American Journal of Emergency Medicine xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem

Original Contribution

Predictors of early death in patients with acute pulmonary embolism☆ Çağdaş Akgüllü, MD a,⁎, İmran Kurt Ömürlü, PhD b, Ufuk Eryılmaz, MD a, Mücahit Avcil, MD c, Evrin Dağtekin, MD a, Mehmet Akdeniz, MD a, Hasan Güngör, MD a, Cemil Zencir, MD a a b c

Department of Cardiology, Medical Faculty, Adnan Menderes University, Aydin 09100, Turkey Department of Biostatistics, Medical Faculty, Adnan Menderes University, Aydin 09100, Turkey Emergency Department, Medical Faculty, Adnan Menderes University, Aydin 09100, Turkey

a r t i c l e

i n f o

Article history: Received 19 June 2014 Received in revised form 10 November 2014 Accepted 15 November 2014 Available online xxxx

a b s t r a c t Aim: We aimed to determine the predictors of early death in the course of acute pulmonary embolism (APE). Materials and methods: We included 206 patients who had been admitted to our hospital between January 2011 and April 2013 with the diagnosis of APE. We derived a new model including corrected QT interval dispersion (QTcd) and P wave dispersion (Pd), echocardiographic findings, laboratory markers, and blood cell count indices to predict early death in patients with APE. Results: Thirty patients (14.5%) died; 176 patients (85.5%) lived after diagnosis of APE. Logistic regression (LR) analysis found that troponin I (odds ratio [OR], 1.084 [95% confidence interval {CI}, 1.009-1.165]), creatinine (OR, 4.153 [95% CI, 1.375-12.541]), mean platelet volume (OR, 1.991 [95% CI, 1.230-3.223]), neutrophil to lymphocyte ratio (NLR) (OR, 1.079 [95% CI, 1.005-1.160]), QTcd (OR, 1.084 [95% CI, 1.043-1.127]), Pd (OR, 1.049 [95% CI, 1.004-1.096]) were associated with early death in APE. New LR model (area under the curve [AUC], 0.970) performed better than the simplified pulmonary embolism severity index (sPESI) score (AUC, 0.859) in predicting early death in APE (P = .021). The predictivity of the sPESI score significantly improved after its single combination with creatinine, QTcd, or troponin I. When the combined model was constructed together with these 6 independent variables and sPESI score, stepwise LR model automatically excluded Pd and NLR, and the AUC from the rest of the combined model was 0.976, which is significantly different from the AUC of sPESI (0.859) (P = .0031). Conclusions: Creatinine, troponin I, and QTcd significantly improves sPESI score. A new model with troponin I, creatinine, mean platelet volume, NLR, QTcd, and Pd seems to have greater prognostic power than the sPESI scoring system. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Acute pulmonary embolism (APE) remains associated with high morbidity and mortality rates despite advanced therapeutic options. This may be rooted partially in deficient initial prognostic assessment of patients. Both European and US guidelines suggest more aggressive treatment, such as administration of thrombolytic agents, for those at high risk for early mortality [1]. Moreover, some studies indicate that thrombolytic therapy may have a place in the management of some moderate-risk patients [2]. Risk stratification of patients with APE is mandatory to allow assessment of the individual prognosis and guide therapeutic decision making. Interestingly, clinical scores have been developed and validated to predict short-term prognosis after APE [3-5]. The simplified pulmonary embolism severity index (sPESI) is the most extensively studied clinical score to date [3]. It includes 6

☆ This report was neither previously submitted elsewhere nor under review process. ⁎ Corresponding author. Adnan Menderes Universitesi, Tıp Fakültesi, Kardiyoloji Anabilim Dali, 090100, Aytepe, Aydin, Turkey. Tel.: +90 256 444 12 56 2215; fax: +90 256 213 60 64. E-mail address: [email protected] (Ç. Akgüllü).

equally weighted variables: older than 80 years, history of cancer, history of chronic cardiopulmonary disease, heart rate greater than 110 beat per minute, systolic blood pressure less than 100 mm Hg, and arterial oxyhemoglobin saturation less than 90%. Other than these useful clinical scores, there is no clear consensus on the use of thrombolytic therapy, especially in intermediate-risk patients, defined as those without shock or hypotension but with adverse event predictors, such as elevated serum markers or right ventricular (RV) dysfunction by diagnosed by transthoracic echocardiography (TTE). To highlight the problem, some studies have suggested that laboratory biomarkers, particularly cardiac troponins but also electrocardiographic (ECG) and echocardiographic parameters and some complete blood cell count indices, could be useful in identifying patients with an elevated risk of death and complications during the acute phase of pulmonary embolism (PE) [6-10]. In addition, data indicate that serum creatinine levels may have prognostic importance in PE [11]. A positive characteristic of these prognostic markers is that all can be achieved easily in the emergency department (ED). To the author's best knowledge, there are no data in the literature about combined use of these prognostic markers in the course of APE. We aimed to determine the predictive abilities of ECG, TTE, laboratory

http://dx.doi.org/10.1016/j.ajem.2014.11.022 0735-6757/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

2

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

markers, blood cell count indices, and associated clinical conditions regarding early death in the course of APE. In addition, we used logistic regression (LR) analyses of independent variables to derive a valuable model to predict those at high risk. We also tested the additive prognostic determination values of these independent variables when combining them to the sPESI score.

patients with atrial fibrillation during admission, excessive noise in ECG, or incomplete TTE data. The study protocol was approved by the Ethics Committee of the University of Adnan Menderes, and we followed all procedures in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008.

2. Materials and methods

2.2. Diagnosis of PE

2.1. Patients

We diagnosed PE by spiral CT pulmonary angiography with the direct visualization of an intraluminal filling defect in the course of acute symptoms and signs suggesting PE.

