Keywords
gastric varices - glue injection - safety and efficacy
Introduction
Gastric varices (GVs) develop in around 15 to 20% of patients with portal hypertension.[1] The bleeding of GVs is independent of hepatic venous portal vein gradient and carries
a short-term mortality risk of 45% and rebleeding risk at 5 years as high as 44%.[1]
[2] Currently, available treatment options include endoscopic modalities like variceal
banding, endoscopic cyanoacrylate (E-CYA) injection, endoscopic ultrasound (EUS)-guided
coil and glue (EUS-CG), and radiological techniques like transjugular intrahepatic
portosystemic shunts (TIPS), and balloon retrograde transvenous obliteration (BRTO).[3]
E-CYA has emerged as an effective treatment with a hemostasis rate of more than 90%.[4] However, this procedure has its limitations. These include the inability to assess
the true size of the GVs based on the endoscopic bulge, limited visualization during
active bleeding, and para-variceal injection leading to tissue necrosis and perforation.[5] EUS-guided therapy involves placing one or more coils into the GVs for obliteration.
This can be done with or without CYA injection. While EUS-guided therapies for GVs
are emerging, these therapies require significant expertise, are expensive, lack long-term
safety data, and are limited only to tertiary care centers. E-CYA has high short-term
rebleed rates that require significant reintervention as compared with EUS-CG.[5] However, no long-term efficacy is available for EUS-CG. On the contrary, E-CYA requires
considerably less expertise and carries few complications with experience.[6] In addition, specific training and modification in cyanoacrylate injection techniques
like lesser or no dilution of glue has been shown to reduce these complications leading
to faster glue polymerization, which does not require confirmation of obliteration
by EUS.[7] Hence, this study was done to explore the long-term efficacy and safety of E-CYA
for the treatment of GVs.
Materials and Methods
In this retrospective analysis, all patients with GVs who underwent E-CYA injection
in the department of gastroenterology from January 2016 to March 2023 were included.
Inclusion Criteria
Exclusion Criteria
-
Patients with a history of BRTO, TIPS, and shunt surgeries
-
GVs not requiring E-CYA
-
Patients with incomplete data
The demographic details, etiology of portal hypertension causing GVs, size, location,
and form of GV, sign of recent stigmata, amount of glue required, rebleed, and number
of reinterventions were entered into a preformatted data sheet.
Definitions
-
(1) GVs classification: GVs were classified according to Sarin et al.[1] GVs were classified into gastroesophageal (GOV) and isolated GV (IGV). GOV was subclassified
into GOV-1 when the GV was continuous with the esophageal varices (EV) and extended
along the lesser curve of the stomach, and GOV-2, where the GV extended from the EV
toward the gastric fundus. IGV was classified as IGV-1 wherein isolated GV occurred
at the fundus, and IGV-2, which were ectopic varices located in the antrum, corpus,
and around the pylorus. The form of the GVs was classified according to Hashizume
et al. It was classified into three types: tortuous (F1), nodular (F2), and tumorous
(F3).[8]
-
(2) Technical success: Injection of cyanoacrylate during endoscopy and hardening of
the GV on palpation with standard biopsy forceps.
-
(3) Rebleed: Clinically significant rebleed was defined as the occurrence of clinical
features of overt gastrointestinal bleeding (hematemesis or melena) in a patient who
had undergone glue injection anytime in the past or a drop of hemoglobin of more than
1 g/dL along with recent stigmata of bleed over the GV on repeat endoscopy. Early
rebleed was considered as bleeding from GVs during the first hospital admission. Any
bleed on a subsequent follow-up was considered a delayed rebleed.
-
(4) Obliteration of GV: Successful obliteration was defined endoscopically by probing
the GV with a tip needle sheath. Complete obliteration was confirmed by a hard consistency
of the injected GV.
-
(5) Failure: Failure was defined as the inability to control the bleeding despite
E-CYA injection or failure to localize the bleeding GVs due to poor visualization.
