Effects of Obesity Surgery on Non–Insulin-Dependent Diabetes Mellitus

Context  Most individuals who have non–insulin-dependent diabetes mellitus are obese. The obese population has proved a frustrating entity regarding weight loss and diabetes control. Results of medical weight loss programs, medications, and behavior therapy have proved disappointing.

Hypothesis  Bariatric surgery is the most effective method of diabetes management and cure in the morbidly obese population. Surgical procedures to cause malabsorption provide a more dramatic effect on diabetes owing to the imparted bypass of the hormonally active foregut.

Data Sources  Pertinent journal articles spanning the last 40 years, as well as textbooks.

Conclusions  Bariatric surgical procedures have proven a much more successful method of weight loss and diabetes control in the obese population than conservative methods. These surgical procedures have proven safe with reported mortality rates of 0% to 1.5%. Bariatric operations may be divided based on the method of weight loss and effect on diabetes. The first category is restrictive and includes vertical banded gastroplasty and adjustable silicone gastric banding. These operations improve diabetes by decreasing food intake and body weight with a slowing of gastric emptying. The second category not only contains restrictive components but also elements of malabsorption. This category includes the Roux-en-Y gastric bypass and biliary-pancreatic diversion, which bypass the foregut. Although all of the surgical procedures for obesity offer improved weight loss and diabetes control compared with conservative methods, the Roux-en-Y gastric bypass and biliary-pancreatic diversion offer superior weight loss and resolution of diabetes. The more dramatic effect seen in the surgical procedures to cause malabsorption is likely secondary to the bypass of the foregut resulting in increased weight loss and elevation of the enteroglucagon level.

DIABETES MELLITUS is a major health problem. In 1993 the National Institutes of Health estimated that diabetes affected 7.8 million people. Diabetes accounts for the death of approximately 150 000 Americans per year and for $100 billion each year in health costs.^1,2 There are 2 types of diabetes, type 1 juvenile-onset–insulin-dependent diabetes mellitus and type 2 non–insulin-dependent diabetes mellitus (NIDDM). More than 85% of those individuals having diabetes are affected by NIDDM.^1,3 Diabetes is the leading cause of blindness, renal failure, and amputations in the United States.^4,5 In addition, diabetes imparts a 3 times greater risk of cardiovascular attack and a 2 times greater incidence of cerebrovascular accident.^4,5 Impaired glucose tolerance (IGT) is a condition characterized by abnormal glucose metabolism that does not meet the qualifying requirements of NIDDM. Impaired glucose tolerance imparts a higher risk of NIDDM and, like NIDDM, also increases the risk of microvascular disease.^6-13 A longitudinal study of the natural history of IGT revealed that one quater of those who have it went on to develop NIDDM at 5 years and two thirds progressed to NIDDM by 10 years.¹⁴

The goal of diabetes management is to achieve euglycemia as well as to delay and prevent development of complications, such as retinopathy and nephropathy. Current therapy for NIDDM includes dietary restriction, exercise, insulin, and oral hypoglycemic agents. Despite these various treatment modalities, NIDDM remains a difficult condition to manage.

Worldwide obesity affects more than 250 million individuals, 34 million of whom are Americans, of which approximately 10 million qualify as morbidly obese, with a body mass index (calculated as weight in kilograms divided by the height in meters squared) greater than 40.^5,8,15-17 The incidence of overweight or obesity increased from 24.3% to 33.3% between 1960 and 1991.^18-20

Non–insulin-dependent diabetes mellitus affects approximately 8% of the adult population in the United States and an additional 11% have IGT.^21,22 The prevalence of NIDDM and IGT is increased by obesity, age, and a sedentary lifestyle.^{5,8,11,12,15,21-28} Obesity is a recognized cause of insulin resistance (IR) that leads to IGT and, in some cases, to NIDDM.^{8,12,23,26,29-32} The chances for developing diabetes doubles for every 20% increase over the desirable weight.³³ Most patients older than 50 years who have NIDDM are obese.^1,26 Patients who have morbid obesity have a 10% to 28% incidence of NIDDM; an additional 10% to 31% have IGT.^{1,3,5,13,23,24,29,34-37}

