GIP Receptor Antagonism Eliminates Paradoxical Growth Hormone Secretion in Some Patients With Acromegaly

Abstract

Context

About 30% of patients with active acromegaly experience paradoxically increased growth hormone (GH) secretion during the diagnostic oral glucose tolerance test (OGTT). Endogenous glucose-dependent insulinotropic polypeptide (GIP) is implicated in this paradoxical secretion.

Objective

We used the GIP receptor (GIPR) antagonist GIP(3-30)NH₂ to test the hypothesis that GIP mediates this paradoxical response when GIPR is abundantly expressed in somatotropinomas.

Methods

A total of 25 treatment-naive patients with acromegaly were enrolled. Each patient underwent one OGTT during simultaneous placebo infusion and one OGTT during a GIP(3-30)NH₂ infusion. Blood samples were drawn at baseline and regularly after infusions to measure GH. We assessed pituitary adenoma size by magnetic resonance imaging and GIPR expression by immunohistochemistry on resected somatotropinomas. For mechanistic confirmation, we applied in vitro and ex vivo approaches. The main outcome measure was the effect of GIP(3-30)NH₂ on paradoxical GH secretion during OGTT as a measure of GIP involvement.

Results

In 4 of 7 patients with paradoxical GH secretion, GIP(3-30)NH₂ infusion completely abolished the paradoxical response (P = .0003). Somatotrophs were available from 3 of 4 of these patients, all showing abundant GIPR expression. Adenoma size did not differ between patients with and without paradoxical GH secretion.

Conclusion

Of 25 patients with acromegaly, 7 had paradoxical GH secretion during OGTT, and pharmaceutical GIPR blockade abolished this secretion in 4. Corresponding somatotroph adenomas abundantly expressed GIPR, suggesting a therapeutic target in this subpopulation of patients. In vitro and ex vivo analyses confirmed the role of GIP and the effects of the antagonist.

Acromegaly is a rare but debilitating disease caused by hypersecretion of growth hormone (GH), most often from a somatotroph pituitary adenoma. Persistent excess of GH and insulin-like growth factor I (IGF-I) leads to characteristic acral and soft tissue overgrowth as well as cardiovascular disease, metabolic complications, and increased risk for neoplastic and respiratory complications, with excess morbidity and mortality if left untreated or poorly controlled (1-7).

The biochemical diagnosis of acromegaly is based on elevated levels of blood GH that cannot be suppressed adequately in response to an oral glucose tolerance test (OGTT), concomitant with elevated serum IGF-I (8, 9). Most patients with active disease are unresponsive to OGTT, but 30% exhibit a paradoxical increase in GH (10-18). This increase is associated with higher IGF-I levels, smaller and less invasive tumors, and a more favorable response to pharmacotherapy (13-16, 19, 20).

Secretion of glucose-dependent insulinotropic polypeptide (GIP), an incretin hormone secreted from enteroendocrine K cells in the proximal small intestine following food ingestion (21, 22), may be associated with the paradoxical increase in GH secretion (10-12, 15). In 1 analysis involving 2 patients with acromegaly, intravenous GIP infusion reproduced the paradoxical GH secretion (11), and abnormally high GIP receptor (GIPR) expression levels have been found in somatotropinomas of most patients with the paradoxical GH response during OGTT (10, 12, 15). In vitro, GIP increases cyclic adenosine monophosphate (cAMP) levels and GH promotor activity in the GH-secreting cell line GH3 (12). GIP also increases GH secretion in certain somatotropinoma-derived primary cultures (10). Whether endogenous GIP secretion, triggered by oral glucose intake, contributes to the paradoxical GH secretion observed in patients with active acromegaly remains to be determined.

We hypothesized that endogenous GIP mediates the paradoxical response when GIPR is abundantly expressed in somatotropinomas. To address the involvement of endogenous GIP in this response, we conducted a placebo-controlled crossover study in 25 patients with active acromegaly to assess the ability of a high-affinity GIPR antagonist, GIP(3-30)NH₂ (23-25), to reduce paradoxically increased GH levels during OGTT. We also applied immunohistochemical assessment of GIPR expression in resected somatotropinomas from affected patients and used in vitro and ex vivo techniques to confirm GIP involvement in GH secretion and GIPR inhibition by GIP(3-30)NH₂.

Materials and Methods

Ethics

Before study initiation, the scientific ethics committees for the Capital Region of Denmark approved the clinical study protocol (H-17034769) and the ex vivo study protocol (H-18042925). The ex vivo human cell culture protocol was also approved by the Cedars-Sinai Institutional Review Board. The clinical protocol was registered and approved by Copenhagen's joint record of biobanks and record of research projects containing personal data under the Danish Data Protection Agency (case No. 504-0023/16-3000, journal No. SUND-2018-27). The clinical protocol also was registered at ClinicalTrials.gov (registration No. NCT03807076). Before inclusion in the study, all participants gave written informed consent. The study was conducted according to the Declaration of Helsinki.

Study Design

The placebo-controlled crossover clinical portion of the study was conducted at the Department of Endocrinology and Metabolism at Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark, during 2019 to 2023. On the first infusion day, treatment-naive patients with suspected acromegaly received a diagnostic OGTT (75-g glucose in 250-mL water after 10 hours of overnight fasting), with an infusion of placebo (volume-matched saline, 9 mg/mL with human serum albumin [HSA] to a final concentration of 0.5%) begun 20 minutes before initiation of the OGTT and continued for the duration of the test. On the second infusion day, patients underwent a second OGTT, along with infusion of the antagonist GIP(3-30)NH₂ (800 pmol/kg/min) begun 20 minutes before OGTT initiation and continued for the duration of the test. Acromegaly was confirmed with the first OGTT when GH levels failed to drop below 0.4 µg/L after glucose ingestion. This portion of the study was not blinded to ensure that standard-of-care initiation, if needed, would be completed on time.

GIP(3-30)NH₂ administration was well tolerated, and no adverse effects were observed.

As proof of concept that GIP triggers GH release, we established somatotroph-derived cell cultures, and linked the increases in GH to elevated cAMP levels in the GH3 cell line.

Study Objectives and End Points

The primary study objective was to investigate if GIP(3-30)NH₂ could block paradoxical GH secretion during an OGTT, implicating endogenous GIP in this paradoxical secretion in patients with acromegaly. To further establish GIP-related causality, we developed in vitro and ex vivo assays using a rat GH3 cell line transiently transfected with human (h)GIPR and primary cultured cells from human somatotropinomas, respectively.

For the OGTT testing, a paradoxical increase in GH secretion with OGTT was arbitrarily defined as an increase of 30% or more from basal levels. This threshold was established based on the average of published thresholds used to determine a paradoxical response at the time our study was initiated in 2019 (10, 13, 15, 16). The primary end point was the relative baseline-subtracted area under the curve (% bs[AUC]) (accounting for both positive and negative regions) of serum GH from 0 minutes (OGTT initiation) until 180 minutes (OGTT ended). As GH levels differed significantly among patients with acromegaly, GH values were expressed as a percentage of basal levels to allow for grouped analysis. The primary end point, that is, the effect of GIP(3-30)NH₂ on the GH response, was evaluated among all patients with acromegaly, although GIP(3-30)NH₂ was hypothesized to affect only paradoxical GH secretion.

Secondary end points were relative ΔbsAUC_{0-180 minutes} for biomarkers relevant for either GIP physiology and/or acromegaly including serum IGF-I and prolactin and ΔbsAUC_{0-180 minutes} for plasma GIP, glucagon-like peptide (GLP)-1, GLP-2, glucagon, ghrelin, insulin, C-peptide, nonesterified fatty acids (NEFAs), glycerol, cholesterols, triglycerides, and thyrotropin (TSH). Adenoma GIPR expression levels, sizes, and cavernous sinus invasion were also secondary end points.

