- Email: [email protected]
Application of solid-phase microextraction to in vitro skin permeation experiments: example using diethyl phthalate
Toxicology in Vitro 19 (2005) 253–259 www.elsevier.com/locate/toxinvit
Application of solid-phase microextraction to in vitro skin permeation experiments: example using diethyl phthalate H. Fred Frasch *, Ana M. Barbero Health Effects Laboratory Division, National Institute for Occupational Safety and Health, 1095 Willowdale Road, Morgantown, WV 26505, USA Received 28 June 2004; accepted 12 October 2004
Abstract The application of automated solid-phase microextraction (SPME) as a sample preparation technique for in vitro studies of skin permeation is described, using diethyl phthalate (DEP) as an example. In vitro diffusion cell experiments and skin–vehicle partition coefficient determinations require quantitative analysis of low-level analytes in aqueous samples. SPME is an ideal candidate for sample preparation for subsequent gas chromatographic analysis, offering numerous advantages over other methods. SPME conditions were optimized and the automated method was found to exhibit adequate sensitivity and good precision (relative standard deviation = 3%). Abdominal skin (dermatomed at 350 lm) from male hairless guinea pigs (n = 6) was used to measure DEP skin permeation parameters. In vitro methods were employed to determine permeability coefficient (kp), time lag (s) and skin–buffer partition coefficient (KSB) for 2 mM DEP in HEPES buffered Hanks Balanced Salt Solution. Measurements (mean ± standard deviations) are: kp, 0.021 ± 0.012 cm/h; s, 0.67 ± 0.18 h; KSB, 4.74 ± 0.68. The skin may be a significant route for the uptake of DEP. 2004 Elsevier Ltd. All rights reserved. Keywords: Permeability; Time lag; Membrane vehicle partition coefficient; Phthalic acid diesters; Gas chromatography methods
1. Introduction In vitro diffusion cell experiments yield quantitative information about the transport of chemicals through skin. The permeability coefficient (kp) is a measure of the conductance of skin to a particular chemical. It is proportional to the steady-state flux of the chemical through skin when a constant concentration difference is imposed between the outer and inner surfaces. The lag time (s) is a measure of the delay observed from the time chemical is applied to the outer surface of the skin until it appears at the inner surface. Both measurements can be obtained from appropriately designed diffusion cell experiments. The skin–vehicle partition *
Corresponding author. Tel.: +1 304 285 5755; fax: +1 304 285 6041. E-mail address: [email protected] (H.F. Frasch). 0887-2333/$ - see front matter 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tiv.2004.10.001
coefficient (Ksv) is another important determinant of the total amount of chemical that penetrates skin. It expresses the relative affinity of the chemical for skin compared with the vehicle in which the chemical is dissolved—often water. A high Ksv means that the chemical has a high affinity for skin compared with vehicle and will readily partition into the skin, where it may then diffuse into the systemic circulation. This parameter can also readily be measured using in vitro methods. Solid-phase microextraction (SPME) is a sample preparation method that is particularly well-suited to the detection of semi-volatile to volatile chemicals in aqueous solutions (Lord and Pawliszyn, 2000). SPME fibers are thin fused silica rods coated with an absorbent or adsorbent polymer stationary phase. Analytes are extracted by a partitioning mechanism following submersion of the fiber into media. The analytes are then thermally desorbed by placing the fiber in the injection
254
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
port of a gas chromatography system. This method is readily automated, and eliminates laborious and organic waste-generating extraction processes such as liquid– liquid, or solid-phase extraction. Phthalic acid diesters are ubiquitous industrial chemicals and environmental contaminants. The higher molecular weight phthalates are primarily used to impart flexibility to plastics. Diethyl phthalate (DEP, CAS Registry no. 84-66-2) is widely used as a stabilizing agent in perfumes and other cosmetic formulations (Api, 2001). It is also used as a plasticizer in cellulose ester plastics, and as an ethanol denaturant. The potential toxicity of phthalate esters is an ongoing concern. Perhaps owing to its relatively low molecular weight (222.2) and moderate lipophilicity (log Kow = 2.38; Staples et al., 1997), the US Environmental Protection Agency and other governmental agencies have expressed concern over the potential for dermal absorption of DEP. DEP was placed on the Interagency Testing CommitteeÕs list of chemicals designated for dermal absorption rate testing (US EPA, 1993), but subsequently was removed (US EPA, 1994). The stated reason for removal was that sufficient information on dermal absorption rate is now available that is likely to meet OSHAÕs needs. The EPA (1994) cites one in vitro study of DEP permeability (Scott et al., 1987) as the source of this information. Scott et al. report low steady-state flux and permeability of neat DEP in human skin. However, 2 years later, the authors published errata (Scott et al., 1989) in which they state a 10-fold increase in steadystate flux over their original published data. These errata have been overlooked in recent reviews of DEP toxicity (Faust, 1994; Api, 2001); that is, these reviews cite the original low steady-state fluxes that are in error. Another point for consideration is that permeability coefficients measured from neat liquid will in general be substantially less than those measured from an aqueous donor, and these are not directly comparable. Since permeability coefficients are traditionally measured from an aqueous solution, it is desirable to perform these experiments. In this study, the application of SPME as a sample preparation method for in vitro skin diffusion studies is described. We measured hairless guinea pig skin permeability coefficient, lag time, and skin–buffer partition coefficient of DEP dissolved in an aqueous buffer.
