Association of non-selective β blockers with the development of renal dysfunction in liver cirrhosis: a systematic review and meta-analysis

Abstract

Background & Aims

Non-selective β blockers (NSBBs) may negatively influence renal function through decreasing heart rate and cardiac output. This study aimed to systematically investigate their association.

Methods

PubMed, EMBASE, and Cochrane library databases were searched to identify all relevant studies evaluating the association of NSBBs with renal dysfunction in cirrhotic patients. Unadjusted and adjusted data were separately extracted. Odds ratios (ORs) and hazard ratios (HRs) were pooled. Subgroup meta-analyses were performed according to the proportions of ascites and Child-Pugh class B/C and the mean model for end-stage liver disease (MELD) score. Quality of evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation framework.

Results

Fourteen studies were finally included. Based on unadjusted data, NSBBs significantly increased the risk of developing renal dysfunction (OR = 1.49; p = 0.03), and this association remained significant in subgroup analyses of studies where the proportions of ascites was >70% and Child-Pugh class B/C was 100%. Based on adjusted data with propensity score matching (adjusted OR = 0.61; p = 0.08) and multivariable regression modelling (adjusted HR = 0.86; p = 0.713), NSBBs did not increase the risk of developing renal dysfunction, and this association remained not significant in subgroup analyses of studies where the proportions of ascites was >70% and <70%, the proportion of Child-Pugh class B/C was <100%, and the mean MELD score was <15. The quality of evidence was very low for all meta-analyses.

Conclusions

NSBBs may not be associated with the development of renal dysfunction in liver cirrhosis. However, more evidence is required to clarify their association in specific populations.

KEY MESSAGES

• Non-selective β blockers (NSBBs) may negatively influence renal function through decreasing heart rate and cardiac output in liver cirrhosis.

• Our meta-analysis failed to support the association of NSBBs with an increased risk of developing renal dysfunction after covariate adjustment.

Keywords:

• Nonselective β blockers
• propranolol
• acute kidney injury
• hepatorenal syndrome
• liver cirrhosis

1. Introduction

Non-selective β blockers (NSBBs) are the mainstay treatment options for the primary and secondary prevention of oesophageal variceal bleeding in cirrhotic patients [1–3]. Besides, its clinical benefits have been clearly observed in terms of prevention from the development of decompensated events [4] and bacterial infection [5]. However, the safety of NSBBs in liver cirrhosis with refractory ascites has been questioned [6]. Additionally, its use seems to be associated with an increased risk of developing portal vein thrombosis [7,8].

Renal dysfunction, such as acute kidney injury (AKI) and hepatorenal syndrome (HRS), commonly occurs in cirrhotic patients with an incidence of 20–42% [9,10] and a prevalence of 14–50% [11]. Such a wide range in epidemiology may be related to the heterogeneity in the severity of liver cirrhosis among studies [9,12]. Renal dysfunction has a negative impact on post-transplant renal function in liver cirrhosis [13] and 6-fold increased risk of in-hospital death [14]. The development of renal dysfunction in liver cirrhosis is primarily secondary to the progression of cirrhosis, development of gastrointestinal bleeding, diarrhoea, or infection, and use of vasodilators, diuretics, non-steroidal anti-inflammatory drugs, or radiocontrast agents [10,15]. Additionally, it has been reported that cardiac insufficiency should be associated with the development of renal dysfunction in decompensated cirrhosis [16], particularly in cirrhotic patients with ascites [17,18] and spontaneous bacterial peritonitis (SBP) [19]. Considering the property of NSBBs for reduction of cardiac output and heart rate by blocking β₁ receptors, it may be hypothesised that NSBBs deteriorate renal function by influencing systemic hemodynamics.

Herein, we have systematically reviewed the current evidence and conducted a meta-analysis to clarify the impact of NSBBs on the development of renal dysfunction in liver cirrhosis.

2. Methods

This study was performed according to the preferred reporting guidelines for systematic reviews and meta-analyses (PRISMA) statement.

2.1. Registration

The PROSPERO registration number of this study was CRD42021226044.

