*2.3. Echocardiography*

Echocardiographic measurements were performed using Toshiba Xario 100 Diagnostic Ultrasound System (Toshiba, Toshiba Medical System Corporation, Tochigi 324-8550, Japan). M-mode and two-dimensional measurements were performed as recommended by the American Society of Echocardiography [18], including left ventricular end-diastolic and end-systolic diameters, intraventricular septum, and posterior wall end-diastolic thickness. Left ventricular mass (LVM) was calculated according to the Devereux formula [19]. LVM was indexed for BSA (LVMI).

#### *2.4. Brachial Artery Flow-Mediated Dilation*

FMD was measured in the morning, after 10 min of lying in a quiet dimmed room using the Toshiba Xario 100. During the examination, the patients rested in a seated position with their forearms and backs supported. A manual sphygmomanometer cuff was placed on the arm without arteriovenous fistula and the diameter (lumen) of the brachial artery was measured with a linear transducer. The cuff was then inflated at approximately 50 mmHg above the current systolic pressure for 5 min [20]. After the deflation of the cuff, the serial measurements during diastole were recorded and the widest dilation of the brachial artery was used for FMD calculation: FMD% = (A − B)/B × 100%, where A is the diameter of the artery during reactive hyperemia, and B is the initial diameter of the artery.

Non-endothelial dependent vasodilation (NMD) was assessed after at least a 15-min rest from the FMD acquisition. Similarly to FMD measurements, vessel diameters were assessed before and after sublingual nitroglycerin (400 μg) application (Nitromint (glyceroli trinitras), Proterapia, Poland). Of importance, as we observed unexpected reduction of NMD values in some patients, we repeated all NMD measurements using a double dose of nitroglycerin, i.e., 800 μg, after approximately 6 months from the previous follow-up series.

#### *2.5. Central Aortic Pressure and Arterial Sti*ff*ness Measurement*

Pulse waveform analysis was performed with the commercially available SphygmoCor 2000 (AtCor Medical, Sydney, Australia). Peripheral pressure waveforms were recorded from the radial artery using applanation tonometry. After the acquisition of at least 20 sequential waveforms, a validated generalized transfer function was used to generate the corresponding central aortic pressure waveform. Central blood pressure measurements, including central aortic systolic and diastolic blood pressure, central pulse pressure, and augmentation index were performed. Augmentation index was then normalized according to the heart rate (AIx@75). Only high-quality recordings, defined as an in-device quality index greater than 80% and visually acceptable curves by the investigator, were included in the analysis. The entire pulse wave analysis was performed in the sitting position under standardized conditions in the morning hours, after at least 15 min of rest in the supine position.

Arterial sti ffness was also assessed using SphygmoCor 2000 placed over the carotid and femoral arteries. Pressure signals were calibrated using brachial BP and PWV was calculated as the time of the pulse wave between the diagnosed points (distance (m)/time (s)).

#### *2.6. Liver Elastography*

The controlled attenuation parameter (CAP) and liver sti ffness measurement (LSM) were performed using transient elastography with a M-probe (FibroScan 502 Touch, Echosense, Paris, France). The operator was a technician certified by Echosense and unaware of patient status. The measurements were performed using a 3.5-MHz standard probe on the right hepatic lobe through the intercostal spaces with the patient lying in a supine position. As recommended by the manufacturer, 10 successful measurements were performed for each patient and only those with a success rate of at least 60% and an interquartile range/median value of less than 0.3 were considered reliable. The final CAP and liver sti ffness were expressed in dB/m and kPa, respectively [21,22].

#### *2.7. Data and Statistical Analysis*

Statistical analyses were performed using the STATISTICA 13.0 PL for Windows software package (StatSoft Poland, Cracow, Poland). The values were presented as mean values with 95% confidence interval (CI) (for variables with normal distribution), medians with Q25-Q75 quartile values (for variables with not normal distribution), or frequencies. Comparisons between groups were done by using the Student t-test for quantitative variables or the χ2 test for qualitative variables. Variables with not normal distribution were compared using the Mann–Whitney U test. The comparison of baseline and follow-up values was performed using the Student t-test, or the Wilcoxon test for variables with not normal distribution. Correlation coe fficients were calculated using the Pearson test. Due to its not normal distribution, FGF-21 data were logarithmized before the correlation analyses. In all statistical tests, '*p*' values below 0.05 were considered as statistically significant.
