Comparative Efficacy of Various Stents for Palliation in Patients with Malignant Extrahepatic Biliary Obstruction: A Systematic Review and Network Meta-Analysis

Although many studies have investigated the efficacy of stent placement for patients with malignant extrahepatic biliary obstruction, the clinical outcomes and adverse events of biliary stenting have not been comprehensively evaluated. We searched all relevant randomized-controlled trials that evaluated the comparative efficacy of biliary stents, including the plastic stents, uncovered self-expandable metal stents (SEMSs), and covered SEMSs in patients with malignant extrahepatic biliary obstructions. Twenty-one studies with 2326 patients were included. Both uncovered and covered SEMSs had a lower risk of recurrent biliary obstruction (RBO) compared to plastic stents (risk ratio (RR) (95% confidence interval [CI]): uncovered vs. plastic, 0.46 (0.35–0.62); covered vs. plastic, 0.46 (0.34–0.62)). A comparison of the groups using SEMSs revealed that tumor ingrowth was common in the uncovered SEMS group, while stent migration, tumor overgrowth, and occlusion by sludge were common in the covered SEMS group; however, the overall risk of RBO did not differ between these groups (RR (95% CI): uncovered vs. covered: 1.02 (0.80–1.30)). Although the main causes of RBO vary across stents, RBO risk was similar between uncovered and covered SEMS groups. Both SEMSs have superior efficacy in terms of RBO compared to plastic stents.


Introduction
Appropriate biliary drainage is important for relieving symptoms and extending life expectancy in patients with unresectable malignant biliary obstruction (MBO) [1]. Biliary drainage techniques involving endoscopic retrograde cholangiopancreatography (ERCP) or percutaneous transhepatic biliary drainage were established approximately 40 years ago [2,3]. In the 1980s, when endoscopic drainage was first introduced, plastic stents were typically used for the procedure [4]. Biliary drainage using plastic stents was as effective as surgical drainage [2,4]. In the 1990s, self-expandable metal stents (SEMSs) were introduced, and their efficacies were compared with those of plastic stents in many studies [5]. Endoscopic drainage using SEMSs prolongs patient survival, lowers adverse events, including recurrent biliary obstruction (RBO) [6], and requires fewer reinterventions compared with drainage using plastic stents [5,7]. Although SEMSs are more expensive than plastic stents, a recent randomized-controlled trial (RCT) showed that the total costs up to one year after stenting did not significantly differ between plastic stent and SEMS groups because of the longer functional time of SEMSs [8].
Regarding subtypes of SEMSs, several meta-analyses compared the efficacies of uncovered and covered SEMSs during endoscopic drainage for patients with extrahepatic MBO [9,10]. According to these studies, using covered SEMSs entailed a lower risk of

Search Strategy
All relevant studies published between January 1990 and January 2020 that evaluated the efficacy of stents for palliation in patients with extrahepatic MBO were retrieved from the MEDLINE, EMBASE, and Cochrane Library databases. The following search string was used: ((bile duct) or (biliary) or (cholangiocarcinoma) or (cholangiocarcinomas) or (pancreatic) or (pancreas)) and ((obstruction) or (obstructive) or (stricture) or (strictures) or (jaundice) or (cholestasis)) and ((nonresectable) or (unresectable) or (palliation) or (palliative) or (palliating) or (inoperable)) and ((stent) or (stents) or (stenting) or (endoprostheses)) and (random*). Appendix ?? shows the detailed search strategies for each database. Additionally, we examined the references of the screened articles to identify additional relevant studies. Our search was last updated at 15 January 2020.

Inclusion/Exclusion Criteria
The inclusion criteria were as follows: (a) population: patients with unresectable extrahepatic MBO, (b) intervention: ERCP with palliative biliary decompression using stents including a plastic, uncovered, or covered stent, (c) comparator: ERCP with palliative biliary decompression with another type of stent, and (d) outcome: technical and clinical success, RBO, and adverse events. Non-human studies, non-original studies, non-RCTs, abstract-only publications, and studies published in languages other than English were excluded.