2. Materials and methods 2.1. Chemicals and materials All chemicals, including HPLC-grade water, were purchased from Sigma-Aldrich (St. Louis, MO) and were of the highest purity available (generally, HPLCgrade), except: HankÕs balanced salt solution was pur-
chased from Gibco Life Technologies (Rockville, MD). SPME fibers and precleaned screw-top glass vials were purchased from Supelco (Bellefonte, PA); crimp seal amber autosampler vials came from Microliter Analytical Supplies (Suwanee, GA). 2.2. Buffer Buffer consisted of HEPES-buffered HankÕs balanced salt solution (HBHBSS). 5.96 g HEPES free acid was stirred into 1000 ml of HankÕs. Then 0.32 g of NaHCO3 and 0.05 g gentamicin sulfate were added. The pH was brought to 7.40 at 32 C by drop wise addition of 3 N NaOH. 2.3. Diethyl phthalate solutions DEP stock solution of 0.1 g in 100 ml methanol was made and stored at 4 C. Calibration standards of DEP over the range of 0.03–30 lg/ml were made by diluting the stock solution in buffer. Solutions of DEP in buffer (2 mM = 444 lg/ml), used as the donor solution for diffusion cell and partition coefficient experiments, were made fresh on the day of the experiment by diluting precisely weighed neat DEP in buffer. 2.4. Diethyl phthalate solubility An excess of DEP was added to buffer in precleaned, screw-top vials. The vials were rapidly mixed in a vortex mixer (Vortex Genie 2, A. Daigger & Co., Vernon Hills, IL) at room temperature for over 24 h. Triplicate samples were pipetted into Eppendorf centrifuge tubes and spun at 10,000 rpm for 30 min. The supernatant was diluted 1:100 in buffer and samples were subjected to SPME and GC analysis as described below. Eight separate determinations were made. 2.5. Instrumentation Gas chromatographic (GC) analyses were performed using a Varian CP-3800 gas chromatograph with electronic flow control, a model 1177 split-splitless injector and flame ionization detector (FID) (Varian Inc., Walnut Creek, CA). The GC was equipped with a Combi Pal autosampler (CTC Analytics, Zwingen, Switzerland). The capillary column was 5% diphenyl, 95% dimethyl polysiloxane, 15 m long · 0.25 mm inner diameter, with 0.25 lm film thickness (Varian). GC conditions were as follows. The injector was set at 270 C isothermal and the detector at 300 C. The oven was programmed from 60 C with a 5 min hold time, then ramped to 270 C at 15 C/min, and held for 5 min (total GC run time = 24.7 min). Nitrogen was used as both carrier and make-up gases at constant flow rates of 1 ml/min and 25 ml/min respectively. The split ratio
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
was programmed: initial split ratio, 10; at 0.01 min, splitless; at 5 min, split ratio 100; at 10 min, split ratio 10. 2.6. SPME procedures Solid-phase microextraction of DEP from aqueous samples was adapted from previously published methods (Luks-Betleg et al., 2001; Proku˚pkova´ et al., 2002). 85 lm polyacrylate fibers were used. Each fiber was preconditioned according the manufacturerÕs recommendations (300 C for 2 h). No more than 100 measurements were made with each fiber. SPME procedures were automated using the Combi Pal autosampler. Direct immersion sampling was used. Sample extraction time (the time the fiber was in the sample) was 45 min, with the incubator set at 40 C with agitation (250 rpm). Desorption time (the time the fiber was in the GC injector) was 10 min. A retention time of 12.8 min was found for DEP under the specified GC conditions. Calibration standards of 0.1–10 lg/ml were included in each HGP experiment. 2.7. SPME fiber conditions Effects of sample extraction time—i.e., the amount of time the fiber is placed in the sample—and sample temperature were investigated. Three calibration standards of 10 lg/ml DEP in buffer were analyzed for each sample point. Total FID response at the 12.8 min peak was determined for samples incubated under slight heating (30 C and 40 C), which may promote faster equilibration (Penton, 1999) and with sample extraction time varied from 10 min to 60 min. All samples were agitated at 250 rpm. 2.8. SPME measurement precision and effect of time until analysis DEP calibration standard (10 lg/ml DEP in buffer) was placed into 10 autosampler vials. The autosampler was programmed to analyze 1 vial every 3 h. Total FID response at the 12.8 min peak was determined for each reading. 2.9. Hairless guinea pig skin Hairless guinea pigs (HGP) of the strain Crl:IAF(HA)-hrBR obtained from Charles River Laboratories (Wilmington, MA) were used for this study. The use of these animals was approved by our Animal Care and Use Committee. HGPÕs were euthanized with CO2 and abdominal skin was harvested and stored at 85 C until used. On the day of use, the skin was thawed at room temperature, pinned to a Teflon backing board, and dermatomed (Padgett Model B, Integra
255
LifeSciences, Plainsboro, NJ) at a setting of 350 lm thickness. A total of six HGPs were used. Skin sections were obtained using circular stainless steel punches, immediately weighed and mounted onto diffusion cells or used for partition coefficient measurements. 2.10. Diffusion cell experiments From each of six HGPs, three skin punches were mounted on Franz-type static diffusion cells (PermeGear, Bethlehem, PA). The receptor volume was 5 ml and the diameter of exposed skin was 9 mm. The receptor compartments were stirred at 1000 rpm. The waterjacketed cells were kept at 37 C via a recirculating water bath; this maintained the donor compartments at 32 C. At time 0, 0.5 ml of 2 mM DEP in HBHBSS was added to each donor compartment. Samples (0.5 ml) were removed from receptor compartments at times 0, 0.5, 1, 2, 3, 4 and 5 h and added to 0.5 ml HBHBSS in 2 ml autosampler vials for subsequent analysis. The same volume of fresh HBHBSS was added back to the receptor compartments to maintain constant volume. Samples were removed and buffer added using gas tight Hamilton syringes fitted to Chaney adaptors (Hamilton Company, Reno, NV) for precise repeatability. Samples of donor compartment solution were also taken and diluted 1:100 in buffer for analysis. Donor compartment solutions were replaced at 2.5 h to maintain ‘‘infinite source’’ conditions. Throughout the experiment, sink conditions were maintained in the receptor compartment: maximum receptor DEP concentrations <3% of donor concentration. The DEP concentrations were analyzed and the cumulative amount of DEP permeating each skin punch was calculated, accounting for the amount of DEP removed with each sample. From each of six HGPs, the mean of three skin punches at each time point was calculated. The permeability coefficient and lag time for each HGP were calculated by nonlinear regression of the following equation (Crank, 1975) through these mean values: 2 2 n 1 12 k p Cs X ð1Þ n p t QðtÞ ¼ k p Ct k p Cs exp 2 2 p 6s n n¼1 ð1Þ Here, Q(t) is the cumulative amount of DEP penetrated per unit area of skin at time t; C is the DEP concentration of the donor compartment. There are two unknowns to be estimated: kp is the permeability coefficient, and s is lag time. Nonlinear regressions were performed using SigmaPlot 2002 (SPSS Inc., Chicago, IL). The equation was truncated to seven terms in the series. Use of Eq. (1) is mathematically equivalent to calculating kp from the slope of the steady-state DEP accumulation curve and s as the intercept of this curve with the
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
time axis. However, use of Eq. (1) eliminates subjectivity of the analyst in these determinations.
