We found that the neural response to high-calorie visual food cues in obese endometrial cancer (EC) survivors, at baseline, was similar to that previously reported for the general adult obese population in both fasted (pre-meal) and fed (post-meal) states. In addition, we found that obese EC patients had decreased activation to high-calorie food cues after a 6 month behavioral lifestyle intervention compared to baseline in regions associated with food reward and motivation.
Our results in obese EC patients at baseline are consistent with previous findings in the general obese population, which are nicely summarized in a recent review by Carnell et al. . More specifically, previous fMRI studies have shown that obese compared to normal weight individuals have greater activation in brain regions involved in food motivation and reward processing including the insula, lateral orbitofrontal cortex (OFC), amygdala, dorsal striatum and putamen in response to visual high-calorie/palatable food stimuli even after eating a meal [38, 39, 41, 42]. In our study, at baseline, we found that obese EC patients, who appear to be slightly more obese than the general adult female population in the U.S. [13–15], had significant increased activation even after consuming a meal in the insula (BA = 13), cingulate gyrus (BA = 31) and precentral gyrus (BA = 4) for the high- vs. low-calorie contrast and, increased activation in the thalamus, posterior cingulate (BA = 29) and precuneus (BA = 7) for the high-calorie vs. non-food contrast. Because this was the first study in obese EC survivors, we also explored the low-calorie vs. non-food contrast and found, at baseline, increased activation in the putamen, claustrum, caudate, anterior PFC (BA = 10) and SFG (BA = 8) in the fasted state but no significant increased activations in the fed state. The increased activation we observed in the fasted state with low-calorie food cues was likely a function of the patients’ heightened hunger (due to fasting greater than 8 hours, on average). We hypothesize that, in the fed state, our lack of increased activation with low-calorie food cues in obese EC patients mimics findings in the general obese population in that exaggerated food-cue reactivity is more pronounced with high-calorie food cues . However, the differential activations we observed could be associated with alternative interpretations. Activations are based upon predicted or anticipatory responses from combined sensory data and prior experiences and, anticipatory and consummatory responses may be different and affected by mood . Although the cingulate gyrus has been hypothesized to affect emotional self-control, particularly as it relates to dietary restraint , activation could also be a function of anticipatory preferences as many pleasant stimuli activate this region . Moreover, although increased activation in the dlPFC has been correlated with eating restraint, the dlPFC may spontaneously engage self-control mechanisms when individuals view food pictures to arrive at perceived socially acceptable (or ‘appropriate’) responses [56, 57].
To our knowledge, there is only one prior prospective neuroimaging study that has evaluated response to visual food cues in obese patients (and normal weight controls) before and after a behavioral lifestyle intervention . In this study, they found significant decreased activation post-treatment (Session 2: after the 12-week ‘EatRight Lifestyle Weight Management Program’, which has the overarching theme to encourage replacement of high-energy-dense foods by low-energy-dense foods through weekly educational sessions) compared to baseline (Session 1: within 3 weeks of starting the EatRight Program) in the obese patients’ response to high-calorie vs. non-food cues in brain regions involved in reward and attentional processing including the medial prefrontal cortex (BA = 32), precuneus (BA = 7), and posterior cingulate (BA = 23, BA = 31). In our study, we found similar decreased activations with high-calorie vs. non-food contrasts in obese EC patients (treatment group) in the fed state when comparing post-treatment to baseline in the precuneus (BA = 31) and posterior cingulate (BA = 29) as well as the thalamus, lateral globus pallidus and cingulate gyrus (BA = 31), which are regions shown to be involved in reward, motivation, emotional processing and feeding decisions [58, 59]. We also found decreased activation for the low-calorie vs. non-food contrast in the fed state in the insula (bilateral, BA = 13), which is involved in taste processing and, the precentral gyrus (BA = 4), which is involved in pre-motor cortex planning and execution. Although our obese EC patient population was slightly heavier at baseline (mean BMI: 35.8 ± 8.1 kg/m2) than that of the adult obese population of Murdaugh et al.  (mean BMI: 32.9 ± 3.8 kg/m2), we observed a similar decrease in percent weight loss (3.4% ± 2.7%) as that found in Murdaugh et al.  (3.5% ± 2.4%). Therefore, it appears that both behavioral lifestyle interventions, which aimed to increase the intake of low-calorie/nutrient-rich foods (e.g., fruits and vegetables) while concomitantly decreasing the intake of high-calorie/nutrient-weak foods (e.g., chips and sweets) produced similar results in terms of weight loss and decreased neural activation in food motivation and reward regions in response to high-calorie food cues after the intervention compared to pre-treatment.
