Introduction
Unresectable malignant hilar biliary obstruction could be caused by a wide variety
of neoplasms [1]
[2]. Certain malignancies such as Klatskin tumor or hilar cholangiocarcinoma can lead
to intrinsic obstruction, while other tumors, including gallbladder carcinoma, hepatomas,
pancreatic adenocarcinoma, and metastases contribute to extrinsic obstruction, which
ultimately lead to obstructive jaundice [3]
[4].
Thus, palliative biliary drainage is absolutely crucial for improving the quality
of life of patients with unresectable malignant hilar biliary obstruction. Percutaneous
or endoscopic drainage as well as surgical bypass are currently the mainstay of palliative
treatment [5]
[6]. Due to the limited survival of these patients, minimally-invasive procedures such
as endoscopic drainage may be more appealing. Furthermore, prior studies have shown
that comparing to surgical bypass procedures, endoscopic drainage could provide better
outcomes and lower complication rate in the palliative management of patients with
unresectable malignant hilar obstruction [7]
[8].
Endoscopic drainage in patients with unresectable malignant hilar obstruction entails
using a plastic or self-expandable metal stent (SEMS). Despite similar technical success
rates, metallic stents have been shown to provide longer patency than their plastic
counterpart [6]; however, there is no consensus regarding the best treatment option for palliative
treatment of patients with unresectable hilar obstruction. In addition, debate remains
on the superiority of bilateral stenting over unilateral stenting. While in unilateral
stenting there are concerns over suboptimal efficacy in relieving jaundice and bacterial
contamination of the undrained segment [9]
[10], the technical complexity of bilateral stenting has been a major drawback [11]. Therefore, given conflicting outcomes with different endoscopic strategies, we
performed a systematic review and meta-analysis to determine the overall technical
and functional success rates, as well as the adverse event (AE) rate of unilateral
and bilateral metallic stenting and to compare outcomes in management of patients
with unresectable malignant hilar obstruction.
Materials and methods
This systematic review was performed in accordance with the guidelines of the Preferred
Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [12] and meta-analysis of observational studies in epidemiology (MOOSE) [13].
Literature search and study selection
A comprehensive literature search strategy was developed by an experienced medical
reference librarian (JN). Four electronic databases – Medline, Embase, Cochrane, and
ISI Web of Sciences – were screened to identify relevant studies published through
June 1, 2018. The following MeSH and keyword search terms were used: “unilateral stent
or drainage, bilateral stent or drainage, malignant hilar obstruction, Klatskin tumor,
biliary tract neoplasm, gallbladder tumor”. Literature search was conducted by two
investigators (MAM, DS), independently. The retrieved records were screened on the
basis of the title and abstract, and the eligible articles were obtained for full-text
review. Only English language articles were included. The bibliography of retrieved
articles were manually reviewed to reveal any additional relevant studies.
Inclusion and exclusion criteria
Inclusion criteria were determined by two investigators (MAM, MAK). Studies were eligible
for inclusion if they described outcomes of unilateral or bilateral endoscopic hepatic
duct drainage by SEMS in management of patients with unresectable malignant hilar
obstruction. Randomized clinical trials (RCTs) or observational studies that reported
technical and functional success rates as well as early and late AEs. Technical success
was defined as successful stent insertion across the hilar stricture. Functional success
was defined as 50 % decrease in bilirubin level within 14 days or a 75 % decrease
in bilirubin level within a month. Early AEs were defined as any procedure-related
complications that occurred within 30 days, while late AEs were described as complications
occurring later than 30 days. Cholangitis was reported if the patient had fever, leukocytosis,
and increasing bilirubin. Stent occlusion was defined as increase in bilirubin, recurrence
of jaundice, and presence of imaging findings suggestive of intrahepatic bile duct
dilation.
Individual case reports or case series with 10 or fewer patients were excluded. Published
abstracts of national or international conferences were not included in the study
because assessment of risk of bias in them was not possible. Studies that only reported
outcomes of plastic stenting were also excluded. Titles and abstracts of obtained
records were screened by two investigators (MAM, DS), independently. All eligible
articles were downloaded into EndNote 7.0 (Thomson ISI ResearchSoft, Philadelphia,
Pennsylvania, United States) and duplicate studies were removed. For final inclusion,
eligible studies were reviewed at the full-text level to determine their eligibility.
