Predictors of Incretin Concentrations in Subjects With Normal, Impaired, and Diabetic Glucose Tolerance

The study protocol was approved by the ethics committee of the medical faculty of the Ruhr-University, Bochum, Germany, before the experiments (registration no. 2013). Written informed consent was obtained from all participants.

A total of 48 subjects participated in the study. Among those, 17 patients had type 2 diabetes (World Health Organization criteria) of relatively short duration (3.2 ± 2.8 years), 17 had impaired glucose tolerance (IGT), and 14 had a normal oral glucose tolerance. Type 2 diabetes was previously known in 16 of the patients, whereas in one case, the diagnosis of type 2 diabetes was made on occasion of the oral glucose tolerance test. Six patients with type 2 diabetes were treated with metformin, and the other patients were on a dietary regimen. Subject characteristics are presented in Table 1.

At a screening visit, blood was drawn from all participants in the fasting state for measurements of standard hematological and clinical chemistry parameters, and a general clinical examination was performed. Subjects with anemia (hemoglobin <12 g/dl), elevation in liver enzymes (alanin-amino-transferase, aspartat-amino-transferase, alkaline-phosphatase, and γ-glutamyl-transferase) to higher activities than double the respective normal value, or elevated creatinine concentrations (>1.5 mg/dl) were excluded. Body height and weight were determined and waist and hip circumference were measured to calculate BMI and the waist-to-hip ratio, respectively. Blood pressure was determined according to the Riva-Rocci method (29). A vibration perception threshold performed on both medial malleoli revealed no major impairment in pallesthesia in the patients with diabetes (5.4 ± 1.6, 5.9 ± 1.2, and 6.1 ± 1.6 times 1/8 in patients with type 2 diabetes, IGT subjects, and control subjects, respectively; P = 0.43).

If subjects met the inclusion criteria, they were studied on two occasions: 1) In an oral glucose challenge, 75 g glucose (O.G.T.; Roche Diagnostics, Mannheim, Germany) was ingested in the overnight fasting state within 5 min. Blood samples were drawn twice in the fasting state and at 30-min intervals over 240 min afterward. 2) In a mixed meal challenge, a large continental breakfast (two European bread rolls, 20 g butter, 40 g gouda cheese, 30 g jam, one egg, 150 g yogurt [3.5% fat content], and 200 ml tea) was ingested over 15 min. The total caloric content of the test meal was 820 kcal (107 kcal from protein, 353 kcal from fat, and 360 kcal from carbohydrates). Blood samples were drawn twice in the fasting state and at 30-min intervals over 240 min afterward. All antidiabetes treatment was withdrawn at least 2 days before study commencement.

Both tests were performed in the morning after an overnight fast with subjects in a supine position throughout the experiments. Both ear lobes were made hyperemic using Finalgon (4 mg/g nonivamid and 25 mg/g nicoboxil). The experiments were started by the ingestion of 1) the oral glucose load or 2) the mixed test meal, and venous blood samples were drawn at t = −30, 0, 30, 60, 90, 120, 150, 180, 210, and 240 min. In addition, capillary blood samples (∼100 μl) were added to NaF (Microvette CB 300; Sarstedt, Nümbrecht, Germany) for the immediate measurement of glucose.

Glucose concentrations were measured in capillary blood samples using a glucose oxidase method with a Glucose Analyser 2 (Beckman Instruments, Munich, Germany) as previously described (10). Insulin was measured as described previously (10) using an insulin microparticle enzyme immunoassay (IMx Insulin; Abbott Laboratories, Wiesbaden, Germany). Intra-assay coefficient of variation was ∼4%.

C-peptide was measured as described previously (10) using an enzyme-linked immunoabsorbent assay from DAKO (Cambridgeshire, U.K.). Intra-assay coefficients of variation were 3.3–5.7%; interassay variation was 4.6–5.7%. Human insulin and C-peptide were used as standards.

Total GLP-1 concentrations were measured using a radioimmunoassay (antiserum no. 89390), which is specific for the COOH-terminal end of the GLP-1 molecule and reacts equally with intact GLP-1 and the primary (NH₂-terminally truncated) metabolite as described previously (27).

