Carvedilol and nonselective-beta-blockers vs endoscopic ligation for prophylaxis of esophageal variceal hemorrhage in cirrhosis: Systematic review and meta-analysis

Abstract

BACKGROUND

Variceal hemorrhage remains a major contributor to morbidity and mortality among patients with liver cirrhosis. Although primary prophylaxis is strongly recommended, uncertainty persists regarding the relative efficacy of carvedilol and other nonselective beta-blockers (NSBBs) vs endoscopic variceal ligation (EVL).

AIM

To compare the efficacy of carvedilol and other NSBBs with EVL for the primary prophylaxis of variceal bleeding in patients with cirrhosis.

METHODS

A systematic search of PubMed, Cochrane Library, EMBASE, and ScienceDirect (August 2025) was conducted to identify randomized controlled trials and prospective cohorts comparing carvedilol or other NSBBs with EVL. Data were pooled using a random-effects meta-analysis, and relative risk (RR) was used as the summary measure. The primary outcome was first variceal bleeding, while the secondary outcome was bleeding-related mortality.

RESULTS

Five studies (n = 752) were included, involving 380 patients receiving carvedilol/NSBBs and 372 patients undergoing EVL. Carvedilol significantly reduced the risk of first variceal bleeding compared with EVL (RR = 0.64, 95% confidence interval [CI]: 0.44-0.94; P = 0.02, I² = 11%). The absolute bleeding rates were 9.7% with carvedilol vs 15.1% with EVL. Bleed-related mortality did not differ significantly (RR = 0.76, 95%CI: 0.37-1.58; P = 0.47, I² = 4%), although the event numbers were low. Adverse events differed by modality, with carvedilol predominantly associated with systemic intolerance, whereas EVL caused procedure-related complications.

CONCLUSION

Carvedilol and other NSBBs were associated with a lower risk of first variceal bleeding compared with EVL when used as primary prophylaxis in patients with cirrhosis.

Key Words: Cirrhosis; Portal hypertension; Carvedilol; Endoscopic variceal ligation; Primary prophylaxis

Core Tip: This systematic review and meta-analysis demonstrate that carvedilol, a nonselective beta-blocker, is superior to endoscopic variceal ligation for the primary prevention of esophageal variceal bleeding in patients with cirrhosis. Carvedilol significantly decreases the risk of first variceal bleeding compared to endoscopic variceal ligation. It is associated with different adverse event profiles, highlighting its potential as the preferred first-line prophylactic therapy, especially in settings where resources are limited. This evidence supports a paradigm shift in global practice toward greater adoption of nonselective beta-blockers such as carvedilol for primary prophylaxis.

INTRODUCTION

Cirrhosis represents the terminal pathway of chronic liver disease and constitutes a significant global health burden, affecting more than 100 million individuals worldwide and ranking among the top causes of morbidity and mortality[1]. One of the most significant clinical consequences of cirrhosis is the development of portal hypertension, which arises from increased intrahepatic resistance and augmented splenic blood flow. Once portal pressure exceeds a critical threshold, the formation of esophageal and gastric varices becomes almost inevitable[2].

Portal hypertension results from a complex interplay between increased intra-hepatic vascular resistance and elevated portal venous inflow. In liver fibrosis, sinusoidal distortion and capillarization narrow the vascular lumen and substantially increase intra-hepatic resistance. Concurrently, splanchnic vasodilation increases portal blood flow. The combination of these hemodynamic changes results in a sustained elevation of portal pressure, which according to the hemodynamic relationship between vascular resistance and blood flow[3-5].

Splanchnic vasodilation also leads to effective arterial hypovolemia, triggering activation of the renin-angiotensin-aldosterone system and sympathetic nervous system. This neurohumoral response promotes sodium retention, plasma volume expansion, and a hyperdynamic circulatory state characterized by increased cardiac output, further sustaining portal hypertension[6,7]. The resulting chronic elevation in portal pressure drives the development of major complications, such as gastroesophageal varices, ascites, and hepatic encephalopathy, which collectively define advanced portal hypertension in cirrhosis[8,9].

It is estimated that nearly half of patients with cirrhosis will develop varices during the course of their disease. Among these, variceal hemorrhage remains the most life-threatening complication, with a 6-week mortality rate approaching 20% despite modern advances in therapy[10]. Preventing the first variceal bleed is therefore a priority in clinical practice, as an initial bleed is associated not only with a high risk of death but also with a profound negative impact on quality of life and long-term survival.

The pathophysiology of variceal bleeding is closely linked to the severity of portal hypertension, which is commonly quantified by the hepatic venous pressure gradient (HVPG) (Figure 1). Patients with HVPG ≥ 12 mmHg are at substantial risk of variceal rupture[2,11]. For several decades, nonselective beta-blockers (NSBBs) have formed the cornerstone of pharmacological prophylaxis. These agents reduce portal venous inflow through β1-adrenergic blockade, which lowers cardiac output, and β2-adrenergic blockade, which decrease splanchnic vasodilation[12].

