Glucose Ingestion Selectively Amplifies ACTH and Cortisol Secretory-Burst Mass and Enhances Their Joint Synchrony in Healthy Men

Abstract

Context:

Glucose intake is associated with a variable increase in adrenal glucocorticoid secretion.

Hypothesis:

Glucose ingestion elevates cortisol secretion by 1) augmenting pulsatile ACTH release; and/or 2) enhancing ACTH-cortisol synchrony or dose-responsiveness.

Subjects:

Fifty-eight healthy men ages 19–78 yr with computed tomography-estimated abdominal visceral fat participated in the study.

Location:

The study was conducted at the Clinical Translational-Research Center and Veterans Affairs Medical Center.

Methods:

We conducted frequent sampling of plasma ACTH and cortisol concentrations after glucose vs. water ingestion in the fasting state, as well as deconvolution, approximate entropy, linear-regression, and dose-response analysis.

Outcomes:

After water ingestion, age was a negative correlate of the mass of ACTH (P = 0.009; R² = 0.119) and of cortisol (P < 0.001; R² = 0.269) secreted per burst. Glucose ingestion abolished both relationships but amplified pulsatile ACTH (P = 0.009) and cortisol (P = 0.001) secretion. Glucose exposure selectively augmented the mass of ACTH (P < 0.001) and of cortisol (P = 0.004) secreted per burst without altering burst number or basal secretion. The increment in pulsatile ACTH strongly predicted the increment in pulsatile cortisol (P < 10⁻⁴; R² = 0.325) secretion. Abdominal visceral fat positively forecast the glucose-induced increment in cortisol secretory-burst mass (P = 0.019). According to approximate entropy analysis, glucose input also enhanced the joint synchrony of ACTH-cortisol secretory patterns (P ≤ 0.001). Caloric intake did not affect analytical dose-response estimates of ACTH potency and efficacy or adrenal sensitivity.

Conclusion:

Conjoint augmentation of the mass of ACTH and cortisol secreted per burst and enhancement of ACTH-cortisol synchrony underlie glucose-induced glucocorticoid secretion in healthy men. Visceral adiposity is a predictor of the glucose-stimulated increment in burst-like cortisol output, suggesting an additional possible mechanism for increased cardiovascular risk in abdominal obesity.

Ingestion of glucose, amino acids, protein, or mixed meals tends to increase serum and salivary cortisol concentrations in healthy adults (1–8). Gender, time of day, and enteric peptides, such as glucagon-like peptide, tachykinins, and glucose-dependent insulinotropic peptide, may modulate such effects (7–10). In pathological states like ACTH-independent macronodular adrenal hyperplasia, anomalous or exaggerated expression of peptidyl and adrenergic receptors may contribute to excessive cortisol secretion with meals (11, 12). However, the precise mechanisms that mediate oral nutrient effects in healthy individuals are not known. Indeed, under physiological conditions, both ACTH-dependent and ACTH-independent mechanisms of food-induced cortisol secretions have been postulated (3, 13–15). To our knowledge, pulsatile ACTH secretion after caloric ingestion has never been quantified adequately by current standards (16). This limitation is significant because pivotal meal-triggered mechanisms could include amplification of basal (nonpulsatile) or pulsatile ACTH secretion, enhancement of ACTH-cortisol synchrony, potentiation of ACTH-cortisol dose-responsiveness, and augmentation of adrenal cortisol secretion independently of ACTH.

The present investigations used a paired within-subject crossover design with frequent (10-min) sampling over 6.5 h to measure time-varying ACTH and cortisol concentrations in 58 adults before and after ingestion of a fixed glucose load or equivalent volume of water. Deconvolution, approximate entropy (ApEn), and ACTH-cortisol dose-response analyses were then applied to test the foregoing hypotheses noninvasively.

Subjects and Methods

Subjects

Fifty-eight healthy men were recruited to participate after providing voluntary written informed consent approved by the local Institutional Review Board. The admissible age range was 19–78 yr, with body mass index of 20–39 kg/m². Exclusion diagnoses were congestive heart failure, acute or chronic liver or renal disease, anemia, hypothalamopituitary disease, neuropsychiatric drug exposure, glucocorticoid use, systemic inflammatory disease, malignancy, substance abuse, intracranial disease, sleep apnea, and diabetes mellitus. Inclusion criteria were community-dwelling, independently living, consenting adults with stable diurnal work habits, body weight (within 2 kg in 3 months), and recreational exercise patterns.

