Assessment of the in vitro skin irritaion by chemicals using the Vitrolife-Skin human skin model

Transcription

1 ORIGINAL ARTICLE Noriyuki Morikawa et al., AATEX 13(1), 11-26, 2008 Assessment of the in vitro skin irritaion by chemicals using the Vitrolife-Skin human skin model Noriyuki Morikawa, Tatsuya Kitagawa and Kenji Tomihata Research & Development Center, GUNZE Ltd., Kyoto, Japan Abstract The European Centre for the Validation of Alternative Methods (ECVAM) is currently supporting formal validation studies of in vitro tests for predicting skin irritancy and corrosivity, including tests employing two human skin models, EPISKIN and EpiDerm. When skin models are used to evaluate skin irritancy, it is important that suitable chemical application procedures are used. We evaluated the skin irritancy of 44 chemicals using the postincubation method (10-minutes treatment and 18-hours post-treatment incubation) that we originally developed to predict skin irritancy which is similar to a refined protocol for EPISKIN proposed in an ECVAM validation study. The sensitivity, specificity, and accuracy obtained with the MTT reduction assay-based prediction model were 77.8%, 76.9%, and 77.3%, respectively while the corresponding values obtained with the interleukin-1α secretion assay-based prediction model were 61.1%, 92.3%, and 79.5%, respectively. Combining these endpoints indicated a clear increase in sensitivity and accuracy to 94.4% and 81.8%, respectively. Vitrolife-Skin showed a basic utility for irritancy testing by this method and it is possible to confidently predict skin irritancy, provided that an appropriate chemical application procedure is used. Key words: human skin model, skin irritancy, Vitrolife-Skin, EPISKIN, ECVAM Introduction The European Centre for the Validation of Alternative Methods (ECVAM) is currently supporting formal validation studies of in vitro tests for predicting skin irritancy, including tests employing three human skin models, EPISKIN (Cotovio et al., 2005, Fentem et al., 2001, Portes et al., 2002, Zuang et al., 2002), EpiDerm (Fentem et al., 2001, Kandarova et al., 2004, Kandarova et al., 2005, Zuang et al., 2002), and SkinEthic RHE (Kandarova et al., 2006). When skin models are used to evaluate skin irritancy, it is important that a suitable chemical application procedure is utilized. We originally developed the postincubation (PI) method (10-minutes chemical treatment and 18-hours post-treatment incubation), a novel chemical application procedure, to predict skin irritancy (Morikawa et al., 2002, Morikawa et al., 2005, Morota et al., 1998, Morota et al., 1999). This procedure is similar to a refined protocol for EPISKIN proposed in an ECVAM validation study, based mainly on 15-minutes chemical treatment and a post-treatment incubation period of 18 hours (Cotovio et al., 2005). An optimized protocol based on 15-minutes chemical treatment and a post-treatment incubation period of 42 hours improved the predictability of skin irritancy. In addition, to improve the MTT viability-based predic- 11

2 tion model, the release of a membrane damage marker, adenylate kinase, and of cytokines, interleukin (IL)-1α and IL-8, and combinations of these endpoints, were also investigated. In this study, we evaluated the skin irritancy of 44 chemicals using both of the original PI procedure (10-minutes treatment and 18-hours post-treatment incubation) and a slightly modified procedure (10-minutes treatment and 42-hours post-treatment incubation) by MTT assay, IL-1α release, IL-8 release, and glucose consumption. Materials and Methods Materials A total of 44 test chemicals were selected from the panel of 48 chemicals tested in the ECVAM skin irritancy validation study (Cotovio et al., 2005). The test chemicals, 18 of which are known to be irritant (I) in vivo while 26 are nonirritant (NI), are listed in Table 1. Many types of chemical were included whenever possible, in order to cover various physical state (liquid, solid, semi-solid, powder), and a wide range of chemical classes and irritation potencies. All were purchased from Aldrich (Milwaukee, WI, USA) and Acros Organics (Morris Plains, NJ, USA), and used as received (The levels of purity were between 95% and 99.95%.). Dulbecco s modified Eagle s medium (DMEM) and fetal bovine serum (FBS) were purchased from Gibco Laboratories (Grand Island, NY, USA). Dulbecco s phosphate-buffered saline (PBS) and 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT) were obtained from Wako Pure Chemical Industries (Osaka, Japan) and Dojindo Laboratories (Kumamoto, Japan), respectively. The Vitrolife-Skin human skin model, consisting of dermis and epidermis with cornified layers, was prepared as previously described (Morikawa et al., 2002, Morota et al., 1998, Morota et al., 1999). Chemical application procedure by the postincubation (PI) method The experimental steps were performed using the PI method with slightly modifications, as described previously (Morikawa et al., 2002, Morikawa et al., 2005, Morota et al., 1998, Morota et al., 1999). The Vitrolife-Skin models were placed in 250 µl of DMEM + 5% FBS in 24-well plates and equilibrated in a 1-hour incubation (37 C, 5% CO 2, 90% humidity). Each test chemical was applied directly to the stratum corneum (8 mm-diameter area) of six replicate models. Liquids (100 µl) were applied using a positive displacement pipette. Solids were crushed to a powder, if necessary, and 50 mg were applied using a spatula; 50 µl of distilled water were added to ensure good contact with the surface. Six replicate models to which 100 µl distilled water alone had been added served as a negative control. After exposure for 10 minutes at room temperature, the models were washed thoroughly with PBS to remove the test chemical from the tissue surface. Each model was then immersed in 1.5 ml of DMEM + 5% FBS in a 24-well plate and 6 replicate models were cultured in triplicate for an additional 18 hours or 42 hours at 37 C in a 5% CO 2, 90% humidity environment. These application procedures (10-minutes chemical exposure followed by an additional 18-hour or 42-hour postincubation) are similar to the refined and optimized EPISKIN skin irritation protocols, respectively, in the ECVAM validation study (Cotovio et al., 2005). Calculation of cell viability Effects of test chemicals on cell viability were determined using the MTT reduction assay. After blotting, the models were incubated (5% CO 2, 90% humidity) in 1.0 ml of DMEM + 5% FBS containing 0.5 mg MTT for an additional 3 hours at 37 C. After washing with PBS, biopsies of the models were taken using a biopsy punch (6 mm diameter). Biopsies were separated from the models using forceps, and placed in isopropanol (1.0 ml), after the removal of excess water with absorbent paper. Precipitated formazan was extracted overnight at room temperature, with protection from light and the absorbance of extracts was measured at 570 nm using a UV-VIS spectrophotometer (UV-160A, Shimadzu, Kyoto, Japan). Cell viability was expressed relative to that in negative controls. Quantification of cytokines Proinflammatory cytokine release was determined, as previously described (Morota et al., 1999). At the end of the postincubation period, culture medium from each treated or negative control model was collected in a polypropylene vial and stored at 30 C until assayed. Medium samples were diluted as appropriate with fresh culture medium (DMEM + 5% FBS), and human IL-1α and IL-8 concentrations were determined using commercially available ELISA kits (Pierce Endogen, Rockford, IL, USA). 12

