The 2016 ASE/EACVI recommendations may be able to more accurately identify patients at risk for diastolic dysfunction in living donor liver transplantation

Background The aim of this study was to compare the prevalence of diastolic dysfunction between the 2016 American Society of Echocardiography (ASE)/European Association of Cardiovascular Imaging and 2009 ASE/European Association of Echocardiography recommendations in patients undergoing living-donor liver transplantation (LDLT). Patients and methods A total of 312 adult patients who underwent LDLT at our hospital from January 2010 to December 2017 were retrospectively analyzed. Exclusion criteria were systolic dysfunction, arrhythmia, myocardial ischemia, and mitral or aortic valvular insufficiency. Results The study population was largely male (68.3%), and the median age was 54 (49–59) years. The median model for end-stage liver disease score was 12 (6–22) points. A predominant difference in the prevalence rates of diastolic dysfunction was observed between the two recommendations. The prevalence rates of diastolic dysfunction and indeterminate diastolic function were lower according to the 2016 recommendations than the 2009 recommendations. The level of concordance between the two recommendations was poor. The proportion of patients with a high brain natriuretic peptide level (> 100 pg/mL) decreased significantly during surgery in the normal and indeterminate groups according to the 2009 recommendations; however, only the normal group showed an intraoperative decrease in the proportion according to the 2016 recommendations. Patients with diastolic dysfunction showed a poorer overall-survival rate than those with normal function according to both recommendations. However, there was a difference in the survival rate in the indeterminate group between the two recommendations. A significant difference in patient survival rate was observed between the dysfunction and indeterminate groups according to the 2009 recommendations; however, the difference was not significant in the 2016 recommendations. Conclusions The 2016 classification may be better able to identify patients with a risk for diastolic dysfunction. Particularly, patients in the 2016 indeterminate group seemed to require a cardiac diastolic functional evaluation more frequently during and after surgery than those in the 2009 indeterminate group.


Results
The study population was largely male (68.3%), and the median age was 54 (49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59) years. The median model for end-stage liver disease score was 12 (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) points. A predominant difference in the prevalence rates of diastolic dysfunction was observed between the two recommendations. The prevalence rates of diastolic dysfunction and indeterminate diastolic function were lower according to the 2016 recommendations than the 2009 recommendations. The level of concordance between the two recommendations was poor. The proportion of patients with a high brain natriuretic peptide level (> 100 pg/mL) decreased significantly during surgery in the normal and indeterminate groups according to the 2009 recommendations; however, only the normal group showed an intraoperative decrease in the proportion according to the 2016 recommendations. Patients with diastolic dysfunction showed a poorer overall-survival rate than those with normal function according to both recommendations. However, there was a difference in the survival rate in the indeterminate PLOS  Introduction Diastolic dysfunction is a major component of cirrhotic cardiomyopathy and more frequently occurs than systolic dysfunction in patients with cirrhotic cardiomyopathy [1]. Diastolic dysfunction frequently leads to the development of heart failure and an increased risk for mortality [2,3]. Even in patients with mild diastolic dysfunction and a preserved ejection fraction (EF), there is an increased risk for cardiovascular events after surgery [4,5]. Because of peripheral vasodilation in patients with end-stage liver disease (ESLD), independent of etiology, latent cardiac dysfunction is masked at rest. An impairment of systolic or diastolic cardiac response is frequently present when a patient is stressed during and after surgery. As many as half of cirrhotic patients showed signs of diastolic dysfunction within the first week after liver transplantation (LT) [6,7]. More than 70% of patients who undergo LT suffer from one or more complications related to the heart after surgery [8]. Cardiac imaging has played an integral role in the assessment of LT candidates, and echocardiography, such as transthoracic (TTE) and/or transesophageal (TEE) echocardiography, has primarily been used in all liver transplant candidates to assess chamber size, hypertrophy, systolic and diastolic function, valvular function, and a left ventricular outflow tract obstruction [9,10]. However, no comprehensive study has fully performed cardiac diastolic assessments of patients scheduled for LT because the classification of diastolic function from a multiplicity of echocardiographic indices is difficult and there is a difference in hemodynamic condition between healthy subjects and patients with ESLD [11,12]. Particularly in South Korea, living donor liver transplantation (LDLT) has been more frequently performed than deceased donor liver transplantation [13]. Preoperative assessments of patients scheduled for LDLT are complex and comprehensive, and a tolerable cardiac condition is an important cornerstone related to improved outcomes [14]. Thus, there is a need to have better tools, including echocardiography, to identify patients at increased risk for diastolic dysfunction in LDLT.
Among cardiac biomarkers, brain natriuretic peptide (BNP) is mainly produced by elevated atrial or ventricular wall stretch and has been a promising factor to measure cardiac dysfunction. Accuracy in the risk stratification for heart failure is excellent with a cut-off value > 100 pg/mL [15]. Serum levels of BNP are prominently related to diastolic dysfunction determined by echocardiography in patients with a heart failure preserved ejection fraction (HFpEF) [16]. Compared with atrial natriuretic peptide, serum levels of BNP are more sensitive to identify the cardiac pathological progress, including heart structure and function [17]. In patients with ESLD, BNP plays a supportive role diagnosing cirrhotic cardiomyopathy, and a higher BNP level is strongly related to cardiac systolic or diastolic dysfunction and a poor survival rate [18,19].

