Bariatric surgery: the challenges with candidate selection, individualizing treatment and clinical outcomes

Patient selection criteria for bariatric surgery include body mass index (BMI), the presence of co-morbidities and a history of prior weight loss attempts. National Institute of Clinical Excellence (NICE) and National Institutes of Health (NIH) guidelines state that bariatric surgery should be offered to patients with a BMI of 35 to 40 kg/m² who have obesity related conditions such as diabetes mellitus or obstructive sleep apnea, or in those with a BMI of 40 kg/m² or greater regardless of weight related co-morbidities [19, 20]. Bariatric surgery for individuals with a BMI less than 35 kg/m² with obesity related co-morbidities is under investigation but is not currently recommended [21].

If a candidate meets the criteria for surgery, then a multi-disciplinary team assessment is made as to the suitability of the candidate. This is a complex process involving psychological, surgical, dietetic and medical review. The individual must be physically and psychologically fit to proceed to surgery. Expectations must be managed and a determination must be made as to the individuals' ability to comply with post-operative care. The decision to operate will take into account the benefits the candidate is likely to gain, and the risks peri-operatively and post-operatively.

This is an individualized assessment, and the role of the psychologist and/or psychiatrist should be central. Some reports suggest an increased risk of suicide after bariatric surgery, although the etiology remains unclear [6]. Major failures of bariatric surgery are due to poor psychological adaptation, especially if the patient's expectations were not adequately managed. All candidates should be given the correct and realistic information on what the procedure can achieve. If this is addressed, then the risk of surgical failure can often be mitigated [22, 23]. This personalized assessment is a vital part of the pre-operative assessment.

For each patient the benefits of the procedure should outweigh the operative risk. In general, obese patients have an increased prevalence of cardiopulmonary disease that may be undiagnosed pre-operatively [24]. An individualized pre-operative assessment should be completed by a multi-disciplinary team [25]. Pre-operative investigations should focus on screening for cardiac arrhythmia, prolonged QT syndrome, and cardiomyopathy [25]. Almost 70% of individuals awaiting bariatric surgery can be diagnosed with obstructive sleep apnea, with over 40% meeting the criteria for severe disease [26, 27]. However, this is not associated with a greater rate of peri-operative complications [28]. Male gender, age older than 50 years, congestive heart failure, peripheral vascular disease and renal impairment are associated with greater mortality [29].

Once the candidate has been selected, then the appropriate procedure must be chosen. Unfortunately, there is no evidence based medical approach for procedure selection, and this remains one of the most frustrating shortfalls in bariatric practice for clinicians and patients. Clinicians take a pragmatic approach to the choice of procedure, and the decision is determined by the individuals' clinical phenotype, the aims of therapy, and peri-operative risk.

Clinicians and patients can often be disappointed when surgical outcomes are not as impressive as may have been hoped and look for objective evidence that can allow them to predict outcomes. The most damaging outcome of surgical procedures that do not perform as expected is to blame the recipient for not performing well enough with regard to diet and exercise, when in fact, the failure is almost always rooted in biology. Weight loss prediction is one of the common aims of these predictive models. Pre-operative weight loss can predict post-operative weight loss, putatively as a marker of 'intrinsic motivation' [30, 31]. These data are contentious and not without bias as many of them are retrospective uncontrolled studies [32]. The available prospective data are of short duration [31, 32]. A systematic review shows that 50% of the published results find that pre-operative weight loss has an association with weight change post-operatively, and the remaining 50% refute these findings [32]. In the context of this evidence, pre-operative weight loss cannot be relied on to predict surgical outcomes at this time.

Genetic disorders such as melanocortin-4 receptor (MC4R) deficiency may have an inherent physiological role in these mixed results, as animals with MC4R deficiency have a resistance to weight loss after bariatric surgery [33]. There is interest in these genotypes, as identifying them may aid the prediction of outcomes. However, identified phenotypes such as MC4R are not common in the obese population, with heterogeneous mutations identified in less than 3% of European and North American obese cohorts [34]. Culprit genes that have been associated with obesity, with approximately 20 implicated to date, are still only found in 5% of obese people [35].

Investigation of genetic factors that may predict individual responses to bariatric surgery is ongoing and controversial [35‐37]. While certain genotypes are associated with improved outcomes after bariatric surgery, they are not procedure specific and, therefore, while potentially aiding prediction of weight loss post-operatively, will not aid procedure selection [35‐37]. It is also well recognized that the genetic influence on obesity may be much more complex than we currently understand, with the inevitable influence of environment making the situation less clear. For now, study of identified genotypes, with a correlation between genotype and treatment outcomes, may answer questions on the clinical utility and predictive ability of genotyping in bariatric surgery [33].

To date, any potential genetic markers or biomarkers of weight loss following bariatric surgery have been limited by clinical utility, and sensitivity and specificity [38, 39]. Some data identifying potential markers of weight loss are inadequately controlled and unmatched [38]. The positive findings are in the context of complex interactions, without clinically usable tests that could be applied in daily practice [39].

