PRO 2000 Gel Inhibits HIV and Herpes Simplex Virus Infection Following Vaginal Application: A Double-Blind Placebo-Controlled Trial
The Journal of Infectious Diseases Jan 1, 2006;193:27-35
Marla J. Keller,1 Bouchra Zerhouni-Layachi,1 Natalia Cheshenko,2 Minnie John,2 Kathleen Hogarty,1 Andrea Kasowitz,1 Cindy L. Goldberg,2 Sylvan Wallenstein,3 Albert T. Profy,4 Mary E. Klotman,1 and Betsy C. Herold2
Departments of 1Medicine, 2Pediatrics, and 3Community Medicine, Mount Sinai School of Medicine, New York, New York; 4Indevus Pharmaceuticals, Lexington, Massachusetts
Background. Microbicides used to prevent the transmission of human immunodeficiency virus (HIV) are advancing to clinical trials on the basis of activity observed in vitro and in animal models. However, no data demonstrate activity of microbicides after application in humans. This study was designed to determine the antiviral activity in cervicovaginal lavage (CVL) samples collected after intravaginal application of 0.5% PRO 2000 gel (Indevus).
Methods. A randomized, double-blind study was conducted to assess the anti-HIV and antiherpes simplex virus (HSV) activity of PRO 2000 in CVL samples obtained at screening (48 hours before) and 1 hour after application of study or placebo gel. HeLa cells or human macrophages were inoculated with CVL samples spiked with replication-defective HIV containing a luciferase indicator gene and pseudotyped with an R5 envelope. Human cervical epithelial cells were inoculated with CVL samples and challenged with HSV-2(G), and the virus titer was then determined.
Results. CVL samples obtained after application of PRO 2000 gel significantly inhibited HIV and HSV infection by at least 1000-fold, compared with CVL samples obtained at screening (P < .001). There were no differences in cytokine levels between the drug and placebo groups.
Conclusions. PRO 2000 gel (0.5%) is sufficiently bioavailable and retains substantial antiviral activity after intravaginal application. This strategy provides a mechanism for testing the efficacy of a microbicide before embarking on large-scale clinical trials.
The prevention of HIV and other sexually transmitted infections is a global health priority. It is estimated that 40 million people worldwide are living with HIV, and women account for nearly half of those infected . Unprotected sex is the predominant mode of transmission, and genital herpes plays a major role in increasing the risk for acquisition and transmission of HIV . The seroprevalence of herpes simplex virus type 2 (HSV-2) in developing countries ranges from 60% to 80% . Although condoms offer protection against sexually transmitted infections, their effectiveness is limited because they require initiation and/or consent by the partners. Thus, there is an urgent need for novel preventative strategies, such as vaginal microbicides. Mathematical modeling predicts that, over the course of 3 years, 2.5 million HIV infections could be averted if a microbicide that is 60% effective against HIV were used by 20% of women in half of all sex acts that do not involve a condom .
Nonoxynol-9 was extensively evaluated as a vaginal microbicide. Clinical trials failed to show any protection against HIV, chlamydial infection, or gonorrhea, and a multicenter placebo-controlled phase 3 trial demonstrated a higher risk of acquisition of HIV after frequent use of nonoxynol-9, relative to placebo, presumably reflecting the inflammatory and cytotoxic impact of repeated applications [5, 6]. Five microbicides are currently in phase 3 trials. Three are polyanions (PRO 2000 [Indevus], cellulose sulfate, and Carraguard [Population Council]), one is a surfactant (C31G [Cellegy Pharmaceuticals]), and one is an acid-buffering agent (BufferGel [ReProtect]). These agents inhibit HIV and HSV in cell culture and/or in animal models. However, there are as yet no data demonstrating whether any microbicide exhibits antiviral activity after application in the human vagina. Therefore, this study was designed to evaluate the antiviral activity of PRO 2000 after intravaginal application, to determine whether results from a relatively inexpensive pilot clinical study could serve as a surrogate marker of bioavailability and potential efficacy.
Microbicides will be used not only by uninfected women to protect themselves from acquisition but also by HIV-infected women, who may be unaware of their HIV-seropositive status, who wish to reduce the risk of HIV transmission to their partner(s), and/or who wish to protect themselves from coinfection with other sexually transmitted pathogens. Therefore, this study was conducted to measure the antiviral activity after vaginal application of gel in HIV-infected women.
