- Email: [email protected]
Prospective randomized trial comparing Tannenbaum Teflon and standard polyethylene stents in distal malignant biliary stenosis
Prospective randomized trial comparing Tannenbaum Teflon and standard polyethylene stents in distal malignant biliary stenosis Vittorio Terruzzi, MD, Umberto Comin, MD, Federico De Grazia, MD, Gian Luigi Toti, MD, Alessandro Zambelli, MD, Sandro Beretta, MD, Giorgio Minoli, MD Como, Magenta, Crema, Milan, Italy
Background: Our aim was to compare the clinical efficacy of the Tannenbaum (TB) biliary prostheses, a recently designed Teflon stent without side holes, with the Cotton-Huibregtse (CH) polyethylene stent. Methods: Fifty-seven patients (26 men, mean age 75.5 years) with unresectable malignant tumors and distal biliary stenosis were included (38 pancreatic head cancer, 17 cholangiocarcinoma, 2 ampullary cancer). Patients were prospectively randomized to have a 10F, 7 cm long TB (29 patients) or CH (28 patients) stent inserted endoscopically. Four patients (2 TB and 2 CH) were excluded: 3 because of the failure of stent insertion and 1 because of a protocol violation. The patients were evaluated clinically and, if necessary, with biochemical tests every month until death or until they needed surgery for symptoms of gastric outlet obstruction. When occlusion or dislocation occurred, the stent was replaced with one of the same type. Cumulative stent patency and patient survival were estimated with the Kaplan-Meier life-table analysis and compared by log-rank chi-square test. Results: The two groups were comparable in mean age, gender, and diagnosis. The patients were followed for a mean of 145.5 days (range 24 to 613); by the end of the study 47 patients (81%) had died or developed symptoms of gastric outlet obstruction. Median survival was 88 days (range 24 to 613) in the TB group and 75.6 days (23 to 486) in the CH group. Stent exchange (occlusion 16, dislocation 3) was necessary for 5 patients in the TB group and 7 in the CH group. No statistical difference was found on comparing the mean duration of function of the first, second, and third stents (analysis of variance, p = 0.80). The median duration of stent function was 96 days (range 11 to 613) in the TB group and 75.5 days (range 23 to 323) in the CH group. No significant difference was found in either survival time (p = 0.48) or stent patency (p = 0.12). Conclusion: This study found no significant advantage of the Tannenbaum prostheses over the standard polyethylene stent in the palliation for patients with distal malignant biliary stenosis with regard to survival or length of stent patency. (Gastrointest Endosc 2000;51:23-7.)
Most patients with biliary obstruction due to cancer of the head of the pancreas and distal cholangiocarcinoma have advanced disease at diagnosis, and only a few are candidates for curative surgery.1 Biliary enteric bypass is unquestionably effective for palliation in patients with biliary or duodenal Received December 23, 1997. For revision April 9, 1998. Accepted September 3, 1999. From Unità di Gastroenterologia, Divisione di Medicina 2, and Unità di Biostatistica, Ospedale Valduce, Como; Servizio di Endoscopia, Ospedale Civile di Magenta, Milano; Servizio di Endoscopia, Ospedale Maggiore, Crema; and Ospedale Civile di Sesto S. Giovanni, Milano, Italy. Presented in part at the annual meeting of the American Society for Gastrointestinal Endoscopy, May 11-14, 1997, Washington, D.C. (Gastrointest Endosc 1997;45:507). Supported in part by Cook Italia. Reprint requests: Vittorio Terruzzi, MD, Divisione di Medicina 2, Ospedale Valduce, via Dante 11, I - 22100 Como, Italy. Copyright © 2000 by the American Society for Gastrointestinal Endoscopy 0016-5107/2000/$12.00 + 0 37/1/102787 VOLUME 51, NO. 1, 2000
obstruction due to advanced pancreatic cancer, but it is associated with high mortality and morbidity rates.2 Since 1979, endoscopic placement of plastic stents has offered a valid alternative to surgery.3 Stent insertion achieves comparable therapeutic results and improves quality of life.4 No difference in patient survival has been found between endoscopic and surgical biliary drainage.5 A persistent problem of endoscopically inserted biliary stents is late occlusion (within 3 to 6 months). Despite efforts to prolong patency with drugs or by changing stent diameter, design, and material, clogging by sludge remains problematic.6-9 Expandable metal stents have a significantly higher patency rate, but high cost precludes extensive use, especially in patients whose life expectancy is shorter than the median patency of the stent.10 In a pilot study, the patency rate of a newly designed Teflon stent without side holes (Tannenbaum stent; Wilson-Cook Medical Inc., Winston-Salem, N.C.) was longer than GASTROINTESTINAL ENDOSCOPY
23
V Terruzzi, U Comin, F De Grazia, et al.
