Low Circulating Levels of Growth Differentiation Factor-15 Before Coronary Artery Bypass Surgery May Predict Postoperative Atrial Fibrillation Olivier Bouchot, MD, PhD,* Charles Guenancia, MD, MSc,†‡ Abdelkader Kahli, MSc,‡ Charline Pujos, MD,* Ghislain Malapert, MD,* Catherine Vergely, PharmD, PhD,‡ and Gabriel Laurent, MD, PhD†§ Objectives: To assess the role of growth differentiation factor-15 (GDF-15) as a potential new predictor of postoperative atrial fibrillation (POAF) after off-pump (OFP) and on-pump (ONP) coronary artery bypass graft (CABG) surgery. Design: Prospective, single-center, observational study. Setting: University teaching hospital. Participants: The first 50 patients planned for OFP surgery and the first 50 patients planned for ONP surgery among patients referred for CABG with the following exclusion criteria: age o18 or 480 years, previous atrial fibrillation/ flutter, previous treatment with amiodarone, previous cardiac surgery, and emergency surgery. Interventions: Included patients were equipped with longduration (7 days) Holter-ECG monitoring. Measurements and Main Results: POAF was defined as an AF episode lasting 430 seconds. All patients underwent preoperative echocardiography to assess left ventricular ejection fraction and left atrial diameter. GDF-15 levels were
assessed after induction of anesthesia and 12 hours after arrival at the intensive care unit. Among the 100 patients, 34 (34%) developed POAF. In Cox multivariate regression analysis, the EuroSCORE, left atrial diameter 445 mm, and low GDF-15 levels at induction were associated independently with the onset of POAF. In contrast, preoperative NT-proBNP levels did not predict POAF. The use of ONP surgery was not associated with a higher incidence of POAF, even though baseline and follow-up characteristics in ONP and OFP patients were identical. Conclusions: In patients with no history of AF, a low plasma level of GDF-15 before CABG surgery was a strong independent predictor of POAF. Moreover, preoperative plasma GDF-15 levels added an incremental predictive value to classic risk factors of POAF. & 2015 Elsevier Inc. All rights reserved.
P
EuroSCORE in 1,458 cardiac surgery patients and an independent predictive factor of postoperative mortality in these patients.8 In a recently published study, the plasma GDF-15 level was associated with nonvalvular, nonpostoperative paroxysmal AF.9 In this observational study, the objective was to assess the role of GDF-15 as a potential new marker of POAF after offpump (OFP) and on-pump (ONP) coronary artery bypass graft (CABG) surgery.
OSTOPERATIVE ATRIAL FIBRILLATION (POAF) has a high prevalence, affecting 20% to 45% of CABG surgery patients within 7 days of the procedure.1,2 POAF increases the duration of hospitalization and healthcare costs, and has been associated with an increased incidence of postoperative stroke, the need for a permanent pacemaker, and early and late mortality.3 Male sex, valvular heart disease, left atrial enlargement, obesity, previous cardiac surgery, chronic lung disease, discontinuation of beta-blockers or ACE inhibitors, AF history, pericarditis, and inflammation are predictive factors for POAF, but advanced age has the strongest and most consistent correlation.3,4 The identification of these risk factors has led to modifications in several intraoperative practices. Because there are many interconnected factors involved in the pathophysiology of POAF, it is difficult to identify the contribution of each to the resulting condition. Therefore, the huge release of inflammation markers triggered by CABG cardiac surgery may not fully explain the occurrence of POAF,5 and there is still a critical need for preoperative clinical predictors. Growth differentiation factor-15 (GDF-15) is a stressresponsive member of the transforming growth factor-β superfamily that originally was cloned as a macrophage-inhibitory cytokine.6 Increased expression of this cytokine can be induced by myocardial stretch, left atrial pressure overload (11), and experimental cardiomyopathy as well as oxidative stress, inflammatory cytokines, and ischemia/reperfusion,7 suggesting that the plasma levels of this cytokine are related not only to myocardial dysfunction but also to circulatory stress. This biomarker thus seems particularly relevant in the setting of cardiac surgery, in which almost all these mechanisms are involved. A recent study has shown that preoperative plasma levels of GDF-15 were useful to improve the risk stratification of the
KEY WORDS: atrial fibrillation, biochemistry, coronary artery bypass grafts, cardiopulmonary bypass, off-pump surgery
MATERIAL AND METHODS
Study Design The independent ethics committee of the University Hospital of Dijon approved the study protocol, and written informed consent was obtained from the patients or their legal representatives. The study was designed as a prospective observational study. All investigations were conducted in accordance with the principles outlined in the Declaration of Helsinki.
From the *Department of Cardiothoracic Surgery, University Hospital, Dijon, France; †Department of Cardiology, University Hospital, Dijon, France; ‡INSERM, U866, LPPCM, Dijon, France; and §CNRS, UMR 5158, Le2I, Dijon, France. Dr. O. Bouchot and Dr. C. Guenancia contributed equally to this work. This work was supported by the University Hospital of Dijon and by grants from Conseil Régional de Bourgogne. Address reprint requests to Charles Guenancia, Service de Cardiologie, CHU Dijon, 14 rue Paul Gaffarel, 21079 Dijon Cedex, France. E-mail: [email protected] © 2015 Elsevier Inc. All rights reserved. 1053-0770/2601-0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2015.01.023
Journal of Cardiothoracic and Vascular Anesthesia, Vol 29, No 5 (October), 2015: pp 1131–1139
1131
1132
Selection of Patients All of the patients operated on by 2 surgeons for CABG surgery at the University Hospital of Dijon from September 2011 to March 2013 were screened for participation in this prospective, observational study. The exclusion criteria were age o18 years or 480 years, previous atrial fibrillation/flutter, previous treatment with amiodarone, previous cardiac surgery, and emergency surgery. This observational study was designed to recruit the first 50 patients planned for on-pump surgery and the first 50 patients planned for off-pump surgery among patients meeting the inclusion criteria. The surgical strategy (ONP v OFP) was decided by the medical and surgical team according to the anatomy of the coronary arteries. Data Collection Exhaustive clinical data were collected at admission, and the following variables were recorded: age, sex, cardiovascular risk factors, cardiovascular and pulmonary diseases, previous regular medication, and preoperative echocardiographic parameters. Medications administered during the hospital stay (but not the doses) were recorded. Previous renal failure was defined as preoperative estimated glomerular filtration rate (eGFR) o30 mL/min/1.73 m2 using the MDRD formula.10
BOUCHOT ET AL
On-pump surgery was performed according to the following protocol: aortic and cavoatrial cannulation, and cardiopulmonary bypass (CBP) was started. Myocardial protection was induced and maintained by warm blood anterograde cardioplegia. Normal body temperature (361C) was used. Depending on the patients’ grafts and characteristics of the target vessels, the right internal mammary artery or the saphenous vein and/or the left internal mammary artery (LIMA) were used for the grafts. All arterial grafts were treated with papaverine to avoid vasospasm. After completion of the CABG (from 1 to 6 CABGs in the patients studied), CPB was discontinued and protamine was given (1:1) for heparin reversal. Off-pump surgery was performed according to the following protocol: cardiac stabilization and displacement was obtained using Cor vasc OPCAB (Coroneo, Montréal, Canada). Anastomoses were constructed using temporary occlusion, with thermoreversible LeGoo gel (SANOFI, Paris, France).12 The LIMA was investigated primarily and anastomosed to the LAD. The right internal thoracic artery was used pediculised or with the Y graft technique on the LIMA and the distal part was anastomosed to the ramus intermedius. After closure of the sternum, the patients were transferred to the postoperative intensive care unit (ICU) and, finally, to the surgery ward.
Echocardiography All of the patients underwent preoperative transthoracic echocardiography (TTE). The left ventricular ejection fraction (LVEF) was calculated using the Simpson method on the apical four-chamber and apical two-chamber views. LVEF was dichotomized at 45% for more clinical relevance. LA anteroposterior diameter was calculated in the parasternal long-axis view, using the M-mode technique, measuring from the trailing edge of the anterior LA wall to the leading edge of the posterior LA wall. For clinical relevance, LA diameter was dichotomized at 45 mm following classic cut-off values.11 Anesthesia and Heart Surgery Procedure Patients were premedicated with midazolam orally plus hydroxyzine 90 minutes before anesthesia. Routine cardiac medications were continued until the morning of the surgery, except for clopidogrel, which was stopped at least 5 days earlier. Before the induction of anesthesia, a complete hemodynamic monitoring system was set up in the operating room. Anesthesia was induced with intravenous midazolam (0.02 mg/ kg), sufentanil (0.2 to 0.5 mg/kg/h), and propofol (1.5 to 2.5 mg/kg). After verifying that manual ventilation was satisfactory, cisatracurium dibesylate (0.06 mg/kg/h) was injected. Patients were intubated orally and ventilated with FIO2: 0.4. Anesthesia was maintained with sufentanil and cisatracurium as required and inhaled desflurane. Surgical Technique Surgical management was standardized. After median sternotomy and graft harvesting (left and right internal mammary arteries ⫾ saphenous graft), the patient received heparin (1.5 mg/kg in the off-pump group versus 3 mg/kg in the onpump group).
