Q&A: The Future of Clinical Trial Data Sharing

The push to increase transparency of clinical trial data continues to build considerable momentum, amid such factors as the explosion of digital health and new requirements by regulators in the U.S. and Europe to disclose study results to the public. Not lost in this movement have been new efforts to break down the barriers to true clinical trial data sharing between researchers in industry, academia, and government. With vast amounts of clinical data available today, many believe that making the information more widely accessible can help advance scientific discovery and research, improve patient care, and avoid unnecessary data duplication.

The quest to “liberate” clinical data in the interest of patient health is one viewed passionately by Dr. Michael Rosenblatt, Executive Vice President and Chief Medical Officer at Merck & Co., who will be speaking on the topic at this month’s DIA 51^st Annual Meeting. The former physician and

educator has spent much of his career generating and teaching about data, and now, as Merck’s primary external advocate on medical issues, he routinely collaborates with outside scientists, organizations, and academic leaders regarding the issue of data sharing. Rosenblatt, who is co-chair of the DIA meeting, will take part in a panel discussion exploring the recent findings from the Institute of Medicine’s (IOM) landmark report on clinical study data sharing. The panelists will also discuss the major risks (privacy and proprietary concerns, IT challenges, etc.) that they believe must be addressed if the potential of this movement is to be realized. Rosenblatt recently touched on these issues and more in an interview with Michael Christel, managing editor of Applied Clinical Trials.

Michael Christel: What are some initial key learnings from the industry’s perspective when reviewing IOM’s data-sharing report?

Michael Rosenblatt: Over the last few years, there has been a growing drumbeat about the need to share data from clinical trials. We and other pharmaceutical companies and people in different sectors of healthcare have felt that that would be very useful, that there are potential benefits. But the IOM is really the first and only group that pulled together representatives from all the key sectors to work on this problem, to develop some principles, and to make some recommendations. They pulled people not only from industry, but academia, editors of medical journals, people from regulatory agencies, and charitable foundations.

They’ve enunciated some important principles. One is, while you know that there is potential benefit to sharing clinical trials data, the data needs to be shared responsibly-access should go to people who have a research question or a research hypothesis, and the information they learn will be reported and made available publicly. Another important takeaway is that these aren’t recommendations that apply to only one sector, like “industry needs to do this.” Anyone who generates clinical trials data should be prepared to share that data.

We live in a world in which drug companies sponsor trials, the NIH sponsors trials, the European governments sponsor trials. If you don’t bring together data from all of those sectors, potentially you get blind spots and distorted information. I think that will be a big part of the sharing-not only sharing it responsibly, but trying to be comprehensive and having one set of principles apply to everyone.  

MC: The IOM committee concludes that sharing study data is in the public interest. In efforts to enhance transparency, to what level will sponsors need to adapt their R&D policies and overall clinical research culture?

MR: For Merck, specifically, we were already doing this. I think we’ll be doing more of it, in part because the demand is going up. We do need to work on a number of technical aspects. How do you share this data in ways in which you can still as much as possible keep the data secure, so that patient privacy and patient confidentially aren’t violated? We need to find platforms that will make the data accessible to those whose requests are going to be approved, and to make the data accessible to all the companies that are participating in this. There are big questions: what’s the type of data? What’s the timeline? The costs are going to go way up. We’re finding ways to do it. But for others, this may be a challenge, just the actual implementation.

In terms of the mindset, we now need to think that every trial we do is something that we should be prepared to share. In a way, that’s a statement to the public and the patients who go into our clinical trials-that we’re not the exclusive owner of that data; others are entitled to make use of that data, too, for patients and for the public good.         

MC: Can more data sharing help improve clinical trial quality, through feeding back into future trial designs?

MR: I think it can. In a way, some of this is a leap of faith. We don’t have a lot of examples yet of actual benefits that we can point to. But everyone believes that when smart people look at data from different perspectives, that will give us insights into what our patients need, which group is responding, which group isn’t responding, do drugs do more or less than we thought they did, in particularly on safety issues? Clinical trials usually involve thousands of patients when you get to a stage III outcomes trial. But when a drug is more rare, the side effects often aren’t seen until the drug is on the market. It may be possible through the sharing of clinical trial data to understand the more rare safety issues because you can aggregate some of that data that won’t appear in any single trial.

