Robert A. Harrington, MD: Hi. This is Bob Harrington from Stanford University, on theheart.org at Medscape Cardiology. Following this year's American College of Cardiology (ACC) meeting, there's been a lot of buzz on theheart.org forum around a clinical trial that was presented near the end of the ACC meeting, namely HEAT-PPCI, which stands for How Effective are Antithrombotic Therapies in Primary PCI [percutaneous coronary intervention]? This was a very interesting trial that was conducted at a single center in Liverpool in the United Kingdom. It took place over 1.5 to 2 years, from February 2012 to November 2013. They enrolled approximately 1800 patients from a single center, with an ST-segment elevation myocardial infarction (STEMI), who were due for a primary angioplasty, and they randomized them to receive unfractionated heparin or the direct thrombin inhibitor bivalirudin (as the base anticoagulant), and then allowed the investigators to use glycoprotein IIb/IIIa inhibitors -- in this case, abciximab -- as bailout therapy in either study arm. The primary endpoint was a 28-day composite of major adverse cardiac events (MACE). The primary safety endpoint was major bleeding using bleeding academic research consortium (BARC) definitions.
The results were really interesting: The group of patients treated with unfractionated heparin fared better in regard to MACE at 28 days than the group of patients receiving bivalirudin. Stent thrombosis was markedly reduced in the heparin group versus the bivalirudin arm. Interestingly (and divergent from prior studies), the bleeding results were comparable between the 2 arms. Needless to say, given that the data are different from previously reported data, this got people's attention. But maybe the thing that caught most people's attention, and has generated a lot of conversation, was the way that the trial was conducted. As I noted, it was done at a single center, but informed consent was delayed until after the procedure, when a detailed informed consent was undertaken in the patients who had survived. For patients who did not survive, their information was available through public death records in the UK. It's a very interesting way of doing a trial, a very interesting way of dealing with the consent issue in acute care, and I thought that the trial design and what it means to the practice of acute care cardiology would be a good one to discuss today.
I'm fortunate to have my good friend and colleague, Magnus Ohman, Professor of Medicine and Director of the Advanced Coronary Disease program at Duke University and Duke University Medical Center, joining me here today. Welcome, Magnus.
E. Magnus Ohman, MD: Thank you, Bob. It's great to be here on the other coast of the country.
Dr. Harrington: It's reminiscent of the many days you and I spent talking about clinical trial results. But in that situation we had our offices right next door to each other, so this will be a little bit different. Magnus, I've described the trial results; you and I certainly talked about this when we saw each other at the ACC meeting. This is a fascinating trial from a design perspective and from a results perspective, and it's a really important trial in terms of the methodology that was employed. Do you want to give your overall comments and then maybe we'll hone in on some of the specific issues?
Dr. Ohman: We almost never see single-center clinical trials with 2000 patients. That's remarkable, and the investigators should be congratulated for a very well-conducted trial. There are aspects of this trial that are different, but the bottom line from my vantage point is that it's terrific to see centers do things like this. Sometimes one wishes we could do more of this type of trial.
Dr. Harrington: That was certainly my initial reaction, Magnus. I, too, compliment the investigators. The lead investigator was Rod Stables, somebody well known in cardiology clinical trials. It was a well-designed, well-considered study. I was blown away by seeing almost 2000 patients enrolled consecutively from a single center.
Dr. Ohman: They practically enrolled every single patient. They had some minor exclusion criteria, but in general they enrolled just about every patient, which is a great way of finding out how a therapy works in your setting. Now, the larger question is, what's the implication of this trial vis à vis other trials that have been done with heparin versus bivalirudin? And let's leave out the HORIZONS-AMI trial for a little bit, because there have been some [similar] trials, at least in elective PCI, comparing heparin and bivalirudin. In fact, Bob, you were involved in some of the very earliest trials. We've been using bivalirudin for PCI for many years, not necessarily in STEMI but for non-STEMI or stable angina. It wasn't approved initially but it was studied very carefully. Those early trials were interesting because, remember, we didn't have glycoprotein IIb/IIIa receptor blockers with the first few trials.
