Model for End Stage Liver Disease Excluding International Normalized Ratio Predicts Severe Right Ventricular Failure After HeartMate 3 Implantation in a Contemporary Cohort

Abstract

Background

Right ventricular failure (RVF) after left ventricular assist devices is associated with significant morbidity and mortality. Therefore, identifying patients at risk for severe RVF is important for clinical decision‐making. Current risk prediction models were not developed in contemporary populations with left ventricular assist devices and have limited clinical applicability. In this study, we sought to evaluate whether the Model for End Stage Liver Disease Excluding International Normalized Ratio (MELD‐XI) can predict severe RVF after HeartMate 3 implantation.

Methods

We retrospectively analyzed all adult patients who received HeartMate 3 left ventricular assist devices as initial implantation at 2 academic medical centers. We assessed whether MELD‐XI is an independent risk factor for severe RVF in multivariate analysis and compared the predictive accuracy of MELD‐XI with previously published risk scores. We also investigated the relationship between MELD‐XI and markers of RV function and whether MELD‐XI was associated with death or pump exchange at 1‐year follow‐up.

Results

Our study included a total of 246 patients, of whom 74 (30%) experienced severe RVF. After adjusting for relevant covariables, MELD‐XI was independently associated with severe RVF (odds ratio [OR], 1.18 [95% CI, 1.09–1.29]; P<0.001) and performed similarly to the EUROMACS (European Registry for Patients with Mechanical Circulatory Support) and Michigan RVF risk scores. In addition, MELD‐XI was not reflective of traditional echocardiographic or hemodynamic measures of right ventricular function. Finally, MELD‐XI ≥14 predicted worse in‐hospital mortality.

Conclusions

Among patients undergoing HeartMate 3implantation, MELD‐XI is independently associated with an increased risk of RVF and in‐hospital mortality.

Nonstandard Abbreviations and Acronyms

HM3

HeartMate 3

MELD‐XI

Model for End Stage Liver Disease Excluding International Normalized Ratio

PAPi

pulmonary artery pulsatility index

PCWP

pulmonary capillary wedge pressure

RAP

right atrial pressure

RVF

right ventricular failure

Clinical Perspective.

What Is New?

• Despite the improving engineering design of left ventricular assist devices leading to better clinical outcomes, RVF after left ventricular assist device implantation remains a major driver of early morbidity and mortality.

• To fully capture the phenotype and risk of RVF, clinicians should integrate measures of both RV dysfunction severity (right atrial pressure/pulmonary capillary wedge pressure ratio) and chronicity (Model for End Stage Liver Disease Excluding International Normalized Ratio) to help select appropriate patients for left ventricular assist devices therapy, guide shared decision making, and optimize perioperative planning.

What Are the Clinical Implications?

• Future studies should focus on validating the combination of right atrial pressure/pulmonary capillary wedge pressure ratio and Model for End Stage Liver Disease Excluding International Normalized Ratio in predicting severe RVF and incorporating these variables into clinical decision‐making.

Left ventricular assist devices (LVADs) are indicated for the treatment of end‐stage heart failure (HF). Improvements in LVAD technology, specifically the use of the HeartMate 3 (HM3) continuous flow device, have significantly improved clinical outcomes. ¹ However, post‐LVAD right ventricular failure (RVF) remains a frequent complication that leads to an increased risk of renal failure, stroke, prolonged hospitalization, and death. ² , ³ , ⁴ , ⁵ Given the high morbidity and mortality associated with RVF, multiple scores have been developed to estimate the risk of severe RVF following LVAD implantation. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ However, these scores were developed predominantly in non‐HM3 populations and external validation attempts have shown reduced predictive accuracy. ¹⁰ , ¹¹ Thus, an easy‐to‐use and accurate method to predict severe RVF in the contemporary era remains an unmet need in the field.

The liver is the first organ upstream of the right‐sided heart chambers and is sensitive to elevated central venous pressures. Congestive hepatopathy is a recognized complication of HF that can lead to hepatic dysfunction and in some cases cardiac cirrhosis. ¹² , ¹³ The Model for End Stage Liver Disease Excluding International Normalized Ratio (MELD‐XI) is a modification of the original MELD that was created for use in patients requiring anticoagulation, which is common among those with HF. ¹⁴ Elevated MELD‐XI has been associated with worse survival in hospitalized patients with HF and LVAD recipients and may predict RVF in a non‐HM3 population. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Whether MELD‐XI is an independent risk factor for severe RVF in HM3 recipients has not been explored.

