Abstract
Context
Few studies have examined the clinical characteristics that predict durable, long-term diabetes remission after bariatric surgery.
Objective
To compare diabetes prevalence and remission rates during 7-year follow-up after Roux-en-Y gastric bypass (RYGB) and laparoscopic gastric banding (LAGB).
Design
An observational cohort of adults with severe obesity recruited between 2006 and 2009 who completed annual research assessments for up to 7 years after RYGB or LAGB.
Setting
Ten US hospitals.
Participants
A total sample of 2256 participants, 827 with known diabetes status at both baseline and at least 1 follow-up visit.
Interventions
Roux-en-Y gastric bypass or LAGB.
Main Outcome Measures
Diabetes rates and associations of patient characteristics with remission status.
Results
Diabetes remission occurred in 57% (46% complete, 11% partial) after RYGB and 22.5% (16.9% complete, 5.6% partial) after LAGB. Following both procedures, remission was greater in younger participants and those with shorter diabetes duration, higher C-peptide levels, higher homeostatic model assessment of β-cell function (HOMA %B), and lower insulin usage at baseline, and with greater postsurgical weight loss. After LAGB, reduced HOMA insulin resistance (IR) was associated with a greater likelihood of diabetes remission, whereas increased HOMA-%B predicted remission after RYGB. Controlling for weight lost, diabetes remission remained nearly 4-fold higher compared with LAGB.
Conclusions
Durable, long-term diabetes remission following bariatric surgery is more likely when performed soon after diagnosis when diabetes medication burden is low and beta-cell function is preserved. A greater weight-independent likelihood of diabetes remission after RYGB than LAGB suggests mechanisms beyond weight loss contribute to improved beta-cell function after RYGB.
Trial Registration clinicaltrials.gov Identifier: NCT00465829.
Keywords: diabetes, gastric bypass, remission, laparoscopic gastric band, beta-cell function
Patients with severe obesity and diabetes who undergo metabolic bariatric surgery often experience both meaningful and sustained weight loss (1–4) and improvement in diabetes control (5–11). Although criteria differ among studies, we and others have reported long-term diabetes remission rates ranging between 50% and 70% following metabolic bariatric surgery (1, 3, 12). When comparing outcomes following the primarily restrictive laparoscopic gastric banding (LAGB) procedure to Roux-en-Y gastric bypass (RYGB) procedure, several studies have concluded that postoperative improvements in glucose control can be primarily accounted for by weight loss (13–15). On the other hand, RYGB influences a number of gut processes linked with improved glucose metabolism that are weight independent, suggesting the potential for unique benefits of diversionary procedures like RYGB on diabetes management (16–18).
We recently published 3-year follow-up data comparing diabetes remission rates and both baseline and longitudinal characteristics that predicted diabetes remission in a large cohort of participants undergoing either LAGB or RYGB (17). In that report we showed that the likelihood of achieving diabetes remission (hemoglobin A1c [HbA1c] < 6.5%, no diabetes medications) was approximately 2-fold greater following RYGB than LAGB, even after controlling for differences in weight loss between procedures. In the present study, we extend our findings to 7 years of follow-up and re-examined baseline and longitudinal characteristics associated with the likelihood of long-term durable diabetes remission, focusing on clinical and laboratory parameters that reflect changes in insulin sensitivity and secretion capacity.
Methods
Study design
Participants.
The Longitudinal Assessment of Bariatric Surgery-2 is an observational (nonrandomized) cohort study that enrolled and completed baseline measurements in 2467 adult participants between 2006 and 2009 at 1/10 centers in 6 geographically diverse US centers. Participants were studied prior to surgery and annually thereafter to evaluate changes in body weight and body composition, assess comorbid disease status, and to obtain fasting blood work for metabolic and other measurements. Of this original cohort, diabetes status could not be determined at baseline in 19 participants, no follow-up visits were available in 83 participants, and 109 underwent a metabolic bariatric surgery procedure other than RYGB or LAGB. Participants who underwent revisional surgery to another procedure (n = 113) during follow-up were analyzed according to their initial procedure choice (intention-to-treat). Following exclusion of these 211 participants, 2256 were eligible for inclusion in the current analysis (Fig. 1) (19). The institutional review board at each study center approved the protocol, and written, informed consent was obtained from all participants before enrollment.
Figure 1.
Decision tree for determining diabetes remission status.
Definition of diabetes and diabetes remission
At baseline, the following criteria were used to determine if participants were included in the group without diabetes: (1) not taking any diabetes medications, (2) not self-reporting a diagnosis of diabetes, and (3) having a measured HbA1c < 39 mmol/M (5.7%) or, if not available, a fasting glucose < 5.6 mmol/L (100 mg/dL). Participants were classified as having diabetes if: (1) their HbA1c was ≥48 mmol/M (6.5%) or, if not available, their fasting glucose was ≥6.9 mmol/L (126 mg/dL); or (2) they self-reported that they currently had diabetes and self-reported use of at least 1 medication for diabetes; or (3) they self-reported currently having diabetes and reported ever having been hospitalized for treatment of a diabetes complication; or (4) self-reported use of any prescription medication for diabetes in the 90 days before surgery. Because metformin is commonly used for the management of prediabetes and polycystic ovary syndrome, participants who met the following criteria were excluded from this analysis if: (1) they self-reported use of metformin but no other diabetes medications, (2) they did not self-report a diagnosis of diabetes or reported a diagnosis of polycystic ovary syndrome, and (3) they had an HbA1c < 48 mmol/M (6.5%).
Partial and complete diabetes remission were defined using the American Diabetes Association Consensus Group (20). Partial remission was defined as an HbA1c between 39 and 48 mmol/M (5.7–6.5%) or, if not available, a fasting glucose 5.6 to 6.9 mmol/l (100–125 mg/dL) in the absence of diabetes pharmacologic therapy. Complete remission was defined as an HbA1c < 39 mmol/M (5.7%) or, if not available, a fasting glucose < 5.6 mmol/l (100 mg/dL) in the absence of diabetes pharmacologic therapy. The decision to continue or discontinue diabetes medications was not standardized but instead left to the discretion of each patient and their clinician.
Weight, circumferences, and weight loss
All measurements were conducted by trained study staff. Body weights, percentage of weight loss, and body mass index (BMI) were measured, as previously reported (21). The percentage of body fat was measured by bioelectrical impedance analysis using a Tanita scale Model TBF-310 (Tanita Corporation, Arlington Heights, Illinois). With the patient standing, the neck circumference was measured at the midpoint between the chin and clavicle and the waist circumference was measured around the abdomen horizontally at the midpoint between the highest point of the iliac crest and the lowest part of the costal margin in the midaxillary line.
Laboratory analyses
Glucose, alanine aminotransferase (ALT), and aspartate aminotransferase (AST) were measured using a Roche autoanalyzer (Roche Diagnostics Inc, Indianapolis, Indiana). Leptin, total ghrelin, and proinsulin levels were determined by radioimmunoassay (EMD Millipore, Inc, St. Charles, Missouri). C-Peptide and insulin levels were measured by a 2-site immunoenzymometeric assay using a Tosoh 2000 auto-analyzer (TOSOH, Biosciences, Inc, South San Francisco, California). HbA1c levels were measured by a dedicated analyzer (TOSOH, Biosciences, Inc, South San Francisco, California). Nonesterified fatty acid levels were determined using a Roche Hitachi Modular P analyzer (Roche Diagnostics Inc, Indianapolis, Indiana). Highly-sensitive C-reactive protein and cystatin C levels were measured using a Siemens Dade Behring BN II Nephelometer (Siemens Inc, Munich, Germany). With the exception of leptin and ghrelin levels, which were measured at baseline and postoperative years 1 to 3, all laboratory studies were assayed at baseline and annually.
Calculations
The HOMA2 calculator (Oxford University, Diabetes Trials Unit [DTU], homeostatic model assessment [HOMA] for β-cell function and insulin sensitivity, a HOMA calculator https://www.dtu.ox.ac.uk/homacalculator/) was used to estimate insulin resistance (HOMA IR) and beta-cell function (HOMA %β) using fasting values for insulin and glucose (22, 23).
