Adolescent and young adult females with AN/AtypAN demonstrated hyperactivation of brain regions involved in top-down and bottom-up neural networks when viewing images of high-calorie foods versus objects, relative to HC. While individuals with EDs and HC both showed robust response to images of palatable foods, consistent with our hypotheses, those with EDs showed increased BOLD activations in neural regions that have been associated with cognitive control and reward processing to food cues relative to HC. We found both main group and appetitive state effects, but no interaction effects (counter to our hypothesis), suggesting that these group effects are robust, whereas the pre-to-post meal attenuation was similar in both groups. Greater neural activation in cognitive control brain regions in individuals with AN (vs. HC) in the context of food images is consistent with previous findings demonstrating that engagement of cognitive control was required for behavioral adaptation to changing environmental demands (15, 16). Yet, it extends previous results, showing that BOLD activations in regions associated with cognitive control were increased even during a passive viewing condition, namely, when cognitive control was not explicitly required. This finding suggests that cognitive control may be maladaptively over-engaged in those with AN and AtypAN, through automatic activation triggered by ED-relevant stimuli.
Interestingly, the observed greater BOLD activation in reward-related brain regions in female adolescents and young adults with AN/AtypAN (vs. HC) in response to food (vs. non-food) images contrasted with previous findings from our group showing less BOLD activation in food motivation brain areas in women with AN (vs. HC) employing the same fMRI paradigm (14). Notably this earlier study included a smaller sample size and was limited to amenorrheic adult females with restricting-type AN, which might account for the differences. Indeed, in the current study, differences within the ED clinical groups emerged as well, elucidating possible neurobiological correlates of divergent symptom presentation. For AtypAN versus AN groups, there were overall differences in dACC, right DLPFC, right hippocampus, right caudate, and right putamen, suggesting greatest recruitment of both cognitive control and reward processing regions in the AtypAN group compared to both AN and HC. When the ED group was split by primary behavioral phenotype, we found bilateral premeal hyperactivation in the left anterior insula, dACC, and right and left DLPFC in those with the BP-type presentation compared to restricting-type and HC groups, suggesting the greatest cognitive control activation in the BP-type group, but in contrast to our hypothesis, we found no differences here in BOLD activation of reward regions.
Taken together, these findings of hyperactivation of cognitive control and reward regions across those with AN/AtypAN, and particularly in AtypAN and those with BP-type illness respectively, suggest that dysfunctional food motivation pathways play a role in symptom presentation in AN, AtypAN and their subtypes. Observed alterations in neural activation in reward and cognitive control brain regions provide important evidence for existing models of aberrant brain activation patterns underlying ED psychopathology (27, 28). Furthermore, our findings fit with the notion that individuals who present with AtypAN (relative to AN) or restrictive eating punctuated by binge-eating episodes and/or purging recruit high levels of cognitive control in efforts to resist eating, whereas for those with restricting-type EDs, restriction may be less effortful perhaps due to lower reward activation in response to food (29). High levels of top-down engagement coupled with increased bottom-up activation is not sustainable indefinitely and this pairing may lead to the relatively higher weight in AtypAN, maintenance of binge-eating and purging behaviors in those with BP-type presentations, and longitudinal diagnostic crossover (30–32).
Study strengths include the interdisciplinary assessment of food motivation via fMRI. Repetition of measures from pre- to post-standardized meal allowed for a nuanced examination of eating behavior in this population. Inclusion of a well-phenotyped, heterogeneous adolescent and young adult sample of females with AN and AtypAN allowed for investigation across and then within clinical presentation compared to HC. However, important limitations warrant acknowledgement. First, the food motivation paradigm we used is well-validated and robustly activates food motivation circuitry, but the nature of the task—involving passive viewing rather than an active assignment—makes conceptualization of DLPFC and dACC activation as reflecting cognitive control inferential rather than conclusive. Notably, activation of the DLPFC has been shown to be associated with inhibition of appetitive responses (e.g., 15, 33, 34–38). Indeed, we chose a well-validated passive viewing of food paradigm rather than an active task to focus on automatic processing that may differentiate those with AN/AtypAN from HC. By definition those with AN/AtypAN exhibit behaviors that drive or maintain weight loss/low-weight; the degree to which abnormal neural activation underlies aberrant eating even when actual food is not present speaks to the automaticity of neural response and may offer novel insights for treatment development. Future studies should include both active and passive food paradigms to comprehensively test questions related to over-engagement of cognitive control regions in this population. Second, while the variance in ED symptomatology was a strength of our study and we were able to split the full ED group in two ways (by degree of low-weight, and by restricting/binge-eating/purging), these groups were limited. As low body weight (defined as ≤ 90% of expected body weight) was a study inclusion criterion, we could not examine the full weight spectrum of individuals captured by the AtypAN diagnosis. Further, our cohort of individuals in the BP-type group was small (n = 16), rendering that set of findings more exploratory. Future studies should include individuals with AtypAN across the weight spectrum, and a larger sample of individuals with a BP-type presentation.
Our results lay the groundwork for future investigations based on large samples to further unveil the shared or symptom-specific characteristics of food reward and cognitive control pathways and which can combine neural data with other biologic modulators of appetite (e.g., endocrine signaling) and their synergistic interplay or disruption thereof in explaining the clinical phenotypes of these complex conditions. This line of research could guide the development of targeted interventions to interrupt symptom consolidation.