ARTICLE IN PRESS Journal of Critical Care (2010) xx, xxx–xxx

Troponin-I as a prognosticator of mortality in severe sepsis patients☆,☆☆,★ Jijo John MD a,⁎, D. Bradley Woodward MD b , Yanping Wang PhD b , S. Betty Yan PhD b , Diana Fisher MS b , Gary T. Kinasewitz MD a , Darell Heiselman DO b a

University of Oklahoma Health Sciences Center, Pulmonary/Critical Care Medicine, Oklahoma City, OK 73104, USA Eli Lilly and Company, Cardiovascular/Critical Care, Lilly Corporate Center, Indianapolis, IN 46285, USA

b

Keywords: Troponin; Severe sepsis; PROWESS; Drotrecogin alfa (activated)

Abstract Purpose: The purpose of this retrospective study was to evaluate cardiac troponin-I (cTnI) as a 28-day mortality prognosticator and predictor for a drotrecogin alfa (activated) (DrotAA) survival benefit in recombinant human activated Protein C Worldwide Evaluation in Severe Sepsis patients. Methods: Cardiac troponin-I was measured using the Access AccuTnI Troponin I assay (Beckman Coulter, Fullerton, CA). There were 598 patients (305 DrotAA, 293 placebo) with baseline cTnI data (cTnI negative [b0.06 ng/mL], n = 147; cTnI positive [≥0.06 ng/mL], n = 451). Results: Cardiac troponin-I–positive patients were older (mean age, 61 vs 56 years; P = .002), were sicker (mean Acute Physiology and Chronic Health Evaluation II, 26.1 vs 22.3; P b .001), had lower baseline protein C levels (mean level, 49% vs 56%; P = .017), and had higher 28-day mortality (32% vs 14%, P b .0001) than cTnI-negative patients. Elevated cTnI was an independent prognosticator of mortality (odds ratio, 2.020; 95% confidence interval, 1.153-3.541) after adjusting for other significant variables. Breslow-Day interaction test between cTnI levels and treatment was not significant (P = .65). Conclusion: This is the largest severe sepsis study reporting an association between elevated cTnI and higher mortality. Cardiac troponin-I elevation was not predictive of a survival benefit with DrotAA treatment. © 2010 Published by Elsevier Inc.

1. Introduction Biomarkers are important tools in the quest to treat patients quickly and efficiently, especially in critical illness ☆

Work was performed at Eli Lilly and Company, Cardiovascular/ Critical Care, Lilly Corporate Center, Indianapolis, IN 46285. ☆☆ Financial support was provided by Eli Lilly and Company. ★ Disclosures: Dr Kinasewitz is a speaker for Eli Lilly's lecture bureau. Drs Woodward, Wang, Yan, and Heiselman and Ms Fisher are employees and shareholders of Eli Lilly and Company. Eli Lilly and Company is the maker of drotrecogin alfa (activated). ⁎ Corresponding author. Tel.: +1 405 271 6173; fax: +1 405 271 5892. E-mail address: [email protected] (J. John). 0883-9441/$ – see front matter © 2010 Published by Elsevier Inc. doi:10.1016/j.jcrc.2009.12.001

where time could be a limiting factor. Compared with other diagnostic tools, biomarkers may provide an efficient way for clinicians to quickly formulate different diagnoses, guide therapy, and provide prognostic values [1]. Cardiac biomarkers are some of the more well-documented biomarkers currently in use, with the ability of cardiac troponin T and cardiac troponin I (cTnI) measurements to predict outcome and guide the therapeutic management of patients with acute coronary syndromes [2]. Specifically, cTnI has been shown to be an indicator of myocardial injury and is an accepted prognosticator of myocardial infarction (MI) [3,4]. Although cTnI is cardiac-specific, its release seems not to be limited to cardiac-related events, but is also detectable in other critical

