Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

Contents lists available at ScienceDirect

Journal of Pharmaceutical Sciences

journal homepage: www.jpharmsci.org

Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Optimization and Application of In Vitro and Ex Vivo Models for

Vaginal Semisolids Safety Evaluation
Rita Monteiro Machado 1, 2, Ana Palmeira-de-Oliveira 1, 2, Luiza Breitenfeld 1,

Martinez-de-Oliveira 1, 3, Rita Palmeira-de-Oliveira 1, 4, *
Jose
1
CICS, UBI d Health Sciences Research Center, Faculty of Health Sciences, University of Beira Interior, Avenida Infante D. Henrique, 6200-506
Covilha~, Portugal
2
Labfit, HPRD d Health Products Research and Development, Lda, Edifício UBIMEDICAL, Estrada Municipal 506, 6200-284 Covilha ~, Portugal
3
Child and Woman's Health Department, Centro Hospitalar Cova da Beira EPE, Quinta do Alvito, 6200-251 Covilha ~, Portugal
4
Center for Neuroscience and Cell Biology, University of Coimbra, Rua Larga, 3004-504 Coimbra, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: Preclinical safety assessment of vaginal products includes cytotoxicity assays upon cell lines. Further-
Received 24 August 2018 more, tissue explants have been considered for application on ex vivo models. In this study, traditional
Revised 19 February 2019 and renewed methods were studied for toxicity assessment of vaginal semisolids upon products
Accepted 21 May 2019
currently used in clinical practice as antimicrobials (Gino-Canesten®, Sertopic®, Dermofix®, Gyno-
Available online 29 May 2019
Pevaryl®, Lomexin®, Gino Travogen®, Dalacin V®), containing estrogens (Ovestin®, Blissel®, Colpo-
trophine®), and reference formulations (Replens®, Universal Placebo). Two in vitro cytotoxicity tests were
Keywords:
vaginal drug delivery performed: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and neutral red uptake
cytotoxicity upon uterine (HEC-1A), cervical (HeLa) and vaginal (VK2 E6/E7) cells, according to ISO/EN 10993-5
in vitro model (in vitro evaluation of medical devices). Similarly, a strategy to determine tissue viability on ex vivo
ex vivo vaginal model
porcine vaginal model (through MTT reduction assay and histological analysis) was developed and
optimized. The vaginal cell line VK2 E6/E7 conducted to the most accurate calculation of half-maximal
toxic concentration among all cells on the MTT assay. However, it was shown not be sensitive to the
neutral red uptake assay. Tissues from the porcine model were collected with approx. 15% variability in
thickness and variation coefficients lower than 25% when testing negative and positive controls were
achieved. These models can improve cost-efficiency in early steps of product development.
© 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

Introduction Vaginal drug delivery systems include liquid, semisolid, and

solid formulations.12 Semisolids (creams, ointments, gels) have
The vagina is a promising route of administration when local been reported as the most preferred dosage form for self-use by
and systemic drug delivery is intended.1-4 The vaginal route com- women concerning HIV prophylaxis.13-17 Concerning all these
prises diverse pharmacokinetic advantages due to its large surface particularities added to women's preference and acceptability
area, rich blood supply, avoidance of the hepatic first-pass effect, patterns for vaginal semisolid formulations,18,19 a safety- and
and relative high permeability to many drugs.5 Nevertheless, suitability-driven study is expected to conduct to a better charac-
several drawbacks, including leakage, inconsistent drug absorption, terization of the already marketed products (which are considered
influence on sexual intercourse, and local irritation, need to be safe based on clinical trials and current use) in view of the devel-
addressed during the early stages of product development.3,6 opment of new products. Furthermore, the improvement of tradi-
Ideally, vaginal drug delivery systems should not interfere with tional characterization methods by considering physiological
either vaginal physiology or women's daily life, while allowing parameters of the vaginal environment can provide predictive tools
obtaining high drug bioavailability with little variability, and low to better characterize new vaginal semisolid formulations and
incidence of side effects.7-11 optimize their acceptability and even enhance cost-efficiency at
early stages of product development.20
Over the last decades, efforts have been made to conduct
* Correspondence to: Rita Palmeira-de-Oliveira (Telephone: þ351 275 329 002). investigation to a level of high comprehension of vaginal epithe-
E-mail address: [email protected] (R. Palmeira-de-Oliveira). lium mechanisms.21 The understanding and application of several

https://doi.org/10.1016/j.xphs.2019.05.026
0022-3549/© 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
3290 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

techniques to predict drug toxicity through the vaginal epithelium to now, the cosmetic industry and pharma industry are still not
contributes to the successful selection of drugs and appropriate focused on validating standardized tissue models or cellular assays
formulations in the early stages of development.22 Thus, it is of for vaginal irritation assessment because this application still rep-
great importance to implement accurate and reproducible methods resents only a small part of business.
to predict drug toxicity. The in vitro and ex vivo approaches should Cytotoxicity assays constitute a gold standard of in vitro pre-
be privileged, and preferably an optimal in vitrodin vivo correlation clinical evaluations of chemicals in cultured cells. The MTT reduc-
should be established.23 The strategy of patient-driven product tion assay is one of the most commonly applied cytotoxicity assays.
development is also reflected in recent approaches such as the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bro-
biopharmaceutics risk assessment roadmap (BioRAM), which aims mide) is a water-soluble yellow tetrazolium salt, which is converted
to optimize drug product development and performance right from to an insoluble purple compound (formazan) due to cleavage,
the early stages.24-26 within the mitochondria, of the tetrazolium ring mainly by succi-
In the field of vaginal drug administration, recent research and nate dehydrogenase. Formazan does not permeate cell membranes,
innovation has been largely focused on microbicide formulations. and consequently accumulates in metabolic viable cells. The MTT
However, the same safety and risk assessments should be trans- assay was tested for its validity in various cell lines36 and has
posed to other vaginal product categories. In fact, preclinical already been applied to uterine,37 cervical,38 and vaginal39 derived
development, that is, nonclinical access to toxicological and phar- cells. Further modifications of the initial protocol described by
macokinetics data, could be optimized to reduce general financial Mosmann36 were proposed40,41 to improve the repeatability and
costs and overall time of product development before commer- the sensitivity of the assay. The MTT assay is the basis of currently
cialization.27-30 validated safety methods, such as the in vitro skin irritation protocol
Despite being an internal cavity, the vaginal epithelium is (OECD TG 439).
constantly exposed to potential pathogenic microbes, microflora, The neutral red uptake (NRU) assay is also a quantitative
chemicals, and hormonal changes. As minor injuries can occur either colorimetric method. It is based on the ability of viable cells to
mechanically or chemically after products usage and administration, incorporate and bind (uptake) the red dye within their lyso-
further damage or infection may be promoted.31 Commonly used somes.42-44 Although they are used extensively as convenient and
vaginal products include drug products (intended to prevent, treat, rapid measurements of cell viability, all these methods should be
or diagnose diseases through a pharmacological, immunological, or carried out with some caution. In fact, an increase in NR uptake can
metabolic mechanism), medical devices (products that prevent, be induced by lysosomal swelling agents such as weak alkaline
treat or diagnose diseases through a mechanism of action that is not substances and osmotic swelling agents. Regarding the MTT assay,
exclusively pharmacological, immunological, or metabolic), or even some reducing agents and respiratory chain inhibitors could affect
cosmetics (products intended to clean, protect, perfume, maintain in the MTT formazan formation. The MTT assay was also found to be
good conditions, or correct body odors). Vaginal pharmaceutical, significantly influenced by a number of parameters such as medium
cosmetic, and personal care products can occasionally induce un- pH, D-glucose concentration in culture media, and cellular con-
desirable local or systemic side effects. Still, the standard method for centration of pyridine nucleotides.45 In addition, the NRU assay
assessing vaginal mucosal irritation is the in vivo rabbit vaginal presents some advantages over the MTT assay. The former pro-
irritation test,22 except for cosmetics for which, in European regu- cedure is more sensitive and readily quantifiable. It is at least 2
latory frame, animal testing is not allowed.32 times cheaper, presents less interferences, and does not use an
Besides the prohibition of the marketing of cosmetic products unstable tetrazolium salt.44 As a general remark, these assays have
tested in animals, the European Commission has undertaken other some limitations in common with other cell culture procedures
regulatory decisions in pursuance of the 3Rs policy (Replacement, concerning the substance chemical characteristics, such as vola-
Reduction, Refinement) such as the approval of the REACH (Regis- tility, unstable or explosive behavior in water, low solubility,
tration, Evaluation and Authorization of Chemicals) legislation colorant nature, or chemical incompatibility between the test
(applicable to raw materials, drug products, medical devices, and substance and the chemical.44,46
even cosmetics). Together with the worldwide increasing concern The first objective of the present study was to compare these 2
with animal welfare, these legal changes have given priority for the cytotoxicity tests, MTT and NRU assay when applied to cervical,
search for validated in vitro methods so that they can replace the uterine, and vaginal cell lines. The method used was in compliance
currently used in vivo methods and be applicable to all regulatory to ISO/EN 10993-5 for in vitro evaluation of medical devices47
classes of products (cosmetics, medical devices, and drugs). In fact, (which recommends these tests upon BALB/3T3 clone A31 and
although the prohibition of using animals for product testing is only NCTC clone 929 cell lines. These cell lines are obtained from mouse
currently applied to cosmetics in the European regulatory context, muscular fibroblasts, not representing a human surrogate). These
the replacement of in vivo methods for in vitro methods represents methods were compared to infer cytotoxicity of a representative
an overall effort for all products.22,33 One of the recent examples group of vaginal semisolid medicines commercialized across the EU
supporting these changes is the validation of a protocol for in vitro and USA. Replens® (Laboratoires Majorelle, France)48-53 and Uni-
skin irritation assessment of medical devices extracts adapted from versal Placebo54-57 were used as reference products because their
that previously approved for cosmetics. This protocol is expected to toxicity profile is largely described on the literature. Second, this
be soon published as part of the ISO 10993 group of standards work aimed to develop and prevalidate a new test to determine
(Biological evaluation of medical devices). tissue viability upon an ex vivo porcine vaginal model using the
Safety concerns are currently spotted right through the early MTT reduction assay and histological analysis. This strategy of
stages of development, especially in what concerns to microbicides performing in vitro techniques on explanted tissues can circumvent
formulations. This concern was even more highlighted after the some of the problems related to the use of animals in science and
unexpected findings of nonoxynol-9 (N-9) in clinical trials,34 industry and also can better help the prediction of in vivo effects.
stressing the need for more appropriate in vitro assays to predict Finally, it is our objective to find correlations between the 2
in vivo safety issues. methodologies in study (in vitro and ex vivo) and to propose an ideal
The cosmetic industry has already adopted validated alternative test combination for preclinical assessment of vaginal semisolid
in vitro methods for outcomes such as eye irritation, phototoxicity, products. A schematic representation of the experimental work is
skin irritation, skin corrosion and skin sensitisation.35 However, up detailed in Figure 1.
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3291