Study Selection
In the first step of the study selection, duplicated articles, which were retrieved through multiple search engines, were excluded. Thereafter, we examined the titles and abstracts of the articles to exclude irrelevant studies. The full text of the remaining articles was then assessed for eligibility. Two investigators (C.H.P. and S.W.P.) independently evaluated the studies for eligibility and resolved any disagreements through discussion. If an agreement could not be reached, a third investigator (J.H.J.) determined study eligibility. The Cochrane Risk of Bias assessment tool was used for assessing the risk of bias in the included RCTs.

Data Extraction and Study Endpoint
Using a data extraction form developed in advance, two reviewers (C.H.P. and S.W.P.) independently extracted the following information: first author name, year of publication, study design, country, study period, publication language, type of biliary stent (plastic stent, uncovered SEMS, or covered SEMS), and clinical outcomes, including clinical success, RBO, and adverse events.
The primary endpoint in this meta-analysis was the comparative efficacy of biliary stents in terms of RBO. Secondary endpoints were causes of RBO, including occlusion by sludge, tumor ingrowth and overgrowth, and stent migration, and stent-related inflammation, including cholangitis, cholecystitis, and pancreatitis.

Statistical Analysis
A direct pairwise meta-analysis was conducted to calculate the risk ratios (RRs) of categorical variables, including those of RBO and adverse events using a random-effects model. For technical and clinical successes, we calculated crude proportions rather than RRs because the technical or clinical success rates were 100% in many studies, and those studies could not be included for calculating pooled RRs. Statistical heterogeneity was assessed using two methods: Cochrane's Q test, in which p-values of <0.1 were considered statistically significant for heterogeneity, and I 2 statistics, wherein values of >50% suggested significant heterogeneity [13]. We assessed publication bias quantitatively using the Begg and Mazumdar adjusted-rank correlation test (publication bias was considered present if p was <0.1) [14]. We also assessed bias qualitatively by inspecting the funnel plots of logarithmic RRs versus their standard errors [15]. When the number of included studies for each pairwise comparison was less than 10, the test for publication bias was not conducted [16]. A direct pairwise meta-analysis was performed using the Review Manager statistical software (version 5.3.5; Cochrane Collaboration, Copenhagen, Denmark) and Comprehensive Meta Analysis (version 2.2.064; Biostat Inc., Englewood, NJ, USA) software.
A frequentist network meta-analysis was performed to calculate direct and indirect estimates and combine mixed estimates [17]. Moreover, each stent type was ranked according to P-scores, which were based solely on the point estimates and standard errors of network estimates [18]. The P-score of each eradication regimen can be interpreted as the mean extent of certainty that a certain regimen was better than another [18]. The network meta-analysis was performed using the R statistical software (version 3.6.2; R Foundation for Statistical Computing, Vienna, Austria) with the netmeta package (version 0.9-1; Rücker et al.). The netmeta package is based on the graph theory methodology used to model the relative treatment effects of multiple treatments under a frequentist framework [19].
The baseline patient and lesion characteristics are summarized in Table 1. Studies were published between 1992 and 2020 with an enrollment period that ranged from 1990 to 2018. Six studies evaluated compared the efficacy of plastic stents and that of uncovered SEMSs [20][21][22][23]25,37], while three studies compared the efficacy of plastic stents and that of covered SEMSs [26,30,33]. The other 11 studies compared the efficacy of uncovered SEMSs and that of covered SEMSs [24,[27][28][29]31,32,[34][35][36]38,39]. The remaining study was a three-arm trial, which compared the efficacy of plastic stents, that of uncovered SEMS, and that of covered SEMS [8].

