Abstract
We carried out a phase 1 trial of a candidate vaginal microbicide gel against HIV-1 and other sexually transmitted diseases, which contained cellulose acetate 1,2-benzenedicarboxylate (also known as cellulose acetate phthalate) in a glycerol-based vehicle. We had to terminate the study after five women had completed dosing, due to all women experiencing unacceptable vulvo-vaginal side-effects. Further investigations showed that the gel had a very high osmolality, which we believe led to excessive fluid transudation across the vaginal mucosa and acute mucosal dysfunction. We also showed that the rheology of the gel changed dramatically on fluid dilution. The osmolality and rheology of candidate microbicides and other genital mucosal products should therefore be analysed and considered at an early stage of product development.
INTRODUCTION
Control of the global HIV-1 pandemic will only be achieved when the incidence of new infections can be significantly reduced, and there continues to be an urgent need for new methods of preventing HIV-1 transmission. 1 Vaginal microbicides are one such promising approach, which if proved effective could be female-initiated and could block heterosexual transmission. One strategy in microbicide development is to investigate drugs or pharmaceutical ingredients that are already approved for other uses. Screening of approved pharmaceutical excipients against HIV-1 infection showed that cellulose acetate phthalate (CAP, also known as cellulose acetate 1,2-benzenedicarboxylate) had moderately potent anti-HIV-1 activity with IC50s in the range of 5–100 µg/mL. CAP, acting as a polyanion, inhibits HIV-1 infection via binding to the envelope glycoprotein gp120 on the viral particle, or via direct virus inactivation. 2,3 In the cervical explant model CAP blocked HIV-1 infection by a range of cell-free and cell-associated primary isolates independent of viral subtype and co-receptor usage. 4 CAP also protected rhesus macaques against vaginal challenge with pathogenic X4 and R5 SHIV. 5 Furthermore, CAP is active in vitro against a range of sexually transmitted disease pathogens, including HSV2, Neisseria gonorrhoeae, Chlamydia trachomatis, Trichomonas vaginalis, Haemophilus ducreyii and Mycoplasma hominis, but is inactive against lactobacilli. 2 Furthermore, CAP, which has been used for enteric coating of tablets and capsules, has an established safety record and is classified as Generally Regarded As Safe (GRAS).
Vaginal microbicides need to be formulated as products that are acceptable for repetitive or continuous vaginal delivery in women. Most microbicide candidates under development are formulated in water-based gelling agents such as hydroxyethylcellulose. 6 CAP is poorly soluble in aqueous solutions at pH < 6.0, and also hydrolyses and becomes inactive during storage in aqueous solutions. Therefore, insoluble micronized CAP was used in the development of the CAP vaginal microbicide gel, with glycerol, an anhydrous organic solvent as the primary vehicle for suspension of micronized CAP, and other GRAS compounds completing the formulation. This final formulation was tested in the 10-day rabbit vaginal toxicity test and classified as non-irritant to the vaginal mucosa.
METHODS
Clinical study
Women aged between 18–45 years were recruited via advertisements in local media and invited for a screening visit. Participants were eligible if they were sexually active with a male partner, HIV negative, at low risk of HIV and using a reliable method of contraception with a regular menstrual cycle (>26 days in duration). We excluded women who were pregnant or within 12 weeks postpartum, had current genital ulceration, were currently experiencing pain during sexual intercourse, were currently participating in another trial or whose male partner was unwilling to use condoms during the gel use. We also excluded those who had experienced post-coital or intermenstrual bleeding, received chemotherapy or immunotherapy, received treatment for cervical-intraepithelial neoplasia within the past three months or who had had gynaecological instrumentation of the cervix within the past one month. Women with untreated gonorrhoea, chlamydia, trichomoniasis or bacterial vaginosis were also ineligible (although participants were eligible after treatment). All participants had to be capable of providing informed consent.
Participants who provided informed consent and were still eligible after screening attended in their next menstrual cycle for the first dose of the CAP vaginal gel. The vaginal gel consisted of 3 mL of extractable volume of gel per dose containing 13% micronized CAP suspended in a glycerin-based gel (Table 1). Each tube of gel dispensed contained one dose and the participant transferred the gel into a vaginal applicator before use.
Formulation of the 13% CAP gel used in the trial
CAP = cellulose acetate phthalate
Baseline investigations were carried out before the first dose, and colposcopy was carried out 6 to 12 hours after the first dose of their gel to check for any immediate genital adverse events (AEs). Providing colposcopy was acceptable at this visit, the participant proceeded to use the gel on a daily basis for the following 13 days (±2 days for ease of scheduling follow-up visits). Follow-up investigations were carried out 7 and 14 (±2) days after the first dose.