2.2. Literature search

The PubMed, EMBASE, and Cochrane library databases were searched to identify all studies which reported the incidence of renal dysfunction (AKI, HRS, and unspecified renal dysfunction) in cirrhotic patients treated with and without NSBBs. Search items were ‘(liver cirrhosis) or (cirrhotic)’ and ‘(acute kidney injury) or (hepatorenal syndrome) or (renal dysfunction)’ and ‘(non-selective beta blockers) or (propranolol) or (carvedilol) or (nadolol)’. The latest search was done on August 4, 2023.

2.3. Study selection

The study selection was performed by two investigators (XX and TW). Disagreements between them were resolved by further discussion. There was no language limitation. All eligible observational and interventional studies should evaluate the association of NSBBs with renal dysfunction in cirrhotic patients. If some procedures or operations that significantly influenced hemodynamic changes were performed during the use of NSBBs, they would not be considered. Exclusion criteria were as follows: (1) duplicates; (2) reviews or meta-analyses; (3) case reports; (4) letters, editorials, or comments; (5) experimental studies; (6) conference abstracts; (7) studies that were irrelevant to study population, intervention, or comparator; (8) studies that did not evaluate the risk of renal dysfunction as an outcome of interest; and (9) studies that did not exclude patients with AKI at baseline.

2.4. Data extraction

Unadjusted data were derived from studies without covariate adjustment. Adjusted data were derived from studies with propensity score matching and multivariable regression modelling. The characteristics of studies were independently extracted by two investigators (XX and TW), as follows: first author, publication year, country, study design, enrolment period, target population, follow-up duration, types of NSBBs drugs in NSBBs groups, interventions in control groups, types of renal dysfunction, number of patients treated with and without NSBBs who developed and did not develop renal dysfunction, indication of NSBBs, proportion of patients with ascites, proportion of patients with Child-Pugh class B/C, and model for end-stage liver disease (MELD) score.

2.5. Study quality assessment

The study quality was independently assessed by two investigators (XX and FG). If the assessment was inconsistent between them, a consensus would be reached through further discussion with another investigator (TW). The quality of cohort and case-control studies were assessed by using the Newcastle-Ottawa Scale (NOS) [20], which includes eight questions in three domains as follows: 1. Selection, 2. Comparability, and 3. Outcome. The highest NOS score should be nine points. Good quality would be considered, if the score at the domain of Selection, Comparability, and Outcome was ≥3 points, ≥1 point, and ≥2 points, respectively. Fair quality would be considered, if the score at the domain of Selection, Comparability, and Outcome was 2 points, ≥1 point, and ≥2 points, respectively. Poor quality would be considered, if the score at the domain of Selection, Comparability, and Outcome was ≤ 1 point, 0 point, or ≤1 point, respectively. The risk of bias of randomized controlled trials (RCTs) was assessed by using the Cochrane Collaboration’s tool [21], which includes seven domains as follows: 1. random sequence generation, 2. allocation concealment, 3. blinding of participants and personnel, 4. blinding of outcome assessment, 5. incomplete outcome data, 6. selective reporting, and 7. other bias. The risk of bias for each domain is divided into low, high, and unclear.

2.6. Definition

Decompensated cirrhosis was defined as Child-Pugh class B/C and/or decompensated events, including ascites, SBP, or variceal bleeding.

2.7. Statistical analyses

Review Manager version 5.3 (Cochrane collaboration, the Nordic Cochrane Centre, Copenhagen, Denmark) and Stata version 12.0 (Stata Corp, Texas, USA) were used for statistical analyses. Only a random-effects model allowing the true effect size varied from study to study was employed [22]. Odds ratios (ORs) and hazard ratios (HRs) with 95% confidence intervals (CIs) were pooled. I² and p value were calculated to assess heterogeneity among studies, and I²>50% and/or p < 0.1 were suggestive of statistically significant heterogeneity. Leave-one-out sensitivity analyses were performed by omitting each study. Meta-regression analyses were performed with covariates, including the proportions of patients with ascites and Child-Pugh class B/C, MELD score, study design, publication year, region, and sample size. If MELD score was expressed as median with range, its mean with standard deviation would be estimated [23]. Subgroup meta-analyses were performed according to the proportion of patients with ascites <70% and >70%, the proportion of patients with Child-Pugh class B/C < 100% and 100%, and the mean MELD score <15 and >15. The interaction between subgroups was tested. Publication bias was evaluated by funnel plot and Egger’s test, and p < 0.1 was considered statistically significant publication bias.

