The Sheep as a Model of Preclinical Safety and Pharmacokinetic Evaluations of Candidate Microbicides

Abstract

When developing novel microbicide products for the prevention of HIV infection, the preclinical safety program must evaluate not only the active pharmaceutical ingredient but also the product itself. To that end, we applied several relatively standard toxicology study methodologies to female sheep, incorporating an assessment of the pharmacokinetics, safety, tolerability, and local toxicity of a dapivirine-containing human vaginal ring formulation (Dapivirine Vaginal Ring-004). We performed a 3-month general toxicology study, a preliminary pharmacokinetic study using drug-loaded vaginal gel, and a detailed assessment of the kinetics of dapivirine delivery to plasma, vaginal, and rectal fluid and rectal, vaginal, and cervical tissue over 28 days of exposure and 3 and 7 days after removal of the ring. The findings of the general toxicology study supported the existing data from both preclinical and clinical studies in that there were no signs of toxicity related to dapivirine. In addition, the presence of the physical dapivirine ring did not alter local or systemic toxicity or the pharmacokinetics of dapivirine. Pharmacokinetic studies indicated that the dapivirine ring produced significant vaginal tissue levels of dapivirine. However, no dapivirine was detected in cervical tissue samples using the methods described here. Plasma and vaginal fluid levels were lower than those in previous clinical studies, while there were detectable dapivirine levels in the rectal tissue and fluid. All tissue and fluid levels tailed off rapidly to undetectable levels following removal of the ring. The sheep represents a very useful model for the assessment of the safety and pharmacokinetics of microbicide drug delivery devices, such as the vaginal ring.

INTRODUCTION

With an estimated 2.3 million new human immunodeficiency virus type 1 (HIV-1) infections in 2012 (1) and a total number of people living with HIV estimated at 35.3 million in the same year, the need for effective prevention strategies remains critical. Although barrier prevention strategies, such as male and female condoms, are effective, and behavioral programs teaching the benefits of monogamy and abstinence have shown some success (2–4), they have not significantly impacted the epidemic. One area of key concern is sub-Saharan Africa, where 57% of HIV-infected people are women (1). In order to provide prevention strategies to women in this high-risk population, locally (vaginally or rectally) applied products containing potent antiretroviral drugs are being developed by many different groups. These microbicides represent one of the most promising prophylactic strategies currently being developed in terms of being simple to self-administer, relatively easy to produce, and by virtue of the lower drug levels required compared to those with systemic approaches, the potential to be within reach of the health system and nongovernmental organization (NGO) budgets in the developing world. A key advantage to the topical microbicide approach is also that the same targeted active ingredients can be delivered in multiple different dosage forms (e.g., gels, vaginal rings, films, and tablets) that will better suit regional and individual use preferences.

As with any pharmaceutical product, the development of microbicides requires extensive safety testing in preclinical species prior to the initiation of clinical trials. For most pharmaceutical products, the focus is on systemic safety and primarily concerns the active pharmaceutical ingredient (API). For microbicides, it is also important to address the systemic toxicity of the product, but since these products are administered topically, there is also a need to evaluate the local effects of both the API and the formulated product intended for clinical use. This is particularly important for products intended to prevent HIV transmission, because there is the potential that even minor levels of local irritation might increase HIV susceptibility through disruption of the epithelium or recruitment of CD4 cells that are the target for HIV.

For vaginal microbicides, local tolerability and irritation are typically addressed using a rabbit vaginal irritation assay (5). The rabbit is preferred by regulatory authorities because of its sensitivity to vaginal irritants and because of the wealth of experience with this model in safety evaluations of spermicides (6). However, histological differences between the rabbit and human vaginal epithelia likely mean that this species is overly sensitive (7). In the rabbit, the epithelium is made up of a single layer of columnar cells (7), whereas in women, the epithelium consists of multiple layers of squamous cells (8). Consequently, alternative animal models that are histologically closer to humans may have greater clinical relevance for testing of vaginal microbicides.

Another critical consideration in the preclinical development of microbicides is the determination of drug concentrations in target tissues. For drugs that target the virus directly, measurement of luminal concentrations may be informative, but for drugs with intracellular mechanisms of action, such as those that target the viral reverse transcriptase, integrase, or protease enzymes, and probably those that interact with proteins expressed on the target cell surface, such as coreceptor antagonists, it is important to establish concentrations within the target tissues. Biopsy specimens of vaginal or cervical tissue can be collected during clinical trials and analyzed for drug concentrations, but data obtained using this approach are limited in their utility, because of the inability to distinguish the quantities of drug present within the living cells of the epithelium from those remaining on the epithelial surface or associated with the dead cells of the keratinized layer. Preclinical studies allow for the collection of whole tissues during postmortem examinations, which can be processed and analyzed in ways not possible with clinical sampling procedures. Again, for these purposes, the rabbit has limited utility because of the interspecies differences between rabbits and humans in anatomy, histology, and physiology.

Here, we present data obtained in toxicology and pharmacokinetic studies in the sheep. The sheep has a number of characteristics that make it suitable for use in preclinical evaluations of microbicides. In particular, the cervix and vagina are anatomically and histologically similar to those of women (9), and since the dimensions of the lower female reproductive tract are also similar in the two species, intravaginal products intended for human use, including medical devices, such as intravaginal rings, can be evaluated in sheep without modification. In addition, sheep are widely available, relatively inexpensive, and have a temperament that makes them easy to handle and examine under laboratory conditions; therefore, they can be used in studies conducted in compliance with the principles of good laboratory practice (GLP). A number of studies utilizing sheep in the preclinical evaluation of microbicides have been reported (10–12).

The studies described here were conducted to evaluate the nonnucleoside reverse transcriptase inhibitor (NNRTI) dapivirine, a substituted di-amino-pyrimidine (DAPY) derivative with potent antiviral activity against HIV-1 (13). They include a 3-month toxicology study with vaginal ring and gel formulations, a 1-month pharmacokinetic study with the same vaginal ring formulation, and a 5-day pharmacokinetic feasibility study using another gel formulation. Dapivirine was licensed to the International Partnership for Microbicides (IPM) by Janssen Pharmaceuticals. Phase III clinical trials of the dapivirine vaginal ring were initiated in early 2012. Pharmacokinetic data are available from multiple clinical studies with the dapivirine vaginal ring (14) and dapivirine gel formulations (15–17) and from studies in rhesus macaques and rabbits using radiolabeled dapivirine in a gel formulation (18).