Dr. Harrington: The first trials, going back to the Bittl trial, were without IIb/IIIa's; it was in the very early days of intense study of antithrombotic therapy in the cath lab.
Dr. Ohman: Right, but even in those early trials there was a signal for lowering bleeding [with bivalirudin]. In fact, there was a meta-analysis published a couple years ago by Olivier Bertrand and others; they took all of these early small trials and other trials from all over the world, and they also found a consistent message of lowering bleeding. They didn't find a consistent message in reduction of ischemic events, but it was close to the line on the forest plot, if you understand what I mean, and that was because the trials were small and the confidence interval was very wide. What's interesting with this trial is that the bleeding signal is not what you'd expect. At least that's my take.
Dr. Harrington: That was my initial take as well. Let's not focus on HORIZONS-AMI, but if we look at many of the earlier trials with bivalirudin versus a strategy of heparin plus IIb/IIIa, there had been the suggestion that there's an early ischemic benefit with heparin plus the potent platelet inhibitor. Over time, some of that is attenuated, but the big difference comes in the bleeding results. Then we've got the HORIZONS-AMI trial, which has very different results.
Dr. Ohman: Yes, but once again, that's against a glycoprotein IIb/IIIa receptor blocker. One of the interesting aspects of this trial is that it's a STEMI trial, so although patients were "pre-loaded," we don't know how much of the drug (either clopidogrel, prasugrel, or ticagrelor) was on board.
Dr. Harrington: There was a lot of ticagrelor used in this study.
Dr. Ohman: Yes. But generally speaking, for these very rapid primary PCI trials, we don't know how much of this therapy gets absorbed. You probably saw this paper published a couple of weeks ago in JACC from a group who looked at the absorption of clopidogrel in the acute setting (using platelet function testing) and noticed that they could almost correlate it to the amount of morphine given. We know that morphine reduces gastric uptake or gastric mobility and therefore reduces [clopidogrel] uptake. This trial is interesting from a lot of different perspectives, in that it sheds light on an area that we thought we had resolved, but like most other things we don't really resolve things. We get to a point where we have some interesting information that has us asking, "Okay, why do we see this?"
Dr. Harrington: Right, and how should we think about it and how do we interpret this in light of previous evidence? Are there issues here that we need to pull apart to explain why it might be different? And then, ultimately, the question always asked is, what do we do with the data? Let's start with the methods. You and I were both impressed and congratulate the investigators for their commitment to pulling this off at a single center. But there's another side to single-center clinical trials, which is the generalizability. Is the generalizability better here, Magnus, because they enrolled such a big sample of patients?
Dr. Ohman: It is, to some extent. They measured activated clotting time (ACT) in both arms and that was lower [than in prior studies]. They used a very different methodology for measuring ACT in this study compared with other studies. They used a device that I don't think is available in the US, but I could be wrong about that. [Editor's Note: The Actalyke® MAX-ACT (Helena Laboratories) point-of-care ACT testing system used in the trial is FDA-cleared for use in the United States]. But that is one of the things that affects generalizability -- namely, if you have a system that is not widely used at other centers, what does that mean? Does that mean the same thing or does it mean that there's something here that may not be quite right?
What was also interesting (and I didn't see this in the slides) was that they mentioned that they had to re-bolus some patients. Now, generally speaking, at least at Duke, and I'm sure it's the same at Stanford, it's quite rare for us to re-bolus patients with bivalirudin.
Dr. Harrington: Yes. In the clinical trials it was very unusual. When we did the early clinical trials that were double-blinded (like REPLACE), repeat dosing was unusual.
