Introduction
Anastomotic biliary stricture is the most common adverse event (AE) after orthotopic
liver transplantation (OLTX), occurring in 5 % to 19 % of patients [1]
[2]
[3]
[4]
[5]. The first-line approach to this type of stricture involves endoscopic retrograde
cholangiopancreatography (ERCP), with placement of multiple plastic stents (MPSs)
[6]
[7]
[8]
[9]. However, the benefits of using fully-covered self-expandable metallic stents (CSEMS)
in this situation are still being discussed [10]
[11]
[12]. The most significant drawback of using MPS is the number of ERCPs needed to complete
the treatment because the plastic stents have to be replaced approximately every 3
months; the number and diameter of the stents is increased until the stricture is
resolved [7]
[13].
CSEMS have been used to treat benign strictures with encouraging results because their
removal is no longer a limiting factor [10]. The initial premise is that only two ERCPs are needed to treat the biliary anastomotic
stricture after OLTX: one to place the metallic stent and another to remove it, leading
to greater acceptance by patients and potentially lesser costs and fewer AEs [13]. Case series with CSEMS have already demonstrated the safety of these devices, and
few studies on this topic have shown similar rates of treatment success, recurrence,
and AEs between the two types of treatment [14]
[15]
[16]
[17].
Therefore, the current systematic review and meta-analysis evaluates all the randomized
studies available in the literature to compare use of MPS and CSEMS with regard to
efficacy, safety, and cost in treating anastomotic biliary stricture after liver transplantation.
Patients and methods
This systematic review and meta-analysis were performed according to the recommendations
of Preferred Reporting Items for Systematic Reviews and Meta-Analyses and registered
in the international PROSPERO database (CRD 42017068478) [18].
Eligibility criteria
Only randomized clinical trials (RCTs) comparing use of MPS with CSEMS in initial
treatment of anastomotic biliary stricture after liver transplantation were included.
There were no restrictions with regard to language. Inclusion criteria were patients
aged > 18 who underwent OLTX and had stricture of the biliary anastomosis confirmed
via cholangiography. Exclusion criteria were recurrent biliary stricture, non-anastomosis
biliary stricture, and stricture in the hepatic hilum.
Search and selection of articles and collection of data
Two independent researchers conducted the search, evaluated and selected the articles;
disagreements were resolved by consensus. Databases searched were Medline (PubMed),
EMBASE, SciELO/LILACS, and Cochrane until October 2017. The search strategy in Medline
and EMBASE was “(Liver transplantation OR Hepatic Transplantation OR Liver Grafting
OR Hepatic Transplantation) AND (biliary stricture OR biliary stenosis OR biliary
stenose OR ERCP OR plastic OR metallic OR stent OR cholangiography OR cholangiopancreatography)”.
The search strategy in SciELO/LILACS and Cochrane was “Liver transplantation AND biliary”.
Outcomes evaluated were as follows: number of ERCPs performed, rate of stricture resolution,
rate of stricture recurrence, total treatment time, stent migration, cost, AEs, and
number of stents.
Risk of bias
Studies were individually evaluated with regard to proper randomization, allocation
concealment, number of losses below 20 %, relevant outcomes, presence of analysis
by intention to treat, risk bias table (RoB) tool recommended by Cochrane and Jadad
scale [19]
[20]. The RoB tool assists in measuring biases through evaluation of blinding, randomization,
and information on losses. Scores on Jadad scale range from 0 to 5, with scores below
3 generally considered to represent studies of low methodological quality. Evidence
quality was analyzed according to the Grading of Recommendations Assessment, Development
and Evaluation Working Group (GRADE) [21].
Statistical measures and analysis
Dichotomous variables were analyzed using risk difference and Mantel-Haenszel test,
whereas continuous variables were analyzed using mean difference and inverse variance.
The fixed effect model was preferred, but we used the random effect model when less
than 50 % heterogeneity was not reached.
When studies reported medians and ranges, these were transformed into mean and standard
deviations using the formula by Hozo et al. [22].
We used a 95 % confidence interval, and heterogeneity was calculated using the method
by Higgins (I-square).
Statistical analyses were performed with the RevMan 5 (Review Manager Version 5.3.5,
Cochrane Collaboration, London, UK) and OpenEpi (Open Source Epidemiologic Statistics
for Public Health) software. The relationship between sample size and effect for each
outcome was graphically analyzed using a forest plot. Funnel plots were used to assess
risk of publication bias or inconsistency between the study outcomes.
For cost analysis, it was necessary to convert the data from one study from Australian
dollars to US dollars [13]. The exchange rate for the month of publication was used (January 2014). Then, we
calculated the mean and standard deviation to perform the Student’s t-test.
