https://orcid.org/0000-0002-4908-0596
Abstract
Pulsatile insulin secretion is impaired in diseases such as type 2 diabetes that are characterized by insulin resistance. This has led to the suggestion that changes in insulin pulsatility directly impair insulin signaling. We sought to examine the effects of pulse characteristics on insulin action in humans, hypothesizing that a decrease in pulse amplitude or frequency is associated with impaired hepatic insulin action.
We studied 29 nondiabetic subjects on two occasions. On 1 occasion, hepatic and peripheral insulin action was measured using a euglycemic clamp. The deuterated water method was used to estimate the contribution of gluconeogenesis to endogenous glucose production. On a separate study day, we utilized nonparametric stochastic deconvolution of frequently sampled peripheral C-peptide concentrations during fasting to reconstruct portal insulin secretion. In addition to measuring basal and pulsatile insulin secretion, we used approximate entropy to measure orderliness and Fourier transform to measure the average, and the dispersion of, insulin pulse frequencies.
In univariate analysis, basal insulin secretion (R2 = 0.16) and insulin pulse amplitude (R2 = 0.09) correlated weakly with insulin-induced suppression of gluconeogenesis. However, after adjustment for age, sex, and weight, these associations were no longer significant. The other pulse characteristics also did not correlate with the ability of insulin to suppress endogenous glucose production (and gluconeogenesis) or to stimulate glucose disappearance.
Overall, our data demonstrate that insulin pulse characteristics, considered independently of other factors, do not correlate with measures of hepatic and peripheral insulin sensitivity in nondiabetic humans.
Insulin is secreted in a pulsatile fashion into the portal vein although this is not readily discernible when sampled in the systemic circulation due to the combined effects of hepatic insulin extraction (which may be directly affected by insulin pulse amplitude (1)) and waveform damping (2). Changes in pulse amplitude and frequency result in changes in overall insulin secretion and indeed conditions such as type 2 diabetes, which are characterized by impaired β-cell function, exhibit decreased pulse amplitude and frequency (2). Similarly, first-degree relatives of people with type 2 diabetes exhibit defective oscillatory insulin release (3), as do women with polycystic ovarian syndrome (4). Increased disorderliness of insulin release and decreased secretory burst mass has been observed in aging individuals (5). Prior work has demonstrated that impaired insulin pulse characteristics impair peripheral (6) and hepatic insulin action (7,8) in dogs and rodents, respectively. This could suggest a mechanism underpinning the observation that in prediabetes impaired insulin secretion is accompanied by impaired insulin action (9-13).
Prior experiments have attempted to address whether pulse characteristics might correlate with measures of insulin sensitivity in humans (14). However, many experiments in humans have utilized artificially generated peripheral insulin pulses (15-19) to demonstrate an effect on hepatic or peripheral insulin signaling or both. These effects have not been universally observed (20,21), and the methods used introduce differences in overall insulin exposure during pulsatile vs continuous delivery. In addition, the use of a hyperinsulinemic, euglycemic clamp where endogenous glucose production (EGP) is completely suppressed precludes measurement of hepatic insulin action (20,22). Recently, we developed a novel method for insulin pulse characterization, which avoids the need for hepatic vein sampling by using nonparametric stochastic deconvolution of peripheral venous C-peptide concentrations to reconstruct insulin secretion after individualized determination of C-peptide clearance (23). This method has the added advantage of avoiding the effects of hepatic insulin extraction on insulin pulse characteristics measured in the systemic circulation (1).
We used this method to demonstrate an effect of diabetes-associated variation in the TCF7L2 locus on insulin pulse orderliness (24). On a separate study day, insulin action was measured using a hyperinsulinemic, euglycemic clamp designed to produce ~50% suppression of EGP to measure the direct effects (independent of changes in insulin and glucagon secretion) of diabetes-associated variation in the TCF7L2 locus on insulin action and hepatic glucose metabolism (25). These data sets provided an opportunity to determine whether insulin pulse characteristics in the fasting state correlated with insulin action. We therefore sought to determine whether increased pulse amplitude or increased pulse frequency was associated with an improved ability of insulin to suppress EGP and stimulate glucose disappearance (Rd). The experimental design also allowed us to examine the effect of insulin on the suppression of gluconeogenesis (measured via the deuterated water method) as previously described (26), modified to correct for the transaldolase reaction (27) as described in (25).