40 oC 30 oC
2
6
2.11. Skin–buffer partition coefficient measurement
FID Response (10 area counts)
256
K SB ¼
C S ðC R V R C B V B Þ=V S ¼ CB CB
ð2Þ
0 15
30
45
60
Extraction Time (min)
Fig. 1. Effect of temperature and sample extraction time on detector response. Flame ionization detector response is displayed as a function of sample absorption time. The SPME fiber was placed in DEP solution (10 lg/ml) for the specified times with stirring at 250 rpm and heating at 30 C or 40 C as indicated. Data points represent means ± standard deviation of three determinations; smooth lines are spline interpolations.
3.2. SPME measurement precision and effect of time until analysis Results from these tests are displayed in Fig. 2. The relative standard deviation of the 10 sequential measurements was 2.9%, and there was no effect of time on the measured concentration over the 27 h analysis time (P = 0.64, Pearson Product Moment test). 3.3. DEP calibration curve and DEP solubility Fig. 3 is a typical calibration curve for DEP. A separate calibration curve was generated for each HGP experiment. The calibration is linear (r2 = 0.997) over three orders of magnitude of concentration, which spanned all measured DEP concentrations except those at the time = 0 h points of the permeation studies. DEP solubility in buffer averaged 879 lg/ml = 4.0 mM.
2.0
RSD = 2.9%
6
where C is concentration, V is volume, and subscripts S, B and R refer to skin, buffer and reference. CB is the DEP concentration in the vial after incubation with skin. CR was taken as the mean from the three reference vials. The volume of skin was calculated from weight assuming a density of 1 g/ml. The mean from three skin samples was taken for each HGP. Preliminary tests demonstrated that the centrifugation and filtration processes did not affect the measured concentration of DEP in reference vials, and that there was no difference in measured KSB when incubations were 2, 3 or 4 h.
1
0
FID Response (10 area counts)
A modification of the vial depletion method (Jepson et al., 1994) was used. From each of the same six HGPÕs used for permeation studies, three pre-weighed skin punches were placed in separate 4 ml pre-cleaned screw-top vials, each containing a 10 mm Teflon-coated magnetic stir bar. 1.5 ml of 2 mM DEP in HBHBSS was added to the vials, and they were placed on a 15-position magnetic stirrer (Variomag Telesystem, H + P Labortechnik, Oberschleißheim, Germany) immersed in a 32 C water bath (Precision model 285, Winchester VA) and incubated for 3 h with gentle agitation (350 rpm, ‘‘shake’’ mode). Three reference vials (no skin punches) were incubated simultaneously. Liquid from the vials was transferred after incubation into Eppendorf centrifuge tubes and spun at 10,000 rpm for 30 min. The supernatant was then filtered through a 0.45 lm hydrophilic polytetrafluoroethylene syringe filter (Millex LCR, Millipore, Bedford, MA) fitted to a 1 ml plastic syringe (BD, Franklin Lakes, NJ). Ten microliters of filtered supernatant was added to 990 ll of HBHBSS in a 2 ml autosampler vial for subsequent determination of DEP concentrations. The skin–buffer partition coefficient, KSB, was calculated from:
3. Results 3.1. SPME fiber conditions Fig. 1 shows the effects of sample temperature and fiber extraction time on detector response. All samples were agitated (250 rpm) for the duration of the extraction time. Data points represent means ± standard deviations for three determinations per point. Lines are spline interpolations between the points.
1.5
1.0 0
9
18
27
x,SD
Time to Analysis (hours)
Fig. 2. SPME precision and effect of time to analysis. DEP solution (10 lg/ml) was placed in 10 autosampler vials, and the samples were analyzed at 3 h intervals. Detector response is shown. x, SD: mean, standard deviation. RSD: relative standard deviation.
8
10
7
10
6
10
5
10
4
0.01
0.2
10
3
0.00
0.0
log Area = 1.000 log [DEP] + 5.468 2 r = 0.997
0.05
1.0
0.04
0.8
7 6
0.02
0.6 0.4
KSB
0.03
τ (hr)
5 4 3 2
0.01
0.1
1
10
100
DEP Concentration (µg/ml)
Fig. 3. Typical DEP calibration. Detector response at the given DEP concentrations is displayed. r2 is the linear correlation coefficient of the given equation.
3.4. DEP permeation coefficient and lag time Fig. 4 displays the accumulation of DEP over time in receptor compartments from three skin punches from one HGP. From this data, kp and s are calculated as described in the Methods. Data points represent the means ± SDÕs of the three skin punches, while the continuous line represents the regression of the data points with Eq. (1). One out of six permeation experiments was deemed to be a failure. In this experiment, DEP concentrations in receptor fluid were much lower than expected for two of the skin punches, and the accumulation curves could not be well-fitted using Eq. (1). For this HGP, the correlation coefficient (r2) of the regression was 0.970; for all others, r2 exceeded 0.995. We speculate that the SPME fiber may have been damaged for these
50 DEP accumulation ( µg/cm2)
257
10
kp (cm/hr)
FID Response (area counts)
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
kp = 0.018 cm/hr τ = 0.62 hr
40
2
r = 0.998 30 20 10 0 0
1
2
3
4
5
Time (hours)
Fig. 4. Typical diffusion cell results. Total DEP accumulation in the receptor compartment is displayed as a function of time. Data points represent means ± standard deviations for three skin punches from one HGP. Smooth line is the regression of these data points with Eq. (1), which yields an estimate of kp (permeation coefficient) and s (time lag). r2 is the nonlinear correlation coefficient.
1 0
Fig. 5. Scatter plot of data. Shown are all experimental determinations of permeation coefficient (kp, n = 5), time lag (s, n = 5), and skin– vehicle partition coefficient (KSB, n = 6). Also shown are means ± standard deviations.
samples. For completeness, the calculated kp for this experiment was 0.0028 cm/h and s was 0.47 h. A scatter plot with means and standard deviations for the remaining five HGPs is shown in Fig. 5. Mean ± SD of kp is 0.021 ± 0.012 cm/h; mean ± SD of s is 0.67 ± 0.18 h. 3.5. Skin–buffer partition coefficient Fig. 5 also shows individual points and the mean and standard deviation of the measured KSB for all six HGPs. Mean ± SD is 4.74 ± 0.68.
