Article

Interleukin-6 at discharge predicts all-cause mortality in patients with sepsis

a b s t r a c t

Purpose: Interleukin-6 (IL-6) is a proinflammatory cytokine that plays a central role in the pathogenesis of sepsis. We aim to investigate the association between IL-6 and all-cause mortality in patients with sepsis. Methods: A cohort of 40 elderly patients with sepsis was identified between March 2009 and June 2010 at Rambam Health Medical Campus, Haifa, Israel. The cohort was followed up for all-cause mortality occurring during the 6 months after hospital discharge. Cox proportional hazard model was used to assess the association between IL-6 and all-cause mortality.

Results: Iinterleukin-6 at discharge had a higher predictive accuracy for all-cause mortality when compared with IL-6 at admission. The area under the curve was 0.752 (P = .015) and 0.545 (P = .661), respectively. Eleven (27.5%) patients died during follow-up; the subjects who died have higher IL-6 levels at discharge (median, 50.6 pg/mL [interquartile range, 39.6-105.9]) compared with survivors at the end of follow-up (median, 35.4 [interquartile range, 15.8-49]; P = .014). The risk of all-cause mortality was higher in subjects with IL-6 levels above the median compared with subjects with lower IL-6 levels (log-rank P = .017). On multivariate Cox proportional analysis, adjusting for the potential confounders, IL-6 at discharge remained an independent predictor for 6 month all-cause mortality (hazard ratio, 6.05 [1.24-24.20]) for levels above the median compared with lower levels.

Conclusions: Iinterleukin-6 at discharge is an independent predictor of all-cause mortality in patients with sepsis. Compared with IL-6 at admission, IL-6 at discharge better predicts all-cause mortality.

(C) 2013

Introduction

Sepsis is a fatal disease that involves approximately 750 000 patients a year in the United States. The incidence is increasing in the elderly and in patients with comorbid conditions [1]. Interleukin-6 (IL-6) is a well-known proinflammatory cytokine that plays a central role in the pathogenesis of sepsis. Elevated levels of IL-6 early in the course of the disease can predict the development of sepsis [2,3]. Several studies had suggested IL-6 as a marker of sepsis, whereas others incorporated it in several models as a predictor of sepsis development and its severity. However, all these studies have addressed IL-6 levels either early or through the course of the disease but not at discharge [4-9].

? This work was supported, in part, by The Rappaport Family Institute for Research in the Medical Sciences, Technion, Israel Institute of Technology, Haifa, Israel.

* Corresponding author. Department of Internal Medicine C, Ha’emek Medical Center, Afula 18101, Israel. Tel.: +972 4 6495132; fax: +972 4 6495134.

E-mail address: [email protected] (W. Saliba).

1 These authors contributed equally to this work.

Yende et al [10] reported that in patients with pneumonia, elevated levels of IL-6 and IL-10 at discharge are associated with increased 1-year mortality. They suggested that elevated levels at the time of discharge could possibly represent persistent inflammation, even if the patients have improved clinically.

Given that the role of IL-6 levels at discharge, as a predictor of mortality, has not been examined in further etiologies of sepsis, other than pneumonia, in this study we aim to investigate the relationship between IL-6 at discharge and 6 month all-cause mortality.

Materials and methods

Patient population

The study was performed at Rambam Health Care Campus, Haifa, Israel, and included patients who were admitted to the department of internal medicine “B” with the primary diagnosis of sepsis of any cause between March 29, 2009, and June 5, 2010. After the initiation of the study, owing to logistic problems, recruitment of patients in the study was withhold and continued after few months. Overall, 40

0735-6757/$ – see front matter (C) 2013 http://dx.doi.org/10.1016/j.ajem.2013.06.011

1362 M. Naffaa et al. / American Journal of Emergency Medicine 31 (2013) 13611364

patients with sepsis were recruited in the study of 117 patients with sepsis who were admitted during the study period. The study was conducted in accordance with the principles of the Declaration of Helsinki and approved by The Rambam Hospital institutional review board.

The diagnosis of sepsis was defined according to the published consensus as systemic inflammatory response syndrome with suspected or proven infection [11]. Subjects with active cancer were excluded.