Any disagreement was resolved through discussion with the senior author (MAK).
Data extraction and quality assessment
The following data were extracted from the included studies: study design, year of
publication, number of patients, age, sex, type of stent (plastic or SEMS), stenting
strategy (unilateral or bilateral), technical success rate, functional success rate,
occlusion rate, early and late AEs, post-procedure cholangitis, stent patency, survival
and follow-up duration. Two investigators (MAM, DS), independently assessed the quality
of included studies using the National Institutes of Health (NIH) quality assessment
tool for before–after (pre-post) studies with no control group and the Cochrane tool
for RCTs [14].
Statistical analysis and data synthesis
Weighted pooled rates (WPR) along with their 95 % confidence intervals (CI) and predictive
interval were calculated for technical success, functional success, early and late
AEs, post-procedure cholangitis, and stent occlusion. The weight of each study was
calculated by the inverse of its variance. Subgroup analysis was performed to compare
effect size in studies that employed unilateral or bilateral stent. Cochran Q test
and I2 were assessed for heterogeneity across studies [15]. Presence of significant heterogeneity was considered when P < 0.1 for the Cochrane Q test. The I2 values of 0–50 %, 50 % to 75 %, and 75 % to 100 % were interpreted as low, moderate,
and high level of heterogeneity, respectively [16]. Due to heterogeneity between types and sizes of stents, DerSimonian–Laird random-effects
model of meta-analysis was applied [17]
[18]. A random effects model was also used to combine studies within each subgroup. P < 0.05 was considered significant.
A sensitivity analysis was performed to determine if one-arm studies had a disparate
effect on the pooled effect. We calculated the pooled effect with only two-arm studies.
With this approach, if the pooled effect does not change significantly, then the one-arm
studies do not disturb the pooled effect. Visual evaluation of funnel plots and the
Begg–Mazumdar regression test were performed for evaluation of presence of publication
bias [19]. If publication bias was present, Duval and Tweedie’s ‘‘trim and fill’’ method was
used to adjust effect size, accordingly. All analysis was performed by one investigator
(MAM), using Comprehensive Meta-Analysis software (version 2.0; Biostat; Englewood,
New Jersey, United States).
Results
Literature search
Based on our search strategy, we identified 35 records from PubMed, 49 records from
Embase, 22 records from Cochrane, and 47 records from ISI Web of Science. Fifty-six
duplicate records were removed and an additional 63 studies were excluded after screening
titles and abstracts. Thirty-four studies were selected for the full-text review and
16 studies were excluded due to our exclusion criteria; one study [20] was excluded due to serial insertion of stents and another study [21] was excluded due to possible population overlap with another included study [22]. Finally, 18 studies with 911 patients were included in our meta-analysis. Our search
strategy and study selection are summarized in a PRISMA flowchart ([Fig. 1]).
Fig. 1 PRISMA flowchart. From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group
(2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA
Statement. PLoS Med 6(7): e1000097. doi:10.1371/journal.pmed1000097.
Study characteristics and quality assessment
Among included studies, 10 [22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31] were retrospective and the rest [32]
[33]
[34]
[35]
[36]
[37]
[38]
[39] were prospective. Two studies [32]
[33] were RCTs. Thirteen studies [23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[34]
[35]
[38]
[39] were single-center and the other five studies [21]
[32]
[33]
[36]
[37] were multicenter. Two studies [38]
[39] evaluated only unilateral stents, 11 studies [21]
[23]
[25]
[26]
[27]
[30]
[31]
[34]
[35]
[36]
[37] evaluated only bilateral stents and five studies [24]
[28]
[29]
[32]
[33] compared unilateral vs. bilateral stents.
Among 911 patients with malignant hilar obstruction, cholangiocarcinoma was the most
frequent etiology (70 %). Seventeen percent of patients had stricture due to gallbladder
carcinoma, and 13 % of patients had obstruction related to other causes. One study
[39] did not report the etiology. According to the Bismuth classification, 29.25 % of
cases were type II, 38.5 % of cases were type III, and 32.25 % of cases were type
IV. Two studies [23]
[28] did not report Bismuth classification. Demographic data and study characteristics
of included studies are described in [Table 1] and [Table 2].
Table 1
Demographic and technical data from single-arm studies.
