Avoidant/restrictive food intake disorder (ARFID) leads to faltering growth and psychosocial impairment. Three phenotypes can co-occur: fear of aversive consequences of eating (ARFID-fear phenotype), sensory sensitivity, and lack of interest in eating/food. We hypothesized that youth with ARFID, especially ARFID-fear phenotype, would show hyperactivation of fear-related regions in response to ARFID-specific fear images, compared to healthy controls (HC), and activation of these regions would positively correlate with ARFID fear severity.

Youth (N=103: 76 ARFID, including 20 ARFID-fear phenotype; 27 HC) underwent functional MRI scanning while viewing ARFID-specific fear (e.g. vomiting, choking) versus neutral images. We compared blood-oxygen-level-dependent (BOLD) response in fear-related region of interests (ROI; e.g. amygdala, hippocampus, insula) between ARFID and ARFID-fear phenotype versus HC. We evaluated the association between brain response and ARFID fear severity in ARFID-fear phenotype.

Across individuals, there was a robust bilateral amygdala response to ARFID-specific fear versus neutral images. Compared to HC, ARFID-fear phenotype showed a greater insula response to ARFID-specific fear versus neutral images (p=0.049). There were no other group differences and no significant relationships between BOLD response and ARFID fear severity in ARFID-fear phenotype.

ARFID-specific fear images elicit amygdala responses across individuals, with greater activation in the insula only in ARFID-fear phenotype versus HC. These findings validate the ARFID-specific fear paradigm and highlight the intriguing possibility that, in the ARFID-fear phenotype, universally feared experiences such as choking and vomiting serve as the unconditioned stimulus in developing ARFID and may partially be mediated by the insular cortex.

DSM-5 differentiates avoidant/restrictive food intake disorder (ARFID) from other eating disorders (EDs) by a lack of overvaluation of body weight/shape driving restrictive eating. However, clinical observations and research demonstrate ARFID and shape/weight motivations sometimes co-occur. To inform classification, we: (1) derived profiles underlying restriction motivation and examined their validity and (2) described diagnostic characterizations of individuals in each profile to explore whether findings support current diagnostic schemes. We expected, consistent with DSM-5, that profiles would comprise individuals endorsing solely ARFID or restraint (i.e. trying to eat less to control shape/weight) motivations.

We applied latent profile analysis to 202 treatment-seeking individuals (ages 10–79 years [M = 26, s.d. = 14], 76% female) with ARFID or a non-ARFID ED, using the Nine-Item ARFID Screen (Picky, Appetite, and Fear subscales) and the Eating Disorder Examination-Questionnaire Restraint subscale as indicators.

A 5-profile solution emerged: Restraint/ARFID-Mixed (n = 24; 8% [n = 2] with ARFID diagnosis); ARFID-2 (with Picky/Appetite; n = 56; 82% ARFID); ARFID-3 (with Picky/Appetite/Fear; n = 40; 68% ARFID); Restraint (n = 45; 11% ARFID); and Non-Endorsers (n = 37; 2% ARFID). Two profiles comprised individuals endorsing solely ARFID motivations (ARFID-2, ARFID-3) and one comprising solely restraint motivations (Restraint), consistent with DSM-5. However, Restraint/ARFID-Mixed (92% non-ARFID ED diagnoses, comprising 18% of those with non-ARFID ED diagnoses in the full sample) endorsed ARFID and restraint motivations.

The heterogeneous profiles identified suggest ARFID and restraint motivations for dietary restriction may overlap somewhat and that individuals with non-ARFID EDs can also endorse high ARFID symptoms. Future research should clarify diagnostic boundaries between ARFID and non-ARFID EDs.