TY - JOUR
AU - Bøttcher,, Morten
AB - Abstract Aims European and North American guidelines currently recommend pre-test probability (PTP) stratification based on simple probability models in patients with suspected coronary artery disease (CAD). However, no unequivocal recommendation has yet been established. We aimed to compare the ability of risk factors and different PTP stratification models to predict haemodynamically obstructive CAD with fractional flow reserve (FFR) as reference in low to intermediate probability patients. Methods and results We prospectively included 1675 patients with low to intermediate risk who had been referred to coronary computed tomography angiography (CTA). Patients with coronary stenosis were subsequently investigated by invasive coronary angiography (ICA) with FFR measurement if indicated. Discrimination and calibration were assessed for four models: the updated Diamond–Forrester (UDF), the CAD Consortium Basic, the Clinical, and the Clinical + Coronary artery calcium score (CACS). At coronary CTA, 24% of patients were diagnosed with a suspected stenosis and 10% had haemodynamically obstructive CAD at the ICA. Calibration for all CAD Consortium models increased compared with the UDF score. However, all models overestimated the probability of haemodynamically obstructive CAD. Discrimination increased by area under the receiver operating curve from 67% to 86% for UDF vs. CAD Consortium Clinical + CACS. The proportion of low-probability patients (pre-test score < 15%) was for the UDF, CAD Consortium Basic, Clinical, and Clinical + CACS: 14%, 58%, 51%, and 66%, respectively. The corresponding negative predictive values were 97%, 94%, 95%, and 98%, respectively. Conclusion CAD Consortium models improve PTP stratification compared with the UDF score, mainly due to superior calibration in low to intermediate probability patients. Adding the coronary calcium score to the models substantially increases discrimination. Clinical Trials. gov identifier NCT02264717. risk factors, risk stratification, coronary computed tomography angiography, coronary angiography, fractional flow reserve Introduction Pre-test probability (PTP) stratification of patients with symptoms suggestive of obstructive coronary artery disease (CAD) remains a challenge. Consequently, initial diagnostic testing is often normal. Several PTP stratification models have been suggested as guidance to whether cardiac testing should be deferred or conducted non-invasively or invasively. Both European and North American guidelines currently recommend PTP stratification based on PTP models, but no unequivocal recommendation has yet been established. The American Heart Association/American College of Cardiology (AHA/ACC) guidelines recommend using the Diamond–Forrester score including the Coronary Artery Surgery Study data model or the Duke Clinical score, which were introduced in 1979 and 1983, respectively.1–3 The Diamond–Forrester score is a simple model including only age, gender, and angina typicality, while the Duke Clinical score adds further risk factors like smoking, dyslipidaemia, diabetes, and electrocardiogram (ECG) changes. In 2011, the Diamond–Forrester score was recalibrated to a new version—the updated Diamond–Forrester score (UDF score) which is currently recommended by the European Society of Cardiology (ESC).4,5 However, all these models have been developed in high-probability patients and have been shown to overestimate the PTP of CAD when used in contemporary patient cohorts being considered for non-invasive testing. In 2012, an international collaborative network, coined the ‘CAD Consortium’, and developed and validated three scores in patients from 18 different hospitals across Europe and the USA: (i) a ‘Basic model’, which is a recalibration of the UDF score; (ii) a ‘Clinical model’ which expands the Basic model with cardiovascular risk factors including diabetes, hypertension, dyslipidaemia, and smoking; and (iii) finally, a ‘Clinical + CACS model’ which included the coronary artery calcium score (CACS) into the Clinical model.6 In contrast to previous models, the CAD Consortium scores have been developed not only in high-probability patients referred to invasive coronary angiography (ICA), but also in low to intermediate probability patient referred to coronary computed tomography angiography (CTA). Both the former and the more recent models have been developed and validated against anatomical stenosis defined as luminal diameter reduction on an ICA or a coronary CTA. However, ICA with fractional flow reserve measurements (ICA-FFR) has been established as the reference standard for lesion-specific ischaemia and revascularization. Previous studies have demonstrated that a substantial proportion of 30–50% and 50–70% diameter stenosis are reclassified as haemodynamically obstructive or non-haemodynamically obstructive, using ICA-FFR as reference.7 It seems reasonable to hypothesize that the predictive value of PTP models might be affected by these reclassifications because symptoms of angina pectoris are driven by myocardial ischaemia and not by anatomical measurements of stenosis severity. Nonetheless, no PTP model has been validated using ICA-FFR as the reference standard. The aim of this study was therefore to investigate the ability of risk factors and PTP stratification models to predict haemodynamically obstructive CAD using ICA-FFR as a reference. Methods Study design In total, 1675 patients without known CAD referred to coronary CTA as first-line diagnostic test due to a history of symptoms suggestive of CAD were consecutively included in a multi-centre cohort trial—the Danish study of Non-Invasive testing in Coronary Artery Disease (Dan-NICAD).8,9 All patients had a systematic interview to assess risk factors and symptoms, and, based on this information and medical record reviews, the PTP stratification models were calculated (Supplementary data online, Table S1). The pre-test probabilities (PTP) of obstructive coronary disease of the UDF and the CAD consortium Basis, Clinical, and Clinical + CACS models for patients aged 35, 45, 55, 65, 75, and 85 years, with none or all risk factors, and with a CACS of 0, 10, 100, or 400 are illustrated in the Supplementary data online, Figure S1. Subsequently, a coronary CTA was performed. Patients with ≥50% diameter coronary stenosis at coronary CTA were randomized 1:1 to cardiac magnetic resonance imaging or single-photon emission computed tomography (CT) myocardial perfusion imaging (MPI) and finally underwent ICA with ICA-FFR measurement. Patients with <50% diameter stenosis at coronary CTA were not investigated further in the study (Figure 1). The study design and imaging protocols have previously been published.8 Figure 1 Open in new tabDownload slide Flow chart of patients included in the study. CACS, coronary artery calcium score; CAD, coronary artery disease; CTA, coronary computed tomography angiography; ICA, invasive coronary