JEH Journal of Endocrinology And Hormones 3070-2313 Open Access Pub United States 10.14302/issn.3070-2313.jeh-17-1756 JEH-17-1756 research-article β-Cell function in type 1 diabetes may not be as low as presumed β-cell function in type 1 diabetes Gonca Tamer 1 * Burcu Dogan 2 Ozge Telci Caklili 3 Division of Endocrinology and Metabolism, Department of Internal Medicine, Medeniyet University, Göztepe Training and Research Hospital, Istanbul. Department of Family Medicine, Goztepe Training and Research Hospital, Istanbul. Department of Internal Medicine, Medeniyet University, Göztepe Training and Research Hospital, Istanbul. Corresponding author Dragan Nikolic 1 University of Belgrade, School of Medicine, dr Subotica 9, 11000 Belgrade, Serbia Gonca Tamer, 1Division of Endocrinology and Metabolism, Department of Internal Medicine, Medeniyet University, Göztepe Training and Research Hospital, Istanbul. Email: hgtamer@yahoo.com

The authors have declared that no competing interests exist.

 13 02 2018 1 1 16 24 13 09 2017 02 02 2018 13 02 2018 © 2018 Gonca Tamer, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. This article is available from http://openaccesspub.org//jeh/article/693 Objective

We aimed to evaluate β-cell function of type 1 diabetic patients (T1DP)s based on fasting and stimulated C-peptide levels.

 Material and methods

Study included 135 T1DPs and 31 healthy subjects. Fasting C-peptide levels were measured in healthy subjects and T1DPs. The Mixed-meal tolerance test (MMTT) was performed in T1DPs. Fasting and stimulated (90 minute post MMTT) C-peptide levels were measured via electrochemiluminescence assay. Two categorizations were made according to fasting (the first categorization ) and at 90th minute MMTT (the second categorization) C-peptide levels. For the first categorization; the groups were classified as follows: patients with undetectable ≤0.1ng/mL(group1); minimal 0.1-0.8ng/mL(group2); and sustained ≥0.8ng/mL (group3) C-peptide levels. For the second categorization, groups were as follows: patients with undetectable ≤0.1 ng/mL(group1); minimal 0.1-0.8 ng/ml (group2); and sustained ≥0.8ng/mL (group3) in which C peptide levels were increased to ≥150% of fasting C-peptide levels at the 90th minute after MMTT.

 Results

For the first category; 41.5%, 40% and 18.5% of T1DPs were in group1, group2 and group3, respectively. For the second category; 34.8%, 20.7% and 44.4% were in group1, group2 and group3, respectively.

In first categorization 58.5% and in second categorization 65.1% of T1DPs had detectable C-peptide levels. 44,4%of the T1DPs had a response to MMTT with C-peptide levels ≥0.8 ng/mL which increased to ≥150% of fasting C-peptide level at the 90th minute after MMTT as it is seen in non-diabetics.

 Conclusion

The present study suggests the presence of functioning β-cells in T1DPs and 44.4 % of T1DPs have a response to MMTT as seen in non-diabetics. (ClinicalTrials.gov number: NCT02199470.)

 β-Cell function type 1 diabetes mixed-meal tolerance test. Introduction

Type 1 diabetes (T1D) is currently defined as autoimmune destruction of β-cells resulting in absolute insulin deficiency 123. The decrease in β-cell function in type 1 diabetic patients (T1DP)s varies individually. Why some T1DPs have functioning β-cells and some do not is probably a result of their unique immunological and genetic characteristics. Some β-cells may be able to regenerate, evade the immunological attack or the inflammation may not be high enough 13456. Type 1 diabetes is a serious disease as it causes severe short and long term complications (short term complications are ketoacidosis, hyperosmolar coma, hypoglycemia and long term complications are macrovascular complications and microvascular complications such as retinopathy, nephropathy and neuropathy) and high mortality rate 1. Increasing endogenous insulin production in T1DPs can improve glycemic control and decrease the complication and mortality rates. Because insulin secretory capacity provide T1DPs for managing themselves intensively in a more effective manner so that to be protected from hypoglycemia and other diabetic complications17. However decrease in mortality and complication rate can be achieved if sufficient β-cell function is present.