Figure 1 Pathophysiology of portal hypertension: Interaction of vascular resistance and blood flow leading to clinical complications. RAAS: Renin-angiotensin-aldosterone system.

Current recommendations advocate beginning NSBBs at a subtherapeutic dose, incrementally increasing the dose at 2-3 days intervals, with a target resting heart rate of 55-60 beats per minute, while preserving a systolic blood pressure of no less than 90 mmHg and adhering to the maximum allowable daily dose[13]. In recent practice guidance, including Baveno VII and the 2023 American Association for the Study of Liver Diseases recommendations, carvedilol has emerged as the preferred NSBBs because of its greater efficacy in reducing HVPG and better tolerability compared to traditional agents such as propranolol. Individuals receiving variceal prophylaxis are generally advised to continue the NSBBs indefinitely. However, those with decompensated cirrhosis should be monitored closely; if the individual develops persistently low systolic blood pressure (i.e. less than 90 mmHg) or spontaneous bacterial peritonitis, the beta-blockers should be temporarily held[14].

Beyond their hemodynamic effects, long-term administration of NSBBs in patients with cirrhosis has demonstrated a substantial reduction in the risk of first variceal bleeding and mortality. Emerging evidence highlight that these benefits extend beyond hemodynamics to include “non-hemodynamic” effects, including reduced bacterial translocation and systemic inflammation – mechanisms that are critical in preventing acute-on-chronic liver failure[15]. Nonetheless, appropriate patient selection remains crucial – particularly in those with advanced decompensation – since NSBBs may paradoxically worsen circulatory dysfunction if systemic perfusion is already severely compromised[16].

In patients with ascites, especially refractory ascites, careful monitoring is required, as NSBBs may blunt the compensatory increase in cardiac output necessary to support renal perfusion. However, recent observational data suggest that NSBBs are generally safe in this population provided that mean arterial pressure remains above 65 mmHg and renal function is preserved. These findings challenge the earlier concept of a strict “therapeutic window” and support an individual approach, with treatment guided by frequent assessment of blood pressure, renal profile, and signs of hypoperfusion rather than routine discontinuation.

Carvedilol, in particular, has emerged as a promising agent because, in addition to these classical effects, it possesses α1-adrenergic antagonism properties. This dual mechanism reduces intrahepatic vascular tone and produces a more potent fall in HVPG compared with propranolol[17]. Clinical studies have shown that carvedilol achieves higher rates of hemodynamic response, and it is increasingly being considered as the preferred NSBB in guidelines[2,17].

Carvedilol is the preferred NSBBs in recent practice guidance (including Baveno VII and American Association for the Study of Liver Diseases 2023) because of its greater efficacy in reducing HVPG and better tolerability compared to traditional agents like propranolol. Individuals receiving variceal prophylaxis continue the NSBBs indefinitely. Persons with decompensated cirrhosis should be monitored closely; if the individual develops persistently low systolic blood pressure (i.e. less than 90 mmHg) or spontaneous bacterial peritonitis, the beta-blockers should be temporarily held[2,14].

In parallel, endoscopic variceal ligation (EVL) has become the standard endoscopic intervention for primary prophylaxis. By mechanically eradicating varices through band placement, EVL reduces the risk of bleeding[18]. Its principal advantage lies in providing an immediate anatomical solution; however, it requires repeated endoscopic sessions, incurs procedural costs, and carries a risk of complications, such as post-banding ulcers, chest pain, and dysphagia[19]. In contrast, NSBBs are inexpensive, widely available, and avoid invasive procedures, but adherence is frequently limited by side effects, including hypotension, fatigue, dizziness, bradycardia or impotence[20], which may lead to poor adherence or discontinuation.

The comparative efficacy of carvedilol vs EVL remains a subject of debate. Early randomized trials, such as that by Tripathi et al[18], suggested carvedilol may be more effective than EVL in preventing a first variceal bleed. In contrast, Shah et al[19] found similar bleeding rates between the two approaches but with differences in side-effect profiles. Studies using propranolol by Lay et al[20], demonstrated comparable efficacy between pharmacological therapy and EVL, underscoring that both strategies are valid first-line options. Observational studies, notably from Reiberger et al[17], further suggest that carvedilol may confer broader benefits by reducing hepatic decompensation beyond variceal bleeding. However, heterogeneity in outcome observed in secondary prophylaxis trials has limited the generalizability of these findings[21].

International consensus statements, including the Baveno VII guidelines, recognize both EVL and NSBBs as appropriate interventions for primary prophylaxis, with a tendency to favor carvedilol due to its superior hemodynamic effects[2]. Nonetheless, these recommendations stop short of recommending one strategy over the other as universally superior, reflecting the ongoing uncertainty in the field. Furthermore, while carvedilol is gaining traction, concerns about tolerability in decompensated patients remain, particularly due to risks of hypotension[22,23]. On the other hand, EVL requires endoscopic expertise and may not be readily accessible in resource-limited settings, which further complicates decision-making[10].

