Artificial Intelligence Enabled CT Based Quantitative Coronary Topography (AI-QCT)/Coronary Plaque Analysis (AI-CPA)

Title XVIII of the Social Security Act, §1862(a)(1)(A) allows coverage and payment for only those services that are reasonable and necessary for the diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member.

Title XVIII of the Social Security Act, §1862(a)(1)(D) Items and services related to research and experimentation.

Title XVIII of the Social Security Act, §1862(a)(7) states Medicare will not cover any services or procedures associated with routine physical checkups.

Title XVIII of the Social Security Act, §1862(I)(5)(D) The 21st Century Cures Act of 2016 (Public Law 114-255) added language directing the Secretary of the Department of Health and Human Services (DHHS) to improve the transparency of the LCD process.

42 CFR §410.32 indicates that diagnostic tests may only be ordered by the treating physician (or other treating practitioner acting within the scope of his or her license and Medicare requirements).

CMS Internet-Only Manual, Pub. 100-03, Medicare National Coverage Determinations (NCD) Manual, Chapter 1, Part 4, §220 Radiology and §220.1 Computed Tomography (CT).

CMS Internet-Only Manual, Pub. 100-08, Medicare Program Integrity Manual, Chapter 13, §13.5.4 Reasonable and Necessary Provisions in LCDs.

Protecting Access to Medicare Act (PAMA) of 2014, Section 218(b), established a new program to increase the rate of appropriate advanced diagnostic imaging services provided to Medicare beneficiaries.

Artificial Intelligence Enabled CT Based Quantitative Coronary Topography/Coronary Plaque Analysis (AI-QCT/AI-CPA) using coronary computed tomography angiography (CCTA)* is considered reasonable and medically necessary as a diagnostic study when:

1. The patient has acute or stable chest pain with no known coronary artery disease (CAD)¹ and is eligible for CCTA*, AND
2. CCTA classifies patient as:
   • Intermediate risk** OR
   • CAD-RADS 1, CAD-RADS 2 or CAD-RADS 3 category*** on CCTA^1,2, AND
3. Cardiac evaluation is negative or inconclusive for acute coronary syndrome (ACS)¹.

*See Local Coverage Determination (LCD), Cardiac Computed Tomography & Angiography (CCTA) L33423 for criteria for CCTA.

AI-QCT/AI-CPA should not be performed until after the base study (CCTA) has been completed and interpreted. Software to perform AI-QCT/AI-CPA must be FDA cleared or approved.

**Intermediate and high-risk as defined in the 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the Evaluation and Diagnosis of Chest Pain¹

*** CAD-RADS 1- CAD-RADS 3 category as defined by CAD-RADS™ 2.0–2022 Coronary Artery Disease Reporting and Data System (CAD-RADS): an Expert Consensus document of the Society of Cardiovascular Computed Tomography (SCCT), the American College of Cardiology (ACC), the American College of Radiology (ACR), and the North America Society of Cardiovascular Imaging (NASCI).^2,3

Limitations

AI-QCT/AI-CPA is NOT considered reasonable or necessary in the following clinical circumstance(s) (non-covered):

1. Screening (i.e., in the absence of signs, symptoms, or disease);
2. When there are contraindications to CCTA*;
3. In conjunction with invasive coronary catheterization;
4. In the presence of normal CCTA results (CAD RADS=0 or no plaque disease);
5. In the presence of high-grade stenosis (>70%) or CAD-RADS 4 and CAD-RADS 5;
6. Within 30 days of a myocardial infarction (MI);
7. In the presence of unstable coronary symptoms; and
8. For disease surveillance.

*See Cardiac Computed Tomography & Angiography (CCTA) L33423 for a full listing of contraindications to CCTA.

Definitions

Artificial Intelligence Enabled CT Based Quantitative Coronary Topography (AI-QCT)/Coronary Plaque Analysis (AI-CPA) - Artificial intelligence application to imaging obtained through coronary computed tomography (CT) scans to calculate coronary artery dimensions and degree of stenosis per vessel and coronary plaque composition and burden.⁴

Calcified Plaque - Higher density plaque, mostly composed of calcium, thought to be associated with lower clinical risk than non-calcified plaque. Traditionally, the overall burden of calcified plaque has been assessed indirectly through a coronary artery calcium score (CACS) or using cut off >350 Hounsfield Units (HU).⁵ There is both calcified and non-calcified tissue present.⁶

Coronary Artery Disease (CAD) - Narrowing of the coronary arteries usually caused by plaque and atherosclerosis that can lead to ischemia of the heart.¹

• Known CAD includes patient with prior anatomic testing with identified nonobstructive atherosclerotic plaque and obstructive CAD.¹

Coronary Artery Disease Reporting and Data System (CAD-RADS) - A standardized method to communicate findings of CCTA.²

Coronary Computed Tomography Angiography (CCTA) - A non-invasive test using advanced computed tomography angiography imaging to view the tissues and blood vessels of the heart. This can be used to determine the presence and extent of CAD.

Coronary Plaque Analysis (CPA) - Analysis of coronary plaque composition and burden.

Fibrotic Plaque - A plaque with density of 131-350 HU.

High Risk Plaque (HRP) - High risk plaque findings include napkin-ring sign, low-attenuation plaque (LAP), positive vessel remodeling, low CT attenuation and spotty calcification.^5,7

Invasive Coronary Angiography (ICA) - Invasive procedure done at the time of cardiac catheterization to look at the arteries of the heart and can determine the presence and extent of CAD.

Low Attenuation Plaque (LAP) - Low density plaque with dark appearance on CCTA and higher lipid content usually defined as attenuation of <30 HU and associated with higher clinical risk. Density is measured in Hounsfield Units (HU) and low attenuation is -50 to 50 HU.^5,8

Major Adverse Cardiac Events (MACE) - Fatal and non-fatal MI. Some studies also include unstable angina requiring hospitalization or revascularization.⁷

Non-Calcified Plaque (NCP) - A lower density plaque, often earlier in development and associated with higher clinical risk with density of 50-130 HU.^5,8 There is no discernible calcification present.⁶

Nonobstructive CAD - CAD with <50% stenosis.¹

Obstructive CAD - CAD with >50% stenosis¹

Plaque- The presence of tissue structures ≥ 1mm² within or adjacent to the coronary artery lumen, identified and at least 2 independent planes, that can be distinguished from the surrounding tissues (epicardial fat) and the lumen.⁶

Quantitative Coronary Plaque Analysis (QCPA) - Imaging technique that provides objective and reproducible measurements of coronary artery dimensions and composition of the atherosclerotic plaques.

Quantitative Coronary Topography (QCT) - CT scan imaging that provides objective and reproducible measurements of the coronary artery dimensions and degree of stenosis per vessel.

Provider Qualifications

The Medicare Program Integrity Manual states services will be considered medically reasonable and necessary only if performed by appropriately trained providers.

Patient safety and quality of care mandate that healthcare professionals who interpret CCTA and QCPA and AI-QCT/AI-CPA are appropriately trained and/or credentialed by a formal residency/fellowship program. Credentialing or privileges are required for procedures performed in inpatient and outpatient settings.⁴

All aspects of care must be within the provider’s medical licensure and scope of practice. Reimbursement for procedures utilizing imaging techniques may be made to providers who meet training requirements for the procedures in this policy only if their respective state allows such in their practice act and formally licenses or certifies the practitioner to use and interpret these imaging modalities. At a minimum, training must cover and develop an understanding of anatomy and drug pharmacodynamics and kinetics as well as proficiency in diagnosis and management of disease, the technical performance of the procedure, and utilization of the required associated imaging modalities. Supervision, interpretation, and reports shall/must be performed by a physician with the advanced training requirements and or credentialing for CCTA and AI-QCT/AI-CPA. The technical and professional portions must meet the criteria for performance for CCTA (L33423).

Providers must also meet the U.S Food and Drug Administration (FDA) requirements which includes “The software is not intended to replace the skill and judgement of a qualified medical practitioner and should only be used by people who have been appropriately trained in the software’s functions, capabilities and limitations.”⁹ Radiology technicians must also meet all training requirements for performance of AI-QCT/AI-CPA.

Notice: Services performed for any given diagnosis must meet all the indications and limitations stated in this LCD, the general requirements for medical necessity as stated in CMS payment policy manuals, all existing CMS national coverage determinations (NCD), and all Medicare payment rules.

A Contractor Advisory Committee (CAC) Meeting “Non-Invasive Technology for Coronary Artery Plaque Analysis” was hosted 5/25/23 by CGS Administrators, Noridian Healthcare Solutions, National Government Services, Palmetto GBA, and WPS Government Health Administrators. The transcript and audio are available on each Medicare Administrative Contractor’s (MACs) website.

Background

CCTA has become an effective gate keeper for invasive angiography helping guide referral for obstructive CAD. It has been demonstrated to be superior to exercise electrocardiography and single-photon emission computed tomography (SPECT) for detection of obstructive CAD (>50% stenosis).^10,11 Studies have demonstrated excellent prognostic value of a normal CCTA for both short and long term mortality rates.¹² The updated 2021 American College of Cardiology and American Heart Association Chest Pain Guideline¹ states CCTA has become a first line tool in evaluation of acute and chronic CAD particularly in symptomatic patients with stable symptoms and intermediate or high pre-test probability of obstructive CAD, or among intermediate-risk acute chest pain patients.

CCTA can also provide information on plaque burden and adverse coronary artery plaque characteristics which has been demonstrated to be an independent predictor of disease and prognosis.¹² The characterization of coronary atherosclerotic plaques can be calculated from CCTA, but the process is time consuming and often with variable results.^4,13 At least 5 different software have been developed as an adjunct to CCTA to aid in the visualization, reduce evaluation time, and improve accuracy of this assessment. The gold standard is considered intravascular ultrasound (IVUS) and optical coherence tomography (OCT) which are the tools that are often used to validate the software.⁵ This is intended to improve clinical diagnosis and management of CAD.¹⁴

Applications and Limitations of CCTA

SCOT-HEART trial demonstrated that the addition of CCTA to standard of care significantly improved diagnostic certainty of angina.^15,16 In this prospective, open labeled, parallel group, multicentered trial 4,146 patients with stable chest pain received standard of care (SOC) plus CCTA (n=2073) or SOC alone (n=2073). Primary end point was death from CAD or nonfatal MI at 5 years. They reported that the 5-year death rate was lower in the CCTA group as compared to the SOC group (2.3% [48 patients] vs. 3.9% [81 patients]; hazard ratio, 0.59; 95% confidence interval [CI], 0.41 to 0.84; p = 0.004). Rates of ICA and coronary revascularization were higher in the CCTA group compared to SOC in the first few months of follow-up, with no differences in the overall use of ICA and coronary revascularization reported at 5 years. During follow up, CCTA assigned patients were more likely to have initiated preventive therapies when compared to the SOC alone group (19.4% [402 patients] vs. 14.7% [305 patients]; odds ratio, 1.40; 95% confidence interval [CI], 1.19 to 1.65) and antianginal therapies (13.2% [273 patients] vs. 10.7% [221 patients]; odds ratio, 1.27; 95% CI, 1.05 to 1.54). The authors conclude the use of CCTA resulted in more correct diagnoses of coronary heart disease than standard care alone, increase use of appropriate therapies, and the resultant change in management lead to fewer clinical events in the CCTA group than in the SOC group. They found CCTA was associated with a 38% reduction in coronary heart disease death and non-fatal MI, however the result fell just short of statistical significance (adjusted HR 0.62, 95% CI 0.38-1.01; p=0.527).¹⁶ In 1,769 patients with stable chest pain, it was reported that incident MI was more frequent in those with greater burden of atherosclerotic plaque consistent with invasive imaging.

