Oesophagogastroduodenoscopy is useful for the diagnosis and treatment of upper gastrointestinal diseases. Endoscopy is more expeditious, sensitive, and specific than are barium radiographic studies, especially for superficial conditions such as oesophagitis and gastritis and for investigating gastrointestinal bleeding. Biopsy specimens may be obtained during endoscopy, further increasing its diagnostic value. In addition, therapeutic manoeuvres are possible with oesophagogastroduodenoscopy, especially staunching upper gastrointestinal bleeding, placing gastrostomy tubes, removing foreign bodies, and treating benign and malignant oesophageal strictures. Endoscopic retrograde cholangiopancreatography (ERCP) is a special type of upper gastrointestinal endoscopy specifically designed to assess the pancreaticobiliary system. In addition to the exquisite diagnostic power of ERCP, therapeutic manoeuvres such as removal of biliary stones and placement of tubular endoprostheses for palliation of malignant obstruction are possible. ERCP is considered in Section 16.2.1. 110

As discussed in Section 27.3 195, surgeons should perform endoscopy. Especially in cases of gastric ulcer or upper gastrointestinal bleeding, if surgery becomes necessary the plan for operation greatly depends upon the specific findings at oesophagogastroduodenoscopy. This information should preferably be first-hand.

Opportunities for therapeutic upper gastrointestinal endoscopy are growing. Surgeon endoscopists are most likely to make impartial choices between open surgery and endoscopic treatment. By temperament and training they are best equipped to perform endoscopic surgery and to deal with the possible complications.

An upper gastrointestinal endoscope is a thin flexible instrument, usually 100 cm long. An image is transmitted from the tip by either a fibreoptic bundle to the lens on the head of the scope or by a tiny video camera on the tip to a television screen. The end can be deflected in a wide arc by manipulating two steering knobs on the endoscopic head. There is a tiny air insufflation channel to produce stomach distension for inspection. Through this channel automatic water lavage of the scope tip may also be carried out to keep debris from obscuring the view. Another larger channel (and sometimes two) is available for suction or for passing instruments (biopsy forceps, grasping forceps, cytology brush, snare, injection needle, cauterizing probe, laser fibre).

The patient must have an empty stomach; nothing by mouth for 6 h beforehand is usually sufficient. Pharyngeal anaesthesia to blunt the gag reflex is achieved with benzocaine or lidocaine spray or gargle. Usually, the patient is also given intravenous sedation (narcotic and benzodiazepine) and is monitored during the procedure with blood pressure determinations, electrocardiography, and pulse oximetry.

To begin oesophagogastroduodenoscopy, most endoscopists place the patient on his or her left side, hold the instrument in the left hand, and alternately manipulate the scope controls and shaft with the right hand (or have an assistant manipulate the endoscope shaft). It is better to place the instrument in a bracket or chest harness, liberating the left hand to hold the shaft and the right hand to control the tip deflection knobs and instruments in the biopsy channel. The ability to control tip and shaft movements simultaneously gives facile and versatile operation, especially for endoscopic surgery. After placing a bite block between the patient's teeth, the endoscope is inserted into his or her mouth. The tongue is followed down to its base where the upper larynx can be seen (Fig. 1) 1355. The oesophagus is best entered under direct vision; the scope is guided posterior to the arytenoid cartilages, where the view becomes obscured by the contracted cricopharyngeus muscle. While maintaining gentle pressure against the muscle with the scope, the patient is asked to swallow. This manoeuvre allows ready passage of the instrument into the upper oesophagus. If not, one should back up into the posterior pharynx, suction out any saliva with the scope, and try once more after the patient catches his breath. Keeping the gut lumen in the centre of the field of view (Fig. 2) 1356 the operator advances the scope distally with deflection of the tip and twisting of the shaft as necessary. Once in the stomach it is best to stay oriented by twisting the scope (usually counter-clockwise at this point) so that the lesser gastric curvature lies at 12 o'clock with respect to the endoscopic view. Prominent rugae are visible in the gastric body (Fig. 3) 1357 and fundus, whereas the antral mucosa is rather flat (Fig. 4) 1358. The scope is then passed into the duodenum. Sometimes a moment of pressure with the instrument tip is required before the pylorus (Fig. 5) 1359 relaxes enough to permit the scope to traverse it. Care must be taken not to slip too quickly through the bulb (Fig. 6) 1360; a proximal lesion (just beyond the pylorus) can be missed.

The second portion of the duodenum is subsequently entered. Sometimes the angle between the first and second portions can be acute; if so, passage is easier if one remembers that the duodenum turns not only inferiorly, but also posteriorly. Circular mucosal folds differentiate the second portion of the duodenum from the smooth bulb (Fig. 7) 1361. Alternating insertion and withdrawal movements advance the endoscope into the second and third portions of the duodenum. The papilla of Vater is rarely seen without a side-viewing (ERCP) scope. Finally, the scope is withdrawn back into the stomach, and the tip is deflected 180° for ‘retroflexed’ inspection (Fig. 8) 1362. This gives the best view of the gastric incisura, the lesser curvature, and the fundus.

Diagnostic osophagogastroduodenoscopy is especially valuable for evaluating upper gastrointestinal tumours, strictures, ulcers, varices, and mucosal changes such as oesophagitis and gastritis. It is less useful for judging functional or motility disorders or extraluminal lesions. Specifically, the major indications for diagnostic upper gastrointestinal endoscopy are:

(1) persistent upper abdominal pain or distress, especially if it is associated with symptoms or signs suggestive of serious disease (anorexia, weight loss, anaemia);

(2)persistent symptoms of gastro-oesophageal reflux despite treatment;

(3)swallowing difficulties;

(4)persistent vomiting of unknown cause;

(5)surveillance for upper gastrointestinal malignancy in high-risk patients;

(6)evaluation of upper gastrointestinal bleeding of unexplained iron deficiency anaemia;

(7)evaluation of ulcers, strictures, and tumours found by a barium meal examination.