Treatment times expressed in months were converted to days.
Results
Search
The literature search yielded 3,322 articles in Medline/PubMed, 2,447 in EMBASE, 127
in SciELO/LILACS, and 55 in Cochrane, totaling 5,951 articles. After repeated articles
were excluded, 4,650 studies remained. Among these, four RCTs were found that met
the inclusion criteria and were included in the meta-analysis, with a total of 205
patients (103 in the CSEMS group and 102 in the MPS group). The article selection
process is illustrated in [Fig. 1].
Fig. 1 Article selection process.
The four selected studies have similar characteristics ([Table 1]) [10]
[13]
[23]
[24]. CSEMS dwelling time ranged from 3 to 6 months and the interval to replace the plastic
stents ranged from 6 to 16 weeks. The study by Kaffes et al. [6] was the only one that used metallic prostheses with greater diameter at the ends
and did not mention minimum follow-up time; the other three studies [10]
[23]
[24] reported follow-up of at least 1 year. Data were extracted from published material
and there was no need to access original data from the studies.
Table 1
Study characteristics.
|
Study
|
P
|
I
|
C
|
Follow-up
|
Metallic stent time
|
Plastic stent exchange time
|
Plastic stents total time
|
Metallic stent characteristics
|
Stricture dilation
|
|
Kaffes A, 2014
|
20
|
10
|
10
|
26 (6 – 40)/25.5 (3.0 – 44) months
|
3 months
|
3 months
|
Up to 12 months (earlier if stricture resolution were observed)
|
Taewoong Medical (10-mm diameter at either end and 8-mm at the center)
|
Endoscopist discretion
|
|
Cote GA, 2016
|
73
|
37
|
36
|
At least 1 year
|
6 months
|
3 – 4 months
|
Up to 12 months (earlier if stricture resolution were observed)
|
Wallflex, Boston Scientific, 8- or 10-mm diameter
|
Dilated all patients from the plastic stent group and at endoscopist discretion at
metallic stent group
|
|
Tal AO, 2017
|
48
|
24
|
24
|
At least 1 year
|
4 – 6 months
|
6 – 12 weeks
|
No information
|
10-mm diameter, no anti-migration flaps
|
Endoscopist discretion
|
|
Martins FP, 2017
|
64
|
32
|
32
|
At least 1 year
|
6 months
|
3 months
|
12 months
|
Wallflex, Boston Scientific, 10-mm diameter, 60- or 80-mm length
|
Dilated all patients from the plastic stent group and at endoscopist discretion at
metallic stent group
|
P: population; I: intervention; C: control
Biases and evidence quality
Risks of individual biases of the four studies were restricted to double blinding,
which was not possible because of the need to evaluate the stricture when the stents
are placed and removed by the physician. In all studies, losses were less than 20 %
and randomization, appropriate allocation, and analysis by intention to treat were
performed; consequently, their Jadad score was 3 ( [Table 2]). [Fig. 2] shows the RoB tool recommended by Cochrane. The GRADE Summary of Findings for evidence
quality can be found in [Annex 2].
Table 2
Risk of bias and Jadad.
|
Study
|
Focal question
|
Appropriate randomization
|
Allocation concealment
|
Double blinding
|
Losts (< 20 %)
|
Prognosis characteristics
|
Outcomes
|
Intention to treat analysis
|
Sample size determination
|
Jadad
|
|
Kaffes 2014
|
Yes
|
Yes
|
Yes
|
No
|
Yes (0 %)
|
Yes
|
Yes
|
Yes
|
No
|
3
|
|
Cote 2016
|
Yes
|
Yes
|
Yes
|
No
|
Yes (8 %)
|
Yes
|
Yes
|
Yes
|
Yes
|
3
|
|
Tal 2017
|
Yes
|
Yes
|
Yes
|
No
|
Yes (17 %)
|
Yes
|
Yes
|
Yes
|
Yes
|
3
|
|
Martins 2017
|
Yes
|
Yes
|
Yes
|
No
|
Yes (7,8 %)
|
Yes
|
Yes
|
Yes
|
Yes
|
3
|
Fig. 2 Risk of bias tool.
Outcomes
We were able to compare several outcomes, such as number of ERCPs performed, stricture
resolution, stricture recurrence, treatment time, migration, AEs, costs, and number
of stents ( [Annex 3]). Tal et al. [11] considered the result after the required crossovers (four in the CSEMS and three
in the MPS groups), whereas Martins et al. [13] considered a failure when crossover was necessary. Because the number of crossovers
was similar and they all resulted in stricture resolution, we agreed with the assessment
by Tal et al. [11] to consider these as successful cases.