Study Selection and Characteristics
Twenty-one studies including 2326 patients were included in our meta-analysis ( Figure 1) [8,[20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39]. The baseline patient and lesion characteristics are summarized in Table 1. Studies were published between 1992 and 2020 with an enrollment period that ranged from 1990 to 2018. Six studies evaluated compared the efficacy of plastic stents and that of uncovered SEMSs [20][21][22][23]25,37], while three studies compared the efficacy of plastic stents and that of covered SEMSs [26,30,33]. The other 11 studies compared the efficacy of uncovered SEMSs and that of covered SEMSs [24,[27][28][29]31,32,[34][35][36]38,39]. The remaining study was a three-arm trial, which compared the efficacy of plastic stents, that of uncovered SEMS, and that of covered SEMS [8].   The evidence network is shown in Figure S1. Lines represent the comparison between stent types. The thickness of these lines and the numbers shown in large represent the number of studies included in each comparison. The numbers in rectangles indicate the number of patients included in each comparison. Because one study was a three-arm trial, the sum of comparisons and that of patients exceeded the number of included studies and that of included patients, respectively. SEMS, self-expandable metal stent.
The risk of bias assessments for individual studies is shown in Figure S2. Among the 21 included studies, 3 (14.3%) had an unclear risk of bias regarding random sequence generation. An unclear risk of bias in terms of allocation concealment was identified in 8 studies (38.1%). All studies were assessed as having a low risk of performance and detection bias because stent-related outcomes, including RBO, are less likely to be affected by the blinding of participants and investigators. Attrition bias was not identified. One study (4.8%) was assessed as having a high risk of reporting bias because it did not report the proportion of RBO, which was a primary endpoint of our meta-analysis.

Technical and Clinical Success According to the Stent Type
Technical and clinical successes were excellent regardless of stent type. The proportion of technical success was 99.4% (95% CI, 98.6-99.7%) in the plastic stent group, 100.0% (95% CI, 98.5-100.0%) in the uncovered SEMS group, and 99.2% (95% CI, 98.6-99.7%) in the covered SEMS group. Additionally, the proportion of clinical success was 90.1% (95% CI, 84.9-93.6%) in the plastic stent group, 94.6% (95% CI, 91.2-96.7%) in the uncovered SEMS group, and 92.3% (95% CI, 88.4-94.9%) in the covered SEMS group. There was no statistical difference in technical and clinical successes among stent types. Between the uncovered and covered SEMS groups, there was no significant difference in terms of RBO, but there was statistical heterogeneity (covered SEMS vs. uncovered SEMS: RR (95% CI) = 0.93 (0.68-1.26), df = 10, p = 0.009, I 2 = 57%). Publication bias was assessed for RBO between the uncovered and covered SEMS groups. We did not detect any corresponding publication bias using the Begg and Mazumdar-adjusted rank correlation tests (p = 0.938). Asymmetry was not observed upon the visual inspection of the funnel plot ( Figure S3).

Direct Meta-Analysis for Recurrent Biliary Obstruction and Stent-Related Inflammation
The white diamond represents the pooled logarithmic RR, with a 95% confidence interval, among observed studies.
RBO, recurrent biliary obstruction; SEMS, self-expandable metal stent; M-H, Mantel-Haenszel; RR, risk ratio Figure S4 shows forest plots for causes of RBO, including stent occlusion by sludge, tumor ingrowth, tumor overgrowth, and stent migration. Stent occlusion by sludge was common in the plastic stent group, tumor ingrowth was common in the uncovered SEMS group, and tumor overgrowth and stent migration tended to be common in the covered SEMS group. M-H, Mantel-Haenszel; CI, confidence interval.
In terms of stent-related inflammation, there was no statistical difference in most comparisons ( Figure S5). Although cholangitis was less common in the covered SEMS group than in the plastic stent group, only one study was included in that comparison. M-H, Mantel-Haenszel; CI, confidence interval. Between the uncovered and covered SEMS groups, there was no significant differ ence in terms of RBO, but there was statistical heterogeneity (covered SEMS vs. uncovered SEMS: RR (95% CI) = 0.93 (0.68-1.26), df = 10, p = 0.009, I 2 = 57%). Publication bias wa assessed for RBO between the uncovered and covered SEMS groups. We did not detec any corresponding publication bias using the Begg and Mazumdar-adjusted rank corre lation tests (p = 0.938). Asymmetry was not observed upon the visual inspection of the funnel plot ( Figure S3).
The white diamond represents the pooled logarithmic RR, with a 95% confidence interval, among observed studies.
RBO, recurrent biliary obstruction; SEMS, self-expandable metal stent; M-H, Mantel-Haenszel; RR, risk ratio Figure S4 shows forest plots for causes of RBO, including stent occlusion by sludge tumor ingrowth, tumor overgrowth, and stent migration. Stent occlusion by sludge wa common in the plastic stent group, tumor ingrowth was common in the uncovered SEMS group, and tumor overgrowth and stent migration tended to be common in the covered SEMS group. M-H, Mantel-Haenszel; CI, confidence interval.
In terms of stent-related inflammation, there was no statistical difference in mos comparisons ( Figure S5). Although cholangitis was less common in the covered SEMS group than in the plastic stent group, only one study was included in that comparison. M

