Patterns of Plasma Glucagon Dynamics Do Not Match Metabolic Phenotypes in Young Women

Patterns of Plasma Glucagon Dynamics Do Not Match Metabolic Phenotypes in Young Women Abstract Context The role of hyperglucagonemia in type 2 diabetes is still debated. Objective We analyzed glucagon dynamics during oral glucose tolerance tests (oGTTs) in young women with one out of three metabolic phenotypes: healthy control (normoglycemic after a normoglycemic pregnancy), normoglycemic high-risk (normoglycemic after a pregnancy complicated by gestational diabetes), and prediabetes/screening-diagnosed type 2 diabetes. We asked if glucagon patterns were homogeneous within the metabolic phenotypes. Design and Setting Five-point oGTT, sandwich enzyme-linked immunosorbent assay for glucagon, and functional data analysis with unsupervised clustering. Participants Cross-sectional analysis of 285 women from the monocenter observational study Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes, recruited between November 2011 and May 2016. Results We found four patterns of glucagon dynamics that did not match the metabolic phenotypes. Elevated fasting glucagon and delayed glucagon suppression was overrepresented with prediabetes/diabetes, but this was only detected in 21% of this group. It also occurred in 8% of the control group. Conclusions We conclude that hyperglucagonemia may contribute to type 2 diabetes in a subgroup of affected individuals but that it is not a sine qua non for the disease. This should be considered in future pathophysiological studies and when testing pharmacotherapies addressing glucagon signaling. Glucagon is the main antagonist of insulin. It raises plasma glucose by reducing glycolysis and increasing gluconeogenesis and glycogenolysis (1, 2). Glucagon secretion from α cells is triggered by hypoglycemia and inhibited by insulin from neighboring β cells. In turn, glucagon inhibits insulin secretion (2). Unger et al. (3, 4) first postulated that elevated glucagon is a sine qua non in the development of diabetes. This marked the departure from an insulinocentric concept of type 2 diabetes pathogenesis to a bihormonal or even glucagonocentric model (4). In a glucagonocentric model, most metabolic derangements of diabetes are caused by the disinhibition of glucagon secretion (resulting from insulin-resistant α cells or impaired insulin release), but not directly by insufficient insulin action in other tissues (4, 5). The issue of the different pathophysiologic models remains unresolved, at least in part due to technical difficulties: glucagon is unstable, difficult to measure because of many similar peptides in plasma (2, 6), and its concentration is very low (7). Furthermore, α cells are harder to isolate than their insulin-producing neighbors (8). This impedes cellular studies. Current data on plasma glucagon levels in (pre)diabetic human subjects are also inconsistent. Several studies have found impaired glucagon suppression during an oral glucose tolerance test (oGTT) in prediabetic and diabetic individuals when compared with healthy controls (9–11). Other studies reported on increased fasting glucagon levels (12, 13). In contrast, Ahrén and Larsson (14) saw no differences between impaired glucose tolerance (IGT) and normoglycemic subjects, and Wagner et al. (15) observed rising glucagon values during an oGTT in 21% to 34% of healthy, insulin-sensitive individuals. These authors even found that this pattern predicted future metabolic health. New sandwich enzyme-linked immunosorbent assays (ELISAs) with improved specificity for glucagon became available recently, and this prompted us to re-examine the issue in a postpregnancy cohort of young women. We compared three groups of study participants with different metabolic phenotypes: a control group (normoglycemic women, who had recently completed a normoglycemic pregnancy), a normoglycemic high-risk group for type 2 diabetes [normoglycemic women after a recent pregnancy complicated by gestational diabetes (GDM)] (16, 17), and a prediabetes/diabetes group (women with prediabetes or screening-diagnosed type 2 diabetes after GDM). We first confirmed that average fasting plasma glucagon was higher and glucagon suppression during an oGTT was impaired in the normoglycemic high-risk and the prediabetes/diabetes groups, similar to what was seen in the majority of previous studies. However, our main research goal was to determine whether glucagon dynamics within each metabolic group were homogeneous or followed heterogeneous patterns. We used functional data analysis and unsupervised clustering to address this question. Research Design and Methods Study cohort Study participants were women enrolled in the prospective, monocenter observational study Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes (PPSDiab) between November 2011 and May 2016 (18). The cohort includes women with GDM during their last pregnancy and women following a normoglycemic pregnancy in a 2:1 ratio, recruited consecutively from the diabetes center and the obstetrics department of the University Hospital (Klinikum der Universität München) in Munich, Germany. Premenopausal women 3 to 16 months after a singleton (n = 295) or twin (n = 9) pregnancy with live birth(s) were eligible to participate. The GDM diagnosis was based on a 75-g oGTT with cut-off values for GDM according to the International Association of the Diabetes and Pregnancy Study Groups recommendations (plasma glucose: fasting 92 mg/dL, 1 hour 180 mg/dL, and 2 hours 153 mg/dL). Women without a history of GDM and either a normal 75-g oGTT (n = 294) or a normal 50-g screening oGTT (<135 mg/dL plasma glucose after 1 hour, n = 10) after the 23rd week of gestation were included in the normoglycemic group. Exclusion criteria for this study were alcohol or substance abuse, prepregnancy diabetes, and chronic diseases requiring continuous medication, except for hypothyroidism (n = 52), bronchial asthma (n = 8), mild hypertension (n = 4), gastroesophageal reflux (n = 2), and history of pulmonary embolism resulting in rivaroxaban prophylaxis (n = 1). Written informed consent was obtained from all study participants, and the protocol was approved by the ethical review committee of the Ludwig-Maximilians-Universität (study ID 300-11). Data used in this analysis were collected at the baseline visit of the PPSDiab study, 3 to 16 months after the index pregnancy. In addition to the baseline visit, post-GDM women also attended yearly follow-up visits with a 75-g oGTT. Groups We compared three groups of women: a control group (women normoglycemic at the baseline visit and after a normoglycemic pregnancy), a normoglycemic high-risk group (women normoglycemic at the baseline visit but with GDM during the preceding pregnancy), and a prediabetes/diabetes group [women with impaired fasting glucose (IFG), IGT, combined IFG plus IGT, or screening-diagnosed type 2 diabetes at the baseline visit and with GDM during the preceding pregnancy]. IFG [fasting plasma glucose ≥100 mg/dL (5.6 mmol/L)], IGT [2-hour plasma glucose ≥140 mg/dL (7.8 mmol/L)], and diabetes [fasting plasma glucose ≥126 mg/dL (7.0 mmol/L) or 2-hour plasma glucose ≥200 mg/dL (11.0 mmol/L)] were defined according to the criteria of the American Diabetes Association (19). Measurements We conducted a five-point 75-g oGTT with measurement of plasma glucose (Glucose HK Gen.3; Roche Diagnostics, Mannheim, Germany), serum insulin (chemiluminescent immunoassay; DiaSorin LIASON Systems, Saluggia, Italy), high-sensitivity C-reactive protein (wide-range C-reactive protein; Siemens Health Care Diagnostics, Erlangen, Germany), and blood lipids [low-density lipoprotein and high-density lipoprotein (HDL) cholesterol and triglycerides] (enzymatic caloric test; Roche Diagnostics, Mannheim, Germany) after an overnight fast. Plasma glucagon was measured at all five time points of oGTT with an ELISA (Glucagon ELISA; Mercodia, Uppsala, Sweden; catalog no: 10-1271-01) and also a radioimmunoassay (RIA) (Merck Millipore, Darmstadt, Germany; catalog no: GL-32K) for 283 subjects. ELISA and RIA measurements gave different results (Supplemental Table 1; Supplemental Fig. 1). In particular, suppression of plasma glucagon during the oGTT was insufficiently represented in the RIA measurement. Sensitivity and specificity of ELISA for pancreatic glucagon (amino acids 33 to 61) have been proven to be superior to RIA (20, 21). Thus, for this analysis, we exclusively used glucagon data measured by ELISA (n = 299). Plasma for glucagon measurements was collected in BD p800 tubes (BD Biosciences, San Jose, CA), which contain specific proteinase inhibitors to stabilize glucagon and other metabolically important hormones. Plasma was immediately separated by centrifugation and directly frozen in aliquots on dry ice, before being transferred to a –80°C freezer within 1 hour from completion of the oGTT. Glucagon measurements were done in one batch and only from aliquots that had not been thawed previously. Height and waist circumference were measured to the nearest 1 cm. Body mass and body fat mass were determined by a bioelectrical impedance analysis scale (Tanita BC-418; Tanita Corporation, Tokyo, Japan) (22, 23). Blood pressure was calculated as the mean out of two measurements in a resting seated position. In addition to these basic tests, all study subjects were asked to participate in a magnetic resonance imaging (MRI) measurement and an intravenous glucose tolerance test (ivGTT) on a voluntary basis. MRI (3 Tesla System, Ingenia, or Achieva; Philips Health Care, Hamburg, Germany) included determination of abdominal visceral adipose tissue volumes and liver fat content, using an mDixon low-fat fraction map. In the ivGTT, a glucose bolus of 0.3 g/kg body weight was injected over 1 minute with subsequent frequent blood sampling at 0, 2, 4, 6, 8, 10, 20, 30, 45, and 60 minutes. The measurements were used for the calculation of first-phase insulin response. A detailed description of the study design, anthropometric, clinical, and MRI measurements, and methodologies of blood sampling and analysis can be found elsewhere (24). Calculations   Mean blood pressure=(diastolic value*2+systolic value)/3 The insulin sensitivity index (ISI) according to Matsuda and DeFronzo (25) was calculated from the oGTT:  ISI=10000/√[(glucose 0,*insulin 0,)*(glucose 0,+2*(glucose 30,+60,+90,)+glucose120,)/8*(insulin0,+2*(insulin30, + 60, + 90,)+insulin 120,)/8]The disposition index (DI) was calculated as (26):  DI=ISI*IR30with  IR30=insulin 30,−insulin0,ISI and insulin release 0’ to 30’ in the oGTT were previously validated with data from ivGTT-euglycemic clamp tests in this cohort (24). Glucagon suppression indices were calculated as (27):  Early suppression=(1-[glucagon 30,/glucagon 0,])*100%  Late suppression=(1-[glucagon 120,/glucagon 30,])*100%  Overall suppression=(1-[glucagon 120,/glucagon 0,])*100%Area under the glucagon curve was calculated using the trapezoidal rule. First-phase insulin response in the ivGTT test was calculated as the incremental area under the insulin curve from 0 to 10 minutes. Statistical analysis All metric and normally distributed variables are reported as mean ± standard deviation; nonnormally distributed variables are presented as median (first quartile to third quartile). Categorical variables are presented as frequency and percentage. The Kruskal-Wallis test was used to compare metric variables, and the χ2 or Fisher’s exact test was used to compare categorical variables. For post hoc analysis, Dunn’s test was used. P values <0.05 were considered statistically significant. Functional data analysis methods were used for the analysis of the oGTT measurements (28). In the first step, the five-point oGTT measurements were converted into continuous, smooth curves based on B-spline basis functions (29). Afterward, a functional principal component analysis was performed based on the fitted curves to analyze the temporal variation (28). In the next step, a cluster analysis was conducted to identify patients with similar plasma glucagon dynamics. Hierarchical clustering was performed on the first three principal components of the functional principal component analysis via the Hierarchical Clustering on Principal Components function of Husson et al. (30). Hierarchical clustering was performed using the Ward’s criterion on the selected principal components. The