The most important indications for therapeutic oesophagogastroduodenoscopy are:

(1)cauterization of injection of bleeding peptic ulcers (and occasionally vascular malformations);

(2)injection sclerosis of oesophageal varices that have bled;

(3)removal of foreign bodies or bezoars;

(4)removal of gastric polyps;

(5)dilatation of oesophageal strictures;

(6)palliative treatment of malignant upper gastrointestinal obstruction by dilatation, laser fulguration, or intubation;

(7)placement of gastrostomy tubes.

There are contraindications to oesophagogastroduodenoscopy. Endoscopy should be avoided if there is a possibility of a perforated viscus. If the stomach is not empty, little useful inspection is possible, and the risk of vomiting and aspiration pneumonia is great. A poor airway contraindicates upper gastrointestinal endoscopy. Respiratory arrest may result if there is a marginal passage (for example, pharyngeal tumour) combined with intravenous sedation, as well as partial occlusion and oedema induced by the endoscope. Endoscopy surgery should not be carried out if the patient is anticoagulated or has a coagulopathy.

Complications due to diagnostic oesophagogastroduodenoscopy are uncommon. Aspiration pneumonia may occur because of the combination of pharyngeal anaesthesia and instrumentation, especially in obtunded patients, or if there is active upper gastrointestinal bleeding. Although adverse cardiorespiratory events are rare, they may occur if the patient is frail or if sedation is excessive. Perforation and bleeding are infrequent unless endoscopic surgery is done. As is the case for colonoscopy, drug reactions, vasovagal reflex, bacteraemia, and superficial phlebitis are other less serious potential problems.

Diagnostic upper gastrointestinal endoscopy is essential in the management of numerous lesions. It is the best way to evaluate the type and severity of oesophagitis (which may not appear on barium examination). A mass, a stricture, or an ulcer of the oesophagus or stomach require endoscopy and, usually removal of a biopsy specimen to determine whether malignancy is responsible. Cytological specimens obtained by brush or needle aspiration can augment simple biopsy. If gastric or oesophageal surgery becomes necessary, the type of incision as well as the extent of resection greatly depends upon the endoscopic assessment. Oesophagogastroduodenoscopy is the best first step in evaluating upper gastrointestinal bleeding. Contrast radiographic studies should be avoided in that situation because they often miss the lesion and because the barium hinders subsequent endoscopy or angiography. Endoscopy also reveals features of a lesion that predict the risk of rebleeding. Whenever small-bowel mucosal biopsies are needed, upper gastrointestinal endoscopy (perhaps with an extra-long scope) is the best way to obtain them. As discussed below, in a few specific circumstances oesophagogastroduodenoscopy is required for periodic surveillance for upper gastrointestinal cancers. Endoscopic ultrasound is a relatively new technique, which employs a small sonar instrument on the tip of a modified scope to provide a cross-sectional image of the oesophageal or stomach wall as well as adjacent viscera. It may be valuable for staging gastro-oesophageal tumours, assessing submucosal masses, and evaluating pancreatic lesions.

Oesophagogastroduodenoscopy is not appropriate for routine screening for upper gastrointestinal cancer, but it is worthwhile for patients at increased risk. Barrett's oesophagus and achalasia are premalignant conditions that warrant periodic (probably yearly) endoscopy. So is Plummer–Vinson syndrome, although it is rare. Gastric adenomatous polyps are unusual but, as in the colon, they may contain or transform to malignancy; they should be removed. Similarly, polyps are associated with a tendency to form metachronous lesions; periodic follow-up oesophagogastroduodenoscopy is indicated. Familial polyposis coli is associated with an excess risk of gastroduodenal neoplasms. After colectomy patients with familial polyposis coli require upper surveillance endoscopy. Atrophic gastritis, a postgastrectomy stomach remnant, oropharyngeal cancer, and lye-induced oesophageal stricture are other conditions that somewhat predispose to gastro-oesophageal cancer, but probably not enough to justify periodic oesophagogastroduodenoscopy.

Barrett's oesophagus is the most important upper gastrointestinal premalignant lesion. In this acquired condition, metaplastic columnar epithelium replaces the normal squamous epithelium in the distal oesophagus. It is strongly associated with gastro-oesophageal reflux, and there may be a concomitant hiatus hernia, oesophagitis, ulcer, or stricture. Nevertheless, reflux symptoms are sometimes mild. The typical endoscopic appearance is of velvety red mucosa extending circumferentially or in linear streaks proximally from the gastro-oesophageal junction. It may be difficult to differentiate a small hiatus hernia from Barrett's oesophagus, but there are usually longitudinal gastric folds in a hiatus hernia whereas the latter is a smooth, tubular structure above the most proximal gastric folds. Usually the metaplastic mucosa is readily discernible, but occasionally biopsy specimens are necessary for discrimination. Toluidine blue or Lugol's iodine sprayed on the mucosa enhances the detection of the columnar epithelium. Adenocarcinoma may arise in the Barrett's mucosa. The prevalence of cancer at initial diagnosis of Barrett's oesophagus is about 10 per cent. There is at least a forty-fold increased risk of developing distal oesophageal adenocarcinoma is patients with Barrett's oesophagus, about 1 to 2 per cent per year. These patients, including patients who have undergone successful antireflux surgery, should undergo yearly screening oesophagogastroduodenoscopy with random biopsies of the Barrett's mucosa. If persistent severe mucosal dysplasia is found, oesophagectomy should be considered.

Gastric polypectomy is different from colonic polypectomy. The stomach wall is thick and vascular. Removal of gastric polyps carries a greater risk of bleeding, but a lesser risk of perforation than in the large bowel. Therefore, more electrocoagulating energy must be applied when snaring gastric lesions. If the endoscope is withdrawn with the polyp suctioned on to its tip, care must be taken that the polyp is not pulled off by the cricopharyngeus muscle and aspirated into the trachea. An endoscope overtube obviates the problem. Because polypectomy leaves an iatrogenic stomach ulcer, a period of antacid therapy is advisable afterwards.

