- Email: [email protected]
Balloon Dilation With Adequate Duration Is Safer Than Sphincterotomy for Extracting Bile Duct Stones: A Systematic Review and Meta-analyses
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2012;10:1101–1109
SYSTEMATIC REVIEWS AND META-ANALYSES Fasiha Kanwal, Section Editor
Balloon Dilation With Adequate Duration Is Safer Than Sphincterotomy for Extracting Bile Duct Stones: A Systematic Review and Meta-analyses WEI–CHIH LIAO,*,‡ YU–KANG TU,‡,§ MING–SHIANG WU,* HSIU–PO WANG,* JAW–TOWN LIN,*,储 JOSEPH W. LEUNG,¶ and KUO–LIONG CHIEN*,‡ *Department of Internal Medicine, National Taiwan University Hospital, National Taiwan University College of Medicine, Taipei, Taiwan; ‡Graduate Institute of Epidemiology and Preventive Medicine, National Taiwan University, Taipei, Taiwan; §Division of Biostatistics, Leeds Institute of Genetics, Health & Therapeutics and Leeds Dental Institute, University of Leeds, Leeds, United Kingdom; 储Department of Internal Medicine, E-DA Hospital and I-Shou University, Kaohsiung County, Taiwan; and ¶Davis Medical Center, University of California and Section of Gastroenterology, Sacramento VA Medical Center, Sacramento, California
BACKGROUND & AIMS: Endoscopic sphincterotomy (EST) is the standard treatment for choledocholithiasis. Endoscopic papillary balloon dilation (EPBD) has a lower risk for bleeding than EST, but EPBD is reserved for patients with bleeding diathesis because some studies reported that it increases the risk for pancreatitis. A short dilation time (ⱕ1 minute) is therefore recommended to reduce pancreatitis. However, there is evidence for an inverse relationship between EPBD duration and pancreatitis, prompting reevaluation of the optimal duration and relative safety of EPBD vs EST. METHODS: We systematically reviewed randomized controlled trials to compare long EPBD (⬎1 minute), short EPBD (ⱕ1 minute), and EST regarding pancreatitis and overall complications. In addition to pairwise meta-analyses, Bayesian network meta-analysis was undertaken to compare the 3 procedures together. Relation between duration and outcome was also analyzed by meta-regression. RESULTS: Compared with EST, short EPBD had a higher risk for pancreatitis (odds ratio [OR] by traditional analysis, 3.87; 95% confidence interval, 1.08 – 13.84 and OR by network meta-analysis, 4.14; 95% credible interval, 1.58 –12.56), but long EPBD did not pose a higher risk than EST (1.14, 0.56 –2.35 and 1.07, 0.38 –2.76). Long EPBD had a lower overall rate of complications than EST (0.61, 0.36 –1.04 and 0.54, 0.20 –1.36). In network meta-analysis, probabilities of being the safest treatment for long EPBD/short EPBD/EST regarding pancreatitis and overall complications were 43.9%/0.2%/55.9% and 90.3%/1.3%/8.4%, respectively. CONCLUSIONS: Duration of EPBD is inversely associated with pancreatitis risk. Currently recommended <1-minute dilation actually increases pancreatitis. EPBD with adequate duration may be preferred over EST because of comparable pancreatitis but lower overall complication rates. Keywords: Gallbladder; Mixed Treatment Comparison; Sphincteroplasty; Risk Factor.
Watch this article’s video abstract and others at http:// tiny.cc/bz9jv. Scan the quick response (QR) code to the left with your mobile device to watch this article’s video abstract and others. Don’t have a QR code reader? Get one at mobiletag.com/en/download.php.
E
ndoscopic retrograde cholangiopancreatography (ERCP) and endoscopic sphincterotomy (EST) are the standard treatment for choledocholithiasis,1–3 which cuts biliary sphincter to open the bile duct orifice for stone extraction.4,5 An alternative to EST is endoscopic papillary balloon dilation (EPBD), which uses a balloon of 6 –10 mm in diameter to dilate biliary sphincter.5–10 Two meta-analyses of randomized controlled trials (RCTs) consistently found that EST and EPBD had comparable overall success rates of stone removal (94.3% vs 96.5%) and similar rates of cholangitis, perforation, mortality, and overall complications (10.5% vs 10.3%) for treating choledocholithiasis.1,11 Those meta-analyses, however, also showed that EPBD was less likely to cause bleeding (0% vs 2%) but more likely to cause pancreatitis (7.4% vs 4.3%) than EST.1,11,12 Therefore, EPBD is reserved for patients with bleeding diathesis in current consensus,1,11,12 and a short dilation duration (ⱕ1 minute) is recommended to reduce pancreatitis.1,2,12 However, the risk of pancreatitis after EPBD in previous trials varied greatly between 0% and 15.4%.1,11,13 Emerging evidence suggests that this disparity may be explained by differences in dilation duration of EPBD in those trials,12,13 and this raises concerns that dilation duration might be a source of confounding14 and prompts reevaluation of the current recommendation with consideration of EPBD duration.12,13 A recent RCT found that the risk of pancreatitis after EPBD with 5-minute duration was comparable to EST and was lower than EPBD with 1-minute duration.13 Contrary to current recommendation, these findings suggest that EPBD with ⱕ1-minute dilation actually increases the risk of pancreatitis, whereas EPBD with adequate duration may have a lower overall complication rate than EST because it reduces bleeding without causing more pancreatitis. Therefore, we reevaluated the relative safety of EPBD over EST and the optimal duration of EPBD by conducting a network meta-analysis to compare shortduration EPBD (ⱕ1 minute), long-duration EPBD (⬎1 minute), and EST, focusing on pancreatitis and overall complications. Abbreviations used in this paper: CI, confidence interval; CrI, credible interval; EPBD, endoscopic papillary balloon dilation; ERCP, endoscopic retrograde cholangiopancreatography; EST, endoscopic sphincterotomy; OR, odds ratio; RCT, randomized controlled trial. © 2012 by the AGA Institute 1542-3565/$36.00 http://dx.doi.org/10.1016/j.cgh.2012.05.017
1102
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
EST was more difficult and required special technique. Finally, 12 RCTs were eligible for analysis.
Data Extraction and Quality Assessment Two investigators (W.-C.L., H.-P.W.) independently reviewed full manuscripts of eligible trials and extracted information into an electronic database including patient characteristics, methods of EPBD and EST, and complications. Dilation duration was treated as a continuous variable and dichotomized into short (ⱕ1 minute) or long (⬎1 minute). For studies with repeated dilations, total duration was used for analysis. Studies were assessed independently by the same reviewers with Cochrane risk of bias tool (Supplementary Figure 1).19 Disagreement was resolved by joint review of the manuscript and consensus.
Data Synthesis and Analysis
Figure 1. Flow chart of literature search and selection.
Methods Data Sources and Searches This systematic review follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline.15 The MEDLINE was searched for RCTs of EPBD for bile duct stone until end of August 2011 without language or date restriction by using the following keywords: endoscopic papillary balloon dilation/dilatation, endoscopic sphincteroplasty, endoscopic balloon sphincter dilation/dilatation, and endoscopic balloon dilation/dilatation. Cochrane Collaboration Central Register of Controlled Clinical Trials, Cochrane Systematic Reviews, http://ClinicalTrials.gov, and bibliographies of included trials and related meta-analyses were manually searched for additional references.
