Gut, 1991,32,685-689

Palliation of proximal malignant biliary obstructionby endoscopic endoprosthesis insertion

A A Polydorou, S R Cairns, J F Dowsett,R C G Russell

A R W Hatfield, P R Salmon, P B Cotton,

Departments ofGastroenterology andSurgery, UniversityCollege and MiddlesexHospital Medical School,LondonA A PolydorouJ F DowsettA RW HatfieldP R SalmonR C G Russell

Royal Sussex CountyHospital, BrightonS R Cairns

Division ofGastroenterology, DukeUniversity MedicalCentre, Durham, NorthCarolina, USAP B CottonCorrespondence to:Dr Stuart R Cairns, RoyalSussex County Hospital,Eastern Road, BrightonBN2 5BE.Accepted for publication16 July 1990

AbstractFor four years up to December 1987, 190patients (median age 73 years) with proximalmalignant biliary obstruction were treatedby endoscopic endoprosthesis insertion.Altogether 101 had cholangiocarcinoma, 21gail bladder carcinoma, 20 local spread ofpancreatic carcinoma, and 48 metastatic malig-nancy. Fifty eight patients had type I, 54 typeII, and 78 type III proximal biliary strictures(Bismuth classification). All patients wereeither unfit or unsuitable for an attempt atcurative surgical resection. A single endo-prosthesis was placed initially, with a furtherstent being placed only if relief of cholestasiswas insufficient or sepsis developed inundrained segments. The combined per-cutaneous-endoscopic technique was used toplace the endoprosthesis when appropriate,after failed endoscopic endoprosthesis inser-tion or for second endoprosthesis placement.Full follow up was available in 97%. Thirteenpatients were still alive at the time of reviewand all but one had been treated within the pastsix months. Initial endoprosthesis insertionsucceeded technically at the first attempt in 127patients, at the second in 30, and at a combinedprocedure in a further 13 (cumulative totalsuccess rate 89% - type I: 93%; type II: 94%;and type HI: 84%). There was adequate biliarydrainage after single endoprosthesis insertionin 152 of the 170 successful placements, givingan overall successful drainage rate of 80%.Three patients had a second stent placed bycombined procedure because of insufficientdrainage, giving an overall successful drainagerate of 82% (155 of 190). The final overalldrainage success rates were type I: 91%; typeI:83%; and type III: 73%. The early complica-tion rates were type I: 7%; type II: 14%; andtype 1I:31%. The principal early complicationwas clinical cholangitis, which occurred in 13patients (7%) and required second stent place-ment in five. The 30 day mortality was 22%overall (type I: 14%; type II: 15%; and type III:32%) but the direct procedure related mortalitywas only 3%. Median survival overall for typesI, II, and III strictures were 21, 12, and 10weeks respectively but survival was signifi-cantly shorter for metastatic than primarymalignancy (p<005). Endoscopic insertion ofa single endoprosthesis will provide good pal-liation of proximal malignant biliary obstruc-tion caused by unresectable malignancy in 80%of patients. Second stents should be placedonly if required. Extensive stricturing becauseof metastatic disease carries a poor prognosisand careful patient selection for treatment isrequired.

Although management ofdistal malignant biliaryobstruction by endoprosthesis insertion is wellestablished,' the place of endoscopic endo-prosthesis insertion for proximal malignantbiliary obstruction (Fig 1) remains controversial.Surgery, involving hepatic resection ifnecessary,offers the only chance of long term cure butcarries considerable procedure related morbidityand mortality.2 Most (90%) patients are, how-ever, unsuitable or unfit for attempted curativeresection.3 Palliative options include surgicalbypass and surgical percutaneous or endoscopicendoprosthesis insertion. ` Internal drainageshould always be preferred to external drain-age.89 Drainage may, if desired, be supple-mented by radiotherapy.`'01

Attempts at palliation, even in elderly patients,are usually worthwhile because the quality of lifemay be improved considerably by the reliefofthepruritus and nausea of cholestasis. As surgicalbypass procedures, in the best hands and in thefittest patients, carry a 33% operative mortality2and surgical intubation carries a 60 day post-operative morbidity of 26% and a 30% mortal-ity,3 a strong argument can be made formanagement by percutaneous or endoscopicendoprosthesis insertion. Large studies of bothtechniques in single centres have been few. Arandomised trial from this institution, however,suggested a lower complication rate with endo-scopic than percutaneous insertion, and theendoscopic route is therefore the preferred pri-mary intervention. 2 Moreover, the developmentof the combined percutaneous-endoscopic tech-nique (CP) has provided an increased endoscopic

Type I Type II

r;*.r.1 Type III

Figure 1: Diagram of types ofproximal biliary obstruction.

