Lung DOI 10.1007/s00408-015-9752-4
Prognostic Value of Biomarkers in Acute Non-massive Pulmonary Embolism: A Systematic Review and Meta-analysis Anurag Bajaj1 • Parul Rathor2 • Vishal Sehgal3 • Besher Kabak3 Ajay Shetty3 • Ossama Al Masalmeh1 • Srikanth Hosur4
•
Received: 9 March 2015 / Accepted: 15 June 2015 Ó Springer Science+Business Media New York 2015
Abstract Background Various biomarkers have been evaluated to risk stratify patients with acute pulmonary embolism (PE). We aimed to summarize the available evidence to compare the prognostic value of three most widely studied biomarkers in normotensive patients with acute PE. Method A systematic literature review of database, including Pubmed, EMBASE and Cochrane, was done.
& Anurag Bajaj [email protected] Parul Rathor [email protected] Vishal Sehgal [email protected] Besher Kabak [email protected] Ajay Shetty [email protected] Ossama Al Masalmeh [email protected] Srikanth Hosur [email protected] 1
The Wright Center for Graduate Medical Education, 501 Madison Avenue, Scranton, PA 18510, USA
2
Zhengzhou University, 100 Ke Xue Da Dao, Zhongyuan, Zhengzhou, Henan, China
3
The Common Wealth Medical College, 525 Pine Street, Scranton, PA, USA
4
Critical Care Division, Geisinger Community Medical Center, 1800 Mulberry Street, Scranton, USA
Studies were included if those were done on patients with acute PE and serum troponin or brain natriuretic peptide and N-terminal proBNP (BNP/NT-proBNP) or Heart-type fatty acid-binding protein (H-FABP) assay was done. The primary end point was short-term all-cause mortality. The secondary end points were PE-related mortality and serious adverse events. Results All three biomarkers were significantly associated with increased risk for short-term all-cause mortality, PE-related mortality and serious adverse events. All-cause mortality: troponin [odds ratio (OR) 4.80; 95 % CI 3.25–7.08, I2 = 54 %], BNP or NT-proBNP (OR 7.98; 95 % CI 4.34–14.67, I2 = 0 %); PE-related mortality: troponin (OR 3.80; 95 % CI 2.74–5.27, I2 = 0 %), BNP or NT-proBNP (OR 7.57; 95 % CI 2.89–19.81, I2 = 0 %) and H-FABP (OR 25.97; 95 % CI 6.63–101.66, I2 = 40 %). H-FABP has the lowest negative likelihood ratio (NLR) of 0.17 for mortality followed by high-sensitive cardiac troponin T (hs-cTnT) with NLR of 0.21. Conclusion None of the biomarker identifies a subgroup of patients who can be managed as an outpatient versus patients who may get benefit from thrombolytics with certainty; however, H-FABP and hs-cTnT showed some promising results and should be investigated further. Keywords Pulmonary embolism Prognosis Metaanalysis Mortality Abbreviations PE BNP/ NT-proBNP H-FABP Hs-cTnT PLR
Pulmonary embolism Brain natriuretic peptide/N-terminal brain natriuretic peptide Heart-type fatty acid-binding protein High-sensitive cardiac troponin T Positive likelihood ratio
123
Lung
NLR OR
Negative likelihood ratio Odds ratio
Methods Study Objectives
Introduction Acute pulmonary embolism (PE) is the third leading cause of cardiovascular death [1]. Acute PE has a broad spectrum of clinical manifestations, ranging from patients who die of sudden cardiac death to patients who are literally asymptomatic. The mortality attributable to PE approximates 50 % in patients who are hypotensive at the time of admission (high risk) and current guidelines recommend using thrombolytics in these patients [2, 3]. About 90 % of all acute PE patients are normotensive at the time of admission, but carry very heterogenous short-term mortality, ranging from\1 % to about 15 % [2, 3]. On one side of the spectrum are the patients with very low risk for death that could be managed as an outpatient with new oral anticoagulants and on the other side of the spectrum are the patients with risk for poor outcomes that needs close monitoring and may benefit from more aggressive treatment like thrombolytics. Early risk stratification of the patients with acute PE is very important for optimal management and improves outcomes. It leads to decrease in health care cost and increase in patient satisfaction [4]. Patients who are hemodynamically stable at the time of admission, but manifest right ventricular dysfunction (intermediate risk), have a poor prognosis when compared to patients without right ventricular dysfunction (low risk) [5, 6]. The ability to predict normotensive patients with poor prognosis would help physician to guide treatment according to individual risk. Efforts have been made to predict poor prognosis in patients with pulmonary embolism using clinical as well as serum biomarkers. Echocardiogram is a useful tool in risk stratification in acute PE by diagnosing right ventricular dysfunction, but the positive predictive value of echocardiogram in predicting shortterm mortality is low and is not able to discriminate between patients who can be treated as an outpatient versus patients who need close monitoring [7]. Various biomarkers like troponin (T and I), brain natriuretic peptides (BNP) or N-terminal pro BNP (NT-proBNP) and Heart-type fatty acid-binding (H-FABP) have been evaluated to identify subgroups of patients with poor outcomes [8–47]. These biomarkers are released into the circulation as a consequence of myocardial injury or wall stress in response to elevated right ventricular afterload because of acute PE. However, selection of the most accurate biomarker is still controversial. We aimed to summarize the results of published studies regarding the prognostic role of three principal serum biomarkers in acute PE.
123
The primary objective of this meta-analysis was to summarize the available evidence for the discriminatory ability of prognostic biomarkers in acute PE. We compared the prognostic value of three principal most widely studied serum biomarkers: troponin (T and I), BNP or NT-proBNP and H-FABP in normotensive patients with acute PE. Data Source and Searches A systematic search of Medline, EMBASE and Cochrane Database was performed. Following key words were used, ‘‘Troponin,’’ ‘‘BNP,’’ ‘‘NT-proBNP,’’ ‘‘Heart-type fatty acid binding protein,’’ ‘‘biomarkers’’ and ‘‘Pulmonary embolism.’’ The references of previous meta-analyses, abstracts from annual meetings, web base and review articles were reviewed to identify any additional relevant studies. The retrieved studies were carefully examined to exclude potentially duplicate or overlapping data. No language restriction was enforced. The manuscripts of all retrieved studies cited before December 2014 were reviewed. Abstracts were excluded from the analysis because non-peer reviewed data may introduce bias into the study. Study Eligibility Studies were considered eligible for the analysis if they fulfilled the following criteria. (1) Design: Prospective or retrospective cohort study design with well-defined study population; (2) Population: Patients admitted to the hospital with objectively confirmed acute PE and were normotensive (blood pressure [90 mmHg) at the time of admission; (3) Cardiac specific troponin (T or I), BNP or NT-proBNP or H-FABP assay was done at the time of admission or within 24 h of hospitalization; (4) Outcomes reported: short-term (in-hospital or 30 days) all-cause mortality, PE-related mortality, short-term serious adverse events; (5) 2 9 2 table can be constructed from the data to compute true positive, false positive, true negative, false negative.
Data Extraction and Validity Assessment Two independent reviewers (AB and PR) independently performed the literature search and identify relevant studies. A third investigator was available for arbitration in the event of discordance of the extracted data. The retrieved studies were carefully examined to exclude potentially
Lung
duplicate or overlapping data. Meeting abstracts were excluded from our analysis (Fig. 1). Authors were contacted to get relevant data. Relevant data on study design, year of publication, patient population, inclusion criteria, exclusion criteria, mean age, gender, follow-up period, type of troponin/BNP/NT-proBNP/H-FABP assay, timing of sampling, cut-off of biomarkers used and outcomes were extracted (Tables 1, 2, 3, 4, 5 and 6). In cases where data in several publications were derived from part or all of the same patient series, only the study presenting the largest patient population or most complete dataset was included. Study Endpoints The primary outcome of interest was in-hospital or 30 days all-cause mortality. Secondary outcomes were in-hospital or 30 days mortality resulting from PE and serious adverse events. Definitions Serious adverse events were defined as composite of death, need for thrombolytics, endotracheal intubation, catecholamine infusion for sustained hypotension, cardiopulmonary resuscitation, or recurrent PE. Normotensive pulmonary embolism was defined as patients with acute PE without sustained hypotension (blood pressure more than 90 mmHg) [3]. Data Synthesis and Statistical Analysis We constructed 2 9 2 tables for in-hospital or 30 days mortality and serious adverse events for all biomarkers. We
computed sensitivity, specificity, positive likelihood ratio and negative likelihood ratio separately for all biomarkers by means of true positive, false positive, true negative and false negative. Likelihood ratios do not depend on prevalence rates, and there is general consensus that a positive likelihood ratio of greater than 10 and a negative likelihood ratio of less than 0.1 provide reliable evidence of satisfactory diagnostic performance [48]. A study level analysis was done using Review manager 5.2 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2008) and Meta-disc 1.4. Odds ratio (OR) and 95 % confidence intervals (95 % CI) were used as summary statistics for all outcomes. Studies were evaluated for heterogeneity by visual inspection of the confidence intervals and by means of I2 [I2 = (Q - df)/Q], where Q is the v2 statistic and df is degree of freedom. As a guide, an I2 [ 30 % was considered as an indicator of statistical heterogeneity among the studies. A Mantel–Haenszel fixed effect model was used to calculate the pooled OR for homogeneous end points. Random-effect (DerSimonian) analysis was reported in the presence of significant heterogeneity. Even if there was little or no evidence of heterogeneity, a random-effect model was used for each outcome because of the arguments put forth by many authors for using random-effect models in medical decision-making contexts, especially in the case of rare events. A p value of less than 0.05 was considered significant. Funnel plots were used to evaluate for publication bias by plotting standard error (log OR) of mortality and serious adverse events. Weighted symmetric summary receiver operating characteristic plots was computed for all biomarkers using Moses–Shapiro–Littenberg method [49]. We separately analysed studies that used troponin T and those that used troponin I to predict allcause mortality. We did a combine and separate analysis of BNP and NT-proBNP. We did a subgroup analysis of studies that used high-sensitive cardiac troponin (hs-cTnT), BNP and NT-proBNP to predict all-cause mortality. This meta-analysis was performed in accordance with the ‘‘Meta-Analysis of Observational Studies in Epidemiology: A Proposal for Reporting’’ (MOOSE) [50]. As suggested previously [51], the studies were not scored based on their quality.
