Bob Leahy: Hello Rupert. Thank you for talking to PositivelIte.com. Rupert, I really was fascinated by your recent webinar, one of the ACT community forum series. (If people want to look at it, and I recommend you do, go here and skip to slide 150). You gave it the title “I’m on antivirals and undetectable: what about safe sex?” Tell us why you chose this title.
Dr. Rupet Kaul: Four of us were speaking in the ACT session, and between us we touched on several issues that people face around starting HIV treatment – when to start treatment, whether treatment can eradicate the virus, what treatment means in terms of HIV transmission, etc. Since the talk was a day or two after the Supreme Court decision, which had highlighted issues regarding antivirals and transmission, I thought this was a good time to explain some of the science that was considered in making that decision.
I like how, right at the start, you make a distinction between “legally” safe sex (having a low viral load, using a condom) and “safe” from a practical point of view. I want to return to that issue in a minute, but do you think there’s room for confusion here?
There is tremendous room for confusion here: when it comes to sex and the risk of HIV transmission, “safe” means very different things to different people. There is no dictionary definition that says a risk less than X is “safe” and a risk more than X is not. The fact that “safe” is subjective and open to interpretation is the fundamental problem facing the community and, of course, the legal system. In some ways the job of a researcher like me is much easier – we know what we know, we know what we don’t know, and we communicate those things as best we can to the relevant stakeholders. Taking that information and converting it into legal guidelines, or into the reality of a person’s sex life, is a much trickier thing.
OK. I understand. I’d like to look at HPTN 052. Now many of us will be familiar with this study – CATIE calls a it a game-changer, but for those who aren’t, can you give us those findings in a nutshell?
Sure. This was a great study that provided us with some really clear and incontrovertible numbers, although it still leaves ample room for debate on some issues. HPTN 052 was a clinical trial performed in four continents, with a slight majority of participants from Africa. The study enrolled heterosexual HIV-infected men and women who had a fairly high CD4 count, so that under the current clinical care guidelines they would not need to start treatment for a while, and these people all had a stable, HIV-uninfected sexual partner (usually a spouse). In addition to frequent couples counselling about how to reduce HIV transmission, participants were randomly assigned to either start antivirals early or to be monitored for a while: the “delayed antivirals” group only started meds if their CD4 count started to fall. Whenever a person became HIV infected the virus was checked to make sure that it came from their partner in the trial, and not from a different partner outside the marriage. The first analysis was done after 28 people had become HIV infected. The team saw that only 1/28 of these people had a partner in the “early antivirals” group, while 27/28 had a partner in the “delayed antivirals” group. Based on this difference, they concluded that early antivirals made transmission 96% less likely, and for ethical reasons they stopped the trial early so that all participants could be offered early antivirals. The one person who became infected in the “early antivirals” group did so early, within the first month of being on treatment. This is similar to an earlier observational study that showed a 92% reduction [Donnell, Lancet 2010], again with only one transmission event within 3 months of starting antivirals. There was also an earlier study that “pooled” information from several small observational studies and showed a 92% reduction in transmission [Attia, AIDS 2009]: this was based on 5 transmissions from people on antivirals in over 5,000 couples, although the timing in relation to starting treatment was not known.
Now that 96% reduction in the risk of transmission is highly significant. I’m curious though why it’s only 96%. I mean, there was only one infection in the group on treatment and that person had only just started on treatment before his female partner was infected. In those circumstances, viral suppression seems unlikely, doesn’t it? I guess I’m suggesting that if the research was more rigorous and eliminated people who’d just started treatment, we could well have seen a 100% reduction in transmissions!
There are two parts to my answer here. First, it is true that if we exclude that person who transmitted HIV very close to the time that he started antivirals, there would be no cases in that arm of the trial. It might seem logical to exclude that person from the analysis, but that is not how the study was set up. Changing the numbers afterwards to exclude people based on things like that is a real “no no” in research: you have to present the results the way you said you would before you started the trial, and then people like us can assess these kind of nuances for ourselves.