Dr. Ohman: Maybe there is something here; I don't know. I don't want to say that this is the reason why the trial is flawed, but it gets back to the fact that if few centers use this methodology for examining the effectiveness of the anticoagulation, then it's hard to extrapolate the findings to your own setting if you don't use the same device. On the heparin side, the ACT levels were about 200, 220, thereabouts. That's kind of low if you're using single heparin.
In David Moliterno's paper many years ago, he pushed that by over 50, and there's been some really beautiful analysis done by David and others from Cleveland Clinic, looking at the optimal range for heparin, and it's over 250. Is it the device or are they underdosed? I really don't know.
Dr. Harrington: Certainly in modern angioplasty, including for STEMI, the speed with which these procedures are done is very rapid. Once you get access, once you get a wire across the lesion, things go very quickly. A lot of those studies that looked at the relationship between ACT and outcome were in an era when the procedures took a lot longer. If you get in there and you get out pretty quickly, how much antithrombotic effect do you need? Do you need to have the level that we thought was appropriate years ago, when we were trying to prevent things like abrupt closure? Now we're in and out pretty quickly.
Dr. Ohman: Most of the procedures in this trial lasted less than 25 minutes, as I recall.
Dr. Harrington: If you look at some of the recent PCI trials, like the CHAMPION studies that both of us were involved with, those procedures had a median time of 20 minutes. Obviously it's a little bit different with STEMI, but the point is that these things are quick.
Let's turn our attention to the consent issue, because to me that is one of the most methodologically interesting things that they did. They got countrywide consent through the UK ethics boards, and then they delayed actual consent from the individual patient until after the procedure. People have talked about this, and in some settings -- CPR, for example, and trauma -- the notion of community consent is something that's been accepted in the clinical research community, but in acute MI research, we've largely still relied upon individual patient consent before the procedure. HEAT-PPCI is very different, and I've got to tell you, I liked it a lot. I think it was innovative, it was bold, they appear to have done it very carefully, and if you want to do acute MI research, isn't this the way to go?
Dr. Ohman: Well, here I differ, Bob, because I served on the Duke internal review board (IRB) for nearly a decade, and yes, they did all the right things if it was resuscitation research, if it was out-of-hospital cardiac arrest research, for example. In those cases, you go through a central IRB that is a community forum and actually discusses the trial in the community.
Dr. Harrington: So you have no problem with the ethics of the trial?
Dr. Ohman: I have an ethical dilemma to some extent, but I'll come back to that in a second. First let me go back in time, because it's always helpful to look back. The ISIS-1 and ISIS-2 trials were both large, randomized trials assessing mortality and were both run out of Oxford.[10,11] They were done with verbal consent; you document in the chart that the patient had verbally consented to the trial. Of course, that wouldn't hold much today, but that's what was done back then, and I think some countries may have had no informed consent. In today's society, if you can get consent (and I think you can get consent in AMI) I think you should. You can't get it in resuscitation research with an unconscious patient -- I fully accept that -- and I understand the leeway (what does the word "informed" mean here). But if I'm about to take you, the patient, to the cath lab to do a cardiac catheterization, there are some risks and some benefits. That's what we say to our patients right now; they sign a piece of paper and we do the procedure. That's the standard of care. I do not believe that you couldn't add an informed consent to that part, telling the patient the same thing, making them understand that there are 2 options for anticoagulant and we don't really know which is best, and here is the study outline, and then have them sign. In many settings you can do an abbreviated informed consent that is just 1 page.
Dr. Harrington: Which has been utilized in acute MI research for a long time and we both have employed it. But, Magnus, let me take a different view here. In the practice of medicine, doctors do all sorts of things, sometimes without a lot of rhyme or reason to it. It's the reduction to practice patent, it's what they feel comfortable with, it's what they've done before, and so the practice of medicine -- in many ways, as you and I know -- is fairly random. When you're asking a question around commonly used drugs that the doctors are going to use anyway, isn't it, in many ways, more ethical to try to learn something from it?
