American Journal of Emergency Medicine (2013) 31, 215-221

Original Contribution

Proadrenomedullin, a useful tool for risk stratification in high pneumonia severity index score community acquired pneumonia^?

Caroline Courtais PharmD ^a, Nils Kuster PharmD ^a, Anne-Marie Dupuy MD, PhD ^a, Margit Folschveiller MD ^b, Riad Jreige MD ^b, Anne-Sophie Bargnoux MD, PhD ^a, Julie Guiot PharmD ^a, Sophie Lefebvre PhD ^b, Jean-Paul Cristol MD, PhD ^a,?,

Mustapha Sebbane MD, PhD ^b

^aDepartment of Biochemistry, Lapeyronie Hospital, 191 Avenue du Doyen Gaston Giraud, 34295 Montpellier cedex 5, France

^bAdult Emergency Department, Lapeyronie University Hospital, Montpellier, France

Received 20 June 2012; revised 13 July 2012; accepted 14 July 2012

Abstract The aim of the present study was, first, to evaluate the prognostic value of mid-regional proadrenomedullin (proADM) in emergency department (ED) patients with a diagnosis of Community acquired pneumonia and, second, to analyze the added value of proADM as a risk stratification tool in comparison with other biomarkers and clinical severity scores.

We evaluated proADM, C-reactive protein and procalcitonin, along with the pneumonia severity index score in consecutive CAP patients. Ability to predict 30-day mortality was assessed using receiver operating characteristic curve analysis, logistic regression, and reclassification metrics for all patients and for patients with high PSI scores. Primary outcome was death within 30 days after ED admission.

One hundred nine patients were included (median age [interquartile range] 71 [27] years). Nine patients died within 30 days. A significant correlation between proADM and PSI was found (? = 0.584, P b .001). PSI and proADM levels were significantly predictive of risk of death. In patients with PSI class IV and V (score N90), proADM levels significantly predicted risk of death (OR [95% CI], 4.681 (1.661-20.221), P = .012) whereas PSI score did not (P = .122). ROC_AUC (area under the receiver operating characteristic curve) was higher for proADM than for PSI score (ROC_AUC [95% CI], 0.810 [0.654-0.965] and 0.669 [0.445-0.893] respectively). Reclassification analysis revealed that combina- tion of PSI and proADM allows a better risk assessment than PSI alone (P = .001).

MR-proADM may be helpful in individual risk stratification of CAP patients with a high PSI score in the ED, allowing to a better identification of patients at risk of death.

(C) 2013

^? Source of support: reagents for proadrenomedullin assays were kindly provided by BRAHMS Laboratories.

* Corresponding author. Tel.: +33 4 67 33 83 14; fax: +33 4 67 33 83 93.

E-mail address: [email protected] (J.-P. Cristol).

Introduction

The lower respiratory tract infections are a major Public health problem because they remain the third leading cause of death worldwide [1] including European

0735-6757/$ – see front matter (C) 2013 http://dx.doi.org/10.1016/j.ajem.2012.07.017

countries (15% of mortality in a French study [2]). In adults, the incidence varies among countries from 1.6 to 11 per 1000 adults, with rates of hospitalization between 40% and 60% [1].

Diagnosis and severity assessment of community ac- quired pneumonia (CAP) are difficult and largely depend on the clinician’s experience since they are based on clinical and radiological arguments [3].

In this context, the risk stratification is a key issue in clinical management of CAP in the emergency department (ED) to choose the most appropriate care setting: outpatient care or admission to a medical ward or to an intensive care unit. The decision regarding hospitalization relies on clinical stability, risk of complications and death (data based on physical examination, laboratory findings and active co-morbidities) and the presence of other medical or social problems that could interfere with an Optimal care.

Four Prognostic scores could be used to evaluate the severity of CAP: the Pneumonia severity index , the CURB-65 (an acronym for each of the risk factors assessed, ie, Confusion, Urea, Respiratory Rate, Blood pressure and age 65 or older) according to the British Thoracic Society’s rules, and the CRB-65 which is a simplified version and the American Thoracic Society’s score [4]. The most widely used prediction score is the PSI based on demographic factors (age, life in nursing homes), comorbidities, clinical examinations, radiologic, and bio- logic data. Indeed, it has a higher Prognostic accuracy [5,6]. However, because of the effect of age in the computation, PSI score tends to overestimate risk in Elderly people [7].

