Evaluating appetite/satiety hormones and eating behaviours as predictors of weight loss maintenance with GLP‐1RA therapy in adolescents with severe obesity

Whilst glucagon‐like peptide‐1 receptor agonists (GLP1‐RAs) are effective for treating adolescent obesity, weight loss maintenance (WLM; preventing weight regain) remains a challenge. Our goal was to investigate appetite/satiety hormones and eating behaviours that may predict WLM with exenatide (a GLP1‐RA) versus placebo in adolescents with severe obesity.


| INTRODUCTION
][9][10][11] However, whilst some adolescents are able to achieve and maintain BMI reduction with AOMs, others experience no change, increased BMI or initial BMI reduction followed by regain. 7,9,10Predicting who will respond to a given intervention and prevent weight regain following initial BMI reduction represents an important step towards tailored obesity treatment with effective and long-term results whilst avoiding ineffective medication exposure.This is of particular relevance as there are now four Food and Drug Administration (FDA)-approved medications for the chronic treatment of pediatric obesity.
Two of the currently FDA-approved medications for pediatric obesity are glucagon-like peptide-1 receptor agonists (GLP-1 RAs).
GLP-1RAs are an attractive pharmacotherapy option to prevent weight regain given their multiple central and peripheral mechanisms of action targeting many of the post-weight loss counterregulatory metabolic adaptations known to induce obesity relapse.3][14][15][16][17][18][19][20][21] Maintenance of a percentage of initial weight reduction following LMT and preventing weight regain back to or above pre-intervention weight is characterized as weight loss maintenance (WLM).A recent trial by our group evaluated the use of a GLP-1RA, exenatide extended release (XR) 2 mg weekly, to aid in WLM. 22WLM was assessed over 12 months following initial BMI reduction (≥5%) from meal replacement therapy for ≤8 weeks.We previously reported that participants assigned to exenatide XR experienced a BMI regain of 4.6% versus 10.1% in the placebo group. 22Although this was a non-significant study, within the exenatide XR group, weight trajectories varied widely with responses ranging from 15% BMI regain to 20% additional BMI reduction over 12 months. 22The purpose of this pre-specified secondary analysis was to investigate if pre-randomization self-reported appetite and satiety, eating behaviour characteristics and/or appetite/satiety regulation hormones were predictive of 52-week WLM in adolescents with severe obesity randomized to exenatide XR treatment versus placebo following at least a 5% BMI reduction.

| Overview
This was a pre-specified secondary analysis data from a single-centre, double-blind, randomized, placebo-controlled clinical trial designed to evaluate the effect of exenatide XR 2 mg weekly on improving WLM after initial dietary-induced weight reduction (ClinicalTrials.govNCT02496611).One-hundred adolescents with severe obesity were enrolled in the parent study and participated in a short-term (≤8 weeks) meal replacement therapy run-in phase with the goal of ≥5% BMI reduction during this time. 23Only those achieving ≥5% BMI reduction at the 4-, 6-or 8-week time point were randomized 1:1 to either exenatide XR or placebo for an additional 52 weeks (a total of 66 participants).Participants who did not have ≥5% BMI reduction were removed from the study, and no further data was collected.
Both treatment arms also received ongoing LMT for 52 weeks.Additional details of the trial, including randomization procedures and LMT, have been previously published. 22We used data from the time of the exenatide/placebo randomization visit (i.e., after an initial 5% weight reduction) and the end of the study visit at 52 weeks for this analysis.The study was approved by the University of Minnesota Institutional Review Board and all parents/guardians and participants provided written informed consent and assent, respectively.

| Participants
Participants were recruited between 2016 and 2020, prior to FDA approval of GLP-1RAs for pediatric obesity.Primary inclusion criteria included ages 12 to <18 years with severe obesity classified as BMI ≥1.2 Â 95th percentile (based on age and sex norms) 24 or ≥35 kg/m 2 , whichever was lower.General exclusion criteria included pre-pubertal status, type 1 or type 2 diabetes mellitus, previous (within 6 months) or current use of medications used primarily for weight loss, history of metabolic/bariatric surgery and dose changes in medications for dyslipidemia, prediabetes or hypertension within the prior 6 months.

