Percutaneous biliary drainage with emphasis on hilar lesions

The mortality rates of surgery and percutaneous transhcpatic biliary drainage (PTHRD) are comparable. Long tenn studies show th.u delayed compltcauons (X:cur in the majority of cases of PTI IBD and surviv,tl is not improved compared to surgery. The many recent advances 111 endnscop1c ancl percutaneous drainage techniques and the recognition that the patient is best served by a noncompetitive multidisciplinary approach will ensure that virtually every patient obtains the most satisfactory drainage possible with a mirnmum of risk and discomfort. Endo:.copic drainage should be the first therapeutic option, with radiologic assistance 111 the 15 ro 25% where endoscopic drainage fails orb incomplete. Can J Gastroenterol l 990;4(9):579-587

H owever, many of these early papers compared the 'surgical' more.ti iry of surgical procedures with the procedural mortality of PTI 11m 20 to 30% and 0 w 4%, respectively (2,3  the mortality rates of surgery anJ rn IBD were comparable ( 4 ). With ume, t he low complication races reported m early series were seen in part to be the result of short follow-up penods. Longer term studies showl·d chat JclaycJ complicanons occurred in the maJority nf c;ises and surv ival was not Improved compared to surgery ( 5).
T he long term surv1h1I of pauents with malignant obstru<.tion was shown to he dependent on dw LonJit ion uf the pat1enr, not on rhe mode of drainage (6). PTI IBl) remained a part of the medical armamcntanum because 1l was a simpler anJ more cost effective way of treat ing patients compared t1) surgery, anJ patients had shorter hosp1rnl stays.
A resurgence of PTI m1) has occurrcJ in the past few years ( 12) as 1l has hecome dear that not all biliary ohs1 ruc-t1ons um bl' handled by enJnscop1sts a lone. Radiologic support is important in paucnts with ohst ruct 1on of the upper gastroinres11nal tract, prtor  In this /Jatienc the right duce obsrrucuon extends u/J to the first division poinc surgery (especially Roux-en-Y loop.s and Billroth II anastomoses) and endoscopic failures. Cremer ( 13) notes that radiological help is required in 25% of hilar lesions.
'Turf battles have been overcome in part by the introduction of combined radio I og ic-e ndosco pi c procedures ( 14,15). Furthermore, a host of new applications has been developed.
Over the past l 5 years there have been many technical developments which have improved the ease, safety and success rates of these procedures. Among the most important are the heavy duty guidewires developed by Lunderquist ( l 6) aml improved by Amplatz, which facilitate dilation of the track and catheter insertion. A variety of wires has also been developed to aiJ in bypassing tight and tortuous strictures such as the Lunderquisc-Ring torque control wire, the variable core wire and the polymer coated 'glide' wire. Drainage catheters are now made of softer, more comfortable materials, have better retention devices and a greater resistance to encrustation. A wide variety of percutaneously place-580 able enJoprosthescs has also been developed for increased ease of placement and longer functioning life.

