Daily prophylactic treatment with one of three tenofovir-based regimens did not reduce the risk of HIV acquisition among women at high risk for the infection, according to a randomized, placebo-controlled trial conducted in South Africa, Uganda and Zimbabwe.1 Failure to follow the treatment protocol may have contributed to the lack of efficacy: Although the mean rate of adherence was 86–90% (depending on the method of assessment), tenofovir was detected in only 25–30% of plasma samples from a random sample of participants assigned to one of the intervention regimens.

Several randomized trials have found preexposure prophylaxis with oral or topical tenofovir to be effective in reducing the risk of HIV acquisition, though none of them focused on women of reproductive age. While those studies were in progress, researchers initiated a randomized, placebo-controlled trial to compare the effectiveness of oral tenofovir with that of topical tenofovir in preventing HIV infection and to assess the safety of the treatments.

Women aged 18–45 at 15 sites in the three study countries were eligible for the trial if they had had vaginal intercourse in the prior three months; were using an effective contraceptive (hormonal method, IUD or sterilization); had normal kidney, hematologic and liver function; and were neither pregnant (or planning to become pregnant in the next two years) nor breast-feeding. Participants were randomly assigned to one of five regimens: 300 mg oral tenofovir and a placebo; oral tenofovir-emtricitabine (300 mg tenofovir and 200 mg emtricitabine) and a placebo; two placebos; vaginal 1% tenofovir gel; or vaginal placebo gel. Participants were instructed to use the products every day, and were provided with HIV risk-reduction counseling, adherence counseling, hepatitis B immunization and condoms. HIV testing was performed monthly, and testing for other STIs was done at enrollment and annually thereafter (or if necessary). Other monitoring included monthly interviews and pregnancy testing, quarterly tenofovir plasma testing and biannual pelvic examinations. Adherence was assessed monthly through interviews and in-clinic counts of empty pill bottles, unused pills or unused vaginal applicators, and every three months via audio computer-assisted self-interview (ACASI); the latter was also used to assess participants’ condom use and sexual practices. Study products were withheld (temporarily or permanently) if women became pregnant, started breast-feeding, or had adverse effects or a positive HIV test.

Women were followed for 12–36 months. However, mean follow-up was only slightly more than a year, as the trial was terminated ahead of schedule (in 2012) because of lack of intervention efficacy. Analyses included Cox proportional-hazard models (to assess time to HIV seroconversion), Kaplan-Meier plots (to estimate the cumulative probability of infection), generalized-estimating-equation models (to identify characteristics associated with tenofovir detection) and Cox regression analyses (to identify characteristics associated with HIV acquisition and assess the association between tenofovir detection and HIV acquisition).

In all, 5,029 women enrolled in the trial. Participants were similar across the five study groups: Their mean age was 25, 21% were married, 92% had at least some secondary education, 85% reported condom use at last sex and 71% used an injectable for contraception. On average, women had had 1.5 live births, and they had had 2.5 episodes of vaginal sex in the week before enrollment. Ninety-one percent of participants remained in the study until their final scheduled visit; overall, the study yielded 5,509 person-years of follow-up. Across groups, the mean rate of adherence was 86% when assessed by remaining product count, 90% when assessed in face-to-face interviews and 88% when assessed by ACASI.

The incidence of HIV infection was 5.7 per 100 person-years, and was particularly high among women who were younger than 25 (8.0 per 100 person-years) or unmarried (7.2 per 100). The effectiveness of oral tenofovir was –49.0%, meaning that the risk of infection was 49% greater among women receiving this regimen than among women receiving placebo (hazard ratio, 1.5), though the association was only marginally significant. Infection rates in the oral tenofovir-emtricitabine and the vaginal tenofovir gel groups did not differ from those in the respective placebo groups.

In a randomly selected subcohort, tenofovir was detected in only 30% of quarterly plasma samples from the oral tenofovir group, 29% of samples from the oral tenofovir-emtricitabine group and 25% of samples from the vaginal tenofovir gel group. Moreover, 58% of participants in the oral tenofovir group, 50% of those in the oral tenofovir-emtricitabine group and 57% of those in the vaginal tenofovir gel group had no tenofovir detected in any quarterly plasma sample.

After adjustment for study site, the baseline characteristics associated with tenofovir detection were being older than 25 (odds ratio, 1.6), being married (2.2), having an independent income (1.4) and having more than one child (1.8). Some supplementary analyses suggested that HIV risk was reduced among participants with better adherence; for example, vaginal tenofovir gel users who had detectable levels of tenofovir at their first quarterly visit were less likely to acquire HIV than were those without detectable levels (hazard ratio, 0.3). Finally, the risk of adverse effects was similar across study groups, with the exception that mildly elevated serum creatinine levels were more common among women in the oral tenofovir-emtricitabine group than among those in the oral placebo group (1.3% vs. 0.2%).

The researchers note that although the lack of efficacy seen in the study contrasts with the findings of most previous trials, it is consistent with results of another trial in which user adherence was low. They add that a limitation of the study is that they were unable to measure or control for women’s likelihood of HIV exposure, and that this likelihood may have differed between women who used the study products and those who did not. They suggest that the lack of both efficacy and adherence has “implications for biomedical prevention research” and “for the implementation of interventions with proven efficacy”; in particular, the results “reaffirm the need for effective and acceptable prevention interventions for women at high risk for sexual acquisition of HIV-1 and suggest that more accurate measures are critical for the estimation of product use during biomedical HIV-1–prevention trials.”—L. Melhado


1. Marrazzo JM et al., Tenofovir-based preexposure prophylaxis for HIV infection among African women, New England Journal of Medicine, 2015, 372(6):509–518.