One of the most talked-about studies presented at CROI 2008 was authored by Caroline Sabin and colleagues on behalf of the large, international D:A:D cohort.1 It focused on the impact of specific antiretrovirals on heart attack risk in HIV-infected patients. I had the chance to catch up with Jens Lundgren, M.D., Chief Physician and Director of the Copenhagen HIV Programme, where the D:A:D cohort is based. He'll summarize the data and discuss the ramifications of the results.
Bonnie Goldman: Dr. Lundgren, can you please summarize your study?
Jens Lundgren, M.D.
Jens Lundgren: We looked at the nucleoside analogs to see whether the nucleoside analogs were associated with an altered risk of a myocardial infarction, or heart attack. Before we started the analysis, our assumption was that, if anything, it would be the two thymidine analogs, zidovudine [AZT, Retrovir] and stavudine [d4T, Zerit], that would be the culprits. The analyses, as they turn out, show that that was not the case. Zidovudine and stavudine were, in fact, not at all associated with that, which really surprised us, because we thought that both those drugs had been associated with dyslipidemia, insulin resistance and diabetes, which are obviously intermediate to an excess risk of cardiovascular disease. So we thought that that should then be happening, but we didn't see that.
I think what we were even more surprised about was that, in proper diligence of analyzing this, we also tested some of the other drugs, assuming that they would, at least, be negative. But we found something different. We then spent four months trying to disentangle what this means, and what these signals are describing. At the moment, we have been able to show that this associated risk with the use of abacavir [ABC, Ziagen] is true, as long as you take the drug. If you go off the drug, that excess risk disappears, suggesting that it has something to do with the drug while it's in your body that may be associated with an altered risk of cardiovascular disease.
If these associations are biologically real, what are the biological explanations? We don't know, but it has something to do with the drug itself, as opposed to something where it sort of leaves permanent damage, also after it discontinues.
I think it's very important to underline that this study is an observational study. It's not a randomized trial. Therefore, the interpretation of the findings needs to be more cautious. Because an observational study cannot prove definitively causality between a drug and a given outcome like myocardial infarction. Clearly, people need to make sure that they appreciate that limitation of the study design when they are reviewing the data. We have, of course, done that, as well, and thought long and hard about whether we can identify reasons other than the drug itself that could explain these associations. I must say, we are not able to come up with a proper alternate explanation. That leaves, of course, the possibility that this could be a biologically true effect.
Bonnie Goldman: Isn't it true, though, that patients were put on either ddI [didanosine, Videx] or abacavir because they might have had risk factors for heart disease and it was assumed that the other drugs might have exacerbated these risks?
Jens Lundgren: We checked for that possibility quite carefully. Indeed, there was, at least in some patients; it appears that those who started using abacavir were at slightly higher risk of cardiovascular disease. But we checked for that in our analysis. What you do is: Statistically you then start to adjust for those factors, and if that was the explanation, you would then expect the association with the drug to disappear. It didn't. It didn't affect it at all.
Also, if it was that explanation, one would think that the risk shouldn't go away when you stop the drug, because then it's a subgroup of the patients with a given increased risk that you're studying. That risk will continue after they stop taking abacavir. However, that's not really consistent with what we are finding -- that it is this so-called channelling bias.
Of course we can't exclude the possibility that this is the explanation, if there were factors that we are not aware of that were leading to preventive use of these drugs. But I think we have done the best we can to understand and appreciate whether that's the explanation, and I must say that we haven't found support for that.
Bonnie Goldman: Could you describe the risk? The numbers are a little difficult to understand.
Jens Lundgren: That's the problem, right? When you translate complicated statistical analysis of something, it's hard to know what it means. The number itself is a 90% increased risk. So what does that mean? Well, it means that we can all predict our underlying cardiovascular risk. There's the Framingham equation, and there are other equations that have been developed to predict risk. These predictions essentially formulate which proportion of a given population that lives in the next 10 years will get a heart attack. Some people will have a very low risk; less than 1% will develop a heart attack in the next 10 years. Other populations will have a very high risk, a 20% or 30% risk.
This 90% increased risk needs to be placed into context with the underlying risk. If your underlying risk is low, say ½%, and then you add 90% on top of that, that's still less than 1%. So that's really a trivial problem, I would say, compared to all the other things that we have to worry about in our lives.
