Prohormones in the Early Diagnosis of Cardiac Syncope.

BACKGROUND: The early detection of cardiac syncope is challenging. We aimed to evaluate the diagnostic value of 4 novel prohormones, quantifying different neurohumoral pathways, possibly involved in the pathophysiological features of cardiac syncope: midregional-pro-A-type natriuretic peptide (MRproANP), C-terminal proendothelin 1, copeptin, and midregional-proadrenomedullin. METHODS AND
RESULTS: We prospectively enrolled unselected patients presenting with syncope to the emergency department (ED) in a diagnostic multicenter study. ED probability of cardiac syncope was quantified by the treating ED physician using a visual analogue scale. Prohormones were measured in a blinded manner. Two independent cardiologists adjudicated the final diagnosis on the basis of all clinical information, including 1-year follow-up. Among 689 patients, cardiac syncope was the adjudicated final diagnosis in 125 (18%). Plasma concentrations of MRproANP, C-terminal proendothelin 1, copeptin, and midregional-proadrenomedullin were all significantly higher in patients with cardiac syncope compared with patients with other causes (P<0.001). The diagnostic accuracies for cardiac syncope, as quantified by the area under the curve, were 0.80 (95% confidence interval [CI], 0.76-0.84), 0.69 (95% CI, 0.64-0.74), 0.58 (95% CI, 0.52-0.63), and 0.68 (95% CI, 0.63-0.73), respectively. In conjunction with the ED probability (0.86; 95% CI, 0.82-0.90), MRproANP, but not the other prohormone, improved the area under the curve to 0.90 (95% CI, 0.87-0.93), which was significantly higher than for the ED probability alone (P=0.003). An algorithm to rule out cardiac syncope combining an MRproANP level of <77 pmol/L and an ED probability of <20% had a sensitivity and a negative predictive value of 99%.
CONCLUSIONS: The use of MRproANP significantly improves the early detection of cardiac syncope among unselected patients presenting to the ED with syncope. CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifier: NCT01548352.

Clinical Perspective

What Is New?

This is the first prospective analysis assessing the diagnostic role of 4 novel prohormones in patients presenting with syncope to the emergency department.

Diagnostic accuracy of midregional–pro‐A‐type natriuretic peptide for cardiac origin among unselected patients presenting with syncope to the emergency department is high and provides significantly incremental diagnostic value on top of clinical judgement and a recommended risk stratification tool.

What Are the Clinical Implications?

An algorithm based on the combination of midregional–pro‐A‐type natriuretic peptide and clinical judgement allowed us to rule out the presence of cardiac syncope with a sensitivity of 99% and a negative predictive value of 99% and may, therefore, improve patient care and logistics.

Although this study represents an important step towards the integration of biomarkers into the clinical management of patients presenting with syncope to the emergency department, external validation in a large diagnostic study of comparable methodological scrutiny is required.

Syncope is a transient loss of consciousness associated with an inability to maintain postural tone attributable to global cerebral hypoperfusion.1, 2 Syncope is common and represents 1% to 2% of all emergency department (ED) visits.3 Establishing the cause is challenging and, therefore, generally time and resource consuming. In the United States, 30% to 40% of such patients are subsequently admitted for further investigation at an annual cost of $2.4 billion, according to the Medicare database.4 The risk of death is doubled among patients with cardiac syncope in comparison to other causes.5, 6 Identification of these patients is, therefore, crucial. Several risk scores,7, 8, 9 and the establishment of specialized syncope units10, 11 in the ED, were proposed to improve the diagnostic yield to identify patients at risk of an adverse outcome. However, these tools have not been implemented in most institutions, at least in part because of their perceived complexity. Therefore, the rapid and accurate detection of cardiac syncope in the ED remains an unmet clinical need.

We hypothesized that blood biomarkers might provide incremental diagnostic value in the rapid and accurate detection of cardiac syncope in the ED, similar to the contribution that they have made for other common presenting symptoms, such as acute chest pain12, 13, 14, 15 and acute dyspnea.16, 17, 18

It is challenging to find a biomarker that could represent the “memory of the cardiac or arrhythmic event” at ED presentation, when the patient has again become asymptomatic and hemodynamically stable. The recent development and clinical introduction of assays that reliably quantify stable prohormone fragments, rather than the more unstable active hormone, have provided a new and unique diagnostic window.19, 20 Using this concept, immunoassays targeting 4 prohormone fragments, including midregional–pro‐A‐type natriuretic peptide (MRproANP), C‐terminal proendothelin 1 (CTproET1), copeptin, and midregional‐proadrenomedullin (MRproADM), were developed and applied in several clinical settings.18, 19, 20, 21, 22, 23 These prohormones allow us to quantify the activation of 4 distinct cardiovascular and biochemical pathways possibly involved in either the pathophysiological features of syncope or the endogenous response to it.18, 19, 20, 21, 22, 23, 24 As a first hint about the possible role of these prohormones in syncope, CTproET1 plasma concentrations were associated with malignant arrhythmias in patients with chronic heart failure.25

We, therefore, performed a large international study evaluating the diagnostic utility of MRproANP, CTproET1, copeptin, and MRproADM in patients presenting with syncope to the ED.

