Abstract Rationale Phenotype-stratified reanalysis of the negative PROWESS-SHOCK trial recently found heterogeneous treatment effects (HTE) of activated Protein C (APC) in septic patients, with reduced mortality in the Hyperinflammatory and increased mortality in the Hypoinflammatory phenotype1. We examined whether similar HTE was present in the earlier PROWESS trial. Next, we exclusively analysed Hyperinflammatory patients derived from a combined cohort of PROWESS and PROWESS-SHOCK to test: (1) Did treatment increase protein C activity; (2) Was higher plasma protein C associated with longer survival; and (3) Did higher plasma protein C levels mediate APC’s effect on survival. Methods PROWESS (N = 1,690) tested APC versus placebo in severe sepsis and found a treatment benefit. We used pretreatment laboratory and vital signs to phenotype patients using a previously validated Clinical Classifier Model2. We tested for HTE using a logistic model with 28-day mortality predicted by phenotypes, treatment groups, and their interaction. PROWESS and PROWESS-SHOCK’s Hyperinflammatory patients were then combined (N = 1,295), and daily differences in Protein C activity (Day 0-Day 4) between treatment groups were tested with mixed-effects regression. The association between longitudinal Protein C increases and mortality risk was then tested in a Bayesian joint model. Finally, mediation analysis using the joint model tested whether greater increases in Protein C activity mediated APC’s effect on survival. Results In PROWESS, compared to Hypoinflammatory patients (N = 933, 55%), Hyperinflammatory patients (N = 757, 45%) had higher APACHE-II scores (mean=37.7±7.5); Hypo=22.3±7.0), vasopressor use (89% vs. 41%), mechanical ventilation rates (80% vs. 72%), IL6 levels, and 28-day mortality (35% vs 25%)(all p 0.001). The interaction of treatment group and phenotype was non-significant (p = 0.28); however, we observed improved survival with APC compared to placebo with increasing Hyperinflammatory probability (Panel A).In the combined Hyperinflammatory cohort from PROWESS and PROWESS-SHOCK, APC treatment conferred a significant survival benefit (Panel B). We also observed (as shown in Panel C) that: (1) Through day 4, daily plasma protein C rose more with APC vs. placebo (p 0.0001; top); (2) Greater plasma protein C increases were associated with reduced mortality hazards (α-mean = -0.12, 95% CI: -(0.17)-(-0.07); middle); and (3) Greater protein C increases mediated APC’s effect on survival time (β mean = -0.53, 95% CI: (-0.76)-(-0.34); bottom). Conclusions We observed a survival benefit from APC among Hyperinflammatory patients in PROWESS and PROWESS-SHOCK. The survival benefit of APC in the Hyperinflammatory phenotype is partially mediated by early treatment-related increases in plasma Protein C activity. This abstract is funded by: R01HL173531, R35GM142992
Bartek et al. (Fri,) studied this question.