Endoscopy is an important early manoeuvre in the management of upper gastrointestinal bleeding. An experienced endoscopist can identify the bleeding source about 90 per cent of the time. A clear demonstration of the type and location of the responsible lesion helps guide subsequent treatment. This is especially important if urgent surgery becomes necessary; the surgeon must know where to focus the operation. The optimal arrangement is the surgeon and endoscopist as one person. Beyond diagnosis, however, treatment of a bleeding site is possible through the endoscope. In as much as upper gastrointestinal bleeding ceases spontaneously in about 80 per cent of cases, therapeutic endoscopy benefits primarily the small remaining group of ‘high-risk’ patients.

Both clinical and endoscopic features determine risk. Those most likely to die from bleeding are elderly patients, patients with other major medical problems, patients who begin bleeding while in hospital, and patients who bleed heavily (large transfusion requirement, haematemesis or haematochezia, hypotension, recurrent bleeding while in the hospital, need for emergency surgery).

Endoscopic findings also predict outcome. Varices or cancer correlate with a high mortality. For peptic ulcers, a visible vessel indicates up to a 50 per cent likelihood of rebleeding, whereas two other ‘stigmata of recent haemorrhage’ (adherent clot or active oozing) are associated with a 20 to 25 per cent risk of recurrent haemorrhage. Large ulcers are probably more likely to rebleed than are small ones, as are ulcers on the posterior wall of the duodenal bulb (overlying the gastroduodenal artery). Oesophageal varices, large vessels, or varices displaying overlying red spots are especially prone to bleed or rebleed. Therefore oesophagogastroduodenoscopy is worthwhile in all upper gastrointestinal bleeders, not only to identify the haemorrhage source, but also to help assess the danger of rebleeding and death. Endoscopic treatment can then be selected only for patients with high-risk features. In these candidates therapeutic endoscopy is of greatest benefit.

Numerous endoscopic techniques for haemostasis make use of instruments passed down the biopsy channel of the scope. The most common causes for upper gastrointestinal bleeding are peptic ulcers (Figs. 9–11) 1363,1364,1365 and oesophageal varices. For varices, injection sclerotherapy (as discussed below) is clearly the best choice. For ulcers, the available effective approaches are thermal coagulation with multipolar electrocautery (‘BICAP’), a heater probe, or a neodymium: yttrium aluminium garnet (Nd: YAG) laser, as well as injection therapy with absolute ethanol (or perhaps adrenaline, hypertonic saline, or a mixture). Monopolar electrocoagulation and argon laser photocoagulation have largely been replaced by the other modalities. Topical agents, such as clotting factors, adrenaline, microcrystalline collagen, and cyanoacrylate glue, are ineffective. No endoscopic method is free of problems. Massive haemorrhage and blood clots may obscure the view, ulcers in a deformed duodenum or elsewhere may be inaccessible, and much endoscopic skill is required. The patient must be able to co-operate with the examination, as well to remain haemodynamically stable enough to permit waiting, if necessary, for the examination and able to receive intravenous sedation. Bleeding may recur. Furthermore, the patient is at risk for aspiration pneumonia, gastrointestinal perforation, and even induction or exacerbation of bleeding due to the procedure.

The multipolar electrocoagulation probe is surrounded by three pairs of electrodes. Current is conducted between any pair of electrodes, and the depth of tissue injury is minimal. A bleeding vessel can be compressed with the probe before heat delivery (coaptive coagulation), which is superior to cauterization alone, especially for larger vessels (over 0.5 mm). Tamponade before coagulation also lets the operator know that the instrument tip is correctly placed. A built-in lavage channel is handy. Electrodes cover both the end and the sides of the probe tip so it can be applied tangentially when an en face approach to the vessel is impossible. The multipolar electrode delivers enough energy to coagulate, but not to vaporize tissue (which might induce bleeding). The equipment is portable and, compared to a laser, relatively cheap. Other attachments can be used with the unit to treat obstructing gastrointestinal tumours and haemorrhoids. Tissue adherence to the endoscope may be a problem, however; when the probe is withdrawn the coagulum may come also, with resumed bleeding.

The heater probe is a tiny Teflon-coated electrical resistance heater capable of delivering a precise amount of energy. It also incorporates an irrigating channel. Its advantages are the same as for the multipolar electrocoagulator, although it works a bit more slowly. Tissue adherence can also be a problem.

Laser photocoagulation takes advantage of the high power and precise control of laser light, which can be guided down a flexible quartz fibre in an endoscope channel. When the light is absorbed by tissue, it is converted to heat; no tissue contact is required. Adherence is therefore not a problem. A coaxial CO&sub2; jet clears the target area of blood and debris. The amount of energy delivered can be precisely determined but light (and heat) penetration is a little deeper than with the heater probe or multipolar unit. This effect can be an advantage when dealing with larger and deeper blood vessels, but might increase the perforation risk. Energy sufficient to vaporize tissue may be delivered, inducing bleeding. Lasers are perhaps the most technically difficult of the instruments to master and apply, and a direct en face approach to the lesion is required. Laser machines are not very portable, and are very expensive. The cost is somewhat mitigated by their versatility; Nd:YAG lasers can also be used for tissue ablation in the gastrointestinal tract and lung. They are ideal for the rapid treatment of multiple superficial mucosal vascular lesions (radiation telangiectasias (Figs. 12, 13) 1366,1367, Osler–Weber–Rendu disease).

Injection therapy (Figs. 14, 15) 1368,1369 makes use of a thin flexible tube with a short retractable needle at its tip. This is the same item that is used for injection sclerosis of oesophageal varices (see Fig. 17 1371) but absolute alcohol, adrenaline, or hypertonic saline is usually employed for ulcer treatment. It is the simplest, cheapest, and most portable of all the treatment options. If precisely applied, it can bring about tissue coagulation with little necrosis, which may minimize the risk of perforation or induced bleeding. Nevertheless, it has no channel for simultaneous irrigation, and cannot easily be applied tangentially.

