Content uploaded by Russel M Walters
Author content
All content in this area was uploaded by Russel M Walters on Nov 13, 2017
Content may be subject to copyright.
Introduction
The current efforts of the consumer products
industry are focused on finding in vitro methods
that can reliably assess the enhanced or improved
mildness of final formulations designed for topical
application to the skin. Traditionally, a combined
strategy based on the Draize rabbit primary der-
mal irritation test (1) and human clinical testing
was used to demonstrate enhanced mildness.
However, ethical concerns over unnecessary ani-
mal use, in addition to the need for efficient, cost-
effective preclinical screening assays for prototype
exploration, have promoted the development of
alternative methods for determining skin irrita-
tion (2–10).
As a fully immunocompetent organ, the skin
responds with an inflammatory reaction to the var-
ious chemicals or formulated products it comes in
contact with, either accidentally or intentionally.
The local inflammatory reaction following expo-
sure to an irritant is believed to start with the
release of primary cytokines (such as interleukin
[IL]-1α) that stimulate the synthesis and release of
multifunctional secondary cytokines, which then
escalate and drive the inflammatory cascade or
serve to repress the reaction through negative
feedback. The release of pro-inflammatory medi-
ators from skin equivalents in response to an irri-
tant has been shown to be useful for the
prediction of skin irritation potential, when in
vitro results were compared with the human clini-
cal response (2, 3). To that end, skin equivalents
like EpiDerm™ (MatTek Corporation, Ashland,
MA, USA) have become the prime choice for alter-
native testing methods.
Personal care cleansers typically contain multi-
ple different surfactant species in solution. The dif-
ferent surfactants self-assemble into mixed
micelles, typically spherical structures containing
the different surfactants. These mixed micelle sur-
factant systems, which often behave differently to
the individual surfactants, can change the thermo-
dynamics of the surfactant system and the extent
to which surfactants penetrate and disrupt the
skin (11–14). Therefore, the resultant properties of
the cleansers are determined by the way the sur-
factant system works. Baby cleansers are typically
In Vitro Assessment of Skin Irritation Potential of
Surfactant-based Formulations by Using a 3-D Skin
Reconstructed Tissue Model and Cytokine Response
Russel M. Walters,1Lisa Gandolfi,2M. Catherine Mack,1Michael Fevola,1Katharine Martin,1
Mathew T. Hamilton,3Allison Hilberer,4Nicole Barnes,4Nathan Wilt,4Jennifer R. Nash,4Hans
A. Raabe4and Gertrude-Emilia Costin4
1Johnson & Johnson Consumer Inc., Skillman, NJ, USA; 2Clariant Corporation, Charlotte, NC, USA;
3Wellspring Worldwide Inc., Chicago, IL, USA; 4Institute for In Vitro Sciences Inc., Gaithersburg, MD, USA
Summary — The personal care industry is focused on developing safe, more efficacious, and increasingly
milder products, that are routinely undergoing preclinical and clinical testing before becoming available for
consumer use on skin. In vitro systems based on skin reconstructed equivalents are now established for the
preclinical assessment of product irritation potential and as alternative testing methods to the classic Draize
rabbit skin irritation test. We have used the 3-D EpiDerm™ model system to evaluate tissue viability and
primary cytokine interleukin-1αrelease as a way to evaluate the potential dermal irritation of 224 non-
ionic, amphoteric and/or anionic surfactant-containing formulations, or individual raw materials. As part of
our testing programme, two representative benchmark materials with known clinical skin irritation poten-
tial were qualified through repeated testing, for use as references for the skin irritation evaluation of for-
mulations containing new surfactant ingredients. We have established a correlation between the in vitro
screening approach and clinical testing, and are continually expanding our database to enhance this cor-
relation. This testing programme integrates the efforts of global manufacturers of personal care products
that focus on the development of increasingly milder formulations to be applied to the skin, without the
use of animal testing.
Key words: 3-D skin equivalents, IL-1
α
, in vitro, irritation, surfactants.
Address for correspondence: Russel M. Walters, Johnson & Johnson Consumer Inc., 199 Grandview Rd,
Skillman, NJ, USA.
E-mail: rwalter2@its.jnj.com
ATLA 44, 523–532, 2016 523
mild to the skin, and they often contain blends of
anionic, amphoteric and non-ionic surfactants.