In this multicenter study, we sought to investigate whether MELD‐XI is associated with an increased risk of severe RVF in patients undergoing HM3 implantation and to compare the performance of MELD‐XI with previously published risk scores. We hypothesized that higher MELD‐XI would be an independent risk factor for severe RVF and 1‐year mortality.

METHODS

Study Population

We retrospectively analyzed all adult patients over the age of 18 who received an HM3 LVAD as initial implantation between 2015 and 2022 at Washington University in St. Louis and Massachusetts General Hospital. We excluded patients who underwent pump exchange or had insufficient clinical data. Patient demographics and comorbidities were extrapolated from institutional LVAD databases. Laboratory values were taken immediately before surgery and echocardiographic and hemodynamic data were obtained from the most recent transthoracic echocardiogram or right heart catheterization (RHC) within 90 days, respectively. Patient consent was waived and this study was approved by the institutional review boards at Washington University in St. Louis (202205108) and Massachusetts General Hospital (2018P000131). All data that support the findings of this study are available upon request.

MELD‐XI and RVF Risk Scores

MELD‐XI, a previously validated measure of liver dysfunction developed for patients who may be on anticoagulation, was calculated as 5.11×Ln(bilirubin)+11.76×Ln(creatinine)+9.44. ¹⁴ The EUROMACS (European Registry for Patients with Mechanical Circulatory Support) score was calculated as right atrial pressure to pulmonary capillary wedge pressure ratio (RAP/PCWP)>0.54 (2 points)+hemoglobin ≤10 g/dL (1 point)+multiple intravenous inotropes (2.5 points)+INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) profile 1 to 3 (2 points)+severe RV dysfunction on echocardiography (2 points). ⁸ The Michigan RVF risk score was calculated as vasopressor requirement (4 points)+aspartate transaminase ≥80 IU/L (2 points)+bilirubin ≥2.0 mg/dL (2.5 points)+creatinine ≥2.3 mg/dL (3 points). ⁹

Outcomes

The primary outcome was severe RVF according to the INTERMACS definition, characterized by clinical symptoms of elevated filling pressures and needing RV assist device implantation, >14 days of inotropes, or >48 hours of pulmonary vasodilators following HM3 implantation. ² Secondary outcomes were in‐hospital death and a composite of postdischarge death or pump exchange. All patients were followed for at least 1 year and censored at the last date of follow‐up or heart transplant, the latter of which was considered a successful end point.

Statistical Analysis

Clinical and demographic variables were compared between subjects with and without severe RVF. Continuous variables are presented as mean±SD or median (interquartile range) and compared using a t test or Wilcoxon test, depending on the distribution. Categorical and ordinal variables are compared using the chi‐square and Cochran–Armitage trend tests, respectively. MELD‐XI, EUROMACS, and Michigan risk scores were individually calculated for each subject, and receiver operator curves are shown with their area under the curve (AUC) for predicting severe RVF and compared using DeLong test. The correlation between MELD‐XI and echocardiographic and hemodynamic measures of RV dysfunction was assessed visually and by the Spearman correlation coefficient. To evaluate the effect of MELD‐XI on severe RVF, we fit a generalized linear mixed effects model with a random study center effect, using binomial distribution. The variables offered into the model were age, sex, race, NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide), RAP/PCWP, pulmonary artery pulsatility index (PAPi), INTERMACS profile, and MELD‐XI. The final model was selected by backward elimination using Akaike information criterion criteria. Survival was estimated using Kaplan–Meier curves, stratified a priori by MELD‐XI ≥ or <14, and compared using log‐rank test. To assess for time‐to‐outcome of in‐hospital and postdischarge mortality or pump exchange, Cox proportional hazards models were fitted with site correlation as a random effect. Initially, we adjusted for age, sex, race, HF cause, body mass index, RAP/PCWP, INTERMACS profile, and MELD‐XI and our final model was selected by backward elimination using P value <0.1. Finally, in‐hospital mortality was further examined by Kaplan–Meier curves and compared using log‐rank test, using the combination of MELD‐XI and RAP/PCWP as predictors. Proportionality assumptions were tested for all Cox models. All analyses were performed in R, version 4.3.1.

