In the present study, we investigated the prognostic value of MR-proANP (mid-regional pro-atrial natriuretic peptide). We consecutively evaluated a catheterization laboratory cohort of 2700 patients with symptomatic CAD (coronary artery disease) [74.1% male; ACS (acute coronary syndrome), n=1316; SAP (stable angina pectoris), n=1384] presenting to the Cardiology Department of a large primary care hospital, all of whom underwent coronary angiography. Serum MR-proANP and other laboratory markers were sampled at the time of presentation or in the catheterization laboratory. Clinical outcome was assessed by hospital chart analysis and telephone interviews. The primary end point was all-cause death at 3 months after enrolment. Follow-up data were complete in 2621 patients (97.1%). Using ROC (receiver operating characteristic) curves, the AUC (area under the curve) of 0.73 [95% CI (confidence interval), 0.67–0.79] for MR-proANP was significantly higher compared with 0.58 (95% CI, 0.55–0.62) for Tn-I (troponin-I; DeLong test, P=0.0024). According to ROC analysis, the optimal cut-off value of MR-proANP was at 236 pmol/l for all-cause death, which helped to find a significantly increased rate of all-cause death (n=76) at 3 months in patients with elevated baseline concentrations (≥236 pmol/l) compared with patients with a lower concentration level in Kaplan–Meier survival analysis (log rank, P<0.001). The predictive performance of MR-proANP was independent of other clinical variables or cardiovascular risk factors, and superior to that of Tn-I or other cardiac biomarkers (all: P<0.0001). MR-proANP may help in the prediction of all-cause death in patients with symptomatic CAD. Further studies should verify its prognostic value and confirm the appropriate cut-off value.
- coronary artery disease
- mid-regional pro-atrial natriuretic peptide (MR-proANP)
- natriuretic peptide
- prognostic biomarker
ACS (acute coronary syndromes) carry a strong risk of death and short-term development of major adverse cardiovascular events [1,2]. A vast array of biomarkers have been suggested to help in the risk stratification of patients with ACS [3–5]. Natriuretic peptides are widely used in the diagnosis and risk stratification of patients with heart failure. This is particularly true for BNP (B-type natriuretic peptide) and NT-proBNP (N-terminal pro-BNP), which are recommended by current heart failure guidelines . In addition, both markers have been suggested for the guidance of treatment in patients with heart failure [7,8]. Although widely available and used in clinical practice, little is known about the prognostic value of natriuretic peptides in patients with CAD (coronary artery disease) without heart failure or in those presenting with ACS. Such information may be important in order to identify patients at high risk of major adverse cardiovascular events. The troponins, usually assessed with the primary intention to help in establishing the correct diagnosis of ACS, have also been suggested to be useful in predicting mortality in these patients with CVD (cardiovascular disease); however, their value is hampered by the need for repeated measurements .
Another problem in the analysis of prognostic markers is their sheer number. Indeed, a multitude of prognostic markers has been suggested over the last several years for risk stratification of patients with ACS or stable CAD, including markers such as MPO (myeloperoxidase) or H-FABP (heart fatty-acid-binding protein) [10,11]. Most of these markers carry the important drawback of either limited clinical availability or of non-standardized assays. BNP and NT-proBNP, both of which are widely clinically available, have been shown to have good short- and long-term prognostic value in patients presenting with ACS . For several years, BNP was ascribed a greater prognostic potential in the assessment of prognosis in patients after AMI (acute myocardial infarction) than the atrium-derived ANP (atrial natriuretic peptide). This early imbalance of the value of prognostication pointed towards an ANP assay instability  that has recently been overcome by the development and establishment of a novel sandwich immunoassay measuring MR-proANP (mid-regional pro-ANP) . Indeed, studies in patients with acute  or CHF (chronic heart failure)  have shown that the diagnostic and prognostic utility of MR-proANP is non-inferior to that of BNP or NT-proBNP. On the basis of on the fact that, compared with MR-proANP, circulating levels of BNP are rather low and only enhanced as much as 200–300-fold , we considered MR-proANP as worthwhile for further investigation of clinical outcome in patients with ischaemic heart disease.
