About 10 per cent of the population in the United States and Great Britain suffer from gastric or duodenal ulcers at some time. Most of these ulcers are not serious or chronic and those that do not heal spontaneously respond to medical therapy. Hospital admission and surgery are required by only a small proportion of people with these diseases. For example, in the Massachusetts General Hospital there were about 500 admissions per year for gastric or duodenal ulcer in 1973; a decade later there were only 300 per year. About 25 per cent of patients admitted with these ulcers now undergo surgical procedures.

Statistics concerning the frequency of the diseases must be regarded with some scepticism. Barium contrast studies cannot identify the many small superficial ulcers that come and go within a space of a few days. Endoscopy now discloses many of these ulcers; however, it may also produce false-negative results if the stomach and duodenum are distorted by a previous surgical procedure or by hypertrophic gastric mucosa. The true incidence of ulcer disease is, therefore, higher than is shown in most statistics.

Although the term peptic ulcer is frequently applied to all of these lesions, it is really a misnomer. Duodenal and stomal ulcers are due primarily to gastric acid. Gastric ulcers have several causes; they may either be benign or malignant and must be regarded with suspicion. There are also other less common lesions, such as stress ulcers, Cushing's ulcer, Curling's ulcer, and ulcers due to hypergastrinaemia that must be considered. At present about 60 per cent of the operations or diagnoses made on surgical wards for ulcer disease are due to duodenal ulcer, 35 per cent to gastric ulcers, and 5 per cent to less common causes.

Recent investigations have led to the conclusion that Helicobacter pylori is the cause of the majority of cases of duodenal and gastric ulcer. In a study by Hentschel et al., eradication of this infection has led to freedom from symptoms and apparent cure in 84 per cent of cases of duodenal ulcer. Therapy has included combinations of antibiotics (amoxicillin and metronidazole) and ranitidine for 6 weeks. Graham has used other antibiotics; he concludes that the major causes of peptic ulcer at the present time are H. pylori infection, excessive use of non-steroidal anti-inflammatory drugs, and hypersecretory states.

The earliest operations involving extirpation of the stomach were performed for cancer. Although the French surgeon Jules Péan performed a pylorectomy for cancer in 1879, the patient died 5 days later. Rydigier, the great Polish surgeon, had a similar unsuccessful case. Billroth was the first surgeon to report survival of a patient following excision of the distal portion of the stomach in 1881; he anastomosed the stomach to the duodenum, thereby establishing what has been known ever since as the Billroth I procedure. The patient lived for 4 months before dying from disseminated cancer. In 1885 his assistant, Wölfler, operating on a patient with gastric cancer, performed a preliminary gastrojejunostomy and, at a second stage, removed the distal stomach and closed the duodenal stump; thereafter this operation was known as the Billroth II. The first gastrojejunostomy for an obstructing cancer was performed in 1891 by Wölfler.

In the last two decades of the nineteenth century duodenal ulcers were rare, although benign gastric ulcers were recognized with increasing frequency. Rydigier, in 1881, performed a gastric resection for a benign gastric ulcer, and the patient survived. He received wide criticism; the editor who published the paper (entitled The first gastric resection for gastric ulcer) added the comment ‘And hopefully the last’ to the article.

The first operation for duodenal ulcer—a gastrojejunostomy—according to Herrington, was performed by Codivilla in 1893. This quickly became the procedure of choice for both duodenal and gastric ulcers. Twenty years later, extensive gastrectomies, fostered by von Haberer and by Finsterer began to displace pylorectomies, which were the earlier operations of choice. Angry controversies developed between proponents of gastroenterostomy and those supporting gastric resection. It was not until the 1930s that it was recognized that a large number of gastrojejunal ulcers appeared after gastrojejunostomy; thereafter gastric resection, as popularized by Finsterer, was finally accepted. Nevertheless, occasional gastrojejunostomies were performed in major clinics in the United States for benign ulcers even as late as 1945.

The heyday for gastric resection in the United States lasted until 1943. At that time Dragstedt and Owens reintroduced bilateral truncal vagotomy for the treatment of duodenal ulcers. This operation had been used before. Brodie, from London, had studied it experimentally in 1814; Latarjet had combined it with a conservative gastric resection in 1922; Schiassi had combined vagotomy with drainage in 1925. However, Dragstedt had undertaken many important laboratory experiments that showed the potential of vagotomy had great potential. His recommendation, made in a two-page report, was based upon only two patients who had been followed for 2 months after operation. Nevertheless, his experiments were so convincing that the course of gastric surgery was immediately and completely changed.

Bilateral truncal vagotomy, performed either through the abdomen or the chest, was soon found to be unsatisfactory: the stomach often failed to empty, and early recurrences occurred in one-third of the patients. Hence, an accompanying drainage procedure either by pyloroplasty or gastrojejunostomy (Dragstedt's operation) became the favourite procedure for duodenal ulcer.

Dissatisfied with the results of radical resections of the stomach for duodenal ulcer, and the obvious successes of vagotomy, two separate groups of surgeons almost simultaneously proposed the operation that combined a partial resection of the stomach with truncal vagotomy. The first of these operations was carried out by Smithwick and Farmer in Boston in October 1946; they called the operation ‘hemigastrectomy and vagotomy’. In January 1947, Edwards and Herrington, in Nashville, unaware of Smithwick's operation, performed the same procedure. They called it ‘conservative gastrectomy’ or ‘antrectomy and vagotomy’. Both groups found a spectacular reduction in acid production because of the elimination of antral gastrin and of parietal cell stimulation. This method essentially eliminated the major problem that followed many previous operations—recurrent ulcer.

Thereafter, attention was diverted from recurrence and was focused on other side-effects of operations for duodenal ulcer disease; as a result proximal gastric vagotomy was developed. Clinical use of this procedure was begun in 1965. Because there have been variations in the technique of this operation the method is discussed in detail below.

Arguments for and against various operations will be discussed below. Suffice it to say here that surgeons are not unanimous in their selection of the best operation for duodenal ulcer. Meanwhile, gastric ulcers continue to be treated with satisfaction by some type of gastric resection, often combined with truncal vagotomy.

Gastric and duodenal ulcers are common throughout the world, although supporting evidence from many countries is unreliable. The incidence and severity of peptic ulcer is decreasing in the Western world; the number of deaths from this cause have declined steadily in the past 60 years in the United States. The number of deaths due to ulcers of the stomach and duodenum has decreased from 6.0 per 100 000 population in 1930 to 5.5 in 1950, 3.7 in 1972 (this period was still before widespread use of H&sub2;-receptor antagonists) and to 2.7 in 1986.

A complete survey of all hospitals in Sweden showed that in 1950 there were 64 elective admissions for ulcer operations per 100 000 hospital admissions, compared with 11 in 1986. The number of operations decreased from 9.4 per 100 000 admissions in 1956 to 6.6 in 1986. Gustavsson et al. observed that the dramatic decrease in peptic ulcer surgery started long before the advent of fibreoptic endoscopy, H&sub2;-receptor antagonists, and highly selective (proximal gastric) vagotomy. They also concluded that the falling incidence of perforation indicates that the ulcer prevalence (or severity) has diminished.

Taylor, commenting on the changing picture of peptic ulcer disease in Great Britain up to 1989, stated that in the 10 years since H&sub2;-receptor antagonists have become available no reduction occurred in the number of deaths due to this disease in England and Wales. Instead, deaths have now shifted to older age groups: 95 per cent of the deaths occur in patients over 55 years of age.

Our studies of ulcer disease from 1974 to 1985 show a sharp decline in elective operations for duodenal ulcer to approximately 10 per year. The number of cases as well as the death rate in those patients operated on for acute massive haemorrhage and acute perforation have remained nearly constant during the entire period.

In summary, operations for peptic ulcer now are chiefly undertaken for massive haemorrhage and acute perforation. Because poor-risk elderly patients are frequently the victims of these catastrophes, the hospital mortality is increased in contrast to the overall reduction in mortality of peptic ulcer in the United States.

Ulcers of the stomach and duodenum are caused chiefly by the effects of hydrochloric acid, produced by the parietal cells of the stomach, and by lack of protection of the mucosa against this acid. Acid production is by far the most important factor as far as duodenal ulcer is concerned, but cannot be the only factor, since the severity of duodenal ulcers and their responses to therapy do not vary directly with the amount of gastric acid secreted.

Additional factors are important in the production of gastric ulcers. Gastric contents can be retained for a much longer period than duodenal contents and are not neutralized as rapidly as by duodenal chyme. The mucous membrane of the stomach must normally be protected continuously from the damaging effect of the hydrochloric acid which it secretes. This protective layer consists of an adherent layer of mucus which is separated from the gastric mucosa by a bicarbonate layer that normally neutralizes any hydrochloric acid diffusing back through the mucous layer. The mucous layer also protects the gastric mucosa against other noxious agents such as bile, alcohol, and aspirin and other irritating drugs.

Secretion of acid by the parietal cells of the stomach is influenced by several mechanisms. Stimulation by the vagus nerves accounts for about 50 per cent of acid secretion. Gastrin that is secreted by the antral mucosa causes 40 to 45 per cent of the secretion and the remaining 5 to 10 per cent is due to gastrin that comes from the intestinal tract from various APUD cells.

Production of gastric mucus should presumably vary according to the blood supply of the organ (Fig. 1) 922. Experimental studies have shown that prostaglandins increase the mucosal blood supply and increase the bicarbonate level and thickness of the mucous layer; they may, therefore, play an important role in the protection of the gastric mucosa.

The stomach contains several types of glands. Mucus-secreting glands are located throughout the stomach; gastrin-producing cells occur in the antrum and in an area that often runs along the lesser curvature up nearly to the oesophagus. Chief cells, located mainly in the upper half of the stomach, secrete pepsin. Parietal cells secrete hydrochloric acid and also are located almost entirely in the fundus.

The stomach is supplied by both sympathetic and parasympathetic nerves. Although the function of the sympathetic nerves is not of great importance to surgeons, during the 1940s it was noted that patients with ulcers who underwent bilateral lumbodorsal sympathectomies for essential hypertension lost their ulcer pain immediately after the operation.

The parasympathetic supply comes from the vagus. Section of both vagus trunks at the oesophagogastric level not only reduces a major stimulus to the secretory activity of the parietal cells but also produces major changes in the motility of the stomach. The secretion of gastrin by antral cells may also be diminished.

