Dermal penetration and metabolism of five glycidyl ethers in human, rat and mouse skin

Article abstract of DOI:10.1080/004982500237488

1. Glycidyl ethers (GE), an important class of industrial chemicals, are considered to be potentially mutagenic in vivo because some GE have been shown to be direct mutagens in short-term in vitro tests. 2. The percutaneous penetration and metabolism of representatives of different classes of GE was studied in the fresh, full-thickness C3H mouse, and dermatomed human and Fisher 344 rat skin to determine th apparent permeability constants, lag times and metabolic profiles. 3. Five different GE, the diglycidyl ethers of bisphenol A (BADGE), 4,4'-dihydroxy-3,3',5,5'-tetramethylbiphenyl (Epikote YX4000) and 1,6-hexanediol (HDDGE) and the GE of 1-dodecanol (C₁₂GE) and o-cresol (o-CGE), were synthesized by reaction of their alcohols with epichlorohydrin. Their radiolabelled analogues were synthesized with a ¹⁴C-label using [U-¹⁴C]-epichlorohydrin. 4. There was a large variation (four orders of magnitude) in percutaneous penetration between the five GE. In general, penetration through full-thickness mouse skin was higher than through dermatomed rat skin, whereas dermatomed human skin was the least permeable. The permeability increased in the order YX4000 < BADGE < C₁₂GE < o-CGE < HDDGE. 5. The relative skin permeability of the five GE could be explained for a significant part by the lipophilicity, expressed as log P(o/w), in combination with the molecular weight of the compounds. 6. During skin penetration, all GE were extensively metabolized to their corresponding (bis-)diols. Virtually no YX4000, and only very small amounts of C₁₂GE and o-CGE penetrated the skin unchanged. For o-CGE, but none of the other GE, the percentage of the applied dose that penetrated the skin unchanged increased over time. 7. The large variation in response observed with the five selected GE indicates that GE should not be considered as a single class of compounds but rather on the basis of their individual properties.

Full text of DOI:10.1080/004982500237488

xenobiotica
vol
no
. 30, . 5, 469±483
, 2000,
Dermal penetration and metabolism of Ž ve glycidyl
ethers in human, rat and mouse skin
P. J. BOOGAARD*, M. A. DENNEMAN
and N. J. VAN SITTERT
Department of Molecular Toxicology, Shell Research and Technology Center,
Amsterdam, Shell International Chemicals BV, PO Box 38000, 1030 BN Amsterdam,
The Netherlands
Received 23 July 1999
1. Glycidyl ethers (GE), an important class of industrial chemicals, are considered to be
potentially mutagenic in vivo because some GE have been shown to be direct mutagens in
-
short term in vitro tests.
å
2. The percutaneous penetration and metabolism of representatives of di erent classes
-
of GE was studied in the fresh, full thickness C3H mouse, and dermatomed human and
Fisher 344 rat skin to determine the apparent permeability constants, lag times and
metabolic proŽ les.
å
-
-
3. Five di erent GE, the diglycidyl ethers of bisphenol A (BADGE), 4,49 dihydroxy
-
-
3,39,5,59 tetramethylbiphenyl (Epikote YX4000) and 1,6 hexanediol (HDDGE) and the
GE of 1 dodecanol (C GE) and o cresol (o CGE), were synthesized by reaction of their
"#
alcohols with epichlorohydrin. Their radiolabelled analogues were synthesized with a
-
-
-
"%
-
C
"%
-
-
label using [U C] epichlorohydrin.
4. There was a large variation (four orders of magnitude) in percutaneous penetration
-
between the Ž ve GE. In general, penetration through full thickness mouse skin was higher
than through dermatomed rat skin, whereas dermatomed human skin was the least
!
!
!
-
GE o
permeable. The permeability increased in the order YX4000 BADGE
CGE HDDGE.
C
"#
!
5. The relative skin permeability of the Ž ve GE could be explained for a signiŽ cant part
by the lipophilicity, expressed as log Po/w , in combination with the molecular weight of the
compounds.
6. During skin penetration, all GE were extensively metabolized to their corresponding
-
(bis )diols. Virtually no YX4000, and only very small amounts of C GE and BADGE,
"#
-
penetrated the skin unchanged, but signiŽ cant amounts of HDDGE and o CGE penetrated
-
the skin unchanged. For o CGE, but none of the other GE, the percentage of the applied
dose that penetrated the skin unchanged increased over time.
7. The large variation in response observed with the Ž ve selected GE indicates that GE
should not be considered as a single class of compounds but rather on the basis of their
individual properties.
Introduction
Glycidyl ethers (GE) form an important and highly diverse group of industrial
chemicals used for manufacturing epoxy resin systems applied in adhesives,
coatings, plastics, electronics and structural composites. Phenyl glycidyl ether
(PGE) and resorcinol diglycidyl ether have been reported as causing tumours in
animal studies and consequently have been classiŽ ed by the IARC as group 2B
carcinogens (possibly carcinogenic to humans) (IARC 1985, 1989, 1999a). Some
other glycidyl ethers, such as BADGE and triethylene glycol diglycidyl ether, were
considered not classiŽ able as to their carcinogenicity to humans (group 3) since only
limited or inadequate evidence in experimental animals and no epidemiological data
in humans relevant to their carcinogenicity were available (IARC 1989, 1999c, d).
-
* Author for correspondence; e mail: Peter.J.Boogaard!opc.shell.com
}
-
-
ISSN 0049 8254 print ISSN 1366 5928 online ’ 2000 Taylor & Francis Ltd
Xenobiotica
}}
}
http: www.tandf.co.uk journals