Figure 1. Schematic representation of the experimental work.

Materials and Methods Belgium), hydroxyethylcellulose (HEC; Natrosol 250 HX, Ashland
Inc.), sorbic acid (Sigma-Aldrich, Germany), and sodium chloride
Tested Formulations (Merck, Germany). All used chemicals and reagents were of analytic
grade or equivalent.
Ten different semisolid vaginal medicinal products that are
commercially available in Europe and the United States were Epithelial Cells
included in this study: Gino-Canesten® (Bayer, Portugal), Sertopic®
(Ferrer, Portugal), Dermofix® (Azevedos Laboratories, Portugal), The cell lines HEC-1A, HeLa, and VK2 E6/E7 were obtained from
Gyno-Pevaryl® (Johnson & Johnson, Portugal), Lomexin® (Jaba the American Type Culture Collection (ATCC-LGC Promochem,
Recordati, Portugal), Gino Travogen® (Bayer, Portugal), Dalacin V® Teddington, United Kingdom). The uterine HEC-1A cells, originated
(Pfizer Laboratories, Portugal), Ovestin® (Aspen Pharma, Portugal), from a line of human endometrial adenocarcinoma,58 were cultured
Blissel® (ITF Medivida, Portugal), Colpotrophine® (Teva Pharma, in RPMI 1640 supplemented with 100 U/mL penicillin, 100 mg/mL
Portugal). Replens® (Laboratoires Majorelle, France),48-53 and Uni- streptomycin, and 10% FBS, further referred to as RPMI complete
versal Placebo54-57 were used as reference products because their medium (Passages 33-38). HeLa cell line is also epithelial, derived
toxicity profile is largely described on the literature. Universal Pla- from human cervical adenocarcinoma.59 These cells were cultured
cebo was prepared according to Tien et al.48 by dissolving of 2.7 g of in DMEM F12 medium supplemented with penicillin (100 U/mL),
hydroxyethylcellulose (2000cP) in 96.3 g of water containing 0.85 g streptomycin (100 mg/mL), and 10% FBS and is further referred to as
of sodium chloride and 0.1 g of sorbic acid. The final pH was adjusted DMEM complete medium (passages 64-68). VK2 E6/E7 cell line is
to 4.4 by adding sodium hydroxide, and the gel was stored at 2 C-8 C. also an epithelial one, derived from human vagina (HPV-16 E6/E7
General characteristics of the studied products are listed in Table 1. All transformed).39 VK2 E6/E7 cells were cultured in keratinocyteese-
products, except placebo and Replens®, are classified as medicines, rum-free medium (keratinocyte-SFM) added of 0.1 ng/mL human
while comprising an active pharmaceutical ingredient (API). Never- recombinant epithelial growth factor, 0.05 mg/mL bovine pituitary
theless, some do not require a medical prescription to be dispensed at extract, and calcium chloride 44.1 mg/L (passages 4-6).
pharmacies, which is the case of Gino-Canesten® and Gyno-Pevaryl®.
Vaginal Tissues Explants
Materials
Complete porcine genitalia, from approximately 6-month-old
Reagents used include Roswell Park Memorial Institute-1640 animals, were collected from a local slaughterhouse, transported
RPMI (RPMI-1640, Sigma, Germany), Dulbecco's Modified Eagle under refrigeration to the laboratory and processed within 3 h of
Medium F12 (DMEM F12, Gibco), keratinocyteeserum-free me- animal death. The vaginal tubes were separated from the sur-
dium (Gibco), human recombinant epithelial growth factor (Gibco), rounding organs using scissors and tweezers. The vagina was cut
bovine pituitary extract (Gibco), 3-(4,5-dimethylthiazol-2-yl)-2,5- longitudinally and the vestibule (caudal vagina) excised. The ves-
diphenyltetrazolium bromide (MTT, Amresco), fetal bovine serum tibule was selected since we found (unpublished data) to be the
(FBS, Merck, Germany), penicillin and streptomycin (SP, Sigma, proximal vagina histologically more similar to the human one.
Germany), phosphate buffer solution (PBS, Sigma, Germany), Tissues were then washed on a prewarmed saline solution (NaCl
dimethyl sulfoxide (DMSO, Fisher Chemical, United Kingdom), 0.9%, 37 C). This section of tissue was then thickness normalized by
Triton X100 (Fisher Chemical, United Kingdom), sodium dodecyl using a manual dermatome (Watson Skin Graft Knife, BBraun,
sulfate (SDS, Acros Organics, Belgium), Nonoxynol-9/Tergitol (N-9, Germany) equipped with Aesculap blades (BA718Rm, BBraun,
Sigma, Germany), neutral red (NR, Acros Organics, Belgium), Germany). Then, epithelial sheets were placed upon aluminum foil
Ethanol (Manuel Vieira e Ca, Portugal), glacial acetic acid (ChemLab, (contact made by the basolateral side) and punched to sections of
3292 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

Table 1
General Characteristics of the Formulations Included in the Study (Information Provided by the Manufacturer)

Excipients Product Dosage Form API Excipients

Antimicrobials Gino-Canesten® vaginal cream Clotrimazole 10 mg/g Benzyl alcohol, cetyl palmitate, cetostearyl alcohol, purified
water, polysorbate 60, sorbitan monostearate,
octyldodecanol
Sertopic® vaginal cream Sertaconazole 2 g/100g Ethylene glycol and polyethylene glycol, stearate palmite,
saturated glycerides polyglycolized, glycerol isostearate,
liquid paraffin, methylparaben, sorbic acid
Dermofix® vaginal cream Sertaconazole 20 mg/g Tefose 63, Labrafil, Peceol, liquid paraffin, Nipagin, sorbic acid
Gyno-pevaryl® vaginal cream Econazole 10 mg/g Stearate pegoxol 7, liquid paraffin, oleic macrogolglycerides,
butylhydroxyanisole (E320), benzoic acid, purified water
Lomexin® vaginal cream Fenticonazole 20 mg/g Propylene glycol, hydrogenated lanolin, sweet almond oil,
polyglycol esters of fatty acids, cetyl alcohol, glyceryl
monostearate, sodium EDTA, purified water
Gino Travogen® vaginal cream Isoconazole 10 mg/g Polysorbate 60, sorbitan stearate, cetostearyl alcohol, thick
paraffin, white vaseline, purified water
Dalacin V® vaginal cream Clindamycin 20 mg/g Propylene glycol, cetostearyl alcohol, liquid paraffin, sorbitan
stearate, benzyl alcohol, cetyl palmitate, stearic acid,
polysorbate 60, purified water
Estrogens Ovestin® vaginal cream Estriol 1 mg/g Octyldodecanol, glycerol, cetyl alcohol
Stearyl alcohol, polysorbate 60, sorbitan stearate, lactic acid,
chlorhexidine hydrochloride, sodium hydroxide, purified
water, synthetic spermaceti
Blissel® vaginal gel Estriol 50 mg/g Glicerol (E 422), p-hydroxybenzoic acid methyl ester, p-
hydroxybenzoic acid propyl ester, polycarbophil, carbomer,
sodium hydroxide, hydrochloric acid, purified water
Colpotrophine® vaginal cream Promestriene 1 g/100 g p-hydroxybenzoic acid methyl ester, p-hydroxybenzoic acid
propyl ester, mono and diglycerides of saturated fatty acids,
polyglycol ether of saturated fatty alcohols, oleic acid decyl
ester, triglycerides of capric and caprylic acids, glycerol,
purified water
Reference products Universal placebo vaginal gel Not applicable Purified water, hydroxyethylcellulose, sodium chloride, sorbic
acid, caramel color, sodium hydroxide
Replens® Vaginal gel Not applicable Purified water, Polycarbophil, paraffin oil, glycerin, palm oil
hydrogenated, carbomer homopolymer type B, sorbic acid,
sodium hydroxide

