Br. J. Surg. 1987, Vol. 74. October, 900-904

J. P. Britton, D. Johnston, D. C. Ward, A. T. R. Axon and M. C. J. Barker

University Department of Surgery, Departments of Nuclear Medicine and Gastroenterology and Gastric Follow Up Clinic. The General Infirmary, Leeds. UK Correspondence to: Professor D. Johnston, University Department of Surgery, The General Infirmary, Leeds LS1 3EX, UK

Gastric emptying and clinical outcome after Roux-en-Y diversion

The results of 48 Roux-en-Y ( R Y ) diversion procedures are reported: 41 were performed as secondary procedures and 7 as part of a primary operation for peptic ulcer. There was no operative mortality, but four patients developed temporary fistulae in the postoperative period and three patients required reoperation. Good clinical results were found when RY diversion was performed as a primary procedure or when the indication for operation was peptic ulceration. The overall results, however, were poor: 24 patients (50 per cent) felt that they had not benefited and 32 patients (67 per cent) remained in Visick grades III or ZV. The main cause of failure was gastric stasis, especially of solid food. Gastric emptying studies were carried out after RY diversion in 22 patients, most of whom had symptoms of stasis. Emptying of liquids was found to be normal in most patients, but emptying of solids was delayed, the median ti for solids being 160 (75-370) min compared with 67 (50- 85) rnin in DU patients. Bilious vomiting improved signi$cantly after RY diversion, but 18 patients (38 per cent) complained of vomiting food and 32 patients (67 per cent) experienced postprandial distress or pain. Loss of the antral mill, vagotomy of the gastric remnant and, perhaps, resistance to gastric emptying by the Roux loop itself may together explain the delay in gastric emptying of solids after RY diversion. Keywords: Alkaline reflux gastritis, Roux-en-Y diversion, gastric emptying

Excessive reflux of duodenal or jejunal content into the stomach is common after both partial gastrectomy (PG) and vagotomy with a drainage procedure (V+D). Though many patients remain asymptomatic in spite of such reflux, others develop symptoms such as epigastric pain, nausea and bilious

Roux-en-Y (RY) diversion, first described by Roux in 1897’, is still widely used in the treatment of bilious vomiting after PG or V + D, and has been reported to yield good results, provided that the diverting loop of jejunum is 4&50 cm long6-*.

It is somewhat unusual, however, for bilious vomiting to occur as an isolated symptom. More often, patients who require revisional surgery after PG or V + D complain of several symptoms, some of which may be unrelated to enterogastric reflux. Our impression at a gastric follow-up clinic was that, though successful in relieving bilious vomiting, RY diversion often yielded disappointing results. In particular, symptoms suggestive of impaired gastric emptying, such as pain or bloating after meals, nausea and vomiting, seemed to be common sequelae. We wondered whether RY diversion was solving one clinical problem at the price of creating other problems. For that reason, we reviewed the outcome of RY diversion in 48 patients, 22 of whom also underwent tests of gastric emptying.

Patients and methods Patients Fortyeight patients underwent RY diversion between 1975 and 1986. Many of them had been referred from elsewhere as failures of gastric surgery. There were 23 men and 25 women, aged from 35 to 75 years (median, 50 years). In 22 patients this was the first reoperation, but 19 patients had undergone 2 or more previous gastric operations (Table I). In seven patients, RY diversion formed part of a primary operative procedure. The median interval between the first operation and RY diversion was 9 years (interquartile range 4-15 years).

Diagnosis Before RY diversion was undertaken, a precise diagnosis of the cause of the patient’s symptoms was sought by means of the history and

endoscopic, radiological and gastric emptying studies. The indications for RY diversion (Table I ) were, predominantly, enterogastric reflux in 28 patients, gastric stasis in 7 and peptic ulceration in 13. Many patients, however, had a variety of symptoms such as pain, nausea, dumping and diarrhoea, such that the neat classification into three groups is somewhat arbitrary. Patients diagnosed as having enterogastric reflux had both the macroscopic appearances of biliary gastritis at endoscopy and the microscopic features of the conditiong. Patients with gastric stasis had been found to have food residues in the stomach at endoscopy after an overnight fast, and gastric emptying studies showed delayed emptying. The 13 patients with peptic ulceration comprised 3 patients with pyloric stenosis, 3 with combined prepyloric and duodenal ulceration and 7 with recurrent peptic ulceration after V + D or PG.

