Surgical Treatment of Obesity

Surgical treatment of obesity (‘bariatric surgery’; anti-obesity surgery) passes the pragmatic test: it works, most of the time. It is also cost-effective and on a cost per-kg-lost basis is superior to any other method of weight loss for class II and III obesity. Most important: the results are durable, defined as providing maintenance of medical significant weight loss for more than 5 years. Why, then, is surgical treatment not more widely appreciated or performed? A recent survey of attendees of a weight-loss clinic showed that most of the obese patients were willing to take a 6% risk of immediate death if they were guaranteed to reach their desired weight and 25% of the patients were willing to take a 21% risk of dying (1). Yet only a small fraction of eligible patients undergo antiobesity surgery, this most effective treatment with a mortality rate below 1%. Men in particular do not have such surgery though their relative risk of dying from obesity is substantially higher than the risk of women of equal body mass index (BMI) (2) and also higher than their risk of dying from anti-obesity surgery. There are many causes for a relative underutilization of anti-obesity surgery, some of which are frankly irrational. Developments during the past decade effectively address earlier concerns over safety and reliability. This text will describe the fundamental principles of the three most common surgical techniques, will discuss safety and will attempt to define crucial problems influencing the outcome of anti-obesity surgery. Recent trends in this field threaten to repeat mistakes from the 1960s and 1970s.


INTRODUCTION
The exponential increase in obesity worldwide has brought into sharp focus the lack of effective methods for treating or preventing the disease. As the prevalence and severity of obesity have been rising, so has the acceptance of increasingly aggressive remedies. It is in this context that the use of antiobesity (bariatric) surgery has grown. 1 During 2006 there were 386 English-language publications on "bariatric surgery in humans" (excluding plastic surgery) compared with an average of approximately 31 yearly articles since recording started in 1969.
Here we describe the generic types of operations, the mechanisms for weight loss, evolving indications for surgery, outcomes, and aspects of patient selection.
patients are particularly well served by minimally invasive approaches, which avoid the tissue trauma of large transabdominal incisions in open surgery. Notably, newer, even less invasive, endoscopic surgical methods, based on techniques for treating gastroesophageal reflux, 2,3 are currently being evaluated for treating obesity.

Purely restrictive procedures
Purely gastric-restrictive operations, such as adjustable gastric banding ( Figure 1A) and vertical banded gastroplasty, cause weight loss by limiting the capacity of the stomach to accommodate food and by slowing the flow of ingested nutrients.

Gastric restriction combined with diversion
Gastric bypass reduces the stomach to approximately 5% of its normal volume, and ingested food bypasses approximately 95% of the stomach, the entire duodenum, and 20 cm of the proximal jejunum ( Figure 1B). Initially (for the first 6-18 months), the operation relies on gastric restriction, much as does banding.

Diversionary malabsorptive procedures
Malabsorptive procedures bypass or resect the stomach and also bypass long segments of small intestine to reduce the area of mucosa available for nutrient absorption. Biliopancreatic diversion, with or without a pylorus-sparing 'duodenal switch', ( Figure 1C) promotes selective malabsorption of fat; 4 weight loss is effective, but this procedure causes more complications than gastric bypass, such as protein malnutrition, diarrhea, and deficiencies of various vitamins. 5

Regulatory metabolic operations
Duodenal-jejunal bypass 6 and ileal interposition 7 are investigational operations, as is truncal vagotomy, 8,9 all with minimal malabsorptive yet significant metabolic effects mediated via neuroregulatory brain-gut peptides. These peptides employ mechanisms believed to cause weight loss through appetitive and metabolic effects present in diversionary operations such as gastric bypass.