This study was conducted in Adnan Menderes University Faculty of Medicine in Aydın, Turkey. We retrospectively reviewed archived data of 288 patients who had been hospitalized in Faculty Hospital between January 2011 and April 2013 with the diagnosis of PE and who received final diagnoses after exact demonstration of thrombus in pulmonary arteries via computed tomography (CT). The study included 206 patients who were in sinus rhythm, suitable for QT and P wave analyses, and who had accessible TTE data and laboratory test results. Eighty-two patients were excluded from the study (36 had incomplete TTE data, TTE was not performed on 12 of them, 11 had atrial fibrillation in the ECG, 9 had excessive noise in the ECG, and 14 patients had missing laboratory data) (Fig. 1). We recorded patients' baseline characteristics (sex, ages, etc), comorbidities, symptoms, hemodynamic conditions, all-cause mortality rates during hospitalization, total hospitalization times, radiographic test results, and laboratory findings, and we evaluated ECGs and TTEs obtained during admission (Table 1). Using collected baseline data at the time of PE diagnosis and the outcome data of this cohort, we retrospectively assessed sPESI scores for all patients. We excluded

2.3. Definitions We defined early death as inhospital mortality. We defined major bleeding as (1) a fall in hemoglobin of 2 g/dL, (2) transfusion of 2 U or more of red blood cells, (3) symptoms in a critical organ (intracranial, intraspinal, intraocular, retroperitoneal, intraarticular or pericardial, or intramuscular with compartment syndrome), or (4) fatal bleeding. Otherwise, we defined bleeding as minor bleeding. We defined the appropriate ECG to have at least 10 analyzable leads for the needed measurements. Otherwise, ECG was defined to have excessive noise. If the TTE data included all needed measurements (such as RV and left ventricular [LV] diameters, ejection fraction, or pulmonary artery pressure), then we defined it to be complete; otherwise, we defined it to be incomplete data. We defined hemodynamic instability as (1) PE causing hypotension (systolic blood pressure ≤90 mm Hg or a reduction of at least 40 mm Hg

Fig. 1. A diagram depicting the inclusion phase of the patients. Abbreviation: AF, atrial fibrillation.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

3

Table 1 Baseline characteristics of the study population and their comparison between the deceased and survivors in the course of APE

Age (y) Sex Female Male Deep venous thrombosis Diabetes mellitus Hypertension Chronic obstructive pulmonary disease Malignancy Smoking Hemodynamic instability Thrombolytic therapy Minor bleeding QTcd (milliseconds) Pd (milliseconds) P pulmonale T-wave abnormalities ST-segment depression S1Q3T3 Incomplete RBBB Complete RBBB Right axis deviation Left axis deviation Precordial low voltage Tachycardia RV hypertrophy Hemoglobin (g/dL) MCV (fl) RDW (%) Platelet count (per mm3) MPV (fl) PDW (%) NLR Troponin I (mg/dL) D-Dimer (μg/dL) Creatinine (mg/dL) Ejection fraction (%) Estimated PAP LV end-diastolic diameter RV end-diastolic diameter RV/LV ratio D-shaped septum sPESI score

All patients

Deaths (n = 30)

Survivors (n = 176)

61.8 ± 11.8

61.8 ± 2.16

59.06 ± 1.04

P .310

109 (53%) 97 (47%) 35 (17%) 61 (30%) 124 (60%) 58 (28%) 43 (21%) 89 (43.2%) 57 (27.7%) 25 (12.1%) 14 (6.79%) 79 (73.5-95.13) 49.5 (37–58) 8 (3.9%) 86 (41.7%) 46 (22.3%) 10 (4.9%) 12 (5.8%) 14 (6.8%) 16 (7.8%) 20 (9.7%) 8 (3.9%) 112 (54.4%) 8 (3.9%) 12.12 ± 1.75 84.9 ± 8.49 15.5 (14.2-16.6) 283 000 (214 250-336 250) 8.8 (8.2-9.8) 50.75 (46.1-56.17) 5 (2.70-8.90) 0.10 (0-2.12) 3200 (1700-5125)

7 (23.3%) 23 (76.7%) 7 (23.3%) 9 (30%) 18 (60%) 7 (23.3%) 15 (50%) 17 (56.7%) 18 (60%) 8 (26.7%) 3 (10.0%) 104 (97.37-118.85) 72.5 (53.75-78.5) 4 (13.3%) 20 (66.7%) 10 (33.3%) 6 (20%) 6 (20%) 5 (16.7%) 6 (20%) 6 (20%) 2 (6.7%) 17 (56.7%) 2 (6.7%) 11.71 ± 2.02 85 (81.23-88.93) 15.6 (14.8-17.2) 227 000 (162 000-338 250) 10.2 (8.75-11.2) 52.8 (50.02-64.4) 12.2 (8.37-21.02) 5.6 (2.5-13.15) 4450 (3200-5900)

102 (58%) 74 (42%) 28 (15.9%) 52 (29.5%) 106 (49.4%) 51 (29%) 28 (15.9%) 72 (40.9%) 39 (22.2%) 17 (9.7%) 11 (6.3%) 78 (68.12-84) 48 (35.25-55.5) 4 (2.3%) 66 (37.5%) 36 (20.5%) 4 (2.3%) 6 (3.4%) 9 (5.1%) 10 (5.7%) 14 (8%) 6 (3.4%) 95 (54%) 6 (3.4%) 12.19 ± 1.69 85.6 (79.8-90.08) 15.5 (14.2-16.6) 284 000 (229 000-336 750) 8.8 (8.2-9.5) 49.75 (45.42-54.4) 3.9 (2.6-7.3) 0 (0.0-1.2) 2700 (1450-4500)

