As expected, obese adolescents had significantly higher ratings of disinhibition, hunger, and cognitive restraint on the TFEQ. Although higher levels of cognitive restraint among obese adolescents would on first inspection appear counterintuitive, it conforms to the described model of “rigid restraint” in which an individual with disinhibited eating and cognitive restraint may tend to restrict food in some situations but grossly overeat in others (20
Our novel neurostructural results among obese adolescents are consistent with findings in the adult literature (8
) demonstrating gray matter volume reductions. In our adolescent sample these decreases were most marked for the orbitofrontal cortex, a brain region important in impulse control, but also showed a weak trend for whole frontal lobe. We speculate that the more subtle volume reductions that exist in other brain regions among obese adolescents may actually reach statistical significance in an expanded sample.
Importantly for this report, we found the group with excess weight to not only have higher disinhibition scores on the TFEQ, but lower performance on cognitive tests reflecting brain functions thought to be central to behavioral inhibition, even when controlling for IQ. Out of the frontal lobe regions and functions we measured, we were particularly interested in ascertaining the relationship between the disinhibition factor of the TFEQ and the OFC, a brain region that is very important for behavioral inhibition (impulse control). We selected the Stroop because it is the only one of our frontal lobe tasks (including those that tap executive functions) that specifically tests the ability to inhibit automated responses. This is the direct cognitive parallel of the behavioral (disinhibition factor of the TFEQ) and brain region (OFC) also involved in inhibition of automatic responses. Our interest was to ascertain the functional (Stroop vs. other frontal tasks that do not measure response inhibition) and anatomic (OFC) specificity of our findings and their association to the disinhibition factor of the TFEQ.
We also found significant associations between disinhibition factor scores and both BMI and OFC volume. When the relationship between disinhibition and OFC volume was examined separately in lean and obese participants, we found a strong negative association only for the lean group. It is possible that obese individuals have already experienced a critical level of disinhibition-(which as we demonstrated is associated with BMI), whereby additional disinhibition is not as clearly reflected in further changes in OFC, but perhaps in different brain regions or networks not assessed as part of this study. Another possibility for these different findings for each of the two weight groups is that given that the obese groups has higher degree of item endorsement, they may be more susceptible to issues of social desirability and therefore they may be less likely to fully report the extent of their behavioral disinhibition in eating, dampening the association in this group. Lastly, it is also possible that range restriction, namely the phenomena of correlations decreasing when the variance is decreased as occurs when we divide our sample in two could be affecting our results.
While our study finds that disinhibition in feeding behavior is associated with reductions in executive functioning and frontal gray matter volumes, the cross-sectional nature of our design does not allow us to address the issue of directionality or causation. With that being said, there are several plausible theories regarding the direction of these associations.
One possibility is that primary structural or functional brain deficits lead to disinhibited eating and reductions in neurocognitive function. This line of reasoning is partially supported by work showing disinhibtion in eating behavior to presage increased caloric intake (21
) and obesity (22
). It is also consistent with functional imaging work demonstrating that individuals, who in response to visualized intake of palatable foods show weaker activation of brain reward circuits, are at elevated risk for future weight gain (23
); perhaps they need a larger stimulus (more food) to derive the same reward response.
Another possible explanation is that brain structural deficits like those demonstrated in this study result from obesity and its associated insulin resistance. This possibility is supported by a 24-year longitudinal study showing increased BMI beginning in middle age correlated with decreased temporal lobe volume in later life (24
). Also supporting this effect order is our own work in adults where we find that hippocampal volumes were associated with impairments in glucose tolerance (25
) as well as that in adolescents with T2DM, where we find cognitive impairments and reductions in frontal lobe volumes and in white matter microstructural integrity (26
). We posit that the obesity-associated insulin resistance exhibited by our group of adolescents with excess weight may contribute to decreased executive function and structural deficits. We have described a possible model for these effects (27
) in which we hypothesize that insulin resistance is associated with decreased brain vascular reactivity related to endothelial dysfunction. We know that during brain activation, such as occurs when performing a cognitive task, there is an increase in synaptic activity in the brain region involved. In the normal brain this results in regional vasodilation and thus an increase in glucose availability to that region to support the increased cognitive demand (28
). Therefore, vascular reactivity, which is integral to well-regulated cerebral blood flow, is key for maintaining an optimal neuronal environment during brain activation (29
). Research showing endothelial dysfunction in obese children, even before the development of diabetes (30
), further supports this premise. In addition, the inflammatory marker C-reactive protein (CRP) was elevated in our obese adolescents. In studies examining large cohorts of adults, investigators have found increased levels of inflammatory cytokines as putative mediators of cognitive decline among individuals with metabolic syndrome (31
). A possible mechanism for these cognitive effects is provided by animal data demonstrating that excess inflammatory cytokines can decrease long term potentiation (LTP), a process understood as essential in consolidation of memory in the hippocampus. Inflammatory cytokines may also cause impairment in neurogenesis and neuroplasticity, processes vital to the formation of memories and the maintenance of structural neural integrity.
A third possibility is that these effects are bidirectional whereby behavioral disinhibition predisposes to obesity, which may negatively impact brain areas responsible for executive function and inhibition of caloric intake, thus causing a vicious cycle of dysfunction. This third possibility could help explain why it so difficult for individuals to lose weight once it has been gained.
We are encouraged by the fact that among the few brain regions that we evaluated, the OFC, a brain region that has been demonstrated as important in behavioral inhibition in both animal and human studies, had the most significant volume reduction among obese adolescents. Our findings, including lower performance on cognitive tests thought to require intact OFC, coupled with volume reductions in this area associated to behavioral disinhibition point to its likely importance in weight gain.
This study has some clear limitations. First, it is a cross-sectional view that does not allow us to comment on clear causality. Second, given our relatively modest sample size we restricted our measurements to brain regions that in previous studies had either been found to have been associated to obesity or disinhibition, or those that we had good theoretical reasons to believe could be involved. Therefore, it is possible that there are other brain areas, which we did not evaluate, that may also be involved. A third limitation of our study is that we only have participants' current weight and we cannot comment on duration of obesity; the sample that we studied is likely to have considerable variability in duration of obesity and its associated insulin resistance. Nevertheless, our study has significant strengths, including the careful matching between groups, the multidimensional evaluations conducted, and the unbiased MRI methods utilized in the analyses of the MRI data.
To better understand the issues described here, future work should evaluate subjects longitudinally, tracking the development of obesity across time while concomitantly measuring cognitive, behavioral, and neurostructural changes. Alternatively, our understanding could also be improved through a study designed to examine the consequences of a successful obesity treatment (e.g. bariatric surgery), and thus ascertain whether some of these deficits are reversible. Furthermore, future work should evaluate other possible associated factors such as pro- and anti-inflammatory cytokines as well as utilize more sensitive MRI techniques such as diffusion tensor imaging (DTI).