This retrospective cohort study was conducted at the Affiliated Drum Tower Hospital, Medical School, Nanjing University, China, from November 2021 to March 2023. Patients who remained obese 6 months after metabolic surgery and finished 1 year of follow-up were enrolled. The inclusion criteria were as follows: individuals aged ≥ 18 years, underwent primary metabolic surgery ≥ 1 year, had a BMI of ≥ 28 kg/m2 at 6 months postoperatively, and could provide informed consent. The exclusion criteria were as follows: presence of an anatomical or endocrinological pathology causing suboptimal weight loss or weight regain (e.g., gastrogastric fistula), calcitonin concentrations of ≥ 50 ng/l, a family or personal history of multiple endocrine neoplasia type 2 or familial medullary thyroid carcinoma, a history of acute or chronic pancreatitis, an estimated glomerular filtration rate of < 30 ml/min/1.73 m2, current pregnancy, inability to maintain adequate contraception, breastfeeding, a personal history of heart failure with New York Association class III or higher, or a personal history of diabetic gastroparesis. The trial was conducted in accordance with the principles embodied in the Declaration of Helsinki and was approved by the Ethics Committee of the Affiliated Drum Tower Hospital, Medical School, Nanjing University (approval no. 2017-030-08). All participants provided informed consent for their data potentially being used in the study, and the anonymity of all participants was preserved. Similar methods have been presented [6].

Patients eligible for the trial were divided into two groups based on their treatment regimen at 6 months postoperatively: (1) control group, which underwent the diet–exercise program, and (2) liraglutide group, which underwent the diet-exercise program plus liraglutide therapy at 6 months postoperatively. In routine clinical practice in our center, patients with a BMI of ≥ 28 kg/m2 at 6 months postoperatively were given an extra diet–exercise program as follows and were advised to receive liraglutide treatment. The starting dose of liraglutide was 0.6 mg per day administered subcutaneously at week 1, 1.2 mg per day as tolerated at week 2, and up to 1.8 mg per day from week 3 to the end of week 12.

The exercise program was designed to meet the World Health Organization (WHO) recommendations of a minimum of 75 min/week of vigorous-intensity aerobic physical activity, 150 min/week of moderate-intensity aerobic physical activity, or an equivalent combination of both [22]. Participants followed a low-calorie diet of 1200 kcal/day. Patients were advised to take one Centrum®, Caltrate®, and alfacalcidol soft capsule (0.5 µg) twice daily. Patients received monthly telephone follow-ups, and follow-up appointments were scheduled in the clinic at 12 and 24 weeks.

Participants with type 2 diabetes mellitus (T2DM) were diagnosed according to the WHO criteria, and the remission of T2DM was defined as having glycosylated hemoglobin (HbA1c) levels of < 6%, fasting blood glucose (FBG) levels of < 5.55 mmol/l, and the absence of anti-diabetic medications [23, 24]. Hypertension was defined as systolic blood pressure (SBP) of ≥ 140 mmHg, diastolic blood pressure of ≥ 90 mmHg, or the use of antihypertensive medication. Additionally, the remission of hypertension was defined as blood pressure (BP) of < 120/80 mmHg without antihypertensive medication [25]. Non-alcoholic fatty liver disease (NAFLD) was diagnosed using abdominal ultrasound, the degree of lipid deposition and liver fibrosis was evaluated using the controlled attenuation parameter, and liver stiffness was measured using FibroTouch® (Wuxi Heskell Medical Technology Co., Ltd., Wuxi, China) [26]. Hyperuricemia was diagnosed with serum uric acid (UA) levels of ≥ 420 mmol/l or using anti-hyperuricemia medications [27]. Furthermore, dyslipidemia was defined as fasting total lipoprotein cholesterol levels of ≥ 5.2 mmol/l, low-density lipoprotein cholesterol (LDL-C) levels of ≥ 3.4 mmol/l, high-density lipoprotein cholesterol (HDL-C) levels of < 1.04 mmol/l, triglyceride levels of ≥ 1.7 mmol/l, or a history of lipid-lowering agents [25]. The remissions of dyslipidemia and hyperuricemia were defined as achieving normal biochemical values for serum lipid and UA levels without medication.

Participants’ data, including height, weight, BP, comorbid conditions, and medication usage, were collected at baseline and at 4, 8, 12, 16, 20, and 24 weeks. Laboratory measurements, including hepatic and renal function, lipid profiles, nutritional status, and HbA1c, FBG, fasting insulin, and fasting C-peptide levels, were obtained at baseline and at 12 and 24 weeks. The Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) was calculated using the following formula: HOMA-IR = (fasting insulin level × FBG level)/22.5 [28]. Visceral fat area was determined using InBody 720 Body Composition Analyzer (BioSpace Co., Ltd., Seoul, Korea).

To evaluate weight loss after metabolic surgery, %TWL, ∆weight, BMI, and ∆BMI were calculated for all participants at every follow-up visit. The following formulas were used: %TWL = ([baseline weight—weight at each follow-up visit]/baseline weight) × 100%; ∆weight = (weight at each follow-up visit)—(baseline weight); BMI = body weight (kg)/height2 (m2); and ∆BMI = (BMI at each follow-up visit)—(baseline BMI).[25]

The primary outcome of our study was the change in %TWL. The secondary outcomes included the proportions of patients achieving %TWL of ≥ 5%, ≥ 10%, and ≥ 15%; changes in weight and BMI; proportions of patients achieving the endpoint of healthy weight (18.5 kg/m2 ≤ BMI < 24 kg/m2), BMI < 28 kg/m2, LDL-C levels < 2.6 mmol/l, HbA1c levels < 6%, and BP < 120/80 mmHg; rate of complication alleviation; and occurrence rates of adverse and severe adverse events.

Using a two-sample t-test and based on preliminary experimental data, the difference in the mean percentage of weight change between the liraglutide and control groups was 4.8%; the group mean standard deviation (SD) was 5.4%, and statistical significance was set at the 5% level. With a sample size of 23 individuals per group, ensuring > 90% power and accounting for an estimated 20% dropout rate, at least 56 participants (28 per group) needed to be recruited. PASS 15.0 (NCSS, LLC., Kaysville, Utah, USA) was used to perform the sample size calculation.

Statistical analyses were performed using IBM SPSS Statistics for Windows version 25.0 (IBM Corp., Armonk, NY, USA). Categorical and continuous variables are presented as percentages and mean ± SD, respectively. Categorical variables are described using absolute and relative frequencies. Continuous variables at baseline were compared using independent sample t-tests. Linear regression models adjusted for baseline values were used to analyze the differences in changes between the liraglutide and control groups at 24 weeks. Within-group differences before and after the intervention were assessed using t-tests. Weight, BMI, and %TWL were examined for differences using repeated-measures analysis of variance. Categorical variables were analyzed using Pearson’s chi-square test and Fisher’s exact test. A significance level of 0.05 was applied to all hypothesis testing.