Skip to main content
Premium Trial:

Request an Annual Quote

Researchers Identify Protein Markers for Predicting Severity, Hospitalization in Flu Cases


A team led by researchers at the St. Jude Children's Research Hospital in Memphis, Tenn., has identified protein biomarkers that could help predict severity of flu infections.

In a study published in December in the American Journal of Respiratory and Critical Care Medicine, the researchers determined that increased patient levels of nasal lavage MCP-3, IFNα2, and plasma IL-10 predicted progression to severe disease while increased plasma IL-10, MCP-3, and IL-6 levels predicted patient hospitalization.

The study also identified MCP-3 as a potential standalone marker for identifying patients likely to be hospitalized, St. Jude researcher Paul Thomas, lead author on the paper, told ProteoMonitor. He and his colleagues are currently exploring the protein as a potential quick and inexpensive aid in determining whether to admit flu patients for hospitalization, he said.

The work was part of a larger five-year effort looking at more than 300 flu patients with the aim of identifying biomarkers for predicting the progression and severity of the infection as well as better understanding the mechanisms underlying the disease. The AJRCCM study looked at 73 flu patients, measuring the levels of 39 and 42 different cytokines in plasma and nasal lavage samples, respectively.

In correlating protein expression with disease progression, the researchers looked at a predetermined subset of 11 cytokines representing a broad range of immune activities, a decision made due to the study's limited sample size, Thomas said.

"We knew that we were going to be measuring 42 different cytokines, and that statistically it was going to be impossible with our patient sample size to test every one of those and retain statistical power," he said. Given this, his collaborators and co-authors at the Fred Hutchinson Cancer Research Center suggested they use a subset of the 42 measurements, Thomas said.

From these measurements, the researchers identified the two predictive profiles. Interestingly, Thomas noted, while both the nasal lavage and plasma measurements appeared to correlate with disease progression, the two didn't correlate particularly well with one another.

"Even the same cytokine in the nasal wash and the plasma weren't correlated," he said, noting that this was to be expected based on mouse models of flu infection.

"We know [from such models] that the airway microenvironment is a completely different environment from the systemic immune environment, which is what you are measuring in plasma," he said. He added that the researchers were able to make better mechanistic sense of their nasal lavage findings, observing that while "in the airways everything you are looking at is relevant to the infection... what we were seeing in the blood was just a bit noisier."

Among the study's most surprising results, Thomas said, was that immune memory didn't appear to correlate with disease severity.

"I think the assumption in the field has been that the susceptibility of children [to flu] relates to their lack of immune memory," he said. The study found, though, that children naive to flu infection "actually mounted quite vigorous immune responses."

Thomas suggested, however, that this finding might be due in large part to the resolution of assays used to measure protective immunity.

"The assays we have are pretty blunt," he said. "We know that there are emerging protective classes of antibodies that don't read out in those assays at all. So you can score completely negative in those assays and still be protected via antibody, and you can also have relatively high levels of those antibodies and not be protected."

Also interesting, Thomas noted, was the researchers' finding that severity of infection did not correlate with viral load.

"Certainly, if you fail to control viral replication, that is going to have clinical consequences," he said. "But what we observed in our cohort is that you had a group of people who got severely ill, and that didn't seem to have any relationship to a difference in viral loads. So it's not that viral load can't be predictive, it's just that in many cases, it's not."

This, Thomas noted, raises the "basic immunological question" of what, if not viral replication, drives pathogenesis of flu, a question the researchers hope their protein analysis can help answer.

"There's a lot of mouse model data [indicating] that immune response plays a role in pathogenesis," he said. "So can we from a human system determine what those key parameters are and try to work out the mechanism of why they are?"

Beyond mechanistic questions, Thomas and his colleagues hope also to further validate their results with the aim of developing a test that doctors could use to help determine whether or not to admit a flu patient to the hospital. He cited the protein MCP-3 as particularly promising, noting that it appeared to predict "reasonably well who would be hospitalized upon enrollment."

He said he envisioned such a test as a cheap, fast assay doctors could run such that a flu patient "walks in the door and get the result back right away."

The researchers have no particular interest in patenting such an assay, Thomas said.

"Certainly there are plenty of commercial vendors out there making tests for that marker, so our interest would really just be in looking to facilitate its use and really validate it as a clinical correlate that is helpful in diagnostics," he said.

"There's a lot of work to be done before it could be adopted," he added, estimating that any such test was several years away.