PRO 2000, a synthetic naphthalene sulfonic acid polymer that interacts with viral envelope glycoproteins to prevent virus entry, was selected because it is included in 2 of the multicenter studies . Phase 1 trials were conducted with 0.5%, 2%, or 4% gel in HIV-uninfected and HIV-infected volunteers [8, 9]. All concentrations were considered to be safe and well tolerated, although there was a trend toward increasing frequency of adverse events in the groups receiving 4% PRO 2000 gel. Because the 0.5% strength was judged to be safe in all clinical studies, it was selected for further investigation.
Subjects. Twenty-five women were assessed for eligibility. Four participants were excluded; 1 had genital herpes, and 3 had T. vaginalis infection diagnosed at screening. One subject did not return for enrollment. Ten women received 0.5% PRO 2000 gel, and 10 received placebo gel. There were no significant differences between drug and placebo groups with respect to age, baseline CD4 cell count, HIV-1 RNA level, or pH of CVL samples obtained after application of gel. Three PRO 2000 gel and 3 placebo gel recipients experienced adverse events. Five were considered to be possibly related to application of gel, were mild, and were similar in frequency and extent to those reported in a prior phase 1 trial of 0.5% PRO 2000 in HIV-uninfected women  and less frequent than those reported in a phase 1 study of 4% PRO 2000 in HIV-infected women . One subject experienced nausea and vomiting, which was judged to be unrelated to gel application but, instead, was associated with a new antiretroviral regimen, which was initiated 1 day after application of PRO 2000 gel.
Anti-HIV activity of PRO 2000 gel within CVL samples.
Recent studies demonstrate that infectious HIV can be isolated from genital secretions, using a modified Magi assay, when the virus load in the genital tract is >10,000 copies/mL . However, the level of viral RNA in the genital tracts of subjects at the screening visit ranged from <75 to 350 copies/mL (branched DNA assay; Bayer Diagnostics). Therefore, rather than culturing the subjects' own virus, we adopted a spiking strategy, using replication-defective pseudotyped luciferase-expressing HIV viruses. In pilot studies, we confirmed that neither PRO 2000 gel nor placebo gel interferes with the luciferase assay. Results demonstrate that a 1:2 dilution of CVL samples collected 1 h after application of PRO 2000 gel markedly inhibited HIV-1 infection, whereas CVL samples collected after application of placebo gel had little or no antiviral activity (figure 1, top and middle panels). The log fold reduction in HIV infection was 4.027 ± 1.33 for the active gel group, compared with 0.245 ± 0.45 for the placebo group (P < .001; Wilcoxon test). To correlate the antiviral activity observed following in vivo application with that predicted by in vitro assays, unformulated PRO 2000 (1-100 ug/mL) was diluted in pooled CVL samples (obtained from 5 subjects at screening) and then mixed with HIV-1 JRFL virus, and infectivity was monitored. Results suggest that the degree of anti-HIV activity observed in the CVL samples obtained after application of PRO 2000 gel was comparable to that observed with 100 ug/mL unformulated PRO 2000 diluted in CVL fluid.
The spiking strategy was expanded to evaluate the activity of CVL samples obtained after application of gel against a virus pseudotyped with an envelope cloned from a primary R5 isolate. R5 viruses predominate early after sexual transmission and may be more readily transmitted via mucosal routes [18, 19]. Because of limited sample availability, these studies were conducted after unblinding and included CVL samples from 5 randomly selected subjects in the active group and 2 subjects in the placebo group. CVL samples obtained after application of PRO 2000 gel inhibited HIV infection by >99% (figure 2, top panel). We also examined the anti-HIV activity of CVL samples on primary human macrophages. At a 1 : 8 dilution, the CVL samples obtained after application of PRO 2000 inhibited HIV infection of primary human macrophages (figure 2, bottom panel). The log fold reduction in HIV infection of macrophages was 2.84 ± 0.54 for the PRO 2000 gel group, compared with 0.245 ± 0.45 for the placebo group.
Anti-HSV activity of PRO 2000 gel after vaginal application.
Parallel studies were conducted with HSV-2. The log fold reduction in HSV yields in the presence of CVL samples (final dilution, 1 : 4) was 3.095 ± 0.178 for PRO 2000, compared with 0.636 ± 0.162 for placebo (P < .001; Wilcoxon test) (figure 3, top and middle panels, respectively). The antiviral activity after application of PRO 2000 gel was comparable to that observed with 100 ug/mL unformulated drug diluted in CVL fluid (figure 3, bottom panel).
Determination of concentration of PRO 2000 in CVL samples.
The concentration of PRO 2000 in CVL samples after application of 0.5% PRO 2000 gel ranged from 115 to 341 ug/mL (table 1), which correlates with that predicted by the observed antiviral activity.