Adult face cleansers are typically also mild, and
may contain blends of anionic, amphoteric and
non-ionic surfactants, whereas adult shampoos are
usually more aggressive and contain primarily
anionic surfactants. The penetration of surfactants
into the stratum corneum of the skin, and the ways
in which surfactants interact with the endogenous
skin lipids, can affect the barrier function of the
skin (11, 13–16) and result in increases in
transepidermal water loss (TEWL; 17–19), which
can be assessed clinically.
A critical component in the development of reli-
able in vitro screening methods for skin irritation is
the correlation of in vitro test results with clinical
skin irritation performance. The more prevalent in
vitro tests used to assess surfactant aggressiveness,
such as the Zein solubilisation test (5, 6) or transep-
ithelial permeation test (also known as fluorescein
dye leakage test; 4, 8, 9), have not been successfully
correlated to in vivo skin irritation models. Thus,
significant effort was put into generating a reliable
correlation between data provided by in vitro assays
based on reconstructed tissue models and clinical
studies for the assessment of skin irritation poten-
tial of surfactant-based formulations.
By employing 3-D reconstructed skin tissues
with tissue viability and IL-1αexpression analysis
as endpoints, we have developed an in vitro test
method to evaluate the potential dermal irritation
of non-ionic, amphoteric and/or anionic surfactant-
containing cleansing formulations or individual
surfactants. An exaggerated patch test with the
TEWL endpoint was used as the clinical assess-
ment for correlation with our in vitro test method.
Due to the successful correlation between in vitro
and clinical results, the EpiDerm-based dermal
irritation test can be used to compare and analyse
the potential dermal irritation of new formulations
as a guideline for formulation development for new
mild(er) skin-cleansing products.
Materials and Methods
Test articles
Sixteen individual surfactants (considered to be
‘raw materials’), 46 commercial surfactant-based
cleansers, and 162 other prototype surfactant-
based cleansers (non-commercialised formula-
tions), were used as test articles, making a total of
224 different test articles.
The test articles were tested as 10% dilutions
(weight/volume percentage [w/v]) in sterile,
deionised water (Quality Biological, Gaithersburg,
MD, USA), unless otherwise specified. As typical
mild surfactant-based cleansers contain about 10%
(w/v) surfactant, the 16 individual surfactants were
diluted to 10% (w/v) in sterile, deionised water, then
tested at 10% (w/v) dilutions, for a final concentra-
tion of 1% (w/v).
The 10% (w/v) dilution mimics the end-user expo-
sure to wash-off products such as shampoos, condi-
tioners, etc. Two test articles were included in nearly
every study and were qualified as benchmark mate-
rials (Mild Cleanser 1 and Mild Cleanser 2). They
were tested 32 and 35 unique times, respectively;
each individual test consisted of three independent
tissue samples. These two benchmark materials
were mild cleansers containing a mixture of anionic,
amphoteric and non-ionic surfactants. Triton®
X-100 (Fisher Scientific, Pittsburgh, PA, USA) was
used as the assay positive control to assess the qual-
ity of the tissue lots used in the experiments. Sterile,
deionised water (Quality Biological) was used as the
assay negative control.
Reconstructed tissues
The 3-D human reconstructed epidermal model
EpiDerm Skin Model (EPI-200) provided by
MatTek Corporation (Ashland, MA, USA) was
used in our experiments. The EpiDerm tissues are
based on normal, human-derived epidermal ker-
atinocytes cultured to form a multilayered, highly
differentiated model of the human epidermis (20).
The EpiDerm tissues were cultured in a Dulbecco’s
Modified Eagle Medium-based culture medium
provided by the tissue supplier. Since the tissues
have a functional stratum corneum, the test arti-
cles were applied directly to the culture surface, at
the air interface.
Treatment of the EpiDerm tissues
The EpiDerm tissues were stored at 2–8°C until
used. The day before treatment, the EpiDerm tis-
sues were cultured in six-well plates containing a
hydrocortisone free-assay medium (HCF-AM), and
equilibrated at 37 ± 1°C in a humidified atmo-
sphere of 5 ± 1% CO2in air (standard culture con-
ditions) overnight.