Methods

Study Design, Setting, and Selection of Participants

BASEL IX (Basel Syncope Evaluation Study) is an ongoing, prospective, international, diagnostic, multicenter study enrolling unselected patients in 13 hospitals in 8 countries (Switzerland, Spain, Germany, Italy, Poland, New Zealand, Australia, and the United States) on 3 continents (Europe, Australia, and North America; Table S1). The study is designed to contribute to improving the management of patients presenting with syncope. Patients aged >40 years presenting to the ED with syncope within the past 12 hours were recruited, after written informed consent was obtained.

For this analysis, the exclusion criteria included patients with missing prohormone fragment measurements, patients with a final diagnosis of a nonsyncopal loss of consciousness, and those in whom the final diagnosis remained unclear even after central adjudication. The study was performed according to the principles of the Declaration of Helsinki preregistered and was approved by the local ethics committees. The authors designed the study, gathered and analyzed the data according to the Standards for the Reporting of Diagnostic Accuracy Studies guidelines for studies of diagnostic accuracy (Data S1), vouch for the data and analysis, wrote the article, and made the decision to submit the article for publication.

Routine Clinical Assessment

Two sets of data were obtained: (1) by the treating ED physician as part of routine clinical care, according to local standard operating procedures; and (2) by an experienced research fellow, using standardized case report forms uniformly collecting predefined details of patient history, the circumstances of syncope, and physical examination findings. A digital 12‐lead ECG was recorded at presentation and stored electronically. Clinical judgment by the ED physician about the presence of cardiac syncope was quantified using a visual analogue scale ranging from 0% to 100%. The treating ED physician estimated cardiac origin probability on the basis of all information available in the individual patient 90 minutes after presentation, including clinical assessment, the 12‐lead ECG, and the routine laboratory test results (but not the ones assessed for the study). To also provide a comparison with a recommended standard, the “Evaluation of Guidelines in Syncope Study” (EGSYS) risk score was calculated (Data S1).1, 2 Clinical decisions were absolutely independent of the present study. All participating centers were using standardized operating procedures for the diagnostic workup of patients presenting with syncope to the ED, according to current European Society of Cardiology Guidelines.2

Biochemical Measurements

Immediately after informed consent was obtained, venous blood was obtained in EDTA plastic tubes, centrifuged, and stored at −80°C. Measurement of MRproANP, CTproET1, copeptin, and MRproADM was performed in a blinded manner in a dedicated core laboratory using validated sandwich immunoassays (Brahms, Hennigsdorf/Berlin, Germany).26, 27, 28, 29 In a subgroup of patients, B‐type natriuretic peptide (BNP) was measured as part of the local clinical standardized operating procedures for patients with syncope from fresh plasma using the Architect system (Abbott, Chicago, IL).

Follow‐Up and Adjudicated Final Diagnosis

Patients were contacted 12 and 24 months after discharge by telephone or in written form. Information about recurrent syncope, hospitalization, and cardiac events during follow‐up was further obtained from the patient's hospital notes, the family physician's records, and national registries on mortality, where possible. To determine the final diagnosis of the index syncopal event in each patient, 2 independent cardiologists reviewed all available medical records from both data sets: the clinical data set and the study‐specific data set after at least 12 months of follow‐up. The clinical data set included the clinical history, findings on physical examination, results of routine laboratory tests, radiologic testing results, ECG findings, and, if available, Holter‐ECG, external and implantable loop device, echocardiography, cardiac exercise test, Schellong test, tilt table testing, coronary angiography, electrophysiological examination, pacemaker control, and further investigation findings during recurrent hospitalizations or ambulatory treatment. A detailed overview of all performed diagnostic tests is given in Table S2. Study‐specific data included standardized forms uniformly collecting predefined details of patient history, the circumstances of syncope, the physical examination findings, and at least 12 months of follow‐up data. In situations of adjudicator disagreement about the diagnosis, cases were reviewed and adjudicated in conjunction with a third cardiologist. Predefined categories for the adjudication included cardiac syncope, reflex syncope, orthostatic syncope, other noncardiac syncope, and unknown cause of syncope (Table S3). According to guidelines,2 cardiac causes of syncope were defined as supraventricular or ventricular arrhythmias, severe structural heart diseases (eg, hypertrophic cardiomyopathy or valvular diseases), pericardial tamponade, congenital myocardial or valvular anomalies, aortic dissection, or acute pulmonary hypertension (eg, attributable to pulmonary embolism), leading to a transient loss of consciousness. It is important to highlight that the presence of cardiac disease (eg, coronary artery disease) was not at all sufficient for the adjudication to a cardiac cause of syncope. The detailed reconstruction of the syncopal event with the study‐specific data set and third‐party anamnesis, and long‐term follow‐up on cardiovascular events and/or recurrent syncope, were critical pillars of the adjudication. Further details on the adjudication are given in Data S1.

