Our goal today is to address three specific questions that have remained controversial throughout the AIDS epidemic. First, based on available scientific evidence, what is the risk of HIV transmission to an HIV-uninfected person who performs oral sex on an insertive male partner who is HIV positive?
Second, what specific factors can affect this risk in a given situation? And finally, what advice is most appropriate for care providers and public health officials to communicate to people who have, or are thinking of having, oral sex? Based on available scientific evidence, what is the risk of HIV transmission to an HIV-uninfected person who performs oral sex on an insertive male partner who is HIV positive?
Thanks to everybody for joining us today. I think it might be useful to begin the discussion by trying to come to some consensus of what the scientific evidence says about the risk of HIV transmission to an HIV-uninfected person who performs oral sex on an insertive male partner who is HIV positive.
In other words, what risk does one have of acquiring HIV by practicing receptive oral sex with or without ejaculation? We have various datasets and literatures that can play to that. So why don't we start by reviewing what those datasets. I think it's appropriate that we refer to receptive sex on a male partner as "fellatio," since that is an appropriate and well described term for the act. Okay, fellatio with ejaculation. Who would like to start? Perhaps we could get fellatio without ejaculation off the table and agree that's an extremely low risk exposure activity?
Well, let's hear about that. Would everyone agree with that? Or, would there be anyone who disagrees with that?
I think when we're talking about risk, we have to talk first about whether transmission has been reported to occur by that route and second, how frequently it occurs. What I would say about receptive fellatio without ejaculation is that there are case reports that it has occurred, but that it is exceedingly rare, and so the risk is very low for that type of exposure. That would be my opinion.
I would say it does occur but very rarely and it is very low risk. Certainly in our Options study and other studies, there are clear reports of transmission through anal sex where there is not ejaculation, so I think pre-ejaculate is potentially contagious. So that suggests that if oral sex transmission occurs, it potentially could happen without ejaculation.
Having said that, though, while biologically plausible, I think it would be very unlikely. The data that's out there suggests that it is very low risk but I think there's not really good data to quantitate that.
We have one report in the Options Project, one oral transmission case in which it's unclear whether there was ejaculation. It's possibly without ejaculation and I'm a little skeptical because of the exact way the report was made whether that's true.
But otherwise, all the cases in which we suspect oral sex transmission included ejaculation. Our data again has a lot of limitations. We're not seeing cases, with one possible exception, that it can be a route of transmission. So we will say that fellatio without ejaculation can happen but it is exceedingly rare. It's not "no risk" but it's relatively low risk. I would say extremely low risk. Extremely low risk, okay. I think there has to be exposure to infectious substance or an infectious inoculum and no, HIV is not transmitted from skin-to-skin contact or skin-to-mucous-membrane contact without any deep cuts or breaks in squamous epithelium.
If there is no infectious pre-cum, which is still a hypothetical route of transmission, and there is no ejaculate, there should be no transmission, there should be no exposure to virus. To follow up on that, there is some evidence that there is virus in pre-ejaculate, although probably low titer in a relatively small volume.
One situation in which you might think that infection without ejaculation could occur orally would be, for instance, if someone had a urethral discharge. In cases of urethral gonorrhea or other inflammatory STDs or potentially ulcerative STDs--where you have an open wound on the penis--you could hypothesize that transmission could occur by that route. In the absence of those types of cofactors, I think I would agree with what Rick said, that transmission of HIV by receptive fellatio without ejaculation is exceedingly rare.
The problem with the discussion, though, continues to revolve around the inability to quantify risk. And because these are cases or, in fact, even uncorroborated cases, of acquiring HIV from fellatio without ejaculation, besides saying "exceedingly low risk" or "very low risk," that's the best you can do.
It is all still hypothetical. Well then, let's move to the case of fellatio with ejaculation and as you, from your various datasets, attempt to quantify the risk, it might be helpful to give a short description of the data upon which the estimates are made.
Susan, do you have anything to say? I think that there's clearly biological plausibility that HIV transmission can occur from receptive fellatio with ejaculation and there are a number of types of databases that support that that occurs. One is animal data, which demonstrates that if you swab tonsillar tissue or expose macaques to SIV via oral mucosa, infection can occur Slide 1. On the other hand, I think that the studies in which comparisons are made of the efficacy of SIV transmission through oral, anal, vaginal, and intravenous routes are potentially flawed.
In animal models, vaginal and rectal exposure are done atraumatically, which doesn't really representing what happens during sexual intercourse. So I wouldn't take away from the SIV models what the relative risks of various types of exposure are, but I would say that infection clearly occurs.
There also are data from animal models describing in detail the histology of tonsillar tissue, which is very similar to the histology of vaginal and rectal tissue, both in being rich in dendritic cells and also having these so-called M-cells that are similar to the MALT--or mucosal-associated lymphoid tissue--of the gut and the rectum. These cells are able to transport antigens to lymphoid tissue in the absence of trauma or inflammation.
So I think that there's good basic science evidence that tonsillar tissue in particular should be susceptible to infection.
We also know that an average ejaculate in the absence of inflammation or STDs has about a million white cells. Now, we still don't know whether HIV is transmitted through free virus or cell-associated virus. But again, if a male partner is ejaculating into someone's mouth and into the area of the tonsillar tissue, the tonsillar tissue is receptive to becoming infected. We know there's HIV in semen.
