Is hunger important to model in fMRI visual food-cue reactivity paradigms in adults with obesity and how should this be done?

We considered 1) influence of self-reported hunger in behavioral and fMRI food-cue reactivity (fMRIFCR) 2) optimal methods to model this. Adults (N = 32; 19-60 years; F = 21; BMI 30-39.9 kg/m(2)) participated in an fMRI-FCR task that required rating 240 images of food and matched objects for 'appeal'. Hunger, satiety, thirst, fullness and emptiness were measured pre- and post-scan (visual analogue scales). Hunger, satiety, fullness and emptiness were combined to form a latent factor (appetite). Post-vs. pre-scores were compared using paired t-tests. In mixed-effects models, appeal/fMRI-FCR responses were regressed on image (i.e. food/ objects), with random intercepts and slopes of image for functional runs nested within subjects. Each of hunger, satiety, thirst, fullness, emptiness and appetite were added as covariates in 4 forms (separate models): 1) change; 2) post- and pre-mean; 3) pre-; 4) change and pre-. Satiety decreased (Delta = 13.39, p = 0.001) and thirst increased (Delta = 11.78, p = 0.006) during the scan. Changes in other constructs were not significant (p's > 0.05). Including covariates did not influence food vs. object contrast of appeal ratings/fMRI-FCR. Significant image X covariate interactions were observed in some fMRI models. However, including these constructs did not improve the overall model fit. While some subjective, self-reported hunger, satiety and related constructs may be moderating fMRIFCR, these constructs do not appear to be salient influences on appeal/fMRI-FCR in people with obesity undergoing fMRI. (C) 2017 Elsevier Ltd. All rights reserved.