The exact physiological basis of acute growth hormone (GH) suppression by oral glucose is not fully understood. Glucose-mediated increase in hypothalamic somatostatin seems to be the most plausible explanation. Attempts to better understand its underlying mechanisms are compromised by species disparities in the response of GH to glucose load. While in humans, glucose inhibits GH release, the acute elevation of circulating glucose levels in rats has either no effect on GH secretion or may be stimulatory. Likewise, chronic hyperglycemia alters GH release in both humans and rats nonetheless in opposite directions. Several factors influence nadir GH concentrations including, age, gender, body mass index, pubertal age, and the type of assay used. Besides the classical suppressive effects of glucose on GH release, a paradoxical GH increase to oral glucose may be observed in around one third of patients with acromegaly as well as in various other disorders. Though its pathophysiology is poorly characterized, an altered interplay between somatostatin and GH-releasing hormone has been suggested and a link with pituitary ectopic expression of glucose-dependent insulinotropic polypeptide receptor has been recently demonstrated. A better understanding of the dynamics mediating GH response to glucose may allow a more optimal use of the OGTT as a diagnostic tool in various conditions, especially acromegaly.
Growth hormone (GH) secretory pattern in rodents and in humans is pulsatile and sexually dimorphic [1, 2]. GH secretion is classically orchestrated by 2 hypothalamic hormones: GH-releasing hormone (GHRH) and somatotropin release-inhibiting factor (SRIF), which exhibit a stimulatory and an inhibitory effect, respectively, on the somatotroph cells in the anterior pituitary . Ghrelin, a gut-derived peptide and a ligand of the GH secretagogue receptor is increasingly recognized as the third regulator of GH secretion with a marked stimulatory action. In addition to its excitatory impact on GHRH release and a weaker inhibitory action on somatostatin, ghrelin directly stimulates GH secretion from pituitary somatotroph cells [4, 5]. IGF-1 has a major inhibitory action on GH release through feedback at both the hypothalamus and the pituitary. It seems to inhibit both spontaneous and GHRH-stimulated GH release and displays stimulatory effects on somatostatin neurons  (Fig. 1). A number of factors including neuropeptides, neurotransmitters, peripheral hormones such as leptin, sexual steroids, glucocorticoids and various metabolic signals are implicated in the complex regulation of GH release.
The impact of glucose on GH secretion has been demonstrated in the early 1960s  and confirmed later by many other authors. It is well established that hypoglycemia stimulates GH release  and hence, insulin-induced hypoglycemia is used clinically to assess the integrity of GH secretion.
On the other hand, oral glucose administration suppresses GH secretion and likewise is the standard method for assessing inhibitory control of GH release . Failure of GH suppression is characteristic of active acromegaly, though it can also be observed in other pathological conditions such as chronic kidney disease, liver disease, and anorexia nervosa (Table 1).
The purpose of this review is to highlight the known mechanisms underlying the effects of glucose on GH in normal and pathological conditions and summarize the factors influencing GH response to glucose overload in healthy individuals and patients with acromegaly.
Methodology for Literature Search
A PubMed search was conducted from the period of 1946 until 2018 using the search terms: “acromegaly,” “diabetes,” “glucose,” “hyperglycemia,” “growth hormone,” “growth hormone releasing hormone,” “nadir growth hormone,” “oral glucose tolerance test,” “paradoxical response,” and “somatostatin.”
The above terms were used in mixed combinations. Boolean operators and truncations were used to expand our search results. References from the selected pertinent articles, and publications available in the authors’ libraries were also used.
GH Response to Oral Glucose and to Chronic Hyperglycemia
The precise mechanisms of GH suppression by oral glucose are not completely understood. Attempts to determine the exact mechanisms are compromised by species disparities observed in the physiological responses of GH to glucose overload or deprivation. In this section, we briefly review existing data in humans and rodents in both acute and chronic hyperglycemia.
Following oral glucose administration in humans, a transient suppression of plasma GH levels for 2–3 h is observed followed by a delayed rise occurring at 3–5 h post glucose ingestion [7, 10]. This initial suppression seems to be related to a glucose-mediated increase in hypothalamic somatostatin release. Evidence supporting this hypothesis emerges from the findings that in healthy individuals, GH secretion in response to GHRH or GH secretagogue is diminished after an oral glucose load [11, 12]. Furthermore, the inhibitory effect of glucose is reversed with the acetylcholinesterase inhibitor pyridostigmine, a substance thought to suppress somatostatin release from the hypothalamus . These findings support the hypothesis that oral glucose load is associated with a somatostatin release into the hypophyseal portal blood suppressing GH levels. The delayed GH rise would result from a decrease in somatostatinergic tone and hence an increase in GHRH . Subsequently, the available pituitary stores of GH are released leading to a rebound rise in GH.