Study
|
Stent
|
Number of patients (male)
|
Age (mean ± SD)
|
Technical success, %
|
Functional success, %
|
Stent patency, (day, median)
|
Survival, (day, median)
|
Follow-up
|
De Palma et al 2003
|
Unilateral stent
|
61 (40)
|
62.5 ± 11.2
|
96.7
|
100
|
169
|
140
|
N/A
|
Singh et al 2004
|
Unilateral stent
|
18 (7)
|
53.7 ± 11.1
|
100
|
100
|
N/A
|
N/A
|
N/A
|
Chahal et al 2010
|
Bilateral stent
|
21 (15)
|
63.7 ± 13.9
|
100
|
N/A
|
189
|
N/A
|
6.14 months
|
Hwang et al 2011
|
Bilateral stent
|
30 (20)
|
68 ± 8
|
86.7
|
100
|
140
|
176
|
5.8 months
|
Kim et al 2012
|
Bilateral stent
|
97 (48)
|
71 median
|
78.4
|
97.4
|
159
|
226
|
234 days
|
Kim et al 2009
|
Bilateral stent
|
34 (19)
|
68.3 median
|
85.3
|
100
|
186
|
239
|
21 months
|
Kitamura et al 2017
|
Bilateral stent
|
17 (9)
|
78 median
|
100
|
82
|
N/A
|
N/A
|
192 days
|
Law et al 2013
|
Bilateral stent
|
24 (19)
|
68 ± 13
|
100
|
N/A
|
N/A
|
N/A
|
97 days
|
Lee et al 2013
|
Bilateral stent
|
84 (48)
|
68.3 ± 15.8
|
95.2
|
97.5
|
238
|
256
|
N/A
|
Park et al 2016
|
Bilateral stent
|
31 (16)
|
67 ± 14
|
83.9
|
92.3
|
188
|
175
|
N/A
|
Park et al 2009
|
Bilateral stent
|
35 (21)
|
66
|
94.3
|
100
|
150
|
180
|
142 days
|
Yang et al 2018
|
Bilateral stent
|
43 (22)
|
72.9 ± 9.1
|
88.4
|
100
|
198
|
300
|
184 days
|
Dumas et al 2000
|
Bilateral stent
|
45 (28)
|
72
|
73.3
|
100
|
N/A
|
N/A
|
8.5 months
|
Table 2
Demographic and technical data from two-arm studies.
Study
|
Number of patients (male)
|
age
|
Technical success%
|
Functional success%
|
Stent patency, (day, median)
|
Survival, (day, median)
|
Uni
|
Bi
|
Uni
|
Bi
|
Uni
|
Bi
|
Uni
|
Bi
|
Uni
|
Bi
|
Uni
|
Bi
|
Lee et al 2017
|
66 (33)
|
67 (36)
|
74.1 ± 10.42
|
73.5 ± 10.42
|
100
|
95.5
|
95.3
|
87.5
|
139
|
252
|
178
|
270
|
Iwano et al 2011
|
63 (35)
|
19 (11)
|
71.6
|
66.6
|
95.2
|
89.4
|
N/A
|
N/A
|
133
|
125
|
170
|
184
|
Naitoh et al 2009
|
17 (9)
|
29 (12)
|
69 ± 14
|
70 ± 11
|
100
|
90
|
94.1
|
96.1
|
210
|
488
|
166
|
205
|
Mukai et al 2012
|
14
|
16
|
N/A
|
N/A
|
100
|
100
|
100
|
100
|
363
|
295
|
N/A
|
N/A
|
Liberato et al 2012
|
35
|
45
|
N/A
|
N/A
|
100
|
93.3
|
N/A
|
N/A
|
24 weeks
|
29 weeks
|
N/A
|
N/A
|
Uni, unilateral; Bi, bilateral
The Cochrane tool for assessing risk of bias was used for quality assessment of the
two RCTs. Performance bias was high due to awareness of endoscopists and patients
about stent type; however, risk of bias in other domains, including selection, detection,
attrition, and reporting was low. Quality assessment of the observational studies
was done by NIH quality assessment tool for Before-After (Pre-Post) Studies With No
Control Group. Nine studies [24]
[25]
[28]
[29]
[30]
[31]
[34]
[35]
[38] had good quality and seven studies [23]
[26]
[27]
[32]
[36]
[37]
[39] had fair quality ([Table 3]).
Table 3
Quality assessment of studies with NIH quality assessment tool for before–after studies
with no control group.
