Introduction

Type 2 diabetes (T2D) is a serious chronic disease that has a rapidly increasing prevalence and causes a significant global burden [1].

Bariatric surgery has been established as a treatment for T2D in patients with obesity after studies have shown its effectiveness for improving glycemic control [2, 3] and a resultant decrease in microvascular complications related to remission following surgery [4]. In patients with obesity and T2D, the addition of bariatric surgery to intensive medical therapy has been shown to be better than intensive medical therapy alone for diabetes remission and other co-morbid risk factors [5, 6].

Laparoscopic sleeve gastrectomy (LSG) and laparoscopic Roux-en-Y gastric bypass (LRYGB) are the two most common primary bariatric operations performed worldwide, accounting for 50.2% and 36.9% of procedures performed from 2016 to 2020 respectively [7]. Both procedures restrict caloric intake by reducing the size of the stomach, but LRYGB, by nature of bypassing the duodenum, also causes changes in gastrointestinal hormones that may enhance diabetes remission [8]. The addition of a silastic ring around the gastric pouch of a LRYGB (SR-LRYGB) is designed to reduce outlet dilatation over time and can improve long-term weight loss by reducing weight regain [9]. It is therefore thought that over time any difference in weight loss may increase favoring banded gastric bypass.

Despite evidence that both LSG and LRYGB can induce diabetes remission, long-term data beyond 5 years directly comparing the two procedures for the outcome of diabetes remission is sparse and limited to secondary endpoint analysis therefore tending to be underpowered to detect clinically significant differences in diabetes remission [10, 11]. There are currently no randomized trials directly comparing LSG and LRYGB for the primary outcome of diabetes that have published data beyond 5 years. We previously reported the 5-year results of the current trial showing SR-LRYGB produced significantly greater long-term remission of T2D and significantly greater weight loss in obese patients compared to LSG [12], although diabetes remission was initially similar at 1 year [13]. We continue to follow our patients longitudinally and here we present results collected at 7 years.

Research Design and Methods

Trial Design

The trial design protocol for this study has been published previously [14]. Briefly, we conducted a single center, prospective, randomized, 2-arm, parallel, double-blinded (patient and assessor) clinical trial in Auckland, New Zealand. The study was approved by the Northern A Health and Disability Ethics Committee, Ministry of Health, New Zealand.

Participants

All participants provided written consent for the trial. Inclusion criteria were participants aged 20–55 years, BMI of 35–65 kg/m2, T2D (as defined by the American Diabetes Association [15]) for at least 6 months duration, suitable for either procedure, and committed to follow-up. Exclusion criteria were C-peptide < 350 pmol/L, pregnancy, type 1 diabetes or secondary diabetes, chronic pancreatitis, oral steroid therapy, current smokers, and those not suitable for general anesthesia.

Randomization and Blinding

Randomization was by computer-generated random number codes used to assign participants 1:1 to either LSG or SR-LRYGB. Stratification was performed according to age category (20–39 years, 40–55 years), BMI category (35–44.9 kg/m2, 45–54.9 kg/m2, 55–65 kg/m2), ethnicity (Maori, Pacific, New Zealand European, other), duration of diabetes diagnosis (< 5 years, 5–10 years, > 10 years), and the presence of insulin therapy. To determine the allocation arm, patient data were entered by the operating surgeon into a secure web-based randomization system (Minim, London, UK) following the induction of general anesthesia. Both operations were performed with identical laparoscopic port placement to maintain allocation concealment and blinding of participants and all other research and clinical team members. At the 5-year mark, as per the original trial protocol, participants and researchers were unblinded.

