NEW YORK – Researchers at the Max Planck Institute of Biochemistry are using mass spectrometry to analyze the plasma proteomes of 50,000 individuals to identify biomarkers linked to maternal and fetal health.
Detailed in a BioRxiv preprint published in March, the effort is one of the largest to date to use mass spectrometry for population-scale plasma proteomics analysis.
Over the last half decade, researchers have conducted a number of large-scale plasma proteomic studies measuring thousands of proteins in tens of thousands of individuals, assessing their links to a variety of health conditions. These studies have not used mass spectrometry, however, but rather affinity-based platforms from Olink (now part of Thermo Fisher Scientific) and SomaLogic (now part of Standard BioTools), which until recently were the only technologies that offered the combination of depth of coverage and throughput required to analyze such large sample cohorts.
Advances in mass spec instrumentation and plasma protein enrichment have made feasible studies analyzing tens of thousands of samples, but, to date, few if any mass spec studies on this scale have been undertaken.
"It's kind of the mass spec answer to the UK Biobank," said Vincent Albrecht, a graduate student in the lab of Max Planck researcher Matthias Mann and first author on the preprint.
The UK Biobank's Pharma Proteomics Project (PPP) has generated one of the largest population proteomics datasets, releasing in 2023 a set consisting of roughly 3,000 proteins measured in blood samples from 54,000 UKB participants produced using Olink's Explore platform. In January, the biobank said it has launched a study that will measure up to 5,400 proteins in 600,000 blood samples.
The samples used in the Max Planck study come from the Multi-Omics for Mothers and Infants (MOMI) Consortium, which is supported by the Bill and Melinda Gates Foundation. The project aims to define the underlying biological risks for adverse pregnancy outcomes and prioritize early interventions.
Albrecht said he and his colleagues have analyzed around three-quarters of the 50,000-sample cohort. Running all 50,000 samples will take around a year total using two mass spectrometry systems, he said.
He said that advances in sample preparation as well as improved instrumentation, including the Evosep One liquid chromatography system and the Thermo Fisher Scientific Orbitrap Astral, had enabled the researchers to tackle a cohort of this size.
For the study, the Max Planck team is using a perchloric acid-based workflow called PCA-N. The approach is based on a method developed by Boston Children's Hospital researchers Judith and Hanno Steen in which plasma samples are treated with perchloric acid to deplete high abundance proteins, allowing for improved depth of coverage.
Albrecht said that he and his colleagues evaluated a number of plasma proteomic workflows before settling on the perchloric acid-based workflow. They chose the workflow in large part due to its robustness and its robustness to sample contamination in particular, Albrecht said.
"We know the plasma is collected at different sites and different clinics, and a key driver for variance is contamination based on platelets, and leukocytes, and coagulation," he said. "If you have different nurses or different phlebotomists, they may take the blood in different ways, and when you start getting, for instance, platelets in your sample, the biology is masked by this sort of contamination."
The MOMI samples were collected at different sites in Zambia, Tanzania, Pakistan, and Bangladesh over multiple years.
Albrecht said that the workflow's low sample requirement, which the researchers enabled by adding a neutralization step, was also key as it made sample prep compatible with a high-throughput 384-well plate (and, potentially, a 1,536-well) format. They were able to process all 50,000 samples in seven days.
Achieving low marginal costs was also important to the team, Albrecht said, noting that they hope for the method to be widely accessible. The cost per sample of the PCA-N workflow is essentially the same as the cost per sample of a neat plasma analysis.
Albrecht said that in the MOMI cohort he and his colleagues are measuring close to 2,000 proteins per sample using the PCA-N approach. The researchers also included in the BioRxiv preprint data on 1,500 quality control samples interspersed among the MOMI cohort samples, reporting intraplate CVs of 17.7 percent across the study's duration of nearly one year.
A number of new mass spec-based plasma proteomic workflows have become available in recent years. Companies including Seer, Biognosys, and PreOmics have developed particle-based enrichment methods as has the lab of University of Washington researcher Michael MacCoss. A recent study by University of Wisconsin-Madison researcher Josh Coon compared several of these approaches, finding that Seer's Proteograph XT platform, PreOmics EnrichPlus product, and University of Washington approach (called Mag-Net) outperformed a perchloric acid-depletion approach both in terms of depth of coverage and reproducibility. (The Biognosys system was not included among the approaches evaluated.) In an email, Seer Senior VP and Scientific Fellow Asim Siddiqui said that the company is in some cases measuring more than 9,000 proteins per plasma sample.
Sara Ahadi, associate director of multiomics and integrative analysis at Grifols subsidiary Alkahest, said that while perchloric acid depletion does not offer the depth of coverage of other methods, its low cost and ease of use make it an attractive option.
"We use it in our lab, and it's a favorite [approach] even though it is not the deepest," she said. "We have seen in our data that it is more robust than nanoparticle or bead enrichment" methods.
She added that the neutralization step incorporated by the Max Planck researchers further streamlines the method while also making it more amenable to high-throughput automation.
In March, Ahadi, who is not involved in the Max Planck MOMI work, and her colleagues published results from a study comparing several commercially available plasma proteomic workflows, though they did not include a perchloric acid-depletion approach.
Ahadi also applauded the Max Planck team's demonstration of its workflow's stability over the course of many samples and months of analysis.
"You have multiple levels of [QC] information — interplate, intraplate, a large number of samples over a long time," she said. "They have to look at all of it in detail. It really shows how you would be able to maintain this quality, [addressing] any factor that could cause robustness to go off."
"I think their attention to detail and the quality of data is very important," Ahadi said. "They basically didn't play the number game, they played the quality game, and that will give us advantages we will see in the biological data that is going to come out of this."
While population proteomics studies have thus far relied almost exclusively on affinity-based methods, the Max Planck-MOMI work suggests that mass spec is now a viable option for such work. Perhaps as important, it indicates that organizations like MOMI who possess large sample cohorts are ready to give mass spec approaches a look.
Ahadi noted that among mass spec's advantages is its ability to provide peptide-level information that can help identify specific protein isoforms present in a sample as well as its ability to provide protein posttranslational modification data that is not typically available with affinity-based approaches.
Effectively generating and making use of that data takes significant expertise, however, which Ahadi suggested has slowed mass spec's move into population studies.
"Mass spectrometry data is not as easy to interpret as what you get from affinity-based readouts," she said. "It has a lot of complexity and levels that, if you can take advantage of them, you can benefit from them a lot … but it requires a high level of understanding of the mass spectrometry data to take advantage of those extra levels."