To assess the role of GIPR in this response, we evaluated the presence of this protein in somatotropinomas and conducted in vitro and ex vivo evaluations of GIP, and the antagonist, GIP(3-30)NH₂ in a transiently transfected rat cell line and in primary cells derived from human somatotropinomas.

Patients and Eligibility Criteria

Adults (aged 18-85 years) under investigation for acromegaly at Danish hospitals were invited to participate in the study, and only patients with confirmed acromegaly following the first OGTT test were included in the final analyses. Exclusion criteria were medical treatment for acromegaly, abnormal liver enzymes (alanine transaminase and aspartate transaminase levels twice the normal ranges), uncontrolled hypertension (>170/100 mm Hg), severe arteriosclerotic heart disease or heart failure (New York Heart Association groups 3 and 4), severe anemia, pregnancy, or any disease or illness that potentially could interfere with the end point analysis. Demographics and anthropometric data, including body mass index, sex, age, height waist-to-hip ratio, and weight were also collected from the patients. Participants were provided with a symbolic honorarium for their involvement in the study.

Our study included men and women. Because of the small numbers in subgroups, sex was not considered in the statistical analyses.

GIP(3-30)NH₂ Synthesis and Preparation

GIP(3-30)NH₂ is a naturally occurring GIPR antagonist used in several human studies (25-34). In this work, GIP(3-30)NH₂ was custom synthesized by Caslo with greater than 98% purity and confirmed as identical to the natural human peptide by high-performance liquid chromatography and mass spectrometry. The synthesized GIP(3-30)NH₂ was dissolved in sodium hydrogen carbonate (10 mM) with 0.5% HSA (CSL Behring) in sterile water, pH 9.25, followed by sterile filtration and endotoxin testing by the Capital Region Pharmacy. GIP(3-30)NH₂ was stored at −20 °C until use. On test days, GIP(3-30)NH₂ was diluted in sodium chloride (Fresenius Kabi) containing 0.5% HSA to a final volume of 500 mL with a suitable concentration for an infusion of 800 pmol/kg/min. For placebo infusion, we used 9 mg/mL sodium chloride with 0.5% HSA.

Sample Collection

Before each OGTT, a cannula was inserted into a cubital vein for blood sampling, and another cannula was inserted into the contralateral cubital vein for infusions. Baseline blood samples were drawn at −60, −45, and −30 minutes before OGTT initiation. At −20 minutes, infusion of placebo or GIP(3-30)NH₂ was initiated, and blood samples were drawn at −15 minutes and 0 minutes, immediately followed by oral glucose consumption (75 g) within 5 minutes. Additional blood samples were drawn at 15, 30, 60, 90, 120, and 180 minutes, when the OGTT ended. Infusions were stopped immediately after the 180-minute blood samples were taken. Study personnel were responsible for handling all blood samples and infusions. Patients remained sedentary in a semi-recumbent position throughout the OGTT.

After the OGTTs, patients diagnosed with acromegaly underwent transsphenoidal surgical resection of the adenoma whenever possible, and these samples were later subjected to immunohistochemical analysis for GIPR expression. Magnetic resonance imaging was conducted in all but one patient prior to resection. Primary cell cultures were developed from somatotropinomas associated with a separate population of patients not included in the infusion portion of the study.

Blood Sample Analysis

Venous blood samples were collected and processed appropriately to separate plasma or serum and stored at −20 °C and −80 °C until analysis. Plasma glucose was measured at the bedside using a glucose analyzer (Contour NEXT, Ascensia Diabetes Care). Heart rate and blood pressure were measured with the patient's arm in a relaxed, nonflexed position, using a standard digital blood pressure monitor (UA-651, A&D Medical). Serum GH and serum IGF-I were analyzed using an IDS-iSYS Multi-Discipline Automated System (Immunodiagnostic Systems) with the automated chemiluminescence immunoassay method. The GH assay range was 0.050 to 100 ng/mL, with an intra-assay variability of 1.1% to 3.4%, and the IGF-I assay range was 10 to 1200 ng/mL, with an intra-assay variability of 1.3% to 3.7%. Blood samples from one patient needed to be diluted to fit these ranges. Plasma samples for total GIP, total GLP-1, and intact GLP-2 were extracted with ethanol before analysis (70%, 70%, and 75%, respectively) and measured in plasma by radioimmunoassay, as previously described (35-37), using C-terminally directed antisera against GIP (code No. 80867, RRID:AB_2892194) and GLP-1 (code No. 89390, RRID:AB_2892195) and N-terminally directed antiserum against GLP-2 (code No. 92160, RRID:AB_2943622). Plasma GIP(3-30)NH₂ was measured as previously described (25) with an in-house–developed radioimmunoassay, similar to the GIP assay, using a polyclonal antibody raised in rabbits against the C-terminal region of human GIP(1-30)NH₂ (code No. 95234, RRID:AB_2892197). Human GIP(1-30)NH₂ was used as the standard (catalog No. 027-30, Phoenix Pharmaceuticals) and ¹²⁵I-labeled human GIP(1-42) as the tracer (catalog No. NEX402, Perkin Elmer). The intra-assay coefficient of variation was less than 10% for all assays.

Plasma glucagon and plasma ghrelin (total) were measured by enzyme-linked immunosorbent assay (ELISA) (glucagon, catalog No. 10-1271-01, RRID:AB_2737304, Mercodia; ghrelin, catalog No. EZGRT-89K, RRID:AB_2892838, Merck KGaA). Plasma insulin, C-peptide, and TSH were measured with electrochemiluminescence assays on a Cobas 8000 (Roche Diagnostics). Plasma triglycerides, total cholesterol, high-density lipoprotein (HDL) cholesterol, and low-density lipoprotein (LDL) cholesterol were measured using photometric assays on a Cobas 8000. Plasma prolactin was measured by an immunofluorescence assay using a B·R·A·H·M·S KRYPTOR Gold Analyzer (824.050, Thermo Scientific), and plasma NEFA and glycerol were measured by enzymatic methods modified to run on a Cobas 6000 automatic analyzer (Roche).

Surgical Sample Analysis

A total of 21 somatotropinomas from patients who underwent the OGTT study were paraffin-embedded immediately after resection and stored in Rigshospitalet's biobank. All samples were handled simultaneously after the last patient included in the study had undergone the last OGTT. Paraffin-embedded somatotropinomas were cut into 5-μm sections, deparaffinized in Histo-Clear, and rehydrated through graded alcohols to water. HEK293 cells transfected with hGIPR and sections from a human pancreas were used as positive controls. HEK293 cells without hGIPR transfection were used as a negative control.

Antigens were retrieved in 10-mM Tris with 0.5-mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetra-acetic acid at pH 9.0 (ie, TEG buffer) by boiling for 15 minutes in a microwave oven. Sections were preincubated for 10 minutes in 2% bovine serum albumin and incubated at 4 °C overnight with a rabbit anti-GIPR polyclonal antibody (1:100; catalog No. PA5-33585, RRID:AB_2550965, Thermo Fisher Scientific) and a mouse anti-GH monoclonal antibody (1:1000; catalog No. SC-374266, RRID:AB_10989917, Santa Cruz Biotechnology). Next, sections for brightfield microscopy were incubated for 40 minutes with biotinylated secondary immunoglobulins (diluted 1:200 for GH; catalog No. BA-2000, RRID:AB_2313581, Vector Laboratories; diluted 1:200 for GIPR; catalog No. BA-1000, RRID:AB_2313606, Vector Laboratories). Then, 3% hydrogen peroxide was added to block endogenous peroxidase, and the third layer was formed by a preformed avidin and biotinylated horseradish peroxidase macromolecular complex (Elite ABC, Code no. PK-6100, Vector Laboratories) for 30 minutes. Finally, the reaction was developed with 3,3-diaminobenzidine (DAB+, SK-4105, Vector Laboratories) for 15 minutes and counterstained with Mayer's hematoxylin.