Statistical Analysis

Continuous variables are presented as mean±SD when normally distributed and median with 25th and 75th percentiles when nonnormally distributed. Categorical variables are expressed as numbers and percentages. The Mann‐Whitney U test was applied for comparison of continuous variables between cardiac and noncardiac syncope, and categorical variables were compared by Pearson χ2 test and Fisher exact test, as appropriate. We calculated the sample size on the basis of our previous experiences with studies on the diagnosis of acute myocardial infarction in patients presenting with chest pain to the ED12, 13, 14, 15, 30, 31 and aimed to enroll a total of 120 patients with cardiac syncope. Using the nomogram of Carley et al and Jones et al,32, 33 the targeted sample size ranged from 660 to 940 patients, estimating a sensitivity of 90% with a 2‐sided 95% confidence interval (CI) of 5% to detect cardiac syncope. To evaluate diagnostic accuracy for the diagnosis of cardiac syncope, receiver‐operating characteristic curves were constructed to assess the sensitivity and specificity of the ED physician's clinical judgment and the EGSYS risk score for cardiac origin probability, alone and in combination with MRproANP, CTproET1, copeptin, and MRproADM. The comparison of areas under the receiver‐operating characteristic curves (AUCs) was performed according to DeLong et al.34 In addition, reclassification tables for net reclassification improvement were used to assess the incremental yield of the additional use of MRproANP at presentation to predict cardiac syncope.35 Logistic regression was used to combine clinical judgement with prohormone plasma concentrations in predicting the final adjudicated diagnosis, generating a graphic displayed as a cardiac syncope diagnosis nomogram. Glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation.36

All hypothesis testing was 2 tailed, and P<0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (SPSS Inc, Chicago, IL) and R 3.3.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Characteristics of Study Subjects

Eight‐hundred eighty‐six consecutive patients were enrolled in the BASEL IX study (Figure S1) from May18th, 2010 until July 25th, 2014. A complete screening log was not kept in all participating sites. In those sites with a screening log, 14% of presenting patients were excluded because they presented to the ED >12 hours after the syncopal event. For this analysis, patients with missing prohormone fragment measurements (n=3), those with a final diagnosis of a nonsyncopal loss of consciousness (n=117), and individuals in whom the final diagnosis remained unclear even after central adjudication (n=77) were excluded, leaving 689 patients for the analysis. A complete follow‐up at 12 months was available in 99% of patients and at 24 months in 89% of patients.

Patients with cardiac syncope were significantly older, more often had a history of coronary artery disease, arrhythmia, or valvular heart disease, and more often experienced syncope during exertion. Further details on patient demographics are provided within Table 1.

Table 1

Baseline Characteristics of the Patients

Characteristics	All Patients (n=689)	Patients With Cardiac Syncope (n=125)	Patients Without Cardiac Syncope (n=564)	P Value
Age, y	70 (57–80)	77 (68–84)	68 (54–78)	<0.001
Time to presentation, min	75 (50–127)	81 (52–150)	75 (50–122)	0.379
Time to blood draw since presentation, min	107 (70–159)	109 (75–162)	107 (70–159)	0.682
Time to blood draw since syncopal event, min	190 (140–290)	187 (156–298)	190 (137–289)	0.464
Male sex	404 (58.6)	76 (60.8)	328 (58.2)	0.658
Inpatient treatment	368 (53.4)	106 (84.8)	262 (46.5)	<0.001
Time hospitalized, d	4 (1–8)	6 (2–9)	3 (1–7)	<0.001
Risk factors
Hypertension	393 (57.1)	79 (63.7)	314 (55.7)	0.124
Hypercholesterolemia	262 (39.3)	57 (47.9)	205 (37.5)	0.045
Diabetes mellitus	98 (14.2)	25 (20.2)	73 (12.9)	0.052
Current smoking	130 (19.0)	15 (12.3)	115 (20.4)	0.051
History of smoking	218 (31.8)	43 (35.2)	175 (31.1)	0.431
History
Coronary artery disease	149 (21.9)	45 (37.2)	104 (18.6)	<0.001
Previous MI	92 (13.4)	28 (22.4)	64 (11.3)	0.002
Arrhythmiaa	138 (20.5)	43 (35.5)	95 (17.2)	<0.001
Valvular heart disease	63 (9.3)	22 (18.0)	41 (7.4)	<0.001
Previous stroke	48 (7.0)	9 (7.3)	39 (7.0)	1
Epilepsy	19 (2.8)	2 (1.6)	17 (3.0)	0.551
Previous syncope	413 (61.4)	74 (59.7)	339 (61.7)	0.745
Syncope situation
Orthostatic	83 (12.3)	8 (6.5)	75 (13.6)	0.043
While standing	271 (40.2)	50 (40.7)	221 (40.1)	0.993
Exertion	76 (11.3)	32 (26.2)	44 (8.0)	<0.001
While sitting	256 (37.9)	41 (33.6)	215 (38.9)	0.325
While lying down	22 (3.3)	3 (2.5)	19 (3.4)	0.781
Presence of injury	97 (14.5)	18 (15.0)	79 (14.4)	0.983
Vital parameters
Heart rate	72 (64–84)	73 (60–92)	72 (64–82)	0.509
Systolic BP	129 (114–145)	131 (117–152)	128 (114–144)	0.062
Diastolic BP	73 (62–82)	73 (60–84)	73 (63–81)	0.972
Pathological ECGb	240 (34.8)	72 (57.6)	168 (29.8)	<0.001
Long‐term medication
Aspirin	242 (32)	52 (41)	190 (29.5)	0.013
Vitamin K antagonists	92 (12)	26 (21)	66 (10.3)	0.002
β‐Blockers	256 (33)	54 (43)	202 (31.4)	0.017
Antiarrhythmics	26 (3.4)	7 (5.6)	19 (3.0)	0.227
ACEIs/ARBs	333 (43)	66 (52)	267 (41.5)	0.031
Calcium antagonists	130 (17)	23 (18)	107 (16.6)	0.755
Diuretics	222 (29)	56 (44)	166 (25.8)	<0.001
Nitroglycerine	44 (5.7)	14 (11)	30 (4.7)	0.008

Data are presented as median (25th–75th percentile) or number (percentage). ACEI indicates angiotensin‐converting enzyme inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; and MI, myocardial infarction.