Patient management
According to our hospital LDLT protocol that was described previously in detail [24,25], the piggyback surgical procedure was applied using the right lobe from the living donor that was larger than 40% of the recipient's standard liver volume or 0.8% of the recipient's body weight [26]. After completing the hepatic vascular anastomosis, intact hepatic circulation was demonstrated using Doppler ultrasonography. Balanced anesthesia care was meticulously performed under multiple vital monitoring with serial laboratory assessments. A triple-drug regimen (i.e., tacrolimus, mycophenolate mofetil, and prednisolone) with basiliximab (i.e., interleukin-2 receptor antagonist) was administered for immunosuppression during the perioperative period and gradually tapered after surgery.

Echocardiographic evaluation
Cardiac function was carefully measured preoperatively to prevent foreshortening the atrium and ventricle and averaged over three serial cardiac cycles using two transthoracic echocardiographic devices (GE Healthcare, Vivid E9, Milwaukee, WI, USA), (PHILIPS Healthcare, iE33, Durham, NC, USA) by experienced cardiologists. Left ventricular (LV) EF was calculated using the biplane method (i.e., modified Simpson's rule) on apical four-and two-chambered views. Lateral or septal mitral annulus velocity (i.e., e' wave) was evaluated using tissue Doppler imaging for LV diastolic function measured on apical views. Mitral inflow velocities (i.e., E and A waves) and deceleration time were measured using pulse-wave Doppler imaging. Peak tricuspid regurgitation (TR) velocity was measured using continuous-wave Doppler imaging. Left atrial (LA) volume was measured using the biplane method (modified Simpson's rule) on apical views and indexed to body surface area (BSA) (i.e., LV volume/BSA = left atrial volume index [LAVI]). LV hypertrophy type was classified by the LV mass index and relative wall thickness [27,28].

Classification of diastolic function
The prevalence of diastolic function was evaluated according to the 2016 and 2009 recommendations [20,21]. According to the four diastolic parameters (i.e., average E/e' > 14; septal e' velocity < 7 cm/s or lateral e' velocity < 10 cm/s; peak TR velocity > 2.8 m/s; LAVI > 34 mL/ m 2 ) in the 2016 recommendations, the patients were classified into three groups, including the normal diastolic function group (i.e., normal group), indeterminate function group (i.e., indeterminate group), and the diastolic dysfunction group (i.e., dysfunction group). Normal diastolic function was determined as the available diastolic parameters < 50%; diastolic dysfunction was determined as the available diastolic parameters > 50%; and indeterminate function was determined as the available diastolic parameters of 50%. In the 2009 recommendations, septal e' velocity < 8 cm/s or lateral e' velocity < 10 cm/s and LAVI � 34 mL/m 2 were used to evaluate diastolic function. The normal group was defined as no available diastolic parameters, and the diastolic group was defined as all available diastolic parameters. When it was not possible to determine diastolic function because of a discrepancy in diastolic parameters, the subjects were placed in the indeterminate group.

BNP measurement
Intraoperative BNP levels were evaluated three times, such as in the preanhepatic phase (i.e., immediately after inducing anesthesia); in the anhepatic phase (i.e., starting at the portal venous anastomosis); and in the neohepatic phase (i.e., starting at peritoneal closure) [24,29]. BNP levels were investigated via enzyme-linked immunosorbent assay using a Siemens ADVIA Centaur (Siemens Healthcare Diagnostics, Erlangen, Germany). The analytical process was a fully automated two-site sandwich immunoassay using direct chemiluminescent technology. The detection range was 2-5,000 pg/mL according to the manufacturer. BNP levels were classified into high vs. low based on a cut-off value of 100 pg/mL [15].