MATERIALS AND METHODS

Animals.

Female Suffolk cross sheep were used in all studies. The sheep were housed at Huntingdon Life Sciences (Huntingdon, Cambridgeshire, United Kingdom) in pens with a concrete floor and concrete/galvanized metal sides (five sheep per pen, according to treatment group). Straw was provided as a bedding material, and the building provided natural light supplemented by overhead fluorescent tubes to provide a 12-h light/12-h dark photoperiod throughout the study. Natural ventilation was supplemented as necessary by roof extractor fans. Each animal was offered a standard dry pelleted concentrate diet (Ewe and Lamb ration; Dodson & Horrell) once daily in the morning. Good-quality meadow hay was offered ad libitum, as was fresh drinking water. The studies were conducted in accordance with applicable sections of the United Kingdom Animals (Scientific Procedures) Act of 1986 and the associated codes of practice for the housing and care of animals used in scientific procedures and the humane killing of animals under schedule 1 to the act, issued under section 21 of the act.

Materials.

Dapivirine was manufactured under a contract to IPM by N. V. Ajinomoto Omnichem S.A. (Wetteren, Belgium). The dapivirine gels were manufactured by IPM (Bethlehem, PA, USA), and the dapivirine rings were manufactured by IPM or by Qpharma (Malmö, Sweden). The excipient and active components of the rings and gels are described in Tables 1 and 2. Tenofovir was provided by Gilead (Foster City, CA). The tenofovir gel was manufactured under a contract to IPM by Particle Sciences, Inc. The rings had an overall ring diameter of 56.0 mm and a cross-sectional diameter of 7.7 mm.

TABLE 1.

Excipients and active components of test gels

Excipient/componenta	Amt (% by weight) in gel containing:
	0.5% (wt/wt) dapivirine (formulation 4789)	0.2% (wt/wt) dapivirine (Gel-002b)	1% (wt/wt) tenofovir
Dapivirine	0.500	0.200	1.000
Methylparaben NF	0.200	0.200	0.180
Propylparaben NF	0.050	0.050	0.020
Propylene glycol NF	5.000		20.000
Polyethylene glycol 300		34.800
Hydroxyethylcellulose NF (Natrosol 250H Pharm)	3.500	2.000	2.500
Polycarbophil (Noveon AA-1)	0.200	0.650	0.050
Carbopol 1382 polymer		0.600
18% NaOH aqueous solution	0.065	0.090	3.750
Citric acid USP			1.000
Final pH	4.7	4.7	4.6
Purified H2Oc	q.s. 100	q.s. 100	q.s. 100

^aNF, national formulary; USP, U.S. Pharmacopeial.

^bPlacebo Gel-002 was formulated exactly as per Gel-002, except that dapivirine was not included and was replaced with additional polyethylene glycol 300.

^cq.s. 100, purified water was added in sufficient quantity to total 100% by weight.

TABLE 2.

Excipient and active components of test rings

Component	Function	Amt per ring (mg)
Dapivirine	Active pharmaceutical ingredient	25
Silicone liquid (NuSil MED-360)	Dispersing agent	175
Silicone elastomer part A (NuSil MED-4870 part A)a		3,900
Vinyl-terminated polydimethylsiloxane (linear) polymers	Polymer
Platinum	Catalyst for cross-linking reaction
∼30% amorphous (noncrystalline) reinforcing silica	Filler
Silicone elastomer part B (NuSil MED-4870 part B)a		3,900
Vinyl-terminated polydimethylsiloxane (linear) polymers	Polymer
Hydride functional polydimethysiloxane polymer	Cross-linker
∼30% amorphous (noncrystalline) reinforcing silica	Filler
Proprietary manufacturing information	Inhibitor

^aPart A and part B refer to a two-part translucent silicone system used with injection molding equipment.

The placebo ring was manufactured with the same components and dimensions as the dapivirine-containing ring, except that it contained U.S. Pharmacopeial (USP) titanium dioxide dispersed in silicone fluid as colorant and no dapivirine. The addition of colorant was for the purpose of maintaining blinded conditions during the clinical evaluation. The content of the colorant in the ring was ≤0.05%.

Since each of the three studies described here focused on different objectives and/or different dosage forms and because this meant that a number of dosages were used, the rationale for the selection of dose level and dosage form is provided below for each study.

Study 1, a 3-month toxicology study with intravaginal ring and gel.

Fifty sheep weighing between 57.0 and 93.0 kg (5 years of age) received either a placebo vaginal ring, the dapivirine vaginal ring, placebo Gel-002, or dapivirine Gel-002 (n = 10 per group). An additional group of 10 animals remained untreated. The dapivirine rings were exactly the same as the rings evaluated in clinical trials and contained 25 mg of dapivirine. A 0.2% (wt/wt) concentration was selected for the dapivirine gel, because this concentration had been used in a previous 9-month intravaginal study in rabbits and caused no local or systemic toxicity. After the first 30 days of treatment, 5 animals from each group were killed and examined in a full necropsy (histopathology tissue list presented in Table S1 in the supplemental material). The remaining 5 animals per group continued treatment for a further 60 days (90 days total dosing) before being killed and examined in a full necropsy.

Animals receiving vaginal rings were restrained in a comfortable standing position, and the rings were carefully inserted into the cranial vagina using a gloved finger and a small amount of medical-grade lubricant, as appropriate. The ring location was confirmed using a speculum. Medical-grade string was attached to each ring prior to application, and the ends of the string were left hanging outside the vagina to facilitate visual checks of the ring presence and to aid ring removal. Thirty days after insertion, the rings were removed and the animals were either euthanized for necropsy or a new ring was inserted for a further 30 days. Animals receiving vaginal gel were dosed daily. Each animal was restrained in a comfortable standing position, and a 2.5-ml volume of gel was administered using a disposable syringe and rubber catheter to reach the cranial vagina. For all dosing procedures, great care was taken to avoid any damage to the vagina.

