Journal of Critical Care (2012) 27, 523.e1–523.e9

The Sequential Organ Failure Assessment scoreand copeptin for predicting survival inventilator-associated pneumonia☆,☆☆

Lucas Boeck MDa, Philippe Eggimann MDb, Nicholas Smyrnios MDc, Hans Pargger MDd,Nehal Thakkar MDc, Martin Siegemund MDd, Nils G. Morgenthaler MDe,Janko Rakic MDa, Michael Tamm MDa, Daiana Stolz MDa,⁎

aClinic of Pulmonary Medicine and Respiratory Cell Research, University Hospital, CH-4031 Basel, SwitzerlandbDepartment of Adult Critical Care Medicine, University Hospital, CH-1011 Lausanne, SwitzerlandcDivision of Pulmonary, Allergy and Critical Care Medicine, UMass Memorial Medical Center, 01655-Worcester, MA, USAdDepartment of Anesthesiology and Surgical Intensive Care Medicine, University Hospital, CH-4031 Basel, SwitzerlandeResearch Department Brahms AG, Hennigsdorf/Berlin, Germany 16761

thGthm

d

0h

Keywords:Biomarker;Prognosis;Infection

AbstractIntroduction: Ventilator-associated pneumonia remains the most common nosocomial infection inthe critically ill and contributes to significant morbidity. Eventual decisions regarding withdrawal ormaximal therapy are demanding and rely on physicians' experience. Additional objective tools forrisk assessment may improve medical judgement. Copeptin, reflecting vasopressin release, as well asthe Sequential Organ Failure Assessment (SOFA) score, reflecting the individual degree of organdysfunction, might qualify for survival prediction in ventilator-associated pneumonia. Weinvestigated the predictive value of the SOFA score and copeptin in ventilator-associated pneumonia.Methods: One hundred one patients with ventilator-associated pneumonia were prospectivelyassessed. Death within 28 days after ventilator-associated pneumonia onset was the primary endpoint.Results: The SOFA score and the copeptin levels at ventilator-associated pneumonia onset weresignificantly elevated in nonsurvivors (P = .002 and P = .017, respectively). Both markers haddifferent time courses in survivors and nonsurvivors (P b .001 and P = .006). Mean SOFA (average

☆ Competing interests: Daiana Stolz was supported by grants from the Swiss National Foundation (PP00P3_128412/1), the Liechtenstein Foundation, ande “Freiwillige Akademische Gesellschaft Basel.” Nehal Thakkar was supported by the Will Rogers Foundation. Funding obtained from Brahms (Hennigsdorf,ermany) was used for assay material and logistics. The sponsors of this investigator-initiated project had no involvement in the design and conduct of the study;e collection, management, analysis, and interpretation of the data; or in the preparation, review, and approval of the manuscript or decision to submit theanuscript. The remaining authors declare no conflict of interests.☆☆ Authors' contribution: Design of study: MT, DS; acquisition of data: PE, HP, MS, SM, JR, DS; statistical analysis: LB; interpretation of data: LB, DS;

rafting of the manuscript: LB, DS; critical revision of the manuscript: PE, HP, MS, NM, SM.⁎ Corresponding author. Tel.: +41 61 265 5184; fax: +41 61 265 4785.E-mail address: dstolz@uhbs.ch (D. Stolz).

883-9441/$ – see front matter © 2012 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.jcrc.2011.07.081

523.e2 L. Boeck et al.

SOFA of 10 days after VAP onset) was superior in predicting 28-day survival as compared withSOFA and copeptin at ventilator-associated pneumonia onset (area under the curve, 0.90 vs 0.73 and0.67, respectively).Conclusions: The predictive value of serial-measured SOFA significantly exceeds those of singleSOFA and copeptin measurements. Serial SOFA scores accurately predict outcome in ventilator-associated pneumonia.© 2012 Elsevier Inc. All rights reserved.

1. Introduction

Annually, more than 300 000 critically ill patientsrequire mechanical ventilation in the United States [1].Despite continuous optimization of preventive measures,ventilator-associated pneumonia (VAP) incidence remainshigh [2-5]. Several trials showed an association of VAPwith morbidity, mortality, and prolonged hospital stay [2].Disease-related costs were estimated to outrange US$40 000 per patient [6]. For humanitarian, ethical, andeconomic reasons, activities have to be concentrated onpatients who most likely benefit. Accordingly, activitiesmay be restricted in patients who would have prolongedsuffering or who would be left with an unacceptablequality of life [7]. Furthermore, individualized therapy inregard to disease severity/outcome is promising [8,9]. Inthis context, differentiating between patients with good andpoor outcome is crucial.