Total GIP was measured, as described previously, using the COOH-terminally directed antiserum R65 (30), which reacts fully with intact GIP and the NH₂-terminally truncated metabolite.

Glucagon was measured by a radioimmunoassay using antibody no. 4305 in ethanol-extracted plasma, as described previously (31). The detection limit was <1 pmol/l. Intra-assay coefficients of variation were <5.0%, and interassay coefficients of variation were 15%.

Triglycerides were measured using an enzymatic color test system (Olympus system reagent triglyceride OSR 6133). Free fatty acids (FFAs) were determined using reagents from Wako Chemicals (Neuss, Germany) by spectrophotometric analysis.

Results are reported as mean ± SE. All statistical calculations were carried out using repeated-measures ANOVA using Statistica version 5.0 (Statsoft Europe, Hamburg, Germany). Values at single time points were compared by one-way ANOVA followed Duncan's post hoc test. A two-sided P value <0.05 was taken to indicate significant differences. Integrated incremental plasma concentrations of insulin, C-peptide, GIP, GLP-1, and triglycerides were calculated according to the trapezoidal rule (baseline subtracted). Integrated glucagon and FFA levels were calculated as the negative (decremental) areas under the curve. Insulin resistance was calculated according to the homeostasis model assessment (HOMA) model (32). Linear regression analyses were performed using GraphPad Prism, version 4.0. Multivariate regression analyses regarding the concentrations of GIP and GLP-1, taking into account the factors age, sex, body weight, glucose (basal and positive area under the curve [AUC]), glucagon (basal and negative AUC), insulin (basal and positive AUC), A1C, and FFAs (basal and negative AUC), were carried out using Statistica version 5.0.

As expected, glucose concentrations were higher in the subjects with IGT and type 2 diabetes both after the oral glucose load and after the mixed meal ingestion (P < 0.0001; Fig. 1). Likewise, insulin and C-peptide concentrations were higher in IGT subjects and patients with diabetes compared with control subjects, indicating a higher degree of insulin resistance (Fig. 1). This impression was confirmed by HOMA analysis, revealing higher levels of insulin resistance in the patients with type 2 diabetes (5.0 ± 0.9) and IGT subjects (3.0 ± 0.5) compared with control subjects (1.8 ± 0.3; P = 0.006).

Glucagon levels were significantly lowered after oral glucose and mixed meal ingestion (P < 0.001; Fig. 2). The time course of glucagon concentrations was significantly different between the groups after the test meal (P < 0.001), with higher glucagon levels in patients with type 2 diabetes during the first 90 min of the test. There were no differences in the time course of glucagon levels between the groups after the oral glucose load, even though glucagon concentrations tended to be higher in the diabetic patients during the first 90 min after glucose ingestion as well.

FFA concentrations were higher in IGT subjects and patients with type 2 diabetes during both experiments (P < 0.0001). These differences were most pronounced in the fasting state, whereas FFA levels were even lower in the patients toward the end of the experiments (significant only after oral glucose ingestion). As expected, triglyceride levels remained unchanged after oral glucose ingestion (P = 0.06) and increased significantly after the mixed test meal (P < 0.0001). There was a trend toward higher triglyceride levels in the patients with type 2 diabetes (Fig. 2).

Plasma concentrations of GIP and GLP-1 increased significantly after oral glucose and meal ingestion (P < 0.0001). Peak levels of GIP were reached 90 min after the oral glucose load and 60 min after mixed meal ingestion. The maximum GLP-1 concentrations were detected at t = 30 and 90 min, respectively (Fig. 3). There were no differences in the plasma concentrations of GIP between the groups after oral glucose ingestion (P = 0.98). In contrast, the initial GIP response 30 min after meal ingestion was significantly increased in the patients with type 2 diabetes, whereas the IGT subjects exhibited higher GIP levels from t = 180 to 240 min after the test meal (P < 0.001). However, despite these differences at individual time points, there was no difference in integrated GIP levels after the meal (P = 0.33).

