Association of epicardial adipose tissue with proteomics, coronary flow reserve, cardiac structure and function, and quality of life in heart failure with preserved ejection fraction: insights from the PROMIS‐HFpEF study

Figures

• 

  Figure 1. Correlation coefficients for the predictive value (x‐axis) and the orthogonal variation (y‐axis) of all variables assessed in PROMIS‐HFpEF. Significant predictors are indicated with a large symbol. OLINK variables (circles) are highlighted together other clinical, laboratory or echocardiographic discriminant variables (triangles). In addition, the following variables are included for comparison: E/e', left ventricular mass index, deceleration time, left atrial reservoir strain and diabetes. Creatinine–cystatin C estimated glomerular filtration rate (Creat‐CysC eGFR) is highlighted as a potential explanation for the orthogonal variation of the OLINK variables loading with high orthogonal correlation coefficients in the model.

• 

  Figure 2. Scores of subjects based on the variable pattern shown in Figure 1 along the discriminant component (x‐axis) and the orthogonal component (y‐axis). Symbols are coloured by class (epicardial adipose tissue ≥9 mm in green, <9 mm in blue) and sized proportionally by individual epicardial adipose tissue (mm). The subject highlighted with an closed arrow is thus predicted to have the most extreme phenotype, whereas the subject marked with an open arrow is predicted to have a phenotype not different from the group of subjects having epicardial adipose tissue <9 mm.

• 

  Figure 3. Top 40 variables associated with epicardial adipose tissue (EAT) ≥9 mm in OPLS‐DA separating subjects with EAT ≥9 versus <9 mm. The selected bars (top 20 in each direction) represent the variables with the highest numerical loading scores (p [1]) along the discriminant component, meaning variables with the best predictive value. Red bars and list on the left are variables that are higher, and blue bars and list on the right are variables that are lower in subjects with EAT ≥9 mm. The lists are sorted by the numerical loading score, meaning that body mass index and insulin‐like growth factor‐binding protein 1 are the top positive and negative predictor, respectively. CFR, coronary flow reserve; LDL, low‐density lipoprotein.