The cause of a partial return of acid secretion after bilateral truncal vagotomy remains a mystery. True regeneration of the nerves has never been demonstrated in the stomach, although Nyhus did describe ‘budding’ of minor branches. Parasympathetic fibres probably run together with the sympathetic nerves along other major arteries such as the right gastroepiploic artery or the splenic artery and they may become more active after vagotomy.

Causes of gastric ulcers vary in importance depending upon the location of these ulcers. In 1965 Johnson observed the different behaviour of ulcers in different portions of the stomach and classified them into three groups. In patients with type 1 ulcers, located in the body of the stomach, gastric acid secretion as determined by secretory tests is low. Here inadequate mucosal protection and back-diffusion of hydrochloric acid through the protective bicarbonate and mucous layer into the gastric mucosa seem to be involved in the disease. Silen suggested it is in this location that alternating waves of gastric acid from above and refluxing alkaline secretions from the duodenum destroy the protective layer of mucus.

Johnson's type 2 gastric ulcers (those combined with a present or past duodenal ulcer) and type 3 (those in the prepyloric area) behave more like duodenal ulcers; here the effects of acid production appear to be more important.

Ulcers that develop after the formation of anastomoses of the stomach with the intestine, or ulcers which recur after various operations for duodenal or gastric ulcers are due to persistent secretion of hydrochloric acid.

Extragastric lesions can lead to increased levels of serum gastrin and consequent acid production, followed by ulceration in the stomach, duodenum, or intestine. The prime example is the gastrinoma—the usual cause of the Zollinger–Ellison syndrome.

The diagnosis of a gastric or duodenal ulcer is made from protean symptoms and signs. If the ulcer is acute and uncomplicated, epigastric pain occurs typically about 3 a.m. and an hour or more after meals. Pain is located in the mid-epigastrium to the right of the midline with duodenal ulcers, and in the midline or slightly to the left of the midline with gastric ulcers. The pain is relieved by food or antacids. Slight nausea or epigastric fullness is common. Vomiting is uncommon unless pyloric obstruction supervenes.

Physical examination is negative except for slight tenderness, usually sharply localized to the area in which the pain is felt. Confirmatory evidence is obtained by endoscopy and, particularly in the case of a gastric ulcer, by biopsy examination. A gastric ulcer should be considered to be malignant unless scrupulous biopsy examination of a large specimen indicates otherwise and unless the ulcer heals on medical therapy. In more chronic cases, barium contrast studies are also helpful. The stool should be examined for occult blood; blood counts should be made to rule out anaemia.

Any complication of ulcer disease raises the question of surgical therapy. Some severe complications occur without previous symptoms, but others may be treated medically, at least for a certain time. Close co-operation between physicians and surgeons is necessary in the care of such cases. These complications, all of which may lead to operative procedures, are acute perforation, bleeding (either massive, repetitive, or occult), pyloric obstruction, intractability, and fistula formation.

Acute perforation is accompanied by the abrupt onset of severe epigastric pain and shock, with hypotension and sweating. There is often a brief period in which the patient feels somewhat better, but the pain recurs with greater intensity. An abdominal radiograph taken 6 h after perforation with the patient upright shows gas beneath the diaphragm in 60 per cent of the patients. In questionable cases, Gastrografin can be injected through a nasogastric tube to determine whether or not there is a perforation. Determining whether or not the patient had previous ulcer symptoms is important: the choice of operation may depend upon this single factor.

Acute massive haemorrhage is defined clinically by the requirement of five units of whole blood to restore normal vital signs and blood counts. The patient vomits blood and exhibits signs of shock. In the great majority of cases there is a history of intake of some drug such as aspirin that irritates the gastric mucosa.

Occult bleeding, characterized by tarry stools or merely by positive guaiac stools, may occur in patients with duodenal ulcers. Diagnosis is obtained most readily by endoscopy.

Obstruction is diagnosed by repeated vomiting, hyponatraemia, and hypochloraemia. If there is accompanying ulcer pain, the obstruction is probably due to oedema about an active ulcer; if there is no accompanying pain, it is probably a result of fibrosis and scar. On examination a greatly distended stomach is palpable. Peristaltic waves can be observed in thin patients, running from left to right. Diagnosis is made by the insertion of a nasogastric tube and confirmed later by barium studies and endoscopy to exclude or confirm the presence of an obstructing tumour.

Intractability refers to chronic ulcers that have not responded to treatment with various medications. Many patients have symptoms that suggest ulcers, but no objective evidence of abnormalities on endoscopy or barium studies. Vigilance is essential before an operation is undertaken since patients in whom objective evidence of ulcer disease is minimal and in whom no relief is obtained on standard medications are apt to do poorly after an operation.

The differential diagnoses in these patients include a variety of possibilities, ranging from functional abnormalities to hiatus hernia with reflux oesophagitis, gallstones, pancreatitis, and coronary disease.

The three major types of peptic ulcers—gastric, duodenal, and stomal (anastomotic)—have certain distinguishing characteristics in addition to those noted above.

Gastric ulcers occur most commonly in elderly women, with the highest incidence between 55 and 60 years of age; duodenal ulcers, in contrast, are more common in men at a younger age.

If preoperative acid studies are undertaken in patients with type 1 ulcers, gastric pH levels approach those indicative of achlorhydria and are distinctly higher than those with types II or III or with duodenal ulcers.

A benign ulcer must be distinguished from an ulcerating cancer. The location of an ulcer in the stomach is not helpful, although ulcerating lesions on the greater curvature are rare and appear to be as likely to be due to cancer as to benign disease. Both advanced age and comparative hypochlorhydria are also characteristic of gastric cancer. A good clinical rule is to suspect that every gastric ulcer is malignant initially. Diagnosis cannot be considered adequate until the results of endoscopy and biopsy are known. Even if these studies suggest the lesion is benign, failure of the ulcer to heal or its recurrence must be regarded as danger signs and an indication for operation.

The symptoms of a gastric ulcer are similar to those of a duodenal ulcer except that in gastric ulcer the epigastric pain tends to be slightly more to the left. Bleeding occurs more commonly and perforation less frequently than is the case with duodenal ulcers: most perforations occurring in what appears to be the immediate prepyloric area are actually duodenal ulcers. Massive haemorrhage from a gastric ulcer is more likely to be fatal than that from a duodenal ulcer.

An unusual complication of a perforation occurs when ulcers on the greater curvature penetrate into the colon, producing a gastrocolic fistula. The usual causes of such fistulae are cancers of the colon or stomach penetrating into the neighbouring viscus. Surgical intervention is necessary.

Duodenal ulcers characteristically occur in young men aged 20 to 50 years, but they may appear at any age. They are usually initially diagnosed clinically by pain in the epigastrium just to the right of the midline. The pain varies in nature and in intensity; it often is described as ‘burning’ and is most likely to occur at night and several hours after eating. The pain is reduced rapidly by intake of food or antacids. Less common symptoms include heartburn from oesophageal reflux, and nausea and vomiting.

Duodenal ulcers are usually located in the first portion of the duodenum, known as the bulb. Although benign ulcers are found lower in the duodenum, cancer must be suspected if ulcerating lesions are found in this unusual location.

The diagnosis of a duodenal ulcer with only minimal symptoms is made on the basis of the typical pain and response to food or drugs that reduce acid secretion or protect the mucosa. In such cases neither endoscopy nor barium contrast radiographs are necessary. Diagnosis in more severe cases is preferably made by endoscopy; repeated endoscopies to monitor healing are not indicated unless the patient responds poorly. Barium contrast studies are helpful, particularly if an operation is contemplated, because the extent of the duodenal deformity can be delineated.

Perforations of the duodenum are usually due to anterior ulcers. Chronic perforations are rare; they usually arise from the posterior wall and are contained by the head of the pancreas. Occasionally, fistulae to other organs, such as the biliary tree or colon, occur.

Stomal ulcers have many characteristics that distinguish them from gastric and duodenal ulcers, and provide difficult diagnostic and technical problems for surgeons. They develop following gastric resection or other procedures, such as gastroenterostomy, that involve the anastomosis of the stomach to some other portion of the gastrointestinal tract. In such cases an ulcer develops near the stoma; it is almost invariably in the efferent limb of the intestine, not in the stomach.

These ulcers are due to the effects of gastric acid on the susceptible intestinal mucosa. There appears to be a direct relationship between vulnerability to acid and distance of the intestine from the pylorus. For example, technical errors have at times led to gastroileal or gastrocolic anastomoses: they are much more likely to be followed by stomal ulcers than are gastroduodenal or gastrojejunal stomas.

Stomal ulcers are nearly always associated with high levels of acid secretion and hence occur after operations for duodenal ulcer, or for type 2 or 3 gastric ulcers. If the previous operation was apparently adequate to reduce acid, the possibility of hypergastrinaemia resulting from the retained antrum syndrome or from a gastrinoma should be considered: serum gastrin levels should be measured. Rarely, such ulcers may be confused with cancer arising in the gastric mucosa near an anastomotic line.

Diagnosis may be difficult. Endoscopy is essential, but may fail to disclose an ulcer after a gastrojejunostomy if the ulcer is remote from the stoma. Barium studies are helpful if the ulcer is in the duodenum, but not after Billroth II resections. If bleeding is a symptom, selective angiography can often confirm the diagnosis.

The symptoms of such ulcers include pain, haemorrhage, and perforation. Pain is usually similar to that associated with the previous ulcer, but occurs more to the left in the epigastrium if the complication follows gastroenterostomy. Medical therapy, similar to that for duodenal ulcer, has been used frequently for this symptom. Although some observers have reported favourable results, most cases recur when therapy is stopped. The indications for surgical therapy are therefore broader than for other types of ulcer disease, and include the presence of a stomal ulcer as well as all of the other standard indications for ulcer surgery.

A serious complication of a stomal ulcer that develops after a Billroth II gastrectomy or a gastrojejunostomy is a gastrojejunocolic fistula. Patients may have relatively little pain, but have sudden uncontrollable diarrhoea. The faeces contains undigested food and vomiting of faeces may occur, with a persistent foul smell on the breath. Loss of weight and malnutrition occur unless prompt surgical relief is provided. A barium enema or upper gastrointestinal series will confirm the diagnosis.

The operation that is required includes resection of the stomach, intestine, and colon involved in the fistula. Antibiotic preparation, both orally (neomycin and erythromycin base) and parenterally (metronidazole and ampicillin, for example, or cefotetan and gentamicin) is necessary, and the colon must be cleansed by administration of laxatives or a polyethylene-glycol based cathartic. After an adequate gastric resection and vagotomy, the continuity of the jejunum is restored, and a new gastrojejunostomy is made. If the patient is in generally good medical condition, continuity of the colon is also restored, but if the patient is badly malnourished both ends of the resected colon can be brought out at colostomies, continuity being restored at a later date.

