The Partners in Prevention HSV/HIV Transmission Study provides the largest and most detailed database to date for modeling the relationship of plasma HIV-1 RNA and risk of heterosexual HIV-1 transmission, allowing us to model a predicted impact of a reduction in plasma HIV-1 RNA on HIV-1 infectiousness. Our model assumes a linear relationship between the logarithm of the risk of HIV-1 transmission and the logarithm of plasma HIV-1 RNA; this assumption is validated by the observed data particularly in the range of 3 to 6 log10
plasma RNA. Thus, plasma HIV-1 RNA reductions of 0.5 log10
and 0.74 log10
copies/mL are associated with 37% and 50% reductions in the risk of HIV-1 transmission, respectively, regardless of the baseline level of plasma HIV-1 RNA. Our findings are in close agreement with those from a HIV-1 serodiscordant couples cohort in Rakai, Uganda for which 0.5 and 0.74 log10
plasma HIV-1 RNA reductions were associated with 36% and 48% reductions in HIV-1 transmission, respectively (reference (2); and personal communication, Dr. Laith Abu-Raddad, Weill Cornell Medical College, Qatar). However, it should be noted that viral sequence-based linkage confirmation performed in our study cohort was not performed in the Rakai study. Linkage confirmation was performed in a prior study of Zambian HIV-1 serodiscordant couples 
, but the specific measures of the relationship of plasma HIV-1 and transmission risk discussed here have not been reported for that cohort. Our findings are also similar to other recent studies reporting reductions in HIV-1 transmission risk of 26%
in association with a 0.5 log10
copies/mL plasma HIV-1 RNA decrease. In addition to incorporating genetic linkage of transmissions, our model is based on a larger cohort of HIV-1 serodiscordant couples, with longer duration of follow-up, than previous studies.
Epidemiologic co-factors for HIV-1 transmission (gender, HSV-2 infection in the HIV-1 uninfected partner, male circumcision in the HIV-1 uninfected partner and history of any unprotected sex) did not interact with or confound the relationship of plasma HIV-1 RNA levels to HIV-1 transmission risk. Since, based on study eligibility criteria, all HIV-1 infected partners in this study cohort were dually infected with HSV-2, we could not assess HSV-2 infection in the HIV-1 infected partner for interaction with the relationship of HIV-1 plasma level and HIV-1 transmission. However, given the close agreement between the model derived from this cohort compared to that derived from the Rakai cohort, where the prevalence of HSV-2 antibody among HIV-1 infected persons was lower (67%),
it is unlikely that HSV-2 infection strongly interacts with this relationship. Therefore, we infer that our model is both accurate and generally applicable to populations with varied characteristics, particularly within the 2 to 6 log10
copies/mL range of plasma HIV-1 levels where this model has greatest power.
Our data suggest that HIV-1 transmission risk increases substantially with no evidence of biological saturation through a range of plasma HIV-1 levels from 3 to 6 log10
copies/mL. In the range of 6–7 log10
copies/mL we had a small number of events (n
3) with few person-years of follow-up (n
43), so we cannot be certain that the log-linear relationship holds at such high levels of plasma HIV-1 RNA. Increased power to ascertain HIV-1 transmission risk in this high plasma HIV-1 stratum would necessitate targeting HIV-1 serodiscordant couples with one acutely infected and one HIV-1 uninfected partner, who are difficult to identify. At the low end of the plasma HIV-1 RNA spectrum (HIV-1 infected partners with plasma HIV-1 levels <3 log10
copies/mL), we only observed three HIV-1 transmission events over 968 person-years of follow-up. Given the small number of events, it is possible that, in some cases, HIV-1 levels immediately prior to HIV-1 transmission were transiently higher than those documented at their scheduled visits. More generally, since plasma HIV-1 levels were measured at 0, 3, 6, 12 months and at study exit, and not at every quarterly visit, our estimate of the relationship between HIV-1 transmission risk and HIV-1 plasma viral load may be somewhat attenuated. However, given the high correlation of plasma HIV-1 RNA levels over time in our dataset, our model is likely robust.
Antiretroviral therapy is currently the only therapeutic intervention that consistently reduces plasma HIV-1 RNA by more than 0.7 log copies/mL.
A recent analysis of our cohort documented that ART use reduced risk of HIV-1 transmission by 92% (p
0.004) and an ongoing randomized trial will assess the long-term effect of ART on HIV-1 transmission
. However, current resources constrain many countries from providing ART to all HIV-1 infected persons meeting current international guidelines for treatment, and recent proposals
to perform HIV-1 testing more broadly (“test and treat”) will accentuate this challenge.
This model demonstrates that a 0.74 log10
copies/mL reduction in HIV-1 plasma RNA reduces HIV-1 transmission by 50%, and that 90% of these infections averted could be achieved by targeting the plasma HIV-1 RNA reduction to a subset of the population with plasma HIV-1 RNA levels ≥4 log10
copies/mL (less than 60% of HIV-1 infected persons, across a range of scenarios). While ART for HIV-1 prevention may be made more cost-effective by targeting persons with plasma HIV-1 RNA ≥4 log10
copies/mL, given the current high cost and limited availability of plasma HIV-1 RNA monitoring, significant advances in this technology would be required before this approach could be implemented in ‘treatment for prevention’ algorithms
in resource-limited settings. In the absence of a low-cost HIV-1 RNA assay, the model and tool we have developed could be used to inform public health resource allocation or to prioritize populations considered for HIV-1 prevention clinical trials.
This model does not explicitly include the impact of stage of HIV-1 infection on HIV-1 transmission potential. Given the long duration of time that individuals with clinically latent HIV-1 infection can remain infectious, their overall contribution to HIV-1 transmission may be higher than individuals at early and advanced HIV-1 disease stages associated with higher plasma HIV-1 RNA levels.
However, the plasma HIV-1 RNA levels contributing to most HIV-1 infections in our study population (i.e., ≥4 log10
copies/mL) encompasses the median plasma HIV-1 level of clinically latent infections potentially mitigating the impact of explicit inclusion of stage of HIV-1 infection in the model.
This model also does not include additional behavioral factors, or viral and host factors that also contribute to HIV-1 infectiousness. However, by modeling the relationship of plasma HIV-1 RNA level to HIV-1 transmission risk, this analysis captures the principal determinant of that risk. We are also in the process of evaluating a per sexual contact model to address differences in sexual frequency and condom use. Finally, our evaluation of HIV-1 infections averted provides an approximate estimate derived by extrapolation from the stable partnerships followed through the clinical trial. More complex models are required to incorporate population-level estimations with multiple risk groups having different saturation levels and the impact of onward transmission.
Non-ART interventions, such as treatment of co-infections that are associated with increased plasma HIV-1 levels, are currently being evaluated to reduce HIV-1 infectiousness and disease progression. If such interventions reduce plasma HIV-1 levels by >0.7 log10 copies/mL, they may also provide a valuable tool for prevention of HIV-1 transmission. HSV-2 suppression with standard doses of acyclovir (as assessed in the Partners in Prevention HSV/HIV Transmission Study) did not meet this goal, and ongoing studies are evaluating whether higher doses of herpes suppression are associated with a greater reduction in plasma HIV-1 levels. Given the costs and challenges of trials evaluating interventions to reduce HIV-1 transmission, there is a need to prioritize which candidate interventions are evaluated. This model of plasma HIV-1 RNA levels and HIV-1 transmission risk provides a useful assessment tool for estimating the potential HIV-1 prevention impact of reductions in plasma HIV-1 levels.