P J Boogaard
. .
470
et al.
Figure 1. Structures of synthesized chemicals with their molecular weights (MW) and logarithm of the
}
æ
octanol water partition coe cient (log Po/w ). *Position of radiolabels.
Nevertheless, glycidyl ethers as a group are under scrutiny, because several
epoxy compounds have been associated with mutagenicity and cancer. It is indeed
likely that the epoxy group is responsible for the direct mutagenic eåects observed
et al
in bacteria exposed to glycidyl ethers (Gardiner
. 1992) and for the DNA
in vitro
et al
et al
. 1998). GE are
adducts observed in
systems (Plna
. 1996, Deforce
often used as reactive diluents and as components of paints and glues. Most of the
GE have a low volatility, and as a consequence one of the primary portals of entry
for GE is the skin through direct contact with the compounds. Any systemic but also
any topical eåects of GE will be highly dependent on the extent of percutaneous
penetration and metabolism. Both penetration and metabolism of a particular
compound are largely determined by its structure and physical chemical properties
that vary widely among the GE.
The main objective of the current study was to compare the percutaneous
absorption and metabolism of a series of selected GE (Ž gure 1) in viable skin
explants of dorsal skin of rat and mouse and human breast skin. Based on previous
experience with subcellular fractions from skin and viable skin, it was decided to
study dermal metabolism in viable skin only since it appeared that viable skin gave

Glycidyl ethers and the skin
471
in vivo
et al
. 1999).
a much better re ection of the
same individual (Boogaard
situation than subcellular fractions from the
Materials and methods
Chemicals
"
}
"%
-
-
-
[U C] epichlorohydrin, radiochemical purity 98%, speciŽ c activity 38 mCi mmol, was custom
"%
-
-
-
-
synthesized by ChemSyn Laboratories (Lenexa, KS, USA). [Ring U C] bis phenol A diglycidyl ether
(radiochemical purity 98% ; speciŽ c activity 11 mCi mmol) was a gift from Dr John Waechter (Dow
Chemical Co., Midland, MI, USA). Recrystallized BADGE (chemical purity 99%) and its bis diol
"
}
"
-
"
-
(bis phenol A diglycerol ether) (chemical purity 96%) were from the Department of Basic &
Exploratory Research (Shell Research & Technology Centre, Amsterdam). Gentamicin solution, Hanks’
å
bu ered salts and bovine serum albumin (fraction V) (BSA) were from Sigma Chemical Co. (St Louis,
-
-
MO, USA). 4 Phenylchalcone oxide (4 PCO) was from ICN Biomedicals BV (Zoetermeer, The
Netherlands); its chemical purity was 99%. Tetraethylammonium hydroxide (20% aqueous solution)
was from E. Merck (Darmstadt, Germany). All other chemicals used were from Aldrich and were at least
"
reagent grade with a purity 95%. Solvents used for HPLC were HPLC grade.
Synthesis of reference materials (Ž gure 1)
-
-
The glycidyl and glycerol ethers of o cresol and 1 dodecanol as well as the diglycidyl and diglycerol
-
-
-
-
ethers of 1,6 hexanediol, 4,49 dihydroxy 3,39,5,59 tetramethylbiphenyl and diphenylol propane were
"%
-
synthesized from the corresponding alcohols and epichlorohydrin (ECH). The C labelled analogues
were synthesized from the same alcohols and [U C] epichlorohydrin, diluted with non labelled ECH
to a speciŽ c activity of 0.2 mCi mmol. The chemical and radiochemical purities for the (di)glycidyl
ethers were: YX4000 97% and 98% ; HDDGE 92% and 98% ; C GE 95% and 98% ;
"#
o CGE 94% and 92% respectively. The chemical purities for the (di)glycerol ethers were: YX4000
"%
-
-
-
}
"
"
"
"
"
"
"
"
"
-
"
"
"
-
97% ; HDDGE 93% ; C GE 90% ; o CGE 85%. Details of synthesis and analysis are given
in Van Elburg et al. (1999).
"#
Liquid scintillation counting (LSC)
Radioactivity in solutions was measured using TriCarb 2200CA liquid scintillation counters
(Canberra Packard, Groningen, The Netherlands) in antistatic scintillation vials after mixing with 5 10
±
vols Ultima Gold Scintillation cocktail (Canberra Packard). The machine was calibrated using
$
"%
commercial H and C internal standard kits for organic solvents (Wallac, Turku, Finland). Calibration
was checked daily by counting a set of quenched standards commercially prepared in sealed glass vials.
æ
Counting e ciency was determined using the spectral index of the internal standard (SIE) and cpm was
automatically transformed to dpm. Samples were corrected for background and counted for periods
!
æ
su ciently long to reach a relative standard deviation of 0.02.
-
High performance liquid chromatography (HPLC)
HPLC analyses were carried out using various systems. For analysis of radiolabelled compounds, a
-
Ramona D or 2000 on line radioactivity detector (Raytest, Straubenhardt, Germany), both equipped
with solid scintillant cells, were used. In addition, because the total radioactivity in individual peaks was
-
in most cases below the limit of quantitation of the on line detectors, fractions were collected using LKB
fraction collectors. The radioactivity in the collected fractions was determined by LSC. HPLC analyses
were carried out using the following equipment and conditions:
System A. Apparatus: HP1050 liquid chromatograph with UV detection (k 230 nm) (Hewlett
5
3
Packard, Amstelveen, The Netherlands); column: Phenomenex Jupiter 300 A, 5 lm ODS, 250 4.6 mm
-
-
(Phenomenex, Torrance, CA, USA) with HiChrom security guard column NC120 5C18 10C
}
(HiChrom, Reading, UK); eluent: puriŽ ed water (A) and methanol (B);  ow: 1.0 ml min; gradient:
0 18 min, linear increase from 25 to 92.5% B; 18 20 min, 92.5% B; 20 22 min: linear increase from
±
±
±
92.5 to 100% B, 30 35 min linear decrease from 100 to 25% B.
±
System B. Apparatus: HP1100 liquid chromatograph with UV detection (k 230 nm) (Hewlett
5
3
Packard); column: Phenomenex Jupiter 300 A, 5 lm ODS, 250 4.6 mm, with Phenomenex Security
3
Guard (widepore 4 3.0 mm ODS cartridge); eluent: puriŽ ed water (A) and acetonitrile (B);  ow:
}
1.0 ml min; gradient: 0 31 min, linear increase from 25 to 48.5 % B; 31 35 min, linear increase from
±
±
48.5% to 100% B; 35 40 min, linear decrease from 100 to 25% B; 40 45 min 25 % B.
±
±
å
System C. As for system B, but with a di erent gradient: 0 15 min, linear increase from 0 to 30% B;
±
15 25 min, linear increase from 30% to 45% B; 25 30 min, linear increase from 45 to 100% B;
±
±
30 33 min, 100% B; 33 35 min linear decrease from 100 to 0% B.
±
±