While there is some potential in this field, usable techniques are still many years away [40].

The factors most consistently negatively associated with post-operative weight loss include higher BMI levels and personality disorders [32]. Given the impact of psychological markers on outcomes, techniques such the artificial neural network, which can incorporate psychological and biological measurements, have been tested to predict surgical outcomes [41]. Such techniques rely on established data that have been shown to effect outcomes, but may incorrectly predict response in as many as 30% of bariatric surgery recipients [41]. These models are multi-factorial, prone to bias and socio-cultural differences [41]. They are also time-consuming and expensive [41]. As the techniques are refined, we may develop a useful model that could be employed in routine practice, although we have not arrived at that point as of yet.

While weight loss remains difficult to predict, there are increasing amounts of data on prediction of diabetes remission. Markers of insulin secretion, such as C-peptide may aid pre-operative prediction of diabetes remission [42]. These results report an increased rate of diabetes remission with higher C-peptide levels [42, 43]. The highest cut-off can predict diabetes remission with a specificity of approximately 90% [42]. Shorter duration of diabetes, lower glycosylated hemoglobin (HbA1c) levels and insulin independence are also associated with a higher post-operative remission rate [43, 44]. These data illustrate a role for C-peptide to be used in conjunction with clinical data to predict diabetes remission. If a validated, sensitive and specific model were developed then it may aid procedure selection. However, the models currently studied can provide great specificity, but only at the cost of sensitivity [45].

The use of incretin and bile acid profiles has been investigated for use in predicting weight loss and metabolic outcomes following bariatric surgery [46, 47]. The findings suggest that the restoration of peptide YY (PYY) and glucagon-like-peptide- 1 (GLP-1) secretion following RYGB contribute to satiety and weight loss [46, 48]. Bile acids also have a role in this process, and the mechanisms underlying this are currently being elucidated [47]. Increased bile acid delivery to the terminal ileum can improve satiety and enhance weight loss [47]. However, these changes occur after surgery, and there is no current evidence that would allow them to be used to select candidates pre-procedure.

Similarly, while the restoration of the PYY and GLP-1 response is associated with satiety and weight loss in RYGB, as opposed to AGB, there are no strong data on differences within this group that allow us to predict the degree of weight loss following RYGB based on the incretin or bile acid response [48, 49]. There are data demonstrating a progressive rise in PYY and GLP-1 following RYGB that is associated with increased satiety but without noted differences within the group [49]. The relationship between absolute incretin or bile acid levels, or trends, and weight loss remains to be determined. At this time, the restoration of incretin secretion and increased serum levels of bile acid are associated with enhanced satiety and weight loss, but they cannot be used to predict weight loss [46‐49].

For now, procedure selection is best informed by the candidates' objectives and by how they want to live their lives after surgery. As a primary aim of surgery, the efficacy of weight loss associated with each procedure must be considered. RYGB results in greater weight loss than AGB, although good quality post-operative care can improve the weight loss following AGB, with results comparable to RYGB [50‐52]. AGB is associated with a lower rate of immediate post-operative complications but also a higher rate of re-operation for insufficient weight loss [50‐52]. The associated mortality rate is higher with RYGB than with AGB but still less than 0.3% [50, 51].

Weight loss is comparable between RYGB and SG in the short term [53, 54]. Some studies suggest that more patients will regain weight in the medium-term after SG [55]. BPD/BPD-DS results in greater weight loss, but with higher complication rates, than RYGB [56, 57]. Therefore, the greatest weight loss would likely be achieved with BPD/BPD-DS. However, this is not generally agreed. BPD/BPD-DS may not be suitable for high-risk operative candidates and some randomized controlled trials have shown no additional benefit of the extra weight loss above RYGB [58].

There are variances in outcome between national health systems, with AGB considered a superior bariatric procedure in systems where there is an excellent post-operative care pathway [52]. This implies that AGB can have comparable results providing that the post-operative care is planned and provided by experienced clinicians.

With regard to diabetes remission or treatment, RYGB offers a greater rate of remission than AGB [50, 51]. SG has a remission rate comparable to RYGB in the short-term, but a higher rate of relapse in the medium-term [59]. BPD and BPD-DS may offer a higher rate of diabetes remission than RYGB or AGB [60‐62]. While AGB is the least effective in inducing diabetes remission, it can offer substantial improvements in diabetes control, which are greater than those offered by medical therapy in obese cohorts [63, 64].

Other conditions can influence a decision on bariatric surgery. Respiratory disease may improve more significantly with greater weight loss [65]. Therefore, those with obstructive sleep apnea (OSA) or asthma could theoretically be considered for more consistently efficacious procedures such as BPD/BPD-DS or RYGB. Conversely, SG and AGB are associated with deteriorations in gastro-esophageal reflux disease (GERD) and, therefore, should be avoided in this cohort [55, 66]. In GERD, RYGB is increasingly considered as the treatment of choice as it can remediate the GERD due to the reduction in the stomach pouch and prevention of esophageal reflux [67].