Protocol

Subjects (n = 58) undertook two 10-min sampling sessions after overnight fasting, beginning at 0800 h. At 0830 h, glucose (75 g) or the same volume of water (10 ounces) was administered orally. Blood sampling continued thereafter for 6 more hours (until 1430 h). Plasma was obtained in chilled tubes containing divalent-metal chelators. An abdominal computed tomography (CT) scan was performed at the L3–4 interspace to estimate abdominal visceral fat (AVF) cross-sectional area, as described (17). In three subjects, there was a delay (not exceeding 1 h) in starting the protocol.

Assays

Circulating concentrations of ACTH, cortisol, and insulin were assayed by Immulite 2000 (Siemens Healthcare Diagnostics, Flanders, NJ), using reagents from the Siemens Healthcare Diagnostics. The assay for cortisol has a detection range of 0.2–50 μg/dl with intra-and interassay coefficients of variation of 7.2–9.4% and 6.3–7.5% at respective concentrations of 3.8–44 and 3.7–41 μg/dl. ACTH assay has a detection range of 5–1250 ng/liter, with intra- and interassay coefficients of variation of 6.1–8.2% and 4.4–5.7% at respective concentrations of 32–417 and 30–446 ng/liter. Single fasting blood specimens were used for the measurements of glucose and insulin. Synchron (Beckman Coulter, Fullerton, CA) and Siemens Dimension Vista autoanalyzers were used for the measurement of serum glucose concentrations.

Analyses

Plasma ACTH and cortisol time series were subjected to automated deconvolution analysis using a Matlab-implemented maximum-likelihood methodology (18). The two-component cortisol half-life model was 2.4 and 56 min (63% slow decay), and that of ACTH was 3.5 and 18 min (63% slow component) (19, 20). Outcome variables were basal (nonpulsatile), pulsatile and total (sum of basal plus pulsatile) secretion, and the mass (concentration units), number (per 6 h after the ingestion), and shape (mode) of ACTH and cortisol secretory bursts.

ApEn was calculated on the last 5.5 h of sampling (beginning 30 min after glucose ingestion). ApEn provides a scale-independent model-free estimate of secretory-pattern reproducibility or regularity, wherein higher ApEn corresponds to greater irregularity or higher process randomness (21). ApEn provides a surrogate measure of changes in feedback control in interconnected systems with high sensitivity and specificity (both >90%) (22). Cross-ApEn is the bivariate counterpart applied to paired time series, where higher values identify greater asynchrony (less pattern coordination), and conversely (23).

Dose-response estimates of ACTH-cortisol drive were performed as recently described (24). The Matlab program regresses deconvolved cortisol secretion rates on reconvolved ACTH concentrations via a four-parameter logistic dose-response model. Analyses used the paired 6.5-h ACTH-cortisol time series in each subject.

Statistics

A paired Student's t test was used to evaluate the effect of oral glucose compared with water ingestion on deconvolution, ApEn, and dose-response parameters. Linear regression analysis was employed to assess effects of age and/or AVF on ACTH and cortisol measures (25). Systat 11 (Systat Inc., Richmond, CA) was the software platform. Data are expressed as the mean ± sem.

Results

Fasting plasma glucose (mg/dl) and insulin (mU/liter) concentrations averaged 94 ± 1.2 and 5.9 ± 0.61, respectively. The age range was 19–78 yr, and body mass index range was 20–39 kg/m². Figure 1 gives the mean (±sem) paired profiles of ACTH and cortisol concentrations measured every 10 min for 6.5 h in the 58 men studied before (0.5 h) and after (6.0 h) water and glucose ingestion. There were visually prominent increases in ACTH and cortisol 2.5–4 h after glucose ingestion.

Fig. 1.

Mean ± sem plasma cortisol (top) and ACTH (bottom) 10-min time series over 6.5 h in the fasting control setting (closed circles) and after glucose ingestion (open circles) in 58 healthy men. Time zero is 0800 h. Glucose (OGT) or water was administered orally at 30 min (vertical arrows).

Deconvolution analysis was used to estimate ACTH and cortisol secretion over the 6-h interval starting with glucose or water ingestion. Figure 2A gives mean ± sem outcomes for ACTH: basal secretion, pulsatile secretion, secretory-burst mass, and number on the control (fasting) and glucose-ingestion days. Paired statistical comparisons (control vs. glucose) disclosed the following: 1) a 31% increase in pulsatile ACTH secretion on the glucose day (P = 0.009), due to commensurately augmented ACTH secretory-burst mass (P < 0.001) rather than number (P = 0.055); and 2) no change in basal (nonpulsatile) ACTH secretion (P = 0.298). In addition, ACTH secretory-burst shape (modal time in minutes of maximal ACTH secretion rate) was similar for control (9.3 ± 0.68) and glucose (9.4 ± 0.73) conditions. Thematically comparable outcomes were observed for glucose-induced cortisol secretion, in which pulsatile secretion rose by 27% (P = 0.001) (Fig. 2B). Glucose selectively augmented the size (mass) of cortisol secretory bursts (P = 0.004) and weakly decreased basal cortisol secretion (P = 0.031) with no effect on cortisol pulse number (P = 0.890). The mode was invariant of condition (grand mean, 12 ± 0.62 min).

