NEW YORK – The Children's Hospital of Philadelphia and the National Cancer Institute's Frederick National Laboratory have launched a new research tool to make good on the promise of the Research to Accelerate Cures and Equity (RACE) for Children Act, which requires companies developing cancer drugs for adults to test them on children when there is a shared molecular target.
The tool, called the Molecular Targets Platform, promises to accelerate precision therapies for children by enabling researchers to more easily identify what those shared targets are and answer a broad range of questions about the significance of biomarkers in childhood cancer and how to best target them in clinical trials.
Signed into law in 2017 as part of the FDA Reauthorization Act, the RACE for Children Act has prompted drugmakers to dramatically increase the number of new drug applications submitted to the FDA with pediatric study plans. Progress in childhood cancer research had previously been hindered by a reluctance among drugmakers to pursue therapeutic development in the comparatively smaller pediatric populations affected. The new law has provided motivation for the companies to carry out these studies.
However, until now the only resource provided to support that effort has been a list published by the FDA of molecular targets in adult cancer that are considered relevant to pediatric cancer.
"That was just a spreadsheet," said John Maris, a professor of pediatric medicine in the division of oncology at the University of Pennsylvania and co-principal investigator on the project. "You can look at it and sort it, but what we wanted to be able to do is integrate that with childhood cancer data, next-generation sequencing data, and [gene] expression data, and have everything together and linkable."
The goal was to develop a portal that would frame the data with a user-friendly interface to allow pediatricians, industry, cancer advocacy groups, and data scientists to ask questions relevant to cancer drug development in children.
Deanne Taylor, director of bioinformatics at CHOP, assembled a team of bioinformaticians and data scientists from there and the NCI to take in data from six major sources, organize it into a common format, and build a computational infrastructure around it. In addition to the FDA's molecular target list, the Molecular Target Platform also draws data from Therapeutically Applicable Research to Generate Effective treatments (TARGETS), Gabriella Miller Kids First Neuroblastoma, Open Pediatric Brain Tumor Atlas, Oncology Knowledge Base (OncoKB) Cancer Gene List, and the Genotype-Tissue Expression (GTex) project. The team plans to bring in data from The Cancer Genome Atlas in the future.
The processing for these diverse data sources includes primary sequence analysis through Cavatica, a collaboration between CHOP and Seven Bridges Genomics to gather and share genomic, clinical, and other biomedical data in the area of pediatric cancer and rare disease. Secondary analysis is carried out in OpenPedCan, an open-source analysis project hosted by CHOP that harmonizes multiple sources of cancer data, meaning that it converts the data into a common format and processes it through a common pipeline so that researchers can compare data of the same type — such as sequencing and gene expression data —originating from different sources.
As a part of its functionality, the Molecular Targets Platform is designed to answer questions from researchers. Taylor said an example of such a query might be to ask how many other tissues might be affected if a particular immunotherapy target were used for a cancer therapy. That query would draw in part on data from the GTex project, which is a collection of samples from 54 non-diseased tissue sites in nearly 1,000 individuals. "You can compare the gene expression of your particular gene with all the normal tissues that are currently in GTex," Taylor said. For the time being, those are adult tissues, but Taylor said the GTex project is working to add children's tissue samples and that that data would be a part of the platform in the future.
Researchers could also compare expression of a gene in one cancer with all other cancers currently in the database. "As we get more diversity and more details, we'll be expanding that list," Taylor said.
Other queries supported by the platform include those related to copy number variation, amplification, fusions, single-nucleotide variations, and deletions of genes.
"We purposely decided to develop the platform in an iterative fashion using real-world use cases," Maris said. "Physicians on the team would write out different scenarios that we've either actually directly experienced in our own research or could imagine."
Those scenarios explored how relevant targets might show up in childhood cancers. "Some drugs are completely dependent upon a mutation in a certain gene. Can we define how many childhood cancers have this mutation and what subsets have mutations?" said Maris. "[These are] very practical things you need to know in order to develop a clinical trial."
As an example, Maris pointed out that approximately 7 percent of adult leukemia patients have rearrangements of KMT2A that are treatable with menin inhibitors, but these are the most common drivers of infant leukemia, representing 70 percent to 80 percent of cases. Maris said that the Molecular Targets Platform could help researchers identify other adult biomarkers that are important in children.
Another key aspect of the Molecular Targets Platform was that the CHOP and NCI teams adapted the Open Targets Platform to create it. Taylor said that because many pharmaceutical companies use Open Targets internally, it would be a good framework to introduce the new resource to pharmaceutical companies in a way that would be useful to them. "We adapted Open Targets and then Frederick National Laboratory stood up prototypes of the platform with changes to add in some pediatric cancer data," Taylor said. "The first round of what we built is currently out on the web. It's the first time all of this data can be viewed together in the public domain."
Taylor noted that while the platform has launched and is ready to use, the teams working on it will be adding more complex data and functionality in the future, including data being processed at other institutions representing about 2,000 total cases, which will be submitted to NCI. "That's all going to be coming in the next year or two," said Taylor. "That will increase the diversity of the available datasets, and the types of cancers."
"There are a lot of different databases out there with pediatric cancer data," Maris added. "But this is the first that integrates the FDA's relevant molecular target list and is not just a data repository, but rather a tool to make decisions about drug development."
Although the Molecular Targets Platform is dedicated to children's cancers, it also contains adult data that was brought in with the Open Targets Platform. That opens up possibilities for cross-referencing searches on childhood cancer targets with other diseases, such as diabetes, to investigate interactions of various therapeutics with childhood cancer targets.
Maris emphasized that a resource like the Molecular Targets Platform will become even more valuable as combination therapies are further explored, especially with legislation introduced in the US House of Representatives in March, called the Give Kids a Chance Act, which would ensure children are able to participate not only in single-drug trials but combination therapy trials, as well.