A fluorescence technique was used for double staining. Each section was incubated with both primary antibodies and then incubated for 1 hour with Alexa Fluor 488–labeled goat antirabbit immunoglobulin G secondary antibody (1:500; catalog No. ab150081, RRID:AB_2734747, Abcam) and Alexa Fluor 568–labeled donkey antimouse immunoglobulin G secondary antibody (1:500; catalog No. ab175700, RRID:AB_3083693, Abcam). All coverslips were mounted with aqueous mounting medium (catalog No. S3023, Agilent).

GIPR expression levels were determined semi-quantitatively, with the paradoxical response status of the samples concealed. The semi-quantification was performed by evaluating the amount and staining intensity of GIPR-expressing cells relative to the overall area of the mounted adenoma section using light microscopy (Leitz Wetzlar). A comparison to GIPR staining in 2 normal pituitaries was integrated into the analysis. If 15% or less cells in an adenoma were evaluated as positive for GIPR expression, the expression level was referred to as “normal.” If 30% or more cells in an adenoma showed GIPR expression, the expression level was characterized as abundant. None of the adenomas had an estimated GIPR expression of 16% to 29%. Colocalization with GH was examined with a fluorescence microscope (Zeiss Axioskop 2 plus) with a 40× objective. All images were acquired using DeltaPix cameras and analyzed with DeltaPix InSight 6.7.5 software.

Radiological Measurements on Imaging

Diagnostic magnetic resonance imaging was performed either immediately before or after the OGTTs and reviewed for adenoma characteristics by a radiologist (R.J.) unaware of the OGTT response. One patient in the nonparadoxical group did not have a magnetic resonance scan performed until months after the initiation of treatment and was therefore excluded from this analysis. Tumors were classified as microadenomas if the maximal diameter was smaller than 10 mm and as macroadenomas if the maximal diameter was larger than 10 mm. Tumor volumes were calculated using the formula: ½ × length × width × height. The invasion of the cavernous sinus was evaluated on coronal sections.

GH3 Cell Line Culturing, Transfection, and Growth Hormone Secretion Assay

To confirm the effect of GIP on GH secretion at the molecular level, we used cultured GH3 rat cells transfected with hGIPR to assess the effects of GIP and GIPR antagonist administration in a GH secretion assay. The GH3 rat cell line (CCL-82.1, lot No. 59257317, ATCC, RRID:CVCL_0273) from female rats was cultured in 10% CO₂ at 37 °C in MEM Eagle m/erles + NaHCO₃ (substrate department, University of Copenhagen) supplemented with 2.5% fetal bovine serum (F2442, Merck), 15% horse serum (16050122, Thermo Fisher Scientific), 1% 2-mM glutamine (substrate department, University of Copenhagen), and 1% 180 units/mL penicillin plus 45 g/mL streptomycin (substrate department, University of Copenhagen). hGIPR complementary DNA was purchased from OriGene (catalog No. SC110906) and cloned into the pCMV-Script vector (Agilent Technologies Denmark). Transient transfections were accomplished using Lipofectamine 2000 (11668027, Invitrogen) according to the manufacturer's instructions. Two days after transfections, the culture medium was removed. Fresh culture medium containing a final concentration of 10 μmol/L of an optimized GIPR antagonist (custom synthesized by Caslo), 10 or 100 nmol/L GIP(1-42) (catalog No. 01-4030658, Bachem), or vehicle (0.5% dimethyl sulfoxide in culture medium) was added in sequence, with the antagonist added 20 minutes before addition of GIP in costimulated wells. After 20 minutes, the culture medium was collected and analyzed for GH content by ELISA (rat/mouse growth hormone ELISA, catalog No. EZRMGH-45K, Merck, RRID:AB_2892711, Millipore). Experiments were conducted in duplicate and repeated in 3 independent experiments.

Cyclic Adenosine Monophosphate Accumulation Assay

To confirm if GIP mediates GH secretion through elevated cAMP levels in GH3 cells, we conducted a cAMP accumulation assay. GH3 cells were seeded into white 96-well plates at a density of 30 000 cells/well. Transient transfections were performed 1 day after seeding, and 2 days after transfections, cells were washed once with HEPES buffered saline (HBS; Gibco, 24040092) and incubated with 0.1% casein, HBS, and 1-mM 3-isobutyl-1-methylxanthine for 30 minutes at 37 °C. GIP was added to the cells in increasing final concentrations of 1 × 10⁻¹² to 1 × 10⁻⁷ M and incubated for 30 minutes at 37 °C. The HitHunter cAMP XS assay (DiscoveRx) was carried out according to the manufacturer's instructions. Experiments were conducted in duplicate and repeated twice. Luminescence was measured using the Perkin Elmer EnVision 2104 Multilabeled Plate Reader with excitation at 320 nm and emission at 665 and 620 nm. The resulting dose-response curve was fitted in GraphPad Prism version 9.5.1 for Windows (GraphPad Software) using nonlinear regression analysis (4-logistic parameter equation) for estimation of potency (EC₅₀).

Primary Somatotroph-derived Cell Cultures

For proof of concept that GIP induces GH release, we established primary cultures from somatotropinomas derived from patients with acromegaly not participating in the clinical OGTT study. A total of 25 freshly resected somatotroph adenomas at Odense University Hospital, Denmark, Aarhus University Hospital Denmark and Cedars-Sinai Medical Center, Los Angeles, California, USA (16 from women and 9 from men) were obtained in DMEM/Ham's F-12 50/50 Mix medium (catalog No. 10-090-CM, Corning) containing 2% Antibiotic:Antimycotic (catalog No. 400-101, Gemini Bio-Products). Adenoma tissue was washed and mechanically separated into single cells by gentle pressure-filtration through a sterile 70-μm cell strainer (catalog No. CLS431751-50EA, Corning, Merck). The mixture was centrifuged at 350 relative centrifugal force for 5 minutes, the supernatant removed, and red blood cell lysis buffer (catalog No. 11814389001, Sigma-Aldrich) added. The mixture then was centrifuged at 350 relative centrifugal force for 3 minutes, the supernatant removed, and the cell pellet resuspended in culture medium (NeuroCult NS-A Basal Medium [human], catalog No. 05750, Stem Cell Technologies) containing 10% NeuroCult NS-A Proliferation Supplement (catalog No. 05753, Stem Cell Technologies) and 2% Antibiotic:Antimycotic. Primary cells were aliquoted into 24-well plates precoated with ECL Cell Attachment Matrix (catalog No. 08-110, Merck) and incubated at 37 °C under 5% CO₂. At 24 hours after seeding, the culture medium was carefully removed. Fresh culture medium containing a final concentration of 10 or 100 nM GIP(1-42) to determine the optimal stimulation concentration) was added (catalog No. 01-4030658, Bachem), or vehicle (0.5% dimethyl sulfoxide in culture medium) was added. After 1 hour, the culture medium was collected and analyzed for GH content by ELISA (hGH ELISA, catalog No. 25-HGHHU-E01, RRID:AB_3101992, ALPCO).

Statistical Analysis and Calculations

Previous clinical data indicated that a dose of 800 pmol/kg/min of GIP(3-30)NH₂ inhibited GIP-induced potentiation of glucose-stimulated insulin secretion by more than 80% (25, 27). Based on these efficacy data, a power calculation with a type II error (β) risk of 20% and a level of statistical significance (α) of 5% suggested that 7 patients with a paradoxical increase in GH secretion to oral glucose should be included in the study to detect a 20% or greater difference in GH secretion. As only approximately 30% of patients with acromegaly have a paradoxical GH secretion, we expected to include approximately 30 patients with acromegaly to account for uncertainties in the actual fraction of patients with this characteristic.

