Using noninvasive clinical parameters to predict mortality and morbidity after cardiac interventions in patients with cirrhosis: A systematic review

Abstract

Background:

Cardiovascular disease commonly affects advanced liver disease patients. They undergo cardiac interventions to improve cardiac outcomes. Cirrhosis increases complication risk, including bleeding, renal and respiratory failure, and further decompensation, including death, posing a clinical dilemma to proceduralists. Predicting outcomes is crucial in managing patients with cirrhosis. Our aim was to systematically review clinical parameters to assess the mortality and complication risk in patients with cirrhosis undergoing cardiac interventions.

Methods:

We searched cirrhosis and cardiovascular intervention terminology in PubMed and Excerpta Medica Database (EMBASE) from inception to January 8, 2023. We included studies reporting clinical scores (e.g. Model for End-stage Liver Disease (MELD), Child–Pugh–Turcotte (CPT), cardiovascular interventions, mortality, and morbidity outcomes). We independently abstracted data from eligible studies and performed qualitative summaries.

Results:

Eight studies met the inclusion criteria. Procedures included tricuspid valve surgery, catheterization-related procedures, aortic valve replacement (AVR), pericardiectomy, and left ventricular assist device (LVAD) placement. MELD primarily predicted mortality (n = 4), followed by CPT (n = 2). Mortality is significantly increased for MELD > 15 after tricuspid valve surgery. Albumin, creatinine, and MELD were significantly associated with increased mortality after transcatheter AVR (TAVR), although specific values lacked stratification. CPT was significantly associated with increased mortality after cardiac catheterization or pericardiectomy. In LVAD placement, increasing MELD increased the unadjusted odds for perioperative mortality.

Conclusions:

Our systematic review showed that clinical parameters predict mortality and morbidity risk in patients with cirrhosis undergoing cardiac procedures.

INTRODUCTION

Cardiovascular disease is a common cause of mortality among the aging population, including those with advanced liver disease. Nonalcoholic fatty liver disease (NAFLD) is associated with metabolic syndrome and, therefore, associated with cardiovascular disease risk.[1] Given the shared risk factors of alcohol-use disorder and obesity, coronary artery disease (CAD) and cirrhosis are rising.[2] As such, these patients undergo cardiac interventions and studies, to treat and prevent adverse cardiac outcomes. Patients with cirrhosis are known to have an increased postoperative complication risk: major bleeding, renal and respiratory failure, and further decompensation. These are true after cardiac procedures and are well studied in patients undergoing cardiac surgery.[3] Tools need to be in place to assess the risks of cardiovascular procedures. Although risk scores such as the Child–Pugh–Turcotte (CPT) and Model for End-stage Liver Disease (MELD) with sodium correction (MELD-Na) have been used to help determine mortality after certain cardiac procedures, little evidence exists to understand their utility in determining complications from these procedures in cirrhotic patients. Little data exists for nonsurgical cardiac procedures, such as coronary catheterization, which poses its own complications in cirrhotic patients due to thrombocytopenia and coagulopathy.[4] Other catheter-based procedures share their risk of bleeding and vascular complications in patients with cirrhosis.[5] Using noninvasive assessments to understand the complication risk is valuable in managing patients with cirrhosis and cardiovascular disease. We systematically reviewed available clinical scores to assess mortality and complication risk in patients with cirrhosis undergoing cardiovascular interventions.

METHODS

We included observational or experimental studies involving the introduction of intervention and measurement of outcomes. There was no language restriction when searching for studies. The authors were not contacted if critical information could not be abstracted from the full manuscript.

We created a search with key terms such as cirrhosis and cardiovascular interventions in PubMed and Excerpta Medica Database (EMBASE) from database inception to January 8, 2023 [Figure 1]. No other reference databases were used. The search utilized a combination of free text and the Medical Subject Headings (MeSH) database of PubMed and EMTREE terms (EMBASE). The included search terms are seen in Supplementary Table 1. Two reviewers (CM and AP) independently assessed the studies.

