Introduction
Esophageal variceal (EV) bleeding is one of the most life-threatening complications
of portal hypertension, with mortality rates of 15 % to 20 % [1]. The current recommendation is to hemodynamically stabilize the patient and promptly
initiate vasoactive drugs like terlipressin and/or somatostatin analogues, followed
by endoscopic ligation of EV, ideally within 12 hours of presentation [2]
[3]. However, 20 % to 30 % of patients can rebleed and become refractory to standard
treatment [4].
In refractory patients, treatment options include surgical and/or non-surgical creation
of a porto-systemic shunt that reduces the portal pressure and controlling the bleeding
by tamponade. Surgical options of shunt creation are as follows: portocaval (portal
vein and vena cava), mesocaval (mesenteric vein and vena cava), spleno-renal shunt
(proximal splenic vein and left renal vein), and externally reinforced shunts that
can be either mesocaval or portocaval. Non-surgical options include balloon tamponade,
trans-jugular intra-hepatic porto-systemic shunt (TIPS) that is created by an interventional
radiologist under fluoroscopic guidance and the placement of a self-expanding metal
stent (SEMS) [2]
[3].
Balloon tamponade was the most commonly used option in the past. However, it has fallen
out of favor due to risk of pressure-induced necrosis of the esophagus and to the
fact that it can only be used for, at most, 24–48 hours [5]. Use of TIPS is limited by technical difficulties and availability [6]. Studies published thus far report the use of SEMS as a ‘bridge’ therapy with majority
of the patients eventually being treated with EV banding and/ or TIPS [3]
[7]
[8]
[9]. Although the current Baveno consensus workshop recommends the use of SEMS, an evidence-based
approach to guide the use of SEMS in potentially life-threatening refractory EV bleeding
is not established.
The goals of this meta-analysis were to study the clinical outcomes of SEMS, and use
the clinical outcomes of TIPS as a comparator, in refractory EV hemorrhage.
Methods
Search strategy
We conducted a comprehensive search of several databases from inception to May 24,
2019. The databases included Ovid MEDLINE® and Epub Ahead of Print, In-Process and
other non-indexed citations, Ovid Embase, Ovid Cochrane Central Register of Controlled
trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. An experienced medical
librarian using inputs from the study authors helped with the literature search. Controlled
vocabulary supplemented with keywords was used to search for studies of interest.
The full search strategy is available in Appendix 1. The PRISMA and MOOSE checklist were followed and are provided in Appendix 2 and 3 [10]
[11].
Study selection
In this meta-analysis, we included studies that evaluated clinical outcomes of SEMS
and studies that evaluated outcomes of TIPS in patients with refractory EV hemorrhage.
Studies were included irrespective of the study sample-size, inpatient/outpatient
setting, and geography as long as they provided data needed for the analysis.
Studies done in the pediatric population (age < 18 years), and studies not published
in English language were our only exclusion criteria. In case of multiple publications
from the same cohort and/or overlapping cohorts, data from the most recent and/or
most appropriate comprehensive report were retained.
Data abstraction and quality assessment
Data on study-related outcomes in individual studies were abstracted onto a standardized
form by at least two authors (SRK, RK), and two authors (BPM, SC) did the quality
scoring independently. Primary study authors were contacted via email as needed for
further information and/or clarification on data.
The Newcastle-Ottawa scale for cohort studies was used to assess the quality of studies
[12]. This quality score consisted of 8 questions, the details of which
are provided in Supplementary Table 1.
Outcomes assessed
-
Pooled rate of all-cause mortality,
-
Pooled rate of immediate EV bleeding control,
-
Pooled rate of rebleeding, and
-
Pooled rate of adverse events.
Comparison analysis: The outcomes with SEMS were compared with the outcomes of TIPS.
