Introduction

Variceal hemorrhage is a serious complication of portal hypertension and represents approximately 60–65% of all bleeding episodes in patients with cirrhosis [1, 2]. The reported mortality rate during the first variceal hemorrhage episode is 15–20%, with higher rates in advanced liver disease [3]. Despite the availability of effective treatment options for acute variceal hemorrhage, the risk for subsequent episodes of hemorrhage and mortality remains substantial. In one study, the risk of rebleeding following an initial variceal hemorrhage was 13% after 5 days and 17% at week 6 with reported mortality of 20% [4].

Management of acute variceal hemorrhage consists of esophageal variceal band ligation (EBL) along with intravenous vasoconstrictors, antibiotics, and proton-pump inhibitor (PPI) followed by the initiation of secondary prophylaxis [5]. Combination therapy with EBL and nonselective beta-blockers are the current standard of care for secondary prophylaxis of variceal hemorrhage [6]. Despite the well-established effectiveness of PPI therapy in a variety of etiologies of upper gastrointestinal bleeding (UGIB), current data are insufficient to support its use in preventing variceal rebleeding or treating portal hypertensive gastropathy [7, 8].

Acid suppression therapy showed to benefit patients with cirrhosis by reducing the size of post-EVL esophageal ulcerations [9] and promoting gastric mucosal healing in peptic ulcer disease [10]. These benefits may explain the common clinical practice of prescribing oral PPI therapy in cirrhotic patients in the absence of supporting data and despite of published associations of long-term PPI use and spontaneous bacterial peritonitis as well as hepatic encephalopathy [11,12,13].

The role of PPI therapy in preventing UGIB in patients with cirrhosis after variceal hemorrhage remains unclear. Our study aimed to systemically analyze the role of PPI in post-band ligation ulcers.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline was fulfilled in this systematic review and meta-analysis [14].

Search strategy

We searched PubMed MEDLINE, Scopus, and Web of Science for studies that measured the effect of PPI for prophylaxis and treatment of post-band ligation ulcers up to July 20, 2021.

The following search terms were used: (“PPI” OR “Proton pump inhibitors” OR “Proton pump inhibit*”) AND (“post band ligation” OR “ligation ulcers” OR “bleeding ulcers” OR “post band ulcers”); moreover, reviewing the reference lists of retrieved articles was used to complement the broad search.

Eligibility criteria

Studies that measured the effect of PPI as treatment or prophylaxis for post-band ligation ulcers and articles that were published in peer-reviewed international journals and had enough data for qualitative and quantitative analysis were included with no language restriction. We excluded conference papers, unpublished articles, reviews, letters to the editor, posters, and animal studies.

Data extraction

We extracted the following data from the included studies as baseline characteristics: name of the first author, publication year, country, study design, gender, mean age, and total sample size (Table 1). For qualitative and quantitative analysis, the received medical treatment, inclusion and exclusion criteria, and conclusion were extracted (Table 2).

Table 1 Baseline characteristics
Full size table
Table 2 Summary of included studies
Full size table

Quality assessment

We used the Newcastle-Ottawa Scale (NOS) [22] to assess the observational studies and ROB-2 risk of bias version 2 for randomized control trials (RCT). The NOS tool judges the studies on three broad perspectives: the selection of the study groups, the comparability of the groups, and the ascertainment of either the exposure or outcome. Furthermore, ROB-2 tool assesses the risk of biases in the following domains: (i) bias arising from the randomization process, (ii) bias due to deviations from intended interventions, (iii) bias due to missing outcome data, (iv) bias in measurement of the outcome, and (v) bias in selection of the reported result. A judgement of “low risk,” “some concerns,” or “high risk” was made for the risk of bias in each domain, allowing an overall risk of bias to be generated for each study using the tools algorithm. Two independent reviewers (A.A and A.A) screened the methodological quality of included studies and in case of discrepancies were resolved by discussion.