Figure 1.

Flowchart of the study up to search completed August 2022

Supplementary Table 1.

Search terms

“Liver disease” OR “Hepatic disease” OR (liver AND disease) OR (hepatic AND disease) OR “chronic liver disease” OR “advanced liver disease” OR “hepatic failure” OR “hepatic dysfunction” OR “liver dysfunction”

Advanced liver fibrosis OR liver fibrosis OR F4 OR hepatic fibrosis Cirrhosis OR cirrhotic

Liver dysfunction OR liver failure

Portal hypertension”

AND

Cardiovascular surgery” OR “Cardiothoracic surgery” OR “cardiac catheterization” OR “cardiac ablation” OR “percutaneous coronary intervention” OR “cardiac stress test” OR “valve replacement” OR “coronary artery bypass” OR “CABG” OR “cardioversion” OR “valve surgery” OR “heart surgery” OR “cardiac surgical procedures” OR “coronary artery bypass grafting” OR “cardiovascular operation” OR “cardiopulmonary bypass” OR “thoracic surgery” OR “cardiac operation” OR “mitral valve replacement” OR “pericardiectomy” OR aortic valve replacement” OR “tricuspid annuloplasty” OR “transcatheter valve replacement” OR “transcatheter valve implantation” OR “surgical valve replacement” OR “surgical valve implantation” OR “transcatheter interventions” OR “TAVR” or “SAVR” OR “Balloon valvuloplasty”

AND

“Risk scores” OR “MELD-Na” OR “Child-Pugh” OR “Child Pugh Turcotte” OR “Model for End Stage Liver Disease” OR “VOCAL Penn” OR “EuroSCORE”

OR

“Serum biomarkers” OR “biological marker” OR “biomarker” OR “biomarkers” OR “markers” OR “blood” OR “serum” or “plasma”

AND

“decompensation” OR “hemorrhage” OR “ascites” OR “encephalopathy” OR “jaundice” OR “varices” OR “mortality” OR “survival” OR “survival rate” OR “death” or “fatality”

Inclusion and exclusion criteria

We included studies reporting clinical scores (e.g., MELD and CPT), cardiovascular interventions, and mortality outcomes or complications. Studies included assessed participants with cirrhosis as defined by the authors, independent of patients’ settings. Excluded articles involved those not stratifying clinical parameters to assess risk by cardiac procedure type. For example, if a study was utilizing a MELD score for preoperative risk of multiple cardiac procedures (surgical and percutaneous) and listed outcomes from the procedures as a whole, and not individually per procedure type, the study was excluded. We excluded abstracts presented in scientific meetings lacking detailed information.

Cardiac procedures

We assessed the following interventions: tricuspid valve surgery (replacement or repair), percutaneous coronary intervention, right and left coronary artery catheterization, transcutaneous aortic valve replacement (AVR) or repair, pericardiectomy, ablation procedures, and left ventricular assist device (LVAD) placement. Studies evaluating multiple cardiac procedures were excluded because when assessing the noninvasive risk factors or scores (e.g. VOCAL-Penn Score[6]) composite data were provided.

Quality assessment of studies

We assessed the articles’ methodological quality using the Newcastle–Ottawa Quality Assessment Scale (NOS) with a maximum score of 8, defining good methodological quality when NOS ≥6 [Supplementary Table 2]. A third reviewer (RH) reconciled disagreements. Article selection was completed using Covidence software. Given clinical heterogeneity, we did not pursue a meta-analysis and provided a qualitative summary of each study.

Supplementary Table 2.