ARTICLE IN PRESS 2 clinical conditions, such as trauma, pulmonary embolism, and severe sepsis. Recent studies have begun examining the role of cTnI in sepsis and septic shock. In a prospective case-control study by Ammann et al [5], of the 20 patients with sepsis, septic shock, or systemic inflammatory response syndrome and without underlying acute coronary syndromes, 85% of the patients had elevated cTnI, whereas no patient in the control group of the study had elevated cTnI (P b .0001). In another prospective study, Arlati et al [6] showed that cTnI was elevated in 11 of 19 sepsis patients. Other sepsis studies, with similarly small numbers of patients (b105), showed that sepsis patients with elevated troponin levels had higher Acute Physiology Age and Chronic Health Evaluation (APACHE) II scores and higher mortality [7-9]. Severe sepsis is the leading cause of death in the noncoronary intensive care unit (ICU) and the 10th leading cause of death overall [10-12]. Drotrecogin alfa (activated) (DrotAA) is approved by the United States Food and Drug Administration for the reduction of mortality in adult patients with severe sepsis and high risk of death. Approval for DrotAA was based on the pivotal phase 3 study, recombinant human activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS), that enrolled 1690 patients (850 DrotAA, 840 placebo) [13]. This current retrospective study (abstract recently published by Heiselman et al [14]) used a subset of the PROWESS severe sepsis patients (n = 598) and evaluated cTnI as a prognosticator of 28-day mortality as well as a predictor for a survival benefit with DrotAA treatment.

2. Methods 2.1. Patient population From July 1998 through June 2000, 1690 severe sepsis patients were enrolled in PROWESS, a randomized, doubleblind, placebo-controlled phase 3 trial for evaluating DrotAA treatment. The PROWESS trial was conducted at 164 centers in 11 countries [13]. The institutional review board at each center approved the protocol, written informed consent was obtained from all participants or their authorized representatives, and the study was conducted according to the principles of the Declaration of Helsinki. Patients must have met all inclusion criteria, including the presence of sepsis-induced organ dysfunction of 24 hours or less. Baseline laboratory specimens were collected after the patient met entry criteria and before study drug administration. For the current study, vasopressor status and serum for troponin level were both obtained at study entry.

2.2. Troponin-I assay Because the measurement of cTnI levels was not prospectively defined in the PROWESS study, baseline

J. John et al. serum samples stored at −70°C from the PROWESS trial were anonymized before cTnI measurement. Cardiac troponin-I was measured using the Access AccuTnI Troponin I assay (Beckman Coulter, Fullerton, CA). Per the manufacturer, the cut point level for upper limit of normal (with an assay coefficient of variation b10%) was 0.06 ng/ mL. For this study, troponin negative (cTnI−) and troponin positive (cTnI+) were defined as levels less than 0.06 ng/mL and greater than or equal to 0.06 ng/mL, respectively.

2.3. Statistical analyses Categorical variables are presented as incidence rates, and comparisons between groups were based on a χ2 test. The Breslow-Day test was used to test homogeneity of treatment effect across subgroups. Continuous variables are presented as mean ± standard deviation (SD) or median, and comparisons between groups were performed with the use of analysis of variance of ranked data. Two-sided 5% significance levels and 95% confidence intervals were used for all analyses. No adjustments were made to P values for multiple comparisons. Computations were performed using version 9 of SAS software (SAS, Cary, NC). A stepwise logistic regression procedure was used to select significant prognosticators of mortality. A large number of baseline variables were considered: age (b65 vs ≥65), sex, race (white vs other), patient location before hospitalization, presumed site of infection, comorbidity (based on APACHE II chronic health points), functional dependency status (based on activities of daily living score), Gram stain of organism cultured, recent surgery history, protein C level, APACHE II score, number of organ failures, renal sequential organ failure assessment (SOFA) score, respiratory SOFA score, cardiovascular SOFA score, hematology SOFA score, shock status, ventilation status, prothrombin time, activated partial thromboplastin time, interleukin-6 (IL-6) level, D-dimer level, antithrombin level, creatinine clearance, and treatment (placebo vs DrotAA). The cTnI level was forced into the final model to test whether it is an independent prognosticator after adjusting for all the significant variables.

3. Results 3.1. Troponin-I as prognosticator of mortality Of the 598 patients, 147 patients (71 placebo, 76 DrotAA) were cTnI− and 451 (222 placebo, 229 DrotAA) were cTnI+ at baseline. The median baseline level of cTnI in the 598 patients was 0.15 ng/mL (range, 0.01-353 ng/mL). There were significant differences in baseline characteristics between the cTnI+ and cTnI− groups (Table 1). The cTnI+ group was slightly older and had worse baseline disease severity (more patients had APACHE II scores

ARTICLE IN PRESS Troponin-I as a prognosticator of mortality in severe sepsis Table 1

3

Mortality outcome and baseline characteristics by troponin-I class for total study population

