Adaptive thermogenesis at the level of resting energy expenditure after diet alone or diet plus bariatric surgery

Abstract

Objective:

To compare the magnitude of adaptive thermogenesis (AT), at the level of resting energy expenditure (REE), after a very-low energy diet (VLED) alone, or combined with Roux-en-Y Gastric bypass (RYGB) or sleeve gastrectomy (SG), and to investigate the association between AT and changes in appetite.

Methods:

44 participants with severe obesity underwent 10 weeks of a VLED alone, or combined with RYGB, or SG. Body weight/composition, REE, subjective appetite feelings, and plasma concentrations of gastrointestinal hormones were measured at baseline and week 11. AT, at the level of REE, was defined as a significant lower measured versus predicted (using a regression model with baseline data) REE.

Results:

Participants lost 18.4±3.9 kg body weight and experienced AT, at the level of REE (−121±188 kcal/day, P<0.001), with no differences among groups. The larger the AT, at the level of REE, the greater the reduction in fasting ghrelin concentrations, and the smaller the reduction in feelings of hunger and desire to eat in the postprandial state.

Conclusion:

Weight loss modality does not seem to modulate the magnitude of AT, at the level of REE. The greater the AT, at the level of REE, the greater the drive to eat following weight loss.

Introduction

Adaptive thermogenesis (AT) is defined as a reduction in total energy expenditure (TEE), or any of its components (resting or non-resting energy expenditure (EE)), lower than predicted, given the loss of fat mass (FM) and fat-free mass (FFM). The existence of AT remains controversial, with some studies reporting it (1, 2) and others not (3, 4). The energy balance (EB) status of the participants when measurements of EE and body composition are taken are likely to explain some of these differences (5, 6). Although AT does not seem to be a risk factor for weight regain (7), it might lead to a resistance to weight loss. Studies have shown that individuals experiencing a larger AT, lose less weight and FM in response to a low-energy diet (7), and take longer to reach their weight loss goals (8).

Bariatric surgery is considered the best treatment option for obesity (9), with Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) being the two most common procedures (10). Like diet-induced weight loss, the existence of AT after bariatric surgery is also inconsistent (11–16). Additionally, some studies show that the magnitude of AT differs between bariatric procedures (14, 15), while others not (16, 17). Timing post-bariatric surgery, magnitude of weight loss and component of EE measured (TEE or REE) are likely to contribute to some of these differences. Indeed, most studies looking at AT post-bariatric surgery have measured REE (13, 15–18), with only a few reporting TEE (11, 12, 14). It remains to be determined if AT after bariatric surgery is the result of energy restriction, weight loss, or the anatomical and physiological changes that occur post-RYGB and SG, or a combination of them. A comparison of AT following weight loss induced by diet or bariatric surgery would help to solve this issue.

Some studies have reported AT to be positively correlated with weight loss after a lifestyle intervention (2, 19) and percent total weight lost (%TWL), as well as FM loss, after RYGB and SG (13–15), while others report no association between AT, and FM regardless of surgical procedure (17, 18). Overall, the association between AT, weight, and FM loss after diet and bariatric surgery remains inconsistent.

Despite the reduction in EE observed with weight reduction, weight loss is also followed by increased hunger feelings (1, 20, 21), likely as a result of increased secretion of the orexigenic hormone, ghrelin (5, 21). Additionally, some studies have shown that weight loss leads to a reduction in the postprandial secretion of satiety hormones, peptide YY (PYY), and cholecystokinin (CCK) (20, 21), while others report an increase in CCK, PYY_3–36 and total glucagon-like peptide-1 (GLP-1) (22, 23). There is some evidence to suggest that AT can be modulated by hormones involved in appetite and metabolism, namely insulin and leptin (24, 25). A decrease in leptin in response to energy restriction has been found to be an important determinant of AT (26, 27). A potential association between AT and changes in appetite could provide insight into how the body adapts to weight loss and whether these two responses on both sides of the energy balance (EB) equation are interconnected.

Therefore, the main aim of this study was to compare the magnitude of AT after diet alone or plus RYGB, or SG. Secondary aims were to investigate potential associations between AT and changes in FM and body weight, as well as changes in appetite markers (gastrointestinal hormones (GI) and subjective feelings of hunger).

