How much does viral load need to fall to reduce infectiousness?

A sub-study of the Partners in Prevention Study¹ was able to calculate that in order to reduce the risk of HIV transmission by 50%, an average 5.5-fold (82%) viral load reduction (0.74 log) is needed.

The authors analysed the 108 linked infections already cited above, with viral load measurements available from the transmitting partner within the preceding three months.

Note that this method cannot account for temporary viral load increases, such as ones caused by infections, that might not be fully captured by the three-month window between measurements, or sexually transmitted infections which would not affect plasma viral load but which might nevertheless increase genital viral load. 

The researchers stratified individuals according to their viral load over time, and then calculated the HIV incidence per 100 person-years for each viral load stratum. 

They found that annual incidence was 5.7% from partners with viral loads over one million copies/ml, 4.7% from partners with viral loads between 100,000 and a million, 2.9% from partners with viral loads between 10,000 and 100,000, 0.7% from partners with viral loads between 1000 and 10,000, and 0.3% (only three transmissions or one in 333 couples per year) from partners with viral loads under 1000.

If the relationship between viral load and infectiousness observed remains the same at very low viral loads, and if an average viral load of 35,000 was cut to just under 40, this would imply a 94% reduction in infectiousness, or a 98% reduction for the kind of average viral load (3.34 copies/ml) actually detected by hypersensitive tests in the average person on suppressive antiretrovirals. Note that this very roughly, in lower viral load strata, equates to a tenfold reduction in infectiousness for every hundredfold decline in viral load. Note also that this does not imply that below a certain viral load, transmission is impossible.

Extrapolating from the data presented, this degree of reduction would clearly have the greatest prevention impact in people with higher levels of viral load, since this is where over 90% of transmission events occurred.

An Australian study, published in the wake of the Swiss Statement (see above), claimed that if condom use was abandoned altogether in cases where one couple had an undetectable viral load, incidence would actually increase fourfold, especially amongst gay men.² They extrapolated figures from the study of transmissions in Rakai, Uganda,³ to calculate that even if the positive partner had a viral load of only ten copies/ml there would still be a one-in-2300 chance, per sex act, of HIV being transmitted between men having anal sex. The equivalent figure in heterosexual couples would be one transmission per 23,000 sex acts from man to woman and one per 45,000 sex acts from woman to man. This would imply that in a gay relationship lasting ten years in which the couple had unprotected sex an average of twice a week throughout (1000 acts) there would be a one-in-three chance that the HIV-positive man would infect his partner.

As the authors of this study themselves comment, this calculation depends on an ‘unverifiable’ assumption that a linear relationship between viral load and transmission probability persists even down to very low viral loads, and there is no threshold below which there is so little HIV in body fluids that transmission becomes impossible.

In addition, their conclusion that “in the light of current knowledge, safe sex is the only way to prevent HIV spread” does not take into account that consistent and correct condom use is actually a behaviour only practised by a bare majority of gay men and by a minority in most gay primary relationships and between heterosexuals. As soon as the proportion of HIV positive people with suppressed viral loads in a population becomes larger than the proportion attempting 100% condom use, then treatment-as-prevention becomes at least as efficient at reducing HIV incidence as safer sex, even in the unlikely scenario that the former completely replaces the latter.