Microbicide Shows No Protective Effect in South African Clinical Trial

H. Ball

First published online:

Carraguard, a microbicide effective at blocking transmission of recombinant simian HIV among rhesus macaques, failed to show a protective effect among women in a multisite, placebo-controlled, double-blind trial in South Africa.1 The study was the fifth major trial to assess the effectiveness of a candidate microbicide, and it differed from its similarly unsuccessful predecessors in part because investigators took steps to more accurately track the microbicide's use. Although fewer seroconversions occurred among women who used the microbicide during heterosexual vaginal intercourse compared with those who used the placebo gel (3.3 vs. 3.8 per 100 woman-years), the difference was not statistically significant.

The study participants were sexually active, HIV-negative women aged 16 or older who were recruited from health clinics, malls, churches and other public venues in Isipingo, Soshanguve and Gugulethu, South Africa. Those who were pregnant or planned to become pregnant in the next two years, had been pregnant in the prior four weeks, had ever had a positive Pap smear or had used illicit injection drugs in the past year were excluded. The 6,202 eligible, consenting participants were randomly assigned to receive either Carraguard, a seaweed-based gel that prevents in vitro mucosal transmission of HIV-1 (and possibly other STIs), or the placebo, a methylcellulose gel chosen for its safety, lack of microbicidal properties and physical similarity to Carraguard.

Participants were instructed to visit their clinic at enrollment, month one, month three and every third month thereafter for a total of 9–24 months. They were tested for HIV at each follow-up visit; they also received a pelvic exam, counseling on HIV risk reduction and family planning, and a pregnancy test at every visit, as well as periodic tests for STIs other than HIV. Those found to be HIV-positive or pregnant were disqualified from further participation in the study. At each visit, women received a supply of their assigned gel and as many male or female condoms as they desired. They were instructed to insert the gel prior to vaginal sex and to use a condom with the gel during sex.

To assess whether the women were using the gel, the researchers asked participants to return all opened and unopened gel applicators at each visit, and they used bar codes to verify that women had returned the same applicators that had been issued to them. In addition, the researchers applied a dye that reacts to vaginal mucous to each opened applicator to verify that it had been inserted into the vagina.

The main outcome variable, time to seroconversion, was defined as the length of time between a participant's first clinic visit and the midpoint between the positive HIV test and the most recent negative test, plus one day. The researchers also measured the incidence of other STIs and vaginal infections and the safety of the microbicide and the placebo.

Women in the Carraguard and placebo groups were similar in their social and demographic characteristics, and they reported similar levels of risky sexual behaviors and contraceptive use. In both groups, most participants were aged 25 or older (63%) and single (63%). On average, participants had had about 11 years of education and earned less than US$150 per month. Twenty-six percent had an STI at baseline. The two groups had experienced similar levels of partner abuse (34–36%) and forced sex (10–11%) in the prior three months; 9% in each group said they had multiple sex partners and 17–18% reported that their main partner had other partners. Among women with multiple sexual partners, 60–63% had used condoms during their most recent sex act, compared with 33% among all women with a steady partner.

During the trial, the frequency of vaginal sex and use of condoms and gel were similar in the two groups. Women engaged in an average of 2.5 sex acts per week; 96% reported using gel at their last sex act, and 64% reported using a condom. Participants returned 92% of the gel applicators, of which 70% had been opened. Dye tests showed that 61% of opened applicators had been inserted vaginally. The researchers estimated that participants had used gel during only 42% of their sex acts.

A total of 285 participants seroconverted. Although the incidence of seroconversion was slightly lower in the microbicide group than in the placebo group (3.3 vs. 3.8 per 100 woman-years), neither this finding nor the difference in time to seroconversion was statistically significant. Participants who were younger, had had an STI at screening or were from the Isipingo research site had elevated odds of contracting HIV. No safety concerns arose, nor did the two groups differ in rates of pregnancy or STIs.

While the results indicate that "Carraguard is unlikely to have a meaningful protective effect as used by study participants," the researchers point out that several factors may have contributed to the lack of efficacy: participants' low level of adherence to the gel-and-condom regimen; HIV transmission through anal sex, an act that may have been under- reported; and the possibility that the methylcellulose gel was not a true placebo and thus offered some protection against the virus. Exit interviews with a subset of participants indicated that nonuse of gel was most commonly due to participants' running out of or forgetting to use the gel, leading the researchers to suggest that "even a highly efficacious coitally-dependent product" may have "insufficient effectiveness in real-life settings." However, they stress that despite the disappointing results in this and previous microbicide trials, "the search for female-controlled HIV-prevention methods must continue."—H. Ball


1. Skoler-Karpoff S et al., Efficacy of Carraguard for prevention of HIV infection in women in South Africa: a randomised, double-blind, placebo-controlled trial, Lancet, 2008, 372(9654):1977–1987.