Results
Fig. 1 Flow diagram for systematic search. PE pulmonary embolism
The abstracts and titles of 3302 articles were identified for initial review using search strategies as described in Fig. 1. We identified 211 articles to be appropriate in terms of evaluation of prognostic biomarkers in pulmonary embolism. For these articles, full-text articles were obtained. Upon further review, 38 studies were eligible for this metaanalysis. Thirty-seven studies included patients with
123
Lung Table 1 Characteristics of the selected troponin studies Study
Year
Study design
Vuilleumier et al. [9]
2014
Prospective
N
230
Bulj et al. [10]
2013
Prospective
104
Becattini et al. [11]
2013
Prospective
1109
Age (years)
Sex male ( %)
Thrombolytics
Outcomes
Follow-up
75 (69–82)
59.1
NA
30 days SAE
30 days
68.7 ? 13.4
36.5
No
Hospital mortality
In-hospital
69 ? 15
44
Yes
In-hospital
Sanchez et al. [12]
2013
Prospective
517
67 (52–77)
47
No
Hospital mortality/ SAE 30 days SAE
Ozsu et al. [13]
2013
Prospective
121
70 (55–76)
43
Yes
30 days mortality
6 months
Vanni et al. [17]
2011
Prospective
510
NA
43.7
Yes
Hospital mortality/ SAE
In-hospital
Lankeit et al. [14]
2011
Prospective
526
71 (55–79)
51
Yes
30 days mortality/ SAE
6 months
Kang et al. [15] s
2011
Retrospective
173
61 ? 14
44
NA
30 day mortality/ SAE
3 months
Jimenez et al. [16]
2010
Prospective
Lankeit et al. [18]
2010
Prospective
156
591
1165
30 days
74 (65–82)
43
No
30 days mortality
30 days
67 (53–75)
45
Yes
30 days SAE
965 days median
Stein et al. [19]
2010
Retrospective
65 ? 17
45
No
Hospital mortality
In-hospital
Ozsu et al. [20]
2010
Prospective
108
70 (21–90)
44
No
30 days mortality
30 days
Singanayagan et al. [21] Kostrubiec et al. [22]
2010
Retrospective
411
NA
43
No
30 days mortality
2010
Prospective
212
64 ? 18
38
Yes
30 days mortality
30 days/ 6 weeks 30 days
Moores et al. [26]
2009
Prospective
567
NA
43
No
30 days mortality
30 days
Vuilleumier et al. [23]
2009
Prospective
146
72 (21–94)
42
NA
3 months SAE
3 months
Post et al. [24]
2009
Prospective
168
56
45
Yes
30 days mortality
30 days
Bova et al. [25]
2009
Prospective
201
68.2
41
Yes
Hospital mortality/ SAE
3 months
Jimenez et al. [27]
2008
Prospective
318
72 (14)
43
No
30 days mortality
30 days
Bova et al. [33]
2005
Prospective
60
65 (18)
61
Yes
Hospital SAE
3 months
Palmiere et al. [28]
2008
Prospective
89
63
22
Yes
Hospital mortality/HI
In-hospital
Tulevski et al. [29]
2007
Prospective
28
53 ? 18
43
No
90 day mortality
3 months
Kline et al. [30]
2006
Prospective
193
53 ? 17
41
No
Hospital mortality
6 months
Doukeitis et al. [31] Kostrubiec et al. [32]
2005 2005
Prospective Prospective
458 100
62 63 ? 18
43 35
No Yes
3 months mortality 40 days mortality/ SAE
3 months 40 days
Prusczyk et al. [34]
2003
Prospective
64
61 ? 17
53
Yes
Hospital mortality
In-hospital
N number of patients, SAE serious adverse events, NA not available
hypotension or hemodynamic instability at the time of admission and excluded from our analysis. There were ten studies which shared patients with the studies included in our final analysis and were excluded. Fourteen studies did not provide required end points and were excluded.
et al. [32] reported outcomes at 40 days, Vuillieumier et al., Ten Wolde et al. and Doukelitis et al. at 3 months and Kline et al. at 6 months [23, 30–32, 35]. Baseline demographic characteristics of included studies are shown in Table 1, 3 and 5. Mean age of the patients is similar across all the studies and varied from 53 to 72 years.
Selected Studies Troponin Assay Among the included studies, thirty-five studies were prospective cohorts and three were retrospective cohorts as mentioned in Tables 1, 2, 3, 4, 5 and 6. All studies except five reported in-hospital or 30 days event rates. Kostrubiec
123
We included studies for both troponin T and I. Twelve studies used troponin T and eleven studies used troponin I. In one study, information on type of troponin was not
Lung Table 2 Characteristics of the troponin assays Study
Year
Troponin
Assay
Manufacturer
Cut-off for abnormal
Time of sampling
Vuilleumier et al. [9]
2014
hs-cTn T
Electrochemiluminescence
Hoffmann-La Roche
14 pg/ml
Within 24 h
Bulj et al. [10]
2013
cTn T
Electrochemiluminescence
Hoffmann-La Roche
0.1 ng/ml
Admission, 6, 24, 72 h
Becattini et al. [11] Sanchez et al. [12]
2013 2013
cTn I or T cTn I
NA Photometric immunoassay
NA Dade behring
NA 0.1 lg/ml
Within 24 h Admission
Ozsu et al. [13]
2013
hs-cTn T
Electrochemiluminescence
Roche
0.014 ng/ml
Admission
Vanni et al. [17]
2011
cTn I
ADVIA centaur
VITALGmbH
0.15 ng/ml
NA
Lankeit et al. [14]
2011
hs-cTn T
Electrochemiluminescence
Roche
14 pg/ml
Admission
Kang et al. [15]
2011
cTn T
Electrochemiluminescence
Roche
0.2 ng/ml
Admission, 12, 24 h
Jimenez et al. [16]
2011
cTn I
MEIA
Abbott
0.1 ng/ml
within 12 h
Lankeit et al. [18]
2010
hs-cTn T
Electrochemiluminescence
Roche
14 pg/ml
Admission
Stein et al. [19]
2010
cTn I
Chemiluminescence
Beckman coulter
0.5 ng/ml
NA
Advia centaur
1.2 ng/ml
Ozsu et al. [20]
2010
c Tn T
Electrochemiluminescence
Roche
0.027 ng/ml
Admission
Singanayagan et al. [21] Kostrubiec et al. [22]
2010
c Tn I
NA
NA
0.2 lg/l
Within 4 h
2010
cTn T
Electrochemiluminescence
Roche
0.01 lg/l
Admission
c Tn I
Dimension RXL
Dade behring
0.01 lg/l
c Tn I
MEIA
Abbott
0.1 ng/ml
ultra Tn I
Moores et al. [26]
2009
Admission
Vuilleumier et al. [23]
2009
cTn I
Electrochemiluminescence
Beckman coulter
0.09 ng/ml
Admission
Post et al. [24]
2009
c Tn T
Electrochemiluminescence
NA
0.1 ng/ml
Admission, 8 h
Bova et al. [25]
2009
c Tn I
Immunoenzymatic method
Beckmann/Dade Behring
0.07/0.15 ng/ml
Within 12, 24,72 h
Jimenez et al. [27]
2008
cTn I
MEIA
Abbott
0.1 ng/ml
Admission
Bova et al. [33]
2005
c Tn T
NA
NA
0.01 lg/l
NA
Palmiere et al. [28]
2008
cTn I
Immunoassay-testPak
Dade behring
0.1 lg/l
Admission, 24, 48 h
Tulevski et al. [29]
2007
cTn T
NA
NA
0.01 lg/l
Admission
Kline et al. [30] Doukeitis et al. [31]
2006 2005
cTn T cTn I
Electrochemiluminescence Immunofluorescence
Roche Abbott
0.1 ng/l 0.5 lg/l
NA Within 24 h
Kostrubiec et al. [32]
2005
c Tn T
Electrochemiluminescence
Roche
0.01 lg/ml
Admission
Prusczyk et al. [34]
2003
cTn T
Electrochemiluminescence
Roche
0.01 ug/l
Admission, 6, 12, 18 h
MEIA microparticle enzyme immunoassay, NA not available
available and two studies included both troponin T and I. All studies used a predefined cut-off to define elevated versus normal troponin levels except eight studies where the best threshold able to define abnormal levels was derived from receiver operating characteristic (ROC) curve construction [14, 15, 18, 20, 22–24, 32]. Information on type of assay, manufacturer and cut-off is shown in Table 2. BNP or NT-proBNP Assay We included seventeen studies for both BNP and NTproBNP. Eight studies used BNP and nine studies used NTproBNP. Information on type of assay, manufacturer and cut-
off is shown in Table 5. Eight studies used predefined cut-off [12, 25, 29, 30, 38–40, 42] and nine studies defined abnormal cut-off after ROC analysis to determine best threshold to predict complicated PE [9, 14, 18, 20, 23, 32, 35–37]. H-FABP Assay We included six studies in this category. Three studies used predefined cut-off for H-FABP assay [43, 44, 47] and in other three studies, the used cut-off was derived from ROC construction to estimate the best threshold [23, 45, 46]. Information on type of assay, manufacturer and cut-off is shown in Table 6.
123
Lung Table 3 Characteristics of the selected BNP or NT-proBNP studies Study
Year
Design
N
Age (years)
Gender male (%)
Thrombolytics
Outcomes
Follow-up 90 days
Ten Wolde et al. [35]
2003
Prospective
110
58 ± 18
na
No
90 days mortality
Pieralli et al. [36]
2006
Prospective
61
75 ± 14
26
Yes
In-hospital mortality
Inhospital
Kline et al. [30]
2006
Prospective
191
53 ? 17
42
No
6 months SAE
6 months 90 days
Tulevski et al. [29]
2007
Prospective
28
53 ± 18
43
No
90 days mortality
Logeart et al. [37]
2007
Prospective
67
64 ± 16
60
Yes
In-hospital mortality
Inhospital
Bova et al. [25]
2009
Prospective
144
68.2
41
Yes
Hospital mortality/SAE
90 days 30 days
Sanchez et al. [12]
2013
Prospective
521
67 (52–77)
47
No
30 days/SAE
Jimenez et al. [42]
2014
Prospective
848
72 (59–80)
49
Yes
30 days mortality/SAE
30 days
Kostrubiec et al. [32]
2005
Prospective
100
62 ± 18
35
Yes
40 days mortality
40 days
Maziere et al. [38]
2007
Prospective
60
72 ? 15
40
No
In-hospital mortality
Inhospital
Vuilleumier et al. [23] Lankeit et al. [18]
2009 2010
Prospective Prospective
146 156
72 (21–94) 67 (53–75)
42 45
NA Yes
90 SAE 30 days mortality/SAE
90 days 965 median
Ozsu et al. [20]
2010
Prospective
108
70 (21–90)
44
No
30 days mortality
30 days
Klok et al. [39]
2010
Prospective
113
56 ± 17
53
Yes
6 weeks SAE
6 weeks Inhospital
Zondag et al. [40]
2011
Prospective
63
63 (50–73)
29
Yes
In-hospital SAE
Vuilleumier et al. [9]
2014
Prospective
230
75 (69–82)
59.1
NA
30 SAE
30 days
Lankeit et al. [41]
2014
Prospective
688
70 (54–78)
47.4
Yes
30 days SAE
30 days
N number of patients, SAE serious adverse events, NA not available
Outcome Measures Troponin All-cause Mortality: Twenty-three studies reported allcause mortality data (Fig. 2). The overall mortality rate in acute PE was 5.7 %. 274 of 2596 patients with elevated troponin levels and 143 out of 4707 with normal levels died (10.5 vs. 3.1 %). Elevated troponin (T or I) levels were significantly associated with higher risk of death as compared to normal troponin levels (OR 4.80; 95 % CI 3.25–7.08, I2 = 54 %). Excluding three retrospective studies revealed the same results (OR 4.69; 95 % CI 2.95–7.44, I2 = 55 %). Eleven studies used troponin T and ten studies used troponin I. The results were consistent for both troponin T (OR 7.58; 95 % CI 4.55–12.61], I2 = 0 %) and I (OR 3.27; 95 % CI 2.01–5.34, I2 = 65 %). Subgroup analysis of studies using high-sensitive cardiac troponin T (hs-cTnT): Three studies which reported all-cause mortality used hs-cTnT. OR for mortality was 7.86 (95 % CI 2.94–21.00, I2 = 0 %). PE-related mortality (Fig. 2): Fifteen studies reported PE-related mortality. Elevated troponin (T and I) levels were significantly associated with higher risk of PE-related death as compared to normal troponin levels (OR 3.80; 95 % CI 2.74–5.27, I2 = 0 %). Serious adverse events (Fig. 2): Ten studies reported serious adverse events rates. 283 out of 3437 patients had
123
serious adverse events. 176 patients out of 1390 with elevated troponin levels had serious adverse events as compared to 107 out of 2047 with normal troponin levels. Elevated troponin levels were significantly associated with higher risk of serious adverse events as compared to normal troponin levels (OR 3.65; 95 % CI 2.41–5.53, I2 = 47 %). Excluding study by Kostrubiec et al. [32] significantly decreased the heterogeneity to zero (OR 4.09; 95 % CI 3.04–5.51, I2 = 0 %).
BNP or NT-proBNP All-cause mortality (Fig. 3): Ten studies reported data on all-cause mortality. The overall mortality rate in acute PE was 5.5 %. 75 of 808 patients with elevated BNP or NTproBNP levels and 12 out of 765 with normal levels died (9.2 vs. 1.5 %). Elevated BNP or NT-proBNP levels were significantly associated with higher risk of death as compared to normal levels (OR 7.98; 95 % CI 4.34–14.67, I2 = 0 %). The results were consistent for both BNP (OR 6.72; 95 % CI 2.21–20.41], I2 = 0 %) and NT-proBNP (OR 8.59; 95 % CI 4.10–17.97, I2 = 1 %). PE-related mortality (Fig. 3): Nine studies reported PErelated mortality. Elevated BNP or NT-proBNP levels were significantly associated with higher risk of PE-related mortality as compared to normal troponin levels (OR 7.57; 95 % CI 2.89–19.81, I2 = 0 %).
