Presented at the Biotherapeutics Analytical Summit 2018, Baltimore, MD, USA
Selecting the best possible monoclonal antibody as drug candidate requires extensive characterization to mitigate the risk of the molecule failing in late stage development. Here we present a rapid and fully automated data processing and analysis approach to leverage intact mass screening for in-depth understanding of the molecule.
The intact mass analysis of antibodies typically includes the identification of the elution time ranges for each fragment and subsequent averaging of LC-MS spectra that are deconvolved for each discrete time range. This often hinders accurate quantification since overlapping peaks result in only partial integration of the total available molecule amount. Further splitting of the elution time ranges and deconvolution is necessary to detect co-eluting lower-abundant versions of the main components. Furthermore, the high number of deconvoluted spectra makes it difficult to gain a complete understanding of the sample since the relationship between the signals is lost. Finally, such a manual approach is extremely time-consuming and may even introduce significant bias.
To address these issues, we developed an automated two-dimensional time-resolved deconvolution algorithm, which allows for the identification and accurate quantification of even trace amounts of large molecules in one single plot, including co-eluting components and in-source decays. It further overcomes the problems of the time-intensive and non-reproducible selection of elution time ranges. Major signals belong to the Fd′ fragment with fully reduced intra-chain disulfide bonds plus its oxidized version were identified. Other modified versions of each main signal were also detected and quantified. Glycation, succinimide formation at an asparagine amino acid, partially reduced and the fully reduced versions of the disulfide bridges, and several ion adducts were quantified. The advantages for performing structural characterization of antibodies by time-resolved deconvolution compared to the classical approach are illustrated.