β-cells were observed in pancreatic autopsy samples obtained from T1DPs and persistent C-peptide secretion following a meal stimulus, was noted in the majority of T1DPs in recent studies 1234. The present study aimed to determine the level of β-cell function in T1DPs based on the fasting C-peptide levels and C-peptide levels after meal stimuli.

 Material and Method Study Population

We recruited 31 healthy subjects, 135 T1DPs, aged 18-56 years, who have T1D ≤25 years. All patients had been on insulin therapy since diagnosis. Ethics committee of our hospital approved the study protocol, which was performed in accordance with the Declaration of Helsinki (24.01.2013, 30/I). Our study was registered with an approved clinical trial registry. The clinical trial registration number of this study is NCT02199470.

Type 1 diabetes was defined as insulin dependency from the time of diagnosis onwards according to WHO criteria 3. Duration of diabetes ranged from 1 month to 25 years and all patients had been on insulin treatment since the diagnosis. Subjects receiving drugs or/and having diseases which influence β-cell function and insulin sensitivity were excluded from the study. Glucose intolerance (fasting glucose level and that after glucose load) was also used as an exclusion criteria in the control group.

Fasting C-peptide levels of healthy subjects and T1DPs and stimulated C-peptide levels of T1DPs were measured. Stimulated C-peptide level was defined as C-peptide level measured 90 minutes after the mixed- meal.

Two categorizations were made according to the fasting (the first categorization) and stimulated (the second categorization) C-peptide levels. For the first categorization; the groups were classified as follows: patients with undetectable ≤0.1 ng/mL (group 1); with minimal 0.1-0.8 ng/ml (group 2); and with sustained ≥0.8 ng/mL (group 3) C-peptide levels. For the second categorization, groups were as follows: patients with undetectable ≤0.1 ng/mL (group 1); with minimal 0.1-0.8 ng/ml (group 2); and with sustained ≥0.8 ng/mL (group 3) C-peptide levels which increased to ≥150% of fasting C-peptide levels at the 90th minute after MMTT.

Mixed meal tolerance test: The MMTT was administered to each patient in the morning of an overnight fast of ≥ 12 hours (86 patients 12 hours, 35 patients 12.5 hours and 14 patients 13 hours) and not smoking since 10.00 p.m. the previous day and not administered their usual morning insulin. Then, venous glucose, C-peptide, HbA1c and lipid levels were measured and MMTT was performed to all patients. After collecting fasting blood samples, a mixed meal containing 33 g of carbohydrate, 15 g of protein, and 6 g of fat (240 kcal total) 2 was ingested in less than 15 min., and then C-peptide level was measured at 90th minute after mixed- meal.

 Laboratory Analysis:

The fasting glucose level was measured via the hexokinase method and the HbA1c level was measured via high performance liquid chromatography (Tosoh G7 and 2.2, Tokyo, Japan). The lipid profile was determined based on standard enzymatic colorimetric methods (using Roche Diagnostics kits) and the C-peptide level was measured via direct electrochemiluminescence immunoassay using mouse monoclonal anti-C- peptide antibody (Immulite 2000,Siemens, Germany).

Statistical Analysis

Statistical analysis was performed using SPSS version 16.0 ( Inc., Chicago, Illinois, USA). The normality of the distribution of continuous variables was determined based on the one-sample Kolmogorov–Smirnov test, and data are presented as median and interquartile range. Categorical variables are shown as frequency and group percentage. Differences in variables not normally distributed between groups were evaluated using the Kruskal Wallis test. Two-group non-parametric comparisons were made using the Mann Whitney-U test. Spearmen’s rho correlation analysis was used to identify possible associations between fasting C-peptide and stimulated C-peptide levels and the other study variables. The level of statistical significance was set at p < 0.05.

 Results

Median age of the patients (53.33% of the T1DPs were male and 46.6% were female) was 29 years (interquartile range: 22-37 years), median age at diagnosis was 23 years (interquartile range: 22-37 years), and median duration of diabetes was 52 months (interquartile range: 19-132 months).