Study Selection We identified 141 studies for review of title and abstract to find RCTs comparing EPBD vs EST or long EPBD vs short EPBD (Figure 1). Studies on endoscopic papillary large balloon dilation were excluded, because it combines EST and dilation with a large balloon for large/difficult stones and has increased risk of complications.16,17 After initial screening, full text of 48 potentially eligible articles was retrieved for detailed assessment. Studies were excluded if duration for EPBD was not clearly reported, definition of post-ERCP complications was different from the consensus guideline,18 or subjects had surgically altered anatomy (eg, Billroth-II anastomosis) in which
The outcomes analyzed were risks of post-ERCP pancreatitis and overall complications. A three-stage strategy was used for analysis. First, pairwise comparisons of long EPBD vs EST and short EPBD vs EST were performed with traditional meta-analysis by using Stata 10.0 (StataCorp, College Station, TX). Pooled odds ratios (ORs) and their 95% confidence intervals (CIs) were estimated with fixed-effect model if no significant heterogeneity existed and with randomeffects model if significant heterogeneity existed. Heterogeneity was evaluated by I2 and Cochran Q.20 Relations between dilation duration and outcomes were analyzed by metaregression with natural logarithm of OR as the dependent variable.21 Potential small study bias was evaluated by funnel plots and by tests of Egger et al.22 Because traditional meta-analysis excludes studies with no event in both treatment arms and can introduce bias when using continuity corrections for studies with no event in one arm,23 we also used Bayesian methods for pairwise meta-analyses. Bayesian methods handle sparse data and zero events without continuity corrections; thus, they include all evidence and provide validation for results of the first stage.24 Finally, a Bayesian network metaanalysis, also known as mixed treatment comparison, was undertaken to compare long EPBD, short EPBD, and EST together. Bayesian network meta-analysis preserves randomization in the trials and combines evidence from direct and indirect comparisons to provide more precise estimates.25–27 Bayesian meta-analyses were performed by using WinBUGS 1.4.3 with Markov Chain Monte Carlo methods (MRC Biostatistics Unit, Cambridge, UK). We used noninformative uniform and normal prior distributions25,27 and 3 different sets of starting values with 50,000 burn-ins and an additional 50,000 iterations for each chain, yielding 150,000 iterations to obtain the posterior distributions of model parameters. Convergence of iterations was evaluated by using Gelman–Rubin–Brooke statistic.28
Results All included RCTs have been published in full manuscript form. Outcomes and detailed characteristics are summarized in Table 1 and Supplementary Table 1. Eleven studies compared EPBD vs EST without significant small study bias (Supplementary Figure 2); four compared short EPBD (total n ⫽ 459) vs EST (total n ⫽ 464), and seven compared long
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
1103
Table 1. Outcomes of RCTs Included in the Analysis Complications, n (%) First author, year Short EPBD (ⱕ1 min) vs EST Bergman,6 1997 Vlavianous,3 2003 Fujita,7 2003 DiSario,2 2004 Long EPBD (⬎1 min) vs EST Minami,5 1995 Ochi,45 1999 Arnold,46 2001 Yasuda,40 2001 Natsui,47 2002 Lin,9 2004 Tanaka,48 2004 Short EPBD vs long EPBD Liao,13 2010
Treatment
EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST
1 min 30 s 15 s 1 min
3 min 3 mina 2 minb 2 minb 2 min 5 min 2 min
EPBD: 1 min EPBD: 5 min
Participant, n
Pancreatitis
All
101 101 103 99 138 144 117 120
7 (6.9) 7 (6.9) 5 (4.9) 1 (1.0) 15 (10.9) 4 (2.8) 18 (15.4) 1 (0.8)
17 (16.8) 24 (23.8) 7 (6.8) 3 (3.0) 20 (14.5) 17 (11.8) 21 (17.9) 4 (3.3)
20 20 55 55 30 30 35 35 70 70 51 53 16 16
2 (10.0) 2 (10.0) 0 (0.0) 2 (3.6) 6 (20.0) 3 (10.0) 2 (5.7) 2 (5.7) 4 (5.7) 3 (4.3) 0 (0.0) 0 (0.0) 3 (18.8) 3 (18.8)
2 (10.0) 2 (10.0) 1 (1.8) 3 (5.5) 9 (30.0) 5 (16.7) 2 (5.7) 3 (8.6) 7 (10.0) 8 (11.4) 1 (2.0) 14 (26.4) 3 (18.8) 4 (25.0)
86 84
13 (15.1) 4 (4.8)
16 (18.6) 4 (4.8)
aThree bTwo
repeated dilations, each for 1 minute. repeated dilations, each for 1 minute.
EPBD (total n ⫽ 277) vs EST (total n ⫽ 279). Only one study compared long EPBD vs short EPBD.
Traditional Meta-analyses: Long Endoscopic Papillary Balloon Dilation vs Endoscopic Sphincterotomy and Short Endoscopic Papillary Balloon Dilation vs Endoscopic Sphincterotomy Pancreatitis. One study9 was excluded because no pancreatitis was reported in either treatment group. Dilation duration and estimated ORs of included trials are summarized in Figure 2. There was significant heterogeneity across studies comparing short EPBD vs EST (P ⫽ .034, I2 ⫽ 65.4%), but not in those comparing long EPBD vs EST (P ⫽ .822, I2 ⫽ 0%). Compared with EST, short EPBD had a significantly higher pancreatitis risk (pooled OR, 3.87; 95% CI, 1.08 –13.84; randomeffects model), but long EPBD did not pose a higher risk (pooled OR, 1.14; 95% CI, 0.56 –2.35; fixed-effect model) (Figure 2). Overall complications. There was significant heterogeneity across studies comparing short EPBD vs EST (P ⫽ .006, I2 ⫽ 76.2%), but not in those comparing long EPBD vs EST (P ⫽ .123, I2 ⫽ 40.3%). Compared with EST, short EPBD had a trend toward higher overall complication rate (pooled OR, 1.71; 95% CI, 0.67– 4.35; random-effects model). By contrast, long EPBD seemed to have a lower overall complication rate (pooled OR, 0.61; 95% CI, 0.36 –1.04; fixed-effect model) (Figure 2).
Meta-regression: Relation Between Dilation Duration and Outcomes For pancreatitis, one study9 was excluded because no pancreatitis occurred. The dilation durations varied between 0.25 and 3 minutes and explained 46.8% of the between-study variance of OR. Variation in OR from heterogeneity across the trials decreased from 34.8% to 7.4% after differences in dilation duration were considered. The regression coefficient of dilation duration was ⫺0.69 (95% CI, ⫺1.47 to 0.09; P ⫽ .076), meaning that every 1-minute increase in dilation duration up to 3 minutes was associated with a 49.8% (95% CI, ⫺9.4% to 77.0%) reduction in OR (Figure 3). For overall complications, dilation duration explained 33.6% of the between-study variance of OR with a regression coefficient of ⫺0.60 (95% CI, ⫺1.15 to ⫺0.05; P ⫽ .037), meaning that every 1-minute increase in dilation duration up to 5 minutes was associated with a 45.1% (95% CI, 4.9%– 68.3%) reduction in OR (Figure 3). The use of single or repeated dilations was not associated with ORs of either outcome.
Bayesian Pairwise Meta-analyses All eligible studies were included in Bayesian metaanalyses, and the results were similar to those of traditional meta-analyses (Table 2, Figure 4). Compared with EST, short EPBD had a higher risk of pancreatitis, whereas long EPBD had a comparable pancreatitis risk.
1104
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Figure 2. Forest plots for ORs of pancreatitis (A) or overall complications (B) stratified according to dilation duration.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
1105
Figure 3. Associations between dilation duration and natural logarithm of OR of pancreatitis (A) or overall complications (B). Circles are sized according to the precision of each OR (inverse of within-study variance).
Network Meta-analyses With Bayesian Mixed Treatment Comparison For pancreatitis, short EPBD had a higher risk than both EST (OR, 4.14; 95% credible interval [CrI], 1.58 –12.56) and long EPBD (OR, 3.85; 95% CrI, 1.22–15.46), whereas long EPBD did not pose a higher risk than EST (OR, 1.07; 95% CrI, 0.38 –2.76) (Table 2, Figure 4). The OR of short vs long
EPBD estimated indirectly from short EPBD vs EST and long EPBD vs EST (3.86) was consistent with estimate from direct comparison (3.85). The probabilities of being ranked as the safest treatment for long EPBD, short EPBD, and EST were 43.9%, 0.2%, and 55.9%, respectively (Supplementary Figure 3). For overall complications, short EPBD had a trend toward higher risk than EST (OR, 1.90; 95% CrI, 0.67–5.70), whereas
1106
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Table 2. Comparisons of Short EPBD, Long EPBD, and EST: Pooled ORs for Pancreatitis and Overall Complications Pairwise meta-analysis
Pancreatitis Short EPBD vs EST Long EPBD vs EST Short vs long EPBD Overall complications Short EPBD vs EST Long EPBD vs EST Short vs long EPBD aOnly
Network meta-analysis
Conventional OR (95% CI)
Bayesian OR (95% CrI)
OR (95% CrI)
3.87 (1.08–13.84) 1.14 (0.56–2.35) 3.56 (1.11–11.42)a
4.46 (1.08–22.49) 1.10 (0.36–2.85) 3.84 (0.25–60.89)a
4.14 (1.58–12.56) 1.07 (0.38–2.76) 3.85 (1.22–15.46)
1.71 (0.67–4.35) 0.61 (0.36–1.04) 4.57 (1.46–14.33)a
1.77 (0.50–7.05) 0.57 (0.18–1.62) 4.99 (0.33–78.18)a
1.90 (0.67–5.70) 0.54 (0.20–1.36) 3.56 (1.01–13.85)
1 trial directly compared short vs long EPBD.
long EPBD had a trend toward lower risk than EST (OR, 0.54; 95% CrI, 0.20 –1.36) (Table 2, Figure 4). Short EPBD had a higher risk than long EPBD (OR, 3.56; 95% CrI, 1.01–13.85). The OR of short vs long EPBD estimated from indirect comparison (3.54) was consistent with estimate from direct comparison (3.56). The probabilities of being ranked as the safest treatment for long EPBD, short EPBD, and EST were 90.3%, 1.7%, and 8.0%, respectively (Supplementary Figure 3).