number of clusters was chosen based on the growth of between-inertia. For the final partitioning, the k-means algorithm was performed with the partition obtained from the hierarchical tree as the initial partition. All statistical calculations were performed using SAS statistical software package version 9.3 (SAS Institute, Inc., Cary, NC) or R version 3.1.3 (www.r-project.org). Results Mean glucagon curves differ between metabolic groups We recruited 304 women into the PPSDiab study cohort but excluded 19 from this analysis. Two women were excluded because of type 1 diabetes diagnosed during follow-up, two because of overt hyperthyroidism, and one because of an acute upper respiratory infection at baseline. Eight women were excluded from the control group due to pathological glucose tolerance at the baseline visit, and six women were excluded due to missing glucagon values. Our final sample consisted of 285 study participants: 93 normoglycemic women after a normoglycemic pregnancy (control group), 121 normoglycemic women who had GDM (normoglycemic high-risk group), and 71 women with IFG, IGT, or newly diagnosed type 2 diabetes (prediabetes/diabetes group). Baseline characteristics of the study cohort are shown in Table 1. Mean age and low-density lipoprotein cholesterol were comparable, but mean blood pressure, waist circumference, triglycerides, c-reactive protein, liver fat content, intra-abdominal fat, and fasting and 2-hour plasma glucose increased and HDL cholesterol and insulin sensitivity decreased from the control over the normoglycemic high-risk to the prediabetes/diabetes groups (all significant over the three groups; results of pairwise post hoc tests shown in Table 1). Table 1. Baseline Characteristics of the PPSDiab Study Sample   Control  Normoglycemic High-Risk  Prediabetes/Diabetes  P Value  n  93  121  71    Glucose status           NGT  93 (100.0%)  121 (100.0%)  —     IFG  —  —  31 (43.7%)     IGT  —  —  22 (31.0%)     IFG + IGT  —  —  12 (16.9%)     Type 2 diabetes  —  —  6 (8.5%)    Age (y)  35.3 ± 4.2  35.2 ± 4.5  35.9 ± 4.5  0.6204  Mean blood pressure (mm Hg) (missing = 1)  85.8 ± 9.0  89.0 ± 8.6a  90.9 ± 10.3a  0.0026  BMI (kg/m2) (missing = 4)  23.7 ± 4.0  25.2 ± 5.8  28.2 ± 7.1a,b  0.0001  Waist circumference (cm) (missing = 5)  78.1 ± 8.9  80.7 ± 11.2  86.6 ± 13.2a,b  0.0002  hsCRP (mg/dL)  0.04 (0.01–0.08)  0.06 (0.02–0.25)a  0.09 (0.02–0.30)a  0.0030  Triglycerides (mg/dL)  61.0 (51.0–77.0)  65.0 (50.0–87.0)  81.0 (62.0–130.0)a,b  <0.0001  HDL cholesterol (mg/dL)  64.0 (57.0–73.0)  63.0 (56.0–73.0)  56.0 (46.0–65.0)a,b  <0.0001  LDL cholesterol (mg/dL)  104.0 (88.0–118.0)  105.0 (89.0–120.0)  104.0 (85.0–124.0)  0.9035  Plasma glucose 0 min (mg/dL)  89.0 (83.0–92.0)  91.0 (87.0–95.0)  102.0 (97.0–106.0)a,b  <0.0001  Plasma glucose 120 min (mg/dL)  93.0 (81.0–108.0)  114.0 (96.0–122.0)a  141.0 (113.0–165.0)a,b  <0.0001  ISI ( missing = 1)  6.8 (5.2–8.6)  5.5 (3.7–7.5)a  3.3 (2.1–4.6)a,b  <0.0001  DI (missing = 1)  297.4 (221.4–363.1)  246.6 (179.7–322.0)  160.0 (111.4–207.6)a,b  <0.0001  FPIR (missing = 152)  2.2 (1.4–3.5)  2.2 (1.6–3.5)  2.3 (1.5–3.9)  0.8218  Liver fat content (%) (missing = 132)  0.2 (0.0–0.8)  0.5 (0.0–1.1)  1.7 (0.0–4.1)a,b  0.0122  Intra-abdominal fat (L) (missing = 124)  1.4 (0.9–2.1)  1.8 (1.1–2.9)a  2.3 (1.3–3.2)a  0.0046  Glucagon 0 min (pmol/L)  6.0 (4.6–8.2)  6.6 (4.5–8.4)  7.7 (5.6–11.2)a,b  0.0069  Glucagon 30 min (pmol/L)  3.0 (2.4–4.7)  3.7 (2.5–4.9)  5.0 (3.0–7.6)a,b  <0.0001  Glucagon 60 min (pmol/L)  1.9 (1.4–3.1)  2.6 (1.8–3.7)  2.9 (2.0–4.4)a  0.0009  Glucagon 90 min (pmol/L)  2.1 (1.3–3.0)  2.1 (1.6–3.2)  2.5 (1.8–3.9)  0.0527  Glucagon 120 min (pmol/L)  2.3 (1.4–3.5)  2.2 (1.5–3.3)  2.3 (1.6–3.5)  0.5239  AUC glucagon  339.4 (248.5–473.6)  392.1 (283.5–518.2)  511.5 (353.4–615.2)a,b  0.0006  Early-suppression glucagon (0–30) (%)  47.6 (32.8–57.9)  41.3 (22.9–58.3)  32.0 (14.5–51.3)a  0.0055  Late-suppression glucagon (30–120) (%)  31.8 (8.9–49.6)  40.9 (14.9–56.7)  47.4 (33.3–63.6)a,b  <0.0001  Suppression glucagon (0–120) (%)  61.2 (48.2–76.9)  64.1 (49.5–74.4)  68.5 (57.3–75.0)  0.3130    Control  Normoglycemic High-Risk  Prediabetes/Diabetes  P Value  n  93  121  71    Glucose status           NGT  93 (100.0%)  121 (100.0%)  —     IFG  —  —  31 (43.7%)     IGT  —  —  22 (31.0%)     IFG + IGT  —  —  12 (16.9%)     Type 2 diabetes  —  —  6 (8.5%)    Age (y)  35.3 ± 4.2  35.2 ± 4.5  35.9 ± 4.5  0.6204  Mean blood pressure (mm Hg) (missing = 1)  85.8 ± 9.0  89.0 ± 8.6a  90.9 ± 10.3a  0.0026  BMI (kg/m2) (missing = 4)  23.7 ± 4.0  25.2 ± 5.8  28.2 ± 7.1a,b  0.0001  Waist circumference (cm) (missing = 5)  78.1 ± 8.9  80.7 ± 11.2  86.6 ± 13.2a,b  0.0002  hsCRP (mg/dL)  0.04 (0.01–0.08)  0.06 (0.02–0.25)a  0.09 (0.02–0.30)a  0.0030  Triglycerides (mg/dL)  61.0 (51.0–77.0)  65.0 (50.0–87.0)  81.0 (62.0–130.0)a,b  <0.0001  HDL cholesterol (mg/dL)  64.0 (57.0–73.0)  63.0 (56.0–73.0)  56.0 (46.0–65.0)a,b  <0.0001  LDL cholesterol (mg/dL)  104.0 (88.0–118.0)  105.0 (89.0–120.0)  104.0 (85.0–124.0)  0.9035  Plasma glucose 0 min (mg/dL)  89.0 (83.0–92.0)  91.0 (87.0–95.0)  102.0 (97.0–106.0)a,b  <0.0001  Plasma glucose 120 min (mg/dL)  93.0 (81.0–108.0)  114.0 (96.0–122.0)a  141.0 (113.0–165.0)a,b  <0.0001  ISI ( missing = 1)  6.8 (5.2–8.6)  5.5 (3.7–7.5)a  3.3 (2.1–4.6)a,b  <0.0001  DI (missing = 1)  297.4 (221.4–363.1)  246.6 (179.7–322.0)  160.0 (111.4–207.6)a,b  <0.0001  FPIR (missing = 152)  2.2 (1.4–3.5)  2.2 (1.6–3.5)  2.3 (1.5–3.9)  0.8218  Liver fat content (%) (missing = 132)  0.2 (0.0–0.8)  0.5 (0.0–1.1)  1.7 (0.0–4.1)a,b  0.0122  Intra-abdominal fat (L) (missing = 124)  1.4 (0.9–2.1)  1.8 (1.1–2.9)a  2.3 (1.3–3.2)a  0.0046  Glucagon 0 min (pmol/L)  6.0 (4.6–8.2)  6.6 (4.5–8.4)  7.7 (5.6–11.2)a,b  0.0069  Glucagon 30 min (pmol/L)  3.0 (2.4–4.7)  3.7 (2.5–4.9)  5.0 (3.0–7.6)a,b  <0.0001  Glucagon 60 min (pmol/L)  1.9 (1.4–3.1)  2.6 (1.8–3.7)  2.9 (2.0–4.4)a  0.0009  Glucagon 90 min (pmol/L)  2.1 (1.3–3.0)  2.1 (1.6–3.2)  2.5 (1.8–3.9)  0.0527  Glucagon 120 min (pmol/L)  2.3 (1.4–3.5)  2.2 (1.5–3.3)  2.3 (1.6–3.5)  0.5239  AUC glucagon  339.4 (248.5–473.6)  392.1 (283.5–518.2)  511.5 (353.4–615.2)a,b  0.0006  Early-suppression glucagon (0–30) (%)  47.6 (32.8–57.9)  41.3 (22.9–58.3)  32.0 (14.5–51.3)a  0.0055  Late-suppression glucagon (30–120) (%)  31.8 (8.9–49.6)  40.9 (14.9–56.7)  47.4 (33.3–63.6)a,b  <0.0001  Suppression glucagon (0–120) (%)  61.2 (48.2–76.9)  64.1 (49.5–74.4)  68.5 (57.3–75.0)  0.3130  Abbreviations: AUC, area under the curve; BMI, body mass index; FPIR, first-phase insulin response; hsCRP, high-sensitivity C-reactive protein; LDL, low-density lipoprotein; NGT, normal glucose tolerance. a Significant post hoc tests vs control. b Significant post hoc tests vs normoglycemic high-risk. View Large We next compared plasma glucagon levels during the oGTT in the three groups (Table 1). Fasting plasma glucagon was significantly elevated, and early glucagon suppression was diminished in the prediabetes/diabetes group compared with the control group [median (Q1 to Q3) for fasting plasma glucagon: 6.0 (4.6 to 8.2) (pmol/L) vs 7.7 (5.6 to 11.2) (pmol/L); early glucagon suppression: 47.6 (32.8 to 57.9) (pmol/L) vs 32.0 (14.5 to 51.3) (pmol/L), respectively]. The normoglycemic high-risk group lay in between for these variables, but closer to the control group and not statistically different from it [median (Q1 to Q3) for fasting plasma glucagon: 6.6 (4.5 to 8.4) (pmol/L); early glucagon suppression: 41.3 (22.9 to 58.3) (pmol/L)] (Fig. 1; Table 1). Total glucagon suppression was similar in all three groups. Figure 1. View largeDownload slide Glucagon during oGTT stratified by risk groups (blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Figure 1. View largeDownload slide Glucagon during oGTT stratified by risk groups (blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Similar to a recent publication by Faerch et al. (27), we further examined fasting glucagon values and glucagon suppression indices in women with isolated IFG compared with those with isolated IGT and combined IFG + IGT (Supplemental Fig. 2; Supplemental Table 2). Late and overall glucagon suppression was smaller in women with isolated IFG compared with both other groups [median (Q1 to Q3) late suppression: 41.8 (16.5 to 50.4) (%) vs 58.1 (43.1 to 71.3) (%) vs 58.9 (46.1 to 69.6) (%) and overall suppression: 58.9 (39.8 to 70.2) (%) vs 71.2 (68.4 to 81.0) (%) vs 73.7 (63.8 to 81.0) (%) in IFG vs IGT vs IFG + IGT, respectively). We observed no significant differences in early glucagon suppression and fasting glucagon. Plasma glucagon patterns are heterogeneous within each metabolic group The five-point glucagon curves in response to oral glucose were heterogeneous between individuals (Fig. 2a). To examine this further, we calculated continuous, smooth curves from the five measurements during the oGTT based on B-spline basis functions (Fig. 2a). Stratified by group, these curves confirmed within-group heterogeneity of plasma glucagon dynamics (Supplemental Fig. 3). To permit pattern identification, we added a principal component analysis of the curves. The first three principal component factors explained 79%, 17%, and 3% of curve variance, respectively (Fig. 2b). We used these three principal components as input for an unsupervised cluster analysis (Fig. 2c). This identified four clusters corresponding to four distinct patterns of plasma glucagon dynamics (Fig. 2d). Figure 2. View largeDownload slide Process of functional data analysis. (a) Based on the five-point oGTT data curves, continuous, smooth curves were calculated (median indicated by black line). (b) Then, a principal component analysis of the curves was conducted (median indicated by solid line; extremes indicated by dotted lines). (c) The three principal components were used as input for an unsupervised cluster analysis (asterisk indicates line types used to represent the clusters in Fig. 3). (d) Fitted glucagon curves during oGTT stratified by the four clusters (colors: original risk groups as used in Table 1 and Fig. 1; blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Figure 2. View largeDownload slide Process of functional data analysis. (a) Based on the five-point oGTT data curves, continuous, smooth curves were calculated (median indicated by black line). (b) Then, a principal component analysis of the curves was conducted (median indicated by solid line; extremes indicated by dotted lines). (c) The three principal components were used as input for an unsupervised cluster analysis (asterisk indicates line types used to represent the clusters in Fig. 3). (d) Fitted glucagon curves during oGTT stratified by the four clusters (colors: original risk groups as used in Table 1 and Fig. 1; blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Cluster 3 was the largest (n = 188; Table 2) and showed low mean fasting glucagon and rapid suppression during the oGTT (Figs. 2d and 3a ). Cluster 2, the second largest (n = 62), had higher mean fasting glucagon but equally rapid suppression. Cluster 1 (n = 21) had high mean fasting glucagon and delayed suppression, and cluster 4 (n = 7) had low mean fasting glucagon and a rising curve after glucose ingestion (Fig. 3a; Table 2). Table 2. Baseline Characteristics of the PPSDiab Study Sample, Stratified by Clusters of Glucagon Dynamics   Cluster 1  Cluster 2  Cluster 3  Cluster 4  P Value  n  28  62  188  7    Risk group             Control  7 (25.0%)  19 (30.7%)  65 (34.6%)  2 (28.6%)  0.0279   Normoglycemic high-risk  6 (21.4%)  27 (43.6%)  84 (44.7%)  4 (57.1%)     Prediabetes/diabetes  15 (53.6%)  16 (25.8%)  39 (20.7%)  1 (14.3%)    Glucose status             NGT  13 (46.4%)  46 (74.2%)  149 (79.3%)  6 (85.7%)  0.0099   IFG  5 (17.9%)  6 (9.7%)  19 (10.1%)  1 (14.3%)     IGT  3 (10.7%)  7 (11.3%)  12 (6.4%)  0     IFG + IGT  3 (10.7%)  3 (4.8%)  6 (3.2%)  0     Type 2 diabetes  4 (14.3%)  0  2 (1.1%)  0    Age (y)  33.5 ± 4.8  35.5 ± 4.4  35.7 ± 4.3a  35.0 ± 4.0  0.0315  Mean blood pressure (mm Hg) (missing = 1)  96.2 ± 8.6  89.4 ± 9.2  87.0 ± 9.0  85.6 ± 7.4  <0.0001  BMI (kg/m2) (missing = 4)  33.3 ± 6.1  26.5 ± 6.4a  24.0 ± 4.6a  21.6 ± 1.5a  <0.0001  Waist circumference (cm) (missing = 5)  96.0 ± 11.9  83.8 ± 12.3a  78.6 ± 9.3a  73.5 ± 4.1a  <0.0001  hsCRP (mg/dL)  0.19 (0.07–0.47)  0.05 (0.01–0.17)a  0.04 (0.01–0.12)a  0.12 (0.05–0.38)  0.0004  Triglycerides (mg/dL)  91.5 (58.5–132.0)  