The operative procedure for a stomal ulcer depends upon the original operation for ulcer disease. If the previous procedure was a gastroenterostomy alone, gastric resection and bilateral vagotomy are necessary in addition to reconstruction of jejunal continuity. If a gastric resection and vagotomy had been performed previously, the duodenal stump must be exposed to ensure that the entire antrum was removed, and any residual intact vagal fibres or trunks must be identified and divided. If the previous resection and vagotomy was apparently adequate, a higher gastric resection is required.

Because the operative field in the abdomen is often obscured by numerous adhesions, a secondary vagotomy may be very difficult or dangerous. A transthoracic vagotomy is advisable in these circumstances.

The medical therapy is similar for both gastric and duodenal ulcer and includes the use of specific medications and elimination of known gastric irritants. Attempts to relieve social problems that include irregular habits of eating, working, and sleeping are also made.

Gastric irritants include tobacco, alcohol, caffeine, aspirin, ibuprofen, indomethacin, and salicylates, as well as many other non-steroidal anti-inflammatory drugs. Although the exact mechanism by which cigarette smoking is related to ulcer disease is not known, experience has shown that smoking leads to the development and prevents healing of ulcers. Alcohol is a severe irritant that promotes the formation of ulcers. Patients who have portal hypertension and are believed to be bleeding from oesophageal varices actually often have ulcers of the stomach or duodenum. Aspirin is also a direct irritant: its effect on platelet adhesiveness means that it not only causes ulcers, but leads to bleeding from them. Ibuprofen and the numerous proprietary drugs in which it is present are also irritants. Orange juice is irritating because it contains citric acid.

For many years diet was considered to be closely related to ulcer disease. Elaborate concoctions, relying chiefly on milk and cream, were considered to be essential in acute stages of the disease. These specific diets are not essential: regular food, taken at regular intervals, is best. Because food tends to reduce the action of acid on the mucosa, small, frequent meals are desirable in acute episodes. A drink of milk or a flavoured milk drink and a few crackers are valuable as a snack prior to bedtime and for rapid control of pain that appears in the middle of the night, despite the secretogenic stimulus of milk on gastric acid secretions.

Specific medications to control ulcers include antacids, H&sub2;-receptor antagonists, mucosal coating agents, omeprazole, and prostaglandins. Pharmacies offer a host of antacid preparations. They contain such compounds as sodium bicarbonate, magnesium carbonate, and aluminium hydroxide. The calcium content of some of these agents potentiates acid secretion, however. In 1973 compounds available as over-the-counter preparations were graded for safety and efficiency by a panel set up by the Food and Drug Administration (FDA): Mylanta II was the most effective.

The most common prescription drugs are the H&sub2;-receptor antagonists, particularly cimetidine (300 mg four times daily) and ranitidine (150 mg twice daily), although several others are being developed.

The most effective mucosal coating agent is sucralfate (1 g four times daily). Omeprazole acts upon the intracellular proton pump and can eliminate gastric acid production altogether.

Among their numerous other properties, prostaglandins increase the blood supply to the gastric mucosa and decrease the damaging effects of non-steroidal anti-inflammatory drugs. The only prostaglandin approved to date by the FDA is misoprostol.

Controlled studies in which various drugs were tested and compared with placebos indicated that antacids, H&sub2;-receptor antagonists, sucralfate, and omeprazole were equally effective in so far as relief for pain was concerned. Using pain control and endoscopy as elements for comparison, about 80 per cent of ulcers healed within 1 month, although healing time was longer with gastric ulcers and in aged patients. The recurrence rate within 6 months was high. A gastric ulcer that either fails to show definite evidence of healing within 1 month or that recurs after therapy has ceased should always be regarded with suspicion, even if biopsy specimens show no sign of cancer. These ulcers should be considered for surgical removal.

Up to 1970 surgical therapy was chiefly undertaken in patients with painful gastric or duodenal ulcers who failed to respond to medical measures. These so-called ‘intractable’ cases are now rare, not only because better therapeutic drugs have been found, but because ulcers are less virulent than they were in the past.

Major attention must now be paid to patients who present as emergencies, with either acute massive bleeding or acute perforations. Both of these catastrophes demand surgical salvage; medical therapy is not an option.

The essential features of an elective operation for gastric ulcer are: (1) preliminary endoscopy and biopsy of the lesion; (2) partial gastrectomy that includes the entire antrum and the ulcer; (3) the addition of bilateral truncal vagotomy except for type 1 ulcers; (4) for juxtaoesophageal ulcers, intragastric biopsy of the lesion, followed by antrectomy and Billroth II anastomosis, leaving the ulcer in situ (Madlener procedure). For ulcers complicated by massive haemorrhage or perforation a distal gastrectomy that includes the ulcer is preferred.

In elective operations performed for intractable pain, antrectomy and bilateral vagotomy is preferred; proximal gastric vagotomy or pyloroplasty and bilateral truncal vagotomy are acceptable, according to criteria specified below. Selection of operations for patients with a massive haemorrhage or acute perforation are discussed below.

Serum gastrin levels should be measured to ascertain the presence of a gastrinoma. If the previous operation was a gastroenterostomy or pyloroplasty, a Billroth II resection and bilateral truncal vagotomy are performed. If the previous operation was a Billroth II gastrectomy, a bilateral truncal vagotomy or higher gastric resection plus vagotomy is recommended.

The major indications for surgical treatment of peptic ulcer include acute perforation, acute massive haemorrhage, gastric outlet obstruction, and intractability. Less common indications include repeated episodes of minor bleeding and fistula formation (gastrocolic, duodenocolic, or from the duodenum into any portion of the biliary tree). Intractability in the case of a gastric ulcer means delayed healing or recurrence after healing, even in the absence of pain: such apparently benign ulcers can actually be malignant.

Preparations for operation may need to be limited to ‘essentials’ in patients who present as emergencies with acute perforation or massive haemorrhage. Nasogastric decompression prior to induction of anaesthesia, intravenous or intramuscular administration of antibiotics, establishment of intravenous lines for administration of electrolytes, and a supply of compatible blood are essential.

In patients likely to require surgery, but in whom there is no emergency, time may be taken for necessary diagnostic measures. Suction on an inlying nasogastric tube is necessary if there is any evidence of obstruction. Thorough medical evaluation, including haematological studies, may indicate the need for preoperative transfusions.

The diagnosis of an acute perforation of a duodenal ulcer is suggested by the sudden onset of severe epigastric pain followed by a variable degree of shock and often slight vomiting. In the untreated patient, the condition tends to improve after a few hours, to be followed shortly thereafter by increasing prostration, pain spreading throughout the abdomen, and cardiovascular collapse. Physical examination typically reveals a board-like abdomen, with tenderness most marked in the mid- or right epigastrium. Peristalsis is absent. The temperature in the first few hours after perforation is normal.

Six hours after onset of symptoms an abdominal radiograph taken in the sitting position shows free air beneath the diaphragm in about 60 per cent of cases (Fig. 3) 924. If the diagnosis is in doubt, Gastrografin injected through an inlying nasogastric tube will show evidence of extravasation through the perforation.

The differential diagnosis includes acute cholecystitis, acute pancreatitis, strangulating intestinal obstruction, acute appendicitis, perforation of some other portion of the intestinal tract, and mesenteric thrombosis. A past history of ulcer disease is predictive, since the chances that the present episode is due to a perforated ulcer are greatly increased in such patients.

Occasionally, a perforated ulcer seals spontaneously and the patient will continue to improve. However, these ‘formes frustes’ are rare, and delay before an operation, other than to institute nasogastric suction, antibiotics, and intravenous administration of fluids and electrolytes, is dangerous.

Exploratory laparotomy is often necessary to confirm the diagnosis. The size of the perforation may vary greatly, from a diameter of only 2 to 3 mm to a hole 2 to 5 cm across.

When the abdomen is opened, the perforation may have sealed spontaneously, covered in most cases by adjacent omentum. Such a finding has led some surgeons to advocate the non-operative treatment of what is believed to be an acutely perforated ulcer, such treatment consisting of nasogastric suction, antibiotics, and intravenous fluids and electrolytes. The diagnosis of an abdominal catastrophe may be missed and this approach must be abandoned if the patient's condition appears to be deteriorating. This injunction means that such patients must be watched carefully and that operation may be necessary at an unfortunate hour.

Although this method of therapy may produce satisfactory results in young, vigorous patients, older persons often have other serious diseases such as mesenteric thrombosis; they cannot withstand the effects of peritonitis, and spontaneous sealing of the perforation does not occur as often as in younger patients. The main indications for such treatment include patients with recent coronary occlusion or those in whom the diagnosis has been delayed and in whom the ulcer has apparently sealed spontaneously.

At the time of operation for an acutely perforated duodenal ulcer the surgeon must choose between a simple closure of the perforation or a definitive procedure designed to prevent future recurrences of ulcer disease. Simple closure may be effected with a few interrupted sutures, an omental or round ligament patch, or by suture closure reinforced with omentum. Care must be taken to avoid obstructing the duodenum with the sutures.

In certain instances a more radical operation is necessary. Such cases include a giant perforation that cannot be closed by suture, a large posterior ulcer, a perforated ulcer on the anterior wall, or perforation accompanied by profuse bleeding. As was noted above, the most important indication for a definitive operation is a history of a previous ulcer.

Definitive operations can be carried out successfully at the time of the great majority of operations for perforation. Appropriate procedures include gastric resection (with or without truncal vagotomy), pyloroplasty and bilateral truncal vagotomy, and proximal gastric vagotomy. Opinions vary concerning the choice of these procedures. Our preference has been for bilateral truncal vagotomy and antrectomy, although some surgeons opt for proximal gastric vagotomy.

Unless operating conditions are optimal, the surgeon must choose a simple closure of a perforation as a life-saving method; if necessary a definitive procedure can be performed later. Follow-up studies after simple plication show that nearly one-third of all patients remain free of symptoms, and about half of those with recurrent symptoms requires a definitive operation for ulcer disease later. In the last Massachusetts General Hospital series of 107 patients, 21 per cent received a definitive operation within 5 years after simple plication.