Discussion Although EPBD has a lower risk of post-ERCP bleeding1,11 and long-term complications than EST11,29 and is easier to perform in altered/difficult anatomy,30,31 EPBD is reserved for patients with bleeding diathesis by current consensus because some studies reported a higher risk of pancreatitis,2,7 and short dilation (ⱕ1 minute) is recommended for presumably lower pancreatitis risk.1,2,12 However, our meta-analyses indicate that short EPBD duration increases rather than decreases the risk of post-ERCP pancreatitis, and EPBD with adequate dura-
tion has a similar pancreatitis risk and a lower overall complication rate compared with EST for choledocholithiasis. This meta-analysis contradicts the common belief that pancreatitis results from direct pancreatic duct compression during balloon dilation, and thus dilation duration should be short.2,12 Recent evidence suggests that EPBD with ⱕ1-minute dilatation carries a higher risk of inadequate sphincter loosening, which increases the risks of pancreatitis and failed stone extraction.13,32 The sphincter of Oddi surrounds ampulla and wraps around the distal end of the bile and pancreatic duct.33 Whereas the sphincter is ablated by EST, it is only loosened by EPBD,13,32 and hemorrhage/inflammation and edema develop at ampulla within 2 hours after EPBD.32 An inadequately loosened sphincter from short-duration EPBD may limit volume expansion of its encircled contents as in a compartment syndrome; thus it worsens compression of pancreatic duct from post-EPBD edema and increases pancreatitis risk.13,33 The compartment theory is supported by our meta-analysis and also explains why pancreatitis is reduced when greater sphincter loosening is
Figure 4. Pooled ORs by traditional meta-analysis, Bayesian meta-analysis, and Bayesian network meta-analysis.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
achieved by a larger balloon,16,34,35 or when the sphincter had been ablated by prior EST.31 Two previous meta-analyses had compared EPBD with EST for treating choledocholithiasis and formed the basis of current consensus.1,11,12 Because the significance of dilation duration had not been recognized until recently, those meta-analyses pooled all studies regardless of dilation duration and concluded that EPBD had a higher pancreatitis risk than EST. Although a nonsignificant Cochran Q seemed to justify pooling of all trials in the above meta-analysis,11 Cochran Q is known to have a poor power in detecting true heterogeneity, especially when the number of studies is small as in the current case.20,36 In this study, we consistently noted that significant heterogeneity caused by differences in dilation duration did exist and must be considered to avoid misleading conclusions. Because short EPBD has a higher pancreatitis risk than EST but long EPBD has a similar risk, pooling short and long EPBD would suggest that EPBD has a higher pancreatitis risk than EST as in previous meta-analyses.1,11 Similarly for the overall complications, pooling higher-risk short EPBD with lower-risk long EPBD would suggest that EPBD and EST have comparable overall complication rates as in previous meta-analyses.1,11 The lower overall complication rate of long EPBD for choledocholithiasis may be attributed to avoidance of bleeding from sphincter cutting in EST1,11,12 without increasing the risk of pancreatitis. Because an estimated 150,000 ESTs are performed annually in the United States with a 2% risk of bleeding,1,37 using EPBD rather than EST as the first-line treatment might prevent 3000 cases of bleeding annually and the associated morbidity/mortality and costs, and this benefit may become greater with the increasing use of antithrombotic agents. EST is a high-risk procedure in patients taking antithrombotic medications, and guidelines recommend discontinuation of antithrombotic agents for 5–10 days before EST.38,39 This causes delayed treatment and prolonged admission, and the interruption of antithrombotic therapy can cause devastating thromboembolic events.38 By contrast, EPBD can be safely performed with continued antithrombotic therapy,39 which simplifies patient management and avoids both the risk of bleeding and thromboembolism. Besides a lower short-term complication rate, emerging evidence indicates that EPBD has also a lower long-term complication rate than EST for treating choledocholithiasis.11,29 Whereas EPBD preserves the sphincter function,5,10,40 EST ablates the sphincter, and subsequent duodenobiliary reflux and bacterial colonization of bile duct predispose to stone recurrence and cholangitis with potential morbidity and need for reintervention,4,5,10,29 which occurred in 10.1% after EPBD compared with 25.0% after EST (P ⫽ .0016) during a median follow-up of 6.7 years in a multicenter RCT.29 Because of comparable overall success of stone extraction,1,11 EPBD may be preferred over EST because of a lower short-term and long-term complication rate. The longest EPBD duration ever reported is 5 minutes, and our meta-regression shows that overall complications decrease with increasing EPBD duration up to 5 minutes. It is unknown whether dilation duration longer than 5 minutes may further reduce pancreatitis risk, but the degree of sphincter loosening can only increase to a certain point and cannot exceed that achieved by EST, which ablates the sphincter; thus, the pancreatitis risk of EPBD is at best comparable with and is unlikely to
1107
be lower than that of EST. Further studies on EPBD with different durations will be helpful to corroborate our findings and determine the optimal duration. An important but controversial issue is whether pancreatic duct should be routinely stented after EPBD to prevent pancreatitis. Although this has been recommended,41 only one retrospective case-control study had evaluated this strategy but found no significant benefit.42 Pancreatic stenting increases procedure complexity and requires additional follow-up to ensure stent passage or removal.41,43,44 In the trials included, only 1 patient received pancreatic stenting2 and had little influence on our results, which indicate that EPBD for an adequate duration (eg, 5 minutes) provides simple and effective prevention for pancreatitis, and routine pancreatic stenting is not warranted. The strengths of our meta-analyses are that our analyses were conducted by separating studies into short and long EPBD durations, and we used both standard and Bayesian approaches to ensure our results are robust. Our results suggest a rethink about how bile duct stones can be best treated and how EPBD should be performed. Our results also give new insights into the elusive mechanism of post-EPBD pancreatitis and the disparities in pancreatitis rates in previous trials. The agreement among different statistical methods and direct/indirect comparisons further strengthens our results. This study also had limitations. Only one trial had directly compared short and long EPBD, and to address this limitation we used Bayesian network meta-analysis to incorporate evidence from indirect comparison of short vs long EPBD by using evidence from trials comparing long EPBD vs EST and trials comparing short EPBD vs EST. We assessed consistency between direct and indirect comparisons and found that results from direct and indirect comparisons are consistent, indicating the conclusion that short EPBD has a higher risk of pancreatitis and overall complications than long EPBD is quite robust. Traditional meta-analyses and meta-regression of trials comparing EPBD with EST also give the same conclusion. Second, we excluded 4 studies that compared EST with EPBD but did not report the duration for EPBD. Although this might result in loss of information, there were also other concerns about those studies; 3 of the 4 studies were abstracts only, and the other study was a brief report. Because of a lack of sufficient information on patient characteristics, methodology, and follow-up, we could not fully assess the eligibility and risk of bias of those studies and therefore decided to exclude them in our analyses. Last, it should be noted that our meta-analyses only pertain to choledocholithiasis, and our conclusions cannot be generalized to other diseases requiring sphincterotomy such as sphincter of Oddi dysfunction.
Conclusions Our meta-analyses suggest that compared with EST, treating bile duct stones with EPBD ⬎1 minute may reduce bleeding/overall complications without increasing the risk of pancreatitis. Currently recommended EPBD with ⱕ1-minute duration has a higher pancreatitis risk than EPBD ⬎1 minute and EST. EPBD with an adequate duration (around 5 minutes) has lower complication rates than the current standard of EST and may be used as the first-line treatment for bile duct stones.
1108
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org, and at http://dx. doi.org/10.1016/j.cgh.2012.05.017.
19.
20.
