The study was a randomized clinical trial. Randomization was performed using sealed envelopes in which patients were allocated in a 1:1 proportion to either a partially covered Wallstent^® Endoscopic Biliary Endoprosthesis with Permalume™ covering comprised of two components: the implantable metallic stent and the Unistep™ Plus Delivery System, (Boston Scientific, Natick, MA, United States) (self-expandable metal stent, SEMS), or a 10 French (Fr) Amsterdam-type polyethylene plastic stent (PS) biliary stent. The sealed envelopes were opened only at the time of intent of stent insertion in the ERCP suite after confirmation that all selection criteria had been fulfilled. The allocation sequence was performed centrally and patient enrolment and participant assignment was carried out by a third party not directly involved with the patient’s care or the measurement of outcomes. Neither patient, treating team, or the evaluators of outcomes were blinded to treatment allocation due to the nature of the intervention and follow-up care required. Each investigator received written approval for the study from his respective Institutional Review Boards prior to study initiation and patient enrollment. The trial did not require prior registration as it was started before 2004 and both stents are FDA approved.

Inclusion criteria were age 18 or older, and the provision of a signed written voluntary informed consent form approved by the Institutional Review Boards at participating centers. All patients demonstrated laboratory, imaging and/or histological evidence of malignant biliary obstruction. The cause of obstruction could be any intrinsic or extrinsic malignancy extending no more proximal than 1cm below the common hepatic ductal bifurcation. A Karnofsky performance scale was applied. A Karnofsky score > 60% is a validated measure of patient function, previously used in Pancreatico-biliary cancer patients[16]. Patients were required to haven anticipated life expectancy that would allow for completion of full follow-up. Exclusion criteria were jaundice related to intrahepatic cholestasis or obstruction, or a prior attempt at a curative surgical resection for the biliary obstructing lesion. There were 6 participating North American university centers (Fletcher Allen Health Care at the University of Vermont, McGill University Health Centre, St. Elizabeth’s Medical Center, Dartmouth Hitchcock Medical Center, Duke University Medical Center, and Thomas Jefferson University).

The primary endpoint of the trial was the time to occurrence of stent failure as defined by the appearance of one or more cholestatic symptoms accompanied by a 50% increase in bilirubin from the lowest post-stent insertion value recorded prior to this follow-up event, and/or cholangitis (defined as the new onset of pain, fever, and jaundice) whether associated with a stent replacement or not. Repeat ERCP for stent replacement or suspected obstruction using a stent of any type was considered to represent a stent failure. Secondary outcomes were death, cholestatic symptoms, laboratory data, technical success (defined as the successful delivery and deployment of the initial stent to the desired location in the biliary tree) measured at the time of the procedure, the presence of adverse events, and the Karnofsky performance score. We also administered the short form (SF)-36 general quality of life measurement scale, which is a questionnaire measuring patient’s perceptions about functional health and well-being previously administered to bilio-pancreatic cancer patients[1].

ERCP was performed by experienced endoscopists; stents were placed with or without prior dilatation or sphincterotomy after sealed-envelope randomization to stent type was done after confirmation of obstruction meeting inclusion criteria. The patient then received either the SEMS or the PS. The length of each type of stent was determined by the biliary anatomy and left to the discretion of the endoscopist as part of the medical effectiveness philosophy of the trial thereby enhancing generalizability of the results. A cholangiogram was performed to document stent patency and position. No prophylactic antibiotics were used. Each patient had one- and three-month follow-up, followed by quarterly scheduled follow-up sessions up to 2 years following stent insertion.

Failed plastic stents were replaced with covered metal stents, while failed covered metal stents were replaced with either one or more plastic or metal stent(s) inserted through the metal stent. The decision for the choice of stent type following stent failure was left to the discretion of the endoscopist and recorded.

Dedicated, standardized electronic case report forms were completed by trained research assistants and downloaded into a web-based remote data entry repository. Internal validity of recorded data and missing data quality checks were performed centrally by trained research personnel. At baseline, investigation variables included any significant medical history, the tumor type, stage, and location, the date of diagnosis, the length and maximum diameter of stricture, and administration of any prior anticancer treatment, as well as the Karnofsky Score. Variables assessed at baseline and at periodic follow-up visits (months 1, 3, 6, 9, 12, and if the patient survived, 3-monthly up to month 24) following the index procedure included a cholestatic symptom assessment, the use of any adjuvant treatment such as radiation and/or chemotherapy, laboratory test results (chemistry, hematology), and the Karnofsky index. All adverse events were recorded, including the occurrences of cholangitis, pancreatitis, and cholecystitis using standardized definitions[17].