Each EpiDerm tissue was considered an inde-
pendent sample. At least 16 hours after initiating
the tissue cultures, the medium was removed from
under the tissues and a 0.9ml aliquot of fresh, pre-
warmed HCF-AM was added to each well. Each
test article (100μl) was applied onto three tissues,
and the negative control (100μl sterile, deionised
H2O) was added to the other three tissues in the
six-well plate. At the end of the 1-hour exposure
period, each tissue was rinsed five times with
approximately 0.5ml per rinse of calcium-free and
magnesium-free Dulbecco’s phosphate-buffered
saline (CMF-DPBS; Quality Biological). After rins-
524 R.M. Walters et al.
ing, each tissue was placed in the designated well
of a new six-well plate containing 0.9ml of fresh
HCF-AM, and incubated under standard culture
conditions for the post-exposure incubation period
(24 hours).
Viability assay
Tissue viability was determined by using a method
based on the reduction of the yellow tetrazolium
salt 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetra-
zolium bromide (MTT) to the purple formazan dye
by mitochondrial succinate dehydrogenase in
viable cells (21). A 1mg/ml solution of MTT in
warm MTT addition medium was prepared no
more than 2 hours before use.
Upon completion of the 24-hour post-exposure
incubation, the tissues were removed from their
incubation medium, rinsed with CMF-DPBS,
blotted dry and transferred into pre-labelled 24-
well plates containing 300μl MTT solution per
well. The medium from under each tissue was
quick-frozen (≤ –60°C) for subsequent cytokine
analysis.
After 3 ± 0.1 hours of incubation in MTT, the
EpiDerm tissues were blotted on absorbent paper
and transferred into 24-well plates containing 2ml
of isopropanol per well and shaken at room tem-
perature. After 2 hours, the absorbance of a 200μl
aliquot of tissue extract was measured at 550nm
(Vmax® Kinetic ELISA microplate reader;
Molecular Devices, Sunnyvale, CA, USA). The via-
bility of the tissues exposed to the test articles was
calculated and expressed as a percentage relative
to the viability of the negative control (i.e.
deionised water-treated tissues). The tissue viabil-
ity value was taken as the mean value from the
three independent wells tested in each experi-
ment.
IL-1αanalysis
The IL-1αconcentration was determined by using
a kit from R&D Systems (Minneapolis, MN, USA),
according to the manufacturer’s instructions.
Thawed media samples, collected as described pre-
viously, were tested neat and as 1:10 dilutions, to
keep the readings within the linear range of the
assay. The IL-1αvalue reported for each test was
the mean value from the three independent tissues
used per test article in each experiment and plated
in duplicate.
Clinical study: Exaggerated patch test
Adult subjects (aged 18–65 years), who had been
diagnosed with atopic dermatitis, and therefore
had skin with impaired barrier function, were
recruited into the study. Each subject gave written
informed consent to participate, and the protocol
for the study was approved by an institutional
review board. Each subject was exposed to the test
articles (diluted to 50% with distilled water) on the
volar forearm under an occlusive patch for
24 hours (n= 25 subjects/panel). Typical in-use
exposure to personal care cleansers occurs over a
varying range of cleanser dilution during the
course of the washing period. A 50% dilution is
likely a relatively high cleanser concentration, and
was used in our clinical study in order to differen-
tiate between relatively mild surfactant systems.
After 24 hours, the patch and cleansing solution
were removed, and a new patch with the same
treatment was reapplied; this operation was
repeated for a total of four days. The skin barrier
function was measured before and after four days
of patching. The TEWL value was measured by
using a VapoMeter (Delfin Technol ogies, Kuopio,
Finland), as described previously (22). Clinical
TEWL results represented the change from the
baseline (i.e. before patching of the test article)
TEWL at Day 4. Each of the personal care cleans -
ers was tested on a panel of subjects; 27 subjects
were recruited into each testing panel, and at
least 25 subjects in each panel completed the
study. The data set of 28 cleansers combines the
results from three different panels that occurred
at different times of the year.