Statistical analysis
Patient demographic and clinical characteristics are reported as means and 95% confidence intervals or counts and percent. All factors were evaluated for normality and skewness. Partial and complete remission was defined only for patients who had diabetes mellitus at baseline and achieved the definition of “without diabetes” for each visit. Differences in patient characteristics between procedure type and remission status were evaluated with t-tests, Wilcoxon rank-sum tests, or chi-square tests, as appropriate. Hedges g and Phi effect sizes were calculated to evaluate the magnitude of differences. The odds of remission (reported as adjusted risk ratios) were evaluated with multivariable mixed effects poisson regression, with robust standard errors and stratified by procedure type. Due to the large number of possible factors contributing to remission, these models were constructed in a 3-step process: (1) 3 preliminary multivariable Poisson models were constructed for each procedure type (1 model containing only laboratory variables, 1 model containing only comorbidities, and 1 model containing only demographic variables), (2) factors associated with remission at any point during follow-up at P < 0.20 were eligible for the final model, and (3) Hosmer and Lemeshow’s variable reduction methods were then used to inform the construction of the final multivariable Poisson model. Data collection for proinsulin and C-peptide stopped at 3 years, so the last observed data point was carried over into subsequent years in which the patient had a recorded visit; models were fit with and without these variables, and only kept in the model if all other associations remained consistent. To evaluate the effect of the change in various clinical characteristics from baseline on remission, univariable risk ratios were estimated from a mixed effects Poisson regression, with robust standard errors and stratified by procedure type. All models were adjusted for the baseline factor. Finally, the adjusted relative risk of diabetes remission was estimated for RYGB compared with LAGB only for those patients with diabetes at baseline using multivariable mixed effects logistic regression, adjusting for weight change from baseline, visit, age at surgery, and a propensity score weight to adjust for differences in patient characteristics between the procedures. The propensity score model included age at surgery, gender, education level (high school only vs postsecondary education), baseline BMI, the number of baseline comorbidities, and center. A small number of subjects underwent surgical revision to another procedure during follow-up. This was most typically involved when converting a LAGB to RYGB. Analysis was conducted retaining a participant’s original surgical assignment, even if they underwent conversion to another procedure (eg, LAGB to RYGB). Significance was assessed at P < 0.05. Analysis was conducted in JMP 14.2 and SAS 9.4 (both SAS Inc, Cary, North Carolina).
Results
Baseline descriptive statistics
Of the 2256 participants at baseline, 827 (37%) had diabetes (Supplemental Table 1; all supplementary material and figures are located in a digital research materials repository) (19). Fifty percent of the 827 participants with diabetes at baseline reported the duration of their diabetes, 43% reported taking just 1 noninsulin diabetes medication, 32% reported taking 2 or more noninsulin medications, and 36% reported taking insulin. Compared to those without diabetes or prediabetes (n = 1111), those with diabetes were older and more likely to be male and non-White, had greater waist and neck circumferences, significantly higher insulin and proinsulin levels, lower estimated insulin secretion (HOMA %B), greater estimated insulin resistance (HOMA IR), higher cytokine and adipokine levels, higher liver enzyme levels (ALT, AST), a higher frequency of abnormal renal function (including higher cystatin C levels), a greater frequency of obstructive sleep apnea and continuous positive airway pressure (CPAP) use, and a greater medication burden for obesity-related complications (Supplemental Table 1) (19).
At baseline, participants with diabetes who achieved complete or partial remission during at least 1 follow-up visit after surgery (n = 562, 68%), compared with those who did not, were, on average, younger, heavier (with a greater percentage of body fat), had shorter diabetes duration, lower HbA1c levels, and were less likely to need more than 1 noninsulin diabetes medication or use insulin. They had similar HOMA IR but higher C-peptide and proinsulin levels as well as HOMA %B. The proportion of subjects using lipid-lowering medications was lower, and the frequency of participants undergoing RYGB versus LAGB was higher (Supplemental Table 2) (19).
Among those with diabetes, most participants with diabetes opted to undergo RYGB (78%) rather than LAGB (22%) (Table 1). Despite the nonrandomized nature of this study, baseline characteristics of these groups were similar, with a few exceptions (Table 1). Similarities between the groups included self-reported duration of diabetes, levels of HbA1c, frequency of insulin and noninsulin diabetes medication use, parameters of insulin resistance and insulin secretion, and levels of free fatty acids, liver enzymes, and cystatin C. On the other hand, compared to the LAGB group, those who underwent RYGB were slightly younger and heavier with greater percent body fat and waist circumferences, had higher leptin and hs-CRP levels, and were less likely to use CPAP even though the frequency of obstructive sleep apnea was not different.
Table 1.
Descriptive statistics for baseline patient characteristics by procedure for diabetes patients only
| Mean (CI) or N (%) | P-value | ||||
|---|---|---|---|---|---|
| Patient Characteristic at Baseline | RYGB (N = 645) | LAGB (N = 182) | |||
| Age at surgery (years) | 49.12 | (48.35, 49.9) | 53.24 | (51.66, 54.82) | <0.0001 |
| Sex (female) | 482 | (74.73%) | 114 | (62.64%) | 0.0013 |
| Race | |||||
| White | 531 | (83.1%) | 165 | (91.67%) | 0.074 |
| Black | 80 | (12.52%) | 9 | (5%) | |
| Other | 28 | (4.38%) | 6 | (3.33%) | |
| Hispanic ethnicity | 27 | (4.19%) | 10 | (5.52%) | 0.473 |
| BMI (kg/m2) | 48.06 | (47.45, 48.66) | 45.12 | (44.05, 46.19) | <0.0001 |
| Body fat, % | 49.80 | (49.23, 50.37) | 48.28 | (47.17, 49.38) | 0.0129 |
| Waist circumference (cm) | 137.24 | (136.01, 138.47) | 133.61 | (131.24, 135.97) | 0.0071 |
| Neck circumference (cm) | 44.04 | (43.64, 44.44) | 44.33 | (43.52, 45.13) | 0.513 |
| Diabetes duration (years) | 8.03 | (7.19, 8.87) | 8.48 | (7.12, 9.84) | 0.577 |
| Glucose (mmol/L) | 7.50 | (7.24, 7.77) | 7.95 | (7.44, 8.46) | 0.122 |
| Hemoglobin A1c (%) | 7.05 | (6.93, 7.17) | 7.09 | (6.89, 7.3) | 0.755 |
| Insulin use (%) | 186 | (36.26%) | 54 | (36.49%) | 0.959 |
| Number of noninsulin medications | |||||
| One (%) | 44 | (6.82%) | 11 | (6.04%) | 0.979 |
| Two or more (%) | 523 | (81.09%) | 153 | (84.07%) | |
| Insulin (pmol/L) | 255.51 | (221.59, 289.43) | 255.68 | (208.23, 303.14) | 0.996 |
| C-peptide (nmol/L) | 1.33 | (1.27, 1.38) | 1.39 | (1.26, 1.52) | 0.333 |
| Proinsulin | 46.42 | (42.94, 49.91) | 50.40 | (42.72, 58.08) | 0.314 |
| Proinsulin/Insulin Ratio | 0.26 | (0.24, 0.28) | 0.31 | (0.22, 0.39) | 0.090 |
| HOMA IR | 3.23 | (3.07, 3.38) | 3.41 | (3.1, 3.72) | 0.274 |
| HOMA %B | 121.12 | (113.78, 128.47) | 111.53 | (97.59, 125.46) | 0.223 |
| Free fatty acids | 0.74 | (0.71, 0.76) | 0.70 | (0.66, 0.74) | 0.170 |
| hs-CRP | 1.03 | (0.95, 1.1) | 0.83 | (0.7, 0.96) | 0.0196 |
| Leptin | 57.92 | (55.78, 60.06) | 51.10 | (47.26, 54.94) | 0.0032 |
| Leptin to fat mass ratio | 0.87 | (0.84, 0.9) | 0.80 | (0.74, 0.87) | 0.069 |
| Ghrelin | 748.01 | (729.19, 766.83) | 738.68 | (694.17, 783.19) | 0.667 |
| ALT (μkat) | 0.62 | (0.58, 0.65) | 0.64 | (0.55, 0.72) | 0.627 |
| AST (μkat) | 0.56 | (0.53, 0.58) | 0.61 | (0.53, 0.68) | 0.119 |
| Abnormal kidney function | 95 | (15.32%) | 21 | (12.21%) | 0.307 |
| Cystatin C | 1.00 | (0.96, 1.03) | 0.96 | (0.89, 1.04) | 0.401 |
| Obstructive sleep apnea | 400 | (62.11%) | 126 | (69.23%) | 0.078 |
| CPAP use | 334 | (83.08%) | 112 | (88.89%) | 0.0358 |
| Antihyperlipidemia medication use | 305 | (60.04%) | 94 | (63.51%) | 0.247 |
| Antihypertension medication use | 523 | (83.41%) | 152 | (87.86%) | 0.296 |
| Antidepression medication use | 245 | (41.88%) | 70 | (40.94%) | 0.826 |
Differences in continuous variables were evaluated with t-tests. Associations with categorical variables were evaluated with chi-square tests. Hedge’s g effect size was calculated for continuous variables and Phi was calculated for categorical variables. *small, **medium, ***large effect size. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CI, confidence interval; CPAP, continuous positive airway pressure; HOMA-%B, homeostatic model assessment of beta cell function; HOMA IR, homeostatic model assessment of insulin resistance; hs-CRP, highly-sensitive C-reactive protein; LAGB, laparoscopic gastric banding; RYGB, Roux-en-Y gastric bypass.