RESULTS

Baseline Characteristics

Our study included 246 HM3 recipients, of whom 28% were Black and 20% were female. In total, 74 (30%) of patients experienced severe RVF by the INTERMACS definition. When comparing those who developed severe RVF to those who did not, baseline demographics and medical comorbidities were similar. However, at time of implantation, patients who developed severe RVF were more likely to be INTERMACS profile 1 or 2 (50% versus 26%, P<0.001), have a higher RAP/PCWP (0.58 versus 0.42, P<0.001), lower PAPi (1.5 versus 2.5, P<0.001), and elevated MELD‐XI (16.56 versus 13.21, P<0.001, Table 1). Between the 2 institutions, those implanted at Washington University in St. Louis had more high‐risk features, with a greater percentage of INTERMACS profile 1 or 2 (42% versus 14%), higher RAP/PCWP (0.53 versus 0.36), lower PAPi (1.80 versus 3.14), higher MELD‐XI (14.54 versus 13.18), and increased incidence of severe RVF (36% versus 19%, Table S1).

Table 1.

Baseline Characteristics of Patients Undergoing HeartMate 3 Implantation

	Overall	Non‐RVF	Severe RVF	P value
No.	246	172	74
Age, y, mean±SD	58±13	59±12	56±14	0.118
Sex, female/male (%)	49/197 (20/80)	31/141 (18/82)	18/56 (24/76)	0.337
Race,White/Black/Other* (%)	177/52/16 (72/21/7)	126/32/14 (73/19/8)	51/20/2 (70/27/3)	0.120
Body mass index (median [IQR])	28 [25–32]	28 [24–32]	29 [26–33]	0.105
Heart failure cause ischemic/nonischemic (%)	102/143 (42/58)	74/97 (43/57)	28/46 (38/62)	0.515
Interagency Registry for Mechanically Assisted Circulatory Support profile (%)				<0.001
1–2	81 (33)	44 (26)	37 (50)
3	118 (48)	87 (51)	31 (42)
4–7	47 (19)	41 (24)	6 (8)
Extracorporeal membrane oxygenation no/yes (%)	233/13 (94.7/5.3)	170/2 (98.8/1.2)	63/11 (85.1/14.9)	<0.001
Left ventricular ejection fraction	18 [14–22]	18 [14–23]	18 [13–20]	0.139
Right atrial pressure/pulmonary capillary wedge pressure	0.47 [0.31–0.66]	0.42 [0.29–0.59]	0.58 [0.44–0.77]	<0.001
Tricuspid annular plane systolic excursion	1.50 [1.30–1.80]	1.50 [1.30–1.80]	1.50 [1.20–1.80]	0.199
Pulmonary artery pulsatility index	2.00 [1.30–3.60]	2.50 [1.60–4.04]	1.50 [1.10–2.44]	<0.001
Log(N‐terminal pro‐B‐type natriuretic peptide)	8.55 [7.96–9.17]	8.53 [7.86–9.13]	8.56 [8.15–9.44]	0.221
Cardiac output	3.73 [3.00–4.50]	3.70 [3.00–4.50]	3.85 [3.00–4.30]	0.750
Bilirubin	0.90 [0.52–1.40]	0.80 [0.50–1.20]	1.20 [0.70–2.08]	<0.001
Creatinine	1.33 [1.05–1.75]	1.27 [1.01–1.68]	1.44 [1.17–1.94]	0.004
Model for End Stage Liver Disease Excluding International Normalized Ratio	14 [11–17]	13 [11–16]	17 [12–19]	<0.001
Michigan RVF	0 [0–11.5]	0 [0–0]	0 [0–2.5]	<0.001
European Registry for Patients with Mechanical Circulatory Support S	5 [3–8]	4 [2–7]	7 [5–9]	<0.001

All continuous variables are expressed as median [IQR]. RVF indicates right ventricular failure. * Other category includes Hispanic ethnicity, Asian, and Native American.