2.8. Meta-analyses quality assessment

The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework was used to determine the quality of evidence in the meta-analysis. The quality of the evidence was categorized as strong, moderate, low, and very low according to the presence of risk of bias, inconsistency, imprecision, indirectness, and other considerations, including publication bias, large effect, plausible confounder, and dose response gradient [24]. A summary of findings table was created by GRADEpro GDT online version.

3. Results

3.1. Study selection

Among the 308 studies initially retrieved from PubMed, EMBASE, and Cochrane library databases, 50 duplicate references, 134 reviews or meta-analyses, 14 case reports, 27 conference abstracts, 17 letters, editorials, or comments, 4 experimental studies, and 47 studies with no data of interest. Thus, 15 studies were potentially eligible [25–39], furthermore, one cohort study was excluded, because it enrolled patients with AKI at baseline [25]. Finally, 14 studies, which reported the data regarding the association of NSBBs with the development of renal dysfunction in cirrhotic patients, were included [26–39] (Figure 1).

Figure 1. Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram of study selection.

3.2. Study characteristics

The study characteristics are summarised in Table 1. The sample size ranged from 61 to 3075. The publication year ranged from 2014 to 2023. Two studies were conducted in Austria [27,32], two Denmark [28,30], two Korea [31,34], two India [36,38], one Belgium [29], one China Taiwan [35], one Greece [26], one Russia [37], one United Kingdom [39], and one the United States of America [33].

Table 1. Study characteristics.

First author (year)	Country	Study design	Enrollment period	Target population	Follow-up duration	NSBBs group	Control group	Types of renal dysfunction
Mandorfer (2014)	Austria	Retrospective cohort	2006–2011	Cirrhosis with ascites and SBP	90 days	Propranolol and carvedilol	No NSBBs	HRS/AKI stages 2/3
Kimer (2015)	Denmark	Retrospective cohort	May 2010– Jan 2014	Cirrhosis with refractory ascites	NA	Propranolol and metoprolol	No NSBBs	HRS
Sersté (2015)	Belgium	Retrospective cohort	Jul 2006–Jul 2014	Adults with cirrhosis and severe, biopsy-proven alcoholic hepatitis	168 days	Propranolol	No NSBBs	AKI/HRS type I (defined by ICA and ADQI)
Bang (2016)	Denmark	Retrospective cohort	1995–2010	Decompensated cirrhosis with ascites and history of paracentesis	2 years	Propranolol	No NSBBs	HRS (ICD-10: K767)
Kalambokis (2016)	Greece	Retrospective cohort	Jan 2002–Dec 2012	Child-Pugh class B/C with small varices without red signs	3 years	Propranolol	No NSBBs	HRS
Kim (2017)	Korea	Nest case-control	Jan 1990–Dec 2010	Adults with cirrhosis who were waitlisted for LT	18.2 months (median)	Propranolol and nadolol	No NSBBs	AKI stages 2/3
Scheiner (2017)	Austria	Retrospective cohort	May 2004– Aug 2012	Cirrhosis with clinically significant portal hypertension	32.9 months (mean)	Propranolol and carvedilol	No NSBBs	AKI stages 2/3 (defined by ICA 2015)
Ngwa (2020)	USA	Retrospective cohort	Jan 2012– Jun 2012	Adults with cirrhosis who were referred and evaluated for LT	282 days (median)	Propranolol, carvedilol, and nadolol	No NSBBs	AKI stages 1/2/3 (defined by ICA 2015)
Yoo (2020)	Korea	Retrospective cohort	Jan 2005– Dec 2016	Child-Pugh class B/C with ascites received primary prevention of esophageal VH	42.1 months (median)	Propranolol + EVL	EVL	HRS (defined by ICA 2015)
Li (2021)	China, Taiwan	Retrospective cohort	2006–2016	Adults with cirrhosis	>4 years	Propranolol, carvedilol, nadolol, and timolol	No NSBBs	AKI (defined by KDIGO)
Maiwall (2021)	India	Retrospective cohort	Aug 2012– Jul 2016	Cirrhosis	379 days (median)	Propranolol and carvedilol	No NSBBs	AKI
Korobka (2022)	Russia	Retrospective cohort	NA	Cirrhosis with ascites and varices who were waitlisted for LT	46.8 months (median)	Propranolol, carvedilol, and nadolol + EVL	EVL	AKI
Singh (2022)	India	RCT	Jul 2015–Dec 2016	Cirrhosis with ≥ grade 2 ascites and varices requiring VH prophylaxis	12 months	Propranolol	EVL	AKI stages 1/2/3
Tittanegro (2023)	United Kingdom	Retrospective cohort	Jan 2016–Jun 2019	Decompensated cirrhosis	3–15 days of hospitalization	Propranolol and carvedilol	No NSBBs	Renal dysfunction