Statistical analysis
The distribution of continuous data was assessed using the Shapiro-Wilk test. The values are expressed as medians (interquartile range) and numbers (proportion). The perioperative data were compared between the normal, indeterminate, and dysfunction groups using the Kruskal-Wallis test with the Mann-Whitney U-test as a post-hoc test. The categorical data were evaluated using χ 2 or Fisher's exact tests, as appropriate. The test for trends was conducted using a linear-by-linear association method. Concordance in the prevalence of diastolic dysfunction according to the 2016 ASE/EACVI and 2009 ASE/EAE recommendations was evaluated using a Cohen's kappa coefficient with 95% confidence intervals (CIs  The prevalence of these diastolic parameters increased from the normal group to the indeterminate and dysfunction groups. The mitral E/A ratio was lower in the indeterminate group and higher in the dysfunction group than in the normal group. Deceleration time was shorter in the dysfunction group than in the normal group. LVEF was comparable between the three groups. The proportion of abnormal chamber geometry (i.e., eccentric hypertrophy, concentric remodeling and hypertrophy) was higher in the indeterminate and dysfunction groups than in the normal group.

Comparison of preoperative MELD score according to diastology in the 2016 ASE/EACVI and the 2009 ASE/EAE recommendations
Higher MELD score was moderately correlated with the degree of diastolic function in the 2016 recommendations, but the correlation between MELD score and diastolic function in the 2009 recommendation was weak (S1 Table). In both recommendations, the proportion of patients with a high MELD score (> 16 points) increased significantly in accordance with degree of diastolic function (S2 Table). Table 4 shows the intraoperative serial changes in the proportion of patients with a high serum levels of BNP (>100 pg/mL) at the preanhepatic, anhepatic, and neohepatic phases in each group, and differences in the proportion of high BNP level at each phase among the three groups. According to the 2016 recommendations, the proportion of patients with a high BNP level decreased significantly from the preanhepatic phase to the anhepatic and neohepatic phase in the normal group, but not in either the indeterminate or dysfunction groups. According to the 2009 recommendations, the proportion of patients with a high BNP level decreased significantly through the surgical phases in both the normal and indeterminate groups, but not in the dysfunction group. According to the 2016 recommendations, the proportion of patients with a high BNP level at the neohepatic phase was higher in the indeterminate and dysfunction groups than in the normal group; however, the proportion was only higher in the dysfunction group than in the normal group according to the 2009 recommendations.

Comparison of postoperative outcomes between the 2016 ASE/EACVI and 2009 ASE/EAE recommendations
Patients with diastolic dysfunction (2016 recommendations) remained in the ICU longer and had a higher incidence of overt HFrEF than those with normal diastolic function ( Table 5). The proportion of patients using mechanical ventilation was higher in the 2016 diastolic dysfunction group than in the normal group, and the proportion of patients who underwent CRRT was higher in the 2016 indeterminate and dysfunction groups than in the normal group. Patients with diastolic dysfunction (2009 recommendations) also suffered from more frequent development of the HFrEF than those with normal diastolic function. Five patients developed overt HFrEF during the follow-up period. In the 2016 recommendations, four patients in the dysfunction group (33.0%) and one patient in the indeterminate group (2.5%) experienced HFrEF. In the 2009 recommendations, three patients in the dysfunction group (5.9%) and two patients in the indeterminate group (1.3%) experienced HFrEF. However, other outcomes (i.e., total hospital stay, and the development of EAD and AKI) were comparable among the three groups according to the 2019 and 2009 recommendations. According to the 2016 recommendations, the survival rate was better in the normal group than in the dysfunction group; however, no differences in survival rates were observed in the indeterminate group compared with the normal and dysfunction groups (Fig 2). According to the 2009 recommendations, the survival rate was better in the normal and indeterminate groups than in the dysfunction group; however, no difference in survival rate was detected   According to the 2016 recommendations, overall patient survival was significantly different between the normal and dysfunction groups (p = 0.007) but did not differ between the normal and indeterminate groups (p = 0.183) or between the indeterminate and dysfunction groups (p = 0.223). The 1-, 3-and 5-year survival rates were 98%, 93%, and 91% in the normal group; 87%, 70%, and 70% in the indeterminate group; and 64%, 54%, and 54% in the dysfunction group, respectively. (B) According to the 2009 recommendations, overall patient survival was significantly different between the normal and dysfunction groups (p = 0.042) and between the indeterminate and dysfunction groups (p = 0.027), but not between the normal and indeterminate groups (p = 0.836). The 1-, 3-, and 5-year survival rates were 94%, 92%, and 83% in the normal group; 95%, 91%, and 87% in the indeterminate group; and 83%, 74%, and 62% in the dysfunction group, respectively.