Therefore, a growing interest has emerged in search for biomarkers allowing improvement of risk stratification alone or in combination with clinical prognostic scores. Previous studies have reported the stable mid-region proadrenomedullin (proADM) as a promising marker [8-10]. proADM directly reftects levels of active peptide adrenomedullin (ADM). ADM has multiple tissue sites of action and pluripotent function including vasodilatory, antimicrobial and anti-inftammatory activities [11]. Two main mechanisms can explain the increase of circulating proADM in infections, including CAP. Firstly, ADM is a member of the calcitonin gene family that is extensively synthesized during infections. Secondarily, a decreased clearance by the kidneys and the lung may partly contribute to the increase in ADM plasma levels observed in CAP [12-15].

In this study, we seek to confirm the potential prognostic role of proADM in CAP by comparison with well-known biomarkers (C-reactive protein [CRP] and procalcitonin [PCT]) in a prospective cohort of ED patients and further demonstrate its potential value for reassignment of CAP patients to risk categories that better reftected their final outcome thanks to net reclassification index (NRI) measure.

Methods

Study design and setting

We conducted a single center and prospective study from June 2009 to July 2010, in an urban-based university hospital ED with a census of 55000 admissions a year.

Patients were consecutive ED patients with a diagnosis of CAP, based on clinical and radiological findings. proADM, CRP, and PCT were measured along with other standard biological parameters.

Primary end point was all causes of death within 30 days. The main objective of this study was to assess the prognostic value of proADM as a risk marker in ED patients with CAP. Secondarily, we aimed to determine whether proADM could improve risk stratification in patients presenting with a high PSI score.

Study was approved by the local ethics committee. Written informed consent was obtained from all patients. Samples collected for this study were registered at the higher education and research ministry Ministere de l’Enseignement Superieur et de la Recherche with the number DC-2009- 1052, according to French law.

Study population and intervention

Eligible subjects were adult patients with a diagnosis of CAP based on clinical examination (including fever, respiratory difficulty and auscultation crackles…) and radiologic (lung opacity) findings.

Non-inclusion criteria were suspicion of hospital-ac- quired pneumonia (or discharged from a hospital within the prior 10 days), previous episode of pneumonia within the past 30 days, Aspiration pneumonia, Immunocompromised patients, pregnancy.

Patient’s management was left to the discretion of the attending ED physician, blinded to proADM results, accord- ing to standard management of CAP in the ED. All patients had a venous blood sample drawn upon arrival for further Laboratory analysis. The PSI score was calculated for each patient. PSI score stratifies patients into five risk categories. Patients in class I, II, and III have a 30-day mortality risk lower than 2.8%, patients in class IV and V have a risk between 9.3% and 31% [19]. All patients hospital records, including ED diagnosis were independently reviewed by a senior ED physician, blinded to the results of proADM.

Patients were followed-up for 30 days with a medical phone check-up at thirtieth day after hospitalization, according to standard management. Cure was defined as resolution of clinical signs of CAP.

Data collection and processing

Patient’s history, clinical examination as well as radio- logical and laboratory results and disposition were prospec- tively collected from the patient’s ED records.

Measurement of proADM and other laboratory parameters

proADM and procalcitonin were measured respectively in EDTA plasma and sera of all patients by a time-resolved amplified cryptate emission technology assay (Kryptor MR- proADM and Kryptor PCT; BRAHMS AG, Hennigsdorf, Germany).C-reactive protein was measured with an immuno- turbidimetric assay (OLYMPUS, Rungis, France) on heparin plasma samples.

Analytical performances of Kryptor MR-proADM assay were carried out according to the Valtec protocol [16]. Three levels of proADM controls were made and stored at 4?C: level 1 (0.34 nmol/L), corresponding to a normal level according to the reference range [8] (0.33 +- 0.7 nmol/L), level 2 (0.51 nmol/L) and level 3 (1.31 nmol/L), close to the optimal decision threshold (1.8 nmol/L) [8]. Within-day imprecision was determined through twenty replicated analyses of each control. Between-day imprecision was assessed on twenty consecutive days using the same controls leading to twenty measures for each level. Intra-assay coefficients of variation were 5.67%, 3.42%, and 1.26 % and inter-assay coefficients of variation were 7.18, 5.80 and 1.97% for levels 1, 2, and 3, respectively.

