Patients in this study were seen or (in fewer than 20 patients) contacted by phone during the first 6 months of 1998. The weights in the patients not seen in the clinic compared logically with the weights recorded in the clinic during the previous year. The minimal follow-up was set at 3 years. No patients were included who were followed up for 3 years and then lost to follow-up.

The IGB technique has been previously reported. ¹ A small, 4-cm-long pouch on the lesser curvature was made adjacent to a 28 or 30 Maloney bougie using a V. Mueller PI-90 stapler (MMM Company, St. Paul, MN) using 4.8-mm staples. This stapler is used to make two double rows of staples with an interval of free tissue in between that permits division by sharp dissection or cautery (Fig. 1) . The pouch side was oversewn with PDS suture, and the gastric side was inverted. Omentum was sutured between the staple lines. A proximal loop of jejunum was divided, the distal end was advanced in a retrocolic retrogastric position, and an anastomosis was made between the end of the gastric pouch and the side of the jejunum using a single running PDS suture. The anastomosis was 0.8 to 1.0 cm in diameter but was always enlarged to the diameter of the adjacent jejunum, as determined by endoscopy after 6 months to 1 year. This enlargement occurred whether absorbable or nonabsorbable suture material was used.

Upper gastrointestinal x-ray examinations were performed in patients who appeared to have a gastrogastric fistula on endoscopy or in patients with less-than-expected weight loss.

We used a modification of the Reinhold classification to document results (Table 1). ³ An excellent or good result (BMI ≤ 35 kg/m², or loss of excess weight > 50%) was considered a success. We compared the results of the patients classified before surgery as obese or morbidly obese and those classified as super-obese. In each group, we further assessed success on an individual basis. Results are reported as mean ± SD.

One patient died on the second postoperative day of pulmonary embolism. There were two late deaths, resulting from suicide at 4.75 years after surgery and alcoholic cirrhosis of the liver at 6 years after surgery. Both patients had a BMI < 30 kg/m² at the time of death.

Table 2 compares the mean BMI in the obese plus morbidly obese group with that of the super-obese group before surgery, at the time of the lowest weight, and at the end of the trial. In the obese and morbidly obese group, the BMI declined from 44 ± 3 kg/m² to 26 ± 4 kg/m² at 2.1 ± 1.4 years after surgery. The super-obese group fell from 56 ± 6 kg/m² to 31 ± 5 kg/m² at 2.3 ± 1.5 years. The mean BMI rose slightly with time, but this did not change the rating for the groups (i.e., excellent for the obese and morbidly obese and good for the super-obese). The final mean BMI was 29 ± 4 kg/m² for the obese and morbidly obese group and 35 ± 7 kg/m² for the super-obese group.

Although 60% of the obese and morbidly obese group achieved an excellent result, 33% were ranked as good and 7% had a final BMI > 35 kg/m². The spread was even greater for the super-obese: 31% obtained a good result and 26% achieved an excellent result, and treatment failed in 43%. Therefore, the IGB succeeded in 93% of the obese or morbidly obese patients and in 57% of the super-obese patients.

Our experience suggested that intestinal bypass was unacceptable because of a high incidence of protein malnutrition, with impaired liver function. ⁵ We have since favored operations that limit intake over procedures designed to decrease absorption. We also found vertical banded gastroplasty (VBG) inadequate because of breakdown of the staple line and patient dissatisfaction. ⁶ We made the VBG collar 45 to 47 mm rather than 50 mm, as recommended by Mason. ⁷ We found that if the staple line did not perforate and an outlet of the pouch was maintained at less than 12 mm, results of VBG were very acceptable. A prospective randomized trial comparing the VBG with division between the staple lines and Roux-en-Y gastric bypass without division showed that the gastric bypass was better. However, the success rate (BMI < 35 kg/m² and loss of excess weight >50%) was only 58% at 33 months after surgery. ¹

In the above-mentioned trial, failures of both VBG and gastric bypass were converted to IGB to decrease the incidence of solid food intolerance in the patients undergoing VBG and to repair and prevent fistula formation in the patients undergoing gastric bypass. ¹ This redo operation, the IGB, proved more successful than gastric bypass or VBG during short-term follow-up (35.8 ± 19.4 months). ⁶ When the IGB was used as an initial operation or for morbid obesity only, the success rate was 96%. If used as a redo procedure or for super-obesity, the success rate was 63%. ⁸

Long-term follow-up of the IGB is indicated for several reasons. There are few studies in which 90% of the patients have been followed up for an average of more than 5 years, and none have provided data on individual patient outcomes. The IGB is constructed so that the gastrojejunostomy is as large as possible. Absorbable sutures are used because nonabsorbable sutures, when used in previous studies, frequently migrated into the lumen and could cause symptomatic erosions at the suture line. Patient satisfaction was also a reason to avoid attempting to limit oral intake, either by the size of the anastomosis or with the use of external bands around the outlet of the pouch. Constructing a very small pouch, which is possible on the lesser curvature, is a distinct feature of this operation. The pouch was always less than 30 mL when measured, and usually less than 15 mL.