Arrhythmia indicates history of supraventricular or ventricular tachycardia.

Pathological ECG was defined as meeting at least one of the following criteria: bifascicular block (left bundle branch block or right bundle branch block [RBBB] combined with left anterior fascicular block), second‐ or third‐degree AV block, asymptomatic inappropriate sinus bradycardia (<50 beats/min) in the absence of negatively chronotropic medications, nonsustained ventricular tachycardia, preexcited QRS complexes, long or short QT intervals (men, >450 ms; women, >470 ms), RBBB pattern with ST elevation in leads V1 to V3 (Brugada syndrome), early repolarization, Q waves suggesting myocardial infarction, negative T waves in right precordial leads, and ε waves and ventricular late potentials suggestive of arrhythmogenic right ventricular dysplasia.2

Adjudicated Final Diagnosis of Syncope

The overall patient population had the following adjudicated syncope causes: 125 (18%) cardiac, 320 (46%) reflex (neurally mediated), 181 (26%) orthostatic hypotension, and 63 (9%) other noncardiac causes. Among patients with cardiac syncope, most (90 [13%]) experienced arrhythmias, of which 58 (8.4%) were bradyarrhythmias and 27 (3.9%) were tachyarrhythmias. Twenty‐eight patients (4.1%) had structural heart disease, including acute myocardial infarction in 13 (1.9%); in 7 patients (1%), other cardiac origin triggered syncope (eg, pulmonary embolism). All final adjudicated diagnoses are listed in Table S3.

Prohormone Plasma Concentrations According to Adjudicated Diagnosis

Plasma concentrations of MRproANP, CTproET1, copeptin, and MRproADM were all significantly higher among patients with cardiac syncope compared with patients with other adjudicated causes of syncope (Table 2, Figure 1).

Table 2

Plasma Concentration of Prohormones According to Adjudicated Diagnosis

Prohormones	Cardiac (n=125)	Reflex (n=320)	Orthostatic (n=181)	Other, Noncardiac (n=63)	P Valuea
MRproANP, pmol/L	246 (141–355)	91 (59–146)	122 (72–197)	111 (62–189)	<0.001
CTproET1, pmol/L	89 (65–118)	61 (53–75)	80 (59–104)	65 (52–83)	<0.001
Copeptin, pmol/L	45 (20–78)	32 (14–64)	33 (14–72)	18 (10–52)	0.01
MRproADM, nmol/L	1.0 (0.7–1.4)	0.6 (0.5–0.9)	0.9 (0.7–1.2)	0.7 (0.6–1.0)	<0.001

Data are given as median (25th–75th percentile). MRproANP: In healthy individuals, the range of MRproANP concentrations was from 9.6 to 313 pmol/L. The median was 45 pmol/L. The 99th percentile was 197.5 pmol/L.26 CTproET1: In healthy individuals, the range of CTproET1 concentrations was from 10.5 to 77.4 pmol/L, and the mean (SD) was 44.3 (10.6) pmol/L. The 99th percentile was 72.8 pmol/L.27 Copeptin: In healthy individuals, the range of copeptin concentrations was from 1 to 13.8 pmol/L. The median was 4.2 pmol/L. The 99th percentile was 13.5 pmol/L.29 MRproADM: In healthy individuals, the range of MRproADM concentrations was from 0.10 to 0.64 nmol/L, and the mean (SD) was 0.33 (0.07) nmol/L. The 99th percentile was 0.52 nmol/L.28 CTproET1 indicates C‐terminal proendothelin 1; MRproADM, midregional‐proadrenomedullin; and MRproANP, midregional–pro‐A‐type natriuretic peptide.

The P values were tested for the comparison across the 4 categories in the table. The Kruskal‐Wallis test was used to conduct this test.

Figure 1

Box plots for all assessed prohormones in patients with a different origin of syncope. The ends of the whisker were defined so that the maximum length of each whisker is 1.5 times the interquartile range. Midregional–pro‐A‐type natriuretic peptide (MRproANP): In healthy individuals, the range of MRproANP concentrations was from 9.6 to 313 pmol/L. The median was 45 pmol/L. The 99th percentile was 197.5 pmol/L.26 C‐terminal proendothelin 1 (CTproET1): In healthy individuals, the range of CTproET1 was from 10.5 to 77.4 pmol/L, and the mean (SD) was 44.3 (10.6) pmol/L. The 99th percentile was 72.8 pmol/L.27 Copeptin: In healthy individuals, the range of copeptin concentrations was from 1 to 13.8 pmol/L. The median was 4.2 pmol/L. The 99th percentile was 13.5 pmol/L.29 Midregional‐proadrenomedullin (MRproADM): In healthy individuals, the range of MRproADM was from 0.10 to 0.64 nmol/L, and the mean (SD) was 0.33 (0.07) nmol/L. The 99th percentile was 0.52 nmol/L.28

