Author + information
- Received October 9, 2004
- Revision received February 24, 2005
- Accepted March 15, 2005
- Published online July 5, 2005.
- Christophe Adrie, MD⁎,⁎ (, )
- Mehran Monchi, MD†,
- Ivan Laurent, MD†,
- Suzan Um, BS‡,
- S. Betty Yan, PhD‡,
- Marie Thuong, MD⁎,
- Alain Cariou, MD§,
- Julien Charpentier, MD§ and
- Jean François Dhainaut, MD§
- ↵⁎Reprint requests and correspondence:
Dr. Christophe Adrie, Service de Réanimation, Hôpital Delafontaine, 2 rue du Dr Delafontaine, Saint Denis, France
Objectives We investigated coagulation abnormalities in out-of-hospital cardiac arrest (OHCA) patients, with special attention to the protein C anticoagulant pathway.
Background Successfully resuscitated cardiac arrest is followed by a systemic inflammatory response and by activation of coagulation, both of which may contribute to organ failure and neurological dysfunction.
Methods Coagulation parameters were measured in all patients admitted after successfully resuscitated OHCA.
Results At admission, 67 patients had a systemic inflammatory response with increased interleukin-6 and coagulation activity (thrombin-antithrombin complex), reduced anticoagulation (antithrombin, protein C, and protein S), activated fibrinolysis (plasmin-antiplasmin complex), and, in some cases, inhibited fibrinolysis (increased plasminogen activator inhibitor-1 with a peak on day 1). These abnormalities were more severe in patients who died within two days (50 of 67, 75%) and were most severe in patients dying from early refractory shock. Protein C and S levels were low compared to healthy volunteers and discriminated OHCA survivors from nonsurvivors. Furthermore, a subgroup of patients had a transient increase in plasma-activated protein C at admission followed by undetectable levels. This, along with an increase in soluble thrombomodulin over time, suggests secondary endothelial injury and dysfunction of the protein C anticoagulant pathway similar to that observed in severe sepsis.
Conclusions Major coagulation abnormalities were found after successful resuscitation of cardiac arrest. These abnormalities are consistent with secondary down-regulation of the thrombomodulin-endothelial protein C receptor pathway.
Sudden death from coronary heart disease before reaching the hospital occurs in about 225,000 people annually in the U.S. The overall survival rate is low, ranging from 4% to 33% depending on the efficacy of the chain of survival (1,2). In successfully resuscitated patients admitted to the intensive care unit (ICU), the prognosis remains poor, and life-threatening disturbances known as “post-resuscitation disease” may lead to multiple organ dysfunction and jeopardize neurological recovery (3). Shock often develops several hours after hospital admission. It is characterized by low cardiac output and is usually reversible within 24 h, suggesting post-resuscitation myocardial dysfunction associated with peripheral vasodilation (4). Early death by multiple organ failure is associated with low cardiac output lasting longer than 24 h; however, the hemodynamic status does not predict the neurological outcome (4). Reperfusion failure, ischemia-reperfusion injury, and cerebral injury may be responsible for an overwhelming systemic inflammatory response associated with elevated plasma cytokines, presence of circulating endotoxin, leukocyte dysregulation, and adrenal dysfunction, a picture similar to that observed in severe sepsis (5,6). Moreover, a similar systemic inflammatory response has been described in severe cardiogenic shock with multiple organ failure (7) or a need for mechanical circulatory support (8).
Cardiopulmonary resuscitation (CPR) and a return to spontaneous circulation are associated with marked activation of blood coagulation, without adequate concomitant activation of endogenous fibrinolysis (9,10). This suggests that intravascular fibrin formation and microvascular thrombosis after cardiac arrest may contribute to organ dysfunction, including neurological impairment. Consistent with this hypothesis, thrombolytic therapy after CPR improved survival in experimental models of induced cardiac arrest (11,12) and allowed the return of spontaneous circulation after failed initial CPR (13).