The great majority of perforated gastric ulcers are located in the immediate prepyloric area. They behave in the same way as perforated duodenal ulcers, and the same considerations are applicable. However, perforations of ulcers elsewhere in the stomach introduce the possibility of malignancy, and immediate definitive resections of the stomach are recommended. If the patient's condition is poor, and only a simple closure is contemplated, biopsy specimens should be taken of the margins of the ulcer.

A patient who has required 5 units of blood during a single hospital admission by definition has a massive haemorrhage. When a patient enters with serious upper gastrointestinal bleeding, the source can usually be determined with reasonable certainty on clinical grounds. Nearly two-thirds of all cases of severe upper gastrointestinal haemorrhage are due to ulcers of the stomach or duodenum. A past history of ulcer disease, a recent history of ingestion of gastric irritants such as aspirin or ibuprofen, sudden profuse vomiting of blood, and later haematochezia are typical. The other major causes of massive upper gastrointestinal bleeding—portal hypertension and oesophageal varices—can usually be eliminated by history and physical examination, but the frequency of concomitant ulcer disease makes other early diagnostic procedures essential. At times, a bleeding duodenal ulcer is manifested merely by tarry stools.

Early upper gastrointestinal endoscopy is the most valuable diagnostic study. However, profuse bleeding may make it useless, since such patients must be taken directly to the operating room. If an Ewald tube is passed into the stomach and irrigation with saline solution is continued until the bleeding slows to a trickle, the time is ripe for endoscopy, and the source can usually be determined.

Selective angiography can be helpful but is not always available. On the other hand injection of 99 m-technetium-labelled erythrocytes is not useful because of the heavy background radiation in the upper abdomen. Although barium contrast studies often show the ulcer, they should not be used before angiography because residual barium might prevent further imaging studies.

The diagnosis of the source of bleeding must often be made by laparotomy. An adequate incision and careful observation of the abdominal contents should indicate the probable site of bleeding. A wide gastrotomy incision that extends across the pylorus, if necessary, gives the best exposure to help find the source.

Therapeutic measures for acute massive bleeding include injection of vasopressin into a peripheral vein, irrigation of the stomach for removal of blood clots, continued nasogastric suction, elimination or oral intake, and intravenous replacement of blood, fluid, and electrolytes. Until an operation is performed or the patient recovers, a careful chart of the vital signs and repeated assays of the haematocrit are essential

Massive bleeding is not well tolerated in patients over 60 years of age: under that age, mortality is not increased by delay and attempts to control the bleeding by endoscopic techniques such as electrocoagulation or laser.

Older patients respond much better to early operation. As shown in a controlled series by Morris et al., delay of definitive treatment by over 48 h increases the mortality rate in patients 60 years of age or over from 2 per cent to 15 per cent. Such delay did not increase mortality in younger patients, however.

Early operation is indicated in patients bleeding from the gastroduodenal artery or a gastric ulcer. Angiographic embolization of the gastroduodenal or the left gastric arteries may be effective, but there are dangers involved. A foreign body may slip from the gastroduodenal into the hepatic artery and lead ot hepatic necrosis; one in the left gastric artery may lead to necrosis of the upper portion of the stomach.

The site of bleeding must be determined. If the source is the gastroduodenal artery, it must be ligated as the first step. This procedure involves suture of the artery, either above the duodenum as it emerges from the hepatic artery or within the duodenal lumen, suture of the caudal portion of the artery within the duodenum or of the two major branches—the superior pancreatoduodenal and the right gastroepiploic—and of the transverse pancreatic artery. Heavy, non-absorbable sutures are used.

In other instances the duodenum may be boggy and bleeding from multiple areas; in such instances it may be impossible to identify single vessels to ligate.

As soon as bleeding has been controlled, a definitive operation can be carried out. We believe that hemigastrectomy and truncal vagotomy followed by a Billroth II anastomosis is best; others prefer the simpler pyloroplasty and truncal vagotomy. A major determinant in the selection is the experience of the surgeon. Gastric resection is more difficult, but gives slightly better control of bleeding because it removes areas of gastritis and duodenitis that are potential sources of postoperative bleeding.

Gastric resection is preferred for bleeding gastric ulcers. A truncal vagotomy should be added if the ulcer is in the prepyloric area or if the patient has a history of a duodenal ulcer. Either a Billroth I or II anastomosis can be made.

In some patients who are very poor operative risks, a local excision of the ulcer may be performed. However, the chances of recurrence within a year approach nearly 50 per cent.

The most difficult stomach ulcer to treat is one located in the proximal portion. In older patients who are poor risks, a distal resection, ligation of the left gastric artery, and intragastric plication of the ulcer is recommended.

Some surgeons have approved simple plication for all bleeding gastric ulcers. We have had no experience with this method. The only patient treated in this way in the Massachusetts General Hospital in the last decade required a second operation within 24 h for recurrent bleeding; he did well after a gastric resection.

Results of operations for massive bleeding from gastric and duodenal ulcer

The reported results of such operations vary greatly. In the Massachusetts General Hospital the mortality in the last study was 22 per cent; many of these patients had bled after other serious operations such as coronary artery bypass. Other groups have reported mortality rates as low as 10 per cent. It is impossible to compare surgical results with other methods of therapy; endoscopy is rarely attempted in such cases and arteriography is used infrequently. Furthermore, patients in whom other methods of therapy have failed are relegated to surgeons in the hope that they can solve the problem.

Gastric outlet obstruction is one of the important indications for surgery for an ulcer of the stomach or duodenum. Such complications must be regarded with care, because gastric cancer may be present, but is difficult to diagnose in these circumstances.

Duodenal ulcers cause obstruction for two reasons: in young patients acute ulcers are associated with much surrounding inflammation and oedema; in the elderly fibrosis develops about an old ulcer. Differentiation between the two modes of obstruction usually can be made by observation during several days of nasogastric suction, intravenous administration of H&sub2;-receptor antagonists, and intravenous alimentation. Oedema usually subsides within a few days, while fibrotic obstruction continues to produce large quantities of gastric aspirate.

In either case the prognosis is poor without operation. However, obstruction due to oedema that subsides rapidly can be treated conservatively for a few weeks, at which time the operation will be much easier and safer.

The operation chosen depends on the condition of the patient. Gastric resection combined with vagotomy is our preference in young patients, although vagotomy and gastroenterostomy may be performed if inflammation about the pylorus is severe. Proximal gastric vagotomy is contraindicated.

Older patients in poor condition who are considered to have fibrotic obstruction and burned-out ulcers, can be treated if necessary by gastroenterostomy under local anaesthesia. A gastric resection will generally be tolerated, but the addition of a vagotomy may lead to a prolonged period of gastric atony, and should be avoided.

Obstructing gastric ulcers should be treated by gastric resection combined with truncal vagotomy.

Intractability implies that a patient with an ulcer has been treated with the best available medical therapy but is still symptomatic. Such ulcers are usually duodenal or prepyloric. It is important that a gastrinoma is ruled out by appropriate tests. The patient should undergo endoscopy while symptoms are most severe; if an ulcer is present, and further medical therapy only leads to temporary relief, a standard ulcer operation should be performed.

There are many patients who complain of ulcer-like symptoms, but in whom no objective evidence of an ulcer can be found: other causes of epigastric distress should be investigated. These patients should be examined carefully by endoscopy and by barium contrast radiography before an operation is undertaken. To do otherwise may mean that the operative findings will be unimpressive and that the patient will complain of the same symptoms postoperatively.

In these patients full medical therapy is recommended before embarking on any possibly curative operation. Under these circumstances the least that can be done is the best. A proximal gastric vagotomy will be the least likely to subject the patient to new, more serious postoperative symptoms, provided that the procedure is uncomplicated.

Few operations are performed for intractability. In the Massachusetts General Hospital about 10 operations for persistent or recurrent gastric ulcer and 10 for recurrent or persistent duodenal ulcer are performed per year. The number is likely to be considerably higher in countries where continued medical therapy is impossible and where operations are performed for less strict criteria.

Repetitive minor haemorrhage is another indication for operation. The source of such bleeding is more likely to be ulcer disease associated with gastritis or duodenitis; major vessels, such as the left gastric or gastroduodenal are not involved. The symptoms usually consist of repeated episodes of tarry stools, which may not be associated with pain.

Other sites of occult bleeding may need to be investigated, and endoscopy during a bleeding episode is essential to confirm the diagnosis. Otherwise, this difficult diagnostic problem may require such measures as preoperative arteriography and intraoperative endoscopy.

When the symptoms are not controlled by medical measures, operation is necessary. Gastric resection with bilateral truncal vagotomy is preferred; the antrum should be completely excised. Careful examination of the entire gastrointestinal tract at the time of operation is essential.

Gastric ulcers can perforate into the transverse colon, producing a gastrocolic fistula, the symptoms of which include vomiting of faeces and severe diarrhoea. Exactly the same type of fistula and symptoms can occur if the underlying disease is either gastric or colonic cancer. The diagnosis of a fistula may be made by barium contrast radiographs or endoscopy. However, endoscopic observation and biopsy to rule out cancer are indicated. If the lesion is benign, gastric and colonic resections are necessary; malignancy of either the stomach or colon requires more extensive resection.

Duodenal ulcers may penetrate into the common bile duct or the ascending or transverse colon; involvement of any other organ is rare. Although the general principles of therapy are the same as for gastric ulcers, choledochoduodenal fistulae furnish special problems. This problem may arise either from a duodenal ulcer in the second portion of the duodenum or from primary disease of the biliary tree. If biliary tract disease can be ruled out and the dissection promises to be dangerous, a low ulcer can be treated by a Billroth II resection; the duodenal stump can be closed cephalad to the fistula, leaving it in place. A vagotomy is added.

This operation is the favourite procedure employed by many surgeons for the treatment of both duodenal and gastric ulcers because of its superiority in elimination of gastric acid, a low recurrence rate, and the absence of most of the troublesome sequelae of more radical gastrectomies.

The indications for the use of this operation include any of the complications of duodenal ulcer, and gastric ulcer located in the pylorus or prepyloric area, or gastric ulcer combined with duodenal ulcer. It also can be used for all other peptic ulcers of the stomach.

This is an operation of considerable complexity and severity. It carries a higher risk of postoperative complications than pyloroplasty and vagotomy, proximal gastric vagotomy, or gastric resection alone. Even skilled surgeons may have to adopt different procedures in very ill or aged persons.

The operation is a combination of bilateral truncal vagotomy and distal gastrectomy. The procedure is completed by either a Billroth I gastroduodenostomy (Fig. 4) 925 or a Billroth II gastroenterostomy (Fig. 5) 926. Antrectomy and hemigastrectomy are considered to be synonymous for practical purposes.

