The double-blind, placebo-controlled randomized clinical trial (RCT) is the gold standard for establishing the efficacy of investigational products. This study design, through blinding study investigators and participants to product assignments, often maximizes the likelihood of obtaining an unbiased estimate of efficacy. However, to obtain an unbiased estimate of efficacy, the product used as a placebo control must have no effect on the outcomes of interest.
In vaginal microbicide research for the prevention of HIV-1 and other sexually transmitted infections (STIs), there are several mechanisms through which a placebo gel could have a protective effect on the outcomes of interest, thereby creating efficacy dilution.1,2 First, a placebo gel could have protective effects by virtue of lubricating properties or by serving as a physical barrier to infection. Second, antimicrobial effects from preservatives contained in the placebo gel could have effects on vaginal flora. Finally, the presence of a placebo gel in the vagina after ejaculation lowers the concentration of the infecting agent.
To date, vaginal microbicide research for the prevention of HIV-1 and other STIs has been limited by a lack of information on the possible protective effects of placebo gels used in clinical trials. The adverse impact of this lack of information on the potential effects of placebo control gels was clearly demonstrated in the UNAIDS sponsored study of the nonoxynol-9–containing gel COL-1492 assessing risk of acquisition of HIV-1.3 The placebo used in that study was a bioadhesive gel with all 3 of the potential mechanisms of actions that could protect against HIV-1 infection.4,5 A higher rate of HIV-1 infection was observed in the COL-1492 group of that study compared with that in the placebo group, a difference that is likely due to a deleterious effect of COL-1492 through epithelial damage with frequent use of N-9.6 However, part of this difference between the randomized arms could also be due to a possible protective effect of the placebo gel. In the absence of information that inclusion of a no-gel arm in that study could have provided, either interpretation is valid.
Although there are several reasons that the use of a placebo gel as a control in vaginal microbicides could result in a biased estimate of efficacy, the use of a no-gel control arm (“condom only”) also presents difficulties when assessing the effectiveness of a vaginal microbicide product. Women assigned to a gel arm may exhibit different sexual behavior than if they were assigned to a “condom-only” arm, perhaps substituting product use for condom use, and this complicates the interpretation of comparisons between the no-gel arm and the active product arm.7–9
A hydroxyethylcellulose (HEC) placebo gel has been formulated to minimize any impact—negative or positive—on vaginal mucosa, acquisition of HIV-1 and several other STIs, and pregnancy, and has been proposed as a “universal placebo” for vaginal microbicide trials.10–14 However, until the HEC placebo gel was tested in vivo in a relevant population against a no-gel arm, it was not possible to assess potential behavioral effects associated with the use of the HEC gel, and assess the potential direct causal effects of the HEC gel with respect to multiple outcome measures including acquisition of HIV-1 and other STIs, pregnancy, and genital safety. HPTN 035 was designed with both a no-gel control arm and an HEC placebo gel control arm.1 Here we provide a detailed comparison of differences in reported behavior, safety, and effectiveness of these 2 control arms. This comparison provides information to help interpret the results of previous vaginal microbicide trials using the HEC placebo gel, and to help determine if the HEC placebo gel is a suitable control for ongoing and future vaginal microbicide studies of newly developed products such as the maraviroc and tenofovir combined gel.
Study Design and Procedures
The study was conducted with the understanding and written informed consent of all participants. The trial (NCT00074425) was approved by 11 institutional review boards that oversee research conducted at the 8 study sites, and regulatory authorities in the United States, South Africa, and Zimbabwe. Details of the HPTN 035 study design and procedures are published elsewhere.15 Briefly, HPTN 035 was a 4-arm randomized placebo-controlled trial including 2 active microbicide products (BufferGel and 0.5% PRO2000 Gel), and 2 control arms (the HEC placebo gel arm and the no-gel arm). HIV-negative nonpregnant women at 7 sites in Blantyre and Lilongwe, Malawi; Durban and Hlabisa, South Africa; Harare and Chitungwiza, Zimbabwe; Lusaka, Zambia; and Philadelphia, United States, were randomized (stratified by site) in a 1:1:1:1 ratio to the 4 arms. Women randomized to a gel arm were instructed to insert 1 applicator of gel in their vagina up to 1 hour before each episode of vaginal intercourse. Women who became pregnant during follow-up temporarily discontinued gel use while pregnant.