During the study, clinical condition, body weight, food consumption, ophthalmic examination, hematology, blood chemistry, urinalysis, toxicokinetics, organ weight, macropathology, and histopathology investigations were undertaken.

This study was performed in compliance with all CFR title 21 part 58 (19) of the good laboratory practice regulations and the applicable GLP regulations in the European Union (EU) and Japan, and according to International Conference on Harmonisation (ICH) guidelines for toxicokinetics and nonclinical safety studies for the conduct of human trials (S3A [20] and M3 [21], respectively). Dose formulation concentrations were confirmed using an ultrahigh-performance liquid chromatography (UPLC) method, and the stability and suitability of the test materials for the duration of the study were demonstrated. The analytical method was validated on a Waters Acquity UPLC system with UV detection and entailed the extraction of dapivirine from the sample using acetonitrile (high-performance liquid chromatography [HPLC] grade) in water (50:50 [vol/vol]) and chromatographic separation using a 1.7-μm, 2.1- by 50-cm Waters Acquity BEH C₁₈ column at a 1.0-ml/min flow rate of acetonitrile/water/o-H₃PO₄. Dapivirine samples were injected (2-μl injection volume), and UV detection at a wavelength of 290 nm was used to determine dapivirine content, with an approximate retention time of 1.43 min. Blood samples (1.0 ml) were drawn from the jugular vein of sheep into tubes containing lithium heparin anticoagulant on day 1 and immediately prior to termination (i.e., 5 animals per group on day 30 and 5 animals per group on day 90) predose and 2, 4, 6, 8, 12, and 24 h after dosing. Plasma samples were analyzed using a validated liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for the detection of dapivirine in sheep plasma (Analytisch Biochemisch Laboratorium BV [ABL], Assen, The Netherlands) with an analytical range of 5.00 to 5,000 pg/ml. Dapivirine and internal standard (D₄ isotope-labeled dapivirine) were extracted from plasma samples using solid-phase extraction (Oasis MCX 1-ml [30-mg] cartridges; Waters). The chromatographic separation was performed on a reversed-phase 3.5-μm C₁₈ column with a gradient elution of a mixture of 100 mM ammonium acetate, water, and acetonitrile. The samples were injected on an Agilent 1100 HPLC system equipped with a Sciex API 4000 MS/MS detector. Dapivirine and the internal standard were quantified in negative ionization mode by multiple reaction monitoring (MRM). The mass transitions m/z (mass-to-charge ratio) 328→142 and m/z 332→146 were used to measure dapivirine and the internal standard, respectively, with a run time of 10 min per analysis.

Study 2, local pharmacokinetics of vaginally administered dapivirine and tenofovir gels.

A second study was performed to evaluate whether dapivirine could be detected in postmortem vaginal tissue slices. In this study, tenofovir was used as a reference control, since it was previously shown to penetrate vaginal tissue in humans (22).

A confounding issue in the assessment of local delivery of active drug to target tissues relates to the potential for locally administered drug to remain on the surface of the tissue or in the dead cells of the keratinized layer. This is particularly a concern for gels, since the relatively high levels of drug in the gels could markedly affect the measured tissue concentration if residual gel remained associated with the analyzed sample. For this study, methods were developed for vaginal lavage, postmortem excision of the vaginal tissue, and dissection to produce tissue blocks with minimized residual gel contaminations at the time of necropsy. These methods (described below) were optimized to constrain any potential source of contamination of deeper tissue levels with surface drug or gel, and then frozen sections were taken in order to determine the tissue levels at different depths relative to the mucosal surface. In this way, the level of drug penetration into the vaginal tissue was determined without the confounding factors of the surface drug levels or drug uptake into the keratinized surface layers.

Four sheep (2 per group) received daily vaginal doses of gel containing either 0.5% dapivirine (2.5-ml dose volume) or 1% tenofovir (4-ml dose volume) for 5 days. The gel composition is provided in Table 1. In this study, the animals were dosed during September, just prior to the initiation of the breeding season in the United Kingdom. There was no investigation to determine whether the animals had entered an active sexual cycle or which stage within the cycle they were. The dapivirine concentration was selected as the maximum that has been shown to be stable in a gel formulation to ensure the greatest likelihood of detecting the compound in the tissue. The same tenofovir formulation and concentration as was shown to penetrate human vaginal tissue (22) were used. The gels were administered using the method described above.

At the end of dosing (2 h ± 10 min after the last dose), the animals were euthanized and the vagina excised. Before dissection, vaginal lavage was carried out using warm (approximately 37°C) saline solution in order to rinse any loosely adhering gel from the vaginal mucosa. The cranial vagina was carefully excised, and four subsamples (2 left [L] and 2 right [R]) were collected, each approximately 2 cm by 2 cm. Each subsample was pinned to a cork board with the mucosa uppermost. One L subsample and one R subsample were swabbed gently using a swab moistened with saline solution to remove any adherent gel formulation, taking care to avoid damage to the vaginal mucosa. The second L and R subsamples were not swabbed. All subsamples were then snap-frozen in prechilled isopentane (separate containers) and stored at −80°C. Each subsample was clearly labeled as L or R and swabbed or unswabbed. Precautions against the contamination of tissues with dose formulation were taken at all times during the necropsy procedure. A transverse section (perpendicular to the mucosal/serosal surfaces) was taken from each tissue block, mounted on slides, and stained with hematoxylin and eosin (H&E) in order to provide morphological orientation/depth location information.

The individual frozen tissue samples were trimmed (using a different precooled disposable scalpel or razor blade per cut made) to remove the edges that may have contained formulated drug from cuts made prior to freezing the tissue. The samples were mounted on cork using a small amount of optimum cutting temperature (OCT) embedding medium. Each sample was serially sectioned with a target section thickness of 20 μm in a cryotome and on a plane parallel to the mucosal/serosal surface. Sectioning started at the mucosal surface for the 2 samples per animal taken from the left side of the vagina (1 swabbed and 1 not swabbed) and at the serosal surface for the 2 samples per animal taken from the right side of the vagina (1 swabbed and 1 not swabbed). The sections were immediately transferred (using clean tweezers) into precooled sample tubes (polypropylene tubes of 1 to 2 cm in diameter) pooled in groups of consecutive sections to provide a minimum of 50 mg of analytical sample size and stored at −80°C. The total wet tissue weight (in whole milligrams) placed in each tube was determined at the time of tissue collection. The target tissue weight per tube was 100 to 150 mg, representing between 15 and 18 slices. Precautions against the cross-contamination of tissue sections were taken at all times during processing/sectioning, including cleaning with alcohol and replacement of the cryotome blade and tweezers between each analytical subset of serial sections and cleaning of the specimen stage between each tube/sample.

