Skin irritation/corrosion



The model predicts the reversible (irritation) and irreversible (corrosion) damage of the skin following the application of a test substance for up to 4 hours.




The training set of the model consists of 856 organic chemicals having according to EU dangerous substance directive (DSD) the risk phrases 34, 35 or 38; global harmonized system (GHS) health hazard H314, H315 or H316; or having primary skin Irritation Indices larger than 1.5. Part of experimental data was provided by the National Institute for Public Health and the Environment (RIVM) Netherlands [1].




The Skin irritation / corrosion  model consists of two major components[2 - 5]:


  • Exclusion rules for skin irritation/corrosion (provided by RIVM).
  • Inclusion rules for Skin irritation/corrosion 


Exclusion rules for skin irritation/corrosion are based on physicochemical cut-off values of the following descriptors: molecular weight; melting point; log KOW; water solubility and vapor pressure.  These rules are coded in a profiling scheme consisting of 28 categories.  Inclusion rules form a profiler consisting of empirically derived structural boundaries.




The domain was determined by splitting training chemicals into correctly and incorrectly predicted chemicals [6].  The applicability domain consists of three layers:


  • General parametric requirements - includes ranges of variation log KOW and MW,
  • Structural domain - based on atom-centered fragments (ACFs).
  • Reliability of the inclusion rules - ratio between the number of correctly classified chemicals and the total number of chemicals


A chemical is considered In Domain if its log KOW and MW are within the specified ranges and if its ACFs are presented in the training chemicals. The information implemented in the applicability domain is extracted from the correctly predicted training chemicals used to build the model and in this respect, the applicability domain determines practically the interpolation space of the model.




The model performance is evaluated by the percent of correctly predicted irritants (sensitivity) - 70 %. The training set of chemicals is not well balanced: from 856 chemicals 854 are positive and only 2 are experimentally observed negative irritants.  In this respect, specificity of the model is not defined, given insufficiency of negatives in training set.  In progress is work aiming to expand the training set and to improve the balance between active and non-active training chemicals.




The predictions of skin irritation/corrosion model could be reported in tab delimited file including the following information for the chemicals:


  • Chemical identity (CAS number, Name, SMILES),
  • Observed  and predicted data for  skin irritation and corrosion,
  • Applicability domain details.





1. Kodithala K, Hopfinger AJ, Thompson ED, Robinson MK. Prediction of skin irritation from organic chemicals using membrane-interaction QSAR analysis.Toxicol Sci. 2002, 66, 336-346.

2. E. Hulzebos,  J.D. Walker, I. Gerner, K. Schlegel, Use of structural alerts to develop rules for identifying chemical substances with skin irritation or skin corrosion potential. QSAR Comb. Sci., 2005, 24(3), 332-342.

3. I. Gerner, M.D. Barratt, S. Zinke, K. Schlegel, E. Schlede, Development and prevalidation of a list of structure-activity relationship rules to be used in expert systems for prediction of the skin-sensitizing properties of chemicals. ATLA, 2004, 32(5), 487-509.

4. A.G. Saliner, I. Tsakovska, M. Pavan, G. Patlewicz, A.P. Worth, Evaluation of SARs for the prediction of skin irritation/corrosion potential: structural inclusion rules in the BfR decision support system.SAR QSAR Environ. Res.,2007, 18(3-4), 331-342.

5. D.A. Basketter, D. Kan-King-Yu, P. Dierkes, I.R. Jowsey, Does irritation potency contribute to the skin sensitization potency of contact allergens. Cutan. Ocul. Toxicol., 2007, 26(4), 279-286.

6. S. Dimitrov, G. Dimitrova, T. Pavlov, N. Dimitrova, G. Patlevisz, J. Niemela and O. Mekenyan, J. Chem. Inf. Model. 45, 2005, 839-849.

Skin irritation/corrosion

Model Features