Role of artificial intelligence in cardiovascular risk prediction and outcomes: comparison of machine-learning and conventional statistical approaches for the analysis of carotid ultrasound features and intra-plaque neovascularization

Amer M JohriLaura E MantellaAnkush D JamthikarLuca SabaJohn R LairdJasjit S Suri

https://pubmed.ncbi.nlm.nih.gov/34050838/

Abstract

The aim of this study was to compare machine learning (ML) methods with conventional statistical methods to investigate the predictive ability of carotid plaque characteristics for assessing the risk of coronary artery disease (CAD) and cardiovascular (CV) events. Focused carotid B-mode ultrasound, contrast-enhanced ultrasound, and coronary angiography were performed on 459 participants. These participants were followed for 30 days. Plaque characteristics such as carotid intima-media thickness (cIMT), maximum plaque height (MPH), total plaque area (TPA), and intraplaque neovascularization (IPN) were measured at baseline. Two ML-based algorithms-random forest (RF) and random survival forest (RSF) were used for CAD and CV event prediction. The performance of these algorithms was compared against (i) univariate and multivariate analysis for CAD prediction using the area-under-the-curve (AUC) and (ii) Cox proportional hazard model for CV event prediction using the concordance index (c-index). There was a significant association between CAD and carotid plaque characteristics [cIMT (odds ratio (OR) = 1.49, p = 0.03), MPH (OR = 2.44, p < 0.0001), TPA (OR = 1.61, p < 0.0001), and IPN (OR = 2.78, p < 0.0001)]. IPN alone reported significant CV event prediction (hazard ratio = 1.24, p < 0.0001). CAD prediction using the RF algorithm reported an improvement in AUC by ~ 3% over the univariate analysis with IPN alone (0.97 vs. 0.94, p < 0.0001). Cardiovascular event prediction using RSF demonstrated an improvement in the c-index by ~ 17.8% over the Cox-based model (0.86 vs. 0.73). Carotid imaging phenotypes and IPN were associated with CAD and CV events. The ML-based system is superior to the conventional statistically-derived approaches for CAD prediction and survival analysis.

Keywords: And cardiovascular event prediction; Coronary artery disease; Focused carotid ultrasound; Intraplaque neovascularization; Machine learning; Risk prediction.

Congratulations Laura!

Congratulations to Laura Mantella who successfully defended her PhD thesis last week (future MD/PhD)! Impressive work on carotid plaque neovascularization associated with coronary artery disease.

Dr. Johri (CINQ) is Awarded CIHR for: Combining Intraplaque Neovascularization with Risk Stratification by Carotid Stress Echo (The CIRCE Study)

Stress echo (SE) is one of the most accessible tests used to select patients for angiography, but suffers from moderate sensitivity and specificity for cardiovascular (CV) outcomes. In Ontario alone, of the >100,000 SEs performed every year, 90% are normal, and patients may have little follow-up. It is estimated that 3-5% (~4,000 patients/year) of those with a negative SE have a major adverse cardiovascular event(MACE) within 3 years. We have demonstrated that adding carotid plaque assessment by ultrasound to the SE is feasible and improves testing with a net reclassification improvement (NRI) of 25% for coronary artery disease (CAD). We reported that activity of plaque adds further discriminatory power for predicting CAD and MACE. Plaques can be assessed by intraplaque neovascularization (IPN), which occurs in response to hypoxia or inflammation within the lesion. IPN occurs when vessels grow from the vasa vasorum into the lesion, resulting in a fragile and leaky network at risk of rupture and hemorrhage. IPN of carotid arterial plaque can now be quantified using ultrasound contrast. We hypothesize that addition of carotid IPN detection at the time of SE will enhance risk prediction for MACE. We plan to demonstrate that plaque inflammatory activity, detected by IPN serves as a powerful imaging biomarker of CAD and CV events to improve the sensitivity of SE.

This is a prospective, parallel, 6-centre study assessing IPN + SE in 1500 consecutive outpatients referred for SE. Patients will be recruited from a community-based cardiac clinic, and the Universities of Queen’s, Dalhousie, Calgary, Toronto, and RUMC(Chicago). The NPV and sensitivity of IPN + SE for ruling out CAD will be determined. Follow-up will be 3-year MACE. The proposal is founded upon a simple, inexpensive, and safe addition to the workflow of an existing non-invasive test. Extensive work to date indicates that plaque assessment added to SE will enhance stratification to reduce referral for unnecessary angiography and better identify patients at risk. This multi-center study of IPN + SE is expected to show increased predictive power for MACE, establishing a new standard for CV risk stratification.

The CIRCE study was the top ranked proposal in its committee and awarded the 2020/2021 CIHR Project grant led by Queen’s University.

Recommendations for the Assessment of Carotid Arterial Plaque by Ultrasound for the Characterization of Atherosclerosis and Evaluation of Cardiovascular Risk: From the American Society of Echocardiography

Amer M Johri, Vijay Nambi, Tasneem Z Naqvi, Steven B Feinstein, Esther S H Kim, Margaret M Park, Harald Becher, Henrik Sillesen

J Am Soc Echocardiogr 2020;33:917-33

https://pubmed.ncbi.nlm.nih.gov/32600741/

Atherosclerotic plaque detection by carotid ultrasound provides cardiovascular disease risk stratification. The advantages and disadvantages of two-dimensional (2D) and three-dimensional (3D) ultrasound methods for carotid arterial plaque quantification are reviewed. Advanced and emerging methods of carotid arterial plaque activity and composition analysis by ultrasound are considered. Recommendations for the standardization of focused 2D and 3D carotid arterial plaque ultrasound image acquisition and measurement for the purpose of cardiovascular disease stratification are formulated. Potential clinical application towards cardiovascular risk stratification of recommended focused carotid arterial plaque quantification approaches are summarized.