Operative procedure Most of the operations were performed by the senior author or his senior registrar. The operative procedure itself usually consisted of a truncal vagotomy with antrectomy, closure of the duodenal stump and formation of a 50 cm RY loop ofjejunum, which was anastomosed to the gastric remnant. In most cases an antecolic anastomosis was fashioned using two layers of continuous chromic catgut and the efferent Ioop sutured to the greater curve. Normally the whole width of the stomach was used to provide a wide stoma. Thirty patients had previously undergone vagotomy (including four who had undergone revagotomy) and six patients had a primary vagotomy added to the RY procedure. Twelve patients did not undergo vagotomy, either at the initial operation or at the time of RY diversion. The antrectomy was relatively conservative, the mean weight of the resected gastric antrum in 15 patients being only 73 g (35-15Og): thus, a relatively large gastric remnant was left behind.

Assessment of results Clinical results. Patients were reviewed annually at a gastric follow-

up clinic, where they were interviewed by a physician and surgeon who did not know what type of operation had been performed. The duration of follow-up after RY diversion ranged from 4 to 132 months, with a mean and median of 26 and 24 months, respectively. The duration of follow-up after the first gastric operation was obviously much longer, except in patients who underwent RY diversion as a primary procedure. Individual symptoms were recorded and the patient’s overall state graded according to the modified Visick classification. The symptoms and Visick status at most recent follow-up were compared with the

900 0007-1323/87/100900-05%3.00 0 1987 Butterworth & Co (Publishers) Ltd

Gastric emptying: J. P. Britton et al.

Table 1 Indication for Roux-en-Y diversion procedure in 48 patients with details of previous operations

Indication

Roux-en-Y as primary or

Nature of previous operations

Number of revisional Vagotomy + Partial patients procedure Vagotomy partial gastrectomy gastrectomy

Enterogastric reflux 28

Gastric stasis 7

Primary or recurrent 13 peptic ulceration pyloric stenosis

Primary 2* Revisional 26 12

Primary 0 Revisional 7 5

Primary 5 Revisional 8 7

10

0

1

* These patients had previously undergone cholecystectomy

Table 2 Patients’ symptoms before and after Roux-en-Y diversion

Vomiting Postprandial Number of distress or Epigastric Gastro-oesophageal

Group patients All types Bile Food fullness pain Nausea reflux or heartburn

Group 1 Enterogastric reflux 28

Before RY 20 20, 3 7 25 22 15 After RY 10 5 9 20 18 17 10

Group 2 Gastric stasis 7

Before RY 6 2 5 5 5 7 7 After RY 5 3 5 5 5 4 1

Group 3 Peptic ulceration 13

Before RY 6 3 5 4 12 10 7 After RY 4 1 4 7 4 4 2

Total 48 Before RY 32 25, 13 16 42 39 29, After RY 19 9 18 32 27 25 13

* Significant difference P<@OOl

patient’s symptoms and Visick grade before RY diversion. Each patient was asked whether or not the operation had been beneficial. Endoscopy was performed if the patient was symptomatic to exclude recurrent ulceration or stenosis at the stoma.

Gastric emptying studies. Twenty-two patients underwent detailed studies of gastric emptying after RY diversion. They were not representative of the entire group of 48 patients, since most of them (19 of 22) were suffering from troublesome symptoms such as vomiting, pain and postprandial satiety and were regarded as having a poor result after RY diversion (Visick grades I11 or IV).

