Primary Roux-Y gastrojejunostomy versus gastroduodenostomy after antrectomy and selective vagotomy

Primary Roux-Y Gastrojejunostomy versus Gastroduodenostomy after Antrectomy and Selective Vagotomy Ulf H. Haglund, MD, PhD, Uppsala,Sweden, Roland L. Jansson, MD, PhD, Bor~s,Sweden, Jacob G.E. Lindhagen, MD, PhD, Norrk6ping,Sweden, Ears R. Eundell, MD, PhD, Gothenburg,Sweden, Erik G. Svartholm, MD, PhD, Maim6,Sweden, Lars C. Olbe, MD, PhO, Gothenburg,Sweden

One hundred twenty-one patients with prepyloric ulcer disease entered a randomized clinical trial comparing gastroduodenostomy with Roux-Y gastrojejunostomy after antrectomy and selective gastric vagotomy. The postoperative course and morbidity were quite similar in the two study groups, as was the postoperative infectious complication rate. Forty-four of the patients with a Billroth I reconstruction and 52 of those with a Roux-Y reconstruction were followed up with a clinical assessment at least 6 months after the operation. The postgastrectomy symptoms were quite frequent, but did not differ between the two study groups. Seventy-five percent of the patients with a Billroth I gastroduodenostomy had symptoms corresponding to Visick grades 1 and 2, compared with 81% of those with Roux-Y reconstruction. Although the latter procedure was very effective in preventing bile reflux to the gastric remnant, no difference was observed in the gastric emptying rate after the two operations.

A

ntrectomy combined with selective gastric vagotomy has been repeatedly shown to be efficacious in preventing recurrent peptic ulcer, with a recurrence rate of only 0% to 1.5%. On the other hand, postgastrectomy complaints of various severity are experienced in 10% to 20% of patients [I]. Because of these postgastrectomy complaints, antrectomy with vagotomy has been abandoned by many surgeons in duodenal ulcer surgery and replaced with proximal gastric vagotomy. However, a high recurrence rate has been reported after proximal gastric vagotomy for prepyloric ulcers [2,3], which is why antrectomy and selective vagotomy has been reserved for patients with this disease. From the Departments of Surgery, General Hospital, Malta6, Sweden; Sahlgren's Hospital, Gothenburg, Sweden; and Central Hospital, Borers, Sweden, and Norrk6ping, Sweden. Requests for reprints should be addressed to Ulf Haglund, MD, PhD, Department of Surgery, Akademiska sjukhuset, S-750 14 Uppsala, Sweden. Manuscript submitted March 31, 1989, revised September 8, 1989, and accepted September 27, 1989. 546

THE AMERICAN JOURNAL OF SURGERY

Five percent to 10% of the complaints following antrectomy with or without vagotomy may be due to symptoms of duodenogastric reflux (nausea, bilious vomiting, and epigastric pain) [4,5]. Patients with symptoms of duodenogastric reflux after partial gastrectomy may have a significantly greater reflux of duodenal contents into the gastric remnant than asymptomatic patients [6,7]. Diversion of duodenal contents from the gastric remnant by revisional surgery with a Roux-Y reconstruction in these patients results in clinical improvement [6-I0], since bilious vomiting is effectively eliminated by this reconstruction, but the more unspecific symptoms such as nausea, epigastric pain, and fullness may remain or return in about 25% of the patients [9-12]. Another 10% of these patients may also develop symptoms of gastric retention [12-14]. It should also be mentioned that the Roux-Y reconstruction must be combined with vagotomy to prevent stomal ulceration [8]. In order to study whether postgastrectomy complaints could be reduced and long-term development of gastritis/ epithelial cellular atypia could be prevented using a primary Roux-Y gastrojejunostomy, patients with prepyloric ulcer underwent antrectomy and selective vagotomy and the reconstruction was randomized to a gastroduodenostomy or a Roux-Y gastrojejunostomy. This paper reports the early postoperative results in this series of patients. PATIENTS AND METHODS Four hospitals participated in the study; 121 patients entered the clinical trial during the time period 1982 to 1986. All patients had an antrectomy and a selective vagotomy for prepyloric ulcer disease. A prepyloric ulcer was def'med as an ulcer located within 3 to 4 cm orally to the pylorus. The indication for surgery varied according to the clinical routines at the individual hospitals. Long duration of the disease and/or complications such as hemorrhage or stenosis were the most common indications. Malignancy was ruled out at preoperative endoscopy with biopsy. The resection was limited to the antrum with a few centimeters of the distal part of the body as a margin. All specimens were routinely investigated to determine the existence of oxyntic gland mucosa in the proximal margin. Randomization to gastroduodenostomy or Roux-Y gastrojejunostomy was performed during surgery when the surgeon was convinced that both reconstruction procedures were technically suitable. Randomization was carried out according to the 'sealed-envelope' principle, and block-randomization was applied. All gastrointestinal anastomoses were suturdd with either

