TIPS vs. endoscopic treatment for prevention of recurrent variceal bleeding: a long-term follow-up of 126 patients

Gastroesophageal variceal bleeding (GEVB) is a severe complication of portal hypertension. In cirrhotic patients with a history of variceal bleeding, the incidence of GEVB within 1 year is 60%, while the mortality from each rebleeding episode is nearly 20%.1 In terms of prevention of recurrent bleeding, current guidelines recommend management of patients with the history of variceal bleeding. The first-line treatment for preventing recurrent variceal bleeding is pharmacologic treatment with non-selective β-adrenergic blockers (NSBB), or a combination of isosorbide mononitrate (ISMN) and nadolol combined with endoscopic treatment (ET), i.e. variceal sclerotherapy and/or variceal ligation.2, 3 In frequently recurring bleeding episodes, in patients unresponsive to pharmacological and endoscopic treatment, transjugular intrahepatic portosystemic shunt (TIPS) or surgical procedures (i.e. portocaval or splenorenal shunt) are the treatments of choice. TIPS is recommended as a “rescue-urgent” treatment if primary haemostasis cannot be obtained with endoscopic and pharmacological treatment, or if uncontrollable early rebleeding occurs within 48 hours.4, 5, 6 TIPS is used as an elective procedure after the second or third (and/or more) recurrent bleeding episode from varices (especially if repeated over short periods of time) in hemodynamically and clinically stable patients with optimally regulated risk factors for the complication of the procedure (i.e. improvement of coagulation factors, elimination or reduction of ascites, regulation of cardiac and renal function, and clinically significant improvement in hepatic encephalopathy). In most of the randomized studies, patients were included in the study 24 to 96 hours from the last bleeding.7, 8, 9, 10, 11 Studies which could evaluate the role of elective TIPS in comparison to combined ET and NSBB treatment in patients with recurrent variceal bleeding episodes are rare.8,12, 13 According to literature, there are even fewer studies which analysed the long-term effect of TIPS vs. ET, 30 months being the longest observation period in terms of survival.14 The purpose of our study was therefore to compare elective TIPS with combined ET and NSBB treatment in terms of their long-term efficacy in preventing recurrent GEVB in patients with portal hypertension.

This retrospective study included 126 patients with liver cirrhosis and recurrent GEVB episodes originating from ruptured esophageal and gastric varices. The inclusion criteria were: (1) at least three gastroesophageal variceal bleedings or two recurrent episodes of bleeding within a less than a month period; (2) < 1 month since the previous bleeding episode; (3) Child-Pugh score < 13; (4) technically successful TIPS procedure; (5) 18 < age < 75; (6) patient’s written informed consent. Exclusion criteria were: (1) patients who did not meet inclusion criteria; (2) chronic occlusion of portal vein; (3) hepatocellular carcinoma (HCC) or/ and other types of cancer (with the exceptions of non-melanoma skin cancer and in situ cervical cancer); (4) acute hepatitis.

Patients who were treated with elective TIPS were included in the study from the time of the procedure, i.e. on average 35 days after the last variceal bleeding episode. Those patients were hemodynamically and otherwise (in terms of disease) stable. ET patients were included in the study after the last variceal bleeding, that is after successful pharmacological and endoscopic eradication, i.e. on average 30 days after the last bleeding episode. All patients were followed-up with clinical evaluations, serum laboratory tests, and Doppler ultrasound before hospital discharge, in the outpatient clinic at 3 months after TIPS and every 6 months thereafter. Portal venography was performed only as an introduction to a re-intervention in patients with suspected or impaired shunt malfunction. Patients were followed-up from inclusion in the study until death, liver transplantation, the last follow-up observation or until the end of our study.

Primary endpoint of our study was rebleeding rate. Bleeding-related mortality and survival were considered as secondary endpoints.

The study took place at the Institute of Radiology and Department of Gastroenterology of University Medical Centre Ljubljana, and at the Department of Gastroenterology of University Medical Centre Maribor.

The study was approved by the National Medical Ethics Committee of the Republic of Slovenia (Number 94/11/11) and was in agreement with the Declaration of Helsinki.

Following the current guidelines for preventing recurrent gastroesophageal variceal bleeding, TIPS should not be used as a first-line treatment and should be limited to those bleedings, which are refractory to pharmacologic and endoscopic treatment.3, 4 This recommendation is mainly because the rate of hepatic encephalopathy is significantly higher in patients undergoing TIPS than in those receiving ET and NSBBs. Moreover, according to meta-analysis, deaths due to all causes do not differ between the two groups of patients.11, 16 Consequently, use of TIPS treatment has been limited worldwide in the last decade. However, in the present study, elective TIPS was found to be more effective than ET in terms of prevention of recurrent GEVB and was associated with a similar rate of encephalopathy, and with a similar survival rate which accords with the results of recently published studies.5,15, 16, 17, 18 These results were also in line with a meta-analysis that included mostly studies from 2000–2010.19 So far, most evidence for the use of TIPS for secondary prevention of GEVB comes from randomized studies published between 1995 and 2002 with patient’s follow-up period, on average 20 months (ranged from 15 to 37 months).7, 8, 9, 10, 11 In twelve studies, patients were included in the study 24 hours to 96 hours from the last bleeding and only in one study two weeks after the last bleeding.8 A meta-analysis of studies showed a small number of recurrent bleeding in the TIPS group (19%, range 9–40% vs. 44.4%, range 21–61% in the ET group). In these studies, they basically did not distinguish urgent TIPS from elective TIPS.