that reported for polyethylene stents with side holes and similar to that reported for metallic expandable biliary prostheses.11 In a non-randomized, prospective study, this “Tannenbaum” stent remained patent significantly longer than a Teflon pigtail stent.12 The aim of this multicenter, prospective, randomized trial, was to compare the clinical efficacy of the “Tannenbaum” Teflon prosthesis (TB), with conventional “Cotton-Huibregtse” polyethylene stents (CH) in patients with previously untreated malignant distal common bile duct obstruction. PATIENTS AND METHODS Patients Patients were included if they had obstructive jaundice due to untreated malignancy involving the distal bile duct. Diagnosis was based on presenting symptoms, biochemical tests, abdominal ultrasound, and ERCP. CT was done as necessary when evaluating patients for surgery. All patients were initially considered for surgery and were only rejected because of disease extension or severe disability (old age and/or other comorbid conditions). Patients with duodenal obstruction or hepatic hilar involvement and those with other previous malignant diseases or a previous drainage procedure or gastric or bile duct surgery were excluded. Patients in poor general condition with a short life expectancy were also excluded. No other therapeutic procedures, such as percutaneous drainage, radiotherapy or chemotherapy, were used for biliary obstruction during the study. Informed consent was obtained from all patients before randomization, and the study was approved by the Internal Review Board of Valduce Hospital. Prostheses The “Tannenbaum” stent is a straight biliary prosthesis made from Teflon tubing, with no side holes and with multiple anchoring flaps that do not perforate the wall.11 This prosthesis was compared with the conventional “Cotton-Huibregtse” polyethylene stent with side holes. Both types were obtained from the same manufacturer (Wilson-Cook). Study design and methods In this multicenter (four digestive-endoscopic centers), prospective, single-blind study (patients were not informed as to the type of prosthesis), patients were consecutively randomized to have either a 10F diameter, 7 cm long TB stent, or a CH stent of the same size inserted endoscopically. Thirty minutes before endoscopy, all patients received cefazoline 1 gm intravenously. Randomization was done by means of a computed schedule blindly generated for each center participating in the study. After diagnostic ERCP the stents were inserted with a three-layer technique (guidewire, guide-catheter, pusher) with or without previous endoscopic sphincterotomy. ERCP complications and those related to stent insertion were assessed within 7 days of the procedure. End points of the study were stent occlusion, death, or the need for palliative surgery for duodenal obstruction due to tumor growth. 24
GASTROINTESTINAL ENDOSCOPY
Tannenbaum Teflon vs. standard polyethylene biliary stents
Follow-up data (clinical and when useful, biochemical data) were obtained at 1-month intervals by telephone from the referring physician, until death, or until surgical treatment was needed for gastric outlet obstruction. When occlusion or dislocation occurred, the stent was removed with a polypectomy snare or a Dormia basket and the same type of stent was used for replacement. Occlusion or dislocation was considered when patients developed cholangitis, jaundice, or flu-like syndrome associated with cholestasis (defined as an episode of biliary pain or abdominal discomfort, with jaundice and/or fever over 38° C, for more than 1 day and/or increase in liver function tests). Stent occlusion was assessed macroscopically after removal. Statistical analysis Stent patency. The number of stent placements required to demonstrate statistical significance with a 5% alpha error and a 90% power was 67 and was calculated on the basis of previous data,11,12 assuming a 40% difference in patency rate at 1-year between the two groups.13 With a lower (80%) power, estimated sample size was 51 placements assuming a 40% difference between the two groups, 65 for a 35% difference and 88 for a 30% difference. The Pearson chi-square test with Yates’ correction and the Fisher’s exact test for categorical comparison of data were used as appropriate. T test was used for continuous data comparison. Analysis of variance (ANOVA) test was applied to compare the mean patency of the first stent with the mean patency of the second and third ones when more than one stent replacement was required in the same patient. Stent patency was calculated in days, as the interval between the time of its replacement. Patient death or the appearance of gastric outlet obstruction due to duodenal stenosis were considered censoring events. Cumulative stent patency was estimated by KaplanMeier life-table analysis and the two groups (TB and CH) were compared by the log rank chi-square test. Cox regression was used to study the effect of other covariates on the time to stent occlusion. Number of previous prosthesis exchanges in the same patient and type (TB vs. CH) were considered potentially explanatory covariates in the stent patency survival function. A step-wise (step-down) regression strategy was used for variable selection. Log cumulative hazard plots were examined to test the proportionality of hazard assumption. Martingale residuals and deviance residuals plot were examined to test the goodness of fit of the model. Regression coefficients and 95% confidence intervals were calculated. A significance test was made using likelihood ratio test. Patient survival. The number of patients required with respect to survival was calculated so that testing would be done at 90% power and 5% level of significance. Assuming, on the basis of previous data,12 a 25% difference in 1-year patient survival between the two groups, estimated sample size was 112. The estimated sample size reduces to 85 patients assuming a 25% difference and 80% VOLUME 51, NO. 1, 2000
Tannenbaum Teflon vs. standard polyethylene biliary stents
V Terruzzi, U Comin, F De Grazia, et al.
power and increases to 137 with the same power and assuming a 20% difference between the two groups.13 Patient death was considered the end-point. Patient survival was calculated in days from the time of stent insertion. Survival time was censored at the last visit for patients lost during follow-up or submitted to palliative surgery for duodenal stenosis. Age, gender (male vs. female), diagnosis (pancreatic carcinoma vs. cholangiocarcinoma or ampullary tumor), and type of stent (TB vs. CH) were examined as potentially explanatory variables. As for stent patency, cumulative patient survival (days from the first stent insertion) and Cox regression were examined. All analyses were made using StatView 4.5 (Abacus Concepts, Inc., Berkeley, Calif.).