References 1. Baron TH, Harewood GC. Endoscopic balloon dilation of the biliary sphincter compared to endoscopic biliary sphincterotomy for removal of common bile duct stones during ERCP: a metaanalysis of randomized, controlled trials. Am J Gastroenterol 2004;99: 1455–1460. 2. Disario JA, Freeman ML, Bjorkman DJ, et al. Endoscopic balloon dilation compared with sphincterotomy for extraction of bile duct stones. Gastroenterology 2004;127:1291–1299. 3. Vlavianos P, Chopra K, Mandalia S, et al. Endoscopic balloon dilatation versus endoscopic sphincterotomy for the removal of bile duct stones: a prospective randomised trial. Gut 2003;52: 1165–1169. 4. Bergman JJ, van Berkel AM, Groen AK, et al. Biliary manometry, bacterial characteristics, bile composition, and histologic changes fifteen to seventeen years after endoscopic sphincterotomy. Gastrointest Endosc 1997;45:400 – 405. 5. Minami A, Nakatsu T, Uchida N, et al. Papillary dilation vs sphincterotomy in endoscopic removal of bile duct stones: a randomized trial with manometric function. Dig Dis Sci 1995;40:2550 –2554. 6. Bergman JJ, Rauws EA, Fockens P, et al. Randomised trial of endoscopic balloon dilation versus endoscopic sphincterotomy for removal of bile duct stones. Lancet 1997;349:1124 –1129. 7. Fujita N, Maguchi H, Komatsu Y, et al. Endoscopic sphincterotomy and endoscopic papillary balloon dilatation for bile duct stones: a prospective randomized controlled multicenter trial. Gastrointest Endosc 2003;57:151–155. 8. Komatsu Y, Kawabe T, Toda N, et al. Endoscopic papillary balloon dilation for the management of common bile duct stones: experience of 226 cases. Endoscopy 1998;30:12–17. 9. Lin CK, Lai KH, Chan HH, et al. Endoscopic balloon dilatation is a safe method in the management of common bile duct stones. Dig Liver Dis 2004;36:68 –72. 10. Sato H, Kodama T, Takaaki J, et al. Endoscopic papillary balloon dilatation may preserve sphincter of Oddi function after common bile duct stone management: evaluation from the viewpoint of endoscopic manometry. Gut 1997;41:541–544. 11. Weinberg BM, Shindy W, Lo S. Endoscopic balloon sphincter dilation (sphincteroplasty) versus sphincterotomy for common bile duct stones. Cochrane Database Syst Rev 2006;4: CD004890. 12. Attam R, Freeman ML. Endoscopic papillary balloon dilation for stone extraction: if, when, and for how long? Gastrointest Endosc 2010;72:1163–1166. 13. Liao WC, Lee CT, Chang CY, et al. Randomized trial of 1-minute versus 5-minute endoscopic balloon dilation for extraction of bile duct stones. Gastrointest Endosc 2010;72:1154 –1162. 14. Fletcher J. What is heterogeneity and is it important? BMJ 2007; 334:94 –96. 15. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009;151:264 –269,W64. 16. Attam R, Freeman ML. Endoscopic papillary large balloon dilation for large common bile duct stones. J Hepatobiliary Pancreat Surg 2009;16:618 – 623. 17. Attasaranya S, Cheon YK, Vittal H, et al. Large-diameter biliary orifice balloon dilation to aid in endoscopic bile duct stone removal: a multicenter series. Gastrointest Endosc 2008;67: 1046 –1052. 18. Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterot-
21. 22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36. 37.
38.
39.
omy complications and their management: an attempt at consensus. Gastrointest Endosc 1991;37:383–393. Higgins JP, Altman DG. Assessing risk of bias in included studies. In: Higgins JP, Green S (eds). Cochrane handbook for systematic reviews of interventions. Chichester, UK: John Wiley & Sons, 2011:607– 631. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557–560. Thompson SG, Sharp SJ. Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med 1999;18:2693–2708. Egger M, Davey Smith G, Schneider M, et al. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997;315: 629 – 634. Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med 2004;23:1351–1375. Sutton AJ, Abrams KR. Bayesian methods in meta-analysis and evidence synthesis. Stat Methods Med Res 2001;10: 277–303. Ades AE, Sculpher M, Sutton A, et al. Bayesian methods for evidence synthesis in cost-effectiveness analysis. Pharmacoeconomics 2006;24:1–19. Song F, Altman DG, Glenny AM, et al. Validity of indirect comparison for estimating efficacy of competing interventions: empirical evidence from published meta-analyses. BMJ 2003;326:472. Sutton A, Ades AE, Cooper N, et al. Use of indirect and mixed treatment comparisons for technology assessment. Pharmacoeconomics 2008;26:753–767. Brooks SP, Gelman A. General methods for monitoring convergence of iterative simulations. J Comput Graph Stat 1998;7: 434 – 455. Yasuda I, Fujita N, Maguchi H, et al. Long-term outcomes after endoscopic sphincterotomy versus endoscopic papillary balloon dilation for bile duct stones. Gastrointest Endosc 2010;72: 1185–1191. de los Campo Fernandez de los Rios R, Brullet Benedí E. Endoscopic treatment of choledocholithiasis in patients with Billroth II gastrectomy by dilatation with papillary balloon. Gastroenterol Hepatol 1998;21:391–393. Liao WC, Huang SP, Wu MS, et al. Comparison of endoscopic papillary balloon dilatation and sphincterotomy for lithotripsy in difficult sphincterotomy. J Clin Gastroenterol 2008;42:295–299. Mac Mathuna P, Siegenberg D, Gibbons D, et al. The acute and long-term effect of balloon sphincteroplasty on papillary structure in pigs. Gastrointest Endosc 1996;44:650 – 655. Elta GH, Elmunzer BJ. Biliary tract motor function and dysfunction. In: Feldman M, Friedman LS, Brandt LJ (eds). Sleisenger & Fordtran’s gastrointestinal and liver disease. 9th ed. Philadelphia: W. B. Saunders, 2010:1067–1076. Ersoz G, Tekesin O, Ozutemiz AO, et al. Biliary sphincterotomy plus dilation with a large balloon for bile duct stones that are difficult to extract. Gastrointest Endosc 2003;57:156 –159. Jeong S, Ki SH, Lee DH, et al. Endoscopic large-balloon sphincteroplasty without preceding sphincterotomy for the removal of large bile duct stones: a preliminary study. Gastrointest Endosc 2009;70:915–922. Hardy RJ, Thompson SG. Detecting and describing heterogeneity in meta-analysis. Stat Med 1998;17:841– 856. Freeman ML, Nelson DB, Sherman S, et al. Complications of endoscopic biliary sphincterotomy. N Engl J Med 1996;335: 909 –918. Anderson MA, Ben-Menachem T, Gan SI, et al. Management of antithrombotic agents for endoscopic procedures. Gastrointest Endosc 2009;70:1060 –1070. Boustière C, Veitch A, Vanbiervliet G, et al. Endoscopy and antiplatelet agents: European Society of Gastrointestinal Endoscopy (ESGE) guideline. Endoscopy 2011;43:445– 461.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
40. Yasuda I, Tomita E, Enya M, et al. Can endoscopic papillary balloon dilation really preserve sphincter of Oddi function? Gut 2001;49:686 – 691. 41. Freeman ML. Pancreatic stents for prevention of post-endoscopic retrograde cholangiopancreatography pancreatitis. Clin Gastroenterol Hepatol 2007;5:1354 –1365. 42. Aizawa T, Ueno N. Stent placement in the pancreatic duct prevents pancreatitis after endoscopic sphincter dilation for removal of bile duct stones. Gastrointest Endosc 2001;54: 209 –213. 43. Freeman ML, Overby C, Qi D. Pancreatic stent insertion: consequences of failure and results of a modified technique to maximize success. Gastrointest Endosc 2004;59:8 –14. 44. Smithline A, Silverman W, Rogers D, et al. Effect of prophylactic main pancreatic duct stenting on the incidence of biliary endoscopic sphincterotomy-induced pancreatitis in high-risk patients. Gastrointest Endosc 1993;39:652– 657. 45. Ochi Y, Mukawa K, Kiyosawa K, et al. Comparing the treatment outcomes of endoscopic papillary dilation and endoscopic sphincterotomy for removal of bile duct stones. J Gastroenterol Hepatol 1999;14:90 –96. 46. Arnold JC, Benz C, Martin WR, et al. Endoscopic papillary balloon dilation vs sphincterotomy for removal of common bile duct stones: a prospective randomized pilot study. Endoscopy 2001;33:563–567. 47. Natsui M, Narisawa R, Motoyama H, et al. What is an appropriate
1109
indication for endoscopic papillary balloon dilation? Eur J Gastroenterol Hepatol 2002;14:635– 640. 48. Tanaka S, Sawayama T, Yoshioka T. Endoscopic papillary balloon dilation and endoscopic sphincterotomy for bile duct stones: long-term outcomes in a prospective randomized controlled trial. Gastrointest Endosc 2004;59:614 – 618.
Reprint requests Address requests for reprints to: Yu-Kang Tu, PhD, Division of Biostatistics, Leeds Institute of Genetics, Health and Therapeutics, Room 8.01, Level 8, Worsley Building, University of Leeds, Leeds, United Kingdom. e-mail: [email protected]; fax: 44-113-3434877; or KuoLiong Chien, PhD, Graduate Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, 5F, No. 17, Hsu Chow Road, Taipei, Taiwan. e-mail: [email protected]; fax: 886-2-23511955. Conflicts of interest The authors disclose no conflicts. Funding Supported by an International Joint Grant from the Royal Society, London, UK, and National Science Council, Taiwan, and by National Taiwan University Hospital.