3 Table 1 Test chemicals. No. Chemical CAS No. 1) Chemical type Form 2) In vivo class 3) PII 4) 1 Sodium lauryl sulfate (20%, aq.) Soaps/surfactant L I Sodium lauryl sulfate (50%, aq.) Soaps/surfactant L I Tetrachloroethylene Chlorinated L I Pottasium hydroxide (5%, aq.) Alkali L I Heptanal Aldehyde L I Lilestralis/lilial Aldehyde L I 4.5/3.5 7 Methyl palmitate Ester S I Bromopentane Brominated L I α-terpineol Alcohol L I 4.4/4.7/4 10 β-citronellol Alcohol L I 4.2/4/ Bromohexane Brominated L I Methyl laurate Ester L I Cinnamaldehyde Aldehyde L I Linalyl acetate Ester L I 3.6/ d-limonene Miscellaneous L I 3.5/ Eugenol Phenolic derivative L I Undecenoic acid Organic acid L I Linalool Alcohol L I 3.3/3.4/ Dimethyl disulfide Sulfur containing L NI Methyl stearate Ester S NI Benzyl alcohol Alcohol L NI 1.5/ cis-cyclooctene Hydrocarbon L NI Ethoxyethyl methacrylate Acrylate/ methacrylate L NI Benzyl benzoate Ester L NI Methyl-4-phenyl-2-butanol Alcohol L NI Benzyl acetate Ester L NI 1.5/ Isopropyl palmitate Ester L NI ,4-Xylidine Amine L NI Sodium metasilicate (10%, aq.) Alkali L NI Isopropyl myristate Ester L NI Hydroxycitronellal Aldehyde L NI 1.1/ n-butyl propionate Ester L NI Sodium bisulfite Inorganic S NI ,6-Dibromohexane Brominated derivative L NI Propanol Alcohol L NI Benzyl salicylate Ester L NI Lauric acid Fatty acid S NI Dipropylene glycol Alcohol L NI Sodium bicarbonate Alkali S NI ,3'-Dithiopropionic acid Sulfur containing S NI ,4'-Methylenebis(2,6-di-tert-butylphenol) Phenolic L NI Amino-1,2,4-triazole Miscellaneous S NI Chloro-3-nitrobenzene Halogenated aromatic S NI Erucamide Amide L NI 0.0 1) CAS No. = Chemical Abstracts Service Registry number. 2) S = solid, L = liquid. 3) I = irritant, NI = non irritant (Cotovio et al., 2005). 4) PII = primary irritation index (Cotovio et al., 2005). 13

4 Quantification of glucose consumption At the end of the postincubation period, culture medium from each treated or negative control model was collected in a polypropylene vial and stored at 30 C until assayed. Glucose concentrations were determined using a commercially available colorimetric enzyme assay kit, Glucose C II-Test Wako (Wako Pure Chemical Industries). Prediction models Predictions of in vitro irritancy/nonirritancy were made according to the refined and optimized prediction models used in the EPISKIN skin irritancy test (Cotovio et al., 2005). Chemicals that reduced cell viability to less than 50% upon exposure to the Vitrolife-Skin model are thus predicted to be irritant in vivo. Results Skin irritancy test using MTT reduction assay Results of cell viability and classification predictions (50% threshold) are shown in Table 2. Comparisons of cell viability after exposure to test chemicals for a 10-minute treatment period followed by 18-hour (10 min/18 hr) or 42-hour (10 min/42 hr) post-treatment periods, together with the primary irritation index (PII), are shown in Figs. 1 and 2, respectively. Using the 10 min/18 hr pro 120 Linalyl acetate Methyl palmitate Cell viability (% of control) Sodium bisulfite r = ,6-Dibromohexane Methyl laurate d -Limonene Primary irritation index (PII) Fig. 1 Cell viability after exposure to test chemicals for 10 minutes followed by 18-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 120 Cell viability (% of control) Sodium bisulfite r = Methyl laurate Linalyl acetate Methyl palmitate ,6-Dibromohexane d -Limonene Primary irritation index (PII) Fig. 2 Cell viability after exposure to test chemicals for 10 minutes followed by 42-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 14

5 Table 2 Results of MTT cell viability assays and classification predictions. No. Chemical 10 min/18 hr 1) 10 min/42 hr 2) Viability (%) In vitro class 3) Viability (%) 1 Sodium lauryl sulfate (20%, aq.) 1.8 ± 0.1 I 1.6 ± 0.2 I 2 Sodium lauryl sulfate (50%, aq.) 0.6 ± 0.1 I 0.9 ± 0.2 I 3 Tetrachloroethylene 3.9 ± 1.5 I 5.5 ± 3.3 I 4 Pottasium hydroxide (5%, aq.) 1.3 ± 0.0 I 0.4 ± 0.2 I 5 Heptanal 9.7 ± 2.8 I 6.6 ± 0.9 I 6 Lilestralis/lilial 4.8 ± 0.8 I 5.4 ± 0.4 I 7 Methyl palmitate ± 11.4 NI 98.9 ± 17.7 NI 8 1-Bromopentane 5.9 ± 0.7 I 4.2 ± 0.9 I 9 α-terpineol 2.6 ± 0.0 I 3.8 ± 0.2 I 10 β-citronellol 4.7 ± 0.3 I 3.8 ± 0.6 I 11 1-Bromohexane 16.0 ± 15.9 I 3.8 ± 0.8 I 12 Methyl laurate ± 4.3 NI 74.2 ± 4.7 NI 13 Cinnamaldehyde 9.5 ± 1.1 I 7.0 ± 1.3 I 14 Linalyl acetate ± 6.6 NI 52.5 ± 30.2 NI 15 d-limonene 51.0 ± 20.7 NI 25.5 ± 12.4 I 16 Eugenol 30.1 ± 5.6 I 26.3 ± 4.6 I Undecenoic acid 13.0 ± 13.0 I 2.0 ± 0.2 I 18 Linalool 5.1 ± 0.2 I 2.7 ± 0.3 I 19 Dimethyl disulfide 2.1 ± 0.1 I 14.2 ± 14.9 I 20 Methyl stearate ± 1.3 NI ± 37.0 NI 21 Benzyl alcohol 85.1 ± 31.3 NI 98.7 ± 27.4 NI 22 cis-cyclooctene 5.3 ± 0.3 I 3.2 ± 0.5 I 23 2-Ethoxyethyl methacrylate 81.0 ± 8.1 NI 62.1 ± 8.9 NI 24 Benzyl benzoate ± 7.9 NI 74.5 ± 5.1 NI 25 2-Methyl-4-phenyl-2-butanol 2.4 ± 0.6 I 3.2 ± 0.5 I 26 Benzyl acetate 99.9 ± 4.7 NI 94.1 ± 13.6 NI 27 Isopropyl palmitate ± 10.1 NI 77.8 ± 8.3 NI 28 2,4-Xylidine 18.7 ± 11.4 I 4.5 ± 0.4 I 29 Sodium metasilicate (10%, aq.) 2.0 ± 0.3 I 1.1 ± 0.2 I 30 Isopropyl myristate ± 6.1 NI 71.9 ± 9.0 NI 31 Hydroxycitronellal 17.0 ± 5.0 I 3.6 ± 0.4 I 32 n-butyl propionate 83.7 ± 23.3 NI 83.6 ± 12.9 NI 33 Sodium bisulfite 63.5 ± 20.9 NI 41.6 ± 11.8 I 34 1,6-Dibromohexane 55.5 ± 11.2 NI 7.2 ± 0.8 I 35 2-Propanol 89.5 ± 6.0 NI 84.8 ± 10.8 NI 36 Benzyl salicylate ± 4.6 NI 77.0 ± 7.2 NI 37 Lauric acid 94.0 ± 5.5 NI 78.9 ± 6.0 NI 38 Dipropylene glycol ± 1.2 NI 83.5 ± 8.8 NI 39 Sodium bicarbonate ± 5.1 NI 81.5 ± 4.2 NI 40 3,3'-Dithiopropionic acid ± 26.1 NI 80.2 ± 6.5 NI 41 4,4'-Methylenebis(2,6-di-tert-butylphenol) ± 29.3 NI 84.0 ± 3.8 NI 42 4-Amino-1,2,4-triazole 92.5 ± 9.3 NI 81.2 ± 23.5 NI 43 1-Chloro-3-nitrobenzene 90.1 ± 2.9 NI ± 23.7 NI 44 Erucamide ± 23.9 NI 70.6 ± 8.9 NI 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) I = irritant (viability < 50 %), NI = nonirritant (viability > 50%). In vitro class 3) 15