The association of obesity, NIDDM, and IR has long been recognized, but the exact mechanism of the association is unknown. The multiple diseases or comorbidities associated with central obesity are termed the "metabolic syndrome," "syndrome X," or the "IR metabolic syndrome of obesity."³⁸ It is postulated that IR leads to an ever-increasing demand on the pancreatic islet cells to produce more insulin. This cycle continues until the islet cells become exhausted and are no longer able to keep up with the increased insulin requirements, leading to hyperglycemia, IGT, and NIDDM.^{1,8,12,14,28-30,32,39}

Two human genes that may play a role in the cause of the metabolic syndrome X have been identified. A quantitative trait locus on chromosome 3 (band 3q27) is strongly linked to the phenotype of the metabolic syndrome X.^26,29 Fasting hyperinsulinemia as a precursor to NIDDM has been shown to be genetically linked to a region on chromosome 3 that possesses the same locus encoding glucose transporter 2 (GLUT2).^29,30 Knowing the genetic basis of this disease state provides the opportunity to discover underlying mechanisms and to improve the safety and efficacy of treatments.

There is also evidence that adipose tissue itself may play a role in the cause of syndrome X. Resistin, an adipocyte-secreted protein found on chromosome 19 in humans, is shown to impair glucose tolerance and insulin action in animals whereas antiresistin antibodies improve the level of blood glucose and insulin action.⁴⁰ In addition, glucose transporter 4 (GLUT4) in adipose tissue mediates insulin action in the liver and muscle of mice.^8,41

Rapid eating rate and subsequent substrate overload may be implicated in the cause of syndrome X. A rapid eating rate, whether it is due to hypothalamic-pituitary axis dysfunction, stress, or related genetic susceptibility leads to a cycle resulting in disordered satiety.^16,29 An increased eating rate results in an increased release of gastrointestinal hormones (eg, insulin or gastric inhibitory polypeptide and pancreatic polypeptide) mediated by the cephalic phase (preabsorptive) as well as intestinal, incretin effects (absorptive).²⁹ Gastric inhibitory polypeptide and pancreatic polypeptide stimulate insulin production from the pancreas.^42-46 Rapid absorption of glucose and lipids leads to substrate overload with resultant glucotoxicity and lipotoxicity.^29,47-49 The substrate overload results in a disturbance of pancreatic and liver function and as IR in the metabolic capacity of adipose tissue, muscle, and liver.^29,47,49 Eventually, the pancreas is unable to meet the increasing insulin requirements thus resulting in hyperglycemia, and eventually IGT and NIDDM.^{1,8,10,12,23,29,30,32,39}

Insulin resistance and its resultant hyperinsulinemia are a central feature of obesity and of the comorbidity NIDDM.^{19,26,28,31,32,49} Owing to the association of obesity and IR, weight loss has been a primary therapy in the management of NIDDM. Sjostrom et al^50-52 showed that weight loss sustained over 2 years in obese individuals resulted in reduction of glucose levels. A study with a stable-weight obese control group showed that a 23% maintained weight loss over 2 years resulted in a 32-fold decrease in the incidence of NIDDM.⁵² Doar et al³⁹ studied 118 obese patients with NIDDM who underwent significant weight loss with diet modification. This weight loss resulted in some improvement of diabetes control with declines in plasma glucose levels from 250 to 170 mg/dL (13.9-9.4 mmol/L) and declines in random plasma glucose to levels less than 140 mg/dL (7.8 mmol/L) in 59% of the 70 patients; however, only 14% of the patients had a normal result for the glucose tolerance test. Evidence suggests that control of the blood glucose level in patients with NIDDM prevents the development of proliferative retinopathy.^53,54 As little as 10% loss of the initial body weight can give substantial health benefits, including improved blood glucose level control, decreased blood pressure, decreased cholesterol level, and lower triglyceride levels.^55-57

