Treatment reduces infections by over 90%: a theme that is here to stay

Simon Collins, HIV i-Base

CROI was important this year because of the profile given to further studies supporting the role of treatment as prevention. Together they support the argument that universal treatment is perhaps the most powerful prevention tool we are likely to have for many years, perhaps with the potential to even eradicate the virus on a population level.

In a lecture prior to the main conference, Brian Williams from the South African Centre for Epidemiological Modelling and Analysis, Stellensbosch, detailed the modeling data for the direct and indirect impact of ARVs on prevention, [1] elaborating on the research paper published last year in the Lancet. [2]

At its most optimistic, this includes the potential for universal treatment to eliminate new infections in South Africa within 5-10 years on a cost neutral budget, at the same time saving millions of lives (and preventing millions of new infections). The science on which the model is based shows an impact on dramatically reducing infections that few can ignore.

The epidemiology for the model included low HIV infectivity (~0.001 per heterosexual encounter), 10-fold individual variability in infectivity, a slow epidemic doubling time (~1-3 years), a long period of potential infectiousness (5-15 years) and an average case reproduction number (~7 additional people infected per case): leading to a calculation showing that virtual eradication of HIV could be achieved if transmission could be reduced by 7-fold.

Viral load is commonly reduced by 10,000 times on treatment, and although infectivity reduces in smaller proportions (roughly in relation to the cube route of viral load), the net impact of treatment on infectivity was estimated to be a 96% reduction.

The impact on reducing TB and for continuous treatment after pregnancy were also included, and for interventions based only on PrEP alone or in combination with ARVs. For South Africa, the model was based on a conservative treatment programme, treating at a CD4 count of 200 cells/mm3, but similar costs and benefits were shown when starting universal treatment at 350, 500 or even at diagnosis. The initial outlay (an adjusted US $60 billion) was compensated by lower cost of hospitalisations and reduced new infections, and saved an additional 3 million lives over 40 years, at stable costs.

The discussion after the presentation stressed the need for pilot operational research on each aspect of a universal treatment model, including willingness to test, virological response rates with earlier treatment, the actual impact on transmission – and the need to develop new heath structures to allow such scale-up.

A first step in confirming treatment reduces HIV transmission in real world settings was shown in results from the Partners in Prevention HSV/HIV Transmission (PARTNERS) Study in over 3400 serodifferent heterosexual couples in seven southern African countries (Botswana, Kenya, Rwanda, South Africa, Tanzania, Uganda, and Zambia).  The HIV-positive partner was a man in 32% and a woman in 68% of couples. [3]

This study previously reported that HSV therapy with daily acyclovir failed to protect against HIV infections, explained by a massive increase in localised CD4 target cells, and persistence for up to two months after the healing of HSV lesions.

All HIV-positive partners entered the study with CD4 counts >250 cells/mm3 and were not on treatment. Over two years, approximately 10% of study participants required HIV treatment for their own care, and this allowed for the HIV transmission rates to be compared by use of ARV treatment. Intensive risk reduction support was supplied throughout the study, to minimise HIV risk for the HIV-negative partners.

People with more advanced HIV at baseline were more likely to start treatment; with higher baseline viral loads (mean 4.4 vs 3.9 log copies/mL, p<0.001), and lower CD4 counts (375 vs 540 cells/mm3, p<0.001). A higher proportion of men than women (12% vs 9%, p=0.01) strated treatment, at slightly lower median CD4 counts (192 vs 204 cells/mm3, p=0.05). People starting treatment were also older (mean 35.2 vs 32.7 years, p<0.001).

ART was initiated at CD4 counts <200 cells/mm3 in 52% patients, between 200 and 349 cells/mm3 in 33%, and >350 cells/mm3in 15% (30% of this group were for prevention of mother to child transmission).

New HIV infections were detected in 151 of the HIV-negative partners, over 24 months of follow-up, with testing and prevention support provided every 3 months. Phylogenetic analysis suggested that slightly less than one third (43/151) of the infections were not from the relationship partner. Five cases were excluded from the transmission analysis due to uncertain use of ARVs.

This left an overall transmission rate in 103 remaining transmissions of 2.1%.

Of these, 102/103 were in the non-ARV group (102/4558 person years; rate 2.24 95%CI 1.82-2.72) compared to 1/103 from partners using ARV treatment (1/233 person years; rate 0.37 95%CI 0.09-2.04). This produced an unadjusted relative risk of 0.17 (p=0.037), which became even more significant when adjusting for time on study and CD4 count, showing a 92% reduction in risk: RR=0.08 (95%CI 0.002, 0.57, p=0.004).

The single transmission case occurred in someone whose partner started treatment 18 days before the 9 month assessment, when they were still HIV-negative (details on whether this was by HIV-antigen or PCR testing were not provided), but who seroconverted by the month 12 evaluations. Viral load was undetectable at month 12 in the HIV-positive partner.

