Open Trial Databases Not Widely Popular – Yet – Duke Researcher Finds

For three major databases allowing researcher access to clinical trial data, few requests were to validate results.

Endpoint validation work may have to become part of a clinical trial's primary analysis plan as it is not popular among researchers using clinical trial data sharing portals – one of the somewhat disappointing implications of an analysis by Ann Marie Navar, of the Duke Clinical Research Institute.

Validation studies were among the primary reasons advocates demanded that sponsors make trial results available to researchers. A number of pharma companies now have systems in place to allow researchers access to the data for legitimate research purposes.

But of 113 research proposals with data-sharing agreements, only five were for a validation study of a primary endpoint, according to a study published March 22 in the Journal of the American Medical Association.

Navar said in an interview with "The Pink Sheet" that a financial support system is needed to support the analyses, along with methods to allow publication regardless of the results.

That could include building the validation study into the clinical trial analytic plan at the beginning. An external group could be identified at that point to conduct the analysis, Navar said.

Validation studies are not popular because there is little grant funding available for the work. And proving something is already correct also will not likely garner much attention, she said.

"Incentive for validation studies may be limited as confirmatory reanalyses are less likely to be published," Navar and her colleagues wrote in the paper. "However, the one publication using these data was a validation study that found contradictory results from the initial article."

That 2015 study reanalyzed SmithKline Beecham's Study 329 that compared paroxetine and imipramine with placebo for treatment of adolescents with unipolar major depression.

It found that efficacy of paroxetine and imipramine was not statistically or clinically significantly different from placebo for the primary and secondary endpoints. Researchers also found clinically significant increases in suicidal ideation and other adverse events among those in the paroxetine group and cardiovascular problems in the imipramine group.

GlaxoSmithKline PLC pled guilty to promoting paroxetine, trade name Paxil, for off-label use as a treatment for depression in patients under age 18 in 2012.

The company agreed to a $3bn settlement with the Department of Justice, which also covered off-label promotion charges for Wellbutrin (buproprion) (Also see "GSK’s DoJ Settlement Requires Execs To Forfeit Bonuses If Misconduct Occurs" - Pink Sheet, 2 Jul, 2012.).

FDA May Not Publicize Its Validation Work

In many cases, FDA cannot be counted on to validate and publicize all trials.

Not all the studies in the open databases have been reviewed by FDA, only those submitted for a new product approval or a label change, Navar said.

And if FDA does a reanalysis of data it does see, it may not be published or made available to the public, Navar added.

Navar said a research group dedicated to confirmatory analyses could be created and funded to ensure validation work is completed.

It still is unclear who would bear the cost of the additional studies, Navar said.

When new disclosure measures were put in place for clinical research, there were concerns that NIH-funded investigators in particular could be forced to cut research in order to ensure data could be released publically (Also see "Clinical Data Sharing Could Increase Research Costs, Industry Fears" - Pink Sheet, 28 Apr, 2014.).

Less Than 16% Of Available Trials Requested

Navar and colleagues reviewed data from three major data-sharing portals: the Yale University Open Data Access Project (YODA), ClinicalStudyDataRequest.com, and the Supporting Open Access for Researchers initiative.

Data across the three sites included more than 3,200 trials from 14 sponsors (see table). More than 44% of the trials were Phase III, another nearly 24% were Phase I, about 18% were Phase II and nearly 13% were Phase IV studies.

Sponsors Making Clinical Trial Data Available
Astellas	Boehringer Ingelheim
Bristol-Myers Squibb	Eisai
Eli Lilly	GlaxoSmithKline
Johnson and Johnson	Medtronic
Novartis	Roche
Sanofi	Takeda
Union Chimique Belge	Viiv Healthcare
Source: JAMA article

A total of 154 research proposals had been approved and 113 had data-sharing agreements in place.

Navar and colleagues found that only 15.5% of the trials available had been requested.

She was surprised at the relatively low usage of the data, particularly given the discussion in scientific literature about the importance of open science.

Many researchers likely are unaware that the data are available because they have not been promoted sufficiently, Navar said.

"You can't build it and they will come," she said. "People have to know they will exist and have support."

Several sponsors were leery of making data publically available. But after establishing the systems, they were looking for ways to make more researchers interested in them (Also see "Clinical Data Disclosure Embraced By Pharma; Now It Needs Partners" - Pink Sheet, 1 Dec, 2014.).

Database Request Rates About The Same

Navar said the work being done with the data now is "important and innovative" and highlights the creativity of researchers.

Requests have included analyses of disease states and secondary analysis of treatment effect. There also were efforts to make trials more efficient.

Joseph Ross, associate professor of medicine at Yale University, who also works with the YODA program, told "The Pink Sheet" that he doesn't think there is a "right" number of requests, in part because the initiative is new.

He said the number YODA has received for J&J data it administers is about the same as those received by ClinicalStudyDataRequest.com for GSK data.

"We seem to be right on track with one another," Ross said.

J&J said in late 2015 that it had received 25 data requests through the YODA portal. But J&J and other sponsors were thinking of developing a single place to handle data sharing (Also see "Clinical Trial Data Sharing: Central Public-Private Portal May Expand Access" - Pink Sheet, 13 Nov, 2015.).

Twelve of the 177 research proposals that had been processed and met initial requirements were rejected, according to the paper. Navar did not have data on the reasons for the rejections.

Ross also agreed that more should be done to ensure the available data is used for public health and scientific advancement.

"We still need funders to step forward and support data users with grants and funding," Ross said. "Even if the data is being made freely available, it still requires the time and effort of an analyst and research team to analyze the data and prepare the findings for publication."

Navar added that there is little incentive for sponsors to continue adding more data to the open databases if usage remains low.

"It's sort of a supply and demand issue too," she said.

The National Academies of Sciences, Engineering and Medicine's Health and Medicine Division, formerly the Institute of Medicine, said in a 2015 report that data sharing should increase, but also allowed several loopholes like intellectual property protections, which made it unclear whether it would lead to any real changes (Also see "Clinical Trial Data Sharing: IOM Envisions Open And Closed Doors" - Pink Sheet, 15 Jan, 2015.).