Systematic review unable to confirm or deny Swiss statement on infection risk with undetectable viral load

Edwin J. Bernard
Published: 08 August 2008

A systematic review of studies of serodiscordant couples where the HIV-positive partner was on antiretroviral treatment could neither confirm not disprove the recent Swiss declaration of a negligible risk of HIV transmission from a person on treatment with undetectable viral load, according to data presented during Thursday afternoon’s late breaker sessions at the XVII International AIDS Conference in Mexico City.

However, the review revealed some new information – that HIV transmission, from an individual not on treatment, had occurred in a 2005 Spanish study at a blood plasma viral load of 362 copies/ml.

The Swiss statement – which has caused controversy since it was published in January 2008 and which was discussed at a pre-conference satellite session last week – asserted that the per sexual act risk of HIV transmission from an HIV-positive individual on treatment with an undetectable viral load and no sexually transmitted infections to their HIV-negative partner is below 1 in 100,000.

To assess whether this was indeed the case, investigators from the University of Bern conducted an extensive search of published studies and conference presentations involving serodiscordant couples since 1996, when effective antiretroviral treatment first became available.

Out of a total of 252 published articles and abstracts only 14 were found to be potentially eligible due to duplication, irrelevant topic or study design, or lack of additional data from the study authors – of these, seven were published or were about to be published, and seven were conference abstracts.

However, their systematic review did not identify any studies that fulfilled the exact criteria of the Swiss statement – where the HIV-positive partner is on antiretroviral treatment, with an undetectable viral load (below 40 copies/ml) for longer than six months with no other sexually transmitted infection (STI), which they defined as syphilis, chlamydia, gonorrhoea, or genital herpes.

Consequently, they included cohorts where the HIV-positive partner was not on treatment; that defined a higher threshold of undetectable viral load (400 copies/ml); and/or where the HIV-positive partner’s STI status was unclear.

The greatest number of discordant couples in these studies came from Africa (1822) with only 424 from Europe. Just one study included gay men or other men who have sex with men (accounting for just 43 couples) with the remaining thirteen cohorts including heterosexual couples. However, none of these studies included detailed information about the specific sexual acts they practised – oral, vaginal and/or anal sex.

Only nine cohorts reported use of antiretroviral therapy in the HIV-positive partner, accounting for 428 couples.

Only one study, from Castilla and colleagues in Spain, published in 2005, reported on HIV-positive individuals on antiretroviral treatment with an undetectable viral load, although the STI status of these individuals remains unclear. In this study, there were no transmissions per 100 person-years from 283.2 person-years of follow-up.

The authors of the systematic review calculated a summary estimate of the HIV transmission rate per 100 person years with a 95% confidence interval (CI) resulting in an upper confidence interval of 1.06 transmissions per 100 person-years.

In the four studies that included HIV-positive individuals with a viral load below 400 copies/ml who were not on treatment but where there was no clear information on STI status, only one transmission occurred over more than 600 person-years. This resulted in a transmission rate of 0.16 per 100 person-years and an upper confidence interval of 1.16 transmissions per 100 person-years.

This single instance of HIV transmission occurred at 362 copies/ml in the blood plasma viral load. This is new information not included in the original article in Journal of Acquired Immune Deficiency Syndromes. Previously, the highest blood viral load threshold published for HIV transmission was around 1500 copies/ml by Quinn and colleagues.

She noted that a recent Australian study, using a mathematic model that assumed there was no threshold viral load level below which transmission would not occur, had concluded that transmission may still occur from gay men who fulfilled the provisos of the Swiss statement and who practised anal sex.

In an online editorial responding to this Australian modelling study, Swiss Federal AIDS Commission President Professor Pietro Vernazza strongly suggests that the Australians’ assumptions are flawed, and highlights the accompanying editorial by Garnett and Gazzard, which pointed out that even if the assumptions in the model were correct, the risks of transmission over 100 sexual acts were similarly low for 100% condom use and unprotected sex on treatment with an undetectable viral load and no STIs.

Systematic review lead author Suzanna Attia conceded that, “a body of indirect evidence suggests that HIV transmission at very low viral loads is very rare. We did not identify any studies or case reports where transmission occurred below 40 copies/ml.” However, on the basis of current evidence, she and her co-authors were unable to confirm or disprove the Swiss statement’s suggested 1 in 100,000 risk of HIV transmission for an individual taking effective HIV treatment with no STIs.

She concluded by stressing that their review was ongoing, and it was hoped that incoming data would allow them to increase the precision of their estimates.