Discussion
The main findings of this study were that there was a predominant difference in the prevalence rate of diastolic dysfunction between The 2016 ASE/EACVI recommendations were proposed to simply identify the patients at risk of diastolic dysfunction using TTE findings in a daily clinical setting [20]. However, in previous studies based on general community populations, poor concordance rates of diastolic function between the 2016 and 2009 recommendations were reported that lowered the prevalence of diastolic dysfunction (i.e., only 1.3% by Huttin et al. and 1.4% by Almeida et al.) in the 2016 recommendations than the prevalence determined by the 2009 recommendations [22,23]. This finding seemed to be largely shared with our result of lower prevalence of diastolic dysfunction in the 2016 recommendations (i.e., 3.8%) than in the 2009 recommendations (i.e., 16.3%). The potential reason for this discrepancy may be related to the inclusion of a new diastolic parameter, such as a TR velocity > 2.8 m/s. TR velocity is a marker of acute or chronic pressure overload, because pulmonary systolic hypertension, derived from the TR velocity, is closely related to overt pulmonary hypertension in patients with diastolic dysfunction [36]. TR velocity may eventually be deemed to play an important role as it represents a more aggravated degree of diastolic function. In our study (2016 recommendations), the total prevalence of patients with a TR velocity > 2.8 m/s was 8.7%, and the proportion for TR velocity > 2.8 m/s was significantly higher in the indeterminate and dysfunction groups than in the normal group. Including TR velocity may have affected the increase in specificity to diagnose diastolic dysfunction in the 2016 recommendations than in the 2009 recommendations. Compared with the previous general population studies [22,23], our prevalence of diastolic dysfunction appeared to be higher, possibly because our study populations suffered from ESLD, which caused pathophysiological changes in cardiac function or geometry [37]. In line with previous work, our study found that a higher MELD score, which represents the severity of hepatic decompensation, was correlated with the degree of diastolic function in both the 2016 and 2009 recommendations [1,6].
BNP is derived from the ventricular wall due to hemodynamic volume or pressure stress and is significantly related to the severity of left ventricular function. Furthermore, a close correlation between serum levels of BNP and diastolic parameters on echocardiography, such as tissue Doppler imaging, is present so that high serum levels of BNP (> 100 pg/mL) are reflected by increased left ventricular filling pressure [38,39]. In previous LT studies, high serum levels of BNP were significantly associated with poor postoperative outcomes, such as graft dysfunction, kidney injury, and patient survival [24,29,40]. In our study, patients with normal diastolic function experienced a significant decrease in the proportion of high BNP level (> 100 pg/mL) through the surgical phases; however, this finding was not evident in the patients with diastolic dysfunction according to both recommendations. Interestingly, there were differences in intraoperative changes in the proportion of patients with a high BNP level in the indeterminate group between the 2016 and 2009 recommendations. Namely, the indeterminate group (2009 recommendations) showed a decrease in the proportion of patients with a high BNP through the surgery, and the proportion of patients with a high BNP level was eventually comparable between the indeterminate and normal groups at the neohepatic phase. However, in the 2016 recommendations, the indeterminate group did not experience a decrease in the proportion of patients with a high BNP level; and the proportion of patients with a high BNP level was higher in the indeterminate group than in the normal group at each surgical phase. These findings potentially explain why only 14 patients (9.0%) in the indeterminate group (2009 recommendations) remained in the indeterminate group (2016 recommendations), and almost all patients (n = 140; 90.3%) in the indeterminate group (2009 recommendations) were reclassified into the normal group (2016 recommendations). One patient in the indeterminate group (2009 recommendations) was reclassified into the dysfunction group (2016 recommendations). Therefore, our results suggest that patients in the indeterminate group (2016 recommendations) may have a higher risk of developing impaired diastolic function than normal function during LDLT compared with the 2009 recommendations. Additionally, patients with indeterminate diastolic findings, defined as 50% of the 2016 diastolic parameters (i.e., average E/e' > 14, septal e' velocity < 7 cm/s or lateral e' velocity < 10 cm/s, TR velocity > 2.8 m/s, and LAVI > 34 mL/ m 2 ), may require meticulously and continuously monitored cardiac status using ECG or laboratory factors during LDLT. Cardiovascular complications are a major cause of mortality in patients undergoing LT, including graft rejection and infection that relates to approximately 20% of post-transplant