Is treatment as prevention starting to work?

One of the earliest demonstrations that expanding treatment in a population has the potential to reduce HIV infections came from Taiwan in 2004.¹

By the end of 2002, the cumulative number of HIV-infected citizens in Taiwan had reached 4390 or about one per 5000 head of population.

After free access to ARVs was established, the estimated HIV-transmission rate decreased by 53% [95% confidence interval, 31%–65%]) and the ratio of new cases to established cases decreased from 3.9 new cases for every ten established cases to 1.8.

Importantly, this appeared to be achieved in a context of no increase in risk behaviour amongst vulnerable populations: there was no statistically significant change in the incidence of syphilis either in the general population or among people living with HIV during this period (the HIV population in Taiwan consists largely of MSM.)

Turning to San Francisco, evidence was presented to CROI 2010 that very high rates of testing and treatment might be starting to reduce HIV incidence amongst gay men there.² (Gay men represent 85-90% of HIV cases in San Francisco, so these data reflect almost completely the population at risk in the city).

Dr Moupali Das from the San Francisco Department of Public Health (DPH) said that the reduction in infections was ultimately due to an increased frequency of HIV testing and a higher rate of linkage to care.

Between 2004 and 2008, the proportion of gay men taking HIV tests who had tested less than a year previously rose from 65% to 72%, and the proportion testing within the last six months from 41% to 53%. It was estimated that during this time the proportion of gay men with HIV unaware of their infection fell from 24% to 14.5%.

The proportion of gay men diagnosed with HIV linked to care rose to nearly 80%, and the proportion of those in care who were on antiretrovirals rose from 78% to 90%, with nearly three-quarters having an undetectable viral load (under 75 copies/ml).

Between 2004 and 2008, Dr Das said, the number of HIV diagnoses in San Francisco fell by 45%, and the average viral load amongst the HIV-positive population (the “community viral load (CVL)”) by 40%, from 23,000 in 2005 to 15,000 in 2008. “Community viral load acts like a virometer,” said Dr Das, “a measure of the temperature of the epidemic.”

At the same time the number of new HIV diagnoses fell from 798 in 2004 to 434 in 2008. The association between reduction in viral load and new diagnoses was statistically significant (p = 0.019). However, it is important to note that this is only a measure of the correlation between viral load and diagnoses: it doesn’t prove one caused the other.

The DPH also estimated that the actual HIV incidence (the true number of new HIV infections, diagnosed and undiagnosed) fell by one-third between 2006 and 2008, from approximately 930 new infections in 2006 to 620 in 2008. However, due to the margin of error in this method of calculating incidence, this was not statistically significant (p = 0.3) so cannot be said to prove that a real decline is yet happening.

The study has one significant limitation in that it could not include the viral load from undiagnosed individuals in its calculation of community viral load.

Interestingly, the reduction in new diagnoses and estimated incidence occurred within a context of significant increases in sexually transmitted infections (STIs) including rectal gonorrhoea and syphilis. The decline in diagnoses and apparent decline in incidence might have been greater if there had not been an increase in STIs, but equally the increase in STIs might be restricted to those already with HIV, due to serosorting practices.

“Our findings support the hypothesis that wide-scale early antiretroviral therapy can have a preventative effect at population level,” commented Dr Das. She said that community viral load (CVL) was a useful ‘upstream’ predictor of the likely number of new infections, and could therefore be used to calculate future resources and prevention needs.

At the same conference, Professor Julio Montaner, the architect of the British Columbia policy, said that an expansion in the diagnosis and on treatment of people with HIV in the province had started to produce modest reductions in HIV diagnoses and in the CVL.³

However, he also told the conference that the ‘second wave’ of increased ARV coverage actually started prior to the adoption of the policy, which in itself does not appear to have further increased treatment access.

Antiretroviral (ARV) coverage started in the province in 1996 and had reached 2500 patients by 1999. After this it reached a plateau, which appeared to be connected to a lot of patients in that era choosing to take treatment interruptions.

Starting from the beginning of 2004, a second wave of treatment uptake began, which continues to this day, and there are now 5000 people in the province on treatment. Many of these people were not drug-naive but re-started treatment after the Centre undertook a campaign of contacting people on treatment interruptions and suggesting they resume.

There is room for considerable further expansion of treatment in British Columbia, as this figure represents less than half of the estimated number of people who have tested HIV-positive, and a third of the estimated total number of people with HIV.

The proportion of patients on treatment with a viral load under 50 copies/ml increased from 66% in 2000 to 88% in 2008.

Since 2004, there has been a modest, but statistically significant, decline in the number of new HIV diagnoses per year, from 440 in 2004 to 370 in 2009. However,  this is entirely accounted for by a decrease in diagnoses in injecting drug users (IDUs), which halved during this period, from 150 in 2004 to 80 in 2009.

Montaner said that the reductions appeared to be driven by antiretroviral take-up, rather than changes in risk behaviour, as British Columbia already has a long history of harm-reduction schemes for IDUs. The reductions coincided with an outreach campaign to get injecting drugs users on to HIV treatment, though the study could not prove that one caused the other.

The proportion of non-IDU patients with a viral load under 500 copies/ml increased from 43% in 2004 to 77% in 2009. The proportion of IDUs with a viral load under 500 copies/ml increased from 34% to 74% – nearly the same as other patients. This represents a considerable achievement in a province with a very specific HIV epidemic with particular concentrations in aboriginal Canadians living in remote communities and injecting drug users.

Montaner also produced an approximate measure of ‘community viral load’: the average viral load within the HIV-positive community at large. Montaner’s way of doing this was to measure the total number of patients ever given a viral load test in the province, minus those known to have died or moved away. This number amounted to 7400 in 2004 and had increased to 10,200 in 2009. He then determined the proportion of all patients who received viral load tests whose viral load was in one of five different viral load strata at the end of any given year.

The absolute number of patients with a viral load test result over 500 copies/ml at last test decreased as a proportion of the total from 65% to 40%. This represents a rough measurement of the proportion of patients likely to be infectious.

“Our results show an association between expanded HAART coverage, decreased provincial plasma viral load, and decreased new HIV diagnoses,” said Montaner.

“Seek, Test, Treat and Retain (STTR) strategies targeting HIV-positive individuals who meet criteria for HAART initiation should proceed expeditiously,” he added.