Inflammatory and procoagulant host responses are closely linked not only to infection, but to all inflammatory processes (14). Inflammatory cytokines activate coagulation and inhibit fibrinolysis, whereas the procoagulant thrombin stimulates multiple inflammatory pathways (14). Activated protein C is an endogenous protein that enhances fibrinolysis, limits thrombin generation, and modulates inflammation. It is converted from its inactive precursor, protein C, by thrombin coupled to thrombomodulin. On the other hand, thrombin can have multiple proinflammatory properties; it activates endothelial cells to express P-selectin, promotes neutrophil and monocyte adhesion, induces endothelial platelet-activating factor formation, and acts as a chemoattractant for polymorphonuclear neutrophils (15). Administration of activated protein C significantly reduces mortality in patients with severe sepsis, a disease of systemic inflammatory activation (16).
In this study, we investigated the inflammation and coagulation responses in patients who were successfully resuscitated after an out-of-hospital cardiac arrest (OHCA). We directed special attention to the protein C pathway.
Patients after OHCA, patients with severe sepsis, and healthy volunteers
This study was performed according to the ethical rules of our institutions (Cochin Hospital, Paris, France, and Delafontaine Hospital, Saint Denis, France), and informed consent was obtained from the next-of-kin of all patients. Cardiac arrest was defined as absence of spontaneous respiration, palpable heartbeat, and responsiveness to stimuli. Consecutive patients older than 16 years of age who were successfully resuscitated after OHCA were prospectively included in the study. Successful resuscitation was defined as recovery of blood pressure and pulse for more than 1 h, with or without a continuous catecholamine infusion. Both the Simplified Acute Physiology score (SAPS II) (17) and the Logistic Organ Dysfunction (LOD) score (18) were calculated. We recorded oral anticoagulants and/or antiplatelet agents given before or after admission, as well as prophylactic heparin therapy. Outcomes identified three groups of patients: survivors, all of whom were conscious; patients who died within four days from early refractory shock with multiple organ failure; and patients who died later from neurological dysfunction with or without a need for initial inotropic support (4–6). Patients were excluded from the study if they received thrombolytic therapy or therapeutic heparin. We also excluded patients with end-stage liver disease identified on the basis of clinical evidence or medical history. To check the validity of our biomarker assay methods, we studied two control groups, one composed of patients with severe sepsis meeting American College of Chest Physicians/Society of Critical Care Medicine consensus criteria (19) and studied within two days of ICU admission, and the other composed of healthy volunteers.
Citrated blood samples (4 ml) were collected and immediately centrifuged at 1,500 gfor 10 min. The plasma was stored at −80°C until analysis. Blood collection was performed at ICU admission (day 0) and daily for the next 7 days (days 1 to 7). In a subset of 16 patients, additional blood samples were collected on days 0, 1, and 2 for measuring plasma endogenous activated protein C. Blood samples were drawn into citrated tubes containing the reversible serine protease inhibitor benzamidine, which blocks the irreversible inhibition of activated protein C by endogenous plasma protease inhibitors (20).
The following assays were performed using an STA Compact coagulation analyzer (Diagnostica Stago, Asnières, France) with Diagnostica Stago test kits. Activated partial thromboplastin time (STA-PTT A), prothrombin time (STA-Neoplastine Cl plus), protein C (Staclot Protein C), and free protein S (Staclot Protein S) were measured using coagulation-based activity assays. D-dimer levels were measured immunoturbidimetrically with the STA Liatest D-DI latex immunoassay. Antithrombin (AT) (Stachrom ATIII) and plasminogen activator inhibitor (PAI)-1 (Stachrom PAI) levels were quantitated using chromogenic activity assays. Soluble thrombomodulin (sTM) (Asserachrom Thrombomodulin, Diagnostica Stago), thrombin-antithrombin complex (TAT) (Enzygnost TAT micro, Dade Behring, Marburg, Germany), plasmin-antiplasmin complex (PAP) (PAP micro ELISA, DRG International Inc., Mountainside, New Jersey), and interleukin (IL)-6 (Quantikine Human IL-6 kit, R & D Systems, Minneapolis, Minnesota) antigen levels were measured by enzyme immunoassays. Plasma-activated protein C levels were measured using immunocapture-amidolytic assays, as previously described by Gruber and Griffin (20). Normal ranges and abbreviations for each of the biomarkers are reported in the Appendix.