Diabetes prevalence, incidence, and remission
During follow-up, the prevalence rates of diabetes dropped after both procedures, more so after RYGB, and incidence rates (defined as the percentage of participants with diabetes at the given time point after surgery who did not have diabetes or prediabetes at baseline) remained low (Table 2).
Table 2.
Observed prevalence, incidence, and remission (complete and partial) of diabetes and BMI by follow-up time point and surgical procedure for all participants undergoing RYGB or LAGB
| Diabetes Status By Procedure |
Baseline | 1 Year | 2 Year | 3 Year | 4 Year | 5 Year | 7 Year | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | Count | % | |
| RYGB | ||||||||||||||
| Prevalence | 645/1667 | 38.7% | 236/1415 | 16.7% | 191/1296 | 14.7% | 211/1326 | 15.9% | 232/1345 | 17.2% | 179/1350 | 13.3% | 149/928 | 16.1% |
| Incidence | NA | NA | 6/866 | 0.7% | 1/785 | 0.1% | 4/811 | 0.5% | 4/810 | 0.5% | 3/799 | 0.4% | 1/567 | 0.2% |
| Complete remission | NA | NA | 258/549 | 47.0% | 261/509 | 51.3% | 249/508 | 49.0% | 239/525 | 45.5% | 236/539 | 43.8% | 159/346 | 46.0% |
| Partial remission | NA | NA | 62/549 | 11.3% | 58/509 | 11.4% | 52/508 | 10.2% | 58/525 | 11.0% | 65/539 | 12.1% | 39/346 | 11.3% |
| Total remission | NA | NA | 320/549 | 58.3% | 319/509 | 62.7% | 301/508 | 59.3% | 297/525 | 56.6% | 301/539 | 55.8% | 198/346 | 57.2% |
| Missing | NA | NA | 96/645 | 14.9% | 136/645 | 21.1% | 137/645 | 21.2% | 120/645 | 18.6% | 106/645 | 16.4% | 299/645 | 46.4% |
| BMI (kg) | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR |
| 46.4 | 9.3 | 30.7 | 7.6 | 30.4 | 7.9 | 31.5 | 8.4 | 32.5 | 8.6 | 32.9 | 9.3 | 33.3 | 9.4 | |
| LAGB | ||||||||||||||
| Prevalence | 182/589 | 30.9% | 127/518 | 24.5% | 104/484 | 21.5% | 104/502 | 20.7% | 106/491 | 21.6% | 111/497 | 22.3% | 62/307 | 20.2% |
| Incidence | NA | NA | 0/356 | 0.0% | 1/324 | 0.3% | 0/345 | 0.0% | 1/329 | 0.3% | 3/333 | 0.9% | 2/205 | 1.0% |
| Complete remission | NA | NA | 26/162 | 16.0% | 35/155 | 22.6% | 30/151 | 19.9% | 30/155 | 19.4% | 24/152 | 15.8% | 15/89 | 16.9% |
| Partial remission | NA | NA | 12/162 | 7.4% | 10/155 | 6.5% | 10/151 | 6.6% | 12/155 | 7.7% | 11/152 | 7.2% | 5/89 | 5.6% |
| Total remission | NA | NA | 38/162 | 23.5% | 45/155 | 29.0% | 40/151 | 26.5% | 42/155 | 27.1% | 35/152 | 23.0% | 20/89 | 22.5% |
| Missing | NA | NA | 20/182 | 11.0% | 27/182 | 14.8% | 31/182 | 17.0% | 27/182 | 14.8% | 30/182 | 16.5% | 93/182 | 51.1% |
| BMI (kg) | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR | Median | IQR |
| 43.8 | 7.8 | 37.0 | 8.2 | 36.6 | 9.1 | 36.9 | 9.2 | 36.5 | 10.2 | 36.5 | 9.6 | 36.8 | 10.5 | |
Baseline sample sizes: RYGB = 645 (diabetes), 794 (no diabetes); LAGB = 182 (diabetes), 317 (no diabetes). Prevalence is the percentage of all participants with diabetes at the given time point who had a recorded diabetes status at baseline (diabetes or no diabetes). Incidence is the percentage of participants with diabetes at the given time point who did not have diabetes at baseline. Partial remission and complete remission are defined in the “Methods” section. Missing is the percentage of participants who were missing follow-up information about diabetes out of those who had diabetes information at baseline. Percentages were calculated from available data and do not include data lost to follow-up.
Abbreviations: BMI, body mass index; IQR, interquartile range; RYGB, Roux-en-Y Gastric Bypass; LAGB, laparoscopic adjustable gastric banding; NA, not applicable.
Diabetes remission (including both complete and partial) peaked 2 to 3 years after both procedures before declining during the 7-year follow-up period (Table 2). This decline from a peak of 62% remission (51% in complete and 11% in partial remission) following RYGB to 57% by year 7 represented a reduction in those in complete remission, which dropped to 46%, whereas the number in partial remission remained stable at 11%. On the other hand, the decline diabetes remission after LAGB resulted from reductions in both complete (22.6–16.9%) and partial (6.5–5.6%) remission categories.
Baseline characteristics associated with diabetes remission: RYGB vs LAGB
In both RYGB and LAGB groups, postoperative diabetes remission was more likely in those who were younger, with shorter diabetes duration, and with better baseline HbA1c levels that required fewer diabetes medications or need for insulin at baseline (Tables 3 and 4). Participants who achieved diabetes remission after both surgeries also had greater baseline estimated beta-cell function by both C-peptide levels and HOMA %B without any difference in HOMA IR, had higher highly-sensitive C-reactive protein (hs-CPR) levels, and were much less likely to use antihyperlipidemia medications compared with those that did not achieve diabetes remission. Using a multivariate mixed model to calculate adjusted risk ratios using all relevant baseline characteristics, individuals from both surgical groups were more likely to achieve diabetes remission during follow-up if they had lower baseline insulin levels and higher C-peptide levels and HOMA %B (Table 5).
Table 3.