Results
Test system reproducibility: Tissue viability
and IL-1αendpoints
To assess the test system reproducibility and to
determine the actual shapes of the distributions,
the tissue viability (MTT endpoint) and cytokine
release (IL-1αendpoint) data were analysed for
two mild cleanser formulas, two positive controls
and one negative control that were repeatedly
tested. Figure 1a shows the experimentally deter-
mined distribution of tissue viability after treat-
ment with the two benchmark materials. In
statistical analysis, commonly normal distribu-
tions are assumed, while biological systems often
produce distributions that are not normal. Due to
the repeated measurement, the distribution can be
experimentally generated for the four different
exposures. For each exposure, the set of responses
were binned (bin width of 5%), and the fraction of
all responses are shown for each bin. The mean per-
centages (± SD) of viable tissues treated with Mild
Cleanser 1 and Mild Cleanser 2 were 100.1 ± 7.1%
and 96.9 ± 6.4%, respectively. Also shown in
Figure 1a are the results of the positive control,
Skin irritation potential of cleansers 525
Figure 1: Tissue viability and IL-1αendpoints
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0
010
a)
20 30 40 50 60 70 80 90 100 110
tissue viability (%)
fraction of tests
0.5
0.4
0.3
0.2
0.1
0
25
c)
31 39 49 61 77 96 120 150 187 234 293
Log IL-1α(pg/ml)
fraction of tests
0.5
0.4
0.3
0.2
0.1
0
025
b)
50 75 100 125 150 175 200 225 250 275
IL-1α(pg/ml)
fraction of tests
= Positive control (8 hours); = positive control (4 hours); = Mild Cleanser 2; = Mild Cleanser 1;
= negative control in c).
Graph a) shows the shape of the distribution and the reproducibility of tissue viability (%). The results of the positive
control, Triton X-100 (4-hour and 8-hour exposures), are also shown. Graph b) shows the IL-1
α
concentration (pg/ml)
for tissues treated with Mild Cleanser 1 and Mild Cleanser 2. IL-1
α
results are shown on a linear scale (graph b) and
log scale (graph c).
526 R.M. Walters et al.
Triton X-100-treated tissue, which, when exposed
for 4 and 8 hours, demonstrated a reduction in via-
bility of 18.9 ± 10.6% and 82.5 ± 10%, respectively.
IL-1αrelease from the same tissues is shown in
Figure 1b, on both linear and log scales. Also shown
in Figure 1b are the results of the negative control
(i.e. deionised water-treated tissues), after the 1-
hour exposure time. Although the viability assay
showed no difference between the two benchmark
materials (Mild Cleansers 1 and 2), there was a
clear difference in IL-1αproduction between them,
indicating that the sensitivity of the IL-1αendpoint
was greater than that of the MTT viability end-
point, and also suggesting that Mild Cleanser 2 had
slightly greater irritation potential than Mild
Cleanser 1. Mean values for IL-1αrelease after
exposure to Cleanser 1 (n= 32) and Cleanser 2 (n=
35) were 69 ± 25pg/ml and 149 ± 53pg/ml, respec-
tively, and the difference between the cleansers in
IL-1αrelease was significant (P < 0.001). Mean val-
ues for IL-1αrelease after exposure to the assay
positive control (Triton X-100) for 4 and 8 hours
were 100 ± 44pg/ml and 320 ± 124pg/ml, respec-
tively, and 20 ± 10pg/ml after exposure to deionised
water, the assay negative control. Therefore, the IL-
1αresponse was used to assess potential dermal
irritation of raw materials and final formulations in
this testing programme.
Skin irritation continuum for surfactant-
based test articles with rinse-off product
applications
The sensitivity of the test system was also investi-
gated for a wide range of surfactant-based cleans-
ing formulations. Figure 2a shows the distribution
of tissue viability values obtained for 224 different
surfactant-based formulations. Most of these test
articles do not impact tissue viability, and the
results are distributed as expected, at around
100%. There is a small set of results with viability
values from 0 to 15%, and a smaller fraction (4% of
the total) with viability from 20% to 70%. The
observed distribution from these surfactant-based
cleansing formulations of tissue viability appears
to be largely binary, either viable or not viable.
Tissue viability does not appear to be a continuous
variable and, as such, traditional statistical
descriptions are not valid. Figure 2b shows the dis-
tribution in the IL-1αresponse across the range of
surfactant-based cleansers. The distribution is
clearly non-linear; there is a high fraction of
results that exhibit low IL-1αresponses and a long
tail of rare results that exhibit very high IL-1α
responses. The insert shows a power law distribu-
tion, which appears to be a more appropriate fit to
the data.
