FDA eyes pharmacogenomics data.

The U.S. Food and Drug Administration (FDA) is looking at how microarray and toxicogenomics data may be incorporated into its drug review process. Field insiders expect microarray data will eventually be a standard component of submissions for both investigational new drug applications (for use in clinical tests) and new drug applications (for marketing new drugs in the United States). However, in the short run, the FDA's capacity to manage a deluge of these additional data is limited. And key questions remain as to exactly how and when the FDA will accept microarray data.

Only about 1 out of every 10 drugs makes it to the first phase of clinical trials, according to industry estimates. Current estimates of the cost to develop a drug run $800 million. Experts contend that if the use of microarray data could even double the efficiency of drug development--for example, by increasing the number of drug candidates that make it to clinical trials--the savings would be substantial. And the potential to increase the efficiency is greater than that, says Leslie Browne, chief operating officer of Iconix Pharmaceuticals, based in Mountain View, California.

Microarray data could also improve drug quality. Research has shown that gene expression data can catch changes early on that traditionally are seen only in pathology. "A tool like this provides an opportunity to weed out compounds early that will have problems," explains Browne. In a rat study, for example, lesions were caught at day 5 compared to day 28 for classical histologic methods. Other retrospective studies have demonstrated the strength of the microarray as a predictive tool across species.

The FDA released draft guidance on pharmacogenomics data submissions in November 2003. "The draft guidance is a great start to this process, and the developing debate will enhance the field," says Browne. By embracing the technology early on, "the FDA in this case has been helping to push this forward," he says.

Norris Alderson, the FDA senior associate commissioner for science, intends to develop one set of standards for use throughout the agency, including all FDA centers.

"We're striving to achieve--as much as possible--harmony within the agency as we move forward to apply genomics in a regulatory setting," explains John Leighton, supervisory pharmacologist in the FDA Division of Oncology Drug Products. "Our thinking is evolving as we see more and more submissions containing genomic data and gain a better understanding of what is useful and what isn't from a regulatory standpoint."

To help develop the guidance and learn how to address microarray data, the FDA Office of Testing and Research has launched two gene expression database projects. The first, a collaboration with Iconix, will familiarize FDA reviewers with microarray basics using Iconix's DrugMatrix toxicogenomics database. So far, DrugMatrix contains findings on 600 compounds at multiple dosage amounts and times. Gene expression data are linked to information on pharmacology, histopathology, clinical chemistry, and toxicology, providing a reference for FDA reviewers to compare findings with known results. Iconix is also training FDA reviewers on quality control for microarray data generation, as well as how to analyze data across multiple microarray product platforms and validate biomarkers from integrated chemogenic data sets.

The second project, in partnership with Schering-Plough and Affymetrix services provider Expression Analysis, based in Durham, North Carolina, is building a database for mock gene expression data submissions. According to the 23 June 2003 edition of the online news source Bioinform, the planned internal gene expression database will help educate FDA reviewers about the format, content, and context of microarray data submissions.

Most experts agree that the FDA has been legitimately conservative in its use of toxicogenomics data so far, because there are real risks in adapting microarrays and similar technologies before they are mature. "The idiosyncratic response of individuals to drugs is still quite unknown, and just because we could measure forty thousand genes at a time doesn't make this problem any easier to solve," says Atul Butte, a physician and instructor of endocrinology and informatics at Children's Hospital Boston and Harvard Medical School.

Initially, Leighton sees microarray technology as an adjunct to traditional drug evaluation tools that will help researchers better understand the underlying mechanisms of toxicity, especially for long-term studies. Moreover, he believes such data will play a greater role, at least initially, as a tool for enhancing an understanding of a compound's pharmacology rather than its toxicologic properties.

But much work needs to be done before the FDA can determine how microarray data should be used in regulation, and standards need to be established before the agency can decide how to use such information in risk assessment. For example, there are no known valid biomarkers to date, as called for in the guidance. Among other technical challenges, a process needs to be established for how a biomarker progresses from "experimental" to "probable" status, and then to being a known biomarker. The FDA and many other research groups are striving to correlate content and format of gene expression microarray data with standard toxicology and pharmacology data.

Industry in general has been slower than the FDA to promote the use of microarrays in the development of new drugs. "Drug companies have been reluctant to embrace it because they have realized the disadvantages," Browne says. But the FDA is working to demonstrate that voluntary submission won't come with penalties. "We hope to overcome the fear by some in industry that the agency won't know how to use the data or make inappropriate use of the data," Leighton says.