MECHANISMS FOR wEIGHT LOSS AFTER BARIATRIC SURGERY
Purely gastric-restrictive procedures cause gastric and/or esophageal distension, eliciting early satiety, nausea and discomfort or even vomiting when pouch capacity is exceeded. 10 Esophageal distension activates afferent nerves stimulated by mechanoreceptors and neuropeptides such as Substance P. 11,12 Positron emission tomography studies of esophageal distension have identified ncpendmet_2006_035f1.eps  The proximal divided jejunum is reattached to the jejunum 75-150 cm below the gastrojejunal anastomosis creating a Roux-en-Y limb. (C) Biliopancreatic diversion, with or without a pylorus-sparing 'duodenal switch' causes malabsorption as pancreatic and biliary secretions are diverted to the distal small intestine approximately 50 cm from the ileocecal valve. Absorption is thus limited to the distal ileum. A 'sleeve' gastrectomy is depicted.
brain loci in proximity to those associated with appetitive signaling 13,14 and with visceral pain; 15 therefore, esophageal signals might participate in regulating (inhibiting) food intake via satiety and/or nimiety signals. Distension of the cardia or the upper stomach also increases activity in the areas involved in appetite regulation. 16 After adjustable gastric banding, weight-stable patients with optimal band tightening experience greater post-prandial satiety than those with looser bands. 17 Interestingly, patients with optimal adjustment experienced less hunger after an overnight fast, implying that the band might affect satiety through unknown mechanisms perhaps resulting from activation of mechanoreceptors under the band. 17 For the first 12-24 months after gastric bypass, the restrictive component decreases as pouch and stoma stretch. 18 Thereafter, mechanisms emanating from the diversionary component of gastric bypass prevent weight regain: the pyloric 'meter' or 'brake' is absent and rapid transit via the gastrojejunostomy is allowed. Maldigestion can also result from the absence of acid and pepsin and the grinding-mixing forces of the stomach. The unphysiological presence of undigested food and liquids rapidly 'dumped' into the small bowel causes nimiety via mechanoreceptors and possibly satiety via chemoreceptors. 19 Gastric bypass causes profound changes in plasma gut peptide levels. Ghrelin, secreted mainly from the stomach, has been shown to stimulate food intake. 20 A majority of studies has shown a decrease in plasma ghrelin levels after gastric bypass, 21 which is consistent with bypass of the stomach and a rapid stimulation of the small intestine. Peptide YY decreased food intake in obese humans when it was given in a truncated formulation (peptide YY 3-36 ); 22 this finding has renewed interest in studying peptide YY. Glucagon-like peptide 1 (GLP-1) also decreases food intake and appetite in obese subjects, as well as stimulating insulin release and inhibiting glucagon secretion. 23 This may be one mechanism whereby such methods as duodenal switch and ileal interposition affect glucose disposal and energy balance.
Both peptide YY 24 and GLP-1 25 are released from cells in the distal gut, and levels increase postprandially after gastric bypass, consistent with the rapid stimulation of the lower gut. These increases in appetitive gut peptides may decrease food intake and contribute to the weight loss after gastric bypass. 26 Interestingly, biliopancreatic diversion exhibits increased levels of ghrelin 27 whereas GLP-1 is increased more after biliopancreatic diversion than after gastric bypass. This finding might explain why biliopancreatic diversion tends to ameliorate noninsulin-dependent diabetes mellitus more than does gastric bypass. 28 New data suggest that ghrelin might influence hedonic feeding. Ghrelin receptors are found in the ventral tegmental area, thought to be a node in the mesolimbic reward pathway with dopaminergic projections to various areas of the brain. Indeed, injection of ghrelin into the ventral tegmental area increases food intake in rats. 29 Furthermore, high central leptin concentrations downregulate hypothalamic leptin receptors and impair leptin signaling; 30 this downregulation modulates the interaction between leptin and other regulators of appetite (e.g. ghrelin) in the obese state. Decreases in both leptin and ghrelin concentrations after gastric bypass might sensitize leptin pathways and reduce the reward of hedonic eating.
Cholecystokinin, a foregut peptide identified in the vagus nerve, and GLP-1 given in combination did not exhibit additive effects on appetite in humans, 31 in contrast to peptide YY 3-36 and GLP-1. 32 Most current diversionary operations reduce and/or exclude the stomach and pylorus, the site of cholecystokinin and other receptors. Some exclude varying lengths and segments of small bowel, creating 'short-cuts' for unabsorbed nutrients, which, therefore, overstimulate or understimulate the duodenum, jejunum or terminal ileum, each with its own complement of peptides ( Figure 2).

