Introduction

Most adult HIV infections in Africa are due to heterosexual transmission [1], and being in a stable HIV discordant sexual relationship is associated with a 10-fold higher risk of HIV transmission than being in a concordant HIV-negative relationship [2,3]. HIV-uninfected individuals in discordant couple relationships are therefore among the most at risk populations (MARPs). HIV serodiscordant couples enrolled in an HIV vaccine feasibility study in Masaka, Uganda, had an HIV incidence rate of 4.3 and 4.4 per 100 person years (PY) in men and women respectively [4].

At the peak of the HIV epidemic, Uganda adopted the promotion and dissemination of several prevention strategies to control HIV transmission including abstinence, being faithful to one’s partner, reducing the number of sexual partners, treatment of sexually transmitted infections (STI), HIV voluntary counseling and testing (plus sharing of results with partners) and consistent and correct use of condoms [5]. These strategies helped to reduce HIV prevalence [6]; however, they have limitations. HIV prevention programs that focus on reducing the number of sexual partners, use of condoms during casual sex and increased fidelity among married partners are not likely to directly decrease the risk of HIV transmission among persons already living in HIV serodiscordant relationships [3]. Therefore, research into new approaches to HIV prevention particularly in HIV discordant couples remains critical.

Recently, several trials of antiretroviral pre-exposure prophylaxis have shown major reductions in HIV acquisition. In the Pre-exposure Prophylaxis Initiative (iPrEx) study - a randomized multinational clinical trial among men who have sex with men (MSM) - daily fixed dose combination regimen of tenofovir disoproxil fumarate combined with emtricitabine (TDF/FTC), reduced HIV acquisition by 44% overall [7]. Efficacy correlated with adherence and detectable drug levels. Pill use on 90% or more of days was associated with 73% efficacy, while detectable drug levels were associated with 92% efficacy. Subsequent pharmacokinetic modeling of the iPrEx data suggest that 7 days per week dosing could achieve 99% efficacy in prevention of HIV infection among MSM, while 4 days per week could still lower risk by 96% [8]. In the TDF2 study – a randomized trial conducted in Botswana among young HIV-uninfected men and women - daily use of TDF/FTC reduced HIV acquisition by 62% [9]. In the Partners PrEP trial – a randomized multinational trial in HIV discordant couples in Kenya and Uganda - daily PrEP of either TDF alone or combined as TDF/FTC reduced the risk of HIV acquisition by 67-75% [10]. Having detectable plasma tenofovir levels was associated with 86 and 90% reduction in HIV acquisition, for the TDF and TDF/FTC groups respectively [11]. Interestingly, two other randomized PrEP trials in at-risk women failed to find a reduction in risk of HIV infection in the treatment group. The FEM-PrEP trial of oral TDF/FTC in at-risk African women was halted early due a likelihood of being unable to demonstrate difference in HIV seroconversion based on interim data safety monitoring board determination, as was the case for oral and vaginal tenofovir arms of the VOICE study [12,13]. The explanations for these contradictory results are not fully understood; however, low adherence appeared to play an important role. In the FEM-PrEP trial, adherence by self-report and pill counts were high, but plasma drug levels revealed that only 15-26% of samples from HIV seroconverters actually had tenofovir detected, as well as only 26-38% of non-seroconverting controls [14]. Thus, study medication adherence was too low to assess its protective effect in the trial. In 2012, the US Food and Drug Administration approved daily Truvada for HIV prevention in individuals at high risk of sexual HIV infection [15].

The above trials all included daily oral regimens which may be challenging as a population-level HIV prevention method. Daily PrEP may be limited by adherence challenges, as well as acceptability, toxicity and cost concerns, although most of the blinded large efficacy trials have shown it to be acceptable and safe. An intermittent dosing regimen may be an alternative strategy if able to achieve adequate drug levels required for HIV prevention. The biological characteristics of TDF and FTC favour usage of an intermittent PrEP regimen. Both TDF and FTC are phosphorylated to active forms that have long half-lives in plasma and peripheral blood mononuclear cells (PBMCs) of ≥60 and 39 hours respectively [16,17]. Additionally, in studies using macaques as animal models, intermittent PrEP, given 2 hours before and 24 hours after each weekly virus challenge, protected animals against SHIV as well as daily PrEP [18].

In this randomized, double-blind study, we evaluated safety, acceptability and adherence to an intermittent PrEP regimen with TDF/FTC compared to a daily regimen among men and women living in HIV serodiscordant couple relationships. The intermittent regimen was defined as twice weekly dosing (Mondays and Fridays) plus a coitally-dependent dose, but not exceeding one pill per day. This regimen was selected based on primate challenge data described above, which suggested that having steady state drug levels in addition to dosing soon after exposure was important for protection [18].