We know there are white cells in semen. It seems eminently reasonable to me that infection could occur by that route. So then we're left with "what do the epidemiologic data tell us? When you do cohort studies in people who are generally having multiple sexual practices, many studies suggest that having receptive fellatio with ejaculation does, in fact, confer some increased risk but that risk is relatively low, and in most cases no longer statistically significant once you control for other riskier factors, such as receptive anal sex.
From data that we have--and I will let Eric describe it in more detail--laying out the per-contact risk of HIV, having receptive anal sex with or without ejaculation is probably on the order of 10 times riskier than having receptive fellatio with ejaculation.
So clearly, the riskiest practices can overwhelm our ability to look at the risk that is associated with having this lower-risk type of exposure. There are case series, which I think Rick will probably describe in greater detail, demonstrating that oral transmission occurs. We also have a cohort study of newly infected people in cities where people were followed prospectively in a cohort study.
We had about a hundred seroconverters who were prospectively identified and were asked about their risk behaviors both prospectively--before their infection status was known--as well as retrospectively, when they were queried in great detail about all of their potential HIV exposures from three months prior to their last negative HIV test all the way through their first positive HIV test. Is it plausible that some proportion of those men under-reported higher risk practices, either because they misclassified the serostatus of their partners or were uncomfortable acknowledging their actual risk behaviors?
The analysis that I worked on was with data for people who were multiply exposed, to try to use multivariate methods essentially to tease out some estimate of the risk of fellatio with ejaculation Slide 3. We worked on a large dataset of an month prospective cohort with three follow-up visits.
One thing to emphasize is that this was lower by a factor of about 7 than receptive anal sex with HIV-positive or unknown sero-status partners, which was the same partner category that was used for the fellatio analysis. The other point to make is that the confidence interval for that estimate for fellatio with ejaculation was also quite wide. I mean, consistent with what Susan was saying about the difficulty of rejecting the hypothesis of finding a statistically significant result is that the information isn't very strong.
So while the lower bound was very low, the upper bound was also high enough to be of some concern. To go back to where Susan started, the evidence is there that there is biological plausibility. Of the case reports--actually 28 published--some have multiple cases, so there are maybe over 40 in the actual literature.
But I would emphasize that the number of case reports is extremely low when one considers the size and the duration of this epidemic and that, from a population perspective, should be kept in mind. The SIV data is also compelling and I think that probably more should be done with that animal model. The problem I have with the SIV model, and I think Susan addressed, is that it's not easy to extrapolate because in fact oral transmission of SIV, non-traumatic oral transmission, was easier to induce than anal infection in the SIV model.
So I'm not entirely a believer in the SIV model vis-a-vis transmission routes. I would argue that the epidemiologic evidence is actually overwhelming in showing that the risk is extremely low, and I go back to a paper published from the early cohorts.
Roger Detels published a wonderful paper in which men out of 2, MSM were followed and two of those reported no anal-genital sex in the seroconversion period in which their infection was detected, but in fact only one of those could be reliably reported as being inside the range of this infection period and so this very early data goes back and shows extremely low risk Slide 4. Obviously, since the beginning of this epidemic, oral sex has been performed by a majority of folks, and not only gay men but heterosexuals.
So it has always been a concern. But I don't think that the epidemiologic evidence in any way supports that it implies risk. In July of this last year, a group in Spain published an excellent paper from serodiscordant couples, who were heterosexuals, where they evaluated for risks of HIV transmission through unprotected oral sex, and in over 19, unprotected oral-genital contacts with HIV-infected partners, there was not a single case of seroconversion to HIV Slide 5.
This included both infected women and infected men, but the majority of the population in this study was infected men. Our data in seroconverters that we've analyzed as well from pooling studies of seroconversion, did find a significant, an elevated risk. This is the study that we published in , where we showed an elevated odds in a model controlling for anal sex, and actually this is what led me to begin to believe that there was increased oral sex risk, finding this elevated odds ratio in association with oral sex in seroconverters Slide 6.
The limitation of my study, which is clearly pointed out and which I come back to as I review more and more evidence, is that I could not ascertain that any of the men in this study did not also have anal sex and I believe that infectivity of anal sex is so high that I think it's almost impossible to unmask that. Eric's study was compelling and I think it was not only compelling because of the significantly low infectivity estimate associated with oral sex, but also, even though not significantly, insertive anal sex was of somewhat higher estimated infectivity, and protected receptive anal sex with a condom was of somewhat higher infectivity.
The data we recently published from my study designed to look at this question--and it may be the only study designed to look at the risk of oral sex--we published a study that showed that among men who practice exclusively fellatio, not one HIV infection had occurred Slide 7. To date, we've now interviewed over men and again find no infections. This represents over 5, acts of oral sex, and preliminary infectivity estimates based on certain assumptions suggest an upper bound of less than what Eric published, which is 0.
And I would say that, from the same time period and from the same population from which we recruited our participants, that HIV prevalence and incidence were extremely high in men who reported anal sex and in men who reported anal sex with a condom. These men were all recruited from HIV testing sites, who tend to be very high risk. There are likely to be certain differences in men who only practice oral sex compared to men who have a larger repertoire of sexual activities.
The last thing I want to note about the epidemiologic studies, is that there are studies which report a high proportion of men who say that they got HIV infection from oral sex and in general, most of those studies rely on behavior which is recorded after folks know the results of their HIV test. And I think in the early years, 10 years ago, we knew that there was a recall bias in that. I would hypothesize, and this is only a hypothesis, and much of we're trying to do here today is based on hypotheses, but I would hypothesize that recall bias may be even higher nowadays.