ExpandableMetal Biliary Stenting inPatientsWithRecurrentPrematurePolyethylene Stent OcclusionKrishna Menon, M.D., F.R.C.P.(C), Joseph Romagnuolo, M.D., F.R.C.P.(C), andAlan N. Barkun, M.D., C.M., F.R.C.P.(C), F.A.C.P., M.Sc.(epid)Division of Gastroenterology, McGill University and the McGill University Health Centre,Montreal, Quebec, Canada

OBJECTIVES: Expandable metal stents are currently used totreat biliary tract obstruction. Few data exist on the role ofmetal stents in patients with recurrent premature plasticbiliary stent occlusion. We report our preliminary resultsusing an expandable metal biliary stent with enhanced stentflexibility in this group of patients. Our aim was to assess theefficacy of the Diamond biliary stent in the treatment ofrecurrent premature biliary plastic stent blockage.

METHODS: From September 1997 to June 1998, six patientswith inoperable biliary obstruction and at least one priorepisode of plastic stent occlusion were treated with anexpandable metal biliary prosthesis. There were four womenand six men, with a mean age of 736 11 yr. Five patientshad malignant biliary obstruction, and one inoperable pa-tient had a biliary stricture secondary to chronic pancreatitis.Patients were followed-up prospectively until either stentocclusion or death. The patency duration of the metal stentswas compared to that of the most recently placed plasticstents.

RESULTS: All stents were successfully inserted endoscopi-cally with visualized biliary drainage. No significant imme-diate postprocedural complications were noted. Mediantime from initial diagnosis of biliary obstruction to metalstent insertion was 35 wk (range 7–142), during which timepatients had occluded a median of 4.5 plastic stents (range2–5). Median patient survival was 117 days (15–312) andmedian time to blockage of the last plastic stent was 25.5days (range 10–90 days). Three cases of metal stent occlu-sion occurred. One patient had early stent occlusion at 9days because of debris and sludge clogging the metal stent.Two other patients developed stent occlusion at 120 and 157days. Two stents remained patent until the patients’ deathsat 15 and 87 days. Overall median stent patency was 139days. The three patients who developed stent occlusion weretreated with successful insertion of one or more plasticstents through the existing metal stent. Duration of metalstent patency was significantly longer than that of the lastplastic stent (58.8 days longer; 95% CI [6.4, 111];p 5 0.03).

CONCLUSIONS: Metal biliary prostheses represent an effec-tive management strategy for recurrent plastic biliary stentobstruction. Patients in this subgroup may have a shorter

duration of metal stent patency than the reported duration ofstent patency in patients receiving initial metal stent place-ment. However the duration of patency still seems to besignificantly longer than that of the most recently placedplastic stent. (Am J Gastroenterol 2001;96:1435–1440.© 2001 by Am. Coll. of Gastroenterology)

INTRODUCTION

Endoscopic biliary stenting is well accepted as the treatmentof choice in patients with inoperable tumors causing biliaryobstruction, and has been associated with reduced morbidityand a shorter hospital stay compared with surgical earlydrainage (1–7). The major problem faced after endoscopicstent insertion is stent occlusion resulting in recurrent jaun-dice or cholangitis, complications seen less frequently inpatients palliated surgically (1–3). Traditionally, polyethyl-ene stents have been used for biliary decompression, butthese tend to occlude within 3–4 months because of forma-tion of a bacterial biofilm and sludge encrustation (8, 9). Thestent patency can be improved by increasing the internaldiameter of the stent, but is limited by the size of theoperating channel of duodenoscopes (10). This problem wasovercome with the development of metal stents that expandfrom a compressed delivery form to a standard maximaldiameter of 10 mm or 30F. This increase in internal diam-eter results in a significantly longer duration of patency, inthe range of 8–9 months, compared to polyethylene stents(11–13). By design, metal stents have a reduced surface areafor bacterial adherence. As a result, the mechanisms of stentocclusion are fundamentally different and include tumoringrowth and overgrowth, and intimal hyperplasia ratherthan sludge formation (8, 11). Covered metal stent proto-types developed to overcome this problem have unfortu-nately thus far met with poor success in further prolongingstent patency (14). Metal stents may improve quality of life,but apart from one retrospective study (8) showing an in-crease in median survival from 4 months with polyethylenestents to 6.5 months with metal stents, no clear survivalbenefit attributable to the use of metal stents has beenshown.