Fig. 2.

ACTH (A) and cortisol (CORT) (B) deconvolution analysis. Bar graphs with paired comparisons give the mean ± sem (n = 58 men) based upon Student's t test. Deconvolution outcomes apply to the 6-h interval immediately after water (fasting) or 75-g glucose (OGT) ingestion (40–390 min, Fig. 1).

To assess whether increased pulsatile cortisol secretion reflected increased pulsatile ACTH secretion, intraindividual glucose-minus-control incremental values for cortisol were regressed on matching incremental values for ACTH (Fig. 3). This yielded Pearson's P < 0.0001 and R² = 0.325. If five Systat-identified high-leverage values were removed, the regression yielded P = 0.0003, R² = 0.227 (Fig. 3, inset). The pulsatile ACTH-pulsatile cortisol relationship was equally strong by nonparametric Spearman's rank correlation, P = 0.000635, rho=0.435 (n = 58). Regression of incremental cortisol secretory-burst mass (dependent variable) on incremental ACTH secretory-burst mass (independent variable) resulted in Pearson's P = 0.046 and R² = 0.070 (n = 58) (without six extreme values, P was 0.0085, R² = 0.130). For nonparametric Spearman's rank correlation, P was 0.00998 and rho was 0.336 (n = 58).

Fig. 3.

Linear regression analysis of the relationship between incremental pulsatile cortisol (dependent variable) and incremental pulsatile ACTH (independent variable) secretion rates (concentration units per 6 h) in 58 men. Each datum is an incremental (glucose minus control) pulsatile secretion rate for ACTH (x-axis) and cortisol (y-axis). Pearson's parametric correlation estimates are stated numerically. The inset tests the same relationship when five high-leverage values are removed.

AVF was a positive determinant of cortisol secretory-burst mass increments (glucose minus control value) with P = 0.019, R² = 0.10 (Fig. 4, top). In addition, AVF was a positive correlate of ACTH secretory-burst mode (duration) with P = 0.023, R²=0.091 (Fig. 4, bottom).

Fig. 4.

Linear regression of incremental cortisol secretory-burst size (mass) (top) and incremental ACTH secretory-burst duration (mode) (bottom) on CT-estimated AVF cross-sectional area. The single boxed datum is an outlier by P < 0.001 for Studentized residuals. Pearson's P and R² values are stated.

Univariate ApEn was applied separately to ACTH and cortisol time series as a model-free measure of altered feedback control (Subjects and Methods). ApEn was higher for ACTH than cortisol in the control and glucose conditions (both P < 0.01), indicating greater process randomness in ACTH than cortisol secretory patterns (Fig. 5). Glucose ingestion decreased ApEn of ACTH (P = 0.021) and cortisol (P = 0.025), signifying reduced process randomness (greater pattern reproducibility). Bivariate cross-ApEn estimates unmasked prominent reductions in both forward ACTH-cortisol (P = 0.001) and reverse cortisol-ACTH (P < 0.001) cross-ApEn, denoting marked joint-synchrony enhancement after glucose exposure. ApEn of ACTH (R² = 0.15; P = 0.035), as well as cross-ApEn of ACTH-cortisol (R² = 0.086; P = 0.026) and cortisol-ACTH (R² = 0.83; P = 0.028) increased with age in the control but not the glucose-administration session (plots not shown).

Fig. 5.

Glucose (OGT) compared with water ingestion regularizes ACTH and cortisol secretion (decreases ApEn) and synchronizes ACTH-cortisol and cortisol-ACTH secretion (decreases cross-ApEn) in healthy men. Regularization and synchrony predict significant feedback adaptations within the corticotropic axis. Connected data are paired values (n = 58 men). P values are paired two-tailed Student's t estimates after natural-logarithmic transformation.

To test the hypothesis that glucose administration enhances ACTH-cortisol feedforward coupling, analytical dose-response estimation was carried out using the sensitivity model (Subjects and Methods). Adrenal sensitivity (slope term), the one-half maximally effective concentration of ACTH [EC₅₀ (ng/liter)], ACTH efficacy (μg/dl · min cortisol secretion), and basal secretion (same units) were all independent of glucose ingestion (Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org).