Continuous data were expressed as medians and interquartile ranges. Undetectable levels (levels below the detection threshold) were assigned the value 0. Because of the high mortality rate within the first few days in the ICU, the statistical analysis of circulating markers was confined to the first two days. Differences between groups were evaluated using Mann-Whitney Utests or chi-square tests. Relationships between two continuous variables were analyzed using Spearman’s rank correlation tests. Repeated measures analysis of variance was used to compare the time-course of coagulation markers between survivors and nonsurvivors. For this analysis, levels of coagulation parameters were normalized by natural log transformation. After natural log transformation, the Shapiro-Wilk W test was used to test for normality, with p values >0.10 indicating a normal distribution. For nonsurvivors, missing values after death were replaced by the last available value. A p value <0.05 was considered statistically significant. Statistical tests were run using Stata 7.0 software (Stata Corp., College Station, Texas).
We included 67 patients (49 men, 73%) admitted after successful resuscitation of cardiac arrest, either to the Delafontaine Hospital ICU (n = 35) or to the Cochin Hospital ICU (n = 32). Among them, 43 were included in a previous study from February 1999 to April 2000 (5); the 24 additional patients were included from December 2001 to July 2002. Median SAPS II score was 66 (56 to 79), median LOD score was 9 (7 to 12), and median Glasgow Coma Scale score was 3 (3 of 3) at admission (Table 1).Cardiac ischemia was the leading cause of cardiac arrest (n = 39, 58%). All patients received endotracheal mechanical ventilation. Median time from cardiac arrest to first blood sampling at ICU admission was 2 h 30 min (interquartile range, 1 h 48 min to 3 h 36 min). Overt infection developed in 23 (34%) patients, 10 survivors (59%), and 13 (26%) nonsurvivors (p = 0.03) with a median time to occurrence of two days (one to three days). All infections but one consisted of pneumonia (22 of 23, 96%), the main cause being aspiration during the episode of cardiac arrest. Of the 17 survivors, all were conscious at ICU discharge, and only one had severe cerebral damage. Of the 50 patients who died, 8 died within one day, 17 within two days, 22 within three days, and 30 within four days. The remaining 20 patients died on day 5 or later. Of the 50 deaths, 20 (40%) were due to early refractory shock (Table 2).No patients received hypothermia treatment. Of the 10 patients on oral anticoagulant therapy before admission, only one survived (p = 0.43). Prior to admission, 8 patients were on chronic antiplatelet treatment; 29 additional patients received antiplatelet therapy at admission. Subcutaneous low-dose heparin was given to 48 patients during the ICU stay.
The positive control group was composed of 12 patients (9 men, 75%) with septic shock, a SAPS II score of 61 (48 to 74), a LOD score of 8 (6.5 to 9), a Glasgow Coma Scale score of 13 (12 to 15), and a need for vasopressor therapy. The lungs (n = 7, 58%) and abdomen (n = 3, 25%) were the main sites of infection. Of these 12 patients, 7 (58%) died. The negative control group comprised 10 healthy volunteers (Table 3).
As described previously (5), IL-6 and lactate levels were high in the OHCA group at admission, and high levels were associated with death (receiver-operating characteristic test: 0.76 and 0.81, respectively). Both IL-6 and lactate levels were correlated to CPR duration (r = 0.46, p = 0.0002; and r = 0.45, p = 0.0005, respectively), but not to the time from collapse to basic life support. The large volume of fluid administered explains the decrease in hematocrit from 42% (39% to 46%) on admission to 36% (35% to 40%) on day 2 (p < 10−4), and 39 patients (58%) required vasopressor therapy within the first two days. However, neither the need for vasopressors nor the magnitude of the hematocrit reduction was associated with mortality.