The aim of the present study was to assess the prognostic value of MR-proANP in a large cardiac catheterization cohort of consecutive patients with confirmed CAD presenting with either acute chest pain or for elective coronary angiography to a large primary care hospital.
MATERIALS AND METHODS
Study design and population
We evaluated 4038 consecutive patients with chest pain presenting to the Cardiology Department of the University Hospital Tübingen, Tübingen, Germany between December 2007 and April 2009. Patients with non-coronary origin of chest pain (n=1338) were excluded from all further analyses. Thus the final dataset consisted of a catheterization laboratory cohort of 2700 patients with confirmed CAD presenting either with ACS or SAP (stable angina pectoris). On presentation, all patients underwent a thorough physical examination, a resting ECG, routine laboratory assessments including full blood count and basic biochemistry and an echocardiogram for the determination of LVEF (left ventricular ejection fraction). Patients with SAP were admitted to the cardiology department to ensure rapid transfer for elective coronary angiography. Thus all patients received coronary angiography within 24 h of arrival. In patients presenting with ACS, the median time between onset of chest pain and hospital presentation was 147 min. Patients with SAP were admitted for elective coronary angiography. The number of coronary vessels affected with significant stenoses (defined as stenosis ≥75%) and all clinical variables are presented in Table 1. ACS was diagnosed according to the joint guidelines issued by the American Heart Association and the American College of Cardiology . In brief, ACS was defined as episodes of angina pectoris of more than 20 min duration with progress over time or with recurrent episodes at rest or with minimal physical strain within the last 24 h. The ECG shows at least one of the following criteria: new ST-depression of at least 0.1 mV or negative T of at least 0.3 mV in two or more leads of standard resting ECG. Serum values of CK (creatine kinase) and Tn-I (troponin-I) levels ranged from normal to several fold increased. STEMI (ST-elevation myocardial infarction) describes episodes of more than 20 min of angina pectoris with the following ECG and laboratory criteria: ST-segment elevation in two or more leads with at least 0.1 mV and a 2-fold elevation of CK compared with the normal level and with a significant CK-MB (CK myocardial band) or an increased Tn-I level, whereas an NSTEMI (non-STEMI) shows episodes of more than 20 min of angina pectoris without ST-segment elevation in the standard ECG. ACS comprised patients' chest pain due to AMI (STEMI or NSTEMI) or unstable angina pectoris (CK- and troponin-negative). In contrast, SAP defines symptoms without progression and stable intensity.
Exclusion criteria included age <18 years, inability to give or lack of informed consent, pregnancy, infectious disease or sepsis, non-coronary origin of chest pain such as arterial hypertension, orthopaedic aetiology, acute pulmonary embolism, Da Costa's syndrome and myocarditis.
The study was approved by the local ethics committee of the University Hospital Tübingen and performed in accordance with the Declaration of Helsinki. All patients provided written informed consent.
Blood samples for biomarker assessment were drawn from an anticubital vein and collected at the time of presentation to the Cardiology Department or in the catheterization laboratory, whichever came first.
Biomarkers such as creatinine, CK, CK-MB and CRP (C-reactive protein) and Tn-I-ultra assay were measured on an Advia-Centaur System (all assays and system from Siemens Healthcare Diagnostics). Since the use of CK/CK-MB assays have gradually become outdated, the AMI diagnosis of the final dataset was based on Tn-I assessment. Serial assessments of Tn-I were performed as judged by the attending physician and according to current guidelines, which suggest an enhanced Tn-I value exceeding the 99th percentile of the normal reference population . The diagnostic threshold for Tn-I-ultra assay is ≥0.04 μg/l according to the manufacturer and was based on previous studies . For technical reasons, Tn-I data were not available in 47 patients with SAP. An additional blood sample was taken into 5 ml of pyrogen-free serum vials and immediately centrifuged at 10000 g for 30 min at 4°C. Immediately thereafter, the supernatant serum was aliquotted and stored at −80°C until analysis. MR-proANP serum values were determined using a novel sandwich immunoluminometric assay (B.R.A.H.M.S GmbH) with a within-run imprecision coefficient of variation of 1.2% and total imprecision of 5.4% as described previously , using an automated immunofluoroscence assay (Kryptor®; B.R.A.H.M.S GmbH): the assay's lower limit of quantification was 4.5 pmol/l. According to the previous description, the median value in healthy blood donors was at 45 (range, 9.6–313) pmol/l . Attending physicians in the Cardiology Department were aware of the patients' Tn-I serum values, but unaware of their respective MR-proANP levels.