Several clinical severity scores such as the AcutePhysiology and Chronic Health Evaluation score andSimplified Acute Physiology Score have been establishedand shown to predict survival [10]. However, most aredesigned for intensive care unit (ICU) admission andignore factors that influence patient outcome during thecourse of ICU stay. The Sequential Organ FailureAssessment (SOFA) score was originally designed as atool for describing the severity of organ dysfunction [11].Accordingly, initial and serial SOFA assessments wereassociated with outcome [12,13]. Although easily calcula-ble, the performance of SOFA at ICU admission is similaras compared with more sophisticated measures [12]. Incontrast to other scores, SOFA is suitable to evaluatechanges in the patient status over time [13]. Importantly,scores were not evaluated for risk stratification ofindividual patients.

Recently, biomarkers have been suggested for predict-ing outcome in critically ill patients. Copeptin, a stablefragment of the vasopressin precursor, is released bymultiple physiological and pathological stimuli such asexercise, pain, hypoglycemia, hypoxia, acidosis, shock,heart disease, stroke, and infection [14-18]. A predictivevalue of copeptin was suggested in several infections andcritical care [19-22].

We investigated the performance of copeptin and theSOFA score for predicting survival in VAP.

2. Materials and methods

2.1. Setting and study population

Predefined secondary end points from a prospectivemulticentric trail including 101 patients were analyzed [23].The study took place in 7 medical and surgical ICUs (UMassMemorial Medical Center, Worcester, MA; UniversityHospital Lausanne, Switzerland; and University HospitalBasel, Switzerland). The trial was approved by theinstitutional review boards of all participating institutionsand registered in the Current Controlled Trials Database as“ProVAP”-Study (ISRCTN61015974). Written informedconsent was obtained from all included patients or theirlegal representatives.

2.2. Diagnostic criteria

Ventilator-associated pneumonia was diagnosed clinical-ly according to the American Thoracic Society guidelines[24,25]. It was defined as a new or progressive infiltrate onchest radiography associated with at least 2 of the following:purulent tracheal secretions; fever (body temperatureN38°C/100.4°F); leukocytosis/leukopenia (leukocyte countN11 000/μL or b3000/μL). Critically ill patients wereeligible for the study if they met all the following criteria:(1) clinically diagnosed VAP, (2) prior mechanical ventila-tion for at least 48 hours, and (3) older than 18 years. Patientswere excluded if they (1) were pregnant, (2) were enrolled inanother trial, (3) had received immunosuppressants or long-term corticosteroid therapy (N0.5 mg/kg per day for N1month), (4) were immunosuppressed, or (5) had a coexistingextrapulmonary infection diagnosed within the first 3 daysand requiring antibiotic therapy for more than 3 days.Microbiologically confirmed VAP was defined by asignificant growth of quantitative cultures of endotrachealaspirates (EA), bronchoalveolar lavage (BAL), or protectedspecimen brush (PSB) specimens [24].

2.3. Baseline assessment and follow-up

At the time of enrolment, the following information wasrecorded from each subject: age, sex, preexisting comorbid-ities, severity of the underlying medical condition(s),

523.e3SOFA and copeptin to predict survival in VAP

primary reason for initiating mechanical ventilation,duration of prior mechanical ventilation, antibiotic usewithin 14 days of VAP onset, body temperature, heart rate,mean arterial pressure (MAP), oxygen saturation, ratio ofpartial pressure of arterial oxygen to the fraction ofinspired oxygen (PaO2/FIO2), leukocyte count (WBC), andcopeptin serum levels. The following indices werecalculated: SOFA score, Organ Dysfunction and/orInfection (ODIN) score, and Clinical Pulmonary InfectionScore (CPIS). During the 28-day follow-up period, thefollowing information was recorded: body temperature,heart rate, MAP, oxygen saturation, PaO2/FIO2, WBC,SOFA, ODIN, CPIS, mechanical ventilation status,antibiotic use, and survival throughout the 28-day studyperiod. The SOFA and copeptin measurements wereassessed on VAP onset (day 0) and daily until day 10.The mean SOFA/copeptin is the calculated arithmeticmean of SOFA/copeptin values comprising a particularnumber of days after VAP onset. If not stated otherwise, itis the mean of the SOFA/copeptin values assessed within10 days after VAP onset. At VAP onset, 1 SOFA and 4copeptin measurements were missing.

2.4. Outcome assessment

Follow-up was performed for 28 days or until death,whichever occurred first. Patients alive at day 28 wereclassified as survivors; all others were classified asnonsurvivors. No patient was lost to follow-up.