Abbreviations: ADQI: acute disease quality initiative; AKI: acute kidney injury; EVL: oesophageal variceal ligation; HRS: hepatorenal syndrome; ICA: international club of ascites; ICD: international classification of diseases; KDIGO: kidney disease improving global outcomes; LT: liver transplantations; NA: not available; NSBBs: non-selective β blockers; RCT: randomised controlled trial; SBP: spontaneous bacterial peritonitis; VH: variceal haemorrhage.

Twelve studies were cohort studies [26–30,32–37,39], one was a case-control study [31] that matched the patients by age, serum creatinine (sCr), and MELD-Na [31], and one was a RCT [38].

The follow-up duration was from 90 days to four years in 12 studies [26,27,29–38] and 3-15 days of hospitalisation in one study [39], but unknown in the remaining one [28].

Type of NSBBs was reported in all included studies, including propranolol alone in five studies [26,29,30,34,38], and propranolol in a combination with carvedilol, nadolol, timolol, or metoprolol in nine other studies [27,28,31–33,35–37,39].

Diagnostic criteria of renal dysfunction were made according to the Acute Disease Quality Initiative [29], Kidney Disease Improving Global Outcomes [35], or International Club of Ascites [29, 32–34]. One study evaluated the development of renal dysfunction [39]. Nine studies evaluated the development of AKI [27,29,31–33,35–38], of which five evaluated AKI stage [27,31–33,38] and four did not provide any information regarding AKI stage [29,35–37]. Six studies evaluated the development of HRS [26–30,34], of which one evaluated HRS type I [29] and five did not provide any information regarding type of HRS [26–28 30,34].

3.3. Study quality

Ten observational studies were of good quality [26–28,30,33–37,39], but the others were of poor quality [29,31,32] (Supplementary Table 1). One RCT had a high risk of bias in the domain of blinding of participants and personnel and low risks of bias in six other domains [38].

3.4. Patient characteristics

Six studies included patients with both decompensated and compensated cirrhosis [29,31–33,35,36], and eight included patients with decompensated cirrhosis alone, including unclassified decompensation event, ascites, SBP, variceal bleeding, or Child-Pugh class B/C [26–28,30,34,37–39].

The indication of NSBBs was primary prophylaxis of variceal bleeding in three studies [26,34,38], both primary and secondary prophylaxis of variceal bleeding in six studies [27,29,32,33,36,37], and unclear in five studies [28,30,31,35,39].

The proportion of cirrhotic patients with ascites was 100% in six studies [27,28,30,34,37,38] and 31.3%-89.4% in eight studies [26,29,31–33,35,36,39]. The proportion of cirrhotic patients with Child-Pugh class B/C was 100% in four studies [26,34,37,38], 45.5%-96.7% in four studies [27,32,33,36], and unclear in six studies [28–31,35,39]. The mean MELD score ranged 12.1–27.3 in nine studies [27–29,31–33,36,38,39], but was unclear in five studies [26,30,34,35,37].

3.5. Unadjusted data

Fourteen cohorts from 13 studies reported unadjusted data [26–34,36–39] (Supplementary Table 2). Meta-analysis demonstrated that NSBBs significantly increased the risk of developing renal dysfunction (unadjusted OR = 1.49, 95%CI = 1.05–2.12; p = 0.03). The heterogeneity was statistically significant (I²=73%; p < 0.00001) (Figure 2). Sensitivity analysis did not find the source of heterogeneity (Supplementary Table 3). Meta-regression analyses demonstrated that the heterogeneity was related to publication year (before 2015 vs. after 2015; p = 0.025) and sample size (>200 vs. <200; p = 0.007), rather than the proportions of patients with ascites (>70% vs. <70%; p = 0.486) and Child-Pugh class B/C (100% vs. <100%; p = 0.184), mean MELD score (>15 vs. <15; p = 0.499), study design (observational studies vs. RCT; p = 0.126), or region (Asia vs. America vs. Europe vs. Oceania; p = 0.347). Publication bias was not statistically significant (Supplementary Figure 1, p = 0.349).