Figure 2c shows the correlation between tissue
viability and IL-1αobtained for the same set of test
articles. In the skin irritation continuum assessed
by the two endpoints, for over 90% of the test arti-
cles, the exhibited tissue viability was around
100%. Within this majority of test articles, a wide
range of IL-1αconcentrations (between 10pg/ml
and approximately 600pg/ml) was observed. For
the remaining approximately 10% of test articles,
the tissue viability was below 85% and was typi-
cally associated with a high IL-1αrelease (Figure
2b). For the remaining formulations tested, IL-1α
production of ≥ 600pg/ml indicated increased toxic-
ity associated with materials that can be classified
as moderate to severe irritants. Overall, our data
confirmed the initial observations that IL-1α
released from the tissues is a more sensitive end-
point for distinguishing and rank-ordering final
formulations that range in irritancy potential.
Skin irritation assessment of commercial
skincare cleansers by the EpiDerm test
system
Previous studies reported that different classes of
surfactants have different skin irritation profiles
(12, 23). In our study, 16 individual surfactants
commonly used in personal care formulations were
assessed. As shown in Figure 3a, there was a wide
range of responses (over two orders of magnitude
in IL-1αrelease). At constant active concentra-
tions, non-ionic surfactants (light grey bars) were
assessed by our test system as the mildest (lowest
IL-1αreleased), whereas the anionic surfactants
(black bars) resulted in the greatest IL-1α release.
Amphoteric surfactants (dark grey bars) had inter-
mediate effects. Among the anionic surfactants,
the sulphate and sulphonate head groups were
most irritating, whereas surfactants with alterna-
tive anionic head groups (such as isethionate or
glutamate) were milder. Glucamide non-ionic sur-
factants with longer carbon chains, such as cocoyl
methyl glucamide and lauroyl methyl glucamide
(C16–18), were milder than shorter carbon chains,
such as capryloyl/caproyl methyl glucamide (C8–
10). Independent of charge, surfactants with long
hydrophilic repeat units, such as polyglycerol or
polyethylene oxide, were among the mildest in
their particular class. Polyglycerol-10 laurate and
sodium laureth-13 carboxylate are representative
of this trend.
We used this in vitro system to determine the
surfactant aggressiveness of 46 different commer-
cial skin cleansers. The commercial cleansers were
grouped into three categories based on their stated
marketing segment: baby washes and shampoos
(n= 21), adult facial cleansers (n= 9), and adult
body washes and shampoos (n= 16). As shown in
Figure 3b, each category displayed a range of irri-
tation responses: baby wash and shampoo tended
to induce less IL-1αrelease than adult body wash
and shampoo, which induced a greater release of
Skin irritation potential of cleansers 527
Figure 2: Sensitivity of the test system to 224 different surfactant-based formulations
Graph a) shows the distribution of the tissue viability response (%) for the 224 surfactant-based test articles in bins of
tissue viability. Graph b) shows the distribution of the IL-1
α
response for the 224 surfactant-based test articles in
bins of IL-1
α
. Graph c) shows the correlation between tissue viability (%) and IL-1
α
concentration (pg/ml) for 224
surfactant-based test articles with rinse-off applications. The dotted line indicates 85% viability.
0.25
0.20
0.15
0.10
0.05
0
010
a)
20 30 40 50 60 70 80 90 100 110 120 130
tissue viability (%)
fraction of tests
0.4
0.3
0.2
0.1
0
0 50 150 250 350 450 550 650 750 850 950 1050 1150 1250 1350 1450 1550 1650
b)
Log IL-1α(pg/ml)
fraction of tests
100
10–1
10–2
10–3
50
500
Log IL-1α(pg/ml)
fraction of tests
120
100
80
60
40
20
0
101102103
c)
IL-1α(pg/ml)
tissue viability (%)
528 R.M. Walters et al.
Figure 3: Cytokine release in response to cleanser exposure
= Anionic; = amphoteric; = non-ionic.
SH = sodium hydrolysed.
Graph a) shows the individual surfactants typically used in commercial cleanser products.
Graph b) shows the three classes of commercial cleanser products. For each class of cleansers, mean values are
represented by horizontal lines. The * in graph b) represents a liquid castile soap considered to be an outlier for the
class of baby washes and shampoo products. Each set of symbols is slightly offset for ease of visualisation.