Materials and Methods
Study subjects
After approval by the Mayo Clinic Institutional Review Board, data from 29 nondiabetic subjects who had provided informed, written consent prior to participation in a series of published experiments (24,25,28) was utilized for this analysis. Their characteristics are described in Table 1. At the time of study they had no active illness, no prior history of diabetes, and were not taking medications that might alter glucose metabolism. Body composition was measured by dual energy X-ray absorptiometry (iDXA scanner; GE, Wauwatosa, WI, USA).
Subject characteristics at the time of screening, during the clamp, and during the insulin pulsatility studies
| Characteristics | Mean ± SEM |
|---|---|
| Demographic characteristics | |
| n (M/F) | 29 (10/19) |
| Age (years) | 46 ± 2 |
| Total body mass (kg) | 78.9 ± 2.6 |
| Lean body mass (kg) | 48.2 ± 2.1 |
| BMI (kg/m2) | 27.8 ± 0.7 |
| n (NGT/IGT) | 17/12 |
| Fasting glucose (mmol/L) | 4.85 ± 0.10 |
| At time of screening (oral glucose test) | |
| Si (10–4 dl/kg/min per μU/ml) | 15 ± 2 |
| Φ (10-9min−1) | 53 ± 2 |
| DI (10–14 dL/kg/min2 per pmol/L) | 1362 ± 216 |
| Clamp study | |
| Fasting EGP (μmol/kg/min) | 14.2 ± 0.5 |
| Fasting gluconeogenesis (μmol/kg/min) | 7.3 ± 0.5 |
| Fasting insulin (pmol/L) | 24 ± 2 |
| Clamp EGP (μmol/kg/min) | 6.8 ± 1.1 |
| Clamp Rd (μmol/kg/min) | 29.7 ± 2.8 |
| Clamp gluconeogenesis (μmol/kg/min) | 2.7 ± 0.4 |
| Clamp insulin (pmol/L) | 76 ± 4 |
| Insulin pulsatility study | |
| Basal insulin secretion (pmol/L/min) | 126 ± 8 |
| Insulin pulse amplitude (pmol/L/min) | 41 ± 3 |
| Fasting pulse interval (min) | 5.7 ± 0.4 |
| Fasting ApEn | 1.14 ± 0.03 |
| Fasting FDI | 0.47 ± 0.01 |
| Characteristics | Mean ± SEM |
|---|---|
| Demographic characteristics | |
| n (M/F) | 29 (10/19) |
| Age (years) | 46 ± 2 |
| Total body mass (kg) | 78.9 ± 2.6 |
| Lean body mass (kg) | 48.2 ± 2.1 |
| BMI (kg/m2) | 27.8 ± 0.7 |
| n (NGT/IGT) | 17/12 |
| Fasting glucose (mmol/L) | 4.85 ± 0.10 |
| At time of screening (oral glucose test) | |
| Si (10–4 dl/kg/min per μU/ml) | 15 ± 2 |
| Φ (10-9min−1) | 53 ± 2 |
| DI (10–14 dL/kg/min2 per pmol/L) | 1362 ± 216 |
| Clamp study | |
| Fasting EGP (μmol/kg/min) | 14.2 ± 0.5 |
| Fasting gluconeogenesis (μmol/kg/min) | 7.3 ± 0.5 |
| Fasting insulin (pmol/L) | 24 ± 2 |
| Clamp EGP (μmol/kg/min) | 6.8 ± 1.1 |
| Clamp Rd (μmol/kg/min) | 29.7 ± 2.8 |
| Clamp gluconeogenesis (μmol/kg/min) | 2.7 ± 0.4 |
| Clamp insulin (pmol/L) | 76 ± 4 |
| Insulin pulsatility study | |
| Basal insulin secretion (pmol/L/min) | 126 ± 8 |
| Insulin pulse amplitude (pmol/L/min) | 41 ± 3 |
| Fasting pulse interval (min) | 5.7 ± 0.4 |
| Fasting ApEn | 1.14 ± 0.03 |
| Fasting FDI | 0.47 ± 0.01 |
Abbreviations: BMI, body mass index, NGT, normal glucose tolerance, IGT, impaired glucose tolerance, Si, insulin action, Φ, beta cell responsivity to glucose, DI, disposition index, EGP, endogenous glucose production, Rd, glucose disappearance, ApEn, approximate entropy, FDI, frequency dispersion index.