MC: How are pharma companies approaching the challenge of weighing greater public/regulatory demands for transparency with proprietary concerns?

MR: We see this as an issue that’s very important, and it’s bigger than ourselves. We were pleased that the IOM recognized some of the issues that were very real to us. One is the recognition that there could be proprietary information that could put us at a competitive disadvantage if it were revealed too early. The recommendations of the report allow for adjustment to be made. Those things do have to be absolutely protected in any implementation plan going forward. The other thing is we all have to do it at the same level. So if academia is coming in and they’re going to be sharing their data, they have the same concerns. That sort of sets the level at which kind of data is available, who’s it available to, etc.         

MC: Your DIA session will explore other barriers to clinical trial data sharing. What are your thoughts on the risks associated with misleading analyses in particular?

MR : It’s important to remember that with much of the analyses-especially when you’re getting to big data-the rules of the road and the methodologies for secondary use of data are still in the early days. You might have a question and you might have various kinds of methodologies that you can use to answer that question, some of which would be valid and some which might not. There isn’t enough experience and standards set yet to know which methodologies clearly should be used and which shouldn’t be used.

The other thing is we can get in situations where a solo researcher, with access to the data, can come up with a finding using his or her methodology that’s a little off the mark-maybe misleading or inaccurate and hasn’t been scrutinized in the same way that either the company or regulatory agency has done. Then the researcher goes out and publishes this-puts it on the web, or in their blog, or whatever they’re going to do that’s not peer reviewed. So the danger is, one, it’s never helpful to medicine to have misleading information circulate around. But you can compromise or even lose drugs that actually provide many patients with benefit. Sometimes that erroneous information never goes away. Think about autism and vaccines, for example. It’s extremely important that there’s integrity around this process and that we get the right information out there.

MC: Do you feel potential compromises in the area of patient privacy might be the biggest risk in this whole effort?

MR: I do. I see this as a potential safety issue. Because even though the data is supposedly de-identified, no one in this world can assure patients and physicians that the data is absolutely secure. The National Security Agency of the United States was hacked, as were the Swiss banks, and as have been a number of major retailers in the United States-and now health systems. Why did we think health systems were somehow protected? What’s at risk is if you reveal that a patient has mental illness or a drug problem or had that in the past, in the U.S. that can be a problem in getting insurance or getting employment. There are places in the world where if you reveal someone’s religion or their ethnicity or their sexual orientation, that’s potentially a fatal outcome.

You have to remember the direction we are going in. We’re going to be layering more and more genomic data into these databases. And with genomic data, you can absolutely identify an individual uniquely based on their genetic data. There are safety concerns that we’ll have to address with informed consent, because we can’t give assurances that the dataset is bulletproof. It’s one thing to generate the data and hold on to it the way clinical researchers have done for decades now. It’s quite another thing to open it up to sharing. We have to find a way. We have to be able to share; it’s really important. But we have to remember the nature of the information we are sharing.

MC: With industry initiatives underway to set up data transparency workstreams, how do you see clinical trial data sharing evolving in the next few years, particularly in establishing the infrastructure necessary to make it work?

MR: Although it wasn’t done intentionally this way, a number of companies and groups have been piloting out projects and ways of doing it in parallel. So I think we’ll start comparing best practices. My guess is we’re going to be moving more and more toward a common platform for sharing. We now have a set of guiding principles that more or less fit where everybody wants to go. Now we’re into the blocking and tackling of implementing this over the next few years.

MC: Will its ultimate success hinge on ensuring the data is shared responsibly, as you alluded to earlier?

MR: That’s our view. Some think there shouldn’t be very much sharing, while others think you should just take all your data and hang it out like laundry and let everybody have at it. If your main goal is to get data shared for research purposes to benefit patients, then I think you wind up somewhere in the middle.