RESULTS Fifty-seven patients were included: 38 had pancreatic head cancer, 17 cholangiocarcinoma, and 2 ampullary tumors. There were 25 men and 32 women. Mean age was 75.5 years (range 41 to 90 years). After randomization 29 patients were treated with the TB stent, and 28 with the CH stent. Two TB and two CH patients were excluded, three because of unsuccessful stent insertion and one for a protocol violation (after the first insertion, the prostheses was replaced with a self-expandable metal stent). No major procedure-related complications, such as pancreatitis, cholangitis, perforation, or significant bleeding arose in either group. The two groups were comparable with regard to age, gender, initial serum bilirubin, and diagnosis (Table 1). The patients were followed for a mean of 145.5 days (range 24 to 613 days) after insertion of the first stent. Stent patency Stent occlusion occurred in 16 patients (7 TB and 9 CH). Stent migration, always into the duodenum, occurred in 2 TB patients and 1 CH. The number of stent exchanges was similar in the two groups (Table 2). When a stent exchange was made after occlusion, no significant difference was found between the mean duration of patency of the first stent (103.2 days), the second (120.8 days), and the third (123.4 days) inserted (p = 0.78). We therefore decided we could justifiably group all of the stents regardless of their placement order and assess the differences between TB and CH groups within the overall case list of 71 stents, thus achieving the necessary statistical power. Median duration of stent patency was 96 days (range 11 to 613) in the TB group and 75.5 days (23 to 323) in the CH group (Fig. 1 and Table 3). No differences were found in the two groups (stents = 72; TB = 34; CH = 38: log-rank chi-square = 3.039; p = VOLUME 51, NO. 1, 2000
Figure 1. Kaplan-Meier plots of patency rates for the “Tannenbaum” (circles) and “Cotton-Huibregtse” stents (squares). Total number of stents: 72 (TB = 34, CH = 38). Statistical analysis: log-rank chi-square test = 3.034; p = 0.0813.
Table 1. Baseline clinical data on 57 patients randomized to two different stents (mean ± SD) No. of patients Age (yr) Male/female Bilirubin (µmol/L) Diagnosis Pancreatic carcinoma Cholangiocarcinoma Ampullary tumor
Tannenbaum
Cotton-Huibregtse
29 72.0 ± 12 12/17 207 ± 89
28 76.3 ± 8 14/14 262 ± 102
20 8 1
18 9 1
All p values, calculated with t-test for continuous data and Pearson chi-square test to proportions, were not significant.
0.0813). Cox regression coefficients with 95% CI are shown in Table 4. The likelihood ratio (lr) was not statistically significant (observations n = 72; lr = 3.83; p = 0.15). Patient survival At the end of the study a total of 47 patients (81%) had either died (36 patients) or had symptoms of gastric outlet obstruction (11 patients: five in the TB group and six in the CH group). Four patients in each group had gone on to palliative gastroenterostomy. Three patients were considered unfit for bypass surgery. Median survival was 88 days (range 24 to 613 days) in the TB group, and 76.5 days (range 23 to 486) in the CH group (Fig. 2 and Table 3). No significant difference was found in patient survival (total number of patients = 53, TB = 27, CH = 26: log rank chi-square = 0.443, p = 0.4825). Cox regression coefficients with 95% confidence intervals are shown in Table 5. The likelihood ratio (lr) was statistically not significant (number of patients 53, lr = 2.35, p = 0.67). GASTROINTESTINAL ENDOSCOPY
25
V Terruzzi, U Comin, F De Grazia, et al.
Tannenbaum Teflon vs. standard polyethylene biliary stents
Table 2. Stent exchanges
No exchanges One exchange Two exchanges
Tannenbaum patients
Cotton-Huibregtse patients
21 2 3
20 3 4
Table 3. Stent patency and patient survival
Figure 2. Kaplan-Meier plots of survival of patients treated with a “Tannenbaum” stent (circles) or “Cotton-Huibregtse” stent (squares). Total number of patients = 53 (TB = 27, CH = 26). Statistical analysis: log-rank chi-square test = 0.493; p = 0.4825.
DISCUSSION Endoscopic drainage of malignant distal biliary stenosis is considered an effective alternative to palliative surgery, with significantly lower rates of early complications, hospitalization, and 1-month mortality.5 Clogging remains a problem with respect to endoscopically inserted plastic stents, and patency rates are often lower than the life expectancy of the patient.14 Metal stents have a considerably higher patency rate,10 but their use is limited by cost (ten times higher than plastic stents), the possibility of tumor ingrowth and because they cannot be removed or even repositioned.14 Recent studies have concentrated on overcoming the problem of clogging with plastic stents.6-9,11,12,14-17 Tested in vitro, Teflon was found to be significantly better than polyethylene and other polymers.6,15 However, Van Berkel et al.16 recently compared in vivo, in a prospective randomized study, Amsterdam-type biliary stents of these two different materials and showed no difference in patency rates. Stents without side holes have performed better both in vitro15 and in vivo; side holes not only increase the amount of biliary sludge in the stents, but also cancel out any difference between the various materials.6 However Sung et al.,17 in a recent prospective, randomized study, suggested that polyethylene stent patency was not improved by the absence of side holes. The underlying substrates of sludge deposition are a biofilm, bacteria, and the binding of biliary proteins, so that other strategies directed at preventing stent occlusion, such as the use of drugs to reduce clogging (antibiotics, aspirin, and ursodeoxycholic acid) have been tested but have been found to 26
GASTROINTESTINAL ENDOSCOPY
No. of patients Excluded Evaluated Median patient survival (range) Stent placement Median stent patency (range)
Tannenbaum
Cotton-Huibregtse
29 2 27
28 2 26
88 days (24-613) 34