Figure 2. Meta-analyses regarding the association of NSBBs with developing renal dysfunction based on unadjusted data.

The association of NSBBs with an increased risk of developing renal dysfunction remained in subgroup analyses of studies where the proportion of patients with ascites was >70% [26–28,30,33,34,36–38] (unadjusted OR = 1.57, 95%CI = 1.05–2.34; p = 0.03) and the proportion of patients with Child-Pugh class B/C was 100% [26,34,37,38] (unadjusted OR = 2.33, 95%CI = 1.25–4.36; p = 0.008), but not in those where the proportion of patients with ascites was <70% [29,31,32,39] (unadjusted OR = 1.37, 95%CI = 0.54–3.48; p = 0.50), the proportion of patients with Child-Pugh class B/C was <100% [27,32,33,36] (unadjusted OR = 1.55, 95%CI = 0.89–2.71; p = 0.12), and mean MELD score was <15 [31,32,36,38] (unadjusted OR = 1.31, 95%CI = 0.97–1.79; p = 0.08) and >15 [27–29, 33, 39] (unadjusted OR = 2.17, 95%CI = 0.88-5.37; p = 0.09) (Figure 2). The interaction between subgroups was not statistically significant according to the proportion of ascites (p = 0.80), the proportion of Child-Pugh class B/C (p = 0.34), and the mean MELD score (p = 0.30). The quality of evidence was very low for unadjusted data (Supplementary Table 4).

3.6. Adjusted data

Nine cohorts from seven studies reported adjusted data [29–32,34,35,39] (Supplementary Table 2). Based on the three cohorts from three studies where propensity score matching was employed [34,35,39], meta-analysis did not demonstrate that NSBBs significantly increased the risk of developing renal dysfunction (adjusted OR = 0.61, 95%CI = 0.35-1.07; p = 0.08). The heterogeneity was not statistically significant (I²=19%; p = 0.29). Subgroup analyses were not further performed due to limited data available in the included three studies [34,35,39].

Based on the six cohorts from four studies where multivariable regression modelling was employed [29–32], meta-analysis did not demonstrate that NSBBs significantly increased the risk of developing renal dysfunction (adjusted HR = 0.86, 95%CI = 0.38–1.94; p = 0.713). The heterogeneity was statistically significant (I²=85.4%; p < 0.001) (Figure 3). Sensitivity analysis did not find the source of heterogeneity (Supplementary Table 3). Meta-regression analyses demonstrated that the heterogeneity was not related to the proportion of patients with ascites (>70% vs. <70%; p = 0.628), mean MELD score (>15 vs. <15; p = 0.551), publication year (before 2015 vs. after 2015; p = 0.423), region (Asia vs. America vs. Europe vs. Oceania; p = 0.915), or sample size (>200 vs. <200; p = 0.341). The association of NSBBs with the risk of developing renal dysfunction remained not significant in subgroup analyses of studies where the proportion of patients with ascites was >70% [30,31] (adjusted HR = 1.08, 95%CI = 0.33–3.55; p = 0.895) and <70% [29,31,32] (adjusted HR = 0.64, 95%CI = 0.14–2.95; p = 0.571), the proportion of patients with Child-Pugh class B/C was <100% [32] (adjusted HR = 0.50, 95%CI = 0.20–1.25; p = 0.142), and mean MELD score was <15 [32] (adjusted HR = 0.50, 95%CI = 0.20–1.25; p = 0.142); by comparison, NSBBs significantly increased the risk of developing renal dysfunction in the subgroup analysis of one study where the mean MELD score was >15 [29] (adjusted HR = 2.27, 95%CI = 1.48–3.48; p < 0.001) (Figure 3). The interaction between subgroups was statistically significant according to the mean MELD score (p = 0.004), but not the proportion of ascites (p = 0.60). The quality of evidence was very low for adjusted data (Supplementary Table 4). Notably, the quality of evidence for subgroup analyses that only included one study was not assessed.