Methods

Study design

The data included in this secondary analysis was previously collected from the parent study: The effect of DIet-induced weight loss versus Sleeve gastrectomy and Gastric bypass on APpetite (DISGAP). The DISGAP study is a three-armed prospective nonrandomized controlled trial, comparing how a similar weight loss induced by very low-energy diet (VLED) alone, or in combination with bariatric surgery, impacts both homeostatic and hedonic appetite markers. This study was approved by the local ethics committee (REK Midt-Norge, Norway) (ID number 2019/252). An outline of the DISGAP study can be seen in Figure 1 (22).

Figure 1.

DISGASP study design [nonrandomized study]. Assessments were performed at baseline (BL) and week 11 (W11). BL: baseline; SG: sleeve gastrectomy; RYGB: Roux-en-Y Gastric Bypass; VLED: very low-energy diet; WL: weight loss; W11: week 11.

Participants

Participants included in this analysis were adults with severe obesity scheduled for SG or RYGB at two local hospitals in the Central Norway Health Region. Patients on a waiting list for bariatric surgery, or who did not meet eligibility criteria for it, and individuals with severe obesity from the local community, comprised the control group (diet intervention alone). Controls were matched for body mass index (BMI), age, and sex of the surgical groups. Recruitment and data collection took place between September 2019 and January 2022. A flow diagram of the DISGAP study can be seen in Figure 2 (22).

Figure 2.

Flow diagram of the study. SG: sleeve gastrectomy; RYGB: Roux-en-Y gastric bypass.

Since the study is a secondary analysis, the inclusion and exclusion criteria were set by the parent study (22). Participants provided written informed consent before joining the study. All participants had to be weight stable (self-reported), with no more than a 2-kg weight change within the last 3 months, and not enrolled in any other obesity treatment or behavioral program.

Detailed Protocol

All participants followed a formula based very low energy diet (VLED) (Lighter Life, Harlow) (750 kcal, 26 E% fat, 36 E% carbohydrates, 5 E% fiber, and 33 E% protein), for 10 weeks under the guidance of a registered dietitian. Patients scheduled for SG and RYGB initiated the VLED 2 weeks prior to surgery and continued for another 8 weeks. The surgical groups were instructed to consume only fluid food packs the first weeks postoperatively, gradually increasing the texture of the food. All participants were asked to fill out a self-reported food diary. At weekly scheduled follow-ups, food diaries were discussed, side effects recorded, body weight monitored, and acetoacetate (a ketone body) measured in urine with ketostix (Bayer Ketostix 2880 Urine Reagent Test Strip, Ascensia Diabetes Care), as a measure of dietary compliance (22). Bariatric surgeries were performed using standard laparoscopic procedures, as previously described (22).

Outcome variables

The following variables were collected at baseline (BL), before the start of the VLED for all groups, and after the 10-week intervention (W11), after an overnight fast (at least 10 hours).

Body weight and composition

Body weight (BW) and composition (FM, and FFM) were measured with air-displacement plethysmography (BodPod, COSMED) (22).

Resting Energy Expenditure

After resting for 15 minutes, REE was measured for 30 minutes with a computerized, open-circuit, indirect calorimetry system with a ventilated canopy (Vmax Encore 29 N; Care Fusion, Baesweiler, Germany). Oxygen uptake (VO₂) and carbon dioxide production (CO₂) were measured continuously, and values were averaged at 1-minute intervals (22).

Plasma concentrations of Gastrointestinal (GI) hormones

Blood samples were drawn in fasting (4-mL EDTA-coated tubes), every 15 minutes for the first hour after a standardized breakfast, and then at 30-minute intervals until 150 minutes. The breakfast was a 200-mL commercial low-glycemic drink (Diben Drink, Fresenius Kabi Norge AS) (300 kcal, 42 E% fat, 35 E% carbohydrates, 3 E% fiber, and 20 E% protein), and participants drank it slowly over a 15-minute period (22).

Details regarding blood collection and handling have been previously reported (22). Samples were analyzed for AG and total PYY using a Human Metabolic Hormone Magnetic Bead Panel (HMHEMAG-34 K, Merck KGaA). CCK and total GLP-1 were analyzed using “in-house” radioimmunoassays. A ketone body assay kit (MAK134, Sigma-Aldrich) was used to measure ß-hydroxybutyric acid (ßHB) plasma concentrations.