The participants were followed monthly for 12–30 months. Pregnancy testing and gel provision were done at monthly study visits, whereas pelvic examinations, HIV-1 testing, and the collection of data on sexual behavior and adherence were all done quarterly. Testing for Chlamydia trachomatis, Neisseria gonorrhoeae, and syphilis was done annually and at study exit; herpes simplex virus 2 (HSV-2) testing was done at enrollment, and at study exit for women testing HSV-2 negative at enrollment.
HIV-1 infection was determined using the standardized algorithm explained in detail elsewhere.15 C. trachomatis and N. gonorrhoeae testing was done using urine strand displacement assay (Becton Dickinson ProbeTec, Becton Dickinson, Franklin Lakes, NJ). Vaginal fluid wet mount was used to test for trichomoniasis (saline prep), candidiasis (saline and KOH prep) and bacterial vaginosis (BV; saline and KOH prep), and BV was defined using Nugent criteria. Syphilis serologic testing was done using a rapid plasma reagin screening test followed by confirmatory Treponema pallidum particle agglutination testing. HSV-2 testing was done using an enzyme-linked immunosorbent assay (Focus Technologies HerpeSelect-2, Focus Technologies, Cypress, CA). Finally, pregnancy testing was done using a urine human chorionic gonadotropin (hCG) test (Quidel QuickView One Step hCG, Quidel Corporation, San Diego, CA).
All analyses were intent to treat and stratified by site. Sexual behavior during follow-up was compared using the t test for continuous variables and χ2-tests for binary variables. Changes in sexual behavior from baseline were tested using the paired t-test. Cox proportional hazard models were used to compare safety outcomes and pregnancy outcomes. Discrete time Cox proportional hazards models were used to compare time to detection of HIV. For outcomes with recurring events [BV, chlamydia, gonorrhea, trichomoniasis, genital ulcer disease (GUD), and pregnancy], the Andersen Gill Proportional Hazards model with robust variance estimates was used. The mean rates of syphilis were compared using Poisson regression with person years of follow-up as an offset. HSV-2, which was only assessed at study entry and exit, was compared using logistic regression.
To determine the magnitude of differences between the 2 randomization arms that was independent of condom use, all models of safety, pregnancy, and STI outcomes were adjusted for the level of reported condom use (condom use during >85% of last vaginal sex acts during follow-up, which was the median reported value).
Study Population Characteristics and Retention
Between February 2005 and September 2008, 771 women were enrolled into the HEC placebo gel arm and 772 into the no-gel arm of the trial. Of these, 11 women in the placebo arm and 10 in the no-gel arm had no follow-up data. Detailed information on baseline characteristics of these 2 groups has been presented previously.15 Briefly, the mean age at enrollment was 26 years, mean number of vaginal sex acts in the past 7 days was 2.9, and 68% of women reported using a condom during their last sex act. The mean follow-up time was 20.5 months in the no-gel arm, and 20.3 months in the placebo arm (P = 0.7). The no-gel arm had an average of 19.2 visits per woman and the HEC placebo gel arm 19.1 visits per woman (P = 0.8). Retention in HPTN 035 overall was very high (93.6%) and did not differ between the no-gel and HEC placebo gel arms (94.0% versus 93.1%, P = 0.3). Baseline characteristics were similar between the 2 arms (Table 1).
Gel Use, Sexual Behavior, and Condom Use
Adherence to gel use in the HEC placebo gel arm was high with gel use reported 81.4% of last vaginal sex acts. Similar levels of gel use were also reported during vaginal sex acts in the past week. The mean reported number of sex partners, mean reported number of vaginal sex acts, and mean reported number of anal sex acts were all similar between the 2 arms during follow-up (Table 2).
