Gay men with overlapping partnerships are more likely to be recently and locally infected

A study analysing transmission patterns between gay men in a cohort of patients in San Diego, California has found the strongest scientific evidence so far that concurrency – having ongoing sexual relationships with more than one partner at the same time – is an independent predictor of HIV transmission.

It found that men who said they had had condomless sex in the last three months with concurrent partners, rather than just serial partners, were 69% more likely to be in a local transmission pair or cluster that was linked genetically, an indication of HIV transmission between the partners, and the average number of partners they were linked to was 40% greater.

The study also found that participants with concurrent partners were 58% more likely to be the likely source of HIV infection seen in another cohort member, rather than the recipient, and 67% more likely to have apparently passed on their HIV in early infection (between one and six months after infection). However these two observations did not reach statistical significance.

The researchers say that concurrency is probably an important contributor to the high ongoing burden of HIV among men who have sex with men and that being in concurrent sexual partnerships should be considered as a criterion for pre-exposure prophylaxis (PrEP).

Background

The background to the idea that concurrency – having more than one regular partner – might be a significant factor in HIV transmission first arose in discussions about heterosexual transmission in Africa, as a way of trying to explain the high prevalence of HIV in some populations. However in that context its influence was very difficult to establish scientifically as frequent testing (to establish the date of infection), genetic testing (to find out if infections in individuals were linked) and behavioural questionnaires (to establish concurrency) are all needed.

If concurrent relationships are common in a population, they speed up HIV transmission because people form more strongly linked sexual networks. Each person not only has more partners but those partners are linked to more third parties. If the predominant pattern of sexual partnerships is serial monogamy then people will only pass on an infection they have themselves acquired when they stop having sex with the person who gave it to them. If the pattern is mainly one-off casual sexual contacts then the chance of acquiring HIV from each person is restricted to one encounter, and if you don’t acquire HIV then you will not pass it on to the next person. But concurrent partnerships, which imply having sex a number of times with at least two people in the same time span, offer the opportunity for multiple exposures to HIV and, if you are infected, multiple opportunities to pass it on.

This is a relationship pattern many gay men observe; as an Australian study found out last year, a high proportion of gay men, rather than just having either committed relationships or casual encounters, have ongoing but less emotionally committed relationships with ‘fuck buddies’ and may have those at the same time as with committed boyfriends and/or casual encounters.

The study

The researchers were able to establish whether concurrency was linked to HIV transmission because they had two important pieces of information. Firstly, they had a group of 985 gay men with HIV from two different studies who were newly diagnosed and not on treatment. These men had a gene in their virus sequenced. Genetic testing like this can be used to investigate many things, including the rate HIV changes over time, the prevalence of drug resistance, and, in this case, whether two or more viruses in a group of people are ‘linked’ – i.e. whether they are so similar that it is reasonable to conclude one of the people might have infected the other(s).

The researchers were also able to compute an estimated date of infection by using incidence assays and other techniques. If two samples were linked, the direction of infection was inferred if the estimated date of infection was more than 30 days after the other, and that the source partner was in recent infection if it was less than 180 days later in one sample than the other.

Secondly, out of these men, they also had 288 who had completed a computer-assisted interview that provided details of up to three of their sexual partners during the last three months. This included information (when known) on the partners’ age, gender, HIV status, whether they were a “main” partner, and whether condomless anal sex took place. Importantly for this study, the men also provided information on when they first, and most recently, had sex with these partners – so it was easy to work out if relationships overlapped.

The researchers then looked at whether people who had concurrent relationships were more likely to be genetically linked to others within the cohort, whether they were more likely to be in early infection, and whether they were likely to be the source of infection.

Results – linked transmissions

The average difference between the 487,578 pairs of viral samples in the group was 5.83%, meaning that on average about one in 20 bases (links in the genetic chain) was different in one sample from a randomly selected second sample. Pairs were judged to be genetically linked if the genetic difference was less than 1.5%, which occurred in only 0.25% of pairs, and of these 698 were judged to be “putative transmissions”.

Of the 985 genetically tested patients, 594 had been infected for less than six months (“recent infection”). Forty-five per cent of the cohort (452 men) were part of a putative transmission pair or group – 49% of those tested during recent infection. In 63% of links, the researchers were able to work out the putative date of transmission.

Sixty-one per cent of linked transmissions were between only two people, but the remainder were in linked clusters, with the biggest cluster numbering 21 genetically linked men. The mean number of men in a linked group was 3.2.

The men in pairs or groups in the study had only had HIV for an average of seven weeks, the researchers estimated.

It was possible to work out a putative direction of transmission in 338 (73%) of links; of these, an existing participant in the study was the transmission source in 189 (32%) of cases. In 18% of cases, including men who entered the study both as source partners and recipients, transmission definitely happened during the source partner’s period of recent infection, and of these, 45 (10%) definitely infected a later participant during their period of recent infection. There were probably more, but many men entered the study after their period of recent infection and in many cases a direction of infection could not be inferred due to dates of infection being too close together or uncertain.

Results – concurrent relationships

Of the 288 men interviewed about their last 2-3 sexual partners in the last three months, 75% had had condomless sex with at least one partner, and 54% had had overlapping relationships during that time. Interestingly, 37% said that they had had more than one “main” partner, which backs up the Australian study’s finding that gay men don’t just think in terms of committed versus casual relationships.

Sixty per cent of men who had concurrent relationships were in a genetically linked pair or group, versus 46% who had not, and men with concurrent relationships had a higher number, on average, of men their HIV was genetically linked to: 2.5 (this includes men with zero links) versus 1.6.

Adjusting for confounders that could also increase a person’s likelihood of infection, having had concurrent relationships raised the risk of being in a cluster by 69% and raised the average number of connections by 40%.

More participants with concurrent partners were inferred to be the source of a genetically linked HIV infection (32% with concurrent partners versus 24% without, a 58% raised risk) and this definitely happened during their period of recent infection in 20% with concurrent partners versus 12% without, a 67% raised risk. However in these last two instances, there were too few people to do a proper multivariate analysis so these two figures cannot be counted as statistically significant.

Comments and conclusions

The researchers conclude that “While our findings do not provide direct evidence that concurrency facilitates HIV transmission during recent infection, they are consistent with the concurrency hypothesis, suggest concurrency during recent infection is associated with HIV transmission among MSM [men who have sex with men]... [They] suggest concurrency may contribute to the disproportionate burden of HIV among MSM in the United States and point to the need for HIV prevention strategies that mitigate its effect.

“One strategy might include the expansion of indications for pre-exposure prophylaxis (PrEP) to include sexual partner concurrency. PrEP use among MSM who report concurrent sexual partners could reduce their risk of HIV acquisition from HIV-infected sexual partners, and thus their subsequent probability of transmitting HIV to uninfected concurrent sexual partners.”