Quality Assessment Tool Using Modified Newcastle–Ottawa Scale

Criteria	Answer yes/no
1. Was the definition of patients based on the presence of cirrhosis?
2. Was the presence of cirrhosis confirmed in all patients?
3. Were data obtained from the electronic medical record?
4. Were noninvasive biomarkers or scores calculated before the cardiac procedure?
5. If there was a comparator group, were groups stratified by a noninvasive biomarker or score?
6. Was mortality and/or morbidity evaluated for all patients?
7. Was the length of follow-up at least 30 days post-procedure?
8. Was the loss of follow-up <10% of the original cohort?
Study	Procedure type	Study design	Q.1	Q.2	Q.3	Q.4	Q.5	Q.6	Q.7	Q.8	Total score (out of 8)
Ailawadi	Tricuspid valve surgery (replacement or repair)	RS	1	1	1	1	1	1	1	1	8
Krill	Percutaneous coronary intervention	RS (case–control)	1	1	1	1	NA	1	1	1	7
Townsend	Right and left heart catheterization	RS	1	1	1	1	1	1	1	1	8
Arai	TAVI	PS	1	1	1	1	1	1	1	1	8
Nakamura	TAVR	RS (observational)	1	1	1	1	NA	1	1	1	7
Peeraphatdit	TAVR	RS (cohort)	1	1	1	1	NA	1	1	1	7
Komoda	Pericardiectomy	RS	1	1	1	1	1	1	1	1	8
Matthews	LVAD placement	RS	1	1	1	1	1	1	1	1	8

Q=Question, 1=Yes, 0=No, NA=Not applicable

Noninvasive risk assessment

We defined noninvasive risk stratification as any laboratory marker (i.e., international normalized ratio (INR), albumin, or creatinine) or score (MELD, MELD excluding INR (MELD-XI), or CPT). Derivations for each score are shown in Supplementary Table 3.

Supplementary Table 3.

Noninvasive risk score calculations

Score	Variables	Calculation
Model for End-Stage Liver Disease (MELD)	Serum bilirubin, creatinine, INR	9.57 × loge (creatinine) + 3.78×loge (total bilirubin) + 11.2 × loge (INR) + 6.43
MELD-Na	Serum bilirubin, creatinine, INR, sodium	MELD +1.32 × (137 - Na) - [0.033 x MELD*(137 - Na)]
MELD-XI	Serum bilirubin, serum creatinine	0.11 × ln (serum bilirubin in mg/dL) ± 11.76 × ln (serum creatinine in mg/dL) ± 9.44
Child–Turcotte–Pugh Score	Serum bilirubin, albumin, INR, ascites, hepatic encephalopathy	Each measure scored 1–3 (3 is most severe). Class A (5–6 points), Class B (7–9 points), Class C (10–15 points)
International normalized ratio (INR)	Patient prothrombin time (PT), control prothrombin time	Patient PT÷control PT

Data extraction

Two independent authors (CM and AP) extracted data using a standard spreadsheet. We extracted procedure type, study design, duration, aim, measured outcomes, total patients, cirrhosis etiologies, cardiac comorbidities, mean patient age, concomitant operations, noninvasive risk factor(s) used, risk factor subgroups, mortality or morbidity outcome, and results with statistical significance [Table 1].

Table 1.