Subject characteristics at the time of screening, during the clamp, and during the insulin pulsatility studies
| Characteristics | Mean ± SEM |
|---|---|
| Demographic characteristics | |
| n (M/F) | 29 (10/19) |
| Age (years) | 46 ± 2 |
| Total body mass (kg) | 78.9 ± 2.6 |
| Lean body mass (kg) | 48.2 ± 2.1 |
| BMI (kg/m2) | 27.8 ± 0.7 |
| n (NGT/IGT) | 17/12 |
| Fasting glucose (mmol/L) | 4.85 ± 0.10 |
| At time of screening (oral glucose test) | |
| Si (10–4 dl/kg/min per μU/ml) | 15 ± 2 |
| Φ (10-9min−1) | 53 ± 2 |
| DI (10–14 dL/kg/min2 per pmol/L) | 1362 ± 216 |
| Clamp study | |
| Fasting EGP (μmol/kg/min) | 14.2 ± 0.5 |
| Fasting gluconeogenesis (μmol/kg/min) | 7.3 ± 0.5 |
| Fasting insulin (pmol/L) | 24 ± 2 |
| Clamp EGP (μmol/kg/min) | 6.8 ± 1.1 |
| Clamp Rd (μmol/kg/min) | 29.7 ± 2.8 |
| Clamp gluconeogenesis (μmol/kg/min) | 2.7 ± 0.4 |
| Clamp insulin (pmol/L) | 76 ± 4 |
| Insulin pulsatility study | |
| Basal insulin secretion (pmol/L/min) | 126 ± 8 |
| Insulin pulse amplitude (pmol/L/min) | 41 ± 3 |
| Fasting pulse interval (min) | 5.7 ± 0.4 |
| Fasting ApEn | 1.14 ± 0.03 |
| Fasting FDI | 0.47 ± 0.01 |
| Characteristics | Mean ± SEM |
|---|---|
| Demographic characteristics | |
| n (M/F) | 29 (10/19) |
| Age (years) | 46 ± 2 |
| Total body mass (kg) | 78.9 ± 2.6 |
| Lean body mass (kg) | 48.2 ± 2.1 |
| BMI (kg/m2) | 27.8 ± 0.7 |
| n (NGT/IGT) | 17/12 |
| Fasting glucose (mmol/L) | 4.85 ± 0.10 |
| At time of screening (oral glucose test) | |
| Si (10–4 dl/kg/min per μU/ml) | 15 ± 2 |
| Φ (10-9min−1) | 53 ± 2 |
| DI (10–14 dL/kg/min2 per pmol/L) | 1362 ± 216 |
| Clamp study | |
| Fasting EGP (μmol/kg/min) | 14.2 ± 0.5 |
| Fasting gluconeogenesis (μmol/kg/min) | 7.3 ± 0.5 |
| Fasting insulin (pmol/L) | 24 ± 2 |
| Clamp EGP (μmol/kg/min) | 6.8 ± 1.1 |
| Clamp Rd (μmol/kg/min) | 29.7 ± 2.8 |
| Clamp gluconeogenesis (μmol/kg/min) | 2.7 ± 0.4 |
| Clamp insulin (pmol/L) | 76 ± 4 |
| Insulin pulsatility study | |
| Basal insulin secretion (pmol/L/min) | 126 ± 8 |
| Insulin pulse amplitude (pmol/L/min) | 41 ± 3 |
| Fasting pulse interval (min) | 5.7 ± 0.4 |
| Fasting ApEn | 1.14 ± 0.03 |
| Fasting FDI | 0.47 ± 0.01 |
Abbreviations: BMI, body mass index, NGT, normal glucose tolerance, IGT, impaired glucose tolerance, Si, insulin action, Φ, beta cell responsivity to glucose, DI, disposition index, EGP, endogenous glucose production, Rd, glucose disappearance, ApEn, approximate entropy, FDI, frequency dispersion index.
