Randomised controlled trials of circumcision as a preventive measure

Three randomised controlled trials (RCTs) of circumcision in (largely young) HIV-negative adult males were initiated in three different countries in Africa: one in the peri-urban township of Orange Farm in South Africa, one in the urban setting of the largely Luo city of Kisumu in Kenya, and the other in the rural setting of Rakai Province in Uganda on the opposite shore of Lake Victoria.

In each case, the trial was stopped early because the results were so conclusive – demonstrating that circumcision has anything between a 50% and a 75% protective effect against HIV infection in men – that it was considered unethical not to offer circumcision to the control group in advance of the planned trial end date.

The first to report was the South African trial,¹ whose preliminary findings were announced at the Third International IAS Conference on HIV Pathogenesis and Treatment in Rio de Janeiro held in July 2005. This trial, conducted by the French research organisation INSERM, randomised 3273 men aged 16 to 24 to be circumcised at the start of the trial or to be offered circumcision at the end of it 21 months later.

At the point when the trial was stopped, it had become apparent that there were only 35% as many infections in the circumcision arm as in the control arm, implying that circumcision can prevent at least six out of ten female-to-male HIV transmissions.

However, when the results were analysed according to true circumcision status rather than by intervention group, the protective effect went up to 75%. This is because there were ‘crossovers’ between the intervention and control arms – some men randomised to be circumcised were not, and some in the control arm were.

Circumcisions in the intervention arm were carried out by a surgeon under local anaesthesia and with post-operative pain relief given. There had been no deaths or permanent adverse effects in any participant. 31% complained of pain and 15% initially had problems with the changed appearance of their penis.

HIV incidence was measured at three and twelve months into the trial and finally at 21 months though the average follow-up period was in fact 20 months due to the premature termination of the trial. Although all participants received intensive safer sex counselling and condoms, there were 51 HIV seroconversions in the control arm versus 18 in the circumcision arm. This translates as HIV incidences of 2.2% and 0.77% a year respectively. In the control arm there were nine, 15 and 27 new infections at three, twelve and 21 months and in the circumcision arm two, seven and nine.

"This is the first RCT demonstrating a strong protective effect of safe male circumcision", said researcher Bertrand Auvert. He could have added that this was not just the first RCT demonstrating the efficacy of circumcision; it was the first RCT that demonstrated the efficacy of any new protective measure against HIV transmission amongst adults since the early trials of condom and needle-exchange efficacy.

The other two trials, involving 2784 men in Kisumu, Kenya² and 4996 men in Rakai, Uganda,³ had not expected to finish data collection till February 2007 and December 2007 respectively.

However both trials were stopped early, in December 2006, because an interim review of trial data revealed similarly convincing results, though not with quite such a strong protective effect as the INSERM trial. In the Kisumu trial the per-protocol protective effect was 53% while in Rakai it was 51%.

Results from the Kisumu trial were published in The Lancet on 23 February 2007.² By the time the trial was stopped, with an average follow-up time of 24 months after circumcision, 22 men out of 1391 in the intervention group and 47 out of 1393 in the control group had tested positive for HIV. The two-year HIV incidence was 2.1% in the circumcision group and 4.2 in the control group (p = 0.0065); the relative risk of HIV infection in circumcised men was 0.47 which corresponds to a reduction in the risk of acquiring an HIV infection of 53% (95% confidence interval, 22%-72%).

Adjusting for non-adherence to treatment and excluding four men found to be HIV positive at enrolment, the protective effect of circumcision was 60% – the same as that seen in the Rakai trial (see below). Adverse events related to the intervention (21 events in 1.5% of those circumcised) resolved quickly.

A study of the way infections accumulated after follow-up seemed to show that the beneficial effect of circumcision was only seen after twelve months – before this time infection rates in the intervention and control arms were identical, but then started to diverge. This supports other data cited below that people may be at elevated risk of HIV acquisition or transmission in the period immediately after circumcision.

There were some signs of behavioural disinhibition in circumcised men. One was the number of partners reported in the previous six months. There was a linear decrease across visits in the proportion of men in the control group reporting two or more partners in the previous six months, whereas the proportion reporting the same behaviour in the circumcision group fell from month zero to month six, but did not fall thereafter. This was statistically significant (p = ~0.03).

Differences between the study groups were also found for unprotected sexual intercourse (p = 0.0349) and consistent condom use (p = 0.0326), with individuals in the control group being safer. The researchers say that ‘notably greater’ numbers of circumcised men reported other risky behaviours, but that differences were non-significant.

Ronald Gray, lead investigator of the Rakai trial, gave details of this trial to the Fourteenth Conference on Retroviruses and Opportunistic Infections (CROI) in Los Angeles in February 2007.^{4 5}

In the Rakai study the 4996 men randomised were aged 15 to 49, a wider age range than in the other two trials which only randomised young men up to the age of 25.They were randomised either to immediate circumcision or to be offered circumcision at the end of the two-year study. Fifty per cent of the men reported extramarital partners and 40% reported (inconsistent) condom use. Men who turned out to be HIV-positive on screening were referred to a parallel and ongoing study which is studying the effect of circumcision on HIV transmission by positive men.

The study was stopped early when interim analysis showed that in an ‘intent-to-treat’ analysis, the incidence of HIV infection in the circumcision group was reduced by 51%. It was 0.66 per 100 person-years in the circumcised men and 1.33 per 100 person-years among uncircumcised men. This difference was statistically significant (p = 0.007).