Coagulation biomarkers at admission
At study entry, all OHCA patients had an inflammatory response and coagulation abnormalities including increased coagulation activation, reduced anticoagulation, and activated fibrinolysis (Table 1). Baseline D-dimer levels were high in all OHCA patients. Coagulation activation (elevated TAT, 100% of patients; decreased protein C, protein S, and AT) was associated with activated fibrinolysis (elevated PAP, 100% of OHCA patients). All OHCA patients had evidence of systemic inflammation (elevated IL-6, 100% of patients). However, at admission, 37% of OHCA patients had reduced fibrinolysis due to PAI-1 elevation, and 28% had high sTM levels indicating endothelial injury, in contrast to results in the patients with septic shock. Furthermore, activation of coagulation and fibrinolysis and reduced anticoagulation at admission were more pronounced in nonsurvivors (Table 1), particularly those dying from early refractory shock (Table 2). These features are similar to those observed in patients with septic shock (Table 3).
Course of coagulation abnormalities over the first two days in the ICU
After natural log transformation of the coagulation parameters, the p values of the Shapiro-Wilk W test were all above 0.10 (ranging from 0.43 to 0.89), indicating a normal distribution. Biomarker levels in OHCA patients over the first two days are shown in Figure 1and Table 3. In the nonsurvivors, coagulation abnormalities were more severe and less likely to resolve within two days. Nonsurvivors also showed more severe acquired deficiencies in anticoagulant factors at admission and less recovery over the first two study days, as compared to survivors. At all time points within the two-day period, protein C was less than the lower limit of normal in 34 (51%) patients overall and 25 (44%) patients who were not taking oral anticoagulation. Survivors had significantly lower levels of TAT over the study period than did the nonsurvivors. They also had significantly less fibrinolysis activation (lower PAP levels; Fig. 1) and less fibrinolysis inhibition (less PAI-1 elevation, data not shown) over time than did nonsurvivors. The ratio between fibrinolysis activation and coagulation activation (PAP/TAT) was higher in survivors than nonsurvivors, suggesting inadequate fibrinolysis in patients who died (p = 0.01; Fig. 1). Additionally, nonsurvivors had higher levels of sTM (marker for endothelial injury) and IL-6 (marker for inflammation) than did survivors throughout the study period (data not shown). For comparison, values obtained from healthy volunteers and patients with severe sepsis are summarized in Table 3.
Of the 16 patients who had assays of plasma endogenous activated protein C, 14 (87%) had detectable levels (>1 ng/ml) at admission. Levels were detectable in 3 of the 12 (25%) patients alive on day 1 and in 1 of the 9 patients alive on day 2 (Fig. 2).Activated protein C levels were detectable in 2 of 6 patients (33%) with septic shock and in none of the 10 healthy volunteers (data not shown). Activated protein C levels were closely correlated with TAT levels (r = 0.68, p = 0.008) and with the LOD score (r = 0.61, p = 0.01).
Evidence of systemic coagulopathy was consistently present in patients successfully resuscitated after cardiac arrest: coagulation was activated, anticoagulant factors diminished, and fibrinolysis increased. This was associated with a transient initial increase of endogenous production of activated protein C likely followed by a secondary endothelial injury and dysfunctional protein C anticoagulant pathway.
Cytokines such as tumor necrosis factor-alpha, IL-1, and IL-6 are released into the circulation, up-regulating the expression of tissue factor (21), a major initiator of intravascular coagulation, on monocytes and endothelial cells. The thrombin generated in this process not only converts fibrinogen to fibrin clot, but also has potent proinflammatory effects. We found IL-6 elevation, which can be related to several mechanisms such as whole-body ischemia-reperfusion syndrome, bacterial or endotoxin translocation from the gut, and pulmonary aspiration (6). This systemic post-resuscitation response constitutes a sepsis-like syndrome in which systemic inflammation (6) and coagulation activation (thrombin generation) may reinforce each other, resulting in fatal multiorgan failure (14,16).