INDICATIONS FOR SURGERY
Indications for surgery evolved in the United States during the 1980s and were formulated in the 1991 National Institutes of Health Consensus Development Conference on gastrointestinal surgery for severe obesity. 33 The weight criterion for undergoing surgery, expressed in terms of BMI, was broadened from a BMI of at least 40 kg/m 2 to include a BMI of at least 35 kg/m 2 in the presence of manifest comorbidity. Individuals with a BMI of at least 35 kg/m 2 sustained for more than 1 year exhibit comorbidity if thoroughly evaluated. Dyslipidemia, impaired glucose metabolism, fatty liver, and poor quality of life appear very early in the natural history of obesity; these conditions are increasingly appearing in children. Age criteria for candidacy for bariatric surgery have widened in parallel with the growing prevalence of severe obesity, the most rapidly increasing class of obesity. Longer duration of exposure to obesity decreases the likelihood of sustainable remission or cure, 34 arguing for early intervention. "Early", in this context, may be defined as shortly after attaining a critical BMI, in analogy with early treatment of diabetes. 35 With the availability of laparoscopic approaches, older patients are increasingly having bariatric surgery. Just as age criteria have widened during recent years, so have the weight criteria. Several studies have used lower BMI limits (BMI 30-35 kg/m 2 ), 36,37 a trend that is growing.

REOPERATIONS
Indications for reoperative bariatric surgery are not defined, and criteria for choice of secondary or 'rescue' procedures have not been developed. In our opinion 'failure' is best defined as recurrence of comorbidity accompanying postoperative weight gain. This has been reported to occur in about 25% of patients, depending on procedure, although data on reoperation rates are frequently absent. It is absolutely essential that the team taking care of a patient with weight regain ascertain probable causes, rather than simply ascribing it to surgical-technical problems.

Weight loss
Weight loss per se is likely to be the least important measure of bariatric surgical outcomes, although it is highly correlated with recognized risk factors and with patient satisfaction. Weight loss is conventionally expressed as percent excess weight loss (%EWL). "Excess" refers to the weight that exceeds actuarial standards of weight adjusted for height ('desirable' or 'ideal' weight) corresponding to minimal mortality. In a recent meta-analysis, mean EWL (and 95% CI) after biliopancreatic diversion was 70 These differences might result from eating behavior and marketing of food, as well as the severity of obesity, in the different populations.

Survival advantage
Reducing mortality is the most important goal of bariatric surgery. In 2006, the Swedish Obese Subjects study, 44 the first prospective controlled study designed specifically to assess mortality, reported significantly lower mortality over 10 years in operated patients than in those treated conventionally. 45 Other studies, using different types of databases, for example national death registries, case-control series and public health agencies, also suggest that mortality is reduced after surgery. [46][47][48][49][50] In Washington State, mortality 15 years after bariatric surgery was 11.3% compared with 16.3% in nonoperated obese subjects. 46 A common finding in all these studies is a relatively low perioperative mortality rate. Surgical method (e.g. restrictive in the Swedish Obese Subjects study, and vertical banded gastroplasty in Iowa 49 versus gastric bypass in Washington State 46 ) and approach (open or laparoscopic) are important when evaluating these data, considering that substantial advances have been made during the last 10 years.

Comorbidity reduction
Most obesity comorbidity is ameliorated postoperatively depending on the type of surgery performed. 51 Purely gastric-restrictive procedures do not have the durability of comorbidity reduction seen after diversionary operations. 34,52,53 As an example, dyslipidemia is best corrected after biliopancreatic diversion, followed in effectiveness by gastric bypass and adjustable gastric banding.
In the Swedish Obese Subjects study, the incidence of hypertriglyceridemia was reduced from 27% to 17% after 10 years. 44 The effect of bariatric surgery on hypertension is complex. Blood pressure is often lower up to 5 years, but in some series hypertension returns after 10 years, probably from weight regain after restrictive operations. 44 Non-insulin-dependent diabetes mellitus responds dramatically. In the Swedish Obese Subjects study the incidence of noninsulin-dependent diabetes mellitus was reduced from 24% to 7% at 10-year follow-up in the control group compared with the surgical group. 44 Other series demonstrate improvements of up to 85% among patients with non-insulin-dependent diabetes mellitus. 38,50,51 Table 1 summarizes changes in four major comorbidities over time according to type of operation. 38 The rapid increase in childhood obesity demonstrates the importance of the intrauterine environment as a precursor of offspring obesity. Pregnancies in which the mother is obese are hazardous for mother, fetus and child, increasing the prevalence of childhood, and ultimately adult, obesity. Several studies show that prior bariatric surgery ameliorates gestation and delivery and fetal and neonatal health (reviewed by Kral 54 ). A recent study of 113 offspring born after maternal bariatric surgery who were followed for 2-18 years revealed a reduction of obesity to local population standards. 55

