Reply to ‘Evidence is still required for treatment as prevention for riskier routes of HIV transmission’

Abstract

We read with interest Dr Wilson's thoughtful review of our article [1] describing a new transmission model in British Columbia. We agree that the parameter describing the reduction in transmission risk associated with decreased viral load is among the most important for accurately estimating the effectiveness and cost-effectiveness of a program to expand treatment with highly active antiretroviral therapy (HAART). In the absence of empirical data quantifying this parameter from other populations, we utilized an estimate of 2.45-fold decrease per log-copy decrease in viral load from studies of discordant heterosexual couples [2], under the assumption that the magnitude of the association would not be modified by route of transmission. Results from a more recent meta-analysis of heterosexual transmission [3] suggest that the transmission reduction may be greater at all levels of viral load (Table 1, last column). Along with other assumptions within the model, we included parameter estimates, which, if biased, would likely have an effect of underestimating the benefits of an increase in HAART use.Table 1: Extrapolated reduction in infectivity associated with a 1 − log change in viral load, stratified by viral load categories.With respect to other routes of infection, more recently published observational data from British Columbia are consistent with the results of our modeling study [4]. Within British Columbia, homosexual and injection-related are both significant routes of transmission. Between 2004 and 2009, corresponding to a period of HAART expansion, there was a 17% reduction in new infections relative to what would be expected if infection rates remained constant throughout the period [4]. Within injection drug users specifically, a notable relationship was observed between decreases in community viral load and subsequent new infections [4]. In the mathematical model, we assumed that 2000 new individuals would be accessing treatment and as a result, 26% fewer infections were projected to occur over 30 years [1]. A Poisson regression analysis of observed data in British Columbia indicated a 0.97 reduction factor in new infections for every 100 individuals on treatment; for 2000 individuals, this would correspond to an even greater projected reduction in new infections, 46% based on 13 years of available data. Despite the uncertainty in transmission reduction parameters for specific subpopulations, the published observational data [4] are consistent with our interpretation of the model results that show the program to be cost-effective even under conservative assumptions. As such, the values from the model likely underestimate the actual economic benefit of expanding treatment. The study by Fisher et al.[5] referenced by Dr Wilson reports a transmission risk reduction factor of 1.61 per log-change in viral load for men who have sex with men. That study employs a novel design, with notable features being that transmission events were assigned probabilistically based on genetic markers, and that the study sample was characterized by individuals who were infected with HIV. Thus, by definition, no empirical data are included for encounters in which individuals were not infected, which may have affected the risk ratio. To assess the impact of this lower estimate of transmission risk reduction, we replaced the reduction factor of 2.45 with 1.61 in the mathematical model, and estimated a reduction in cumulative of net benefit of 42%, from US$ 900 million to 525 million. Although changing the monetary benefits, the program remains cost-effective intervention, and became so within 2 years of implementation. We agree that obtaining empirical estimates to quantify the relationship between viral load and reduction in transmission risk specifically for homosexual and injection-related transmission would represent an important refinement in model-based results. In the absence of such estimates, however, the available evidence continues to suggest societal and economic benefits of providing HAART to all those with a medical indication.