Why the proportion of transmission during early-stage HIV infection does not predict the long-term impact of treatment on HIV incidence Jeffrey W. Eaton1 and Timothy B. Hallett Department of Infectious Disease Epidemiology, Imperial College London, St Mary’s Hospital, London W2 1PG, United Kingdom Edited by Alan S. Perelson, Los Alamos National Laboratory, Los Alamos, NM, and accepted by the Editorial Board September 5, 2014 (received for review December 23, 2013) Antiretroviral therapy (ART) reduces the infectiousness of HIV- been infected recently affects the impact of treatment scale-up infected persons, but only after testing, linkage to care, and success- on HIV incidence. The model is calibrated to longitudinal HIV ful viral suppression. Thus, a large proportion of HIV transmission prevalence data from South Africa using a Bayesian framework. during a period of high infectiousness in the first few months after Thus, the model accounts for not only the early epidemic growth infection (“early transmission”) is perceived as a threat to the impact rate highlighted in previous research (5, 9, 18), but also the of HIV “treatment-as-prevention” strategies. We created a mathe- heterogeneity and sexual behavior change to explain the peak matical model of a heterosexual HIV epidemic to investigate how and decline in HIV incidence observed in sub-Saharan African the proportion of early transmission affects the impact of ART on HIV epidemics (32, 33). reducing HIV incidence. The model includes stages of HIV infection, The model calibration allows uncertainty about factors that flexible sexual mixing, and changes in risk behavior over the epi- determine the amount of early transmission, including the relative demic. The model was calibrated to HIV prevalence data from South infectiousness during early infection, heterogeneity in propensity Africa using a Bayesian framework. Immediately after ART was in- for sexual risk behavior, assortativity in sexual partner selection, troduced, more early transmission was associated with a smaller re- reduction in risk propensity over the life course, and population- duction in HIV incidence rate—consistent with the concern that wide reductions in risk behavior in response to the epidemic (32, 33). This results in multiple combinations of parameter values that a large amount of early transmission reduces the impact of treatment are consistent with the observed epidemic and variation in the on incidence. However, the proportion of early transmission was not amount of early transmission. We simulated the impact of a strongly related to the long-term reduction in incidence. This was treatment intervention and report how the proportion of early because more early transmission resulted in a shorter generation transmission correlates with the reduction in HIV incidence from time, in which case lower values for the basic reproductive number the intervention over the short- and long-term. (R0) are consistent with observed epidemic growth, and R0 was neg- atively correlated with long-term intervention impact. The fraction Results of early transmission depends on biological factors, behavioral pat- Model Calibration and Transmission During Epidemic Stage. The terns, and epidemic stage and alone does not predict long-term in- mathematical model was calibrated to the time series of HIV tervention impacts. However, early transmission may be an important prevalence among pregnant women in South Africa from 1990 determinant in the outcome of short-term trials and evaluation through 2008 (34) and HIV prevalence among adult men and of programs. womenaged15–49 y in three national household surveys in 2002, 2005, and 2008 (Fig. 1A). The HIV incidence rate grows rapidly HIV incidence | antiretroviral therapy | early infection | in the early 1990s, peaking in 1997 and declining by around HIV prevention intervention | mathematical model Significance ecent studies have confirmed that effective antiretroviral Rtherapy (ART) reduces the transmission of HIV among stable Antiretroviral treatment (ART), which prevents both morbidity – heterosexual couples (1 3). This finding has generated interest in and HIV transmission for persons infected with HIV, is now understanding the population-level impact of HIV treatment on thought to be central to strategies for controlling the spread of reducing the rate of new HIV infections in generalized epidemic HIV. One major concern has been that if a large proportion of settings (4). Research, including mathematical modeling (5–10), HIV transmission occurs early in infection, before persons can be implementation research (11), and major randomized controlled diagnosed and treated, the impact of treatment on reducing trials (12–14), are focused on how ART