Diagnostic Performance

The diagnostic accuracy of MRproANP levels for determining a cardiac syncope cause, as quantified by the AUC curve, was 0.80 (95% CI, 0.76–0.84; Figure 2). Diagnostic performance measures of receiver‐operating characteristic curve–derived cutoff points achieving predefined target sensitivities for rule out and rule in are summarized in Tables 3 and 4. At a threshold of <77 pmol/L, MRproANP ruled out 30% of individuals (n=211), with a sensitivity of 95% and a negative predictive value of 97%. An algorithm to rule out cardiac syncope, combining an MRproANP level of <77 pmol/L and an ED probability of <20%, had a sensitivity of 99% and a negative predictive value of 99% and allowed us to triage 18% of patients toward rule out. At a threshold of >181 pmol/L, MRproANP ruled in 28% of individuals (n=192), with a specificity of 80% and a positive predictive value of 41%. An algorithm to rule in cardiac syncope, combining an MRproANP level of >181 pmol/L and an ED probability of >80%, had a specificity of 98% and a positive predictive value of 81% and allowed us to triage 8% of patients toward rule in.

Figure 2

Receiver‐operating characteristic (ROC) curves for detection of cardiac syncope for midregional–pro‐A‐type natriuretic peptide (MRproANP), alone and in combination with either clinical judgement (left) or Evaluation of Guidelines in Syncope Study (EGSYS) risk score (right). ROC curves for the diagnostic performance of MRproANP, alone and in combination with either clinical judgment (left) or EGSYS risk score (right), for diagnosing cardiac syncope. The red curve displays the biomarker alone; the green curve, emergency department (ED) probability (visual analogue scale [VAS]; left) or EGSYS risk score (right) for detecting cardiac syncope; and the blue curve, the combination of the biomarker and ED probability (left) or EGSYS risk score (right). The black arrows show the comparison of areas under the ROC curves (AUCs) for the combination of ED probability with MRproANP against ED probability alone (left) and for the combination of EGSYS risk score with MRproANP against the EGSYS risk score alone (right). CI indicates confidence interval.

Table 3

Diagnostic Test Characteristics of Prespecified Cut Point Values for Rule Out of Cardiac Syncope

Target Sensitivity, %	No. (%) of Patients	Cut Point for MRproANP, pmol/L	Sensitivity, %	Specificity, %	NPV, %	PPV, %	Accuracy, %
70	450 (65)	<159	70 (62–78)	73 (69–77)	92 (90–94)	37 (33–41)	72 (69–76)
80	380 (55)	<130	80 (73–87)	63 (59–67)	94 (91–96)	33 (30–36)	63 (63–70)
90	289 (42)	<98	90 (85–94)	49 (45–53)	96 (93–98)	28 (26–30)	56 (53–60)
95	211 (31)	<77	95 (91–98)	36 (32–40)	97 (95–99)	25 (24–26)	47 (44–50)
Combination	124 (18)	<77 (VAS score, <20%)	99 (96–100)	22 (19–26)	99 (96–100)	22 (19–25)	36 (32–39)

Numbers represent percentage (95% confidence interval) unless otherwise indicated. MRproANP indicates midregional–pro‐A‐type natriuretic peptide; NPV, negative predictive value; PPV, positive predictive value; and VAS, visual analogue scale.

Table 4

Diagnostic Test Characteristics of Prespecified Cut Point Values for Rule In of Cardiac Syncope

Target Specificity, %	No. (%) of Patients	Cut Point for MRproANP, pmol/L	Sensitivity, %	Specificity, %	NPV, %	PPV, %	Accuracy, %
70	262 (38)	>147	74 (66–82)	70 (66–74)	92 (90–94)	35 (31–39)	71 (67–74)
80	192 (28)	>181	63 (54–71)	80 (77–83)	91 (89–93)	41 (36–47)	77 (74–80)
90	121 (18)	>243	51 (42–60)	90 (88–93)	89 (88–91)	53 (46–62)	83 (81–86)
95	55 (8)	>373	22 (14–29)	95 (93–97)	85 (83–86)	49 (37–62)	82 (80–84)
Combination	53 (8)	>181 (VAS score, >80%)	34 (27–43)	98 (97–99)	81 (69–90)	87 (84–90)	87 (84–89)

Numbers represent percentage (95% confidence interval) unless otherwise indicated. MRproANP indicates midregional–pro‐A‐type natriuretic peptide; NPV, negative predictive value; PPV, positive predictive value; and VAS, visual analogue scale.

The diagnostic accuracy for the other prohormone markers for detecting cardiac syncope, presented as AUC, was 0.69 (95% CI, 0.64–0.74) for CTproET1, 0.58 (95% CI, 0.52–0.63) for copeptin, and 0.68 (95% CI, 0.63–0.73) for MRproADM. The combination of CTproET1 and MRproANP did not improve diagnostic accuracy compared with that of MRproANP alone (P=0.90).