This study was conducted and completed before the release of any PrEP efficacy studies referenced above. A parallel study with the same design was conducted in Kenyan MSM and female sex workers (FSW) and results published elsewhere [19].

Methods

The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1 and Protocol S1.

Results

Discussion

In this study of both daily and intermittent PrEP regimens among HIV serodiscordant couples, intermittent PrEP was found to be acceptable and safe. Adherence to daily and fixed intermittent dosing was high, but significantly lower for post-coital dosing. To our knowledge, this is the first study to report on safety, adherence and acceptability of an intermittent oral PrEP regimen among heterosexual HIV serodiscordant couples.

Both intermittent and daily PrEP regimens were well tolerated, with a similar proportion of volunteers reporting mild and moderate adverse events in each regimen group. This observation is in agreement with reports from other larger oral daily PrEP studies [7,9,10,12,25,26]. There were similar proportions of volunteers with gastrointestinal complaints among active and placebo, and between daily and intermittent groups in the study. This differed from the iPrEx study that reported more frequent moderate (grade 2 and above) nausea and unintentional weight loss in the active group [7]. This difference could have been due to a shorter follow-up period and fewer volunteers involved in the current study. A few volunteers in the active group had self-limited elevated creatinine (mild to moderate), and several volunteers in both active and placebo groups had reduced creatinine clearance. This finding was similar to reports from other PrEP trials [7,9,26].

Median adherence rate as measured by MEMS for daily use was significantly higher than that for fixed doses within the intermittent regimen; whereas adherence to coitally-dependant dosing was considerably worse. Despite the statistical difference between daily and intermittent fixed arm, adherence rate was in the 90’s for each. There was no change in adherence after adjusting for reported curiosity openings or pocket doses. Based on MEMS data, volunteers in the daily group took the pill on almost every day of the week. This high adherence is similar to that reported in Partners PrEP trial (92-99% as measured by clinic-based pill counts, unannounced pill counts and MEMS) [27] – (Haberer in press) and higher than that reported in iPrEx (89-95% by clinic count and self-report) [7]. In contrast, adherence to both the daily and fixed component in the same intermittent regimen was lower among MSM and FSW in a parallel trial in Kenya. Median MEMS adherence rates were 83% (IQR: 63-92) for daily dosing and 55% (IQR: 28-78) for fixed intermittent dosing, while adherence to any post-coital doses was 26% [IQR: 14-50] [19]. The higher adherence in the current study may be due to providing PrEP in the context of stable, socially supported relationships as well as the uninfected partner’s hope of maintaining good health while at the same time preserving their serodiscordant relationship, as proposed within the Partners PrEP study [28]. These factors may be less likely to play a role in the MSM population studied in the Kenyan trial [29].

Intermittent fixed dosing, at least in this discordant couple population, appears to have comparable adherence to daily dosing, suggesting that intermittent PrEP regimens may be feasible in certain populations. However, the number of doses per week necessary for >90% efficacy is likely to be greater than the 2 doses evaluated in this study. Using data from the iPrEx study, pharmacokinetic modeling suggested 97% efficacy for 4 doses per week, which dropped to 76% with 2 doses per week [8]. These data, derived from protection against HIV infection in MSM, may not be readily applicable to heterosexual transmission. Comparable data on efficacy of less than daily dosing for heterosexual transmission are not available yet, but are anticipated from the Partners PrEP study.

Median post-coital adherence as measured by MEMS was low at 45%. Post-coital dosing among MSM and FSW in the Kenyan study was even lower at 26% [19]. These findings are consistent with the incomplete adherence observed with a pre- and post-coital dosing strategy for vaginally administered tenofovir microbicide gel in the CAPRISA 004 microbicide trial. In that study, which observed a 39% reduction in HIV incidence in the tenofovir group compared to the placebo group, the median adherence was 60%, and 42% of the women were classified as having <50% adherence to two doses of gel for self-reported sex acts in the last 30 days [30]. The low post-coital adherence in the current study could have been due to behavior and/or measurement error. It may be easier to remember to take pills daily, because a routine behaviour is established. Unfamiliarity with use of SMS to collect sexual behaviour data coupled with technical difficulties (network outages and a user interface with multiple questions) may have affected post-coital adherence monitoring even though the technology appeared to be relatively well accepted by study volunteers. Quantitative data collected such as reasons for missed pills, travel away from home and pill sharing did not explain why post-coital dosing adherence was lower. Qualitative data from a parallel trial of daily and intermittent PrEP among MSM and FSW in Kenya identified several factors influencing post-coital dosing adherence: alcohol use around the time of sex, moving locations frequently and transactional sex work [28].

