As part of an ongoing effort to encourage the submission of microarray-based toxicogenomic and pharmacogenomics data in the drug approval process, the US Food and Drug Administration’s Center for Drug Evaluation and Research has enlisted the aid of Gene Logic in a project to create a “reference standard” of genes that are reproducible across platforms and tissue types.
Gene Logic is the latest FDA collaborator in an initiative that so far involves Affymetrix, Rosetta Inpharmatics, Iconix, and FDA’s National Center for Toxicologic Research. FDA researchers are mining Rosetta’s and Iconix’s databases to select sets of rat tissues from the Affymetrix, Amersham, and Agilent microarray platforms that have the greatest number of unique genes between them to create mixed tissue samples. Gene Logic will then analyze these tissue samples against its database of expression data from control and vehicle-treated samples and rank the genes according to their variance of expression by tissue type — liver, heart, kidney, and testes.
“It’s a bioinformatics analysis,” said Doug Dolginow, Gene Logic’s senior vice president of pharmacogenomics. “We can look through the database across all these tissues and see of the thousands of genes that are common among the three microarray platforms — which of them are consistently found at a rank order level that is the same from tissue type to tissue type — and then we can use that standard so that whenever data is submitted to the FDA for a pharmacogenomics or toxicogenomic study, you would expect to see those same genes present in the rank order that’s expected, and you would expect to have some assurance that the technical quality of the data is good.”
The final reference set of genes, which the FDA plans to make publicly available by next June, would act much like any other calibration standard in the laboratory industry, Dolginow explained: “There’s a control material that you run, and whenever you run an experiment you see if the control is in the range of what you’d expect, so then you know the data is reliable. It’s the same approach here.”
Acknowledging the challenges of site-to-site reproducibility and other standardization hurdles of microarray data in addition to cross-platform comparisons, Karol Thompson, the primary CDER investigator on the project, admitted, “we don’t really even know if this is feasible, but we’re hoping to get that information.”
Dolginow, however, was confident that a standard set of genes should be attainable. Gene Logic has already identified what it calls “invariant genes” in human and rat tissues across a variety of tissues that it uses for quality control purposes. “We know it can be done,” Dolginow said. “Now we just need to see that it works across three platforms.”
According to Thompson, the project — even if successful — will only be one step in a much more complex process to standardize microarray data submitted with drug applications. “These performance standards could in the future be a recommended part of a nonclinical genomics portion of a drug submission, which is currently considered ancillary data,” she explained. More immediately, she said, they should encourage the voluntary submission of microarray data.
“The real advantage to us is that the industry moves forward,” said Dolginow. “If in the course of doing this it allows data to be submitted to the FDA with a clear understanding of when it’s going to have regulatory impact and when it’s not, and it helps foster the adoption of these technologies in pharmacogenomics and clinical trials and safety studies, then I think that’s good for everybody.”