Modelling the impact of antiretroviral therapy on the epidemic of HIV

Abstract

Thirty years after HIV first appeared it has killed close to 30 million people but transmission continues unchecked. In 2009, an estimated 1.8 million lives were lost and 2.6 million more people were infected with HIV [1]. To cut transmission, many social, behavioural and biomedical interventions have been developed, tested and tried but have had little impact on the epidemic in most countries. One substantial success has been the development of combination antiretroviral therapy (ART) that reduces viral load and restores immune function. This raises the possibility of using ART not only to treat people but also to prevent new HIV infections. Here we consider the impact of ART on the transmission of HIV and show how it could help to control the epidemic. Much needs to be known and understood concerning the impact of early treatment with ART on the prognosis for individual patients and on transmission. We review the current literature on factors associated with modelling treatment for prevention and illustrate the potential impact using existing models. We focus on generalized epidemics in sub- Saharan Africa, with an emphasis on South Africa, where transmission is mainly heterosexual and which account for an estimated 17% of all people living with HIV. We also make reference to epidemics among men who have sex with men and injection drug users where appropriate. We discuss ways in which using treatment as prevention can be taken forward knowing that this can only be the beginning of what must become an inclusive dialogue among all of those concerned to stop acquired immune deficiency syndrome (AIDS).

Fig. (1)

The epidemic of HIV in adults in South Africa. HIV prevalence (red), incidence (blue), mortality (black), data (red dots).

Fig. (2)

Impact of treatment on HIV in South Africa. People start antiretroviral therapy (ART) when their CD4 cell count is below 200 cells/mm³ (top left), 350 cells/mm³ (top right), 500/ cells/mm³ (bottom left) or if they start immediately, irrespective of CD4 cell count (bottom right). HIV prevalence not on ART (red), data (red dots) and on ART (pink); HIV incidence (blue); AIDS-related mortality not on ART (black) and on ART (grey).

Fig. (3)

Impact of pre-exposure prophylaxis (PREP) on HIV in South Africa. All HIV-negative people are put onto antiretroviral therapy (ART). HIV prevalence (red); data (red dots); HIV incidence (blue); AIDS-related mortality (black).

Fig. (4)

Survival of postpartum mothers in Harare, Zimbabwe (1997 to 2000) as a function of CD4 cell counts without treatment [80,81]. Dots: HIV-positive data; Fitted line: model; Upper horizontal line: asymptote; Lower horizontal line: HIV-negative mortality from data.

Fig. (5)

Reduction in maternal deaths after one year (black) and of infected children at 6 weeks (gray) in South Africa. For illustrative purposes the data are scaled to 1 without antiretroviral therapy (ART); the inset numbers give the annual rates. Starting treatment immediately reduces one-year mortality in mothers from 6.8 thousand to 3.4 thousand and the number of children infected at 6 weeks from 58 thousand to 10 thousand.

Fig. (6)

Impact of antiretroviral therapy (ART) on tuberculosis (TB) in South Africa. People start ART when their CD4 cell count is below 200 (top left), 350 (top right), 500 cells/mm³ (bottom left) or if they start immediately, irrespective of CD4 cell count (bottom right). TB incidence from HIV-negative people (dark blue); including HIV-positive people not on ART (pink) or HIV-positive people on ART (light blue); total (red); data (red dots).

Fig. (7)

Annual probability of infection in discordant couples as a function of viral load for the three studies discussed in the text. Dots are point estimates with 95% confidence limits. Lines are maximum likelihood fits with 95% confidence bands. A: from a meta-analysis [106]; B: from HIV/HSV Transmission Study [109], C: Serodiscordant couples in East and Southern Africa [107]. The asymptotic transmission probabilities are A: 8.3% ± 1.8% per year; B: 3.2% ± 0.8% per year; C: 5.3% ± 1.3% per year. The saturation viraemiae where the extrapolated initial linear increase meets the asymptote, expressed as log₁₀ copies/mL, are A: 4.05 ± 0.33; B: 4.40 ± 0.26; C: 4.56 ± 0.21.

Fig. (8)

Trends in acquired drug resistance (top) and plasma viral load suppression (bottom) in British Columbia [148]. The proportion of HIV-positive people with good viral load suppression increased linearly between 1995 and 2007 (lower line) while the incidence of drug resistance fell exponentially from about 20% in 1995 to 2% in 2007 (upper line).