Characteristics of studies

Author, year	Procedure type	Design	Years studied	Aim	Total patients	Cirrhosis etiologies	Cardiac comorbidities
Ailawadi, Ann Thoracic Surg 2009	Tricuspid valve surgery (replacement or repair)	RS	1994–2008	To determine whether liver disease affects outcomes in tricuspid valve surgery	168	Alcohol related (33%), hepatitis B (11%), hepatitis C (14%)	DM (21%), PVD (12%), renal failure (21%), congestive heart failure (70%), myocardial infarction (18%)
Krill, AP&T 2017	Percutaneous coronary intervention	RS (case–control)	2000–2015	Compare GIB and mortality rates in cirrhotic patients with CAD	148 (68 with stent placement)	Hepatitis C (20.6%), alcohol (27.9%) and nonalcoholic (25%) steatohepatitis, other (4.4%)	DM, hypertension, dyslipidemia, pulmonary embolism
Townsend, Am J Cardiol 2012	Right and left heart catheterization	RS	2003–2009	If elevated INR is predictive of complications from cardiac catheterization in patients with ESLD	240 (157 with RHC and 83 with LHC)	NR	NR
Arai, Int J Cardiol 2017	TAVI	PS	2013–2015	Assess the prognostic value of MELD-XI for patients undergoing TAVI	749	NR	DM (25.2%), hypertension (75.6%), dyslipidemia (42.7%), atrial fibrillation (6.6%), CHF (47.8%), CKD (67.2%)
Nakamura, Structural Heart 2019	TAVR	RS (observational)	2009–2016	Liver cirrhosis affecting postoperative outcomes of TAVR	430 (30 with liver cirrhosis)	Viral (73.3%), alcohol (6.7%), viral and alcohol (3.3%), unknown, 6.7%), other (10.0%)	Prior cardiac surgery (13.3%), PVD (36.4%), DM (27.3%), HTN (27.3%), CAD (31.8%)
Peeraphatdit, Hepatology 2020	TAVR	RS (cohort)	2008–2016	Compare outcomes following TAVR and SAVR in patients with cirrhosis to inform the preferred intervention	105 (55 with TAVR)	Alcohol (10.9%), nonalcohol (50.9%), hepatitis C (7.3%), other (30.9%)	DM (56.4%), HTN (85.5%), PVD (50.9%), CHF (81.8%)
Komoda, Ann Thoracic Surg 2013	Pericardiectomy	RS	1996–2011	If scoring systems can predict mortality after radical pericardiectomy in patients with constrictive pericarditis	64	NR	Atrial fibrillation (46.9%), DM (18.8), ESRD (4.7%)
Matthews, Circulation 2010	LVAD placement	RS	1996–2007	Assess the ability of MELD to predict operative morbidity and mortality in LVAD recipients	211	NR	DM (25.1%), ischemic cardiomyopathy (51.7%)

AS: aortic stenosis, CAD: coronary artery disease, CHF: congestive heart failure, DM: diabetes mellitus, ESLD: end-stage liver disease, GIBL: gastrointestinal bleeding, HTN: hypertension, LVAD: left ventricular assist device, NR: not reported, PS: prospective study, PVD: peripheral vascular disease, SAVR: surgical aortic valve replacement, TAVI: transcatheter aortic valve implantation, TAVR: transcatheter aortic valve replacement, RS: retrospective study

If a study provided a specific risk factor cutoff, we extracted the distinct categories as specified by the study authors and reported the statistical significance reported by the authors. We used qualitative summary estimates for outcomes (e.g., odds ratio [OR] and 95% confidence intervals [CIs]) [Table 2].

Table 2.