Multiple evaluations of this data support a correlation with plaque burden and cardiovascular outcomes. LAP burden correlated weakly with cardiovascular risk score (r=0.34; P<0.001), strongly with coronary artery calcium score (r=0.62; P<0.001), and very strongly with the severity of luminal coronary stenosis (area stenosis, r=0.83; P<0.001). They conclude LAP burden is the strongest predictor of fatal or non-fatal MI.⁸ A post-hoc analysis aimed to investigate adverse coronary plaque characteristics in patients with suspected CAD found that among study participants 608 of 1,769 had 1 or more adverse plaque features. Coronary heart disease death or non-fatal MI was 3 times more frequent in patients with adverse plaque and twice as frequent in those with obstructive disease. The combination of obstructive disease and adverse plaque had the highest event rate with a 10-fold increase in MI compared to normal coronary arteries (HR: 11.50; 95% CI: 3.39 to 39.04; p < 0.001). There was no difference in the risk of events in patients with CACS > 100 AU. However for those with CACS <100 AU, adverse plaque was demonstrated as an increased risk of coronary heart disease death or non-fatal MI compared to those without adverse plaques (HR: 3.38; 95% CI: 1.13 to 10.08; p = 0.03).The authors explain that CACS has been used as a surrogate measure of overall plaque burden and plaque stability may be a more predictive marker than calcifications which may represent more stable plaques.¹⁷ Another post-hoc analysis reported CCTA had a stronger association with 5-year coronary heart disease death or nonfatal MI (hazard ratio, 10.63; 95% CI, 2.32-48.70; P = .002) than exercise electrocardiography (ECG) alone with the greatest difference in those with inconclusive results of exercise ECG (6 of 283 [2%] vs 18 of 283 [6%]), although this did not reach statistical significance (log-rank p=0.05).¹⁸ A study found that low-density plaque was associated with the greatest risk of future MI, independent of other risk factors including cardiovascular risk or, CACS and stenosis severity. A low-density plaque burden of >4% was associated with almost 5x increase risk of MI (HR, 4.65 [95% CI, 2.06–10.5]; P<0.001) and those with non-obstructive CAD were 6x more likely to experience an MI ((HR, 6.61 [95% CI, 1.91 to 22.82]; P=0.003).^8,19 Another report derived from the SCOT-HEART trial calculates a threshold of ≥238.5 mm³ is associated with a relative hazard of 5.4 (p=0.0042) determined from the 41 subjects in the cohort who experienced MI during the study period.²⁰

The Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) trial compared CCTA to functional testing for stable chest pain. This randomized control trial (RCT) enrolled 10,003 patients with stable chest pain who underwent CCTA or functional testing (exercise ECG, nuclear stress testing, or stress ECG) and followed them for 25 months. The RCT concluded that initial evaluation with CCTA did not improve clinical outcomes compared to functional testing.²¹ Sub-analysis of this population to explore prognostic value of these tests found the prevalence of normal test results and incidence rate of events in these patients were lower in the CCTA group (n=4500) in comparison to functional testing (n=4602) (33.4% versus 78.0%, and 0.9% versus 2.1%, respectively; both p<0.001). They reported that CCTA offered higher discriminatory ability in predicting events over functional test.²² A observational cohort within this same cohort explored the role of HRP in predicting MACE in this population. They found HRP was present in 676 (15%) and carried a 70% increased risk of MACE independent of cardiovascular risk factors (6.4% vs. 2.4%; hazard ratio, 2.73; 95% CI, 1.89-3.93).²³ They also reported there was no significant difference in MACE in patients with significant stenosis and HRP as opposed to those with significant stenosis without HRPs, and HRPs are a stronger predictor of MACE in women and younger patients. This data is limited by the low absolute MACE rate within the population and low positive predictive value of HRPs. Compared to the SCOT-HEART trial the shorter duration of follow-up (2 vs. 5 years) and low event rate in the PROMISE trial likely contributed to the variability within these results. This study is pertinent as it explores the role of plaques in predictors of MACE.

A systematic review and meta-analysis sought to better understand the impact of statins on percent atheroma volume (PAV) measured by IVUS. The end points of this investigation were change in PAV and MACE. Seventeen prospective studies were included (n=6333) with a duration from 11-104 weeks. Mean change in PAV, across the study arms, ranged from −5.6% to 3.1%. MACE ranged from 0 to 72 events per study arm: 13 study arms (38%) reported no events, 8 (24%) reported 1–2 events and 13 (38%) reported 3 or more events. Meta regression demonstrates a decrease in the odds of MACE associated with a reduction in mean PAV. The authors conclude based on the analysis that “a 1% reduction in mean PAV as induced by dyslipidemia therapies was associated with a 20% reduction in the odds of MACE.” This study is pertinent as it demonstrates a reduction in plaque volume utilizing the gold standard of IVUS as a result of medical management.²⁴

The role of plaque burden was further explored in the Progression of Atherosclerotic Plaque Determined by Computed Tomographic Angiography Imaging registry (PARADIGM study).²⁵ This was a prospective, multinational study that enrolled 2,252 patients at 13 sites without history of CAD who underwent serial CCTA at an interscan interval of >2 years. Plaques were analyzed for the percent diameter stenosis, PAV, plaque composition, and presence of HRP, defined by the presence of 2 or more of the following: LAP, positive arterial remodeling, or spotty calcifications. The population was further divided into statin-naive (n=474) and statin-taking (n=781) patients. They found that the group on statins experience a slower rate of overall PAV progression (1.76 ± 2.40% per year vs. 2.04 ± 2.37% per year, respectively; p=0.002), and annual incidence of new HRP features were lower at 0.9% per year vs. 1.6% per year, respectively; all p < 0.001). They reported more rapid progression of calcified PAV (1.27 ± 1.54% per year vs. 0.98 ± 1.27% per year, respectively; p < 0.001) with slower progression of noncalcified PAV in statin-taking patients (0.49 ± 2.39% per year vs. 1.06 ± 2.42% per year and 0.9% per year vs. 1.6% per year, respectively; all p < 0.001). The rates of progression to >50% diameter stenosis was not different (1.0% vs. 1.4%, respectively; p > 0.05) concluding that statins did not affect the progression of percentage of stenosis severity of coronary artery lesions but induced phenotypic plaque transformation. They conclude statins were associated with slower progression of overall coronary atherosclerosis volume, with increased plaque calcification and reduction of HRP features. This is pertinent as this study demonstrates a potential role of plaque analysis in clinical management of coronary lesions. A sub-analysis evaluated 1,166 patients who experienced 129 MACE events over 8.2 years of follow-result. They found whole heart PAV increased from baseline (2.32% to 4.04%) and was independently associated with MACE (OR 1.23 [95% CI 1.08, 1.39]).²⁶ Another analysis of 2,458 coronary lesions among 857 patients on statins reported that statin use was associated with greater rates of transformation towards more dense calcium and decrease in low-attenuation and fibro-fatty plaque volumes.²⁷

Several trials explore the effect of drugs for those already on statins and in need of escalating treatment.

The Effect of Alirocumab on Atherosclerotic Plaque Volume, Architecture and Composition (ARCHITECT) study was a phase IV, open-label, multicenter, single-arm clinical trial designed to access plaque burden in patients with familial hypercholesterolemia.²⁸ The investigators explored the change in plaque burden in patients (n=104) being treated with alirocumab in addition to statins. They reported that the global coronary plaque burden changed from 34.6% (32.5%–36.8%) at entry to 30.4% (27.4%–33.4%) at follow-up, which represents a –4.6% (–7.7% to –1.9%) statistically significant regression p<0.001). Plaque burden was measured with QAngio CT (Research Edition V2.1.16.1; Medis Specials). Limitations include mean age of patients below Medicare population (mean 53.3), lack of control arm, short term follow-up and uncertainty in how these findings impact clinical outcomes. This study is pertinent in that it demonstrates a role for measuring plaque burden for atherosclerotic CAD and that intervention can impact plaque burden and composition. A study investigated the effect of evolocumab on 136 high risk coronary plaques in patients already taking statins. They found the addition of evolocumab reduce the remodeling index, decrease the percent stenosis, and independently correlated with a change in the minimum CT density. These findings suggest stabilization of the vulnerable coronary plaques with size reduction and may play a role in protecting against CAD progression in patients taking statins.²⁹ Another RCT also reported a positive effect of evolocumab in statin-treated patients on PAV.³⁰ The EVAPORATE study reports icosapent ethyl (IPE), a highly purified eicosapentaenoic acid ethyl ester, is effective in reducing LAP compared to placebo in statin-taking subjects.¹⁰ A nonrandomized study (n=80) reports a potential benefit of low-dose colchicine therapy in reducing LAP volume.³¹

Limitations of CCTA based interpretations is that they are dependent on good to excellent range quality of the CCTA study. In the ISCHEMIA trial, 35% of CCTA’s were suboptimal.³²

Artificial Intelligence Enabled CT based Quantitative Coronary Plaque Analysis (AI-CPA)

Studies have identified higher plaque burden as a risk factor for MACE. Plaque burden is challenging to calculate as it is very time intensive with high variability even among expert readers. Software programs to aid in the calculation of coronary artery stenosis, plaque analysis and fractional flow reserve (FFR) have been developed. Challenges with automation is that the quantitative software programs are variable and only studied in specified patient groups, so generalizability of results is not established. Despite enthusiasm for the technology, further validation is necessary especially in varying size vessels prior to widespread implementation to ensure accuracy. Additionally, variations among gender, ethnicity, age, weight, vessel size, equipment used, definitions within studies and other factors may potentially influence results and warrants further investigation.¹⁹

Clinical Validity

The Rule Out Myocardial Infarction/Ischemia Using Computer Assisted Tomography (ROMICAT) study performed quantitative plaque analysis in 260 patients who presented to the emergency department with suspected ACS. Utilizing a novel semi-automated quantitative measurement, they determined HRP features were an independent predictor of ACS.^33,34 This study was designed to validate the diagnostic value of a score of high-risk coronary features called the ROMICAT score. This earlier tool required manual adjustments of the vessels and plaque boundaries and could take 30 to 60 minutes in patients with multiple plaques.

The CT Evaluation by Artificial Intelligence for Atherosclerosis, Stenosis and Vascular Morphology (CLARIFY) study was a prospective, blinded diagnostic cohort study designed for external validation. It was conducted at multiple sites and included 232 patients who underwent CCTA for acute and stable chest pain.⁴ The CCTA was followed by evaluation with FDA-cleared software service that performs AI-driven coronary artery segmentation and labeling, lumen and vessel wall determination, plaque quantification (Cleerly, Inc.). The mean age of the subjects was 60 ± 12 years with 37% females and most with co-morbidities. After the CCTA was performed, the results were downloaded and AI-aided CCTA analyses were conducted in a blinded manner using the Cleerly software platform. The CCTA was also analyzed by 3 blinded Level 3 readers and consensus of the individual reads was considered the “ground truth” for the study.

