Study participants were recruited from areas within each city that were considered high HIV transmission areas, including markets, bars, and hotels. Although we did not specifically ask as part of the clinical trial procedures if the participants were sex workers, most exchanged sex for money. Special ethical considerations were taken into account because of the potential vulnerability of this population. We developed strategies to protect the confidentiality and autonomy of the participants, increase/ensure comprehension of the informed consent and research methods, and promote access to resources and services during and after the trial.
We enrolled HIV-antibody-negative women 18 to 35 y old who were at risk of HIV infection by virtue of having an average of three or more coital acts per week and four or more sexual partners per month. Participants had to be willing to use the study drug as directed and participate for up to 12 mo of follow-up. Because TDF has been associated with rare episodes of renal disorders, increased liver enzymes, and hypophosphatemia, participants were also required to have adequate renal function (serum creatinine < 1.5 mg/dl), liver function (aspartate aminotransferase [AST] and alanine aminotransferase [ALT] < 43 U/l), and serum phosphorus (≥2.2 mg/dl) at their screening visit. Since no adequate and well-controlled studies of TDF have been conducted in pregnant women, participants were not enrolled if pregnant or breastfeeding, or wishing to become pregnant during the 12 mo of study participation.
During recruitment, study staff explained the general purpose of the study and the eligibility requirements. If eligible, women were referred to the study clinic at each of the three study sites. At the screening visit, women completed written informed consent, received HIV pretest and condom counseling, and underwent oral mucosal transudate (OMT) rapid HIV testing. All participants received HIV post-test counseling, a physical and pelvic examination, urine pregnancy tests, and assessment of hepatic and renal function. Women with reactive OMT rapid HIV tests received ELISA to confirm HIV status. Potential participants were asked to return 4 wk after their screening visit to receive the results of their hepatic and renal function tests and, if applicable, a confirmatory HIV test. At this second visit, participants signed or marked a consent form for enrollment, received HIV counseling, provided urine for pregnancy testing, and provided another OMT sample for HIV testing.
All participants provided written informed consent in their preferred language. Illiterate participants were read the informed consent forms verbatim in the presence of a witness, and provided a mark or thumbprint in lieu of signature. The study protocol and informed consent forms were approved by the Ghana Health Service Ethical Review Committee, Accra, Ghana; the National Ethics Review Committee, Ministry of Public Health, Yaoundé, Cameroon; the University of Ibadan/University College Hospital Institutional Review Board, Ibadan, Nigeria; and the Protection of Human Subjects Committee, Family Health International (FHI), Durham, North Carolina, United States.
At each monthly follow-up visit, participants underwent OMT HIV and pregnancy testing, adverse event (AE) assessment, risk reduction counseling, and study drug and condom re-supply. Clinic staff counseled participants to take one pill every day, distributed condoms, and emphasized that condoms should be used for all sexual contacts with all partners because they were using a product of unknown or no effectiveness for preventing HIV. Participants responded to structured questionnaires on their sexual behavior and their experience taking the pills during the previous 30 d, and were reminded of study concepts discussed during the informed consent process. At months 1, 3, 6, 9, 12, and as needed, participants underwent physical examination and blood was drawn for laboratory assessment of hepatic and renal function. We discontinued product use for any participant who had a reactive rapid HIV or positive pregnancy test result. Pregnant women were allowed to resume study drug use after their pregnancy had ended, providing they were not breastfeeding. Study staff at each study site referred and offered to escort participants who became infected with HIV during the study to HIV-related psychological, social, and medical services, such as viral load, CD4 level, and HIV resistance testing in their communities, as well as antiretroviral drug provision when needed. Immediately after a participant missed a scheduled follow-up appointment, study staff made up to three attempts to contact that participant and reschedule the clinic appointment (ideally to occur within 1 wk of the original appointment).
Objective and Interventions
The objective of this trial was to investigate the safety and preliminary effectiveness of a daily dose of 300 mg of TDF versus placebo in HIV-uninfected women.
Protocol-defined primary safety endpoints included grade 2 or higher serum creatinine elevations (>2.0 mg/dl) for renal function, grade 3 or 4 AST or ALT elevations (>170 U/l) for hepatic function, and grade 3 or 4 phosphorus abnormalities (<1.5 mg/dl).
After the study began, the TDF prescribing information was amended to reflect a concern that people with chronic hepatitis B virus (HBV) infection are at risk of developing reactivation (i.e., flares) of hepatic disease after stopping use of TDF. To monitor for potential flares, we amended our protocol to add HBV surface antigen (HBsAg) testing for all enrolled participants at the time of product discontinuation, and ALT/AST monitoring for 3 mo after product discontinuation among HBsAg-positive participants. This amendment was implemented only at the Ghana site in August 2005, which was the only active study site (i.e., on drug) at the time.
