Abstract Background Microvascular injury (MVI) in patients with acute myocardial infarction (AMI) after successful percutaneous coronary intervention (PCI) are independent predictors of poor prognosis, particularly due to microvascular obstruction (MVO) and intramyocardial hemorrhage (IMH). Cardiac magnetic resonance imaging (CMR) is currently regarded as the gold standard for diagnosing MVI, but its clinical application faces limitations. Magnetocardiography (MCG) has shown promising potential in the early diagnosis of coronary artery disease. As a highly sensitive, non-invasive, and radiation-free diagnostic tool, MCG has not yet been reported in terms of its application value for assessing post-MI myocardial perfusion. Purpose This study aimed to develop and validate an MCG-based model for myocardial microvascular perfusion dysfunction and exploratory analysis its clinical value in predicting prognosis in AMI patients post PCI, by comparing and calibrating with CMR. Methods Patients diagnosed with AMI post PCI were recruited from 2023 Feb to 2024 Jun. All patients completed MCG within 3-7 days after AMI, with those who simultaneously underwent CMR constituting the development cohort. According to CMR findings, patients were categorized into those without MVI (MVO-/IMH-), those with MVO but no IMH (MVO+/IMH-), and those with IMH (IMH-). Machine learning algorithms were employed to construct the MCG predictive model, incorporating MCG features, clinical data and laboratory results. During the 90-day follow-up after discharge, all patients underwent at least one proBNP testing. Results In this study, 284 patients were included in the development cohort (123 patients, mean age, 59 years ± 13 SD; 17 13.8% female) and the external validation cohort (161 patients (mean age, 61 years ± 13 SD; 24 14.9% female) who only completed MCG. In the development cohort, MVI occurred in 83 (67%) patients, of whom 30 (24%) had an MVO+/IMH- pattern and 53 (43%) had an IMH+ pattern. According to CMR categories, the MCG-based model accurately predicted MVI in the development cohort, as measured by area under the ROC curve (MVO+/IMH- group AUC; 0.77 95% CI [0.68–0.86, IMH+ group AUC; 0.88 95% CI [0.81–0.94)(Figure 1). The MCG model was effective in identifying individuals at risk of a 10% increase in proBNP levels during follow-up versus at discharge in the IMH+ group. The odds ratio in the IMH+ group was 10.7 (95% CI 1.11–312) compared with the other two groups in the development cohort and 11.2 (95% CI 1.37–249) in the external validation cohort (p0.05), which both showed a significant correlation with ejection fraction (EF). Conclusion We have established an innovative "one-stop" clinical follow-up protocol for MVI assessment post-AMI. This MCG-based model has been applied to assess MVI and predict adverse myocardial remodeling in AMI patients, which serves as a tool for rapid and easy screening in clinical practice.
Shen et al. (Sat,) studied this question.