Background and aims Carvedilol is effective in the primary prophylaxis for large oesophageal varices. We investigated its use in preventing progression of small to large oesophageal varices.
Methods Consecutive cirrhotics with small oesophageal varices were prospectively randomised to either carvedilol (n=70) or placebo (n=70) and followed up for a minimum of 24 months. Endoscopy was done at baseline and six monthly intervals. Hepatic vein pressure gradient (HVPG) was measured at baseline and at 12 months. The primary endpoint was development of large varices.
Results Baseline characteristics in two groups were comparable. The predominant aetiology of cirrhosis was non-alcoholic fatty liver disease in both the groups. The mean dose of carvedilol administered was 12±1.67 mg/day and the target heart rate achieved was 58±3 bpm. A higher proportion of patients in carvedilol group had non-progression to large varices than placebo (79.4% vs 61.4%; p=0.04); the mean time of non-progression to large varices was 20.8 months (95% CI 19.4 to 22.4) in carvedilol group and 18.7 months (95% CI 17.1 to 20.4) in placebo group (p=0.04). There was a modest reduction of HVPG at 1 year in carvedilol group (−8.64%) compared with placebo (+0.33%) (p=0.22). None of the patients in either group died of variceal bleeding or liver-related causes. No major adverse events were observed in either group.
Conclusions Carvedilol is safe and effective in delaying the progression of small to large oesophageal varices in patients with cirrhosis.
Trial registration number NCT01196507; post-results.
- PORTAL HYPERTENSION
- OESOPHAGEAL VARICES
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
One of the important predictors of variceal haemorrhage is the size of varices, with the highest risk of first haemorrhage occurring in patients with large varices with red colour signs (15%–30% at 2 years).
Non-selective β-blockers like propranolol have not been shown to be effective in preventing the progression of small to large varices in meta-analysis.
Carvedilol, a potent non-cardioselective vasodilating β-blocker, with intrinsic anti-α1-adrenergic activity produces more significant reduction in hepatic vein pressure gradient (HVPG) than propranolol and has been shown to be as effective as endoscopic variceal ligation in primary prophylaxis of variceal bleeding in large oesophageal varices.
What are the new findings?
Carvedilol achieved a reduction in the proportion of patients progressing to large oesophageal varices over a treatment and follow-up period of 24 months compared with placebo.
Carvedilol slowed the time to progression of small to large varices from 18.7 months (in placebo group) to 20.8 months.
There was a modest reduction of 8.64% in the HVPG at 1 year from baseline with the use of carvedilol compared with 0.33% increase with placebo, although this was not statistically significant.
How might it impact on clinical practice in the foreseeable future?
This is the first randomised controlled trial showing the efficacy of carvedilol in early primary prophylaxis by delaying the progression from small to large varices in patients with cirrhotic portal hypertension.
In cirrhotics with small varices and mild-to-moderate hyperdynamic circulation, HVPG reduction by 8%–10% though not significant is helpful. Use of carvedilol should be considered in these patients.
Oesophageal variceal haemorrhage is one of the major complications of portal hypertension. It occurs in about one third of the cirrhotics during lifetime and accounts for nearly 80% of all bleeding episodes. The mortality of acute variceal haemorrhage has decreased significantly due to improved medical and endoscopic management but still ranges from 15% to 20%.1
The annual incidence of development of oesophageal varices in patients with cirrhosis and no varices is approximately 5%.2 The risk of bleeding is approximately 7% at 2 years in patients with small (≤5 mm) varices, and 5%–23% of these patients develop larger varices (≥5 mm) at 1 year.3 ,4 Alcoholic cirrhosis, Child-Pugh's grade B or C and presence of red wale markings at the time of initial endoscopy are the main factors associated with the progression from small to large varices.5 Variceal haemorrhage occurs at a yearly rate of 5%–15%, and the most important predictor of haemorrhage is the size of varices, with the highest risk of first haemorrhage occurring in patients with large varices (15%–30% at 2 years).6 It seems, therefore, quite logical to prevent or delay the growth of small oesophageal varices to reduce the incidence of acute bleed and overall mortality to large size by reducing portal pressure.