INDICATIONS AND APPLICATIONS
Currently in the author's institution and many others, the main indication for ITHBD is obstructed biliary tree following failure of endoscopic drainage. In many other centres where enJoscopic drainage is not available, PTI IBO remains the first-line approach to the obstructed biliary tree.
Some early studies suggested that routine preoperative drainage of surgical candidates with hyperbilirubinernia was a helpful procedure ( 48), but other studies have not borne this ouL (49), and preoperative drainage has fallen out of favour. TECHNIQUE PTI nm is painful, so standard PTI lllD technique begins with sedation of the patient and good analgesia. Mo~t patients arc given intravenous narcotics, usually morphine or fentanyl, with the dose titrated to the patient's pain level.
A variety of other methods of pain control are applied as required. Some authors add mtercostal nerve blocks tn the usual extensive infiltration of local anesthetic in order to control somattc ~anglion blockade can reduce deep 'v1scernl' pain (51), and some authors use it routinely Inrrapleural hlock can reduce both somatic anJ visceral pam (52). It c.an also be performed during rrrHRD if nec.es~ary (53). Rarely, epidural or general anesthesia may be rt'quired for patients en toleratt' the pro-ccJure ( 50). Some rad1olog1sts use routine antibiotic prophylaxis (54 ); others Jo not.
A fine needle cholangiogram is usually performed as the initial step to outline the anatomy and select an appropriate duct (55,56). This part of the procedure may be Jone blindly using only abdominal wall and radiologic landmarks, or it may be guiJe<l by ultrasound (57). Care must be taken ro inject the least amount of contrast necessary to plan the procedure. Undue elevation of intraJuctal pres~ure is thought to contribute ro rhe incidence of sepsis (58). Carbon d1ox1de may be mtro<luced to show the left duct system, since it will rise anteriorly with the pat tent supine. Recause there ,s a risk of sepsis with inJection, some radiologists forego the Jiagnosttc and 'mapping' chol.rngwgram anJ go directly to drn 111age A variety of mst ruments have been Jc, e lopeJ for ·~, ng k st ,c.k' drainage ( 59).
After the cholnngiogram has shPwn the duct sy~tem, an appropriate dutt ts selected for puncture ( Figure I). The puncture stte ts Just anterior to the m,<l axillary line below the I 0th nb to a\'0t<l lung and pleural space. A lower puncture site should he chosen tf this will result in an appropriate, preferably straight or downward approach to the chosen duLt, nr if the lung c.learly comes below the 10th rih. The ,iuthor performs lateral fluomscopy m order to ensure a horizontal approach to the chosen duct. Once the duct is punctured, a gu1dewire 1s introduced as far as possible to secure the track, which is then dilated and a torquahk, catheter is introduced. The catheter and wire are then manipulated beyond the ohstruction mto the duoJenum. When the guiding c:-itheter is removed, the Percutaneous biliary drainage maintam liver functton in patients with both ducts obstructed by h1lar lesions. lf the ob~tructton 1s below the h,lum then the entire biliary tree can he Jramed. Several advantages art• realized hy a leftsided approach. The lung, pleura anJ diaphragm are all avoided, and therefore chest complicauons do not occur (62). Pleurittc and shoulder up pam anJ splinting are also almost cl1mmated. Reduced O\'erall 1.:ompltcations have also hecn documented (61 ).
In the author's experience patients have much less long tl·rm d1su101fort, possibly because the catheter is less disturbed hy respiratory movement. A further advantage is that the left hepauc duct has a longer length of unhrancheJ duct above the confluence than docs the right. In practical terms this means that a ctrcumferenually cxpandtng lesion cakes longer to obstruct secondary branches on the left than on the nghr. Left-s1<lcJ catheters should therefore have a longer useful life before progressl\·e segmental 1solatton occur~. Some authors (63) recommend dram mg both duc1 systt·ms rout mely 111 patients with htlar nhsrruct1on to prevent the development of cholangit is m the undrained lohe. The author Joes this routmely tn pattents m whom pm)r endoscopic drainage has hecn attempted However, when hoth ducts arc drained, the risks are add itive. If the biliary system has not been contaminated by the en<loscopist, generally only a single (left) PTHBD catheter is placed. Should the patient show any evidence of sepsis unresponsive LO conservative therapy or not achieve adequate palliation of symptoms, the author proceeds co decompression of the undrained lobe (Figure 4).
A variety of ingenious procedures have been Jevcloped to reduce the number of PTHBO catheters and/or percutaneous punctures. In early cases the author placed internal right and external left drains through the same right lateral puncture. Druy and Melville (64) used asimilar technique but linked the tubes via an external connector so that the left duct drainage passed into the right internal drain anc.l therefore also drained internally. By laterconver- Uflacker (65) modified this 'crossover' technique further with placement of two internal stems ( Figure 5). His group also used a similar technique (65) in which the 'crossover' endoprosthesis did not pass from right to left via the duct system hut through a transparenchymal tract ( Figure 6). Even T tubes have been used to drain both ducts from one puncture (66). Burke and McLean (67) modified a Cope loop gastrostomy catheter. The catheter could be placed so that some holes were m both right and left ducts and some in the jejunum, allowing effective internal drainage of ana~tomotic choledochojejunal stricture via a single right puncture. Theoretically this technique can also be used in the case of hilar tumours if the duel below is large enough to accept the loop.
If single puncture drainage of both ducts fail~ or is not feasible, it may still he possible to place an internal drainage catheter from one side, and a crossover catheter or prosthesis from the other, which will still allow bilateral internal drainage. As a fallhack position, the patient can be left with one internal and one external drain (Figure 4 ). Bilateral external drainage is the 'worse case' scenario, but will still result in clinical palliation, albeit al the cost of two external tubes and bile replacement therapy. The large number of options available and the ingenuity of some of thc~e techniques can lead to technically effective drainage in the majoriLy of obstructed patients. However, it must always be kept in mind that most patients requiring these techniques have limited lifespans no mauer how technically successful the procedure. The interventionalist should be careful