VOLUME 159 JUNE 1990

BILLROTH I VERSUS ROUX-Y

Dexon or Vicryl sutures, and the Roux limb was placed retrocolically with a length of 50 cm. The patients were then followed according to a detailed protocol describing the postoperative course as well as the clinical status. After discharge from the hospital, they were regularly seen (3 and 6 months postoperatively and once yearly) at the outpatient clinic. The patients were asked about postgastrectomy complaints (dumping, nausea, vomiting, epigastric fullness, diarrhea). Dumping was graded into three grades: mild (occasional and/or slight dizziness after meals), moderate (occasional need to lay down after a meal, although no incapacitating complaints), and severe (need to lay down after each meal, interfering with a normal social life). Diarrhea and vomiting were defined subjectively, i.e., if the patient considered either of these to be a significant complaint, it was so noted. The relatively low incidence of these complaints prohibited further analysis regarding frequency. A Visick grading was carried out. Gastroscopy was performed if ulcer symptoms recurred, otherwise every second year. Thirty-nine patients underwent postoperative scintigraphy with dimethyl iminodiacetic acid (HIDA) to quantitate duodenogastric bile reflux. These patients also underwent a gastric emptying test during the first 6 months postoperatively. Semisolid food (toast and technetium99m-labeled egg) was used [15]. A half-life time less than 45 minutes was used as the upper limit of the normal range. Statistical evaluation of the clinical data was performed by application of the chi-square method. The study protocol was approved by the local ethical committees, and informed consent was obtained from each patient. RESULTS Fifty-seven patients underwent operation with a gastroduodenostomy (Billroth I), and 64 patients were randomized to a Roux-Y gastrojejunostomy. There was no difference between these two groups of patients regarding relevant demographic data (Table I). The duration of ulcer disease was similar in the two groups, with a history of more than 5 years of complaints in the majority of patients. Few patients had a prepyloric recurrence after a previous proximal gastric vagotomy (4% and 6%, respectively). A history of ulcer complications (hemorrhage, stenosis, perforation) was experienced by as many as 37% and 38%, respectively, of the patients in the two study groups. The duration of the surgical procedure (median: 3.5 versus 4.5 hours; range: 2 to 7 hours versus 2 to 9 hours) and the intraoperative blood loss (median: 300 mL for both groups; range: 100 to 2,100 mL versus 100 to 1,800 mL), as well as the frequency of splenectomies (9% and 11%, respectively), did not differ between the Billroth I and Roux-Y groups. Most of the splenectomies were performed during the first 2 of the 5 years of this study. Prophylactic antibiotics were given to the majority of patients. The postoperative course (Table II) was very similar

TABLE I Demographic Data on Patients Operated on with BIIIroth I or Roux-Y Reconstruction

Age (mean 4- SD), years Sex (F:M) Brocas index > 1.1 Smokers Complicating medical disorders Duodenal & prepyloric ulcer

Billroth I (n -- 57)

Roux-Y (n -- 64)

