Research ArticleCirculating MicroRNA-223 Serum Levels Do Not PredictSepsis or Survival in Patients with Critical Illness

Fabian Benz1 Frank Tacke1 Mark Luedde2 Christian Trautwein1

Tom Luedde1 Alexander Koch1 and Christoph Roderburg1

1Department of Medicine III University Hospital RWTH Aachen Pauwelsstrasse 30 52074 Aachen Germany2Department of Cardiology and Angiology University of Kiel Schittenhelmstrasse 12 24105 Kiel Germany

Correspondence should be addressed to Christoph Roderburg croderburgukaachende

Received 11 December 2014 Accepted 4 February 2015

Academic Editor Luisella Bocchio-Chiavetto

Copyright copy 2015 Fabian Benz et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Background and Aims Dysregulation of miR-223 was recently linked to various diseases associated with systemic inflammatoryresponses such as type 2 diabetes cancer and bacterial infections However contradictory results are available on potentialalterations of miR-223 serum levels during sepsis We thus aimed to evaluate the diagnostic and prognostic value of miR-223 serumconcentrations in patients with critical illness and sepsis Methods We used iv injection of lipopolysaccharide (LPS) as well ascecal pole ligation and puncture (CLP) for induction of polymicrobial sepsis in mice and measured alterations in serum levels ofmiR-223These results frommice were translated into a large and well-characterized cohort of critically ill patients admitted to themedical intensive care unit (ICU) Finally results from analysis in patients were correlated with clinical data and extensive sets ofroutine and experimental biomarkers Results Although LPS injection induced moderately elevated serummiR-223 levels in miceno significant alterations in miR-223 serum levels were found in mice after CLP-induced sepsis In accordance with these resultsfrom animal models serummiR-223 levels did not differ between critically ill patients and healthy controls However ICU patientswith more severe disease (APACHE-II score) showed moderately reduced circulating miR-223 Strikingly no differences in miR-223 levels were found in critically ill patients with or without sepsis and serum levels of miR-223 did not correlate with classicalmarkers of inflammation or bacterial infection Finally low miR-223 serum levels were moderately associated with an unfavorableprognosis of patients during the ICU treatment but did not predict long-term mortality Conclusion Recent reports on alterationsin miR-223 serum levels during sepsis revealed contradictory results preventing a potential use of this miRNA in clinical routineWe clearly show that miR-223 serum levels do not reflect the presence of sepsis neither in mouse models nor in a large cohortof ICU patients and do not indicate clinical outcome of critically ill patients Thus miR-223 serum levels should not be used as abiomarker in this setting

1 Introduction

Sepsis represents a major cause of death for critically illpatients during intensive care unit (ICU) treatment [1] Inthis setting early diagnosis and initiation of specific thera-peuticmeasureswere shown to considerably reducemortalityin critically ill patients with sepsis Thus in addition toestablished laboratory parameters and clinical scores novelbiomarkersmay significantly improve treatment and progno-sis assessment in patients admitted to the intensive care unit[2]

MicroRNAs (miRNAs) represent small RNAs of 22nucleotides in length that do not withhold the sequentialinformation to transcribe for proteins but function as criticalregulators of gene expression in eukaryotes [3 4] miRNAsare involved in various pathophysiological processes such ascell injury proliferation or carcinogenesis [5ndash7] A specificderegulation of microRNA miR-223 was described in differ-ent disease states associated with a systemic inflammatoryresponse such as bacterial infections or autoimmune diseasesAlthough this may represent an epiphenomenon of thesediseases recent evidence suggests that miR-223 is actively

Hindawi Publishing CorporationDisease MarkersVolume 2015 Article ID 384208 10 pageshttpdxdoiorg1011552015384208

2 Disease Markers

involved in regulation of inflammatory processes by limitingthe inflammation to prevent collateral damage and tissueinjury [8] In line with this assumption various targets of themiR-223 belong to the superfamily of inflammatory genessuch as granzyme B IKK-alpha and STAT3 Moreover theexpression of miR-223 itself is regulated by inflammatorypathways such as NF-120581B or the TLR4 pathway suggesting adeep integration of this miRNA in inflammatory signalingcascades driving inflammation and anti-infectious responsesin general [8]

Besides their functional involvement in gene expressionmiRNAs can be released to the extracellular compartmentand are easily detectable in body fluids such as blood sweatand urine However at present little is known about miRNAsderegulated in the serum of patients with critical illness andsepsis It was demonstrated that serum levels of differentmiRNAs such as miR-133a miR-150 and miR-146a are sig-nificantly altered in critically ill patients compared to healthycontrols [9ndash11] Importantly alterations in miR-223 serumlevels were described in patients with sepsis but have yetled to conflicting results While Wang et al described higherlevels of miR-223 in septic patients [12] and suggest a directassociation between high miR-223 serum levels and patientsrsquooutcome [13] Wang et al reported lower levels of miR-223 inthe serum of patients with septic disease [11] thus highlight-ing the need for further studies clarifying the regulation ofmiR-223 during bacterial infection and sepsis

Considering the complexity of septic disease in humanswe first analysed serum levels of miR-223 in different well-established models of experimental polymicrobial sepsis inmice such as cecal ligation and puncture (CLP) surgery andlipopolysaccharide (LPS) injection To translate our findingsfrom these animal models to human pathogenesis we ana-lyzed miR-223 serum levels in a large well-characterizedcohort of 221 critically ill patients (with and without sepsis)demonstrating that serum levels of miR-223 do not reflect thepresence of septic disease and are not associatedwith the clin-ical outcome of patients during intensive care unit treatment

2 Materials and Methods

21 Study Design In this present study we prospectivelyrecruited 221 patients (141 male 80 females) with a medianage of 63 years (range 18 to 89 years) that were consecutivelyadmitted to the General Internal Medicine Intensive CareUnit (ICU) at the University Hospital Aachen [14] Bloodsamples were collected upon admission to the ICU (prior totherapeutic interventions) centrifuged for 10min at 2000 gand serum sampleswere stored atminus80∘Cuntil use In addition75 healthy blood donors (47 male 29 female median age 33years and range 18ndash67) with normal values for blood countsC-reactive protein and liver enzymeswere recruited from thelocal blood donation center as healthy controls [14] Patientswere included in the study upon providing written informedconsent and the ethics committees approved this consentprocedure The study protocol is in line with Declaration ofHelsinki and was approved by the local ethics committee(Ethics Committee of the University Hospital Aachen

RWTH University Aachen Germany reference number EK15006) We did not include patients who were expected tohave a short-term (lt72 h) intensive care treatment due topostinterventional observation or acute intoxication into thisstudy [14] Patients who fulfilled the criteria proposed bythe American College of Chest Physicians and the Societyof Critical Care Medicine Consensus Conference Committeefor severe sepsis and septic shock were categorized as sepsispatients and the others as nonsepsis patients All patientswere treated in accordance with current guidelines for treat-ment of sepsis (Surviving Sepsis Campaign) and specificguidelines of the respective boards The clinical course ofpatients was followed up for a period of three years bydirectly contacting the patients the patientsrsquo relatives or theirprimary care physician [10]

22 Patient Characteristics The criteria of bacterial sepsis atthe time point of admission to the ICU were fulfilled by 137of the 221 patients (Table 1) and pneumonia represented themost common site of infection (Table 2) Sepsis patients weremore often and for longer terms in need of mechanical ven-tilation and had significantly higher levels of routinely usedbiomarkers of inflammation (C-reactive protein procalci-tonin and white blood cell count) as compared to the non-sepsis patientsrsquo cohort (Table 1) In nonsepsis patients car-diopulmonary diseases (myocardial infarction pulmonaryembolism and cardiac pulmonary edema) decompensatedliver cirrhosis or other critical conditions represented thepredominant etiologies which did not differ in age or sexfrom sepsis patients Both groups did not differ in AcutePhysiology and ChronicHealth Evaluation (APACHE) II andSimplified Acute Physiology Score 2 (SAPS2) score vaso-pressor demand or laboratory parameters indicating liver orrenal dysfunction (Table 1)

23 Cecal Ligation and Puncture (CLP) Themouse model ofcecal ligation and puncture (CLP) as a well-establishedmodelfor polymicrobial sepsis was used Male C57Bl6 mice (119899 =14 6ndash8wk of age) were purchased fromThe Jackson Labora-tory (Bar Harbor ME) and were subjected to CLP surgery asdescribed previously [15] Blood was taken before and 24 hafter surgery and serum was stored in minus80∘C until use Ani-mals received humane care according to European nationaland institutional regulations

24 LPS Injection Endotoxin mediated sepsis was inducedby lipopolysaccharide (LPS) which is a component of Gram-negative bacteria 25 120583g per gram bodyweight was injectedintraperitoneally inmale C57BL6mice andmice were killed8 hours later

25 miRNA Isolation from Serum For miRNA isolation weused 400120583L serum from human or 70120583L serum from miceAs described previously [16] serumwas spiked withmiScriptmiRNA Mimic SV40 (Qiagen 2 120583M 1 120583L100 120583L serum) forsample normalization 800 120583L phenol (Qiazol) and 200120583Lchloroform were added to the sample and mixed vigorouslyfor 15 sec followed by incubation at room temperature for

