- Email: [email protected]
Results of the surgical treatment of obesity
Results of the Surgical Treatment of Obesity Lloyd D. MacLean, MD, FACS, Barbara M. Rhode, PDt, MSc (Nutr), CNSD, John Sampalis, PhD, Montreal,Canada, R. Armour Forse, MDCM,FACS,Boston,Massachusetts
A prospective, randomiT~d trial comparing vertical banded gastroplasty (VBG) and vertical gastric bypass (GB) for obesity was completed in 106 patients who did not differ in baseline body mass index (BMI = kg/m 2) or length of follow-up. The goal of this surgery was to return patients to within 50% of their ideal weight, i.e., a body mass index less than 35 kg/m 2, and to accomplish this while maintaining a low risk for malnutrition as well as other morbidity and mortality. Success was defined as a BMI less than 35 kg/m 2 because the mortality risk increases rapidly above this degree of obesity. Surgical failures were encountered in 43% of the 54 patients in the VBG group, all of whom had division between the vertical staple lines. The main causes of failure were stenosis and enlargement of the gastroplasty orifice. Surgery failed in 23% of the GB-treated patients, due to perforation of the vertical staple line. An isolated gastric bypass (IGB) not dependent on staples was performed as the remedial operation for the failures of both VBG and GB. IGB was significantly better than VBG or GB, with a success rate of 83% compared with 39% for VBG and 58% for GB. Subsequent experience since completion of this randomi,Jed trial in 54 consecutive patients supports IGB for primary, as well as remedial, operations for the morbidly obese (BMI = 40 to 50 kg/m2), as well as for patients who are super obese (BMI greater than 50 kg/m2).
Fromthe Departmentof Surgery(LDM,BMR,JS), McGillUniversity, Montreal,Canada,and the DepartmentofSurgery(RAF),Harvard MedicalSchool,Boston,Massachusetts. Requests for reprints shouldbe addressedto LloydD. MacLean, MD, RoyalVictoriaHospital,Room$10.26, 687 Pine AvenueWest, Montreal,PQ, CanadaH3A IA1. Presentedat the 33rdAnnualMeetingof the Societyfor Surgeryof the AlimentaryTract, San Francisco,California,May 11-13, 1992.
ertical banded gastroplasty (VBG) and gastric V bypass (GB) are the two operations most frequently performed for severe obesity. Our experience with VBG has been unsatisfactory mainly due to breakdown of the vertical staple line. This occurred in 48% of 201 patients recently reported [1]. Many of these patients required reoperation including GB, which resulted in a higher success rate [2]. We also observed over long periods of follow-up (over 4 years) that if the vertical staple line in VBG remained intact and if the orifice size remained less than 11 mm in diameter, the results in 47 patients who were morbidly obese preoperatively were highly satisfactory (0% to 25% excess weight) in 57% and good (26% to 50% excess weight) in an additional 32%. Therefore, a successful result, defined by Reinhold [3] as good or excellent (less than 50% excess weight or a body mass index less than 35 kg/m2), was achieved in 89% of these highly selected patients. We reasoned, therefore, that an ideal operation would require surgical isolation of the pouch with VBG. The purpose of the current study was to compare a VBG not dependent upon staples, i.e., made with a division between the vertical staple lines, with Roux-en-Y GB that had a vertical lesser curvature pouch the same size as that constructed for the VBG (less than 15 mL). The GB was constructed with a large orifice, and staple lines were not divided. PATIENTS AND METHODS A total of 54 patients were randomized to undergo VBG, and 52 patients were randomized to undergo GB. Randomization took place immediately after the surgery had commenced. The study period was from October 1987 to February 1990. All operations were performed through an upper abdominal midline incision using a table-mounted retractor (Gomez retractor, Pilling Canada, Markham, Ontario). For the gastroplasty, the vertical staple line was made with a V. Mueller PI-90 stapler (3M Company, St. Paul, MN) using 4.8-mm staples. This stapler introduces two double rows of staples with an interval of free tissue in between that permits division between the staple lines (Figure 1). The cut edges were sutured to establish hemostasis and prevent leakage. Omentum was sutured between the staple lines. The proximate ILS 25-mm instrument (Ethicon Inc., Somerville, N J) was fired adjacent to a 28 Maloney bougie 4 cm distal to the angle of His. A drain was placed from the left side of the esophagogastric junction to the lesser omental sac distal to the left gastric vessels to identify the angle of His accurately. The volume of this pouch was infrequently measured but was always less than 15 mL and often less than 10 mL. The banding of the VBG was done with polypropylene mesh cut 60 mm in length and 15 mm in width, overlapped and
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993 155
MACLEAN ET AL
Figure 1. Vertical banded gas~'opiasty wil~h division between vertical staple lines. A Maloney bougie no. 28 is in place, and the polypropylene mesh band is 45 to 47 mm in external circumference.
Figure 3. Isolated gastric bypass with separation of the small
gastric pouch from the remaining stomach.
156
Figure 2. Gastric bypass made with TA-90B stapler. The vertical line was oversewn as illustrated. The Roux limb was 40 cm in length and was placed in the retrocolic, retrogastric position.
sutured in place, to achieve a 45- to 47-mm external circumference. This was done with a double piece of polypropylene. The patients who underwent GB had a small pouch of the same size constructed in the same location, but the vertical staple line was constructed with the TA-90B stapler (Auto Suture, U. S. Surgical Corporation, Norwalk, CT), which places four rows of closely approximated staples. The Roux limb of jejunum was made 40 cm in length and was brought to the small gastric pouch in the retrocolic, retrogastric position. The end of the jejunum was dosed, and the anastomosis was made using a single layer of either Prolene 3-0 or P.D.S. 3-0 suture in a running fashion, i.e., end of the gastric pouch to the side of the jejunum (Figure 2). This anastomosis was made approximately 1.0 to 1.5 cm in diameter. In patients in whom either operation failed, a third procedure was performed (Figure 3), in which an isolated gastric bypass (IGB) was performed. In these cases, the Mueller stapler with division between the staple lines was used, thereby separating the small gastric pouch from the remaining stomach. All patients had a closed system drain placed in the left upper quadrant for 2 to 3 days and a gastrostomy to the bypassed stomach that was retained for 6 weeks, at which time the patients underwent initial gastroscopy. Table I summarizes the characteristics of the two groups of patients. Thirty-seven (69%) of the patients who underwent VBG were morbidly obese (BMI = 40 to 50
THE AMERICANJOURNALOF SURGERY VOLUME165 JANUARY1993
SURGICAL TREATMENT OF OBESITY
kg/m2), and 17 (31%) were super obese (BMI greater than 50 kg/m2). Twenty-nine (56%) of the patients who underwent GB were morbidly obese, and 23 (44%) were super obese. The preoperative incidence of super obesity was not statistically different in the two groups (Fisber's exact test; p = 0.12).
Weight loss monitoring: We saw patients once a month for the first 3 months, every 3 months for the first year, and semiannually thereafter. At each visit, patients were wdghed and examined, and their body mass index was calculated. Final foUow-up data were obtained in March 1992. Gastroscopy was performed with a 9.8- or 9-mm instrument at 6 weeks, 6 months, and annually thereafter. Additional gastroscopic examinations were performed when required by a change in clinical status. We made estimates of the orifice size and of the status of the staple line at each endoscopic operation. Urgent endoscopy was performed in patients who developed acute episodes of vomiting after operation. This occurred only in the patients who underwent VBG. If the blockage was caused by food or pills, these were removed by endoscopic techniques. If stenosis of the orifice caused repeated vomiting (VBG only), we ruled out obstruction due to a foreign body and placed the patient on a clear liquid diet. We did not attempt dilatation because previous experience has shown that this technique is unsuccessful. If the vomiting persisted, patients underwent another operation using an IGB. If a hole developed in the staple line, which mainly occurred in the GB group, and the patient was gaining weight or developed a stomal ulcer, IGB was performed. Classification of weight loss: The Reinhold [3] classification was used to describe the postoperative results. Ideal weight was determined by the use of the Metropolitan Life Insurance tables [4]. Results, which were determined at the times of evaluation, were expressed as excellent when the patient had 0% to 25% excess weight; a good result was an excess of 26% to 50%; a fair result, 51% to 75% excess; and a poor result, 76% to 100~ excess. Failure was defined as more than 100% excess weight. For the purposes of this study, we considered surgical treatment successful if the patient's results were categorized as excellent or good and a re.operation was not required to achieve these results. Surgical treatment was defined as a failure if a reoperation was required, regardless of the ultimate outcome. Statistical methods: Differences between groups with respect to continuous variables were evaluated using Student's t-test (two groups) or analysis of variance (ANOVA) (more than two groups). For categoric variables, the x 2 test was used. In addition, odds ratios were calculated for binary outcomes. Ninety-five percent confidence intervals (CIs) were used to assess the precision and statistical significance of the odds ratios. Multiple linear regression and ANOVA were used to test the difference between surgical procedures with re-
TABLE
I
Baseline Characteristics of Study Patients* Vertical Banded Gastroplasty
Gastric Bypass
p Value
52 38.8 • 9.5 278 • 41 48 _-. 7
52 40.1 _+ 7.7 295 -+ 53 50 -+ 7
-0.43 0.07 0.20
No. of patients Age (y) Weight (Ib) Body mass index (kg/m 2) *Valuesexpressedas mean • SD.
