Identification and risk assessment of unknown contaminants migrating from Food Contact Materials

Abstract

The exposure of humans to possibly thousands of chemical compounds through food poses a health risk that is questioning our ability to ensure high standards for food safety. Food contact materials (FCM) are a major source of extraneous chemical compounds in food, yet not much knowledge is available on all compounds present due to FCM. The practicability of comprehensive studies, like risk assessment (RA), is questionable for an increasingly large number of chemical compounds. As a consequence, most research is focused on a small number of well-studied chemical compounds, but little is dedicated to the much larger number of unknown compounds. How are FCM safe for use if the greater part of it is unidentified, unassessed, and possibly completely unknown? Here, development is shown for two new analytical strategies: semi-quantification and tentative identification; along with possible application for FCM RA. Paper and board FCM were extracted for migratable content, followed by analysis by liquid chromatography (LC) high resolution quadrupole-time of flight (Q-TOF) mass spectrometry (MS). Compounds were semi-quantified by comparing to non-identical reference standards after dedicated system optimisation. For identification, Q-TOF MS/MS utilizing automated precursor selection was used to actively collect non-target fragmentation spectra of compounds in the chromatogram. A risk prioritization approach that classified chemical compounds according to expected risk was developed based on applied tentative data and subsequent data interpretation by expert assessors. Semi-quantification was demonstrated to work for almost any compound detected by LCQTOF-MS analysis (Manuscript A). The errors in the predicted concentrations were at maximum up to 3-fold error with average around up to 2-fold error. These errors were attainable after dedicated optimisation of the LC-MS system to produce uniform responses that favour improving response of weak-response compounds. Semi-quantification did not require chemical identification or standard matching. For a single sample, more than 300 compounds were simultaneously quantified. Consequently, semi-quantification is a valuable strategy for prioritization based on concentration and for acquiring quantitative data without prior identification or available reference standards. The tentative identification of compounds was demonstrated by non-targeted structural data acquisition (Manuscript B). Fragmentation spectra collected by Q-TOF MS/MS were correlated with in silico generated spectra using chemical structure databases to find the best-matching chemical compound to the spectrum, thereby removing the need for a reference standard. A total of five structure databases were used, resulting in structure prediction for over 130 compounds discovered in a recycled paper and board pizza box. For most of the 130 compounds, structure predictions were successful with good correlation scores, resulting in an impression of the chemical structure. The tentative identity of some compounds was evaluated for possible risk based on concentration and existing hazard data. Tentative identification is a promising strategy used to obtain significant chemical information about compounds in complex samples. Tentative data was used to prioritize risk of identified compounds, differentiating between high-risk and low-risk compounds based on predicted exposure and predicted hazards (Manuscript C). This approach mimics RA procedures by converting tentative data to hazard- and exposure estimates, followed by a combined assessment based on expertise judgement. The expertise of several trained risk assessors was used to assign risk profiles to compounds in order to achieve a risk ranking. Although tentative data contains uncertainty, interpretation by experts produces a viable risk ranking of known and unknown chemical compounds by implementing a consensus model of expert interpretations. Risk prioritization is successful in classifying estimated risk based on predicted exposure and predicted hazard, and is valuable for to preliminary RA studies. The overarching strategy in this study shows that explorative techniques are valuable tools to help ensure food safety in the future. Tentative data and risk prioritization are key concepts that, when developed further in combination with predictive tools like structureactivity modeling or migration modeling, could be at the forefront of identifying present and future risks. The need for these strategies is clear: tentative and explorative data is needed because the current alternative is often no data.