Clinical Features and Outcomes of Repeated Endoscopic Therapy for Esophagogastric Variceal Hemorrhage in Cirrhotic Patients: Ten-Year Real-World Analysis

Objective This study is aimed at evaluating the survival of cirrhotic patients with different etiologies after endoscopic therapy for acute variceal bleeding and the effect of repeated endotherapy on patients' prognosis. Methods We retrospectively evaluated the clinical features and outcomes between cirrhotic patients with chronic HBV or HCV infections and other etiologies. The 3-year and 5-year survival rates and rehemorrhage rate in one year between the viral and nonviral cirrhosis patients were compared by Kaplan-Meier curves and log-rank test. Cox analysis was used to identify the impact factors that affect the long-term survival of patients with cirrhosis and variceal bleeding after endotherapy. Results Out of 2665 patients with liver cirrhosis and variceal hemorrhage selected from our medical center between September 2008 and December 2017, a total of 1342 patients were included for analysis. The median follow-up duration was 32.9 months (range 0.16-111.4 months), the 3- and 5-year cumulative survival rates were 75.3% and 52.8%, respectively. The median survival time was significantly longer in viral cirrhosis patients (47.1 months [95% CI: 24.9-69.1]) compared with nonviral cirrhosis patients (37.0 months [95% CI: 25.0-56.0], p = 0.001). The 3-year and 5-year survival rates of the viral group were higher than the nonviral group. The rehemorrhage rate at one year was higher in nonviral patients than in viral patients (p < 0.001). Conclusion Repeated endotherapy combined with effective antiviral therapy is helpful for long-term survival of cirrhotic population with variceal hemorrhage and HBV or HCV infection.


Introduction
Acute variceal bleeding (AVB) is a fatal complication in patients with liver cirrhosis, which is associated with increased mortality of about 20% at six weeks, despite recent progress in management. Nearly 50% of patients with newly diagnosed liver cirrhosis have accompanying varices, and new varices develop every year or preexisting varices worsen in 7-8% of patients, and the first bleeding occurs in 12% of patients each year [1,2]. Most cirrhotic patients cannot receive liver transplantation after variceal hemorrhage and other cirrhotic complications in China, since few donated livers are used and it is too expensive for most patients to receive transplantation. Consequently, endoscopic therapy is one of most common techniques to control esophageal variceal bleeding in such patients. However, the clinical out-comes of cirrhotic patients with different etiologies after repeated endotherapy for AVB remain unknown.
The past 2-3 decades have seen significant advances in the management of portal hypertension and AVB, and this has led to marked improvement in survival [3][4][5]. The hemostatic rate achieved by emergency EIS ranged between 60 and 100% [6]. EVL has been extensively evaluated in the treatment of variceal bleeding and in the prophylaxis of first bleeding [7,8]. Cyanoacrylate injection is globally accepted as the primary intervention for gastric variceal bleeding. It was reported that 92% of patients admitted for acute gastric variceal bleeding achieved successful hemostasis after the injection of cyanoacrylate [9]. Combined treatment with vasoactive drugs, prophylactic antibiotics, and endoscopic techniques is the recommended standard of care for patients with AVB [10]. Endoscopic therapy has been considered the mainstay treatment for AVB. EVL combined with a vasoactive drug is considered the standard care for AVB, which is recommended by BAVENO VI [11]. However, the longterm effect of ligation plus cyanoacrylate injection with or without sclerotherapy of variceal bleeding in cirrhotic patients with concomitant esophageal and gastric varices remains unclear.
In the present study, we retrospectively analyzed 1342 liver cirrhosis patients with variceal bleeding after endotherapy in our hospital during the past decade. The aim of this study was to analyze patients with cirrhosis receiving standard therapies for the different etiologies and complications in a real-life setting. Survival and rebleeding analysis was used to evaluate the prognostic impact of viral liver cirrhosis and nonviral cirrhosis. Factors that affect the clinical outcomes were analyzed in patients with liver cirrhosis and variceal bleeding after endoscopic therapy.

Subjects.
This retrospective single-center study was completed at the Beijing Ditan Hospital of Capital Medical University. The clinical data was collected between September 2008 and December 2017. Eventually, 1342 liver cirrhosis patients with variceal bleeding who were treated with endotherapy were included in the analysis. The exclusion criteria were as follows: (i) patients with liver carcinoma and other serious concurrent illnesses; (ii) patients who received TIPS, splenic embolization, pericardial devascularization with splenectomy, or other interventions for varices; and (iii) patients who were lost to follow-up.
The study protocol was approved by the Ethics Committee of Beijing Ditan Hospital, Capital Medical University [JDLKZ (2018-021-01)]. The study has been registered in the Chinese Clinical Trial Registry (ChiCTR1800019265).

