See also (in French): Virémie et contamination: Bernard Hirschel démonte les arguments fallacieux de Michel Ohayon, spécialiste de la prévention à Sida Info Service

This is a controversial statement that does not reflect the views of most doctors. The scientific evidence is incomplete, so we can’t suggest that transmission is impossible for people whose treatment is working well. We have known for a long time that successful treatment does reduce the risk, but we also know that the level of virus in the body can go up between medical visits.

You say “this is a controversial statement that does not reflect the views of most doctors". Have you polled them? Have you been careful to distinguish what they say to their patients in private from what they say in public? What they say in public is largely influenced by peer pressure, and what I’d call the asymmetry of risk assessment: Denying a risk that turns nonetheless out to be real endangers your career, while making the opposite error is of no consequence. No wonder that all recommendations by committees protect, first and foremost, those who make them!

The scientific evidence will forever be incomplete, because one cannot prove the absence of a risk. If you have “known for a long time that successful treatment does reduce the risk" why have you not said so? Because, I suppose, the evidence is incomplete. You are locking yourself into a position that is very difficult to change, even in the face of evidence which would convince most people.

“…that the level of virus in the body can go up between medical visits": As any medical student who has passed an exam knows, if the question is “can�, the answer is “yes�. Most anything can happen in medicine; the question is whether it is likely and important, or unlikely and not of consequence. Most doctors would agree that if three measures of viral loads spanning a period of 6 months or so are all negative, and if the patient continues taking his drugs regularly, the probability that "the level of virus in the body goes up between medical visits" is extremely small.

Personally I also think it’s extremely problematic that he has launched these ideas into the mainstream media single-handed, rather than seek a consensus amongst other stakeholders. Would he be able to generate a consensus around his position, given the available evidence?

We’ll see. But of course, such a consensus cannot emerge without bringing the issue into the open. The evidence, as you say, has been available for a long time – at least since Quinn’s article in 1999 [1], but consensus makers agree most of all in doing nothing.

Nonetheless, such a consensus is emerging. A paper has been approved by the Swiss AIDS commission, with a view to publication in early 2008, which is very close to what I said for AIDS day.

I agree with giving nuanced and evidence based advice, according to individual circumstances. But if that is the objective, I’m unsure that interviews in mainstream newspapers, followed by the inevitable Chinese whispers are the best way to achieve that.

It is possible to convey a nuanced and factual position in a mainstream newspaper, as you can see if you read a verbatim translation of the interview in Le Temps [2]. I’m not comfortable with the idea that there is one truth for the initiated, and another for the masses.

My main problem with this is it is based on undetectable viral load from a blood test, and 5-10% of people with undetectable viral load in blood are likely to have detectable viral load in semen, vaginal secretions and the tissue in the anus.

That is a misleading statement. The best of these studies, from my colleague P. Vernazza in St. Gallen [3], shows detectable virus in semen of 2 of 114 patients with undetectable viral load in the blood. And the two exceptions are instructive: The first was treated for only 8 weeks with a combination of stavudine, saquinavir, and ritonavir, and the second with didanosine, stavudine and hydroxyurea. Both cases were not treated long enough, and both received unconventional treatment which would not qualify as effective HAART in 2007. Regarding the issue of rectal secretions, see below. Measuring viral load in vaginal secretions is difficult; what evidence there is suggests that effective HAART causes most measurements to be negative [4] [5].

Local infections, especially STIs also increase viral load in genital fluids. For me, this is more important than the chance that viral load may increase or fluctuate between medical visits.

This has been shown in untreated patients, but rarely if ever, to my knowledge, in treated patients with stably suppressed viremia [6] [7].

I think Hirschel is wrong to say the post exposure prophylaxis (PEP) should not be given if a partner has an undetectable viral load.

Tell me why. Chance of infection in a person exposed by sex or needle stick to a treated patient with undetectable viremia may well be zero, but let’s, for the sake of argument, assume it may be as high as exposure to an untreated patient with a low viral load of 1000 per ml (risk of infection approximately 1:10000 [8]). At these odds you’d kill a few with the drugs, before saving one from HIV. And: PEP costs at least 1500 US$. The cost of preventing one HIV infection with PEP would be 1500 * 10000, or 15’000’000$. Is this an effective use of health-care resources? .

Also, transmission risk can be different for vaginal sex compared to anal sex (whether heterosexual or gay).

Could be, but what data there are from untreated patients suggests that the risk is similar [9].