Data analysis

We conducted our double-arm meta-analysis using RevMan version 5. Random-effects meta-analysis models were employed to estimate the effect of PPI for bleeding, bleeding-related death, and hospitalization. Heterogeneity was evaluated using the inconsistency (I2) and chi-squared (χ2) test. I2 > 50% was considered substantial heterogeneity in the studies, and a P value less than 0.05 was considered statistically significant. The data was continuous, and we used the standardized mean difference (MD) and risk ratio (RR) with a 95% confidence interval to assess the estimated effect measure.

Results

Search results

Our search strategy resulted in a total number of 127 studies. After removing the duplicates, 79 articles were screened for title and abstract screening, and 25 full-text articles were evaluated for eligibility. Following the full-text screening, 7 [15,16,17,18,19,20,21] papers met our criteria and were included in our meta-analysis (Fig. 1). Four studies were randomized control trials; three were retrospective cohort.

Fig. 1
figure 1

Full-text screening

Full size image

Baseline characteristics/summary of the included studies

A total of 2030 patients were included in our study of which 1480 participants were males (72%) and 550 females (18%). Mean age was 59.7 years old. Various types of proton pump inhibitors (PPI) were used including pantoprazole, rabeprazole, or omeprazole—the most used PPI. All PPI were used for treatment of post-band ligation ulcers occur for hospitalized patients (Tables 1 and 2).

Quality assessment

ROB-1 was performed assessing the risk of bias for randomized controlled trials; out of our 4 studies, 2 showed low risk of bias and 2 unclear (Figs. 2 and 3). While for cohort studies, judged by following New castle Ottawa (NOS) guidelines, our three cohort studies were of good quality due to matching of the cases and controls regarding the confounders and well selection of controls with detailed description (Table 3).

Fig. 2
figure 2

Low and unclear risk of bias

Full size image
Fig. 3
figure 3

Studies showing the risk of bias

Full size image
Table 3 Newcastle-Ottawa Scale (NOS) for assessing the quality of observational studies
Full size table

Data analysis

Forest plot of a random-effects meta-analysis on post-band ligation variceal bleed compares PPI with placebo. Values are risk ratios (95% CIs). The shaded boxes represent the point estimate for each individual trial, and the horizontal line extending from each box represents the upper and lower limits of the 95% CI. The size of the shaded circle indicates the relative weight of the trial in the meta-analysis. The diamonds represent the overall pooled risk ratio.

In our first analysis, all our seven studies including 2030 patients, rebleeding post-band ligation compared PPI and placebo with significant favor PPI (p = 0.00001). The pooled risk ratio was 0.53 (95% CI of 0.41, 0.68), showing a protective effect from rebleeding with PPI. Heterogeneity analysis demonstrated low-moderate statistical evidence for heterogeneity (I2 = 23%, p = 0.26) (Fig. 4).

Fig. 4
figure 4

Forest plot for rebleeding post-band ligation

Full size image

In the second analysis, seven studies including 2030 patients bleeding-related death at a 1-month period compared PPI and placebo with significant favor for PPI (p = 0.00001). The pooled risk ratio was significant at 0.33 (95% CI of 0.20, 0.53), showing a protective effect from bleeding-related death with PPI. No heterogeneity analysis was found as evidence for heterogeneity (I2 = 0%, p = 0.69) (Fig. 5).

Fig. 5
figure 5

Forest plot of bleeding related death

Full size image

In the third analysis, four studies including 1141 patients’ length of hospital stay postoperative compared PPI and placebo. The cumulative mean difference was insignificant at 0.13 (95% CI of −1.13, 1.39), showing no effect either for PPI or placebo on length of hospital stay. Heterogeneity analysis demonstrated no evidence for heterogeneity (I2 = 0%, p = 0.84) (Fig. 6).

Fig. 6
figure 6

Forest plot of length of hospital stay 

Full size image

Discussion

Our study demonstrates a significant reduction in the rate of bleeding and bleeding-related deaths with the use of PPIs rather than placebo following EVL. In addition, there is no evidence that this benefit comes at the cost of a longer hospital stay. Thus, our analysis shows that PPIs may be a valuable option following EVL as they are a cheap and widely available class of drugs that may significantly reduce complications and mortality following the procedure [22, 23].