TABLE lI
Reasons for Reoperation Vertical Banded Gastroplasty
Gastric Bypass
No. of patients Reoperation
54 23 (43%)
52 12 (23%)
Indications Stenosis Enlarged orifice Staple line fistula Normal but clinical failure Abscess Stomal ulcer
11 7 2 2 1 --
--12 --7
TABLE IlI
Outcome at Final Assessment of Operations
No. of procedures No. of patients Revised to normal No. lost to follow-up Follow-up (mo)* Patient outcome (%)t Excellent Good Fair Poor Failure
Vertical Banded Gastroplasty
Gastric Bypass
Isolated Gastric Bypass
54 31 5 0 38.6 _+ 8.5
52 40 0 0 33.1 -+ 12.4
30 30 0 1 21.8 - 9.3
7 14 6 2 25
(13) (26) (11) (4) (46) ~
Overall Success (%) Failure (%)
12 18 5 3 14
(23) (35) (10) (6) (27)
15 (50) 9 (30) 2 (7) 2 (7) 1 (3)
p--0.27 I p=0.034-~ p = 0.00015
21 (39) 33 (61)
30 (58) 22 (42)
I ~
p = 0.0003
24 (83) 5 (17)
I
*Valuesexpressedas mean -+ SD. lp Valuesare basedon • analysis,
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
157
MACLEAN ET AL
TABLE IV
Multivariate Analysis on Change in Body Mass Index (BMI) F Ratio
p ValUe
Variable Age Follow-up duration Baseline BMI
0.22 1.77 36.95
0.64 0.19 0.0001
Surgery Isolated bypass versus gastroplastry Gastric bypass versus gastroplasty Isolated bypass versus gastric bypass
5.17 11.25 0.93 6.04
0.007 0.001 0.34 0.016
*Values were calculated by analysis of variance.
TABLE V
Patient Characteristics No. of patients No. lost to follow-up Follow-up (me _+ 1 SD) Body mass index (kg/m 2 -+ 1 SD)
54 1 15.9 --- 3.8 48 _+ 7
TABLE Vl
Results of Isolated Gastric Bypass Since Completion of Randomized Trial Parameter New operation Revision operation Morbid obese Super obese
No.
Success*
36 17 34 19
32 14 32 14
(89%) (82%) (94%) (74%)
*Success = < 35 kg/m 2 = Good + Excellent = < 50% excessweight.
spect to the change in BMI from baseline while controlling for the patient's age, the duration of follow-up, and the baseline BMI. For this analysis, planned contrasts were used to evaluate all possible pairwise comparisons of the three surgical procedures. RESULTS There were no deaths. One patient has been lost to follow-up. Five patients who were converted to normal from VBG are classified as failures. These patients did not undergo IGB. Forty-three percent of 54 patients who underwent VBG required re.operation for the reasons outlined in Table IL Eighteen of the 23 patients were converted to IGB. Twelve or 23% of patients who underwent GB developed a hole in the staple line and were converted to an IGB not dependent on staples. Therefore, a total of 54 VBG operations and 82 GB or IGB operations were performed on the 106 patients. The success rate in the combined bypass group was 67% compared with 39% for the VBG group (p --- 0.003, odds ratio (OR) = 3.1; 95% CI -- 1.45 to 6.87). The final outcomes are shown in Table 158
THE AMERICAN JOURNAL OF SURGERY
HI, which is an analysis by surgical procedure. The outcome is classified according to the patients' weight at the time of final evaluation as described above. The results in Table III show that significantly better results were achieved with the remedial operation IGB than with either VBG or GB. Table III also lists the success rate for each procedure, which was clef'reed in this study as an excellent or good outcome, no requirement for re.operation, a BMI less than 35 kg/m 2, or less than 50% excess weight. These data show that the rate of successful outcome was significantly higher with IGB compared with the other two procedures. Erosion of the Marlex mesh was associated with both stenosis and enlargement of the orifice after gastroplasty in 13 of 23 patients who required reoperation (Table II). Staple line dehiscence was followed by stomal ulcer in 7 of 12 patients who required operation after GB. These ulcers all healed after reoperation and conversion to IGB. In one patient, the single failure, reoperation was necessary because severe stenosis at the previous ulcer site did not respond to dilatation. The surgery was reversed in this patient. Multivariate analysis comparing the three procedures while controlling for the age of the patient, the baseline BMI, and the duration of follow-up showed that IGB resulted in a significantly higher reduction in the BMI, This analysis emphasized the importance of the baseline BMI as a potential confounder (Table IV). Since completing this study, 54 consecutive patients have undergone IGB (Tables V and VI) and have been followed for a minimum of 1 year (mean: 15.9 4- 3.8 months). These results, albeit with short follow-up times, are better than any previous experience at our clinic and support the use of this procedure both for morbidly as well as for super obese patients and for initial as well as remedial surgery. Serious morbidity due to subphrenic abscess or leak from the anastomosis or staple line occurred in 6 patients among the total of 106 patients in the prospective randomized trial plus the 54 patients who have undergone surgery since completion of the trial (3.8% of patients). These 160 patients underwent a total of 195 operations (3.1% of operations). Leak or intraperitoneal abscess formation occurred after the first operation in three patients (two VBGs and one GB). In the remaining three patients, leak or abscess formation followed the remedial operation, which was IGB. In five of the six patients, laparotomy was necessary, and, in one patient, aspiration alone was successful. COMMENTS This randomized trial strongly supports GB over VBG, especially when staple line disruption is eliminated as a complication in the former operation. This preference of GB over VBG has previously been supported by the excellent results obtained in prospective trials by Sugerman et al [5] and by Hall et al [6]. The operation we prefer, which has evolved from the data herein reported and our previous experience, embod-
VOLUME165
JANUARY1993
SURGICAL TREATMENT OF OBESITY
ies the following characteristics: a small pouch less than 15 mL in volume; a pouch completely separated from the remaining stomach and not dependent upon staples; a pouch that empties dependently so that there is no stasis that can lead to regurgitation or vomiting and enlargement of the pouch due to distention; and an orifice from the pouch not dependent on a foreign body to maintain its size or diameter and of a size that permits the ingestion of solid food. In our slow pursuit of the best operation to treat obesity, we have adopted a procedure that closely approximates that preferred by Linner and Drew [7], with the exception that we do not use external support of the gastrojejunostomy. Linner and Drew [7] reported that 53% of patients were within 25% of their ideal weight and another 30% had 26% to 50% excess weight (an 83% success rate) after 5 or more years. The incidence of super obesity was not given by these authors. We believe that a vertically placed pouch is less likely to dilate because of prompt emptying. The pouch should be separated from the main stomach and not be dependent on staples in order to prevent fistula formation. A hand-sutured anastomosis is done because it is easier than using stapling instruments on a very small pouch in the confined space available. In our experience, attempts to limit the size of the anastomosis by using a running, continuous, nonabsorbable suture fail because the suture material migrates into the lumen, and the anastomosis over subsequent months becomes the same size as the jejunum beyond the anastomosis. Patients who have successful results after VBG, as we have performed the operation, usually ingest a pureed or liquid diet. After GB or IGB, patients can eat anything but in limited quantities. Diarrhea has occurred in a minority of GB patients and only during the first month postoperatively. A feeling of fullness, dizziness, and a desire to lie down after eating (dumping syndrome) is frequently encountered, but since these symptoms are associated with decreased intake and weight loss, most patients do not complain about them. No patient has required specific medication for these symptoms. The results of gastric restrictive surgery have been poorer for patients who were super obese preoperatively than for patients who were morbidly obese preoperatively [8]. Brolin et al [9] prospectively studied a modification of the Roux-en-Y gastric bypass for super obese patients. In their conventional bypass, the Roux limb was 75 cm in length, and the jejunojejunostomy was constructed 15 cm beyond the ligament of Treitz. In the experimental group, the Roux limb was 150 cm in length, and the jejunojejunostomy extended 30 cm beyond the ligament of Treitz. The latter resulted in significantly better weight loss at 2 and 3 years postoperatively and without additional metabolic sequelae or diarrhea. There was not a difference in results between these operations at 4 years, but this may be due to the small number of patients followed that long. It is still unclear whether patients with super obesity are able to eat more than other patients after the same operation or whether they have a significantly lower energy expenditure. Static measurement of resting metabolic