ø ¼ 3 mm using a biopsy punch (Stiefel, GSK, United Kingdom). were added (see Test products preparation section). After this
These sections were maintained on a warmed saline solution until period, cells were washed with PBS and incubated for 4 h with a 0.5
use in the MTT assay. Tissues thickness was assessed to maintain mg/mL solution of MTT reagent prepared in the respective culture
homogeneity between samples. For this purpose, a digital micro- medium without supplementation. Subsequently to formazan
meter (Vogel, Germany) was used by placing the tissue between 2 crystals formation, extraction was accomplished with 200 mL of
microscopy slides, measuring total thickness (mm) and then sub- DMSO for 15 min, through mild agitation on an orbital shaker,
tracting the thickness of the slides to the total measured thickness. protected from light. Absorbances were then measured at 590 nm
and 630 nm for background deduction, using a microplate spectro-
Test Products Preparation photometer (BIORAD xMark). In addition, cells were observed and
photographed before and after the application of the MTT reagent
Test products were diluted to 20%, 5%, 1%, 0.4%, 0.2%, 0.1% (w/v), solution to evidence the formation of the formazan crystals, using an
in medium without supplementation but containing 0.5% (v/v) of inverted microscope (Olympus IX51, Japan, equipped with OCTAX
DMSO to ensure proper solubility of formulations. DMSO was Eyeware v.1.5 Build 406, Germany)ddata not shown. Triton X-100,
selected based on the recommendations of the ISO-10993.47 To SDS, and N-9 were used as positive controls because they are widely
assure that this substance was not cytotoxic by itself, further con- known inducers of cytotoxic effects, with N-9 being a toxic vaginal
trols were performed with the used concentration in culture me- compound, as assessed in vivo.31 The negative control consists of
dium. A negative control was included in all experiments (cells/ cells without any treatment (just culture media along the assay),
tissues only with media/DMSO) and also positive controls for which was considered as the 100% viability reference for products
toxicity were included (SDS 5% (w/v), Triton X-100 1% (v/v), and toxicity calculation. Furthermore, the concentration that was toxic to
Nonoxynol-9 (N-9) 2% (v/v)). Formulations were left in contact with half of the culture (cell/tissue), the half-maximal toxic concentration
the testing system (cell or tissue) for 24 h. (TC50), was calculated by logistic regression using GraphPad Prism
Version 6.0 (Copyright, 2015).
Cellular Toxicity
NRU Assay
MTT Assay The NRU assay was performed according to the
The MTT reduction assay was performed as previously literature44,46,62,63 and ISO/EN 10993-5 for in vitro evaluation of
described40,46,60,61 and according to ISO/EN 10993-5 for in vitro medical devices.47 Cells were seeded into 96-well plates (100,000
evaluation of medical devices.47 Cells were seeded onto 96-well cells/mL) and maintained in culture until a semiconfluent mono-
plates (100,000 cells/mL) with the respective incomplete culture layer was achieved (37 C, 5% CO2, 24 h). They were then exposed to
media. Cells were let to adhere for 24 h at 37 C, 5% CO2. After the test compounds (100 mL) over the range of concentrations
obtaining a half-confluent culture, 100 mL of the test formulations described previously. After 24 h of exposure, wells were washed
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3293

once with 150 mL of prewarmed PBS, then 100 mL of NR medium was dehydration before embedding in paraffin. Tissue blocks were cut
added to each well and the plates were incubated at 37 C in a to have about 3 mm thickness and stained with hematoxylin and
humidified atmosphere of 5% CO2 for 3 h. After incubation, the NR eosin. Slides were observed and microphotographs were taken
medium was removed, and cells were washed with 150 mL of PBS. using a Zeiss microscope (Axio Imager A1, Zeiss, Germany) equip-
Finally, 150 mL NR desorb solution (ETOH/acetic acid) was added to ped with a digital camera (Axiocam, Zeiss, Germany).
all wells, including blanks, to extract the dye. Plates were rapidly
shaken on a microtiter plate shaker for 10 min to allow NR to be Data Processing and Statistical Analysis
extracted from the cells forming a homogeneous solution. The
absorbance of the resulting colored solution at 540 nm was Data were analyzed to produce arithmetic means with standard
measured in a microtiter plate reader (BIORAD xMark). NRU was deviations using Microsoft Excel. Analysis of variance and Dun-
determined for each concentration and compared to the one nett's multiple comparisons test were performed to determine the
determined in control cultures. For the study, an NR stock solution significance of the difference between sets of data (p < 0.05). Lo-
was previously prepared by dissolving 0.4 g of the dye in 100 mL of gistic regression analysis was performed using GraphPad Prism
milliQ water. The NR medium was prepared in the day prior of software to calculate the half-maximal toxic concentration (TC50) in
usage by adding 1 mL of NR stock solution to 79 mL of culture toxicity assays (Sigmoidal dose-response, best fit values with 95%
media, followed by incubation overnight at 37 C and filtration confidence interval).
through a 0.2 mm filter before adding to the cells, to be free of
crystals. The NR desorb/extraction solution is composed of 1% Results
glacial acetic acid solution, 50% ethanol, and 49% water.
Cellular Toxicity
Tissue Toxicity
Cytotoxicity was assessed by means of the MTT reduction assay
Method Optimization and NRU performed upon 3 different cell lines (HeLa, HEC-1A, and
Tissue explants were first tested for homogeneity and response VK2 E6/E7). Results are presented as the mean of 2 independent
to MTT assay. Therefore, tissues were submitted to a preliminary experiments. All assays were tracked microscopically for cell
experiment, using different animals and applying the viability test integrity, density, and morphology, as stated in the respective
on fresh and frozen tissues. Frozen explants were obtained as protocol.47 Cell cultures acquired half-confluency and were able to
described (see section “Vaginal tissue explants”) but tissue prepa- reduce MTT in the mitochondria and to integrate NR in the lyso-
ration occurred after defrosting; upon reception from the slaugh- somes. Results are expressed as viability percentage of the negative
terhouse, the vaginas were opened longitudinally (from vulva to control (cells treated only with media). For all experiments, positive
cervix), washed in an HBSS solution at pH 4.2, wrapped in controls were used to assure the occurrence of induced toxic effects
aluminum foil and stored in airtight bags to be frozen at 20 C. For and were in accordance with those expected toxic results. For NRU
viability studies performed to establish the method (fresh tissue: assay, cell viabilities obtained for HeLa and HEC-1A after contact
46 animals, at least n ¼ 2 for each animal, 6 independent experi- with the positive controls (Triton X100 1%, SDS 5%, and N-9 2%)
ments; frozen tissue: 12 animals, n ¼ 3, 1 experiment), tissues were were as follows: 2.75 ± 0.26%; 3.62 ± 0.278%; 3.79 ± 0.30%; 17.23 ±
left in culture for 24 h and then the MTT assay was performed. For 0.91%; 29.75 ± 9.85%; 19.10 ± 1.82%, respectively. Standard de-
toxicity studies (only performed on fresh tissue: 23 animals, at least viations for negative controls (cells with media) were 10.93% and
n ¼ 2 for each animal, 5 independent experiments), tissues were 16.30% for HeLa and HEC-1A, respectively. The results for NRU assay
put in contact with SDS 5% (w/v) during 24 h, and then the MTT using the VK2 E6/E7 are not showed because they were inconclu-
assay was performed. SDS 5% was selected as a positive control for sive. Regarding the MTT assay, cell viabilities for HeLa, HEC-1A, and
this test because it is generally used as a positive control in tissue VK2 E6/E7 in contact with positive controls (Triton X100 1%, SDS
toxicity assays and recognized to have a toxic effect even on 5%, and N-9 2%) were as follows: 2.75 ± 0.26%; 3.62 ± 0.28%; 3.79 ±
epithelial vaginal tissue.37 0.30%; 5.82 ± 0.34%; 5.86 ± 0.32%; 6.54 ± 0.50%, 26.52 ± 1.45%;
13.35 ± 3.93%; and 22.12 ± 2.17%, respectively. Standard deviations
MTT Assay for negative controls were 10.93%, 11.51%, and 14.63% for HeLa and
Tissue explants were placed in 96-well flat-bottomed tissue HEC-1A and VK2 E6/E7, respectively.
culture plates and treated with the test formulations diluted on In Figure 2, all profiles for cytotoxicity are represented both for
RPMI media without supplementation at 37 C, 5% CO2 during 24 h, the NRU and the MTT assays. As a general picture, tests performed
n ¼ 6. In addition, negative controls were included for each animal. upon HEC-1A and VK2 E6/E7 cell lines conducted to higher via-
SDS 5% (w/v), Triton X-100 1% (v/v), and N-9 2% (v/v) were used as bilities than those performed on HeLa cells, being higher in the NRU
positive controls. Subsequently, tissues were washed twice with assay. Antimicrobials showed a linear decrease of viabilities with
prewarmed PBS. MTT solution at a final concentration of 0.5 mg/mL increasing concentration of products, as expected. Furthermore,
(freshly prepared in culture medium) was added to each well and this behavior occurred consistently on the 3 cell lines and within
incubated 1 h at 37 C, 5% CO2. Tissues were then transferred into the 2 assays, exception made for antimicrobials studied in HEC-1A
new plates and extracted with 200 mL of isopropanol during 1.5 h at using the NRU assay, in which viabilities do not seem to be affected
room temperature under gentle shaking. Finally, tissues were by the dilution factor.
discarded for plate reading at 590 nm against the background at The estrogen-containing formulations, Ovestin® and Colpo-
630 nm (BIORAD xMark). The tissue viability was calculated as trophine®, presented the higher extents of cell toxicity, indepen-
percentage from the negative control (tissue with no formulation). dently of the tested concentrations (just a slight increase was
observed in the 0.1% concentration). Profiles for the same cell line
Histological Analysis were concordant between the assays. Blissel® presented an odd
One tissue per formulation, at its highest concentration, 20% profile, having a high increase on viability at 5%, which is reduced at
(w/v), was reserved for histological analysis. Tissues were fixed in a intermediate concentrations, and then at low concentrations con-
balanced 10% formalin solution. Subsequently, fixed tissues were ducts again to high cell viabilities. This behavior might be related
run in a set of ethanol solutions of increasing concentrations for with the Blissel® polymeric composition (polycarbophil/carbomer)
3294 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