Experimental design—clamp study
Participants were admitted to the Clinical Research and Translation Unit at 5 pm on the day prior to study. During an overnight fast 1.67g/kg wt. of deuterated water (2H2O) was then given in 3 divided doses at 10 pm, 12 am, and 2 am. The following morning at 6 am, a dorsal hand vein was cannulated and placed in a heated Plexiglas box and maintained at 55oC to allow sampling of arterialized venous blood. The contralateral forearm vein was cannulated for tracer, glucose, and hormone infusions. At 6:30 am (−180 min) a primed, continuous infusion of [3-3H] glucose (12 μCi prime, 0.12 μCi/ min continuous) and [1-13C] acetate (2.5 μmol/kg/min) was started and continued for the duration of the experiment. At 9:30 am (0 min), an infusion of somatostatin (60 ng/kg/min), glucagon (0.65 ng/kg/min), and growth hormone (0.25 ng/kg/min) was started and maintained for the duration of study. Insulin was also infused at 0.30 mU/kg/min. At this time a variable infusion of 50% dextrose containing [3-3H] glucose commenced with the infusion rate varied to maintain glucose at ~5.5 mmol/L over the period of study (25). At 255 min, a bolus of C-peptide (60 pmol/kg) was administered and blood sampled over the subsequent 2 h to enable individual calculation of C-peptide clearance (28).
Experimental design—insulin pulsatility study
Subjects were admitted to the Clinical Research and Translation Unit at 5 pm on the day prior to study. The following morning (at approximately 6:30 am), an 18-g cannula was inserted retrogradely into a dorsal hand vein. This was then placed in a heated Plexiglas box maintained at 55oC to allow sampling of arterialized venous blood. Subjects were then moved to the radiology suite where a hepatic vein catheter was placed via the femoral vein under fluoroscopic guidance. Following their return, at 8 am (0 min) blood was sampled at 1-min intervals from the arterialized hand vein and from the hepatic vein at 2-min intervals over a 45-min period (fasting phase). At 8:46 am (46 min) glucose infusion commenced, and the infusion rate was adjusted to rapidly achieve and maintain peripheral glucose concentrations of ~9.8mmol/L. Following this 30 min (9:15 am), blood was sampled during the hyperglycemic phase for an additional 45 min (23,24).
Analytic techniques
Plasma samples were placed on ice, centrifuged at 4°C, separated, and stored at −20°C until assayed. Glucose concentrations were measured using a glucose oxidase method (Yellow Springs Instruments, Yellow Springs, OH, USA). Plasma insulin was measured using a chemiluminescence assay (Access Assay; Beckman, Chaska, MN, USA). C-peptide was measured using a 2-site immunenzymatic sandwich assay (Roche Diagnostics, Indianapolis, IN, USA) in accordance with the manufacturer’s instructions. Plasma glucagon was measured by radio-immunoassay (Linco Research, St. Louis, MO, USA). Plasma [3-3H] glucose specific activity was measured by liquid scintillation counting. The method of Jones et al (29) was used to measure and analyze deuterium enrichment on the second and fifth carbon of plasma glucose. To correct for errors introduced by transaldolase exchange (27), estimation of transaldolase exchange was accomplished by measurement of [3-13C] glucose and [4-13C] glucose enrichment by nuclear magnetic resonance spectroscopy as previously described (25).
Calculations
Glucose appearance and disappearance during the clamp study were calculated using the steady state equations of Steele (30) where the actual tracer infusion rate was utilized. The volume of distribution of glucose was assumed to be 200 mL/kg with a pool correction factor equal to 0.65. EGP was calculated by subtracting the glucose infusion rate from the tracer-determined rate of glucose appearance. Fasting and clamp rates of EGP represent the mean of the −30 to 0 min and the 210 to 240 min values, respectively. All rates of infusion and turnover were expressed per kilogram of lean body mass. The rate of gluconeogenesis was calculated by multiplying the C5/C2 ratio by the respective EGP. (25). Hepatic insulin action was estimated from the relationship of EGP to insulin (Si EGP) as previously described (31). A similar analysis was undertaken for gluconeogenesis (Si GNG) and for Rd (Si Rd).