Variable 28-d mortality, n (%) cTnI levels, ng/mL Mean ± SD Median (25%, 75%) Age (mean ± SD) Sex Male APACHE II score (mean ± SD) APACHE II score ≥25, n (%) Protein C levels Mean ± SD Median (25%, 75%) Protein C deficiency status b Yes No Unknown Severe protein C deficiency status b Yes No Unknown Recent surgery (within last 30 d) No Yes Unknown Vasopressor status c Yes Ventilation status c Yes No. of organ failures 1 2 3 ≥4 ARDS status Yes SOFA score (mean ± SD) Cardiovascular Hematology Hepatic Renal Respiratory Central laboratory platelet count (×103/mm3), n Median (25%, 75%) Prothrombin time (s), n Median (25%, 75%) IL-6 (ng/mL), n Median (25%, 75%) D-Dimer (μg/mL), n Median (25%, 75%)

Troponin-I+ (n = 451) 145 (32.2%)

Troponin-I− (n = 147) 20 (13.6%)

2.74 ± 18.02 0.27 (0.11, 0.84) 61.5 ± 17.0

0.03 ± 0.01 0.04 (0.03, 0.04) 56.3 ± 18.4

252 (55.9%) 26.1 ± 7.5 259 (57.4%)

81 (55.1%) 22.3 ± 6.8 58 (39.5%)

0.49 ± 0.26 0.45 (0.30-0.64)

0.56 ± 0.30 0.54 (0.34-0.73)

P value a b.0001 b.0001

.002 .870 b.001 b.001 .017 .053

384 (85.1%) 47 (10.4%) 20 (4.4%)

119 (81.0%) 25 (17.0%) 3 (2.0%)

180 (39.9%) 251 (55.7%) 20 (4.4%)

47 (32.0%) 97 (66.0%) 3 (2.0%)

319 (70.7%) 129 (28.6%) 3 (0.7%)

105 (71.4%) 42 (28.6%)

288 (63.9%)

73 (49.7%)

352 (78.0%)

112 (76.2%)

111 (24.6%) 137 (30.4%) 110 (24.4%) 93 (20.6%)

44 (29.9%) 59 (40.1%) 27 (18.4%) 17 (11.6%)

85 (18.8%)

29 (19.7%)

.063

.611

.002 .639 .025

.813

2.70 ± 1.50 0.80 ± 0.99 0.61 ± 0.86 1.30 ± 1.17 2.72 ± 1.06 375 169 (112, 232) 429 18.8 (16.4, 22.6) 446 590 (180, 3579) 426 4.44 (2.38, 8.92)

2.21 ± 1.50 0.40 ± 0.76 0.34 ± 0.67 0.86 ± 1.04 2.67 ± 1.09 127 210 (158, 311) 142 18.1 (15.9, 20.7) 147 411 (94.1, 1489) 141 3.27 (1.53, 5.66)

b.001 b.001 b.001 b.001 .765 b.001 .006 .004 b.001

ARDS indicates acute respiratory distress syndrome. a P values are based on the χ2 test for categorical variables and on the analysis of variance of ranked data for continuous variables. b Protein C deficiency was defined in the PROWESS study [13] as less than 81%, and severe protein C deficiency was defined as less than or equal to 40%. c Baseline vasopressor status and ventilation status represent other indicators of disease severity. The criteria to determine the presence of these disease severity measures were prospectively defined in the PROWESS study [13].

ARTICLE IN PRESS 4

J. John et al.

≥25, more organ dysfunction and vasopressor use, and higher SOFA scores, as well as higher mean APACHE II scores). In addition, the cTnI+ group had lower baseline endogenous protein C levels, more patients classified with severe protein C deficiency, and higher baseline levels of IL6 and D-dimer. The cTnI− group had significantly higher baseline platelet counts. In the overall group, the 28-day mortality was significantly higher in the cTnI+ patients as compared with the cTnI− patients (32.2% vs 13.6%, P b .0001). The significantly higher 28-day mortality rate in cTnI+ patients as compared with cTnI− patients was observed within the placebo-treated patients (35.6% vs 16.9%, P = .003) and within the DrotAAtreated patients (28.8% vs 10.5%, P = .001).