Diagnosis of POAF All patients underwent Holter ECG monitoring (Spider Flash, Sorin Group France) started immediately after inclusion. The Holter ECG device was programmed to record every arrhythmic event for 7 days. The device was consistent with routine use in the ICU, because only 3 electrodes were required, thus providing a 2-lead recording, and was easily removable if necessary for patient care or to perform imaging exams. The Holter ECG monitor was removed after 7 days of recording, or at death (whichever occurred first). An experienced cardiologist who was blinded to the patient’s clinical data analyzed the Holter ECG recordings. If the diagnosis was uncertain, a second cardiologist, blinded to the first results, also analyzed the tracings. No cases of discordance between the 2 analyses occurred. AF was diagnosed according to the guidelines of the European Society of Cardiology for the interpretation of Holter ECG recordings (ie, any arrhythmia that presents the ECG characteristics of AF; namely, “the surface ECG shows ‘absolutely’ irregular RR intervals, there are no distinct P waves on the surface ECG, and the atrial cycle length [when visible], ie, the interval between 2 atrial activations, is usually variable and o200 ms [4300 bpm]”) and lasting at least 30 seconds on a rhythm strip, should be considered AF.9 Silent AF was defined as the occurrence of AF on the Holter ECG recording, regardless of the duration or number of episodes, and in the absence of any mention of AF in the medical file during the first 7 days of the hospital stay (ie, AF not diagnosed by the ICU physicians). In contrast, symptomatic AF was defined as any AF diagnosed by a physician during the hospital stay. When AF was diagnosed on the Holter ECG recording, the following data were collected: date and time of the first
1133
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
episode, ventricular response to the first episode and mean ventricular response to AF, the total burden of AF (total duration of all episodes of AF), and the number of episodes. Nonsustained supraventricular tachycardia (o30 s) also was recorded from the automated count for all patients, and the daily mean ventricular rate was calculated. Heart rate variability was calculated automatically by the Holter software on sinus rhythm strips at day 1 and day 7, as was the mean value for the whole recording. Treatment of New-Onset AF After discharge, each patient’s daily medical file and treatment sheet were analyzed by a physician unaware of the Holter ECG monitoring results, to evaluate the number of AF cases not diagnosed by the clinicians (silent AF) and the treatment choices when AF was identified (rate or rhythm control, medical or electrical cardioversion, anticoagulant therapy). Blood Samples The following measurements were made at admission and every 24 hours during the ICU stay: total blood count, electrolytes, corrected calcemia, creatinine, cardiac troponin Ic, NT-proBNP, CRP, and lactates and blood protein levels. Determination of GDF-15 Arterial blood samples were taken from the arterial catheter immediately after the induction of general anesthesia and 12 hours after arrival at the ICU.13 The blood samples were centrifuged immediately after collection and the plasma was frozen immediately in liquid nitrogen and stored at –801C until analysis. Plasma GDF-15 concentrations were measured by quantitative sandwich enzyme immunoassay (Human GDF-15, Quantikine, R&D Systems Europe, Lille, France) with a linear range from 200 to 50,000 ng/L. The color intensity, relative to GDF-15 concentration, was measured at 450 nm with a spectrophotometer (VictorV3, Perkin Elmer, Courtaboeuf, France). Complete data sets including GDF-15 levels were available from all 100 patients Clinical Follow-Up Lastly, clinical outcomes were recorded over 1-year follow-up, namely: length of stay in the ICU and in the hospital, death in the ICU, at discharge (or at day 28, whichever came first) and at 1 year. The presence of AF and the occurrence of stroke between discharge and the follow-up time points were recorded. Follow-up information at 1 year was obtained by phone call to either the patient or the patient’s family or the general practitioner. One-year follow-up was obtained for all patients.
correlations were computed to assess the relationship between variables. In accordance with the recommendations of Harrell et al on multivariate prognostic modeling,14,15 variables were selected on the basis of their proven clinical, pathophysiologic, and epidemiologic relevance to the endpoint in question. Before the construction of the multivariate models, collinearity between variables was excluded. The correlation matrix of multivariate model showed that the correlation between variables (r-coefficient) did not exceed positive or negative value of 0.25. Because the purpose of this study was to determine whether GDF-15 predicted AF in the study cohort, this variable was included in the analyses. Three more variables, age, left atrial diameter o45 mm, and the EuroSCORE, also were chosen on the basis of their univariate association with POAF in the present study. Given the relatively small size of the population, the EuroSCORE was chosen because it demonstrated a strong univariate association with AF occurrence and included several univariate variables that were significantly different between the POAF and SR groups (previous myocardial infarction, lung disease, LVEF). Because age correlated with high GDF-15 plasma levels, and despite its presence in the EuroSCORE, this variable was included in the multivariate model. This strategy already has been used in the particular setting of POAF in a study in which EuroSCORE and age were shown to be predictive factors of POAF.16 The exploratory identification of factors associated with AF was performed using Cox multivariate descending conditional regression, with inclusion and exclusion cut-offs at 5%. To improve the robustness of results, the POAF patients then were matched 1:1 with the SR patients using nearest neighbor matching on the linear propensity score with a maximum caliper width of 0.2 standard deviations. The same Cox proportional-hazard model was used to compare AF with SR patients with regard to time to the first occurrence of AF. Moreover, to test for the incremental predictive value of NT-proBNP and GDF-15 levels on classic predictors of POAF, a multivariate model that included age, the EuroSCORE and left atrial diameter, based on data in the literature, was built for the whole set of patients (Model 1). These 2 biomarkers then were added to the model and the changes in the global χ2 of the model were established. All multivariate models were tested for multicollinearity and were found to be stable: the tolerance ranged from 0.47 for EuroSCORE to 0.96 for LA 445 mm. The variation inflation factor ranged from 1.04 for LA 445 mm to 2.13 for EuroSCORE. All the tests were two-sided, and a p value o0.05 was considered significant. All analyses were performed using SPSS 20.0 0 (SPSS, Inc., Chicago, IL). RESULTS
Statistical Analysis
Patient Characteristics
Continuous variables are presented as means ⫾ standard deviations when normally distributed or medians and ranges otherwise; and categoric variables as numbers (percentages). Patients were classified into the AF or sinus rhythm (SR) group according to the onset of AF on the Holter tracings. The characteristics of both groups were compared using the exact Mann-Whitney U test or t test for continuous variables and χ2 or Fisher’s exact test for categoric variables. Spearman rank
During the study period, POAF occurred in 34 patients (34%) (AF group); the remaining 66 patients (66%) remained in permanent sinus rhythm (SR group). Factors Associated With POAF Comparing baseline characteristics (Table 1), patients developing POAF were significantly less likely to be smokers (18% v 38%, p ¼ 0.04), and were more likely to have previous
1134
BOUCHOT ET AL
Table 1. Baseline Characteristics According to the Occurrence of Atrial Fibrillation
Characteristics
Risk factors Age, years Female sex Systemic hypertension Diabetes mellitus Smoking Obesity (BMI Z30 kg/m2) Hypercholesterolemia Medical coronary history Recent MI Previous chronic renal failure Previous chronic respiratory failure Previous sleep apnea Previous treatments Beta blockers Diuretics Statins ACE inhibitors VKA Clinical data NYHA41 EuroSCORE LVEF (%) LVEF o45% LA diameter LA diameter 445 mm Serum creatinine, mmol/L NT-proBNP, pg/mL
AF Group
SR Group
(n ¼ 34)
(n ¼ 66)
p Value
66 ⫾ 7 3 (9) 21 (62) 11 (32) 6 (18) 19 (56) 27 (79)
63 ⫾ 9 5 (8) 43 (65) 25 (38) 25 (38) 32 (48) 50 (76)
0.06 1 0.74 0.59 0.04 0.48 0.68
15 (44) 3 (9)
18 (27) 7 (11)
0.09 1
2 (6)
7 (11)
0.71
8 (24)
5 (8)
0.03
31 14 33 28 2
(91) (41) (97) (82) (6)
15 (44) 4 (2-6.3) 54 ⫾ 12 7 (21) 42 (37-45) 7 (21) 88.5 (77-101) 150 (77-1061)
53 16 64 54 3
(80) (24) (97) (82) (5)
0.25 0.08 1 1 1
31 (47) 3 (1-5) 57 ⫾ 11 2 (3) 39.5 (35-43) 3 (5) 88 (74-102.5) 146 (61-491)
0.79 0.07 0.22 0.007 0.08 0.03 0.86 0.33
NOTE. Values expressed as n (%), median (interquartile range), or mean ⫾ SD. Abbreviations: ACE, angiotensin conversion enzyme; AF, atrial fibrillation; BMI, body mass index; EuroSCORE, European System for Cardiac Operative Risk Evaluation; LA, left atrial; LVEF, left ventricular ejection fraction; MI, myocardial infarction; NT-proBNP, N terminal pro brain natriuretic peptide; NYHA, New York Heart Association; SR, sinus rhythm; VKA, vitamin K antagonists.
sleep apnea (24% v 8%, p ¼ 0.03), left atrial dilation (21% v 5%, p ¼ 0.03), and LVEF impairment (21 v 3%, p ¼ 0.007). Although not statistically significant, POAF patients were older, had a higher EuroSCORE, and were more likely to have previous myocardial infarction and to be treated with diuretics. Moreover, POAF patients were comparable to SR patients regarding surgery characteristics, including the proportion of off-pump and on-pump procedures. ICU management differed significantly concerning the need for inotropic support (catecholamine infusion) (27% v 8%, p ¼ 0.02) and the prescription of amiodarone (50% v 20%, p ¼ 0.002). In univariate testing, all biologic variables in the 2 groups, including GDF-15 at induction and in the ICU, were comparable. Tables 2 and 3 Characteristics of POAF Approximately one-third of AF (13 of 34) patients were defined as having silent AF, because they were not diagnosed
by clinicians during the hospital stay. When compared with symptomatic AF, silent AF was associated significantly significantly associated with a lower AF burden (345 ⫾ 774 min v 1,542 ⫾ 1,987 min, p ¼ 0.02) and a longer time between surgery and the first AF episode (3.4 ⫾ 3 days v 1.7 ⫾ 1.1 days, p ¼ 0.03). The first episode of AF occurred within the first 48 hours of monitoring in 23 of the 34 AF patients (67%), and the mean burden of AF was 18 ⫾ 28 hours. Half of the AF patients received an anti-arrhythmic drug during their hospital stay (amiodarone).
