Shelley Wood; Jessica L. Mega, MD, MPH; E. Magnus Ohman, MD


March 20, 2014

This feature requires the newest version of Flash. You can download it here.

Negative or Neutral Clinical Trials

Shelley Wood: Hi. I am Shelley Wood, Managing Editor of heartwire, and my guests today are Dr. Mega and Dr. Ohman.

Our topic is negative or neutral trials that have positive (or at least interesting) secondary analyses. We have seen quite a few of these at recent major cardiology meetings, and it leaves not just the journalists but practicing physicians with questions about what to take from those trials that is meaningful in day-to-day practice. Both of you have some examples of trials where you thought this was important. Dr. Mega?

Jessica L. Mega, MD: This is critical as we start to look at these studies. There is a huge investment in these studies, and obviously as clinical trialists, we would agree that the top-line message has to be respected -- so when you do secondary analyses, it's in the context of a neutral trial.

However, that doesn't mean that there are not important biological observations, and one that stood out for me at this meeting was the TOPCAT trial.[1] This is a study looking at patients with heart failure and preserved ejection fraction. Spironolactone overall did not reduce the primary endpoint, but if you looked at patients who were biomarker-positive -- particularly those who had myocardial stretch and elevated B-type natriuretic peptide (BNP) levels -- in that group, there was more of a signal. We need to be thinking about the types of patients who come into the trial, and there is more to be said for these studies.

E. Magnus Ohman, MD: There is another twist to this. Not only is it the selection of a certain subgroup, but the secondary endpoints are very important. Using TOPCAT as an example, there was a fairly significant reduction in rehospitalization rates. Although the primary endpoint was not achieved, there was improvement in an area that all of us would like to consider: keeping patients out of the hospital.

You have to use a little bit of judgment. The challenge to us as academic physicians is how to use this to inform our guidelines for our practice. Typically for guidelines, we respect only the primary results, because that is scientifically the most correct way to do it, but as a physician who practices, you have to remember that you are doing more than treating to guidelines. You are treating an individual patient, so you have to weigh these things out.

In many areas, the analyses from trials with important secondary endpoints can be just as important. We have to be careful not to put too much weight on them, but as an individual physician, you should use them freely and I'm sure you do in your practice, Jessica.

Dr. Mega: I do it all the time. Another example that stood out is the CURRENT-OASIS 7 trial.[2] Overall, looking at double-dose clopidogrel vs standard dose vs aspirin, the investigators showed that if patients who were going to undergo an invasive strategy or percutaneous coronary intervention, there may have been a benefit.

We have to take these types of things into account when we think about the individual patient in front of us, and so I look at the totality of the data. Not only that, we also have to look at who is being in enrolled in these trials to begin with, and so generalizability becomes very important.

The Power of the Composite Endpoint

Ms. Wood: It seems that as we get better at treating patients, it becomes harder to show a benefit with a new therapy or a new strategy, so we see more of these composite primary endpoints. Others have told me that the problem is that you have a great deal of difficulty showing a benefit with a composite endpoint. One part of that endpoint might be very important, but then your trial isn't powered to show a benefit for that particular item within the endpoint.

How do you tease that out? You might not get the funding to do another trial just looking at that one endpoint. I am not sure what the solution is.

Dr. Ohman: I don't know that we will have an answer for you today, but what we have to remember is that when we do these trials in the first place, we should do adequately powered trials. You are absolutely right. Many trials are not adequately powered to show meaningful differences, and that is a shame, but it's also a reflection of fiscal restraint and the difficulties in obtaining funding. Sometimes having some data is better than none whatsoever, and that is where the secondary analysis could be of some importance, even though it may not direct all approaches to this particular disease entity.

Dr. Mega: This concept brings up the question of what is a composite endpoint? Obviously, death is going to carry a certain amount of weight, but if we were to poll the 3 of us, how much is a myocardial infarction worth? How much is a stroke worth? What about a stroke with a certain Rankin scale, and then what about safety? With a lot of the antithrombotic agents, how do we balance the bleeding? Composite end points are necessary. It's the only way with these trials that we are going to end up having adequate power, but we have to think about how we rate each of these elements.

