Abstract
The Ivy Glioblastoma Atlas provides detailed information about genes expressed in different anatomic regions of human brain tumors. Researchers are already drawing upon the resource to identify novel therapeutic targets for this deadly cancer.
A first-of-its-kind online resource offers a platform for clinicians and scientists to explore the genetic basis of glioblastoma down to the subanatomic level.
The Ivy Glioblastoma Atlas—posted online 3 years ago, but analyzed and described in detail only this week—includes comprehensive maps of genomic alterations and gene expression profiles across the various histologic substructures of brain tumors collected from 41 patients.
“It's an extremely valuable resource,” says Roel Verhaak, PhD, a computational biologist at the Jackson Laboratory for Genomic Medicine in Farmington, CT, who was not involved in building the atlas. “Because it provides a detailed look into what the different cells are at a very granular level, it really allows us a way to much better interpret glioblastoma datasets.”
Unlike previous systematic efforts to characterize the genetic drivers of this deadly brain cancer—projects such as The Cancer Genome Atlas, which used bulk tissue samples, or various single-cell profiling studies, for which the anatomic origin of dissociated cells is typically unknown—the glioblastoma atlas specifies where certain gene sets in a tumor show enriched expression and novel gene mutations.
As such, it lets researchers probe the spatial dimensions of glioblastoma development, diagnosis, and treatment, says project lead Ralph Puchalski, PhD, from the Swedish Neuroscience Institute (SNI) in Seattle, WA. “You can determine not only where your genes of interest are expressed,” he says, “but also where cell types and phenotypes reside within the anatomic features.”
Puchalski led a team at the Allen Institute for Brain Science in Seattle, where he worked until 2015, to create the atlas. The researchers used laser microdissection to sort tumors according to classic histologic features, and profiled RNA levels in each region with sequencing and in situ hybridization assays. Nameeta Shah, PhD, and her colleagues at the SNI also characterized DNA heterogeneity within tumors by analyzing point mutations and copy-number changes in key glioblastoma-associated genes.
Together, the groups documented distinct genetic signatures in four discrete tumor regions: the outer boundary tissue, the tumor core, the densely packed cells around sites of necrosis, and the vascular lesions.
In unpublished analyses available online, the researchers screened for RNAs linked previously to glioma stem cells (GSC). The gene expression patterns revealed three populations of these tumor-initiating cells: one in the cancer's low-oxygen center; one in the perivascular region near the tumor's edges; and a third distributed throughout the malignancy.
The findings suggest that drug companies may have to “target all different kinds [of GSCs] at the same time to have a better chance of fighting the disease,” says Shah, now at the Mazumdar Shaw Center for Translational Research in Bangalore, India.
Support for this idea came last year when Jeremy Rich, MD, a neuro-oncologist at the University of California, San Diego, led a team that referenced the atlas to validate two promising drug targets for disrupting two spatially distinct populations of GSCs. In a separate 2017 study, the atlas helped Rich and his colleagues confirm that another potential drug target for glioblastoma is most highly expressed in the tumor's hypoxic core.
To date, at least a dozen other research teams have also drawn upon the raw atlas data and published papers on their findings—thanks, researchers say, to the atlas architects' open data policy. “I applaud them for not restricting access before their own publication was out,” says Verhaak, who mined the data last year to pare down a list of putative glioblastoma genes. “It's really already been picked up by the field.” –Elie Dolgin
