OBJECTIVE

Bariatric surgery is an effective treatment option for individuals with obesity and type 2 diabetes (T2D). However, whether outcomes in subtypes of individuals at risk for T2D and/or comorbidities (Tübingen Clusters) differ, is unknown. Of these, cluster 5 (C5) and cluster 6 (C6) are high-risk clusters for developing T2D and/or comorbidities, while cluster 4 (C4) is a low-risk cluster. We investigated bariatric surgery outcomes, hypothesizing that high-risk clusters benefit most due to great potential for metabolic improvement.

RESEARCH DESIGN AND METHODS

We allocated participants without T2D but at risk for T2D, defined by elevated BMI, to the Tübingen Clusters. Participants had normal glucose regulation or prediabetes according to American Diabetes Association criteria. Two cohorts underwent bariatric surgery: a discovery (Lille, France) and a replication cohort (Rome, Italy). A control cohort (Tübingen, Germany) received behavioral modification counseling. Main outcomes included alteration of glucose regulation parameters and prediabetes remission.

RESULTS

In the discovery cohort, 15.0% of participants (n = 121) were allocated to C4, 22.3% (n = 180) to C5, and 62.4% (n = 503) to C6. Relative body weight loss was similar among all clusters; however, reduction of insulin resistance and improvement of β-cell function were strongest in C5. Prediabetes remission rate was lowest in low-risk C4 and highest in high-risk C5. Individuals from high-risk clusters changed to low-risk clusters in both bariatric surgery cohorts but not in the control cohort.

CONCLUSIONS

Participants in C5 had the highest benefit from bariatric surgery in terms of improvement in insulin resistance, β-cell function, and prediabetes remission. This novel classification might help identify individuals who will benefit specifically from bariatric surgery.

Worldwide, 2.5 billion adults are affected by overweight and obesity (1). Many of these individuals develop type 2 diabetes (T2D), which is among the leading causes of death globally (2). Most individuals have already established T2D-associated comorbidities, such as nephropathy or macrovascular disease, at the time of diagnosis (3). Therefore, it is of utmost importance to improve T2D prevention and treat individuals earlier in the course of metabolic disease, specifically in prediabetes, since lifetime T2D risk in people aged >35 years with prediabetes is >70% (4). Bariatric surgery is a well-established therapy to decelerate disease progression of obesity-associated sequelae, such as cardiovascular events, renal disease, or mortality, via body weight reduction. It has been shown to improve glycemic control in people with T2D and even promote T2D remission. A recent data-driven classification of individuals with T2D has defined five clusters that show different disease progression patterns and diverging risks of diabetes complications (5). Importantly, T2D clusters benefit from bariatric surgery to different extents and in differential manners (6). Participants from the severe insulin-resistant diabetes (SIRD) cluster benefited more from bariatric surgery in terms of both T2D remission and renal function compared with the remaining clusters.

However, complications such as nephropathy can occur even before the onset of T2D (3). Thus, it is important to further characterize and understand subphenotypes and treatment responses before T2D onset. In individuals at risk for T2D, defined by prediabetes, a history of gestational diabetes mellitus, familial risk for T2D, and/or elevated BMI, data-driven clusters differing in T2D risk and related complications have been identified. Clustering variables include anthropometrics, glucose and insulin measures from oral glucose tolerance tests (OGTTs), and fasting lipid levels (7). From these six clusters, cluster 4 (C4), cluster 5 (C5), and cluster 6 (C6) are linked with obesity, and cluster 3 (C3), C5, and C6 show a high risk for developing T2D and/or complications. C5 is characterized by high liver fat content and insulin resistance and high cardiovascular risk (“high liver fat content and insulin resistance-related cluster”). C6 shows a high nephropathy risk and high insulin secretion despite a relatively low risk to develop T2D (“nephropathy risk and high insulin secretion-related cluster”). C4, however, belongs to the low-risk clusters associated with severe obesity but with a low risk of developing T2D or related complications (“low risk obesity cluster”). An overview of these Tübingen Clusters is provided in Supplementary Fig. 1. Similar to individuals with T2D, individuals with prediabetes can achieve a reduction of insulin resistance via weight loss after bariatric surgery and may thereby significantly reduce their elevated T2D risk, as has been shown previously (8). While current guidelines recognize T2D as a comorbidity guiding clinical decision-making about bariatric surgery, prediabetes representing an independent risk factor for cardiovascular events, kidney disease, or even mortality is currently not recognized as a relevant obesity-associated comorbidity, partially due to its heterogeneity (2,9). To account for this heterogeneity and to evaluate therapeutic responses to bariatric surgery, participants with prediabetes were assigned to the Tübingen Clusters. In this study, we aimed to examine the cluster-specific impact of bariatric surgery on glucose regulation parameters, prediabetes remission, and cluster change in individuals with an elevated risk for T2D, specifically in two cohorts undergoing bariatric surgery and a control cohort (10). We hypothesized that high-risk clusters benefit most from bariatric surgery due to their potential to improve a previously deleterious metabolic state.

Study Design and Participants

In this multicohort study, we investigated the postoperative outcomes of novel data-driven subphenotypes of individuals without T2D but at risk for T2D (defined by elevated BMI >27 kg/m2) in two cohorts undergoing bariatric surgery: the A Biological Atlas of Severe Obesity (ABOS) cohort in Lille, France (ClinicalTrials.gov identifier: NCT01129297); the Bariatric Surgery and Reactive Hypoglycemia study cohort in Rome, Italy (ClinicalTrials.gov identifier: NCT01581801); and the Clinical and Metabolic Characterization of Long-Term Courses of Obesity Patients (AdipFollowup) cohort in Tübingen, Germany (ClinicalTrials.gov identifier: NCT04375371) as control. ABOS participants were followed up after 1 and 2 years, Rome cohort participants were followed up for a mean (SD) of 15.3 (4.5) months, and control cohort participants were followed up for ∼10 years (mean [SD] 128.9 [30.1] months).

ABOS is an ongoing prospective study that aims to identify determinants of outcomes of bariatric surgery. ABOS participants without T2D (n = 806) who underwent Roux-en-Y gastric bypass, sleeve gastrectomy, or gastric banding between 1 January 2006 and 12 December 2017 were included in the current study (Supplementary Fig. 2). Participant data were prospectively collected at the time of surgery and 1 and 2 years after surgery. A 75-g OGTT was performed at baseline and at follow-up. A description of the laboratory assessments has been published previously (11).