Taken together, existing evidence indicates that both carvedilol and EVL are effective for the primary prophylaxis of variceal bleeding, yet, important questions remain regarding their relative efficacy, safety, and feasibility for different patient populations. The available randomized and cohort studies are limited by small sample sizes and heterogenous results, creating a clear evidence gap. Therefore, a meta-analysis comparing carvedilol (as a representative NSBB with enhanced portal pressure-lowering activity) with EVL is warranted to clarify their relative roles in the primary prophylaxis of variceal hemorrhage in patients with cirrhosis. Such an analysis may inform clinical guidelines, optimize patient outcomes, and direct future research toward more individualized management of portal hypertension.

MATERIALS AND METHODS

This systematic review and meta-analysis adhered to the standards set forth by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). To strengthen transparency and ensure reproducibility, the protocol was formulated prior to study initiation. The review question was framed using the Population (Patient/Problem), Intervention, Comparison, and Outcome (PICO) framework: (1) Adult patients with cirrhosis and esophageal varices (population); (2) Carvedilol or other NSBBs (intervention); (3) EVL (comparator); and (4) Outcomes including first variceal bleeding, bleed-related mortality, and treatment-related complications (outcomes). Only studies of primary prophylaxis were included; trials limited to secondary prophylaxis or studies on variceal progression without bleeding outcomes were excluded. The protocol was prospectively registered in the International Prospective Register of Systematic Reviews (Registration No. CRD42024542285).

Search strategy

A comprehensive literature search was performed in these major electronic databases: (1) PubMed/MEDLINE; (2) Cochrane Library; (3) EMBASE; and (4) ScienceDirect. The searches encompassed all available records up to August 2025. This search strategy combined Medical Subject Headings (MeSH) and free-text terms: (“Liver Cirrhosis”[Mesh] OR “decompensated cirrhosis” OR “advanced cirrhosis” OR “hepatic cirrhosis”) AND (“Esophageal and Gastric Varices”[Mesh] OR “esophageal varices” OR “variceal bleeding” OR “variceal hemorrhage”) AND (“Carvedilol”[Mesh] OR carvedilol OR “non-selective beta blocker”) AND (“Esophageal Varices/surgery”[Mesh] OR “Endoscopic Band Ligation” OR EBL OR “endoscopic therapy” OR “endoscopic variceal ligation”). No language restrictions were applied. A manual search of the reference lists from relevant articles and review papers was also performed to retrieve any additional studies that qualified for inclusion.

Study selection

Two reviewers independently screened titles and abstracts for eligibility. Full-text articles were reviewed to confirm their inclusion. Discrepancies were resolved through discussions and consensus. Eligible studies were randomized controlled trials (RCTs) or prospective cohort studies that directly compared carvedilol or other NSBBs with EVL in the primary prophylaxis of esophageal variceal bleeding. Studies exclusively focused on secondary prophylaxis, small varices progression without bleeding endpoints, or those with placebo comparators were excluded. Studies evaluating propranolol were included, as it belongs to the NSBB class and shares portal pressure-lowering mechanisms with carvedilol.

Data extraction

Data from each study were independently extracted by two reviewers using standardized forms. The extracted variables included study author, year of publication, country, study design, population size, baseline characteristics, type and dose of intervention, comparator details, follow-up duration, and outcomes. The following outcomes were extracted: (1) First variceal bleeding; (2) Bleeding-related mortality; and (3) Treatment-related complications. Hazard ratios (HRs) with 95% confidence interval (CI) were extracted when available. For dichotomous outcomes, event counts were collected and transformed into odds ratios.

Risk of bias assessment

The methodological quality of included studies was independently assessed by two reviewers. The Cochrane Risk of Bias 2.0 tool was used for RCTs, while the Newcastle-Ottawa Scale was used for cohort studies. The assessed domains included sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other potential sources of bias. Discrepancies in quality assessment were resolved by consensus.

Statistical analyses

Outcomes were pooled using a random-effects model (DerSimonian-Laird method) to account for clinical and methodological heterogeneity. HRs were log-transformed and analyzed separately from odds ratios when appropriate, with standard errors derived from the reported 95%CI. The degree of heterogeneity was quantified using the I² statistic, where values of 25%, 50%, and 75% were considered indicative of low, moderate, and high heterogeneity, respectively. When at least 10 studies were available, publication bias was examined through visual inspection of funnel plots and Egger’s test. Subgroup analyses were pre-specified by type of NSBB (carvedilol vs propranolol) and study design (RCT vs cohort). Sensitivity analyses were performed by sequentially excluding each of the studies.

RESULTS

Study selection

The electronic database search yielded 125 potentially relevant articles (PubMed: 26; Cochrane Library: 31; EMBASE: 10; ScienceDirect: 58). Following removal of 11 duplicates (3 identified and excluded), 114 unique records remained for screening. After evaluating titles and abstracts, a majority were excluded because they were irrelevant to the research question, primarily due to addressing unrelated interventions (e.g., transjugular intrahepatic portosystemic shunt [TIPS], sclerotherapy, or other pharmacological agents) or involving secondary prophylaxis cohorts. Eight full-text articles were retrieved for detailed assessment. Among these, three were excluded: (1) One compared carvedilol with placebo in patients with small varices and thus did not report bleeding outcomes; (2) Another focused exclusively on secondary prophylaxis; and (3) The third used a non-comparative design. In total, five studies met the eligibility criteria and were included in the quantitative synthesis (four RCTs and one prospective cohort study), as illustrated in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram (Figure 2).