In summary, candidate selection and preparation is key to achieving good surgical outcomes. Each procedure should be considered for each individual, and the data to date do not support the application of a generic selection based on body weight, diabetes or other co-morbidities. The choice of procedure is a complex process with the patient and their interests at its core. The surgeon's experience to deal with the inevitable complications of each procedure and to manage long-term follow-up care remain dominant considerations.

In those with diabetes, BPD/BPD-DS offers the highest rate of remission, but also the highest complication rates. RYGB and SG are comparably efficacious in treating diabetes in the short-term, but questions remain regarding the medium to long-term. It should be noted that the volume of data for RYGB is greater than that for SG and BPD/BPD-DS. For those with GERD, RYGB should be the treatment of choice. SG and AGB should be avoided.

AGB can also lead to weight loss and diabetes remission and can offer greater control than medical therapy, even if remission is not achieved. It should be noted that the choice of AGB should take into account the availability of good quality post-operative care. AGB may be suitable for those who wish to lose weight and improve diabetes control, but not remission, and are at higher surgical risk.

The outcomes following bariatric surgery vary between procedures, and predicting individual outcomes following bariatric surgery is difficult, for the reasons outlined in the previous sub-sections. We will review each outcome individually and compare between modalities below.

Surgical complications can be defined as early or late, depending on if they occur within the first thirty post-operative days or afterwards. There is a wide range in the reported complication rates. The benchmark for bariatric centers of excellence is the Longitudinal Assessment of Bariatric Surgery consortium [114]. Mortality rates after bariatric surgery are low with a mortality rate after RYGB of 0.3% [114]. Pulmonary and venous thrombo-embolism are early complications and occur in less than 0.5% of bariatric surgery recipients [115]. Other complications can be specific to the modality and include the following.

Deficiencies of iron, vitamin B12, folate, and fat-soluble vitamins can occur after bariatric surgery and are best described in RYGB, BPD and BPD-DS [140]. Vitamin D deficiency can persist despite prescribed replacement in BPD and may tend towards secondary hyperparathyroidism [140]. The risk of nutritional deficiencies depends on postoperative weight loss, the surgical procedure performed and patient compliance with follow up [140, 141].

Vomiting is frequent after bariatric surgery but must always be considered to be pathological until proven otherwise after RYGB. An examination and appropriate radiological studies to screen for stricture, stoma stenosis or herniation must be completed. If no pathological cause is found then treatment should be conservative with replacement of fluid and electrolytes [141]. Often, vomiting can be the result of overeating or rapid eating. The patient should be re-educated on eating habits and kept under review.

Diarrhea is reported in up to 40% following bariatric surgery [142]. More than 30% of bariatric surgical recipients report worsening bowel function in the post-operative period and some develop fecal incontinence [143]. The etiology of this is unclear and treatment is based on appropriate dietary modification and anti-diarrheal pharmacotherapy.

There is a variable incidence of the dumping syndrome after bariatric surgery, particularly in RYGB [144]. Dumping syndrome remains a 'waste-basket diagnosis', with the clinical presentation generally considered to include early abdominal pain, diarrhea, nausea, bloating, fatigue, facial flushing, palpitations, hypotension and syncope after high glycemic index meals. These symptoms usually occur within an hour of eating. Similar symptoms that occur two or three hours after a meal include perspiration, palpitations, hunger, tremor, agitation, and syncope [144]. These have been blamed on hypoglycemia and GLP-1, although a definitive etiology remains to be established [145, 146].

The treatment of early dumping syndrome is usually straightforward dietary modification, with small regular meals containing protein and carbohydrate with a very low glycemic index. Treatment of the symptoms that occur within two or three hours of a meal also rely on the same dietary modifications, with the added aim of 2 or 3 kg of weight gain which often abolishes the symptoms secondary to the small amount of increased insulin resistance. Pharmacotherapy with acarbose or somatostatin analogues may be needed [147], with transient enteral feeding required in severe cases [144].

There is some overlap in the hypoglycemic syndromes associated with bariatric surgery, with a number of mechanisms likely contributing to each. Obesity-related beta-cell hypertrophy that does not fully regress after weight loss, with improved GLP-1 dynamics and improved peripheral sensitivity, all probably contribute to each syndrome [148]. There can be an exaggerated incretin response in those with hypoglycemic syndromes [145, 146]. However, the extent to which the incretin effect is involved can vary by syndrome and even by case [148]. If a post-operative patient presents with dumping syndrome or hypoglycemia that is unresponsive to dietary modification or 3 kg weight gain, or with atypical features such as fasting symptoms, then a full investigation of their insulin dynamics is needed.

Other post-operative complications include alopecia, cholelithiasis and hypoglycemia. Postoperative hair loss has been reported in up to 4.5% of bariatric surgical candidates [149]. This is usually mild and non-progressive. Cholelithiasis can occur in up to 2% of individuals in the months after surgery [150]. Ursodeoxycholic acid has been recommended for prevention [140].