685

Polydorou, Cairns, Dowsett, Hatfield, Salmon, Cotton, Russell

success rate, while still avoiding large borehepatic puncture. 3 It has become our second lineintervention.We report our experience with endoscopic

endoprosthesis placement in 190 patients witheither primary or secondary proximal malignantbiliary obstruction at the hilum. In the analysiswe have attempted to address several points ofimportance to the clinician. These include theinitial success rate, the number of endo-prostheses required to provide adequate relief ofcholestasis, the relative merits of stent placementinto the left or right lobe, the prevalence of earlyand late septic complications, the prevalence oflate stent change either because of stent blockageor tumour progression with overgrowth, andfinally the value of stenting patients with secon-dary malignancy.

MethodsBetween October 1983 and December 1987, 190consecutive patients with biliary obstruction atthe hilum because of primary or secondarymalignancy were treated by attempted endo-scopic endoprosthesis insertion. Details, includ-ing intention to treat and the result of treatment,were prospectively entered onto a computerprogramme (PEDRO). There were 74 men and116 women with a median (range) age of 73 (35-94) years. The mean (range) duration of jaundicebefore the first endoscopy was 5 (1-14) weeks andthe mean (range) serum bilirubin was 361 (22-913) mmol/l, normal < 17. Other data included amean (range) serum alkaline phosphatase of 1540(360-4600) IU/1, normal <280, mean (range)serum albumin of 34 (20-43) g/l, normal 35-53,mean (range) serum creatinine of 106 (82-420)[imol/l, normal 50-125, and mean (range)haemoglobin of 11.7 (8'0-14-3) g/dl.

All patients were considered unsuitable forresection on the basis of medical fitness ortumour extent, or both, as defined by ultra-sound, computed tomogram (tumour extendinginto both lobes or major vascular involvement) orcholangiography (type II/III obstruction), orboth. The proximal obstruction was subdividedaccording to Bismuth's classification on the basisof cholangiographic or ultrasound information(Fig 1), or both, and according to proved orpresumed histology (Table I). It is stressed thatno attempt was made at any stage to obtain a'complete' diagnostic cholangiogram. Contrastmedium injection before stricture passage wasstopped once a duct deemed suitable for stentingwas identified and was continued only after

TABLE I Type ofstricture in study group

Type of biliary obsruction*

I II III Total

No (%) of patients 58 (31) 54 (28) 78 (41) 190(M/F) 19/39 23/31 32/46 74/116

Mean age (years) (range) 72 (41-94) 70 (35-91) 71 (42-92) 71 (35-94)Mean bilirubin (,umol/l)

(range) 292 (22-714) 444 (140-913) 343 (31-728) 361(22-913)Cholangiocarcinoma 25 28 48 101Pancreatic carcinoma 12 5 3 20Metastatic carcinoma 14 16 18 48Gall bladder carcinoma 7 5 9 21

*Bismuth classification.

secure access to the desired segment wasobtained with a 5 FG catheter. The rationalebehind this approach was to avoid contrastmedium injection into segments that would notbe adequately drained by the inserted stent. Onestent only was placed in all but two patients, inwhom the first segment stented was thoughtinadequate at the time of initial stenting.