Figure 2. Cellular viability profiles for all tested formulations at dilutions ranging from 0.1% to 20% (w/v). Products were divided according to their therapeutic activity (from left to
right: antimicrobial products, estrogens, and reference or placebo products). (a) Represents results for the NRU assay. (b) Represents results for the MTT assay. Cell viability is
represented as percentage of the control treated only with culture media. Results are presented as the mean and bars represent standard deviations from 2 independent exper-
iments in which each condition was tested in triplicate (total n ¼ 6 wells). To facilitate the interpretation of the figure, * denotes NO statistical difference from the control (2-way
ANOVA, Dunnett's multiple comparisons test, p < 0.05).

which in the lower concentrations promotes lower toxicity and a high pH-buffering capacity and high osmolality could explain the
then promotes somehow a negative effect on cell proliferation at high level of cytotoxicity presented.20
intermediate concentrations (actually this dilution could promote a Regarding VK2 E6/E7 cells, no studies were found available on
better diffusion of estriol). Concerning the reference products, low the literature presenting results in the NRU assay, despite these
cytotoxic profiles were observed, as expected. Universal Placebo gel cells being largely used to study vaginal toxicity mechanisms
led to most stable profile of viabilities across concentrations. All through other methods. In fact, we found irregular results with
concentrations conducted to viabilities over 50% of cell viability. On these cells on the NRU method (n ¼ 3, 2 independent experiments),
the other hand, Replens® only showed to reach high viabilities at and this could be due to an increased NR uptake induced by lyso-
concentrations lower than 5%. This fact may be related with the low somal swelling agents such as weakly basic substances and by os-
pH presented by this formulation, around 2.8, which together with motic swelling agents such as polyols, as was demonstrated in
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3295

Table 2
Optimization and Prevalidation Studies on Tissues Viability and Toxicity

Viability Studies (Abs at 570 nm) Culture Media Toxicity Studies (Abs at 570 nm) SDS 5%

Fresh Tissue (Mean ± SD) Frozen (Mean ± SD) Fresh Tissue (Mean ± SD)

0.5788 ± 0.1412 0.1015 ± 0.0144a 0.0581 ± 0.0133a

Variation coefficient: 24.39% Variation coefficient: 14.20% Variation coefficient: 22.83%
46 animals n ¼ 2-3 12 animals n ¼ 2 23 animals n ¼ 2-3
6 independent experiments 1 experiment 5 independent experiments

The viability study corresponds to a negative control (tissues plus culture media) and the toxicity study was performed using SDS 5%, recognized to have a toxic effect on
epithelial vaginal tissue.
a
Denotes significance on one-way ANOVA Dunnett's multiple comparisons test (p < 0.05).

previous studies.64,65 Taking this into account, the lysosomal Blissel®, Colpotrophine®) and reference products (Replens and
swelling may lead to an underestimation of the cytotoxicity when Universal Placebo) induced only minimal changes to the tissues. In
the NR assay is used.44,45 addition, Dalacin V®, an antibacterial product, was able to preserve
epithelial layers' integrity. On the other hand, antifungals showed
Tissue Toxicity extensive desquamation effects, particularly evident for Gino Can-
esten®. As expected, positive controls (Triton, SDS, and N-9) con-
Method Optimization ducted to extensive damage on the epithelial layers. Damage caused
Optimization experiments were performed to assure that after by Triton and N-9 was not as extensive as for SDS, which completely
collection and cutting procedures, tissues would have enough destroyed the epithelial and basal layers of the explant.
viability to be used in a toxicity assay such as the MTT reduction
assay. In addition, this step resulted in better handling of the sur-
Models Comparison
gical instruments and consequently more precise cutting technique
for the operator. Results for viability and toxicity studies are shown
Calculated half-maximal toxic concentrations (TC50) either for
in Table 2. Tissue viability was performed on either fresh or frozen
cellular models and tissue models are represented in Table 3. The
tissue. As expected, frozen tissues returned low absorbances, which
ex vivo porcine vaginal tissue model allowed for the calculation of
are not adequate to a final MTT assay, because this would have been
the TC50 for almost all products. This means, on the one hand that
the maximum absorbance to obtain and could not distinguish
the formulations concentration in test is suitable to the test
significant differences between formulations. For this reason, per-
method, and on the other hand that this method is less sensitive in
forming such toxicity studies in frozen-thawed tissues is not
comparison with the cellular model. Nevertheless, for Gino-Can-
possible. Furthermore, the performance of a toxic substance (SDS)
esten® and Gino Travogen®, the TC50 confidence interval was very
confirmed that chemicals can induce negative responses in ex vivo
wide, which could be related with the high slopes and standard
tissues. One-way analysis of variance with Dunnett's multiple
deviations that are shown on the toxicity profiles for these products
comparisons test was applied to assess differences between the
between concentrations 5% and 20%. These 3 cell lines were chosen
fresh tissue, frozen tissue, and tissue treated with SDS (p < 0.05 was
because they were representative of 3 different epithelia that are
accepted as denoting significance). On the fresh tissue experiment,
present in the vaginal cavity, so they would allow a complete
it was clear that high absorbances can be reached using an ex vivo
toxicological profile of the formulation after administration by the
model, that is, suitable for toxicity testing of formulations.
vaginal route. Concerning the MTT assay, the most sensitive cell line
Moreover, an acceptable variation coefficient can be held with the
was the HeLa (cervical cells), while VK2 E6/E7 (vaginal cells) was
proposed tissue preparation technique and culture procedure
found to be more robust. Because it is expected that cellular models
(24.39%; see Table 2). After measuring tissues thickness (6 animals,
are more sensitive to toxic effects than tissue models (on account of
n ¼ 6, 3 independent experiments), the variation coefficient (based
the more complex and organized structure of the latter) and that
on the mean and standard deviation) was calculated to be 15.12%.
the tissue, in turn, is more related to the in vivo environment, VK2
E6/E7 may provide more reliable results of the vaginal toxicity of
MTT Assay
formulations when performing the MTT assay. Neutral red uptake
Tissue viability results after exposure to the test products are
assay should be performed to confirm previous results from the
shown in Figure 3. Only Universal Placebo and Ovestin® were found
MTT assay, gathering the maximum data for preclinical safety
to conduct to viabilities >50% by the first dilution. All other for-
characterization of formulations, or to circumvent problems due to
mulations, except Dalacin V®, returned viabilities above 50% upon
chemical incompatibility. When possible, final formulations that
the second dilution (1:20). Standard deviations were relatively high
were selected to proceed for further preclinical and clinical stages
but acceptable considering that the surrogate used for this test is
should be tested under an ex vivo model, being the technique
biological. Positive controls, Triton X100 1%, SDS 5%, and N-9 2%,
herein presented a valuable alternative test method.
conducted to viabilities of 3.22 ± 0.77%, 2.51 ± 0.81%, and 2.56 ±
0.60%, respectively, confirming that they have toxic effects on the
vaginal epithelium (metabolic toxicity). Discussion

Histological Analysis The vaginal epithelium is a common route for topical adminis-
tration of antimicrobials, estrogens, lubricants, and hygiene
Histological analysis performed on tissues after exposure to for- feminine-care products.2 Nonetheless, these products could
mulations was found to be a useful complement to tissue viability potentially conduct to acute irritation of the epithelial surfaces. For
assessment. Representative images of explants histology after example, viral and bacterial infection potentiation subsequent to
exposure to the test formulations are presented in Figure 4. All for- the use of some vaginal lubricants has been demonstrated, rein-
mulations seem to have induced epithelial alterations of the tissue forcing the need of extensive safety characterizations both for
when compared to the control. Nevertheless, estrogenic (Ovestin®, medicinal and cosmetic or hygiene products.31,34,50,66-69
3296 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

Figure 3. Tissue viability profiles for each tested formulation at dilutions ranging from 0.1% to 20% (w/v). Viabilities are presented as percentage of the control tissue treated only
with culture media. Results are presented as the mean and bars represent standard deviations from 2 experiments in which each condition was tested in triplicate (total n ¼ 6
tissues). To facilitate the interpretation of the figure, * represents NO statistical difference from the control (2-way ANOVA, Dunnett's multiple comparisons test, p < 0.05).