angiogram; ICA-FFR, invasive coronary angiogram with fractional flow reserve; MPI, myocardial perfusion imaging. Figure 1 Open in new tabDownload slide Flow chart of patients included in the study. CACS, coronary artery calcium score; CAD, coronary artery disease; CTA, coronary computed tomography angiography; ICA, invasive coronary angiogram; ICA-FFR, invasive coronary angiogram with fractional flow reserve; MPI, myocardial perfusion imaging. The exclusion criteria were (i) age <40; (ii) previous coronary revascularization or myocardial infarction; (iii) unstable angina pectoris; (iv) estimated glomerular filtration rate <40 mL/min; (v) pregnancy; and (vi) contraindication for iodine-containing contrast medium, magnetic resonance imaging, or adenosine (severe asthma, advanced atrioventricular block, or critical aortic stenosis). The study was approved by The Danish Data Protection Agency and The Central Denmark Region Committees on Health Research Ethics. All patients signed a written informed consent form. CACS and coronary CTA acquisition and interpretation All scans were performed on a 320-slice volume CT scanner (Aquilion One, Toshiba Medical Systems, Japan) in accordance with usual clinical guidelines. Imaging analysis included an Agatston CACS and visual evaluation of luminal diameter stenosis estimation in each coronary segment. Stenosis severity was classified from diameter stenosis reduction in all segments with a reference vessel diameter >2 mm as no CAD: 0%; mild to moderate CAD: >0 to <50%; and severe CAD: ≥50% diameter stenosis reduction. ICA and ICA-FFR acquisition and interpretation Approximately 4 weeks after referral to coronary CTA, ICA was performed in patients with a suspected severe CAD at coronary CTA. ICA-FFR was performed in all segments with a 30–90% diameter stenosis at the ICA. 2D quantitative coronary analysis (QCA) was performed in an independent core lab (ClinFact, Leiden, The Netherlands). Haemodynamically obstructive CAD was defined as high-grade stenosis by visual assessment (>90% diameter stenosis) or by an ICA-FFR ≤0.80 in a vessel ≥2.0 mm diameter. If ICA-FFR was indicated but not technically possible, QCA-based stenosis assessment was used with a cut-off of ≥50% diameter reduction. All other patients, including patient without severe CAD at coronary CTA were classified as having no haemodynamically obstructive CAD. Statistical analysis Continuous variables were presented as mean (± standard deviation) and categorical variables as n (%). Correlations were evaluated with Spearman’s rho. Calibration and discrimination of the PTP models were evaluated according to previous recommendation.10 First, a calibration plot of the mean predicted probability and the mean observed proportion of obstructive CAD with flexible calibration (Loess bandwidth 0.8) were illustrated and evaluated. Perfect predictions should be on the ideal line in the calibration plot, statistically described with an intercept alpha of 0 (‘calibration-in-the-large’) and slope beta of 1 (‘calibration slope’). The discrimination C-statistic included the area under the receiver operating characteristic curve (AUC). Diagnostic accuracy was evaluated using sensitivity, specificity, positive and negative predictive values (PPV and NPV), and likelihood ratios. Patients with missing CTA or ICA data were excluded from analysis of these specific endpoints. Statistical analysis was performed using STATA-15 (StataCorp, College Station, TX, USA). Results Of the 1675 patients included, 1653 (99%) completed the coronary CTA. At coronary CTA, 391 (24%) were diagnosed with suspected severe CAD and referred to ICA with ICA-FFR measurements. Of the referred patients, 362 (93%) completed the ICA and 160 (44%) were diagnosed with haemodynamically obstructive CAD (Figure 1). The mean duration from inclusion to the ICA was 33 ± 20 days. Patient demographics and imaging characteristics are presented in Tables 1 and 2. Data from patients with missing ICA are presented in Supplementary data online, Table S2. Table 1 Patient demographics Haemodynamically obstructive CADa Total No Yes 1657 1464 160 Characteristics  Race, Caucasian 1644 (99.2) 1452 (99.2) 159 (99.4)  Sex, male 802 (48.4) 665 (45.4) 112 (70.0)  Age (years) 57.2 ± 8.8 56.8 ± 8.7 60.1 ± 9.1  Genetic pre-dispositionb 613 (37.0) 537 (36.7) 65 (40.6)  Body mass index (kg/m²) 26.8 ± 4.3 26.7 ± 4.2 27.2 ± 4.5  Smoking   Never 781 (47.2) 708 (48.4) 64 (40.0)   Former 611 (36.9) 532 (36.4) 64 (40.0)   Active 264 (15.9) 223 (15.2) 32 (20.0)  Heart rate 65.6 ± 11.2 65.5 ± 11.3 66.3 ± 10.8  Blood pressure   Systolic 138 ± 19 138 ± 19 145 ± 20   Diastolic 83 ± 11 83 ± 11 85 ± 12 Cardiac symptoms  Chest pain type   Non-specific 641 (38.7) 616 (39.6) 49 (30.6)   Atypical chest pain 562 (33.9) 503 (34.4) 50 (31.3)   Typical chest pain 454 (27.4) 381 (26.0) 61 (38.1)  Chest pain strongest location   Retrosternal 725 (43.8) 629 (43.0) 79 (49.4)   Let side of chest 393 (23.8) 353 (24.1) 36 (22.5)   Other location 203 (11.7) 173 (11.8) 18 (11.3)   No pain (dyspnoea or arrhythmia) 345 (20.7) 309 (21.1) 27 (16.9)  Chest pain alleviated by rest or nitroglycerinec 559 (42.6) 470 (40.7) 80 (60.1)  Chest pain triggered by physical or mental effortc 636 (48.1) 547 (47.4) 80 (60.1) Medically treated  Diabetes 93 (5.6) 68 (4.7) 20 (12.5)  Hypertension 578 (34.9) 489 (33.4) 578 (34.9)   Mean number of antihypertensive 1.9 ± 1.0 1.8 ± 1.0 2.1 ± 1.1  Hypercholesterolaemia 388 (23.3) 329 (22.9) 53 (33.3) Biochemistryd  Cholesterol (total, mmol/L) 5.4 ± 1.1 5.4 ± 1.0 5.5 ± 1.5  Glucose (fasting, mmol/L) 5.9 ± 1.1 5.9 ± 1.0 6.2 ± 1.1  HbA1c (mmol/mol) 37.9 ± 7.4 37.5 ± 6.8 40.9 ± 10.9  Creatinine (mmol/L) 75.5 ± 14.5 74.8 ± 14.0 79.7 ± 16.3 Haemodynamically obstructive CADa Total No Yes 1657 1464 160 Characteristics  Race, Caucasian 1644 (99.2) 1452 (99.2) 159 (99.4)  Sex, male 802 (48.4) 665 (45.4) 112 (70.0)  Age (years) 57.2 ± 8.8 56.8 ± 8.7 60.1 ± 9.1  Genetic pre-dispositionb 613 (37.0) 537 (36.7) 65 (40.6)  Body mass index (kg/m²) 26.8 ± 4.3 26.7 ± 4.2 27.2 ± 4.5  Smoking   Never 781 (47.2) 708 (48.4) 64 (40.0)   Former 611 (36.9) 532 (36.4) 64 (40.0)   Active 264 (15.9) 223 (15.2) 32 (20.0)  Heart rate 65.6 ± 11.2 65.5 ± 11.3 66.3 ± 10.8  Blood pressure   Systolic 138 ± 19 138 ± 19 145 ± 20   Diastolic 83 ± 11 83 ± 11 85 ± 12 Cardiac symptoms  Chest pain type   Non-specific 641 (38.7) 616 (39.6) 49 (30.6)   Atypical chest pain 562 (33.9) 503 (34.4) 50 (31.3)   Typical chest pain 454 (27.4) 381 (26.0) 61 (38.1)  Chest pain strongest location   Retrosternal 725 (43.8) 629 (43.0) 79 (49.4)   Let side of chest 393 (23.8) 353 (24.1) 36 (22.5)   Other location 203 (11.7) 173 (11.8) 18 (11.3)   No pain (dyspnoea or arrhythmia) 345 (20.7) 309 (21.1) 27 (16.9)  Chest pain alleviated by rest or nitroglycerinec 559 (42.6) 470 (40.7) 80 (60.1)  Chest pain triggered by physical or mental effortc 636 (48.1) 547 (47.4) 80 (60.1) Medically treated  Diabetes 93 (5.6) 68 (4.7) 20 (12.5)  Hypertension 