The tissue samples were analyzed for tenofovir (Analytisch Biochemisch Laboratorium BV [ABL], Assen, The Netherlands) by LC-MS/MS in the range of 1 to 100 ng/g of tissue using a method based on that previously described by Choi et al. (23). The tissue samples were analyzed for dapivirine (ABV) by LC-MS/MS in the range of 0.05 to 100 μg/g of tissue. Tissue samples of approximately 20 mg were accurately weighed and subsequently homogenized with 0.9% NaCl solution. The homogenate was sonicated by placing it in an ultrasonic bath for 10 min in order to disrupt any intact cells. The homogenate was alkalized by adding NH₄OH, and 4 ml of tertiary butyl methyl ether was added to each sample prior to extraction on a rotation table for 20 min. The water phase was discarded, and the organic phase was evaporated under a stream of nitrogen. The samples were reconstituted in 100 mM ammonium acetate in 60% acetonitrile (ACN). The extracts were analyzed using methods similar to those described above for plasma samples.

The tissue levels of dapivirine and tenofovir were determined, and the depth of each sample from the mucosal surface was approximated based on the number of slices in each sample tube and an assumed slice thickness of 20 μm. The tissue levels were correlated against the shallowest depth attributed to that tube (for examples, samples ranging from 300 to 600 μm from the mucosal surface were represented by a depth of 300 μm).

Study 3, local pharmacokinetics of dapivirine delivered from a vaginal ring.

Following establishment of the feasibility of the tissue penetration methodology, another study was conducted using the same technique to determine the local concentrations of dapivirine achieved using the vaginal ring formulation that has progressed into phase III clinical trials and was evaluated in the 90-day toxicology study (see Tables 1 and 2). The vaginal ring was placed in the cranial vagina of three groups of three female Suffolk cross sheep on day 1, in the same manner as described above, and left in place for 28 days. This study was conducted in October, during the first month of the United Kingdom breeding season, and as such, it was assumed that all animals were actively cycling. However, the stage of the sexual cycle was not monitored, and therefore, it cannot be ruled out that this may have influenced the pharmacokinetic profile and interanimal variations observed. Group 1 animals were euthanized within 10 min of ring removal on day 29, group 2 animals were euthanized 3 days after ring removal (day 32), and group 3 animals were euthanized 7 days after ring removal (day 36). The time points at which samples of plasma, vaginal fluid, vaginal tissue, cervical tissue, rectal fluid, and rectal tissue for determination of dapivirine concentrations were collected are provided in Table S2 in the supplemental material).

Blood samples (1.0 ml) were collected from the jugular vein into tubes containing lithium heparin anticoagulant. The samples were spun at 1,500 × g and 4°C for 10 min to separate the plasma from the blood cells and then frozen at −20°C within 1 h of sampling. Vaginal fluid was sampled by means of Weck-Cel sponges held in place in the vagina for 1 min. Rectal fluid was sampled by means of Weck-Cel sponges prewetted with 100 μl of a 0.9% sodium chloride solution and held in place in the rectum for 1 min. Owing to the thin epithelial layers in the rectum and the small volume of rectal mucus or secretions, the collection of rectal fluid samples occasionally resulted in slight rupture of the epithelium and the collection of small volumes of blood from the resultant capillary damage. While the volume of blood on the sponges was very small (traces of red visible), whenever rectal samples appeared to be contaminated with blood, a repeat sample was taken ≥1 h after the original sample (and ≥1 h prior to the next sample). The resampling was performed on the basis that the blood levels of drug may have been considerably higher than rectal fluid levels and as such would have biased any results. All sponges were placed immediately into stoppered containers and stored (and transported) at −20°C to prevent the potential loss of fluids or materials during storage and transport to the analytical facility.

Tissues were collected immediately postmortem, according to similar methods as described above for the vaginal tissue, with the exception that all mucosal surfaces were swabbed gently with a saline-moistened swab, and the tissues were sectioned from the serosal surface to the mucosal surface. Rectal tissue samples were not sliced but rather snap-frozen in prechilled isopentane as a tissue block, since no drug was delivered rectally and no surface contamination was anticipated, and so slicing was considered unnecessary. All tissue samples were snap-frozen in prechilled isopentane prior to freezing at −80°C for storage and transport to the analytical laboratory.

All samples of fluid, plasma, and tissue were analyzed using GLP-validated LC/MS-MS methods by ABL (Assen, The Netherlands).

Sheep plasma samples were analyzed for dapivirine using the validated LC-MS/MS method described above, with an analytical range of 5.00 to 5,000 pg/ml. Sheep vaginal and rectal fluid samples on Weck-Cel sponges were analyzed for dapivirine using a validated LC-MS/MS method with an analytical range of 1 to 1,000 pg/sponge. Dapivirine and the internal standard (D₄ isotope-labeled dapivirine) were extracted from the sponges using a mixture of 1% formic acid and methanol (1:3), followed by a second extraction with methanol. The extract was evaporated at 70°C under a stream of nitrogen, and the residue was reconstituted in a mixture of 0.1% NH₄OH and acetonitrile (2:3). The extracts were analyzed using the methods described above. The concentrations of dapivirine in sheep vaginal fluids are expressed as mass per gram of fluid collected (sponges were weighed before and after sampling at the animal facility). Rectal fluid levels are expressed as the absolute drug level per sponge, because it is not uncommon for prewetted sponges to weigh less after sampling the rectum than they did before insertion (our unpublished observations), making it impractical to express drug levels per unit of mass or volume.

Sheep vaginal tissue, cervical tissue, and rectal tissue samples were analyzed for dapivirine using the validated LC-MS/MS method described above in the range of 1 to 1,000 ng/g of tissue.