76.5 days (23-323) 38
96 days (11-613)
75.5 days (23-323)
be without clinically significant benefit.14 The new Teflon prosthesis without side holes, known as the “Tannenbaum” stent, was recently compared with a pigtail Teflon stent with side holes in a prospective but non-randomized clinical trial; TB stents were associated with a remarkable increase (almost double) in patency rate of the first stent and a significantly lower number of stent exchanges.12 In a similar pilot study, a median patency rate of 64 weeks was observed with this new Teflon prosthesis without side holes; these results compare favorably with historic data concerning polyethylene stents with side holes and are similar to those for self-expanding metal stents.11 Our prospective, randomized, multicenter trial did not confirm the expectations of these earlier nonrandomized studies,11,12 and our data are supported by a recent preliminary report concerning a study of a large number of patients with results similar to ours.18 The median survival of patients with a TB stent was 88 days in our study and 117 in this study by Martin et al., while patients with polyethylene prostheses with side holes survived, respectively, 76.5 days and 151 days. The two studies cannot be compared with regard to stent patency rates because only the median patency of the first prosthesis is included in the preliminary report of Martin et al. In any case, the difference is never significant.18 The statistical analysis used in our study was not based on an intention-to-treat approach because one of the end-points was stent occlusion: in fact, the reason for exclusion of three of the four patients excluded was stent insertion failure. Moreover the low incidence of exclusions in both groups (two in TB and two in CH group) did not significantly influence the final analysis. The statistical power of our study is lower than VOLUME 51, NO. 1, 2000
Tannenbaum Teflon vs. standard polyethylene biliary stents
Table 4. Hazard ratios and 95% confidence intervals of variables considered in stent patency analysis Hazard ratios No. of previous change Type (TB)
95% Lower
95% Upper
0.177 0.353
1.114 1.480
0.444 0.723
Table 5. Hazard ratios and 95% confidence intervals of variables considered in survival analysis Hazard ratios Gender (male) Type (TB) Age Diagnosis (pancreatic carcinoma)
95% Lower
V Terruzzi, U Comin, F De Grazia, et al.
5.
6.
7.
95% Upper
0.998 0.741 0.981
0.509 0.370 0.950
1.955 1.485 1.014
0.692
0.340
1.407
8.
9.
10.
needed with respect to patient survival, but our results do not differ from previously published data.16-18 Even considering the crude numbers, which suggest a better patency rate of the TB prosthesis in both our study and that of Martin et al., survival is only a few days longer so that use of the TB prosthesis is not justified. In conclusion, the TB stent appears to offer no particular advantage over the standard polyethylene stent.
11.
12.
13. 14. 15.
ACKNOWLEDGMENT We thank J. D. Baggott for assistance with editing the manuscript. REFERENCES 1. Van Heerden JA. Pancreatic resection for carcinoma of the pancreas: Whipple versus total pancreatectomy. An institutional perspective. World J Surg 1984;8:880-8. 2. Sarr MG, Cameron JL. Surgical management of unresectable carcinoma of the pancreas. Surgery 1982;91:123-33. 3. Soehendra N, Reynders-Frederix V. Palliative gallengangdrainage. Dtsch Med Wochenschr 1979;104:206-9. 4. Ballinger AB, McHugh M, Catnach SM. Symptom relief and
VOLUME 51, NO. 1, 2000
16.
17.
18.
quality of life after stenting for malignant bile duct obstruction. Gut 1994;35:467-70. Smith AC, Dowsett JF, Russell RCG, Hatfield ARW, Cotton PB. Randomized trial of endoscopy stenting versus surgical bypass in malignant low bile duct obstruction. Lancet 1994; 344:1655-60. Coene PPLO, Groen AK, Cheng J, Out MMJ, Tytgat GNJ, Huibregtse K. Clogging of biliary endoprosthesis: a new perspective. Gut 1990;31:913-7. Speer AG, Cotton PB, McRae KD. Endoscopic management of malignant biliary obstruction: stent of l0 French gauge are preferable to stent of 8 French gauge. Gastrointest Endosc 1988;34:412-7. Ghosh S, Palmer KR. Prevention of biliary stent occlusion using cyclical antibiotics and ursodeoxycholic acid. Gut 1994; 35:1757-9. Smit JM, Out MMJ, Groen AK. A placebo controlled study on the efficacy of aspirin and doxycycline in preventing clogging of biliary endoprosthesis. Gastrointest Endosc 1985;35:485-9. Knyrim K, Wagner HJ, Pausch J, Vakil N. A prospective randomized controlled trial of metal stents for malignant biliary obstruction of the common bile duct. Endoscopy 1993;25:207-12. Binmoeller KF, Seitz U, Seifert H, Thonke F, Sikka S, Soehendra N. The Tannenbaum stent: a new plastic biliary stent without side holes. Am J Gastroenterol 1995;90:1764-8. Seitz U, Vadeyar H, Soehendra N. Prolonged patency with a new designed teflon biliary prosthesis. Endoscopy 1994;26: 478-82. Freedman LS. Tables of number of patients required in clinical trial, using the log-rank test. Stat Med 1982;1:121-9. Libby ED, Leung JW. Prevention of biliary stent clogging: a clinical review. Am J Gastroenterol 1996;91:1301-8. Rey JF, Maupetit P, Greff M. Experimental study of biliary endoprosthesis efficiency. Endoscopy 1985;17:14:5-8. Van Berkel AM, Boland C, Redekop WK, Bergman JJGHM, Groen AK, Tytgat GNJ, et al. A prospective randomized trial of Teflon versus Polyethylene stents for distal malignant biliary obstruction. Endoscopy 1998;30:681-6. Sung JJY, Chung SCS, CP Tsui, Co AL, Li AK. Omitting sideholes in biliary stents does not improve drainage of the obstructed biliary system: a prospective randomized trial. Gastrointest Endosc 1994;40:321-5. Martin DF, Frost R, Sheridan M, Freman A, Lawrie B, Deakin M. A prospective randomized trial comparing 10 Fr Tannenbaum stents with 10 Fr Cotton-Leung stents in patients with malignant common duct strictures [abstract]. Endoscopy 1997;29:E51.
