Advertisement

Machine learning integration of circulating and imaging biomarkers for explainable patient-specific prediction of cardiac events: A prospective study

      Highlights

      • We used machine learning (ML) to integrate clinical data, CT measures, and serum biomarkers for cardiac prognostication.
      • The computed ML risk score outperformed current risk assessment tools for the long-term prediction of hard cardiac events.
      • Serum biomarkers provided incremental prognostic value beyond clinical and imaging features in an ML model.
      • Biomarkers of inflammation, extracellular matrix remodeling, and fibrosis had high variable importance for ML prediction.
      • Our ML model can provide individualized, patient-specific explanations of its predictions.

      Abstract

      Background and aims

      We sought to assess the performance of a comprehensive machine learning (ML) risk score integrating circulating biomarkers and computed tomography (CT) measures for the long-term prediction of hard cardiac events in asymptomatic subjects.

      Methods

      We studied 1069 subjects (age 58.2 ± 8.2 years, 54.0% males) from the prospective EISNER trial who underwent coronary artery calcium (CAC) scoring CT, serum biomarker assessment, and long-term follow-up. Epicardial adipose tissue (EAT) was quantified from CT using fully automated deep learning software. Forty-eight serum biomarkers, both established and novel, were assayed. An ML algorithm (XGBoost) was trained using clinical risk factors, CT measures (CAC score, number of coronary lesions, aortic valve calcium score, EAT volume and attenuation), and circulating biomarkers, and validated using repeated 10-fold cross validation.

      Results

      At 14.5 ± 2.0 years, there were 50 hard cardiac events (myocardial infarction or cardiac death). The ML risk score (area under the receiver operator characteristic curve [AUC] 0.81) outperformed the CAC score (0.75) and ASCVD risk score (0.74; both p = 0.02) for the prediction of hard cardiac events. Serum biomarkers provided incremental prognostic value beyond clinical data and CT measures in the ML model (net reclassification index 0.53 [95% CI: 0.23–0.81], p < 0.0001). Among novel biomarkers, MMP-9, pentraxin 3, PIGR, and GDF-15 had highest variable importance for ML and reflect the pathways of inflammation, extracellular matrix remodeling, and fibrosis.

      Conclusions

      In this prospective study, ML integration of novel circulating biomarkers and noninvasive imaging measures provided superior long-term risk prediction for cardiac events compared to current risk assessment tools.

      Graphical abstract

      Keywords

      To read this article in full you will need to make a payment

      Purchase one-time access:

      Academic & Personal: 24 hour online accessCorporate R&D Professionals: 24 hour online access
      One-time access price info
      • For academic or personal research use, select 'Academic and Personal'
      • For corporate R&D use, select 'Corporate R&D Professionals'

      Subscribe:

      Subscribe to Atherosclerosis
      Already a print subscriber? Claim online access
      Already an online subscriber? Sign in
      Institutional Access: Sign in to ScienceDirect