For the analysis of prediabetes remission, all individuals with prediabetes at baseline were included (n = 423). Prediabetes status was defined at baseline based on a fasting plasma glucose (PG) of 100–125 mg/dL (5.6–6.9 mmol/L) and/or a 2-h PG of 140–199 mg/dL (7.8–11.0 mmol/L) during OGTT and/or an HbA1c of 5.7–6.4% (39–46 mmol/mol) according to American Diabetes Association recommendations (12). Glucose area under the curve (AUC) during OGTT was determined using the trapezoidal rule (AUCGlucose 0–120 min) (13). Peripheral insulin sensitivity was estimated by the modified Matsuda index (ISIMatsmod) according to the following equation: 10,000 / √([glucose0 min × insulin 0 min] × [(glucose0 min + glucose30 min + glucose120 min) / 3] × [(insulin0 min + insulin30 min + insulin120 min) / 3]) (14,15). Insulin resistance was assessed by the HOMA of insulin resistance (HOMA-IR), which was calculated according to the following equation: insulin0 min × glucose0 min / 22.5 (16). The disposition index as a measure of β-cell function was calculated as the product of the C-peptidogenic index and Matsuda index (17) and the C-peptide/glucose AUC using the trapezoidal rule as a proxy for insulin secretion (AUCC-peptide 0–30 min / AUCGlucose 0–30 min) (18).

The independent replication cohort from Rome consisted of 60 individuals with obesity without T2D who were randomly assigned 1:1 to either Roux-en-Y gastric bypass or sleeve gastrectomy at the Catholic University School of Medicine in Rome between December 2012 and December 2014 (19). A description of the analytic procedures of samples has been published previously (19).

The independent control cohort from Tübingen consisted of 46 individuals with obesity who received behavioral modification counseling for body weight reduction. This included 10 group sessions with nutrition, physical activity, and lifestyle counseling over 6 months. A detailed description of the laboratory assessments has been published elsewhere (20). Individuals of the control cohort were retrospectively contacted for rephenotyping between 27 January and 23 October 2020.

The studies were reviewed and approved by the regional human ethics committees (Lille: Comité de Protection des Personnes Nord Ouest VI; Rome: Rome Catholic University Ethical Committee; Tübingen: Ethics Committee at the Eberhard-Karls University of Tübingen) in accordance with national guidelines and the provisions of the Declaration of Helsinki as revised in 2000. All participants provided written informed consent to participate in the respective studies.

Clustering

Clinical clustering variables of the Tübingen Clusters were BMI, hip and waist circumference, fasting PG and insulin, 2-h PG and insulin, fasting triglycerides, and HDL cholesterol (HDL-C) levels (7). The Tübingen Clusters were named after the first cohort in which they were described (Tübingen Family Study) (7,21). An online application for personal use or research purposes is accessible at https://prediabclusters.idm-tuebingen.org.

Participants were assigned to C3, C4, C5, or C6 at baseline. Owing to the low number of participants assigned to C3 (n = 2 each in ABOS and the control cohort), it was excluded from further analysis.

Surgery

All bariatric surgery procedures were done laparoscopically, as described previously (6).

Outcomes

Prediabetes remission was defined according to current American Diabetes Association criteria for normal glucose regulation, as below the described cutoffs for prediabetes, and without the use of glucose-lowering drugs (8,10,12). Chronic kidney disease (CKD) was assessed based on the estimated glomerular filtration rate (eGFR) calculated according to the MDRD equation (22). The general cardiovascular risk profile was estimated according to the Framingham sex-specific multivariable risk algorithm (23). Liver biopsies were done as previously described (24). The Nonalcoholic Fatty Liver Disease (NAFLD) Activity Score was defined as the unweighted summed scores for steatosis (0–3), lobular inflammation (0–3), and ballooning (25). Noninvasive tests (fatty liver index, AST-to-platelet ratio index [APRI], and NAFLD Fibrosis Score) were computed as described previously (24).

Statistical Analysis

Statistical analyses were performed using the R Version 4.2.2 software. Data are presented as mean (95% CI), median (interquartile range), or n (%) unless otherwise specified. Between-group comparisons over time were analyzed using linear mixed-effects models, with participant as a random effect using the lme4 package, or cross-sectionally using two-way ANOVA with a post hoc test for multiple comparisons (least significant difference), applying Bonferroni correction, or Wilcoxon signed rank or χ2 test, as appropriate. The model included cluster, time point, and the interaction between the two as model terms, and main outcomes were evaluated in models with BMI, age, sex, and type of surgery as fixed effects and in the case of insulin secretion, with insulin sensitivity.

Data and Resource Availability

The data sets generated or analyzed during the current study are not publicly available since they are subject to national data protection laws and restrictions imposed by the ethics committees to ensure data privacy of study participants. They can be applied for through an individual project agreement with the principal investigator of the respective university hospital.

In all cohorts, the most abundant cluster was high-risk C6 (Fig. 1). Participant anthropometric and metabolic characteristics of ABOS at baseline are summarized in Table 1, and those of the remaining cohorts are combined in Supplementary Table 1. Low-risk C4 participants had lower BMI, liver fat content, triglycerides, insulin resistance, insulin secretion, and glycemia and higher HDL-C compared with C5 and C6 participants at baseline.

Figure 1

Tübingen Clusters distribution at baseline and change of cluster assignment over time in the ABOS cohort (A), Rome cohort (B), and control cohort (C).

Figure 1

Tübingen Clusters distribution at baseline and change of cluster assignment over time in the ABOS cohort (A), Rome cohort (B), and control cohort (C).