Network Meta-Analysis for Recurrent Biliary Obstruction and Stent-Related Inflammation
In Figure 3, forest plots show the network estimates of RBO and its causes in the uncovered and covered SEMS groups compared with the plastic stent group. Both uncovered and covered SEMS groups showed a lower risk of RBO compared to the plastic stent group (RR (95% CI): uncovered SEMS vs. plastic stent, 0.46 (0.35-0.62); and covered SEMS vs. plastic stent, 0.46 (0.34-0.62)). There was no significant difference between the uncovered and covered SEMS groups (uncovered vs. covered: RR (95% CI) = 1.02 (0.80-1.30)). Plastic stents had the highest P-score (>99%), followed by uncovered SEMS (28%) and covered SEMS (22%).
The detailed efficacy profiles of stents are summarized in Table S1. There was no network inconsistency in the comparison.
Both uncovered and covered SEMS groups also had a lower risk of stent occlusion by sludge compared to the plastic stent group (RR (95% CI): uncovered SEMS vs. plastic stent, 0.09 (0.04-0.18); and covered SEMS vs. plastic stent, 0.17 (0.08-0.37)). Additionally, the risk of stent occlusion by sludge in the uncovered SEMS group was lower than that in the covered SEMS group (uncovered vs. covered: RR (95% CI) = 0.51 (0.31-0.83)). However, the uncovered SEMS group had a higher risk of tumor ingrowth compared to the covered SEMS and plastic stent groups (RR (95% CI): uncovered SEMS vs. covered SEMS, 4.49 (2.21-9.09); and uncovered SEMS vs. plastic stent, 23.79 (2.54-222.81)). In terms of tumor overgrowth, both uncovered and covered SEMS groups tended to be higher risk compared to the plastic stent group. However, the covered SEMS group had a significantly higher risk of tumor overgrowth compared to the uncovered SEMS group (covered vs. uncovered: RR (95% CI) = 1. 98 (1.18-3.34)). Stent migration was common in the covered SEMS group than in the uncovered SEMS group (covered vs. uncovered: RR (95% CI) = 8.38 (2.83-24.84)). There was no network inconsistency in any comparison.

Network Meta-Analysis for Recurrent Biliary Obstruction and Stent-Related Inflammation
In Figure 3, forest plots show the network estimates of RBO and its causes in the uncovered and covered SEMS groups compared with the plastic stent group. Both uncovered and covered SEMS groups showed a lower risk of RBO compared to the plastic stent group (RR (95% CI): uncovered SEMS vs. plastic stent, 0.46 (0.35-0.62); and covered SEMS vs. plastic stent, 0.46 (0.34-0.62)). There was no significant difference between the uncovered and covered SEMS groups (uncovered vs. covered: RR (95% CI) = 1.02 (0.80-1.30)). Plastic stents had the highest P-score (>99%), followed by uncovered SEMS (28%) and covered SEMS (22%). Comparative efficacy for recurrent biliary obstruction and its causes according to the type of stents from the network meta-analysis. The P-score indicates the mean extent of certainty that one type of stent is better than another. RBO, recurrent biliary obstruction; SEMS, self-expandable metal stent; RR, risk ratio; CI, confidence interval. Comparative efficacy for recurrent biliary obstruction and its causes according to the type of stents from the network meta-analysis. The P-score indicates the mean extent of certainty that one type of stent is better than another. RBO, recurrent biliary obstruction; SEMS, self-expandable metal stent; RR, risk ratio; CI, confidence interval. Figure 4 shows the forest plots for stent-related inflammation. Although cholangitis was marginally more common in the plastic stent group than in the covered SEMS group (RR (95% CI) = 2.37 (1.003-5.60)), there was no significant difference or network inconsistency in any other comparison.
tumor overgrowth, both uncovered and covered SEMS groups tended to be higher risk compared to the plastic stent group. However, the covered SEMS group had a significantly higher risk of tumor overgrowth compared to the uncovered SEMS group (covered vs. uncovered: RR (95% CI) = 1.98 (1.18-3.34)). Stent migration was common in the covered SEMS group than in the uncovered SEMS group (covered vs. uncovered: RR (95% CI) = 8.38 (2.83-24.84)). There was no network inconsistency in any comparison. Figure 4 shows the forest plots for stent-related inflammation. Although cholangitis was marginally more common in the plastic stent group than in the covered SEMS group (RR (95% CI) = 2.37 (1.003-5.60)), there was no significant difference or network inconsistency in any other comparison. Comparative efficacy for stent-related inflammation according to the type of stents from the network metaanalysis. The P-score indicates the mean extent of certainty that one type of stent is better than another. SEMS, self-expandable metal stent; RR, risk ratio; CI, confidence interval.

