The application of NTS and SS methods for article, chemical product and recycled material screening is a relatively new but presumably fast-growing field as suggested by the majority of studies reviewed here being published in the last couple of years. Most of these studies have focused on screening different types of products aiming to identify hazardous (or not) chemicals in these products, often by developing and applying new improved workflows. Some studies have also looked at which chemicals can migrate from the products under exposure-relevant scenarios as well as their transformation products. Besides identifying chemical substances with a well characterized hazard profile that may also be regulated, prioritized or under assessment, there are few studies that have also investigated potential hazard properties of the identified chemicals using in silico approaches or tested their cocktail effect under experimental conditions.
To use NTS and SS data for identifying compounds of regulatory concern, the data needs to be searched through regulatory and hazard information databases or applied to in silico prediction tools. This type of processing is highly dependent on chemical identifiers such as CAS-numbers, SMIILES, InChI keys, chemical names, and molecular formulas. Thus, ambiguities in these may lead to inconsistent or incorrect conclusions about hazardous properties and chemical risks. Challenges observed in this study regarding data from NTS and SS studies included the lack or inconsistent usage of chemical identifiers. The challenge becomes more complex as regulatory and hazard information databases often utilize a single CAS-number as a chemical identifier for a compound which may have many CAS-numbers assigned. A sufficient data curation process is therefore required to assign unique and comprehensive chemicals identifiers in order to minimize the risk of failing to assign hazardous properties to chemicals identified with NTS and SS methods.
Our evaluation of the data has shown that a focus on the most frequently detected compounds in a specific product or material category might result in a rather low number of compounds, and that these compounds may be influenced by different types of bias. This is supported by the interlaboratory comparisons presented in the discussion. Future work, building on the gathered data from this study could include in silico hazard assessment of compounds which were detected frequently with sufficient reliability but were not found in the regulatory and hazard information databases, to further investigate the extent of which NTS and SS techniques are capable of detecting compounds of emerging concern and potential regulatory relevance.
Non-target and suspect screening approaches have the potential for being suitable and effective for identifying regulatory-relevant substances in products and materials, but they are probably better suited for early warning screening purposes at this stage given their current limitations especially with the identification uncertainty/confidence. Positive identifications may require large efforts for verification, especially since not all chemicals on the market are available for purchase as reference standards that could facilitate a confirmation of the identity. As long as a compound has not been positively identified using a reference standard the risk that it may constitute a false positive identification will exist and therefore also requires preparedness from regulators that this may be the case. False negative identification of hazardous compounds presents an additional challenge as there is a hazardous substance that remains undetected. However, multiple developments could reduce such risks. The introduction of more standardized workflows, with well-defined QA/QC criteria, are needed to increase the overall screening method robustness and the comparability between results from different labs. This may involve introduction of QC-samples to verify the performance of the method within the desired chemical space or providing information on, for instance, false negative results.