Assessment of the incretin effect in healthy subjects: concordance between clamp and OGTT methods

Abstract

The incretin effect describes the insulin response to nutrient ingestion that exceeds the response to glycemia per se. It is mediated by gastrointestinal factors and is necessary to maintain postprandial glucose homeostasis. The incretin effect results in a more than twofold increase of the insulin response to a meal in healthy people and two different techniques have been used in the past to measure its magnitude. Most studies employ an OGTT on 1 day, followed by a matching glucose infusion on a separate day. Another study design employs a hyperglycemic glucose clamp that is maintained after oral ingestion of glucose. Both protocols allow quantification of the incretin effect by comparing the insulin response to an identical glycemic stimulus. Here we performed a within-subject comparison of both techniques to quantify the incretin effect and suggest different calculation methods to interpret the results derived from the clamp experiment in a cohort of healthy young adults (n = 10, age 33 ± 4 yr). All subjects participated on four different study days: 1) OGTT, 2) isoglycemic glucose infusion (Iso-IV), 3) hyperglycemic clamp with oral glucose ingestion (clamp-OGTT), and 4) hyperglycemic clamp (clamp). With the classic OGTT/Iso-IV method, the insulin response to glucose ingestion increased more than twofold and was 60 ± 6% and 49 ± 5% for insulin and c-peptide. Different estimates of the incretin effect based on the clamp method ranged from 58% to 79% for insulin and 38% to 61% for c-peptide, both significantly higher than values derived from the OGTT/isoglycemic infusion method. However, when the effect of continuous hyperglycemia on insulin secretion was accounted for, using extrapolation from early time points of the clamp, good agreement was noted between the two methods. Based on these results, both techniques seem to be equally suited to measure the incretin effect and should be employed according to the scientific questions, experimental contingencies, and investigator experience.

NEW & NOTEWORTHY This proof-of-concept study shows that the incretin effect can be reliably assessed by two different methods with similar quantitative results. A single-day hyperglycemic clamp with oral glucose ingestion allows the determination of the incretin effect with fewer study days and less day-to-day variability.

INTRODUCTION

The stimulation of insulin secretion by factors secreted from the gastrointestinal tract was one of the first demonstrations of endocrine physiology (1, 2). This gut-islet connection, termed the incretin effect (3, 4) and/or the entero-insular axis (5), has been attributed primarily to two peptides produced in the intestinal mucosa, glucose-dependent insulinotropic polypeptide (GIP), and glucagon-like peptide 1 (GLP-1). These factors are released in response to nutrient ingestion and are essential for normal glucose tolerance (6–9), allowing insulin secretion proportional to nutrient absorption at prandial levels of glycemia (10–12). Because of their important physiology, GLP-1 and GIP have been the basis of active pharmaceutical development over the past two decades leading to new classes of medication to treat diabetes (13). A number of studies have characterized the incretin effect in various pathological states (14–18), findings that support the incretin system as a critical component in glucose regulation (19).

A method for quantifying the incretin effect in diabetic and nondiabetic subjects has become a classic technique in metabolic physiology (12, 20, 21). In this approach, a subject has an oral glucose tolerance test (OGTT) as the first study, using frequent blood sampling to define a comprehensive glycemic profile. On a second day, the subject returns and receives intravenous glucose adjusted to match the glucose excursion during the OGTT. This generates a within-subject comparison, whereby the effects of plasma glucose on insulin secretion are controlled, and differences in β-cell secretion, reflected in plasma insulin or c-peptide, can be attributed to the route of glucose administration. Using this technique, the incretin effect, the difference in insulin release between the OGTT and the isoglycemic infusion, accounts for 30%–70% of prandial β-cell output (19, 22). An alternative to the 2-day method for determining the incretin effect is an approach in which blood glucose is clamped with an intravenous infusion that is maintained over several hours before and after an oral glucose tolerance test (9, 23, 24). Fundamentally, this method follows the same principles as the classical procedure, comparing insulin secretion at a fixed glucose level before and after enteral release of the incretins. Although slightly more complicated to perform, the clamp method has the advantage of deriving an estimate of the incretin effect in 1 day.

We have previously demonstrated that the relative magnitude of the incretin effect determined with the 1-day clamp method is reproducible within subjects and is similar in magnitude at different levels of glycemia (25). However, there remains some question as to the accuracy of this method since insulin secretion rises steadily when stimulated by fixed, steady-state hyperglycemia. In this study, we hypothesized that the incretin effect would be a stable measure in healthy subjects independent of the method used to determine it.

