Endoscopic retrograde cholangiopancreatography (ERCP) is a combined endoscopic and radiological procedure which plays an essential role in the diagnosis and the management of diseases of the biliary tract and pancreas. The diagnostic technique was first described in 1968 and this was quickly followed by the development of therapeutic endoscopic procedures for the relief of obstructive jaundice and the extraction of stones from the bile duct. Although ERCP requires expensive equipment and technical skill, it is now widely available and it should be regarded as complementary to percutaneous transhepatic techniques.

Modern duodenoscopes are sophisticated side-viewing endoscopes, which are designed specifically for use within the duodenum. Older instruments produce an optical image from a congruent fibreoptic cable to which a television camera can be attached. In the future, video duodenoscopes, which incorporate the camera in the tip of the endoscope, will become the standard instrument. The television image is a real advantage as the endoscopy assistants are able to see what they are doing, and teaching the technique is made much easier. Endoscopes with instrument channels up to 4.2 mm in diameter are available. They are fully insulated and completely immersible, so that the whole instrument can be sterilized.

The basic diagnostic cannula is a plastic tube, 200 cm long, with an outside diameter of 1.6 mm, and a single terminal opening with a rounded tip marked at centimetre intervals for the last 5 cm (Fig. 1) 1375. The most important therapeutic instrument is the sphincterotomy knife (Fig. 2) 1376. This consists of a fine wire running down inside a plastic cannula and attached at the tip. For the final 3 cm the wire runs outside the plastic tubing. When the wire is shortened within the plastic tubing at the proximal end, the wire at the distal end ‘bowstrings’ across the tubing (Fig. 3) 1377. Other instruments are flexible guidewires, baskets, and balloons of a conventional design and suitable length and strength (Fig. 1) 1375.

A good-quality radiological screening unit with an image intensifier, which will also take still radiographs, is essential. A television screen should be placed behind the patient's head where the picture is comfortably visible to the endoscopist and the radiologist.

The common indications for ERCP are extrahepatic obstructive jaundice and cholangitis. In most circumstances the endoscopist will attempt to relieve the obstruction or to remove the stones at the same examination. A diagnostic examination is appropriate in any patient in whom biliary or pancreatic disease is suspected, provided the same information cannot be obtained in a simpler or safer way. A preliminary ultrasound examination is always essential, and may, in expert hands, lead to a diagnosis. Even if diagnosis is not possible, the findings may influence management during the endoscopy.

There are no absolute contraindications. An ERCP cannot be performed if the patient has oesophageal, gastric, or duodenal obstruction. Patients with hepatitis or other infectious conditions should be examined at the end of the list, and the endoscopes should be cleaned and sterilized immediately afterwards.

Most patients of any age and in any condition can tolerate an ERCP. Those with respiratory impairment require careful sedation, but patients who are severely ill, from cholangitis for example, are often better treated, after resuscitation, by this method rather than conventional surgery.

Blood coagulation must be normal. The prothrombin time should be less than 1.5 times the control value, and the platelet count must be within the normal range. Jaundiced patients receive vitamin K routinely, and fresh, frozen plasma is given immediately before the endoscopy if the prothrombin time remains abnormal. Patients are starved for at least 6 h beforehand and all of them receive intravenous fluids even though renal failure following an ERCP in a jaundiced patient is rare. When a therapeutic procedure is planned, the patient must receive an intravenous antibiotic 1 h beforehand, so as to minimize the risk of subsequent cholangitis. At the present time we use cefuroxime (1.5 g) or piperacillin (1 g).

Premedication is generally unnecessary. Children are best examined under general anaesthesia but adults are sedated with intravenous pethidine (50 mg) and midazolam (10 mg). One endoscopy assistant, equipped with a dedicated sucker, is solely responsible for looking after the patient throughout the procedure. Supplementary oxygen via nasal catheters is given to all the patients because respiratory depression can occur, and, since most of the procedure is conducted in semidarkness, physiological monitoring is essential. At the end of the examination the action of the sedative drugs is routinely reversed with naloxone (0.4 mg IV and 0.4 mg IM) and flumazenil (100 mcg IV).

Patients can eat and drink as soon as they have fully recovered from the sedation. They should stay in hospital for the night after the examination, since any complications usually develop within 12 h of the procedure.

ERCP is technically the most difficult of all the common flexible endoscopic procedures. Patience, persistence, and perseverance are absolutely essential. Furthermore, a different endoscopist on a different occasion may succeed after an initial failure.

