S166 Abstracts / Toxicology Letters 238S (2015) S56–S383 parameters. As well as the prediction of renal clearance, the model can contribute to toxicocokinetic profile predictions when incorpo- rated into a whole body physiologically based toxicokinetic model. Such an in silico model can contribute not only to the prediction of systemic exposure-driven TD but has, with further elaboration, the potential to support prediction of nephrotoxicity through mod- elling of local, intra-renal xenobiotic concentrations. http://dx.doi.org/10.1016/j.toxlet.2015.08.565 P06-005 Optimization of curation of the dataset with data on repeated dose toxicity U. Gundert-Remy 1,∗ , M. Batke 2 , A. Bitsch 2 , M. Gütlein 3 , S. Kramer 3 , F. Partosch 1 , M. Seeland 3 1 Charité, Berlin, Germany 2 ITEM Fraunhofer, Hannover, Germany 3 Johannes Gutenberg-Universität Mainz, Mainz, Germany Introduction: For some areas of risk assessment, the use of alter- native methods is supported by current directives and guidance (e.g. REACH, Cosmetics, BPD, PPP). According to OECD principles alternative methods need to be scientifically valid. Methods: Within a project on grouping and development of predictive models sup- ported by a grant of Federal Ministry of Education and Research, we curated a dataset based on RepDose and ELINCS database. The final dataset consists of rat repeated dose toxicity studies for 1022 com- pounds representing 28 endpoints as organ-effect-combinations. Toxicological and modelling experts did jointly the curation and selection of endpoints as an iterative process. Results: Missing val- ues for endpoints of the dataset were the main problem to be handled. Endpoints such as thyroid gland contain specific informa- tion in contrast to unspecific endpoints such as liver/body weight. Unfortunately, for specific endpoints data is often missing (>90%) in the dataset. Several attempts were made to fill the data gaps. Finally, a statistic imputation procedure gave best results for group- ing and modelling. We decided to include endpoints in the dataset only if the number of data points were sufficient to make precise predictions. The toxicological profile of a substance is determined not only by the affected endpoints but also by the potency. Hence, we decided to include the information on the potency as measured by LOELs. We explored several statistical procedures. Best results were obtained by an equal frequency distribution of LOEL values per endpoint. Smiles codes for substructures and reactive groups were selected as structural representation of the substances. The final evaluation showed that this structural description has major impact on the outcome of the grouping, e.g. ethylene glycols and aliphatic alcohols were not separated due to missing specific Smiles codes. By grouping and modelling the dataset it was experienced that specific physico-chemical parameters should be included in order to gain toxicological meaningful results. Discussion: Overall, within this project, we were able to show the impact of a carefully cured dataset on the usefulness and quality of grouping and pre- dictive models build upon this dataset. For further information see http://mlc-reach.informatik.uni-mainz.de http://dx.doi.org/10.1016/j.toxlet.2015.08.566 P06-006 Developing read-across for toxicity prediction within an AOP framework to support the transition from the analogue approach to local QSARs F.P. Steinmetz ∗ , D.J. Ebbrell, S.J. Enoch, J.C. Madden, M.D. Nelms, M.T.D. Cronin Liverpool John Moores University, Liverpool, United Kingdom Read-across for toxicity prediction is developing from a simplis- tic approach where one analogue is chemically similar to another, to supporting more complex groupings of molecules based on both biological and chemical similarity. The latter approach provides the opportunity to develop local, mechanistically based quantita- tive structure-activity relationships (QSARs) to make predictions even for the most difficult endpoints e.g. chronic and reproductive toxicity. The aim of this study was to illustrate how “local” QSARs can be developed within robust groupings. The QSARs were devel- oped from knowledge of the molecular initiating event (MIE) from putative Adverse Outcome Pathways (AOPs). The concept has been illustrated with reference to cytotoxicity data to Aliivibrio fischeri grouped for protein binding (representing reactive toxicity) and mitochondrial toxicity mechanisms (including respiratory uncou- pling) using structural alerts. Sufficient cytotoxicity data were retrieved (n = 36 and 45 respectively) to develop QSARs. For the pro- tein binding category, a hydrophobicity and reactivity-based QSAR, employing hydrophobicity (log P) and molecular orbital energies (HOMO–LUMO gap), was applied. For mitochondrial toxicity, a permeability-based QSAR, using molecular size (molecular weight) and log P, was utilised. Both approaches demonstrated that highly significant QSARs could be developed where data and mechanistic understanding are available. Future developments will be based on QSARs building on the grouping approach, which could also support the development of (semi-)quantitative AOPs. As docu- mentation and justification of predictions is the key to scientific and regulatory acceptability, this approach provides the possibility of greater acceptance for in silico methods. The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007–2013) COSMOS Project under Grant Agreement N ◦ 266835 and Cosmetics Europe. http://dx.doi.org/10.1016/j.toxlet.2015.08.567 P06-007 Advanced QSAR models for use in toxicokinetic modelling D. Sarigiannis ∗ , K. Papadaki, P. Kontoroupis, S. Karakitsios Aristotle University of Thessaloniki, Chemical Engineering, Thessaloniki, Greece A current limitation to further introducing PBBK models in the risk assessment arena is the lack of generic character of these mod- els. A critical limiting factor of describing the ADME process for a large chemical space is the proper parameterization for “data poor” compounds. In order to expand the applicability of PBBK models to cover as much as possible the chemical space, model parameteri- zation for data poor chemicals is done using advanced quantitative structure–activity relationships (QSARs). Several QSAR modeling approaches have been investigated, including (a) an algorithm based on fractional content of and the lipophilicity of the com- pound of interest; (b) the molecular fractions algorithm that takes