Combination of Overall Clinical Judgement With MRproANP

The AUC for the combination of ED probability with MRproANP was significantly higher (0.90; 95% CI, 0.87–0.93) than for the ED probability alone (AUC, 0.86; 95% CI, 0.82–0.90; P=0.003). The net reclassification improvement of MRproANP was calculated at 0.216 (P<0.001). Integrated discriminatory improvement was 0.035 (P=0.001; Table S4). The combination of the other prohormones with ED probability did not provide a significant improvement in diagnostic accuracy when compared with ED probability alone (CTproET1, P=0.05; copeptin, P=0.09; and MRproADM, P=0.08, for the comparisons of ED probability plus biomarker versus ED probability alone).

Cardiac Syncope Using the Fagan Nomogram

Figure 3 is a Fagan nomogram for cardiac syncope to visualize the additive value of MRproANP to clinical judgement. As indicated by the nomogram, MRproANP has the greatest value as a diagnostic test in patients with intermediate pretest probability. In this category (visual analogue scale score ≥10% and ≤60%), 34 of 424 patients (8%) experienced cardiac syncope. MRproANP ≥175 pmol/L correctly classified 77% of all patients as having cardiac syncope or not having cardiac syncope. In 358 patients with a low ED probability of cardiac syncope (≤20%), 12 of 358 (3.4%) had a final adjudicated diagnosis of cardiac syncope. Of these 12 individuals, 11 (99.3%) could have had the misdiagnosis corrected if the additional information of MRproANP ≥77 pmol/L had been available. Specific patient characteristics of these 11 patients are provided in Table S5. A diagnosis other than cardiac syncope was initially suspected in the ED in the presence of prodromal symptoms (8 of 11 patients) and a normal ECG at presentation (7 of 11 patients) or in the absence of cardiac risk factors, such as diabetes mellitus (2 of 11 patients) or hypertension (3 of 11 patients).

Figure 3

Cardiac syncope diagnosis nomogram. Clinical judgment by the emergency department (ED) physician about the presence of cardiac syncope is displayed as “Pre.” The treating ED physician estimated cardiac origin probability on the basis of all information available in the individual patient 90 minutes after presentation, including clinical assessment, the 12‐lead ECG, and the routine laboratory test results. The middle line represents midregional–pro‐A‐type natriuretic peptide (MRproANP) level (in pmol/L) at presentation. When a straight line is drawn through the pretest probability and MRproANP level, the posttest probability is found on the right line (“Post”). For example, an ED probability of 50% with an MRproANP level of 1000 pmol/L yields an ≈78% probability of cardiac syncope on the basis of these 2 predictors (blue‐dotted line).

Combination of MRproANP With EGSYS Risk Score

The diagnostic accuracy of MRproANP levels in combination with the EGSYS risk score in detecting cardiac syncope is shown in Figure 2. The AUC for the combination of MRproANP and the EGSYS risk score was significantly higher (0.81; 95% CI, 0.77–0.85) than for the EGSYS risk score alone (AUC, 0.67; 95% CI, 0.63–0.72; P<0.001).

Prediction of Cardiac Syncope

Logistic regression analysis confirmed MRproANP as a predictor of cardiac syncope in both univariate and multivariate analyses (Table 5). In multivariable analysis, only a pathological ECG (according to the European Society of Cardiology Guidelines), impaired renal function, and MRproANP (odds ratio, 4.51; 95% CI, 2.84–7.16) were independent predictors of cardiac syncope.

Table 5

Logistic Regression

Variable	Univariable Logistic Regression				Multivariable Logistic Regression
	Odds Ratio	95% CI (Lower‐Upper)		P Value	Odds Ratio	95% CI (Lower‐Upper)		P Value
Agea	1663	1412	1958	<0.001	1293	1012	1654	0.040
Male sex	1124	757	1669	0.535
Valvular disease	2747	1569	4809	<0.001	1726	901	3307	0.01
Charlson comorbidity index	1252	1164	1347	<0.001
Palpitations	1025	502	2093	0.945
ECG pathological	3236	2176	4812	<0.001	2020	1268	3218	0.003
GFR, CKD‐EPI	981	973	989	<0.001	1023	1008	1038	0.002
MRproANP (ln transformed)	4569	3329	6271	<0.001	4506	2837	7156	<0.001
CTproET1 (ln transformed)	5522	3380	9022	<0.001	1700	765	3744	0.193

CI indicates confidence interval; CKD‐EPI indicates Chronic Kidney Disease Epidemiology Collaboration; CTproET1, C‐terminal proendothelin 1; GFR, glomerular filtration rate; ln, natural log; and MRproANP, midregional–pro‐A‐type natriuretic peptide.

Per 10‐year increase.

Direct Comparison of MRproANP With BNP

In the subgroup of patients with both biochemical signals available (n=393 [57%]), the AUC was 0.77 (95% CI, 0.71–0.83) for MRproANP and 0.74 (95% CI, 0.68–0.81) for BNP (P=0.278) for detecting cardiac syncope.