Morbidly obese individuals who have NIDDM often pose a frustrating problem because they often gain weight in response to insulin and sulfonylurea therapy and lose half the expected weight while making dietary modifications and taking obesity medications.⁵⁵ Conservative therapies available for weight loss include diet therapy, physical activity, behavior modification, and pharmacotherapy.⁵⁸ A low-calorie diet (1000-1500 kcal/d) provides approximately 8% loss of initial body weight at 4 to 6 months.⁵⁹ A very low-calorie diet (<800 kcal/d) results in 15% to 20% loss of initial body weight at 4 months, imparting improved glycemic control.^60-64 However, at 1 year a very low-calorie diet compared with a low-calorie diet provides an equal amount of weight loss, and the significance of improved glycemic control is lost because of poor compliance and subsequent weight regain.^58,60-63 Increased physical activity is more effective for maintenance of weight loss and is not an effective approach for initial weight loss, because only a minimal amount of energy is consumed with the activities that most morbidly obese patients can perform.^58,60,65 Behavioral modification therapy results in approximately a 10% weight loss at 6 months, with regain of one third of the lost weight within 1 year of treatment cessation.^66,67 There are only 2 Food and Drug Administration–approved agents for the long-term treatment of obesity, orlistat and sibutramine hydrochloride, that only achieve about a 10% weight loss.^68-76 Medically treated subjects with diabetes, in contradistinction to those treated surgically, lose approximately half that lost by nondiabetic subjects.⁵⁵ Patients who complete medical weight loss programs lose 10% of their initial body weight only to regain one third within 1 year and nearly all of it within 5 years.⁵⁹ Weight loss by conservative methods has proven exceedingly difficult to maintain over long periods; however, surgical intervention gives substantial weight loss that seems to be equal in diabetic and nondiabetic subjects.^{3,4,18,35,37,52,58,77-80}

Based on the disappointing results of weight loss in morbidly obese patients treated using conservative methods, the National Institutes of Health has endorsed 2 surgical procedures for weight reduction in the morbidly obese, Roux-en-Y gastric bypass (RYGB) and vertical band gastroplasty (VBG).⁵⁸ Surgical procedures for obesity based on gastric restriction, to achieve weight loss by limiting the amount and rate of food ingestion, as well as intestinal bypass, based on weight loss due to malabsorption, first began in the 1950s and have continued to progress. Various operations have included the ileocolic bypass, ileojejunal bypass, gastric plasties, gastric bypasses, gastric bandings, and gastric resections with intestinal bypass. Many of these surgical procedures have been modified or adapted over time. The 4 main operations being performed for morbid obesity are VBG, RYGB, biliopancreatic diversion (BPD), and adjustable silicone gastric banding (ASGB). Approximately 40 000 bariatric surgical procedures are performed each year in the United States.⁸¹ Success has been defined as a greater than 50% of excess body weight loss (EWL).^77,78

The RYGB consists of a small restrictive gastric reservoir associated with early satiety, coupled with a Roux-en-Y loop to provide an element of malabsorption. Excess weight loss greater than or equal to 50% is maintained in approximately 85% of the patients at 2 years and 60% at 5 years.^77,81-83 The average percentage of EWL with the RYGB is 60% to 85%.^{5,17,34,37,79,82-96} Pories et al³ reported that in their series of 608 patients treated with RYGB, patient follow-up was an extraordinary 97%. These patients had a maximum weight loss at 24 months (73% EWL) with only a 10% weight regain after 14 years.⁶ The RYGB provides sustained weight loss with mean weights falling from 138 kg (304 lb) preoperativly to 87 kg (192 lb) by 1 year, and was maintained at 93 kg (205 lb), 94 kg (206.5 lb), and 93 kg (204.7 lb) at 5, 10, and 14 years, respectively.⁵ Owing to the mild malabsorptive component of the operation, vitamin, iron, calcium, and zinc supplements are mandatory; however, these metabolic deficiencies are controllable with supplements.^{36,77,82,91,95}