Treatments and Endoscopic
Therapy. All patients with advanced cirrhosis and AVB were started on vasoactive drugs including vasopressin, somatostatin, and their analogs as the initial treatment before endotherapy. Antibiotics were also administered as soon as the patients arrived at the hospital. Endotherapy was performed when the patients achieved hemodynamic stability, 12-24 hours after hospital admission.
The endotherapy was performed using the following methods: (i) endoscopic variceal ligation (EVL) was used for patients with the first esophageal variceal bleeding; (ii) cyanoacrylate injection was used for patients with gastric variceal hemorrhage; (iii) the "sandwich therapy" was performed in patients with esophageal gastric variceal hemorrhage. Patients were injected with 2 mL polidocanol, 0.5 mL n-butyl-2 cyanoacrylate, and 2 mL polidocanol (repeated injection if necessary). Most patients with GOV1 and GOV2 varices received repeated endotherapy every 3-6 months during follow-up. Endoscopic injection sclerotherapy (EIS) was also performed for some patients with esophageal variceal rebleeding based on the clinical guidelines [12,13] and the specialists' preference and experience for choosing the best endotherapy. Nonselective beta blockers (NSBBs) were also administered for the secondary prevention of rebleeding after achieving hemodynamic stability. 2.5. Statistical Analysis. The dataset was divided into different groups for analysis. The clinical characteristics of viral and nonviral cirrhosis patients were analyzed. Quantitative data was detected for normal distribution. Student's t-test was used if the data accorded with normal distribution, and the Mann-Whitney U test was used if the data did not accord with normal distribution. Mean ± standard deviation and interquartile range were used as appropriate. The chisquare test was used for analyzing qualitative data, and the data were expressed in percentage.
To examine the survival differences among the groups, the Kaplan-Meier method was used to generate the survival curves. The log-rank method was used to compare the differences between the groups. Univariate and multivariate Cox analyses were conducted to identify the impact factors that influenced the long-term survival. Two-sided p values < 0.05 were considered statistically significant. Statistical analyses were performed using SPSS 19.0 (IBM Corp., Armonk, NY, United States) and GraphPad Prism 7.0 (GraphPad Prism Software Inc., San Diego, CA, USA).

Clinical Characteristics.
A total of 2665 patients with liver cirrhosis and variceal bleeding after endotherapy were assessed in the study. After excluding 363 patients lost to follow-up; 376 patients with preexisting HCC and other serious concurrent illnesses; and 584 patients who received TIPS, splenic embolism, splenectomy, and liver transplantation, 281/584 patients received these interventions after repeated endotherapy, and these patients were further interviewed, of which 78 patients received the intervention accompanied with indications for other treatments, 203 patients were for other reasons, such as being combined with cholecystolithiasis requiring cholecystectomy; being combined with Among the 1342 patients included in the study, 68.5% (919 of 1342) were male, and the average age was 52.2 years (range 18-83 years). The median follow-up duration was 65 months (range 0.16-126.40 months), and the 3and 5-year cumulative survival rates were 75.3% and 52.8%, respectively. A total of 1010 patients were followed up for three years, of which 692 had viral cirrhosis and 318 had nonviral cirrhosis, while nine patients died of noncirrhosis-related causes. A total of 709 patients were followed up for five years, of which 500 had viral cirrhosis and 209 had nonviral cirrhosis. The flowchart of the patient selection process is shown in Figure 1. The baseline characteristics of patients who were followed up for three years are shown in Table 1.

Comparison of Survival between Viral and Nonviral
Populations. To evaluate the long-term outcome of endotherapy, the survival rates between the viral and nonviral groups were analyzed. The result showed that the median survival time was significantly longer in viral cirrhosis patients (47.1 months [95% CI: 24.9-69.1]) compared with nonviral cirrho-sis patients (37.0 months [95% CI: 25.0-56.0]) (p = 0:001). The 3-year survival rate of the viral group was 81.2%, which was higher than that of the nonviral group (73.3%) (p = 0:01) (Figure 2(a)). The 5-year survival rate of the viral group was 66.6%, which was significantly higher than that of the nonviral group (48.3%) (p < 0:001) (Figure 2(b)). The viral cirrhosis patients had longer survival time than the nonviral cirrhosis patients. However, there was an intersection in the survival curves of both groups at 15 months after endotherapy. So we further compared the survival rate before and after 15 months in the viral and nonviral groups. The result showed that the survival rate was significantly higher in the nonviral group than in the viral group within 15 months after patients received endotherapy (p = 0:008) (Figure 2(c)). In contrast, the survival rate was significantly higher in the viral group than in the nonviral group after 15 months after endotherapy (p < 0:001) (Figure 2(d)).
Since different etiological factors may influence the prognosis of cirrhosis, we further assessed the survival rates of subgroups of nonviral cirrhosis patients.