Number of ERCPS performed
The four studies were included in the analysis of the number of ERCPs performed, with
a total of 205 patients. The number of ERCPs necessary for treatment was lower (MD:
−1.86; 95 %CI [−3.12 to −0.6]) in the patients in whom CSEMS were used. Heterogeneity
observed in the analysis exceeded 50 %, and funnel plot was used; no outliers were
identified, and treatment heterogeneity was not possible; therefore, we used the random
effect model for analysis ([Fig. 3]).
Fig. 3 Forest Plot of number of ERCPs performed.
Number of stents per patient
The number of stents per patient was lower (MD: −10.633; 95 %CI [−20.82 to −0.44])
in the patients in whom CSEMS were used. Heterogeneity observed in the analysis exceeded
50 %, and because only two studies were involved, treatment heterogeneity was not
possible and it was necessary to use the random effect model for the analysis ([Fig. 4]).
Fig. 4 Forest Plot of number of stents per patient.
Stricture resolution
Four studies were included in analysis of the stricture resolution, with a total of
205 patients. There was no statistically significant difference in the rate of stricture
resolution between the two groups, which showed equivalence for the initial success
of the treatment (RD: 0.01; 95 %CI [−0.08 – 0.10]) ([Fig. 5]).
Fig. 5 Forest Plot of number of strictures resolved.
Stricture recurrence
No statistically significant difference was found between the two groups in evaluation
of the rate of recurrence in 181 patients who had successful initial stricture treatment
(RD: 0.13; 95 %CI [−0.03 to 0.28]). The heterogeneity observed in the analysis exceeded
50 %, and funnel plot was used; no outliers were identified and treatment heterogeneity
was not possible; therefore, we used the random effect model for the analysis ([Fig. 6]).
Fig. 6 Forest Plot of stricture recurrence.
Treatment time
Treatment time (number of days until all stents were removed) was less in the CSEMS
group (MD: −105.07; 95 %CI [−202.38 to −7.76 days]). Heterogeneity observed in the
analysis exceeded 50 %, and funnel plot was used; no outliers were identified and
heterogeneity correction was not possible; therefore, the random effect model was
used for the analysis ( [Fig. 7]).
Fig. 7 Forest Plot of treatment time.
Adverse events
Two studies were used in the analysis of AEs not related to migration, with a total
of 84 patients. No statistically significant difference was observed (RD: −0.10; 95 %CI
[−0.65 to 0.44]). Heterogeneity observed in the analysis was greater than 50 %, and
the random effect model was used for the analysis ( [Fig. 8]).
Fig. 8 Forest Plot of adverse events.
Stent migration was individually analyzed and involved two studies with a total of
84 patients. Again, this analysis did not demonstrate a statistically significant
difference between the materials (RD: −0.05; 95 %CI [−0.18 to 0.08]).
Cost
Assessment of treatment cost included two studies [13]
[24] that reported the average cost of each treatment. Treatment with CSEMS was less
expensive than that with MPS (average of $ 8,288.00 and $ 18,580.00, respectively;
P < 0.001) ([Fig. 9]).
Fig. 9 Graph of the cost means.
Discussion
Treatment of biliary stricture after liver transplantation with CSEMS has aroused
interest in follow-up of transplanted patients because traditional protocols with
plastic stents have an approximate duration of 1 year and several ERCPs are required
to complete treatment. A study with a high level of evidence is still necessary [25], such as this meta-analysis including only RCTs because new randomized trials are
available in the literature, two of which [23]
[24] were published recently (2017). This thereby shows the need for and interest in
determining whether CSEMS are adequate and yield results similar to MPSs. All studies
included in this systematic review and meta-analysis had adequate randomization and
allocation, few losses, and good quality.
As expected, the number of ERCPs was considerably lower in the CSEMS cases because
this treatment involves only two procedures, one to place the stent and another to
remove it. This result was consistent among studies, and this issue is already considered
to be resolved. Therefore, we need to know if fewer ERCPs are associated with lower
AE rates, lower costs, and greater patient acceptance.
Similar to the number of ERCPs, duration of treatment was also shorter in the CSEMS
cases because most MPS protocols last 1 year and CSEMS protocols last a maximum of
6 months.
Stricture resolution was similar among all studies, with no statistical difference,
showing equivalence in initial treatment success between the methods [10]
[13]
[23]
[24]. However, as stricture resolution after the initial treatment was determined by
cholangiography at time of stent removal, it is not likely to be very informative
concerning a relevant clinical endpoint because cholangiography at time of removal
is very likely to have improved enough to warrant stent removal.