P J Boogaard
. .
472
et al.
-
-
Liquid chromatography mass spectrometry (LC MS)
-
LC MS analyses were performed using system A on a HP 1050 liquid chromatograph coupled to a
-
Micromass Quattro quadrupole mass spectrometer.Electrospray was used as the LC MS interfacing and
}
ionization technique (positive ions). The scan range was 100 1500 Dalton with 0.5 scan s.
±
Human skin
-
Fresh full thickness healthy human breast skin was from Ž ve female Caucasian patients, aged 17 41
±
years, undergoing mammoplastic reduction surgery in a regional hospital (MCA, Alkmaar, The
Netherlands). The skin was placed on ice in a Petri dish and transported to the laboratory where it was
used within 3 h from the time it was surgically removed. The use of human tissue in the present studies
was according to the guidelines of the 1984 Declaration of Helsinki. Patients gave written, informed
consent and the hospital’s medical ethical committee approved the use of viable human tissue for this
±
study.
Rodent skin
C
C
Male Fisher 344 rats ( 250 g) and male C3H mice ( 25 g) were from the Central Laboratories for
the Blood Banks (CLB) (Amsterdam, The Netherlands). Animals were kept on a 12 h light dark cycle in
}
-
-
-
-
humidity (60 15% relative humidity) and temperature (22 2 C) controlled mass air displacement
Ê
6
6
-
rooms. A standard rodent diet (VRF I, Broekman Instituut [Charles River Laboratories], Someren, The
-
Netherlands) and de ionized water were supplied ad libitum. Following an acclimation of 1 week, the
animals were used as the source for viable rodent skin. Animals were killed with a lethal dose of sodium
pentobarbital, and the dorsal region shaved. The skin was removed and placed in a Petri dish on ice until
used (within 2 h from removal).
Skin preparation
å
Skin penetration and metabolism was studied using viable skin. Three di erent skin donors (both for
human and animal skin) were used for each of the Ž ve GE under study. Human skin was dissected from
the tissue removedduring mammoplasticreduction surgery using a scalpel and cleanedwith physiological
saline. The shaved area from the rodent skin was cut out with dissecting scissors. Excess subcutaneous
tissue was removed from the skin samples with a scalpel. The skin was placed ona Te ondissecting board
-
and full thickness skin circles were cut out using a circular sharpened steel cutter. Mouse skin was always
used at full thickness since it is so thin that it can not be dermatomed. Human and rat skin was only used
following dermatomization. The skin, epidermis up, was stretched over a small band of Te on and a sheet
å
-
of skin was cut o using a hand held keratotome (Storz Instrument Co., US). Skin circles were cut out
of this sheet using small dissection scissors. The gross integrity of the skin was assessed visually. Skin
membranes were rejected if very high quantities of the applied dose appeared in the receptor  uid
immediately after application. Previous studies have shown that the enzymes essential for inactivation of
glycidyl compounds are present and active in skin membranes prepared with this procedure (Boogaard
et al. 1999).
å
Di usion system
-
å
The skin was mounted, epidermis uppermost, in a  ow through di usion cell system that has been
described (Boogaard et al. 1999) and which is based on the system developed by Bronaugh and Stewart
å
(1984) and Hotchkiss et al. (1990). The system consisted of (1) 12 stainless steel di usion cells, which
were stirred continuously with magnetic stirrer bars (Variomag HP 15 Multipoint, HP Labortechnik,
Munich, Germany) to prevent irregular penetration due to air bubbles underneath the skin membrane,
(2) a fraction collector holding 20 ml glass scintillation vials, (3) a thermostatically controlled water
circulator (PolyScience, USA) to heat the jackets holding the cells in such a way that the skin surface
temperature was kept at 32 1 C, and (4) a cassette peristaltic pump (Watson Marlow, Falmouth, UK).
Ê
6
#
The exposed skin surface was 0.636 cm . The underside of the skin was exposed to continuously  owing
C
}
( ow rate 1.5 ml h) receptor  uid consisting of Hanks’ balanced salt solution, without phenol red and
}
å
sodium bicarbonate, bu ered with 25 mm HEPES (pH 7.40) and containing 2.5% w v BSA and 50 lg
gentamicin per ml. The receptor  uid was transported to the cells and from the cells, through the pump,
to the fraction collector through PVC tubing (Accurated tubing, Elkay). In some experiments the cells
were occluded by placing a polycarbonate lid over the cells.
Dosing and determination of penetration
-
-
In each experiment, which was carried out in triplicate, three to four cells with skin from the same
donor (human or animal) were used to study the penetration of each of the GE under study.
Aliquots of 5.0 ll GE dissolved in acetone at a concentration of 636 mm were applied to the exposed
}
#
skin surface using a micropipette (Ž nal concentration 5.0 lmol cm ). The receptor  uid was collected in