6 tocol, 14 of the 18 chemicals known to be irritant in vivo were correctly classified as such in vitro. Among these, d-limonene (number 15) appeared borderline irritant, with a viability result of 51.0%, close to the 50% defined threshold. In contrast, the viability observed following incubation with methyl palmitate (number 7), methyl laurate (number 12), and linalyl acetate (number 14) was close to 100%. In comparison, 15 of the 18 chemicals classified as irritant in vivo were correctly classified as such using the 10 min/42 hr protocol. d-limonene (number 15) for which a borderline result had been obtained using the 10 min/18 hr 30 protocol was correctly identified as irritant, with viability clearly under the 50% threshold (25.5%). A borderline result close to the 50% defined threshold (52.5%) was obtained for linalyl acetate (number 14), for which viability remained very high using the 10 min/18 hr protocol. Of the 26 chemicals known to be nonirritant in vivo, 20 were correctly classified as such using the 10 min/18 hr protocol. Sodium bisulfite (number 33) and 1,6-dibromohexane (number 34), correctly classified as nonirritant using the 10 min/18 hr protocol, were classified as irritant using the 10 min/42 hr protocol. IL-1α release (fold control) n -Butyl propionate d -Limonene Linalyl acetate Methyl laurate r = Primary irritation index (PII) Fig. 3 IL-1α release after exposure to test chemicals for 10 minutes followed by 18-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 9 8 n -Butyl propionate IL-1α release (fold control) Linalyl acetate d -Limonene r = Methyl laurate Primary irritation index (PII) Fig. 4 IL-1α release after exposure to test chemicals for 10 minutes followed by 42-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 16

7 Table 3 Results of interleukin-1α (IL-1α) release assays and classification predictions. 10 min/18 hr 1) 10 min/42 hr 2) No. Chemical In vitro In vitro IL-1α release 3) IL-1α release 3) class 4) class 4) 1 Sodium lauryl sulfate (20%, aq.) 6.06 ± 1.75 I 4.10 ± 2.00 I 2 Sodium lauryl sulfate (50%, aq.) 1.83 ± 0.74 NI 0.86 ± 0.17 NI 3 Tetrachloroethylene ± 1.44 I 7.99 ± 1.00 I 4 Pottasium hydroxide (5%, aq.) 1.16 ± 0.08 NI 0.76 ± 0.00 NI 5 Heptanal 3.27 ± 1.47 NI 5.25 ± 0.69 I 6 Lilestralis/lilial 3.28 ± 0.57 NI 3.13 ± 0.53 NI 7 Methyl palmitate 2.01 ± 0.09 NI 1.03 ± 0.12 NI 8 1-Bromopentane 5.21 ± 0.92 I 3.33 ± 1.03 NI 9 α-terpineol 2.84 ± 0.27 NI 7.03 ± 0.71 I 10 β-citronellol 6.04 ± 1.09 I 4.02 ± 1.92 I 11 1-Bromohexane 6.25 ± 0.27 I 3.76 ± 0.99 I 12 Methyl laurate 6.64 ± 0.37 I 0.95 ± 0.27 NI 13 Cinnamaldehyde 1.29 ± 0.06 NI 0.86 ± 0.04 NI 14 Linalyl acetate 3.57 ± 1.07 I 6.10 ± 0.91 I 15 d-limonene 5.53 ± 1.55 I 5.46 ± 2.48 I 16 Eugenol ± 4.11 I 2.56 ± 0.43 NI Undecenoic acid 7.19 ± 0.95 I 6.90 ± 3.21 I 18 Linalool 4.07 ± 0.11 I 7.18 ± 0.84 I 19 Dimethyl disulfide 1.68 ± 0.49 NI 2.79 ± 1.17 NI 20 Methyl stearate 1.00 ± 0.11 NI 0.66 ± 0.26 NI 21 Benzyl alcohol 1.21 ± 0.23 NI 1.93 ± 0.25 NI 22 cis-cyclooctene 3.15 ± 0.13 NI 3.77 ± 0.78 I 23 2-Ethoxyethyl methacrylate 1.10 ± 0.44 NI 2.72 ± 0.44 NI 24 Benzyl benzoate 1.17 ± 0.36 NI 0.17 ± 0.24 NI 25 2-Methyl-4-phenyl-2-butanol 2.91 ± 0.90 NI 2.77 ± 0.77 NI 26 Benzyl acetate 1.64 ± 0.38 NI 1.95 ± 0.59 NI 27 Isopropyl palmitate 0.96 ± 0.14 NI 0.17 ± 0.19 NI 28 2,4-Xylidine 8.69 ± 3.33 I 2.39 ± 0.32 NI 29 Sodium metasilicate (10%, aq.) 0.25 ± 0.04 NI 1.16 ± 0.02 NI 30 Isopropyl myristate 1.09 ± 0.09 NI 0.28 ± 0.07 NI 31 Hydroxycitronellal 1.55 ± 0.30 NI 1.04 ± 0.18 NI 32 n-butyl propionate 7.72 ± 0.99 I 7.70 ± 2.59 I 33 Sodium bisulfite 3.29 ± 1.00 NI 2.99 ± 0.04 NI 34 1,6-Dibromohexane 2.49 ± 0.38 NI 4.00 ± 0.91 I 35 2-Propanol 2.56 ± 0.12 NI 2.77 ± 0.53 NI 36 Benzyl salicylate 2.55 ± 0.29 NI 2.97 ± 0.82 NI 37 Lauric acid 1.25 ± 0.12 NI 1.10 ± 0.09 NI 38 Dipropylene glycol 2.47 ± 0.06 NI 3.47 ± 1.09 NI 39 Sodium bicarbonate 1.63 ± 0.35 NI 1.06 ± 0.25 NI 40 3,3'-Dithiopropionic acid 1.34 ± 0.24 NI 1.42 ± 0.46 NI 41 4,4'-Methylenebis(2,6-di-tert-butylphenol) 1.13 ± 0.20 NI 1.28 ± 0.26 NI 42 4-Amino-1,2,4-triazole 2.20 ± 0.23 NI 1.48 ± 0.56 NI 43 1-Chloro-3-nitrobenzene 0.88 ± 0.16 NI 1.00 ± 0.11 NI 44 Erucamide 0.75 ± 0.16 NI 0.63 ± 0.27 NI 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) IL-1α release; results are expressed as fold increase release to control; control mean ± SD release = ± 44.8 pg/18 hr and 87.3 ± 23.8 pg/42 hr. 4) I = irritant (IL-1α release > 3.5-fold), NI = nonirritant (IL-1α release < 3.5-fold). 17

