Introduction
Regardless of achievements in critical care the overwhelming inflammatory response [
1] in patients with severe acute pancreatitis (SAP) still leads to multiple organ dysfunction (MOD) in over 60% [
2] and to hospital death in 6 to 47% [
2,
3] of cases. In MOD patients with at least two failing organs, according to the Sequential Organ Failure Assessment (SOFA) [
4], the hospital mortality rates may be as high as 50 to 91% [
2]. Furthermore, systemic activation of the coagulation system commonly occurs in over half of the critically ill patients with SAP [
2]. In rabbits an immediate activation of protein C (PC) is a specific characteristic of hemostatic activation in SAP [
5]. We have also previously shown that plasma samples in SAP drawn before MOD showed low PC levels and high endogenous activated protein C (APC) to PC ratios [
6].
Exogenous recombinant human activated protein C (drotrecogin alpha activated) has been shown to significantly reduce myeloperoxidase levels in acute experimental pancreatitis in rats [
7]. Based on an experimental model inhibition of expression of pancreatic p38 MAPK and JNK and up-regulation of ERK1/2 expression by APC treatment may theoretically protect against pancreatic injury and ameliorate the disease [
8]. In addition, APC has improved the severity of pancreatic tissue histology, and decreased the super-infection rate and serum markers of inflammation during the course of experimental acute necrotizing pancreatitis [
9].
Unlike other pharmacologic treatments APC has been shown to reduce 28-day mortality by 6.1% in a large randomized controlled trial (RCT) in severe sepsis [
10]. Thereafter, the benefits of the treatment have been questioned with regard to the excess bleeding [
11‐
13], and confirmatory RCTs are ongoing to confirm or refute the benefit in septic shock [
14]. Based on the positive RCT in severe sepsis [
10] and the above mentioned laboratory studies in SAP it has been suggested that APC may also be beneficial in SAP [
15]. However, no human RCT focusing on APC in SAP is available to date. Thus, no evidence exists to justify the use of APC in the early stages of SAP with specific concerns of risk of hemorrhage in patients with prolonged pancreatitis and pancreatic necrosis [
16].
Accordingly, we conducted a prospective randomized double-blind pilot study in 32 patients with SAP (and without severe sepsis). We hypothesized that APC would be associated with decrease in SOFA during the treatment and also aimed to assess organ-failure free days to detect potential signs of benefit of APC in SAP for further hypothesis generating purposes. We also recorded the number of serious cases of bleeding in both study groups.
Discussion
We undertook a single-center double-blind pilot RCT in 32 patients with severe acute pancreatitis, and found that patients randomized to receive APC had no more bleedings compared to the placebo group. In addition, our study revealed no beneficial effect of APC on MOD. The serum total and conjugated bilirubin levels increased in the APC group compared to the placebo group.
In our study we found no severe bleeding associated with APC treatment. The main purpose of this pilot study was to focus on this issue, because the increased risk of bleeding with APC has raised concerns [
11]. The data from randomized trials in severe sepsis patients have shown a slightly increased risk of death from serious bleeding with APC (0.7% vs. 0.2% with placebo) and a rate of 2.7% to 5.5% of serious bleeding in surgical patients during the APC infusion [
13]. Given that the patients with SAP are supposed to be prone to thrombocytopenia [
2], disseminated intravascular coagulation [
25] and intra-abdominal bleedings these concerns seem to be justified. However, based on
post hoc subgroup analysis of patients with severe sepsis and acute pancreatitis in the PROWESS trial [
10], APC has been recommended for SAP patients with severe sepsis bearing in mind the concern of retroperitoneal haemorrage [
19]. Fortunately, none of our 16 patients with APC had retroperitoneal haemorrhage or other serious bleeding in CT scans and in none of the three expired patients in this group autopsy could verify any bleeding to be cause of the fatal outcome. Of note, however, the absence of serious bleeding in both groups should be seen in the light of our study design excluding patients with decreased platelet count and the low number of patients with operative treatment (2 of 16 in the APC group).
Our study detected a significant increase in serum bilirubin levels in the APC group, both in total and conjugated values. The reason for this increase remained unsolved, because no difference in patient-related plasma hemoglobin levels, serum gamma-GT, pre-albumin, or in DST was detected. In the literature, early bilirubin increase during Days 1 to 7 has been detected previously also in severe sepsis patients treated with APC [
26]. Despite the comparable baseline hepatic SOFA both the averaged hepatic SOFA scores for Days 1 to 7 and the prevalence of hepatic organ dysfunction were increased in APC group compared to placebo in the PROWESS trial [
26]. Our results, which showed increase in total bilirubin levels with APC, are in agreement with the previous findings in a different patient group of severe sepsis.