Figure 2 Diagram Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow of literature search strategy for systematic reviews and meta-analyses shows the number of articles (n) in the different selection stages of the selection process.

Characteristics of included studies

The five studies collectively enrolled 752 patients with cirrhosis and medium to large esophageal varices who had not yet experienced a variceal hemorrhage. Of these, 380 patients received carvedilol or other NSBB therapy, while 372 underwent EVL. The studies varied in geographic origin. Tripathi et al[18] and Tripathi et al[21] were multicenter randomized RCTs from the United Kingdom. Shah et al[19] was a multicenter trial in Pakistan, Lay et al[20] was conducted across three centers in Taiwan, and Reiberger et al[17] was a prospective cohort from Austria.

Although all studies addressed primary prophylaxis, methodological differences were evident. Tripathi et al[18] enrolled patients with grade II varices and followed them for a median of 20 months. Tripathi et al[21] included patients with medium-large varices with a median follow-up of 12 months; however, it was stopped early due to recruitment challenges during the coronavirus disease 2019 pandemic. Shah et al[19] followed patients for an average of 13 months and incorporated a high proportion of viral cirrhosis, whereas Lay et al[20] recruited mostly hepatitis-related cirrhosis with a longer mean follow-up of nearly three years. Reiberger et al[17] adopted an HVPG-guided approach, enrolling patients initially treated with propranolol but switched to carvedilol if they failed to achieve hemodynamic response, offering a unique insight into the utility of carvedilol in NSBB non-responders.

Baseline disease severity varied across studies. While most patients were Child-Pugh class A or B, the proportion of patients with compensated cirrhosis was higher in Lay et al[20] and Shah et al[19]. In contrast, Reiberger et al[17] and Tripathi et al[21] included a subset with decompensation at baseline, typically initiated at 6.25 mg and titrated to 12.5 mg daily. EVL was generally performed at intervals of 2-4 weeks until variceal eradication, followed by surveillance endoscopies. A detailed summary of study characteristics is presented in Table 1[17-21].

Table 1 Characteristics and results of the included studies.

Ref.	Country/setting	Study design	Population (n)	Cirrhosis stage	Intervention (carvedilol/other nonselective beta-blocker)	Comparator (EVL)	Follow-up duration	Bleeding	Lower CI	Higher CI	Bleed related mortality	Lower CI	Higher CI	Bleeding (carvedilol, number/total)	Bleeding (EBL, number/total)	Bleeding-related mortality (carvedilol, number/total)	Bleeding-related mortality (EBL, number/total)
Shah et al[19]	Pakistan (3 tertiary centers: Aga Khan University, Dow University, Jinnah Postgrad Medical Centre, Karachi)	Multicenter RCT	168 cirrhotic patients with medium-large varices (82 carvedilol, 86 EVL)	Mostly viral cirrhosis (approximately 90%); hepatitis C virus most common (approximately 73% carvedilol, 76% EVL); Child-Pugh mean 7.3 (B/C in approximately 55%); ascites approximately 40%	Carvedilol 6.25 mg daily, increase to 12.5 mg/day after 1 week, maintained	EVL sessions every 3 weeks until variceal eradication, then 6-monthly surveillance	Mean follow-up 13.3 ± 12.1 months (range 1-50)	1.61 (HR)	0.27	9.69	1.53	0.71	3.3	7/82	6/86	4/82	4/86
Lay et al[20]	Taiwan (China Medical University Hospital, Taichung Veterans Gen Hospital, Taipei Veterans Gen Hospital)	Prospective RCT	100 cirrhotic patients with high-risk esophageal varices (50 EVL, 50 propranolol)	Mostly viral hepatitis (approximately 72%-74%), 20%-22% alcoholic, others 6%; Child-Pugh A: Approximately 45%, Child-Pugh B: Approximately 40%, Child-Pugh C: Approximately 15%	Propranolol titrated from 40 mg twice daily, mean 68 mg/day; dose adjusted to 20%-25% HR reduction	EVL (weekly × 3, then every 2-3 weeks until eradication; approximately 3.2 sessions, approximately 9.8 bands)	Mean follow-up approximately 35 months	0.89 (odds ratio)	0.36	2.18	1.23	0.5	3.04	11/50	12/50	5/50	4/50
Reiberger et al[17]	Austria (Vienna Hepatic Hemodynamic Lab + Hietzing Hospital)	Prospective cohort with HVPG-guided treatment	104 cirrhotic patients with hepatic venous pressure gradient > 12 mmHg and varices	Alcoholic 55%, viral 33%, metabolic dysfunction-associated steatohepatitis 10%; Child-Pugh A/B/C: 66/30/8	Carvedilol 6.25-50 mg/day (median 12.5 mg), titrated by blood pressure/heart rate, used in propranolol non-responders; 38 patients	EVL (29 patients) every 2-4 weeks until eradication; surveillance every 6 months	mean 19.5 ± 9.7 months (up to 2 years)							3/38	7/29	1/38	5/29
Tripathi et al[18]	UK (5 centers: Edinburgh, Glasgow, Paisley, Birmingham)	Multicenter RCT	152 cirrhotic patients with grade II + esophageal varices (77 carvedilol, 75 VBL)	Alcoholic liver disease 73%; median age 54 years; median Child-Pugh 8 (A/B/C: Approximately 38%/25%/37%); ascites approximately 50%	Carvedilol 6.25 mg/day increase to 12.5 mg if tolerated	Variceal band ligation every 2 weeks until eradication, then surveillance	Median follow-up 20 months (range up to 50 months)	0.41 (HR)	0.19	0.96	1.98	0.59	6.59	8/77	18/75	2/77	2/75
Tripathi et al[21]	United Kingdom, 52 sites	Multicentre, randomized, controlled, open-label trial	265 cirrhotic patients with medium-large esophageal varices (133 carvedilol, 132 VBL)	mean age 59.6 years; 68.3% male; median Child-Pugh 5 (A: Approximately 79%, B: Approximately 21%); model for end-stage liver disease 8; alcohol-related disease 46.4%; 24% with decompensation at baseline	Carvedilol 12.5 mg once daily	Variceal band ligation as per British Society of Gastroenterology guidelines	12 months (median follow-up 1 year)	0.58 (HR)	0.24	1.41				8/133	13/132