The VBG developed in 1980 consists of a small gastric reservoir with a restricted outlet. Weight loss stabilizes at 1 year with a mean EWL of approximately 50% to 60%.^{77,79,89,90,92,95,97-103} Over time, there seems to be slow weight regain,^{89,98,104-106} with 50% of the patients maintaining 50% or more EWL at 5 years and 40% EWL at 10 years.^98,104,105 The Mayo Clinic, Rochester, Minn, reported only 26% of the 71 patients maintained an EWL of more than 50% 10 years following VBG.¹⁰⁶ Mason et al¹⁰⁷ reported that the morbidly obese patients operated on with their latest generation of VBG, the VBG5, averaged 200% of ideal body weight preoperatively, 145% at 2 years, 151% at 5 years, and 159% at 10 years. Advantages of the VBG include that it is an operation that is easier to perform than the RYGB and BPD, and it maintains normal gastrointestinal anatomy for radiographic and endoscopic procedures.^77,107,108 The VBG also maintains normal digestion and absorption in the patient to avoid nutritional deficiencies associated with bypass procedures.^107,108 The weight loss, however, is less than that afforded by the RYGB.^{5,46,58,77,79,83,85,87-90,92,95,98,101,102,106,109-113}

The BPD was developed by Scopinaro, and like the RYGB consists of both a restrictive and malabsorptive component. A subtotal gastrectomy is performed, comprising the restrictive portion, and the small bowel is divided 250 cm proximal to the ileocecal valve, with the distal limb being anastomosed to the gastric pouch and the proximal (biliopancreatic conduit) anastomosed to the distal limb 50 cm proximal to the ileocecal valve.⁷⁷ The BPD, thus, confers a large component of malabsorption with only the distal 50 cm of common channel as an area for absorption of nutrients. The BPD is the most effective bariatric procedure to date resulting in a mean EWL of approximately 80% which is maintained.^77,114-116 In a series of 2241 patients operated on over a 21-year period, the BPD provided a mean permanent reduction of about 75% of the initial excess weight.¹¹⁵ However, the BPD is the bariatric operation with the greatest amount of anatomical restructuring and requires additional therapy and monitoring owing to the metabolic adverse effects imparted by its malabsorption.

Gastric banding with an ASGB to create a smaller gastric reservoir was introduced in the 1980s. This procedure is solely restrictive and imparts an EWL greater than or equal to 50% at 5 years in approximately 65% of the patients.⁷⁷ The ASGB is generally felt to provide weight loss comparable to the VBG, with an average EWL of 40% to 60%.^117-119 Kuzmak¹²⁰ reported 57% of the patients experienced a 60% EWL at 3 years. Kirby et al¹²¹ reported a 69.7% EWL at 2 years, but a high rate of reoperations, approaching 50%. This operation has been adapted more frequently to a laproscopic technique than the previously discussed procedures. The laproscopic adjustable silicone gastric banding (LASGB) experience has shown similar results as the open-approach procedure, with a 40% to 70% EWL.^{117,119,122-124} Most favorable studies have been performed in other countries.^{21,117,122,123} The experience in the United States has been less favorable. DeMaria¹⁷ and DeMaria et al¹¹³ reported 41% of the patients in their series required removal of the ASGB, generally secondary to inadequate weight loss. Those with intact bands had a mean weight loss of 38%, and only 11% achieved a weight loss of at least 50% EWL. At the last Food and Drug Administration Advisory Panel Session, only 115 patients had been followed up for longer than 3 years after undergoing LASGB, showing an average EWL of 33%, with one third requiring revision or removal.¹²⁵ This is the least invasive of the bariatric procedures, and most easily reversed; however, there is considerable variation in the results among various surgeons, with the EWL varying from 23% to 70%.^{17,35,77,113,117,119,123-125}

The National Institutes of Health endorses the RYGB and VBG.⁵⁸ Owing to the approximately 10% to 20% more weight loss and the greater longevity of weight loss, the RYGB rather than the VBG has become the preferred procedure of many surgeons, and has become the gold standard of bariatric surgery.^{5,17,46,58,77,79,83,85,87-90,94,95,99,101,102,110,111,113}

A membership survey of the American Society for Bariatric Surgery revealed that RYGB constitutes 70% of the bariatric procedures performed each year; the VBG, ASGB, and LASGB constitute 16%; and the BPD constitutes 10% of the operations performed.¹²⁶ The BPD provides the most effective weight loss to date followed by the RYGB that provides more effective weight loss than the VBG, that, in turn, seems to provide more consistent weight loss than the ASGB and LASGB.