Assessment methodology and definitions
The collected data was matched between the groups (SEMS and TIPS) before statistical
analysis. Although, this model of comparison is indirect, the approach is comparable
to a retrospective case-control study with matched groups [13]. Refractory EV bleeding was defined according to the Baveno IV and V guidelines:
fresh hematemesis or aspiration of > 100 ml of fresh blood via the nasogastric tube
beyond 2 hours after endoscopy and/or a drop of 3 g/dL in hemoglobin without blood
transfusion [5]
[14].
Rebleeding was defined as per the Baveno V guidelines: evidence of rebleeding from
portal hypertensive sources (hematemesis, malena, aspiration of > 100 mL of fresh
blood via the nasogastric tube beyond two hours after endoscopy and/or a drop of 3 g/dL
in hemoglobin without blood transfusion [14].
Statistical analysis
We used meta-analysis techniques to calculate the pooled estimates in each case following
the methods suggested by DerSimonian and Laird using the random-effects model [15]. When incidence of an outcome was zero in a study, a continuity correction of 0.5
was added to the number of incident cases before statistical analysis [16]. We assessed heterogeneity between study-specific estimates by using Cochran Q statistical
test for heterogeneity, 95 % prediction interval (PI), which deals with the dispersion
of the effects [17]
[18]
[19]. and the I2 statistics [20]
[21]. In this, values < 30 %, 30 % to 60 %, 61 % to 75 %, and > 75 % were suggestive
of low, moderate, substantial, and considerable heterogeneity, respectively [22]. Publication bias was ascertained, qualitatively, by visual inspection of funnel
plot and quantitatively, by the Egger test [23]. When publication bias was present, further statistics using the fail-Safe N test
and Duval and Tweedie’s “Trim and Fill” test was used to ascertain the impact of the
bias [24]. Three levels of impact were reported based on the concordance between the reported
results and the actual estimate if there were no bias. The impact was reported as
minimal if both versions were estimated to be same, modest if effect size changed
substantially but the final finding would still remain the same, and severe if basic
final conclusion of the analysis is threatened by the bias [25].
All analyses were performed using Comprehensive Meta-Analysis (CMA) software, version
3 (BioStat, Englewood, New Jersey, United States).
Results
Search results and population characteristics
From an initial 395 studies, 214 records were screened and 33 full-length articles
were assessed. Twenty-one studies were included in the final analysis, of which 12
studies reported on the outcomes with SEMS [7]
[8]
[9]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34] and 9 reported on the outcomes with TIPS [6]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42].
The schematic diagram of study selection is illustrated in Supplementary Fig. 1. Two SEMS studies had cohort overlap and the most comprehensive one was retained
for the analysis (Wright 2010 with Hogan 2009 and Zehetner 2008 with Hubmann 2006)
[32]
[34]
[43]
[44].
Baseline population characteristics were comparable between the SEMS and TIPS groups.
The mean and/or median age ranged from 46 years to 69 years, with a predominantly
male population (74 %). Twelve percent of the patients were Child’s A, 39 % were Child’s
B and the rest were Child’s C cirrhotics. The mean MELD score ranged from 16 to 29.37
in SEMS group and 15.5 to 20.9 in TIPS groups. The ELLA-CS stent (SX-ELLA Stent Danis,
Hradec Kralove, Czech Republic) was used in all SEMS studies with a stent indwell
time ranging from 1 to 30 days. Data was not available on the severity of the bleeding
and the interventional procedure was done on an emergency basis. The population characteristics
are described in [Table 1].
Table 1
Study and population characteristics.
|
Study
|
Design, period, center, country
|
Intervention
|
Age (years)
|
Total patients (n)
|
M/F
|
Meld
|
Child–Pugh score
|
Rebleeding
|
Mortality (all-cause)
|
Adverse events – total
|
Stent migration
|
Stent indwell time
|
Follow-up (d = days)
|
|
Dechene, 2012 [26]
|
Retrospective Case series, August 2007 to March 2011, Single center, Germany.
