IN SILICO TOXICITY PREDICTION OF COSMETICS’ COLORANTS USING ADMET PREDICTOR, OECD QSAR TOOLBOX, TOXICITY ESTIMATING SOFTWARE TOOL AND TOXTREE

• Main Author: Dwi Aryati, Widya
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/44513
• Institution: Institut Teknologi Bandung

Cosmetics on the market have to be guaranteed as safe to human. One of the safety assessment of cosmetics is toxicity profile of its ingredients such as colorants, preservatives and UV-filters. Data of toxicity can be obtained from in vitro and in vivo studies. However, the use of test animals in cosmetics’toxicity has been limited. Therefore, the use of methods to predict toxicity of a compound is increasing to support in vitro and in vivo studies. In silico toxicology is one of alternative methods that applies computer technology to analyze existing data and models to predict toxicological properties of a compound. The aim of this study is to predict toxicity of colorants in cosmetics. In this study, mutagenicity, carcinogenicity, and skin sensitization prediction of colorants in cosmetics were done using Toxicity Estimating Software Tool (T.E.S.T.) v4.2, ToxTree v2.6.13, ADMET Predictor™v.7.1.0013, and OECD QSAR Toolbox v3.4. Verification of prediction methods was carried out using positive and negative control from CLP Regulation using classification model. Cooper statistical parameters were used for verification of prediction methods. The value of Cooper statistical parameters obtained for sensitivity, specificity, accuracy, positive and negative predictivity is ranged between 0.6 and 1.0 for mutagenicity prediction methods, between 0.7 and 0.8 for carcinogenicity prediction methods, and between 0.4 and 0.7 for skin sensitization prediction methods. The value of false positive and negative rate is ranged between 0-0.3 for mutagenicity prediction methods, between 0.1 and 0.2 for carcinogenicity prediction methods, and between 0.2 and 0.5 for skin sensitization prediction methods. The value of positive and negative ROC is ranged between 2- ?for mutagenicity prediction methods, between 2-6 for carcinogenicity prediction methods and between 0.96 and 2.55 for skin sensitization prediction methods. From this study, 38 colorants were predicted as nontoxic and 96 colorants as toxic which consist of 21 colorants as carcinogen, mutagen and skin sensitizer, 14 colorants as carcinogen and skin sensitizer, 8 colorants as carcinogen and mutagen with structural alerts anthrone, xanthone, thioxanthone or acridone, aromatic diazo, aromatic amine, aromatic nitro, hydrazine, polycyclic aromatic hidrocarbons, and quinone, 29 colorants as carcinogen with structural alert metal, benzenesulfonic ethers, imidazole or benzeimidazole, and halogenated benzene, and 24 colorants as skin sensitizer with structural alert amide, ketone, aromatic carbonyl, 1,2- dicarbonyls, 1,3-dicarbonyls, pyrazolones and pyrazolidinones and quinone methide(s)/imines, quinoide oxime structure, nitroquinones, naphthoquinone(s)/imines. This study showed that in silico method can be used to predict carcinogenicity, mutagenicity and skin sensitization of colorants in cosmetics and can be used as supporting method for safety evaluation of colorants in cosmetics.