All four haemostatic techniques seem to be useful and effective. Nevertheless, results from clinical trials are often conflicting and inconclusive because of insufficient data or confounding variables. Few studies compare directly the various techniques, and the results from individual tests are roughly comparable. At present, a selection may be made based on the advantages and disadvantages enumerated above, and personal preference. In experienced centres, for patients with clinically severe bleeding from peptic ulcers displaying ‘stigmata of recent haemorrhage’, any one of these techniques will result in initial haemostasis in 80 to 100 per cent of patients, with rebleeding in 10 to 20 per cent. Reapplication produces an ultimate control rate of approximately 80 to 90 per cent, with the remainder of patients proceeding to emergency surgery (about 10 per cent) or death. Perforation occurs in 0 to 2 per cent, and uncontrollable bleeding is rarely induced. Trials comparing medical treatment alone to endoscopic treatment usually demonstrate superior haemostasis with the endoscopic method (a reduced proportion of patients with continued or recurrent bleeding, and a decreased number of transfusions). Often, but not always, the need for emergency surgery is decreased, and mortality improved. Even if surgery must eventually be performed, there is benefit from temporary haemostasis by endoscopic therapy; the patient can be resuscitated, stabilized, and operated upon electively rather than as an emergency.

A successful technique for endoscopic therapy begins with aggressive stomach lavage with a large orogastric tube to remove obscuring blood and clots. Changes in the patient's position can also shift blood out of the way. A double-channel endoscope is helpful because it allows simultaneous suction and use of the instrument. The likelihood of success is greatest when a clear and complete view (preferably en face) of the lesion is obtained before starting treatment. A heater probe or multipolar coagulator requires exact tamponade of the bleeding point with the instrument tip. Laser photocoagulation and injection therapy (0.2 ml increments of ethanol) begin in the immediately surrounding tissue to create a rim of oedema or coagulation to slow inflow, followed by targeting the actual bleeder. Treatment must be expeditious, before induced duodenal oedema obscures the view, and before the patient becomes excessively distended with air.

Endoscopic treatment of non-variceal upper gastrointestinal bleeding is feasible and safe, and reasonably (although not dramatically) effective. The most promising modalities are multipolar electrocoagulation, the heater probe, Nd:YAG laser photocoagulation, and injection therapy with absolute alcohol. Nevertheless, if the bleeding is truly torrential, endoscopic treatment is unlikely to be successful, and the patient should be taken directly to surgery. A posterior duodenal bulb bleeder or visible vessel may be the gastroduodenal artery and is also best treated surgically to avoid massive induced bleeding. Patients with active but not massive haemorrhage, and those with adherent clots or visible vessels, are ideal candidates for endoscopic therapy, especially if they are poor operative risks. Patients without stigmata of recent haemorrhage are not worth treating, since they rarely rebleed.

Bleeding oesophageal varices (Fig. 16) 1370 are difficult to treat and often fatal. Medical therapy, including transfusion, intravenous vasopressin, and balloon compression (Sengstaken–Blakemore tube) staunches the bleeding only some of the time. At least half the patients start bleeding again when the tube is deflated or when the infusion is stopped. Even without immediate rebleeding, there is a substantial likelihood of later recurrent haemorrhage and death, especially if liver function is poor. Urgent portasystemic shunting reliably cures the bleeding, but at the cost of a high morbidity and mortality.

Endoscopic injection sclerosis of varices is an attractive alternative. By means of a retractable needle inside a thin plastic tube (Fig. 17) 1371, which is passed down the biopsy channel of the scope, a small amount of sclerosing solution can be injected in and around the varices (Fig. 18) 1372. The injection causes thrombosis within the vein; inflammation and some necrosis of the endothelium, submucosa, and mucosa; and subsequent fibrosis and shrinkage of the varix. Large varices, and varices displaying overlying ‘red spots’ (red wales) are especially prone to bleed or to rebleed. Endoscopically determined variceal pressure (which correlates with vein wall tension) also predicts a tendency to bleed. Most important, the degree of cirrhosis or liver dysfunction correlates with variceal haemorrhage risk.

Rates of survival, rebleeding, encephalopathy, and other complications are difficult to study in these heterogeneous and infirm patients. It has not been proved that survival after shunt surgery is prolonged compared to medical treatment; death from hepatic failure merely replaces death from haemorrhage. The stay in hospital for emergency surgery is typically long, complicated, and costly; encephalopathy is a considerable risk for survivors. In contrast, endoscopic injection sclerosis effectively reduces the frequency of rebleeding episodes and improves survival compared with medical therapy. Sclerotherapy also seems to result in improved survival and rebleeding rates compared with shunted historical controls. Nevertheless, little information is available to compare directly endoscopic therapy with surgery. The measure is made more difficult by the numerous variations in patient characteristics, as well as types and timing of surgical treatment. The few available comparisons of sclerosis with surgery do not clearly favour one over the other. Although acute and chronic control of bleeding is better with surgery, at least initial morbidity and mortality rates, as well as costs, are much lower with endoscopic therapy. With time, however, sclerosis patients probably catch up to their surgically treated partners, as rebleeding episodes gradually increase statistics for morbidity, mortality, and cost. In the final analysis, the outcome is determined more by liver function than by the type of treatment; the choice is whether to pay a price all at once or in instalments. It is reasonable to try the more benign endoscopic treatment first and to operate upon the failures. The primary disadvantage of injection sclerotherapy is that repeated treatments are required to obliterate the varices, and lifelong follow-up is necessary. A compliant patient is helpful. Until varices are eradicated, there is also the inconvenience of having occasionally to deal with recurrent bleeding episodes.