Statistical analysis

Univariate descriptive statistics are expressed as median (interquartile range [IQR]), unless otherwise indicated. Correlations were assessed using Spearman’s rank correla- tion test. Comparison between 2 groups was performed using Mann-Whitney U-Wilcoxon nonparametric test for contin- uous data and Fisher’s exact test for categorical data.

Logistic regression was used in order to determine ability of sex, age, CRP, PCT, proADM level, and PSI score to efficiently predict risk of death within 30 days. Biological marker values were transformed on a logarithmic scale in order to normalize distribution.

Performance of proADM, Clinical score or association of both to predict outcome was assessed using area under the receiver operating characteristic curve (ROC_AUC). 95% Confidence intervals (95% CI) of AUC were calculated according to DeLong’s method [17]. Comparison between ROC_AUC was performed through DeLong’s test for corre- lated ROC curves.

Added usefulness of proADM, in comparison to clinical score alone for risk stratification, was also assessed through reclassification tables metrics [18]. Therefore, net reclassi- fication improvement with categories (NRI) was computed, using the estimated risk of death based on the PSI score [19] as risk limits for reclassification.

For individuals who died within 30 days after hospital- ization, risk classification was considered improved if individuals moved to a higher risk category with the addition of proADM and worsened if the individuals moved to a lower one. For individuals still alive at 30th day after

hospitalization, the converse is true. In patients who died within 30 days after ED admission, the difference in the proportion of individuals moving up and down a category was calculated and in patients still alive at 30th day; the proportion of individuals moving down minus the proportion moving up a category was calculated.

Because a PSI score higher than 90 (PSI class IV and V) is a strong argument for hospitalization and adverse event including mortality risk, statistical analysis was performed both for the total population and for the subgroup of patients with a high PSI score.

Significance level was set to P b .05. All analyses were performed using R 2.11 (R Foundation for Statistical Computing, Vienna, Austria). Packages ROCR and pROC were used to analyze ROC curves.

Results

Patient characteristics

115 consecutive ED patients were prospectively included in the study. Six patients were excluded for missing data (3 patients were lost to follow-up, and proADM was not tested in 3 patients), leaving 109 patients for analysis. CRP and/or PCT levels were not available for 7 additional patients.

Baseline characteristics of the study cohort are summarized in Table 1. Median (IQR) age of patients was 71 years (27). Twenty-four patients (22%) were classified into PSI class II, 24 (22%) to class III, 38 (34.9%) to class IV and 23 (21.1%) to class V. Seven (7%) patients were discharged home with ambulatory care, 102 (94%) were hospitalized, among which 24 (22%) patients were admitted to intensive care units. Nine (8.3%) patients died within 30 days after ED admission.

Performances of biomarkers in risk assessment of 30-day mortality

As shown in Fig. 1, significant correlation between proADM and PSI was found (? = 0.584, P b .001) whereas

Table 1 Patients’ characteristics

Variable	Alive, n = 100	Dead, n = 9	P
Median age, y (IQR)	70.5 (53-79)	78 (59-84)	.224
Gender, n (%)			.158
Female	37 (37%)	1 (11.1%)
Male	63 (63%)	8 (88.9%)
Median CRP,	145.3	102.4	.334
mg/L (IQR)	(62.25-264.25)	(72.4-211.7)
Median PCT,	0.65 (0.19-3.8)	3.62	.09
ng/mL (IQR)		(0.62-8.08)
Median PSI score (IQR)	97 (73-123)	115 (105-158)	.036
Median Pro-ADM,	1.035	3.017	.003
nmol/L (IQR)	(0.714-1.788)	(2.403-5.787)
Data are shown as median (IQR) or number (%).

Clinical scores“>48% of PSI class IV and V patients had proADM levels greater than 1.8 nmol/L.