Descriptive data are reported as means ± SDs for normally distributed data and as medians (25th-75th percentiles) for nonparametric data. Between-infusion differences (placebo vs GIP(3-30NH₂)) and between-group (paradoxical vs nonparadoxical) results are presented as means (95% CIs). Visual assessment and the Shapiro-Wilk test were used to evaluate all descriptive data for normal distribution. bsAUC was calculated for 0 to 180 minutes for all biomarkers. The baseline was the mean of the −60-, −45-, and −30-minute values, and the bsAUC considered both incremental and decremental changes from the baseline.

The Matsuda index (a measure of insulin resistance) was calculated using the following formula:

$$\mathrm{Matsuda}\mathrm{index}={\displaystyle \frac{10000}{\sqrt{\mathrm{basal}\mathrm{glucose}(\mathrm{mg}/\mathrm{dL})\times \mathrm{basal}\mathrm{insulin}(\mathrm{mIU}/\mathrm{L}))\times (\mathrm{mean}\mathrm{glucose}(\mathrm{mg}/\mathrm{dL})\times \mathrm{mean}\mathrm{insulin}(\mathrm{mIU}/\mathrm{L})}}}$$

The Pearson correlation method was used for correlation analysis of anthropometrics and biomarkers with basal GH levels, placebo GH bsAUC_{0-180 minutes}, or GH ΔbsAUC_{0-180 minutes}.

Differences in demographic and anthropometric data and results of radiological measurements between the paradoxical and nonparadoxical groups were evaluated by Fisher exact test for categorical variables and unpaired t tests for continuous variables. Differences in basal values between groups and test days and effects of GIP(3-30)NH₂ infusion on peak and bsAUC_{0-180 minutes} were evaluated by matched 2-way repeated-measures analysis of variance (ANOVA) and the Šidák multiple comparison test when relevant. Skewed data were log-transformed before analysis and back-transformed for presentation.

All cell line studies were performed in duplicate in 3 independent experiments. Primary cell studies were performed in duplicate to sextuplicate (depending on the size of the resected adenoma) in 1 independent experiment due to the nature of primary cells, which do not divide and have a limited lifespan and therefore cannot be used for repeated experiments. Data were analyzed with unpaired t tests for primary cell studies and one-way ANOVA with Holm-Šídák's multiple comparisons test for cell line studies.

All statistical tests were 2-sided and considered statistically significant when P was less than .05. Statistical analysis was performed and graphics developed with GraphPad Prism version 9.5.1 for Windows (GraphPad Software).

Results

Patients and Clinical Characteristics

A total of 27 patients undergoing workup for the diagnosis of acromegaly were initially included in the OGTT + infusion part of the study. Two patients were excluded because of GH levels below the diagnostic threshold (0.4 µg/L) following the first OGTT. Thus, 25 patients completed both OGTTs and infusions, the first a placebo infusion on test day 1 and the second a GIP(3-30)NH₂ infusion on test day 2 (Fig. 1A).

Figure 1.

Study design and plasma levels of GIP(3-30)NH₂ in the paradoxical and nonparadoxical groups. A) Blood sample time points are indicated with red drops, and yellow stars indicate blood pressure and heart rate measurements. The duration of infusion is highlighted with a yellow bar. B) Plasma levels of GIP(3-30)NH₂. Black circles, paradoxical group (n = 7); open circles, nonparadoxical group (n = 18). C) Corresponding bsAUC_{0-180 minutes}. Data are presented as mean ± SEM, and bsAUC_{0-180 minutes} were assessed for statistical differences using an unpaired t test. No statistically significant differences were found.

Abbreviations: GIPR, glucose-dependent insulinotropic polypeptide receptor; OGTT, oral glucose tolerance test.

Seven patients (28%) had a paradoxical increase in GH secretion during the first OGTT (referred to as the paradoxical group), predefined as an increase of 30% or greater from basal levels (taken as the mean of GH concentrations at −60, −45, and −30 minutes). For all 7 patients, the rise in GH secretion occurred within the first 90 minutes after glucose consumption (Table 1). No differences in demographics or anthropometrics were found between the paradoxical and nonparadoxical groups (Supplementary Table S1 (38)).

Table 1.

Growth hormone secretion patterns and levels

Patient No.	Sex	Time to GH increase, min	Time to GH peak, min	Basal GH, µg/L		Peak GH, %		bsAUC GH, % × min		Basal glucose, mmol/L		Peak glucose, mmol/L		bsAUC glucose, mmol/L × min
				Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2
Paradoxical group
1	Male	15	60	3.3 ± 0.00	3.0 ± 0.00	233	−34	23 078	−4673	6.4	6.30	11.8	10.9	399	350
4	Female	60	90	7.4 ± 0.00	6.9 ± 0.00	42	−21	3615	−1208	7.4	7.50	13.8	14.3	698	651
6	Male	60	60	28.4 ± 0.00	37.6 ± 0.00	127	84	12 863	7493	6.5	7.10	12.2	10.8	404	279
13	Female	30	60	8.8 ± 0.10	7.8 ± 1.6	363	59	36 968	4455	5.37	5.70	11.2	12.9	564	693
17	Male	60	60	2.7 ± 0.04	2.4 ± 0.33	44	74	1603	1853	5.67	5.77	10.9	12.4	526	700
18	Male	90	90	21.1 ± 1.6	36.5 ± 5.3	110	−54	5070	−7088	6.7	5.77	11.3	11.3	275	418
27	Male	60	60	2.0 ± 0.22	2.4 ± 0.31	50	51	3263	1988	5.83	5.77	10.1	10.3	170	392
Nonparadoxical group
2	Male	—	—	9.7 ± 0.00	7.7 ± 0.00	−23	33	−2475	615	6.3	6.1	9.4	10.2	137	270
3	Male	—	—	10.1 ± 0.00	9.8 ± 0.00	−28	−40	−3248	−5255	5.9	5.7	7.5	10.6	96	293
5	Female	—	—	78.1 ± 0.00	75.5 ± 0.00	−47	−48	−5513	−5565	4.5	4.6	10.6	8.9	730	532
7	Female	—	—	2.2 ± 0.00	2.8 ± 0.00	−19	−55	−1883	−9248	6.6	7.1	12.2	13.3	421	476
8	Male	—	—	11.9 ± 0.00	16.6 ± 0.00	−18	−17	−1388	−1943	7.4	6.9	14.7	16.8	628	1204
9	Male	—	—	145 ± 8.5	126 ± 5.0	−34	−29	−3173	−2355	6.3	6.2	8.6	8.6	161	95
10	Female	—	—	2.2 ± 0.01	1.9 ± 0.08	−55	−53	−6758	−6443	6.2	6.2	11.7	11.9	371	514
11	Female	—	—	34.6 ± 0.90	24.4 ± 2.1	−45	−40	−5280	−6083	6.4	6.6	10.8	11.1	422	419
12	Female	—	—	2.0 ± 0.62	1.7 ± 0.21	−42	−38	−3128	−4568	5.0	5.2	11.0	8.4	516	207
14	Male	—	—	9.2 ± 1.2	6.7 ± 1.4	−20	25	−2363	2160	6.7	6.7	9.7	10.4	338	390
16	Male	—	—	11.5 ± 0.30	9.8 ± 0.30	23	17	908	308	6.2	6.3	12.1	15.9	313	616
19	Female	—	—	4.2 ± 1.8	5.4 ± 0.79	−83	−85	−8483	−9248	5.2	5.3	9.4	10.4	166	550
21	Male	—	—	14.6 ± 0.99	13.9 ± 0.60	−18	−8	−368	−548	5.7	5.7	9.2	9.2	99	201
22	Male	—	—	4.3 ± 0.41	4.4 ± 0.13	−25	−24	−2513	−263	5.9	6.2	10.3	9.5	337	239
23	Male	—	—	1.5 ± 0.10	1.5 ± 0.07	−41	−31	−4223	−2423	6.5	6.0	10.8	11.4	291	473
24	Female	—	—	11.9 ± 0.66	16.3 ± 0.22	−31	−35	−2318	−2745	6.7	6.7	10.9	9.9	236	56
25	Female	—	—	2.5 ± 0.11	2.4 ± 0.05	−32	10	−2925	263	6.7	6.3	12.4	12.7	374	466
26	Male	—	—	3.3 ± 0.05	3.6 ± 0.08	−33	−19	−3675	−1958	6.3	6.5	11.1	10.5	134	199

Individual GH and glucose measurements for all patients (paradoxical group, n = 7; nonparadoxical group, n = 18). Time to increase (min) indicates the first time GH levels increased above basal levels during placebo OGTT. Time to peak is when the most significant relative (%) increase from basal levels was observed during placebo OGTT. Responders to GIP(3-30)NH₂ infusion (n = 4) are highlighted in bold. Data are presented as mean ± SD for basal values and absolute numbers (%) used for the remaining variables.