The planned enrollment was 120 patients. Sample size predictions were calculated using a model of binomial proportions and independent samples. Assuming a 25% improvement in stent patency duration with expandable metallic stenting and using a 1-sided type I error rate of 5% and a type II error rate of 20%, approximately 60 patients were thought to be needed in each group.

Amongst descriptive variables, continuous variables are reported as means and standard deviations as well as medians where appropriate, and categorical variables as proportions. Inferential testing was carried out using t-tests for continuous and χ² for categorical variables. Karnofsky scores and quality of life scores were assessed for both intra- and between-group differences comparing baseline values to the last visit on record at the 1-mo, the 3-mo visit, and the 6-mo visits; both within and across groups. We used a t-test with either the pooled or Satterthwaite method, depending on the results of the equality of variances test at each follow-up period.

Survival analyses were performed for both stent failure and patient survival using both intention-to-treat (ITT) and per protocol (PP) analyses. In the first group, only subjects who had at least a 50% drop in bilirubin at 1 mo were included. In the second, all subjects were included as originally randomized. Kaplan-Meier curves were created for the SEMS and PS groups, and compared with a log-rank test. Cox proportional hazards regression models were also used to determine if significant covariates were associated with either time-to-stent failure or time-to-death. The proportional hazards assumption was tested with the use of a Kolmogorov-Smirnov Supremum test. The following covariates were included for these analyses in addition to stent randomization group: obstruction (for prediction of mortality), tumor type, known metastatic cancer, chemotherapy or radiation therapy, and the baseline Karnofsky score. Covariates that were associated with the outcome with a P-value of 0.15 or less in a univariate model were entered into a multivariable model. There was no planned interim analysis.

A total of 85 patients were accrued over 37 mo. The study was closed prior to completion of enrollment of the estimated 120 patients due to a marked slowing of patient accrual (trial fatigue). Of the 85 patients included, 42 were randomized to the SEMS group, and 43 to the PS group. Three patients had evaluable baseline patient data but were excluded from further analyses because of inclusion protocol violations (2 were never stented, and one received a metal covered stent when in fact randomized to plastic stent). Of the 82 patients with analyzable outcomes data, 41 received a SEMS and 41 a PS; the CONSORT diagram is shown in Figure 1. Population characteristics at baseline for both groups are shown in Table 1.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Consolidated standards of reporting trials diagram. SEMS: Self-expandable metal stent.

Twenty-nine point eight percent were inpatients; amongst these, the mean hospital stays related to the procedure were 2.5 ± 1.6 d in the SEMS group, and 4.9 ± 4.7 d in the PS group.

The mean length of Wallstents used was 61.4 ± 11.2 mm (median: 60 mm, range: 40-80 mm); 95.1% patients received a 10Fr diameter and 4.9% an 8F diameter. In the PC group, the stent length was 76.0 ± 18.2 mm (median: 70, range: 50-120 mm); all patients had a 10Fr diameter.

In ITT analysis, the time to stent failure was 385.3 ± 52.5 d in the SEMS and 153.3 ± 19.8 d in the PS group (P = 0.006) (Figure 2A). Corresponding results were 396.5 ± 56.8 d and 164.3 ± 24.1 d, respectively (P = 0.025) using a PP approach. After adjustment for possible confounding variables, in ITT analysis, the only independent significant predictor of a failed stent was the stent group allocation (HR = 0.29, 95%CI: 0.12-0.75, P = 0.011); similar findings were noted with the PP analysis (HR = 0.22, 95%CI: 0.06-0.80, P = 0.013)

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Survival analysis. A: Time to stent failure for partially covered self-expanding metal stent and plastic stent patient groups; B: Time to death for partially covered self-expanding metal stent and plastic stent patient groups.

Procedural outcomes: No differences in intra-procedural events were noted. Overall, 69.4% of all stent insertions were carried out in an out-patient setting. Optimal stent insertion and positioning was noted in 95.3% of patient with in the SEMS and 97.4% of patients in the PS groups, respectively. The length of the SEMS used was 61.4 ± 11.2 mm (median: 60 mm, range: 40-80 mm) and a diameter of 59.64 ± 11.2 mm (median: 60 mm, range: 40-80 mm); 95.1% patients received a 10Fr diameter and 4.9% a 8F diameter. In the PC group, the stent length was 76.0 ± 18.2 mm (median: 70 mm, range: 50-120 mm); All patients had a 10Fr diameter. Sphincterotomy was carried out prior to stent insertion in 18.8% of cases, and balloon dilatation in 3.5%. The distal end of the stent was positioned outside the CBD into the duodenum in 93.0%. Some form of tissue sampling was carried out at the time of ERCP in 56.0% of patients.