12

In patients with high PSI scores (PSI higher than 90), logistic regression revealed that proADM levels significantly predicted 30 day mortality risk whereas PSI score did not (P =

10

.012 and P = .12, respectively, Table 2). ROC_AUC was higher for proADM than for PSI score (ROC_AUC [95% CI], 0.810 [0.654-0.965] and 0.669 (0.445-0.893) respectively) albeit the difference did not reach significance (P = .094, Fig. 2B). Moreover, reclassification analysis (Table 3B) showed that proADM improved risk assessment for 28 patients out of 61, leading to a NRI of 1.016 (P = .001) with both significant events and non-events components (P = .046 and P b .001,

Pro-ADM (nmol/L)

2

4

6

8

respectively).

1-2 3

0

4 5

PSI class

Discussion

Fig. 1 Distribution of proADM values according to the PSI class. Horizontal line represents the 1.8 nmol/L decision threshold proposed by Christ-Crain et al 15.

the relationship between PCT and PSI score did not reach significance (? = 0.180, P = .070). In contrast, no relationship was observed for CRP (? = -0.051, P = .614). In the whole patient cohort, considering mortality within 30 days after inclusion, logistic regression models displayed no significant Predictive power for variables sex, age, CRP or PCT levels. PSI and proADM levels were significantly

predictive of risk of death in univariate analysis (Table 2).

ROC curves for the PSI score alone, proADM, or both combined are shown in Fig. 2A. ROC_AUC (95% CI) were 0.712 (0.537-0.887), 0.803 (0.601-1.000) and 0.804 (0.603-

Beta OR (95% CI) Study population Male gender 1.547 4.698 (0.815-88.922)			P .152
Age a	0.151	1.163	.214
		(0.938-1.527)
CRP b	-0.191	0.826	.415
		(0.525-1.346)
PCT b	0.216	1.241	.070

1.000), respectively. PSI score ROC_AUC was not different

Our results indicate that proADM outperforms other biomarkers, including CRP and PCT in severity assessment of CAP. Furthermore, proADM could improve accuracy of risk stratification, particularly in non-event patients with high PSI scores.

Severity of CAP and clinical scores

Different scores have been developed with the aim to assess pneumonia severity and improve Care decisions, of

Table 2 Univariate logistic regression analysis

Ability of clinical and Biological characteristics of patients to predict death within 30 days after admission.

Odds ratios are shown for a 2-fold increase.

^a For a 5-year increase.

^b Biological marker values were log-transformed (base 2) to normalize distributions before modeling.

from proADM ROC

AUC

(P = .328). In our population,

sensitivity (95% CI) for PSI score higher than 90 or proADM level higher than 1.8 nmol/L was identical (77.8% [40.0- 97.2]). However, proADM had a much higher specificity

(75.0% [65.3-83.1] vs 46.0% [36.0-56.3], P b .001). Reclassification tables for total study population results	PSI score	0.021	(0.986-1.588) 1.021	.026
are shown in Table 3A. Overall, 38 patients out of 109			(1.003-1.041)
were net correctly reclassified, leading to a statistically significant NRI value of 0.784 (P = .005). Individual	proADM b	1.292	3.641 (1.797-9.005)	.001

components of NRI revealed that improvement in classi- fication mainly occurs in non-events (P b .001) rather than in events (P = .102).

PSI	PCT b	0.173	1.188	.207
Score N90			(0.908-1.578)
	PSI score	0.021	1.022	.122
			(0.994-1.051)
	proADM b	1.544	4.681	.012
			(1.661-20.221)

3.3. Usefulness of proADM for patients with high PSI scores

As shown in Fig. 1, 94% of patients with low PSI scores (PSI score lower than 90, that is, PSI classes I, II, III), had proADM levels below 1.8 nmol/L. In contrast, such consistency was not observed in higher PSI classes. Only

A B

PSI, AUC=0.710 ProADM, AUC=0.803

PSI + ProADM, AUC=0.804

PSI, AUC=0.669 ProADM, AUC=0.810

PSI + ProADM, AUC=0.810

Sensitivity

0.6

0.8

1.0

Sensitivity

0.6

0.8

1.0

1.0 0.8 0.6 0.4 0.2 0.0

0.0

0.2

0.4

0.0

0.2

0.4

Specificity

1.0 0.8 0.6 0.4 0.2 0.0

Specificity

Fig. 2 Receiver operating curve analysis of proADM and PSI to predict 30-day mortality risk in all ED patient cohort (A) and in patients with PSI score N90 (B).

which the most important are the PSI, the CURB-65, the CRB-65, and the American Thoracic Society’s score. Since PSI score is the most widely used severity scoring system in CAP [20], we decided to use this score as the reference tool for risk stratification in this study.

