Prognostic Value of Coronary CTA in Stable Chest Pain: CAD-RADS, CAC, and Cardiovascular Events in PROMISE
Original Research
Central Illustration
Abstract
Objectives
The purpose of this study was to compare Coronary Artery Disease Reporting and Data System (CAD-RADS) to traditional stenosis categories and the coronary artery calcium score (CACS) for predicting cardiovascular events in patients with stable chest pain and suspected coronary artery disease (CAD).
Background
The 2016 CAD-RADS has been established to standardize the reporting of CAD on coronary CT angiography (CTA).
Methods
PROMISE (Prospective Multicenter Imaging Study for Evaluation of Chest Pain) trial participants’ CTAs were assessed by a central CT core laboratory for CACS, traditional stenosis-based categories, and modified CAD-RADS grade including high-risk coronary plaque (HRP) features. Traditional stenosis categories and CAD-RADS grade were compared for the prediction of the composite endpoint of death, myocardial infarction, or hospitalization for unstable angina over a median follow-up of 25 months. Incremental prognostic value over traditional risk factors and CACS was assessed.
Results
In 3,840 eligible patients (mean age: 60.4 ± 8.2 years; 49% men), 3.0% (115) experienced events. CAD-RADS (concordance statistic [C-statistic] 0.747) had significantly higher discriminatory value than traditional stenosis-based assessments (C-statistic 0.698 to 0.717; all p for comparison ≤0.001). With no plaque (CAD-RADS 0) as the baseline, the hazard ratio (HR) for an event increased from 2.43 (95% confidence interval [CI]: 1.16 to 5.08) for CAD-RADS 1 to 21.84 (95% CI: 8.63 to 55.26) for CAD-RADS 4b and 5. In stepwise nested models, CAD-RADS added incremental prognostic value beyond ASCVD risk score and CACS (C-statistic 0.776 vs. 0.682; p < 0.001), and added incremental value persisted in all CACS strata.
Conclusions
These data from a large representative contemporary cohort of patients undergoing coronary CTA for stable chest pain support the prognostic value of CAD-RADS as a standard reporting system for coronary CTA.
Introduction
The current 2012 American College of Cardiology/American Heart Association (ACC/AHA) guidelines give a Class IIb indication for coronary computed tomography angiography (CTA) for patients with stable chest pain (1). In 2015, 2 large randomized controlled trials (PROMISE [Prospective Multicenter Imaging Study for Evaluation of Chest Pain] and SCOT-HEART [Scottish Computed Tomography of the HEART), demonstrated that a diagnostic strategy including CTA has similar to superior health outcomes to functional testing in stable chest pain, again confirmed by the 5-year follow-up data from SCOT-HEART in 2018 (2–4). In 2016, these results led the UK National Institute for Health and Clinical Excellence (NICE) to recommend coronary CTA as the first-line test for patients with atypical or typical anginal symptoms (5). Based on these developments, it seems likely that coronary CTA will take an increasingly important role in the evaluation of chest pain.
Over the last 20 years, many studies have established the high diagnostic accuracy of coronary CTA to detect obstructive coronary artery disease (CAD) on invasive coronary angiography (6–8). Likewise CTA can also identify nonobstructive plaque and high-risk plaque (HRP) seen on intravascular ultrasound (9–11). Obstructive plaque, nonobstructive plaque, and HRP are independent predictors of major adverse coronary events (MACE) and have the potential to guide management decisions (2,4,12,13). Initially, prognostic studies were limited to assessing the value of stenosis, which remains the strongest predictor of future MACE. More recently, the value of nonobstructive CAD and HRP has been established (12,14). The coronary artery calcium score (CACS) is often acquired before coronary CTA, and although the prognostic value of CACS is best established in asymptomatic populations, it also has strong prognostic value in patients with stable chest pain (15,16).
To standardize and facilitate the reporting of CAD on coronary CTA, in 2016 the Society of Cardiovascular Computed Tomography (SCCT), the American College of Radiology (ACR), and the North American Society for Cardiovascular Imaging (NASCI) established the Coronary Artery Disease Reporting and Data System (CAD-RADS) (17). Compared with the prevailing traditional cut points for per-lesion stenosis on CTA (0%, 1% to 49%, 50% to 69%, 70% to 100%), CAD-RADS adds additional categories intended for risk stratification: 1) explicitly differentiating between minimal (1% to 29%) and mild (30% to 49%) stenosis; 2) adding categories for left main and multivessel stenosis; and 3) including HRP features. However, the prognostic value of CAD-RADS including HRP and whether there is incremental value beyond existing traditional stenosis categories, ASCVD risk score, or CACS is not known.
Thus, the aim of this study was to determine the prognostic value of CAD-RADS and compare with established predictors of MACE in a large contemporary population with stable chest pain.
Methods
Study design and population
The PROMISE trial was a randomized comparative effectiveness trial in stable outpatients with chest pain who required noninvasive cardiac testing to determine the presence of obstructive CAD or myocardial ischemia. The study population and inclusion and exclusion criteria are detailed elsewhere (2,18). In brief, 10,003 patients from 193 sites across North America with expertise in the fields of cardiology, primary care, radiology, and anesthesia were included in PROMISE between July 2010, and September 2013. Patients were randomly assigned to either anatomic coronary CTA or functional testing (exercise electrocardiography, stress echocardiography, or nuclear stress testing), with interpretation of testing and subsequent decision making by the local physicians. Local institutional review boards approved the study, and all patients provided written informed consent.