All rings used in this study were stored in individual bags and shipped at room temperature for analysis of the remaining drug content and residual drug levels as an indication of the total drug delivered to each animal over the 28-day period. In brief, rings were cut into sections to increase the surface area, and dapivirine was extracted by incubation of the ring segments in acetone for 24 h at ambient temperature. The acetone was then evaporated under a gentle stream of nitrogen, and the remaining dry residue was dissolved in 100% methanol. HPLC determination of the dapivirine content of a 10-μl aliquot was determined using UV detection at 257 nm in an Agilent 1100 system with a no. 178 Luna C₁₈ 30- by 4.6-mm analytical column fitted with a Phenomenex SecurityGuard system containing a C₁₈ (4- by 3-mm) precolumn kept at ∼25°C and a 65:35 (vol/vol) methanol-to-water mobile phase. In this assay system, the dapivirine retention time was 7.20 min.

RESULTS

Study 1, a 90-day toxicology study with intravaginal ring and gel.

Dapivirine administered intravaginally to sheep for up to 90 days as either a ring or gel formulation produced no signs of local or systemic toxicity in any parameter measured. No animals died or were killed prematurely, and there were no clinical signs indicative of a reaction to treatment. There was no effect of treatment on body weight or food consumption, and there were no ophthalmic findings. Hematology, blood chemistry, and urinalysis parameters were all within the normal range for sheep of this age and weight range. There were no macroscopic changes or differences in organ weights and no microscopic/histopathologic findings indicative of toxicity. After 30 days, one animal given the dapivirine ring had depressions on the vagina, while after 90 days, another animal given the dapivirine ring had a raised area in the vagina, but there were no corresponding histopathological observations. One animal treated with dapivirine gel had a dark area in the vagina, which correlated with brown pigmented macrophages in the submucosa and muscle. Cysts were observed after 90 days in one animal treated with the dapivirine ring and one animal given the placebo ring, and these findings correlated with histopathological findings of microscopic cysts or squamous cysts. Three animals had cysts that were not observed macroscopically. These and the other minor changes seen in the vagina (subepithelial lymphoid foci, inflammation of superficial layers of epithelium, epithelial/subepithelial inflammation, submucosal inflammation, and dilated/cystic ducts) were considered to be consistent with normal background findings and were not attributed to either the gels, rings, or dapivirine.

The mean maximum plasma concentration of dapivirine (C_max) and the mean area under the plasma concentration-time curve estimated up to 24 h (AUC_0–24) for days 1 and 30 or 90 are presented in Table 3. Systemic exposure to dapivirine (C_max and AUC_0–24) was higher for the gel than for the ring on days 1, 30, and 90 of treatment. The time at which the maximum plasma concentration occurred (T_max) in animals treated with the dapivirine vaginal ring was generally 2 h after dosing on day 1 (range, 2 to 4 h) but was more variable on days 30 and 90 and occurred in the range of 2 to 24 h after dosing. In the animals receiving dapivirine Gel-002, the T_max was generally 4 or 6 h after dosing and in the range of 2 to 12 h after dosing. For the ring, the C_max and AUC values were lower at the end of each 30-day period of use (i.e., day 30 and 90) than that on day 1. In contrast, repeat dosing with dapivirine Gel-002 resulted in higher C_max and AUC values on day 30 than those on day 1, but by day 90 of dosing, these values were lower than those on day 1. The terminal half-life could not be estimated for any animal treated with the dapivirine ring; however, for those treated with dapivirine Gel-002, the terminal half-life was in the range of 4.6 to 7.9 h (where it could be estimated).

TABLE 3.

Toxicokinetic parameters for female sheep administered the dapivirine vaginal ring or dapivirine Gel-002 and compared with human pharmacokinetic parameters with the same dapivirine vaginal ring formulation (14)^a

Treatment	Measurementb	Cmax (pg/ml)				Tmax (h)				AUC0–24 (pg · h/ml)
		D1		D30	D90	D1		D30	D90	D1		D30	D90
		I	T	I	T	I	T	I	T	I	T	I	T
Vaginal ring	Median	55.0	74.7	30.5	29	2	2	8	12	1,030	1,400	643	862
	Min	50.5	53.4	16.1	18.6	2	2	2	2	960	1,020	460	622
	Max	59.6	138	37.7	50.5	4	2	24	24	1,110	2,740	813	4,409
Gel-002	Median	1,100	861	1,230	414	6	4	8	4	9,020	7,790	16,700	3,320
	Min	161	701	634	237	2	4	6	2	16,200	12,700	21,500	18,300
	Max	1,490	1,360	1,450	1,140	6	6	12	8	766	5,050	7,410	1,890
Vaginal ring (human)	Median	348 (72)–359 (52)				169–336				2,939 (760)–3,103 (833)

^aThe sheep data presented are the median for 5 animals per group from the interim kill (I) on day 30 (D30) or the terminal kill (T) on day 90 (D90).

^bMin, minimum; Max, maximum.

Study 2, local pharmacokinetics of vaginally administered dapivirine and tenofovir gels.

The feasibility study in which gels containing either 1% tenofovir (4-ml dose volume) as a reference control or 0.5% dapivirine (2.5-ml dose volume) were administered for 5 days demonstrated that both compounds were detectable in vaginal tissue sections. Both animals that received 1% tenofovir gel had quantifiable concentrations of tenofovir in virtually all vaginal tissue samples analyzed throughout the full depth of tissue (Table 4 and Fig. 1). In one animal, there was an overall trend indicating a decline in tissue concentrations from the mucosal to the serosal surface. However, the opposite trend was indicated by the samples from the other animal, with a slight tendency toward an increase in levels from the mucosal to the serosal surface. In the two animals that received 0.5% dapivirine gel, dapivirine was quantifiable in all samples of vaginal tissue throughout the full depth of the vaginal wall (Table 4 and Fig. 1). There was an overall trend indicating a decline in tissue concentrations from the mucosal to the serosal surface. One sample (left unswabbed) from one animal presented very high concentrations of dapivirine (one to two orders of magnitude higher than those in other samples at the mucosal surface). It is likely that this reflects contamination of the analytical sample with residual gel, and therefore, this data point has been excluded from the graphical presentation of the mean data.

TABLE 4.