Baseline descriptive characteristics for patients with diabetes undergoing RYGB
| Never in Remission (N = 162) | Remission (N = 483) | P-value and Effect Size | ||||
|---|---|---|---|---|---|---|
| Age at Surgery (years) | 51.57 | 49.99, 53.14 | 48.30 | 47.42, 49.19 | 0.0002 | * |
| Sex (female) | 119 | 73.46% | 363 | 75.16% | 0.67 | * |
| Race | ||||||
| White | 135 | 84.91% | 396 | 82.5% | 0.77 | * |
| Black | 18 | 11.32% | 62 | 12.92% | ||
| Other | 6 | 3.77% | 22 | 4.58% | ||
| Hispanic ethnicity | 7 | 4.32% | 20 | 4.14% | 0.25 | * |
| BMI (kg/m2) | 47.35 | 46.1, 48.59 | 48.29 | 47.61, 48.98 | 0.08 | * |
| Body fat, % | 47.93 | 46.61, 49.24 | 50.44 | 49.83, 51.05 | 0.0012 | * |
| Waist circumference (cm) | 137.19 | 134.68, 139.7 | 137.25 | 135.84, 138.67 | 0.83 | * |
| Neck circumference (cm) | 44.02 | 43.24, 44.8 | 44.04 | 43.58, 44.51 | 0.83 | * |
| Diabetes duration (years) | 13.51 | 11.55, 15.46 | 6.07 | 5.32, 6.82 | <0.0001 | *** |
| Glucose (mmol/L) | 8.04 | 7.39, 8.69 | 7.32 | 7.05, 7.6 | 0.07 | * |
| Hemoglobin A1c (%) | 7.72 | 7.49, 7.96 | 6.83 | 6.7, 6.96 | <0.0001 | ** |
| Insulin use (%) | 97 | 66.44% | 89 | 24.25% | <0.0001 | ** |
| Number of noninsulin medications | ||||||
| One (%) | 63 | 40.38% | 204 | 43.87% | <0.0001 | * |
| Two or more (%) | 60 | 38.46% | 138 | 29.68% | ||
| Insulin (pmol/L) | 342.12 | 220.46, 463.79 | 226.52 | 206.54, 246.49 | 0.30 | * |
| C-peptide (nmol/L) | 2.92 | 2.56, 3.28 | 4.66 | 4.47, 4.86 | <0.0001 | *** |
| Proinsulin | 35.91 | 28.97, 42.86 | 49.94 | 45.94, 53.94 | <0.0001 | * |
| Proinsulin/insulin Ratio | 0.21 | 0.17, 0.25 | 0.27 | 0.25, 0.29 | <0.0001 | * |
| HOMA IR | 3.22 | 2.88, 3.56 | 3.23 | 3.05, 3.4 | 0.93 | * |
| HOMA %B | 107.77 | 91.6, 123.95 | 124.91 | 116.65, 133.16 | 0.0062 | * |
| Free fatty acids | 0.70 | 0.65, 0.75 | 0.75 | 0.72, 0.78 | 0.0164 | * |
| hs-CRP | 0.85 | 0.7, 1 | 1.08 | 0.99, 1.18 | 0.0003 | * |
| Leptin | 61.83 | 56.73, 66.94 | 56.61 | 54.32, 58.9 | 0.19 | * |
| Leptin to fat mass ratio | 0.95 | 0.87, 1.03 | 0.84 | 0.8, 0.88 | 0.01 | * |
| Ghrelin | 752.31 | 714.94, 789.67 | 746.57 | 724.69, 768.45 | 0.78 | * |
| ALT (μkat) | 34.24 | 31.33, 37.15 | 37.02 | 34.67, 39.36 | 0.44 | * |
| AST (μkat) | 31.52 | 29.09, 33.95 | 33.16 | 31.24, 35.08 | 0.64 | * |
| Abnormal kidney function | 35 | 22.58% | 60 | 12.9% | 0.0051 | * |
| Cystatin C | 1.08 | 0.99, 1.18 | 0.97 | 0.93, 1 | 0.03 | * |
| Obstructive sleep apnea | 96 | 59.26% | 303 | 62.86% | 0.42 | * |
| CPAP use | 81 | 82.65% | 253 | 83.22% | 0.90 | * |
| Dyslipidemia | 105 | 75% | 309 | 73.05% | 0.65 | * |
| Antihyperlipidemia medication use | 114 | 71.7% | 234 | 49.68% | <0.0001 | * |
| Antihypertension medication use | 137 | 85.09% | 361 | 76.16% | 0.0142 | * |
| Antidepression medication use | 63 | 41.45% | 182 | 42.03% | 0.90 | * |
Results are mean (95% CI) or n (%). Differences in continuous variables were evaluated with t-tests or Wilcoxon Rank-sum tests, as appropriate. Associations with categorical variables were evaluated with chi-square tests. Hedge’s g effect size was reported for continuous variables and Phi was reported for categorical variables. *small, **medium, ***large effect size. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CI, confidence interval; CPAP, continuous positive airway pressure; HOMA-%B, homeostatic model assessment of beta cell function; HOMA IR, homeostatic model assessment of insulin resistance; hs-CRP, highly-sensitive C-reactive protein; RYGB, Roux-en-Y gastric bypass.
Table 4.
Baseline descriptive characteristics for patients with diabetes undergoing LAGB
| Never in Remission (N = 103) | Remission (N = 79) | P-value and Effect Size | ||||
|---|---|---|---|---|---|---|
| Age at surgery (years) | 55.76 | 53.91, 57.61 | 49.95 | 47.34, 52.56 | 0.0005 | ** |
| Sex (female) | 41 | 39.81% | 27 | 34.18% | 0.44 | * |
| Race | ||||||
| White | 93 | 91.18% | 72 | 92.31% | 0.29 | * |
| Black | 4 | 3.92% | 5 | 6.41% | ||
| Other | 5 | 4.9% | 1 | 1.28% | ||
| Hispanic ethnicity | 6 | 5.83% | 4 | 5.06% | 0.42 | * |
| BMI (kg/m2) | 45.79 | 44.19, 47.38 | 44.25 | 42.88, 45.61 | 0.26 | * |
| Body fat, % | 47.22 | 45.61, 48.82 | 49.41 | 47.89, 50.93 | 0.12 | * |
| Waist circumference (cm) | 134.76 | 131.64, 137.88 | 132.21 | 128.52, 135.91 | 0.19 | * |
| Neck circumference (cm) | 45.24 | 44.12, 46.36 | 43.22 | 42.08, 44.36 | 0.0185 | * |
| Diabetes duration (years) | 11.08 | 9.12, 13.03 | 4.71 | 3.63, 5.78 | <0.0001 | *** |
| Glucose (mmol/L) | 8.47 | 7.74, 9.2 | 7.28 | 6.58, 7.98 | 0.0006 | * |
| Hemoglobin A1c (%) | 7.53 | 7.24, 7.81 | 6.51 | 6.28, 6.75 | <0.0001 | *** |
| Insulin use (%) | 48 | 53.93% | 6 | 10.17% | <0.0001 | ** |
| Number of noninsulin medications | ||||||
| One (%) | 45 | 45.92% | 33 | 41.77% | 0.0001 | * |
| Two or more (%) | 35 | 35.71% | 22 | 27.85% | ||
| Insulin (pmol/L) | 276.14 | 199.29, 352.98 | 229.58 | 182.11, 277.05 | 0.92 | * |
| C-peptide (nmol/L) | 3.90 | 3.28, 4.52 | 5.10 | 4.58, 5.62 | <0.0001 | * |
| Proinsulin | 46.43 | 38.06, 54.79 | 55.62 | 41.4, 69.84 | 0.15 | * |
| Proinsulin/insulin ratio | 0.32 | 0.18, 0.47 | 0.29 | 0.23, 0.34 | 0.26 | * |
| HOMA IR | 3.43 | 2.98, 3.88 | 3.38 | 2.96, 3.81 | 0.70 | * |
| HOMA %B | 105.42 | 84.04, 126.8 | 119.70 | 103.36, 136.05 | 0.002 | * |
| Free fatty acids | 0.71 | 0.65, 0.77 | 0.69 | 0.63, 0.75 | 0.93 | * |
| hs-CRP | 0.73 | 0.55, 0.91 | 0.95 | 0.76, 1.14 | 0.0319 | * |
| Leptin | 53.64 | 48.08, 59.21 | 47.76 | 42.62, 52.89 | 0.28 | * |
| Leptin to fat mass ratio | 0.84 | 0.74, 0.94 | 0.76 | 0.68, 0.84 | 0.49 | * |
| Ghrelin | 746.68 | 679.44, 813.91 | 728.16 | 672.94, 783.39 | 0.71 | * |
| ALT (μkat) | 33.47 | 28.14, 38.81 | 42.52 | 33.07, 51.97 | 0.28 | * |
| AST (μkat) | 34.97 | 29.42, 40.52 | 36.56 | 29.92, 43.2 | 0.98 | * |
| Abnormal kidney function | 13 | 13.54% | 8 | 10.53% | 0.55 | * |
| Cystatin C | 1.01 | 0.89, 1.14 | 0.90 | 0.84, 0.95 | 0.23 | * |
| Obstructive sleep apnea | 77 | 74.76% | 51 | 64.56% | 0.14 | * |
| CPAP use | 69 | 90.79% | 43 | 86% | 0.41 | * |
| Dyslipidemia | 60 | 67.42% | 47 | 67.14% | 0.97 | * |
| Antihyperlipidemia medication use | 73 | 73.74% | 34 | 43.04% | <0.0001 | ** |
| Antihypertension medication use | 86 | 86.87% | 60 | 75.95% | 0.06 | * |
| Antidepression medication use | 45 | 47.37% | 25 | 32.89% | 0.05 | * |
Results are mean (95% CI) or n (%). Differences in continuous variables were evaluated with t-tests or Wilcoxon rank-sum tests, as appropriate. Associations with categorical variables were evaluated with chi-square tests. Hedge’s g effect size was reported for continuous variables and Phi was reported for categorical variables. *small, **medium, ***large effect size. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CI, confidence interval; CPAP, continuous positive airway pressure; HOMA-%B, homeostatic model assessment of beta cell function; HOMA IR, homeostatic model assessment of insulin resistance; hs-CRP, highly-sensitive C-reactive protein; LAGB, laparoscopic gastric banding; RYGB, Roux-en-Y gastric bypass.