AI performance was excellent for accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) as follows: >70% stenosis: 99.7%, 90.9%, 99.8%, 93.3%, 99.9%, respectively; >50% stenosis: 94.8%, 80.0%, 97.0, 80.0%, 97.0%, respectively. The CAD-RADs categorization comparing expert readers to the AI results reported 78% agreement (182/232) and 98.3% (228/232) agreed within 1 category. The most common disagreement was between CAD-RADs 0 and AI CAD-RADS 1 (n= 29 12.5% per patient, n=161 17.4% per vessel). There was >99% category agreement between expert readers and AI-read studies for CAD-RADS 0-3 and CAD-RADS 4-5 using a threshold of >70% stenosis. Bland-Altman plots depict agreement between expert reader and AI, reporting HRP features were found in 49/232 (21.1%) patients using AI and in 31/232 (13.4%) by consensus expert readers which had an 82% agreement. The software analysis time was 9.7 ± 3.2 minutes and time to AI-QCT/AI-CPA analysis and report generation 23.7 ± 6.4 minutes.

The authors conclude that the AI determined reading had highest correlation to the consensus of the expert readers rather than an individual reader suggesting improved accuracy over an individual reader alone. They found the AI approach identified a wide range of atherosclerosis plaque volume and plaque composition in all coronary arteries and their branches, which may offer a benefit over individual readers where this assessment is dependent on the image phase and other factors which limit consistency especially with less experienced readers. Limitations of this study include lack of control group, ground truth is consensus of 3 expert readers without validation to invasive approaches, lack of a guideline basis reference standard for CCTA atherosclerosis quantification, and sample size too small to validate in high-risk population as only 15% of the studied population had anatomically obstructive stenosis.

A challenge with CCTA is the risk of overestimation of CAD and studies have demonstrated that less experienced readers have a higher rate of overestimation as compared to expert readers and risk an increase in unnecessary invasive procedures.³⁵ Investigations to reduce this are underway and include the addition of AI-enabled solutions. In a retrospective report data from the CREDENCE trial,³⁶ 612 study participants with stable chest pain without a history of CAD referred for nonemergent invasive coronary angiogram (ICA) underwent CCTA, ICA and invasive FFR. CCTA images were interpreted on per-lesion and per-segment basis for lumen and vessel volume, diameter stenosis, plaque composition and volume, number of lesions, and the presence of HRP features using semiautomated plaque analysis software (QAngioCT Research Edition, version 3.1.4.1; Medis Medical Imaging) and FFR measurements by Heartflow. The authors conclude that quantification of obstructive and nonobstructive plaque was superior to functional imaging and improved predictions of stress induced alterations in perfusion.

A retrospective international, multicentered study evaluated the data on 921 patients who underwent CCTA. The data was used to develop a novel deep learning convolutional neural network to evaluate segment coronary plaque. This was applied to an independent test set for external validation as compared to IVUS. The authors reported excellent or good agreement, respectively between deep learning and expert reader measurements of total plaque volume (TPV) (intraclass correlation coefficient [ICC] 0.964) and percent diameter stenosis (ICC 0.879; both p<0.0001). The correlation between the deep learning tool and IVUS for plaque volume was excellent (ICC 0.949). The mean per patient analysis time for expert readers was 25.66 minutes (SD 6.79 minutes) as compared to 5.67 seconds (SD 1.87 seconds) for the deep learning tool.²⁰

Myocardial perfusion imaging (MPI) is the most common non-invasive stress imaging modality applied during stress testing in the United States (U.S.). However, MPI has been reported to have a limited performance in ischemia detection.^37,38 A retrospective, multicenter, diagnostic cohort study enrolled 301 subjects and was designed to compare diagnostic performance of MPI and CCTA with AI-QCT using Cleerly, Inc. software.³⁹ The primary endpoint being evaluated was detection of obstructive CAD. The study was a retrospective post hoc analysis from the prospective 23-center Computed Tomographic Evaluation of Atherosclerotic Determinants of Myocardial Ischemia (CREDENCE) trial. The mean study age was 64.4 ± 10.3 with stable symptoms of myocardial ischemia referred for nonemergent invasive angiography.

For patients with no ischemia on MPI (n=102) AI-QCT/AI-CPA identified obstruction ≥50% in 54% and included 20% with severe stenosis (≥ 70%). For the 199 patients with ischemia on MIP, AI-QCT identified nonobstructive stenosis (<50%) in 23%. They reported that AI-QCT had a significantly higher AUC than MPI for predicting ≥ 50% stenosis by CCTA (0.88 vs. 066), ≥ 70% (0.92 vs 0.81) and FFR <0.8- (0.90 vs 0.71) (p<0.001). An AI-QCT result of ≥ 50% stenosis and ischemia on stress MPI had sensitivity of 95% versus 74% and specificity of 63% versus 43% for detecting ≥ 50% stenosis by CCTA measurement. The authors conclude that CCTA with AI-QCT/AI-CPA has a higher diagnostic performance than MPI for detecting obstructive CAD. They suggest that “a scenario of performing coronary CCTA with AI-QCT in all patients and those showing ≥ 70% stenosis undergoing invasive angiography would reduce invasive angiography utilization by 39%; a scenario of performing MPI in all patients and those showing ischemia undergoing coronary CCTA with AI-QCT and those with ≥ 70% stenosis on AI-QCT undergoing invasive angiography would reduce invasive angiography utilization by 49%.” The study is limited by retrospective design which is not a sufficient study design to establish non-inferiority, lack of long-term outcome data, lack of generalizability (mostly men and younger than Medicare population therefore not necessarily applicable to real world practice).

A sub study of 303 subjects from the CREDENCE Trial were evaluated with AI-QCT/AI-CPA with software by Cleerly, Inc.⁴⁰ In this cohort, they compared AI-QCT/AI-CPA to core lab–interpreted coronary CTA, core lab quantitative coronary angiography (QCA), and invasive FFR. They report the prevalence of stenosis ≥50% was observed in 67% (n=202 of 303) of patients and 36% (n=308 of 848) of vessels, while presence of stenosis ≥ 70% was observed in 39% (n=119 of 303) of patients and 19% (n=157 of 848) of vessels. The per-patient sensitivity, specificity, PPV, NPV, and accuracy for ≥50% stenosis was 94%, 68%, 81%, 90%, and 84%, respectively, and for detection of ≥70% stenosis was 94%, 82%, 69%, 97%, and 86%, respectively. Correlations of AI-QCT vs. QCA for % stenosis on a per-territory and per-patient basis were 0.728 and 0.717, respectively (p < 0.0001 for both). Evaluation of predominately calcified (≥50%) vs. predominately non-calcified vessels (<50%) reported significantly lower per vessel specificity in the predominately calcified vessels as compared to predominately non-calcified (86% calcified vs 95.3% noncalcified; p < 0.0001) at 50% threshold and even lower for 70% threshold (92.7% calcified vs 92.9% noncalcified; p < 0.0001). There was discordance of >30% between the AI-QCT determined stenosis as compared to the QCA in 8.1% (74 of 909 vessels) in which 1 or both stenoses was ≥50%. Of the 157 vessels that were ≥70% by AI-QCT, 60.5% (n= 95) had a concordant QCA of ≥70% and 62 were discordant. FFR QCA and AI-QCT had similar accuracy (85% and 86.2%; P ¼ 0.217), respectively for predicting a FFR of <0.8. False positive rate with AI-QCT/AI-CPA was 39.4%, with 62/157 vessels reported to have stenosis ≥70% by AI-QCT/AI-CPA but found to have <70% on invasive CCTA. While invasive FFR was more accurate with 66.1% having FFR of <0.8.

A retrospective review of 406 lesions evaluated with CCTA and coronary flow indexes looked at FFR, resting distal coronary artery pressure (Pd)-to-aortic pressure (Pa) ratio (hereafter, Pd/Pa), coronary flow reserve (CFR), hyperemic flow (1/hyperemic mean transit time [Tmn]), resting flow (1/resting Tmn), and index of microcirculatory resistance (IMR). They report high disease burden (hazard ratio [HR], 4.0; P = 0.004) and adverse plaque (HR, 2.7; P = 0.02) were associated with a higher risk of MACE (n = 27) over median 2.9-year follow-up. This suggest HRP burden and features are associated with cardiovascular events independent of coronary hemodynamic parameters.⁴¹ Another retrospective review called the Assessing Diagnostic Value of Noninvasive FFRCT in Coronary Care (ADVANCE) registry reported on 4,430 patients who had CCTA and AI-QCT/AI-CPA over 1 year. There were 55 MACES. The investigators calculated a TPV of >564 mm³ to be associated with MACE (adjusted HR, 1.515 [95% CI, 1.093-2.099], p=0.0126). Calcified, noncalcified, and low-attenuation PAV above optimal cutoff were associated with adverse outcomes but not cardiac death/MI.⁴²

A population of 303 patients from the CREDENCE trial who underwent CCTA prior to ICA and FFR were evaluated with AI-QCT/AI-CPA.⁴³ They correlated PAV in patients with 50% stenosis on ICA with non-obstructive CAD into single vessel, 2 vessels, and 3 vessels/left main disease groups and further classified by ischemic or non-ischemic. Definition of plaque stage thresholds of 0, 250, 750 mm³ and 0, 5, and 15% PAV resulted in 4 clinically distinct stages in which patients with no, nonobstructive, single VD and multi-vessel disease were optimally distributed. They proposed the following staging criteria based on atherosclerotic plaque burden by QTC related to stenosis severity: Stage 0 (Normal, 0% PAV, 0 mm³ TPV), Stage 1 (Mild, >0–5% PAV or >0–250 mm³ TPV), Stage 2 (Moderate, >5–15% PAV or >250–750 mm³ TPV) and Stage 3 (Severe, >15% PAV or >750 mm³ TPV).

The authors conclude lower specificity, PPV, and accuracy when the AI-based evaluation(s) were compared with QCA vs the previously employed reference standard of consensus of L3 expert readers, primarily because of an increase in false positive diagnoses by the AI-based evaluation.⁴ They report discriminatory power of the AI-based evaluation appears to be high, with a per-patient AUC of 0.88 for ≥50% stenosis and of 0.92 for ≥70% stenosis threshold, and a per-vessel AUC of 0.90 for ≥50% stenosis and of 0.95 for ≥70% stenosis. While these results were higher than that reported in the CLARIFY study, they conclude the composite of this data demonstrates evidence-based evaluation is an important adjunct to CCTA results.

Limitations of the study include retrospective design, focused on the outcome measure of stenosis severity, and did not include evaluation for plaque characteristics which is being investigated in a separate study, does not include evaluation of mild lesions, population not generalizable (mostly male and younger than the Medicare population), and small sample size in patients with more severe disease.

A longitudinal study followed 1,577 patients who underwent coronary CCTA for cardiovascular events over 10.5 years (range 6-11.4). For each subject they calculated Morise Score, CAD severity and coronary TPV. They reported 3.7% (59/1577) had cardiac death or ACS during the study period. They reported that coronary TPV provided additive prognostic value over clinical risk assessed with the Morise Score and CAD severity (rise in C-index from 0.744 to 0.769, P= 0.03). The use of Morise Score and TPV was superior with reclassification of 800 (51%) of patients compared to Morise score alone.⁴⁴ The authors conclude that coronary TPV up to 10 years can be used to reclassify patients into different risk groups compared with clinical risk alone. The study was limited, as no information was available regarding changes in medical therapy after coronary CCTA, although aspirin (ASA) and statin therapy was recommended when signs of CAD were evident.

