Aggregate Clinical and Biomarker-Based Model Predicts Adverse Outcomes in Patients With Coronary Artery Disease

• DOI: 10.1016/j.amjcard.2023.06.115
• Publication Date: 2023-07-28T14:45:59Z
• Authors: Shivang R. Desai; Devinder S. Dhindsa; Yi-An Ko; Pratik B. Sandesara; Anurag Mehta; Chang Liu; Ayman S. Tahhan; Salim S. Hayek; Kiran Ejaz; Ananya Hooda; Ayman Alkhoder; Shabatun J. Islam; Steven C. Rogers; Agim Beshiri; Gillian Murtagh; Jonathan H. Kim; Peter Wilson; Zakaria Almuwaqqat; Laurence S. Sperling; Arshed A. Quyyumi
• Keywords: Male; Predictive Value of Tests; Risk Factors; Myocardial Infarction; Humans; Female; Coronary Artery Disease; Middle Aged; Prognosis; Risk Assessment; Article; Biomarkers

Despite guideline-based therapy, patients with coronary artery disease (CAD) are at widely variable risk for cardiovascular events. This variability demands a more individualized assessment. Herein, we evaluate the prognostic value of 6 biomarkers: high-sensitivity C-reactive protein, heat shock protein-70, fibrin degradation products, soluble urokinase plasminogen activator receptor, troponin I, and B-type natriuretic peptide. We then develop multi-biomarker-based event prediction model stable CAD. In total, 3,115 subjects CAD who underwent cardiac catheterization Emory (mean age 62.8 years, 17% Black, 35% female, 57% obstructive CAD, 31% diabetes mellitus) were randomized into training cohort to identify biomarker cutoff values validation Main outcomes included (1) all-cause death (2) composite nonfatal myocardial infarction (MI) within 5 years. Elevation each level was associated higher rates in cohort. A score created using optimal cutoffs, ranging from 0 exceeding its cutoff. cohort, unit increase independently (hazard ratio 1.62, 95% confidence interval [CI] 1.45 1.80) death/MI 1.52, CI 1.35 1.71). improved c-statistic (∆ 6.4%, 3.9 8.8) net reclassification index by 31.1% (95% 24 37), compared clinical factors alone. Integrating multiple biomarkers variables refines assessment Patients established high recurrent standardization secondary prevention, heterogeneous events exists that approach management.1Bhatt DL Eagle KA Ohman EM Hirsch AT Goto S Mahoney Wilson PW Alberts MJ D'Agostino R Liau CS Mas JL Röther J Smith Jr, SC Salette G Contant CF Massaro JM Steg PG REACH Registry InvestigatorsComparative determinants 4-year outpatients or atherothrombosis.JAMA. 2010; 304: 1350-1357Crossref PubMed Scopus (595) Google Scholar Traditional measures such as left ventricular ejection fraction (LVEF) commonly used assess but fail reliably estimate residual events.2Sabatine MS Morrow DA de Lemos JA Gibson CM Murphy SA Rifai N McCabe C Antman Cannon CP Braunwald E. Multimarker stratification non-ST elevation acute syndromes: simultaneous peptide.Circulation. 2002; 105: 1760-1763Crossref (654) Scholar,3Vittinghoff E Shlipak MG Varosy PD Furberg CD Ireland CC Khan SS Blumenthal Barrett-Connor Hulley Heart Estrogen/progestin Replacement Study Research Group. Risk prevention women heart disease: estrogen/progestin Study.Ann Intern Med. 2003; 138: 81-89Crossref The transition plaque disruption is complex involves activation pathophysiologic pathways including inflammation, immune activation, cellular stress, coagulation, stretch, ischemia.4Eapen DJ Manocha P Ghasemzadeh Patel RS Al Kassem H Hammadah M Veledar Le NA Pielak T Thorball CW Velegraki Kremastinos DT Lerakis Sperling L Quyyumi AA Soluble receptor an independent predictor presence severity future adverse events.J Am Assoc. 2014; 3e001118Crossref Scholar, 5Eapen Wassel Nanjundappa RA Sikora Malayter D Epstein SE. Aggregate based on markers cell coagulation outcomes.J Coll Cardiol. 