Determinants of GDF-15 at Baseline Figure 1 shows the distribution of GDF-15 according to the time of blood sampling, to the onset of POAF, and to the type of surgery. Preoperative circulating plasma levels of GDF-15 correlated significantly with the EuroSCORE (r ¼ 0.54; p o 0.001), with age (r ¼ 0.40; p o 0.001), with preoperative NT-proBNP levels (r ¼ 0.31, p ¼ 0.002), and with creatinine (r ¼ 0.36; p o 0.001). In contrast, no correlation was observed between GDF-15 levels and echocardiographic parameters. Figure 2 shows the relationship scatterplots of GDF-15 immediately after the induction of general anesthesia with age, EuroSCORE, and LA diameter. Regarding categoric baseline variables, GDF-15 levels were significantly higher in patients with diabetes (1.3 ⫾ 0.7 ng/L v 1 ⫾ 0.4 ng/L, p ¼ 0.003), a history of high blood pressure (1.2 ⫾ 0.6 ng/L v 1 ⫾ 0.5 ng/L, p ¼ 0.05), previous renal failure (1.8 ⫾ 0.8 ng/ L v 1 ⫾ 0.5 ng/L, p o 0.001), peripheral artery disease (1.5 ⫾ 0.7 ng/L v 1 ⫾ 0.5 ng/L, p o 0.001), and dyspnea as assessed by NYHA 41 (1.3 ⫾ 0.6 ng/L v 0.9 ⫾ 0.5 ng/L, p ¼ 0.002). The determinants of baseline levels of GDF-15 in the present study were consistent with those in a previous study conducted in the authors’ laboratory, in which associations were found between GDF-15 levels at induction before CABG surgery and age, the EuroSCORE, diabetes, renal function, and NT-proBNP.13 Moreover, in patients free of clinically overt cardiovascular disease, several studies found that GDF-15 levels were linked to age, diabetes, hypertension, impaired renal function, and NT-proBNP levels.17,18
Cox Multivariate Regression After Cox multivariate regression analysis (Table 3), the EuroSCORE, LA diameter 445 mm, and GDF-15 immediately after anesthesia induction were associated independently with the onset of POAF. Figure 3 provides unadjusted (A) and adjusted (B) POAF-free survival curves according to the median baseline GDF-15 level (¼ 1,013 ng/L). After propensity score matching, the EuroSCORE (odds ratio [OR]: 1.29, 95% confidence interval [CI]: 1.07-1.55, p ¼ 0.008), GDF-15 immediately after anesthesia induction (OR: 0.41, 95% CI: 0.17-0.95, p ¼ 0.038) and LA diameter 445 mm (OR: 0.39, 95% CI: 0.16-0.93, p ¼ 0.034) remained associated independently with the onset of POAF. When comparing the incremental prognostic value of GDF15 according to the time of blood sampling, it was found that
1135
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
Table 2. Clinical Characteristics According to the Occurrence of Atrial Fibrillation AF Group (n ¼ 34)
Characteristics
Surgery data Number of CABG 1 2 3 4 5 6 CBP CBP duration (min) Aortic cross-clamp duration (min) ICU management Mechanical ventilation duration (hours) ICU stay duration (hours) Norepinephrine Inotropic catecholamine Amiodarone Biologic data pH 3 hours after surgery pH 24 hours after surgery Troponin 3 hours after surgery, ng/mL Troponin 24 hours after surgery, ng/mL CRP peak, mg/L NT-proBNP peak, pg/mL GDF-15 at induction, ng/L GDF-15 12 hours after arrival to the ICU, ng/L In-hospital follow-up Renal failure Sepsis Stroke Myocardial infarction Treatment at discharge Beta-blockers Statins ACE inhibitors VKA
SR Group (n ¼ 66)
p Value
0.43 1 (3) 10 (29) 12 (35) 8 (24) 2 (6) 1 (3) 15 (44) 104 ⫾ 26 79 ⫾ 22
1 (2) 11 (17) 28 (42) 22 (33) 4 (6) 0 35 (53) 94 ⫾ 28 71 ⫾ 22
0.40 0.24 0.23
6⫾4 68 ⫾ 84 12 (35) 9 (27) 17 (50)
6⫾5 65 ⫾ 39 31 (47) 5 (8) 13 (20)
0.55 0.81 0.26 0.02 0.002
7.37 ⫾ 0.06 7.37 ⫾ 0.04 1.95 (0.49-6.35) 1.9 (1.1-3.4) 203.3 ⫾ 99.6 156 (75-1865) 1,002 (788-1230) 2,544 (1868-3207) 5 (15) 3 (9) 0 0 28 29 21 9
(85) (88) (64) (27)
7.39 ⫾ 0.06 7.37 ⫾ 0.04 3.5 (0.9-5.8) 2.3 (1.7-5) 194.3 ⫾ 85.8 150 (65-531) 1,027 (687-1365) 2,759 (1,861-4,261)
0.12 0.66 0.33 0.19 0.64 0.32 0.74 0.57
14 (21) 1 (2) 0 3 (5)
0.59 0.11 — 0.55
60 (91) 36 (96) 41 (62) 0
0.49 0.22 0.88 o0.001
NOTE. Values expressed as n (%), median (interquartile range), or mean ⫾ SD. Abbreviations: ACE, angiotensin converting enzyme; AF, atrial fibrillation; CABG, coronary artery bypass graft; CBP, cardiopulmonary bypass; CRP, C-reactive protein; GDF-15, growth differentiation factor-15; ICU, intensive care unit; NT-proBNP, N terminal pro-brain natriuretic peptide; SR, sinus rhythm; VKA, vitamin K antagonists.
GDF-15 levels immediately after anesthesia induction had a better predictive value than those 12 hours after arrival at the ICU (Fig 4). In contrast, preoperative NT-proBNP levels did not make the model more discriminative.
Outcomes A small number of outcomes were recorded during the hospital stay, and there was no significant difference between the 2 groups, except for a trend toward more sepsis events in
Table 3. Factors Associated With Atrial Fibrillation by Cox Multivariate Analysis Univariate Analysis OR (95% CI)
Age EuroSCORE Left atrial diameter 445 mm GDF-15 at induction, mg/mL
1.04 1.21 3.12 0.87
(1-1.08) (1.05-1.39) (1.35-7.21) (0.48-1.58)
Multivariate Analysis p Value
0.05 0.007 0.008 0.64
OR (95% CI)
1.02 1.44 4.18 0.29
(0.97-1.08) (1.21-1.72) (1.75-9.99) (0.13-0.67)
p Value
0.43 o0.001 0.001 0.004
Abbreviation: CI, confidence interval; EuroSCORE, European System for Cardiac Operative Risk Evaluation; GDF-15, growth differentiation factor-15; OR, odds ratio.
1136
BOUCHOT ET AL
Fig 1. Distribution of GDF-15 plasma levels according to the time of blood sampling (A), to the onset of POAF (B), and to the type of surgery (C). Bars represent interquartile ratios (IQRs), and lines mark medians. Whiskers extend from the box up to the smallest/highest observations that lie within 1.5 IQR from the quartiles. Observations that lie further from the quartiles are marked by circles (1.5-3 IQR). Abbreviations: GDF, growth differentiation factor-15; ICU, intensive care unit.
AF patients. No strokes occurred during the in-hospital follow-up. At 1-year follow-up, the characteristics of POAF patients did not differ significantly from those of SR patients, even though there was a trend toward a higher rate of paroxysmal or persistent AF in the former (18% v 7%, p ¼ 0.11). DISCUSSION
The present study showed that in a selected group of 100 patients with no history of AF, POAF incidence within the week after a planned CABG was not associated with the type of cardiac surgery (ONP v OFP), the systemic inflammation status, and preoperative NT-proBNP levels. The incidence of POAF in the ONP-CABG and OFPCABG patients was similar. This result also was found in subgroup analysis according to the AF modalities (silent AF, AF within the first 2 days). Moreover, no difference was found between the 2 types of procedure regarding the duration or the number of AF episodes. After a comprehensive exploration of AF (number of episodes, AF duration, and mean and maximal heart rate in AF and in sinus rhythm) through 7 days of Holter monitoring, the authors suggest that the use of ONP is not associated with higher rates of AF, especially because the baseline and follow-up characteristics of patients in the ONP and OFP groups were identical. The only minor difference observed was a lower number of coronary artery grafts per
procedure in the OFP group (OFP: 2.8 ⫾ 0.8 v ONP: 3.6 ⫾ 0.9, p o 0.001), although this did not influence any of the procedure times. The present results were in accordance with several studies in which both groups were comparable at baseline19–21 and highlight the need to reconsider the roles of inflammation and ischemia in the pathophysiology of POAF. The acute inflammation burst and cardiac ischemia induced by the surgery, and reflected by a 200% mean increase in GDF15 values between these 2 analyses, did not appear to be associated with the onset of POAF. This finding was reinforced by the absence of any association between CRP levels and the development of POAF. A recent review article was in agreement with this finding, thus suggesting that inflammation does not play a major role in the onset of POAF.22 Preoperative NT-proBNP levels were not associated with POAF and did not improve the discrimination value in the multivariate model (Fig 1). In a recent meta-analysis, Cai et al concluded that the predictive value of the postoperative NTproBNP assessment was better than that of the preoperative assessment.23 Moreover, another review article based on 10 studies found that the association between brain natriuretic peptides measured before cardiac surgery and the onset of POAF was modest, with an AUC of 0.61 (95% CI 0.55-0.64) on the ROC curve (1,491 patients analyzed).24 Thus, given the relatively small number of patients included in the present study, the present result was consistent with these reports.
Fig 2. Scatterplots representing the relationship between GDF-15 immediately after the induction of general anesthesia and (A) age, (B) EuroSCORE, and (C) left atrial diameter, in the whole set of patients. Abbreviation: GDF, growth differentiation factor-15.
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
1137
Fig 3. Estimated postoperative atrial fibrillation-free survival rates after ICU admission according to the median baseline GDF-15 value. (A) Unadjusted data. (B) Adjusted for differences in clinical characteristics and concomitant diseases by Cox multivariate hazards regression (age, EuroSCORE, and left atrial diameter 445 mm). P value refers to Cox model. Abbreviations: GDF, growth differentiation factor-15; ICU, intensive care unit.