|
SEMS
|
63.8 (11)
|
8
|
6/2
|
29.37 (16–40)
|
A: 0, B:0, C:8
|
3
|
5n (60 d); 6n (Total)
|
0
|
0
|
11 d (4–17 d)
|
60 d
|
|
Drastich, 2016 [27]
|
Retrospective Case series, Jun 2011 to Nov 2014, Single Center, Czech Republic.
|
SEMS
|
52.9 (27–80)
|
14
|
7/7
|
nr
|
nr
|
nr
|
7
|
nr
|
2/13
|
9.5 d (7–26 d)
|
nr
|
|
Escorsell, 2016 [28]
|
Randomised controlled trial, March 2009 to January 2013 Multicenter, Spain.
|
SEMS
|
69 (40–81)
|
13
|
13/0
|
16.5 (9–32)
|
A:3, B:10, C:10
|
2/13 (15 d), 6/13 (42 d)
|
6n (42 d)
|
6
|
7
|
7 d (2–14 d)
|
42 d (6-weeks)
|
|
Fierz, 2013 [29]
|
Retrospective Case series, Oct 2010 to Oct 2011, Multicenter, Switzerland.
|
SEMS
|
57 (41–68)
|
7
|
5/2
|
27 (11–37)
|
A:0, B: 2, C: 5
|
0
|
2n (42 d)
|
2
|
2
|
12 h–5 d
|
42 d
|
|
Goenka, 2017 [7]
|
Retrospective, Apr 2012 to May 2016, Single center, India.
|
SEMS
|
53 ± 13.7 (27 to 72)
|
12
|
11/1
|
20.17 ± 5.97 (14 to 35)
|
nr
|
1
|
5
|
0
|
0
|
17.5 d (7–30 d)
|
30 d
|
|
Holster, 2013 [8]
|
Prospective Case series, Feb 2012 to Oct 2012, Single center, Netherlands.
|
SEMS
|
58(48–78)
|
5
|
3/2
|
21 (11–28)
|
A:1, B:1, C:1
|
1
|
3
|
1
|
1
|
11 d (6–214 d)
|
180 d (6 months)
|
|
Muller, 2015 [31]
|
Retrospective Case series, May 2011 to Mar 2014, Single center, Germany.
|
SEMS
|
64.2 (43–79)
|
11
|
8/3
|
16 (8–36)
|
A:1, B:6, C:3
|
1
|
3 (42 d)
|
9
|
7
|
12.1 d (5–24 d)
|
42 d
|
|
Mishin, 2013 [30]
|
Retrospective Case series, Moldova.
|
SEMS
|
46.92 (24–62)
|
12
|
8/4
|
nr
|
nr
|
1
|
3
|
5
|
5
|
nr
|
30 d
|
|
Pfisterer, 2019 [9]
|
Retrospective, Jan 2009 to Dec 2016, Multicenter, Austria.
|
SEMS
|
55.5 (SD 11.5)
|
34
|
28/6
|
Median = 18(IQR 10)
|
A:1, B:10, C:8
|
22
|
22
|
17
|
13
|
5 d (pt survived ≥ 14 days)
|
42 d (6 weeks)
|
|
Wright, 2010 [32]
|
Retrospective Case series, Mar 2007 to Jul 2008, Single center, United Kingdom.
|
SEMS
|
49.4 (18–70)
|
10
|
9/1
|
32 (23–39)
|
nr
|
3
|
5 (42 d)
|
1
|
0
|
9 d (6–14 d)
|
42 d
|
|
Zakaria, 2013 [33]
|
Retrospective Case series, Jan 2008 to Dec 2009, Single center, Egypt.
|
SEMS
|
55.60 ± 5.62 (18–65)
|
16
|
14/2
|
nr
|
A:2, B:8, C:6
|
2
|
4
|
7
|
6
|
2–4 d
|
nr
|
|
Zehetner, 2008 [34]
|
Retrospective, Jan 2003 to Aug 2006, Austria.
|
SEMS
|
56 (32–91)
|
34
|
33/1
|
nr
|
A:0, B:13, C:21
|
0
|
10 (60 d)
|
8
|
7
|
5 d (1–14 d)
|
60 d
|
|
Cello, 1997 [35]
|
Randomised controlled trial, Nov 1991 to Dec 1995, Multicenter, USA.