Disease Markers 3

Table 1 Baseline patient characteristics

Parameter All patients Nonsepsis Sepsis 119875 valueNumber 221 84 137 naSex (malefemale) 14180 5628 8552 nsAge median (range) [years] 63 (18ndash89) 62 (18ndash85) 64 (20ndash89) nsAPACHE-II score median (range) 17 (2ndash40) 15 (2ndash33) 180 (3ndash40) nsSAPS2 score median (range) 425 (0ndash79) 41 (13ndash72) 43 (0ndash79) nsICU days median (range) 7 (1ndash83) 5 (1ndash45) 10 (1ndash83) lt0001Death during ICU [] 222 131 277 lt005Ventilation [] 671 605 712 lt001Body mass index 2608 (166ndash865) 261 (166ndash533) 261 (183ndash865) nsCreatinine 13 (0ndash15) 10 (03ndash150) 15 (00ndash107) nsAlbumin 270 (152ndash522) 292 (00ndash522) 258 (00ndash410) nsWBC median (range) [times103120583L] 122 (01ndash674) 116 (18ndash277) 127 (01ndash674) nsCRP median (range) [mgdL] 940 (lt5ndash230) 17 (5ndash230) 165 (lt5ndash230) lt0001Procalcitonin median (range) [120583gL] 07 (0ndash1806) 02 (01ndash1000) 23 (00ndash1806) lt0001Interleukin-6 median (range) [pgmL] 105 (0ndash83000) 63 (4ndash83000) 220 (0ndash28000) lt0001Tumor necrosis factor median [pgmL] 19 (49ndash140) 165 (80ndash100) 235 (49ndash140) lt0001INR 118 (0ndash92) 116 (090ndash432 118 (000ndash92) nsAPACHE Acute Physiology and Chronic Health Evaluation CRP C-reactive protein ICU intensive care unit INR international normalized ratio SAPSsimplified acute physiology score WBC white blood cell count

Table 2 Disease etiology of the study population

Sepsis Nonsepsis119899 = 137 119899 = 84

Sepsis critical illness (119899 ())Source of infectionPulmonary 74 (540)Abdominal 28 (204)Urogenital 3 (22)Other 32 (234)

Nonsepsis critical illness (119899 ())Cardiopulmonary disease 30 (357)Decompensated liver cirrhosis 12 (143)Nonsepsis other 42 (500)

10min Then samples were centrifuged for 15min at 12000 gand the aqueous phase containing total RNA was precipi-tated with 500120583L 100 isopropanol and 2120583L glycogen (Fer-mentas St Leon-Rot Germany) overnight at minus20∘C Aftercentrifugation at 4∘C for 30min (12000 g) the pellets werewashed once with 70 ethanol and precipitated RNA wasresuspended in 30 120583L RNase-free water (Ambion AustinTX) [9]

26 miRNA Isolation from Tissue As described previously[16] total RNA was purified from tissue using Trizol reagent(Invitrogen Carlsbad CA USA) and RNeasy Mini kit (Qia-gen Hilden Germany) The quantity and quality of the RNAwere determined spectroscopically using a nanodrop (Ther-moScientific Waltham MA USA)

27 Quantitative Real-Time PCR Relative expression of miR-223 in serum and tissue was measured with quantitativereal-time PCR (qPCR) as recently described [16] In detailtotal RNA (1 120583g for tissue RNA 5120583L for serum RNA) wasused to synthesize cDNA utilizing miScript Reverse Tran-scriptase Kit (Qiagen) according to the manufacturerrsquos pro-tocol and was then resuspended in suitable amounts ofH2O cDNA samples (2 120583L) were used for qPCR in a total

volume of 25 120583L using the miScript SYBR Green PCR Kit(Qiagen) and miRNA-specific primers (Qiagen sequence51015840UGUCAGUUUGUCAAAUACCCCA) on a qPCR mach-ine (Applied Biosystems 7300 Sequence Detection SystemApplied Biosystems Foster City CA) All real-time PCRreactions were performed in duplicate Data were analyzedusing the SDS 23 and RQ Manager 12 software packagesand relative gene expression was generated using the 2minusΔΔCTmethod (ΔCT target gene minusΔCT control gene)

28 Statistical Analysis All statistical analyses were per-formed with SPSS 20 (IBM SPSS Statistics) as recentlydescribed [9] Data are displayed as median and range con-sidering the skewed distribution of most parameters Gaus-sian distribution was tested with Kolmogorov-Smirnov testDifferences between two groups were assessed by Studentrsquos119905-test or Wilcoxon rank-sum test and multiple comparisonsbetween more than two groups have been conducted byANOVA with Bonferronirsquos test or Dunnrsquos test for post hocanalysis Box-whisker-plot graphics illustrate a statisticalsummaryHere the box represents themedianwith interquar-tile range (IQR) and the ldquowhiskersrdquo include all values smallerthan the upper quartile plus 15 lowast IQR and larger than thelower quartileminus 15lowast IQRValues outside of thewhiskers

4 Disease Markersm

iR-2

23 re

lativ

e exp

ress

ion

Control0

2

4

6

LPS

lowastlowast

8h after

(a)

Before CLP

0

02

04

06

08

10

miR

-233

relat

ive e

xpre

ssio

n ns

CLP

0

05

10

15

miR

-233

relat

ive e

xpre

ssio

n

CLP CLP24h after 24h afterBefore

(b)

miR

-223

relat

ive e

xpre

ssio

n

CLP-OPBefore

0

1

2

3

4

5

ns

6h 24h

(c)

Brain

CLP

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

Con

trol

Heart

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

Liver

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

0

1

2

3

4

5

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

PBMC

0

1

2

3

ns

Lung

0

10

20

30

40m

iR-2

23 re

lativ

e exp

ress

ion

CLP

Con

trol

lowast

Kidney

0

1

2

3

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

lowastlowast

Spleen

(d)

Figure 1 miR-223 serum levels in murine models of septic diseases (a) Serum was obtained from C57Bl6j wild-type mice at baseline and8 hours after injection of 25 120583g per gram body weight miR-223 serum levels were measured The bar graphs represent mean plusmn SEM from119899 = 5 animals per group (b and c) Serum was obtained from C57Bl6j wild-type mice at baseline and 8 hours and 24 hours after inductionof polymicrobial sepsis by CLP miR-223 serum levels were measured The bar graphs represent mean plusmn SEM the line graphs display pairedpre- and postoperative values for individual animals (119899 = 14) (d) Relative expression of miR-233 in different organs in mice after SHAM orCLP surgery (kidney spleen liver lung heart brain and peripheral blood mononuclear cells (PBMC)) from C57Bl6j wild-type mice

are displayed as separate points and represent outliers Allvalues including outliers have been included for statisticalanalyses Correlations between variables have been analysedusing the Spearman correlation test Receiver operating char-acteristic (ROC) curve analysis and the derived area underthe curve (AUC) statistic provide a global and standardizedappreciation of the accuracy of a marker or a composite scorefor predicting an event ROC curves were generated by plot-ting sensitivity against 1 minus specificity Kaplan-Meier curveswere plotted to display the impact on survival and between-group differences were assessed using the log-rank testCox regressions were used to identify factors predicting ICUmortality or overallmortality All reported119875 valueswere two-tailed and a 119875 value less than 005 was considered to indicatestatistical significance

3 Results

31 miR-223 SerumLevels inMurineModels of Septic DiseasesBased on the contradictory data on alterations of miR-223serum levels in patients with sepsis we decided to first analysemiR-223 serum concentrations in highly standardizedmousemodels of septic disease Therefore CLP procedures and LPSinjections were performed in C57Bl6 mice As determinedby miRNA-specific qPCR miR-223 levels were moderatelybut significantly elevated 8 h after injection of LPS (Fig-ure 1(a)) In contrast levels of circulating miR-223 remainedunaffected at different time points after induction of sepsisby using the CLP model which closely resembles humansepsis (Figures 1(b) and 1(c)) Finally miR-223 expressionwasanalysed in different organs after induction of sepsis by CLP

Disease Markers 5

surgery to further determine potential mechanism regulatingmiR-223 serum levels in sepsis These analyses revealed asignificant upregulation in the lung while expression of miR-223 was downregulated in kidney and unaffected from theinduction of septic disease in the other organs including liverbrain heart and muscle (Figure 1(d))

32 miR-223 SerumLevels in Critically Ill Patients andHealthyControls We next analyzed levels of circulating miR-223in sera of 221 patients at admission to the ICU as wellas in 75 healthy volunteers miR-223 concentrations wereslightly lower in critically ill patients compared to healthycontrols (119875 = 0141 Figure 2(a)) When we analyzed serumlevels of miR-223 with respect to disease severity we foundsignificantly lower levels in patients with more severe diseaseaccording to higher APACHE-II scores (gt10) compared topatients with lower APACHE-II scores (lt10) (119875 = 0043Figure 2(b))

The metabolic status of patients was shown to influencethe outcome of critically ill patients [17] Since miR-223was shown to be involved in the pathophysiology of type 2diabetes [8] we analyzed potential correlations betweenmiR-223 serum levels and the presence of obesity or type 2diabetes Importantly we found no significant differences incirculating miR-223 between patients with obesity or normalbody weight (Figure 2(c)) and those with or without type 2diabetes mellitus respectively (Figure 2(d)) In addition nodifferences were foundwhen patients were compared by theirage or gender (data not shown)