CI: Confidence interval; EVL: Endoscopic variceal ligation; HR: Hazard ratio; RCT: Randomized controlled trial; VBL: Variceal band ligation; EBL: Endoscopic band ligation.

Quality assessment

Overall, the RCTs showed low to moderate risk of bias, mainly related to blinding. The observational studies demonstrated moderate methodological quality according to the Newcastle-Ottawa Scale.

Primary outcome: Risk of first variceal bleeding

All five studies reported first bleeding events, which constituted the primary endpoint of the meta-analysis. In total, 37 bleeding events occurred among 380 patients receiving carvedilol (9.7%), compared with 56 events among 372 patients treated with EVL (15.1%). Pooled analysis demonstrated that carvedilol significantly reduced the risk of first variceal bleeding, with a 36% relative risk (RR) reduction (RR = 0.64, 95%CI: 0.44-0.94; P = 0.02) (Figure 3A)[17-21]. Importantly, heterogeneity was low (I² = 11%, P = 0.34), indicating consistent findings across trials despite methodological and population differences.

Figure 3 Pooled result. A: Pooled result for the risk of variceal bleeding between carvedilol and endoscopic variceal ligation groups; B: Pooled result for the bleed-related mortality of variceal bleeding between carvedilol and endoscopic variceal ligation groups. EBL: Endoscopic band ligation.

Examining individual trials provides further context for the pooled effect. Tripathi et al[18], one of the earlier RCTs, found a significantly lower risk of first bleed with carvedilol (10% vs 23%; HR = 0.41, 95%CI: 0.19-0.96), although this benefit was attenuated in the per-protocol analysis. Shah et al[19] observed no significant difference between carvedilol and EVL (8.5% vs 6.9%; HR = 1.61, 95%CI: 0.27-9.69), but finding likely influenced by limited by limited smaller sample size and low number of bleeding events. Lay et al[20], which compared propranolol with EVL, reported comparable bleeding rates between treatment arms (22% vs 24%); however, the extended follow-up highlighted equivalence of pharmacological and endoscopic prophylaxis strategy. In contrast, Reiberger et al[17] demonstrated a marked advantage of carvedilol, with rebleeding rates of 8% compared with 24% for EVL (P = 0.043), particularly in HVPG responders. Finally, Tripathi et al[21], although underpowered due to early termination, showed a trend favoring carvedilol (3.8% vs 7.6%; HR = 0.46, P = 0.159), consistent with the overall meta-analysis.

Taken together, the consistent direction of effect across trials, combined with the low heterogeneity, supports the conclusion that carvedilol is at least as effective as EVL and may offer a favorable option for the prevention of first variceal bleeding in cirrhosis.

Secondary outcome: Bleed-related mortality

Four studies (Reiberger et al[17], Tripathi et al[18], Shah et al[19], and Lay et al[20]) reported mortality specifically attributable to variceal bleeding. Across these studies, 12 bleeding-related mortality were recorded among 247 patients receiving carvedilol/NSBBs (4.8%), compared with 15 deaths among 240 patients treated with EVL (6.3%). Pooled analysis revealed no statistically significant difference in bleed-related mortality (RR = 0.76, 95%CI: 0.37-1.58; P = 0.47) (Figure 3B)[17-20]. Statistical heterogeneity was negligible (I² = 4%, P = 0.37).