Close modal
Table 1

Baseline characteristics of ABOS cohort

CharacteristicC4 (n = 121)C5 (n = 180)C6 (n = 503)P
Age (years)     
 Mean (SD) 37.5 (10.6) 41.1 (11.6) 36.9 (11.3) <0.001 
 Median (IQR) 37.0 (15.0) 41.0 (17.3) 35.0 (18.0)  
Sex     
 Female 103 (85.1%) 152 (84.4%) 389 (77.3%) 0.0396 
 Male 18 (14.9%) 28 (15.6%) 114 (22.7%)  
BMI (kg/m2    
 Mean (SD) 42.7 (5.2) 46.4 (6.6) 47.5 (7.6) <0.001 
 Median (IQR) 41.3 (4.5) 45.2 (7.8) 45.7 (9.0)  
Waist-to-hip ratio     
 Mean (SD) 0.9 (0.1) 0.9 (0.1) 0.9 (0.1) <0.001 
 Median (IQR) 0.9 (0.1) 0.9 (0.1) 0.9 (0.1)  
eGFR by MDRD (mL/min)     
 Mean (SD) 100.3 (21.6) 95.9 (20.2) 101.0 (21.3) 0.0195 
 Median (IQR) 97.2 (26.7) 95.3 (24.3) 98.3 (24.7)  
Liver fat content (%)     
 Mean (SD) 14.8 (22.6) 28.8 (24.9) 21.6 (21.9) <0.001 
 Median (IQR) 5.0 (19.0) 20.0 (35.0) 15.0 (26.0)  
Type of surgery     
 Gastric banding 36 (29.8%) 34 (18.9%) 112 (22.3%) 0.0546 
 Gastric bypass 67 (55.4%) 105 (58.3%) 264 (52.5%)  
 Sleeve gastrectomy 18 (14.9%) 41 (22.8%) 127 (25.2%)  
Triglycerides (mmol/L)     
 Mean (SD) 1.1 (0.5) 1.9 (0.9) 1.3 (0.5) <0.001 
 Median (IQR) 1.0 (0.6) 1.7 (0.9) 1.2 (0.6)  
HDL-C (mmol/L)     
 Mean (SD) 1.3 (0.3) 1.1 (0.2) 1.2 (0.2) <0.001 
 Median (IQR) 1.2 (0.4) 1.0 (0.3) 1.2 (0.3)  
HOMA-IR     
 Mean (SD) 1.7 (0.6) 5.0 (3.0) 3.8 (2.0) <0.001 
 Median (IQR) 1.7 (0.9) 4.4 (2.8) 3.3 (2.2)  
HOMA of β-cell function     
 Mean (SD) 108.3 (77.5) 184.0 (112.1) 213.4 (118.7) <0.001 
 Median (IQR) 98.7 (52.2) 155.2 (114.9) 186.5 (130.4)  
Glucose AUC     
 Mean (SD) 780.1 (113.2) 1061.0 (107.6) 862.0 (109.4) <0.001 
 Median (IQR) 784.6 (116.0) 1067.2 (132.5) 865.5 (139.9)  
Fasting PG (mmol/L)     
 Mean (SD) 5.0 (0.4) 5.8 (0.6) 5.2 (0.5) <0.001 
 Median (IQR) 5.1 (0.5) 5.8 (0.9) 5.2 (0.7)  
2-h PG (mmol/L)     
 Mean (SD) 5.2 (1.1) 8.6 (1.2) 6.2 (1.2) <0.001 
 Median (IQR) 5.1 (1.4) 8.5 (1.5) 6.2 (1.6)  
Fasting plasma insulin (mU/L)     
 Mean (SD) 7.5 (2.6) 19.6 (11.0) 16.5 (8.1) <0.001 
 Median (IQR) 7.6 (3.6) 17.3 (12.0) 14.7 (9.1)  
2-h Plasma insulin (mU/L)     
 Mean (SD) 21.9 (14.2) 133.0 (104.0) 74.9 (62.1) <0.001 
 Median (IQR) 18.2 (16.0) 101.2 (104.5) 62.2 (57.9)  
Fasting c-peptide (ng/mL)     
 Mean (SD) 2.9 (1.9) 4.0 (1.1) 3.7 (1.1) <0.001 
 Median (IQR) 2.7 (1.0) 3.8 (1.3) 3.5 (1.3)  
2-h C-peptide (ng/mL)     
 Mean (SD) 7.3 (4.7) 12.9 (3.9) 10.2 (3.2) <0.001 
 Median (IQR) 6.5 (3.0) 12.7 (4.1) 10.2 (4.2)  
 HbA1c (%)     
 Mean (SD) 5.5 (0.4) 5.7 (0.4) 5.6 (0.4) <0.001 
 Median (IQR) 5.5 (0.5) 5.8 (0.5) 5.6 (0.5)  
CharacteristicC4 (n = 121)C5 (n = 180)C6 (n = 503)P
Age (years)     
 Mean (SD) 37.5 (10.6) 41.1 (11.6) 36.9 (11.3) <0.001 
 Median (IQR) 37.0 (15.0) 41.0 (17.3) 35.0 (18.0)  
Sex     
 Female 103 (85.1%) 152 (84.4%) 389 (77.3%) 0.0396 
 Male 18 (14.9%) 28 (15.6%) 114 (22.7%)  
BMI (kg/m2    
 Mean (SD) 42.7 (5.2) 46.4 (6.6) 47.5 (7.6) <0.001 
 Median (IQR) 41.3 (4.5) 45.2 (7.8) 45.7 (9.0)  
Waist-to-hip ratio     
 Mean (SD) 0.9 (0.1) 0.9 (0.1) 0.9 (0.1) <0.001 
 Median (IQR) 0.9 (0.1) 0.9 (0.1) 0.9 (0.1)  
eGFR by MDRD (mL/min)     
 Mean (SD) 100.3 (21.6) 95.9 (20.2) 101.0 (21.3) 0.0195 
 Median (IQR) 97.2 (26.7) 95.3 (24.3) 98.3 (24.7)  
Liver fat content (%)     
 Mean (SD) 14.8 (22.6) 28.8 (24.9) 21.6 (21.9) <0.001 
 Median (IQR) 5.0 (19.0) 20.0 (35.0) 15.0 (26.0)  
Type of surgery     
 Gastric banding 36 (29.8%) 34 (18.9%) 112 (22.3%) 0.0546 
 Gastric bypass 67 (55.4%) 105 (58.3%) 264 (52.5%)  
 Sleeve gastrectomy 18 (14.9%) 41 (22.8%) 127 (25.2%)  
Triglycerides (mmol/L)     
 Mean (SD) 1.1 (0.5) 1.9 (0.9) 1.3 (0.5) <0.001 
 Median (IQR) 1.0 (0.6) 1.7 (0.9) 1.2 (0.6)  
HDL-C (mmol/L)     
 Mean (SD) 1.3 (0.3) 1.1 (0.2) 1.2 (0.2) <0.001 
 Median (IQR) 1.2 (0.4) 1.0 (0.3) 1.2 (0.3)  
HOMA-IR     
 Mean (SD) 1.7 (0.6) 5.0 (3.0) 3.8 (2.0) <0.001 
 Median (IQR) 1.7 (0.9) 4.4 (2.8) 3.3 (2.2)  
HOMA of β-cell function     
 Mean (SD) 108.3 (77.5) 184.0 (112.1) 213.4 (118.7) <0.001 
 Median (IQR) 98.7 (52.2) 155.2 (114.9) 186.5 (130.4)  
Glucose AUC     
 Mean (SD) 780.1 (113.2) 1061.0 (107.6) 862.0 (109.4) <0.001 
 Median (IQR) 784.6 (116.0) 1067.2 (132.5) 865.5 (139.9)  
Fasting PG (mmol/L)     
 Mean (SD) 5.0 (0.4) 5.8 (0.6) 5.2 (0.5) <0.001 
 Median (IQR) 5.1 (0.5) 5.8 (0.9) 5.2 (0.7)  
2-h PG (mmol/L)     
 Mean (SD) 5.2 (1.1) 8.6 (1.2) 6.2 (1.2) <0.001 
 Median (IQR) 5.1 (1.4) 8.5 (1.5) 6.2 (1.6)  
Fasting plasma insulin (mU/L)     
 Mean (SD) 7.5 (2.6) 19.6 (11.0) 16.5 (8.1) <0.001 
 Median (IQR) 7.6 (3.6) 17.3 (12.0) 14.7 (9.1)  
2-h Plasma insulin (mU/L)     
 Mean (SD) 21.9 (14.2) 133.0 (104.0) 74.9 (62.1) <0.001 
 Median (IQR) 18.2 (16.0) 101.2 (104.5) 62.2 (57.9)  
Fasting c-peptide (ng/mL)     
 Mean (SD) 2.9 (1.9) 4.0 (1.1) 3.7 (1.1) <0.001 
 Median (IQR) 2.7 (1.0) 3.8 (1.3) 3.5 (1.3)  
2-h C-peptide (ng/mL)     
 Mean (SD) 7.3 (4.7) 12.9 (3.9) 10.2 (3.2) <0.001 
 Median (IQR) 6.5 (3.0) 12.7 (4.1) 10.2 (4.2)  
 HbA1c (%)     
 Mean (SD) 5.5 (0.4) 5.7 (0.4) 5.6 (0.4) <0.001 
 Median (IQR) 5.5 (0.5) 5.8 (0.5) 5.6 (0.5)  