deaths [41]. More than 70% of patients who undergo LT suffer from cardiovascular complications, and approximately 7% of these patients are aggravated to severe heart failure [8]. Preoperative diastolic dysfunction is significantly associated with an increased risk of developing perioperative heart failure and higher rates of 1-year mortality [42,43]. However, successful LT has a positive effect on cardiac functional recovery after LT, as diastolic dysfunction gradually improves during the first year after surgery, together with the cardiac response to stressful stimuli [44][45][46]. Our study shared previous findings [42,43,47] that patients with diastolic dysfunction (both 2016 and 2009 recommendations) developed an aggravation of cardiac function to overt heart failure and poor overall patient survival than those with normal diastolic function. However, there was a difference in overall patient survival in the indeterminate group between the recommendations. Unlike the 2009 recommendations, there was no difference in the overall survival rate between the dysfunction and indeterminate groups in the 2016 recommendations. The potential explanation is that inclusion of TR velocity > 2.8 m/s played a role to more clearly stratify the mortality risk after surgery. A study by Bushyhead et al. [48] suggested that TR (more than a mild degree) is significantly associated with worse patient survival after LT, and another study by Kia et al. [49] showed that among various echocardiographic variables, only TR (more than a mild degree) plays a predictive role in patient and graft survival. Ford et al. [50] suggested that the cut-off level of TR velocity on 1-year LT mortality was 3.0 m/s, and that backward-pressure, derived from increased and prolonged TR, may cause persistent graft edema related to graft failure and long-term complications. Therefore, TR may be an indicator of hemodynamic load that becomes aggravated when patients are in precarious conditions, such as sepsis [51].
Some limitations of our study should be discussed. First, our study included only patients with a preserved EF (> 50%); therefore, our findings do not apply to patients with a reduced EF. Particularly, because of peripheral vasodilatation, cardiac systolic dysfunction in patients with ESLD is latent at rest [52]. Further study is required to accurately measure systolic dysfunction when patients are challenged during LT. Second, the patients did not routinely undergo post-transplant cardiac function testing using TTE, which was only performed based on the decision of the attending intensive care physician. Eventually, because of the possibility of an under-estimate of the incidence of overt heart failure after surgery, we were unable to investigate patient survival rate related to heart-specific complications, and evaluated all-cause patient survival rate, including graft rejection and infection. A prospective study for heart originating deaths would further clarify the differences between the 2016 and 2009 recommendations. Third, the gold standard to diagnose diastolic dysfunction has not been fully established in clinical settings. Particularly, advanced liver disease causes pathophysiological changes in hemodynamic circulation, such as splanchnic vasodilatation [6]. The diagnostic accuracy of diastolic dysfunction in both recommendations has not been fully demonstrated in patients who underwent LDLT. Finally, because of the small number of patients with diastolic dysfunction in the 2016 recommendations, we were unable to evaluate intraoperative changes in BNP and postoperative outcomes according to the degree of diastolic dysfunction. A further large cohort study is required to clarify the effect of diastolic severity on prognosis in patients who underwent LDLT.

Conclusions
Clinical application of diastology according to the 2016 ASE/EACVI recommendations in patients who underwent LDLT resulted in a lower prevalence of indeterminate function and overt diastolic dysfunction, and a higher prevalence of normal diastolic function. Therefore, the concordance between previous and current recommendations was poor, which was caused by reclassification of the diastolic functional evaluation. The current 2016 classification may be able to more clearly identify patients at risk for diastolic dysfunction that may result from inclusion of TR velocity > 2.8 m/s. This finding was supported by the intraoperative BNP level, as a diastolic marker. Particularly, patients in the 2016 indeterminate group seem to require an evaluation of cardiac diastolic function more frequently than those in the 2009 indeterminate group, during and after surgery, because the 2016 indeterminate group included ill patients who most likely should be evaluated and treated like the diastolic dysfunction group. Additionally, this finding is not limited to patients undergoing LDLT, but also can be addressed in patients undergoing deceased donor LT [53,54]. The prognostic impact of patients with indeterminate diastolic function and overt diastolic dysfunction needs further investigation in patients who undergo LDLT.
Supporting information S1