==== Front JACC Asia JACC Asia JACC Asia 2772-3747 Elsevier S2772-3747(23)00090-X 10.1016/j.jacasi.2023.03.008 Original Research Growth Differentiation Factor-15 and Clinical Outcomes in Japanese Patients With Ischemic Heart Disease Kobayashi Yuta MD Otaki Yoichiro MD, PhD y-otaki@med.id.yamagata-u.ac.jp ∗ Watanabe Tetsu MD, PhD Tachibana Shingo MD Sato Junya MD Saito Yuji MD, PhD Aono Tomonori MD, PhD Goto Jun MD, PhD Kato Shigehiko MD, PhD Tamura Harutoshi MD, PhD Nishiyama Satoshi MD, PhD Arimoto Takanori MD, PhD Takahashi Hiroki MD, PhD Watanabe Masafumi MD, PhD Department of Cardiology, Pulmonology, and Nephrology, Yamagata University School of Medicine, Yamagata, Japan ∗ Address for correspondence: Dr Yoichiro Otaki, Department of Cardiology, Pulmonology and Nephrology, Yamagata University School of Medicine, 2-2-2 Iida-Nishi, Yamagata 990-9585, Japan. y-otaki@med.id.yamagata-u.ac.jp 09 5 2023 6 2023 09 5 2023 3 3 457471 8 11 2022 6 3 2023 6 3 2023 © 2023 The Authors 2023 https://creativecommons.org/licenses/by/4.0/ This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Background Despite a reduction in the rate of thrombotic events, ischemic heart disease (IHD) remains a key medical problem associated with high major bleeding and mortality in Asian patients with IHD. Growth differentiation factor (GDF)-15, a stress-response cytokine belonging to the transforming growth factor beta superfamily, is reportedly associated with poor clinical outcomes in Western patients with IHD. However, the clinical significance of GDF-15 in Asian patients with IHD has not yet been fully elucidated. Objectives The aim of the present study was to examine the impact of serum GDF-15 on clinical outcomes in Japanese patients with IHD. Methods Serum GDF-15 levels were evaluated in 632 consecutive patients with IHD. All patients were followed up for a median period of 2.8 years. The primary endpoint was the all-cause mortality rate. Secondary endpoints were major adverse cardiovascular events (MACE), heart failure (HF)–related rehospitalization, bleeding, and thrombotic events. Results Serum GDF-15 levels were elevated in acute coronary syndrome, severe coronary artery disease, and the major Japanese version of the high bleeding risk criteria. Multivariate Cox proportional hazards regression analysis demonstrated that GDF-15 was an independent predictor of all-cause mortality, MACE, HF-related rehospitalizations, and bleeding events after adjusting for confounding risk factors but not for thrombotic events. Adding GDF-15 to risk factors significantly improved the net reclassification index and integrated discrimination improvement for all-cause deaths, MACE, HF-related rehospitalization, and bleeding events. Conclusions Serum GDF-15 could be a feasible marker for major bleeding and adverse clinical outcomes in Japanese patients with IHD. Central Illustration Key Words clinical outcomes GDF-15 high bleeding risk ischemic heart disease Abbreviations and Acronyms ACS acute coronary syndrome(s) AF atrial fibrillation BNP brain natriuretic peptide CCS chronic coronary syndrome CKD chronic kidney disease GDF growth differentiation factor HBR high bleeding risk HF heart failure hsCRP high-sensitivity C-reactive protein IDI integrated discrimination improvement IHD ischemic heart disease J-HBR Japanese version of high bleeding risk MACE major adverse cardiovascular event(s) NRI net reclassification index PCI percutaneous coronary intervention PVD peripheral vascular disease TnT troponin T ==== Body pmcIschemic heart disease (IHD) is a leading cause of death worldwide. Because of advances in revascularization and antithrombotic medicine, the thrombotic event rate has decreased significantly among patients with IHD who undergo revascularization.1 However, mortality in patients with IHD who undergo revascularization remains high. Therefore, early identification and risk stratification of patients at high risk for all-cause death are of critical importance. Bleeding events in patients with IHD are associated with high mortality risk.2,3 Dual antiplatelet therapy is the standard of care for patients with IHD who undergo revascularization, which may increase the tendency for bleeding.4 Furthermore, Asian patients with IHD are characterized by high bleeding risk (HBR) compared with Caucasian patients with IHD.5 Therefore, an assessment of bleeding risk using the Japanese version of HBR (J-HBR) is recommended to determine the intensity and duration of antiplatelet and anticoagulation therapy according to the Japanese Circulation Society 2020 guidelines focused on antithrombotic therapy in patients with coronary artery disease.6 Some components of J-HBR, such as chronic kidney disease (CKD), anemia, heart failure (HF), atrial fibrillation (AF) and/or oral anticoagulation, and peripheral vascular disease (PVD), were also risk factors for thrombotic events. Therefore, a biomarker is required to discriminate between thrombotic and bleeding risks and to predict subsequent all-cause mortality. Growth differentiation factor (GDF)-15 is a member of the transforming growth factor-beta superfamily. Although GDF-15 is appreciably expressed in the liver and placenta at baseline, circulating GDF-15 levels are generally low in healthy individuals. GDF-15 is induced by several factors, such as oxidative stress, inflammation, and mechanical stress in various tissues, and is released into circulation.7,8 As circulating GDF-15 is reportedly elevated in advanced age, cancer, and cardiovascular disease,9 it is considered a stress-response cytokine. Accumulating evidence has demonstrated that GDF-15 can help predict mortality, major adverse cardiovascular events (MACE), and HF hospitalizations in patients with IHD.10, 11, 12, 13 Moreover, GDF-15 