Variations in diabetes remission rates after bariatric surgery in Spanish adults according to the use of different diagnostic criteria for diabetes

Worldwide prevalence of type 2 diabetes (DM2) continues to increase simultaneously with obesity rates [1, 2]. Over 60% of DM2 patients are obese [1], and this tandem is now a public health problem. Recent studies have shown that medical therapy for DM2 and comorbid obesity is not as successful as bariatric surgery, suggesting that modest weight loss and DM2 control is hard to achieve [3, 4]. American Diabetes Association (ADA) recommended for the first time in their 2010 edition of their standards of care [5] to consider bariatric surgery in the treatment of DM2 patients with a body mass index (BMI) ≥ 35 kg/m². This indication is now widely accepted by leading international associations [6‐9], particularly when DM2 or comorbidities are refractory to life style changes and pharmacological treatment.

The choice of criteria to define remission of DM2 after bariatric surgery is still widely debated. Some authors consider the withdrawal of medication to be the best criteria [10], while others suggest using various fasting plasma glucose (FPG) and glycated haemoglobin (HbA1c) cut-off points [11‐13], or a combination of both [14‐16]. After publication of the meta-analysis by Buchwald et al. in 2004, DM2 remission criteria recommended by these authors (HbA1c <6% and FPG 100 mg/dl) were the most used [17].

However, a consensus group from the ADA (Buse’s consensus group [18]), consisting of experts in endocrinology, diabetes education, bariatric surgery and haematology-oncology, proposed in 2009 a new, far more stringent standard set of definition criteria based on biochemical (HbA1c and FPG levels) and clinical parameters (need for pharmacological treatment and duration of remission). In 2013, scientific associations in Spain published a position statement on metabolic surgery in patients with DM2 stating their set of criteria for define DM2 remission, which were similar to those approved by the ADA consensus group [19]. Considering the difficulty in establishing remission of DM2, Sánchez- Pernaute and Scopinaro have proposed using the ADA HbA1c cut-off point to diagnose DM2 [11, 20].

The aim of this study is to compare DM2 remission rates 5 years after bariatric surgery using the criteria approved by Spanish associations and ADA diagnostic criteria for DM2.

We conducted a retrospective observational study in 127 patients undergoing bariatric surgery in a single centre (Hospital Universitario Reina Sofía, Córdoba, Spain) between January 2001 and December 2009. All patients were diagnosed with DM2 and obesity (BMI ≥ 35 kg/m²) before surgery, and had at least 3 years of follow-up with documentation of FPG, HbA1c levels, and body weight. Three types of bariatric surgery were performed: roux-en-Y gastric bypass, sleeve gastrectomy and gastric band. The type of surgery was chosen depending on the patient’s preoperative characteristics. Those undergoing bariatric surgery with a different technique were also excluded because of their low incidence. Patients who underwent a second bariatric surgery (including conversion, revisional and reversal procedures) were also excluded.

All procedures in studies involving human participants were performed in accordance with the ethical standards of the institutional review board of Hospital Universitario Reina Sofía and with the principles of the Declaration of Helsinki 2013. For this type of study formal consent was not required. This study was approved by the Ethics Committee of the Hospital Universitario Reina Sofía.

Demographic (age and sex), anthropometric (weight, height, BMI) and analytical (FPG and HbA1c) variables were collected during follow-up.

Remission of DM2 was defined using 2 different criteria: (1) those approved by Spanish scientific associations, based on HbA1c and FPG cut-off levels and need for diabetes medication, which classify their status as complete, prolonged or partial remission, improvement, or no remission, as shown in Table 1 [19]; and (2) ADA criteria, based on HbA1c levels and need for diabetes medication, which classify patients’ status as complete or partial remission, optimal control, or no remission, as shown in Table 2 [20].

For the sake of statistical analysis, the complete and prolonged remission groups defined by the Spanish associations were included in the same group, given that prolonged remission according to the Spanish associations’ criteria is included in the criteria for complete remission. Complete and partial remission and no remission criteria from both classification systems were compared. The improvement and optimal control criteria from the Spanish and ADA systems, respectively, were not compared separately, as these were both included in their respective no-remission classification.

Excess weight loss percentage (EWL%) was calculated using the formula: [(initial weight- follow-up weight)/(initial weight – ideal weight)] X 100. Ideal weight was calculated for a BMI of 21 kg/m²in women, and 23 kg/m² in men. DM2 was diagnosed according to ADA criteria: FPG > 126 mg/dl, HbA1c ≥ 6.5%, random blood glucose ≥200 mg/dl, or use of insulin or oral diabetes medication.

Glucose was measured in mg/dl, and HbA1c results are reported in NGSP/DCCT (%) units (to 1 decimal point) [21].