Recently, the involvement of ghrelin in the regulation of post-glucose GH has been suggested . In a multivariate analysis, ghrelin was the only predictor for fasting and peak GH levels following oral glucose load in women  (Fig. 1). Of note, some authors have shown that the nadir GH levels obtained following glucose intake are not specific to glucose and also occur after water ingestion or even spontaneous GH measures. They showed that glucose rather inhibits spontaneous GH surges [17, 18]. The exact molecular mechanisms by which glucose modulates GH release are yet to be determined.
Stanley et al.  have demonstrated that changes in glucose and specifically hypoglycemia activate GHRH neurons through the glucokinase activation in EGFP-tagged ribosomal protein constructed mice. However, the role of direct glucose sensing in GHRH neurons on GH modulation is questionable, since hypoglycemia in mice causes a reduction in GH release.
While in humans, glucose load inhibits GH release, the acute elevation of circulating glucose levels in rats has either no effect on GH secretion or may be stimulatory (Table 2). A possible direct pituitary action of glucose cannot be ruled out. Although little effect of high glucose on basal and GHRH-stimulated GH levels has been observed in rat anterior pituitary cells in static incubation  or in perifusion , sustained glucose elevation (72 h) increased GH secretion in cultured cells . In in vivo rat models, GH secretion is inhibited by both hypo and hyperglycemia most likely via the stimulation of somatostatin release. Both acute hypo- and hyperglycemia stimulate SRIF mRNA, while GHRH mRNA is stimulated only by hyperglycemia .
Chronic hyperglycemia, manifested clinically by diabetes mellitus, alters GH release in both humans and rats though in opposite directions.
Type 1 diabetes mellitus (T1D) patients display an increased pulsatile GH secretion and an exaggerated GH increase after GHRH administration [24, 25]. Failure of T1D patients to increase their GH responses to GHRH following pyridostigmine treatment suggests a decrease in hypothalamic somatostatin release in these patients . Evidence suggests that portal vein insulin deficiency contributes to GH dysregulation by downregulating hepatic GH receptors explaining the state of “GH resistance” in this population . Furthermore, decreased hepatic production of IGF-1 observed in diabetic patients results in excess GH secretion by lack of negative feedback action .
Data in patients with T2D have yielded conflicting results. Spontaneous GH secretion as well as GHRH stimulated GH may be increased, normal or decreased . One of the main determinants of these differences was obesity where obese T2D patients display significantly reduced GH responses to GHRH compared to lean individuals and to non-obese diabetic patients [29, 30].
Diabetes mellitus in rodents seems to reduce pulsatile GH secretion and attenuate GH secretory response to GHRH (Table 3). In streptozotocin (STZ)-induced type 1 diabetic rat models, GH secretion is decreased along with a reduction in GHRH mRNA and SRIF mRNA, suggesting a differential effect of acute and chronic hyperglycemia on hypothalamic-pituitary GH regulation [31, 32]. Hyperglycemia seems to directly impact pituitary somatotroph cells by blunting GH release in response to GHRH or by increasing somatostatin release [33-36]. This attenuated GH response is restored after treatment with somatostatin antiserum or pentobarbital anesthesia, which presumably suppresses somatostatin release [37, 38]. In isolated pituitary cells from STZ-induced diabetic mice, a decrease in basal GH levels and in GHRH-stimulated GH release as well as a resistance to inhibitory action of SRIF on GHRH stimulated GH release have been reported . However, STZ dose and time of assessment after treatment may alter GH release  and should be taken into account when interpreting results. Indeed, Liu et al. have demonstrated that high STZ dose (100–200 mg/kg) impairs GH secretion by blocking GH secretory granules in somatotroph cells often inducing cell rupture. This suggests that some of the decrease in pituitary GH release in high dose STZ-treated rats could be secondary to the toxic destruction of the somatotroph cells.
Factors Influencing GH Suppression to Oral Glucose
Nadir GH concentrations are influenced by physiological factors such as age, gender, and body mass index (BMI) and vary according to the GH assay used. Table 4 lists the mean nadir GH in different studies investigating plasma GH after OGTT in healthy subjects by different GH assays.