IQR, interquartile range.

Since Tübingen Clusters depend on modifiable metabolic measures, we hypothesized that bariatric surgery would lead to reassignment of high-risk clusters to low-risk clusters. In the ABOS and Rome cohorts, bariatric surgery led to a switch from high-risk to low-risk clusters in most participants (Fig. 1A and B). Equal fractions of C5 and C6 participants remained in high-risk clusters in ABOS, but nearly all C5 participants who did not convert to a low-risk cluster converted to C6 (Fig. 1A). Finally, in the control cohort, most participants stayed in high-risk clusters, and nearly one-half of C4 participants converted to C6 after long-term follow-up (Fig. 1C).

Next, we investigated cluster-specific anthropometric outcomes of bariatric surgery. One year after surgery, C5 and C6 participants had a higher BMI than C4 participants (mean [SD]: C5 33.2 [6.2] vs. C6 33.6 [6.2] [P > 0.99] and C4 30.6 [5.6] [each P vs. C5 and C6 <0.001]), while relative body weight loss was similar among all clusters (mean [SD]: C4 27.9% [11.0%] vs. C5 27.7% [10.9%] vs. C6 27.9% [11.1%]; P = 0.81) (Fig. 2A).

Figure 2

Weight and glucose regulation trajectories in ABOS (Lille) cohort. Percent change of the following parameters: weight loss (A), AUCGlucose 0–120 min (B), HOMA-IR (C), Matsuda index (D), disposition index (E), and AUCC-peptide 0–30 min / AUCGlucose 0–30 min (F). PG and insulin trajectories over OGTT at baseline (G and H) and after 1 year (I and J). Data are mean (95% CI). Analyses were performed using two-way ANOVA or Wilcoxon signed rank test, as applicable. *P < 0.05, **P < 0.01, ***P < 0.001. Color of asterisks indicates comparison cluster (GJ).

Figure 2

Weight and glucose regulation trajectories in ABOS (Lille) cohort. Percent change of the following parameters: weight loss (A), AUCGlucose 0–120 min (B), HOMA-IR (C), Matsuda index (D), disposition index (E), and AUCC-peptide 0–30 min / AUCGlucose 0–30 min (F). PG and insulin trajectories over OGTT at baseline (G and H) and after 1 year (I and J). Data are mean (95% CI). Analyses were performed using two-way ANOVA or Wilcoxon signed rank test, as applicable. *P < 0.05, **P < 0.01, ***P < 0.001. Color of asterisks indicates comparison cluster (GJ).

Close modal

We further examined parameters of glucose regulation and insulin sensitivity. Initially, AUCGlucose 0–120 min was highest in C5 and C6 (Table 1) and reduced in both clusters after surgery, with the most pronounced relative reduction in C5 versus the other clusters (Fig. 2B). AUCGlucose 0–120 min was lowest in C4 (Table 1), but AUCGlucose 0–120 min did not change significantly 1 year after bariatric surgery in this cluster (mean [SD]: 761 [155] min × mmol/L; P = 0.9). PG levels after bariatric surgery decreased to similar values between C6 and C4 but remained slightly higher in C5 versus C4 and C6 (Fig. 2G and I). As expected based on cluster characteristics, HOMA-IR was highest in C5 and lowest in C4 (P for each comparison <0.001) (Table 1). After bariatric surgery, C5 achieved the most pronounced reductions in HOMA-IR, with both C5 and C6 exhibiting stronger reductions in HOMA-IR than C4 (ΔHOMA-IR: C4 −29.52% vs. C5 −53.82% vs. C6 −53.00%; P C4 vs. C5 or C6 <0.001, P C5 vs. C6 >0.99) (Fig. 2C). Similarly, ISIMatsmod increased more strongly in C5 and C6 after bariatric surgery (ΔISIMatsmod: C4 66.44% vs. C5 227.88% vs. C6 150.80%; P for each comparison <0.001) (Fig. 2D). However, HOMA-IR and ISIMatsmod did not change significantly in C4 (Fig. 2C and D). The disposition index was highest in C4 and C6 at baseline, while C4 was the only cluster not to show a significant increase in β-cell function after bariatric surgery (mean [SD]: C4 104.93% [274.70%] vs. C5 254.73% [311.10%] vs. C6 133.52% [394.26%]; C4 vs. C5 P = 0.03, C4 vs. C6 P = 0.62, C5 vs. C6 P = 0.01) (Fig. 2E). AUCC-peptide 0–30 min / AUCGlucose 0–30 min was highest in C6 and lowest in C5 and C4 (Supplementary Fig. 6D). Only C5 increased insulin sensitivity–adjusted insulin secretion after bariatric surgery, which was not observed in C4 and C6, while the latter was characterized by high insulin secretion (Fig. 2F, Supplementary Fig. 5F, and Supplementary Fig. 6D). Changes in the remaining glucose regulation parameters of the Rome cohort are shown in Supplementary Fig. 5.

We next analyzed glucose regulation trajectories of C4, C5, and C6. Changes in glucose regulation status 1 year after bariatric surgery are shown in Fig. 3A–C. Despite exhibiting the lowest prediabetes prevalence, C4 had the lowest prediabetes remission rate of all participants who met prediabetes criteria at baseline, while C5 and C6 had the highest remission rate 1 year (C4 60% vs. C5 77% [P = 0.045]; C4 vs. C6 74% [P = 0.09]) (Fig. 3D) and 2 years (C4 61% vs C5 82% [P = 0.007]; C4 vs. C6 79% [P = 0.047]) (Fig. 3E) after bariatric surgery.