Glycidyl ethers and the skin
473
-
3 hourly fractions for 24 h following application of the GE. The fractions were weighed for determination
of the actual  ow rates and the radioactivity in an aliquot of the fractions was determined by LSC. The
remaining portions of the collected fractions were extracted twice with a double volume of ethyl acetate.
An aliquot of DMSO was added to the combined fractions and a gentle stream of nitrogen evaporated the
ethyl acetate. The DMSO residues were mixed with an equal volume of methanol and subsequently
-
analysed by HPLC with radiodetection. For studies with o CGE and C GE system A was used, for
"#
BADGE and YX4000 system B was used, for HDDGE system C was used. The (radioactive) products
-
-
formed were identiŽ ed by co elution with reference standards and by LC MS.
At the end of the experiment the skin was removed from the cell and soaked twice for 12 h in 2 ml ethyl
acetate to dissolve GE remaining on the skin. The combined ethyl acetate fractions were dissolved in
Ultima Gold scintillation cocktail and the radioactivity determined by LSC. The skin itself was dissolved
over 48 h in 1.0 ml tetraethylammonium hydroxide and the total radioactivity in the resulting solution
was determined by LSC. The receptor  uid remaining in the cell and the tubing was collected; the tubing
was subsequently  ushed with receptor  uid and water. The combined  uids were counted by LSC.
Finally, to obtain a mass balance, the complete cells (including supportive grids and stirrer bars) were
individually sonicated in 100 ml ethyl acetate for 30 min to dissolve any traces of the GE, and an aliquot
of the ethyl acetate was counted by LSC.
}
æ
Estimation of the octanol water partition coe cient (P
)
/
o w
"
Compounds with a log P /
2 may accumulate in the fatty compartment of the dermis. To study the
skin permeation of such compounds and avoid the dermis acting as a reservoir, the skin of rat and
humans, which have a relatively thick dermis, should be dermatomed. For BADGE, a log P / 4.0 was
o w
o w 5
determined by HPLC (Eadsforth 1983). For the other GE, however, no log Po/w has been determined. For
these GE log Po/wwas estimated using the KOWWIN software package, v.1.57 (Syracuse Research
Software, Whitstable, UK). The following were determined: YX4000 log P /
5.2, C GE log P /
5
"#
5
o w
o w
-
5.0, BADGE log P /
3.8, o CGE log P /
2.2 and HDDGE log P /
0.84.
o w 5
o w 5
o w 5
Calculations and statistics
Multiple linear regression analyses were performed using the LINEST function of the Excel for
Windows 95 software package (v.7.0a; Microsoft Co., Redmond, WA, USA).
Results
Percutaneous penetration
Percutaneous penetration varies considerably between the various GE and also
between the three species. In Ž gure 2A±E the actual dermal penetration, as
percentage of the applied dose, for the Ž ve GE through the skin is depicted. The
mass balances show that the recovery of YX4000 and BADGE is complete, but that
o-
there is a small loss of C GE and HDDGE and a signiŽ cant loss of CGE (table 1).
"#
The incomplete mass balance is due to the volatility of the GE: whereas YX4000and
BADGE have virtually no vapour pressure and C GE and HDDGE have a low
"#
o-
vapour pressure, CGE is relatively volatile. The skin permeation of YX4000,
BADGE and C GE is linear with time over the entire 24 h for human, rat and
"#
o-
mouse skin. For CGE and HDDGE a substantial deviation from linearity was
C
o-
observed after 6 h in rat and mouse skin. In human skin the absorption of GCE
also deviated from linearity, but the absorption of HDDGE through the human skin
o-
was linear with time over the entire 24 h. With CGE, the dermal penetration
C
reached a plateau at absorption of
10, 18 and 23% of the applied dose with
human, rat and mouse skin respectively. In all three cases, the curves started
C
levelling oå after 12 h. Based on the mass balance and the relatively high volatility
o-
o-
of CGE, it is assumed that at this point nearly all CGE had evaporated from the
skin. For HDDGE, the curves for rat and mouse skin also seemed to reach a plateau,
but at a later point in time and at signiŽ cantly higher absorption levels (65±80% of
applied dose). Occlusion of the skin signiŽ cantly enhanced the dermal penetration of
o-
CGE through human, rat as well as mouse skin: the percentage of the applied dose

P J Boogaard
. .
474
et al.
(a)
(b)
(c)
(d)
(e)
Figure 2. Penetration of glycidyl ethers through viable skin. Human and rat skin were used
-
dermatomed, whereas mouse skin was used at full thickness (undermatomed). Data are averages
å
of three separate experiments with skin from di erent human or animal donors (each experiment

Glycidyl ethers and the skin
475
Table 1. Mass balance of skin permeation experiments percentage of applied dose of GE recovered
from the surface of the skin, from the skin itself, in the receptor  uid and from the percutaneous
penetration equipment.
-
Species
YX4000
99.6 0.6
BADGE
102 10
C
GE
o CGE
HDDGE
83 13
"#
Human
Rat
Mouse
70.9 0.6
88
70
13
22
26
3
3
5
6
6
6
6
6
6
6
105
104
5
8
98 11
6
3
90
89 10
6
6
6
6
6
6
6
98
8
6
6
Data are means SD of triplicate determinations.
6
(a)
(b)
-
Figure 3. Penetration of glycidyl ethers through occluded and non occluded viable skin. Human and rat
-
skin were used dermatomed, whereas mouse skin was used at full thickness (undermatomed).
-
Data for non occluded skin are identical to those in Ž gure 2, but without error bars for clarity.
Mouse skin U, ; rat skin + , * ; human skin, E , D . Filled symbols indicate occluded;
V
-
-
-
open symbols indicate non occluded skin. (a) o CGE (o cresyl glycidyl ether). (b) HDDGE
-
(1,6 hexanediol diglycidyl ether).
that had penetrated through the skin was roughly doubled by occlusion (Ž gure 3A).
In contrast, the dermal penetration of HDDGE was not signiŽ cantly in uenced by
o-
occlusion of the skin (Ž gure 3B). The data on the occlusion of CGE and HDDGE
indicate that the increased dermal penetration is mainly due to diminished
volatilization of the GE and not to alterations of the skin, such as increased
hydratation, caused by the occlusion.
-
-
with three to four separate cells). Values are means SE (n 3). Mouse skin, ; rat skin, * ; D ;
V
6
5
-
-
-
human skin. (a) YX4000 (4,49 dihydroxy 3,39,5,59 tetramethylbiphenyl diglycidyl ether). Lines
are obtained by linear regression analysis on all data points; R
human, rat and mouse skin respectively. (b) BADGE (bisphenol A diglycidyl ether). Lines are
obtained by linear regression analysis on all data points; R
human, rat and mouse skin respectively. (c) C GE (1 dodecyl glycidyl ether). Lines are obtained
by linear regression analysis on all data points; R
mouse skin respectively. (d) o CGE (o cresyl glycidyl ether). Lines are obtained by extrapolation
#
0.9977, 0.9999 and 0.9992 for
5
#
0.9916, 0.9996 and 0.9997 for
5
-
"#
#
0.9958, 0.9996 and 0.9993 for human, rat and
5
-
-
-
of the linear part of the permeation curves. (e) HDDGE (1,6 hexanediol diglycidyl ether). The
line representing the penetration through human skin is obtained by linear regression analysis on
#
all data points; R
0.9999. Lines representing the penetration through rodent skin are obtained
5
by extrapolation of the initial, linear part of the penetration curve.