Table 5.
Multivariable adjusted risk ratios for baseline risk factors associated with diabetes mellitus remission among participants undergoing either RYGB or LAGB procedures
| RYGB, N = 587 | LAGB, N = 165 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Adjusted | 95% CI | Adjusted | 95% CI | ||||||
| Risk Factor | Referent | Risk Ratio | Lower | Upper | P-value | Risk Ratio | Lower | Upper | P-value |
| Age at surgery | per ↑ 5 year | 0.99 | 0.96 | 1.03 | 0.710 | 0.94 | 0.84 | 1.05 | 0.274 |
| BMI | per ↓ 1 | 1.02 | 1.00 | 1.03 | 0.022 | – | – | – | – |
| Baseline weight | per ↓ 5 kg | 0.98 | 0.97 | 0.99 | 0.003 | – | – | – | – |
| Weight change from baseline (%) | per ↓ 10% | – | – | – | – | 1.29 | 0.99 | 1.67 | 0.057 |
| Waist circumference | per ↓ 10 cm | 1.01 | 0.96 | 1.07 | 0.612 | 0.88 | 0.75 | 1.02 | 0.088 |
| Noninsulin medications at baseline | None vs any | 0.85 | 0.65 | 1.12 | 0.254 | 1.03 | 0.52 | 2.03 | 0.943 |
| Insulin use at baselineb | Yes vs No | 0.37 | 0.29 | 0.49 | <0.0001 | – | – | – | – |
| Insulin | per ↓ of 30 pmol | 1.07 | 1.01 | 1.14 | 0.020 | 1.16 | 1.03 | 1.30 | 0.014 |
| C-peptidea | per ↑ 1 nmol | 1.00 | 1.00 | 1.01 | 0.036 | 1.02 | 1.00 | 1.04 | 0.047 |
| Proinsulina | per ↓ of 5 | 1.03 | 1.01 | 1.05 | 0.009 | 1.07 | 0.98 | 1.18 | 0.144 |
| HOMA %B | per ↑ 20 unit | 1.06 | 1.03 | 1.09 | <0.0001 | 1.10 | 1.03 | 1.18 | 0.003 |
| Cystatin-C | per ↓ of 0.50 | 1.05 | 0.95 | 1.15 | 0.336 | 1.50 | 1.06 | 2.12 | 0.023 |
| hs-CRP | per ↓ of 0.50 | – | – | – | – | 1.03 | 0.99 | 1.08 | 0.091 |
| Education | HS graduation or better | – | – | – | – | 0.49 | 0.32 | 0.76 | 0.002 |
| Smoking | Never vs former/current | 1.20 | 0.91 | 1.60 | 0.201 | 0.75 | 0.48 | 1.17 | 0.209 |
| Hyperlipidemia | No vs Yes | 1.25 | 1.09 | 1.43 | 0.001 | – | – | – | – |
Analysis was stratified by procedure; there were 2 multivariable mixed models in total. Hosmer and Lemeshow’s variable reduction methods contributed to the construction of the multivariate models from univariate factors associated with remission at any point during follow-up. Abbreviations: BMI, body mass index; CI, confidence interval; HOMA-%B, homeostatic model assessment of beta cell function; hs-CPR, highly-sensitive C-reactive protein; LAGB, laparoscopic gastric banding; RYGB, Roux-en-Y gastric bypass.
a Data collection for proinsulin and C-peptide stopped at 3 years; therefore, the last observed data point was carried over into subsequent years in which the patient had a recorded visit. Models were fit with and without these variables, and were only kept in the model if all other associations remained consistent.
b Could not be estimated in the LAGB model because only 2 participants who were on insulin at baseline ever went into remission.
On the other hand, baseline characteristics that were selectively associated with diabetes remission after RYGB included higher free fatty acid levels, lower leptin-to-fat mass ratio (despite having a greater BMI and percentage of body fat), lower proinsulin and proinsulin/insulin levels, lower rates of antihypertension medication use, and a much lower rate of abnormal kidney function (Table 3). In the multivariable setting, baseline lower BMI, lower proinsulin levels, and not having hyperlipidemia was associated with a higher adjusted risk ratio of diabetes remission after RYGB but not LAGB, whereas baseline lower cystatin C was associated with a greater likelihood of diabetes remission after LAGB than RYGB (Table 5).
Postoperative characteristics associated with diabetes remission: RYGB vs LAGB
During the follow-up period, the likelihood for achieving diabetes remission (either complete or partial) during 7 years of follow-up for both surgical groups diminished with time since surgery and was associated with greater percentage of weight loss and reductions in waist and neck circumferences (Table 6). Although insulin resistance improved following both procedures and was greater after RYBP than LAGB (Supplemental Table 3) (19), reductions in insulin levels and insulin resistance during follow-up were associated with a greater likelihood of achieving diabetes remission only after LAGB, whereas improved estimated beta-cell secretion was significantly associated with diabetes remission after RYGB (Table 6).
Table 6.
Univariable risk ratios for diabetes remission among participants undergoing either RYGB or LAGB procedures based on changes from baseline across 3 or 7 years follow-up
| Changeb | RYGB | LAGB | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| N | RR | Lower | Upper | P-value | N | RR | Lower | Upper | P-value | ||
| Time since surgery | Per 5 years | 1667 | 0.93 | 0.91 | 0.96 | <0.0001 | 589 | 0.83 | 0.75 | 0.91 | <0.0001 |
| Weight | Per -10% | 1667 | 1.11 | 1.07 | 1.15 | <0.0001 | 589 | 1.32 | 1.16 | 1.49 | <0.0001 |
| Waist circumference | Per -10% | 1619 | 1.12 | 1.07 | 1.18 | <0.0001 | 569 | 1.47 | 1.23 | 1.76 | <0.0001 |
| Neck circumference | Per -5% | 1626 | 1.07 | 1.04 | 1.11 | <0.0001 | 570 | 1.21 | 1.09 | 1.33 | 0.0003 |
| Body fat, % | Per -10% | 1411 | 1.04 | 1.01 | 1.07 | 0.0056 | 518 | 1.11 | 0.98 | 1.27 | 0.11 |
| Insulin | Per -20% | 1603 | 1.00 | 0.99 | 1.01 | 0.95 | 565 | 1.05 | 1.01 | 1.09 | 0.0238 |
| HOMA IR | Per -15% | 1265 | 1.02 | 1.00 | 1.05 | 0.08 | 435 | 1.23 | 1.12 | 1.34 | <0.0001 |
| HOMA %B | Per -35% | 1265 | 0.97 | 0.96 | 0.99 | 0.0022 | 435 | 1.01 | 0.94 | 1.09 | 0.74 |
| hs-CRP | Per -20% | 1602 | 1.00 | 1.00 | 1.01 | 0.35 | 566 | 1.00 | 0.99 | 1.02 | 0.47 |
| ALTa | Per -20% | 1592 | 1.00 | 0.99 | 1.02 | 0.58 | 563 | 1.15 | 1.04 | 1.28 | 0.0062 |
| ASTa | Per -15% | 1592 | 1.01 | 0.99 | 1.03 | 0.32 | 563 | 1.02 | 0.91 | 1.15 | 0.70 |
| Cystatin-C | Per -10% | 1602 | 1.00 | 1.00 | 1.00 | 0.053 | 566 | 1.04 | 0.97 | 1.12 | 0.22 |
Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; HOMA-%B, homeostatic model assessment of beta cell function; HOMA IR, homeostatic model assessment of insulin resistance; hs-CPR, highly-sensitive C-reactive protein; LAGB, laparoscopic gastric banding; RR, relative risk; RYGB, Roux-en-Y gastric bypass.