The studies were carried out in the Department ofNuclear Medicine using technetium, 99mTc, labelled liquid and solid meals. Serial images were obtained using a gamma camera (Elscint Dymax LF, Berinsfield, Oxfordshire, UK) interfaced to a digital computer (Digital Equipment Corporation, Reading, UK, PDP-11. In the studies of emptying of liquids, the patient drank 200 ml of water to which was added 10 MBq of 99mTc-DTPA. Thirty frames, each of 1 min duration, were acquired. Using a region of interest over the stomach, seen on early frames, an activity-time curve was generated. One hour after completion of the liquid meal, the patient ate a solid meal consisting of one scrambled egg on two rounds of toast with 25 g of butter and a cup (150ml) of tea. During cooking 10 MBq of 99mTc labelled macro-aggregated albumen was added to the egg. Thirty lmin frames were acquired from the beginning of the meal and activity-time curves generated. A further 30min of data were acquired (‘late phase’) beginning 1 hour after the start of the meal. The liquid and late phase solid emptying curves were both assumed to bemono-exponential and, using a curve fitting routine, the times taken for half the meal to leave the stomach ( t i ) were calculated.

Statistical analysis. Results were analysed by the Mann-Whitney U test and x 2 test, with Yates’s correction where necessary.

Results Operat ive mortal i ty There was no operative mortality. One patient has since died of cerebral haemorrhage.

Postoperative morbidity Eight patients developed specific complications of the RY diversion. Two patients suffered significant haemorrhage from the suture line, one requiring reoperation. Two patients were shown to have a physical obstruction at the gastrojejunal anastomosis which required refashioning. Three patients developed an upper gastrointestinal fistula and one a faecal fistula but all eventually closed with conservative management.

Functional results (Table 2) In group 1 patients with enterogastric reflux (28 patients), bilious vomiting improved significantly after operation ( P < 0.001), though five patients still experienced this symptom to a lesser degree. Other symptoms, however, were not improved. In particular, symptoms suggestive of gastric stasis such as food vomiting, postprandial fullness, epigastric pain and nausea were all common. In addition, 16 patients complained of inability to eat a normal sized meal.

Group 2 patients with gastric stasis (7 patients) showed no significant improvement in their symptoms after RY diversion.

In group 3 patients with peptic ulceration (13 patients), symtoms improved after RY diversion through several still

Br. J. Surg., Vol. 74, No. 10, October 1987 901

Gastric emptying: J. P. Britton et al.

Table 3 Results of Roux-en-Y diversion expressed according to the Visick classification

Pre-operative Postoperative Visick grade Visick grade

Group patients I I1 111 IV I 11 111 IV Number of

Group 1 Enterogastric reflux 28 0 0 8 20 5 3 812

Group 2 Gastric stasis 7 0 0 1 6 1 1 1 4

Group 3 Peptic ulceration 13 0 0 0 1 3 2 4 6 1

Total 48 0 0 9 3 9 8 8 1 5 1 7 ~~~ -

The pre-operative Visick grade is indicated

Table 4 Patients’ assesment of the outcome of Roux-en-Y diversion

Number of Group patients Improved Not improved

Group 1

Group 2

Group 3 Peptic ulceration 13 12 (92) 1 (8)

Total 48 24 (50) 24 (50)

Enterogastric reflux 28 10 (36) 18 (64)

Gastric stasis 7 2 (29) 5 (71)

Numbers in parentheses are percentages

experienced some symptoms, particularly postprandial fullness and food vomiting.

When the 48 patients were considered together, symptoms of heartburn or reflux improved significantly after RY diversion. Although there was a reduction in the number of patients with vomiting, enquiry as to the type of vomiting revealed a significant improvement in bilious vomiting but food vomiting became more common. More patients complained of postprandial fullness after operation and many patients continued to complain of epigastric pain and nausea. Food vomiting, postprandial fullness, epigastric pain and nausea are all symptoms suggestive of gastric stasis and usually occur in combination. Thirty-five patients (73 per cent) from the series had one or more of these symptoms after RY diversion and twenty-three patients (48 per cent) complained of having to limit the size of their meals.

Visick grades (Table 3 ) Pre-operatively, 39 patients were classified Visick grade IV and 9 patients were classified Visick grade 111. After RY diversion, 17 patients (35 per cent) were classified as having a Visick grade IV result and 32 patients (67 per cent) were in Visick grades 111 or IV. Good results, however, were obtained by the 13 patients in group 3, the peptic ulcer group, only 1 of whom was in Visick grade IV; in contrast, 12 of 28 patients in group 1 (P< 0.05) and 4 of 7 patients in group 2 were in Visick grade IV.