Complications and side-effects
Perioperative mortality Severe complications have decreased over time; deaths generally occur in less than 1% of patients. The Swedish Obese Subjects study reported 5 deaths in 2,000 patients, corresponding to a mortality rate of 0.2%; 44 however, single series and population studies have reported significantly higher mortality rates approaching 2% [56][57][58] with the highest figures pertaining to complex procedures.
Low perioperative mortality rates affect the interpretation of the role of weight reduction per se on mortality, now demonstrated to be reduced after bariatric surgery. 47-50 Operative mortality is strongly related to surgical volume 59 and has decreased dramatically with the wide adoption of laparoscopic procedures during the last 6 years. An excellent systematic review comparing shortterm mortality among 5,780 patients with laparoscopic adjustable gastric banding (0.05%), 2,858 vertical banded gastroplasty (0.31%) and 9,258 gastric bypass (0.50%) patients demonstrated significantly lower mortality rates with laparoscopic adjustable gastric banding. Vertical banded gastroplasty and gastric bypass groups were not, however, analyzed according to approach (open versus laparoscopic), and percentage of follow-up was not tabulated. 60 The most common causes of death are pulmonary embolism, (unrecognized) intra-abdominal leaks and myocardial infarction, although the incidence of fatal myocardial infarction is lower after surgical than medical treatment of obesity. 56

Surgical complications
Some complications are procedure-specific: adjustable gastric banding does not entail entering the gastrointestinal tract and does not affect bowel function. Band erosion can cause pain and ulcers. Pouch enlargement and band slippage can result in acid reflux, and slippage can also contribute to vomiting. Acid reflux is uncommon after gastric bypass because most of the stomach and acidproducing cells are excluded, but it can occur after biliopancreatic diversion, owing to a larger proximal stomach and diversion of bile. Similarly, surgical approach (open versus laparoscopic) affects the types of complications. A recent review of gastric bypass 61 found statistically significant differences in numerous complications, particularly wound problems in 6.6% of open procedures compared with 3.0% in laparoscopic cases. Incisional hernias are more common after open surgery, whereas other complications are more common after laparoscopic surgery: small-bowel obstruction (3.1% versus 2.1%), anastomotic stenosis (4.7% versus 0.7%) and gastrointestinal bleeding (1.9% versus 0.6%). There were no differences in the number of leaks (1.2%), pulmonary emboli (<1%) or pneumonia (0.1-0.3%). The cause of the higher incidence of small-bowel obstruction early after laparoscopic surgery is probably internal hernia.

Deficiencies
The most important long-term adverse effects are deficiencies of vitamins and minerals, especially after diversionary malabsorptive operations.
Gastric-restrictive operations have mostly caused iron deficiency in menstruating women, owing to reduced intake of meat, and excessive vomiting can cause thiamine deficiency resulting in neuropathy. Gastric bypass and biliopancreatic diversion are associated with deficiencies of Vitamin B 12 , 62 folate, calcium and vitamin D. 63,64 In the long term, peripheral neuropathy can occur (up to 16% in some series). Patients are urged to have blood nutrient levels monitored and to take oral vitamin supplementation. The varying efficacy of oral vitamin B 12 supplementation is partly related to patient adherence, although some studies have demonstrated decreased uptake after surgery. 62 Bone loss tends to occur during the first year after gastric bypass surgery and then stabilizes with unchanged vitamin D levels. 64 Weight-loss failure The most problematic long-term complication of bariatric surgery is poor or inadequate weight loss. Reoperations are more difficult than primary procedures and have higher perioperative complication rates. 65 Failed gastric-restrictive procedures are, nevertheless, best handled by conversion to a diversionary operation. 66 Long-term failure of diversionary operations requires adjustment of intestinal limb lengths to create more malabsorption, with the risk of creating frank malnutrition. Should this occur, one or more trials of parenteral nutrition support are sufficient to enable non-bypassed intestine to recover its compensatory capacity, in our experience decreasing the risk of recurrent malnutrition.