.001 .461 1.000 1.000 .678 b.001 .158 b.001 .015 .435 b.001 b.001 .017 .005 .184 .001 .003 .036 .016 .086 .329 .940 .329 .162 .833 .477 .051 b.001 .001 b.001 b.001 b.001

0.81 (0.70-1.20) 66 (62-68) 36 (33-42) 48 (46-51) 46 (43.75-48) 0.96 (0.91-1) 77 (37.4%) 2 (1-2)

1.7 (0.9-2.32) 64 (59.5-66.5) 45.5 (36-56.5) 49 (47-51.25) 49 (45.75-51.25) 1.01 (0.95-1.06) 22 (73.3%) 3 (2-4)

0.8 (0.69-1.0) 66 (63-68) 36 (33-39.75) 48 (46-50.75) 46 (43-48) 0.95 (0.90-1.0) 55 (31.3%) 2 (1-2)

b.001 .015 b.001 .240 b.001 .003 b.001 b.001

Abbreviations: PDW, platelet distribution width; MCV, mean corpuscular volume; PAP, pulmonary arterial pressure.

for at least 15 minutes); (2) tachycardia (pulse rate ≥ 110/min); (3) hypoxemia (Pa O 2 b 60 mm Hg); (4) blurred consciousness; or (5) syncope associated with hypotension; oliguria; or cool, clammy extremities. We used the definition of “cool and clammy extremities” as a result of the physical examination notes of the physicians who had first contact in the ED.

study did not have access to the equipment to perform off-line analyses from digitally stored prior images and, thus, did not use this useful echocardiographic parameter, which has become accepted as a good predictor of RV longitudinal function.

2.4. Transthoracic echocardiography

To the best of our knowledge, there are no data in the literature to support the usefulness of P wave dispersion (Pd) in the determination of prognosis of APE. However, both Pd and corrected QT interval dispersion (QTcd) were previously shown to be increased in the setting of pulmonary hypertension [13,14]. Moreover, recently, QTcd was shown to be useful in the prognostic determination of APE [15]. To make the analysis, we followed similar ECG measurement techniques. We analyzed ECGs that had been recorded on admission using a supine, standard 12-lead ECG tracing at 25 mm/s paper speed at 10 mm/mV amplitude. We performed measurements manually with the help of a magnifying glass by 2 experienced cardiologists who were blind to clinical data of the patients. We used 3 consecutive beats for the analyses where at least 10 leads were analyzable in ECGs. We measured QT intervals from the onset of the QRS to the end of the T wave, defined as the return to the T-P isoelectric baseline. Besides,

In this study, we used recorded data from comprehensive TTEs performed on admission on subjects in the left lateral decubitus position (Philips HD11XE and Envisor HD; Phillips USA, Andover, MA). We acquired images from standard echocardiographic views in accordance with the recommendations of the American Society of Echocardiography [12]. To calculate the RV/LV ratio, we used archived measurements of the midcavitary end-diastolic diameters recorded from the standard, apical 4-chamber view. In addition, we used archived data about pulmonary artery systolic pressure, which was derived as the sum of the tricuspid regurgitant gradient obtained from continuous wave Doppler and the right atrial pressure as estimated from the inferior vena cava [12]. A great deal of data pertaining to RV tricuspid annular plane systolic excursion measurements was missing from the archives. In addition, the

2.5. Analyses of ECG, corrected QT interval dispersion, and P wave dispersion measurements

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

4

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

we accepted only monophasic, well-defined T waves for measurement. When U waves were present, we measured the QT interval to the nadir of the curve between the T and the U waves. We defined QTcd as the difference between the maximum and minimum QT values. Bazett's formula (QT corrected [QTc] = QT/√RR) was used for the QTc and QTcd analyses [16]. We measured P wave duration from the point of junction between the isoelectric line and the beginning of P wave deflection to the point of junction between the end of P wave deflection and the isoelectric line. We defined Pd as the difference between the maximum (Pmax) and minimum (Pmin) P wave durations occurring in any of the 12 leads. To estimate intraobserver and interobserver variability, we reanalyzed 33 randomized ECGs, and we repeated the measurements by 2 cardiologists. To determine intraobserver variability, we evaluated 93 ECG tracings by the same investigator on 2 occasions. To assess interobserver variation, all ECG tracings were analyzed by a second, independent investigator who was blind to the results obtained by the first. 2.6. Statistical analyses We used the Kolmogorov-Smirnov test to assess the normality of numeric variables. For those that were normally distributed, we made comparison between 2 groups by independent-sample t test, and we present descriptive statistics as mean ± SD. For numeric variables that were not normally distributed, we made comparison between 2 groups by Mann-Whitney U test, and we present descriptive statistics as median (25-75 percentiles). To analyze categorical data, we used a χ 2 test, and we present descriptive statistics as frequency (percentages). To determine the reliability and reproducibility of ECG parameters, we used the Spearman's ρ correlation analysis to evaluate their interobserver and intraobserver variability. The study used LR with a forward stepwise variable selection to predict the probability of early death in APE. We assessed fit by χ2 statistics proposed by a Hosmer-Lemeshow goodness-of-fit test. Furthermore, we used the receiver operating characteristic (ROC) curve to determine the cutoff point, the area under the curve (AUC) of sPESI, and predictors in the best LR model. In addition, the study tested the statistical significance of the difference between the areas under 2 ROC curves, the best LR model, and sPESI score, using the method proposed by Hanley and McNeil. In conclusion, we considered the P values less than .05 statistically significant. 3. Results We excluded 1 patient who had died because of a malignancy and 2 patients who had had pneumonia, as their baseline ECGs were not appropriate for further analyses. In addition, we excluded 2 patients who had died of sepsis because they had had atrial fibrillation. In addition, we excluded 1 patient who had died of PE and 2 patients who had died because of major bleeding from the study because they lacked initial echocardiographic evaluations. After excluding these inappropriate deaths, all deaths included in the study were related to PE. Among the study population, 30 patients (14.5%) died, and 176 patients (85.5%) lived after diagnosis of APE. Table 1 shows baseline characteristics of the study population. Mean age was 61.8 ± 11.8 years. Frequency of females in the population was 53% (n = 109), and 47% were male (n = 97). Frequency of hypertension was 60% (n = 124), diabetes mellitus was 30% (n = 61), chronic obstructive pulmonary disease was 28% (n = 58), deep venous thrombosis was 17% (n = 35), and history of malignancy was 21% (n = 43). In addition, 43% (n = 89) were smokers. Mean total hospitalization time was 7.0 (4.0-9.0) days, and mean hospitalization to death time was 3.0 (2.0-5.0) days. Mean hemoglobin of the population was 12.12 ± 1.75 g/dL, mean troponin I was 0.10 (0-2.12), and mean D-dimer was 3200 (1700-5125). Mean QTcd was 83.59 ± 20.01 milliseconds, and mean

