Customize your JAMA Network experience by selecting one or more topics from the list below.
Dixon JB, O’Brien PE, Playfair J, et al. Adjustable Gastric Banding and Conventional Therapy for Type 2 Diabetes: A Randomized Controlled Trial. JAMA. 2008;299(3):316–323. doi:https://doi.org/10.1001/jama.299.3.316
Author Affiliations: Centre for Obesity Research and Education (Drs Dixon, O’Brien, Chapman, Schachter,
and Skinner and Mss Playfair and Anderson) and Department of Epidemiology and Preventive Medicine (Dr Bailey), Monash University, Melbourne,
Australia; and Department of Medicine (AH/NH), University of Melbourne,
Melbourne (Dr Proietto).
Context Observational studies suggest that surgically induced loss of weight may be effective therapy for type 2 diabetes.
Objective To determine if surgically induced weight loss results in better glycemic control and less need for diabetes medications than conventional approaches to weight loss and diabetes control.
Design, Setting, and Participants
Unblinded randomized controlled trial conducted from December 2002 through December 2006 at the University Obesity Research Center in Australia, with general community recruitment to established treatment programs. Participants were 60 obese patients (BMI >30 and <40)
with recently diagnosed (<2 years) type 2 diabetes.
Interventions Conventional diabetes therapy with a focus on weight loss by lifestyle change vs laparoscopic adjustable gastric banding with conventional diabetes care.
Main Outcome Measures
Remission of type 2 diabetes (fasting glucose level <126
mg/dL [7.0 mmol/L] and glycated hemoglobin [HbA1c] value <6.2% while taking no glycemic therapy). Secondary measures included weight and components of the metabolic syndrome. Analysis was by intention-to-treat.
Of the 60 patients enrolled, 55 (92%) completed the 2-year follow-up.
Remission of type 2 diabetes was achieved by 22 (73%) in the surgical group and 4 (13%) in the conventional-therapy group. Relative risk of remission for the surgical group was 5.5 (95% confidence interval,
2.2-14.0). Surgical and conventional-therapy groups lost a mean (SD)
of 20.7% (8.6%) and 1.7% (5.2%) of weight, respectively, at 2 years (P < .001). Remission of type 2 diabetes was related to weight loss (R2 = 0.46, P < .001) and lower baseline HbA1c levels (combined R2 = 0.52, P < .001). There were no serious complications in either group.
Participants randomized to surgical therapy were more likely to achieve remission of type 2 diabetes through greater weight loss.
These results need to be confirmed in a larger, more diverse population and have long-term efficacy assessed.
actr.org Identifier: ACTRN012605000159651
Significant sustained weight loss achieved using bariatric surgery has never been formally investigated as a treatment for type 2 diabetes in obese participants. Several observational studies suggest substantial benefit, but these have generally been restricted to severely obese participants; to our knowledge, there have been no published randomized controlled trials.
Obesity and type 2 diabetes are likely to be the 2 greatest public health problems of the coming decades.1
The conditions are strongly linked, with the increased prevalence of diabetes correlating with the increased prevalence of obesity.2
The adjusted relative risk of developing type 2 diabetes in participants with a body mass index (BMI) greater than 35 (calculated as weight in kilograms divided by height in meters squared) is 93 (95% confidence interval [CI],
81-107) for women3
and 42 (95% CI,
22-81) for men,4
compared with participants with a BMI less than 22 and less than 23, respectively. Approximately half of those diagnosed with type 2 diabetes are obese.5
Early and intensive treatment of type 2 diabetes is known to improve health outcomes and quality of life.6-9
Weight control comprises perhaps the most important aspect of type 2 diabetes management, with weight loss reducing morbidity and mortality.10
Recent evidence indicates that improvement in blood glucose control is related to degree of weight loss.11
Unfortunately, currently available lifestyle and pharmacological strategies provide only small to modest levels of weight loss, a problem compounded by patients with diabetes experiencing greater difficulty in losing weight than those without diabetes.12-14
The costs associated with medical weight loss therapies for obese patients with type 2 diabetes are high and ongoing.15
Despite observational evidence suggesting that weight-loss surgery is associated with a 60% to 80% diabetes remission rate in severely obese persons and that earlier interventions are more likely to provide remission,16
bariatric procedures fail to generate significant attention in diabetes guidelines.17 Concerns exist regarding the lack of randomized controlled evidence, as well as the safety, invasiveness, and cost-effectiveness of surgical weight loss procedures. Providing appropriate evidence has previously proved problematic, because the invasive nature of the surgery makes participant recruitment difficult. However, with the advent of safer, less invasive surgical weight-loss procedures,
randomized controlled trials are now feasible, and studies examining mild to moderate obesity, which is responsible for much of the diabetes burden, are possible.
