Applying (Q)SAR tools in the prioritization of potentially genotoxic substances migrating from printed paper and board food contact materials

Recent food contamination issues due to leakage of food packaging substances into food have raised concerns towards potential adverse health effects caused by exposure to these chemicals (e.g. bisphenol A). In 2011, the European Commission adopted a new regulation listing the chemicals that can be used to manufacture plastics intended to come into contact with food. However, to this day, no such EU regulation exists on chemicals used in other types of food contact materials (FCM). Consequently, non-plastic FCM are only covered by national legislation or by the general recommendations of the Council of Europe, containing ‘inventory lists’ of additives, monomers, solvents and other compounds that might be present in non-plastic FCM. Since no (recent) safety evaluation has been carried out for thousands of substances present in these inventory lists, identification of substances requiring further safety data is urgently needed. The current work attempts to identify substances of highest concern used in printed paper and board FCM. First, information on substances present in the inventory lists of the Council of Europe or from national legislation was assembled in a database. Among the 6068 unique compounds, 4694 or 77% have not been officially evaluated. Secondly, the non-evaluated single substances were selected for further prioritisation. Over 75% of them have physicochemical characteristics in favour of both migration to food and toxicity. Next, their genotoxic potential was analysed in silico using multiple (Q)SAR models (ToxTree, VEGA Consensus, Derek NexusTM, Sarah NexusTM), the initial focus being the mutagenicity endpoint. Whenever possible, an applicability domain assessment was performed to ensure fitting of the test compounds in the models applied. The analysis yielded a number of 106 substances positive in all 4 tools (substances of highest concern), whereas 1107 substances were negative in all 4 tools (lowest concern). Another 515 substances were either positive in at least one tool, or could not be analysed in one or several tools (medium concern). These results indicate that several hundreds of non-evaluated single substances used in printed paper and board are predicted genotoxic by multiple in silico tools. Moreover, inspection of their physicochemical properties shows considerable migration and toxicity potential. The substances containing a structural alert for genotoxicity were selected for further more in depth study, combining extensive literature review with in vitro genotoxicity testing.