Figure 3

Glucose regulation trajectories, prediabetes remission, and glucose regulation indices by response in ABOS cohort. Glucose regulation status at baseline and 1 and 2 years after surgery in C4 (A), C5 (B), and C6 (C). Prediabetes remission after 1 year (D) and 2 years (E), weight (F), HOMA-IR (G), Matsuda index (H), HOMA of β-cell function (HOMA-B) (I), disposition index (J), and AUCC-peptide 0–30 min / AUCGlucose 0–30 min (K). This multivariable mixed-effects linear model included age, sex, type of surgery, time point, cluster, and the interaction between cluster and time point as main effects. Data are mean (95% CI). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3

Glucose regulation trajectories, prediabetes remission, and glucose regulation indices by response in ABOS cohort. Glucose regulation status at baseline and 1 and 2 years after surgery in C4 (A), C5 (B), and C6 (C). Prediabetes remission after 1 year (D) and 2 years (E), weight (F), HOMA-IR (G), Matsuda index (H), HOMA of β-cell function (HOMA-B) (I), disposition index (J), and AUCC-peptide 0–30 min / AUCGlucose 0–30 min (K). This multivariable mixed-effects linear model included age, sex, type of surgery, time point, cluster, and the interaction between cluster and time point as main effects. Data are mean (95% CI). *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal

To further understand the underlying mechanisms contributing to prediabetes remission, we investigated anthropometric and metabolic parameters by prediabetes remission status (i.e., responder vs. nonresponder). In all clusters, responders had more body weight loss than nonresponders (mean [SD] Δbody weight: C4 responders 35 [5] kg vs. nonresponders 29 [8] kg [P group over time <0.05]; C5 responders 38 [3] kg vs. nonresponders 25 [6] kg [P group over time <0.001]; C6 responders 42 [2.5] kg vs. nonresponders 26 [5] kg [P group over time <0.001]) (Fig. 3F), indicating that weight loss is important for prediabetes remission in all clusters. Insulin resistance and sensitivity (HOMA-IR and ISIMatsmod) showed a stronger improvement in C6 responders than in nonresponders (Fig. 3G and H); however, improvement did not differ between C5 and C4 nonresponders and responders. Furthermore, β-cell function improved more strongly in C5 responders than nonresponders but was not different between C4 responders and nonresponders (disposition index: C4 responders vs. nonresponders P group over time = 0.07; C5 responders vs. nonresponders P group over time = 0.02; C6 responders vs. nonresponders P group over time = 0.33) (Fig. 3J). AUCC-peptide 0–30 min/AUCGlucose 0–30 min increased only in C5 responders.

Next, since glucose levels are critically regulated by the liver and partly depend on hepatic lipid content, local inflammation, and fibrosis, we analyzed in a subgroup of participants who underwent initial and rebiopsy of the liver 1 year after bariatric surgery (n = 104). Here, C5 showed the highest liver steatosis severity and NAFLD Activity Score but not Kleiner Liver Fibrosis Score at baseline (liver steatosis severity: C4 vs. C5 P < 0.001; C4 vs. C6 P < 0.001; C5 vs. C6 P = 0.003) (Supplementary Fig. 3AC). Most C5 and C6 participants achieved a liver fat percentage reduction into the normal or near-normal range (Supplementary Fig. 3D). C4 had a liver fat content corresponding to grade 1 macroscopic steatosis, which decreased by trend into the normal range. These findings were similar when assessed by noninvasive tests for steatosis and fibrosis (Supplementary Fig. 3EG). Scores for advanced metabolic dysfunction–associated steatotic liver disease (MASLD), such as the APRI and Fibrosis-4 (FIB-4), were significantly reduced in C6 nonresponders, while the FIB-4 score increased in C5 and C6 responders (Supplementary Fig. 4AC).

The highest prediabetes remission and the high conversion rates from high- to low-risk clusters in C5 were also accompanied by the strongest reduction in relative Framingham Risk Score (rFRS), although C5 participants were older (Supplementary Fig. 6A). Furthermore, C5 and C6 participants had a slight increase in eGFR (Supplementary Fig. 6B). Trajectories of insulin sensitivity and secretion of ABOS are shown in Supplementary Fig. 6C and D, and glucose regulation trajectories and prediabetes remission rates of the Rome and control cohorts in Supplementary Fig. 7.

In this study, we show that novel, data-driven clusters of individuals at risk for T2D (Tübingen Clusters) differ in their response to bariatric surgery. As body weight loss was similar among all clusters, differences in the alteration of glucose regulation parameters were independent of differences in relative body weight loss. While high-risk C5 and C6 ameliorated both insulin resistance and β-cell function, specifically low-risk C4 participants with prediabetes did not benefit from bariatric surgery to the same extent, which is demonstrated by their lower prediabetes remission rate. C4 achieved a moderate improvement in insulin resistance (despite basal HOMA-IR being in the normal range), but neither C4 responders nor nonresponders had improved insulin secretion or β-cell function after bariatric surgery. Whether this was due to, for example, a lack of improved β-cell sensitivity to incretins or changes in hepatic VLDL-palmitate export remains to be demonstrated (26,27). However, metabolic dysfunction was less severe in C4 already before surgery and may reflect a state of metabolically healthy obesity (28). Nonetheless, since prediabetes remission via weight loss is beneficial in terms of T2D risk reduction and potential complications (8), C4 individuals with prediabetes may benefit from prediabetes remission despite the overall less severe metabolic dysfunction of the whole cluster. C5 benefited most from bariatric surgery in terms of improvement in insulin resistance, prediabetes remission, and cardiovascular risk as reflected by the rFRS. Despite that overall prediabetes remission rates were lower than expected after bariatric surgery, participants achieving prediabetes remission primarily had improved insulin sensitivity in C6 and insulin secretion in C5. Thus, for C6 individuals, improving insulin sensitivity in light of already high insulin secretion is sufficient to achieve prediabetes remission. Vice versa, increasing insulin secretion appears to be a key mechanistic underpinning for those achieving remission in C5. Considering the slightly lower insulin secretion in C5 versus C6 at baseline, we cannot rule out that cluster definition may be associated with this outcome.