      References

        • Pencina M.J.
        • D'Agostino RB Sr
        • Larson M.G.
        • Massaro J.M.
        • Vasan R.S.
        Predicting the 30-year risk of cardiovascular disease: the framingham heart study.
        Circulation. 2009; 119: 3078-3084
        • Goff Jr., D.C.
        • Lloyd-Jones D.M.
        • Bennett G.
        • Coady S.
        • D'Agostino R.B.
        • Gibbons R.
        • Greenland P.
        • Lackland D.T.
        • Levy D.
        • O'Donnell C.J.
        • Robinson J.G.
        • Schwartz J.S.
        • Shero S.T.
        • Smith Jr., S.C.
        • Sorlie P.
        • Stone N.J.
        • Wilson P.W.
        • Jordan H.S.
        • Nevo L.
        • Wnek J.
        • Anderson J.L.
        • Halperin J.L.
        • Albert N.M.
        • Bozkurt B.
        • Brindis R.G.
        • Curtis L.H.
        • DeMets D.
        • Hochman J.S.
        • Kovacs R.J.
        • Ohman E.M.
        • Pressler S.J.
        • Sellke F.W.
        • Shen W.K.
        • Smith Jr., S.C.
        • Tomaselli G.F.
        2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of cardiology/American heart association task force on practice guidelines.
        Circulation. 2014; 129: S49-S73
        • Kelkar A.A.
        • Schultz W.M.
        • Khosa F.
        • Schulman-Marcus J.
        • O'Hartaigh B.W.
        • Gransar H.
        • Blaha M.J.
        • Knapper J.T.
        • Berman D.S.
        • Quyyumi A.
        • Budoff M.J.
        • Callister T.Q.
        • Min J.K.
        • Shaw L.J.
        Long-term prognosis after coronary artery calcium scoring among low-intermediate risk women and men.
        Circ Cardiovasc Imaging. 2016; 9e003742
        • Mahabadi A.A.
        • Lehmann N.
        • Mohlenkamp S.
        • Pundt N.
        • Dykun I.
        • Roggenbuck U.
        • Moebus S.
        • Jockel K.H.
        • Erbel R.
        • Kalsch H.
        • Heinz Nixdorf Investigative G.
        Noncoronary measures enhance the predictive value of cardiac CT above traditional risk factors and CAC score in the general population.
        JACC Cardiovasc Imaging. 2016; 9: 1177-1185
        • Kaptoge S.
        • Di Angelantonio E.
        • Lowe G.
        • Pepys M.B.
        • Thompson S.G.
        • Collins R.
        • Danesh J.
        C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis.
        Lancet. 2010; 375: 132-140
        • Wong Y.K.
        • Cheung C.Y.Y.
        • Tang C.S.
        • Au K.W.
        • Hai J.S.H.
        • Lee C.H.
        • Lau K.K.
        • Cheung B.M.Y.
        • Sham P.C.
        • Xu A.
        • Lam K.S.L.
        • Tse H.F.
        Age-biomarkers-clinical risk factors for prediction of cardiovascular events in patients with coronary artery disease.
        Arterioscler. Thromb. Vasc. Biol. 2018; 38: 2519-2527
        • Tromp J.
        • Khan M.A.
        • Klip I.T.
        • Meyer S.
        • de Boer R.A.
        • Jaarsma T.
        • Hillege H.
        • van Veldhuisen D.J.
        • van der Meer P.
        • Voors A.A.
        Biomarker profiles in heart failure patients with preserved and reduced ejection fraction.
        J Am Heart Assoc. 2017; 6
        • Rozanski A.
        • Gransar H.
        • Shaw L.J.
        • Kim J.
        • Miranda-Peats L.
        • Wong N.D.
        • Rana J.S.
        • Orakzai R.
        • Hayes S.W.
        • Friedman J.D.
        • Thomson L.E.J.
        • Polk D.
        • Min J.
        • Budoff M.J.
        • Berman D.S.
        Impact of coronary artery calcium scanning on coronary risk factors and downstream testing.
        J. Am. Coll. Cardiol. 2011; 57: 1622
        • Commandeur F.
        • Slomka P.J.
        • Goeller M.
        • Chen X.
        • Cadet S.
        • Razipour A.
        • McElhinney P.
        • Gransar H.
        • Cantu S.
        • Miller R.J.H.
        • Rozanski A.
        • Achenbach S.
        • Tamarappoo B.K.
        • Berman D.S.