The major drawbacks to metal stent insertion are their

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 96, No. 5, 2001© 2001 by Am. Coll. of Gastroenterology ISSN 0002-9270/01/$20.00Published by Elsevier Science Inc. PII S0002-9270(01)02358-9

inability to be removed in the event of occlusion and theirup-front cost. Newer prototypes (spiral or coiled stents) canbe removed endoscopically but display a shorter duration ofpatency than metal mesh stents (15). The significant costs ofmetal stents (US $1000–1200) have limited their wide-spread use in the Canadian health care setting, and thus theyare only considered selectively in patients who have initiallyhad insertion of a polyethylene stent, as proposed by someinvestigators (16). This subgroup of patients is therefore ofparticular interest; however, the performance of metal stentsin patients with recurrent plastic stent occlusion has not beenwell studied (8). Because replacement polyethylene stentsdisplay an even shorter duration of patency than the originalstent, there is reason to believe that the performance ofmetal stents in these patients may not be the same as thatseen in patients in whom a metal stent was inserted as initialtherapy (13, 17). We therefore sought to identify whethermetal stent insertion would provide better palliation byincreasing the duration of stent patency in this particularsubgroup of patients at high risk for reocclusion.

The commercially available Diamond stent (Ultraflex Di-amond Biliary Stent, Microvasive, Boston Scientific Cor-poration, Natick, MA) is an uncovered metal mesh thatexpands to a final diameter of 10 mm after insertion and isavailable in three lengths (4, 6, and 8 cm). It is made ofnitinol and is inserted in compressed form through a con-trolled release delivery system. This report details our pre-liminary experience with the Diamond biliary stent in asubgroup of patients with recurrent plastic biliary stentocclusion, and compares this to the patency duration of thepatients’ most recently placed plastic stent.

MATERIALS AND METHODS

PatientsFrom September 1997 to June 1998, six patients weretreated for biliary obstruction with a Diamond biliary pros-thesis at our institution. Patient data were collected prospec-tively and, in part, gathered retrospectively by reviewingpatient charts and procedure records. Patients were con-tacted to obtain current clinical status and to document anysymptoms of stent obstruction.

Endoscopic TechniqueERCPs were performed using conscious sedation with me-peridine and midazolami.v. as required. A therapeutic chan-nel duodenoscope was used in all cases. Patients weretreated with prophylactic antibiotics intravenously beforethe procedure. After opacification of the common bile duct,a 0.035-inch guidewire was inserted. All metal stents wereinserted over a guidewire in compressed form using theenclosed 10F delivery catheter system. Under fluoroscopiccontrol, each stent was released at the site of stricture.Placement was considered successful if the stent was radio-logically visualized to be positioned across the stricture andfree flow of bile was noted into the duodenum. Any com-

plications occurring within 7 days post-ERCP were re-corded and considered as immediate procedure-related com-plications.

OutcomesThe primary outcome of interest was stent dysfunction orblockage. Secondary outcomes included death and proce-dural complications. Stent patency represented the intervalbetween time of stent insertion and stent dysfunction orpatient death. Stent dysfunction was suspected in any patientwith recurrent jaundice or cholangitis and confirmed byERCP. Early stent dysfunction was defined as the presenceof stent dysfunction before the clinical resolution of jaun-dice. Post-ERCP complications were graded as mild, mod-erate, or severe according to the previously described crite-ria (18).

Statistical AnalysisDescriptive statistics are reported as proportions, and asmean6 SD or median and range. A comparison of theduration of patency between the last plastic stent placed andthe metal stent was made using the pairedt test, and 95%confidence intervals were constructed for the difference induration of stent patency. Survival analysis for patency ofthe metal stents and the most recent plastic stent was per-formed using standard Kaplan-Meier methodology. In keep-ing with this methodology, patients who died or who com-pleted follow-up before reaching the primary outcome (stentocclusion) were considered “censored” at the time of theirdeath for the purpose of survival analysis and calculation ofmedian stent survival. The two curves were compared withthe log-rank test.