Let’s say we do decide to exclude that person, though. An important thing that we need to ask is whether, since the first 27 transmissions happened in the “delayed antivirals” group, that means that no transmissions could happen even if we waited until many more people had been infected (which the trial did not do, because it would have been unethical). What I mean is, the trial was set up to show whether antivirals can reduce HIV transmission, and it absolutely proved that they can. What it was not set up to show was whether transmission is impossible when you are taking antivirals. An easy analogy is tossing a coin. If you get 27 “heads” in a row then you have proved that heads are much more likely, and we would all be happy to bet on the next result being “heads”. But does that mean that it is impossible to get tails with this coin? That is a different question, and you would need to toss it a lot more times to prove that. The good news is that, even though early antivirals have now been offered to all participants, the HPTN 052 study will continue to follow all couples until 2015. This will allow the study to give us a much more precise measure of exactly how (un)likely transmission is to occur when somebody is taking antivirals.
It’s been said many times over that HPTN 052 results really couldn’t be extrapolated to gay/bi men because this was a study of heterosexual couples and that the risk of anal sex just isn’t comparable with that of vaginal sex. But this raises two questions in my mind. Firstly, heterosexual couples have anal sex too. HPTN 052 must surely have included (straight) couples who were having anal sex and the risk reduction was still 96%. So I’m not getting why this study, we are repeatedly told, doesn’t reflect the rigours of anal sex, the nature of the rectum lining, etc. etc., so it means no good news for gay men. Can you comment on that?
Plenty of heterosexual couples have anal sex. It is reported by couples a lot less commonly in Africa, but even if you think those reports may not be accurate, it’s pretty easy to show that anal sex was probably not a big factor in HIV transmissions within HPTN 052. We know that all the 28 HIV transmissions happened between a woman and man within the couples, because they matched the virus to that of the infected partner. And we know that when people have anal sex, the receptive partner is at a much higher risk than the insertive partner (about ten times more). So, since the receptive partner in a heterosexual couple is always the woman, if anal sex was common we would expect to see more transmission from men to women in the study. However, it was actually the other way round, with slightly more transmission from women to men. This says to me that there was probably not much transmission through anal sex in the HPTN 052 study.
OK. Got it. My second question. We know that the risk of insertive anal sex is more or less the same as insertive vaginal sex. It’s in the one in 500 to one in 1,000 range, without antiretrovirals or condoms complicating the picture. So . . . for many gay men – those who practice only insertive sex, the HPTN 052 results are extremely relevant, no? Why would we suggest those results don’t have applicability to gay men, period?
The risks that you are talking about are right on: the risk of transmission during unprotected anal sex is about 1/20 – 1/300 for receptive anal sex, and 1/500 – 1/1000 or so for insertive anal sex. Although we don’t have specific data about anal sex from the studies to date, my best guess is that if you are an HIV uninfected man who only practices insertive anal sex (studies say that is about 10% or less of gay men) then if your partner goes on antivirals this will have the same protective benefit for you as was seen in the HPTN 052 study.
However, in terms of the recent Supreme Court “prescription” about disclosure, remember that for a gay man with HIV who only practices insertive anal sex, this means that (with him) all his partners are only practicing receptive anal sex. Receptive anal sex is riskier in general, and is also where we have less information about the protective effect of antivirals. My best guess is that we would probably see the same kind of reduction – roughly twenty-fold less, or a risk of 1/2000 or so – but different people have theoretical reasons about why the protection might be less and why it might be more. Again, the good news is that at least two observational studies in Europe and Australia are enrolling HIV-discordant male-male couples to specifically assess the effect of antivirals on transmission in gay men.
Can we try to pin this down? If someone is on antiretrovirals, say for two years, virus fully suppressed and without STI’s, the risk of transmission through insertive anal sex is about one in what range?
That is a good question that nobody has a precise answer to, yet. We hope that it will come from the several studies that are underway and that I mentioned earlier. Specifically, it will be interesting to see whether the amount of reduction is any lower than the 96% reduction seen in HPTN 052.
And that risk is sufficient for ASOs to suggest that condoms must always be used? (I’m not talking here from a legal perspective.)
I hope that ASOs consider the legal perspective in the advice that they give out! However, what level of risk is “enough risk” for ASOs or Supreme Courts to make recommendations one way or another is the fundamental question here. It is important to realize that this is not a “science question” – instead, this is a very subjective question that has many different answers for many different people.