8 Skin irritancy test using IL-1α release assay Results of IL-1α release assays are expressed relative to negative controls. Comparisons of IL-1α release after exposure to test chemicals for 10 minutes followed by 18-hour or 42-hour postincubation periods, together with PII values, are shown in Figs. 3 and 4, respectively. Most chemicals nonirritant in vivo produced a less than 3.5-fold increase in IL-1α release. In this study, therefore, a less than 3.5-fold increase in IL-1α release was considered indicative of a nonirritant chemical. The results of IL-1α release assays and classification predictions are shown in Table 3. Methyl laurate (number 12), linalyl acetate (number 14), and d-limonene (number 15), wrongly identified as nonirritant by the MTT assay, were correctly classified as irritant in vitro by IL-1α release assay using the 10 min/18 hr protocol. Additionally, linalyl acetate (number 14), wrongly identified as nonirritant by the MTT assay, was correctly classified as irritant in vitro by IL-1α release assay using the 10 min/42 hr protocol. On the other hand, n-butyl propionate (number 32), correctly identified as nonirritant by the MTT assay, was wrongly classified as irritant in vitro by IL-1α release assay in both the 10 min/18 hr protocol and 10 min/42 hr protocols. 2.5 IL-8 release (fold control) r = Methyl laurate Linalyl acetate Methyl palmitate d -Limonene Primary irritation index (PII) Fig. 5 IL-8 release after exposure to test chemicals for 10 minutes followed by 18-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). IL-8 release (fold control) Linalyl acetate Methyl laurate r = Methyl palmitate Primary irritation index (PII) Fig. 6 IL-8 release after exposure to test chemicals for 10 minutes followed by 42-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 18

9 Table 4 Results of interleukin-8 (IL-8) release assays and classification predictions. 10 min/18 hr 1) 10 min/42 hr 2) No. Chemical In vitro In vitro IL-8 release 3) class 4) IL-8 release 3) class 4) 1 Sodium lauryl sulfate (20%, aq.) 0.33 ± 0.22 I 0.11 ± 0.05 I 2 Sodium lauryl sulfate (50%, aq.) 0.17 ± 0.02 I 0.13 ± 0.02 I 3 Tetrachloroethylene 0.12 ± 0.08 I 0.15 ± 0.05 I 4 Pottasium hydroxide (5%, aq.) 1.16 ± 0.04 I 0.16 ± 0.03 I 5 Heptanal 0.15 ± 0.04 I 0.13 ± 0.01 I 6 Lilestralis/lilial 0.31 ± 0.07 I 0.10 ± 0.01 I 7 Methyl palmitate 1.26 ± 0.23 NI 0.94 ± 0.67 NI 8 1-Bromopentane 0.25 ± 0.07 I 0.12 ± 0.03 I 9 α-terpineol 0.08 ± 0.01 I 0.13 ± 0.00 I 10 β-citronellol 0.37 ± 0.15 I 0.13 ± 0.04 I 11 1-Bromohexane 0.45 ± 0.16 I 0.14 ± 0.02 I 12 Methyl laurate 1.52 ± 0.02 NI 5.75 ± 1.14 NI 13 Cinnamaldehyde 0.03 ± 0.00 I 0.11 ± 0.01 I 14 Linalyl acetate 1.36 ± 0.37 NI 8.48 ± 3.68 NI 15 d-limonene 1.21 ± 0.82 NI 0.25 ± 0.11 I 16 Eugenol 0.09 ± 0.01 I 0.12 ± 0.01 I Undecenoic acid 0.19 ± 0.05 I 0.24 ± 0.01 I 18 Linalool 0.26 ± 0.02 I 0.14 ± 0.01 I 19 Dimethyl disulfide 0.10 ± 0.03 I 0.32 ± 0.38 I 20 Methyl stearate 0.79 ± 0.32 NI 5.39 ± 1.30 NI 21 Benzyl alcohol 1.08 ± 0.66 NI 3.03 ± 0.37 NI 22 cis-cyclooctene 0.32 ± 0.30 I 0.12 ± 0.00 I 23 2-Ethoxyethyl methacrylate 1.07 ± 0.04 NI 2.56 ± 0.89 NI 24 Benzyl benzoate 1.63 ± 0.09 NI 3.46 ± 1.36 NI 25 2-Methyl-4-phenyl-2-butanol 0.18 ± 0.00 I 0.13 ± 0.01 I 26 Benzyl acetate 1.43 ± 0.31 NI 3.38 ± 0.70 NI 27 Isopropyl palmitate 1.32 ± 0.51 NI 1.45 ± 0.56 NI 28 2,4-Xylidine 0.26 ± 0.09 I 0.13 ± 0.04 I 29 Sodium metasilicate (10%, aq.) 0.04 ± 0.01 I 0.52 ± 0.16 I 30 Isopropyl myristate 1.64 ± 0.08 NI 2.87 ± 0.95 NI 31 Hydroxycitronellal 0.04 ± 0.01 I 0.07 ± 0.00 I 32 n-butyl propionate 1.36 ± 0.12 NI 1.58 ± 0.38 NI 33 Sodium bisulfite 0.85 ± 0.42 NI 1.09 ± 0.33 NI 34 1,6-Dibromohexane 0.35 ± 0.09 I 0.10 ± 0.01 I 35 2-Propanol 1.65 ± 0.06 NI 3.84 ± 1.59 NI 36 Benzyl salicylate 1.68 ± 0.11 NI 7.04 ± 1.62 NI 37 Lauric acid 2.13 ± 0.46 NI 1.79 ± 0.70 NI 38 Dipropylene glycol 1.40 ± 0.31 NI 2.62 ± 0.46 NI 39 Sodium bicarbonate 1.35 ± 0.34 NI 2.97 ± 1.73 NI 40 3,3'-Dithiopropionic acid 1.29 ± 0.11 NI 4.51 ± 1.53 NI 41 4,4'-Methylenebis(2,6-di-tert-butylphenol) 1.20 ± 0.22 NI 4.13 ± 2.30 NI 42 4-Amino-1,2,4-triazole 1.26 ± 0.28 NI 2.99 ± 0.90 NI 43 1-Chloro-3-nitrobenzene 1.30 ± 0.10 NI 1.94 ± 1.11 NI 44 Erucamide 1.30 ± 0.45 NI 3.25 ± 1.91 NI 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) IL-8 release; results are expressed as fold increase release to control; control mean ± SD release = ± ng/18 hr and ± ng/42 hr. 4) I = irritant (IL-8 release < 0.7-fold), NI = nonirritant (IL-8 release > 0.7-fold). 19