The no-gel arm reported significantly higher condom use during follow-up compared with the HEC placebo gel arm. Specifically, 47% of women in the HEC placebo gel arm reported high condom use (defined as using a condom during ≥85% of last vaginal sex acts), compared with 59% of women in the no-gel arm (P < 0.001). In addition, this differential reported condom use was fairly consistent throughout the 30 months of the study (Fig. 1).
To further assess changes in behavior during follow-up in the HEC gel arm and the no-gel arm, baseline reported sexual behavior was compared with reported sexual behavior during follow-up. There was no significant change from baseline in the reported number of sex partners in either arm (P = 0.1 for the HEC placebo arm, and P = 0.9 for the no-gel arm), and women in both arms reported fewer vaginal sex acts during follow-up compared with baseline (P = 0.007 for the HEC placebo arm, and P < 0.001 for the no-gel arm). However, there were differences between the 2 arms in the change from baseline in reported condom use. Although the HEC placebo arm did not have a significant change from baseline in reported condom use [mean increase from baseline in the percentage of vaginal sex acts reported to be protected by a condom of 1.58% (P = 0.4)], the no-gel arm had a significant increase from baseline in reported condom use [mean increase from baseline in the percentage of vaginal sex acts reported to be protected by a condom of 15.1% (P < 0.001)].
The incidence rates of safety outcomes during follow-up in the 2 arms are presented in Table 3. The overall incidence of deep epithelial disruption, defined as lesions penetrating into and exposing the subepithelial tissue and possibly blood vessels, was similar in the HEC placebo gel arm and the no-gel arm (1.93 versus 1.48 per 100 person years), although there was limited statistical power to assess this endpoint. There was no significant difference between the 2 arms in the percentage of women with at least one pelvic examination finding, the percentage of women experiencing at least one grade 3 or higher adverse event, and the rate of adverse pregnancy outcomes between the 2 arms, although statistical power for the pregnancy outcomes was limited (Table 3). Results were similar after adjustment for differential condom use during follow-up (data not shown).
Sexually Transmitted Diseases, Genital Tract Conditions, and Pregnancy
There was no difference in the rate of acquisition of HIV-1 in the 2 arms [hazard ratio = 0.97, 95% confidence interval (CI): (0.66, 1.44); Table 4]. There were also no differences in the rates of C. trachomatis, N. gonorrhoeae, bacterial vaginosis, trichomoniasis, GUD, syphilis, HSV-2, and pregnancy (Table 4). Results were similar after adjusting for differential condom use (data not shown).
HPTN 035 is the only large-scale vaginal microbicide effectiveness study designed with women randomized to either the HEC placebo gel or no gel. This design provides a unique opportunity to compare sexual behavior and safety measurements during follow-up in these 2 control arms, and to assess the activity of the HEC placebo gel against HIV-1 and other STIs. In this RCT, we found no significant differences between the HEC placebo gel and the no-gel arms in the rates of genital adverse events, pregnancy outcomes, or STIs, including HIV-1. However, there were significantly higher rates of condom use during follow-up in the no-gel arm. After controlling for this differential condom use, results were similar to the unadjusted results.
Our finding of a higher rate of reported condom use in the no-gel arm has been seen in other STI prevention studies using a “no product” or “condom-use alone” arm as a control.7–9 This phenomenon is thought to mainly occur through “product substitution” or “condom migration” in which women in an intervention arm substitute using the intervention product for condoms while, at the same time, women in the condom-use only arm maintain high rates of condom use.16 In HPTN 035, we did not see evidence of condom migration (ie, a decrease in reported condom use from baseline rates in the HEC placebo gel arm). Rather, we saw reported rates of condom use remain relatively stable from baseline through follow-up in the HEC placebo gel arm, and an increase from baseline in reported rates in the no-gel arm, plausibly because of intensive counseling regarding condom use during the trial and also plausibly due to reporting bias. Regardless of the cause, the phenomenon of differential condom use between arms has been postulated as 1 reason most HIV/STI prevention trials using an open-label “condom-use alone” arm as a control have not yielded a positive estimate of efficacy.17,18
The positive safety profile of the HEC placebo gel seen in HPTN 035, along with the lack of any large effects on HIV/STI incidence, help interpret the findings of previous vaginal microbicide trials using the HEC placebo as a control. Recently, 4 phase III randomized double-blind, placebo-controlled HIV prevention trials using the HEC placebo gel as a control have reported results.12–14,19 Two of these studies, one assessing 1.0% C31G (SAVVY) gel and one assessing Cellulose Sulfate gel, were halted early and reported a trend for a higher rate of HIV incidence in women randomized to the active gel arm.12,14 The results from HPTN 035 provide additional evidence to suggest that this apparent increased risk of HIV was likely not due to a highly protective effect of the HEC placebo gel. The findings also suggest that the efficacy observed in the CAPRISA 004 trial18 was due to the active tenofovir gel and not due to a harmful placebo.