P J Boogaard
. . et al.
476
±
Table 2. Apparent permeability constanta (in 10 cm h) in dermatomed human and rat skin and full
thickness mouse skin of Ž ve glycidyl ethers.
’
}
-
-
Species
YX4000
0.047 0.02
BADGE
0.48 0.22
C
GE
o CGE
HDDGE
136 28
402 22
6
577 86
6
"#
Human
Rat
Mouse
3.31 0.19
29.0 7.3
46.1 17.2
93 11
134 15
176 32
6
6
6
6
6
6
6
6
6
0.98 0.02
5.5 1.5
6
6
2.63 0.55
8.6 1.6
6
6
Data are means SE of triplicate determinations.
A Ž nite amount of the glycidyl ethers, 5 lmol cm , was applied to the skin surface, therefore, P is
6
a
}
#
the apparent permeability constant.
Table 3. Percutaneous penetration (as percentageof the applied dose) through dermatomedhuman and
-
rat skin and full thickness mouse skin of Ž ve glycidyl ethers over 24 h
-
Species
YX4000
0.011 0.003
BADGE
0.137 0.055
C
GE
o CGE
HDDGE
37.8 8.5
67.2 0.5
6
80.3 8.3
6
"#
Human
Rat
Mouse
0.870 0.065
7.73 2.09
10.97 3.63
10.2 1.6
17.4 1.8
22.7 4.6
6
6
6
6
6
6
6
6
6
0.275 0.010
1.57 0.41
6
6
0.726 0.113
2.98 0.80
6
6
Data are means SE of triplicate determinations.
6
Data represent total radioactivity, i.e. parent compounds plus metabolites.
Table 4. Lag time (h) in dermal penetration of Ž ve glycidyl ethers through dermatomed human and rat
-
skin and full thickness mouse skin.
-
Species
YX4000
2.4 1.5
BADGE
6.1 1.7
C
GE
o CGE
HDDGE
0.92 0.18
"#
HumanHuman
Rat
Mouse
3.8 1.6
3.0 0.6
5.3 0.7
0.96 0.23
0.36 0.26
0.51 0.47
6
6
6
6
6
6
6
6
6
1.87 0.09
1.8 0.5
0.40 0.18
6
6
6
0.92 0.50
2.3 1.0
0.62 0.23
6
6
6
Data are means SE of triplicate determinations.
6
æ
Table 5. Correlation between the logarithm of the apparent permeability coe cient (log P) of Ž ve
-
glycidyl ethers (YX4000, BADGE, C GE, o CGE and HDDGE) and their intrinsic physical
"#
}
æ
parameters molecular weight (MW) and logarithm of the octanol water partition coe cient (log
Po/w ).
æ
Regression coe cient ( SE) for
6
Regression constant
SE)
Species
MW
log P /
(
R
6
o w
Human
Rat
Mouse
0.386 0.114
0.0113 0.0026
4.91 0.58
0.987
0.957
0.955
0.880
2
2
2
2
2
2
2
2
6
6
6
0.283 0.145
0.0073 0.0034
4.30 0.74
6
6
6
0.261 0.134
0.0065 0.0031
4.22 0.69
6
6
6
All
0.310 0.118
0.0084 0.0027
4.48 0.61
6
6
6
"
-
}
The apparent permeability constant, in non occluded skin, varies over
4
orders of magnitude, from 4.7 (SE 0.2)3 10± cm h, for YX4000 in human skin, to
)
5.8 (SE 0.9)3 10± cm h, for HDDGE in mouse skin (table 2). Regardless of
%
}
!
!
!
o-
species, the permeability increased in the order: YX4000 BADGE C GE
"#
!
-
-
CGE HDDGE. For all GE, full thickness mouse skin had the highest per
meability and human dermatomed skin was the least permeable (tables 2 and 3). The
lag time is longest in human skin for all GE, with the exception of C GE, where the
"#
lag time in mouse skin is longer (table 4). Rather poor linear correlations were

Glycidyl ethers and the skin
477
æ
Table 6. Correlation between the logarithm of the apparent permeability coe cient (log P) of Ž ve
-
glycidyl ethers (YX4000, BADGE, C GE, o CGE and HDDGE) and their intrinsic physical
"#
parameters square root of molecular weight ( MW) and logarithm of the octanol water partition
}
o
æ
coe cient (log Po/w ).
æ
Regression coe cient ( SE) for
6
Regression constant
SE)
Species
MW
log P /
(
R
o w
6
Human
Rat
Mouse
0.396 0.136
0.351 0.102
7.62 1.42
0.981
0.948
0.946
0.872
2
2
2
2
2
2
2
2
6
6
6
0.293 0.159
0.223 0.120
5.99 1.67
6
6
6
0.271 0.147
0.197 0.111
5.71 1.54
6
6
6
All
0.320 0.122
0.257 0.092
6.44 1.28
6
6
6
(a)
(b)
Figure 4. Correlation between the logarithm of the apparent permeability constant (log P) and the
}
æ
logarithm of the octanol water partition coe cient (log Po/w ) for the aromatic glycidyl ethers (A)
and the bifunctional glycidyl ethers (B). Log P was determined experimentally (table 2) and log
P / was estimated (the Materials and methods) Symbols are the log P determined in viable
o w
dermatomed human skin (rat skin (E ) or undermatomed mouse skin (+). (A) Aromatic glycidyl
-
ethers: o CGE, YX4000 and BADGE. Lines are obtained by linear regression analysis;
#
R
0.975, 0.988 and 0.967 for human, rat and mouse skin respectively. (B) Bifunctional glycidyl
ethers: YX4000, BADGE and HDDGE. Lines are obtained by linear regression analysis;
5
#
R
0.946, 0.922 and 0.958 for human, rat and mouse skin respectively.
5
P
obtained by plotting the logarithm of the apparent permeability constant (log ) for
#
P
R
the 5 GE against their log
(
0.75, 0.72 and 0.73 for human, rat and mouse
/
o w
5
#
R
respectively) or their molecular weight (MW) (
0.83, 0.75 and 0.76 for human,
5
rat and mouse respectively). MW gave slightly better correlations than the square
root of MW ( MW) (tables 5 and 6). In contrast, highly signiŽ cant linear
o
P
P
/
o w
correlations were found between the log and log
if only the bifunctional GE
or only the aromatic GE were considered (Ž gure 4). Multiple regression analysis
P
with the log obtained for all Ž ve GE under study as dependent variables and their
P
intrinsic physical properties log
and either MW (table 5) or MW (table 6) as
/
o w
o
independent variables showed statistically highly signiŽ cant correlations for each of
the species separately as well as for all data grouped together.