GASTROINTESTINAL ENDOSCOPY
27
Background: Our aim was to compare the clinical efficacy of the Tannenbaum (TB) biliary prostheses, a recently designed Teflon stent without side holes, with the Cotton-Huibregtse (CH) polyethylene stent. Methods: Fifty-seven patients (26 men, mean age 75.5 years) with unresectable malignant tumors and distal biliary stenosis were included (38 pancreatic head cancer, 17 cholangiocarcinoma, 2 ampullary cancer). Patients were prospectively randomized to have a 10F, 7 cm long TB (29 patients) or CH (28 patients) stent inserted endoscopically. Four patients (2 TB and 2 CH) were excluded: 3 because of the failure of stent insertion and 1 because of a protocol violation. The patients were evaluated clinically and, if necessary, with biochemical tests every month until death or until they needed surgery for symptoms of gastric outlet obstruction. When occlusion or dislocation occurred, the stent was replaced with one of the same type. Cumulative stent patency and patient survival were estimated with the Kaplan-Meier life-table analysis and compared by log-rank chi-square test. Results: The two groups were comparable in mean age, gender, and diagnosis. The patients were followed for a mean of 145.5 days (range 24 to 613); by the end of the study 47 patients (81%) had died or developed symptoms of gastric outlet obstruction. Median survival was 88 days (range 24 to 613) in the TB group and 75.6 days (23 to 486) in the CH group. Stent exchange (occlusion 16, dislocation 3) was necessary for 5 patients in the TB group and 7 in the CH group. No statistical difference was found on comparing the mean duration of function of the first, second, and third stents (analysis of variance, p = 0.80). The median duration of stent function was 96 days (range 11 to 613) in the TB group and 75.5 days (range 23 to 323) in the CH group. No significant difference was found in either survival time (p = 0.48) or stent patency (p = 0.12). Conclusion: This study found no significant advantage of the Tannenbaum prostheses over the standard polyethylene stent in the palliation for patients with distal malignant biliary stenosis with regard to survival or length of stent patency. (Gastrointest Endosc 2000;51:23-7.)
Most patients with biliary obstruction due to cancer of the head of the pancreas and distal cholangiocarcinoma have advanced disease at diagnosis, and only a few are candidates for curative surgery.1 Biliary enteric bypass is unquestionably effective for palliation in patients with biliary or duodenal Received December 23, 1997. For revision April 9, 1998. Accepted September 3, 1999. From Unità di Gastroenterologia, Divisione di Medicina 2, and Unità di Biostatistica, Ospedale Valduce, Como; Servizio di Endoscopia, Ospedale Civile di Magenta, Milano; Servizio di Endoscopia, Ospedale Maggiore, Crema; and Ospedale Civile di Sesto S. Giovanni, Milano, Italy. Presented in part at the annual meeting of the American Society for Gastrointestinal Endoscopy, May 11-14, 1997, Washington, D.C. (Gastrointest Endosc 1997;45:507). Supported in part by Cook Italia. Reprint requests: Vittorio Terruzzi, MD, Divisione di Medicina 2, Ospedale Valduce, via Dante 11, I - 22100 Como, Italy. Copyright © 2000 by the American Society for Gastrointestinal Endoscopy 0016-5107/2000/$12.00 + 0 37/1/102787 VOLUME 51, NO. 1, 2000
obstruction due to advanced pancreatic cancer, but it is associated with high mortality and morbidity rates.2 Since 1979, endoscopic placement of plastic stents has offered a valid alternative to surgery.3 Stent insertion achieves comparable therapeutic results and improves quality of life.4 No difference in patient survival has been found between endoscopic and surgical biliary drainage.5 A persistent problem of endoscopically inserted biliary stents is late occlusion (within 3 to 6 months). Despite efforts to prolong patency with drugs or by changing stent diameter, design, and material, clogging by sludge remains problematic.6-9 Expandable metal stents have a significantly higher patency rate, but high cost precludes extensive use, especially in patients whose life expectancy is shorter than the median patency of the stent.10 In a pilot study, the patency rate of a newly designed Teflon stent without side holes (Tannenbaum stent; Wilson-Cook Medical Inc., Winston-Salem, N.C.) was longer than GASTROINTESTINAL ENDOSCOPY
23
V Terruzzi, U Comin, F De Grazia, et al.