Pd was 49.50 (37-58) milliseconds. The mean longest P wave duration was 101 (87-112) milliseconds, and the mean shortest was 47 (42-60) milliseconds. In addition, 27.7% (n = 57) of patients were hemodynamically unstable, and 12.1% (n = 25) had had thrombolytic therapy. There was no major bleeding, and minor bleeding occurred in 6.8% (n = 14) of patients. Mean total hospitalization time was 7.00 (4.00-9.00) days, and mean hospitalization to time of death was 3.00 (2.00-5.00) days (Table 1). Table 1 also shows a comparison of baseline characteristics of the deceased and survivors in the course of APE. There was no difference between survivors and deceased in age; smoking habits; and accompanying diseases, such as diabetes, hypertension, chronic obstructive pulmonary disease, and deep venous thrombosis (P N .05) (Table 1). However, there was a significant difference in sex, with more males dying (76.7% vs 23.3%; P = .001). Malignancy was more frequent in the deceased (50%; n = 15) than in survivors (15.9%; n = 28) (P b .001). Hemodynamic instability was a significant factor, with those having it who died being 60% (n = 18) and those having it who lived being 22% (n = 39) (P b .001). Frequency of thrombolytic therapy was higher among those who died (26.7%; n = 8) than among those who lived (9.7%; n = 17) (P = .015). Total hospitalization time, at 3 days (2-5.25), was less among those who died than the 7 days (5-9.75) among those who survived (P b .001). The sPESI scores of those who died, at 3 (2-4), were higher than those of the survivors, at 2 (1-2) (P b .001). D -Dimer levels of those who died were higher, at 4450 μg/dL (3200-5900), than D -dimer levels of survivors, at 2700 μg/dL (1450-4500) (P ≤ .001). In addition, troponin I serum levels were higher, at 5.6 mg/dL (2.5-13.15), in those who died than in those who lived, in which they were 0 mg/dL (0.0-1.2) (P b .001). Creatinine levels were also higher among those who died (1.70 mg/dL [0.90-2.32] vs 0.80 mg/dL [0.69-1.00]; P b .001). There was no significant difference between the groups in hematological parameters, such as hemoglobin, mean corpuscular volume, and red blood cell distribution width (RDW) levels. However, platelet distribution width (P = .001), mean platelet volume (MPV) (P b .001), and neutrophil to lymphocyte ratio (NLR) (P b .001) were significantly higher in those who died (Table 2). There was no difference between the groups in ECG findings, such as left axis deviation, ST-segment depression, low voltage, findings of RV hypertrophy, and tachycardia (Table 1). However, the incidence on ECGs of complete and incomplete right bundle-branch block (RBBB) was higher in those who died (P = .036 and P = .003, respectively). Electrocardiographic findings of nonspecific T-wave abnormalities (P = .005), right axis deviation (P = .016), P pulmonale (P = .017), and S1Q3T3 (P = .001) were higher in those who died (Table 1). P wave dispersion values of those who died were higher than they were in those who survived: 52.80 (50.02-64.40) milliseconds and 49.75 (45.42-54.40) milliseconds, respectively (P b .001). In addition, QTcd values of those who died were higher than those of survivors: 104.00 (97.37-118.85) milliseconds and 78.00 (68.12-84.00) milliseconds, respectively (P b .001). Reproducibility of the determination of QTcd and Pd was high in both intraobserver and interobserver comparisons. To estimate Table 2 Multivariable LR analysis model to predict early death in APE Variables

OR

95% CI

P

Troponin Creatinine MPV NLR QTcd Pd

1.084 4.153 1.991 1.079 1.084 1.049

1.009-1165 1.375-12.541 1.230-3.223 1.005-1.160 1.043-1.127 1.004-1.096

.027 .012 .005 .037 b.001 .031

Odds ratios with respective 95% CIs for early mortality were calculated by LR analysis.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

intraobserver and interobserver variability, we reanalyzed 93 randomized ECGs, and measurements were repeated by both cardiologists. Intraobserver variability for QTcd (r = 0.976; P b .001) and for Pd (r = 0.988; P b .001) was not significant. Interobserver variability between the 2 cardiologists estimating QTcd (r = 0.982; P b .001) and Pd (r = 0.994; P b .001) was not significant. Among TTE parameters, there was no difference between the groups in LV end-diastolic diameter (P = .240). However, RV end-diastolic diameter, pulmonary artery systolic pressure, presence of D-shaped septum (P b .001), and RV/LV ratio (P = .003) were higher among those who died. Ejection fraction was lower in those who died (P = .015) (Table 1). Using independent predictors, we used LR analysis with a forward stepwise variable selection to predict the probability of early death in APE. We assessed fit by χ 2 statistics proposed by the HosmerLemeshow goodness-of-fit test. The LR model showed a good fit based on the χ2 of Hosmer-Lemeshow goodness-of-fit statistics (P N .05). Results of LR analysis showed that troponin I (odds ratio [OR], 1.084 [95% confidence interval {CI}, 1.009-1.165]), creatinine (OR, 4.153 [95% CI, 1.375-12.541]), MPV (OR, 1.991 [95% CI, 1.230-3.223]), NLR (OR, 1.079 [95% CI, 1.005-1.160]), QTcd (OR, 1.084 [95% CI, 1.043-1.127]), and Pd (OR, 1.049 [95% CI, 1.004-1.096]) were associated with early death in acute PE. The study found the best model by LR analysis (Table 2). P ðearly deathÞ ¼