Our analysis included PROMISE patients who were randomized to the coronary CTA arm and received diagnostic noncontrast CT for calcium scoring and contrast-enhanced coronary CTA (Figure 1). In the parent PROMISE trial, the noncontrast CT before the coronary CTA was not mandatory but left to the discretion of site investigators.
Consort Diagram
Consort diagram showing included and excluded patients originating from the coronary CTA-arm of the parent PROMISE (Prospective Multicenter Imaging Study for Evaluation of Chest Pain) trial.
CACS, CTA and HRP
Coronary artery calcification was assessed on all available noncontrast CT datasets by 1 of 7 readers blinded to clinical information and outcome. We quantified the CACS using the Agatston method and dedicated software (Syngo.via, Siemens Healthcare, Forchheim, Germany). Upfront interobserver agreement was assessed in 30 randomly selected noncontrast CT datasets with an intraclass correlation coefficient ranging from 0.99 to 1.00 among all readers.
Coronary CTA datasets were analyzed by 1 of 6 expert readers in coronary CTA who were also blinded to clinical information and outcome (19). Upfront interobserver reliability was assessed using 50 randomly selected coronary CTAs among all readers (≥70% stenosis or left main ≥50% stenosis: kappa = 0.69; HRP: kappa = 0.56). Evaluable coronary artery segments were assessed for the presence of stenosis using 5 predefined categories: 0%, 1% to 29%, 30% to 49%, 50% to 69%, or ≥70% stenosis. Stenosis was categorized in 3 ways.
First, stenosis were categorized using CAD-RADS (17). CAD-RADS was introduced after the start of the core laboratory measurements; to translate our stenosis categories to those in CAD-RADS, we made a minor modification to CAD-RADS category 1 (stenosis 1% to 29% instead of 1% to 24%) and category 2 (stenosis 30% to 49% instead of 25% to 49%). Thus, this analysis evaluates a slightly modified CAD-RADS; the term CAD-RADS is used throughout for readability.
Second, we defined traditional stenosis in 2 ways. Traditional definition 1 corresponds to the pre-existing standard for coronary CTA including stenosis categories of no CAD (0%), mild CAD (1% to 49% stenosis), moderate CAD (50% to 69% stenosis in any major vessels/branch), severe CAD (≥50% stenosis of left main [LM] or ≥70% in any major vessel/branch). Traditional definition 2 was defined according to the obstructive CAD definition used in the PROMISE trial (14): normal (absence of coronary atherosclerosis), mildly abnormal (nonobstructive CAD: 1% to 69% stenosis in any major vessels/branch, or <50% LM stenosis), moderately abnormal (obstructive CAD: ≥70% stenosis in 1 major vessel/branch), and severely abnormal (high-risk CAD: 2 or more vessel disease (≥70%), or ≥50% LM stenosis, or ≥70% proximal left anterior descending [LAD] stenosis).
Beyond stenosis, all coronary segments were assessed for HRP features (positive remodeling, spotty calcification, low CT attenuation <30 HU and napkin-ring sign) as previously defined (12). As per the CAD-RADS definition, the “V” modifier for “vulnerable plaque” was defined as 2 or more HRP features in at least 1 coronary plaque/segment (17).
Study endpoint
The endpoint was a composite of death from any cause, myocardial infarction (MI), or hospitalization for unstable angina (UAP). An independent clinical events committee adjudicated all endpoints in a blinded fashion on the basis of standard prospectively determined definitions (2,18).
Statistical analysis
Continuous variables are presented as mean ± SD. Categorical and ordinal variables are presented as frequencies and proportions. Comparisons among groups were performed using an independent sample Student's t-test for continuous variables, the Fisher’s exact test for categorical variables, and the Wilcoxon rank-sum test for ordinal variables.
Cox proportional hazards regression models were used to calculate hazard ratios unadjusted and adjusted for atherosclerotic cardiovascular disease (ASCVD) risk score with 95% confidence intervals and assess the relationship of test results to the time to the first clinical event (or censoring) (20). Cumulative event rates based on test results were computed for each testing strategy (CACS, stenosis, or CAD-RADS) using the Kaplan-Meier method (21).
The discriminatory value of traditional and CAD-RADS grading schemes for the composite outcome was assessed using the C-statistic (22,23). Because of low individual prevalence, CAD-RADS categories 4b and 5 were combined as 1 composite category. A stepwise C-statistic comparison between nested models assessed the incremental prognostic value of ASCVD, CACS, and CAD-RADS over ASCVD risk score. C-statistics were compared using “somersd” and “lincom” packages in Stata (SE 14.2, StataCorp LP, College Station, Texas). The Stata routines used to compare the C-statistics account for the nested-model structure.
A 2-sided p value of <0.05 was considered to indicate statistical significance. All analyses were performed using Stata.
Results
Of 4,996 PROMISE patients randomized to an anatomic testing strategy (CTA), 3,840 patients (77%) were included in the analysis. The reasons for exclusion are provided in Figure 1. Excluded patients were older, differed in risk profile and presenting symptoms, and had a higher prevalence of statin therapy compared with included patients (Supplemental Table 1).