Primary and secondary end points of clinical follow-up
The primary end point was all-cause death at 3 months after enrolment. Follow-up for 3 months also documented composite events of recurrent non-fatal AMI, stroke, and all-cause death as a secondary combined end point. Clinical outcome was evaluated by hospital chart analysis and a predefined structured telephone interview. Follow-up data were complete in 2621 patients (97.1%).
Values are presented as means±S.D. Non-normally distributed biomarker data were log-transformed to achieve a normal distribution. Post-hoc testing using Bonferroni's test was based on ANOVA of log-transformed data. Cox proportional-hazards regression was used to analyse the effect of risk factors on outcome in single predictor and multivariable analyses. To evaluate the added predictive ability of the new marker, an NRI (net reclassification improvement) was applied .
In a subgroup analysis, we evaluated the GRACE (Global Registry of Acute Coronary Events) score for risk assessment, which comprises predictive factors of Killip class, systolic BP (blood pressure), HR (heart rate), age, creatinine and other risk factors such as cardiac arrest at admission, ST segment elevation and elevated cardiac enzyme levels, as described previously . Survival curves plotted by the Kaplan–Meier method were used for illustrative purposes. Time-dependent ROC (receiver operating characteristic) curves and time-dependent AUC (area under the curve) values are determined from censored survival data using the Kaplan–Meier method . To test whether MR-proANP provides additive information to Tn-I, clinical variables or cardiovascular risk factors, we used the likelihood ratio χ2 test. Using the ROC decision plot, we determined the optimal cut-off value of MR-proANP by maximizing the product of sensitivity×specificity.
A P<0.05 was considered statistically significant and evaluated with appropriate non-parametric tests. To assess a statistical difference between the AUCs in ROC analysis, we applied a DeLong test using Analyse-it for Microsoft Excel (version 2.20; Analyse-it Software Ltd). All other statistical analyses were performed using R version 2.5.1 (http://www.r-project.org, library Design, Hmisc, ROCR) and SPSS Statistics software for Windows version 19.
Our study population comprised 2700 consecutive patients with symptomatic CAD, all of whom underwent coronary angiography. 1316 patients (48.7%) received a final diagnosis of ACS, 1384 (51.3%) a diagnosis of SAP. Detailed information of demographic data are presented in Table 1.
A total of 435 patients (16.1%) presented with elevated levels of creatinine above the respective reference limits for males (>1.1 mg/dl) or females (>0.8 mg/dl). Renal function assessed by eGFR (estimated glomerular filtration rate) showed no significant difference between patients with ACS and SAP (P>0.05; Table 1). Patients with ACS presented with a significantly lower LVEF than those without ACS (P=0.0005; Table 1).
The median Tn-I level at baseline was 0.05 [IQR (interquartile range), 0.029–0.25] μg/l and the median of MR-proANP at baseline was 172 pmol/l with an IQR of 107–289 pmol/l (Table 1). Although Tn-I levels were significantly different between patients with ACS and those with SAP (P<0.001), no such difference was detected for MR-proANP (P=0.34).
Primary and secondary end points
All-cause death was defined as the primary end point, which occurred in 76 patients (2.8%) at 3-month follow-up. A total of 183 (6.8%) patients experienced the composite secondary end point of recurrent AMI in 98 (3.6%), ischaemic stroke in 19 (0.7%) and all-cause mortality. Ten patients experienced more than one event during follow-up. A total of 299 (11.1%) patients underwent a revascularization procedure within 3 months of follow-up, which did not influence clinical outcome in survival analysis (results not shown).