MELD‐XI Is an Independent Risk Factor for Severe RVF After HM3 Implantation

We next analyzed whether MELD‐XI is an independent risk factor for severe RVF. In unadjusted analysis, each 1‐point increase in MELD‐XI was associated with 19% risk of severe RVF (95% CI, 1.11–1.29; P<0.001). Our initial multivariate model adjusted for age, sex, race, NT‐proBNP, INTERMACS profile, RAP/PCWP, PAPi, and MELD‐XI. After backward elimination using Akaike information criterion criteria, age, sex, race, NT‐proBNP, and PAPi were all found to be nonsignificant in a stepwise fashion and thus our final model incorporated only INTERMACS profile, RAP/PCWP, and MELD‐XI to optimize performance in predicting the primary outcome. We found that MELD‐XI was independently associated with severe RVF, with each 1‐point increase in MELD‐XI corresponding to an 18% increased risk of severe RVF (P<0.001, Table 2). This finding was consistent across both institutions (Table S2). In addition, RAP/PCWP (odds ratio [OR], 1.33 for every 0.1‐point increase in RAP/PCWP ratio; P<0.001) and INTERMACS profile 1 or 2, when compared with 4 (OR, 3.12; P=0.032), were risk factors for severe RVF as well.

Table 2.

MELD‐XI Is an Independent Predictor of Severe Right Ventricular Failure After HeartMate 3 Implantation

Variable	OR	95% CI	P value
MELD‐XI*	1.18	1.09–1.29	<0.001
RAP/PCWP†	1.33	1.14–1.54	<0.001
INTERMACS profile 1–2‡	3.12	1.10–8.84	0.032
INTERMACS profile 3‡	1.58	0.57–4.36	0.377

In multivariate analysis, our initial model adjusted for age, sex, race, N‐terminal pro‐B‐type natriuretic peptide, INTERMACS profile, RAP/PCWP, pulmonary artery pulsatility index, and MELD‐XI. After backwards elimination of nonsignificant variables, our final model incorporated MELD‐XI, RAP/PCWP, and INTERMACS profile. INTERMACS indicates Interagency Registry for Mechanically Assisted Circulatory Support; MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio; OR, odds ratio; and RAP/PCWP, right atrial pressure/pulmonary capillary wedge pressure.

^*OR per 1‐point increase in MELD‐XI score.

^† OR per 0.1 increase in RAP/PCWP ratio.

^‡ OR compared with INTERMACS 4–7.

MELD‐XI Performs as Well as Previously Published Risk Scores

Because MELD‐XI was associated with the risk of severe RVF in our patient population, we sought to compare the predictive accuracy of MELD‐XI to previously published risk prediction tools. Two of the most commonly used RVF risk scores are the EUROMACS, which is the most contemporary and only model that included HM3 recipients, and the Michigan RVF score, which is the most widely validated to date. ¹¹ We first validated the EUROMACS (AUC 0.68) and Michigan RVF score (AUC 0.66) in predicting severe RVF in our cohort and subsequently found that MELD‐XI performed similarly (AUC 0.69, Figure 1).

Figure 1. MELD‐XI performs similarly to previously published risk scores to predict severe right ventricular failure after left ventricular assist device implantation.

Comparison of AUC between EUROMACS (blue), Michigan RVF (green), and MELD‐XI (red) in this multi‐institutional cohort of HeartMate 3 recipients. AUC indicates area under the curve; EUROMACS, European Registry for Patients with Mechanical Circulatory Support; MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio; and RVF, right ventricular failure.

MELD‐XI Does Not Correlate With Measures of RV Function

As MELD‐XI is a surrogate marker of hepatorenal disease, which could reflect chronic congestion in patients with HF, we next asked whether it correlated with echocardiographic or hemodynamic measures of RV dysfunction. ² , ¹¹ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ The median time between echocardiogram and and LVAD implantation was 13 and 9 days, respectively. Frequently used parameters in assessing RV dysfunction include tricuspid annular plane systolic excursion (TAPSE), RAP/PCWP ratio, and PAPi. Interestingly, we found that MELD‐XI did not or poorly correlated with all of the following variables: TAPSE (rho=−0.15), RAP (rho=0.22), RAP/PCWP ratio (rho=0.16), mean pulmonary artery pressure (rho=0.27), PAPi (rho=−0.09), and pulmonary vascular resistance (rho=0.07, Figure 2).

