#20 Bloodstream Infection is Associated with Prior Gut Microbiome Colonization of the Same Strain in the Neonatal Intensive Care Unit

Abstract

Abstract Background Infants hospitalized in neonatal intensive care units (NICUs) have high incidence of bacterial bloodstream infections (BSIs) leading to up to 15% mortality and substantial morbidity. Common etiologies of BSI in the NICU are pathogens also capable of colonizing the preterm infant gut. However, differentiating susceptible neonates from those in the same environment remains challenging. We hypothesized that the same strains causing BSI increase in abundance in the gut of affected neonates prior to bacteremia and increased above controls in the same environment. Method We performed shotgun metagenomic sequencing on 462 fecal samples from 19 infants with BSI and 37 controls in the same NICUs paired for gestational age, days of life, antibiotic exposure, delivery mode, and sex. We whole genome sequenced the BSI isolate from 19 cases (6 Staphylococcus aureus, 3 Streptococcus agalactiae, 3 Serratia marcescens, 2 Enterococcus faecalis, 2 Klebsiella pneumoniae, 2 Escherichia coli, and 1 Enterobacter cloacae). We utilized MetaPhLan3 to infer taxonomy and InStrain to track the whole genome sequence of the BSI pathogen within the fecal metagenomes longitudinally. We performed generalized linear mixed effect models with MaAsLin2 to determine significant associations between the microbiome and metadata with subject as a random effect. Results We found no differences in gut microbiome diversity or richness between cases and controls relative to bacteremia day of life. However, infants who experience a BSI possess significantly higher median relative abundances of the causative species in their gut relative to paired controls in the 2 weeks prior to bacteremia (Figure 1A, median 21.4% IQR 0.33-52 vs. 0.4% IQR 0.23-10.5, p<0.05 Wilcoxon). 10/19 infants produced at least 1 stool before bacteremia containing an organism with 0-1 genomic substitutions from the BSI isolate, establishing isogenicity between gut residing and invasive pathogen. The coverage (abundance) of the BSI isolate increased in the stool of affected neonates in the 10 days before infection to an average maximum of 0.48/1.0 (Figure 1B, linear mixed effect model p<0.05). We observed the strongest associations between high gut microbiome abundance of E. faecalis, Klebsiella spp., E. coli, and S. marcescens in the gut microbiome of cases whose BSI was caused by that organism compared to all other samples (p<0.01). Conclusion We show high relative abundance of the same strain in stool prior to bacteremia in hospitalized neonates. The relative abundance of the causative species is higher in cases versus paired controls. These data suggest the potential of gut microbial surveillance and management to identify blooms of gut pathogens to limit the incidence of bacteremia.