33 miR-223 Serum Levels Do Not Indicate Sepsis in CriticallyIll Patients Elevated levels of miR-223 were suggested todiscriminate between SIRS and sepsis patients with highsensitivity and specificity [11] Our cohort of ICU patientsfeatured both patients with sepsis (119899 = 137) and patients thatdid not fulfill sepsis criteria (119899 = 84)Thus we further investi-gated the impact of sepsis on miR-223 serum concentrationsin our cohort In these analyses no significant differencesin miR-223 levels between septic and nonseptic patientswere evident (119875 = 0529 Figure 3(a)) Of note the factthat circulating miR-223 is independent of the presence ofsepsis was further substantiated by correlation analysesrevealing that serum miR-223 levels were not correlated toestablished markers of systemic inflammation and bacterialinfection such as C-reactive protein (CRP) procalcitonin(PCT) interleukin-6 (IL-6) interleukin-10 (IL-10) or tumornecrosis factor (TNF) in critically ill patients (Table 3)

In order to investigate the impact of the underlyingetiology of sepsiscritical illness on miR-223 serum levelsmore precisely we again performed subgroup analyses Thecohort of sepsis patients was subdivided into a pulmonaryand a nonpulmonary site of infection and the nonsepsispatients were categorized into liver cirrhosis cardiovasculardisorders and others However also this analysis revealedno differences in miR-223 serum concentration between thedifferent subgroups thus excluding the fact that potentialalterations in miR-223 levels might only exist in a specific

Table 3 Correlations of miR-223 serum levels at ICU admissionwith other laboratory markers

Parameter ICU patients119877 119875

Markers of liver functionCholinesterase 0135 nsProtein 0110 nsAlbumin 0089 nsGGT 0084 nsGLDH 0116 nsAP minus0074 nsINR minus0028 nsBilirubin total minus0014 nsBilirubin direct 0059 nsAST 0091 nsALT 0167 0013

Markers of inflammationC-reactive protein minus0010 nsProcalcitonin minus0085 nsIL-6 0171 nsIL-10 0118 nsTNF-alpha minus0152 nsAmylase 0218 nsLipase 0111 nsLeucocyte counts 0155 0021

Markers of renal functionCreatinine minus0391 lt0001cystatin C minus0384 lt0001cystatin C GFR 0379 lt0001Urea minus0435 lt0001

Others variablesLactate minus0146 0031NT-proCNP minus0432 lt0001suPAR minus0187 0022

Clinical scoringAPACHE-II minus0139 nsSOFA minus0187 0030ICU days minus0051 nsDays alive minus0190 ns119877 correlation coefficient 119875 119875 value 119877 and 119875 values by Spearman rankcorrelation INR international normalized ratio IL-6 interleukin-6 IL-10interleukin-10 TNF tumour necrosis factor APACHE-II Acute Physiologyand Chronic Health Evaluation II SOFA Sequential Organ Failure Assess-ment Score

group of patients and be masked if these patients are mergedwith other patients (Figure 3(b) Table 2)

In summary our analysis reveals that in contrast to pre-vious studies reporting a downregulation [11] or upregulation[12] of miR-223 in serum of patients with septic disease miR-223 levels are only slightly altered in critical illness and sepsisindicating that miR-223 measurements from serum are notsuitable to detect sepsis

34 miR-223 Serum Concentrations Do Not Predict Survivalin Critically Ill Patients Multiple organ failure represents a

6 Disease Markers

Control Patients

0

100

200

300

400

500m

iR-2

23 se

rum

leve

ls (A

U)

ns

(a)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 0043

le10 gt10

APACHE-II

(b)

No YesType 2 diabetes

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

(c)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

BMI (kgm2)

le30 gt30

(d)

Figure 2 miR-223 serum levels in critically ill patients and healthy controls (a) Serum concentrations of miR-223 were determined by qPCRin RNA extracts from serum of patients at admission to the intensive care unit (ICU) miR-223 levels were unchanged in critically ill patients(119899 = 221) compared with healthy controls (119899 = 75) (b) SerummiR-223 concentrations at admission to the ICU were significantly changed incritically ill patients with high initial Acute Physiology and Chronic Health Evaluation (APACHE) II scores (gt10) in comparison to patientswith low APACHE-II scores (le10) (c) Serum miR-223 concentrations at admission to the ICU are unchanged in patients with or withouttype 2 diabetes mellitus (d) SerummiR-223 levels at admission to the medical ICU are independent of the presence of obesity in critically illpatients Box plots are displayed where the bold line indicates the median per group the box represents 50 of the values and the horizontallines show minimum and maximum values of the calculated nonoutlier values asterisks and open circles indicate outlier values

fearful complication of sepsis and sepsis shock syndromeoften leading to death in critically ill patients To deter-mine whether miR-223 serum levels might be indicative forpatientsrsquo prognosis during and after ICU treatment we firstperformed correlation analysis between miR-223 levels and

classicalmarkers of organ dysfunctionWhilemiR-223 serumlevels showed a significant correlation with decreased renalfunction (Table 3) no correlation with acute liver injury or animpaired liver synthesis capacity could be established More-over we found no correlation to classical prognosis scores

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

Disease Markers 3

Table 1 Baseline patient characteristics

Parameter All patients Nonsepsis Sepsis 119875 valueNumber 221 84 137 naSex (malefemale) 14180 5628 8552 nsAge median (range) [years] 63 (18ndash89) 62 (18ndash85) 64 (20ndash89) nsAPACHE-II score median (range) 17 (2ndash40) 15 (2ndash33) 180 (3ndash40) nsSAPS2 score median (range) 425 (0ndash79) 41 (13ndash72) 43 (0ndash79) nsICU days median (range) 7 (1ndash83) 5 (1ndash45) 10 (1ndash83) lt0001Death during ICU [] 222 131 277 lt005Ventilation [] 671 605 712 lt001Body mass index 2608 (166ndash865) 261 (166ndash533) 261 (183ndash865) nsCreatinine 13 (0ndash15) 10 (03ndash150) 15 (00ndash107) nsAlbumin 270 (152ndash522) 292 (00ndash522) 258 (00ndash410) nsWBC median (range) [times103120583L] 122 (01ndash674) 116 (18ndash277) 127 (01ndash674) nsCRP median (range) [mgdL] 940 (lt5ndash230) 17 (5ndash230) 165 (lt5ndash230) lt0001Procalcitonin median (range) [120583gL] 07 (0ndash1806) 02 (01ndash1000) 23 (00ndash1806) lt0001Interleukin-6 median (range) [pgmL] 105 (0ndash83000) 63 (4ndash83000) 220 (0ndash28000) lt0001Tumor necrosis factor median [pgmL] 19 (49ndash140) 165 (80ndash100) 235 (49ndash140) lt0001INR 118 (0ndash92) 116 (090ndash432 118 (000ndash92) nsAPACHE Acute Physiology and Chronic Health Evaluation CRP C-reactive protein ICU intensive care unit INR international normalized ratio SAPSsimplified acute physiology score WBC white blood cell count

Table 2 Disease etiology of the study population

Sepsis Nonsepsis119899 = 137 119899 = 84

Sepsis critical illness (119899 ())Source of infectionPulmonary 74 (540)Abdominal 28 (204)Urogenital 3 (22)Other 32 (234)

Nonsepsis critical illness (119899 ())Cardiopulmonary disease 30 (357)Decompensated liver cirrhosis 12 (143)Nonsepsis other 42 (500)

10min Then samples were centrifuged for 15min at 12000 gand the aqueous phase containing total RNA was precipi-tated with 500120583L 100 isopropanol and 2120583L glycogen (Fer-mentas St Leon-Rot Germany) overnight at minus20∘C Aftercentrifugation at 4∘C for 30min (12000 g) the pellets werewashed once with 70 ethanol and precipitated RNA wasresuspended in 30 120583L RNase-free water (Ambion AustinTX) [9]

26 miRNA Isolation from Tissue As described previously[16] total RNA was purified from tissue using Trizol reagent(Invitrogen Carlsbad CA USA) and RNeasy Mini kit (Qia-gen Hilden Germany) The quantity and quality of the RNAwere determined spectroscopically using a nanodrop (Ther-moScientific Waltham MA USA)

27 Quantitative Real-Time PCR Relative expression of miR-223 in serum and tissue was measured with quantitativereal-time PCR (qPCR) as recently described [16] In detailtotal RNA (1 120583g for tissue RNA 5120583L for serum RNA) wasused to synthesize cDNA utilizing miScript Reverse Tran-scriptase Kit (Qiagen) according to the manufacturerrsquos pro-tocol and was then resuspended in suitable amounts ofH2O cDNA samples (2 120583L) were used for qPCR in a total

volume of 25 120583L using the miScript SYBR Green PCR Kit(Qiagen) and miRNA-specific primers (Qiagen sequence51015840UGUCAGUUUGUCAAAUACCCCA) on a qPCR mach-ine (Applied Biosystems 7300 Sequence Detection SystemApplied Biosystems Foster City CA) All real-time PCRreactions were performed in duplicate Data were analyzedusing the SDS 23 and RQ Manager 12 software packagesand relative gene expression was generated using the 2minusΔΔCTmethod (ΔCT target gene minusΔCT control gene)

