The US Food and Drug Administration plans to issue a preliminary list of genes that are common to the three leading microarray platforms — Affymetrix’s GeneChips, Agilent Technologies’ microarrays, and Amersham’s CodeLink — some time this summer, Karol Thompson of the FDA told BioArray News this week.
Last week, Gene Logic of Gaithersburg, Md., delivered a CD-ROM to the FDA Center for Drug Evaluation and Research containing data derived from the company’s library of gene-expression studies of rat liver — a collaborative project initiated in August.
The data represents an effort to tease out a list of genes that could be used as a standard benchmark for microarray laboratory processes.
“I would hope it would be something that people can use as a proficiency test in the near term, and maybe something they can send along to the FDA along with their genomic data to give assurance of how well they can run an assay,” Thompson, an investigator with FDA’s CDER, told BioArray News. “Right now, there is really no way to do that except comparing their data to their historic data.”
The standard being developed in this collaboration applies to the probes that are etched or spotted onto industrially mass-produced microarrays. It is part of an multi-pronged effort conducted by industry, academia, and government to develop standards that are seen as key hurdles to overcome in moving this nearly 10-year-old technology from the research labs to clinical applications, which require inherently greater accuracy and reproducibility.
Thompson said that in addition to this project, the FDA is also involved with a standards effort for mixed tissue analysis.
The data from this project, however, will serve as a something that people can use quickly, Thompson said.
“We want to have something up and running and usable within a short period of time,” Thompson said.
This standardization effort could be likened to shooting a basketball off a moving ship into a trash can as the technology is constantly and rapidly evolving. Last summer, when this project started, single whole-human-genome microarrays were just ready to come into the market. Now, all the major manufacturers have this product. The probes used on mass-manufactured microarrays range from 25 to 60 bases long, and many of them can only fairly be described as scientifically grounded best guesses as to the gene sequences they represent..
The data Gene Logic delivered was derived from a comparison of a list of FDA-selected gene fragments against the company’s library of gene-expression data, which was produced by analyses of vehicle-treated rat liver samples using the Affymetrix GeneChip platform.
“This interaction was to try to identify invariant genes in rat tissue and find genes that always exist in a certain order of plenty,” said Donna Mendrick, vice president of toxicology for Gene Logic, who has spearheaded this collaboration. She said that the FDA provided the company with a list of thousands of genes to start the project. She declined to quantify the list the company returned to the agency.
The FDA hopes to produce a list of genes that have a stable rank order and good quality, Thompson said: “We are using liver samples at the moment. It is our test case.”
Thompson said FDA researchers would compare the Gene Logic data against other data to evaluate it for consistency.
“We will probably publish the results and let other people look in their own systems and give us an outside evaluation,” said Thompson. “We want to put out the first phase of it this summer and get the rest of the data out as soon as possible because we want people to use these as we come up with the usable benchmark genes and then go and evaluate them in their lab, or make the standards themselves, and see what kind of results they get.”
Gene Logic is one of several collaborators working with the FDA on microarray data.
At the end, this project may produce a proficiency metric for laboratories. At this point, it is unclear exactly what that measurement will be, Thompson said.
“We are thinking about showing a linear range and certain precision of measurement,” Thompson said. “This will be voluntary, not a requirement.”
Mendrick told BioArray News that she was unsure if the eventual goal of a standardized list of invariant genes that extends across the major microarray platforms is achievable.
“I’m not sure that anyone is going to get a list that will work across platforms,” she said. “Maybe what we will end up with is a rank order for Affymetrix.”
Working with microarray data will benefit FDA, she said.
“Part of the interaction is helping them see the data,” Mendrick said. “I can tell them from personal experience that you don’t understand microarray data until you try to work with it yourself. Take a file with 8,000 rows and 12 columns [that] hits your e-mail. You just look at it and say this is hopeless. This data may help them get their internal people up to speed [on microarray data] and what can go wrong and what can go right. We want to continue with Karol Thompson’s group, and expand this agreement. I’m a strong believer that microarrays are a key to understanding toxicity and it’s to everyone’s benefit to get the FDA up to speed.”