Prognostic Value of Machine Learning-based Time-to-Event Analysis Using Coronary CT Angiography in Patients with Suspected Coronary Artery Disease.

Purpose: To assess the long-term prognostic value of a machine learning (ML) approach in time-to-event analyses incorporating coronary CT angiography (CCTA)-derived and clinical parameters in patients with suspected coronary artery disease. Materials and Methods: The retrospective analysis included patients with suspected coronary artery disease who underwent CCTA between October 2004 and December 2017. Major adverse cardiovascular events were defined as the composite of all-cause death, myocardial infarction, unstable angina, or late revascularization (>90 days after index scan). Clinical and CCTA-derived parameters were assessed as predictors of major adverse cardiovascular events and incorporated into two models: a Cox proportional hazards model with recursive feature elimination and an ML model based on random survival forests. Both models were trained and validated by employing repeated nested cross-validation. Harrell concordance index (C-index) was used to assess the predictive power. Results: A total of 5457 patients (mean age, 61 years ± 11 [SD]; 3648 male patients) were evaluated. The predictive power of the ML model (C-index, 0.74; 95% CI: 0.71, 0.76) was significantly higher than the Cox model (C-index, 0.71; 95% CI: 0.68, 0.74; P = .02). The ML model also outperformed the segment stenosis score (C-index, 0.69; 95% CI: 0.66, 0.72; P < .001), which was the best performing CCTA-derived parameter, and patient age (C-index, 0.66; 95% CI: 0.63, 0.69; P < .001), the best performing clinical parameter. Conclusion: An ML model for time-to-event analysis based on random survival forests had higher performance in predicting major adverse cardiovascular events compared with established clinical or CCTA-derived metrics and a conventional Cox model.Keywords: Machine Learning, CT Angiography, Cardiac, Arteries, Heart, Arteriosclerosis, Coronary Artery DiseaseSupplemental material is available for this article.© RSNA, 2023.

Convolutional neural networks on risk stratification of patients with suspected coronary artery disease undergoing coronary computed tomography angiography.

To assess the prognostic value of convolutional neural networks (CNN) on coronary computed tomography angiography (CCTA) in comparison to conventional computed tomography (CT) reporting and clinical risk scores. 5468 patients who underwent CCTA with suspected coronary artery disease (CAD) were included. Primary endpoint was defined as a composite of all-cause death, myocardial infarction, unstable angina or late revascularization (> 90 days after CCTA). Early revascularization was additionally included as a training endpoint for the CNN algorithm. Cardiovascular risk stratification was based on Morise score and the extent of CAD (eoCAD) as assessed on CCTA. Semiautomatic post-processing was performed for vessel delineation and annotation of calcified and non-calcified plaque areas. Using a two-step training of a DenseNet-121 CNN the entire network was trained with the training endpoint, followed by training the feature layer with the primary endpoint. During a median follow-up of 7.2 years, the primary endpoint occurred in 334 patients. CNN showed an AUC of 0.631 ± 0.015 for prediction of the combined primary endpoint, while combining it with conventional CT and clinical risk scores showed an improvement of AUC from 0.646 ± 0.014 (based on eoCAD only) to 0.680 ± 0.015 (p < 0.0001) and from 0.619 ± 0.0149 (based on Morise Score only) to 0.6812 ± 0.0145 (p < 0.0001), respectively. In a stepwise model including all prediction methods, it was found an AUC of 0.680 ± 0.0148. CNN analysis showed to improve conventional CCTA-derived and clinical risk stratification when evaluating CCTA of patients with suspected CAD.

Radiation Dose and Image Quality of a High-Pitch Prospective Spiral First Approach in Coronary Computed Tomography Angiography (CCTA).