rate expressed in terms of body cell mass does not show a difference between patients with morbid and super obesity. Measurement of energy expenditure over several days may show a difference that could account for the failure of the super obese patients to return to normal weight. Methods are now available to study energy expenditure over long periods [I0,11]. Marginal or stomal ulceration occurred exclusively in patients with GB and only in those who developed a perforation of the vertical staple line, which permitted acid from the bypassed stomach to enter the small gastric pouch and jejunum. Seven of 12 patients with a staple line perforation developed a stomal ulcer, and healing was achieved in all after the staple line was closed, Jordan et al [12], in a report of 34 patients with stomal ulceration after GB, emphasized the importance of staple line disruption as a predisposing cause (22 of 34 patients). They advised resection of the ulcer and the Roux limb with gastrogastrostomy. In the remaining 12 patients, stomal ulceration probably developed due to a large pouch. These results underscore the importance of separating the gastric pouch from the bypassed stomach and constructing a small pouch initially. CONCLUSION IGB was significantly better than GB or VBG in producing successful weight loss. IGB proved to be efficacious in treating super obesity as well as morbid obesity and when used either as a primary or as a remedial operation. REFERENCES 1. MacLean LD, Rhode BM, Forsr RA. Late results of vertical banded gastroplastyfor morbid and super obesity. Surgery 1990; 107: 20-7. 2. MacLean LD, Rhode BM, Forse RA. A gastroplasty which avoids stapling in continuity.Surgery (In press). 3. ReinholdRB. Criticalanalysisof long-termweightlossfollowing gastric bypass. Surg Gynecol Obstct 1982; 155: 385-94. 4. MetropolitanLife Foundation.Height and weight tables. New York: Metropolitan Life Insurance Co., 1983. 5. Sugerman HJ, Starkey JV, Birkenhaucr R. A randomizedprospectivr trial of gastric bypass versus vertical banded gastroplasty for morbid obesity and their effects on swe~ts versus nonswe~ts eaters. Ann Surg 1987; 205: 613-22. 6. Hall JC, Watts JM, O'Brien PE, et al. Gastric surgery for morbidobesity.The AdelaideStudy. Ann Surg 1990; 211:419-27. 7. Linner JH, Drew RL. Why the operationwe prefer is the RouxY gastric bypass. Obesity Surgery 1991; 1: 305-6. 8. Mason EE, Doherty C, Maher JW, Scott DH, RodriguczEM, BlommersTJ. Super obesityand gastricreductionprocedures.Gastrocnterol Clin North Am 1987; 16: 495-502. 9. Brolin RE, Kenlcr HA, German JH, Cody RP. Long-limb gastric bypass in the superobese. A prospectiverandomizedstudy. Ann Surg 1992; 215: 387-95. 10. LifsonN. Theoryof use of the turnoverrates of bodywater for measuring energy and material balance. J Theor Biol 1966; 12: 46-74. 11. BouchardC, TrcrnblayA, Nadeau A, et al. Long-termexercise trainingwith constantenergyintake. 1: Effecton bodycomposition and selected metabolicvariables. Int J Obes 1990; 14: 57-73. 12. Jordan JH, HockingMP, Rout WR, WoodwardER. Marginal
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 JANUARY1993 159
MACLEANETAL
ulcer followinggastric bypass for morbid obesity.Am Surg 1991; 57: 286-8. DISCUSSION John G. Kral (Brooklyn, NY): Your study confirms the results of the prospective Adelaide Study as well as the study by Sugarman in which gastroplasty was compared with gastric bypass. Given the absence of any effective nonsurgical treatment option for severe obesity, you do your results a disservice through your definition of "success." The mortality data from the American Cancer Society can't be applied to a population that has been severely obese and subsequently lost weight. The Metropolitan Life Insurance standards do not pertain to previously obese individuals. In fact, very moderate weight loss, even in the absence of achieving "normal" weight, significantly reduces morbidity and mortality to within the "normal, range. A better definition of "success" should take into account the reductions in co-morbidity and mortality rates compared with severely obese patients treated by nonsurgicai means. Lloyd D. MaeLean: Life expectancy is not usually documented because these patients can't get life insurance, and so actuarial techniques are not available. Quality of life is more likely to be improved as proximity to normal weights is achieved. I don't see why morbidity and mortality in these patients should be different than in acturial studies when the obese patients return to a weight for which we have data. Robert E. Brolin (New Brunswick, N J): I have two questions. First, do you still recommend the use of vertical banded gastroplasty? Second, how do you obtain such a complete work-up? Lloyd D. MaeLean: I do not think that vertical banded gastroplasty is appropriate based on the results of our own work and those of previous studies. The randomized studies of Dr. Sugerman and one from Australia support this point of view. Dr. Ed Mason has reported the results of vertical banded gastroplasty in morbidly obese patients after 5 years (Mason EE. Gastric surgery for morbid obesity. Surg Clin North Am 1992; 72: 501-13). Thirty percent of his patients remained morbidly obese. Based on these data, I conclude that vertical banded gastroplasty is not as successful as other techniques. Patients return because we have a full-time staff. It is probably attributable to the Canadian health care delivery system. There's no cost for patients to return, and we are able to perform endoscopy at follow-up. B. G. Wolfe (Sacramento, CA): We have seen perforated duodenal ulcer after gastric bypass and have been concerned that other patients may have duodenal ulcers, since the parietal cell mass, as you indicated, is below the staple row. If you're going to divide the stomach, why not resect it? Have you ever reused the stomach for re-anastomosis? Lloyd D. MaeLean: We had to perform a re-anastomosis in a patient with an ulcer. We divided between the staple line, and a stenosis developed that did not respond to dilatation. This occurred over a period of several weeks, 160
after which time we decided to excise the area of the ulcer and part of the jejunum and convert it back to normal gastric anatomy. B. G. Woffe: For such patients, esophagojejunostomy with gastrectomy has been a more common remedial procedure. Lloyd D. MaeLean: We always do a gastrostomy to the bypassed stomach in these patients. If access to the bypassed stomach or duodenal area is needed to evaluate for a cancer or ulcer, the radiologist can approach through the area of the gastrostomy scar, gain access to the bypassed stomach, and dilate the tract, thereby permitting endoscopy. We have done that on a few occasions, including cases in which we had to gain access to the duodenum. Dr. Kelly. Did you think that removing the rest of the stomach would increase the morbidity of the operation? Lloyd D. MaeLean: I would have thought so. John M. Kellum (Richmond, VA): At the Medical College of Virginia, we are following more than 700 patients. Of these patients, 200 have been followed for more than 5 years. This is the series of Dr. Sugerman to which you alluded. Our results Show that for gastric bypass, the percentage of excess weight loss at 5 years is 55% versus 40% or less for vertical banded gastroplasty. We perform the operation as described by Ed Mason. There are patients in whom we still perform vertical banded gastroplasty. These are usually well-motivated women who are about 100 lb overweight and who We think will avoid sweets. The use of vertical banded gastroplasty circumvents the potential iron-deficiency problem. Such patients seem to do reasonably well with vertical banded gastroplasty. I disagree withone point. We use a triple superimposed staple line for our gastric bypasses, arid we have had a low rate of disruption (approximately 5%). Would this be a better operation initially, perhaps preserving the divided pouch for the second operation? Lloyd D. MaeLean: I'm concerned about staple line breakdown. We have used every conceivable means to keep the staple lines intact, even by oversewing them. Use of the TA-90B stapler is widely accepted. But, as I talk with physicians who endoscopically examine their patients on a regular basis, I find that staple line perforation is much more common than is generally reported. M. P. Hocking (Gainesville, FL): In my practice in Florida, we used the old two-row stapler and had less than a 10% incidence of recognized staple line dehiscence, although we have not systematically restudied our patients. Could oversewing the staple line contribute to dehiscence? Maybe the staple line isn't failing, but it's actually the fact that small fistulas are developing at the suture sites. How do you explain your high staple line dehiscence rate? The second question relates to the morbidity. What's the downside of dividing the stomach? Clearly, you obtained better results, but was there any excess morbidity in the early postoperative period as a result of division of the stomach?
