Article Text

Download PDFPDF

A randomised placebo-controlled trial to explore the effect of suppressive therapy with acyclovir on genital shedding of HIV-1 and herpes simplex virus type 2 among Zimbabwean sex workers
  1. F M Cowan1,
  2. S J Pascoe2,
  3. K L Barlow3,
  4. L F Langhaug2,
  5. S Jaffar2,
  6. J W Hargrove2,
  7. N J Robinson5,
  8. M T Bassett4,
  9. D Wilson4,
  10. D W G Brown3,
  11. R J Hayes2
  1. 1
    Royal Free and University College Medical School, London, UK
  2. 2
    London School of Hygiene and Tropical Medicine, London, UK
  3. 3
    Health Protection Agency (HPA), London, UK
  4. 4
    University of Zimbabwe, Harare, Zimbabwe
  5. 5
    GlaxoSmithKline Research and Development, Brentford, UK
  1. Dr F M Cowan, Centre for Sexual Health and HIV Research, Royal Free and University College Medical School, University College London, Mortimer Market Centre, off Capper Street, London WC1E 6AU, UK; frances{at}


Objectives: To determine the effect of daily acyclovir on genital shedding of HIV-1 and herpes simplex virus type 2 (HSV-2) in a randomised placebo-controlled trial among rural Zimbabwean sex workers.

Methods: 214 women were recruited and tested for HIV-1 and HSV-2 antibodies, HIV plasma viral load, CD4 lymphocyte count and genital swabs for qualitative detection of HIV-1 and HSV-2 genital shedding. Women were randomly assigned to acyclovir 400 mg twice a day for 12 weeks or matching placebo and were followed weekly to detect HIV-1 or HSV-2 genital shedding. Shedding analyses were only undertaken on 125 women co-infected with HSV-2 and HIV-1. Data were analysed using logistic regression, with random effects modelling used to account for repeated measurements on the same women.

Results: All women were randomly assigned to acyclovir or placebo; 125 of whom were co-infected with HIV-1 and HSV-2. 69 women were randomly assigned to acyclovir and 56 to placebo. Although twice daily acyclovir reduced rates of HSV-2 genital shedding, (adjusted odds ratio (AOR) 0.24; 95% CI 0.12 to 0.48; less than p<0.001), it had no effect on the proportion of visits at which HIV-1 shedding was detected (AOR 1.08; 95% CI 0.48 to 2.42; p = 0.9). Adherence varied between participants but even when adherence was high (as determined by pill count and extent of HSV-2 suppression) HIV-1 shedding was not reduced.

Conclusion: Among these HIV-1 and HSV-2-seropositive women, suppressive acyclovir therapy had no effect on the rate of HIV genital shedding despite a reduction in genital HSV-2. Treatment adherence and its measurement clearly affect the interpretation of these results.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

  • Among HIV-positive individuals, co-infection with HSV-2 has been shown, in some but not all studies, to increase the frequency and quantity of HIV-1 genital shedding.

  • In this placebo controlled randomised controlled trial, daily acyclovir failed to suppress HIV-1 genital shedding despite effecting a substantial and significant reduction in HSV-2 genital shedding.

Herpes simplex virus type 2 (HSV-2) facilitates the sexual acquisition of HIV-1.13 A recent meta-analysis that examined the effect of HSV-2 on HIV acquisition found an adjusted risk ratio of 2.7 (95% CI 1.9 to 3.9) in men and 3.1 (95% CI 1.7 to 5.6) in women.2 The role of HSV-2 infection in facilitating the transmission of HIV-1 is, however, less clear. Only one study in HIV-1-serodiscordant couples in Uganda has been able to examine this directly.4 5 It found that the per-contact risk of HIV-1 transmission was increased fivefold (0.0062 versus 0.0012) in those with recent genital ulceration compared with those without (p = 0.002). HSV-2 serostatus per se was not associated with HIV-1 transmission. Studies that have examined whether HSV-2 infection increases the risk of HIV genital shedding have had mixed results, with some studies reporting increased HIV-1 shedding and others more equivocal findings.6 7 The reason for these differing results is complex and probably includes issues of measurement, assay sensitivity and sample collection. Disentangling the relative influence of these different factors and then adequately controlling for them is problematical. An alternative approach to exploring this relationship is to examine whether suppressing HSV-2 genital shedding using antiviral therapy results in a concomitant reduction in HIV shedding.

