Weight Loss Directly Influences Intermediate-Term Remission of Diabetes Mellitus After Bariatric Surgery: A Retrospective Case-Control Study

Abstract

Purpose

Roux en Y gastric bypass surgery (RYGB) is an effective therapy for patients with severe obesity. It induces both significant weight loss and rapid improvements of metabolic complications. This study was undertaken to better define the direct role of weight loss in the metabolic improvements.

Methods

A retrospective, case-control study of a cohort of 649 patients with obesity who underwent RYGB, comparing higher and lower responders at 2 years after surgery (n = 100 pairs). Pairs of patients were matched for age, gender, and initial BMI. The rates of remission of diabetes, hypertension, dyslipidemia, and hyperuricemia were compared using a mixed effects logistic regression analysis. Results Diabetes before surgery was present in 12/100 lower responders and 17/100 higher responders. Remission at 2 years was observed in 4/12 (33%) of lower responders, compared to 15/17 (88%) of higher responders. Thus, the odds of diabetes remission was significantly smaller in lower responders (OR = 0.067, 95% CI 0.01–0.447). A mixed model regression analysis of all the parameters for each patient showed that the odds of achieving remission of any comorbidity was significantly lower in lower responders (OR = 0.62, 95% CI = 0.39–0.97).

Conclusion

We could demonstrate that weight loss is a significant determinant of the remission of diabetes 2 years after RYGB. These data underline the importance of weight loss in the benefits of this procedure.

Introduction

Obesity is defined by WHO as an abnormal or excessive fat accumulation that presents a risk to health [1], mostly because comorbidities such as type 2 diabetes mellitus (T2DM), hypertension (HTN), or dyslipidemia [2]. Weight loss achieved with conservative multidisciplinary therapy, averaging 5–10% at most, usually allows to achieve improvement of most comorbidities but is rarely sustained over time [3].

Bariatric surgery, including Roux-en-Y gastric bypass (RYGB), has been an accepted therapy in patients with severe obesity since 1991 [4]. It is an effective procedure that induces a rapid and important decrease in body weight that can amount to a loss of up to 70% of the excess weight (or ~ 30% of the pre-surgery weight) in the long-term. The nadir of the weight curve often occurs between 18 and 24 months [5].. It also allows for an improvement, and at times even resolution, of comorbidities [5–9]. Indeed, remission rates of approximately 70%, 60%, and 40% have been reported between 3 and 7 years after RYGB for T2DM, dyslipidemia, and HTN, respectively [3, 6, 10].

However, little is known of the relationship between body weight trajectories over the initial 2 years following surgery and the remission of somatic comorbidities. Therefore, the present study was specifically designed to assess the role, if any, played by the amount of weight loss achieved during these initial 2 years after surgery on the evolution of comorbidities of obesity. We hypothesized that regardless of the type of comorbidity, their resolution would be greater in the patients losing the most weight.

Materials and Methods

Study Population

Data from this study were derived from the COOL cohort (for Cohort Obesity Lausanne), an ongoing observational bariatric study approved by the Ethics Committee for Clinical Research of the Canton de Vaud (authorization CER-VD #304/15). Patients from this cohort underwent RYGB by a single surgical team at the Lausanne University Hospital between 1999 and 2012 (N = 649).

All patients were evaluated and prepared before surgery by a multidisciplinary team including an endocrinologist, dietitians, psychologists, a psychiatrist, and a bariatric surgeon. All patients were submitted to preoperative teaching focusing on the principles and risks of bariatric surgery, dietary, and behavioral aspects. All patients were submitted to upper gastrointestinal endoscopy and abdominal echography before surgery, and were evaluated using nocturnal oximetry for sleep apnea. Other investigations (i.e., cardiological evaluation) were performed only if deemed necessary. After surgery, patients were seen after 1 month, then every 3 months during the first year, biannually during the second year, then annually. Each visit included a clinical evaluation and laboratory tests to evaluate comorbidities and micronutritional deficiencies. The latter were supplemented as required. Patients were also seen by the dietician and, if deemed necessary, by the psychologist. The comprehensive work-up and follow-up protocol did not change over time.

Roux-en-Y gastric bypass was performed by laparoscopy using 6 ports. Briefly, the procedure involved formation of a small 10–20 ml gastric pouch using linear staplers, a 30–50 cm biliopancreatic limb, and a 100 cm Roux limb. The latter was brought up in a retrocolic and retrogastric fashion. The gastrojejunostomy was performed using a 21 mm circular stapler, and the jejunojejunoistomy with a 45 mm linear stapler. All mesenteric defects were closed. The technique remained the same over time, but for small purely technical details (stapler brand, choice of suturing material).