Risk reduction
Handling patients at increased risk, such as those who are very obese or those with severe comorbidities, is a continuing problem. 67 The cost-effectiveness of a staged surgical strategy for operative risk-reduction has not been compared with nonoperative methods, such as conventional outpatient diet, exercise, behavior modification, intragastric balloon, intermaxillary fixation ('jaw wiring'), pharmacotherapy or monitored in-patient calorie control using very low calorie diets. On a cost per unit weight loss per unit time basis, surgical methods are nevertheless superior both over the short and the long term.
In our opinion, older, heavier patients, with more serious comorbidities, might benefit from a laparoscopically placed adjustable gastric band, the simplest, quickest and least complication-prone operation, until medically significant weight loss KRAL AND NÄSLUND AUGUST 2007 vOL 3 NO 8 www.nature.com/clinicalpractice/endmet has occurred. Thereafter, a more complex and durable, diversionary, malabsorptive procedure should be considered in order to increase the likelihood of sustaining the weight loss. Gagner and colleagues suggested using sleeve gastrectomy in a two-staged approach in this setting, 68 although any resective operation involves more risk of complications than simple band placement.

Cost-benefit analyses in bariatric surgery
As with all therapy, there has been a long-standing interest in cost efficiency analyses for bariatric surgery. 69,70 There are no published cost-benefit comparisons between, for example, adjustable gastric banding and other simple operations, accounting for cost of device and postoperative visits for band adjustment. Careful review of economic, health, quality of life and survival benefits of bariatric surgery would have to be balanced against the costs of side-effects, complications and death, taking into consideration geographic, socioeconomic and setting-specific costs that can vary between countries with different accounting practices and methods for delivery of care. Although the Swedish Obese Subjects study might be well suited to perform a cost-benefit analysis (and has, indeed, studied sick leave, disability and medication costs), given the uniformity of delivery of care in the tightly regulated socialized system of a relatively small country with excellent demographic resources, it is difficult to extrapolate the findings to other countries or settings. There are, furthermore, concerns about the relevance of the treatment arm of this study because it is mainly based on an obsolete operation. 71

Standards for bariatric surgery
As with all surgery there is a learning curve, particularly for laparoscopic bariatric surgery. Data suggest that competency requires the performance of 75-100 cases. 72 High-volume (>100 cases per year) centers have better results than low-volume centers (<50 cases per year). 59 This is true for length of stay, overall complications, costs and mortality rate and is especially true for older patients (>60 years). Owing to great variations in the quality of care in the United States, several initiatives have been taken to define minimal performance parameters for hospitals to qualify for insurance reimbursement. A 'center of excellence' designation requires the meeting of standards for volume, equipment, human resources, reporting and complication rates. Equally important as the number of cases performed is the availability of a dedicated, competent team to assess, educate and take care of patients before and after surgery.

PATIENT SELECTION Outcome predictors
There are several different classes of outcomes after bariatric surgery: perioperative and postoperative complications and adverse effects, comorbidity reduction, weight-loss maintenance, improved economy, psychosocial improvement, and objective and subjective quality of life changes. They are interrelated and associated with weight loss, which alone is an inadequate outcome measure. Just as quality of life is multidimensional, 73 so is the definition of 'successful' outcome, 74 making it difficult to identify specific predictors useful for patient selection. 75 The largest source of variance in studies of outcome and its predictors is duration of post-treatment observation, including percentage of patients lost to follow-up, sadly neglected in the preponderance of literature on weight loss regardless of method. The majority of publications on bariatric surgery, whether observational or mechanistic, include patients during various phases of weight change, which might explain conflicting data especially during rapid weight loss.
Classic predictors of outcome such as age, weight and sex have been studied with regard to bariatric surgery. One systematic review of age effects demonstrated inconsistent results: six studies found younger patients lost more weight, whereas four did not. Heavier patients lost less weight. Male sex and high age and preoperative weight are risk factors for increased postoperative complications. 76 Five postoperative categories associated with poor long-term outcomes of bariatric surgery have been identified: poor patient knowledge; psychosocial maladaptation; anatomic complications; gastrointestinal pathophysiology; and weight-related symptomatology. 77 There is considerable interaction and overlap among these categories, but they are equally important to recognize and address in patient selection. 78 African-American patients seem to experience less weight loss after bariatric surgery than white patients; 79 however, the improvement in comorbidity was similar in one study. 80