Development of oesophageal varices, their growth and subsequent risk of variceal haemorrhage is determined by portal pressure, assessed by hepatic venous pressure gradient (HVPG). Use of non-selective β-blockers (NSBBs) is the recommended therapy for reducing portal pressure and for primary prophylaxis against variceal haemorrhage in cirrhotic patients with large oesophageal varices.7 Early reduction of portal pressure could ameliorate the development of portal hypertension-related complications and reduce portosystemic shunting.8 Although NSBBs are ineffective in preventing development of varices, they may be effective in a subgroup of patients with small varices with high risk of variceal progression or bleed. In this regard, two previous studies using propranolol and nadolol for early primary prophylaxis (preventing growth of varices) reported contradicting results.9 ,10 A recent meta-analysis comprising six randomised controlled trials (RCTs) has shown that NSBBs like propranolol are not effective in preventing the progression of small to large varices.11
Carvedilol, a non-cardioselective vasodilating β-blocker, with intrinsic anti-α1-adrenergic activity use may lead to more significant reduction in HVPG than propranolol and has been shown to be as effective as endoscopic variceal ligation (EVL) in prophylaxis for large oesophageal varices.12 It has also been suggested that carvedilol may have a potential cardiovascular organ protection effect mediated by its antioxidant activity and by its ability to inhibit vascular smooth muscle cell proliferation.13 Carvedilol is rapidly absorbed achieving peak plasma concentrations between 1 and 2 hours after oral administration.14 Thus, carvedilol by combining β-blocker and anti-α1-adrenergic activities has been documented to have the potential of achieving greater reductions of portal pressure than propranolol.15
However, there is scarcity of data to demonstrate the efficacy of carvedilol in preventing the progression of small oesophageal varices. In the present study, we have compared carvedilol with placebo for early primary prophylaxis to prevent the progression of small oesophageal varices in cirrhotic patients.
Patients and methods
Patients with cirrhosis presenting to the Department of Hepatology, Institute of Liver and Biliary Sciences (ILBS), from November 2010 to December 2012 were consecutively screened and those fulfilling the criteria were enrolled in 1:1 ratio into the randomised placebo-controlled clinical trial. Patients eligible for the trial included cirrhotics with small oesophageal varices (≤5 mm in diameter) without any history of previous variceal bleeding.16 Cirrhosis was diagnosed on clinical, radiological or laboratory parameters and/or liver biopsy. The exclusion criteria were: age <18 years or >75 years, non-cirrhotic portal hypertension, Child-Turcotte-Pugh score >12, refractory ascites, acute kidney injury, history of prior EVL or sclerotherapy, history of surgery for portal hypertension, significant cardiopulmonary comorbidity, uncontrolled diabetes, peripheral vascular disease, presence of any malignancy that significantly affects survival, evidence of alcoholic hepatitis or active alcohol abuse with last intake ≤1 month or refusal to participate in the study. A period of 1 month was based on studies where short-term mortality is assessed at 28 days, often with abstinence. Furthermore, it has been shown that the immunological injury starts to decline within 28 days from the last drink.17 At each endoscopy, we had determined the size of the oesophageal varices during the withdrawal of the endoscope, with adequate air insufflation of the oesophagus and stomach. The size of the varices was carefully assessed as small (<5 mm) or large (>5 mm) and, if in doubt, it was measured against the gap between the jaws of the open endoscopic biopsy forceps (5 mm). If varices were smaller than the open biopsy forceps, they were identified as small varices. They were further graded as grade I and II based on the presence of varices in one or both phases of respiration and with full air insufflation or not. Multiple endoscopic pictures for assessment of grade of oesophageal varices were recorded at each endoscopy session. The interobserver variability in the assessment of the size of oesophageal varices was determined. In case of any disagreement, the recorded images were retrieved and reviewed by a third endoscopist.
The study protocol conformed to the Declaration of Helsinki of 1975, revised in 1983 and was approved by the Institutional Ethics Committee (file no: F.1/8/22/AC/admin/ILBS/2010/2902). The study followed the Consolidated Standards of Reporting Trials (CONSORT) guidelines for RCTs and was registered at the ClinicalTrials.gov with full protocol access (identifier: NCT01196507). An informed written consent was taken from each patient at the time of enrolment. All authors had access to the study data and had reviewed and approved the final manuscript.