Gender: A higher nadir level of GH during OGTT has been observed in women compared to men in healthy populations [42-47] and in patients with acromegaly [48, 49]. These higher GH concentrations after OGTT seem to be a consequence of higher basal values rather than a lower suppressive effect of glucose. Indeed, greater basal GH concentrations have been documented in premenopausal women and in women treated with oral estrogens [50-52]. However, these gender-specific differences have not been consistent in all studies [50, 53-55].
Body Mass Index
Similarly, an inverse relationship between GH concentrations and BMI has been demonstrated [43, 56, 57] especially in obese subjects with a BMI > 30 kg/m2 . In 381 subjects with normal pituitary function, BMI was the major determinant of GH nadirs following oral glucose. When stratified according to BMI, GH nadirs were significantly different across all groups . In general, investigations in obese subjects showed reduction in both spontaneous [60, 61] and stimulated GH secretion . Specifically, the mass of GH secreted per burst is diminished and the metabolic clearance rate of GH is increased .
GH secretion decreases with age at a rate of 14% per decade . The decrease in GH results from a marked reduction in GH pulse amplitude rather than frequency  most likely related to a relative deficiency in GHRH and ghrelin secretion and an increase in SRIF release . Likewise, some authors have reported lower nadir GH levels in older individuals [47, 57, 65]. The use of age-adjusted GH and nadir GH cut-off values to define biochemical remission of acromegaly following surgery has been suggested . However, the effect of age on GH nadir levels has not been consistently observed across all studies [50, 59].
The influence of pubertal stage on GH nadirs after oral glucose has been investigated. Higher nadir GH levels were reported in mid-pubertal girls and boys compared to children in other pubertal stages .
In HIV-infected patients, those with lipodystrophy did not display a rebound of GH during a 3-h OGTT suggesting that in the setting of fat redistribution, patients exhibit prolonged post glucose GH suppression .
With the progression from older polyclonal radio-immunoassay to more sensitive and specific GH assays, lower GH cut-off values have been suggested .
The influence of assay methods on nadir GH concentrations has been repeatedly demonstrated during the last decades [43-45, 72, 73] (Table 4). Although nadir GH levels results from all assays strongly correlate, mean GH concentrations vary widely hindering the establishment of optimal normative data. For instance, Arafat et al.  demonstrated that mean GH concentrations obtained with the immunochemiluminetric assay, Immulite, were two to threefold higher than the Nichols immunochemiluminetric assay and six-fold higher than the ELISA assay. Furthermore, the use of different standard preparations for GH calibration also accounts for inter-assay variability in different laboratories. Finally, the reporting of results in biological activity (mUI/L) or mass units (µg/L) together with the use of variable conversion factors constitutes an additional cause for assay variability. Currently, all GH assays should be calibrated using the recombinant calibrator 98/574 and all GH assay results should be reported in mass units [74, 75].
OGTT in the Diagnosis of Acromegaly
In contrast to healthy subjects, oral glucose fails to suppress GH in acromegaly. The criteria for GH suppression after oral glucose have evolved in recent years with the availability of more sensitive techniques. However, despite several reports highlighting the unmet need for standardization of GH assays during the last decades, little progress has been noted to date [76-78]. The cutoff for nadir GH after OGTT seems to largely depend on the GH assay used . The 1 µg/L cutoff used to define acromegaly probably needs to be reduced to 0.3 µg/L or even lower [79-81]. Indeed, with the generalized use of sensitive assays (chemiluminescence or fluorometric assays with very low detection limits [0.10–0.30 µg/L]), a few patients with clear clinical signs of acromegaly and high IGF-I levels, may demonstrate suppressed GH levels (< 1 µg/L during the OGTT) due to their low GH output [43, 82-85].
Of note, in around one third of patients with acromegaly, GH levels may paradoxically increase in response to oral glucose [84, 86-90]. This phenomenon was initially described by Beck et al.  in acromegaly patients and later reported in several other conditions and disorders (Table 5). There is currently no agreement on the criteria to define a paradoxical GH response pattern. Indeed, some studies regarded an early rise of GH to oral glucose as opposed to a normal suppression as a paradoxical response. Other reports described an increase of more than 20–100% above basal levels (Table 6). It should be noted however that most reports were early studies, involving different study populations with various pathologies [17, 92-94]. The OGTT was not standardized, as 50–100 g of oral glucose was administered and variable GH assays were used. Additionally, data on the reproducibility of the paradoxical GH response to oral glucose are lacking, as it may be influenced by the well-documented spontaneous GH fluctuations in acromegalic patients .