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY1993
SURGICAL TREATMENT OF OBESITY
Lloyd D. MaeLean: Division of the stomach has not had any effect on the morbidity rate. I suspect that oversewing doesn't contribute to stable line dehiscence, but I haven't examined the data in respect to that question. John Sonneland (Spokane, WA): Dr. MacLean and Dr. Kellum agree that better weight reduction is achieved with gastric bypass than with vertical banded gastroplasty. However, I would like to speak in defense of vertical gastroplasty. Isn't vertical banded gastroplasty a simpler operation with fewer intraoperative complications? Why did 43% of your patients who underwent vertical gastroplasty require a reoperation? In a personal series of more than 40 patients, only 1 of my patients required a reoperation. I have yet to have staple line disruption with a four staple line vertical gastroplasty. Why was there suture line dehiscence if you were using four rows of staples? As a suggestion, would you consider a Silastic ring as an alternative to Marlex, which is known to have problems? If you can advance a gastroscope through the stoma of a vertical banded gastroplasty, would you consider stenosis to be present? Lloyd D. MaeLean: I don't think that there is a difference in the complication rate between vertical banded gastroplasty and gastric bypass. We do not have expehence with any other banding material other than Marlex. We encounter problems with stenosis because we make the opening smaller (45 to 47 mm). We found that when the opening was larger (e.g., 50 mm), patients didn't lose weight. We may have less restrictive indications for reoperating on patients than other surgeons. The goal of our surgery is to help patients lose enough weight so that they're eventually within 50% of their ideal weight. If they don't achieve this goal and there is a technical problem, I reoperate. With the operation as we've now been doing it for about 2 years, we have only had to reoperate on one patient. My explanation for the stenosis is that it's the size we make the Marlex band. If you have also divided between the staple lines, everything must pass through that orifice. It's the only way, in our hands, that we can help patients return to a reasonable weight. The Marlex will erode in an appreciable number of those patients. We used Silastic years ago on a type of horizontal gastric bypass with a Roux-en Y to a pouch at the top of the stomach. We made the anastomosis on the greater curvature, and we put a Silastic ring around the anastomosis. A large number of these rings, even though they were not attached to the stomach or jejunum, eroded into the stomach. Consequently, I am skeptical about foreign materials in these positions. Linner and Drew [7] do put fascia around their anastomoses, without reported difficulty. B. D. Sehirmer (Charlottesville, VA): In my series of nearly 200 gastric bypass operations, I have come to the conclusion that the divided gastric bypass is a superior operation, due to the low but additive incidence of staple
line disruption, which I observed over about 5 years. Are your failures (disruptions) in the gastric bypasses apparent at 1 year, or is this an additive phenomenon? Lloyd D. MaeLean: They occurred principally over the first year after the operation, and this is different from our experience with vertical banded gastroplasties. We have performed vertical banded gastroplasties in which the staple line has perforated 4 or 5 years later. Patients usually notice when perforation occurs because they suddenly find that they can eat anything they want. B. E. Terry (Columbia, MO): The problem that you have with staple line disruption, which occurred in 20% of the gastric bypasses, and the problem that you have with stenosis after gastroplasties might stem from the failure to make the stoma large enough. The TA-90B stapler has a 9-cm staple line, which results in a small pouch. Have you measured the volume of the pouch and at what pressures have these measurements been obtained? In pursuing your quest for a near-ideal weight as the end result, you have used the gastric bypass. After a few years with the unprotected stoma of that small pouch, these patients will develop an accommodation (i.e., a pouch in which the volume is equal to the volumes of the small gastric remnant and the jejunum), which can result in malnutrition. Lloyd D. MaeLean: In my opinion, the size of the stoma doesn't make any difference in weight loss after gastric bypass. The results we've achieved are with a large stoma. Endoscopy performed a year after gastric bypass shows that the stoma is the same size as the jejunum just beyond it. We make the stomas smaller when we start because the pouch itself is so small. We have used nonabsorbable polypropylene around the anastomosis, and endoscopic examination 6 months later showed that all the polypropylene was inside the lumen. I don't think that the results that we have obtained are due to the size of the stoma. The stoma should be large because the patients should eat solid food. The best resuits we have had with gastroplasty have been in patients who were on a pureed liquid diet. We have measured pouch size, but this is not easily done, and we haven't measured pouch size in all our patients. One method is to insert a tube via the esophagus and then to put a tape around the esophagogastric junction to prevent regurgitation. The tube, which has a 70cm column on it, is filled, and the volume is measured before the stomach is divided. I've also instilled fluid until it stops dripping, aspirated it, and measured the volume. In almost every case, the pouch has measured less than 15 mL. B. E. Terry: My point is that if patients have such a small pouch, particularly those who have undergone gastroplasty, they will be forced to eat liquids, usually highcaloric ones. This may account for the failure of surgical treatment. Lloyd D. MaeLean: I agree with your comment as it refers to gastroplasty, but patients who have undergone gastric bypass can eat anything; they just eat smaller quantities. They are totally different from patients who have undergone vertical banded gastrop!asty.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
161
MACLEAN ET AL
Keith A, Kelly (Rochester, MN): I notice that you brought your Roux limb up behind the stomach, rather than anterior to it. Would you comment on that? When you anastomose the jejunal limb to the gastric pouch, do you use staples or sutures for the pouch jejunal anastomosis? Lloyd D. MaeLean: I always do the procedure by hand. It isn't possible to get a stapling device in there. I use the retroeolic, retrogastric approach for the Roux-enY because it is such a short distance. I never have to divide a single jejunum vessel. The distance from the ligament of Treitz up to a lesser curve pouch is very short. E. B. Cabot (Boston, MA): Your incidence of staple line dehiscence is very high in all groups of patients. It is considerably higher (200/0) than any of the reported series in the literature and certainly higher than my own personal experience with more than 200 patients (below 5%). Could the discrepancy in dehiscence rates be related to the size of the outlet of the pouch? Dr. Mason tried decreasing the size of the pouch outlet in the vertical banded gastroplasty to 4.5 em and encountered stenosis and some dehiscence. He returned to the 5-cm size. Lloyd D. MaeLean: A tube is left in the stomach, and
162
THE AMERICAN JOURNAL OF SURGERY
we perform endoscopy in all the patients at 6 weeks, 6 months, and annually, or if their symptoms change. In patients who have undergone a gastric bypass, a 9.8-mm endoscope passes very easily at 6 weeks, which is not the case for patients who have undergone gastroplasty. Yet, both groups of patients have a significant fistula rate, unless the vertical staple lines are separated. Fistulas can occur whether there is significant back pressure from a small orifice with the gastroplasty or when there is minimal back pressure with the gastric bypass. Another Canadian investigator who has used the TA-90B stapler in several hundred patients has a 20% fistula rate. Unless the gastric mucosa is completely destroyed at the time that the staples are applied, the potential for a gastrogastric fistula exists if the staples migrate. Prolonged pressure (for several minutes) at the time the stapler is fired may decrease the incidence of this complication. We prefer to divide between the staple lines to eliminate fistula formation. Unless these patients undergo careful endoscopy or have an upper gastrointestinal series, it can't be assumed that the staple lines are intact.
