Food Reward after Bariatric Surgery and Weight Loss Outcomes: An Exploratory Study

Changes in food preferences after bariatric surgery may alter its effectiveness as a treatment for obesity. We aimed to compare food reward for a comprehensive variety of food categories between patients who received a sleeve gastrectomy (SG) or a Roux-en-Y gastric bypass (RYGB) and to explore whether food reward differs according to weight loss. In this cross-sectional exploratory study, food reward was assessed using the Leeds Food Preference Questionnaire (LFPQ) in patients at 6, 12, or 24 months after SG or RYGB. We assessed the liking and wanting of 11 food categories. Comparisons were done regarding the type of surgery and total weight loss (TWL; based on tertile distribution). Fifty-six patients (30 SG and 26 RYGB) were included (women: 70%; age: 44.0 (11.1) y). Regarding the type of surgery, scores were not significantly different between SG and RYGB, except for ‘non-dairy products—without color’ explicit liking (p = 0.04). Regarding TWL outcomes, explicit liking, explicit wanting, and implicit wanting, scores were significantly higher for good responders than low responders for ‘No meat—High fat’ (post-hoc corrected p-value: 0.04, 0.03, and 0.04, respectively). Together, our results failed to identify major differences in liking and wanting between the types of surgery and tended to indicate that higher weight loss might be related to a higher reward for high protein-content food. Rather focus only on palatable foods, future studies should also consider a broader range of food items, including protein reward.


Introduction
Bariatric surgery is considered the most effective treatment in case of severe complicated or morbid obesity, to achieve sustained weight loss, reduce comorbidities and mortality [1][2][3]. The most commonly performed surgical techniques are sleeve gastrectomy (SG) and Roux-en-Y gastric bypass (RYGB) [4]. Both types of procedures give similar weight loss patterns at 5-year follow-up [5].
Bariatric surgery has a direct effect on food intake, including reducing ingested volume [6] and total energy intake [6,7], as well as reducing hunger and increasing satiety [8]. In addition, our recently published systematic review and meta-analyses found that food Among the various tools that exist to measure food reward, we used the Leeds Food Preference Questionnaire (LFPQ) [17], in a version adapted by Van der Meij et al. [30] as it gives a behavioral measure of food reward quantifying both explicit liking and implicit wanting. Explicit liking was measured using visual analog scales (VAS) on the question "How pleasant would it be to taste some of this food now?" Implicit wanting was computed using a forced-choice task during which a spontaneous choice must have been made between all combinations of food items presented consecutively in pairs. An implicit wanting score for a given food category is calculated from the frequency of choice and non-choice of this food category and the reaction time of the participants [18]. This score ranges between −100 and 100 and is interpreted in relation to the scores of the other food categories evaluated in the task. A positive score indicates that the considered food category was chosen more often and faster than the others. A negative score indicates the opposite.
Although the LFPQ as adapted by Van der Meij [30] was initially intended for use among the elderly, this version seemed relevant to explore food preferences among patients after bariatric surgery. In particular, it was composed of six stages which allowed us to assess liking and wanting for 11 food categories varying in nutritional composition (high/low in carbohydrates, high/low in fat, high/low in protein, dairy/nondairy, with/without fiber, with/without meat) and/or in taste (savory/sweet), appearance (with/without color, with/without variation, with/without sauce), and texture (solid/fluid) ( Table 1). These food categories were combined within a task (i.e., high fat-sweet, high fat-savory, low fat-sweet, low fat-savory).
We adapted this version for its use in a French population by translating the instructions from English to French and adapting the food items to French eating habits [18]. The new set of food pictures was validated by asking 20 health professionals specialized in nutrition or dietetics and 20 non-specialized counterparts whether the food pictures correspond to what they can eat in France using a nine-point Likert-type scale where 1 represented 'Not at all' and 9 'Extremely'. Food items with a score ≤5 were considered not representative of French eating habits and were replaced. The list of food items used is available in Table 1.