62.5 (53.0–83.0)  67.5 (53.0–88.5)  63.0 (58.0–91.0)  0.0898  HDL cholesterol (mg/dL)  49.0 (44.5–61.5)  62.0 (51.0–73.0)  63.0 (56.0–73.0)  65.0 (56.0–70.0)  0.0012  Plasma glucose 0 min (mg/dL)  97.5 (90.5–106.0)  91.0 (88.0–97.0)  91.0 (86.0–97.0)  87.0 (82.0–92.0)  0.0078  Plasma glucose 120 min (mg/dL)  127.0 (115.5–154.5)  113.5 (95.0–130.0)a  106.5 (90.0–121.5)a  80.0 (74.0–92.0)a,b  <0.0001  ISI (missing = 1)  2.5 (1.9–4.3)  5.0 (3.3–6.9)a  5.8 (4.2–8.1)a  7.9 (5.6–8.3)a  <0.0001  DI (missing = 1)  152.0 (96.5–247.8)  230.2 (165.3–392.0)a  252.8 (176.7–324.4)a  232.9 (156.2–276.4)  0.0007  IR30 (missing = 1)  55.7 (37.1–82.2)  50.3 (36.4–86.1)  41.6 (30.9–60.1)  28.7 (26.2–41.3)a,b  0.0023  FPIR (missing = 152)c  3.9 (2.2–6.2)  3.3 (2.2–4.3)  2.1 (1.4–3.1)  2.1 (1.0–2.7) (n = 3)  0.0140  Liver fat content (%) (missing = 131)  2.4 (1.1–6.4)  0.7 (0.0–1.7)a  0.3 (0.0–0.8)a  0.1 (0.0–0.5)a  <0.0001  Intra-abdominal fat (L) (missing = 124)  3.4 (2.9–4.4)  2.0 (1.5–3.0)a  1.5 (1.0–2.3)a,b  1.1 (0.9–1.6)a,b  <0.0001    Cluster 1  Cluster 2  Cluster 3  Cluster 4  P Value  n  28  62  188  7    Risk group             Control  7 (25.0%)  19 (30.7%)  65 (34.6%)  2 (28.6%)  0.0279   Normoglycemic high-risk  6 (21.4%)  27 (43.6%)  84 (44.7%)  4 (57.1%)     Prediabetes/diabetes  15 (53.6%)  16 (25.8%)  39 (20.7%)  1 (14.3%)    Glucose status             NGT  13 (46.4%)  46 (74.2%)  149 (79.3%)  6 (85.7%)  0.0099   IFG  5 (17.9%)  6 (9.7%)  19 (10.1%)  1 (14.3%)     IGT  3 (10.7%)  7 (11.3%)  12 (6.4%)  0     IFG + IGT  3 (10.7%)  3 (4.8%)  6 (3.2%)  0     Type 2 diabetes  4 (14.3%)  0  2 (1.1%)  0    Age (y)  33.5 ± 4.8  35.5 ± 4.4  35.7 ± 4.3a  35.0 ± 4.0  0.0315  Mean blood pressure (mm Hg) (missing = 1)  96.2 ± 8.6  89.4 ± 9.2  87.0 ± 9.0  85.6 ± 7.4  <0.0001  BMI (kg/m2) (missing = 4)  33.3 ± 6.1  26.5 ± 6.4a  24.0 ± 4.6a  21.6 ± 1.5a  <0.0001  Waist circumference (cm) (missing = 5)  96.0 ± 11.9  83.8 ± 12.3a  78.6 ± 9.3a  73.5 ± 4.1a  <0.0001  hsCRP (mg/dL)  0.19 (0.07–0.47)  0.05 (0.01–0.17)a  0.04 (0.01–0.12)a  0.12 (0.05–0.38)  0.0004  Triglycerides (mg/dL)  91.5 (58.5–132.0)  62.5 (53.0–83.0)  67.5 (53.0–88.5)  63.0 (58.0–91.0)  0.0898  HDL cholesterol (mg/dL)  49.0 (44.5–61.5)  62.0 (51.0–73.0)  63.0 (56.0–73.0)  65.0 (56.0–70.0)  0.0012  Plasma glucose 0 min (mg/dL)  97.5 (90.5–106.0)  91.0 (88.0–97.0)  91.0 (86.0–97.0)  87.0 (82.0–92.0)  0.0078  Plasma glucose 120 min (mg/dL)  127.0 (115.5–154.5)  113.5 (95.0–130.0)a  106.5 (90.0–121.5)a  80.0 (74.0–92.0)a,b  <0.0001  ISI (missing = 1)  2.5 (1.9–4.3)  5.0 (3.3–6.9)a  5.8 (4.2–8.1)a  7.9 (5.6–8.3)a  <0.0001  DI (missing = 1)  152.0 (96.5–247.8)  230.2 (165.3–392.0)a  252.8 (176.7–324.4)a  232.9 (156.2–276.4)  0.0007  IR30 (missing = 1)  55.7 (37.1–82.2)  50.3 (36.4–86.1)  41.6 (30.9–60.1)  28.7 (26.2–41.3)a,b  0.0023  FPIR (missing = 152)c  3.9 (2.2–6.2)  3.3 (2.2–4.3)  2.1 (1.4–3.1)  2.1 (1.0–2.7) (n = 3)  0.0140  Liver fat content (%) (missing = 131)  2.4 (1.1–6.4)  0.7 (0.0–1.7)a  0.3 (0.0–0.8)a  0.1 (0.0–0.5)a  <0.0001  Intra-abdominal fat (L) (missing = 124)  3.4 (2.9–4.4)  2.0 (1.5–3.0)a  1.5 (1.0–2.3)a,b  1.1 (0.9–1.6)a,b  <0.0001  Abbreviations: BMI, body mass index; FPIR, first-phase insulin response; hsCRP, high-sensitivity C-reactive protein; IR30, insulin release 0’ to 30’ in the oGTT; NGT, normal glucose tolerance. a Significant post hoc test: significant vs cluster 1. b Significant post hoc test: significant vs cluster 2. c The post hoc test for FPIR was conducted both including cluster 4 and after exclusion of cluster 4 (due to the small group size in cluster 4); in any case, the post hoc test has not reached significance. View Large Figure 3. View largeDownload slide Means of (a) glucagon, (b) glucose, (c) insulin, and (d) c-peptide curves during oGTT stratified by the four clusters derived from the glucagon curves (Fig. 2). Figure 3. View largeDownload slide Means of (a) glucagon, (b) glucose, (c) insulin, and (d) c-peptide curves during oGTT stratified by the four clusters derived from the glucagon curves (Fig. 2). Cluster 1 contained the highest proportion of women from the prediabetes/diabetes group (53%), followed by cluster 2, cluster 3, and cluster 4. Women in cluster 1 had significantly higher body mass index, waist circumference, triglycerides, liver fat content, and intra-abdominal fat and lower HDL cholesterol and ISI than those in the other three clusters. The DI of cluster 1 was significantly lower than those of clusters 2 and 3 (Table 2). Cluster 4 included lean, insulin-sensitive women with a tendency toward low glucose values (Fig. 3b and 3c; Table 2). Discussion In our first analysis, we found that women with prediabetes/screening-diagnosed type 2 diabetes had higher fasting glucagon and delayed glucagon suppression during an oGTT compared with healthy control subjects (normoglycemic women after a normoglycemic pregnancy). Normoglycemic women after GDM, a high-risk group for type 2 diabetes (16, 17), lay in between, with values closer to and not statistically different from the control group. These results are in line with most previous studies that saw the highest fasting glucagon and most impaired glucagon suppression in subjects with diabetes, followed by those with prediabetes, and, at the low end, normoglycemic individuals (10–13, 27). In several nondiabetic cohorts, fasting glucagon was higher in insulin-resistant than in insulin-sensitive subjects (31–33). A majority of studies also found a positive association of plasma glucagon with obesity in groups with similar glucose tolerance (11, 13, 31). Some earlier studies had different findings. Ahrén and Larsson (14) reported that fasting and postprandial glucagon did not differ between IGT and normoglycemic subjects in 84 postmenopausal women. Wagner et al. (15) analyzed cohorts of nondiabetic individuals and found that, in 21% to 34% of subjects, glucagon was not suppressed until 120 minutes into the oGTT. These individuals were lean and insulin-sensitive, and also had a favorable prognosis of insulin sensitivity over time. In their recent study, Faerch et al. (27) described that glucagon curves differed between individuals with IFG and those with IGT. They found a smaller overall decrease in glucagon during an oGTT in the group with isolated IFG compared with isolated IGT and combined IFG + IGT. Our analysis confirms this result, with the difference in overall glucagon suppression mainly caused by the late phase of the oGTT (Supplemental Fig. 2; Supplemental Table 2). In our second analysis, we saw that plasma glucagon dynamics in the study cohort followed four different patterns, based on an unsupervised cluster analysis. The clusters detected did not fully or even closely match the predefined metabolic groups. We consider this the main finding of this paper. Subjects from the prediabetes/diabetes group were overrepresented in cluster 1 (with high fasting glucagon and diminished suppression), but still only made up 50% of this cluster, which also contained 25% control subjects. Conversely, the majority of women from the prediabetes/diabetes group (n = 39; 55%) fell into cluster 3, the “most normal” cluster (with low fasting glucagon and rapid suppression). Therefore, hyperglucagonemia was not a universal prerequisite for impaired glucose metabolism or early type 2 diabetes. It only affected a subgroup of individuals. Delayed glucagon suppression was clearly associated with obesity and metabolic syndrome markers in our study. This is evident from the clinical characteristics (e.g., waist circumference, blood lipids, plasma glucose, and intra-abdominal and liver fat) of the subjects in cluster 1 compared with the other clusters (Table 2). Hepatic steatosis may even be a cause of hyperglucagonemia, as it disrupts hepatic glucagon sensitivity and probably leads to reactive hypersecretion of the hormone (34). The association of liver fat and hyperglucagonemia was found independent of the presence of disrupted glucose metabolism (34, 35). Impaired early insulin secretion could be another cause of delayed postprandial glucagon suppression, but we do not find evidence for this relationship. Early insulin and c-peptide levels during the oGTT and first-phase insulin secretion in the ivGTT were not reduced in the women in cluster 1. The reduced DI results from lower insulin sensitivity (ISI) in this cluster, but not from reduced early insulin secretion (insulin release 0’ to 30’ in the oGTT) (Table 2). The α cell resistance to inhibition by insulin or a reactive glucagon hypersecretion due to a resistance of the liver is therefore the most likely explanations for our findings. .Another noteworthy observation was the small cluster 4 (n = 7; 2.5% of participants), with low fasting glucagon, but rising glucagon levels during the oGTT. The women in this cluster were lean and insulin-sensitive and had low glucose levels. In this group, the rising glucagon probably is a physiologic response to avoid postchallenge hypoglycemia as a result of an overactive insulin response, which is not uncommon in lean, young women (36). Wagner et al. (15) associated rising glucagon during an oGTT with a favorable metabolic prognosis. Our small and probably not representative sample does not confirm this finding. Five of the 7 women in cluster 4 had had GDM (Table 2), and all of these 5 women developed prediabetes or diabetes during the follow-up of this study (mean duration of follow-up was 38.2 months; data not shown). In our cohort, this phenotype is also much less common than reported in the previous publication. However, given the small number of subjects in cluster 4, we find these observations interesting and worth following up on, but we do not claim that they constitute scientific evidence by themselves. Finally, we believe it is important to use highly specific glucagon assays, in particular to study postprandial glucagon dynamics. We initially used a standard RIA, which strongly underestimated glucagon suppression (Supplemental Fig. 1). This was probably due to cross-reactivity with other peptides cleaved from proglucagon, such as oxyntomodulin, glicentin 1-61 (N-terminally elongated glucagon), and miniglucagon. Intestinal secretion of these peptides increases in the postprandial state, masking glucagon suppression (21, 37–39). Sandwich ELISAs, with antibodies against the N- and the C-terminal end of the glucagon molecule, circumvent this problem. Strengths of this study are optimal preanalytic and analytic techniques plus a cohort homogeneous for age and sex and with little medication and concomitant diseases. We used functional data analysis to interpret glucagon dynamics and also consider this a strength of our work. This method can extract more of the information contained in a function than classic multivariate statistical techniques (40–42). Together with a subsequent cluster analysis, it permits the grouping of data sets according to their curve shapes. Using a recent history of GDM to identify a high-risk cohort early in the process of type 2 diabetes development should have limited secondary metabolic abnormalities to the minimal extent possible in a human study. At the same time, the study cohort can also be interpreted as a weakness, because results may not apply to the general population. Another limitation of this analysis is its cross-sectional design, which precludes the clarification of cause-effect relationships. We conclude that fasting hyperglucagonemia and delayed postprandial glucagon suppression associate with insulin resistance, prediabetes, and diabetes, but are, in reality, only present in subgroups of individuals. Dysglycemia can develop without elevated plasma glucagon, and elevated glucagon does not preclude normoglycemia. Fasting hyperglucagonemia and delayed suppression are strongly linked to obesity and metabolic syndrome. Rising glucagon during an oGTT may be a rare phenomenon. It occurs in insulin-sensitive individuals with a tendency toward hypoglycemia, but does not necessarily indicate metabolic health. Our results have consequences for the pathophysiologic understanding of type 2 diabetes and for the development of precision treatments. At present, glucagon agonists and antagonists are evaluated for diabetes therapy (1, 2, 43, 44). Based on our findings, patients should probably be stratified by glucagon values for such treatments. For those patients with hyperglucagonemia, glucagon antagonists could be an appropriate therapy, whereas for others, agonists may be useful to induce beneficial effects mediated through the glucagon receptor, such as weight loss (2, 44). Abbreviations: DI disposition index ELISA enzyme-linked immunosorbent assay GDM gestational diabetes HDL high-density lipoprotein IFG impaired fasting glucose IGT impaired glucose tolerance ISI insulin sensitivity index ivGTT intravenous glucose tolerance test MRI magnetic resonance imaging oGTT oral glucose tolerance test PPSDiab Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes RIA radioimmunoassay. Acknowledgments We thank all participants in the PPSDiab study and to the diabetes care team of the Medizinische Klinik IV. Financial Support: This work was supported by the Helmholtz Zentrum für München, Klinikum der Universität München, and the German Center for Diabetes Research (to A.L.). Author Contributions: Conceptualization and methodology, A.L. and J.S.; formal analysis, M.R.; investigation, C.G., V.S., F.B., S.M., N.H., and D.P.; resources, C.H.; data curation, V.S. and C.G.; writing—original draft, C.G.; writing—review and editing, A.L.; and visualization, M.R., C.G., and A.L. All authors reviewed and approved the final version of the manuscript, revising it critically for important intellectual content. Disclosure Summary: The authors have nothing to disclose. References 1. Jiang G, Zhang BB. Glucagon and regulation of glucose metabolism. Am J Physiol Endocrinol Metab . 