52 4- 10.8 36:21 12 34 17 12

52 4- 10.2 41:23 9 45 23 19

10 39 35

20 35 40

10 8

9 9

disease Duration of ulcer disease 2-6 years >5 years No ulcer complication Complications Hemorrhage Obstruction

TABLE II

Postoperative Course and Morbidity

Prolonged nasogastric drainage Parenteral nutrition >--7 days Abdominal infections (total) Intra-abdominal abscess Extra-abdominal infections Postoperative hemorrhage Reoperatlon Other clinically important

Billroth I (n = 57)

Roux-Y (n = 64)

6 7 7 5 7 0 4 2

7 7 8 1 6 2 6 6

complications

in the two groups of patients, with an equal median postoperative hospital stay (9 days). There was one postoperative death, occurring in a patient with a Billroth I gastroduodenostomy who died of cardiac insufficiency and pulmonary edema. However, an anastomotic leakage with a small intra-abdominal abscess was seen at autopsy. There were two anastomotic leakages (one in each group) recorded, but the other patient had a reoperation with suturing of the leakage and then had an uneventful postoperative course. The postoperative morbidity is outlined in Table II, showing a complication rate with no difference between the two types of gastrointestinal reconstructions. We fobnd a slightly higher intra-abdominal infection rate in patients with a Billroth I anastomosis, but the total abdominal infection rate did not differ between the two groups. The postoperative follow-up period was 6 months or more in 44 of the patients with a Billroth I reconstruction and in 52 of those with a Roux-Y reconstruction. The median follow-up was 26 months in the former group and 31 months in the latter. As seen in Table III, the postgastrectomy symptoms were quite frequent, but most importantly, the symptoms did not differ between the two types of reconstruction after antrectomy and selective vagotomy. One patient in both series had recurrent ulceration. Forty-three percent of the patients operated on with a

T H E A M E R I C A N J O U R N A L OF S U R G E R Y

V O L U M E 159

J U N E 1990

547

HAGLUND

ET AL

TABLE III Clinical Outcome and Postgastrectomy Complaints

Visick grade 1 2 3 4 Diarrhea Nausea & vomiting Dumping

Billroth I (n = 44)

Roux-Y (n = 52)

2

(81%) 17 9 1" 7 (14%) 4 (8%) 16 (31%)

(75%)

10 1" 7 (16%) 8 (18%) 8 (18%)

* Recurrence,

gastroduodenostomy complained of some kind of postgastrectomy symptom, compared with 56% in the RouxY group. It should, however, be pointed out that the great majority of symptoms were mild and in no case severe or incapacitating (Table III). Applying a modified Visick scale (1 = no complaints, 2 = occasional complaints easily dealt with, 3 = frequent complaints demanding medication, 4 = severe complaints affecting employment and/or social life), we found that 75% of the patients operated on with a BiUroth I gastroduodenostomy had symptoms corresponding to Visick grades 1 and 2, with a similar figure found among those having a Roux-Y reconstruction (81%). No significant difference was encountered in overall clinical judgment after the operations. Bile reflux as studied by HIDA-scintigraphy revealed that 14 of 21 patients who had a Billroth I reconstruction had abnormal bile reflux to the gastric remnant, compared with only 1 of 18 patients studied after a Roux-Y reconstruction (p <0.001). Postoperative complaints were not more common among patients with bile reflux compared with those without reflux. Delayed gastric emptying of semisolid food was found in 39% of the patients with Billroth I reconstruction compared with 33% of the Roux-Y patients. COMMENTS The present clinical trial is the first study that approaches on a randomized basis the problem of primary Roux-Y gastrointestinal reconstruction after antrectomy and selective gastric vagotomy for peptic ulcer disease, with respect to the immediate as well as the long-term clinical course. The immediate postoperative course was found to be very similar in both groups, although the postoperative morbidity may look less impressive, with two anastomotic leakages and a rather high frequency of postoperative abdominal infections. It should, however, be pointed out that the operations were not performed by surgeons with a special interest and experience in gastric surgery, although they were supervised by such surgeons. Instead, many surgeons in training were involved in the treatment of these patients. We therefore believe that the present results are representative for routine surgical praxis. Both operative procedures were effective in prevent548

