Finding a Way to Predict the Severity of COVID-19 in Children

Children with severe SARS-CoV-2 infection had significantly higher levels of a trio of salivary cytokines, according to a preliminary analysis.

In the pilot study with 180 children, those with severe COVID-19 had higher levels of CXCL10, MIG, and TNF R-1 in their saliva against children with mild COVID-19, reported Usha Sethuraman, MD, of University Central Michigan at Mount Pleasure, at the American Academy of Pediatrics (AAP) virtual meeting.

Although specific cytokine profiles may not be related to the various symptoms of COVID-19 infection – in part because of a small sample size – the researchers hope that additional data will enable them to distinguish biomarkers of COVID-19 pediatric patients, he noted.

“Many of these children with COVID have symptoms very similar to those with other common viral infections,” says Sethuraman MedPage Today. “And there’s no way for us to tell which child is going to get sick or not. So if we have a method to decide which child is going to develop [multisystem inflammatory syndrome (MIS-C)] or severe COVID, that would help us start treatment earlier. “

“So in that sense, this would be a game changer,” he added.

The incidence of COVID-19 in children is generally mild, but some younger patients may develop MIS-C, Kawasaki disease, or respiratory failure. In the 24 states and New York City that report on children’s hospitals, children are hospitalized in 0.1% -1.9% of all COVID-19 pediatric cases, according to September 2021 AAP data.

Currently, there are no established biomarkers that can predict COVID-19 progression and severity in children, although studies have linked disease severity to improved cytokine levels in adults.

Sethuraman said his group chose saliva cytokines because of the ease of sampling. “The main purpose of this study was to make it as noninvasive as possible,” he said.

The study began in March 2021 and will run until June 2022. Initial data from March through May came from Michigan Children’s Hospital in Detroit and UPMC Children’s Hospital in Pittsburgh. The analysis of saliva samples is being done at Penn State College of Medicine in Hershey, Pennsylvania, while the development of models using artificial intelligence (AI) is being done at Wayne State University in Detroit.

Of the 180 children (mean age 7.1 years; 49.9% female) enrolled in the study to date, 60 were hospitalized and 40 were categorized as having severe COVID-19. Of those 40, five exhibited cardiac symptoms, 26 had severe respiratory symptoms, while the others exhibited neither, but “were predominantly diagnosed with MIS-C,” according to Sethuraman.

Six salivary cytokines – TNF R-1, IL-13, IL-15, CCL7, CXCL10, and MIG (CXCL9) – were measured, and all three of the elevated cytokines in patients with severe disease were pro-inflammatory , researchers reported.

They also found that cytokines did not vary significantly (P.> 0.05) between various COVID-19 phenotypes, and that hierarchical logistic regression shows that a three-cytokine model accounts for “only 4.2% of the variance between severe and non-severe COVID groups.”

“The combined model of the three cytokines with age, sex and asthma status showed moderate accuracy (75%) and poor sensitivity (21%),” the researchers noted, which was the limitation of the study, along with the number small number of patients.

Sethuraman and colleagues also measured miRNA levels in 129 saliva samples. They found that levels of 63 miRNAs varied significantly between severe and non-severe cases in their initial analysis, and showed promise as a predictive model.

“Our ultimate goal is to incorporate all the features, including social determinants of health and clinical knowledge, and use AI to make a model,” Sethuraman said. “[We want] to narrow down to a few miRNAs and some cytokines, and develop a bedside gadget [that] predicts for us whether a child has a serious disease. “

  • author['full_name']

    Lei Lei Wu is a news intern for Medpage Today. She is based in New Jersey. Follow


The study was supported by the Eunice Kennedy Shriver National Institute for Child Health and Human Development / NIH Rapid Diagnostics Acceleration Program (RADx).

Sethuraman did not disclose any relationships with industry. Co-authors revealed relationships with Quadrant Biosciences, Moderna, and Novavax.