Recurrence is one of the most debated issues in this topic. Although it appears that
the question of stricture resolution has been answered, more information is still
needed with regard to recurrence. In this analysis, there was no statistical difference
in rates of recurrence. However, the study by Martins et al.[24] was the only one that showed a significantly higher rate of recurrence in the CSEMS
group, and the authors raised the hypothesis that the shorter dwelling time of the
CSEMS in the bile duct (6 months) was responsible for the rate of recurrence being
higher in the CSEMS group than in the MPS group (dwelling time of 12 months) in their
study. Although this was not the main objective of their study, Martins et al. [24] followed these patients, and six of the eight patients who had recurrence after
CSEMS were successfully treated after a new endoscopic treatment with MPS for 1 year.
Cote et al. [10], Martins et al. [24], and Tal et al. [23] clearly indicated a minimum follow-up period of 1 year, whereas Kaffes et al.[13] did not report a minimum follow-up period. Given the benign characteristics of the
underlying disease and associated life expectancy, even a follow-up of 12 months is
probably not enough to make firm assumptions on long-term efficacy (especially in
the CSEMS group because more is known about the chance of recurrence in the MPS group).
Another major concern is stent migration because a totally covered CSEMS would be
more likely to migrate, causing AEs and affecting the final outcome of treatment.
We did not find statistical significant difference related to migration, but this
analysis was performed taking into consideration the number of patients instead of
the number of stents used. Kaffes et al. [13] used a specific metallic prosthesis with a larger diameter at the ends, which may
have reduced the number of migrations in that study. The different types of stents
used may limit overall generalization. Migration of CSEMS does not seem to cause major
repercussions: Tal et al. [23] reported eight migrations of CSEMS; four did not present further stricture (among
these, one patient had recurrence and was again treated with CSEMS successfully) and
the other four that still had stricture crossed over to the MPS group.
Data related to AEs are scarce, and the studies that mention them [13]
[24] do not clarify whether any patient presented more than one AE. Both procedures are
relatively safe and do not have AE rates that prevent their execution. The rate of
pancreatitis after CSEMS decreases considerably when papillotomy is performed, and
this procedure is also indicated for deployment of CSEMS [24].
The most unexpected result, which is an argument in favor of use of CSEMS, was related
to costs. Treatment with CSEMS was widely believed to be more expensive because of
the costs of the stents [23], but the current study showed, via a meta-analysis of two studies on different continents
(Kaffes et al. [13] and Martins et al. [24]), that use of CSEMS was consistently advantageous, with a cost that was less than
half of that of the MPS treatment (average of $ 8,288.50 versus $ 18,580.00, respectively;
P < 0.001).
One of the limitations was the way each study was analyzed. Martins et al. [24] considered the need for crossover as treatment failure whereas Tal et al. [23] did not consider this as failure and conducted the analysis by intention to treat
at the end. However, Tal et al. [23] had a similar number of crossovers in both groups (four crossovers from CSEMS to
MPS and three from MPS to CSEMS) and all cases resulted in stricture resolution; therefore,
we don’t believe that crossover affected the final analysis.
Diagnosis and treatment success were determined by assessment of cholangiography,
and none of the studies established a relationship between diagnosis and resolution
of biliary stricture and laboratory tests (hepatobiliary enzymes).
The different types of stents used and the 1-year follow-up can reveal unreliable
data once it affects the migration rate and recurrence rate.
Data on AEs are scarce, and because some studies reported only some AEs, these could
not be comprehensively included in the meta-analysis.
This is the first systematic review and meta-analysis to include only randomized studies
on this topic with an adequate number of patients (205). It showed statistical differences
with regard to several outcomes, thereby answering several questions related to this
topic.
Conclusion
Rates of resolution and recurrence of biliary stricture after liver transplantation
were similar, and the number of ERCPs performed, number of stents used, duration of
treatment, and costs were lower in the CSEMS group than in the MPS group. These results
show that use of CSEMS is a valid option for initial treatment of post-OLTX anastomotic
biliary stricture. Whether it should be standard of care should ideally depend on
larger RCTs with adequate follow-up (i. e. 2 years or longer) and one type of stent.