Absolute Risk of Recurrent Biliary Obstruction and Adverse Events
To understand the absolute risk of RBO and adverse events, the incidence of each outcome was summarized in Table S2. The proportion of RBO was 47.4% (95% CI, 42.8- Figure 4. Comparative efficacy for stent-related inflammation according to the type of stents from the network meta-analysis. The P-score indicates the mean extent of certainty that one type of stent is better than another. SEMS, self-expandable metal stent; RR, risk ratio; CI, confidence interval.

Discussion
This network meta-analysis study is the first to comprehensively compare three types of stents for the endoscopic drainage of extrahepatic MBO. Compared to the traditional pairwise meta-analytic approach, our study has the innovative advantage of filling the literature gap between RCTs, offering the application of appropriate stents to the clinical field through novel comparisons. We showed that all three types of stents have extremely high technical success rates of more than 99%. Additionally, all stent types showed high clinical success rates of more than 90%. However, the clinical success rate of the plastic stent tended to be 4.5% lower than that of the uncovered SEMS (90.1% vs. 94.6%). This might be derived from the vulnerability of the plastic stent and sludge occlusion, followed by biliary obstruction, subsequent cholangitis, and the need for stent revision [40,41].
Regarding primary outcomes related to RBO, both uncovered and covered SEMSs showed a lower risk compared to plastic stents. The diameter of the stent may play an essential role in maintaining stent patency; thus, a 10-Fr plastic stent would be better than 7-or 8-Fr stents [42]. However, duodenoscopes have a limitation of delivery for plastic stents greater than about 11.5-Fr stents through the working channel. In contrast, biliary SEMSs with larger internal diameters (up to 10 mm) can be placed with a 7-or 8-Fr delivery system, which is enough to pass through the working channel [43].
RBO did not differ between uncovered and covered SEMSs. However, it was affected by different mechanisms. Covered SEMSs were more likely to result in RBO from sludge occlusion, tumor overgrowth, and stent migration, whereas uncovered SEMSs were mainly influenced by tumor ingrowth. The coating membrane of the covered SEMS was invented to prevent tumor ingrowth; however, it offers a medium for the attachment and colonization of a bacterial biofilm [9], which lead to stent sludge occlusion. Simultaneously, it can disturb the embedding and anchoring of the stent in situ [44]. This can consequently lead to an 8.4-fold higher chance of stent migration compared with uncovered SEMSs. However, the absolute risk of stent migration was relatively low even in patients who underwent covered SEMS placement (5.0%) compared with that of tumor ingrowth caused by uncovered SEMSs (17.3%).
Although SEMSs have a relatively good stent patency, their revision for RBO is challenging compared to plastic stents because of difficulties in stent removal or replacement, especially when an uncovered SEMS was placed. Thus, plastic stents might be appropriate for patients with a life expectancy of less than six months, whereas SEMS can be more favorable for patients with a life expectancy of six months or more [45].
Among stent-related inflammations, including cholangitis, cholecystitis, and pancreatitis, cholangitis may be the most concerning because it is the most common. The risk of cholangitis in the plastic stent group was higher than that in the covered SEMS group. Because biliary drainage through plastic stents may require more re-interventions than drainage through SEMSs, the risk of cholangitis can be increased when plastic stents are placed. Regarding cholecystitis, several studies reported that covered SEMSs might have a higher cholecystitis risk after biliary stent placement than after uncovered SEMS placement [46,47]. Additionally, our meta-analysis showed a tendency of increased cholecystitis risk in the covered SEMS group compared to the plastic stent group; however, there was no significant difference between groups. It was difficult to identify the small differences regarding cholecystitis risk among stents because incidences of stent-related cholecystitis were very low (0.5% for plastic stents, 1.6% for uncovered SEMSs, and 3.1% for covered SEMSs). Additionally, it has been suggested that tumor invasions to the feeding artery of the gallbladder or the orifice of the cystic duct, rather than the stent itself, might contribute to the development of cholecystitis [48]. Meanwhile, there is also a concern regarding acute pancreatitis surrounding SEMSs because the opening of the pancreatic duct could, theoretically, be occluded by the axial force of the stent [49,50]. However, our analysis showed no statistical difference in the risk of pancreatitis among stents.