VOLUME 165
JANUARY 1993
A prospective, randomiT~d trial comparing vertical banded gastroplasty (VBG) and vertical gastric bypass (GB) for obesity was completed in 106 patients who did not differ in baseline body mass index (BMI = kg/m 2) or length of follow-up. The goal of this surgery was to return patients to within 50% of their ideal weight, i.e., a body mass index less than 35 kg/m 2, and to accomplish this while maintaining a low risk for malnutrition as well as other morbidity and mortality. Success was defined as a BMI less than 35 kg/m 2 because the mortality risk increases rapidly above this degree of obesity. Surgical failures were encountered in 43% of the 54 patients in the VBG group, all of whom had division between the vertical staple lines. The main causes of failure were stenosis and enlargement of the gastroplasty orifice. Surgery failed in 23% of the GB-treated patients, due to perforation of the vertical staple line. An isolated gastric bypass (IGB) not dependent on staples was performed as the remedial operation for the failures of both VBG and GB. IGB was significantly better than VBG or GB, with a success rate of 83% compared with 39% for VBG and 58% for GB. Subsequent experience since completion of this randomi,Jed trial in 54 consecutive patients supports IGB for primary, as well as remedial, operations for the morbidly obese (BMI = 40 to 50 kg/m2), as well as for patients who are super obese (BMI greater than 50 kg/m2).
Fromthe Departmentof Surgery(LDM,BMR,JS), McGillUniversity, Montreal,Canada,and the DepartmentofSurgery(RAF),Harvard MedicalSchool,Boston,Massachusetts. Requests for reprints shouldbe addressedto LloydD. MacLean, MD, RoyalVictoriaHospital,Room$10.26, 687 Pine AvenueWest, Montreal,PQ, CanadaH3A IA1. Presentedat the 33rdAnnualMeetingof the Societyfor Surgeryof the AlimentaryTract, San Francisco,California,May 11-13, 1992.
ertical banded gastroplasty (VBG) and gastric V bypass (GB) are the two operations most frequently performed for severe obesity. Our experience with VBG has been unsatisfactory mainly due to breakdown of the vertical staple line. This occurred in 48% of 201 patients recently reported [1]. Many of these patients required reoperation including GB, which resulted in a higher success rate [2]. We also observed over long periods of follow-up (over 4 years) that if the vertical staple line in VBG remained intact and if the orifice size remained less than 11 mm in diameter, the results in 47 patients who were morbidly obese preoperatively were highly satisfactory (0% to 25% excess weight) in 57% and good (26% to 50% excess weight) in an additional 32%. Therefore, a successful result, defined by Reinhold [3] as good or excellent (less than 50% excess weight or a body mass index less than 35 kg/m2), was achieved in 89% of these highly selected patients. We reasoned, therefore, that an ideal operation would require surgical isolation of the pouch with VBG. The purpose of the current study was to compare a VBG not dependent upon staples, i.e., made with a division between the vertical staple lines, with Roux-en-Y GB that had a vertical lesser curvature pouch the same size as that constructed for the VBG (less than 15 mL). The GB was constructed with a large orifice, and staple lines were not divided. PATIENTS AND METHODS A total of 54 patients were randomized to undergo VBG, and 52 patients were randomized to undergo GB. Randomization took place immediately after the surgery had commenced. The study period was from October 1987 to February 1990. All operations were performed through an upper abdominal midline incision using a table-mounted retractor (Gomez retractor, Pilling Canada, Markham, Ontario). For the gastroplasty, the vertical staple line was made with a V. Mueller PI-90 stapler (3M Company, St. Paul, MN) using 4.8-mm staples. This stapler introduces two double rows of staples with an interval of free tissue in between that permits division between the staple lines (Figure 1). The cut edges were sutured to establish hemostasis and prevent leakage. Omentum was sutured between the staple lines. The proximate ILS 25-mm instrument (Ethicon Inc., Somerville, N J) was fired adjacent to a 28 Maloney bougie 4 cm distal to the angle of His. A drain was placed from the left side of the esophagogastric junction to the lesser omental sac distal to the left gastric vessels to identify the angle of His accurately. The volume of this pouch was infrequently measured but was always less than 15 mL and often less than 10 mL. The banding of the VBG was done with polypropylene mesh cut 60 mm in length and 15 mm in width, overlapped and
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993 155
MACLEAN ET AL
Figure 1. Vertical banded gas~'opiasty wil~h division between vertical staple lines. A Maloney bougie no. 28 is in place, and the polypropylene mesh band is 45 to 47 mm in external circumference.
Figure 3. Isolated gastric bypass with separation of the small
gastric pouch from the remaining stomach.
156
Figure 2. Gastric bypass made with TA-90B stapler. The vertical line was oversewn as illustrated. The Roux limb was 40 cm in length and was placed in the retrocolic, retrogastric position.
sutured in place, to achieve a 45- to 47-mm external circumference. This was done with a double piece of polypropylene. The patients who underwent GB had a small pouch of the same size constructed in the same location, but the vertical staple line was constructed with the TA-90B stapler (Auto Suture, U. S. Surgical Corporation, Norwalk, CT), which places four rows of closely approximated staples. The Roux limb of jejunum was made 40 cm in length and was brought to the small gastric pouch in the retrocolic, retrogastric position. The end of the jejunum was dosed, and the anastomosis was made using a single layer of either Prolene 3-0 or P.D.S. 3-0 suture in a running fashion, i.e., end of the gastric pouch to the side of the jejunum (Figure 2). This anastomosis was made approximately 1.0 to 1.5 cm in diameter. In patients in whom either operation failed, a third procedure was performed (Figure 3), in which an isolated gastric bypass (IGB) was performed. In these cases, the Mueller stapler with division between the staple lines was used, thereby separating the small gastric pouch from the remaining stomach. All patients had a closed system drain placed in the left upper quadrant for 2 to 3 days and a gastrostomy to the bypassed stomach that was retained for 6 weeks, at which time the patients underwent initial gastroscopy. Table I summarizes the characteristics of the two groups of patients. Thirty-seven (69%) of the patients who underwent VBG were morbidly obese (BMI = 40 to 50
THE AMERICANJOURNALOF SURGERY VOLUME165 JANUARY1993
SURGICAL TREATMENT OF OBESITY
kg/m2), and 17 (31%) were super obese (BMI greater than 50 kg/m2). Twenty-nine (56%) of the patients who underwent GB were morbidly obese, and 23 (44%) were super obese. The preoperative incidence of super obesity was not statistically different in the two groups (Fisber's exact test; p = 0.12).
Weight loss monitoring: We saw patients once a month for the first 3 months, every 3 months for the first year, and semiannually thereafter. At each visit, patients were wdghed and examined, and their body mass index was calculated. Final foUow-up data were obtained in March 1992. Gastroscopy was performed with a 9.8- or 9-mm instrument at 6 weeks, 6 months, and annually thereafter. Additional gastroscopic examinations were performed when required by a change in clinical status. We made estimates of the orifice size and of the status of the staple line at each endoscopic operation. Urgent endoscopy was performed in patients who developed acute episodes of vomiting after operation. This occurred only in the patients who underwent VBG. If the blockage was caused by food or pills, these were removed by endoscopic techniques. If stenosis of the orifice caused repeated vomiting (VBG only), we ruled out obstruction due to a foreign body and placed the patient on a clear liquid diet. We did not attempt dilatation because previous experience has shown that this technique is unsuccessful. If the vomiting persisted, patients underwent another operation using an IGB. If a hole developed in the staple line, which mainly occurred in the GB group, and the patient was gaining weight or developed a stomal ulcer, IGB was performed. Classification of weight loss: The Reinhold [3] classification was used to describe the postoperative results. Ideal weight was determined by the use of the Metropolitan Life Insurance tables [4]. Results, which were determined at the times of evaluation, were expressed as excellent when the patient had 0% to 25% excess weight; a good result was an excess of 26% to 50%; a fair result, 51% to 75% excess; and a poor result, 76% to 100~ excess. Failure was defined as more than 100% excess weight. For the purposes of this study, we considered surgical treatment successful if the patient's results were categorized as excellent or good and a re.operation was not required to achieve these results. Surgical treatment was defined as a failure if a reoperation was required, regardless of the ultimate outcome. Statistical methods: Differences between groups with respect to continuous variables were evaluated using Student's t-test (two groups) or analysis of variance (ANOVA) (more than two groups). For categoric variables, the x 2 test was used. In addition, odds ratios were calculated for binary outcomes. Ninety-five percent confidence intervals (CIs) were used to assess the precision and statistical significance of the odds ratios. Multiple linear regression and ANOVA were used to test the difference between surgical procedures with re-
TABLE
I
Baseline Characteristics of Study Patients* Vertical Banded Gastroplasty
Gastric Bypass
p Value
52 38.8 • 9.5 278 • 41 48 _-. 7
52 40.1 _+ 7.7 295 -+ 53 50 -+ 7
-0.43 0.07 0.20
No. of patients Age (y) Weight (Ib) Body mass index (kg/m 2) *Valuesexpressedas mean • SD.
