**4. Discussion**

The present study was aimed at examining the involvement of different peripheral obesity-related biomarkers (namely, ghrelin, IR, and leptin/adiponectin ratio) to explain different eating styles, i.e., external, emotional, and restrained patterns, in childhood and adolescence obesity.

Our findings show that plasma ghrelin concentrations were negatively related to restrained eating after controlling for sex and age (nonsignificant) effects. In other words, a poor obesity-related biomarker (i.e., lower ghrelin concentrations) was related to restrained eating in childhood and adolescence obesity, as expected. These findings extend previous research that supports the restraint theory [21,23] by suggesting that restrained eating is not only an "obese" eating style, but also a "higher-risk-obesity" eating style. It should be noted that our results contrast with some previous studies conducted in normal-weight adults, indicating that restrained eating is positively correlated [41] or not correlated [40] with ghrelin. According to preliminary research, this discrepancy in the direction of the relationship could be due to the obesity status [63]. Moreover, different measures of restrained eating could reflect distinct conceptualizations of what it means, which might also explain inconsistencies in the literature in this regard. Mainly, scales measuring restrained eating differ in the purpose for which authors developed them [22], i.e., to assess chronic dieters who cyclically restrain their intake or over-eat [64], or individuals who success in reducing their intake [11]. Future studies should be cautious in this concern and simultaneously include individuals with normal and excess weight through different life course stages to disentangle the nature of this relationship. Finally, IR and leptin/adiponectin ratio were not significantly related to restrained eating. In the study conducted by Schur and colleagues [41], leptin and insulin were also measured, but they were not related to cognitive restraint, similar to our results.

External eating was hypothesized to be underlaid by high-risk peripheral biomarkers, specifically lower ghrelin levels, higher IR, and higher inflammation indicators (i.e., leptin/adiponectin ratio). However, our results showed no significant contribution from these biochemical indices to explain external eating once sex and age effects were partialized out. Although there are no previous studies examining the role of ghrelin, IR, leptin, and adiponectin in self-report eating styles in obesity, preliminary evidence exists regarding the influence of ghrelin on the brain activity involved in processing external food cues. In this regard, Malik and colleagues found an increased brain activity response to food-related visual cues after ghrelin administration, which was indeed correlated with self-reported hunger ratings [65]. However, our results suggest that this effect is not extended to the habitual external eating style in childhood obesity.

Similarly, contrary to what was hypothesized, emotional eating was not significantly related to any obesity-related biomarkers in our sample. Sex was the only statistically significant predictor of emotional eating, indicating that women showed higher scores on emotional eating than men, which is consistent with previous findings [20]. Noteworthy in this regard is a previous cohort study that found that higher leptin was cross-sectionally related to emotional overeating at age 7 years, but it was not prospectively associated with emotional overeating three years later after controlling for BMIz. However, this study used a parent-reported instead of a self-report measure of eating styles, that is, parents (or main caregivers) responded to the questionnaire instead of the children on their own (e.g., "My child eats more when annoyed"). Therefore, some biases of their parents could have influenced these results.

To our knowledge, this is the first study to comprehensively consider both physiological risk indicators of obesity and eating styles in a young sample. Theoretically, this study suggests that metabolic dysfunction of adipokines and IR might not yet be involved in eating behavior patterns in obesity in the early stages of development (i.e., childhood and adolescence). This fact promisingly implies that these hormonal disruptions could not have impaired the usual eating responses yet, even if they might be affecting current satiety perceptions, as pointed out by previous research [66]. Given the predictive role of eating styles on prospective weight in adults [15], it can be thought that the clinical condition of obese children and adolescents could still be reversed without consequences in their way of relating to food. In particular, this can be applied to eating styles that involve appropriate processing of interoceptive satiety signals, namely the absence of eating as a maladaptive emotion regulation strategy (emotional eating) and the adaptive levels of sensitivity towards food-related rewards (external eating). However, this is not the case with the cognitive suppression of these internal hunger signals, as evidenced by the negative association between restrained eating and ghrelin levels. In other words, it seems

that cognitively ignoring the interoceptive signs of hunger is not a good eating strategy in children and adolescents with obesity, since it is related to a poorer ghrelin profile.

Although interoceptive cues might be ignored at a cognitive level in those with lower ghrelin concentrations, it might be plausible that interoceptive abilities are not yet disturbed in young obese individuals, unlike obese adults [27]. This could explain why emotional and external eating styles were not related to metabolic biomarkers in our sample. Taking all into account, and in light of preliminary research on the role of interoceptive processing as an underlying mechanism of eating styles in adults [67], future studies should examine the interaction between peripheral biomarkers involved in satiety, their perception and processing in the central nervous system (i.e., interoception), and eating styles, in childhood obesity. From a clinical point of view, our findings suggest that dietary treatments mainly focused on restricting the amount of food intake (which inevitably involves ignoring the sensations of appetite) rather than on eating healthy food are contraindicated. In this line, previous research has shown the negative effects of dieting for healthy management of obesity during childhood and adolescence in the long term [68].

The current study presents some limitations that should be noted. First, the sample size was not very large, so the statistical power might be compromised, leading to not finding a statistical effect in this sample that could exist in the population. Second, the age range of participants in this study covered two different developmental stages (childhood and adolescence). Although statistical control has been applied over the potential effect of age (and sex) on the relationships of interest, future studies should replicate this research throughout the different development stages. Third, our results refer only to children and adolescents with moderate and severe obesity (BMIz > 2), so they cannot be generalized to other populations such as children and adolescents with normal weight or overweight (BMIz < 2), nor to adults. Fourth, our findings may be biased by the recruitment period, which was characterized by the COVID-19 pandemic for a portion of our sample. In this regard, a recent longitudinal study showed that external, emotional, and restrained eating styles remained stable during the lockdown in college students [69], supporting the notion of eating styles as trait-like rather than state-like patterns. Although these findings might be different in children and adolescents with obesity, to our best knowledge, there is no evidence to think that the associations between eating styles and biomarkers may vary according to circumstantial factors. Finally, given the cross-sectional nature of the design of the current study, causal relationships cannot be established. Future studies should adopt longitudinal designs to determine the directionality between ghrelin and restrained eating in childhood obesity and the mediating mechanisms by which this relationship occurs.

Despite these limitations, this study contributes substantially to a better understanding of the interactions between two different types of crucial factors (biochemical and psychological) involved in childhood and adolescent obesity, namely circulating hormones and eating styles.