We also undertook a model-independent analysis of insulin action; EGP (and subsequently gluconeogenesis and Rd) during fasting and the clamp and their respective insulin concentrations were loge-transformed and the gradient of loge EGP vs Loge insulin in each individual (fasting vs clamp) was used as a measure of hepatic insulin action.
During the pulsatility study, the insulin secretion rate over time was derived from peripheral C-peptide concentrations, measured at 2-min intervals (23) using nonparametric deconvolution (32), and C-peptide kinetics were directly calculated from individual experimental data of C-peptide decay as previously described (23). Once the pulsatile insulin secretion rate over time was reconstructed, we analyzed the euglycemic and hyperglycemic portion of the signal separately to derive measures of basal insulin secretion, pulse amplitude, and pulse orderliness using approximate entropy (ApEn). We then used the Fast Fourier Transform algorithm implemented in MATLAB R2017b to obtain a periodogram for each individual. This was used to calculate main pulse frequency and to better characterize the contribution of different pulse frequencies to overall secretion using a frequency dispersion index (FDI). This was validated against hepatic vein insulin concentrations (23). Data from the euglycemic portion of the experiment are utilized in the current analysis.
Statistical analysis
The primary hypothesis of this analysis based on prior work (7) was that decreased insulin pulse amplitude (1° hypothesis) and a decreased insulin pulse frequency (2° hypothesis) are associated with impaired hepatic insulin sensitivity. In that paper, Matveyenko et al reported an R2 > 0.80 for the relationship of pulse amplitude with clamp M-value and for the relationship of pulse amplitude with glucose disposal (7). Assuming an α = 0.05, 29 subjects would provide ~85% power to detect an effect size of R2 = 0.20.
Data are presented as mean ± SEM. Multivariate analysis adjusting for the effects of age, sex, and weight was performed in JMP Pro 11 (SAS Institute Inc., Cary, NC, USA). Residuals for the conditional logistic regression of a particular parameter with the covariates were used to confirm or refute the contribution of that parameter to variation in insulin action. A P-value < 0.05 was considered to be statistically significant.
Results
Subject characteristics
Subject characteristics are recorded in Table 1.
EGP, gluconeogenesis, and glucose disappearance as a function of insulin concentrations during fasting and the clamp study
Fasting insulin concentrations (24 ± 2 pmol/L) rose to 76 ± 4 pmol/L during the clamp (Fig. 1A-1C. This suppressed EGP from 14 ± 1 μmol/kg/min to 7 ± 1 μmol/kg/min (Fig. 1A). Gluconeogenesis was also suppressed by insulin (Fig. 1B). Rd increased to 30 ± 3 μmol/kg/min during the clamp (Fig. 1C).
The relationship of endogenous glucose production (A), gluconeogenesis (B) and glucose disappearance (Rd) (C) with insulin concentrations observed during fasting (○) and during the clamp (● ) phase of the experiment.
Relationship of insulin sensitivity of EGP and of gluconeogenesis as well as suppression of glucose disappearance with insulin pulse characteristics
ApEn, basal insulin secretion, pulse amplitude, pulse interval, and FDI did not correlate with fasting EGP; there was no correlation with insulin-induced suppression of EGP (Si EGP) (Fig. 2A-2E, respectively) after adjusting for the covariates of age, sex, and weight.
The ability of insulin to suppress endogenous glucose production was correlated with approximate entropy (ApEn) (A), basal insulin secretion (basal) (B), mean insulin pulse amplitude (amplitude) (C), pulse interval (interval) (D), and frequency dispersion index (FDI) (E). R2 and P-values represent results from a multivariate analysis adjusting for the effects of age, sex, and weight.
Similarly, ApEn, basal insulin secretion, pulse amplitude, pulse interval and FDI did not correlate with fasting rates of gluconeogenesis; there was also no correlation with insulin-induced suppression of gluconeogenesis (Si GNG) (Fig. 3A-3E, respectively) after adjusting for the covariates of age, sex, and weight.
