Sensitization of the reinforcing value of high energy density foods is associated with increased zBMI gain in adolescents

The study investigates whether sensitization of the relative reinforcing value (RRV) of food predicts changes in standardized BMI (zBMI) during adolescence, a critical developmental period marked by heightened reward sensitivity and less mature inhibitory control. RRV reflects motivation to work for food and is linked to higher BMI and weight gain. Delay discounting (DD), a preference for immediate over delayed rewards, can interact with RRV (reinforcement pathology) to predict obesity-related outcomes. Prior adult studies indicate that repeated intake of highly palatable, high energy density (HED) foods can increase their reinforcing value (sensitization) particularly among individuals with higher weight status, whereas low energy density (LED) foods rarely produce sensitization. The study aims to determine whether HED or LED sensitization predicts adolescent zBMI trajectories over two years and whether DD moderates these relationships.

Prior work shows higher food RRV associates with higher BMI and weight gain, and DD moderates this relationship, forming reinforcement pathology linked to obesity risk and treatment resistance. Sensitization—an increase in RRV after repeated intake—parallels mechanisms described in substance use literature and has been associated with higher BMI and weight gain in adults and higher cross-sectional zBMI in adolescents. Adult studies demonstrate that daily intake of highly liked HED snacks can decrease RRV in individuals without obesity but increase it in individuals with obesity; LED foods rarely induce sensitization and show limited associations with weight outcomes. Research also suggests highly palatable, high sugar/fat foods strongly activate reward circuitry. Evidence on LED foods indicates low rates of sensitization, even with intensified exposure protocols, and no clear protective effect on weight. These findings motivate examining HED versus LED sensitization, their interaction, and potential moderation by DD in adolescents.

Design: A 2-year prospective observational cohort of 201 adolescents. Baseline comprised 5 visits over 5–6 weeks; follow-ups occurred at 6, 15, and 24 months. Participants: Adolescents aged 12–14 at baseline (zBMI between −1.5 and 2.0; non-obese) and 14–16 at 24 months. Inclusion required moderate liking of at least one HED and one LED food and willingness to consume them daily for two 2-week periods, English proficiency, and likelihood to remain in the region for 2 years. Exclusions included medical conditions/medications affecting appetite and food allergies. Consent/assent obtained. IRB-approved. Anthropometrics: Height and weight measured using digital SECA stadiometer and scale, without shoes and in light clothing. zBMI computed using CDC age- and sex-specific growth charts. RRV Task: Assessed at visits 2–5 using a validated computer-based progressive ratio task offering choices between responses for a preferred food (HED or LED) versus 2 minutes of a highly liked sedentary activity. LED options (ED ≤ 1.0 kcal/g): fruit cups, applesauce, low-fat yogurt; average portion 62.2 g (~48.3 kcal). HED options (>4 kcal/g): chips, cookies, chocolate candy; average portion 16.8 g (~90 kcal). Response requirements doubled across schedules (20, 40, 80, … up to 5120 clicks). Area under the curve (AUC) of responses across schedules quantified RRV. Sensitization Paradigm: Participants received 14 daily portions of their chosen HED or LED food for each 2-week period between visits 2–3 and 4–5; order randomized. HED portions ~58 g/300 kcal; LED portions ~200 g/160 kcal. Daily text reminders; participants confirmed consumption and timing by text/phone. Compliance >70% for both foods required. Sensitization was computed as post-daily intake AUC minus baseline AUC. Categorization: increase >1 labeled “sensitization”; ≤1 labeled “satiation.” Delay Discounting (DD): Measured at baseline and each follow-up via an adjusting amount task choosing immediate amounts versus $50 after delays (1 day to 6 months). Indifference points determined per delay; discounting rate (k) derived using Mazur’s hyperbolic model. Nonsystematic data excluded per standard algorithm. Area under the curve from baseline used for analyses. Questionnaires: Demographics (race/ethnicity, income, education, marital status, occupation), Pubertal Development Questionnaire (adolescent and parent reports averaged), subjective ratings (hunger, thirst, liking, wanting) via 100 mm VAS around preload and RRV tasks, and the Dutch Eating Behavior Questionnaire (DEBQ) for dietary restraint. Sample Size and Analysis: Power based on prior effect size (0.19), alpha 0.05, power 0.80, requiring ~180 participants. Multilevel models with appointments (level 1) nested within individuals (level 2) examined zBMI at baseline, 6, 15, and 24 months. Time modeled as months since baseline. Fixed effects included baseline zBMI, sex, DEBQ, pubertal development; primary predictors were sensitization categories and interactions with time. Models: (1) HED sensitization × time; (2) LED sensitization × time; (3) HED and LED sensitization and their interactions × time; (4) inclusion of DD and its interactions with HED sensitization and time. Missingness analyses indicated no baseline differences between completers and non-completers on key covariates.