Hepatic phenotypes of Tübingen Clusters previously assessed by 1H magnetic resonance spectroscopy are validated here for the first time via liver biopsies (7). We observed that C5 participants indeed had the highest liver fat content, followed by C6, which reduced into the near-normal range after bariatric surgery. The differential response to bariatric surgery may result from stronger metabolic perturbances in C5 at baseline combined with the high effectiveness of bariatric surgery in improving MASLD, as shown to be associated with higher remission of T2D after bariatric surgery (29). This reduction in liver fat may be associated with an increase in insulin secretion, as has been demonstrated previously (30–32). Importantly, the noninvasive fatty liver index shows a similar pattern as liver steatosis determined by liver biopsy. However, liver fibrosis assessed both by liver biopsy (Kleiner Liver Fibrosis Score) and noninvasively (APRI and NAFLD Fibrosis Score) did not differ among clusters, reflecting the relatively low prevalence of liver fibrosis in this cohort. Additionally, noninvasive tests for hepatic fibrosis in MASLD may not be suitable for adequately reflecting an elevated MASLD risk after bariatric surgery (e.g., increasing FIB-4 scores in C5 and C6 responders) (24).

Weight loss was higher in individuals achieving prediabetes remission in all clusters, indicating that weight loss can mediate prediabetes remission, which has previously been demonstrated for lifestyle intervention (8). Although weight loss mediated prediabetes remission at least in part in all clusters, mechanisms resulting in prediabetes remission differed among clusters. Specifically, in C4, insulin resistance only improved marginally, while in C6, change in insulin sensitivity was a discriminator between response and nonresponse. This was not the case in C5. Furthermore, neither C4 nonresponders nor responders significantly increased insulin secretion or β-cell function, while in C5 and C6, responders increased β-cell function in particular. As the prediabetic state delineates a higher risk for T2D, the primary aim, besides weight loss, should be remission of prediabetes as one of the most effective ways to reduce T2D risk (8,10). As shown here, prediabetes remission rates after bariatric surgery differed between low-risk and high-risk clusters. This is surprising since C4 participants were younger compared with C5 and younger age has been associated with higher prediabetes remission rates after bariatric surgery (33). Overall remission rates of prediabetes were similar to those observed after bariatric surgery in individuals with overt T2D (34). Since in prediabetes, metabolic derangements are not as severe as in T2D, a higher feasibility of prediabetes remission compared with T2D remission could be expected. However, similar remission rates suggest that prediabetes remission is as difficult to achieve as T2D remission. Similar to our findings, evidence on bariatric surgery–induced T2D remission in T2D subphenotypes showed that bariatric surgery induced a lower remission rate in low-risk mild obesity–related diabetes compared with the high-risk cluster SIRD (6). These findings indicate that other mechanisms apart from weight loss, which was similar among clusters in our study, might play a role in improving glucose regulation in C4. Specific characteristics of C4 that cannot or can only partly be improved by bariatric surgery may be decisive for this. As C4 participants had near-normal HOMA-IR values at baseline, insulin resistance may not have been the main driver of the prediabetic state in this cluster. Even though C4 participants had a slightly better β-cell function at baseline compared with C5 and C6 participants, a higher proportion in C4 still did not manage to gain further improvements in β-cell function as opposed to C5 and C6. Individuals with severe insulin deficient diabetes have been shown to have the lowest diabetes remission rate compared with mild obesity–related diabetes and SIRD (6). In line with this, lacking the ability to increase β-cell function could prevent C4 individuals from returning to normal glucose regulation (35). In previous studies by our group, C4 participants did not have a specific genetic risk of β-cell dysfunction (7). Since first-phase insulin secretion after bariatric surgery is also orchestrated by release of glucagon-like peptide 1, C4 might not promote or even be able to increase glucagon-like peptide 1 secretion after bariatric surgery as strongly as the other clusters (36). Therefore, further studies examining incretin responses after bariatric surgery between subphenotypes are needed to investigate whether tailored treatment (e.g., with incretin-based medication) might be a more effective treatment option in terms of prediabetes remission, specifically in C4 (37).

After short-term follow-up, most participants converted from high-risk to low-risk clusters, while those from low-risk clusters stayed in low-risk clusters, as expected. This cluster change is similar in both cohorts undergoing bariatric surgery (ABOS and Rome). However, in the control cohort, most participants changed to high-risk clusters over time, imposing an increasing metabolic risk without surgical intervention. This study is the first in our knowledge to show that individuals change clusters after bariatric surgery, which may reflect the reduced cardiometabolic risk upon bariatric surgery assessed by a reduction of the rFRS. Interestingly, rFRS was reduced to similar levels among all three clusters, although C5 had the highest rFRS before surgery. This is particularly important since C5 participants were the oldest, with age being part of the bariatric surgery–independent variables of the rFRS. Furthermore, both C5 and C6 participants increased kidney function represented by eGFR, which might reflect the kidney-protective effect of bariatric surgery specifically in these clusters and may not be the case for C4. After long-term follow-up in participants who did not undergo bariatric surgery, a switch from a high- to a low-risk cluster was rare, and many of the participants with former low risk converted to high-risk clusters. Thereby, reassignment to Tübingen Clusters could help reassess risk for T2D and complications after bariatric surgery and may guide therapeutic approaches postoperatively.

Our study has some limitations. First, most study participants were White Europeans, which may limit generalizability. Additionally, this study is the first to show cluster reassignment after bariatric surgery, which could be affected by alterations in gastrointestinal glucose absorption and insulin secretion. Still, high-risk clusters had the strongest improvement of rFRS and eGFR, possibly reflecting the reduced risk represented by cluster change. Similarly, not all dynamic glucose regulation indices have been validated after bariatric surgery. However, OGTT curves have been successfully compared with hyperinsulinemic-euglycemic clamp tests before and after bariatric surgery in individuals without diabetes (38). Thus, applied indices most likely reflect actual metabolic changes after bariatric surgery. Finally, different prediabetes definitions, for example, by World Health Organization criteria with fasting glucose 110–125 mg/dL (6.1–6.9 mmol/L) and without an HbA1c cutoff or by 1-h PG, may result in different prediabetes remission rates (39,40).

In conclusion, our results support the relevance of this novel T2D risk classification in individuals with severe obesity and identified differing responses to surgery. Our analysis shows that participants classified as high-risk C5 benefited most from bariatric surgery in terms of amelioration of insulin resistance, insulin secretion, prediabetes remission, and risk cluster change. Low-risk C4 participants had the lowest prediabetes remission rate, suggesting that reaching weight loss targets may not be sufficient for achieving normal glucose regulation in this cluster. These findings may help advance precision medicine approaches in bariatric surgery.

This article contains supplementary material online at https://doi.org/10.2337/figshare.28654844.