that reported for polyethylene stents with side holes and similar to that reported for metallic expandable biliary prostheses.11 In a non-randomized, prospective study, this “Tannenbaum” stent remained patent significantly longer than a Teflon pigtail stent.12 The aim of this multicenter, prospective, randomized trial, was to compare the clinical efficacy of the “Tannenbaum” Teflon prosthesis (TB), with conventional “Cotton-Huibregtse” polyethylene stents (CH) in patients with previously untreated malignant distal common bile duct obstruction. PATIENTS AND METHODS Patients Patients were included if they had obstructive jaundice due to untreated malignancy involving the distal bile duct. Diagnosis was based on presenting symptoms, biochemical tests, abdominal ultrasound, and ERCP. CT was done as necessary when evaluating patients for surgery. All patients were initially considered for surgery and were only rejected because of disease extension or severe disability (old age and/or other comorbid conditions). Patients with duodenal obstruction or hepatic hilar involvement and those with other previous malignant diseases or a previous drainage procedure or gastric or bile duct surgery were excluded. Patients in poor general condition with a short life expectancy were also excluded. No other therapeutic procedures, such as percutaneous drainage, radiotherapy or chemotherapy, were used for biliary obstruction during the study. Informed consent was obtained from all patients before randomization, and the study was approved by the Internal Review Board of Valduce Hospital. Prostheses The “Tannenbaum” stent is a straight biliary prosthesis made from Teflon tubing, with no side holes and with multiple anchoring flaps that do not perforate the wall.11 This prosthesis was compared with the conventional “Cotton-Huibregtse” polyethylene stent with side holes. Both types were obtained from the same manufacturer (Wilson-Cook). Study design and methods In this multicenter (four digestive-endoscopic centers), prospective, single-blind study (patients were not informed as to the type of prosthesis), patients were consecutively randomized to have either a 10F diameter, 7 cm long TB stent, or a CH stent of the same size inserted endoscopically. Thirty minutes before endoscopy, all patients received cefazoline 1 gm intravenously. Randomization was done by means of a computed schedule blindly generated for each center participating in the study. After diagnostic ERCP the stents were inserted with a three-layer technique (guidewire, guide-catheter, pusher) with or without previous endoscopic sphincterotomy. ERCP complications and those related to stent insertion were assessed within 7 days of the procedure. End points of the study were stent occlusion, death, or the need for palliative surgery for duodenal obstruction due to tumor growth. 24
GASTROINTESTINAL ENDOSCOPY
Tannenbaum Teflon vs. standard polyethylene biliary stents
Follow-up data (clinical and when useful, biochemical data) were obtained at 1-month intervals by telephone from the referring physician, until death, or until surgical treatment was needed for gastric outlet obstruction. When occlusion or dislocation occurred, the stent was removed with a polypectomy snare or a Dormia basket and the same type of stent was used for replacement. Occlusion or dislocation was considered when patients developed cholangitis, jaundice, or flu-like syndrome associated with cholestasis (defined as an episode of biliary pain or abdominal discomfort, with jaundice and/or fever over 38° C, for more than 1 day and/or increase in liver function tests). Stent occlusion was assessed macroscopically after removal. Statistical analysis Stent patency. The number of stent placements required to demonstrate statistical significance with a 5% alpha error and a 90% power was 67 and was calculated on the basis of previous data,11,12 assuming a 40% difference in patency rate at 1-year between the two groups.13 With a lower (80%) power, estimated sample size was 51 placements assuming a 40% difference between the two groups, 65 for a 35% difference and 88 for a 30% difference. The Pearson chi-square test with Yates’ correction and the Fisher’s exact test for categorical comparison of data were used as appropriate. T test was used for continuous data comparison. Analysis of variance (ANOVA) test was applied to compare the mean patency of the first stent with the mean patency of the second and third ones when more than one stent replacement was required in the same patient. Stent patency was calculated in days, as the interval between the time of its replacement. Patient death or the appearance of gastric outlet obstruction due to duodenal stenosis were considered censoring events. Cumulative stent patency was estimated by KaplanMeier life-table analysis and the two groups (TB and CH) were compared by the log rank chi-square test. Cox regression was used to study the effect of other covariates on the time to stent occlusion. Number of previous prosthesis exchanges in the same patient and type (TB vs. CH) were considered potentially explanatory covariates in the stent patency survival function. A step-wise (step-down) regression strategy was used for variable selection. Log cumulative hazard plots were examined to test the proportionality of hazard assumption. Martingale residuals and deviance residuals plot were examined to test the goodness of fit of the model. Regression coefficients and 95% confidence intervals were calculated. A significance test was made using likelihood ratio test. Patient survival. The number of patients required with respect to survival was calculated so that testing would be done at 90% power and 5% level of significance. Assuming, on the basis of previous data,12 a 25% difference in 1-year patient survival between the two groups, estimated sample size was 112. The estimated sample size reduces to 85 patients assuming a 25% difference and 80% VOLUME 51, NO. 1, 2000
Tannenbaum Teflon vs. standard polyethylene biliary stents
V Terruzzi, U Comin, F De Grazia, et al.
power and increases to 137 with the same power and assuming a 20% difference between the two groups.13 Patient death was considered the end-point. Patient survival was calculated in days from the time of stent insertion. Survival time was censored at the last visit for patients lost during follow-up or submitted to palliative surgery for duodenal stenosis. Age, gender (male vs. female), diagnosis (pancreatic carcinoma vs. cholangiocarcinoma or ampullary tumor), and type of stent (TB vs. CH) were examined as potentially explanatory variables. As for stent patency, cumulative patient survival (days from the first stent insertion) and Cox regression were examined. All analyses were made using StatView 4.5 (Abacus Concepts, Inc., Berkeley, Calif.).