TABLE lI
Reasons for Reoperation Vertical Banded Gastroplasty
Gastric Bypass
No. of patients Reoperation
54 23 (43%)
52 12 (23%)
Indications Stenosis Enlarged orifice Staple line fistula Normal but clinical failure Abscess Stomal ulcer
11 7 2 2 1 --
--12 --7
TABLE IlI
Outcome at Final Assessment of Operations
No. of procedures No. of patients Revised to normal No. lost to follow-up Follow-up (mo)* Patient outcome (%)t Excellent Good Fair Poor Failure
Vertical Banded Gastroplasty
Gastric Bypass
Isolated Gastric Bypass
54 31 5 0 38.6 _+ 8.5
52 40 0 0 33.1 -+ 12.4
30 30 0 1 21.8 - 9.3
7 14 6 2 25
(13) (26) (11) (4) (46) ~
Overall Success (%) Failure (%)
12 18 5 3 14
(23) (35) (10) (6) (27)
15 (50) 9 (30) 2 (7) 2 (7) 1 (3)
p--0.27 I p=0.034-~ p = 0.00015
21 (39) 33 (61)
30 (58) 22 (42)
I ~
p = 0.0003
24 (83) 5 (17)
I
*Valuesexpressedas mean -+ SD. lp Valuesare basedon • analysis,
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
157
MACLEAN ET AL
TABLE IV
Multivariate Analysis on Change in Body Mass Index (BMI) F Ratio
p ValUe
Variable Age Follow-up duration Baseline BMI
0.22 1.77 36.95
0.64 0.19 0.0001
Surgery Isolated bypass versus gastroplastry Gastric bypass versus gastroplasty Isolated bypass versus gastric bypass
5.17 11.25 0.93 6.04
0.007 0.001 0.34 0.016
*Values were calculated by analysis of variance.
TABLE V
Patient Characteristics No. of patients No. lost to follow-up Follow-up (me _+ 1 SD) Body mass index (kg/m 2 -+ 1 SD)
54 1 15.9 --- 3.8 48 _+ 7
TABLE Vl
Results of Isolated Gastric Bypass Since Completion of Randomized Trial Parameter New operation Revision operation Morbid obese Super obese
No.
Success*
36 17 34 19
32 14 32 14
(89%) (82%) (94%) (74%)
*Success = < 35 kg/m 2 = Good + Excellent = < 50% excessweight.
spect to the change in BMI from baseline while controlling for the patient's age, the duration of follow-up, and the baseline BMI. For this analysis, planned contrasts were used to evaluate all possible pairwise comparisons of the three surgical procedures. RESULTS There were no deaths. One patient has been lost to follow-up. Five patients who were converted to normal from VBG are classified as failures. These patients did not undergo IGB. Forty-three percent of 54 patients who underwent VBG required re.operation for the reasons outlined in Table IL Eighteen of the 23 patients were converted to IGB. Twelve or 23% of patients who underwent GB developed a hole in the staple line and were converted to an IGB not dependent on staples. Therefore, a total of 54 VBG operations and 82 GB or IGB operations were performed on the 106 patients. The success rate in the combined bypass group was 67% compared with 39% for the VBG group (p --- 0.003, odds ratio (OR) = 3.1; 95% CI -- 1.45 to 6.87). The final outcomes are shown in Table 158
THE AMERICAN JOURNAL OF SURGERY
HI, which is an analysis by surgical procedure. The outcome is classified according to the patients' weight at the time of final evaluation as described above. The results in Table III show that significantly better results were achieved with the remedial operation IGB than with either VBG or GB. Table III also lists the success rate for each procedure, which was clef'reed in this study as an excellent or good outcome, no requirement for re.operation, a BMI less than 35 kg/m 2, or less than 50% excess weight. These data show that the rate of successful outcome was significantly higher with IGB compared with the other two procedures. Erosion of the Marlex mesh was associated with both stenosis and enlargement of the orifice after gastroplasty in 13 of 23 patients who required reoperation (Table II). Staple line dehiscence was followed by stomal ulcer in 7 of 12 patients who required operation after GB. These ulcers all healed after reoperation and conversion to IGB. In one patient, the single failure, reoperation was necessary because severe stenosis at the previous ulcer site did not respond to dilatation. The surgery was reversed in this patient. Multivariate analysis comparing the three procedures while controlling for the age of the patient, the baseline BMI, and the duration of follow-up showed that IGB resulted in a significantly higher reduction in the BMI, This analysis emphasized the importance of the baseline BMI as a potential confounder (Table IV). Since completing this study, 54 consecutive patients have undergone IGB (Tables V and VI) and have been followed for a minimum of 1 year (mean: 15.9 4- 3.8 months). These results, albeit with short follow-up times, are better than any previous experience at our clinic and support the use of this procedure both for morbidly as well as for super obese patients and for initial as well as remedial surgery. Serious morbidity due to subphrenic abscess or leak from the anastomosis or staple line occurred in 6 patients among the total of 106 patients in the prospective randomized trial plus the 54 patients who have undergone surgery since completion of the trial (3.8% of patients). These 160 patients underwent a total of 195 operations (3.1% of operations). Leak or intraperitoneal abscess formation occurred after the first operation in three patients (two VBGs and one GB). In the remaining three patients, leak or abscess formation followed the remedial operation, which was IGB. In five of the six patients, laparotomy was necessary, and, in one patient, aspiration alone was successful. COMMENTS This randomized trial strongly supports GB over VBG, especially when staple line disruption is eliminated as a complication in the former operation. This preference of GB over VBG has previously been supported by the excellent results obtained in prospective trials by Sugerman et al [5] and by Hall et al [6]. The operation we prefer, which has evolved from the data herein reported and our previous experience, embod-
VOLUME165
JANUARY1993
SURGICAL TREATMENT OF OBESITY
ies the following characteristics: a small pouch less than 15 mL in volume; a pouch completely separated from the remaining stomach and not dependent upon staples; a pouch that empties dependently so that there is no stasis that can lead to regurgitation or vomiting and enlargement of the pouch due to distention; and an orifice from the pouch not dependent on a foreign body to maintain its size or diameter and of a size that permits the ingestion of solid food. In our slow pursuit of the best operation to treat obesity, we have adopted a procedure that closely approximates that preferred by Linner and Drew [7], with the exception that we do not use external support of the gastrojejunostomy. Linner and Drew [7] reported that 53% of patients were within 25% of their ideal weight and another 30% had 26% to 50% excess weight (an 83% success rate) after 5 or more years. The incidence of super obesity was not given by these authors. We believe that a vertically placed pouch is less likely to dilate because of prompt emptying. The pouch should be separated from the main stomach and not be dependent on staples in order to prevent fistula formation. A hand-sutured anastomosis is done because it is easier than using stapling instruments on a very small pouch in the confined space available. In our experience, attempts to limit the size of the anastomosis by using a running, continuous, nonabsorbable suture fail because the suture material migrates into the lumen, and the anastomosis over subsequent months becomes the same size as the jejunum beyond the anastomosis. Patients who have successful results after VBG, as we have performed the operation, usually ingest a pureed or liquid diet. After GB or IGB, patients can eat anything but in limited quantities. Diarrhea has occurred in a minority of GB patients and only during the first month postoperatively. A feeling of fullness, dizziness, and a desire to lie down after eating (dumping syndrome) is frequently encountered, but since these symptoms are associated with decreased intake and weight loss, most patients do not complain about them. No patient has required specific medication for these symptoms. The results of gastric restrictive surgery have been poorer for patients who were super obese preoperatively than for patients who were morbidly obese preoperatively [8]. Brolin et al [9] prospectively studied a modification of the Roux-en-Y gastric bypass for super obese patients. In their conventional bypass, the Roux limb was 75 cm in length, and the jejunojejunostomy was constructed 15 cm beyond the ligament of Treitz. In the experimental group, the Roux limb was 150 cm in length, and the jejunojejunostomy extended 30 cm beyond the ligament of Treitz. The latter resulted in significantly better weight loss at 2 and 3 years postoperatively and without additional metabolic sequelae or diarrhea. There was not a difference in results between these operations at 4 years, but this may be due to the small number of patients followed that long. It is still unclear whether patients with super obesity are able to eat more than other patients after the same operation or whether they have a significantly lower energy expenditure. Static measurement of resting metabolic