Conversely, the authors have shown in this study that POAF was associated clearly and independently with a higher EuroSCORE and left atrial dilation and a lower baseline plasma GDF-15 level. Higher EuroSCORE and left atrial dilation previously have been described as predictors of POAF. Helgadottir et al proved that the EuroSCORE was an independent predictor of POAF in a retrospective study based on 744 patients.16 In recent studies, left atrial diameter, LA volume, and strain also were shown to be associated with AF after CABG.25–27 Because high levels of GDF-15 have been described as predictors of cardiovascular events (death, heart failure, myocardial ischemia) in several settings, including cardiac surgery, the present results seem controversial. In fact, results showed that after adjustment for confounding factors of GDF-15 (age, left atrial diameter, EuroSCORE), low plasma levels of GDF15 are associated with a higher rate of AF. However, this finding was closely in accordance with the large amount of data in the literature describing the in vitro and in vivo cardioprotective effects of GDF-15. It has been proven that GDF-15 promotes anti-apoptotic defenses28 via specific SMAD pathways and plays a key role in animal models of heart failure, in which it prevents myocardial remodeling through its antihypertrophic and anti-inflammatory properties.29 Thus, although GDF-15 acts as a cardioprotective cytokine in experimental cellular and animal models, higher levels of GDF-15 may be associated with adverse outcomes in patients. The situation is somewhat reminiscent of BNP, which is produced in the heart in response to ischemia or increases in wall stress. BNP promotes salutary effects in animal models but is indicative of a poor prognosis in patients with acute coronary syndrome and heart failure.30 On the basis of these findings, Wollert hypothesized that GDF-15 also may reflect part of an adaptive response that is overridden by the severity of the underlying disease.31 Thus, though plasma GDF-15 levels correlate strongly with the long-term prognosis in patients, there is no pathophysiologic explanation so far. The long-term prognostic value of GDF-15 may be somewhat different from its potential protective short-term effects in the particular setting of AF after CABG surgery.
Interestingly, its prognostic value appeared only after adjustment for other AF risk factors, such as the EuroSCORE (which includes age, cardiac systolic function, cardiovascular risk factors) and left atrial dimensions. In fact, as GDF-15 levels correlate with age, renal function, and heart failure, the baseline levels were not associated with AF. However, its prognostic value is additional to classic risk factors and levels at anesthesia induction were linked more closely to POAF than were levels in the ICU 12 hours after surgery.
Fig 4. Incremental value of preoperative NT-proBNP and of GDF15 immediately after the induction of general anesthesia or 12 hours after arrival at the ICU to predict AF after CABG. Bar graph illustrates the change in global χ2 value by the addition of preoperative NTproBNP levels or GDF-15 levels to a Cox proportional hazard model that comprised age, the EuroSCORE, and left atrial diameter 445 mm (model 1). The addition of GDF-15 (but not NT-proBNP) significantly improved the global χ2, thereby demonstrating the incremental value of GDF-15 to predict AF after CABG. The best model included GDF-15 at induction. Abbreviations: NT-proBNP, N terminal pro brain natriuretic peptide; GDF, growth differentiation factor-15; ICU, intensive care unit; INC, induction of general anesthesia.
1138
BOUCHOT ET AL
In patients undergoing cardiac surgery, accurate risk assessment is of paramount importance for clinical audits, benchmarking, and research and to identify high-risk patients who may benefit from prophylactic interventions to reduce postoperative adverse outcomes.24 Because GDF-15 has to be adjusted to confounding factors to demonstrate its predictive value, it seems difficult at this point in time to use it as a clinical tool to predict POAF. However, this biomarker could be of interest if added to a predictive preoperative score, in which the confounding factors could be balanced with different weights. Limitations The present study had several limitations. First of all, given the relatively small number of patients (because inclusion criteria were designed to recruit a homogeneous and “otherwise healthy” cohort of patients), the present results should be confirmed in larger population studies. Another limitation must be highlighted: even though the patients undergoing on-pump and off-pump surgeries were comparable, the study was not designed as a randomized controlled trial to compare the incidence of POAF in these 2 groups. Moreover, the
assessment of LA dimensions could have been improved by the measurement of LA surface area and volume. CONCLUSIONS
This study clearly showed that in selected patients with no history of AF, a low plasma level of GDF-15 before cardiac surgery was a strong independent predictor of POAF. Moreover, the preoperative assessment of plasma GDF-15 levels added an incremental predictive value to classical risk factors of POAF. Although GDF-15 is known to be a cardioprotective cytokine in several ischemia and heart failure models, the precise mechanisms involved remain unclear. Experimental studies are warranted to explore the specific interaction between GDF-15 and atrial myocytes, regarding both inflammatory status and electrophysiologic remodeling. ACKNOWLEDGMENTS
The authors thank Frédérique Debomy for technical assistance, Philip Bastable for English revision of the manuscript, and Pr. Christine Binquet for statistical assistance.
REFERENCES 1. Ommen SR, Odell JA, Stanton MS: Atrial arrhythmias after cardiothoracic surgery. N Engl J Med 336:1429-1434, 1997 2. Willems S, Weiss C, Meinertz T: Tachyarrhythmias following coronary artery bypass graft surgery: Epidemiology, mechanisms, and current therapeutic strategies. Thorac Cardiovasc Surg 45: 232-237, 1997 3. Creswell LL, Schuessler RB, Rosenbloom M, et al: Hazards of postoperative atrial arrhythmias. Ann Thorac Surg 56:539-549, 1993 4. Kaireviciute D, Aidietis A, Lip GY: Atrial fibrillation following cardiac surgery: Clinical features and preventative strategies. Eur Heart J 30:410-425, 2009 5. Ishida K, Kimura F, Imamaki M, et al: Relation of inflammatory cytokines to atrial fibrillation after off-pump coronary artery bypass grafting. Eur J Cardiothorac Surg 29:501-505, 2006 6. Bootcov MR, Bauskin AR, Valenzuela SM, et al: MIC-1, a novel macrophage inhibitory cytokine, is a divergent member of the TGFbeta superfamily. Proc Natl Acad Sci USA 94:11514-11519, 1997 7. Kempf T, Wollert KC: Growth-differentiation factor-15 in heart failure. Heart Fail Clin 5:537-547, 2009 8. Heringlake M, Charitos EI, Gatz N, et al: Growth differentiation factor 15: A novel risk marker adjunct to the EuroSCORE for risk stratification in cardiac surgery patients. J Am Coll Cardiol 61: 672-681, 2013 9. Shao Q, Liu H, Ng CY, et al: Circulating serum levels of growth differentiation factor-15 and neuregulin-1 in patients with paroxysmal non-valvular atrial fibrillation. Int J Cardiol 172:e311-e313, 2014 10. Levey AS, Bosch JP, Lewis JB, et al: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130:461-470, 1999 11. Raitt MH, Volgman AS, Zoble RG, et al: Prediction of the recurrence of atrial fibrillation after cardioversion in the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study. Am Heart J 151:390-396, 2006 12. Bouchot O, Berger RL, Berne JP, et al: Clinical experience with a novel thermosensitive temporary coronary artery occluder (LeGoo). Ann Thorac Surg 89:1912-1917, 2010
13. Kahli A, Guenancia C, Zeller M, et al: Growth differentiation factor-15 (GDF-15) levels are associated with cardiac and renal injury in patients undergoing coronary artery bypass grafting with cardiopulmonary bypass. PLoS One 9:e105759, 2014 14. Harrell FE Jr, Lee KL, Califf RM, et al: Regression modelling strategies for improved prognostic prediction. Stat Med 3:143-152, 1984 15. Harrell FE Jr., Lee KL, Mark DB: Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 15:361-387, 1996 16. Helgadottir S, Sigurdsson MI, Ingvarsdottir IL, et al: Atrial fibrillation following cardiac surgery: Risk analysis and long-term survival. J Cardiothorac Surg 7:87, 2012 17. Eggers KM, Kempf T, Wallentin L, et al: Change in growth differentiation factor 15 concentrations over time independently predicts mortality in community-dwelling elderly individuals. Clin Chem 59:1091-1098, 2013 18. Ho JE, Mahajan A, Chen MH, et al: Clinical and genetic correlates of growth differentiation factor 15 in the community. Clin Chem 58:1582-1591, 2012 19. Turk T, Vural H, Eris C, et al: Atrial fibrillation after off-pump coronary artery surgery: A prospective, matched study. J Int Med Res 35:134-142, 2007 20. Archbold RA, Curzen NP: Off-pump coronary artery bypass graft surgery: The incidence of postoperative atrial fibrillation. Heart 89:1134-1137, 2003 21. Møller CH, Penninga L, Wetterslev J, et al: Off-pump versus onpump coronary artery bypass grafting for ischaemic heart disease. Cochrane Database Syst Rev 3: CD007224, 2012 22. Jacob KA, Nathoe HM, Dieleman JM, et al: Inflammation in new-onset atrial fibrillation after cardiac surgery: A systematic review. Eur J Clin Invest 44:402-428, 2014 23. Cai G-L, Chen J, Hu C-B, et al: Value of plasma brain natriuretic peptide levels for predicting postoperative atrial fibrillation: A systemic review and meta-analysis. World J Surg 38:51-59, 2014 24. Litton E, Ho KM: The use of preoperative brain natriuretic peptides as a predictor of adverse outcomes after cardiac surgery: A systematic review and meta-analysis. Eur J Cardiothorac Surg 41:525-534, 2012
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
25. Drossos G, Koutsogiannidis C-P, Ananiadou O, et al: Pericardial fat is strongly associated with atrial fibrillation after coronary artery bypass graft surgery. Eur J Cardiothorac Surg 46:1014-1020, 2014 26. Elawady MA, Bashandy M: Clinical and echocardiographic predicators of postoperative atrial fibrillation. Asian Cardiovasc Thorac Ann 22:655-659, 2013 27. Her A-Y, Kim J-Y, Kim YH, et al: Left atrial strain assessed by speckle tracking imaging is related to new-onset atrial fibrillation after coronary artery bypass grafting. Can J Cardiol 29:377-383, 2013
1139
28. Xu J, Kimball TR, Lorenz JN, et al: GDF15/MIC-1 functions as a protective and antihypertrophic factor released from the myocardium in association with SMAD protein activation. Circ Res 98:342-350, 2006 29. Xu X, Li Z, Gao W: Growth differentiation factor 15 in cardiovascular diseases: From bench to bedside. Biomarkers 16:466-475, 2011 30. de Lemos JA, McGuire DK, Drazner MH: B-type natriuretic peptide in cardiovascular disease. Lancet 362:316-322, 2003 31. Wollert KC: Growth-differentiation factor-15 in cardiovascular disease: From bench to bedside, and back. Basic Res Cardiol 102: 412-415, 2007
assessed after induction of anesthesia and 12 hours after arrival at the intensive care unit. Among the 100 patients, 34 (34%) developed POAF. In Cox multivariate regression analysis, the EuroSCORE, left atrial diameter 445 mm, and low GDF-15 levels at induction were associated independently with the onset of POAF. In contrast, preoperative NT-proBNP levels did not predict POAF. The use of ONP surgery was not associated with a higher incidence of POAF, even though baseline and follow-up characteristics in ONP and OFP patients were identical. Conclusions: In patients with no history of AF, a low plasma level of GDF-15 before CABG surgery was a strong independent predictor of POAF. Moreover, preoperative plasma GDF-15 levels added an incremental predictive value to classic risk factors of POAF. & 2015 Elsevier Inc. All rights reserved.
