### INTRODUCTION

### MATERIALS AND METHODS

_{1c}≥ 6.5% were classified as having DM. Those without any diagnostic test results were classified into the control group. The severity of ESLD was evaluated using the Model for End-stage Liver Disease (MELD) score based on serum bilirubin, creatinine, and international normalized ratio. Patients aged < 18 years, with a history of liver transplant, with concomitant chronic kidney disease, with a history of cardiac surgery or moderate to severe valvular disease, and without a preoperative echocardiogram were excluded. Demographic data, patient status before liver transplant, blood test results, and hemodynamic data were collected from electronic medical records. Transthoracic echocardiography was performed by a skilled technician using a Hewlett-Packard Sonos 2500 or 5500 imaging system (Hewlett-Packard Inc., USA) with a 2.5-MHz transducer, and the measurements were confirmed by a cardiologist. The 2-dimensional variables measured included end-systolic interventricular diameter, end-diastolic interventricular diameter, end-systolic left ventricular posterior wall thickness, end-diastolic left ventricular posterior wall thickness, left ventricular mass index, and left atrial diameter. The Teichholz method or biplane modified Simpson’s rule was used as appropriate to measure the left ventricular end-diastolic volume (EDV) and left ventricular end-systolic volume (ESV), which were used to calculate the SV (= EDV − ESV) and left ventricular ejection fraction (LVEF). The index of each measured value was calculated by dividing each value by the body surface area. In the apical 4-chamber view, pulsed-wave Doppler was used to measure early mitral inflow velocity (E), late mitral inflow velocity (A), deceleration time of the E wave, and E/A ratio. The tissue Doppler image was used to measure the velocities of systolic wave (s’), early diastolic wave (e’), and late diastolic wave (a’) on the luminal side of the septal mitral annulus. The E/e’ ratio, which reflects the left ventricular end-diastolic pressure, was calculated (Fig. 1). For the evaluation of more complicated left ventricular end-systolic function, end-systolic elastance was calculated as ESV / end-systolic pressure. End systolic pressure, a reflection of aortic pressure, was calculated as noninvasive systolic pressure × 0.9. To evaluate vascular resistance, arterial elastance was calculated as end-systolic pressure / SV [11].

*t*-test or Mann-Whitney test, as appropriate. Categorical variables were analyzed using the chi-square test or Fisher’s exact test. To minimize the difference in baseline characteristics of the 2 groups, a 1:2 matched propensity score analysis [12] was performed. The propensity score, which is the probability of each subject to be assigned to the treatment group according to a given covariance, was calculated using a propensity score model via logistic regression analysis with the patient’s age, sex, body mass index, MELD score, hypertension, history of cardiovascular disease, history of beta-blocker use, and btype natriuretic protein, among the characteristics specified in Table 1. Caliper matching was performed in a 1:2 ratio by using this propensity score in the nonrandom package (http://www.rdocumentation.org/packages/nonrandom) of R, and the caliper was set as standard deviation*0.2. The model test for propensity scores was performed using Cstatistics and the Hosmer-Lemeshow test. R (version 3.3.1; R Foundation for Statistical Computing, Austria) was used for statistical analysis, and paired

*t*-test was used to analyze the difference between the 2 matched groups. P < 0.05 was deemed statistically significant.

##### Table 1

### RESULTS

##### Table 2

^{} Values are presented as median (1Q, 3Q), mean ± SD, or number (%). LV: left ventricle, LVMI: left ventricular mass index, E/A: early and late diastolic velocity ratio, E/e’: ratio of early diastolic to tissue doppler imaging velocities, TR: tricuspid regurgitation, RV: right ventricle, PGsys: systolic pressure gradient.

^{2}= 12.113, df = 8, and P = 0.1462. Baseline variables that showed a difference between the 2 groups before matching showed P > 0.05 after matching (Table 3). On echocardiograms, systolic function indices that were significant before matching (LVEF, s’ velocity; both P > 0.05) did not show a significant difference after matching (Fig. 2). However, diastolic function indices including E/A ratio (1.09 [0.87-1.37] vs. 1.02 [0.85-1.28], P = 0.017) and e’ velocity (7.4 [6.4-8.5] cm/s vs. 7.0 [5.9-8.1] cm/s, P < 0.001) were still low in DM-ESLD patients, whereas the E/e’ ratio, a marker of left ventricular end-diastolic pressure, was still high (10.1 ± 3.0 vs.10.9 ± 3.2, P < 0.001) (Table 4, Fig. 3).

##### Table 3

^{} Values are presented as mean ± SD, number (%), or median (1Q, 3Q). Propensity score model discrimination was evaluated with c-statistics (c = 0.662) and calibration was assessed with Hosmer-Lemeshow statistics (X-squared = 12.113, df = 8, P value = 0.1462). SMD: standardized mean difference, INR: international normalized ratio.

##### Table 4

^{} Values are presented as median (1Q, 3Q), mean ± SD, or number (%). LV: left ventricle, LVMI: left ventricular mass index, E/A: early and late diastolic velocity ratio, E/e’: ratio of early diastolic to tissue doppler imaging velocities, TR: tricuspid regurgitation, RV: right ventricle, PGsys: systolic pressure gradient.