4. Discussion Solid-phase microextraction offers advantages over other means of sample preparation for the quantification of permeants in in vitro diffusion experiments. SPME replaces labor intensive, waste-generating multistep sample clean up and preparation techniques. For slightly volatile analytes such as DEP in aqueous samples, liquid–liquid or solid-phase extraction represent reasonable alternatives to SPME. These involve the use of organic solvents and sorbent columns to concentrate the analytes into a solvent that is compatible with direct GC injection. These techniques are manual or semi-automated at best. Another alternative is the use of radiolabeled DEP; however this has its own issues, including license requirements, generation of radioactive waste, uncertainty regarding radiochemical purity, and risk of contamination to laboratory equipment and personnel. The convenience of automation of GC procedures with SPME is invaluable. All sample extraction and measurement steps can be programmed using the autosampler. These diffusion cell experiments use three cells with seven time samples for each HGP used, plus three samples for the donor cell concentrations. Additionally, six samples from each HGP are required to measure skin–vehicle partition coefficient. Manual preparation
258
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
of these 30 samples for each HGP would be quite laborious. We investigated several parameters that affect the sensitivity of the SPME method. Fig. 1 shows the effects of sample temperature and fiber extraction time on detector response. The data demonstrate that slight sample heating and longer extraction times enhance sensitivity. The figure also shows that equilibration is not yet reached, even at 60 min extraction time. This is acceptable as long as extraction time is precisely regulated (Lord and Pawliszyn, 2000). The 45 min extraction time for this study was selected as a compromise between the competing demands of sensitivity and sample run time. If fiber conditions are controlled precisely, then good precision of the method should be expected. Fig. 2 demonstrates a relative standard deviation of 3%. Furthermore, there was no deterioration of the samples over a 27 h period, which exceeds the amount of time required for a full experimental run. The calibration of DEP in buffer using SPME is linear (Fig. 3), and the measured experimental concentrations fell well within the range of calibration standards that was tested. The inter-specimen variability in the measured permeability coefficients (Fig. 5) is well documented in these types of studies (Southwell et al., 1984; Williams et al., 1992). For example, Southwell et al. (1984) report an average of 66% in inter-specimen coefficient of variation (standard deviation divided by mean) for human skin studies of several permeants. This compares with our 57%. Although it might be expected that the variability will be greater for human skin specimens than for inbred species such as the HGP, perhaps a 57% coefficient of variation is not unreasonable. The permeability of DEP measured here (2.1 · 102 cm/h) is typical of a chemical with its physicochemical properties. The Frasch model (Frasch, 2002) predicts a kp of 5.1 · 103 cm/h— 1/4 of the value measured here—while the Potts and Guy (1992) equation predicts 3.9 · 103 cm/h. Both models are based on permeability data from human skin, which has generally been found to be less permeable than HGP skin. There is apparently only one reported in vitro measurement of DEP permeability in the peerreviewed literature. Scott et al. (1987, errata 1989) report a steady-state absorption rate (flux) of the neat chemical of 12.8 lg cm2 h1 for human epidermis and 414 lg cm2 h1 for rat epidermis. These compare to our mean value of 8.0 lg cm2 h1 (calculated by multiplying permeability coefficient by donor concentration). Estimated maximum flux from a saturated solution (4 mM or 2· the concentration used here) would therefore be 16 lg cm2 h1. Because the thermodynamic potential of a chemical is the same from a saturated solution as from the neat chemical, maximum fluxes from infinite doses of the saturated aqueous solution
and from the neat chemical should be equal, if there are no vehicle effects or permeant effects on permeability. Scott et al. (1987, errata 1989) also report a kp value of 1.14 · 105 cm/h. This is calculated as the steady-state flux divided by the donor concentration, which in this case is given by the density of DEP. It is important to emphasize that values of permeability coefficients measured from neat chemical should not be compared with those measured from aqueous solutions. Uptake of DEP by the dermal route may be significant. The corrected steady-state flux values reported by Scott et al. (1987, errata 1989) and those reported herein, are available for the evaluation of percutaneous absorption of DEP.
References Api, A.M., 2001. Toxicological profile of diethyl phthalate: a vehicle for fragrance and cosmetic ingredients. Food and Chemical Toxicology 39, 97–108. Crank, J., 1975. The Mathematics of Diffusion, second ed. Clarendon Press, Oxford, p. 51. Faust, R.A., 1994. Toxicity Summary for Diethyl Phthalate. Oak Ridge National Laboratory, Center for Integrated Risk Assessment http://riskassessment.ornl.gov/documents/DEPHTHAL.pdf. Frasch, H.F., 2002. A random walk model of skin permeation. Risk Analysis 22, 265–276. Jepson, G.W., Hoover, D.K., Black, R.K., McCafferty, J.D., Mahle, D.A., Gearhart, J.M., 1994. A partition coefficient determination method for nonvolatile chemicals in biological tissues. Fundamental and Applied Toxicology 22, 519–524. Lord, H., Pawliszyn, J., 2000. Evolution of solid-phase microextraction technology. Journal of Chromatography A 885, 153–193. Luks-Betleg, K., Popp, P., Janoszka, B., Paschke, H., 2001. Solidphase microextraction of phthalates from water. Journal of Chromatography A 938, 93–101. Penton, Z., 1999. Method development with solid phase microextraction. In: Wercinski, S.A.S. (Ed.), Solid Phase Microextraction: A Practical Guide. Marcel Dekker, New York. Potts, R.O., Guy, R.H., 1992. Predicting skin permeability. Pharmaceutical Research 9, 663–669. Proku˚pkova´, G., Holadova´, K., Poustka, J., Hajsˇlova´, J., 2002. Development of a solid-phase microextraction method for the determination of phthalic acid esters in water. Analytica Chimica Acta 457, 211–223. Scott, R.C., Dugard, P.H., Ramsey, J.D., Rhodes, C., 1987. In vitro absorption of some o-phthalate diesters through human and rat skin. Environmental Health Perspectives 74, 223–227. Scott, R.C., Dugard, P.H., Ramsey, J.D., Rhodes, C., 1989. Errata in ‘‘In vitro absorption of some o-phthalate diesters through human and rat skin’’. Environmental Health Perspectives 79, 323. Southwell, D., Barry, B.W., Woodford, R., 1984. Variations in permeability of human skin within and between specimens. International Journal of Pharmaceutics 18, 299–309. Staples, C.A., Peterson, D.R., Parkerton, T.F., Adams, W.J., 1997. The environmental fate of phthalate esters: a literature review. Chemosphere 35, 667–749. US EPA, 1993. Thirty-second report of the TSCA interagency testing committee to the administrator; receipt of report, request for comments notice of opportunity to initiate negotiations for TSCA section 4 testing consent agreements. Federal Register 58, 38490– 38496.
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259 US EPA, 1994. Thirty-fourth report of the TSCA interagency testing committee to the administrator receipt of report and request for comments. Federal Register 59, 35720–35728.
259
Williams, A.C., Cornwell, P.A., Barry, B.W., 1992. On the nongaussian distribution of human skin permeabilities. International Journal of Pharmaceutics 86, 69–77.