Procedures

All participants were prescribed a very-low-calorie diet with three servings of OPTIFAST (Nestle, Vevey, Switzerland) containing approximately 150 cal plus vegetables for 2 weeks before surgery. Operations were performed by bariatric surgical fellows under the direct supervision of an experienced bariatric surgeon (M.W.C.B). LSG consisted of a gastric resection using an Echelon flex 45 stapler (Ethicon) over a 36-Fr orogastric bougie, commencing 2 cm proximal from the pylorus. For SR-LRYGB, a lesser curve-based gastric pouch was fashioned over a 32-Fr orogastric tube, with a 50-cm biliopancreatic limb and 100-cm antecolic Roux limb. The gastrojejunostomy was single-layered, hand-sewn, join over a 32-Fr orogastric tube. A 6.5-cm silastic ring was placed 2–3 cm proximal to the gastrojejunal anastomosis. The mesenteric defects were routinely closed. Methylene blue dye leak test was performed in all procedures. All patients were prescribed a twice-daily multivitamin (Centrum Plus; Pfizer; New Zealand) containing 200 mg elemental calcium and 600 IU vitamin D3 post-operatively.

Objective

The objective of this 7-year secondary analysis was to assess diabetes remission in patients who were randomized to either LSG or SR-LRYGB. Additionally, we assessed weight loss and morbidity and mortality complications of the two surgeries.

Outcome Measures

The primary outcome for this trial was the proportion of patients achieving diabetes remission at 5 years, as previously reported [12]. We report here the proportions of patients with diabetes remission at 7 years, defined as an HbA1C < 6.0% (42 mmol/mol) and off diabetes medications for a minimum of 6 months beforehand. Other pre-defined secondary endpoints for diabetes included alternative diabetic thresholds of HbA1c levels < 7% (52 mmol/ mol), < 6.5% (48 mmol/mol), and ≤ 5.6% (38 mmol/mol), each without the need for any glucose-lowering medication.

Additional secondary outcomes assessed at 7 years included percentage absolute weight loss, calculated as ([baseline weight – follow-up weight]/baseline weight) × 100, percentage excess weight loss, calculated as ([baseline weight – follow-up weight]/ [baseline weight – ideal weight for BMI 25 kg/m2]) × 100, and surgical-related adverse events, classified as minor or major in accordance with standard outcome reporting guidelines [16], and recorded as either occurring early (< 30 days after surgery) or late (> 30 days after surgery).

Follow-up

Post-operative care and follow-up were identical for both groups. All pharmacological agents for diabetes were stopped at the time of surgery. Glucose-lowering therapy was restarted if mean post-operative capillary glucose was > 12 mmol/L. All patients were reviewed by an endocrinologist up to 5 years post-operatively for adjustment of metabolic medications according to a prespecified adjustment protocol and following this management of diabetic medications was transferred to general practitioners. Patients were offered follow-up appointments with the surgical team at yearly intervals for symptom review and data collection. Laboratory and medication dispensing data were also collected from an online medical record database analysis and accepted if within a window of ± 6 months from the 7-year post-operative anniversary date.

Statistical Analysis

Normally distributed continuous variables are reported as means and SDs. Intention to treat analysis was performed. Differences in the proportions of participants achieving diabetes remission were compared between SR-LRYGB and LSG groups at 7 years adjusting for stratification variables using logistic regression. Within-group changes and between-group differences are reported with point estimates and 95% CIs. For HbA1c and body weight assessed at baseline and on multiple occasions over 7 years, a repeated-measures mixed-effects model with adjustment for stratification variables was used to compare changes from baseline in the two groups. Least squares means with SEs were plotted graphically. Categorical data were compared between SR-LRYGB and LSG groups using Fisher’s exact test. Two-sided p < 0.05 was considered to indicate statistical significance. Analysis was performed using SAS 9.4 statistical software (SAS Institute, Cary, NC).

Results

Trial Patients

From September 2011 to October 2014, a total of 114 patients were randomized to undergo LSG (n = 58) and SR-LRYGB (n = 56). The flow of patients is shown in Fig. 1, and the baseline characteristics of each group summarized in Table 1. There were six deaths over the 7-year follow-up period, four from the LSG arm, and two from the SR-LRYGB arm. No deaths were directly attributed to the surgical intervention received (three because of malignancy, one because of a motor vehicle accident, one attributed to myocardial infarction, and one attributed to non-ischemic cardiomyopathy). HbA1c laboratory and medication dispensing data within ± 6 months of the 7-year post-operative timepoint for classifying diabetes remission was available in 89 patients of the remaining 108 patients (82.4%), 39 (72.2%) in the LSG group and 50 (92.6%) in the SR-LRYGB group. Weight and height measurement was available for 49 patients (45.4%).