10 Table 5 Results of glucose consumption assays and classification predictions. No. Chemical glucose consumption 3) 10 min/18 hr 1) 10 min/42 hr 2) In vitro class 4) glucose consumption 3) 1 Sodium lauryl sulfate (20%, aq.) 0.24 ± 0.08 I 0.09 ± 0.08 I 2 Sodium lauryl sulfate (50%, aq.) 0.09 ± 0.08 I 0.16 ± 0.03 I 3 Tetrachloroethylene 0.28 ± 0.06 I 0.11 ± 0.03 I 4 Pottasium hydroxide (5%, aq.) 0.48 ± 0.08 I 0.26 ± 0.10 I 5 Heptanal 0.17 ± 0.01 I 0.13 ± 0.02 I 6 Lilestralis/lilial 0.52 ± 0.07 I 0.16 ± 0.05 I 7 Methyl palmitate 1.00 ± 0.08 NI 1.15 ± 0.06 NI 8 1-Bromopentane 0.24 ± 0.06 I 0.12 ± 0.04 I 9 α-terpineol 0.12 ± 0.07 I 0.15 ± 0.05 I 10 β-citronellol 0.06 ± 0.12 I 0.07 ± 0.02 I 11 1-Bromohexane 0.37 ± 0.10 I 0.28 ± 0.03 I 12 Methyl laurate 1.04 ± 0.08 NI 1.15 ± 0.06 NI 13 Cinnamaldehyde 0.20 ± 0.05 I 0.17 ± 0.06 I 14 Linalyl acetate 0.69 ± 0.01 NI 0.73 ± 0.17 NI 15 d-limonene 0.69 ± 0.10 NI 0.28 ± 0.10 I 16 Eugenol 0.55 ± 0.02 I 0.39 ± 0.06 I Undecenoic acid 0.50 ± 0.14 I 0.23 ± 0.03 I 18 Linalool 0.13 ± 0.03 I 0.10 ± 0.03 I 19 Dimethyl disulfide 0.13 ± 0.01 I 0.19 ± 0.20 I 20 Methyl stearate 1.17 ± 0.14 NI 0.98 ± 0.03 NI 21 Benzyl alcohol 0.72 ± 0.16 NI 0.66 ± 0.04 NI 22 cis-cyclooctene 0.30 ± 0.21 I 0.14 ± 0.02 I 23 2-Ethoxyethyl methacrylate 0.92 ± 0.05 NI 0.70 ± 0.05 NI 24 Benzyl benzoate 1.10 ± 0.04 NI 0.99 ± 0.03 NI 25 2-Methyl-4-phenyl-2-butanol 0.20 ± 0.04 I 0.12 ± 0.00 I 26 Benzyl acetate 0.73 ± 0.06 NI 0.71 ± 0.05 NI 27 Isopropyl palmitate 1.10 ± 0.05 NI 0.97 ± 0.07 NI 28 2,4-Xylidine 0.76 ± 0.10 NI 0.33 ± 0.08 I 29 Sodium metasilicate (10%, aq.) 0.30 ± 0.08 I 0.48 ± 0.05 I 30 Isopropyl myristate 1.02 ± 0.01 NI 0.96 ± 0.06 NI 31 Hydroxycitronellal 0.26 ± 0.10 I 0.08 ± 0.09 I 32 n-butyl propionate 0.54 ± 0.09 I 0.38 ± 0.04 I 33 Sodium bisulfite 0.77 ± 0.02 NI 0.68 ± 0.07 NI 34 1,6-Dibromohexane 0.49 ± 0.03 I 0.41 ± 0.13 I 35 2-Propanol 1.01 ± 0.04 NI 0.99 ± 0.00 NI 36 Benzyl salicylate 1.03 ± 0.05 NI 0.97 ± 0.02 NI 37 Lauric acid 0.98 ± 0.02 NI 1.04 ± 0.03 NI 38 Dipropylene glycol 0.98 ± 0.02 NI 0.97 ± 0.02 NI 39 Sodium bicarbonate 1.06 ± 0.03 NI 1.05 ± 0.04 NI 40 3,3'-Dithiopropionic acid 0.96 ± 0.06 NI 1.03 ± 0.01 NI 41 4,4'-Methylenebis(2,6-di-tert-butylphenol) 1.08 ± 0.08 NI 1.05 ± 0.01 NI 42 4-Amino-1,2,4-triazole 0.98 ± 0.07 NI 1.04 ± 0.03 NI 43 1-Chloro-3-nitrobenzene 1.09 ± 0.03 NI 1.04 ± 0.04 NI 44 Erucamide 1.14 ± 0.12 NI 0.99 ± 0.05 NI 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) Glucose consumption; results are expressed as fold increase consumption to control; control mean ± SD consumption = 2.62 ± 0.29 mg/18 hr and 4.08 ± 0.21 mg/42 hr. 4) I = irritant (glucose consumption < 0.6-fold), NI = nonirritant (glucose consumption > 0.6-fold). In vitro class 4) 20