Consistent with previous results described by Böttiger et al. (9), we found marked activation of coagulation and fibrinolysis in patients after CPR. However, D-dimers were only slightly increased in the study by Böttiger et al. (9), suggesting that activation of blood coagulation was not adequately balanced by activation of endogenous fibrinolysis, whereas D-dimers were dramatically elevated in our study. This discrepancy can be ascribed in part to differences in patient populations: most of the patients studied by Böttiger et al. (9) did not recover spontaneous circulation, and none survived beyond 48 h, whereas we included only patients who returned to spontaneous circulation, and we had about one-fourth of our patients discharged alive and conscious from the ICU. Thus, activation of endogenous fibrinolysis may lag behind fibrin formation.
We found decreases in protein C and S levels, with the lowest levels in nonsurvivors. Hemodilution may contribute to decreased protein C and S levels. However, these levels were already low at admission and remained low over the two-day study period, whereas the hematocrit fell sharply over time (Table 3). This suggests that early hemodilution (4) did not account for the very early protein C and S consumption during and/or just after CPR. A similar early decrease in protein C levels has been reported several hours before the onset of clinical signs of severe sepsis (22).
We found a number of differences between patients with OHCA and severe sepsis regarding the profile of coagulation/fibrinolysis activation: protein C and S depletion was less marked in the OHCA patients than in the patients with severe sepsis (Table 3). Severe sepsis often begins insidiously, whereas the acute insult associated with cardiac arrest may lead to a rapid spike in these biomarkers followed by more moderate but sustained abnormalities. Another explanation may be related to the higher early mortality rate in OHCA patients (especially those with early refractory shock); patients with the most abnormal values died early (within the first 48 h) giving the “artifactual” impression of the median levels being rapidly “normalized” in the surviving patients.
Plasma concentrations of endogenous activated protein C in patients and normal volunteers were usually undetectable. Only a very short-lived and early increase was observed in OHCA patients at admission. The generation of activated protein C in plasma in healthy humans is dependent on circulating concentrations of both protein C and thrombin (23). However, in patients with severe sepsis, conversion of endogenous protein C to activated protein C may be impaired because of endothelial dysfunction with down-regulation of thrombomodulin and endothelial protein C receptor (23). Cardiac arrest is an acute event occurring at a well-defined time, which allows detection of early changes in systemic biomarkers. We speculate that early endothelial stimulation with thrombin generation is responsible for the tremendous increase in protein C conversion to activated protein C, and that this phase is rapidly followed by endothelial dysfunction characterized by an inability to generate an adequate amount of activated protein C. This is consistent with the increase in sTM levels over time (Table 3), as sTM is a marker for endothelial injury. Moreover, in an experimental baboon sepsis model, Taylor et al. (24) observed a similar profile, with a transient increase in endogenous activated protein C followed by a decrease. An alternative hypothesis is that, among the five serine protease inhibitors (namely, protein C inhibitor, alpha1-antitrypsin, alpha2-antiplasmin, alpha2-macroglobulin, and PAI-1) known to inhibit activated protein C, the last four are acute-phase reactants and increase during systemic inflammatory responses, which may diminish free levels of activated protein C (25,26). Support for this hypothesis comes from the significant increase in PAI-1 with a transient peak on day 1 seen in our patients (Table 3).
At admission, activated protein C levels were closely correlated with thrombin generation, as reflected by TAT complex levels and organ dysfunction assessed by the LOD score in the subset of patients whose activated protein C levels were assayed. This suggests that increased generation of thrombin and activated protein C reflect the progression of organ dysfunction and disease severity. The rise in activated protein C may reflect a natural compensatory mechanism that dampens the coagulation activation and inflammatory response. Disseminated intravascular coagulation is characterized by thrombin generation and fibrin deposition, resulting in widespread microvascular thrombosis responsible for multiorgan failure, including neurological dysfunction. Interestingly, activated protein C has been shown to minimize ischemia/reperfusion injury to the damaged spinal cord, and to the brain in stroke models (27–29). This protective effect may be related to the anticoagulant, anti-inflammatory, and antiapoptotic properties of activated protein C (29). This should encourage clinical studies to determine whether administration of thrombolytic or anticoagulant agents improves neurological outcome in OHCA patients (13). Furthermore, early hemodynamic instability is independent from subsequent neurological events (4), suggesting that aggressive initial treatment may be in order until a reliable neurological assessment can be performed.