The outstanding feature of this operation is that the incidence of recurrent ulceration is less than 1 per cent after 5 years in all large series. Furthermore, long-term data are now available, and indicate no increase in the rate of recurrence.

In the last Massachusetts General Hospital study, the operative mortality in elective cases was 0.5 per cent; this increased when the indications for operation were acute massive haemorrhage or acute perforation. Morbidity from early complications led to reoperation in 8 per cent of cases; reoperations for recurrent ulcer were necessary in 0.3 per cent of patients. Reoperations for other reasons (chiefly alkaline gastritis or motility problems) were necessary at a late period in 5 per cent of patients.

Smithwick found in his early studies that reduction in fasting and stimulated acid secretion was nearly complete, that long-range dumping symptoms occurred in about 5 per cent but were disabling in only 1 per cent, that significant weight loss occurred in 3 per cent, and persistent diarrhoea was troublesome in 3 per cent.

Herrington reported that in his last series of 341 patients treated between 1975 and 1984 65 per cent obtained an excellent result, 20 per cent a good result, 12 per cent a fair result, and 3 per cent a poor result.

Three types of vagotomy require consideration. Bilateral truncal vagotomy denervates the entire stomach and the gastrointestinal tract to the midcolon (Fig. 6) 927. When combined with other operations it is extremely effective in reducing the number of recurrent ulcers. However it does carry some deleterious side-effects, reducing the ability of the stomach to empty and being followed by other late motility disturbances, occasional vagus diarrhoea, and reflux alkaline gastritis.

Selective vagotomy denervates the entire stomach but leaves nerves to the gallbladder, pylorus, and bowel intact. From a practical point of view this operation is more difficult and time-consuming than truncal vagotomy, appears to have nearly equivalent results, and has attracted few supporters.

Proximal gastric vagotomy, the most recently developed type of vagotomy, has received enormous support and is the only type of vagotomy that can be used without pyloroplasty, gastroenterostomy, or antrectomy.

This surgical procedure, which many observers today believe is the best available operation for duodenal ulcer, was developed relatively recently. Conceived in America, first tried in Germany in combination with procedures designed to reduce spasm of the pyloric muscle, and used shortly afterward without any interference with the pylorus simultaneously in Denmark and England, it has a peculiarly international flavour. For this reason, the operation first had several different names, and differences in technique led to acrimonious exchanges between various pioneers.

The world was looking for a superior operation for duodenal ulcer when, in 1957, Griffiths and Harken in Seattle, performed the original laboratory investigations in dogs that proved the feasibility of the operation. Truncal vagotomy alone at that time was known to lead to gastric paresis, with severe gastric retention and accentuation of the gastrin factor, actually leading to an increase in the activity of a duodenal ulcer unless an emptying procedure—either pyloroplasty or gastroenterostomy—was added. The new operation, which allowed the retention of motor function in the antrum and a functioning pylorus, appeared to be the answer.

When Holle, in Munich, first described the procedure in 1965, he was concerned that the antrum would not empty properly, and performed additional procedures, such as section of the pyloric muscle. Such additions were viewed with alarm by Amdrup and Jensen in Copenhagen and Johnston in Leeds; their publications in 1970 included descriptions of the operation as it is now performed. They believed that Holle's additions were actually types of pyloroplasty, and that the known evils of pyloroplasty of dumping and bile reflux would be repeated. However, it should be recognized that pyloric obstruction is now considered as a contraindication for proximal gastric vagotomy. The term proximal gastric vagotomy now is used in preference to its synonyms—highly selective vagotomy, superselective vagotomy, selective proximal vagotomy, or parietal cell vagotomy.

Proximal gastric vagotomy has been used for the treatment of uncomplicated duodenal ulcers, in combination with closure of an acute perforation, or in combination with ligation of the gastroduodenal artery for massive haemorrhage. We propose a limited role for the procedure: namely, a procedure to be undertaken in thin, poorly nourished patients who require surgery because of pain unresponsive to medical therapy. It is contraindicated for duodenal ulcers causing obstruction, for pyloric ulcers, and for gastric ulcers. Although nearly all surgeons agree it is contraindicated for any gastric ulcer, Jordan has combined proximal gastric vagotomy with ulcer excision in selected patients.

The operation comprises division of the branches of the anterior and posterior vagus nerves to the stomach, identified from the lower 7 cm of the oesophagus down to the ‘crow's foot’—the vagal branches in the distal 8 cm of the stomach (Fig. 7) 928. The operation is meticulous and time-consuming.

Because of the high rate of recurrent ulceration, many modifications of the procedure have been suggested. Operative proof of section of the nerves has been demonstrated by stimulation of the denervated lower oesophagus or by intragastric measurements of pH. Dissection of the vagal trunks from the lower 7 cm of oesophagus is necessary to eliminate the ‘criminal nerve of Grassi’; otherwise, appreciable vagal innervation would remain. The nerves running along the right gastroepiploic vessels are also divided by some surgeons, but this has met with little favour. Taylor prefers a posterior truncal vagotomy and an extramucosal division of the branches of the anterior vagus down as low as the crow's foot by means of a long incision through the serosa and mucosa of the anterior stomach wall near the lesser curvature.

The results of this operation have been studied extensively in the laboratory and in clinical practice. One of the most careful and complete investigations was carried out by a collaborative group in Switzerland, France, and Germany under the direction of Allgöwer and Muller. Five years after the original operation, complete investigations were carried out and these were repeated at the end of 10 years. There was an 85 per cent follow-up after 5 years, and three-quarters of the patients also underwent endoscopy. One of their most important findings was a high incidence of asymptomatic ulcers on endoscopy, although the significance of this finding is not clear.

From their observations as well as those of others the following conclusions seem justified. Firstly, the operative mortality is extremely low (<0.5 per cent), and postoperative complications are rare. The development of gastric ulcers can be prevented by careful closure of the peritoneum over the area from which vagal branches have been removed. Secondly, when proximal gastric vagotomy is performed for the indications listed above, the recurrence rate at the end of 5 years is about 10 per cent, although the recurrence rate continues to rise thereafter and, according to Jensen in Copenhagen, is 39 per cent at the end of 15 years.

In 1942 Visick established criteria for judging postoperative results that, after minor modifications, have been widely accepted. They are: Visick I, totally asymptomatic; Visick II, minor symptoms, not interfering with daily activities; Visick III, periods of inability to perform daily tasks, or continuous symptoms interfering with life; Visick IV, totally incapacitated, or secondary operation required for relief of symptoms. Results reported by some surgeons indicate that 5 years after proximal gastric vagotomy 95 per cent of patients are rated Visick I or Visick II. Other observers are not so optimistic: 90 per cent is a more common figure.

Preoperative studies of gastric secretion are of little or no value, except to exclude patients with Zollinger–Ellison syndrome. Postoperatively, basal acid reduction is reduced by about 60 per cent, and secretion due to pentagastrin stimulation by about 50 per cent. Postoperatively, weight loss is uncommon, and weight gain often occurs, blood studies are normal, with no evidence of anaemia, serum calcium levels remain normal, and dumping syndrome, diarrhoea, and bile gastritis are rare. Recurrences are usually associated with only minor symptoms, and most of them can be treated medically. An operation is necessary in about 3 per cent of cases, and antrectomy is safe and has excellent results. In the Basle series, there was a clinical recurrence rate of about 5 per cent symptomatic, asymptomatic recurrences in a total of 14.9 per cent, and later gastric resection in 2 per cent of all patients.

The continued presence of gastric acid prevents colonization of the stomach with Gram-negative bacilli and may be a deterrent to later development of cancer of the stomach.

Pyloroplasty and vagotomy is the simplest surgical treatment for duodenal ulcer and in selected instances it can be used in the treatment of bleeding, obstruction, perforation, and intractability. It also allows a stomach that failed to empty after bilateral vagotomy of any type to drain more effectively. The operation carries low morbidity and mortality rates.

With so many recommendations, it is no wonder that pyloroplasty-vagotomy was the favourite procedure for duodenal ulcer for many years after Dragstedt's popularization of vagotomy. However, as is usual with most operations for peptic ulcer, late complications have reduced its popularity.

The operation usually consists of a Heineke–Mikulicz procedure (Fig. 8) 929. A longitudinal incision is made across the distal stomach and the proximal duodenum and closed transversely, so that the action of the pyloric valve is obliterated. In other cases, when the proximal duodenum is badly deformed by scar, a Finney procedure (essentially a side-to-side gastroduodenostomy) or a Jaboulay procedure (gastric resection plus side-to-side gastroduodenostomy) is used.

Some surgeons report a recurrence rate of 5 per cent. Others found it to be much higher. Herrington's figure was 18 per cent, and the average is about 10 per cent. If reoperation is necessary, the dissection is difficult because of the proximity of the pancreas and the common bile duct. The great majority of surgeons find it unsatisfactory for the treatment of gastric ulcer. The absence of a pylorus leads to a high incidence of alkaline gastritis; this is increased if a simultaneous cholecystectomy is done. Dumping syndrome and diarrhoea are common sequelae. The popularity of this operation has waned, and it has been succeeded by proximal gastric vagotomy or by vagotomy and antrectomy.

The usual procedure involves resection of the distal two-thirds of the stomach followed by either a Billroth I or a Billroth II anastomosis (Fig. 9) 930. In our opinion the operation is neater and more accurate when sutures are inserted by hand, although stapling instruments can be used.

For type 1 gastric ulcer this is a standard operation for patients 70 years of age or over. In younger patients a vagotomy is usually added.

For gastric or duodenal ulcer, relative indications for this operation include obesity, an abnormally high diaphragm, portal hypertension, emergency operations for massive bleeding, or perforation in which a definitive procedure is wise but in which operative conditions are less than optimal. It is also performed for the treatment of long-standing pyloric obstruction in aged patients.

The main disadvantages of this operation include a recurrence rate of at least 2 to 3 per cent of cases, and a weight loss that averages 4 kg in men and 6 kg in women. The recurrence rate is lower than that of any of the standard operations other than vagotomy-antrectomy. Late complications such as motility problems secondary to vagotomy are also avoided.

Surgeons vary in their preference of an elective operation for duodenal ulcer. There is an indication for each of the four operations described above: our preference is usually vagotomy-antrectomy.

There are definite contraindications to proximal gastric vagotomy, as noted above, while contraindications to vagotomy-antrectomy are rare, except that the surgeon should avoid resective procedures, when possible, in thin, malnourished patients, and in those in whom subjective symptoms appear to be out of proportion to objective findings. Women also do less well than do men, particularly in so far as weight loss is concerned.