On the other hand, if you take a patient with an underlying risk of 20% or 30%, and then you add 90% to that, then it becomes a real deal. At least that's our take on it at the moment -- that in people with a high underlying risk of cardiovascular disease, that's where you at least should start to consider whether this is the right drug to use.
Bonnie Goldman: Could you characterize the 20% Framingham risk for somebody who doesn't check for those risk factors in a patient? Would it be someone who is overweight?
Jens Lundgren: I would say that, if you haven't checked it, it's about time that you start -- not only because of this study, but because I think that it is pivotal information to have when you are assessing preventive strategies. Because we know that there are many modifiable risk factors, like smoking, cholesterol levels, diabetes and hypertension, that you could do something about. So you need to do that.
But anyway, to answer your question directly: A patient who is a male of 45, 50 years of age, a patient who smokes, a patient who is diabetic, those are the sorts of patients that you would think about.
Bonnie Goldman: Tell me a little bit about the cohort. It consists of 33,000 patients?
Jens Lundgren: It's actually a collaboration of cohorts that has been ongoing since 1999. It was put together by a group of physicians, statisticians and other good people that were concerned about what would happen if we continue to use these drugs for a long period of time. So it's a rather unique collaboration internationally, in which we focus very much on quality, the quality of the data that we put into the database. Each of the cohorts does their own research. We then merge the data into one common database. And we cull it to issue other endpoints, to make sure that we get it right.
Bonnie Goldman: Are there other HIV medications that you found that increase the risk?
Jens Lundgren: That's an ongoing project. We published in the New England Journal of Medicine in April 2007 that the protease inhibitor drug class was associated with an increased risk.2 But that was studying a class of drugs, not individual drugs. So we are currently working on studying which drugs in the protease inhibitor class are potentially the explanation for this, but we haven't got there yet.
Bonnie Goldman: Are there any plans for a randomized study now?
Jens Lundgren: I would love to do a randomized trial. As a matter of fact, I would love to see many more randomized trials to assess these issues. Of course, the issue about randomized trials is that these have to be very large, and they need to be long-term. And therefore, randomized trials cost a lot of money. But if we can generate the funds to do these trials, I think that's absolutely the best way to go. Because that provides much more definitive information about these issues.
The problem of course is that there are many different permutations of drug combinations that you can start to compare. And if each one of them requires a 10,000-patient study, you're going to use billions of dollars. So although that's the sort of state-of-the-art and the preferred way of doing it, it may not all happen. As a matter of fact, there are very few randomized studies that include several thousand patients. The biggest one in recent times is the SMART study that included 5,400 patients.3 That's a little bit of our dilemma at the moment, since we want to get these drugs on the market. We want to get them quickly on the market, because they help people to survive. And we should not put too much of a blockage from a regulatory point of view, before they are released into the market.
But as a consequence, the studies that are done to prove effect, what you call biological endpoint studies, are not clinical endpoint studies. So you rely on these relatively small-sized studies to fully describe the benefits and the risks of the drugs. And clearly, that may or may not be the case. Certainly, our study, which we're now reporting, raises the question of whether we actually have good enough information from the randomized trials that are used to get the drugs on the market to really ensure that we fully understand the benefits and the risks of using these drugs. I think that's a fundamental dilemma that we have. I think it is an important area of discussion, given the fact that we have to use these drugs for the rest of a patient's life. This is not a short-term treatment we're talking about, of course. It's decades of using these drugs. We need to understand -- I need to understand which drugs are causing which problems in my patients so I can start to talk to my patients about these problems, and make sure that we address them as proactively as we can. I think that's the challenge that we are faced with.
Bonnie Goldman: I think this study, as well as other studies at the conference, are showing that you have to look at the full health history of a patient before even prescribing HIV medications, because of all these factors. Because certain drugs will have this problem. And particularly with a combination of three, it's kind of an art to figure out which regimen will not cause issues.
Jens Lundgren: But isn't that amazing? Five years ago, we were all upset about CD4 cell count and viral load, and we still should be, of course. But now we're starting to discuss underlying cardiovascular risk. It's a new mindset. I think it shows the maturity of the field. We're getting into a more mature understanding of the issues.