Figure 2. MELD‐XI does not or poorly correlate with echocardiographic and hemodynamic measures of right ventricular function.

Box plots of median days between right heart catheterization (9 days) and echocardiogram (13 days) and LVAD implantation (top). Dot plots of the relationship between MELD‐XI and different measures of right ventricular function and pulmonary artery hemodynamics (bottom). Each dot represents an individual patient. LVAD indicates left ventricular assist device; MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio; mPAP, mean pulmonary artery pressure; PAPi, pulmonary artery pulsatility index; PCWP, pulmonary capillary wedge pressure; PVR, pulmonary vascular resistance; RAP, right atrial pressure; RHC, right heart catheterization; and TAPSE, tricuspid annular planar systolic excursion.

MELD‐XI Predicts Early But Not Late Mortality After HM3 Implantation

Finally, we assessed whether preoperative MELD‐XI can predict mortality after HM3 implantation. In our cohort, 45 (18%) of patients died within 1 year of follow‐up with median time‐to‐death of 32 days. When stratified by MELD‐XI ≥14 or<14 a priori, we found that MELD‐XI ≥14 was associated with significantly worse in‐hospital mortality after LVAD implantation (log‐rank P=0.002). However, MELD‐XI did not predict the risk of death or pump exchange at 1‐year follow‐up among patients who survived index hospitalization (log‐rank P=0.9, Figure 3). The effect of MELD‐XI was confirmed in the multivariate Cox proportional hazard model. In unadjusted analysis, each 1‐point increse in MELD‐XI was associated with a 9% increase in mortality during index hospitalization (95% CI, 1.03–1.14; P=0.002). Our initial Cox model adjusted for age, sex, race, body mass index, HF cause, INTERMACS profile, RAP/PCWP, and MELD‐XI. After backward elimination, age, body mass index, RAP/PCWP, and MELD‐XI were significant predictors of in‐hospital mortality, with each 1‐point increase in MELD‐XI corresponding to 11% increased risk of death during index hospitalization (P=0.003, Table 3). On the other hand, MELD‐XI did not predict mortality or pump exchange after discharge (hazard ratio, 1.05 [95% CI, 0.94–1.16]; P=0.38). Given the changing demographics of LVAD recipients after the 2018 United Network for Organ Sharing heart transplant allocation system change, we analyzed whether MELD‐XI still performs well in the contemporary era. ²⁵ , ²⁶ We found that MELD‐XI ≥14 consistently predicted early but not late mortality before and after the date of the allocation system change (Figure 4). Finally, as both MELD‐XI and RAP/PCWP predicted severe RVF and death, we asked whether the combination is a better marker of adverse clinical outcomes. Indeed, we found that the combination of MELD‐XI and RAP/PCWP predicted severe RVF with improved discrimination (AUC 0.76) and patients with high MELD‐XI and RAP/PCWP had the worst survival (log rank P<0.01, Figure 5A and 5B).

Figure 3. MELD‐XI predicts early but not late mortality or pump exchange after HeartMate 3 implantation.

Kaplan–Meier curves are shown, stratified by MELD‐XI ≥14 (red) and <14 (blue). Patients with higher MELD‐XI had worse in‐hospital mortality (left, log rank P=0.002), but MELD‐XI did not predict survival after discharge from index hospitalization (right, log rank P=0.9). MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio.

Table 3.

MELD‐XI Is Associated With Increased Mortality During Index Hospitalization

Variable	HR	95% CI	P value
MELD‐XI*	1.11	1.04–1.20	0.003
Age†	1.65	1.09–2.49	0.018
RAP/PCWP‡	1.21	1.02–1.45	0.031
BMI§	1.07	1.00–1.14	0.048

In multivariate analysis, our initial Cox‐proportional hazards model was adjusted for age, sex, race, BMI, heart failure cause, Interagency Registry for Mechanically Assisted Circulatory Support profile, RAP/PCWP, and MELD‐XI. After backward elimination using P<0.1 as cutoff, age, MELD‐XI, RAP/PCWP, and BMI were independent predictors of in‐hospital mortality. BMI indicates body mass index; HR, hazard ratio; MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio; and RAP/PCWP, right atrial pressure/pulmonary capillary wedge pressure.

