When, how and to what extent LC-MS should be used for Host Cell Protein (HCP) analysis is a hot topic. The debate over the degree to which LC-MS-HCP analysis can be relied upon for HCP detection and quantitation is an ongoing one and for the moment, will be largely defined by the specific product and the process employed to generate an Active Pharmaceutical Ingredient. At BioPharmaSpec, we have developed an understanding of the tools that are available to assist in generating a greater depth of LC-MS analysis and in doing so, reduce HCP oversight.
With any method there are benefits and limitations. In this blog we have highlighted some key aspects of LC-MS HCP detection to help in the development of a robust HCP detection strategy. Armed with the right knowledge, we hope to guide our clients through the complex process of assessing the extent to which LC-MS analysis might be used as a comparative tool and to progress their drug products through this challenging regulatory landscape.
To date, the industry ‘gold-standard’ has been immunoassay, typically enzyme-linked-immunosorbent assays (ELISAs), which provide high sensitivity and throughput. ELISAs provide straightforward testing but generic or platform kits won’t detect proteins particular to the manufacturing process, unless the assay has been developed to be process-specific, which can be costly and take time to develop. There is also the risk of missing any low-reactive/immunogenic HCPs.
As LC-MS specialists, our focus at BioPharmaSpec is to deliver an increased breadth of HCP identification with high reproducibility and specificity. HCP-MS can detect a wider population of individual HCPs than ELISA within a single run, reaching down to sub 10ppm or single digit ppm levels depending on the quantitation methods employed. HCP-MS is also highly flexible, so if there are any changes to upstream processing steps which can risk changes to the HCP profile, then MS is a time-efficient way (in comparison to process-specific ELISAs) to monitor the population and clearance.
The limitations include the fact LC-MS requires skilled operators with experience to perform the analysis and interpret the data. Also, in label-free quantitation methods, the HCP detection is dependent on the ionization capability of the protein fragments. Use of relative quantitation with known spike standards is also not as accurate due to ionization differences between the spike and HCP peptides.
In HCP analysis, the principal aim is to identify and quantitate down to very low abundance proteins, as we are trying to eliminate as many and as much as possible from the final drug product. Targeted mass spectrometric techniques have been developed with this specific focus in mind, using both labeling and label-free methods like multiple-reaction-monitoring (MRM) or Sequential Window Acquisition of all Theoretical mass spectra (SWATH-MS). These methods have produced the best results for these types of analyses, and it’s the reason why they are frequently adopted for HCP-MS study. MS does have an incredible dynamic range - up to 5 times sensitivity - which means it can detect low levels of HCPs amongst the high concentration of drug product, but there can still be suppression of peptides of interest or ionization inefficiencies or perhaps incomplete digestion or other interference.
For these reasons, we favor a two-pronged approach in which ELISA is run alongside LC-MS and/or other orthogonal methods like SDS-PAGE/Western blotting, in order to bring depth and more thorough characterization to the product’s HCP profiles.
Bioinformatics tools are fundamental for the peak-peaking and filtering of large HCP-MS datasets, usually comprising many replicates, sometimes with thousands of HCP identifications. Instrument-specific processing software is tailored for a specific set-up and raw file type. We use both SCIEX and Waters systems and their respective software programs, Progenesis QI or SWATH Acquisition MicroApp to perform these analyses and calculate the peptide areas that provide the HCP quantitation estimates. As these software packages can produce vast amounts of data, we have taken time to develop further data-filtering methods following the initial instrument/software-specific processing. This helps to target the real HCP identifications and eliminate false positives.
The algorithms that sort through these spectra are critical to the HCP-MS strategy in terms of time-efficiency, but there are potential errors that can lead to HCP mis-assignment or false positives as a result of the challenges of interpreting data over a five order of magnitude sensitivity range. To avoid too heavy a reliance on automated data review, we have found that manual review/human scrutiny of any HCPs that can have a significant impact on total HCP estimations is very valuable and helps to supplement the automated hits.
LC-MS is not a replacement for ELISA, but depending on the HCP coverage that is achieved through a commercial and/or process-specific kit, HCP-MS will assist in the characterization and understanding of HCP profiles. It may be too early to say whether the regulators favor a particular MS method over another. In theory, LC-MS has the sensitivity and the depth to identify differences with the HCP population from any upstream changes, and the regulators are starting to recognize its suitability in HCP analysis. Currently, however, HCP-ELISA analysis is a requirement and MS should be used to increase the breadth of HCP identification as an orthogonal technique.