Figure 3. Meta-analyses regarding the association of NSBBs with developing renal dysfunction based on adjusted data with multivariable regression modelling.

4. Discussion

Discontinuation of NSBBs has been clearly indicated in the event of renal dysfunction. A post-hoc study showed that the development of HRS was one of statistically significant predictors of NSBBs discontinuation [40]. Another multicentre observational study also revealed that the development of AKI during hospitalization determined the discontinuation of NSBBs in cirrhotic patients with ascites [41]. Accordingly, the recent EASL guideline recommends that NSBBs should be stopped once AKI developed in cirrhosis [2], and Baveno VII consensus also recommends that NSBBs should be dose-reduced or discontinued in cirrhotic patients with ascites when systolic blood pressure <90 mmHg, mean arterial blood pressure (MAP) <65 mmHg, and/or HRS-AKI develops [3]. Meanwhile, both of them suggest that NSBBs can be re-titrated or re-initiated once blood pressure returns to baseline and/or HRS-AKI resolves [2,3]. However, the impact of NSBBs as a precipitating factor for the development of renal dysfunction has not been acknowledged by any practice guideline or consensus. Indeed, a recent study has demonstrated that NSBBs slow sympathetic overdrive, reduce cardiac output, and decrease renal perfusion pressure below a critical threshold of 65 mmHg, thereby impairing renal function in cirrhotic patients with refractory ascites [42]. By contrast, another study did not find a negative effect of NSBBs on renal function in patients with decompensated cirrhosis [39]. This discrepancy among studies should be mainly attributed to some following characteristics. First, all patients had refractory ascites in the study by Téllez et al. [42], and 68% of patients had ascites in the study by Tittanegro et al. [39]. Second, the follow-up duration is 4 weeks in the study by Téllez et al. [42], but 3-15 days of hospitalization in the study by Tittanegro et al. [39]. Third, the information regarding use of other drugs except for NSBBs is unavailable in the study by Téllez et al. [42], but all patients received human albumin infusion in the study by Tittanegro et al. [39], which would improve renal function [43]. Fourth, renal dysfunction is evaluated by significant changes in eGFR, serum cystatin C, serum creatinine, and renal hemodynamics in the study by Téllez et al. [42], but a certain threshold in serum creatinine in the study by Tittanegro et al. [39].

To the best of our knowledge, only a previous meta-analysis, which was published in 2018, explored the outcome of cirrhotic patients with ascites treated with and without NSBBs. Based on only four studies, it did not demonstrate any association of NSBBs with the development of HRS in cirrhotic patients (OR = 1.22; p = 0.67) [44]. By comparison, in our current meta-analysis, the type of renal dysfunction explored was more comprehensive, including AKI, HRS, and unspecified renal dysfunction. Based on a total of 14 studies, we demonstrated that NSBBs significantly increased the risk of developing renal dysfunction in cirrhotic patients in the meta-analysis of unadjusted data alone, rather than adjusted data.

Patients with liver cirrhosis exhibit hyperdynamic circulation characterised by high cardiac output and heart rate to counteract splanchnic and systemic vasodilatation. However, with progression of liver cirrhosis, high-output heart failure develops, which is characterised by decreased cardiac output that is insufficient to maintain systemic and renal perfusion [10,45]. In this setting, the use of NSBBs further influences systemic hemodynamics through deceasing heart rate [46–48], cardiac output [47, 48,], cardiac index [49], and MAP [46–48] as well as increasing systemic vascular resistance [47,49], thereby aggravating renal injury. Taken together, NSBBs may have a potential impact on the development of renal dysfunction. However, this side effect may be transient, because the decrease in heart rate and systolic arterial pressure can be reversed after the discontinuation of NSBBs [50]. Furthermore, the use of NSBBs do not directly affect renal blood flow [46,49] and renal vascular resistance [49].