Our study has several limitations. It was completed before the introduction of routine therapeutic hypothermia, which may interfere with coagulation (30). However, this treatment is currently recommended only in a very small subgroup of patients (<10%) (30). Patients with cardiac arrest constitute a heterogeneous population with a variety of underlying diseases (dominated by causes of cardiac ischemia) and often multiple treatments that may interact with platelet aggregation or coagulation (as their intended effect or as a side effect) before and/or during the ICU stay. Moreover, lifestyle, diet, and congenital deficiencies may affect coagulation pathways. Nevertheless, we found no significant influence of drugs known to alter coagulation systems, and we believe that our patients are representative of the overall population of successfully resuscitated OHCA patients.
In conclusion, systemic coagulation abnormalities were consistently found in patients after recovery of spontaneous circulation after cardiac arrest, and the profile of these abnormalities was similar to that in patients with severe sepsis. The protein C depletion, transient increase in endogenous activated protein C, gradual elevation in sTM, and systemic inflammatory response suggest that the protein C anticoagulant pathway may contribute to the high mortality seen after CPR, as observed in patients with severe sepsis.
The authors are indebted to A. Wolfe, MD, for helping with this manuscript.
Appendix Abbreviations List and Normal Ranges
|IL-6 and Coagulation Parameters||Normal Ranges|
|Interleukin-6 (IL-6)||Less than the detection threshold (0.7 pg/ml)|
|Activated protein C||Less than the detection threshold (1 ng/ml)|
|Protein C (clot-based activity assay)||64%–161%|
|Protein S (clot-based activity assay)||55%–152%|
|Soluble thrombomodulin (sTM)||18–53 ng/ml|
|Antithrombin (AT) (chromogenic-based activity assay)||83%–130%|
|Prothrombin time (PT)||12.5–14.9 s|
|Activated partial thromboplastin time (APTT)||26.9–35.1 s|
|Thrombin-antithrombin complex (TAT)||0–3.0 ng/ml|
|Plasmin-antiplasmin complex (PAP)||42–306 μg/ml|
|Plasminogen activator inhibitor-1 (PAI-1)||4.0–37.8 AU/ml|
Suzan Um and Dr. Yan are employees of Eli Lilly and Company, and Dr. Dhainaut has served as a consultant for Eli Lilly and Company. However, this study was a research collaboration effort and was not funded by Lilly.
- Abbreviations and Acronyms
- cardiopulmonary resuscitation
- intensive care unit
- Logistic Organ Dysfunction score
- out-of-hospital cardiac arrest
- plasminogen activator inhibitor
- plasmin-antiplasmin complex
- SAPS II
- Simplified Acute Physiology score
- soluble thrombomodulin
- thrombin-antithrombin complex
- Received October 9, 2004.
- Revision received February 24, 2005.
- Accepted March 15, 2005.
- American College of Cardiology Foundation
- Laurent I.,
- Monchi M.,
- Chiche J.D.,
- et al.
- Adrie C.,
- Adib-Conquy M.,
- Laurent I.,
- et al.
- Hasper D.,
- Hummel M.,
- Kleber F.X.,
- Reindl I.,
- Volk H.D.
- Böttiger B.W.,
- Motsch J.,
- Böhrer H.,
- et al.
- Gando S.,
- Kameue T.,
- Nanzaki S.,
- et al.
- Bone R.C.,
- Balk R.A.,
- Cerra F.B.,
- et al.,
- The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine
- Gruber A.,
- Griffin J.H.
- Taylor F.B. Jr..,
- Wada H.,
- Kinasewitz G.
- Taoka Y.,
- Okajima K.,
- Uchiba M.,
- et al.
- Nolan J.P.,
- Morley P.T.,
- Vanden Hoeck T.L.,
- Hickey R.W.