Endoscopy
Procedures were performed in the interventional endoscopy unit of the hospital. Esophagogastroduodenoscopy
(EGD) was done by three consultant interventional gastroenterologists with a minimum
experience of 5 years of E-CYA injections. All patients were started on octreotide
infusion as per institutional protocol. All the patients received prophylactic intravenous
ceftriaxone 30 minutes before the procedure. All endoscopists performed the procedure
with the intention of obliterating GVs. If the initial hemostasis failed, TIPS/BRTO
was offered to the patients as rescue therapy. Bleeding from GVs is defined as a varix
having any of the following features:
Technique
EGD was done using the Olympus-190 series with an injection needle flushed with distilled
water. After selecting the GV, the base of the GV was punctured using a 21G needle
and undiluted n-butyl 2 cyanoacrylate (Histoacryl glue [n-butyl-2-cyanoacrylate],
B. Braun Medical Inc., Bethlehem, Pennsylvania, United States) was injected in aliquots
of 0.5 to 1 mL. The amount of n-butyl 2 cyanoacrylate depended on the size of the
varices, but usually not more than 5 mL of glue was injected in a single varix in
one setting. Following injection, the catheter was immediately flushed with 1 to 2 mL
of distilled water to clear the glue from the catheter into the varix. The needle
was then retracted, followed by flushing of the needle with distilled water to keep
it patent.
All patients underwent a second endoscopy after 24 hours. Complete obliteration was
confirmed by a hard consistency of the injected varix on probing with standard biopsy
forceps. The soft varices were deemed patent and additional aliquots of E-CYA were
administered to achieve complete obliteration of all patent GVs. Octreotide infusion
was continued for a minimum of 3 days and all patients received 5 days of intravenous
ceftriaxone. All patients in our study were prescribed nonselective β-blockers following
endotherapy as part of standard secondary prophylaxis for gastric variceal bleeding.
Propranolol was titrated to achieve a target heart rate of 55 to 60 beats per minute,
while carvedilol dosing was adjusted to maintain a mean arterial pressure of above
65 mm Hg.
Follow-Up
Patients underwent repeat endoscopy at 1 month, 6 months, and 1 year postprocedure.
If a patient experienced rebleeding from GVs at any time between scheduled follow-up
visits, an additional E-CYA injection was administered. In cases where patent varices
were detected during follow-up endoscopy, a repeat E-CYA injection was performed to
achieve obliteration of the GVs.
Outcome Measures
The primary outcome was to evaluate the primary hemostasis, rebleeding during index
admission or at follow-up, mortality, and procedure-related complications. Primary
hemostasis was defined as the complete cessation of bleeding by the end of the procedure.
Rebleeding was identified by a drop in hemoglobin of more than 2 g/dL accompanied
by clinical signs of bleeding. All complications related to E-CYA were reviewed, including
embolic events and infections. The secondary outcome was a rebleeding rate based on
the etiology of portal hypertension examined in terms of the relationship to the Child-Turcotte-Pugh
(CTP) status, the type, form, and size of GVs.
Statistical Analysis
The statistical analysis was performed using SPSS software (IBM Corp. Released 2019,
IBM SPSS Statistics for Windows, Version 26.0, IBM Corp., Armonk, New York, United
States). Patients with missing data were excluded. Data normality was assessed with
the Shapiro–Wilk test. Categorical variables were evaluated with chi-square or K-Fisher's
exact test (if expected count < 5). The Kaplan–Meier method was applied to estimate
the time to rebleeding-free survival following successful variceal obliteration. Statistical
significance was determined at 5% level of significance.
Results
Demographics
Of the 113 patients who presented with gastric variceal bleeding, 96 met the inclusion
and exclusion criteria. A flowchart depicting the study design is given in [Fig. 1]. The majority was male (83.3%).
Fig. 1 Flowchart depicting the study design.