For couples who want to have a baby, the risk is very low (probably less than one in 3-20,000)...

Where does that number come from? Do you know of a single documented case of transmission?

...but this is really in the context of a more careful and limited approach including:

  checking viral load is undetectable in semen

  using urine LH testing to determine ovulation and only limiting conception attempts to these 1-2 days when the woman is most likely to conceive

  using short course PrEP and PEP, perhaps single dose tenofovir/FTC, as additional protection

  not stopping condom use at other times

I have no doubt that you can scare some couples into taking all these measures, but this is not really what happens in real life.

In the context of limited access to sperm-washing, and the low conception rate from sperm-washing (around 15% I think) I think it is important that there is a wider discussion of these very low risks - so that adults can make informed decisions. Many people want to have children and this information should be available to support their choices. In practice this is something very few doctors want to discuss openly - and historically there has not been much enthusiasm for HIV-positive people to have children anyway.

The reason that “very few doctors want to discuss openly� is precisely the opprobrium cast on those who suggest that treatment might eliminate risk.

I don’t think that long-term relationships have anything to do with HIV risk compared to casual partners - other than you are more likely to have more reliable information about the partners health and sexual health - and these things impact on transmission risk.

To the contrary, long-term relationships have everything to do with estimating HIV risk, because (1) treatment and adherence to treatment determine viral load, (2) viral load determines risk, and (3) only in a long-term relationship can an uninfected partner judge adherence.

The transmission risk study that Hirshel is thinking of is probably the Rakai study from Uganda, which was a heterosexual study of about 400 serodifferent couples where no cases of transmission were seen when viral load was less that about 1500 copies/mL.

Yes [10]

I’m not aware of any similar study in gay men and don’t think these results would be seen in a study of MSM. One of the most important and under-reported issues involve viral load levels in the anal tissue and secretions, see [11] [12]

In the absence of studies one must use logic: There is really no good biological reason why vaginal transmission should be any different from rectal transmission

Indeed, the summary of Zuckerman’s article [13] says: " regardless of ART use, median HIV RNA levels were higher in rectal secretions (4.96 log10 copies/mL) than in blood plasma". However, you really need to read the paper itself. There were 63 patients, of which only 27 were on anti-retroviral treatment. Of these 27, 9 had a detectable plasma viral load, leaving only 18 with undetectable viral load. Actually, within the limits of this very small sample, there was good correlation between plasma viral load and presence of HIV in seminal secretion, and somewhat less good correlation between plasma viral load and HIV in the rectal mucosa. Two essential pieces of information are missing, (1) how many treated patients with undetectable plasma viral load had virus in their rectal biopsies, (2) how long was their plasma VL non-detectable before biopsy, and (3) what type of ART was given.

As explained above, while there may be some differences between anus and vagina, it remains uncertain if and how these differences influence transmission.

I don’t think this is out of character for Bernard Hirshel. He has spoken before about all HIV-positive people using treatment in order to reduce transmission on a population level.

Not true. The credit goes to Julio Montaner who gave a plenary conference on this topic at the Toronto conference. I also saw him present confidential study results without realising it. Thank you for the compliment; I hope I have never done worse. Perhaps if this generates much press, a consensus statement from a more representative group may follow.

Yes, I agree that Hirschel is oversimplifying. We know from Steve Taylor’s work in the UK that something like 12% of both women and gay men have higher viral loads in their semen or vaginal secretions than their blood – and that some people who have an undetectable viral load in the blood have a detectable VL in their genital secretions. Taylor called these ‘super-shedders’ which is an unfortunate term because it’s a) stigmatising and b) implies they’re always the same people, when in fact shedding HIV into the genital tract may be occasional, and more to do with having STIs etc (including asymptomatic herpes).

I suppose this refers to Taylor’s paper in Annals of Internal Medicine [14]. The work was done in Brazil, where patients received double or triple ART. According to Figure 1 of the paper, after 6 months of therapy, 44% still had a detectable viral load in plasma, and 25 percent in semen. These patients (1) were not receiving HAART as we now define it, and (2) no conclusion is possible about the presence of virus in semen, in patients with suppressed plasma viremia due to standard HAART.

Regarding the "super shedders": None of these were on treatment.

There is a strong correlation between undetectability in blood and in genital secretions – so yes, there’s a high likelihood that people on suppressive HAART are usually only marginally infectious. But the studies haven’t been done to find out how common exceptions are to this rule.