Time to death: The time to death did not differ between both groups: 192.3 ± 23.4 d for SEMS vs 211.5 ± 28.0 d for PS (P = 0.70) using an ITT approach (Figure 2B). Similar conclusions were reached using the PP approach 248.5 ± 26.8 d vs 251.3 ± 32.5 d, respectively (P = 0.66). After adjustment for possible confounding variables, no significant predictor of time to death was found in ITT or PP analysis.

Additional secondary outcomes: Complications including the development of pancreatitis, cholangitis, and cholecystitis are (2.4% vs 2.4%, P = 1.0000; 4.9% vs 24.5%, P = 0.029; 4.8% vs 0.0%, P = 0.4741). Only cholangitis differed, with a greater frequency in the PS group.

The percentage reduction in bilirubin value from baseline to the 1-mo visit was no different in SEMS than in the PS group [74.0%, (95%CI: 60.0-87.9) vs 63.7% (95%CI: 45.5-81.9), respectively, P = 0.37]. No statistical differences in Karnoksy performance scores were noted between the two treatment groups when comparing the differences in scores for the last, 1, 3, and 6 mo visits compared to baseline. In the pre-planned paired analysis to assess intra-group differences, patients receiving the SEMS, showed significant improvements noted at 6 mo and at the last available visit (P = 0.015, and P = 0.022, respectively). There were also significant improvements noted for patients in the PS group compared to baseline both at 1 mo and at the last available date of follow-up (P = 0.045, and P = 0.0014, respectively; full data available upon request).

Overall, 29.4% of patients had one or more cholestatic symptoms at follow-up, 24.3% for the SEMS group and 33.3% for the PS group (P = 0.213). Additional symptom reporting showed no difference between both groups with regards to individual cholestatic symptoms (data not shown).

Quality of life-SF-36 measures: Seventy-four patients answered the quality of life questionnaires over a total of 174 visits during a 12-mo follow-up (31 patients answered only once to the SF-36 questionnaire, 17 answered to two questionnaire and 13 responded to 3 questionnaires). Among these, 38 had received a SEMS and 36 a PS. At baseline, patients in the SEMS group exhibited lower means than those in the PS group for all 8 summary scores, indicating worse quality of life parameters; the differences however were not statistically significant except for physical functioning (46.4 vs 63.9, P = 0.008). The SEMS group scores improved gradually such that, by 9 mo, most were arithmetically greater than scores from the PS group although without significant differences. There remained, however, only a very small number of patients able to complete the questionnaires in follow-up (9 patients at 9 mo, and 5 at 12 mo). In paired analysis, statistical significant improvements were noted amongst SEMS patients in physical functioning (6 mo vs baseline), and vitality (1 mo vs baseline). Significant bettering of quality of life was noted amongst PS patients at 1 month vs baseline for bodily pain, social functioning, and mental health, as well as in vitality for the 9 mo vs baseline comparison (full quality of life scores are available upon request).

Despite existing evidence in the literature favoring metal over platis stenting in distal biliary obstruction palliation, few data exist assessing partially covered metal stents, especially in a contemporary setting.

Plastuc stent palliation remains widespread, while partially covered metal biliary stents appear to migrate more frequently than their uncovered counterparts, but have gained popularity.

There have been 2 previously published studies in the literature that have compared partially covered metal to plastic biliary stenting with limited generalizability, suggesting the superiority of the metal stent alternative.

The current article validates the conclusion that partially covered metal stents result in a significantly longer time to stent patency, without a prolonged survival time in patients undergoing palliation for distal biliary malignant obstruction. Additional and summary data are required to confirm the robustness of the latter finding, while the advent of completely covered metal stents require further assessments, owing to the theoretical benefit of decreased stent in growth versus the possible increased risk of stent migration.

The findings of this randomized trial appear to confirm that of two previous studies in the literature and is limited by the small sample size, even though a signififcant difference in the primary outcome measure was noted. The achieved statistical power limits the interpretation of other endpoints.