Development and validation of a risk calculator predicting exercise-induced ventricular arrhythmia in patients with cardiovascular disease

A Machine Learning Model to Predict Cardiovascular Events during Exercise Evaluation in Patients with Coronary Heart Disease

OBJECTIVE

To develop and optimize a machine learning prediction model for cardiovascular events during exercise evaluation in patients with coronary heart disease (CHD).

METHODS

16,645 cases of cardiopulmonary exercise testing (CPET) conducted in patients with CHD from January 2016 to September 2019 were retrospectively included. Clinical data before testing and data during exercise were collected and analyzed.

RESULTS

Cardiovascular events occurred during 505 CPETs (3.0%). No death was reported. Predictive accuracy of the model was evaluated by area under the curve (AUC). AUCs for the SVM, logistic regression, GBDT and XGBoost were 0.686, 0.778, 0.784, and 0.794 respectively.

CONCLUSIONS

Machine learning methods (especially XGBoost) can effectively predict cardiovascular events during exercise evaluation in CHD patients. Cardiovascular events were associated with age, male, diabetes and duration of diabetes, myocardial infarction history, smoking history, hyperlipidemia history, hypertension history, oxygen uptake, and ventilation efficiency indicators.

Ventricular arrhythmias not meeting criteria for terminating cardiopulmonary exercise testing stratify prognosis and disease severity in heart failure of preserved, midrange, and reduced ejection fraction

Background

Continued high mortality in heart failure patients indicates the need for additional methods of risk stratification and phenotyping.

Hypothesis

We hypothesized that ventricular arrhythmias that do not meet test‐termination criteria (non‐terminating ventricular arrhythmias [NTVA]) during cardiopulmonary exercise testing (CPET) may help in phenotyping disease severity and prognosis in heart failure with reduced (HFrEF) and midrange (HFmrEF)/preserved (HFpEF) left ventricular ejection fraction (LVEF).

Methods

About 319 patients with heart failure (199 HFrEF; 80 HFmrEF; 41 HFpEF) underwent CPET. Tricuspid annular plane systolic excursion (TAPSE) and pulmonary artery systolic pressure (PASP) were measured by echocardiography. B‐type natriuretic peptide (BNP) at rest and peak exercise was also determined. The patients were tracked for primary (cardiac death) and secondary composite outcomes (all‐cause death, heart transplantation/left ventricular assist device implantation, hospitalization for cardiac reasons).

Results

Forty‐seven (15%) of the patients demonstrated NTVA during CPET, regardless of coronary artery disease prevalence. Patients without arrhythmias had a significantly higher LVEF (P < .05), TAPSE/PASP ratio (P < .001), peak oxygen consumption (P < .01), lower resting and peak BNP (P < .001), and the minute ventilation/carbon dioxide production slope (P < .001) compared to those with NTVA. Seventy‐one patients died during the tracking period, 54 for cardiac reasons. NTVA during CPET was a significant predictor of primary and secondary outcomes in the total heart failure cohort (HR: 5.3, 3.7; 95% CI: 3.1‐9.1, 2.4‐5.5; P < .001, respectively), as well as in subgroups categorized according to reduced and middle‐range/preserved LVEF (P < .001).

Conclusion

Exercise‐induced ventricular arrhythmias that do not reach test‐termination criteria are nonetheless indicative of an advanced disease severity phenotype and worse prognosis.