P
EuroSCORE in 1,458 cardiac surgery patients and an independent predictive factor of postoperative mortality in these patients.8 In a recently published study, the plasma GDF-15 level was associated with nonvalvular, nonpostoperative paroxysmal AF.9 In this observational study, the objective was to assess the role of GDF-15 as a potential new marker of POAF after offpump (OFP) and on-pump (ONP) coronary artery bypass graft (CABG) surgery.
OSTOPERATIVE ATRIAL FIBRILLATION (POAF) has a high prevalence, affecting 20% to 45% of CABG surgery patients within 7 days of the procedure.1,2 POAF increases the duration of hospitalization and healthcare costs, and has been associated with an increased incidence of postoperative stroke, the need for a permanent pacemaker, and early and late mortality.3 Male sex, valvular heart disease, left atrial enlargement, obesity, previous cardiac surgery, chronic lung disease, discontinuation of beta-blockers or ACE inhibitors, AF history, pericarditis, and inflammation are predictive factors for POAF, but advanced age has the strongest and most consistent correlation.3,4 The identification of these risk factors has led to modifications in several intraoperative practices. Because there are many interconnected factors involved in the pathophysiology of POAF, it is difficult to identify the contribution of each to the resulting condition. Therefore, the huge release of inflammation markers triggered by CABG cardiac surgery may not fully explain the occurrence of POAF,5 and there is still a critical need for preoperative clinical predictors. Growth differentiation factor-15 (GDF-15) is a stressresponsive member of the transforming growth factor-β superfamily that originally was cloned as a macrophage-inhibitory cytokine.6 Increased expression of this cytokine can be induced by myocardial stretch, left atrial pressure overload (11), and experimental cardiomyopathy as well as oxidative stress, inflammatory cytokines, and ischemia/reperfusion,7 suggesting that the plasma levels of this cytokine are related not only to myocardial dysfunction but also to circulatory stress. This biomarker thus seems particularly relevant in the setting of cardiac surgery, in which almost all these mechanisms are involved. A recent study has shown that preoperative plasma levels of GDF-15 were useful to improve the risk stratification of the
KEY WORDS: atrial fibrillation, biochemistry, coronary artery bypass grafts, cardiopulmonary bypass, off-pump surgery
MATERIAL AND METHODS
Study Design The independent ethics committee of the University Hospital of Dijon approved the study protocol, and written informed consent was obtained from the patients or their legal representatives. The study was designed as a prospective observational study. All investigations were conducted in accordance with the principles outlined in the Declaration of Helsinki.
From the *Department of Cardiothoracic Surgery, University Hospital, Dijon, France; †Department of Cardiology, University Hospital, Dijon, France; ‡INSERM, U866, LPPCM, Dijon, France; and §CNRS, UMR 5158, Le2I, Dijon, France. Dr. O. Bouchot and Dr. C. Guenancia contributed equally to this work. This work was supported by the University Hospital of Dijon and by grants from Conseil Régional de Bourgogne. Address reprint requests to Charles Guenancia, Service de Cardiologie, CHU Dijon, 14 rue Paul Gaffarel, 21079 Dijon Cedex, France. E-mail: [email protected] © 2015 Elsevier Inc. All rights reserved. 1053-0770/2601-0001$36.00/0 http://dx.doi.org/10.1053/j.jvca.2015.01.023
Journal of Cardiothoracic and Vascular Anesthesia, Vol 29, No 5 (October), 2015: pp 1131–1139
1131
1132
Selection of Patients All of the patients operated on by 2 surgeons for CABG surgery at the University Hospital of Dijon from September 2011 to March 2013 were screened for participation in this prospective, observational study. The exclusion criteria were age o18 years or 480 years, previous atrial fibrillation/flutter, previous treatment with amiodarone, previous cardiac surgery, and emergency surgery. This observational study was designed to recruit the first 50 patients planned for on-pump surgery and the first 50 patients planned for off-pump surgery among patients meeting the inclusion criteria. The surgical strategy (ONP v OFP) was decided by the medical and surgical team according to the anatomy of the coronary arteries. Data Collection Exhaustive clinical data were collected at admission, and the following variables were recorded: age, sex, cardiovascular risk factors, cardiovascular and pulmonary diseases, previous regular medication, and preoperative echocardiographic parameters. Medications administered during the hospital stay (but not the doses) were recorded. Previous renal failure was defined as preoperative estimated glomerular filtration rate (eGFR) o30 mL/min/1.73 m2 using the MDRD formula.10
BOUCHOT ET AL
On-pump surgery was performed according to the following protocol: aortic and cavoatrial cannulation, and cardiopulmonary bypass (CBP) was started. Myocardial protection was induced and maintained by warm blood anterograde cardioplegia. Normal body temperature (361C) was used. Depending on the patients’ grafts and characteristics of the target vessels, the right internal mammary artery or the saphenous vein and/or the left internal mammary artery (LIMA) were used for the grafts. All arterial grafts were treated with papaverine to avoid vasospasm. After completion of the CABG (from 1 to 6 CABGs in the patients studied), CPB was discontinued and protamine was given (1:1) for heparin reversal. Off-pump surgery was performed according to the following protocol: cardiac stabilization and displacement was obtained using Cor vasc OPCAB (Coroneo, Montréal, Canada). Anastomoses were constructed using temporary occlusion, with thermoreversible LeGoo gel (SANOFI, Paris, France).12 The LIMA was investigated primarily and anastomosed to the LAD. The right internal thoracic artery was used pediculised or with the Y graft technique on the LIMA and the distal part was anastomosed to the ramus intermedius. After closure of the sternum, the patients were transferred to the postoperative intensive care unit (ICU) and, finally, to the surgery ward.
Echocardiography All of the patients underwent preoperative transthoracic echocardiography (TTE). The left ventricular ejection fraction (LVEF) was calculated using the Simpson method on the apical four-chamber and apical two-chamber views. LVEF was dichotomized at 45% for more clinical relevance. LA anteroposterior diameter was calculated in the parasternal long-axis view, using the M-mode technique, measuring from the trailing edge of the anterior LA wall to the leading edge of the posterior LA wall. For clinical relevance, LA diameter was dichotomized at 45 mm following classic cut-off values.11 Anesthesia and Heart Surgery Procedure Patients were premedicated with midazolam orally plus hydroxyzine 90 minutes before anesthesia. Routine cardiac medications were continued until the morning of the surgery, except for clopidogrel, which was stopped at least 5 days earlier. Before the induction of anesthesia, a complete hemodynamic monitoring system was set up in the operating room. Anesthesia was induced with intravenous midazolam (0.02 mg/ kg), sufentanil (0.2 to 0.5 mg/kg/h), and propofol (1.5 to 2.5 mg/kg). After verifying that manual ventilation was satisfactory, cisatracurium dibesylate (0.06 mg/kg/h) was injected. Patients were intubated orally and ventilated with FIO2: 0.4. Anesthesia was maintained with sufentanil and cisatracurium as required and inhaled desflurane. Surgical Technique Surgical management was standardized. After median sternotomy and graft harvesting (left and right internal mammary arteries ⫾ saphenous graft), the patient received heparin (1.5 mg/kg in the off-pump group versus 3 mg/kg in the onpump group).