The effectiveness endpoint was infection with HIV-1 or HIV-2, measured by detecting antibodies in OMT (OraQuick ADVANCE Rapid HIV-1/2 Antibody Test, Orasure Technologies, http://www.orasure.com
) and confirmed by an enzyme-linked immunosorbent assay (ELISA) (Genetic Systems HIV-1/HIV-2 Plus O ELISA, Bio-Rad, http://www.bio-rad.com
) or Western blot (Genetic Systems HIV-1 Western Blot, BioRad) from a finger prick or blood serum specimen. Discordant results between the rapid antibody and the ELISA assays were resolved with Western blot testing.
We designed the study to have 90% power to conclude with 95% confidence that TDF reduced the rate of HIV infection by 50% (i.e., power to obtain a one-sided 95% upper confidence limit for the rate ratio that is less than 0.5) if the true rate of reduction due to TDF was at least 83%. The planned sample size was 1,200 participants (400 per site), with 12 mo of follow-up for each participant. Based on prior work with high-risk women in Cameroon [19
], we assumed that the HIV incidence rate in the placebo group would be no less than five per 100 person-years, and that follow-up would be at least 80% at 12 mo. Under these assumptions, we expected that we would reach the required number of incident HIV infections (30) to achieve 90% power. Owing to premature closures of the Cameroon and Nigeria study sites, however, the planned person-years of follow-up and study power could not be achieved. Consequently, the primary effectiveness analysis was revised to a traditional two-sided test of non-zero effectiveness at the 0.05 significance level. The decision to change the primary effectiveness analysis was made by the project leader and statistician before unblinding the study.
For non-HIV safety outcomes, the planned sample size would have provided approximately 90% power to detect a doubling in the proportion of women experiencing any particular adverse reaction, if the proportion experiencing the reaction in the placebo arm was 0.05 or more. However, due to early termination of the study in Cameroon and Nigeria, we had only approximately 70% power to detect such a difference.
Randomization and Blinding
A randomization manager not otherwise involved in the study developed a random allocation sequence using a permuted-block design stratified by site, with random block sizes of 12, 18, and 32. A copy of the randomization list was sent to the manufacturer, who filled each drug bottle with a 30-d supply of TDF or placebo (12 bottles per randomization number). Placebo tablets were identical to the TDF tablets in size, shape, color, and taste. Each drug bottle was marked with a sequential randomization number, but no product identifier. Participants, field study staff, monitors, statisticians, and other FHI staff involved in the trial were blinded to drug assignment. Study staff assigned each eligible participant the next sequential number, and gave her the first month's supply of study drug after she had fully qualified for the study and signed or marked the enrollment consent form.
Primary safety analyses were based on time to grade 2 or higher abnormalities for creatinine or grade 3 or higher abnormalities for ALT, AST, and phosphorus, by site, using exact two-sided log-rank tests computed in StatXact Version 7 (Cytel, http://www.cytel.com
). Because of irregularities in the performance of the laboratory in Nigeria, data from that site were excluded from the primary safety analyses. The project leader and study statistician made this decision before unblinding the study.
The primary effectiveness analysis was based on an exact log-rank test of no difference in the distribution of HIV-free survival times, with data pooled across all three sites. Each laboratory abnormality and HIV infection date was estimated as the midpoint between the date of the first positive test result and the previous, negative test date. A right censoring time of 12 mo was applied in Ghana, corresponding to the treatment regimen. For all participants in Cameroon and Nigeria, the censoring date was the date of permanent product withdrawal (4 February 2005 and 7 March 2005, respectively). Secondarily, exact confidence intervals for the relative risk of HIV, laboratory abnormalities, and AEs within system organ classes were computed under a Poisson assumption for the event rates in each treatment group. All primary and secondary analyses were based on two-sided tests and conducted at the 0.05 significance level.
Protocol-specified interim safety analyses occurred in April and December 2005. An independent Data Monitoring Committee, with access to treatment assignments, reviewed AEs and primary safety and effectiveness outcomes. The committee did not recommend changes in study procedures after either analysis.
FHI monitors (trained in Good Clinical Practice) visited the sites regularly to review study eligibility, informed consent, protocol compliance, laboratory procedures, source documents, and AEs. We attempted to get original hospital records, when available, for serious adverse events (SAEs) and deaths.