RESULTS Fifty-seven patients were included: 38 had pancreatic head cancer, 17 cholangiocarcinoma, and 2 ampullary tumors. There were 25 men and 32 women. Mean age was 75.5 years (range 41 to 90 years). After randomization 29 patients were treated with the TB stent, and 28 with the CH stent. Two TB and two CH patients were excluded, three because of unsuccessful stent insertion and one for a protocol violation (after the first insertion, the prostheses was replaced with a self-expandable metal stent). No major procedure-related complications, such as pancreatitis, cholangitis, perforation, or significant bleeding arose in either group. The two groups were comparable with regard to age, gender, initial serum bilirubin, and diagnosis (Table 1). The patients were followed for a mean of 145.5 days (range 24 to 613 days) after insertion of the first stent. Stent patency Stent occlusion occurred in 16 patients (7 TB and 9 CH). Stent migration, always into the duodenum, occurred in 2 TB patients and 1 CH. The number of stent exchanges was similar in the two groups (Table 2). When a stent exchange was made after occlusion, no significant difference was found between the mean duration of patency of the first stent (103.2 days), the second (120.8 days), and the third (123.4 days) inserted (p = 0.78). We therefore decided we could justifiably group all of the stents regardless of their placement order and assess the differences between TB and CH groups within the overall case list of 71 stents, thus achieving the necessary statistical power. Median duration of stent patency was 96 days (range 11 to 613) in the TB group and 75.5 days (23 to 323) in the CH group (Fig. 1 and Table 3). No differences were found in the two groups (stents = 72; TB = 34; CH = 38: log-rank chi-square = 3.039; p = VOLUME 51, NO. 1, 2000
Figure 1. Kaplan-Meier plots of patency rates for the “Tannenbaum” (circles) and “Cotton-Huibregtse” stents (squares). Total number of stents: 72 (TB = 34, CH = 38). Statistical analysis: log-rank chi-square test = 3.034; p = 0.0813.
Table 1. Baseline clinical data on 57 patients randomized to two different stents (mean ± SD) No. of patients Age (yr) Male/female Bilirubin (µmol/L) Diagnosis Pancreatic carcinoma Cholangiocarcinoma Ampullary tumor
Tannenbaum
Cotton-Huibregtse
29 72.0 ± 12 12/17 207 ± 89
28 76.3 ± 8 14/14 262 ± 102
20 8 1
18 9 1
All p values, calculated with t-test for continuous data and Pearson chi-square test to proportions, were not significant.
0.0813). Cox regression coefficients with 95% CI are shown in Table 4. The likelihood ratio (lr) was not statistically significant (observations n = 72; lr = 3.83; p = 0.15). Patient survival At the end of the study a total of 47 patients (81%) had either died (36 patients) or had symptoms of gastric outlet obstruction (11 patients: five in the TB group and six in the CH group). Four patients in each group had gone on to palliative gastroenterostomy. Three patients were considered unfit for bypass surgery. Median survival was 88 days (range 24 to 613 days) in the TB group, and 76.5 days (range 23 to 486) in the CH group (Fig. 2 and Table 3). No significant difference was found in patient survival (total number of patients = 53, TB = 27, CH = 26: log rank chi-square = 0.443, p = 0.4825). Cox regression coefficients with 95% confidence intervals are shown in Table 5. The likelihood ratio (lr) was statistically not significant (number of patients 53, lr = 2.35, p = 0.67). GASTROINTESTINAL ENDOSCOPY
25
V Terruzzi, U Comin, F De Grazia, et al.
Tannenbaum Teflon vs. standard polyethylene biliary stents
Table 2. Stent exchanges
No exchanges One exchange Two exchanges
Tannenbaum patients
Cotton-Huibregtse patients
21 2 3
20 3 4
Table 3. Stent patency and patient survival
Figure 2. Kaplan-Meier plots of survival of patients treated with a “Tannenbaum” stent (circles) or “Cotton-Huibregtse” stent (squares). Total number of patients = 53 (TB = 27, CH = 26). Statistical analysis: log-rank chi-square test = 0.493; p = 0.4825.
DISCUSSION Endoscopic drainage of malignant distal biliary stenosis is considered an effective alternative to palliative surgery, with significantly lower rates of early complications, hospitalization, and 1-month mortality.5 Clogging remains a problem with respect to endoscopically inserted plastic stents, and patency rates are often lower than the life expectancy of the patient.14 Metal stents have a considerably higher patency rate,10 but their use is limited by cost (ten times higher than plastic stents), the possibility of tumor ingrowth and because they cannot be removed or even repositioned.14 Recent studies have concentrated on overcoming the problem of clogging with plastic stents.6-9,11,12,14-17 Tested in vitro, Teflon was found to be significantly better than polyethylene and other polymers.6,15 However, Van Berkel et al.16 recently compared in vivo, in a prospective randomized study, Amsterdam-type biliary stents of these two different materials and showed no difference in patency rates. Stents without side holes have performed better both in vitro15 and in vivo; side holes not only increase the amount of biliary sludge in the stents, but also cancel out any difference between the various materials.6 However Sung et al.,17 in a recent prospective, randomized study, suggested that polyethylene stent patency was not improved by the absence of side holes. The underlying substrates of sludge deposition are a biofilm, bacteria, and the binding of biliary proteins, so that other strategies directed at preventing stent occlusion, such as the use of drugs to reduce clogging (antibiotics, aspirin, and ursodeoxycholic acid) have been tested but have been found to 26
GASTROINTESTINAL ENDOSCOPY
No. of patients Excluded Evaluated Median patient survival (range) Stent placement Median stent patency (range)
Tannenbaum
Cotton-Huibregtse
29 2 27
28 2 26
88 days (24-613) 34
76.5 days (23-323) 38
96 days (11-613)
75.5 days (23-323)
be without clinically significant benefit.14 The new Teflon prosthesis without side holes, known as the “Tannenbaum” stent, was recently compared with a pigtail Teflon stent with side holes in a prospective but non-randomized clinical trial; TB stents were associated with a remarkable increase (almost double) in patency rate of the first stent and a significantly lower number of stent exchanges.12 In a similar pilot study, a median patency rate of 64 weeks was observed with this new Teflon prosthesis without side holes; these results compare favorably with historic data concerning polyethylene stents with side holes and are similar to those for self-expanding metal stents.11 Our prospective, randomized, multicenter trial did not confirm the expectations of these earlier nonrandomized studies,11,12 and our data are supported by a recent preliminary report concerning a study of a large number of patients with results similar to ours.18 The median survival of patients with a TB stent was 88 days in our study and 117 in this study by Martin et al., while patients with polyethylene prostheses with side holes survived, respectively, 76.5 days and 151 days. The two studies cannot be compared with regard to stent patency rates because only the median patency of the first prosthesis is included in the preliminary report of Martin et al. In any case, the difference is never significant.18 The statistical analysis used in our study was not based on an intention-to-treat approach because one of the end-points was stent occlusion: in fact, the reason for exclusion of three of the four patients excluded was stent insertion failure. Moreover the low incidence of exclusions in both groups (two in TB and two in CH group) did not significantly influence the final analysis. The statistical power of our study is lower than VOLUME 51, NO. 1, 2000
Tannenbaum Teflon vs. standard polyethylene biliary stents
Table 4. Hazard ratios and 95% confidence intervals of variables considered in stent patency analysis Hazard ratios No. of previous change Type (TB)
95% Lower
95% Upper
0.177 0.353
1.114 1.480
0.444 0.723
Table 5. Hazard ratios and 95% confidence intervals of variables considered in survival analysis Hazard ratios Gender (male) Type (TB) Age Diagnosis (pancreatic carcinoma)
95% Lower
V Terruzzi, U Comin, F De Grazia, et al.