It has been argued that the ideal design for a RCT of a vaginal microbicide for STI prevention would contain both a placebo control arm and a no-gel arm.1 In such a trial, the comparison of the microbicide with the placebo control arm provides estimates of the antimicrobial effects of the microbicide, whereas comparison of the microbicide with the no-gel arm allows the assessment of “real-world” effectiveness, which includes physical barrier effects, lubrication effects, effects on vaginal flora, and effects on risk behavior, including condom use. Although these considerations may be applicable, with limited resources for HIV/STI prevention research, designing every large RCT of a vaginal microbicide with 2 control arms, thereby substantially increasing the size and cost of the trial, is not feasible. The results of HPTN 035 indicate that in the presence of high adherence to the study product, future and ongoing studies do not require both control arms. Microbicide trials designed to use the HEC placebo gel as their lone control arm have not limited their ability to produce a valid result or their statistical power to detect an effective method of HIV/STI prevention.
The main strength of our study is the study design—an RCT with women randomized to 2 control arms. Other strengths include the high rates of retention in both control arms, high reported adherence to gel in the HEC placebo gel arm, and frequent collection of behavioral data to aid interpretation of the results. However, the main weakness of the study is also the design—HPTN 035 was not designed as an equivalence trial for the 2 control arms. Thus, although the HR comparing the rates of HIV-1 acquisition in the 2 arms was near unity, the 95% CI ranged from 0.66 to 1.44, highlighting the inability of this study to determine equivalence of these 2 control arms. However, for some of the safety endpoints (eg, pelvic examination findings), and some of the STI endpoints (eg, BV and Trichomoniasis), we had high power to see fairly small effects, and for most other STIs, including HIV, we had high power to rule out large differences between the 2 arms. Other weaknesses include the open-label nature of the no-gel arm, and the reliance on self-reported data on sexual risk behavior and adherence.
There are many challenges in the design of, conduct of, and interpretation of results from HIV prevention trials.20 Just one of these challenges is the choice of an appropriate control arm. In this large RCT, we found no significant differences between the no-gel and HEC placebo gel arms in the rates of genital safety events, pregnancy outcomes, or STIs. These results aid interpretation of the results of previous vaginal microbicide trials that used the HEC gel as a control, and suggest that the HEC gel is a suitable control for future and ongoing vaginal microbicide studies, including antiretroviral containing microbicides such as the newly developed maraviroc and tenofovir combined gel.
Sponsors: US National Institutes of Health (R. Black, L. Soto-Torres, S. Estep), Indevus Pharmaceuticals (A. Profy), ReProtect (T. Moench).
Protocol Chair: Salim S Abdool Karim.
Blantyre, Malawi: T. Taha, N. Kumwenda, B. Makanani, S. Hurst, C. Nkhoma, L. Seyama Durban and Hlabisa, South Africa: G. Ramjee, R. Govinden, N. Coumi, N. Dladla-Qwabe, S. Ganesh Harare-Chitungwiza, Zimbabwe: Z.M. Chirenje, N. Padian, A. van der Straten, T. Magure, M. Mlingo, N. Mgodi
Lilongwe, Malawi: I. Hoffman, F. Martinson, T. Tembo, L. Chinula, T. Mvalo
Lusaka, Zambia: G. Parham, M. Kapina, C. Reid, M. Kasaro Philadelphia, United States: L. Maslankowski, J. Prince, N. Tustin, S. Whittington, E. Yu.