Median MR-proANP levels in patients who experienced against those who did not experience the primary outcome were 357 (IQR, 181–663) pmol/l compared with 170 (IQR, 106–282) pmol/l respectively (P<0.001). MR-proANP concentrations were also higher for the composite outcome (all post-hoc P values <0.001), compared with patients without an event within 90 days (Figure 1A). For Tn-I, only all-cause death showed significantly higher concentrations in patients who experienced an event compared with those without events (post-hoc P=0.024, other comparisons P>0.2) (Figure 1B).
Results of single predictor and multivariable Cox regression analysis are summarized in Table 2. MR-proANP is the strongest predictor for both end points in single predictor analysis, significantly stronger than Tn-I, and stronger than all other clinical variables combined, that is age, gender, LVEF as well as five cardiovascular risk factors (Table 2). MR-proANP adds significantly to the clinical model (joint χ2 90.3, P<0.0001 for outcome death from any cause, and joint χ2 78.9, P<0.0001 for the combined end point). This was not the case for Tn-I (Table 2). These results remained essentially unchanged when subgroups of patients with ACS and SAP were analysed separately. No material change was noted after the addition of the variables BMI (body mass index) and creatinine, both of which were available for only a limited number of patients, 50.5% and 46.8% respectively. For the entire cohort, the non-significant joint χ2 for death from any cause for the clinical model including BMI, creatinine and Tn-I increased from 14.1 (P=0.23) to 35.4 (P=0.0004) after the addition of MR-proANP. The corresponding χ2 value for the combined end point was 12.7 (P=0.31), which increased to 45.4 (P<0.0001) after the addition of MR-proANP. The added value of MR-proANP to the GRACE score was higher than that of Tn-I (Table 3).
The added value of MR-proANP/Tn-I at high risk as compared with the clinical model provided for composite end point NRI=15.3% (MR-proANP), NRI=0% (Tn-I) and for all-cause death NRI=26.6% (MR-proANP), NRI=9% (Tn-I). In a subgroup analysis of 1384 patients, we applied the GRACE score for risk assessment (Table 3). The added value of MR-proANP/Tn-I at high risk as compared with the GRACE score provided for a composite end point NRI=23.5% (MR-proANP), NRI=10.9% (Tn-I), and for all-cause death NRI=32.9% (MR-proANP), NRI=16% (Tn-I).
Time-dependent ROC curves of MR-proANP and Tn-I were compared. For MR-proANP, we calculated an AUC of 0.73 [95% CI (confidence interval), 0.67–0.79] for all-cause death after 3 months (Figure 2A), and an AUC of 0.63 (95% CI, 0.59–0.68) for the combined secondary end point (myocardial reinfarction, ischaemic stroke and all-cause death). The corresponding AUCs of the Tn-I values were 0.58 (95% CI, 0.55–0.62) for all-cause death (Figure 2A), and 0.54 (95% CI, 0.51–0.56) for the composite of adverse events. AUCs of MR-proANP were significantly higher compared with the corresponding AUCs of Tn-I for all-cause death (DeLong test, P=0.0024) and for the combined end point (DeLong test, P=0.0038). Missing variables for Tn-I scarcely affected the significant difference of Tn-I levels between patients with ACS and SAP (P<0.001).
According to the ROC decision plot for all-cause death, we found that the optimal cut-off value to correctly predict the primary outcome at 90 days was an MR-proANP concentration of 236 pmol/l, which yielded a sensitivity of 65.8% (95% CI, 54.0–76.3) and a specificity of 66.9% (95% CI, 65.0–68.7) (Figure 2B).
Applying the optimal cut-off value into the Kaplan–Meier survival analysis, patients with an elevated baseline MR-proANP concentration (≥236 pmol/l) showed a significantly increased rate of all-cause death at 3 months as compared with patients with a lower concentration level (log rank, P<0.001) (Figure 3A). Illustration of the predictive performance via tertiles in Kaplan–Meier analysis demonstrated a cut-off value on a similar level at 240 pmol/l (third tertile of MR-proANP) revealing a significantly enhanced rate of all-cause death at 3 months in patients with enhanced MR-proANP concentration compared with patients with lower concentration (log rank, P<0.001) (Figure 3B).