Since the earliest of the surgical procedures for obesity, the resolution of diabetes has been observed in addition to the desired weight loss (Table 1). In 87 subjects followed up for 16 years, who underwent the now-abandoned jejunoileal bypass (JIB), none developed diabetes, compared with 6.6% to 9.8% in a comparable population.⁸⁰ The Swedish Obese Subjects study^35,52 showed that patients treated with a surgical procedure for obesity consisting of RYGB, VBG, or ASGB experienced 47% resolution of diabetes in comparison to 17% in the conservatively treated, matched-control group.The 2-year incidence of NIDDM was reduced by 30-fold.

Pories et al ^3,5 have shown that RYGB corrects abnormal fasting plasma glucose and insulin concentrations, as well as the glycosylated hemoglobin level and the intravenous glucose tolerance test result, with maintenance of euglycemia for as long as 14 years in 91% of the patients. Pories et al³ and Hickey et al⁴ reported a 14-year experience with RYGB with an extraordinary 97% follow-up in which 121 (82.9%) of the 146 patients with NIDDM and 150 (99%) of the 152 patients with IGT returned to euglycemia with complete normalization of their glucose metabolism. In a comparison study of morbidly obese patients undergoing RYGB and morbidly obese controls, Long et al¹³ showed the RYGB imparted a greater than 30-fold decrease in risk of developing NIDDM after weight loss. Cowan and Buffington⁸¹ performed RYGB in 82 patients, of whom 33 (40%) had an elevated fasting glucose level, and 13 (16%) had diabetes. Postoperatively, fasting blood glucose levels were reduced in all patients, and diabetes was resolved in 12 (92.3%) of the 13 cases.

Like the RYGB, the VBG also leads to resolution of NIDDM in most of the treated patients. Sjostrom et al observed 346 patients undergoing a variety of surgical procedures for obesity (VBG, 227; ASGB, 86; and RYGB, 33) and 346 matched obese controls over 8 years for development of NIDDM. The prevalence of diabetes increased from 7.8% to 24.9% in the controls while the prevalence of diabetes in the surgically treated group remained relatively stable at 10.8% and 10.5%, showing that maintained weight reduction strongly counteracted development of NIDDM. Sjostrom et al⁵⁰ reported a 2-year unadjusted NIDDM incidence of 4.7% in controls vs 0.0% in surgically treated patients and an 18.5% vs a 3.6% incidence at 8 years with an odds ratio of the surgically treated group developing NIDDM after 8 years of 0.16. Bourdages et al¹²⁷ reported the resolution of NIDDM in 5 (83%) of 6 subjects following VBG. Yashkov et al¹⁰⁰ reported the Russian experience with VBG in which 3 of 4 patients with NIDDM had complete resolution of diabetes and the fourth patient experienced improvement in glucose metabolism. Similarly, Haciyanli et al¹⁰³ reported resolution of NIDDM in 75% of diabetic patients following VBG. Vertical band gastroplasty imparted a 76% resolution of diabetes over 1 to 4 years compared with an 84% resolution of diabetes at 10 years with the RYGB.⁴ The RYGB seems to not only confer improved weight loss but also improved resolution of NIDDM in comparison with the VBG.^23,46,88,95