is associated with major bleeding in patients with AF across different geographic areas, including Asia.14,15 GDF-15 measurement is used clinically for risk stratification of clinical outcomes in patients with acute coronary syndrome (ACS) and chronic HF and assessment of bleeding risk in patients with AF in Western countries. However, the clinical usefulness of serum GDF-15 levels in Asian populations has not yet been fully elucidated. Notably, it remains undetermined whether GDF-15 could identify Asian patients with IHD at high risk for bleeding events. Considering the HBR in Asian patients, we focused on an association of serum GDF-15 with bleeding events and all-cause death in Asian patients with IHD. The aims of the present study were to: 1) examine the association between serum GDF-15 levels and coronary artery disease severity and J-HBR; and 2) examine the impact of serum GDF-15 on clinical outcomes such as mortality, MACE, HF-related rehospitalization, bleeding events, and thrombotic events in Japanese patients with IHD. Methods Ethics statement All procedures involving participants were undertaken in accordance with the ethical, institutional, and/or national research committee guidelines of the centers at which the studies were conducted (Yamagata University, 2021-298), and all procedures complied with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The study was approved by the institutional ethics committee of Yamagata University School of Medicine. Study subjects The study subjects were 632 patients with IHD who were admitted to Yamagata University Hospital for treatment between 2012 and 2018. The diagnoses of ACS and chronic coronary syndrome (CCS) were based on fourth universal definition of myocardial infarction and 2019 European Society of Cardiology guidelines, respectively.16,17 Invasive treatments, such as percutaneous coronary intervention (PCI) and/or coronary artery bypass graft surgery, were determined by expert cardiologists and cardiac surgeons according to the guidelines.18,19 Clinical data such as age, sex, hypertension, diabetes mellitus, dyslipidemia, smoking status, medical history, and medication use at discharge were obtained from the patients’ medical records and interviews. Serum GDF-15 level measurement and ABC-AF-bleeding score calculation Blood samples were collected in the early morning within 24 hours of hospital admission. The samples were centrifuged at 3,000 rpm for 15 minutes at 4 °C, and the obtained serum was stored at −80 °C until analysis. The GDF-15 level assay was performed using an electrochemiluminescent immunoassay (Roche Diagnostics). Brain natriuretic peptide (BNP) levels were evaluated using a chemiluminescent enzyme immunoassay. Troponin T (TnT) and high sensitivity C-reactive protein (hsCRP) levels were evaluated using the electrochemiluminescent immunoassay and latex agglutination methods, respectively. Hemoglobin levels were measured simultaneously. The ABC-AF-bleeding score is composed of age, TnT, hemoglobin, GDF-15, and previous bleeding.15 The ABC-AF-bleeding score was evaluated using an online calculator. Cardiovascular risk factors Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or antihypertensive medication use. Diabetes mellitus was defined as a fasting blood glucose level ≥126 mg/dL, glycated hemoglobin ≥6.5%, or antidiabetic medication use. Dyslipidemia was defined as high-density lipoprotein cholesterol <40 mg/dL, low-density lipoprotein cholesterol ≥140 mg/dL, triglyceride ≥150 mg/dL, or use of lipid-lowering medication. J-HBR Bleeding risk was assessed using the J-HBR criteria according to the Japanese Circulation Society 2020 guideline recommendations.6 The primary criteria for J-HBR include low body weight (<55 kg in men, <50 kg in women), severe CKD (estimated glomerular filtration rate <30 mL/min/1.73 m2 or hemodialysis), moderate to severe anemia (hemoglobin <11 g/dL), HF, anticipated use of long-term oral anticoagulation, PVD, history of nontraumatic bleeding events, history of ischemic stroke or intracranial hemorrhage, thrombocytopenia (platelet count <100 × 109/L), active malignancy, liver cirrhosis with portal hypertension, chronic bleeding diatheses, nondeferrable major surgery on dual antiplatelet therapy, and recent major surgery or major trauma within 30 days before PCI. Endpoints and follow-up period All patients were prospectively followed up by telephone or by reviewing medical records twice a year for a median period of 2.8 years (IQR: 0.9-5.0 years). The primary endpoint was the all-cause mortality rate. The secondary endpoints were MACE, HF-related rehospitalization, bleeding, and thrombotic events. MACE includes ACS, stroke, cerebral hemorrhage, HF-related rehospitalization, fatal arrhythmias, and cardiac death, defined as death due to progressive ACS, HF, or sudden cardiac death. HF-related rehospitalization was defined as hospitalization for HF progression. Bleeding events were defined as Bleeding Academic Research Consortium type 3 or 5. Thrombotic events were defined as ACS, stroke, and peripheral vascular thrombosis. Statistical analysis The normality of continuous variables was confirmed using the Shapiro-Wilk test. Results are expressed as mean ± SD for continuous variables and as number (percentage) for categorical variables. Skewed values are presented as median (IQR). Student’s t-test and the chi-square test were used to compare continuous and categorical variables, respectively. Differences in age, hemoglobin, and estimated glomerular filtration rate among the groups were analyzed using analysis of variance with Tukey’s post hoc test. Differences in TnT, BNP, hsCRP, and ABC-AF-bleeding score among groups were analyzed using the Steel-Dwass test.20 Survival curves were