However, this score suffers from different drawbacks. Scoring may depend on both clinician’s experience and interpretation [3], as it is based on clinical, biologic, and radiologic arguments, making its use cumbersome in ED [21]. Furthermore, Schuetz et al [10] have observed in a large cohort that PSI as well as CURB-65 scores overestimate risk of death. In our population, more than one out of 2 patients was thereby classified in PSI class IV or V. This over- estimation probably originates, at least partially, from the weight accorded to age. As a consequence, severity scales are characterized by high negative predictive values but low positive predictive values [22], therefore leading to unnec- essary consumption of high cost health care. These limitations of clinical score, along with the necessity to control health expenditure, highlight the need of improving risk stratification in CAP, especially in high-scored patients.

Clinical scores versus biomarkers

In this context, biomarkers, which are simple and objective, may be useful. Among different biomarkers recently proposed [7,8], proADM seems to be one of the most promising predicting factor. In agreement with previous studies [23,24], our results indicate that both proADM and PSI effectively predict risk of death in our CAP population (Fig. 2A) and are superior to inftammatory (CRP) and sepsis (PCT) markers.

Confticting results have been reported from previous studies comparing predictive power of PSI score and proADM, separately or in combination. Some studies tended to prove that ROC_AUC was increased with proADM [8] whereas others did not found such an improvement [9]. These differences could be explained by differences in methodology. First of all, considering Composite end points provides more events and may therefore make it easier to demonstrate an improvement. Furthermore, CAP severity of patients at admission may also strongly inftuence findings. Indeed, our results seem to indicate that proADM could be of particular interest in populations with high PSI scores.

Not surprisingly, we observed a significant correlation

between proADM level and PSI score. PSI class I and II- patients, patients thus had low proADM levels. However, a wide range of proADM levels were observed in higher PSI classes (IV and V). In our population, more than one patient with a PSI score higher than 90 out of two presented a proADM level lower than 1.8 nmol/L. To the best of our knowledge, few studies have analyzed predictive value of proADM according to PSI classes. Huang’s study [9] suggests that measuring proADM is particularly interesting in PSI classes IV and V, leading us to assess its predictive power in patients with high PSI score.

Value of proADM for patients with high PSI scores

When only patients with a PSI score higher than 90 were considered, we found a decrease in ROC_AUC for PSI score, whereas no decrease was found for proADM (Fig. 2B). Furthermore, logistic regression models showed that PSI

PSI	b2.8% 2.8%-9.2%		9.3%-30.9%	N31%	Total		Increased risk	Decreased risk	Net correctly reclassified
Patients with	outcome
b2.8%	0	0	0	0	0
2.8%-9.2%	1	1	2	1	5
9.3%-30.9%	0	0	2	2	4		5	1	4
N31%	0	0	0	0	0
Total	1	1	4	3	9
Patients without outcome
b2.8%	14	1	0	0	15
2.8%-9.2%	27	22	6	0	55
9.3%-30.9%	7	9	11	2	29		9	43	34
N31%	0	0	0	1	1
Total	48	32	17	3	100
						Total	14	45	38
B. Patients with PSI score N90, n = 61

PSI	b2.8%	2.8%-9.2%	9.3%-30.9%	N31%	Total	Increased risk	Decreased risk	Net correctly reclassified
Patients with	outcome
b2.8%	0	0	0	0	0
2.8%-9.2%	0	1	2	0	3
9.3%-30.9%	0	0	2	2	4	4	0	4
N31%	0	0	0	0	0
Total	0	1	4	2	7
Patients without outcome
b2.8%	0	0	0	0	0
2.8%-9.2%	17	5	5	1	28
9.3%-30.9%	7	8	8	2	25	8	32	24
N31%	0	0	0	1	1

score was not a significant predictor of the risk of death in these patients categories (IV and V), whereas proADM still allowed risk assessment.