A sub-analysis of the SCOT-HEART trial evaluated 1,769 participants (43% female) who underwent CCTA and explored sex difference in stenosis, adverse plaque characteristics and quantitative assessments of total, calcified, non-calcified and LAP burden. The authors report that females were more likely to have normal coronary arteries and less likely to have adverse plaque characteristics (p<0.001 for all measurements). While the percentage who went on to have MI was low over 4.7 years, 1.4% in women and 3% in men, the women who experienced MI had similar findings to the men. LAP burden was a strong predictor of MI in both men and women.⁴⁵

An international, multicenter, retrospective study included 9 cohorts of patients undergoing CCTA at 11 sites. A novel deep learning convolutional neural network was trained to segment coronary plaque in 921 patients. An external independent test site validated 175 patients and found good or excellent agreement between the deep-learning network and expert reader measurements of TPV (intraclass correlation coefficient [ICC] 0.964) and percent diameter stenosis (ICC 0.879; both p<0.0001). In addition, 50 patients that were assessed by IVUS were also found to have excellent agreement for deep learning TPV (ICC 0.949) and minimal luminal area (ICC 0.904). The deep learning plaque analysis time was 5.65 seconds (SD 1.87) versus 25.66 minutes (6.79) taken by experts. The subjects were followed for a median of 4.7 years and investigators found a deep learning-based TPV of 238.5 mm³ or higher was associated with an increased risk of MI (HR 5.36, 95% CI 1.70–16.86; p=0.0042) after adjustment for the presence of deep learning-based obstructive stenosis (HR 2.49, 1.07–5.50; p=0.0089) and the ASSIGN clinical risk score (HR 1.01, 0.99–1.04; p=0.35). The authors concluded that the deep learning model has prognostic value for future MI risk.²⁰

CCTA has a Level 1A recommendation for initial evaluation of acute and stable chest pain in patients without known but suspected CAD per the AHA/ACC Guidelines. However despite CCTA’s high sensitivity, a moderate specificity suggest a risk of false positives and overestimation of stenosis.⁴⁶ A sub-analysis from the CREDENCE trial aims to compare AI-QCT, CCTA and FFR-CT for discriminating coronary ischemia at the patient and vessel levels.⁴⁷ In this study AI-QCT/AI-CPA was calculated using Cleerly, Inc. software and in the CREDENCE trial FFR-CT was calculated by Heartflow, Inc. software in a blinded fashion. Area under receiver operative characteristics curve (AUC) with 95% confidence intervals was used to compare the modalities.

In comparing the 3 modalities for discriminating ischemia at a ≥505 stenosis threshold they report the following: AI-QCT with accuracy, sensitivity, specificity, PPV, and NPV of 82%, 95%, 66%, 76%, and 92%, respectively, and FFR-CT’s of 69%, 59%, 82%, 79%, and 63%, respectively. Comparatively, CCTA achieved intermediate range outcomes with performance measures, 73%, 75%, 71%, 75% and 71%, respectively. They conclude accuracy was greatest for AI-QCT/AI-CPA. FFR-CT demonstrated high specificity and PPV, but weaker in other measures and comparable to the results reported with CCTA.

In comparing for discriminating ischemia at the vessel level they report accuracy, sensitivity, specificity, PPV and NPV for AI-QCT at the vessel level were 84%, 89%, 83%, 65%, and 95%, FFR-CT’s of 7%, 60%, 83%, 56% and 85%, and CCTA 79%, 65%, 84%, 59% and 87%, respectively. The area under the receiver operative characteristics curve (AUC) for discriminating patient-level ischemia by AI-QCT, CCTA and FFR-CT was 0.90, 0.77 and 0.73, respectively (p<0.001) and 0.93, 0.83 and 0.79, respectively, at the vessel level (p<0.001). The author concludes AI-QCT/AI-CPA achieved equal specificity to FFR-CT for discriminating ischemia at the vessel level while maintaining superior sensitivity and overall accuracy compared to both CCTA and FFR-CT.

The study is limited by all the limitations that impacted the CREDENCE trial from which the data was obtained and the retrospective study design which is not sufficient to determine superiority or non-inferiority between these different modalities.

A retrospective report analyzed 79 patients with end stage renal disease (ESRD) referred for CCTA and underwent AI-QCT/AI-CPA with Cleerly, Inc. software.⁴⁸ This study provided disease distribution and plaque analysis specific to the ESRD population. They reported higher low-density non-calcified-plaque, non-calcified-plaque, calcified-plaque, length, and TPV in patients with >50% stenosis and obstructive lesions. They found more calcified-plaque and PAV in patients >65 years old.

A staging criterion for plaque volume is proposed by Min et al.⁴³ This is an important step for measuring clinical utility and outcome related to plaque in the future. The authors report that while atherosclerotic plaque characterization by CCTA enables quantification of CAD and has been demonstrated as a strong predictor of future risks of MACE, a clinically useful threshold to understand patient disease burden to guide diagnosis and management was lacking. A population of 303 patients from the CREDENCE trial³⁶ who had CCTA and FFR prior to ICA data was analyzed with AI-QCT/AI-CPA with Cleerly, Inc. software. They correlated PAV in patients with 50% stenosis on ICA with non-obstructive CAD into single vessel, 2 vessels, and 3 vessels/left main disease groups and further classified by ischemic or non-ischemic. Definition of plaque stage thresholds of 0, 250, 750 mm³ and 0, 5, and 15% PAV resulted in 4 clinically distinct stages in which patients with no, nonobstructive, single VD and multi-vessel disease were optimally distributed. They proposed the following staging criteria based on atherosclerotic plaque burden by QCT related to stenosis severity: Stage 0 (Normal, 0% PAV, 0 mm³ TPV), Stage 1 (Mild, >0–5% PAV or >0–250 mm³ TPV), Stage 2 (Moderate, >5–15% PAV or >250–750 mm³ TPV) and Stage 3 (Severe, >15% PAV or >750 mm³ TPV).

Another post-hoc analysis from the CREDENCE study³⁶ also investigated the relationship been coronary stenosis and plaque characteristics and age using AI-QCT/AI-CPA with Cleerly, Inc. software. The cohort of patients >65 (n=154) had more plaque volume and calcified plaque than those <65 (n=139).⁴⁹ On a per lesion level they reported those >65 had more calcified plaque in both obstructive and non-obstructive lesions while the <65 cohort had more plaque volume, non-calcified plaques and low-density non-calcified plaques and lesion length in obstructive lesions. The authors conclude this may aid in approaches to management.

A large observational, retrospective, consecutive, international, multicenter cohort study included 11,808 patients who underwent clinically indicated CCTA. Analysis for TPV and composition was performed with AI-CPA with software from Heartflow. Mean age was 62.7 ± 12.2 years with 45.9% women. The authors report median TPV was 223 mm³ (IQR: 29-614 mm³) and was significantly higher in male participants (360 mm³; IQR: 78-805 mm³) compared with female participants (108 mm³; IQR: 10-388 mm³) (P < 0.0001). These results do not address clinical outcomes of the cohort but do provide nomograms. The authors call for future investigation to validate the use of the nomograms to confirm the potential role of AI-CPA as a risk tool for guiding clinical decisions.⁵⁰

Omori et al.⁵¹ conducted the multicenter study to investigate the performance of AI-QCT in the detection of low-density noncalcified plaque (LD-NCP) using near-infrared spectroscopy-intravascular ultrasound (NIRS-IVUS). Patients were enrolled if they were indicated to undergo CCTA and IVUS, NIRS-IVUS, or optical coherence tomography. A total of 133 atherosclerotic plaques (n=47) were evaluated. The area under the curve of LD-NCP<30HU was 0.97 (95% confidence interval [CI]: 0.93–1.00] with an optimal volume threshold of 2.3 mm³. Using <30 HU and 2.3 mm³, accuracy, sensitivity, and specificity were 94% (95% CI: 88–96%), 93% (95% CI: 76–98%), and 94% (95% CI: 88–98%), respectively. As compared to 42%, 100%, and 27% using <30 HU and >0 mm³ volume of LD-NCP (p < 0.001). A strong correlation was reported between AI-QCT and IVUS measurements which included vessel area (r² = 0.87), lumen area (r² = 0.87), plaque burden (r² = 0.78) and lesion length (r² = 0.88), respectively. Authors concluded AI-QCT served as an adequate tool to detect significant LD-NCP with maxLCBI_4mm ≥ 400 as the reference standard.

Hakim et al.⁵² conducted a study to compare the accuracy of endothelial shear stress (ESS) computation of local ESS metrics by CCTA vs IVUS imaging from a registry of 59 patients who underwent both IVUS and CCTA for suspected CAD. IVUS and CCTA measurement of plaque characteristics correlated when measured and included vessel, lumen, plaque area and minimal luminal area per artery, resulting in 12.7 + 4.3 vs 10.7 + 4.5 mm², r = 0.63; 6.8 +2.7 vs 5.6 + 2.7 mm², r = 0.43; 5.9 + 2.9 vs 5.1 +3.2 mm², r = 0.52; 4.5 + 1.3 vs 4.1 + 1.5 mm², r = 0.67 respectively. ESS metrics of local minimal, maximal, and average ESS were moderately correlated when measured with IVUS and CCTA resulting in 2.0 +1.4 vs 2.5 + 2.6 Pa, r = 0.28; 3.3 + 1.6 vs 4.2 + 3.6 Pa, r = 0.42; 2.6 + 1.5 vs 3.3 + 3.0 Pa, r = 0.35, respectively. Authors conclude CCTA and IVUS produce similar outcomes when evaluating ESS.

A post-hoc analysis for the CT- PRECISION registry trial evaluated 377 bifurcation lesions from 340 patients. They determined that side branch occlusion occurred in 28/377 bifurcation lesions (7.5%). They found that non calcified plaque burden and low-density plaque burden are associated with improvements in predicting side clearance occlusion. This study contributes useful information in correlating CCTA findings with invasive catheterization, but how this improves diagnosis over invasive catheterization and if this results in a change to clinical management and patient outcomes is not established.⁵³

Narula et al.⁵⁴ reports on the REVEALPLAQUE study which included 237 patients with known CAD comparing an AI-enabled plaque analysis (Heartflow) to quantify and characterize total plaque, vessel, lumen, calcified plaque, non-calcified plaque and LAP volumes derived from CCTA and compared to IVUS measurements. AI-enabled CCTA quantification and characterization of atherosclerosis demonstrated strong agreement with IVUS reference standard measurements. The correlation was significant for TPV, calcified and noncalcified plaque volumes. The concordance between AI-QCPA and IVUS segmentation was excellent, with correlation coefficients and linear regression slopes nearing 1.0 and Y intercepts nearing 0.

