Single-use systems (SUSs) are widely used in pharmaceutical manufacturing, packaging and storage [1,2]. These systems consist of tubing, filters, sensors, connectors, bags, and other components. However, the use of various chemical additives to enhance the properties of plastics and elastomers in SUSs can lead to the migration of impurities into the final product [3,4]. This is particularly concerning for therapeutic proteins, as they are susceptible to chemical modifications that can cause structural damage and aggregation [5], [6], [7]. To ensure product quality and patient safety, it is essential to identify and control these extractables and leachables (E&L) during the drug development stage.

Given the structural complexity of these contaminants, a combination of analytical methods is necessary to obtain a comprehensive E&L profile [8,9]. Both targeted and non-target approaches are utilized for E&L investigation. Targeted analysis methods are developed to detect specific compounds that exhibit extreme genotoxicity, hepatotoxicity, teratogenicity or cytotoxicity [3,10]. Examples of such compounds include polyaromatic hydrocarbons (PAHs) [10,11], N-nitrosamines [12], [13], [14], 2-mercaptobenzothiazole (2-MBT) [11] and antioxidants [15,16]. Non-targeted strategies, on the other hand, are employed to identify unexpected or comprehensive compounds and typically involve the analysis of volatile (VOC), semivolatile (SVOC), nonvolatile (NVOC) organic compounds, elemental entities (ELE), and inorganic anions and low-molecular-weight organic acids [8,10,17].

Structural elucidation plays a crucial role in E&L studies as it impacts subsequent toxicological assessment, which is essential for selecting manufacturing/packaging components and ensuring the safety of a drug. Headspace gas chromatography-mass spectrometry (GC–MS) and direct injection GC–MS are applied for the characterization of VOCs and SVOCs, respectively. While NVOCs are analyzed using reversed-phase liquid chromatography-mass spectrometry (RPLC-MS). The identification of compounds in GC–MS spectra typically relies on spectral matching with the NIST library, with a library match factor of no less than 80 [18,19]. However, due to the greater variation in mass spectra generated by LC-MS to GC–MS, only online LC-MS databases (such as mzCloud and MassBank) are used for qualitative analysis [19,20]. Although commercial E&L-specific databases, such as Agilent E&L Personal Compound Database and Library (PCDL) and Thermo Compound Discoverer E&L HRAM Compound Database, are available, they have a limited number of compounds (approximately 1000) and lack MS/MS spectra [18,21]. To accurately interpret mass spectra, the development of an internal database [22], [23], [24], along with expert interpretation [19,25], is highly needed. Furthermore, it is essential to report the confidence level of an identified analyte to inform other investigators about the reliability of the result. Identification categories can be classified into four levels: confirmed, confident, tentative and unidentified [10,26,27].

Numerous reports have been published on the extractables study of NVOCs, which are utilized to simulate the leachables study in worst-case scenario [3,[28], [29], [30], [31], [32], [33], [34]]. Although a few leachable studies utilized actual drug products such as vaccine [22], antibody/ protein [35], [36], [37] and peptides [38,39], the scope of leachables screening has been limited to compounds that were identified in extractable experiments of specific material systems or focused on specific impurities detected in drugs. However, a comprehensive leachables study of non-target NVOCs in antibody drugs has yet to be reported.

On the other hand, therapeutic protein formulations typically contain various excipients such as buffer salts, antioxidants, chelating agents and surfactants [40]. Polysorbate 20 and 80 (PS 20 and 80) are crucial excipients that protect the antibody protein from denaturation, aggregation, and surface adsorption [41]. However, the high ionizability of PS may interfere with the accurate quantitation of leachables. Therefore, it is recommended to prepare drug samples by precipitating the protein and removing PS prior to LC-MS analysis. Despite a previous report demonstrating that PS 80 can be reduced through complexation with a metal ion followed by precipitation with thiocyanate [42], currently, this method has not yet been applied to the study of leachables in biopharmaceuticals.

The objective of this study was to detect and identify all NVOCs observed in two antibody drugs using UPLC-QTOF MS. The structures of leachables were identified by searching against the self-built E&L database (SELD) and web-based libraries. Additionally, we proposed a procedure for tentatively identifying impurities that are not present in any of the databases. Lastly, a quantitative method based on complexation-precipitation assisted extraction for targeted analysis was developed. To the best of our knowledge, this is the first report on the non-target screening, identification, and targeted quantification of leached NVOCs from antibody drugs.