rate expressed in terms of body cell mass does not show a difference between patients with morbid and super obesity. Measurement of energy expenditure over several days may show a difference that could account for the failure of the super obese patients to return to normal weight. Methods are now available to study energy expenditure over long periods [I0,11]. Marginal or stomal ulceration occurred exclusively in patients with GB and only in those who developed a perforation of the vertical staple line, which permitted acid from the bypassed stomach to enter the small gastric pouch and jejunum. Seven of 12 patients with a staple line perforation developed a stomal ulcer, and healing was achieved in all after the staple line was closed, Jordan et al [12], in a report of 34 patients with stomal ulceration after GB, emphasized the importance of staple line disruption as a predisposing cause (22 of 34 patients). They advised resection of the ulcer and the Roux limb with gastrogastrostomy. In the remaining 12 patients, stomal ulceration probably developed due to a large pouch. These results underscore the importance of separating the gastric pouch from the bypassed stomach and constructing a small pouch initially. CONCLUSION IGB was significantly better than GB or VBG in producing successful weight loss. IGB proved to be efficacious in treating super obesity as well as morbid obesity and when used either as a primary or as a remedial operation. REFERENCES 1. MacLean LD, Rhode BM, Forsr RA. Late results of vertical banded gastroplastyfor morbid and super obesity. Surgery 1990; 107: 20-7. 2. MacLean LD, Rhode BM, Forse RA. A gastroplasty which avoids stapling in continuity.Surgery (In press). 3. ReinholdRB. Criticalanalysisof long-termweightlossfollowing gastric bypass. Surg Gynecol Obstct 1982; 155: 385-94. 4. MetropolitanLife Foundation.Height and weight tables. New York: Metropolitan Life Insurance Co., 1983. 5. Sugerman HJ, Starkey JV, Birkenhaucr R. A randomizedprospectivr trial of gastric bypass versus vertical banded gastroplasty for morbid obesity and their effects on swe~ts versus nonswe~ts eaters. Ann Surg 1987; 205: 613-22. 6. Hall JC, Watts JM, O'Brien PE, et al. Gastric surgery for morbidobesity.The AdelaideStudy. Ann Surg 1990; 211:419-27. 7. Linner JH, Drew RL. Why the operationwe prefer is the RouxY gastric bypass. Obesity Surgery 1991; 1: 305-6. 8. Mason EE, Doherty C, Maher JW, Scott DH, RodriguczEM, BlommersTJ. Super obesityand gastricreductionprocedures.Gastrocnterol Clin North Am 1987; 16: 495-502. 9. Brolin RE, Kenlcr HA, German JH, Cody RP. Long-limb gastric bypass in the superobese. A prospectiverandomizedstudy. Ann Surg 1992; 215: 387-95. 10. LifsonN. Theoryof use of the turnoverrates of bodywater for measuring energy and material balance. J Theor Biol 1966; 12: 46-74. 11. BouchardC, TrcrnblayA, Nadeau A, et al. Long-termexercise trainingwith constantenergyintake. 1: Effecton bodycomposition and selected metabolicvariables. Int J Obes 1990; 14: 57-73. 12. Jordan JH, HockingMP, Rout WR, WoodwardER. Marginal
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 JANUARY1993 159
MACLEANETAL
ulcer followinggastric bypass for morbid obesity.Am Surg 1991; 57: 286-8. DISCUSSION John G. Kral (Brooklyn, NY): Your study confirms the results of the prospective Adelaide Study as well as the study by Sugarman in which gastroplasty was compared with gastric bypass. Given the absence of any effective nonsurgical treatment option for severe obesity, you do your results a disservice through your definition of "success." The mortality data from the American Cancer Society can't be applied to a population that has been severely obese and subsequently lost weight. The Metropolitan Life Insurance standards do not pertain to previously obese individuals. In fact, very moderate weight loss, even in the absence of achieving "normal" weight, significantly reduces morbidity and mortality to within the "normal, range. A better definition of "success" should take into account the reductions in co-morbidity and mortality rates compared with severely obese patients treated by nonsurgicai means. Lloyd D. MaeLean: Life expectancy is not usually documented because these patients can't get life insurance, and so actuarial techniques are not available. Quality of life is more likely to be improved as proximity to normal weights is achieved. I don't see why morbidity and mortality in these patients should be different than in acturial studies when the obese patients return to a weight for which we have data. Robert E. Brolin (New Brunswick, N J): I have two questions. First, do you still recommend the use of vertical banded gastroplasty? Second, how do you obtain such a complete work-up? Lloyd D. MaeLean: I do not think that vertical banded gastroplasty is appropriate based on the results of our own work and those of previous studies. The randomized studies of Dr. Sugerman and one from Australia support this point of view. Dr. Ed Mason has reported the results of vertical banded gastroplasty in morbidly obese patients after 5 years (Mason EE. Gastric surgery for morbid obesity. Surg Clin North Am 1992; 72: 501-13). Thirty percent of his patients remained morbidly obese. Based on these data, I conclude that vertical banded gastroplasty is not as successful as other techniques. Patients return because we have a full-time staff. It is probably attributable to the Canadian health care delivery system. There's no cost for patients to return, and we are able to perform endoscopy at follow-up. B. G. Wolfe (Sacramento, CA): We have seen perforated duodenal ulcer after gastric bypass and have been concerned that other patients may have duodenal ulcers, since the parietal cell mass, as you indicated, is below the staple row. If you're going to divide the stomach, why not resect it? Have you ever reused the stomach for re-anastomosis? Lloyd D. MaeLean: We had to perform a re-anastomosis in a patient with an ulcer. We divided between the staple line, and a stenosis developed that did not respond to dilatation. This occurred over a period of several weeks, 160
after which time we decided to excise the area of the ulcer and part of the jejunum and convert it back to normal gastric anatomy. B. G. Woffe: For such patients, esophagojejunostomy with gastrectomy has been a more common remedial procedure. Lloyd D. MaeLean: We always do a gastrostomy to the bypassed stomach in these patients. If access to the bypassed stomach or duodenal area is needed to evaluate for a cancer or ulcer, the radiologist can approach through the area of the gastrostomy scar, gain access to the bypassed stomach, and dilate the tract, thereby permitting endoscopy. We have done that on a few occasions, including cases in which we had to gain access to the duodenum. Dr. Kelly. Did you think that removing the rest of the stomach would increase the morbidity of the operation? Lloyd D. MaeLean: I would have thought so. John M. Kellum (Richmond, VA): At the Medical College of Virginia, we are following more than 700 patients. Of these patients, 200 have been followed for more than 5 years. This is the series of Dr. Sugerman to which you alluded. Our results Show that for gastric bypass, the percentage of excess weight loss at 5 years is 55% versus 40% or less for vertical banded gastroplasty. We perform the operation as described by Ed Mason. There are patients in whom we still perform vertical banded gastroplasty. These are usually well-motivated women who are about 100 lb overweight and who We think will avoid sweets. The use of vertical banded gastroplasty circumvents the potential iron-deficiency problem. Such patients seem to do reasonably well with vertical banded gastroplasty. I disagree withone point. We use a triple superimposed staple line for our gastric bypasses, arid we have had a low rate of disruption (approximately 5%). Would this be a better operation initially, perhaps preserving the divided pouch for the second operation? Lloyd D. MaeLean: I'm concerned about staple line breakdown. We have used every conceivable means to keep the staple lines intact, even by oversewing them. Use of the TA-90B stapler is widely accepted. But, as I talk with physicians who endoscopically examine their patients on a regular basis, I find that staple line perforation is much more common than is generally reported. M. P. Hocking (Gainesville, FL): In my practice in Florida, we used the old two-row stapler and had less than a 10% incidence of recognized staple line dehiscence, although we have not systematically restudied our patients. Could oversewing the staple line contribute to dehiscence? Maybe the staple line isn't failing, but it's actually the fact that small fistulas are developing at the suture sites. How do you explain your high staple line dehiscence rate? The second question relates to the morbidity. What's the downside of dividing the stomach? Clearly, you obtained better results, but was there any excess morbidity in the early postoperative period as a result of division of the stomach?