When a patient apparently has a long life ahead, or one that might take him to countries where drug therapy is not available in case of recurrent symptoms, vagotomy-antrectomy is more likely to fill his needs (Fig. 10) 931. Neurotic patients who magnify symptoms do better with proximal gastric vagotomy.

Patients must be informed of the side-effects of both operations and the high incidence of recurrent ulcer after proximal gastric vagotomy compared with vagotomy-antrectomy.

Postoperative complications may be divided into two major groups—early and late. Early complications follow within a month after the date of initial surgery. Late sequelae occur after that time.

In addition to relatively uncommon problems involving the cardiovascular, pulmonary, hepatic, or renal systems, there are certain complications referable to specific operative procedures on the stomach or duodenum. They include postoperative haemorrhage, suture line leakage, damage to adjacent tissues or organs, and stomal delay.

If there are signs of postoperative bleeding such as falling blood pressure, rising pulse rate and fall in haematocrit soon after operation the surgeon must consider several possibilities. If there is no blood in the aspirate from a nasogastric tube, extragastric bleeding is likely. Splenic rupture or damage to the vessels in the hilum are the most likely causes, particularly following a vagotomy.

If there is blood in the nasogastric aspirate, other causes must be considered. The initial bleeding site may not have been identified or adequately controlled by the first operation. If the initial site has been treated satisfactorily, bleeding from another area in the mucosa is not uncommon in the presence of extensive gastritis. The bleeding may have originated from a suture line. Bleeding from the site of insertion of a gastrostomy tube is not infrequent.

Bleeding usually occurs within the first few hours or the first day after the original operation. Whether or not a second laparotomy must be carried out depends upon the patient's response to conservative measures, including continued nasogastric tube irrigation and aspiration, and intravenous replacement of fluids, electrolytes, and blood. Severe or continued bleeding is an indication for re-exploration. Following a gastrectomy, the stomach is opened, previous suture lines are inspected, and the gastric remnant is examined for other bleeding sites. If a source is not discovered in the stomach, the duodenum is explored.

If all findings, including oesophageal and distal duodenal endoscopy are negative, and the patient has been vomiting blood, haemorrhage probably arose from the stomach. Tiny gastric ulcers or cirsoid aneurysms are the most likely causes, and a generous subtotal gastrectomy (75 per cent) is indicated. This is one of the few cases in which a so-called ‘blind gastrectomy’ has to be performed in the hope that the bleeding lesion will be excised to save the patient's life.

By far the most common problem is duodenal stump leakage following a Billroth II gastrectomy. This catastrophe occurs in 1 per cent of such operations; unless a remedial operation is undertaken immediately, the mortality rate rises to nearly 50 per cent.

Leakage can be avoided or minimized by several methods. If a gastrectomy is planned, but the inflammation about the duodenum appears particularly severe, the surgeon should consider performing a vagotomy and gastroenterostomy. A controlled duodenal fistula can be made at the time of the original operation by catheter duodenostomy (Fig. 11) 932. Alternatively, a drain may be placed near the duodenal stump; if a leak occurs it will probably track into the site of the drain.

Drainage from an incompletely sutured duodenal stump or rupture of the suture line may occur at any time after the operation but is rare after 7 days. Rupture is diagnosed by the sudden onset of severe pain and tenderness in the right upper quadrant. If a drain is already in place, spontaneous drainage of duodenal contents may occur. If no drain has been inserted at the time of operation, the treatment of such a leak is immediate operation and the insertion of drains near the duodenal stump. Attempts to identify the site of drainage in the stump may lead to further damage. A catheter can occasionally be placed in the stump itself.

A leak from a gastroduodenostomy or a gastrojejunostomy is suggested by unusually severe postoperative local pain and tenderness. If it is detected at an early stage, the opening can be repaired and the suture line can be bolstered with omentum; at a later stage, a fistula is almost certain to develop. Fistulae with only a small output may heal spontaneously if nasogastric suction and hyperalimentation are used. Other fistulae will require surgical repair.

Gastric resections for duodenal ulcer may be followed by pancreatitis if the pancreas has been damaged. Fortunately, pancreatitis is usually mild and subsides after conservative treatment. Vagotomy Vagotomy, in addition to splenic injury, may lead to perforation of the oesophagus or rupture of the diaphragm. If an oesophageal perforation is suspected, an immediate drink of Gastrografin is indicated. An oesophageal perforation requires closure by overlapping the gastric fundus or by suturing a pleural flap or omentum over the defect. Wide drainage of the left upper quadrant and often of the left pleural cavity is essential.

If a gastroduodenostomy or gastrojejunostomy is not functioning normally 10 days after operation, the patient has stomal delay. A Massachusetts General Hospital study showed this to occur after gastric resection in 4.7 per cent of patients; the addition of truncal vagotomy increased the incidence to 8.6 per cent. Function is likely to be regained slowly in elderly patients who received preoperative treatment with cimetidine, who have had a long period of obstruction, or who have had a vagotomy combined with a procedure on the stomach. A low serum potassium or albumin level can lead to malfunction. Starch peritonitis has also been implicated; fat necrosis in a large omentum, and mechanical obstruction from adhesions just distal to a gastrojejunostomy are other causes.

Endoscopy should be performed to ascertain that the anastomosis is patent; some authors have reported that gentle passage of the scope into and down the efferent loop produces immediate improvement (Fig. 12) 933. The patient is usually treated with central alimentation and careful observation; medications such as bethanecol (Urecholine) or metoclopramide tend not to be helpful. Reoperation is recommended if there is no improvement in 2 weeks.

Acceptable mortality rates are 0.5 per cent for elective proximal gastric vagotomy and 1 per cent for gastric resections.

The late sequelae of operations for duodenal and gastric ulcer vary, depending upon the substance of the original procedure. The majority follow some type of gastrectomy; others are due to vagotomy or to loss of the pyloric sphincter. They can be roughly classified into three groups—those chiefly of historical significance, those that occur today but are rare, and those that occur today in 5 per cent or more of patients or are of current interest.

These include the afferent loop syndrome, a mechanical problem that arises when a long jejunal afferent loop is used for a Billroth II resection or as a result of an inadequate anastomotic opening. Typical symptoms are sudden attacks of severe vomiting of bile, unmixed with food in acute cases. Persistent dilatation of the afferent loop leads to retention of duodenal contents, secondary proliferation of bacteria, and symptoms, such as anaemia and malnutrition, suggestive of a blind loop syndrome. Surgical conversion to a Billroth I anastomosis is curative.

Sequelae due to technical errors, such as gastroileostomy rather than gastrojejunostomy, are also of historical interest. They result in severe diarrhoea and late defects in absorption of food, including vitamins. A beri-beri syndrome has been reported. The error is most likely to occur when there is non-rotation of the small intestine and there is no ligament of Treitz.

Microcytic anaemia was common when the duodenum was bypassed and the oral intake of iron was poor. Macrocytic anaemia can follow high subtotal resections; it is universal after total gastrectomy because the loss of intrinsic factor prevents the absorption of vitamin B&sub1;&sub2;. Regular injections of cyanocobalamin (vitamin B&sub1;&sub2;) are necessary for the remainder of the patient's life.

Absorption of fats, proteins, and/or carbohydrates decreases directly with an increase in the amount of stomach resected. Iron absorption can be diminished following construction of Billroth II anastomoses since the duodenum is a major site of iron absorption. Calcium can be lost due to persistent diarrhoea, with consequent osteoporosis. An increased incidence of certain diseases such as tuberculosis has been reported after extensive gastrectomy in poorly nourished patients.

For the interested reader a detailed discussion of these problems was made in 1963 by Stammers and Williams. Their careful studies served to reduce the standard amount of stomach that was resected for ulcer disease.

These include technical errors: an unusually small gastroduodenostomy can result in symptoms of partial obstruction with epigastric fullness, heartburn, and episodes of vomiting; conversion of a Billroth I to a Billroth II anastomosis relieves symptoms. An inadequate gastrojejunostomy also requires a secondary operation.

Vagal diarrhoea is a complication in which the aetiology is not clear, but severe, almost uncontrollable explosive diarrhoea occurs after about 1 per cent of bilateral truncal vagotomies. Other causes of diarrhoea, such as colonization of the stomach with Gram-negative bacilli or giardiasis, must be excluded. Herrington and Sawyers have successfully employed a reversed loop of the jejunum for the treatment of a few such patients.

A patient with dental plates and a Billroth II anastomosis may unknowingly swallow a large amount of undigestible material such as orange pulp, and develop intestinal obstruction from a bezoar. After truncal vagotomy, endoscopy may disclose undigested masses of vegetable material in the stomach; most are of no significance.

Small gastric reservoirs can lead to malnutrition. Such operations are now rarely performed for ulcer disease, except when total gastrectomy is necessary for hypergastrinaemia, when they are usually tolerated well. In some instances Scott has proved the value of a gastric pouch fashioned with two to three loops of jejunum: this can improve the ability of a patient to eat larger meals.

Some sequelae occur today in 5 per cent of patients or are of current interest. These are of variable significance and range from potentially serious, such as a persistent or recurrent ulcer (Table 2) 304, to those in which symptoms appear to be functional with no accompanying pathological change. Precise definition of some of these syndromes is difficult and the relative emphasis placed upon them by different observers differs. We prefer to emphasize persistent or recurrent ulceration because of its potential for future serious problems and because it can be confirmed by objective rather than by subjective evidence.

Careful questioning of patients who have undergone gastric surgery uncovers a large number of symptoms, many of which can also be elicited from individuals who have never had an operation. The incidence and significance of dyspeptic symptoms, such as postprandial fullness, weight loss, loss of appetite, and nausea, that are secondary to the operative procedure is, therefore, hard to assess.

Satisfaction after an operation for duodenal ulcer is usually about 90 per cent, regardless of the procedure performed. Many patients do, however, have minor symptoms that put them into Visick grade II. Müller found that 64 per cent of patients were grade I (asymptomatic) following proximal gastric vagotomy, while 28 per cent were grade II (minor symptoms). Hoffman found that 15 per cent of patients were Visick class II, regardless of the type of vagotomy used.

The percentage of patients in Visick class II appears to be nearly the same after vagotomy-antrectomy as after proximal gastric vagotomy, truncal vagotomy-pyloroplasty, and selective vagotomy with pyloroplasty. It is only in the proportion of more unfavourable results (Visick III or IV) that there is a variation between the results of different procedures.