Multiple case series have demonstrated that laparoscopic adjustable gastric banding (LAGB) results in significant weight loss,18-20
with medium-term weight loss of approximately 20% of body weight21
; perioperative mortality is approximately 0.05% for LAGB.21
Using the LAGB intervention, we conducted a 2-year randomized controlled trial involving 60 obese participants (BMI >30 and <40)
to compare surgically induced weight loss with conventional therapy for the management of recently diagnosed type 2 diabetes (<2 years).
Patients were recruited via newspaper advertisement and were not paid to participate, nor did they pay any medical costs. The study was reviewed and approved by the human ethics committees of The Alfred Hospital, The Avenue Hospital, and Monash University in accordance with the guidelines of the National Health and Medical Research Council and the Helsinki Declaration, as revised in 2000. Recruitment commenced in December 2002, the last participant was randomized in November 2004, and all data were available for analysis in December 2006. All participants provided written informed consent to participate in the study. Additional written informed consent was obtained prior to any surgical procedure.
Patients were eligible if they were aged between 20 and 60 years,
had a body mass index of 30 to 40, had been diagnosed with clearly documented type 2 diabetes within the previous 2 years, had no evidence of renal impairment or diabetic retinopathy, and were able to understand and comply with the study process.
Candidates were excluded if they had a history of type 1 diabetes,
diabetes secondary to a specific disease, or previous bariatric surgery;
a history of medical problems such as mental impairment, drug or alcohol addiction, recent major vascular event, internal malignancy, or portal hypertension; or a contraindication for either study group. Participants were excluded if they did not attend 2 initial information visits.
In addition to any assessments required for inclusion, each potential participant was assessed by a dietitian, a general physician,
and a consultant endocrinologist specializing in diabetes (L.C.) to suggest any changes required to maximize current management. A run-in period of at least 3 months was undertaken in which further alterations to eating, exercise, glucose self-monitoring, and medications were suggested. During this time, study compliance was assessed using attendance at appointments and completion of questionnaires. The endocrinologist independently determined when a patient was ready for randomization.
Baseline weight, blood pressure, anthropometric measures, and biochemical data (levels of fasting plasma glucose, glycated hemoglobin [HbA1c],
C-peptide, and serum insulin, and a lipid profile) were measured immediately prior to randomization.
Randomization was computer derived, with blocking into 3 groups to allow for orderly recruitment into both study groups and to reduce the risk of uneven recruitment late in the series. The study was not blinded.
Conventional-Therapy Program. This program delivered best available medical practice for the treatment,
education, and follow-up of patients with type 2 diabetes. Patients had open access to a general physician, dietitian, nurse, and diabetes educator and had visits with at least 1 team member every 6 weeks throughout the 2 years. Medical therapies, including pharmaceutical agents, were determined by an experienced diabetologist on an individual basis.
Lifestyle modification programs were individually structured to reduce energy intake, to reduce intake of fat (<30%) and saturated fats, and to encourage intake of low glycemic index and high-fiber foods. Physical activity advice encouraged 10 000 steps per day and 200 minutes per week of structured activity, including moderate-intensity aerobic activity and resistance exercise. Lifestyle was the primary approach to weight loss, but very low-calorie diets and medications were discussed with all patients and used after consultation with the dietitian or general physician if the patient expressed a desire to use additional measures.
