Activated protein C in septic shock: a propensity-matched analysis

In the United States alone, approximately 750,000 cases of sepsis occur each year, of which at least 225,000 are fatal. One study evaluating the epidemiology of sepsis between 1979 and 2000 showed an 8.7% increase in the annual incidence of sepsis. The cost of management of a septic patient is approximately $50,000 amounting to annual costs estimated at $17 billion. Sepsis is the second leading cause of death in noncoronary Intensive Care Units (ICUs), and the 10^th leading cause of death overall. Organ failure occurs in 33.6% of patients with sepsis. Severe sepsis carries an estimated 30 to 50% mortality. Seventy percent of patients with three or more organ failures die. Those who survive severe sepsis have a lower quality of life compared to the age- and gender-adjusted general population, as much as 1.5 years later [1‐7].

Sepsis is a complex syndrome that remains incompletely understood. It has proven very difficult to tailor new therapies for severe sepsis. One such potential therapy is Recombinant Human Activated Protein C (rhAPC). The use of rhAPC for the treatment of severe sepsis remains controversial despite multiple reported trials [8‐13]. The Prospective Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial reported a 6% absolute reduction in mortality together with a 1.5% absolute increase in the risk of serious bleeding in patients receiving rhAPC compared to those receiving placebo [8]. In a subgroup analysis of the PROWESS trial, patients with higher risk of death had a larger drop in hospital mortality compared to patients with a lower risk of death. In patients with Acute Physiologic and Chronic Health Evaluation II (APACHE II) score >25, hospital mortality dropped from 47.9% to 36.7% (P = 0.002) with rhAPC treatment [9]. In patients with three, four, or five organ failures, hospital mortality dropped from 37.0% to 30.5%, 50.1% to 44.9%, and 63.3% to 32.5%, respectively [9]. In contrast, a subsequent trial suggested that patients with lower mortality risk did not benefit [10]. Widespread clinical use of rhAPC has likely also been limited by its cost estimated at $7,500 to $9,000 per course of therapy [14‐16].

Two large multicenter trials evaluating the effect of rhAPC in septic shock patients are currently underway [17, 18]. We examined the association between rhAPC and outcomes in septic shock patients in our institution, with the hypothesis that patients with septic shock who were treated with rhAPC had an improved in-hospital mortality compared to patients with septic shock with similar acuity who did not receive rhAPC.

A total of 563 septic shock patients were included (110 received rhAPC and 453 did not). Treated and untreated groups were matched in patient characteristics, comorbidities, physiologic variables, organ failures, and APACHE II scores (Table 2). Mean APACHE II scores were 24.5 for the matched treated and 23.9 for the matched untreated group (P = 0.54). In a 1:1 propensity-matched analysis, receipt of rhAPC was associated with reduced in-hospital mortality (35.2% vs. 53.8%, P = 0.005), similar mean days on vasopressors (2 vs. 2, P = 0.90), similar mean days on mechanical ventilation (9 vs. 8.7, P = 0.80), similar mean length of ICU stay in days (11.0 vs. 11.3, P = 0.9), and similar mean length of hospital stay in days (19.5 vs 27, P = 0.11). No patients in either group had intracranial bleeding. Differences in major transfusions (3.7% vs 9.3%, P = 0.09), minor transfusions (35.2% vs 41.7%, P = 0.27), any transfusion (minor and/or major, 38% vs 46.3%, P = 0.16) and gastrointestinal bleeding (2.8% vs 0%, P = 1.00) were not statistically significant (Table 3).

In this cohort propensity-matched analysis study, we found that patients with septic shock who were treated with rhAPC had a reduced in-hospital mortality compared with patients with septic shock with similar acuity who were not treated with rhAPC, with a relative risk reduction of 34.6%, and absolute risk reduction of 18.6%. This translates into a number needed to treat (NNT) to prevent a death of 5.4. We also found that the rate of bleeding in our cohort was similar to other published studies. In the major clinical trials, over the 28-day study period, serious bleeding events were observed in 3.5 to 6.5% of patients receiving rhAPC as compared with 2.0 to 5.0% of patients receiving the placebo [23]. Rates of intracranial hemorrhage were 0 to 1.5% with rhAPC as compared with 0 to 0.7% with the placebo.

The first large randomized trial studied a dose of 24 μg per kilogram of body weight per hour in the phase 3 PROWESS study [8]. This placebo-controlled, randomized, double-blind, multicenter trial included 1,690 patients. Treatment with rhAPC within 24 hours after diagnosis was associated with a mortality rate of 24.7% at 28 days versus 30.8% with placebo (P = 0.005). The overall incidence of at least one bleeding event was 24.9% in the treatment group and 17.7% in the placebo group (P = 0.001) [24]. The incidence of serious bleeding in the PROWESS study was also higher in the rhAPC group than in the placebo group (3.5% vs. 2.0%, P = 0.06). In subgroup analyses, most of the benefit of rhAPC treatment was seen in patients at increased risk for death, including those with APACHE II scores of 25 or greater. A follow-up study, the Administration of Drotrecogin Alfa (Activated) in Early Stage Severe Sepsis (ADDRESS) trial, evaluated the role of rhAPC in patients with severe sepsis, associated with either single-organ failure or an APACHE II score below 25 [10]. The study was stopped, after enrolling 2,640 patients, because there was no indication of a positive effect. The 28-day rate of death from any cause was 18.5% in the rhAPC group and 17% in the placebo group (P = 0.34). The in-hospital mortality rate was 20.6% in the rhAPC group and 20.5% in the placebo group (P = 0.98). Serious bleeding occurred in 2.4% of patients receiving rhAPC and in 1.2% of patients receiving the placebo (P = 0.02) during the drug-infusion period. No significant difference was observed in the rate of intracranial hemorrhage (0.3% with rhAPC and 0.2% with the placebo) [10].