Abbreviations: bsAUC, baseline-subtracted area under the curve; GH, growth hormone; GIP, glucose-dependent insulinotropic polypeptide; OGTT, oral glucose tolerance test.

None of the patients were treated with antidiabetics, dual GLP-1/GIP receptor agonists, or steroids, all of which can potentially affect circulating GIP levels. Two patients in the nonparadoxical group had ulcerative colitis and were treated with mesalamine (Asacol, which has a suggested anti-inflammatory effect on colonic epithelial cells); one patient in the paradoxical group previously had surgery for appendicitis, and one in the nonparadoxical group previously had surgery for intestinal obstruction. No other bowel operations or intestinal diseases were reported for any of the patients.

GIP(3-30)NH₂ Infusion

Basal GIP(3-30)NH₂ levels and levels during saline infusion were below the detection limit. Steady-state levels of GIP(3-30)NH₂ during infusion were reached after 20 minutes, corresponding to the time point of 0 minutes, when the OGTT was initiated. Mean GIP(3-30)NH₂ concentrations from time point 0 minutes until 180 minutes were 85 ± 4.9 nmol/L in the paradoxical group and 85 ± 4.2 nmol/L in the nonparadoxical group (Fig. 1B), and the two groups also had similar AUCs (paradoxical: 17 787 ± 3555; nonparadoxical: 16 681 ± 5001 nmol/L × min; P = .60) (Fig. 1C). No patients reported discomfort related to the infusion, and neither GIP(3-30)NH₂ infusion nor paradoxical status was associated with changes in heart rate or blood pressure (Supplementary Table S2 (38)).

Paradoxical Growth Hormone Increase During Oral Glucose Tolerance Test Blocked by GIP(3-30)NH₂

Basal GH levels were similar for all patients (Table 2). GH levels peaked at 138% ± 120% from basal values within 90 minutes during placebo OGTT in the paradoxical group (Figs. 2A and 3A). In contrast, the nonparadoxical group had a drop in GH levels (−32% ± 21%) during the placebo OGTT (Figs. 2C and 3A). The time to rise, peak, and GH amplitude varied among the 7 patients in the paradoxical group (see Fig. 2A and Supplementary Table S2 (38)), and the largest peak GH levels were not associated with the largest peaks in plasma glucose levels (see Table 1). Four patients in the paradoxical group responded to the GIP(3-30)NH₂ infusion with a significantly lower GH secretion during the second OGTT, implying a GIP-mediated response (referred to here as responders; assessed by 2-way ANOVA, statistics not shown) (see Table 1). The remaining 3 patients in the paradoxical group had a minor and nonsignificant change in GH levels during the second OGTT with GIP(3-30)NH₂ infusion, implying a non–GIP-mediated paradoxical response (designated here as nonresponders) (see Fig. 2B and Table 1). As the study was not powered to subdivide the paradoxical group into responders and nonresponders, we did not statistically analyze these data.

Table 2.

Growth hormone and prolactin levels stratified by paradoxical status

	Paradoxical (n = 7)								Nonparadoxical (n = 18)				Paradoxical vs nonparadoxical
					Responders (n = 4)		Nonresponders (n = 3)
	Placebo	GIP(3-30)NH2	Diff.	P	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Placebo	GIP(3-30)NH2	Diff.	P	Diff.	P
GH
Basal, µg/L	7.4 (2.7 to 21.1)	6.9 (2.4 to 36.5)	8.6 (−11 to 25)a	.49	8.1 (4.3 to 18)	7.4 (4.0 to 29)	2.7 (2.0 to 28)	2.4 (2.4 to 38)	9.4 (2.5 to 13)	7.2 (2.7 to 16)	−2.7 (−16 to 9.2)a	.85	15 (−199 to 76)a	.94
Peak, %	138 ± 120	29 ± 54	−109 (−174 to −45)	.001	187 ± 142	−1.3 ± 52	74 ± 46	70 ± 17	−32 ± 21	−24 ± 31	7.4 (−33 to 48)	.89	−170 (−224 to −116)	<.0001
bsAUC, % × min	12 351 ± 13 221	403 ± 5095	−11 949 (−18 323 to −5574)	.0003	17 183 ± 15 885	−2129 ± 5009	5910 ± 6079	3778 ± 3218	−3267 ± 2222	−3072 ± 3354	195 (−3780 to 4170)	.99	−15 618 (−21 459 to −9778)	<.0001
Prolactin
Basal, mIU/L	149 (135 to 157)	152 (141 to 175)	11 (0.10 to 24)a	.048	156 (115 to 159)	129 (155 to 211)	147 (135 to 149)	152 (141 to 175)	327 (163 to 679)	282 (170 to 604)	2.4 (−4.2 to 9.4)a	.64	132 (7.3 to 401)a	.03
Peak, %	13 ± 27	−12 ± 22	−25 (−55 to 5.3)	.12	17 ± 26	−13 ± 27	7.7 ± 34	−10 ± 16	4.0 ± 34	−8.0 ± 21	−12 (−31 to 6.7)	.26	−9.1 (−24 to 32)	.7
bsAUC, % × min	800 ± 2179	−1553 ± 2000	−2353 (−4089 to −617)	.007	979 ± 2491	−2054 ± 2477	562 ± 2186	−885 ± 1277	−247 ± 2334	−961 ± 2048	−715 (−1797 to 368)	.24	−1047 (−3280 to 1186)	.49

Peak levels denote the most noticeable change (increase or decrease) from basal levels, and bsAUC_{0-180 minutes} represents the sum of incremental and decremental changes from basal levels during the period 0 to 180 minutes. Descriptive statistics are presented as mean ± SD, or median (25th-75th percentiles) for data that did not follow a normal distribution. Differences between placebo and GIP(3-30)NH₂ infusions and between the paradoxical and nonparadoxical groups are presented as means (95% CIs). To assess statistical significance between infusions in the paradoxical and nonparadoxical groups and between paradoxical-placebo (n = 7) and nonparadoxical-placebo (n = 18), a matched 2-way repeated-measures analysis of variance with Šidák multiple comparisons test was applied. Log transformation was applied to nonparametric data. Statistical significance is denoted by P less than .05, and significant results are highlighted in bold.

Abbreviations: bsAUC, baseline-subtracted area under the curve; diff., difference; GH, growth hormone; GIP, glucose-dependent insulinotropic polypeptide.

^a Back-transformed data reported as a relative change from placebo (%).

Figure 2.

GH secretion patterns during OGTTs. A) and B) Paradoxical group (n = 7). Solid lines highlight responders (n = 4) and dashed lines nonresponders (n = 3). C) and D) Nonparadoxical group (n = 18). Data are presented as relative (%) changes in GH secretion from basal levels. Each curve represents a corresponding patient, and identical colors reflect data from the same patient during OGTT with placebo and GIP(3-30)NH₂ infusion, respectively.

Abbreviations: Responder, paradoxical GH secretion is abolished by GIP(3-30)NH₂; nonresponder, paradoxical GH secretion not abolished by GIP(3-30)NH₂; GH, growth hormone; GIP, glucose-dependent insulinotropic polypeptide; OGTT, oral glucose tolerance test.