All patients received parenteral antibiotics(usually mezlocillin 2 g, eight hourly, given onehour before the procedure and continued for atleast 36 hours after successful stent placement oruntil successful drainage was achieved). Anexperienced senior endoscopist was presentduring most of the procedures. Amsterdam type10 FG polyethylene endoprostheses wereroutinely inserted but 10% of patients had 8 or11.5 FG stents placed either initially or at stentchange. Stents were inserted over a guide wireand coaxial catheter via an Olympus TJFIO 4.2mm channel endoscope, using standard tech-niques.'4 Endoscopic sphincterotomy was per-formed in most patients both to assist stentinsertion and to facilitate later change. When astent could not be placed endoscopically, thecombined percutaneous-endoscopic technique(CP) was used.'5 All patients were carefullymonitored clinically after the procedures withparticular attention to fever, serial biochemistry,and ultrasound evidence of decompression.Patients were withdrawn from repeated attemptsat palliation and designated 'failed palliation' ifnon-specific general deterioration occurred or if,on discussion with the patient, the risks offurther intervention were agreed to outweigh thepossible benefit.

Successful endoprosthesis placement wasdefined as passage of the prosthesis across thestricture with good radiological positioning,immediate bile passage down the stent, andaerocholia on subsequent plain x ray or ultra-sound.6 17 Most patients also had segmentdecompression assessed by ultrasonography,usually 36-72 hours after the procedure.

Successful drainage was defined as loss ofpruritus, if present, plus a fall in the bilirubinconcentration greater than 30% of the pretreat-ment value within 30 days (but usually within 10days). It should be stressed that the aim wasreduction in bilirubin to acceptable concentra-tions (usually < 100 ,umol/l) where it was likely torelieve symptoms of cholestasis such as pruritusand nausea. 8

Early complications were defined as thoseoccurring within 30 days of endoprosthesisplacement. Cholangitis was diagnosed if a feverabove 38°C developed without other cause andwas associated with rigors or persisted for longerthan 48 hours after the procedure. That is,postprocedural fever on the night of the pro-cedure, which is undoubtedly caused by a briefbacteraemia and settles rapidly either spon-taneously or because of antibiotic prophylaxis,was not labelled cholangitis. Complications ofpercutaneous biopsies are not included and didnot contribute to the overall mortality andmorbidity.

Thirty day mortality was defined as deathwithin 30 days of the first attempt at endoscopicendoprosthesis insertion, whether successful or

686

Palliation ofproximal malignant biliary obstruction by endoscopic endoprosthesis insertion

TABLE II Results in 190 patients with proximal biliary obstruction treated by endoscopicendoprosthesis insertion

Type ofstricture*

l II III Total

No of patients1st attempt 58 54 78 190

Technical success:1st attempt (S/F) 45/13 (78%) 36/18 (67%) 46/32 (59%) 127/63 (67%)No further attempt 4 2 10 162nd attempt (S/F) 5/4(9%) 12/1 (22%) 13/5(17%) 30/10 (16%)Combined procedure (S/F) 4/0 (7%) 3/1 (6%) 6/2 (8%) 13/3 (7%)Total 54 (93%) 51(94%) 65 (84%) 170 (89%)

Successful drainage:Single stent 53 (91%) 44 (81%) 55 (71%) 152 (80%)Two stents - 1 (2%) 2 (3%) 155 (82%)

Early complications:Noofpatients 4(7%) 8(15%) 24(31%) 36(19%)No of events 6 (10%) 16 (30%) 28 (36%) 50 (26%)

30 Day mortality 8 (14%) 8 (15%) 25 (32%) 41(22%)Survival:Mean (wks) 23 22 17 20Median (range) (wks) 21 (2-85) 12 (1-94) 10 (1-115) 12 (1-115)

Stent change:No of patients 27 (47%) 25 (46%) 28 (36%) 80 (42%)No of events (S/F) 48/4 36/5 46/11 130/20

*Bismuth classification.

not. Procedure related mortality was defined asdeath directly related to a complication of endo-prosthesis insertion, including the complicationsof papillary access (sphincterotomy) or hepaticpuncture (CP). All patients were carefullyinformed of the possible symptoms of endo-prosthesis blockage and change was onlyperformed when clinical or ultrasonographicevidence, or both, of endoprosthesis blockagewas present.

Histological confirmation of malignancy wasattempted only after biliary decompression andwas obtained in 108 patients (57%) either byaspiration cytology or by Biopty gun trucutbiopsy. This was normally done at the time ofthe ultrasonographic decompression check.Diagnosis of the histological origin of the biliaryobstruction was surmised from available clinicaland imaging data in the remainder of cases.Follow up was obtained by examination of the

case notes and from referring consultants andgeneral practitioners. Statistical analysis wasperformed using X' with Yates's correction andMann-Whitney U test.