28 Statistical Analysis All statistical analyses were per-formed with SPSS 20 (IBM SPSS Statistics) as recentlydescribed [9] Data are displayed as median and range con-sidering the skewed distribution of most parameters Gaus-sian distribution was tested with Kolmogorov-Smirnov testDifferences between two groups were assessed by Studentrsquos119905-test or Wilcoxon rank-sum test and multiple comparisonsbetween more than two groups have been conducted byANOVA with Bonferronirsquos test or Dunnrsquos test for post hocanalysis Box-whisker-plot graphics illustrate a statisticalsummaryHere the box represents themedianwith interquar-tile range (IQR) and the ldquowhiskersrdquo include all values smallerthan the upper quartile plus 15 lowast IQR and larger than thelower quartileminus 15lowast IQRValues outside of thewhiskers

4 Disease Markersm

iR-2

23 re

lativ

e exp

ress

ion

Control0

2

4

6

LPS

lowastlowast

8h after

(a)

Before CLP

0

02

04

06

08

10

miR

-233

relat

ive e

xpre

ssio

n ns

CLP

0

05

10

15

miR

-233

relat

ive e

xpre

ssio

n

CLP CLP24h after 24h afterBefore

(b)

miR

-223

relat

ive e

xpre

ssio

n

CLP-OPBefore

0

1

2

3

4

5

ns

6h 24h

(c)

Brain

CLP

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

Con

trol

Heart

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

Liver

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

0

1

2

3

4

5

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

PBMC

0

1

2

3

ns

Lung

0

10

20

30

40m

iR-2

23 re

lativ

e exp

ress

ion

CLP

Con

trol

lowast

Kidney

0

1

2

3

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

lowastlowast

Spleen

(d)

Figure 1 miR-223 serum levels in murine models of septic diseases (a) Serum was obtained from C57Bl6j wild-type mice at baseline and8 hours after injection of 25 120583g per gram body weight miR-223 serum levels were measured The bar graphs represent mean plusmn SEM from119899 = 5 animals per group (b and c) Serum was obtained from C57Bl6j wild-type mice at baseline and 8 hours and 24 hours after inductionof polymicrobial sepsis by CLP miR-223 serum levels were measured The bar graphs represent mean plusmn SEM the line graphs display pairedpre- and postoperative values for individual animals (119899 = 14) (d) Relative expression of miR-233 in different organs in mice after SHAM orCLP surgery (kidney spleen liver lung heart brain and peripheral blood mononuclear cells (PBMC)) from C57Bl6j wild-type mice

are displayed as separate points and represent outliers Allvalues including outliers have been included for statisticalanalyses Correlations between variables have been analysedusing the Spearman correlation test Receiver operating char-acteristic (ROC) curve analysis and the derived area underthe curve (AUC) statistic provide a global and standardizedappreciation of the accuracy of a marker or a composite scorefor predicting an event ROC curves were generated by plot-ting sensitivity against 1 minus specificity Kaplan-Meier curveswere plotted to display the impact on survival and between-group differences were assessed using the log-rank testCox regressions were used to identify factors predicting ICUmortality or overallmortality All reported119875 valueswere two-tailed and a 119875 value less than 005 was considered to indicatestatistical significance

3 Results

31 miR-223 SerumLevels inMurineModels of Septic DiseasesBased on the contradictory data on alterations of miR-223serum levels in patients with sepsis we decided to first analysemiR-223 serum concentrations in highly standardizedmousemodels of septic disease Therefore CLP procedures and LPSinjections were performed in C57Bl6 mice As determinedby miRNA-specific qPCR miR-223 levels were moderatelybut significantly elevated 8 h after injection of LPS (Fig-ure 1(a)) In contrast levels of circulating miR-223 remainedunaffected at different time points after induction of sepsisby using the CLP model which closely resembles humansepsis (Figures 1(b) and 1(c)) Finally miR-223 expressionwasanalysed in different organs after induction of sepsis by CLP

Disease Markers 5

surgery to further determine potential mechanism regulatingmiR-223 serum levels in sepsis These analyses revealed asignificant upregulation in the lung while expression of miR-223 was downregulated in kidney and unaffected from theinduction of septic disease in the other organs including liverbrain heart and muscle (Figure 1(d))

32 miR-223 SerumLevels in Critically Ill Patients andHealthyControls We next analyzed levels of circulating miR-223in sera of 221 patients at admission to the ICU as wellas in 75 healthy volunteers miR-223 concentrations wereslightly lower in critically ill patients compared to healthycontrols (119875 = 0141 Figure 2(a)) When we analyzed serumlevels of miR-223 with respect to disease severity we foundsignificantly lower levels in patients with more severe diseaseaccording to higher APACHE-II scores (gt10) compared topatients with lower APACHE-II scores (lt10) (119875 = 0043Figure 2(b))

The metabolic status of patients was shown to influencethe outcome of critically ill patients [17] Since miR-223was shown to be involved in the pathophysiology of type 2diabetes [8] we analyzed potential correlations betweenmiR-223 serum levels and the presence of obesity or type 2diabetes Importantly we found no significant differences incirculating miR-223 between patients with obesity or normalbody weight (Figure 2(c)) and those with or without type 2diabetes mellitus respectively (Figure 2(d)) In addition nodifferences were foundwhen patients were compared by theirage or gender (data not shown)

33 miR-223 Serum Levels Do Not Indicate Sepsis in CriticallyIll Patients Elevated levels of miR-223 were suggested todiscriminate between SIRS and sepsis patients with highsensitivity and specificity [11] Our cohort of ICU patientsfeatured both patients with sepsis (119899 = 137) and patients thatdid not fulfill sepsis criteria (119899 = 84)Thus we further investi-gated the impact of sepsis on miR-223 serum concentrationsin our cohort In these analyses no significant differencesin miR-223 levels between septic and nonseptic patientswere evident (119875 = 0529 Figure 3(a)) Of note the factthat circulating miR-223 is independent of the presence ofsepsis was further substantiated by correlation analysesrevealing that serum miR-223 levels were not correlated toestablished markers of systemic inflammation and bacterialinfection such as C-reactive protein (CRP) procalcitonin(PCT) interleukin-6 (IL-6) interleukin-10 (IL-10) or tumornecrosis factor (TNF) in critically ill patients (Table 3)

In order to investigate the impact of the underlyingetiology of sepsiscritical illness on miR-223 serum levelsmore precisely we again performed subgroup analyses Thecohort of sepsis patients was subdivided into a pulmonaryand a nonpulmonary site of infection and the nonsepsispatients were categorized into liver cirrhosis cardiovasculardisorders and others However also this analysis revealedno differences in miR-223 serum concentration between thedifferent subgroups thus excluding the fact that potentialalterations in miR-223 levels might only exist in a specific

Table 3 Correlations of miR-223 serum levels at ICU admissionwith other laboratory markers

Parameter ICU patients119877 119875

Markers of liver functionCholinesterase 0135 nsProtein 0110 nsAlbumin 0089 nsGGT 0084 nsGLDH 0116 nsAP minus0074 nsINR minus0028 nsBilirubin total minus0014 nsBilirubin direct 0059 nsAST 0091 nsALT 0167 0013

Markers of inflammationC-reactive protein minus0010 nsProcalcitonin minus0085 nsIL-6 0171 nsIL-10 0118 nsTNF-alpha minus0152 nsAmylase 0218 nsLipase 0111 nsLeucocyte counts 0155 0021

Markers of renal functionCreatinine minus0391 lt0001cystatin C minus0384 lt0001cystatin C GFR 0379 lt0001Urea minus0435 lt0001

Others variablesLactate minus0146 0031NT-proCNP minus0432 lt0001suPAR minus0187 0022

Clinical scoringAPACHE-II minus0139 nsSOFA minus0187 0030ICU days minus0051 nsDays alive minus0190 ns119877 correlation coefficient 119875 119875 value 119877 and 119875 values by Spearman rankcorrelation INR international normalized ratio IL-6 interleukin-6 IL-10interleukin-10 TNF tumour necrosis factor APACHE-II Acute Physiologyand Chronic Health Evaluation II SOFA Sequential Organ Failure Assess-ment Score

group of patients and be masked if these patients are mergedwith other patients (Figure 3(b) Table 2)

In summary our analysis reveals that in contrast to pre-vious studies reporting a downregulation [11] or upregulation[12] of miR-223 in serum of patients with septic disease miR-223 levels are only slightly altered in critical illness and sepsisindicating that miR-223 measurements from serum are notsuitable to detect sepsis

34 miR-223 Serum Concentrations Do Not Predict Survivalin Critically Ill Patients Multiple organ failure represents a

6 Disease Markers

Control Patients

0

100

200

300

400

500m

iR-2

23 se

rum

leve

ls (A

U)

ns

(a)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 0043

le10 gt10

APACHE-II

(b)

No YesType 2 diabetes

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

(c)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

BMI (kgm2)

le30 gt30

(d)