        • Dey D.
        Machine learning to predict the long-term risk of myocardial infarction and cardiac death based on clinical risk, coronary calcium, and epicardial adipose tissue: a prospective study.
        Cardiovasc. Res. 2019; https://doi.org/10.1093/cvr/cvz321
        • Agatston A.S.
        • Janowitz W.R.
        • Hildner F.J.
        • Zusmer N.R.
        • Viamonte Jr., M.
        • Detrano R.
        Quantification of coronary artery calcium using ultrafast computed tomography.
        J. Am. Coll. Cardiol. 1990; 15: 827-832
        • Commandeur F.
        • Goeller M.
        • Razipour A.
        • Cadet S.
        • Hell M.M.
        • Kwiecinski J.
        • Chen X.
        • Chang H.-J.
        • Marwan M.
        • Achenbach S.
        • Berman D.S.
        • Slomka P.J.
        • Tamarappoo B.K.
        • Dey D.
        Fully automated CT quantification of epicardial adipose tissue by deep learning: a multicenter study.
        Radiology: Artif. Intell. 2019; 1e190045
        • Guyon I.
        • Elisseeff A.
        An introduction to variable and feature selection.
        J. Mach. Learn. Res. 2003; 3: 1157-1182
        • Al'Aref S.J.
        • Singh G.
        • Choi J.W.
        • Xu Z.
        • Maliakal G.
        • van Rosendael A.R.
        • Lee B.C.
        • Fatima Z.
        • Andreini D.
        • Bax J.J.
        • Cademartiri F.
        • Chinnaiyan K.
        • Chow B.J.W.
        • Conte E.
        • Cury R.C.
        • Feuchtner G.
        • Hadamitzky M.
        • Kim Y.-J.
        • Lee S.-E.
        • Leipsic J.A.
        • Maffei E.
        • Marques H.
        • Plank F.
        • Pontone G.
        • Raff G.L.
        • Villines T.C.
        • Weirich H.G.
        • Cho I.
        • Danad I.
        • Han D.
        • Heo R.
        • Lee J.H.
        • Rizvi A.
        • Stuijfzand W.J.
        • Gransar H.
        • Lu Y.
        • Sung J.M.
        • Park H.-B.
        • Berman D.S.
        • Budoff M.J.
        • Samady H.
        • Stone P.H.
        • Virmani R.
        • Narula J.
        • Chang H.-J.
        • Lin F.Y.
        • Baskaran L.
        • Shaw L.J.
        • Min J.K.
        A Boosted Ensemble Algorithm for Determination of Plaque Stability in High-Risk Patients on Coronary CTA. JACC: Cardiovascular Imaging.
        2020: 3447
        • Molinaro A.M.
        • Simon R.
        • Pfeiffer R.M.
        Prediction error estimation: a comparison of resampling methods.
        Bioinformatics. 2005; 21: 3301-3307
        • DeLong E.R.
        • DeLong D.M.
        • Clarke-Pearson D.L.
        Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach.
        Biometrics. 1988; 44: 837-845
        • Pencina M.J.
        • D'Agostino R.B.
        • Sr
        • Steyerberg E.W.
        Extensions of net reclassification improvement calculations to measure usefulness of new biomarkers.
        Stat. Med. 2011; 30: 11-21
        • Omland T.
        • Kullo I.J.
        Biomarker-based risk models to risk stratify patients with stable coronary heart disease.
        J. Am. Coll. Cardiol. 2017; 70: 827-829
        • Lehmann N.
        • Erbel R.
        • Mahabadi A.A.
        • Rauwolf M.
        • Mohlenkamp S.
        • Moebus S.
        • Kalsch H.
        • Budde T.
        • Schmermund A.
        • Stang A.
        • Fuhrer-Sakel D.
        • Weimar C.
        • Roggenbuck U.
        • Dragano N.
        • Jockel K.H.
        Value of progression of coronary artery calcification for risk prediction of coronary and cardiovascular events: result of the HNR study (heinz nixdorf recall).
        Circulation. 2018; 137: 665-679
        • Williams M.
        • Shaw L.J.
        • Raggi P.
        • Morris D.
        • Vaccarino V.
        • Liu S.T.
        • Weinstein S.R.
        • Mosler T.P.
        • Tseng P.H.
        • Flores F.R.
        • Nasir K.
        • Budoff M.
        Prognostic value of number and site of calcified coronary lesions compared with the total score.