Objective: To investigate a high-pitch spiral first (HPSF) approach for coronary computed tomography angiography (CCTA) in an unselected patient cohort and compare diagnostic yield and radiation exposure to CCTAs acquired via conventional, non-high-pitch spiral first (NHPSF) scan regimes. Materials and Methods: All consecutive patients from 1 January 2015 to 31 December 2017 were included. Two investigation protocols (HPSF/NHPSF) were used with the aim to achieve diagnostic image quality of all coronary segments. Low-pitch secondary scans followed the initial examination if image quality was unsatisfactory. Dosage and image quality were compared between both regimes. Results: 1410 patients were subject to a HPSF and 236 patients to a NHPSF approach. While the HPSF approach led to a higher fraction of re-scans (35% vs. 11%, p < 0.001), the fraction of aggregate scans that remained non-diagnostic after considering the initial and secondary scan was comparably low for the HPSF and NHPSF approach (0.78 vs. 0%, p = 0.18). Aggregate radiation exposure in the HPSF protocol was significantly lower (1.12 mSv (IQR: 0.73, 2.10) vs. 3.96 mSv (IQR: 2.23, 8.33) p < 0.001). Conclusions: In spite of a higher number of re-scans, a HPSF approach leads to a reduction in overall radiation exposure with diagnostic yields similar to a NHPSF approach.

Sex- and age-specific differences in the long-term prognostic value of morphological plaque features detected by coronary computed tomography angiography.

BACKGROUND: Although sex- and age-specific differences in coronary plaque features detected by coronary computed tomography angiography (CCTA) are known, insufficient information regarding the long-term prognostic value of these findings exists. METHODS: A total of 1615 patients with suspected but not previously diagnosed coronary artery disease (CAD) were examined by CCTA and coronary plaque features were assessed. The median follow-up period was 10.5 (IQR 9.2-11.4) years. Cox proportional-hazards analysis was used for the combined endpoint of cardiac death or nonfatal myocardial infarction. RESULTS: The endpoint occurred more often in patients older than 65 years (5.66% vs. 2.05%; p = 0.00029) but similarly between female (3.34%) and male (3.07%) patients (p = 0.76). Both sexes displayed a similar prevalence for noncalcified (female vs. male: 0.77 ± 1.38 vs. 0.89 ± 1.41; p = 0.098) and low-attenuation (female vs. male: 2.6% vs. 4.37%; p = 0.096) plaques. As assessed by p for interaction CADRADS (p for interaction = 0.013), noncalcified plaques (p for interaction = 0.022) and low-attenuation plaques (p for interaction = 0.045) had a better primary endpoint association in women than in men. Concerning age, no difference in outcome association was apparent as evaluated by p for interaction. CONCLUSION: CCTA demonstrates excellent long-term prognostic value irrespective of sex and age and independent from the higher prevalence of atherosclerotic plaques in men and patients older than 65 years. Although similarly prevalent in both sexes, noncalcified and low-attenuation plaques exhibit a better prognostic value in women.

Long-term prognostic value of morphological plaque features on coronary computed tomography angiography.

AIMS: To investigate the incremental prognostic value of morphological plaque features beyond clinical risk and coronary stenosis levels. Although associated with the degree of coronary stenosis, most cardiac events occur on the basis of ruptured non-obstructive plaques and consecutive vessel thrombosis. As such, identification of vulnerable plaques is paramount for cardiovascular risk prediction and treatment decisions. METHODS AND RESULTS: A total of 1615 patients with suspected but not previously diagnosed coronary artery disease (CAD) were examined by coronary computed tomography angiography and morphological plaque features were assessed. Mean follow-up was 10.5 (interquartile range 9.2-11.4) years. Cox proportional hazards analysis was used for the composite endpoint of cardiac death and non-fatal myocardial infarction. The study endpoint was reached in 51 patients (36 cardiac deaths, 15 non-fatal myocardial infarctions). In addition to quantitative parameters (presence of any calcified/non-calcified plaque or elevated plaque load), morphologic plaque features such as a spotty or gross calcification pattern and napkin-ring sign (NRS) were predictive for events. However, only spotty calcified plaques and NRS could confer additive prognostic value beyond clinical risk and coronary stenosis level. In a stepwise approach, endpoint prediction beyond clinical risk (Morise score) could be improved by inclusion of CAD severity (χ2 of 27.5, P < 0.001) and further discrimination for spotty calcified plaques (χ2 of 3.89, P = 0.049). CONCLUSION: Improved cardiovascular risk prediction beyond clinical risk and coronary stenosis levels can be made by discriminating for the presence of spotty calcified plaques. Thus, an intensified prophylactic anti-atherosclerotic treatment appears to be warranted in patients with coronary plaques that show spotty calcifications.