Discussion

This prospective multicenter study using long‐term follow‐up and central adjudication aimed to contribute to advancing the rapid and accurate diagnosis of patients presenting with syncope to the ED. The study evaluated the diagnostic utility of prohormones quantifying 4 different neurohumoral pathways possibly involved in the pathophysiological characteristics of cardiac syncope. Prohormones were selected on the basis of the hypothesis that they could represent the “memory of the cardiac and in fact often arrhythmic event” at ED presentation and because commonly they are more stable analytically compared with active hormones.16, 17, 18

We report 7 major findings: First, in patients with an adjudicated diagnosis of cardiac syncope, prohormone plasma levels (MRproANP, CTproET1, copeptin, and MRproADM) were significantly higher than in patients with noncardiac causes of syncope. Second, although MRproANP levels were nearly exclusively elevated in patients with cardiac syncope, CTproET1 and MRproADM levels were also elevated in patients with syncope attributable to orthostatic hypotension. Third, diagnostic accuracy for cardiac syncope of prohormones, as quantified by the AUC curve, was high for MRproANP, moderate for CTproET1 and MRproADM, and poor for copeptin. Fourth, MRproANP provided significant incremental value when combined with clinical judgment in the ED, which was mainly based on the clinical assessment and the 12‐lead ECG. Diagnostic accuracy for the combination of MRproANP and ED probability was high (AUC, 0.90; 95% CI, 0.87–0.93) and significantly higher than for ED probability alone. Although the 0.04 higher C‐statistic represents a modest increase, a confirmed diagnosis in a population fraught with diagnostic uncertainty suggests important potential clinical utility. MRproANP concentrations might add information of a more objective and more specific marker of heart disease than a subjective clinical history or examination alone. An algorithm incorporating MRproANP levels seems to perform better than existing protocols based on international syncope guidelines1, 2 (eg, the substantial incremental value of MRproANP was confirmed using the EGSYS risk score as an alternative method of reflecting the standard of care). The magnitude of the increase in diagnostic accuracy provided by MRproANP was identical compared with the increase in diagnostic accuracy in the diagnosis of myocardial infarction provided by high‐sensitivity cardiac troponin compared with sensitive cardiac troponin.37 None of the other biochemical signals provided incremental value on top of clinical judgment. Fifth, increased MRproANP levels were independently associated with cardiac causes of syncope (odds ratio, 4.51; 95% CI, 2.84–7.16) and were a stronger predictor than clinical features. Sixth, an algorithm based on the combination of MRproANP and clinical judgement allowed use to rule out the presence of cardiac syncope, with a sensitivity of 99% and a negative predictive value of 99%. Seventh, subgroup analysis in patients in whom BNP was measured as part of the clinical standardized operating procedures documented comparable diagnostic accuracy of MRproANP and BNP. This suggests that our findings about the incremental diagnostic value of MRproANP likely can be extrapolated to BNP. Further studies are warranted to appropriately test this hypothesis.

These findings extend and corroborate previous pilot data on the possible clinical utility of biomarkers quantifying neurohormonal pathways.22, 38, 39, 40, 41, 42 Several studies have demonstrated that ANP and BNP plasma concentrations are increased in heart failure and cardiac arrhythmias.43, 44, 45 For example, in a pilot study of 18 patients, a significant increase of BNP and NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide) plasma concentrations was observed after the induction of ventricular fibrillation in patients with an implantable cardioverter defibrillator.45 Accordingly, BNP and NT‐proBNP have been proposed to help identify patients with syncope of cardiac origin and to be good markers for risk stratification in this population.8, 39, 46, 47, 48 Finally, retrospective single‐center studies reported higher BNP and NT‐pro‐BNP levels in patients admitted for syncope to a cardiology department in whom a cardiac cause was found during in‐hospital workup compared with other causes.31, 45

Detection of cardiac syncope has immediate consequences for patient management and, in general, triggers hospital admission, ECG‐rhythm monitoring, and cardiology consultation.1, 2 ECG‐rhythm monitoring is mandatory because the next episode of bradyarrhythmia or tachyarrhythmia may be fatal if undetected and untreated.1, 2 Cardiology consultation is necessary for the rapid initiation of the appropriate further diagnostic and therapeutic measures. These include implantation of a pacemaker in patients with documented symptomatic bradycardia (eg, high‐degree AV block) and transthoracic echocardiography in patients with suspected severe aortic stenosis, followed by coronary angiography and computed tomography–angiography once severe aortic stenosis is confirmed in the preparation of aortic valve replacement. They also include implantation of a cardioverter‐defibrillator in patients with coronary artery disease and severely impaired left ventricular ejection fraction or further risk stratification by an electrophysiological study for the assessment of ventricular arrhythmias. Moreover, if the risk of recurrence of a cardiac syncope is estimated to be high (eg, in patients with intermittent high‐degree AV block), bed rest would seem prudent to avoid the possible harm of injuries related to falls in case of recurrence.

Our findings suggest that plasma levels of MRproANP can be considered a marker of cardiac syncope. Therefore, MRproANP can possibly offer diagnostic information that would improve management of syncopal patients in the ED. As demonstrated by the nomogram, and by proposing an algorithm implementing the combination of a low ED probability and low plasma levels of MRproANP (with sensitivity and negative predictive values of 99% for cardiac syncope), the benefits of MRproANP application may be predominately in ruling out a cardiac cause of syncope and defining low‐risk patients suitable for outpatient management. Avoiding unnecessary hospital admissions and extensive cardiac workup by a more accurate diagnosis likely will result in cost savings. This approach mirrors the use of D‐dimers in the rule out of venous thromboembolism.49 Further studies specifically addressing cost‐effectiveness of MRproANP are warranted to evaluate whether the cost savings of the biomarker‐guided approach are similar to cost savings observed with the use of natriuretic peptides in patients presenting with acute dyspnea.50, 51

Further studies are also necessary to decipher the exact pathophysiological link between the vasoconstrictor prohormone CTproET1 and the vasodilator prohormone MRproADM and orthostatic hypotension.