Fig. 1
figure 1

Patient flow through the trial

Full size image
Table 1 Baseline characteristics of patients
Full size table

Diabetes Remission

Diabetes remission at 7 years, as defined by a measured HbA1c < 6% (42 mmol/mol) without glucose-lowering medications, was seen in 46.0% (23 of 50 participants) in the SR-LRYGB group compared with 30.8% (12 of 39 participants) in the LSG group (adjusted OR 4.64; 95% CI 1.39, 15.52; p = 0.013) (Fig. 2). Using alternative diabetic thresholds, participants achieving HbA1c ≤ 5.6% (38 mmol/mol), < 6.5% (48 mmol/mol), or < 7% (52 mmol/ mol), without the use of diabetic medications in LSG vs SR-LRYGB groups were 15 vs 28% (p = 0.062), 36 vs 62% (p = 0.002), and 44 vs 66% (p = 0.006), respectively (Table 2).

Fig. 2
figure 2

Percentage of patients in remission from type-2 diabetes over time (*p < 0.01)

Full size image
Table 2 End points measured at 7 years
Full size table

Weight Loss

Participants in the SR-LRYGB group had a significantly greater percentage total body weight loss (%TBWL) from baseline than those in the LSG group at 7 years (26.2% vs 13.4%; absolute difference 12.8%; 95% CI 7.2%, 18.2%; p < 0.0001) (Fig. 3). Similarly, the SR-LRYGB group had a significantly greater percentage of excess weight loss (%EBWL) at 7 years than the LSG group (66% vs 37.8%; absolute difference 28.2%; 95% CI 14.3, 42.1; p = 0.0002). The difference in BMI from baseline was also significantly greater for SR-LRYGB than LSG (− 11.4 vs − 5.7; absolute difference − 5.7; 95% CI − 8.4, − 3.0; p < 0.0001).

Fig. 3
figure 3

Percentage total weight change (from baseline) over time

Full size image

Surgical Adverse Events

Early (< 30 days after surgery) and late (> 30 days after surgery) complications are summarized in Table 3. The early minor morbidity rate was 12.5% for SR-LRYGB and 5.2% for LSG (n = 7 vs 3 respectively; p = 0.20), while the early major morbidity rate was 3.6% for SR-LRYGB and 8.6% for LSG (n = 2 vs 5; p = 0.44). The late minor morbidity rate at 7 years was 21.4% for SR-LRYGB and 13.8% for LSG (n = 12 vs 8; p = 0.33), while the late major morbidity rate was 16.1% for the SR-LRYGB and 10.3% for the LSG (n = 9 vs 6; p = 0.68). Minor morbidity occurring between 5 and 7 years post-operatively included one stenosis requiring dilatation in the LSG group and one marginal ulceration in the SR-LRYGB group. Major morbidity occurring between 5 and 7 years included one bowel obstruction requiring operation in the SR-LRYGB group, and one removal of silastic ring in the LSG group (in a patient that had been converted to SR-LRYGB at 16 months for severe reflux).

Table 3 Complications for SR-LRYGB and LSG
Full size table

Discussion

This randomized trial shows that SR-LRYGB produces significantly greater T2D remission maintenance at 7 years compared with LSG. We used the threshold of HbA1c < 6.0% without diabetic medications to define remission, in accord with the consensus definition of remission at the time of the trial protocol publication [17]. In recent years, expert consensus has changed this threshold to HbA1c < 6.5% without diabetic medications [18], and our alternative threshold analysis still shows a significant advantage for SR-LRYGB over LSG at this cut-off. These findings support our previously published 5-year analysis [12], and this study is the first of its kind to show the persisting advantage of SR-LRYGB over LSG out to 7 years for diabetes outcomes.