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY1993
SURGICAL TREATMENT OF OBESITY
Lloyd D. MaeLean: Division of the stomach has not had any effect on the morbidity rate. I suspect that oversewing doesn't contribute to stable line dehiscence, but I haven't examined the data in respect to that question. John Sonneland (Spokane, WA): Dr. MacLean and Dr. Kellum agree that better weight reduction is achieved with gastric bypass than with vertical banded gastroplasty. However, I would like to speak in defense of vertical gastroplasty. Isn't vertical banded gastroplasty a simpler operation with fewer intraoperative complications? Why did 43% of your patients who underwent vertical gastroplasty require a reoperation? In a personal series of more than 40 patients, only 1 of my patients required a reoperation. I have yet to have staple line disruption with a four staple line vertical gastroplasty. Why was there suture line dehiscence if you were using four rows of staples? As a suggestion, would you consider a Silastic ring as an alternative to Marlex, which is known to have problems? If you can advance a gastroscope through the stoma of a vertical banded gastroplasty, would you consider stenosis to be present? Lloyd D. MaeLean: I don't think that there is a difference in the complication rate between vertical banded gastroplasty and gastric bypass. We do not have expehence with any other banding material other than Marlex. We encounter problems with stenosis because we make the opening smaller (45 to 47 mm). We found that when the opening was larger (e.g., 50 mm), patients didn't lose weight. We may have less restrictive indications for reoperating on patients than other surgeons. The goal of our surgery is to help patients lose enough weight so that they're eventually within 50% of their ideal weight. If they don't achieve this goal and there is a technical problem, I reoperate. With the operation as we've now been doing it for about 2 years, we have only had to reoperate on one patient. My explanation for the stenosis is that it's the size we make the Marlex band. If you have also divided between the staple lines, everything must pass through that orifice. It's the only way, in our hands, that we can help patients return to a reasonable weight. The Marlex will erode in an appreciable number of those patients. We used Silastic years ago on a type of horizontal gastric bypass with a Roux-en Y to a pouch at the top of the stomach. We made the anastomosis on the greater curvature, and we put a Silastic ring around the anastomosis. A large number of these rings, even though they were not attached to the stomach or jejunum, eroded into the stomach. Consequently, I am skeptical about foreign materials in these positions. Linner and Drew [7] do put fascia around their anastomoses, without reported difficulty. B. D. Sehirmer (Charlottesville, VA): In my series of nearly 200 gastric bypass operations, I have come to the conclusion that the divided gastric bypass is a superior operation, due to the low but additive incidence of staple
line disruption, which I observed over about 5 years. Are your failures (disruptions) in the gastric bypasses apparent at 1 year, or is this an additive phenomenon? Lloyd D. MaeLean: They occurred principally over the first year after the operation, and this is different from our experience with vertical banded gastroplasties. We have performed vertical banded gastroplasties in which the staple line has perforated 4 or 5 years later. Patients usually notice when perforation occurs because they suddenly find that they can eat anything they want. B. E. Terry (Columbia, MO): The problem that you have with staple line disruption, which occurred in 20% of the gastric bypasses, and the problem that you have with stenosis after gastroplasties might stem from the failure to make the stoma large enough. The TA-90B stapler has a 9-cm staple line, which results in a small pouch. Have you measured the volume of the pouch and at what pressures have these measurements been obtained? In pursuing your quest for a near-ideal weight as the end result, you have used the gastric bypass. After a few years with the unprotected stoma of that small pouch, these patients will develop an accommodation (i.e., a pouch in which the volume is equal to the volumes of the small gastric remnant and the jejunum), which can result in malnutrition. Lloyd D. MaeLean: In my opinion, the size of the stoma doesn't make any difference in weight loss after gastric bypass. The results we've achieved are with a large stoma. Endoscopy performed a year after gastric bypass shows that the stoma is the same size as the jejunum just beyond it. We make the stomas smaller when we start because the pouch itself is so small. We have used nonabsorbable polypropylene around the anastomosis, and endoscopic examination 6 months later showed that all the polypropylene was inside the lumen. I don't think that the results that we have obtained are due to the size of the stoma. The stoma should be large because the patients should eat solid food. The best resuits we have had with gastroplasty have been in patients who were on a pureed liquid diet. We have measured pouch size, but this is not easily done, and we haven't measured pouch size in all our patients. One method is to insert a tube via the esophagus and then to put a tape around the esophagogastric junction to prevent regurgitation. The tube, which has a 70cm column on it, is filled, and the volume is measured before the stomach is divided. I've also instilled fluid until it stops dripping, aspirated it, and measured the volume. In almost every case, the pouch has measured less than 15 mL. B. E. Terry: My point is that if patients have such a small pouch, particularly those who have undergone gastroplasty, they will be forced to eat liquids, usually highcaloric ones. This may account for the failure of surgical treatment. Lloyd D. MaeLean: I agree with your comment as it refers to gastroplasty, but patients who have undergone gastric bypass can eat anything; they just eat smaller quantities. They are totally different from patients who have undergone vertical banded gastrop!asty.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
161
MACLEAN ET AL
Keith A, Kelly (Rochester, MN): I notice that you brought your Roux limb up behind the stomach, rather than anterior to it. Would you comment on that? When you anastomose the jejunal limb to the gastric pouch, do you use staples or sutures for the pouch jejunal anastomosis? Lloyd D. MaeLean: I always do the procedure by hand. It isn't possible to get a stapling device in there. I use the retroeolic, retrogastric approach for the Roux-enY because it is such a short distance. I never have to divide a single jejunum vessel. The distance from the ligament of Treitz up to a lesser curve pouch is very short. E. B. Cabot (Boston, MA): Your incidence of staple line dehiscence is very high in all groups of patients. It is considerably higher (200/0) than any of the reported series in the literature and certainly higher than my own personal experience with more than 200 patients (below 5%). Could the discrepancy in dehiscence rates be related to the size of the outlet of the pouch? Dr. Mason tried decreasing the size of the pouch outlet in the vertical banded gastroplasty to 4.5 em and encountered stenosis and some dehiscence. He returned to the 5-cm size. Lloyd D. MaeLean: A tube is left in the stomach, and
162
THE AMERICAN JOURNAL OF SURGERY
we perform endoscopy in all the patients at 6 weeks, 6 months, and annually, or if their symptoms change. In patients who have undergone a gastric bypass, a 9.8-mm endoscope passes very easily at 6 weeks, which is not the case for patients who have undergone gastroplasty. Yet, both groups of patients have a significant fistula rate, unless the vertical staple lines are separated. Fistulas can occur whether there is significant back pressure from a small orifice with the gastroplasty or when there is minimal back pressure with the gastric bypass. Another Canadian investigator who has used the TA-90B stapler in several hundred patients has a 20% fistula rate. Unless the gastric mucosa is completely destroyed at the time that the staples are applied, the potential for a gastrogastric fistula exists if the staples migrate. Prolonged pressure (for several minutes) at the time the stapler is fired may decrease the incidence of this complication. We prefer to divide between the staple lines to eliminate fistula formation. Unless these patients undergo careful endoscopy or have an upper gastrointestinal series, it can't be assumed that the staple lines are intact.
VOLUME 165
JANUARY 1993