Of course, we all appreciate that these drugs are saving lives. They should be used, but they should be used appropriately. We are doing the right thing, irrespective of what drugs we're using, of course, but the challenge becomes: Can we do better? Back in '96, '97, you needed the randomized trial of 100 patients to show a benefit of a new drug. I mean, it was trivial. It was obvious that this works. Now we talk about several thousands of patients. If you look at cardiovascular research, they started off with a small number of studies, because the benefits were obvious. But now they also do very large studies. So I think that's where the field has to move, because that's where the research questions now reside, and these large studies are needed to answer those questions.
Bonnie Goldman: I'd like to further discuss the abacavir link. Was there any link seen with the duration of the use of abacavir?
Jens Lundgren: No. That was one of the surprises, because that's what we found when we studied the protease inhibitors -- that the longer you had been on, the bigger the problem became. But it was a very different signal that we found with abacavir. It appears that there is a fairly immediate onset of excess risk that is then maintained at a similar level, irrespective of how long a time you have actually used the drug -- not a higher and higher risk the longer you've been using it. It disappears, as I mentioned, once you go off the drug.
Bonnie Goldman: That was true with ddI, as well.
Jens Lundgren: It looks like that's true for ddI. I would say that if you're used to looking at large numbers and many results and so forth, there are some results that, although they are statistically significant, you are [less] comfortable about than others. That's not to [discredit] what we have been reporting, but I'm more comfortable about the abacavir signal than the didanosine signal. It's a stronger signal, and therefore, I'm more confident about it.
Bonnie Goldman: Tenofovir [TDF, Viread] and FTC [emtricitabine, Emtriva] were not compared in this, because not enough patients were taking it, I understand?
Jens Lundgren: There are a lot of patients now taking tenofovir, but that's not going to cut it. You also need to have been using the drugs for a reasonable period of time.
Bonnie Goldman: When you say "reasonable," what do you mean?
Jens Lundgren: That depends on how many patients are on it. The more patients on it, the shorter period of time you need until you can start to analyze it. So I suspect that we will, within the next year or two, be able to look into tenofovir, as well. But I don't think it is helpful to put out preliminary findings. That's certainly not how we have gone about reporting our findings. We want to report findings when we're confident in the results. And we're just not confident in the tenofovir results yet. There's only one way to fix that, and that is to wait a little bit, and have a little bit of patience. But I can guarantee you that we will report it when we feel confident, as we did with this.
Bonnie Goldman: I see you've put out a position statement by the D:A:D Steering Committee.4 Why did you decide to do that?
Jens Lundgren: I think I can reflect the mood of the steering committee quite precisely by saying that we essentially didn't believe it. We've tried over three months -- from October, when we first saw this, until December -- to disprove what we have found.
Bonnie Goldman: Meaning your results were surprising to everybody.
Jens Lundgren: Yes. We just couldn't believe it.
Bonnie Goldman: Because it was counterintuitive.
Jens Lundgren: Not counterintuitive. Because I wouldn't say that there's evidence that it would not be, but it was highly unexpected. And therefore, you start to then worry: Did we err? Did we do something wrong in our analysis? These are huge statistical programs that deal with these huge datasets. There are many things that can go wrong in that process.
But with due diligence -- three months -- we came up with the same signal. So we said, we're not going to sit on this anymore. We're going to put this out there. But we knew -- at least, we suspected -- that people would get word about this. An abstract is about two hundred, three hundred words. It's a very limited space in which to explain this. And of course, what we would hate to see would be that people reacted in what I would characterize as an irrational manner with regards to this report. For example, "Oh, I can't take these drugs anymore. I need to drop that drug." That would be terrible.
We felt, as a group, that we had a certain responsibility. That was why we put this position statement up. So that people at least have the opportunity to read and understand where we're coming from, what we have found, and our thinking around this -- including the limitations of the study, our take on this, as well as some advice on how to discuss the results with their treating physician.
Bonnie Goldman: What do you think the biggest misunderstanding will be about this study?
Jens Lundgren: I'm concerned about a couple of things, which we are obviously trying to prevent. The first thing is that people will be angry or frustrated. That would just be terrible.
Bonnie Goldman: Meaning, clinicians and patients will just say all HIV drugs are problematic.