The Visick grade 6 months after operation was compared with the Visick grade at the most recent visit to the clinic. Symptoms were found to have worsened in 12 patients and improved in 5 patients. Overall, when the final outcome was compared with the result recorded during the first year after RY diversion, no evidence of significant change in the patients’ functional results was found.

Results of RY diversion a s a primary and as a revisional procedure Only 1 of the 7 patients who underwent RY diversion as the primary operative procedure had a poor, Visick grade IV, result, whereas 16 of 41 patients who underwent RY as a revisional procedure had a poor result.

R Y diversion with and without vagotomy compared When the outcome according to the Visick classification was compared in these two groups, there was no evidence that addition of vagotomy improved, or worsened, the clinical result. None of the 12 patients without vagotomy developed recurrent ulceration; one of the vagotomized patients developed a recurrent, stoma1 ulcer.

Patients’ opinion Of the 48 patients, 24 (50percent) felt that they had been improved by RY diversion (Table 4 ) . Of the 13 patients in group 3, 12 considered that they had been improved by the operation, whereas only 12 of the remaining 35 patients felt that they had been improved.

Gastric emptying studies Solid f o o d . The median t+ Pnterquartile range) for solids

was 160 (75-370) min after RY dlversion, compared with 67 (5& 85) min in control (DU) patients (P< 0.01) (Figure 1). Of the 22 studies, 15 showed delayed gastric emptying of solids, and 14 of these 15 patients complained of food vomiting and/or epigastric fullness. Five patients who had normal empyting of solids according to the results of the test nevertheless had symptoms suggestive of stasis; three patients complained of epigastric fullness alone, but two patients described food vomiting.

Five of the seven patients with pre-operative stasis were investigated by gastric emptying studies both before and after RY diversion. The rate of solid emptying slowed in three cases and improved in two cases.

Liquids. The median ti and interquartile range for liquids was 8.7 (3.3-18)min after RY diversion, and the results of most tests were within the normal range of ti (5-10 min) for this laboratory. However, 8 of the 22 studies showed delayed gastric emptying of liquids and each of these 8 patients had clinical symptoms of stasis and were in Visick grades I11 or IV.

100

90

80

70 *.’

a 00 60

E - m X .-

50 .4- 0 0)

4- 8 40 C

E 30

20

10

0 60 64 68 72 76 80 84 88

Time after start of solid meal ( m h ) Figure 1 Mean curves for the gastric emptying of a solid meal in patients with uncomplicated duodenal ulcer disease and in patients ajitrr Roux-en-Y diversion. 0 , R Y patients; 0, uncomplicated DU patients

902 Br. J. Surg., Vol. 74, No. 10, October 1987

Gastric emptying: J. P. Britton et al.

who were studied converted to a fed state after a solid meal. Hence, pain, nausea and vomiting after RY diversion may be secondary to a defect in motor function, the Roux limb acting as an area of functional obstruction.

Many authors who have studied the effects of ‘remedial’ operations in patients with side effects of gastric surgery have found that good results cannot be guaranteed. That has certainly been our own and others’ experience with the RY procedure, though our results seem to be the worst ever reported after this operation. With the wisdom of hindsight, we think that several patients in this series were injudiciously selected for RY conversion. Some were polysymptomatic and somewhat neurotic, though whether the neurosis was secondary to the side effects of gastric surgery or preceded such surgery was unclear. Many patients evinced a variety of symptoms, and it was difficult to be certain that excessive reflux of ‘bile’ into the stomach was the principal defect. The use of an instillation test to infuse autologous enteric juices on to the gastric mucosa has been suggested as a possible way forward in the selection of patients for revisional surgery”. We did not attempt to quantify enterogastric reflux in any systematic fashion before the RY procedure, partly because the appropriate techniques were not available in the earlier years, and partly because of a certain scepticism about the clinical value of such measurements”. Certainly, many patients have copious reflux, but few symptoms.