The standard preclinical test for the assessment of vaginal irrita- been stated which model better predicts in vivo toxicity.71 On the
tion and toxicity is the rabbit vaginal irritation model. Similar to many one hand, ex vivo assays are closer to fully represent the in vivo
other in vivo models, the reproducibility of results from this model mechanisms. On the other hand, in vitro techniques are easier to
(either between animals or between tests) is suboptimal. On account handle, quicker, more sensitive, and able for high-throughput
of the high level of variability, standard rabbit vaginal irritation pro- screening in early steps of discovery and development. During
tocols use 10 rabbits per test article. The large number of animals product development, it has been observed that even slight mod-
required is undesirable from an animal welfare point of view and the ifications in final formulations can either modify the efficacy or
variability decrease the confidence in the assay results. These prob- toxicity of a vaginal preparation.72,73
lems are even magnified when primates are used for testing.70 In this work, the toxicity profiles of final formulations belonging
Several in vitro and ex vivo models have been used to study the to different therapeutic classes (antimicrobials, estrogens, and
toxicity of substances and products, although it has not already reference products) were assessed. The formulations included in
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3297

Figure 4. Impact of the tested formulations on the porcine ex vivo vaginal epithelium after 24 h of exposure. Histological images are representative of the higher concentration
tested (20%), that is, the worst-case scenario for the dilutions tested for tissue toxicity. Hematoxylin and eosin staining. Magnification 100.

this study are already commercialized (approved for human use) so 2 test models. This profile is thought to be due to the presence of
their main purpose of this work was to propose a rational tool that estriol. It is actually consistent to the general profile encountered
could rapidly disclose final products toxicity before getting to ani- for estrogens. In fact, Blissel® is the estrogenic formulation with the
mal studies or even replacing the currently available animal lower concentration of estriol (50 mg/g vs. 1 mg/g in Ovestin® and
studies. Because these formulations are drug products, a risk/ Colpotrophine®), and this difference may explain why Ovestin® and
benefit assessment underlies its marketing so despite being rather Colpotrophine® present cytotoxic profiles among all formulation
safe, they can present some sort of in vitro toxicity. This actually dilutions in the cellular model (even at 0.1%, the estriol content is
means that relatively negative results in in vitro stages of safety probably sufficiently high to saturate all estrogenic receptors con-
testing do not necessary completely eliminate a product from the ducting to oxidative stress).74 Blissel® showed an estrogen-driven
pipeline of development. Some of the results herein presented are behavior, conducting high cellular viabilities in low concentra-
not fully explained because the authors did not have access to the tions and low viabilities at high concentrations. Moreover, it was
total quantitative information of the formulations. This is the case clear that the cellular models were affected by dilution effects
for Blissel® that presented an odd profile, although consistent in the (inconsistencies after certain dilutions), which did not represent a

Table 3
Half-Maximal Toxic Concentrations (%-TC50 and 95% CI) Calculated for Cellular and Tissue Models Using 3 Different Cell Lines (HeLa, HEC-1A, and VK2 E6/E7) and Porcine
Vaginal Explants

Product Cellular Toxicity TC50 (%) Tissue Toxicity TC50 (%)

MTT Assay NRU Assay MTT Assay

HeLa HEC-1A VK2 E6/E7 HeLa HEC-1A Porcine Vaginal Explants

Gino-Canesten® <0.1 <0.1 7.039 (3.825-12.950) <0.1 <0.1 18.530 (2.436-141.000)

Sertopic® <0.1 <0.1 1.118 (0.550-2.272) <0.1 <0.1 1.607 (0.568-4.549)
Dermofix® <0.1 <0.1 5.243 (2.511-10.950) <0.1 <0.1 >20
Gyno-Pevaryl® <0.1 0.196 (0.114-0.336) <0.1 <0.1 <0.1 >20
Lomexin® <0.1 0.232 (0.142-0.381) 0.296 (0.198-0.441) <0.1 <0.1 >20
Gino Travogen® <0.1 0.141 (0.114-0.173) <0.1 <0.1 <0.1 11.950 (0.949-150.400)
Dalacin V® <0.1 0.974 (0.420-2.256) 0.386 (0.302-0.495) <0.1 <0.1 1.233 (0.628-2.422)
Ovestin® <0.1 <0.1 2.163 (1.206-3.882) <0.1 <0.1 <0.1
Blissel® 0.579 (0.288-1.167) 6.249 (2.953-13.220) 8.008 (3.293-19.480) 0.825 (0.436-1.561) <0.1 3.030 (2.000-4.591)
Colpotrophine® <0.1 <0.1 <0.1 <0.1 <0.1 0.911 (0.591-1.405)
Universal placebo >20 >20 >20 >20 2.982 (0.722-12.320) >20
Replens® 5.186 (2.203-12.210) 10.000 (8.877-11.260) 4.847 (3.270-7.186) 5.731 (3.851-8.529) >20 >20

Within the cellular models, MTT and NRU assays were issued, as for tissues, only MTT assay was performed.
3298 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