Diagnosis of POAF All patients underwent Holter ECG monitoring (Spider Flash, Sorin Group France) started immediately after inclusion. The Holter ECG device was programmed to record every arrhythmic event for 7 days. The device was consistent with routine use in the ICU, because only 3 electrodes were required, thus providing a 2-lead recording, and was easily removable if necessary for patient care or to perform imaging exams. The Holter ECG monitor was removed after 7 days of recording, or at death (whichever occurred first). An experienced cardiologist who was blinded to the patient’s clinical data analyzed the Holter ECG recordings. If the diagnosis was uncertain, a second cardiologist, blinded to the first results, also analyzed the tracings. No cases of discordance between the 2 analyses occurred. AF was diagnosed according to the guidelines of the European Society of Cardiology for the interpretation of Holter ECG recordings (ie, any arrhythmia that presents the ECG characteristics of AF; namely, “the surface ECG shows ‘absolutely’ irregular RR intervals, there are no distinct P waves on the surface ECG, and the atrial cycle length [when visible], ie, the interval between 2 atrial activations, is usually variable and o200 ms [4300 bpm]”) and lasting at least 30 seconds on a rhythm strip, should be considered AF.9 Silent AF was defined as the occurrence of AF on the Holter ECG recording, regardless of the duration or number of episodes, and in the absence of any mention of AF in the medical file during the first 7 days of the hospital stay (ie, AF not diagnosed by the ICU physicians). In contrast, symptomatic AF was defined as any AF diagnosed by a physician during the hospital stay. When AF was diagnosed on the Holter ECG recording, the following data were collected: date and time of the first
1133
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
episode, ventricular response to the first episode and mean ventricular response to AF, the total burden of AF (total duration of all episodes of AF), and the number of episodes. Nonsustained supraventricular tachycardia (o30 s) also was recorded from the automated count for all patients, and the daily mean ventricular rate was calculated. Heart rate variability was calculated automatically by the Holter software on sinus rhythm strips at day 1 and day 7, as was the mean value for the whole recording. Treatment of New-Onset AF After discharge, each patient’s daily medical file and treatment sheet were analyzed by a physician unaware of the Holter ECG monitoring results, to evaluate the number of AF cases not diagnosed by the clinicians (silent AF) and the treatment choices when AF was identified (rate or rhythm control, medical or electrical cardioversion, anticoagulant therapy). Blood Samples The following measurements were made at admission and every 24 hours during the ICU stay: total blood count, electrolytes, corrected calcemia, creatinine, cardiac troponin Ic, NT-proBNP, CRP, and lactates and blood protein levels. Determination of GDF-15 Arterial blood samples were taken from the arterial catheter immediately after the induction of general anesthesia and 12 hours after arrival at the ICU.13 The blood samples were centrifuged immediately after collection and the plasma was frozen immediately in liquid nitrogen and stored at –801C until analysis. Plasma GDF-15 concentrations were measured by quantitative sandwich enzyme immunoassay (Human GDF-15, Quantikine, R&D Systems Europe, Lille, France) with a linear range from 200 to 50,000 ng/L. The color intensity, relative to GDF-15 concentration, was measured at 450 nm with a spectrophotometer (VictorV3, Perkin Elmer, Courtaboeuf, France). Complete data sets including GDF-15 levels were available from all 100 patients Clinical Follow-Up Lastly, clinical outcomes were recorded over 1-year follow-up, namely: length of stay in the ICU and in the hospital, death in the ICU, at discharge (or at day 28, whichever came first) and at 1 year. The presence of AF and the occurrence of stroke between discharge and the follow-up time points were recorded. Follow-up information at 1 year was obtained by phone call to either the patient or the patient’s family or the general practitioner. One-year follow-up was obtained for all patients.
correlations were computed to assess the relationship between variables. In accordance with the recommendations of Harrell et al on multivariate prognostic modeling,14,15 variables were selected on the basis of their proven clinical, pathophysiologic, and epidemiologic relevance to the endpoint in question. Before the construction of the multivariate models, collinearity between variables was excluded. The correlation matrix of multivariate model showed that the correlation between variables (r-coefficient) did not exceed positive or negative value of 0.25. Because the purpose of this study was to determine whether GDF-15 predicted AF in the study cohort, this variable was included in the analyses. Three more variables, age, left atrial diameter o45 mm, and the EuroSCORE, also were chosen on the basis of their univariate association with POAF in the present study. Given the relatively small size of the population, the EuroSCORE was chosen because it demonstrated a strong univariate association with AF occurrence and included several univariate variables that were significantly different between the POAF and SR groups (previous myocardial infarction, lung disease, LVEF). Because age correlated with high GDF-15 plasma levels, and despite its presence in the EuroSCORE, this variable was included in the multivariate model. This strategy already has been used in the particular setting of POAF in a study in which EuroSCORE and age were shown to be predictive factors of POAF.16 The exploratory identification of factors associated with AF was performed using Cox multivariate descending conditional regression, with inclusion and exclusion cut-offs at 5%. To improve the robustness of results, the POAF patients then were matched 1:1 with the SR patients using nearest neighbor matching on the linear propensity score with a maximum caliper width of 0.2 standard deviations. The same Cox proportional-hazard model was used to compare AF with SR patients with regard to time to the first occurrence of AF. Moreover, to test for the incremental predictive value of NT-proBNP and GDF-15 levels on classic predictors of POAF, a multivariate model that included age, the EuroSCORE and left atrial diameter, based on data in the literature, was built for the whole set of patients (Model 1). These 2 biomarkers then were added to the model and the changes in the global χ2 of the model were established. All multivariate models were tested for multicollinearity and were found to be stable: the tolerance ranged from 0.47 for EuroSCORE to 0.96 for LA 445 mm. The variation inflation factor ranged from 1.04 for LA 445 mm to 2.13 for EuroSCORE. All the tests were two-sided, and a p value o0.05 was considered significant. All analyses were performed using SPSS 20.0 0 (SPSS, Inc., Chicago, IL). RESULTS
Statistical Analysis
Patient Characteristics
Continuous variables are presented as means ⫾ standard deviations when normally distributed or medians and ranges otherwise; and categoric variables as numbers (percentages). Patients were classified into the AF or sinus rhythm (SR) group according to the onset of AF on the Holter tracings. The characteristics of both groups were compared using the exact Mann-Whitney U test or t test for continuous variables and χ2 or Fisher’s exact test for categoric variables. Spearman rank
During the study period, POAF occurred in 34 patients (34%) (AF group); the remaining 66 patients (66%) remained in permanent sinus rhythm (SR group). Factors Associated With POAF Comparing baseline characteristics (Table 1), patients developing POAF were significantly less likely to be smokers (18% v 38%, p ¼ 0.04), and were more likely to have previous
1134
BOUCHOT ET AL
Table 1. Baseline Characteristics According to the Occurrence of Atrial Fibrillation
Characteristics
Risk factors Age, years Female sex Systemic hypertension Diabetes mellitus Smoking Obesity (BMI Z30 kg/m2) Hypercholesterolemia Medical coronary history Recent MI Previous chronic renal failure Previous chronic respiratory failure Previous sleep apnea Previous treatments Beta blockers Diuretics Statins ACE inhibitors VKA Clinical data NYHA41 EuroSCORE LVEF (%) LVEF o45% LA diameter LA diameter 445 mm Serum creatinine, mmol/L NT-proBNP, pg/mL
AF Group
SR Group
(n ¼ 34)
(n ¼ 66)
p Value
66 ⫾ 7 3 (9) 21 (62) 11 (32) 6 (18) 19 (56) 27 (79)
63 ⫾ 9 5 (8) 43 (65) 25 (38) 25 (38) 32 (48) 50 (76)
0.06 1 0.74 0.59 0.04 0.48 0.68
15 (44) 3 (9)
18 (27) 7 (11)
0.09 1
2 (6)
7 (11)
0.71
8 (24)
5 (8)
0.03
31 14 33 28 2
(91) (41) (97) (82) (6)
15 (44) 4 (2-6.3) 54 ⫾ 12 7 (21) 42 (37-45) 7 (21) 88.5 (77-101) 150 (77-1061)
53 16 64 54 3
(80) (24) (97) (82) (5)
0.25 0.08 1 1 1
31 (47) 3 (1-5) 57 ⫾ 11 2 (3) 39.5 (35-43) 3 (5) 88 (74-102.5) 146 (61-491)
0.79 0.07 0.22 0.007 0.08 0.03 0.86 0.33
NOTE. Values expressed as n (%), median (interquartile range), or mean ⫾ SD. Abbreviations: ACE, angiotensin conversion enzyme; AF, atrial fibrillation; BMI, body mass index; EuroSCORE, European System for Cardiac Operative Risk Evaluation; LA, left atrial; LVEF, left ventricular ejection fraction; MI, myocardial infarction; NT-proBNP, N terminal pro brain natriuretic peptide; NYHA, New York Heart Association; SR, sinus rhythm; VKA, vitamin K antagonists.
sleep apnea (24% v 8%, p ¼ 0.03), left atrial dilation (21% v 5%, p ¼ 0.03), and LVEF impairment (21 v 3%, p ¼ 0.007). Although not statistically significant, POAF patients were older, had a higher EuroSCORE, and were more likely to have previous myocardial infarction and to be treated with diuretics. Moreover, POAF patients were comparable to SR patients regarding surgery characteristics, including the proportion of off-pump and on-pump procedures. ICU management differed significantly concerning the need for inotropic support (catecholamine infusion) (27% v 8%, p ¼ 0.02) and the prescription of amiodarone (50% v 20%, p ¼ 0.002). In univariate testing, all biologic variables in the 2 groups, including GDF-15 at induction and in the ICU, were comparable. Tables 2 and 3 Characteristics of POAF Approximately one-third of AF (13 of 34) patients were defined as having silent AF, because they were not diagnosed
by clinicians during the hospital stay. When compared with symptomatic AF, silent AF was associated significantly significantly associated with a lower AF burden (345 ⫾ 774 min v 1,542 ⫾ 1,987 min, p ¼ 0.02) and a longer time between surgery and the first AF episode (3.4 ⫾ 3 days v 1.7 ⫾ 1.1 days, p ¼ 0.03). The first episode of AF occurred within the first 48 hours of monitoring in 23 of the 34 AF patients (67%), and the mean burden of AF was 18 ⫾ 28 hours. Half of the AF patients received an anti-arrhythmic drug during their hospital stay (amiodarone).