Figure 2 miR-223 serum levels in critically ill patients and healthy controls (a) Serum concentrations of miR-223 were determined by qPCRin RNA extracts from serum of patients at admission to the intensive care unit (ICU) miR-223 levels were unchanged in critically ill patients(119899 = 221) compared with healthy controls (119899 = 75) (b) SerummiR-223 concentrations at admission to the ICU were significantly changed incritically ill patients with high initial Acute Physiology and Chronic Health Evaluation (APACHE) II scores (gt10) in comparison to patientswith low APACHE-II scores (le10) (c) Serum miR-223 concentrations at admission to the ICU are unchanged in patients with or withouttype 2 diabetes mellitus (d) SerummiR-223 levels at admission to the medical ICU are independent of the presence of obesity in critically illpatients Box plots are displayed where the bold line indicates the median per group the box represents 50 of the values and the horizontallines show minimum and maximum values of the calculated nonoutlier values asterisks and open circles indicate outlier values

fearful complication of sepsis and sepsis shock syndromeoften leading to death in critically ill patients To deter-mine whether miR-223 serum levels might be indicative forpatientsrsquo prognosis during and after ICU treatment we firstperformed correlation analysis between miR-223 levels and

classicalmarkers of organ dysfunctionWhilemiR-223 serumlevels showed a significant correlation with decreased renalfunction (Table 3) no correlation with acute liver injury or animpaired liver synthesis capacity could be established More-over we found no correlation to classical prognosis scores

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

4 Disease Markersm

iR-2

23 re

lativ

e exp

ress

ion

Control0

2

4

6

LPS

lowastlowast

8h after

(a)

Before CLP

0

02

04

06

08

10

miR

-233

relat

ive e

xpre

ssio

n ns

CLP

0

05

10

15

miR

-233

relat

ive e

xpre

ssio

n

CLP CLP24h after 24h afterBefore

(b)

miR

-223

relat

ive e

xpre

ssio

n

CLP-OPBefore

0

1

2

3

4

5

ns

6h 24h

(c)

Brain

CLP

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

Con

trol

Heart

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

Liver

0

1

2

3

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

0

1

2

3

4

5

ns

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

PBMC

0

1

2

3

ns

Lung

0

10

20

30

40m

iR-2

23 re

lativ

e exp

ress

ion

CLP

Con

trol

lowast

Kidney

0

1

2

3

miR

-223

relat

ive e

xpre

ssio

n

CLP

Con

trol

lowastlowast

Spleen

(d)

Figure 1 miR-223 serum levels in murine models of septic diseases (a) Serum was obtained from C57Bl6j wild-type mice at baseline and8 hours after injection of 25 120583g per gram body weight miR-223 serum levels were measured The bar graphs represent mean plusmn SEM from119899 = 5 animals per group (b and c) Serum was obtained from C57Bl6j wild-type mice at baseline and 8 hours and 24 hours after inductionof polymicrobial sepsis by CLP miR-223 serum levels were measured The bar graphs represent mean plusmn SEM the line graphs display pairedpre- and postoperative values for individual animals (119899 = 14) (d) Relative expression of miR-233 in different organs in mice after SHAM orCLP surgery (kidney spleen liver lung heart brain and peripheral blood mononuclear cells (PBMC)) from C57Bl6j wild-type mice

are displayed as separate points and represent outliers Allvalues including outliers have been included for statisticalanalyses Correlations between variables have been analysedusing the Spearman correlation test Receiver operating char-acteristic (ROC) curve analysis and the derived area underthe curve (AUC) statistic provide a global and standardizedappreciation of the accuracy of a marker or a composite scorefor predicting an event ROC curves were generated by plot-ting sensitivity against 1 minus specificity Kaplan-Meier curveswere plotted to display the impact on survival and between-group differences were assessed using the log-rank testCox regressions were used to identify factors predicting ICUmortality or overallmortality All reported119875 valueswere two-tailed and a 119875 value less than 005 was considered to indicatestatistical significance

3 Results

31 miR-223 SerumLevels inMurineModels of Septic DiseasesBased on the contradictory data on alterations of miR-223serum levels in patients with sepsis we decided to first analysemiR-223 serum concentrations in highly standardizedmousemodels of septic disease Therefore CLP procedures and LPSinjections were performed in C57Bl6 mice As determinedby miRNA-specific qPCR miR-223 levels were moderatelybut significantly elevated 8 h after injection of LPS (Fig-ure 1(a)) In contrast levels of circulating miR-223 remainedunaffected at different time points after induction of sepsisby using the CLP model which closely resembles humansepsis (Figures 1(b) and 1(c)) Finally miR-223 expressionwasanalysed in different organs after induction of sepsis by CLP

Disease Markers 5

surgery to further determine potential mechanism regulatingmiR-223 serum levels in sepsis These analyses revealed asignificant upregulation in the lung while expression of miR-223 was downregulated in kidney and unaffected from theinduction of septic disease in the other organs including liverbrain heart and muscle (Figure 1(d))

32 miR-223 SerumLevels in Critically Ill Patients andHealthyControls We next analyzed levels of circulating miR-223in sera of 221 patients at admission to the ICU as wellas in 75 healthy volunteers miR-223 concentrations wereslightly lower in critically ill patients compared to healthycontrols (119875 = 0141 Figure 2(a)) When we analyzed serumlevels of miR-223 with respect to disease severity we foundsignificantly lower levels in patients with more severe diseaseaccording to higher APACHE-II scores (gt10) compared topatients with lower APACHE-II scores (lt10) (119875 = 0043Figure 2(b))

The metabolic status of patients was shown to influencethe outcome of critically ill patients [17] Since miR-223was shown to be involved in the pathophysiology of type 2diabetes [8] we analyzed potential correlations betweenmiR-223 serum levels and the presence of obesity or type 2diabetes Importantly we found no significant differences incirculating miR-223 between patients with obesity or normalbody weight (Figure 2(c)) and those with or without type 2diabetes mellitus respectively (Figure 2(d)) In addition nodifferences were foundwhen patients were compared by theirage or gender (data not shown)

33 miR-223 Serum Levels Do Not Indicate Sepsis in CriticallyIll Patients Elevated levels of miR-223 were suggested todiscriminate between SIRS and sepsis patients with highsensitivity and specificity [11] Our cohort of ICU patientsfeatured both patients with sepsis (119899 = 137) and patients thatdid not fulfill sepsis criteria (119899 = 84)Thus we further investi-gated the impact of sepsis on miR-223 serum concentrationsin our cohort In these analyses no significant differencesin miR-223 levels between septic and nonseptic patientswere evident (119875 = 0529 Figure 3(a)) Of note the factthat circulating miR-223 is independent of the presence ofsepsis was further substantiated by correlation analysesrevealing that serum miR-223 levels were not correlated toestablished markers of systemic inflammation and bacterialinfection such as C-reactive protein (CRP) procalcitonin(PCT) interleukin-6 (IL-6) interleukin-10 (IL-10) or tumornecrosis factor (TNF) in critically ill patients (Table 3)

In order to investigate the impact of the underlyingetiology of sepsiscritical illness on miR-223 serum levelsmore precisely we again performed subgroup analyses Thecohort of sepsis patients was subdivided into a pulmonaryand a nonpulmonary site of infection and the nonsepsispatients were categorized into liver cirrhosis cardiovasculardisorders and others However also this analysis revealedno differences in miR-223 serum concentration between thedifferent subgroups thus excluding the fact that potentialalterations in miR-223 levels might only exist in a specific

Table 3 Correlations of miR-223 serum levels at ICU admissionwith other laboratory markers

Parameter ICU patients119877 119875

Markers of liver functionCholinesterase 0135 nsProtein 0110 nsAlbumin 0089 nsGGT 0084 nsGLDH 0116 nsAP minus0074 nsINR minus0028 nsBilirubin total minus0014 nsBilirubin direct 0059 nsAST 0091 nsALT 0167 0013

Markers of inflammationC-reactive protein minus0010 nsProcalcitonin minus0085 nsIL-6 0171 nsIL-10 0118 nsTNF-alpha minus0152 nsAmylase 0218 nsLipase 0111 nsLeucocyte counts 0155 0021

Markers of renal functionCreatinine minus0391 lt0001cystatin C minus0384 lt0001cystatin C GFR 0379 lt0001Urea minus0435 lt0001

Others variablesLactate minus0146 0031NT-proCNP minus0432 lt0001suPAR minus0187 0022

Clinical scoringAPACHE-II minus0139 nsSOFA minus0187 0030ICU days minus0051 nsDays alive minus0190 ns119877 correlation coefficient 119875 119875 value 119877 and 119875 values by Spearman rankcorrelation INR international normalized ratio IL-6 interleukin-6 IL-10interleukin-10 TNF tumour necrosis factor APACHE-II Acute Physiologyand Chronic Health Evaluation II SOFA Sequential Organ Failure Assess-ment Score

group of patients and be masked if these patients are mergedwith other patients (Figure 3(b) Table 2)

In summary our analysis reveals that in contrast to pre-vious studies reporting a downregulation [11] or upregulation[12] of miR-223 in serum of patients with septic disease miR-223 levels are only slightly altered in critical illness and sepsisindicating that miR-223 measurements from serum are notsuitable to detect sepsis

34 miR-223 Serum Concentrations Do Not Predict Survivalin Critically Ill Patients Multiple organ failure represents a

6 Disease Markers

Control Patients

0

100

200

300

400

500m

iR-2

23 se

rum

leve

ls (A

U)

ns

(a)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 0043

le10 gt10

APACHE-II

(b)

No YesType 2 diabetes

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

(c)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

BMI (kgm2)

le30 gt30

(d)