Lung Table 4 Characteristics of the BNP or NT-proBNP assays Study
Assay
Type of assay
Manufacturer
Cutt off for abnormal
Timing of sampling
Ten Wolde et al. [35]
BNP
Immunoradiometric
Shionoria
21.7 pg/ml
Admission
Pieralli et al. [36]
BNP
Fluorescence immunoassay
Biosite
527 ng/ml
First hour within admission
Kline et al. [30]
BNP
Two-site ‘‘sandwich’’ immunoassay and chemiluminescence
Biosite
90 pg/ml
Admission
Tulevski et al. [29]
BNP
Immunoradiometric
Shionoria
35 pg/ml
Admission
Logeart et al. [37]
BNP
Fluorescence immunoassay
Biosite
100 pg/ml
Admission
Bova et al. [25]
BNP
chemiluminescence
Biosite
100 pg/ml
Admission, 12, 24, 48 h
Immunofluorescence
Roche
50 pg/ml
Sanchez et al. [12]
BNP
Electrochemiluminescence
Beckman coulter
100 pg/ml
Jimenez et al. [42]
BNP
MEIA
Abbott
100 pg/ml
Kostrubiec et al. [32]
NT-BNP
Electrochemiluminescence
Roche
600 pg/ml
Maziere et al. [38]
NT-BNP
Electrochemiluminescence
Roche
1000 pg/ml
Admission
Vuilleumier et al. [23]
NT-BNP
Electrochemiluminescence
Roche
300 pg/ml
Admission
Admission Admission
Lankeit et al. [18]
NT-BNP
Electrochemiluminescence
Roche
1000 pg/ml
Admission
Ozsu et al. [20]
NT-BNP
Enzyme immunoassay
Biomedica
300 pmol/ml
Admission
Klok et al. [39]
NT-BNP
Electrochemiluminescence
Roche
600 pg/ml
Admission
Zondag et al. [40]
NT-BNP
Electrochemiluminescence
Roche
500 pg/ml
Admission and every morning
Vuilleumier et al. [9] Lankeit et al. [41]
NT-BNP NT-BNP
Electrochemiluminescence Electrochemiluminescence
Hoffman la roche Roche
300 pg/ml 600 pg/ml
Admission Admission
Table 5 Characteristics of the selected H-FABP studies Study
Year
Study design
N
Age
Gender male (%)
Thrombolytics
Outcome
Follow-up
Gul et al. [47]
2014
Prospective
80
62 ? 17
40
Yes
30 days SAE
30 days
Chen et al. [45]
2013
Prospective
90
61.1 ? 14.6
41
Yes
30 days SAE
30 days
Dellas et al. [46]
2014
Prospective
271
70 (52–76)
44
Yes
30 days SAE
30 days
Vuilleumier et al. [23]
2009
Prospective
146
72 (21–94)
42
Yes
90 days SAE
90 days
Gul et al. [43]
2012
Prospective
61
62 ? 17
37.7
Yes
RVD
30 days
Boscheri et al. [44]
2010
Prospective
101
70 ? 13
43.5
Yes
30 days mortality
30 days
H-FABP heart-type fatty acid-binding protein, N number of patients, SAE serious adverse events, NA not available
Serious adverse events (Fig. 3): Twelve studies reported data on serious adverse events. Elevated BNP or NT-proBNP levels were significantly associated with higher risk of serious adverse events as compared to normal levels (OR 4.66; 95 % CI 3.21–6.77, I2 = 10 %). The results were consistent for both BNP (OR 3.69; 95 % CI 2.51–5.42, I2 = 0 %) and NTproBNP (OR 7.03; 95 % CI 3.67–13.15, I2 = 0 %). H-FABP All-cause mortality: None of the studies reported all-cause mortality.
PE-related mortality (Fig. 4): Four studies reported PErelated mortality. None of the studies reported all-cause mortality. The PE-related mortality in acute PE was 9.3 %. Elevated H-FABP levels were significantly associated with higher risk of PE-related mortality as compared to normal H-FABP levels (OR 25.97; 95 % CI 6.63–101.66, I2 = 40 %). If we exclude the study of Boscheri et al. [44], the heterogeneity becomes zero (OR 14.14; 95 % CI 5.27–37.92, I2 = 0 %). Serious adverse events (Fig. 4): Six studies reported serious adverse events. Elevated H-FABP levels were significantly associated with higher risk of serious adverse
123
Lung Table 6 Characteristics of the H-FABP assays Study
Year
H-FABP Assay
Cut-off for abnormal (ng/ml)
Manufacturer
Time of sampling
Gul et al. [47]
2014
Immunochromatographic
7
HiSens card
Admission
Chen et al. [45]
2013
Sandwich Elisa
6
NA
Admission
Dellas et al. [46]
2014
Elisa/Immunoturbidimetric
6
HyCult
Admission
Vuilleumier et al. [23]
2009
Sandwich Elisa
6
HyCult
Admission
Gul et al. [43]
2012
Immunochromatographic
7
HiSens card
Admission
Boscheri et al. [44]
2010
Sandwich Elisa
7
Rennescens GmbH
Admission
Fig. 2 Forest plot of pooled analysis comparing outcomes between elevated troponin and normal troponin levels in acute Pulmonary embolism. a All-cause mortality; b PE-related mortality; c Serious adverse events. The squares and horizontal lines indicate odds ratio for random-effect model and the 95 % CI for each trial included. The
size of each square is proportional to the statistical weight of a trial. The diamond indicates effect estimate derived from meta-analysis with centre indicates the point estimate and the right and left end indicate 95 % CI
events as compared to normal H-FABP levels (OR 16.81; 95 % CI 5.88–48.09, I2 = 61 %). If we exclude the study of Boscheri et al. and Vuilleumier et al. [23, 44] then heterogeneity becomes zero(OR 16.13; 95 % CI 7.52–34.59, I2 = 0 %).
Publication Bias
123
The funnel plots were created to ascertain publication bias in our study comparing mortality and serious adverse events between elevated biomarker levels and the normal
Lung
Fig. 3 Forest plot of pooled analysis comparing outcomes between elevated BNP or NT-proBNP and normal BNP or NT-proBNP levels in acute Pulmonary embolism. a All-cause mortality; b PE-related mortality; c serious adverse events
levels in acute PE. We drew the funnel plots only for troponin and BNP or NT-proBNP. We did not draw the funnel plots for H-FABP analysis because the numbers of studies were very few. The plots were asymmetrical in our study for all outcomes (all-cause mortality, PE-related mortality and serious adverse events) for both troponin and BNP indicating that there may be a publication bias. It was roughly symmetrical for PE-related mortality in troponin. However, apart from publication bias, it is important to consider other source of selection bias, heterogeneity of studies and methodological limitations of the studies.
Discussion Our meta-analysis indicates that elevated levels of all three biomarkers, including troponin, BNP or NT-proBNP and H-FABP, are associated with increased risk for short-
term mortality as well as serious adverse events. To our knowledge, this is the first ever meta-analysis in which prognostic value of three most studied principal biomarkers is compared simultaneously in normotensive patients with acute PE. There is general consensus that a positive likelihood ratio of greater than 10 and a negative likelihood ratio of less than 0.1 provide reliable evidence of satisfactory prognostic performance. None of the three biomarkers fulfilled the above-mentioned criteria to be able to satisfactorily discriminate between patients with low risk for death versus high risk for death (Table 7). H-FABP, with NLR of 0.17, provides the closest NLR for satisfactory performance to be able to identify patients who are low risk for death and can be treated as an outpatient. However, the number of studies included is very few, only four for mortality and six for serious adverse events. Also, there is some heterogeneity among the studies. These two factors weaken the conclusion.
123
Lung
Fig. 4 Forest plot of pooled analysis comparing outcomes between elevated H-FABP and normal H-FABP levels in acute Pulmonary embolism. a PE-related mortality; b serious adverse events
However, H-FABP has some additional advantage over other biomarkers. H-FABP is a cytoplasmic protein released from cardiac myocytes into the circulation as early as 90 min after PE; therefore, it is useful in early risk stratification when the patients still in the ER [43– 46]. As compared to H-FABP, troponin and BNP or NTproBNP is released into the circulation a little later. Troponin takes at least 6 h to rise so normal troponin levels at the time of admission do not rule out poor outcomes in PE. Also, neither PLR nor NLR is extreme enough to identify patients who have favourable versus unfavourable outcomes. Three of the troponin studies used hs-cTnT for all-cause mortality. NLR of hs-cTnT in predicting all-cause mortality is 0.21. The NLR of hs-cTnT is small as compared to other troponins and could be used to identify patients who can be treated as an outpatient; however, the numbers of studies are very few and results cannot be generalized at this point, more studies are needed to clarify the role of hs-cTnT in acute PE. BNP or NT-proBNP has low NLR of 0.28 for mortality and 0.41 for serious adverse events and it could be a potential biomarker to identify patients with low risk of adverse events; however, the levels of BNP are altered in many clinical conditions like renal failure, elderly age and there is no standardized cut-off to define normal versus abnormal levels. Combining biomarkers like H-FABP with BNP or hs-cTnT or with clinical prognostic tool like sPESI (simplified pulmonary embolism severity index
123
score) may be able to provide us more satisfactory prognostic tool. Combining biomarkers such as Troponin with H-FABP sequentially can increase the combine specificity and may identify patients who need more close monitoring. Similarly, combining biomarkers such as hscTnT with the BNP or H-FABP simultaneously can increase combine sensitivity and may identify patients who can be treated as an outpatient. However, combining biomarkers are beyond the scope of the current study and further studies are needed to clarify their role.
Strengths We did an extensive search for the relevant studies. We included a larger number of studies by searching the grey literature and contacting the authors when data were missing to complete the analysis. Studies were searched and reviewed independently by two authors and clarified by a third author for methods and outcomes. We screened the studies in detail and also contacted author to remove any duplicated or partially duplicated cohorts. We used random-effect models for wider confidence interval and between study and within study variation. We calculated the likelihood ratios which best describe the discriminatory power of the test because it does not depend on prevalence of the disease and it helps clinicians in area of clinical uncertainty [48].
Lung Table 7 Pooled summary results of prognostic value of biomarkers in acute Pulmonary embolism Biomarker
Outcomes
ORs (95 % CI)
Sensitivity
Specificity
PLR
NLR
AUC
Troponin
All-cause mortality
4.80 (3.25–7.08)
0.66 (0.61–0.70)
0.66 (0.65–0.67)
2.13 (1.84–2.47)
0.51 (0.40–0.60)
0.7393
PE-related mortality
3.80 (2.74–5.27)
0.69 (0.61–0.75)
0.62 (0.61–0.64)
2.05 (1.69–2.49)
0.55 (0.45–0.67)
Serious adverse events
3.65 (2.41–5.53)
0.64 (0.58–0.70)
0.64 (0.62–0.65)
2.06 (1.62–2.62)
0.53 (0.39–0.72)
hs-cTn T
All-cause mortality
7.86 (2.94–21.00)
0.92 (0.81–0.98)
0.43 (0.39–0.46).
1.61 (1.42–1.86)
0.21 (0.09–0.51)
BNP
All-cause mortality
7.98 (4.34–14.67)
0.86 (0.77–0.93)
0.51 (0.48–0.53)
1.77 (1.40–2.23)
0.33 (0.21–0.52)
PE-related mortality
7.57 (2.89–19.81)
0.97 (0.85–1)
0.50 (0.47–0.54)
1.88 (1.31–2.70)
0.28 (0.13–0.64)
Serious adverse events
4.66 (3.21–6.77)
0.78 (0.72–0.82)
0.54 (0.52–0.56)
1.75 (1.58–1.93)
0.41 (0.29–0.58)
PE-related mortality
25.97 (6.63–101.66)
0.88 (0.75–0.95)
0.70 (0.65–0.70)
4.53 (1.93–10.66)
0.17 (0.03–0.55)
Serious adverse events
16.81 (5.88–48.09)
0.72 (0.62–0.80)
0.76 (0.73–0.80)
4.88 (2.35–10.17)
0.32 (0.16–0.60)
H–FABP
0.8367
0.9028
PLR positive likelihood ratio, NLR negative likelihood ratio, CI confidence interval, AUC area under the curve
Limitations It was a study level analysis. Studies included were nonrandomized. There was diversity among studies leading to heterogeneity in the final outcome and weaken the conclusion of the analysis. Our pooled analysis did not take into account variables like age, gender, type of assay and cut-off.