Data extraction of studies with noninvasive scores used and published outcomes

Author, year	Design	Outcomes measured	Total patients	Mean age (years)	Noninvasive risk factor (s) used	Noninvasive risk factor subgroups	Mortality outcome (unless specified otherwise)	P		P
TAVR
							Life-threatening bleeding		30-day survival
Arai, Int J Cardiol 2017	PS	Complications during TAVI	749	84.3±5.2	MELD-XI	MELD-XI >10 (n=136)	11%	0.03	95.6%	0.03
						MELD-XI ≤10 (n=613)	5.9%		98.5%
							Hazard ratio (2-year mortality)
Nakamura, Structural Heart 2019	RS	All-cause mortality and major adverse cardiac events1	430 (30 with liver cirrhosis)	78±7	Albumin	<3.0 g/dL	0.065 (95% CI 0.0053–0.42)	<0.005
							Hazard ratio2
Peeraphatdit, Hepatology 2020	RS (cohort)	Compare outcomes in patients with and without cirrhosis, predictors for survival, create the selection algorithm	55	75.4±9.4	Multiple3	MELD	1.13 (95% CI 1.05–1.21)	0.002
						MELD-XI	1.11 (1.06–1.17)	<0.001
						Cr	1.36 (1.08–1.72)	0.01
						Albumin	0.6 (.38–.94)	0.03
Catheterization-related procedures
							Odds Ratio4
Krill, AP&T 2017	RS, Case–control	Rates of GIB (30 days, 90 days, 1 year, 2 years) and mortality (30 days, 90 days, 1 year, and 2 years)	148 (68 patients with stent placement)	60.4±0.6	Child–Pugh–Turcotte	Not stratified	2.43 (1.58–3.73)	<0.0001
							Bleeding (pre- and post-procedure Hgb (g/dL))
Townsend, Am J Cardiol 2012	RS	Rates of bleeding complications, blood product administration	240 (157 with RHC and 83 with LHC)	RHC: INR ≤1.5 55±7.9	INR	RHC: INR ≤1.5	9.7 vs 9.7	0.98
				RHC: INR >1.5 52±10		RHC: INR >1.5	11.1 vs 10.9	0.10
				LHC: INR ≤1.5 57±9.2		LHC: INR ≤1.5	9.5 vs 9.6	0.87
				LHC: INR >1.5 52±7.8		LHC: INR >1.5	9.7 vs 9.7	0.99
Pericardiectomy
							Survival (5-year)	0.000
Komoda, Ann Thoracic Surg 2013	RS	Actuarial survival and prognostic factors	45	52.4±15.0	Child–Pugh–Turcotte	CPT-A	80.8%
			19	62.9±9.7		CPT-B/C	37.9%
							Survival (30-day)	0.000
						CPT-A	2.2%
						CPT-B/C	36.8%
LVAD
							6 months Adjusted odds ratio		Hazard ratio (95% CI)
Matthews, Circulation 2010	RS	MELD scores to predict perioperative blood product administered, operative death	211	Operative death: 55±17 Survivors 49±13	MELD	Each 5-unit increase in MELD increased the unadjusted odds	1.6 (95% CI 1.1–2.3)	0.013
						Survival rate
						UMHS MELD <17	88±3%	0.009	2.5 (1.2–5.3)
						UMHS ≥17	74±6%
						INTERMACS MELD <17	82±3%	0.032	2.5 (1.1–5.4)
						INTERMACS MELD ≥17	67±5%
Tricuspid valve surgery
								Rate (30-day)
Ailawadi, Ann Thoracic Surg 2009	RS	Use of MELD in stratifying patients undergoing surgery and predict mortality	168	61±14 years	CABG (n=32), mitral valve replacement (n=76), mitral valve repair (n=14), and aortic valve replacement (n=42)	MELD	MELD <10 (n=50)	1.90%
							MELD 10–14.9 (n=44)	6.80%
							MELD 15–19.9 (n=11)	27.3%
							MELD >20 (n=13)	30.8%	0.002

1. Major adverse cardiac events defined by readmission due to cardiac events, such as heart failure, arrhythmia, coronary artery disease, and prosthetic valve dysfunction. 2. No specific time frame was identified for mortality. 3. MELD, Cr, and albumin were not stratified for mortality in the cohort. In the subgroup of patients with a MELD score <12, the TAVR group had reduced survival compared with the SAVR group (median survival of 2.8 vs. 4.4 years; P=0.047). In those with a MELD score ≥12, survival after TAVR, SAVR, and medical therapy was similar (1.3 vs. 2.1 vs. 1.6 years, respectively; P=0.53). MELD-XI: MELD excluding INR score (created for patients with elevated INR due to anticoagulation). 4. No specific time period was identified for mortality

RESULTS

Overall results of studies

Our search identified 9014 studies, and eight (2,115 patients) met the inclusion criteria [Table 1]. Three studies included transcutaneous AVR or repair (1,284 patients). Independent studies were included for tricuspid valve surgery (168 patients), percutaneous coronary intervention (148 patients), right and left catheterization (240 patients), pericardiectomy (64 patients), and LVAD placement (211 patients). Of the 2,115 patients in our studies, most included a form of MELD (1,183 patients) versus CPT in 212 patients. Other markers included international normalized ratio (INR) in 157 patients, as well as creatinine and albumin in 55 patients. Six studies reported mortality rate as the primary outcome, and two studies reported bleeding risk as the primary outcome. The overall quality of the studies based on NOS was 7.6 (ranging from 7 to 8).

