Predicting Coronary Artery Lesion Severity using Pulse Wave Harmonics: A SYNTAX Score-based Study

This study aimed to investigate the correlation between the differences in pulse wave harmonic indices between the left and right hands and the SYNTAX score and to explore the potential of pulse wave harmonics in predicting the degree of coronary artery lesions.

The arterial pressure wave signals from both hands of the patients scheduled for coronary angiography were recorded using photoplethysmography. According to the "visceral resonance theory", taking integer multiples of the heartbeat from 0 to 11 as the resonance frequencies, the collected arterial pressure waves were decomposed into the 0th to 11th harmonics via the Fourier transform method. The harmonic characteristics were quantified by amplitude (Cn), phase (Pn), and energy (Dn) (n is the harmonic serial number), and the coefficient of variation of the indices was calculated and suffixed as CV. The difference between the measured values of the left- and right-hand parameters of the same patient was calculated (ΔCn, ΔPn, ΔDn, ΔCnCV, ΔPnCV), and the absolute value of the difference was obtained (|ΔCn|, |ΔPn|, |ΔDn|, |ΔCnCV|, |ΔPnCV|). Based on the coronary angiography imaging data, SYNTAX scores were computed for all participants, who were stratified by gender into male and female cohorts. For each group, logistic regression models were established with SYNTAX scoreΔ22 as the dependent variable, and harmonic index differences as the independent variables. To determine the best prediction model, the Akaike Information Criterion (AIC) was used and model with the lowest score was selected. Finally, the discriminant ability of the prediction model was evaluated using the ROC curve analysis and the Bootstrap internal validation method.

A total of 348 patients were included, with 249 males and 99 females. In the male group, the discriminant model was based on |ΔC10|, ΔD6, |ΔD9|, |ΔD10|, |ΔP8|, |ΔP10|, ΔP1CV, and ΔC9CV, with the minimum AIC value of 105.47, the area under the ROC curve (AUC) of 0.89, and the average AUC of 0.85 in the Bootstrap internal validation. In the female group, the discriminant model was based on |ΔD2|, |ΔD3|, |ΔD5|, |ΔD6|, |ΔD9|, |ΔC2CV|, |ΔC4CV|, |ΔC5CV|, | ΔC6CV|, and |ΔC9CV|, with the minimum AIC value of 59.34. The AUC of the ROC curve of this prediction model was 0.92, and the average AUC in the Bootstrap internal validation was 0.84.

In this study, the degree of coronary artery occlusion was evaluated through a noninvasive method combined with the SYNTAX score, providing a valuable noninvasive tool for clinical evaluation of CAD. This detection method is easy to operate, has high repeatability, and the equipment is small in size, making it suitable for various environments, it can be operated independently by the patients. Yet, the current study, being cross-sectional, only found an association rather than a causal relationship, calling for future prospective studies to clarify the causal link.

The different characteristics of pulse wave harmonics between the left and right hands can effectively reflect the degree of coronary artery lesions. Through the analysis of pulse wave harmonics, a diagnostic model with good discriminant ability for predicting the degree of coronary artery lesions can be constructed, which may offer a valuable non-invasive tool for the clinical assessment of CAD.