Patients were randomised to receive either carvedilol or placebo. An independent statistician generated allocation sequence. Using block randomisation method with the block size of 10, randomisation was done. We intended for a double-blind trial design, which minimises bias in assessment of outcome but implies evident difficulties in the management of dose adjustments that are to be done in relation to heart rate during follow-up, or to use a single-blind design and limit the blindness to the subject. We decided, in agreement with the ethics committee, to use a single-blind design, also because it was considered unrealistic that blindness could be kept with a drug with an evident effect on heart rate. Enrolment of participants, assessing eligibility and obtaining consent was carried out by one of the two authors (AB or CKK).
Carvedilol was administered orally at a start dose of 3.125 mg twice daily. This was gradually increased and at 1 week, it was increased to 6.25 mg twice daily and if required, further increased up to a maximum dose of 12.5 mg twice daily with a caution that the systolic blood pressure did not fall below 100 mm Hg and heart rate <55 bpm. Patients were asked to coordinate with the primary care physician to monitor the heart rate and blood pressure and also to keep a diary to report any adverse effects.
The placebo tablets were identical to the carvedilol tablets. First, the patients were given placebo tablets in a dose of one tablet twice a day. Then, this was gradually increased to two tablets, three tablets and finally four, twice a day in the next 3 weeks. Heart rate and blood pressure recording was done in each visit. Since the maximum dose of carvedilol was meant to be 25 mg per day or a maximum of eight tablets per day, we did not increase the placebo tablets to more than this. Patients were asked to report any adverse effects.
The initial clinic visit was 1 week after initiation of the protocol and then at 6, 12, 18 and 24 months. It was decided that every patient must complete at least three endoscopic assessments during the follow-up. During each follow-up visit, detailed clinical examination, complete blood counts, kidney and liver function tests were done. Upper GI endoscopy (UGIE) was done at baseline and at 6–9 monthly intervals. The involved endoscopists in the study were well trained and experienced in diagnostic and advanced therapeutic endoscopic procedures. The schedule of the follow-up endoscopies is described in online supplementary table S1. Two endoscopists saw the images each time and noted their findings. In case of any discordance in the findings, the stored endoscopic videos were studied in detail by a third endoscopist to decrease the interobserver variability. Second, all endoscopists were kept unaware of the treatment arm to which the patient was randomised. HVPG was done at baseline and at 1 year of follow-up. Compliance to the drugs was assessed through pill count and periodic telephonic conversations every fortnightly. Patient data were censored if lost to follow-up, withdrew from the study, died or underwent a liver transplantation. After recruitment of the last patient, the follow-up was continued in both the treatment arms.
Hepatic venous pressure gradient
HVPG was measured in the Hepatic Hemodynamic Laboratory, by introducing a 7 Fr. Swan-Ganz catheter via the transfemoral approach into a major hepatic vein as described earlier.18 The catheter was advanced until it was wedged into the hepatic vein. The occluded position of the catheter was checked by the absence of reflux after the injection of 1–2 mL of contrast medium and appearance of a sinusoidogram. A mean of three readings was taken. All HVPGs were done by well-trained hepatologists in the study, who were blinded to the patients' clinical data and did not follow-up in outpatient department.
Primary and secondary outcomes
The primary endpoint was the non-progression from small to large oesophageal varices at 2 years. The secondary endpoints were: (i) reduction in HVPG at 1 year, (ii) survival at 24 months and (iii) treatment-related adverse events. HVPG non-responders were defined as those with HVPG reduction of <10% compared with baseline and those with ≥20% reduction from baseline were defined as responders. The partial responders were considered with HVPG reduction between 10% and 20%.
Treatment after failure to achieve primary endpoint
Patients developing large oesophageal varices in either group during the treatment protocol were subjected to EVL as per the departmental protocol. These patients underwent repeat UGIE every 3 weeks to assess for EVL after the initial EVL session, until variceal eradication was achieved. They were followed up with 3–6 monthly once surveillance to see for variceal recurrence. Bleeding from oesophageal varices was to be treated by a combination of medical and endoscopic treatment.