Dapivirine levels in postmortem vaginal tissues of sheep at various sampling depths

Approximate depth from mucosal surface (μm)	Drug levels (mean [SEM]) (μg/g of wet tissue) in:
	Swabbed tissue		Unswabbed tissue
	Tenofovir	Dapivirine	Tenofovir	Dapivirine
350	6.6 (1.9)	12.8 (6.0)	4.9 (0.8)	175.9 (174)
650	6.7 (1.6)	9.6 (4.3)	5.0 (1.3)	88.9 (88.7)
950	6.4 (1.1)	5.1 (1.9)	2.8 (0.8)	24.4 (25.4)
1,250	5.9 (1.9)	2.8 (1.7)	2.5 (0.2)	4.6 (2.2)
1,550	4.3 (1.1)	2.1 (1.2)	0.97 (0.2)	1.8 (1.8)
1,850	3.8 (1.1)	1.7 (1.5)	0.81 (0.4)	7.1 (7.5)
2,150	3.6 (1.6)	1.9 (1.3)	0.84 (0.7)	17.7 (19.1)
2,450	5.5 (2.9)	4.1 (4.0)	2.1 (1.7)	14.0 (12.5)
2,750	7.4 (3.4)	1.9 (0.7)	4.1 (3.4)	3.2a
3,050	7.6 (4.5)	1.7 (0.6)	6.1 (6.4)	3.1a
3,350	0.2a	3.5a	7.1 (9.5)

^aActual value (only 1 tissue sample was available).

FIG 1.

Tissue levels (mean ± SEM) of dapivirine and tenofovir in vaginal tissue samples. The tissue levels of dapivirine and tenofovir were determined in sheep vaginal tissue following daily vaginal administration of drug-containing gel (0.5% dapivirine or 1.0% tenofovir; n = 2 per group).Vaginal tissue samples were collected at various depths in the vaginal wall through serial sectioning in a cryotome, homogenized, and assayed using LC-MS/MS methods.

Examination of H&E-stained transverse sections of the tissue blocks indicated a distinct variability in thickness of the overall vaginal wall and the individual tissue layers, although the thickness of the different tissue layers was relatively proportionate to the overall sample thickness (Table 5).

TABLE 5.

Overall vaginal wall and tissue layer thickness of sheep

Tissue	Thickness rangea	% of total thickness
Overall vaginal wall	1.97–4.66 mm
Epithelium	73.5–179.9 μm	2–6
Lamina propria	204.9–683.7 μm	7–19
Muscle layer	1.08–2.7 mm	37–72
Adventitia	318–1,350 μm	12–37

^an = 4 sheep, 4 samples per sheep.

Study 3, local pharmacokinetics of dapivirine delivered from a vaginal ring.

In the main pharmacokinetic study in which groups of 3 sheep were treated with vaginal rings containing 25 mg of dapivirine for 28 days, the levels of dapivirine were determined in samples from blood plasma, vaginal fluid, vaginal tissue, cervical tissue, rectal fluid, and rectal tissue.

Plasma.

Maximum plasma concentrations (up to 108 pg/ml) occurred 2 h after ring insertion (the first sampling time point) in two out of three animals sampled at this time. All other animals were sampled at later time points, but in general, the highest concentrations occurred at the earliest time that samples were collected. Subsequently, the plasma concentrations declined over the remaining period in which the ring remained inserted but were still quantifiable in the range of 14.6 to 35.4 pg/ml just prior to ring removal after 28 days. In animals terminated 3 or 7 days after ring removal, plasma levels declined sharply in the first few hours after ring removal and were no longer detectable immediately prior to termination (Fig. 2).

FIG 2.

Plasma levels (mean ± SEM) of dapivirine in sheep following insertion of Dapivirine Vaginal Ring-004. Dapivirine Vaginal Ring-004 was placed in the vagina of female sheep, and plasma samples were collected from the jugular vein over the 28 days that the ring remained in place and for the subsequent 0 (group 1), 3 (group 2), and 7 (group 3) days following ring removal (n = 3 per group).

Vaginal fluid.

Quantifiable concentrations of dapivirine were detected in all vaginal fluid samples collected from animals with the ring in place. Maximum concentrations (7,870 ng/g of fluid) were detected as early as 8 h after ring placement in animals sampled during the first day of treatment. For all other animals, the maximum concentrations (up to 21,800 ng/g of fluid) were detected at the earliest 2 or 3 time points tested. All animals had detectable levels of dapivirine in vaginal fluid prior to ring removal, although the concentrations declined rapidly in the first few hours following ring removal. Of the animals terminated 3 days after ring removal, only one had quantifiable levels of dapivirine in the vaginal fluid at the time of termination (459 pg/g of fluid), although all three of the animals terminated 7 days after ring removal had detectable levels in vaginal fluid at the time of termination (1,340 to 6,350 pg/g of fluid) (Fig. 3).

FIG 3.

Vaginal fluid levels (mean ± SEM) of dapivirine in sheep following insertion of Dapivirine Vaginal Ring-004. Dapivirine Vaginal Ring-004 was placed in the vagina of female sheep, and vaginal fluid samples were collected over the 28 days that the ring remained in place and for the subsequent 0 (group 1), 3 (group 2), and 7 (group 3) days following ring removal (n = 3 per group).

Rectal fluid.

Dapivirine was detected at quantifiable levels in rectal fluid samples collected both while the ring was in place and also 3 and 7 days following ring removal (Fig. 4). Maximum concentrations (up to 2,350 pg/spear) were detected at 24 h after ring placement for 6 of the 9 animals. There was no clear pattern to the timing of the peak concentration in the other three animals. Three and 7 days after ring removal, the dapivirine levels in the rectal fluid samples had dropped substantially but were still quantifiable (1.56 to 29.7 pg/spear and 8.45 to 103 pg/spear, respectively).

FIG 4.

Rectal fluid levels (mean ± SEM) of dapivirine in sheep following insertion of Dapivirine Vaginal Ring-004. Dapivirine Vaginal Ring-004 was placed in the vagina of female sheep, and rectal fluid samples were collected over the 28 days that the ring remained in place and for the subsequent 0 (group 1), 3 (group 2), and 7 (group 3) days following ring removal (n = 3 per group).

Vaginal tissue.