Funding. This research was supported by the “Programme d’Investissement d’Avenir” (PRECINASH, ANR-16-RHUS-0006; European Genomic Institute for Diabetes, ANR-10-LABX-0046), Lille University (WILL-CHAlRES-23-001), Fondation de la Recherche Médicale (EQU202303016330 PATTOU), and EU Horizon 2020 research and innovation program (Innovative Medicines Initiative 2, project SOPHIA 875534). This work was also supported by German Center for Diabetes Research grant 01GI0925 via the Federal Ministry of Education and Research and Helmholtz Munich. Additionally, R.J.v.S. was supported by Helmholtz Young Investigator Group grant VH-NG-1619 of the Helmholtz Center Munich and the Helmholtz Society and Cluster of Excellence Controlling Microbes to Fight Infections grant 03.007. A.L.B is supported by German Research Foundation grants GRK2816 and BI1292/9-1. L.S. is supported by the German Research Association as a clinician scientist. M.H. receives grants from the German Diabetes Association (DDG), the German Research Association, the Eurpean Research Council, the Novo Nordisk Foundation, and the German Diabetes Center (DZD e.V.).

Duality of Interest. G.M. receives consulting fees from Novo Nodisk, Eli Lilly, Boheringer Ingelheim, Johnson & Johnson, Medtronic, Fractyl Inc, RecorInc, is part of boards of Keyron Ltd, Jemyl Ltd., Metadeq Inc, and GHP Scientific Ltd. M.H. receives consulting fees or honoraria by Chiesi/Amryt, Boehringer Ingelheim, Sanofi, AstraZeneca, Bayer, Novo Nordisk, Eli Lilly, and Novartis, and is part of the DDG board. R.W. is part of advisory boards of Eli Lilly, Novo Nordisk and Sanofi, receives honoraria from Eli Lilly, Novo Nordisk, Sanofi, Boehringer Ingelheim and Synlab and support for meeting attendences from Novo Nordisk ad Sanofi. A.F. receives consulting fees from health insurance “AOK” Germany and honoraria by Abbott, Novo Nordisk and AstraZeneca. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.S., V.R., A.S., P.B., E.C., H.V., G.M., C.G., O.V., K.Z., R.A., A.Mi., R.C., G.B., C.M., M.C., M.G., V.M.F., M.H., L.F., A.Mo., K.K., A.P., R.L., A.F., N.S., H.P., A.L.B., R.J.v.S., and F.P. contributed substantially to the detailed conception and design of the study, the acquisition of data, or the data analysis and interpretation. L.S., V.R., R.J.v.S., and F.P. drafted the manuscript. L.S., R.W., and K.P. performed the cluster classification. A.L.B. and F.P. conceptualized the work. All authors contributed to the interpretation of data and critical revision and approval of the manuscript. R.J.v.S. and F.P. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 84th Scientific Sessions of the American Diabetes Association, Orlando, FL, 21–24 June 2024, the 58th Annual Conference of the German Diabetes Association, Berlin, Germany, 8–11 May 2024, the 59th Annual Meeting of the European Association for the Study of Diabetes, Hamburg, Germany, 2–6 October 2023, the 83rd Scientific Sessions of the American Diabetes Association, San Diego, CA, 23–26 June 2023, and the 57th Annual Conference of the German Diabetes Association, Berlin, Germany, 27–20 May 2023.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Cheryl A.M. Anderson and Ania M. Jastreboff.