There are a few limitations of the statistical analysis interms of the comparision of patency duration between themetal stent and last plastic stent. The two comparisongroups are not independent and are, in fact, paired. Thus, wechose to compare the differences in two ways. First, stentpatency durations were compared using a pairedt test andcalculation of a 95% confidence interval for the observeddifference, with the assumption that the paired differenceswould follow a roughly normal distribution (in contrast tothe original patency duration distributions, which may beskewed or even bimodal). This type of paired analysisignores censoring, and assumes that the last follow-up of themetal stents is equivalent to the time at which they becameblocked. In the case of three patients whose follow-up endedbefore stent occlusion occurred, this biases the differencetoward the null and conservatively biases against metalstents for this comparison. The order of the cross-over,which is in one direction only (i.e., plastic to metal) alsolikely biases against metal stents. Secondly, the nonpara-metric log-rank test was used to compare the Kaplan-Meiercurves. This method inherently accounts for censoring; but,strictly speaking, this type of survival analysis is classicallyperformed on two independent samples.

1436 Menon et al. AJG – Vol. 96, No. 5, 2001

RESULTS

As shown in Table 1, indications for stent insertion werepancreatic carcinoma in two patients, and one case each ofmetastatic adenocarcinoma with nodal compression of thedistal common bile duct, islet cell carcinoma, ampullarycarcinoma, and benign biliary stricture secondary to chronicpancreatitis. No patients were considered to be surgicalcandidates, either because of advanced disease or significantcomorbidity. The mean age of the patients was 736 10.5 yr;there were two men and four women. Each of the sixpatients had plastic stents that were clogged at enrollmentand removed at the time of Diamond stent insertion. Themedian time from diagnosis of biliary obstruction to metalstent insertion was 35 wk (range 7–142 wk); during thistime, patients occluded a median of 4.5 plastic stents (range2–5 stents). The median time from the placement of the lastplastic stent until blockage of that stent was 25.5 days (range10–90 days). These characteristics are detailed in Table 1.All metal stents were successfully inserted endoscopicallywith adequate biliary drainage visualized. No patient expe-rienced a significant immediate complication. In all cases,the location of the stricture allowed positioning of the stentacross the ampulla with two to three rows of the diamondmesh protruding into the duodenum. All patients had normalcoagulation parameters. One patient with portal gastropathyand suspected portal hypertension had some bleedingaround the ampulla after stent insertion. This resolved spon-taneously without the need for endoscopic therapy or bloodtransfusion.

Median patient survival was 117 days (range 15–312days). Median duration of stent patency was 139 days (range9–214 days) (Fig. 1). One patient had early stent dysfunc-tion 9 days after insertion; at repeat ERCP, proteinaceousmaterial was seen to be clogging the stent. Exclusion of thispatient resulted in a median stent patency of 157 days (range120–214 days). Two other patients had stent occlusion.Patient 5, who had a a benign biliary stricture, occluded at157 days because of sludge and debris clogging the stent.

Patient 6 occluded at 120 days; at ERCP, tumor ingrowth atthe region of the ampulla was visualized. All three patientsin whom metal stent occlusion occurred were treated withthe insertion of a plastic stent through the existing metalstent. Two patients expired with no evidence of stent dys-function at 15 and 87 days. One stent is currently still patentwith a follow-up 214 days, and the patient has remainedclinically stable. Overall, the metal stents lasted signifi-cantly longer (58.8 days longer; 95% CI [6.4,111];p 50.03) than the most recently placed plastic stents. The sur-vival curves (Fig. 1) of the metal stents and the last plasticstent in place were significantly different (log-rank test,p 50.02).

DISCUSSION

Although metal stents have a significantly longer duration ofpatency compared to plastic stents, the up-front cost asso-ciated with metal stents have limited their widespread use inthe Canadian health care setting. Thus, we have examinedtheir selective role in patients who achieve poor results withpolyethylene stents.

In our series, all metal stents were successfully insertedendoscopically, with good resultant endoscopic drainage.Although direct visualization of the lower part of the stentfor the distal biliary strictures in this study resulted insuccessful placement in all cases, poor radiographic visibil-ity of the stent may limit its accuracy of placement inproximal biliary tumors. No significant complications werenoted immediately postinsertion. Overall, in this prelimi-nary report, the median duration of stent patency of 139days is shorter than that reported in the literature for metalstents. In a case series by Yooet al. (19), the Diamond stentseemed to be as effective as the Gianturco biliary Z-stent. Incontrast, Siqueiraet al. (20) reported, in abstract form, ashorter median duration of patency of 56 days for theDiamond stent compared to 247 days for the Wallstent,although the reasons for the shortened duration of patency

Table 1. Patient Outcome Using the Diamond Biliary Stent

PatientNo.