2.5. Copeptin measurements

Copeptin (C-terminal proAVP) was measured with 50 μLserum using a sandwich immunoluminometric assay using 2polyclonal antibodies to amino acids 132 to 164 ofpreprovasopressin (CT-proAVP LIA; BRAHMS AG, Hen-ningsdorf/Berlin, Germany) [14]. Since the initial publica-tion, the assay was modified as follows: the capture antibodywas replaced by a murine monoclonal antibody directed toamino acids 137 to 144 (GPAGAL) of proAVP. Thismodification improved the sensitivity of the assay. Thelower detection limit was 0.4 pmol/L, and the functionalassay sensitivity (b20% interassay coefficient of variation)was less than 1 pmol/L [26]. A median copeptin level of4.2 pmol/L (95% confidence interval [CI], 4.0-4.4 pmol/L)was measured in healthy individuals [14].

2.6. Statistical analyses

Discrete variables are expressed as counts (percentages);and continuous variables, as median (interquartile range[IQR]). Comparability of groups was analysed by χ2 test,Fisher exact test, or Mann-Whitney U test, as appropriate.Correlation analyses were performed using Spearman rankcorrelation coefficient (ρ). To detect the time course of the

biomarkers across survivors and nonsurvivors, a linearmixed-effect model with fixed factors day and group andrandom factor subject was performed on the log-trans-formed parameters. To study possible different timecourses, the interaction between day and group wasdetermined. The relation of predictors with survival wasinvestigated using logistic regression models. Time todeath was analysed by Kaplan-Meier survival curves andcompared by the log-rank test. Area under the curve(AUC) values of receiver operating characteristic analyses,a measure of discrimination, were calculated. Combinedparameters were entered in a nonlinear logistic regressionmodel where each predictor is modeled as a 5-knot cubicspline [27]. The AUC values were reported withcorresponding 95% CIs. All tests were 2-tailed; P b .05was defined as significant. Data were analyzed usingstatistical software (Statistical Package for Social Sciences,version 16 for Windows; SPSS, Chicago, IL).

3. Results

A total of 101 patients with a median [IQR] age of 57 [43-70] years were included in the study. Baseline characteristicsare summarized in Table 1. A causative organism wasidentified in 74 patients (76%). The most frequently isolatedbacteria were Staphylococcus aureus (30%), Pseudomonasaeruginosa (25%), and Klebsiella species (13%). Within 28days after VAP onset, 20 patients died. Deaths were due totraumatic brain injury/subarachnoid hemorrhage (8), respi-ratory failure/acute respiratory distress syndrome (5), septicshock (3), cardiogenic shock (2), multiorgan failure (1), andacute liver failure (1).

3.1. Predictive value of SOFA in VAP

At VAP onset, the median [IQR] SOFA score was 7[6.0-9.8]. Cardiac comorbidity was associated with anelevated SOFA score (P = .03). Age and pulmonary,renal, and hematologic/oncologic comorbidities did notdiffer according to score values. The SOFA wassignificantly different between survivors and nonsurvivors(median [IQR], 6 [5-9] vs 9 [7-14]; P = .002). There wasa significant interaction between day and group, reflectinga different decrease of SOFA, in survivors and non-survivors (P b .001, Fig. 1A). The SOFA ratio of VAPonset and day 3 (SOFA day 3/SOFA VAP onset) differedsignificantly in survivors and nonsurvivors (0.80 [0.54-1.00] vs 0.93 [0.88-1.29]; P = .024). The Kaplan-Meiersurvival was significantly different across the SOFAquartiles (log rank: SOFA at VAP onset P = .048,mean SOFA P b .001; Fig. 2A,B). Regression analyseswere performed using 32 variables including age, sex,comorbidities, vital parameters, complications, ODIN,CPIS, SOFA, and copeptin (Table 2). In multivariate