For the first category; 56(41.5%), 54(40%), and 25(18.5% ) patients of T1DPs were in group 1, group 2 and group 3, respectively. For the second category; 47(34.8%) patients of T1DPs were in group 1, 28 (20.7% ) of them in group 2 and 60 (44.4 %) of them in group 3.

The characterizations of all the groups and healthy subjects were in Table 1, Table 2 and Table 3. Among participants 58.5% of the patients had a detectable fasting C-peptide level and 65.1% had a detectable stimulated C-peptide level. Forty four point four percent of the T1DPs (group 3) had a response to mixed- meal stimulus with a C-peptide levels ≥0.8 ng/mL which increased to ≥150% of the fasting C-peptide level similar to non-diabetics. Moreover, there were no difference between fasting C-peptide levels of group 3 and those of healthy subjects (p=0,5 for the comparison of two groups according to the first categorization and p=0.1 for the comparison of these two groups according to the second categorization). As shown in Table 2 and Table 3, according to the first and second categorizations, group 3 were the oldest at time of diagnosis (p=0.03, p=0.01 respectively) and had the shortest duration of diabetes (p<0.001, p<0.001 respectively). There were no differences among the three groups of T1DPs in gender, age, HbA1C, fasting glucose, total cholesterol, LDL-C, or triglyceride levels(p>0.05, p>0.05, p>0.05, p>0.05, p>0.05, p>0.05, p>0.05 respectively).

Demographic and clinical characteristics of patients and controls
 Variables Patients with type 1 diabetes(n=135) Control Group (n=31) Pvalue
Age (year) Median Interquartile range  2921-36  3027-38 0.13
Gender (Male/Female) 72/63 13/18 0.56
Fasting blood glucose (mg/dL) Median Interquartile range  175115-263  8684-92  <0.001
HbA1c (%) Median Interquartile range  8.47.1-10.1  5.25.1-5.4  <0.001
Fasting C peptide (nmol/L) Median Interquartile range  0.230.01-1.74  1.351.05-1.66  <0.001
Total Cholesterol (mg/dL) Median Interquartile range  168150-194  170157-189  0,97
Low Density Lipoprotein Cholesterol (mg/dL) Median Interquartile range   9680-122   9873-115   0,32
High Density Lipoprotein Cholesterol (mg/dL) Median Interquartile range   5344-65   5746-73   0,09
Triglyceride (mg/dL) Median Interquartile range  7356-103  7362-86  0,8
Charesteristics of the groups according to the first categorization
  Group 1(n=56) Group 2(n=54) Group 3(n=25) P
Age (year) Median Interquartile range  2823-37  2923-36  2921-39  0.82
Gender (n/n) Female/Male  31/25  24/30  8/17  0.14
Duration of diabetes, (month) Median Interquartile range  11459-168  3712-92  182-28  <0.001
Fasting Blood Glucose (mg/dL) Median Interquartile range  180115-263  171136-251  144124-237  0.68
HbA1c mmol/mol (DCCT units%) Median Interquartile range  63.9 (8.0)55.2-82.5 (7.2-9.7)  65(8.15)50.8-85.5 (6.8-10.0)  69.4 (8.5)46.4-86.9 (6.4-10.1)  0.97
Fasting C peptide (nmol/L) Median Interquartile range  0.100.10-0.10  0.280.21-0.52  1.260.93-1.74  <0.001 
Total cholesterol (mg/dL) Median Interquartile range  171155-194  175151-190  162134-190  0.31
LDL-C (mg/dL) Median Interquartile range  9581-122  9983-126  8969-118  0.29
HDL-C (mg/dL) Median Interquartile range  5746-68  5344-65  4638-54  0.008
Triglyceride (mg/dL) Median Interquartile range  7559-105  7056-90  7864-148  0.37 
Age at diagnosis (year) Median Interquartile range  2115-28  2417-31  2619-34  0.03
Charesteristics of the groups according to the second categorization
  Group 1(n=47) Group 2(n=28) Group 3(n=60) P
Age (year) Median Interquartile range  3025-40  2621-31  2822-38  0.27
Gender (n/n) Female/Male  28/21  13/15  21/33  0.17
Duration of diabetes, (month) Median Interquartile range  13875-170  3814-90  244-56  <0.001
Fasting Blood Glucose (mg/dL) Median Interquartile range  200115-282  193144-260  143118-186  0.04
HbA1c mmol/mol (DCCT units%) Median Interquartile range  63.9 (8.0)55.2-79.2 (7.2-9.4)  68.3(8.4)57.4-59.6 (7.4-7.6)  59.6(7.6)46.4-80.3 (6.4-9.5)  0.21
90-minute C peptide (nmol/L) Median Interquartile range  0.100.10-0.10  0.280.24-0.65  2.101.40-3.77  <0.001
Total cholesterol (mg/dL) Median Interquartile range  169154-195  170152-199  164138-190  0.32
LDL-C (mg/dL) Median Interquartile range  9478-123  9781-127  9679-121  0.92
HDL-C (mg/dL) Median Interquartile range  5545-68  5748-65  4842-58  0.06
Triglyceride (mg/dL) Median Interquartile range  7667-112  6048-85  7358-94  0.31
Age at diagnosis (year) Median Interquartile range  1913-26  2416-29  2619-35  0.01