Surgical Program. In addition to all aspects of the conventional-therapy program, the surgical group underwent placement of a laparoscopic adjustable gastric band via the pars flaccida technique by 1 of 2 experienced surgeons within 1 month of randomization.22
Progress was reviewed by the bariatric surgical team every 4 to 6 weeks throughout the study, and adjustments to band volume were made using standard clinical criteria.23
The primary end points of the study related to glycemic control at 2 years after randomization. These were assessed as the proportion of participants achieving remission (exceptional glycemic control)
of type 2 diabetes, defined as fasting plasma glucose levels less than 126 mg/dL (to convert to mmol/L, multiply by 0.0555) in addition to HbA1c values less than 6.2% without the use of oral hypoglycemics or insulin. The use of metformin posed a dilemma, because it may be recommended in the remitted diabetic state. Our protocol recommended cessation of metformin, if prescribed, when the fasting insulin concentration was normal (<17.0 uIU/mL [to convert to pmol/L,
multiply by 6.945]) and the HbA1c value was less than 6.2%
with a normal fasting plasma glucose level less than 108.0 mg/dL.
Secondary outcome measures included percentage change in HbA1C levels, weight, blood pressure, waist circumference, and levels of fasting lipids, including total cholesterol, triglycerides, and high-density lipoprotein cholesterol. Changes in medication use, changes in the proportion of participants with the metabolic syndrome as defined by the National Cholesterol Education Program Adult Treatment Panel III criteria,24 and changes in indirect measures of insulin resistance using the homeostatic model assessment method were assessed. Adverse events and effects were recorded.
Sample Size. Sample size was selected to provide a statistical power of 80% to detect a 1% difference in HbA1c values between the groups at 2 years (group mean,
6.5% vs 7.5% [SD, 1.3%]) (P < .05).25
The study was also powered for diabetes remission rates on the basis that we expected approximately 60% remission in the surgical group and 20% in the conventional-therapy group.25 To allow for these scenarios, a minimum of 27 patients were required in each study group. Recruitment size was therefore set at 60.
Data Analysis. Univariate statistical analysis was performed using SPSS version 12.01 (SPSS Inc, Chicago,
Illinois), with baseline comparisons made using χ2 tests for equal proportion, t tests for normally distributed outcomes, or Mann-Whitney U tests otherwise. Multivariate longitudinal analysis was performed to assess weight and biochemical changes with time using the PROC Mixed procedure in SAS version 8.2
(SAS Institute Inc, Cary, North Carolina). All data were analyzed using a true intention-to-treat analysis for all 60 patients. Continuous variables were expressed as mean (standard deviation), with differences expressed as mean (95% CI). Binary logistic regression was used to examine the associates of diabetes remission. A 2-sided P value of .05 was considered statistically significant.
The flow of participants through the study is shown in Figure 1. One patient randomized to surgery withdrew from the study on the evening prior to scheduled operation and did not agree to be further followed up. The remaining 29 surgically treated patients (97%) completed the 2-year program. Of the conventionally treated patients, 26 (87%) completed the 2-year assessment. The baseline characteristics of the groups are shown in Table 1. There were no statistically significant differences in demographics or values contributing to study outcomes between the groups. There were only 13 participants with a baseline BMI less than 35—6 randomized to surgery and 7 to the conventional-therapy group. The mean BMI of those recruited to the study was 37.1.
All bands were placed laparoscopically, with a mean procedure time of 54 minutes (SD, 10.8; range, 40-74), and hospital admissions lengths were 1 day (23 [80%]), 2 days (5 [17%]), and 4 days (1 [3%]).
The patient who stayed 4 days had the band removed on day 15 due to band intolerance. This patient underwent follow-up for 2 years, in accordance with the intention-to-treat analysis. Sixteen of the 26
completers in the conventional-therapy group elected to use a very low-calorie diet (n = 11) or sibutramine (n = 7)
at some stage during the 2 years. None elected to use orlistat.
Remission of type 2 diabetes was achieved by 26 study participants (43%) at 2 years (22/30 [73%] randomized to the surgical program and 4/30 [13%] to the conventional-therapy program) (P < .001).
This represented 76% and 15% remission rates among completers in the surgery and conventional-therapy groups, respectively. A most conservative analysis using the assumption that all 4 noncompleters in the conventional-therapy group achieved remission and that the only noncompleter in the surgical group did not indicates significantly greater remission in the surgically treated group (22/30 [73%] for surgery vs 8/30 [27%] for conventional therapy, P < .001).
Greater percentage of weight loss at 2 years and lower baseline HbA1c values were independently associated with remission (Cox-Snell R2 = 0.50, P < .001), but percentage of weight loss alone explained most of the variance (Cox-Snell R2 = 0.46, P < .001).
