Signaling an important milestone in its effort to apply pharmacogenomics technologies to the drug-discovery process, the US Food and Drug Administration has enlisted Gene Logic to help it standardize certain gene-expression data.
The move, announced earlier this week, may embolden pharmaceutical companies as early as next summer to voluntarily submit microarray-based toxicogenomic data as part of their pre-clinical trials.
However, the standards will not be completed in time to become part of a pharmacogenomics guidance the FDA plans to issue in the fall, according to Karol Thompson, the CDER investigator who will be working with Gene Logic. The fall guidance, which will be drafted by a group headed by CDER Director Janet Woodcock, will be “much broader in scope,” Thompson said.
Gene Logic and Thompson’s team will work for the next two months to standardize gene-expression data requirements, and to fulfill the FDA’s broader effort to encourage drug makers to submit microarray-based toxicological data without fear of those data coming back to haunt them downstream.
“The main problem that researchers have run into while looking at pharmacogenomics and toxicogenomics data over the past several years has been that there are a lot of variables that cause the data to be different,” said Doug Dolignow, senior vice president of pharmacogenomics at Gene Logic.
The team said it will attempt to solve this problem by comparing gene-expression data found in Gene Logic’s GeneExpress library of control and vehicle-treated tissue samples against genes in microarrays manufactured by Affymetrix, Agilent, and Amersham Biosciences. The goal is to locate genes that are common across each of the platforms, and to identify the expression of these genes in Gene Logic’s GeneExpress library.
By looking across the samples — Gene Logic and the FDA said they will initially focus on genes linked to organs associated with toxicity (liver, brain, kidneys, and testes) the partners hope to rank genes by their expression frequency across each of the platforms.
“This [collaboration] aims at creating some sort of metric to measure data submission,” said Dolignow. Researchers will “have a set of genes that no matter what microarray platform you look at, the genes are present, and they’re rank-ordered in approximately the same way.”
The idea, he said, is that pharma companies performing toxicity studies that want to submit toxicogenomic data to the FDA would be able to take their control sample and run it across their microarray, showing that the invariant genes are represented at the right rank order and the right level when compared against the study set provided by the three platform companies.
“That would be an indicator that the quality of the data being submitted is consistent with the quality of the data been submitted by other studies,” Dolignow explained.
He said the FDA collaboration will likely wrap up in two months, and that results of the study will be made public by the FDA. He stressed that Gene Logic data used in the partnership will remain proprietary.
Thompson declined to comment on whether a new standard will appear in an FDA guidance down the road. However, “we will be able to assess whether this type of standard will be of value to propose as a voluntary part of a genomics submission” by June 2004, she said.
“This is kind of a short-hand way of knowing how [pharmas] can … get the right biology out of the experiment at the end of the day,” Thompson said in an interview with SNPtech Reporter. “It’s a way for us to know that the platform they’re choosing and the methods they’re choosing have the right performance standards we’re looking for in toxicogenomics studies.”
In the meantime, Thompson’s group will send the new standard to “a lot” of disparate labs that will investigate how it stands up across different platforms. She said these participants include drug makers, academic labs, bioinformaticists, contract research organizations, members of the National Life Sciences Institute, and a variety of gene-expression tool vendors. She declined to name the labs but confirmed that Affy, Iconix, Rosetta, Amersham, and Agilent are among them.
The FDA and Gene Logic have been interacting on this project for more than a year and a half through a collaboration with Donna Mendrick, Gene Logic’s vice president of toxicology. In fact, the company and the agency have been quietly working on the toxicology-standards project since June, said Thompson. According to Dolginow, it was Gene Logic’s massive database of rat gene-expression data — the company has more than 25,000 tissue samples — that led the agency to tap it for help.
“If they didn’t do it with us, it would take many years and a lot of money to do this. We’ve already generated the data,” he said. “Now it’s just an analysis.”
Dolginow said Gene Logic can take a drug compound, dose it in sub-toxic and toxic levels in rats, and take the expression data and predict whether the drug is hepatotoxic at 90 percent accuracy. “We’ve presented that data to the FDA. An important component in the future of [pharmacogenomics] is that you can actually demonstrate you can do this.”
Thompson agreed, saying that Gene Logic “has a large database of … control rats, and we thought that they would be a logical source to tap for some of this information to help us design these standards.” Rat genes are being used because nearly all toxicogenomic studies rely on genes from this organism rather than on human genes.
The issue of voluntary submission of microarray-based toxicology data has been a thorn in the side of many drug makers, and many say it is the linchpin in the FDA’s broader strategy of infusing pharmacogenomics technologies into drug trials. Some large drug makers, in fact, said they refuse to voluntarily submit gene expression-based toxicology data without clear guidance from the FDA, fearing that the data might inadvertently hurt their approval chances.
Pfizer, the world’s largest drug maker and a robust consumer of pharmacogenomics technology, is one such pharma. Bruce Littman, executive director of the drug giant’s clinical sciences global R&D unit, admitted at a pharmacogenomics conference at Yale in May that Pfizer actively shies away from submitting early-stage toxicology data for fear that some of the data, like those depicting activated proto-oncogenes, may be used against it down the road.
Woodcock, in an interview with SNPtech Reporter following that symposium, stressed that this concern is unfounded, and said she hopes her group’s fall guidance will adjust pharma’s perception: Early-stage gene-expression studies in many drugs will activate proto-oncogenes, she explained.
“We need to reassure the industry and make them feel very comfortable that we’re not going to therefore use that information in our decision making about their product,” Woodcock said.
The FDA’s plan has quieted Pfizer’s anxiety somewhat. “One of the barriers to growth of the science of toxicogenomics and pharmacogenomics has been the relative difficulty to transfer data and share data between laboratories,” Jim Mayne, the company’s executive director of toxicogenomics science, told SNPtech Reporter this week.
“Creating standards that allow sponsors to transfer data and analysis to the agency and allow laboratories ... to exchange data and ideas is going to help the pace of the development in the field as a whole,” Mayne said.