Resolution of NIDDM is likewise observed following the BPD. In a series of 1773 patients undergoing BPD with a minimum follow-up of 1 year, the 248 (14%) with preoperative simple hyperglycemia, the 108 (6.1%) with NIDDM manageable with the use of oral hypoglycemic therapy, and the 32 (1.8%) with NIDDM requiring insulin therapy, all experienced normalization of the levels of glucose and insulin and of insulin sensitivity, and became independent from all medications.¹²⁸ The experience of Marceau et al¹²⁹ with the duodenal switch, a variation of the BPD showed 69 of the 72 diabetic patients had resolution of their NIDDM and the need for medications. Marceau et al⁴² found that the blood glucose level decreased even in the absence of NIDDM, suggesting a beneficial effect prior to qualifying with an "abnormal value." Following BPD, the blood glucose level decreased even in euglycemic patients (mean [SD], 0.2 [0.02]-0.3 [0.4] mg/dL [5.2 (0.5)-4.7 (0.7) mmol/L]) (P<.003).

Similar to some reports of the weight loss associated with the LASGB and ASGB, the effects on NIDDM have not been optimal. In a series of European and Australian patients only 40% of the patients with NIDDM had resolution after LASGB.^17,113,125 Similar results were observed in the Swedish Obese Subjects study,³⁵ with resolution of NIDDM in only 9 (47%) of the19 patients treated with ASGB banding.

Weight loss secondary to the RYGB prevents progression to NIDDM.^3,13 In a study of 61 morbidly obese patients without NIDDM, 50 underwent RYGB. At 8 years' follow-up, 6 (55%) of the 11 who did not undergo surgery developed NIDDM, whereas, at 10.2 years' follow-up none of the 50 patient who underwent RYGB developed NIDDM.¹³ Roux-en-Y gastric bypass was the first form of obesity therapy shown to actually reduce the mortality associated with NIDDM. In a comparison study in which 154 patients undergoing RYGB and 78 morbidly obese surgical candidates acted as unoperated on control subjects owing to insurance problems or personal choice, the number of patients requiring medication for treatment of NIDDM increased from 56.4% to 87.5% in the control group, whereas, the surgical patients requiring medical management decreased from 31% to 8.6 %. In addition, mortality was 28% in the control group vs 9% in the surgically treated group. Every year of follow-up, the chance of dying was 4.5% for the control group vs 1% in the RYGB group.^4,130

The exact mechanism for the dramatic effect of surgical procedures for obesity on NIDDM remains unknown. Hypotheses include weight reduction, decreased caloric intake, and in the case of the RYGB and BPD, bypass of the hormonally active foregut.

Initially, the resolution of NIDDM was felt to be secondary to the weight loss imparted by these operations; however, there are observations that suggest that other factors are likely to play a role. The return to euglycemia is rapid and is observed within 10 days postoperatively following RYGB, before any significant weight loss occurs.^3,4,131 Scopinaro et al¹¹⁵ observed that serum glucose level normalized in patients with preoperative NIDDM as early as 1 month postoperatively after BPD, when their excess weight was still more than 80%.In addition, despite significant weight loss, many of these patients remain obese by definition. Hickey et al⁴ measured the levels of fasting plasma insulin, glucose, and leptin; insulin sensitivity; and dietary habits in 6 morbidly obese women after an RYGB who had a stable weight and 6 morbidly obese preoperative control subjects who also had a stable weight. Despite matching these patients for weight, body mass index, percentage of body fat, body fat distribution, metabolic rates, and age, the surgical patients had significantly lower serum leptin levels, fasting glucose levels, fasting insulin levels, increased insulin sensitivity, and decreased food intake. This study demonstrated that despite patients having the same weight loss, age, and fat mass, the surgical patients had improved control of NIDDM over the nonsurgical controls, suggesting that the RYGB's effect on NIDDM is likely secondary to decreased caloric intake or bypass of hormonally active antrum/duodenum/jejunum rather than weight loss alone.⁴

Support of decreased caloric intake as a mechanism for obesity surgery's effect on NIDDM is shown in a study by Pories et al³ consisting of a sham operation. A patient who was taken to the operating room for an RYGB was unable to undergo completion secondary to a full stomach. Postoperatively this patient received the same postoperative diet as those patients who had undergone the RYGB. The same normalization of plasma glucose and insulin levels was observed in this patient as in the patients who underwent RYGB while he remained on the diet.³ This experimental design is similar to pair-feeding experiments in animals and suggests that caloric restriction is sufficient to explain improvement in NIDDM after RYGB during active weight loss.