^*HR per 1‐point increase in MELD‐XI.

^† HR per 10‐year increase in age.

^‡ HR per 0.1 increase in RAP/PCWP.

^§ HR per 1 kg/m² increase in BMI.

Figure 4. MELD‐XI remains predictive of in‐hospital mortality in the contemporary era.

Kaplan–Meier curves are shown before (left) and after (right) the 2018 heart transplant allocation system change. Patients with MELD‐XI ≥14 (red) had higher in‐hospital mortality or pump exchange than those with MELD‐XI <14 (blue) in both eras (log rank P=0.039 and 0.021 for before and after, respectively). MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio.

Figure 5. MELD‐XI and right atrial pressure/pulmonary capillary wedge pressure ratio is associated with the worst outcomes after HeartMate 3 implantation.

The combination of MELD‐XI ≥14 and RAP/PCWP ≥0.55 predicted severe RVF with the greatest discrimination (A) and portended the worst in‐hospital mortality (B). AUC indicates area under the curve; MELD‐XI, Model for End Stage Liver Disease Excluding International Normalized Ratio; RAP/PCWP, right atrial pressure/pulmonary capillary wedge pressure; and RVF, right ventricular failure.

DISCUSSION

In this multicenter retrospective analysis, we found that MELD‐XI was independently associated with severe RVF and in‐hospital mortality after HM3 implantation. In addition, MELD‐XI performed just as well as previously validated RVF risk scores but interestingly, did not correlate with traditional echocardiographic and hemodynamic measures used to assess RV function. These findings shed important insights into predictors of clinical outcomes in contemporary LVAD recipients and the syndrome of RVF.

RVF Is the Major Cause of Early Mortality and Previous Risk Scores Have Limited Applicability in the Contemporary Era

The occurrence of severe RVF after LVAD increases the risk of in‐hospital mortality as well as significant morbidity including the need for renal replacement therapy, stroke, and gastrointestinal bleeding. ² , ³ , ⁴ , ²⁷ , ²⁸ , ²⁹ Thus, identification of patients at increased risk of RVF is paramount in the evaluation of patients for LVAD candidacy, shared decision‐making to proceed with HM3 implantation, and perioperative planning. Previous studies have analyzed preoperative risk factors that increase the likelihood of severe RVF, using a combination of hemodynamic, echocardiographic, and clinical variables to generate a weighted risk score. Of all the scores, EUROMACS is the only one to include HM3 recipients; however, patients with this device comprised only 9% of the derivation cohort. ⁸ The Michigan RVF risk score, on the other hand, is the most validated tool to date but was developed during a time when only 14% of patients had continuous flow LVAD devices. ⁹ , ¹¹ Furthermore, multiple external validation attempts have yielded disappointing results. ¹⁰ , ³⁰ Therefore, identifying an easy‐to‐use risk prediction tool applicable to HM3 recipients remains an unmet clinical need.

MELD‐XI Is Simple, Widely Available, and Has Broad Application

MELD‐XI was originally introduced by Heuman et al. to account for the high prevalence of anticoagulation use in patients with HF, which correlated well with MELD in patients with cirrhosis. ¹⁴ Elevated MELD and MELD‐XI have been shown to be associated with a higher risk of mortality in patients undergoing LVAD. ¹⁶ , ¹⁷ , ¹⁸ , ³¹ , ³² However, these studies were largely single centered and conducted in an era of non‐HM3 devices. In the present study, we show that higher MELD‐XI is associated with an increased likelihood of severe RVF and the discriminatory power for severe RVF is comparable to previous risk scores, many of which require 4 or more variables. On the other hand, MELD‐XI is easy to calculate and relies on only bilirubin and creatinine, 2 readily available clinical laboratory values. Moreover, we found that MELD‐XI predicted in‐hospital mortality and was not affected by the United Network for Organ Sharing allocation system change in October 2018, making it a powerful tool for preoperative risk assessment in patients undergoing evaluation for LVAD in the contemporary era. We also found that MELD‐XI was not associated with mortality beyond index hospitalization. First, this further supports our primary finding that MELD‐XI is a robust predictor of severe RVF, which affects early mortality. We believe this observation adds nuance to the existing literature that although higher MELD‐XI is associated with worse 1‐year mortality, the biggest risk is seen early. Second, HM3 support may lead to improvement in hepatorenal disease over time if patients survive the initial postoperative period. This is in line with Yang et al., who showed that patients who experienced an improvement in MELD‐XI post LVAD had similar survival after heart transplant to those who had a low MELD‐XI to begin with. ³³ More recently, Mehra et al. derived and validated the HM3 risk score in MOMENTUM 3 (Multicenter Study of MagLev Technology in Patients Undergoing Mechanical Circulatory Support Therapy With HeartMate 3) clinical trial population. ³⁴ The HM3 risk score also incorporates measures of both renal and RV dysfunction (serum urea nitrogen and RAP/PCWP, respectively). However, exploratory analysis in our cohort showed that HM3 risk score did not predict severe RVF (AUC 0.48), possibly related to its not incorporating markers of liver disease.

