Introduction

Cirrhotic cardiomyopathy (CCM) is defined as a chronic cardiac dysfunction in cirrhosis patients without underlying primary cardiac disease. Left ventricular diastolic dysfunction (LVDD) is an important mechanism that contributes to heart failure with preserved ejection fraction (HFpEF). LVDD is more common manifestation of CCM than LV systolic dysfunction in patients with cirrhosis [1, 2]. CCM may be related to portal hypertension, increased blood volume, high cardiac output, and pulmonary hypertension. CCM can be present in patients with cirrhosis caused by any chronic liver disease; however, some underlying liver diseases—such as nonalcoholic fatty liver disease (NAFLD) may in theory independently increase its risk [3,4,5,6,7,8,9,10]. This likely reflects shared risk factors and etiologic conditions such as obesity, insulin resistance, and inflammation.

CCM typically exists in a latent state but can become unmasked during times of physiologic stress (such as hepatic decompensation or infection) or after transjugular intrahepatic portosystemic shunt (TIPS) placement [11]. Cardiac dysfunction is a recognized complication after TIPS as well as after liver transplant, and risk is likely higher in patients with underlying CCM [11]. Most importantly, CCM may be an independent predictor of mortality both before and after liver transplant [12,13,14]. Therefore, identification of cardiac dysfunction in cirrhosis patients is important.

A diagnosis of CCM is made primarily by transthoracic echocardiography (TTE)—according to criteria defined in 2020 by the Cirrhotic Cardiomyopathy Consortium [1], Fig. 1. These newer criteria were created to reflect advances in echocardiography with implementation of speckle tracking strain imaging and enhancements in tissue Doppler imaging. Despite advances in TTE based imaging, diagnosis of CCM remains a challenge, complicating our understanding of its prevalence and clinical significance. Echocardiography is operator dependent and this, along with various patient characteristics, can contribute to high interobserver variability for some parameters like assessment of LV ejection fraction. On the other hand, assessment of Doppler parameters tends to have lower interobserver variability. Additionally, many of the unique hemodynamic changes that occur in patients with cirrhosis such as reduced systemic vascular resistance and high cardiac output may obscure cardiac pathology since LV ejection fraction may be higher than expected, even with myocardial dysfunction. This is particularly challenging in the early stages of the disease process. Therefore, additional tools are needed to help identify cirrhosis patients at risk of cardiac complications. The H2FPEF score (Heavy, 2 or more Hypertensive drugs, atrial Fibrillation, Pulmonary hypertension, Elderly, and elevated Filling pressure) is a noninvasive predictive model used to estimate the probability of underlying HFpEF [15, 16]. We sought to evaluate clinical determinants of the score in patients with cirrhosis at our institution who had a TTE prior to liver transplant as well as to determine if higher H2FPEF scores are associated with cardiac complications postoperatively.

Fig. 1
figure 1

Consort diagram

Full size image

Methods

Patient Demographics

We retrospectively studied 261 adult patients who had a liver transplant and a pre-transplant cardiac assessment that included a TTE and dobutamine stress test at the Medical University of South Carolina between January 2010 and October 2018. Since E/e′ is required for the diagnosis of left ventricular diastolic dysfunction (LVDD) and for the calculation of the H2FPEF score, only those subjects with this data included in the TTE report were included for analyses (n = 166, Fig. 1). The following pre-transplant variables were collected from the pre-transplant evaluation: demographics, etiology of chronic liver disease, laboratory testing results, calculated model of end stage liver disease (MELD) score, cardiovascular comorbidities (diabetes, hypertension, hyperlipidemia, atrial fibrillation, and coronary artery disease), and cardiac medications. Post-transplant outcomes included new-onset systolic heart failure (defined as an ejection fraction < 50% on post-transplant echocardiography), and death (either due to heart failure or any cause). A cardiac death was defined as death from cardiogenic shock, requiring left ventricular assist device, ventricular arrythmia, or out-of-hospital sudden cardiac death). The study was approved by the Institutional Review Board.

Echocardiography

TTE was performed according to MUSC’s institutional standard protocol. Left ventricular DD was defined according to the 2020 Cirrhotic Cardiomyopathy Consortium, that use four echocardiographic variables (Table 1): annual e′ velocity (septal e′ < 7 cm/s or lateral e′ < 10 cm/s), average E/e′ ratio > 14, left atrial volume index (LAVI) > 34 mL/m2, and tricuspid valve regurgitation velocity > 2.8 m/s). If ≥ 3 variables are abnormal, then LVDD is present.

Table 1 Cirrhotic cardiomyopathy consortium 2019 diagnostic criteria
Full size table

The H2FPEF score was calculated for each patient using the following six clinical and echocardiographic variables: (i) body mass index (BMI) > 30 kg/m2 (H), (ii) the use of ≥ 2 antihypertensive medications (H), (iii) presence of atrial fibrillation (F), (iv) pulmonary hypertension defined as echocardiographic estimated right ventricular systolic pressure > 35 mmHg (P), (v) age > 60 years (E), and (vi) elevated left ventricular filling pressure evidenced by E/e′ > 9 (F). Atrial fibrillation provides 3 points, obesity 2 points, and the remaining four variables each yield 1 point for a maximum possible score of 9, Fig. 2. Only patients with sufficient data to calculate the H2FPEF score were included for analysis (n = 166 subjects).