provision might be expanded new infections will be less. We use a mathematical model to strategically to maximize its public health benefits (15, 16). quantify how the proportion of early transmission will affect One concern is that if a large fraction of HIV transmission occurs the impact of intervention strategies, and explain underlying shortly after a person becomes infected, before the person can be epidemiological mechanisms for this. Model simulations sug- diagnosed and initiated on ART, this will limit the potential impact gest that—counter to expectations—the proportion of early of HIV treatment on reducing HIV incidence (9, 17, 18). Data transmission at the start of an ART intervention does not pre- suggest that persons are more infectious during a short period of dict the long-term intervention impact. “early infection” after becoming infected with HIV (19–22), al- though there is debate about the extent, duration, and determinants Author contributions: J.W.E. and T.B.H. designed research, performed research, contrib- of elevated infectiousness (18, 23). The amount of transmission that uted new reagents/analytic tools, analyzed data, and wrote the paper. occurs also will depend on patterns of sexual behavior and sexual This article is a PNAS Direct Submission. A.S.P. is a guest editor invited by the Editorial networks (17, 24–27). There have been estimates for the contri- Board. bution of early infection to transmission from mathematical models Freely available online through the PNAS open access option. (7, 17, 21, 24–26) and phylogenetic analyses (28–31), but these vary See Commentary on page 15867. widely, from 5% to above 50% (23). 1To whom correspondence should be addressed. Email: [email protected]. In this study, we use a mathematical model to quantify how the This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. proportion of transmission that comes from persons who have 1073/pnas.1323007111/-/DCSupplemental. 16202–16207 | PNAS | November 11, 2014 | vol. 111 | no. 45 www.pnas.org/cgi/doi/10.1073/pnas.1323007111 Downloaded by guest on October 1, 2021 37% (95% credible interval 30–43) over the next decade A B SEE COMMENTARY B 15 30% 2 (Fig. 1 ). The decline in incidence is the result of two factors: R = 0.56 infection saturation in the higher-risk groups and a reduction 25% in the unprotected sexual contact rate over time. It is esti- 10 20% mated that the contact rate declines by 28% (95% CI, 11–43; 15% parameter posterior distributions in Fig. S1). Density 5 The fraction of transmission occurring from individuals in each 10% Early transmission stage of infection evolves over the course of the epidemic (Fig. 0 5% 1C). During the early growth phase of the epidemic, most infected 10% 20% 30% 7x 10x 15x 30x 50x persons have been infected recently, and so most new infections Early transmission in 2010 Relative early infect. (log scale) result from highly infectious persons in early HIV infection (early Fig. 2. (A) Distribution in the percentage of transmission during early in- transmission). As the epidemic matures, the contribution of early fection in 2010. (B) Correlation between the relative infectiousness during transmission declines whereas the contribution of advanced stages early infection (on the log scale) and the proportion of early transmission. of infection grows. In the absence of ART scale-up, early transmission was estimated to account for 17% (95% CI, 12–24) of all transmission in 2010 (Fig. A Fig. 3 illustrates the correlation between the proportion of early 2 ). Much of the variation in early transmission was explained by transmission at the start of the intervention and the reduction in the relative infectiousness during early infection compared to R2 = B incidence over time. In the first year of the intervention, the re- chronic infection ( 56%; Fig. 2 ). Adjusting for the influence of duction in incidence was negatively correlated with the fraction of infectiousness during early infection, higher rates of moving from early transmission (Pearson’s r = −0.49) (Fig. 3A). This is con- higher- to lower-risk groups resulted in more early transmission sistent with the conventional wisdom that more early transmission because faster movement to low-risk groups results in more will result in less impact of treatment on reducing incidence. transmission occurring early after infection. These two parame- However, in the long term, the proportion of early transmission ters together explained 89% of the variation in early transmission A was not strongly predictive of the reduction in incidence, ex- in 2010 (Fig. S2 ). plaining only about 5% of the variation in intervention impact. In fact, after the 10th year, there was a modest positive correlation Early Transmission