RNA-expression biosignatures derived from the patient’s peripheral blood distinguish bacterial from viral causes of fever in young children, according to two separate preliminary studies published online Aug. 23 in JAMA.
Several studies have suggested that the source of infection in febrile children might be identified by examining the pattern of host genes that are either activated or suppressed during the body’s inflammatory response. Distinguishing the relatively few but potentially life-threatening bacterial infections from the more common but milder, self-resolving viral infections is difficult, and current practice is to admit “ill-appearing” febrile children to the hospital and administer parenteral antibiotics while awaiting the results of blood and tissue cultures. Those results are often ambiguous, and the whole process represents a large burden on health care resources as well as contributing to inappropriate antibiotic treatment.
Dr. Jethro A. Herberg
Two multinational research groups developed different techniques for detecting RNA biosignatures in patients’ blood samples, then assessed the accuracy of those tests in validation cohorts. One group focused on ruling out bacterial infection as the source of fever in young children (median age, 19 months), while the other investigated whether the host responses of the youngest children (aged 60 days and younger), who have immature immune systems, are robust enough to allow detection of RNA biosignatures.
In the discovery phase of the first study, analysis of RNA gene expression was performed on blood samples obtained from 240 children at admission to hospitals in the United Kingdom, Spain, and the United States during a 4-year period. A total of 8,565 RNA transcript signatures were identified as potential biomarkers to discriminate between viral and bacterial infection. This was narrowed down to 38 transcript signatures, and then to only 2 – IFI44L and FAM89A – that were used to devise a Disease Risk Score (DRS) for each patient, said Jethro A. Herberg, PhD, of the division of infectious diseases, Imperial College London, and his associates.
IFI44L expression was increased in patients who had viral infection, while FAM89A expression was increased in those who had bacterial infection, as compared with healthy children. (In previous studies, IFI44L was reported to be up-regulated in interferon-mediated antiviral responses and FAM89A was reported to be elevated among children with septic shock.)
The DRS showed 90% sensitivity in distinguishing viral from bacterial infection in the discovery cohort. It then showed 96.4% sensitivity in a validation cohort of 130 febrile children (mean age, 17 months). The DRS also identified bacterial infection in a validation cohort of 24 children with meningococcal infection (91.7% sensitivity and 96.0% specificity), and distinguished it from inflammatory conditions in another cohort of 30 children with juvenile idiopathic arthritis and 18 with Henoch-Schönlein purpura (90.0% sensitivity and 95.8% specificity).
The DRS discriminated among viral, bacterial, and inflammatory diseases including systemic lupus erythematosus in a further validation cohort, a published dataset from children and adults who had all three types of illness. It was accurate regardless of the severity of infection and regardless of the duration of infection, as well as in cases where patients were coinfected with both virus and bacteria, the investigators said (JAMA. 2016 Aug 23. doi:10.1001/jama.2016.11236).
“The DRS signature, distinguishing viral from bacterial infections with only two transcripts, has potential to be translated into a clinically applicable test using current technology. Furthermore, new methods for rapid detection of nucleic acids, including nanoparticles and electrical impedance, have potential for low-cost, rapid analysis of multitranscript signatures,” Dr. Herberg and his associates noted.
Further research is needed to assess the accuracy and clinical utility of this technique in different settings, they added.
In the second study, RNA gene expression was analyzed from blood samples from 1,883 febrile infants (median age, 37 days) “who posed diagnostic quandaries” at admission to 22 emergency departments during a 2-year period. The discovery phase involved 89 of these infants who ultimately were found to have bacterial infections (bacteremia or UTIs), 190 who didn’t have bacterial infections (enterovirus, influenza, or other viruses), and 19 healthy control infants, said Prashant Mahajan, MD, division chief and research director, pediatric emergency medicine, Children’s Hospital of Michigan, Detroit, and his associates.
Dr. Prashant Mahajan
The investigators identified 3,753 RNA transcript signatures that could potentially identify or rule out bacterial sources of infection, which they then narrowed down to 66. This set of 66 signatures showed 82% sensitivity and 88% specificity in the discovery cohort and 87% sensitivity and 89% specificity in a validation cohort.
“The bacterial RNA biosignature was notably more predictive of bacterial infection than clinical examination” and use of the Yale Observation Score, and it “added significantly to prediction beyond the YOS alone,” Dr. Mahajan and his associates said (JAMA. 2016 Aug 23. doi: 10.1001/jama.2016.9207).