The patient lies on the X-ray table on his left side with his left arm behind him. The right arm lies up by the head with the intravenous cannula readily accessible to the endoscopist. The right knee and hip are fully flexed to begin with, while the left leg lies straight. The pelvis then lies more or less vertical and leaves the abdomen free. From this position it is easy to turn the patient on to his face when the endoscope is in the duodenum, simply by straightening the right leg. The weight of the patient on the abdomen then maintains the position of the endoscope across the stomach and within the duodenum. By good fortune, the X-ray beam then usually lies in exactly the correct plane for radiology of both the bile and the pancreatic ducts. A diathermy plate is placed around the left thigh when required.

Insertion of the duodenoscope requires practice. The endoscopist starts by facing the patient and passing the endoscope into the stomach. The easiest way to enter the duodenum is to push the endoscope around the greater curve of the stomach and obtain a face-on view of the pylorus. The tip of the instrument is then elevated and the endoscope advanced. The pylorus disappears from view, like the setting sun, and the endoscope enters the duodenum. The tip of the endoscope is then rotated and locked to the right, the endoscopist twists his body and thus the whole instrument to the right, and the excess length is withdrawn while maintaining a view of the duodenal lumen by manipulating the up–down control. During this manoeuvre the ampulla initially recedes from view, but then the tip of the endoscope advances and comes to lie immediately opposite the ampulla.

In this ‘short’ position, the endoscope lies in almost a straight line from the mouth to the duodenum, with the ampulla about 65 cm from the incisor teeth (Fig. 4) 1378. This is the easiest position in which to cannulate the ampulla. Occasionally, the ‘long’ position is better, particularly for cannulation of the bile duct. Here the original position of the endoscope around the greater curve of the stomach is maintained and virtually the full length of the endoscope may lie within the patient (Fig. 5) 1379.

The key to successful cannulation is to position the endoscope correctly in relation to the ampulla. Until the endoscopist has a clear, face-on view of the ampulla (Fig. 6) 1380 attempts at cannulation are less likely to be successful. Duodenal peristalsis is a nuisance. Persistent insufflation of air may overcome the contractions, but the bowel can be paralysed with intravenous hyoscine- N-butylbromide (40 mg). Intravenous glucagon (1 mg) is an alternative.

Some manoeuvres may be helpful in achieving the correct position. Repeatedly pushing the endoscope further down the duodenum and then withdrawing, as described before, may help. Hooking the tip of the endoscope around the corner into the third part of the duodenum, at the same time as shortening the endoscope, is sometimes useful; and maintaining the patient on his left side occasionally helps. Patients who have had a previous Billroth II or Polya gastrectomy have to be intubated through the gastroenterostomy and backwards along the afferent loop. This may sometimes be easier with an end-viewing endoscope.

Orientation during endoscopy is always difficult. It is best not to try and orient oneself in relation to the position of the patient. It is helpful to imagine sitting in the second part of the duodenum facing the ampulla on the medial wall with your legs lying along the third part of the duodenum towards the duodenojejunal flexure. From this position the pancreatic duct runs almost horizontally away from the ampulla towards the spleen, while the bile duct passes vertically upwards behind the medial wall of the duodenum towards the liver which is above the endoscopist's head.

If the ampulla cannot be seen easily, the endoscopist should search the medial wall of the duodenum systematically, starting distally and working proximally. The endoscope is gently twisted from side to side and any suspicious mucosal folds should be elevated with a cannula. Diverticula require special attention and it may be necessary to place the tip of the endoscope actually within a diverticulum to find the ampulla. The papillary opening itself is usually obvious, except with ampullary tumours, when gentle probing with a cannula may help. The accessory ampulla can usually be seen 1 to 2 cm proximal and slightly to the right of the main ampulla. Once the main ampulla has been found, the tip of the cannula should be placed just within the orifice and a small bolus of contrast injected while screening the upper abdomen. If reflux of contrast into the duodenal lumen occurs, injection should stop immediately and the position of the cannula very gently adjusted, followed by a further injection of contrast. When a duct fills, suitable radiographic pictures are taken. The endoscopist should then withdraw the cannula a little, reposition the endoscope and the cannula, bearing in mind the comments in the paragraph above, and so fill the other duct (Figs. 7, 8) 1381,1382. Sometimes it is immediately possible to advance the cannula some way into a duct. If this happens, it is wise to withdraw the cannula a little immediately prior to injecting contrast, in order to be sure that the tip of the cannula is not jammed into a side branch of the pancreatic duct.