3.2. Troponin-I as independent prognosticator of mortality outcome Based on the data of all the patients, the following variables were selected (P ≤ .05) by the stepwise logistic regression procedure: treatment, age, race, functional dependency status, APACHE II score, activated partial thromboplastin time, and IL-6. When adjusting for these significant variables, the P value for troponin level was 0.014 (odds ratio, 2.020; 95% confidence interval, 1.1533.541), demonstrating that troponin level was an independent prognosticator. Of note, in this multivariable model, treatment was also a significant factor (P = .0237) in favor of DrotAA.

3.3. Troponin-I as predictor of drotrecogin alfa (activated) treatment effect The interaction between treatment and cTnI levels was evaluated by the Breslow-Day test (P = .65) and was not statistically significant (Table 2). There were, however, a few baseline differences between the treatment groups. In the cTnI− patients, the DrotAA-treated group had more patients with at least 4 organ failures at baseline as compared with the placebo-treated patients (17.1% vs 5.6%, P = .030). For the cTnI+ patients, the placebo-treated group had more baseline vasopressor use (68.9% vs 59.0%, P = .028) and higher baseline mean cardiovascular SOFA scores (mean, 2.88 vs 2.52; P = .012) as compared with the DrotAA-treated patients.

Table 2 28-day mortality outcome by troponin-I levels and DrotAA treatment 28-d Mortality −

cTnI cTn-I+

DrotAA

Placebo

P value

8/76 (10.5%) 66/229 (28.8%)

12/71 (16.9%) 79/222 (35.6%)

.260 .124

Breslow-Day interaction test for the interaction between treatment and troponin levels = 0.65.

3.4. Correlation of troponin-I with ICU-free and hospital-free days The cTnI− patients had more ICU-free days than cTnI+ patients in all subgroups (total population: mean days, 15 vs 11; P b .001; placebo-treated: mean days, 16 vs 11; P b .001; DrotAA-treated: mean days, 14 vs 11; P = .023). cTnI− patients also had more hospital-free days in the total population (mean days, 7 vs 5; P = .005) and placebo-treated subgroup (mean days, 7 vs 5; P = .044). Within the cTnI+ and cTnI− groups, there were no significant differences in ICUand hospital-free days between the placebo- and DrotAAtreated patients.

4. Discussion In evaluating serum samples from a large cohort of patients with severe sepsis, we were hoping to address 2 questions: (1) Is cTnI a marker of disease severity in severe sepsis? (2) Could troponin be used to guide therapy? This is the largest study (n = 598) to date wherein baseline cTnI measurement was used as a prognostic marker in patients with severe sepsis. Elevated cTnI values were observed in 75% (451 of 598) of patients, which is consistent with previous studies of severe sepsis patients in which troponin was noted to be elevated in 43% to 85% of patients [7-9,15-20]. These earlier studies were either retrospective analyses or small prospective observational studies, and troponin was measured within 72 hours after the onset of organ failure. In studies involving patients with septic shock, there was an incidence of elevated cTnI ranging from 43% to 80% [15,19,20]. In a heterogeneous population of patients with sepsis, severe sepsis, and septic shock, the incidence of elevated troponin ranged from 55% to 85% [7,9,16-18]. The exact mechanism leading to elevated cTnI in sepsis remains uncertain. Sepsis-induced myocardial dysfunction is well recognized, occurring in 40% to 50% of patients with septic shock. Ver Elst et al [15] noted that 78% of cTnI+ patients had reduced left ventricular ejection fraction compared with 9% of cTnI− individuals. Fernandes et al [17] observed similar findings. Although some investigators have reported that a history of coronary artery disease is more common in patients with elevated cTnI [21], objective testing has not supported the concept that flow-limiting coronary disease is the cause of the troponin release. In the study by ver Elst et al [15], autopsy was performed on 12 of the 21 patients with septic shock who died. Six patients who were cTnI+ and 4 cTnI− patients had nonspecific findings. One patient who was cTnI− had a ruptured left ventricular wall, and 1 cTnI+ patient had an anterior wall MI. In the study published by Ammann et al [5] involving patients with sepsis, 17 of the 20 patients had elevated troponin. Of these patients, 4 cTnI+ patients