Subjective appetite ratings

Appetite ratings (hunger, fullness, desire to eat, and prospective food consumption [PFC]) were assessed using a 10-cm visual analog scale at fasting, immediately after the standardized breakfast, and every 30 minutes for a period of 2.5 hours.

Statistical analysis

Statistical analysis was performed with SPSS version 29 (SPSS Inc., Chicago, IL), and data presented as mean ± SD. Statistical significance was set at P < 0.05 and normality of variables was assessed by visual inspection of histogram and Shapiro-Wilk test. Participants were included in the analysis if they had REE data (n=44) measurements at baseline and week 11.

Changes in BW, FM, and FFM over time (baseline and week 11) and differences between groups were analyzed using a mixed ANOVA, with Bonferroni correction for post hoc pairwise comparisons. The existence of AT at the level of REE was tested using a paired t-test comparing measured REE (REE_m) and predicted REE (REE_p) at baseline and week 11.

An equation to predict REE was derived from baseline data of all the participants.

$$\mathrm{Model}:{\mathrm{REE}}_{\text{p}}\left(\mathrm{kcal}/\text{d}\right)=318–\left(5.0\times \mathrm{Age}\right)+\left(30.3\times \mathrm{FFM}\left[\mathrm{kg}\right]\right)+\left(3.0\times \mathrm{FM}\left[\mathrm{kg}\right]\right)+\left(65.6\times \mathrm{Sex}\right)$$

R² = 0.70; p < 0.001

An additional equation to predict REE was derived from baseline data of all participants without FM.

$$\mathrm{Model}:{\mathrm{REE}}_{\text{p}}\left(\mathrm{kcal}/\text{d}\right)=458–\left(5.5\times \mathrm{Age}\right)+\left(31.3\times \mathrm{FFM}\left[\mathrm{kg}\right]\right)+(54.5\times \mathrm{Sex})$$

R² = 0.70; p < 0.001

Bivariate correlation analyses were performed between AT at week 11 and BW, FM and FFM loss, as well as changes in appetite feelings and GI hormones using Pearson or Spearmen correlation coefficients, depending on the distribution of the variables. Changes over time were calculated by subtracting baseline data from week 11, with negative values indicating a reduction and positive values indicating an increase over time.

Linear regression analyses were conducted to determine if AT, at the level of REE, was a significant predictor of changes in appetite markers (subjective feelings of hunger and GI hormones) after adjusting for age, sex, and group.

Results

The demographic and anthropometric characteristics of the participants at baseline and after the intervention (week 11) can be seen in Table 1. Forty-four participants (n = 15 VLED, n = 15 SG, n = 14 RYGB), mainly females (70.4%), were included in the analysis. They had an average BMI of 40.7 ± 3.9 kg/m² and an age of 43.5 ± 11.3 years. There were no significant differences in sex distribution, or any of the anthropometric variables measured at baseline, among groups. However, participants in the RYGB group were significantly older than those in the SG group (P=0.045).

Table 1.

Demographic and anthropometric characteristics of the participants by group at baseline and week 11

	Baseline			Week 11 (W11)			P-value Time	P-value Group
	Control	SG	RYGB	Control	SG	RYGB
n	15	15	14	15	15	14
Age (yr)	44.5 ± 11.1	38.0 ± 9.6	48.1 ± 11.5	__	__	__		0.045
Females, %	73	73	64	__	__	__		NS
BMI (kg/m2)	40.0 ± 3.7	40.5 ± 3.1	41.6 ± 4.9	33.7 ± 4.0	34.1 ± 2.9	35.0 ± 5.2	<0.001	NS
Weight (kg)	115.1 ± 20.8	120.9 ± 14.7	120.6 ± 13.5	97.8 ± 19.6	101.8 ± 12.6	101.7 ± 12.1	<0.001	NS
FFM (%)	52.7 ± 5.4	51.8 ± 4.2	52.9 ± 7.0	58.8 ± 6.2	56.7 ± 5.1	57.3 ± 8.3	<0.001	NS
FFM (kg)	61.2 ± 13.2	62.5 ± 8.9	63.7 ± 10.0	57.3 ± 12.1	57.6 ± 7.9	57.9 ± 8.7	<0.001	NS
FM (%)	46.9 ± 5.1	48.2 ± 4.2	47.0 ± 7.0	41.2 ± 5.1	43.3 ± 5.1	42.7 ± 8.3	<0.001	NS
FM (kg)	53.9 ± 11.2	58.4 ± 8.9	56.7 ± 11.5	40.5 ± 11.2	44.1 ± 8.2	43.8 ± 11.6	<0.001	NS
βHB (mM)	0.2 ± 0.2	0.2 ± 0.4	0.2 ± 0.3	0.8 ± 0.4	1.1 ± 0.9	0.6 ± 0.9	<0.001	NS

Note: Data presented as mean ± SD. There were significant differences between BMI, weight, FM, and FFM over time (p < 0.001) with no significant time*group interaction.