Genital, clinical and laboratory AEs were recorded and classified according to a predefined grading system (grade 1 = mild, 2 = moderate, 3 = severe, 4 = extreme). Genital AEs were solicited using questions related to unexpected vaginal bleeding, genital discomfort (itching, burning), dysuria, dyspareunia and from examination findings including erythema, oedema, epithelial disruption and vascular lesions. Data on laboratory AEs were collected through routine tests or in response to a clinical event. Data on other clinical events were collected through open questions and information recorded by participants on a diary card.
The impact of CAP on vaginal flora was assessed by Nugent score at each follow-up visit. At the end of the dosing period, participants were asked whether they had any problems with using the gel, whether they experienced any abnormal vaginal discharge or whether any features of the gel (such as taste, smell) bothered them.
The primary endpoint was defined as the number of participants experiencing a grade 3 or above genital, clinical or laboratory AE or any event leading to discontinuation of the gel. Secondary endpoints were defined as grade 1 or 2 AEs attributable to CAP and whether participants would be willing to use the gel in the future.
We aimed to enrol 10 participants into the study. Due to events described in the Results section, we only enrolled five out of the planned 10, and then conducted further osmolality studies.
Osmolality and rheology studies
An osmole (Osm) is a unit of measurement of the number of moles of a chemical that contributes to a solution's osmotic pressure. Osmolality is the measure of the number of osmoles of solute per kilogram of solvent, and is expressed as mOsm/kg. We measured the osmolality of the 13% CAP gel and a variety of other over-the-counter (OTC) products used for female hygiene and other topical uses, using a Vapor Pressure Osmometer (model 5520, Wescore Inc, Logan, UT, USA), according to the manufacturer's instructions. Distilled water, physiological saline, phosphate-buffered saline and vaginal fluid stimulant were assessed as references. Glycerol at different dilutions (in distilled water) was also tested. Each sample was tested in triplicate and three readings were recorded for each test. Finally, we measured the viscosity of both the original CAP 13% gel, and the gel diluted 1:1 with water, at 23 ± 2°C, using a Brookfield Viscometer, Model RVDV I+, using Spindle 14 or other appropriate spindle at 5 rpm.
RESULTS
Clinical
Five women were screened and recruited into the study. Their mean age was 30 years (range 23–40) and all participants were of white ethnicity. All five participants were nulliparous although one woman had had two previous pregnancies. No participants had any significant medical history or abnormal findings on physical or gynaecological examination.
All participants completed all follow-up visits. Four participants used 14 doses of the gel, and one 11 doses during the two-week dosing period. No abnormalities were found on haematology and biochemistry screening during or after the dosing period.
One participant stopped using gel on her own decision on day 11 of the dosing period following the development of a number of symptoms related to gel usage. This subject had Candida albicans isolated at day 7 of dosing, had vulval oedema at day 14, and thus fulfilled the primary endpoint as defined above. All participants experienced at least one local genital AE (number of events per participant ranged from 1 to 8), all of which were deemed to be attributable (possibly/probably/definitely related) to CAP. AEs experienced were: itching in genital area (4/5 participants); burning or hotness in genital area (2/5); dyspareunia (1/5); dysuria (1/5) and candida vulvovaginitis (2/5 – one was diagnosed after the dosing period at the follow-up visit). Four out of five participants experienced clinical (non-genital) AEs attributable to CAP (nausea, 2/5; lower abdominal discomfort, 2/5 and dizziness, 1/5). All local and clinical AEs attributable to CAP were mild and with the exception of one participant, these all lasted less than one week.
When questioned about the acceptability aspects of the gel, four women spontaneously reported what the investigators felt was an unusual symptom. This was that they noticed a distinct feeling of ‘wetness’ in the hours after gel application. As the gel was usually applied before going to bed, this occurred at night time, and three subjects reported a large amount of gel leakage for several hours after application, necessitating the use of pads, knickers or pyjamas. The investigators thus reviewed the study findings after recruitment of five participants, and decided that the level of leakage observed was unacceptable, that therefore the product was not acceptable, and that the study should be terminated. It was also decided that we needed to investigate the characteristics and rheology of the gel to try and determine the reasons for the excess leakage and wetness of the gel that had not been anticipated from the preclinical studies.