The paradoxical GH response to OGTT may have clinical and prognostic significance in patients with acromegaly. We have recently reported that a paradoxical GH response to oral glucose occurred in patients with GNAS-mutation negative tumors, mainly of the pure somatotroph densely granulated phenotype and harboring high cytogenetic alterations [96, 97].
Very recently, a large study analyzed 496 patients with acromegaly and investigated the association between glucose-induced GH response and endocrine profiles, clinical manifestations, and response to therapy . Patients in the paradoxical response group were older, had smaller and less invasive tumors, and showed a better response to somatostatin analogues. Therefore, a paradoxical rise of GH to oral glucose in acromegaly may reflect some important biological characteristics of pituitary tumors.
As for the underlying mechanisms of the paradoxical GH response, to date they remain incompletely characterized. An altered interplay between somatostatin and GHRH has been suggested  but not well documented. Recently, in analogy to food-dependent Cushing’s syndrome [99-101], a role for ectopic expression of the glucose-dependent insulinotropic polypeptide receptor (GIPR) in somatotroph adenomas for mediating the paradoxical GH response to OGTT has been evoked [88, 102]. We have recently shown that among 41 pituitary adenomas, all 10 samples from patients with paradoxical GH responses displayed ectopic GIPR expression . In a previous study, GIP infusion was shown to increase GH secretion in 2 patients with acromegaly whose GH showed a paradoxical response to OGTT . Furthermore, in GH3 cells transfected with GIPR, GIP stimulation increased cAMP levels and GH transcription . Indeed, acromegalic patients exhibiting paradoxical GH responses to oral glucose load increased their plasma GH levels in response to intravenous GIP stimulation. Furthermore, loss of this paradoxical GH response when glucose was administered intravenously supported the hypothesis of the implication of a gastrointestinal hormone . In addition, GIP stimulation of GIPR-expressing somatotroph adenomas in primary culture increased GH release in 80% of these adenomas, with 60% reaching statistical significance .
OGTT in the Follow-Up of Acromegaly
Disease control in acromegaly is defined by the normalization of IGF-1 levels and GH nadir < 0.4 μg/L after OGTT using ultrasensitive assays . As previously mentioned, GH nadir levels during an OGTT are affected by multiple factors such as biological factors, analytical variations in addition to treatment-specific differences in biochemical responses. These factors result in discordant IGF-1 and GH levels and should be considered when interpreting results of the OGTT test.
Specifically, the utility of a GH nadir during OGTT in monitoring disease activity in somatostatin analogue treated patients has been questioned [43, 105] and is generally not recommended [104, 106]. In these patients, IGF-1 levels do not seem to correlate with GH secretion as opposed to healthy individuals and to patients with acromegaly not receiving somatostatin analogues. Therefore, in this patient population, normalizing IGF-1 has been recommended as the goal of therapy and suggested as the best reflection of disease activity .
In contrast, another group has questioned the validity of IGF-1 as an adequate marker of disease activity in somatostatin analogue-treated patients, as somatostatin analogues may exert a suppressive effect on hepatic IGF1 production resulting in normal IGF-1 levels despite continued disease activity induced by circulating GH [107-109]. The authors reported significantly higher GH nadir levels along with worse symptoms and quality of life in controlled patients treated with somatostatin analogues compared to surgery despite comparable IGF-I and fasting GH levels . Dose escalation of somatostatin analogues minimized the discordance between IGF-1 and nadir GH levels with, however, no improvement in quality of life . Furthermore, higher GH nadir levels after an OGTT as well as after mixed meals were observed in patients treated with somatostatin analogues compared to patients successfully treated with surgery and matched to IGF-1 levels suggesting a residual disease activity in somatostatin analogue treated patients despite normalized IGF-I levels . The authors thus conclude that measuring GH levels during OGTT may unmask insufficient disease control with somatostatin analogues despite normalized IGF-I levels.
Although the impact of glucose on GH secretion has been known since decades, the exact underlying mechanisms are still elusive. Besides the classical suppressive effects on GH release, glucose exhibits stimulatory effects termed “paradoxical” reported in acromegaly and in various other disorders. The paradoxical response of GH to oral glucose in acromegaly may help delineate specific phenotypic tumor characteristics that may influence therapy. Further understanding of the dynamics mediating GH response to glucose may allow the better use of the OGTT as a diagnostic tool in various conditions especially acromegaly. Potential biological factors modifying GH secretions and GH assays should always be taken into account in the interpretation of GH values during OGTT.
The authors declare that they have no conflicts of interest to disclose.