|
TIPS
|
48.8 (2.0)
|
24
|
19/5
|
nr
|
nr
|
3
|
5n (30 d)
|
nr
|
nr
|
nr
|
574.5 ± 109 d
|
|
Garcia-Pagan, 2010 [6]
|
Randomised controlled trial, May 2004 to Mar 2007, Spain.
|
TIPS
|
52 ± 10
|
32
|
21/11
|
15.5 ± 5
|
A:0, B:16, C:16
|
0
|
4
|
nr
|
nr
|
nr
|
14.6 ± 8.4 months
|
|
Garcıa-Pagan, 2013 [36]
|
Retrospective (Randomised controlled trial), Mar 2007 to Jan 2011, Spain.
|
TIPS
|
56 ± 12
|
45
|
34/11
|
16.5 ± 5
|
A:0, B:18, C:27
|
2
|
6
|
nr
|
nr
|
nr
|
13.1 ± 12 months
|
|
Monescillo, 2004 [37]
|
Randomised controlled trial, Jun 1997 to Nov 2000, Single center, Spain.
|
TIPS
|
56 ± 12 (32–75)
|
26
|
22/4
|
nr
|
A:3, B:11, C:12
|
3
|
8
|
nr
|
nr
|
nr
|
1 years
|
|
Orloff, 2012 [42]
|
Prospective Randomized controlled trial, Jul 1996 to Jul 2011, USA.
|
TIPS
|
49 (30–84)
|
78
|
56/22
|
nr
|
A:16, B:39, C:23
|
nr
|
66
|
nr
|
nr
|
nr
|
3 to 10 years
|
|
Popovic, 2010 [38]
|
Retrospective, Apr 1994 to Jan 2000, Single center, Slovenia.
|
TIPS
|
52.0 ± 13.2
|
50
|
29/21
|
nr
|
A:8, B:26, C:16
|
3
|
13
|
16
|
nr
|
nr
|
35.5 ± 19.6 months
|
|
Rudler, 2014 [39]
|
Prospective, Mar 2011 to Feb 2013, Single center, France.
|
TIPS
|
53.2 ± 9.0
|
31
|
24/7
|
20.9 ± 6.9
|
A:0, B:7, C:24
|
0
|
9
|
14
|
nr
|
nr
|
7.8 months
|
|
Shi, 2014 [40]
|
Retrospective, Jan 2006 to Dec 2011, Single center, China
|
TIPS
|
49.7 ± 9.0
|
48
|
30/18
|
NR
|
A:11, B:28, C:9
|
9
|
16
|
23
|
nr
|
nr
|
35.4 ± 18.7 months
|
|
Xue, 2012 [41]
|
Retrospective, Jan 2007 to Jun 2010, Single center, China
|
TIPS
|
51 ± 13
|
64
|
42/22
|
NR
|
A:23, B:30, C:11
|
nr
|
8
|
nr
|
nr
|
nr
|
20.7 ± 1.3 months
|
SEMS, self-expanding metal stent; TIPS, transjugular intrahepatic portosystemic stent;
MELD. model for endstage liver disease, nr, not reported
Characteristics and quality of included studies
One SEMS study was prospective, whereas rest were retrospective [8]. Three were from multicenter data [9]
[28]
[29]. No studies were population-based. Overall, four studies were considered high quality
and the rest were medium quality [9]
[28]
[29]
[34]. There were no low-quality studies. All TIPS studies in the comparator group were
considered high quality. The detailed assessment of study quality can be found in
Supplementary Table 1.
Meta-analysis outcomes
A total of 574 patients were included in the analysis from 21 studies. 176 patients
were treated with SEMS in 12 studies and 398 patients were treated with TIPS in 9
studies.