Abbreviations: N: sample size; NS: non-significant; BMI: body mass index; FFM: fat-free mass; FM: fat mass; SG: sleeve gastrectomy; RYGB: Roux-en-Y gastric bypass.

Throughout the 10-week intervention, participants lost on average 18.4 ± 3.9 kg BW, 13.5 ± 3.2 kg FM and 4.8 ± 2.2 kg FFM (P<0.001 for all), with no significant differences among groups. Participants were all in nutritional-induced ketosis at week 11, as seen by ßHB plasma concentrations above 0.3 mM, and no differences between groups were observed. Participants experienced a significant reduction in REE_m of 322 ± 270 kcal/day (P<0.001) over time, with no significant differences among groups. REE_m was significantly lower than REE_p post-intervention (−121±188 kcal/day, P<0.001), indicating AT, at the level of REE, with no significant differences among groups (Table 2). AT indicates changes in REE associated with the interventions (VLED alone or VLED plus surgery) after accounting for the loss of FM and FFM.

Table 2.

Resting energy expenditure (REE) and adaptive thermogenesis at the level of REE over time

	Baseline			Week 11 (W11)			P-value Time	P-value Group
	Control	SG	RYGB	Control	SG	RYGB
n	15	15	14	15	15	14
REEm (kcal/d)	2261.6±543.5	2218.8±325.5	2251.7±376.7	1934.7±387.6	1891.5±357.5	1940.5±351.4	<0.001	NS
REEp (kcal/d)	2175.7±445.8	2262.4±305.2	2247.4±302.0	2019.1±417.1	2072.6±267.0	2035.9±261.4	<0.001	NS
REEm-REEp (kcal/day)				−84.4±217.5	−181.1±181.1	−95.4±156.6		NS

Note: Data presented as mean ± SD. There was no effect of group or time*group interaction was seen for RMR_m and RMR_p.

Abbreviations: NS: non-significant; REE: resting energy expenditure; REE_m: REE measured; REE_p: REE predicted.

No significant correlation was seen between AT (kcal/day) at the level of REE and BW [kg] (r=−0.076, P=0.623), FM [kg] (r=−0.081, P=0.602), or FFM loss [kg] (r=−0.021, P=0.892). Additionally, when REE was predicted without accounting for FM, no association was seen between AT at W11 (−171±188 kcal/day, P<0.001) and weight or FM loss (data not shown). However, there was a significant correlation between AT (kcal/day), at the level of REE, at week 11 and changes in fasted ghrelin plasma concentrations (r=−0.310, P=0.046, n=42), indicating that the greater the AT the greater the decrease in ghrelin (see Figure 3A). Additionally, there was a significant correlation between AT (kcal/day), at the level of REE, at week 11 and changes in ratings of hunger (see Figure 3B) (r=0.332, P=0.028) and desire to eat (see Figure 3C) (r=0.396, P=0.008) in the postprandial state, indicating that the greater the AT, the smaller the reduction in postprandial ratings of hunger and desire to eat.

Figure 3.

Scatterplots for the association between AT at the level of REE and changes in plasma concentrations of fasting ghrelin (A) in the fasting state, and changes in the ratings of hunger (B), and desire to eat in the postprandial state (C). VLED (black filled circle), SG (open circle), RYGB (light grey filled square). A negative value on the y axis indicates metabolic adaptation.

Further, regression analysis showed that AT, at the level of REE, was a significant predictor of changes in fasted ghrelin (P=0.044), hunger (P=0.029) and DTE (P=0.008) in the postprandial state, even Safter adjusting for covariates (see Table 3).

Table 3.