The patients’ sex, age, baseline BMI, C-peptide level, time spent engaged in planned physical activity, and the group to which they were randomized were not predictive of remission after controlling for percentage of weight loss. Figure 2 shows individual percentage weight loss as well as baseline and 2-year measures of weight for patients randomized to each group.
Only 4 of 34 patients (12%) who lost less than 10% of body weight were in remission at 2 years (Figure 2).
This group was characterized by significantly lower baseline HbA1c levels when compared with others who lost 10% or more. All were below 6.9%,
with the median 6.45%, compared with median 7.6% for the remaining 30 (P = .02 by Mann-Whitney U test).
Only 4 of 26 (15%) losing more than 10% body weight did not achieve remission. There were no significant baseline predictors of this small group.
The surgical group achieved a mean 20.0% (SD, 9.4%) body weight loss at 2 years, compared with 1.4% (SD, 4.9%) among the conventional-therapy group (P < .001) (Figure 2). This represents a loss of 62.5% of excess weight (using BMI of 25 as ideal weight) in the surgical group compared with 4.3% in the conventional-therapy group and a reduction of BMI from 36.9 to 29.5 compared with a reduction from 37.1 to 36.6.
Individual and grouped mean (SD) weight changes for both groups are shown in Figure 2. The surgical group had a greater weight loss (P < .001),
and the difference between the groups increased with time (P < .001).
There was a positive correlation between average times participants reported performing planned physical activity each week throughout the study and weight loss (Spearman r = 0.39, P = .003). Participants (n = 29)
who reported more than 3 periods of physical activity of greater than 30 minutes per week (median reported) had better mean weight loss (13.9 [SD, 10.9] kg compared with 7.8 [SD, 12.3] kg, P = .046)
and a higher likelihood of diabetes remission (odds ratio, 3.4; 95%
CI, 1.2-10.1; P = .02). However,
reported physical activity was not an independent predictor of diabetes remission.
Mean levels of HbA1c and fasting plasma glucose were significantly lower in the surgical group at 2 years (P < .001 for both). At baseline there were 2 (7%)
surgically treated and 4 (13%) conventionally treated participants with HbA1c levels less than 6.2%; at 2 years, the numbers were 24 (80%) and 6 (20%), respectively. The proportion with HbA1c levels less than 6.2% improved significantly (P < .001)
in the surgical group but not in the conventional-therapy group.
There was a significant reduction in the use of pharmacotherapy for glycemic control in the surgical group at 2 years. At baseline,
2 surgically treated and 4 conventionally treated patients were not using pharmacotherapy; at 2 years, the numbers were 26 and 8, respectively.
In the surgical group at 2 years there were fewer using metformin (3 vs 28, P < .001) and other hypoglycemic therapy (1 vs 9, P = .006).
The 1 surgical patient using insulin at baseline was in remission at 2 years. There were no significant changes in the use of therapy in the group randomized to receive conventional therapy. Metformin was used by 26 and 18, other oral hypoglycemic agents by 8 and 7,
and insulin by 0 and 3, at baseline and 2 years, respectively.
Table 2 shows changes in some clinical and laboratory measures of health at 24 months. The surgical group had a significantly greater improvement at 2 years in insulin resistance and in levels of triglycerides and high-density lipoprotein cholesterol. The metabolic syndrome was present in 29
patients (97%) in each group prior to commencement of treatment, and 21 (70%) of surgically treated and 4 (13%) conventionally treated participants did not fulfill the National Cholesterol Education Program Adult Treatment Panel III criteria at 2 years (P < .001).
The reduction in the metabolic syndrome was significant in the surgical group (P < .001) but not in the conventional-therapy group (P = .23).
Caution is required in interpreting these results, as the study was not powered to assess multiple outcome measures.
Although there was no significant blood pressure difference between completers in the surgical and conventional-therapy groups,
there was a significant (P = .005)
reduction in use of antihypertensive agents in the surgical group (20/29 at baseline and 6/29 at 2 years, P < .001)
compared with the conventional-therapy group (15/26 at baseline and at 2 years). There also was a reduction in the use of lipid-lowering medications in the surgical group (12/29 at baseline and 4/29 at 2
years, P = .02) but no significant change in the conventional-therapy group (8/26 at baseline and 7/26
at 2 years).
