Sepsis, defined as the systemic inflammatory response syndrome (SIRS) in the setting of suspected infection, is directly responsible for at least 200,000 deaths per year in the United States, 1 accounting for more attributable deaths than breast cancer, colon cancer, leukemia, lymphoma, and ovarian cancer combined. 2 By consensus, sepsis is defined as infection combined with the systemic inflammatory response syndrome (SIRS). 3 The pathophysiology of sepsis is complex, and the relative contributions of host-mediated responses and the specific causative pathogen to poor patient outcomes are unclear. In a patient population with clearly defined bloodstream infections, we sought to determine whether acute bloodstream infection with different pathogens was associated with different concentrations of host biomarkers reflective of different physiologic processes. From the standpoint of the host response, there are substantial differences between the degree of acute inflammation induced by specific bacterial species in pre-clinical animal model systems with controlled inocula of pathogen. 4–8 In pediatric populations with hematologic malignancy, differences were observed in inflammatory cytokine production in patients with gram-positive cocci when compared to gram-negative bacilli infections. 9 In adults, higher levels of pro-inflammatory IL-6 have been reported in patients with gramnegative vs. gram-positive bacteremia when measured during routine clinical care. 10 Higher levels of the proinflammatory cytokines IL1b, IL-6, and IL18 responses have been reported in human leukocytes stimulated with LPS when compared with heat-killed S. aureus, a finding mirrored in the elevations of the serum of patients with unspecified “infections” from gram-positive pathogens. 11 The in vitro differences are contrasted by in vivo findings from a different group who found no difference in IL-6 between patients with sepsis from pure gram-positive infection when compared to pure gram-negative infections. 12 No differences were found at the level of mRNA expression of neutrophils from patients with gram-positive versus gram-negative infections. 13 In terms of clinical outcomes using administrative databases, authors have reported either no difference in outcome 14 or a roughly 6% crude mortality difference with respect to the inciting pathogen. 15

In experimental systems, the acute hyperinflammatory response has traditionally been thought to be followed by a subsequent compensatory antiflammatory response syndrome (CARS) where patients are thought to be at risk from pathogens of normally low virulence. 16, 17 This two-phase response has been brought into question recently as markers associated with both inflammation and an anti-inflammatory response (eg IL-10) are elevated simultaneously in both routine clinical care and experimental systems. 18, 19 We are aware of no data examining the host response to pathogens of normally low virulence.