SYSTEMATIC REVIEWS AND META-ANALYSES Fasiha Kanwal, Section Editor
Balloon Dilation With Adequate Duration Is Safer Than Sphincterotomy for Extracting Bile Duct Stones: A Systematic Review and Meta-analyses WEI–CHIH LIAO,*,‡ YU–KANG TU,‡,§ MING–SHIANG WU,* HSIU–PO WANG,* JAW–TOWN LIN,*,储 JOSEPH W. LEUNG,¶ and KUO–LIONG CHIEN*,‡ *Department of Internal Medicine, National Taiwan University Hospital, National Taiwan University College of Medicine, Taipei, Taiwan; ‡Graduate Institute of Epidemiology and Preventive Medicine, National Taiwan University, Taipei, Taiwan; §Division of Biostatistics, Leeds Institute of Genetics, Health & Therapeutics and Leeds Dental Institute, University of Leeds, Leeds, United Kingdom; 储Department of Internal Medicine, E-DA Hospital and I-Shou University, Kaohsiung County, Taiwan; and ¶Davis Medical Center, University of California and Section of Gastroenterology, Sacramento VA Medical Center, Sacramento, California
BACKGROUND & AIMS: Endoscopic sphincterotomy (EST) is the standard treatment for choledocholithiasis. Endoscopic papillary balloon dilation (EPBD) has a lower risk for bleeding than EST, but EPBD is reserved for patients with bleeding diathesis because some studies reported that it increases the risk for pancreatitis. A short dilation time (ⱕ1 minute) is therefore recommended to reduce pancreatitis. However, there is evidence for an inverse relationship between EPBD duration and pancreatitis, prompting reevaluation of the optimal duration and relative safety of EPBD vs EST. METHODS: We systematically reviewed randomized controlled trials to compare long EPBD (⬎1 minute), short EPBD (ⱕ1 minute), and EST regarding pancreatitis and overall complications. In addition to pairwise meta-analyses, Bayesian network meta-analysis was undertaken to compare the 3 procedures together. Relation between duration and outcome was also analyzed by meta-regression. RESULTS: Compared with EST, short EPBD had a higher risk for pancreatitis (odds ratio [OR] by traditional analysis, 3.87; 95% confidence interval, 1.08 – 13.84 and OR by network meta-analysis, 4.14; 95% credible interval, 1.58 –12.56), but long EPBD did not pose a higher risk than EST (1.14, 0.56 –2.35 and 1.07, 0.38 –2.76). Long EPBD had a lower overall rate of complications than EST (0.61, 0.36 –1.04 and 0.54, 0.20 –1.36). In network meta-analysis, probabilities of being the safest treatment for long EPBD/short EPBD/EST regarding pancreatitis and overall complications were 43.9%/0.2%/55.9% and 90.3%/1.3%/8.4%, respectively. CONCLUSIONS: Duration of EPBD is inversely associated with pancreatitis risk. Currently recommended <1-minute dilation actually increases pancreatitis. EPBD with adequate duration may be preferred over EST because of comparable pancreatitis but lower overall complication rates. Keywords: Gallbladder; Mixed Treatment Comparison; Sphincteroplasty; Risk Factor.
Watch this article’s video abstract and others at http:// tiny.cc/bz9jv. Scan the quick response (QR) code to the left with your mobile device to watch this article’s video abstract and others. Don’t have a QR code reader? Get one at mobiletag.com/en/download.php.
E
ndoscopic retrograde cholangiopancreatography (ERCP) and endoscopic sphincterotomy (EST) are the standard treatment for choledocholithiasis,1–3 which cuts biliary sphincter to open the bile duct orifice for stone extraction.4,5 An alternative to EST is endoscopic papillary balloon dilation (EPBD), which uses a balloon of 6 –10 mm in diameter to dilate biliary sphincter.5–10 Two meta-analyses of randomized controlled trials (RCTs) consistently found that EST and EPBD had comparable overall success rates of stone removal (94.3% vs 96.5%) and similar rates of cholangitis, perforation, mortality, and overall complications (10.5% vs 10.3%) for treating choledocholithiasis.1,11 Those meta-analyses, however, also showed that EPBD was less likely to cause bleeding (0% vs 2%) but more likely to cause pancreatitis (7.4% vs 4.3%) than EST.1,11,12 Therefore, EPBD is reserved for patients with bleeding diathesis in current consensus,1,11,12 and a short dilation duration (ⱕ1 minute) is recommended to reduce pancreatitis.1,2,12 However, the risk of pancreatitis after EPBD in previous trials varied greatly between 0% and 15.4%.1,11,13 Emerging evidence suggests that this disparity may be explained by differences in dilation duration of EPBD in those trials,12,13 and this raises concerns that dilation duration might be a source of confounding14 and prompts reevaluation of the current recommendation with consideration of EPBD duration.12,13 A recent RCT found that the risk of pancreatitis after EPBD with 5-minute duration was comparable to EST and was lower than EPBD with 1-minute duration.13 Contrary to current recommendation, these findings suggest that EPBD with ⱕ1-minute dilation actually increases the risk of pancreatitis, whereas EPBD with adequate duration may have a lower overall complication rate than EST because it reduces bleeding without causing more pancreatitis. Therefore, we reevaluated the relative safety of EPBD over EST and the optimal duration of EPBD by conducting a network meta-analysis to compare shortduration EPBD (ⱕ1 minute), long-duration EPBD (⬎1 minute), and EST, focusing on pancreatitis and overall complications. Abbreviations used in this paper: CI, confidence interval; CrI, credible interval; EPBD, endoscopic papillary balloon dilation; ERCP, endoscopic retrograde cholangiopancreatography; EST, endoscopic sphincterotomy; OR, odds ratio; RCT, randomized controlled trial. © 2012 by the AGA Institute 1542-3565/$36.00 http://dx.doi.org/10.1016/j.cgh.2012.05.017
1102
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
EST was more difficult and required special technique. Finally, 12 RCTs were eligible for analysis.
Data Extraction and Quality Assessment Two investigators (W.-C.L., H.-P.W.) independently reviewed full manuscripts of eligible trials and extracted information into an electronic database including patient characteristics, methods of EPBD and EST, and complications. Dilation duration was treated as a continuous variable and dichotomized into short (ⱕ1 minute) or long (⬎1 minute). For studies with repeated dilations, total duration was used for analysis. Studies were assessed independently by the same reviewers with Cochrane risk of bias tool (Supplementary Figure 1).19 Disagreement was resolved by joint review of the manuscript and consensus.
Data Synthesis and Analysis
Figure 1. Flow chart of literature search and selection.
Methods Data Sources and Searches This systematic review follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline.15 The MEDLINE was searched for RCTs of EPBD for bile duct stone until end of August 2011 without language or date restriction by using the following keywords: endoscopic papillary balloon dilation/dilatation, endoscopic sphincteroplasty, endoscopic balloon sphincter dilation/dilatation, and endoscopic balloon dilation/dilatation. Cochrane Collaboration Central Register of Controlled Clinical Trials, Cochrane Systematic Reviews, http://ClinicalTrials.gov, and bibliographies of included trials and related meta-analyses were manually searched for additional references.