Surgical Group. One patient developed a superficial wound infection over the access port site 2 weeks postplacement,
which resolved with intravenous antibiotics. Two patients developed gastric pouch enlargement, both at 10 months after placement, and were treated with nonurgent laparoscopic revisional surgery to remove and replace the band. One patient experienced eating difficulties and persistent regurgitation with no saline in the band and no impedance of flow on contrast study. The band was removed 15 days after placement.
Hospital stay for each revisional procedure was less than 1 day, and there were no complications. Other adverse events reported were postoperative febrile episodes in 1 patient. No cause was found, and the fever resolved.
A minor hypoglycemic episode occurred in 1 patient and gastrointestinal tract intolerance to metformin in another.
Conventional-Therapy Group. Two patients had minor gastrointestinal tract adverse effects, and 1 had persistent diarrhea with metformin. One patient developed vasculitic rash, possibly related to rosiglitazone. All problems resolved when the medications were discontinued. One patient had multiple hypoglycemic episodes,
and another was admitted to hospital with angina and a transient cerebral ischemic episode. Two patients were intolerant of very low-calorie meal replacement.
This study, to our knowledge the first randomized trial comparing surgically induced loss of weight with conventional therapy for management of type 2 diabetes in obese participants, demonstrates superior glycemic control and diabetes remission rates with adjustable gastric banding.
After 2 years, the surgical group displayed a 5 times higher remission rate and 4 times greater reduction in HbA1c values than the conventional-therapy group. This study confirms the findings of several observational studies,25,27-30
including the case-controlled Swedish Obese Subjects Study,31 that found heightened diabetes remission rates among 2000 patients undergoing gastric stapling with follow-up for 10 years.
In addition to superior glycemic control, this study also demonstrates higher rates of resolution of the metabolic syndrome, confirming the finding of an earlier randomized controlled trial,32 as well as improvements in insulin sensitivity and concentrations of triglycerides and high-density lipoprotein cholesterol in the surgical group. These changes in secondary outcome measures need to be examined with caution and are hypothesis-generating only,
as the study was not powered for an examination of multiple secondary outcome measures. There also appeared to be significant reductions in the use of oral hypoglycemic, antihypertensive, and lipid-lowering medications in patients randomized to surgery. Although the physician responsible for altering medications was not blinded to treatment group, blood pressure and biochemical data indicate that decision-making was determined by clinical need.
An important finding of this study is that degree of weight loss, not the method, appears to be the major driver of glycemic improvement and diabetes remission in obese participants. This has important implications,
as it suggests that intensive weight-loss therapy may be a more effective first step in the management of diabetes than simple lifestyle change.
This study shows that few participants achieved remission with a body weight loss of less than 10%, a level expected to produce important health benefits.33
The aim of this current trial was to compare 2 established treatment programs that involve widely available and accepted therapies. The use of bariatric surgery is rapidly increasing, and more than 90%
of procedures performed in Australia are LAGB. At the time this study was carried out, early intensive use of insulin was unusual practice in Australia, and access to thiozoladinedione medications was restricted for patients with poor glycemic control. The hypoglycemic agent exenatide was also unavailable. The results achieved by the surgical group in this study closely resemble those now being targeted by experimental intensive medical therapies using multiple hypoglycemic strategies,
including exenatide and early insulin use.34 It is worth noting that the results achieved by weight-loss surgery come without the risks of hypoglycemia and weight gain often associated with medical therapies.
Importantly, this study is the first, to our knowledge, to formally document change in glycemic control in patients with diabetes and BMI of 30 to 35 following surgically induced weight loss. Bariatric surgical guidelines include only BMI greater than 35 with comorbid type 2 diabetes as an indication for weight loss surgery.35,36 In this study we found the benefits of weight loss as efficacious for participants with BMI in the 30 to 35 range as for those with BMI in the 35 to 40 range; however,
analysis and conclusions are limited, as there were only 13 participants (22%) in this BMI category.
Our study did not include participants with a BMI greater that 40. We believed it inappropriate to recruit those with a BMI greater than 40 into the study, because a number of observational studies have shown effectiveness of bariatric surgery in these patients.