ARTICLE IN PRESS Troponin-I as a prognosticator of mortality in severe sepsis underwent autopsy, 3 with normal coronaries and 1 with 2-vessel disease. An autopsy was performed on 1 cTnI− patient who was noted to have normal coronaries. Seven cTnI+ patients had stress echocardiograms that were reported as normal. In a subsequent study by Ammann et al [16], 32 of 51 patients with sepsis had elevated troponin. In 72% of these cTnI+ patients, flow-limiting coronary artery disease was excluded by dobutamine stress echocardiogram (n = 16), coronary angiogram (n = 1), and autopsy (n = 6). In 1 patient, subacute MI was noted; and another had high-grade coronary artery stenosis with individual cardiomyocyte necrosis at autopsy. One coronary angiogram and 1 stress echocardiogram showed evidence of an old myocardial scar, but no signs of actual ischemia. Unfortunately, we do not have electrocardiograms to correlate with the changes in troponin in our patients. However, several previous studies of severe sepsis patients with elevated troponin have found nonspecific electrocardiographic changes in patients with elevated cTnI [5,7]. John et al found that the incidence of wall motion abnormalities on echocardiogram in patients with severe sepsis were similar irrespective of their cTnI status [7]. Mehta et al [20] noted that wall motion abnormalities were significantly higher in individuals who were cTnI+. During severe sepsis, a combination of protein C consumption, protein S inactivation, and reduction in activity of the protein C activation complex by oxidation, cytokinemediated down-regulation, and proteolytic release of the activation components sets in motion conditions that would favor an acquired defect in the protein C pathway. This setting would favor microvascular thrombosis, increased leukocyte adhesion, and increased cytokine formation [22]. In this current study, patients who had elevated cTnI had lower protein C levels. However, 10% of the cTnI+ patients had normal protein C levels, implying there are other mechanisms that can lead to cTnI release. Patients with elevated cTnI were more critically ill as reflected by higher APACHE II scores at study entry and were more likely to have 3 or more organ failures. The APACHE II scores of patients with cTnI+ in previous studies [5-7,15-20], when available, varied between 26.3 and 28.1 and were significantly higher than those in the cTnI− group. Nonetheless, cTnI remained an independent predictor of mortality in the multivariate model, which included APACHE II scores. It is possible that there could be other factors not entered into the model that would affect cTnI being an independent prognosticator of mortality. Previous studies have demonstrated an association between cTnI levels and hospital or ICU length of stay. Mehta et al [20] demonstrated that cTnI (along with APACHE II) was an independent predictor of death and length of ICU stay; Lim et al [23] also showed that elevated cTn was associated with an increased length of ICU stay and hospital stay. In addition, Stein et al [24] showed that intermediate cTnI was independently associated with in-hospital mortality and length of ICU stay. In the current study, cTnI+ patients had

5 fewer ICU-free days possibly because of the increased mortality in this group. The ICU-free days and hospital-free days were similar in survivors. In a previous single-center study, John et al [7] found that patients with elevated cTnI were more critically ill, had a higher mortality, and also had a greater benefit from DrotAA. In the current study, the elevated cTnI group also was more critically ill and had a higher mortality. However, in contrast to the findings by John et al, a relationship between cTnI positivity and a survival benefit with DrotAA treatment could not be established in this population (in the former study, patients were prospectively identified; but cTnI was measured at the discretion of the primary physician, which may have led to selection bias). The patients in that study were also sicker (mean APACHE II, 27); and mortality for the entire group (40%), as well as in the group with elevated cTnI (52%), was higher than the current study. Although the current study measured cTnI in samples from a larger randomized cohort, only 598 of the 1690 PROWESS patients had samples available to be tested. The fact that this was a retrospective examination of a subgroup with available samples is a limitation of this study. One or more of these factors may have contributed to the difference in outcome to DrotAA in patients with elevated cTnI.

5. Conclusion We found, in a subset of patients from the PROWESS trial, that elevated cTnI was associated with a higher mortality. Using multivariable analysis, troponin was an independent prognosticator of mortality, but was not a predictor of the survival benefit observed with DrotAA administration. This study further supports the association between increased cTnI levels and increased hospital and ICU lengths of stay.

Acknowledgments This study was funded by Eli Lilly and Company. Drs John, Woodward, Wang, Yan, Kinasewitz, and Heiselman, and Ms Fisher contributed to the interpretation of the data and the drafting of the manuscript. Dr Wang performed the statistical analyses for the manuscript. All authors reviewed and approved the final version of the manuscript. The authors wish to acknowledge the following Lilly colleagues for their assistance: Suzane Um for managing and coordinating the entire sample and clinical data deidentification process; Betty Yan, Suzane Um, Christine Clauss, Kelly Sweeney, Brad Woodward, and Jeri Dalka for deidentifying the frozen samples; and Alan Degner, Brenda Hanssen, Michele Langsford, and Bruce Beechler for providing the measurements of cTnI.

