Print BCTV: Opening the Books -- The future of open access to pharma's clinical trial data

Opening the Books

Transcript of BioCentury This Week TV Episode 134




Dr. Martha Brumfield, CEO, Critical Path Institute, Tucson, Ariz.

Peter Doshi, Johns Hopkins University, Baltimore, Md.

Perry Nisen, SVP Science and Innovation, GlaxoSmithKline plc (LSE:GSK; NYSE:GSK), London, U.K.



Advair fluticasone propionate/salmeterol, GSK

Avandia rosiglitazone, GSK

European Medicines Agency (EMA), London, U.K.

National Institutes of Health (NIH), Bethesda, Md.

Paxil paroxetine, GSK

U.S. Food and Drug Administration (FDA), Silver Spring, Md.



Steve Usdin, Senior Editor




STEVE USDIN: Drug companies are under pressure to open their files. How much should they reveal? And what will the public learn? I'm Steve Usdin. Welcome to BioCentury This Week.




NARRATOR: Your trusted source for biotechnology information and analysis, BioCentury This Week.


STEVE USDIN: Pharmaceutical companies gather huge amounts of data about their drugs. Traditionally, they've kept most of this information secret. Those days are ending.


Under pressure from academics, medical journals and governments, drug companies have started to share vast databases with independent researchers. Watchdogs argue it's necessary to verify whether companies and regulators came to the right conclusions about safety and efficacy. Some researchers want to mine the data for new discoveries about benefits or risks and understand why clinical trials fail.


Now the debate is shifting from whether drug companies should share raw data to how much should be shared, who gets access and how to protect patient privacy. The benchmark case is GlaxoSmithKline. The giant U.K. pharma has launched the most comprehensive initiative to date to make data available to independent researchers.


Today, we'll hear from Peter Doshi of Johns Hopkins. He's a leading proponent of clinical trial transparency, and from Martha Brumfield, CEO of The Critical Path Institute. C-Path works with industry and academic researchers to make data available and useful. First, we're joined by Perry Nisen, Senior Vice President for Science and Innovation at GlaxoSmithKline, to discuss his company's data sharing initiative. Peter, I want to ask you, to start with, to describe what it is that you're doing and how you're doing it, because you're not just making your files open for anybody to start rummaging around in them, right?


PERRY NISEN: Yes. Thank you. So what we will be doing is making available anonymized patient-level data from our clinical trials. We've made the decision to take, in the first instance, an intermediate path, where working through an independent panel to evaluate research proposals, the veracity of what people want to do, the qualifications of people doing that work, to make available the results of the primary data from all of our clinical trials, which we have anonymized the personalized information about the individual -- their names, their date of birth, et cetera. And we will make that available through an independent channel.


STEVE USDIN: So can you give some examples, maybe, of the kinds of research that people are going to be able to do, that you envision that they will be able to do, using this very detailed information that you're making available, as opposed to the summary information that's published in medical journals that people are used to seeing?


PERRY NISEN: I think people are interested in all kinds of things, everything from testing methodology, for interrogating placebo data, for ultimately comparing results in studies and meta-analyses of different studies to affirm or refute conclusions made by sponsors and regulators, as well, to look at the data in whatever ways that they choose to look at the data. So I think there's a wide spectrum of those possibilities.


STEVE USDIN: And you said there's going to be an independent panel that's going to determine what research can be conducted using your database. What are their criteria going to be for determining what's legitimate and what isn't legitimate?


PERRY NISEN: Yeah. So thank you for that. So this is really going to be a first step in a journey.


Our real intent, and we hope to be able to move to that by year end, maybe, as quickly as that, is to move to an independent custodian for all of those data to which other sponsors and other participants could submit their data as well. But in that first step that we'll be taking is we've identified an independent committee of individuals, a well-known and independent physician scientist, epidemiologist and ethicist, a statistician and a patient advocate, where we will have a website to which people, researchers, investigators, could submit their proposal, very short one, their research proposal, what their analytical plan is, to disclose how they're managing whatever conflicts of interest they may have.


