Development of an in vitro alternative assay method for vaginal irritation

Abstract

The vaginal mucosa is commonly exposed to chemicals and therapeutic agents that may result in irritation and/or inflammation. In addition to acute effects, vaginal irritation and inflammation can make women more susceptible to infections such as HIV-1 and herpes simplex virus 2 (HSV-2). Hence, the vaginal irritation potential of feminine care formulations and vaginally administered therapeutic agents is a significant public health concern. Traditionally, testing of such materials has been performed using the rabbit vaginal irritation (RVI) assay. In the current study, we investigated whether the organotypic, highly differentiated EpiVaginal™ tissue could be used as a non-animal alternative to the RVI test. The EpiVaginal tissue was exposed to a single application of ingredients commonly found in feminine hygiene products and the effects on tissue viability (MTT assay), barrier disruption (measured by transepithelial electrical resistance, TEER and sodium fluorescein (NaFl) leakage), and inflammatory cytokine release (interleukin (IL)-1α, IL-1β, IL-6, and IL-8) patterns were examined. When compared to untreated controls, two irritating ingredients, nonoxynol 9 and benzalkonium chloride, reduced tissue viability to <40% and TEER to <60% while increasing NaFl leakage by 11–24% and IL-1α and IL-1β release by >100%. Four other non-irritating materials had minimal effects on these parameters. Assay reproducibility was confirmed by testing the chemicals using three different tissue production lots and by using tissues reconstructed from cells obtained from three different donors. Coefficients of variation between tissue lots reconstructed with cells obtained from the same donor or lots reconstructed with cells obtained from different donors were less than 10% and 12%, respectively. In conclusion, decreases in tissue viability and barrier function and increases in IL-1α and IL-1β release appear to be useful endpoints for preclinical screening of topically applied chemicals and formulations for their vaginal irritation potential.

Introduction

The number of alternative in vitro irritation assay models that are reproducible, less time consuming, relatively inexpensive, and more predictive of human response than animal models is increasing. For instance, reconstructed in vitro tissue culture skin models such as EpiDerm (MatTek Corporation, Ashland, MA) and EPISKIN (SkinEthic Laboratories, Nice, France) have been validated and accepted as validated tests by the European Union (EU) and Organization for Economic Cooperation and Development to identify skin irritants and corrosives (Spielmann et al., 2007). These models address a significant market need since the 7th Amendment to the Cosmetics Directive banned the testing of cosmetic ingredients or formulations in the European Union as of March 2009 (http://www.aavs.org/lawsProductsSeventh.html). While a lot of emphasis has been placed on developing alternatives to skin and ocular testing, in vitro alternatives to the rabbit vaginal irritation (RVI) test are limited and the RVI assay remains the predominant means of determining vaginal irritation potential.

The vaginal epithelium forms an uninterrupted mucosal barrier between the external micro-environment of the vaginal canal and the underlying tissues. Exposure of the vaginal-ectocervical tissue to chemical insult may cause damage and/or inflammation at the site of application. The public health risk caused by such reactions is signficant since increased infection rates for sexually transmitted infections (STI) such as HIV-1 can result (Fichorova et al., 2004, Weber et al., 2005). The increased infectivity is due to: (a) compromised tissue barrier which allows viral entry, (b) recruitment of susceptible target cells to the site of inflammation (Catalone et al., 2005) and/or (c) induction of inflammatory cytokines such as IL-1β and TNF-α which are known to activate HIV LTR via the NFkB pathway (Osborn et al., 1989).

Animal models such as the rabbit, slug, pig, and mouse have been used to screen chemicals/formulations for their irritancy and inflammation potential (Tsai et al., 2003, D’Cruz and Uckun, 2002, Achilles et al., 2002, Galen et al., 2007). Among the various models, the RVI assay is the most commonly used test and the only FDA approved animal model for vaginal irritancy testing. In the RVI assay, visual observations and the subjective parameters of erythema, edema, and discharge are used to evaluate injury and inflammation. In addition, histological cross-sections from exposed tissues are scored for epithelial exfoliation, leukocyte infiltration, thickening of the lamina propria (edema), and vascular congestion. However, the rabbit cervicovaginal tissue is markedly different from the human in terms of anatomy since it lacks the cyclic reproductive human (menstrual) stages and it is unresponsive to most human genital pathogens (Belec et al., 1995, Noguchi et al., 2003). Due to these species related differences, the RVI test is at times not predictive of human irritancy and toxicity. For instance, the RVI test failed to predict the disruptive effects of nonoxynol-9 (N9) which led to increased HIV-1 infection rates (Stafford et al., 1998, Centers for Disease Control and Prevention, 2000). In recent years, researchers have also used pig as an animal model to study inflammatory responses induced by candidate spermicides and microbicide formulations. The availability of specific probes to study inflammatory responses in pigs and the physiological and histological similarity of the genital tract of pigs with that of the human vagina were among the important factors cited to promote the use of pig as a suitable animal model for vaginal irritation studies (Squier et al., 2007). However, the pig remains an animal model and it is not suitable for high throughput screening of chemicals and formulations. Also, existing animal protocols vary substantially in relation to number of animals to be used, dose volume, and the depth of application. A comparative summary of the use of organotypic tissue and animal models for vaginal irritation testing is presented in Table 1.

Vaginal explant cultures have also been used in many laboratories to predict in vivo irritancy of vaginally applied chemicals and feminine hygiene products. However, the scarcity of normal human tissue, short survival time (due to their rapid deterioration and loss of tissue integrity when cultured), and donor-to-donor variation makes explant tissues impractical for determining the irritancy potential of topically applied chemicals or formulations. The organotypic EpiVaginal tissue model has advantages over explants tissues in that: (1) EpiVaginal can be cultured for relatively longer time period (e.g. 2–3 weeks), (2) a large number of tissues can be produced from a single donor, (3) tissue-to-tissue variability is reduced, (4) there are fewer regulatory constraints (Cummins and Doncel, 2009), and (5) the tissue can be reconstructed in the presence/absence of endocrine hormones or different cell types to mimic its in vivo counterpart. Vaginal and ectocervical monolayer cells or transformed cell lines have also been used to assess toxicity of feminine care products and spermicides. However, assays that utilize submerged monolayer epithelial cultures do not take into account the three-dimensional differentiated structure of the native vaginal epithelium and, as a result, have limited value. Other factors such as differences in gene and protein expression patterns between monolayer cultures and vaginal epithelium and limitations of testing aqueous incompatible materials such as gels, films, and vaginal rings also make monolayer cultures poor models for vaginal irritation or toxicity studies (Ayehunie et al., 2006). Further, versus animal models, the human cell-based EpiVaginal tissue is directly infectible with human pathogens such as HIV-1 (Cole et al., 2007) while animal models are infectible only with viruses that resemble HIV-1.

In this study, we used an organotypic 3D vaginal tissue model reconstructed from normal human cells to predict irritation of ingredients commonly found in vaginally administered products. Tissue viability, structure, barrier function, and cytokine release patterns were monitored to correlate in vitro results to in vivo irritation. Using these biomarkers, nonoxynol-9 (N9), which induced mild irritation in the RVI assay, and benzalkonium chloride, known to be a vaginal irritant (Fichorova et al., 2004), were correctly identified as irritants.