Absence of acute inflammatory response to single application.
Inflammatory responses may increase susceptibility to HIV-1 and limit the beneficial effect of vaginal microbicides. To determine whether a single application in vivo induced a mucosal response, the concentrations of IL-1beta, IL-8, and SLPI in CVL samples at screening and after intravaginal application were measured. IL-1beta and IL-8 may be released from intracellular stores within 1 h after exposure to an inflammatory trigger , and we previously observed that IL-8 and SLPI were significantly reduced, relative to media, following exposure of immortalized human endocervical cells in vitro to unformulated PRO 2000 . Moreover, all 3 of these mediators may have effects on HIV infection. Drug, matched placebo, and CVL samples did not interfere with their detection by ELISA. Results indicate that there was no difference in the baseline concentrations of these mediators in CVL samples obtained at screening between subjects who received PRO 2000 gel or placebo gel (figure 4). However, there was a marked reduction in the levels of all of these mediators following application of either gel (not significant by analysis of covariance). These results suggest that there is no influx or activation of inflammatory cells, which may induce cytokine release. The screening CVL samples may have washed away the proteins, and 48 h may be insufficient time to replenish vaginal secretions in the absence of an inflammatory trigger. Consistent with this notion, we observed no increase in polymorphonuclear leukocytes in the cell pellets of CVL samples.
This study demonstrates for the first time that a candidate microbicide is sufficiently bioavailable and retains substantial anti-HIV and anti-HSV activity after intravaginal application. The activity correlates with the concentration of PRO 2000 in CVL samples, measured using a fluorescence assay, which may provide a useful tool for clinical trials to assess adherence to the protocol. The measured antiviral activity and concentrations of PRO 2000 in CVL samples likely underestimate the actual amount present in the genital tract. Alternative techniques for recovering genital tract secretions, such as wicking methods (Sno-Strip [Akorn] or TearFlo [Wilson Ophthalmic]), may have provided less dilute samples. However, wicking methods may induce an inflammatory response, and, in addition, a recent study suggested that lavage yielded more reproducible cytokine levels than did wicking methods . In the lavage procedure used, the applied 2-mL dose of gel and resident vaginal secretions are diluted with 10 mL of saline. This may be similar to what happens after sexual intercourse, when the vaginal contents are diluted with 2-6 mL of semen.
These studies provide critical data that may help interpret the results of ongoing phase 3 clinical trials. For example, our data suggest that, if PRO 2000 gel fails to reduce transmission of HIV or HSV, this failure is not due to reduced bioavailability or loss of biological activity in the genital tract environment. Alternative explanations include problems with consistent and proper use or, perhaps, loss of activity in the presence of semen. Notably, subjects enrolled in the present study were asked to abstain from sexual intercourse. Thus, a decrease in antiviral activity of formulated microbicide in the presence of seminal fluid could not be assessed. Studies to assess the impact of seminal fluid on antiviral activity are ongoing.
The extent of antiviral activity needed to reduce transmission is not known. Studies conducted among couples with discordant HIV infection status reproducibly demonstrate that the risk of HIV transmission correlates with virus loads , which may be influenced by the menstrual cycle, contraception, and vaginal flora and may be increased in the setting of acute HIV infection or in the presence of coinfection with other sexually transmitted pathogens . The findings that diluted CVL samples collected 1 h after application of PRO 2000 gel inhibit HIV-1 infection of HeLa cells by >4 logs and HIV infection of primary macrophages by about 3 logs suggests that the antiviral activity following vaginal application is sufficient to reduce HIV transmission. How these findings translate in vivo and whether this degree of in vitro anti-HIV activity will be effective even in the presence of acute HIV or other sexually transmitted infections is difficult to predict.
The finding that CVL samples obtained after application of PRO 2000 gel also significantly inhibited infection by a virus pseudotyped with an envelope cloned from a primary R5 isolate is also important, because R5 viruses predominate early during infection. Prior studies suggested that some polyanions, such as dextran sulfate, have greater activity against X4 isolates than against R5 viruses because of differences in the charge on the V3 loop of the gp120 proteins [19, 23]. However, our data to date suggest that PRO 2000 is highly active against R5 isolates. Given the diversity of HIV envelopes, this strategy should be expanded to include additional viruses pseudotyped with a panel of envelopes cloned from primary R5 and R5X4 strains isolated from different clades, because these drugs will be used worldwide.