plotted using the Kaplan-Meier method and compared using the log-rank test. Because of the skewed distributions of TnT, BNP, hsCRP, and GDF-15 values, we checked supreme tests for linear form. Thus, we used log-transformed TnT, BNP, hsCRP, and GDF-15 values for the Cox proportional hazards regression analysis. Cox proportional hazards analysis was used to determine independent predictors of the primary and secondary endpoints. Multicollinearity was assessed using the variance inflation factor. Significant predictors selected in the univariate Cox proportional hazards regression analysis for each clinical outcome were entered into the multivariate analysis for all-cause mortality, MACE, HF-related rehospitalization, and thrombotic events. The proportionality assumption in the Cox model was assessed using the Schoenfeld residuals test. As BNP was used for the diagnosis of HF in the present study, there was a close relationship between BNP and HF. BNP is also a prognostic marker for IHD, so we entered BNP into multivariate analysis, but not HF. The major components of the J-HBR criteria were entered into the multivariate analysis for bleeding events. Thus, we entered HF into the multivariate analysis for bleeding events, but not BNP. As the definition of anemia differs by sex, we used hemoglobin in the Cox proportional hazards model. Receiver-operating characteristic curves for all-cause mortality, MACE, HF-related rehospitalization, and bleeding events were plotted and used to measure the predictive accuracy of serum GDF-15 level for these clinical outcomes. We calculated the net reclassification index (NRI) and integrated discrimination improvement (IDI) to measure the incremental value of serum GDF-15 in classifying patients at high or low risk for all-cause mortality, MACE, HF-related rehospitalizations, and bleeding events. Statistical significance was set at a P value of <0.05. All statistical analyses were performed using JMP version 14.2 (SAS Institute) and EZR (Saitama Medical Center, Jichi Medical University). Results Baseline characteristics of all patients with IHD and comparisons of clinical characteristics among the GDF-15 tertiles The baseline characteristics of the 632 patients are summarized in Table 1. There were 484 patients (76.6%) with ACS and 148 (23.4%) with CCS. Significant stenosis in left main trunk lesions and 3-vessel diseases were observed in 53 (8.5%) and 147 (23.6%) patients, respectively. Hypertension, diabetes mellitus, dyslipidemia, and current smoking were identified in 480 (76%), 246 (39%), 371 (58.7%), and 400 (63.3%) patients, respectively. The median serum GDF-15 level was 1,991 pg/mL. PCI and coronary artery bypass graft surgery were performed in 565 (89.4%) and 41 (6.5%) patients with IHD, respectively, indicating that up to 94.9% of the patients underwent revascularization in this study. Major criteria for J-HBR, such as low body weight, severe CKD, moderate to severe anemia, HF, anticipated use of long-term oral anticoagulation, and PVD, were identified in 156 (24.7%), 97 (15.3%), 78 (12.3%), 160 (25.3%), 97 (15.3%), and 68 (10.8%) patients, respectively.Table 1 Baseline Characteristics and Comparisons of Clinical Characteristics Among 3 Groups on the Basis of GDF-15 Level All Subjects (N = 632) First Tertile (<1,509 pg/mL) (n = 211) Second Tertile (1,509-2,604 pg/mL) (n = 211) Third Tertile (>2,604 pg/mL) (n = 210) P Value Age, y 69.5 ± 11.5 61.8 ± 10.4 71.0 ± 10.0a 75.6 ± 9.5a,b <0.0001 Men 479 (75.8) 170 (80.6) 159 (75.4) 150 (71.4) 0.0896 ACS/CCS 484/148 154/57 149/62 181/29 0.0003 LMT lesion 53 (8.5) 11 (5.2) 15 (7.2) 27 (13.3) 0.0228 3-vessel disease 147 (23.6) 33 (15.6) 49 (23.4) 65 (32.0) 0.0005 Hypertension 480 (76.0) 137 (64.9) 170 (80.6) 173 (82.4) <0.0001 Diabetes mellitus 246 (39.0) 60 (28.0) 91 (43.3) 95 (45.2) 0.0006 Dyslipidemia 371 (58.7) 131 (62.1) 121 (57.4) 119 (56.7) 0.4688 Smoking 400 (63.3) 148 (70.1) 127 (60.2) 125 (59.5) 0.0404 Prior myocardial infarction 66 (10.4) 19 (9.0) 28 (13.3) 19 (9.1) 0.2583 Prior revascularization 111 (17.6) 38 (18.0) 43 (20.4) 30 (14.3) 0.2538 Prior atrial fibrillation 70 (11.1) 13 (6.2) 22 (10.4) 35 (16.7) 0.0026 Prior heart failure 35 (5.5) 4 (1.9) 12 (5.7) 19 (9.1) 0.0058 Prior bleeding 24 (3.8) 5 (2.4) 7 (3.3) 12 (5.7) 0.1806 Prior stroke 75 (11.9) 12 (5.7) 29 (13.7) 34 (16.2) 0.0023 Biochemical data  eGFR, mL/min/1.73 m2 69.3 ± 25.1 80.7 ± 19.4 71.9 ± 19.1a 55.2 ± 28.5a,b <0.0001  Hemoglobin, g/dL 13.3 ± 2.1 14.1 ± 1.7 13.5 ± 1.8a 12.5 ± 2.3a,b <0.0001  Troponin T, pg/mL 1.24 (0.10-4.41) 0.69 (0.04-3.19) 0.72 (0.03-3.11) 2.61 (0.64-5.71)c,d <0.0001  BNP, pg/mL 62.8 (23.1-205.7) 29.5 (16.1-71.2) 52.7 (22.0-170.7)c 166.8 (63.8-481.6)c,d <0.0001  hsCRP, μg/mL 0.59 (0.12-2.67) 0.27 (0.07-1.58) 0.35 (0.08-2.07) 1.42 (0.45-5.14)c,d <0.0001  GDF-15, pg/mL 1,991 (1,289-3,060)  ABC-AF-bleeding score 4.96 (2.74-8.17) 2.89 (1.33-4.01) 5.01 (2.95-6.36)c 9.96 (6.77-13.57)c,d <0.0001 Medications  ACE inhibitors and/or ARBs 516 (81.6) 169 (80.1) 176 (83.4) 171 (81.4) 0.6740  β-blockers 378 (59.8) 123 (58.3) 121 (57.4) 134 (63.8) 0.3421  MRAs 53 (8.4) 9 (4.3) 11 (5.2) 33 (15.7) <0.0001  Statins 553 (87.8) 198 (94.3) 191 (91.0) 164 (78.1) <0.0001  Aspirin 596 (94.3) 204 (96.7) 201 (95.3) 191 (91.0) 0.0306  P2Y12 receptor antagonists 540 (85.4) 192 (90.5) 181 (85.8) 167 (79.5) 0.0038 Revascularization 600 (94.9) 205 (97.2) 202 (95.7) 193 (91.9) 0.0396  PCI 565 (89.4) 198 (93.8) 191 (90.5) 176 (83.8) 0.0030  CABG 41 (6.5) 9 (4.3) 14 (6.6) 18 (8.6) 0.1900 Japanese version of HBR criteria  Low body weight 156 (24.7) 35 (16.6) 57 (27.3) 64 (31.7) 0.0013  Severe CKD 97 (15.3) 6 (2.8) 21 (10.0) 70 (33.3) <0.0001  Moderate to severe anemia 78 (12.3) 6 (2.8) 18 (8.5) 54 (25.7) <0.0001  Heart failure 160 (25.3) 20 (9.5) 48 (22.8) 92 (43.8) <0.0001  Anticipated use of long-term oral anticoagulation 97 (15.3) 21 (10.0) 32 (15.2) 44 (21.0) 0.0074  Peripheral vascular disease 68 (10.8) 11 (5.2) 23 (10.9) 34 (16.2) 0.0014  History of nontraumatic bleeding events 10 (1.6) 2 (1.0) 1 (0.5) 7 (3.3) 0.0419  Previous ischemic stroke or ICH 13 (2.1) 3 (1.4) 5 (2.4) 5 (2.4) 0.7282  Thrombocytopenia (platelet count <100 × 109/L) 8 (1.3) 3 (1.4) 0 (0.0) 5 (2.4) 0.0891  Active malignancy 27 (4.3) 4 (1.9) 7 (3.3) 16 (7.6) 0.0104  Liver cirrhosis with portal hypertension 4 (0.6) 0 (0.0) 1 (0.5) 3 (1.4) 0.1701  Chronic bleeding diatheses 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)  Nondeferrable major surgery on DAPT 27 (4.3) 6 (2.8) 11 (5.2) 10 (4.8) 0.4420  Recent major surgery or major trauma within 30 d before PCI 4 (0.6) 0 (0.0) 3 (1.4) 1 (0.5) 0.1726 Values are mean ± SD, n (%), n, or median (IQR). ACE = angiotensin-converting enzyme inhibitor; ACS = acute coronary syndrome; ARB = angiotensin receptor blocker; BNP = brain natriuretic peptide; CABG = coronary artery bypass graft; CCS = chronic coronary syndrome; CKD = chronic kidney disease; DAPT = dual antiplatelet therapy; eGFR = estimated glomerular filtration rate; GDF= growth differentiation factor; HBR = high bleeding risk; hsCRP = high-sensitivity C-reactive protein; ICH = intracranial hemorrhage; LMT = left main trunk; MRA = mineral corticoid receptor antagonist; PCI = percutaneous coronary intervention. a P < 0.05 vs first tertile. b P < 0.05 vs second tertile by analysis of variance with Tukey’s post hoc test. c P < 0.05 vs first tertile. d P < 0.05 vs second tertile by Steel-Dwass test. As shown in Figure 1, all patients were divided into 3 groups according to serum GDF-15 tertiles: first tertile group, GDF-15 <1,509 pg/mL (n = 211); second tertile group, GDF-15 1,509 to 2,604 pg/mL (n = 211); and third tertile group, GDF-15 >2,604 pg/mL (n = 210). The third tertile group was older and had higher prevalence rates of ACS, left main trunk lesions, 3-vessel disease, hypertension, diabetes mellitus, prior AF, prior HF, and prior stroke compared with the other groups. Additionally, the third tertile group showed lower estimated glomerular filtration rate and hemoglobin levels, higher TnT, BNP, hsCRP levels, and higher ABC-AF-bleeding scores than the other groups. The third tertile group had higher prevalence rates of low body weight, severe CKD, moderate to severe anemia, HF, anticipated use of long-term oral anticoagulation, PVD, history of nontraumatic bleeding events, and active malignancy than the other groups (Table 1).Figure 1 Study Flowchart Flowchart of the present study. GDF= growth differentiation factor; HF = heart failure; IHD = ischemic heart disease; MACE = major adverse cardiovascular event(s). Association of serum GDF-15 with coronary artery disease severity and J-HBR criteria Serum GDF-15 levels were elevated in patients with severe coronary artery disease, such as left main trunk lesions and 3-vessel disease (Figures 2A and 2B). Furthermore, serum GDF-15 levels were higher in patients with ACS than in those with CCS (Figure 2C). As shown in Figures 2D to 2I, serum GDF-15 levels were significantly elevated in patients with low body weight, severe CKD, moderate to severe anemia, HF, anticipated use of long-term oral anticoagulation, and PVD. Notably, serum GDF-15 levels increased according to the J-HBR criteria (Figure 2J, Central Illustration).Figure 2 Associations of Serum GDF-15 With IHD Severity and J-HBR Associations of serum GDF-15 level with left main trunk (LMT) lesion (A), 3-vessel disease (B), and diagnosis of IHD (C). Associations of serum GDF-15 level with low body weight (Bw) (D), severe chronic kidney disease (CKD) (E), moderate to severe anemia (F), heart failure (G), anticipated use of long-term oral anticoagulation (H), and peripheral vascular disease (PVD) (I). Association between serum GDF-15 level and an accumulation of the Japanese version of high bleeding risk (J-HBR). ACS = acute coronary syndrome; CCS = chronic coronary syndrome; HBR = high bleeding risk; other abbreviations as in Figure 1. Central Illustration GDF-15 and Clinical Outcomes in Japanese Patients With IHD Serum growth differentiation factor (GDF)-15 elevation was related to the Japanese version of high bleeding risk. GDF-15 could stratify patients with ischemic heart disease (IHD) at high risk for all-cause death and major bleeding events. ACS = acute coronary syndrome; Bw = body weight; CCS = chronic coronary syndrome; CKD = chronic kidney disease; PVD = peripheral vascular disease. Serum GDF-15 levels and primary endpoint During the follow-up period, 78 all-cause deaths occurred in this study. As shown in Figure 3A, Kaplan-Meier analysis revealed that the highest tertile group had the greatest risk for all-cause mortality among the 3 groups (Central Illustration). Multivariate Cox proportional hazards regression analysis demonstrated that serum GDF-15 level was an independent predictor of all-cause mortality after adjusting for age, sex, hemoglobin, log TnT, log BNP, log hsCRP, revascularization, and severe CKD (HR: 1.42; 95% CI: 1.11-1.79; P = 0.0038) (Table 2). To examine whether the model fit and discrimination improved when serum GDF-15 was added to predictors selected in univariate Cox proportional hazards regression analysis, we evaluated improvements in the C index, NRI, and IDI. Serum GDF-15 significantly improved the C index, with significant NRI and IDI (Figure 3B).Figure 3 Prediction Capacity of Serum GDF-15 for Primary Outcome (A) Kaplan-Meier analysis for all-cause mortality among serum growth differentiation factor (GDF)-15 tertiles. (B) Receiver-operating characteristic curve for all-cause mortality. Comparison of C index, net