2003; 284( 4): E671– E678. Google Scholar CrossRef Search ADS PubMed  2. Wewer Albrechtsen NJ, Kuhre RE, Pedersen J, Knop FK, Holst JJ. The biology of glucagon and the consequences of hyperglucagonemia. Biomarkers Med . 2016; 10( 11): 1141– 1151. Google Scholar CrossRef Search ADS   3. Lee Y, Berglund ED, Wang MY, Fu X, Yu X, Charron MJ, Burgess SC, Unger RH. Metabolic manifestations of insulin deficiency do not occur without glucagon action. Proc Natl Acad Sci USA . 2012; 109( 37): 14972– 14976. Google Scholar CrossRef Search ADS PubMed  4. Unger RH, Cherrington AD. Glucagonocentric restructuring of diabetes: a pathophysiologic and therapeutic makeover. J Clin Invest . 2012; 122( 1): 4– 12. Google Scholar CrossRef Search ADS PubMed  5. Braaten JT, Faloona GR, Unger RH. The effect of insulin on the alpha-cell response to hyperglycemia in long-standing alloxan diabetes. J Clin Invest . 1974; 53( 4): 1017– 1021. Google Scholar CrossRef Search ADS PubMed  6. Bataille D, Dalle S. The forgotten members of the glucagon family. Diabetes Res Clin Pract . 2014; 106( 1): 1– 10. Google Scholar CrossRef Search ADS PubMed  7. Ruiter M, La Fleur SE, van Heijningen C, van der Vliet J, Kalsbeek A, Buijs RM. The daily rhythm in plasma glucagon concentrations in the rat is modulated by the biological clock and by feeding behavior. Diabetes . 2003; 52( 7): 1709– 1715. Google Scholar CrossRef Search ADS PubMed  8. Gaisano HY, Macdonald PE, Vranic M. Glucagon secretion and signaling in the development of diabetes. Front Physiol . 2012; 3: 349. Google Scholar CrossRef Search ADS PubMed  9. Beis C, Grigorakis SI, Philippou G, Alevizaki M, Anastasiou E. Lack of suppression of plasma glucagon levels in late pregnancy persists postpartum only in women with previous gestational diabetes mellitus. Acta Diabetol . 2005; 42( 1): 31– 35. Google Scholar CrossRef Search ADS PubMed  10. Abdul-Ghani M, DeFronzo RA. Fasting hyperglycemia impairs glucose- but not insulin-mediated suppression of glucagon secretion. J Clin Endocrinol Metab . 2007; 92( 5): 1778– 1784. Google Scholar CrossRef Search ADS PubMed  11. Knop FK, Aaboe K, Vilsbøll T, Vølund A, Holst JJ, Krarup T, Madsbad S. Impaired incretin effect and fasting hyperglucagonaemia characterizing type 2 diabetic subjects are early signs of dysmetabolism in obesity. Diabetes Obes Metab . 2012; 14( 6): 500– 510. Google Scholar CrossRef Search ADS PubMed  12. Shen J, Chen Z, Chen C, Zhu X, Han Y. Impact of incretin on early-phase insulin secretion and glucose excursion. Endocrine . 2013; 44( 2): 403– 410. Google Scholar CrossRef Search ADS PubMed  13. Manell H, Staaf J, Manukyan L, Kristinsson H, Cen J, Stenlid R, Ciba I, Forslund A, Bergsten P. Altered plasma levels of glucagon, GLP-1 and glicentin during OGTT in adolescents with obesity and type 2 diabetes. J Clin Endocrinol Metab . 2016; 101( 3): 1181– 1189. Google Scholar CrossRef Search ADS PubMed  14. Ahrén B, Larsson H. Impaired glucose tolerance (IGT) is associated with reduced insulin-induced suppression of glucagon concentrations. Diabetologia . 2001; 44( 11): 1998– 2003. Google Scholar CrossRef Search ADS PubMed  15. Wagner R, Hakaste LH, Ahlqvist E, Heni M, Machann J, Schick F, Van Obberghen E, Stefan N, Gallwitz B, Tuomi T, Haring HU, Groop L, Fritsche A. Non-suppressed glucagon after glucose challenge as a potential predictor for glucose tolerance. Diabetes . 2017; 66( 5): 1373– 1379. Google Scholar CrossRef Search ADS PubMed  16. Bellamy L, Casas JP, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet . 2009; 373( 9677): 1773– 1779. Google Scholar CrossRef Search ADS PubMed  17. Feig DS, Zinman B, Wang X, Hux JE. Risk of development of diabetes mellitus after diagnosis of gestational diabetes. CMAJ . 2008; 179( 3): 229– 234. Google Scholar CrossRef Search ADS PubMed  18. Gar C, Rottenkolber M, Grallert H, Banning F, Freibothe I, Sacco V, Wichmann C, Reif S, Potzel A, Dauber V, Schendell C, Sommer NN, Wolfarth B, Seissler J, Lechner A, Ferrari U. Physical fitness and plasma leptin in women with recent gestational diabetes. PLoS One . 2017; 12( 6): e0179128. Google Scholar CrossRef Search ADS PubMed  19. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care . 2011; 34( Suppl 1): S62– S69. CrossRef Search ADS PubMed  20. Raines M, King C, Christensen S, Amin P, Schranz D. 2014 Glucagon quantification: comparison of radioimmunoassay and sandwich ELISA methods. In: American Association for Clinical Chemistry : AACC; Poster A-213. 21. Matsuo T, Miyagawa J, Kusunoki Y, Miuchi M, Ikawa T, Akagami T, Tokuda M, Katsuno T, Kushida A, Inagaki T, Namba M. Postabsorptive hyperglucagonemia in patients with type 2 diabetes mellitus analyzed with a novel enzyme-linked immunosorbent assay. J Diabetes Investig . 2016; 7( 3): 324– 331. Google Scholar CrossRef Search ADS PubMed  22. Bosy-Westphal A, Later W, Hitze B, Sato T, Kossel E, Gluer CC, Heller M, Muller MJ. Accuracy of bioelectrical impedance consumer devices for measurement of body composition in comparison to whole body magnetic resonance imaging and dual X-ray absorptiometry. Obes Facts . 2008; 1( 6): 319– 324. Google Scholar CrossRef Search ADS PubMed  23. Bosy-Westphal A, Schautz B, Later W, Kehayias JJ, Gallagher D, Müller MJ. What makes a BIA equation unique? Validity of eight-electrode multifrequency BIA to estimate body composition in a healthy adult population. Eur J Clin Nutr . 2013; 67( Suppl 1): S14– S21. Google Scholar CrossRef Search ADS PubMed  24. Rottenkolber M, Ferrari U, Holland L, Aertsen S, Kammer NN, Hetterich H, Fugmann M, Banning F, Weise M, Sacco V, Kohn D, Freibothe I, Hutter S, Hasbargen U, Lehmann R, Grallert H, Parhofer KG, Seissler J, Lechner A. The diabetes risk phenotype of young women with recent gestational diabetes. J Clin Endocrinol Metab . 2015; 100( 6): E910– E918. Google Scholar CrossRef Search ADS PubMed  25. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care . 1999; 22( 9): 1462– 1470. Google Scholar CrossRef Search ADS PubMed  26. Kahn SE, Prigeon RL, McCulloch DK, Boyko EJ, Bergman RN, Schwartz MW, Neifing JL, Ward WK, Beard JC, Palmer JP. Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function. Diabetes . 1993; 42( 11): 1663– 1672. Google Scholar CrossRef Search ADS PubMed  27. Færch K, Vistisen D, Pacini G, Torekov SS, Johansen NB, Witte DR, Jonsson A, Pedersen O, Hansen T, Lauritzen T, Jørgensen ME, Ahrén B, Holst JJ. Insulin resistance is accompanied by increased fasting glucagon and delayed glucagon suppression in individuals with normal and impaired glucose regulation. Diabetes . 2016; 65( 11): 3473– 3481. Google Scholar CrossRef Search ADS PubMed  28. Ramsay JO, Silverman BW. Functional Data Analysis. 2 . New York: Springer; 2005. Google Scholar CrossRef Search ADS   29. Ramsay JO, Hooker G, Graves S. Functional Data Analysis With R and MATLAB . New York: Springer Science & Business Media; 2009. CrossRef Search ADS   30. Husson F, Josse J, Pagès J. Principal component methods – hierarchical clustering – partitional clustering: why would we need to choose for visualizing data? Available at: http://factominer.free.fr/more/HCPC_husson_josse.pdf. Accessed 31 May 2017. 31. Calanna S, Piro S, Di Pino A, Maria Zagami R, Urbano F, Purrello F, Maria Rabuazzo A. Beta and alpha cell function in metabolically healthy but obese subjects: relationship with entero-insular axis. Obesity (Silver Spring) . 2013; 21( 2): 320– 325. Google Scholar CrossRef Search ADS PubMed  32. Ferrannini E, Muscelli E, Natali A, Gabriel R, Mitrakou A, Flyvbjerg A, Golay A, Hojlund K; Relationship between Insulin Sensitivity and Cardiovascular Disease Risk (RISC) Project Investigators. Association of fasting glucagon and proinsulin concentrations with insulin resistance. Diabetologia . 2007; 50( 11): 2342– 2347. Google Scholar CrossRef Search ADS PubMed  33. Weiss R, D’Adamo E, Santoro N, Hershkop K, Caprio S. Basal alpha-cell up-regulation in obese insulin-resistant adolescents. J Clin Endocrinol Metab . 2011; 96( 1): 91– 97. Google Scholar CrossRef Search ADS PubMed  34. Junker AE, Gluud L, Holst JJ, Knop FK, Vilsbøll T. Diabetic and nondiabetic patients with nonalcoholic fatty liver disease have an impaired incretin effect and fasting hyperglucagonaemia. J Intern Med . 2016; 279( 5): 485– 493. Google Scholar CrossRef Search ADS PubMed  35. Bernsmeier C, Meyer-Gerspach AC, Blaser LS, Jeker L, Steinert RE, Heim MH, Beglinger C. Glucose-induced glucagon-like Peptide 1 secretion is deficient in patients with non-alcoholic fatty liver disease. PLoS One . 2014; 9( 1): e87488. Google Scholar CrossRef Search ADS PubMed  36. Brun JF, Fedou C, Mercier J. Postprandial reactive hypoglycemia. Diabetes Metab . 2000; 26( 5): 337– 351. Google Scholar PubMed  37. Dunning BE, Gerich JE. The role of alpha-cell dysregulation in fasting and postprandial hyperglycemia in type 2 diabetes and therapeutic implications. Endocr Rev . 2007; 28( 3): 253– 283. Google Scholar CrossRef Search ADS PubMed  38. Orskov C, Jeppesen J, Madsbad S, Holst JJ. Proglucagon products in plasma of noninsulin-dependent diabetics and nondiabetic controls in the fasting state and after oral glucose and intravenous arginine. J Clin Invest . 1991; 87( 2): 415– 423. Google Scholar CrossRef Search ADS PubMed  39. Holst JJ, Christensen M, Lund A, de Heer J, Svendsen B, Kielgast U, Knop FK. Regulation of glucagon secretion by incretins. Diabetes Obes Metab . 2011; 13( Suppl 1): 89– 94. Google Scholar CrossRef Search ADS PubMed  40. Ferraty F, Mas A, Vieu P. Advances on nonparametric regression for functional variables. Aust N Z J Stat . 2007; 49: 1– 20. Google Scholar CrossRef Search ADS   41. Mas A, Pumo B. Functional linear regression with derivatives. J Nonparametr Stat . 2009; 21( 1): 19– 40. Google Scholar CrossRef Search ADS   42. Frøslie KF, Røislien J, Qvigstad E, Godang K, Bollerslev J, Voldner N, Henriksen T, Veierød MB. Shape information from glucose curves: functional data analysis compared with traditional summary measures. BMC Med Res Methodol . 2013; 13( 1): 6. Google Scholar CrossRef Search ADS PubMed  43. Guzman CB, Zhang XM, Liu R, Regev A, Shankar S, Garhyan P, Pillai SG, Kazda C, Chalasani N, Hardy TA. Treatment with LY2409021, a glucagon receptor antagonist, increases liver fat in patients with type 2 diabetes. Diabetes Obes Metab . 2017; 19( 11): 1521– 1528. Google Scholar CrossRef Search ADS PubMed  44. Lefèbvre PJ, Paquot N, Scheen AJ. Inhibiting or antagonizing glucagon: making progress in diabetes care. Diabetes Obes Metab . 2015; 17( 8): 720– 725. Google Scholar CrossRef Search ADS PubMed  Copyright © 2018 Endocrine Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Clinical Endocrinology and Metabolism Oxford University Press

Loading next page...
 
/lp/ou_press/patterns-of-plasma-glucagon-dynamics-do-not-match-metabolic-phenotypes-uem0f2PzN6
Publisher
Oxford University Press
Copyright
Copyright © 2018 Endocrine Society
ISSN
0021-972X
eISSN
1945-7197
D.O.I.