And it's really a completely independent assessment by this group of, is there a research plan. Is it an answerable question? Do they have individual or individuals with the expertise in terms of data management and the statistical expertise to conduct that study? And that's really it.


And that's really completely independent of GSK. We will post the questions being asked and what's done and what's not being done. But we won't be taking those decisions as a company.


STEVE USDIN: Will your databases be open to your competitors? Will companies developing drugs that potentially could compete with yours be able to query your database, learn what you've learned on a level of the individual patients, and then use that for their drug development programs?


PERRY NISEN: I think need to make their research, their information available. As I said, we think this is a first step to making these anonymized, patient-level data available. We hope this is going to nucleate efforts amongst other sponsors to submit their data and agree to a mechanism where personalized information can be made available, because there are challenges related to retaining confidentiality, respecting informed consent, misuse of the data and rendering conclusions, for example, about safety or benefit that may not be conducted in a scientific way.


STEVE USDIN: So one of the things that most of the press stories that have come up about your initiative have mentioned the fact that GSK had a large settlement with the government. And part of that was around allegations that GSK admitted of not disclosing information about a drug called Avandia. Has that informed what you're doing here? Has that changed the culture at GSK around transparency and sharing?


PERRY NISEN: Our decision to make patient-level data, and our evolution toward transparency and making information available, is independent of that, completely independent. And we believe this to be the right thing to do. It is the right thing to do for science. It's the right thing to do for society. And it's completely consistent with the values that the company has to be transparent and make that information available, clinical study reports available, the trials that we're doing available as well.


STEVE USDIN: Well, thanks. We're going to be joined by Martha Brumfield and Peter Doshi in a moment. First, transparency campaigners focus on industry. But academic and government researchers have a worse record when it comes to disclosure.


U.S. law requires publication of the results of clinical trials on within a year after a study is finished. Less than 10% of government-funded studies are posted on time. Here are the data from the British Medical Journal.




NARRATOR: You're watching BioCentury This Week.




STEVE USDIN: We're continuing our discussion with GSKs Perry Nisen and from Martha Brumfield, CEO of the Critical Path Institute. C-Path works with industry and academic researchers to make data available and useful. And Peter Doshi, a researcher at Johns Hopkins University. Peter has advocated for greater clinical trial transparency and participated in reviews of industry data.


Peter, I want to start with you. You heard Perry describe what GSK is doing now. You've been fighting for greater data transparency for a long time. What are your views about what GSK and other companies, where they're at today?


PETER DOSHI: Well, I think I'm very pleased that we're sitting at this table together. I'm really happy that we're having this discussion. And I think it's really good that GSK, in particular, has a let's learn as we go attitude about this, because I don't think we're there today. I don't think that GSK's plan, as they presented it, is what we all hope for.


So I took the time yesterday to go over in detail about what GSK has presented. And just in terms of the what, what data are we talking about, a few things stuck out. One is it's only trials since 2000. So that is no Paxil, no Avandia, Advair. We're not talking about any of those drugs.


The other one was no unpublished studies will be available, only published studies. And the other one was that no off-label uses where the drug has been tested in a trial for an off-label use. That won't be on the table. So we're talking, actually, about a much more limited set of trials than I had assumed, based on the press releases.


STEVE USDIN: And Perry, what will be your response to that?


PERRY NISEN: So our intentions are to make all the data available. What we're doing is moving forward, as you said, in a stepwise manner. We can't do it all at once, because we're not ready to do it all at once.


We'll have all of our studies that have been done available and accessible for what they are. And then we'll work with researchers, if they have questions that are endorsed by the independent committee and independent custodian when we get there. And we'll do our best to go back and see if we can find those data. Maybe many of those studies that you're referring to, especially the older ones, are going to take pretty considerable amounts of effort to go back to what are often paper report forms and in archives and in various places.