For active bleeders, management begins with resuscitation, and the sclerosis procedure is best done with the patient monitored in an intensive care unit or operating room. Administration of blood, clotting factors (fresh frozen plasma), and vitamin K may suffice to stem the haemorrhage. If bleeding persists, intravenous vasopressin or perhaps somatostatin and then balloon compression are appropriate manoeuvres to stabilize the patient's condition before gastroscopy (which can be done immediately or 6–24 h later). The rate of bleeding can at least be transiently checked in 80 to 90 per cent of patients, and the improved view greatly facilitates the endoscopic injection. Nevertheless, it is also reasonable, and perhaps more efficient, to skip medical manoeuvres and to carry out immediate endoscopic sclerosis, if equipment and personnel can be marshalled quickly. Encephalopathy, instrumentation, and iatrogenic sedation and pharyngeal anaesthesia predispose the patient to a disastrous aspiration of blood or vomitus; therefore, endotracheal intubation is wise. Gastric lavage with a large orogastric tube is necessary to clear the stomach of clots before the scope is passed. Flexible fibreoptic endoscopy has supplanted rigid oesophagoscopy for sclerotherapy, since it is easier, safer, and does not require general anaesthesia. Unless varices are clearly identified as the bleeding source, a brief oesophagogastroduodenoscopy before starting injection sclerosis is prudent. Cirrhotic patients sometimes bleed from peptic ulcers, gastritis, or from Mallory–Weiss tears. If non-bleeding varices are present, and there are no other important upper gastrointestinal lesions, the patient has probably bled from the varices and requires treatment accordingly. Gastric varices may also be seen. Often they are simply inferior extensions of oesophageal varices (being within 2–3 cm of the squamocolumnar junction). These usually disappear when the oesophageal varices are obliterated. More distant (fundal) gastric varices are unusual but problematic sources of gastrointestinal bleeding. Diagnosis can be difficult because they may resemble rugal folds or may flatten when the stomach is insufflated for endoscopy. Injection sclerosis of these ‘true’ gastric varices is usually fruitless; rebleeding often occurs, and there is probably an excess risk of perforation and ulceration.

The scope is with drawn into the distal oesophagus for injection. A small balloon placed around and just proximal to the endoscope tip is optional, but when inflated may produce relative stasis within the varices to hold the sclerosant in place until thrombosis occurs. Although blood flows in various directions within a varix, it most often moves cephalad. The balloon is also handy for tamponade to prevent back-bleeding from the injection needle hole. Injections are best aimed as low in the squamous mucosa over a varix as possible, since bleeding varices are always near the gastro-oesophageal junction. Sclerosant (1–3 ml) is injected into the lumen of three to five oesophageal varices. Intravariceal rather than submucosal paravariceal injections are the most widely advocated and probably minimize mucosal slough. Further injections at 3- to 4-cm intervals from distal to proximal oesophagus are optional. After withdrawing the injection needle, the needle hole is momentarily occluded with the needle sheath, the endoscope tip, or the balloon on the scope. Care must be taken not to overdistend the intestine with air during a long and difficult sclerosis procedure, especially in an unintubated patient; since it makes breathing more difficult. A nasogastric tube is preferably omitted after sclerosis to avoid mucosal erosion and rebleeding. Injection is repeated at 48 hours and at 1 week; the patient can then be discharged if stable.

Subsequent injection procedures can usually be done in an outpatient setting, unless active bleeding recurs. Since variceal size correlates positively with bleeding risk, the goal is to continue periodic injections until the varices are gone. After that, periodic surveillance endoscopy is wise, to check for recurrent varix formation. A reasonable schedule for follow-up sclerosis after an episode of acute bleeding is at 48 h, 1 week, every 4 to 8 weeks until the varices have shrunk, and then as necessary every 6 to 12 months for life. The sequence is restarted if the patient rebleeds.

Endoscopists differ about many technical details of injection sclerosis. Little objective information is available comparing the possible regimens, especially because so many variables can affect the outcome of sclerotherapy. The most common sclerosant solutions used in the United States are 5 per cent sodium morrhuate and 0.5 to 3.0 per cent sodium tetradecyl sulphate. Elsewhere, physicians also use 5 per cent ethanolamine oleate, 1 to 3 per cent polidocanol, and 50 to 95 per cent ethanol. Thrombin, a cephalosporin antibiotic, and 20 to 50 per cent dextrose are sometimes mixed in. Many endoscopists omit the balloon on the scope tip. Some workers use a flexible overtube with the fibreoptic endoscope; the tube has a distal slot to allow isolated prolapse of the target varix and compression of the remainder. Injections can be paravariceal, but most workers aim for an intravariceal location. The optimum number of injections and overall volume of sclerosant to administer per session is unknown. Similarly, the time interval between follow-up injections varies among sclerotherapists from a week to a few months. Numerous, frequent, and voluminous injections obliterate varices more rapidly, making rebleeding less likely. But intensive treatment is traded against an increased risk of local complications (ulceration, perforation, stricture) (Fig. 19) 1373. Laser photocoagulation of varices is of little benefit because of recurrent haemorrhage. Endoscopic elastic banding of varices has had reasonable early results (Fig. 20) 1374.

Although injection sclerosis is the safest non-medical treatment for bleeding varices, complications do occur. The sclerosis technique, the skill of the endoscopist, and the clinical state of the patient influence the morbidity rate. Transient substernal chest pain (probably from oesophageal spasm) odynophagia, fever, and pleural effusions are common in the first 24 to 72 h after injection, and are almost always inconsequential. Oesophageal mucosal ulceration due to tissue necrosis from the sclerosant is not uncommon, especially if follow-up endoscopies take place at short intervals. The ulcers usually heal spontaneously. Rarely, recurrent, bleeding issues from the ulcer. If an ulcer is seen at follow-up endoscopy, adjacent injections should be avoided. In a patient whose variceal bleeding has ceased, instrumentation can induce recurrent haemorrhage that might not be controllable endoscopically—fortunately an unusual problem. Instrumental acute oesophageal perforation does not occur, except when rigid oesophagoscopes are used; delayed necrosis and perforation with mediastinitis happens in 0 to 2 per cent. Aspiration pneumonia occurs in 2 to 10 per cent of patients, especially when sclerotherapy is undertaken during active bleeding. Although some sclerosant reaches the pulmonary circulation, respiratory function does not suffer. Even when thrombin is mixed with the sclerosant, no subclinical or overt systemic clotting abnormality results. Symptomatic distal oesophageal stricture develops in 2 to 20 per cent, especially when sclerotherapy is frequent and prolonged. Strictures usually respond well to dilatation.

