Our study has two main findings. First, we show a similar course in plasma glucose after glucagon and arginine stimulation, while macimorelin had no effect on glucose levels, and second, we show a strong correlation between the decrease in glucose levels and increase in copeptin after glucagon and arginine stimulation.
Arginine vasopressin (AVP) is produced by hypothalamic magnocellular and parvocellular neurosecretory cells projecting to the posterior pituitary . While magnocellular neurons are supposed to be involved in the osmoregulation, parvocellular neurons contain co-packaged AVP and corticotropin releasing hormone (CRH) in their secretory granules and stimulate adrenocorticotropin (ACTH) secretion from the anterior pituitary after non-osmotic stimuli [12, 13]. Hypoglycemia is one of the most potent non-osmotic stimuli for the pituitary gland and induces an acute increase in prolactin, ACTH, growth hormone (GH)  and AVP . Specifically, for AVP, Kacheva et al. showed an increase in copeptin after hypoglycemia induced by an insulin tolerance test with maximum copeptin levels after 47 minutes, while the glucose nadir was at 30 minutes . Therefore, the insulin-hypoglycemia test can be used to assess anterior and posterior pituitary functions.
Whether a drop in glucose levels itself directly stimulates AVP release or whether this rapid drop triggers multisynaptic stress pathways  leading to AVP release remains elusive and cannot be answered with our correlation study. A study in primates suggests that the mammillary nuclei and the lateral hypothalamic nuclei might be glucose-sensing areas, independently from osmolality and epinephrine secretion . The pathway connecting the glucose-sensing nuclei to GH release might include the ventromedial nucleus and the median eminence of the hypothalamus . Our data demonstrate a clear correlation between a decrease in glucose and an increase in copeptin levels upon glucagon and arginine administration. We observed a rapid drop of glucose levels, but no hypoglycemia, which could support the hypothesis of a glucose-drop activating central sensing area rather than hypoglycemia-like stress response.
The glucagon stimulation test is considered as an alternative to the insulin-hypoglycemia test for the evaluation of anterior pituitary function in adults and children . One major advantage is the combined evaluation of ACTH and GH secretion . The exact mechanism of glucagon-induced GH secretion is unclear; however, already in 1975 Mitchell and colleagues suggested a relationship between the rapid drop of elevated glucose levels observed after administration . Glucose course after glucagon injection might mimic an insulin-induced hypoglycemia state, characterized by rapid decrease of glucose levels, but without leading to absolute hypoglycemia. In support of this, the same group demonstrated a suppressed GH peak upon glucagon stimulation if the participants received an additional continuous glucose infusion preventing the rapid drop of glucose levels. We have recently shown a marked increase of copeptin after subcutaneous glucagon injection simultaneously to GH, supporting the hypothesis of a common stimulus, e.g., glucose drop induced stress response, for both hormones .
Arginine might act in two possible synergistic ways, in the first part of the test via direct central hypothalamic/pituitary stimulation, and second half of the test via the change in glucose levels. In support of this, a continuous increase of copeptin was observed within the first 30 minutes after arginine infusion, whereas the increase of copeptin after glucagon injection was only observed after glucose nadir. More precisely, L-arginine conversion into L-citrulline is catalyzed by nitric oxide synthases (NOS) located in many cells including hypothalamic AVP producing neurons with nitric oxide (NO) as the product . In vivo studies have proposed a direct stimulatory effect of NO on AVP release . Interestingly, administration of NOS inhibitors to rats induced a mild polyuria-polydipsia syndrome probably due to deficient AVP secretion . In contrast, the exogenous administration of a NO donor, i.e., arginine, could explain the direct acute release of AVP.
Glucose course after arginine infusion was very similar to the one observed after glucagon injection, i.e., an initial increase followed by a rapid decrease to low-normal levels, although with a smaller drop (delta) in glucose. Arginine leads to depolarization of pancreatic beta cells and consequently voltage-dependent calcium channels, which stimulates insulin secretion , possibly explaining the later drop in glucose. Besides its direct central effect, arginine might stimulate AVP in addition via drop in glucose levels explaining the later constant plateau of copeptin in the second half of the test.
Macimorelin, a synthetic ghrelin receptor agonist, was recently proposed as a simplified test in the diagnosis of GH deficiency . It is characterized by rapid absorption after oral intake with a plasma peak concentration of around 60 minutes , at which a GH peak is observed in healthy adults. We recently showed no effect on copeptin levels and now show that there is no effect on glucose levels. Macimorelin might act directly on hypothalamic and pituitary ghrelin receptors, leading to a direct and selective GH release or modulation of growth hormone releasing hormone (GHRH) . Both hypothalamus and pituitary are potential sites of action; in animal models, hypothalamic-pituitary disconnection led to decreased GH release after administration of ghrelin receptor agonists supporting the hypothesis of hypothalamic receptor action. Conversely, pituitary cell cultures show a marked increase of GH after incubation with ghrelin receptor agonists . In summary, we assume that macimorelin selectively stimulates GH release via specific receptors rather than a direct glucose-, a secondary glucagon- or stress-induced GH and AVP/copeptin secretion.
Our study has limitations. The main limitation of our analysis is that we could not measure plasma insulin and glucagon levels, because the preanalytic requirements were not met. Second, this is a post-hoc analysis of three prospective studies.
In conclusion, this study shows a correlation between the drop in glucose levels and the increase in copeptin levels upon arginine and glucagon stimulation. In contrast, upon macimorelin stimulation, glucose and copeptin levels remain unchanged. These findings suggest that a drop in glucose levels might be a possible explanation for the increase in copeptin levels in the first two stimulation tests. Whether the drop in glucose levels itself leads to copeptin secretion, or whether it mirrors other stimuli such as hypoglycemia-induced stress cannot be answered with our study and could be better defined in further research measuring glucagon and insulin levels throughout the stimulation tests.