Most patients presenting with syncope to the ED had mildly elevated plasma concentrations of copeptin, a stable peptide derived from the precursor of vasopressin, irrespective of the cause of syncope. This suggests that the arginine‐vasopressin system may be activated to a similar extent by several mechanisms leading to syncope. Accordingly, copeptin does not seem to have a role as a diagnostic biomarker in this setting.41, 52

This study has 3 important methodological strengths that differentiate it from previous studies on syncope: global representation of patients attributable to enrollment in 8 countries on 3 continents, long‐term follow‐up, and central adjudication by 2 independent cardiologists to maximize the accuracy of the reference standard diagnosis of cardiac versus noncardiac syncope.

Although this study represents an important step towards the integration of biomarkers into the clinical management of patients presenting with syncope to the ED, external validation in a large diagnostic study of comparable methodological scrutiny is required.

This applies to the diagnostic accuracy of MRproANP in general and the suggested cutoff levels for early rule out in particular. Because patients with syncope show a wide age range (44–94 years in this study), an even larger data set may allow derivation of age‐optimized cutoff levels to further improve the sensitivity and/or the effectiveness of the biomarker‐based rule‐out approach.53 Also, additional studies are warranted to evaluate other biochemical and/or electrocardiographic signatures, in combination with clinical judgment, to improve the early diagnosis of cardiac syncope. These studies are needed because diagnostic uncertainties in patients presenting to the ED with syncope have become the focus of debate about inappropriate use of resources, increasing healthcare costs, and, most important, patient safety.54, 55

Potential limitations of the present study merit consideration. First, this diagnostic study using central adjudication required informed consent. This may have introduced a small, but unavoidable, selection bias. Because this study enrolled all patients presenting with syncope to the ED, irrespective of the pretest probability for a cardiac cause, any selection bias should have been minimized. Second, we recruited patients presenting to the ED. Therefore, it is unknown whether our findings can be extrapolated to patients presenting to primary care. Third, we cannot comment on patients who present >12 hours after symptom onset because these patients were excluded from our study. However, only a few patients presented >12 hours after the syncopal event. Fourth, prohormone levels were measured once. Further studies are warranted to evaluate whether serial sampling would allow a further increase in diagnostic accuracy. Fifth, despite using the most stringent and unprecedented method of central adjudication of the final diagnosis by 2 independent cardiologists who had access to the whole clinical workup and the study‐specific data set, third‐party anamnesis, and long‐term follow‐up of cardiovascular events and/or recurrent syncope, a few patients may still have been misclassified. This invariably may lead to a slight underestimation of the true accuracy of the prohormones tested.

In conclusion, this large multicenter diagnostic study suggests that the plasma level of MRproANP may be a quantitative marker of cardiac syncope. Using it in conjunction with the ED probability, summarizing all information commonly available at 90 minutes in the ED, including the 12‐lead ECG, improves the early rule out and/or rule in of cardiac syncope. In contrast, the other prohormones tested did not seem to have diagnostic utility.

Sources of Funding

This study was supported by research grants from the Swiss National Science Foundation, the Swiss Heart Foundation, the Cardiovascular Research Foundation Basel, the University of Basel, BRAHMS, Singulex, and the University Hospital Basel.

Disclosures

Twerenbold received research support from the Swiss National Science Foundation (P300PB‐167803/1) and speaker honoraria/consulting honoraria from Roche, Abbott, and Brahms. Reichlin has received research grants from the Goldschmidt‐Jacobson‐Foundation, the Swiss National Science Foundation (PASMP3‐136995), the Swiss Heart Foundation, the Professor Max Cloëtta Foundation, the Uniscientia Foundation Vaduz, the University of Basel, and the Department of Internal Medicine, University Hospital Basel, and speaker honoraria from Brahms and Roche. Rubini has received a research grant from the Swiss Heart Foundation and speaker's honoraria from Abbott. Martin‐Sanchez received advisory/consulting fees from Novartis, MSD, Pfizer, The Medicine Company, and Otsuka and research grants from the Spanish Ministry of Health, FEDER, Novartis, Abbott, and Orion‐Pharma. Mueller has received research support from the Swiss National Science Foundation, the Swiss Heart Foundation, Abbott, Alere, Brahms, Nanosphere, Roche, Siemens, 8sense, Nanosphere, and the Department of Internal Medicine, University Hospital Basel, and speaker honoraria from Abbott, Alere, Brahms, Novartis, Roche, and Siemens. All other authors declare that they have no conflict of interest with this study.

Data S1. Supplemental Methods.

Table S1. Enrollment Across Regions

Table S2. Performed Diagnostic Tests at Admission or During Long‐Term Follow‐Up in All 689 Patients

Table S3. All Final Adjudicated Diagnoses, n (%)

Table S4. Reclassification by Means of MRproANP

Table S5. Overview of Patients With Low Initial ED‐Probability (VAS‐Score ≤20%) and a Final Diagnosis of Cardiac Syncope

Figure S1. Patient flow diagram for all patients.

Click here for additional data file.