Clinical Utility

Min et al.⁵⁵ reported on outcomes from the CONFIRM trial which was comprised of 23,854 consecutively enrolled patients who underwent CCTA for suspected CAD at 12 centers. Thirty-four percent of the registry was asymptomatic and had low pre-test probability of CAD at enrollment. CAD by CCTA was defined as none (0% stenosis), mild (1% to 49% stenosis), moderate (50% to 69% stenosis), or severe (>70% stenosis). Severity of CAD was determined by considering individual patient, per-vessel, and per-segment. The study cohort had a high prevalence of cardiovascular risk factors and symptoms who were ages 57 + 13 years and were 54% male, with most presenting with intermediate or high pretest likelihood of obstructive CAD. A total of 404 deaths were reported at a mean survival examined at 2.3 + 1.1 years (median 2.1 years; interquartile range: 1.5 to 3.1 years). Authors concluded that higher rates of mortality are associated in individuals without known CAD, nonobstructive and obstructive CAD by CCTA and vary further depending on age and sex. The risk of death in individuals without CAD by CCTA was very low. A post hoc analysis using a test sample of 17,793 patients and validation sample of 2,506 patients with suspected CAD from the CONFIRM registry were followed a median of 2.3 years. There were 347 deaths during this time. The investigators reported a correlation between mortality and the number of proximal segments with mixed or calcified plaques (C-index 0.64, p<0.0001) and the number of proximal segments with a stenosis > 50% (C-index 0.56, p=0.002) equivalent to a 5-year increase in age or risk of smoking.⁵⁶ A 6-year follow-up in an asymptomatic population who had CCTA, plaque analysis and CACS reported improved prognostication of 6-year all-cause mortality beyond a set of conventional risk factors alone but no further incremental value was offered by CCTA findings were added to a model incorporating risk factors + CACS.⁵⁷

Chang et al.⁵⁸ conducted the ICONIC study which was a nested case-control study within the CONFIRM study consisting of a cohort of 25,251 patients undergoing CCTA to identify atherosclerotic features associated with precursors of ACS. Duration of follow up was 3.4 + 2.1 years. ACS patients were propensity matched 1:1 with non-event patients with no prior CAD for risk factors and CCTA- obstructive (>50%) CAD. A total of 234 patients with ACS and control pairs were included. At baseline, >65% of patients with ACS had nonobstructive CAD while 52% had HRP, meaning the group with non-obstructive CAD at baseline represented the group that experienced the greatest number of MACE. Characteristics such as percent diameter stenosis (%DS), cross-sectional plaque burden, fibrofatty and necrotic core volume, and HRP resulted in an increased hazard ration of ACS (1.010 per %DS, 95% confidence interval [CI]: 1.005 to 1.015). Plaque analysis identified 129 “culprit lesion precursors”. Three quarters of these had <50% stenosis and 31% showing HRP. Authors concluded that evaluating plaque composition assists with identifying high risk cases. A post hoc analysis reported that plaque volumes > 1000 HU (called 1K plaque) are associated with a lower risk for ACS.⁵⁹

A 2018 systematic review included 13 studies that reported on CCTA derived CPA and MACE and included 552 MACES in 13,977 patients for a rate of 3.9%. In terms of plaque morphology, the strongest association was observed for noncalcified plaque (HR, 1.45; 95% confidence interval [CI], 1.24–1.70; P<0.001), with weaker associations found for partially calcified (HR, 1.37; 95% CI, 1.18–1.60; P<0.001) and calcified plaques (HR, 1.23; 95% CI, 1.16–1.30; P<0.001).⁷ HRP findings include napkin-ring sign, LAP , positive remodeling, and spotty calcification. All HRP findings were strongly associated with MACE and ≥ 2 HRP findings had the highest risk of MACE (HR, 9.17; 95% CI, 4.10–20.50; P<0.001). The authors conclude HRP is an independent predictor of MACE but acknowledges that clinical impact of these findings was not established.

A post hoc analysis of the SCOT-HEART trial assessed whether noncalcified LAP burden on CCTA might be a predictor of the future risk of MI. Authors investigated the association between the future risk MACE and LAP burden (% plaque to vessel volume), cardiovascular risk score, CACS or obstructive coronary artery stenoses. Quantitative assessment of atherosclerotic plaque subtypes was performed using standardized semiautomatic software (Autoplaque, Version 2.5, Cedars-Sinai Medical Center). They found LAP burden, CACS, obstructive CAD were all higher in the individual who experienced MI. In patients (n=1769) followed for median of 4.7 years with LAP burden greater than 4% were nearly 5 times more likely to have subsequent MI than those below this threshold (HR, 4.65; 95% CI, 2.06–10.5; P<0.001).⁸ They conclude that LAP can provide incremental prediction of MI to standard assessment cardiovascular risk scores, computed tomography calcium scoring or luminal stenosis severity. The strength of this study was it was multicentered, RCT with long-term follow-up and without industry funding reducing potential bias. Limitations include use of single technique to analyze plaque volume, potential influence on decision making for participants in the trial from the original CCTA results, however this may have reduced incidence of MACE so would not diminish the value of the findings. Authors state further investigation is needed to see if this can be used for clinical decision making leading to improved patient outcomes. This study is pertinent as it identifies the potential role of plaque analysis as a predictor of future MACE.

A sub analysis of the Computed Tomographic Angiography for Selective Cardiac Catheterization trial (CONSERVE)⁴⁶ assessed 747 patients who underwent CCTA prior to non-emergent ICA with AI-QCT/AI-CPA software by Cleerly, Inc.⁴⁶ The AI-QCT reported on stenosis determination, coronary vascular measurements and quantification and characterization of atherosclerotic plaque. CCTA interpretation and AI-QCT/AI-CPA guided findings were evaluated for MACE at 1-year follow-up.

AI-QCT/AI-CPA reported no CAD in 9% compared to 34% with CCTA alone. For intermediate stenosis (50-69%) AI-QCT/AI-CPA identified 8% of patients (60/747) as compared to Level II/III readers in which 27% (205/747) were reported to have ≥50% stenosis, 16% (117/747) with ≥70% stenosis, and 12% (88/787) with intermediate stenosis resulting in referral for invasive intervention. The rates of safe ICA deferral from AI-QCT were significantly higher than those based upon Level II/III reader interpretation of CCTA. They reported that no patient experienced MACE for 1 year following who had been quantified as having non-severe stenosis by AI-QCT. When categorizing stenosis severity as 0%, 1%−24%, 26%−49%, 50%−69%, >70%, stenosis severity to predict MACE events was similar between AI-QCT (AUC of 0.61; 95% CI 0.52−0.70) and Level II/III CCTA interpretation (AUC of 0.63; 95% CI 0.53−0.73; p = 0.64). The authors conclude that for patients meeting criteria for non-emergent ICA based on ACC/ AHA Guidelines, that adoption of an AI-QCT approach could reduce unnecessary ICA by 87%−95% based upon stenosis severity thresholds.

For CPA there was a linear correlation between the quantification of atherosclerotic plaque and absolute plaque volume and MACE. Plaque volumes were categorized using HU ranges with noncalcified plaque (NCP) defined as HU between −30 and +350; low density-NCP (LD-NCP) defined as plaques < 30 HU; and calcified plaque (CP) defined as >350 HU.⁴⁰

As the HR for each plaque volume increased so did the observed rate of MACE. For patients with plaque volume between 0 and 300 mm³ (n = 509), 301−750 mm³ (n = 174) and ≥750 mm³ (n =64), there was an observed MACE rate of 2.6%, 7%, and 9.4%, respectively, (p = 0.001).

The authors conclude that AI-QCT/AI-CPA referral management can avoid unnecessary ICA in patients with stable CAD with <50% or <70% stenosis. Limitations of the study include indirectness as mean age was 60 ± 12.2 which is younger than the Medicare population, the population was 86% Asian and not reflective of US population, post hoc analysis, risk of bias due to lack of blinding for CCTA core laboratory and short-term follow-up.

A long-term follow-up included 536 patients enrolled in studies evaluating the long-term prognostic value of high-risk plaques.^62-64 All patients underwent CCTA for suspected stable CAD between 2008-2014. AI-QCT/AI-CPA analysis was conducted using Cleerly, Inc. platform to analyze the CCTA images. This study population was in Amsterdam where national registry database is maintained, was utilized to determine prognostic data from follow-up visits between 2021 and 2022 providing follow-up data for 508 patients. They were evaluated for MACE and outcomes based on plaque staging.⁴³ Coronary plaque volume was normalized to the total per–patient vessel volume to account for variation in coronary artery volume by calculating as plaque volume /vessel volume x 100%. These normalized volumes were reported as PAV, percentage noncalcified plaque volume (NCPV), and percentage calcified plaque volume (CPV).

AI-QCT analysis showed 343 patients (64%) had nonobstructive CAD (<50% stenosis), 88 (16%) had moderate obstructive CAD (≥50%-69% stenosis) and 105 (20%) had a severe obstructive stenosis of ≥70%. Plaque volume was stratified with 15 patients without plaque (CAD stage 0), 257 patients with a PAV between 0% and 5% (CAD stage 1), 149 patients with a PAV between 5% and 15% (CAD stage 2), and 115 patients with ≥15% PAV (CAD stage 3). Patients were also stratified based on their NCPV: 15 patients without NCPV (NCPV stage 0), 212 patients with a NCPV volume between 0% and 2.5%, 160 patients with NCPV between 2.5% and 7.5%, and 149 patients with >7.5% NCPV. Higher PAV stages correlated with worse survival for both MACE and secondary outcomes (p<0.001). Higher NCPV stages also showed worse survival. Those with no plaque volume did not experience any events during follow-up. The addition of AI-QCT to a risk model with clinical risk factors and CACS improved risk discrimination for MACE at 2, 5, and 10 years of follow-up (10-year AUC: 0.82 [95% CI: 0.780.87] vs 0.73 [95% CI: 0.68-0.79]; p< 0.001; NRI: 0.21 [95% CI: 0.09-0.38]). The authors reported AI-QCT/AI-CPA outperformed manual stenosis grading and segment involvement score in the prediction of MACE at 10 years (AUC:0.82 [95% CI: 0.78-0.87] vs 0.78 [95% CI: 0.73-0.83]; P ¼ 0.040; NRI: 0.04 [95% CI: 0.05 to 0.27]). Additionally, they were able to calculate predictive risk for 10-year survival across subgroups based on plaque burden by stenosis grade.

The authors conclude that quantitative coronary plaque staging with AI-QCT may improve risk stratification for long term MACE. Previous studies (ICONIC) demonstrated >75% of culprit lesions prior to MI were nonobstructive. This study found plaque volumes were a greater determinant of 10-year ASCVD risk in patients with low CAD-RADS stenosis score compared to those with high scores suggesting that high-plaque burden, even in absence of obstructive lesions, are independent prognostic risk factors. This may have implications for treatment as these higher risk individuals may benefit from an intensified therapy. With the emergence of additional therapeutic options, the technology may be able to improve identification to apply treatments to those who need them the most and avoid in those who may not benefit.

This study benefits from a 10-year follow-up offering valuable long term outcome data. Limitations include indirectness as the population was younger than Medicare population (mean age 58 ± 9.2), population in non-US population, risk of bias from missing outcome data inherent to retrospective study design, imprecision related to relatively small sample size, and concerns related to study quality and follow-up of the cohort over time.

A large observational, retrospective, consecutive, international, multicenter cohort study included 11,808 patients who underwent clinically indicated CCTA. Analysis for TPV and composition was performed with AI-CPA using Heartflow, Inc. platform. Mean age was 62.7 ± 12.2 years with 45.9% women. The authors report median TPV was 223 mm³ (IQR: 29-614 mm³) and was significantly higher in male participants (360 mm³; IQR: 78-805 mm³) compared with female participants (108 mm³; IQR: 10-388 mm³) (P < 0.0001).⁵⁰

In “CAD-RADS™ 2.0 - 2022 Coronary Artery Disease-Reporting and Data System”² management considerations are made based on the CAD-RADS category 0-5. Management is impacted by the calculated plaque burden which is grades P1-P4 representing mild, moderate, severe, and extensive, respectively. For CAD-RADS 1 & 2 management varies based on the plaque burden recommending the following “P1: consider risk factor modification and preventive pharmacotherapy, P2: risk factor modification and preventive pharmacotherapy, and P3 or P4: aggressive risk factor modification and preventive pharmacotherapy”. For CAD-RADS 3-5 management recommendations are not changed based on plaque volume specifically with P1, P2, P3 or P4 all recommended to receive “aggressive risk factor modification and preventive pharmacotherapy.”