The efficacy of rhAPC in severe sepsis remains a matter of dispute. Some experts even criticized the Food and Drug Administration's decision to approve rhAPC [24‐26]. The main criticisms were aimed at intercurrent changes made in the PROWESS protocol, among them the exclusion of participants who were thought to be likely to die within 28 days because of the severity of the underlying disease and the introduction of new cell lines for growing the rhAPC protein. The use of a subgroup analysis of data from a major trial to identify suitable candidates for treatment was also heavily criticized. Mackenzie, Carlet and others suggested that a general recommendation for the clinical use of rhAPC was premature or not justified [27, 28]. On the other hand, Surviving Sepsis Guidelines suggested the use of rhAPC for severe septic patients who have multiorgan failure or a high risk of death [21]. As a result, two large multicenter, randomized, double-blind trials investigating the use of rhAPC in septic shock are currently underway [17, 18].

The patients in our trial had mortality rates higher than those observed in the PROWESS trial, possibly because we only studied septic shock patients that might be sicker than the patient population in PROWESS. In our study, we had fewer patients with single organ failure, a similar number of patients with two or three organ failures, but more patients with four or five organ failures than PROWESS. In our study, the more organ failures, the higher was the mortality in both groups. Across all organ failure subgroups (one, two, three, four, or five organ failure subgroups), in-hospital mortality was persistently higher in the group that did not receive rhAPC than the group that received rhAPC (data not shown). The higher relative risk reduction in mortality in our study (33.9% vs 13% in PROWESS) is likely due to the fact that rhAPC is most beneficial in the sickest septic shock patients, with multiple organ failures. The bleeding rate in our study was smaller than that observed in PROWESS, but similar to that observed in other major clinical trials [23].

In our study, there was no difference between the matched treated and matched untreated groups in days on vasopressors, days on mechanical ventilation, length of ICU stay in days, and length of hospital stay in days (Table 3). These findings are not surprising and are similar to those found in PROWESS [9]. This adds to the evidence that survivors of treated versus untreated groups look fairly similar, but more patients survived in the treated group. This also suggests that rhAPC improves hospital survival without the consumption of additional resources.

Our study is limited in that it is a retrospective analysis and that the number of patients is relatively small. However, this propensity analysis provides an estimate of the benefit of rhAPC in a natural hospital setting. That makes it different from an estimate made from a randomized controlled trial (RCT), where conditions are more controlled. An RCT tells us that there is a causal effect of a certain size. A propensity analysis of retrospective data tells us what size effect we can expect under normal hospital conditions. Also, fewer patients (110) received rhAPC than those who did not (453). This could be explained by the possibility that treatment was preferentially directed toward patients thought to have a higher likelihood of meaningful survival, a lower risk of bleeding, or contraindication to rhAPC by the managing intensivist. However, it is not explained by differences in the types of patients receiving it versus not since we used all possible recorded pre-treatment variables. In addition, the 108 untreated patients are also candidates for rhAPC therapy in the sense that they have propensities to receive rhAPC that match the propensities of those 108 patients actually receiving rhAPC. There was also no evidence of a difference in any of the bleeding variables between the two groups (Table 3). Usage of rhAPC also depended upon whether the intensivist caring for the patient was critical or supportive of rhAPC. However, usage of rhAPC has remained constant over the study period, seen graphically and by non-significant hypothesis tests (t-test, Wilcoxon test and logistic regression all have P-values near 0.5). Our primary outcome measure was in-hospital mortality; we did not have access to longer-term outcomes, and, thus, our survival analyses assumed patients who were discharged alive from the hospital were in fact alive at Day 28. However, this study is made stronger by the propensity-matched analysis. Our primary outcome was in-hospital mortality, and rhAPC began to show its effect in the ICU (Table 3). ICU mortality comes close to statistical significance with P = 0.09; however, if we make the test one-sided in the hypothesized direction, the P-value becomes 0.045. In-hospital mortality shows fuller statistical significance (which of course includes ICU mortality). A recent similar study conducted by Lindenauer et al. showed similar results [29]. In a propensity-matched sample in which all covariates achieved balance, receipt of rhAPC was associated with reduced hospital mortality (40.7% vs. 46.6%; risk ratio, 0.87; 95% confidence interval, 0.80 to 0.95). Four rhAPC-treated patients (0.25%) had hemorrhagic stroke, 107 (6.8%) had gastrointestinal bleeding, and five (0.3%) required major transfusion [29]. An important limitation of the study by Lindenauer et al. is that all the data on their patients were based on billing records, and not direct clinical measurements.

Patients with septic shock who were treated with rhAPC had a reduced in-hospital mortality compared with patients with septic shock with similar acuity who were not treated with rhAPC. In addition, time on mechanical ventilation, time on vasopressors, length of ICU stay and bleeding complications did not differ between the groups. Pending results from large multicenter studies evaluating rhAPC in septic shock patients, our analysis adds to the evidence that early initiation of rhAPC may improve survival of patients with septic shock who are at low risk of bleeding.