Using matrix assisted laser desorption ionisation mass spectrometry combined with machine learning for vaccine authenticity screening

The global population is increasingly reliant on a wide range of vaccines to maintain population health with billions of doses used annually in immunisation programs worldwide. There is also a rise in substandard and falsified (SF) vaccines in supply chains which threatens individual and public health. SF vaccines can be caused by degradation of authentic vaccines, for example, due to incorrect storage conditions or inappropriate cold chain management, but there are also an increasing number of reports of deliberately falsified vaccine products. Unfortunately, there is currently no global infrastructure in place to monitor supply chains and no screening methods have been developed to identify falsified vaccines. In this study, we aimed to help address this by developing and validating a matrix-assisted laser desorption/ionisation-mass spectrometry (MALDI-MS) method that could distinguish authentic and falsified vaccines. We used Bruker MALDI Biotyper Sirius and bioMérieux VITEK MS instruments which have a worldwide distribution and are designed for clinical applications. Here we present a workflow, applicable on both instruments, that combines MALDI-MS analysis with open-source machine learning and statistical analysis that can differentiate authentic and falsified vaccines. In addition, multivariate data modelling and identification of diagnostic mass spectral peaks, are shown to have the potential for developing a vaccine authentication database. Our validated MALDI-MS method for vaccine authenticity testing provides proof-of-concept that such an approach could be scaled to address the latent need for more effective global vaccine supply-chain screening.