Our policy is to publish all of our work. So we're not intending to hide anything. What we're trying to do is move forward in a way to get started, as you said, learn as we go, and make information available as we can.


The resources necessary to secure the data, to anonymize it, the older it is, it's harder to get. And it's really a matter of getting started and doing what we can. There is not an intention to just make certain stuff available.


MARTHA BRUMFIELD: I'm curious as to whether one of the limitations you've encountered, because it's something we have in our programs, is that the informed consent that was originally signed by the research subject doesn't specifically allow for secondary research use of their data. And sometimes that has been a limitation in being able to have access to data.


PERRY NISEN: So I think that's a very important question. And our position on that is, when individuals offer informed consent, it's within certain boundaries. Our position is to respect those boundaries. That being said, as we move forward with current studies, we endeavor to be more comprehensive and open as to the ways that those data and information can be used.


But you're exactly correct. We do have to respect the boundaries. We believe we need to respect the boundaries of the informed consent that people signed at the time they did it.


STEVE USDIN: One of the things I'm interested in, both Martha and Peter, your view on it, is the difference between being able to query data, which is essentially what GSK is doing, and demands to actually have the data in hand and what are the pros and cons on both sides of that?


MARTHA BRUMFIELD: Well, I'll be happy to start. And I'll speak from the perspective of what we do at Critical Path Institute. Our goal is to aggregate data, so that we can query it to answer a specific research question. So in order to aggregate that data, it's very important, first, that the data be de-identified, that the informed consents allow this to happen. But then the data have to be in a format that allows them to talk to each other. So we have to often standardize them.


We see that the value proposition that, by aggregating large sets of data across multiple trials and multiple sponsors -- industry, academia, government laboratories such as NIH -- that that allows us to learn much faster about the intricacies of disease and, hopefully, will inform more efficacious and novel therapies going forward.


STEVE USDIN: And Peter, we're going to have to get to you just after we take a break. The Coalition Against Major Diseases, led by the Critical Path Institute, runs a collaboration among drug companies, academic researchers and government regulators to share data from Alzheimer's trials. We'll talk about how data sharing can advance the search for cures in a moment. First, a snapshot of the Coalition Against Major Diseases.






NARRATOR: Now, back to BioCentury This Week.


STEVE USDIN: We're talking with Perry Nisen, Martha Brumfield and Peter Doshi about unleashing clinical data to find new therapies. Peter, you wanted to make a point about the distinction between being able to query a database with clinical trial data and actually having the database and the data in hand.


PETER DOSHI: Right. So I think those are the right words to use, querying versus accessing or querying versus sharing, because the GSK policy, as I understand it, won't give people, even after they've signed a contract, won't give people the data for them to take home and analyze on their own computer. Rather, it's setting up a kind of protected portal, a sandbox, where GSK will give people access to that data set on GSK's computers or somebody else's computer, but not their own. And that access will only last one year.


And so what I see as sort of, underlying all this is the assumption that these data are too dangerous in the wrong person's hands, that we can't have real sharing, and that there should be gatekeeper mechanisms all along the way, from the panel that's going to adjudicate, who can see the data and what data they'll see, all the way to how they'll be accessing it, which is not on their own computer, but on some other computer.


STEVE USDIN: So Perry, what are the reasons for that? What are the concerns that you have with just having the database out there, putting it on CDs and sending it around?


PERRY NISEN: I think our main issues are those related to doing what we can to try to assure confidentiality. And what we've chosen to do is to have those data available and all the software, the SAS and there are programs available to do whatever studies people want to do with them, but to keep them on a closed server separate from GSK. And actually, our intentions are moving toward an independent custodian altogether in the first instance.


And it's not the purpose to clarify what the independent panel is, as we move toward a fully independent custodian, to be the gatekeeper over that. It's to make sure -- to make sure -- to assess that there is a question being asked that there is a protocol present and that it's answerable, approachable, that the work will be published and made available. Those are our intentions. It's not to control it. The point of why can't it just be downloaded, made available, just full-out free access for all, and our concerns are really in the service of trying to reasonably assure confidentiality, and same so for a data sharing agreement, to ask the investigator to use the data for the purposes that they say that they will.