11 Glucose consumption (fold control) Linalyl acetate r = Skin irritancy test using IL-8 release assay Results of IL-8 release assays are expressed relative to negative controls. Comparisons of IL-8 release after exposure to test chemicals for 10 minutes followed by 18-hour or 42-hour postincubation periods, together with PII values, are shown in Figs. 5 and 6, respectively. Most chemicals irritant of in vivo produced a less than 0.7-fold increase in IL-8 release. In this study, therefore, a less than 0.7-fold increase in IL-8 release was considered indicative of an irritant chemical. The results of IL-8 release assays and classification predictions are shown in Table 4. Using the 10 min/18 hr protocol, 14 of the 18 chemicals known to be irritant in vivo were correctly classified as such in vitro. Methyl palmitate (number 7), methyl laurate (number 12), linalyl acetate (number 14), and d-limonene (number 15), wrongly identified as nonirritant by the MTT assay, were similarly wrongly classified by the IL-8 release assay in vitro with the 10 min/18 hr protocol. Methyl palmitate (number 7), methyl laurate (number 12), and linalyl acetate (number 14) were also wrongly classified by the IL-8 release assay with the 10 min/42 hr protocol. Methyl laurate d -Limonene Methyl palmitate PII Fig. 7 Glucose consumption after exposure to test chemicals for 10 minutes followed by 18-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 1.2 Methyl laurate Methyl palmitate Glucose consumption (fold control) r = Linalyl acetate PII Fig. 8 Glucose consumption after exposure to test chemicals for 10 minutes followed by 42-hours postincubation, together with PII values. Closed circles, in vivo irritant (I); open circles, in vivo nonirritant (NI). 21

12 Skin irritancy test using glucose consumption assay Results of glucose consumption assays are expressed relative to negative controls. Comparisons of glucose consumption after exposure to test chemicals for 10 minutes followed by 18-hour or 42-hour postincubation periods, together with PII values, are shown in Figs. 7 and 8, respectively. Most chemicals irritant in vivo produced a less than 0.6-fold increase in glucose consumption. In this study, therefore, a less than 0.6-fold increase in glucose consumption value was considered indicative of an irritant chemical. The results of glucose consumption assays and classification predictions are shown in Table 5. Using the 10 min/18 hr protocol, 14 of the 18 chemicals known to be irritant in vivo were correctly classified as such in vitro. Methyl palmitate (number 7), methyl laurate (number 12), linalyl acetate (number 14), and d-limonene (number 15), wrongly identified as nonirritant by the MTT assay, were similarly wrongly classified by the glucose consumption assay in vitro using the 10 min/18 hr protocol. Methyl palmitate (number 7), methyl laurate (number 12), and linalyl acetate (number 14) were also wrongly classified by the glucose consumption assay with the 10 min/42 hr protocol. Discussion Table 6 shows the key performance parameters for the two protocols. The sensitivity of the MTT assay with the 10 min/42 hr protocol increased to 83.3% from 77.8% (10 min/18 hr protocol), thus showing an improvement. On the other hand, its specificity decreased from 76.9% (10 min/18 hr protocol) to 69.2%. Its accuracy was also slightly decreased from 77.3% (10 min/18 hr protocol) to 75.0%. In the EPISKIN validation study, the sensitivity, specificity, and accuracy of the MTT assay were improved from 75.0%, 75.0%, and 75.0% (15 min/18 hr protocol) to 85.0%, 78.6%, and 81.3% (15 min/42 hr protocol), respectively (Cotovio et al., 2005). The difference between these results is considered to be due to difference in the barrier function between EPISKIN and Vitrolife-Skin, or to difference in the chemical application procedures (chemical volume and washing method). Although the surface area of the Vitrolife-Skin (0.50 cm 2 ) is similar to that of EPISKIN (0.38 cm 2 ), 10 µl of liquid chemical Table 6 Comparison of statistical parameters for key performance indicators using MTT, IL-1α, IL-8, and glucose individually and in combination. Parameter 10 min/18 hr 1) 10 min/42 hr 2) MTT 3) IL-1α 4) MTT + IL-1α 5) IL-8 6) glucose 7) MTT 3) IL-1α 4) MTT + IL-1α 5) IL-8 6) glucose 7) Sensitivity 14/18 11/18 17/18 14/18 14/18 15/18 10/18 16/18 15/18 15/ % 61.1% 94.4% 77.8% 77.8% 83.3% 55.6% 88.9% 83.3% 83.3% Specificity 20/26 24/26 19/26 19/26 19/26 18/26 23/26 17/26 19/26 18/ % 92.3% 73.1% 73.1% 73.1% 69.2% 88.5% 65.4% 73.1% 69.2% Accuracy 34/44 35/44 36/44 33/44 33/44 33/44 33/44 33/44 34/44 33/ % 79.5% 81.8% 75.0% 75.0% 75.0% 75.0% 75.0% 77.3% 75.0% Positive 14/20 11/13 17/24 14/21 14/21 15/23 10/13 16/25 15/22 15/23 predictivity 70.0% 84.6% 70.8% 66.7% 66.7% 65.2% 76.9% 64.0% 68.2% 65.2% Negative 20/24 24/31 19/20 19/23 19/23 18/21 23/31 17/19 19/22 18/21 predictivity 83.3% 77.4% 95.0% 82.6% 82.6% 85.7% 74.2% 89.5% 86.4% 85.7% False negatives 4/24 7/31 1/20 4/23 4/23 3/21 8/31 2/19 3/22 3/ % 22.6% 5.0% 17.4% 17.4% 14.3% 25.8% 10.5% 13.6% 14.3% False positives 6/20 2/13 7/24 7/21 7/21 8/23 3/13 9/25 7/22 8/ % 15.4% 29.2% 33.3% 33.3% 34.8% 23.1% 36.0% 31.8% 34.8% 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) MTT; 50 % viability cut-off (Table 2, Figures 1 & 2). 4) IL-1α; 3.5-fold release cut-off (Table 3, Figures 3 & 4). 5) MTT + IL-1α; Two-step selection (all chemicals first sorted according to MTT assay results and then by IL-1α release assay results of chemicals identified as nonirritant) according to the EPISKIN prediction model in the ECVAM validation study (Cotovio et al., 2005) (Table 7). 6) IL-8; 0.7-fold release cut-off (Table 4, Figures 5 & 6). 7) Glucose; 0.6-fold release cut-off (Table 5, Figures 7 & 8). 22