Table 3 NRI analysis in total population (A) and high-PSI score patients (B), according to 30-day-mortality risks related to PSI classes

A. Total population, n = 109

30-d mortality risk PSI + proADM Reclassified

30-d mortality risk PSI + proADM

Reclassified

Total

24

13

13

4

54

Total 12

32

28

Outcome was death within thirty days after ED admission. b2.8, 2.8-9.2, 9.3-30.9, and N31 are 30-day mortality risks according to PSI classes, expressed in

%. In total population, NRI = 0.784, P = .005. In PSI score N90 patients, NRI = 1.016, P = .001.

recent developments in statistical methods have lead to consider other metrics for assessing usefulness of a new marker. So-called “reclassification metrics” have therefore been proposed. Whereas ROC_AUC consider both events and non events at the same time, the net reclassification improvement method distinguishes events and non events. Total NRI could thus been split into an event-based and a non-event-based statistic. In our entire ED patient cohort, a significant NRI was found, suggesting that classification was improved when proADM level was taken into account. Interestingly, significance was only reached for non-events, but not for events. proADM improvement is thus linked to a more accurate risk assessment in patients who are still alive 30 days after ED admission. When NRI

method was applied to patients with a high PSI score, both events and non-events individual components were found significant.

Dosing proADM in ED patients may be of added value to clinical scores, especially in patients with high PSI scores, allowing to better estimate CAP severity and avoid unnecessary intensive care unit hospitalizations.

As recently suggested, combination of biomarkers and physiological scores significantly improves prediction abilities [25]. Our data could be the starting point for a new decision algorithm based on a combination of proADM and PSI, in CAP patients with high PSI score, as recently proposed by Albrich [26] using combination of proADM and CURB-65. Further large-scale studies including medico-economical data collection are therefore necessary in order to demonstrate cost-effective benefice of proADM.

Limitations

This study has some limitations. First, it was conducted in a single center. Therefore, results are based on CAP patients PSI scores and mortality risk observed in our center (8.3% 30-day mortality in this study) and reftect current manage- ment at our institution, where most patients diagnosed with CAP are referred for hospitalization. Indeed, our study population included 56% of patients (n = 61) with class IV and V PSI scores, who should be hospitalized, according to this risk stratification tool. The remaining class I, II, and III patients could have been considered for outpatient care. However, proADM usefulness has also been demonstrated in higher risk populations [27]. Furthermore, the relatively small number of patients included and low number of events do not allow us to propose a new decision tree. Finally, only admission level of proADM was measured while recent studies have suggested that serial measurement could bring significant information [28].

Conclusions

proADM could thus be a helpful marker, particularly in patients with a high PSI score. Inclusion of this biomarker in decision strategies could provide a more precise evaluation of risk of death, thus avoiding expensive unnecessary hospitalizations.

Acknowledgments

The authors thank BRAHMS Laboratories for kindly providing reagents. The authors also thank both the ED and biochemistry lab staff for their helpful contribution to the study.