Of those patients included in the study, the mean age was 60.4 ± 8.2 years, and 49% (1,868 of 3,840) were men (Table 1). Over median follow-up of 25 months (interquartile range 18 to 34 months), 115 patients (3.0%) experienced the composite outcome, including 53 (1.4%) all-cause deaths, 29 (0.8%) cardiovascular deaths, 18 (0.5%) MIs, and 46 (1.2%) admissions for UAP.
| Demographics | |
| Age (yrs) | 60.4 ± 8.2 |
| Male | 1,868/3,840 (48.7) |
| Racial or ethnic minority | 861/3,814 (22.6) |
| Cardiac risk factors | |
| BMI (kg/m2)∗ | 30.3 ± 5.8 |
| Hypertension | 2,461/3,840 (64.1) |
| Diabetes | 778/3,840 (20.3) |
| Dyslipidemia | 2,588/3,840 (67.4) |
| Family history of premature CAD† | 1,272/3,829 (33.2) |
| Peripheral or cerebrovascular disease | 192/3,839 (5.0) |
| CAD equivalent‡ | 920/3,840 (24.0) |
| History of heart failure | 150/3840 (3.9) |
| Metabolic syndrome§ | 1,393/3,840 (36.3) |
| Current or past tobacco use | 1,977/3,839 (51.5) |
| Sedentary lifestyle‖ | 1,844/3.832 (48.1) |
| History of depression | 732/3,840 (19.1) |
| Risk factor burden and risk score¶ | |
| No risk factors | 97/3,840 (2.5) |
| Risk factor burden | 2.4 ± 1.1 |
| Combined Diamond-Forrester and Coronary Artery surgery risk score# | 53.0 ± 21.1 |
| Framingham risk score | |
| Low risk (<6%) | 253/3,834 (6.6) |
| Intermediate risk (6%-20%) | 2,006/3,834 (52.3) |
| High risk (>20%) | 1,575/3,834 (41.1) |
| ASCVD pooled cohort risk prediction (2013) | |
| Low risk (<7.5%) | 1,249/3,795 (32.9) |
| Elevated risk (≥7.5%) | 2,546/3,795 (67.1) |
| Relevant medications | |
| Beta-blocker | 911/3,675 (24.8) |
| ACE or ARB | 1,586/3,675 (43.2) |
| Statin | 1,670/3,675 (45.4) |
| Aspirin | 1,648/3,675 (44.8) |
| Clopidogrel | 48/3,675 (1.3) |
| Prasugrel | 1/3,675 (0.03) |
| Warfarin | 53/3,675 (1.4) |
| Primary presenting symptom and anginal type | |
| Chest pain | 2,810/3,837 (73.2) |
| Dyspnea on exertion | 547/3,837 (14.3) |
| Anginal type, site-reported∗∗ | |
| Typical | 408/3,840 (10.6) |
| Atypical | 3,021/3,840 (78.7) |
| Nonanginal | 411/3,840 (10.7) |
Prevalence of CAD
On coronary CTA, 1,303 (34%) of patients had no visible CAD, whereas 186 (4.8%) showed 70% to 99% stenosis (CAD-RADS 4a), and 54 (1.4%) showed ≥50% LM or ≥70% stenosis in 3 vessels (CAD-RADS 4b and 5). Using the traditional definition 1, 294 (7.7%) and 240 (6.3%) patients had moderate (50% to 69% stenosis) and severe stenosis (≥50% stenosis of LM or ≥70% in any major vessel), respectively. Using the traditional definition 2, moderately abnormal (≥70% stenosis in 1 major vessel) and severely abnormal (≥2 vessel disease (≥70%), or ≥50% LM stenosis, or ≥70% proximal LAD stenosis) were detected in 145 (3.8%) and 95 (2.5%) patients, respectively.
Any HRP feature was present in 1,938 (50.5%) patients, of whom 416 (21.5%) had at least 1 coronary segment with 2 or more HRP features. The number of patients with presence of 2 or more HRP features per segment gradually increased across CAD-RADS categories from 9.1% (112 of 1,236) for CAD-RADS 1% to 37.0% (20 of 54) for CAD-RADS 4b and 5 (p < 0.001).
Downstream invasive angiography and revascularization
Overall, 437 (11.4%) patients underwent invasive coronary angiography (ICA), of whom 217 (49.7%) were revascularized. As displayed in Supplemental Table 2, the rate of ICA as well as the percent leading to revascularization increased across CAD-RADS categories (both p < 0.001), with CAD-RADS 4a and 4b and 5 showing the highest rates of ICA (61.8% [115 of 186] and 64.8% [35 of 54]) and revascularization (50.5% [94 of 186] and 46.3% [25 of 54]), respectively. The proportion of ICA leading to revascularization increased from 10% (6 of 56) for CAD-RADS 1% to 82% (94 of 115) for CAD-RADS 4a and 71% (25 of 35) for CAD-RADS 4b and 5.
Prognostic value of presence and extent of CAD
Higher CAD stenosis category by CTA was significantly associated with the composite endpoint in univariate and multivariate analysis (adjusted for ASCVD risk score) for all definitions used to categorize degree of stenosis (traditional definition 1, traditional definition 2, and CAD-RADS categories) as displayed in Table 2. The risk for the composite endpoint increased from HR 2.43 (95% CI: 1.16 to 5.08) for CAD-RADS 1 to HR 21.84 (95% CI: 8.63 to 55.26) for CAD-RADS 4b and 5. The presence of HRP features showed significant associations to the time to event (Supplemental Figure 1) and was significantly associated with a higher hazard for the composite endpoint in univariate (HR 3.08; 95% CI: 2.04 to 4.65; p < 0.001) and multivariate analysis (HR 2.61; 95% CI: 1.71 to 3.98); p < 0.001 (Table 2).