Results

ENDOPROSTHESIS PLACEMENT AND FUNCTIONEndoscopic placement of an endoprosthesis wasachieved in 127 patients at the first attempt, in 30at the second, and in 13 patients at CP. As shown

TABLE III Early complications in 190 patients withproximal malignant biliary obstruction treated by endoscopicendoprosthesis insertion

Type of biliary obstruction*

Complication I II III Total

Cholangitis 2 3 8 13Pancreatitis - - 3 3Retroperitoneal

perforation 2 2 3 7Gastrointestinal

bleeding - 2 5 7Stent migration 2 4 2 8Bile leakage - 3 5 8Respiratory failure - 1 1 2Renal failure - 1 1 2

*Bismuth classification.

in Table II, 16 patients had no further attemptsafter a failed first procedure (type I: four; type II:two; type III: 10). Seven patients proceeded toCP immediately after an initial failure of endo-scopic stent placement (type II: three; type III:four), and all patients except one who had a typeIII obstruction proceeded to CP after a secondfailed endoscopic attempt. The overall successrate for stent placement was 170 of 190 (89%)(Table II). Of the 20 patients with failed stentplacement, two had external drainage only afterfailed CP and the remaining 18 patients had nofurther intervention. The site of stent placementwas identified in 116 patients with type II or IIIstrictures. In 70 patients the stent was placed inthe right hepatic duct and in 39 the left duct. Nodefinite location was identified in the remainingseven cases.

Successful drainage was obtained in 80% ofpatients overall (type I: 91%; type II: 81%; typeIII: 71%) (Table II). That is, inadequate relief ofcholestasis despite adequate placement ofa singlestent occurred overall in 7% of patients (2%,13%, and 13% for types I, II, and III stricturesrespectively). Failure of drainage was signifi-cantly more common after-technically successfulendoprosthesis placement in type III than type Iobstruction (p<0 01). A second stent was placedby CP in three patients (type II: one; type III:two) because of failure of resolution of jaundicedespite segmental decompression by the firststent - all ofwhich provided successful drainage.The success rate of drainage overall thusincreased to 83% for type II strictures and 74%for type III strictures.

EARLY COMPLICATIONSComplications occurring within 30 days of theprocedure were present in 36 patients (19%)(Table III). Cholangitis was the principal majorcomplication occurring in 13 (7%), and if notassociated with endoprosthesis insertion failure,it was presumed to be caused by sepsis inundrained segments. Cholangitis resolved onparenteral antibiotics alone in five patients, withreplacement of the single stent in three, and infive the placement of an additional second stentinto the presumed undrained segment was per-formed (this was achieved endoscopically in twoand via CP in three patients). All CP wereperformed via the ducts of the left lobe of theliver and all three patients had had initial endo-scopic endoprosthesis insertion into right lobeducts.The prevalence of complications after stent

placement in patients with type II or III obstruc-tion was not significantly different in patientswith an endoprosthesis placed in the right ductscompared with those with an endoprosthesisplaced in the left hepatic duct (p>0Q05).

MORTALITYFollow up to 30 days was obtained for allpatients. The 30 day mortality was 22% overall(41 patients) and was 14% for type I obstruction,15% for type II, and 32% for type III (Table II).There was a significant difference in 30 daymortality between patients presenting with type I

687

Polydorou, Cairns, Dowsett, Hatfield, Salmon, Cotton, Russell

W

CL

0)0.

* Type I&-^ Type 11

a Type III

Months

01 \ t X-~~~Metastaticcarcinoma60-

0.-0 40-

20-

0

0 3 6 9 1'2 15Months

Figure 2: (A) Patient survival according to type ofstrictureActuarial 18 month survival after endoscopic placement ofendoprosthesis in patients with unresectable malignant liverhilar strictures (n= 172). P (50% survival): I v II=0 288;v III=0 006; II v III=0 162. (B) Patient survivalaccording to cause ofstricture. Actuarial survival ofpatien:with hilar stricture who were dead at the time ofreview.according to histological diagnosis (n=172).

and type III hilar stricturing (p<0Q05) anbetween patients with type II and type Istrictures (p<0 05) (Fig 2).Death within 30 days was caused by a pr(

cedure related complication in five patients (30/(bleeding four, sepsis one). The remainirpatients died from progressive malignant disearrather than a complication of the procedure.