Although the analysis was underpowered to detect modest differences, the overall pattern of results is noteworthy. Lay et al[20] reported comparable rates of bleeding-related death rates between propranolol and EVL (10% vs 8%), a finding mirrored by Shah et al[19], who observed nearly identical mortality rates (4.8% vs 4.6%). Tripathi et al[18] reported slightly higher bleeding-related mortality in the carvedilol group (3% vs 2%); however, the absolute number of events was too small to draw firm conclusions. Tripathi et al[21] did not provide bleed-specific mortality, thereby limiting broader comparisons. Pre-specified subgroup analyses according to NSBB type and study design could not be meaningfully performed because of the limited number of included studies.

Collectively, these findings suggest that although carvedilol reduces the incidence of first bleeding, this effect does not clearly translate into lower bleeding-related mortality. This likely attributable to the low overall event rates and rescue therapies such as TIPS are widely available.

Other outcomes

Adverse event reporting was variable across studies, precluding quantitative synthesis but providing important qualitative insights. Carvedilol was associated with systemic side effects, such as hypotension, dizziness, bradycardia, and fatigue, leading to discontinuation in up to 10%-20% of patients in some trials (Tripathi et al[18], Shah et al[19], and Tripathi et al[21]). In contrast, EVL-related adverse events were largely procedure-specific and included post-banding ulcer bleeding, chest pain, and dysphagia (Tripathi et al[18] and Lay et al[20]). In Tripathi et al[21], adverse events occurred in 46% of patients on carvedilol, with 11% discontinuing treatment, compared to 10% dysphagia and one serious adverse event in the EVL group.

These contrasting safety profiles highlight the trade-offs between pharmacological and endoscopic prophylaxis. Carvedilol offers a non-invasive and inexpensive but may not be tolerated in all patients, while EVL, although invasive, is generally well tolerated, albeit requiring repeated procedures and specialized resources.

Sensitivity analyses employing a leave-one-out approach showed no substantial change in the pooled estimates, indicating the robustness of the findings. Formal assessment of publication bias was not feasible due to the limited number of included studies.

DISCUSSION

This meta-analysis provides one of the most comprehensive syntheses to date comparing carvedilol and EVL for the primary prophylaxis of esophageal variceal hemorrhage in patients with cirrhosis. Across five controlled studies involving 752 patients, carvedilol significantly reduced the risk of first variceal bleeding compared to EVL, with a pooled RR reduction of 36%. Importantly, this effect was consistent across diverse study populations from Asia, Europe, and South Asia, with low statistical heterogeneity, underscoring the robustness of the finding. The magnitude of benefit with carvedilol is clinically meaningful, as the absolute risk reduction of approximately 5% translates to a number needed to treat of about 20 to prevent one bleeding episode.

In contrast, bleed-related mortality was not significantly different between the two interventions. This is not unexpected, given the relatively low number of fatal variceal bleeds in modern practice and the availability of rescue therapies such as TIPS. Nevertheless, the prevention of first bleeding is itself a critical achievement, as initial variceal hemorrhage carries substantial risks of decompensation, rebleeding, and death over the long term[1,2].

Taken together, the findings highlight that carvedilol is not only non-inferior but likely comparable to EVL as a strategy for primary prophylaxis of variceal bleeding. By addressing both the structural and dynamic components of portal hypertension, carvedilol appears to offer broader protection than a purely anatomical intervention such as EVL.

Mechanistic interpretation

The observed advantage of carvedilol is rooted in its unique pharmacologic properties. Unlike traditional NSBBs such as propranolol or nadolol, carvedilol combines nonselective beta-adrenergic blockade with α1-adrenergic antagonism[17,24,25]. This dual action produces a greater reduction in HVPG, which is the strongest predictor of variceal bleeding risk. Reductions of ≥ 20% from baseline or to < 12 mmHg are considered hemodynamic responses, and carvedilol achieves these targets in a larger proportion of patients compared with propranolol[2,17,26].

Carvedilol decreases portal inflow through its β1-blockade and β2-blockade while simultaneously lowering intrahepatic vascular resistance via α1-blockade. This dual effect addresses both the dynamic (vasoconstriction) and fixed (fibrosis, sinusoidal distortion) contributors to portal hypertension[25,27]. In contrast, EVL eradicates existing varices but does not alter the underlying portal hypertension. Varices frequently recur, and patients remain at risk of bleeding from new varices or from gastric varices, which are not amenable to banding[18].

Moreover, portal hypertension drives not only variceal bleeding but also ascites, spontaneous bacterial peritonitis, and hepatic encephalopathy[10]. By directly reducing portal pressure, carvedilol has the potential to influence the broader natural history of cirrhosis, whereas EVL is limited to the prevention of a single complication. This broader disease-modifying effect may explain why carvedilol is increasingly favored in international guidelines and why observational studies suggest benefits extending beyond variceal hemorrhage[27].