Although the examination is primarily radiological, the endoscopic findings should not be ignored. A formal examination of the oesophagus and stomach is better done on a separate occasion, but the first and second parts of the duodenum are always seen clearly. Benign ulceration, malignant infiltration, and overt cancer may all be visible. Distortion of the duodenal anatomy and stenosis of the lumen may also be appreciated and may, on occasion, make it difficult to position the endoscope. Observation is particularly important in consideration of a carcinoma of the ampulla because, initially, histology and cytology may not make a definitive diagnosis (Fig. 9) 1383. All biopsy specimens should be taken just before removing the endoscope, when any bleeding will not interfere with cannulating the ampulla. The ampulla itself has a variable appearance. The most common appearance is shown in Fig. 6 1380, where the ampulla appears as a nipple. Small mucosal fronds often pout out from the opening and there may be mild inflammation. Sometimes the ampulla is very flat and it is then often hard to find. A patulous, markedly inflamed, or oedematous appearance suggests either the recent passage of a stone or obstruction to lymphatic drainage by a tumour. Very rarely, there may be two separate orifices corresponding to the openings to the pancreatic and bile ducts.

If chronic pancreatitis is suspected, a plain abdominal radiograph must be taken first, as calcification is rapidly obscured by contrast. During the procedure the assistance of a radiologist is invaluable so that the endoscopist can concentrate on the endoscopic appearance while the radiologist can position the patient and the X-ray beam to provide the best radiological image. Any of the standard contrast media can be used, but the concentration is important. If the contrast is too dense, stones in the bile duct may be obscured, and if it is too dilute, there is a risk of overfilling the pancreatic duct. We use 30 per cent and 60 per cent W/V meglumine iothalamate.

Attention to detail during the examination is essential and requires careful co-ordination between the endoscopy assistant, the endoscopist, and the radiologist. The pancreatic duct must be filled right to the tail of the gland, with some filling of the side branches but without filling the acini and producing a parenchymogram, which will almost always lead to pancreatitis.

During cholangiography early screening may be helpful but it may also be deceptive. If a stone is present in the bile duct it may rise up in the duct as the heavy contrast is injected, but the lucent whirlpool which is created when a fine jet of contrast is injected into a dilated duct may mimic a stone. Certainly, no therapeutic procedure should be performed until the diagnosis of choledocholithiasis has been definitely established on an radiographic film (Fig. 10) 1384.

Ideally, contrast should outline both sides of a stricture because the diagnosis may rest almost entirely on the radiographic appearances. If this does not occur in the pancreatic duct, the proximal side-branches must be well filled, or even slightly overfilled, before an apparent narrowing or obstruction of the main duct can be accepted as significant. In the bile duct it may be necessary to jam the cannula deliberately into the bottom end of a stricture and then to inject contrast under a little pressure in order to obtain satisfactory pictures (Fig. 11) 1385.

Sometimes it is impossible to outline any duct. If there is a definite abnormality on ultrasound, it is usually justified to make a small precut in the tip of the ampulla, either with a needle knife (Fig. 12) 1386 or a sphincterotomy knife. Immediate cannulation of one or both ducts may then be possible, but it is often easier a few days later when there is some oedema of the ampulla. The use of this manoeuvre lessens with increasing experience.

The basic therapeutic manoeuvre is to divide the ampullary sphincter and gain access to the bile duct. The relaxed sphincterotomy knife (Fig. 2) 1376 is inserted into the bile duct and the position confirmed on the television screen. The knife is slowly withdrawn until about half the wire is visible outside the ampulla. It is then gently tightened by the endoscopy assistant and the endoscopist divides the sphincter by using short bursts of cutting diathermy current. Three manoeuvres will extend the cut upwards through the sphincter—elevating the bridge, elevating the tip of the endoscope, and tightening the wire. Only one of these manoeuvres is used at a time so that the sphincter is divided slowly and in a controlled way (Fig. 13) 1387. If they are combined, there is a real risk that the wire will cut too far too fast, and the larger the cut the greater the risk of haemorrhage and perforation. It is rarely necessary to enlarge a sphincterotomy beyond the transverse mucosal fold, which lies immediately above the ampulla, and it is normally sufficient if the fully tightened knife will come through the opening easily. A sphincterotomy needs only to be big enough to allow removal of the largest stone.