Linear regression model predicting changes in ghrelin plasma concentrations in the fasting state, and subjective ratings of hunger and desire to eat in the postprandial state

	Predictors	β Coefficient (95% CI)	P-value	R2, adjusted
Model A			<0.001	0.40
Ghrelin	Constant	(−138.81 to −1.47)	0.046
	Age	0.31 (0.28 to 2.94)	0.019
	Sex	0.29 (4.14 to 68.23)	0.028
	Group	−0.44 (−50.13 to −13.69)	0.001
	AT_W11	0.33 (0.03 to 0.18)	0.010
Model B			0.085	0.07
Hunger	Constant	(−5855.72 to 1001.98)	0.160
	Group	−0.06 (−1883.32 to 1251.50)	0.686
	AT_W11	−0.33 (−14.50 to −0.83)	0.029
Model C			0.020	0.13
Desire to eat	Constant	(−2942.21 to 1339.82)	0.454
	Group	−0.13 (−1416.60 to 540.82)	0.371
	AT_W11	−0.40 (−10.21 to −1.68)	0.008

Note: Model A: fasting ghrelin change; Model B: change in hunger (tAUC); Model C: change in DTE (tAUC). Sex: 1 for female and 2 for male.

Abbreviations: AT: adaptive thermogenesis at the level of REE

Changes in ßHB plasma concentrations were negatively correlated with insulin both in the fasting and postprandial (tAUC) states (r=−0.427, p=0.005 and r=−0.340, p=0.03, respectively), as well as fasting glucose (r=−0.512, p=<0.001) at week 11. Additionally, a positive correlation was seen with changes in postprandial feelings of fullness (tAUC) (r=0.362, p=0.02). No correlation was observed between changes in ßHB and AT, at the level of REE.

Discussion

This secondary analysis aimed to compare the magnitude of AT, at the level of REE, following diet and bariatric surgery, with energy and macronutrient composition of the diet, as well as changes in BW, FM and FFM being similar across groups. We found no differences in the magnitude of AT between VLED alone, and VLED in combination with bariatric surgery. Secondly, no correlation was seen between AT and BW, FM, or FFM loss. Lastly, there was a significant positive correlation between AT and changes in ghrelin concentration in the fasting state and a negative correlation with changes in subjective feelings of hunger and desire to eat in the postprandial state.

AT, estimated as changes in REE associated with the interventions (VLED alone or VLED plus surgery) after accounting for the loss of FM and FFM, was seen in the present study (−121±188 kcal/day, P<0.001), after an average weight loss of 18.4±3.9 kg, and with participants in negative EB. Additionally, no significant differences between groups were observed for AT. Previous studies have consistently shown AT 6 months after RYGB and SG, both at the level of REE and TEE (11, 12). Additionally, Tam et al showed AT (TEE) at 8 weeks and 1 year post RYGB and SG, with no differences between procedures (14). Of the studies reporting AT at the level of REE, some report no differences between RYGB and SG (17), while others show differences between the two bariatric procedures (14, 15). However, from our knowledge only one study has compared AT after bariatric surgery and diet. Tam and colleagues found no differences in AT at the level of TEE, between RYGB, SG, and diet at 8 weeks and 1 year follow-up, despite AT at the level of REE being greater after SG compared with RYGB and diet at both time points (14). In this study, weight loss cannot explain the differences in the magnitude of AT between SG and RYGB as weight loss was similar between the two procedures at both time points. Nonetheless, the study is limited given that the composition of the diet and magnitude of weight loss were not matched across groups.

No correlation was seen between AT and BW, FM, or FFM loss in the present study, which is in line with other studies looking at AT post-bariatric surgery (17, 18). Studies from our research group show AT to be associated with less weight and FM loss in response to energy restriction, in individuals with overweight or obesity (7), and to slow down weight loss rate (28), suggesting that AT might contribute to resistance to weight loss (7, 28). However, one study found a positive correlation between AT and FM, but not FFM loss post-RYGB (13). Fothergill and colleagues showed, in the Biggest Loser study, that individuals who maintained a greater weight loss at 6 years experienced the largest AT (2). Overall, evidence regarding the correlation between AT following diet-induced weight loss, or obesity surgery, and BW, FM and FFM loss is inconclusive, and more research is needed.