The DECODE study explores the role of AI-QCT/AI-CPA in clinical decision making. A cohort of 100 patients with suspected CAD who underwent clinically indicated CCTA were included. Three Level 3 CCTA readers reviewed the CCTA reports, clinical and laboratory data, and management plans using a recently published management hierarchy. The cardiologists were then provided AI-QCT/AI-CPA data and had the opportunity to alter management plans with this additional information. Management Plan Reclassification Rate (RR) following AI-QCT/AI-CPA review was 66% (66/100) (95% CI 56.72%-75.28%). The authors report that when AI-QCT/AI-CPA information was added most management plans were intensified with reclassification rates ranging from 47% in patients with CAC=0 to 96% in patients with CAC>400, and 89.5% in patients with coronary stenosis >50%. Broken down by CAD-RADs score the RR was 89.5% for CAD-RADS ≥ 3 (>50% stenosis) and 51.6% with <50% stenosis.⁴⁶ Limitations of the study include small sample size, potential variability among readers, lack of control group or randomization, potential risk of bias, and lack of knowledge of the changes in management impact on patient outcomes.

The CERTAIN study compares the clinical utility of AI-QCT to CCTA. This is a multi-centered crossover study in which 750 patients had CCTA performed and providers made diagnosis and management plans based on test results. Next, AI-QCT/AI-CPA results using Cleerly Inc. platform were provided to the providers on the same patient and they were asked if there would be any changes in management based on this information. The authors report that AI-QCT improved physician's confidence 2.5-fold at every step in the care pathway and was associated with a change in management for 57.1% (428/750) of patients (p<0.001). This resulted in a reduction in noninvasive and invasive downstream testing by 37% and the initiation of statin or ASA treatment by 28.1% and 23% respectively (p<0.001). Limitations include short term follow up with lack of data on if patients received additional testing in the future and additional management guidelines that were changed after the initial decisions were made and the study was not designed to compare this management approach to standard of care.⁶⁶

The PREVENT study is a multicentered RCT with non-flow limiting (FFR >0.80) vulnerable coronary plaques identified by intracoronary imaging or randomized to percutaneous coronary intervention plus optical medical therapy (n=803) or optical medical therapy alone (n=803). At 2 years, 3 patients experienced MACE in the percutaneous coronary intervention group (0.4%) and in 27 patients in the medical therapy group (3.4%) (absolute difference –3·0 percentage points [95% CI –4.4 to –1.8]; p=0.0003). The authors concluded that in patients with non-flow limiting high-risk coronary plaques that preventive percutaneous coronary intervention reduced major adverse cardiac events better than medical therapy alone.⁶⁷

Societal Guidance

Society of Cardiovascular Computed Tomography (SCCT), the American College of Cardiology (ACC), the American College of Radiology (ACR) and the North American Society of Cardiovascular Imaging (NASCI)²

Practice Guidelines in the form of a consensus document was published by SCCT, ACC, ACR and NASCI. This document reports on the standardized reporting system for CCTA called CAD-RADS. Additionally, they proposed management considerations including further cardiac investigation and escalating clinical management, for the different levels of disease based on the CAD-RADS classification. They acknowledge emerging technologies for performing quantitative and reproducible assessments for total plaque burden and type but are not widely available and lack consensus thresholds for patient management.

The 2021 Chest Pain Guidelines from the American Heart Association (AHA) and AAC¹ state:

• For intermediate-high risk patients with stable chest pain and no known CAD, CCTA is effective for diagnosis for CAD, for risk stratification, and for guiding treatment decisions (1A recommendation).
• For intermediate-risk patients with acute chest pain and no known CAD eligible for diagnostic testing after a negative or inclusive evaluation for ACS, CCTA is useful for exclusion of atherosclerotic plaque and obstructive CAD (1A recommendation).
• For symptomatic patients with known nonobstructive CAD who have stable chest pain, CCTA is reasonable for determining atherosclerotic plaque burden and progression to obstructive CAD and guiding therapeutic decision making (2A recommendation, LOE: B-NR).
• For intermediate-risk patients with acute chest pain and known nonobstructive CAD, CCTA can be useful to determine progression of atherosclerotic plaque and obstructive CAD (2A recommendation, LOE:B-NR)

The American College of Cardiology⁶⁸

The ACC Innovations in Prevention Working Group published Atherosclerosis Treatment Algorithms with the aim to personalize medical intervention(s) based on findings from CCTA and risk factors. The “Atherosclerosis Treatment Algorithms” includes recommended treatments based on plaque staging.

Atherosclerosis stages were categorized as:^36,43

• Stage 0 = 0 mm³ (0% PAV).
• Stage 1 = >0-250 mm³ (>0-5% PAV).
• Stage 2 = >250-750 mm³ (>5%-15% PAV).
• Stage 3 = >750 mm³ (>15% PAV).

Multiple interventions are reviewed with Dietary Approaches to Stop Hypertension (DASH) diet, physical activity and Icosapent ethyl supported by RCT, statins supported by observational cohort study, colchicine supposed by a prospective study and Evolocumab by a retrospective study. The treatment algorithms also provide recommendations for treatment and serial CCTA to monitor disease progression based on stage of disease.

These guidelines are limited as they have not been validated as to impact on patient outcomes, and some of the recommended treatments are supported by low quality literature. The ACC Foundation has an investment in the software utilized in the algorithms.

American Society of Preventive Cardiology (ASPC)⁵

A clinical practice statement was developed by an expert panel on CCTA and emerging applications of CCTA. Within this report the authors discuss the role of CCTA in calculation of plaque volume and that when plaque volumes decrease, as seen on serial CCTA with CPA, improved clinical outcomes are demonstrated. They conclude that ability to distinguish plaque composition and identify vulnerable plaques are clinically applicable for the evaluation and management of CAD. AI-QCT/AI-CPA has been shown to improve/personalize preventive therapies, reduce time, and improve accuracy for CPA calculation. They state there is a high correlation demonstrated with 98% agreement with expert readers for CAD-RADS category on per patient and 99% of per vessel basis with >95% sensitivity for detection of obstructive stenosis.⁴ They state confidence for future clinical trials to utilize AI-QCT/AI-CPA for plaque quantification and evaluation of response to therapies. They also state: “AI solutions should be validated in multicenter clinical trials against appropriate ground truth standards in order to ensure accuracy, precision, and generalizability” and “the effect of individualized preventive therapies that is guided by the identification of AI-identified CCTA adverse plaque characteristics require study in future prospective randomized trials”.

Society of Cardiovascular Computer Tomography (SCCT)⁶

A 2024 expert consensus document which aims to provide a framework for standardization and nomenclature, methodology and reporting of quantitative CCTA analyses. The report offers multiple recommendations to aid in accuracy and consistency in future study designs.

Asian Society of Cardiovascular Imaging⁶⁹

This expert consensus guideline seeks to resolve ambiguities in CCTA interpretations by leveraging the expertise of specialists for the evaluation of coronary stenosis and plaque in CT angiography.

CAD Frontiers Atherosclerosis CT Imaging Outcome Consortium Accelerating Atherosclerosis Drug Development group (ACTION A2D2)⁷⁰

This group works towards developing consensus on CCTA plaque measurements that relate to the causal pathway for MACE. They are working towards standardization of plaque endpoint to improve the quality of studies. This document offers a summary of clinical evidence and application to trial design to aid investigators in future research especially those exploring potential drugs.

Health Care Disparities

The demographic of CAD varies between genders, race, and age. A sub-analysis of the CONFIRM study found a higher prevalence of obstructive CAD in men (42% vs. 26%, p<0.001). A strong association between increase MACE risk and non-obstructive CAD was reported and there was not a difference based on gender for MACE risk.⁷¹  Women have been found to have a higher rate of non-obstructive CAD which is highly prognostic (2-fold increased risk) in women of future MACE. Women were also found to carry an increased risk associated with non-obstructive left main disease, presence of HRP compared to men.⁵ Studies are beginning to show a difference in plaque composition between men and women which may impact disease progression. The SCCT has developed an expert consensus statement about the role of CCTA in women and future investigations are needed to further delineate these differences. Additionally, most studies did not include a population that represents the Medicare population on age and co-morbidities. Differences in the diagnostic and management related to ethnicity needs further investigation as these demographics have not been well represented in the studies.  A special report from AHA on risk assessment tools to guide decision making in primary prevention of atherosclerotic cardiovascular disease advocates that future studies explore risk factors and include Hispanics, East Asians, South Asians, patients >75 years, underrepresented minority groups and those with social deprivations to allow more targeted risk assessments including diverse racial/ethnic groups.⁷²

Several studies investigate specific at-risk population. A post-hoc analysis from the SCOT-HEART study reports on patients with diabetes mellitus demonstrating a higher rate of all-cause mortality and major adverse cardiac events at 5 years compared to non-diabetic subjects. This risk was not present in those with diabetes mellitus without CAD according to the CCTA results.⁷³ Another study found elevated fibro-fatty plaque volume was a predictor of events over age, gender and traditional risk factors in diabetic patients at high risk for CV disease.⁷⁴ A paper seeks to understand the role of statins to reduce non-calcified plaque volume and high-risk coronary plaques in HIV-infected patients who are at elevated risk of MACE. They reported a positive impact of statins in this population.⁷⁵ A double blinded placebo-controlled RCT with 774 subjects found a reduction in non-calcified plaque volume and progression over a 24-month period in HIV positive patients using pitavastatin.^76-77

It is well established that CCTA is an important tool for cardiac evaluation and can provide clinically relevant information for the care of patients with existing or suspected CAD. CCTA has been found beneficial to clarify the diagnosis, guide intervention and potentially reduce the risk of future MI.¹⁶ Investigators have sought to further expand the information gathered from CCTA to offer non-invasive means to evaluate the coronary arteries with the potential to reduce the need for invasive angiography or provide further diagnostic information that may impact management. This is pertinent as the many patients with lower-risk findings may not receive as aggressive interventions and go on to experience MACE demonstrating a need for further technologies that can aid in evaluating additional risk factors.

Emerging evidence has established that the quantity and composition of plaque can play a role in identifying high risk cases. Multiple studies have identified that CCTA plaque burden is associated with cardiovascular outcomes including the ICONIC, CONFIRM and PARADIGM studies.^25,55,58 Plaque composition has emerging evidence to provide information that can aid in identification of this at-risk group and offer additional management options to reduce risk. The addition of plaque quantification was demonstrated to improve the diagnosis and prognostic risk stratification, beyond the capability of CCTA alone.⁶² While this information can be calculated by expert readers it is very time consuming and variability among readers has limited its utility. The ability to calculate this accurately with enhanced technologies offers the potential to incorporate into the decision making more practically. Studies have demonstrated that statins are associated with slower progression of overall coronary atherosclerosis volume, with increased plaque calcification and reduction of high-risk plaque features demonstrating there are interventions that may improve outcomes based on these findings.²⁵ Symptomatic patients with intermediate risk and no known CAD and negative evaluation have been identified as a group at potential elevated risk. The Chest Pain Guidelines support the role of plaque assessment in this group to further delineate risk and offer escalating intervention if indicated. This is supported by a strong recommendation and high-quality evidence (1A). Cardiac ischemia is decreased blood flow and oxygen to the heart muscle and the current societal Chest Pain Guidelines recommend non-invasive testing for cardiac ischemia with a functional stress test (exercise or pharmacologic echocardiography or MPI) Class I recommendation. The use of AI-QCT/AI-CPA does not extend to evaluation for cardiac ischemia defined within the limited coverage criteria of this LCD as there is currently insufficient evidence to support this role.