13 was insufficient to wet the surface. In this study, therefore, the application volume of liquids was increased to 100 µl. In the ECVAM validation study, a nylon mesh was placed on the surface of the tissue after the test chemical had been applied, to improve liquid spreading (Kandarova et al., 2004, Kandarova et al., 2005, Kandarova et al., 2006). Table 7 shows the classification predictions obtained from the MTT viability assay and the IL-1α release assay. The MTT viability assay was used as the first step, mainly separate irritant from nonirritant chemicals. Chemicals qualified for classification as irritant in the first step were defi- Table 7 Classification predictions obtained from MTT cell viability and IL-1α release assays in combination. No. Chemical Viability (MTT) min/18 hr 1) 10 min/42 hr 2) IL-1α release MTT + IL-1α 3) Viability (MTT) IL-1α release MTT + IL-1α 3) 1 Sodium lauryl sulfate (20%, aq.) I I I I I I 2 Sodium lauryl sulfate (50%, aq.) I NI I I NI I 3 Tetrachloroethylene I I I I I I 4 Pottasium hydroxide (5%, aq.) I NI I I NI I 5 Heptanal I NI I I I I 6 Lilestralis/lilial I NI I I NI I 7 Methyl palmitate NI NI NI NI NI NI 8 1-Bromopentane I I I I NI I 9 α-terpineol I NI I I I I 10 β-citronellol I I I I I I 11 1-Bromohexane I I I I I I 12 Methyl laurate NI I I NI NI NI 13 Cinnamaldehyde I NI I I NI I 14 Linalyl acetate NI I I NI I I 15 d-limonene NI I I I I I 16 Eugenol I I I I NI I Undecenoic acid I I I I I I 18 Linalool I I I I I I 19 Dimethyl disulfide I NI I I NI I 20 Methyl stearate NI NI NI NI NI NI 21 Benzyl alcohol NI NI NI NI NI NI 22 cis-cyclooctene I NI I I I I 23 2-Ethoxyethyl methacrylate NI NI NI NI NI NI 24 Benzyl benzoate NI NI NI NI NI NI 25 2-Methyl-4-phenyl-2-butanol I NI I I NI I 26 Benzyl acetate NI NI NI NI NI NI 27 Isopropyl palmitate NI NI NI NI NI NI 28 2,4-Xylidine I I I I NI I 29 Sodium metasilicate (10%, aq.) I NI I I NI I 30 Isopropyl myristate NI NI NI NI NI NI 31 Hydroxycitronellal I NI I I NI I 32 n-butyl propionate NI I I NI I I 33 Sodium bisulfite NI NI NI I NI I 34 1,6-Dibromohexane NI NI NI I I I 35 2-Propanol NI NI NI NI NI NI 36 Benzyl salicylate NI NI NI NI NI NI 37 Lauric acid NI NI NI NI NI NI 38 Dipropylene glycol NI NI NI NI NI NI 39 Sodium bicarbonate NI NI NI NI NI NI 40 3,3'-Dithiopropionic acid NI NI NI NI NI NI 41 4,4'-Methylenebis(2,6-di-tert-butylphenol) NI NI NI NI NI NI 42 4-Amino-1,2,4-triazole NI NI NI NI NI NI 43 1-Chloro-3-nitrobenzene NI NI NI NI NI NI 44 Erucamide NI NI NI NI NI NI 1) 10 min/18 hr; 10-minute treatment period followed by 18-hour post-treatment period. 2) 10 min/42 hr; 10-minute treatment period followed by 42-hour post-treatment period. 3) MTT + IL-1α; Two-step selection (all chemicals first sorted according to MTT assay results and then by IL-1α release assay results of chemicals identified as nonirritant) according to the EPISKIN prediction model in the ECVAM validation study (Cotovio et al., 2005).

14 nitely out of the process, and only nonirritant chemicals should qualify to go through the second step. After the first selection, nonirritant chemicals were subjected to a second step using IL-1α release assay in order to detect and discard possible false negatives. Using this two-step selection process, as described for the EPISKIN prediction model in the ECVAM validation study (Cotovio et al., 2005), this resulted in clear increases in both sensitivity and accuracy, to 94.4% and 81.8%, respectively, with the 10 min/18 hr protocol and slight increases in each, to 88.9% and 75.0%, respectively, with the 10 min/42 hr protocol (Table 6). These results are similar to those obtained in the EPISKIN validation study (Cotovio et al., 2005). Comparison of Tables 2 and 4 shows that classification on the basis of IL-8 release was quite similar to that on the basis of cell viability, except in the case of 1,6-dibromohexane (number 34) with the 10 min/18 hr protocol and sodium bisulfite (number 33) with the 10 min/42 hr protocol. Key statistical parameters for IL-8 release and cell viability were consequently also quite similar (Table 6). These results differ from those obtained in the EPISKIN validation study (Cotovio et al., 2005). IL-8 release was significantly correlated with cell viability (r = for the 10 min/18 hr protocol and r = for the 10 min/42 hr protocol) (Data not shown.). Conversely, IL-1α release was not so highly correlated with cell viability (r = for the 10 min/18 hr protocol and r = IL-1α release (pg) Post-treatment incubation time (hr) Fig. 9 Time course of IL-1α release (mean ± SD) from undamaged Vitrolife-Skin human skin model. IL-8 release (ng) Post-treatment incubation time (hr) Fig. 10 Time course of IL-8 release (mean ± SD) from undamaged Vitrolife-Skin human skin and Vitrolife-Dermis human dermal models. Closed circles, Vitrolife-Skin; open circles, Vitrolife-Dermis. 24

