The effect of drotrecogin alfa (activated) on long-term survival after severe sepsis

Abstract

Objective: To determine long-term survival for subjects with severe sepsis enrolled in the previous multiple-center trial (PROWESS) of drotrecogin alfa (activated) (DrotAA) vs. placebo.

Design: Retrospective, cross-sectional, blinded follow-up of subjects enrolled in a previous randomized, controlled trial.

Setting: One hundred sixty-four tertiary care institutions in 11 countries.

Participants: The 1,690 subjects with severe sepsis enrolled and treated with study drug in PROWESS, of whom 1,220 were alive at 28 days (the end of the original PROWESS follow-up).

Interventions: DrotAA (n = 850), 24 mug/kg/hr for 96 hrs, or placebo (n = 840).

Measurements and main results: Long-term survival data were collected. We had follow-up information on 100% of subjects at 28 days, 98% at hospital discharge, 94% at 3 months, and 93% at 1 yr. The longest follow-up was 3.6 yrs. Hospital survival was higher with DrotAA vs. placebo (70.3% vs. 65.1%, p = .03). There was no statistically significant difference in duration of survival time or in landmark survival rates in subjects who received DrotAA compared with those who received placebo (median duration of survival = 1113 days vs. 846 days for DrotAA vs. placebo, p = .10; landmark survival rates for DrotAA vs. placebo, 66.1% vs. 62.4% at 3 months [p = .11], 62.2% vs. 60.3% at 6 months [p = .44], 58.9% vs. 57.2% at 1 yr [p = .49], and 52.6% vs. 49.3% at 2(1/2) yrs [p = .21]). There was a significant interaction (p = .0008) between treatment assignment and baseline Acute Physiology and Chronic Health Evaluation (APACHE) II scores, suggesting qualitative differences in treatment effect with severity of illness. Subjects with APACHE II >/=25 had better survival time with DrotAA (median duration of survival: 450 vs. 71 days, p =.0005). Survival rates were also higher at landmark time points (DrotAA vs. placebo, 58.9% vs. 48.4% at 3 months [p = .003], 55.2% vs. 45.3% at 6 months [p = .005], 52.1% vs. 41.3% at 1 yr [p = .002], and 45.6% vs. 33.8% at 2(1/2) yrs [p = .001]). In the APACHE II <25 group there was no significant difference in survival time or survival rates at landmark time points except at 1 yr (DrotAA vs. placebo, 65.5% vs. 72.0% at 1 yr, p = .04).

Conclusions: The acute survival benefit observed in subjects with severe sepsis who received DrotAA persists to hospital discharge. The survival benefit loses statistical significance thereafter. Post hoc analysis suggests the effect of DrotAA varies by APACHE II score with improved long-term survival in subjects with APACHE II scores >/=25 but no benefit in those with lower scores.

Figure 1. Subjects

Flow of PROWESS primary analysis cohort through long-term follow-up study.

Figure 2. Survival for overall cohort

Kaplan-Meier survival curves in PROWESS primary analysis population [Peto-Wilcoxon, p=0.10; Mantel-Haenszel, p=0.15 for the comparison of drotrecogin alfa (activated) and placebo survival curves]. The dashed line is placebo and the solid line is drotrecogin alfa (activated). The curves are truncated at 2½ years where the proportion of the overall cohort still alive and in follow-up was 20%, as per recent recommendations (25). The number of subjects at risk are indicated below the x–axis.

Figure 3. Survival for pre-specified subgroups

Kaplan-Meier survival curves in PROWESS subgroups of age (<60y, N=741 vs. ≥60y, N=949), functional status [independent, N=1220 vs. dependent, N=470, as measured by pre-enrollment ADL score=0 or not (32)], APACHE II score (<25, N=873 vs. ≥25, N=817), and degree of organ dysfunction (single, N=419 vs. multiple, N=1271). The dashed line is placebo and the solid line is drotrecogin alfa (activated). All four subgroup analyses were pre-specified. There was a statistically significant qualitative difference in treatment effect by baseline APACHE II score evaluated dichotomously (< or ≥25, p=0.0008). There were no significant treatment assignment-by-subgroup interactions for age, functional status, or degree of organ dysfunction subgroups (p values ranging from 0.55–0.91 for all interaction terms introduced into the Cox models). OD, organ dysfunction.