For many years postoperative recurrent ulcers have been regarded as the cardinal sign of an unsuccessful operation. Some doubt has been cast upon this conclusion by the fact that many recurrences following proximal gastric vagotomy can be treated successfully with H&sub2;-receptor antagonists. Such treatment may, however, not be readily available in developing countries, where the first operative procedure needs to be curative.

Absolute figures concerning recurrences after the same type of operation often show wide variations, and these depend upon factors that are impossible to analyse. Some of these variations are shown in Table 2 304. However, certain generalizations are warranted.

The recurrence rate after gastric resections for type 1 gastric ulcer, even without vagotomy, is negligible; other late sequelae are also minimal. Pyloric ulcers, prepyloric ulcers, and gastric ulcers associated with duodenal ulcers do not respond well to proximal gastric vagotomy; they should be treated by antrectomy plus bilateral truncal vagotomy.

The recurrence rate after gastric resection and bilateral truncal vagotomy is 1 per cent or less after 5 years and does not appear to increase later. The highest rate of recurrence is after proximal gastric vagotomy, and this appears to rise with time.

Billroth I and II gastric resections have a similar rate of recurrence if a bilateral truncal vagotomy is added; without a vagotomy, the recurrence rate after a Billroth II is lower.

Endoscopic examination 5 years after proximal gastric vagotomy shows that active ulcers are present in approximately twice as many individuals as those who have symptoms. The fate of these silent ulcers is not known.

Recurrent ulcers after proximal gastric vagotomy can be successfully treated by a gastric resection.

This syndrome consists of a combination of one or more of the following: fall in blood pressure, sweating, pallor, abdominal distension, cramps, diarrhoea, weakness, and a desire to lie down. Early dumping occurs within 15 to 30 min after eating; late dumping occurs about 2 h after eating.

Early dumping occurs in some persons who have not had an operation. However it is accentuated by any mechanism that destroys the action of the pyloric valve, so that gastric contents are discharged rapidly into the duodenum or jejunum. Liquids with a high osmolality, such as a chocolate milk shake, are most likely to initiate symptoms. It is most common after gastric resections and least common after proximal gastric vagotomy. It occurs in about 25 per cent of patients who have had a gastric operation other than proximal gastric vagotomy in the early weeks after operation and tends to disappear with time.

Preoperative administration of a hypertonic glucose solution may identify those who are likely to suffer from postoperative dumping. The mechanism is believed to be the rapid fluid and electrolyte shift into the intestinal lumen in response to this challenge. Increased activity of hormone secreting cells in the intestine also may be a factor.

Medical therapy of the syndrome consists of solid rather than liquid meals, small but repeated servings, and a low intake of carbohydrates and of dairy products. Additional aids, such as separate intake of solids and liquids, and avoidance of very warm or very cold liquids, can help. After a patient finds that certain foods are tolerated well, others may be added one at a time.

Surgical treatment has generally been unsatisfactory. The operation most likely to succeed consists of an interposition of a retroperistaltic 10-cm loop of jejunum between the distal end of the stomach and the duodenum; a vagotomy must be added if one has not already been performed.

Because dumping symptoms tend to disappear after months or even several years, a conservative attitude is warranted in the treatment of these patients.

Late dumping occurs 2 to 3 h after eating, and affects only 2 per cent of patients. The symptoms are also related to carbohydrate ingestion; 2 h after eating the hyperglycaemia induces an oversupply of insulin which then leads to a reactive hypoglycaemia attended by symptoms of prostration and sweating. Moderation of carbohydrate intake usually controls the symptoms.

The causes and classification of postgastrectomy symptoms are severe enough to interfere with an individual's lifestyle and are extremely difficult to analyse. Nowhere is the problem more difficult than with alkaline gastritis.

This syndrome arises because of regurgitation of bile and duodenal contents back into the stomach. The existence of such a syndrome is proven by the fact that diversion of bile relieves the symptoms, yet no individual symptoms or objective findings serve to distinguish the syndrome prior to operation. Epigastric distress, particularly after eating, nausea, and occasional vomiting and/or diarrhoea are common but totally non-specific. What appears to be obvious gastritis to the endoscopist is often normal according to the pathologist.

In an attempt to find some preoperative test that would help to make the diagnosis, Warshaw injected various solutions into the stomach through a nasogastric tube. Patients with the syndrome developed pain after injection of a strong alkali solution, but not after saline or acid infusion. Such patients appeared to have better results from a reparative operation than others who had a negative test.

The syndrome is most common after a Billroth II gastrectomy, but is nearly as common after a Billroth I or pyloroplasty. It occurs rarely after a proximal gastric vagotomy. Since the symptoms often tend to diminish with the passage of time, a conservative approach is indicated.

To detour upper intestinal contents from the stomach, a Roux-en-Y anastomosis is made (Fig. 13) 934. The jejunum must be divided and the proximal end anastomosed to the descending jejunum 45 cm below the stomach. Although this procedure received great enthusiasm when it was first introduced, a satisfactory result is found in less than two-thirds of patients treated in this way. Some of those with unsatisfactory results prove to have motility problems that may be helped with a higher or even a total gastric resection. However, a large proportion of the unsatisfactory results occur in patients with multiple psychiatric complaints. In some of these the original operation appears to have been performed after inadequate evaluation.

Disturbances of the normal motility of the stomach and/or duodenum can lead to unduly rapid emptying or to paresis. Rapid emptying, known as dumping, is considered above. Slow emptying of the stomach can lead to stomal delay, obstructive vomiting, the formation of gastric bezoars, poor nutrition, and loss of appetite and weight.

Primary motility problems involving the stomach are rare, but they often appear after surgery, particularly when a vagotomy has been performed. In patients who have not undergone any operative procedure, diseases such as diabetes, scleroderma, amyloid disease, central nervous system abnormalities, and Chagas' disease must be considered. In this section disturbances secondary to operations for gastric or duodenal ulcer will be described.

Motility of the stomach is activated by the vagus nerve. Little peristalsis is demonstrable on radiographic studies of the fundus, even in normal persons. One or more pacemakers near the lower portion of the fundus initiate impulses which lead to antral peristalsis and emptying of the stomach. Because the pacemakers are activated by the vagus nerves, vagotomy without pyloroplasty or gastroenterostomy often causes emptying of the stomach to become unco-ordinated and weak, and marked gastric dilatation can occur.

When food reaches the duodenum another pacemaker in the proximal portion institutes peristaltic waves that progress smoothly down through the duodenum and the jejunum. Interruption as, for example, by a Roux-en-Y loop of jejunum, creates an additional factor that leads to poor gastric emptying. If any of the necessary anastomoses are technically inadequate, a mechanical factor contributes to the problem.

The diagnosis of inadequate emptying of the stomach can be made at an early stage by barium studies. Radionuclides may be mixed with food and the amount and rate of emptying quantitated.

Treatment includes small meals at regular intervals. Urecholine (bethanechol) is occasionally effective. We have not found metoclopramide particularly valuable, but have had no experience with domperidone. Erythromycin has proved valuable in treatment of diabetic gastroparesis and could be helpful in some cases.

Surgical therapy is required only rarely. Vogel has advised the use of a higher subtotal gastrectomy with a Roux-en-Y anastomosis. We have had one patient in whom a total gastrectomy provided a spectacular cure after repeated high resections and anastomoses had failed.

Several operations on the stomach lead to reflux of intestinal contents into the stomach or the stomach remnant. ‘Intestinalization’ of the mucosa of the stomach (a change from the histologic picture of gastric mucosa to that of intestinal mucosa) follows in many instances and is believed to be a precursor to the development of cancer.

Cancer of the gastric stump (colloquially called ‘stump cancer’) has been studied extensively. Although there is an increased risk of developing gastric cancer in patients who have had a gastrectomy, compared with the risk of the normal population, this increase is not great and, in view of the rarity of gastric cancer in the United States today, presents no important threat. Stump cancer does not appear for at least 10 years after a gastrectomy, and the appearance of most cases is even more delayed. Stump cancer is not frequent enough to warrant routine endoscopies.

These lesions were described originally by Harvey Cushing in 1932. They consist of deep, frequently perforated ulcers that arise in the oesophagus, stomach, or duodenum. They occur most typically after neurosurgical illnesses.

These are typical duodenal ulcers which essentially occur in burned patients. When they were first described (1842) duodenal ulcers were uncommon. Whether they represented previous ulcers activated after the burn, or whether they are associated with true stress ulcers is not clear.

Aspirin is a prime example of an irritant. In addition to its effect on platelet adherence, it directly irritates the gastric mucosa and leads to superficial ulcerations. Ibuprofen and alcohol provoke similar local injuries. Accidental or suicidal ingestion of acids or alkalines can lead to rapid mass destruction and perforation of the oesophagus and/or stomach. Alkalis exert their effect mainly on the oesophagus and acids mainly on the body of the stomach.

Whether or not steroids lead to ulceration of the stomach is controversial. However, there appears to be a clear association between such therapy and perforations of the colon.

True stress ulcers are superficial lesions that appear after burns, sepsis, severe trauma, protracted hypotension, or serious operations. They may involve only the mucosa (and are often therefore called mucosal erosions), or they may penetrate into the submucosa. Perforation is rare; massive bleeding is the usual complication. Stress ulcers are always found in the fundus, but many also occur throughout the stomach and proximal duodenum.

Prevention of these ulcers is important: bleeding carries a high mortality rate. Treatment includes prevention of gastric distension (usually by nasogastric intubation and suction), respiratory care, and the administration of either antacids, H&sub2;-receptor antagonists, or mucosal protectants. Controlled studies suggest that antacids are the most effective treatment, but intravenous cimetidine is easier to use and is more economical of nurses' time. Another factor that still needs investigation is whether attempts to make the stomach achlorhydric and hence, poorly resistant to growth of bacteria, can lead to colonization of the normally sterile stomach with enteric organisms. If so, aspiration of gastric contents could lead to devastating pneumonitis.

Diagnosis of stress ulcers can be made by endoscopy. The endoscope can also be used to treat bleeding ulcerations, by electrocoagulation which is safer than the laser beam.

Selective arteriography shows extreme dilatation of arterioles particularly in the fundus of the stomach. However, embolization or even selective perfusion with vasopressin is dangerous because of the possibility of necrosis of the gastric wall.