GLP-1 levels were not different between the groups both after oral glucose and meal ingestion. GIP responses were 186 ± 17% higher after mixed meal ingestion than after the oral glucose load (P < 0.0001; Fig. 4). In contrast, there were no differences in GLP-1 plasma concentrations between the experiments with oral glucose or mixed meal ingestion (P = 0.06; Fig. 4). The patterns of GIP and GLP-1 levels were not significantly different when the six patients pretreated with metformin were excluded from the analyses (details not shown).

There was a strong negative association between fasting glucagon levels and GLP-1 levels, as assessed by the incremental AUCs, after oral glucose and meal ingestion (Table 2; Fig. 5). In contrast, the integrated decremental glucagon levels after glucose or test meal ingestion were positively associated with the increases in GLP-1 levels. Conversely, there was a positive relationship between fasting FFA levels and subsequent GLP-1 concentrations, whereas the integrated decremental FFA levels were inversely related to the integrated GLP-1 levels. There also was a significant positive relationship between GLP-1 levels and increasing age and a negative association with higher BMI levels. These associations were stronger after oral glucose ingestion than after mixed meal ingestion (Table 2).

GIP levels were positively related to fasting FFA levels and integrated decremental glucagon concentrations, whereas integrated decremental FFA levels were negatively associated with GIP concentrations (Table 3). There also was a significant positive relationship between age and GIP levels.

GLP-1 plasma concentrations were significantly higher in female than in male subjects after oral glucose (3,491 ± 491 vs. 2,061 ± 296 pmol · l⁻¹ · min⁻¹; P = 0.04) and after mixed meal ingestion (4,153 ± 485 vs. 2,639 ± 419 pmol · l⁻¹ · min⁻¹; P = 0.037), whereas GIP levels were similar in both groups (8,846 ± 618 vs. 8,090 ± 865 pmol · l⁻¹ · min⁻¹, respectively, after oral glucose ingestion; P = 0.47; 22,951 ± 1,153 vs. 19,327 ± 1,524 pmol · l⁻¹ · min⁻¹, respectively, after mixed meal ingestion; P = 0.063). GLP-1 concentrations were similar in diabetic patients treated with metformin or with a dietary regimen (details not shown).

There was no detectable association between the measures of glucose control (fasting and 120-min glucose, A1C), insulin secretion (HOMA of β-cell function), or insulin sensitivity (HOMA of insulin resistance) and the plasma concentrations of GIP or GLP-1 (Tables 1 and 2).

In a multivariate regression analysis, the GLP-1 levels after the oral glucose load were positively related to age (P = 0.015) and the decremental AUC_glucagon (P = 0.0012), whereas no significant association with GIP levels was detected.

The present studies were undertaken to examine whether defects in the plasma levels of incretin hormones in patients with type 2 diabetes and IGT were more prominent after an oral glucose load than after mixed meal ingestion. To our own surprise, we did not find any alterations in GLP-1 concentrations after oral glucose or mixed meal ingestion in the individuals with IGT and type 2 diabetes. In contrast, GIP levels were similar in all groups after the oral glucose load and even slightly higher in subjects with IGT and type 2 diabetes after the meal test.