METHODS

Subjects

Ten healthy subjects (7 M/3F), aged 33.3 ± 4.1 yr, with a mean BMI of 26.0 ± 1.6 kg/m² and normal glucose tolerance (HbA1c: 5.0 ± 0.1%) were recruited for this study. All subjects were free of chronic medical conditions, had no family history of diabetes, and did not use any regular medication except for birth control pills among some of the women. Subjects had stable body weights over the previous 3 mo and were asked to refrain from strenuous physical exercise and consume their usual diet for 3 days before each day of study. The treatment and experimental protocols described in Experimental Procedures were approved by the institutional review board of the University of Cincinnati, and all participants provided written informed consent before the studies.

Experimental Procedures

Subjects were admitted to the Clinical Research Unit (CRU) at the Cincinnati VA Hospital in the morning after an overnight fast of at least 10 h. Each subject participated in four separate days of study. The OGTT was completed first in all subjects followed by the Iso-IV study within 7–20 days. Clamp-OGTTs and hyperglycemic clamps were performed randomly within 1 wk. All subjects had the 4 days of study completed within 4–5 wk. Body weight and height were measured, and pregnancy was excluded in female subjects. Intravenous catheters were placed in each forearm for the withdrawal of blood samples and infusion of glucose or saline. The arm and hand used for blood sampling were wrapped in a heating pad to maintain blood flow and a stable degree of arterialization of blood. At time 0, following the removal of four fasting blood samples over 20 min, subjects had:

• 1) An oral glucose tolerance test (OGTT). Subjects consumed 300 mL of a solution containing 75 g of glucose and 10 g of d-xylose over 10 min. Post glucose blood was sampled at 15, 30, 45, 60, 75, 90, 105, 120, 150 and 180 min.
• 2) An isoglycemic infusion (Iso-IV). Infusion of 20% dextrose solution was adjusted every 5 min to match arterialized venous glucose levels during the OGTT; blood sampling was identical to the OGTT.
• 3) A hyperglycemic clamp with oral glucose consumption (clamp-OGTT). A primed continuous infusion of 20% dextrose was started to clamp blood glucose at 10 mM for 270 min. At 90 min, subjects ingested 300 mL of a solution containing 75 g glucose and 10 g of d-xylose, and the dextrose infusion was adjusted to maintain the glucose clamp. Blood was sampled at 10-min intervals for the duration of the study.
• 4) A hyperglycemic clamp (clamp). A glucose infusion was given as above to clamp blood glucose at 10 mM for 270 min; blood sampling was identical to the clamp-OGTT.

Adjustments to the glucose infusion rate (GIR) were made empirically by the two investigators authoring this manuscript in response to changes in blood glucose measured every 3–5 min, and no precalculated algorithm was employed. Mean GIR are depicted in Fig. 1A and Fig. 2A.

Figure 1.

Glucose (A), insulin (B), and c-peptide (C) response to an oral glucose tolerance test (OGTT, full circle, black) or an isoglycemic intravenous glucose infusion (Iso-IV, open circle, white). Glucose infusion rate (GIR, A, right y-axis) during Iso-IV is depicted by the continuous line and striped area. Glucose infusion generally matched well between the two studies. Glucose AUC (right) was not different between the two study days (P = 0.2871). With the same glycemic stimulus insulin, AUC (P = 0.0029) and c-peptide AUC (P < 0.0001) were significantly greater after oral glucose load compared to intravenous glucose administration. Data are means ± SE. **P < 0.01, ***P < 0.001. AUC, area under the curve.

Figure 2.

Glucose (A), insulin (B), and c-peptide (C) response to a hyperglycemic clamp with oral glucose ingested at 90 min (clamp-OGTT, solid circles, black) or a hyperglycemic clamp alone (clamp, open circles, white) for 270 min. Glucose infusion rates (GIR, A, right y-axis) are shown as a continuous line with the dotted area (clamp-OGTT, fewer dots) and an interrupted line with the dotted area (clamp, dots). Glycemia did not differ between the two studies (P = 0.3731). The addition of oral glucose resulted in a significantly greater insulin and c-peptide response [average insulin (B) and c-peptide (C) from 100 to 270 min, P < 0.001] than hyperglycemia alone. Data are means ± SE. ***P < 0.001. OGTT, oral glucose tolerance test.