References

1. Polverino E, Torres Marti A. Community-acquired pneumonia. Minerva Anestesiol 2011;77:196-211.
2. Vives L, Biel P, Maler G, et al. Acute community-acquired pneumonia of moderate and grave severity investigated by bronchoscopy. Analysis of 193 cases hospitalized in a general hospital. Rev Mal Respir 1996;13:175-82.
3. Fine MJ, Hough LJ, Medsger AR, et al. The hospital admission decision for patients with community-acquired pneumonia. Results from the pneumonia Patient Outcomes Research Team cohort study. Arch Intern Med 1997;157:36-44.
4. Singanayagam A, Chalmers JD, Hill AT. Severity assessment in community-acquired pneumonia: a review. QJM 2009;102:379-88.
5. Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an interna- tional derivation and validation study. Thorax 2003;58:377-82.
6. Lim W, Macfarlane J, Boswell T, et al. Study of community acquired pneumonia aetiology (SCAPA) in adults admitted to hospital: implications for management guidelines. Thorax 2001;56:296-301.
7. Menendez R, Martinez R, Reyes S, et al. Biomarkers improve Mortality prediction by prognostic scales in community-acquired pneumonia. Thorax 2009;64:587-91.
8. Christ-Crain M, Morgenthaler NG, Stolz D, et al. Pro-adrenomedullin to predict severity and outcome in community-acquired pneumonia [ISRCTN04176397]. Crit Care 2006;10:R96.
9. Huang DT, Angus DC, Kellum JA, et al. Midregional proadrenome- dullin as a Prognostic tool in community-acquired pneumonia. Chest 2009;136:823-31.
10. Schuetz P, Wolbers M, Christ-Crain M, et al. Prohormones for prediction of adverse medical outcome in community-acquired pneumonia and lower respiratory tract infections. Crit Care 2010;14: R106.
11. Samson WK. Proadrenomedullin-derived peptides. Front Neuroendo- crinol 1998;19:100-27.
12. Becker KL, Nylen ES, White JC, et al. Procalcitonin and the calcitonin gene family of peptides in inflammation, infection, and sepsis: a journey from calcitonin back to its precursors. J Clin Endocrinol Metabol 2004;89:1512-25.
13. Hirata Y, Mitaka C, Sato K, et al. Increased circulating adrenome- dullin, a novel vasodilatory peptide, in sepsis. J Clin Endocrinol Metabol 1996;81:1449-53.
14. Nishikimi T, Kitamura K, Saito Y, et al. Clinical studies on the sites of production and clearance of circulating adrenomedullin in human subjects. Hypertension 1994;24:600-4.
15. Christ-Crain M, Morgenthaler NG, Struck J, et al. Mid-regional pro- adrenomedullin as a prognostic marker in sepsis: an observational study. Crit Care 2005;9:R816-24.
16. Vassault A, Baud M, Castanier M, et al. Recommendations for a comparison of techniques. Ann Biol Clin 1992;50:727-30.
17. DeLong ER, Vernon WB, Bollinger RR. Sensitivity and specificity of a monitoring test. Biometrics 1985;41:947-58.
18. Pencina M, D’Agostino R, Vasan R. Statistical methods for assessment of added usefulness of new biomarkers. Clin Chem Lab Med 2010;48: 1703-11.
19. Aujesky D, Fine MJ. The Pneumonia Severity Index: a decade after the initial derivation and validation. Clin Infect Dis 2008;47:S133-9.
20. Metlay J, Fine M. Testing strategies in the initial management of patients with community-acquired pneumonia. Ann Intern Med 2003;138:109-18.
21. Krueger P, Loeb M, Kelly C, et al. Assessing, treating and preventing community acquired pneumonia in older adults: findings from a community-wide survey of emergency room and family physicians. BMC Fam Pract 2005;6:32.
22. Loke YK, Kwok CS, Niruban A, et al. Value of severity scales in predicting mortality from community-acquired pneumonia: systematic review and meta-analysis. Thorax 2010;65:884-90.
23. Bello S, Lasierra AB, Minchole E, et al. Prognostic power of proadrenomedullin in community-acquired pneumonia is independent on etiology. Eur Respir J 2011.
24. Kruger S, Ewig S, Giersdorf S, et al. Cardiovascular and inflammatory biomarkers to predict short- and long-term survival in community- acquired pneumonia: results from the German Competence Network, CAPNETZ. Am J Respir Crit Care Med 2010;182:1426-34.
25. Renaud B, Schuetz P, Claessens Y-E, et al. Proadrenomedullin improves REA-ICU score for predicting Early Severe Community-Acquired Pneumonia. Chest. 2012 In Press http://dx.doi.org/10.1378/chest.11- 2574.
26. Albrich WC, Dusemund F, Ruegger K, et al. Enhancement of CURB65 score with proadrenomedullin (CURB65-A) for outcome prediction in lower respiratory tract infections: derivation of a clinical algorithm. BMC Infect Dis 2011;11:112.
27. Suberviola B, Castellanos-Ortega A, Llorca J, et al. Prognostic value of proadrenomedullin in severe sepsis and septic shock patients with community-acquired pneumonia. Swiss Med Wkly 2012;142: w13542.
28. Maisel A, Mueller C, Nowak RM, et al. Midregion prohormone adrenomedullin and prognosis in patients presenting with acute dyspnea: results from the BACH (Biomarkers in Acute Heart Failure) trial. J Am Coll Cardiol 2011;58:1057-67.