Zoe Berg, Contributing Illustartor

A recent study from the Foxman lab at the Yale School of Medicine has indicated that screening patients for a certain cytokine biomarker could be a key way to identify new and dangerous viral pathogens emerging in human populations, improving current public health surveillance systems. 

According to Ellen Foxman, an associate professor of laboratory medicine and immunobiology at the Yale School of Medicine, there are three current methods most used for detecting new viral pathogens: surveying for outbreaks of infections not explained by a known virus and sequencing samples of interest, monitoring for viruses in animals that have the biological potential to jump into human populations and sequencing random clinical samples to find genetic sequences that could correspond to viral infections. 

The findings of this study would mainly improve upon the third method of surveillance. Testing for this cytokine biomarker, called CXCL10, could help to drastically reduce the number of samples needing to be sequenced in order to detect new viral infections.

“All of the methods are great, and they all complement each other, but it’s good to come up with finding new methods that help to make this process of finding viruses that should be on our radar screen more efficient,” Foxman told the News. “And that’s what this project was about- making a more efficient system to screen for viruses that we are not aware of.”

This study builds off of a 2017 paper from the Foxman lab, where the researchers discovered the significance of CXCL10 as a biomarker. At many hospitals, including the Yale New Haven Hospital, nasal swab samples are often gathered from symptomatically ill patients in order to test for 15 of the most common viral infections, such as influenza and the common cold. The researchers found that levels of CXCL10 were typically significantly elevated in patients ill with any type of viral infection.

According to Foxman, this is due to the innate immune system — the part of the immune system that works to detect any pathogenic invaders. While adaptive immunity defends against a specific viral infection such as the common cold, influenza or COVID-19 based on previous encounters with these viruses, the innate immune system simply launches an immune response against any detected pathogen.

 The innate immune system can only tell the difference between broad categories of infection, such as bacterial and viral, and would react similarly to any type of viral infection. Foxman described the process of looking at CXCL10 levels as “eavesdropping” on the innate immune system in order to improve our detection of disease.

Since high CXCL10 levels were found to correspond to viral infection, the researchers believed that this marker could be used to help identify new viral infections. The idea was that if a patient tested negative for all viruses being screened for but also had high levels of CXCL10, their sample should be genetically sequenced to see if the patient was actually infected by a new viral infection that was not being screened for. 

The researchers first tested this process in nasopharyngeal samples that were screened for viruses at the Yale New Haven Hospital between Jan. 23 and Jan. 29, 2017. The researchers utilized machine learning and clustering methods on the sequenced samples to find that there was a significant difference between CXCL10 levels in patients who tested positive for viral infections and those who did not. 

“Machine learning allows a thorough interrogation of high dimensional data to arrive at a meaningful answer as to whether or not it is possible to differentiate different infections via their cytokine features, as well as the relative importance of those features,” said Jason Bisha GRD ’26, a doctoral student in the department of microbial pathogenesis at Yale and one of the lead authors of the study. “Through identifying and extracting the most important features that drive the classification decisions, it is possible to get a better understanding of the biological particulars of an infection as well as create diagnostic criteria for clinical use.”

Furthermore, after testing for this difference, the researchers obtained eight samples that had significant levels of CXCL10 but tested negative for all viral infection. They then sequenced these samples, believing that it was possible that they may have been infected by a new virus. The team detected the presence of influenza C in one of the samples, which is a rare type of influenza that was not screened for in the patients.

This process was repeated in samples from the first week of March 2020, and the researchers were able to find four previously undetected cases of COVID-19 in samples that tested for significant levels of CXCL10 but tested negative for all other screened viral infections.

Foxman believes that testing for CXCL10 could be “piggybacked” onto routine screenings for viral illnesses in order to help with public health surveillance. Instead of randomly choosing samples to be screened for new viral infections, researchers could focus on sequencing samples that test negative for all other viruses but test positive for high CXCL10 levels, since these are the samples that are most likely to have an unknown viral infection. According to Foxman, this could cut down on 90 percent of sequencings and significantly increase the odds of detecting new viruses. 

Foxman hopes that screenings for CXCL10 will expand outside of the New Haven Hospital in order to improve overall levels of public health surveillance methods.

“I think that we have a capacity to do this in New Haven and our hospital here, but I really hope that other groups trying to do surveillance for emerging or undiagnosed respiratory viruses will also implement this technique with samples from other places,” Foxman said. “Or, that we can set up collaborations with people to look at samples that are coming from other countries or other regions.”

According to Foxman, early detection of new viral pathogens is especially important in this day and age, in which increasing population density and global travel makes a pandemic situation such as with COVID-19 much more likely.

Additionally, Jorge Alfaro-Murillo, an associate research scientist in the department of biostatistics who specializes in modeling of infectious diseases, described that improved methods for public health surveillance like Foxman’s work can not only help to prevent the next pandemic, but also help to combat other growing public health issues in the United States, such as antibiotic resistance.

“Early detection of pathogens is vital for global public health because the effect of public health policies can be increased substantially if those policies are implemented as early as possible during an outbreak,” Alfaro-Murillo wrote. “Additionally, implementation of methods for viral pathogen detection in clinical settings would help to reduce antimicrobial resistance by decreasing antibiotics overprescription. Antimicrobial resistance is one of the top 10 global public health threats facing humanity according to the World Health Organization.”

Funding for this research was received from National Institutes of Health, the Hartwell Foundation, the Gruber Foundation, Fast Grants for COVID-19 research from the Mercatus Center and the Huffman Family Donor Advised Fund.

Jessica Kasamoto covers the Yale School of Public Health for the SciTech desk. She is a graduate student in computational biology and bioinformatics.