10.1210/jc.2017-02014
Publisher site
See Article on Publisher Site

Abstract

Abstract Context The role of hyperglucagonemia in type 2 diabetes is still debated. Objective We analyzed glucagon dynamics during oral glucose tolerance tests (oGTTs) in young women with one out of three metabolic phenotypes: healthy control (normoglycemic after a normoglycemic pregnancy), normoglycemic high-risk (normoglycemic after a pregnancy complicated by gestational diabetes), and prediabetes/screening-diagnosed type 2 diabetes. We asked if glucagon patterns were homogeneous within the metabolic phenotypes. Design and Setting Five-point oGTT, sandwich enzyme-linked immunosorbent assay for glucagon, and functional data analysis with unsupervised clustering. Participants Cross-sectional analysis of 285 women from the monocenter observational study Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes, recruited between November 2011 and May 2016. Results We found four patterns of glucagon dynamics that did not match the metabolic phenotypes. Elevated fasting glucagon and delayed glucagon suppression was overrepresented with prediabetes/diabetes, but this was only detected in 21% of this group. It also occurred in 8% of the control group. Conclusions We conclude that hyperglucagonemia may contribute to type 2 diabetes in a subgroup of affected individuals but that it is not a sine qua non for the disease. This should be considered in future pathophysiological studies and when testing pharmacotherapies addressing glucagon signaling. Glucagon is the main antagonist of insulin. It raises plasma glucose by reducing glycolysis and increasing gluconeogenesis and glycogenolysis (1, 2). Glucagon secretion from α cells is triggered by hypoglycemia and inhibited by insulin from neighboring β cells. In turn, glucagon inhibits insulin secretion (2). Unger et al. (3, 4) first postulated that elevated glucagon is a sine qua non in the development of diabetes. This marked the departure from an insulinocentric concept of type 2 diabetes pathogenesis to a bihormonal or even glucagonocentric model (4). In a glucagonocentric model, most metabolic derangements of diabetes are caused by the disinhibition of glucagon secretion (resulting from insulin-resistant α cells or impaired insulin release), but not directly by insufficient insulin action in other tissues (4, 5). The issue of the different pathophysiologic models remains unresolved, at least in part due to technical difficulties: glucagon is unstable, difficult to measure because of many similar peptides in plasma (2, 6), and its concentration is very low (7). Furthermore, α cells are harder to isolate than their insulin-producing neighbors (8). This impedes cellular studies. Current data on plasma glucagon levels in (pre)diabetic human subjects are also inconsistent. Several studies have found impaired glucagon suppression during an oral glucose tolerance test (oGTT) in prediabetic and diabetic individuals when compared with healthy controls (9–11). Other studies reported on increased fasting glucagon levels (12, 13). In contrast, Ahrén and Larsson (14) saw no differences between impaired glucose tolerance (IGT) and normoglycemic subjects, and Wagner et al. (15) observed rising glucagon values during an oGTT in 21% to 34% of healthy, insulin-sensitive individuals. These authors even found that this pattern predicted future metabolic health. New sandwich enzyme-linked immunosorbent assays (ELISAs) with improved specificity for glucagon became available recently, and this prompted us to re-examine the issue in a postpregnancy cohort of young women. We compared three groups of study participants with different metabolic phenotypes: a control group (normoglycemic women, who had recently completed a normoglycemic pregnancy), a normoglycemic high-risk group for type 2 diabetes [normoglycemic women after a recent pregnancy complicated by gestational diabetes (GDM)] (16, 17), and a prediabetes/diabetes group (women with prediabetes or screening-diagnosed type 2 diabetes after GDM). We first confirmed that average fasting plasma glucagon was higher and glucagon suppression during an oGTT was impaired in the normoglycemic high-risk and the prediabetes/diabetes groups, similar to what was seen in the majority of previous studies. However, our main research goal was to determine whether glucagon dynamics within each metabolic group were homogeneous or followed heterogeneous patterns. We used functional data analysis and unsupervised clustering to address this question. Research Design and Methods Study cohort Study participants were women enrolled in the prospective, monocenter observational study Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes (PPSDiab) between November 2011 and May 2016 (18). The cohort includes women with GDM during their last pregnancy and women following a normoglycemic pregnancy in a 2:1 ratio, recruited consecutively from the diabetes center and the obstetrics department of the University Hospital (Klinikum der Universität München) in Munich, Germany. Premenopausal women 3 to 16 months after a singleton (n = 295) or twin (n = 9) pregnancy with live birth(s) were eligible to participate. The GDM diagnosis was based on a 75-g oGTT with cut-off values for GDM according to the International Association of the Diabetes and Pregnancy Study Groups recommendations (plasma glucose: fasting 92 mg/dL, 1 hour 180 mg/dL, and 2 hours 153 mg/dL). Women without a history of GDM and either a normal 75-g oGTT (n = 294) or a normal 50-g screening oGTT (<135 mg/dL plasma glucose after 1 hour, n = 10) after the 23rd week of gestation were included in the normoglycemic group. Exclusion criteria for this study were alcohol or substance abuse, prepregnancy diabetes, and chronic diseases requiring continuous medication, except for hypothyroidism (n = 52), bronchial asthma (n = 8), mild hypertension (n = 4), gastroesophageal reflux (n = 2), and history of pulmonary embolism resulting in rivaroxaban prophylaxis (n = 1). Written informed consent was obtained from all study participants, and the protocol was approved by the ethical review committee of the Ludwig-Maximilians-Universität (study ID 300-11). Data used in this analysis were collected at the baseline visit of the PPSDiab study, 3 to 16 months after the index pregnancy. In addition to the baseline visit, post-GDM women also attended yearly follow-up visits with a 75-g oGTT. Groups We compared three groups of women: a control group (women normoglycemic at the baseline visit and after a normoglycemic pregnancy), a normoglycemic high-risk group (women normoglycemic at the baseline visit but with GDM during the preceding pregnancy), and a prediabetes/diabetes group [women with impaired fasting glucose (IFG), IGT, combined IFG plus IGT, or screening-diagnosed type 2 diabetes at the baseline visit and with GDM during the preceding pregnancy]. IFG [fasting plasma glucose ≥100 mg/dL (5.6 mmol/L)], IGT [2-hour plasma glucose ≥140 mg/dL (7.8 mmol/L)], and diabetes [fasting plasma glucose ≥126 mg/dL (7.0 mmol/L) or 2-hour plasma glucose ≥200 mg/dL (11.0 mmol/L)] were defined according to the criteria of the American Diabetes Association (19). Measurements We conducted a five-point 75-g oGTT with measurement of plasma glucose (Glucose HK Gen.3; Roche Diagnostics, Mannheim, Germany), serum insulin (chemiluminescent immunoassay; DiaSorin LIASON Systems, Saluggia, Italy), high-sensitivity C-reactive protein (wide-range C-reactive protein; Siemens Health Care Diagnostics, Erlangen, Germany), and blood lipids [low-density lipoprotein and high-density lipoprotein (HDL) cholesterol and triglycerides] (enzymatic caloric test; Roche Diagnostics, Mannheim, Germany) after an overnight fast. Plasma glucagon was measured at all five time points of oGTT with an ELISA (Glucagon ELISA; Mercodia, Uppsala, Sweden; catalog no: 10-1271-01) and also a radioimmunoassay (RIA) (Merck Millipore, Darmstadt, Germany; catalog no: GL-32K) for 283 subjects. ELISA and RIA measurements gave different results (Supplemental Table 1; Supplemental Fig. 1). In particular, suppression of plasma glucagon during the oGTT was insufficiently represented in the RIA measurement. Sensitivity and specificity of ELISA for pancreatic glucagon (amino acids 33 to 61) have been proven to be superior to RIA (20, 21). Thus, for this analysis, we exclusively used glucagon data measured by ELISA (n = 299). Plasma for glucagon measurements was collected in BD p800 tubes (BD Biosciences, San Jose, CA), which contain specific proteinase inhibitors to stabilize glucagon and other metabolically important hormones. Plasma was immediately separated by centrifugation and directly frozen in aliquots on dry ice, before being transferred to a –80°C freezer within 1 hour from completion of the oGTT. Glucagon measurements were done in one batch and only from aliquots that had not been thawed previously. Height and waist circumference were measured to the nearest 1 cm. Body mass and body fat mass were determined by a bioelectrical impedance analysis scale (Tanita BC-418; Tanita Corporation, Tokyo, Japan) (22, 23). Blood pressure was calculated as the mean out of two measurements in a resting seated position. In addition to these basic tests, all study subjects were asked to participate in a magnetic resonance imaging (MRI) measurement and an intravenous glucose tolerance test (ivGTT) on a voluntary basis. MRI (3 Tesla System, Ingenia, or Achieva; Philips Health Care, Hamburg, Germany) included determination of abdominal visceral adipose tissue volumes and liver fat content, using an mDixon low-fat fraction map. In the ivGTT, a glucose bolus of 0.3 g/kg body weight was injected over 1 minute with subsequent frequent blood sampling at 0, 2, 4, 6, 8, 10, 20, 30, 45, and 60 minutes. The measurements were used for the calculation of first-phase insulin response. A detailed description of the study design, anthropometric, clinical, and MRI measurements, and methodologies of blood sampling and analysis can be found elsewhere (24). Calculations   Mean blood pressure=(diastolic value*2+systolic value)/3 The insulin sensitivity index (ISI) according to Matsuda and DeFronzo (25) was calculated from the oGTT:  ISI=10000/√[(glucose 0,*insulin 0,)*(glucose 0,+2*(glucose 30,+60,+90,)+glucose120,)/8*(insulin0,+2*(insulin30, + 60, + 90,)+insulin 120,)/8]The disposition index (DI) was calculated as (26):  DI=ISI*IR30with  IR30=insulin 30,−insulin0,ISI and insulin release 0’ to 30’ in the oGTT were previously validated with data from ivGTT-euglycemic clamp tests in this cohort (24). Glucagon suppression indices were calculated as (27):  Early suppression=(1-[glucagon 30,/glucagon 0,])*100%  Late suppression=(1-[glucagon 120,/glucagon 30,])*100%  Overall suppression=(1-[glucagon 120,/glucagon 0,])*100%Area under the glucagon curve was calculated using the trapezoidal rule. First-phase insulin response in the ivGTT test was calculated as the incremental area under the insulin curve from 0 to 10 minutes. Statistical analysis All metric and normally distributed variables are reported as mean ± standard deviation; nonnormally distributed variables are presented as median (first quartile to third quartile). Categorical variables are presented as frequency and percentage. The Kruskal-Wallis test was used to compare metric variables, and the χ2 or Fisher’s exact test was used to compare categorical variables. For post hoc analysis, Dunn’s test was used. P values <0.05 were considered statistically significant. Functional data analysis methods were used for the analysis of the oGTT measurements (28). In the first step, the five-point oGTT measurements were converted into continuous, smooth curves based on B-spline basis functions (29). Afterward, a functional principal component analysis was performed based on the fitted curves to analyze the temporal variation (28). In the next step, a cluster analysis was conducted to identify patients with similar plasma glucagon dynamics. Hierarchical clustering was performed on the first three principal components of the functional principal component analysis via the Hierarchical Clustering on Principal Components function of Husson et al. (30). Hierarchical clustering was performed using the Ward’s criterion on the selected principal components. The number of clusters was