5.
6.
7.
95% Upper
0.998 0.741 0.981
0.509 0.370 0.950
1.955 1.485 1.014
0.692
0.340
1.407
8.
9.
10.
needed with respect to patient survival, but our results do not differ from previously published data.16-18 Even considering the crude numbers, which suggest a better patency rate of the TB prosthesis in both our study and that of Martin et al., survival is only a few days longer so that use of the TB prosthesis is not justified. In conclusion, the TB stent appears to offer no particular advantage over the standard polyethylene stent.
11.
12.
13. 14. 15.
ACKNOWLEDGMENT We thank J. D. Baggott for assistance with editing the manuscript. REFERENCES 1. Van Heerden JA. Pancreatic resection for carcinoma of the pancreas: Whipple versus total pancreatectomy. An institutional perspective. World J Surg 1984;8:880-8. 2. Sarr MG, Cameron JL. Surgical management of unresectable carcinoma of the pancreas. Surgery 1982;91:123-33. 3. Soehendra N, Reynders-Frederix V. Palliative gallengangdrainage. Dtsch Med Wochenschr 1979;104:206-9. 4. Ballinger AB, McHugh M, Catnach SM. Symptom relief and
VOLUME 51, NO. 1, 2000
16.
17.
18.
quality of life after stenting for malignant bile duct obstruction. Gut 1994;35:467-70. Smith AC, Dowsett JF, Russell RCG, Hatfield ARW, Cotton PB. Randomized trial of endoscopy stenting versus surgical bypass in malignant low bile duct obstruction. Lancet 1994; 344:1655-60. Coene PPLO, Groen AK, Cheng J, Out MMJ, Tytgat GNJ, Huibregtse K. Clogging of biliary endoprosthesis: a new perspective. Gut 1990;31:913-7. Speer AG, Cotton PB, McRae KD. Endoscopic management of malignant biliary obstruction: stent of l0 French gauge are preferable to stent of 8 French gauge. Gastrointest Endosc 1988;34:412-7. Ghosh S, Palmer KR. Prevention of biliary stent occlusion using cyclical antibiotics and ursodeoxycholic acid. Gut 1994; 35:1757-9. Smit JM, Out MMJ, Groen AK. A placebo controlled study on the efficacy of aspirin and doxycycline in preventing clogging of biliary endoprosthesis. Gastrointest Endosc 1985;35:485-9. Knyrim K, Wagner HJ, Pausch J, Vakil N. A prospective randomized controlled trial of metal stents for malignant biliary obstruction of the common bile duct. Endoscopy 1993;25:207-12. Binmoeller KF, Seitz U, Seifert H, Thonke F, Sikka S, Soehendra N. The Tannenbaum stent: a new plastic biliary stent without side holes. Am J Gastroenterol 1995;90:1764-8. Seitz U, Vadeyar H, Soehendra N. Prolonged patency with a new designed teflon biliary prosthesis. Endoscopy 1994;26: 478-82. Freedman LS. Tables of number of patients required in clinical trial, using the log-rank test. Stat Med 1982;1:121-9. Libby ED, Leung JW. Prevention of biliary stent clogging: a clinical review. Am J Gastroenterol 1996;91:1301-8. Rey JF, Maupetit P, Greff M. Experimental study of biliary endoprosthesis efficiency. Endoscopy 1985;17:14:5-8. Van Berkel AM, Boland C, Redekop WK, Bergman JJGHM, Groen AK, Tytgat GNJ, et al. A prospective randomized trial of Teflon versus Polyethylene stents for distal malignant biliary obstruction. Endoscopy 1998;30:681-6. Sung JJY, Chung SCS, CP Tsui, Co AL, Li AK. Omitting sideholes in biliary stents does not improve drainage of the obstructed biliary system: a prospective randomized trial. Gastrointest Endosc 1994;40:321-5. Martin DF, Frost R, Sheridan M, Freman A, Lawrie B, Deakin M. A prospective randomized trial comparing 10 Fr Tannenbaum stents with 10 Fr Cotton-Leung stents in patients with malignant common duct strictures [abstract]. Endoscopy 1997;29:E51.
GASTROINTESTINAL ENDOSCOPY
27