Considering the impact of AKI on worse outcomes [51], there are several considerations for assessing the underlying risk of development of renal dysfunction in cirrhotic patients treated with NSBBs in routine clinical practice. First, among patients with ascites, NSBBs responders, who are characterised as a reduction of hepatic venous pressure gradient to >20% from baseline or <12mmHg, have significantly lower odds of HRS than non-responders (OR = 0.27) [52]. Thus, non-response to NSBBs may carry a higher risk of renal dysfunction. Second, treatment with NSBBs is associated with an increase in sCr concentration and a high frequency of severe AKI in SBP patients with a MAP <65mmHg [53]. Additionally, a renal perfusion pressure <65mmHg is considered the most valuable parameter for the failure of renal autoregulation [54]. Thus, a low MAP and/or renal perfusion pressure during NSBBs intake may carry a higher risk of renal dysfunction. Third, NSBBs induce a higher risk of developing AKI in patients with acute‑on‑chronic liver failure during a 28- or 60-day follow-up period, but not a 90-day follow-up period [55]. Thus, a close monitoring of renal dysfunction should be emphasised in early stage of using NSBBs. Fourth, it has been reported that carvedilol significantly increased glomerular filtration rate and renal blood flow [56], decreased renal vascular resistance [56], and had little impact on sCr concentration [56] and sCr clearance [48,57] in cirrhotic patients compared with propranolol. Thus, the type of NSBBs probably influences its impact on renal dysfunction.

Our study had several limitations. First, a relatively small number of studies were included. Most of them were observational with substantial heterogeneity, which affected the reliability of our statistical results. We attempted to obtain conservative results by using a random-effects model and explore the source of heterogeneity by performing subgroups analyses, sensitivity analyses, and meta-regression analyses. However, we acknowledged that the quality of evidence for our findings according to GRADE framework was very low. Second, the association of NSBBs with renal dysfunction was not the primary objective in some of the included studies and the type of renal dysfunction was varied among studies, which may cause potential heterogeneity in outcome measurement. Third, the information regarding type, dosage, and duration of NSBBs was very limited. Therefore, their impacts on the development of renal dysfunction could not be further explored by subgroup analyses. Fourth, the volume of ascites was not available in all included studies and only one study reported that all included patients had refractory ascites [28]. Furthermore, only two studies including patients with pure decompensated cirrhosis provided adjusted data with propensity score matching [34,39], and only one study including patients with pure decompensated cirrhosis provided adjusted data with multivariable regression modelling [30]. Therefore, due to limited data, we could not perform further subgroup analyses according to the characteristics of ascites and the stage of liver disease. Fifth, some confounders, such as use of diuretics and non-steroidal anti-inflammatory drugs or occurrence of variceal bleeding during NSBBs intake, might influence the risk of developing renal dysfunction, but could not be sufficiently adjusted in our meta-analyses.

5. Conclusions

In our systematic review and meta-analysis,NSBBs did not increase the risk of developing renal dysfunction in liver cirrhosis after adjusting for confounding factors, however, it should be acknowledged that the heterogeneity in patient characteristics among studies, such as compensated vs. decompensated cirrhosis or diuretic-responsive vs. refractory ascites, would lead to uncertainty in the statistical results. Therefore, future high-quality studies are required to elucidate their association in specific populations. Nonetheless, the use of NSBBs should be more cautious in more advanced cirrhosis. Future studies should identify which group of NSBBs users is at a high risk of developing renal dysfunction and explore how to monitor and prevent the risk of developing renal dysfunction during the use of NSBBs.

Authors’ contributions

Conception and design: Xiangbo Xu and Xingshun Qi;

Collection and assembly of data: Xiangbo Xu, Fangbo Gao, and Ting Wang;

Data analysis and interpretation: Xiangbo Xu, Fangbo Gao, Ting Wang, Zuyao Yang, and Xingshun Qi;

Manuscript writing: Xiangbo Xu, Fangbo Gao, Ting Wang, Zuyao Yang, Qingchun Zhao, and Xingshun Qi;

Final approval of manuscript: Xiangbo Xu, Fangbo Gao, Ting Wang, Zuyao Yang, Qingchun Zhao, and Xingshun Qi;

Supervision: Qingchun Zhao and Xingshun Qi.

Acknowledgments

None.

Disclosure statement

Xiangbo Xu, Fangbo Gao, Ting Wang, Zuyao Yang, Qingchun Zhao, and Xingshun Qi declare that they have no conflict of interest.

Data availability statement

Not applicable.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.