a Indicates only 3 years of follow-up.
b All factors represent the change in that factor from baseline value, and all models were adjusted for baseline value.
Give that weight loss is greater following RYGB than LAGB, we evaluated the probability of achieving diabetes remission as a function of weight loss (or gain) following both procedures. Adjusting for the percentage of weight loss differences, the modeled probability of diabetes remission was greater for RYGB than LAGB at all follow-up time points, increasing from an adjusted relative risk (aRR) of 1.86 the first postoperative year to 3.96 after 7 years of follow-up (Table 7 and Fig. 2).
Table 7.
Adjusted aRR of diabetes remission for participants undergoing RYGB compared with patients who undergo LAGB by change in weight from baseline over 7 years postsurgical follow-up
| Visit | Number Remitted/Total with Baseline Diabetes | Percent Remitted | aRR | 95% CI for aRR | P-value | |
|---|---|---|---|---|---|---|
| Lower | Upper | |||||
| 12-month | 324/654 | 50% | 1.86 | 1.03 | 3.35 | 0.0408 |
| 24-month | 323/607 | 53% | 2.19 | 1.26 | 3.82 | 0.0055 |
| 36-month | 303/598 | 51% | 1.97 | 1.15 | 3.38 | 0.0132 |
| 48-month | 303/622 | 49% | 2.05 | 1.26 | 3.33 | 0.0041 |
| 60-month | 299/633 | 47% | 2.35 | 1.43 | 3.87 | 0.0008 |
| 84-month | 206/416 | 50% | 3.96 | 2.10 | 7.46 | <0.0001 |
Diabetes patients only. The relative risk for remission of diabetes mellitus was evaluated separately at each time point for a total of 6 models. Each model included the percentage of weight change at the given follow-up visit, baseline weight, procedure type, and a propensity score for the procedure type. The propensity score was estimated for all participants in a single logistic regression model that included age, sex, education, baseline BMI, the number of comorbid conditions at baseline, and the hospital site. Some participants were excluded due to incomplete data. Abbreviations: aRR, adjusted relative risk; BMI, body mass index; CI, confidence interval; LAGB, laparoscopic gastric banding; RYGB, Roux-en-Y gastric bypass.
Figure 2.
Modeled probabilities and 95% confidence intervals for diabetes remission for each postoperative year (labeled in months) of follow-up as a function of the percentage of weight loss in participants undergoing laparoscopic gastric banding (LAGB, orange lines) and Roux-en-Y gastric bypass (RYGB, blue lines). Adjusted relative risk (aRR) estimates and 95% confidence intervals for the association between surgical type (RYGB vs LAGB) and diabetes remission are adjusted for the percentage of weight change from baseline and the propensity score consisting of baseline demographic and clinical characteristics associated with type of bariatric surgical procedure. aRR is greater for RYGB than LAGB at each postoperative year. For actual aRR and P-values see Table 7.
Discussion
Utilizing a more expanded dataset than was previously available (1) and American Diabetes Association consensus group definitions for partial and complete remission (20), we found that total diabetes remission in participants with severe obesity peaked 2 years after surgery before declining slightly in both the RYBP and LAGB groups. Complete diabetes remission accounted for the majority of these peak remission events, reaching 81% of the total remission events after RYGB and 78% after LAGB.
Findings in this study extend observations we previously published on the likelihood of achieving diabetes remission after RYGB and LAGB from 3 to 7 years of follow-up (17). We not only show that the weight-loss independent likelihood of achieving diabetes remission remains greater after RYGB than after LAGB during extended follow-up, but that it continues to increase over time, from being nearly 2-fold more likely 3 years after surgery to being nearly 4-fold more likely 7 years after their surgery. This widening diabetes remission rate difference between procedures was not due to progressively greater numbers of participants achieving diabetes remission following RYGB than LAGB over time. Instead, peak diabetes remission rates occurred in the 2nd postoperative year and then declined progressively to year 7 in both groups. However, this rate of decline was less after RYGB (62.7% in year 2 down to 57.2% by year 7) than LAGB (29% in year 2 down to 22.5% by year 7).
Baseline participant characteristics associated with long-term diabetes remission (complete and partial remission combined) during 7 years of follow-up were similar to what we and others have reported previously (17, 24–27). These included younger age, shorter diabetes duration, better baseline control (lower HbA1c level), absence of the need for insulin treatment and/or use of fewer noninsulin diabetes medications, as well as biochemical indicators of greater insulin secretory capacity (higher C-peptide levels and HOMA-%B). Additionally, we found that treatment with lipid-lowering medications at baseline is associated with lower diabetes remission rates in both surgical groups. Interestingly, statins, which make up the majority of the drugs in this class, have been associated with increased risk for type 2 diabetes during long-term therapy (28, 29). Two other reports that have evaluated the association between baseline statin use and diabetes status in cohort studies of patients after bariatric surgery have shown either no relationship (30) or higher diabetes remission rates among those taking statins (31). As with any cohort analyses, however, it cannot be determined whether lipid-lowering medications (and more specifically statins) exerted independent (causative) effects on diabetes remission risk and, as such, these findings remain exploratory.
Risk factors of diabetes remission that were similar for both surgical groups included the greater passage of time since surgery (lower likelihood) and greater weight loss, as well as decreases of both waist and neck circumferences (all greater likelihoods). Weight and neck circumference measurements have been linked with insulin resistance directly (32–34) as well as by association with obstructive sleep apnea risk (35). However, improvements in estimated insulin sensitivity were associated with diabetes remission only after LAGB, whereas an improvement in estimated beta-cell response was significantly associated with diabetes remission only after RYGB. These findings extend our previous findings from to 3 years to 7 years and are in alignment with our hypothesis that weight-independent processes are occurring after RYGB that disproportionately lead to improvements in insulin secretory capacity and disposition index (18, 36).
A recent study that matched a small number of participants with type 2 diabetes who lost weight with a low-calorie diet versus RYGB (roughly 18% weight loss for each group) demonstrated no differences between these groups in diabetes remission status or need or diabetes medications, endogenous glucose production, glucose disposal, or meal-stimulated insulin secretion (37). Their results support the conclusion that metabolic benefits after RYGB are not weight independent, in contrast with our findings. Several differences in study population and design could explain these discrepancies. Our study included many more participants with slightly shorter diabetes duration and better glycemic control at baseline, representing a population who might have started off with a better likelihood of postoperative beta-cell responsiveness. Our primary outcome was diabetes remission under free-living conditions, as opposed to standardized experimental conditions that measure insulin sensitivity (hyperinsulinemic euglycemic clamp) and secretion, but which may fall short of replicating true physiological environments or the interplay of these 2 dynamic processes in the real world. In addition, our participants experienced a wider range of weight loss and the duration of follow-up was much longer, so we were able to explore both the variability and durability of their response as opposed to just 1 time point. Examination of the relative risk curves for each procedure (Fig. 2) reveal that the greatest differences in diabetes remission between the procedures occurred at lower weight-loss levels and converged (no difference) as weight losses approached 40% of baseline. Since improvement in insulin resistance, but much less so insulin secretion, tracks closely with the amount of weight lost (14, 36), weight loss approaching 40% may allow for enough improvement in insulin sensitivity that the residual beta-cell function is sufficient to establish glucose levels below the diabetes diagnosis threshold regardless of procedure.