1 1 þ e−ð−9:894þ0:081troponinþ1:424creatinineþ0:689MPVþ0:076NLRþ0:081QTcdþ0:048PdÞ

Table 3 shows a comparison of the cutoff value, AUC, 95% CIs of AUC, sensitivity, and specificity for predictors in the best LR model. Troponin I, creatinine, MPV, NLR, QTcd, and Pd all had AUCs, sensitivities, and specificities in the ranges of 0.719 to 0.895, 60.00% to 96.67%, and 77.84% to 95.45%, respectively. As Fig. 2 shows, the LR model (AUC, 0.970) performed better than sPESI scores (AUC, 0.859) in predicting early death in APE (P = .021). We created LR models to determine the predictive abilities of 6 variables together with sPESI. We added our new independent variables one by one to the sPESI score to derive a new LR model each time to demonstrate the contribution values of each variable to the sPESI. Then, ROC analyses were performed using the predictive probabilities of these 6 LR models (Table 4). The most valuable contributions came from creatinine, QTcd, troponin I, and MPV, whereas Pd and NLR seemed to have less valuable contributions to the predictivity of the sPESI (Fig. 3). We also checked whether the difference between the ROC analyses of each new LR model and the sPESI score had any statistical significance. We demonstrated that the predictive value of the ROC analyses of the LR models, sPESI + creatinine (P = .0317), sPESI + QTcd (P = .0132), and sPESI + troponin (P = .0245), were different from the predictive value of the ROC analyses of the sPESI, where sPESI + MPV (P = .0587) had borderline P value and sPESI + Pd (P = .2846) and sPESI + NLR (P = .0793) had no difference. We added all 6 variables of our model together to the sPESI to demonstrate their combined predictive value. When we ran a stepwise LR Table 3 Receiver operating characteristic curve analysis of variables that significantly predict early death Variables and cutoff values AUC Troponin N2.3 Creatinine N1.35 MPV N9.7 NLR N8.4 QTcd N86 Pd N64.5 sPESI N2

0.879 0.823 0.719 0.825 0.895 0.795 0.859

0.827-0.920 0.764-0.873 0.653-0.779 0.766-0.875 0.845-0.933 0.733-0.848 0.804-0.903

P

Sensitivity Specificity

b.0001 b.0001 .0001 b.0001 b.0001 b.0001 b.0001

83.33 66.67 63.33 76.67 96.67 60.0 73.33

88.07 88.64 81.25 81.82 77.84 95.45 87.5

5

model for the combined 6 variables and sPESI together, Pd and NLR were excluded automatically from the analysis model. The AUC of the combined model was 0.976 (creatinine, MPV, QTcd, troponin I, and sPESI score together) (Table 4) (Fig. 2). The difference between the AUC of the sPESI score and the combined model was significant (P = .0031). The difference between the AUC of our new LR model (6 variables together) and combined model was not significant (P = .8020). We have demonstrated that our model increased the predictivity of the sPESI score, but the sPESI score did not improve the predictivity in our model. 4. Discussion With a unicenter cohort of 206 patients with APE, we demonstrated that ECG findings of QTc and Pd values, serum creatinine and troponin I levels, and hematological parameters of the MPV and NLR, all of which are easily detected in the ED, constitute a useful model to predict early death in the course of APE. In addition, we demonstrated that this model is significantly more powerful than the sPESI scoring system in predicting high-risk patients with APE. Clinical presentation of APE varies from a small, asymptomatic pulmonary embolus with low mortality to a massive PE resulting in RV heart failure, shock, and/or death. Incidence of mortality is directly related to management strategy, which, in turn, is related to accurate risk stratification of patients. The crucial importance of identifying patients who are at high risk has given rise to studies focusing on detection of powerful, new predictive tools. Interestingly, some studies have demonstrated that laboratory markers, such as cardiac troponins, D-dimer, natriuretic peptides, heart-type fatty acids, and serum homocysteine levels, can be useful prediction tools [17-22]. In addition, some parameters among imaging modalities, especially focusing on the assessment of RV and right atrial size and function, have been shown to be useful [23-25]. In addition, reports indicate that renal failure may play a crucial role in the risk stratification of patients with APE [11,26]. Some reports evaluate the usefulness of complete blood count indices, such as NLR, MPV, and RDW, in predicting patients with high risk [2730]. Finally, reports evaluate the QT dispersion parameter of the surface ECG as a useful predictive tool in the course of APE [15,10]. Several classical ECG abnormalities are associated with APE, but none are highly specific or sensitive. In patients with RV failure, the ECG may show sinus tachycardia, signs of RV strain, repolarization abnormalities, ischemia including complete or incomplete RBBB, a rightward axis greater than 90°, an S1Q3T3 pattern, a Qr in lead V1, ST elevation in V1, or precordial T-wave inversions. However, approximately two-thirds of patients with massive or submassive PEs exhibit no such classical changes on ECG [31]. Another parameter of the surface ECG, QTcd, has been shown to have an important role in predicting mortality in APE [15,10]. Corrected QT interval dispersion is suggested as a good marker for detection of inhomogeneity of ventricular recovery times. The rationale for this close relation of QTcd and prognosis of PE may be explained partly by acute repolarization changes of the overloaded RV. In the context of acute RV failure, advanced heterogeneity of repolarization may lead to increased QTcd that can be detected easily on a surface ECG [32]. Interestingly, Ermis et al [15] suggested that the sensitivity of QTcd greater than 71.5 milliseconds for prediction of mortality was 71%, with a specificity of 73% in patients with APE [15]. Thus, it is not surprising that this study found that QTcd had significant importance in the LR analyses. Another parameter of the surface ECG is Pd. This ECG parameter, to the author's best knowledge, has not been studied previously as a prognostic marker in the context of APE. P wave dispersion is regarded as a marker of atrial depolarization heterogeneity and has been suggested as reflecting abnormalities of atrial size and structure [33,34]. During acute RV failure, as RV end-diastolic volumes and pressures increase, RV wall stress increases, leading to reduced RV stroke volume. In addition, elevated RV end-diastolic volumes promote tricuspid annular