The mean age was 48.17 ± 6.07 years. Portal hypertension was attributed to cirrhosis
in 82 patients (85.4%), while 14 patients (14.6%) had extrahepatic portal vein obstruction
(EHPVO). Among the cirrhotic patients, 28 (34.1%) were classified as CTP class A,
40 (48.8%) as Child–Pugh class B, and 14 (17.1%) as Child–Pugh class C. The most common
type of GVs was GOV-2, present in 41 patients (42.7%), followed by GOV-1 in 28 patients
(29.2%). The majority of varices were classified as F2 in 51 patients (53.1%), followed
by F3 in 45 patients (45.8%) ([Table 1]).
Table 1
Demographic data, etiology of portal hypertension, classification of cirrhosis (CTP),
types and forms of gastric varices, and treatment interventions in the study population
|
Variable
|
Frequency (%)
|
|
Sex
|
|
Male
|
80 (83.3)
|
|
Female
|
16 (16.7)
|
|
Age (mean ± SD)
|
48.17 ± 6.07
|
|
Etiology of portal hypertension
|
|
Cirrhosis
|
82 (85.4)
|
|
EHPVO
|
14 (14.6)
|
|
CTP class
|
|
A
|
28 (29.2)
|
|
B
|
40 (41.7)
|
|
C
|
14 (14.6)
|
|
Concomitant esophageal varices
|
90 (93.75)
|
|
Portal hypertensive gastropathy
|
|
|
Mild
|
91 (94.8)
|
|
Severe
|
5 (5.2)
|
|
Endoscopic variceal ligation done for esophageal varices
|
61 (63.5)
|
|
Sarin's class
|
|
GOV-1
|
27 (28.1)
|
|
GOV-2
|
41 (42.7)
|
|
IGV-1
|
23 (24)
|
|
IGV-2
|
5 (5.2)
|
|
Form of varix
|
|
F1
|
1 (1)
|
|
F2
|
51 (53.1)
|
|
F3
|
44 (45.8)
|
|
Size of varix more than 1 cm
|
81 (84.4)
|
|
Glue aliquot injected per session (mL) (IQR)
|
3 (2–4)
|
|
Number of patients who underwent TIPS
|
1 (1.04)
|
Abbreviations: CTP, Child-Turcotte-Pugh; EHPVO, extrahepatic portal vein obstruction;
GOV, gastroesophageal varices; IGV, isolated gastric varices; IQR, interquartile range;
SD, standard deviation; TIPS, transjugular intrahepatic portosystemic shunt.
Rebleed and Mortality
The median follow-up period was 30.5 (12–49) months. Technical success, defined as
primary hemostasis achieved during the index episode of bleeding, was 98.9%. One patient
experienced a failed E-CYA injection due to a high number of GVs and massive bleeding
at the time of the procedure, which hindered adequate visualization. This patient,
who had alcohol-related cirrhosis, was the only case of intraprocedure mortality.
During the first hospital stay two patients had a rebleed. Both these patients were
managed successfully with a repeat E-CYA injection.
At follow-up evaluations, 62 of the 96 patients (64.58%) did not require further glue
injections. At the 1-month follow-up, four patients required additional aliquots of
E-CYA injections due to the presence of patent GVs, as confirmed by probing with biopsy
forceps. Clinically significant rebleeding occurred in 10 patients (10.4%) within
the first month, while 6 patients (6.3%) experienced rebleeding within 6 months, and
2 patients (2.1%) had rebleeding after 1 year. Additionally, two patients had rebleed
after 1 year (2.1%). Rebleed of all but one patient was managed successfully with
repeat E-CYA injection. One patient underwent TIPS due to recurrent bleed electively.
Furthermore, 10 patients (10.41%) experienced disease-related mortality during the
follow-up period. Of these 10 patients, 5 patients died due to sepsis secondary to
spontaneous bacterial peritonitis (SBP) and 2 patients due to sepsis other than SBP.
The other three patients died due to liver failure. There was no procedure-related
mortality. A flowchart depicting rebleed and treatment outcomes is given in [Fig. 2]. The median time to recurrence of GVs was 11 months.