578 (34.9) 489 (33.4) 578 (34.9)   Mean number of antihypertensive 1.9 ± 1.0 1.8 ± 1.0 2.1 ± 1.1  Hypercholesterolaemia 388 (23.3) 329 (22.9) 53 (33.3) Biochemistryd  Cholesterol (total, mmol/L) 5.4 ± 1.1 5.4 ± 1.0 5.5 ± 1.5  Glucose (fasting, mmol/L) 5.9 ± 1.1 5.9 ± 1.0 6.2 ± 1.1  HbA1c (mmol/mol) 37.9 ± 7.4 37.5 ± 6.8 40.9 ± 10.9  Creatinine (mmol/L) 75.5 ± 14.5 74.8 ± 14.0 79.7 ± 16.3 Values are presented as n (%) or mean ± SD or (range). a Haemodynamic disease status were missing in 33 patients due to dropout between coronary CTA and ICA. See these patients demographic in the Supplementary data online. b CAD among first degree relatives aged <60 years. c Of patients with any chest pain. d Data available in cholesterol 94%, glucose fasting 8%, HbA1c 78%, and creatinine 99% of the cohort. Open in new tab Table 1 Patient demographics Haemodynamically obstructive CADa Total No Yes 1657 1464 160 Characteristics  Race, Caucasian 1644 (99.2) 1452 (99.2) 159 (99.4)  Sex, male 802 (48.4) 665 (45.4) 112 (70.0)  Age (years) 57.2 ± 8.8 56.8 ± 8.7 60.1 ± 9.1  Genetic pre-dispositionb 613 (37.0) 537 (36.7) 65 (40.6)  Body mass index (kg/m²) 26.8 ± 4.3 26.7 ± 4.2 27.2 ± 4.5  Smoking   Never 781 (47.2) 708 (48.4) 64 (40.0)   Former 611 (36.9) 532 (36.4) 64 (40.0)   Active 264 (15.9) 223 (15.2) 32 (20.0)  Heart rate 65.6 ± 11.2 65.5 ± 11.3 66.3 ± 10.8  Blood pressure   Systolic 138 ± 19 138 ± 19 145 ± 20   Diastolic 83 ± 11 83 ± 11 85 ± 12 Cardiac symptoms  Chest pain type   Non-specific 641 (38.7) 616 (39.6) 49 (30.6)   Atypical chest pain 562 (33.9) 503 (34.4) 50 (31.3)   Typical chest pain 454 (27.4) 381 (26.0) 61 (38.1)  Chest pain strongest location   Retrosternal 725 (43.8) 629 (43.0) 79 (49.4)   Let side of chest 393 (23.8) 353 (24.1) 36 (22.5)   Other location 203 (11.7) 173 (11.8) 18 (11.3)   No pain (dyspnoea or arrhythmia) 345 (20.7) 309 (21.1) 27 (16.9)  Chest pain alleviated by rest or nitroglycerinec 559 (42.6) 470 (40.7) 80 (60.1)  Chest pain triggered by physical or mental effortc 636 (48.1) 547 (47.4) 80 (60.1) Medically treated  Diabetes 93 (5.6) 68 (4.7) 20 (12.5)  Hypertension 578 (34.9) 489 (33.4) 578 (34.9)   Mean number of antihypertensive 1.9 ± 1.0 1.8 ± 1.0 2.1 ± 1.1  Hypercholesterolaemia 388 (23.3) 329 (22.9) 53 (33.3) Biochemistryd  Cholesterol (total, mmol/L) 5.4 ± 1.1 5.4 ± 1.0 5.5 ± 1.5  Glucose (fasting, mmol/L) 5.9 ± 1.1 5.9 ± 1.0 6.2 ± 1.1  HbA1c (mmol/mol) 37.9 ± 7.4 37.5 ± 6.8 40.9 ± 10.9  Creatinine (mmol/L) 75.5 ± 14.5 74.8 ± 14.0 79.7 ± 16.3 Haemodynamically obstructive CADa Total No Yes 1657 1464 160 Characteristics  Race, Caucasian 1644 (99.2) 1452 (99.2) 159 (99.4)  Sex, male 802 (48.4) 665 (45.4) 112 (70.0)  Age (years) 57.2 ± 8.8 56.8 ± 8.7 60.1 ± 9.1  Genetic pre-dispositionb 613 (37.0) 537 (36.7) 65 (40.6)  Body mass index (kg/m²) 26.8 ± 4.3 26.7 ± 4.2 27.2 ± 4.5  Smoking   Never 781 (47.2) 708 (48.4) 64 (40.0)   Former 611 (36.9) 532 (36.4) 64 (40.0)   Active 264 (15.9) 223 (15.2) 32 (20.0)  Heart rate 65.6 ± 11.2 65.5 ± 11.3 66.3 ± 10.8  Blood pressure   Systolic 138 ± 19 138 ± 19 145 ± 20   Diastolic 83 ± 11 83 ± 11 85 ± 12 Cardiac symptoms  Chest pain type   Non-specific 641 (38.7) 616 (39.6) 49 (30.6)   Atypical chest pain 562 (33.9) 503 (34.4) 50 (31.3)   Typical chest pain 454 (27.4) 381 (26.0) 61 (38.1)  Chest pain strongest location   Retrosternal 725 (43.8) 629 (43.0) 79 (49.4)   Let side of chest 393 (23.8) 353 (24.1) 36 (22.5)   Other location 203 (11.7) 173 (11.8) 18 (11.3)   No pain (dyspnoea or arrhythmia) 345 (20.7) 309 (21.1) 27 (16.9)  Chest pain alleviated by rest or nitroglycerinec 559 (42.6) 470 (40.7) 80 (60.1)  Chest pain triggered by physical or mental effortc 636 (48.1) 547 (47.4) 80 (60.1) Medically treated  Diabetes 93 (5.6) 68 (4.7) 20 (12.5)  Hypertension 578 (34.9) 489 (33.4) 578 (34.9)   Mean number of antihypertensive 1.9 ± 1.0 1.8 ± 1.0 2.1 ± 1.1  Hypercholesterolaemia 388 (23.3) 329 (22.9) 53 (33.3) Biochemistryd  Cholesterol (total, mmol/L) 5.4 ± 1.1 5.4 ± 1.0 5.5 ± 1.5  Glucose (fasting, mmol/L) 5.9 ± 1.1 5.9 ± 1.0 6.2 ± 1.1  HbA1c (mmol/mol) 37.9 ± 7.4 37.5 ± 6.8 40.9 ± 10.9  Creatinine (mmol/L) 75.5 ± 14.5 74.8 ± 14.0 79.7 ± 16.3 Values are presented as n (%) or mean ± SD or (range). a Haemodynamic disease status were missing in 33 patients due to dropout between coronary CTA and ICA. See these patients demographic in the Supplementary data online. b CAD among first degree relatives aged <60 years. c Of patients with any chest pain. d Data available in cholesterol 94%, glucose fasting 8%, HbA1c 78%, and creatinine 99% of the cohort. Open in new tab Table 2 Imaging study characteristics Coronary artery calcium score 0 (0–82)  =0 841 (50.7)  >0–399 646 (40.0)  ≥400 170 (10.3) Coronary CTA  Disease severitya   None 783 (47.4)   Mild to moderate 479 (28.9)   Severe 391 (23.7)  Segment stenosis score 1 (0–4)  Abnormal test result   1-Vessel disease 218 (13.1)   2-Vessel disease 93 (5.6)   3-Vessel disease or left main 80 (4.8) ICA quantitative analysis  Stenosis degree (%)   Diameter stenosis <30 83 (22.9)   Diameter stenosis 30–49 109 (30.1)   Diameter stenosis 50–69 99 (26.2)   Diameter stenosis 70–89 45 (12.4)   Diameter stenosis ≥90 30 (8.3) FFRb  ICA-FFR 0.84 (0.75–0.88)  ICA-FFR in patient with ICA-FFR >0.80 0.87 (0.84–0.91)  ICA-FFR in patient with ICA-FFR ≤0.80 0.73 (0.68–0.77) Haemodynamically obstructive CAD at ICAc  Patients with stenosis 160 (9.6)   1-Vessel disease 97 (5.8)   2-Vessel disease 30 (2.5)   3-Vessel disease or left main 30 (1.3) Coronary artery calcium score 0 (0–82)  =0 841 (50.7)  >0–399 646 (40.0)  ≥400 170 (10.3) Coronary CTA  Disease severitya   None 783 (47.4)   Mild to moderate 479 (28.9)   Severe 391 (23.7)  Segment stenosis score 1 (0–4)  Abnormal test result   1-Vessel disease 218 (13.1)   2-Vessel disease 93 (5.6)   3-Vessel disease or left main 80 (4.8) ICA quantitative analysis  Stenosis degree (%)   Diameter stenosis <30 83 (22.9)   Diameter stenosis 30–49 109 (30.1)   Diameter stenosis 50–69 99 (26.2)   Diameter stenosis 70–89 45 (12.4)   Diameter stenosis ≥90 30 (8.3) FFRb  ICA-FFR 0.84 (0.75–0.88)  ICA-FFR in patient with ICA-FFR >0.80 0.87 (0.84–0.91)  ICA-FFR in patient with ICA-FFR ≤0.80 0.73 (0.68–0.77) Haemodynamically obstructive CAD at ICAc  Patients with stenosis 160 (9.6)   1-Vessel disease 97 (5.8)   2-Vessel disease 30 (2.5)   3-Vessel disease or left main 30 (1.3) Values are presented as n (%), mean ± SD, or median (interquartile range). a No stenosis, mild to moderate: diameter stenosis >0 to <50%, and severe: diameter stenosis ≥50%. b FFR is measured in patients with a visual stenosis of between 30% and 90%. cDefined as visual high-grade stenosis (>90% diameter stenosis), an ICA-FFR ≤ 0.80 or an diameter stenosis >50% by quantitative coronary analysis if ICA-FFR was indicated but not technically possible. Open in new tab Table 2 Imaging study characteristics Coronary artery calcium score 0 (0–82)  =0 841 (50.7)  >0–399 646 (40.0)  ≥400 170 (10.3) Coronary CTA  