SURVIVALThirteen patients were alive at the time ofrevievAll except one had been treated within tIprevious six months. The single long term suvivor still alive was first treated 132 weeks befo:this review. Five patients were lost to long ternfollow up. Data about patients who have die(172) is provided in Table II. All patients who dinot have biliary drainage established died withisix weeks. Survival was significantly longer fitype I than type III obstruction (p<001) (Fig 2Shorter survival was seen in patients wilobstruction caused by metastatic disease or paicreatic carcinoma than in those with eithcholangiocarcinoma or gall bladder carcinonr

TABLE IV Median and mean survival and 30 day mortality according to extent and cause ohilar stricture. (Only patients with a histological diagnosis are included, n= 108. )

Mean (median) (weeks) p 30 day mortality (%)

C+GB P+M C+GB P+M

I 26 (22) 19 (12) (0-291) 10 19II 23 (16) 17 (10) (0.507) 12 23III 19 (12) 8 (6) (0.048) 28 42

C=cholangiocarcinoma; GB=gall bladder cancer; P=pancreatic cancer; M=metastatic cancer.

(Table IV; Fig 2). Survival of patients withsuccessfully drained type II or III obstructionwas not significantly longer in those treated byinsertion ofa stent in the left rather than the righthepatic duct system (p>005).

LATE COMPLICATIONSEighty patients required between one and sixstent changes (one change: 54 patients; twochanges: 13 patients; >two changes: 13 patients)(Table II).

1 DiscussionThere are numerous alternative palliative treat-ments available for patients with proximal malig-

_J nant biliary obstruction for whom attemptedcurative surgical resection is not possible.'920These include surgical bypass24 2' and surgical,3 22percutaneous, and endoscopic` 11 20 23 endo-prosthesis insertion. Internal drainage is prefer-red to external drainage for fluid and electrolyte,tumour seeding, and cosmetic/psychologicalreasons, though it probably carries a higher risk

8 of sepsis.24 Extensive obstructive stricturing isnot, however, amenable to conventional surgicalbypass and requires special surgical techniques.25Although surgical bypass offers the possibility of

r a single permanent definitive palliative pro-I cedure, it carries a high procedural morbidity

ts and mortality,22' has only been reported in theminority subgroup of fit, younger patients, andalthough generally technically straightforward, itinvolves a general anaesthetic and tissue dissec-

id tion. Surgical intubation has been performedII when surgical bypass has not been possible and

carries all the disadvantages of surgery and noo- proved benefit over the non-surgical alter-to) natives.3 For most patients, who are eitherig elderly, unfit, or have extensive bilobar disease,se percutaneous or endoscopic endoprosthesis

insertion are valid and often the only alternatives.As a randomised study has shown the complica-tion rate to be lower with the latter,'2 endoscopicstent placement is now our treatment of choice

v. for all patients not amenable to attempted resec-he tion. If this fails, the CP technique may be usedr- to place the endoprosthesis, with the advantagere over pure percutaneous placement of a smallerm bore liver puncture and presumably less bleeding~d and bile leakage. 31'id In this series of 190 patients treated endo-in scopically, the only group with potentiallyor resectable disease was those with type I!). cholangiocarcinoma (n=25). Eighteen of theseth patients were over the age of 70 years, four hadn- intrahepatic metastases, and 12 had stronger medical contraindications to surgery. The rarityna of potentially resectable hilar cancer is well

illustrated by Bismuth's series of 213 patientsf referred to a specialist liver surgical unit over 25

years,3 with only 16 (7 5%) patients undergoingcurative resection.