- Sample: 201 adolescents (52% female), predominantly non-Hispanic White (71%), upper-middle income; 165 satiators and 36 sensitizers (HED categorization). No baseline differences by sensitization in demographics or baseline zBMI; sensitizers had higher baseline RRV of seated activity (F(1,198)=4.07; p=0.045). - HED Sensitization: Significant interaction with time predicting zBMI increase (Visit × HED sensitization β=0.0070; SE=0.0033; p=0.035). Baseline zBMI (β=0.9058; p<0.0001) and pubertal development (β=0.0113; p=0.039) were significant positive predictors of zBMI. No main effect of HED sensitization or sex interactions. - LED Sensitization: No main effect on zBMI (β=−0.0824; p=0.173) and no interaction with time (β=0.0013; p=0.639). A sex × LED sensitization interaction was significant (β=0.1743; p=0.019), but there was no time interaction, indicating no effect on zBMI trajectories. - HED × LED Interaction: No significant interactions between HED and LED sensitization or their interaction with time on zBMI (all p>0.39). - Delay Discounting (DD): No main effect of DD on zBMI (β=−0.0007; p=0.857), no interaction with time (β=−0.0001; p=0.611), and no moderation of HED sensitization effects (all interactions p≥0.668). Including DD reduced the HED sensitization × time effect to a trend (β=0.0102; p=0.096), likely reflecting reduced power in the more complex model. - Trajectories: Model-estimated zBMI increased over time among HED sensitizers and slightly decreased among satiators (Figure 2). Group means suggested greater zBMI change among sensitizers (mean change ~0.197±0.09) than satiators (~0.04±0.04), though unadjusted difference testing was not the focus.

Findings support that sensitization of the reinforcing value of HED foods is a behavioral phenotype that prospectively predicts accelerated zBMI gain during adolescence, extending prior adult findings and earlier cross-sectional adolescent observations. LED sensitization showed no protective association with zBMI change, and interactions between HED and LED sensitization were not significant, indicating HED sensitization is the primary risk factor among the RRV-based phenotypes evaluated. DD did not moderate the relationship between HED sensitization and zBMI change, possibly due to high variability in adolescent DD, the dominant influence of motivational systems over inhibitory control during adolescence, and potential overlap between constructs. The results highlight sensitization to HED foods as a promising, novel intervention target for preventing excess weight gain in youth, potentially through strategies that reduce sensitization or enhance the relative value of non-food alternatives.

Sensitization of the RRV of HED foods prospectively predicts increased zBMI gain over two years in adolescents without obesity, replicating adult findings and extending them developmentally. Sensitization of LED foods does not confer protection, and DD does not moderate the HED sensitization–zBMI relationship. Sensitization thus represents a novel, reproducible behavioral risk phenotype and potential intervention target. Future research should test behavioral strategies to reduce HED sensitization, assess whether increasing the RRV of non-food alternatives mitigates weight gain, and evaluate the stability and modifiability of sensitization across developmental stages and diverse populations.

- Sample demographics: Predominantly White and upper-middle class, limiting generalizability and power to test interactions with socioeconomic status, race/ethnicity, or food insecurity. - Developmental timing: Girls entered further along in pubertal development than boys, potentially creating different growth trajectories at baseline. - Group imbalance: Fewer sensitizers than satiators reduced statistical power for interaction tests. - COVID-19 impact: Laboratory shutdown (March 15–July 1, 2020) affected timing of 24-month assessments; although 80% retention was achieved, higher retention would strengthen findings.