Criteria
|
De Palma et al 2002
|
Singh et al
|
Chahal et al
|
Hwang et al
|
Kim et al 2012
|
Kim et al 2009
|
Kitamura et al
|
Law et al
|
Lee et el 2013
|
Park et al 2016
|
Park et al 2009
|
Yang et al
|
Iwano et al
|
Naitoh et al
|
Liberato et al
|
Dumas et al
|
1. Was the study question or objective clearly stated?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
2. Were eligibility/selection criteria for the study population prespecified and clearly
described?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
3. Were the participants in the study representative of those who would be eligible
for the test/service/intervention in the general or clinical population of interest?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
4. Were all eligible participants that met the prespecified entry criteria enrolled?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
5. Was the sample size sufficiently large to provide confidence in the findings?
|
Yes
|
No
|
No
|
Yes
|
Yes
|
Yes
|
No
|
No
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
6. Was the test/service/intervention clearly described and delivered consistently
across the study population?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
7. Were the outcome measures prespecified, clearly defined, valid, reliable, and assessed
consistently across all study participants?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
8. Were the people assessing the outcomes blinded to the participants' exposures/interventions?
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
9. Was the loss to follow-up after baseline 20 % or less? Were those lost to follow-up
accounted for in the analysis?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
10. Did the statistical methods examine changes in outcome measures from before to
after the intervention? Were statistical tests done that provided p values for the
pre-to-post changes?
|
Yes
|
No
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
No
|
Yes
|
No
|
No
|
Yes
|
Yes
|
Yes
|
Yes
|
11. Were outcome measures of interest taken multiple times before the intervention
and multiple times after the intervention (i. e., did they use an interrupted time-series
design)?
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
12. If the intervention was conducted at a group level (e. g., a whole hospital, a
community, etc.) did the statistical analysis take into account the use of individual-level
data to determine effects at the group level?
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
N/A
|
Results
|
Good
|
Fair
|
Fair
|
Good
|
Good
|
Good
|
Fair
|
Fair
|
Fair
|
Good
|
Fair
|
Fair
|
Good
|
Good
|
Good
|
Good
|
NIH, National Institutes of Health
Meta-analysis
Unilateral vs. bilateral metallic stent
Among included studies, 18 [22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39] recruited the self-expandable metallic stent. The weighted pooled rate (WPR) of
technical success was 91 % (95 %CI: 88 –94 %, prediction interval: 72 –98 %), Cochrane
Q test P = 0.001, I2 = 55 %. The Begg–Mazumdar test revealed no publication bias (tau = 0.20, P = 0.17, two-tailed) (Supplementary Fig. 1). Subgroup analysis was used to compare effect size between unilateral and bilateral
stenting. The technical success rate was 97 % (95 %CI: 93 –98 %, prediction interval:
91 –97 %) and 89 % (95 %CI: 84 –92 %, prediction variable: 69 –97 %) in the unilateral
and bilateral stenting, respectively. The technical success rate was significantly
higher in patients who underwent unilateral metallic stenting as compared to those
who underwent bilateral metallic stenting (P = 0.003) ([Fig. 2a]). Sensitivity analysis by pooling data from two-arms studies also showed the higher
technical success rate in unilateral group (RR = 3.82, P = 0.03).
Fig. 2 Forest plots displaying weighted pool rate of a technical success and b functional success.
The WPR of functional success was 95 % (95 %CI: 92 –96 %, predictive interval: 87 –98 %),
Cochrane Q test P = 0.24, I2 = 18%. There was no publication bias according to the Begg–Mazumdar regression test
(tau = 0.05, P = 0.74, two-tailed). The WPR of functional success of unilateral stenting was 96 %
(95 %CI: 90 –98 %, predictive interval: 84 –99 %) and bilateral stenting was 94 %
(95 %CI: 91 –96 %, predictive interval: 79 –98 %). There was no statistically significant
difference between the functional success rate of unilateral and bilateral stenting
(P = 0.481) ([Fig. 2b]). Sensitivity analysis also showed no statistically significant difference between
two groups (RR = 2.18, P = 0.22).