Transcatheter AVR (TAVR)/repair

Arai et al.[7] prospectively studied 749 patients to evaluate the MELD-XI score in prognosticating complications of transcatheter aortic valve implantation (TAVI). The incidence of life-threatening bleed was higher in the high MELD-XI group (MELD-XI >10). 30-day and 6-month survival was significantly lower in the high MELD-XI group. Predictors of 6-month mortality using the Cox regression analysis were prior coronary artery bypass graft surgery (CABG), estimated glomerular filtrate rate <60 mL/min/1.73 m², MELD-XI score (hazard ratio [HR] 3.50, 95% CI 1.87–6.56, P < 0.01), and chronic atrial fibrillation. On multivariate analysis, the MELD-XI score was a predictor of 6-month cumulative mortality (HR 2.21, 95% CI 1.06–5.54, P < 0.02). Receiver operating characteristic analysis revealed that MELD-XI showed better accuracy in predicting 6-month mortality when compared to Logistic European System for Cardiac Operative Risk Evaluation, European System for Cardiac Operative Risk Evaluation II, and Society of Thoracic Surgeons scores (area under the curve = 0.67, 0.58, 0.57, and 0.60, respectively).

Nakamura et al.[8] studied outcomes in TAVR procedures using a retrospective study of 430 patients. No significant difference was found in early mortality, hospital stay, or major complications between cirrhosis and non-cirrhosis patients. The midterm overall survival was significantly lower in the cirrhosis group, but major adverse cardiac events were not significantly different. Univariate and multivariate Cox regression analysis found cirrhosis as a significant predictor for postoperative survival. Serum albumin <3.0 g/dL had a significantly lower 2-year survival rate than those with levels >3.0 g/dL.

Peeraphatdit et al.[9] retrospectively studied outcomes following TAVR and surgical AVR (SAVR) in 105 patients. The postoperative complication rates between the TAVR and SAVR were not statistically significant among stratified groups. Using univariate Cox regression, MELD score, MELD-XI, creatinine, and albumin were factors significantly associated with post-AVR mortality. In multivariate analysis, MELD score was associated with mortality. Based on the Kaplan–Meier curves, survival benefit was noted in SAVR over TAVR in MELD scores <12.

Catheterization-related procedures

Krill et al.[10] evaluated cirrhosis patients who underwent coronary interventions and the effect of dual antiplatelet therapy (DAPT), which is known to cause gastrointestinal bleeding (GIB), after the procedure on mortality, compared with those who underwent medical therapy for CAD. In this retrospective study of 148 patients, 68 were treated with stent placement. Indications for stent placement were stable angina, acute coronary syndromes, congestive heart failure (CHF) workup, and liver transplant workup. Those who underwent stent placement compared with medical therapy had significant GIB at the 1- and 2-year mark (Krill), with the majority of bleeding from upper gastrointestinal sources. Age, prior myocardial infarction, proton pump inhibitor (PPI) use, and race were independent predictors of GI bleeding. There was no significant difference in mortality between the two groups at any time point. Age, varices, and CPT score were independent factors for mortality.

Townsend et al.[11] investigated the predictive effects of elevated INR in patients undergoing right and left heart catheterization. This retrospective study included 240 patients. There were no significant changes in pre- and post-procedural hemoglobin, even in the stratified INR groups. In the right heart catheterization (RHC) group, two cases had anemia due to GIB. In the left heart catheterization (LHC) group, three patients had anemia due to GIB.