Alpha olefin sulphonate
Sodium lauryl sulphate
Sodium lauroyl methyl isethionate
Sodium cocoyl glycinate
Coco-betaine
Sodium cocoyl glutamate
Sodium cocoyl isethionate
Capryloyl/caproyl methyl glucamide
Lauroyl methyl glucamide and ethanol
Cocamidopropyl betaine
Sodium-laureth-13 carboxylate
Cocoyl methyl glucamide
Coco-glucoside
Lauroyl methyl glucamide
Polyglyceryl-10 laurate
SH potato starch dodecenylsuccinate
101102
IL-1α(pg/ml)
103104
103
102
101
0baby wash
and shampoo
IL-1α(pg/ml)
adult facial
cleanser
adult body
wash and
shampoo
*
b)
a)
Skin irritation potential of cleansers 529
the cytokine. The adult facial cleanser group had
the widest range of irritation responses; this was
expected, as this group contained mixed categories
of cleansers with some intended for use on sensi-
tive skin, described as mild products, and some
more aggressive products, such as acne cleansers.
In the baby wash and shampoo class, although
most commercial formulations induced little IL-1α
release (median 61pg/ml), one of the commercial
baby cleansers appeared to be an outlier, inducing
the release of 504pg/ml of IL-1α(depicted as an
asterisk in Figure 3b). This outlier product was a
liquid castile soap with a high pH of 9.6. Castile
soaps and other high-pH cleansers are well known
to be harsh, and thus are not recommended for
babies by the Association of Women’s Health,
Obstetric and Neonatal Nurses (AWHONN).
AWHONN guidelines for neonatal skin care (24) rec-
ommend neutral to slightly acidic pH (5.5 to 7.0)
cleansers for babies. Although this outlier product is
marketed as a baby cleanser, it would not be consid-
ered by AWHONN as an appropriate baby cleanser
product, and the high level of IL-1αrelease identifies
the product as a potential irritant.
Correlation of preclinical and clinical test
results
Twenty-eight different personal care cleansers
were evaluated in the in vitro IL-1α release test
and in a human clinical study that tested the
effects of the cleansers on the skin barrier. Figure
4 shows the correlation between IL-1αrelease from
the EpiDerm reconstructed tissues and barrier dis-
ruption (assessed by change in TEWL) in the clin-
ical study. The range of IL-1αcytokine release
spanned the full range of cleanser aggressiveness
that was observed in the commercial cleanser set
in Figure 3b. Over this wide range of surfactant-
based skin cleansers, a positive correlation could
be established between the results generated by
the in vitro test system and in the human clinical
study. For example, increased TEWL corresponded
to high IL-1αproduction, whereas minimal
changes in TEWL corresponded to a lower cytokine
response. These results suggested that the in vitro
test successfully predicted the change in TEWL, a
known human clinical response to surfactant-
based personal care cleansers. A calculation based
on this data set showed an R2= 0.66, which indi-
cates a strong correlation (P < 0.01) when consider-
ing the variation that is typically present in
human test models.
Discussion
Reproducible in vitro test systems can be used to
accurately assess the irritation potential of raw
materials and formulated products, thus avoiding
the use of animals for testing. The test system pre-
sented here, based on the 3-D EpiDerm model
(MatTek Corporation), and employing tissue viabil-
ity and IL-1αrelease as the endpoints, has proven to
be a useful tool for new formulation development
and provided a testing platform for Johnson &
Johnson’s skin irritation assessment programme.
During our testing programme, two representa-
tive benchmark materials were qualified for use as
reference materials in the evaluation of skin irrita-
tion of formulations containing new surfactant
ingredients. We have developed a database and
established a range of potential irritation
responses to the benchmarks, by using the IL-1α
release endpoint (Figures 1a and 1b). Comparison
of the dermal irritation potential of new formula-
tions with that of existing mild formulations can
guide the formulation development of new mild
cleansing products.
