Abstract
Background Epidemiologic studies have linked antibiotic exposure to subsequent sepsis, suggesting that microbiome disruption may be in the causal pathway and an independent risk factor. This study tests whether variation in the gut microbiota associates with risk of sepsis onset and its outcomes.
Methods Using a validated surveillance definition, patients with an archived rectal swab from intensive care and hematology units were screened for sepsis. After confirmation by chart review, cases were matched to controls in a 1:2 ratio based on age, gender, and collection date. Relative taxon abundance was measured by sequence analysis of 16S rRNA gene amplicons; total bacterial abundance was measured by qPCR of the 23S rRNA gene. Conditional logistic regression identified clinical and microbiota variables associated with sepsis.
Results There were 103 sepsis cases matched to 206 controls. In a final model adjusting for exposure to broad-spectrum antibiotics and indwelling vascular catheters, high relative abundance (RA) of Enterococcus (Odds Ratio (OR) 1.36 per 10% increase, P=.016) and high total bacterial abundance (OR 1.50 per 10-fold increase in 23S copies/μL, P =.001) were independently associated with sepsis. Decreased RA of butyrate-producing bacteria also independently associated with sepsis (OR 1.20 for 10% decrease in RA, P =.041), and mortality in unadjusted analysis (OR=1.47 for 10% decrease in RA, P=.034).
Conclusions This study indicates that the microbiota is altered at sepsis onset. The decreased RA of butyrate-producing bacteria in sepsis also associates with mortality, suggesting a therapeutic role for prebiotics and probiotics in the prevention and treatment of sepsis.
Importance Early detection of patients at risk for sepsis could enable interventions to prevent or rapidly treat this life-threatening condition. Prior antibiotic treatment is associated with sepsis, suggesting that disruption of the bacterial population in the gut (the intestinal microbiome) could be an important step leading to disease. To investigate this theory, we matched hospitalized patients with and without sepsis and characterized the patients’ microbiomes close to or at onset of sepsis. We found that several microbiome alterations, including having more total bacteria in the gut was associated with onset, regardless of prior antibiotic treatment. This signature of microbiome disruption brings us closer to identifying the biological causes of sepsis and could be used to develop new diagnostic tests to identify patients at risk of sepsis.