Application of solid-phase microextraction to in vitro skin permeation experiments: example using diethyl phthalate H. Fred Frasch *, Ana M. Barbero Health Effects Laboratory Division, National Institute for Occupational Safety and Health, 1095 Willowdale Road, Morgantown, WV 26505, USA Received 28 June 2004; accepted 12 October 2004
Abstract The application of automated solid-phase microextraction (SPME) as a sample preparation technique for in vitro studies of skin permeation is described, using diethyl phthalate (DEP) as an example. In vitro diffusion cell experiments and skin–vehicle partition coefficient determinations require quantitative analysis of low-level analytes in aqueous samples. SPME is an ideal candidate for sample preparation for subsequent gas chromatographic analysis, offering numerous advantages over other methods. SPME conditions were optimized and the automated method was found to exhibit adequate sensitivity and good precision (relative standard deviation = 3%). Abdominal skin (dermatomed at 350 lm) from male hairless guinea pigs (n = 6) was used to measure DEP skin permeation parameters. In vitro methods were employed to determine permeability coefficient (kp), time lag (s) and skin–buffer partition coefficient (KSB) for 2 mM DEP in HEPES buffered Hanks Balanced Salt Solution. Measurements (mean ± standard deviations) are: kp, 0.021 ± 0.012 cm/h; s, 0.67 ± 0.18 h; KSB, 4.74 ± 0.68. The skin may be a significant route for the uptake of DEP. 2004 Elsevier Ltd. All rights reserved. Keywords: Permeability; Time lag; Membrane vehicle partition coefficient; Phthalic acid diesters; Gas chromatography methods
1. Introduction In vitro diffusion cell experiments yield quantitative information about the transport of chemicals through skin. The permeability coefficient (kp) is a measure of the conductance of skin to a particular chemical. It is proportional to the steady-state flux of the chemical through skin when a constant concentration difference is imposed between the outer and inner surfaces. The lag time (s) is a measure of the delay observed from the time chemical is applied to the outer surface of the skin until it appears at the inner surface. Both measurements can be obtained from appropriately designed diffusion cell experiments. The skin–vehicle partition *
Corresponding author. Tel.: +1 304 285 5755; fax: +1 304 285 6041. E-mail address: [email protected] (H.F. Frasch). 0887-2333/$ - see front matter 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tiv.2004.10.001
coefficient (Ksv) is another important determinant of the total amount of chemical that penetrates skin. It expresses the relative affinity of the chemical for skin compared with the vehicle in which the chemical is dissolved—often water. A high Ksv means that the chemical has a high affinity for skin compared with vehicle and will readily partition into the skin, where it may then diffuse into the systemic circulation. This parameter can also readily be measured using in vitro methods. Solid-phase microextraction (SPME) is a sample preparation method that is particularly well-suited to the detection of semi-volatile to volatile chemicals in aqueous solutions (Lord and Pawliszyn, 2000). SPME fibers are thin fused silica rods coated with an absorbent or adsorbent polymer stationary phase. Analytes are extracted by a partitioning mechanism following submersion of the fiber into media. The analytes are then thermally desorbed by placing the fiber in the injection
254
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
port of a gas chromatography system. This method is readily automated, and eliminates laborious and organic waste-generating extraction processes such as liquid– liquid, or solid-phase extraction. Phthalic acid diesters are ubiquitous industrial chemicals and environmental contaminants. The higher molecular weight phthalates are primarily used to impart flexibility to plastics. Diethyl phthalate (DEP, CAS Registry no. 84-66-2) is widely used as a stabilizing agent in perfumes and other cosmetic formulations (Api, 2001). It is also used as a plasticizer in cellulose ester plastics, and as an ethanol denaturant. The potential toxicity of phthalate esters is an ongoing concern. Perhaps owing to its relatively low molecular weight (222.2) and moderate lipophilicity (log Kow = 2.38; Staples et al., 1997), the US Environmental Protection Agency and other governmental agencies have expressed concern over the potential for dermal absorption of DEP. DEP was placed on the Interagency Testing CommitteeÕs list of chemicals designated for dermal absorption rate testing (US EPA, 1993), but subsequently was removed (US EPA, 1994). The stated reason for removal was that sufficient information on dermal absorption rate is now available that is likely to meet OSHAÕs needs. The EPA (1994) cites one in vitro study of DEP permeability (Scott et al., 1987) as the source of this information. Scott et al. report low steady-state flux and permeability of neat DEP in human skin. However, 2 years later, the authors published errata (Scott et al., 1989) in which they state a 10-fold increase in steadystate flux over their original published data. These errata have been overlooked in recent reviews of DEP toxicity (Faust, 1994; Api, 2001); that is, these reviews cite the original low steady-state fluxes that are in error. Another point for consideration is that permeability coefficients measured from neat liquid will in general be substantially less than those measured from an aqueous donor, and these are not directly comparable. Since permeability coefficients are traditionally measured from an aqueous solution, it is desirable to perform these experiments. In this study, the application of SPME as a sample preparation method for in vitro skin diffusion studies is described. We measured hairless guinea pig skin permeability coefficient, lag time, and skin–buffer partition coefficient of DEP dissolved in an aqueous buffer.