We also report on a large in vitro data set of indi-
vidual surfactants and typical skincare cleansing
systems. Although the tissue viability endpoint
could identify relatively harsh personal care
cleansers, IL-1αwas found to be the more sensitive
endpoint in discriminating mild cleansers (Figures
2 and 3). Values of tissue viability of ≤ 80% are
Figure 4: The correlation between cytokine
release in vitro and clinical TEWL
measurement
R2= coefficient of determination = 0.66;
TEWL = transepidermal water loss.
The graph shows the correlation between in vitro IL-1
α
concentration (pg/ml) and the change in TEWL after
4 days’ treatment in the exaggerated patch clinical model.
50.0
5.0
0.5
100101
in vitro: IL-1α(pg/ml)
clinical: TEWL (mg/cm2/h)
102103
530 R.M. Walters et al.
generally associated with unpredictable, unreli-
able, and difficult-to-interpret levels of IL-1α
released by the affected tissues (3). Within our
testing programme, the model was valuable for the
comparison of internal products with products pro-
duced by others, and was adaptable to a variety of
rinse-off products with a range of mildness pro-
files. Additionally, we found a promising correla-
tion between our large set of in vitro results and
the corresponding human clinical results that
assessed changes in skin barrier function by using
TEWL as an endpoint, demonstrating the value of
the in vitro testing platform in predicting clinical
results (Figure 4).
It is well known that surfactant concentration
and surfactant combinations in fully formulated
cleansers affect the overall dermal irritation poten-
tial (12, 14, 23). Thus, the irritation assessment of
individual surfactants is typically used as a cur-
sory pre-screening technique before formulation
development. It is necessary to also evaluate the
irritation potential of fully formulated prototypes
or products. In addition to providing a measure of
dermal irritation potential of individual surfac-
tants for formulation guidance, the test system
based on the EpiDerm model has been a useful
research tool for evaluating the dermal irritation
potential of internal products and prototypes.
Overall, our data showed that the in vitro der-
mal irritation test specifically designed to address
the testing needs for the classes of raw materials
and formulated products, described here, has util-
ity for developing an understanding of individual
ingredient skin irritation potential, and for pre-
dicting clinical irritation profiles, as well as for
guiding the formulation development of non-irri-
tating rinse-off products.
Acknowledgements
This work was supported in full by Johnson &
Johnson Consumer Inc., Skillman, NJ, USA.
The authors would like to acknowledge the clin-
ical subjects participating in these studies. Medical
writing and editorial assistance was provided by
Alex Loeb, PhD, CMPP, of Evidence Scientific
Solutions (Philadelphia, PA, USA), and was
funded by Johnson & Johnson Consumer Inc.
Received 05.05.16; received in final form 25.10.16;
accepted for publication 27.10.16.
References
1. Draize, J.H., Woodard, G. & Calvery, H.O. (1944).
Methods for the study of irritation and toxicity of
substances applied topically to the skin and mucous
membranes. Journal of Pharmacology & Experi -
mental Therapeutics 82, 377–390.
2. Bernhofer, L.P., Barkovic, S., Appa, Y. & Martin,
K.M. (1999). IL-1αand IL-1ra secretion from epi-
dermal equivalents and the prediction of the irrita-
tion potential of mild soap and surfactant-based
consumer products. Toxicology in Vitro 13, 231–
239.
3. Bernhofer, L.P., Seiberg, M. & Martin, K.M. (1999).
The influence of the response of skin equivalent sys-
tems to topically applied consumer products by
epithelial–mesenchymal interactions. Toxicology in
Vitro 13, 219–229.
4. Cottin, M. & Zanvit, A. (1997). Fluorescein leakage
test: A useful tool in ocular safety assessment.
Toxicology in Vitro 11, 399–405.
5. Deo, N., Jockusch, S., Turro, N.J. & Somasundaran,
P. (2003). Surfactant interactions with Zein protein.
Langmuir 19, 5083–5088.
6. Gotte, E. (1964). Skin Compatibility of Tensides
Measured by Their Capacity for Dissolving Zein
Protein. [Presentation at Proceedings of the 4th Inter -
national Congress of Surface Active Substances,
Brussels, Belgium, 7–12 September 1964].
7. Lee, J.K., Kim, D.B., Kim, J.I. & Kim, P.Y. (2000).
In vitro cytotoxicity tests on cultured human skin
fibroblasts to predict skin irritation potential of sur-
factants. Toxicology in Vitro 14, 345–349.