Comparison of Rebleeding between Viral and Nonviral
Populations. The rebleeding rate at one year was compared between the viral and nonviral groups. The viral population had a lower rebleeding rate than the nonviral population (23.0% versus 33.0%, p < 0:001) (Figure 4(a)).
The rebleeding rate at one year was also different in the subgroups of the nonviral population (p = 0:032) (Figure 4(b)). When we further compared it in pairs, the rebleeding rate was significantly higher in the alcohol group than in the cryptogenic group (p = 0:046) (Figure 4(c)).

Predictors of Long-Term
Survival. Among the 1342 liver cirrhosis patients with AVB after endotherapy, 709 patients were followed up for five years, of whom 500 were viral cirrhosis patients, while 209 were nonviral cirrhosis patients.
Factors associated with 5-year mortality were analyzed by univariate and multivariate Cox regression analyses. Univariate analysis showed that age, gender, WBC, NEU%, LYM%, HGB, GGT, TBIL, ALB, Cr, INR, MELD score, and Child-Pugh score significantly impacted the incidence of death in five years. Multivariate analysis showed that age, LYM%, HB, GGT, DBIL, ALB, and MELD score were independent impact factors for the high risk of mortality (Table 2).

Discussion
Endoscopic therapy has been used to treat variceal hemorrhage for nearly half century. However, the long-tern survival of cirrhotic patients with repeated variceal bleeding remains unknown. Liver transplantation is unavailable for the   Gastroenterology Research and Practice majority of cirrhotic patients in China. When these patients suffer from AVB, endotherapy is their major treatment option. It is necessary to evaluate the survival after repeated endotherapy in liver cirrhosis patients with AVB. In this study, survival analysis was performed to evaluate the prognosis of viral and nonviral liver cirrhosis patients after endotherapy. Westaby et al. conducted a prospective, randomized, controlled trial of injection sclerotherapy in the long-term management of variceal bleeding in 1977. They found a significant improvement in survival in the sclerotherapytreated patients [14]. After variceal obliteration was achieved, episodes of rebleeding were reduced by almost 10-fold, which showed that patients benefit from endotherapy, since variceal bleeding was the leading cause of death in such patients.
The median survival after decompensation was 2-4 years, which was demonstrated by the first appearance of variceal bleeding, ascites, encephalopathy, and jaundice [15]. A cohort study of 174 patients hospitalized for GOV bleeding, with an overall mean follow-up of 22 months, showed that the cumulative rebleeding rate at one year was 30% and the cumulative survival rates at 1 year and 5 years were 73.4% and 37%, respectively [16]. Another study enrolled 39 patients with Child C who had undergone endoscopic treatment and showed that the 3-year survival was 28.2% [17]. Yet another study enrolled 49 patients, with a mean followup duration of 30 months and showed that the rebleeding rate at one year after endotherapy was 45% [18]. In this study, which had longer follow-up duration and larger sample size, the 3-and 5-year cumulative survival rates were 75.3% and 52.8%, respectively, while the rebleeding rate at one year was 33% in nonviral cirrhosis patients and 23% in viral cirrhosis patients. The prognosis seemed better than previously reported studies. For patients with repeated variceal hemorrhage, endotherapy may be helpful for long-term survival. Meanwhile, the exclusion of preexisting HCC may have a slight effect on the results.
The most important finding of this study was that the viral cirrhosis population had a higher survival rate than the nonviral cirrhosis population. In the 15 months after endotherapy, the survival rate was higher in the nonviral cirrhosis population, but after 15 months, it was higher in the viral cirrhosis population. The reason might be that the long-term use of antiviral treatment could improve the patients' survival. The rebleeding rate was much lower in the viral cirrhosis patients than in the nonviral cirrhosis patients.