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

6

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

Fig. 2. Comparison of ROC curves of the new model, sPESI score, and combined model in predicting early mortality in APE. Variables of logistic regression model: troponin I, creatinine, MPV, NLR, QTcd, and Pd. Variables of sPESI: older than 80 years, history of cancer, history of chronic cardiopulmonary disease, heart rate greater than 110 beats per minute, systolic blood pressure less than 100 mm Hg, and arterial oxyhemoglobin saturation less than 90%. Variables of the combined model: troponin I, creatinine, MPV, QTcd, older than 80 years, history of cancer, history of chronic cardiopulmonary disease, heart rate greater than 110 beats per minute, systolic blood pressure less than 100 mm Hg, and arterial oxyhemoglobin saturation less than 90%.

dilatation, which worsens tricuspid valve insufficiency and increases right atrial pressure and atrial wall tension. Previously, Barra et al [35] demonstrated that, in the course of APE, atrial fibrillation is a predictor of bad prognosis. In addition, it has been shown recently that Pd is a predictor of atrial fibrillation [36]. Thus, in this study, we sought to determine Pd's prognostic power in APE and demonstrated that Pd can be a useful ECG predictor in APE. In addition, we showed the myocardial injury marker troponin I to be a significantly important prognostic marker in our LR analysis

model. Its importance to risk stratification has been shown previously in APE [19,37]. Moreover, recent studies have also demonstrated that the use of new, highly sensitive cardiac troponins improves diagnostic sensitivity in APE [38]. Interestingly, Lankeit et al [39] have shown that a combination of highly sensitive troponin T and sPESI score improved risk stratification of normotensive patients with APE. We have shown that serum creatinine levels play a significant role in detecting patients at high risk. Interestingly, Kostrubiec et al [11] have shown that rapid improvement of renal function in patients with APE

Table 4 Receiver operating characteristic curve analysis of variables combined with sPESI score to predict early death LR models

Sensitivity

Specificity

AUC

95% CI for AUC

P

sPESI LR model (Cr + MPV + NLR + QTcd + Pd + troponin I) Combined model (sPESI + Cr + troponin I + QTcd + MPV) sPESI + Cr sPESI + MPV sPESI + NLR sPESI + Pd sPESI + QTcd sPESI + troponin I

73.33 90.0 96.7 90.00 86.67 90 80 96.67 93.33

87.5 93.7 92.6 92.05 85.23 79.55 86.36 73.3 82.39

0.859 0.970 0.976 0.933 0.909 0.906 0.892 0.928 0.921

0.804-0.903 0.937-0.989 0.945-0.992 0.890-0.963 0.861-0.944 0.858-0.942 0.842-0.931 0.883-0.959 0.875-0.954

b.001 b.001 b.001 b.001 b.001 b.001 b.001 b.001 b.001

Abbreviation: Cr, creatinine.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

7

Fig. 3. Comparison of ROC curves of the independent variables when combined with sPESI score in predicting early mortality in APE.

may indicate a favorable short-term prognosis. In another study, they showed that impaired kidney function is related to all-cause mortality in APE. They found that, in initially normotensive patients, a glomerular filtration rate (GFR) less than 35 mL/min predicts 30-day mortality. They concluded that GFR assessment improved troponin-based risk stratification of APE [26]. In this study, we did not use GFR; however, we found that serum creatinine levels can be used as a good prognostic tool in APE. Lastly, we found that complete blood count indices, such as MPV and NLR, play a significant role in prediction of prognosis in APE. The balance between neutrophils and lymphocytes has been regarded as a marker of systemic inflammation [40]. Kayrak et al [27] previously suggested that NLR may play an important role in predicting high-risk patients. On the other hand, MPV, which is accepted as an indicator of platelet activation, has been shown to have prognostic value in APE [29,41]. We suggest that both indices, which can be derived from a simple complete blood count test, can be good prognostic tools in the context of APE. In this study, analyses of QTcd and Pd were performed manually by 2 experienced cardiologists. We demonstrated that reproducibility of the determination of QTcd and Pd was high in both intraobserver and interobserver comparisons. However, in real life, it may be a little time consuming to perform this analysis manually, especially in emergency situations. Of interest, it was previously shown that automatic QTcd measurements may be as reliable as manual QTcd measurement techniques [42-44]. There are no data in the literature about automatic Pd measurement techniques or comparison of automatic vs

manual analyses. However, it may be possible in the future after the essential studies. Current guidelines recommend prognostic evaluation to stratify patients into risk groups after a diagnosis of APE. Among the recommended evaluation systems are sPESI, Geneva, and Wells [4]. However, none of these scoring systems take into account the troponins, creatinine levels, MPV, NLR, QTcd, or Pd. In this study, we concluded that combined use of these predictors may have greater prognostic power than the sPESI scoring system. The most prominent limitation of our study was relatively small number of outcome of interest (death number) among the study population. The power of our analysis might have been affected from the relatively small number. Large, randomized studies are needed to assess the importance of these markers. In addition, in future studies, it would be valuable to assess the probable improvement power of these prognostic markers when combined with TTE or CT parameters, such as tricuspid annular plane systolic excursion or CT obstruction indices, or with laboratory markers, such as heart-type fatty acids, binding proteins, or natriuretics.