STEVE USDIN: And I want to bring Martha in and ask you again about your experience and what it is that you've been doing and some of the complexities about gathering and sharing this kind of data.


MARTHA BRUMFIELD: Mm-hm. Well, thank you for asking. We are a non-profit. And that enables us to work in the space of being a neutral third party, so that we can bring together, in the pre-competitive space, researchers from academia, industry and government labs, EMA and FDA, so they can collectively decide what type of research question they want to ask, similar to what Perry is describing. And then they determine what type of data are needed to try and answer that question.


STEVE USDIN: And can you give some examples, because we've been talking very much in the abstract, about the kinds of questions that can be asked and answered that are valuable with these kinds of data.


MARTHA BRUMFIELD: Absolutely. So with our Coalition Against Major Diseases, we started focusing on Alzheimer's disease and because so many companies have experienced major failures in their Phase III trials, and we're not seeing a lot of efficacy from the treatments that are available, it was thought that we really need to understand the disease progression. And the best way to do that was to aggregate data across a number of trials.


So we had nine companies that were willing to share their placebo arm control arm data, which we pooled in a database. It took months to do this, because they had to all be in a common format. But from that data, we've been able to develop a clinical trial simulation tool, which now will be publicly available to any researcher. And they can use that algorithm, along with the specifics of their drug compound, to help them design a clinical trial that is more likely to be able to show differentiation between the treatment effect of their active compound and the comparator.


STEVE USDIN: So we've just got a short time left in this segment, but is that the kind of question that you envision, Peter, people asking? Or you're talking with different kinds of questions?


PETER DOSHI: I think people are going to query these data in many, many ways, not just this one particular way. I think some people are going to look to replicate studies. Some people are going to look to verify the accuracy of publications in the medical literature. And that's a big concern. Academics have been looking at published literature for decades now, thinking that that was the gold standard.


STEVE USDIN: Is that a legitimate demand, to be able to look at this data, to be able to replicate it independently?


PERRY NISEN: I think, fair enough, the scientific method is generally to repeat an experiment. But for large studies, how you repeat those experiments? So independent assessment and evaluation of those data, either for new approaches, new endpoints, better understanding of the disease, and to replicate the findings, seems completely valid.


STEVE USDIN: Peter, we've been talking about the experience in the United States. But actually, Europe is ahead of the United States, in terms of clinical trial transparency, aren't they?


PETER DOSHI: Absolutely. Since November 2010, EMA has put in a new policy.


STEVE USDIN: EMA is the European Medicines Agency, the equivalent of FDA.


PETER DOSHI: Exactly. Right. And they put in a new policy since that date that allows anybody, don't have to be a good researcher with a good research question, anybody can write them by email, say I'd like these data from this drug. If they have it, they will give it to you. They will give it to you free of charge. And they'll give you PDFs. And it happens over the course of anywhere from a day to months. And they've put out 1.6 million pages of documents at this point already. And they're moving towards even greater transparency in the future, called prospective release of data, where you won't even need to request. Or perhaps there will be another mechanism of request.


STEVE USDIN: And Perry, it's interesting, one of the things that's come up in Europe. Some of the pharmaceutical companies have actually sued the EMA to try to prevent the release of information, because they're saying it's just their competitors that are trying to get the information, not independent scientists. What's your viewpoint about whether competitors should be able to get information, clinical trial information? Or should it be limited to scientific researchers?


PERRY NISEN: I think we need to make data available. There are investigators in companies. There are investigators in academia. I think the pursuit of answering questions, of verifying information, is important for society, is important for patients. And that's what we have to go to in the spirit of transparency and making information available. And to make these distinctions between what sponsors do, the scientists there, and what academics do, I think, is unreasonable. And I think it would be interesting to comment upon the extent to which the academic community is participating, similarly, in making information available in a transparent manner as well.