Details on CD4 count in the HIV-positive partner showed transmissions at all CD4 levels, with a considerably higher risk when the partner had a count <200 cells/mm3 (rate = 8.79 vs 2.79 at 200-350 and 1.7 at 350-500).

This is likely to be an indirect marker of higher viral load relating to more advanced infections, but surprisingly, the presentation provided no further information on viral load levels of the source partner, other than showing that after a median of 7 months treatment (IQR 3-12months) the median viral load dropped to undetectable, indicating excellent responses.

Importantly, and perhaps showing the positive results from the behavioural interventions, the percentage of visits at which people reported unprotected sex dropped from 6.2% to 3.7% at the pre- and post-treatment visits, respectively, with no change in frequency of sex.

Two other studies at CROI, in a largely MSM population in San Francisco, supported the impact of ARVs to reduce transmission.

Moupali Das-Douglas and colleagues from the San Francisco Department of Public Health and the University of California presented results from a model that estimated values for average and total community viral load (CVL) from 2004-2008 and then compared these with the expected and actual number of new diagnoses over the same period. [4]

Average CVL was defined as the mean of the most recent viral load of all reported HIV-positive individuals in a particular population, divided by the number of reported HIV-positive individuals in the population. Total CVL was the sum of the most recent viral loads of all HIV-positive individuals in a particular population.

The context for this study was an effective ‘test and treat’ programme that from 2004 to 2008 increased the percentage of MSM testing within 12 months from 65% to 72% and within 6 months from 41% to 53%. The percentage of HIV-positive MSM unaware of their status dropped from 24% to 14.5% (comparable UK figures vary from 30-50%). By 2008, 90% of patients in care were on HAART, with 72% virologically suppressed (<75 copies/mL).

The decreases in mean CVL and reductions in actual diagnoses (from 798 in 2004 to 434 in 2008) were both statistically significant (p=0.005), as were the decreases in total CVL (p=0.019) and percentage of virologically suppressed patients (p=0.002). The presentation acknowledged that a limitation in these results is that cases may be diagnosed chronic rather than new infections, which was addressed in methodology for expected and actual incidence rates.

However, using a more conservative meta-regression analysis (different to the reported abstract), the 30% reduction in CVL and almost 40% reduction in incidence (rather than cases) was not significant (p=0.3) due to the degree of imprecision in the estimates.

While this makes it too early to link CVL with incidence, the reductions in newly diagnosed and reported cases, at the same time as increased testing, greater ARV coverage and greater virological suppression strongly support close following of subsequent data from this model.

In a related poster, Edwin Charlebois and colleagues modeled the impact of earlier treatment and broader test and treat programmes in San Francisco, suggesting that HIV prevalence could fall from the current 25% to around 10% by 2030 if the programme shifted to universal test and treat. [5]

As this issue of HTB went to press, a policy shift in San Francisco to offer HIV treatment to all newly diagnosed patients, regardless of CD4 count or viral load, was announced by public health officials. [6]


The positive correlation between viral load and risk of transmission for every route, whether sexual, from shared injection equipment, during pregnancy, at birth and from breast milk, and from needle stick exposure to health workers, is now convincingly demonstrated. 

For some of these transmission routes, antiretroviral treatment to reduce viral load is already widely used to reduce transmission (principally for mother to child transmission, PEP and PEPSE).

Treatment dramatically extends life, reduces morbidity and should now be additionally valued for reducing transmission. An estimated 70% of HIV-positive people globally in need of immediate treatment for their own care are still unable to access it.


Unless stated otherwise, all references are to the Programme and Abstracts of the 17th Conference on Retroviruses and Opportunistic Infections. 16-10 February 2010, San Francisco. All oral abstracts are available as webcasts.

  1. Williams B and Dye C. Put your money where your model is: ART for the prevention and treatment of HIV/AIDS.  Webcast: Guiding the global response. Tuesday 2.30pm.
  2. Granich RM et al. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. The Lancet, Volume 373, Issue 9657, p48- 57, 3 January 2009. doi:10.1016/S0140-6736(08)61697-9.
  3. Donnell D et al. ART and risk of heterosexual HIV-1 transmission in HIV-1 serodiscordant African couples: a multinational prospective study. 17th CROI 2010. Oral abstract 136.
    Webcast: New Insights into Transmission and Acute Infection. Friday 9.30am.
  4. Das-Douglas M et al. Decreases in community viral load are associated with a reduction in new HIV diagnoses in San Francisco. 17th CROI 2010. Oral abstract 33.
    Webcast: Testing and Transmission. Wednesday 9.30am.
  5. Charlebois E et al. Effect of Expanded ART Strategies on the MSM HIV Epidemic in San Francisco. 17th CROI 2010. Poster abstract 996.
  6. San Francisco endorses starting HIV treatment immediately after diagnosis. (5 April 2010)

Links to other websites are current at date of posting but not maintained.