molecules simultaneously should allow for a uniquely potent and selective targeting of neuroblastoma, and success would demonstrate a generalizable approach to the treatment of almost any cancer, even those that fail to express a cancer-specific antigen.