In animals terminated immediately after ring removal on day 29, quantifiable concentrations of dapivirine were observed in all vaginal tissue samples from all animals, ranging from 8.59 to 738 ng/g of tissue (Fig. 5). The concentrations were generally highest in samples at or adjacent to the mucosal surface, declining toward the serosal surface. In animals terminated 3 and 7 days after ring removal, all vaginal tissue samples were below the lower level of quantification (LLOQ), with the exception of one sample from one animal terminated 3 days after ring removal.

FIG 5.

Vaginal tissue levels (mean ± SEM) of dapivirine after 28 days of exposure to Dapivirine Vaginal Ring-004. Dapivirine Vaginal Ring-004 was placed in the vagina of female sheep, and vaginal tissue samples were collected postmortem immediately following ring removal (group 1 animals, n = 3). The data are presented as the median and range.

Cervical tissue.

Immediately after ring removal on day 29, quantifiable concentrations of dapivirine were observed in three (of 13) cervical tissue samples from one animal only and ranged from 1.29 to 6.03 ng/g of tissue. All cervical tissue samples from other animals terminated at this time point were below the LLOQ, as were the samples for animals terminated 3 or 7 days after ring removal.

Rectal tissue.

Quantifiable concentrations of dapivirine were observed in one (of five) and three (of five) rectal tissue samples from two animals terminated immediately after ring removal and ranged from 1.08 to 5.12 ng/g of tissue. In the other animal terminated immediately after ring removal and for all animals terminated 3 or 7 days after ring removal, the rectal tissue samples were all below the LLOQ.

Residual drug levels.

Quantification of the drug remaining in the used rings from this study indicated a uniform level of drug delivery across all nine animals tested. The dapivirine level in the unused reference ring that was included in the residual drug analysis was determined to be 24.2 mg. The mean (± standard error of the mean [SEM]) ring load after 28 days of use in a sheep was 20.78 (± 0.096) mg, indicating that all rings delivered between 2.9 and 3.7 mg (13.6 and 18% of the total drug load, respectively) across the 28 days of use (Table 6). The mean drug delivery over 28 days was 4.22 mg.

TABLE 6.

Residual drug levels in vaginal rings after 28 days of use in sheep

Animal no.	Test group	Residual drug level (mg)	Total drug delivered (mg)	Estimated mean daily dose (mg/day)
1	1	21.0	4.0	0.143
2	1	20.7	4.3	0.154
3	1	21.3	3.7	0.132
4	2	20.5	4.5	0.161
5	2	20.5	4.5	0.164
6	2	20.8	4.2	0.150
7	3	21.0	4.0	0.143
8	3	20.6	4.4	0.157
9	3	20.6	4.4	0.157
Total mean drug delivery		20.78	4.22	0.151

DISCUSSION

Administration of the dapivirine vaginal ring to female sheep for 30 or 90 days produced no detectable local or systemic toxicity. These data are consistent with intravaginal toxicity studies in rabbits of up to 9 months duration using gel formulations of dapivirine in which no toxicity was seen, and they support the conclusion that the physical presence of the silicone ring does not potentiate the toxicity of dapivirine. They also suggest that the presence of the ring does not cause changes that could increase susceptibility to HIV.

The toxicokinetics data from the 90-day toxicity study demonstrated that more dapivirine was absorbed from the gel than from the ring, but systemic levels were relatively low for both formulations. The dapivirine gel contained 0.2% (wt/wt) dapivirine, and with a daily dosing volume of 2.5 ml (∼2.5 g), the total daily dose was 5 mg, whereas the dapivirine ring had a total drug load of 25 mg, and each ring was used for a period of 30 days. Furthermore, analysis of the residual dapivirine levels in the used rings from the pharmacokinetic study showed that only approximately 4 mg was released from the ring over the 28-day period of use. Therefore, based on the amount of drug released from the ring, the total dose administered over 28 days was about 35 times higher for the gel than that for the ring. Also, when comparing the C_max data, it is important to consider that the gel represents a bolus dose, whereas the ring is a sustained delivery device, and consequently, the AUC data provide a more representative comparison of the exposures achieved.

One of the specific challenges for the development of microbicides is the potential for local irritation or other toxicities at the site of application, including effects that might increase susceptibility to HIV infection. Therefore, there is a need for relevant preclinical models that are sufficiently sensitive to identify potentially harmful candidate microbicides without being overly sensitive, such that products that might be effective are eliminated because of a toxicity signal that is not clinically relevant. The standard model for vaginal irritancy testing is the rabbit, but this is likely overly sensitive and is not suitable for testing some products, such as vaginal rings, because of the dimensions of the vagina. Given the anatomical and histological similarity between the ovine and human lower female reproductive tracts, the sheep has the potential to be an important preclinical model in the development of microbicides.

The toxicity study described here adds further weight to the utility of this species in the preclinical assessment of local toxicities. The animals were amenable to the types of investigative procedures routinely incorporated in toxicity studies, and despite a number of background histopathological lesions, it was the opinion of the examining pathologist that treatment-related findings could be readily distinguished from the background findings observed. Administration of the dapivirine vaginal ring and vaginal gel formulations did not identify any toxicity, and the absence of any notable findings is consistent with observations from clinical trials involving the same ring formulation (24) and previous preclinical studies using the same gel. However, in order to determine whether this species is adequately sensitive for routine evaluation of vaginal irritation, further studies should be conducted using materials with known irritant effects.

Another key challenge for microbicides that is typically not faced by other types of pharmaceutical products is the lack of opportunity to establish the efficacious dose in a modest sized phase II dose range-finding clinical trial. Consequently, developers look to pharmacokinetic and pharmacodynamic assessments in early clinical trials and preclinical models to guide the dose level selection process. With antiretroviral drugs that target the intracellular mechanisms of the viral replication process, it is highly likely that delivery of the drug to the cellular layers of the vaginal wall is critical. However, there are many complications to accurately assessing tissue levels of a drug, such as residual drug on the surface of tissue samples (e.g., in vaginal fluid or excess gel) and/or drug held in the keratinized surface layers that do not present a viable site for viral replication. The studies described here show that the sheep can be used for determining drug concentrations at different depths within the vaginal wall in a model that is anatomically and histologically similar to humans. In addition, the actual products intended for human use can be tested without modification.