methodological issue in our ex vivo model. These findings along are composed of an oily phase and an aqueous phase), it was
with the difficulties in attributing the toxic actions to a single difficult to solubilize and wash the cells during the tests. Actually,
substance (because all formulations correspond to mixtures of we have circumvented this problem by using 0.5% (v/v) of DMSO in
drugs and excipients) let us state that this screening of final the final formulation dilution with culture media to prevent phase
formulation should be proceeded by individualized excipient and separation of these type of products in test. Furthermore, with
active substances responses. A study conducted by Gali et al. (2010) handling practice and constant microscopy monitoring, we were
discussed the toxic effects of different classes of excipients for the able to confirm that cell detachment after washing steps did not
vaginal route and the results obtained offer a useful guidance to occur. Still, it was also observed that washing steps were even
select the most promising excipients regarding their toxicity easier when performing the tests on the ex vivo model.
in vitro. A criterion for selection appointed by the authors is that the In 2006, MatTek™ Corporation presented an in vitro vaginal
concentration of a specific excipient should be below the CC50.37 human tissue model with low variation coefficients intrabatch and
Transposing that assumption to this work, we could estimate the interbatch (<10 and <15%, respectively) as an innovative and
product concentration on the vaginal fluid at each moment, improved toxicological test system70 that could be applied to
0.75 mL,75 and adjust the recommended doses to minimize toxicity, vaginal drug delivery assays, bacterial adhesion, and omics.79 Epi-
although warranting the therapeutic effect. Similarly, results in the Vaginal™ is available as an epithelial tissue (grown from normal
study by Gali et al. point to interesting correlations between the human vaginal epithelial cells) and as a full thickness tissue
data obtained in the various assays. This implies that the different (including epithelial cells and a fibroblast-containing lamina
assays do not generate independent data, although their sensitiv- propria). These 2 tissues could also include immune-competent
ities can differ. Therefore, to begin with screening, the simplest dendritic cells for inflammatory studies. Another company
assay should be preferred and all excipients and APIs should be providing this type of tools in Europe is EpiSkin™: HVE (recon-
checked for biocompatibility. Thereafter, the most promising for- structed human vaginal epithelium) is composed of A431 cells
mulations could then be confirmed in a more relevant model, such (derived from a vulvar epidermoid carcinoma). These tools are
as the ex vivo model, and after that an in vivo model, if relevant and clearly advantageous when in comparison with other in vitro and
necessary. ex vivo models, not only in terms of sensitivity and reproducibility,
It is well established that data obtained using in vitro models but also in terms of technology, quality control, and even technical
during the preclinical stages of development will not match assistance. Nevertheless, they have main applicability to industry
entirely the outcomes of clinical trials. This happens because, first, screening platforms, rather than academic research, because of the
the settings are different on the 2 stages (API concentration and economical expenditure. That is a key point why ex vivo tissues
length of exposure and, of course, the biological environment), and, could represent valuable tools for scientific research, when early
second, the parameters evaluated in clinical trials are wider stages of the preclinical development of drugs and products are the
(including, e.g., histopathological evaluation, assessment of the focus. Moreover, the ex vivo models take into consideration an
vaginal inflammatory condition, pH, and microflora appreciation). interindividual variability that is not present in manufactured
Consequently, although there is no certainty about the in vivo reconstructed models and that in fact is closer to the variability
standard for safety assessment, potential safety issues detected on encountered further on in vivo studies.
in vitro and ex vivo assays should be seriously considered. Several Despite having a full cell structure (epithelial, connective, im-
international research groups and organizations have reunited ef- mune) and better tolerance to formulations, tissue explants also
forts to identify vaginal biomarkers, including cytokines and che- have some drawbacks like variability, and the fact that they are
mokines, in an attempt to correlate in vitro and in vivo safety testing technically more demanding. In addition, ex vivo tissue is not
more properly. Fichorova et al.,76 in 2001, have established a cor- entirely representative of the in vivo situation due to lack of tissue
relation between mucosal toxicity and increased levels of the regeneration, lack of immune cells recruitment, and independence
proinflammatory chemokine IL-8 in vaginal washings of from hormones.69,77 If human explants are used, they can also be of
spermicide-treated rabbits. Thus, quantification of IL-8 might be limited number and require an institutional review board
used as a sensitive analyte to complement in vitro toxicity testing. approval.80
The strategy of testing semisolids formulations (i.e., final for- In this study, a good overlapping profile could be found between
mulations) toxicity using both in vitro and ex vivo models has been the cellular and the tissue model, although cellular testing was
previously explored by other groups. For example, Rohan et al.77 more sensitive conducting to lower viabilities, and even making
showed that a tenofovir gel and a placebo gel composed of impossible the calculation of TC50. Indeed, the ability to calculate
hydroxyethylcellulose, EDTA, citric acid, glycerin, and the pre- TC50 for almost all formulations in the ex vivo model indicates that
servatives methylparaben and propylparaben were detrimental this model is valuable to assure direct comparisons, even allowing
toward epithelial cells and explants, causing a reduced viability and the use of more concentrated dilutions of formulations, if experi-
epithelial layer integrity of HEC-1A and Caco-2 cells and cervical mentally possible.
tissue explants. Similar observations were made by Dezzutti et al.,78 There is a great deal of data that validate the porcine model of
assessing the toxicity and epithelial layer integrity of HEC-1A and vaginal mucosa in terms of structure, function, and reactivity in
Caco-2 cell lines after exposure to PRO-2000 gel, UC-781 gel, and comparison to human tissue. Both have stratified squamous
the placebo gels methylcellulose and Vena Gel. Furthermore, our epithelium supported by connective tissue. The use of small,
workgroup has previously tested several vaginal lubricants ex vivo, specimens provides convenience, efficiency, and high
commercially available, under a safety perspective, concerning throughput for screening.81,82 Samples of porcine tissue are inex-
cytotoxicity, pH, and osmolality.31 Earlier, a similar study has been pensive to obtain (in abundance on slaughterhouses) and handling
performed by Dezzuti et al. on 10 aqueous-based lubricants to test is easy when compared to the use of whole animals.22,81 Ex vivo
not only cellular toxicity and damage to epithelial monolayers but porcine tissues have been largely tested for drug permeation. But,
also their toxicity upon human explants epithelium. Main findings despite appearing to be a good ex vivo permeability model for
of toxicity to the microflora were attributed to preservatives con- human vaginal tissue extrapolation, it has already been shown that
tained in the formulations.69 Similarly to the study herein pre- its reliability may vary upon substances' chemical characteristics.
sented, these authors report difficulties in performing the toxicity For example, for hydrophilic molecules (water and vasopressin,
tests in the cellular models because, when testing creams (which e.g.), the porcine vaginal tissue is an accurate in vitro permeability
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3299

model of human tissue, whereas for more lipophilic molecules development. It could have application not only on the pharma-
(such as oxytocin), the flux could be higher than the corresponding ceutical industry and research but also to the cosmetic, hygiene,
estimated value for human tissue.83 Another limitation related to and medical devices industries. Moreover, it could further be
the ex vivo vaginal model is the fact that some authors use cervical applied to the development of primary cell cultures, as a surrogate
and uterine explants, to resemble the vaginal epithelia.37,77,84,85 for permeation and metabolic studies, and to originate cocultures
This may be due to the difficulty in accessing human vaginal with microorganisms.
epithelium and also based on the assumption that vaginal formu-
lations shall be safe not only to the vagina but also to the uterus. The Acknowledgments
porcine model can relatively circumvent this problem of accessing
specimens for testing while providing a good similitude with the This work was supported by FEDER funds through the
human vagina itself. Added to that, tissue collection and culture POCIdCOMPETE 2020dOperational Programme Competitiveness
procedures had not been standardized and optimized, before this and Internationalisation in Axis IdStrengthening research, tech-
research. Previous studies within our group have been focused on nological development and innovation (Project POCI-01-0145-
disclosing the likeliness of porcine vaginal tissue and the human FEDER-007491) and National Funds by FCTdFoundation for Sci-
tissue, and even disclosing which anatomical region of the porcine ence and Technology (Project UID/Multi/00709/2013), Portugal.
vagina is more appropriate to collect to comply with human simi- Financial support was also provided by Labfit, HPRD, Lda. and
larity (in preparation for publication). FCTdFoundation for Science and Technology through a PhD
The tetrazolium-based MTT assay has long been regarded as the fellowship (grant reference SFRH/BDE/111544/2015), Portugal.
gold standard of cytotoxicity assays as it is highly sensitive and has Rita Palmeira de Oliveira acknowledges FCT for financial support
been miniaturized for use as a high-throughput screening assay. (grant SFRH/BPD/124437/2016), Portugal.
The first use of the MTT assay upon tissue explants goes back to the Finally, the authors thank the slaughterhouse for kindly
90s, when it was applied to different types of tissues, for instance providing the porcine tissues and Dr. Catarina Ferreira for her
buccal mucosa.86,87 This method is also the one recommended by precious help with microscopy slides preparation and staining.
the manufacturers in the EpiVaginal™ and HVE™ reconstructed
models. Recently, Van Tonder et al., in 2015, conducted a study References
comparing several toxicity assays, including the MTT and NRU as-
says in MDA-MB-231, MCF-7, and MCF-12A cell lines. Results indi- 1. Woolfson AD, Malcolm RK, Gallagher R. Drug delivery by the intravaginal route.
Crit Rev Ther Drug Carrier Syst. 2000;17(5):509-555.
cated that the NRU showed one of the smallest variability across the
2. Alexander NJ, Baker E, Kaptein M, Karck U, Miller L, Zampaglione E. Why
linear range, while the largest variation was observed for the MTT consider vaginal drug administration? Fertil Steril. 2004;82(1):1-12.
assay. This means that this assay would more accurately detect 3. Hussain A, Ahsan F. The vagina as a route for systemic drug delivery. J Control
small changes in cell number than the MTT assay. Furthermore, the Release. 2005;103(2):301-313.
4. Srikrishna S, Cardozo L. The vagina as a route for drug delivery: a review. Int
MTT assay was the one to have more interferences with test Urogynecol J. 2013;24(4):537-543.
products.88 Although seeming less sensitive and with higher 5. Vermani K, Garg S. The scope and potential of vaginal drug delivery. Pharm Sci
number of interferences, MTT has a large historical use in cells and Technolo Today. 2000;3(10):359-364.
6. Nappi RE, Liekens G, Brandenburg U. Attitudes, perceptions and knowledge
tissues, and researchers are familiar with the methodology. about the vagina: the international vagina dialogue survey. Contraception.
Furthermore, standard documents like those from the “Interna- 2006;73(5):493-500.
tional Organization for Standardization” recommend this proced- 7. Sassi AB, McCullough KD, Cost MR, Hillier SL, Rohan LC. Permeability of tritiated
water through human cervical and vaginal tissue. J Pharm Sci. 2004;93(8):
ure for cytotoxic assessment of medical devices. 2009-2016.
8. Ensign LM, Cone R, Hanes J. Nanoparticle-based drug delivery to the vagina: a
Conclusions review. J Control Release. 2014;190:500-514.
9. Anderson MJ, Scholz MT, Parks PJ, Peterson ML. Ex vivo porcine vaginal
mucosal model of infection for determining effectiveness and toxicity of anti-
This work presented some preclinical toxicity results obtained septics. J Appl Microbiol. 2013;115(3):679-688.
for a panel of vaginal semisolid formulations, already regarded as 10. Chatterton BE, Penglis S, Kovacs JC, Presnell B, Hunt B. Retention and distri-
bution of two 99mTc-DTPA labelled vaginal dosage forms. Int J Pharm.
safe for human use. Regarding the cytotoxicity studies in 3 cell lines
2004;271(1):137-143.
representative of the female reproductive tract, with 2 different 11. Dukhin SS, Labib ME. Convective diffusion of nanoparticles from the epithelial
methods (MTT and NRU), we concluded that VK2 E6/E7 was barrier toward regional lymph nodes. Adv Colloid Interface Sci. 2013;199-200:
regarded as the more robust to use with the MTT assay, although it 23-43.
12. Allen LV, Popovich NG, Ansel HC. Ansel ’ S Pharmaceutical Dosage Forms and
was not sensitive to the neutral red assay. In addition, it was Drug Delivery Systems. 9th ed. Philadelphia, PA: Williams & Wilkins; 2011.
concluded that the ideal product concentration used to these assays 13. Hardy E, de P
adua KS, Jime
nez AL, Zaneveld LJ. Women’s preferences for
should be lower than 0.1% and up to 10% (w/v) to allow for assertive vaginal antimicrobial contraceptives II. Contraception. 1998;58(4):239-244.
14. Hardy E, de P
adua KS, Osis MJD, Jime
nez AL, Zaneveld LJ. Women’s preferences
calculations of the TC50 in all cell lines. Furthermore, this work not for vaginal antimicrobial contraceptives IV. Contraception. 1998;58(4):251-255.
only represents an optimization strategy to perform in vitro cellular 15. Woodsong C, Holt JDS. Acceptability and preferences for vaginal dosage forms
toxicity assays using semisolid aqueous and nonaqueous formula- intended for prevention of HIV or HIV and pregnancy *. Adv Drug Deliv Rev.
2015;92:146-154.
tions but also achieved the development of a preparation technique 16. Hardy E, Jime
Nez AL, De Pa
Dua KS, Zaneveld LJD. Women’s preferences for
for vaginal ex vivo porcine tissues. The later showed to be a vaginal antimicrobial contraceptives III. Contraception. 1998;58:245-249.
promising strategy for ex vivo toxicity testing because it demon- 17. Hardy E, Hebling EM, Sousa MH, Almeida AF, Amaral E. Delivery of micro-
bicides to the vagina: difficulties reported with the use of three devices,
strated high reproducibility and sensitivity, with low variability. adherence to use and preferences. Contraception. 2007;76:126-131.
Consistent results were obtained with the in vitro and ex vivo ap- 18. Palmeira-de-Oliveira R, Duarte P, Palmeira-de-Oliveira A, et al. What do Por-
proaches. Therefore, the combination of these 2 strategies (in vitro tuguese women prefer regarding vaginal products? Results from a cross-
sectional web-based survey. Pharmaceutics. 2014;6(4):543-556.
and ex vivo) could represent a powerful tool in the preclinical stages
19. Palmeira-de-Oliveira R, Duarte P, Palmeira-de-Oliveira A, et al. Women’s ex-
of product development not only for final formulations, as periences, preferences and perceptions regarding vaginal products: results
demonstrated in this work, but also for new substances or excipi- from a cross-sectional web-based survey in Portugal. Eur J Contracept Reprod
ents. The use of this methodology in preclinical safety assessments Health Care. 2015;20:259-271.
20. Machado RM, Palmeira-de-Oliveira A, Martinez-de-Oliveira J, Palmeira-de-
may optimize cost-efficiency of new formulations development by Oliveira R. Vaginal semisolid products: technological performance considering
predicting efficacy and safety profiles at early stages of product physiologic parameters. Eur J Pharm Sci. 2017;109:556-568.
3300 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301