In total, 60 early complications happened in the 517 patients. The WPRs of early complication
were 15 % (95 %CI: 7–30 %) and 9 % (95 %CI: 6 –15 %) in unilateral and bilateral SEMS,
respectively ([Fig. 3a]). Cholangitis was the most common type of early complication, which occurred in
5.5 % of unilateral and 6.3 % of bilateral groups. In 550 patients, 550 late complications
were reported. The WPR of late complication was 43 % (95 %CI: 25 –64 %) in the unilateral
group and 45 % (95 %CI: 33 –56 %) in the bilateral group ([Fig. 3b]). Obstruction after 30 days occurred in 33.7 % of the unilateral group and 33.1 %
in the bilateral group. Fifty-two percent of cases with stent occlusion were managed
by insertion of a plastic stent through the metal stent. The remaining patients were
managed by SEMS (20 %), percutaneous drainage (24 %), nasobiliary drainage (2%), and
cleaning of sludge (2 %). Subgroup analysis demonstrated no statistically significant
difference in AEs between unilateral and bilateral metal stent drainage. No mortality
related to the procedure was reported.
Fig. 3 Forest plots displaying weighted pool rate of a early complications and b late complication.
Discussion
The majority of patients with malignant hilar biliary obstruction present with non-curative
disease which makes palliative care the mainstay of management. Surgical biliary bypass
procedures have a higher risk of morbidity and mortality with comparable efficacy
to other therapeutic options [40]
[41]
[42]. Therefore, endoscopic and percutaneous stent insertions are frequently used for
management of advanced hilar carcinoma. Percutaneous stenting requires external tube
placement and is associated with longer hospital stays, which could be less comfortable
for patients [7]. Thus, endoscopic biliary drainage has been introduced as a less invasive alternative
for palliation of biliary obstruction and is considered as the first choice for management
of malignant hilar obstruction [43].
The findings from our meta-analysis revealed that both unilateral and bilateral hilar
stenting are associated with comparable efficacy and safety for palliative management
of patients with malignant hilar obstruction. It should also be noted that a high
level of heterogeneity was revealed in the technical success rate of metallic stenting.
This could be explained by the different etiologies of malignant hilar obstruction
in the patient populations of the included studies. In addition, hilar cholangiocarcinoma,
which was the most common etiology among included patients, is anatomically categorized
by Bismuth classification into five different groups and the pooled population consisted
of patients with various Bismuth classification. Unfortunately, we were unable to
perform subgroup analysis based on etiology or Bismuth classification because the
included studies did not report their outcome based on these covariates.
Our findings also suggest that the technical success rate of unilateral metallic stenting
was significantly higher than that for bilateral metallic stenting. This may be due
to the challenges that are associated with insertion of a bilateral metallic stent.
Prior studies have also reported on the technical challenges of insertion of a bilateral
metal stent, and recommended stricture dilation before stent insertion as a solution,
to improve the technical success rate of second metallic stent insertion [44]. In addition, multiple techniques have been described for bilateral stent insertion.
Most of them were developed as to “stent-by-stent” or “stent-in-stent” techniques
[27]. However, detailed categorization based on different insertion techniques was not
possible because the included studies did not report their outcomes based on these
categories. Further investigations are warranted to compare outcomes of different
insertion techniques and determine the most effective method. In our meta-analysis
no significant difference was observed in terms of functional success or early and
late complications between the unilateral and bilateral metallic stenting groups.
These findings are in contrast with a previous systematic review which showed lower
overall, early and late complication rates for bilateral metallic stent [44]. This discrepancy may be explained by the distinct differences in the methodology
and execution of these two studies. The mentioned systematic review included published
abstracts and no quality assessment was performed for observational studies whereas
in our meta-analysis, we excluded published abstracts due to the discrepancy between
full-text articles and published abstracts [46]
[46]. Small case series were also excluded to decrease the effects of small studies on
the results. We also performed a quality assessment for both RCTs and observational
studies.
We acknowledge that our study has several limitations. First, the majority of included
studies were observational, which could raise concerns about presence of selection
bias. Second, we could not categorize our results based on type of malignant neoplasm
or bismuth classification because the included studies did not report patient outcomes
based on their etiologies and classification. Finally, included studies used different
types, brands, and sizes of stents, which couln increase the level of heterogeneity
of our results.
Conclusion
In conclusion, the findings from our study suggest that both unilateral and bilateral
drainage of malignant hilar obstruction can effectively relieve biliary obstruction.
Although unilateral metallic stent insertion had a significantly higher technical
success rate, no significant difference was observed in terms of functional success
rates. Early and late complication rates were also comparable between the two groups.