Table 1 Baseline characteristics of 101 VAP patients

TotalN = 101

Survivorsn = 81

Nonsurvivorsn = 20

P value

Sex, male 74 (74) 61 (76) 13 (65) .459Age, y 57 [43-70] 55 [42-68] 67 [52-75] .033Admission categoryMedical 53 (53) 41 (51) 12 (60) .615Elective surgery 4 (4) 4 (5) 0 (0) .582Emergency surgery 43 (43) 35 (43) 8 (40) .994ComorbiditiesCardiac 48 (48) 37 (46) 11 (55) .619Pulmonary 19 (19) 14 (17) 5 (25) .523Renal 16 (16) 11 (14) 5 (25) .302Hematologic/oncologic 8 (8) 4 (5) 4 (20) .047Reason for mechanical ventilationCardiovascular failure 32 (32) 25 (31) 7 (35) .930Acute respiratory failure 53 (53) 42 (52) 11 (55) .998Trauma 33 (33) 28 (35) 5 (25) .582Neurologic failure 30 (30) 26 (32) 4 (20) .431Sepsis 25 (25) 22 (27) 3 (15) .387Miscellaneous 10 (10) 9 (11) 1 (5) .682Vital parameters at VAP onsetHeart rate, per min 90 [79-103] 90 [80-110] 88 [76-95] .344MAP, mm Hg 77 [70-90] 77 [70-90] 75 [68-85] .436Temperature, °C 38.0 [37.4-38.6] 38.1 [37.6-38.8] 37.4 [37.0-38.3] .010PaO2/Fio2 188 [135-253] 192 [140-254] 173 [126-225] .290SaO2, % 97 [94-99] 97 [94-99] 98 [96-99] .468WBC, 109/L 11.6 [8.0-15.0] 11.2 [8.0-14.6] 13.3 [9.3-18.0] .209Biomarker at VAP onsetCopeptin 31.0 (15.2-54.7) 29.7 (15.2-47.6) 50.6 (16.9-89.4) .017Clinical scores at VAP onsetODIN score 2 [1-3] 2 [1-2] 3 [1-4] .050SOFA score 7.0 [6.0-9.8] 6 [5-9] 9 [7-14] .002CPIS 7.5 [6.0-9.0] 8.0 [6.0-9.0] 7.0 [6.0-8.0] .799MicrobiologyPositive microbiological cultures (EA, BAL, PSB) 74 (76) 58 (73) 16 (89) .226Positive blood cultures 34 (34) 29 (36) 5 (25) .515

Values are given as numbers (%) or median [IQR]. Fever was defined as a body temperature greater than 38°C/100.4°F; leukocytosis/leukopenia, as aleukocyte count higher than 11 000/μL or lower than 3000/μL. SaO2 indicates oxygen saturation. Bold stands for statistically significant results.

523.e4 L. Boeck et al.

logistic regression, including at most 2 parameters, meanSOFA and hematologic/oncologic comorbidities remainedthe only independent predictors of 28-day survival. TheAUC was 0.73 (95% CI, 0.60-0.86) for SOFA at VAPonset and 0.90 (95% CI, 0.80-0.99; Fig. 3A) for meanSOFA. The AUC value of SOFA at VAP onset forpredicting survival continuously increased, by includingadditional SOFA days (Fig. 3A). Restricting the popula-tion to microbiologically confirmed VAP patients (n = 74)did not alter survival prediction (AUC for SOFA at VAPonset, 0.71; mean SOFA, 0.90). Fig. 4 displays thepredictive values for survivors and nonsurvivors lessthan/greater than specific mean SOFA thresholds. A meanSOFA less than 7 was associated with a sensitivity of81% and a positive predictive value of 97% for survival.In contrast, a mean SOFA greater than 13 had a

sensitivity of 20% and a positive predictive value of100% for death.

3.2. Predictive value of copeptin in VAP

Median [IQR] copeptin at VAP onset was 31.0 [15.2-54.7] pmol/L and was significantly elevated in nonsurvi-vors as compared with survivors (50.6 pmol/L [16.9-89.4]vs 29.7 pmol/L [15.2-47.6]; P = .017). Age, MAP, as wellas cardiac, pulmonary, and renal comorbidity wereassociated with copeptin levels. Patients with sepsis hada significantly elevated copeptin at VAP onset (37 [30-66]pmol/L vs 25 [12-53] pmol/L; P = .032). Mean arterialpressure correlated negatively with copeptin (Spearmanrank correlation coefficient −0.299; P = .003) at VAPonset. The trend of copeptin levels over time differed

16

14

12

10

8

6

4

2

0

survivors non-survivors

survivors non-survivors

150

125

100

75

50

25

0day0 2 3 4 5 6 7 8 9 10 day0 2 3 4 5 6 7 8 9 10

day0 2 3 4 5 6 7 8 9 10 day0 2 3 4 5 6 7 8 9 10

P < .001

P = .006

A

B

Fig. 1 Trend of SOFA score (A) and copeptin (in picomoles per liter; B) during the first 10 days after VAP onset in survivors (n = 81) andnonsurvivors (n = 20).