In the present study, 88.6% of 274 patients were followed up for 3 to 8.4 years (mean 5.5 years). The overall results were excellent for patients who were obese or morbidly obese before surgery and good for those classified as super-obese. Vomiting and diarrhea were unusual and never persisted.

Weight loss in these patients cannot be attributed to the dumping syndrome, which was unusual after 1 year and could be avoided entirely if patients did not ingest dairy products. There was increased satiety after decreased intake. With the exception of dairy products, the patients could eat anything, and mincing or liquefying foods was not required. Diarrhea, abdominal bloating, clinically relevant malodorous stools, and excessive flatus were not sequelae of this operation.

The excellent publication by Marceau et al ⁹ provides enough data, with a follow-up of more than 90% of patients, to make a comparison between the IGB and the biliopancreatic diversion with duodenal switch (DS). We could not find any other report in the literature of Roux-en-Y gastric bypass with or without division with a nearly 90% follow-up that provided data on the BMI before and after surgery together with weight loss, maximum weight lost, and number of patients with a BMI < 35 kg/m² at final assessment. Table 3 compares these two operations. The DS operation incorporates a sleeve resection of the stomach with retention of the vagi, the antrum, and a short segment of the duodenum. The excision of gastric tissue is equal to that removed by subtotal distal gastrectomy. The ileoileal anastomosis is made 100 cm proximal to the ileocecal valve, resulting in a 100-cm common limb. The whole alimentary limb is 250 mm (Fig. 2). This is the operation favored over the standard biliopancreatic diversion by this group. The DS operation produced greater weight loss than the standard biliopancreatic diversion, with fewer side effects, in an assessment of 690 patients. ⁹ Comparison of IGB and DS shows a longer follow-up, higher preoperative weight, and greater weight loss and greater maximum weight loss for IGB. The final average BMI was 30 ± 7 kg/m² for DS versus 31 ± 6 kg/m² for IGB. The final percentage of patients with a BMI < 35 kg/m² was 81% for DS and 79% for IGB.

There was a gain in weight from the nadir weight after both operations. The maximum weight loss was 53 ± 20 kg after DS and 56 ± 17 kg after IGB. At final analysis, these were 46 ± 20 kg and 48 ± 18 kg, respectively. The weight gain from nadir was comparable for the two procedures: 7 kg for DS and 8 kg for IGB.

Although IGB avoids diarrhea, clinically important malodorous stools, flatus, and probably malabsorption of calcium, iron deficiency is common and not easily prevented or treated. ¹⁰ However, severe anemia is unusual. The vertically placed gastric bypass in IGB permits the formation of a very small gastric pouch without stasis that remains small over time. ² This has not been the case in our experience with horizontally placed gastric pouches or with vertically placed pouches with constrictions to outflow. In a previous study, we performed upper gastrointestinal x-ray examinations 1 year apart to measure the size of the gastric pouch. Pouch volume remained <30 mL during this period, and the mean BMI was 31.2 kg/m² before and 32.0 kg/m² after 1 year in 18 patients. ² A small pouch that stays small is probably an important reason for the improved results seen in this study, which make it comparable to biliopancreatic diversions. X-ray examinations performed 6 to 8 years after surgery continue to show very small pouches.

Although the average result of IGB for morbidly obese patients is excellent, some obtain only a good result, and treatment fails for unexplained reasons in a few (7%). Likewise, the average result for the super-obese patients is good, but treatment fails in many (43%); surprisingly, 26% obtained an excellent result. These varying responses cannot be predicted before surgery, nor can they be explained after surgery at present. Genetic predisposition may be a significant factor. Childhood onset and parental obesity are known to be strong predictors of obesity in families. The prevalence of obesity (BMI > 40 kg/m²) has recently been compared among 235 families of patients who were candidates for surgical treatment of obesity and 152 families of normal-weight control subjects matched for sex and age. ¹¹ Results showed that obese patients were 25 times more likely to have a first-degree relative with morbid obesity than were those in the control groups (OR = 24.5; 95% CI = 10.4–57.2). In the families of the obese patients, mothers were twice as likely to be severely obese than fathers, but most (57%) of the siblings within the families were not obese (BMI < 30 kg/m²). These results do not help in patient selection. The super-obese do not do as well as the obese and morbidly obese in terms of approximation to normal weight, but we are not convinced that adding a malabsorption component to the procedure would add to their well-being. Many super-obese patients already obtain an excellent result with IGB alone. However, advances in our knowledge of genetic predisposition to obesity may in the future make it possible to predict successful major weight loss, approximation to normal weight, and risk of relapse.

Although the incidence of gastrogastric fistula was much lower in this study (2%) than in other reports, there was still a disturbing incidence of ventral hernia (16%). This operation can be performed laparoscopically, and the long-term results reported here make this approach attractive if for no other reason than to avoid ventral hernia (personal communications from P. Schauer, University of Pittsburgh, Pittsburgh, PA, 1999, and M. Gagner, Mount Sinai Hospital, New York, NY, 1999). ¹²