Coronary computed tomography angiography as a tool for long-term cardiovascular risk stratification in diabetic patients.

Objectives of the study were to examine the long-term prognostic power of coronary computed tomography angiography (CCTA) to predict death or myocardial infarction in patients with diabetes mellitus (DM). The prognostic value of CCTA in diabetic patients has been confirmed for short- and intermediate follow-up durations. The slowly progressing nature of coronary artery disease (CAD), however, underlines the necessity to validate CCTA for longer observation periods in this high-risk population. A total of 132 patients with DM and 1781 without DM were examined by CCTA and followed for a median duration of 9.7 (IQR 6.9, 11.2) and 9.9 (IQR 6.9, 11.1) years, respectively. Cox proportional hazards analysis was used for the composite endpoint of death and myocardial infarction. Warranty period was defined as the number of years that an individual stays in a low-risk group with a cumulative probability for the endpoint below 1% and calculated for patients with/without DM and rising degrees of CAD. The study endpoint was reached in 12 (9.1%) patients with and 87 (4.9%) patients without DM (p = 0.024). Quantification of coronary stenosis by CADRADS or CAD severity (normal/non-obstructive/obstructive) was incremental for endpoint prediction with a multivariate (+Morise) χ2 of 3.90 and 3.85, respectively. The lowest annual event rate of 0.19% was noted in non-diabetic patients with no CAD, translating to a warranty period of 5.26 years. The highest annual event rate of 1.73% was found in diabetic patients with obstructive CAD, corresponding to a warranty period of 0.58 years. Compared to patients with no DM and no CAD, the risk of death or myocardial infarction in diabetic patients increased with rising levels of coronary obstruction at multivariate hazard ratios (HR) of 3.28 [95% CI 2.32, 4.64 (p < 0.001)], 3.02 [95% CI 2.19, 4.17 (p < 0.001)] and 9.40 [95% CI 4.90, 18.03 (p < 0.001)] for normal coronary arteries, non-obstructive CAD and obstructive CAD. This study validates the long-term prognostic utility of CCTA-assessed CAD for predicting death or myocardial infarction in a population of patients with DM. The rates of death or myocardial infarction rise with CAD severity in diabetic and non-diabetic patients, identifying the highest risk group of patients with DM and obstructive CAD.

10-Year Follow-Up After Coronary Computed Tomography Angiography in Patients With Suspected Coronary Artery Disease.

OBJECTIVES: The aim of this study was to determine the long-term prognostic power of coronary computed tomography angiography (CTA) to predict cardiac death and nonfatal myocardial infarction. BACKGROUND: Prognostic usefulness of coronary CTA has been confirmed for short- and intermediate-term follow-up. However, long-term data for prognostic usefulness is still lacking, but is paramount because of the slowly progressing nature of coronary artery disease (CAD). METHODS: A total of 2,011 patients with suspected but not previously diagnosed CAD were examined by coronary CTA. Mean follow-up was 10.0 years (interquartile range [IQR]: 8.1 to 11.2 years). Cox proportional hazards analysis was used for the composite endpoint of cardiac death and nonfatal myocardial infarction. Event-free survival, which was defined as the years it took to reach a cumulative 1% risk for the composite endpoint and reclassification from clinical risk, was calculated. RESULTS: The study endpoint was reached in 58 patients (42 cardiac deaths, 16 nonfatal myocardial infarctions). Coronary CTA-assessed CAD severity (normal, nonobstructive, or obstructive) showed the best correlation with the endpoint, with an adjusted c-index of 0.704, compared with a univariate c-index of 0.622 for the clinical risk model (Morise score) alone. The annual event rate for patients with normal coronary arteries on baseline coronary CTA was 0.04%, which translated to an event-free survival period of 10 years. The highest annual event rate of 1.33% was found in patients with 3-vessel obstructive CAD. Reclassification from clinical risk (Morise score) was possible in approximately two-thirds of all patients (68%; p < 0.0001), which led to a substantial reduction of the intermediate-risk group (reduction from 74% to 15%) in favor of the low-risk group (increase from 20% to 83%). CONCLUSIONS: Patients with normal coronary CTA results benefitted from an event-free survival period of 10 years against cardiac death and nonfatal myocardial infarction. Risk stratification according to coronary CTA results allowed for the delineation of clearly diverging prognostic groups and reclassified approximately two-thirds of all patients from clinical risk groups.