523.e5SOFA and copeptin to predict survival in VAP

significantly among survivors and nonsurvivors (P = .006;Fig. 1B). However, the copeptin ratio of VAP onset andday 3 (copeptin day 3/copeptin VAP onset) was similar insurvivors and nonsurvivors (P = .51). Copeptin quartiles atVAP onset did not differ, whereas mean copeptin quartilesdiffered significantly in outcome (log-rank: P = .079 vsP = .003; Fig. 2C/D). Patients with mean copeptin levels

in the highest quartile (N41.5 pmol/L) had a more than40% chance of death, whereas patients in the lowerquartiles had a risk not to survive of between 8% and 16%.The AUC of copeptin at VAP onset was 0.67 (95% CI,0.53-0.81), and that of mean copeptin was 0.70 (95% CI,0.57-0.84; Fig. 3B). The AUC values in the subgroup ofmicrobiologically confirmed VAP patients did not differ

1.0

0.8

0.6

0.4

0.2

0.0

0 5 10 15 20 25 30

prob

abili

ty o

f sur

viva

l

time (days)

1.0

0.8

0.6

0.4

0.2

0.0

0 5 10 15 20 25 30

prob

abili

ty o

f sur

viva

l

time (days)

1.0

0.8

0.6

0.4

0.2

0.0

0 5 10 15 20 25 30

prob

abili

ty o

f sur

viva

l

time (days)

1.0

0.8

0.6

0.4

0.2

0.0

0 5 10 15 20 25 30

prob

abili

ty o

f sur

viva

l

time (days)

under 66 - 77 - 9.75over 9.75 P = .048 log rank

under 3.73.7 - 5.75.7 - 8.0over 8.0

P < .001 log rank

under 15.015.1 - 31.531.6 - 55.0over 55.1 P = .079 log rank

under 13.914.0 - 23.023.1 - 41.4over 41.5 P = .003 log rank

A B

C D

Fig. 2 Kaplan-Meier estimates of survival within 28 days after VAP onset. Quartiles of SOFA at VAP onset (A), mean SOFA (B), copeptinat VAP onset (in picomoles per liter; C), and mean copeptin (in picomoles per liter; D) are compared.

523.e6 L. Boeck et al.

from those in the general study population (0.67 and0.72, respectively).

3.3. Combination of SOFA and copeptin in VAP

Predictive properties could not be improved by includingboth SOFA and copeptin in analysis. The AUC values of thecombination SOFA-copeptin at VAP onset (0.73; 95% CI,0.60-0.87) and of the combination SOFA mean–copeptinmean (0.89; 95% CI, 0.80-0.99) were not superior ascompared with the AUC value of SOFA.

4. Discussion

We report 3 important findings. The SOFA score andcopeptin values at VAP onset differ significantly in survivorsand nonsurvivors. Single SOFA or copeptin measurementshave only a moderate prognostic accuracy regarding survivalin VAP. And most importantly, serial SOFA assessmentscontinuously improve predictive performance over time.

Organ failure is a continuous deviation from the normalphysiological state of health that may evolve to recovery orworsen until death. The assessments of SOFA and copeptin

Table 2 Univariate logistic regression for predicting death

Oddsratio

95% CI P value

Age, y/10 1.44 1.03-2.01 .035Sex, male 0.58 0.20-1.66 .308Coexisting illnessesCardiac 1.45 0.54-3.89 .456Pulmonary 1.60 0.50-5.11 .432Renal 2.12 0.64-7.01 .218Hematologic/oncologic 4.81 1.09-21.29 .038Antibiotics within 14 dbefore VAP onset

2.11 0.56-7.92 .267

Vital parameters at VAP onsetHeart rate, per min/10 0.87 0.67-1.13 .305MAP, mm Hg/10 0.86 0.62-1.20 .376Temperature, °C 0.48 0.27-0.83 .009PaO2/Fio2, /10 0.99 0.94-1.04 .657SaO2,, % 1.12 0.92-1.36 .260WBC, 109/L 1.08 1.00-1.16 .059Biomarker at VAP onsetCopeptin at VAP onset/10 1.07 0.99-1.16 .086Mean copeptin/10 1.17 1.02-1.23 .023Clinical scores at VAP onsetODIN score 1.91 1.15-3.17 .013CPIS 0.99 0.76-1.29 .950SOFA at VAP onset 1.28 1.10-1.49 .001Mean SOFA 1.61 1.29-2.00 b.001MicrobiologyPositive microbiologicalcultures (EA, BAL, PSB)

2.90 0.61-13.68 .179

Positive blood cultures 0.75 0.08-7.13 .802

The odds ratio represents the change in odds of being in one category ofoutcome (nonsurvivor) when the value of predictor increases by 1 unit.Bold stands for statistically significant results.

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.2

0.0

0.0 0.2 0.4

Sen

sitiv

ity

1 - Specificity

SOFA day0 (AUC 0.73)mean SOFA day0-day2 (0.81)mean SOFA day0-day5 (0.86)mean SOFA day0-day10 (0.90)

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.2

0.0

0.0 0.2 0.4

Sen

sitiv

ity

1 - Specificity

Copeptin day0 (AUC 0.67)mean Copeptin within day10 (0.70)

A

B

Fig. 3 Receiver operating characteristic curves of SOFA (A)and copeptin (B) representing a single measurement at VAP onsetor a calculated mean comprising a particular number of days afterVAP onset.

