Risk of rebleeding from gastroesophageal varices after initial treatment with cyanoacrylate; a systematic review and pooled analysis

Background Cyanoacrylate alone or in combination with other interventions, can be used to achieve variable rates of success in preventing rebleeding. Our study aims to assess the pooled risk of gastric and esophageal varices rebleeding after an initial treatment with cyanoacrylate alone and/or in combination with other treatments, by a systematic review of the literature and pooled analysis. Methods PubMed, EMBASE, SCOPUS, and the Cochrane library were searched for studies that reported the risk of rebleeding during the follow-up period after treatment of gastric or esophageal varices with either cyanoacrylate alone or in combination with other treatments. Standard error, upper and lower confidence intervals at 95% confidence interval for the risk were obtained using STATA Version 15 which was also used to generate forest plots for pooled analysis. The random or fixed effect model was applied depending on the heterogeneity (I2). Results A total of 39 studies were found to report treatment of either gastric or esophageal varices with either cyanoacrylate alone or in combination with other treatments. When gastric varices are treated with cyanoacrylate alone, the risk of rebleeding during the follow-up period is 0.15(Confidence Interval: 0.11–0.18). When combined with lipiodol; polidocanol or sclerotherapy the rebleeding risks are 0.13 (CI:0.03–0.22), 0.10(CI:0.02–0.19), and 0.10(CI:0.05–0.18), respectively. When combined with percutaneous transhepatic variceal embolization; percutaneous transhepatic variceal embolization; endoscopic ultrasound guided coils; or with ethanolamine, the rebleeding risk are 0.10(CI:0.03–0.17), 0.10(CI:0.03–0.17), 0.07(CI:0.03–0.11) and 0.08(CI:0.02–0.14), respectively. When esophageal varices are treated with cyanoacrylate alone, the risk of rebleeding is 0.29(CI:0.11–0.47). When combined with percutaneous transhepatic variceal embolization; sclerotherapy; or band ligation, the risks of rebleeding are 0.16(CI:0.10–0.22), 0.12(CI:0.04–0.20) and 0.10(CI:0.04–0.24), respectively. When combined with a transjugular intrahepatic portosystemic shunt; or ethanolamine, the risks of rebleeding are 0.06(CI: − 0.01-0.12) and 0.02 (CI: − 0.02-0.05), respectively. Conclusion In treating both gastric and esophageal varices, cyanoacrylate produces better results in terms of lower risk of rebleeding when combined with other treatments than when used alone. The combination of cyanoacrylate with ethanolamine or with endoscopic ultrasound guided coils produces the lowest risk of rebleeding in esophageal and gastric varices, respectively. We call upon randomized trials to test these hypotheses.


Background
Liver cirrhosis is the leading cause of portal hypertension which in turn, leads to portal hypertension and gastrointestinal varices. Up to 17% of liver cirrhosis patients will develop esophageal varices, while 15% will develop gastric varices. Up to 30%, gastroesophageal varices will bleed within 2 years [1]. Bleeding from varices is one among gastrointestinal emergencies that account for the majority of mortalities and morbidities among portal hypertension patients despite the cause [2]. About 50 to 80% of patients who survive the first episode of variceal hemorrhage will have a recurrent early or late rebleeding episode [3]. Up to 20% of patients with a rebleeding episode will not survive [4].
From older literature, half of the variceal hemorrhages would stop spontaneously however, the risk of rebleeding and mortality increases significantly [5]. Current studies, however, report that in patients with cirrhosis Child-Pugh of class C or with hepatic venous pressure of higher than 20 mm Hg are less likely to spontaneous stoppage of bleeding. These patients would require interventional hemostatic measures with pharmacological drugs such as octreotide, somatostatin and beta blockers; endoscopic sclerotherapy, band ligation, or tissue adhesives injection; and/or shunting by surgery or by transjugular intrahepatic portosystemic shunt to achieve hemostasis. A selective combination of these approaches has also been reported [1]. Different hemostatic approaches differ in terms of their success rates in achieving hemostasis, preventing rebleeding, and reducing mortality and morbidity. With advancing technology, each approach has evolved, and tissue adhesives have increasingly being used as the first line of therapy during the last decades [6,7].
Also known as "tissue glue", tissue adhesives were approved by the United States of America's Food and Drug Authority in 1998, however, there have been previous studies reporting their use as back as the year 1981 [8]. Primarily containing n-butyl-2 cyanoacrylate or 2-octyl cyanoacrylate, tissue adhesives are liquid monomers that undergo chemical reactions upon contact with moisture, to form polymers that can strongly attach to tissue [9]. Despite a number of reported complications associated with their use such as embolism and needle impaction [10], cyanoacrylate has been reported to have higher hemostasis and lower rebleeding rates than traditional band ligation and sclerotherapy in gastroesophageal varices [2]. Moreover, they have been reported to have antibiotic activity towards gram-positive bacteria [11].
Cyanoacrylate can be used alone or in combination with other interventions, to achieve variable rates of successes in hemostasis, reducing mortality and prevention of rebleeding. Our study was aimed at assessing the overall risk of gastroesophageal rebleeding after an initial treatment with cyanoacrylate alone and/or in combination with other treatments, by a systematic review of literature and pooled analysis.