Disease severitya   None 783 (47.4)   Mild to moderate 479 (28.9)   Severe 391 (23.7)  Segment stenosis score 1 (0–4)  Abnormal test result   1-Vessel disease 218 (13.1)   2-Vessel disease 93 (5.6)   3-Vessel disease or left main 80 (4.8) ICA quantitative analysis  Stenosis degree (%)   Diameter stenosis <30 83 (22.9)   Diameter stenosis 30–49 109 (30.1)   Diameter stenosis 50–69 99 (26.2)   Diameter stenosis 70–89 45 (12.4)   Diameter stenosis ≥90 30 (8.3) FFRb  ICA-FFR 0.84 (0.75–0.88)  ICA-FFR in patient with ICA-FFR >0.80 0.87 (0.84–0.91)  ICA-FFR in patient with ICA-FFR ≤0.80 0.73 (0.68–0.77) Haemodynamically obstructive CAD at ICAc  Patients with stenosis 160 (9.6)   1-Vessel disease 97 (5.8)   2-Vessel disease 30 (2.5)   3-Vessel disease or left main 30 (1.3) Coronary artery calcium score 0 (0–82)  =0 841 (50.7)  >0–399 646 (40.0)  ≥400 170 (10.3) Coronary CTA  Disease severitya   None 783 (47.4)   Mild to moderate 479 (28.9)   Severe 391 (23.7)  Segment stenosis score 1 (0–4)  Abnormal test result   1-Vessel disease 218 (13.1)   2-Vessel disease 93 (5.6)   3-Vessel disease or left main 80 (4.8) ICA quantitative analysis  Stenosis degree (%)   Diameter stenosis <30 83 (22.9)   Diameter stenosis 30–49 109 (30.1)   Diameter stenosis 50–69 99 (26.2)   Diameter stenosis 70–89 45 (12.4)   Diameter stenosis ≥90 30 (8.3) FFRb  ICA-FFR 0.84 (0.75–0.88)  ICA-FFR in patient with ICA-FFR >0.80 0.87 (0.84–0.91)  ICA-FFR in patient with ICA-FFR ≤0.80 0.73 (0.68–0.77) Haemodynamically obstructive CAD at ICAc  Patients with stenosis 160 (9.6)   1-Vessel disease 97 (5.8)   2-Vessel disease 30 (2.5)   3-Vessel disease or left main 30 (1.3) Values are presented as n (%), mean ± SD, or median (interquartile range). a No stenosis, mild to moderate: diameter stenosis >0 to <50%, and severe: diameter stenosis ≥50%. b FFR is measured in patients with a visual stenosis of between 30% and 90%. cDefined as visual high-grade stenosis (>90% diameter stenosis), an ICA-FFR ≤ 0.80 or an diameter stenosis >50% by quantitative coronary analysis if ICA-FFR was indicated but not technically possible. Open in new tab The relative risk ratio for haemodynamically obstructive CAD for each of the baseline risk factors is illustrated in Figure 2, and the exact values and confidence interval are available in Supplementary data online, Table S3. Figure 2 Open in new tabDownload slide The relative probability ratio of baseline characteristics, risk factors, symptoms, ECG and coronary calcium score for the presence of a haemodynamically obstructive CAD at ICA. A base-10 log scale is used for the Y-axis. Values and confidence interval are presented in Supplementary data online, Table S3. References for each category are age <50, body mass index <25, chest pain typicality = non-cardiac, coronary artery calcium score = 0. ECG is categorized as abnormal if pathological Q-waves or ST-depression were present as defined in the Duke Clinical score. Figure 2 Open in new tabDownload slide The relative probability ratio of baseline characteristics, risk factors, symptoms, ECG and coronary calcium score for the presence of a haemodynamically obstructive CAD at ICA. A base-10 log scale is used for the Y-axis. Values and confidence interval are presented in Supplementary data online, Table S3. References for each category are age <50, body mass index <25, chest pain typicality = non-cardiac, coronary artery calcium score = 0. ECG is categorized as abnormal if pathological Q-waves or ST-depression were present as defined in the Duke Clinical score. The distribution of the PTP scores according to patients with no, mild to moderate, and severe CAD at coronary CTA and haemodynamically obstructive CAD at ICA are listed in Table 3 and illustrated in Figure 3 and in Supplementary data online, Figure S2. Figure 3 Open in new tabDownload slide Distribution of the patients (n = 1653) divided by the present of CAD severity at coronary CTA (left) and presence of haemodynamically obstructive CAD at ICA (right) according to the categories of the four pre-test probability stratification models (A-H). Figure 3 Open in new tabDownload slide Distribution of the patients (n = 1653) divided by the present of CAD severity at coronary CTA (left) and presence of haemodynamically obstructive CAD at ICA (right) according to the categories of the four pre-test probability stratification models (A-H). Table 3 Pre-test probability models calibration and discrimination analysis Reference Diamond–Forrester score (updated) CAD Consortium Basic Clinical Clinical + CACS Median value Median pre-test probability models value  Total cohort 36 (20–53) 12 (6–25) 13 (6–28) 6 (2–26)  No CAD CTAa 29 (16–44) 9 (4–17) 9 (4–17) 2 (1–4)  Mild/moderate CAD CTAa 41 (24–55) 15 (8–26) 16 (9–30) 16 (6–30)  Severe CAD CTAa 46 (29–63) 20 (9–36) 24 (11–40) 37 (19–61)  Non-haemodynamically obstructive ICAb 35 (19–51) 12 (6–22) 12 (6–25) 4 (1–17)  Haemodynamically obstructive CAD ICAb 50 (34–71) 23 (11–39) 30 (13–47) 47 (24–72) Calibration Calibration-in-the-largec CTAa −0.89 0.42 0.30 0.65 ICAb −2.09 −0.80 −0.95 −0.87 Calibration slopec CTAa 0.51 0.52 0.57 0.94 ICAb 0.60 0.57 0.60 0.84 Discrimination Area under the receiver operating characteristic curve CTAa 65 (61–68) 66 (63–69) 69 (66–72) 87 (85–89) ICAb 67 (63–72) 68 (63–72) 70 (66–74) 86 (83–89) Pre-test probability cut-off <5  Number of patients below cut-off 0 310 (19) 310 (19) 781 (48)  Number of diseased patients below cut-offd 0 11 (4) 9 (3) 10 (1)  Sensitivity ICAb NA 93 (88–97) 94 (90–97) 94 (89–97)  Specificity ICAb NA 20 (18–23) 21 (19–23) 53 (50–55)  Positive predictive value ICAb NA 11 (10–13) 12 (10–13) 18 (15–21)  Negative predictive value ICAb NA 96 (94–98) 97 (95–99) 99 (98–99)  Positive likelihood ratio ICAb NA 1.17 (1.11–1.23) 1.19 (1.13–1.24) 1.98 (1.85–2.12)  Negative likelihood ratio ICAb NA 0.34 (0.19–0.60) 0.27 (0.14–0.52) 0.12 (0.06–0.22) Pre-test probability cut-off <10  Number of patients below cut-off 69 (4) 695 (43) 658 (41) 981 (60)  Number of diseased patients below cut-offd 1 (1) 38 (5) 32 (5) 15 (2)  Sensitivity ICAb 99 (97–100) 76 (69–83) 80 (73–86) 91 (85–95)  Specificity ICAb 5 (4–6) 45 (42–48) 43 (40–45) 66 (64–68)  Positive predictive value ICAb 10 (9–12) 13 (11–16) 13 (11–16) 23 (19–26)  Negative predictive value ICAb 99 (92–100) 95 (93–96) 95 (93–97) 99 (98–99)  Positive likelihood ratio ICAb 1.04 (1.02–1.06) 1.38 (1.25–1.53) 1.40 (1.28–1.53) 2.66 (2.44–2.91)  Negative likelihood ratio ICAb 0.13 (0.02–0.96) 0.53 (0.40–0.70) 0.47 (0.34–0.64) 0.14 (0.09–0.23) Pre-test probability cut-off <15  Number of patients below cut-off 231 (14) 949 (58) 898 (55) 1098 (68)  Number of diseased patients below cut-offd 8 (3) 56 (6) 49 (5) 23 (2)  Sensitivity ICAb 95 (90–98) 65 (57–72) 69 (62–76) 86 (79–90)  Specificity ICAb 15 (13–17) 61 (58–64) 58 (55–61) 73 (71–76)  Positive predictive value ICAb 11 (9–13) 15 (13–18) 15 (13–18) 26 (22–30)  Negative predictive value ICAb 97 (93–99) 94 (93–96) 95 (93–96) 98 (97–99)  