2. Materials and methods 2.1. Chemicals and materials All chemicals, including HPLC-grade water, were purchased from Sigma-Aldrich (St. Louis, MO) and were of the highest purity available (generally, HPLCgrade), except: HankÕs balanced salt solution was pur-
chased from Gibco Life Technologies (Rockville, MD). SPME fibers and precleaned screw-top glass vials were purchased from Supelco (Bellefonte, PA); crimp seal amber autosampler vials came from Microliter Analytical Supplies (Suwanee, GA). 2.2. Buffer Buffer consisted of HEPES-buffered HankÕs balanced salt solution (HBHBSS). 5.96 g HEPES free acid was stirred into 1000 ml of HankÕs. Then 0.32 g of NaHCO3 and 0.05 g gentamicin sulfate were added. The pH was brought to 7.40 at 32 C by drop wise addition of 3 N NaOH. 2.3. Diethyl phthalate solutions DEP stock solution of 0.1 g in 100 ml methanol was made and stored at 4 C. Calibration standards of DEP over the range of 0.03–30 lg/ml were made by diluting the stock solution in buffer. Solutions of DEP in buffer (2 mM = 444 lg/ml), used as the donor solution for diffusion cell and partition coefficient experiments, were made fresh on the day of the experiment by diluting precisely weighed neat DEP in buffer. 2.4. Diethyl phthalate solubility An excess of DEP was added to buffer in precleaned, screw-top vials. The vials were rapidly mixed in a vortex mixer (Vortex Genie 2, A. Daigger & Co., Vernon Hills, IL) at room temperature for over 24 h. Triplicate samples were pipetted into Eppendorf centrifuge tubes and spun at 10,000 rpm for 30 min. The supernatant was diluted 1:100 in buffer and samples were subjected to SPME and GC analysis as described below. Eight separate determinations were made. 2.5. Instrumentation Gas chromatographic (GC) analyses were performed using a Varian CP-3800 gas chromatograph with electronic flow control, a model 1177 split-splitless injector and flame ionization detector (FID) (Varian Inc., Walnut Creek, CA). The GC was equipped with a Combi Pal autosampler (CTC Analytics, Zwingen, Switzerland). The capillary column was 5% diphenyl, 95% dimethyl polysiloxane, 15 m long · 0.25 mm inner diameter, with 0.25 lm film thickness (Varian). GC conditions were as follows. The injector was set at 270 C isothermal and the detector at 300 C. The oven was programmed from 60 C with a 5 min hold time, then ramped to 270 C at 15 C/min, and held for 5 min (total GC run time = 24.7 min). Nitrogen was used as both carrier and make-up gases at constant flow rates of 1 ml/min and 25 ml/min respectively. The split ratio
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
was programmed: initial split ratio, 10; at 0.01 min, splitless; at 5 min, split ratio 100; at 10 min, split ratio 10. 2.6. SPME procedures Solid-phase microextraction of DEP from aqueous samples was adapted from previously published methods (Luks-Betleg et al., 2001; Proku˚pkova´ et al., 2002). 85 lm polyacrylate fibers were used. Each fiber was preconditioned according the manufacturerÕs recommendations (300 C for 2 h). No more than 100 measurements were made with each fiber. SPME procedures were automated using the Combi Pal autosampler. Direct immersion sampling was used. Sample extraction time (the time the fiber was in the sample) was 45 min, with the incubator set at 40 C with agitation (250 rpm). Desorption time (the time the fiber was in the GC injector) was 10 min. A retention time of 12.8 min was found for DEP under the specified GC conditions. Calibration standards of 0.1–10 lg/ml were included in each HGP experiment. 2.7. SPME fiber conditions Effects of sample extraction time—i.e., the amount of time the fiber is placed in the sample—and sample temperature were investigated. Three calibration standards of 10 lg/ml DEP in buffer were analyzed for each sample point. Total FID response at the 12.8 min peak was determined for samples incubated under slight heating (30 C and 40 C), which may promote faster equilibration (Penton, 1999) and with sample extraction time varied from 10 min to 60 min. All samples were agitated at 250 rpm. 2.8. SPME measurement precision and effect of time until analysis DEP calibration standard (10 lg/ml DEP in buffer) was placed into 10 autosampler vials. The autosampler was programmed to analyze 1 vial every 3 h. Total FID response at the 12.8 min peak was determined for each reading. 2.9. Hairless guinea pig skin Hairless guinea pigs (HGP) of the strain Crl:IAF(HA)-hrBR obtained from Charles River Laboratories (Wilmington, MA) were used for this study. The use of these animals was approved by our Animal Care and Use Committee. HGPÕs were euthanized with CO2 and abdominal skin was harvested and stored at 85 C until used. On the day of use, the skin was thawed at room temperature, pinned to a Teflon backing board, and dermatomed (Padgett Model B, Integra
255
LifeSciences, Plainsboro, NJ) at a setting of 350 lm thickness. A total of six HGPs were used. Skin sections were obtained using circular stainless steel punches, immediately weighed and mounted onto diffusion cells or used for partition coefficient measurements. 2.10. Diffusion cell experiments From each of six HGPs, three skin punches were mounted on Franz-type static diffusion cells (PermeGear, Bethlehem, PA). The receptor volume was 5 ml and the diameter of exposed skin was 9 mm. The receptor compartments were stirred at 1000 rpm. The waterjacketed cells were kept at 37 C via a recirculating water bath; this maintained the donor compartments at 32 C. At time 0, 0.5 ml of 2 mM DEP in HBHBSS was added to each donor compartment. Samples (0.5 ml) were removed from receptor compartments at times 0, 0.5, 1, 2, 3, 4 and 5 h and added to 0.5 ml HBHBSS in 2 ml autosampler vials for subsequent analysis. The same volume of fresh HBHBSS was added back to the receptor compartments to maintain constant volume. Samples were removed and buffer added using gas tight Hamilton syringes fitted to Chaney adaptors (Hamilton Company, Reno, NV) for precise repeatability. Samples of donor compartment solution were also taken and diluted 1:100 in buffer for analysis. Donor compartment solutions were replaced at 2.5 h to maintain ‘‘infinite source’’ conditions. Throughout the experiment, sink conditions were maintained in the receptor compartment: maximum receptor DEP concentrations <3% of donor concentration. The DEP concentrations were analyzed and the cumulative amount of DEP permeating each skin punch was calculated, accounting for the amount of DEP removed with each sample. From each of six HGPs, the mean of three skin punches at each time point was calculated. The permeability coefficient and lag time for each HGP were calculated by nonlinear regression of the following equation (Crank, 1975) through these mean values: 2 2 n 1 12 k p Cs X ð1Þ n p t QðtÞ ¼ k p Ct k p Cs exp 2 2 p 6s n n¼1 ð1Þ Here, Q(t) is the cumulative amount of DEP penetrated per unit area of skin at time t; C is the DEP concentration of the donor compartment. There are two unknowns to be estimated: kp is the permeability coefficient, and s is lag time. Nonlinear regressions were performed using SigmaPlot 2002 (SPSS Inc., Chicago, IL). The equation was truncated to seven terms in the series. Use of Eq. (1) is mathematically equivalent to calculating kp from the slope of the steady-state DEP accumulation curve and s as the intercept of this curve with the
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
time axis. However, use of Eq. (1) eliminates subjectivity of the analyst in these determinations.
40 oC 30 oC
2
6
2.11. Skin–buffer partition coefficient measurement
FID Response (10 area counts)
256
K SB ¼
C S ðC R V R C B V B Þ=V S ¼ CB CB
ð2Þ
0 15
30
45
60
Extraction Time (min)
Fig. 1. Effect of temperature and sample extraction time on detector response. Flame ionization detector response is displayed as a function of sample absorption time. The SPME fiber was placed in DEP solution (10 lg/ml) for the specified times with stirring at 250 rpm and heating at 30 C or 40 C as indicated. Data points represent means ± standard deviation of three determinations; smooth lines are spline interpolations.