| Events | Unadjusted | Adjusted for ASCVD Risk Score (Continuous Variable) | |||
|---|---|---|---|---|---|
| HR (95% CI) | p Value | HR (95% CI) | p Value | ||
| Degree of stenosis by CTA with following definitions | |||||
| Traditional definition 1 | |||||
| No CAD | 10/1,303 (0.8) | Base | --- | Base | --- |
| Mild CAD (1%–49% stenosis) | 58/2,003 (2.9) | 3.82 (1.95–7.48) | <0.001 | 3.40 (1.73–6.70) | <0.001 |
| Moderate CAD (50%–69% stenosis) | 18/294 (6.1) | 8.25 (3.81–17.86) | <0.001 | 6.91 (3.14–15.18) | <0.001 |
| Severe CAD (≥50% stenosis of LM or ≥70% in any major vessel) | 29/240 (12.1) | 17.61 (8.58–36.14) | <0.001 | 13.26 (6.28–28.00) | <0.001 |
| Traditional definition 2 | |||||
| Normal (No CAD) | 10/1,303 (0.8) | Base | --- | Base | --- |
| Mildly abnormal (1%–69% stenosis in any major vessels or <50% LM stenosis) | 76/2,297 (3.3) | 4.38 (2.26–8.46) | <0.001 | 3.82 (1.96–7.44) | <0.001 |
| Moderately abnormal (≥70% in 1 major vessel) | 19/145 (13.1) | 19.21 (8.93–41.32) | <0.001 | 15.18 (6.92–33.29) | <0.001 |
| Severely abnormal (≥2 vessel disease [≥70%] or ≥50% left main stenosis or ≥70% proximal LAD stenosis) | 10/95 (10.5) | 15.20 (6.33–36.53) | <0.001 | 9.62 (3.72–24.84) | <0.001 |
| CAD-RADS stenosis categories | |||||
| No plaque/stenosis: CAD-RADS 0 | 10/1,303 (0.8) | Base | --- | Base | --- |
| 1%–29%: CAD-RADS 1 | 25/1,236 (2.0) | 2.66 (1.28–5.54) | 0.009 | 2.43 (1.16–5.08) | 0.019 |
| 30%–49%: CAD-RADS 2 | 33/767 (4.3) | 5.70 (2.81–11.57) | <0.001 | 5.02 (2.45–10.28) | <0.001 |
| 50%–69%: CAD-RADS 3 | 18/294 (6.1) | 8.25 (3.81–17.87) | <0.001 | 7.03 (3.20–15.47) | <0.001 |
| 70%–99%: CAD-RADS 4a | 19/186 (10.2) | 14.38 (6.69–30.93) | <0.001 | 11.39 (5.16–25.12) | <0.001 |
| ≥50% LM or ≥70% in 3 vessels or total occlusion: CAD-RADS 4b + 5 | 10/54 (18.5) | 30.80 (12.81–74.07) | <0.001 | 21.84 (8.63–55.26) | <0.001 |
| High-risk plaque features∗ | |||||
| Absence of “vulnerable plaque” | 84/3,424 (2.5) | Base | --- | Base | --- |
| Presence of “vulnerable plaque” | 31/416 (7.5) | 3.08 (2.04–4.65) | <0.001 | 2.61 (1.71–3.98) | <0.001 |
Discriminatory capacity of stratification of CAD to predict events
The capacity to discriminate future events (all-cause death, MI, UAP) for the traditional definition 1, traditional definition 2, and CAD-RADS categories were c-statistic 0.717 (95% CI: 0.673 to 0.760), c-statistic 0.698 (95% CI: 0.658 to 0.739), and c-statistic 0.747 (95% CI: 0.703 to 0.792) respectively. CAD-RADS had significantly higher discriminatory value compared with both traditional definitions (p ≤ 0.001).
As determined by log-rank test in Kaplan-Meier estimates, CAD-RADS categories also showed significant associations to the time to event, with an increase in risk for the composite endpoint with the next higher category (p < 0.001) (Figures 2A to 2C).
Kaplan-Meier Estimates
Composite outcome estimates (death, MI, and UAP) by severity of stenosis using the (A) traditional definition 1,* (B) traditional definition 2,† and (C) CAD-RADS. *No CAD (0%); mild CAD (1% to 49% stenosis); moderate CAD (50% to 69% stenosis in any major vessels/branch); severe CAD (≥50% stenosis of left main [LM] or ≥70% in any major vessel/branch) †Normal (absence of coronary atherosclerosis), mildly abnormal (nonobstructive CAD: 1% to 69% stenosis in any major vessels/branch or <50% LM stenosis), moderately abnormal (obstructive CAD: ≥70% stenosis in 1 major vessel/branch), and severely abnormal (high-risk CAD: 2 or more vessel disease (≥70%) or ≥50% LM stenosis or ≥70% proximal LAD stenosis) CAD-RADS = coronary artery disease reporting and data system; CAD-RADS 0 = no plaque/stenosis; CAD-RADS 1 = 1% to 29% stenosis; CAD-RADS 2 = 30% to 49% stenosis; CAD-RADS 3 = 50% to 69% stenosis; CAD-RADS 4a = 70% to 99% stenosis; CAD-RADS 4b and 5: ≥50% LM stenosis or ≥70% stenosis in 3 vessels or total occlusion.
