Value of high-density lipoprotein cholesterol, myocardial perfusion index, and global longitudinal strain derived from cardiac magnetic resonance imaging in predicting coronary slow flow in patients with nonobstructive coronary artery disease.

Abstract

Background: Coronary slow flow (CSF) is associated with dyslipidemias, smoking, and increased body mass index (BMI), yet its diagnosis through noninvasive methods remains challenging. Cardiac magnetic resonance (CMR) is a multimodal imaging technique that enables the simultaneous assessment of impaired myocardial perfusion and deteriorated ventricular function in patients with cardiac disease. This study aimed to demonstrate altered perfusion and deformation parameters on CMR and to evaluate the value of CMR parameters for predicting CSF.

Methods: Participants without obstructive epicardial arterial disease who underwent CMR imaging and coronary angiography (CAG) for typical angina symptoms were enrolled in this retrospective study. CSF was defined by the presence of at least one CAG showing corrected thrombolysis in myocardial infarction frame count (CTFC) >27 frames. The myocardial perfusion index (PI) was analyzed via semiquantitative resting first-pass perfusion. Left ventricular (LV) performance was assessed via CMR feature tracking (CMR-FT) cine imaging, including global longitudinal strain (GLS), global circumferential strain (GCS), and global radial strain (GRS). Baseline clinical factors were collected, including sex, age, and traditional cardiovascular risk factors, along with levels of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), and serum creatinine. Multivariate logistic regression analysis was performed to identify independent predictors of CSF, and a combined prediction model for CSF was developed. The predictive accuracy of the parameters was evaluated via receiver operating characteristic (ROC) curves.

Results: A total of 146 participants who underwent CAG and CMR were included and divided into CSF (n=73; 78.1% male; age 49.44±9.59 years) and control (n=73; 57.5% male; age 47.32±13.57 years) groups based on CTFC. Patients with CSF were more likely to have a higher BMI, hyperuricemia, peripheral arterial disease, and a smoking habit, as well as lower HDL-C levels and elevated TGs as compared to controls. Compared with controls, patients with CSF had impaired GLS (-12.09%±2.69% vs. -14.38%±2.36%) and GCS (-18.70%±3.24% vs. -19.80%±2.21%) (all P values <0.05). Global LV PI was significantly decreased in patients with CSF as compared with controls (11.34%±4.24% vs. 15.25%±8.50%; P<0.001). After adjustments were made for clinical factors and imaging indices, multivariate analysis indicated that the independent predictors of CSF were HDL-C [odds ratio (OR) 0.119; 95% confidence interval (CI): 0.016-0.897; P=0.039], GLS (OR 1.339; 95% CI: 1.112-1.613; P=0.002), and global LV PI (OR 0.456; 95% CI: 0.209-0.994; P=0.048). Moreover, in predicting CSF, the combination of PI, GLS, and HDL-C yielded the best area under the curve (with an 84.9% sensitivity and a 60.3% specificity) as compared to PI (0.783 vs. 0.616; P<0.001), GLS (0.783 vs. 0.742; P=0.130), and HDL-C (0.783 vs. 0.654; P=0.003), respectively.

Conclusions: Reduced HDL-C, decreased PI, and GLS derived from CMR may serve as predictors of CSF. Further multicenter, randomized controlled trials with larger sample sizes are needed to validate these findings.