Jens Lundgren: Yes, and being very pessimistic about the whole thing and so forth. That's not the point here. The point is that the scientific process means that you identify something. You report it to your colleagues in order to allow for discussions with them, to allow for colleagues to then look at their own dataset, and, whether or not they find something like this, to allow for a consensus at some point. It takes time, but the scientific process starts by us reporting a new and novel finding like this. People just need to understand that this process takes time to resolve itself. It will resolve itself. I can guarantee you, whether it takes a year or two years, it will. There will be studies done. There will be analyses done to try to understand what the biological mechanism is, if any, and so forth. This will stimulate a lot of research.
The purpose is not to make people's lives miserable, but to help in the longer run to understand what this all means.
The second concern would be people not appreciating relative and absolute risk. These are difficult and complex things to think about. With respect to this 90% increased risk, people said, "Wow, that's a big risk!" But, as we talked about earlier, it depends on your underlying risk. If your underlying risk is low, a 90% increased risk remains low.
Bonnie Goldman: Under 1%.
Jens Lundgren: In that given example, right. But we are not used to discussing these issues. I don't think we, as a physician and patient community, have spent enough time understanding the difference between relative risk and absolute risk, which depends on the individual. We need to have more discussions about that so people can understand that, and can intuitively react to that. I hope we will avoid that confusion.
Somebody was saying, "So, we need to switch all our patients away from abacavir?" I said, "Heavens, no. That would be the last thing I would be suggesting that people should do." But there are certain subgroups of patients -- those with high underlying risks -- that you want to focus on and consider carefully. Why is this patient on abacavir? Do they have any other drug options left?
I hope that people will reflect, ask questions, and have discussions. If people have other questions about the study, we're very happy to discuss that and provide clarity. Our purpose here is to try to provide a transparent reporting of our findings. We're reporting them in complete honesty through the process we used, and they are here to be discussed.
Bonnie Goldman: A few more questions about the study population: Was this an older population? What was the average age?
Jens Lundgren: When the patients came into the study a couple of years ago, they were around 38 to 40 years old. They have obviously grown older, as a consequence of the study continuing. So they are now 43, 44, 45 years old. But that's the typical HIV population.
Bonnie Goldman: Right. It's an older population that has a higher risk, in any case.
Jens Lundgren: The risk gradually grows as you age. Cardiovascular disease is an age-related risk. That's when you see the heart attacks: in your 50s, if you are a male. That risk is delayed by around 10 years if you're female. That's how cardiovascular risk behaves in the background population, in the general population, and certainly, it reacts in the same way in HIV patients.
Bonnie Goldman: So, if I'm a 25-year-old HIV-positive male, or I'm a clinician with lots of patients in their twenties ...
Jens Lundgren: This is irrelevant. There are many, many other issues in 25-year-olds that are much, much more important to focus on.
Bonnie Goldman: Than this possible risk.
Jens Lundgren: There are many, many other things that you need to put into consideration. But clearly, if you are talking to a 55-year-old male who smokes, you need to have a serious discussion with this person. Not only about abacavir, but also about not smoking. Maybe he should exercise more. Maybe he should be put on a statin to lower his cholesterol level. Maybe we should check his blood pressure? Is that OK? Is it elevated? Do we need to treat that? Those with HIV are affected by all the other diseases that the general population is affected by. We know that. If we think a little bit more carefully about this: hypertension is a problem; diabetes is a problem in an older population, especially if they are obese.
These factors behave in the same way if you are infected with HIV, as opposed to not infected. HIV is not protecting you from all these problems.
Bonnie Goldman: Are you making any changes with your patients?
Jens Lundgren: I will carefully talk about the results with certain patients and then come to a joint conclusion about possibly switching meds. But I would say that's a decision between me and my patient. I just need to make sure that the patient understands what the findings are.
Bonnie Goldman: But would you talk to only the older patients, and those with higher risks? Or would you also talk to the patients who are just infected?
Jens Lundgren: I would certainly focus my attention on the older patients. In the younger, I would probably be much more focused if they are smokers, for example. I mean, why don't you quit your smoking? That's good advice. I think it's very helpful, not only for your cardiovascular risk, but a variety of others. Fifty percent of the HIV population that we are studying are smokers. So that's a real health problem. That would probably be my main discussion point with the younger patient population. It would be the same with older patients, but I may want to add, "Why don't we switch this drug, if we have other viable options?"
Bonnie Goldman: I want you to tell our audience a little bit about this amazing cohort of 33,000 people, in many countries. How did it start?