| Intervention
All participants in the parent study were instructed to adhere to the meal replacement therapy for at least 4 weeks, and up to 8 weeks, with the goal of achieving ≥5% BMI reduction. 23The prescribed eating plan consisted of three liquid shakes of balanced macronutrient composition, two pre-packaged low-calorie frozen entrée meals for dinner and two servings of fruit and three servings of vegetables per day for a total of approximately 1400 kcals per day.Only participants who achieved ≥5% BMI reduction after 4-8 weeks of the meal replacement therapy were randomized 1:1 to either exenatide XR 2 mg plus LMT or placebo plus LMT for an additional 52 weeks.Only data from participants who were randomized to exenatide XR or placebo were included in this analysis.

| Primary outcome
Anthropometrics Height and weight were measured at randomization and week 52 visits with a calibrated, wall-mounted stadiometer and an electronic scale, respectively.WLM was defined as the mean percent change in BMI (weight in kg divided by height in metres squared) from randomization (post-meal replacement therapy run-in phase) to 52 weeks, meaning a positive value for WLM is weight regain whilst a negative value for WLM is additional weight loss.Total fat mass (kg) was measured by dual-energy X-ray absorptiometry (DXA; GE Healthcare) at randomization and week 52.DXA measurements, using standard positioning techniques, were conducted and analysed by trained staff.

Child Eating Behaviour Questionnaire
The CEBQ is a 35-item questionnaire assessing eating behaviours. 25,26 this study, questions were adapted to accommodate adolescents reporting on their own eating behaviour (i.e., changed from third person ['my child…'] to first person ['I…']). 27The CEBQ comprises eight eating domains of which we focused on three 'food approach behaviors' (Food Responsiveness, Emotional Overeating, Enjoyment of Food) and one 'food avoid behavior' (Satiety Responsiveness).Each item in the CEBQ is rated on a five-point Likert-type scale ranging from 'never' (1) to 'always' (5).Scores for each eating domain were calculated as the averages of the scores for the individual survey items pertaining to that specific domain, with a higher score indicating greater endorsement of the specific eating behaviour. 26When a response to a question was missing, the average of complete responses within each scale was used as long as at least 75% of the questions within the scale had responses.

Power of Food Scale
The PFS is a 15-item self-reported questionnaire to assess hedonic hunger or preoccupation with and desire to eat foods for pleasure in the absence of hunger. 28,29Each item on the PFS is answered on a 5-point Likert-type scale with 1 indicating 'I don't agree at all' and 5 indicating 'I strongly agree'.The PFS includes three subscales: Food Available, Food Present, and Food Tasted.Scoring the PFS includes an aggregate score and subscale scores. 28The aggregate score is the mean score per response across all items; subscale scores are the mean score per response in each subscale.For the aggregate score and each of the subscale scores, 1 is the minimum score indicating lowest hedonic hunger, and 5 is the maximum score indicating highest hedonic hunger. 28,29nge eating behaviours Binge eating behaviours were measured using the Eating Disorder Examine Questionnaire (EDE-Q).30 For this study, the following two items were used to assess binge eating: 'Over the past 28 days, how many times have you eaten what other people would regard as an unusually large amount of food (given the circumstances)?' and 'On how many of these times did you have a sense of having lost control over your eating (at the time that you were eating)?'These items were summarized by an indicator variable of whether or not the individual engaged in any binge eating in the last 28 days (i.e., any participant who responded with a number greater than 0 for both questions was considered positive for binge eating).
GLP-1 levels were analysed using the ELISA platform (Mercodia, Uppsala, Sweden).Participants rated their hunger, fullness and urge to eat on a 15-cm visual analogue scale from 'not at all' to 'extremely' at the 15-, 30-, 45-, 60-, 90-and 120-min time points.This method has been validated for use in appetite research, and was utilized by Sysko et al. in a study of adolescents with severe obesity. 31,32The area-under-the-curve (AUC) for all the biomarkers and the visual analogue scale was computed for the entire standardized meal test.

| Statistical analysis
Descriptive analyses of characteristics and outcomes included means with standard deviations for continuous variables and counts with frequencies for categorical variables.Amongst randomized participants, regression analyses used 52-week BMI percent and total body fat mass change from randomization (post-meal replacement therapy period) as the primary outcomes with interaction terms between each predictor and the treatment group.Interaction terms were summarized at the 25th and 75th percentiles of the within-study population for each predictor.Unadjusted model predictors included age, BMI at randomization and sex.Those variables were all adjusted for in models evaluating the following predictors: CEBQ, PFS, EDE-Q, AUC for appetite regulation hormones and VAS.Predictors were chosen based on previous similar studies of response to AOM. 27,33 All analyses were performed using R (v4.2.3; R Core Team 2023). 34