Our study has several limitations. Participants were mostly women, mostly white, and on average heavier than nonsurgical weight loss studies in patients with type 2 diabetes (38), which limits the generalizability of our findings. This was a nonrandomized, observational cohort where participants and their surgeons decided which procedure they would undergo. However, despite differences in BMI and waist circumferences, the 2 surgical groups were remarkably well matched for baseline metabolic characteristics, especially those that are potentially important influencers of diabetes remission status. Participants were recruited at a time when sleeve gastrectomy was uncommon and estimates of insulin sensitivity and secretion were derived using fasting insulin and glucose levels and the nonrandomized HOMA2 calculator (39), which has been validated in nonsurgical cohorts (40) but not in those who have undergone bariatric procedures. Finally, the study was closed out midway through recruitment for the final study visit, resulting in missing data in just over 45% of participants for year 7. Modeled prevalence data was performed on missing versus recorded diabetes prevalence at each visit, and the data from year 7 was essentially unchanged from immediate previous visits, was not different by surgical procedure, and was determined to have sufficient statistical power for the main outcomes of weight change and prevalence of comorbid conditions (1). The strengths of our study include a standardized and comprehensive data compilation that was collected prospectively.
In conclusion, we confirm and extend our previous findings in participants with severe obesity out to 7 years after bariatric surgery, showing that the likelihood of achieving and sustaining both complete and partial diabetes remission in those with severe obesity is increased when bariatric surgery is performed on younger individuals, soon after diabetes diagnosis, those with better diabetes control achieved by fewer medications and no insulin, and those who undergo RYGB as opposed to LAGB. Our data also suggest that delaying bariatric surgery until patients have “failed” medical therapy is not optimal for patient care and should be reconsidered in current treatment guidelines/algorithms. However, this does not mean that a patient with poorly controlled diabetes or who requires insulin will not benefit from bariatric surgery. Rather, the benefit they might experience would come in the form of better glucose control while requiring fewer diabetes medications (including a lower insulin dose) (41) as opposed to complete or partial diabetes remission. Weight-independent benefits on glucose control and preservation of beta-cell insulin secretory capacity following RYGB are important areas for continued research and hold promise for the development of novel treatments for diabetes and its complications.
Acknowledgments
LABS personnel contributing to the study include: Columbia University Medical Center, New York, New York: Paul D. Berk, MD; Marc Bessler, MD; Amna Daud ; Harrison Lobdell IV; Jemela Mwelu ; Beth Schrope, MD, PhD; and Akuezunkpa Ude, MD; Cornell University Medical Center, New York, New York: Jamie Honohan BA; Michelle Capasso, BA; Ricardo Costa, BS; Greg Dakin, MD; Faith Ebel RD, MPH; Michel Gagner, MD; Jane Hsieh BS; Alfons Pomp, MD; and Gladys Strain, PhD; East Carolina Medical Center, Greenville, North Carolina: Rita Bowden, RN; William Chapman, MD, FACS; Blair Cundiff, BS; Mallory Ball, BS; Emily Cunningham, BA; Lynis Dohm, PhD; John Pender MD; and Walter Pories, MD, FACS; Neuropsychiatric Research Institute, Fargo, North Dakota: Jennifer Barker, MBA; Michael Howell, MD; Luis Garcia, MD, FACS, MBA; Kathy Lancaster, BA; Erika Lovaas, BS; James E. Mitchell, MD; and Tim Monson, MD; Oregon Health & Science University, Portland, Oregon: Chelsea Cassady, BS; Emily Coburn, MPH; Emily Moher, MPH; Clifford Deveney, MD; Katherine Elder, PhD; Stefanie Greene, Jonathan Purnell, MD; Robert O’Rourke, MD; Chad Sorenson ; and Bruce M. Wolfe, MD; Legacy Good Samaritan Hospital, Portland, Oregon: Emma Patterson, MD; William Raum, MD; Lisa VanDerWerff, PAC; and Jason Kwiatkowski, PAC; University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania: Anita P. Courcoulas, MD, MPH, FACS; William Gourash, MSN, CRNP; Carol A. McCloskey, MD; Ramesh Ramanathan, MD; Melissa Kalarchian PhD; Marsha Marcus PhD; Eleanor Shirley, MA; and Angela Turo, BS; University of Washington, Seattle, Washington: David R. Flum, MD, MPH; E. Patchen Dellinger, MD; Saurabh Khandelwal, MD; Skye D. Stewart, MS, CCRC; Morgan M. Cooley ; Rebecca Blissell ; and Megan J. Miller, MEd; Virginia Mason Medical Center, Seattle, Washington: Richard Thirlby, MD; Lily Chang, MD; Jeffrey Hunter, MD; Ravi Moonka, MD; and Debbie Ng, MPH, MA; Data Coordinating Center, Graduate School of Public Health at the University of Pittsburgh, Pittsburgh, Pennsylvania: Steven H. Belle, PhD, MScHyg; Wendy C. King, PhD; Debbie Martin, BA; Rocco Mercurio, MBA; Abdus Wahed, PhD; and Frani Averbach, MPH, RDN; National Institute of Diabetes and Digestive and Kidney Diseases: Mary Horlick, MD; Carolyn W. Miles, PhD; Myrlene A. Staten, MD; and Susan Z. Yanovski, MD; National Cancer Institute: David E. Kleiner, MD, PhD.
Financial Support: This clinical study was a cooperative agreement funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Grant numbers: DCC – U01 DK066557, Columbia – U01-DK66667 (in collaboration with Cornell University Medical Center CTRC, Grant UL1-RR024996); University of Washington – U01-DK66568 (in collaboration with CTRC, Grant M01RR-00037); Neuropsychiatric Research Institute – U01-DK66471; East Carolina University – U01-DK66526; University of Pittsburgh Medical Center – U01-DK66585 (in collaboration with CTRC, Grant UL1-RR024153); Oregon Health & Science University – U01-DK66555. Additional funding support came from R01 DK103842 (J.Q.P.).
Additional Information
Disclosure Summary: Dr Courcoulas reports grant from Allurion Inc, outside the submitted work. Dr Purnell reports consulting for Novo Nordisk. Drs Dewey, Laferrère, Selzer, Flum, Mitchell, Pomp, Pories, Inge, and Wolfe have no disclosures related to this work.
Data Availability
All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.