Right, but I think people’s approach to risk is different from HIV prevention folk. Prevention people – and even the Supreme Court for that matter – seem to glom on to exceptions to the rule. In other words, they have to think about what is the absolute worst that can happen. But we don’t usually think like that. That’s why we cross the road. We tend to think what is the likely outcome of any risk behaviour. And that gap between them (the “exceptionalists”) and us (the “realists”) is difficult to bridge, no?
Sometimes it seems like there is a “them” and an “us” in the debate, but I am not sure that is really the case. In fact, there is a tremendous spectrum across people in their answer to the most basic question at issue here – which is “how much risk is too much risk?” No matter what the numbers say – and hopefully we have done a good job of going through the available numbers for your readers – this is where the science falls short.
I think this is really well illustrated by an anecdote that I told in the question period after that ACT forum. I often see people in my clinical practice who may have been exposed to HIV in one setting or another – sex, needle stick injury, whatever. We talk about what happened, come up with a “best guess” as to the risk of transmission, and then talk about whether a short course of antiviral medications is indicated to reduce that risk. In one case we arrived at a risk of 1/100 of acquiring HIV, and this person said that this risk was much lower than they had expected, and that they did not think it was significant enough to take antivirals for month. On the other hand, in another case we arrived at a number of 1/1,000,000 (one in a million), and the person was adamant that this was very significant, and that they wanted to take the antivirals even though they were not covered and it would cost them well over $1,000 out of pocket. To my mind, there can be no clearer indication that getting everybody to agree on what constitutes a “significant” risk of HIV transmission is just an impossible job.
I want to talk about semen now, Rupert. Because you’ve said that viral load in the semen is a key predictor of HIV transmission. I think we’ve led people to believe, and I talked about it here, that if you have an undetectable viral load in the blood this means nothing when it comes to what’s in your semen. But it’s not that simple, is it. We do have a good idea of the risk.
Both the level of HIV in the blood and the level of HIV in the semen are each predictors of transmission risk – and for each one, a higher level means a higher risk of transmission. There is no validated “threshold level” for semen HIV and transmission, especially since semen virus levels go up and down much more than blood viral levels, but we think that a level over 5,000 RNA copies/ml is most relevant in terms of transmission risk. Our own studies have followed men for six months around when they start antivirals, with semen and blood samples collected every month. What we saw was that virus levels in the blood and semen generally did the same thing, and that was to become undetectable, although virus could occasionally be found in semen despite a person being undetectable in blood. In fact, the reduction in the frequency of “high level” semen virus was 92%, which is very similar to the 92-96% reduction in actual HIV transmission that was seen in the clinical studies. We have done much less work in men who have been taking antivirals for longer periods, but it seems that the frequency of semen virus gets less and less the longer that you have been on treatment.
OK, let me ask you this. I’m undetectable. No STIs. What are the chances that I have virus in my semen AND it’s capable of infecting someone else? (I think you call this a high level of virus, i.e. over 5,000 copies.)
There are two caveats here. First, we don’t know as much about semen levels in men who have been on treatment for a long time. Second, it’s important to say that we cannot be certain how capable any semen sample is of infecting another person “in the real world”. What we can say is that, after relatively short term antiviral treatment in men with no STI and an undetectable blood viral load, we saw high levels of virus at roughly one in twenty-five visits. Preliminary work suggests that after a longer duration of antiviral treatment, the chance of finding virus in semen is likely to be lower than this.
And that risk would change much if I had an STI?
We know that when untreated men in Malawi had gonorrhea, the level of HIV in their semen went up about 8-fold, and that it came down again after the gonorrhea was treated. There are not many studies looking at the impact of STIs on HIV levels in the semen of men on antivirals and who are undetectable in blood. However, one recent study did show that three-quarters of such men with an active STI had detectable virus in semen (although numbers were small: this was six men out of eight), and concern over such increases is the reason for the lack of STIs to be explicitly mentioned in the Supreme Court decision and the prior “Swiss Guidelines”. Also, it is worth pointing out that STIs are generally infrequent within married couples, and so these low rates might have contributed to the success of the HPTN 052 study.
So it depends too in part on how long a person is on effective treatment whether it’s likely that virus will be found in the semen? Longer than two years on treatment and you haven’t found any?
