Percentage of Low-Attenuation Epicardial Adipose Volume (LAAV%) as an Independent Predictor of Coronary Microvascular Dysfunction: A Retrospective Cohort Study

Figures

• 

  Fig. 1.. Flowchart of patient enrollment, exclusion, and study design. After applying the predefined exclusion criteria to the initial 178 patients with suspected INOCA, a final cohort of 155 patients (with one target vessel analyzed per patient) was included for comprehensive CCTA and invasive physiological assessment. Abbreviations: INOCA, ischemia and non-obstructive coronary arteries; CCTA, coronary computed tomography angiography; EAT, epicardial adipose tissue; LAAV, low-attenuation adipose volume; FFR, fractional flow reserve; IMR, index of microcirculatory resistance; CFR, coronary flow reserve; CMD, coronary microvascular dysfunction; HU, Hounsfield units; LAAD, low-attenuation adipose density; CAD, coronary artery disease.

• 

  Fig. 2.. Illustrative CCTA slices for quantifying EAT volumes and densities. (A) EFV segmented from –190 to –15 HU. (B) LAAV from –190 to –75 HU. (C) MAAV from –75 to –45 HU. (D) HAAV from –45 to –15 HU. Abbreviations: CCTA, coronary computed tomography angiography; EAT, epicardial adipose tissue; HU, Hounsfield units; EFV, epicardial adipose volume; LAAV, low-attenuation adipose volume; MAAV, medium-attenuation adipose volume; HAAV, high-attenuation adipose volume.

• 

  Fig. 3.. Distribution of EAT characteristics and clinical parameters between groups. Comparison of (A) LAAV%, (B) age, (C) BMI, (D) HbA1c, (E) triglycerides, and (F) smoking history rate between the CMD and non-CMD groups. Box and violin plots show median, interquartile range, and distribution density; bar chart shows proportion. Abbreviations: BMI, body mass index; CMD, coronary microvascular dysfunction; EAT, epicardial adipose tissue; HbA1c, glycated hemoglobin; LAAV%, percentage of low-attenuation adipose volume; TG, triglycerides.

• 

  Fig. 4.. Correlation matrix of EAT metrics and clinical parameters. Spearman’s rank correlation (ρ) showing associations between CCTA-derived EAT phenotypes and clinical/laboratory parameters. Red and blue indicate positive and negative correlations, respectively. *p < 0.05, **p < 0.01, ***p < 0.001. Abbreviations: BMI, body mass index; CRP, C-reactive protein; EAT, epicardial adipose tissue; EFV, epicardial adipose volume; FBG, fasting blood glucose; FIB, fibrinogen; HbA1c, glycated hemoglobin; LAAV, low-attenuation adipose volume; LAAV%, percentage of LAAV; MB, myocardial bridging; TG, triglycerides.

• 

  Fig. 5.. Forest plot of independent risk factors for CMD. The plot presents the adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs) for predictors identified in the parsimonious multivariable logistic regression model. Notably, the OR for LAAV% is standardized per standard deviation (SD) increase to allow for direct comparison of effect sizes among predictors, while clinical factors (e.g., BMI and smoking history) are reported per unit increase. The vertical dashed line at OR = 1 indicates the null reference value. Red diamonds represent point estimates with OR >1, while blue diamonds represent point estimates with OR <1. LAAV, low-attenuation adipose volume; BMI, body mass index; CMD, coronary microvascular dysfunction; HbA1c, glycated hemoglobin.

• 

  Fig. 6.. ROC analysis for predicting CMD: comparative diagnostic performance of incremental predictive models. The ROC curves illustrate the accuracy of three models in identifying coronary microvascular dysfunction (CMD). Model 1 includes LAAV% (AUC = 0.710). Model 2 includes clinical risk factors (AUC = 0.756). Model 3 incorporates both (AUC = 0.810, p < 0.05 vs. Model 1 and Model 2). These results identify LAAV% as a superior, independent imaging biomarker for CMD. Abbreviations: AUC, area under the curve; CMD, coronary microvascular dysfunction; EAT, epicardial adipose tissue; LAAV%, percentage of low-attenuation adipose volume.

• 

  Fig. 7.. RCS analysis reveals a non-linear relationship between LAAV% and BMI with the risk of CMD. Restricted cubic spline (RCS) curves illustrate the predicted probability of CMD according to (A) LAAV% and (B) BMI. For LAAV%, a distinct J-shaped association is observed, with a biological ‘tipping point’ is identified at 41.7% (calculated via RCS model), beyond which the risk of CMD increases exponentially. For BMI, the RCS curve shows the dose-response relationship with CMD risk. (See Fig. 6 for corresponding diagnostic performance). Abbreviations: CMD, coronary microvascular dysfunction; LAAV, low-attenuation adipose volume; BMI, body mass index.