Study Design

This study was designed as a retrospective, case-control study of the COOL patients comparing higher responders and lower responders at 2 years after surgery. To this end, all the patients of the cohort were first stratified according to the loss of weight achieved at 2 years, expressed in percent of excess weight loss (%EWL = [(initial weight) − (postoperative weight)]/[(initial weight) − (ideal weight)]), with ideal weight considered as the weight corresponding to a BMI = 25 kg/m² [11].

Then, one hundred patients with the highest %EWL (higher responders) were matched for pre-intervention BMI, age, and gender to one hundred patients with the lowest %EWL (lower responders), allowing to build 100 case-controls pairs (Fig. 1). For the matching, a mean difference of 2.4 (± 1.8) BMI points and 8.9 (± 7) years of age was accepted between two members of a pair.

Fig. 1.

Each bar represents the percent of excess weight loss (%EWL) at 2 years of all patients from the COOL cohort. Higher responders of the present study in red, lower responders in blue (N = 649)

Study End Points

Comorbidities assessed before and 2 years after surgery were prediabetes and T2DM, dyslipidemia, HTN, and hyperuricemia. The clinical criteria used to define a pathologic condition, or the cutoff laboratory values applied to the biological data, are listed in Table 1. All data were retrieved from electronic medical records, and then de-identified for analyses.

Table 1.

Clinical or laboratory criteria used to define comorbidities

Condition	Definition
Hyperuricemia1	> 339 (mmol/l) or use of medication
Hypertension	> 140/902 or use of medication
Prediabetes3	Fasting glucose : > 5.6 < 7 mmol/l, or HbA1c : 5.7–6.4 %, or OGTT4 120’ : 7.9–11.0 mmol/l
Diabetes2	Fasting glucose : ≥ 7 mmol/l, or HbA1c : ≥ 6.5%, or OGTT 120’ > = 11.1 mmol/l, or use of any glucose lowering drug
Dyslipidemia	According to ATPIII5 :
	Hypercholesterolemia : LDL-cholesterol ≥ 2.6 mmol/l
	Isolated hypertriglyceridemia : TG ≥ 1.7 mmol/L
	Mixed dyslipidemia: LDL ≥ 2.6 mmol/L and TG ≥1.7 mmol/L or use of any lipid lowering drug

¹Normal reference range used in hospital laboratory

²Data’s collection was done before the new AHA’s 2017 Hypertension Clinical Guidelines

³American Diabetes Association. Standards of medical care in diabetes–2018 [28]

⁴OGTT oral glucose tolerance test

⁵Expert panel on detection, evaluation, and treatment of high blood cholesterol in adults–2001 [29]

For patients diagnosed with any of these comorbidities preoperatively, the definition of remission was as follows: achievement of normal biological values (according to cut off values of Table 1), in absence of any medical therapy, at the time of evaluation. According to this definition, remission criteria were identical for prediabetes and diabetes, and required normalization of the biological work-up (fasting glucose ≤5.6 mmol/l and/or HbA1c <5.7%, and/or OGTT 120’ < 7.9 mmol/l, depending on the available values).

Statistical Analysis

Numerical variables are presented as means ± SD, and categorical variables as numbers (%). First, the rate of remission of each comorbidity was calculated separately for higher and lower responders. It was calculated as the number of patients experiencing a remission at 2 years divided by the number of patients presenting the comorbidity preoperatively, expressed in percent values. These remission rates were compared between the 2 groups, separately for each comorbidity, using logistic regression analysis to assess the likelihood of remission of lower responders relatively to higher responders. To do this, individual remissions were considered as a binary (presence or absence of remission) outcome variable.

Then, all comorbidity remissions were analysed together, using a mixed effects logistic regression analysis with random intercept to take into account the likely lack of independence between comorbidities within patients.

Due to inherent imperfections in the matching design, each analysis was finally adjusted for the following potential confounding factors; age (years), sex, and initial BMI.

Results

Pre-surgery

The clinical characteristics of patients before surgery are described in Table 2. Patients were predominantly women (90%), with a mean age of 36.1 ± 9.2 years (range 19–54) for lower responders and 37.7 ± 9.7 years (range 20–65) for higher responders (p = 0.2). By study design, the mean preoperative BMI was identical between the two patients groups: 42.5 ± 3.8 kg/m² (range 34.0–56.8) for higher responders and 43.1 ± 4.2 kg/m² (range 36.3–59.2) for lower responders. Despite the matching by BMI, prevalence of both HTN (53% and 38% for lower and higher responders, respectively; p = 0.05) and hypertriglyceridemia alone (16% and 5%, respectively; p = 0.03) were significantly higher in lower responders compared to higher responders at baseline. All other biochemical characteristics were identical.