Assuming that non-progression to large varices in carvedilol group is 80% and 50% in placebo.10 With α as 5%, power 90% we need to include 114 cases. Further taking 20% attrition rate, it was decided to enrol 70 cases in each arm and allocated randomly with block randomisation method with block size taken as 10. Baseline parametric data were expressed as the proportion, mean±SD and median with IQR. The intention-to-treat and per-protocol analyses were also carried out. The differences in the groups were analysed using χ2 test, unpaired t test or Mann–Whitney test as appropriate. Paired t test was used to compare the change before and after the intervention. The cumulative probability of development of large varices and overall survival was determined by the Kaplan–Meier plots and compared by the log-rank test. The agreement between the two endoscopists was seen by κ statistics. The change in the heart rate and mean arterial pressure measurements was compared between the two treatment groups at different time intervals using repeated measures followed by post hoc comparison by least square method.
p Value <0.05 was considered statistically significant. No additional analyses were done. All statistical tests were performed using SPSS for Windows V.22 (IBM, Armonk, New York, USA).
A total of 157 consecutive cirrhotics with small oesophageal varices were screened. One hundred and forty patients who fulfilled the inclusion and exclusion criteria were randomised in two groups; 70 patients were assigned to carvedilol group and equal number to placebo group (figure 1). The two study groups were comparable with respect to demographic characteristics, clinical, laboratory data and severity scores (table 1). The most common aetiology of cirrhosis was cryptogenic followed by alcohol and viral-related cirrhosis in the two groups as shown in table 1. Patients with alcoholic liver disease were strictly abstinent from alcohol for at least >4 weeks to avoid any possibility of underlying alcoholic hepatitis. Patients with hepatitis B-related cirrhosis were on either tenofovir 300 mg or entecavir 0.5 mg orally once daily if their HBV DNA was ≥2000 IU/mL as they were compensated cirrhosis (APASL guidelines 201519). Patients with hepatitis C-related cirrhosis had received a combination of PEG-IFN α-2a/2b and weight-based ribavirin for 24/48 weeks. All patients had undergone baseline UGIE before enrolment and underwent repeat endoscopies at 6–9 monthly intervals. The agreement (κ statistics) between the two endoscopists was found as 0.83. The mean carvedilol dose administered was 12±1.67 mg/day. The compliance was stringently monitored by pill count in each follow-up and over periodic telephonic interviews. Two patients were lost to follow-up and one was non-compliant in the placebo group, while three were non-compliant and one was intolerant to therapy in the carvedilol group. Hence, the intention-to-treat analysis was performed on 140 patients (70 in each group) and per-protocol analysis was performed on 133 patients (66 in carvedilol group and 67 in placebo group). Two patients in the placebo group underwent liver transplantation after last follow-up of 6 and 12 months. The mean follow-up in the carvedilol group was 21.67±5.13 months and 21.08±5.02 months in the placebo group.
Overall, 56 (80%) patients in carvedilol group and 45 (64.3%) patients in placebo group were free from progression to large varices (p=0.04). Those with progression to large varices underwent EVL. Using log-rank (Mantel Cox) test in intention-to-treat analysis, the cumulative probability of non-progression to large varices was 79.4% in the carvedilol group and 61.4% in the placebo group, respectively, and the mean time to non-progression to large varices being 20.8 (95% CI 19.4 to 22.4) months in carvedilol group and 18.7 (95% CI 17.1 to 20.4) months in placebo group (p=0.04) (figure 2).
In per-protocol analysis by log-rank (Mantel Cox) test, the cumulative probability of non-progression to large varices was 78.5% in carvedilol group and 59.8% in placebo group respectively, and the mean time to non-progression of large varices was 20.7 months (95% CI 19.2 to 22.3) in carvedilol group and 18.5 months (95% CI 16.8 to 20.3) in placebo group (p=0.03).
A total of 13 patients had gastric varices, 6 at baseline (two patients in carvedilol and four in placebo group) and 7 during the study period. In the carvedilol group, one patient was diagnosed with new gastric varices at 6 months, one at 12 months and two at 24 months of follow-up. In the placebo group, one patient was diagnosed with gastric varix at 6 months and two at 12 months of follow-up. Since the number of the patients with gastric varices was quite small, the effects of carvedilol on the prevention of growth of these varices could not be studied.