In this paper we report the results of a 12-week randomised double-blind placebo-controlled trial of suppressive acyclovir among sex workers in rural Zimbabwe. The aims were to determine whether the suppression of genital HSV-2 infection had any effect on the prevalence of genital shedding of HIV-1 and to explore adherence to daily suppressive acyclovir among participants.


Study setting and participants

Participants were female sex workers who had previously enrolled in an HIV prevention study in rural Zimbabwe.8 All women had recently received presumptive treatment for bacterial sexually transmitted infections, including Neiserria gonorrhoea, Chlamydia trachomatis, Trichomonas vaginalis, bacterial vaginosis, chancroid and syphilis. At baseline, participants were tested for antibodies to HIV-1 and HSV-2, HIV-1 plasma viral load and CD4 lymphocyte count. In addition, cervicovaginal, vulval and perianal swabs were taken to determine the rate of HSV-2 and HIV-1 genital shedding. All three swabs were collected into a single vial of viral transport medium and it was therefore not possible to differentiate the site of genital viral shedding. Cervicovaginal lavage (CVL) was undertaken using 3 ml normal saline to quantify genital HIV-1 viral shedding. All specimens were stored at between 2 and 8°C and were transported to the laboratory within 8 h. Results of the baseline characteristics of participants have already been published.8 9


All participants were randomly assigned to receive acyclovir 400 mg twice a day or matching placebo for 12 weeks, regardless of their HIV and HSV-2 antibody status. Randomisation was undertaken by GlaxoSmithKline Research and Development (GSK). Study drug was delivered to Zimbabwe in sequentially numbered courses of treatment; each course comprised three identical bottles of 72 tablets. Courses were allocated to study sites in permuted blocks of four. Participants were consecutively allocated to the next numbered course of treatment. At the baseline visit they were given the first bottle of study drug from their allocated course, which contained a month’s supply of either acyclovir or matching placebo. Women were asked to attend 12-weekly follow-up visits and received a new bottle of study drug at week 5 and week 9. The drug allocation code was forwarded to the study statistician by GSK after follow-up of all women and laboratory testing were completed.

Follow-up visits

At each weekly follow-up visit three swabs were taken from the vulval, perianal and cervicovaginal areas and put into a single vial of viral transport medium. CVL was not done at follow-up.

Laboratory methods

Blood was tested for HIV-1 antibodies using GacPAT. Positive samples were confirmed using GACELISA. HSV-2 antibody was detected using HerpeSelect enzyme immunoassay (Focus Technologies, Cypress, California, USA). Plasma HIV-1-RNA levels were tested using the Standard Cobas Amplicor HIV-1 Monitor test, version 1.5 (Roche Molecular Systems, Branchburg, New Jersey, USA). CD4 lymphocyte counts were analysed using a FACS counter. CVL specimens were spun. The supernatant and pellet were stored separately at −80°C.

Only women who were co-infected with HIV-1 and HSV-2 were assessed for evidence of genital viral shedding. Quantitative HIV-1-RNA viral loads were obtained from the baseline CVL supernatant using the UltraSensitive Cobas Amplicor HIV-1 monitor protocol as described in the baseline paper.9 HSV-2 PCR was performed using the LightCycler and B-globin PCR (PC04/GH20) to exclude amplification inhibitors. Multistix 10 SG reagent strips (Bayer Diagnostics, Tarrytown, New York, USA) were used to assess haem contamination of CVL supernatant and genital swab specimens. Y chromosome PCR was done on CVL supernatant as a marker of semen contamination but not on follow-up specimens.