Diet-induced weight loss is associated with increased hunger and drive to eat, likely due to an increase in the secretion of ghrelin, a hunger-stimulating hormone (20). Changes in the postprandial secretion of satiety hormones following diet-induced weight loss remain inconsistent (21, 23), but postprandial feelings of fullness have been reported to increase (20). Of relevance, the increase in hunger feelings and ghrelin concentration seen with weight loss is minimized or absent if participants are under nutritional-induce ketosis (29–31), a metabolic state that results from lack of carbohydrates and/or energy in the diet, such as when following a VLED. Indeed, in the present DISGAP study, participants were ketotic at week 11, as seen by their βHB plasma concentrations (a marker of ketosis), and the diet group experienced no change in plasma ghrelin concentrations, either in the fasting or postprandial states (22). Both RYGB and SG have been shown to increase the postprandial secretion of satiety peptides GLP-1 and PYY (32), but to have different effects on ghrelin (32, 33). SG is followed by a reduction in ghrelin concentrations due to the removal of the fundus of the stomach, where most of ghrelin is produced, while changes in ghrelin following RYGB remain inconsistent (33). In the DISGAP study, both the SG and RYGB experienced a reduction in ghrelin in the fasting state, and an increase in GLP-1 and PYY in the postprandial state (22). Overall, diet-induced weight loss seems to be associated with increased drive to eat in the fasting state, while bariatric surgery is associated with an improved postprandial satiety response.

A positive moderate correlation was seen between AT and changes in ghrelin in the fasting state, and a negative correlation with changes in subjective feelings of hunger and desire to eat in the postprandial state. In line with this, our research group found in individuals with obesity who underwent a low-energy diet, a larger AT at the level of REE was correlated with a greater increase in hunger and desire to eat in the fasting state, and a smaller suppression of hunger and desire to eat in the postprandial state (34). A few studies have looked at potential association between AT and hormones involved in metabolism and appetite (2, 14, 25, 26, 35). Müller and colleagues found no association between AT and changes in ghrelin, leptin, or triiodothyronine (T3) after caloric restriction, but an association with decreased insulin secretion (25). Other studies have also found no associations between AT and leptin, T3, thyroxine (T4), or thyroid-stimulating hormone (TSH) plasma concentrations after combined lifestyle interventions (2, 14, 25). However, AT was reported to be associated with changes in fasted leptin plasma concentrations in individuals who underwent RYGB, or a combined lifestyle intervention (35), or 24-hour leptin in participants undergoing a diet intervention (26). In line with our current study, no association between AT and changes PYY plasma concentrations were reported by McNeil and colleagues (36). Inconsistencies between AT and appetite hormones might be due to the hormonal fractions measured, as well as the component of EE analyzed.

Our results indicate that AT occurs regardless of weight loss modality, with an average reduction in REE 120 kcal/day below predicted values. If diet composition and weight, FM and FFM loss are matched, bariatric surgery does not seem to offer any benefit regarding the expression, or magnitude, of AT. Despite a positive correlation between AT and changes in ghrelin in the fasting state, with a greater AT being associated with a greater reduction in fasted ghrelin plasma concentrations, this did not translate into correlations with subjective feelings of hunger in the fasting state. A greater AT was associated with a smaller reduction (or greater increase) in feelings of hunger and desire to eat after the standardized breakfast. If patients who experience AT also feel hungrier and have a greater desire to eat in the postprandial state, they may initiate the following meal earlier and consume more food, therefore compromising diet adherence and weight loss outcomes. This needs, of course, to be interpreted in the context of ketosis, known to suppress appetite (29, 31, 37). Future studies should replicate these findings outside of ketosis, and with participants in EB, and to try to identify individuals at risk of experiencing a larger AT following weight loss, so that extra measures could be put in place to improve diet adherence and weight loss outcomes, namely increasing dietary fiber and protein, known to increase satiety (38).

We have previously suggested that the changes in appetite and EE seen with diet-induced weight loss are not a compensatory mechanism driving weight regain, but instead a normalization towards a lower body weight and FM (5). Even though AT, at the level of REE, has been found in the present study, it needs to be taken into account that participants were in negative EB, and this variable is likely to modulate the magnitude of AT, as previously shown by our research group (19, 39). Indeed, if measurements were repeated after a period of weight stabilization, AT would be expected to be minimal or inexistent (19, 39). Even though in the present study an association between AT and changes in appetite has been found, in line with our previous findings (40), suggesting an interplay between EE and appetite, by which those individuals experiencing a larger AT also experience a greater drive to eat in response to weight loss, it needs to be acknowledged that AT was only measured at the level of REE. Further studies are needed to better understand the interplay between AT, at the level of TEE, and increased appetite under conditions of EB, and outside of ketosis, in obese-reduced individuals, and their potential role in weight regain, to better understand if these adaptations, on both sides of the EB equation, are indeed a normalization or not.

