Exploratory analysis of eating- and physical activity-related outcomes from a randomized controlled trial for weight loss maintenance with exercise and liraglutide single or combination treatment

Weight regain following intentional weight loss is common and often occurs within five years. Biological adaptations to weight loss (e.g., reductions in resting and non-resting energy expenditure beyond what is expected from reduced body mass, and changes in appetite-regulating hormones) promote increased energy intake. Eating behaviors linked with obesity (low cognitive restraint, higher emotional and uncontrolled eating, heightened reward responses to energy-dense foods) and sedentary behavior may further contribute to weight regain. While GLP-1 receptor agonists (GLP-1RAs) can reduce appetite and promote satiety in the short term, their long-term effects on appetite during weight loss maintenance are understudied. Exercise is recommended to counteract reductions in energy expenditure after weight loss and may improve postprandial satiation, but could also increase appetite to match higher energy needs. The research aim was to determine how one-year maintenance strategies—moderate-to-vigorous intensity exercise, liraglutide 3.0 mg, or their combination—affect appetite, eating behavior, sedentary behavior, and non-exercise physical activity after an initial low-calorie diet-induced weight loss, compared with placebo.

Prior research indicates that appetite and energy expenditure adapt to weight loss, promoting weight regain (e.g., adaptive thermogenesis and hormonal changes). Obesity-associated eating behavior patterns include low cognitive restraint and elevated emotional/uncontrolled eating, with increased hedonic responses to high-fat foods. Sedentary behavior contributes to obesity risk; debates persist regarding whether weight loss induces decreases in non-exercise activity or increases in sedentary time. GLP-1RAs reduce hunger and increase fullness acutely and over weeks; semaglutide has been shown to reduce food liking/wanting for energy-dense foods over 12 weeks, while one-year liraglutide adjunct to intensive behavioral therapy showed early appetite effects that waned by 1 year despite greater weight loss. Exercise may enhance meal-induced satiety and reduce liking/wanting for high-fat foods, but can also increase appetite commensurate with energy expenditure. Previous primary results from this trial showed that an 8-week low-calorie diet-induced ~13 kg weight loss was maintained with exercise or liraglutide alone, and further reduced with their combination versus weight regain with placebo after one year.