        JACC Cardiovasc Imaging. 2008; 1: 61-69
        • Owens D.S.
        • Budoff M.J.
        • Katz R.
        • Takasu J.
        • Shavelle D.M.
        • Carr J.J.
        • Heckbert S.R.
        • Otto C.M.
        • Probstfield J.L.
        • Kronmal R.A.
        • O'Brien K.D.
        Aortic valve calcium independently predicts coronary and cardiovascular events in a primary prevention population.
        JACC (J. Am. Coll. Cardiol.): Cardiovascular Imaging. 2012; 5: 619-625
        • Goeller M.
        • Achenbach S.
        • Marwan M.
        • Doris M.K.
        • Cadet S.
        • Commandeur F.
        • Chen X.
        • Slomka P.J.
        • Gransar H.
        • Cao J.J.
        • Wong N.D.
        • Albrecht M.H.
        • Rozanski A.
        • Tamarappoo B.K.
        • Berman D.S.
        • Dey D.
        Epicardial adipose tissue density and volume are related to subclinical atherosclerosis, inflammation and major adverse cardiac events in asymptomatic subjects.
        J Cardiovasc Comput Tomogr. 2018; 12: 67-73
        • Schnabel Renate B.
        • Yin X.
        • Larson Martin G.
        • Yamamoto Jennifer F.
        • Fontes João D.
        • Kathiresan S.
        • Rong J.
        • Levy D.
        • Keaney John F.
        • Wang Thomas J.
        • Murabito Joanne M.
        • Vasan Ramachandran S.
        • Benjamin Emelia J.
        Multiple inflammatory biomarkers in relation to cardiovascular events and mortality in the community.
        Arterioscler. Thromb. Vasc. Biol. 2013; 33: 1728-1733
        • van Holten T.C.
        • Waanders L.F.
        • de Groot P.G.
        • Vissers J.
        • Hoefer I.E.
        • Pasterkamp G.
        • Prins M.W.J.
        • Roest M.
        Circulating biomarkers for predicting cardiovascular disease risk; a systematic review and comprehensive overview of meta-analyses.
        PloS One. 2013; 8 (e62080-e62080)
        • Rana J.S.
        • Gransar H.
        • Wong N.D.
        • Shaw L.
        • Pencina M.
        • Nasir K.
        • Rozanski A.
        • Hayes S.W.
        • Thomson L.E.
        • Friedman J.D.
        • Min J.K.
        • Berman D.S.
        Comparative value of coronary artery calcium and multiple blood biomarkers for prognostication of cardiovascular events.
        Am. J. Cardiol. 2012; 109: 1449-1453
        • Wang T.J.
        • Gona P.
        • Larson M.G.
        • Tofler G.H.
        • Levy D.
        • Newton-Cheh C.
        • Jacques P.F.
        • Rifai N.
        • Selhub J.
        • Robins S.J.
        • Benjamin E.J.
        • D'Agostino R.B.
        • Vasan R.S.
        Multiple biomarkers for the prediction of first major cardiovascular events and death.
        N. Engl. J. Med. 2006; 355: 2631-2639
        • Folsom A.R.
        • Chambless L.E.
        • Ballantyne C.M.
        • Coresh J.
        • Heiss G.
        • Wu K.K.
        • Boerwinkle E.
        • Mosley Jr., T.H.
        • Sorlie P.
        • Diao G.
        • Sharrett A.R.
        An assessment of incremental coronary risk prediction using C-reactive protein and other novel risk markers: the atherosclerosis risk in communities study.
        Arch. Intern. Med. 2006; 166: 1368-1373
        • Ference B.A.
        • Ginsberg H.N.
        • Graham I.
        • Ray K.K.
        • Packard C.J.
        • Bruckert E.
        • Hegele R.A.
        • Krauss R.M.
        • Raal F.J.
        • Schunkert H.
        • Watts G.F.
        • Borén J.
        • Fazio S.
        • Horton J.D.
        • Masana L.
        • Nicholls S.J.
        • Nordestgaard B.G.
        • van de Sluis B.
        • Taskinen M.-R.
        • Tokgözoğlu L.
        • Landmesser U.
        • Laufs U.
        • Wiklund O.
        • Stock J.K.
        • Chapman M.J.
        • Catapano A.L.
        Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel.
        Eur. Heart J. 2017; 38: 2459-2472
        • Lowe G.D.