Beyond its hemodynamic effects, carvedilol may exert broader disease-modifying actions that are relevant across different etiologies of cirrhosis, including chronic hepatitis C. Experimental and clinical data suggest that NSBBs, particularly carvedilol, can reduce bacterial translocation, systemic inflammation, and potentially fibrogenesis, all of which contribute to portal hypertension and decompensation regardless of whether the underlying cause is viral, alcoholic, or metabolic[28,29]. These pleiotropic effects provide a biologic rationale for the survival benefits observed in cohort studies of patients with compensated cirrhosis and clinically significant portal hypertension who receive long-term carvedilol therapy[30].

Etiology remains crucial, however, when balancing the benefits and risks of carvedilol vs EVL. In hepatitis B-related cirrhosis, etiologic control with long-term antiviral therapy reduces but does not abolish the risk of clinically significant portal hypertension and variceal bleeding, particularly in patients with established advanced fibrosis or cirrhosis at the time of diagnosis[30]. For these patients, carvedilol offers an attractive option for primary prophylaxis because of its potent portal pressure-lowering effect and once-daily or twice-daily oral dosing, while EVL remains an important alternative when there are contraindications or intolerance to NSBBs[30,31].

In hepatitis C-related cirrhosis, patients often present at a relatively younger age and may have fewer cardiometabolic comorbidities than those with alcohol-associated or metabolic-associated cirrhosis, which can enhance tolerability of carvedilol and allow uptitration to hemodynamically effective doses[32,33]. Even after sustained virologic response with direct-acting antivirals, many hepatitis C virus (HCV) cirrhosis patients retain clinically significant portal hypertension and established varices, so primary prophylaxis with carvedilol or EVL remains indicated according to current guidance. In this setting, carvedilol not only reduces portal pressure and variceal bleeding risk but may also decrease decompensation and hepatocellular carcinoma incidence over time, whereas EVL provides a non-systemic, procedure-based option for patients with asthma, severe bradyarrhythmias, or hypotension that limit NSBB use[31,34].

In non-viral, metabolically driven cirrhosis such as metabolic dysfunction-associated steatotic liver disease/metabolic dysfunction-associated steatohepatitis, patients frequently have multiple comorbidities, including obesity, diabetes, hypertension, and cardiovascular disease, which complicate the risk-benefit assessment for carvedilol[6,14,35]. While these patients still benefit from portal pressure reduction with NSBBs, the higher baseline prevalence of hypotension, autonomic dysfunction, and polypharmacy may necessitate more cautious dose titration and closer monitoring, making EVL a reasonable initial strategy in selected high-risk or frail individuals. Thus, in metabolic cirrhosis, individualized decisions that integrate cardiovascular risk, blood pressure profile, and local endoscopic expertise are essential when choosing between carvedilol, EVL, or a combined approach for primary prophylaxis of variceal hemorrhage[1,8,36].

Clinical implications

The clinical implications of these findings are highly relevant to current practice. Carvedilol represents a non-invasive, cost-effective, and widely accessible option for primary prophylaxis. Its oral administration allows for immediate initiation, even in community or resource-limited settings where endoscopic facilities may not be readily available[10,37]. This is particularly pertinent in low- and middle-income countries, where cirrhosis burden is high but access to endoscopy is restricted[38].

At the same time, carvedilol is not without limitations. Hypotension, dizziness, fatigue, and bradycardia are common side effects, and a proportion of patients require dose reduction or discontinuation[18-20]. In the study of Tripathi et al[21], nearly one-fifth of patients required dose adjustment and 11% discontinued therapy due to intolerance. These side effects may limit carvedilol’s use in patients with advanced decompensation, refractory ascites, or low baseline blood pressure[20,39].

EVL, while invasive, has its own advantages. It is highly effective in eradicating varices, well tolerated in most patients, and not limited by systemic hemodynamic effects. In patients with contraindications to NSBBs, such as asthma or severe bradyarrhythmias, EVL remains indispensable. Moreover, in centers with established endoscopic expertise, EVL can be performed safely with low complication rates. Thus, rather than viewing the two strategies as mutually exclusive, they should be considered complementary, with therapy individualized according to patient profile, tolerability, and healthcare resources.

The etiology of cirrhosis also plays an important role in selecting the optimal prophylactic strategy. In patients with cirrhosis due to hepatitis C, alcohol, or mixed etiologies, the risk of portal hypertension and variceal bleeding remains high, so the need for primary prophylaxis with carvedilol or EVL persists even after etiologic therapy has been introduced[30,40]. Clinical trials and meta-analyses comparing carvedilol and EVL generally include populations with a substantial proportion of alcoholic and hepatitis C-related cirrhosis, allowing their results to be extrapolated to these subgroups, with appropriate adjustment for the degree of decompensation and comorbidities[41,42].

In hepatitis C-related cirrhosis specifically, viral eradication with direct-acting antivirals can improve liver function and reduce portal pressure, but does not consistently abolish the risk of bleeding once clinically significant portal hypertension is established[32,42,43]. In patients with HCV-cirrhosis with established varices, primary prophylaxis therefore remains recommended according to current guidelines, with the choice between carvedilol and EVL guided by tolerability, baseline blood pressure, and endoscopic availability. Moreover, the use of NSBBs in HCV-related cirrhosis has been associated with a reduced risk of decompensation and a potential decrease in hepatocellular carcinoma incidence, providing long-term benefits that extend beyond variceal bleeding prevention alone[44,45].