The ability of insulin to suppress gluconeogenesis was correlated with approximate entropy (ApEn) (A), basal insulin secretion (basal) (B), mean insulin pulse amplitude (amplitude) (C), pulse interval (interval) (D), and frequency dispersion index (FDI) (E). R2 and P-values represent results from a multivariate analysis adjusting for the effects of age, sex, and weight.
Finally, the ability of insulin to stimulate Rd (Si Rd) exhibited no relationship with ApEn, basal insulin secretion, pulse amplitude, pulse interval and FDI (Fig. 4A-4E, respectively) after adjustment for the covariates of age, sex, and weight.
The ability of insulin to stimulate glucose disappearance was correlated with approximate entropy (ApEn) (A), basal insulin secretion (basal) (B), mean insulin pulse amplitude (amplitude) (C), pulse interval (interval) (D), and frequency dispersion index (FDI) (E). R2 and P-values represent results from a multivariate analysis adjusting for the effects of age, sex, and weight.
The model-independent method using loge-transformation to quantify insulin action on EGP, gluconeogenesis, and Rd (see previous discussion) likewise failed to demonstrate a relationship with insulin pulse characteristics (data not shown).
Discussion
In this series of experiments, we report that there is no correlation of the ability of insulin to suppress EGP (and specifically the component of EGP attributable to gluconeogenesis) as well as the ability of insulin to stimulate Rd with portal insulin pulse characteristics in nondiabetic subjects. This data set is somewhat unique in that it uses state-of-the-art measurement of insulin pulse characteristics as well as insulin sensitivity in a nondiabetic population to test the primary hypothesis that insulin pulse characteristics alter insulin action. In cross-sectional studies, insulin secretion and β-cell function decline in concert as glucose tolerance worsens, but the mechanism(s) underlying these observations have been unclear.
In this analysis, we had chosen to examine the association of insulin action with insulin pulse characteristics under euglycemic conditions since these are more likely to be representative of the portal insulin concentrations that the liver is exposed to after an overnight fast. However, we also undertook an exploratory analysis of the relationship between insulin pulse characteristics during hyperglycemia and hepatic and extrahepatic insulin action. The ApEn of insulin pulse characteristics during hyperglycemia was weakly associated with the ability of insulin to suppress EGP (R2 = 0.12, P = 0.05). This association is somewhat unexpected and its significance is unclear given a lack of association with the suppression of gluconeogenesis by insulin. None of the other pulse characteristics measured during hyperglycemia were associated with the ability of insulin to suppress EGP and stimulate Rd.
An earlier experiment that suggested this hypothesis reported that a 50% pancreatectomy in dogs induced peripheral insulin resistance. This decline in insulin action was strongly correlated with the decrease in insulin pulse amplitude—a consequence of the surgically induced decrease in beta cell mass (6). In a subsequent experiment, Matveyenko et al examined hepatic insulin signaling in rodents and dogs after exposure to intraportal insulin infusion, which was pulsatile, constant, or reproducing the pattern observed in type 2 diabetes (7). Endogenous islet secretion was inhibited by somatostatin and glucagon replaced at basal concentrations. The absence of insulin pulses impaired insulin receptor substrate 1 and 2 activation and downstream signaling as well as decreased expression of hepatic glucokinase (7).
The data from human studies to date have been more heterogeneous. Differences in the methodology used to quantify insulin pulse characteristics may be one explanation. For example, O’Rahilly et al used a Fourier transform in a manner similar to our methodology (3). On the other hand, multiple other studies utilized qualitative measures of insulin concentrations or parametric deconvolution of insulin concentrations (33). Other studies used the observed oscillation in frequently sampled peripheral insulin concentrations to describe insulin pulse characteristics (14). All of these methodologies, to some degree or another, ignore the confounding effect of hepatic insulin extraction on insulin concentrations in the hepatic vein and systemic circulation (1). The current method, based on the deconvolution of C-peptide concentrations, avoids these limitations (23).
Other studies examined insulin action in the presence and absence of artificial pulse generation. For example, Courtney et al compared uniform pulses at a frequency of 7 or 14 min with continuous insulin infusion and concluded that insulin action was unaffected (20). In contrast, Komjati et al reported that insulin pulses were more effective in suppressing EGP during continuous glucagon infusion as compared to continuous insulin delivery (8). Several other authors have suggested that pulsatile insulin delivery has advantages over continuous insulin infusion (15-18), but the relevance of these experimental conditions to normal physiology are uncertain.