Data that were collected include the following: (1) demographic and clinical characteristics (age, sex, and duration of hospitaliza- tions), (2) comorbidities (ischemic heart disease and diabetes mellitus), and (3) laboratory variables (at least 2 sets of blood samples were obtained from all participants in the study, at the day of admission and at the day of discharge). The first blood sample was collected in the ward of internal medicine during the first 24 hours from the patient presentation to the emergency department. Hemoglobin level, mean corpuscular volume, and red cell distribution width (RDW) were measured using the Advia 120 Hematology Analyzer (Siemens Healthcare Diagnostics, Deerfield, IL). Glucose, Blood urea nitrogen , and Creatinine levels were measured using the “Dimension” (Siemens Healthcare Diagnostics). Interleukin-6 levels were determined using the “Quantikine human interleukin 6 Elisa assay” (R&D Systems, Minneapolis, MN). procalcitonin levels were analyzed using an automated microparticle immunoassay for procalcitonin (Liaison Brahms PCT; Brahms Diagnostics, Berlin, Germany). Brain natri- uretic peptide (BNP) levels were determined using the BNP microparticle enzyme immunoassay using an AXSYM system (Abott Laboratories, Abbott Park, IL).

Study end point

All patients were followed up for 6 months after hospital discharge. The primary end point of the study was all-cause mortality. Mortality data were retrieved from the database of our hospital and the Ministry Of Health.

Statistical analyses

Continuous variables are presented as means (SD), along with medians and interquartile range (IQR). Categorical variable are presented as proportions. The Mann-Whitney test was used to compare means, and the association between categorical variables was tested using the ?2 test or the Fischer exact test, as appropriate. spearman correlation coefficient was used to test the relation between continuous variables. Subjects were followed up for 6 months from their discharge, and time to death was calculated from their discharge until death. The distribution of time to death is presented by Kaplan-Meier curves and compared with log-rank test. The association between time to death and potential predictors of mortality was analyzed by the univariate Cox proportional hazard model. Variables with a P value of .1 or less in the univariate model were included in the multivariate Cox proportional hazard model using the backward selection. Multiple linear regression was used to study the association between BNP level at the time of discharge and demographic variables, clinical variables, and Biochemical biomarkers. For this purpose, BNP and other skewed continuous variables were logarithmically trans- formed. Receiver operating characteristic curve was constructed, and the area under the curve was calculated to compare the predictive accuracy for 6-month all-cause mortality of IL-6 at

Results

Study population characteristics

Forty patients with sepsis were included in the study. The median age was 79.0 years (IQR, 68.2-84.0), and 15 (37.5%) were female. Using the median cut point of IL-6 levels at the time of discharge, we categorized the study patients into 2 groups: subjects with IL-6 below the median (<=38.6 pg/mL) and subjects with IL-6 above the median (N 38.6 pg/mL). The baseline characteristics of the 2 groups were similar (Table 1).

The association between IL-6 and 6 month all-cause mortality

Eleven (27.5%) patients died during the follow-up period. Subjects who died had higher IL-6 levels (median, 50.6 pg/mL [IQR, 39.6- 105.9]), as compared with survivors at the end of follow-up period (median, 35.4 pg/mL [IQR, 15.8-49.6]; P = .014).

The risk of 6-month all-cause was higher in subjects with IL-6 levels higher than 38.6 pg/mL compared with subjects with IL-6 levels of 38.6 pg/mL or less (log-rank P = .017; Fig. 1). On multivariate Cox proportional analysis, IL-6 remained an independent predictor for 6- month mortality (hazard ratio [HR], 6.05 [1.24-24.20], for levels N 38.6 pg/mL compared with lower levels; Table 2). Red cell distribution width and serum BUN were directly associated with increased risk of mortality and were the only variables along with IL-6 that reached statistical significance on multivariate analyses (Table 2).

Procalcitonin and BNP did not reach statistical significance on univariate Cox proportional hazard model, neither when they were forced into the multivariate model.