Age (yr),Sex Diagnosis

Time SinceDiagnosis

(wk)

No. ofPrevious

Plastic Stents

Patency of LastPlastic Stent

(days)

DiamondStent Patency

(days)

Survival AfterStent Insertion

(days) Outcome

1 79, F Islet cellcarcinoma

142 .5 90 214 Alive Stent patent andpatient well

2 79, F Pancreaticcarcinoma

13 2 13 15 15 Stent patent untilpatient’s death

3 73, F Metastaticcarcinoma

7 2 10 9 142 Early occlusion

4 63, M Benignstricture

132 .5 85 157 Alive Occlusion

5 84, F Ampullarycarcinoma

46 5 26 120 Alive Occlusion due totumor ingrowth

6 57, M Pancreaticcarcinoma

24 4 25 87 87 Stent patent untilpatient’s death

F 5 female, M5 male.

1437AJG – May, 2001 Metal Biliary Stenting and Plastic Stent Occlusion

were unclear. A second study comparing these two stenttypes in 46 patients showed no difference in stent patency,but there was a significantly lower need for subsequentstents in the Wallstent group (21).

All patients in our study had previously undergone plasticstent insertion and had experienced subsequent clogging.This has been suggested to shorten the patency of subse-quently inserted plastic stents (13, 22). Our study suggeststhat this may also shorten the patency of subsequentlyinserted metal stents. In choosing this subgroup of patients,we are selecting out patients with advanced disease that mayinclude factors with a detrimental effect on duration of stentpatency (such as accumulated sludge and debris in the bileducts or a tendency to sludge formation) (22–24). Schmass-mann et al. (8) showed, in a retrospective study of 165patients, that the median stent patency of second metalstents was slightly less than that of first metal stents (10vs8 months), but that the superiority of metal over plasticstents persisted: second metal stents had a median patencyof 8 monthsversus3 months with second plastic stents. Inour study, all patients had occluded at least two stentspreviously, and so the metal stent was at least their thirdstent. The incidence of early stent dysfunction is unclear,although it has been reported in several other studies (8, 15,25). Two of the three patients who suffered metal stentocclusion in our study had evidence of sludge and debris inthe biliary system. It is likely that insertion of a metal stentunder these conditions may lead to an increased incidence ofearly stent occlusion because of the persistence of thesefactors. It seems, however, that despite the lower duration ofmetal stent patency in our study, the time to dysfunction is

still significantly longer compared with that of the lastplastic stent placed.

There are a few potential limitations of our analysiscomparing the duration of patency of the metal stent to thelast plastic stent placed (as discussed in Materials and Meth-ods). The results of both analyses agree that there is asignificant difference in stent longevity between the lastplastic stent and the subsequent metal stent, but most likelythe difference has been underestimated by our methods.

The goal of palliation is to provide patient comfort and toimprove quality of life. Pratet al. (26) showed that metalstent insertion was comparable to plastic stent insertion withelective exchange and that both of these strategies weresuperior to plastic stent exchange performed only in theevent of stent dysfunction with regard to complication-freesurvival. Schmassmannet al.(8) hypothesized that the smallincrease in survival seen in their study could be attributed tothe fact that longer symptomatic relief because of prolongedmetal stent patency rendered later procedures more accept-able, and thus improved compliance and avoided death fromuntreated stent dysfunction.

Several studies have attempted to address the cost-effec-tive use of metal stents for malignant biliary obstruction. Astudy from Switzerland showed the cost-effectiveness ratioin their retrospective study was $2105 per ERCP saved (8).However, this study only addressed the cost of the stentingitself and not differences in cost of hospitalization for suchfactors as stent dysfunction and complications, and it mea-sured effectiveness only in terms of procedures saved andnot in terms of quality-adjusted survival. The use of metalbiliary stents for malignant biliary obstruction would be

Figure 1. Cumulative patency of the Diamond biliary stent (solid line) in our series of six patients and the corresponding patency of themost recently placed plastic stent (broken line) using life table analysis.

1438 Menon et al. AJG – Vol. 96, No. 5, 2001

most cost-effective in those patients with a predicted longersurvival (.6–9 months) (8, 26). Yeohet al. (27) used aMarkov model to show that initial placement of a plasticstent, followed by metal stent placement at first occlusion,was an economical option. Also, this study showed that ifmetal stents were to cost less than half of the cost of anERCP, or if one were able to accurately predict those pa-tients who were going to survive.4 months, then initialmetal stent insertion would be preferred. In this model,however, the occlusion rate of metal stents placed afterplastic stent occlusion was assumed to be the same as that ofan initial metal stent. Our study shows that this assumptionmay not be a valid one. Also, the option of allowing two ormore trials with a plastic stent, before resorting to a metalstent, was not modeled.