Eligibility criteria
The current study involved participants with bleeding gastroesophageal varices who underwent hemostasis by cyanoacrylate injection alone or in combination with other treatments. Observational and interventional studies reporting the risk of rebleeding after hemostasis treatment were included. Extending the external validity, eligible English published literature from across the world were included.

Information sources
Four online databases, namely PubMed, EMBASE, SCO-PUS, and the Cochrane library were systematically searched with no time range specified. Secondary referencing of eligible studies extended the search scope. The last search was conducted on 4th March 2020.

Study selection process
Two authors screened titles and abstracts of all articles from online database searches to identify the most relevant articles in line with our study question. The relevant articles were sought for full texts and finally included studies were identified after thorough reading full text articles to assess inclusion and exclusion criteria. This process was done by two authors; ZH and JS with the third author, TL assisting to resolve discrepancies. The search, screening, and study identification process are summarized in Fig. 1.

Data extraction
Before the data extraction process from full-text articles meeting eligibility criteria for inclusion, assessment for methodological biases was done by using the Joanna Briggs institute meta-analysis of statistics assessment and review instrument. PRISMA [12](preferred reporting items for systematic reviews and meta-analyses) tool was used to minimize reporting bias upon the write-up of this study.
Data collected included author name, year of publication, country of study, study design, comparison groups involved, varicose lesion location, study sample size, definitive diagnoses, number/ proportion of rebleeding events among followed up patients, the name of tissue adhesive utilized, treatment urgency, extent of live damage prior to treatment and follow-up duration. This was independently performed by two authors, namely; ZH and DZ with SL to resolve discrepancies. The current study had one outcome, the risk of rebleeding.

Analysis
The risk of rebleeding among gastric and esophageal varices patients were analyzed separately. Moreover, the risk of rebleeding in gastric or esophageal varices groups was analyzed separately depending on whether the cyanoacrylate was utilized alone or in combination with other treatments. This gave rise to five separate analyses on which quantitative analysis was conducted: [1] Analyzing the pooled risk of rebleeding in gastric varices treated with cyanoacrylate alone [2]; analyzing the pooled risk of rebleeding in esophageal varices treated with cyanoacrylate alone [3]; analyzing the pooled risk of rebleeding in gastric varices treated with cyanoacrylate with ethanolamine [4]; analyzing the pooled risk of rebleeding in gastric varices treated with cyanoacrylate with endoscopic ultrasound guided coils; and [5] analyzing a pooled risk of rebleeding in esophageal varices treated with cyanoacrylate with percutaneous transhepatic variceal embolization. A qualitative narrative (i.e. descriptive) approach was utilized in assessing the risk of rebleeding in gastroesophageal varices treated with cyanoacrylate with sclerotherapy as the eligible studies involved different participants.
The risk of rebleeding was calculated dividing the number of patients rebleeding during the follow-up period after endoscopic hemostasis by the total number of patients that initially underwent the endoscopic hemostasis procedure. The denominator did not include patients lost during the follow up. Standard error, upper and lower confidence intervals (at 95% confidence interval) for the risk, were obtained from the "generate command" in computer software STATA Version 15 which was also used to generate forest plots for pooled analysis. The software was customized to a random or fixed effect model depending on the heterogeneity (I 2 ) of the studies when analyzing the outcomes. The fixed effect model was used when I 2 was less than 50% and the random effect model was used when I 2 was more than 50% indicating significant heterogeneity.