A few studies have prospectively evaluated response to visual food cues before and after bariatric and gastric banding surgery. Although surgical and behavioral therapy studies are not directly comparable due, in part, to the substantially higher weight loss in surgical therapies (ranging from 11.8%  to 25.2% ± 8.4% ), we observed differential activation in some of the same brain regions in our study with only 3.4% ± 2.7% weight loss. For example, Ochner et al. , found decreased activation in the precuneus, posterior cingulate and thalamus in response to high-calorie food cues in the fed state among Class III obese patients (n = 10; mean BMI: 45.0 ± 5.0 kg/m2) 1-month after bariatric surgery compared to 1-month prior to their surgery. Furthermore, Bruce at al.  found that 12 weeks after gastric banding surgery, Class III obese patients (n = 10; mean BMI: 40.6 ± 1.96 kg/m2) showed decreased activation in food motivation and reward regions including the insula, inferior middle gyrus and middle frontal gyrus. Interesting, we and Bruce et al.  found increased activation post-treatment compared to baseline in the superior frontal gyrus (SFG), which could suggest increased self-regulation post-treatment but might also represent the continued struggle with cognitive control when viewing high-calorie food cues. Bruce et al.  also observed increased activation in the middle frontal gyrus which is a region previously found to have leptin-reversible increased neural activity in response to food cues following weight loss . Rosenbaum et al.  suggested that weight loss reflects a state of leptin-deficiency and a phenotype with greater emotional and sensory responsiveness to food cues, make maintaining weight loss more difficult. The results with leptin administration observed by Rosenbaum et al.  seem to conflict with the results we and others have observed in behavioral and surgical weight loss interventions. Nevertheless, we believe our results, taken together with the results of other studies that have prospectively examined response to visual food cues before and after behavioral lifestyle  and surgical interventions [60, 61], suggest these types of interventions may help to dampen the response to high-calorie food cues, although more aggressive and/or longer-term interventions may be required to fully alleviate the incentive salience of high-calorie food cues corresponding to the increased activation found post-treatment relative to pre-treatment in attentional processing areas.
When evaluating the post-treatment (PostTx) scan separately, we found increased activation in only a few regions including the declive, culmen and middle frontal gyrus in the fasted state when comparing high-calorie vs. non-food images. In the post-meal state, when comparing high- vs. low-calorie images, we found increased activation predominantly in the frontal regions including the middle frontal gyrus and inferior frontal gyrus. Furthermore, we found decreased activation post-meal in several brain regions associated with food reward and motivation (insula, anterior cingulate, cingulated gyrus, middle frontal gyrus, inferior frontal gyrus) when comparing low-calorie vs. non-food images. Taken together, these results suggests that obese EC patients may be viewing food cues differently post-intervention with increased cognitive attention and, perhaps, less rewarding value, to high-calorie food images, particularly after satiation. We cannot compare our PostTx scan results to previous prospective lifestyle  and surgical [60, 61] intervention studies because previous studies only present their post-treatment results relative to the pre-treatment condition, as the neural adaptations in response to the intervention may be better elucidated when comparing the post-treatment to the pre-treatment condition.
With regard to our exploratory studies evaluating correlations with weight loss, we found positive correlations between baseline activations in high-calorie vs. non-food contrasts and percent weight change similar to Murdaugh et al. . However, our findings were limited to two frontal regions (OFC; medial frontal gyrus) while Murdaugh et al.  reported significant correlations in visual areas including the superior parietal lobe and middle frontal gyrus (BA = 8). When comparing PostTx to baseline scans, we did not observe any significant correlations between regions showing significant activation with high-calorie compared to non-food contrasts and percent weight loss. Murdaugh et al.  also did not report any significant correlations between percent weight change and activation in regions of interest at 12 weeks post-treatment (Session 2) compared to pre-treatment (Session 1) but they did observe significant negative correlations between percent weight change at 9 months (6 months after the intervention was completed) and change in activation (Session 1 minus Session 2) when comparing high-calorie vs. non-food images in several regions including the insula and thalamus. Bruce et al.  reported that they found no significant correlations between percent weight change and activation in regions of interest when examining visual food cues in obese adults before and after gastric banding surgery. Given the small sample size in our study and in Bruce et al. , our studies may have been underpowered to detect the significant correlations observed in Murdaugh et al. , which had over two times the number of patients. On the other hand, given the number of exploratory tests we performed without correcting for multiple testing, it is possible that the significant correlations we found at baseline were due to chance. Clearly, additional, larger prospective lifestyle and surgical intervention studies are needed to better elucidate the brain regions that may be most amenable to neural adaptations that drive sustained weight loss.
Overall, our results suggest the SUCCEED behavioral lifestyle intervention, which focused on replacing high-calorie/nutrient-weak foods (e.g., chips and sweets) with low-calorie/nutrient-dense foods (e.g., fruits and vegetables), may help reduce the rewarding value of high-calorie food cues, particularly after eating meal, in obese endometrial cancer survivors. However, given the increased activation we observed in frontal regions post-treatment compared to baseline, which could suggest increased self-regulation post-treatment but might also represent the continued struggle with cognitive control when viewing high-calorie food cues, more aggressive and/or longer-term interventions may be required to fully alleviate the incentive salience of high-calorie food cues. If the necessary neural adaptations in brain regions implicated in food reward and motivation could be maintained and long-term weight loss sustained, we could ultimately increase the survival of obese EC patients, who are at a very high risk of death from the co-morbidities associated with their obesity [14, 17].
Major limitations of our study in obese EC patients include the lack of a control group and a very small sample size. Therefore, our results should be considered preliminary and interpreted with caution until replicated in a larger study. Nevertheless, the differential activations we observed at baseline and for post-treatment compared to baseline are consistent with other previously published studies conducted in the general obese population. Although we provided the obese EC patients with a 750 kcal standardized meal, they only consumed, on average, about 500-550 kcals. Patients had the ability to select the type of sandwich (turkey, roastbeef, vegetarian) but they did not have the opportunity to select the type of fruit and vegetable provided; and, the items most likely to remain uneaten were vegetables and condiments (mayonnaise, mustard). Nevertheless, patients reported that they were no longer hungry after the meal and their food preference ratings were high for both high- and low-calorie foods. Interestingly, previous studies in obese populations have provided much smaller meals (e.g., 250 kcal liquid meal ) and have found similar activations in response to high-calorie food cues, suggesting that the total calories consumed may not have a material effect.