Conclusion Our meta-analysis demonstrates that elevated levels of all three biomarkers, including troponin, BNP or NT-proBNP and H-FABP, are associated with increased risk for shortterm mortality and serious adverse events; however, none of the markers provide reliable evidence of satisfactory prognostic performance. Based on the results of our analysis, we cannot discriminate by certainty patients who can be managed as an outpatient versus patients who may get benefit from thrombolytics. H-FABP and hs-cTnT showed promising results with very low NLR for short-term mortality and should be investigated further in larger trials to confirm their prognostic value. Conflict of interest Anurag Bajaj, Parul Rathor, Vishal Sehgal, Besher Kabak, Ajay Shetty, Ossama Al Masalmeh and Srikant Hosur state that there is no conflict of interest involved in this study.
References 1. Goldhaber SZ, Bounameaux H (2012) Pulmonary embolism and deep vein thrombosis. Lancet 379(9828):1835–1846 2. Goldhaber SZ, Visani L, De Rosa M (1999) Acute pulmonary embolism: clinical outcomes in the International Cooperative Pulmonary Embolism Registry (ICOPER). Lancet 353:1386– 1389 3. Jaff MR, McMurtry MS, Archer SL, On behalf of the American Heart Association Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation, Council on Peripheral Vascular Disease, and Council on Arteriosclerosis, Thrombosis and Vascular Biology. et al (2011) Management of massive and submassive pulmonary embolism, iliofemoral deep vein thrombosis, and chronic thromboembolic pulmonary hypertension. Circulation 123:1788–1830 4. Aujesky D, Smith KJ, Cornuz J, Roberts MS (2005) Cost-effectiveness of low-molecular-weight heparin for treatment of pulmonary embolism. Chest 128:1601–1610 5. Kucher N, Rossi E, De Rosa M, Goldhaber SZ (2005) Prognostic role of echocardiography among patients with acute pulmonary embolism and a systolic arterial pressure of 90 mm Hg or higher. Arch Intern Med 165:1777–1781 6. Grifoni S, Olivotto I, Cecchini P et al (2000) Short-term clinical outcome of patients with acute pulmonary embolism, normal blood pressure, and echocardiographic right ventricular dysfunction. Circulation 101:2817–2822 7. Coutance G, Cauderlier E, Ehtisham J, Hamon M, Hamon M (2011) The prognostic value of markers of right ventricular dysfunction in pulmonary embolism: a meta-analysis. Crit Care 15:R103 8. Bajaj A (2014) The prognostic value of plasma heart-type fatty acid-binding protein in acute pulmonary embolism. Chest 146:e236–e237
123
Lung 9. Vuilleumier N, Limacher A, Me´an M et al (2015) Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-high-risk pulmonary embolism. J Intern Med 277(6):707–716 10. Bulj N, Potocˇnjak I, Sharma M, Pintaric H, Degoricija V (2013) Timing of troponin T measurements in triage of pulmonary embolism patients. Croat Med J 54(6):561–568 11. Becattinii C, Casazza F, Forgione C, Imperadore F, Bongarzoni A, Agnelli G et al (2013) Acute pulmonary embolism: external validation of an integrated risk stratification model. Chest 144(5):1539–1545 12. Sanchez O, Trinquart L, Planquette B et al (2013) Echocardiography and pulmonary embolism severity index have independent prognostic roles in pulmonary embolism. Eur Respir J 42(3):681–688 13. Ozsu S, Abul Y, Orem A (2013) Predictive value of troponins and simplified pulmonary embolism severity index in patients with normotensive pulmonary embolism. Multidiscip Respir Med 8(1):34 14. Lankeit M, Jime´nez D, Kostrubiec M et al (2011) Predictive value of the high-sensitivity troponin T assay and the simplified Pulmonary Embolism Severity Index in hemodynamically stable patients with acute pulmonary embolism: a prospective validation study. Circulation 124(24):2716–2724 15. Kang DK, Sun JS, Park KJ, Lim HS (2011) Usefulness of combined assessment with computed tomographic signs of right ventricular dysfunction and cardiac troponin T for risk stratification of acute pulmonary embolism. Am J Cardiol 108(1):133–140 16. Jime´nez D, Aujesky D, Moores L et al (2011) Combinations of prognostic tools for identification of high-risk normotensive patients with acute symptomatic pulmonary embolism. Thorax 66(1):75–81 17. Vanni S, Nazerian P, Pepe G (2011) Comparison of two prognostic models for acute pulmonary embolism; clinical vs right ventricular dysfunction-guided approach. J Thromb Haemost 9:1916–1923 18. Lankeit M, Friesen D, Aschoff J et al (2010) Highly sensitive troponin T assay in normotensive patients with acute pulmonary embolism. Eur Heart J 31(15):1836–1844 19. Stein PD, Matta F, Janjua M, Yaekoub AY, Jaweesh F, Alrifai A (2010) Outcome in stable patients with acute pulmonary embolism who had right ventricular enlargement and/or elevated levels of troponin I. Am J Cardiol 106(4):558–563 20. Ozsu S, Karaman K, Mentese A et al (2010) Combined risk stratification with computerized tomography/echocardiography and biomarkers in patients with normotensive pulmonary embolism. Thromb Res 126(6):486–492 21. Singanayagam A, Scally C, Al-Khairalla MZ et al (2011) Are biomarkers additive to pulmonary embolism severity index for severity assessment in normotensive patients with acute pulmonary embolism? QJM 104(2):125–131 22. Kostrubiec M, Łabyk A, Pedowska-Włoszek J et al (2010) Assessment of renal dysfunction improves troponin-based shortterm prognosis in patients with acute symptomatic pulmonary embolism. J Thromb Haemost 8(4):651–658 23. Vuilleumier N, Le Gal G, Verschuren F et al (2009) Cardiac biomarkers for risk stratification in non-massive pulmonary embolism: a multicenter prospective study. J Thromb Haemost 7(3):391–398 24. Post F, Mertens D, Sinning C, Peetz D, Mu¨nzel T (2009) Decision for aggressive therapy in acute pulmonary embolism: implication of elevated troponin T. Clin Res Cardiol 98(6):401–408 25. Bova C, Pesavento R, Marchiori A et al (2009) TELESIO Study Group. Risk stratification and outcomes in hemodynamically
123
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
stable patients with acute pulmonary embolism: a prospective, multicentre, cohort study with three months of follow-up. J Thromb Haemost 7(6):938–944 Moores L, Aujesky D, Jime´nez D et al (2010) Pulmonary Embolism Severity Index and troponin testing for the selection of low-risk patients with acute symptomatic pulmonary embolism. J Thromb Haemost 8(3):517–522 Jime´nez D, Dı´az G, Molina J et al (2008) Troponin I and risk stratification of patients with acute nonmassive pulmonary embolism. Eur Respir J 31(4):847–853 Palmieri V, Gallotta G, Rendina D et al (2008) Troponin I and right ventricular dysfunction for risk assessment in patients with nonmassive pulmonary embolismin the Emergency Department in combination with clinically based risk score. Intern Emerg Med 3(2):131–138 Tulevski II, ten Wolde M, van Veldhuisen DJ et al (2007) Combined utility of brain natriuretic peptide and cardiac troponin T may improve rapid triage and risk stratification in normotensive patients with pulmonary embolism. Int J Cardiol 116:161–166 Kline JA, Hernandez-Nino J, Rose GA, Norton HJ, Camargo CA Jr (2006) Surrogate markers for adverse outcomes in normotensive patients with pulmonary embolism. Crit Care Med 34:2773–2780 Douketis JD, Leeuwenkamp O, Grobara P et al (2005) The incidence and prognostic significance of elevated cardiac troponins in patients with submassive pulmonary embolism. J Thromb Haemost 3:508–513 Kostrubiec M, Pruszczyk P, Bochowicz A et al (2005) Biomarker-based risk assessment model in acute pulmonary embolism. Eur Heart J 26:2166–2172 Bova C, Crocco F, Ricchio R, Serafini O, Greco F, Noto A (2005) Importance of troponin T for the risk stratification of normotensive patients with pulmonary embolism: a prospective, cohort study with a three-month follow-up. Haematologica 90:423–424 Pruszczyk P, Bochowicz A, Torbicki A et al (2003) Cardiac troponin T monitoring identifies high-risk group of normotensive patients with acute pulmonary embolism. Chest 123:1947–1952 Ten Wolde M, Tulevski II, Mulder JW et al (2003) Brain natriuretic peptide as a predictor of adverse outcome in patients with pulmonary embolism. Circulation 107:2082–2084 Pieralli F, Olivotto I, Vanni S et al (2006) Usefulness of bedside testing for brain natriuretic peptide to identify right ventricular dysfunction and outcome in normotensive patients with acute pulmonary embolism. Am J Cardiol 97:1386–1390 Logeart D, Lecuyer L, Thabut G et al (2007) Biomarker-based strategy for screening right ventricular dysfunction in patients with non-massive pulmonary embolism. Intensive Care Med 33:286–292 Maziere F, Birolleau S, Medimagh S et al (2007) Comparison of troponin I and N-terminal-pro Btype natriuretic peptide for risk stratification in patients with pulmonary embolism. Eur J Emerg Med 14:207–211 Klok FA, Van Der Bijl N, Eikenboom HC et al (2010) Comparison of CT assessed right ventricular size and cardiac biomarkers for predicting short-term clinical outcome in normotensive patients suspected of having acute pulmonary embolism. J Thromb Haemost 8(4):853–856 Zondag W, Agterof MJ, Schutgens RE et al (2011) Repeated NTproBNP testing and risk for adverse outcome after acute pulmonary embolism. Thromb Haemost 106(6):1226–1227 Lankeit M, Jimenez D, Kostrubiec M et al (2014) Validation of N-terminal pro-brain natriuretic peptide cut-off values for risk stratification of pulmonary embolism. Eur Respir J 43(6):1669–1677 Jimenes D, Lobo JL, Monreal M et al (2014) PROTECT investigators. Prognostic significance of multidetector CT in normotensive paients with pulmonary embolism: results of the protect study. Thorax 69(2):109–115
Lung 43. Gu¨l E, Can I, Gu¨ler I et al (2012) Association of pulmonary artery obstruction index with elevated heart-type fatty acid binding protein and short-term mortality in patients with pulmonary embolism at intermediate risk. Diagn Interv Radiol 18:531–536 44. Boscheri A, Wunderlich C, Langer M et al (2010) Correlation of heart-type fatty acid-binding protein with mortality and echocardiographic data in patients with pulmonary embolism at intermediate risk. Am Heart J 160:294–300 45. Chen Y, Liu S, Guo W, Wang ZZ (2013) Heart-type fatty acid binding protein for the assessment of the short-term prognosis in acute pulmonary embolism patients with hemodynamic stability on admission. Zhonghua Jie He He Hu Xi Za Zhi. 36:516–521 46. Dellas C, Tschepe M, Seeber V (2014) A novel H-FABP assay and a fast prognostic score for risk assessment of normotensive pulmonary embolism. Thromb Haemost 111:996–1003 47. Gul EE, Can I, Kayrak M, Abdulhalikov T, Erdogan HI, Altunbas G, Ozdemir K, Gok H (2014) Thrombolysis in patients
48. 49.
50.
51.
with pulmonary embolism and elevated heart-type fatty acidbinding protein levels. J Thromb Thrombolysis 37:483–489 Grimes DA, Schulz KF (2005) Refining clinical diagnosis with likelihood ratios. Lancet 365:1500–1505 Moses LE, Shapiro D, Littenberg B (1993) Combining independent studies of a diagnostic test into a summary ROC curve: dataanalytic approaches and some additional considerations. Stat Med 12:1293–1316 Stroup DF, Berlin JA, Morton SC et al (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting: meta-analysis of observational studies in epidemiology (MOOSE) group. JAMA 283:2008–2012 Herbison P, Hay-Smith J, Gillespie WJ (2006) Adjustment of meta-analyses on the basis of quality scores should be abandoned. J Clin Epidemiol 59(12):1249–1256
123
Prognostic Value of Biomarkers in Acute Non-massive Pulmonary Embolism: A Systematic Review and Meta-analysis Anurag Bajaj1 • Parul Rathor2 • Vishal Sehgal3 • Besher Kabak3 Ajay Shetty3 • Ossama Al Masalmeh1 • Srikanth Hosur4
•
Received: 9 March 2015 / Accepted: 15 June 2015 Ó Springer Science+Business Media New York 2015
Abstract Background Various biomarkers have been evaluated to risk stratify patients with acute pulmonary embolism (PE). We aimed to summarize the available evidence to compare the prognostic value of three most widely studied biomarkers in normotensive patients with acute PE. Method A systematic literature review of database, including Pubmed, EMBASE and Cochrane, was done.