chosen based on the growth of between-inertia. For the final partitioning, the k-means algorithm was performed with the partition obtained from the hierarchical tree as the initial partition. All statistical calculations were performed using SAS statistical software package version 9.3 (SAS Institute, Inc., Cary, NC) or R version 3.1.3 (www.r-project.org). Results Mean glucagon curves differ between metabolic groups We recruited 304 women into the PPSDiab study cohort but excluded 19 from this analysis. Two women were excluded because of type 1 diabetes diagnosed during follow-up, two because of overt hyperthyroidism, and one because of an acute upper respiratory infection at baseline. Eight women were excluded from the control group due to pathological glucose tolerance at the baseline visit, and six women were excluded due to missing glucagon values. Our final sample consisted of 285 study participants: 93 normoglycemic women after a normoglycemic pregnancy (control group), 121 normoglycemic women who had GDM (normoglycemic high-risk group), and 71 women with IFG, IGT, or newly diagnosed type 2 diabetes (prediabetes/diabetes group). Baseline characteristics of the study cohort are shown in Table 1. Mean age and low-density lipoprotein cholesterol were comparable, but mean blood pressure, waist circumference, triglycerides, c-reactive protein, liver fat content, intra-abdominal fat, and fasting and 2-hour plasma glucose increased and HDL cholesterol and insulin sensitivity decreased from the control over the normoglycemic high-risk to the prediabetes/diabetes groups (all significant over the three groups; results of pairwise post hoc tests shown in Table 1). Table 1. Baseline Characteristics of the PPSDiab Study Sample   Control  Normoglycemic High-Risk  Prediabetes/Diabetes  P Value  n  93  121  71    Glucose status           NGT  93 (100.0%)  121 (100.0%)  —     IFG  —  —  31 (43.7%)     IGT  —  —  22 (31.0%)     IFG + IGT  —  —  12 (16.9%)     Type 2 diabetes  —  —  6 (8.5%)    Age (y)  35.3 ± 4.2  35.2 ± 4.5  35.9 ± 4.5  0.6204  Mean blood pressure (mm Hg) (missing = 1)  85.8 ± 9.0  89.0 ± 8.6a  90.9 ± 10.3a  0.0026  BMI (kg/m2) (missing = 4)  23.7 ± 4.0  25.2 ± 5.8  28.2 ± 7.1a,b  0.0001  Waist circumference (cm) (missing = 5)  78.1 ± 8.9  80.7 ± 11.2  86.6 ± 13.2a,b  0.0002  hsCRP (mg/dL)  0.04 (0.01–0.08)  0.06 (0.02–0.25)a  0.09 (0.02–0.30)a  0.0030  Triglycerides (mg/dL)  61.0 (51.0–77.0)  65.0 (50.0–87.0)  81.0 (62.0–130.0)a,b  <0.0001  HDL cholesterol (mg/dL)  64.0 (57.0–73.0)  63.0 (56.0–73.0)  56.0 (46.0–65.0)a,b  <0.0001  LDL cholesterol (mg/dL)  104.0 (88.0–118.0)  105.0 (89.0–120.0)  104.0 (85.0–124.0)  0.9035  Plasma glucose 0 min (mg/dL)  89.0 (83.0–92.0)  91.0 (87.0–95.0)  102.0 (97.0–106.0)a,b  <0.0001  Plasma glucose 120 min (mg/dL)  93.0 (81.0–108.0)  114.0 (96.0–122.0)a  141.0 (113.0–165.0)a,b  <0.0001  ISI ( missing = 1)  6.8 (5.2–8.6)  5.5 (3.7–7.5)a  3.3 (2.1–4.6)a,b  <0.0001  DI (missing = 1)  297.4 (221.4–363.1)  246.6 (179.7–322.0)  160.0 (111.4–207.6)a,b  <0.0001  FPIR (missing = 152)  2.2 (1.4–3.5)  2.2 (1.6–3.5)  2.3 (1.5–3.9)  0.8218  Liver fat content (%) (missing = 132)  0.2 (0.0–0.8)  0.5 (0.0–1.1)  1.7 (0.0–4.1)a,b  0.0122  Intra-abdominal fat (L) (missing = 124)  1.4 (0.9–2.1)  1.8 (1.1–2.9)a  2.3 (1.3–3.2)a  0.0046  Glucagon 0 min (pmol/L)  6.0 (4.6–8.2)  6.6 (4.5–8.4)  7.7 (5.6–11.2)a,b  0.0069  Glucagon 30 min (pmol/L)  3.0 (2.4–4.7)  3.7 (2.5–4.9)  5.0 (3.0–7.6)a,b  <0.0001  Glucagon 60 min (pmol/L)  1.9 (1.4–3.1)  2.6 (1.8–3.7)  2.9 (2.0–4.4)a  0.0009  Glucagon 90 min (pmol/L)  2.1 (1.3–3.0)  2.1 (1.6–3.2)  2.5 (1.8–3.9)  0.0527  Glucagon 120 min (pmol/L)  2.3 (1.4–3.5)  2.2 (1.5–3.3)  2.3 (1.6–3.5)  0.5239  AUC glucagon  339.4 (248.5–473.6)  392.1 (283.5–518.2)  511.5 (353.4–615.2)a,b  0.0006  Early-suppression glucagon (0–30) (%)  47.6 (32.8–57.9)  41.3 (22.9–58.3)  32.0 (14.5–51.3)a  0.0055  Late-suppression glucagon (30–120) (%)  31.8 (8.9–49.6)  40.9 (14.9–56.7)  47.4 (33.3–63.6)a,b  <0.0001  Suppression glucagon (0–120) (%)  61.2 (48.2–76.9)  64.1 (49.5–74.4)  68.5 (57.3–75.0)  0.3130    Control  Normoglycemic High-Risk  Prediabetes/Diabetes  P Value  n  93  121  71    Glucose status           NGT  93 (100.0%)  121 (100.0%)  —     IFG  —  —  31 (43.7%)     IGT  —  —  22 (31.0%)     IFG + IGT  —  —  12 (16.9%)     Type 2 diabetes  —  —  6 (8.5%)    Age (y)  35.3 ± 4.2  35.2 ± 4.5  35.9 ± 4.5  0.6204  Mean blood pressure (mm Hg) (missing = 1)  85.8 ± 9.0  89.0 ± 8.6a  90.9 ± 10.3a  0.0026  BMI (kg/m2) (missing = 4)  23.7 ± 4.0  25.2 ± 5.8  28.2 ± 7.1a,b  0.0001  Waist circumference (cm) (missing = 5)  78.1 ± 8.9  80.7 ± 11.2  86.6 ± 13.2a,b  0.0002  hsCRP (mg/dL)  0.04 (0.01–0.08)  0.06 (0.02–0.25)a  0.09 (0.02–0.30)a  0.0030  Triglycerides (mg/dL)  61.0 (51.0–77.0)  65.0 (50.0–87.0)  81.0 (62.0–130.0)a,b  <0.0001  HDL cholesterol (mg/dL)  64.0 (57.0–73.0)  63.0 (56.0–73.0)  56.0 (46.0–65.0)a,b  <0.0001  LDL cholesterol (mg/dL)  104.0 (88.0–118.0)  105.0 (89.0–120.0)  104.0 (85.0–124.0)  0.9035  Plasma glucose 0 min (mg/dL)  89.0 (83.0–92.0)  91.0 (87.0–95.0)  102.0 (97.0–106.0)a,b  <0.0001  Plasma glucose 120 min (mg/dL)  93.0 (81.0–108.0)  114.0 (96.0–122.0)a  141.0 (113.0–165.0)a,b  <0.0001  ISI ( missing = 1)  6.8 (5.2–8.6)  5.5 (3.7–7.5)a  3.3 (2.1–4.6)a,b  <0.0001  DI (missing = 1)  297.4 (221.4–363.1)  246.6 (179.7–322.0)  160.0 (111.4–207.6)a,b  <0.0001  FPIR (missing = 152)  2.2 (1.4–3.5)  2.2 (1.6–3.5)  2.3 (1.5–3.9)  0.8218  Liver fat content (%) (missing = 132)  0.2 (0.0–0.8)  0.5 (0.0–1.1)  1.7 (0.0–4.1)a,b  0.0122  Intra-abdominal fat (L) (missing = 124)  1.4 (0.9–2.1)  1.8 (1.1–2.9)a  2.3 (1.3–3.2)a  0.0046  Glucagon 0 min (pmol/L)  6.0 (4.6–8.2)  6.6 (4.5–8.4)  7.7 (5.6–11.2)a,b  0.0069  Glucagon 30 min (pmol/L)  3.0 (2.4–4.7)  3.7 (2.5–4.9)  5.0 (3.0–7.6)a,b  <0.0001  Glucagon 60 min (pmol/L)  1.9 (1.4–3.1)  2.6 (1.8–3.7)  2.9 (2.0–4.4)a  0.0009  Glucagon 90 min (pmol/L)  2.1 (1.3–3.0)  2.1 (1.6–3.2)  2.5 (1.8–3.9)  0.0527  Glucagon 120 min (pmol/L)  2.3 (1.4–3.5)  2.2 (1.5–3.3)  2.3 (1.6–3.5)  0.5239  AUC glucagon  339.4 (248.5–473.6)  392.1 (283.5–518.2)  511.5 (353.4–615.2)a,b  0.0006  Early-suppression glucagon (0–30) (%)  47.6 (32.8–57.9)  41.3 (22.9–58.3)  32.0 (14.5–51.3)a  0.0055  Late-suppression glucagon (30–120) (%)  31.8 (8.9–49.6)  40.9 (14.9–56.7)  47.4 (33.3–63.6)a,b  <0.0001  Suppression glucagon (0–120) (%)  61.2 (48.2–76.9)  64.1 (49.5–74.4)  68.5 (57.3–75.0)  0.3130  Abbreviations: AUC, area under the curve; BMI, body mass index; FPIR, first-phase insulin response; hsCRP, high-sensitivity C-reactive protein; LDL, low-density lipoprotein; NGT, normal glucose tolerance. a Significant post hoc tests vs control. b Significant post hoc tests vs normoglycemic high-risk. View Large We next compared plasma glucagon levels during the oGTT in the three groups (Table 1). Fasting plasma glucagon was significantly elevated, and early glucagon suppression was diminished in the prediabetes/diabetes group compared with the control group [median (Q1 to Q3) for fasting plasma glucagon: 6.0 (4.6 to 8.2) (pmol/L) vs 7.7 (5.6 to 11.2) (pmol/L); early glucagon suppression: 47.6 (32.8 to 57.9) (pmol/L) vs 32.0 (14.5 to 51.3) (pmol/L), respectively]. The normoglycemic high-risk group lay in between for these variables, but closer to the control group and not statistically different from it [median (Q1 to Q3) for fasting plasma glucagon: 6.6 (4.5 to 8.4) (pmol/L); early glucagon suppression: 41.3 (22.9 to 58.3) (pmol/L)] (Fig. 1; Table 1). Total glucagon suppression was similar in all three groups. Figure 1. View largeDownload slide Glucagon during oGTT stratified by risk groups (blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Figure 1. View largeDownload slide Glucagon during oGTT stratified by risk groups (blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Similar to a recent publication by Faerch et al. (27), we further examined fasting glucagon values and glucagon suppression indices in women with isolated IFG compared with those with isolated IGT and combined IFG + IGT (Supplemental Fig. 2; Supplemental Table 2). Late and overall glucagon suppression was smaller in women with isolated IFG compared with both other groups [median (Q1 to Q3) late suppression: 41.8 (16.5 to 50.4) (%) vs 58.1 (43.1 to 71.3) (%) vs 58.9 (46.1 to 69.6) (%) and overall suppression: 58.9 (39.8 to 70.2) (%) vs 71.2 (68.4 to 81.0) (%) vs 73.7 (63.8 to 81.0) (%) in IFG vs IGT vs IFG + IGT, respectively). We observed no significant differences in early glucagon suppression and fasting glucagon. Plasma glucagon patterns are heterogeneous within each metabolic group The five-point glucagon curves in response to oral glucose were heterogeneous between individuals (Fig. 2a). To examine this further, we calculated continuous, smooth curves from the five measurements during the oGTT based on B-spline basis functions (Fig. 2a). Stratified by group, these curves confirmed within-group heterogeneity of plasma glucagon dynamics (Supplemental Fig. 3). To permit pattern identification, we added a principal component analysis of the curves. The first three principal component factors explained 79%, 17%, and 3% of curve variance, respectively (Fig. 2b). We used these three principal components as input for an unsupervised cluster analysis (Fig. 2c). This identified four clusters corresponding to four distinct patterns of plasma glucagon dynamics (Fig. 2d). Figure 2. View largeDownload slide Process of functional data analysis. (a) Based on the five-point oGTT data curves, continuous, smooth curves were calculated (median indicated by black line). (b) Then, a principal component analysis of the curves was conducted (median indicated by solid line; extremes indicated by dotted lines). (c) The three principal components were used as input for an unsupervised cluster analysis (asterisk indicates line types used to represent the clusters in Fig. 3). (d) Fitted glucagon curves during oGTT stratified by the four clusters (colors: original risk groups as used in Table 1 and Fig. 1; blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Figure 2. View largeDownload slide Process of functional data analysis. (a) Based on the five-point oGTT data curves, continuous, smooth curves were calculated (median indicated by black line). (b) Then, a principal component analysis of the curves was conducted (median indicated by solid line; extremes indicated by dotted lines). (c) The three principal components were used as input for an unsupervised cluster analysis (asterisk indicates line types used to represent the clusters in Fig. 3). (d) Fitted glucagon curves during oGTT stratified by the four clusters (colors: original risk groups as used in Table 1 and Fig. 1; blue = controls, gray = normoglycemic high-risk, red = prediabetes/diabetes). Cluster 3 was the largest (n = 188; Table 2) and showed low mean fasting glucagon and rapid suppression during the oGTT (Figs. 2d and 3a ). Cluster 2, the second largest (n = 62), had higher mean fasting glucagon but equally rapid suppression. Cluster 1 (n = 21) had high mean fasting glucagon and delayed suppression, and cluster 4 (n = 7) had low mean fasting glucagon and a rising curve after glucose ingestion (Fig. 3a; Table 2). Table 2. Baseline Characteristics of the PPSDiab Study Sample, Stratified by Clusters of Glucagon Dynamics   Cluster 1  Cluster 2  Cluster 3  Cluster 4  P Value  n  28  62  188  7    Risk group             Control  7 (25.0%)  19 (30.7%)  65 (34.6%)  2 (28.6%)  0.0279   Normoglycemic high-risk  6 (21.4%)  27 (43.6%)  84 (44.7%)  4 (57.1%)     Prediabetes/diabetes  15 (53.6%)  16 (25.8%)  39 (20.7%)  1 (14.3%)    Glucose status             NGT  13 (46.4%)  46 (74.2%)  149 (79.3%)  6 (85.7%)  0.0099   IFG  5 (17.9%)  6 (9.7%)  19 (10.1%)  1 (14.3%)     IGT  3 (10.7%)  7 (11.3%)  12 (6.4%)  0     IFG + IGT  3 (10.7%)  3 (4.8%)  6 (3.2%)  0     Type 2 diabetes  4 (14.3%)  0  2 (1.1%)  0    Age (y)  33.5 ± 4.8  35.5 ± 4.4  35.7 ± 4.3a  35.0 ± 4.0  0.0315  Mean blood pressure (mm Hg) (missing = 1)  96.2 ± 8.6  89.4 ± 9.2  87.0 ± 9.0  85.6 ± 7.4  <0.0001  BMI (kg/m2) (missing = 4)  33.3 ± 6.1  26.5 ± 6.4a  24.0 ± 4.6a  21.6 ± 1.5a  <0.0001  Waist circumference (cm) (missing = 5)  96.0 ± 11.9  83.8 ± 12.3a  78.6 ± 9.3a  73.5 ± 4.1a  <0.0001  hsCRP (mg/dL)  0.19 (0.07–0.47)  0.05 (0.01–0.17)a  0.04 (0.01–0.12)a  