523.e7SOFA and copeptin to predict survival in VAP

are 2 different approaches estimating organ failure andtherefore survival.

Herein we confirm that the SOFA score assessed on theday of VAP onset is higher in nonsurvivors with an AUCvalue for predicting survival within the range of thosepreviously reported (AUC, 0.61-0.88) [12,28]. Furthermore,SOFA remained elevated in nonsurvivors throughout theICU stay. In general, it was shown that, in ICU patients,serial assessments of SOFA add information to a singlemeasurement [13]. We could demonstrate that serial SOFAmeasurements continuously improve SOFA from a limited toan excellent predictive tool in VAP. Merely the combinationof the first 2 days has shown a clear improvement in survivalprediction as compared with a single assessment at VAPonset. Including 5 or 10 days into the model furtherimproved predictive properties and provided remarkableprognostic information. We used the mean of dailymeasurements, starting at VAP onset. The mean SOFAintegrates the trend of organ dysfunction and minimizes theinfluence of variable daily measurements. Noteworthy, serialmeasurements delay risk evaluation and miss patients whoare discharged or deceased at an earlier stage. Compared

with the SOFA scores reported in general ICU patients, weassessed higher SOFA values [13]. However, it is known thatinfection per se accounts for a greater degree of organdysfunction [29]. Thus, cutoff values need to be adjusted fordifferent groups of critically ill patients.

In our VAP cohort, we could identify patients with lessthan/equal to a mean SOFA of 7 to be at very low risk fordying, whereas patients with greater than/equal to a meanSOFA of 13 had a particularly poor outcome. Cutoff valueswere reached early in the course of disease and might havemajor clinical implications.

Sur

vivo

rs

Non

-sur

v.

100 97 91 89 86

n = 50 n = 68 n = 74 n = 81 n = 87

55

n = 11

56

n = 9

71

n = 7

100

n = 4

value predicting survival in % value predicting death in %

Patients below mean SOFA Patients above mean SOFA

6 7 8 9 10 10 11 12 13

Fig. 4 Proportions of survivors and nonsurvivors with scores greater than/less than a mean SOFA threshold. Patients with scores less than amean SOFA of 7 survived in 99%. In contrast, 100% of patients with scores greater than a mean SOFA of 13 died.

523.e8 L. Boeck et al.

In this prospective observational study, we demonstratethat copeptin at VAP onset is significantly elevated innonsurvivors and moderately predicts survival. Ourfindings are in line with those of Seligman et al [20],who revealed similar predictive properties. The arginine-vasopressin system, as mirrored by copeptin, is influencedby many physiological and pathophysiological stimuli.Particularly in severe diseases, the association betweenplasma osmolality and vasopressin is lost; and vasopressinis released as a result of changes in blood pressure andstimulation of the pituitary stress response [16]. Inaddition, changes in pulmonary gas exchange andinflammatory mediators cause alterations in the arginine-vasopressin system [30]. Copeptin, a vasopressin precursorfragment, reflects vasopressin release favorably. Recentdata suggest a predictive value of copeptin in neurological,cardiac, and pulmonary disease [18,21,31]. Furthermore,copeptin was related to ??COPD prognosis, pneumoniaseverity, sepsis, and shock [19,21,22,32,33]. Therefore, weexpected copeptin to predict survival successfully. Unfor-tunately, because of the complexity of vasopressin effectsin VAP, the difference of copeptin levels betweensurvivors and nonsurvivors was small. Clear cutoff valuescould not be defined. Moreover, copeptin did not improveSOFA's performance. Thus, copeptin might insufficientlyreflect risk in VAP.

A marker of outcome should combine at least 3 mainfeatures. The marker needs to be elevated in disease ascompared with healthy conditions. The degree of anyorgan dysfunction has to be associated with the marker'sheight. And finally, the marker should reflect the overallstate of health. Copeptin seems to achieve the first 2 stepsunder many circumstances. However, weighting of severalorgan dysfunctions is limited. Predictive strength possiblygets lost in the heterogeneous picture of VAP. In contrast,mean SOFA clearly reflects the overall health in VAPpatients. We suggest that serial SOFA is superior tocopeptin in merging diverse morbidities in the intensivecare population.