The predictive accuracy values, for 6-month all-cause mortality, of IL-6 level at the time of admission and the time of discharge were compared using receiver operating characteristic curve analyses. The

Table 1

Baseline characteristics of patients with sepsis according to IL-6 levels (median

cut points)

Variables

IL-6 below the median,

<=38.6 pg/mL (n = 20)

IL-6 above the median,

N 38.6 pg/mL (n = 20)

P

Age (y) Mean +- SD

72.7 +- 14.7

77.0 +- 9.7

.394

Median (IQR)

Sex, n (%)

78.5 (65.7-83.5)

80.0 (71.2-85.5)

.327

Female

9 (45.0)

6 (30.0)

Male

BNP (pg/mL) Mean +- SD

11 (56.0)

607.2 +- 785.5

14 (70.0)

899.8 +- 1092.5

.745

Median (IQR) Procalcitonin (ng/mL)

Mean +- SD

231.9 (133.3-849.2)

1.20 +- 2.30

346.8 (59.5-1432.0)

4.75 +- 10.11

.384

Median (IQR)

0.33 (b0.1-1.64)

0.43 (b0.1-4.1)

Creatinine (mg/dL)

Mean +- SD

1.43 +- 1.04

1.80 +- 1.92

.944

Median (IQR) BUN (mg/dL)

Mean +- SD

1.06 (0.83-1.48)

22.0 +- 17.4

1.17 (0.78-1.80)

27.0 +- 17.7

.346

Median (IQR)

Hemoglobin (g/dL) Mean +- SD

19.0 (11.0-25.0)

11.8 +- 1.3

21.0 (13.2-36.5)

12.2 +- 1.9

.542

Median (IQR) RDW (%)

Mean +- SD

11.7 (10.6-12.8)

15.2 +- 2.1

11.7 (11.0-12.8)

14.7 +- 1.2

.786

Median (IQR)

Duration of

14.4 (13.7-16.9)

14.4 (13.9-15.2)

.786

hospitalization (per 1 d)

Mean +- SD 6.8 +- 4.2 8.5 +- 7.7

admission and IL-6 at discharge. For all analyses, a P value less

Median (IQR)

6.0 (3.2-10.5)

6.0 (3.0-12.7)

than .05 for 2-tailed tests was considered statistically significant.

Diabetes mellitus, n (%)

10 (50)

7 (35)

.337

All statistical analyses were performed using SPSS 18.0 (SPSS Inc,

Chicago, IL).

Ischemic heart disease,

n (%)

9 (45)

7 (35)

.519

M. Naffaa et al. / American Journal of Emergency Medicine 31 (2013) 13611364 1363

IL-6 ?38.6 pg/ml

IL-6 >38.6 pg/ml

Log-rank P value = .017

Fig. 1. Kaplan-Meier curves for 6-month all-cause mortality according to IL-6 levels (median cut points) among patients with sepsis.

Fig. 2. Receiver operating characteristic curves for the performance of IL-6 at the time of admission and at the time of discharge to predict 6-month all-cause mortality in patients with sepsis.

area under the curve was 0.545 (P = .661) for IL-6 level at admission and 0.752 (P = .015) for IL-6 level at discharge (Fig. 2). Using an IL-6 threshold of 38.6 pg/mL at discharge, the positive predictive value (PPV) and the negative predictive value (NPV) for 6-month all- cause mortality were 45% and 90%, respectively. Notably, the threshold of 19.8 pg/mL for IL-6 level has an NPV of 100% for 6- month all-cause mortality.

Predictors of high BNP levels in sepsis

Brain natriuretic peptide levels were high in our cohort with sepsis (median, 263 pg/mL [IQR, 90-1226]) and was significantly correlated with procalcitonin (Spearman correlation coefficient r = 0.528, P b

.001), creatinine (r = 0.54, P b .001), BUN (r = 0.476, P = .002), RDW

(r = 0.454, P = .003), and hemoglobin (r = -426, P = .006). Brain natriuretic peptide was not significantly correlated with IL-6 at discharge (r = 0.284, P = .075).

Table 2

Univariate and multivariate Cox proportional hazard analysis for 6-month all-cause mortality among patients with sepsis

Variables Univariatea Multivariateb

HR (95% CI) P HR (95% CI) P

IL-6

<=38.6 pg/mL Reference Reference

N 38.6 pg/mL 5.22 (1.12-24.20) .035 6.05 (1.24-24.20) .026

On multiple linear regression model, procalcitonin, creatinine, hemoglobin, sex, and history of ischemic heart disease were independent predictors of BNP levels in patients with sepsis. The presence of an underlying ischemic heart disease was the strongest independent factor of BNP in sepsis (Table 3). This model explained 65.5% of the variability in BNP levels in patients with sepsis.