The major findings of the present study are: (i) patients with an increased baseline MR-proANP concentration showed a significantly elevated rate of fatal outcome (all-cause death), as well as of the combined secondary end point comprising recurrent AMI, stroke and death at 3 months of follow-up as compared with patients with a lower level; (ii) the predictive performance of MR-proANP is superior to that of clinical variables, LVEF and classical cardiovascular risk factors; (iii) patients with an enhanced baseline MR-proANP concentration (≥236 pmol/l) are more likely to reach the primary end point of all-cause death than patients with initially increased Tn-I levels.
The use of current troponin assays is hampered by a relatively late onset of biomarker detectability , even with recently introduced high-sensitivity troponin assays [25,26]. Thus several other diagnostic biomarkers have been tested recently, and interesting candidate markers include the so-called upstream biomarkers, which are detected earlier and independent of cell necrosis may have great potential for an appropriate and rapid clinical decision-making in patients with imminent ACS. Our group has described previously an increased surface expression of platelet collagen receptor glycoprotein VI in patients with ACS and ischaemic stroke, which was associated with major adverse cerebrovascular and cardiovascular events [27–30].
Natriuretic peptides such as MR-proANP and BNP are released during haemodynamic stress or myocardial injury, prior to myocardial necrosis in patients with AMI. More importantly, MR-proANP is able to identify high-risk patients irrespective of the clinical circumstances, that is, it remains a reliable prognostic marker in patients with SAP but also in those with ACS. In all of our methods, MR-proANP performed better than troponin in assessing the patients' risk of death or of the composite outcome (Table 2). This was also true when not only the clinical variables alone but also the GRACE score were used in combination with MR-proANP compared with Tn-I, as is visible from the higher χ2 value (Table 3). The synthesis and release of natriuretic peptides as counter-regulatory hormones during haemodynamic stress is associated with reduced LVEF and independent study groups have found that in particular MR-proANP provides an incremental value of clinical outcome compared with BNP in patients with acute as well as CHF (chronic heart failure) . Intriguingly, patients with a first anterior AMI have been shown to benefit from an intravenous ANP administration, which helps in preventing left ventricular remodelling .
Very recently, a Dutch group published parts of a large population-based observational cohort study showing that MR-proANP is associated with all-cause death and cardiovascular events in the general population . Another study focused on the prognostic value of MR-proANP in patients with AMI. However, this study evaluated the biomarkers after the confirmed diagnosis of AMI, was restricted to AMI, and follow-up of patients showed a large variation ranging from 0 to more than 700 days . However, our study was not designed to confirm the diagnostic value of MR-proANP in AMI, but focused on the prognostic value for the prediction of all-cause death in patients presenting with either ACS or SAP to the cardiac catheterization laboratory. To assess the prognostic power of MR-proANP compared with other markers of clinical outcome such as troponin, χ2 value derived from Cox proportional hazard analysis proves to be a helpful statistical tool .
Determination of the optimal cut-off value remains challenging and is usually considered as a trade-off between true positives and true negatives, and should provide both the highest sensitivity and specificity, as an ROC curve offers the graphical illustration of these trade-offs [34,35]. Thus rather simple statistical tools applying the 50th percentile or a bisecting line may help to determine the optimal cut-off value. Using ROC decision plot analysis, we chose the method of a bisecting line. Maximizing the product of sensitivity and specificity, the optimal cut-off value of MR-proANP for all-cause death after 3 months was 236 pmol/l. This cut-off value helped to predict a significantly higher event rate of all-cause death in patients with a baseline increased MR-proANP concentration and has been found on a similar level at 240 pmol/l (third tertile of MR-proANP) via tertiles in Kaplan–Maier survival analysis. Not surprisingly, this cut-off value is significantly higher than the cut-off that has been suggested to help in the correct rule-out of acute heart failure in patients presenting to the emergency department with acute shortness of breath, which is 120 pmol/l .
The AUCs of the ROC curves were significantly higher for MR-proANP than for Tn-I. This difference may be seen both for the combined secondary end point of major adverse cardiovascular events such as recurrent AMI, ischaemic stroke and all-cause death after 90 days and in particular for the primary end point of all-cause death alone.