Studies which could evaluate the role of elective TIPS in comparison to ET in patients with recurrent variceal bleeding episodes are rare.12 There is even fewer studies which could analyse the longterm effect of TIPS vs. ET.13, 14

In our study, there was less recurrent bleeding episodes in the TIPS group of patients in comparison to the ET group of patients (21.4% vs. 66.1%, respectively), which accords with the results of previous comparable studies.17, 18, 19, 20 Most of the patients in our study had recurrent bleeding episode within the first two years (46.6% in the TIPS group vs. 83.7% in the ET group). Moreover, a total number of recurrent bleeding episodes (which required endoscopic intervention and hospitalization) was lower in the TIPS group; the difference between the two groups was statistically significant. Multivariate analysis identified ET as the only significant independent predictor of recurrent bleeding. The main advantage of TIPS procedure compared to ET seems to be due to the direct and controlled reduction of hepatic venous pressure gradient (HVPG) during procedure below the threshold value for variceal rupture and bleeding (i.e. < 12 mmHg) or ≥ 20% reduction of baseline HVPG value.20, 21 By reducing HVPG, we not only improve the rate of variceal rebleeding, but also reduce other complications of PH, such as ascites, and improve liver and kidney function. In our study, the mean reduction of HVPG was 35.9%. This correlates to previous studies, which showed that reducing HVPG > 20% below baseline value contributes to lower risk for recurrent bleeding, spontaneous bacterial peritonitis, ascites and death.21 Reducing of HVPG is therefore crucial for higher quality and longer survival of patients with liver cirrhosis. There are also fewer recurrent bleeding episodes in patients who are treated with surgical portosystemic shunts, which correlates with our asumption of direct impact of reduced HVPG on the course of disease.22 The recurrent bleeding in the TIPS group occurred most frequently due to shunt malfunction, and in more than half of patients, shunt malfunctions were successfully repaired with an additional balloon dilatation and stent or stent graft insertion. In the majority of patients, shunt malfunction was determined by Doppler ultrasound before the appearance of clinical signs. Fewer reintervention and rebleeding episodes were reported for studies where patients, as in the present study, had regular ultrasound monitoring.23

Because rebleeding is associated with increased risk of mortality, preventing variceal rebleeding may be a substitute outcome of survival.24, 25, 26 Based on our study, recurrent bleeding episode seems to be the leading cause of death in the ET group of patients and liver failure in the TIPS group of patients; differences were statistically significant (Table 3). Despite statistically significant lower mortality rate due to recurrent variceal bleeding episode in the TIPS group of patients, it did not result in improved long-term survival between the two groups. The TIPS group of patients had better 2-year survival rates, but the difference was not statistically significant. We assumed that other factors than rebleeding may have contributed to the observed mortality in both groups. Liver failure was the leading cause of death in patients who survived more than 2 years, which suggests that preserved liver function is the main predicting factor of long-term survival in patients with liver cirrhosis and portal hypertension, whereas occurrence of recurrent variceal bleeding only has a minor effect (5–10% based on our study).17

The incidence of hepatic encephalopathy before joining the study was the same in both groups and is comparable to literature.14,27, 28 Causes of the same frequency of HE are the most likely comparable clinical characteristics of patients prior to being included in the study: hemodynamically stable and under-conditions, patients with similar liver cirrhosis, the same number of patients with Child B and Child C hepatic impairment, and similar age of patients. Compared to previous studies, the incidence of HE in the TIPS group is lower in our study and slightly higher than in the ET control group, but the difference is not statistically significant, which accords with the results of comparable studies.27, 28

Our study has limitations which have to be mentioned. First, the most important is its retrospective nature, so the analysis is subject to potential patient selection bias. Our data recording was limited to the available medical records and documentation, so we cannot exclude some degree of underreporting due to inherent limitations of non-standardized clinical documentations outside of clinical studies. Second, because the primary endpoint is variceal rebleeding, the power calculation is primarily based on a difference in the rate of variceal rebleeding between both groups. Thus, the data regarding mortality should not be overemphasized. Third, this study is being conducted in a single tertiary centre with the TIPS technique experience. Accordingly, our findings might not be promptly generalized to other centres with less experience. Fourth, treatment of GEVB has improved over the past few decades in all fields of medicine, including the treatment of complications of liver cirrhosis and portal hypertension, EVL and the use of endoprosthesis in the TIPS group in later years as compared to earlier years, when EST and stent were commonly used. Despite these limitations, there was a large number of patients enrolled in this study, and they were followed for a long time period. Acorrding to available international literature, our study was the longest in terms of observation time of both, TIPS and ET, groups of patients (median observation time 47 months for the TIPS group and 40 months for the ET group).

In conclusion, TIPS compared to ET in combination with NSBB was more effective in the prevention of recurrent gastroesophageal variceal bleeding, had significantly lower mortality due to recurrent variceal bleeding, but did not result in long-term survival benefit. The incidence of hepatic encephalopathy was similar in both groups. Liver failure was the leading cause of death in patients surviving more than 2 years, which suggests that preserved liver function is the main predicting factor of long-term survival, whereas occurrence of recurrent variceal bleeding only has a minor effect.