Fig. 2
figure 2

Point allocations for each clinical variable in the H2FPEF score [15] (top table) along with the associated probability of having heart failure with preserved ejection fraction (HFpEF) based on the resulting total score (bottom graph)

Full size image

Statistical Analyses

Continuous variables are expressed as median (interquartile range) and comparisons were made using the Mann–Whitney U and Kruskal–Wallis test. Correlations between H2FPEF scores and continuous variables were done with Spearman correlation. One-way ANOVA was used for univariate comparisons with H2FPEF since it is ordinal, and generalized linear models were created for multivariable analyses for predictors of the H2FPEF score. Categorical variables are expressed as numbers (percentages), and comparisons were made using a Chi-square test. Statistical analyses were performed using SPSS.

Results

Demographics

Of the 166 patients included in the study, the majority were men (65%) and Caucasian (85%) with a median age of 60 (53–65) years (Table 2). The most common cause of chronic liver disease was NASH (41%) followed by alcohol (34%) and others (25%). Liver disease was advanced with a median MELD score of 22.0 (17.0–28.0) and bilirubin level 3.2 mg/dL (2.0–5.7). Diabetes and hypertension were present in 40% and 44% of the subjects, respectively, and obesity was common (BMI > 30 in 68/166, 41%). Hyperlipidemia was present in 16% of the subjects, but atrial fibrillation was uncommon (7/165, 4%). Patients with NASH cirrhosis were older and had a higher prevalence of the metabolic syndrome (diabetes, hypertension, and hyperlipidemia) than patients with cirrhosis from alcohol or other etiologies (Table 3), but there was no difference between the etiologies with regards to the presence of atrial fibrillation or severity of liver disease. Twenty-eight patients (17%) had LVDD diagnosed by TTE. Fifteen patients (9%) died within the first year of transplant and 6 (4%) developed new-onset systolic heart failure in the post-operative period.

Table 2 Study population demographics
Full size table
Table 3 Comparison of cirrhosis etiologies
Full size table

H2FPEF Score

The median H2FPEF score for the population was 2 (1.0–4.0). The range of H2FPEF scores in the study population is presented in Fig. 3. H2FPEF scores of 1 (38/166, 23%) and 2 (34/166, 20%) had the highest frequency, and the majority of subjects (99/166, 56%) had an H2FPEF score < 3 that correlates with an HFpEF probability < 50%. A high H2FPEF score ≥ 5 that correlates with at least an 80% probability of HFpEF [16] was present in 22 (13%) subjects.

Fig. 3
figure 3

Distribution of the H2FPEF scores in the study population

Full size image

As expected, H2FPEF scores correlated with age (r = 0.47, < 0.001), BMI (r = 0.56, < 0.001) and HgA1c (r = 0.23, 0.003); however, they were not influenced by severity of liver disease (bilirubin, INR, and MELD score) or serum creatinine (Table 4). There was a weak correlation with systolic blood pressure (r = 0.216, p = 0.006) but not with diastolic blood pressure or heart rate. H2FPEF scores were higher in patients with NASH (3.22, 2.79–3.64) compared to alcohol (1.89, 1.5–2.29, p < 0.001) and other causes of chronic liver disease (1.73, 1.28–2.18, p < 0.001, Fig. 4). Additionally, all subjects with H2FPEF scores > 6 (n = 6) had NASH cirrhosis. The association between NASH and higher H2FPEF scores remained even after correcting for age, BMI as well as the presence of hypertension and diabetes (Wald chi square 4.672, p = 0.03, Table 5).

Table 4 Correlation between demographic/clinical variables and H2FPEF scores
Full size table
Fig. 4
figure 4

Pairwise comparison of H2FPEF scores according to cirrhosis etiology. H2FPEF scores were higher in cirrhosis patients with NASH compared to alcohol (3.22, 2.79–3.64 vs. 1.89, 1.5–2.29, p < 0.001) and other chronic liver disease etiologies (3.22, 2.79–3.64 vs. 1.73, 1.28–2.18, p < 0.001). H2FPEF scores were not statistically different between alcohol cirrhosis other chronic liver disease patients (1.89, 1.5–2.29 vs. 1.73, 1.28–2.18, p = 0.88). NS not significant

Full size image
Table 5 Generalized linear model for H2FPEF score
Full size table

On TTE, there was not a correlation between H2FPEF scores and LVEF, peak TR velocity, left ventricular mass or left-sided chamber dimensions (Table 6). The H2FPEF score was higher in patients with DD compared to those without DD (3.11, 2.34–3.87 vs. 2.26, 1.98–2.54 respectively, p = 0.018). Subjects with DD were no different from non-DD subjects with respect to demographics (age, gender, race), components of the metabolic syndrome (diabetes, hypertension, and hyperlipidemia), and serum creatinine. There was also no difference in DD prevalence according to etiology of liver disease or measures of liver dysfunction (bilirubin, INR, and MELD score). Aside from 1 subject, all patients with LVDD had a prolonged QTc. The median QTc was 485.7 (476.3–495.1) in subjects with DD compared to 466.4 (460.8–471.9) in subjects without DD, p = 0.004.