P J Boogaard
. .
478
et al.
(a)
(b)
Figure 5. For legend see facing page.
Percutaneous metabolism
Extensive metabolism of all Ž ve GE was observed in the skin. As can be seen in
bis-
Ž gure 5A and B, YX4000 and BADGE are almost completely converted to their
diol during absorption. Note that the relative metabolic conversion in Ž gure 5 is
given as a percentage of the dose penetrated, which is more than an order of
magnitude lower in humans than in rodents. Therefore, the absolute amount of
YX4000, BADGE or C GE that penetrates the skin unchanged is much lower for
"#
human skin than either rat or mouse skin. The major part of the limited amount of
bis-
YX4000 that penetrates the skin is metabolized to the
diol (77 2, 87.9 0.2 and
6 6
80 4 % of the penetrated dose for human, rat and mouse skin respectively) and
6
-
only a minor portion of the penetrated material is present as the mono diol (20 1,
6
5.1 0.3 and 10 4% of the penetrated dose for human, rat and mouse skin
6
6
respectively). The percentage of the penetrated dose that permeated the skin as

Glycidyl ethers and the skin
479
(c)
(d)
Figure 5. Hydrolysis of glycidyl ethers in viable skin. Human and rat skin were used dermatomed,
-
whereas mouse skin was used at full thickness (undermatomed). Bars are means SE (n 3) of
6
5
the metabolic conversion of each glycidyl ether to its corresponding diol, and for bifunctional
-
glycidyl ethers also their bis diols, as a percentage of the total penetrated dose in 24 h. (a) YX4000
-
-
-
(4,49 dihydroxy 3,39,5,59 tetramethylbiphenyl diglycidyl ether). (b) BADGE (bisphenol
A
-
-
diglycidyl ether). (c) C GE (1 dodecyl glycidyl ether). (d) HDDGE (1,6 hexanediol diglycidyl
ether).
"#
YX4000 was only 2.1 0.1, 0.67 0.03 and 3.9 1.7% in human, rat and mouse
6
6
6
respectively. For BADGE very similar results were found: 79 8, 92.9 0.1 and
6
6
bis-
diol in human, rat and
96 1% of the penetrated dose was converted to the
6
mouse skin respectively, 10 5, 3.8 0.1 and 2.0 0.2 % of the penetrated dose was
6
6
6
-
hydrolysed to the mono diol in human, rat and mouse skin respectively, and
1.1 0.7, 0.29 0.02 and 0.15 0.07 % of the penetrated dose was recovered as
6
6
6
BADGE itself in human, rat and mouse skin. For the third bifunctional GE under
study, HDDGE, a very diåerent picture was observed. First, HDDGE penetrates

P J Boogaard
. .
480
et al.
-
Figure 6. Hydrolysis of o cresyl glycidyl ether in viable skin over time. Human and rat skin were used
-
dermatomed, whereas mouse skin was used at full thickness (undermatomed). Bars are
-
means SE (n 3) of the metabolic conversion of o cresyl glycidyl ether to mephenesin (the diol)
6
5
as a percentage of the total amount permeated during the 24 h of the experiment.
the skin rapidly and eæciently and, second, only a minor portion of the penetrated
bis-
amount is fully hydrolysed to the
diol. Especially in rat only very little
metabolism takes place and 78 1 % of the absorbed dose penetrates through the
6
bis-
skin unaltered. Virtually no
diol (0.41 0.25% of the absorbed dose) was formed
6
in rat skin during penetration. Mouse and human are very similar with 13.4 0.3
6
and 20 1% of the absorbed dose penetrating unchanged respectively, and
6
-
bis-
79.8 0.7 and 72 3% as the mono diol and 4.8 0.4 and 4.5 1.0% as the
diol
6
6
6
6
respectively. C GE was almost completely hydrolysed with 86 7, 95.1 0.7 and
6
6
"#
96.5 0.3% of the absorbed dose being recoveredas the diol and only 7 4, 2.8 0.4
6
6
6
and 2.1 0.2% of the absorbed dose as unchanged C GE in human, rat and mouse
6
"#
skin respectively. For YX4000, BADGE, C GE and HDDGE the relative amount
"#
of various hydrolysis products in the receptor  uid did not change during the 24 h
o-
of the study. For CGE, however, the picture was very dissimilar (Ž gure 6): a
considerable amount of the penetrated dose (86±88%), was hydrolysed to
o-
mephenesin (the glycerol ether of cresol) during the Ž rst hours of penetration in
o-
human, rat as well as mouse skin, but as the study progressed less CGE was
hydrolysed and more unchanged compound was recovered from the receptor  uid
(39 26, 45 15 and 29 8% of the penetrated dose from 20 to 24 h of the exposure
6
6
6
in human, rat and mouse skin respectively). This eåect was more pronounced in
o-
human skin than rodent skin, but the absolute amounts of CGE penetrating the
skin were lower in human skin than in either rat or mouse skin.
Discussion
In the present study, the percutaneous penetration and metabolism of selected
GE, representatives of diåerent classes of GE were investigated. Since considerable
exposure to GE may also occur through inhalation, either as aerosol or as vapour for