Results of sclerotherapy are good in experienced centres. Active bleeding can be stopped 80 to 95 per cent of the time. About 20 per cent of patients rebleed during the initial period in hospital, but sclerotherapy can be repeated, giving a final acute control rate of 85 to 90 per cent. There is a 20 to 30 per cent short-term or ‘hospital’ mortality, usually from hepatic failure rather than bleeding. Endoscopic sclerotherapy is useful not only for acute haemostasis, but also for elective, long-term management of varices. Repeated injections, usually in an outpatient setting, gradually eradicate the lesions. Within the first 1 to 2 years of such a regimen, 20 to 30 per cent of patients rebleed; reinjection, however, ultimately controls the haemorrhage in 80 to 90 per cent of patients. Rebleeding most often appears during the first few months of treatment, before varices have become obliterated. Compliant patients do much better than average; those who faithfully attend follow-up endoscopy sessions rebleed less than 10 per cent of the time (these patients are also more likely to follow advice to stop drinking alcohol). Overall survival is about 50 per cent; again, death usually results from liver failure, not from bleeding. Hepatic function is the best predictor of death or rebleeding. Prophylactic endoscopic sclerosis of varices that have never bled is not worthwhile.

Endoscopic injection sclerosis is effective and safe, and is indicated for any patient with bleeding oesophageal varices. Once the haemorrhage has been staunched, the patient can recover, and decisions can be made about definitive elective treatment. Depending upon patient and physician preference, a shunt or continued sclerosis may be selected. Most cases can be nicely treated as outpatients with periodic injection sclerosis, but a non-compliant patient or the presence of gastric varices argues for surgery. Liver transplantation is now an alternative for selected cirrhotic patients. Episodes of repeat haemorrhage most often occur within the first few months of initiating sclerotherapy, before varices are obliterated. Occasionally, if rebleeding is frequent, or if the varices are not shrinking despite the injections, or if the patient is unco-operative, sclerosis should be abandoned in favour of surgery. Endoscopically uncontrollable bleeding (recurrent variceal haemorrhage despite two contiguous emergency injection treatments) leaves no alternative but shunt or oesophageal transection, but such patients usually have poor liver function and do poorly no matter what treatment is given them.

Ingested foreign bodies usually occur in children. The remainder appear in adults who are most often edentulous, prisoners, or psychiatric patients. Fortunately, 80 per cent of foreign bodies pass spontaneously through the gastrointestinal tract. Nevertheless, if the object seems unlikely to pass, or fails to pass, or would be dangerous to allow to pass, it should be extracted. Fibreoptic endoscopy is effective and safe for this purpose; under most circumstances there is no longer any need for rigid oesophagoscopy and general anaesthesia. The success rate for foreign body extraction with a fibreoptic upper gastrointestinal scope is 90 to 95 per cent.

Decisions about management depend upon the type of foreign body, and its location in the body. Plain radiographs often provide the information. For oesophageal foreign bodies, lateral views are important, to be sure that the object is not in the trachea instead. In children, or anyone who cannot give an accurate history, radiographs encompassing the gut from mouth to anus are necessary to display the number and types of items present. Barium radiographic studies, however, should not be done, since the barium will later obscure the endoscopic view. If the object is in the cricopharyngeal area or oesophagus (foreign body or meat bolus), extraction should be carried out as soon as possible, to avoid possible aspiration or perforation. Removal is especially urgent if the object is sharp (such as a piece of bone) or if it is a miniature disc battery (serious corrosive damage can develop within hours). Attempted digestion of an impacted meat bolus with papain (meat tenderizer) is unwise because it can worsen oesophagitis (most of these patients have strictures), and it is dangerous if aspirated. An item in the stomach can often be left to pass spontaneously unless it is large or sharp. Passage usually occurs within 3 days, although the patient can be observed (with periodic radiographs) for up to 2 weeks. Surgery is required if the foreign body fails to progress through the gastrointestinal tract, or if signs of inflammation develop. If the object is thicker than about 2 cm or longer than about 5 cm, it often will not exit from the stomach; endoscopic removal is wise. Sharp, irregular, or pointed objects such as toothpicks, chicken bones, razor blades, or dental bridges should be extracted. If a sharp or pointed item has passed through the stomach into the small bowel, the patient should be admitted to the hospital for observation and daily radiographs. Laparotomy is necessary if the foreign body does not progress over a few days, or if signs of inflammation appear. If drugs in packets (cocaine, heroin) have been ingested for smuggling, they should not be endoscopically removed because packet rupture can cause rapid death from drug overdose. Laparotomy is necessary instead. Caustic (for example, lye or acid) ingestion should be promptly evaluated endoscopically. Removal is impossible of course, but it is helpful to assess the degree and extent of damage to the oesophagus and stomach. The scope should be advanced only to the most proximal level of major damage.