21. Wu SJ, Robinson JR. Vaginal epithelial models. Pharm Biotechnol. 1996;8:409- 51. Caramella CM, Rossi S, Ferrari F, Bonferoni MC, Sandri G. Mucoadhesive and
424. thermogelling systems for vaginal drug delivery. Adv Drug Deliv Rev. 2015;92:
22. Costin G, Raabe H, Priston R. Vaginal irritation models: the current status of 39-52.
available alternative and in vitro tests. Altern Lab Anim. 2011;39(4):317-337. 52. Acartürk F. Mucoadhesive vaginal drug delivery systems. Recent Pat Drug Deliv
23. Ardolino LI, Meloni M, Brugali G, Corsini E, Galli CL. Preclinical evaluation of Formul. 2009;3(3):193-205.
Tolerability of a selective, Bacteriostatic, locally active vaginal formulation. Curr 53. Valenta C. The use of mucoadhesive polymers in vaginal delivery. Adv Drug
Ther Res Clin Exp. 2016;83:13-21. Deliv Rev. 2005;57(11):1692-1712.
24. Selen A. Biopharmaceutics and Quality by Design e Perspectives from FDA. CRS 54. Bygdeman M, Swahn ML. Replens versus dienoestrol cream in the symptom-
Product Development Forum: Poorly Soluble Drugs Unlocking the Door to New atic treatment of vaginal atrophy in postmenopausal women. Maturitas.
Paradigms in Drug Delivery January 27-29, 2011. Miami, FL. 2011. 1996;23(3):259-263.
25. Selen A, Dickinson PA, Müllertz A, et al. The biopharmaceutics risk assessment 55. Nachtigall LE. Comparative study: replens versus local estrogen in menopausal
roadmap for optimizing clinical drug product performance. J Pharm Sci. women. Fertil Steril. 1994;61(1):178-180.
2014;103(11):3377-3397. 56. Clark MR, McCormick TJ, Doncel GF, Friend DR. Preclinical evaluation of
26. Dickinson P, Kesisoglou F, Flanagan T. Optimizing clinical drug product per- UC781 microbicide vaginal drug delivery. Drug Deliv Transl Res. 2011;1(2):
formance: applying biopharmaceutics risk assessment roadmap (BioRAM) and 175-182.
the BioRAM scoring grid. J Pharm Sci. 2016;105:3243-3255. 57. Garg S, Goldman D, Krumme M, Rohan LC, Smoot S, Friend DR. Advances in
27. Fern
andez-Romero JA, Teleshova N, Zydowsky TM, Robbiani M. Preclinical development, scale-up and manufacturing of microbicide gels, films, and tab-
assessments of vaginal microbicide candidate safety and efficacy *. Adv Drug lets. Antiviral Res. 2010;88(Suppl 1):S19-S29.
Deliv Rev. 2014;92:27-38. 58. Kuramoto H, Tamura S, Notake Y. Establishment of a cell line of human
28. Lackman-Smith CS, Snyder BA, Marotte KM, et al. Safety and anti-HIV assess- endometrial adenocarcinoma in vitro. Am J Obstet Gynecol. 1972;114(8):1012-
ments of natural vaginal cleansing products in an established topical micro- 1019.
bicides in vitro testing algorithm. AIDS Res Ther. 2010;7:22. 59. Jones HW. Record of the first physician to see henrietta lacks at the Johns
29. Lard-Whiteford SL, Matecka D, O’rear JJ, Yuen IS, Litterst C, Reichelderfer P. hopkins hospital: history of the beginning of the HeLa cell line. Am J Obstet
Recommendations for the nonclinical development of topical microbicides for Gynecol. 1997;176(6):s227-s228.
prevention of HIV transmission: an update. J Acquir Immune Defic Syndr. 60. Plumb JA, Milroy R, Kaye SB. Effects of the pH dependence of 3-(4,5-
2004;36:541-552. dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide-formazan absorp-
30. Morrow KM, Fava JL, Rosen RK, et al. Designing preclinical perceptibility tion on chemosensitivity determined by a novel tetrazolium-based assay.
measures to evaluate topical vaginal gel formulations: relating user sensory Cancer Res. 1989;49(16):4435-4440.
perceptions and experiences to formulation properties. AIDS Res Hum Retro- 61. Berridge MV, Herst PM, Tan AS. Tetrazolium dyes as tools in cell biology: new
viruses. 2014;30(1):78-91. insights into their cellular reduction. Biotechnol Annu Rev. 2005;11:127-152.
31. Cunha AR, Machado RM, Palmeira-de-Oliveira A, Martinez-de-Oliveira J, das 62. Oecd. In Vitro 3T3 NRU phototoxicity test. Test. 2004;(April):1-15.
Neves J, Palmeira-de-Oliveira R. Characterization of commercially available 63. Yang X, Zhang W, Yang Y, Xiong X, Xie X, Tan X. Preliminary study on neutral
vaginal lubricants: a safety perspective. Pharmaceutics. 2014;6(3):530-542. red uptake assay as an alternative method for eye irritation test. Aatex.
32. Regulation (EC) No 1223/2009 of the European Parliament and of the Council of 2008;14:509-514.
30 November 2009 on Cosmetic Products and All Further Amendments. 64. Ohkuma S, Poole B. Cytoplasmic vacuolation of mouse peritoneal macrophages
33. Liebsch M, Grune B, Seiler A, et al. Alternatives to animal testing: current status and the uptake into lysosomes of weakly basic substances. J Cell Biol.
and future perspectives. Arch Toxicol. 2011;85:841-858. 1981;90(3):656-664.
34. Van Damme L, Ramjee G, Alary M, et al. Effectiveness of COL-1492, a 65. Olivier P, Testard P, Marzin D, Abbott D. Effect of high polyol concentrations on
nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: a the neutral red absorption assay and tetrazolium-MTT test of rat hepatocytes
randomised controlled trial. Lancet. 2002;360(9338):971-977. in primary culture. Toxicol In Vitro. 1995;9(2):133-138.
35. European Union Reference Laboratory for alternatives to animal testing (EURL- 66. Maguire RA, Bergman N, Phillips DM. Comparison of microbicides for efficacy
ECVAM). In: Topical Toxicity - Validation and Regulatory Acceptance; 2016. in protecting mice against vaginal challenge with herpes simplex virus type 2,
Available at: https://eurl-ecvam.jrc.ec.europa.eu/validation-regulatory-accepta cytotoxicity, antibacterial properties, and sperm immobilization. Sex Transm
nce/topical-toxicity/. Accessed October 31, 2016. Dis. 2001;28(5):259-265.
36. Mosmann T. Rapid colorimetric assay for cellular growth and survival: application 67. Moench TR, Mumper RJ, Hoen TE, Sun M, Cone RA. Microbicide excipients can
to proliferation and cytotoxicity assays. J Immunol Methods. 1983;65(1-2):55-63. greatly increase susceptibility to genital herpes transmission in the mouse.
37. Gali Y, Delezay O, Brouwers J, et al. In vitro evaluation of viability, integrity, and BMC Infect Dis. 2010;10:331.
inflammation in genital epithelia upon exposure to pharmaceutical excipients and 68. Begay O, Jean-Pierre N, Abraham CJ, et al. Identification of personal lubricants
candidate microbicides. Antimicrob Agents Chemother. 2010;54(12):5105-5114. that can cause rectal epithelial cell damage and enhance HIV type 1 replication
38. Dobaria NB, Badhan AC, Mashru RC. A novel itraconazole bioadhesive film for in vitro. AIDS Res Hum Retroviruses. 2011;27(9):1019-1024.
vaginal delivery: design, optimization, and physicodynamic characterization. 69. Dezzutti CS, Brown ER, Moncla B, et al. Is wetter better? An evaluation of over-
AAPS PharmSciTech. 2009;10(3):951-959. the-counter personal lubricants for safety and anti-HIV-1 activity. PLoS One.
39. Fichorova RN, Rheinwald JG, Anderson DJ. Generation of papillomavirus- 2012;7(11):e48328.
immortalized cell lines from normal human ectocervical, endocervical, and 70. Ayehunie S, Cannon C, Lamore S, et al. Organotypic human vaginal-ectocervical
vaginal epithelium that maintain expression of tissue-specific differentiation tissue model for irritation studies of spermicides, microbicides, and feminine-
proteins. Biol Reprod. 1997;57(4):847-855. care products. Toxicol In Vitro. 2006;20(5):689-698.
40. Denizot F, Lang R. Rapid colorimetric assay for cell growth and survival. €n KK, Praet M, et al. Development of an in vitro dual-chamber model
71. Gali Y, Arie
Modifications to the tetrazolium dye procedure giving improved sensitivity of the female genital tract as a screening tool for epithelial toxicity. J Virol
and reliability. J Immunol Methods. 1986;89(2):271-277. Methods. 2010;165(2):186-197.
41. Hansen MB, Nielsen SE, Berg K. Re-examination and further development of a 72. Gali Y, Delezay O, Brouwers J, Addad N, Augustijns P, Bourlet T. In vitro eval-
precise and rapid dye method for measuring cell growth/cell kill. J Immunol uation of viability, integrity, and inflammation in genital epithelia upon
Methods. 1989;119(2):203-210. exposure to pharmaceutical excipients and candidate microbicides. Antimicrob
42. Borenfreund E, Babich H, Martin-Alguacil N. Rapid chemosensitivity assay with Agents Chemother. 2010;54(12):5105.
human normal and tumor cells in vitro. In Vitro Cell Dev Biol. 1990;26(11): 73. Goeman J, Ndoye I, Sakho LM, et al. Frequent use of menfegol spermicidal
1030-1034. vaginal foaming tablets associated with a high incidence of genital lesions.
43. Fotakis G, Timbrell JA. In vitro cytotoxicity assays: comparison of LDH, neutral J Infect Dis. 1995;171(6):1611-1614.
red, MTT and protein assay in hepatoma cell lines following exposure to cad- 74. Bhat HK, Calaf G, Hei TK, Loya T, Vadgama JV. Critical role of oxidative stress in
mium chloride. Toxicol Lett. 2006;160(2):171-177. estrogen-induced carcinogenesis. Proc Natl Acad Sci U S A. 2003;100(7):3913-
44. Repetto G, del Peso A, Zurita JL. Neutral red uptake assay for the estimation of 3918.
cell viability/cytotoxicity. Nat Protoc. 2008;3(7):1125-1131. 75. Owen DH, Katz DF. A vaginal fluid simulant. Contraception. 1999;59(2):91-95.
45. Chiba K, Kawakami K, Tohyama K. Simultaneous evaluation of cell viability by 76. Fichorova RN, Tucker LD, Anderson DJ. The molecular basis of nonoxynol-9-
neutral red, MTT and crystal violet staining assays of the same cells. Toxicol induced vaginal inflammation and its possible relevance to human immuno-
In Vitro. 1998;12(3):251-258. deficiency virus type 1 transmission. J Infect Dis. 2001;184(4):418-428.
46. Freshney RI. Culture of animal cells set. Med Clin (Barc). 2000;82:273-280. 77. Rohan LC, Moncla BJ, Kunjara Na Ayudhya RP, et al. In vitro and ex vivo testing
47. ISO/EN10993-5. B. Int Stand ISO 10993-5 Biol Eval Med devices - Part 5 Tests of tenofovir shows it is effective as an HIV-1 microbicide. PLoS One. 2010;5(2):
Cytotox Vitr methods. 3rd ed. Geneva, Switzerland: ISO (International Organi- e9310.
zation for Standardization); 2009:42. 78. Dezzutti CS, James VN, Ramos A, et al. In vitro comparison of topical micro-
48. Tien D, Schnaare RL, Kang F, et al. In vitro and in vivo characterization of a bicides for prevention of human immunodeficiency virus type 1 transmission.
potential universal placebo designed for use in vaginal microbicide clinical Antimicrob Agents Chemother. 2004;48(10):3834-3844.
trials. AIDS Res Hum Retroviruses. 2005;21(10):845-853. 79. Garg S, Tambwekar KR, Vermani K, Garg A. Compendium of pharmaceutical
49. Schwartz JL, Ballagh SA, Kwok C, et al. Fourteen-day safety and acceptability excipients for vaginal formulations. Pharm Technol. 2001:14.
study of the universal placebo gel. Contraception. 2007;75(2):136-141. 80. Gorodeski GI. The cultured human cervical epithelium: a new model for
50. Adriaens E, Remon JP. Mucosal irritation potential of personal lubricants relates studying paracellular transport. J Soc Gynecol Investig. 1996;3(5):267-280.
to product osmolality as detected by the slug mucosal irritation assay. Sex 81. Squier C, Mantz M. Porcine vagina ex vivo as a model for studying permeability
Transm Dis. 2008;35(5):512-516. and pathogenesis in mucosa. J Pharm Sci. 2008;97(1):9-21.
 R.M. Machado et al. / Journal of Pharmaceutical Sciences 108 (2019) 3289-3301 3301

