Coupling a genome-wide association study with a novel computational method, researchers at the Children's Hospital of Philadelphia identified four genes — DUSP12, DDX4, IL31RA, and HSD17B12 — that harbor variants linked to the development of neuroblastoma, a common childhood cancer. The team, led by CHOP's John Maris, concentrated on the low-risk form of the disease, which has not been as extensively studied as the high-risk form.
"We used a straightforward GWAS approach using relatively standard methodologies. However, what we did that was different here were two things: we used our relatively robust phenotyping information to do subset analyses, to ask if the association signals were more enriched in patients who lived or died or patients who had localized disease versus more advanced disease," Maris says. "We [also] developed a methodology — knowing that we're doing this for translational purposes — to do a gene-centric GWAS and consider genes as units, and thereby reduce the multiple-testing problem." The way the methodology works is to consider each gene as a unit that has possible permutations — SNPs — and to test for associations at the gene level.
The team, which published its study in PLoS Genetics in March, used a SNP chip with 600,000 variants, but instead of considering each variant as a variable, the researchers considered each gene as one. Because genes are different sizes, Maris says, there could be anywhere from a handful of SNPs to hundreds of them at any given gene. The more SNPs there are, the greater the chance that some of them will show up as false positives and send the researchers looking in the wrong directions. "You need to control for the number of SNPs, and that's what we did," Maris says.
The genes the researchers found in this study — which had never been linked to either cancer or neuroblastoma before — could lead to therapeutic targets for treating neuroblastoma, Maris says. The pathways those genes are in could be targeted by drugs for a higher likelihood of survival and recovery for the patient. "For example, one of the major genes discovered in this effort, DUSP12, we were able to very quickly learn is an important phosphatase that regulates signaling through the MAP-kinase and RAS pathways," Maris says. "We've wondered why, in neuroblastomas, mutations in those pathways — which are very common in some other cancers — are largely absent. So, it may be that over-expression and activation of DUSP12 leads to activation of these pathways, both in the developing nervous system that leads to a predisposition of the cancer and then — once the cancer forms — this is selected for and further amplifies activation of these pathways."
A few years ago, researchers did not know very much about the genes that affect neuroblastoma, he adds, and studies like this add to the general knowledge of the disease as well as to the list of possible treatments. Maris' lab also has ongoing collaborative research agreements and material transfer agreements with many of the large pharmaceutical companies to facilitate the process of getting these markers from the lab to the clinic.
As pertinent as the team's immediate discoveries about neuroblastoma-associated genes are, their novel computational method could be applied to fields other than cancer research. "This method is agnostic to the disease, and that is one of the important points — it could be used in any sort of GWAS where one is more concerned about signals within genes than signals that may occur elsewhere in the genome," Maris says.