If RY diversion yields such disappointing results, what can be done for patients such as those described in this paper? First, the problem should be kept to a minimum; most patients with duodenal or benign gastric ulceration can now be treated medically, without operation. If elective operation is indicated, use of highly selective vagotomy with preservation of the antral mill and pylorus will eliminate the problem of ‘bile’ re flu^'^*^^. For patients who suffer from dumping and enterogastric reflux after vagotomy with a drainage procedure, closure of the gastro- enterostomy or reversal of the pyloroplasty is less dangerous than RY diversion and will slow gastric emptying and diminish reflux, though some patients will develop gastric retention. If bilious vomiting is the predominant side effect after partial gastrectomy, RY diversion will help many patients, but the potential benefit of adding a vagotomy of the gastric remnant must be weighed against the considerable risk that gastric stasis may ensue. For patients with severe stasis after vagotomy- antrectomy with RY reconstruction, a higher gastric resection may have to be considered, and indeed we have two patients whose symptoms were improved by total gastrectomy and a Roux link-up. These are extreme measures, however, and dangerous to life. The main lesson is that reflux of ‘bile’ into the stomach should be prevented.

Discussion The main findings were that, overall, the clinical results of RY diversion were poor: two-thirds of the 48 patients remained in Visick grades I11 or IV and only half the patients felt that they had benefited. Symptoms of bilious vomiting and heartburn or gastro-oesophageal reflux improved significantly, but nausea and epigastric pain did not, vomiting of food became more common, and postprandial discomfort or distress, which had been experienced by 16 patients (33 per cent) before reoperation, was experienced by 32 patients (67 per cent) after RY diversion. Neither the patients’ symptoms nor their overall Visick grades improved with the passage of time after RY diversion.

Certain subgroups of patients, however, fared better than the average. Thus, five of the seven patients in whom RY diversion was performed as part of a primary operation for peptic ulcer obtained good results, and only one fared badly. Other authors have reported good results after vagotomy and antrectomy with RY reconstruction as a primary operation for duodenal ulcer’ o- ”. We have also used the RY method of reconstruction as a primary procedure in 13 patients after subtotal gastrectomy for gastric carcinoma, and none of these patients had a bad, Visick grade IV, result. In this study, there were 13 patients who underwent RY diversion for peptic ulceration, whether primary or recurrent (group 3), and these patients did fairly well, only one having a bad result.

It was interesting that the addition of vagotomy, which is universally recommended, appeared to make no difference to the outcome. Indeed, the only patient to develop a recurrent (stomal) ulcer was in the group of patients who underwent vagotomy. However, the numbers of patients were too small and follow-up too short to permit definite conclusions.

Results were poor in the 41 patients who underwent RY diversion as a revisional procedure. Despite a significant improvement in bilious vomiting and in heartburn or reflux, most of these patients developed symptoms suggestive of gastric stasis (Table 2 ) and no fewer than 30 of them (73 per cent) had a poor functional result (Visick grades I11 or IV). These observations are in agreement with previous reports of a high incidence of gastric stasis after RY di~ers ion’~-’~ .

The clinical evidence of gastric stasis after RY diversion was supported by the results of the gastric emptying studies, which showed, on average, normal emptying of liquids but slow gastric emptying of solids. Median ti for solids was 160min after RY diversion, compared with 67 min in patients with duodenal ulcer before operation. These studies were admittedly carried out in the 19 patients with the worst symptoms of stasis (and in 3 other patients with mild symptoms) but the patients tested formed a high proportion, almost 50percent, of the 41 patients who underwent RY diversion as a revisional procedure. Some of the other patients might also have been found to have slow gastric emptying, had they been tested. An inadequate gastrojejunal stoma was excluded by endoscopy.

We believe that gastric stasis of solids after RY diversion is attributable to three factors: the antrectomy, the vagotomy, and the presence of the RY loop ofjejunum. The gastric antral mill is responsible for grinding solid food into chyme, and its removal by antrectomy or partial gastrectomy may be expected to impair the breakdown of solid food into smaller particles. The addition of vagotomy will render the gastric remnant atonic, and thus less capable of expelling solid food into the jejunum. As Hocking et al. commented, ‘a vagally denervated small gastric pouch (after RY) serves as an atonic reservoir which impairs the emptying of ~olids’’~. This idea fits with our clinical observation at endoscopy that many of the patients still have food in the gastric remnant after an overnight fast. Although our results show that the addition of vagotomy makes no difference to outcome the numbers are too small to question this argument, particularly as we feel that vagotomy is only part of a multifactorial process responsible for stasis. Finally, studies in man’ have shown that in the fasted state the migrating motor complex is either absent or disrupted in the Roux limb, and none of the seven subjects

References 1.