Genital swabs were extracted using the MagNAPure total nucleic acid isolation kit (Roche), or manually using QIAamp DNA blood mini kit (Qiagen, Hilden, Germany) for HSV-2 and a variation of the Boom method for HIV-1.10 HIV-1 gag and/or pol was amplified as previously described10 to provide qualitative evidence of HIV shedding. HSV-2 was detected using a LightCycler assay, a modified version of the method of Espy et al.11 Equivocal results were confirmed using a slightly modified version of a nested, block-based type-specific PCR targeting the gD gene.12

Statistical considerations

The proposed sample size of 220 sex workers assumed that 80% (n  =  176) of sex workers would be HIV positive, that HSV-2 infection among HIV-positive sex workers would be universal and that 80% of genital swabs taken from HIV-positive sex workers in the placebo arm would have HIV detected. With this number, the study would have 80% power to detect a 25% reduction in HIV shedding as measured over all follow-up visits in the acyclovir arm of the study.

All data were double entered in Microsoft Access. Range and consistency checks were performed. Plasma viral load was dichotomised at 50 000 copies per millilitre and CD4 cell count at 200 per microlitre. The frequency of shedding was compared between groups using χ2 tests and logistic regression, with correlations due to repeated measurements of the same women taken into account using random effects models. All odds ratios (OR) were calculated using random effects models. Data were analysed using Stata 9.0.

Ethical considerations

The study was approved by the Medical Research Council of Zimbabwe and the Medicines Control Authority of Zimbabwe. In addition the study had approval from the ethics committees of University College London and London School of Hygiene and Tropical Medicine. All women provided written consent for their participation. Women took part in the study anonymously and so could not be provided with their test results. All women received presumptive treatment for bacterial sexually transmitted infections and women with positive rapid plasma reagin assays at the time of their baseline visit received treatment for syphilis. All women received study drug (acyclovir or matching placebo) regardless of HSV-2 antibody status. HSV-2 suppressive therapy had been in use in developed countries for many years and was known to be safe and well tolerated. Antiretroviral therapy was rarely available within Zimbabwe in 2001.


A total of 214 women were enrolled (see fig 1). Overall, 182 (85.4%) women were HSV-2 infected (95% CI 80.7% to 90.2%) and 131 (61.2%) were HIV-1 infected (95% CI 54.6% to 67.8%); 125 women (57.9%; 95% CI 51.3% to 64.6%) were co-infected with both viruses, which was fewer than originally anticipated. Of note, 95.4% of HIV-1-infected women were co-infected with HSV-2 (age-adjusted OR 10.7; 95% CI 3.9 to 29.2). The baseline characteristics of all women have been published previously.9

Figure 1 The consort flowchart. LFU, lost to follow-up.

Table 1 shows the baseline characteristics of the 125 co-infected women by treatment allocation. Sixty-nine women were allocated to receive acyclovir and 56 placebo. Women in the acyclovir arm appeared to have spent longer in sex work, to be less well educated, have a slightly higher plasma and cervical HIV viral load and to have lower CD4 cell counts. Approximately 80% of women were shedding HIV-1 at baseline and approximately 20% were shedding HSV-2. No association was found between HSV-2 genital shedding and HIV genital shedding at baseline (data not shown).9

Table 1 Baseline characteristics

In table 2, the number and proportion of participants with genital shedding of HSV-2 and HSV-1 detected at each follow-up visit are presented. There was no significant difference in the rate of HIV-1 shedding between treatment arms (univariate OR 1.18; 95% CI 0.48 to 2.90; p = 0.7); adjusted OR (adjusting for CD4 lymphocyte count and plasma viral load at baseline) 1.08 (95% CI 0.48 to 2.42; p = 0.9). The proportion of women who ever had HIV shedding detected was also not different between the arms of the study (47/55 (85%) placebo and 61/69 (88%) acyclovir (p = 0.6)).