References
- 1. Courcoulas AP, King WC, Belle SH, et al. Seven-year weight trajectories and health outcomes in the longitudinal assessment of bariatric surgery (LABS) study. JAMA Surg. 2018;153(5):427–434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Maciejewski ML, Arterburn DE, Van Scoyoc L, et al. Bariatric surgery and long-term durability of weight loss. JAMA Surg. 2016;151(11):1046–1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Adams TD, Davidson LE, Litwin SE, et al. Weight and metabolic outcomes 12 years after gastric bypass. N Engl J Med. 2017;377(12):1143–1155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Sjöström L, Narbro K, Sjöström CD, et al. ; Swedish Obese Subjects Study . Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med. 2007;357(8):741-752. [DOI] [PubMed] [Google Scholar]
- 5. 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] [PubMed] [Google Scholar]
- 6. Mingrone G, Panunzi S, De Gaetano A, et al. Bariatric-metabolic surgery versus conventional medical treatment in obese patients with type 2 diabetes: 5 year follow-up of an open-label, single-centre, randomised controlled trial. Lancet. 2015;386(9997):964–973. [DOI] [PubMed] [Google Scholar]
- 7. Cummings DE, Arterburn DE, Westbrook EO, et al. Gastric bypass surgery vs intensive lifestyle and medical intervention for type 2 diabetes: the CROSSROADS randomised controlled trial. Diabetologia. 2016;59(5):945–953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Schauer PR, Bhatt DL, Kirwan JP, et al. ; STAMPEDE Investigators . Bariatric surgery versus intensive medical therapy for diabetes - 5-year outcomes. N Engl J Med. 2017;376(7):641–651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Simonson DC, Halperin F, Foster K, Vernon A, Goldfine AB. Clinical and patient-centered outcomes in obese patients with type 2 diabetes 3 years after randomization to Roux-en-Y gastric bypass surgery versus intensive lifestyle management: the SLIMM-T2D study. Diabetes Care. 2018;41(4):670–679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ikramuddin S, Korner J, Lee WJ, et al. Lifestyle intervention and medical management with vs without Roux-en-Y gastric bypass and control of hemoglobin A1c, LDL cholesterol, and systolic blood pressure at 5 years in the diabetes surgery study. JAMA. 2018;319(3):266–278. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Courcoulas AP, Gallagher JW, Neiberg RH, et al. Bariatric surgery vs lifestyle intervention for diabetes treatment: 5-year outcomes from a randomized trial. J Clin Endocrinol Metab. 2020;105(3):866–876. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sjöström L, Peltonen M, Jacobson P, et al. Association of bariatric surgery with long-term remission of type 2 diabetes and with microvascular and macrovascular complications. JAMA. 2014;311(22):2297–2304. [DOI] [PubMed] [Google Scholar]
- 13. Sjöholm K, Sjöström E, Carlsson LM, Peltonen M. Weight change-adjusted effects of gastric bypass surgery on glucose metabolism: 2- and 10-year results from the Swedish obese subjects (SOS) study. Diabetes Care. 2016;39(4):625–631. [DOI] [PubMed] [Google Scholar]
- 14. Nannipieri M, Mari A, Anselmino M, et al. The role of beta-cell function and insulin sensitivity in the remission of type 2 diabetes after gastric bypass surgery. J Clin Endocrinol Metab. 2011;96(9):E1372–E1379. [DOI] [PubMed] [Google Scholar]
- 15. Bradley D, Conte C, Mittendorfer B, et al. Gastric bypass and banding equally improve insulin sensitivity and β cell function. J Clin Invest. 2012;122(12):4667–4674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Batterham RL, Cummings DE. Mechanisms of diabetes improvement following bariatric/metabolic surgery. Diabetes Care. 2016;39(6):893–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Purnell JQ, Selzer F, Wahed AS, et al. Type 2 diabetes remission rates after laparoscopic gastric bypass and gastric banding: results of the longitudinal assessment of bariatric surgery study. Diabetes Care. 2016;39(7):1101–1107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Laferrère B, Pattou F. Weight-independent mechanisms of glucose control after Roux-en-Y gastric bypass. Front Endocrinol (Lausanne). 2018;9:530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. 10.5061/dryad.1zcrjdfqz. [DOI] [Google Scholar]
- 20. Buse JB, Caprio S, Cefalu WT, et al. How do we define cure of diabetes? Diabetes Care. 2009;32(11):2133–2135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Belle SH, Berk PD, Courcoulas AP, et al. ; LABS Consortium Weight Change Writing Group . Reporting weight change: standardized reporting accounting for baseline weight. Surg Obes Relat Dis. 2013;9(5):782–789. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care. 2004;27(6):1487–1495. [DOI] [PubMed] [Google Scholar]
- 23. Caumo A, Perseghin G, Brunani A, Luzi L. New insights on the simultaneous assessment of insulin sensitivity and beta-cell function with the HOMA2 method. Diabetes Care. 2006;29(12):2733–2734. [DOI] [PubMed] [Google Scholar]
- 24. Hamza N, Abbas MH, Darwish A, Shafeek Z, New J, Ammori BJ. Predictors of remission of type 2 diabetes mellitus after laparoscopic gastric banding and bypass. Surg Obes Relat Dis. 2011;7(6):691–696. [DOI] [PubMed] [Google Scholar]
- 25. Blackstone R, Bunt JC, Cortés MC, Sugerman HJ. Type 2 diabetes after gastric bypass: remission in five models using HbA1c, fasting blood glucose, and medication status. Surg Obes Relat Dis. 2012;8(5):548–555. [DOI] [PubMed] [Google Scholar]
- 26. Still CD, Wood GC, Benotti P, et al. Preoperative prediction of type 2 diabetes remission after Roux-en-Y gastric bypass surgery: a retrospective cohort study. Lancet Diabetes Endocrinol. 2014;2(1):38–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Aung L, Lee WJ, Chen SC, et al. Bariatric surgery for patients with early-onset vs late-onset type 2 diabetes. JAMA Surg. 2016;151(9):798–805. [DOI] [PubMed] [Google Scholar]
- 28. Preiss D, Seshasai SR, Welsh P, et al. Risk of incident diabetes with intensive-dose compared with moderate-dose statin therapy: a meta-analysis. JAMA. 2011;305(24):2556–2564. [DOI] [PubMed] [Google Scholar]
- 29. Ridker PM, Pradhan A, MacFadyen JG, Libby P, Glynn RJ. Cardiovascular benefits and diabetes risks of statin therapy in primary prevention: an analysis from the JUPITER trial. Lancet. 2012;380(9841):565–571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Perna M, Baker M, Byrne TK, Morgan K. Statins and the bariatric patient: characterization and perioperative effects of statin therapy in the gastric bypass patient. Am Surg. 2011;77(1):44–47. [PubMed] [Google Scholar]
- 31. Taylor BA, Ng J, Stone A, Thompson PD, Papasavas PK, Tishler DS. Effects of statin therapy on weight loss and diabetes in bariatric patients. Surg Obes Relat Dis. 2017;13(4):674–680. [DOI] [PubMed] [Google Scholar]
- 32. Bragg F, Tang K, Guo Y, et al. ; China Kadoorie Biobank (CKB) Collaborative Group . Associations of general and central adiposity with incident diabetes in Chinese men and women. Diabetes Care. 2018;41(3):494–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Carey VJ, Walters EE, Colditz GA, et al. Body fat distribution and risk of non-insulin-dependent diabetes mellitus in women. The Nurses’ Health Study. Am J Epidemiol. 1997;145(7):614–619. [DOI] [PubMed] [Google Scholar]
- 34. Karter AJ, Mayer-Davis EJ, Selby JV, et al. Insulin sensitivity and abdominal obesity in African-American, Hispanic, and non-Hispanic white men and women. The Insulin Resistance and Atherosclerosis Study. Diabetes. 1996;45(11):1547–1555. [DOI] [PubMed] [Google Scholar]
- 35. Onat A, Hergenç G, Yüksel H, et al. Neck circumference as a measure of central obesity: associations with metabolic syndrome and obstructive sleep apnea syndrome beyond waist circumference. Clin Nutr. 2009;28(1):46–51. [DOI] [PubMed] [Google Scholar]
- 36. Purnell JQ, Johnson GS, Wahed AS, et al. Prospective evaluation of insulin and incretin dynamics in obese adults with and without diabetes for 2 years after Roux-en-Y gastric bypass. Diabetologia. 2018;61(5):1142–1154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Yoshino M, Kayser BD, Yoshino J, et al. Effects of diet versus gastric bypass on metabolic function in diabetes. N Engl J Med. 2020;383(8):721–732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Wing RR, Bolin P, Brancati FL, et al. ; Look AHEAD Research Group . Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med. 2013;369(2):145–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Levy JC, Matthews DR, Hermans MP. Correct homeostasis model assessment (HOMA) evaluation uses the computer program. Diabetes Care. 1998;21(12):2191–2192. [DOI] [PubMed] [Google Scholar]
- 40. Pisprasert V, Ingram KH, Lopez-Davila MF, Munoz AJ, Garvey WT. Limitations in the use of indices using glucose and insulin levels to predict insulin sensitivity: impact of race and gender and superiority of the indices derived from oral glucose tolerance test in African Americans. Diabetes Care. 2013;36(4):845–853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Kadera BE, Lum K, Grant J, Pryor AD, Portenier DD, DeMaria EJ. Remission of type 2 diabetes after Roux-en-Y gastric bypass is associated with greater weight loss. Surg Obes Relat Dis. 2009;5(3):305–309. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.