Positive likelihood ratio ICAb 1.12 (1.07–1.17) 1.67 (1.46–1.90) 1.65 (1.47–1.86) 3.22 (2.90–3.58)  Negative likelihood ratio ICAb 0.33 (0.17–0.65) 0.57 (0.46–0.71) 0.53 (0.42–0.67) 0.20 (0.13–0.29) Reference Diamond–Forrester score (updated) CAD Consortium Basic Clinical Clinical + CACS Median value Median pre-test probability models value  Total cohort 36 (20–53) 12 (6–25) 13 (6–28) 6 (2–26)  No CAD CTAa 29 (16–44) 9 (4–17) 9 (4–17) 2 (1–4)  Mild/moderate CAD CTAa 41 (24–55) 15 (8–26) 16 (9–30) 16 (6–30)  Severe CAD CTAa 46 (29–63) 20 (9–36) 24 (11–40) 37 (19–61)  Non-haemodynamically obstructive ICAb 35 (19–51) 12 (6–22) 12 (6–25) 4 (1–17)  Haemodynamically obstructive CAD ICAb 50 (34–71) 23 (11–39) 30 (13–47) 47 (24–72) Calibration Calibration-in-the-largec CTAa −0.89 0.42 0.30 0.65 ICAb −2.09 −0.80 −0.95 −0.87 Calibration slopec CTAa 0.51 0.52 0.57 0.94 ICAb 0.60 0.57 0.60 0.84 Discrimination Area under the receiver operating characteristic curve CTAa 65 (61–68) 66 (63–69) 69 (66–72) 87 (85–89) ICAb 67 (63–72) 68 (63–72) 70 (66–74) 86 (83–89) Pre-test probability cut-off <5  Number of patients below cut-off 0 310 (19) 310 (19) 781 (48)  Number of diseased patients below cut-offd 0 11 (4) 9 (3) 10 (1)  Sensitivity ICAb NA 93 (88–97) 94 (90–97) 94 (89–97)  Specificity ICAb NA 20 (18–23) 21 (19–23) 53 (50–55)  Positive predictive value ICAb NA 11 (10–13) 12 (10–13) 18 (15–21)  Negative predictive value ICAb NA 96 (94–98) 97 (95–99) 99 (98–99)  Positive likelihood ratio ICAb NA 1.17 (1.11–1.23) 1.19 (1.13–1.24) 1.98 (1.85–2.12)  Negative likelihood ratio ICAb NA 0.34 (0.19–0.60) 0.27 (0.14–0.52) 0.12 (0.06–0.22) Pre-test probability cut-off <10  Number of patients below cut-off 69 (4) 695 (43) 658 (41) 981 (60)  Number of diseased patients below cut-offd 1 (1) 38 (5) 32 (5) 15 (2)  Sensitivity ICAb 99 (97–100) 76 (69–83) 80 (73–86) 91 (85–95)  Specificity ICAb 5 (4–6) 45 (42–48) 43 (40–45) 66 (64–68)  Positive predictive value ICAb 10 (9–12) 13 (11–16) 13 (11–16) 23 (19–26)  Negative predictive value ICAb 99 (92–100) 95 (93–96) 95 (93–97) 99 (98–99)  Positive likelihood ratio ICAb 1.04 (1.02–1.06) 1.38 (1.25–1.53) 1.40 (1.28–1.53) 2.66 (2.44–2.91)  Negative likelihood ratio ICAb 0.13 (0.02–0.96) 0.53 (0.40–0.70) 0.47 (0.34–0.64) 0.14 (0.09–0.23) Pre-test probability cut-off <15  Number of patients below cut-off 231 (14) 949 (58) 898 (55) 1098 (68)  Number of diseased patients below cut-offd 8 (3) 56 (6) 49 (5) 23 (2)  Sensitivity ICAb 95 (90–98) 65 (57–72) 69 (62–76) 86 (79–90)  Specificity ICAb 15 (13–17) 61 (58–64) 58 (55–61) 73 (71–76)  Positive predictive value ICAb 11 (9–13) 15 (13–18) 15 (13–18) 26 (22–30)  Negative predictive value ICAb 97 (93–99) 94 (93–96) 95 (93–96) 98 (97–99)  Positive likelihood ratio ICAb 1.12 (1.07–1.17) 1.67 (1.46–1.90) 1.65 (1.47–1.86) 3.22 (2.90–3.58)  Negative likelihood ratio ICAb 0.33 (0.17–0.65) 0.57 (0.46–0.71) 0.53 (0.42–0.67) 0.20 (0.13–0.29) Values are presented as intercepts/slopes, %, n (%), or median (interquartile range). a Severe CAD defined from coronary CTA. b Haemodynamically obstructive CAD defined from ICA with fractional flow reserve measurement. c Perfect predictions should be on the ideal line in a calibration plot, described with an intercept alpha of 0 (calibration-in-the-large) and slope beta of 1 (calibration slope). d Number of patients with haemodynamically obstructive CAD at ICA with a pre-test probability below the cut-off. NA, not available. Open in new tab Table 3 Pre-test probability models calibration and discrimination analysis Reference Diamond–Forrester score (updated) CAD Consortium Basic Clinical Clinical + CACS Median value Median pre-test probability models value  Total cohort 36 (20–53) 12 (6–25) 13 (6–28) 6 (2–26)  No CAD CTAa 29 (16–44) 9 (4–17) 9 (4–17) 2 (1–4)  Mild/moderate CAD CTAa 41 (24–55) 15 (8–26) 16 (9–30) 16 (6–30)  Severe CAD CTAa 46 (29–63) 20 (9–36) 24 (11–40) 37 (19–61)  Non-haemodynamically obstructive ICAb 35 (19–51) 12 (6–22) 12 (6–25) 4 (1–17)  Haemodynamically obstructive CAD ICAb 50 (34–71) 23 (11–39) 30 (13–47) 47 (24–72) Calibration Calibration-in-the-largec CTAa −0.89 0.42 0.30 0.65 ICAb −2.09 −0.80 −0.95 −0.87 Calibration slopec CTAa 0.51 0.52 0.57 0.94 ICAb 0.60 0.57 0.60 0.84 Discrimination Area under the receiver operating characteristic curve CTAa 65 (61–68) 66 (63–69) 69 (66–72) 87 (85–89) ICAb 67 (63–72) 68 (63–72) 70 (66–74) 86 (83–89) Pre-test probability cut-off <5  Number of patients below cut-off 0 310 (19) 310 (19) 781 (48)  Number of diseased patients below cut-offd 0 11 (4) 9 (3) 10 (1)  Sensitivity ICAb NA 93 (88–97) 94 (90–97) 94 (89–97)  Specificity ICAb NA 20 (18–23) 21 (19–23) 53 (50–55)  Positive predictive value ICAb NA 11 (10–13) 12 (10–13) 18 (15–21)  Negative predictive value ICAb NA 96 (94–98) 97 (95–99) 99 (98–99)  Positive likelihood ratio ICAb NA 1.17 (1.11–1.23) 1.19 (1.13–1.24) 1.98 (1.85–2.12)  Negative likelihood ratio ICAb NA 0.34 (0.19–0.60) 0.27 (0.14–0.52) 0.12 (0.06–0.22) Pre-test probability cut-off <10  Number of patients below cut-off 69 (4) 695 (43) 658 (41) 981 (60)  Number of diseased patients below cut-offd 1 (1) 38 (5) 32 (5) 15 (2)  Sensitivity ICAb 99 (97–100) 76 (69–83) 80 (73–86) 91 (85–95)  Specificity ICAb 5 (4–6) 45 (42–48) 43 (40–45) 66 (64–68)  Positive predictive value ICAb 10 (9–12) 13 (11–16) 13 (11–16) 23 (19–26)  Negative predictive value ICAb 99 (92–100) 95 (93–96) 95 (93–97) 99 (98–99)  Positive likelihood ratio ICAb 1.04 (1.02–1.06) 1.38 (1.25–1.53) 1.40 (1.28–1.53) 2.66 (2.44–2.91)  Negative likelihood ratio ICAb 0.13 (0.02–0.96) 0.53 (0.40–0.70) 0.47 (0.34–0.64) 0.14 (0.09–0.23) Pre-test probability cut-off <15  Number of patients below cut-off 231 (14) 949 (58) 898 (55) 1098 (68)  Number of diseased patients below cut-offd 8 (3) 56 (6) 49 (5) 23 (2)  Sensitivity ICAb 95 (90–98) 65 (57–72) 69 (62–76) 86 (79–90)  Specificity ICAb 15 (13–17) 61 (58–64) 58 (55–61) 73 (71–76)  Positive predictive value ICAb 11 (9–13) 15 (13–18) 15 (13–18) 26 (22–30)  Negative predictive value ICAb 97 (93–99) 94 (93–96) 95 (93–96) 98 (97–99)  Positive likelihood ratio ICAb 1.12 (1.07–1.17) 1.67 (1.46–1.90) 1.65 (1.47–1.86) 3.22 (2.90–3.58)  Negative likelihood ratio ICAb 0.33 (0.17–0.65) 0.57 (0.46–0.71) 0.53 (0.42–0.67) 0.20 (0.13–0.29) Reference Diamond–Forrester score (updated) CAD Consortium Basic Clinical Clinical + CACS Median value Median pre-test probability models value  Total cohort 36 (20–53) 12 (6–25) 13 (6–28) 6 (2–26)  No CAD CTAa 29 (16–44) 9 (4–17) 9 (4–17) 2 (1–4)  Mild/moderate CAD CTAa 41 (24–55) 15 (8–26) 16 (9–30) 16 (6–30)  Severe CAD CTAa 46 (29–63) 20 (9–36) 24 (11–40) 37 (19–61)  Non-haemodynamically obstructive ICAb 35 (19–51) 12 (6–22) 12 (6–25) 4 (1–17)  Haemodynamically obstructive CAD ICAb 50 (34–71) 23 (11–39) 30 (13–47) 47 (24–72) Calibration Calibration-in-the-largec CTAa −0.89 0.42 0.30 0.65 ICAb −2.09 −0.80 −0.95 −0.87 Calibration slopec CTAa 0.51 0.52 0.57 0.94 ICAb 0.60 0.57 0.60 0.84 Discrimination Area under