Table 2.

Pre-surgery clinical characteristics of the 200 patients

	Lower responders (n = 100)	Higher responders (n = 100)	P value
Baseline BMI (kg/m2)	43.1 ± 4.2	42.5 ± 3.8	0.3a
Male/female (%)	10/90	10/90
Age (years)	36.1 ± 9.2	37.8 ± 9.6	0.2a
Comorbidities (%)
Ethnicityc
Caucasian	87	92	0.36b
Hispanic	8	7	1b
African	3	0	0.45b
Dyslipidemia	90	81	0.11b
Hypercholesterolemia	40	37	0.88b
Hypertriglyceridemia	16	5	0.03b
Mixed dyslipidemia	34	39	0.22b
Taking medicationd	8	11	0.34b
Prediabetes	37	25	0.09b
Diabetes	12	17	0.42b
On oral medication	5	12	0.14b
On insulin	5	3	0.24b
Hypertension	53	38	0.05b
Taking medicatione	22	25	0.4b
Hyperuricemia	43	39	0.67b

^aUnpaired bilateral student’s t test

^bFisher’s exact test

^cUnknown data for ethnicity was, respectively, of 2% and 1%

^dUnknow data 3%

^eUnknow data 1.5%

As a result of this pairing, the mean %EWL at 2 years was lower in the lower responders (79.4 ± 14.1%; range 35.0–99.5) than in the higher responders (109.5 ± 8.4%; range 100–141.7; p < 0.001). Similarly, the mean BMI at 2 years was higher in the lower responders (29 ± 3.3 kg/m2; range 25.1–40.5) than in the higher responders (23.4 ± 1.2 kg/m2; range 18.7–25.0, p < 0.001). The percent of total body weight loss (%TWL) was also significantly different between both groups, with a mean %TWL of 32.8% ± 6.1 (range 15.6–46.8) for lower responders, and 44.5% ± 5 (range 37.4–56.9) for higher responders (p < 0.001, Fig. 2).

Fig. 2.

BMI, percent of extra weight loss (%EWL) and percent of total weight loss (%TWL) at 2 years after surgery. All parameters significantly different between higher and lower responders by unpaired Student’s T test (p < 0.001)

Remission Rate of Metabolic Comorbidities After RYGB

Remission rates for all comorbidities in higher and lower responders are displayed in Table 3. The remission rate of T2DM was significantly lower in lower responders (OR = 0.067, 95% CI 0.01–0.447). 17 patients among the higher responders and 12 among the lower responders (on a total of 200 patients) had diabetes before surgery. After 2 years, diabetes was in remission in 88% of them in the higher responders group and 33% in the lower responder one (OR = 0.067, 95% CI 0.01–0.447). In contrast, no statistically significant difference between the 2 groups was noted for the rate of resolution of prediabetes, dyslipidemia, HTN, and hyperuricermia, even after adjusting for age, gender and initial BMI. (Table 3).

Table 3.

Results of logistic regression analysis to assess the risk of metabolic comorbidities’ remission after 2 years of RY GB between higher and lower responders (N = 200; corresponding to 100 higher responders and 100 lower responders)

	Patients affected % (N)	Patients not affected % (N)	Missing 2-year data N (lower/higher)	Total patients in remission after 2 years %	Lower responders in remission %	Higher responders in remission %	Odds ratio	95% CI
Diabetes	14.5 (29)	85.5 (171)	0	66	33	88	0.067	0.01–0.447
Prediabetes	31 (62)	69 (138)	0	87	87	88	0.873	0.189–4.034
Dyslipidemia	85.5(171)	14.5 (29)	4 (1/3)	62	58	65	0.744	0.397–1.395
Hypercholesterolemia	38.5 (77)	14.5 (29)	4(1/3)	65	63	67	0.834	0.324–2.141
Isolated hypertriglyceridemia	10.5 (21)	14.5 (29)	4(1/3)	61	75	60	0.450b
Mixed dyslipidemia	36.5 (73)	14.5 (29)	4(1/3)	56	46	65	0.451	0.173–1–181
Hypertension	45.5 (91)	54 (108)	18(8/10)	77	72	86	0.423	0.123–1.459
Hyperuricemia	41(82)	59 (118)	12(4/8)	80	80	81	0.93	0.285–3–034