Surgical therapy of these ulcers has been quite unsatisfactory for several reasons, not least of which is the fact that the patients are already very ill. A high subtotal gastrectomy and vagotomy is usually preferred, but even this may be followed by bleeding from the gastric remnant. The alternative, total gastrectomy, is an operation of considerable magnitude but must be chosen if the condition cannot be controlled by less radical measures.

Fifty years ago ulcerating cancers of the stomach frequently were confused with benign ulcers. Today endoscopic examinations and biopsies have augmented barium contrast studies of the stomach, and the incidence of cancer of the stomach has declined spectacularly in frequency in the United States as well as in western Europe. This problem of diagnosis now arises only rarely. Nevertheless, endoscopists have been known to miss a lesion in the gastric fundus or may fail to obtain a satisfactory biopsy.

Complete healing of an ulcerating cancer occurs rarely, if ever. There is no evidence that a benign gastric ulcer can degenerate into cancer. Therefore, it is necessary to be certain that a gastric ulcer that appears to be benign actually heals. Furthermore, if a gastric ulcer recurs after therapy the possibility of underlying malignancy is increased.

Malignant lymphomas of the stomach are prone to develop superficial ulcers; however some may be deep and may perforate. Because these ulcers may respond to the medical therapy for a benign ulcer, unless biopsy specimens are taken, the diagnosis may be missed. Barium studies may show a thick-walled stomach with hypertrophic mucosal folds. A CT scan may demonstrate enlarged lymph nodes in the area of the coeliac axis. The preferred treatment is gastric resection followed by radiation and/or chemotherapy.

The definition of hypergastrinaemia rests on laboratory studies which show an elevated serum gastrin level. Gastrinomas and the retained antrum syndrome are the only common causes that are surgically important.

The first description of such lesions was made by Zollinger and Ellison in 1955, who reported patients with the combination of severe peptic ulcers, often located in unusual locations, excessive amounts of gastric secretion of high acidity, and non-&bgr;-cell pancreatic tumours. The definition of the Zollinger–Ellison syndrome was modified in ensuing years to include less virulent ulcers. Tumours may be found outside the pancreas or duodenum, and about 25 per cent of patients have the multiple endocrine neoplasia type I syndrome and no tumour in the peripancreatic area.

Clinical symptoms typical of severe peptic ulcer disease—epigastric pain, weight loss, and vomiting—occur in 90 per cent of patients; severe diarrhoea occurs in one-third. The symptoms may progress slowly or very rapidly. The most important laboratory finding is an elevated serum gastrin level: the upper normal limit is 150 pg/ml, but patients with the Zollinger–Ellison syndrome may have levels over 350 pg/ml at some time of the day. Additional diagnostic evidence is provided by determination of the basal acid output of the stomach, which is not normally over 4 mEq/h for women and 6 mEq/h for men. Although the normal maximal acid output after stimulation by Histalog (a histamine analogue) is 30 mEq/h for women and 40 mEq/h for men, a level of over 15 mEq/h raises the suspicion of a gastrinoma. The diagnosis is confirmed if the injection of intravenous secretin elevates serum gastrin to at least 200 pg/ml. According to Townsend and Thompson, various other causes of hypergastrinaemia, including G-cell hyperplasia, retained antrum, pyloric obstruction, pernicious anaemia, and duodenal ulcers, do not cause this rise after injection of secretin. Gastrin levels may also be elevated after treatment with H&sub2;-receptor antagonists and omeprazole.

Since over 60 per cent of peripancreatic gastrinomas are malignant, and the major symptoms arise from the severe ulcer disease, the treatment of gastrinomas for many years consisted of total gastrectomy. However, excision of some gastrinomas may result in permanently normal gastrin levels, and cure of the patient without any procedure on the stomach. The tumour may be located preoperatively by CT scanning, ultrasound, selective angiography, and selective venous sampling. These measures generally fail to identify small tumours, such as the micropolypoid gastrinomas, only 1 to 8 mm in diameter, which are found in the duodenum.

Symptoms associated with excessive acid secretion have been controlled by the administration of H&sub2;-receptor antagonists or omeprazole; such medical therapy has to continue throughout the life of the patient. At the present time, the pendulum has swung back to surgical therapy since a gastrinoma may be discovered and the patients may be cured by its excision.

If the tumour cannot be found or if liver metastases are present, some type of surgery must be undertaken on the stomach; these procedures have ranged from proximal gastric vagotomy to total gastrectomy. Total gastrectomy is the procedure that is most likely to succeed: postoperatively, patients often gain weight. Townsend and Thompson are strong advocates of total gastrectomy for all of these patients; there were no operative deaths in their series. Even in the presence of metastatic disease, progression of the tumour may be retarded by chemotherapy consisting of 5-fluorouracil and streptozotocin; somatostatin analogues are also effective.

An easy solution to the problems of the difficult duodenum encountered in Billroth II resections has eluded many surgeons. As explained previously, our preference is to make a catheter duodenostomy.

In 1895, von Eiselsberg suggested that the treatment of inoperable cancer should involve exclusion and closure of the antrum, which is left in situ. This relatively simple exclusion procedure was later applied by some surgeons to the treatment of peptic ulcer; unfortunately antral exclusion was followed by a high incidence of gastrojejunal ulcers. In a personal series treated by Allen in the late 1930s, seven out of eight patients with duodenal or prepyloric ulcers developed anastomotic ulcers within a year.

The same problem proved to be a serious argument against the two-stage gastrectomy of McKittrick. Such patients underwent Billroth II resections with antral exclusion at the first stage, and excision of the antrum at the second. If the second operation was not performed within 6 weeks, nearly all of these patients developed a gastrojejunal ulcer.

Dragstedt proved that the antrum, when it was in an alkaline rather than an acid medium, produced unusually large amounts of gastrin. If a vagotomy had not been performed, stimulation of the parietal cells resulted in excess acid production and a new ulcer.

Even the addition of bilateral truncal vagotomy to the two-stage operation, as carried out by Waddell and Bartlett, proved to be dangerous. Other complications such as intussusception of the antrum followed and the operation was discarded.

The conclusion is clear: a retained antrum is ulcerogenic, and whenever a patient who has undergone a previous gastric resection for ulcer disease is operated on for a recurrent ulcer, the surgeon must look carefully and must excise any antral tissue that remains proximal to the duodenum.

In addition to gastrinomas and the retained antrum syndrome there are many other unusual causes of hypergastrinaemia. G-cell hyperplasia is characterized by increased production of gastrin by the antral mucosa because of an increase in number of gastrin-producing cells. To establish the diagnosis, other unusual causes of hypergastrinaemia, such as pernicious anaemia, achlorhydria, atrophic gastritis, renal failure, and treatment with H&sub2;-receptor antagonists must be eliminated; the reasons for the hypergastrinaemia seen in these conditions is not clear.

Donahue PE, Bombeck CT, Condon RE, Nyhus LM. Proximal gastric vagotomy versus selective vagotomy with antrectomy: results of a prospective, randomized clinical trial after four to twelve years. Surgery, 1984; 96: 515–91.

Edwards LW, Herrington JL Jr. Vagotomy and gastroenterostomy—vagotomy and conservative gastrectomy. Ann Surg, 1953; 137: 873– 83.

Emas S, Fernstrom M. Prospective, randomized trial of selective vagotomy with pyloroplasty and selective proximal vagotomy with and without pyloroplasty in the treatment of duodenal, pyloric and prepyloric ulcers. Am J Surg, 1985; 149: 236–43.

Farmer DA, Howe CW, Porell WJ, Smithwick RH. The effect of various surgical procedures upon the acidity of the gastric contents of ulcer patients. Ann Surg, 1951; 134: 319–31.

Fischer AB. Twenty-five years after Billroth II gastrectomy for duodenal ulcer. World J Surg, 1984; 8: 293–302.

Goligher JC. A technique for highly selective (parietal cell or proximal gastric) vagotomy for duodenal ulcer. Br J Surg, 1974; 61: 337–45.

Goligher JC, Hill GL, Kenny TE, Nutter E. Several standard elective operations for duodenal ulcer. Ten to 16 year clinical results. Br J Surg, 1978; 65: 145–51.

Graham DY. Treatment of peptic ulcers caused by Helicobacter pylori. N Engl J Med 1993; 328: 349–50.

Hentschel E, et al. Effect of ranitidine and amoxicillin on the eradication of Helicobacter pylori and the recurrence of duodenal ulcer. N Engl J Med, 1993; 328: 308–12.

Herrington JL Jr. Historical aspects of gastrointestinal surgery. In Scott HW Jr, Sawyers JL, eds. Surgery of the stomach, duodenum and small intestine. London: Blackwell, 1987: 3–24.

Hoffmann J, Jensen H-E, Christiansen J, Olesen A, Loud FB, Heauch O. Prospective controlled vagotomy trial for duodenal ulcer; results after 11–15 years. Ann Surg, 1989; 209: 40–5.

Johnson HD. Gastric ulcer. Ann Surg, 1965; 162: 996–1004.

Jordan PH Jr. Duodenal ulcers and their surgical treatment: where did they come from? Am J Surg, 1985; 149: 2–14.

Muller C, Martinoli S. Die proximal-selektive vagotomie. Berlin: Springer-Verlag, 1985.

Nyhus LM, Wastell C. Surgery of the stomach and duodenum. 4th edn. Boston: Little Brown and Co., 1986.

Scott HW Jr, Sawyers JL. Surgery of the stomach, duodenum, and small intestine. London: Blackwell, 1987.

Stammers FAR, Williams JA. Partial gastrectomy. London: Butterworths, 1963.

Tanner NC. The surgical treatment of peptic ulcer. Br J Surg, 1964; 51: 5–23.

Taylor T. Lesser curve superficial seromyotomy—an operation for chronic duodenal ulcer. Br J Surg, 1979; 66: 733–7.

Townsend CM Jr, Thompson JC. Up-to-date treatment of the patient with hypergastrinemia. Adv Surg, 1987; 20: 155–81.

Visick AH. A study of the failures after gastrectomy. Ann R Coll Surg Engl, 1948; 3: 288–94.

Welch CE. Surgery of the Stomach and Duodenum. 5th edn. Chicago: Yearbook Publishers, 1973.

Welch CE. Gastric Resection for duodenal ulcer. In Scott HW Jr, Sawyers JL. Surgery of the Stomach, Duodenum, and Small Intestine. Ch. 35. London: Blackwell Scientific Publications, 1987: 681–95.

Welch CE, Rodkey GV, Gryska PR. A thousand operations for ulcer disease. Ann Surg, 1986; 204: 454–67.