References [1] Torbicki A, Perrier A, Konstantinides S, Agnelli G, Galiè N, Pruszczyk P, et al. Guidelines on the diagnosis and management of acute pulmonary embolism. The Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J 2008;29:2276–315.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

8

Ç. Akgüllü et al. / American Journal of Emergency Medicine xxx (2014) xxx–xxx

[2] Chatterjee S, Chakraborty A, Weinberg I, Kadakia M, Wilensky RL, Sardar P, et al. Thrombolysis for pulmonary embolism and risk of all-cause mortality, major bleeding, and intracranial hemorrhage: a meta-analysis. JAMA 2014;311(23):2414–21. [3] Jimenez D, Aujesky D, Moores L, Gomez V, Lobo JL, Uresandi F, et al. Simplification of the pulmonary embolism severity index for prognostication in patients with acute symptomatic pulmonary embolism. Arch Intern Med 2010;170:1383–9. [4] Aujesky D, Roy PM, Le Manach CP, Verschuren F, Meyer G, Obrosky DS, et al. Validation of a model to predict adverse outcomes in patients with pulmonary embolism. Eur Heart J 2006;27:476–81. [5] Wicki J, Perrier A, Perneger TV, Bounameaux H, Junod AF. Predicting adverse outcome in patients with acute pulmonary embolism: a risk score. Thromb Haemost 2000;84:548–52. [6] Becattini C, Vedovati MC, Agnelli G. Prognostic value of troponins in acute pulmonary embolism: a meta-analysis. Circulation 2007;116:427–33. [7] Chow V, Ng AC, Chung T, Thomas L, Kritharides L. Right atrial to left atrial area ratio on early echocardiography predicts long-term survival after acute pulmonary embolism. Cardiovasc Ultrasound 2013;31:11–7. [8] Pruszczyk P, Goliszek S, Lichodziejewska B, Kostrubiec M, Ciurzyński M, Kurnicka K, et al. Prognostic value of echocardiography in normotensive patients with acute pulmonary embolism. JACC Cardiovasc Imaging 2014;7(6):553–60. [9] Sen HS, Abakay O, Tanrikulu AC, Sezgi C, Taylan M, Abakay A, et al. Is a complete blood cell count useful in determining the prognosis of pulmonary embolism? Wien Klin Wochenschr 2014;126(11-12):347–54. [10] Ding X, Zhang S, Pei Z. QT dispersion in acute pulmonary embolism. Zhong Nan Da Xue Xue Bao Yi Xue Ban 2013;38(4):395–9. [11] Kostrubiec M, Łabyk A, Pedowska-Włoszek J, Pacho S, Dzikowska-Diduch O, Dul P, et al. Rapid improvement of renal function in patients with acute pulmonary embolism indicates favorable short term prognosis. Thromb Res 2012;130(3):37–42. [12] Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, et al. Recommendations for chamber quantification: a report from the American Society of Echocardiography's guidelines and standards committee and the chamber quantification writing group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:1440–63. [13] Sap F, Karataş Z, Altin H, Alp H, Oran B, Baysal T, et al. Dispersion durations of P-wave and QT interval in children with congenital heart disease and pulmonary arterial hypertension. Pediatr Cardiol 2013;34(3):591–6. [14] Hong-liang Z, Qin L, Zhi-hong L, Zhi-hui Z, Chang-ming X, Xin-hai N, et al. Heart ratecorrected QT interval and QT dispersion in patients with pulmonary hypertension. Wien Klin Wochenschr 2009;121:330–3. [15] Ermis N, Ermis H, Sen N, Kepez A, Cuglan B. QT dispersion in patients with pulmonary embolism. Wien Klin Wochenschr 2010;122(23–24):691–7. [16] Bazett HC. An analysis of the time-relations of electrocardiograms. Heart 1920;7:353–70. [17] Zhu H, Xue H, Wang GY, Xing YH, Li TS, Qian YY, et al. Association between serum homocysteine and in-hospital death in patients with acute pulmonary embolism. Zhonghua Xin Xue Guan Bing Za Zhi 2013;41(9):756–60. [18] Ng PL, Kwok CH, Tsui SH, Tong HK. Raised troponin I in Chinese patients with acute pulmonary embolism suggests complicated clinical course and may predict intensive care unit admission. Singapore Med J 2013;54(2):86–9. [19] Walter T, Apfaltrer P, Weilbacher F, Meyer M, Schoenberg SO, Fink C, et al. Predictive value of high-sensitivity troponin I and D-dimer assays for adverse outcome in patients with acute pulmonary embolism. Exp Ther Med 2013;5(2):586–90. [20] Dellas C, Tschepe M, Seeber V, Zwiener I, Kuhnert K, Schäfer K, et al. A novel H-FABP assay and a fast prognostic score for risk assessment of normotensive pulmonary embolism. Thromb Haemost 2014;111(5):996–1003. [21] Lankeit M, Jiménez D, Kostrubiec M, Dellas C, Kuhnert K, Hasenfuß G, et al. Validation of N-terminal pro-brain natriuretic peptide cut-off values for risk stratification of pulmonary embolism. Eur Respir J 2014;43(6):1669–77. [22] Klok FA, Djurabi RK, Nijkeuter M, Eikenboom HC, Leebeek FW, Kramer MH, et al. High D-dimer level is associated with increased 15-d and 3 months mortality through a more central localization of pulmonary emboli and serious comorbidity. Br J Haematol 2008;140(2):218–22.