| RESULTS
Sixty-six adolescents (mean age 16.0 ± 1.49; 47% female) were included with 33 participants in each of the exenatide XR and placebo groups (Table 1).BMI, BMI percentile of the 95th percentile T A B L E 1 Demographics, eating behaviours, appetite and appetite/satiety response to meal testing at randomization. and responses to the EDE-Q, CEBQ, PFS, visual analogue scale and AUC for the appetite/satiety regulation hormones prerandomization did not differ meaningfully between the groups.Prespecified predictors as continuous variables are summarized for the 25th and 75th intra-study population percentiles for participants (Table 2).
3.1 | Predictors of WLM for the intra-study population was significantly associated with WLM with exenatide XR versus placebo.BMI percent of the 95th percentile at randomization was also not statistically significantly associated with WLM with exenatide XR versus placebo (Table 2).Sex was not statistically significantly associated with the percentage change in BMI at 52 weeks (WLM) with exenatide XR versus placebo (À4.75% for males vs. À6.39%for females; 1.64, 95% CI [À7.91, 11.18]; p = 0.737).BMI at randomization was not significantly associated with total body fat mass change at 52 weeks (Table 3).

| Eating behaviours
Responses at the 25th or 75th percentile for the intra-study population regarding the EDE-Q (Binge Eating Behaviours), CEBQ (Food Responsiveness, Emotional Overeating, Enjoyment of Food or Satiety Responsiveness) or PFS (Hedonic Eating/Love of Food) at randomization were not statistically significantly associated with WLM in terms of BMI percent or total body fat mass change with exenatide XR versus placebo (Tables 2 and 3).
T A B L E 3 Predictors of exenatide-placebo treatment effect of WLM (total fat change [kg] at 52 weeks).b Binge eating question 2: On how many of these times did you have a sense of having lost control over your eating (at the time that you were eating)?

| Visual Analogue Scale
Self-reported feelings of hunger, urge to eat or fullness; or changes in hunger, fullness or urge to eat (AUC) during a meal were not significantly associated with WLM in terms of BMI percent or total body fat mass change with exenatide XR versus placebo (Tables 2 and 3).No other appetite/satiety hormone meal response at randomization was significantly associated with WLM in terms of BMI percent or total body fat mass change at 52 weeks (Tables 2 and 3).