As mentioned above, the ability to predict longer survi-vors may be critical to developing a cost-effective approachto malignant biliary obstruction. Unfortunately, adequatemodels to predict patient survival are lacking. Pratet al.(16)showed that tumor diameter assessed by ultrasound, CT, orERCP was the best discriminator of patient survival in agroup of 101 patients with nonhilar malignant biliary ob-struction. The group of patients with a tumor diameter,30mm had a median survival of 6.6 months as opposed to 3.2months for the group with a larger tumor diameter. Theauthors suggest an algorithm whereby patients with tumors,30 mm in diameter receive metal stents, and those withtumors of.30 mm in diameter receive plastic stents. Thisis in keeping with a previous study that included onlypatients with pancreatic cancer, in which a tumor diameterof .50 mm was predictive of a median survival of 2.5months (28). The presence of metastases was also a poorprognostic indicator, as previously suggested (7). Clearly,further studies analyzing subgroups with common etiologiesof malignant biliary obstruction are required to develop anoptimal management algorithm.

In summary, the Diamond metal stent is an effectivemeans of treatment for recurrent plastic biliary stent occlu-sion, but stent patency seems to be less than that reported inpatients receiving a metal stent as their first biliary prosthe-sis. However, because of the small sample size and diversepatient mix, firm conclusions regarding the cost-effective-ness of metal stenting after polyethylene stent occlusioncannot be made based solely on this study. Despite thisdecreased patency compared with that in previous reports,the metal stents still seemed to last significantly longer thanthe plastic stents that they replaced and, therefore, may stillbe cost-effective in the setting of recurrent polyethylenestent occlusion. Mathematical models may not be able toassume similar patency amongde novometal stents andthose placed after failure with one or more plastic stents.Prospective studies validating predictive survival modelswould facilitate the development of alternate cost-effectivestrategies for the use of expandable metal biliary stents forthe palliation of malignant biliary obstruction.

ACKNOWLEDGMENTS

The authors thank Microvasive (Boston Scientific Corpora-tion, Natick, MA) for providing the Ultraflex Diamondbiliary stents used in this study. Dr. Barkun is a researchscholar funded by the Fonds de la Recherche en Sante´ duQuebec. Dr. Romagnuolo is a research fellow funded by theAlberta Heritage Foundation for Medical Research.

Reprint requests and correspondence:Alan N. Barkun, M.D.,Director, Division of Gastroenterology, Montreal General Hospi-tal, 1650 Cedar Avenue, D7-148, Montreal, Quebec H3G 1A4,Canada.

Received July 25, 2000; accepted Dec. 13, 2000.

REFERENCES

1. Shepherd HA, Royle G, Ross APR, et al. Endoscopic biliaryendoprosthesis in the palliation of malignant obstruction of thedistal common bile duct: A randomized trial. Br J Surg 1988;75:1166–8.

2. Smith AC, Dowsett JF, Russell RCG, et al. Randomised trialof endoscopic stenting versus surgical by-pass in malignantlow bile duct obstruction. Lancet 1994;344:1655–60.

3. Andersen JR, Sorensen SM, Kruse A, et al. Randomised trialof endoscopic endoprosthesis versus operative bypass in ma-lignant obstructive jaundice. Gut 1989;30:1132–5.

4. Deviere J, Baize M, de Toeuf J, et al. Long-term follow-up ofpatients with hilar malignant stricture treated by endoscopicinternal biliary drainage. Gastrointest Endosc 1988;32:95–101.

5. Luman W, Cull A, Palmer KR. Quality of life in patientsstented for malignant biliary obstructions. Eur J GastroenterolHepatol 1997;9:481–4.

6. Ballinger AB, McHugh M, Catnach SM, et al. Symptom reliefand quality of life after stenting for malignant bile duct ob-struction. Gut 1994;35:467–70.

7. van den Bosch RP, Van der Schelling GP, Klinkenbijl JHG, etal. Guidelines for the application of surgery and endoprosthe-sis in the palliation of obstructive jaundice in advanced cancerof the pancreas. Ann Surg 1994;219:18–24.