ing ulcer recurrences. On the other hand, postgastrectomy symptoms were often experienced, although with similar frequency in the two groups under study. However, the postgastrectomy complaints were usually mild, and no patient experienced incapacitating problems. The overall clinical assessment as judged by the Visick grading suggested successful results in the great majority of patients instead. During the past few years, an association between delayed gastric emptying and Roux-Y diversion has been reported [11,14]. Severe gastric retention has been observed in patients who underwent a Roux-Y biliary diversion for alkaline reflux gastritis. It is interesting to note that in the immediate postoperative course, we found vomiting to be slightly more frequent in patients with Billroth I reconstruction than in those randomized to Roux-Y gastrojejunostomy. In addition, we were unable to demonstrate a need for a longer postoperative period with nasogastric drainage and larger drainage volumes in patients with Roux-Y reconstruction. A number of patients also underwent a gastric emptying study during the first 6 postoperative months that did not reveal a difference between patients with a gastroduodenostomy compared with those having a Roux-Y gastrojejunostomy, confirming previous observations obtained in patients 1 year or more after the actual operations [16]. We found prolonged gastric retention in a significant portion of the patients, although with similar frequency in both groups. This particular problem tended to subside after a few weeks in almost all patients. Roux-Y reconstruction is no doubt very effective in preventing bile reflux to the gastric remnant [7,8,10]. The clinical and pathophysiologic importance of bile reflux will, however, be highlighted by the long-term results of the present study, including observations on the development of chronic gastritis with the possible appearance of epithelial atypia. So far, however, our results would suggest that primary Roux-Y reconstruction, although effectively eliminating bile reflux, does not result in fewer early postgastrectomy complaints than Billroth I reconstruction. REFERENCES 1. SawyersJL, Herrington JL Jr. Vagotomyand antr~tomy. In: Nyhus LM, WastellC, eds.Surgeryof the stomachand duodenum. 4th edition. Boston: Little, Brown, 1986: 395-417. 2. Andersen D, Amdrup E, HSstrup H, Hanberg-S6rensenF. The Aarhus countyvagotomytrial: trends in the problem of recurrent ulcer after parietal cell vagotomyand selective gastric vagotomy with drainage. World J Surg 1982; 6: 86-92. 3. HollinsheadJW, Smith RC, Gillett DJ. Parietal cell vagotomy: experience with 114 patients with prepylorir or duodenal ulcer. World J Surg 1982; 6: 596-602. 4. Rehnberg O, Olbe L. Early complicationsand late postgastrectomy syndromesin peptic ulcer patients treated by antrectomyand gastroduodenostomywith or withoutvagotomy.Acta Chir Scand 1983; 515(suppl): 28-36. 5. GoligherJC, Feather DB, Hall R, et al. Severalstandardelective operations for duodenal ulcer. Ann Surg 1979; 189: 18-24. 6. Hoare AM, McLeishA, ThompsonH, Alexander-WilliamsJ. Selection of patients for bile diversion surgery: use of bile acid measurement in fasting gastric aspirants. Gut 1978; 19: 163-5.

THE AMERICAN JOURNAL OF SURGERY VOLUME159 JUNE 1990

EDITORIAL COMMENT

7. Sch66n I-M, Andersson H, Faxen A, Olbe L. Gastric bile acids before and after Rous-en-Y transposition for bile reflux gastritis and in asymptomatic controls. Scand J Gastroenterol 1979; 14: 969-76. 8. Herrington JL Jr, Sawyers JL, Whitebead WA. Surgical management of reflux gastritis. Ann Surg 1974; 180: 526-37. 9. Malagelada J-R, Phillips SF, Shorter RG, et al. Postoperative reflux gastritis: pathophysiology and long-term outcome after Roux-en-Y diversion. Ann Intern Med 1985; 103: 178-83. 10. Ritchie WP. Alkaline reflux gastritis. Ann Surg 1986; 203: 537-44. 11. Boren CH, Way LH. Alkaline reflux gastritis: a re-evaluation. Am J Surg 1980; 140: 40-6. 12. Mathias JR, Fernandez A, Sninsky CA, Clench MH, Davis RH. Nausea, vomiting and abdominal pain after Roux-en-Y anas-