COMPUCA TIONS OF PERCUTANEOUS DRAINAGE
The cm11nr complications of percutaneous Jrainage, whether hy internal or external catheter or by endoprosthcsis, are related to the creation of a track from the skin to the hik duct. Thb track, of necessity, passes through the liver and peritoneal space. It may a lso pass through the hem1diaphragm and pleural space if a right lateral approach 1s used. Bleeding, a major cause of mortality and morbidity, is undoubtedly the result of lacerarion of blood vessels and/or liver parenchyma. Such bions occur in up tu 33% of rTI !BJ) patients (69). Hcmohilia is the most common complication in some series and may be fatal (70). LcsserJegreesofhemorrhage may require transfusion or emholotherapy tn control blcedmg (7 1 ).
The types of lesions requiring therapeutic 111tervcntH1n included hcpallc arrery ,ineurysms, hcpauc anery-porrnl vein fistUlae and venous var ices ( 71 ). immediate sepsis is likely the result of tramient biliary venous fistulae created during rhe procedure (58). Delayed infccnon, ic, cholangitis, occurs Lil up to 47% (72). In Carrnsco's series ( 58,72) cholangit1s was more common 1t1 patients with mternally versus externally draming cmhcters. In the present authm\ experience of over 300 cases, chnlangms is almost 111variab ly the result oi poordramage, and if patients live long enough, all catheters will hecome occluded or function polxly. For this reason all rTI IBD catheters are routinely exchanged at 60 co 90 <lay intervals. Using this routine the author has almost been able to eliminate cholangitis in patients with indwelling rTI mo catheters. Cholangit is can also develop with multiple strictures and undrained segments (73 ).
Bile leakage is another complication of the transhepalic track. It may come about as a result of poor technique, catheter obstruction by clots or debris or catheter d islo<lgement ( 72). Bi le peritonitis can occur if the catheter dislodges before a fibrous track to the skm forms, and can be fatal particularly when the bile is infected. Bile leakage along the track around an indwelling catheter can generally be controlled by placement of a larger catheter. If Lhe catheter dislodges after ;.i good biliarycutaneous track has formed, replacing the catheter alo ng the pre-existing route is generally simple (74,75). lf a track has not formed, then a new hiliary drainage procedure may be required.
Laparotomy may also be needed if sufficient bile leakage into the peritoneal space has occurred.
Tumour can grow along the catheter track and spread toskm (78), peritoneal space (79) and pleural space (77). Long Lenn catheters can also be complicated by erosion of ribs (80).
Some skin and abdominal wall discomfort is common and may in some patients require nerve block or local anesthesia for relief (81 ).
Oth er complications are the resu lts of technical problems, eg, duodenal perforation (82) or knotted catheters (83 ). Still others are of uncertain etiology, such as acute c holecystitis (84 ).
Yee and Ho (85) compared complication rates in patients with benign and malignant obstructions and found substantially fewer major complications (2 versus 7%) and deaths (0 versus 2%) in the benign group. Presumably these differences reflect the poor condition and more advanced age of t he malignancy group. The death rate in the multiple series reviewed by Y ce ranged from 0 to 6%. Deaths in most series resulted from hemorrhage, sepsis or bile peritonitis.