Determinants of GDF-15 at Baseline Figure 1 shows the distribution of GDF-15 according to the time of blood sampling, to the onset of POAF, and to the type of surgery. Preoperative circulating plasma levels of GDF-15 correlated significantly with the EuroSCORE (r ¼ 0.54; p o 0.001), with age (r ¼ 0.40; p o 0.001), with preoperative NT-proBNP levels (r ¼ 0.31, p ¼ 0.002), and with creatinine (r ¼ 0.36; p o 0.001). In contrast, no correlation was observed between GDF-15 levels and echocardiographic parameters. Figure 2 shows the relationship scatterplots of GDF-15 immediately after the induction of general anesthesia with age, EuroSCORE, and LA diameter. Regarding categoric baseline variables, GDF-15 levels were significantly higher in patients with diabetes (1.3 ⫾ 0.7 ng/L v 1 ⫾ 0.4 ng/L, p ¼ 0.003), a history of high blood pressure (1.2 ⫾ 0.6 ng/L v 1 ⫾ 0.5 ng/L, p ¼ 0.05), previous renal failure (1.8 ⫾ 0.8 ng/ L v 1 ⫾ 0.5 ng/L, p o 0.001), peripheral artery disease (1.5 ⫾ 0.7 ng/L v 1 ⫾ 0.5 ng/L, p o 0.001), and dyspnea as assessed by NYHA 41 (1.3 ⫾ 0.6 ng/L v 0.9 ⫾ 0.5 ng/L, p ¼ 0.002). The determinants of baseline levels of GDF-15 in the present study were consistent with those in a previous study conducted in the authors’ laboratory, in which associations were found between GDF-15 levels at induction before CABG surgery and age, the EuroSCORE, diabetes, renal function, and NT-proBNP.13 Moreover, in patients free of clinically overt cardiovascular disease, several studies found that GDF-15 levels were linked to age, diabetes, hypertension, impaired renal function, and NT-proBNP levels.17,18
Cox Multivariate Regression After Cox multivariate regression analysis (Table 3), the EuroSCORE, LA diameter 445 mm, and GDF-15 immediately after anesthesia induction were associated independently with the onset of POAF. Figure 3 provides unadjusted (A) and adjusted (B) POAF-free survival curves according to the median baseline GDF-15 level (¼ 1,013 ng/L). After propensity score matching, the EuroSCORE (odds ratio [OR]: 1.29, 95% confidence interval [CI]: 1.07-1.55, p ¼ 0.008), GDF-15 immediately after anesthesia induction (OR: 0.41, 95% CI: 0.17-0.95, p ¼ 0.038) and LA diameter 445 mm (OR: 0.39, 95% CI: 0.16-0.93, p ¼ 0.034) remained associated independently with the onset of POAF. When comparing the incremental prognostic value of GDF15 according to the time of blood sampling, it was found that
1135
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
Table 2. Clinical Characteristics According to the Occurrence of Atrial Fibrillation AF Group (n ¼ 34)
Characteristics
Surgery data Number of CABG 1 2 3 4 5 6 CBP CBP duration (min) Aortic cross-clamp duration (min) ICU management Mechanical ventilation duration (hours) ICU stay duration (hours) Norepinephrine Inotropic catecholamine Amiodarone Biologic data pH 3 hours after surgery pH 24 hours after surgery Troponin 3 hours after surgery, ng/mL Troponin 24 hours after surgery, ng/mL CRP peak, mg/L NT-proBNP peak, pg/mL GDF-15 at induction, ng/L GDF-15 12 hours after arrival to the ICU, ng/L In-hospital follow-up Renal failure Sepsis Stroke Myocardial infarction Treatment at discharge Beta-blockers Statins ACE inhibitors VKA
SR Group (n ¼ 66)
p Value
0.43 1 (3) 10 (29) 12 (35) 8 (24) 2 (6) 1 (3) 15 (44) 104 ⫾ 26 79 ⫾ 22
1 (2) 11 (17) 28 (42) 22 (33) 4 (6) 0 35 (53) 94 ⫾ 28 71 ⫾ 22
0.40 0.24 0.23
6⫾4 68 ⫾ 84 12 (35) 9 (27) 17 (50)
6⫾5 65 ⫾ 39 31 (47) 5 (8) 13 (20)
0.55 0.81 0.26 0.02 0.002
7.37 ⫾ 0.06 7.37 ⫾ 0.04 1.95 (0.49-6.35) 1.9 (1.1-3.4) 203.3 ⫾ 99.6 156 (75-1865) 1,002 (788-1230) 2,544 (1868-3207) 5 (15) 3 (9) 0 0 28 29 21 9
(85) (88) (64) (27)
7.39 ⫾ 0.06 7.37 ⫾ 0.04 3.5 (0.9-5.8) 2.3 (1.7-5) 194.3 ⫾ 85.8 150 (65-531) 1,027 (687-1365) 2,759 (1,861-4,261)
0.12 0.66 0.33 0.19 0.64 0.32 0.74 0.57
14 (21) 1 (2) 0 3 (5)
0.59 0.11 — 0.55
60 (91) 36 (96) 41 (62) 0
0.49 0.22 0.88 o0.001
NOTE. Values expressed as n (%), median (interquartile range), or mean ⫾ SD. Abbreviations: ACE, angiotensin converting enzyme; AF, atrial fibrillation; CABG, coronary artery bypass graft; CBP, cardiopulmonary bypass; CRP, C-reactive protein; GDF-15, growth differentiation factor-15; ICU, intensive care unit; NT-proBNP, N terminal pro-brain natriuretic peptide; SR, sinus rhythm; VKA, vitamin K antagonists.
GDF-15 levels immediately after anesthesia induction had a better predictive value than those 12 hours after arrival at the ICU (Fig 4). In contrast, preoperative NT-proBNP levels did not make the model more discriminative.
Outcomes A small number of outcomes were recorded during the hospital stay, and there was no significant difference between the 2 groups, except for a trend toward more sepsis events in
Table 3. Factors Associated With Atrial Fibrillation by Cox Multivariate Analysis Univariate Analysis OR (95% CI)
Age EuroSCORE Left atrial diameter 445 mm GDF-15 at induction, mg/mL
1.04 1.21 3.12 0.87
(1-1.08) (1.05-1.39) (1.35-7.21) (0.48-1.58)
Multivariate Analysis p Value
0.05 0.007 0.008 0.64
OR (95% CI)
1.02 1.44 4.18 0.29
(0.97-1.08) (1.21-1.72) (1.75-9.99) (0.13-0.67)
p Value
0.43 o0.001 0.001 0.004
Abbreviation: CI, confidence interval; EuroSCORE, European System for Cardiac Operative Risk Evaluation; GDF-15, growth differentiation factor-15; OR, odds ratio.
1136
BOUCHOT ET AL
Fig 1. Distribution of GDF-15 plasma levels according to the time of blood sampling (A), to the onset of POAF (B), and to the type of surgery (C). Bars represent interquartile ratios (IQRs), and lines mark medians. Whiskers extend from the box up to the smallest/highest observations that lie within 1.5 IQR from the quartiles. Observations that lie further from the quartiles are marked by circles (1.5-3 IQR). Abbreviations: GDF, growth differentiation factor-15; ICU, intensive care unit.
AF patients. No strokes occurred during the in-hospital follow-up. At 1-year follow-up, the characteristics of POAF patients did not differ significantly from those of SR patients, even though there was a trend toward a higher rate of paroxysmal or persistent AF in the former (18% v 7%, p ¼ 0.11). DISCUSSION
The present study showed that in a selected group of 100 patients with no history of AF, POAF incidence within the week after a planned CABG was not associated with the type of cardiac surgery (ONP v OFP), the systemic inflammation status, and preoperative NT-proBNP levels. The incidence of POAF in the ONP-CABG and OFPCABG patients was similar. This result also was found in subgroup analysis according to the AF modalities (silent AF, AF within the first 2 days). Moreover, no difference was found between the 2 types of procedure regarding the duration or the number of AF episodes. After a comprehensive exploration of AF (number of episodes, AF duration, and mean and maximal heart rate in AF and in sinus rhythm) through 7 days of Holter monitoring, the authors suggest that the use of ONP is not associated with higher rates of AF, especially because the baseline and follow-up characteristics of patients in the ONP and OFP groups were identical. The only minor difference observed was a lower number of coronary artery grafts per
procedure in the OFP group (OFP: 2.8 ⫾ 0.8 v ONP: 3.6 ⫾ 0.9, p o 0.001), although this did not influence any of the procedure times. The present results were in accordance with several studies in which both groups were comparable at baseline19–21 and highlight the need to reconsider the roles of inflammation and ischemia in the pathophysiology of POAF. The acute inflammation burst and cardiac ischemia induced by the surgery, and reflected by a 200% mean increase in GDF15 values between these 2 analyses, did not appear to be associated with the onset of POAF. This finding was reinforced by the absence of any association between CRP levels and the development of POAF. A recent review article was in agreement with this finding, thus suggesting that inflammation does not play a major role in the onset of POAF.22 Preoperative NT-proBNP levels were not associated with POAF and did not improve the discrimination value in the multivariate model (Fig 1). In a recent meta-analysis, Cai et al concluded that the predictive value of the postoperative NTproBNP assessment was better than that of the preoperative assessment.23 Moreover, another review article based on 10 studies found that the association between brain natriuretic peptides measured before cardiac surgery and the onset of POAF was modest, with an AUC of 0.61 (95% CI 0.55-0.64) on the ROC curve (1,491 patients analyzed).24 Thus, given the relatively small number of patients included in the present study, the present result was consistent with these reports.
Fig 2. Scatterplots representing the relationship between GDF-15 immediately after the induction of general anesthesia and (A) age, (B) EuroSCORE, and (C) left atrial diameter, in the whole set of patients. Abbreviation: GDF, growth differentiation factor-15.
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
1137
Fig 3. Estimated postoperative atrial fibrillation-free survival rates after ICU admission according to the median baseline GDF-15 value. (A) Unadjusted data. (B) Adjusted for differences in clinical characteristics and concomitant diseases by Cox multivariate hazards regression (age, EuroSCORE, and left atrial diameter 445 mm). P value refers to Cox model. Abbreviations: GDF, growth differentiation factor-15; ICU, intensive care unit.