Pericardiectomy

Komoda et al.[12] studied the prognostic effects of scoring systems for mortality after radical pericardiectomy in patients with constrictive pericarditis. In a retrospective review of 64 patients, a Kaplan–Meier model evaluated mortality, with univariate and multivariate Cox regression analysis for risk factors. Patients with CPT-B and CPT-C had significantly lower 30-day and 5-year survival rates compared with CPT-A. Significant prognostic factors for mortality based on univariate analysis included age, serum creatinine >1.1, and CPT score of 7 or more. A CPT score greater than 7 was an independent prognostic factor, while MELD was not.

LVAD placement

Matthews et al.[13] studied predictive outcomes of preoperative MELD scores for patients undergoing LVAD placement. Their retrospective study of 211 patients showed that each 5-unit increase in preoperative MELD score was associated with an additional 20 ± 4 units of total blood product exposure (TPBE) in patients undergoing LVAD placement (P < 0.0001). Furthermore, each 5-unit increase in preoperative MELD score increased the odds of operative mortality by 60% (adjusted OR 1.6, 95% CI: 1.1–2.3, P = 0.013).

Further stratification with a MELD score ≥17 showed a decreased survival rate. The risk-adjusted hazard ratio for death in subjects with a MELD score ≥17 was 2.5 (95% CI 1.2–5.3) times that of those with lower scores.

Tricuspid valve surgery

Ailawadi et al.[14] investigated the mortality risk of chronic liver disease, defined by prior diagnosis of cirrhosis or preoperative MELD >15, undergoing tricuspid valve surgery. In a retrospective study, a cohort of 168 patients showed that preoperative MELD >15 predisposes to a significant increase in mortality, as well as higher bypass time (P = 0.027) and multi-organ failure (P = 0.019).

This study utilized a multivariate logistic regression analysis including statistically significant variables, that included MELD >15. The only predictor of mortality in the created model was MELD >15 with an OR of 9.38 (95% CI 2.25–30, P = 0.002). The model created was notable for R² of 0.10 and C statistic of 0.85.

DISCUSSION

Our systematic review has three major findings. First, MELD, MELD-XI, and CPT are mortality predictors in patients undergoing tricuspid valve surgery, cardiac catheterization, pericardiectomy, and LVAD placement. Second, laboratory markers, including INR, creatinine, and albumin, are predictors of bleeding risk and mortality in cardiac catheterization and TAVR. Finally, high-risk MELD groups (defined as MELD >15, 17, or 20) predicted a higher mortality risk, suggesting that certain MELD cutoffs can identify high-risk groups for cardiac procedures, which offers higher clinical utility.

MELD and CPT scores predicted outcomes in cardiovascular procedures. The CPT score was designed to assess the operative risk of portosystemic shunt surgery for variceal bleeding, while the MELD score predicted survival after transjugular intrahepatic shunt portosystemic shunt placement.[14] Newer studies are adapting these risk scores to predict outcomes in cardiovascular procedures, including 30-day mortality.[15] MELD cutoffs greater than 15 and CPT-B and CPT-C were associated with more mortality [Table 1]. Creatinine and albumin were used to predict mortality outcome with patients undergoing AVR.[8] The main complication reviewed was bleeding risk after cardiac catheterization. Bleeding increased with elevated INR although not significantly.[10] Due to the lack of stratification of risk variables in our studies, certain complications, such as hepatic decompensation, were not formally evaluated. One study compared the GI bleed risk in cirrhotic patients with CAD undergoing stent placement, and bleeding risk was not associated with cirrhotic risk factors.[9]

Our study is complementary to the previously studied VOCAL-Penn and Mayo risk scores.[6] These scores were evaluated in surgeries without stratifying by procedure type. These scores can holistically risk-stratify cirrhotic patients; our study highlights the use of alternative scores or laboratory markers (i.e., MELD, MELD-XI, CPT, INR, and creatinine albumin) for risk stratification of specific procedures (i.e., TAVR and cardiac catheterization).