Qualitative data collected during focus group and individual interviews in the present trial may be informative and are being prepared for publication.

Measurement of adherence is challenging with self-report often providing overestimates when compared to objective measures, such as electronic monitoring, unannounced home-based pill counts and random drug levels [31-36]. Indeed in this study, adherence to post-coital dosing as measured by the self report calendar and SMS report of sexual events alone (without MEMS) was very high in comparison with MEMS. MEMS adherence, however, is also limited by the fact that it records bottle opening rather than actual medication use. Therefore, the low rate of post-coital pill use as measured by MEMS in this study could reflect low adherence to PrEP or the inaccuracy of the methods used to measure it. Other PrEP trials have also found discrepancies between conventional methods such as self-report and biologic measures [7,34]. Adherence rate in the iPrEx trial as measured by clinic pill count and self report was high at 93%, yet drug levels showed that only 50% of volunteers were actually swallowing the pills. Subsequent analyses found low levels of drug detection, with only 18% of volunteers with drug levels reflective of daily use [37]. In addition, self-report and clinic pill counts were poorly predictive of drug exposure [36].

Testing for drug levels was included in this study and the findings will be published when available. Participants’ actual use of trial interventions is often considerably lower than their reported use hence, raising the issue of social desirability. Low adherence is likely to compromise potential efficacy of the intervention and may make interpretation of trial data difficult; since intent-to-treat efficacy and as-treated efficacy will be different. Low adherence may also limit detection of adverse events therefore, misrepresenting safety concerns.

The main reasons given for missing pills were volunteers not having pills with them while away from home and forgetfulness, and these reasons were similar for both daily and intermittent regimen groups. These reasons are similar to those cited in a phase 2 oral PrEP trial conducted in West Africa [38].

Acceptability of oral PrEP was high with 99% of participants reporting willingness to use the pills most of the time or all the time if PrEP was shown to be safe, effective, inexpensive or free. There was no significant difference between those willing to use intermittent (97%) and daily regimens (100%).

There was no change in HIV sexual risk behaviour during the study. Although it is theoretically possible that individuals may feel protected and abandon condom use and other protective measures, none of the oral or topical PrEP trials to date have observed increased HIV risk behavior that would suggest risk compensation [7,9,10,14,30]. Despite encouraging data from animal studies, we found no evidence of HIV-specific immune responses as measured by a validated IFN-γ ELISPOT in volunteers on PrEP It is possible that a longer duration of PrEP and/or more intensive mucosal exposures are necessary to develop these responses, or that too few volunteers were exposed to HIV in this trial. Alternatively, if PrEP-related protective responses do exist, they may be located in the mucosa and may not be evaluable with T-cell assays in PBMCs such as IFN-γ ELISPOTs.

Summary

Our study demonstrated that FTC/TDF was safe when taken either daily or intermittently for HIV pre-exposure prophylaxis by HIV-uninfected volunteers living in coupled serodiscordant relationships. Adherence was high to both daily and fixed intermittent dosing, but not for coitally-dependent dosing, the measurement of which may have been imperfect. These data, together with findings from the Kenyan intermittent PrEP trial [19], suggest that post-coital dosing adherence will be difficult to measure accurately. If the low post-coital adherence observed in these two studies is accurate, post-coital dosing is unlikely to be effective in these populations.

The study was limited by the short follow-up period and relatively small number of volunteers which hamper generalizability of results. Volunteers with creatinine clearance <80 ml/min were not eligible for enrollment. Seventy-three percent of ineligible volunteers were excluded for lab abnormalities, with creatinine clearance <80 ml/min accounting for 43% of these exclusions. The percentage of all screened volunteers who were excluded due to decreased creatinine clearance was 20%; this may limit the generalizability of the results in part to populations with high creatinine clearance. Study volunteers were blinded to their active drug or placebo assignment and at the time of conduct of the study, PrEP had not yet been shown to be effective. Adherence, acceptability, risk compensation behavior and other outcomes may differ when PrEP which is known to be effective is available in the community. Volunteers were recruited from a pre-existing HIV serodiscordant couple cohort in which participants received regular couple and individual HIV risk reduction counseling and other preventive measures. Consistent HIV testing and counseling offered during HIV biomedical intervention studies has been perceived by volunteers to affect their risk behaviour in a positive way, by giving them knowledge and awareness of their HIV-negative status, hence added motivation to negotiate and use condoms more regularly [39]. Therefore, their behaviour may not reflect that of the general population.

Supporting Information

Acknowledgments

We would like to acknowledge the contributions of the study participants to this research. We would like to thank the study staff. Lorna Clark conducted and Josephine Cox oversaw the IFN-γ ELISPOT analyses at the IAVI Human Immunology Laboratory.