8. Martin, K. & Stott, C. (1992). The Trans-Epithelial
Permeability Assay as an In Vitro Assay for
Predicting Ocular Irritation of Surfactant Form -
ulations. [Presentation at World Congress on Cell
and Tissue Culture, Washington, DC, USA, 20–25
June 1992].
9. Tchao, R. (1988). Trans-epithelial permeability of
fluorescein in vitro as an assay to determine eye
irritants. In Alternative Methods in Toxicology (ed.
A.M. Goldberg), pp. 271–283. New York, NY, USA:
Mary Ann Liebert, Inc.
10. Vinardell, M.P. & Mitjans, M. (2008). Alternative
methods for eye and skin irritation tests: An over -
view. Journal of Pharmaceutical Sciences 97, 46–
59.
11. Ghosh, S. & Blankschtein, D. (2007). The role of
sodium dodecyl sulfate (SDS) micelles in inducing
skin barrier perturbation in the presence of glyc-
erol. Journal of Cosmetic Science 58, 109–133.
12. LiBrizzi, J.J., Walters, R.M., Fevola, M. & Tamar -
selvy, K. (2010). Low-irritation compositions and
methods of making the same. United States Patent
7,803,403, 28 September 2010.
13. Moore, P.N., Shiloach, A., Puvvada, S. & Blank -
schtein, D. (2003). Penetration of mixed micelles
into the epidermis: Effect of mixing sodium dodecyl
sulfate with dodecyl hexa(ethylene oxide). Journal
of Cosmetic Science 54, 143–159.
14. Moore, P.N., Puvvada, S. & Blankschtein, D. (2003).
Challenging the surfactant monomer skin penetra-
tion model: Penetration of sodium dodecyl sulfate
micelles into the epidermis. Journal of Cosmetic
Science 54, 29–46.
15. Mao, G., Flach, C.R., Mendelsohn, R. & Walters, R.M.
(2012). Imaging the distribution of sodium dodecyl
sulfate in skin by confocal Raman and infra red
microspectroscopy. Pharmaceutical Research 29,
2189–2201.
16. Saad, P., Flach, C.R., Walters, R.M. & Mendelsohn,
R. (2012). Infrared spectroscopic studies of sodium
dodecyl sulphate permeation and interaction with
stratum corneum lipids in skin. International
Journal of Cosmetic Science 34, 36–43.
Skin irritation potential of cleansers 531
17. de Jongh, C.M., Verberk, M.M., Spiekstra, S.W.,
Gibbs, S. & Kezic, S. (2007). Cytokines at different
stratum corneum levels in normal and sodium lau-
ryl sulphate-irritated skin. Skin Research &
Technology 13, 390–398.
18. Lévêque, J.L., de Rigal, J., Saint-Léger, D. & Billy,
D. (1993). How does sodium lauryl sulfate alter the
skin barrier function in man? A multiparametric
approach. Skin Pharmacology 6, 111–115.
19. van der Valk, P.G., Nater, J.P. & Bleumink, E.
(1984). Skin irritancy of surfactants as assessed by
water vapor loss measurements. Journal of Invest -
igative Dermatology 82, 291–293.
20. Cannon, C.L., Neal, P.J., Southee, J.A., Kubilus, J. &
Klausner, M. (1994). New epidermal model for der-
mal irritancy testing. Toxicology in Vitro 8, 889–891.
21. Berridge, M.V., Tan, A.S., McCoy, K.D. & Wang, R.
(1996). The biochemical and cellular basis of cell
proliferation assays that use tetrazolium salts.
Biochemica 4, 14–19.
22. Pappas, A., Fantasia, J. & Chen, T. (2013). Age and
ethnic variations in sebaceous lipids. Dermato-
Endocrinology 5, 319–324.
23. Walters, R.M., Fevola, M.J., LiBrizzi, J.J. & Martin,
K. (2008). Designing cleansers for the unique needs
of baby skin. Cosmetics & Toiletries 123, 53–60.
24. Association of Women’s Health Obstetric and Neo -
natal Nurses (2013). Neonatal Skin Care: Evidence-
based Clinical Practice Guideline, 3rd edn, 81pp.
Washington, DC, USA: Association of Women’s
Health, Obstetric and Neonatal Nurses.
532 R.M. Walters et al.