The American Gastroenterological Association (AGA), American College of Gastroenterology (ACG), American Heart Association (AHA), or American College of Cardiology (ACC) does not have set guidelines for risk stratification or management of cirrhotic patients undergoing cardiac procedures specifically. However, the American Association for the Study of Liver Disease (AASLD) provides bleeding risk stratification by procedure type and provides recommendations to minimize procedural bleeding. Few guidelines are available to assist clinicians with patients having cirrhosis with underlying cardiac disease. Although the AGA provides guidelines for anticoagulation for patients with atrial fibrillation, no further recommendations are made regarding the risk of ablations.[16] Although AASLD provides management recommendations for patients with NAFLD, which is associated with cardiovascular disease, no risk factor prognostication is provided when undergoing invasive procedures.[17]

A meta-analysis was not performed due to significant heterogeneity in the data regarding noninvasive parameters, such as MELD, CPT, INR, or other variables, including their cutoff values, related to assessing the mortality risk of cirrhosis patients undergoing cardiac procedures. Heterogeneity stems from differences in patient populations across studies, including age, etiology of cirrhosis, severity of cardiac disease, comorbidities, and other clinical characteristics. Variability in study design, data collection, and statistical methods used in different studies also leads to heterogeneity, making it difficult to synthesize the results coherently, also attributed to differences in how outcomes were reported for short-term and long-term mortality. Given these sources of heterogeneity, it may not be possible to obtain a meaningful summary estimate through a meta-analysis.

We believe that noninvasive parameters are effective in predicting outcomes in patients with cirrhosis undergoing cardiac procedures. A clinical framework, such as described in Figure 2, can aid clinicians and patients in understanding the risks of invasive cardiac procedures and guide decision-making. Easy tools to predict outcomes streamline pre-procedural workup and focus on optimizing patients’ risk before the procedure and minimizing post-procedural complications.

Figure 2.

Proposed workflow to utilize noninvasive risk scores to optimize planning for patient with cirrhosis undergoing cardiac procedures

The inference from the studies is limited by several factors. Most studies were retrospective, which reported only major areas of decompensation. Studies were limited by sample size and follow-up duration. Physician selection bias was present, with some choosing medical management of diseases rather than invasive intervention. Some studies did not provide a definition of cirrhosis, and for some patients, the effect of renal failure was difficult to distinguish from MELD prognostication, given that MELD relies on kidney function. The VOCAL-Penn and Mayo risk scores were not compared to any other risk factors for specific procedures, which can be an outcome of interest in future studies.

Several factors limited our review. We found that it is necessary to ensure that variables used to assess risk were stratified by procedure type. The studies initially screened were not specific to procedure type. For example, when referencing cardiac surgery, the risk factor used (MELD, CPT, etc.) was not stratified to a type of surgery (bypass, surgical valve replacement, etc.). Most studies involved cardiac surgery and its subtypes. Limited studies involved percutaneous interventions including coronary catheterization, arrhythmia ablation, and transcatheter valve replacements. Studies involving heart transplant, LVAD placement, or extracorporeal membrane oxygenation (ECMO) placement were limited in number.

We recommend future studies to assess the risk of cardiac procedures in patients with end-stage liver disease. As the prevalence of diseases, such as NAFLD, rises,[17] patients are expected to undergo further cardiac procedures. Further studies should include more catheter-based procedures and structural and electrophysiologic procedures, including assessments of serum laboratory-level associations. Implementation of newer scores, including MELD 3.0, which aims to improve mortality prediction among cirrhosis patients,[18] should be assessed in patients undergoing cardiovascular procedures.

In conclusion, our systematic review showed that noninvasive clinical parameters can predict mortality risk in patients with cirrhosis undergoing cardiac procedures. Future studies should include more information about the risks of specific cirrhosis complications, such as the development of acute-on-chronic liver failure. This information will aid clinicians implementing optimal practices in patients with advanced liver and cardiac diseases.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.