Figure 3.

GH and prolactin responses during OGTT with corresponding bsAUC_{0-180 minutes}. A) Serum GH levels presented as a relative (%) change from basal concentrations (mean of time points −60, −45, and −30 minutes; n = 7 in paradoxical group and n = 18 in nonparadoxical group). B) Serum GH levels (%) in responders (n = 4) and nonresponders (n = 3) of the paradoxical group. C) Serum prolactin levels presented as a relative (%) change from basal concentrations and D) serum prolactin levels (%) in responders (n = 4) and nonresponders (n = 3) of the paradoxical group. Data are presented as mean ± SEM. To assess statistical significance between bsAUC_{0-180 minutes} differences in the paradoxical and nonparadoxical groups and between the paradoxical-placebo and nonparadoxical-placebo groups, a matched 2-way repeated-measures ANOVA with Šidák multiple comparisons test was applied. To assess statistical significance between bsAUC_{0-180 minutes} differences in paradoxical-placebo vs nonparadoxical-GIP(3-30)NH₂ infusion and paradoxical-GIP(3-30)NH₂ infusion vs nonparadoxical-placebo infusion, a 2-way repeated-measures ANOVA with Tukey multiple comparisons test was applied. *P less than .05; ***P less than .001; ****P less than .0001.

Abbreviations: ANOVA, analysis of variance; responder, paradoxical GH secretion is abolished by GIP(3-30)NH₂; nonresponder, paradoxical GH secretion not abolished by GIP(3-30)NH₂; GH, growth hormone; GIP, glucose-dependent insulinotropic polypeptide; OGTT, oral glucose tolerance test.

In the paradoxical group, the GH response during OGTT (time 0-180 minutes) was almost eliminated by GIP(3-30)NH₂ infusion (AUC 403% ± 5095% × min with GIP(3-30)NH₂ infusion vs 12 351% ± 13 221% × min with placebo infusion; P = .0003), but the GH response was unaltered in the nonparadoxical group (see Table 2 and Fig. 3A). Among the 7 patients in the paradoxical group, GIP(3-30)NH₂ infusion eliminated the paradoxical GH response in the 4 responders (see Table 2 and Fig. 3B). Of note, among the responders, GH levels had already declined by 24% during the GIP(3-30)NH₂ infusion at −15 minutes (ie, 5 minutes after initiation of the GIP(3-30)NH₂ infusion and 15 minutes before the OGTT began), indicating an OGTT-independent effect of GIP(3-30)NH₂ on GH levels (see Fig. 3B).

Prolactin Changes With GIP(3-30)NH₂ Infusion in the Paradoxical Group and Other Blood Biomarkers Are Unaffected by Paradoxical Status or GIP(3-30)NH₂ Infusion

Median (25th-75th percentile) basal prolactin levels were higher in the nonparadoxical group compared to the paradoxical group (327 [163-679] vs 149 [135-157] mIU/L; P = .03). In the paradoxical group, prolactin levels increased nonsignificantly to a mean peak of 13% ± 27% above basal levels during placebo infusion/the first OGTT. The maximal mean prolactin increase in the nonparadoxical group was 4% ± 34% during placebo OGTT. With GIP(3-30)NH₂ infusion, prolactin secretion fell below basal concentrations in the paradoxical group (AUC: −1553% ± 2000% × min (GIP(3-30)NH₂ infusion) vs 800% ± 2179% × min with placebo infusion; P = .007) (see Table 2 and Fig. 3C).

The 4 responders to the GIPR antagonist infusion had a more prominent peak in prolactin levels during placebo infusion (17% ± 26%) than nonresponders (7.7% ± 34%; no statistical comparisons conducted). With GIP(3-30)NH₂ infusion, prolactin levels decreased both in responders and nonresponders, and the effect was numerically twice as large in responders (see Table 2 and Fig. 3D). Individual prolactin results revealed that 3 responders and 1 nonresponder had increased prolactin secretion during the OGTT, all reduced by GIP(3-30)NH₂ (data not shown).

The groups did not differ in other biomarkers, including glucose, glucagon, insulin, C-peptide, Matsuda index (Supplementary Table S4, Supplementary Fig. S1 (38)), GIP, GLP-1, IGF-I, ghrelin, TSH, GLP-2 (Supplementary Table S5, Supplementary Fig. S2 (38)), NEFAs, glycerol, cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides (Supplementary Table S6, Supplementary Fig. S3 (38)) besides from insulin and C-peptide that decreased with GIP(3-30)NH2 infusion in the nonparadoxical group (see Supplementary Table S4 and Supplementary Fig. S1 (38)), Biomarkers were also similar between responders and nonresponders within the paradoxical group (Supplementary Tables S7 and S8, Supplementary Figs. S4 and S5 (38)).Of note, mean age- and sex-adjusted basal IGF-I levels were 171% ± 62% above the upper normal limit in the paradoxical group and 143% ± 98% above this limit in the nonparadoxical group.

Given the relevance of the GIP system in the present work, the lack of differences in GIP levels between the paradoxical and nonparadoxical groups is of particular interest.

Abundant Somatotroph Glucose-dependent Insulinotropic Polypeptide Receptor Expression in Responders

Of the 25 patients receiving GIP(3-30)NH₂ infusions, 21 underwent transsphenoidal surgical resection of the adenoma (3 responders, 2 nonresponders, and 16 in the nonparadoxical group). Adenomas were evaluated for somatotrophic GIPR expression using a semi-quantitative method with brightfield and fluorescence microscopy. Samples from the 3 responders abundantly expressed GIPR. The 2 nonresponders had normal GIPR expression levels, 5 adenomas from the nonparadoxical group also had abundant GIPR expression levels, and the remaining 11 adenomas from the nonparadoxical group had normal GIPR expression levels (Supplementary Table S3 (38)). With immunohistochemistry, we confirmed a high degree of overlap between GIPR- and GH-expressing cells (Fig. 4); however, GIPR-positive overlapping with GH-negative cells was also apparent.

Figure 4.

Representative images of adenomas with normal and abundant GIPR expression. Normal GIPR (glucose-dependent insulinotropic polypeptide receptor) expression visualized by immunohistochemistry in a nonparadoxical patient by A) brightfield microscopy, and C) and D) fluorescence microscopy. Abundant GIPR expression visualized in a responder by E, brightfield microscopy, and G and H, fluorescence microscopy. B) and F) Growth hormone (GH) visualization in red; C) and G) GIPR visualization in green; and D) and H) merged red and green staining showing GIPR colocalization with GH in the somatotrophs. All images were acquired with 40× magnification. Scale bars (white), 20 μm.

Imaging Characteristics

Imaging for 24 of the 25 included patients was evaluated. One patient (from the nonparadoxical group) did not have a scan performed until months after the initiation of treatment and was therefore excluded from this analysis. No statistically significant imaging differences were evident between the paradoxical and nonparadoxical groups in terms of microadenoma or macroadenoma prevalence, tumor volume, maximal diameter, or cavernous sinus invasion (Table 3). In the paradoxical group, a nonsignificant trend toward larger adenomas in responders was evident based on the prevalence of macroadenomas (responders, 100%; nonresponders, 33%), median tumor volume (responders, 2.7 cm³; nonresponders, 0.22 cm³), and diameter (responders, 19 mm; nonresponders, 12 mm) (see Table 3).

Table 3.