ARTICLE IN PRESS 6

References [1] McLean AS, Huang SJ, Salter M. Bench-to bedside review: the value of cardiac biomarkers in the intensive care patient. Crit Care 2008;12: 215. [2] Collinson P, Gaze D. Cardiac troponins in intensive care. Crit Care 2005;9:345-6. [3] Fromm R. Cardiac troponins in the intensive care unit: common causes of increased levels and interpretation. Crit Care Med 2007;35: 584-8. [4] Alpert JS, Thygesen K, Antman E, et al. Myocardial infarction redefined—a consensus document of The Joint European Society of Cardiology/American College of Cardiology Committee for the redefinition of myocardial infarction. J Am Coll Cardiol 2000;36: 959-69. [5] Ammann P, Fehr T, Minder EI, et al. Elevation of troponin I in sepsis and septic shock. Intensive Care Med 2001;27:965-9. [6] Arlati S, Brenna S, Prencipe L, et al. Myocardial necrosis in ICU patients with acute non-cardiac disease: a prospective study. Intensive Care Med 2000;26:31-7. [7] John J, Awab A, Norman D, et al. Activated protein C improves survival in severe sepsis patients with elevated troponin. Intensive Care Med 2007;33:2122-8. [8] Mannam P, Devarakonda VS, Wittbrodt ET, et al. Association of troponin I concentrations with outcomes in sepsis. Chest 2004; 865S:126. [9] Maeder M, Fehr T, Rickli H, et al. Sepsis-associated myocardial dysfunction: diagnostic and prognostic impact of cardiac troponins and natriuretic peptides. Chest 2006;129:1349-66. [10] Sands KE, Bates DW, Lanken PN, et al. Academic Medical Center Consortium Sepsis Project Working Group. Epidemiology of sepsis syndrome in 8 academic medical centers. JAMA 2007;278:234-40. [11] Miniño AM, Heron MP, Murphy SL, et al. Disease Control and Prevention National Center for Health Statistics National Vital Statistics System. Deaths: final data for 2004. Natl Vital Stat Rep 2007;55:1-119.

J. John et al. [12] Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001;29:1303-10. [13] Bernard GR, Vincent JL, Laterre PF, et al. Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001;344:699-709. [14] Heiselman DE, Woodward DB, Wang Y, Fisher DL, Kinasewitz G, John J. Troponin as a prognosticator of mortality in severe sepsis patients. Crit Care Med 2008;36(12 Suppl):A160. [15] ver Elst KM, Spapen HD, Nguyen DN, et al. Cardiac troponins I and T are biological markers of left ventricular dysfunction in septic shock. Clin Chem 2000;46:650-7. [16] Ammann P, Maggiorini M, Bertel O, et al. Troponin as a risk factor for mortality in critically ill patients without acute coronary syndromes. J Am Coll Cardiol 2003;41:2004-9. [17] Fernandes Jr CJ, Akamine N, Knobel E. Cardiac troponin: a new serum marker of myocardial injury in sepsis. Intensive Care Med 1999; 25:1165-8. [18] Spies C, Haude V, Fitzner R, et al. Serum cardiac troponin T as a prognostic marker in early sepsis. Chest 1998;113:1055-63. [19] Turner A, Tsamitros M, Bellomo R. Myocardial cell injury in septic shock. Crit Care Med 1999;27:1775-80. [20] Mehta NJ, Khan IA, Gupta V, et al. Cardiac troponin I predicts myocardial dysfunction and adverse outcome in septic shock. Int J Cardiol 2004;95:13-7. [21] Scott EC, Ho HC, Yu M, et al. Pre-existing cardiac disease, troponin I elevation and mortality in patients with severe sepsis and septic shock. Anaesth Intensive Care 2008;36:515-9. [22] Esmon CT. The protein C pathway. Chest 2003;124(3 Suppl):26S-32S. [23] Lim W, Qushmag I, Devereaux PJ, et al. Elevated cardiac troponin measurements in critically ill patients. Arch Inten Med 2006;166: 2446-54. [24] Stein R, Gupta B, Agarwal S, et al. Prognostic implications of normal (b0.10 ng/ml) and borderline (0.10 to 1.49 ng/ml) troponin elevation levels in critically ill patients without acute coronary syndrome. Am J Cardiol 2008;102:509-12.