The Food and Drug Administration is laying the groundwork to build a massive database that one day may provide reviewers and in-house researchers with instant access to data collected from all clinical trials for new drugs and devices ever submitted to the agency.

The database, dubbed Janus after the Roman god of gateways, could lead to faster reviews of new drug and device applications. It also could give FDA safety researchers a powerful new tool for identifying safety problems that cut across classes of drugs or devices.

It already is hastening the day when companies that submit clinical trial data to the agency will use the same format for similar questions on a common electronic platform. "This is finally going to happen, and they're going to do it right," said an enthusiastic Rebecca Kush, president of the Clinical Data Interchange Standards Consortium.

The 240 member organizations in CDISC include the 30 largest pharmaceutical companies in the world, contract research organizations, academic medical centers, medical journals and potential vendors like Hewlett-Packard, IBM and Microsoft. The group was formed in 1997 to push for adoption of common electronic filing standards at FDA.

A Boon For Personalized Medicine

The agency earlier this year released a request for proposals to construct Janus, with applications due on June 18. Funded from the $1.1 billion appropriated for comparative effectiveness research under the American Recovery and Reinvestment Act, the data warehouse is being pitched as a tool for conducting subgroup analyses on different therapies for the same disease in order to determine which drugs work best in which patients - so-called personalized medicine.

"We are using part of the funds to support building a database that will support standardized clinical data, with an initial focus on data that will be transformed when being put in this database to support the comparative effectiveness agenda," said Vickie Seyfert-Margolis, a senior adviser to the chief scientist at the FDA. "In our case, it is really about personalized medicine and finding patient subtypes that have better safety or efficacy profiles by looking across multiple studies."

That ultimately could wind up limiting the approved indications for some companies' drugs. But the CDISC consortium is hoping the database will provide a financially rewarding benefit for every company seeking a new approval from the FDA - significantly lower clinical trial costs and faster turnaround times at the agency.

Codifying Efficacy Variables

CDISC already has created the Study Data Tabulation Model for use in Janus, which is a standard reporting format that will allow FDA reviewers to go into the database and immediately draw out and compile data from multiple studies. They currently have to transfer data manually, often from tables submitted by companies in formats that can't be manipulated.

CDISC also is hard at work developing common standards for reporting the individual elements of trial data sent via SDTM. It can be as simple as using "0" and "1" for male and female instead of "M" and "F." The already completed Clinical Data Acquisitions Standards Harmonization, or CDASH, includes 18 variables for patient demographic and safety data that are common to every drug or device application. These 18 variables comprise about 80 percent of all clinical trial data, Kush said.

The next phase for the consortium is to develop common elements for efficacy data, which are far more complicated than demographics and safety because there are hundreds of diseases with approved products. It already has codified more than 400 efficacy variables.

It was this complexity that in part led to the collapse of a prior effort at data submission standardization and database creation under former FDA Commissioner Mark McClellan. The Bush administration's Office of Management and Budget ruled in 2004 that it was a costly imposition on business after assuming it would require every company to add an additional step at the end of its clinical trial process to recode their data for electronic submission.

Deputy FDA Commissioner Joshua Sharfstein announced in February that the agency would renew its attempt to set a common electronic standard for submissions. According to Kush, the agency sent a draft of a proposed rule to OMB for review last month. She expects a better outcome this time around. "We have business cases showing that if you collect data in this way in the first instance, you can save 60 to 80 percent of the time and costs in clinical trials," she said.

"We've done over 200 trials with SDTM and if we use our version, it's not a problem," agreed Paula Brown Stafford, executive vice president of integrated clinical services at Quintiles and the next chairman of CDISC. "Some of our customers have come to us with their own interpretation of SDTM standards, though. If we can get pharma companies to design their trials using protocols and patient report forms using CDASH standards, they'll save costs throughout the process. If they had used CDASH at the beginning, they don't have to do that conversion at the end. You cut out costs and time."

However, moving to electronic data submissions that are easily manipulated by reviewers could trigger tougher reviews. "Currently, when we receive sponsor data, it comes in a variety of forms," said FDA's Seyfert-Margolis. "Some of it is flat - a piece of paper or a pdf - as opposed to data in a database. If you're a reviewer and you want to re-analyze it and it's sitting on a piece of paper, you don't have that data available to you.

"The idea is to have a database by which if you're a reviewer of a current submission and there is a related product or study that was done, it would be available to you without having to search through your file drawers," she said. "It's a way to provide for a more scientific review by allowing FDA researchers to look across trials."

A Look Backwards At Health Outcomes

Of course, given the source of funds, the initial uses of Janus will be to look backwards at trials already in the FDA's hands. The contract solicitation calls for prospective bidders to host workshops with stakeholders to come up with candidate studies and methods of analysis. At least two major projects are contemplated in the first three years.

The contractor will construct Janus by identifying FDA-held clinical trials relevant to the priority areas and putting them into the database. The contractor also can look for health outcomes data from other sources. "The FDA houses the largest known repository of clinical study data - including data on drug biologics and devices safety, effectiveness and performance," the solicitation notes. "These data are an invaluable resource for comparative effectiveness research."

The winning bidder will be expected to develop analytical methods of using the database for comparative effectiveness research, and training FDA staff in those methods. The agency also wants the contractor to develop "best practices for submissions of studies to the FDA when they involve product comparisons."

Unfortunately for consumer groups and academicians, outside researchers will not have access to Janus. "We understand that anything that comes from sponsors will have to be maintained as confidential," said FDA's Seyfert-Margolis. "This is not a public repository. It's for the FDA's use."

Beyond the two studies contemplated under the current contract - no dollar amount was attached to the proposal - how much of the massive amount of past clinical trial data sitting in FDA file cabinets and on FDA hard drives will get entered into Janus will depend on how much money Congress appropriates for the task. ARRA may provide enough money to get Janus up and running and conduct a couple of studies, but it will remain an empty vessel with respect to the vast majority of already submitted clinical trials without more resources.

Filling The Vessel

FDA officials are cautiously optimistic that it won't be as expensive as it sounds to convert all its data to the SDTM format. "You don't have to rekey it. You write code to reformat it," said Seyfert-Margolis. "You might move it from Excel to a program that puts it in the right table form for the database."

"We're constraining this within ARRA-funded activities right now to clinical data already submitted to the agency that meet the requirements that the Agency for Healthcare Research and Quality put forth for comparative effectiveness research based on the Institute of Medicine's priorities," she said. "FDA is going to convert only data related to particular questions that fall within those priorities."

The targeted funding "allows us to build this database, test it with real large datasets and make sure it works," she continued. "We'll be doing actual analysis on this old data," she said.

"Longer term, they want to build a database to put all these datasets in so they can do meta-analyses or comparative effectiveness trials," agreed Quintiles' Stafford. But that will take years of reconfiguring data to SDTM standards.

When FDA officials met with industry representatives in February, they told a different story about Janus' implications. "What I heard them say is that they wanted to decrease the review time and get products to market," Stafford said. "Their focus is getting those SDTM guidelines utilized so their reviewers can cut down their review time."

- Merrill Goozner ( ¹ pinkeditor@elsevier.com )

God Of Gateways: Will FDA's Massive Janus Database Drive Review Efficiency Or Comparative Effectiveness Research?

More from Archive

More from Pink Sheet