& Anurag Bajaj [email protected] Parul Rathor [email protected] Vishal Sehgal [email protected] Besher Kabak [email protected] Ajay Shetty [email protected] Ossama Al Masalmeh [email protected] Srikanth Hosur [email protected] 1
The Wright Center for Graduate Medical Education, 501 Madison Avenue, Scranton, PA 18510, USA
2
Zhengzhou University, 100 Ke Xue Da Dao, Zhongyuan, Zhengzhou, Henan, China
3
The Common Wealth Medical College, 525 Pine Street, Scranton, PA, USA
4
Critical Care Division, Geisinger Community Medical Center, 1800 Mulberry Street, Scranton, USA
Studies were included if those were done on patients with acute PE and serum troponin or brain natriuretic peptide and N-terminal proBNP (BNP/NT-proBNP) or Heart-type fatty acid-binding protein (H-FABP) assay was done. The primary end point was short-term all-cause mortality. The secondary end points were PE-related mortality and serious adverse events. Results All three biomarkers were significantly associated with increased risk for short-term all-cause mortality, PE-related mortality and serious adverse events. All-cause mortality: troponin [odds ratio (OR) 4.80; 95 % CI 3.25–7.08, I2 = 54 %], BNP or NT-proBNP (OR 7.98; 95 % CI 4.34–14.67, I2 = 0 %); PE-related mortality: troponin (OR 3.80; 95 % CI 2.74–5.27, I2 = 0 %), BNP or NT-proBNP (OR 7.57; 95 % CI 2.89–19.81, I2 = 0 %) and H-FABP (OR 25.97; 95 % CI 6.63–101.66, I2 = 40 %). H-FABP has the lowest negative likelihood ratio (NLR) of 0.17 for mortality followed by high-sensitive cardiac troponin T (hs-cTnT) with NLR of 0.21. Conclusion None of the biomarker identifies a subgroup of patients who can be managed as an outpatient versus patients who may get benefit from thrombolytics with certainty; however, H-FABP and hs-cTnT showed some promising results and should be investigated further. Keywords Pulmonary embolism Prognosis Metaanalysis Mortality Abbreviations PE BNP/ NT-proBNP H-FABP Hs-cTnT PLR
Pulmonary embolism Brain natriuretic peptide/N-terminal brain natriuretic peptide Heart-type fatty acid-binding protein High-sensitive cardiac troponin T Positive likelihood ratio
123
Lung
NLR OR
Negative likelihood ratio Odds ratio
Methods Study Objectives
Introduction Acute pulmonary embolism (PE) is the third leading cause of cardiovascular death [1]. Acute PE has a broad spectrum of clinical manifestations, ranging from patients who die of sudden cardiac death to patients who are literally asymptomatic. The mortality attributable to PE approximates 50 % in patients who are hypotensive at the time of admission (high risk) and current guidelines recommend using thrombolytics in these patients [2, 3]. About 90 % of all acute PE patients are normotensive at the time of admission, but carry very heterogenous short-term mortality, ranging from\1 % to about 15 % [2, 3]. On one side of the spectrum are the patients with very low risk for death that could be managed as an outpatient with new oral anticoagulants and on the other side of the spectrum are the patients with risk for poor outcomes that needs close monitoring and may benefit from more aggressive treatment like thrombolytics. Early risk stratification of the patients with acute PE is very important for optimal management and improves outcomes. It leads to decrease in health care cost and increase in patient satisfaction [4]. Patients who are hemodynamically stable at the time of admission, but manifest right ventricular dysfunction (intermediate risk), have a poor prognosis when compared to patients without right ventricular dysfunction (low risk) [5, 6]. The ability to predict normotensive patients with poor prognosis would help physician to guide treatment according to individual risk. Efforts have been made to predict poor prognosis in patients with pulmonary embolism using clinical as well as serum biomarkers. Echocardiogram is a useful tool in risk stratification in acute PE by diagnosing right ventricular dysfunction, but the positive predictive value of echocardiogram in predicting shortterm mortality is low and is not able to discriminate between patients who can be treated as an outpatient versus patients who need close monitoring [7]. Various biomarkers like troponin (T and I), brain natriuretic peptides (BNP) or N-terminal pro BNP (NT-proBNP) and Heart-type fatty acid-binding (H-FABP) have been evaluated to identify subgroups of patients with poor outcomes [8–47]. These biomarkers are released into the circulation as a consequence of myocardial injury or wall stress in response to elevated right ventricular afterload because of acute PE. However, selection of the most accurate biomarker is still controversial. We aimed to summarize the results of published studies regarding the prognostic role of three principal serum biomarkers in acute PE.
123
The primary objective of this meta-analysis was to summarize the available evidence for the discriminatory ability of prognostic biomarkers in acute PE. We compared the prognostic value of three principal most widely studied serum biomarkers: troponin (T and I), BNP or NT-proBNP and H-FABP in normotensive patients with acute PE. Data Source and Searches A systematic search of Medline, EMBASE and Cochrane Database was performed. Following key words were used, ‘‘Troponin,’’ ‘‘BNP,’’ ‘‘NT-proBNP,’’ ‘‘Heart-type fatty acid binding protein,’’ ‘‘biomarkers’’ and ‘‘Pulmonary embolism.’’ The references of previous meta-analyses, abstracts from annual meetings, web base and review articles were reviewed to identify any additional relevant studies. The retrieved studies were carefully examined to exclude potentially duplicate or overlapping data. No language restriction was enforced. The manuscripts of all retrieved studies cited before December 2014 were reviewed. Abstracts were excluded from the analysis because non-peer reviewed data may introduce bias into the study. Study Eligibility Studies were considered eligible for the analysis if they fulfilled the following criteria. (1) Design: Prospective or retrospective cohort study design with well-defined study population; (2) Population: Patients admitted to the hospital with objectively confirmed acute PE and were normotensive (blood pressure [90 mmHg) at the time of admission; (3) Cardiac specific troponin (T or I), BNP or NT-proBNP or H-FABP assay was done at the time of admission or within 24 h of hospitalization; (4) Outcomes reported: short-term (in-hospital or 30 days) all-cause mortality, PE-related mortality, short-term serious adverse events; (5) 2 9 2 table can be constructed from the data to compute true positive, false positive, true negative, false negative.
Data Extraction and Validity Assessment Two independent reviewers (AB and PR) independently performed the literature search and identify relevant studies. A third investigator was available for arbitration in the event of discordance of the extracted data. The retrieved studies were carefully examined to exclude potentially
Lung
duplicate or overlapping data. Meeting abstracts were excluded from our analysis (Fig. 1). Authors were contacted to get relevant data. Relevant data on study design, year of publication, patient population, inclusion criteria, exclusion criteria, mean age, gender, follow-up period, type of troponin/BNP/NT-proBNP/H-FABP assay, timing of sampling, cut-off of biomarkers used and outcomes were extracted (Tables 1, 2, 3, 4, 5 and 6). In cases where data in several publications were derived from part or all of the same patient series, only the study presenting the largest patient population or most complete dataset was included. Study Endpoints The primary outcome of interest was in-hospital or 30 days all-cause mortality. Secondary outcomes were in-hospital or 30 days mortality resulting from PE and serious adverse events. Definitions Serious adverse events were defined as composite of death, need for thrombolytics, endotracheal intubation, catecholamine infusion for sustained hypotension, cardiopulmonary resuscitation, or recurrent PE. Normotensive pulmonary embolism was defined as patients with acute PE without sustained hypotension (blood pressure more than 90 mmHg) [3]. Data Synthesis and Statistical Analysis We constructed 2 9 2 tables for in-hospital or 30 days mortality and serious adverse events for all biomarkers. We
computed sensitivity, specificity, positive likelihood ratio and negative likelihood ratio separately for all biomarkers by means of true positive, false positive, true negative and false negative. Likelihood ratios do not depend on prevalence rates, and there is general consensus that a positive likelihood ratio of greater than 10 and a negative likelihood ratio of less than 0.1 provide reliable evidence of satisfactory diagnostic performance [48]. A study level analysis was done using Review manager 5.2 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2008) and Meta-disc 1.4. Odds ratio (OR) and 95 % confidence intervals (95 % CI) were used as summary statistics for all outcomes. Studies were evaluated for heterogeneity by visual inspection of the confidence intervals and by means of I2 [I2 = (Q - df)/Q], where Q is the v2 statistic and df is degree of freedom. As a guide, an I2 [ 30 % was considered as an indicator of statistical heterogeneity among the studies. A Mantel–Haenszel fixed effect model was used to calculate the pooled OR for homogeneous end points. Random-effect (DerSimonian) analysis was reported in the presence of significant heterogeneity. Even if there was little or no evidence of heterogeneity, a random-effect model was used for each outcome because of the arguments put forth by many authors for using random-effect models in medical decision-making contexts, especially in the case of rare events. A p value of less than 0.05 was considered significant. Funnel plots were used to evaluate for publication bias by plotting standard error (log OR) of mortality and serious adverse events. Weighted symmetric summary receiver operating characteristic plots was computed for all biomarkers using Moses–Shapiro–Littenberg method [49]. We separately analysed studies that used troponin T and those that used troponin I to predict allcause mortality. We did a combine and separate analysis of BNP and NT-proBNP. We did a subgroup analysis of studies that used high-sensitive cardiac troponin (hs-cTnT), BNP and NT-proBNP to predict all-cause mortality. This meta-analysis was performed in accordance with the ‘‘Meta-Analysis of Observational Studies in Epidemiology: A Proposal for Reporting’’ (MOOSE) [50]. As suggested previously [51], the studies were not scored based on their quality.
Results
Fig. 1 Flow diagram for systematic search. PE pulmonary embolism
The abstracts and titles of 3302 articles were identified for initial review using search strategies as described in Fig. 1. We identified 211 articles to be appropriate in terms of evaluation of prognostic biomarkers in pulmonary embolism. For these articles, full-text articles were obtained. Upon further review, 38 studies were eligible for this metaanalysis. Thirty-seven studies included patients with
123
Lung Table 1 Characteristics of the selected troponin studies Study
Year
Study design
Vuilleumier et al. [9]
2014
Prospective
N
230
Bulj et al. [10]
2013
Prospective
104
Becattini et al. [11]
2013
Prospective
1109
Age (years)
Sex male ( %)
Thrombolytics
Outcomes
Follow-up
75 (69–82)
59.1
NA
30 days SAE
30 days
68.7 ? 13.4
36.5
No
Hospital mortality
In-hospital
69 ? 15
44
Yes
In-hospital
Sanchez et al. [12]
2013
Prospective
517
67 (52–77)
47
No
Hospital mortality/ SAE 30 days SAE
Ozsu et al. [13]
2013
Prospective
121
70 (55–76)
43
Yes
30 days mortality
6 months
Vanni et al. [17]
2011
Prospective
510
NA
43.7
Yes
Hospital mortality/ SAE
In-hospital
Lankeit et al. [14]
2011
Prospective
526
71 (55–79)
51
Yes
30 days mortality/ SAE
6 months
Kang et al. [15] s
2011
Retrospective
173
61 ? 14
44
NA
30 day mortality/ SAE
3 months
Jimenez et al. [16]
2010
Prospective
Lankeit et al. [18]
2010
Prospective
156
591
1165
30 days
74 (65–82)
43
No
30 days mortality
30 days
67 (53–75)
45
Yes
30 days SAE
965 days median
Stein et al. [19]
2010
Retrospective
65 ? 17
45
No
Hospital mortality
In-hospital
Ozsu et al. [20]
2010
Prospective
108
70 (21–90)
44
No
30 days mortality
30 days
Singanayagan et al. [21] Kostrubiec et al. [22]
2010
Retrospective
411
NA
43
No
30 days mortality
2010
Prospective
212
64 ? 18
38
Yes
30 days mortality
30 days/ 6 weeks 30 days
Moores et al. [26]
2009
Prospective
567
NA
43
No
30 days mortality
30 days
Vuilleumier et al. [23]
2009
Prospective
146
72 (21–94)
42
NA
3 months SAE
3 months
Post et al. [24]
2009
Prospective
168
56
45
Yes
30 days mortality
30 days
Bova et al. [25]
2009
Prospective
201
68.2
41
Yes
Hospital mortality/ SAE
3 months
Jimenez et al. [27]
2008
Prospective
318
72 (14)
43
No
30 days mortality
30 days
Bova et al. [33]
2005
Prospective
60
65 (18)
61
Yes
Hospital SAE
3 months
Palmiere et al. [28]
2008
Prospective
89
63
22
Yes
Hospital mortality/HI
In-hospital
Tulevski et al. [29]
2007
Prospective
28
53 ? 18
43
No
90 day mortality
3 months
Kline et al. [30]
2006
Prospective
193
53 ? 17
41
No
Hospital mortality
6 months
Doukeitis et al. [31] Kostrubiec et al. [32]
2005 2005
Prospective Prospective
458 100
62 63 ? 18
43 35
No Yes
3 months mortality 40 days mortality/ SAE
3 months 40 days
Prusczyk et al. [34]
2003
Prospective
64
61 ? 17
53
Yes
Hospital mortality
In-hospital
N number of patients, SAE serious adverse events, NA not available
hypotension or hemodynamic instability at the time of admission and excluded from our analysis. There were ten studies which shared patients with the studies included in our final analysis and were excluded. Fourteen studies did not provide required end points and were excluded.