Coordinating Center: A. Coletti, K. Gomez, R. White Statistical Center: M. Cianciola, C. Kelly, C. Leburg, B. Mâsse, B. Richardson, K. Román.
Network Laboratory: S. Hillier, E. Piwowar-Manning, L. Rabe.
1. Fleming TR, Richardson BA. Some design issues in trials of microbicides for the prevention of HIV infection. J Infect Dis. 2004;190:666–674.
2. Masse BR, Boily M-C, Dimitrov D, et al.. Efficacy dilution in randomized placebo-controlled vaginal microbicide trials. Emerg Themes Epidemiol. 2009;6:5.
3. Van Damme L, Ramjee G, Alary M, et al.. Effectiveness of COL-1492, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: a randomised controlled trial. Lancet. 2002;360:971–977.
4. Nguyen D, Lee H, Poast J, et al.. Preventing sexual transmission of HIV: anti-HIV bioregulatory and homeostatic components of commercial sexual lubricants. J Biol Regul Homeost Agents. 2004;18:268–274.
5. Wu JP, Fielding SL, Fiscella K. The effect of polycarbophil gel (Replens) on bacterial vaginosis: a pilot study. Eur J Obstet Gynecol Reprod Biol. 2007;130:132–136.
6. WHO. WHO/CONRAD Technical Consultation on Nonoxynol-9 Summary Report. Geneva, Switzerland: World Health Organization; 2003.
7. Ramjee G, van der Straten A, Chipato T, et al.. The diaphragm and lubricant gel for prevention of cervical sexually transmitted infections: results of a randomized controlled trial. PLoS One. 2008;3:e3488.
8. Padian NS, van der Straten A, Ramjee G, et al.. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: a randomized controlled trial. Lancet. 2007;370:251–261.
9. Roddy RE, Zekeng L, Ryan KA, et al.. The effect of nonoxynol-9 gel on urogenital gonorrhea and chlamydial infection: randomized controlled trial. JAMA. 2002;287:1117–1122.
10. Schwartz JL, Ballagh SA, Kwok C, et al.. Fourteen-day safety and acceptability study of the universal placebo gel. Contraception. 2007;75:136–141.
11. Tien D, Schnaare RL, Kang F, et al.. In vitro and in vivo characterization of a potential universal placebo designed for use in vaginal microbicide clinical trials. AIDS Res Hum Retroviruses. 2005;21:845–853.
12. Feldblum PJ, Adeiga A, Bakare R, et al.. SAVVY vaginal gel (C31G) for prevention of HIV infection: a randomized controlled trial in Nigeria. PLoS One. 2008;3:e1474.
13. Peterson L, Nanda K, Opoku BK, et al.. SAVVY (C31G) gel for prevention of HIV infection in women: a phase 3, double-blind, randomized, placebo-controlled trial in Ghana. PLoS One. 2007;2:e1312.
14. Van Damme L, Govinden R, Mirembe FM, et al.. Lack of effectiveness of cellulose sulfate gel for the prevention of vaginal HIV transmission. N Engl J Med. 2008;359:463–472.
15. Abdool Karim S, Richardson B, Ramjee G, et al.. Safety and effectiveness of Buffergel and 0.5% PRO2000 gel for the prevention of HIV infection in women: results of the HPTN 035 microbicide trial. AIDS. 2011;25:957–966.
16. van der Straten A, Cheng H, Moore J, et al.. The use of the diaphragm instead of condoms in a phase III diaphragm trial. AIDS Behav. 2009;13:564–572.
17. Gray RH, Wawer MJ. Randomised trials of HIV prevention. Lancet. 2007;370:200–201.
18. Trussell J, Dominik R. Will microbicide trials yield unbiased estimates of microbicide efficacy? Contraception. 2005;72:408–413.
19. Abdool Karim Q, Abdool Karim SS, Frohlich JA, et al.. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science. 2010;329:1168–1174.
20. Lagakos SW, Gable AR. Challenges to HIV prevention—seeking effective measures in the absence of a vaccine. N Engl J Med. 2008;358:1543–1545.