The observation that the BPD and RYGB result in control of glucose and insulin levels that is superior to the VBG and ASGB suggests that the surgical bypass prcedures may have a role in addition to weight loss in the resolution of NIDDM. Sugerman et al^3,88 observed more profound changes in gut hormone profiles and greater resolution of NIDDM with the RYGB compared with the VBG. Diabetes may be a disease of the foregut in which overstimulation of the foregut by food in susceptible individuals results in increased neuroendocrine signals to the pancreas, resulting in increased insulin secretion by the islet cells. This increase in insulin secretion, in turn, results in IR, and this cycle continues until the pancreas is unable to keep up and diabetes ensues.^4,5,29 If overstimulation of the foregut is the cause of diabetes, then the substantial slowing of food ingestion, by gastric restrictive surgical procedures, such as VBG, ASGB, and LASGB might allow for establishment of substrate-product equilibrium.^4,29 If diabetes is a disease of the foregut, then a bypass of the hormonally active foregut afforded by the RYGB and BPD in addition to their restrictive components, might explain their superiority over the VBG, ASGB, and LASGB in the resolution of NIDDM.^4,45 Serum gastric inhibitory polypeptide, which increases insulin secretion, is reduced in patients following RYGB.⁴⁵ The serum gastric inhibitory polypeptide and pancreatic polypeptide concentration after BPD shows a substantially flat curve in response to the test meal along with normalization of basal and meal-stimulated serum insulin levels.⁴⁴ Enteroglucagon (glucagonlike peptide 1 [GLP-1]), which is secreted in the distal intestine, decreases the rate of glucose production by the liver and increases insulin production, is increased by BPD.⁴⁴ This suggests that the BPD may bypass the enteroinsullar axis, which is felt to be basic to the syndrome X disorder.^49,128 Similar findings were observed in patients 20 years after a JIB, in which glucose and insulin responses were like lean controls, but with large increases of GLP-1 and gastric inhibitory polypeptide.¹³²

One can divide the surgical procedures for obesity into 2 categories depending on their effect on diabetes. The first category (VBG, ASGB, and LASGB) contains those procedures that improve diabetes through a decrease in food intake and body weight with a slowing of gastric emptying. The second category (JIB, RYGB, and BPD) contains procedures that not only decrease food intake and body weight with variable slowing of gastric emptying, but also possess a malabsorptive component that results in increased weight loss and an increase in enteroglucagon–GLP-1 (Table 2).

Enteroglucagon and GLP-1 are probably the same compound and are formed in the upper gastrointestinal tract. There is 1 gene in mammals for the production of 160 amino acid proglucagon, which contains both glucagon and GLP-1. Glucagon (amino acid 33-61) is secreted by the pancreas while enteroglucagon–GLP-1 (amino acid 78-107) is secreted by the terminal ileum in response to a mixed meal.¹³³ Enteroglucagon is elevated after the JIB, and the levels increase through a 20-year postoperative time span.^132,134 Enteroglucagon is increased after the BPD, the JIB, and the RYGB.^44,46 Not only is greater improvement seen in diabetes after RYGB compared with VBG, but there is also a rise in the level of enteroglucagon after the RYGB that is not present after the VBG. The association of higher enteroglucagon levels in subjects with surgical procedures bypassing part of the upper gastrointestinal tract most likely reflects the more rapid presentation of the meal to the distal ileum where enteroglucagon is secreted.¹³³