reclassification improvement (NRI), and integrated discrimination improvement (IDI) for all-cause mortality between baseline model with and without serum GDF-15. The baseline model includes age, sex, hemoglobin, troponin T, brain natriuretic peptide, high-sensitivity C-reactive protein, revascularization, and severe chronic kidney disease. Ref = reference. Table 2 Univariate and Multivariate Cox Proportional Hazards Analyses for All-Cause Mortality in Patients With Ischemic Heart Disease Univariate Analysis Multivariate Analysis HR 95% CI P Value HR 95% CI P Value Agea 2.80 2.11-3.76 <0.0001 1.78 1.32-2.43 0.0002 Men vs women 0.42 0.27-0.66 0.0002 0.87 0.52-1.43 0.5722 ACS vs CCS 1.57 0.89-2.76 0.1178 Hypertension 1.85 0.98-3.50 0.0591 Diabetes mellitus 0.95 0.60-1.50 0.8310 Hba 0.45 0.37-0.55 <0.0001 0.81 0.62-1.06 0.1303 Log TnTa 1.39 1.09-1.79 0.0095 1.11 0.81-1.55 0.5144 Log BNPa 2.82 2.24-3.56 <0.0001 1.99 1.46-2.75 <0.0001 Log hsCRPa 1.54 1.21-1.95 0.0004 1.05 0.77-1.44 0.7787 Log GDF-15a 2.06 1.75-2.40 <0.0001 1.42 1.11-1.79 0.0038 Revascularization 0.29 0.15-0.58 0.0005 0.76 0.35-1.66 0.4946 Low body weight 1.46 0.88-2.42 0.1460 Severe CKD 2.02 1.23-3.34 0.0058 0.59 0.34-1.02 0.0589 Moderate to severe anemia 4.00 2.48-6.44 <0.0001 Heart failure 5.18 3.30-8.14 <0.0001 Anticipated use of long-term oral anticoagulation 0.64 0.31-1.33 0.2345 PVD 1.01 0.51-2.03 0.9724 Hb = hemoglobin; PVD = peripheral vascular disease; TnT = troponin T; other abbreviations as in Table 1. a Per 1 SD increase. Serum GDF-15 levels and secondary endpoints During the follow-up period, 148 MACE, 90 HF-related hospitalizations, 78 bleeding events, and 51 thrombotic events occurred. As shown in Figure 4, Kaplan-Meier analysis revealed that the highest tertile group had the greatest risk for MACE, HF-related hospitalization, and bleeding events among the 3 groups (Central Illustration). In contrast, the thrombotic event rate was not significantly different among the serum GDF-15 tertile groups.Figure 4 Different Risk for Secondary Outcomes Among Serum Growth Differentiation Factor-15 Tertiles Kaplan-Meier analysis for major adverse cardiovascular events (MACE) (A), heart failure (HF)-related rehospitalization (B), bleeding events (C), and thrombotic events (D) among serum growth differentiation factor-15 tertiles. Multivariate Cox proportional hazards regression analysis demonstrated that serum GDF-15 level was an independent predictor for MACE, HF-related hospitalization, and bleeding events after adjustment for confounding risk factors (Table 3). Similarly, ABC-AF-bleeding score was an independent predictor for bleeding events. However, a significant association was not maintained in the multivariate Cox proportional hazards regression analysis for thrombotic events.Table 3 Univariate and Multivariate Cox Proportional Hazards Analyses for Secondary Endpoints in Patients With Ischemic Heart Disease Univariate Analysis Multivariate Analysis HR 95% CI P Value HR 95% CI P Value MACE  Agea 1.54 1.28-1.85 <0.0001 0.99 0.81-1.22 0.9379  Men vs women 0.55 0.39-0.78 0.0007 0.69 0.47-1.01 0.0540  ACS vs CCS 1.39 0.94-2.06 0.1019  Hypertension 1.61 1.04-2.49 0.0336 0.96 0.61-1.50 0.8455  Diabetes mellitus 1.33 0.96-1.84 0.0813  Hba 0.61 0.52-0.71 <0.0001 0.97 0.80-1.17 0.7318  Log TnTa 1.14 1.13-1.60 0.0011 1.12 0.90-1.41 0.3118  Log BNPa 2.43 2.06-2.88 <0.0001 1.87 1.51-2.34 <0.0001  Log hsCRPa 1.43 1.20-1.69 <0.0001 1.03 0.83-1.29 0.7950  Log GDF-15a 2.02 1.76-2.30 <0.0001 1.43 1.17-1.74 0.0003  Revascularization 0.52 0.27-0.99 0.0473 1.09 0.55-2.16 0.8063  Low body weight 1.40 0.98-2.01 0.0657  Severe CKD 2.83 2.00-4.00 <0.0001 1.08 0.71-1.66 0.7089  Moderate to severe anemia 2.77 1.90-4.06 <0.0001  Heart failure 4.05 2.92-5.60 <0.0001  Anticipated use of long-term oral anticoagulation 1.85 1.26-2.71 0.0017 1.38 0.93-2.06 0.1082  PVD 1.67 1.09-2.57 0.0194 1.47 0.93-2.32 0.0997 HF-related rehospitalization  Agea 1.90 1.49-2.45 <0.0001 1.05 0.79-1.41 0.7323  Men vs women 0.48 0.31-0.73 0.0007 0.66 0.40-1.10 0.1115  ACS vs CCS 1.47 0.88-2.47 0.1440  Hypertension 1.73 0.98-3.05 0.0604  Diabetes mellitus 1.50 0.99-2.67 0.0550  Hba 0.57 0.86-1.60 <0.0001 0.87 0.67-1.12 0.2907  Log TnTa 1.51 1.20-1.93 0.0007 1.25 0.93-1.71 0.1474  Log BNPa 3.38 2.72-4.22 <0.0001 2.23 1.67-3.01 <0.0001  Log hsCRPa 1.53 1.23-1.91 0.0002 0.85 0.64-1.14 0.2752  Log GDF-15a 2.51 2.13-2.95 <0.0001 1.61 1.26-2.05 0.0001  Revascularization 0.57 0.25-1.30 0.1798  Low body weight 1.82 1.16-2.85 0.0089 0.88 0.53-1.47 0.6301  Severe CKD 4.25 2.79-6.47 <0.0001 1.18 0.69-2.01 0.5393  Moderate to severe anemia 4.28 2.74-6.67 <0.0001  Heart failure 7.53 4.86-11.65 <0.0001  Anticipated use of long-term oral anticoagulation 2.41 1.53-3.81 0.0002 2.00 1.24-3.21 0.0044  PVD 1.23 0.67-2.25 0.5127 Bleeding events  Agea 1.68 1.30-2.18 <0.0001 1.26 0.94-1.70 0.1273  Men vs women 0.66 0.40-1.07 0.0931  ACS vs CCS 1.08 0.65-1.81 0.7554  Hypertension 1.85 0.98-3.49 0.0596  Diabetes mellitus 0.83 0.53-1.32 0.4349  Prior bleeding 1.32 0.48-3.60 0.5937  Hba 0.57 0.46-0.71 <0.0001 0.74 0.56-0.96 0.0240  Log TnTa 1.14 0.91-1.44 0.2580  Log BNP 1.80 1.44-2.26 <0.0001  Log hsCRP 1.35 1.07-1.70 0.0119  Log GDF-15a 1.78 1.48-2.11 <0.0001 1.57 1.24-1.96 0.0001  Low body weight 1.72 1.08-2.75 0.0229 0.89 0.53-1.50 0.6691  Severe CKD 1.80 1.07-3.02 0.0257 0.66 0.36-1.20 0.1732  Moderate to severe anemia 3.14 1.89-5.23 <0.0001  Heart failure 2.72 1.74-4.28 <0.0001 1.49 0.87-2.54 0.1463  Anticipated use of long-term oral anticoagulation 1.58 0.91-2.73 0.1050 1.33 0.76-2.33 0.3254  PVD 1.43 0.77-2.64 0.2589 1.14 0.61-2.13 0.6797 Thrombotic events  Agea 1.19 0.89-1.62 0.2415  Men vs women 0.67 0.37-1.22 0.1888  ACS vs CCS 0.92 0.50-1.68 0.7781  Hypertension 1.40 0.68-2.87 