Study Selection We identified 141 studies for review of title and abstract to find RCTs comparing EPBD vs EST or long EPBD vs short EPBD (Figure 1). Studies on endoscopic papillary large balloon dilation were excluded, because it combines EST and dilation with a large balloon for large/difficult stones and has increased risk of complications.16,17 After initial screening, full text of 48 potentially eligible articles was retrieved for detailed assessment. Studies were excluded if duration for EPBD was not clearly reported, definition of post-ERCP complications was different from the consensus guideline,18 or subjects had surgically altered anatomy (eg, Billroth-II anastomosis) in which
The outcomes analyzed were risks of post-ERCP pancreatitis and overall complications. A three-stage strategy was used for analysis. First, pairwise comparisons of long EPBD vs EST and short EPBD vs EST were performed with traditional meta-analysis by using Stata 10.0 (StataCorp, College Station, TX). Pooled odds ratios (ORs) and their 95% confidence intervals (CIs) were estimated with fixed-effect model if no significant heterogeneity existed and with randomeffects model if significant heterogeneity existed. Heterogeneity was evaluated by I2 and Cochran Q.20 Relations between dilation duration and outcomes were analyzed by metaregression with natural logarithm of OR as the dependent variable.21 Potential small study bias was evaluated by funnel plots and by tests of Egger et al.22 Because traditional meta-analysis excludes studies with no event in both treatment arms and can introduce bias when using continuity corrections for studies with no event in one arm,23 we also used Bayesian methods for pairwise meta-analyses. Bayesian methods handle sparse data and zero events without continuity corrections; thus, they include all evidence and provide validation for results of the first stage.24 Finally, a Bayesian network metaanalysis, also known as mixed treatment comparison, was undertaken to compare long EPBD, short EPBD, and EST together. Bayesian network meta-analysis preserves randomization in the trials and combines evidence from direct and indirect comparisons to provide more precise estimates.25–27 Bayesian meta-analyses were performed by using WinBUGS 1.4.3 with Markov Chain Monte Carlo methods (MRC Biostatistics Unit, Cambridge, UK). We used noninformative uniform and normal prior distributions25,27 and 3 different sets of starting values with 50,000 burn-ins and an additional 50,000 iterations for each chain, yielding 150,000 iterations to obtain the posterior distributions of model parameters. Convergence of iterations was evaluated by using Gelman–Rubin–Brooke statistic.28
Results All included RCTs have been published in full manuscript form. Outcomes and detailed characteristics are summarized in Table 1 and Supplementary Table 1. Eleven studies compared EPBD vs EST without significant small study bias (Supplementary Figure 2); four compared short EPBD (total n ⫽ 459) vs EST (total n ⫽ 464), and seven compared long
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
1103
Table 1. Outcomes of RCTs Included in the Analysis Complications, n (%) First author, year Short EPBD (ⱕ1 min) vs EST Bergman,6 1997 Vlavianous,3 2003 Fujita,7 2003 DiSario,2 2004 Long EPBD (⬎1 min) vs EST Minami,5 1995 Ochi,45 1999 Arnold,46 2001 Yasuda,40 2001 Natsui,47 2002 Lin,9 2004 Tanaka,48 2004 Short EPBD vs long EPBD Liao,13 2010
Treatment
EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST EPBD: EST
1 min 30 s 15 s 1 min
3 min 3 mina 2 minb 2 minb 2 min 5 min 2 min
EPBD: 1 min EPBD: 5 min
Participant, n
Pancreatitis
All
101 101 103 99 138 144 117 120
7 (6.9) 7 (6.9) 5 (4.9) 1 (1.0) 15 (10.9) 4 (2.8) 18 (15.4) 1 (0.8)
17 (16.8) 24 (23.8) 7 (6.8) 3 (3.0) 20 (14.5) 17 (11.8) 21 (17.9) 4 (3.3)
20 20 55 55 30 30 35 35 70 70 51 53 16 16
2 (10.0) 2 (10.0) 0 (0.0) 2 (3.6) 6 (20.0) 3 (10.0) 2 (5.7) 2 (5.7) 4 (5.7) 3 (4.3) 0 (0.0) 0 (0.0) 3 (18.8) 3 (18.8)
2 (10.0) 2 (10.0) 1 (1.8) 3 (5.5) 9 (30.0) 5 (16.7) 2 (5.7) 3 (8.6) 7 (10.0) 8 (11.4) 1 (2.0) 14 (26.4) 3 (18.8) 4 (25.0)
86 84
13 (15.1) 4 (4.8)
16 (18.6) 4 (4.8)
aThree bTwo
repeated dilations, each for 1 minute. repeated dilations, each for 1 minute.
EPBD (total n ⫽ 277) vs EST (total n ⫽ 279). Only one study compared long EPBD vs short EPBD.
Traditional Meta-analyses: Long Endoscopic Papillary Balloon Dilation vs Endoscopic Sphincterotomy and Short Endoscopic Papillary Balloon Dilation vs Endoscopic Sphincterotomy Pancreatitis. One study9 was excluded because no pancreatitis was reported in either treatment group. Dilation duration and estimated ORs of included trials are summarized in Figure 2. There was significant heterogeneity across studies comparing short EPBD vs EST (P ⫽ .034, I2 ⫽ 65.4%), but not in those comparing long EPBD vs EST (P ⫽ .822, I2 ⫽ 0%). Compared with EST, short EPBD had a significantly higher pancreatitis risk (pooled OR, 3.87; 95% CI, 1.08 –13.84; randomeffects model), but long EPBD did not pose a higher risk (pooled OR, 1.14; 95% CI, 0.56 –2.35; fixed-effect model) (Figure 2). Overall complications. There was significant heterogeneity across studies comparing short EPBD vs EST (P ⫽ .006, I2 ⫽ 76.2%), but not in those comparing long EPBD vs EST (P ⫽ .123, I2 ⫽ 40.3%). Compared with EST, short EPBD had a trend toward higher overall complication rate (pooled OR, 1.71; 95% CI, 0.67– 4.35; random-effects model). By contrast, long EPBD seemed to have a lower overall complication rate (pooled OR, 0.61; 95% CI, 0.36 –1.04; fixed-effect model) (Figure 2).
Meta-regression: Relation Between Dilation Duration and Outcomes For pancreatitis, one study9 was excluded because no pancreatitis occurred. The dilation durations varied between 0.25 and 3 minutes and explained 46.8% of the between-study variance of OR. Variation in OR from heterogeneity across the trials decreased from 34.8% to 7.4% after differences in dilation duration were considered. The regression coefficient of dilation duration was ⫺0.69 (95% CI, ⫺1.47 to 0.09; P ⫽ .076), meaning that every 1-minute increase in dilation duration up to 3 minutes was associated with a 49.8% (95% CI, ⫺9.4% to 77.0%) reduction in OR (Figure 3). For overall complications, dilation duration explained 33.6% of the between-study variance of OR with a regression coefficient of ⫺0.60 (95% CI, ⫺1.15 to ⫺0.05; P ⫽ .037), meaning that every 1-minute increase in dilation duration up to 5 minutes was associated with a 45.1% (95% CI, 4.9%– 68.3%) reduction in OR (Figure 3). The use of single or repeated dilations was not associated with ORs of either outcome.
Bayesian Pairwise Meta-analyses All eligible studies were included in Bayesian metaanalyses, and the results were similar to those of traditional meta-analyses (Table 2, Figure 4). Compared with EST, short EPBD had a higher risk of pancreatitis, whereas long EPBD had a comparable pancreatitis risk.
1104
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Figure 2. Forest plots for ORs of pancreatitis (A) or overall complications (B) stratified according to dilation duration.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
1105
Figure 3. Associations between dilation duration and natural logarithm of OR of pancreatitis (A) or overall complications (B). Circles are sized according to the precision of each OR (inverse of within-study variance).
Network Meta-analyses With Bayesian Mixed Treatment Comparison For pancreatitis, short EPBD had a higher risk than both EST (OR, 4.14; 95% credible interval [CrI], 1.58 –12.56) and long EPBD (OR, 3.85; 95% CrI, 1.22–15.46), whereas long EPBD did not pose a higher risk than EST (OR, 1.07; 95% CrI, 0.38 –2.76) (Table 2, Figure 4). The OR of short vs long
EPBD estimated indirectly from short EPBD vs EST and long EPBD vs EST (3.86) was consistent with estimate from direct comparison (3.85). The probabilities of being ranked as the safest treatment for long EPBD, short EPBD, and EST were 43.9%, 0.2%, and 55.9%, respectively (Supplementary Figure 3). For overall complications, short EPBD had a trend toward higher risk than EST (OR, 1.90; 95% CrI, 0.67–5.70), whereas
1106
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Table 2. Comparisons of Short EPBD, Long EPBD, and EST: Pooled ORs for Pancreatitis and Overall Complications Pairwise meta-analysis
Pancreatitis Short EPBD vs EST Long EPBD vs EST Short vs long EPBD Overall complications Short EPBD vs EST Long EPBD vs EST Short vs long EPBD aOnly
Network meta-analysis
Conventional OR (95% CI)
Bayesian OR (95% CrI)
OR (95% CrI)
3.87 (1.08–13.84) 1.14 (0.56–2.35) 3.56 (1.11–11.42)a
4.46 (1.08–22.49) 1.10 (0.36–2.85) 3.84 (0.25–60.89)a
4.14 (1.58–12.56) 1.07 (0.38–2.76) 3.85 (1.22–15.46)
1.71 (0.67–4.35) 0.61 (0.36–1.04) 4.57 (1.46–14.33)a
1.77 (0.50–7.05) 0.57 (0.18–1.62) 4.99 (0.33–78.18)a
1.90 (0.67–5.70) 0.54 (0.20–1.36) 3.56 (1.01–13.85)
1 trial directly compared short vs long EPBD.
long EPBD had a trend toward lower risk than EST (OR, 0.54; 95% CrI, 0.20 –1.36) (Table 2, Figure 4). Short EPBD had a higher risk than long EPBD (OR, 3.56; 95% CrI, 1.01–13.85). The OR of short vs long EPBD estimated from indirect comparison (3.54) was consistent with estimate from direct comparison (3.56). The probabilities of being ranked as the safest treatment for long EPBD, short EPBD, and EST were 90.3%, 1.7%, and 8.0%, respectively (Supplementary Figure 3).