The present results are at variance with some (17,18) but not all (20) previous studies reporting incretin hormone levels in patients with type 2 diabetes. Toft-Nielsen et al. (17) reported ∼20 and ∼30% lower postprandial GLP-1 levels in IGT subjects and patients with type 2 diabetes compared with normal oral glucose–tolerant subjects, respectively. In line with these data, Vilsboll et al. (18) found not only total but also intact GLP-1 levels to be reduced in patients with type 2 diabetes. However, in subsequent studies, the same group of investigators failed to detect such differences in GLP-1 levels in another group of patients with type 2 diabetes (20). In the present studies, only total GIP and GLP-1 levels were measured, meaning that differences in the degradation of incretin hormones cannot be excluded. In addition to these studies in patients with overt diabetes, we and others have found similar GIP and GLP-1 levels in high-risk groups, such as first-degree relatives of patients with diabetes and women with prior gestational diabetes (26–28). The reasons for the dissimilar results in different cohorts of patients are difficult to explain, but because in all of these studies, incretin levels were measured in the same laboratory, technical issues are rather unlikely. It is therefore important to compare the subject characteristics in more detail. Thus, in the present studies, patients with a rather short diabetes duration (3.2 ± 2.8 years) and in relatively good glycemic control (A1C 6.8 ± 0.9%) were examined. In contrast, the patients studied by Vilsboll et al. (18) and Toft-Nielsen et al. (17) had a longer diabetes duration and exhibited higher A1C levels (17,18). It is therefore possible that defects in GLP-1 secretion develop later during the pathogenesis of type 2 diabetes. In this context, another factor with a potential impact on postprandial GLP-1 levels is the velocity of gastric emptying (33,34). Thus, any deceleration of gastric emptying might blunt the subsequent incretin responses (35). In the present studies, patients with a relatively short duration of diabetes and no signs of diabetic neuropathy were examined. However, because gastric emptying is significantly delayed in some patients with long-standing type 2 diabetes (36), it is possible that the impairment in postprandial GLP-1 concentrations reported earlier was partly driven by a delay in gastric motility.

Another possible factor that might impact on GLP-1 secretion in diabetic patients is the presence of hyperglucagonemia. In fact, in this as well as in previous studies, high glucagon levels were found to be associated with lower GLP-1 concentrations (37). This might explain why in the present study with rather modest differences in glucagon levels between the groups, GLP-1 plasma concentrations were similar in patients with and without diabetes, whereas previous studies examining patients with a longer diabetes duration and with more pronounced differences in glucagon secretion had found a significant impairment in GLP-1 levels in diabetic patients (17). Thus, it seems possible that in healthy subjects GLP-1 secretion is tonically inhibited by glucagon and that the postprandial decline in α-cell secretion, along with other nutrient signals, allows for the rise in GLP-1 levels. In patients with long-standing type 2 diabetes, the lack of glucagon suppression by glucose or meal ingestion might lead to an impairment in GLP-1 secretion (38,39). However, it is important to emphasize that the relationship between glucagon and GLP-1 levels observed herein was purely associative and does not allow for safe conclusions regarding a direct interaction between both hormones. Thus, hyperglucagonemia and reduced GLP-1 levels might also be independent conditions in patients with type 2 diabetes without a causal relationship to each other. Therefore, it appears worthwhile to directly examine the effects of elevated glucagon levels on GLP-1 secretion in more detail.

An alternative explanation is that the impairment in GLP-1 levels can only be provoked by certain test meals, depending of the respective caloric content and nutrient composition. In the present study, GLP-1 plasma concentrations in patients with diabetes were normal after an oral glucose load and after a mixed test meal representative of a typical European continental breakfast. However, it cannot be excluded that incretin response in patients with diabetes and control subjects would be different after the ingestion of a test meal containing rather high fat. Finally, it cannot be excluded that in our present studies, minor differences in GLP-1 concentrations at certain time points were overlooked because of the 30-min sampling intervals.

On a cautionary note, 6 of the 17 patients with type 2 diabetes in this study were pretreated with metformin, and even though metformin was withdrawn 2 days before the experiments, it cannot be fully ruled out that GLP-1 concentrations were affected by this treatment (40). However, the patterns of incretin concentrations were similar when these six patients were excluded from the analyses, and in this relatively small group of subjects, there were no significant differences in GLP-1 levels between patients treated with or without metformin.

Despite the somewhat discrepant results in different studies, the present data showing normal incretin responses in a group of patients with type 2 diabetes suggest that defects in GLP-1 secretion are probably not an important factor in the early pathogenesis of the disease. Along these lines, it is worth noting that the differences in GLP-1 levels reported in the prior studies were only apparent in the late postprandial period (∼120–240 min after meal ingestion) (17,18), whereas defects in insulin secretion typically occur in the early postprandial phase (41).