Incremental value of CAD-RADS stenosis categories beyond CACS
Prevalence of CAC and the association to clinical events
Of 3,840 patients, 1,498 (39%) had CACS of 0. The CACS was 1 to 100 in 1,160 (30%), 101 to 400 in 676 (18%), and >400 in 506 (13%) patients. Across CACS categories, the prevalence of CAD significantly increased (p < 0.001) (Table 3). In patients with CACS of 0, 87% were free of CAD (CAD-RADS 0), whereas 12.3% (184 of 1,498) had nonobstructive disease (CAD-RADS 1 to 3), and 0.7% (11of 1,498) had obstructive disease (CAD-RADS 4 and 5). In 9.2% (17 of 184) of those patients with CACS of zero and nonobstructive disease (CAD-RADS 1 to 3), 2 or more HRP features were present, whereas HRP were present in 54.5% (6 of 11) of patients with a CAC of zero and obstructive disease (CAD-RADS 4 and 5).
| CAD-RADS Stenosis Categories | All Patients (N = 3,840) | CACS Strata | |||
|---|---|---|---|---|---|
| CACS 0 (n = 1,498) | CACS 1–100 (n = 1,160) | CACS >100–400 (n = 676) | CACS >400 (n = 506) | ||
| No plaque/stenosis: CAD-RADS 0 | 1,303 (33.9) | 1,303 (87.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| 1%–29%: CAD-RADS 1 | 1,236 (32.2) | 142 (9.5) | 774 (66.7) | 254 (37.6) | 66 (13.0) |
| 30%–49%: CAD-RADS 2 | 767 (20.0) | 35 (2.3) | 273 (23.5) | 261 (38.6) | 198 (39.1) |
| 50%–69%: CAD-RADS 3 | 294 (7.7) | 7 (0.5) | 68 (5.9) | 93 (13.8) | 126 (24.9) |
| 70%–99%: CAD-RADS 4a | 186 (4.8) | 10 (0.7) | 36 (3.1) | 54 (8.0) | 86 (17.0) |
| ≥50% LM or ≥70% in 3 vessels or total occlusion CAD-RADS 4b+5 | 54 (1.4) | 1 (0.1) | 9 (0.8) | 14 (2.1) | 30 (5.9) |
Overall, the incidence of the composite endpoint increased across CACS categories (p < 0.001) as well as CAD-RADS stenosis categories within each CAC group (p < 0.001), as displayed in Table 4. Among patients with CAC scores of zero, 1.5% (22 of 1,498) experienced events. Ten of these patients had no visible CAD on CT (CAD-RADS 0). Nevertheless, the incidence of the primary endpoint increased with severity of CAD, reflected by increasing hazard ratios from 5.7 (95% CI: 2.3 to 14.5) for nonobstructive CAD (CAD-RADS 1 to 3) to 58.0 (95% CI: 18.1 to 185.3) for obstructive CAD (CAD-RADS 4 and 5), with a p < 0.001 for comparison with patients without plaque.
| CAD-RADS Stenosis Categories | All Patients (N = 3,840) | CACS Strata | |||
|---|---|---|---|---|---|
| CACS 0 (n = 1,498) | CACS 1–100 (n = 1,160) | CACS >100–400 (n = 676) | CACS >400 (n = 506) | ||
| No plaque/stenosis: CAD-RADS 0 | 10/1,303 (0.8) | 10/1,303 (0.8) | 0/0 (---) | 0/0 (---) | 0/0 (---) |
| 1%–29%: CAD-RADS 1 | 25/1,236 (2.0) | 6/142 (4.2) | 12/774 (1.6) | 5/254 (2.0) | 2/66 (3.0) |
| 30%–49%: CAD-RADS 2 | 33/767 (4.3) | 2/35 (5.7) | 9/273 (3.3) | 14/261 (5.4) | 8/198 (4.0) |
| 50%–69%: CAD-RADS 3 | 18/294 (6.1) | 0/7 (0.0) | 2/68 (2.9) | 8/93 (8.6) | 8/126 (6.4) |
| 70%–99%: CAD-RADS 4a | 19/186 (10.2) | 3/10 (30.0) | 2/36 (5.6) | 7/54 (13.0) | 7/86 (8.1) |
| ≥50% LM or ≥70% in 3 vessels or total occlusion CAD-RADS 4b+5 | 10/54 (18.5) | 1/1 (100.0) | 2/9 (22.2) | 3/14 (21.4) | 4/30 (13.3) |
| Total | 115/3,840 (3.0) | 22/1,498 (1.5) | 27/1,160 (2.3) | 37/676 (5.5) | 29/506 (5.7) |
Incremental value of CTA using CAD-RADS categories beyond CACS to predict events
Information about the burden of CAC (strata) significantly improved the discriminatory capacity of the ASCVD pooled cohort risk calculator to predict the composite endpoint of all-cause death, MI, and UAP (c-statistic 0.629 [95% CI: 0.572 to 0.687] vs. 0.682 [95% CI: 0.629 to 0.735]; p = 0.008), as displayed in Tables 5 and 6. Data from CTA using CAD-RADS categories (including HRP features) further incrementally increased the prognostic value to predict the composite endpoint (C-statistic 0.776; 95% CI: 0.734 to 0.818; p < 0.001) (Central Illustration). Stratified by CAC categories, the significant incremental value of coronary CTA over ASCVD risk stratification persisted as listed in Table 6.