Radiological imaging characteristics stratified by paradoxical status

Variable	Paradoxical (n = 7)	Nonparadoxical (n = 17)a	P	Responders (n = 4)	Nonresponders (n = 3)
Microadenoma/macroadenoma, n (%)	2/5 (29/71)	3/14 (18/82)	.61	0/4 (0/100)	2/1 (67/33)
Median tumor volume, cm3	1.3 (0.22-4.0)	1.5 (0.38-2.9)	.95	2.7 (0.73-4.6)	0.22 (0.14-2.9)
Mean maximal tumor diameter, mm	16 ± 7.3	17 ± 7.0	.86	19 ± 7.1	12 ± 6.7
Cavernous sinus invasion, yes/no, n (%)	1/6 (14/86)	7/10 (41/59)	.35	1/3 (25/75)	0/3 (0/100)

Normally distributed data presented as mean ± SD, nonnormally distributed data as median (25th-75th percentiles), and categorical data as n (%). P values reflect differences between the paradoxical and nonparadoxical groups, analyzed by the Fisher exact test for categorical data, unpaired Mann-Whitney t test for tumor volume, and unpaired t test for maximal diameter.

^a One patient in the nonparadoxical group did not have a magnetic resonance imaging scan performed until months after the initiation of treatment and was excluded from this analysis.

Correlation Analysis

To evaluate if any of the anthropometric data (age, sex, height, weight, body mass index, waist-hip ratio) or biomarkers (GH, prolactin, glucose, glucagon, insulin, C-peptide, GIP, GLP-1, IGF-I, ghrelin, TSH, GLP-2, NEFAs, glycerol, cholesterol, LDL cholesterol, HDL cholesterol, triglycerides) correlated with basal GH levels, placebo GH bsAUC_{0-180 minutes}, or GH ΔbsAUC_{0-180 minutes}, we calculated a correlation matrix using the Pearson correlation coefficient (r). Basal GH levels correlated significantly with basal IGF-I levels in the nonparadoxical group (Pearson r = 0.63; P = .005), but we identified no other statistically significant correlations with GH.

Increased Growth Hormone Secretion Under Glucose-dependent Insulinotropic Polypeptide Stimulation In Vitro and Ex Vivo

We used a GH3 rat tumor cell line and cells derived from 25 resected somatotropinomas (16 from women and 9 from men) to assess the role of GIP in GH secretion in vitro and ex vivo, respectively. As noted, the human somatotropinoma cells were derived from patients with acromegaly who did not participate in the clinical GIP(3-30)NH₂ infusion portion of this work.

In the ex vivo study, of 25 somatotropinomas used as source material, 5 (20%) were derived from patients with confirmed paradoxical GH secretion (4 women and 1 man). GIP stimulation increased GH secretion from 4 (16%) among the 25 studied primary cultures (Fig. 5A); however, only 2 of these 4 cultures were derived from patients with a confirmed paradoxical GH secretion to OGTT, and OGTT results for the other 2 were unavailable and their paradoxical status unknown. The remaining 3 cultures derived from patients with confirmed paradoxical GH secretion to OGTT did not significantly respond to GIP stimulation (samples 2, 8, and 25 in Fig. 5A). GIP stimulation also led to a nonsignificant increase in GH secretion from 3 other cultures (samples 4, 11, and 25 in Fig. 5A), with a substantial variation among replicates and vehicle-treated samples that likely prevented these results from reaching statistical significance. The limited quantity of resected somatotropinomas, necessitating the use of all available material for the stimulation study, prevented further correlation with GIPR expression levels.

Figure 5.

Effect of GIP on GH secretion ex vivo and in vitro and on cAMP accumulation in vitro. A) GH secretion from somatotropinoma-derived primary cultures expressed as relative secretion from vehicle. Light blue bars, GIP 10 nM stimulation; dark blue bars, GIP 100 nM stimulation; black arrows, confirmed paradoxical GH secretion to OGTT. B) GIP dose response of cAMP accumulation in hGIPR transiently transfected GH3 cells. Results are expressed as relative cAMP accumulation compared to the maximal GIP-induced cAMP response. Solid line, closed circles: hGIPR vector; dashed line, open squares: empty pCMV vector. C) GIP and GIP + GIPR antagonist-induced GH secretion from hGIPR transiently transfected GH3 cells expressed as relative secretion with vehicle. Data are presented as mean ± SEM. One-way analysis of variance with Holm-Šídák multiple comparison test was used to assess statistical significance in the GH3 cell line, and an unpaired t test to assess statistical significance in primary cultures. *P less than .05 increase from vehicle. ^ΨP less than .05 decrease from GIP-stimulated cells.

Abbreviations: cAMP, cyclic adenosine monophosphate; GH, growth hormone; GIP, glucose-dependent insulinotropic polypeptide; (h)GIPR, (human) glucose-dependent insulinotropic polypeptide receptor.

In the in vitro study, GIP increased intracellular cAMP levels in GH3 cells transiently transfected with hGIPR with a potency (EC₅₀) of 300 pmol/L (log EC₅₀ −9.5 ± 0.41) (Fig. 5B). At 10 nmol/L GIP, GH secretion from GH3 cells significantly increased by 43% compared to the response to vehicle (P = .03). The GIP-mediated GH secretion was reduced to 8% when cells were preincubated with a synthetically optimized peptide-based analogue of GIP(3-30)NH₂ (10 μmol/L) before GIP stimulation (P = .03) (Fig. 5C).

Discussion

A subgroup of patients with acromegaly is characterized by a paradoxical GH increase during diagnostic OGTT. Here, in a set of clinical, in vitro, and ex vivo studies, we tested our hypothesis that endogenous GIP mediates this paradoxical response when the somatotropinomas show abundant expression of GIPR. Exogenous GIP may drive GH production in this patient subgroup (11), and paradoxical GH responsiveness has been positively associated with somatotropinoma GIPR expression (10, 12, 15). We show that endogenous GIP mediated the entire paradoxical GH secretory profile in 4 of 7 patients, as determined by the response to an infusion of the specific GIPR antagonist GIP(3-30)NH₂. These responders to GIP(3-30)NH₂ infusion all displayed abundant somatotroph GIPR expression and, in most cases, prolactin cosecretion. This study thus demonstrates that in approximately half of patients with acromegaly who have a paradoxical GH response to OGTT, endogenous GIP is responsible when there is abundant somatotroph GIPR expression.

The GIPR antagonist did not reduce GH concentrations in all patients with OGTT-induced paradoxical GH responsiveness. This finding is in accordance with previous reports of patients whose somatotroph GIPR expression was not congruent with their paradoxical status (10, 12, 15). Paradoxical GH responsiveness after OGTT is also seen in patients with anorexia nervosa, chronic liver disease, impaired glucose tolerance, diabetes, protein malnutrition, and renal failure (39). In patients with acromegaly, thyrotropin-releasing hormone, luteinizing hormone–releasing hormone, corticotropin-releasing hormone, and gonadotropin-releasing hormone injection all may sometimes induce GH (11, 16, 40, 41), suggesting factors other than GIP as potential explanations for some paradoxical cases.

Here, we observed abundant somatotroph GIPR expression in patients who showed a GIP-mediated paradoxical response (ie, responders to antagonist infusion) and normal somatotroph GIPR expression in those with a paradoxical response but no response to antagonist infusion (nonresponders). Abundant somatotroph GIPR expression was also present in 5 patients in the nonparadoxical group, as previously reported in a small subset of patients with acromegaly without paradoxical GH secretion (10, 12, 15). Whether these adenomas express dysfunctional GIPR variants or low GIPR coupling to cAMP signaling remains to be determined.

Paradoxical GH responsiveness correlates with less hyperprolactinemia and plurihormonal secretion (13, 16), and abundant GIPR expression has been associated with pure somatotropinomas (12, 15). In keeping with previous findings, we confirmed higher basal prolactin levels in the nonparadoxical group. In the paradoxical group, prolactin secretion declined with GIP(3-30)NH₂ infusion, whereas insulin and C-peptide concentrations declined with antagonist infusion in the nonparadoxical group. No other biomarkers were influenced by paradoxical status or GIPR antagonist infusion. In contrast, previous studies have shown increased IGF-I (13, 15, 16), glucose, and insulin levels in patients with paradoxical GH secretion (16). The reason for this difference between studies is not apparent. From a clinical perspective, there is currently no simple way to identify GIP-mediated paradoxical GH secretion. The utility of measuring prolactin during a diagnostic OGTT or of a detailed analysis of GH patterns before and after a glucose load is not clear.