2.

3.

4.

5.

6.

7.

8.

9.

Toye DKM, Williams JA. Postgastrectomy bile vomiting. Lancet 1965; ii: 524-6. Keighley MRB, Asquith P, Alexander-Williams J. Duodenogastric reflux: a cause of gastric mucosal hyperemia and symptoms after operations for peptic ulceration. Gut 1975; 16:

Bushkin FL, Wickbom G, Deford JW et al. Postoperative alkaline reflux gastritis. Surg Gynecol Obstet 1974; 138: 933-9. Ritchie WP Jr. Alkaline reflux gastritis: an objectiveassessment of its diagnosis and treatment. Ann Surg 1980; 192: 288-98. Roux C. De la gastro-enterostomie. Rev Gynecol Chir Abdom 1897; 1: 95. Lawson HH. The reversibility of postgastrectomy alkaline reflux gastritis by a Roux en Y loop. Br J Surg 1972; 59: 13-15. Donovan IA, Fielding JWL, Bradby H et al. Bile diversion after total gastrectomy. Br J Surg 1982; 69: 389-90. Kennedy T, Green R. Roux diversion for bile reflux following gastric surgery. Br J Surg 1978; 65: 323-5. Dixon MF, O’Connor HJ, Axon ATR, King RFGJ, Johnston D. Reflux gastritis: distinct histopathological entity? J Clin Pathol 1986; 39: 52430.

28-32.

Br. J. Surg., Vol. 74, No. 10, October 1987 903

Gastric emptying: J. P. Britton et al.

10.

11.

12.

13.

14.

15.

Cooper G, Bell G. Combined antrectomy and Roux-en-Y anastomosis in the surgical treatment of recurrent peptic ulceration. Br J Surg 1982; 69: 646-50. Herrington JL Jr, Scott HW Jr, Sawyers JL. Experience with vagotomy-antrectomy and Roux-en-Y gastrojejunostomy in surgical treatment of duodenal, gastric, and stoma1 ulcers. Ann Surg 1984; 199: 590-7. Karlqvist PA, Anderberg B, Olaison G, Sjodahl R. Early and late results after antrectomy, selective vagotomy and Roux-en-Y reconstruction for severe peptic ulcer disease. Acta Chir Scand

Boren CH, Way LW. Alkaline reflux gastritis. A reevaluation. Am J Surg 1980; 139: 4&3. Hocking MP, Vogel SB, Falasca CA, Woodward ER. Delayed gastric emptying of liquids and solids following Roux-en-Y biliary diversion. Ann Surg 1981; 194: 494-501. Pellegrini CA, Patti MG, Lewin M, Lawrence WW. Alkaline reflux gastritis and the effect of biliary diversion on gastric emptying of solid food. Am J Surg 1985; 150: 16671.

1986; 152: 357-61.

16.

17.

18.

19.

20.

Mathias JR, Fernandez A, Sninsky CA, Clench MH, Davis RH. Nausea, vomiting and abdominal pain after Roux-en-Y anastomosis: motility of the jejunal limb. Gastroenteroll985; 88: 101-7. Owen WJ, Hollands MJ. The use of an instillation test to define alkaline reflux gastritis. Scand J Gastroenrerol 1984; 19 (suppl.

Earlam R. Bile reflux and the Roux en Y anastomosis. Br J Surg 1983; 70: 393-7. Keighley MRB, Asquith P, Edwards JAC eta/. The importance of an innervated and intact antrum and pylorus in preventing postoperative duodenogastric reflux and gastritis. Br J Surg 1975; 62: 845-9. Dewar P, King R, Johnston D. Bile acid and lysolecithin concentrations in the stomach in patients with duodenal ulcer before operation and after treatment by highly selective vagotomy, partial gastrectomy, or truncal vagotomy and drainage. Gut 1982; 23: 569-77.

92): 257-8.

904 Br. J. Surg., Vol. 74, No. 10, October 1987