Discussion

The present analyses in 58 healthy men demonstrate concomitant amplification of pulsatile ACTH and pulsatile cortisol secretion after 75-g glucose ingestion in the morning. The mechanism entailed selective augmentation of ACTH and cortisol secretory-burst mass. Indeed, ACTH and cortisol burst size rose comparably by 27–31% after oral glucose administration (R² = 0.325; P < 0.0001). Basal (nonpulsatile) ACTH secretion and cortisol secretory-burst number and shape were unaffected at good statistical power (β > 0.85 for α = 0.05 and δ = 30%). Model-free ApEn and cross-ApEn analyses further unveiled enhancement of joint ACTH-cortisol and cortisol-ACTH secretory synchrony (both P ≤ 0.001). In contrast, analytical estimates of endogenous ACTH-cortisol dose-response properties were unaffected by oral glucose exposure. In ensemble, these data demonstrate that morning glucose ingestion stimulates pulsatile cortisol and ACTH secretion, thereby elevating their mean concentrations; the increase in pulsatile cortisol is directly proportionate to the increase in pulsatile ACTH secretion; both increments are due to selective augmentation of secretory-burst size; the glucose effect includes enhancement of ACTH-cortisol secretory synchrony; and the glucose effect does not require a commensurate change in ACTH-cortisol dose-response properties.

Exploratory regression analyses revealed that CT-estimated AVF positively predicts incremental pulsatile cortisol secretion in response to a glucose load. A plausible mechanism is that relative obesity augments the release of gut-derived insulinotropic peptides, like glucagon-like peptide-1, gastric inhibitory peptide, and glucagon, which amplify secretion of not only insulin but also cortisol and CRH (9, 10, 12, 14). These gut peptides were not measured here. In fact, the extent of gut peptide-derived facilitation of cortisol secretion in obesity is not known, but any effects might be reduced by the tendency for lower endogenous peptide levels, at least in diabetic individuals (e.g. glucagon-like peptide-1) (26). In the present study, neither glucose nor insulin was a strong predictor of the incremental rise in pulsatile ACTH or cortisol secretion. However, adiponectin was a positive correlate of incremental adrenal sensitivity to ACTH after glucose ingestion (P = 0.027; R² = 0.084; n = 58 men) (Supplemental Fig. 1). As surrogate measures of adverse metabolic risk (27), AVF and omental fat-cell size have been associated in some but not all studies with increased activity or expression of 11β-hydroxysteroid dehydrogenase type 1, which promotes conversion of inactive cortisone to active cortisol (28–31). If hepatic, splanchnic, or whole-body cortisone activation does increase with AVF, this mechanism might amplify conversion of glucose-induced pulsatile cortisone to cortisol. Other mechanisms could include modulation of adrenal stimulation by splanchnic neurotransmitters (α-2, dopamine, β-adrenergic, GABAergic, nitric oxide), insulin, and nonenteric peptides, such as adiponectin, leptin, IGF-II, IGF-I, or TNFα (11, 32–38). Whatever the mechanism(s), an underlying pulsatile ACTH input to the adrenal cortex seems required to augment glucocorticoid pulses per se (39), as observed here. Moreover, an enteric route of glucose delivery appears critical because overnight iv glucose actually blunts the normal early-morning increase in cortisol secretion (40).

Caveats include the need for further studies to ascertain interactions among age, gender, and glucose ingestion; to discriminate ACTH-independent vis-à-vis ACTH-coupled cortisol secretion; and to define how glucose-induced ACTH/cortisol pulses are altered in the metabolic syndrome. The delayed ACTH-cortisol response to oral glucose suggests that glucose requires uptake and metabolism to initiate the activating effect. In particular, glucose may induce a cascade of intermediate responses, e.g. via adipocytokines, which take time to evolve sequentially (26). Although only glucose was evaluated here, the capability of protein or fat loads to induce similar changes in ACTH-cortisol secretion could be studied by the new methodology implemented here. In addition, whether hypothalamic-pituitary-adrenal reactivity to other stressors is altered by glucose ingestion is not known at this time.

In conclusion, glucose ingestion in the morning selectively augments burst-like ACTH and cortisol secretion and markedly synchronizes ACTH and cortisol secretory patterns. Glucose exposure does not alter analytical estimates of pulsatile ACTH-cortisol dose-responsiveness. Abdominal visceral adiposity positively predicts glucose-evoked increments in pulsatile cortisol secretion. Together, these data elucidate important interactions among glucose ingestion, body composition, and pulsatile secretion of ACTH and cortisol.