No acute inflammatory response was elicited following a single application of drug or placebo. Rather, a decrease in polymorphonuclear leukocytes, IL-1beta, IL-8, and SLPI was detected following application of either drug or placebo gel, suggesting the absence of any inflammatory trigger. Using a similar protocol design, Ficharova et al.  observed a modest increase in IL-8 in CVL samples obtained 12 h after application of a single dose of nonoxynol-9 gel, compared with that in baseline CVL samples obtained 24 h earlier. These findings suggest that, if an inflammatory trigger is present, increases in proinflammatory cytokines will be observed in CVL samples, even when repeated samples are obtained within 24 h. Notably, the women enrolled in this study were in an older reproductive age group, and the innate cervicovaginal environment may differ from that of younger women. A 14-day double-blind placebo-controlled study currently in progress to assess the impact of repeated applications of 0.5% PRO 2000 gel, compared with placebo gel, on mucosal immune mediators is enrolling younger women.
In summary, this trial demonstrates for the first time that a candidate microbicide is sufficiently bioavailable and retains substantial anti-HIV and anti-HSV activity after intravaginal application. Testing additional compounds in the presently planned phase 3 trials would allow validation of this approach and could determine whether this safe and inexpensive strategy provides a surrogate marker to predict efficacy. If confirmed, this strategy could be used to determine which future-generation microbicides should go forward in development, possibly resulting in major cost savings.
MATERIALS AND METHODS
Participants. With informed consent, and pursuant to human experimentation guidelines of the Mount Sinai School of Medicine, HIV-infected women between the ages of 18 and 50 years were recruited from clinics in the New York metropolitan area between October 2003 and June 2004. Enrollment criteria included a level of HIV-1 RNA in plasma of about 4.7 log10 copies/mL (Amplicor HIV-1 Monitor, version 1.5; Roche) within 2 months of screening, no change in antiretroviral therapy within the previous 2 months, and a negative result or atypical squamous cells of undetermined significance on Pap smear documented within 6 months of participation. Exclusion criteria included pregnancy, breast-feeding, menopause, a recent history of intermenstrual bleeding, gynecological surgery, vaginitis, use of hormonal contraception within the previous 2 months, administration of any vaginal products, urinary tract infection, or sexually transmitted infection. The screening visit was scheduled at least 2 days after the cessation of menses.
At the screening visit, CD4 cell count, plasma HIV-1 RNA level, and rapid plasma reagin measurements were obtained. All participants had a urinalysis and culture, pregnancy test, and gynecological examination for detection of bacterial vaginosis, Trichomonas vaginalis, Candida species, and semen. The presence of Neisseria gonorrhoeae and Chlamydia trachomatis infection was determined by use of a DNA probe. Vaginal pH was measured using short-range (3.0-5.5) pHydrion paper (Micro Essential Laboratory). Cervicovaginal lavage (CVL) was done with about 10 mL of sterile normal saline (pH, about 5.5). The study visit was scheduled 48 h later. Subjects were asked to abstain from sexual intercourse for the 48 h before each visit. Participants were randomized in a double-blind fashion to receive 0.5% PRO 2000 gel or placebo gel, administered intravaginally by the study clinicians. Subjects were permitted to ambulate as desired, and 1 h later, CVL was performed. Each enrolled participant was contacted by telephone 4-7 days after administration of gel with regard to any adverse events.
Study drugs. PRO 2000 gel is an aqueous vaginal gel formulation containing 0.5% (wt/wt) PRO 2000. Placebo gel contains all of the same components except PRO 2000; caramel is added to placebo gel for color-matching purposes. PRO 2000 gel 0.5% and placebo gel were packaged in single-dose (2-mL) lacquer-lined aluminum tubes.
CVL samples. CVL samples were transported to the laboratory on ice and centrifuged at about 1000 g for 10-15 min at 4 C. A cocktail of antibiotics (final concentrations: penicillin, 500 U/mL; streptomycin, 50 ug/mL; and amphotericin, 0.5 ug/mL) was added to the supernatants, which were subdivided and stored at -80 C. The protein concentration (BCA Protein Kit) and pH of each sample were determined.