et al. [32] reported outcomes at 40 days, Vuillieumier et al., Ten Wolde et al. and Doukelitis et al. at 3 months and Kline et al. at 6 months [23, 30–32, 35]. Baseline demographic characteristics of included studies are shown in Table 1, 3 and 5. Mean age of the patients is similar across all the studies and varied from 53 to 72 years.
Selected Studies Troponin Assay Among the included studies, thirty-five studies were prospective cohorts and three were retrospective cohorts as mentioned in Tables 1, 2, 3, 4, 5 and 6. All studies except five reported in-hospital or 30 days event rates. Kostrubiec
123
We included studies for both troponin T and I. Twelve studies used troponin T and eleven studies used troponin I. In one study, information on type of troponin was not
Lung Table 2 Characteristics of the troponin assays Study
Year
Troponin
Assay
Manufacturer
Cut-off for abnormal
Time of sampling
Vuilleumier et al. [9]
2014
hs-cTn T
Electrochemiluminescence
Hoffmann-La Roche
14 pg/ml
Within 24 h
Bulj et al. [10]
2013
cTn T
Electrochemiluminescence
Hoffmann-La Roche
0.1 ng/ml
Admission, 6, 24, 72 h
Becattini et al. [11] Sanchez et al. [12]
2013 2013
cTn I or T cTn I
NA Photometric immunoassay
NA Dade behring
NA 0.1 lg/ml
Within 24 h Admission
Ozsu et al. [13]
2013
hs-cTn T
Electrochemiluminescence
Roche
0.014 ng/ml
Admission
Vanni et al. [17]
2011
cTn I
ADVIA centaur
VITALGmbH
0.15 ng/ml
NA
Lankeit et al. [14]
2011
hs-cTn T
Electrochemiluminescence
Roche
14 pg/ml
Admission
Kang et al. [15]
2011
cTn T
Electrochemiluminescence
Roche
0.2 ng/ml
Admission, 12, 24 h
Jimenez et al. [16]
2011
cTn I
MEIA
Abbott
0.1 ng/ml
within 12 h
Lankeit et al. [18]
2010
hs-cTn T
Electrochemiluminescence
Roche
14 pg/ml
Admission
Stein et al. [19]
2010
cTn I
Chemiluminescence
Beckman coulter
0.5 ng/ml
NA
Advia centaur
1.2 ng/ml
Ozsu et al. [20]
2010
c Tn T
Electrochemiluminescence
Roche
0.027 ng/ml
Admission
Singanayagan et al. [21] Kostrubiec et al. [22]
2010
c Tn I
NA
NA
0.2 lg/l
Within 4 h
2010
cTn T
Electrochemiluminescence
Roche
0.01 lg/l
Admission
c Tn I
Dimension RXL
Dade behring
0.01 lg/l
c Tn I
MEIA
Abbott
0.1 ng/ml
ultra Tn I
Moores et al. [26]
2009
Admission
Vuilleumier et al. [23]
2009
cTn I
Electrochemiluminescence
Beckman coulter
0.09 ng/ml
Admission
Post et al. [24]
2009
c Tn T
Electrochemiluminescence
NA
0.1 ng/ml
Admission, 8 h
Bova et al. [25]
2009
c Tn I
Immunoenzymatic method
Beckmann/Dade Behring
0.07/0.15 ng/ml
Within 12, 24,72 h
Jimenez et al. [27]
2008
cTn I
MEIA
Abbott
0.1 ng/ml
Admission
Bova et al. [33]
2005
c Tn T
NA
NA
0.01 lg/l
NA
Palmiere et al. [28]
2008
cTn I
Immunoassay-testPak
Dade behring
0.1 lg/l
Admission, 24, 48 h
Tulevski et al. [29]
2007
cTn T
NA
NA
0.01 lg/l
Admission
Kline et al. [30] Doukeitis et al. [31]
2006 2005
cTn T cTn I
Electrochemiluminescence Immunofluorescence
Roche Abbott
0.1 ng/l 0.5 lg/l
NA Within 24 h
Kostrubiec et al. [32]
2005
c Tn T
Electrochemiluminescence
Roche
0.01 lg/ml
Admission
Prusczyk et al. [34]
2003
cTn T
Electrochemiluminescence
Roche
0.01 ug/l
Admission, 6, 12, 18 h
MEIA microparticle enzyme immunoassay, NA not available
available and two studies included both troponin T and I. All studies used a predefined cut-off to define elevated versus normal troponin levels except eight studies where the best threshold able to define abnormal levels was derived from receiver operating characteristic (ROC) curve construction [14, 15, 18, 20, 22–24, 32]. Information on type of assay, manufacturer and cut-off is shown in Table 2. BNP or NT-proBNP Assay We included seventeen studies for both BNP and NTproBNP. Eight studies used BNP and nine studies used NTproBNP. Information on type of assay, manufacturer and cut-
off is shown in Table 5. Eight studies used predefined cut-off [12, 25, 29, 30, 38–40, 42] and nine studies defined abnormal cut-off after ROC analysis to determine best threshold to predict complicated PE [9, 14, 18, 20, 23, 32, 35–37]. H-FABP Assay We included six studies in this category. Three studies used predefined cut-off for H-FABP assay [43, 44, 47] and in other three studies, the used cut-off was derived from ROC construction to estimate the best threshold [23, 45, 46]. Information on type of assay, manufacturer and cut-off is shown in Table 6.
123
Lung Table 3 Characteristics of the selected BNP or NT-proBNP studies Study
Year
Design
N
Age (years)
Gender male (%)
Thrombolytics
Outcomes
Follow-up 90 days
Ten Wolde et al. [35]
2003
Prospective
110
58 ± 18
na
No
90 days mortality
Pieralli et al. [36]
2006
Prospective
61
75 ± 14
26
Yes
In-hospital mortality
Inhospital
Kline et al. [30]
2006
Prospective
191
53 ? 17
42
No
6 months SAE
6 months 90 days
Tulevski et al. [29]
2007
Prospective
28
53 ± 18
43
No
90 days mortality
Logeart et al. [37]
2007
Prospective
67
64 ± 16
60
Yes
In-hospital mortality
Inhospital
Bova et al. [25]
2009
Prospective
144
68.2
41
Yes
Hospital mortality/SAE
90 days 30 days
Sanchez et al. [12]
2013
Prospective
521
67 (52–77)
47
No
30 days/SAE
Jimenez et al. [42]
2014
Prospective
848
72 (59–80)
49
Yes
30 days mortality/SAE
30 days
Kostrubiec et al. [32]
2005
Prospective
100
62 ± 18
35
Yes
40 days mortality
40 days
Maziere et al. [38]
2007
Prospective
60
72 ? 15
40
No
In-hospital mortality
Inhospital
Vuilleumier et al. [23] Lankeit et al. [18]
2009 2010
Prospective Prospective
146 156
72 (21–94) 67 (53–75)
42 45
NA Yes
90 SAE 30 days mortality/SAE
90 days 965 median
Ozsu et al. [20]
2010
Prospective
108
70 (21–90)
44
No
30 days mortality
30 days
Klok et al. [39]
2010
Prospective
113
56 ± 17
53
Yes
6 weeks SAE
6 weeks Inhospital
Zondag et al. [40]
2011
Prospective
63
63 (50–73)
29
Yes
In-hospital SAE
Vuilleumier et al. [9]
2014
Prospective
230
75 (69–82)
59.1
NA
30 SAE
30 days
Lankeit et al. [41]
2014
Prospective
688
70 (54–78)
47.4
Yes
30 days SAE
30 days
N number of patients, SAE serious adverse events, NA not available
Outcome Measures Troponin All-cause Mortality: Twenty-three studies reported allcause mortality data (Fig. 2). The overall mortality rate in acute PE was 5.7 %. 274 of 2596 patients with elevated troponin levels and 143 out of 4707 with normal levels died (10.5 vs. 3.1 %). Elevated troponin (T or I) levels were significantly associated with higher risk of death as compared to normal troponin levels (OR 4.80; 95 % CI 3.25–7.08, I2 = 54 %). Excluding three retrospective studies revealed the same results (OR 4.69; 95 % CI 2.95–7.44, I2 = 55 %). Eleven studies used troponin T and ten studies used troponin I. The results were consistent for both troponin T (OR 7.58; 95 % CI 4.55–12.61], I2 = 0 %) and I (OR 3.27; 95 % CI 2.01–5.34, I2 = 65 %). Subgroup analysis of studies using high-sensitive cardiac troponin T (hs-cTnT): Three studies which reported all-cause mortality used hs-cTnT. OR for mortality was 7.86 (95 % CI 2.94–21.00, I2 = 0 %). PE-related mortality (Fig. 2): Fifteen studies reported PE-related mortality. Elevated troponin (T and I) levels were significantly associated with higher risk of PE-related death as compared to normal troponin levels (OR 3.80; 95 % CI 2.74–5.27, I2 = 0 %). Serious adverse events (Fig. 2): Ten studies reported serious adverse events rates. 283 out of 3437 patients had
123
serious adverse events. 176 patients out of 1390 with elevated troponin levels had serious adverse events as compared to 107 out of 2047 with normal troponin levels. Elevated troponin levels were significantly associated with higher risk of serious adverse events as compared to normal troponin levels (OR 3.65; 95 % CI 2.41–5.53, I2 = 47 %). Excluding study by Kostrubiec et al. [32] significantly decreased the heterogeneity to zero (OR 4.09; 95 % CI 3.04–5.51, I2 = 0 %).
BNP or NT-proBNP All-cause mortality (Fig. 3): Ten studies reported data on all-cause mortality. The overall mortality rate in acute PE was 5.5 %. 75 of 808 patients with elevated BNP or NTproBNP levels and 12 out of 765 with normal levels died (9.2 vs. 1.5 %). Elevated BNP or NT-proBNP levels were significantly associated with higher risk of death as compared to normal levels (OR 7.98; 95 % CI 4.34–14.67, I2 = 0 %). The results were consistent for both BNP (OR 6.72; 95 % CI 2.21–20.41], I2 = 0 %) and NT-proBNP (OR 8.59; 95 % CI 4.10–17.97, I2 = 1 %). PE-related mortality (Fig. 3): Nine studies reported PErelated mortality. Elevated BNP or NT-proBNP levels were significantly associated with higher risk of PE-related mortality as compared to normal troponin levels (OR 7.57; 95 % CI 2.89–19.81, I2 = 0 %).