Our preliminary studies do show that the chance of finding virus in semen is lower after longer term treatment, but so far we have enrolled less than twenty men in those studies and so we can’t say how much lower with much accuracy.
So this is suggesting to me some very low risk scenarios here. I’m thinking of a poz man on antiretrovirals for two years, adherent say, whose blood viral load is undetectable, who has no STIs, who is engaging in unprotected insertive anal sex? Now is it heresy to suggest that may be very low risk?
I don’t think that anybody would argue that the risk of transmission will be much lower than if that poz man was not on antivirals, and that is a great thing from a public health perspective. However, there are several things that we don’t know in this scenario, and to my mind two are the most important. The first is exactly how much lower the risk will be, compared to the roughly 1/100 risk without antivirals, and the second is how much lower it would need to be for his partner to agree. As I said earlier, my guess is that the risk without condoms will fall by twenty-fold or more, that is, to similar degree to the fall in the HPTN 052 study. If it did, that would bring the risk down from 1/100 to 1/2000, which is within the range of risk for unprotected vaginal sex without antivirals. Is that low enough that it would be deemed “insignificant risk”? Would 1/20,000 be “insignificant risk”? I don’t know the answer, and my guess is that different people will give you different answers. The big thing that came out of the Supreme Court decision was a concrete ruling on what does constitute “insignificant risk” – and that was the risk of transmission if a person is on antivirals, has no STI and also uses a condom.
So back to condoms. You end your presentation with a slide that says “still need to recommend condoms after Supreme Court decision” but then you qualify it verbally that the need is from a legal perspective, and that in theory the risk is very low. Would you lump gay/bi men in with that statement because traditionally it’s been a “we don’t know” answer when it comes to them.
I have done my best to talk about the probability of transmission in different scenarios without giving my opinion on what risk is “significant” and what is “insignificant”. That is because, once people are informed about what the actual risks are, what constitutes “significant” becomes a personal thing and my opinion is no more meaningful than anybody else’s. However, now that the Supreme Court has ruled on the matter, I think that we all have a responsibility to inform the community about the legal perspective on this.
Bob: Right, I think I'm hearing risk is very subjective. I agree. But then we are told what is low risk/high risk all the time. I'm thinking, for instance, about the CAS guidelines (I know they are outdated, and are being revised, but we still use them), or more recent attempts to describe the risk of various sexual acts like HIM's. Aren't all those tools very subjective too? For instance, both those tools I've mentioned call acts "high risk" where there is A risk of transmission and where transmission has occured (at least once, I believe). By that standard, crossing the road is high risk, because accidents have occurred. Riding on an airplane is high risk because airplanes have crashed. Don't the prevention community's "low-/high-type" risk parameters reflect an entirely different mind set to that of the average person to whom they are addressed?
Rupert: Of course, when you look at those guides they are very subjective, in that they simply list things as “very low risk”, “high risk”, and so on without assigning an actual number to that risk. That allows people to bring their pre-conceived notions of risk to the table, and to make assumptions that may or may not be accurate.
One of the interesting things about doing basic and clinical research on HIV transmission is that people from across the spectrum of the prevention community often sit down and talk with me about the latest studies, and so I see firsthand how diverse opinions can be when it comes to preconceptions of risk. While it’s a bit off topic, I’d like to take this opportunity to make the observation that people at the opposite ends of that spectrum tend not to communicate with each other very much. While this topic naturally engenders strong and emotional responses, we really need to have that communication happen if the prevention community – and anybody reading this is part of that community – is going to be able to digest what the science has to say in a balanced way and then move the field forward.
OK, Rupert. That's a good place to end. You’ve been very generous with your answers, and I really appreciate your frankness - and patience. I’ve learned a lot from you. Thank you so much for talking to PositiveLite.com.
Dr. Rupert Kaul is a clinician-scientist based at the University of Toronto and has a clinical HIV practice at the Immunodeficiency Clinic, University Health Network. His research interests include: (1) how local factors in the genital tract, either immune factors or other co-infections, may increase HIV transmission; (2) the interactions between HIV and herpes simplex virus type 2 (genital herpes); (3) the nature of immune responses that provide protection against HIV acquisition and/or progression; and (4) the effects of microbicides in clinical testing on the immunology of the genital tract. This work is made possible by the help of research participants from Toronto and Kenya.