Prognostic Value of Coronary CTA Using Coronary Artery Disease Reporting and Data System in PROMISE
In this large contemporary trial of patients with stable chest pain randomized to coronary CTA, the Coronary Artery Disease Reporting and Data System (CAD-RADS) including information about presence of high-risk plaque features (HRP) had higher prognostic value and discriminatory ability for future MACE (c-statistic 0.747) as compared to traditional stenosis-based assessment (c-statistic 0.698 to 0.717; p for comparison <0.001), (Kaplan-Meier curves for CAD-RADS for the prediction of the composite endpoint of death, myocardial infarction, or hospitalization for unstable angina over a median follow-up of 25 months on the top). Moreover, CAD-RADS added prognostic value over the ASCVD risk score and CAC score (table on the bottom), across all CAC strata.
| Univariable Model | C-Statistic | Multivariable Models | C-Statistic | p Value (Difference Between Models)∗ |
|---|---|---|---|---|
| ASCVD | 0.629 (0.572–0.687) | |||
| CACS | 0.657 (0.606–0.708) | ASCVD+CACS | 0.682 (0.629–0.735) | 0.008 |
| CAD-RADS (+HRP) | 0.747 (0.703–0.792) | ASCVD+CACS+CAD-RADS(+HRP) | 0.776 (0.734–0.818) | <0.001 |
| CACS Strata | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| CACS 0 | CACS 1–100 | CACS >100–400 | CACS >400 | ||||||
| C-Stat | p Value | C-Stat | p Value | C-Stat | p Value | C-Stat | p Value | ||
| Model 1: ASCVD | 0.539 | 0.607 | 0.551 | 0.572 | |||||
| Model 2: ASCVD + CAD-RADS (+HRP) | 0.751 | 0.005 | 0.775 | 0.023 | 0.673 | 0.031 | 0.697 | 0.041 | |
Discussion
In a large contemporary trial of patients with stable chest pain randomized to coronary CTA, we found that the CAD-RADS reporting system had greater prognostic value and discriminatory ability for future MACE than previous traditional stenosis-based categories. This can be explained by CAD-RADS’s more granular grading of nonobstructive and obstructive CAD, inclusion of both stenosis and plaque burden components, and the inclusion of HRP features. Our second major finding is that CAD-RADS adds substantial prognostic value over the ASCVD risk score and CACS across all CACS strata. Together, these results support the prognostic value of CAD-RADS for standardized reporting of coronary CTA.
Strengths of this study include that it was conducted within a large multicenter randomized controlled trial at 193 sites, with prospective enrollment of patients, collection of CTA, and independent adjudication of adverse events. In this analysis, coronary CT was interpreted for CAC, coronary artery stenosis, and high-risk coronary plaque features by a central core laboratory with expert CT readers blinded to clinical information and outcomes. These factors may explain why our results differ from that in a recent analysis of the CONFIRM registry, which found CAD-RADS did not have greater discriminatory value for MI or death (C-statistic CAD-RADS 0.705 vs. traditional 0.710; p = 0.78) (24). In the CONFIRM analysis, CTA stenosis was graded by local site physicians, who tend to call severe stenosis more often than blinded expert core laboratory readers (19). Furthermore, the CONFIRM analysis did not include HRP features, which are a part of CAD-RADS, because of their known prognostic value (12,25–28).
Our results should be interpreted in the context of previous PROMISE publications that assessed the prognostic value of CTA. First, Hoffmann et al.(14) compared the prognostic value of CTA with functional testing, finding that CTA had greater prognostic value. In contrast to the current study, this analysis used the local site interpretations of CTA and functional testing as well as disease categories tailored to allow comparison between CTA and functional testing, which may not reflect how coronary CTA is currently interpreted. A subsequent paper by Lu et al. found that blinded expert central core laboratory interpretation of coronary CTA found 41% fewer patients with stenosis ≥50% than the site readers, yet with better accuracy using quantitative invasive coronary angiography as the reference standard (19). Ferencik et al. (12) found that HRP features on coronary CTA were associated with MACE after adjustment for stenosis and ASCVD risk score; in this study, HRP was defined differently (any plaque with at least 1 high-risk feature, not including spotty calcification) than in CAD-RADS (2 HRP features in a single segment, including spotty calcification). Finally, Budoff et al. (15) compared the coronary artery calcium score to functional testing for estimating prognosis, finding that most patients having events had calcium scores >0 compared with fewer than one-half with an abnormality on functional testing. However, whether coronary CTA adds prognostic value beyond coronary calcium was not assessed.
A second major finding of our study was that CAD-RADS had greater prognostic value than the CACS. Although the CACS is one of the best-studied prognostic imaging biomarkers in asymptomatic populations (29–31), how it relates to stenosis, HRP, and events in symptomatic chest pain populations is not as well established. Indeed, CACS had limited value for diagnosing stenosis, with only one-half of patients with CACS >400 having stenoses ≥50% (Table 3). For those with a CACS between 101 and 400, only a quarter had a stenosis ≥50%. On the other hand, 13% (195 of 1,498) of patients with CACS of zero had detectable coronary plaque, and 6% (12 of 195) of patients with CACS of zero but detectable plaque on CTA experienced events in our analysis, an event rate twice as high as the overall population. It should be noted that of the 1,498 studies with CACS of zero, only 21 had events (15), yet a majority of these (12 of 21) occurred in the group with CACS of zero and nonobstructive CAD on CTA. This finding is complementary to CONFIRM observational data, finding that a minority of symptomatic patients with CACS of zero have coronary plaque on CTA and that this noncalcified plaque is associated with increased cardiac events (32). CAD-RADS had substantial incremental prognostic value over CACS, and this was true in all CACS strata (Table 5). One mechanistic explanation could be the fact that the prevalence of HRP in patients with severe stenosis was the highest in patients without CAC (55%; 6 of 11) and decreased with increasing CACS (45%; 52 of 116), reflecting the potentially higher vulnerability of plaque in patients with less CAC. The CRESCENT (Computed Tomography versus Exercise Testing in Suspected Coronary Artery Disease) randomized controlled trial, which used a tiered CT approach in patients with stable angina, suggested no need for CTA in patients without CAC and low pre-test probability (<70%) for obstructive CAD (16). In this trial, 98 of 242 patients (39%) without CAC and low pre-test probability did not undergo downstream testing, and no adverse events occurred after a follow-up of 1 year. The shorter follow-up in CRESCENT in comparison with PROMISE (median follow-up: 25 months) might partially explain the demonstrated value of CTA in patients without calcification in our analysis, as cardiovascular events in low-risk groups might not be apparent for years. Also, it is possible that events in PROMISE patients would have been even higher without the use of CTA, as CTA was shown to be associated with a higher proportion of patients newly initiated on aspirin and statins compared with standard of care (4). This demonstrates the need for longer follow-up, especially in cohorts with stable chest pain and expected low incidence of events.