The initial pharmacokinetic feasibility study in which female sheep were dosed for 5 days with either dapivirine or tenofovir gel demonstrated that both drugs penetrated the vaginal tissue at all depths and at concentrations that were measurable using the methods employed. It was also interesting that the drug concentrations in the tissues were very similar despite the differences in the total drug administered. This would suggest better tissue penetration for dapivirine; however, the gels had different physicochemical properties and were dosed using different volumes. These differences cannot be ruled out as having influenced the drug penetration of the vaginal tissues. Additionally, physiological changes associated with the various phases of the estrous cycle were not assessed in this study but may also have influenced the pharmacokinetic profiles and contributed to the interanimal differences observed. The subsequent study using the dapivirine ring demonstrated that good penetration of dapivirine into the vaginal tissues also occurred with the ring formulation developed for human use. The mean levels in the vaginal tissue ranged from 12.4 to 474.7 ng/g of tissue. In vitro assays in TZM-bl cells using clonal HIV-1 from subtypes B and C gave 50% inhibitory concentrations (IC₅₀s) for dapivirine that were equivalent to 0.11 (subtype B) and 0.19 (subtype C) ng/ml (25). Therefore, the levels measured in vaginal tissue are 65 to 4,315 times higher than the IC₅₀s in these strains. In comparison, the levels of tenofovir achieved using the 1% gel that has shown protection in women were 0.2 to 7.6 μg/g of tissue. In the same assays described above, the IC₅₀s for tenofovir were equivalent to 1.07 (subtype B) and 1.75 (subtype C) μg/ml, which are only 0.11 to 7.10 times higher than the IC₅₀ in vitro.

It appeared that there was more dapivirine measured for the tissue specimens that were not swabbed (based on the samples below the surface of the tissue) than for those that were swabbed. While the unswabbed specimens are likely to have resulted in greater surface contamination that may have been carried down to the lower tissues, it is unclear whether this was due to mechanical contamination or continued drug migration, but it suggests that swabbing the tissue prior to sampling may be optimal.

Identification of the tissue types across the vaginal wall indicated that the thickness of the layers was highly variable, and so a precise determination of the penetration into the different tissue layers based on depth measurements was not possible. Indeed, the variation in the thickness of the various tissue layers appeared to be as much due to the variability in the vaginal wall thickness and individual sections taken within each animal as it was to between-animal variation. However, the epithelial thickness recorded in this study (73.5 to 179.9 μm) was similar to that reported elsewhere for sheep (86 to 114 μm) and generally thinner than that reported for humans (175 to 284 μm) (26–28).

Despite the similarities in the human and ovine female lower reproductive tracts, there were differences between the species in the pharmacokinetics of dapivirine. Recent clinical studies with Dapivirine Vaginal Ring-004 indicate that the drug exposure in women is higher than that in sheep, both systemically and locally. In women, the highest mean plasma C_max and AUC_0–24 values were 392 pg/ml and 8.379 ng · h/ml, respectively (14), whereas in sheep, the corresponding values were 82.6 pg/ml and 1.59 ng · h/ml. Similarly, the mean vaginal fluid C_max in women was 79.9 μg/g of fluid, but in sheep, the mean C_max was 7.72 μg/g of fluid. These values in women are approximately 5- to 10-fold higher than those observed in the sheep, although interestingly, the residual levels of drug in the used rings were very similar for women (14) and sheep, suggesting that the performance of the dapivirine ring is similar in these species.

There were some observations during the pharmacokinetic evaluations that appear to be anomalous. First, while vaginal tissue and fluid samples were shown to contain relatively high levels of dapivirine, quantifiable drug levels were detected in only 3 of 13 cervical tissue samples from 1 of 3 animals. Confirmation of the anatomical location of the cervical samples indicated that the samples included the ectocervical region closely bordering the vaginal mucosa and were not exclusively the more distal endocervical tissue. A review of the data and sampling and analytical processing methods did not identify any likely artifactual basis for the absence of drug in these samples. Consequently, the reason why dapivirine was not detected in these samples remains unclear. Second, an analysis of the rectal fluid samples from sheep administered the ring formulation demonstrated the presence of substantial levels of dapivirine. This is consistent with studies in macaques using 1% tenofovir gel, which showed rapid distribution of the drug to the rectal compartment after vaginal dosing, and vice versa (29). However, although rectal fluid levels of dapivirine on a per-sponge basis were in the range of 1.78 to 778 pg, the rectal tissue collected immediately after ring removal had quantifiable levels of dapivirine in only a minority of samples from 2 of the 3 animals tested. Since the most likely route of transfer to the rectal fluid would be through secretion from the rectal tissue, the reason for the difference between the concentrations in the fluids and tissue is also unclear.

An additional question of great interest in addressing the use of a vaginal microbicide ring product is the potential duration of action. The data presented here show that dapivirine levels generally dropped rapidly following removal of the ring. By 3 days after ring removal, there were no quantifiable drug levels in the vaginal tissue nor in the plasma, although the levels remained detectable in the vaginal fluid and, somewhat surprisingly, in the rectal fluid at 7 days after ring removal. While there was an appreciable (100-fold) decrease in drug levels in vaginal fluid from those observed with the ring in situ, the levels at 7 days after ring removal were still in the range of 1,340 to 6,350 pg/g of fluid, which is not much lower than the range in vaginal tissue with the ring in place (7,380 to 8,590 pg/g of tissue). However, based on the expectation that the tissue levels are critical for efficacy, it may be that any protection achieved with the dapivirine ring is lost within a relatively short time after removal of the ring. The clinical development of Dapivirine Vaginal Ring-004 and its anticipated mode of use once approved, however, have always been based on continuous use, with immediate replacement of the old ring with a new one after 28 days; as such, these findings do not significantly impact the intended clinical use of the dapivirine vaginal ring.

Overall, the data presented here demonstrate that the dapivirine vaginal ring is safe and that dapivirine is delivered effectively from the ring to the vaginal tissue; the data therefore support further evaluation of the dapivirine vaginal ring in clinical efficacy trials. In addition, the studies illustrate the value of the sheep as a nonclinical model for addressing the safety and pharmacokinetics of the clinical dosage form that are difficult to address in clinical trials. Additional studies to further characterize the model would help in defining the utility of this species.