Performance of native and contrast-enhanced T1 mapping to detect myocardial damage in patients with suspected myocarditis: a head-to-head comparison of different cardiovascular magnetic resonance techniques.

Myocardial T1 mapping is a novel technique that has proven to be superior to standard imaging for differentiation between healthy individuals in acute myocarditis. Aim of this study was comparison of T1 mapping with a clinical biomarker. We retrospectively investigated 171 patients undergoing cardiovascular magnetic resonance (CMR) examination with suspected myocarditis by performing native and contrast enhanced T1-mapping. Additionally, T2w and T1w images and late gadolinium enhancement sequences (LGE) were utilized for myocardial evaluation; Lake Louise Criteria comprise T1w, T2w and LGE imaging in a score. Reference for positive myocarditis diagnosis was a ten-fold increase of troponin level above normal (0.14 ng/ml). Native T1 and extracellular volume (ECV) showed good association with relevant troponin elevations. Area under the curve (AUC) was 81% (p = 0.0001) for native T1 with an optimal threshold of 979 ms and 86% (p < 0.0001) for ECV with an optimal cutoff of 32.4%. AUC for T2w imaging (T2-signal intensity ratio to skeletal muscle) was 77% (p = 0.0003). AUC for T2w imaging (T2-signal intensity compared to remote myocardium) was 69% (p = 0.012). Additionally, we found positive correlation for native T1 and ECV with the Lake Louise Criteria (r = 0.44, p = 0.0001 for native T1 and r = 0.45, p = 0.0001 for ECV). Correlated to troponin as biomarker, ECV and native T1 mapping perform at least equally well in comparison to established CMR-techniques LGE, T2w imaging and the combined Lake Louise Criteria in detecting acute myocardial damage. Normal ECV values rule out myocardial damage with very high certainty. T1 mapping qualifies for further prospective evaluations to evolve as a separate biomarker.

Prognostic value of T1-mapping in TAVR patients: extra-cellular volume as a possible predictor for peri- and post-TAVR adverse events.

The benefit of a transcatheter aortic valve replacement (TAVR) can differ in patients, and therapy bears severe risks. High-degree aortic stenosis can lead to cardiac damage such as diffuse myocardial fibrosis, evaluable by extra-cellular volume (ECV) in CMR. Therefore, fibrosis might be a possible risk factor for unfavorable outcome after TAVR. We sought to assess the prognostic value of T1-mapping and ECV to predict adverse events during and after TAVR. The study population consisted of patients undergoing clinically indicated TAVR by performing additional CMR with native and contrast-enhanced T1-mapping sequences for additional evaluation of ECV. Study endpoints were congestive heart failure (CHF) and TAVR-associated conduction abnormalities defined as new onset of left bundle branch block (LBBB), AV-Block or implantation of a pacemaker. 94 patients were examined and followed. Median follow up time was 187 days (IQR 79-357 days). ECV was increased (>30 %) in 38 patients (40 %). There was no significant correlation between ECV and death, Hazard ratio (HR) 0.847 (95 % CI 0.335; 2.14), p = 0.72. ECV in patients with subsequent CHF was higher than in those without an event (33.5 ± 4.6 and 29.1 ± 4.1 %, respectively), but the difference just did not reach the level of significance HR 2.16 (95 % CI 0.969; 4.84), p = 0.06. Patients with post-TAVR conduction abnormality (LBBB, AV-block or pacemaker implantation) had statistically relevant lower ECV values compared to those without an event. Patients with an event had a mean ECV of 28.1 ± 3.16 %; patients without an event had a mean ECV of 29.8 ± 4.53, HR 0.56 (95 % CI 0.32; 0.96), p = 0.036. In this study, elevated myocardial ECV is a predictor of CHF by trend; CMR may be helpful in identifying patients with a high risk for post-TAVR cardiac decompensation benefitting from an intensified post-interventional surveillance. Patients with post-TAVR conductions abnormalities have a significantly decreased ECV. Nevertheless, it remains unclear which precise molecular tissue alteration is the protective factor or risk factor in this case.