Glycidyl ethers and the skin
481
the more volatile GE, or through ingestion of inhaled GE or food stuås containing
GE, the metabolism of GE in subcellular fractions ofliver and lung were investigated
et al
. 2000).
in a separate study (Boogaard
-
-
The dermal penetration studies demonstrate substantial diåerences in per
cutaneous penetration of the various GE. In general, the penetration through full
thickness mouse skin is higher than through dermatomed rat skin, whereas
dermatomed human skin is the least permeable of all. The percutaneous penetration
!
!
!
of the various GE increases in the following order: YX4000 BADGE C GE
"#
!
o-
CGE HDDGE. The corresponding apparent skin permeability constants vary
±
"
’
}
over 4 orders of magnitude, from 0.0473 10 cm h for YX4000 through human
skin and HDDGE to 5773 10± cm h through mouse skin respectively. In general,
’
}
}
the skin permeation of a compound correlates with its octanol water partition
}
P
et al
. 1995). In
coeæcient ( o w ) and or with its molecular weight (MW) (Wilschut
/
P
/
o w
the current study, percutaneous penetration was indeed correlated with the log
as well as with the MW of the GE. These correlations were most clear if only the
aromatic or only the bifunctional GE were taken into account (Ž gure 6). A number
of mathematical models have been constructed to describe this correlation.
Compared with several other models, the relatively simple model by Guy and Potts
(1993) gave one of the best predictions for the skin permeation of a large set of
et al
. 1995). Using this model(which described the permeability
chemicals(Wilschut
constant as a linear function of log
P /
o w
and the square root of MW) a highly
signiŽ cant correlation for the 5 GE was found. Using MW directly, instead of its
square root, gave a slightly better correlation, which is probably due to the fact that
a relatively small and homogeneous group of compounds is being modelled here.
"
R
The correlations were very strong within one species (
0.95) but still highly
R
signiŽ cant if data for human and rodent skin were pooled (
0.88). These results
5
indicate that the skin penetration of other GE will probably be highly predictable as
well.
Most of the GE were eåectively metabolized during absorption through the skin.
Virtually no YX4000 penetrated human skin unchanged: only 0.00024% of the
applied dose permeated unchanged. Although still low, these percentages were
signiŽ cantly higher in rat and mouse skin: 0.0018 and 0.029% of the applied dose
of YX4000 respectively. For BADGE the amounts were very similar for rat and
mouse skin (0.0046 and 0.0045% of the applied dose) and substantially higher than
for human skin, through which only 0.0015% of the applied dose permeated
unchanged. Most of the absorbed YX4000 and BADGE was completely hydrolysed
bis-
in vivo
to the corresponding
knowledge the only
diol in human as well as rodent skin. To the best of our
studies on the metabolism of GE following dermal
et al et al
application were carried out with BADGE (Climie
. 1981a, b, Steiner
.
et al
1992a, b). Only the study by Climie
. presents data on the disposition of
BADGE following dermal occluded application in CF1 mouse. It was reported that
the total urinary and faecal excretion of radioactivity over the Ž rst 24 h following
C
occluded application of radiolabelled BADGE amounted to 4%, of which 99%
in
was metabolized to more polar metabolites. This matches well with the present
vitro
results, where 3.0 0.8% of BADGE penetrated the skin after 24 h unoccluded
6
bis-
application, of which 96 1 and 2.0 0.2% was converted to the
diol and the
6
6
-
mono diol respectively.
The picture seen with HDDGE, the third bifunctional glycidyl ether studied,
was very diåerent from that of YX4000 and BADGE. Quite a signiŽ cant amount of

P J Boogaard
. .
482
et al.
the applied dose penetrated the skin unchanged: 5.07, 52.4 and 15.9% of the applied
dose for human, rat and mouse skin respectively. The high percentage of HDDGE
that penetrated the rat skin unchanged was particularly striking and suggests that
there is relatively little metabolism of HDDGE in rat skin. This is, however, in
agreement with the observations in subcellular fractions of rat liver and lung where
signiŽ cantly less hydrolysis also occurred than in human or mouse fractions
et al
. 2000). Most of the HDDGE was hydrolysed during penetration to
(Boogaard
-
bis-
diol. For C GE, as for
"#
the mono diol and only a small percentage to the
BADGE, there was only a marginal diåerence in the percentage of the applied dose
that penetrated unchanged (0.23 and 0.21% for mouse and rat skin respectively)but
this amount was much larger than for human skin (only 0.057% of the applied dose
penetrated unchanged). Virtually all C GE was being hydrolysed to its diol during
"#
its penetration of the skin.
o-
Like HDDGE, CGE was extensively absorbed and, compared with the other
o-
GE, was not very extensively hydrolysed. In addition, the hydrolysis of CGE
diminished over time. For instance, in the Ž rst 9 h of the experiment only 0.69% of
the applied dose had permeated human skin unchanged, but in the following period
of 9 h this rose to 2.9% and in the last 5 h to 3.9%. In rodents the same phenomenon
was observed; in rat the percentage rose from 1.9% to 6.3 and 7.8% and in mouse
o-
from 2.6 to 4.1 and 6.7%. In addition, CGE evaporated from the skin, thus
o-
reducing the amount of CGE on the skin. Occlusion of the skin increased the
o-
dermal uptake of CGE about twice but had no signiŽ cant in uence on the extent
o-
of metabolism: the same percentage of CGE passed the skin unchanged. This
o-
suggests that CGE in uences its own metabolism over time. This phenomenon is
probably not due to an increased permeability of the skin, thus reducing the time for
metabolism but rather to an inhibitory eåect. The mass balance data (table 1) and
C
the shape of the permeation curve (Ž gure 2D) indicate that after 12 h virtually all
o- o-
CGE has evaporated from the surface of the skin. This means that most of the
CGE entering the receptor  uid in the last two periods over which metabolism was
studied (10±19and 20±24h; Ž gure 6) was already absorbed into the skin and released
o-
unchanged to the receptor  uid. This eåect might be due to CGE itself, inhibiting
its own metabolism for instance through alkylation of the active site of the epoxide
hydrolase (EH), or by mephenesin, for instance through allosteric inhibition or
competition for the binding site which would increase when the concentration of
o-
mephenesin in the skin increases. Elucidation of the exact mechanism by which
CGE in uences its own dermal metabolism requires additional research.
From the present study it can be concluded that there is a substantial variation
in the extent of dermal penetration of the GE studied. The permeability increases in
!
!
!
!
o-
CGE HDDGE. For all GE, full
-
the order: YX4000 BADGE C GE
"#
thickness mouse skin is more permeable than dermatomed rat skin and that human
skin is the least permeable of all. During skin penetration, all GE are extensively
bis-
metabolized to their corresponding (
)diols. The large variation in response
observed with the Ž ve selected GE indicates that GE should not be considered as a
single class of compounds but rather on the basis of their individual properties.
Acknowledgements
The authors are grateful to Drs Klaassens and Stevens, plastic surgeons at the
Medical Centre Alkmaar, for their kind and eæcient cooperation in the supplying of
fresh human skin samples.