Discussion Although EPBD has a lower risk of post-ERCP bleeding1,11 and long-term complications than EST11,29 and is easier to perform in altered/difficult anatomy,30,31 EPBD is reserved for patients with bleeding diathesis by current consensus because some studies reported a higher risk of pancreatitis,2,7 and short dilation (ⱕ1 minute) is recommended for presumably lower pancreatitis risk.1,2,12 However, our meta-analyses indicate that short EPBD duration increases rather than decreases the risk of post-ERCP pancreatitis, and EPBD with adequate dura-
tion has a similar pancreatitis risk and a lower overall complication rate compared with EST for choledocholithiasis. This meta-analysis contradicts the common belief that pancreatitis results from direct pancreatic duct compression during balloon dilation, and thus dilation duration should be short.2,12 Recent evidence suggests that EPBD with ⱕ1-minute dilatation carries a higher risk of inadequate sphincter loosening, which increases the risks of pancreatitis and failed stone extraction.13,32 The sphincter of Oddi surrounds ampulla and wraps around the distal end of the bile and pancreatic duct.33 Whereas the sphincter is ablated by EST, it is only loosened by EPBD,13,32 and hemorrhage/inflammation and edema develop at ampulla within 2 hours after EPBD.32 An inadequately loosened sphincter from short-duration EPBD may limit volume expansion of its encircled contents as in a compartment syndrome; thus it worsens compression of pancreatic duct from post-EPBD edema and increases pancreatitis risk.13,33 The compartment theory is supported by our meta-analysis and also explains why pancreatitis is reduced when greater sphincter loosening is
Figure 4. Pooled ORs by traditional meta-analysis, Bayesian meta-analysis, and Bayesian network meta-analysis.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
achieved by a larger balloon,16,34,35 or when the sphincter had been ablated by prior EST.31 Two previous meta-analyses had compared EPBD with EST for treating choledocholithiasis and formed the basis of current consensus.1,11,12 Because the significance of dilation duration had not been recognized until recently, those meta-analyses pooled all studies regardless of dilation duration and concluded that EPBD had a higher pancreatitis risk than EST. Although a nonsignificant Cochran Q seemed to justify pooling of all trials in the above meta-analysis,11 Cochran Q is known to have a poor power in detecting true heterogeneity, especially when the number of studies is small as in the current case.20,36 In this study, we consistently noted that significant heterogeneity caused by differences in dilation duration did exist and must be considered to avoid misleading conclusions. Because short EPBD has a higher pancreatitis risk than EST but long EPBD has a similar risk, pooling short and long EPBD would suggest that EPBD has a higher pancreatitis risk than EST as in previous meta-analyses.1,11 Similarly for the overall complications, pooling higher-risk short EPBD with lower-risk long EPBD would suggest that EPBD and EST have comparable overall complication rates as in previous meta-analyses.1,11 The lower overall complication rate of long EPBD for choledocholithiasis may be attributed to avoidance of bleeding from sphincter cutting in EST1,11,12 without increasing the risk of pancreatitis. Because an estimated 150,000 ESTs are performed annually in the United States with a 2% risk of bleeding,1,37 using EPBD rather than EST as the first-line treatment might prevent 3000 cases of bleeding annually and the associated morbidity/mortality and costs, and this benefit may become greater with the increasing use of antithrombotic agents. EST is a high-risk procedure in patients taking antithrombotic medications, and guidelines recommend discontinuation of antithrombotic agents for 5–10 days before EST.38,39 This causes delayed treatment and prolonged admission, and the interruption of antithrombotic therapy can cause devastating thromboembolic events.38 By contrast, EPBD can be safely performed with continued antithrombotic therapy,39 which simplifies patient management and avoids both the risk of bleeding and thromboembolism. Besides a lower short-term complication rate, emerging evidence indicates that EPBD has also a lower long-term complication rate than EST for treating choledocholithiasis.11,29 Whereas EPBD preserves the sphincter function,5,10,40 EST ablates the sphincter, and subsequent duodenobiliary reflux and bacterial colonization of bile duct predispose to stone recurrence and cholangitis with potential morbidity and need for reintervention,4,5,10,29 which occurred in 10.1% after EPBD compared with 25.0% after EST (P ⫽ .0016) during a median follow-up of 6.7 years in a multicenter RCT.29 Because of comparable overall success of stone extraction,1,11 EPBD may be preferred over EST because of a lower short-term and long-term complication rate. The longest EPBD duration ever reported is 5 minutes, and our meta-regression shows that overall complications decrease with increasing EPBD duration up to 5 minutes. It is unknown whether dilation duration longer than 5 minutes may further reduce pancreatitis risk, but the degree of sphincter loosening can only increase to a certain point and cannot exceed that achieved by EST, which ablates the sphincter; thus, the pancreatitis risk of EPBD is at best comparable with and is unlikely to
1107
be lower than that of EST. Further studies on EPBD with different durations will be helpful to corroborate our findings and determine the optimal duration. An important but controversial issue is whether pancreatic duct should be routinely stented after EPBD to prevent pancreatitis. Although this has been recommended,41 only one retrospective case-control study had evaluated this strategy but found no significant benefit.42 Pancreatic stenting increases procedure complexity and requires additional follow-up to ensure stent passage or removal.41,43,44 In the trials included, only 1 patient received pancreatic stenting2 and had little influence on our results, which indicate that EPBD for an adequate duration (eg, 5 minutes) provides simple and effective prevention for pancreatitis, and routine pancreatic stenting is not warranted. The strengths of our meta-analyses are that our analyses were conducted by separating studies into short and long EPBD durations, and we used both standard and Bayesian approaches to ensure our results are robust. Our results suggest a rethink about how bile duct stones can be best treated and how EPBD should be performed. Our results also give new insights into the elusive mechanism of post-EPBD pancreatitis and the disparities in pancreatitis rates in previous trials. The agreement among different statistical methods and direct/indirect comparisons further strengthens our results. This study also had limitations. Only one trial had directly compared short and long EPBD, and to address this limitation we used Bayesian network meta-analysis to incorporate evidence from indirect comparison of short vs long EPBD by using evidence from trials comparing long EPBD vs EST and trials comparing short EPBD vs EST. We assessed consistency between direct and indirect comparisons and found that results from direct and indirect comparisons are consistent, indicating the conclusion that short EPBD has a higher risk of pancreatitis and overall complications than long EPBD is quite robust. Traditional meta-analyses and meta-regression of trials comparing EPBD with EST also give the same conclusion. Second, we excluded 4 studies that compared EST with EPBD but did not report the duration for EPBD. Although this might result in loss of information, there were also other concerns about those studies; 3 of the 4 studies were abstracts only, and the other study was a brief report. Because of a lack of sufficient information on patient characteristics, methodology, and follow-up, we could not fully assess the eligibility and risk of bias of those studies and therefore decided to exclude them in our analyses. Last, it should be noted that our meta-analyses only pertain to choledocholithiasis, and our conclusions cannot be generalized to other diseases requiring sphincterotomy such as sphincter of Oddi dysfunction.
Conclusions Our meta-analyses suggest that compared with EST, treating bile duct stones with EPBD ⬎1 minute may reduce bleeding/overall complications without increasing the risk of pancreatitis. Currently recommended EPBD with ⱕ1-minute duration has a higher pancreatitis risk than EPBD ⬎1 minute and EST. EPBD with an adequate duration (around 5 minutes) has lower complication rates than the current standard of EST and may be used as the first-line treatment for bile duct stones.
1108
LIAO ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 10
Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org, and at http://dx. doi.org/10.1016/j.cgh.2012.05.017.
19.
20.