Conversely, the authors have shown in this study that POAF was associated clearly and independently with a higher EuroSCORE and left atrial dilation and a lower baseline plasma GDF-15 level. Higher EuroSCORE and left atrial dilation previously have been described as predictors of POAF. Helgadottir et al proved that the EuroSCORE was an independent predictor of POAF in a retrospective study based on 744 patients.16 In recent studies, left atrial diameter, LA volume, and strain also were shown to be associated with AF after CABG.25–27 Because high levels of GDF-15 have been described as predictors of cardiovascular events (death, heart failure, myocardial ischemia) in several settings, including cardiac surgery, the present results seem controversial. In fact, results showed that after adjustment for confounding factors of GDF-15 (age, left atrial diameter, EuroSCORE), low plasma levels of GDF15 are associated with a higher rate of AF. However, this finding was closely in accordance with the large amount of data in the literature describing the in vitro and in vivo cardioprotective effects of GDF-15. It has been proven that GDF-15 promotes anti-apoptotic defenses28 via specific SMAD pathways and plays a key role in animal models of heart failure, in which it prevents myocardial remodeling through its antihypertrophic and anti-inflammatory properties.29 Thus, although GDF-15 acts as a cardioprotective cytokine in experimental cellular and animal models, higher levels of GDF-15 may be associated with adverse outcomes in patients. The situation is somewhat reminiscent of BNP, which is produced in the heart in response to ischemia or increases in wall stress. BNP promotes salutary effects in animal models but is indicative of a poor prognosis in patients with acute coronary syndrome and heart failure.30 On the basis of these findings, Wollert hypothesized that GDF-15 also may reflect part of an adaptive response that is overridden by the severity of the underlying disease.31 Thus, though plasma GDF-15 levels correlate strongly with the long-term prognosis in patients, there is no pathophysiologic explanation so far. The long-term prognostic value of GDF-15 may be somewhat different from its potential protective short-term effects in the particular setting of AF after CABG surgery.
Interestingly, its prognostic value appeared only after adjustment for other AF risk factors, such as the EuroSCORE (which includes age, cardiac systolic function, cardiovascular risk factors) and left atrial dimensions. In fact, as GDF-15 levels correlate with age, renal function, and heart failure, the baseline levels were not associated with AF. However, its prognostic value is additional to classic risk factors and levels at anesthesia induction were linked more closely to POAF than were levels in the ICU 12 hours after surgery.
Fig 4. Incremental value of preoperative NT-proBNP and of GDF15 immediately after the induction of general anesthesia or 12 hours after arrival at the ICU to predict AF after CABG. Bar graph illustrates the change in global χ2 value by the addition of preoperative NTproBNP levels or GDF-15 levels to a Cox proportional hazard model that comprised age, the EuroSCORE, and left atrial diameter 445 mm (model 1). The addition of GDF-15 (but not NT-proBNP) significantly improved the global χ2, thereby demonstrating the incremental value of GDF-15 to predict AF after CABG. The best model included GDF-15 at induction. Abbreviations: NT-proBNP, N terminal pro brain natriuretic peptide; GDF, growth differentiation factor-15; ICU, intensive care unit; INC, induction of general anesthesia.
1138
BOUCHOT ET AL
In patients undergoing cardiac surgery, accurate risk assessment is of paramount importance for clinical audits, benchmarking, and research and to identify high-risk patients who may benefit from prophylactic interventions to reduce postoperative adverse outcomes.24 Because GDF-15 has to be adjusted to confounding factors to demonstrate its predictive value, it seems difficult at this point in time to use it as a clinical tool to predict POAF. However, this biomarker could be of interest if added to a predictive preoperative score, in which the confounding factors could be balanced with different weights. Limitations The present study had several limitations. First of all, given the relatively small number of patients (because inclusion criteria were designed to recruit a homogeneous and “otherwise healthy” cohort of patients), the present results should be confirmed in larger population studies. Another limitation must be highlighted: even though the patients undergoing on-pump and off-pump surgeries were comparable, the study was not designed as a randomized controlled trial to compare the incidence of POAF in these 2 groups. Moreover, the
assessment of LA dimensions could have been improved by the measurement of LA surface area and volume. CONCLUSIONS
This study clearly showed that in selected patients with no history of AF, a low plasma level of GDF-15 before cardiac surgery was a strong independent predictor of POAF. Moreover, the preoperative assessment of plasma GDF-15 levels added an incremental predictive value to classical risk factors of POAF. Although GDF-15 is known to be a cardioprotective cytokine in several ischemia and heart failure models, the precise mechanisms involved remain unclear. Experimental studies are warranted to explore the specific interaction between GDF-15 and atrial myocytes, regarding both inflammatory status and electrophysiologic remodeling. ACKNOWLEDGMENTS
The authors thank Frédérique Debomy for technical assistance, Philip Bastable for English revision of the manuscript, and Pr. Christine Binquet for statistical assistance.
REFERENCES 1. Ommen SR, Odell JA, Stanton MS: Atrial arrhythmias after cardiothoracic surgery. N Engl J Med 336:1429-1434, 1997 2. Willems S, Weiss C, Meinertz T: Tachyarrhythmias following coronary artery bypass graft surgery: Epidemiology, mechanisms, and current therapeutic strategies. Thorac Cardiovasc Surg 45: 232-237, 1997 3. Creswell LL, Schuessler RB, Rosenbloom M, et al: Hazards of postoperative atrial arrhythmias. Ann Thorac Surg 56:539-549, 1993 4. Kaireviciute D, Aidietis A, Lip GY: Atrial fibrillation following cardiac surgery: Clinical features and preventative strategies. Eur Heart J 30:410-425, 2009 5. Ishida K, Kimura F, Imamaki M, et al: Relation of inflammatory cytokines to atrial fibrillation after off-pump coronary artery bypass grafting. Eur J Cardiothorac Surg 29:501-505, 2006 6. Bootcov MR, Bauskin AR, Valenzuela SM, et al: MIC-1, a novel macrophage inhibitory cytokine, is a divergent member of the TGFbeta superfamily. Proc Natl Acad Sci USA 94:11514-11519, 1997 7. Kempf T, Wollert KC: Growth-differentiation factor-15 in heart failure. Heart Fail Clin 5:537-547, 2009 8. Heringlake M, Charitos EI, Gatz N, et al: Growth differentiation factor 15: A novel risk marker adjunct to the EuroSCORE for risk stratification in cardiac surgery patients. J Am Coll Cardiol 61: 672-681, 2013 9. Shao Q, Liu H, Ng CY, et al: Circulating serum levels of growth differentiation factor-15 and neuregulin-1 in patients with paroxysmal non-valvular atrial fibrillation. Int J Cardiol 172:e311-e313, 2014 10. Levey AS, Bosch JP, Lewis JB, et al: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130:461-470, 1999 11. Raitt MH, Volgman AS, Zoble RG, et al: Prediction of the recurrence of atrial fibrillation after cardioversion in the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study. Am Heart J 151:390-396, 2006 12. Bouchot O, Berger RL, Berne JP, et al: Clinical experience with a novel thermosensitive temporary coronary artery occluder (LeGoo). Ann Thorac Surg 89:1912-1917, 2010
13. Kahli A, Guenancia C, Zeller M, et al: Growth differentiation factor-15 (GDF-15) levels are associated with cardiac and renal injury in patients undergoing coronary artery bypass grafting with cardiopulmonary bypass. PLoS One 9:e105759, 2014 14. Harrell FE Jr, Lee KL, Califf RM, et al: Regression modelling strategies for improved prognostic prediction. Stat Med 3:143-152, 1984 15. Harrell FE Jr., Lee KL, Mark DB: Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 15:361-387, 1996 16. Helgadottir S, Sigurdsson MI, Ingvarsdottir IL, et al: Atrial fibrillation following cardiac surgery: Risk analysis and long-term survival. J Cardiothorac Surg 7:87, 2012 17. Eggers KM, Kempf T, Wallentin L, et al: Change in growth differentiation factor 15 concentrations over time independently predicts mortality in community-dwelling elderly individuals. Clin Chem 59:1091-1098, 2013 18. Ho JE, Mahajan A, Chen MH, et al: Clinical and genetic correlates of growth differentiation factor 15 in the community. Clin Chem 58:1582-1591, 2012 19. Turk T, Vural H, Eris C, et al: Atrial fibrillation after off-pump coronary artery surgery: A prospective, matched study. J Int Med Res 35:134-142, 2007 20. Archbold RA, Curzen NP: Off-pump coronary artery bypass graft surgery: The incidence of postoperative atrial fibrillation. Heart 89:1134-1137, 2003 21. Møller CH, Penninga L, Wetterslev J, et al: Off-pump versus onpump coronary artery bypass grafting for ischaemic heart disease. Cochrane Database Syst Rev 3: CD007224, 2012 22. Jacob KA, Nathoe HM, Dieleman JM, et al: Inflammation in new-onset atrial fibrillation after cardiac surgery: A systematic review. Eur J Clin Invest 44:402-428, 2014 23. Cai G-L, Chen J, Hu C-B, et al: Value of plasma brain natriuretic peptide levels for predicting postoperative atrial fibrillation: A systemic review and meta-analysis. World J Surg 38:51-59, 2014 24. Litton E, Ho KM: The use of preoperative brain natriuretic peptides as a predictor of adverse outcomes after cardiac surgery: A systematic review and meta-analysis. Eur J Cardiothorac Surg 41:525-534, 2012
GDF-15 AND POST-CABG ATRIAL FIBRILLATION
25. Drossos G, Koutsogiannidis C-P, Ananiadou O, et al: Pericardial fat is strongly associated with atrial fibrillation after coronary artery bypass graft surgery. Eur J Cardiothorac Surg 46:1014-1020, 2014 26. Elawady MA, Bashandy M: Clinical and echocardiographic predicators of postoperative atrial fibrillation. Asian Cardiovasc Thorac Ann 22:655-659, 2013 27. Her A-Y, Kim J-Y, Kim YH, et al: Left atrial strain assessed by speckle tracking imaging is related to new-onset atrial fibrillation after coronary artery bypass grafting. Can J Cardiol 29:377-383, 2013
1139
28. Xu J, Kimball TR, Lorenz JN, et al: GDF15/MIC-1 functions as a protective and antihypertrophic factor released from the myocardium in association with SMAD protein activation. Circ Res 98:342-350, 2006 29. Xu X, Li Z, Gao W: Growth differentiation factor 15 in cardiovascular diseases: From bench to bedside. Biomarkers 16:466-475, 2011 30. de Lemos JA, McGuire DK, Drazner MH: B-type natriuretic peptide in cardiovascular disease. Lancet 362:316-322, 2003 31. Wollert KC: Growth-differentiation factor-15 in cardiovascular disease: From bench to bedside, and back. Basic Res Cardiol 102: 412-415, 2007
