Multiomics: A new model for identifying biomarkers to predict severe COVID-19 outcomes  


Eye on Innovation

In a world of ever-faster technical change, Mayo Clinic Laboratories is uniquely positioned to innovate. Collaboration with clinicians pinpoints unmet patient needs and facilitates the development of diagnostic testing that provides answers.

In a groundbreaking study published in The Lancet Digital Health, Mayo Clinic investigators have developed a multiomic molecular method to predict clinical COVID-19 (SARS-CoV-2) outcomes better than traditional cytokines. Using a machine-learning-based prediction model, the team identified 102 biomarkers, which include several novel cytokines and other proteins, lipids, and metabolites. The discovery may help clinicians reliably predict a more severe course of COVID-19 before the patient gets sick enough to be hospitalized. This could also significantly improve clinical care because, up until now, there have been no biomarkers that can reliably predict which patients are more likely to have severe illness.

The multiomic molecular method employs a combination of two or more “omics” approaches to create a multidimensional, or layered, perspective. This is because there are many small chemicals in a blood sample that have different biochemical properties and, as such, reveal different information. 

“We needed a different analytical pipeline for this study,” says Akhilesh Pandey, M.D., Ph.D., professor of pharmacology and professor of laboratory medicine and pathology in the Department of Laboratory Medicine and Pathology at Mayo Clinic, and lead author of the study paper. “We have looked at thousands of these small molecules in our blood. We have looked at metabolites, lipids, and proteins in a global fashion.   

Akhilesh Pandey, M.D., Ph.D.

“I think this model is a nice demonstration of the use of multiomics, which combines the disciplines of metabolomics, lipidomics, cytokine profiling (Olink), glycoproteomics, and proteomics, with machine learning to develop a panel of predictive biomarkers. This is probably the largest multiomics study of plasma samples so far from Mayo Clinic. I believe it could serve as a guide to future, even more ambitious, efforts.”   

By applying multiple disciplines, the team created a collective “map” of how aspects of biological systems in the body are connected. Each “omicsmap offers a sampling of different types of biological molecules. Such maps can be used in biomedical research to understand what is happening on a molecular and cellular level in normal and disease states. 

Dr. Pandey continues, “We also identified additional plasma protein markers that have been implicated in inflammation, apoptosis, and other important cellular processes, including several that have not previously been described in the context of COVID-19.”  

Clinical implications 

Plasma samples were taken from 637 individuals. Of those, 455 patients tested positive for COVID-19 — either asymptomatic or showing varying degrees of symptoms. The Mayo Clinic study is novel in that it included in-depth details on each patient’s clinical status and outcome. Further, the level of technical expertise and rigor applied to the individual assets is unprecedented. 

Andrew Badley, M.D., is chair of Mayo Clinic’s Molecular Medicine Department and an infectious diseases clinician. He is also chair of Mayo Clinic’s COVID-19 Task Force. “The clinical value is really threefold: one, is to identify therapeutic targets,” he says. “Knowing what proteins or cytokines or chemokines are altered, and when, can inform us on which of those should be targeted. For example, if interleukin (IL)-6 is an early inflammatory target, that validates the use of IL-6 inhibitors for therapy. The second clinical use is to provide a roadmap to design rational combination therapies. Currently in the COVID space, the broad scientific community has done clinical trials on dozens and dozens of agents. And we’ve seen signals of effectiveness in a number of them. And in the history of infectious diseases, once you find viable therapies, the next step is to combine them.”

Andrew Badley, M.D.

Dr. Badley continues, “So understanding which protein values go up, where and when, coupled with the understanding of which therapeutic agents have signals of effectiveness, can help us design those rational combination strategies. The third clinical use is to identify patients who are at highest risk, or lowest risk, of developing severe or complex disease. People with severe disease are most in need of therapeutic interventions. So, we’re very interested in the changes that occur in patients with severe illness from COVID.” 

Once persons of higher or lower risk are identified, clinicians can then closely watch those individuals who fall into the high-risk category. Clinicians could also identify people who, because a particular biomarker’s value is so elevated, are at high risk of disease progression and, thus, could prioritize therapies based on those markers.

Going beyond “freeze-frame” pictures

Another novel aspect of this molecular model is, whereas other COVID-19 studies took plasma samples from patients after they were already severely ill and hospitalized, the Mayo Clinic team took plasma samples from many individuals who had yet to manifest COVID symptoms or were barely showing symptoms.

“What other studies have done is, they collected blood from patients who were already in ICU,” says Dr. Pandey. “Their study question was not, ‘How will they fare as the disease progresses?’ Instead, they wanted to find out what was wrong with these very sick patients in a ‘freeze-frame picture.’ So to get that freeze-frame picture, the patient samples were taken very late, after severe illness had already occurred. With that type of sample, the disease has already progressed, so you basically have the wrong sample to make any predictions.”

Mayo Clinic Laboratories Biomarker

In the Mayo Clinic study, some of the subjects did end up in the hospital, in which case their disease progression and outcome were closely followed. Dr. Pandey continues, “We took blood samples from people before they got hospitalized. Because only samples taken early like this can put you in a position to predict outcomes. And the only thing that we did with those patients, after taking their samples, was to collect data on what happened to them. But we didn't collect blood from them while they were in the hospital. We were just tracking them so that we could connect each sample to each person’s outcome, and how they fared.”

A “multiorgan disease”

At the outset of the pandemic, COVID-19 was known to be a big problem for the lungs if it progressed to the lower respiratory system. But it also become quickly apparent that the disease can do damage throughout the body. 

“There are many things going wrong inside these patients,” says Dr. Pandey. “So we still need to study them. What exactly is happening? What is going to cause kidney damage? And people are talking about brain fog and symptoms like that. So, yes, the original textbook description was that COVID-19 is a big problem for lungs, but we realized early on that this is a multiorgan disease. And we would like to learn more about how our metabolism is altered. The best way to study that is to take blood samples and try to see what proteins we can measure, what metabolites and lipids we can measure, that will basically tell us what is causing the damage. And, of course, we want to be able to predict that before such damage occurs.”

Dr. Badley sums up the overreaching purpose of such multiomic studies this way: “We’re looking to find changes in biomarker values that can inform novel therapeutic targets and combination therapies. We want to lower the patient’s risk of ending up in the hospital, especially the ICU.”

More from Eye on Innovation

 

Chris J. Bahnsen (@cjbahnsen)

Chris J. Bahnsen

Chris J. Bahnsen covers emerging research and discovery for Mayo Clinic Laboratories. His writing has also appeared in The New York Times, Los Angeles Times, and Smithsonian Air & Space. He divides his time between Southern California and Northwest Ohio.