Lung Table 4 Characteristics of the BNP or NT-proBNP assays Study
Assay
Type of assay
Manufacturer
Cutt off for abnormal
Timing of sampling
Ten Wolde et al. [35]
BNP
Immunoradiometric
Shionoria
21.7 pg/ml
Admission
Pieralli et al. [36]
BNP
Fluorescence immunoassay
Biosite
527 ng/ml
First hour within admission
Kline et al. [30]
BNP
Two-site ‘‘sandwich’’ immunoassay and chemiluminescence
Biosite
90 pg/ml
Admission
Tulevski et al. [29]
BNP
Immunoradiometric
Shionoria
35 pg/ml
Admission
Logeart et al. [37]
BNP
Fluorescence immunoassay
Biosite
100 pg/ml
Admission
Bova et al. [25]
BNP
chemiluminescence
Biosite
100 pg/ml
Admission, 12, 24, 48 h
Immunofluorescence
Roche
50 pg/ml
Sanchez et al. [12]
BNP
Electrochemiluminescence
Beckman coulter
100 pg/ml
Jimenez et al. [42]
BNP
MEIA
Abbott
100 pg/ml
Kostrubiec et al. [32]
NT-BNP
Electrochemiluminescence
Roche
600 pg/ml
Maziere et al. [38]
NT-BNP
Electrochemiluminescence
Roche
1000 pg/ml
Admission
Vuilleumier et al. [23]
NT-BNP
Electrochemiluminescence
Roche
300 pg/ml
Admission
Admission Admission
Lankeit et al. [18]
NT-BNP
Electrochemiluminescence
Roche
1000 pg/ml
Admission
Ozsu et al. [20]
NT-BNP
Enzyme immunoassay
Biomedica
300 pmol/ml
Admission
Klok et al. [39]
NT-BNP
Electrochemiluminescence
Roche
600 pg/ml
Admission
Zondag et al. [40]
NT-BNP
Electrochemiluminescence
Roche
500 pg/ml
Admission and every morning
Vuilleumier et al. [9] Lankeit et al. [41]
NT-BNP NT-BNP
Electrochemiluminescence Electrochemiluminescence
Hoffman la roche Roche
300 pg/ml 600 pg/ml
Admission Admission
Table 5 Characteristics of the selected H-FABP studies Study
Year
Study design
N
Age
Gender male (%)
Thrombolytics
Outcome
Follow-up
Gul et al. [47]
2014
Prospective
80
62 ? 17
40
Yes
30 days SAE
30 days
Chen et al. [45]
2013
Prospective
90
61.1 ? 14.6
41
Yes
30 days SAE
30 days
Dellas et al. [46]
2014
Prospective
271
70 (52–76)
44
Yes
30 days SAE
30 days
Vuilleumier et al. [23]
2009
Prospective
146
72 (21–94)
42
Yes
90 days SAE
90 days
Gul et al. [43]
2012
Prospective
61
62 ? 17
37.7
Yes
RVD
30 days
Boscheri et al. [44]
2010
Prospective
101
70 ? 13
43.5
Yes
30 days mortality
30 days
H-FABP heart-type fatty acid-binding protein, N number of patients, SAE serious adverse events, NA not available
Serious adverse events (Fig. 3): Twelve studies reported data on serious adverse events. Elevated BNP or NT-proBNP levels were significantly associated with higher risk of serious adverse events as compared to normal levels (OR 4.66; 95 % CI 3.21–6.77, I2 = 10 %). The results were consistent for both BNP (OR 3.69; 95 % CI 2.51–5.42, I2 = 0 %) and NTproBNP (OR 7.03; 95 % CI 3.67–13.15, I2 = 0 %). H-FABP All-cause mortality: None of the studies reported all-cause mortality.
PE-related mortality (Fig. 4): Four studies reported PErelated mortality. None of the studies reported all-cause mortality. The PE-related mortality in acute PE was 9.3 %. Elevated H-FABP levels were significantly associated with higher risk of PE-related mortality as compared to normal H-FABP levels (OR 25.97; 95 % CI 6.63–101.66, I2 = 40 %). If we exclude the study of Boscheri et al. [44], the heterogeneity becomes zero (OR 14.14; 95 % CI 5.27–37.92, I2 = 0 %). Serious adverse events (Fig. 4): Six studies reported serious adverse events. Elevated H-FABP levels were significantly associated with higher risk of serious adverse
123
Lung Table 6 Characteristics of the H-FABP assays Study
Year
H-FABP Assay
Cut-off for abnormal (ng/ml)
Manufacturer
Time of sampling
Gul et al. [47]
2014
Immunochromatographic
7
HiSens card
Admission
Chen et al. [45]
2013
Sandwich Elisa
6
NA
Admission
Dellas et al. [46]
2014
Elisa/Immunoturbidimetric
6
HyCult
Admission
Vuilleumier et al. [23]
2009
Sandwich Elisa
6
HyCult
Admission
Gul et al. [43]
2012
Immunochromatographic
7
HiSens card
Admission
Boscheri et al. [44]
2010
Sandwich Elisa
7
Rennescens GmbH
Admission
Fig. 2 Forest plot of pooled analysis comparing outcomes between elevated troponin and normal troponin levels in acute Pulmonary embolism. a All-cause mortality; b PE-related mortality; c Serious adverse events. The squares and horizontal lines indicate odds ratio for random-effect model and the 95 % CI for each trial included. The
size of each square is proportional to the statistical weight of a trial. The diamond indicates effect estimate derived from meta-analysis with centre indicates the point estimate and the right and left end indicate 95 % CI
events as compared to normal H-FABP levels (OR 16.81; 95 % CI 5.88–48.09, I2 = 61 %). If we exclude the study of Boscheri et al. and Vuilleumier et al. [23, 44] then heterogeneity becomes zero(OR 16.13; 95 % CI 7.52–34.59, I2 = 0 %).
Publication Bias
123
The funnel plots were created to ascertain publication bias in our study comparing mortality and serious adverse events between elevated biomarker levels and the normal
Lung
Fig. 3 Forest plot of pooled analysis comparing outcomes between elevated BNP or NT-proBNP and normal BNP or NT-proBNP levels in acute Pulmonary embolism. a All-cause mortality; b PE-related mortality; c serious adverse events
levels in acute PE. We drew the funnel plots only for troponin and BNP or NT-proBNP. We did not draw the funnel plots for H-FABP analysis because the numbers of studies were very few. The plots were asymmetrical in our study for all outcomes (all-cause mortality, PE-related mortality and serious adverse events) for both troponin and BNP indicating that there may be a publication bias. It was roughly symmetrical for PE-related mortality in troponin. However, apart from publication bias, it is important to consider other source of selection bias, heterogeneity of studies and methodological limitations of the studies.
Discussion Our meta-analysis indicates that elevated levels of all three biomarkers, including troponin, BNP or NT-proBNP and H-FABP, are associated with increased risk for short-
term mortality as well as serious adverse events. To our knowledge, this is the first ever meta-analysis in which prognostic value of three most studied principal biomarkers is compared simultaneously in normotensive patients with acute PE. There is general consensus that a positive likelihood ratio of greater than 10 and a negative likelihood ratio of less than 0.1 provide reliable evidence of satisfactory prognostic performance. None of the three biomarkers fulfilled the above-mentioned criteria to be able to satisfactorily discriminate between patients with low risk for death versus high risk for death (Table 7). H-FABP, with NLR of 0.17, provides the closest NLR for satisfactory performance to be able to identify patients who are low risk for death and can be treated as an outpatient. However, the number of studies included is very few, only four for mortality and six for serious adverse events. Also, there is some heterogeneity among the studies. These two factors weaken the conclusion.
123
Lung
Fig. 4 Forest plot of pooled analysis comparing outcomes between elevated H-FABP and normal H-FABP levels in acute Pulmonary embolism. a PE-related mortality; b serious adverse events
However, H-FABP has some additional advantage over other biomarkers. H-FABP is a cytoplasmic protein released from cardiac myocytes into the circulation as early as 90 min after PE; therefore, it is useful in early risk stratification when the patients still in the ER [43– 46]. As compared to H-FABP, troponin and BNP or NTproBNP is released into the circulation a little later. Troponin takes at least 6 h to rise so normal troponin levels at the time of admission do not rule out poor outcomes in PE. Also, neither PLR nor NLR is extreme enough to identify patients who have favourable versus unfavourable outcomes. Three of the troponin studies used hs-cTnT for all-cause mortality. NLR of hs-cTnT in predicting all-cause mortality is 0.21. The NLR of hs-cTnT is small as compared to other troponins and could be used to identify patients who can be treated as an outpatient; however, the numbers of studies are very few and results cannot be generalized at this point, more studies are needed to clarify the role of hs-cTnT in acute PE. BNP or NT-proBNP has low NLR of 0.28 for mortality and 0.41 for serious adverse events and it could be a potential biomarker to identify patients with low risk of adverse events; however, the levels of BNP are altered in many clinical conditions like renal failure, elderly age and there is no standardized cut-off to define normal versus abnormal levels. Combining biomarkers like H-FABP with BNP or hs-cTnT or with clinical prognostic tool like sPESI (simplified pulmonary embolism severity index
123
score) may be able to provide us more satisfactory prognostic tool. Combining biomarkers such as Troponin with H-FABP sequentially can increase the combine specificity and may identify patients who need more close monitoring. Similarly, combining biomarkers such as hscTnT with the BNP or H-FABP simultaneously can increase combine sensitivity and may identify patients who can be treated as an outpatient. However, combining biomarkers are beyond the scope of the current study and further studies are needed to clarify their role.
Strengths We did an extensive search for the relevant studies. We included a larger number of studies by searching the grey literature and contacting the authors when data were missing to complete the analysis. Studies were searched and reviewed independently by two authors and clarified by a third author for methods and outcomes. We screened the studies in detail and also contacted author to remove any duplicated or partially duplicated cohorts. We used random-effect models for wider confidence interval and between study and within study variation. We calculated the likelihood ratios which best describe the discriminatory power of the test because it does not depend on prevalence of the disease and it helps clinicians in area of clinical uncertainty [48].
Lung Table 7 Pooled summary results of prognostic value of biomarkers in acute Pulmonary embolism Biomarker
Outcomes
ORs (95 % CI)
Sensitivity
Specificity
PLR
NLR
AUC
Troponin
All-cause mortality
4.80 (3.25–7.08)
0.66 (0.61–0.70)
0.66 (0.65–0.67)
2.13 (1.84–2.47)
0.51 (0.40–0.60)
0.7393
PE-related mortality
3.80 (2.74–5.27)
0.69 (0.61–0.75)
0.62 (0.61–0.64)
2.05 (1.69–2.49)
0.55 (0.45–0.67)
Serious adverse events
3.65 (2.41–5.53)
0.64 (0.58–0.70)
0.64 (0.62–0.65)
2.06 (1.62–2.62)
0.53 (0.39–0.72)
hs-cTn T
All-cause mortality
7.86 (2.94–21.00)
0.92 (0.81–0.98)
0.43 (0.39–0.46).
1.61 (1.42–1.86)
0.21 (0.09–0.51)
BNP
All-cause mortality
7.98 (4.34–14.67)
0.86 (0.77–0.93)
0.51 (0.48–0.53)
1.77 (1.40–2.23)
0.33 (0.21–0.52)
PE-related mortality
7.57 (2.89–19.81)
0.97 (0.85–1)
0.50 (0.47–0.54)
1.88 (1.31–2.70)
0.28 (0.13–0.64)
Serious adverse events
4.66 (3.21–6.77)
0.78 (0.72–0.82)
0.54 (0.52–0.56)
1.75 (1.58–1.93)
0.41 (0.29–0.58)
PE-related mortality
25.97 (6.63–101.66)
0.88 (0.75–0.95)
0.70 (0.65–0.70)
4.53 (1.93–10.66)
0.17 (0.03–0.55)
Serious adverse events
16.81 (5.88–48.09)
0.72 (0.62–0.80)
0.76 (0.73–0.80)
4.88 (2.35–10.17)
0.32 (0.16–0.60)
H–FABP
0.8367
0.9028
PLR positive likelihood ratio, NLR negative likelihood ratio, CI confidence interval, AUC area under the curve
Limitations It was a study level analysis. Studies included were nonrandomized. There was diversity among studies leading to heterogeneity in the final outcome and weaken the conclusion of the analysis. Our pooled analysis did not take into account variables like age, gender, type of assay and cut-off.