A total of 127 patients (62.2% women) with DM2 were included in the study. Baseline and postoperative characteristics of patients are shown in Table 3. Mean age at the time of surgery was 47 ± 8 years; mean preoperative BMI was 50 ± 7 kg/m², FPG was 134 ± 53 mg/dl and HbA1c was 8 ± 7%. As shown in Table 3, patients presenting complete remission are younger, with lower postoperative BMI, and lower pre- and postoperative FPG and HbA1c levels. These intra-group differences were found in both classifications (ANOVA test). Patients with no remission using both reclassification systems were more likely to have been on insulin (Fig. 1). Roux-en-Y gastric bypass was performed in 96% of patients, sleeve gastrectomy in 3%, and gastric band in the remaining 1%. Before surgery, 39 patients were in treatment with metformin, 6 with insulin, and 13 with other oral diabetes medication (ODM). Nineteen (19) patients were in treatment with a combination of metformin and insulin, and 14 with metformin and other ODM. Of the remaining patients, 36 were following a dietary and exercise programme, as DM2 had been diagnosed less than 6 months before surgery. After a mean follow-up of 5 ± 2 years, mean BMI, FPG and HbA1c were 37 ± 6 kg/m², 100 ± 28 mg/dl and 6 ± 1%, respectively (Table 3). The difference between these and variables before and after surgery was statistically significant (paired samples t-tests; p = 0.001, p = 0.007 and p = 0.003, respectively).

There is a good correlation between the reclassification systems (Spearman’s rho = 0.781, p < 0.001). Fifty-nine (59) patients showed complete remission, 3 partial remission, and 35 no remission under both sets of criteria (Table 4).

There is a certain amount of controversy regarding the best criteria to define diabetes remission following bariatric surgery. EWL% is known to be a predictive factor of DM2 remission: the higher the EWL%, the greater the likelihood of remission [22]. This was found to be true in our series. Between the partial remission and no remission groups according to the Spanish criteria, although there were differences, they were not statistically significant. Remission is also associated with age and the degree of DM2 control. Our study has shown that complete DM2 remission is more common in young patients and in those with better-controlled mean HbA1c and FPG levels. Our findings also show that more stringent criteria for defining DM2 remission results in a lower rate of complete remission categorised as total remission. Instead, they would be categorised as partial remission (63.8% with complete remission under Spanish criteria vs. 52% complete remissions following the ADA’s more stringent criteria [HbA1c < 5.7%]). Some authors consider DM2 to be in remission purely on the basis of clinical criteria (no need for diabetes medication) or a single analytical finding: FPG or HbA1c [2, 23]. In the meta-analysis performed by Buchwald et al. [24], remission rates varied considerably (54.9-95.1%), depending on the type of surgery or definition criteria used (FPG < 100 mg/dl or HbA1c < 6%). Other studies and systematic reviews with high remission rates also used less demanding criteria [2, 23] and relatively short (1 to 2 years) follow-up periods. Different approaches to define remission have also led to false expectations with regard to the true percentage of patients achieving long-term diabetes remission. For example in the SOS study the 2-year DM2 remission rate of 72% declined to a 36% remission rate after 10 years [14]. In our series, after a mean follow-up of 64 months, 52% of patients achieved complete remission under ADA criteria, i.e. HbA1c < 5.7% and no need for diabetes medication. This is consistent with other studies [25‐27] reporting remission rates of 40.6, 43.6 and 50%. Specifically, Ramos-Levi [27] et al., in a retrospective cohort of 110 patients, compared DM2 remission rates based on the same definition criteria used in our study. There were some differences between these studies; though mean BMI in their patients was lower than in our study (43.6 ± 5.5 vs 50.9 ± 7.6 kg/m2) and the percentage of patients on insulin in their study was higher than in our study (44.5% vs 19.7%). They found no statistically significant differences between remission rates (50% of total remission using both the Spanish and ADA criteria). However, follow-up was limited to 18 months. In our 5-year follow-up study, we found that remission rates did differ depending on the definition criteria used. This is probably due to the fact that our follow-up period is one of the longest of all studies in DM2 remission. This begs the question whether these remission rates can be sustained over a long-term follow-up of 10 years or more. ADA remission criteria are the most stringent and the most widely used. To confirm DM2 remission, therefore, these same criteria should be applied and updated according to established standards.

One limitation of our study lies in the fact that retrospective cohorts are more susceptible to bias, such as loss to follow-up. In addition, some relatively important data, such as time from onset of diabetes to bariatric surgery, were missing. Another possible limitation is that we did not take into consideration differences between surgical techniques, because very few sleeve gastrectomies were performed during the study period due to the preferences and experience of the surgical team. Into the bargain we cannot know whether patients with new diabetes drugs such as SGLT-2 inhibitors or GLP-1 analogues would have similar or unrelated remission rates, because during the years in which these patients (2001-2009) were included, these drugs were not available in Spain.

In conclusion, we believe that strict criteria based on those established by the ADA with the addition of the Spanish prolonged remission criteria should be used to determine true remission of diabetes following bariatric surgery. Further studies in larger cohorts with longer follow-up periods are needed to conclusively show the best criteria for defining post-bariatric surgery diabetes remission.