Association of Brain Reward Learning Response With Harm Avoidance, Weight Gain, and Hypothalamic Effective Connectivity in Adolescent Anorexia Nervosa

Abstract

Anorexia nervosa (AN) is associated with adolescent onset, severe low body weight, and high mortality as well as high harm avoidance. The brain reward system could have an important role in the perplexing drive for thinness and food avoidance in AN. To test whether brain reward learning response to taste in adolescent AN is altered and associated with treatment response, striatal-hypothalamic connectivity, and elevated harm avoidance. In this cross-sectional multimodal brain imaging study, adolescents and young adults with AN were matched with healthy controls at a university brain imaging facility and eating disorder treatment program. During a sucrose taste classical conditioning paradigm, violations of learned associations between conditioned visual and unconditioned taste stimuli evoked the dopamine-related prediction error (PE). Dynamic effective connectivity during sweet taste receipt was studied to investigate hierarchical brain activation across the brain network that regulates eating. The study was conducted from July 2012 to May 2017, and data were analyzed from June 2017 to December 2017. Prediction error brain reward response across the insula, caudate, and orbitofrontal cortex; dynamic effective connectivity between hypothalamus and ventral striatum; and treatment weight gain, harm avoidance scores, and salivary cortisol levels and their correlations with PE brain response. Of 56 female participants with AN included in the study, the mean (SD) age was 16.6 (2.5) years, and the mean (SD) body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) was 15.9 (0.9); of 52 matched female controls, the mean (SD) age was 16.0 (2.8) years, and the mean (SD) BMI was 20.9 (2.1). Prediction error response was elevated in participants with AN in the caudate head, nucleus accumbens, and insula (multivariate analysis of covariance: Wilks λ, 0.707; P = .02; partial η2 = 0.296), which correlated negatively with sucrose taste pleasantness. Bilateral AN orbitofrontal gyrus rectus PE response was positively correlated with harm avoidance (right ρ, 0.317; 95% CI, 0.091 to 0.539; P < .02; left ρ, 0.336; 95% CI, 0.112 to 0.550; P < .01) but negatively correlated with treatment BMI change (right ρ, -0.282; 95% CI, -0.534 to -0.014; P < .04; left ρ, -0.268; 95% CI, -0.509 to -0.018; P < .045). Participants with AN showed effective connectivity from ventral striatum to hypothalamus, and connectivity strength was positively correlated with insula and orbitofrontal PE response. Right frontal cortex PE response was associated with cortisol, which correlated with body dissatisfaction. These results further support elevated PE signal in AN and suggest a link between PE and elevated harm avoidance, brain connectivity, and weight gain in AN. Prediction error may have a central role in adolescent AN in driving anxiety and ventral striatal-hypothalamus circuit-controlled food avoidance.