Abstract
OBJECTIVES: The Protocol-based Care for Early Septic Shock trial found no differences across alternative resuscitation strategies in all-cause mortality. A separate aim was to determine whether differences in resuscitation strategies affected trajectories of biomarkers of key pathways associated with downstream clinical outcomes of sepsis and whether there were differences in survival across treatment arms for patients with different baseline biomarker profiles. DESIGN: Secondary analysis of a large randomized clinical trial. SETTING: Thirty-one U.S. hospitals. PATIENTS: Six hundred twenty-eight patients with septic shock. INTERVENTIONS: Two resuscitation protocols versus usual care. MEASUREMENTS AND MAIN RESULTS: We measured a panel of biomarkers representing four pathophysiologic domains: "inflammation" (tumor necrosis factor, interleukin-6, and -10); "coagulation" (D-dimers, thrombin-antithrombin complex); "oxidative stress" (urine isoprostane); and "tissue hypoxia" (lactate) at 0, 6, 24, and 72 hours after treatment. We analyzed whether alternative resuscitation strategies affected biomarker trajectories over 72 hours and whether effects on 90-day hospital mortality varied by baseline (time 0) biomarker profiles-both using regression models with interaction terms for treatment arms. For all baseline biomarkers, higher concentrations were associated with increased risk of death by 90 days. However, there was no significant effect of treatment assignment on subsequent biomarker trajectories. We did find evidence for heterogeneity of treatment effect of protocol-based care on mortality for patients with different baseline [interleukin-6] and [interleukin-6] × [interleukin-10] profiles, whereas patients with the lowest quartiles fared better with protocol-based care (odds ratios, 0.32 [0.13-075]; p = 0.01 and 0.32 [0.14-0.73]; p = 0.01, respectively). CONCLUSIONS: In patients with septic shock, alterations in inflammation, coagulation, oxidative stress, and tissue hypoxia are common and associated with adverse outcomes but are not influenced by protocol-based resuscitation compared with usual care. However, contrary to expectation, protocol-based resuscitation appeared to be superior in patients with lower concentrations of inflammatory biomarkers. The mechanisms responsible for this effect are unclear.