None of the patients died due to variceal bleed or hepatic causes in either of the two groups. There were two deaths in the carvedilol group and eight deaths in the placebo group. The cumulative probability of survival at 24 months in carvedilol and placebo groups was 96.3% and 85.8% respectively (p=0.034) in intention-to-treat analysis. In carvedilol group, one died from renal cell carcinoma and another died from urosepsis. In placebo group, the causes of death were intracranial haemorrhage (n=1), pneumonia followed by septic shock (n=3), acute coronary syndrome (n=2) and with renal failure (n=2). None was liver-related morbidity or mortality.
Reduction in HVPG
Fifty two of seventy (74.29%) patients in the carvedilol group and forty eight of seventy (68.6%) in the placebo group underwent repeat HVPG study.
The baseline HVPG was 14.49±4.31 and 14.97±4.76 mm Hg in carvedilol and placebo arm, respectively (p=0.52) (table 1). The mean reduction in HVPG in carvedilol group at the end of 1 year was from 14.5±4.3 to13.4±5.8 mm Hg (p=0.07). In the placebo group, the HVPG changed from 14.66±4.7 to 13.9±5.1 mm Hg (p=0.29) (figure 3).
The overall mean percentage reduction in the HVPG was 8.64% in the carvedilol group compared with an overall 0.33% increase in HVPG in the placebo group. However, this difference was not significant (p=0.22).
We did subgroup analysis to study the effect of antivirals treatment in cirrhotics of viral aetiology (chronic hepatitis B or chronic hepatitis C) on HVPG reduction in the two groups. Thirty eight patients with cirrhosis of viral aetiology and 102 patients with cirrhosis of non-viral aetiology had undergone HVPG at baseline. As shown in online supplementary table S2, there was no significant difference in HVPG reduction in these patients on antivirals compared with those with cirrhosis of non-viral aetiology.
Of the 70 patients in carvedilol group, 3 patients were non-compliant to the protocol. Two patients presented with initial lightheadedness that later resolved spontaneously. One patient in the carvedilol group had intolerable side effects (bradycardia with hypotension) and hence the drug had to be discontinued. One patient was non-compliant in placebo group and two patients were lost to follow-up. There was a significant reduction in the mean heart rate in carvedilol group at each follow-up (see online supplementary table S3) and mean arterial pressure (see online supplementary table S4). No major adverse events were reported in the placebo group. None of the patients bled in either group.
Portal hypertension is the single most important predictor of complications in cirrhosis of liver. Portosystemic collaterals develop above a portal pressure gradient of 10 mm Hg, and bleeding could occur from varices when the pressure rises above 12 mm Hg.20 In the Baveno III consensus meeting,21 a portal pressure above 10 mm Hg was taken as the cut-off for ‘clinically significant portal hypertension’. HVPG is the gold standard method for measurement of portal hypertension and prediction of complications.22 As per the staging classification of cirrhosis proposed in Baveno IV consensus meeting,23 stage 2 cirrhosis comprised patients with varices without ascites or variceal bleeding. Mortality in this stage is 3.4% per year and constitutes transition phase to decompensated cirrhosis namely stage 3 onwards with a mortality of 20% per year. This exemplifies the emphasis on the concept of early primary prophylaxis where NSBBs are used to prevent progression of small to large varices by control of portal pressure and hence a better survival.
In our previous study,9 no obvious benefit could be demonstrated with the use of propanolol to prevent progression to large varices from no varices or small varices. Subsequently, in another study, Merkel et al 10 showed a significant reduction in risk of development of large oesophageal varices with use of nadolol compared with placebo. None of these studies carefully studies the HVPG response with the progression of varices. In presence of these contrasting results, use of β-blockers in early primary prophylaxis remained a grey area in the latest Baveno V recommendations.7
In this novel study, we found that a lower proportion of patients progressed to large oesophageal varices over a follow-up of 24 months by use of carvedilol compared with placebo. There was also evidence showing that carvedilol had slowed the time to non-progression of small to large varices from 18.7 months (in placebo group) to 20.8 months by using carvedilol (p=0.04). Patients were administered the maximum tolerated dose of carvedilol, and periodic monitoring was done.