Assays of Substrates and Hormones

Glucose measurements were made at the bedside in 5-min intervals using a glucose analyzer (YSI Life Sciences, Yellow Springs, OH). Samples for determination of insulin, c-peptide, and d-xylose were collected in prechilled heparinized tubes and stored immediately on ice. Samples for determination of total GLP-1 and GIP were collected in prechilled tubes containing a mix of heparin, EDTA, and aprotinin. All samples were centrifuged within 1 h of withdrawal, frozen, and stored at −80°C for later measurements of peptide hormone concentrations. Insulin and c-peptide concentrations were assayed by RIA (26), and d-xylose plasma concentration with a colorimetric method, as previously described (27). Total GLP-1 was assayed by a commercial RIA as previously described (26). GIP was assayed using a commercial ELISA from Millipore (Millipore, St. Charles, MO) as previously described (18).

Calculations and Statistical Analysis

Fasting insulin and glucose levels for each subject on the different days of the study were the mean of three values taken before glucose administration. The β-cell responses to hyperglycemia were summed as the area under the curve (AUC) above baseline in the 180 min after glucose consumption or isoglycemic glucose infusion (OGTT and Iso-IV), or as the mean concentrations during specific intervals during the glucose clamp; in the clamp-OGTT, 60–90 min was used to reflect the effects of hyperglycemia alone and 100–270 min as the effects of hyperglycemia plus incretins. Insulin sensitivity was estimated during the clamp studies using the mean GIR and plasma insulin concentrations from 60 to 90 min of the glucose clamps. HOMA2-IR was determined for all four studies as described [http://www.dtu.ox.ac.uk/homacalculator/index.php (28)].

The incretin effect was calculated using several approaches. The classical method, method 1, was applied using the difference between insulin and c-peptide AUC from the OGTT and iso-IV studies as a % of the OGTT AUC (18, 29).

For method 2, the incretin effect was calculated from the average insulin (or c-peptide) responses during the final 180 min (100–270 min) of both clamp experiments:

(Ave. ins_{[100–270 min]} clamp-OGTT – Ave. ins_{[100–270 min]} clamp)/Ave. ins_{[100–270 min]} clamp-OGTT × 100%.

With method 3, the incretin effect was calculated from the clamp-OGTT only, using the difference between the postprandial (100–270 min) and preprandial (60–90 min) insulin (or c-peptide) response:

(Ave. ins_{[100–270 min]} – Ave. ins_{[60–90 min]} clamp-OGTT)/Ave. ins _{[100–270 min]} clamp-OGTT × 100%

This method allows for an estimate of the incretin effect in a single-day experiment and has been employed in several previously published studies (9, 27).

Method 4 used the data from the clamp-OGTT experiment similar to method 3, but corrected for the increment of constant hyperglycemia by extrapolating the preprandial (60–90 min) insulin levels throughout the 270 min of the clamp (for each time point from 100 to 270 min, an x-value was calculated using the slope and y-intersect from the linear regression line for the timepoints 60–90 min):

(Ave. ins_{[100–270 min]} – Ave. ins_{[100–270 min extrapol.]} clamp-OGTT)/Ave. ins _{[100–270 min]} clamp-OGTT × 100%

Finally, methods 2–4 were used to calculate the incretin effect based on the results of postprandial insulin and c-peptide levels only during the initial 60 min after glucose ingestion [100–150 min] (methods 2–4 short).

These values for the incretin effect were compared using ANOVA and Tukey’s test for multiple comparisons as appropriate. The normal distribution of the calculated incretin effects was confirmed by KS normality test. Paired observations (incretin effect calculated by long vs. short method) were compared using a paired t test. Slopes of insulin and c-peptide increments were calculated from 60 to 90 min of the clamps to assess the known effect of continuous hyperglycemic glucose infusions on insulin secretion (30). The area under the curve (AUC) above the baseline was calculated using the trapezoidal method. Data are presented as means ± SE unless otherwise noted. For statistical analysis and plotting of graphs, GraphPad Prism 5 (GraphPad Software, San Diego, CA) was used.

RESULTS

Fasting parameters of the subjects at the outset of the four studies are shown in Table 1. Fasting glucose, insulin, and c-peptide did not differ on the 4 days of study, and all subjects had fasting glucose values < 5.1 mM. Insulin sensitivity determined during the glucose clamp did not differ between the two clamp studies (clamp-OGTT 3.38 ± 0.43 and clamp alone 3.61 ± 0.53 mg/kg/min × nM⁻¹; r = 0.73, P = 0.0176 between the two studies), but varied ∼fourfold among the subjects (range 1.33–6.50 mg/kg/min × nM⁻¹). HOMA2-IR did not differ across the 4 days of study.