Overall, the results in this series of patientswith proximal malignant biliary obstructiontreated by endoscopic endoprosthesis insertionwere similar to results of previous series.'820 26

- Successful stenting was achieved in 170 (89%)patients and successful drainage of the biliary

688

Palliation ofproximal malignant biliary obstruction by endoscopic endoprosthesis insertion 689

tree in 155 (82%). Early complications occurredoverall in 36 patients (19%), and 41 (22%) diedwithin 30 days. Cholangitis was the principalcomplication occurring in 13 patients (7%).Patients with failed stent insertion or drainage inthis series were generally managed further byconservative means only, although three hadexternal drainage. It is probable that the tech-nical success rate of stent placement and thedrainage success rate could have been increased ifall patients had had the combined procedureperformed after initial or second endoscopicfailure. However, continuation of palliativeattempts was deemed inappropriate for 16patients because of a very poor clinical state(Table II).There are two prominent areas of difference

between this and previous series. Firstly, theimmediate periprocedural cholangitis rate waslower than previously reported.20 26 This wasprobably the result of our use of prophylacticantibiotics combined with the policy of attempt-ing to inject contrast medium only into thesegment subsequently stented. It is our beliefthat the filling of all segments to obtain acomplete 'diagnostic' cholangiogram should beavoided and only sufficient contrast should beinjected to obtain imaging of the duct to bestented. Applying this rule, only five patientsrequired placement of a second endoprosthesisfor cholangitis. Secondly, only a single endo-prosthesis was placed unless either sepsisdeveloped in undrained segments or resolution ofcholestasis was insufficient. The former occurredin five patients as described and the latter in afurther three patients. That is, a single endo-prosthesis provided palliation in all but eight ofthe patients successfully treated. Most of thesesecond endoprostheses were placed in the leftlobe using the CP procedure. Thus, routineplacement of more than one stent to achievedrainage of all obstructed liver segments asadvocated by Deviere et aPF6 would seem unneces-sary, and if attempted but unsuccessful may bethe cause of a raised prevalence of cholangitis.2Moreover, the placement of a nasobiliary drainabove the stent to irrigate the liver, as advocatedby some groups, would seem more likely to causesepsis in undrained segments than prevent it.26The least intervention for the desired result isalways the best.

Distant or local metastatic cancer causingproximal biliary obstruction may be managed inan identical manner to primary malignancy,except that resection is not suitable. Themorbidity and mortality associated with endo-prosthesis insertion, however, is considerablyhigher for these patients and the survival issignificantly shorter (Fig 2; Table IV). This isundoubtedly because of the influence of theunderlying disease, yet even a short term survivalmay be considerably improved in terms of qual-ity of life by relief from pruritus or nausea. Thepoor prognosis in this group does, however,demand careful patient selection for treatmentand we would not advocate intervention on thebasis of serum biochemistry alone.

In conclusion, most patients with unresectable

malignant proximal biliary obstruction may betreated successfully with an acceptably low com-plication rate by the endoscopic insertion of asingle endoprosthesis. Both primary and second-ary malignancy may be managed successfully.The prevalence of cholangitis is lower than withmore complex interventions. Surgical interven-tion should probably be limited to the minority ofyounger and fitter patients in whom the tumourseems to be resectable.A A Polydorou is supported by the State Scholarships Foundationof Greece and J F Dowsett by the Cancer Research Campaign.The authors thank Mr M J McMahon and Mr E M Elliot of the

Department of Surgery, Leeds General Infirmary, for their help inpreparation of this manuscript.

1 Dowsett JF, Polydorou AA, Vaira D, et al. Endoscopicstenting for malignant biliary obstruction: how good really?A review of 641 consecutive patients. Gut 1988; 29: 1458.

2 Blumgart LH, Hadjis NS, Benjamin IS, Beazley R. Surgicalapproaches to cholangiocarcinoma at confluence of hepaticducts. Lancet 1984; i: 66-70.

3 Bismuth H, Castaing D, Traynor 0. Resection or palliation:priority of surgery in the treatment of hilar cancer. WorldJ7Surg 1988; 12: 39-47.

4 Bismuth H, Corlette MB. Intrahepatic cholangioentericanastomosis in carcinoma of the hilus of the liver. SurgGynecol Obstet 1975; 140: 170-8.

5 Nakayama T, Ikeda A, Okuda K. Percutaneous transhepaticdrainage of the biliary tract: techniques and results in 104cases. Gastroenterology 1978; 74: 554-9.