Comparison with existing literature

Our findings align with the most recent Baveno VII consensus, which endorses either NSBBs or EVL as one of the prophylaxis, while emphasizing carvedilol as the preferred NSBB due to its superior hemodynamic efficacy[2]. Previous meta-analyses that pooled NSBBs and EVL generally concluded similar efficacy between the two approaches[30,46,47]. However, many of these earlier reviews were dominated by propranolol studies, potentially underestimating the benefit of carvedilol. By focusing specifically on carvedilol, this analysis provides a more contemporary and clinically relevant synthesis.

The individual trial results illustrate the variability that has historically contributed to uncertainty. Tripathi et al[18] reported significantly fewer first bleeds with carvedilol than with EVL, while Shah et al[19] found similar bleeding rates but noted differences in side-effect profiles. Lay et al[20] using propranolol rather than carvedilol, showed equivalent efficacy to EVL, reinforcing the class effect of NSBBs. Reiberger et al[17] demonstrated clear advantages of carvedilol in HVPG responders, while Tripathi et al[21], though underpowered, showed a consistent direction of effect favoring carvedilol.

Recent pharmacological reviews also highlight carvedilol’s superiority over traditional NSBBs in reducing HVPG and its broader benefits in delaying decompensation[27,30]. Thus, the weight of evidence increasingly supports carvedilol not only as an acceptable alternative to EVL but as the more effective prophylactic strategy for most patients.

Clinical practice emphasizes patient education regarding adherence, home blood pressure measurement, and recognition of hypotensive symptoms, as these factors substantially influence outcomes. Multidisciplinary follow-up – combining hepatologist supervision with nursing support – enhances adherence and safety, particularly when titrating carvedilol[16,48,49]. Overall, NSBB therapy remains a cornerstone of primary prophylaxis against variceal hemorrhage and should be continued as tolerated, balancing between portal pressure reduction and preservation of systemic perfusion to improve survival[3,31,50].

Limitations

This systematic review and meta-analysis have several important limitations that should be acknowledged. First, the total number of included studies was limited, and although the pooled sample size exceeded that of any individual trial, it remains modest relative to the global burden of cirrhosis. This limitation precluded meaningful subgroup analyses according to NSBB type or study design and also rendered formal assessment of publication bias infeasible because fewer than ten studies were available. Second, heterogeneity in study design and patient populations may have influenced the pooled estimates, despite low statistical heterogeneity for the primary bleeding outcome. Notably, one study evaluated propranolol rather than carvedilol, and another employed an HVPG-guided cohort design rather than strict randomization. Third, adverse events were not consistently defined or systematically reported across studies, preventing quantitative synthesis of treatment tolerability. Fourth, the largest and most recent trial was prematurely terminated due to the coronavirus disease 2019 pandemic, limiting its statistical power. In addition, bleeding-related mortality events were infrequent, and the absence of a significant difference between treatment strategies likely reflects limited power rather than true equivalence. Finally, the use of a DerSimonian-Laird random-effects model in a small number of studies may have overestimated precision, and the certainty of evidence was not formally assessed using the GRADE approach. Taken together, these limitations warrant cautious interpretation of the findings.

Future directions

Future studies should aim to address these gaps with adequately powered, multicenter RCTs. Standardized reporting of outcomes, including bleeding, mortality, adverse events, and quality of life, will be critical for more precise comparisons. Subgroup analyses stratified by cirrhosis etiology, baseline HVPG, and degree of decompensation could help identify which patients benefit most from carvedilol vs EVL. Moreover, as carvedilol may influence other complications of portal hypertension, future trials should extend beyond bleeding to include ascites, encephalopathy, and survival endpoints[27,51,52].

Additionally, cost-effectiveness analyses and patient-reported outcome measures should be incorporated, as these factors are increasingly important in clinical decision-making. With growing evidence favoring carvedilol, guideline committees may consider recommending it as the default strategy for primary prophylaxis, reserving EVL for patient’s intolerant to NSBBs or with contraindications. Ultimately, the goal should be to develop an individualized, patient-centered approach that maximizes efficacy, minimizes harm, and optimizes resource use in diverse healthcare settings.

CONCLUSION

This meta-analysis suggests that carvedilol and other NSBBs are associated with a lower risk of first variceal hemorrhage compared with EVL for primary prophylaxis in patients with cirrhosis. Despite no significant difference in bleeding-related mortality was observed, the reduction in first bleeding represents a clinically relevant benefit. Given the limited number of studies and heterogeneity in study design, these findings should be cautiously interpreted. Differences in safety profiles characterized by systemic adverse effects related to NSBBs and procedure-related complications associated with endoscopic ligation, highlight the importance of individualized treatment decisions. Future large-scale randomized trials should confirm these findings and explore the broader disease-modifying potential of carvedilol beyond bleeding prevention.