Table 2 Number genital shedding HSV-2 and HIV-1 at each follow-up visit

At each follow-up visit, the number of women shedding HSV-2 was less among those taking acyclovir than those taking placebo and this was statistically significant on visits 1, 2, 3, 9 and 10 (table 2). Overall, women in the acyclovir arm shed HSV-2 less frequently than women on placebo; HSV-2 shedding was detected at 64 of 646 (10%) visits in the acyclovir arm compared with 123 of 531 visits in the placebo arm (23%) (univariate OR 0.24; 95% CI 0.12 to 0.50; less than p<0.001); adjusted OR (adjusting for CD4 lymphocyte count and plasma viral load at baseline) 0.24 (95% CI 0.12 to 0.48; less than p<0.001). The proportion of women ever shedding HSV-2 was also significantly reduced in the acyclovir arm (27/69 (39%) acyclovir versus 36/55 (65%) placebo (p = 0.004)).

Table 3 shows comparison of HIV-1 shedding by treatment arm at each visit by whether or not women were shedding HSV-2. Although there was a trend towards increased HIV shedding in women who were shedding HSV-2 during follow-up, there was no evidence that acyclovir treatment reduced HIV shedding overall or when stratified by HSV-2 shedding.

Table 3 Number (%) HIV-1 genital shedding, stratified by HSV-2 shedding by visit

Table 4 shows the proportion of women who were adherent to treatment based on self-report and pill counting at each visit. First, the number of women who did not report missing at least one dose by visit is shown. There was some variability by visit, but no clear difference between treatment arm nor any evidence that self-reported adherence decreased over time. On average approximately 10% of women reported missing a dose at any visit. The second measure of adherence was pill counting. Women were issued with 72 tablets at baseline. They were then given a further 72 tablets at either visit 4 or visit 5 (depending on the number of tablets left at visit 4) and visit 9. Women were asked to finish the tablets in one bottle before starting the next. Women were defined as compliant if the number of tablets observed at that visit was within five below or above the expected number. If women missed a visit the number of observed and expected tablets observed at the subsequent visit was calculated. Using this definition, the proportion of women who were adherent reduced over the course of the study. Of note, among the non-adherent women (26% overall), pill counting showed 15% of women to have returned more tablets than expected and that 11% had returned fewer than expected (indicating that they took more tablets than prescribed or gave them away).

Table 4 Proportion adherent at each visit based on self-report and pill count by treatment arm

We examined the association between the level of adherence defined by pill counting and HIV shedding by study arm, but found no evidence to suggest that more adherent women were less likely to shed HIV (data not shown).


This small trial was designed to determine the effect of acyclovir suppressive therapy on HIV shedding. Women were randomly assigned to receive either acyclovir 400 mg twice a day or matching placebo for 12 weeks and had weekly genital swabs taken over that time. If HIV-1 shedding is increased by the presence of clinical or subclinical HSV-2 shedding then suppressive antiviral therapy might be expected to reduce the rate of HIV-1 shedding. However, in this small study we found no evidence that suppressive acyclovir reduced HIV shedding, although it clearly did reduce the rate of HSV-2 shedding. Of note, no relationship was found between HIV-1 shedding and HSV-2 before starting acyclovir either qualitatively or quantitatively.9

One explanation for the lack of effect is that women did not take their study drug as prescribed. Measuring adherence is difficult, and we therefore used two measures: self-report and pill counting. As others have shown, women in this study appeared to overestimate their adherence using self-reporting when compared with pill counting (table 4). The pill count data showed a clear drop in adherence over the course of the study. Of note, pill counting suggested that among non-adherent women almost as many women returned fewer tablets than expected as returned more, suggesting that these women were either taking extra doses or were giving tablets away. Whereas the pill count and self-report data indicate suboptimal adherence, the demonstrated reduction in HSV-2 shedding in the acyclovir arm suggests that some women were taking enough acyclovir to suppress HSV-2, and it is thus disappointing not to have demonstrated a trend towards a reduction in HIV shedding. It may be that much more consistent HSV-2 suppression adherence is required to have an effect on HIV shedding.