STEVE USDIN: And what are you alluding to there? Basically where you're going is saying that they're not doing it, right?


PERRY NISEN: I think it's challenging. And I think we're doing what we can -- we, where I work -- to get ahead on this and to move forward in that journey and that direction. And I think it would be interesting to hear from the two of you who have more exposure to that, your views on that regard.


MARTHA BRUMFIELD: Well, I can share that, in a number of our consortia where we are forming these pooled databases, sometimes it's much more difficult to get academic researchers and government researchers to share than it is the companies.


STEVE USDIN: And why do you think that is?


MARTHA BRUMFIELD: I don't know. I don't know what that means. Probably tied into what the incentives are for what they need to do with the data to get publications, to write grants, et cetera.


STEVE USDIN: Well next, we're going to get some final thoughts from our panel about clinical trial transparency.


NARRATOR: Now in its 21st year, visit for the most in depth biotech news and analysis. And visit for exclusive free content.




STEVE USDIN: We've been talking today about clinical trial transparency. In our remaining time, let's hear a few final thoughts from our panel. Peter, where do you think we are now? Where do you think we need to go?


PETER DOSHI: I think we're far ahead of where we were three years ago. I think GSK's public expressions of interest to take on this complex topic are great. I think we have a long way to go. And I think we need to stop considering this such a dangerous stuff and that the data are dangerous. And I think there's very few examples of where there's an actual public health harm caused by bad analyses of data.


STEVE USDIN: Mm-hm. Martha?


MARTHA BRUMFIELD: I think society, as a whole, is moving towards greater transparency. And given the rapid pace at which knowledge is generated and advances in medicine and technology, we're at a tipping point where we have to start sharing knowledge and data, if we're going to continue to learn scientifically, so that we can advance society.




PERRY NISEN: Making patient-level data, anonymized data is a journey. We're trying to do the right thing. This is a step for us to do so in a way that we respect the needs and expectations of patients, of science, of society. And as I said, it's a journey. And we want to get there together.


STEVE USDIN: And you say the expectations of patients and the needs of society. It seems to be one of the needs of society and certainly expectations of patients is that the data that they're contributing is going to have the maximum benefit. And it's going to lead to, hopefully, to personalized medicine. Is this kind of patient-level data going to be useful for third parties to get us there?


MARTHA BRUMFIELD: Well, my perspective is it will be. But I think we have to be good stewards of that data. The research subjects that agree to these studies and the incredible investment that the researchers make in acquiring this data, we have to be responsible about how we go about it. I think we're going to move incrementally. And we'll get there.


STEVE USDIN: And talking about responsibility, one of the concerns that's raised is that people will troll around in data and will raise false scares like vaccine deniers or something like that, and will scare people away from using medicines that they should be using. Peter, do you think that's a legitimate concern?


PETER DOSHI: I think the concern is legitimate. But I think empirically, if you've looked at it, and I've asked this at every conference I've gone to on this topic, what are the examples of public health harm that have been caused by sharing clinical trial data? And nobody's been able to give me an answer. Not a single example has come up. And this is really important.


PERRY NISEN: I think, if there are unverified claims of safety or unverified claims of benefit that are undertaken and stated in a non-scientific way, that does pose a risk to society. Eventually, the truth will come forward. Eventually, there will be verification by third parties of those data. We present -- we, sponsors --present their data, analyze their data, submit them to regulators. But it is a challenge.


STEVE USDIN: We're going to have to leave it at that. The truth will come out, when the data is out. That's this week's show.


I'd like to thank Perry Nisen of GSK, Peter Doshi from Johns Hopkins and Martha Brumfield of the Critical Path Institute. Remember to share your thoughts about today's show on Twitter. Join the conversation by using the hashtag #BioCenturyTV. I'm Steve Usdin. And thanks for watching.