Design: Randomized, placebo-controlled, double-blind (for medication), 2×2 factorial trial of one-year weight loss maintenance following an initial low-calorie diet. Registration: EudraCT 2015-005585-32; NCT04122716. Ethics: Approved by Regional Ethics Committee for the Capital Region of Denmark (H-16027082) and Danish Medicines Agency; conducted per Declaration of Helsinki and ICH GCP; written informed consent obtained. Participants: Adults aged 18–65 years with obesity (BMI 32–43 kg/m²), without diabetes, recruited at Hvidovre University Hospital and the University of Copenhagen, Denmark (enrolled 2016–2018). Of 215 enrolled, 195 achieved ≥5% weight loss on diet and were randomized; 166 completed one-year assessments; 130 adhered to interventions and were included in per-protocol analyses. Baseline (per-protocol): mean BMI 36.9±2.9 kg/m², age 45±12 years, 62% women; sedentary time 10.9±1.5 h/day. Interventions: All underwent an 8-week low-calorie diet (~800 kcal/day; two Nutricia Nutridrink meals plus guidance). Post-diet, participants were randomized 1:1:1:1 (stratified by sex and age <40/≥40) to: (1) Exercise+placebo injection; (2) Liraglutide 3.0 mg/day; (3) Exercise+liraglutide; (4) Placebo injection. Liraglutide/placebo dose titration: start 0.6 mg/day with weekly 0.6 mg increments to 3.0 mg/day or maximally tolerated. Blinding applied to liraglutide vs placebo. Exercise arm: 7-week ramp-up, then two supervised vigorous indoor cycling (30 min) plus 15 min circuit training sessions per week, and two additional individual sessions at moderate-to-vigorous intensity per week. Participants in non-exercise arms were instructed to maintain usual physical activity. Outcomes and Assessments: Measured pre- and post-diet and after 1 year of maintenance. - Eating behavior: Three-Factor Eating Questionnaire-R18 (TFEQ-R18), yielding Cognitive Restraint, Emotional Eating, and Uncontrolled Eating scores (0–100 scale; higher indicates greater tendency). - Appetite: Standardized liquid mixed meal test (600 kcal; 74 g CHO, 23 g fat, 24 g protein). Visual analog scales (0–100 mm) for prospective food consumption, hunger, fullness, satiety at fasting and repeatedly over 180 min postprandial; area under the curve computed. Overall Appetite Suppression (OAS) score = (satiety + fullness + [100 − hunger] + [100 − PFC]) / 4. - Food preferences: Leeds Food Preference Questionnaire (culturally adapted) assessing explicit liking (VAS) and implicit wanting (response time–weighted algorithm) for four categories: high-fat sweet, low-fat sweet, high-fat savory, low-fat savory. - Physical activity/sedentary behavior: Wrist-worn triaxial accelerometers (GENEActiv; 75 Hz) worn for 7 consecutive days at five time points (pre-diet, end-of-diet, and weeks 13, 26, 52). Data processed with GGIR (autocalibration, non-wear detection, Euclidean norm minus gravity, 5-s epochs). Inclusion required >16 h/day on ≥3 days/week. Time spent sedentary, light-intensity, and MVPA (moderate-to-vigorous physical activity) derived using established cut-points. Because cycling may be underdetected by wrist accelerometry, the IPAQ-Short Form was also administered for self-reported sitting time, walking, and MVPA. Exercise sessions used sports watches with heart rate chest straps (Polar A300) to estimate exercise energy expenditure, duration (min/week), and intensity (% max HR). Adherence definitions (per-protocol): Medication—administered 2.4 or 3.0 mg/day liraglutide/placebo for ≥75% of intervention period. Exercise—achieved at least 75% of WHO adult recommendations (≥150 min moderate, or ≥75 min vigorous, or equivalent weekly). Statistics: Primary outcome (body weight) previously published; current outcomes were pre-specified exploratory/secondary. Linear mixed models with fixed effects of time, group, sex, age group, and time×group interaction; unstructured covariance; repeated measure at participant level. Non-normal outcomes log-transformed and back-transformed (reported as ratios). Analyses performed for per-protocol and intention-to-treat (all randomized) populations; missing at random assumed. Multiple linear regression tested associations between changes in behavioral/activity measures and percent body weight change, adjusting for age, gender, study group, and baseline weight. Alpha 0.05; exploratory without multiplicity adjustment. SAS EG v7.15 used. Harms: Gastrointestinal adverse events (nausea, diarrhea, vomiting) more frequent with liraglutide and combination than exercise or placebo; decreased appetite reported by 4% placebo, 8% exercise, 37% liraglutide, 33% combination; 16 serious adverse events (8%), with 5 leading to discontinuation (3 exercise, 1 liraglutide, 1 combination).