The duration of diabetes was negatively correlated with fasting and stimulated C-peptide levels (r= -0.57;p<0.001, r= -58; p<0.001 respectively). Group 1 with the longest diabetes duration (more than 92 months) did not have detectable (41.5 %) and stimulated (34.8 %) C-peptide levels. Age at diagnosis was positively correlated with fasting C-peptide levels and stimulated C-peptide levels (r=0.32;p<0.001, r=0.3;p=0.001 respectively).

Fasting and stimulated C-peptide levels of the patients whose diagnosis of diabetes before puberty were significantly lower than those of the patients whose diagnosis of diabetes after puberty (p<0.001, p<0.001).

Discussion

The majority of patients with type 1 diabetes was observed to produce insulin endogenously based on fasting and stimulated C-peptide levels. Additionally, the prevalence of a detectable fasting C-peptide level was 58.5% and the prevalence of a detectable stimulated C-peptide level was 65.1%. Furthermore, almost 44.44 % of T1DPs had a response to mixed-meal stimulus with C-peptide levels ≥0.8 ng/mL which increased to ≥150% of fasting C-peptide levels as in non-diabetics. In the present study, no difference was found between these patients including 44.44 % of T1DPs and healthy subjects in fasting C-peptide levels.

Based on the present findings, we suggest that the fasting and stimulated C-peptide levels in patients with type 1diabetes may be higher than presumed before.

Consistent with Pipeleers’s 3 and Keenan’s findings 4, in the present study, fasting C-peptide levels and stimulated C-peptide levels were found to be positively correlated with age at diagnosis whereas Lohr and Kloppel 5 did not observe a correlation between age of onset of diabetes and residual β-cells. Additionally, in the present study, fasting and stimulated C-peptide levels decreased as the duration of diabetes increased. This finding is also consistent with Oram’s findings 6.

Interestingly, based on a murine model Thorel et al showed that in β-cell-depleted mice, α-cells could differentiate into β-cells in mice after prolonged duration of diabetes 7. Similarly, Chera et al observed that pancreas reconstituted new insulin–producing cells in the absence of autoimmunity in mice in which β-cells were completely ablated. They also reported that glucagon-producing α-cells could begin to produce insulin via a process of reprogramming (transdifferentiation) without proliferation and posited that these phenomena might be translatable to humans, because efficient β-cell regeneration had been determined in children with type 1 diabetes or after pancreatectomy 89 and glucagon/insulin bihormonal human cells were observed following epigenetic manipulation ex vivo 10, and in patients with diabetes 1112. Moreover, Zhou et al reported that aciner cells were capable of conversion into β-cells invivo when administered an adenovirus cocktail of some transcription factors 13. β-cells in patients with long duration of type 1 diabetes have been reported in histological studies since 1965 1415. These findings can explain the increase in the C-peptide levels after mixed-meal stimuli in T1DPs observed in the present study, and the Joslin Medalists4 and Oram’s 6 studies. The increase in the C-peptide levels after mixed-meal stimuli in the present study strongly indicates the presence of functioning β-cells, however, in the present study, there was no histological examination showing that aciner cells and α-cells were capable of conversion into β-cells. This was a limitation of the present study.