Cells and viruses. HeLa-CD4-CCR5 cells, which stably express CD4 and CCR5, were a gift from Shawn E. Kuhmann . Monocyte-derived macrophages were prepared from Ficoll-Hypaque-purified peripheral blood mononuclear cells and purified by adherence after 7-14 days in culture in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum. Replication-defective pseudotyped HIV was prepared in 293 T cells, using a previously described 3-plasmid cotransfection system: the packaging construct, pCMV delta R8.2 ; the HIV reporter construct, pNL4-3.Luc.R-E (obtained from Dr. Nathaniel Landau through the AIDS Research Reference Reagent Program) [12, 13]; and either of the 2 envelope-expressing constructs, pJRFL expressing an R5 envelope (gift from D. Littman, Skirball Institute, New York University, New York) or pRB23-1 expressing a primary R5 envelope . Transfections were done using Lipofectamine 2000 (Invitrogen). Virus supernatants were harvested 48 h after transfection. The amount of p24 antigen in the virus preparations was determined by use of an ELISA obtained from the National Cancer Institute; virus titers were determined in the HeLa-CD4-CCR5 cell line and primary macrophages. HSV-2(G) was used for all HSV studies . CaSki (human cervical epithelial) cells were obtained from the American Type Culture Collection .
Infectivity assays. Supernatants of CVL samples obtained at screening and after application of study gels were diluted with serum-free DMEM to a final dilution of 1 : 2, 1 : 8, or 1 : 64 and then were mixed with HIV-1 JRFL at a concentration of 0.05 pg/cell (or about 10 infectious particles/cell) for 5 min at 37 C before inoculation of HeLa-CD4-CCR5 cells. After a 2-h incubation, the cells were washed extensively and cultured in DMEM with 10% fetal bovine serum for 72 h. The cells were lysed and tested with the luciferase assay system (Promega). Luciferase expression was quantified as relative light units after background readings obtained from uninfected cells were subtracted. Additional experiments were done using the RB23-1 pseudotyped virus (0.05 pg/cell) in HeLa-CD4-CCR5 infection and using primary macrophages as the target cells. For comparison, unformulated PRO 2000 was diluted directly in CVL samples (pooled from 5 subjects at screening) and the antiviral activity was determined. To determine whether either of the gels interfered with the luciferase assay, the gels were added directly to the lysis buffer, and luciferase activity was measured.
For HSV-2 studies, CaSki cells in 96-well plates were exposed for 5 min to CVL samples diluted 1 : 2 and 1 : 8 in DMEM (50 uL/well) and then challenged with serial 10-fold dilutions of HSV-2(G) (0.01-1000 pfu/cell) (50 uL/well; final total volume, 100 L/well). After a 1-h adsorption, the inoculum was removed; cells were washed and overlaid with DMEM containing 0.5% methylcellulose. The virus titer in the presence of CVL samples (before or after application of gel) was determined by counting plaques 48 h after infection, after immunostaining by black plaque assay . Only wells with 25-100 plaques were used to calculate the virus titer.
Measurement of PRO 2000 levels in CVL samples.
A working stock was prepared by diluting 0.5% PRO 2000 gel 1 : 5 (final concentration, 1 mg/mL) in CVL fluid (pooled from screening samples). Standards were then generated by preparing serial 2-fold dilutions of the working stock in PBS (concentration range, 3.125-500 ug/mL). Next, 20-uL aliquots of the standard solutions or the test CVL samples were combined with 620 uL of buffer (0.1 mol/L sodium phosphate [pH 7] containing 0.1 mol/L SDS), and the fluorescence was measured in triplicate. The fluorescence readings (fluorescence spectrophotometer [Perkin-Elmer model LS-5] with excitation set at 330 nm and emission at 355 nm) for the standard solutions were plotted against the concentrations of PRO 2000, and the best straight line was calculated. Unknown PRO 2000 concentrations were calculated by reference to the standard curve.
Cytokine and secretory leukocyte protease inhibitor (SLPI) measurement. Levels of interleukin (IL)1beta, IL-8, and SLPI in CVL samples were measured in triplicate by use of commercial ELISAs (R & D Systems). To determine whether CVL itself or either of the study gels interfered with cytokine detection, cytokine standards were diluted in saline buffer, pooled CVL samples (obtained at screening), pooled CVL samples containing 100 ug/mL 0.5% PRO 2000 gel, or pooled CVL samples containing 100 ug/mL placebo gel, and results were compared.
Statistical analysis. Differences between groups with respect to age, CD4 cell count, virus load, pH of CVL samples, and baseline amounts of IL-1beta, IL-8, and SLPI were compared by t tests. Differences between groups with respect to fold reduction in viral infection after application of drug or placebo gel relative to titers in screening CVL samples were determined by Wilcoxon nonparametric test (with exact P values) on log-transformed data. Differences between groups with respect to cytokines and SLPI were compared by analysis of covariance and were adjusted for baseline (after log transformation of data). Calculations were done using SAS software (version 9; SAS Institute).