0.3645  Diabetes mellitus 1.08 0.62-1.88 0.7830  Hba 0.82 0.62-1.08 0.1524  Log TnTa 1.10 0.83-1.48 0.5045  Log BNPa 1.80 1.36-2.39 <0.0001 1.70 1.22-2.37 0.0018  Log hsCRPa 1.21 0.91-1.61 0.1936  Log GDF-15a 1.45 1.12-1.83 0.0035 1.07 0.75-1.47 0.7001  Revascularization 0.56 0.17-1.79 0.3267  Low body weight 1.19 0.64-2.22 0.5723  Severe CKD 2.05 1.11-3.80 0.0220 1.08 0.53-2.24 0.8261  Moderate to severe anemia 1.62 0.76-3.44 0.2127  Heart failure 2.41 1.37-4.23 0.0022  Anticipated use of long-term oral anticoagulation 1.10 0.52-2.33 0.8125  PVD 3.02 1.63-5.60 0.0004 2.74 1.46-5.16 0.0018 Bleeding events and ABC-AF-bleeding score  Low body weight 1.72 1.08-2.75 0.0229 1.13 0.68-1.85 0.6419  Severe CKD 1.80 1.07-3.02 0.0257 0.81 0.44-1.49 0.5031  Heart failure 2.72 1.74-4.28 <0.0001 1.85 1.10-3.10 0.0197  Anticipated use of long-term oral anticoagulation 1.58 0.91-2.73 0.1050 1.32 0.76-2.32 0.3244  PVD 1.43 0.77-2.64 0.2589 1.37 0.73-2.56 0.3314  ABC-AF-bleeding scorea 1.79 1.46-2.24 <0.0001 1.61 1.25-2.06 0.0002 MACE = major adverse cardiovascular event(s); other abbreviations as in Tables 1 and 2. a Per 1 SD increase. To examine whether model fit and discrimination improved when serum GDF-15 was added to predictors selected in univariate Cox proportional hazards regression analysis, we evaluated improvements in the C index, NRI, and IDI for MACE and HF-related rehospitalizations. Similarly, we examined whether the addition of GDF-15 to J-HBR could improve the C index, NRI, and IDI for bleeding events. There were no significant differences in the C index for MACE, HF rehospitalization, and bleeding events between the baseline model with and without serum GDF-15. However, NRI and IDI for MACE, HF rehospitalization, and bleeding events improved significantly after adding serum GDF-15 (Figures 5A to 5C).Figure 5 Prediction Capacity of Serum Growth Differentiation Factor-15 for Secondary Outcomes Receiver-operating characteristic curves for MACE (A), HF-related rehospitalizations (B), and bleeding events (C). Comparison of C index, NRI, and IDI for MACE, HF-related rehospitalization, and bleeding events between baseline model with and without serum growth differentiation factor-15. The baseline model for MACE includes age, sex, hypertension, hemoglobin, troponin T (TnT), brain natriuretic peptide (BNP), and high-sensitivity C-reactive protein (hsCRP), revascularization, severe CKD, oral anticoagulation, and PVD. The baseline model for HF-related rehospitalization includes age, sex, hemoglobin, TnT, BNP, hsCRP, low body weight, severe CKD, and oral anticoagulation. The baseline model for bleeding events includes age, hemoglobin, low body weight, severe CKD, HF, oral anticoagulation, and PVD. Abbreviations as in Figures 1, 2, and 3. Discussion The novel findings from the present study are as follows: 1) serum GDF-15 elevation was involved in severe coronary artery disease and J-HBR criteria in patients with IHD; 2) multivariate Cox proportional hazards regression analysis demonstrated that serum GDF-15 was an independent predictor of all-cause death, MACE, HF-related rehospitalization, and bleeding events but not of thrombotic events; and 3) serum GDF-15 improved prediction ability for all-cause death, MACE, HF-related rehospitalization, and bleeding events. Serum GDF-15 level and HBR The association between serum GDF-15 and major bleeding in Asian patients with IHD has not yet been examined. Low body weight is a specific bleeding risk factor in Japanese patients with IHD.6 An experimental study has indicated that GDF-15 regulates appetite and body weight by binding to the glial cell–derived neurotrophic factor receptor alpha-like, expressed mainly in postrema neurons.21 A study comparing serum GDF-15 levels in twin pairs demonstrated a relationship between GDF-15 levels and weight loss in humans.22 The more severe the CKD, the higher the risk for bleeding events in patients with IHD.6 A high GDF-15 level was reportedly associated with the progression and incidence of CKD and mortality in patients undergoing hemodialysis.23, 24, 25 Anemia involved in renal dysfunction and chronic inflammation is reportedly associated with high GDF-15 levels, while iron deficiency anemia is not.26,27 An inverse correlation between serum GDF-15 level and hemoglobin has been reported in patients with HF and in heart allograft recipients.28 HF is reportedly associated with elevated circulating GDF-15 levels.29 Previous reports have demonstrated a significant association of high GDF-15 with mortality and HF rehospitalization in patients with HF.30 In the present study, AF was the primary reason for the anticipated use of long-term oral anticoagulation. It has been reported that plasma and atrial tissue expression of GDF-15 are higher in patients with AF than in those with sinus rhythm.31 GDF-15 is also related to clinical outcomes, including mortality, major bleeding, and stroke, in patients with AF.14 PVD is a common comorbidity in patients with IHD. De Haan et al32 demonstrated that serum GDF-15 level was correlated with the severity of PVD and was significantly associated with major amputation and all-cause mortality in patients with PVD. We found that serum GDF-15 levels increased with increasing HBR risk, suggesting that serum GDF-15 levels, in part, reflect the accumulation of HBRs. As the significant association between serum GDF-15 and bleeding events was maintained in the multivariate analysis after adjustment for J-HBR, another mechanism should explain the close association between serum GDF-15 level and bleeding events. An experimental study demonstrated that GDF-15 could prevent platelet aggregation by inhibiting the activation of glycoprotein IIb/IIIa,33 which is the most abundant receptor in human