Glycidyl ethers and the skin
483
References
Boogaard, P. J., Van Elburg, P. A., De Kloe, K. P., Watson , W. P. and van Sittert , N. J., 1999,
-
-
-
-
Metabolic inactivation of 2 oxiranomethyl 2 ethyl 2,5 dimethyl hexanoate (C GE) in skin,
"!
lungs, and liver of humans, rats and mice. Xenobiotica, 29, 987 1006.
±
Boogaard, P. J., De Kloe, K. P., Denneman , M. A., Bierau, J., Kuiken , G., Borkulo, P. E. D. and
van Sittert , N. J., 2000, Metabolic inactivation of Ž ve glycidyl ethers in lungs and liver of
humans, rats and mice. Xenobiotica, 30, 485 502.
±
Bronaugh , R. L. and Stewart, R. F., 1984. Methods for in vitro percutaneous absorption studies. IV:
å
the  ow through di usion cell. Journal of Pharmaceutical Science, 74, 64 67.
±
-
Climie , I. J. G., Hutson , D. H. and Stoydin , G., 1981a, Metabolism of the epoxy resin component 2,3
-
-
bis[4 (2,3 epoxypropoxy)phenyl]propane, the diglycidyl ether of bisphenol A (DGEBPA) in the
"%
-
mouse. I. A comparison of the fate of a single dermal application and of a single oral dose of
DGEBPA. Xenobiotica, 11, 391 399.
C
±
-
Climie , I. J. G, Hutson , D. H. and Stoydin , G., 1981b, Metabolism of the epoxy resin component 2,3
-
-
bis[4 (2,3 epoxypropoxy)phenyl]propane, the diglycidyl ether of bisphenol A (DGEBPA) in the
"%
-
C
mouse. II. IdentiŽ cation of metabolites in urine and faeces following of a single oral dose of
DGEBPA. Xenobiotica, 11, 401 424.
±
Deforce , D. L. D., Lemie re, F., Hoes, I., Millecamps , R. E. M., Esmans , E. L., De Leenheer , A. and
van den Eeckhout , E. G., 1988, Analysis of the DNA adducts of phenyl glycidyl ether in a calf
-
thymus DNA hydrolysate by capillary zone electrophoresis electrospray mass spectrometry:
evidence for phosphate alkylation. Carcinogenesis, 19, 1077 1086.
±
}
-
æ
Eadsforth , C. V., 1983, EPIKOTE 828: Determination of the N octanol Water Partition Coe cient
-
using a Reverse phase HPLC Method. Report no. SBGR 83.385 (Sittingbourne : Shell Research
Centre).
Gardiner , T. H., Waechter , J. M., Wiedow , M. A. and Solomon , W. T., 1992, Glycidyloxy
compounds used in epoxy resin systems: a toxicology review. Reviews in Toxicology and
Pharmacology, 15, S1 77.
±
Guy, R. H. and Potts , R. O., 1993, Penetration of industrial chemicals across the skin: a predictive
model. American Journal of Industrial Medicine, 23, 711 719.
±
Hotchkiss , S. A., Chidgey , M. A. J., Rose, S. and Caldwell , J., 1990, Percutaneous absorption of
benzyl acetate through rat skin in vitro. 1. Validation of an in vitro model against in vivo data. Food
and Chemical Toxicology, 28, 443 447.
±
IARC, 1985, Allyl Compounds, Aldehydes, Epoxides and Peroxides. Diglycidyl Resorcinol Ether (Lyons:
IARC), 36, 181 188.
±
IARC, 1989,Some Organic Solvents, Resin Monomers and Related Compounds, Pigments and Occupational
Exposures in Paint Manufacture and Painting. Some Glycidyl Ethers (Lyon: IARC), 47, 237 261.
±
-
IARC, 1999a, Re evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide. Phenyl
Glycidyl Ether (Lyon: IARC), 71, 1525.
-
IARC, 1999b, Re evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide. Diglycidyl
Resorcinol Ether (Lyon: IARC), 71, 1417.
-
IARC, 1999c, Re evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide. Bisphenol A
Diglycidyl Ether (Lyon: IARC), 71, 1265.
-
IARC, 1999d, Re evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide. Triethylene
Glycol Diglycidyl Etheri (Lyon: IARC), 71, 1285.
Plna, K., Segerback, D. and Schweda, E. K. H., 1996, DNA adduct formation by allyl glycidyl ether.
Carcinogenesis, 17, 1465 1471.
±
Steiner , S., Crane, A. E. and Watson , W. P., 1992a, Molecular dosimetry of DNA adducts in C3H
mice treated with glycidaldehyde. Carcinogenesis, 13, 119 124.
±
Steiner , S., Honger, G. and Sagelsdorff , P., 1992b, Molecular dosimetry of DNA adducts in C3H
mice treated with bisphenol A diglycidyl ether. Carcinogenesis, 13, 969 972.
Van Elburg, P. A., Ormskerk , O. A., de Kloe, K. P. and Boogaard, P. J., 2000, Synthesis of [ C]
±
"%
-
labelled glycidyl and glycerol ethers of aliphatic and aromatic alcohols. Journal of Labelled
Compound Radiopharmacology, 43, 147 167.
±
Wilschut , A., Ten Berge, W. F., Robinson , P. J. and McKone, T. E., 1995, Estimating skin
permeation, the validation of Ž ve mathematical skin permeation models. Chemosphere, 30,
1275 1296.
±