0.12 (0.05–0.38)  0.0004  Triglycerides (mg/dL)  91.5 (58.5–132.0)  62.5 (53.0–83.0)  67.5 (53.0–88.5)  63.0 (58.0–91.0)  0.0898  HDL cholesterol (mg/dL)  49.0 (44.5–61.5)  62.0 (51.0–73.0)  63.0 (56.0–73.0)  65.0 (56.0–70.0)  0.0012  Plasma glucose 0 min (mg/dL)  97.5 (90.5–106.0)  91.0 (88.0–97.0)  91.0 (86.0–97.0)  87.0 (82.0–92.0)  0.0078  Plasma glucose 120 min (mg/dL)  127.0 (115.5–154.5)  113.5 (95.0–130.0)a  106.5 (90.0–121.5)a  80.0 (74.0–92.0)a,b  <0.0001  ISI (missing = 1)  2.5 (1.9–4.3)  5.0 (3.3–6.9)a  5.8 (4.2–8.1)a  7.9 (5.6–8.3)a  <0.0001  DI (missing = 1)  152.0 (96.5–247.8)  230.2 (165.3–392.0)a  252.8 (176.7–324.4)a  232.9 (156.2–276.4)  0.0007  IR30 (missing = 1)  55.7 (37.1–82.2)  50.3 (36.4–86.1)  41.6 (30.9–60.1)  28.7 (26.2–41.3)a,b  0.0023  FPIR (missing = 152)c  3.9 (2.2–6.2)  3.3 (2.2–4.3)  2.1 (1.4–3.1)  2.1 (1.0–2.7) (n = 3)  0.0140  Liver fat content (%) (missing = 131)  2.4 (1.1–6.4)  0.7 (0.0–1.7)a  0.3 (0.0–0.8)a  0.1 (0.0–0.5)a  <0.0001  Intra-abdominal fat (L) (missing = 124)  3.4 (2.9–4.4)  2.0 (1.5–3.0)a  1.5 (1.0–2.3)a,b  1.1 (0.9–1.6)a,b  <0.0001    Cluster 1  Cluster 2  Cluster 3  Cluster 4  P Value  n  28  62  188  7    Risk group             Control  7 (25.0%)  19 (30.7%)  65 (34.6%)  2 (28.6%)  0.0279   Normoglycemic high-risk  6 (21.4%)  27 (43.6%)  84 (44.7%)  4 (57.1%)     Prediabetes/diabetes  15 (53.6%)  16 (25.8%)  39 (20.7%)  1 (14.3%)    Glucose status             NGT  13 (46.4%)  46 (74.2%)  149 (79.3%)  6 (85.7%)  0.0099   IFG  5 (17.9%)  6 (9.7%)  19 (10.1%)  1 (14.3%)     IGT  3 (10.7%)  7 (11.3%)  12 (6.4%)  0     IFG + IGT  3 (10.7%)  3 (4.8%)  6 (3.2%)  0     Type 2 diabetes  4 (14.3%)  0  2 (1.1%)  0    Age (y)  33.5 ± 4.8  35.5 ± 4.4  35.7 ± 4.3a  35.0 ± 4.0  0.0315  Mean blood pressure (mm Hg) (missing = 1)  96.2 ± 8.6  89.4 ± 9.2  87.0 ± 9.0  85.6 ± 7.4  <0.0001  BMI (kg/m2) (missing = 4)  33.3 ± 6.1  26.5 ± 6.4a  24.0 ± 4.6a  21.6 ± 1.5a  <0.0001  Waist circumference (cm) (missing = 5)  96.0 ± 11.9  83.8 ± 12.3a  78.6 ± 9.3a  73.5 ± 4.1a  <0.0001  hsCRP (mg/dL)  0.19 (0.07–0.47)  0.05 (0.01–0.17)a  0.04 (0.01–0.12)a  0.12 (0.05–0.38)  0.0004  Triglycerides (mg/dL)  91.5 (58.5–132.0)  62.5 (53.0–83.0)  67.5 (53.0–88.5)  63.0 (58.0–91.0)  0.0898  HDL cholesterol (mg/dL)  49.0 (44.5–61.5)  62.0 (51.0–73.0)  63.0 (56.0–73.0)  65.0 (56.0–70.0)  0.0012  Plasma glucose 0 min (mg/dL)  97.5 (90.5–106.0)  91.0 (88.0–97.0)  91.0 (86.0–97.0)  87.0 (82.0–92.0)  0.0078  Plasma glucose 120 min (mg/dL)  127.0 (115.5–154.5)  113.5 (95.0–130.0)a  106.5 (90.0–121.5)a  80.0 (74.0–92.0)a,b  <0.0001  ISI (missing = 1)  2.5 (1.9–4.3)  5.0 (3.3–6.9)a  5.8 (4.2–8.1)a  7.9 (5.6–8.3)a  <0.0001  DI (missing = 1)  152.0 (96.5–247.8)  230.2 (165.3–392.0)a  252.8 (176.7–324.4)a  232.9 (156.2–276.4)  0.0007  IR30 (missing = 1)  55.7 (37.1–82.2)  50.3 (36.4–86.1)  41.6 (30.9–60.1)  28.7 (26.2–41.3)a,b  0.0023  FPIR (missing = 152)c  3.9 (2.2–6.2)  3.3 (2.2–4.3)  2.1 (1.4–3.1)  2.1 (1.0–2.7) (n = 3)  0.0140  Liver fat content (%) (missing = 131)  2.4 (1.1–6.4)  0.7 (0.0–1.7)a  0.3 (0.0–0.8)a  0.1 (0.0–0.5)a  <0.0001  Intra-abdominal fat (L) (missing = 124)  3.4 (2.9–4.4)  2.0 (1.5–3.0)a  1.5 (1.0–2.3)a,b  1.1 (0.9–1.6)a,b  <0.0001  Abbreviations: BMI, body mass index; FPIR, first-phase insulin response; hsCRP, high-sensitivity C-reactive protein; IR30, insulin release 0’ to 30’ in the oGTT; NGT, normal glucose tolerance. a Significant post hoc test: significant vs cluster 1. b Significant post hoc test: significant vs cluster 2. c The post hoc test for FPIR was conducted both including cluster 4 and after exclusion of cluster 4 (due to the small group size in cluster 4); in any case, the post hoc test has not reached significance. View Large Figure 3. View largeDownload slide Means of (a) glucagon, (b) glucose, (c) insulin, and (d) c-peptide curves during oGTT stratified by the four clusters derived from the glucagon curves (Fig. 2). Figure 3. View largeDownload slide Means of (a) glucagon, (b) glucose, (c) insulin, and (d) c-peptide curves during oGTT stratified by the four clusters derived from the glucagon curves (Fig. 2). Cluster 1 contained the highest proportion of women from the prediabetes/diabetes group (53%), followed by cluster 2, cluster 3, and cluster 4. Women in cluster 1 had significantly higher body mass index, waist circumference, triglycerides, liver fat content, and intra-abdominal fat and lower HDL cholesterol and ISI than those in the other three clusters. The DI of cluster 1 was significantly lower than those of clusters 2 and 3 (Table 2). Cluster 4 included lean, insulin-sensitive women with a tendency toward low glucose values (Fig. 3b and 3c; Table 2). Discussion In our first analysis, we found that women with prediabetes/screening-diagnosed type 2 diabetes had higher fasting glucagon and delayed glucagon suppression during an oGTT compared with healthy control subjects (normoglycemic women after a normoglycemic pregnancy). Normoglycemic women after GDM, a high-risk group for type 2 diabetes (16, 17), lay in between, with values closer to and not statistically different from the control group. These results are in line with most previous studies that saw the highest fasting glucagon and most impaired glucagon suppression in subjects with diabetes, followed by those with prediabetes, and, at the low end, normoglycemic individuals (10–13, 27). In several nondiabetic cohorts, fasting glucagon was higher in insulin-resistant than in insulin-sensitive subjects (31–33). A majority of studies also found a positive association of plasma glucagon with obesity in groups with similar glucose tolerance (11, 13, 31). Some earlier studies had different findings. Ahrén and Larsson (14) reported that fasting and postprandial glucagon did not differ between IGT and normoglycemic subjects in 84 postmenopausal women. Wagner et al. (15) analyzed cohorts of nondiabetic individuals and found that, in 21% to 34% of subjects, glucagon was not suppressed until 120 minutes into the oGTT. These individuals were lean and insulin-sensitive, and also had a favorable prognosis of insulin sensitivity over time. In their recent study, Faerch et al. (27) described that glucagon curves differed between individuals with IFG and those with IGT. They found a smaller overall decrease in glucagon during an oGTT in the group with isolated IFG compared with isolated IGT and combined IFG + IGT. Our analysis confirms this result, with the difference in overall glucagon suppression mainly caused by the late phase of the oGTT (Supplemental Fig. 2; Supplemental Table 2). In our second analysis, we saw that plasma glucagon dynamics in the study cohort followed four different patterns, based on an unsupervised cluster analysis. The clusters detected did not fully or even closely match the predefined metabolic groups. We consider this the main finding of this paper. Subjects from the prediabetes/diabetes group were overrepresented in cluster 1 (with high fasting glucagon and diminished suppression), but still only made up 50% of this cluster, which also contained 25% control subjects. Conversely, the majority of women from the prediabetes/diabetes group (n = 39; 55%) fell into cluster 3, the “most normal” cluster (with low fasting glucagon and rapid suppression). Therefore, hyperglucagonemia was not a universal prerequisite for impaired glucose metabolism or early type 2 diabetes. It only affected a subgroup of individuals. Delayed glucagon suppression was clearly associated with obesity and metabolic syndrome markers in our study. This is evident from the clinical characteristics (e.g., waist circumference, blood lipids, plasma glucose, and intra-abdominal and liver fat) of the subjects in cluster 1 compared with the other clusters (Table 2). Hepatic steatosis may even be a cause of hyperglucagonemia, as it disrupts hepatic glucagon sensitivity and probably leads to reactive hypersecretion of the hormone (34). The association of liver fat and hyperglucagonemia was found independent of the presence of disrupted glucose metabolism (34, 35). Impaired early insulin secretion could be another cause of delayed postprandial glucagon suppression, but we do not find evidence for this relationship. Early insulin and c-peptide levels during the oGTT and first-phase insulin secretion in the ivGTT were not reduced in the women in cluster 1. The reduced DI results from lower insulin sensitivity (ISI) in this cluster, but not from reduced early insulin secretion (insulin release 0’ to 30’ in the oGTT) (Table 2). The α cell resistance to inhibition by insulin or a reactive glucagon hypersecretion due to a resistance of the liver is therefore the most likely explanations for our findings. .Another noteworthy observation was the small cluster 4 (n = 7; 2.5% of participants), with low fasting glucagon, but rising glucagon levels during the oGTT. The women in this cluster were lean and insulin-sensitive and had low glucose levels. In this group, the rising glucagon probably is a physiologic response to avoid postchallenge hypoglycemia as a result of an overactive insulin response, which is not uncommon in lean, young women (36). Wagner et al. (15) associated rising glucagon during an oGTT with a favorable metabolic prognosis. Our small and probably not representative sample does not confirm this finding. Five of the 7 women in cluster 4 had had GDM (Table 2), and all of these 5 women developed prediabetes or diabetes during the follow-up of this study (mean duration of follow-up was 38.2 months; data not shown). In our cohort, this phenotype is also much less common than reported in the previous publication. However, given the small number of subjects in cluster 4, we find these observations interesting and worth following up on, but we do not claim that they constitute scientific evidence by themselves. Finally, we believe it is important to use highly specific glucagon assays, in particular to study postprandial glucagon dynamics. We initially used a standard RIA, which strongly underestimated glucagon suppression (Supplemental Fig. 1). This was probably due to cross-reactivity with other peptides cleaved from proglucagon, such as oxyntomodulin, glicentin 1-61 (N-terminally elongated glucagon), and miniglucagon. Intestinal secretion of these peptides increases in the postprandial state, masking glucagon suppression (21, 37–39). Sandwich ELISAs, with antibodies against the N- and the C-terminal end of the glucagon molecule, circumvent this problem. Strengths of this study are optimal preanalytic and analytic techniques plus a cohort homogeneous for age and sex and with little medication and concomitant diseases. We used functional data analysis to interpret glucagon dynamics and also consider this a strength of our work. This method can extract more of the information contained in a function than classic multivariate statistical techniques (40–42). Together with a subsequent cluster analysis, it permits the grouping of data sets according to their curve shapes. Using a recent history of GDM to identify a high-risk cohort early in the process of type 2 diabetes development should have limited secondary metabolic abnormalities to the minimal extent possible in a human study. At the same time, the study cohort can also be interpreted as a weakness, because results may not apply to the general population. Another limitation of this analysis is its cross-sectional design, which precludes the clarification of cause-effect relationships. We conclude that fasting hyperglucagonemia and delayed postprandial glucagon suppression associate with insulin resistance, prediabetes, and diabetes, but are, in reality, only present in subgroups of individuals. Dysglycemia can develop without elevated plasma glucagon, and elevated glucagon does not preclude normoglycemia. Fasting hyperglucagonemia and delayed suppression are strongly linked to obesity and metabolic syndrome. Rising glucagon during an oGTT may be a rare phenomenon. It occurs in insulin-sensitive individuals with a tendency toward hypoglycemia, but does not necessarily indicate metabolic health. Our results have consequences for the pathophysiologic understanding of type 2 diabetes and for the development of precision treatments. At present, glucagon agonists and antagonists are evaluated for diabetes therapy (1, 2, 43, 44). Based on our findings, patients should probably be stratified by glucagon values for such treatments. For those patients with hyperglucagonemia, glucagon antagonists could be an appropriate therapy, whereas for others, agonists may be useful to induce beneficial effects mediated through the glucagon receptor, such as weight loss (2, 44). Abbreviations: DI disposition index ELISA enzyme-linked immunosorbent assay GDM gestational diabetes HDL high-density lipoprotein IFG impaired fasting glucose IGT impaired glucose tolerance ISI insulin sensitivity index ivGTT intravenous glucose tolerance test MRI magnetic resonance imaging oGTT oral glucose tolerance test PPSDiab Prediction, Prevention, and Subclassification of gestational and type 2 Diabetes RIA radioimmunoassay. Acknowledgments We thank all participants in the PPSDiab study and to the diabetes care team of the Medizinische Klinik IV. Financial Support: This work was supported by the Helmholtz Zentrum für München, Klinikum der Universität München, and the German Center for Diabetes Research (to A.L.). Author Contributions: Conceptualization and methodology, A.L. and J.S.; formal analysis, M.R.; investigation, C.G., V.S., F.B., S.M., N.H., and D.P.; resources, C.H.; data curation, V.S. and C.G.; writing—original draft, C.G.; writing—review and editing, A.L.; and visualization, M.R., C.G., and A.L. All authors reviewed and approved the final version of the manuscript, revising it critically for important intellectual content. Disclosure Summary: The authors have nothing to disclose. References 1. Jiang G, Zhang BB. Glucagon and regulation of glucose metabolism. Am J Physiol Endocrinol Metab . 