The technique of endoscopic removal begins with examination of an object similar to the one swallowed, if possible, in order to plan and practise the best way to grasp it. While making preparations for endoscopy, the patient should lie on his left side to discourage passage of the foreign body out of the stomach. The patient's airway is the most important consideration during the procedure. If during the process of removal the object is lost at the cricopharyngeus muscle, it may occlude the airway or be aspirated into the trachea. Concern is especially great in children, or any unco-operative patient. Items that are large (such as a coin in a child), irregular, or pointed, or smooth and difficult to grasp are also risky. Placing the patient in Trendelenburg position during withdrawal discourages aspiration. An overtube fitted on the outside of the endoscope is useful; once the endoscope is in place the tube can be slid down over the scope until its lower end is on the stomach or distal oesophagus. The endoscope and foreign body can be withdrawn (with or without simultaneous removal of the overtube) so that the cricopharyngeus cannot dislodge the object. The overtube also protects the oesophagus from laceration if the item is sharp, and can make multiple reinsertions easy (as for meat in the oesophagus, or multiple foreign bodies). If there is any doubt about maintaining an airway, the safest course is to do the endoscopy under general anaesthesia with orotracheal intubation.

If the object is at the level of the posterior pharynx or cricopharyngeus, removal is usually possible with an open rigid laryngoscope (anaesthesia type) and a plain forceps or clamp. For any other location, the flexible endoscope is needed. Extraction can be accomplished with an endoscopic snare, endoscopic forceps (alligator or rat-tooth), dormia basket, balloon catheter, or three-pronged polyp retriever. Hollow objects (such as rings) may be removed after passing a heavy thread through them with the scope instruments. A sharp or irregular object (such as an open safety pin) may need to be repositioned so that the pointed end is trailing when it is withdrawn, or so that it can be pulled into an overtube. Sometimes it is best to push an object from the oesophagus into the stomach, so that it can be turned. It is preferable to withdraw a pointed object and the endoscope through an overtube. If manipulation in that way is impossible, the next best manoeuvre is to at least pull the sharp end into the overtube, and remove the scope, overtube, and foreign body all together. Meat stuck in the oesophagus can usually be readily extracted with a snare. Sometimes it can simply be pushed into the stomach (gently). Most patients with impacted meat have an oesophageal stricture, which may not have been recognized previously. Gastric bezoars may be broken up with a snare, biopsy forceps, or irrigation. A lesion may be present that prevents normal gastric emptying.

Oesophagogastroduodenoscopy should be carried out on all patients with oesophageal strictures. Inspection and biopsy are necessary to be certain that cancer is not responsible. The oesophagus and stomach can be evaluated for associated lesions (such as oesophagitis, hiatus hernia, or peptic ulcer), which may not be apparent on upper gastrointestinal radiography. Beyond diagnosis, dilatation of oesophageal strictures is possible with the aid of endoscopy. Although fibrotic strictures from gastro-oesophageal reflux are the most common lesions treated, bougienage can also be applied for the dilatation of anastomotic strictures, malignant strictures, Schatzki rings, and, occasionally, caustic strictures.

Three types of dilating instruments are available: mercury-filled rubber bougies, guidewire dilators, and inflatable dilators that are passed through the biopsy channel of the endoscope. Mercury-filled dilators (Maloney, Hurst) are passed blindly (or occasionally under fluoroscopy) and are simple to use. No simultaneous endoscopy is required. They are appropriate for mild to moderately narrow strictures that are relatively short, smooth, and straight. For tight, long, or irregular strictures, or strictures associated with a diverticulum, there is a greater risk of instrumental perforation, so wire-guided dilators are preferable. Eder–Puestow bougies are metal, ‘olive tipped’ graduated instruments. They are effective, but have largely been replaced by long, tapered, plastic dilators (Savary–Gillard; American). Inflatable ‘through the scope’ dilators are the latest addition to the array. These are passed down the endoscope channel and through the narrowed area under direct endoscopic view (no guidewire), and are expanded. Like guidewire bougies, they are appropriate for complex strictures, but are somewhat more troublesome to use. They are also expensive and have a limited lifespan. Their best application is in regions not accessible to other types of dilators, such as the pylorus, bile duct, or colon. Dilators are often sized in even-number increments of the French scale; each unit approximates 0.44 mm diameter (for example, 18 French = 6 mm diameter).

The procedure for dilatation begins with diagnostic upper gastrointestinal endoscopy. Rubber mercury-filled dilators are thereafter passed sequentially, with the patient sitting up so that gravity aids the process. For guidewire dilators, the wire is passed down the channel of the endoscope and through the stricture. The scope is then withdrawn, leaving the guidewire in place for the dilators to be passed over it. The goal is to position the wire completely through the stricture, with its end lying smoothly along the greater curve of the stomach (but not into the pylorus). Care must be taken that the wire does not perforate the oesophagus, curl above the stricture or in a hiatus hernia, or traverse only a portion of the stricture. The best assurance of proper placement is to first pass the endoscope (often the paediatric instrument) completely through the stricture. If that is not possible, fluoroscopic guidance is advisable (unless the stricture is short and straight). Indeed, whenever there is doubt about positioning a guidewire, or any difficulty with dilatation, fluoroscopy should be employed. Generally, in a single session only two or three dilator sizes (that is, 4–6 French units) should be passed beyond the first dilator that meets resistance. When passing wire-guided bougies, attention must be paid to the wire so that it does not become displaced.

Although many strictures recur, symptomatic relief after dilatation of benign strictures can be achieved in at least 80 per cent of patients. Dysphagia is usually relieved when dilatation to 40 to 45 French units is achieved (13–15 mm diameter). Symptomatic success is sometimes better than the objective change in stricture diameter. Perforation is the most important complication, and occurs in about 0.5 per cent of procedures. Strictures due to lye ingestion are especially fragile and prone to perforation during dilatation. Aspiration pneumonia, bleeding, and bacteraemia are other possible complications.

Oesophageal carcinoma often has poor prognosis, and palliative treatment is sometime all that can be offered. The best palliation is an uncomplicated resection if possible. Radiation and chemotherapy or both, are also available. Another option is endoscopic treatment by dilatation, tumour ablation, or placement of an indwelling tubular stent. Some physicians resort to these measures only after failure of surgery or radiation, but endoscopic treatment may also be used initially, especially if the patient's life expectancy is short.