Discussion

This study shows that high IL-6 levels at discharge are associated with increased 6-month all-cause mortality. Compared with IL-6 at the time of admission, IL-6 at discharge is a better predictor of 6- month all-cause mortality in patients with sepsis.

Sepsis continues to be a disease with high fatality, and efforts are directed toward early diagnosis and treatment [1]. The need to identify patients with severe condition or high-risk patients has led to the emergence of different models aimed to predict mortality in such patients [8,9,12]. Many of these models have involved IL-6 level as a predictor of mortality, either on admission or early in the course of the disease. However, the role of IL-6 at discharge has not been widely examined.

Table 3

Multiplea Linear regression analysis for the prediction of Ln BNP (dependent variable) among patients with sepsis

BUN (per 10-mg/dL increase)

RDW

1.24 (0.96-1.59)

.09

1.44 (1.07-1.94)

.015

Variables

Regression coefficient (SE)

P

<=14.5

Reference

Reference

Ln procalcitonin

0.234 (0.103)

.029

N 14.5

2.77 (0.81-9.51)

.1

4.27 (1.13-16.12)

.033

Ln creatinine

0.673 (0.283)

.023

Duration of hospitalization

1.06 (0.99-1.13)

.067

Hemoglobin (per 1-g increase)

-0.232 (0.097)

.022

(per 1 d)

Male

-0.722 (0.341)

.042

a Only variables with a P value of .1 or less are shown. Variables considered for univariate analysis include age, sex, history of diabetes and ischemic heart disease, duration of hospitalization, and blood tests at time of discharge (IL-6, procalcitonin, BNP, hemoglobin, RDW, BUN, and creatinine).

b Variables with a P value of .1 or less in univariate analysis were included in the multivariate model. Only variables with a P value less than .05 are shown.

Ischemic heart disease 1.023 (0.325) .004

a Variables considered for univariate analysis include age, sex, history of diabetes and ischemic heart disease, duration of hospitalization, and blood tests at the time of discharge (IL-6, procalcitonin, hemoglobin, RDW, BUN, and creatinine). Variables with a P value of .05 or less in univariate analysis were included in the multiple linear model using backward selection.

1364 M. Naffaa et al. / American Journal of Emergency Medicine 31 (2013) 13611364

Our findings are in line with the finding of Yende et al [10], who have shown that elevated levels of IL-6 and IL-10 at hospital discharge were associated with increased 1-year mortality in patients hospital- ized with pneumonia. They suggested persistent inflammation as an explanation, reflected by the elevated levels of these interleukins. They demonstrated persistent, up-regulated proinflammatory re- sponse to be associated with death caused by cardiovascular disease, renal failure, infections, and cancer.

Interestingly, the time required to achieve clinical improvement in patients with sepsis was not different between patients with high and low IL-6 level, as evidenced by the similar duration of hospitalization between the 2 groups (Table 1). In addition, the duration of hospitalization was not an independent predictor of all-cause mortality in our study (Table 2). Our findings suggest that persistent inflamma- tion that lags behind the clinical improvement may better reflect severity and predict worse prognosis despite clinical improvement.

Interestingly in our study, procalcitonin, a known marker of sepsis, was not associated with 6-month all-cause mortality. Procalcitonin is useful for differentiating between systemic inflammatory response and sepsis and was found to be in correlation with severity of the clinical condition [13,14]. It may be suggested that unlike IL-6, which may be a marker of inflammation that persists after clinical improvement, procalcitonin is marker of inflammation in the early stage of sepsis and better predicts the Short-term outcome.

Another finding of the current study that merits to be mentioned is the independent association of RDW with all-cause mortality. Red cell distribution width emerged in recent years as a predictor of mortality in a variety of clinical conditions including of congestive heart failure and sepsis from different causes [15].

This study revealed high levels of BNP in patients with sepsis, which may reflect myocardial injury in these patients. Although, a preexisting ischemic heart disease in patients with sepsis was the strongest predictor of high BNP levels, high BNP levels were also independently associated with female sex, lower Hemoglobin levels, serum creatinine, and procalcitonin levels (Table 3). In agreement with our findings, it was reported that BNP levels are elevated in the acute phase of community-acquired microbial infections without severe sepsis or septic shock [16]. Nevertheless, our study did not show an association between BNP levels and all-cause mortality, and larger studies are needed before drawing conclusions.