the receiver operating characteristic curve CTAa 65 (61–68) 66 (63–69) 69 (66–72) 87 (85–89) ICAb 67 (63–72) 68 (63–72) 70 (66–74) 86 (83–89) Pre-test probability cut-off <5  Number of patients below cut-off 0 310 (19) 310 (19) 781 (48)  Number of diseased patients below cut-offd 0 11 (4) 9 (3) 10 (1)  Sensitivity ICAb NA 93 (88–97) 94 (90–97) 94 (89–97)  Specificity ICAb NA 20 (18–23) 21 (19–23) 53 (50–55)  Positive predictive value ICAb NA 11 (10–13) 12 (10–13) 18 (15–21)  Negative predictive value ICAb NA 96 (94–98) 97 (95–99) 99 (98–99)  Positive likelihood ratio ICAb NA 1.17 (1.11–1.23) 1.19 (1.13–1.24) 1.98 (1.85–2.12)  Negative likelihood ratio ICAb NA 0.34 (0.19–0.60) 0.27 (0.14–0.52) 0.12 (0.06–0.22) Pre-test probability cut-off <10  Number of patients below cut-off 69 (4) 695 (43) 658 (41) 981 (60)  Number of diseased patients below cut-offd 1 (1) 38 (5) 32 (5) 15 (2)  Sensitivity ICAb 99 (97–100) 76 (69–83) 80 (73–86) 91 (85–95)  Specificity ICAb 5 (4–6) 45 (42–48) 43 (40–45) 66 (64–68)  Positive predictive value ICAb 10 (9–12) 13 (11–16) 13 (11–16) 23 (19–26)  Negative predictive value ICAb 99 (92–100) 95 (93–96) 95 (93–97) 99 (98–99)  Positive likelihood ratio ICAb 1.04 (1.02–1.06) 1.38 (1.25–1.53) 1.40 (1.28–1.53) 2.66 (2.44–2.91)  Negative likelihood ratio ICAb 0.13 (0.02–0.96) 0.53 (0.40–0.70) 0.47 (0.34–0.64) 0.14 (0.09–0.23) Pre-test probability cut-off <15  Number of patients below cut-off 231 (14) 949 (58) 898 (55) 1098 (68)  Number of diseased patients below cut-offd 8 (3) 56 (6) 49 (5) 23 (2)  Sensitivity ICAb 95 (90–98) 65 (57–72) 69 (62–76) 86 (79–90)  Specificity ICAb 15 (13–17) 61 (58–64) 58 (55–61) 73 (71–76)  Positive predictive value ICAb 11 (9–13) 15 (13–18) 15 (13–18) 26 (22–30)  Negative predictive value ICAb 97 (93–99) 94 (93–96) 95 (93–96) 98 (97–99)  Positive likelihood ratio ICAb 1.12 (1.07–1.17) 1.67 (1.46–1.90) 1.65 (1.47–1.86) 3.22 (2.90–3.58)  Negative likelihood ratio ICAb 0.33 (0.17–0.65) 0.57 (0.46–0.71) 0.53 (0.42–0.67) 0.20 (0.13–0.29) Values are presented as intercepts/slopes, %, n (%), or median (interquartile range). a Severe CAD defined from coronary CTA. b Haemodynamically obstructive CAD defined from ICA with fractional flow reserve measurement. c Perfect predictions should be on the ideal line in a calibration plot, described with an intercept alpha of 0 (calibration-in-the-large) and slope beta of 1 (calibration slope). d Number of patients with haemodynamically obstructive CAD at ICA with a pre-test probability below the cut-off. NA, not available. Open in new tab Calibration and discrimination The calibration and discrimination of severe CAD at coronary CTA and haemodynamically obstructive CAD at ICA for the different PTP models are presented in Table 3 and Figure 4. When compared to severe CAD evaluated with coronary CTA, the overall calibration was satisfactory for all three CAD Consortium models, but the UDF score overestimated the PTP. Discrimination of severe CAD at coronary CTA evaluated with AUC increased significantly with the more advanced CAD Consortium Clinical, 69% (66–72), and Clinical + CACS, AUC 87% (85–89), models compared with the UDF score, AUC 65% (61–68), (P < 0.001). CAD Consortium Basic model, AUC 66% (63–69), did not increase compared to UDF score. Figure 4 Open in new tabDownload slide Calibration plot and AUC with present of severe CAD at coronary CTA and presence of haemodynamically obstructive CAD at ICA as a reference for the four pre-test probability stratification models (A-D) (n = 1653). Pre-test probability models calibration and discrimination analysis are presented in Table 3. Figure 4 Open in new tabDownload slide Calibration plot and AUC with present of severe CAD at coronary CTA and presence of haemodynamically obstructive CAD at ICA as a reference for the four pre-test probability stratification models (A-D) (n = 1653). Pre-test probability models calibration and discrimination analysis are presented in Table 3. The calibration and discrimination curves with myocardial perfusion defects as endpoint, either cardiac magnetic resonance imaging or single-photon emission CT MPI endpoint, as presented in the Supplementary data online, Figure S3. When compared with haemodynamically obstructive CAD evaluated by ICA-FFR, the overall calibration for all models overestimated the PTP of haemodynamically obstructive CAD. Nonetheless, calibrations were better for the CAD Consortium Basic and Clinical scores than for the UDF-score, and the CAD Consortium Clinical + CACS was superior to the three other scores. Discrimination evaluated with AUC did not increase between the UDF score, AUC 67% (63–72), and the CAD Consortium Basic score, AUC 68% (63–72) (P = 0.54). However, the AUC of the CAD Consortium Clinical score, 70% (66–74), increased significantly compared with both the UDF score and the CAD Consortium Basic score (P < 0.001). The CAD Consortium Clinical + CACS, AUC 86% (83–89), increased more than all other models (P < 0.001) (Table 3 and Figure 4.). However, the CAD Consortium Clinical + CACS did not increase compared to CACS alone, AUC 0.86 (0.83–0.89) vs. 0.86 (0.83–0.89), P = 0.85. The diagnostic accuracy evaluated by sensitivity and specificity for the PTP models is illustrated in Figure 5. Figure 5 Open in new tabDownload slide Sensitivity and specificity plot for the four pre-test probability stratification models (A-D). The dotted line represents a cut-off for low-probability patients of 15% as proposed by the European guidelines, and the vertical arrows present the sensitivity and specificity at this cut-off. Figure 5 Open in new tabDownload slide Sensitivity and specificity plot for the four pre-test probability stratification models (A-D). The dotted line represents a cut-off for low-probability patients of 15% as proposed by the European guidelines, and the vertical arrows present the sensitivity and specificity at this cut-off. When low PTP was defined as a pre-test score cut-off of <15%, the total number of low-probability patients was as follows for the UDF, the CAD Consortium Basic, the Clinical, and the Clinical + CACS score: 231 (14%), 949 (58%), 898 (55%), and 1098 (68%), respectively, with corresponding NPVs of 97% (93–99), 94% (92–96), 95% (93–96), and 98% (97–99), respectively. Diagnostic accuracy parameters for a PTP score with a cut-off of <5, <10, and <15 are listed in Table 3. When high PTP was defined as a pre-test score cut-off of >85%, the total number of high-probability patients was as follows for the UDF, the CAD Consortium Basic, the Clinical, and the Clinical + CACS scores: 38 (2%), 0 (0%), 2 (0%), and 23 (1%), respectively, with corresponding PPVs of 34% (20–51), not available, 100% (16–100), and 71% (49–87), respectively. Discussion We studied PTP stratification models against haemodynamically obstructive CAD using ICA-FFR as a reference. In contrast to previous studies, we examined a low to intermediate probability cohort without known CAD who was referred to coronary