Table 1.

Demographic and anthropometric data as well as fasting glucose metabolism of the study cohort

	OGTT	Iso-IV	Clamp/OGTT	Clamp	P Value
Age, yr	33.3 ± 4.1
BMI, kg/m²	26.0 ± 1.6
HbA1c, %	5.0 ± 0.1
Fasting glucose, mM	4.41 ± 0.10	4.36 ± 0.07	4.51 ± 0.10	4.45 ± 0.07	0.4377
Fasting insulin, pM	46 ± 8.5	42 ± 5.9	36 ± 5.6	45 ± 6.3	0.1310
Fasting c-peptide, nM	0.51 ± 0.05	0.49 ± 0.05	0.47 ± 0.04	0.53 ± 0.06	0.8860
GIR/insulin, mg/kg/min × pM−1			30.6 ± 5.2	25.9 ± 3.3	0.3772
HOMA2-IR	0.83 ± 0.15	0.75 ± 0.10	0.65 ± 0.10	0.81 ± 0.11	0.1428

Data are presented as means ± SE; N = 10. Parameters of fasting glucose metabolism were not statistically different on the four separate study days. P values were determined by one-way ANOVA for repeated measures or a paired t test as appropriate. GIR, glucose infusion rate; OGTT, oral glucose tolerance test.

All subjects had normal oral glucose tolerance, with no glucose value at 120 min greater than 6.8 mM. Glycemia during the iso-IV infusion was comparable to the levels after the OGTT, with a coefficient of variation (CV) of 5.1 ± 1.3% in the difference between paired values, and there were no significant differences glycemic AUC values(Fig. 1A). Insulin secretion, reflected by either insulin (Fig. 1B) or c-peptide responses (Fig. 1C), was significantly greater during the OGTT, and the calculated incretin effects (method 1) were 60 ± 6% with insulin AUC and 49 ± 5% using c-peptide AUC (Table 2).

Table 2.

Glucose, insulin, c-peptide, and incretin hormones on the four different study days

	OGTT	Iso-IV	Clamp/OGTT	Clamp	P Value
Glucose AUC, mmol/L × min	302.7 ± 46.0	330.5 ± 39.5			0.2871
Insulin AUC, nmol/L × min	40.9 ± 9.1	15.7 ± 3.4			0.0029
C-peptide AUC, nmol/L × min	244.1 ± 24.8	124.9 ± 14.8			<0.0001
Ave. glucose, mM 60–90 min			10.2 ± 0.21	10.1 ± 0.16	0.1814
Ave. glucose, mM 100–270 min			10.2 ± 0.14	10.1 ± 014	0.5349
Ave. insulin, pM 60–90 min			258 ± 52	312 ± 48	0.0943
Ave. insulin, pM 100–270 min			1,317 ± 255	597 ± 124	0.0078
Ave. insulin, pM 100–150 min			1,114 ± 256	428 ± 71	0.0139
Ave. c-peptide, nM 60–90 min			1.97 ± 0.22	2.08 ± 0.22	0.3885
Ave. c-peptide, nM 100–270 min			5.06 ± 0.55	3.17 ± 0.36	0.0007
Ave. c-peptide, nM 100–150 min			4.51 ± 0.56	2.79 ± 0.30	0.0019
GLP-1 AUC, pmol/L × 120 min	365 ± 110		374 ± 167		0.8785
GIP AUC, pmol/L × 180 min	7,352 ± 1,666		7,338 ± 1,382		0.9375

Data are presented as means ± SE. Glucose AUC and glucose average did not differ between the OGTT vs. Iso-IV and clamp/OGTT vs. Clamp, respectively (equal glycemic stimulus). Insulin and c-peptide AUC during the OGTT and average insulin and c-peptide during the clamp/OGTT were significantly higher compared to the corresponding iv-glucose study. P values refer to paired t tests for equal variances. AUC, area under the curve; GIR, glucose infusion rate; OGTT, oral glucose tolerance test.