This study has several strengths. First, the study design is unique, as weight loss, diet composition, and level of nutritional-induced ketosis were similar across groups, allowing the identification of the impact of SG or RYGB alone on the outcome variables. Secondly, all participants were in negative EB, thereby validating comparisons between groups, as the magnitude of AT has been shown to be modulated by the EB status of the participants (7, 39). Thirdly, sex distribution, baseline anthropometric variables, and physical activity levels were similar across groups and therefore unlikely to have affected the variables of interest (22). Lastly, the significance level was adjusted for multiple comparisons, using Bonferroni adjustment. Despite the strengths of the study, there are also some limitations. First, we relied on self-reported data at baseline to determine if participants were weight stable, and as such cannot guarantee that the participants were in EB at baseline. Second, EE was measured only at rest. Third, body composition was measured using air displacement plethysmography, a two-compartment model, which does not consider either water or organ loss, with implications in the estimation of REE and AT, as previously discussed. Lastly, it needs to be acknowledged that at week 11 our participants were ketotic which could impact not only appetite, but also FFM loss, and the magnitude of AT. Weight loss in the present study was induced by a VLED and analysis of βHB plasma concentrations at week 11 showed that participants were in nutritional-induced ketosis (0.86 ± 0.78 mM). Ketosis is accompanied by glycogen depletion and water loss (41, 42). As a result, in the present study, loss of FFM also includes water (between 33-52%), meaning that the prediction of REE at week 11, based on FM and FFM, might have been overestimated, resulting in an overestimation of the magnitude of AT. Additionally, FFM is not a homogenous compartment and includes bone, skeletal muscle (SM) and organs (43). Loss of organ mass has been reported to occur following weight loss (44), and since organs have a much higher metabolic rate than SM (44), its loss would lead to a greater reduction in REE, not accounted for in the prediction on REE, which again would lead to an overestimation of REE and the magnitude of AT following weight loss. This is supported by Müller and colleagues who showed that if the composition of FFM is accounted for, AT following weight loss in minimal or inexistent (45).

Conclusion

Weight loss modality does not seem to modulate the magnitude of AT, at the level of REE, during negative EB, once diet composition, weight, FM and FFM loss are matched. However, individuals with the largest AT at the level of REE, also experience the weakest suppression in the drive to eat in the postprandial state, which can further compromise weight loss outcomes. In the future, AT at the level of TEE, needs to be estimated to determine if its magnitude differs between diet and bariatric surgery, and to explore its potential association with weight regain in the long-term.

What is already known about this subject?

• The existence, magnitude, and clinical relevance of adaptive thermogenesis (AT) following weight loss induced by diet and bariatric surgery remain controversial.

• The magnitude of AT seems to be modulated by the energy balance (EB) status of the participants, being minimal or inexistent when measurements are done in conditions of EB.

• AT seems to be associated with resistance to weight loss during negative EB.

What are the new findings in your manuscript?

• AT, at the level of resting energy expenditure, occurs regardless of weight loss modality and does not differ between diet alone and bariatric surgery (both sleeve gastrectomy and Roux-en-Y gastric bypass), when energy restriction, macronutrient composition of the diet, weight, fat mass and fat-free mass loss are matched.

• Individuals who experience a greater magnitude of AT report a greater increase in feelings of hunger and desire to eat after a meal.

How might your results change the direction of research or the focus of clinical practice?

• Bariatric surgery does not seem to impact AT, at the level of REE, beyond weight loss.

• Since individuals who experience a larger AT also feel hungrier in the postprandial state, they may initiate the following meal earlier and consume more food, therefore compromising diet adherence. Future studies should try to identify people at risk of experiencing a large AT, so that measures can be implemented aiming at improving dietary adherence and weight loss outcomes.

Data sharing statement:

Individual participant data that underlie the results reported in this article can be shared with other researchers, after deidentification (text, tables, and figures). These researchers need to have a specific aim for the data that has been approved by an independent review committee identified for this purpose. The data can also be shared with the purpose of meta-analysis.