Figure 2 miR-223 serum levels in critically ill patients and healthy controls (a) Serum concentrations of miR-223 were determined by qPCRin RNA extracts from serum of patients at admission to the intensive care unit (ICU) miR-223 levels were unchanged in critically ill patients(119899 = 221) compared with healthy controls (119899 = 75) (b) SerummiR-223 concentrations at admission to the ICU were significantly changed incritically ill patients with high initial Acute Physiology and Chronic Health Evaluation (APACHE) II scores (gt10) in comparison to patientswith low APACHE-II scores (le10) (c) Serum miR-223 concentrations at admission to the ICU are unchanged in patients with or withouttype 2 diabetes mellitus (d) SerummiR-223 levels at admission to the medical ICU are independent of the presence of obesity in critically illpatients Box plots are displayed where the bold line indicates the median per group the box represents 50 of the values and the horizontallines show minimum and maximum values of the calculated nonoutlier values asterisks and open circles indicate outlier values

fearful complication of sepsis and sepsis shock syndromeoften leading to death in critically ill patients To deter-mine whether miR-223 serum levels might be indicative forpatientsrsquo prognosis during and after ICU treatment we firstperformed correlation analysis between miR-223 levels and

classicalmarkers of organ dysfunctionWhilemiR-223 serumlevels showed a significant correlation with decreased renalfunction (Table 3) no correlation with acute liver injury or animpaired liver synthesis capacity could be established More-over we found no correlation to classical prognosis scores

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

Disease Markers 5

surgery to further determine potential mechanism regulatingmiR-223 serum levels in sepsis These analyses revealed asignificant upregulation in the lung while expression of miR-223 was downregulated in kidney and unaffected from theinduction of septic disease in the other organs including liverbrain heart and muscle (Figure 1(d))

32 miR-223 SerumLevels in Critically Ill Patients andHealthyControls We next analyzed levels of circulating miR-223in sera of 221 patients at admission to the ICU as wellas in 75 healthy volunteers miR-223 concentrations wereslightly lower in critically ill patients compared to healthycontrols (119875 = 0141 Figure 2(a)) When we analyzed serumlevels of miR-223 with respect to disease severity we foundsignificantly lower levels in patients with more severe diseaseaccording to higher APACHE-II scores (gt10) compared topatients with lower APACHE-II scores (lt10) (119875 = 0043Figure 2(b))

The metabolic status of patients was shown to influencethe outcome of critically ill patients [17] Since miR-223was shown to be involved in the pathophysiology of type 2diabetes [8] we analyzed potential correlations betweenmiR-223 serum levels and the presence of obesity or type 2diabetes Importantly we found no significant differences incirculating miR-223 between patients with obesity or normalbody weight (Figure 2(c)) and those with or without type 2diabetes mellitus respectively (Figure 2(d)) In addition nodifferences were foundwhen patients were compared by theirage or gender (data not shown)

33 miR-223 Serum Levels Do Not Indicate Sepsis in CriticallyIll Patients Elevated levels of miR-223 were suggested todiscriminate between SIRS and sepsis patients with highsensitivity and specificity [11] Our cohort of ICU patientsfeatured both patients with sepsis (119899 = 137) and patients thatdid not fulfill sepsis criteria (119899 = 84)Thus we further investi-gated the impact of sepsis on miR-223 serum concentrationsin our cohort In these analyses no significant differencesin miR-223 levels between septic and nonseptic patientswere evident (119875 = 0529 Figure 3(a)) Of note the factthat circulating miR-223 is independent of the presence ofsepsis was further substantiated by correlation analysesrevealing that serum miR-223 levels were not correlated toestablished markers of systemic inflammation and bacterialinfection such as C-reactive protein (CRP) procalcitonin(PCT) interleukin-6 (IL-6) interleukin-10 (IL-10) or tumornecrosis factor (TNF) in critically ill patients (Table 3)

In order to investigate the impact of the underlyingetiology of sepsiscritical illness on miR-223 serum levelsmore precisely we again performed subgroup analyses Thecohort of sepsis patients was subdivided into a pulmonaryand a nonpulmonary site of infection and the nonsepsispatients were categorized into liver cirrhosis cardiovasculardisorders and others However also this analysis revealedno differences in miR-223 serum concentration between thedifferent subgroups thus excluding the fact that potentialalterations in miR-223 levels might only exist in a specific

Table 3 Correlations of miR-223 serum levels at ICU admissionwith other laboratory markers

Parameter ICU patients119877 119875

Markers of liver functionCholinesterase 0135 nsProtein 0110 nsAlbumin 0089 nsGGT 0084 nsGLDH 0116 nsAP minus0074 nsINR minus0028 nsBilirubin total minus0014 nsBilirubin direct 0059 nsAST 0091 nsALT 0167 0013

Markers of inflammationC-reactive protein minus0010 nsProcalcitonin minus0085 nsIL-6 0171 nsIL-10 0118 nsTNF-alpha minus0152 nsAmylase 0218 nsLipase 0111 nsLeucocyte counts 0155 0021

Markers of renal functionCreatinine minus0391 lt0001cystatin C minus0384 lt0001cystatin C GFR 0379 lt0001Urea minus0435 lt0001

Others variablesLactate minus0146 0031NT-proCNP minus0432 lt0001suPAR minus0187 0022

Clinical scoringAPACHE-II minus0139 nsSOFA minus0187 0030ICU days minus0051 nsDays alive minus0190 ns119877 correlation coefficient 119875 119875 value 119877 and 119875 values by Spearman rankcorrelation INR international normalized ratio IL-6 interleukin-6 IL-10interleukin-10 TNF tumour necrosis factor APACHE-II Acute Physiologyand Chronic Health Evaluation II SOFA Sequential Organ Failure Assess-ment Score

group of patients and be masked if these patients are mergedwith other patients (Figure 3(b) Table 2)

In summary our analysis reveals that in contrast to pre-vious studies reporting a downregulation [11] or upregulation[12] of miR-223 in serum of patients with septic disease miR-223 levels are only slightly altered in critical illness and sepsisindicating that miR-223 measurements from serum are notsuitable to detect sepsis

34 miR-223 Serum Concentrations Do Not Predict Survivalin Critically Ill Patients Multiple organ failure represents a

6 Disease Markers

Control Patients

0

100

200

300

400

500m

iR-2

23 se

rum

leve

ls (A

U)

ns

(a)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 0043

le10 gt10

APACHE-II

(b)

No YesType 2 diabetes

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

(c)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

BMI (kgm2)

le30 gt30

(d)

Figure 2 miR-223 serum levels in critically ill patients and healthy controls (a) Serum concentrations of miR-223 were determined by qPCRin RNA extracts from serum of patients at admission to the intensive care unit (ICU) miR-223 levels were unchanged in critically ill patients(119899 = 221) compared with healthy controls (119899 = 75) (b) SerummiR-223 concentrations at admission to the ICU were significantly changed incritically ill patients with high initial Acute Physiology and Chronic Health Evaluation (APACHE) II scores (gt10) in comparison to patientswith low APACHE-II scores (le10) (c) Serum miR-223 concentrations at admission to the ICU are unchanged in patients with or withouttype 2 diabetes mellitus (d) SerummiR-223 levels at admission to the medical ICU are independent of the presence of obesity in critically illpatients Box plots are displayed where the bold line indicates the median per group the box represents 50 of the values and the horizontallines show minimum and maximum values of the calculated nonoutlier values asterisks and open circles indicate outlier values

fearful complication of sepsis and sepsis shock syndromeoften leading to death in critically ill patients To deter-mine whether miR-223 serum levels might be indicative forpatientsrsquo prognosis during and after ICU treatment we firstperformed correlation analysis between miR-223 levels and

classicalmarkers of organ dysfunctionWhilemiR-223 serumlevels showed a significant correlation with decreased renalfunction (Table 3) no correlation with acute liver injury or animpaired liver synthesis capacity could be established More-over we found no correlation to classical prognosis scores

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

6 Disease Markers

Control Patients

0

100

200

300

400

500m

iR-2

23 se

rum

leve

ls (A

U)

ns

(a)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 0043

le10 gt10

APACHE-II

(b)

No YesType 2 diabetes

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

(c)

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

BMI (kgm2)

le30 gt30

(d)

Figure 2 miR-223 serum levels in critically ill patients and healthy controls (a) Serum concentrations of miR-223 were determined by qPCRin RNA extracts from serum of patients at admission to the intensive care unit (ICU) miR-223 levels were unchanged in critically ill patients(119899 = 221) compared with healthy controls (119899 = 75) (b) SerummiR-223 concentrations at admission to the ICU were significantly changed incritically ill patients with high initial Acute Physiology and Chronic Health Evaluation (APACHE) II scores (gt10) in comparison to patientswith low APACHE-II scores (le10) (c) Serum miR-223 concentrations at admission to the ICU are unchanged in patients with or withouttype 2 diabetes mellitus (d) SerummiR-223 levels at admission to the medical ICU are independent of the presence of obesity in critically illpatients Box plots are displayed where the bold line indicates the median per group the box represents 50 of the values and the horizontallines show minimum and maximum values of the calculated nonoutlier values asterisks and open circles indicate outlier values

fearful complication of sepsis and sepsis shock syndromeoften leading to death in critically ill patients To deter-mine whether miR-223 serum levels might be indicative forpatientsrsquo prognosis during and after ICU treatment we firstperformed correlation analysis between miR-223 levels and

classicalmarkers of organ dysfunctionWhilemiR-223 serumlevels showed a significant correlation with decreased renalfunction (Table 3) no correlation with acute liver injury or animpaired liver synthesis capacity could be established More-over we found no correlation to classical prognosis scores