References 1. Baron TH, Harewood GC. Endoscopic balloon dilation of the biliary sphincter compared to endoscopic biliary sphincterotomy for removal of common bile duct stones during ERCP: a metaanalysis of randomized, controlled trials. Am J Gastroenterol 2004;99: 1455–1460. 2. Disario JA, Freeman ML, Bjorkman DJ, et al. Endoscopic balloon dilation compared with sphincterotomy for extraction of bile duct stones. Gastroenterology 2004;127:1291–1299. 3. Vlavianos P, Chopra K, Mandalia S, et al. Endoscopic balloon dilatation versus endoscopic sphincterotomy for the removal of bile duct stones: a prospective randomised trial. Gut 2003;52: 1165–1169. 4. Bergman JJ, van Berkel AM, Groen AK, et al. Biliary manometry, bacterial characteristics, bile composition, and histologic changes fifteen to seventeen years after endoscopic sphincterotomy. Gastrointest Endosc 1997;45:400 – 405. 5. Minami A, Nakatsu T, Uchida N, et al. Papillary dilation vs sphincterotomy in endoscopic removal of bile duct stones: a randomized trial with manometric function. Dig Dis Sci 1995;40:2550 –2554. 6. Bergman JJ, Rauws EA, Fockens P, et al. Randomised trial of endoscopic balloon dilation versus endoscopic sphincterotomy for removal of bile duct stones. Lancet 1997;349:1124 –1129. 7. Fujita N, Maguchi H, Komatsu Y, et al. Endoscopic sphincterotomy and endoscopic papillary balloon dilatation for bile duct stones: a prospective randomized controlled multicenter trial. Gastrointest Endosc 2003;57:151–155. 8. Komatsu Y, Kawabe T, Toda N, et al. Endoscopic papillary balloon dilation for the management of common bile duct stones: experience of 226 cases. Endoscopy 1998;30:12–17. 9. Lin CK, Lai KH, Chan HH, et al. Endoscopic balloon dilatation is a safe method in the management of common bile duct stones. Dig Liver Dis 2004;36:68 –72. 10. Sato H, Kodama T, Takaaki J, et al. Endoscopic papillary balloon dilatation may preserve sphincter of Oddi function after common bile duct stone management: evaluation from the viewpoint of endoscopic manometry. Gut 1997;41:541–544. 11. Weinberg BM, Shindy W, Lo S. Endoscopic balloon sphincter dilation (sphincteroplasty) versus sphincterotomy for common bile duct stones. Cochrane Database Syst Rev 2006;4: CD004890. 12. Attam R, Freeman ML. Endoscopic papillary balloon dilation for stone extraction: if, when, and for how long? Gastrointest Endosc 2010;72:1163–1166. 13. Liao WC, Lee CT, Chang CY, et al. Randomized trial of 1-minute versus 5-minute endoscopic balloon dilation for extraction of bile duct stones. Gastrointest Endosc 2010;72:1154 –1162. 14. Fletcher J. What is heterogeneity and is it important? BMJ 2007; 334:94 –96. 15. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009;151:264 –269,W64. 16. Attam R, Freeman ML. Endoscopic papillary large balloon dilation for large common bile duct stones. J Hepatobiliary Pancreat Surg 2009;16:618 – 623. 17. Attasaranya S, Cheon YK, Vittal H, et al. Large-diameter biliary orifice balloon dilation to aid in endoscopic bile duct stone removal: a multicenter series. Gastrointest Endosc 2008;67: 1046 –1052. 18. Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterot-
21. 22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36. 37.
38.
39.
omy complications and their management: an attempt at consensus. Gastrointest Endosc 1991;37:383–393. Higgins JP, Altman DG. Assessing risk of bias in included studies. In: Higgins JP, Green S (eds). Cochrane handbook for systematic reviews of interventions. Chichester, UK: John Wiley & Sons, 2011:607– 631. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557–560. Thompson SG, Sharp SJ. Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med 1999;18:2693–2708. Egger M, Davey Smith G, Schneider M, et al. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997;315: 629 – 634. Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med 2004;23:1351–1375. Sutton AJ, Abrams KR. Bayesian methods in meta-analysis and evidence synthesis. Stat Methods Med Res 2001;10: 277–303. Ades AE, Sculpher M, Sutton A, et al. Bayesian methods for evidence synthesis in cost-effectiveness analysis. Pharmacoeconomics 2006;24:1–19. Song F, Altman DG, Glenny AM, et al. Validity of indirect comparison for estimating efficacy of competing interventions: empirical evidence from published meta-analyses. BMJ 2003;326:472. Sutton A, Ades AE, Cooper N, et al. Use of indirect and mixed treatment comparisons for technology assessment. Pharmacoeconomics 2008;26:753–767. Brooks SP, Gelman A. General methods for monitoring convergence of iterative simulations. J Comput Graph Stat 1998;7: 434 – 455. Yasuda I, Fujita N, Maguchi H, et al. Long-term outcomes after endoscopic sphincterotomy versus endoscopic papillary balloon dilation for bile duct stones. Gastrointest Endosc 2010;72: 1185–1191. de los Campo Fernandez de los Rios R, Brullet Benedí E. Endoscopic treatment of choledocholithiasis in patients with Billroth II gastrectomy by dilatation with papillary balloon. Gastroenterol Hepatol 1998;21:391–393. Liao WC, Huang SP, Wu MS, et al. Comparison of endoscopic papillary balloon dilatation and sphincterotomy for lithotripsy in difficult sphincterotomy. J Clin Gastroenterol 2008;42:295–299. Mac Mathuna P, Siegenberg D, Gibbons D, et al. The acute and long-term effect of balloon sphincteroplasty on papillary structure in pigs. Gastrointest Endosc 1996;44:650 – 655. Elta GH, Elmunzer BJ. Biliary tract motor function and dysfunction. In: Feldman M, Friedman LS, Brandt LJ (eds). Sleisenger & Fordtran’s gastrointestinal and liver disease. 9th ed. Philadelphia: W. B. Saunders, 2010:1067–1076. Ersoz G, Tekesin O, Ozutemiz AO, et al. Biliary sphincterotomy plus dilation with a large balloon for bile duct stones that are difficult to extract. Gastrointest Endosc 2003;57:156 –159. Jeong S, Ki SH, Lee DH, et al. Endoscopic large-balloon sphincteroplasty without preceding sphincterotomy for the removal of large bile duct stones: a preliminary study. Gastrointest Endosc 2009;70:915–922. Hardy RJ, Thompson SG. Detecting and describing heterogeneity in meta-analysis. Stat Med 1998;17:841– 856. Freeman ML, Nelson DB, Sherman S, et al. Complications of endoscopic biliary sphincterotomy. N Engl J Med 1996;335: 909 –918. Anderson MA, Ben-Menachem T, Gan SI, et al. Management of antithrombotic agents for endoscopic procedures. Gastrointest Endosc 2009;70:1060 –1070. Boustière C, Veitch A, Vanbiervliet G, et al. Endoscopy and antiplatelet agents: European Society of Gastrointestinal Endoscopy (ESGE) guideline. Endoscopy 2011;43:445– 461.
October 2012
BALLOON DILATION VS SPHINCTEROTOMY FOR CHOLEDOCHOLITHIASIS
40. Yasuda I, Tomita E, Enya M, et al. Can endoscopic papillary balloon dilation really preserve sphincter of Oddi function? Gut 2001;49:686 – 691. 41. Freeman ML. Pancreatic stents for prevention of post-endoscopic retrograde cholangiopancreatography pancreatitis. Clin Gastroenterol Hepatol 2007;5:1354 –1365. 42. Aizawa T, Ueno N. Stent placement in the pancreatic duct prevents pancreatitis after endoscopic sphincter dilation for removal of bile duct stones. Gastrointest Endosc 2001;54: 209 –213. 43. Freeman ML, Overby C, Qi D. Pancreatic stent insertion: consequences of failure and results of a modified technique to maximize success. Gastrointest Endosc 2004;59:8 –14. 44. Smithline A, Silverman W, Rogers D, et al. Effect of prophylactic main pancreatic duct stenting on the incidence of biliary endoscopic sphincterotomy-induced pancreatitis in high-risk patients. Gastrointest Endosc 1993;39:652– 657. 45. Ochi Y, Mukawa K, Kiyosawa K, et al. Comparing the treatment outcomes of endoscopic papillary dilation and endoscopic sphincterotomy for removal of bile duct stones. J Gastroenterol Hepatol 1999;14:90 –96. 46. Arnold JC, Benz C, Martin WR, et al. Endoscopic papillary balloon dilation vs sphincterotomy for removal of common bile duct stones: a prospective randomized pilot study. Endoscopy 2001;33:563–567. 47. Natsui M, Narisawa R, Motoyama H, et al. What is an appropriate
1109
indication for endoscopic papillary balloon dilation? Eur J Gastroenterol Hepatol 2002;14:635– 640. 48. Tanaka S, Sawayama T, Yoshioka T. Endoscopic papillary balloon dilation and endoscopic sphincterotomy for bile duct stones: long-term outcomes in a prospective randomized controlled trial. Gastrointest Endosc 2004;59:614 – 618.
Reprint requests Address requests for reprints to: Yu-Kang Tu, PhD, Division of Biostatistics, Leeds Institute of Genetics, Health and Therapeutics, Room 8.01, Level 8, Worsley Building, University of Leeds, Leeds, United Kingdom. e-mail: [email protected]; fax: 44-113-3434877; or KuoLiong Chien, PhD, Graduate Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, 5F, No. 17, Hsu Chow Road, Taipei, Taiwan. e-mail: [email protected]; fax: 886-2-23511955. Conflicts of interest The authors disclose no conflicts. Funding Supported by an International Joint Grant from the Royal Society, London, UK, and National Science Council, Taiwan, and by National Taiwan University Hospital.