Another objective of these studies was to uncover endogenous predictors of GLP-1 levels. Using linear regression analysis, we identified fasting glucagon levels as a strong negative predictor of subsequent GLP-1 concentrations after oral glucose and after mixed meal ingestion. However, despite this inverse relationship in the fasting state, there was a strong positive association between GLP-1 levels and postprandial glucagon concentrations. This observation is even more surprising because GLP-1 is known to exert glucagonostatic actions (42,43). Therefore, the question arises: what is the driving force in this relationship? Given the inverse association between glucagon and GLP-1 levels in the fasting state, any stimulation of GLP-1 release by glucagon seems rather unlikely (37). A more plausible explanation is that the positive relationship between postprandial GLP-1 and glucagon levels was indirectly mediated by the secretion of GLP-2 (44). Both GLP-1 and GLP-2 are secreted at equimolar amounts from entero-endocrine l-cells (45), but because of its slower degradation, circulating GLP-2 levels exceed those of GLP-1 by severalfold (46,47). It is therefore likely that the potent glucagonotropic actions of GLP-2 shown previously by our group have supervened the glucagonostatic actions of GLP-1 (44,48). Thus, under physiological conditions, the GLP-1–induced suppression of glucagon secretion might be outweighed by the glucagonotropic actions of GLP-2.

One question directly arising from this hypothesis is why DPP-4 inhibitors act to suppress rather than to increase glucagon levels in patients with type 2 diabetes, despite their actions on intact GIP and GLP-2 levels. Although due to their correlative nature, the present studies cannot provide a direct explanation for this phenomenon, two possible reasons come to mind: First, the affinity of DPP-4 for GLP-1 is much higher than for GLP-2 and GIP, meaning that the impact of DPP-4 inhibition on the intact plasma levels of these hormones is different as well. This may lead to an overproportional rise in intact GLP-1 levels (49). Second, the actions of GIP and GLP-1 on glucagon secretion have been shown to be strictly glucose dependent (42,50,51). It is therefore conceivable that in hyperglycemic patients with type 2 diabetes, the glucagonostatic effects of GLP-1 outweigh the actions of GIP and GLP-2 during DPP-4 inhibitor administration. In contrast, in nondiabetic subjects studied at euglycemia, glucagon release has been shown to be suppressed to an even lesser extent after oral than during intravenous glucose administration (37).

In addition to glucagon, fasting FFA levels and age were identified as positive predictors of GLP-1 concentrations, whereas body weight and integrated FFA levels were inversely related. The present analyses also revealed significantly greater GLP-1 responses in female subjects, consistent with the findings of Toft-Nielesen et al. (17).

Although the ingestion of the mixed test meal elicited a significantly greater response of GIP levels than that elicited by the oral glucose load, GLP-1 levels were not different between both experiments. This suggests that GIP and GLP-1 are not entirely cosecreted and that different factors regulate the secretion of GIP and GLP-1. Because both meals differed with respect to the overall caloric and the nutrient composition, it is likely that the increased GIP concentrations after the mixed meal ingestion were largely due to the additional lipid content, which might have had no additional impact on GLP-1 secretion. An alternative explanation is that GLP-1 secretion was already stimulated to a maximum extent, thereby not allowing for an additional gain in GLP-1 levels with increasing amounts of nutrients ingested. Given the diverging actions of both incretin hormones on glucagon secretion (42,51), gastric emptying (52,53), and lipid homeostasis (16,54), the distinct relative responses of GIP and GLP-1 secretion in response to different test meals might serve as a fine regulator of postprandial metabolism.

In conclusion, the present studies have shown that deteriorations in glucose homeostasis can develop in the absence of any impairments in GIP or GLP-1 levels. This suggests that the impairments in postprandial GLP-1 concentrations previously described in some patients with long-standing type 2 diabetes are likely secondary to other metabolic alterations, such as hyperglucagonemia. The unequal dependency of GIP and GLP-1 responses on meal size and composition implies that both hormones are largely released independently of each other and that their respective secretion is regulated by different factors.

J.J.H. has received an unrestricted grant from the Danish Medical Research Council. J.J.M. has received unrestricted grants from the Ruhr-Universität Bochum (FoRUM Grant F-515-06), Novo Nordisk, and the Deutsche Forschungsgemeinschaft (Grant Me 2096/5-1).

We greatly acknowledge the excellent technical assistance of Elisabeth Frick and Lone Bagger.