A virtual encyclopedia
BASEL, Switzerland—Armed with the knowledge that cancer is a genetic disease and that cell lines reflect the genetic disturbances which drive it, Novartis and the Broad Institute have developed an online data repository that catalogues the genetic and molecular profiles of nearly 1,000 human cancer cell lines.
This resource, dubbed the Cancer Cell Line Encyclopedia (CCLE), provides a powerful tool for the design of cancer drug trials and will help researchers identify patients who could benefit most from specific drugs in development, according to the two organizations, which published the results of their collaboration March 28 in the journal Nature. The CCLE project was specifically a collaboration between the Broad Institute, the Novartis Institutes for Biomedical Research (NIBR) and the Genomics Institute of the Novartis Research Foundation to conduct a detailed genetic and pharmacologic characterization of a large panel of human cancer models, develop integrated computational analyses that link distinct pharmacologic vulnerabilities to genomic patterns and translate cell line integrative genomics into cancer patient stratification.
"The systematic translation of cancer genomic data into knowledge of tumor biology and therapeutic possibilities remains challenging," the CCLE's creators wrote in their article, "The Cancer Cell Line Encyclopedia enables predictive modeling of anticancer drug sensitivity."
"Such efforts," the researchers added, "should be greatly aided by robust preclinical model systems that reflect the genomic diversity of human cancers and for which detailed genetic and pharmacological annotation is available. Here, we describe the Cancer Cell Line Encyclopedia (CCLE): a compilation of gene expression, chromosomal copy number and massively parallel sequencing data from 947 human cancer cell lines. When coupled with pharmacological profiles for 24 anticancer drugs across 479 of the cell lines, this collection allowed identification of genetic, lineage and gene-expression-based predictors of drug sensitivity."
On the website, http://www.broadinstitute.org/ccle—which anyone can access—researchers can enter a keyword to search for genes, news items and publications; search results for a gene, including links to annotations and analyses; and browse, analyze and download studies and data sets.
"The goals of the CCLE project are twofold: one is to assemble a resource of genomic characteristics of 1,000 cancer cell lines, and the second is to develop tools to predict the sensitivity of those cell lines to cancer drugs based on genomic alterations," explains Dr. Nicolas Stransky, a computational biologist in the Cancer Program at the Broad and a co-first author of the paper. "Certainly, people have been doing these kinds of things in the past, but on a much smaller scale. There are many applications here. One is to better inform the clinical trials that are taking place in the development of drugs. The use of the CCLE here would be to better select which patients are more likely to respond when they are given a specific drug, because you can tell which cancer cell lines are being killed by a drug."
The team purchased cell lines and their associated information directly from several commercial vendors in the United States, Europe, Japan and Korea, says William Sellers, global head of oncology at the NIBR. Cell lines represent a diverse picture of cancer as a disease, as they include many subtypes of both common and rare forms of cancer.
"If someone buys a vial of cells that we characterized, it should be very close with as minimal drift as possible to what we used in our genetic studies. It was expensive to do this, but it was done to make this a publicly valuable resource," he adds.
Each cell line was genetically characterized through a series of high- throughput analyses at the Broad Institute, including global RNA expression patterns, changes in DNA copy number, as well as DNA sequence variations in about 1,600 genes associated with cancer, and pharmacologic profiling for several drugs in about half of the cell lines. Algorithms were developed to predict drug responses based on the genetic and molecular makeup of cancer cells.
The collaboration was "exciting" for Novartis "because we are in the drug discovery arena," says Sellers. "We are good at team-oriented and project- oriented science. It turned out to be a lot of fun because both teams worked together as a single project team."
In fact, a number of Novartis' clinical trials have already been influenced by the data generated during the collaboration, he adds. For example, Novartis used the data in the development of BYL719, a novel, oral, targeted anticancer agent that selectively inhibits the phosphatidylinositol-3-kinase (PI3K) pathway. The compound, which is being investigated in advanced solid tumor patients, has shown significant cell growth inhibition and induction of apoptosis in a variety of tumor cell lines as well as in animal models. In addition, in preclinical models, it has been shown to possess antiangiogenic properties.
"While this result wasn't unexpected, the power of the encyclopedia results motivated a specific trial design," Sellers says. "The strength of association was so compelling, suggesting that not only was the molecule a good molecule, but also that the best thing to do with respect to its clinical development was to focus the trial on patients who had a certain mutation, so we had the best chance of seeing early efficacy in patients."
Sellers acknowledges that "there is a lot of debate about whether cell lines are OK to use in cancer research." While human cancer cell lines represent a mainstay of tumor biology and drug discovery through facile experimental manipulation, global and detailed mechanistic studies and various high-throughput applications, many previous efforts have been limited in their depth of genetic characterization and pharmacological interrogation, he notes.
"It is important to remember that while cell lines are not always the best tools, they are the most widely used by cancer researchers," he stresses. "People know about their limitations, but it is important to remember that they are tremendously useful tools for studying cancer therapies and how drugs work. In certain cases, cell lines don't fully represent the genomic heterogeneity of cancers. Again, in certain cases, looking at cell lines won't give you a complete picture, but what we show in our paper is that in many cases, they are reasonable models."
"Our biggest hope for this project is that the data will be used by the community, but also that the biggest discoveries in the data are yet to come. This is likely to generate a lot of enthusiasm and be widely used," says Sellers.
Stransky notes that the CCLE is still an ongoing project, and its repository of data is neither final nor complete. In the second phase of the project, "our goal is to perform a much deeper genomic characterization using several sequencing techniques including whole-genome, transcriptome, exome sequencing, etc. We're also looking at other data types such as epigenetic alterations, phosphoproteomics and metabolomics," he says.
Pairing this information with ways to rapidly genotype patient tumor samples represents the next step in the effort to enable the personalization of cancer treatment, according to the researchers. Some major research hospitals already genetically profile cancer patients' tumors routinely, and many more are likely to follow, says the CCLE team.
"What we're trying to do here," Stransky concludes, "is lay down the basis of what personalized medicine would be in the future, which means we're trying to have the best match between a particular drug and which tumors are likely to respond. This is going to be tremendously useful for drug development in the future."