        • Yarnell J.W.
        • Sweetnam P.M.
        • Rumley A.
        • Thomas H.F.
        • Elwood P.C.
        Fibrin D-dimer, tissue plasminogen activator, plasminogen activator inhibitor, and the risk of major ischaemic heart disease in the Caerphilly Study.
        Thromb. Haemostasis. 1998; 79: 129-133
        • Blankenberg S.
        • Rupprecht H.J.
        • Poirier O.
        • Bickel C.
        • Smieja M.
        • Hafner G.
        • Meyer J.
        • Cambien F.
        • Tiret L.
        • AtheroGene I.
        Plasma concentrations and genetic variation of matrix metalloproteinase 9 and prognosis of patients with cardiovascular disease.
        Circulation. 2003; 107: 1579-1585
        • Morishita T.
        • Uzui H.
        • Mitsuke Y.
        • Amaya N.
        • Kaseno K.
        • Ishida K.
        • Fukuoka Y.
        • Ikeda H.
        • Tama N.
        • Yamazaki T.
        • Lee J.D.
        • Tada H.
        Association between matrix metalloproteinase-9 and worsening heart failure events in patients with chronic heart failure.
        ESC Heart Fail. 2017; 4: 321-330
        • Soeki T.
        • Niki T.
        • Kusunose K.
        • Bando S.
        • Hirata Y.
        • Tomita N.
        • Yamaguchi K.
        • Koshiba K.
        • Yagi S.
        • Taketani Y.
        • Iwase T.
        • Yamada H.
        • Wakatsuki T.
        • Akaike M.
        • Sata M.
        Elevated concentrations of pentraxin 3 are associated with coronary plaque vulnerability.
        J. Cardiol. 2011; 58: 151-157
        • Latini R.
        • Gullestad L.
        • Masson S.
        • Nymo S.H.
        • Ueland T.
        • Cuccovillo I.
        • Vårdal M.
        • Bottazzi B.
        • Mantovani A.
        • Lucci D.
        • Masuda N.
        • Sudo Y.
        • Wikstrand J.
        • Tognoni G.
        • Aukrust P.
        • Tavazzi L.
        On behalf of the Investigators of the Controlled Rosuvastatin Multinational Trial in Heart F and trials GI-HF. Pentraxin-3 in chronic heart failure: the CORONA and GISSI-HF trials.
        Eur. J. Heart Fail. 2012; 14: 992-999
        • Eisen A.
        • Kornowski R.
        • Hamdan A.
        • Talmor-Barkan Y.
        • Witberg G.
        • Deshpande K.
        • Paixao A.
        • Ayers C.
        • Joshi P.
        • Rohatgi A.
        • Khera A.
        • Lemos J.
        • Neeland I.
        Novel biomarkers OF coronary artery calcium incidence or progression: insights from the dallas heart study.
        J. Am. Coll. Cardiol. 2019; 73: 1783
        • Cotter G.
        • Voors A.A.
        • Prescott M.F.
        • Felker G.M.
        • Filippatos G.
        • Greenberg B.H.
        • Pang P.S.
        • Ponikowski P.
        • Milo O.
        • Hua T.A.
        • Qian M.
        • Severin T.M.
        • Teerlink J.R.
        • Metra M.
        • Davison B.A.
        Growth differentiation factor 15 (GDF-15) in patients admitted for acute heart failure: results from the RELAX-AHF study.
        Eur. J. Heart Fail. 2015; 17: 1133-1143
        • Kim J.-H.
        Estimating classification error rate: repeated cross-validation, repeated hold-out and bootstrap.
        Comput. Stat. Data Anal. 2009; 53: 3735-3745

      Linked Article

      • Machine learning to predict cardiac events in asymptomatic individuals
        AtherosclerosisVol. 318
        • Preview
          Identifying which patients are at risk of cardiac events is an important goal and enables the application of interventions to prevent future cardiovascular events. Current cardiovascular risk scores incorporate clinical features and biochemical measures to stratify patients into risk groups [1,2]. However, they have a number of limitations, including differences when applied to populations in different countries [3], overestimation of risk in contemporary multi-ethnic cohorts [4] and underestimation of risk in women [5].
        • Full-Text
        • PDF