1.
World Health Organization
.
Obesity and overweight
. Accessed 21 February 2025. Available from https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight
2.
Sandforth
L
,
Kullmann
S
,
Sandforth
A
, et al
.
Prediabetes remission to reduce the global burden of type 2 diabetes
.
Trends Endocrinol Metab.
14 February
2025
[Epub ahead of print]. DOI:10.1016/j.tem.2025.01.004
3.
Birkenfeld
AL
,
Franks
PW
,
Mohan
V.
Precision medicine in people at risk for diabetes and atherosclerotic cardiovascular disease: a fresh perspective on prevention
.
Circulation
2024
;
150
:
1910
1912
4.
Ligthart
S
,
van Herpt
TTW
,
Leening
MJG
, et al
.
Lifetime risk of developing impaired glucose metabolism and eventual progression from prediabetes to type 2 diabetes: a prospective cohort study
.
Lancet Diabetes Endocrinol
2016
;
4
:
44
51
5.
Ahlqvist
E
,
Storm
P
,
Käräjämäki
A
, et al
.
Novel subgroups of adult-onset diabetes and their association with outcomes: a data-driven cluster analysis of six variables
.
Lancet Diabetes Endocrinol
2018
;
6
:
361
369
6.
Raverdy
V
,
Cohen
RV
,
Caiazzo
R
, et al
.
Data-driven subgroups of type 2 diabetes, metabolic response, and renal risk profile after bariatric surgery: a retrospective cohort study
.
Lancet Diabetes Endocrinol
2022
;
10
:
167
176
7.
Wagner
R
,
Heni
M
,
Tabák
AG
, et al
.
Pathophysiology-based subphenotyping of individuals at elevated risk for type 2 diabetes
.
Nat Med
2021
;
27
:
49
57
8.
Sandforth
A
,
von Schwartzenberg
RJ
,
Arreola
EV
, et al
.
Mechanisms of weight loss-induced remission in people with prediabetes: a post-hoc analysis of the randomised, controlled, multicentre Prediabetes Lifestyle Intervention Study (PLIS)
.
Lancet Diabetes Endocrinol
2023
;
11
:
798
810
9.
Schlesinger
S
,
Neuenschwander
M
,
Barbaresko
J
, et al
.
Prediabetes and risk of mortality, diabetes-related complications and comorbidities: umbrella review of meta-analyses of prospective studies
.
Diabetologia
2022
;
65
:
275
285
10.
Birkenfeld
AL
,
Mohan
V.
Prediabetes remission for type 2 diabetes mellitus prevention
.
Nat Rev Endocrinol
2024
;
20
:
441
442
11.
Raverdy
V
,
Baud
G
,
Pigeyre
M
, et al
.
Incidence and predictive factors of postprandial hyperinsulinemic hypoglycemia after Roux-en-Y gastric bypass: a five year longitudinal study
.
Ann Surg
2016
;
264
:
878
885
12.
American Diabetes Association
.
2. Classification and diagnosis of diabetes: Standards of Care in Diabetes—2023
.
Diabetes Care
2023
;
46
(
Suppl. 1
):
S19
S40
13.
Le Floch
JP
,
Escuyer
P
,
Baudin
E
,
Baudon
D
,
Perlemuter
L.
Blood glucose area under the curve. Methodological aspects
.
Diabetes Care
1990
;
13
:
172
175
14.
Lorenzo
C
,
Haffner
SM
,
Stančáková
A
,
Kuusisto
J
,
Laakso
M.
Fasting and OGTT-derived measures of insulin resistance as compared with the euglycemic-hyperinsulinemic clamp in nondiabetic Finnish offspring of type 2 diabetic individuals
.
J Clin Endocrinol Metab
2015
;
100
:
544
550
15.
DeFronzo
RA
,
Matsuda
M.
Reduced time points to calculate the composite index
.
Diabetes Care
2010
;
33
:
e93
16.
Song
Y
,
Manson
JE
,
Tinker
L
, et al
.
Insulin sensitivity and insulin secretion determined by homeostasis model assessment and risk of diabetes in a multiethnic cohort of women: the Women’s Health Initiative Observational Study
.
Diabetes Care
2007
;
30
:
1747
1752
17.
Kim
JD
,
Kang
SJ
,
Lee
MK
, et al
.
C-peptide-based index is more related to incident type 2 diabetes in non-diabetic subjects than insulin-based index
.
Endocrinol Metab (Seoul)
2016
;
31
:
320
327
18.
Tura
A
,
Kautzky-Willer
A
,
Pacini
G.
Insulinogenic indices from insulin and C-peptide: comparison of beta-cell function from OGTT and IVGTT
.
Diabetes Res Clin Pract
2006
;
72
:
298
301
19.
Capristo
E
,
Panunzi
S
,
De Gaetano
A
, et al
.
Incidence of hypoglycemia after gastric bypass vs sleeve gastrectomy: a randomized trial
.
J Clin Endocrinol Metab
2018
;
103
:
2136
2146
20.
Fritsche
A
,
Wagner
R
,
Heni
M
, et al
.
Different effects of lifestyle intervention in high- and low-risk prediabetes: results of the randomized controlled Prediabetes Lifestyle Intervention Study (PLIS)
.
Diabetes
2021
;
70
:
2785
2795
21.
Stumvoll
M
,
Tschritter
O
,
Fritsche
A
, et al
.
Association of the T-G polymorphism in adiponectin (exon 2) with obesity and insulin sensitivity: interaction with family history of type 2 diabetes
.
Diabetes
2002
;
51
:
37
41
22.
Levey
AS
,
Coresh
J
,
Balk
E
, et al.;
National Kidney Foundation
.
National Kidney Foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification
.
Ann Intern Med
2003
;
139
:
137
147
23.
D’Agostino
RB
,
Vasan
RS
,
Pencina
MJ
, et al
.
General cardiovascular risk profile for use in primary care: the Framingham Heart Study
.
Circulation
2008
;
117
:
743
753
24.
Raverdy
V
,
Tavaglione
F
,
Chatelain
E
, et al
.
Performance of non-invasive tests for liver fibrosis resolution after bariatric surgery
.
Metabolism
2024
;
153
:
155790
25.
Kleiner
DE
,
Brunt
EM
,
Van Natta
M
, et al.;
Nonalcoholic Steatohepatitis Clinical Research Network
.
Design and validation of a histological scoring system for nonalcoholic fatty liver disease
.
Hepatology
2005
;
41
:
1313
1321
26.
Bagger
JI
,
Grøndahl
MFG
,
Lund
A
,
Holst
JJ
,
Vilsbøll
T
,
Knop
FK.
Glucagonostatic potency of GLP-1 in patients with type 2 diabetes, patients with type 1 diabetes, and healthy control subjects
.
Diabetes
2021
;
70
:
1347
1356
27.
Al-Mrabeh
A
,
Zhyzhneuskaya
SV
,
Peters
C
, et al
.
Hepatic lipoprotein export and remission of human type 2 diabetes after weight loss
.
Cell Metab
2020
;
31
:
233
249.e4
04
28.
Schulze
MB
,
Stefan
N.
Metabolically healthy obesity: from epidemiology and mechanisms to clinical implications
.
Nat Rev Endocrinol
2024
;
20
:
633
646
29.
Vangoitsenhoven
R
,
Wilson
RL
,
Cherla
DV
, et al
.
Presence of liver steatosis is associated with greater diabetes remission after gastric bypass surgery
.
Diabetes Care
2021
;
44
:
321
325
30.
Wagner
R
,
Heni
M
,
Kantartzis
K
, et al
.
Lower hepatic fat is associated with improved insulin secretion in a high-risk prediabetes subphenotype during lifestyle intervention
.
Diabetes
2023
;
72
:
362
366
31.
Birkenfeld
AL
,
Shulman
GI.
Nonalcoholic fatty liver disease, hepatic insulin resistance, and type 2 diabetes
.
Hepatology
2014
;
59
:
713
723
32.
Taylor
R
,
Al-Mrabeh
A
,
Zhyzhneuskaya
S
, et al
.
Remission of human type 2 diabetes requires decrease in liver and pancreas fat content but is dependent upon capacity for β cell recovery
.
Cell Metab
2018
;
28
:
547
556.e3
33.
Borges-Canha
M
,
Neves
JS
,
Silva
MM
, et al.;
CRIO group
.
Prediabetes remission after bariatric surgery: a 4-years follow-up study
.
BMC Endocr Disord
2024
;
24
:
7
34.
Madsen
LR
,
Baggesen
LM
,
Richelsen
B
,
Thomsen
RW.
Effect of Roux-en-Y gastric bypass surgery on diabetes remission and complications in individuals with type 2 diabetes: a Danish population-based matched cohort study
.
Diabetologia
2019
;
62
:
611
620
35.
Souteiro
P
,
Belo
S
,
Neves
JS
, et al
.
Preoperative beta cell function is predictive of diabetes remission after bariatric surgery
.
Obes Surg
2017
;
27
:
288
294
36.
Larraufie
P
,
Roberts
GP
,
McGavigan
AK
, et al
.
Important role of the GLP-1 axis for glucose homeostasis after bariatric surgery
.
Cell Rep
2019
;
26
:
1399
1408.e6
37.
Jastreboff
AM
,
Le Roux
CW
,
Stefanski
A
, et al.;
SURMOUNT-1 Investigators
.
Tirzepatide for obesity treatment and diabetes prevention
.
N Engl J Med
2025
;
392
:
958
971
38.
Anderwald
C-H
,
Tura
A
,
Promintzer-Schifferl
M
, et al
.
Alterations in gastrointestinal, endocrine, and metabolic processes after bariatric Roux-en-Y gastric bypass surgery
.
Diabetes Care
2012
;
35
:
2580
2587
39.
Bergman
M
,
Manco
M
,
Satman
I
, et al
.
International Diabetes Federation position statement on the 1-hour post-load plasma glucose for the diagnosis of intermediate hyperglycaemia and type 2 diabetes
.
Diabetes Res Clin Pract
2024
;
209
:
111589
40.
Bergman
M
,
Abdul-Ghani
M
,
Chan
J
, et al
.
Staging schema for early diagnosis of prediabetes
.
Lancet Diabetes Endocrinol
2024
;
12
:
873
876
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.