During the clamp-OGTT, blood glucose levels reached a plateau from 60 to 270 min of 10.2 ± 0.14 mM with a mean CV for the clamp values of the 10 subjects of 7.4% (Fig. 2A). Glycemia was not different before and after the oral glucose load (60–90 vs. 100–270 min, P = 0.6654). Plasma insulin levels increased from 36 ± 5.6 pM in the fasting state to 258 ± 52 pM in response to hyperglycemia alone (60–90 min, Fig. 2B). Following ingestion of the oral glucose solution, mean insulin levels significantly increased another ∼sixfold to 1,317 ± 255 pM. C-peptide concentrations followed a similar pattern during the fasting, 60–90 and 100–270 min periods (0.47 ± 0.04, 1.97 ± 0.22, and 5.06 ± 0.55 nM; Fig. 2C). Plasma insulin and c-peptide increased significantly at each stage of the clamp: fasting, IV hyperglycemia, and IV hyperglycemia plus OGTT (P < 0.001). Blood glucose during the clamp alone was comparable to the clamp-OGTT, reaching a mean of 10.1 ± 0.14 mM (P = 0.3731 vs. clamp-OGTT) with a CV of 6.8% (Fig. 2A). Glucose infusion rates (GIR) and grams of glucose infused were similar during the clamp-OGTT and the clamp experiment (326 ± 28 vs. 313 ± 27 g, P = 0.62, Fig. 2A). Insulin and c-peptide values increased similarly to the clamp-OGTT through 90 min and continued to rise gradually through the remainder of the experiment (Fig. 2, B and C). The incretin effect calculated from the difference in average insulin and c-peptide responses from 100 to 270 min of the two clamps was 59% ± 6% and 41% ± 3% (method 2), respectively, and did not differ significantly from the OGTT/Iso-IV studies (method 1).

The incretin effect calculated using insulin and c-peptide values during the clamp-OGTT in relation to the preprandial plasma concentrations was 79% ± 2% and 61% ± 2%, respectively (method 3). The calculated incretin effect based on insulin trended to be higher than with methods 1, 2, and 4 and was significantly greater based on c-peptide compared to methods 2 and 4 (P < 0.01).

The effect of continuous hyperglycemia on β-cell function increased steadily during the clamp alone, with insulin and c-peptide levels approaching 70% of the clamp-OGTT levels by 270 min. We calculated the slope of the continuously rising insulin levels from 60 to 90 min in both clamp experiments and compared these to the slope of the insulin response to hyperglycemia alone in the clamp experiment from 60 to 270 min. The latter showed a steady and linear rise of plasma insulin concentrations with continuous hyperglycemia over time (deviation from 0 P < 0.0001; deviation from linearity P = 0.8671). The slopes were 2.766 ± 2.101 for the clamp (60–90 min), 2.802 ± 0.4664 for the clamp (60–270 min), and 2.475 ± 2.387 for the clamp-OGTT (60–90 min) and did not differ between these calculations. The plasma insulin concentrations from 60 to 270 min of both clamp experiments and the extrapolated linear regression lines from 60–90 min are depicted in Fig. 3.

Figure 3.

Plasma insulin levels during the clamp-OGTT (solid circles) and the clamp (open circles) from 60 to 270 min of the clamp. Linear regression lines are extrapolated from the plasma insulin concentrations from 60 to 90 min of the clamp-OGTT (straight line) and the clamp (dotted line). Both lines are not statistically different from the linear regression of plasma insulin concentrations from 60 to 270 min of the clamp experiment. OGTT, oral glucose tolerance test.

Hence, we calculated the incretin effect using the extrapolated preprandial insulin levels for each clamp-OGTT, correcting the incretin effect in method 3 for the steady linear increase in insulin secretion during continued hyperglycemia (method 4). This resulted in incretin effects of 58% ± 10% and 38% ± 5% for insulin and c-peptide, respectively. Both were not statistically different from the incretin effects calculated by methods 1 and 2 but significantly lower than c-peptide calculations based on method 3 (P < 0.01). The pattern of insulin and c-peptide plasma levels during the clamp-OGTT experiments with a steep increase of both insulin and c-peptide within the first 60 min of glucose ingestion but no clear tendency to return to preprandial levels prompted us to investigate a shorter interval. Calculations of the incretin effect were performed for methods 2, 3, and 4 considering only the first 60 min [100–150] after glucose ingestion. Although methods 2 and 4 resulted in virtually identical incretin effects with the short protocol (P > 0.8) compared to the longer version, short method 3 produced a significantly lower incretin effect compared to its long version (P < 0.001) (Table 3).