CTA and subsequently ICA-FFR if stenosis was suspected at coronary CTA. The study population is a de novo cohort, as coronary CTA is the established first-line test in all eligible patients in Denmark. The main finding is the superior calibration of the CAD Consortium PTP models compared with the UDF score. Consequently, with a PTP cut-off of 15%, >50% of the cohort would be categorized as being at low probability by the CAD Consortium models in contrast to only 14% of the cohort if the UDF score had been used. Hence, well-calibrated PTP models may significantly reduce the need for further diagnostic testing and they seem to be a safe alternative due to the preserved high NPV of ≥94%. The 2013 ESC guidelines on the management of stable CAD recommend using the UDF score and defined low-probability patients as those with a PTP of <15%.5 The argument is as follows: ‘Non-invasive, imaging-based diagnostic methods for CAD have typical sensitivities and specificities of approximately 85%. Hence, 15% of all diagnostic results will be false and, as a consequence, performing no test at all will provide fewer incorrect diagnoses in patients with a pre-test probability <15% (assuming all patients to be healthy) or a pre-test probability >85% (assuming all patients to be diseased). In these situations, testing should only be done for compelling reasons. This is the reason why this Task Force recommends no testing in patients with (i) a low pre-test probability <15% and (ii) a high pre-test probability >85%. In such patients, it is safe to assume that they have (i) no obstructive CAD or (ii) obstructive CAD’. In line with this, the AHA/ACC guidelines of 2012 stated that non-invasive test is most useful in patients with a probability of CAD between 20% and 70%.3 Accurate PTP assessment is essential for patients’ safety and for the yield and cost-effectiveness of subsequent non-invasive and invasive testing.11 Hence, our study illustrates the impact of choosing the best-calibrated PTP model for identification of low-probability patients. In line with our study, previous studies show that the original Diamond–Forrester score, the UDF score, and the Duke Clinical score all overestimate the PTP of CAD, in particular in low to intermediate probability cohorts, and the diagnostic accuracy by AUC is between 65% and 75% with diameter stenosis as a reference.4,6,12,13 Since the development of the CAD Consortium models in 2012,6 the models have been externally validated in four studies using 50% diameter stenosis as a reference at either coronary CTA or ICA. The Duke Clinical Score and the CAD Consortium Basic and Clinical model were evaluated in 2234 high-probability patients referred to ICA for suspected CAD in a study by Almeid et al.14 The prevalence of CAD was 59%. AUCs were 69%, 66%, and 68%, respectively. Performances were significantly better for the Duke Clinical Score and the CAD Consortium Clinical model than for the CAD Consortium Basic model. The Duke Clinical Score performed best in the high-probability subgroup, and the CAD Consortium Clinical model was superior in the low-probability group. Bittencourt et al.15 evaluated 2274 patients without prior known CAD referred to coronary CTA. They compared the UDF score with the CAD Consortium Basic and Clinical model. The prevalence of obstructive CAD was 22%. AUCs were 71%, 75%, and 79%, respectively. As in our study, the UDF score overestimated the PTP of obstructive CAD compared with the two CAD Consortium models. However, in all models, the discriminatory ability was substantially higher than in our study when coronary CTA was used as reference despite a similar CAD prevalence. Nonetheless, this study also demonstrated a three-fold increase in the number of individuals who would be categorized as being at low probability with the CAD consortium models (31%) compared to the UDF score (9%) and subsequently would require no additional testing. Of note, Bittencourt et al. also compared the ability of the model to predict major adverse cardiovascular events and here the CAD consortium models outperformed the UDF score. The AUCs of events for the UDF score, the CAD Consortium Basic, and the Clinical model were 62%, 64%, and 69%, respectively. In the SCOT-HEART trial (Scottish COmputed Tomography of the HEART trial), the CAD Consortium Clinical model was superior to the UDF in 1738 patients evaluated with coronary CTA.16 The prevalence of CAD was high, 38%, in the SCOT-HEART compared with our study and the others studies in coronary CTA cohorts. This is likely due to referral bias because 85% of the patients in the overall cohort of this trial underwent exercise treadmill testing prior to enrolment. Finally, the CAD Consortium Basic, the Clinical, and the Clinical + CACS model were assessed in 3468 patients referred to coronary CTA as part of the PROMISE (Prospective Multicenter Imaging Study for Evaluation of Chest Pain) trial.13 At coronary CTA, 23% of the patients had CAD, and 13% of the patients were referred for ICA, similarly to the prevalence observed in our study. In this study, the authors imputed missing ICA results based on clinical variables and CTA results. In total, obstructive CAD on ICA was imputed for 61% of patients. With ICA as reference the CAD Consortium Basic, Clinical, and Clinical + CACS model had AUCs of 69%, 72%, and 86%. These AUCs are comparable to our results. Interestingly, the Basic and the Clinical model calibrations were better in men than in women, but calibration of the Clinical + CACS score model was nearly perfect regardless of sex in the PROMISE study. In our study, single risk factors or symptoms had no substantial predictive power compared with coronary artery calcification. Hence, the inclusion of CACS to the CAD Consortium models explains the superiority of CAD Consortium Clinical + CACS. CACS is currently measured before coronary CTA and some nuclear MPI modalities. Our results therefore challenge the use of further investigation with iodinated contrast, stress agents, and radiation in patients with a very low PTP score in the Consortium Clinical + CACS model. In fact, a setup for a future rapid chest pain clinic with a nurses/technicians visit including ECG, Echo, and CACS before clinical assessment may rule-out CAD in 60% of the patients with a 99% NPV by using CAD Consortium Clinical + CACS with a PTP cut-off <10. Changing the current guideline recommendation of using the UDF score to using the CAD Consortium models seems to have clinical implications. By recommending the CAD Consortium models with a cut-off of 5% or 10%, the need for non-invasive imaging may not be increased compared with a UDF score cut-off of 15%. However, the optimal cut-off for low-risk defined with CAD Consortium models needs to be further investigated. Strengths and limitations Our study was conducted in a single region of Denmark, the population was overall healthy and almost all participants were Caucasians, which limits the