In fact, there's been a great series of papers coming out from the Johns Hopkins group, published first in a Hastings Center report, that asks, "Why wouldn't you incorporate learning into the system in a more formal way than we currently do?" Maybe HEAT-PPCI is the type of clinical trial that fits the construct of the so-called learning healthcare system, where you take 2 things that are commonly done, and instead of letting the doctors pick, you randomly assign. So then you have the scientific ability to compare A versus B.
Dr. Ohman: I think that's reasonable on one level. On another level, if you're a patient, you expect us -- you and me as doctors -- to pick the treatment that we think is best for that patient. Now, we could actually say, "Mr. Jones, I'm clueless about what to do." But that's not how we practice medicine today. We do make choices, and they might be random for all that we know, and I think that there might even be data to suggest that. But we're supposed to take the best tack for the patient in that individual setting. So my thinking is that it's a little bit of a leap to say, "Well, I'm going to do this procedure, primary PCI, but I have no clue what to do."
Dr. Harrington: I'm not saying it's quite that stark. I'm saying that in this setting, you have drugs that are widely used and there is random variation in how people use them, so why not try to learn something in the system? I actually really like the idea of delayed informed consent in this particular setting for a trial like this. Where I wouldn't be in favor of it is if there were a brand-new procedure or drug and nobody knew how it worked; I would have a different perspective then. But in a setting like this, where you pretty much know the characteristics of the 2 therapies but you want to see how they perform head-to-head, I like the approach that they took. To me, it fits with what you and I both said at the outset: We like the idea of a center enrolling all of their patients into a clinical trial. Why not make it easy for them?
Dr. Ohman: I'm all about making it easier, but I still think we should be honest when we are experimenting -- that's where I come down on it. It differs from your point of view, and I think the world is split on this.
Dr. Harrington: That's why it's good to have the discussion, because I have a sense that a lot of medical practice in some way is experimental, but we don't learn from it because we don't have an ability to make inferences when things are not randomly assigned. It is challenging, because as you and I have done a lot of clinical research, we certainly don't want to return to an era when human experiments were done in secret on patients and other vulnerable individuals, when trust was misused, etc. I'm certainly by no means recommending a return to that, but this is an opportunity to think differently about how we might do clinical investigation in some settings.
Dr. Ohman: And the aspect that is really key here is that, clearly in this setting, in Liverpool, there was equipoise because they had to have that in order to even do the trial.
Dr. Harrington: Absolutely. In order to randomize the patients, there had to be equipoise.
Dr. Ohman: Right, so when we have equipoise we should try to answer those questions. They set out to do that, and they have their answer for their practice of medicine and their setting -- there's no doubt about that. Now we're back to generalizability; how can we generalize this to other settings? That's harder to do. That's why a good multicenter clinical trial is so powerful, and that actually defines guidelines in this country, pretty much. One aspect is approval for the FDA, but the other part of research trials is that they inform the guidelines, so that we pick the best treatment for most of our patients. I wish more people would do this.
Dr. Harrington: This has been exactly the kind of discussion that I wanted to have, Magnus, and I appreciate you joining me. The topic's a tough one, and I hope we get some conversation going in the broader community about how one thinks about studies like this. I think we both agree that these investigators should be congratulated for undertaking that. We're both intrigued by the outcome results, which are different from some of the previous studies'. And we're both intrigued but we differ in our interpretation of the way that the trial was done from a consent perspective. Is that a fair summary?
Dr. Ohman: I think that's a great summary.
Dr. Harrington: Magnus, thanks for joining me here on theheart.org and Medscape Cardiology. This has been a discussion on the HEAT-PPCI study out of Liverpool, UK. My guest tonight has been Magnus Ohman, Professor of Medicine at Duke University. Magnus, thanks again for joining us.
Dr. Ohman: Thank you, Bob. It was fun.
Cite this: HEAT-PPCI: Two Different Points of View - Medscape - Apr 11, 2014.