Other studies looking at the association between HSV-2 and HIV genital shedding have had differing results, with some able to demonstrate an association and others not.7 13 Studies that have quantified HSV-2 shedding do seem to have shown that women shedding greater quantities of HSV-2 are more likely to shed HIV-1. This study only examined the effect of suppressive acyclovir on qualitative and not quantitative HIV-1 and HSV-2 shedding, and therefore missed the opportunity to demonstrate a reduction in the quantity of virus shed, which given the negative findings overall would obviously be important to do. Of note, none of the women in this study were seen with symptomatic genital ulceration, which is associated with much higher levels of HSV-2 shedding than asymptomatic infection.

This was a small study and had less power than originally estimated because of the lower than anticipated HIV-1 prevalence among participants (125 co-infected women instead of 176). This reduced both the precision of effect estimates and probably contributed to the imbalance in the number of women between arms and in baseline characteristics. A greater proportion of women allocated to the acyclovir arm were shedding HIV and HSV-2 at baseline although these differences were not statistically significant.

Nagot and colleagues14 also examined the effect of suppressive antiviral therapy on HIV-1 shedding. They found that among high-risk women from Burkina Faso, twice daily valacyclovir reduced both the frequency and quantity of HIV genital shedding and in addition reduced HIV plasma viral load. In that study, a greater proportion of women were shedding HSV-2 at baseline than in our study (44% versus 20%), and one third of women had a symptomatic episode during follow-up, which may possibly have accounted for the difference between our findings. In addition, that study was conducted using twice daily valacyclovir, which has a longer half-life than acyclovir and therefore the dosing interval is less critical. Compliance as measured by pill count in that study was very high (>95%). Zuckerman and colleagues15 examined the effect of twice daily valacyclovir on rectal shedding of HIV in men who have sex with men in Peru. They also found significantly reduced levels of rectal HIV shedding and a reduction in HIV plasma viral load in the valacyclovir arm.15 In another trial of suppressive acyclovir from South Africa,16 investigators were unable to demonstrate a reduction in the rate or quantity of HIV shedding detected at 3 months. However, unlike this study, they did demonstrate a reduction in the frequency of HIV shedding as well as a reduction in plasma viral load.

In summary, among these HIV-1 and HSV-2-seropositive women, we found no evidence that twice daily acyclovir reduced the frequency of genital shedding of HIV-1. It is possible that we missed an association either because of the small numbers in the trial, low levels of tablet adherence and/or because we were unable to quantify the amount of virus shed. A large clinical trial is underway to see if suppressive antiviral therapy can prevent the transmission of HIV-1 between HIV-1-discordant couples. The results of that trial will overcome the difficulties in interpreting genital shedding studies and in deciding how the results should be applied more broadly to HIV prevention initiatives.


The authors would like to thank Dr Kirstin Edwards of the Genomics, Proteomics and Bioinformatics Unit of the Health Protection Agency, London, for the implementation or development of the HSV assays used in this study and invaluable advice.


View Abstract


  • Funding: This study was part of a larger study funded by a grant from the Wellcome Trust. The research for this substudy was funded by GlaxoSmithKline Research and Development.

  • Competing interests: None.

  • Ethics approval: The study was approved by the Medical Research Council of Zimbabwe and the Medicines Control Authority of Zimbabwe. In addition the study had approval from the ethics committees of University College London and London School of Hygiene and Tropical Medicine.

  • Patient consent: Obtained.

  • Contributors: FMC drafted the paper and all authors commented on the draft. FMC designed the study and oversaw all aspects of the project implementation and evaluation. SJP was responsible for the preliminary statistical analysis for the paper. KLB validated and then undertook all the HIV and HSV-2 PCR analyses on genital and plasma samples. LFL supervised the day to day running of the project. SJ assisted with the study design and undertook statistical analysis. JWH was responsible for the initial statistical analysis and data management. NJR assisted with study design and analysis. MTB was involved in study design and facilitated implementation. DW was involved in study design and facilitated implementation. DWGB was closely involved in study design and the design of laboratory analyses. RJH was involved in study design and oversaw the statistical analysis.