Some limitations of our study need to be mentioned. First,we investigated prespecified secondary end points. Conse-quently, our results need to be considered as hypothesisgenerating. Second, an independent validation set was notincluded; and the small sample size might have altereddiscrimination and calibration. In particular, the number ofevents was small and significantly decreased within the daysof observation. Third, no score was established to determinethe prognosis of individual patients; and the scores in thisstudy were assessed at VAP onset in contrast to ICUadmission [34]. However, a good performance is supposable;and alternative objective options to assess individual risk arelargely absent [35]. Fourth, clinical diagnosis of VAP has alimited specificity [2,36]. Noteworthy, only the clinicalcriteria suggested by the American Thoracic Society arepracticable and achieve the obligation of a high sensitivity[24,25]. Moreover, restricting the population to microbio-logically confirmed cases did not alter the results. In anycase, certain pathogens might have modified biomarkerlevels and therefore risk stratification. Noteworthy, adjust-ment for P aeruginosa did not alter results (data not shown).

5. Conclusions

We conclude that SOFA and copeptin predict survivalat VAP onset. Whereas copeptin's predictive value isstable over time, the performance of the SOFA scoremarkedly increases.

Instruments such as serial SOFA assessments mightreduce prognostic uncertainties concerning survival in VAPand thus improve quality of decisions at the ICU.

Acknowledgments

We thank Anja Meyer, Stephen Heard, Jaclyn Longtine,and Karin Longtine for data supply and most helpful supportduring the study and Andy Schötzau for statistical advice.

523.e9SOFA and copeptin to predict survival in VAP

References

[1] Kahn JM, Goss CH, Heagerty PJ, Kramer AA, O'Brien CR, RubenfeldGD. Hospital volume and the outcomes of mechanical ventilation. NEngl J Med 2006;355(1):41-50.

[2] Chastre J, Fagon JY. Ventilator-associated pneumonia. Am J RespirCrit Care Med 2002;165(7):867-903.

[3] Cook DJ, Walter SD, Cook RJ, Griffith LE, Guyatt GH, Leasa D, et al.Incidence of and risk factors for ventilator-associated pneumonia incritically ill patients. Ann Intern Med 1998;129(6):433-40.

[4] Langer M, Mosconi P, Cigada M, Mandelli M. Long-term respiratorysupport and risk of pneumonia in critically ill patients. IntensiveCareUnitGroup of Infection Control. Am Rev Respir Dis 1989;140(2):302-5.

[5] Terragni PP, Antonelli M, Fumagalli R, Faggiano C, Berardino M,Pallavicini FB, et al. Early vs late tracheotomy for prevention ofpneumonia in mechanically ventilated adult ICU patients: a random-ized controlled trial. JAMA 2010;303(15):1483-9.

[6] Rello J, Ollendorf DA, Oster G, Vera-Llonch M, Bellm L, Redman R,et al. Epidemiology and outcomes of ventilator-associated pneumoniain a large US database. Chest 2002;122(6):2115-21.

[7] Murray LS, Teasdale GM, Murray GD, Jennett B, Miller JD, PickardJD, et al. Does prediction of outcome alter patient management?Lancet 1993;341(8859):1487-91.

[8] Eichacker PQ, Parent C, Kalil A, Esposito C, Cui X, Banks SM, et al.Risk and the efficacy of antiinflammatory agents: retrospective andconfirmatory studies of sepsis. Am J Respir Crit Care Med 2002;166(9):1197-205.

[9] Angus DC, Laterre PF, Helterbrand J, Ely EW, Ball DE, Garg R, et al.The effect of drotrecogin alfa (activated) on long-term survival aftersevere sepsis. Crit Care Med 2004;32(11):2199-206.

[10] Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: aseverity of disease classification system. Crit Care Med 1985;13(10):818-29.

[11] Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, BruiningH, et al. The SOFA (Sepsis-related Organ Failure Assessment) score todescribe organ dysfunction/failure. On behalf of the Working Groupon Sepsis-Related Problems of the European Society of Intensive CareMedicine. Intensive Care Med 1996;22(7):707-10.

[12] Minne L, Abu-Hanna A, de Jonge E. Evaluation of SOFA-basedmodels for predicting mortality in the ICU: a systematic review. CritCare 2008;12(6):R161.

[13] Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL. Serial evaluationof the SOFA score to predict outcome in critically ill patients. JAMA2001;286(14):1754-8.

[14] Morgenthaler NG, Struck J, Alonso C, Bergmann A. Assay for themeasurement of copeptin, a stable peptide derived from the precursorof vasopressin. Clin Chem 2006;52(1):112-9.

[15] Holmes CL, Patel BM, Russell JA,Walley KR. Physiology of vasopressinrelevant to management of septic shock. Chest 2001;120(3):989-1002.

[16] Morgenthaler NG, Struck J, Jochberger S, Dunser MW. Copeptin:clinical use of a new biomarker. Trends Endocrinol Metab 2008;19(2):43-9.

[17] Reichlin T, Hochholzer W, Stelzig C, Laule K, Freidank H,Morgenthaler NG, et al. Incremental value of copeptin for rapid ruleout of acute myocardial infarction. J AmColl Cardiol 2009;54(1):60-8.