Following EVL, bleeding due to ligation ulcers is a common complication occurring after 2.8 to 7.8% of procedures [24,25,26,27], although this rate varies depending on the setting (elective versus emergency) of the EVL session, with emergent EVL carrying a much greater risk of rebleeding [26]. Such bleeding is not only severely debilitating to the patient, but may also be fatal, with a 6-month mortality rate of 58.6% in one study [28].

One potentially important cause of post-EVL bleeding is acid reflux, which has been associated in one study with a significantly increased risk of post-EVL bleeding in patients receiving prophylactic ligation6. Therefore, a possible mechanism by which PPIs may reduce post-EVL bleeding is the reduction of epithelial exposure to acid following the procedure.

Currently, there are no clear recommendations on the use of PPIs in patients with cirrhosis. For instance, the 2015 UK guidelines do not recommend proton pump inhibitor use for the control of an acute variceal bleed or for the prevention of post-EVL bleeding [29]. These recommendations are primarily based on data associating PPI use with severe adverse events. For instance, a 2014 propensity-matched cohort study showed a significantly higher rate of spontaneous bacterial peritonitis (SBP) in patients using PP [30]; however, findings on this risk have been conflicting, with two recent studies not reporting a positive association between PPI use and SBP [30, 31], and one study reinforcing the finding of the 2014 propensity matched cohort study by showing a positive association [32, 33]. In addition, an observational study linked PPI use in patients with cirrhosis to a higher mortality rate [34]. However, patients taking PPIs had a higher baseline severity of disease, and although the authors used multivariate models to adjust for potential confounders, it is doubtful that all potential confounders were adequately adjusted for.

In addition to PPIs, another option for post-EVL bleeding prophylaxis is sucralfate. A study by Sakr et al. showed that sucralfate prophylaxis, compared to placebo, was associated with a nearly 50% relative reduction in the number of patients having post-banding ulcers [35]. Further, the mean size of ulcers in the sucralfate group was also significantly lower. Recently, a trial by Seo et al. showed that combination therapy with EVL and beta-blockers, for the primary prophylaxis of variceal bleeding, significantly reduced the 2-year recurrence rate of bleeding compared to either option alone by nearly four-folds [36]. However, there was no signal of a mortality benefit. To our knowledge, both studies are only available as abstracts and should accordingly be interpreted with caution. A small earlier randomized trial by Nijhawan et al. (30 patients) did not show that the use of sucralfate did not result in enhanced healing [37]. Another trial investigating simvastatin did not show a significant reduction in the rates of bleeding; however, it was a relatively small trial of 59 patients, and the simvastatin group had significant reductions in portal pressure [38, 39].

Ultimately, because of the association of PPI with SBP, mortality, and a consequently unclear net clinical benefit, it may be rational to target high-risk patients for PPI therapy then to use them for all-comers. A number of risk factors have been associated with rebleeding after EVL, including Child-Pugh C status, bacterial infections, bilirubin levels, coagulation indices, the extent of ascites, varices, and the number of bands placed during EVL [40, 41]. In the future, randomized trials enrolling those patients at the highest risk of post-EVL rebleeding may show a net clinical benefit to the use of PPIs following EVL.

Our study has some limitations which ought to be acknowledged. First, a substantial portion of the evidence was derived from observational studies. Second, although statistical heterogeneity was low, there was some significant clinical heterogeneity as not all studies enrolled patients with a similar baseline severity or for the same purposes of primary vs secondary prophylaxis. Third, our meta-analysis cannot be used to determine the net clinical benefit to using PPIs, as side effects of PPI use were not evaluated in our analysis. Finally, it is unclear from our analysis what the optimal duration of PPI therapy is.

In conclusion, our analysis suggests a twofold reduction in the risk of bleeding and a threefold reduction in the risk of bleeding-related death with the use of PPI following EVL. However, a significant portion of the evidence was derived from observational studies, and previous studies have raised concern about the association of PPIs with SBP. Accordingly, future randomized trials targeting high-risk patients are needed to inform clinical practice.