Although this study was the first network meta-analysis to compare the efficacies of various stents in the palliative drainage for extrahepatic MBO, it has several limitations. First, significant heterogeneity was identified in the comparisons of uncovered and covered SEMSs regarding RBO and tumor ingrowth; however, there was no significant heterogeneity in any other comparisons. Second, the specific models of the compared stents varied across the included studies. Although the inclusion of various stent models may help generalize our study findings, it may be a potential confounding variable. A multicenter-cohort study reported that covered WallFlex stents were superior to covered Wallstents in terms of the cumulative incidence of RBO [51]. A Korean multicenter study showed different migration risk between covered SEMS with flaps and conventional covered SEMS in patients with benign biliary strictures [52]. Third, biliary obstructions in the included patients were caused by various underlying malignant diseases, including pancreatic cancer, bile duct cancer, gallbladder cancer, and ampullary cancer. Because clinical outcomes and adverse events in each included study were reported without the consideration of underlying diseases, we could not perform a subgroup analysis according to the types of malignant diseases.
Despite these limitations, the present network meta-analysis provides a better understanding of clinical outcomes and adverse events according to the type of stents in patients with extrahepatic MBO. In conclusion, we recommend that the use of an SEM should be considered as the first option to treat extrahepatic MBOs because SEMs have longer stent patency, lower sludge occlusion rate, and lower risk of stent-related cholangitis, compared with plastic stents. Uncovered SEMSs had a high risk of tumor ingrowth, whereas covered SEMSs had a high risk of stent migration, tumor overgrowth, and sludge occlusion. However, the overall risk of RBO was similar between the uncovered and covered SEMSs.
Supplementary Materials: The following are available online at https://www.mdpi.com/2075-4 426/11/2/86/s1, Figure S1: Evidence network of different stents. Lines represent the comparison between stent types. The thickness of these lines and the numbers shown in large represent the number of studies included in each comparison. The numbers in rectangles indicate the number of patients included in each comparison. Because one study was a three-arm trial, the sum of comparisons and that of patients exceeded the number of included studies and that of included patients, respectively. SEMS, self-expandable metal stent. Figure S2: Risk of bias summary: the review authors' judgments regarding the risk of bias items for each included study. Figure S3: Funnel plot for the analysis of recurrent biliary obstruction between the uncovered and covered SEMS groups. Figure S4 Table S1: Network estimates of the palliative bile duct stent in terms of recurrent biliary obstruction and adverse events. Table S2: Incidence of recurrent biliary obstruction and adverse events according to the stent type. Institutional Review Board Statement: Ethical review and approval were waived for this study because this is a meta-analysis based on the previously published studies.

Informed Consent Statement: Not applicable.
Data Availability Statement: All relevant data are included in the study and supplementary information. 1: (('bile duct' or biliary or cholangiocarcinoma or cholangiocarcinomas or pancreatic or pancreas) and (obstruction or obstructive or stricture or strictures or jaundice or cholestasis) and (nonresectable or unresectable or palliation or palliative or palliating or inoperable) and (stent or stents or stenting or endoprostheses) and random*).ab, ti.