Conclusion Our meta-analysis demonstrates that elevated levels of all three biomarkers, including troponin, BNP or NT-proBNP and H-FABP, are associated with increased risk for shortterm mortality and serious adverse events; however, none of the markers provide reliable evidence of satisfactory prognostic performance. Based on the results of our analysis, we cannot discriminate by certainty patients who can be managed as an outpatient versus patients who may get benefit from thrombolytics. H-FABP and hs-cTnT showed promising results with very low NLR for short-term mortality and should be investigated further in larger trials to confirm their prognostic value. Conflict of interest Anurag Bajaj, Parul Rathor, Vishal Sehgal, Besher Kabak, Ajay Shetty, Ossama Al Masalmeh and Srikant Hosur state that there is no conflict of interest involved in this study.
References 1. Goldhaber SZ, Bounameaux H (2012) Pulmonary embolism and deep vein thrombosis. Lancet 379(9828):1835–1846 2. Goldhaber SZ, Visani L, De Rosa M (1999) Acute pulmonary embolism: clinical outcomes in the International Cooperative Pulmonary Embolism Registry (ICOPER). Lancet 353:1386– 1389 3. Jaff MR, McMurtry MS, Archer SL, On behalf of the American Heart Association Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation, Council on Peripheral Vascular Disease, and Council on Arteriosclerosis, Thrombosis and Vascular Biology. et al (2011) Management of massive and submassive pulmonary embolism, iliofemoral deep vein thrombosis, and chronic thromboembolic pulmonary hypertension. Circulation 123:1788–1830 4. Aujesky D, Smith KJ, Cornuz J, Roberts MS (2005) Cost-effectiveness of low-molecular-weight heparin for treatment of pulmonary embolism. Chest 128:1601–1610 5. Kucher N, Rossi E, De Rosa M, Goldhaber SZ (2005) Prognostic role of echocardiography among patients with acute pulmonary embolism and a systolic arterial pressure of 90 mm Hg or higher. Arch Intern Med 165:1777–1781 6. Grifoni S, Olivotto I, Cecchini P et al (2000) Short-term clinical outcome of patients with acute pulmonary embolism, normal blood pressure, and echocardiographic right ventricular dysfunction. Circulation 101:2817–2822 7. Coutance G, Cauderlier E, Ehtisham J, Hamon M, Hamon M (2011) The prognostic value of markers of right ventricular dysfunction in pulmonary embolism: a meta-analysis. Crit Care 15:R103 8. Bajaj A (2014) The prognostic value of plasma heart-type fatty acid-binding protein in acute pulmonary embolism. Chest 146:e236–e237
123
Lung 9. Vuilleumier N, Limacher A, Me´an M et al (2015) Cardiac biomarkers and clinical scores for risk stratification in elderly patients with non-high-risk pulmonary embolism. J Intern Med 277(6):707–716 10. Bulj N, Potocˇnjak I, Sharma M, Pintaric H, Degoricija V (2013) Timing of troponin T measurements in triage of pulmonary embolism patients. Croat Med J 54(6):561–568 11. Becattinii C, Casazza F, Forgione C, Imperadore F, Bongarzoni A, Agnelli G et al (2013) Acute pulmonary embolism: external validation of an integrated risk stratification model. Chest 144(5):1539–1545 12. Sanchez O, Trinquart L, Planquette B et al (2013) Echocardiography and pulmonary embolism severity index have independent prognostic roles in pulmonary embolism. Eur Respir J 42(3):681–688 13. Ozsu S, Abul Y, Orem A (2013) Predictive value of troponins and simplified pulmonary embolism severity index in patients with normotensive pulmonary embolism. Multidiscip Respir Med 8(1):34 14. Lankeit M, Jime´nez D, Kostrubiec M et al (2011) Predictive value of the high-sensitivity troponin T assay and the simplified Pulmonary Embolism Severity Index in hemodynamically stable patients with acute pulmonary embolism: a prospective validation study. Circulation 124(24):2716–2724 15. Kang DK, Sun JS, Park KJ, Lim HS (2011) Usefulness of combined assessment with computed tomographic signs of right ventricular dysfunction and cardiac troponin T for risk stratification of acute pulmonary embolism. Am J Cardiol 108(1):133–140 16. Jime´nez D, Aujesky D, Moores L et al (2011) Combinations of prognostic tools for identification of high-risk normotensive patients with acute symptomatic pulmonary embolism. Thorax 66(1):75–81 17. Vanni S, Nazerian P, Pepe G (2011) Comparison of two prognostic models for acute pulmonary embolism; clinical vs right ventricular dysfunction-guided approach. J Thromb Haemost 9:1916–1923 18. Lankeit M, Friesen D, Aschoff J et al (2010) Highly sensitive troponin T assay in normotensive patients with acute pulmonary embolism. Eur Heart J 31(15):1836–1844 19. Stein PD, Matta F, Janjua M, Yaekoub AY, Jaweesh F, Alrifai A (2010) Outcome in stable patients with acute pulmonary embolism who had right ventricular enlargement and/or elevated levels of troponin I. Am J Cardiol 106(4):558–563 20. Ozsu S, Karaman K, Mentese A et al (2010) Combined risk stratification with computerized tomography/echocardiography and biomarkers in patients with normotensive pulmonary embolism. Thromb Res 126(6):486–492 21. Singanayagam A, Scally C, Al-Khairalla MZ et al (2011) Are biomarkers additive to pulmonary embolism severity index for severity assessment in normotensive patients with acute pulmonary embolism? QJM 104(2):125–131 22. Kostrubiec M, Łabyk A, Pedowska-Włoszek J et al (2010) Assessment of renal dysfunction improves troponin-based shortterm prognosis in patients with acute symptomatic pulmonary embolism. J Thromb Haemost 8(4):651–658 23. Vuilleumier N, Le Gal G, Verschuren F et al (2009) Cardiac biomarkers for risk stratification in non-massive pulmonary embolism: a multicenter prospective study. J Thromb Haemost 7(3):391–398 24. Post F, Mertens D, Sinning C, Peetz D, Mu¨nzel T (2009) Decision for aggressive therapy in acute pulmonary embolism: implication of elevated troponin T. Clin Res Cardiol 98(6):401–408 25. Bova C, Pesavento R, Marchiori A et al (2009) TELESIO Study Group. Risk stratification and outcomes in hemodynamically
123
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
stable patients with acute pulmonary embolism: a prospective, multicentre, cohort study with three months of follow-up. J Thromb Haemost 7(6):938–944 Moores L, Aujesky D, Jime´nez D et al (2010) Pulmonary Embolism Severity Index and troponin testing for the selection of low-risk patients with acute symptomatic pulmonary embolism. J Thromb Haemost 8(3):517–522 Jime´nez D, Dı´az G, Molina J et al (2008) Troponin I and risk stratification of patients with acute nonmassive pulmonary embolism. Eur Respir J 31(4):847–853 Palmieri V, Gallotta G, Rendina D et al (2008) Troponin I and right ventricular dysfunction for risk assessment in patients with nonmassive pulmonary embolismin the Emergency Department in combination with clinically based risk score. Intern Emerg Med 3(2):131–138 Tulevski II, ten Wolde M, van Veldhuisen DJ et al (2007) Combined utility of brain natriuretic peptide and cardiac troponin T may improve rapid triage and risk stratification in normotensive patients with pulmonary embolism. Int J Cardiol 116:161–166 Kline JA, Hernandez-Nino J, Rose GA, Norton HJ, Camargo CA Jr (2006) Surrogate markers for adverse outcomes in normotensive patients with pulmonary embolism. Crit Care Med 34:2773–2780 Douketis JD, Leeuwenkamp O, Grobara P et al (2005) The incidence and prognostic significance of elevated cardiac troponins in patients with submassive pulmonary embolism. J Thromb Haemost 3:508–513 Kostrubiec M, Pruszczyk P, Bochowicz A et al (2005) Biomarker-based risk assessment model in acute pulmonary embolism. Eur Heart J 26:2166–2172 Bova C, Crocco F, Ricchio R, Serafini O, Greco F, Noto A (2005) Importance of troponin T for the risk stratification of normotensive patients with pulmonary embolism: a prospective, cohort study with a three-month follow-up. Haematologica 90:423–424 Pruszczyk P, Bochowicz A, Torbicki A et al (2003) Cardiac troponin T monitoring identifies high-risk group of normotensive patients with acute pulmonary embolism. Chest 123:1947–1952 Ten Wolde M, Tulevski II, Mulder JW et al (2003) Brain natriuretic peptide as a predictor of adverse outcome in patients with pulmonary embolism. Circulation 107:2082–2084 Pieralli F, Olivotto I, Vanni S et al (2006) Usefulness of bedside testing for brain natriuretic peptide to identify right ventricular dysfunction and outcome in normotensive patients with acute pulmonary embolism. Am J Cardiol 97:1386–1390 Logeart D, Lecuyer L, Thabut G et al (2007) Biomarker-based strategy for screening right ventricular dysfunction in patients with non-massive pulmonary embolism. Intensive Care Med 33:286–292 Maziere F, Birolleau S, Medimagh S et al (2007) Comparison of troponin I and N-terminal-pro Btype natriuretic peptide for risk stratification in patients with pulmonary embolism. Eur J Emerg Med 14:207–211 Klok FA, Van Der Bijl N, Eikenboom HC et al (2010) Comparison of CT assessed right ventricular size and cardiac biomarkers for predicting short-term clinical outcome in normotensive patients suspected of having acute pulmonary embolism. J Thromb Haemost 8(4):853–856 Zondag W, Agterof MJ, Schutgens RE et al (2011) Repeated NTproBNP testing and risk for adverse outcome after acute pulmonary embolism. Thromb Haemost 106(6):1226–1227 Lankeit M, Jimenez D, Kostrubiec M et al (2014) Validation of N-terminal pro-brain natriuretic peptide cut-off values for risk stratification of pulmonary embolism. Eur Respir J 43(6):1669–1677 Jimenes D, Lobo JL, Monreal M et al (2014) PROTECT investigators. Prognostic significance of multidetector CT in normotensive paients with pulmonary embolism: results of the protect study. Thorax 69(2):109–115
Lung 43. Gu¨l E, Can I, Gu¨ler I et al (2012) Association of pulmonary artery obstruction index with elevated heart-type fatty acid binding protein and short-term mortality in patients with pulmonary embolism at intermediate risk. Diagn Interv Radiol 18:531–536 44. Boscheri A, Wunderlich C, Langer M et al (2010) Correlation of heart-type fatty acid-binding protein with mortality and echocardiographic data in patients with pulmonary embolism at intermediate risk. Am Heart J 160:294–300 45. Chen Y, Liu S, Guo W, Wang ZZ (2013) Heart-type fatty acid binding protein for the assessment of the short-term prognosis in acute pulmonary embolism patients with hemodynamic stability on admission. Zhonghua Jie He He Hu Xi Za Zhi. 36:516–521 46. Dellas C, Tschepe M, Seeber V (2014) A novel H-FABP assay and a fast prognostic score for risk assessment of normotensive pulmonary embolism. Thromb Haemost 111:996–1003 47. Gul EE, Can I, Kayrak M, Abdulhalikov T, Erdogan HI, Altunbas G, Ozdemir K, Gok H (2014) Thrombolysis in patients
48. 49.
50.
51.
with pulmonary embolism and elevated heart-type fatty acidbinding protein levels. J Thromb Thrombolysis 37:483–489 Grimes DA, Schulz KF (2005) Refining clinical diagnosis with likelihood ratios. Lancet 365:1500–1505 Moses LE, Shapiro D, Littenberg B (1993) Combining independent studies of a diagnostic test into a summary ROC curve: dataanalytic approaches and some additional considerations. Stat Med 12:1293–1316 Stroup DF, Berlin JA, Morton SC et al (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting: meta-analysis of observational studies in epidemiology (MOOSE) group. JAMA 283:2008–2012 Herbison P, Hay-Smith J, Gillespie WJ (2006) Adjustment of meta-analyses on the basis of quality scores should be abandoned. J Clin Epidemiol 59(12):1249–1256
123