3.2. SPME measurement precision and effect of time until analysis Results from these tests are displayed in Fig. 2. The relative standard deviation of the 10 sequential measurements was 2.9%, and there was no effect of time on the measured concentration over the 27 h analysis time (P = 0.64, Pearson Product Moment test). 3.3. DEP calibration curve and DEP solubility Fig. 3 is a typical calibration curve for DEP. A separate calibration curve was generated for each HGP experiment. The calibration is linear (r2 = 0.997) over three orders of magnitude of concentration, which spanned all measured DEP concentrations except those at the time = 0 h points of the permeation studies. DEP solubility in buffer averaged 879 lg/ml = 4.0 mM.
2.0
RSD = 2.9%
6
where C is concentration, V is volume, and subscripts S, B and R refer to skin, buffer and reference. CB is the DEP concentration in the vial after incubation with skin. CR was taken as the mean from the three reference vials. The volume of skin was calculated from weight assuming a density of 1 g/ml. The mean from three skin samples was taken for each HGP. Preliminary tests demonstrated that the centrifugation and filtration processes did not affect the measured concentration of DEP in reference vials, and that there was no difference in measured KSB when incubations were 2, 3 or 4 h.
1
0
FID Response (10 area counts)
A modification of the vial depletion method (Jepson et al., 1994) was used. From each of the same six HGPÕs used for permeation studies, three pre-weighed skin punches were placed in separate 4 ml pre-cleaned screw-top vials, each containing a 10 mm Teflon-coated magnetic stir bar. 1.5 ml of 2 mM DEP in HBHBSS was added to the vials, and they were placed on a 15-position magnetic stirrer (Variomag Telesystem, H + P Labortechnik, Oberschleißheim, Germany) immersed in a 32 C water bath (Precision model 285, Winchester VA) and incubated for 3 h with gentle agitation (350 rpm, ‘‘shake’’ mode). Three reference vials (no skin punches) were incubated simultaneously. Liquid from the vials was transferred after incubation into Eppendorf centrifuge tubes and spun at 10,000 rpm for 30 min. The supernatant was then filtered through a 0.45 lm hydrophilic polytetrafluoroethylene syringe filter (Millex LCR, Millipore, Bedford, MA) fitted to a 1 ml plastic syringe (BD, Franklin Lakes, NJ). Ten microliters of filtered supernatant was added to 990 ll of HBHBSS in a 2 ml autosampler vial for subsequent determination of DEP concentrations. The skin–buffer partition coefficient, KSB, was calculated from:
3. Results 3.1. SPME fiber conditions Fig. 1 shows the effects of sample temperature and fiber extraction time on detector response. All samples were agitated (250 rpm) for the duration of the extraction time. Data points represent means ± standard deviations for three determinations per point. Lines are spline interpolations between the points.
1.5
1.0 0
9
18
27
x,SD
Time to Analysis (hours)
Fig. 2. SPME precision and effect of time to analysis. DEP solution (10 lg/ml) was placed in 10 autosampler vials, and the samples were analyzed at 3 h intervals. Detector response is shown. x, SD: mean, standard deviation. RSD: relative standard deviation.
8
10
7
10
6
10
5
10
4
0.01
0.2
10
3
0.00
0.0
log Area = 1.000 log [DEP] + 5.468 2 r = 0.997
0.05
1.0
0.04
0.8
7 6
0.02
0.6 0.4
KSB
0.03
τ (hr)
5 4 3 2
0.01
0.1
1
10
100
DEP Concentration (µg/ml)
Fig. 3. Typical DEP calibration. Detector response at the given DEP concentrations is displayed. r2 is the linear correlation coefficient of the given equation.
3.4. DEP permeation coefficient and lag time Fig. 4 displays the accumulation of DEP over time in receptor compartments from three skin punches from one HGP. From this data, kp and s are calculated as described in the Methods. Data points represent the means ± SDÕs of the three skin punches, while the continuous line represents the regression of the data points with Eq. (1). One out of six permeation experiments was deemed to be a failure. In this experiment, DEP concentrations in receptor fluid were much lower than expected for two of the skin punches, and the accumulation curves could not be well-fitted using Eq. (1). For this HGP, the correlation coefficient (r2) of the regression was 0.970; for all others, r2 exceeded 0.995. We speculate that the SPME fiber may have been damaged for these
50 DEP accumulation ( µg/cm2)
257
10
kp (cm/hr)
FID Response (area counts)
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
kp = 0.018 cm/hr τ = 0.62 hr
40
2
r = 0.998 30 20 10 0 0
1
2
3
4
5
Time (hours)
Fig. 4. Typical diffusion cell results. Total DEP accumulation in the receptor compartment is displayed as a function of time. Data points represent means ± standard deviations for three skin punches from one HGP. Smooth line is the regression of these data points with Eq. (1), which yields an estimate of kp (permeation coefficient) and s (time lag). r2 is the nonlinear correlation coefficient.
1 0
Fig. 5. Scatter plot of data. Shown are all experimental determinations of permeation coefficient (kp, n = 5), time lag (s, n = 5), and skin– vehicle partition coefficient (KSB, n = 6). Also shown are means ± standard deviations.
samples. For completeness, the calculated kp for this experiment was 0.0028 cm/h and s was 0.47 h. A scatter plot with means and standard deviations for the remaining five HGPs is shown in Fig. 5. Mean ± SD of kp is 0.021 ± 0.012 cm/h; mean ± SD of s is 0.67 ± 0.18 h. 3.5. Skin–buffer partition coefficient Fig. 5 also shows individual points and the mean and standard deviation of the measured KSB for all six HGPs. Mean ± SD is 4.74 ± 0.68.