8. Schmassmann A, von Gunten E, Knuchel J, et al. Wallstentsversus plastic stents in malignant biliary obstruction: Effectsof stent patency of the first and second stent on patient com-pliance and survival. Am J Gastroenterol 1996;91:654–9.

9. Leung J, Ling T, Kung J, et al. The role of bacteria in theblockage of biliary stents. Gastrointest Endosc 1988;34:19–22.

10. Speer AG, Cotton PB, MacRae KD. Endoscopic managementof malignant biliary obstruction: Stents of 10 French gauge arepreferable to stents of 8 Fr gauge. Gastrointest Endosc 1988;34:412–7.

11. O’Brien S, Hatfield A, Craig P, et al. A three year follow upof self expanding metal stents in the endoscopic palliation oflongterm survivors with malignant biliary obstruction. Gut1995;36:618–21.

12. Knyrim K, Wagner HJ, Pausch J, et al. A prospective ran-domized controlled trial of metal stents for malignant obstruc-tion of the common bile duct. Endosc 1993;25:207–12.

13. Davids P, Groen A, Rauws E, et al. Randomised trial ofself-expanding stents versus polyethylene stents for distal ma-lignant biliary obstruction. Lancet 1992;340:1488–92.

14. Born P, Neuhaus H, Rosch T, et al. Initial experience with a

1439AJG – May, 2001 Metal Biliary Stenting and Plastic Stent Occlusion

new, partially covered wallstent for malignant biliary obstruc-tion. Endoscopy 1996;28:699–702.

15. Smits M, Huibregtse K, Tytgat G. Results of the new nitinolself-expandable stents for distal biliary strictures. Endoscopy1995;27:505–8.

16. Prat F, Chapat O, Ponchon T, et al. Predictive factors forsurvival of patients with inoperable malignant distal biliarystrictures: A practical management guideline. Gut 1998;42:76–80.

17. Dumonceau JM, Deviere J, Delhaye M, et al. Plastic and metalstents for postoperative benign bile duct strictures: The bestand the worst. Gastrointest Endosc 1998;47:8–17.

18. Cotton PB. Outcomes of endoscopic procedures: Strugglingtowards definitions. Gastrointest Endosc 1994;40:514–8(editorial).

19. Yoo B, Kim J, Park J, et al. Prospective study for the useful-ness of self-expanding nitinol biliary stents in patients withmalignant obstruction of extrahepatic bile duct. GastrointestEndosc 1999;47:AB237(abstract).

20. Siqueira E, Martin JA, Vargas H, et al Prospective evaluationof a new metal stent for treating malignant biliary obstruction.Gastrointest Endosc 1999;47:AB236(abstract).

21. Dumonceau JM, Cremer M, Auroux J, et al. A comparison ofUltraflex Diamond stents and Wallstents for palliation of distal

malignant biliary strictures. Am J Gastroenterology 2000;95:670–6.

22. Matsuda Y, Shimakura K, Akamatsu T. Factors affecting thepatency of stents in malignant biliary obstructive disease:Univariate and multivariate analysis. Am J Gastroenterol1991;86:843–9.

23. Gillams A, Dick R, Dooley JS, et al. Self-expandable stainlesssteel braided endoprosthesis for biliary strictures. Radiology1990;174:137–40.

24. Tham TCK, Carr-Locke DL, Vandervoort J, et al. Manage-ment of occluded biliary Wallstents. Gut 1998;42:703–7.

25. Speer AG, Cotton PB, Russel RCG, et al. Randomized trial ofendoscopic versus percutaneous stent insertion in malignantobstructive jaundice. Lancet 1987;2(8550):57–62.

26. Prat F, Chapat O, Ducot B, et al. A randomized trial ofendoscopic drainage methods for inoperable malignant stric-tures of the common bile duct. Gastrointest Endosc 1009;47:1–7.

27. Yeoh KG, Zimmerman MJ, Cunningham JT, et al. Compara-tive costs of metal versus plastic biliary stent strategies formalignant obstructive jaundice by decision analysis. Gastroi-ntest Endosc 1999;49:466–71.

28. Petrek JA, Sandberg WA, Bean PK, et al. Can survival inpancreatic adenocarcinoma be predicted by primary size orstage? Am Surg 1985;51:42–51.

1440 Menon et al. AJG – Vol. 96, No. 5, 2001