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

Disease Markers 7

No YesSepsis

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

(a)

Cp

Oth

er

Oth

er

0

50

100

150

200

250

Sepsis No sepsis

miR

-223

seru

m le

vels

(AU

)

Pulm

onar

y

Abdo

min

al

Uro

geni

tal

Cirr

hosis

(b)

Figure 3 miR-223 serum levels do not indicate sepsis in critically ill patients (a) Serum levels of miR-223 were not different in patientsthat fulfilled sepsis criteria (119899 = 157) compared to patients with nonseptic etiology of critical illness (b) miR-223 serum levels did not varybetween the different etiologies of septic or nonseptic disease

or other parameters indicating patientsrsquo outcome on ICUtreatment such as ventilation time or time spent on ICUtreatment (Table 3)

Despite these negative correlation analyses and the factthat miR-223 serum levels did not significantly vary betweencritically ill patients and healthy controls or between septicand nonseptic patients we next analyzed whether they mightbe useful in predicting mortality in critically ill patients Inlinewith the reducedmiR-223 concentrations in patientswithhigh APACHE-II scores (see Figure 2(b)) patients that diedduring the ICU treatment showed lower levels of miR-223levels compared to survivors supporting a role for miR-223in estimating patientsrsquo outcome during ICU treatment (119875 =0010 Figure 4(a)) Kaplan-Meier curve analysis revealed thatpatients with lower miR-223 levels (eg of the lower quartile)demonstrated an impaired ICU survival compared to patientswith miR-223 concentrations within the upper quartile of allpatients however these differences failed to reach statisticalsignificance in Cox regression analyses

In our cohort of critically ill patients 49 died on the ICUwhile additional 45 patients died after release from ICU Incontrast to the data on ICU survival patients that died duringlong-term follow-up showed no differences in miR-223 levelscompared to survivors (119875 = 0386 Figure 4(c)) In line withthat Kaplan-Meier curve analysis revealed no role for miR-223 serummeasurements in determining patientsrsquo long-termprognosis (Figure 4(d)) thus arguing against a strong role ofmiR-223 as a blood based marker for prediction of criticallyill patientsrsquo prognosis

4 Discussion

Despite enormous advances in diagnosis modalities triagingpatients at the emergency room for relocation to the ICU orguiding therapeutic decisions within the first week of ICUtreatment represents a major challenge in the treatment of

critically ill patients However such decisions are of tremen-dous relevance for critically ill patients [18] In this settingmarkers allowing decision about patientsrsquo treatment andclinical course may be of significant benefit Various authorsdemonstrated that commonly used markers like CRP andPCT might represent diagnostic and prognostic biomarkersin this setting especially when used in combination withclinical severity scores or multimarker approaches [19 20]Besides CRP and PCT a variety of different protein-basedmarkers such as suPAR inducible protein 10 (IP10) neu-trophil gelatinase-associated lipocalin (NGAL) natriureticpeptides mature adrenomedullin (ADM) and thrombopoi-etin was tested however none of these experimental markerscould be translated into clinical use [14 21] Several authorstherefore speculated that novel for example miRNA-basedmarkers might perform better and could therefore enterclinical routine

miRNAs have recently been associated with the patho-genesis of systemic inflammation and infection [22] Besideothers it was shown that miR-223 is critically involved in thedifferentiation and maturation of key players of the innateimmune response An increased immune response towardsinfectious agents such as Candida albicans was shown inmiR-223mutantmice (miR-223minusY)Moreover in endotoxin-challenge models miR-223minusY mice demonstrated elevatedtissue destruction thus highlighting a potential role of miR-223 in the pathophysiology of septic diseases and providingevidence for analysis of miR-223 in serum of sepsis patients

miR-223 has recently been proposed as a novel serumbiomarker for sepsis and septic shock disease in two differentAsian populations of sepsis patients Of note results fromthese studies were conflicting on the one hand it wasdemonstrated in a cohort of 116 patients (43 with mild sepsisand 73 with severe sepsisseptic shock) that elevated miR-223levels are indicative of the presence of septic disease [12] andcorrelate with an impaired prognosis in these patients [13]

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

8 Disease Markers

Yes No

0

50

100

150

200

250

miR

-223

seru

m le

vels

(AU

)

P = 001

Yes No

0

50

100

150

200

250 ns

miR

-223

seru

m le

vels

(AU

)

0 20 40 60 80 100Time (days)

00

02

04

06

08

10

miR-223 lower 25

miR-223 upper 25

log rank 309 (ns)

miR-223 25ndash75

Time (days)0 200 400 600 800 1000

00

02

04

06

08

10

log rank 197 (ns)

Ove

rall

surv

ival

miR-223 lower 25

miR-223 upper 25miR-223 25ndash75

(a) (c)

(b) (d)

Ove

rall

ICU

surv

ival

Overall survival

ICU survivor

Figure 4 miR-223 serum concentrations do not predict survival in critically ill patients (a) Patients that died during the course of ICUtreatment had lower miR-223 serum levels on admittance to ICU compared to survivors (119875 = 0010 119880-test) (b) Kaplan-Meier survivalcurves of ICU patients are displayed showing that miR-223 serum levels had no significant (Cox regression) influence on ICU or long-termsurvival in critically ill patients (c) Patients that died during long-term follow-up had similar miR-223 serum concentration on admittanceto ICU compared to surviving patients (119875 = 0386 119880-test) (d) Kaplan-Meier survival curves of ICU patients are displayed showing thatmiR-223 serum levels had no influence on long-term survival in critically ill patients In (a) and (b) box plots are displayed where the boldline indicates the median per group the box represents 50 of the values and the horizontal lines show minimum and maximum values ofthe calculated nonoutlier values

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

Disease Markers 9

On the other hand in a cohort of 80 patients (30 withSIRS and 50 with sepsis) miR-223 levels were decreased inpatients that fulfilled criteria of sepsis compared to healthycontrols [11] In the present studywe analysedmiR-223 serumlevels in a well-defined cohort of 223 critically ill patientsHowever despite the large number of samples analysed wefailed to demonstrate significant alterations in serum miR-223 concentrations in critical illness and sepsis

The conflicting results of previously published results andour current findings might partly be explained by differencesof experimental procedures in the different studies Whilewe used spiked-in RNA (SV40) for normalization of miR-223 serum levels Wang et al used an internal reference genenamely snU6 for normalization in their studies [12 13] Weand other groups recently demonstrated that snU6 might beregulated itself in inflammatory diseases potentially affectingthese previous results Of note we found a trend towardslower levels of miR-223 in critical illness which is in line withthe data of Wang et al [11] who also used spiked-in RNA fornormalization Moreover the differences between the differ-ent studiesmight also be related to the size and characteristicsof the patient cohorts analyzed in the different studies Wereport data from a large consecutively recruited cohort thatcovered a broad spectrum of critically ill patients with regardto severity of illness as reflected by APACHE-II and SOFAscores Of note besides sepsis conflicting results of miR-223serum levels have also been described in patients with HCCor chronic hepatitis B [23ndash25] highlighting the need forfurther efforts in defining standards for sample preparationdata normalization and data analysis in this setting

We recently found a concordant regulation of miR-133ain the serum of patients with sepsis and mice after inductionof septic disease [9] In the present study we demonstratethat miR-223 levels remained unchanged in mice after CLPsurgery thus reflecting the situation in patients In contrastto the results from CLP procedure we found elevated serumlevels of miR-223 inmice after LPS treatment which is in linewith previous reports These on the first view contradictoryresults might be explained by differences between the twomodels LPS administration induces systemic inflammationthat mimics some of the initial clinical features of sepsis(such as increases in proinflammatory cytokines) but doesnot feature bacteremia representing the sepsis defining eventin humandisease [26ndash28]Moreover LPS causesmuch earlierand higher peak levels of cytokine expression compared withlevels observed in human sepsis [27 29 30] ConsequentlyLPS-induced murine sepsis fails to mimic the differentimmunological stages of human sepsis a proinflammatoryphase and a compensatory anti-inflammatory phase In con-trast CLP-induced sepsis increased lymphocyte apoptosiswhich mimics immunosuppression at the later phase ofhuman sepsis [31ndash33] In this respect CLP-induced sepsis iscompletely different from LPS-induced sepsis and moreclosely mimics human sepsis [34 35]

In summary miR-223 serum levels of critically ill andsepsis patients are not significantly regulated compared tohealthy controls and are only modestly associated with dis-ease severity or outcomeOur data thus strongly argue againsta potential use of miR-223 as a blood based biomarker forseptic disease

Disclosure

The funders had no role in study design data collection andanalysis decision to publish or preparation of the paper

Conflict of Interests

The authors declare that they have no competing interests

Authorsrsquo Contribution

Fabian Benz Frank Tacke Alexander Koch and ChristophRoderburg contributed equally to this paper These authorsshare first or senior authorship respectively