Other Covariates
Information on gender, age, and body mass index (BMI) kg/m 2 before surgery and at assessment time point were collected based on the medical record of the patients. Patients self-reported 'having a budget constraint to purchase food' and 'smoking status'. Ratings of hunger, fullness, and desire to eat were assessed immediately prior to processing the LFPQ using visual analog scales (100 mm) from 'Not at all' to 'Extremely' of the following questions: "How hungry do you feel now?", "How full do you feel now?", and "How strong is your desire to eat now?". Time since the last meal was also recorded. The weight loss results were expressed as a percentage of the total weight loss (%TWL = [(Initial Weight) − (Postoperative Weight)]/[(Initial Weight)] × 100). The %TWL metric has been considered a better metric than the percentage excess weight loss because its calculation does not need to define an ideal body weight and did not differ between lower and higher baseline BMI [31] and is a better predictor of metabolic outcome [32,33].

Statistical Analysis
General characteristics, appetite sensations, and time since the last meal of patients were compared using independent t-test, Pearson's χ 2 test or Fisher's exact test for normal  In the present study, patients having a wanting score <−100 and >100 in a food category were considered as outliers, and results for the food category were excluded from the analyses. Explicit liking, explicit wanting and implicit wanting scores for all the studied food categories were compared between the groups of patients (SG and RYGB) using Student's t-test. We also compared liking and wanting scores between patients at 6, 12, and 24 month follow-up using one-way ANOVA. As mean and range for %TWL was expected to differ at 6, 12, and 24 months after bariatric surgery, we calculated the tertile separately for each post-surgery follow-up duration group (6, 12, and 24 months). Participants in the first tertile (<33%) were considered as low responders, in the middle tertile (33-66%) as middle responders and, in the upper tertile (>66%) as good responders. All participants were analyzed together using one-way ANOVA. Post-hoc analyses were Bonferroni corrected using the SPSS software post-hoc option and post-hoc corrected p-value provided for all ANOVA p-value < 0.1.
Data are presented as mean (SD). All the statistical analyses were conducted using SPPS software version 26.0.0.1 for Mac. All tests were two-sided and p < 0.05 was considered significant.

Characteristics of the Subjects
In the study, 61 patients were initially recruited. Five participants were excluded from the study (two for antidepressant medication, one for actual major depressive disorder, and two for a follow up duration >24 months). A total of 56 participants (30 SG and 26 RYGB) were included in the present analysis. Table 2 presents the characteristics of the patients according to their surgery group. Overall, 75% of the patients were women. Patients with RYGB were significantly older (50.5 y, SD = 9.0) compared to those with SG (38.3 y, SD = 9.6). The mean BMI before surgery of the SG group (45.5 kg/m 2 , SD = 5.9) was significantly higher compared to the RYGB group (41.4 kg/m 2 , SD = 4.4). The number of participants for each follow-up duration was similar between SG (n = 10 for 6, 12, and 24 months) and RYGB (n = 10 for 6 months, n = 7 for 12 months, and n = 9 for 24 months; p = 0.862). The other parameters, including %TWL, appetite sensations scores, were not significantly different between the two groups of surgery.

Relationship between Liking and Wanting for Foods and Type of Bariatric Surgery
Explicit liking, explicit and implicit wanting scores regarding surgery type are illustrated in Figure 1 and detailed in Supplementary Material Table S1. Considering all the studied food categories, scores were not significantly different between SG and RYGB, except for 'non-dairy products-without color' explicit liking (26.5, SD = 24.7 for SG and 14.8, SD = 15.7 for RYGB; p = 0.038). We also observed a trend for a significant difference for 'nondairy products-without color' implicit wanting (−17.0, SD = 23.7 for SG and −27.6, SD = 21.7 for RYGB; p = 0.089), 'non-dairy products-no color' explicit liking (28.3, SD = 23.5 for SG and 18.5, SD = 16.7 for RYGB; p = 0.087), and for foods 'low in carbohydrates-fluid' implicit wanting (−19.5, SD = 32.0 for SG and −2.3, SD = 32.8 for RYGB; p = 0.053). A negative implicit wanting score means that the food category items were more often rejected relatively to the items in the other food categories. Moreover, explicit liking, explicit wanting and implicit wanting scores for each follow-up duration (6, 12, and 24 months) were not significantly different for any of the food categories (data not shown).
Nutrients 2021, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/nutrients no color' explicit liking (28.3, SD = 23.5 for SG and 18.5, SD = 16.7 for RYGB; p = 0.087), and for foods 'low in carbohydrates-fluid' implicit wanting (−19.5, SD = 32.0 for SG and −2.3, SD = 32.8 for RYGB; p = 0.053). A negative implicit wanting score means that the food category items were more often rejected relatively to the items in the other food categories. Moreover, explicit liking, explicit wanting and implicit wanting scores for each follow-up duration (6, 12, and 24 months) were not significantly different for any of the food categories (data not shown).

Relationship between Liking and Wanting for Foods Regarding Weight Loss Outcomes
Regarding comparison between %TWL groups, Bonferroni post-hoc test revealed that differences were significant only between Low and Good responders. Detailed results are available in Tables 3-5 (characteristics of the subjects for each tertile are presented in  Supplementary Material Table S2).  Scores were significantly higher for good responders than Low responders for 'No meat-High fat' and this for the three types of measures, explicit liking, explicit wanting and implicit wanting (post-hoc corrected p-value: 0.037, 0.033 and 0.043, respectively). Regarding 'High protein-Variation' items, explicit liking was higher in good responders than in Low responders (post-hoc corrected p-value = 0.030).
Regarding implicit wanting scores, more differences were found between good and low responders. In addition to the higher scores mentioned above for the 'No meat-High fat' category, good responders also had significant higher scores than Low responders for 'Low carb-Solid' (post-hoc corrected p-value = 0.011) and tended to be significantly higher for 'High protein-No variation' items (post-hoc corrected p-value = 0.061). A trend to significant difference was also observed for 'Low protein-Variation' items with good responders having lower scores than low responders. Taking together, these results suggest that participants with higher %TWL tended to present higher implicit wanting for food items with higher protein content.