This pilot study was underpowered to detect small differences in MOD assessed by SOFA score. Given the wide range in individual change in SOFA (-7 to +10 SOFA points), our study had a moderate chance to detect a three-point decrease in patient-related change in SOFA score during the first five days. However, in this pilot we endeavored to seek a trend for hypothesis generating purposes. Regrettably, no such positive trend in SOFA score was detected in either days free of organ dysfunction or in a change of SOFA score. On the contrary, we detected a non-significant difference in means of 2.3 in SOFA score in advantage of placebo, a trend not explained by the bilirubin changes. To sum up, although the results of pilot studies should be considered with caution, we could not detect any beneficial sign to justify a larger RCT with APC in SAP, at least with similar inclusion criteria, in alcohol-induced acute pancreatitis (> 90%), and with a similar timing, dose and duration of APC treatment.
No previous RCT in humans evaluating the safety and efficacy of APC in SAP has been conducted. Only a few cases of acute pancreatitis accompanied by severe sepsis have been treated with APC [
10,
27]. Previous animal models of SAP have suggested decreased inflammation, up-regulation of expressions of endothelial cell protein C receptor and thrombomodulin [
28], and improved survival with APC in rats [
7], although reports have been contradictory [
29]. Other proposed mechanisms for possible benefits in SAP have been the regulation of mitogen-activated kinases [
8], a decrease in leukocyte-endothelial interaction [
30], improved intestinal microcirculation [
31], and a decrease in infection of mesenteric lymph nodes [
9]. In addition, APC has been shown to modulate the systematic inflammatory response by inhibition of tumor necrosis factor and nuclear factor-kappa-B (NF-κB) release [
32], among other proposed beneficial mechanisms in severe inflammatory disorders [
33]. Given that previous studies have detected lower levels of APC and AT III in non-survivors [
34] and low endogenous PC levels and high APC:PC ratios are found in MOD patients in human SAP [
6], we previously agreed to a biologically plausible rationale for the use of APC in SAP. In keeping with this, APC has been assessed as a potential treatment option for SAP [
15,
16]. Accordingly, we undertook a randomized pilot trial but found no signs of benefits of APC in SAP.
Our study has several strengths. The study was a double-blinded trial, the concomitant treatment did not change during the study period, the disease severity of SAP in terms of SOFA and APACHE II scores was comparable to our previous observational study [
2], and patients not included in the study were registered according to the CONSORT guidelines [
35]. In addition, the placebo group received the best available standard treatment. Our study thus avoided the potential bias caused by under-treatment of the placebo group [
36] and the mortality in the placebo group was in the range of best available practice (< 20%). We also used the same dose of APC as in the two first large trials in severe sepsis [
10,
18].
Our study also has some limitations. First, the groups were not totally comparable after randomization. The patients in the APC group had a higher serum amylase level (although not reflecting the severity of SAP), and those three patients who died of MOD already had high SOFA scores (10 to 14) before the study drug infusion. Extrapolated from the severe sepsis trials we reasoned that the most severely ill SAP patients might also gain the most possible benefit and, therefore, did not exclude patients with severe MOD at baseline. It could be argued that some of our patients were too sick or too early or late in the course of their disease to have any advantage of APC. Compared to a recent study of 55 patients with acute pancreatitis (with a 30-day mortality of 27%) [
37], the patients in this study had higher baseline SOFA (6.4 vs. 5.8) and APACHE II (15.8 vs. 12.2) scores. Furthermore, we used only patient-related changes in our pre-defined analysis to overcome the problem with slight trends to differences at baseline. In addition, SAP patients with high SOFA scores have the possibility to survive [
2] as also shown in the placebo group (SOFA range up to 16) in the present study. However, although the median of maximum SOFA scores was 10 in both treatment groups indicating MOD, due to a limited sample size we cannot definitely rule out the benefit of APC in SAP patients with even higher baseline SOFA scores and more severe MOD. Second, the choice of a similar timing, dose and duration of APC treatment as in severe sepsis trials may be criticized. It is possible that patients with SAP would need a higher dose due to an overwhelming inflammatory response. For a pilot study without any previous human studies, we found it safer to use the dose previously used in RCTs of severe sepsis. Third, the outcome in the placebo group was better than expected which may have caused bias to assessment of the SOFA scores. Fourth, due to sample size we can only conclude that we could exclude a difference in incidence of serious bleeding exceeding 17%. A larger sample size is needed to exclude possible smaller differences in the safety of APC in SAP. Fifth, an alternative study approach could have been to include only a subgroup of SAP patients with low PC and/or antithrombin levels at baseline. Finally, we evaluated the change in SOFA in five days [
20], which may be too short a time to detect the difference in MOD. However, no long-term outcome measure showed a benefit either.
Competing interests
This study was supported by EVO-grant TYH 5226 from Helsinki University Hospital. Eli Lilly in part provided the study drug for this investigator-initiated study, but had no influence on the study design, data analysis or report. The investigators take full responsibility of the integrity and content of this paper.
Authors' contributions
VP, LK and MLK executed the study and drafted the manuscript. VP, MLK, AL, MT, PP, HR and EK participated in the original design and coordination of the study, and in writing the original protocol. VP, LK and MLK analyzed the data. VP, LK, MLK, AL, MT, PP, HR and EK assisted in drafting the manuscript. AM analyzed the CT scans. All authors read and approved the final manuscript.