ENDOPROSTHESESVERSUS CATHETER DRAINAGE
Many radiologists prefer to place endop ros theses rathe r than PTHBD catheters, and in the preceding Jiscussion Lhe terms 'stent' or 'endoprosthesis' can serve in place of'PTI IBD catheter' or 'drain.' Endoprostheses have proven effective when placed by endoscopist (86,87) or radiologist (88,89), although Speer (90) has shown mortality lO be less with endoscopic drainage stems than catheters. le is not the author's intention to debate the relative merits of PTHBD and endoprosthcses. Many factors enter into such a discussion, among them rhe technical sk ills of the enJoscopisc and radiologist, che psychological state, preferences and life expectancy of t he patient, where the patient lives and how he will be monitored, and t he number, type and sites of the obstructing lesions (91,92).
Radiologists can place scents in virtuall y any Juct system in which they have obtained internal drainage. If the sLenL cannot be placed in t he usual an-tegra<le manner, Lhen a radiological method of transpapi llary placement has been developed which is analogous to endoscopic placement, but docs not re, quire an endoscope (93). A wire placed percutaneously is retrieved from the duodenum by a perorally placed basket which is manipulated into position unJer fluoroscopic gu idance. One end of the wire is t hen pulled up and the endoprmthesis attached. By a combination of pushing and pulling the en<loprosthcsis is drawn back into the gut and then into position across t he duct obstruction.
Alternatively a wire placed via PT! IBD catheter is retrieved by a basket under endoscopic guidance and the cndoprosthesis placed in a similar manner (14,94). A third method, which is often used in the author's Jepartment, is for the endoscop ist co place a wire alongside the rn IBD catheter ( Figure   7). The wire will often cross the previously impassable stricture, presumably hecause the catheter has either straightened the track or has slightly eroded the obstructing lesion as a result of continuous movement with respiration. The percutaneous track can be Figure 7) Endoscopic retrograde cholangiopancreatography. Top Attempts to pass a guide111ire were unsuccessful. The wire kept passing out the T-tube track (arrowhead) and would not negotiate the 9Cf' bend, the upper end of which was blocked by a stone. Bottom Following percutaneous transhepatic biliary drainage, the bend in the duct has been srrarghcened by the catheter, and the stone pushed below the area of stenosis. The endoscopist was able to pass a guidewire and endoprosthesis alongside the catheter without difficulty plugged with gelfoam or coils when the PTHBD catheter is removed.
Placing stencs by any of these retrograde methods has the advantage ofallowing a large Stent to be positioned without creation of a large hole in the liver, a major cause of mortality with percutaneous Stent placement (91 ).
The pain associated with dilation of the liver track is eliminated and, theoretically at least, there is a decreased risk of bleeding. The author prefers therefore to let the endoscopists place stents rather than place them radiologically.
Another reason to place endoprostheses by one of these methods is to ensure that endoscopic replacement will be possible if the prosthesis occludes.
While occluded endoprostheses can be exchanged radiologically (95), the author elects not to place them in patients where endoscopic exchange is not feasible, because replacement would require an additional PTHBD procedure with all the attendant risks. An endoprosthesis which is not endoscopically changeable is placed only if the patient has a very short life expectancy and will likely die before occlusion of the endoprosthesis occurs. [n summary, if a scent cannot be placed endoscopically, it is generally preferable to leave a PT! !BO catheter in place. Exchange of an occluded or displaced PTIIBD catheter is a quick painless outpatient procedure.

RECENT DEVELOPMENTS AND FUTURE DIRECTIONS
Large diameter metal stents (up to 30 French) have been developed (96). These stents can be placed through 8 to !OF delivery systems. The theory behind them is that larger diameter scents will be less likely to occlude than the 10 to 14F stems in general use today.
Preliminary reports from Britain suggest that this hope may not be realized. Gillams (97) reported that recurrent jaundice occurred in 42% of patients treated with "Wallstents." Tumour growth through the interstices of the metal mesh or beyond the ends of the stent was among the reasons.
A variation which may prevent tumour growth between the wires of the scent involves a polymer coating which in effect makes the walls of the scent one solid piece without holes. This type of stem is not self-expanding and requi res balloon expansion, accomplished by a catheter-mounted balloon similar to that used for angioplasty (98). Tumour growth through or above the CAN J GASTROENTEROL VOL 4 N o 9 DECEMBER 1990 Percutaneous biliary drainage stent is clearly not a problem in benign diseases, and the expected beneficial result of larger stents has been realized in a small group of patients with nonmalignant obstructions (99).

CONCLUSIONS
In general, the author places endoprostheses endoscopically as a firstline approach. Any of the methods described above may be used to place a scent endoscopically after PTHBD if the original endoscopic drainage fails. Rarely, a scent must be placed radiologically in a patient in whom, for technical reasons (eg, esophageal obstruction), these methods cannot be used. If the stent cannot be placed endoscopically, it cannot be changed endoscopically either. Occlusion of the stent will then require another PTHBD with all the attendant risks. Therefore, in this special circumstance, radiological Stent placement will be performed if and only if the patient's life expectancy is less than the expected stem life. If the patient is likely to have a longer survival, then the author leaves him with a PTHBD catheter, which can be readily and safely exchanged on an outpatient basis. Such catheters are routinely changed at 90 day intervals.
The many recent advances in endoscopic and percutaneous drainage techniques, and the recognition that the patient is best served by a noncompetitive multidisciplinary approach (15) will ensure that virtually every patient obtains the most satisfactory drainage possible with a minimum of risk and discomfort. Endoscopic drainage should be the first therapeutic option, with radiologic assistance in the 15 to 25% where endoscopic drainage fails or is incomplete.