Changes in brain and behavior during food-based decision-making following treatment of anorexia nervosa

Karin Foerde, B. Timothy Walsh, Maya Dalack, Nathaniel D. Daw, Daphna Shohamy, Joanna SteinglassJournal of Eating Disorders 2021 · DOI

TL;DR

This study investigated how brain activity related to food choices changes in individuals with anorexia nervosa (AN) after weight restoration treatment. They found that while food choices didn't significantly change, a decrease in caudate nucleus activity after treatment was linked to an increased preference for high-fat foods, suggesting a complex interplay between brain mechanisms and persistent restrictive eating behaviors.

What Problem Does This Paper Address?

Anorexia nervosa (AN) is a severe eating disorder characterized by persistent food restriction, leading to significant health risks and high relapse rates even after successful weight restoration. While behavioral interventions can achieve weight normalization, they often fail to alter the underlying restrictive eating patterns. This disconnect highlights a critical gap in understanding the neural mechanisms that drive these maladaptive eating behaviors and whether they can be modified by treatment. Previous research has identified differences in brain activity during food-based decision-making between individuals with AN and healthy controls, particularly in the caudate nucleus. This study aims to explore whether these neural differences change with weight restoration treatment and if these changes correlate with behavioral outcomes.

Methods

The researchers employed a longitudinal design using functional Magnetic Resonance Imaging (fMRI) to examine brain activity during a Food Choice Task. Participants included healthy controls (HC) and women hospitalized with AN. Both groups underwent fMRI scanning at two time points: Time 1 (upon admission for AN patients, and a matched interval for HC) and Time 2 (after AN patients achieved normal weight, and a matched interval for HC).

The Food Choice Task involved three phases: healthiness rating, tastiness rating, and a choice phase. During the choice phase, participants indicated their preference between a presented food item and a neutral reference item. Parametric analyses were used to assess choice-related caudate activation, focusing on an a priori region of interest. Whole-brain analyses were also conducted to explore other potential brain regions involved. Statistical analyses included repeated measures ANOVAs, t-tests, and multilevel regression models to compare groups and changes over time, as well as Pearson correlations to examine relationships between neural activity and behavioral choices.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Caudate Activity and High-Fat Food Preference: The most striking finding is the inverse relationship observed in AN patients: as caudate nucleus activation decreased after treatment, their selection of high-fat foods increased. This suggests that while weight restoration may not directly curb restrictive tendencies, it might be associated with a shift in how the brain processes food choices, potentially favoring more palatable, higher-energy options.
  • Persistence of Restrictive Eating: Despite significant improvements in psychological symptoms and BMI after treatment, the actual food choices (proportion of high-fat foods selected) did not change significantly for the AN group. This reinforces the idea that restrictive eating is a deeply ingrained behavior that is resistant to change even with successful weight restoration.
  • Differential Brain Engagement: Before treatment, individuals with AN showed greater choice-related activation in the anterior caudate compared to healthy controls. This difference diminished after treatment, suggesting some normalization of this specific neural response.
  • Parietal Region Normalization: Exploratory whole-brain analyses revealed a parietal/precuneus region that showed greater choice-related activation in AN patients at Time 1, which normalized by Time 2. This region's activity change was also correlated with changes in food choice and caudate activity, hinting at a broader neural network involved in decision-making that is affected by treatment.

Key Results

  • Food Choices: No significant change in the proportion of high-fat foods chosen from Time 1 to Time 2 for the AN group (p > 0.47). The AN group consistently chose high-fat foods less often than the HC group.
  • Caudate Activation:
    • At Time 1, caudate activation differed significantly between AN and HC groups (p = 0.045).
  • At Time 2, this difference was no longer significant (p = 0.49).
  • In the AN group, decreased caudate activation from Time 1 to Time 2 was associated with an increased selection of high-fat foods (r² = -0.43, p = 0.037). This correlation was not found in the HC group.
  • Healthiness and Tastiness Ratings: No significant changes over time. High-fat foods were rated as less healthy and less tasty by the AN group compared to HC.
  • Parietal Region: Change in activation in the parietal region from Time 1 to Time 2 was significantly correlated with changes in high-fat food choices (r² = -0.476, p = 0.019) and also correlated with changes in caudate activity in the AN group (r = 0.745, p < 0.001).

Limitations and Future Work

The authors acknowledge several limitations:

  • Modest Sample Size: The study had a relatively small sample size, which may limit the power to detect subtle treatment effects. They suggest pooling data from multiple studies to increase statistical power.
  • Timing of Assessment: Patients were assessed at the end of acute inpatient treatment, shortly after weight restoration. This may not be the optimal time to capture all neural changes, as full psychological and behavioral symptom resolution may take longer.
  • A Priori ROI Focus: While they explored whole-brain analyses, the primary focus was on pre-defined regions of interest (VMPFC, DLPFC). It's possible that other relevant brain regions were not fully captured.

Future research directions include:

  • Longer-Term Follow-up: Investigating whether food choice behavior and associated neural mechanisms change with longer-term remission.
  • Targeting Neural Mechanisms: Developing treatment strategies that specifically target the identified neural mechanisms, such as the caudate nucleus, to more effectively alter eating behavior.
  • Exploring Parietal Region's Role: Further investigating the specific contribution of the parietal/precuneus region to decision-making processes in AN.

Summary updated

Abstract

Caudate activity was more strongly associated with decisions about food among individuals with anorexia nervosa relative to healthy comparison individuals prior to treatment, and decreases in caudate engagement among individuals with anorexia nervosa undergoing treatment were associated with increases in high-fat food choices. The findings underscore the need for treatment development that more successfully alters both eating behavior and the neural mechanisms that guide it.
BibTeX
@article{foerde2021changes,
  author = {Karin Foerde and B. Timothy Walsh and Maya Dalack and Nathaniel D. Daw and Daphna Shohamy and Joanna Steinglass},
  title = {{Changes in brain and behavior during food-based decision-making following treatment of anorexia nervosa}},
  journal = {Journal of Eating Disorders},
  year = {2021},
  doi = {10.1186/s40337-021-00402-y},
}
Figure