[23] Wolde M, Söhne M, Quak E, Mac Gillavry M, Büller HR. Prognostic value of echocardiographically assessed right ventricular dysfunction in patients with pulmonary embolism. Arch Intern Med 2004;164:1685–9. [24] Zhao DJ, Ma DQ, He W, Wang JJ, Xu Y, Guan CS. Cardiovascular parameters to assess the severity of acute pulmonary embolism with computed tomography. Acta Radiol 2010;51(4):413–9. [25] Chow V, Ng AC, Chung T, Thomas L, Kritharides L. Right atrial to left atrial area ratio on early echocardiography predicts long-term survival after acute pulmonary embolism. Cardiovasc Ultrasound 2013;11:17. [26] Kostrubiec M, Łabyk A, Pedowska-Włoszek J, Pacho S, Wojciechowski A, Jankowski K, et al. Assessment of renal dysfunction improves troponin-based short-term prognosis in patients with acute symptomatic pulmonary embolism. J Thromb Haemost 2010;8(4):651–8. [27] Kayrak M, Erdoğan HI, Solak Y, Akilli H, Gül EE, Yildirim O, et al. Prognostic value of neutrophil to lymphocyte ratio in patients with acute pulmonary embolism: a retrospective study. Heart Lung Circ 2014;23(1):56–62. [28] Günay E, Sarinc Ulasli S, Kacar E, Halici B, Unlu E, Tünay K, et al. Can platelet indices predict obstruction level of pulmonary vascular bed in patients with acute pulmonary embolism? Clin Respir J 2014;8(1):33–40. [29] Kostrubiec M, Łabyk A, Pedowska-Włoszek J, Hrynkiewicz-Szymańska A, Pacho S, Jankowski K, et al. Mean platelet volume predicts early death in acute pulmonary embolism. Heart 2010;96(6):460–5. [30] Abul Y, Ozsu S, Korkmaz A, Bulbul Y, Orem A, Ozlu T. Red cell distribution width: a new predictor for chronic thromboembolic pulmonary hypertension after pulmonary embolism. Chron Respir Dis 2014;11(2):73–81. [31] Tapson VF. Acute pulmonary embolism. Cardiol Clin 2004;22:353–65. [32] Zabel M, Portnoy S, Franz MR. Electrocardiographic indexes of dispersion of ventricular repolarization: an isolated heart validation study. J Am Coll Cardiol 1995;25: 746–52. [33] Dogan U, Dogan EA, Tekinalp M, Tokgoz OS, Aribas A, Akilli H, et al. P-wave dispersion for predicting paroxysmal atrial fibrillation in acute ischemic stroke. Int J Med Sci 2012;9(1):108–14. [34] HashemiJazi M, Amirpour A, Zavvar R, Behjati M, Gharipour M. Predictive value of Pwave duration and dispersion in post coronary artery bypass surgery atrial fibrillation. ARYA Atheroscler 2012;8(2):59–62. [35] Barra SN, Paiva LV, Providência R, Fernandes A, Leitão Marques A. Atrial fibrillation in acute pulmonary embolism: prognostic considerations. Emerg Med J 2014;31(4): 308–12. [36] Elmoniem AA, El-Hefny N, Wadi W. P wave dispersion (PWD) as a predictor of atrial fibrillation (AF). Int J Health Sci (Qassim) 2011;5:25–6. [37] Tong C, Zhang Z, Ma C, Zhu Y, Zhang X. The value of simplified pulmonary embolism severity index and biomarkers in evaluating pulmonary embolism prognosis. Zhonghua Jie He He Hu Xi Za Zhi 2014;37(2):104–8. [38] Ozsu S, Abul Y, Orem A, Oztuna F, Bulbul Y, Yaman H, et al. Predictive value of troponins and simplified pulmonary embolism severity index in patients with normotensive pulmonary embolism. Multidiscip Respir Med 2013;8(1):34. [39] Lankeit M, Jiménez D, Kostrubiec M, Dellas C, Hasenfuss G, Pruszczyk P, et al. 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 2011;124(24):2716–24. [40] Zahorec R. Ratio of neutrophil to lymphocyte counts rapid and simple parameter of systemic inflammation and stress in critically ill. Bratisl Lek Listy 2001;102(1):5–14. [41] Hilal E, Neslihan Y, Gazi G, Sinan T, Zeynep Ayfer A. Does the mean platelet volume have any importance in patients with acute pulmonary embolism? Wien Klin Wochenschr 2013;125(13–14):381–5. [42] Murray A, McLaughlin NB, Campbell RW. Measuring QT dispersion: man versus machine. Heart 1997;77(6):539–42. [43] McLaughlin NB, Campbell RW, Murray A. Accuracy of four automatic QT measurement techniques in cardiac patients and healthy subjects. Heart 1996;76(5):422–6. [44] Takase B, Masaki N, Hattori H, Ishihara M, Kurita A. Usefulness of automatic QT dispersion measurement for detecting exercise-induced myocardial ischemia. Anadolu Kardiyol Derg 2009;9(3):189–95.

Please cite this article as: Akgüllü Ç, et al, Predictors of early death in patients with acute pulmonary embolism, Am J Emerg Med (2014), http://dx. doi.org/10.1016/j.ajem.2014.11.022

Predictors of early death in patients with acute pulmonary embolism.

We aimed to determine the predictors of early death in the course of acute pulmonary embolism (APE)...
707KB Sizes 4 Downloads 7 Views