Eating behavior
Eating behavior has been relatively neglected although bariatric surgery to a great extent is 'behavioral surgery'. Preoperative education about how to eat and what to do if vomiting occurs, as it can, after adjustable gastric banding or early after gastric bypass, is essential. 78 Studies on the importance of binge eating disorder (BED) are equivocal. Two retrospective studies suggested that BED is a negative predictor, 81,82 whereas two prospective studies did not demonstrate any relationship between BED and postoperative outcome. 83,84 One retrospective study suggested that biliopancreatic diversion might correct BED. 85 Mutations in the melanocortin 4 receptor gene have been implicated in BED, 86 although the findings are disputed. 87 Outcomes 3 years after adjustable gastric banding in 300 patients nevertheless clearly demonstrated poorer weight loss, less amelioration of comorbidity and more complications in patients who have mutations in their melanocortin 4 receptor than in patients who have BED but no known mutations in the receptor. 88 A small subset of patients with polymorphisms who were converted to gastric bypass seemed to have a poorer weight loss than those without.
It is too early to determine an optimal strategy that is based on eating behavior or BED with or without gene abnormalities. Operative treatment is nevertheless more effective than any other method in this context: BED is not a contraindication for bariatric surgery. Heavier patients (BMI >50 kg/m 2 ) have a more serious eating disorder, accounting for variations in published results.

Expectations
Confusion among surgeons 89 and other healthcare professionals 90 about goal weights pertains to determining 'sufficient' or 'optimal' weight goals for patients who are, or have been, severely obese. How much weight loss is enough?
Unfortunately, there is a widespread misconception that, after surgery, a BMI of greater than 30 kg/m 2 represents an unhealthy body weight in patients who have lost weight-that the inability to bring patients under this arbitrary threshold constitutes failure of surgery. There is sufficient, although sparse, actuarial evidence to conclude that sustained voluntary weight loss is associated with lower morbidity and mortality than the morbidity or mortality of the general population at the same level of BMI. 91 Patients with a BMI above 30 kg/m 2 and who have lost weight after surgery after previously having had BMIs of at least 35-40 kg/m 2 have lower risk of death or disease than those who never lost weight, contrary to the case with 'failure' of nonsurgical weight-loss treatment.
Step-care strategy In 1988 we introduced the concept of a staged surgical approach based on experience with 'rescue' operations after failed gastric-restrictive surgery. 92 Because of the introduction of the less invasive laparoscopic approaches and development of the adjustable gastric band, in 1991 we proposed the use of a staged approach as a method for patient selection, 93 given the high failure rate of gastric restriction and the inability to predict which patients might benefit most from such operations. This approach allows the more complex, complication-prone operations to be reserved for patients with failure of, or unable to tolerate, gastric restriction, and was subsequently embraced by Mason, 94 the inventor of gastric bypass and vertical banded gastroplasty, and incorporated into a systematic treatment algorithm for rescuing patients with failure of laparoscopic adjustable banding. 95

CONCLUSIONS
Although techniques can be developed to outsmart energy-saving genes and polymorphisms, it is not likely that the second law of thermodynamics will be repealed in the foreseeable future. After obesity surgery there is a clear relationship between the magnitude of weight loss and the durability and extent of improvement of comorbidity, modified by the duration of disease exposure before surgery. For now, bariatric surgery has an important role to play and should be provided early in the disease process, when other methods have failed.

■
Bariatric surgery is safe; complications are fewer after restrictive procedures but weight loss and comorbidity reduction are greater after diversionary procedures ■ Most obesity comorbidity is durably (>10 years) ameliorated after surgery and mortality is less than after nonsurgical care ■ After bariatric surgery most patients do not reach 'normal' weight; however, the weight loss induced by surgery is sufficient to improve morbidity and mortality ■ A dedicated, comprehensive team is needed to assess, educate and manage the patient before and after surgery