Table 3.

Differences in the calculated incretin effects and parameters of the experimental procedure

	AUC (Ins/cp) OGTT vs. Iso-IV (Method 1)	Ave. (Ins/cp) Clamp/OGTT vs. Clamp (Method 2)	Ave. (Ins/cp) Clamp/OGTT Pre- vs. Postprandial (Method 3)	Ave. (Ins/cp) Clamp/OGTT Interpol. vs. postprandial (Method 4)	Method 2—Short	Method 3—Short	Method 4—Short	IE across Methods 1–4 Compared by One-Way ANOVA [P Value]
IE insulin, %	60.3 ± 19.3	59.0 ± 16.6	79.1 ± 6.7	58.1 ± 30.7	53.8 ± 25.6	72.8 ± 10.1#	63.9 ± 19.1	0.0845
IE c-peptide, %	48.6 ± 15.6	40.8 ± 9.8 **	60.9 ± 5.8	37.9 ± 12.8 **	40.2 ± 13.1	55.4 ± 9.1#	39.0 ± 16.7	0.0007
Range IE insulin, %	16–86	35–80	69–92	6–99	3–85	58–89	31–83
Range IE c-peptide, %	14–71	28–53	53–70	24–61	18–60	38–67	6–58
Number of separate study days	2	2	1	1	2	1	1
Test duration, min	180	270	270	270	150	150	150

Data are presented as means (SD). The incretin effect (IE) describes the % increment of insulin and c-peptide in the circulation after oral glucose uptake that exceeds the response to glucose per se and was calculated by the different methods as described in the METHODS section. Although the IE calculated with methods 1, 2, and 4 do not differ, estimates of the IE with method 3 trend toward higher results vs. methods 1, 2, and 4 for insulin and are significantly higher than methods 2 and 4 for c-peptide (one-way ANOVA). **P < 0.01 vs. method 3 (Tukey’s multiple comparison test). The calculated IE was normally distributed across all methods (KS-normality test). The clamp methods were additionally analyzed with a shorter protocol looking at postprandial changes within 60 min of oral glucose ingestion (100–150 min). Long (270 min) and short (150 min) methods were compared with paired t tests for each method. While shortening of methods 2 and 4 did not result in significantly different IE (P > 0.33), the short version of method 3 resulted in significantly lower IE for both insulin and c-peptide (#P < 0.01). AUC, area under the curve; Ave., average; cp, c-peptide; Ins, insulin.

Gastric emptying was measured indirectly by d-xylose absorption. d-xylose AUC was significantly higher during the OGTT compared to the clamp-OGTT (68.7 ± 5.6 vs. 40.7 ± 3.9 mmol/L × min, P = 0.0028), indicating incomplete gastric emptying during the clamp-OGTT in contrast to the OGTT alone. However, initial gastric emptying was comparable during the first hour with similar time-to-peak duration (86.7 ± 3.3 vs. 75.6 ± 9.6 min, P = 0.4124) and slope of d-xylose appearance during the first 60 min (7.58 × 10⁻³±0.70 × 10⁻³ vs. 7.95 × 10⁻³±0.99 × 10⁻³, P = 0.8007) (Fig. 4).

Figure 4.

Gastric emptying measured by postprandial plasma d-xylose concentrations during the OGTT (full circle) and the clamp-OGTT (open circle) (A), postprandial AUC above baseline (B), time-to-peak (C), and slope from 0 to 60 min (D). Although total AUC was significantly lower during the clamp-OGTT compared to the OGTT alone (P = 0.0028) (B), time-to-peak (P = 0.4124) (C) and slope during the first 60 min (P = 0.8007) (D) did not differ significantly between the studies. Data are means ± SE. **P < 0.01. AUC, area under the curve; OGTT, oral glucose tolerance test.

Postprandial plasma concentrations of the incretins GLP-1 and GIP did not differ between the OGTT and the clamp-OGTT with very similar AUCs for GLP-1 and GIP (Table 2).