Determination of atherosclerotic plaque burden for intermediate risk patients with acute and stable chest pain with no known CAD (1A recommendation) is included in the recommendations within the AHA/ACC/ASE/CHEST Guidelines¹. HRP features are strong predictors of adverse events, even in patients with nonobstructive CAD, as demonstrated by the PARADIGM study and PROMISE trial.^23,78 The PROMISE trial demonstrates that in patients with nonobstructive CAD, CCTA provides better prognostic information than functional testing in patients with stable chest pain and low burden of obstructive CAD, MI and events.²² Multiple studies have demonstrated that LAP burden is a predictor of future MI and culprit lesions are often nonobstructive.^8,58,79,80 A post-hoc analysis of the SCOT-HEART trial demonstrates a correlation of LAP burden as a predictive tool for future MI and more accurate than exercise stress test.¹⁸ The DECODE and CERTAIN studies both show that management changes occur as a result of this additional information that may improve patient outcomes.^65,66 Literature has identified that escalated interventions are correlated with improved patient outcomes, such as statins being shown to reduce HRP feature.²⁵ Based on the potential improved diagnostic capabilities and role for change in management based on the study’s results, the use of AI-QCT/AI-CPA is reasonable for stable chest pain with no known CAD. Additionally, since literature shows 40% of MACE occur in patients with <25% diameter stenosis, suggesting minimal plaque burden may predispose MACE the policy is expanded to cover CADS-RAD1 given that understanding plaque burden and characterizations may identify higher risk patients and provide the opportunity for more aggressive intervention as reported in the DECODE and CERTAIN studies.^65,66

CAD-RADS 4A represents a population with stenosis of 1 or 2 vessels between 70-99%, and CAD-RADS 4B represents > 50% of the left main artery or 3 vessels obstructive disease (≥70%) and likely represents ACS. Management for this population includes consideration of hospital admission with ICA except in 4A where ICA or functional assessment is recommended.² Functional assessment in the AHA/ACC/ASE/CHEST Guidelines¹ is defined as CT-FFR, myocardial CT perfusion study, stress testing (exercise tolerance test, stress ECG, single photon emission CT, positron emission tomography, Cardiac MRI) or invasive FFR and AI-QCT/AI-CPA is not included under the options.¹ Additionally, plaque attenuation may be affected by the presence of extensive and dense coronary calcification impacting plaque analysis requiring further validation in this setting. This Guideline recommends a potential benefit of plaque analysis in symptomatic patients with known nonobstructive CAD with stable chest pain with a Grade 2A recommendation (probably recommended). This is supported by B-NR evidence which means at least 1 non-randomized, observational or registry study. Therefore, it is not established how this test will change management or improve outcomes for CAD-RADS 4 or higher or in symptomatic stable known obstructive CAD.¹⁹ While the DECODE trial⁶⁵ suggest management changes in both groups, escalation of therapy is expected in this group regardless of AI-QCT/AI-CPA results; so how this information directly impacts patient outcomes is not clear. The population within this study with CAD-RADS 4 was too small to determine outcomes. Future research may provide evidence and understanding of potential roles of AI-QCT/AI-CPA in these populations.

Therefore, the policy is limited coverage to ensure access to these sub-sets where benefit has been established while further investigation defines the role of this emerging technology beyond this population. The role of this technology for surveillance or monitoring response to treatment has not been established in the literature, and therefore is limited to initial diagnosis.

Related LCDs

L33423 Cardiac Computed Tomography and Angiography

L38278 Non-Invasive Fractional Flow Reserve (FFR) for Stable Ischemic Heart Disease

1. Gulati M, Levy PD, Mukherjee D, et al. 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the evaluation and diagnosis of chest pain: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. J Cardiovasc Comput Tomogr. 2021;16(1):54-122
2. Cury RC, Leipsic J, Abbara S, et al. CAD-RADS™ 2.0–2022 coronary artery disease-reporting and data system: an expert consensus document of the society of cardiovascular computed tomography (SCCT), the American College of Cardiology (ACC), the American College of Radiology (ACR), and the North America society of cardiovascular imaging (NASCI). J Am Coll Cardiol Img. 2022;15(11):1974-2001.
3. Correction to: 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the evaluation and diagnosis of chest pain: a report of the American College of Cardiology/American Heart Association joint committee on clinical practice guidelines. 2021;144(22):e455.
4. Choi AD, Marques H, Kumar V, et al. CT evaluation by artificial intelligence for atherosclerosis, stenosis and vascular morphology (CLARIFY): a multi-center, international study. J Cardiovasc Comput Tomogr. 2021;15(6):470-476.
5. Budoff MJ, Lakshmanan S, Toth PP, et al. Cardiac CT angiography in current practice: an american society for preventive cardiology clinical practice statement*. Am J Prev Cardiol. 2022;9:100318.
6. Nieman K, García-García HM, Hideo-Kajita A, et al. Standards for quantitative assessments by coronary computed tomography angiography (CCTA): An expert consensus document of the society of cardiovascular computed tomography (SCCT). J Cardiovasc Comput Tomogr. 2024;S1934-5925(24).
7. Nerlekar N, Ha FJ, Cheshire C, et al. Computed tomographic coronary angiography-derived plaque characteristics predict major adverse cardiovascular events: a systematic review and meta-analysis. Circ Cardiovasc Imaging. 2018;11(1):e006973.
8. Williams MC, Kwiecinski J, Doris M, et al. Low-Attenuation noncalcified plaque on coronary computed tomography angiography predicts myocardial infarction: results from the multicenter SCOT-HEART trial (Scottish Computed Tomography of the HEART). Circulation. 2020;141(18):1452-1462.
9. U.S. Food and Drug Administration (FDA). Artificial Intelligence and Machine Learning in Software as a Medical Device. Accessed 9/4/2024.
10. Budoff MJ, Bhatt DL, Kinninger A, et al. Effect of icosapent ethyl on progression of coronary atherosclerosis in patients with elevated triglycerides on statin therapy: final results of the EVAPORATE trial. Eur Heart J. 2020;41(40):3925-3932.
11. Nielsen LH, Ortner N, Nørgaard BL, Achenbach S, Leipsic J, Abdulla J. The diagnostic accuracy and outcomes after coronary computed tomography angiography vs. conventional functional testing in patients with stable angina pectoris: a systematic review and meta-analysis. Eur Heart J Cardiovasc Imaging. 2014;15(9):961-971.
12. Narula J, Chandrashekhar Y, Ahmadi A, et al. SCCT 2021 expert consensus document on coronary computed tomographic angiography: a report of the Society of Cardiovascular Computed Tomography. J Cardiovasc Comput Tomogr. 2021;15(3):192-217.
13. Nakanishi R, Motoyama S, Leipsic J, Budoff MJ. How accurate is atherosclerosis imaging by coronary computed tomography angiography? J Cardiovasc Comput Tomogr. 2019;13(5):254-260.
14. Cleerly. Clinical Evidence for Personalized Heart Attack Prevention.
15. Newby DE, Adamson PD, Berry C, et al. Coronary CT angiography and 5-Year risk of myocardial infarction. N Engl J Med. 2018;379(10):924-933.
16. 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(9985):2383-2391.
17. Williams MC, Moss AJ, Dweck M, et al. Coronary artery plaque characteristics associated with adverse outcomes in the SCOT-HEART study. J Am Coll Cardiol. 2019;73(3):291-301.
18. Singh T, Bing R, Dweck MR, et al. Exercise electrocardiography and computed tomography coronary angiography for patients with suspected stable angina pectoris: a post hoc analysis of the randomized SCOT-HEART trial. JAMA Cardiol. 2020;5(8):920-928.
19. Bienstock S, Lin F, Blankstein R, et al. Advances in coronary computed tomographic angiographic imaging of atherosclerosis for risk stratification and preventive care. JACC Cardiovasc Imaging. 2023;16(8):1099-1115.
20. Lin A, Manral N, McElhinney P, et al. Deep learning-enabled coronary CT angiography for plaque and stenosis quantification and cardiac risk prediction: an international multicentre study. Lancet Digit Health. 2022;4(4):e256-e265.
21. Douglas PS, Hoffmann U, Patel MR, et al. Outcomes of anatomical versus functional testing for coronary artery disease. N Engl J Med. 2015;372(14):1291-1300.
22. Hoffmann U, Ferencik M, Udelson JE, et al. 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(24):2320-2332.
23. Ferencik M, Mayrhofer T, Bittner D, et al. 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(2):144-152
24. Bhindi R, Guan M, Zhao Y, Humphries KH, Mancini GBJ. Coronary atheroma regression and adverse cardiac events: a systematic review and meta-regression analysis. Atherosclerosis. 2019;284:194-201.
25. Lee SE, Chang HJ, Sung JM, et al. Effects of statins on coronary atherosclerotic plaques: the PARADIGM study. JACC Cardiovasc Imaging. 2018;11(10):1475-1484.
26. van Rosendael AR, Lin FY, van den Hoogen IJ, et al. Progression of whole-heart atherosclerosis by coronary CT and major adverse cardiovascular events. J Cardiovasc Comput Tomogr. 2021;15(4):322-330.
27. van Rosendael AR, van den Hoogen IJ, Gianni U, et al. Association of statin treatment with progression of coronary atherosclerotic plaque composition. JAMA Cardiol. 2021;6(11):1257-1266.
28. Perez de Isla L, Diaz-Diaz JL, Romero MJ, et al. Alirocumab and coronary atherosclerosis in asymptomatic patients with familial hypercholesterolemia: the ARCHITECT study. Circulation. 2023;147(19):1436-1443.
29. Hirai K, Imamura S, Hirai A, Ookawara S, Morishita Y. Effect of evolocumab on vulnerable coronary plaques: a serial coronary computed tomography angiography study. J Clin Med. 2020;9(10):3338.
30. Nicholls SJ, Puri R, Anderson T, et al. Effect of evolocumab on progression of coronary disease in statin-treated patients: the GLAGOV randomized clinical trial. JAMA. 2016;316(22):2373-2384.
31. Vaidya K, Arnott C, Martínez GJ, et al. Colchicine therapy and plaque stabilization in patients with acute coronary syndrome: a CT coronary angiography study. JACC Cardiovasc Imaging. 2018;11(2 Pt 2):305-316.
32. Reynolds HR, Shaw LJ, Min JK, et al. Outcomes in the ISCHEMIA trial based on coronary artery disease and ischemia severity. Circulation. 2021;144(13):1024-1038.
33. Ferencik M, Mayrhofer T, Puchner SB, et al. Computed tomography-based high-risk coronary plaque score to predict acute coronary syndrome among patients with acute chest pain—results from the ROMICAT II trial. J Cardiovasc Comput Tomogr. 2015;9(6):538-545.
34. Puchner SB, Liu T, Mayrhofer T, et al. 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(7):684-692.
35. Lu MT, Meyersohn NM, Mayrhofer T, et al. Central core laboratory versus site interpretation of coronary CT angiography: agreement and association with cardiovascular events in the PROMISE trial. Radiology. 2018;287(1):87-95.
36. Stuijfzand WJ, van Rosendael AR, Lin FY, et al. Stress myocardial perfusion imaging vs coronary computed tomographic angiography for diagnosis of invasive vessel-specific coronary physiology: predictive modeling results from the computed tomographic evaluation of atherosclerotic determinants of myocardial ischemia (CREDENCE) trial. JAMA Cardiol. 2020;5(12):1338-1348.
37. Greenwood JP, Maredia N, Younger JF, et al. Cardiovascular magnetic resonance and single-photon emission computed tomography for diagnosis of coronary heart disease (CE-MARC): a prospective trial. Lancet. 2012;379(9814):453-460.
38. Maron DJ, Hochman JS, Reynolds HR, et al. Initial invasive or conservative strategy for stable coronary disease. N Engl J Med. 2020;382(15):1395-1407.
39. Lipkin I, Telluri A, Kim Y, et al. Coronary CTA with AI-QCT interpretation: comparison with myocardial perfusion imaging for detection of obstructive stenosis using invasive angiography as reference standard. AJR Am J Roentgenol. 2022;219(3):407-419.
40. Griffin WF, Choi AD, Riess JS, et al. AI evaluation of stenosis on coronary CTA, comparison with quantitative coronary angiography and fractional flow reserve: a CREDENCE trial substudy. JACC Cardiovasc Imaging. 2023;16(2):193-205.
41. Yang S, Hoshino M, Koo B-K, et al. Relationship of plaque features at coronary CT to coronary hemodynamics and cardiovascular events. Radiology. 2022;305(3):578-587.
42. Dundas J, Leipsic J, Fairbairn T, et al. Interaction of AI-enabled quantitative coronary plaque volumes on coronary CT angiography, FFRCT, and clinical outcomes: a retrospective analysis of the ADVANCE registry. Circulation: Cardiovascular Imaging. 2024;17(3):e016143.
43. Min JK, Chang HJ, Andreini D, et al. Coronary CTA plaque volume severity stages according to invasive coronary angiography and FFR. J Cardiovasc Comput Tomogr. 2022;16(5):415-422.
44. Deseive S, Kupke M, Straub R, et al. Quantified coronary total plaque volume from computed tomography angiography provides superior 10-year risk stratification. Eur Heart J Cardiovasc Imaging. 2021;22(3):314-321.
45. Williams MC, Kwiecinski J, Doris M, et al. Sex-specific computed tomography coronary plaque characterization and risk of myocardial infarction. JACC: Cardiovasc Imaging. 2021;14(9):1804-1814.
46. Budoff MJ, Dowe D, Jollis JG, et al. 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(21):1724-1732.
47. Jonas RA, Barkovich E, Choi AD, et al. The effect of scan and patient parameters on the diagnostic performance of AI for detecting coronary stenosis on coronary CT angiography. Clin Imaging. 2022;84:149-158.
48. Cho GW, Ghanem AK, Quesada CG, et al. Quantitative plaque analysis with A.I. augmented CCTA in end-stage renal disease and complex CAD. Clin Imaging. 2022;89:155-161.
49. Jonas R, Earls J, Marques H, et al. Relationship of age, atherosclerosis and angiographic stenosis using artificial intelligence. Open Heart. 2021;8(2):e001832.
50. Tzimas G, Gulsin GS, Everett RJ, et al. Age-and sex-specific nomographic CT quantitative plaque data from a large international cohort. JACC: Cardiovasc Imaging. 2024;17(2):165-175.
51. Omori H, Matsuo H, Fujimoto S, et al. Determination of lipid-rich plaques by artificial intelligence-enabled quantitative computed tomography using near-infrared spectroscopy as reference. Atherosclerosis. 2023;386:117363.
52. Hakim D, Coskun AU, Maynard C, et al. Endothelial shear stress computed from coronary computed tomography angiography: A direct comparison to intravascular ultrasound. J Cardiovasc Comput Tomogr. 2023;17(3):201-210.
53. Grodecki K, Cadet S, Staruch AD, et al. Noncalcified plaque burden quantified from coronary computed tomography angiography improves prediction of side branch occlusion after main vessel stenting in bifurcation lesions: results from the CT-PRECISION registry. Clin Res Cardiol. 2021;110(1):114-123.
54. Narula J, Stuckey TD, Nakazawa G, et al. Prospective deep learning-based quantitative assessment of coronary plaque by computed tomography angiography compared with intravascular ultrasound: the REVEALPLAQUE study. Eur Heart J Cardiovasc Imaging. 2024:25(9):1287-1295.
55. Min JK, Dunning A, Lin FY, et al. Age-and sex-related differences in all-cause mortality risk based on coronary computed tomography angiography findings: results from the international multicenter [CONFIRM] (coronary CT angiography evaluation for clinical outcomes: an international multicenter registry) of 23,854 patients without known coronary artery disease. J Am Coll Cardiol. 2011;58(8):849-860.
56. Hadamitzky M, Achenbach S, Al-Mallah M, et al. Optimized prognostic score for coronary computed tomographic angiography: results from the CONFIRM registry (coronary CT angiography evaluation for clinical outcomes: an international multicenter registry). J Am Coll Cardiol. 2013;62(5):468-476.
57. Cho I, Al'Aref SJ, Berger A, et al. Prognostic value of coronary computed tomographic angiography findings in asymptomatic individuals: a 6-year follow-up from the prospective multicentre international CONFIRM study. Eur Heart J. 2018;39(11):934-941.
58. Chang HJ, Lin FY, Lee SE, et al. Coronary atherosclerotic precursors of acute coronary syndromes. J Am Coll Cardiol. 2018;71(22):2511-2522.
59. van Rosendael AR, Narula J, Lin FY, et al. Association of high-density calcified 1K plaque with risk of acute coronary syndrome. JAMA Cardiol. 2020;5(3):282-290.
60. Chang HJ, Lin FY, Gebow D, et al. Selective referral using CCTA versus direct referral for individuals referred to invasive coronary angiography for suspected CAD: a randomized, controlled, open-label trial. JACC Cardiovasc Imaging. 2019;12(7 Part 2):1303-1312.
61. Kim Y, Choi AD, Telluri A, et al. Atherosclerosis imaging quantitative computed tomography (AI-QCT) to guide referral to invasive coronary angiography in the randomized controlled CONSERVE trial. Clin Cardiol. 2023;46(5):477-483.
62. Nurmohamed NS, Bom MJ, Jukema RA, et al. AI guided quantitative plaque staging predicts long-term cardiovascular outcomes in patients at risk for atherosclerotic CVD. JACC Cardiovasc Imaging. 2024;17(3):269-280.
63. van Diemen PA, Bom MJ, Driessen RS, et al. Prognostic value of RCA pericoronary adipose tissue CT-attenuation beyond high-risk plaques, plaque volume, and ischemia. JACC Cardiovasc Imaging. 2021;14(8):1598-1610.
64. Driessen RS, Bom MJ, van Diemen PA, et al. Incremental prognostic value of hybrid [¹⁵O] H2O positron emission tomography–computed tomography: combining myocardial blood flow, coronary stenosis severity, and high-risk plaque morphology. Eur Heart J Cardiovasc Imaging. 2020;21(10):1105-1113.
65. Rinehart S, Raible S, Ng N, Mullen S, Huey W, Pursnani A. Utility of AI plaque quantification: results of the decisions for treating coronary disease are changed in patients evaluated with quantified plaque analysis (DECODE) study. J Cardiovasc Comput Tomogr. 2023;17(4):S33.
66. Nurmohamed NS, Cole JH, Budoff MJ, et al. Impact of atherosclerosis imaging-quantitative computed tomography on diagnostic certainty, downstream testing, coronary revascularization and medical therapy: the CERTAIN study. Eur Heart J Cardiovasc Imaging. 2024;25(6):857-866.
67. Park S-J, Ahn J-M, Kang D-Y, et al. Preventive percutaneous coronary intervention versus optimal medical therapy alone for the treatment of vulnerable atherosclerotic coronary plaques (PREVENT): a multicentre, open-label, randomised controlled trial. Lancet. 2024;403(10438):1753-1765.
68. Freeman AM, Raman SV, Aggarwal M, et al. Integrating coronary atherosclerosis burden and progression with coronary artery disease risk factors to guide therapeutic decision making. Am J Med. 2023;136(3):260-269.
69. Kim C, Park CH, Lee BY, et al. 2024 consensus statement on coronary stenosis and plaque evaluation in CT angiography from the Asian Society of Cardiovascular Imaging-Practical Tutorial (ASCI-PT). Korean J Radiol. 2024;25(4):331-342.
70. Howden N, Branch K, Douglas P, et al. Computed tomographic angiography measures of coronary plaque in clinical trials: opportunities and considerations to accelerate drug translation. Front Cardiovasc Med. 2024;11:1359500.
71. Schulman-Marcus J, Hartaigh B, Gransar H, et al. Sex-specific associations between coronary artery plaque extent and risk of major adverse cardiovascular events: the CONFIRM long-term registry. JACC Cardiovasc Imaging. 2016;9(4):364-372.
72. Lloyd-Jones DM, Braun LT, Ndumele CE, et al. Use of risk assessment tools to guide decision-making in the primary prevention of atherosclerotic cardiovascular disease: a special report from the American Heart Association and American College of Cardiology. Circulation. 2019;139(25):e1162-e1177.
73. Blanke P, Naoum C, Ahmadi A, et al. Long-term prognostic utility of coronary CT angiography in stable patients with diabetes mellitus. JACC Cardiovasc Imaging. 2016;9(11):1280-1288.
74. Andreini D, Conte E, Mushtaq S, et al. Plaque assessment by coronary CT angiography may predict cardiac events in high risk and very high risk diabetic patients: a long-term follow-up study. Nutr Metab Cardiovasc Dis. 2022;32(3):586-595.
75. Lo J, Lu MT, Ihenachor EJ, et al. Effects of statin therapy on coronary artery plaque volume and high-risk plaque morphology in HIV-infected patients with subclinical atherosclerosis: a randomised, double-blind, placebo-controlled trial. Lancet HIV. 2015;2(2):e52-63.
76. Lu MT, Ribaudo H, Foldyna B, et al. Effects of pitavastatin on coronary artery disease and inflammatory biomarkers in HIV: mechanistic substudy of the REPRIEVE randomized clinical trial. JAMA Cardiol. 2024;9(4):323-334.
77. Fuchs A, Kühl JT, Sigvardsen PE, et al. Subclinical coronary atherosclerosis and risk for myocardial infarction in a danish cohort: a prospective observational cohort study. Ann Intern Med. 2023;176(4):433-442.
78. Bax AM, van Rosendael AR, Ma X, et al. Comparative differences in the atherosclerotic disease burden between the epicardial coronary arteries: quantitative plaque analysis on coronary computed tomography angiography. Eur Heart J Cardiovasc Imaging. 2021;22(3):322-330.
79. Motoyama S, Ito H, Sarai M, et al. 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(4):337-346.
80. Lee SE, Sung JM, Rizvi A, et al. Quantification of coronary atherosclerosis in the assessment of coronary artery disease. Circ Cardiovasc Imaging. 2018;11(7):e007562.

• Quantitative Coronary Plaque Analysis
• Artificial Intelligence Enabled CT
• AI-QCT
• AI-CPA