2003; 284( 4): E671– E678. Google Scholar CrossRef Search ADS PubMed  2. Wewer Albrechtsen NJ, Kuhre RE, Pedersen J, Knop FK, Holst JJ. The biology of glucagon and the consequences of hyperglucagonemia. Biomarkers Med . 2016; 10( 11): 1141– 1151. Google Scholar CrossRef Search ADS   3. Lee Y, Berglund ED, Wang MY, Fu X, Yu X, Charron MJ, Burgess SC, Unger RH. Metabolic manifestations of insulin deficiency do not occur without glucagon action. Proc Natl Acad Sci USA . 2012; 109( 37): 14972– 14976. Google Scholar CrossRef Search ADS PubMed  4. Unger RH, Cherrington AD. Glucagonocentric restructuring of diabetes: a pathophysiologic and therapeutic makeover. J Clin Invest . 2012; 122( 1): 4– 12. Google Scholar CrossRef Search ADS PubMed  5. Braaten JT, Faloona GR, Unger RH. The effect of insulin on the alpha-cell response to hyperglycemia in long-standing alloxan diabetes. J Clin Invest . 1974; 53( 4): 1017– 1021. Google Scholar CrossRef Search ADS PubMed  6. Bataille D, Dalle S. The forgotten members of the glucagon family. Diabetes Res Clin Pract . 2014; 106( 1): 1– 10. Google Scholar CrossRef Search ADS PubMed  7. Ruiter M, La Fleur SE, van Heijningen C, van der Vliet J, Kalsbeek A, Buijs RM. The daily rhythm in plasma glucagon concentrations in the rat is modulated by the biological clock and by feeding behavior. Diabetes . 2003; 52( 7): 1709– 1715. Google Scholar CrossRef Search ADS PubMed  8. Gaisano HY, Macdonald PE, Vranic M. Glucagon secretion and signaling in the development of diabetes. Front Physiol . 2012; 3: 349. Google Scholar CrossRef Search ADS PubMed  9. Beis C, Grigorakis SI, Philippou G, Alevizaki M, Anastasiou E. Lack of suppression of plasma glucagon levels in late pregnancy persists postpartum only in women with previous gestational diabetes mellitus. Acta Diabetol . 2005; 42( 1): 31– 35. Google Scholar CrossRef Search ADS PubMed  10. Abdul-Ghani M, DeFronzo RA. Fasting hyperglycemia impairs glucose- but not insulin-mediated suppression of glucagon secretion. J Clin Endocrinol Metab . 2007; 92( 5): 1778– 1784. Google Scholar CrossRef Search ADS PubMed  11. Knop FK, Aaboe K, Vilsbøll T, Vølund A, Holst JJ, Krarup T, Madsbad S. Impaired incretin effect and fasting hyperglucagonaemia characterizing type 2 diabetic subjects are early signs of dysmetabolism in obesity. Diabetes Obes Metab . 2012; 14( 6): 500– 510. Google Scholar CrossRef Search ADS PubMed  12. Shen J, Chen Z, Chen C, Zhu X, Han Y. Impact of incretin on early-phase insulin secretion and glucose excursion. Endocrine . 2013; 44( 2): 403– 410. Google Scholar CrossRef Search ADS PubMed  13. Manell H, Staaf J, Manukyan L, Kristinsson H, Cen J, Stenlid R, Ciba I, Forslund A, Bergsten P. Altered plasma levels of glucagon, GLP-1 and glicentin during OGTT in adolescents with obesity and type 2 diabetes. J Clin Endocrinol Metab . 2016; 101( 3): 1181– 1189. Google Scholar CrossRef Search ADS PubMed  14. Ahrén B, Larsson H. Impaired glucose tolerance (IGT) is associated with reduced insulin-induced suppression of glucagon concentrations. Diabetologia . 2001; 44( 11): 1998– 2003. Google Scholar CrossRef Search ADS PubMed  15. Wagner R, Hakaste LH, Ahlqvist E, Heni M, Machann J, Schick F, Van Obberghen E, Stefan N, Gallwitz B, Tuomi T, Haring HU, Groop L, Fritsche A. Non-suppressed glucagon after glucose challenge as a potential predictor for glucose tolerance. Diabetes . 2017; 66( 5): 1373– 1379. Google Scholar CrossRef Search ADS PubMed  16. Bellamy L, Casas JP, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet . 2009; 373( 9677): 1773– 1779. Google Scholar CrossRef Search ADS PubMed  17. Feig DS, Zinman B, Wang X, Hux JE. Risk of development of diabetes mellitus after diagnosis of gestational diabetes. CMAJ . 2008; 179( 3): 229– 234. Google Scholar CrossRef Search ADS PubMed  18. Gar C, Rottenkolber M, Grallert H, Banning F, Freibothe I, Sacco V, Wichmann C, Reif S, Potzel A, Dauber V, Schendell C, Sommer NN, Wolfarth B, Seissler J, Lechner A, Ferrari U. Physical fitness and plasma leptin in women with recent gestational diabetes. PLoS One . 2017; 12( 6): e0179128. Google Scholar CrossRef Search ADS PubMed  19. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care . 2011; 34( Suppl 1): S62– S69. CrossRef Search ADS PubMed  20. Raines M, King C, Christensen S, Amin P, Schranz D. 2014 Glucagon quantification: comparison of radioimmunoassay and sandwich ELISA methods. In: American Association for Clinical Chemistry : AACC; Poster A-213. 21. Matsuo T, Miyagawa J, Kusunoki Y, Miuchi M, Ikawa T, Akagami T, Tokuda M, Katsuno T, Kushida A, Inagaki T, Namba M. Postabsorptive hyperglucagonemia in patients with type 2 diabetes mellitus analyzed with a novel enzyme-linked immunosorbent assay. J Diabetes Investig . 2016; 7( 3): 324– 331. Google Scholar CrossRef Search ADS PubMed  22. Bosy-Westphal A, Later W, Hitze B, Sato T, Kossel E, Gluer CC, Heller M, Muller MJ. Accuracy of bioelectrical impedance consumer devices for measurement of body composition in comparison to whole body magnetic resonance imaging and dual X-ray absorptiometry. Obes Facts . 2008; 1( 6): 319– 324. Google Scholar CrossRef Search ADS PubMed  23. Bosy-Westphal A, Schautz B, Later W, Kehayias JJ, Gallagher D, Müller MJ. What makes a BIA equation unique? Validity of eight-electrode multifrequency BIA to estimate body composition in a healthy adult population. Eur J Clin Nutr . 2013; 67( Suppl 1): S14– S21. Google Scholar CrossRef Search ADS PubMed  24. Rottenkolber M, Ferrari U, Holland L, Aertsen S, Kammer NN, Hetterich H, Fugmann M, Banning F, Weise M, Sacco V, Kohn D, Freibothe I, Hutter S, Hasbargen U, Lehmann R, Grallert H, Parhofer KG, Seissler J, Lechner A. The diabetes risk phenotype of young women with recent gestational diabetes. J Clin Endocrinol Metab . 2015; 100( 6): E910– E918. Google Scholar CrossRef Search ADS PubMed  25. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care . 1999; 22( 9): 1462– 1470. Google Scholar CrossRef Search ADS PubMed  26. Kahn SE, Prigeon RL, McCulloch DK, Boyko EJ, Bergman RN, Schwartz MW, Neifing JL, Ward WK, Beard JC, Palmer JP. Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function. Diabetes . 1993; 42( 11): 1663– 1672. Google Scholar CrossRef Search ADS PubMed  27. Færch K, Vistisen D, Pacini G, Torekov SS, Johansen NB, Witte DR, Jonsson A, Pedersen O, Hansen T, Lauritzen T, Jørgensen ME, Ahrén B, Holst JJ. Insulin resistance is accompanied by increased fasting glucagon and delayed glucagon suppression in individuals with normal and impaired glucose regulation. Diabetes . 2016; 65( 11): 3473– 3481. Google Scholar CrossRef Search ADS PubMed  28. Ramsay JO, Silverman BW. Functional Data Analysis. 2 . New York: Springer; 2005. Google Scholar CrossRef Search ADS   29. Ramsay JO, Hooker G, Graves S. Functional Data Analysis With R and MATLAB . New York: Springer Science & Business Media; 2009. CrossRef Search ADS   30. Husson F, Josse J, Pagès J. Principal component methods – hierarchical clustering – partitional clustering: why would we need to choose for visualizing data? Available at: http://factominer.free.fr/more/HCPC_husson_josse.pdf. Accessed 31 May 2017. 31. Calanna S, Piro S, Di Pino A, Maria Zagami R, Urbano F, Purrello F, Maria Rabuazzo A. Beta and alpha cell function in metabolically healthy but obese subjects: relationship with entero-insular axis. Obesity (Silver Spring) . 2013; 21( 2): 320– 325. Google Scholar CrossRef Search ADS PubMed  32. Ferrannini E, Muscelli E, Natali A, Gabriel R, Mitrakou A, Flyvbjerg A, Golay A, Hojlund K; Relationship between Insulin Sensitivity and Cardiovascular Disease Risk (RISC) Project Investigators. Association of fasting glucagon and proinsulin concentrations with insulin resistance. Diabetologia . 2007; 50( 11): 2342– 2347. Google Scholar CrossRef Search ADS PubMed  33. Weiss R, D’Adamo E, Santoro N, Hershkop K, Caprio S. Basal alpha-cell up-regulation in obese insulin-resistant adolescents. J Clin Endocrinol Metab . 2011; 96( 1): 91– 97. Google Scholar CrossRef Search ADS PubMed  34. Junker AE, Gluud L, Holst JJ, Knop FK, Vilsbøll T. Diabetic and nondiabetic patients with nonalcoholic fatty liver disease have an impaired incretin effect and fasting hyperglucagonaemia. J Intern Med . 2016; 279( 5): 485– 493. Google Scholar CrossRef Search ADS PubMed  35. Bernsmeier C, Meyer-Gerspach AC, Blaser LS, Jeker L, Steinert RE, Heim MH, Beglinger C. Glucose-induced glucagon-like Peptide 1 secretion is deficient in patients with non-alcoholic fatty liver disease. PLoS One . 2014; 9( 1): e87488. Google Scholar CrossRef Search ADS PubMed  36. Brun JF, Fedou C, Mercier J. Postprandial reactive hypoglycemia. Diabetes Metab . 2000; 26( 5): 337– 351. Google Scholar PubMed  37. Dunning BE, Gerich JE. The role of alpha-cell dysregulation in fasting and postprandial hyperglycemia in type 2 diabetes and therapeutic implications. Endocr Rev . 2007; 28( 3): 253– 283. Google Scholar CrossRef Search ADS PubMed  38. Orskov C, Jeppesen J, Madsbad S, Holst JJ. Proglucagon products in plasma of noninsulin-dependent diabetics and nondiabetic controls in the fasting state and after oral glucose and intravenous arginine. J Clin Invest . 1991; 87( 2): 415– 423. Google Scholar CrossRef Search ADS PubMed  39. Holst JJ, Christensen M, Lund A, de Heer J, Svendsen B, Kielgast U, Knop FK. Regulation of glucagon secretion by incretins. Diabetes Obes Metab . 2011; 13( Suppl 1): 89– 94. Google Scholar CrossRef Search ADS PubMed  40. Ferraty F, Mas A, Vieu P. Advances on nonparametric regression for functional variables. Aust N Z J Stat . 2007; 49: 1– 20. Google Scholar CrossRef Search ADS   41. Mas A, Pumo B. Functional linear regression with derivatives. J Nonparametr Stat . 2009; 21( 1): 19– 40. Google Scholar CrossRef Search ADS   42. Frøslie KF, Røislien J, Qvigstad E, Godang K, Bollerslev J, Voldner N, Henriksen T, Veierød MB. Shape information from glucose curves: functional data analysis compared with traditional summary measures. BMC Med Res Methodol . 2013; 13( 1): 6. Google Scholar CrossRef Search ADS PubMed  43. Guzman CB, Zhang XM, Liu R, Regev A, Shankar S, Garhyan P, Pillai SG, Kazda C, Chalasani N, Hardy TA. Treatment with LY2409021, a glucagon receptor antagonist, increases liver fat in patients with type 2 diabetes. Diabetes Obes Metab . 2017; 19( 11): 1521– 1528. Google Scholar CrossRef Search ADS PubMed  44. Lefèbvre PJ, Paquot N, Scheen AJ. Inhibiting or antagonizing glucagon: making progress in diabetes care. Diabetes Obes Metab . 2015; 17( 8): 720– 725. Google Scholar CrossRef Search ADS PubMed  Copyright © 2018 Endocrine Society

Journal

Journal of Clinical Endocrinology and MetabolismOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off