A recent 1,769-patient analysis of the SCOT-HEART trial (33) found that both obstructive (≥70% stenosis) CAD and adverse plaque were associated with cardiac events; however, these associations were not independent of the CACS. This contrasts with our result that CAD-RADS had significant incremental prognostic value beyond both the ASCVD risk score and CAC. Besides the differences in demographics and outcomes between the 2 trials, differences in how stenosis and HRP were defined may explain the discrepancy. In SCOT-HEART, stenosis was categorized as normal (0%), nonobstructive (1% to 69%), or obstructive (≥70%). This approach provides less granular information about the degree of nonobstructive stenosis and presence of multivessel disease than CAD-RADS. Furthermore, SCOT-HEART defined adverse plaque as the presence of at least 1 plaque with positive remodeling or low attenuation, in contrast to the CAD-RADS “V” modifier that requires at least 2 HRP features in a single coronary plaque segment. Nevertheless, the unadjusted hazard ratio for adverse/high-risk plaque was similar between the 2 trials (SCOT-HEART HR 3.01 [1.61 to 5.63]; p < 0.001, PROMISE HR 3.08 [2.04 to 4.65]; p < 0.001). In the end, CAD-RADS is currently the preferred reporting system for coronary CTA by the Society of Cardiovascular Computed Tomography (SCCT), American College of Radiology (ACR), North -American Society for Cardiovascular Imaging (NASCI), and American College of Cardiology (ACC); therefore, we believe it has the greatest clinical relevance.
Study limitations
First, PROMISE was a pragmatic trial in which the results of CTA were shared with patients and influenced management (34). A slightly modified CAD-RADS was evaluated, with a threshold of 30% instead of 25% between CAD-RADS 1 and 2. Overall, the prevalence of obstructive CAD was rather low, and CAD-RADS categories 4b and 5 were conflated because of small numbers. These data from a contemporary cohort of patients with stable chest pain reflects real-world practice. Patients with known CAD or previous interventions were excluded from PROMISE, and thus we could not investigate the prognostic value of the CAD-RADS S (stent) or G (graft) category modifiers.
Conclusions
In PROMISE, a large prospective trial of coronary CTA in patients with stable chest pain and suspected CAD, the CAD-RADS reporting system had greater prognostic value than ASCVD score, CACS, and traditional stenosis-based grading schemes.
Perspectives
COMPETENCY IN PATIENT CARE: For patients undergoing coronary CTA for stable chest pain, the CAD-RADS has greater prognostic value than traditional stenosis categories, the ASCVD risk score, and the CACS.
TRANSLATIONAL OUTLOOK: CAD-RADS is currently recommended by several societies for the reporting of CAD on CTA; further research is necessary to determine whether improved prognostic value translates into improved clinical decision making and outcomes.
Abbreviations and Acronyms
| ASCVD | atherosclerotic cardiovascular disease |
| CACS | coronary artery calcium score |
| CAD | coronary artery disease |
| CAD-RADS | Coronary Artery Disease Reporting and Data System |
| CTA | computed tomography angiography |
| HRP | high-risk plaque |
| LM | left main coronary artery |
| MACE | major adverse cardiovascular event |
| UA | unstable angina |
References
1. "2012 ACCF/AHA/ACP/AATS/PCNA/SCAI/STS Guideline for the diagnosis and management of patients with stable ischemic heart disease: 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:e44-e164.
2. "Outcomes of anatomical versus functional testing for coronary artery disease". N Engl J Med 2015;372:1291-1300.
3. The SCOT-HEART Investigators. "CT coronary angiography in patients with suspected angina due to coronary heart disease (SCOT-HEART): an open-label, parallel-group, multicentre trial". Lancet 2015;385:2383-2391.
4. "Coronary CT angiography and 5-year risk of myocardial infarction". N Engl J Med 2018;379:924-933.
5. National Institute for Health and Clinical Excellence. "Chest pain of recent onset: assessment and diagnosis of recent onset chest pain or discomfort of suspected cardiac origin (update)". CG95 London: National Institute for Health and Clinical Excellence, 2016.
6. "Diagnostic accuracy of 64-slice computed tomography coronary angiography: a prospective, multicenter, multivendor study". J Am Coll Cardiol 2008;52:2135-2144.
7. "Diagnostic performance of coronary angiography by 64-row CT". N Engl J Med 2008;359:2324-2336.
8. "Diagnostic performance of 64-multidetector row coronary computed tomographic angiography for evaluation of coronary artery stenosis in individuals without known coronary artery disease: results from the prospective multicenter ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) trial". J Am Coll Cardiol 2008;52:1724-1732.
9. "In vivo CT detection of lipid-rich coronary artery atherosclerotic plaques using quantitative histogram analysis: a head to head comparison with IVUS". Atherosclerosis 2011;215:110-115.
10. "Comprehensive plaque assessment by coronary CT angiography". Nat Rev Cardiol 2014;11:390-402.