Assumptions
Participants were considered to have been correctly diagnosed with upper gastrointestinal bleeding due to gastric or esophageal varices, and not due to other causes such as Mallory-Weiss tear or gastritis. Despite the country under which treatment was given, all patients were considered to have received standard care.

Results
A total of sixty (60) studies that seemed to be relevant to our study basing on screening titles and abstract, were sought for full texts. Five of these were eliminated after thorough full-text reading.  [16] used Korean and Chinese language, respectively. A total of 39 studies were included in the systematic review while 32 studies were pooled for statistical analysis. Table 1 illustrates the characteristics of all included studies in our pooled analysis. These were published between the years 1989 and 2019 from countries in Africa, Europe, Asia, and North America. Eleven studies were retrospective observational; 16 were prospective observational; two were case series, and ten were randomized clinical trials. Thirteen studies were comparative, one arm of which was cyanoacrylate. Eleven studies were non-comparative involving only cyanoacrylate outcome assessment while of the two studies, one involved comparing different doses of cyanoacrylate (i.e. 0.5mls versus 1.0mls) while another compared diluted versus undiluted cyanoacrylate. Follow-up duration after treatment with cyanoacrylate ranged from 6 weeks to 15 years in another study. One study did not report the duration of follow-up.

Characteristics of included studies
A total of 39 studies reported 3630 who had either gastric or esophageal variceal and underwent hemostasis with cyanoacrylate alone or in combination with other treatments. A total of 497 had gastric or esophageal recurrent bleeding episodes during the follow-up period.
Pooled risk of rebleeding in gastric varices treated with cyanoacrylate alone Figure 2 illustrates a forest plot of the pooled risk of rebleeding for gastric varices after cyanoacrylate treatment. A total of 25 studies reported 2590 gastric variceal patients, of whom 402 had had rebleeding after initial treatment with cyanoacrylate hemostasis. The risk ranged from a minimum of 0.04 (4%) to a maximum of 0.99 (99%) in another study. Two studies were excluded for not having rebleeding incidences during the follow up period. The pooled overall risk of rebleeding was 0.30 (confidence interval: 0.30-0.31).
There was a significant heterogeneity observed with I 2 of 99.7%, p-Value< 0.05. This led us to conduct sensitivity analysis, eliminating peculiar studies from the analysis. Figure [42]. Thakeeb reported 3 (i.e. risk = 0.052) rebleeding events among gastric variceal patients; and one (risk = 0.017) rebleeding events among esophageal varices patients. Maruayama reported 10 (i.e. risk =0.5) rebleeding events among gastric varices patients. Figure 5 illustrates a forest plot of   Table 2 summarizes risks of rebleeding in gastric and esophageal varices when treated with cyanoacrylate alone or in combination with other treatments as discussed earlier.

