Can Machine Learning Aid the Selection of Percutaneous vs Surgical Revascularization?

Ten-year survival benefit and appropriateness of surgical versus percutaneous revascularization in synergy between percutaneous coronary intervention with Taxus and cardiac surgery randomized trial

OBJECTIVES

Average treatment effects from randomized trials do not reflect the heterogeneity of an individual's response to treatment. This study evaluates the appropriate proportions of patients for coronary artery bypass grafting, or percutaneous intervention based on the predicted/observed ratio of 10-year all-cause mortality in the SYNTAX population.

METHODS

The study included 1800 randomized patients and 1275 patients in the nested percutaneous (n = 198) or surgical (n = 1077) registries. The primary end point was 10-year all-cause mortality. The SYNTAX score II-2020 was validated internally in the randomized cohort and externally in the registry cohort. Proportions of patients with survival benefits from coronary artery bypass grafting or percutaneous intervention were determined using SYNTAX score II-2020.

RESULTS

Ten-year mortality was 23.8% for coronary artery bypass grafting, 28.6% for percutaneous intervention in the randomized cohort, 27.6% for coronary artery bypass grafting and 55.4% for percutaneous intervention in the registries. In the coronary artery bypass grafting registry, the SYNTAX score II-2020 predicted 10-year mortality with helpful calibration and discrimination (C-index: 0.70, intercept: 0.00, slope: 0.76). The proportion of patients deriving a predicted survival benefit from coronary artery bypass grafting over percutaneous intervention was 82.4% (2143/2602) and 17.7% (459/2602) for the entire SYNTAX trial population, translating into a 4.7 to 1 appropriate ratio of treatment allocation to coronary artery bypass grafting and percutaneous intervention.

CONCLUSIONS

Choosing a revascularization modality should depend on an individual's long-term prognosis rather than average treatment effects. Additionally, patients should be informed about their predicted prognosis.

TRIAL REGISTRATION

Registered on clinicaltrial.gov.

SYNTAXES

NCT03417050 (https://clinicaltrials.gov/ct2/show/NCT03417050).

SYNTAX

NCT00114972 (https://www.clinicaltrials.gov/ct2/show/NCT00114972).

Exploring the predictive values of CRP and lymphocytes in coronary artery disease based on a machine learning and Mendelian randomization

Purpose

To investigate the predictive value of leukocyte subsets and C-reactive protein (CRP) in coronary artery disease (CAD).

Methods

We conducted a Mendelian randomization analysis (MR) on leukocyte subsets, C-reactive protein (CRP) and CAD, incorporating data from 68,624 patients who underwent coronary angiography from 2010 to 2022. After initial screening, clinical data from 46,664 patients were analyzed. Techniques employed included propensity score matching (PSM), logistic regression, lasso regression, and random forest algorithms (RF). Risk factors were assessed, and the sensitivity and specificity of the models were evaluated using receiver operating characteristic (ROC) curves. Additionally, survival analysis was conducted based on a 36-month follow-up period.

Results

The inverse variance weight (IVW) analysis showed that basophil count (OR 0.92, 95% CI: 0.84-1.00, P = 0.048), CRP levels (OR 0.87, 95% CI: 0.73-1.00, P = 0.040), and lymphocyte count (OR 1.10, 95% CI: 1.04-1.16, P = 0.001) are significant risk factors for CAD. Using LASSO regression, logistic regression, and RF analysis, both CRP and lymphocyte counts were consistently identified as risk factors for CAD, prior to and following PSM. The ROC curve analysis indicated that the combination of lymphocyte and CRP levels after PSM achieves a higher diagnostic value (0.85). Survival analysis revealed that high lymphocyte counts and low CRP levels are associated with a decreased risk of Major Adverse Cardiovascular Events (MACE) (P < 0.001). Conversely, a higher CRP level combined with lymphocyte counts correlates with a poorer prognosis.

Conclusion

There is a causal relationship between lymphocytes, CRP and CAD. The combined assessment of CRP and lymphocytes offers diagnostic value for CAD. Furthermore, high CRP levels coupled with low lymphocyte counts are associated with a poor prognosis.

Editor-in-Chief's Top Picks From 2023

Each week, I record audio summaries for every paper in JACC, as well as an issue summary. This process has become a true labor of love due to the time they require, but I am motivated by the sheer number of listeners (16M+), and it has allowed me to familiarize myself with every paper that we publish. Thus, I have selected the top 100 papers (Original Investigations, Review Articles, Society Documents, and the Global Burden of Diseases) from distinct specialties each year. In addition to my personal choices, I have included papers that have been the most accessed or downloaded on our websites, as well as those selected by the JACC Editorial Board members. In order to present the full breadth of this important research in a consumable fashion, we will present these abstracts in this issue of JACC, as well as their Central Illustrations∗ and podcasts. The highlights comprise the following sections: Aorta; Basic and Translational Science; Cardiac Failure, Myocarditis, and Pericarditis; Cardiomyopathies and Genetics; Congenital Heart Disease; Coronary, Peripheral, and Structural Interventions; Coronavirus; Health Promotion and Preventive Cardiology; Imaging; Metabolic and Lipid Disorders; Neurovascular Disease and Dementia; Rhythm Disorders and Thromboembolism; and Valvular Heart Disease.1-104 ∗ To view the full manuscript, including the full-sized Central Illustration, please refer to the original publication in JACC.

Predicting the risk of subclinical atherosclerosis based on interpretable machine models in a Chinese T2DM population

Background

Cardiovascular disease (CVD) has emerged as a global public health concern. Identifying and preventing subclinical atherosclerosis (SCAS), an early indicator of CVD, is critical for improving cardiovascular outcomes. This study aimed to construct interpretable machine learning models for predicting SCAS risk in type 2 diabetes mellitus (T2DM) patients.

Methods

This study included 3084 T2DM individuals who received health care at Zhenhai Lianhua Hospital, Ningbo, China, from January 2018 to December 2022. The least absolute shrinkage and selection operator combined with random forest-recursive feature elimination were used to screen for characteristic variables. Linear discriminant analysis, logistic regression, Naive Bayes, random forest, support vector machine, and extreme gradient boosting were employed in constructing risk prediction models for SCAS in T2DM patients. The area under the receiver operating characteristic curve (AUC) was employed to assess the predictive capacity of the model through 10-fold cross-validation. Additionally, the SHapley Additive exPlanations were utilized to interpret the best-performing model.

Results

The percentage of SCAS was 38.46% (n=1186) in the study population. Fourteen variables, including age, white blood cell count, and basophil count, were identified as independent risk factors for SCAS. Nine predictors, including age, albumin, and total protein, were screened for the construction of risk prediction models. After validation, the random forest model exhibited the best clinical predictive value in the training set with an AUC of 0.729 (95% CI: 0.709-0.749), and it also demonstrated good predictive value in the internal validation set [AUC: 0.715 (95% CI: 0.688-0.742)]. The model interpretation revealed that age, albumin, total protein, total cholesterol, and serum creatinine were the top five variables contributing to the prediction model.

Conclusion

The construction of SCAS risk models based on the Chinese T2DM population contributes to its early prevention and intervention, which would reduce the incidence of adverse cardiovascular prognostic events.