11. "Histogram analysis of lipid-core plaques in coronary computed tomographic angiography: ex vivo validation against histology". Invest Radiol 2013;48:646-653.
12. "Use of high-risk coronary atherosclerotic plaque detection for risk stratification of patients with stable chest pain: a secondary analysis of the PROMISE randomized clinical trial". JAMA Cardiol 2018;3:144-152.
13. "Plaque characterization by coronary computed tomography angiography and the likelihood of acute coronary events in mid-term follow-up". J Am Coll Cardiol 2015;66:337-346.
14. "Prognostic value of noninvasive cardiovascular testing in patients with stable chest pain: insights from the PROMISE trial (Prospective Multicenter Imaging Study for Evaluation of Chest Pain)". Circulation 2017;135:2320-2332.
15. "Prognostic value of coronary artery calcium in the PROMISE study (Prospective Multicenter Imaging Study for Evaluation of Chest Pain)". Circulation 2017;136:1993-2005.
16. "Calcium imaging and selective computed tomography angiography in comparison to functional testing for suspected coronary artery disease: the multicentre, randomized CRESCENT trial". Eur Heart J 2016;37:1232-1243.
17. "CAD-RADS Coronary Artery Disease Reporting and Data System: an expert consensus document of the Society of Cardiovascular Computed Tomography (SCCT), the American College of Radiology (ACR) and the North American Society for Cardiovascular Imaging (NASCI). Endorsed by the American College of Cardiology". J Cardiovasc Comput Tomogr 2016;10:269-281.
18. "PROspective Multicenter Imaging Study for Evaluation of chest pain: rationale and design of the PROMISE trial". Am Heart J 2014;167:796-803.e1.
19. "Central core laboratory versus site interpretation of coronary CT angiography: agreement and association with cardiovascular events in the PROMISE trial". Radiology 2018;287:87-95.
20. "Regression models and life-tables". J R Stat Soc 1972;34:187-220.
21. "Nonparametric estimation from incomplete observations". J Am Stat Assoc 1958;53:457-481.
22. "Evaluating the yield of medical tests". JAMA 1982;247:2543-2546.
23. "Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors". Stat Med 1996;15:361-387.
24. "The coronary artery disease-reporting and data system (CAD-RADS): prognostic and clinical implications associated with standardized coronary computed tomography angiography reporting". J Am Coll Cardiol Img 2018;11:78-89.
25. "Computed tomographic angiography characteristics of atherosclerotic plaques subsequently resulting in acute coronary syndrome". J Am Coll Cardiol 2009;54:49-57.
26. "High-risk plaque detected on coronary CT angiography predicts acute coronary syndromes independent of significant stenosis in acute chest pain: results from the ROMICAT-II trial". J Am Coll Cardiol 2014;64:684-692.
27. "Characteristics of high-risk coronary plaques identified by computed tomographic angiography and associated prognosis: a systematic review and meta-analysis". Eur Heart J Cardiovasc Img 2016;17:120-129.
28. "Computed tomographic coronary angiography-derived plaque characteristics predict major adverse cardiovascular events: a systematic review and meta-analysis". Circ Cardiovasc Img 2018;11:e006973.
29. "Coronary calcium as a predictor of coronary events in four racial or ethnic groups". N Engl J Med 2008;358:1336-1345.
30. "Cardiovascular event prediction and risk reclassification by coronary, aortic, and valvular calcification in the Framingham Heart Study". J Am Heart Assoc 2016;5.
31. "Ten-year association of coronary artery calcium with atherosclerotic cardiovascular disease (ASCVD) events: the multi-ethnic study of atherosclerosis (MESA)". Eur Heart J 2018;39:2401-2408.
32. "Prevalence and severity of coronary artery disease and adverse events among symptomatic patients with coronary artery calcification scores of zero undergoing coronary computed tomography angiography: results from the CONFIRM (Coronary CT Angiography Evaluation for Clinical Outcomes: An International Multicenter) registry". J Am Coll Cardiol 2011;58:2533-2540.
33. "Coronary artery plaque characteristics associated with adverse outcomes in the SCOT-HEART study". J Am Coll Cardiol 2019;73:291-301.
34. "Changes in medical therapy and lifestyle after anatomical or functional testing for coronary artery disease". J Am Heart Assoc 2016;5:e003807.
Appendix
For a supplemental figure and tables, please see the online version of this paper.
Footnotes
The parent PROMISE trial was supported by National Heart, Lung, and Blood Institute grants R01HL098237, R01HL098236, R01HL98305, and R01HL098235. Dr. Bittner was supported by NIH/NHLBI 5K24HL113128. Dr. Budoff has received grants from NIH during the conduct of the study and grants from General Electric outside the submitted work. Dr. Ferencik was supported by American Heart Association Grant 13FTF16450001; and has received grants from American Heart Association outside the submitted work. Dr. Douglas has received grants from HeartFlow outside the submitted work. Dr. Hoffmann was supported by K24HL113128; and has received grants from NIH-NHLBI National Heart, Lung, and Blood Institute, during the conduct of the study; grants from Duke University/Abbott US; grants from HeartFlow, Inc. and Kowa Company, Ltd.; grants and nonfinancial support from MedImmne, LLC.; and personal fees and nonfinancial support from Abbott US, outside the submitted work. Dr. Lu was supported by the American Heart Association Precision Medicine Institute 18UNPG34030172 and the Harvard University Center for- AIDS Research 5P30AI060354-14; and has received research support to his institution from the American Heart Association, Nvidia, Kowa, and Medimmune, outside the submitted work. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
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 JACC: Cardiovascular Imaging author instructions page.