The adverse events observed in this trial were in line with expectations. Among patients undergoing LAGB surgery, rates of postoperative wound infection around the port were approximately 1% to 2%, and reoperation rates for gastric pouch enlargement were approximately 5%.
Several limitations of our study need to be recognized. First,
we restricted the study to participants with a recent diagnosis (<2
years) of type 2 diabetes, and therefore results may not apply to those with a longer history of disease due to deterioration of β-cell function with time. Second, the experience of our bariatric surgical team with the LAGB procedure is extensive. Systematic review shows inverse correlation between the experience of the LAGB surgical team and incidence of early and late complications.21 Third, our study was not powered for safety or to detect differences in hard end points, such as mortality or cardiovascular events. Such studies would require many more participants followed up over a much longer period. Our study only followed up participants for 2 years, and results cannot be readily extrapolated for longer periods. Clearly, weight regain or simply the passage of time puts those achieving remission of type 2 diabetes at risk of relapsing back into diabetic status.
One patient from the surgical group and 4 from the conventional-therapy group did not complete the 2-year follow-up. To account for this missing information, the most conservative case scenario was considered, with the 4 noncompleters from the conventional-therapy group assumed to have achieved diabetic remission and the noncompleter from the surgical group assumed to have not achieved remission. Under this scenario,
the difference between the surgical and the conventional-therapy groups for those achieving remission still remained highly significant (22/30 [73%]
for surgery vs 8/30 [27%] for conventional therapy, P < .001).
In conclusion, this randomized trial demonstrates that weight loss associated with adjustable gastric banding results in diabetes remission in the majority of obese participants recently diagnosed as having diabetes and was associated with greater improvements in features of the metabolic syndrome and use of related medications.
While caution is required in interpreting the longer-term benefits of surgery and weight loss, this study presents strong evidence to support the early consideration of surgically induced loss of weight in the treatment of obese patients with type 2 diabetes.
Corresponding Author: John B. Dixon,
MBBS, PhD, FRACGP, Centre for Obesity Research and Education, Monash University Medical School, The Alfred Hospital, Melbourne, Victoria,
3004, Australia (firstname.lastname@example.org).
Author Contributions: Drs Dixon and Bailey and Ms Anderson had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Dixon, O’Brien,
Acquisition of data: Dixon, O’Brien,
Playfair, Chapman, Schachter, Skinner, Anderson.
Analysis and interpretation of data:
Dixon, O’Brien, Playfair, Bailey, Anderson.
Drafting of the manuscript: Dixon,
Critical revision of the manuscript for important intellectual content: Dixon, O’Brien, Playfair, Chapman,
Schachter, Skinner, Proietto, Bailey, Anderson.
Statistical analysis: Dixon, Bailey,
Obtained funding: Dixon, O’Brien.
Administrative, technical, or material support:
Dixon, O’Brien, Playfair, Schachter, Skinner, Proietto, Anderson.
Study supervision: Dixon, O’Brien,
Financial Disclosures: Dr Dixon reported having received research grants from the National Health and Medical Research Council, Allergan Health, and Novartis Australia; having received compensation for serving on the speakers panel of Novartis Australia and Allergan Health; and serving on the medical advisory board of Novartis Australia and Allergan Health.
Dr O’Brien reported receiving research grants from the National Health and Medical Research Council, Allergan Health, and Novartis Australia; having received compensation for serving on the speakers panel of Allergan Health; and serving on the medical advisory board of Allergan Health. Dr Chapman reported receiving travel grants and honoraria from Eli Lilly, Novo Nordisc, Sanofi Aventis, and Alphapharm.
Dr Proietto reported serving on the medical advisory boards of Novartis Australia, Eli Lilly Australia, Abbott Australia, and Sanofi-Aventis Australia. No other disclosures were reported.
Funding/Support: This study was funded by Monash University, which has received an unrestricted grant from Allergan Health. The laparoscopic adjustable gastric bands (Allergan Health) and the laparoscopic ports (Applied Medical) were provided without charge by the manufacturers.
Role of the Sponsors: Allergan Health and Applied Medical had no role in the design and conduct of the study;
the collection, analysis, and interpretation of the data; or the preparation,
review, or approval of the manuscript.
Create a personal account or sign in to: