Introduction

Bariatric surgery is the collective term for a range of surgical procedures used as therapeutic interventions for the treatment of morbid obesity. Vertical sleeve gastrectomy (VSG), adjustable gastric banding (AGB) and Roux-en-Y gastric bypass (RYGB) are the principal procedures used in contemporary practice. Bariatric surgery remains the most effective and sustainable weight loss treatment for obesity [1]. Weight loss achieved using lifestyle interventions and anti-obesity medications is lesser than that seen after bariatric surgery and is minimally effective with regard to longer term weight loss maintenance. RYGB is the most commonly performed procedure [2] and leads to a 20–30% decrease in body weight which is generally maintained over a period of 20 years [3]. VSG and AGB patients achieve 15–25% and 10–15% long-term body weight loss, respectively [4].

RYGB involves the division of the stomach into two parts—an upper stomach pouch and a lower remnant stomach [5]. The stomach pouch (15–30 ml) is anastomosed to the distal limb of a mid-jejunal transection 30 to 75 cm distal to the ligament of Treitz. Jejunojejunostomy is then completed by anastomosis of the proximal limb of the jejunotomy at a point 75–150 cm distal to the gastrojejunostomy in order to preserve continuity of the intestine. Therefore, food transits straight to the small intestine and confluence with the gastric and biliopancreatic secretions is delayed until the common channel which runs distal to the jejuonjejunostomy [6].

Anatomical reconfiguration of the GI tract after bariatric surgery is linked to multiple endocrine changes, including some which may impact eating behaviour. Assessment of food preferences after RYGB is of clinical interest as eating the ‘right’ types of food can contribute to long-term weight loss and weight loss maintenance [7]; understanding the mechanisms behind these changes could lead to the development of less invasive weight loss interventions.

The aim of this narrative review is therefore to explore the literature documenting changes in food preference after RYGB in both human and animal studies and to outline current thinking with regard to the key mechanisms driving the described shifts in preference.

Indirect Assessment of Food Preferences After Bariatric Surgery in Humans

The first study to document changes in food preferences after RYGB was based on a 24-h dietary recall method and showed a decrease in the consumption of carbohydrate- and fat-rich food early after surgery [8]. A subsequent study showed that at 12-month follow-up [9], only fat consumption remained lowered after RYGB, while the intake of other macronutrients increased [9]. Such changes in food preference after RYGB have been reported to be durable in the long-term up to post-operative year 8 [10] (Table 1).

Table 1 Overview of changes in food preferences after RYGB in human studies
Full size table

RYGB has been associated with more robust reductions in the consumption of foods with high-sugar content than vertical banded gastroplasty (VBG) or AGB [11, 12, 13]. The literature is however not completely consistent on this point. As another study documented, there was sustained lowering of intake of carbohydrates, lipids and protein after RYGB with no evidence of a more marked change in preference for any one of the macronutrient groups [14]. In a further departure, despite decreased consumption during the first 9–12 months after surgery, Trostler et al. (1995) have suggested that patients go back to their baseline dietary habits by the end of their first year, including the consumption of high-fat food [15].

The Swedish Obese Study (SOS) questionnaire was developed by the SOS group to assess the consumption frequency of different food items as a measure of food preferences. Olbers et al. (2006) compared RYGB and VSG and showed that RYGB was superior to VSG in significantly decreasing the intake of cheese, sausages, desserts, cookies and candies, while increasing fruits and vegetables [16]. The reported decrease in fat was mirrored elsewhere in the results of a study comparing preferences at 6 years after RYGB and VBG [17, 18].

Food Frequency Questionnaires (FFQ) have also been deployed to assess dietary change after surgery and resulted in noting of a reduction in the consumption of dessert items and an increase in intake of vegetable and protein-rich foods such as fish, poultry and eggs [19]. Using the Power of Food Scale (PFS) at 16 months post-surgery, coherent changes were [21] identified upon assessment of the impact of thoughts, feeling and motivation around food [20].

Adaptation and tolerance to 236 food items from 3 months up to 2.5 years post-operatively showed a higher preference for lower fat food items after RYGB [21]. Dietary diversity was also reduced as patients were only consuming 41% of the food items in the list. The most recent study using the FFQ approach showed increase in intake of yogurt and fish alongside decreased intake of pizza and hamburger at 6 months after RYGB [22•].

Although Functional Magnetic Resonance Imaging (fMRI) does not provide a direct measure of behaviour, it does allow objective assessment of the differences in the activation of brain regions involved with the motivation to eat permitting inferences about correlations with verbal report and direct measures to be made [23]. Optimally, fMRI protocols should be paired with direct measures of behaviour [24].

Comparing neural activation between RYGB, VSG and weight stable controls exposed to images of high-calorie and low-calorie foods pre- and 6 months post-surgery showed that both RYGB and VSG reduce brain activation in response to pictures of high-calorie foods. Specifically, the activatory response in reward centres decreases significantly after RYGB [25, 27]. A reduction in neural responsivity concurring with reductions in the rating of “wanting and liking” of highly palatable foods has also been shown in a cohort study of patients at 1 month after RYGB surgery [26•, 28•].

RYGB has also been compared to AGB using fMRI scans. Results showed more marked activation in the brain reward centres in the RYGB group in response to images of high-calorie foods, corroborating the picture of choices that emerged from food diaries in the study [27].

Direct Assessment of Food Preferences After Bariatric Surgery in Humans and Animals

Human studies are mostly reliant on indirect reported measures while animal studies allow more direct measures to evaluate changes in food preferences. The only published study reporting on the use of a direct measurement of food preference after surgery assessed changes in food selection after RYGB and sleeve gastrectomy (SG). Participants were offered an ad libitum buffet meal, 3 months before and 6 months after surgery. The buffet consisted of 20 common Danish food items representing a combination of both sweet and savoury and low- and high-fat items. The results showed a decrease in overall energy intake by 54% but no significant difference in food preferences at 6 months. Alongside this, a picture display test was also conducted. This showed increased preference for low-fat savoury food items post-surgery. The pre-surgery test was carried out before nutritional counselling and thus, although the researchers attempted to divert the attention of subjects away from being concerned about the opinions of the researchers on their food choices, it remains challenging to mimic usual behaviour around food intake in an experimental setting [28•].

The two-bottle preference test has been used to obtain direct measurements of changes in preference after bariatric surgery in rats [19, 29]. The test consists of ad libitum exposure of rats to two sources of drinking water, one bottle with the test solution (fatty/sweet solution) and another with water. Preference is measured by calculating the relative amounts of solution consumed over a 48-h period. Both studies using the two-bottle test after RYGB in rats recorded a reduction in consumption of the calorie-dense solutions relative to that of water in RYGB rats compared to the sham-operated rats [20, 31]. Elsewhere, when a vegetable drink was used as an alternative to water versus high-fat and high-sucrose test solution, an increase in preference for the vegetable drink was shown to occur after RYGB [30] (Table 2).

Table 2 Overview of changes in food preferences after RYGB in animal studies
Full size table

In line with results from two-bottle tests when fatty-sweet options are presented in liquid form, rats after RYGB prefer low-fat chow in comparison to high-fat chow. Interestingly, Saeidi et al. (2012) reported an initial increase in high-fat diet consumption by RYGB rats in the first meal after surgery relative to sham rats. However, by the end of a 24-h period, high-fat diet intake in RYGB rats decreased to 51% compared to that observed after sham surgery in rats [31], suggesting that the rats still liked the high-fat diet but adapting a conditioned avoidance to that meal. Additionally, Hao et al. (2016) monitored preference to high-fat diet chow and normal chow in sham- and RYGB-operated animals. Pre-operatively, the RYGB rats had a 5% preference to normal chow, which rose to 8% post-operatively. Meanwhile, there was no difference in preference in the sham group [32].

A direct study of food intake after RYGB in rats consuming a cafeteria diet consisting of four different food items showed that RYGB rats reduced their consumption of fat from 57% of total calories pre-surgery to 37% of total calories after surgery [33•].

Considering all the evidence from human and animal experiments, there may be a change in food preference after bariatric surgery, particularly RYGB although some discrepancies remain in the human studies which may be overcome in the future by studies making more direct measures of preferences, choices and total intake. Weight loss achieved after surgery does not appear to be the only causative factor for changes in the observed food preferences.

Potential Mechanism 1: Taste Sensitivity

Taste sensation depends on activation of three sensory responses. Firstly, as food reacts with the taste receptors, the type of taste, i.e. sweet, salty, sour, bitter or umami, is identified; secondly, signals are sent to the nucleus tractus solitaries (NTS) and other appetite centres in the brain to determine how much the food is wanted ‘appetitive behaviour’ and how much it is liked ‘consummatory behaviour’. Simultaneously, physiological cephalic responses are produced to prepare for food ingestion. These functions are generally classified into three taste sub-domains: stimulus identification, ingestive motivation and digestive preparation.

Some studies show a change in taste after RYGB. Testing the recognition thresholds for the four basic tastes (salt, sweet, sour and bitter), using a modification of the Henkin forced-choice three-stimulus technique, in women trend toward a reduction in salt and sweet detection thresholds after RYGB [34]. A staircase method of stimulus presentation in humans found supporting results [35]. Bueter et al. (2011) also found increased sensitivity for sweet solutions in humans, this time using a unique approach consisting of a method of constant stimuli [29].

Some studies do not show a change in taste after bariatric surgery. The changes observed after RYGB regarding salty and sweet taste as well as smell were not reported as being reproduced in patients after VSG [36]. The study was followed up for 5 years and the changes were recorded through self-reporting [24]. Two studies also report no change in detection thresholds for any of the taste qualities after surgery [37, 38].

The current literature lacks taste sensitivity studies in animals, possibly due to methodological difficulties in the assessment of this domain. Tichansky et al. (2011) attempted to assess taste sensitivity in rats using a brief access test with gustometer consisting of different sucrose concentrations. However, this experiment may not be appropriately designed for the assessment of ‘taste sensitivity’ as the outcome may reflect more of the ‘appetitive behaviour’ to different sweet concentrations.

Spector et al. used the same paradigm in rats as used by Miras et al. (2012), but were not able to reproduce the data [39, 40]. The discrepancy between the human and rat studies using the same methodology is not clear, although the rats were tested once they became weight stable, while the humans were tested when they were in the steepest part of their weight loss curve after surgery.

The enteroendocrine hormone glucagon-like peptide-1 (GLP-1) is released by the taste buds when exposed to taste stimuli. GLP-1 receptor knockout mice exhibit a decrease in sensitivity to lower concentrations of sugar [41]. Takai et al. (2015) demonstrated that sugary substances cause a release of GLP-1 through activation of taste receptor (T1R3) signalling, which in turn could cause activation of the gustatory nerve fibres that are sensitive to sugar [42]. The GLP-1 agonist, exendin 4, reduces food intake in rats while the antagonist, exendin 9–39, caused an increase in food intake [43], despite a lack of altered response to sugar [39]. It can thus be speculated that modulation of taste by gut hormones plays a role in observed reductions in food intake after RYGB. Administration of the exogenous GLP-1 analogue liraglutide does not however alter food preferences in rats when offered the cafeteria diet [33•, 44, 46].

Potential Mechanism 2: Ingestive Motivation

The ingestive motivation domain is easier to assess in both human and animals. Progressive ratio task (PRT) to assess the appetitive sub-domain in RYGB patients found that surgery selectively reduces the reward value of a sweet and fat tastant [45]. Sweet taste palatability tests have been used to explore consummatory taste behaviour in RYGB patients. Visual analogue scales (VAS) or global label magnitude scales (gLMS) used to assess palatability have shown mixed findings with some studies showing a shift in sweetness palatability from pleasant to unpleasant [37], but this has not been consistent [29].

Functional MRI studies are used to see whether hedonic reward motivations are changed in the brain centres perceiving taste. Activation of sensory brain areas as well as areas responsible for reward when presented with salt and sweet stimuli prior and after surgery has been assessed showing increased activation at reward centres for salt taste and decreased activation for sweet taste stimuli [46].

The appetitive sub-domain or behaviour can be investigated in more depth in animal studies, because the social pressures present in human studies are nullified. RYGB rats had a significant reduction in mean licks over 10 s for a sweet stimulus albeit overall mean licks for five different concentrations of sucrose solution did not differ significantly [47]. In 2012, Mathes et al. also used a brief access test with a gustometer. Surprisingly, they showed almost a doubling in the appetitive behaviour of rats after RYGB for sweet stimuli [48, 49].

Facial expressions are used in ‘taste reactivity tests’ to assess liking and disliking of the taste solutions. To our knowledge, this test has not been used in RYGB human studies yet. One study documented that in rats, after RYGB, fewer orofacial responses occur in response to high versus low concentrations of sucrose [50]. On the contrary, another study testing orofacial responses of RYGB rats to different sucrose concentrations found a flat response curve after, indicating no change in consummatory behaviour [51].

PRT lick operant assessment-based study of the effect of RYGB on the reinforcement value of palatable fluid stimuli has been studied in rats [39, 52]. After surgery, RYGB-operated rats do not show lower breakpoints than sham-operated rats for reinforcers consisting of Ensure, sucrose or 5% Intralipid despite having a blunted preference for these caloric fluids versus water in two-bottle preference tests [39]. Collectively, these studies have started to shift our thinking that bariatric surgery may not cause changes in food preferences due to altered ‘wanting’ properties but rather due to a learned adjustment to altered post-ingestive feedback.

Potential Mechanism 3: Conditioned Avoidance

Taste aversion is a form of classical conditioning to a stimulus that elicits aversive reactions such as nausea [53]. Avoidance, on the other hand, is consciously avoiding foods which remain palatable but result in adverse reactions. When RYGB- and sham-operated rats were presented with sugar solution and fat solution for an hour, their initial consumption was comparable [54]. However, over time, RYGB-treated rats reduced their intake, but not their initial lick rate. This suggests avoidance and not aversion as the operative mechanism in reduced fat intake after surgery [55]. Similar results have been seen in human experiments where patients reduced consumption of foods containing sugar and fats to possibly avoid unpleasant post-ingestive unpleasant reactions when larger amounts of the same foods are consumed [10, 18, 25].

Conclusion

A clear shift in food preference is observed in animals and humans after RYGB. Taste sensitivity may change after RYGB but conditioned avoidance and associated learned changes in ingestive motivation may be more likely candidate mediators of changes in food preferences. Residential settings to record food intake and preference by direct measurement in humans have the potential to significantly advance knowledge in the field by avoiding the confounding influence of reporting bias and thus permitting definitive determination of the nature and origin of changes in food preference after RYGB.