Stones are removed from the duct with a balloon or a basket (Fig. 14) 1388. Very large stones and stones which are wider than the diameter of the proximal bile duct are not easy to remove. Such stones can sometimes be reduced in size by the use of dissolving agents instilled down a nasobiliary drain (Fig. 15) 1389, crushing baskets, laser light, ultrasound, and extracorporeal shock waves. None of these is particularly successful. If the duct cannot be completely cleared, it may be appropriate to leave a nasobiliary drain or a stent in place to allow adequate drainage while further management is considered.

The other fundamental procedure is intubation of a duct with a flexible guidewire carried within a cannula. This is the basis for inserting any form of drainage into a duct, and is most often used to place a stent across an obstruction in the bile duct caused by a carcinoma of the pancreas (Fig. 16) 1390. The tip of the cannula is placed below the obstruction. The guide wire is manipulated across the stricture by careful co-ordination between the endoscopy assistant and the endoscopist, and is then followed by the cannula. The stent is ‘railroaded’ over the cannula and guidewire across the stricture. The proximal tip of the stent is placed well above the stricture, while the distal end lies just within the duodenum, so allowing bile to drain. Various diameters and lengths of stent are available, and the larger the diameter the longer they last. Our standard stent is 3.3 mm in outside diameter and 15 cm long. They last about 4 months. Recurrence of jaundice or cholangitis are signs that the stent needs changing, which is easy to do by removing the old stent with a basket or snare and inserting a new one using the original technique.

The ampulla can be found by experienced endoscopists in 98 to 99 per cent of patients. They will expect to cannulate both ducts in 90 per cent of patients, although the inexperienced find it slightly easier to cannulate the pancreatic duct. Most series report a 95 per cent success rate in performing a sphincterotomy and complete clearance of stones from the bile duct in nearly 9 out of 10 patients. Half the failures are because of large stones. Endoscopic intubation of the bile duct for malignant obstruction is rather less successful at present, with up to a quarter of attempts failing. Strictures high up and very low down in the bile duct are particularly difficult, but newer techniques, such as a combined percutaneous and endoscopic approach, may improve the results in the future.

The main complications are cholangitis, pancreatitis, and haemorrhage. The morbidity and mortality for a diagnostic ERCP are about 1 per cent and 0.1 per cent respectively, whereas after an endoscopic sphincterotomy the figures are 10 per cent and 1 per cent. There is a particular risk of haemorrhage after a sphincterotomy, and about half these patients will require a laparotomy. Less frequent problems are retroperitoneal perforation of the duodenal wall and impaction of a basket because a stone is trapped but is too large to remove.

Haemorrhage is usually apparent immediately, but pancreatitis becomes clinically obvious some hours later, and cholangitis can develop at any time. For these reasons we normally give any patient who has had a therapeutic procedure a second dose of antibiotics 12 h after treatment, and we also record the pulse and blood pressure half-hourly for 8 h afterwards.

Bickerstaff KI, Berry AR, Chapman RW, Britton BJ. Endoscopic sphincterotomy for bile duct stones. An institutional review. Ann R Coll Surg 1989; 71: 384–6.

Carr-Lock DL, Cotton PB. Biliary tract and pancreas. In: Miller RA, Wickham JEA, eds. Endoscopic surgery. Br Med Bull 1986; 42: 257–64.

Classen M, Demlin L. Endoskopische Sphinkterotomie der Papilla Vateri und Steinextraktion aus dem Ductus choledochus. Deutsch Med Wochenschr 1974; 99: 496–7.

Cotton PB. Progress report ERCP. Gut 1977; 18: 316–41.

Cotton PB, Williams CB. Practical gastrointestinal endoscopy. 3rd edn. Oxford: Blackwell Scientific Publications, 1990.

McCune WS, Shorb PE, Moscovitz H. Endoscopic cannulation of the ampulla of Vater: a preliminary report. Ann Surg 1968; 167: 752–6.

Shepherd HA, Royle G, Ross APR, Diba A, Arthur M, Colin-Jones D. Endoscopic biliary endoprosthesis in the palliation of malignant obstruction of the distal common bile duct; a randomized trial. Br J Surg 1988; 75: 1166–8.

Soehendra N, Reynders-Frederix V. Palliative bile duct drainage—a new method of introducing a transpapillary drain. Endoscopy 1980; 12: 8–11.