^aOR for estimate risk of remission for lower responders relatively to higher responders

^bFisher’s exact test because of small number of subjects with preoperative hypertriglyceridemia that could not be used to estimate an OR with acceptable accuracy

A total of 402 pre-surgery comorbidities were recorded among 182 patients, an average of 2.2 comorbidities per patient. In a mixed model regression analysis on data adjusted for age, gender, and initial BMI, the odds of achieving remission of any comorbidity was significantly lower in lower responders than in higher responders (OR = 0.62, 95% CI = 0.39–0.97).

Discussion

RGYB induces significant weight loss and major metabolic improvements linked to modifications in gastrointestinal tract hormones [12–14]. In this case-control study, we tested the hypothesis that the amount of weight loss achieved after surgery can influence the rate of remission of the metabolic comorbidities of obesity. We could thus demonstrate that the rate of remission of T2DM 2 years after RYGB was significantly higher in patients who lost most weight. This finding is consistent with our initial hypothesis, demonstrating the role of weight loss in the resolution of T2DM after RYGB.

Initial observations had suggested the role of factors other than weight loss in the evolution of T2DM after RYGB [15–18]. These studies allowed to identify the crucial role played by rapid changes in GI tract hormones. However, the well-recognized direct relationship between overweight/obesity and insulin resistance may also play a role in the long-term effects. Campos et al. [19] compared non-diabetic obese patients after RYGB to obese controls submitted to calorie restriction and losing the same amount of weight. Peripheral insulin sensitivity evolved in an identical way in the two groups, in a manner that was only dependent upon long-term weight loss. These data and others [20] demonstrate the very important role of weight loss in long-term diabetes remission. Long-term 10 year data from the SOS study [21] which include patients after lap banding, vertical band gastrectomy, or RYGB show that the amount of weight lost was significantly correlated with the reduction in fasting glycemia, and to insulin secretion and sensitivity. The impact of changes in upper gastrointestinal track on diabetes resolution [15] is also challenged by the similar rates of weight loss and diabetes remission observed after RYGB and sleeve gastrectomy [22].

Although collected solely after RYGB, the present data are entirely consistent with these past observations, demonstrating the importance of the amount of weight loss for intermediate-term diabetes remission, regardless of the surgical procedure [22]. One of the mechanisms implicated in diabetes remission could be the loss of visceral adipose tissue (VAT). Indeed our group recently showed in patients from the same cohort that the reduction of visceral adipose tissue correlates directly with the amelioration of glucose metabolism [23]. In addition, weight loss has also a direct impact upon the amount of liver and pancreatic fat accumulation, and decreases in these fat depots have been associated with improvements in insulin sensitivity and fasting plasma glucose levels [24, 25].

Here, we show no effect of the amount of weight loss on the remission rate of dyslipidemia. In rodents, RYGB is associated with a significant decrease in cholesterol absorption paralleled with an increase in its stool excretion [26]. As a result, circulating cholesterol levels decrease after RYGB, but not sleeve gastrectomy [10]. In the human as well, RYGB decreases cholesterol absorption and improves hypercholesterolemia independently from weight changes [27]. Therefore, the present data may reflect the preponderance of this specific mechanism of RYGB on the metabolism of biliary salts, independent of the total weight loss amount.

This study has several potential limitations. Matching the two groups for their pre-surgery BMI resulted in a relatively large range of weight loss within each group. Thus, and although the two groups remained significantly different by all parameters evaluated, some of the patients in the lower response group had an absolute weight loss close to some in the higher response group. Another potential limitation lies in the relatively small number of patient pairs included, although the study design can at least partially compensate for that. Also, a relatively low number of patients were diagnosed with diabetes before surgery, again potentially resulting in a loss of statistical power. Furthermore, we have no information upon diabetes duration prior to surgery, which may also constitute a determinant of remission [18].

Despite these potential limitations, we could demonstrate here that there is an overall significantly greater amelioration of all the parameters measured in patients with the greatest initial weight loss, using a mixed model regression analysis. This latter observation constitutes a strong argument in favor of long-term multidisciplinary follow-up of patients after bariatric surgery, with the aim to improve weight maintenance.

In conclusion, our data showed that in patients with obesity who have undergone RYGB, the rate of remission of diabetes mellitus 2 years after surgery is significantly correlated with the amount of excess weight lost at 2 years.