Supplementary Material for: Magnetocardiography for Distinguishing Obstructive Coronary Heart Disease and Coronary Microvascular Dysfunction: A Feasibility Study

‌Objective‌ To evaluate the capability of magnetocardiography (MCG) in distinguishing obstructive coronary heart disease (OCHD) from coronary microvascular dysfunction (CMVD). Methods Patients with chest pain suspected of having CAD were consecutively enrolled in according to the inclusion and exclusion criteria, All patients underwent coronary angiography and coronary blood flow analysis, and were divided into OCHD group (n=447) and CMVD group (n=332) based on the results of imaging analysis. All patients also underwent MCG examination before or after coronarg angiology in 48 hours, the magnetocardiography (MCG) parameters were analyzed by an independent third party. Univariate and multivariate logistic regression analyses were performed to identify significant clinical and MCG parameters between groups. Diagnostic models were constructed using selected MCG parameters alone or in combination with clinical indicators. Performance was assessed using receiver operating characteristic (ROC) curves, accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Clinical utility and model performance were further validated via calibration curves and decision curve analysis. ‌Results In OCHD group, the prevalence of type 2 diabetes mellitus (T2DM) (35.79% vs. 25.30%, p=0.022) and interventricular septal thickness (mm) [11.00 (10.25-12.00) vs. 11.00 (10.00-11.00), p<0.001] were significantly higher than those in the CMVD group, while high-density lipoprotein cholesterol (HDL-C) (mmol/L) was significantly lower [1.01 (0.85-1.16) vs. 1.06 (0.93-1.22), p<0.001]. Seven MCG parameters showed significant intergroup differences: QR-enav[10.08 (9.02-11.22) vs 9.36 (8.38-10.41), p<0.001], RS-enav[10.29 (9.32-11.33) vs 9.66 (8.71-10.73), p<0.001], TT-dicar[0.04 (0.02-0.14) vs 0.02 (0.01-0.09), p<0.001], TT-dicd[41.40 (25.75-77.55) vs 34.40 (20.83-62.90), p<0.001], TT-didav[60.70 (55.50-66.40) vs 63.80 (58.02-67.80), p<0.001], TT-didmse[3.70 (2.10-6.35) vs 2.90 (1.70-4.77), p<0.001], and TT-enav[10.44 (9.20-11.57) vs 9.96 (8.79-11.01), p<0.001]. RS-enav achieved the highest AUC (0.615), with sensitivity, specificity, PPV, and NPV of 48.77%, 69.21%, 70.10%, and 47.72%, respectively. TT-didmse exhibited specificity of 73.51% and PPV of 69.41%, while TT-enav showed specificity of 60.26% and PPV of 67.21%. TT-dicd demonstrated the highest diagnostic sensitivity (78.30%) in distinguishing OCHD patients from CMVD. The integrated model combining seven MCG parameters yielded an AUC of 0.681 (95% CI: 0.642–0.720), which improved to 0.713 (95% CI: 0.674–0.751) in the combined model. Calibration curves confirmed high agreement between nomogram predictions and observed outcomes, and decision curve analysis indicated better net benefit for the combined model. ‌Conclusion‌ MCG parameters, either alone or in combination with clinical indicators, demonstrate feasibility in distinguishing OCHD from CMVD. As a novel non-invasive diagnostic tool, MCG holds potential value in discriminating between OCHD and CMVD.