platelets and acts as a binding site for fibrinogen and von Willebrand factor, resulting in interaction with coagulation factors and other platelets.34 Clinical studies have suggested that GDF-15 could be a marker for bleeding events in patients with IHD.35 Importantly, we showed that the prediction model for bleeding events was significantly improved after adding GDF-15, as evidenced by an improvement in the NRI and IDI. This result confirmed the clinical importance of GDF-15 in bleeding events and raised the possibility that GDF-15 could be a clinical factor in addition to J-HBR in patients with IHD. Serum GDF-15 level and clinical outcomes Accumulating evidence indicates that GDF-15 predicts mortality and HF development.30 Similarly, we demonstrated for the first time that serum GDF-15 could identify and risk-stratify Japanese patients with IHD at high risk for all-cause mortality, MACE, and HF-related rehospitalization. Despite the close relationship between serum GDF-15 levels and severe coronary artery disease, GDF-15 was not significantly associated with thrombotic events. One report showed that GDF-15 could predict thrombotic events.35,36 Others have shown no significant association between GDF-15 level and thrombotic events in patients with AF.14,37 A lower incidence of thrombotic events among Asian patients with IHD may yield this result.5 Therefore, thrombotic events were not the major cause of death in the high GDF-15 group. A recent study indicated the importance of bleeding events in all-cause mortality.1 Thus, we focused on the association of serum GDF-15 levels with major J-HBR criteria and bleeding events. As mentioned previously, serum GDF-15 levels were related to the clinical outcomes of CKD, HF, AF, and PVD.14,24,30,32 Therefore, the high mortality observed in the highest GDF-15 tertile group may be explained by exacerbation of HBR and bleeding events. Importantly, we showed that the prediction models for all-cause mortality, MACE, and HF-related rehospitalization improved significantly after adding serum GDF-15 levels, as evidenced by improvements in NRI and IDI. This result confirmed the clinical importance of GDF-15 in clinical outcomes and raises the possibility that GDF-15 could be an additional clinical factor to consider in patients with IHD. Study limitations First, as this was a prospective observational study, we could not determine the causal relationship between serum GDF-15 levels and clinical outcomes. To eliminate the effect of GDF-15 measurement on antithrombotic medicine, optimization of medical therapy was performed by physicians who were blinded to GDF-15 value throughout the study period. Second, frailty, which is also a major criterion for J-HBR, was not assessed. Third, because cardiologists prescribed antiplatelet medication according to the guideline recommendations, the intensity and duration of antiplatelet therapy were altered during the study period. This may have affected the bleeding event rate. Finally, the optimal cutoff value for serum GDF-15 remains unclear. Conclusions Our study revealed a close association among serum GDF-15 level, J-HBR, and bleeding events in Japanese patients with IHD. Serum GDF-15 level improved the prediction capacity for all-cause mortality, MACE, HF-related rehospitalization, and bleeding events. Serum GDF-15 could serve as a discriminatory factor between thrombotic and bleeding risk and is a feasible marker for major bleeding and adverse clinical outcomes in Asian patients with IHD.Perspectives COMPETENCY IN MEDICAL KNOWLEDGE: We demonstrate, for the first time, the prognostic usefulness of serum GDF-15 for mortality, MACE (except for thrombotic events), HF development, and major bleeding events in Asian patients with IHD. This knowledge serves as a management tool for patients with HBR and early identification and risk stratification of high-risk patients for poor clinical outcomes. TRANSLATIONAL OUTLOOK: Future study is required to confirm the feasibility and clinical utility of GDF-15 in patients with IHD across different geographic areas. Funding Support and Author Disclosures The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Acknowledgments The authors thank Editage for English language editing. The serum GDF-15 assay was provided by Roche, which had no role in the design of the study, the analysis of the data, the preparation of the manuscript, or the decision to submit the manuscript for publication. The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center. ==== Refs References 1 Mauri L. Kereiakes D.J. Yeh R.W. Twelve or 30 months of dual antiplatelet therapy after drug-eluting stents N Engl J Med 371 2014 2155 2166 25399658 2 Rao S.V. O’Grady K. Pieper K.S. Impact of bleeding severity on clinical outcomes among patients with acute coronary syndromes Am J Cardiol 96 2005 1200 1206 16253582 3 Eikelboom J.W. Mehta S.R. Anand S.S. Xie C. Fox K.A. Yusuf S. Adverse impact of bleeding on prognosis in patients with acute coronary syndromes Circulation 114 2006 774 782 16908769 4 Palmerini T. Bacchi Reggiani L. Della Riva D. Bleeding-related deaths in relation to the duration of dual-antiplatelet therapy after coronary stenting J Am Coll Cardiol 69 2017 2011 2022 28427576 5 Kang J. Park K.W. Palmerini T. Racial differences in ischaemia/bleeding risk trade-off during anti-platelet therapy: individual patient level landmark meta-analysis from seven RCTs Thromb Haemost 119 2019 149 162 30597509 6 Nakamura M. Kimura K. Kimura T. JCS 2020 guideline focused update on antithrombotic therapy in patients with coronary artery disease Circ J 84 2020 831 865 32173684 7 Baek S.J. Eling T. Growth differentiation factor 15 (GDF15): a survival protein with therapeutic potential in metabolic diseases Pharmacol Ther 