Notably, basal GIP levels and GIP responses to OGTT did not differ significantly between the paradoxical and nonparadoxical groups or between the GIP-mediated (responders) and non–GIP-mediated (nonresponders) cases. Thus, somatotroph GIPR expression, not GIP hormone levels, seems to determine GIP-mediated paradoxical GH secretion. As no other ligands than GIP are known to bind the GIPR at physiologically relevant concentrations, it is unlikely that alternative ligands could mediate the paradoxical GH secretion through the GIPR. However, we cannot rule out that such a ligand exists, yet to be disclosed as an alternative endogenous agonist for the GIPR.

Differences in adenoma size showed some promise in distinguishing GIP-mediated from non–GIP-mediated responses in the paradoxical group. In previous studies, smaller and less invasive adenomas have been found in patients with paradoxical compared to nonparadoxical GH secretion (13, 14). Although we observed no statistically significant changes in adenoma size or volume between the paradoxical and nonparadoxical groups in the present study, we did find tendencies between the responders (GIP-mediated) and nonresponders (non–GIP-mediated) in the paradoxical group. Larger sample sizes are needed to substantiate these findings.

The complete blockade of paradoxical GH secretion by GIP(3-30)NH₂ in responders may suggest a therapeutic potential of GIPR antagonism for this specific patient group. Patients with paradoxical GH responsiveness are older at diagnosis, with higher IGF-I levels, and they tend to harbor smaller adenomas with a more favorable response to somatostatin receptor ligand (SRL) treatment (13-16, 19). Whether a different treatment strategy would benefit those with paradoxical GH secretion is unclear, but blocking this secretion can reasonably be hypothesized to offer a benefit by lowering IGF-I levels. Moreover, we detected an onset of basal GH suppression by GIP(3-30)NH₂ in responders before the OGTT began (see Fig. 3B). Of note, SRL treatment also seems to suppress paradoxical GH secretion (42), which may be induced by suppressing GIP secretion from intestinal K cells (43, 44). In keeping with this effect, circulating GIP levels are lower in SRL-treated patients with acromegaly (45). A combination of treatments using a GIPR antagonist, which has no known side effects, and SRLs could potentially allow for dose reductions and in turn reduce SRL-associated side effects.

Supporting the clinical results and in agreement with previous studies, we confirmed that GIP stimulation induced GH secretion in GIPR-transfected GH3 cells in vitro and in primary-derived somatotroph cultures ex vivo (10, 12). Our finding that GIPR antagonism blocks GIP-induced GH secretion in GH3 cells supports exploration of GIPR antagonism as a potential treatment option in acromegaly.

In summary, using a GIPR antagonist, we demonstrate that endogenous GIP mediates the paradoxical GH response to OGTT in patients with acromegaly who have abundant somatotroph GIPR expression. We showed, in vivo and in vitro, that GIP-mediated GH secretion is blocked by GIPR antagonism. As different mediators can cause paradoxical GH secretion, subdivision of this patient population into GIP-mediated (antagonist responders) and non–GIP-mediated (antagonist nonresponders) groups could identify those who might benefit from GIPR antagonist treatment, either as a stand-alone option or as an adjunct to SRL treatment.

Abbreviations

ANOVA

analysis of variance

bsAUC

baseline-subtracted area under the curve

cAMP

cyclic adenosine monophosphate

ELISA

enzyme-linked immunosorbent assay

GH

growth hormone

GIP

glucose-dependent insulinotropic polypeptide

GIPR

GIP receptor

GLP

glucagon-like peptide

HDL

high-density lipoprotein

HSA

human serum albumin

IGF-I

insulin-like growth factor I

LDL

low-density lipoprotein

NEFAs

nonesterified fatty acids

OGTT

oral glucose tolerance test

SRL

somatostatin receptor ligand

TSH

thyrotropin

Funding

This work was supported by Innovation Fund Denmark, Antag Therapeutics Aps, Augustinus Fonden, and Læge Sofus Carl Emil Friis og hustru Olga Doris Friis’ Legat. U.F.R.'s research salary was supported by a grant from The Kirsten and Freddy Johansen's Fund.

Author Contributions

M.H.J., L.S.G., F.K.K., A.K.R., M.C.K., U.F.R., A.H.S.U., and M.M.R. designed the clinical research study and wrote the clinical study protocol. M.H.J., C.Z., S.M., A.H.S.U., and M.M.R. designed the in vitro and ex vivo research studies and wrote the in vitro and ex vivo approved study protocols. M.H.J., L.S.G., K.S.J., C.B.C., S.B., and B.Ho. performed the clinical study. M.H.J. performed in vitro experiments. M.H.J., C.Z., and S.J.S. performed ex vivo experiments. M.H.J., K.S.J., F.D., J.J.H., and B.Har. measured biomarkers. R.J. performed the radiological imaging data acquisition. M.H.J., J.C.B.J., and S.S.P. performed immunohistochemistry data acquisition. A.K.R., M.C.K., and U.F.R. recruited patients for the clinical study. M.S.A., J.O.L.J., S.M., and M.C.A.S. recruited patients for the ex vivo study and acquired corresponding OGTT data. F.R.P., C.B., M.A., and B.Hal. collected the adenoma specimens. M.H.J., J.J.H., A.H.S.U., and M.M.R. were responsible for the funding acquisition. M.H.J. conducted the data and statistical analysis and wrote the original manuscript. All authors revised and approved the final manuscript.

Disclosures

GIP(3-30)NH₂ as a therapeutic agent is protected by intellectual property rights owned by the University of Copenhagen (PCT/DK2015/050266). M.H.J. and A.H.S.U. are employed by Antag Therapeutics Aps. L.S.G., J.J.H., A.H.S.U., and M.M.R. are inventors of the following patent: WO 2018/220123; A.H.S.U. and M.M.R. are inventors of the following patents: WO 2020/115048 and WO 2020/115049; and S.J.S., A.H.S.U., and M.M.R. are inventors of the following patent: WO 2021/110845. M.H.J., L.S.G., S.J.S., F.K.K., J.J.H., M.M.R., and A.H.S.U. hold warrants in Antag Therapeutics Aps. S.J.S. holds warrants in ResoTher Pharma. M.M.R. consults for Antag Therapeutics Aps, consults for and is a member of the board for Synklino, and is a shareholder and chair of the board of Bainan Biotech Aps. B.Har. is a shareholder of Bainan Biotech Aps. J.J.H. is a board member of Antag Therapeutics Aps and has served as a consultant or advisor for Novo Nordisk, Roche, Novartis Pharmaceuticals, and Merck Sharp & Dohme and received fees for lectures from Merck Sharp & Dohme and Novo Nordisk. F.K.K. has served on scientific advisory panels, on speakers’ bureaus, and as a consultant for and/or received research support from 89bio, Amgen, AstraZeneca, Boehringer Ingelheim, Carmot Therapeutics, Eli Lilly and Company, Gubra, Bristol Myers Squibb, Gilead, MedImmune, Merck Sharp & Dohme/Merck, Norgine, Novo Nordisk, Sanofi, SNIPR Biome, Structure Therapeutics, Zealand Pharma, and Zucara, and has been an employee of Novo Nordisk since December 1, 2023. J.O.L.J. serves on a Novo Nordisk advisory board and receives unrestricted research grants from Novo Nordisk and Pfizer. J.O.L.J. is an editorial board member and an editor of The Journal of Clinical Endocrinology & Metabolism and played no role in the journal's evaluation of the manuscript. The remaining authors have nothing to declare.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.

Clinical Trial Information

Clinical trial registration number NCT03807076 (registered January 1, 2019).