DISCUSSION

A role for the GI tract in the regulation of prandial glycemia, which was refined as the incretin concept, has a long history in physiology preceding even the discovery of insulin (31). The incretin effect is fundamental in understanding the normal disposition of ingested carbohydrates and has provided important insights into the pathology of diabetes and other disorders of prandial metabolism. Because of the importance of accurately determining the incretin effect in experimental work, we sought to compare the two major methods that have been used to measure this entero-insular connection. The results of the studies described herein indicate that the relative augmentation of insulin secretion by ingestion of glucose in healthy subjects is comparable when determined either by the classic 2-day method and a 1-day clamp procedure. Moreover, with some simple accounting for the inherent effects of continuous hyperglycemia on insulin secretion, the burden for test subjects and patients as well as resources can be reduced. These observations provide guidance for further studies of the incretin effect and add some physiological insight into this mechanism of glucose regulation.

The incretin effect, as it is defined today—the increased β-cell secretion in response to glucose absorbed from the gut compared to an equi-glycemia IV infusion of glucose—emerged shortly after the development of assays for insulin (32, 33). Although early studies reported estimates that the incretin effect accounted for ∼ 50% of prandial insulin secretion (21), truly quantitative approaches to this measure with well-matched conditions were developed only later. Elahi and colleagues (23) reported a clamp-OGTT method to demonstrate an ∼2.5-fold augmentation of insulin secretion in lean, obese, and diabetic subjects after oral glucose consumption during IV-induced hyperglycemia, which they attributed to GIP. This was the first study to determine the effect of the route of glucose administration on insulin secretion under conditions of carefully controlled glycemia. These authors accounted for the known effect of continuous IV glucose infusions to cause a steady rise of plasma insulin using a method of extrapolation similar to method 4 used to analyze our clamp-OGTT results. Soon after this report, in a now-classic paper, Nauck and coworkers (12) used isoglycemic infusions to precisely match the blood glucose levels of an OGTT in 2 separate days of study and computed the incretin effect using insulin secretion rates derived from measures of c-peptide. This method allowed an accounting of the major parameters contributing to insulin secretion in the 2 days of study, and because the design and results provide such a clear description of the incretin effect, this has become the most accepted protocol. Although in the following years, both approaches have been used to determine the incretin effect, no study directly compared those two methods. We have applied both experimental approaches in our within-subjects design and extended the original clamp-OGTT protocol by extending the periods of sampling and adding a glucose-alone clamp.

Our group of volunteers was young, healthy, and based on simple parameters—fasting glucose and insulin and HOMA estimates of insulin sensitivity, metabolically stable on the 4 days they participated in experiments. Although the studies with targeted glucose infusions were relatively labor intensive, the targets were readily achieved, and glycemia was well-matched during the paired OGTT and isoglycemic glucose infusion, and clamp-OGTT and clamp-alone studies. Using comparisons of β-cell secretion from the two pairs of studies with and without glucose ingestion gave comparable estimates of the incretin effect (methods 1 and 2, Table 3). We previously reported that the magnitude of the incretin effect was similar when estimated by the clamp-OGTT method at either 8 or 10 mM glucose (25). The current results extend this observation by demonstrating that the pattern of glycemia during which the incretin effect is measured also does not affect its computation, so long as the glycemic stimulus is matched, with and without glucose ingestion. This is in contrast to the effect of the varying oral glucose loads, in which greater glucose ingestion causes a proportionately greater incretin effect (12). It seems reasonable to conclude that in healthy subjects, the size of the incretin effect is dictated by the amount of enteral stimulus and is independent of the glycemic level, consistent with a system adapted to maintain prandial glucose relatively constant meal-to-meal.

The measurement of the incretin effect using paired data from the clamp-OGTT plus clamp-alone has little advantage over the classic OGTT/Iso-IV approach, both requiring 2 days. Moreover, the clamps are more strenuous and time consuming for test subjects and investigators. Therefore, we explored different analytic approaches to compute the incretin effect from the single clamp-OGTT procedure (methods 3 and 4). We used the mean insulin or c-peptide values from time 60 to 90, a period where blood glucose steady state was easily attained as an estimate of insulin secretion driven by the glycemic stimulus of the clamp. The incretin plus glucose component was taken as the mean insulin or c-peptide across the full 3 h of the clamp after glucose ingestion. Method 3 gave an overestimate of the incretin effect relative to the 2-day methods (methods 1 and 2), likely because it does not account for the effect of continuous hyperglycemia to enhance insulin secretion (30), an effect demonstrated in our clamp-alone experiment. This supposition is supported by method 4, in which we extrapolated the rise of plasma insulin or c-peptide to correct β-cell output during an extended glucose clamp, similar to the approach of Elahi et al. (23). Determination of the incretin effect using this adjustment gave mean values comparable to the OGTT/Iso-IV infusion method with either insulin or c-peptide. Adjustment for continuous hyperglycemia was accurate in this study since the insulin levels from 60 to 90 min of both clamps were almost identical to the slope of the insulin response to 60–270 min of constant hyperglycemia (Fig. 3). Thus, a single-day method with an OGTT during a glucose clamp seems to be both a feasible and accurate means of estimating the incretin effect, while the magnitude of the reported incretin effect in previous studies employing method 3 was likely overestimated.