The pooled rate of all-cause mortality with SEMS was 43.6 % (95 % CI 28.6–59.8) and
with TIPS was 27.9 % (95 % CI 16.3–43.6) ([Fig. 1]). The pooled rate of immediate bleeding control with SEMS was 84.5 % (95 % CI 74–91.2)
and with TIPS was 97.9% (95 % CI 87.7–99.7) ([Fig. 2]). The pooled rate of rebleeding with SEMS was 19.4 % (95 % CI 11.9–30.4) and with
TIPS was 8.8 % (95 % CI 4.8–15.7) ([Fig. 3]).
Fig. 1 Forest plot, mortality.
Fig. 2 Forest plot, immediate bleeding control.
Fig. 3 Forest plot, rebleeding.
The pooled rate of technical success with SEMS was 88.3 % (95 % CI 81.7–92.7) and
with TIPS was 91 % (95 % CI 86.2–94.2). The pooled rate of all adverse events with
SEMS was 36.9 % (95 % CI 26–49.2) and with TIPS was 41.4 % (95 % CI 26.5–58.1). The
pooled rate of stent migration was 31.8 % (95% CI 22–43.5). (Supplementary Fig. 2,
3 and 4) The pooled results are summarized in [Table 2].
Table 2
Summary of pooled rates.
|
Outcomes
|
SEMS
|
TIPS
|
|
Pooled rate (95 % confidence interval)
95 % Prediction interval (PI), I2 values
|
|
Mortality
|
43.6 % (28.6–59.8)
PI: 18 to 73, 38
|
27.9 % (16.3–43.6)
PI: 2 to 88, 91
|
|
Immediate bleeding control
|
84.5 % (74–91.2)
PI: 50 to 97, 40
|
97.9 % (87.7–99.7)
PI: NA (due to limited number of studies)
|
|
Rebleeding
|
19.4 % (11.9–30.4)
PI: 6 to 50, 32
|
8.8 % (4.8–15.7)
PI: 2 to 33, 40
|
|
Technical success
|
88.3 % (81.7–92.7)
PI: 80 to 93, 0
|
91 % (86.2–94.2)
PI: 73 to 97, 26
|
|
All adverse events
|
36.9 % (26–49.2)
PI: 11 to 74, 52
|
41.4 % (26.5–58.1)
PI: 0.4 to 99, 29
|
|
Stent migration
|
31.8 % (22–43.5)
PI: 11 to 63, 41
|
NA
|
SEMS, self-expanding metal stent, TIPS: transjugular intrahepatic portosystemic shunt,
NA: not applicable
Validation of meta-analysis results
Sensitivity analysis
To assess whether any one study had a dominant effect on the meta-analysis, we excluded
one study at a time and analyzed its effect on the main summary estimate. On this
analysis, no single study significantly affected the outcome or the heterogeneity.
Heterogeneity
We assessed dispersion of the calculated rates using the prediction interval (PI)
and I2 percentage values. The PI gives an idea of the range of the dispersion and I2 tell us what proportion of the dispersion is true vs chance [19]. The calculated PIs and corresponding I2 values are reported in [Table 2]. The calculated PI was wide except for the pooled rates of technical success. However,
the I2 heterogeneity was mild to moderate except for the morality rate with TIPS. This means
that the reported pooled clinical outcomes may or may not be valid to the real-world
scenario.
Publication bias
Based on visual inspection of the funnel plot as well as quantitative measurement
that used the Egger regression test, there was evidence of publication bias (supplementary
figure 5, Eggers 2-talied P = 0.04). Further statistics using the fail-Safe N test and Duval and Tweedie’s “Trim
and Fill” test revealed that the impact of the possible publication bias appeared
to be minimal and would not change the calculated estimate or the conclusion of this
meta-analysis.
Quality of evidence
The quality of evidence was rated for results from the meta-analysis according to
the GRADE working group approach [45]. Observational studies begin with a low-quality rating and based on the risk of
bias, indirectness, heterogeneity, and publication bias, the quality of this meta-analysis
would be considered as low-quality evidence.