82. Machado RM, Palmeira-de-Oliveira A, Gaspar C, Martinez-de-Oliveira J, Pal- 86. Imbert D, Cullander C. Buccal mucosa in vitro experiments. J Control Release.
meira-de-Oliveira R. Studies and methodologies on vaginal drug permeation. 1999;58(1):39-50.
Adv Drug Deliv Rev. 2015;92:14-26. 87. Radosevich JA, Haines GK, Elseth KM, Shambaugh GE, Maker VK. A new method
83. van Eyk AD, van der Bijl P. Porcine vaginal mucosa as an in vitro permeability for the detection of viable cells in tissue sections using 3-[4,5-dimethylthiazol-
model for human vaginal mucosa. Int J Pharm. 2005;305(1-2):105-111. 2-yl]-2,5-diphenyltetrazolium bromide (MTT): an application in the assess-
84. Cummins JE, Guarner J, Flowers L, et al. Preclinical testing of candidate topical ment of tissue damage by surgical instruments. Virchows Arch B Cell Pathol Incl
microbicides for anti-human immunodeficiency virus type 1 activity and tissue Mol Pathol. 1993;63(6):345-350.
toxicity in a human cervical explant culture. Antimicrob Agents Chemother. 88. van Tonder A, Joubert AM, Cromarty AD. Limitations of the 3-(4,5-
2007;51(5):1770-1779. dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay
85. Merbah M, Introini A, Fitzgerald W, et al. Cervico-vaginal tissue ex vivo as a model when compared to three commonly used cell enumeration assays. BMC Res
to study early events in HIV-1 infection. Am J Reprod Immunol. 2011;65(3):268-278. Notes. 2015;8(1):47.