| DISCUSSION
Previously, our group reported on the results of a randomized, placebo-controlled trial, which demonstrated that exenatide XR 2 mg weekly partly, but not statistically significantly, mitigated the propensity towards BMI regain after initial ≥5% BMI reduction via meal replacement therapy in adolescents with severe obesity. 22In this pre-specified secondary analysis examining predictors of WLM with exenatide XR, we found that a lower meal-induced leptin response following at least a 5% BMI reduction was associated with superior WLM in terms of BMI percent or total body fat mass change for participants on exenatide XR compared to participants on placebo ($9% BMI change difference between exenatide XR and placebo groups) at 52 weeks.
Leptin helps decrease food intake and regulates energy balance partially by inhibiting the production of neuropeptide Y (a hunger/ appetite-stimulating hormone). 35Adipocytes decrease leptin secretion during caloric restriction or fasting as a metabolic adaptation to prevent further weight loss or restore weight. 35Decreased leptin production is part of a metabolic adaptation to weight loss and is thought to be a contributing factor to rebound weight gain.For example, calorie restriction during religious fasting was associated with decreased leptin concentrations. 36,37Although short-term weight loss has been reported with calorie restriction, religious fasting observers regained weight once typical eating patterns were resumed. 38,39A lower leptin response to meals in some participants in this study after initial BMI reduction with meal replacement therapy may be a similar metabolic adaption to promote weight regain.
Although lower leptin levels can be associated with low satiety response, this self-reported eating behaviour was not found to be a statistically significant predictor of WLM with exenatide XR versus placebo.However, participants who endorsed a higher feeling of 'urge exenatide vs. placebo) of À8.42% compared to À1.89% BMI change for participants who had lower levels of 'urge to eat' at 52 weeks (À6.53 [À14.74,1.67]; p = 0.119).Although the association is not statistically significant, the treatment effect for the exenatide and placebo groups suggests that following an initial weight loss, participants with a stronger 'urge to eat' may have more WLM on exenatide XR than those on placebo.As GLP-1RAs alter insulin and leptin signalling in the hypothalamus and amygdala and partially counteract this specific metabolic adaptation to regain weight, it follows that those with lower leptin levels following weight loss may experience more WLM on exenatide XR than those on placebo.When interpreting these results, it should be acknowledged that lower leptin AUC may be related to lower BMI as adiposity mass is associated with leptin production; as adiposity decreases, so do leptin levels. 35However, lower leptin AUC prior to randomization was also associated with continued fat mass loss with exenatide versus gain with placebo.
These results are aligned with findings from our group previously showing that self-reported appetite was a significant predictor of BMI reduction with exenatide at 3 months. 27Specifically in the Nathan et al. study, participants who reported high measures of appetite had a greater loss in BMI on exenatide compared to placebo (À4.28% vs. 1.02%, p = 0.028). 27Comparisons between these studies need to be interpreted with caution since the focus of the current study was on WLM, not initial weight loss.However, whether it is to induce BMI reduction or to maintain BMI reduction following LMT, exenatide may be particularly effective in adolescents with a strong hunger urge or a propensity to have weight regain following an initial intervention.
Strengths of this study included a comprehensive assessment of self-reported and hormonal measures of appetite and satiety to describe predictors of WLM within the context of a double-blind, randomized, placebo-controlled trial.Both BMI as a surrogate measure of adiposity and total body fat mass were used.Limitations included a limited sample size and the relatively limited array of psychosocial variables measured.Race and ethnicity were not included as covariates in this analysis as they are social constructs and not biological variables.However, the sample size was 82% white and 88% non-Hispanic, which may limit the generalizability of these findings to other populations. 22Data on other confounding factors, such as physical activity, sleep quality and duration and food preferences, were not obtained in the parent study.This was a secondary analysis, and we did not adjust for multiple comparisons.Therefore, the findings of this study should be considered hypothesis-generating.
to eat' on the visual analogue scale at the beginning of a meal had BMI percent change from randomization (treatment effect of F I G U R E 1 Percent change in BMI from randomization to 52 weeks (WLM) by randomization leptin response.Participants whose leptin meal response at randomization was in the lower 25th percentile of the intra-study population had more WLM on exenatide XR than those on placebo.MRT, meal-replacement therapy.
Future studies should consider the potential influence of social and psychological factors (e.g., depression, family environment and social support) on WLM in larger, more diverse samples.As there are now two GLP-1RAs (semaglutide 2.4 mg/week and liraglutide 3.0 mg/day) FDA-approved for pediatric obesity and three GLP-1RAs (dulaglutide, exenatide XR, liraglutide 3.0 mg/day) FDAapproved for pediatric type 2 diabetes mellitus, increasingly prevalent obesity-related comorbidity, the ability to tailor obesity treatment with effective and long-term results is important.More potent GLP-1RAs have been FDA-approved for pediatric obesity since the recruitment and conclusion of this trial.Nonetheless, this study offers important self-reported and hormonal insight into possible predictors of WLM or weight regain with a GLP-1RA following initial dietaryinduced weight loss.This study also demonstrates the potential importance of metabolic response to weight loss (i.e., hunger symptoms; 'urge to eat') in tailoring long-term anti-obesity therapy.Clinicians treating adolescents with obesity may consider GLP-1RA after initial weight loss to prevent rebound weight gain, particularly in adolescent patients with low meal satiety or a strong pre-meal urge to eat.However, the mostly null findings of this study suggest that GLP1-RA treatment may potentially be a useful adjuvant therapy to LMT to maintain initial BMI and fat mass reduction for adolescents with obesity regardless of their sex, age, BMI, eating behaviours or other appetite regulation hormones, besides leptin.
Predictors of exenatide-placebo treatment effect of WLM (BMI percent change at 52-weeks).
3.1.1| Demographics Neither age (0.58, 95% CI [À5.67, 6.83]; 0.856) nor BMI at randomization (2.44, 95% CI [À4.93, 9.82]; 0.516) at the 25th or 75th percentile T A B L E 2 Abbreviations: AUC, area-under-the-curve; BMI, body mass index; CCK, cholecystokinin; GIP, glucose-dependent insulinotropic polypeptide; GLP-1, glucagon-like peptide-1; PP, pancreatic polypeptide; PYY, peptide YY. a Binge eating question 1: Over the past 28 days, how many times have you eaten what other people would regard as an unusually large amount of food (given the circumstances)?b Binge eating question 2: On how many of these times did you have a sense of having lost control over your eating (at the time that you were eating)?