2013; 62: 329-337Crossref (69) 6Ghasemzedah Hayek Ko YA Eapen Khayata Zafari AM Vaccarino V SE Pathway-specific aggregate burden predicts near-term death.Circ Cardiovasc Qual Outcomes. 2017; 10e001493Crossref (20) Circulating representing dysregulation these may help greatest thus aid individualization preventive treatment. High-sensitivity protein (hs-CRP) inflammation,7Sabatine Jablonski Rice MM Warnica JW Domanski Hsia Gersh BJ Ridker PM Pfeffer MA Investigators PEACE Prognostic significance Centers Disease Control/American Association cut points other disease.Circulation. 2007; 115: 1528-1536Crossref (305) (suPAR) also activation,4Eapen Scholar,8Hayek Sever Trachtman Awad Wadhwani Altintas Wei Hotton AL French LS Reiser J. chronic kidney disease.N Engl 2015; 373: 1916-1925Crossref (284) proteins (HSPs) stress,9Pockley AG Frostegård Heat related measurements.Heart. 2005; 91: 1124-1126Crossref (15) products (FDPs) coagulation,5Eapen Scholar,10Kannel WB Wolf PA Castelli WP RB Fibrinogen disease. Framingham study.JAMA. 1987; 258: 1183-1186Crossref troponins (hs-TnIs) stress/injury,11Westermann Neumann JT Sörensen Blankenberg S. assays disease.Nat Rev 14: 472-483Crossref (117) peptides marker stretch12Wang TJ Larson Levy Benjamin EJ Leip EP Omland Vasan RS. Plasma peptide levels death.N 2004; 350: 655-663Crossref (1275) 13European Liver (EASL), European Diabetes (EASD), Obesity (EASO)EASL-EASD-EASO practice guidelines management non-alcoholic fatty liver disease.J Hepatol. 2016; 64: 1388-1402Abstract Full Text PDF (2811) 14Berger Huelsman Strecker K Bojic Moser Stanek B Pacher R. sudden failure.Circulation. 2392-2397Crossref (625) 15Tapanainen Lindgren KS Mäkikallio TH Vuolteenaho O Leppäluoto Huikuri HV. Natriuretic predictors non-sudden after beta-blocking era.J 43: 757-763Crossref (102) have shown Herein pathophysiology-specific biomarkers, validate biomarker-based Participants recruited Cardiovascular Biobank, prospective registry diagnosis suspected 3 Healthcare sites between 2004 2010. Details study population been published previously.5Eapen Briefly, ages 20 90 years confirmed interviewed collect demographic information, medical history, factor prevalence, medication use, behavioral habits previously described.5Eapen Subjects syndrome, congenital disease, transplantation, severe valvular anemia, recent blood transfusion, myocarditis, active inflammatory diseases, cancer, limited follow-up time (<30 days), missing data excluded. approved University Institutional Review Board. All provided written informed consent. Gender race self-reported. Anginal symptom self-reported Seattle Angina Questionnaire. hypertension, hyperlipidemia, mellitus (DM), peripheral vascular atrial fibrillation determined physician and/or treatment prescribed chart. Smoking classified nonsmoker current/past smoker. Blood pressure, weight, height measured enrollment date. Body mass (BMI) defined weight kilograms divided meters. Serum creatinine measurements before obtained routine clinic visits hospitalizations system. estimated glomerular filtration rate (eGFR) computed (CKD) Epidemiology Collaboration equation.16Levey AS Stevens LA Schmid CH Zhang YL Castro 3rd, AF Feldman HI Kusek Eggers Van Lente F Greene Coresh CKD-EPI (Chronic Kidney Collaboration). new equation rate.Ann 2009; 150: 604-612Crossref (17079) CKD eGFR <60 ml/min/1.73 m2. Echocardiographic LVEF abstracted reviewing records. Medications history followed review relevant following medications this analysis: aspirin, β blockers, angiotensin-converting enzyme inhibitor/angiotensin II blocker (ACEi/ARB), statins. Luminal narrowing arteries quantified modified American Association/American College Cardiology classification Gensini score.17Austen WG Edwards JE Frye RL GG Gott VL Griffith McGoon DC ML Roe BB. reporting system evaluated Report Ad Hoc Committee Grading Coronary Artery Disease, council surgery, Association.Circulation. 