Acknowledgments

The authors would like to express their gratitude to DrJane Beger Michaela Roderburg-Goor and members of theLuedde-Lab for helpful discussionsThis work was supportedby grants from the German Research Foundation (SFB-TRR57P06) a Mildred-Scheel Endowed Professorship and asingle project Grant (110043) from the German Cancer Aid(Deutsche Krebshilfe) an ERC Starting Grant (ERC-2007-Stg208237-Luedde-Med3-Aachen) a grant from the EMBOYoung Investigator Program the Interdisciplinary Centre forClinical Research (IZKF) Aachen Germany the Ernst JungFoundation Hamburg and a grant from the Medical Fac-ulty of the RWTH Aachen University to Tom Luedde andChristoph Roderburg and a project grant from the GermanResearch Foundation to Christoph Roderburg (DFG RO43174-1)

References

[1] D CAngus andT van der Poll ldquoSevere sepsis and septic shockrdquoThe New England Journal of Medicine vol 369 no 9 pp 840ndash851 2013

[2] M Sandquist and H R Wong ldquoBiomarkers of sepsis and theirpotential value in diagnosis prognosis and treatmentrdquo ExpertReview of Clinical Immunology vol 10 no 10 pp 1349ndash13562014

[3] V N Kim ldquoMicroRNA biogenesis coordinated cropping anddicingrdquo Nature Reviews Molecular Cell Biology vol 6 no 5 pp376ndash385 2005

[4] M Lagos-Quintana R Rauhut W Lendeckel and T TuschlldquoIdentification of novel genes coding for small expressedRNAsrdquoScience vol 294 no 5543 pp 853ndash858 2001

[5] G A Calin and C M Croce ldquoMicroRNA signatures in humancancersrdquo Nature Reviews Cancer vol 6 no 11 pp 857ndash8662006

[6] W P Kloosterman and R H A Plasterk ldquoThe diverse functionsof microRNAs in animal development and diseaserdquo Develop-mental Cell vol 11 no 4 pp 441ndash450 2006

[7] V Ambros ldquoThe functions of animal microRNAsrdquo Nature vol431 no 7006 pp 350ndash355 2004

[8] M Haneklaus M Gerlic L A J OrsquoNeill and S L MastersldquoMiR-223 Infection inflammation and cancerrdquo Journal ofInternal Medicine vol 274 no 3 pp 215ndash226 2013

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009

10 Disease Markers

[9] F Tacke C Roderburg F Benz et al ldquoLevels of circulating mir-133a are elevated in sepsis and predict mortality in critically illpatientsrdquo Critical Care Medicine vol 42 no 5 pp 1096ndash11042014

[10] C Roderburg M Luedde D Vargas Cardenas et al ldquoCircu-lating microRNA-150 serum levels predict survival in patientswith critical illness and sepsisrdquo PLoS ONE vol 8 no 1 ArticleID e54612 2013

[11] J-F Wang M-L Yu G Yu et al ldquoSerum miR-146a and miR-223 as potential new biomarkers for sepsisrdquo Biochemical andBiophysical Research Communications vol 394 no 1 pp 184ndash188 2010

[12] H-JWang P-J ZhangW-J Chen D Feng Y-H Jia and L-XXie ldquoFour serummicroRNAs identified as diagnostic biomark-ers of sepsisrdquo The Journal of Trauma and Acute Care Surgeryvol 73 no 4 pp 850ndash854 2012

[13] HWang P ZhangW Chen D Feng Y Jia and L Xie ldquoSerummicroRNA signatures identified by Solexa sequencing predictsepsis patientsrsquo mortality a prospective observational studyrdquoPLoS ONE vol 7 no 6 Article ID e38885 2012

[14] A Koch S Voigt C Kruschinski et al ldquoCirculating solubleurokinase plasminogen activator receptor is stably elevated dur-ing the first week of treatment in the intensive care unit and pre-dicts mortality in critically ill patientsrdquoCritical Care vol 15 no1 article R63 2011

[15] C Roderburg F BenzDVargas Cardenas et al ldquoElevatedmiR-122 serum levels are an independent marker of liver injury ininflammatory diseasesrdquo Liver International 2014

[16] F Benz C Roderburg D Vargas Cardenas et al ldquoU6 is unsuit-able for normalization of serum miRNA levels in patients withsepsis or liver fibrosisrdquo Experimental ampMolecularMedicine vol45 article e42 2013

[17] A Koch E Sanson S Voigt A Helm C Trautwein and FTacke ldquoSerumadiponectin upon admission to the intensive careunit may predict mortality in critically ill patientsrdquo Journal ofCritical Care vol 26 no 2 pp 166ndash174 2011

[18] D F Gaieski M EMikkelsen R A Band et al ldquoImpact of timeto antibiotics on survival in patients with severe sepsis or septicshock in whom early goal-directed therapy was initiated in theemergency departmentrdquo Critical Care Medicine vol 38 no 4pp 1045ndash1053 2010

[19] L Magrini G Gagliano F Travaglino et al ldquoComparisonbetween white blood cell count procalcitonin and C reactiveprotein as diagnostic and prognostic biomarkers of infection orsepsis in patients presenting to emergency departmentrdquoClinicalChemistry and Laboratory Medicine vol 52 no 10 pp 1465ndash1472 2014

[20] M Hur H Kim S Lee et al ldquoDiagnostic and prognosticutilities of multimarkers approach using procalcitonin B-type natriuretic peptide and neutrophil gelatinase-associatedlipocalin in critically ill patients with suspected sepsisrdquo BMCInfectious Diseases vol 14 no 1 article 224 2014

[21] S Di Somma L Magrini F Travaglino et al ldquoOpinion paperon innovative approach of biomarkers for infectious diseasesand sepsis management in the emergency departmentrdquo ClinicalChemistry and Laboratory Medicine vol 51 no 6 pp 1167ndash11752013

[22] E Sonkoly M Stahle and A Pivarcsi ldquoMicroRNAs and immu-nity novel players in the regulation of normal immune functionand inflammationrdquo Seminars in Cancer Biology vol 18 no 2 pp131ndash140 2008

[23] J Xu C Wu X Che et al ldquoCirculating MicroRNAs miR-21miR-122 and miR-223 in patients with hepatocellular carci-noma or chronic hepatitisrdquo Molecular Carcinogenesis vol 50no 2 pp 136ndash142 2011

[24] F Ji B Yang X Peng H Ding H You and P Tien ldquoCirculatingmicroRNAs in hepatitis B virus-infected patientsrdquo Journal ofViral Hepatitis vol 18 no 7 pp e242ndashe251 2011

[25] P Qi S-Q Cheng H Wang N Li Y-F Chen and C-F Gao ldquoSerum microRNAs as biomarkers for hepatocellularcarcinoma in Chinese patients with chronic hepatitis B virusinfectionrdquo PLoS ONE vol 6 no 12 Article ID e28486 2011

[26] K A Wichterman A E Baue and I H Chaudry ldquoSepsis andseptic shockmdasha review of laboratory models and a proposalrdquoJournal of Surgical Research vol 29 no 2 pp 189ndash201 1980

[27] D G Remick D E Newcomb G L Bolgos and D R CallldquoComparison of the mortality and inflammatory response oftwo models of sepsis lipopolysaccharide vs cecal ligation andpuncturerdquo Shock (Augusta Ga) vol 13 no 2 pp 110ndash116 2000

[28] H R Michie K R Manogue D R Spriggs et al ldquoDetection ofcirculating tumor necrosis factor after endotoxin administra-tionrdquoThe New England Journal of Medicine vol 318 no 23 pp1481ndash1486 1988

[29] M K Eskandari G Bolgos C Miller D T Nguyen L EDeForce and D G Remick ldquoAnti-tumor necrosis factor anti-body therapy fails to prevent lethality after cecal ligation andpuncture or endotoxemiardquo Journal of Immunology vol 148 no9 pp 2724ndash2730 1992

[30] P Brandtzaeg P Kierulf P Gaustad J Dobloug T E Mollnesand K Sirnes ldquoSystemic meningococcal disease a modelinfection to study acute endotoxinemia in manrdquo Progress inClinical and Biological Research vol 272 pp 263ndash271 1988

[31] R S Hotchkiss and I E Karl ldquoThe pathophysiology andtreatment of sepsisrdquoThe New England Journal of Medicine vol348 no 2 pp 138ndash150 2003

[32] A Ayala and I H Chaudry ldquoImmune dysfunction in murinepolymicrobial sepsis mediators macrophages lymphocytesand apoptosisrdquo Shock vol 6 supplement 1 pp S27ndashS38 1996

[33] JW Dear H Yasuda X Hu et al ldquoSepsis-induced organ failureis mediated by different pathways in the kidney and liveracute renal failure is dependent on MyD88 but not renal cellapoptosisrdquo Kidney International vol 69 no 5 pp 832ndash8362006

[34] K Doi A Leelahavanichkul P S T Yuen and R A StarldquoAnimal models of sepsis and sepsis-induced kidney injuryrdquoJournal of Clinical Investigation vol 119 no 10 pp 2868ndash28782009

[35] A Dyson and M Singer ldquoAnimal models of sepsis why doespreclinical efficacy fail to translate to the clinical settingrdquoCritical Care Medicine vol 37 no 1 pp S30ndashS37 2009