The clamp-OGTT method for generating the incretin effect has several potential advantages, the most obvious of which is that it is a single-day experiment. This is particularly relevant for studies that compare an intervention with a control study where the clamp-OGTT can reduce study days from 4 to 2. In addition, the use of a hyperglycemic clamp allows the determination of first-phase insulin secretion (23), which has utility in studies of subjects with diabetes. Although not as accurate as euglycemic hyperinsulinemic clamps for measuring insulin sensitivity, GIR during steady-state hyperglycemia (i.e., 60–90 min) allows a fairly good direct measurement of insulin sensitivity (34). The ability to keep the blood glucose level fixed using the clamp-OGTT allows measures of the relative incretin effect during interventions that alter glucose tolerance. We and others have used the clamp-OGTT technique with the specific GLP-1 receptor antagonist exendin-9 (9, 27, 35). This peptide is known to impair oral glucose tolerance, resulting in a greater glucose AUC compared to saline (8). In this setting, matched OGTTs and isoglycemic glucose infusions would be confounded by differences in glycemia. Finally, using the clamp-OGTT incretin secretion can be stimulated with a mixed nutrient meal instead of oral glucose (36).

There are also some drawbacks and unresolved questions regarding the clamp-OGTT technique to determine the incretin effect. For one thing, most of the published studies addressing the incretin effect have been done with the OGTT/isoglycemic infusion approach. For another, constant hyperglycemia is not truly physiological compared to an oral glucose tolerance test and may impact the results. Moreover, the β-cell response to steady hyperglycemia may differ in subject groups, e.g., diabetic and nondiabetic, confounding comparisons. Lastly, it is known that hyperglycemia reduces the rate of gastric emptying and may result in delayed enteral delivery of glucose or other nutrients that stimulate incretin release. In fact, the indirect measurement of gastric emptying by d-xylose absorption does suggest incomplete gastric emptying over the period of the clamp procedure. However, the initial emptying rate and release of incretins were comparable in this study so gastric emptying does not seem to be a major confounding factor of our results.

Our study has several limitations worthy of discussion. Due to a demanding study protocol (4 experimental days, clamps over several hours), the cohort is small with limited statistical power to compare some parameters. Also, we only tested young and healthy adults and did not confirm our results in obese subjects or patients with impaired glucose metabolism (18, 37). Given that the incretin effect is impaired in many people with diabetes, a direct extension of these results should be made to that group with caution. Lastly, more work is required to establish that shorter protocols (methods 2–4 short) are amenable to correction for the response to continuous hyperglycemia.

In conclusion, we have demonstrated in a proof-of-concept study that the incretin effect can be reliably measured by two different experimental protocols. The clamp procedure using pre- and postprandial insulin levels from the same day to calculate the incretin effect is technically challenging but helps save time and resources and reduces day-to-day variability in parameters of glucose tolerance in clinical studies. With correction for the effects of continuous hyperglycemia on insulin release, the clamp-OGTT is as reliable and accurate as the OGTT/isoglycemic infusion method for determining the incretin effect in healthy subjects. It is plausible that the clamp-OGTT could be truncated to a shorter protocol that is more convenient. These findings should be extended to other cohorts with known defects of the incretin effect like patients with type 2 diabetes.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This research was supported by a Merit Award from the Department of Veterans Affairs and US PHS Grant R01 57900, both to D.A.D. This study was also supported by the National Institutes of Health (NIH) Clinical and Translational Science Award (CTSA) program, Grant 8 UL1 TR000077-05 to the University of Cincinnati and the Cincinnati Children’s Hospital.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.A.A. and D.A.D. conceived and designed research; B.A.A. performed experiments; B.A.A. and D.A.D. analyzed data; B.A.A. and D.A.D. interpreted results of experiments; B.A.A. prepared figures; B.A.A. drafted manuscript; B.A.A. and D.A.D. edited and revised manuscript; B.A.A. and D.A.D. approved final version of manuscript.