4. Discussion Solid-phase microextraction offers advantages over other means of sample preparation for the quantification of permeants in in vitro diffusion experiments. SPME replaces labor intensive, waste-generating multistep sample clean up and preparation techniques. For slightly volatile analytes such as DEP in aqueous samples, liquid–liquid or solid-phase extraction represent reasonable alternatives to SPME. These involve the use of organic solvents and sorbent columns to concentrate the analytes into a solvent that is compatible with direct GC injection. These techniques are manual or semi-automated at best. Another alternative is the use of radiolabeled DEP; however this has its own issues, including license requirements, generation of radioactive waste, uncertainty regarding radiochemical purity, and risk of contamination to laboratory equipment and personnel. The convenience of automation of GC procedures with SPME is invaluable. All sample extraction and measurement steps can be programmed using the autosampler. These diffusion cell experiments use three cells with seven time samples for each HGP used, plus three samples for the donor cell concentrations. Additionally, six samples from each HGP are required to measure skin–vehicle partition coefficient. Manual preparation
258
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259
of these 30 samples for each HGP would be quite laborious. We investigated several parameters that affect the sensitivity of the SPME method. Fig. 1 shows the effects of sample temperature and fiber extraction time on detector response. The data demonstrate that slight sample heating and longer extraction times enhance sensitivity. The figure also shows that equilibration is not yet reached, even at 60 min extraction time. This is acceptable as long as extraction time is precisely regulated (Lord and Pawliszyn, 2000). The 45 min extraction time for this study was selected as a compromise between the competing demands of sensitivity and sample run time. If fiber conditions are controlled precisely, then good precision of the method should be expected. Fig. 2 demonstrates a relative standard deviation of 3%. Furthermore, there was no deterioration of the samples over a 27 h period, which exceeds the amount of time required for a full experimental run. The calibration of DEP in buffer using SPME is linear (Fig. 3), and the measured experimental concentrations fell well within the range of calibration standards that was tested. The inter-specimen variability in the measured permeability coefficients (Fig. 5) is well documented in these types of studies (Southwell et al., 1984; Williams et al., 1992). For example, Southwell et al. (1984) report an average of 66% in inter-specimen coefficient of variation (standard deviation divided by mean) for human skin studies of several permeants. This compares with our 57%. Although it might be expected that the variability will be greater for human skin specimens than for inbred species such as the HGP, perhaps a 57% coefficient of variation is not unreasonable. The permeability of DEP measured here (2.1 · 102 cm/h) is typical of a chemical with its physicochemical properties. The Frasch model (Frasch, 2002) predicts a kp of 5.1 · 103 cm/h— 1/4 of the value measured here—while the Potts and Guy (1992) equation predicts 3.9 · 103 cm/h. Both models are based on permeability data from human skin, which has generally been found to be less permeable than HGP skin. There is apparently only one reported in vitro measurement of DEP permeability in the peerreviewed literature. Scott et al. (1987, errata 1989) report a steady-state absorption rate (flux) of the neat chemical of 12.8 lg cm2 h1 for human epidermis and 414 lg cm2 h1 for rat epidermis. These compare to our mean value of 8.0 lg cm2 h1 (calculated by multiplying permeability coefficient by donor concentration). Estimated maximum flux from a saturated solution (4 mM or 2· the concentration used here) would therefore be 16 lg cm2 h1. Because the thermodynamic potential of a chemical is the same from a saturated solution as from the neat chemical, maximum fluxes from infinite doses of the saturated aqueous solution
and from the neat chemical should be equal, if there are no vehicle effects or permeant effects on permeability. Scott et al. (1987, errata 1989) also report a kp value of 1.14 · 105 cm/h. This is calculated as the steady-state flux divided by the donor concentration, which in this case is given by the density of DEP. It is important to emphasize that values of permeability coefficients measured from neat chemical should not be compared with those measured from aqueous solutions. Uptake of DEP by the dermal route may be significant. The corrected steady-state flux values reported by Scott et al. (1987, errata 1989) and those reported herein, are available for the evaluation of percutaneous absorption of DEP.
References Api, A.M., 2001. Toxicological profile of diethyl phthalate: a vehicle for fragrance and cosmetic ingredients. Food and Chemical Toxicology 39, 97–108. Crank, J., 1975. The Mathematics of Diffusion, second ed. Clarendon Press, Oxford, p. 51. Faust, R.A., 1994. Toxicity Summary for Diethyl Phthalate. Oak Ridge National Laboratory, Center for Integrated Risk Assessment http://riskassessment.ornl.gov/documents/DEPHTHAL.pdf. Frasch, H.F., 2002. A random walk model of skin permeation. Risk Analysis 22, 265–276. Jepson, G.W., Hoover, D.K., Black, R.K., McCafferty, J.D., Mahle, D.A., Gearhart, J.M., 1994. A partition coefficient determination method for nonvolatile chemicals in biological tissues. Fundamental and Applied Toxicology 22, 519–524. Lord, H., Pawliszyn, J., 2000. Evolution of solid-phase microextraction technology. Journal of Chromatography A 885, 153–193. Luks-Betleg, K., Popp, P., Janoszka, B., Paschke, H., 2001. Solidphase microextraction of phthalates from water. Journal of Chromatography A 938, 93–101. Penton, Z., 1999. Method development with solid phase microextraction. In: Wercinski, S.A.S. (Ed.), Solid Phase Microextraction: A Practical Guide. Marcel Dekker, New York. Potts, R.O., Guy, R.H., 1992. Predicting skin permeability. Pharmaceutical Research 9, 663–669. Proku˚pkova´, G., Holadova´, K., Poustka, J., Hajsˇlova´, J., 2002. Development of a solid-phase microextraction method for the determination of phthalic acid esters in water. Analytica Chimica Acta 457, 211–223. Scott, R.C., Dugard, P.H., Ramsey, J.D., Rhodes, C., 1987. In vitro absorption of some o-phthalate diesters through human and rat skin. Environmental Health Perspectives 74, 223–227. Scott, R.C., Dugard, P.H., Ramsey, J.D., Rhodes, C., 1989. Errata in ‘‘In vitro absorption of some o-phthalate diesters through human and rat skin’’. Environmental Health Perspectives 79, 323. Southwell, D., Barry, B.W., Woodford, R., 1984. Variations in permeability of human skin within and between specimens. International Journal of Pharmaceutics 18, 299–309. Staples, C.A., Peterson, D.R., Parkerton, T.F., Adams, W.J., 1997. The environmental fate of phthalate esters: a literature review. Chemosphere 35, 667–749. US EPA, 1993. Thirty-second report of the TSCA interagency testing committee to the administrator; receipt of report, request for comments notice of opportunity to initiate negotiations for TSCA section 4 testing consent agreements. Federal Register 58, 38490– 38496.
H.F. Frasch, A.M. Barbero / Toxicology in Vitro 19 (2005) 253–259 US EPA, 1994. Thirty-fourth report of the TSCA interagency testing committee to the administrator receipt of report and request for comments. Federal Register 59, 35720–35728.
259
Williams, A.C., Cornwell, P.A., Barry, B.W., 1992. On the nongaussian distribution of human skin permeabilities. International Journal of Pharmaceutics 86, 69–77.