Is endogenous GLP-1 a major influence on the orbitofrontal cortex?

In this issue of Molecular Metabolism, Martin Heni and colleagues report that the level of glucagon-like peptide-1 (GLP-1) in the blood following a glucose challenge is associated with a lower level of neural activity elicited by the presentation of food cues in the orbitofrontal cortex (OFC) of humans [1]. They conclude that their data “support a model in which GLP-1 is secreted after food intake, and subsequently reaches the brain where it regulates food-related activity of the orbitofrontal cortex, independent of insulin.” Given the growing recognition of the importance of GLP-1 in mediating satiation and other metabolic consequences, it is important to understand how the authors actually conducted their studies, so as to better inform the design of future experiments.

Heni et al. used a paradigm in which fasted subjects are given an oral glucose tolerance test, with blood samples taken at 0, 30, and 120 min, coupled with the presentation of food-related pictures before and during the glucose tolerance test. Brain activity associated with the food cues was recorded in brain areas by fMRI. In an earlier publication on the same data set [2], Heni et al. established that pictures of high-calorie foods induced activation of the hypothalamus more strongly in individuals who had smaller increases of blood glucose levels during the glucose tolerance test. They also observed an inverse correlation between food-cue-induced-OFC activity and the change in plasma insulin levels, but only at the 120-min time point.

The authors re-examined their original data when writing the present manuscript and now report an inverse correlation between the food cue-induced OFC activity and the increase in plasma insulin levels 30 min following glucose ingestion in lean, but not in obese individuals. For some reason, the authors did not include the 120-min time point in this more recent analysis of insulin, although insulin appears to correlate with OFC activity in obese individuals even more strongly than in lean individuals at this time, based on visual inspection of their plotted data [2]. What is novel in the present report is that they have now measured total GLP-1 levels in the same blood samples used in the previous experiment and reached their conclusion that GLP-1 is significantly, negatively correlated with OFC activity and with the reward value of food at particular times in both lean and obese individuals. The important point is that their interpretation that obese individuals are “insulin resistant” and are simultaneously “sensitive to GLP-1” based on these data cannot be supported, due to the experimental design and the different time points that were examined. Such a conclusion would require injecting the specified molecules or their respective antagonists in a new series of experiments.

Related to this, assessments of activity in the OFC were made at only two times after ingesting the glucose, and it is not clear how the inclusion or exclusion of other times might alter the correlations with brain activity. Another noteworthy point is that a control condition (water ingestion) was included in the actual experiment reported earlier [2], but never mentioned in the later report, raising the question of whether the conclusions might look different if that analysis were used to fully control the current study. Due to the experimental design and the analytical techniques employed, an endless array of interpretations could be made using the same data set. Thus, care must be employed to avoid over-analysis or over-interpretation of any results obtained from an approach of this sort.

In the present report, insulin is used as a comparator to GLP-1 for influencing activity in the OFC. Heni et al. presumably made the comparison because they already had (and had published) insulin data from the same experiment. Both GLP-1 and insulin have short half-lives, and so the levels measured in this experiment, as sampled at specific snapshots of time, reflect mainly what happened in the previous few minutes. However, the assays used were for active insulin, but mainly for inactive GLP-1, since total GLP-1 (active plus inactive) was what was assessed. Hence, any reference to one-time levels of GLP-1 vs. insulin activity relative to brain activity becomes somewhat imprecise and arbitrary.

If, as stated, the goal of the experiment was to determine whether GLP-1 is important in influencing OFC activity, this relationship could have been established by including a group that was provided a GLP-1 antagonist (along with appropriate controls). Indeed, consistent with the conclusion of Heni et al., a prior report by van Bloemendaal et al. has already established a causal role for GLP-1 receptor agonists in the suppression of OFC activity in response to food-induced cues in both normal subjects and patients with type-2 diabetes, the effects of which were blocked via pre-treatment with a GLP-1 receptor antagonist [3]. An interesting exercise that could be addressed by Heni et al. would be to identify why particular individuals in their study had more or less GLP-1 production from the same load of oral glucose (or reduced degradation).

When considering the interpretation of the Heni et al. data, it is also important to note that the levels of dozens of hormones, nutrients, metabolites and other compounds in the blood and elsewhere in the body change when food (glucose in this instance) is consumed. Measuring only one or a few such compounds and then implying meaningfulness with regard to changes of brain activity and concluding that the results are “independent of insulin” at particular specific time points is a stretch. Had the authors assessed PYY, CCK, glucagon, ghrelin, amylin, pancreatic polypeptide, or any of numerous other meal-related compounds, there might also be changes that correlate with orbitofrontal activity. There is nothing special about GLP-1 other than the fact that it was measured; i.e., each of the other hormones mentioned above also influences food intake and/or energy metabolism, but they were not measured.

In sum, while the findings of Heni et al. are interesting, conclusions concerning endogenous GLP-1's influence on the orbitofrontal cortex should be considered premature until a different paradigm is applied.