Study of the possible migration risks of food contact materials for children under 3 years

Bisphenol-A (BPA) has been used for many years as a monomer for polycarbonate (PC) polymers (water and infant feeding bottles) and epoxy resins (canned food packaging) from which it can be released into the food, the major exposure source of BPA to humans. Since BPA has endocrine disrupting properties, its use was prohibited for the production of polymers for food contact materials for children younger than 3 years old (European Commission, regulation No. 10/2011). Furthermore, in a recent opinion, the Superior Health Council of Belgium expressed its concern regarding the possible risks associated with the used alternatives to PC (No. 8697, 11.03.2010). Consequently, alternatives to PC food contact materials (FCMs) for infants, such as polypropylene (PP), polyethersulphone (PES), polyamide (PA), Tritan™ or silicone baby bottles, have appeared on the market. Migration of BPA from PC has already been extensively studied. Unfortunately, the nature and amounts of substances migrating from the polymeric alternatives other than PC is much less known. The principal aim of this PhD was the identification and quantification of the major and most toxic compounds migrating from baby bottles, in the frame of a Belgian governmental project (ALTPOLYCARB) involving several Belgian universities. The first experimental chapter of this thesis (Chapter 3) describes the possible alternatives to PC FCMs for children under 3 y old on the Belgian market. These articles were documented by an initial market survey in baby-shops, supermarkets and pharmacies. 24 baby bottle types from different manufacturers were encountered here. The polymers used in the manufacture were, in order of importance, PP, PES, PA, Tritan™ and silicone. Some PC baby bottles have also still been encountered. Baby cups, teats, dinnerware and other infant FCMs were studied as well. However, for the latter items, a major percentage of the polymers used for their production could generally not be identified. Given the lack of information for the other FCMs categories than baby bottles, and considering the major importance of the latter for infant feeding, the migration tests were started on a selection of representative baby bottles. For migration testing, the use of simulants is prescribed in the legislation to mimic the testing of real foods. Specifically, a mixture of water-EtOH (50:50, v/v) is recommended as a simulant for milk. After sterilisation of the bottle during ten minutes with boiling water, three migrations were performed during 2h at 70 °C. Firstly, a liquid-liquid extraction (LLE) of the simulant with a mixture of common organic compounds (Chapter 4.1) was optimised. To develop a robust and general method, a mixture of 17 chemicals (identified in the literature as possible migrants from FCMs) covering a wide variety in polarity and chemical functionality was chosen to evaluate the extraction efficiency of n-hexane, iso-octane, ethyl acetate (EtOAc)-n-hexane (1:1 and 1:3), MTBE and dichloromethane (DCM)-n-hexane (1:1 and 1:3). The extracts resulting from the LLE step were analysed on GC-(EI)MS by monitoring specific ions for each analyte and for the internal standard. EtOAc-n-hexane (1:1) and DCM-n-hexane (1:1) were the most efficient extraction solvents. Consequently, there was opted for the nonchlorinated solvents and EtOAc-n-hexane (1:1) was selected for the application to real samples. We have assessed the possible release of unknown chemicals from PP, PES, PA, Tritan™ and silicone baby bottles. The migration solutions from the baby bottles were extracted and analysed on GC-(EI)MS performing an untargeted database search using Wiley® and NIST® libraries. Although the concentrations observed were rather low, various compounds, such as alkanes, phthalates, amides, etc. were detected based on this library search. In Chapter 4.2, unidentified peaks were further investigated by advanced mass spectrometric techniques, such as GC-(EI)TOF-MS and GC-(APCI)QTOF-MS, to specifically elucidate the structure of these unknown compounds. The expected presence of the accurate mass molecular ion and/or protonated molecule in APCI together with the fragmentation pattern observed in both techniques were used for elucidation purposes. By developing an identification strategy based on the combination of these analytical techniques, compounds (e.g. dicyclopentyl-(dimethoxy)silane, Irganox 1010, etc.) that could not be identified before were elucidated here. Additionally, the same extracts were analysed also by LC-QTOF-MS under MSE mode. The full-spectrum accurate mass data of both (de)protonated molecule and fragment ions were acquired simultaneously. Data were automatically processed using a homemade database containing around 1200 chemicals present on the list provided by the EU Regulation No. 10/2011 and expected migrating compounds, such as anti-oxidants, plasticisers, etc. When a peak was detected, and its reference standard was not available in the lab, a tentative