- Population and weight outcomes (per-protocol): After an ~13.7 kg diet-induced weight loss, at 1 year the placebo group regained 6.1 kg. Weight loss was maintained with exercise (0.7 kg; difference vs placebo −5.3 kg; 95% CI, −9.4 to −1.2) and liraglutide (−1.9 kg; difference vs placebo −8.0 kg; 95% CI, −11.7 to −4.2). The combination group lost an additional −6.0 kg (difference vs placebo −12.1 kg; 95% CI, −16.1 to −8.1). - Placebo arm natural history: Postprandial appetite increased versus post-diet baseline: prospective food consumption +19.2% (P=0.01) and hunger +21.9% (P=0.002); fullness −13.4% (P=0.01) and satiety −13.6% (P=0.008). Overall Appetite Suppression (OAS) decreased −14.0% (P=0.001). Sedentary time increased by 31 min/day (P=0.03) and self-reported sitting by 25 min/day. - Liraglutide: Prevented increases in postprandial prospective food consumption (−2.0% vs +19.2% placebo; P<0.05) and hunger (−1.7% vs +21.9% placebo; P=0.02). Maintained OAS (−0.3% vs −14.0% placebo; P=0.02). Fasting satiety increased vs placebo (+24.7% vs −9.4%; P=0.02). Reduced explicit liking for high-fat sweet foods (−16.1%; P=0.04) and decreased explicit liking and implicit wanting for high-fat savory foods (both P<0.001). No significant changes in sedentary time or accelerometer-derived activity. - Exercise: Program achieved sustained exercise energy expenditure (mean ~1661±738 kcal/week) with duration ~156±54 min/week at ~78±4% HRmax. OAS decreased −12.1%, similar to placebo (P=0.9). Implicit wanting for high-fat sweet foods increased (+35%; P=0.03), comparable to placebo. Despite increased exercise energy expenditure, no increase beyond placebo in appetite or sedentary time; weight loss was maintained (treatment effect −5.3 kg vs placebo). - Combination (exercise + liraglutide): Cognitive restraint increased vs placebo (+12.8% vs −8.6%; P=0.04). Sedentary time decreased by 10 min/day versus +31 min/day with placebo (difference −41 min/day; 95% CI, −82.3 to −0.2; P=0.049); self-reported sitting time decreased more than placebo (P=0.003). OAS change was not significantly different from placebo (−7.9% vs −14.0%; P=0.37). Achieved additional weight loss (difference −12.1 kg vs placebo). In ITT analyses, self-reported sitting remained lower vs placebo, while accelerometer sedentary time difference was attenuated due to non-adherence. - Associations with weight change (ITT): Greater increases in cognitive restraint and MVPA were associated with less weight regain during maintenance (β=−0.17 per unit CR change; P=0.003; and β=−0.078 for MVPA; P=0.016). Increased uncontrolled eating tended to relate to weight regain (P=0.07).

The study shows that after diet-induced weight loss, untreated maintenance (placebo) is accompanied by increased postprandial appetite and increased sedentary time, likely contributing to substantial weight regain within one year. Liraglutide counteracted the weight loss–induced rise in appetite (reducing hunger and prospective consumption, increasing satiety) and maintained weight loss, with additional reductions in hedonic preference for high-fat foods; activity levels were unchanged. Exercise produced substantial added energy expenditure without further elevating appetite beyond placebo and helped maintain weight loss; importantly, sedentary time did not increase relative to placebo. The combination of exercise and liraglutide appeared to target both domains: it improved cognitive restraint (an intent to control intake) and reduced sedentary time, aligning with the greatest additional weight loss. Across participants, increases in cognitive restraint and MVPA were associated with better weight maintenance, highlighting behavioral and activity targets. Collectively, targeting appetite regulation and sedentary behavior concurrently may offer the most effective strategy to prevent weight regain following weight loss.

Following weight loss, increases in appetite and sedentary time threaten maintenance. Liraglutide prevented appetite increases and supported weight loss maintenance; exercise increased energy expenditure without worsening appetite and aided maintenance; their combination enhanced cognitive restraint and reduced sedentary time, aligning with the greatest additional weight loss. These findings suggest that comprehensive maintenance strategies addressing both eating-related behavior and sedentary behavior are optimal to prevent weight regain. Future research should evaluate longer-term durability beyond one year, mechanisms linking GLP-1RA with hedonic eating changes, optimization of combined pharmacologic-exercise regimens, and generalizability across populations and settings.

- Exploratory analyses without adjustment for multiplicity limit definitive causal inference. - Per-protocol analyses (adherent participants) may introduce subset selection bias; ITT analyses were also provided but show some attenuation due to non-adherence. - Wrist-worn accelerometry may underdetect cycling-based exercise, potentially underestimating MVPA; IPAQ-SF was included to complement objective measures. - Missing data assumed missing at random. - Sample size powered for body weight (primary outcome), not specifically for all behavioral secondary outcomes. - The study population (adults with obesity without diabetes in Denmark) may limit generalizability to other populations or clinical contexts.