Table 6 Correlation Between Echocardiographic Variables and H2FPEF Scores
Full size table

Post-transplant Outcomes

Within the first year of transplant, 7 patients developed new-onset systolic heart failure and there were 14 deaths. H2FPEF scores were higher in patients who developed new-onset systolic HF (4.0, 3.1–4.9) compared to those who did not develop HF (2.3, 2.1–2.6, p = 0.015, Table 7). Similarly, patients with LVDD were at increased risk of post-transplant HF compared to those without LVDD (4/28, 14.3% vs. 3/138, 2.2%, respectively, p = 0.016). When assessing the individual components of (the H2FPEF score, both the RVSP and presence of atrial fibrillation were associated with an increased risk of post-transplant HF, but not age, BMI, hypertension, or E/e′. Higher H2FPEF scores were not associated with an increased risk of 1-year post-transplant death.

Table 7 Post-transplant systolic heart failure
Full size table

Discussion

In this study, we evaluated clinical determinants of the H2FPEF score in patients with cirrhosis and its potential for identifying those patients at risk for developing systolic heart failure after LT. Overall, H2FPEF scores were modestly elevated in our population and did not appear to be related to the severity of liver dysfunction. However, patients with a diagnosis of NASH had higher H2FPEF scores compared to other causes of chronic liver disease, including alcohol-induced cirrhosis. This association between NASH and higher H2FPEF scores was maintained even after adjusting for hypertension and obesity (which are components of the score but are also major contributors to non-alcoholic fatty liver disease [17,18,19]).

Although there is an established association between NAFLD and heart failure, particularly HFpEF [3,4,5,6,7,8,9, 20], this association has not been identified in patients with cirrhosis. Further investigation is needed to determine whether higher H2FPEF scores in patients with NASH cirrhosis indicates a greater risk of HF complications such as acute kidney injury or HF after procedures that cause cardiac stress such as TIPS. Indeed, there does appear to be an increased risk of renal insufficiency after TIPS in patients with NASH cirrhosis, but it is not clear that this is due to an increased risk of cardiac dysfunction in this population [21].

We were unable to determine whether there is a correlation between H2FPEF score and the severity of portal hypertension or presence of complications of portal hypertension such as varices, ascites, and hepatic encephalopathy. Although portal hypertension-derived, pro-inflammatory cytokines are implicated in the pathophysiology of CCM, there is not an established direct association with severity of portal hypertension and risk of developing CCM [22]. Additionally, our study was not able to determine whether higher H2FPEF scores are associated with increased mortality in cirrhosis patients—since all patients underwent transplant. Although we cannot exclude the possibility that CCM affects mortality in cirrhosis, an association is possible since prior studies have demonstrated that patients with decompensated cirrhosis and an E/e′ > 10 may have reduced survival [23]. Since E/e′ is a component of the H2FPEF score, it is possible that the score could help estimate the mortality risk in cirrhosis as well.

Finally, higher H2FPEF scores were associated with an increased risk of post-transplant systolic HF in our population. Since systolic HF is the most common cardiac complication after liver transplant [24], identifying at-risk patients is important. Although it was designed to help diagnose HFpEF, the H2FPEF score is an attractive candidate predictive tool in this population since it is easily calculated, relying only on readily available clinical data and echocardiographic measurements that are obtained on all patients being evaluated for transplant. However, the number of post-transplant HF events in our study was small, so we were unable to perform multivariable analyses to better assess the influence H2FPEF had on the outcome after controlling for other predictors of post-transplant HF. Additionally, the predictability of the H2FPEF score would need to be compared to established scoring systems that predict the risk of cardiac complications after liver transplant.

Our study is limited by the fact that it is a retrospective analysis from a single center that was predominantly Caucasian. However, aside from the limitation, the population is typical of others in large transplant referral centers. Importantly cardiac testing (including the TTE) was obtained during an outpatient transplant evaluation and not during an acute complication of liver disease such as AKI or infection where hemodynamic changes and therapeutic interventions such as volume expansion could affect the TTE results.

In conclusion, the H2FPEF score in our patient population was increased in NASH cirrhosis patients as well as in those with DD as defined by the Cirrhotic Cardiomyopathy Consortium. Additionally, the H2FPEF score was associated with an increased risk of post-transplant systolic HF. Further investigation is needed to examine the potentially increased risk of HF in NASH cirrhosis compared to causes of liver disease and to determine whether the H2FPEF score could be a useful predictive tool in cirrhosis, both before and after liver transplantation.