6 Soehendra N, Reyders-Frederix V. Palliative bile duct drain-age - a new endoscopic method of introducing a trans-papillary drain. Endoscopy 1980; 12: 8-11.

7 Soehendra N, Grimm H. Endoscopic retrograde drainage forbile duct cancer. WorldJ Surg 1988; 12: 85-90.

8 Burcharth F, Jensen LI, Olesen K. Endoprosthesis for internaldrainage of the biliary tract. Gastroenterology 1979; 77:133-7.

9 McPherson GAD, Benjamin IS, Hodgson HJF, Bowley NB,Allison DJ, Blumgart LH. Preoperative percutaneous trans-hepatic biliary drainage: the results of a controlled trial. BrJSurg 1984; 71: 371-5.

10 Mayer WC, Scott Jones R. Internal radiation for bile ductcancer. WorldJr Surg 1988; 12: 99-104.

11 Levitt MD, Laurence BH, Cameron F, Klemp PFB. Trans-papillary iridium-192 wire in the treatment of malignant bileduct obstruction. Gut 1988; 29: 149-52.

12 Speer AG, Cotton PB, Russell RCG, et al. Randomised trial ofendoscopic versus percutaneous stent insertion in malignantobstructive jaundice. Lancet 1987; ii: 57-62.

13 Dowsett J, Vaira D, Hatfield A, et al. Endoscopic biliarytherapy using combined percutaneous and endoscopic tech-niques. Gastroenterology 1989; 96: 1180-6.

14 Cotton PB. Endoscopic method for relief ofmalignant obstruc-tive jaundice. WorldJ Surg 1984; 8: 854-61.

15 Dowsett JF, Vaira D, Hatfield ARW, et al. Endoscopic biliarytherapy using combined percutaneous and endoscopic tech-niques. Gastroenterology 1989; 96: 1180-6.

16 Laurence BH, Cotton PB. Decompression of malignant biliaryobstruction by duodenoscopic intubation of bile ducts. BMJ1980; 280: 522-3.

17 Cotton PB. Duodenoscopic placement of biliary prostheses torelieve malignant obstructive jaundice. BrJI Surg 1982; 69:501-3.

18 Leung JWC, Emery R, Cotton PB, Russell RCG, Vallon AG,Mason RR. Management of malignant obstructive jaundiceat The Middlesex Hospital. BrJr Surg 1983; 70: 584-6.

19 Pichlmayer R, Ringe B, Lauchart W, Bechstein WO, Guber-natis G, Wagner E. Radical resection and liver grafting as thetwo main components of surgical strategy in the treatment ofproximal bile duct cancer. World J3 Surg 1988; 12: 68-77.

20 Tytgat GNJ, Bartelsman JFWM, Den Hartog jaker FCA,Huibregste K, Mathus-Vliegen EMH. Upper intestinal andbiliary tract endoprosthesis. Dig Dis Sci 1986; 31: 57-76.

21 Miyazaki K, Nagafuchi K, Nakayama F. Bypass procedure forbile duct cancer. WorldJ7 Surg 1988; 12: 64-7.

22 Black K, Hanna SS, Langer B, Jirch DW, Rider WD.Management of carcinoma of the extrahepatic bile ducts.Can JSurg 1978; 21: 542.

23 Kiil J, Kruse A, Rokkjaer M. Endoscopic biliary drainage.BrJ3 Surg 1987; 74: 1087-90.

24 Riemann JF, Lux G, Rasch W, Beickert-Sterba A. Nonsurgical biliary drainage - technique, indications and results.Endoscopy 1981; 13: 157-61.

25 Choi TK, Fan ST, Lai ECS, Wong J. Malignant hilar biliaryobstruction treated by segmental bilioenteric anastomosis.Surgeiy 1988; 104: 525-9.

26 Deviere J, Baize M, De ToeufJ, Cremer M. Long-term follow-up of patients with hilar malignant stricture treated byendoscopic internal biliary drainage. Gastrointestinal Endo-scopy 1988; 34: 95-101.

27 Deviere J, Motte 5, Dumonceau JM, Serruys E, Thys JP,Cremer M. Risk factors for septisaemia after biliaryandoprosthesis placement (BEP). Endoscopy 1988; 20: 79.