Discussion
Through decades-long progressive improvements in the treatment of gastroesophageal varices, cyanoacrylate has evolved to be one of the favored first lines of treatment. The current study was aimed at utilizing a systematic review of literature and pooled analysis to assess the overall risk of gastroesophageal rebleeding after an initial treatment with cyanoacrylate alone and/or in combination with other treatments.
Following the treatment of gastric varices with cyanoacrylate alone, 25 studies demonstrated different risks of rebleeding from the minimum of 0.04 to a maximum of 0.99 in another study, with the overall pooled risk of 0.30 (confidence interval: 0.30-0.31). However, after getting rid of peculiar studies that increased heterogeneity, the resulting overall pooled risk was 0.15 (Confidence interval: 0.11-0.18). This risk of rebleeding coincides with that previously reported by Hou et al. (2009) [28] but differed from the majority of other studies. Authors believe that the reason for the differences among studies to be technological advancement with time. This can be demonstrated the majority of studies from the year 2010 forward having a lower risk of rebleeding than studies before 2010. Different sample sizes and different study methodologies could also explain the differences.
Esophageal varices treated with cyanoacrylate alone showed the risk of rebleeding ranging from 0.25 to 0.38 in different studies with the pooled overall risk of 0.29 (confidence interval: 0.11-0.47). Following a fewer number of studies, a meta regression could not be conducted. However, authors believe that the reason for the differences between studies to be due to different methodological approaches between the studies as Rivet et al.  [22]. The authors of this study hypothesize that gastric varices respond better to cyanoacrylate as compared to esophageal varices in terms of lower risk of rebleeding. We call upon randomized clinical trials comparing the risk of rebleeding between gastric varices and esophageal varices treated with cyanoacrylate alone. When cyanoacrylate is combined with ethanolamine in the treatment of gastric varices the pooled risk of rebleeding after treatment is 0.08 (Confidence interval: 0.02-0.14). The result aligns with that reported by Thakeb et al. (1995) but differs from Maruyama who reported a higher risk of 0.5. The difference is accounted for by fewer sample size by Maruyama. On the other hand, when the combination is used to treat esophageal varices the risk of rebleeding is 0.017(confidence interval: − 0.02-0.05). From an otherwise weak basis, we hypothesis that esophageal varices in contrast to gastric varices, respond better to the combination of cyanoacrylate and ethanolamine, in terms of lower risk of rebleeding. We call upon clinical randomized clinical trials to test this hypothesis.
From our findings, when cyanoacrylate is combined with endoscopic ultrasound guided coils to treat gastric varices the pooled risk of rebleeding is 0.07(confidence interval: 0.03-0.11). This finding is more or less similar to that reported by Bhat et al. (2016) [43] but is higher than that reported by Robles-Medranda et al. (2019) [44]. The reason for the differences could be explained by different sample sizes among studies pooled. One study had nearly five times the sample size used by the other.
When esophageal varices are treated with a combination of cyanoacrylate and percutaneous transhepatic variceal embolization the pooled risk of rebleeding is 0.16(confidence interval: 0.10-0.22). This is coinciding with findings previously reported by Zhang et al. (2007) [16]. In another study by Tian et al. (2011) [46] when the combination is used to treat gastric varices, the risk of rebleeding is 0.10(confidence interval: 0.03-0.17). We hypothesize that esophageal varices in contrast to gastric varices, respond better to the combination of cyanoacrylate and percutaneous transhepatic variceal embolization in terms of lower risk of rebleeding. The authors call upon randomized clinical trials to test this hypothesis.
The risk of rebleeding in gastric varices treated with cyanoacrylate with sclerotherapy was lower by 0.02 from that of esophageal varices treated with the same combination. The difference could partly be due to more or less the same number of sample sizes among the two studies descriptively analyzed. In combination with other treatments such as transjugular intrahepatic portosystemic shunt and balloon-occluded retrograde transvenous obliteration, it is evident that cyanoacrylate improves the efficacy of the treatment of gastroesophageal varices in terms of lowering rebleeding risk.

Study limitations, measures taken, and recommendations
Our study search was limited to English published literature; involved pooling of studies with different sample sizes, different study designs, and different follow-up durations. As demonstrated by Child-Pugh or the model for end-stage liver disease (MELD) classifications, different studies involved participants with different extents of liver damage/cirrhosis. Despite a few studies involving either emergent [49] or elective [50] participants only, the majority of studies combined the two [16,48]. Moreover, from older literature by Sarin et al. (1992) [53], the risk of rebleeding varied with lesion's location on the gastric wall. From the study, isolated varices bled more often as compared to cardia and fundal varices.
These were thought to introduce heterogeneity in the pooled analysis. However, authors appraised eligible studies; performed sensitivity analyses, meta-regression, study exclusion, and used random effect models to deal with high heterogeneity among pooled studies. We also utilized PRISMA tools to minimize reporting biases.
We call upon robust randomized studies taking into account biases encountered in our study and adequately matching participants by the extent of liver damage/cirrhosis; treatment urgency whether elective or emergency; lesion location and follow-up duration.

Conclusion
In treating both gastric and esophageal varices, cyanoacrylate produces better results in terms of lower risk of rebleeding when combined with other treatments than when used alone. The combination of cyanoacrylate with ethanolamine or with endoscopic ultrasound guided coils produces the lowest risk of rebleeding in esophageal and gastric varices, respectively. We call upon randomized trials to test these hypotheses.