198 2019 46 58 30790643 8 Bootcov M.R. Bauskin A.R. Valenzuela S.M. MIC-1, a novel macrophage inhibitory cytokine, is a divergent member of the TGF-beta superfamily Proc Natl Acad Sci U S A 94 1997 11514 11519 9326641 9 Corre J. Hebraud B. Bourin P. Concise review: growth differentiation factor 15 in pathology: a clinical role? Stem Cells Transl Med 2 2013 946 952 24191265 10 Kempf T. Björklund E. Olofsson S. Growth-differentiation factor-15 improves risk stratification in ST-segment elevation myocardial infarction Eur Heart J 28 2007 2858 2865 17977844 11 Wollert K.C. Kempf T. Peter T. Prognostic value of growth-differentiation factor-15 in patients with non-ST-elevation acute coronary syndrome Circulation 115 2007 962 971 17283261 12 Kempf T. Sinning J.M. Quint A. Growth-differentiation factor-15 for risk stratification in patients with stable and unstable coronary heart disease: results from the AtheroGene study Circ Cardiovasc Genet 2 2009 286 292 20031597 13 Hagstrom E. Held C. Stewart R.A. Growth differentiation factor 15 predicts all-cause morbidity and mortality in stable coronary heart disease Clin Chem 63 2017 325 333 27811204 14 Wallentin L. Hijazi Z. Andersson U. Growth differentiation factor 15, a marker of oxidative stress and inflammation, for risk assessment in patients with atrial fibrillation: insights from the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial Circulation 130 2014 1847 1858 25294786 15 Pol T. Hijazi Z. Lindbäck J. Evaluation of the prognostic value of GDF-15, ABC-AF-bleeding score and ABC-AF-death score in patients with atrial fibrillation across different geographical areas Open Heart 8 2021 e001471 16 Thygesen K. Alpert J.S. Jaffe A.S. Fourth universal definition of myocardial infarction (2018) Circulation 138 2018 e618 e651 30571511 17 Knuuti J. Wijns W. Saraste A. 2019 ESC guidelines for the diagnosis and management of chronic coronary syndromes Eur Heart J 41 2020 407 477 31504439 18 Kimura K. Kimura T. Ishihara M. JCS 2018 guideline on diagnosis and treatment of acute coronary syndrome Circ J 83 2019 1085 1196 30930428 19 Neumann F.J. Sousa-Uva M. Ahlsson A. 2018 ESC/EACTS guidelines on myocardial revascularization Eur Heart J 40 2019 87 165 30165437 20 Yoshida K. Tanabe Y. Kido T. Characteristics of the left ventricular three-dimensional maximum principal strain using cardiac computed tomography: reference values from subjects with normal cardiac function Eur Radiol 30 2020 6109 6117 32556462 21 Mullican S.E. Lin-Schmidt X. Chin C.N. GFRAL is the receptor for GDF15 and the ligand promotes weight loss in mice and nonhuman primates Nat Med 23 2017 1150 1157 28846097 22 Tsai V.W. Macia L. Feinle-Bisset C. Serum levels of human MIC-1/GDF15 vary in a diurnal pattern, do not display a profile suggestive of a satiety factor and are related to BMI PLoS ONE 10 2015 e0133362 23 Nair V. Robinson-Cohen C. Smith M.R. Growth differentiation factor-15 and risk of CKD progression J Am Soc Nephrol 28 2017 2233 2240 28159780 24 Ho J.E. Hwang S.J. Wollert K.C. Biomarkers of cardiovascular stress and incident chronic kidney disease Clin Chem 59 2013 1613 1620 23873716 25 You A.S. Kalantar-Zadeh K. Lerner L. Association of growth differentiation factor 15 with mortality in a prospective hemodialysis cohort Cardiorenal Med 7 2017 158 168 28611789 26 Waalen J. von Löhneysen K. Lee P. Xu X. Friedman J.S. Erythropoietin, GDF15, IL6, hepcidin and testosterone levels in a large cohort of elderly individuals with anaemia of known and unknown cause Eur J Haematol 87 2011 107 116 21535154 27 Theurl I. Finkenstedt A. Schroll A. Growth differentiation factor 15 in anaemia of chronic disease, iron deficiency anaemia and mixed type anaemia Br J Haematol 148 2010 449 455 19863534 28 Przybyłowski P. Wasilewski G. Bachorzewska-Gajewska H. Golabek K. Dobrzycki S. Małyszko J. Growth differentiation factor 15 is related to anemia and iron metabolism in heart allograft recipients and patients with chronic heart failure Transplant Proc 46 2014 2852 2855 25380934 29 Chan M.M. Santhanakrishnan R. Chong J.P. Growth differentiation factor 15 in heart failure with preserved vs. reduced ejection fraction Eur J Heart Fail 18 2016 81 88 26497848 30 Wollert K.C. Kempf T. Wallentin L. Growth differentiation factor 15 as a biomarker in cardiovascular disease Clin Chem 63 2017 140 151 28062617 31 Li N. Feng Q. Yu F. Zhou J. Guo X. Plasma growth differentiation factor-15 in patients with “lone” atrial fibrillation J Clin Lab Anal 36 2022 e24373 32 De Haan J.J. Haitjema S. den Ruijter H.M. Growth differentiation factor 15 is associated with major amputation and mortality in patients with peripheral artery disease J Am Heart Assoc 6 2017 e006225 33 Tscharre M. Wittmann F. Kitzmantl D. Growth differentiation factor-15 correlates inversely with protease-activated receptor-1-mediated platelet reactivity in patients with left ventricular assist devices Pharmaceuticals (Basel) 15 2022 484 35455481 34 Tscharre M. Michelson A.D. Gremmel T. Novel antiplatelet agents in cardiovascular disease J Cardiovasc Pharmacol Ther 25 2020 191 200 31960728 35 Hagstrom E. James S.K. Bertilsson M. Growth differentiation factor-15 level predicts major bleeding and cardiovascular events in patients with acute coronary syndromes: results from the PLATO study Eur Heart J 37 2016 1325 1333 26417057 36 Lindholm D. Hagström E. James S.K. Growth differentiation factor 15 at 1 month after an acute coronary syndrome is associated with increased risk of major bleeding J Am Heart Assoc 6 2017 e005580 37 Hijazi Z. Oldgren J. Andersson U. Growth-differentiation factor 15 and risk of major bleeding in atrial fibrillation: insights from the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial Am Heart J 190 2017 94 103 28760218