Jens Lundgren: It started as a consequence of discussions that a small group of us had with the European Medicine Evaluation Agency, the MEA, which is the European equivalent of the FDA [U.S. Food and Drug Administration]. We were slightly concerned that, although it was great that all these drugs came about, in '96, '97, the randomized trials that were the basis of getting them on the market were very small, and very short-term. We were concerned that there was not a plan to monitor and assess what would happen in the longer run when you're using these drugs. We started to get interested in the metabolic problems, the dyslipidemia, etc., and there were a lot of reports coming out from '97 to '99. So we got together and discussed this and agreed that this is something that needs to be focused on.
Bonnie Goldman: Who got together?
Jens Lundgren: We were a group of people that got together in parallel with a discussion between me and Professor Ian Weller from University College, London, as well as the European Medicine Evaluation Agency. There was a quite famous meeting in March 1999 at the MEA in London, where this was put on the table as a problem that needed to get attention. Then we got all the cohorts together. We surveyed essentially the world for who was able and had an existing cohort that provided good quality information. Then we discussed these cohorts together.
There is a very good international network, among cohorts, particularly in Europe, but also in the United States and Australia. We agreed that we need to team up, because these are going to be relatively rare events. None of the cohorts could, by themselves, address this question.
So we teamed up as a group. Eleven cohorts got together, merged our respective databases, agreed on a common platform of data collection of the myocardial infarctions and other outcomes. Then it sort of evolved itself.
The primary focus to begin with was myocardial infarction, but now we're moving into liver disease; we're moving into non-AIDS-defining malignancies; we're moving into chronic kidney failure. These events that are of concern at the moment, but where we still don't really know exactly how big of a problem this is. So we are trying to develop this into a more comprehensive surveillance system, to make sure that we are able to identify emerging safety problems.
Bonnie Goldman: I understand the entire HIV-positive population of the Netherlands is included in this cohort.
Jens Lundgren: There are national cohorts with very good infrastructure and very good data quality in Holland. In Switzerland, for example, also, there's an extraordinarily strong group. There are other areas where not necessarily all patients are involved, but a fat chunk of the patients are involved, in various cohort studies. We needed a lot of patients. We could calculate that, before we even got the study started, we needed a lot of patients in order to have enough endpoints to start saying something about this. It wouldn't be enough to have a couple of thousand, so we needed a lot of patients.
But the trick here, if you want to get a study like this going and get it sustainable, is to make sure that the people who are involved in collecting the data are providing good quality data, and are able to follow the patients over a long period of time. This is a chronic disease; therefore, long-term follow-up of patients is vital to maintain the integrity of these studies. There are a lot of good groups out there, who are doing a lot of work. There are over 200 clinics around the world who are participating in D:A:D. So there are a huge number of people who are involved. If they see a patient with myocardial infarction, then they report that finding into the cohort that then reports it to us in Copenhagen, which is the coordinating center. It's a huge international network, which has now been in existence for eight years. I hope we can continue to do this sort of work, because I find this to be very important. We are obviously a big group that is in consensus on that, and want to focus on this as an area of research. Not the only one, but one area of research.
Bonnie Goldman: Very interesting. Dr. Lundgren, thank you very much.
Jens Lundgren: You're welcome.
This transcript has been lightly edited for clarity.
- Sabin C, Worm S, Weber R, et al, and the D:A:D Study Group. Do thymidine analogues, abacavir, didanosine and lamivudine contribute to the risk of myocardial infarction? The D:A:D study. In: Program and abstracts of the 15th Conference on Retroviruses and Opportunistic Infections; February 3-6, 2008; Boston, Mass. Abstract 957c.
- The D:A:D study group. Class of antiretroviral drugs and the risk of myocardial infarction. N Engl J Med. April 26, 2007;356(17):1723-1735.
- Strategies for Management of Antiretroviral Therapy (SMART) Study Group, El-Sadr WM, Lundgren JD, et al. CD4+ count-guided interruption of antiretroviral treatment. N Engl J Med. November 30, 2006;355(22):2283-2296.
- The D:A:D Steering Committee. Position statement on recent use of abacavir and didanosine, but not of thymidine analogues, is associated with risk of myocardial infarction. Copenhagen HIV Programme. February 4, 2008.
Download the audio file