To the best of our knowledge, this is the first study reporting a clear relationship between IL-6 levels at discharge and 6-month all- cause mortality in patients with the primary diagnosis of sepsis caused by different etiologies including pneumonia.

According to the results of our study, along with results reported by Yende et al, we suggest IL-6 levels at discharge as a predictor of mortality. It should be incorporated in future models to predict prognosis of sepsis. We recommend that patients with higher levels of IL-6 should have further evaluation before discharge.

The major limitation of our study is the small number of patients. In addition. we did not have data on the cause of death, and therefore, we were not able to assess the association with cause-specific mortality. Despite these limitations, our study provide new and unique information that may serve for hypothesis generation to be investigated in larger studies.

In conclusion, IL-6 at discharge is an independent predictor of 6- month all-cause mortality in patients with sepsis. Compared with IL-6 at admission, IL-6 at discharge better predicted all-cause mortality. Our findings and the role of BNP in patients with sepsis need to be addressed in larger studies.

References

  1. Russell JA. Management of sepsis. N Engl J Med 2006;355:1699-713.
  2. Bagdade J, Pedersen BK, Schwenke D, et al. Acute effects of interleukin-6 infusion on apo-B-containing lipoProtein Subclasses in humans. Scand J Clin Lab Invest 2011;71:449-55.
  3. Gouel-Cheron A, Allaouchiche B, Guignant C, et al. Early interleukin-6 and slope of monocyte human leukocyte antigen-DR: a powerful association to predict the development of sepsis after major trauma. PLoS One 2012;7:e33095.
  4. Andaluz-Ojeda D, Bobillo F, Iglesias V, et al. A combined score of pro- and anti- inflammatory interleukins improves Mortality prediction in severe sepsis. Cytokine 2012;57:332-6.
  5. Badiu DC, Paunescu V, Aungurenci A, et al. proinflammatory cytokines in peritonitis. J Med Life 2011;4:158-62.
  6. Nierhaus A, Linssen J, Wichmann D, et al. Use of a weighted, automated analysis of the differential blood count to differentiate sepsis from non-infectious systemic inflammation: the intensive care infection score (ICIS). Inflamm Allergy Drug Targets 2012;11:109-15.
  7. Ravishankaran P, Shah AM, Bhat R. Correlation of interleukin-6, serum lactate, and C-reactive protein to inflammation, complication, and outcome during the surgical course of patients with acute abdomen. J Interferon Cytokine Res 2011;31:685-90.
  8. Tschaikowsky K, Hedwig-Geissing M, Braun GG, et al. Predictive value of procalcitonin, interleukin-6, and C-reactive protein for survival in postoperative patients with severe sepsis. J Crit Care 2011;26:54-64.
  9. Uusitalo-Seppala R, Koskinen P, Leino A, et al. Early detection of severe sepsis in the emergency room: diagnostic value of plasma C-reactive protein, procalcitonin, and interleukin-6. Scand J Infect Dis 2011;43:883-90.
  10. Yende S, D’Angelo G, Mayr F, et al. Elevated hemostasis markers after pneumonia increases one-year risk of all-cause and Cardiovascular deaths. PLoS One 2011;6: e22847.
  11. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. 1992. Chest 2009;136(5 Suppl):e28.
  12. de Pablo R, Monserrat J, Reyes E, et al. Circulating sICAM-1 and sE-selectin as biomarker of infection and prognosis in patients with systemic inflammatory response syndrome. Eur J Intern Med 2013;24:132-8.
  13. Reinhart K, Carlet J. Procalcitonin–a new marker of severe infection and sepsis. Intensive Care Med 2000;26(Suppl 2):S145.
  14. BalcI C, Sungurtekin H, Gurses E, et al. Usefulness of procalcitonin for diagnosis of sepsis in the intensive care unit. Crit Care 2003;7:85-90.
  15. Braun E, Domany E, Kenig Y, et al. Elevated red cell distribution width predicts poor outcome in young patients with community acquired pneumonia. Crit Care 2011;15:R194.
  16. Nikolaou NI, Goritsas C, Dede M, et al. Brain natriuretic peptide increases in septic patients without severe sepsis or shock. Eur J Intern Med 2007;18:535-41.

Leave a Reply

Your email address will not be published. Required fields are marked *