Fig. 2 Rebleed and treatment outcomes in patients with gastric varices (GVs).
A Cox proportional hazard model was used to predict the recurrence of GVs. The recurrence
of GVs was dependent on the CTP score and size of the varix of more than 5 mm. It
was independent of location, size, and form of varix (p > 0.005) ([Table 2]). A 1-year recurrence of bleeding was plotted as a Kaplan–Meier curve with groups
of EHPVO and cirrhosis ([Fig. 3A]) and CTP class ([Fig. 3B]). The difference was assessed by a log-rank test. There were no complications related
to glue therapy in these patients.
Table 2
Comparison of rebleeding with the status of cirrhosis and varices
|
Patients
|
Rebleeding
|
p-Value
|
|
Absent (%)
|
Present (%)
|
|
|
CTP score
|
|
A
|
23 (82.14)
|
5 (17.85)
|
0.005
|
|
B
|
25 (62.5)
|
15 (37.5)
|
|
C
|
2 (14.28)
|
12 (85.71)
|
|
Sarin class
|
|
GOV-1
|
17 (62.96)
|
10 (37.03)
|
0.379
|
|
GOV-2
|
26 (63.41)
|
15 (36.58)
|
|
IGV-1
|
18 (78.26)
|
5 (21.73)
|
|
IGV-2
|
3 (60)
|
2 (40)
|
|
Form of the GVs
|
|
F1
|
1 (100)
|
0 (0)
|
0.794
|
|
F2
|
35 (68.62)
|
16 (31.81)
|
|
F3
|
30 (68.18)
|
14 (31.81)
|
|
Size of the varix
|
|
Less than 1 cm
|
11 (73.33)
|
4 (26.66)
|
< 0.005
|
|
More than 1 cm
|
65 (80.24)
|
16 (19.75)
|
Abbreviations: CTP, Child-Turcotte-Pugh; GOV, gastroesophageal varices; GV, gastric
varices; IGV, isolated gastric varices.
Note: p < 0.005 was considered significant.
Fig. 3 (A and B) Kaplan–Meier graph of gastric variceal rebleeding on follow-up: Child-Turcotte-Pugh
(CTP) A vs. B/C patients (p = 0.005). Kaplan–Meier graph of patients with extrahepatic portal vein obstruction
(EHPVO) vs. cirrhosis free of bleeding (p = 0.094).
Discussion
Currently, the recommended first-line treatment for acute gastric variceal bleeding
is E-CYA injection. This n-butyl 2-cyanoacrylate once in contact with blood solidifies
within a few seconds essentially plugging the varix. A glue cast extravasation is
seen after 2 to 4 weeks.[9] In our 10-year retrospective study, primary hemostasis was achieved in 98.9% of
the patients. E-CYA injection has been reported to have a high primary hemostasis
rate of 88 to 100%.[10]
[11]
[12]
[13]
[14]
The rebleeding rate after primary hemostasis in our study was 20.83% over a median
follow-up of 30.5 months. This is similar to the reports published in the literature.[9]
[10]
[15] In our study, five patients rebled from recurrence of GVs among CTP A patients.