Enteroglucagon–GLP-1 influences glucose metabolism is several ways. It inhibits pancreatic glucagon secretion, a hormone that raises glucose levels in the blood, and it stimulates insulin secretion, a hormone that lowers the blood glucose level. Insulin resistance is decreased under its influence, gastric emptying is prolonged, and intestinal motility decreased.¹³⁵ Glucose disposal is increased after JIB with a normal oral and intravenous insulin response to glucose therapy.¹³⁶ The overproduction of glucose in type 2 diabetes that contributes to IR is driven by hyperglucagonemia.¹³⁷ Not only does GLP-1 infusion decrease the glucose level in insulin-deprived diabetic dogs,¹³⁸ but GLP-1 has also been shown to be the most potent of the incretins, gut peptides that stimulate insulin release.¹³⁹

Glucagonlike peptide 1 has also been evaluated in human diabetes. Infusion of GLP-1 in patients with poorly controlled diabetes stimulated insulin secretion, decreased the level of glucagon, and slowed gastric emptying.¹⁴⁰ The inhibition of pancreatic glucagon decreased glucose use and decreased insulin requirements with a drop in plasma insulin concentrations.¹⁴¹ Since the surgical procedures that bypass a portion of the upper small bowel increase GLP-1 and this hormone has been shown to improve type 2 diabetes, GLP-1 is the prime candidate to account for the greater improvement of type II diabetes with bypass obesity surgery compared with the purely restrictive procedures.

Ghrelin is a recently discovered peptide hormone produced by the stomach that stimlates growth hormone secretion.¹⁴² Ghrelin has been shown to cause hyperphagia and obesity when given to rats.¹⁴³ Although ghrelin levels are less in the obese subject than in the lean subject, ghrelin levels increase prior to meals and this has been suggested to be the signal for meal initiation.¹⁴⁴ Following gastric bypass, ghrelin levels are reduced and do not rise or fall with meals; a similar situation exists with total parenteral nutrition.¹⁴⁴ Although the lack of the ghrelin signal to initiate meals may be an explanation for the weight loss seen with bypass surgery, ghrelin has not been shown to have an independent effect on diabetes outside of its postulated role in weight loss.

Most individuals who have NIDDM are obese.^1,26 The obese population has proved to be a frustrating entity regarding control of NIDDM. Results of medical weight loss programs, medications, and behavior therapy have proved disappointing. Bariatric surgery has proven a much more successful method for weight loss, and control of diabetes in the obese population. We conclude that the morbidly obese diabetic patient, who is refractory to nonoperative weight loss, is best served by a bariatric surgical procedure. These surgical procedures have been proven safe with reported mortality rates of 0% to 1.5% and early complication rates of 1% to 20%.^{3,22,36,82,83,85-91,94,95,101,114,145} With the weight loss imparted by these operations, like NIDDM, other obesity-associated comorbidities resolve, such as, gastroesophageal reflux disease, obstructive sleep apnea, hypertension, hypercholesterolemia, hypertriglyceridemia, degenerative joint disease, and venous stasis disease. Minimally invasive methods of obesity surgery are rapidly developing, and all of the previously mentioned operations are being performed laproscopically with reported mortality, morbidity, and complications equal to or less than the corresponding open approach procedure.^146,147 With obesity surgical procedures moving toward minimally invasive techniques with reported decreased complications, the body mass index limitations of more than 40, or more than 35 with associated comorbidities, may need to be readjusted to a lower value in the future.^{3,22,36,82,83,85-91,94,95,101,114,145}

Of the bariatric operations performed, the RYGB and BPD offer the greatest weight loss and most dramatic effects on NIDDM. The superior effect of these surgical procedures over the VBG, ASGB, and LASGB is likely secondary to the bypass of the foregut, which results in greater weight loss owing to malabsorption, and results in increased enteroglucagon levels. Defining the contribution of the bypassed foregut will require clinical trials comparing restrictive and malabsorptive procedures while controlling for weight loss and other potential confounding variables.

Corresponding author and reprints: Scott E. Greenway, MD, c/o Stanley Klein, MD, Department of Surgery, Harbor–University of California, Los Angeles Medical Center, 1000 W Carson St, Box 15, Torrance, CA 90509.