Numerous endoscopic options are available. Oesophageal dilatation, especially with a guidewire technique, is the simplest palliative treatment. Stenosis rapidly recurs, however, making frequent retreatment necessary. Tumour ablation with a laser (Nd:YAG) or bipolar electrocoagulator (‘BICAP tumour probe’) gives somewhat longer-lasting relief. The laser is applied under direct endoscopic view to vaporize intraluminal tumours. Boring a passageway sometimes requires several sessions, each a few days apart, especially if the tumour is treated from above down. Often, however, treatment may be completed in one or two sittings if the malignant stricture is first dilated, the scope passed through, and treatment applied from distal to proximal. This plan also makes it easier for the operator to stay properly oriented within the oesophageal lumen. The bipolar electrocoagulation instrument is passed through the tumour (usually after dilatation) under fluoroscopic guidance. As the probe traverses the tumour and power is applied, the cancer is circumferentially electrocoagulated; thus, it is appropriate only for circumferential lesions.

Tumour ablation is reasonably successful. It does have problems, however. Long or angled lesions are difficult to treat. Upper oesophageal tumours respond poorly even if a lumen is restored, probably because of dysmotility. Bothersome gastro-oesophageal reflux often results when a distal oesophageal or proximal gastric cancer is cored out. As tumours regrow, repeat treatments are required. Furthermore, therapy becomes increasingly difficult and ineffective as the extraluminal portion of the cancer enlarges. Fistulae and perforations occur, of course, especially if the tumour is large, or if radiation has been given previously. Enthusiasts report approximately 90 per cent technical success (restored lumen) with tumour ablation, but only 70 to 80 per cent functional success (improved dysphagia). Tracheo-oesophageal fistulae develop in 10 per cent, and perforations in 5 per cent. Bleeding is an occasional problem. Median survival is in the range of 3 to 12 months.

Oesophageal tube implantation is attractive because a single procedure can give immediate improvement in swallowing, which is more durable than tumour ablation. It is especially useful for the palliation of a malignant tracheo-oesophageal fistula. After initial dilatation or laser of the malignant stricture, a wire-guided introducer (like a dilator) is passed through the tumour. The prosthesis is then slid down over the introducer and positioned across the tumour under fluoroscopy. Alternatively, the endoscope is passed through the malignancy, and the implant is pushed down around the scope and positioned under endoscopic view. Although intubation is perhaps the best endoscopic palliation, the tubes can sometimes be difficult to place accurately. Oesophageal erosion or tube migration may occur, especially if the patient is subsequently radiated. Tumour overgrowth or food may obstruct the prosthesis (the patient should remain on a liquid diet indefinitely). As with tumour fulguration, intubation works poorly in the upper oesophagus, and promotes reflux in the lower oesophagus. In the limited available literature, 80 to 90 per cent of patients are reported to be improved.

American Society for gastrointestinal endoscopy. Gastrointestinal endoscopy: diagnostic and therapeutic procedures: an information resource manual. Manchester, MA.

Branicki FJ, et al. Bleeding duodenal ulcers: a prospective evaluation of risk factors for rebleeding and death. Ann Surg 1990; 211: 411–18.

Dent TL, Kukora JS, Buinewicz BR. Endoscopic screening and surveillance for gastrointestinal malignancy. Surg Clin N Am 1989; 69: 1205–25.

Fleischer D. Endoscopic therapy of upper gastrointestinal bleeding in humans. Gastroenterology 1986; 90: 217–34.

Fleischer D, Sivak MV. Endoscopic ND:YAG laser therapy as palliation for esophagogastric cancer. Gastroenterology 1985; 89: 827–31.

Hunter JG. Endoscopic laser application in the gastrointestinal tract. Surg Clin N Am 1989; 69: 1147–66.

Infante-Rivard C, Esnaola S, Villeneuve JP. Role of endoscopic variceal sclerotherapy in the long term management of variceal bleeding: a meta-analysis. Gastroenterology 1989; 96: 1087–92.

Overholt BF. Laser treatment of oesophageal cancer. Am J Gastroenterol 1985; 80: 719–20.

Overholt BF. Laser treatment of upper gastrointestinal haemorrhage. Am J Gastroenterol 1985; 80: 721–6.

Sacks HS, et al. Endoscopic hemostasis: and effective therapy for bleeding peptic ulcers. JAMA 1990; 264: 494–9.

Schwesinger WH. Endoscopic diagnosis and treatment of mucosal lesions of the oesophagus. Surg Clin N Am 1989; 69: 1185–1203.

Selivanov V, et al. Management of foreign body ingestion. Ann Surg 1984; 199: 187–91.

Snady H. The role of sclerotherapy in the treatment of oesophageal varices: personal experience and a review of randomized trials. Am J Gastroenterol 1987; 82: 813–22.

Snady H, Feinman L. Prediction of variceal haemorrhage: a prospective study. Am J Gastroenterol 1988; 83: 519–25.

Steele RJC. Endoscopic haemostasis for nonvariceal upper gastrointestinal hemorrhage. Br J Surg 1989; 76: 219–25.

Sugawa C. Endoscopic diagnosis and treatment of upper gastrointestinal bleeding. Surg Clin N Am 1989; 69: 1167–83.

Terblanche J, Kahn D, Bornman PC. Long term injection sclerotherapy treatment for oesophageal varices. Ann Surg 1989; 210: 725–31.

Terblanche J, Krige JEJ, Bornman PC. Endoscopic sclerotherapy. Surg Clin N Am 1990; 70: 341–59.

Tytgat GNJ. Dilation therapy of benign oesophageal stenoses. World J Surg 1989; 13: 142–8.

Webb WA. Oesophageal dilation: personal experience with current instruments and techniques. Am J Gastroenterol 1988; 83: 471–5.

Webb WA. Management of foreign bodies of the upper gastrointestinal tract. Gastroenterology 1988; 94: 204–16.