Hepatorenal Dysfunction and the Syndrome of Right Heart Failure

Intriguingly, we found that although MELD‐XI predicted severe RVF, it does not correlate with echocardiographic or hemodynamic measures of RV function used clinically. In predominantly single‐centered studies, lower TAPSE and PAPi and higher RAP/PCWP have all been shown to increase the risk of severe RVF, which is in line with our finding that RAP/PCWP (although not TAPSE/PAPi) is an independent risk factor for severe RVF. In our cohort, MELD‐XI did not correlate or correlated poorly with measures of RV systolic function (TAPSE), right‐sided filling pressure (RAP), pulmonary artery pressure, or relative RV performance (PAPi, RAP/PCWP). Taken together, MELD‐XI seems to provide additional and incremental information about the risk of RVF in patients undergoing LVAD implantation independent of aforementioned measures of RV function. Increasingly, RVF is recognized as a systemic syndrome characterized by end‐organ dysfunction due to chronic venous congestion. ¹³ , ³⁴ , ³⁵ More recently, our group has shown that patients with clinical RVF have a unique systemic inflammatory profile and some of these circulating factors may be released by Kupffer cells, which are resident hepatic macrophages. ³⁶ We hypothesize that MELD‐XI reflects the global impact of RVF over time in a way that is not captured by snapshot measurements inherent in echocardiographic or hemodynamic parameters. The greatest risk of severe RVF and worst outcomes were observed in patients with high MELD‐XI and RAP/PCWP. Therefore, integrating RV dysfunction severity (RAP/PCWP ratio) and its downstream consequences (MELD‐XI) may better capture the overall phenotypic profile of RVF than either alone.

Limitations

Our study has several limitations. First, it is a retrospective analysis of patients treated at 2 academic medical centers and may not represent other practices. However, this is one of the largest HM3‐exclusive studies to date and the primary outcome of MELD‐XI predicting severe RVF was consistent across both institutions, increasing the generalizability of our findings. Second, we acknowledge that there are multiple risk scores to predict severe RVF. ¹¹ We picked the EUROMACS and Michigan scores because they are the only ones to incorporate HM3 recipients and have been the most widely validated, respectively. Third, although we found that MELD‐XI does not correlate with traditional RV parameters, we acknowledge that these measurements were obtained between 1 and 2 weeks before LVAD and may represent a “decompensated” state. It is possible that after optimization, echocardiographic or hemodynamic values may better correlate with MELD‐XI, consistent with Gonzalez et al. who showed that optimized PAPi is a better predictor of RVF post LVAD. ³⁷ However, this limitation is nevertheless consistent with current clinical practice and has been used in previous studies. Understanding the dynamic changes in MELD‐XI and other parameters should be an area of future research. Finally, although MELD‐XI is associated with severe RVF, our study suggests that combining MELD‐XI and RAP/PCWP ratio may offer even more predictive accuracy. This needs to be validated in an external cohort.

CONCLUSIONS

MELD‐XI is independently associated with an increased risk of severe RVF and in‐hospital mortality after HM3 implantation.

Sources of Funding

This study was funded by the Mentors in Medicine program at Washington University School of Medicine in St. Louis.

Disclosures

None.

Supporting information

Tables S1–S2