What would you consider a convincing study? What methodology to use? Consider Jin F et al: How homosexual men believe they became infected with HIV [15]. In Australia some 143 recently infected men expressed an opinion as to how they became infected. 9 claimed a “man with suppressed viremiaâ€? as a source, whereas 20 said that their probable source was someone who had told them that he was seronegative. I don’t see any other method that the follow-up of stable sero-discordant couples (1), [16], and mother-to-child transmission [17]. But mainly, it seems to me, just use your head: How could transmission occur in the absence of virus?

Another problem Hirschel isn’t considering is that in long-term serodiscordant couples, the negative partner does build up a degree of immunity to their partner’s HIV. This wouldn’t apply to a new or casual relationship. Barry Peters at St Thomas’s pioneered work on this.

It could well be that immunity contributes to the lack of transmission, but wouldn’t you agree that it’s the virus, mostly? How could undetectable virus cause infections?

Whether Hirschel’s right about most cases or not, I also worry about statements such as this being used to rationalise unprotected sex. Some studies have found that gay men are using viral load as a guide as to whether to have unprotected sex but of course your only know that you were undetectable at your last test – which could have been months ago.

For the likelihood that the third test after two previous undetectables will also be undetectable, see above.

It also lays HIV+ people open to unwitting transmission.

What do you mean by that?

I’m not saying that serodiscordant couples should never choose to have unprotected sex. In a relationship where risk is discussed, an informed choice to run such a risk is valid, perhaps in conjunction with other choices like seropositioning (i.e. only without condoms if the HIV+ partner is bottom). But it needs to be an informed one on both sides.

Take a deep breath and say it in public – experience shows the sky won’t fall in.

I also think that if absolutely everyone diagnosed with HIV took ARVs, then incidence would be a lot lower – but it wouldn’t be zero because a) not everyone would and b) possibly a half of all infections are acquired from people who’ve been very recently infected themselves and are usually undiagnosed – see the Quebec study below for this. Even with perfect diagnosis, there would remain some transmission from people in the ‘window period’.

Nothing is perfect, but the models presented at the Toronto conference by Montaner et al. from British Columbia , and at the 14th CROI from Amsterdam [18] show that without HAART, new infections might well be, today, twice as frequent as they are. We do not now have, and will not in the future have a vaccine, and it behoves everyone, and especially the THT, to consider all means of prevention, including treatment.

Somebody is sure to publish, sooner or later, a case report of a transmission originating in a patient with suppressed viremia. What then ?

1) "Hard cases make bad law". Recommendations from Public Health bodies have to weigh benefits and risks for the typical case, not the exceptions.

2) Official recommendations properly address behaviours influencing epidemics – for HIV/AIDS, unprotected intercourse in the absence of effective treatment. The rationale for intervention is less certain when behaviour is associated with risks that are so small that they cannot be measured, or only inferred from anecdotes.

3) Anecdotes regarding HIV infection are inherently problematic, because their interpretation depends on what people tell you about their sex lives. Unfortunately, one cannot rely on this kind of information .

Many studies have asked heterosexual men and women about the number of sex partners they have had. In all countries, men indicate that they have more sex partners than women. But because it takes 1 man plus 1 woman to have sex, this is mathematically impossible. Men exaggerate, or women forget; one or the other, or both, are not telling the truth.

Now consider this hypothetical scenario: A is treated since 2003, with undetectable viremia measured every 4 months. In 2004, A meets B, and they have unprotected sex. B tests HIV-negative in 2004, and in January of 2006, but HIV-positive in March of 2007. A and B declare that they are faithful to each other and have no other sexual partners. The viruses of A and B are closely related. Is this proof that A infected B while having undetectable viremia?

There are two alternative explanations:

1) There is a third individual, C, who infected A in 2003, and B in 2006. This would imply that B was not always faithful to A, and there may be powerful psychological incentives not to admit to infidelity

2) A might have stopped taking his pills for a few weeks without telling B. Rebound, often to extremely high levels of viremia, occurs within two weeks in more than 90 percent of patients who stop their drugs. Again, there may be a reason why A would not admit his lack of compliance: B might say: "What, you stopped taking treatment but didn’t tell me?"

How could you find out whether your original interpretation (transmission originating from a patient with undetectable viremia), or alternative explanations 1) or 2) apply? I cannot think of any method to do that. And being the physician of A, or B, you will not want to pry – it puts you in an impossible position of doubting the word of your patient. Much easier to believe what you are told!

But would you want to issue Public Health Recommendations on such flimsy evidence?