The other piece is that sometimes when we go into a trial, we do our best to try to figure out what the composite should be in terms of what we care about clinically, and also how we think this drug is going to modify these outcomes. Sometimes we don't have the luxury of knowing, so you may have a composite that includes the 3 things we talked about and maybe there is a reduction in revascularization. Only through these trials do we understand more about the biology.

Dr. Ohman: The best example that I was involved with recently is the TRILOGY ACS (Acute Coronary Syndromes) trial.[3] It was a trial of the medically managed ACS population, which has never been studied directly before, and we had very surprising findings. Now that we have gone back, we identified a group of patients who were undergoing cardiac catheterization before randomization, so there was no bias and essentially we found a treatment effect consistent with many of the other trials in that particular subgroup. We won't rewrite the history of the trial, but it helps us for the future. It actually tells us the reverse: In medically managed patients not undergoing revascularization, we do not know what to do. We have no clear evidence, so that will hopefully get us to do trials in that setting.

Ms. Wood: We always hear the phrase "hypothesis-generating," but I am not sure how often the rationale for looking into something further can be supported when the major trial has come out negative. The appetite for funding a trial for the secondary endpoint starts to wane.

Secondary End Points and Spin

Dr. Mega: Not to mention the fact that some of these secondary analyses don't have the same glitter. When we talk about publication and publication bias, in recent months there was a very nice article in the New England Journal of Medicine[4] highlighting the fact that sometimes it's more challenging to get some of these neutral messages, or even secondary messages, out there. I am sure you have seen this as well.

Ms. Wood: Those were all National Institutes of Health-sponsored trials, and there was no difference between the publication of positive and negative trials if they were large, well-powered trials answering a question that everybody wants to know the answer to. However, for a smaller, non-hard clinical endpoint trial, a negative trial was less likely to get published.

Dr. Ohman: There are certain comparative effectiveness analyses that we would like to do. For example, of all the new oral anticoagulants, which one is the best? We don't know, so somebody outside of industry has to fund such a project. Certain areas will always be in what I will call the "underfunded mandate" -- in other words, we will get some, but not enough, information, and we will have to do the best we can with that information.

Ms. Wood: The final issue that I would love your opinion on is the question of "spin." When certain trials come out, they have a neutral primary endpoint, but a secondary endpoint is positive. The journalists get a press release. In most cases, the press release says, "Hospitalizations were reduced, but there was no effect on mortality," and unless you go back to to see what the primary prespecified endpoints and secondary endpoints were, you might (especially with the pressure on the media to get a story done as quickly as possible) end up with stories out there in the public domain that say, "Hospitalizations were reduced."

Some of those secondary endpoints are meaningful to physicians treating patients day to day, but how do you separate that from the spin? Companies that pay a lot of money for these studies are trying to milk a little positivity out of them. What do we do?

Dr. Ohman: Shelley, I would have to give you credit among many other medical writers and reporters, because I saw a lot more spin 15 years ago. Everybody has become more savvy about the fact that you are not going to get away with it. You have to own up to a certain set of standards, and the standards are by and large not driven by guidelines. The primary endpoint matters very much, because if it's not in the primary endpoint, it's unlikely to change guidelines in a major way.

In a way, we have tightened that up, but I still see it as do you, and it is problematic -- but remember that the investigators are enthusiastic. They spent 6 or 7 years doing this study. It's human nature. We try to make this a positive story. It's awfully hard to sit down and say, "Well, we did this project for 7 years with absolutely no benefit." It's not in our nature, so I understand the positivity and I would like to encourage it, but not oversell it.

Dr. Mega: I would like to echo that and say that you have done an amazing job of being balanced, and my personal take is that the truth always works. You start with a primary endpoint, and then in that context explore the secondary endpoints. Sometimes we run into time pressure. You come into a meeting with the top-line results, and there may be all these other interesting nuances, and so to encourage ourselves and the general population of physicians, we sometimes have to be a little bit patient. There will be headline and then maybe a second story.

Ms. Wood: Certainly has focused on the big story -- the top-line results -- only to realize 4 years later that the investigators went back and did aother trial with that secondary endpoint. I'm sure it's gratifying for investigators and is a reminder to us to pay attention to those secondary signals, because we might see them again. That's the hope.

Thanks to both of you for speaking to me today. It is an interesting topic.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.