Participants

The primary results of this study have been published [12]. Herein, we present secondary outcome data of a subsample of female participants (N = 17) that had DLW measured. Baseline characteristics of participants included in this analysis are shown in Table 1.

Table 1 Participant baseline characteristics.Energy expenditure at baseline

At baseline TDEE was comparable between BSG (2903 ± 552 kcal/d) and LCD (2679 ± 275 kcal/d). The prediction equation to estimate TDEE (kcal/d) was 1495 + 12.13 * body mass (in kg) –2.61 * height (in cm) + 3.03 * age (in years), R2 = 0.51, P = 0.02.

Change in body mass and energy imbalance

Participants in BSG lost an average of −16.0 ± 3.5 kg (equating to 12.0 ± 2.2% of baseline body mass) and −44.3 ± 16.4 kg (31.9 ± 6.5%) at 8 and 52 weeks, respectively; compared to weight loss of −8.8 ± 3.4 kg (7.2 ± 3.1%) and −4.3 ± 6.7 kg (3.1 ± 4.9%) observed in LCD (Fig. 1A). There was no statistically significant difference in weight loss after 8 weeks between the groups (−7.2 kg; 95% CI: −17.1, 2.7 kg; P = 0.15); however, participants in BSG showed a significantly larger decrease in body mass at 52 weeks than individuals in LCD (−40.5 kg; 95% CI: −50.7, −30.3 kg; P < 0.001).

Fig. 1: Intervention effects on body mass and energy expenditure.

Change in A body mass (kg) and B total daily energy expenditure (kcal/day) from baseline following bariatric surgery and low-calorie diet-induced weight loss. Data presented as mean (SD). Stars denote statistically significant (within-group) changes compared to baseline, assessed via a linear mixed model, including fixed factors for time and group, the interaction thereof, and a random factor for participant. Between-group differences listed in-text. *P < 0.05; **P < 0.01; ***P < 0.001.

Energy imbalance was calculated as the change in body energy stores (i.e., change in fat and fat-free mass). Over the first 8 weeks, daily energy imbalance was greater in BSG (−1768 ± 629 kcal/d) than in LCD (−654 ± 291 kcal/d) (−1114 kcal/d; 95% CI: −1451, −778 kcal/d; P < 0.001). Similarly, across the entire 52 weeks, daily energy imbalance in BSG (−914 ± 272 kcal/d) was greater than in LCD (−2 ± 44 kcal/d) (−907 kcal/d; 95% CI: −1253, −561 kcal/d; P < 0.001). Hence, daily energy imbalance from week 8 to week 52 in BSG (−753 ± 243 kcal/d) was also greater than in LCD (101 ± 67 kcal/d), as mirrored by the differential change in body mass (−851 kcal/d; 95% CI: −1178, −527 kcal/d; P < 0.001).

Changes in absolute total daily energy expenditure

The change in absolute TDEE somewhat followed the change in body mass (Fig. 1B, Table 2). In LCD, TDEE decreased by −256 ± 239 kcal/d at 8 weeks and recovered close to baseline levels (−84 ± 285 kcal/d) after weight also returned close to baseline at 52 weeks. In contrast, TDEE decreased in BSG by −552 ± 319 kcal/d at 8 weeks and this decrease was maintained at 52 weeks (−583 ± 418 kcal/d). Although there was some evidence of a between-group difference in change of total from baseline after 8 weeks (−296 kcal/d; 95% CI: −605,14 kcal/d; P = 0.06), the difference did not meet conventional levels of statistical significance; however, there was a statistically significant difference after 52 weeks (−499 kcal/d; 95% CI: −809, −190 kcal/d; P = 0.003), indicative of a greater decrease in TDEE in BSG.

Table 2 Measured compared to predicted total daily energy expenditure before and after surgery and diet-induced weight loss.Metabolic adaptation

After adjustment for change in body mass (Table 2), actual TDEE measured using DLW at 8 weeks was significantly lower than predicted in BSG (−358 kcal/d; −651, −64 kcal/d; P = 0.03), suggesting a metabolic adaptation (i.e., difference between measured and predicted TDEE) in response to this weight loss intervention. In contrast, there was no statistically significant difference in LCD (−149 kcal/d; −406, 107 kcal/d; P = 0.28). Similarly, there was no evidence of metabolic adaptation at 52 weeks in either BSG (−45 kcal/d; −339, 248 kcal/d; P = 0.77) or LCD (−68 kcal/d; −330,194 kcal/d; P = 0.63). Results were similar when using regression models that included body composition or body surface area (data not shown). There was no correlation between energy imbalance (i.e., change in body energy stores as determined by the amount of fat and fat-free mass lost and their respective energy values) and metabolic adaptation at either the 8-week (r = 0.27; 95% CI: −0.30, 0.70; P = 0.34) or the 52-week (r = 0.00; 95% CI: −0.57, 0.58; P = 0.99) time point across both groups pooled together or within each group separately (all P > 0.40).

Energy intake across the intervention period

During the initial 8 weeks, daily energy intake in BSG (788 ± 420 kcal/d) was significantly lower than in LCD (1896 ± 574 kcal/d) (−1069 kcal/d; 95% CI: −1534, −603 kcal/d; P < 0.001). Similarly, daily energy intake from week 8 to week 52 in BSG (1520 ± 292 kcal/d) was lower than in LCD (2662 ± 349 kcal/d) (−1086 kcal/d; 95% CI: −1544, −628 kcal/d; P < 0.001). There was also a time effect such that the average energy intake was significantly greater from week 8 to week 52 compared to the time between baseline and week 8 in both BSG (741 kcal/d; 95% CI: 385, 1097 kcal/d; P < 0.001) and LCD (758 kcal/d; 95% CI: 375, 1141 kcal/d; P < 0.001).

Physical activity in response to the interventions

Across both groups and time points, there were no changes (either within- or between-group) from baseline in these objective measures of physical activity (Table 3), except for week 52 in BSG, at which point step count was increased by over 50% and activity minutes were tripled compared to baseline (both P = 0.03). This increase in physical activity was also reflected by an increase in estimated AEE of similar magnitude at 8 weeks in BSG, although this did not reach statistical significance (P = 0.09). There were no correlations between step count and weight loss, energy imbalance, or metabolic adaptation at either time point. However, there was a significant relationship between minutes of physical activity and energy imbalance (r = −0.72; 95% CI: −0.95, −0.03; P = 0.04) at the last time point (Fig. 2A), which was mirrored by the relationship between minutes of physical activity and weight loss (r = −0.63; 95% CI: −0.91, 0.06; P = 0.07; Fig. 2B), although the latter did not reach the conventional threshold for statistical significance. Further, this relationship was driven by one individual who had the greatest weight loss (−69 kg, equating to 43% of baseline body mass) and had the highest level of physical activity (135 min/d). Removal of this individual from the data set resulted in a statistically non-significant relationship (P > 0.05). Similarly, there was a statistically significant relationship between activity energy expenditure and weight loss (r = −0.58; 95% CI: −0.86, −0.05; P = 0.04; Fig. 2C), which was similar to the relationship between activity energy expenditure and energy imbalance (r = −0.57, 95% CI: −0.86, 0.01; P = 0.05; Fig. 2D) at week 8, though the latter did not meet the threshold for statistical significance.

Table 3 Measures of free-living physical activity after surgery and diet-induced weight loss.Fig. 2: Relationship between physical activity and energy expenditure.

Correlation plots between A physical activity and energy imbalance or B change in body mass; and between C activity energy expenditure and change in body mass or D energy imbalance.