Discussion
The present exploratory study aimed to investigate thoroughly liking and wanting for food items (i.e., food reward components) in participants after bariatric surgery using a computerized behavioral task. The originality of this study relies on the use of a wide range of food items varying in nutrition composition, appearance, taste, and texture and on its comparative design considering the types of surgery (SG and RYGB) and the weight loss outcome.
The present work is among the few studies using a distinct comparative design towards the two most common types of bariatric surgery. Regarding socio-demographic data, participants with SG or RYGB did not differ significantly except for age (SG = 38.3 years and RYGB = 50.5 years, p < 0.001). Age and gender influence on taste has been documented and might have influenced our results [34]. Although statistically significant, this difference appears to be of small clinical relevance given that all patients belong to the same middle-age category and may not influence food preferences. Our findings suggest that there is no major difference between the two types of surgery regarding liking and wanting for all foods categories assessed. This is in accordance with studies, showing a similar decrease in preference for fat [35] and liking for high-calorie dense foods [36], as well as an increase in diet quality [37] and dietary intake [38] after both SG and RYGB. However, another study showed a decrease in the hedonic rating for high fat after SG and RYGB, but the liking of high sugar remained unchanged after RYGB [35]. On the other hand, a recent study published by Lewis et al. suggests that RYGB tends to be associated with lower enjoyment and craving for highly palatable foods than SG [39]. RYGB induces more dumping syndrome than SG [40], and thus the digestive discomfort caused by the consumption of foods rich in carbohydrates may have resulted in stronger conditioned taste avoidance in RYGB than in SG. Although this is a frequent issue in clinical practice, our results do not indicate any relevant changes in relation to carbohydrates liking or wanting, in any form (i.e., solid or liquid).
Considering weight loss outcomes (assessed by %TWL), comparisons of explicit liking and explicit and implicit wanting scores between low and good weight loss responders suggest differences in reward depending on food protein content. In our study, participants with a better weight loss trajectory tend to prefer foods with higher protein content.
Interestingly, these results could be explained within the framework of the protein leverage hypothesis. This hypothesis was primarily developed to explain overeating in the context of obesity [41][42][43]. This theory assumes that protein intake is prioritized over carbohydrates and fat intake. Accordingly, participants eat until their protein needs are met, regardless of the energy content of the food. Thus, participants with higher weight loss may have a stronger desire (and consumption) for protein and therefore, consume less palatable products (with high energy density but low protein content). Further confirmation of such findings in a prospective study would be of importance, especially to determine whether a higher pre-surgery protein wanting causes a greater weight loss or whether such modifications appear after bariatric surgery.
In contrast to our expectations, patients with lower weight loss did not show higher liking or wanting for products rich in fat or carbohydrates, except for 'Low carb-Solid' implicit wanting. A systematic review showed that food reward decreased during dietary, exercise, pharmacological, cognitive, and behavioral/multidisciplinary weight-loss interventions [44]. This suggests that more than the bariatric surgery itself, the weight loss and therefore the energy imbalance might have an impact on the motivation to eat and food choices. Most patients experience a 'honeymoon' in the first post-operative year (12-18 months), where they have a reduced appetite, limiting portion sizes, and disinterest in foods [45], which is concomitant with the great and rapid weight loss phase. Our results might be explained considering the time points of assessment reflecting the time since surgery. Indeed, in a previous study, we observed that participants had a higher appreciation of highly palatable foods after 2 years of follow-up than those having been operated on more recently [13]. Other studies also suggest a return of dietary preferences to prior patterns with increasing time since surgery [37,46,47]. Using the Leeds Food Preference Questionnaire, Oustric et al. also observed such a backward step after weight loss programs [48]. Prospective and long-term assessments (after 2 years of follow-up) of food rewards are necessary to better understand if those modifications are sustained and/or if they are related to the weight regain that occurs in 20% of patients within 1 to 3 years after surgery [49].
Some limitations must be considered, especially the cross-sectional design of the study and the small sample size. The transversal design does not allow us to test the causal effect of bariatric surgery on the differences of food reward found between lower weight loss responders and good or normal weight loss responders. The relatively small sample size and the sampling method (routine daycare) is also an important limitation, especially considering the important number of statistical analyses conducted. Larger sample sizes are needed to achieve the required power to apply a proper correction for multiple testing. Another limitation is the lack of controlled laboratory conditions when measuring food preferences as the visits occurred in the hospitalization unit and the time of arrival of the patients were based on their pre-registered clinical consultancy, which prevented rigorously standardizing fasting status, appetite levels, and time since the last meal in patients. However, when comparing the type of surgery or the weight loss trajectory status, the time from last meal or appetite sensations did not differ between groups. Our results showed a significant difference in age between patients with SG and RYGB (38 vs. 51 years old respectively). Finally, food preferences could have been influenced by other parameters, such as the socio-cultural background and ethnicity, which could be considered in future research given their influence on food choices [50,51]. Despite these limitations, our study, by addressing a wider range of food items, highlights the need to consider adding protein items in future research on food rewards, especially in the bariatric surgery field.

Conclusions
To conclude, liking and wanting for a comprehensive range of food items were not different in participants with either SG or RYGB. Wanting for protein items might be related to the weight loss response after bariatric surgery, while no such association was found for usual palatable foods. Adding protein items to usual palatable food items could be clinically relevant to identify postoperative alterations in food reward and guide caregivers to give personalized advice to patients in the context of precision bariatric medicine.