generalizability. Referral bias is limited because the Danish healthcare system is without direct payment for all citizens, and coronary CTA is the recommended first-line diagnostic test for low to intermediate probability. However, for patients with symptoms suggestive of CAD, the first contact to the healthcare system is most often the general practitioner, who refers patients with relevant symptoms to the hospital outpatient clinic. Our results are based on patients admitted to hospital outpatient clinics. Conclusions No single risk factor or symptom convincingly predicts haemodynamically obstructive CAD in patients with symptoms suggestive of CAD and a low to intermediate probability profile. Nonetheless, CAD Consortium PTP models improve the PTP stratification substantially compared with the UDF score mainly owed to improved calibration in this low to intermediate probability cohort. However, only adding the CACS to the models substantially increases the discriminatory power. Acknowledgements The authors would like to thank the Dan-NICAD study collaboration including study nurses and clinical staff at the enrolling centres. Funding The Danish Heart Foundation (grant no. 15-R99-A5837-22920), the Health Research Fund of the Central Denmark Region, and Acarix A/S. Conflict of interest: none declared. References 1 Diamond GA , Forrester JS. Analysis of probability as an aid in the clinical diagnosis of coronary-artery disease . N Engl J Med 1979 ; 300 : 1350 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Pryor DB , Shaw L , McCants CB , Lee KL , Mark DB , Harrell FE Jr et al. Value of the history and physical in identifying patients at increased risk for coronary artery disease . Ann Intern Med 1993 ; 118 : 81 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Fihn SD , Gardin JM , Abrams J , Berra K , Blankenship JC , Dallas AP et al. 2012 ACCF/AHA/ACP/AATS/PCNA/SCAI/STS Guideline for the diagnosis and management of patients with stable ischemic heart disease: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines, and the American College of Physicians, American Association for Thoracic Surgery, Preventive Cardiovascular Nurses Association, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons . J Am Coll Cardiol 2012 ; 60 : 2564 – 603 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Genders TS , Steyerberg EW , Alkadhi H , Leschka S , Desbiolles L , Nieman K et al. A clinical prediction rule for the diagnosis of coronary artery disease: validation, updating, and extension . Eur Heart J 2011 ; 32 : 1316 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Task Force M , Montalescot G , Sechtem U , Achenbach S , Andreotti F , Arden C et al. 2013 ESC guidelines on the management of stable coronary artery disease: the Task Force on the management of stable coronary artery disease of the European Society of Cardiology . Eur Heart J 2013 ; 34 : 2949 – 3003 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Genders TS , Steyerberg EW , Hunink MG , Nieman K , Galema TW , Mollet NR et al. Prediction model to estimate presence of coronary artery disease: retrospective pooled analysis of existing cohorts . BMJ 2012 ; 344 : e3485. Google Scholar Crossref Search ADS PubMed WorldCat 7 Park SJ , Kang SJ , Ahn JM , Shim EB , Kim YT , Yun SC et al. Visual-functional mismatch between coronary angiography and fractional flow reserve . JACC Cardiovasc Interv 2012 ; 5 : 1029 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Nissen L , Winther S , Isaksen C , Ejlersen JA , Brix L , Urbonaviciene G et al. Danish study of Non-Invasive testing in Coronary Artery Disease (Dan-NICAD): study protocol for a randomised controlled trial . Trials 2016 ; 17 : 262. Google Scholar Crossref Search ADS PubMed WorldCat 9 Nissen L , Winther S , Westra J , Ejlersen JA , Isaksen C , Rossi A et al. Diagnosing coronary artery disease after a positive coronary computed tomography angiography: the Dan-NICAD open label, parallel, head to head, randomized controlled diagnostic accuracy trial of cardiovascular magnetic resonance and myocardial perfusion scintigraphy . Eur Heart J Cardiovasc Imaging 2018 ; 19 : 369 – 77 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Steyerberg EW , Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation . Eur Heart J 2014 ; 35 : 1925 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Knuuti J , Ballo H , Juarez-Orozco LE , Saraste A , Kolh P , Rutjes AWS et al. The performance of non-invasive tests to rule-in and rule-out significant coronary artery stenosis in patients with stable angina: a meta-analysis focused on post-test disease probability . Eur Heart J 2018 ; 39 : 3322 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Jensen JM , Voss M , Hansen VB , Andersen LK , Johansen PB , Munkholm H et al. Risk stratification of patients suspected of coronary artery disease: comparison of five different models . Atherosclerosis 2012 ; 220 : 557 – 62 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Genders TSS , Coles A , Hoffmann U , Patel MR , Mark DB , Lee KL et al. The external validity of prediction models for the diagnosis of obstructive coronary artery disease in patients with stable chest pain: insights from the PROMISE Trial . JACC Cardiovasc Imaging 2018 ; 11 : 437 – 46 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Almeida J , Fonseca P , Dias T , Ladeiras-Lopes R , Bettencourt N , Ribeiro J et al. Comparison of coronary artery disease consortium 1 and 2 scores and duke clinical score to predict obstructive coronary disease by invasive coronary angiography . Clin Cardiol 2016 ; 39 : 223 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Bittencourt MS , Hulten E , Polonsky TS , Hoffman U , Nasir K , Abbara S et al. European Society of Cardiology-recommended coronary artery disease consortium pretest probability scores more accurately predict obstructive coronary disease and cardiovascular events than the Diamond and Forrester Score: the Partners Registry . Circulation 2016 ; 134 : 201 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Baskaran L , Danad I , Gransar H , O Hartaigh B , Schulman-Marcus J , Lin FY et al. A comparison of the updated Diamond-Forrester, CAD Consortium, and CONFIRM history-based risk scores for predicting obstructive coronary artery disease in patients with stable chest pain: the SCOT-HEART coronary CTA cohort . JACC Cardiovasc Imaging 2018 ;doi:10.1016/j.jcmg.2018.02.020. WorldCat Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2019. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Pre-test probability prediction in patients with a low to intermediate probability of coronary artery disease: a prospective study with a fractional flow reserve endpoint
JF - European Heart Journal - Cardiovascular Imaging
DO - 10.1093/ehjci/jez058
DA - 2019-11-01
UR - https://www.deepdyve.com/lp/oxford-university-press/pre-test-probability-prediction-in-patients-with-a-low-to-intermediate-TT40fOv0Pc
SP - 1208
VL - 20
IS - 11
DP - DeepDyve
ER -