1975; 51: 5-40Crossref Scholar,18Sinning Lillpopp Appelbaum Ojeda Zeller Schnabel Lubos Jagodzinski Keller Munzel Bickel Angiographic improves prediction: SYNTAX application.Clin Res 102: 495-503Crossref (127) Angiography extracted reports catheterization. Fasting arterial samples serum plasma drawn stored −80°C (mean, 4.9 years). described.4Eapen Scholar,5Eapen hs-CRP FDP sandwich immunoassay (FirstMark Inc., San Diego, California). HSP-70 enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, Minnesota) optimized FirstMark. suPAR commercially available kits (suPARnostic kit, Virogates, Copenhagen, Denmark). hs-TnI Abbott ARCHITECT analyzer (Abbott Laboratories, North Chicago, Illinois). (BNP) BNP chemiluminescent microparticle reported pg/ml). limit detection 1.2 pg/ml interassay coefficient variation <10% 4.7 pg/ml. Minimum detectable hs-CRP, FDP, HSP-70, suPAR, 0.1 mg/L, 0.06 μg/ml, 0.313 ng/ml, 10 pg/ml, respectively. Primary end point adjudicated (Supplementary Materials). Major death, hospitalization failure, MI, stroke entire analysis set (n = 3,115) randomly split equally 1,557) 1,558) sets similar baseline characteristics. Using set, Cox regression Fine-Gray association (treating noncardiovascular competing event) (simultaneously continuous variables). Covariates had age, gender, (Black vs non-Black), BMI, smoking, LVEF, DM, (categorical quartiles), eGFR, use ACEi/ARBs, identified gender-specific hs-cTnI. dichotomized low according point. An aggregated (BRS) patient indicate number elevated 6. Subsequently, applied BRS. Interactions BRS CKD, statins examined. attempted 5-year death/MI. listed Table 1 considered development regression. Final models selected smallest Akaike information criterion values. gender considered. Internal final proposed performed 100 bootstrap samples. For samples, parameter estimates observed set.Table 1Baseline demographics, factors, events, profilesBaseline CharacteristicsAll (n=3115)Training Set (n=1557)Validation (n=1558)Age (years)62.8 (11.6)62.7 (11.6)63.0 (11.5)Male2013 (64.6%)995 (63.9%)1018 (65.3%)Black Race513 (16.5%)242 (15.5%)271 (17.4%)BMI, kg/m229.7 (6.0)29.8 (6.1)29.6 (6.0)Current Former Smoker2040 (65.5%)1003 (64.4%)1037 (66.6%)Obstructive CAD*Obstructive ≥50% stenosis least one major epicardial vessel. Non-obstructive stenosis1770 (56.8%)894 (57.4%)876 (56.2%)Non-obstructive stenosis431 (13.8%)221 (14.2%)210 (13.5%)Gensini Score8 [0-34]8 [0-32]8 [0-35]Angiogram Medical management1584 (59.8%)789 (59.5%)795 (60.1%) Stent445 (16.8%)226 (17.0%)219 (16.6%) Angioplasty404 (15.3%)198 (14.9%)206 (15.6%) Surgery195 (7.4%)105 (7.9%)90 (6.8%) Other21 (0.8%)8 (0.6%)13 (1.0%)Seattle angina questionnaire None over past 4 weeks1295 (46.5%)648 (46.9%)647 (46.1%) <1 weekly352 (12.6%)182 (13.2%)170 (12.1%) 1-2 times per week312 (11.2%)164 (11.9%)148 (10.5%) ≥ week344 (12.3%)179 (13.0%)165 (11.8%) 1-3 day318 (11.4%)137 (9.9%)181 (12.9%) day165 (5.9%)72 (5.2%)93 (6.6%)History MI702 (22.6%)340 (21.9%)362 (23.3%)History failure932 (29.9%)451 (29.0%)481 (30.9%)History PCI1339 (43.0%)661 (42.5%)678 (43.5%)History CABG789 (25.3%)394 (25.3%)395 (25.4%)History PVD471 (15.1%)232 (14.9%)239 (15.3%)History fibrillation213 (6.8%)107 (6.9%)106 (6.8%)Hyperlipidemia2178 (70.1%)1074 (69.3%)1104 (70.9%)Hypertension2136 (72.5%)1119 (72.2%)1124 (72.4%)Diabetes919 (31.2%)494 (31.9%)479 (30.8%)LV Ejection Fraction, %54.4 (11.4)54.6 (11.7)54.2 (11.2)eGFR, mL/min/1.73 m273.5 (22.0)72.1 (22.2)74.0 (22.2)Total Cholesterol, mg/dL164 [139-194]163 [139-194]165 [140-195]LDL-C, mg/dL91 [71-117]91 [71-117]92 [71-117]HDL-C, mg/dL40 [33-48]40 [34-49]Triglycerides, mg/dL127 [85-189]129 [84-193]125 [87-184]Medications ACEi/ARB Use1804 (57.9%)908 (58.3%)896 (57.5%) Beta-blocker Use1940 (62.3%)977 (62.8%)963 (61.8%) Statin Use2217 (71.2%)1114 (71.6%)1103 (70.8%) Aspirin Use2334 (74.9%)1181 (75.9%)1153 (74.0%) Plavix Use1421 (45.6%)716 (46.0%)705 (45.3%)Events All-cause Death440 (14.1%)219 (14.1%)221 (14.2%) Death263 (8.4%)124 (8.0%)139 (8.9%) Myocardial Infarction104 (3.3%)42 (2.7%)62 (4.0%) MACE672 (21.6%)321 (20.6%)351 (22.5%)Biomarkers Hs-CRP (mg/L)2.7 [1.2-6.6]2.7 [1.1-6.3]2.7 [1.2-6.8] (>0)†Percentage non-zeros range reported. CABG bypass graft; disease; rate; products; protein; hs-cTnI I; 70; MACE events; MI infarction; PCI percutaneous intervention; PVD receptor.19.9% [0-8120]19.8% [0-6743]20.0% [0-8120] (ug/mL)0.54 [0.36-0.82]0.53 [0.36-0.82]0.54 [0.36-0.81] (pg/mL)3016 [2363-3973]3033 [2383-4044]2989 [2351-3889] Hs-cTnI (ng/mL)4.8 [2.7-10.9]5.0 [2.8-11.4]4.6 [2.6-10.2] (mg/dL)62.5 [24.9-144.0]66.4 [26.4-147.2]58.4 [23.4-138.0]Mean (standard deviation), median [interquartile range], frequency count (percentage) Obstructive stenosis† Percentage reported.CABG receptor. Open table tab Mean Discrimination testing c-statistic, improvement (NRI), integrated discrimination metrics. p <0.05 statistically significant. Statistical analyses SAS (version 9.3; Institute, Cary, Carolina) 3.4.1 (The Foundation Computing, Vienna, Austria). appear Supplemental Materials. total complete (hs-CRP, hs-cTnI, BNP) analysis, excluding (hs-cTnI > 12 ng/ml) those days). Their mean (±11.6 years), 63% (>50% stenosis), DM (Table 1). Overall, during (interquartile 3.3 there 440 (14.1%) deaths, 263 (8.4%) 104 (3.3%) MIs, 672 (21.6%) Spearman correlation coefficients individual weakly significant, Supplementary 1. Figure shows Kaplan–Meier curves significant differences survival stratified set. 2 lists hazard (HR) subdistribution HR (sHR) corresponding 10th 90th percentile adjustment mentioned covariates. results significantly (while present model) increased adjusted model. death/MI, (sHR 2.25, 1.29 3.95), 3.05, 1.48 6.31), (HR 2.43, 1.20 4.93) remained significant.Table 2Predictive ability setOutcomeBiomarker (per percentile)HR/sHR limits)P-valueDeathHs-CRP1.64 (1.06,2.52)0.025HSP-701.30 (0.97,1.74)0.078FDP2.23 (1.40,3.56)0.0008SuPAR4.83 (2.76,8.45)<.0001Hs-cTnI1.97 (1.18,3.30)0.010BNP1.98 (1.19,3.30)0.009CV Death/MIHs-CRP1.27 (0.72,2.23)0.404HSP-701.35 (0.92,1.98)0.123FDP2.25 (1.29,3.95)0.005SuPAR3.05 (1.48,6.31)0.003Hs-cTnI1.72 (0.88,3.36)0.114BNP2.43 (1.20,4.93)0.014Model (black), diabetes, score, beta-blocker all simultaneously model.BNP peptide; CV cardiovascular; ratio; sHR Model (as variables) (LVEF, failure) reached statistical models. generate predicted patient. Adequate goodness-of-fit (p 0.46) Hosmer-Lemeshow test. To whether aligns (model calibration), deciles showed well alone (Figure 1).Table 3Biomarker yearsCoefficient EstimateHazard Ratio Confidence Interval)P-valueLn(hs-CRP)0.071.08 (1.01, 1.15)0.020HSP-70 >11.3 ng/mL0.431.54 (1.20, 1.96)0.001Ln(FDP)0.061.06 (1.00, 1.13)0.044Ln(suPAR)0.571.78 (1.45, 2.17)<.0001Ln(hs-cTnI)0.081.08 (1.02, 1.14)0.014Ln(BNP)0.161.17 (1.10, 1.26)<.0001History failure0.311.36 (1.04, 1.78)0.024LVEF 10% less)-0.120.89 (0.81, 0.98)0.015CKD0.271.30 1.68)0.039Hypertension0.221.24 (0.93, 1.67)0.150Hyperlipidemia0.261.30 (0.99, 1.69)0.067All covariates six (natural log (Ln) transformed except HSP-70) Stepwise selection AIC create variables. Baseline probability 0.99, which calculated likelihood setting covariate 0. Based maximum approach, derived 1,557 hs-TnI, 11.3 0.82 µg/ml, 2,