Discussion
Our study demonstrates that use of SEMS is associated with a pooled all-cause mortality
rate of 44 %, immediate bleeding control rate of 85 % and a rebleeding rate of 19 %.
A recent multicenter study by Pfiesterer et al reported a mortality rate of 47 % with
the use of SEMS in refractory EV bleeding and our results are on par with this study
[9]. We report a pooled all-cause mortality rate of 28 % with TIPS. The mortality rates
with SEMS and TIPS seemed comparable.
The pooled immediate bleeding control and rebleeding rates are the key findings of
this study. 85 % of patients achieved immediate bleeding control with SEMS, whereas
with TIPS 98 % of patients achieved immediate bleeding control. The pooled rebleeding
rate with SEMS was 19 % and with TIPS was 9 %. Based on our comparison method, the
probability that the proportion of patients undergoing TIPS having a successful outcome
seemed to be more than the ones having a SEMS placed. The pooled rate of technical
success with SEMS was 88 % and with TIPS was 91 %. Although the technical success
rates were comparable, the prompt availability of emergent TIPS continues to be an
issue at many centers due to limited resources and experienced personnel.
The Baveno VI recommendation on the use of SEMS in refractory EV hemorrhage is based
on its favorable safety profile when compared to balloon-tamponade [3]. Our analysis of the adverse events with SEMS revealed a pooled rate of 37 % and
was comparable to the pooled adverse event rate with TIPS, which was 41 %. Stent migration
is a significant problem and our analysis revealed that the stent migrated in approximately
one-third of the patients (31 %).
How does our study compare to other published reviews? The meta-analysis by McCarty
and Njei reported a pooled bleeding control rate of 96 % [46]. The pooled outcomes reported in that study is of questionable validity due to the
inclusion of studies that had overlapping cohorts (Hubmann 2006 with Zehetner 2008
and Hogan 2009 with Wright 2010) [32]
[34]
[43]
[44]. Another meta-analysis by Marot et al., had similar limitation and only reported
the mortality and adverse events [47]. Our study, on the contrary, has avoided studies with overlapping cohorts and we
have presented the pooled results in perspective to the pooled outcomes of TIPS in
refractory EV bleeding, thereby enabling a side-by-side comparison. The meta-analysis
by Shao et al., report that SEMS may be considered in patients with EV bleeding refractory
to conventional therapy and their pooled rates are comparable to this study [48]. The meta-analysis by Qi eta l on TIPS in acute EV bleeding reported that TIPS with
covered stents might improve the overall survival of high-risk patients with acute
EV bleeding [49]. The results are comparable to the pooled mortality rates reported in this study.
The strengths of this review are as follows: systematic literature search with well-defined
inclusion criteria, careful exclusion of redundant studies, inclusion of good quality
studies with detailed extraction of data and rigorous evaluation of study quality.
Our pooled rates are calculated from 176 patients treated with SEMS and 398 patients
treated with TIPS. There are limitations to this study, most of which are inherent
to any meta-analysis. The included studies were not entirely representative of the
general population and community practice, with most studies being performed in tertiary-care
referral centers. Our analysis had studies that were retrospective in nature contributing
to selection bias. It is practically impossible to compare SEMS to TIPS in refractory
EV bleeding by RCT methods and a network meta-analysis is not possible due to lack
of studies with a common comparator. Our study presents the results of SEMS and TIPS
side by side, however, our analysis has the limitation of retrospective comparison
and therefore we do not comment on the superiority and/ or inferiority of one modality
to other. Nevertheless, our study is the best available estimate in literature thus
far with respect to the clinical outcomes of SEMS and TIPS in refractory EV bleeding.
Conclusion
In conclusion, based on our meta-analysis, the use of SEMS in refractory EV bleeding
demonstrates acceptable technical success and immediate bleeding control. However,
the pooled mortality rate and rebleeding rate with TIPS seem to be lesser than SEMS.
We, unfortunately, are unable to validate the results of comparison between the two
modalities due to the limitations in our retrospective comparison methodology.