identification was performed using the accurate mass of the observed fragment ions. Several compounds which were previously not identified by GC-MS analysis were elucidated. In parallel with the identification process, the genotoxicity and endocrine activity of the identified migrants were evaluated by the partner universities participating in the ALTPOLYCARB project using a battery of in vitro assays. Chapter 5.1 describes the optimisation of a LLE method for a number of migrating compounds that were selected based on the outcomes of this toxicity screening/scoring and the migrating abundances observed in Chapter 4.1. Monitoring and quantification of these compounds was done using GC- and LC-QqQ-MS methods, for which several validation parameters were determined (sensitivity, selectivity, linearity, accuracy, precision, recoveries and matrix effects). Analysis of the 3rd migration step of the standard migration conditions (3 migrations, 2 h at 70°C) applied on the baby bottles (which has to comply with the EU legislative migration limits) showed that for some baby bottles, several not authorised compounds exceeded the generally adapted “no-detection limit” of 10 µg kg-1 . Substances, such as 2,4-di-tertbutylphenol (up to 118 µg kg-1 ), 2-butoxyethyl acetate (up to 945 µg kg-1 ) and 4- propylbenzaldehyde (up to 27 µg kg-1 ) were detected in several bottles, as well as some phthalates. The silicone bottle even exhibited concentrations of 2,2,4-trimethyl-1,3- pentanediol diisobutyrate (TXIB) around 350 µg kg-1 . For all detected compounds authorised by the EU Regulation No. 10/2011 with a specific migration limit (SML), such as benzophenone (600 µg kg-1 , found up to 97 µg kg-1 ), concentrations in the migration solutions were below the SMLs. In Chapter 5.2, an evaluation of the effect of several “real-life use conditions” by means of duration tests such as microwave, sterilisation and dishwasher treatment on the profile of the different migrants was determined and compared with a reference treatment (30 min at 40°C) and the standard EU “repetitive use conditions”. Analysis of the extracts from the microwave experiments showed a modest increase in the concentrations of the observed migrants (e.g. azacyclotridecan-2-one from the PA bottle: 124 μg kg-1 after the first microwave heating vs. 70 μg kg-1 in the first referenceexperiment). Moreover, a prolonged release of the target compounds was also observed whereas these migrants disappeared significantly faster in the reference experiment. The dishwasher treatment resulted also in a slight increase of some of the target compounds, whilst others exhibited lower concentrations than the reference experiment. This was most probably due to the fact that they were already partially washed away during dishwashing. Steam sterilisation showed a quick removal of the monitored compounds and the detected concentrations were lower (e.g. TXIB from silicone bottle 28 vs. 118 μg kg-1 ) or similar (di(iso)butyl phthalate) to those seen after the reference treatment. The tendency observed for the steam sterilisation indicated a clear advantage of performing this treatment in order to eliminate residual chemicals that might still be present in the polymer. For the cooking sterilisation, generally the same observations as for the steam sterilisation were done, since also here a preliminary removal of the chemicals was seen. However, an increased release of some target compounds was seen for the silicone bottle, suggesting that this treatment was not suitable for this material. For all duration tests, a downwards tendency of the measured concentrations was observed through the subsequent cycles. The target compounds observed after the different duration tests were in accordance with those seen before in the EU repetitive use experiment; however here the observed concentrations were significantly higher for most compounds. Although the repetitive use experiment therefore seemed to overestimate the actual migration from baby bottles under real-life use conditions, it has to be mentioned that all migration tests were performed individually, and that the combination thereof might result in a higher release. Chapter 5.3 presents the results of a fingerprinting study that was made on the initial samples and on those exposed to a high number of duration tests. To this end, not only the influence of real-life treatments on migration of the target compounds, but also on the rest of possible migrants (e.g. degradation products) was investigated. It was shown that under these real-life use treatments no degradation of the polymers took place, although the steam sterilisation resulted after 10 cycles in an increased release of an antioxidant (BHT) from the PP bottles compared to the first cycle. In Chapter 6, a critical discussion was done on the outcomes of the presented research. Special focus was made on the detection of multiple non-intentionally added substances in the baby bottles. Moreover, suggestions were made on how the presence of these compounds could be safeguarded in future studies.