Most previous randomized studies directly comparing LSG and LRYGB have not used diabetes remission as a primary outcome, and therefore have been underpowered to detect differences between the two groups. The non-blinded Swiss Multicenter Bypass or Sleeve Study (SM-BOSS) included 217 patients with the primary outcome being weight loss; however, only 54 of those patients had T2D at baseline and there was no significant difference at 5 years for diabetes remission between the LSG (62%; 16 of 26) and the LRYGB (68%; 19 of 28) groups [10]. The Finnish multicenter SLEEVEPASS trial included 240 patients, of which 101 patients had T2D at baseline. This trial found no significant difference for T2D remission at 10 years when comparing LSG and LRYGB, with remission achieved in 26% (11 of 42) and 33% (13 of 39) respectively [11]. The STAMPEDE trial was another non-blinded randomized study, aiming to compare intensive medical therapy vs bariatric surgery plus intensive medical therapy, with the bariatric surgery group having arms for both LSG and LRYGB [5]. The 5-year rates of patients achieving HbA1c < 6.0% without diabetic medications were not significantly different between the LSG (15%; 7 of 47) and LRYGB (22%; 11 of 49) groups. One randomized trial that has shown a difference between LSG and LRYGB for diabetes remission is the Norwegian Oseberg study [19]. This trial showed diabetes remission in 48% (26 of 54) in the LSG group and 75% (40 of 53) in the LRYGB group at 1 year. Although these results are sort term, the methodology of this study was comparable to our current study, being a blinded randomized trial, using the same definition of diabetes remission, and comparing LSG and LRYGB for the primary outcome of T2D remission.

Our study shows an advantage for LRYGB over LSG for weight loss, a finding that corresponds with results of the above trials. The STAMPEDE trial, which had total body weight loss as a secondary endpoint, showed a significant difference between the LSG and LRYGB arms at 5 years (19 vs 23 kg, p < 0.05) [5]. In the 10-year SLEEVEPASS trial results, mean %EBWL was higher in the LRYGB group by 8.4% than the LSG group (95% CI 3.1, 13.6), and mean %TBWL was 3.5% higher in the LRYGB group than the LSG group (95% CI 1.6, 5.4) [11]. The SM-BOSS failed to show a significant difference in weight loss between the two groups at 5 years [10]; however, a pooled analysis with the SLEEVEPASS results does show a significant difference in %TBWL favoring LRYGB at 5 years (3.2%, 95% CI 1.6, 4.7) [20].

Our trial shows a larger %TBWL difference between the two groups (12.8%) than the above studies, a finding that could possibly be accounted for by the addition of the silastic ring to our LRYGB. This procedure has previously shown superior weight loss at 5 years compared to LYRGB alone [9]. One criticism of the addition of a silastic ring is the possible increase in adverse events related its insertion. Three patients (4.7%) in the SR-LRYGB arm required adjustment or removal of the silastic ring, a rate consistent with previous published data [9, 21]. One patient in the LSG group required a conversion to SR-LRYGB for severe reflux, and this patient also went on to have their silastic ring removed for dysphagia. We otherwise saw similar rates of early and late complications between groups in the trial.

The strengths of this study lie in its double-blinded, randomized design, and length of follow-up. Excluding the SLEEVEPASS trial [11], there are no other randomized trials comparing LSG and LRYGB outcomes past 5 years, and this is the only trial that primarily focuses on diabetes remission with data that is mature. As the pre-defined primary endpoint of the trial was at 5 years, patients and assessors were no longer blinded at the 7-year point, which may have added to reporting bias; however, results that we saw at 7 years are also consistent with our results reported at 5 years [12]. While follow-up rates for the outcome of diabetes remission were available for 82.4% of participants at 7 years, this was primarily due to the availability of data stored in online clinical and medication prescribing records. Data collection for secondary outcomes presented in previous iterations of this study, including blood pressure and quality-of-life surveys, was limited due to the COVID-19 pandemic and the inability to see patients in person at the 7-year follow-up due to regional lockdowns. The lack of collected data for these parameters meant that we could not meaningfully analyze differences between groups, so the decision was made to exclude them from analysis and reporting at this timepoint. We hope to have much higher follow-up when reporting the 10-year results.

Conclusion

This trial continues to show that SR-LRYGB is superior to LSG for long-term diabetes remission and weight loss for patients with obesity and T2D.