Incremental prognostic value of quantitative plaque assessment in coronary CT angiography during 5 years of follow up.

OBJECTIVE: We sought to assess the incremental prognostic value of quantitative plaque characterization beyond established CT risk scores. BACKGROUND: Several plaque characteristics detectable by coronary computed tomographic angiography (coronary CTA) are thought to be indicative of vulnerable plaques and subsequent cardiac events, particularly low attenuation plaque volume (LAPV), positive remodeling and the napkin-ring sign which is high density vascular adhesion with a small center of low density. It is unknown how quantitative plaque assessment can contribute to the long-term prediction of cardiovascular events in relation to established CT risk scores such as the calcium score or Segment Stenosis Score (SSS). METHODS: In 1168 consecutive patients with suspected coronary artery disease (CAD), calcium score measurement and coronary plaque characterization was performed comprising the presence of calcified, non-calcified, and partially calcified plaques on a per-segment basis. In all non-calcified or partially calcified plaques, semi-automated plaque analysis was performed to quantify low attenuation plaque volume (density <30 HU), total non-calcified plaque volume (<150 HU, TNCPV) and remodeling index. The presence of the napkin-ring sign was assessed visually. The study endpoint was the occurrence of major adverse cardiac events (MACE), a composite of cardiac death, myocardial infarction and coronary revascularization more than 90 days after coronary CTA. RESULTS: During a clinical follow up of 5.7 years, MACE was observed in 46 patients (3.9%). All plaque characteristics were associated with MACE. The strongest association was observed for LAPV (HR 1.12, p < 0.0001). LAPV showed incremental prognostic value in a stepwise multivariable model including the Morise Score for clinical risk, calcium score and SSS (p = 0.036). CONCLUSION: LAPV, TPV, PR and presence of the napkin-ring sign are predictors of MACE independently of clinical risk presentation. LAPV carries slight additional prognostic information beyond the calcium score and conventional coronary CTA analysis. It may therefore improve risk prediction after CT imaging.

Prognostic value of coronary CT angiography in diabetic patients: a 5-year follow up study.

Prevalence of coronary artery disease (CAD) is high in diabetic patients while diagnosis of early stage of CAD remains demanding. This study evaluates prognostic value of coronary computed tomography angiography (CCTA) for long-term outcome to predict cardiac events in oligosymptomatic diabetic patients. A cohort of 108 consecutive diabetic patients without angina pectoris or known CAD, undergoing CCTA was included. 1379 consecutive patients without diabetes were defined as a control group. Coronary artery calcium score (CACS), segment involvement score (SIS) and the segment stenosis score (SSS) were documented. The end point was a composite of cardiac events defined as all-cause death, nonfatal myocardial infarction, or unstable angina requiring hospitalization. Follow up period was 66.0 ± 14.2 month. 98% of initially enrolled patient were followed. During follow-up period 10 cardiac events within the diabetic cohort and 48 within the non-diabetic cohort were observed. Annual event rate in diabetic and non-diabetic patients was 1.74 and 0.64% respectively. In diabetic patients a multivariate analysis showed significant prognostic value over Framingham Score for SIS with a hazard ratio (HR) of 2.98 (95% CI 1.02, 8.72; p = 0.047) and SSS (HR 4.47, 95% CI 1.21, 16.49; p = 0.025), while CACS did not add prognostic value in this cohort. Annual event rate was 0% in diabetic patients with SIS = 0 and 3.9% in diabetic patients with SIS ≥ 8. CCTA allows for improved risk prediction for subsequent cardiac events in oligosymptomatic diabetic patients.

Routine clinical cardiovascular magnetic resonance in paediatric and adult congenital heart disease: patients, protocols, questions asked and contributions made.

BACKGROUND: Cardiovascular magnetic resonance (CMR) of patients with congenital heart disease (CHD) has become routine clinical practice. However, existing CMR protocols focus predominantly on patients with ischemic heart disease, and information is limited on the types of patient with CHD who benefit from CMR investigation, and in what ways. Therefore the aim of this study was to answer the questions: What type of patients were studied by CMR in a centre specializing in paediatric and adult CHD management? What questions were asked, which protocols were used and were the questions successfully answered? To answer these questions, we conducted a cohort study of all 362 patients that received routine clinical CMR during 2007 at the Department of Paediatric Cardiology and Congenital Heart Disease at the Deutsches Herzzentrum München. RESULTS: Underlying diagnosis was in 33% Fallot's tetralogy, 17% aortic coarctation, 8% Ebstein's disease, 6% Marfan's disease, 4% single ventricle with Fontan-like circulation, and 32% others. Median age was 26 years (7 days - 75 years). Ventricular volumes were assessed in 67% of the patients; flow in 74%; unknown anatomy only in 9%; specific individual morphology of known anatomy in 83%; myocardial fibrosis in 8%; stress-induced myocardial perfusion defects in 1%. Only in 3% of the cases the question could not be fully answered. CONCLUSION: Contrary to common belief, routine CMR of patients with CHD was not requested to address global anatomical questions so much as to clarify specific questions of morphology and function of known anatomy. The CMR protocols used differed markedly from those widely used in patients with ischemic heart disease.

Reduction of radiation dose estimates in cardiac 64-slice CT angiography in patients after coronary artery bypass graft surgery.

OBJECTIVES: The objectives of this prospective investigation in patients after bypass graft surgery were (1) to estimate radiation dose for routine bypass graft computed tomography (CT) angiography, (2) to study the impact of anatomically adapted and ECG-controlled tube current modulation on radiation dose estimates, and (3) effects on qualitative and quantitative image quality parameters. METHODS: Radiation dose was estimated for 194 consecutive patients undergoing 64-slice CT angiography (Somatom Sensation 64 Cardiac, Siemens Medical Solutions). The impact of anatomically adapted tube current modulation was studied in 2 consecutive patients groups. Furthermore, the impact of ECG-controlled tube current modulation, applied as indicated, was evaluated in both groups. RESULTS: Mean radiation dose estimate for a 64-slice CT bypass graft study was 18.9 +/- 6.0 mSv (CTDIvol 42.3 +/- 12.9 mGy). The application of anatomically adapted tube current modulation had no effect on dose parameters (CTDIvol 43.3 +/- 13.2 vs. 40.1 +/- 12.1 mGy for those with versus those without anatomically adapted tube current modulation, P = 0.1). Additional implementation of ECG-controlled tube current modulation resulted in reduced dose parameters (CTDIvol of 32.9 +/- 2.6 vs. 58.9 +/- 3.9 mGy and radiation dose estimates: 14.7 +/- 1.9 mSv vs. 26.5 +/- 2.1 mSv for those with versus those without ECG pulsing, both P < 0.01). There was no deterioration in image quality with use of tube current modulation algorithms. CONCLUSIONS: The radiation burden associated with 64-slice bypass graft CT angiographies is substantial. Anatomically adapted tube current modulation does not reduce radiation dose parameters, whereas ECG-controlled tube current modulation was associated with a 45% reduction in dose estimates. Application of both tube current modulation algorithms did not result in reduced image quality.