tomosis: motility of the jejunal limb. Gastroenterology 1985; 88: 101-7. 13. Herrington JL Jr, Scott HW Jr, Sawyers JL. Experience with vagotomy-antrectomy and Roux-en-Y gastrojejunostomy in surgical treatment of duodenal, gastric and stomal ulcers. Ann Surg 1984; 199: 590-7. 14. Hocking MP, Vogel SB, Falasca CA, Woodward ER. Delayed gastric emptying of liquids and solids following Roux-Y biliary diversion. Ann Surg 1981; 194: 494-501. 15. Naeslund L Gastric by-pass vs gastroplasty. A prospective study of differences in two surgical procedures for morbid obesity. Acta Chir Scand [Suppl] 1987; 1-60. 16. Lundell L, Alpsten M, Andersson D. Effect of Roux-en-Y gastroenterostomy on gastric emptying of the solid meal following antrectomy and vagotomy. Ann Chir Gynaecol 1986; 75: 201-4.

EDITORIAL COMMENT J. Lynwood Herrington, Jr., MD, Nashville,Tennessee

The preceding paper by Haglund and co-workers is an interesting randomized study comparing primary Roux-Y gastrointestinal reconstruction with end-to-end gastroduodenostomy (Billroth I), with selective gastric vagotomy performed in each group. The various parameters used in assessing results appear comparable in both groups. In my experience and in the reports of others, a RouxY gastrojejunostomy may result in normal, rapid, or delayed emptying. Delayed transit, however, is not uncommon; in contrast to the authors' experience, we have demonstrated a need for a longer period of postoperative nasogastric decompression. Postoperative bile reflux after a BiUroth I reconstruction in my experience has not usually led to the severe gastric remnant inflammatory changes that

From the Department of Surgery, Vanderbilt University, Nashville, Tennessee.

at times follow a Billroth II reconstruction. Also, I have been impressed with the fact that delay in gastric emptying following a Billroth I reconstruction is much less frequently encountered than with a Roux-Y gastrojejunostomy. Although the authors report approximately equal results with either reconstruction, they do not believe that the Roux-Y method should supplant end-to-end gastroduodenostomy. I wholeheartedly agree with this statement. The Roux method is associated with an increased morbidity, as three intestinal suture or staple lines are required, and two separate anastomotic sites exist for the possibility of postoperative obstruction. However, when the disease in the pyloroduodenal canal is extensive, and when a perforation of a giant duodenal ulcer is encountered, the Roux-Y reconstruction is preferred. Following duodenal stump closure, decompression of the afferent jejunal limb through the Roux-Y limb reduces intraluminal afferent limb pressure,

with diminished tension being applied to the closed duodenal stump. One must always remember that when the Roux principle is used in gastric surgery, the gastric antrum must be adequately removed and the vagotomy must be complete in order to protect against recurrent ulceration. I question the authors' use of selective gastric vagotomy over truncal vagotomy, as few centers currently use the former technique. Also, in the hands of the occasional gastric surgeon, the selective technique may result in an incomplete vagotomy. This randomized clinical study is of great interest, but the Billroth I reconstruction still holds a prominent place as a primary method of reconstruction after partial extirpation of the stomach for gastroduodenal ulcer. The Roux-Y method is a splendid alternative when faced with disease in the pyloroduodenal canal that would render a Billroth I anastomosis unsafe. Currently, I have all but abandoned the Billroth II method of reconstruction.

THE AMERICAN JOURNAL OF SURGERY VOLUME 159 JUNE 1990 549