As ROC analysis has not been considered as the ideal tool for the assessment of novel predictors due to a possible lack of sensitivity, new statistical tools [e.g. NRI and IDI (integrated discrimination improvement)] have been introduced lately to investigate the ‘added value’ of new biomarkers . Thus we preferred to use NRI instead of IDI, since an IDI would not show a clear indication of what is a clinically important value .
First, validity of the optimized diagnostic cut-off value needs to be tested in independent, blind comparisons with a reference standard among a consecutive series of patients suspected (but not known) to have the target disorder, including missing and indeterminate results, and replicating studies in other settings, since both specificity and sensitivity may change, as the same diagnostic test is applied in primary, secondary and tertiary care. There were no Tn-I data available in 47 patients with SAP, which may have influenced the prognostic assessment, although the missing variables scarcely affected the significant difference of Tn-I levels between patients with ACS and SAP (P<0.001). Likewise, values for BMI and baseline creatinine were only available in 50.5% and 46.8% of patients respectively, which forced us to calculate Cox proportional hazard models without these variables. Recalculation including the missing two variables in the remaining patients, however, verified our findings.
Although we encountered an acceptable return of information by hospital chart analysis and telephone interviews (97.1% follow-up data completion) and found significant differences in Kaplan–Meier survival analysis, the statistical power could have been improved with higher event rates or longer follow-up. Therefore larger multicentre studies should verify the results in catheterization laboratory cohorts with the suggested cut-off value and substantiate the prognostic value of MR-proANP.
Moreover, future studies may design a multimarker panel of MR-proANP and several additive prognostic biomarkers to improve risk stratification in patients with symptomatic CAD. Thus the comparison with BNP or NT-proBNP may have provided additional prognostic information. To date, despite some promising issues of multimarker assessments providing additive prognostic information for adverse clinical outcome, the results of the studies have been inconsistent [5,37,38].
In conclusion, MR-proANP is a novel biomarker that may help in the prediction of all-cause mortality and major adverse cardiovascular events in patients presenting with symptomatic CAD. Further studies should substantiate its prognostic value and confirm the appropriateness of the optimal cut-off value determined in this study.
This study was supported by the Deutsche Forschungsgemeinschaft [grant number MA2186/3-1&GK794 (to M.G.); grant number Klinische Forschergruppe KFO274: Platelets-Molecular Mechanisms and Translational Implications] and the German Cardiac Society (DGK) ‘molecular imaging of atheroslerotic plaques’ (grant to B.B.).
Stephan von Haehling conceived the hypothesis and wrote the first draft of the paper. Jana Papassotiriou and Oliver Hartmann conducted the statistical analyses. Meinrad Gawaz, Wolfram Doehner and Rene Botnar provided critical input at all stages and substantially revised the paper. Tobias Geisler, Thomas Wurster and Andreas Schuster provided critical input and revised the paper. Boris Bigalke and Konstantinos Stellos contributed to the accumulation, analysis and interpretation of the data, and substantially revised the paper. All authors have read and approved the manuscript. Stephan von Haehling and Boris Bigalke are responsible for the final approval of the submitted paper.
We thank Franka März, Ariadne Seither and Hanna Schnell for expert technical assistance.
Abbreviations: ACS, acute coronary syndromes; AMI, acute myocardial infarction; ANP, atrial natriuretic peptide; AUC, area under the curve; BMI, body mass index; BNP, B-type natriuretic peptide; CAD, coronary artery disease; CHF, chronic heart failure; CI, confidence interval; CK, creatine kinase; CK-MB, CK myocardial band; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; GRACE, Global Registry of Acute Coronary Events; IDI, integrated discrimination improvement; IQR, interquartile range; LVEF, left ventricular ejection fraction; MR-proANP, mid-regional pro-ANP; NRI, net reclassification improvement; NT-proBNP, N-terminal pro-BNP; ROC, receiver operating characteristic; SAP, stable angina pectoris; STEMI, ST-elevation myocardial infarction; NSTEMI, non-STEMI; Tn-I, troponin-I
- © The Authors Journal compilation © 2012 Biochemical Society