Gastroenterology Research and Practice
In China, most of the viral cirrhosis patients suffer from CHB, and the approval of oral antiviral agents has greatly improved the prognosis [19,20]. In CHB patients, it was reported that hepatic decompensation was associated with high mortality and the 5-year survival rate was 14-35% [21], and the overall survival was 80% at two years after the onset of decompensation [22]. In 2015, Jang et al. reported that patients with decompensated hepatitis B virus-related cirrhosis who took early antiviral treatment had a 5-year survival probability of 59.7% [23], which was lower than 67.1% reported in the present study. This may be owing to improved medical management of cirrhosis and the repeated endotherapy procedures. With standardized antiviral therapy and repeated endotherapy, fewer patients died of AVB, thus improving the survival rate.
We further analyzed the survival difference between each subgroup in the nonviral cirrhosis population. The long-term survival rate in the autoimmune cirrhosis patients was much lower, while the survival rates of cryptogenic cirrhosis patients and alcoholic cirrhosis patients were nearly identical. The rebleeding rate was significantly higher in the alco-holic cirrhosis patients than in the cryptogenic cirrhosis patients and autoimmune cirrhosis patients.
Cryptogenic cirrhosis is a diagnosis of exclusion when there is no other known identifiable etiology [24]. It is suggested that many cryptogenic cirrhosis patients may have evolved from NASH [25]. In this study, there was a high proportion of patients with cryptogenic cirrhosis (94 patients). The reason for this is that our medical center is a national referral center for liver diseases, so the number of unexplained cirrhosis patients referred to our center is far more than those of other hospitals. We interviewed and studied these patients and found that a large proportion of these cryptogenic cirrhosis patients had type 2 diabetes mellitus (66/95) and present or previous obesity (41/95). Hence, we hypothesized that the majority of cryptogenic cirrhosis patients in this study were actually "burnt-out NASH".
In the present study, most of the autoimmune liver cirrhosis patients were primary biliary cirrhosis (58.5%). A unique feature of primary biliary cirrhosis is the development of varices before the onset of cirrhosis. Esophageal varices after initial variceal bleeding is about 50% [26,27]. Esophageal varices were a predictor of higher mortality risk, which may be the reason for the low survival rate of the autoimmune subgroup.

Gastroenterology Research and Practice
Alcoholic liver cirrhosis is a "malignant" disease with an unfavorable prognosis involving a mortality of 49% and 90% after 1 and 15 years of follow-up, respectively. Almost 60% of patients died as a result of bleeding, hepatic coma, or a combination of these two complications. The 5-year survival of alcoholic liver cirrhosis with variceal bleeding was 27% [28,29]. Another populationbased cohort study of 11873 persons with diagnosis of alcoholic liver disease found that the absolute 5-year survival rate in alcoholic liver cirrhosis was 0.29 (95% CI: 0.28-0.30). The majority of alcoholic liver disease patients (65%) died from alcohol-related cause, i.e., alcoholic liver disease and its direct complications [30]. The high proportion of persistent drinkers among patients with variceal bleeding contributed to the high mortality. Abstinence is the simplest way to control the etiology. Among the 211 alcoholic cirrhotic patients, 148 were nondrinkers while 63 were drinkers, the median survival time of the abstinence group was 30.6 months (95% CI: 22.6-38.7), and the median survival time of the drinking group was 33.2 months (95% CI: 23.2-43.2), with no significant differences between the two groups (p = 0:675). Besides, logistic regression analysis showed that neutrophils is the only risk factor for variceal rebleeding (univariate analysis: p = 0:022, OR = 1:128, 95% CI: 1.017-1.25; multivariate analysis: p = 0:026, OR = 1:152, 95% CI: 1.017-1.300). No dose-response relationship was found between alcohol consumption and liver injury, so alcohol consumption may have no effect on variceal rebleeding [31].
As in previous studies, indicators of the severity of hemorrhage (hemoglobin) and severity of liver dysfunction (MELD score, albumin, and bilirubin) were prognostic fac-tors for cirrhotic patients with variceal bleeding. These factors were predictors for short-term outcomes (such as 5day treatment failure and 6-week mortality), as well as long-term outcome (5-year mortality) [32]. The predictive role of GGT for 5-year mortality may suggest liver damage. GGT is routinely used in clinical practice as an indicator of liver injury, and plays an important role in protecting cells against oxidants produced during normal metabolism [33]. GGT is also associated with the risk of all-cause mortality [34]. NEU% but not LYM% was included in the factors associated with 5-year mortality in the multivariate analysis, perhaps because LYM% mainly indicates hypersplenism and indirectly responds to portal hypertension. Indicators of 5-year mortality were related to liver disease per se rather than the severity of bleeding in general.
This study had several limitations. First, due to its retrospective study design, there may have been missing data due to recall bias among patients who were contacted by telephone. Second, endotherapy was strictly in accordance with the guidelines, no data suggested that combined endotherapy has any effect on rebleeding rate, so the GOV type and treatment option were not considered in this study. Third, the study was performed in a single center, and the choice of endoscopic treatment had subjective bias. Hence, further verification in other centers is needed. Fourth, the aim of this retrospective study was to evaluate the survival of patients with repeated endotherapy. Patients with other interventions for varices were excluded, 363 patients were lost to follow-up, and there were slight differences in MELD score between the viral and nonviral groups (4 vs. 6). Hence, selection bias may exist.

Conclusion
Repeated endotherapy combined with effective antiviral therapy is helpful for long-term survival of cirrhotic population with variceal hemorrhage and HBV or HCV infection.