[18] Katan M, Fluri F, Morgenthaler NG, Schuetz P, Zweifel C, BingisserR, et al. Copeptin: a novel, independent prognostic marker in patientswith ischemic stroke. Ann Neurol 2009;66(6):799-808.

[19] Muller B, Morgenthaler N, Stolz D, Schuetz P, Muller C, Bingisser R,et al. Circulating levels of copeptin, a novel biomarker, in lowerrespiratory tract infections. Eur J Clin Invest 2007;37(2):145-52.

[20] Seligman R, Papassotiriou J, Morgenthaler NG, Meisner M, TeixeiraPJ. Copeptin, a novel prognostic biomarker in ventilator-associatedpneumonia. Crit Care 2008;12(1):R11.

[21] Stolz D, Christ-Crain M, Morgenthaler NG, Leuppi J, Miedinger D,Bingisser R, et al. Copeptin, C-reactive protein, and procalcitonin asprognostic biomarkers in acute exacerbation of COPD. Chest 2007;131(4):1058-67.

[22] Jochberger S, Velik-Salchner C, Mayr VD, Luckner G, Wenzel V,Falkensammer G, et al. The vasopressin and copeptin response inpatients with vasodilatory shock after cardiac surgery: a prospec-tive, controlled study. Intensive Care Med 2009;35(3):489-97.

[23] Stolz D, Smyrnios N, Eggimann P, Pargger H, Thakkar N,Siegemund M, et al. Procalcitonin for reduced antibiotic exposurein ventilator associated pneumonia—a randomized study. Eur RespirJ 2009.

[24] Guidelines for the management of adults with hospital-acquired,ventilator-associated, and healthcare-associated pneumonia. Am JRespir Crit Care Med 2005;171(4):388-416.

[25] Fabregas N, Ewig S, Torres A, El-Ebiary M, Ramirez J, de LaBellacasa JP, et al. Clinical diagnosis of ventilator associatedpneumonia revisited: comparative validation using immediate post-mortem lung biopsies. Thorax 1999;54(10):867-73.

[26] Fenske W, Stork S, Blechschmidt A, Maier SG, Morgenthaler NG,Allolio B. Copeptin in the differential diagnosis of hyponatremia. JClin Endocrinol Metab 2009;94(1):123-9.

[27] Harrell FE. Regression modeling strategies: with applications to linearmodels, logistic regression, and survival analysis. New York:Springer; 2001.

[28] Gursel G, Demirtas S. Value of APACHE II, SOFA and CPIS scores inpredicting prognosis in patients with ventilator-associated pneumonia.Respiration 2006;73(4):503-8.

[29] Vincent JL, de Mendonca A, Cantraine F, Moreno R, Takala J, SuterPM, et al. Use of the SOFA score to assess the incidence of organdysfunction/failure in intensive care units: results of a multicenter,prospective study. Working group on “sepsis-related problems” of theEuropean Society of Intensive Care Medicine. Crit Care Med1998;26(11):1793-800.

[30] Chikanza IC, Petrou P, Chrousos G. Perturbations of argininevasopressin secretion during inflammatory stress. Pathophysiologicimplications. Ann N Y Acad Sci 2000;917:825-34.

[31] Stoiser B, Mortl D, Hulsmann M, Berger R, Struck J, MorgenthalerNG, et al. Copeptin, a fragment of the vasopressin precursor, as a novelpredictor of outcome in heart failure. Eur J Clin Invest 2006;36(11):771-8.

[32] Kruger S, Papassotiriou J, Marre R, Richter K, Schumann C, vonBaum H, et al. Pro-atrial natriuretic peptide and pro-vasopressin topredict severity and prognosis in community-acquired pneumonia:results from the German competence network CAPNETZ. IntensiveCare Med 2007;33(12):2069-78.

[33] Morgenthaler NG, Muller B, Struck J, Bergmann A, Redl H, Christ-Crain M. Copeptin, a stable peptide of the arginine vasopressinprecursor, is elevated in hemorrhagic and septic shock. Shock2007;28(2):219-26.

[34] Vincent JL, Opal SM, Marshall JC. Ten reasons why we should NOTuse severity scores as entry criteria for clinical trials or in our treatmentdecisions. Crit Care Med 2010;38(1):283-7.

[35] Cohen NH. The real reasons not to rely on severity scores. Crit CareMed 2010;38(1):334-5.

[36] Fagon JY, Chastre J, Wolff M, Gervais C, Parer-Aubas S, Stephan F,et al. Invasive and noninvasive strategies for management of suspectedventilator-associated pneumonia. A randomized trial. Ann Intern Med2000;132(8):621-30.