An important consideration in experiments recreating pulsatile insulin secretion is that glucagon has actions on hepatic glucose metabolism that are opposite to those of insulin and are exacerbated by relative insulin deficiency (34). The hormone is secreted in a pulsatile fashion into the portal vein in a manner that may be reciprocal to insulin secretion in nondiabetic humans (35). It is therefore possible that when measuring the effect of pulsatile insulin infusion on hepatic insulin action, these actions may be enhanced by changes in endogenous glucagon secretion (16,17,19), which would also affect hepatic insulin action at the time of the experiment.
Steiner et al reported that the absence of first-phase insulin secretion enhances the gluconeogenic response to glucagon (36). Similarly, in healthy humans, the absence of a first-phase insulin secretory response delayed suppression of EGP by insulin (37). Moreover, as type 2 diabetes is characterized by an acquired defect of glucokinase and higher rates of gluconeogenesis, we also explored whether pulse characteristics might directly affect the suppression of gluconeogenesis (38). As was the case for EGP, pulse characteristics did not correlate with fasting rates of gluconeogenesis or with the ability of insulin to suppress gluconeogenesis.
In a negative study such as this, it is important to examine the limitations of this series of experiments. The first is that in our study population, although there was significant heterogeneity of hepatic insulin action, none of our subjects had overtly impaired fasting glucose. The absence of significant defects in fasting insulin secretion may imply that marked abnormalities of fasting insulin pulse amplitude and frequency are necessary to modulate hepatic insulin action. On the other hand, our methodology was capable of detecting small defects in pulse orderliness and frequency (measured by ApEn and FDI) attributable to a genetic variant (TCF7L2) genotype, which has relatively small effects on diabetes predisposition (24). This implies that our methodology is sensitive enough to detect subtle defects in insulin secretion. Finally, this is a correlation analysis based on experiments performed on 2 separate days—a design necessitated by the use of a euglycemic clamp with somatostatin to inhibit endogenous islet hormone secretion. This is an important consideration given the potential intrasubject variability in insulin secretion and action. However, to minimize day-to-day variation, we conducted all experiments at the same circadian time of the day (39). Such measures seem to be stable over extended periods of time (12 weeks) in our prior experiments (40). To ensure that our model-based measures of insulin action did not affect our conclusions, we also utilized a model-independent measure with similar results.
We conclude that within the spectrum of variation in insulin pulse characteristics seen in nondiabetic humans, there is no clear evidence that insulin pulse characteristics directly correlate with measures of hepatic glucose metabolism or insulin action. This does not preclude the possibility that more significant defects in insulin pulse amplitude and frequency, reflecting significant impairments of insulin secretion, are associated with alterations in hepatic metabolism or insulin action. More widespread adoption of this novel method for measuring insulin pulsatility without the need for a hepatic vein catheterization should facilitate the conduct of the requisite studies examining the effect of significant impairment in fasting insulin secretion on insulin action.
Acknowledgments
Financial Support: This study was supported by US National Institutes of Health (DK78646, DK116231), MIUR (Italian Minister for Education) under the initiative “Departments of Excellence” (Law 232/2016) and Mayo Clinic General Clinical Research Center (UL1 TR000135).
Author contributions: M.C.L. and C.D.M undertook mathematical modeling of insulin secretion and measurement of pulse characteristics; R.T.V. researched data and ran the studies; J.C.A. placed the hepatic vein catheters, contributed to the discussion, and reviewed/edited manuscript; J.G.J. and C.B. measured gluconeogenesis using the deuterated water method; R.A.R. and A.M. contributed to the discussion and reviewed/edited manuscript; G.D.N. and C.C. oversaw the mathematical models, contributed to the discussion and reviewed/edited manuscript, and A.V. designed the study, oversaw its conduct, researched data, and wrote the first draft of the manuscript. A.V. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Additional Information
Disclosures: The authors have declared that no conflict of interest exists.
Data Availability
The data sets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
References
