In all, 213 subjects were enrolled; of these, 100 had definitive microbiologic diagnoses and 94 had transcriptomic data passing quality control metrics. Among the 94 patients, the mean age was 61 years, and 92 patients had at least 1 chronic medical condition. Asthma exacerbation was the diagnosis in 17 of the subjects, while 25, 21, 23, and 8 were diagnosed with bronchitis, acute exacerbation of COPD, pneumonia, and bacteremia, respectively. Overall, 41 subjects were classified with “bacterial” infection (27 bacterial only and 14 mixed viral/bacterial) and 53 subjects were classified with “nonbacterial” infection (viral infection alone).
In subsequent genetic analysis, the University of Rochester team assessed the expression of 10 marker genes (BTN3A3, IFI27, RSAD2, KIAA1618, OAS2, IFIT3, IFI44, OASL, IFIT2, and PARP9) identified in earlier research
to determine their ability “to distinguish bacterial from nonbacterial illness.” In all, they found that 8 of the 10 genes (IFI27, RSAD2, KIAA1618, OAS2, IFIT3, IFI44, OASL, and IFIT2) “demonstrated significant differences between groups” based on Wilcoxon Rank test at a false discovery rate (FDR) or q <.05. In addition, based on qPCR, they noted that all 10 demonstrated “significant difference between bacterial and nonbacterial groups” according to Wilcoxon Rank test, at a nominal P
value of <.05—and that expression of all 10 is associated with nonbacterial infection.
In addition, the authors identified 141 genes “that are differentially expressed between subjects with bacterial versus nonbacterial infections,” with most having higher expression in subjects with bacterial infection. They identified reduced expression of these genes in study subjects with a clinical diagnosis of asthma and bronchitis, and increased expression in those with pneumonia and bacteremia (COPD yielded a mixed pattern).
Finally, the research team identified 3 pathways (lymphocyte, α-linoleic acid metabolism, IGF regulation) and 11 genes (ICAM1, ITGAL, ITGB2, PECAM1, IGFBP6, IGFBP2, CTSG, MMP2, ACOX3, FADS2, and PLA2G4A) with which they were able to construct a classifier for bacterial LRTIs with 90% sensitivity and 83% specificity.
“Our results are encouraging [but] this was a small study,” Dr. Falsey said. “We have to further refine our predictive genes; [however], if our results can be prospectively validated, we believe it can be a really promising approach for differentiating bacterial infections clinically.”
Brian P. Dunleavy is a medical writer and editor based in New York. His work has appeared in numerous healthcare-related publications. He is the former editor of Infectious Disease Special Edition.
To stay informed on the latest in infectious disease news and developments, please sign up for our weekly newsletter.