This beneficial effect can be explained by the dual effect of carvedilol on α1 and β receptors. The Spanish group24 showed that the number of HVPG responders was significantly higher with use of carvedilol compared with propranolol over 11.1±4.1 weeks (54% vs 23%; p<0.05). Another recent study by Hobolth et al 25 showed that overall HVPG reduction at 90 days was higher with carvedilol compared with propanolol (19.3±16.1% and 12.5±16.7%, p=0.21).
In our study, although there was a modest HVPG reduction of 8.64% at 1 year from baseline with carvedilol compared with 0.33% increase with placebo, it was statistically not significant. A study by Groszmann et al 26 also observed no significant reduction of HVPG with NSBB (timolol) compared with placebo at 1 year in patients with no varices and early portal hypertension. In the present study also, majority of the patients had Child's A or B cirrhosis and hence would fall into this category of mild-to-moderate splanchnic and portal hyperdynamic circulatory state with possibility of only mild-to-modest HVPG reduction. This may put forth an important point for consideration; whether we should use >20% reduction in portal pressure as a criteria for response in cirrhotic patients with small varices (early primary prophylaxis) or patients who have not developed varices (preprimary prophylaxis). It is likely that a modest reduction of portal pressure by 8%–10% maintained over a long period of time may have a significant effect on the natural history of this select group of patients. Groszmann et al 26 also showed that HVPG reduction of >10% from baseline was significantly associated with lower incidence of development of varices or variceal haemorrhage. Studies with longer follow-up and close haemodynamic monitoring would be of great help in evaluating the benefits of modest reduction in HVPG in cirrhotic populations.
Only modest reduction in portal pressure achievable by carvedilol, the most promising available drug, emphasises the need for search for newer treatment options to reduce portal pressure and hepatic fibrosis. Recently, Rosado et al 27 showed that terutroban significantly reduced portal pressure in CCl4–cirrhotic rats by action at thromboxane-A2/prostaglandin-endoperoxide receptors on the hepatic vascular endothelium leading to amelioration of fibrosis. Hence, it would be logical to do future studies on combination of carvedilol with newer anti-fibrotic approaches for effective control of portal hypertension.
There were no deaths related to variceal bleed or liver-related causes in either of the groups. An overall better survival at 2 years was noted by use of carvedilol, as there were only two deaths in carvedilol group and eight in placebo group; however, none of them were liver related. The relative safety of the long-term use of carvedilol in this subset of cirrhotic patients was a reassuring finding. Only, one patient had developed significant bradycardia with hypotension by use of carvedilol, which reversed on stoppage of the drug.
The strength of the present study is that a homogenous set of patients was included. Further, nearly 70% of the patients had repeat HVPG performed at 1 year; one of the largest in any early primary prophylaxis study so far; reaffirming the importance of haemodynamic monitoring in the pharmacological treatment of portal hypertension.
The present study could have been further strengthened if HVPG could have been repeated after 2 years of carvedilol use. However, since HVPG is an invasive procedure, one needs to have reliable and validated assessment by non-invasive methods like transient elastography.28 Such methods would also allow physicians at large to assess response to therapy by evaluating reduction in portal pressure at their place of practice.
In conclusion, this is the first RCT (NCT01196507) showing the efficacy of carvedilol in early primary prophylaxis by delaying the progression from small to large varices in patients with cirrhotic portal hypertension. Further multicentric studies would help to reaffirm the observations of this study.
AB, CKK and SKS are co-first authors.
The work was presented in part as an oral presentation at the AASLD, The Liver Meeting, 2013.
Contributors Guarantor of the article: SKS. SKS designed and made the study concept and performed critical revision of manuscript for important intellectual content; AB, CKK and CV performed the acquisition of data; GK, ASB and AB performed analysis and interpretation; AB and CKK drafted the manuscript. AC, SMS, RM, MK and VB provided inputs for drafting the manuscript and correcting it.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Institutional Ethics Committee, Institute of Liver and Biliary Sciences, New Delhi, India.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.