15 for the 10 min/42 hr protocol). As shown in Fig. 9, IL-1α is stored intracellularly as pro-il-1α and released only from damaged leaky cells, so that its release from undamaged Vitrolife-Skin was low. In contrast, IL-8 is not stored intracellularly, so time is required for its expression and subsequent release. As shown in Fig. 10, IL-8 release from undamaged Vitrolife-Skin increased gradually, while IL-8 release from Vitrolife-Dermis, comprising only dermis (fibroblasts), was constant. IL-8 release from the only epidermal (keratinocytes) model was lower (< 1 ng) than that from Vitrolife-Skin and Vitrolife-Dermis. IL-1α release from Vitrolife-Skin was similar to that from EPISKIN (Cotovio et al., 2005), whereas IL-8 release from Vitrolife-Skin and Vitrolife-Dermis was higher than that from EPISKIN (Cotovio et al., 2005). It is considered that the differences in IL-8 production profiles are caused to the differences in IL-8 assay results between Vitrolife-Skin and EPISKIN. Glucose consumption was significantly correlated with both cell viability (r = for the 10 min/18 hr protocol and r = for the 10 min/42 hr protocol) and IL-8 release (r = for the 10 min/18 hr protocol and r = for the 10 min/42 hr protocol) (Data not shown.). Comparison of Tables 2 and 5 shows that classification on the basis of glucose consumption was quite similar to that on the basis of cell viability, except in case of 2,4-xylidine (number 28), n-butyl propionate (number 32), and 1,6-dibromohexane (number 34) with the 10 min/18 hr protocol, and n-butyl propionate (number 32) and sodium bisulfite (number 33) with the 10 min/42 hr protocol. Comparison of Tables 4 and 5 shows that classification on the basis of glucose consumption was also quite similar to that on the basis of IL-8 release, except in the case of 2,4-xylidine (number 28) and n-butyl propionate (number 32) with the 10 min/18 hr protocol, and n-butyl propionate (number 32) with the 10 min/42 hr protocol. Key statistical parameters for glucose consumption, cell viability, and IL-8 release were also quite similar (Table 6). Glycolytic metabolism was demonstrated by increased glycolytic enzyme activity along with increased glucose consumption. Glucose consumption is thus considered to be reflected by cell viability. Methyl palmitate (number 7) and methyl laurate (number 12) (Table 2) were incorrectly classified as nonirritant by MTT assay in all cultured skin models and test protocols in which they were tested in the ECVAM validation study (Cotovio et al., 2005, Kandarova et al., 2005, Kandarova et al., 2006) and our previous study (Morikawa et al., 2005). Each of these chemicals belongs to the same chemical group, fatty acid methyl esters. Although they were predicted to be severe irritants by the rabbit skin test, they were either nonirritant or only very slightly irritating in the human patch test (Kandarova et al., 2005). Linalyl acetate (number 14) (Table 2) was also wrongly classified as nonirritant by MTT assay in some cultured skin models and test protocols in which it was tested in the ECVAM validation study (Cotovio et al., 2005, Kandarova et al., 2005). Dimethyl disulfide (number 19), 2-methyl-4- phenyl-2-butanol (number 25), 2,4-xylidine (number 28), and 1,6-dibromohexane (number 34) (Table 2) were also consistently incorrectly predicted to be irritants by MTT assay in all cultured skin models and test protocols in which they were tested in the ECVAM validation study (Cotovio et al., 2005, Kandarova et al., 2005, Kandarova et al., 2006). These chemicals are lipid-soluble or lipid solvents and therefore might readily pass through the skin lipid barrier and penetrate into the deeper layers of the epidermis (Kandarova et al., 2005). Dimethyl disulfide (number 19) was excluded from the statistical analysis due to insufficient evidence of a correct in vivo classification in some ECVAM validation studies (Kandarova et al., 2005, Kandarova et al., 2006). 2,4-Xylidine, a known carcinogen, is sensitive to oxidation and, depending on experimental conditions, can be subjected to structural alterations that could affect its activities in vitro and in vivo in different ways. This may explain its high cytotoxicity observed in most in vitro studies (Cotovio et al., 2005, Kandarova et al., 2005, Kandarova et al., 2006, Portes et al., 2002, Zuang et al., 2002). The irritancy of ciscyclooctene (number 22) was overpredicted using Vitrolife-Skin (Table 2), EPISKIN (Cotovio et al., 2005), and SkinEthic RHE (Kandarova et al., 2006) models, but it was classified as nonirritant using EpiDerm (Kandarova et al., 2005). Whereas benzyl acetate (number 26) was correctly classified as nonirritant using Vitrolife-Skin (Table 2) and EpiDerm (Kandarova et al., 2005) models, its nonirritancy was overpredicted using EPISKIN (Cotovio et al., 2005). In contrast, while the irritancy of sodium metasilicate (10%, aq.) (number 29) and hydroxycitronellal (number 31) were overpredicted using Vitrolife-Skin (Table 2), they were classified as nonirritant us- 25

16 ing EPISKIN (Cotovio et al., 2005) and Epi- Derm (Kandarova et al., 2005) models. Kandarova et al. discussed the interesting results obtained for hydroxycitronellal (number 31). They tested three different samples of this chemical and obtained three different predictions, from which they concluded that, although the purity of the samples was higher than 95%, impurities and/or the ages of the different samples must have been responsible for the different predictions (Kandarova et al., 2006). In conclusion, Vitrolife-Skin showed a basic utility for irritancy testing and it is possible to confidently predict skin irritancy provided that an appropriate chemical application procedure and a combination of suitable endpoints, such as MTT cell viability and IL-1α release assays, are used. Acknowledgements The authors would like to thank Dr. Hajime Kojima, Director of the Japanese Center for the Validation of Alternative Methods (JaCVAM), Division of Pharmacology, Biological Safety Research Center, National Institute of Health Science (NIHS), for his advices. References Cotovio J, Grandidier MH, Portes P, Roguet R, Rubinstenn G. (2005) The in vitro skin irritation of chemicals: optimisation of the EPISKIN prediction model within the framework of the ECVAM validation process. Altern Lab Anim, 33(4), Fentem JH, Briggs D, Chesné C, Elliott GR, Harbell JW, Heylings JR, Portes P, Roguet R, van de Sandt JJM, Botham PA. (2001) A prevalidation study on in vitro tests for acute skin irritation: results and evaluation by the Management Team. Toxic. in Vitro, 15(1), Kandarova H, Liebsch M, Genschow E, Gerner I, Traue D, Slawik B, Spielmann H. (2004) Optimisation of the EpiDerm test protocol for the upcoming ECVAM validation study on in vitro skin irritation tests. ALTEX, 21(3), Kandarova H, Liebsch M, Gerner I, Schmidt E, Genschow E, Traue D, Spielmann H. (2005) The EpiDerm test protocol for the upcoming ECVAM validation study on in vitro skin irritation tests--an assessment of the performance of the optimised test. Altern Lab Anim, 33(4), Kandarova H, Liebsch M, Schmidt E, Genschow E, Traue D, Spielmann H, Meyer K, Steinhoff C, Tornier C, De Wever B, Rosdy M. (2006) Assessment of the skin irritation potential of chemicals by using the SkinEthic reconstructed human epidermal model and the common skin irritation protocol evaluated in the ECVAM skin irritation validation study. Altern Lab Anim, 34(4), Morikawa N, Morota K, Morita S, Kojima H, Nakata S, Konishi H. (2002) Prediction of human skin irritancy using a cultured human skin model: comparison of chemical application procedures and development of a novel chemical application procedure using the Vitrolife-Skin model. AATEX, 9(1), Morikawa N, Morota K, Suzuki M, Kojima H, Nakata S, Konishi H. (2005) Experimental study on a novel chemical application procedure for in vitro skin corrosivity testing using the Vitrolife-Skin human skin model. AATEX, 11(1), Morota K, Morikawa N, Morita S, Kojima H, Konishi H. (1998) Development and evaluation of the cultured skin model. Tiss Cult Res Commun, 17(2), Morota K, Morikawa N, Morita S, Kojima H, Konishi H. (1999) Alternative to primary Draize skin irritation test using cultured human skin model: comparison of six end points. AATEX, 6(1), Portes P, Grandidier MH, Cohen C, Roguet R. (2002) Refinement of the Episkin protocol for the assessment of acute skin irritation of chemicals: follow-up to the ECVAM prevalidation study. Toxicol In Vitro, 16(6), Zuang V, Balls M, Botham PA, Coquette A, Corsini E, Curren RD, Elliott GR, Fentem JH, Heylings JR, Liebsch M, Medina J, Roguet R, van de Sandt JJ, Wiemann C, Worth AP. (2002) Follow-up to the ECVAM prevalidation study on in vitro tests for acute skin irritation. The European Centre for the Validation of Alternative Methods Skin Irritation Task Force report 2. Altern Lab Anim, 30(1), (Received: September 20, 2007/ Accepted: December 27, 2007) Corresponding author: Noriyuki Morikawa Research & Development Center, GUNZE Ltd., 1 Ishiburo, Inokura-shinmachi, Ayabe, Kyoto , Japan Tel: Fax: noriyuki.morikawa@gunze.co.jp 26