This was significantly lower than that of CTP B/C patients. This is similar to two
other cohort studies that reported rebleeding rates of 29 to 35% after initial hemostasis
rates of over 95%.[16]
[17] However, few studies did not show significant differences in rebleeding rates among
cirrhotic patients with CTP grades.[16]
In the present study, we observed no complications associated with the E-CYA injection
procedure. A large study from China noted several intraprocedural complications. Needle
adhesion to the varix occurred in 1.4% and catheter blockage occurred during the injection
process in 2.7%. The overall complication rate varies between 1 and 10%.[18]
[19]
[20] Other complications include portal vein thrombosis, pulmonary thromboembolism, transient
fever, and gastric ulcers.[20] These complications are likely related to procedural factors—such as injection technique,
volume of glue employed, and rate of injection—rather than the intrinsic properties
of n-butyl 2-cyanoacrylate. By adhering to the appropriate injection protocols, including
immediate catheter flushing, utilizing the minimum volume necessary for effective
hemostasis, and administering the glue in small, intermittent aliquots, the incidence
of these complications can be significantly reduced.[21]
Interventional radiological procedures for the treatment of GVs include TIPS and BRTO.[2]
[22]
[23]
[24] Both these procedures have demonstrated efficacy in controlling bleeding. However,
these procedures are relatively invasive and require specialized expertise, which
is not always readily available, particularly in acute settings. The TIPS procedure
involves the creation of a portosystemic shunt between the portal vein and hepatic
vein to reduce portal pressure, but it carries the risk of complications, such as
increased hepatic encephalopathy. Due to its invasive nature and the need for specialized
expertise, TIPS is generally considered a second-line therapy for gastric variceal
bleeding, typically reserved for cases where endoscopic treatment has failed.[9] In our study, only one patient required TIPS for rebleeding during follow-up, and
the procedure was successful. Utilization of TIPS at our center has been limited,
which may have influenced patient selection. A randomized trial compared TIPS with
E-CYA injection for gastric variceal bleeding. E-CYA injection demonstrated superior
variceal obliteration; however, it was associated with a higher rate of rebleeding.[25] E-CYA injections are straightforward, effective, and can be performed bedside in
intensive care unit settings. Additionally, the criteria for patient selection regarding
TIPS/BRTO are influenced by CTP and Model for End-Stage Liver Disease scores, resulting
in more stringent eligibility requirements compared with those for cyanoacrylate injection.
In this study, more than 65% of patients did not require further instances of E-CYA
injection. This result is similar to other studies. One of the challenges during follow-up
is that not all GVs are visible during standard endoscopy. Deeper varices are often
detected through imaging techniques such as computed tomography (CT) or EUS.[15]
[26] In the absence of good randomized controlled trials, optimal approach for bleeding
GVs remains unaddressed.
In recent years, EUS-guided coil injection combined with cyanoacrylate or thrombin
has emerged as a promising alternative for treating patients with a history of gastric
variceal hemorrhage. Randomized trials have indicated that this approach is more effective
and associated with minimal complications.[27] However, implementing EUS-guided combination therapy in cases of acute gastric variceal
bleeding, especially with active spurting or oozing, may present challenges. The need
to switch endoscopic devices could introduce delays in endotherapy, potentially leading
to adverse outcomes. Additionally, this technique is associated with higher costs
and necessitates a level of expertise that may not readily be available in all centers
limiting its widespread applicability.
This study has certain limitations. It is a single-center, retrospective study with
a limited sample size. The retrospective design may have introduced variability in
follow-up adherence. Routine imaging was not obtained preprocedure or postprocedure.
CT imaging plays a valuable adjunct role in the assessment of bleeding GVs by identifying
their size, location, and feeding vessels, which can guide treatment decisions such
as cyanoacrylate injection, TIPS, or BRTO. It also helps detect spontaneous portosystemic
shunts and posttreatment complications. CT scans may delay urgent endoscopic intervention,
potentially wasting critical time in unstable patients. Cost and accessibility further
restrict its use in emergency settings. While CT was not routinely performed in this
study, it remains a useful adjunct in selected cases for optimizing management strategies.
The presence of ascites and concurrent hepatic encephalopathy were heterogeneous to
assess. However, this study demonstrates the efficacy of E-CYA injection for managing
bleeding GVs. The long follow-up of patients provided a detailed efficacy of E-CYA
injections and the natural history and recurrence of GVs.
Conclusion
E-CYA injection is an effective and safe procedure in bleeding GVs. The overall serious
short-term and long-term adverse events of E-CYA injections are rare. Even though
newer modalities like EUS-guided endotherapy has shown excellent results, their long-term
efficacy has not been well established. Our findings reinforce the role of E-CYA as
a primary treatment modality, while the potential benefits of EUS-guided therapy warrant
further investigation in future studies.