Osmolality and rheology studies
The results of the measurements of the osmolality of the 13% CAP gel and various OTC products are shown in Table 2. It can be seen that the CAP gel has a high osmolality, 8511 mOsm, which is ∼30 times higher than vaginal secretions or other physiological fluids. This is also ∼12 times higher than a commonly used OTC vaginal lubricant. Figure 1 shows that this high osmolality is principally due to the glycerol component of the CAP gel, which comprised 57% of the product used in the clinical trial. The results of the investigation of the viscosity of the gel, and the effect of dilution are shown in Table 3. This showed that the CAP 13% gel itself had a very high viscosity, i.e. was ‘stiff’ in its original form, but dilution with water revealed a dramatic fall of more than a thousand fold in viscosity.
Osmolality measurements of a series of glycerol dilutions
Osmolality of the 13% CAP gel, other reference solutes and over-the-counter products
CAP = cellulose acetate phthalate
Viscosity of the CAP 13% gel and a 1:1 dilution
CAP = cellulose acetate phthalate
DISCUSSION
We here report a phase 1 trial of a candidate vaginal microbicide, cellulose acetate 2,3-benzenedicarboxylate. The protocol that was approved for the trial envisaged two parts; Part A, open label, studying 10 HIV-negative women, with post-first-dose colposcopy, and then if interim results were encouraging proceeding to Part B, studying 50 women, including 10 women with HIV infection. However, we were surprised to observe AEs and significant problems with acceptability of the gel in four of the first five women in the trial, and therefore decided to stop the clinical trial.
Further studies revealed that the 13% CAP gel had a high osmolality, caused by the high glycerol concentration, which was significantly in excess of similar marketed vaginal products, and ∼30 times in excess of serum osmolality. We propose that the 13% CAP gel administered into the vagina exerted a strong osmotic pressure across the cervico-vaginal epithelium and stroma causing transudation of tissue- and serum-derived fluids into the vaginal lumen. We were even more surprised by the results of rheology studies, which showed the viscosity of the CAP gel dropped precipitously upon dilution with water, resulting in a virtual ‘collapse’ of the rheology of the original gel. We propose that the administration of the hyperosmolar CAP gel into the vagina caused a brisk transudation of fluid across the cervico-vaginal epithelium into the vaginal lumen, which in turn caused the CAP gel to become hypo-viscous, resulting in the wetness and watery discharge observed by most of the participants. In two of the five participants further mucosal abnormalities including vulval oedema and candida vulvovaginitis were observed.
Related observations have recently been made in regard to the use of hyperosmolar lubricants administered into the distal colon. 7 These investigators administered radio-labelled iso-osmolar (283 mOsm/kg) and hyperosmolar (3429 mOsm/kg) rectal lubricants and observed greater epithelial denudation and greater production of luminal secretions with the hyperosmolar gel compared with the isoosmolar gel. The importance of epithelial barrier function in protection against HIV infection, even when the epithelium is intact without overt disruption, has also been recently demonstrated. 8 Two microbicides, nonoxynol-9 and cellulose sulphate, which have been associated with an increased risk of HIV acquisition, were shown to cause disruption of epithelial tight junctions, and increased transcytosis of HIV across such affected epithelia in model systems.
We therefore suggest that osmolality and rheological characteristics of candidate microbicides and other mucosally applied products should be measured at a very early stage of their product development. This is not only to maximize their acceptability in terms of clinical characteristics during use, but also to design formulations that are most likely to have no negative characteristics in terms of HIV transmission or mucosal toxicity – primum non nocere (first do no harm). The laboratory analyses we made of the CAP formulation seem to explain all the negative findings observed during the phase trial. Identical formulations of the 13% CAP gel had been tested vaginally, prior to the human trial, in both rabbits and macaques without adverse effects being observed. 9 This would appear to indicate that both the rabbit and macaque models may not be a good predictor of certain aspects of human vaginal physiology in relation to microbicides, and reinforces the need for preclinical formulation testing. Our belief is that the formulation itself was the cause of the side-effects observed, and we have no evidence that any toxicity was caused by the active moiety, cellulose acetate 2,3-benzenedicarboxlate. This leaves the way open to further attempts to formulate CAP as a vaginal microbicide.
Footnotes
ACKNOWLEDGEMENTS
We thank all the female volunteers for their participation in the trial, and also Dr Robert Neurath, Dr Jim Turpin and Prof Jonathan Weber for their support and expert advice. We acknowledge the support of the Microbicides Development Programme, funded by the UK Medical Research Council and the Department for International Development, and the National Institutes of Health of the United States (U19 AI 076964).