Annex 1
|
MEDLINE e EMBASE
|
(Liver transplantation OR Hepatic Transplantation OR Liver Grafting OR Hepatic Transplantation)
AND (biliary stricture OR biliary stenosis OR biliary stenose OR ercp OR plastic OR
metallic OR stent OR cholangiography OR cholangiopancreatography)
|
|
Scielo/Lilacs e Cochrane
|
Liver transplantation AND biliary
|
Annex 2
GRADE Summary of Findings for evidence quality
|
Outcomes
|
Anticipated absolute effects[1] (95 % CI)
|
Relative effect
(95 % CI)
|
№ of participants
(studies)
|
Certainty of the evidence
(GRADE)
|
|
Risk with MPS
|
Risk with cSEMS
|
|
Number of ERCPs
|
The mean number of ERCPs was 5.75
|
The mean number of ERCPs in the intervention group was 1,86 lower (3,12 lower to 0,6
lower)
|
–
|
205
(4 RCTs)
|
⊗⊗⊗○
MODERATE [2]
|
|
Number of stents
|
The mean number of stents was 12
|
The mean number of stents in the intervention group was 10,63 lower (20,82 lower to
0,44 lower)
|
–
|
112
(2 RCTs)
|
⊗⊗○○
LOW [2]
,
[3]
|
|
Stricture Resolution
|
88 per 100
|
1 per 100
(–7 to 9)
|
RD 0.01
(–0.08 to 0.10)
|
205
(4 RCTs)
|
⊗⊗⊗○
MODERATE [4]
|
|
Stricture Recurrence
|
10 per 100
|
1 per 100
(0 to 3)
|
RD 0.13
(-0.03 to 0.28)
|
181
(4 RCTs)
|
⊗⊗○○
LOW [2]
,
[4]
|
|
Treatment Time
|
The mean treatment Time was 252 days
|
The mean treatment Time in the intervention group was 105,07 days lower (202,38 lower
to 7,76 lower)
|
–
|
205
(4 RCTs)
|
⊗⊗⊗○
MODERATE [2]
|
|
Adverse Events
|
33 per 100
|
–3 per 100
(–22 to 15)
|
RD – 0.10
(–0.65 to 0.44)
|
84
(2 RCTs)
|
⊗⊗○○
LOW [2]
,
[5]
|
CI: Confidence interval; MD: Mean difference; GRADE Working Group grades of evidence: High certainty: We are very confident that the true effect lies close to that of the estimate of
the effect, Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to
be close to the estimate of the effect, but there is a possibility that it is substantially
different, Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially
different from the estimate of the effect, Very low certainty: We have very little confidence in the effect estimate: The true effect is likely
to be substantially different from the estimate of effect
1
The risk in the intervention group (and its 95 % confidence interval) is based on the assumed risk in the comparison
group and the relative effect of the intervention (and its 95 % CI).
2 i2 > 50 %
3 Tal et al. considered the crossovers for the measures, while Martins et al. didn't
4 Absent of endoscopist blindness
5 Small number of patients included
Annex 3
|
|
Kaffes A, 2014
|
|
Cote GA, 2016
|
|
Tal AO, 2017
|
|
Martins FP, 2017
|
|
|
Results
|
|
|
FCSEMS
|
MPS
|
FCSEMS
|
MPS
|
FCSEMS
|
MPS
|
FCSEMS
|
MPS
|
|
Patients
|
10
|
10
|
37
|
36
|
24
|
24
|
32
|
32
|
|
Number of ERCPs
|
2.0 (2.0 – 2.0)
|
4.5 (2.0 – 6.0)
|
2,21 (± 0,48)
|
3,13 (± 0,88)
|
2.0 (2.0 – 12.0)
|
4.0 (3.0 – 12.0)
|
2 (2 – 2)
|
5 (4 – 6)
|
|
Number of stents per patient
|
|
|
|
|
1.0 (1.0 – 24.0)
|
8.0 (2.0 – 32.0)
|
1 (1 – 1)
|
16 (6 – 30)
|
|
Stricture resolution
|
10
|
8
|
34
|
34
|
24
|
23
|
25
|
28
|
|
Stricture recurrence
|
3/10
|
3/8
|
5/33
|
1/30
|
5/24
|
5/23
|
8/25
|
0/28
|
|
Treatment time (median/range)
|
3.8 (2.5 – 5.0) months
|
10.1 (4.0 – 13.0) months
|
158.2 (± 89,7)
|
193.5 (± 88.7)
|
178.5 (65 – 551)
|
229.5 (59 – 490)
|
158,5 (9 – 239)
|
354 (222 – 42)
|
|
Adverse events
|
1 (cholangitis)
|
5 (4 cholangitis, 1 pain)
|
|
|
|
|
14/60
|
9/141
|
|
Migration
|
0
|
1
|
|
|
|
|
3/30
|
4/141
|
|
Cost
|
AUD: 10830,00 (USD: 9674,00)
|
AUD: 23580,00 (USD: 21065,00)
|
|
|
|
|
USD: 6903,00
|
USD: 16095,00
|