The source of the functioning β-cells observed in the present study and in two otherearlier studies mentioned here in, is not clear, however there might be multiple sources. β-cell proliferation and redifferentiation and dedifferentiation of α-cells, δ cells and aciner cells into β-cells may result in functioning β-cells in T1DPs with long disease duration.

An important finding of the present study is that fasting and at 90th minute post MMTT C-peptide levels in the patients diagnosed as diabetes before puberty were significantly lower than in those diagnosed after puberty, which might be explained with Chera’s mouse study, in which α-cells reprogramed to produce insulin from puberty to adulthood, and somatostatin- producing δ cells were able to convert to insulin-producing β-cells during puberty. Chera et al. suggested that the source of regenerated β-cells after puberty and even a long time after β-cells loss, was α-cells and that prior to puberty the source of β-cells was somatostatin- producing δ cells and reconstituted β-cells 916. The present findings support their suggestion that dedifferentiation of α-cells and δ-cells into β-cells could be possible in humans.

To the best of our knowledge the present study is the first to evaluate the effect of puberty on fasting and stimulated C-peptide levels.

Moreover, the present study has been the first to have evaluated the prevalence of T1DPs having a response to mixed-meal stimulus like non-diabetics, up to now, too. The Physiologic C-peptide response to MMTT in non-diabetics is accepted as a C-peptide level ≥0.8 ng/mL which increases to ≥150% of the fasting C-peptide level 19. The number of T1DPs with a non-diabetic-like response to mixed-meal stimulus were evaluated in only the present study, but was not evaluated in the DCCT (Diabetes Control and Complications Trial) 17, McGee’18, Greenbaum’ 19, Oram’ study 6 or Joslin Medalist study4. It was reported that in Joslin Medalist study 4, 13 (41.93%) of 31 T1DPs who returned for MMTT and in Greenbaum’s study19, almost all the T1DPs responded with a doubling of the C-Peptide level as compared to the fasting level. However it was not evaluated that those C-peptide levels after MMTT were in normal or below the normal range in both of the aforementioned studies 419. We found that 60 (44.4%) of 135 T1DPs responded to meal stimuli with C-peptide levels ≥0.8 ng/mL which increased to ≥150% of fasting C-peptide levels as non-diabetics.

Additionally, the present study administered the MMTT has had the third largest group number of T1DPs after DCCT (included 855 patients) and Greenbaum’s (included 143 patients) studies up to now 46171819.

There are several differences between the present study and the earlier studies cited. The present study is different from Joslin Medalist study and Greenbaum’s study. Since serum C-peptide level was measured via radioimmunoassay in Medalist study 4 and measured via the assay which was unable to measure a value below the lower limit of quantification in Greenbaum’s 19, whereas in the present study it was measured via a direct electrochemiluminescence immunoassay which is more sensitive than other assays used in the two aforementioned studies 4619. Oram’s study 6 included type 1diabetics with the disease duration of > 5 years whereas the present study included type 1 diabetics with the disease duration of ≤ 25 years. Moreover, in the DCCT study 17 mean duration of diabetes was 2.3 ± 0.3 years and in Greenbaum’s , it was ≤ 4 years less than in the present study. Detectable C-peptide levels after mixed-meal stimuli were noted in 65.1% of T1DPs in the present study, versus 73% of those in Oram’s study 6, 67.4% in the Joslin Medalist Study 4, 35.43% in the DCCT 17, 87-95% in Greenbaum’s study 19, and 17% in McGee’s study 18 which included the patients screened via MMTT in the DCCT study 30 years later. It is interesting that the prevalence of detectable C-peptide levels after mixed- meal stimuli in the Joslin Medalist, Oram’s and the present studies were higher than that in the DCCT study as type 1 diabetics in the DCCT study had the shortest duration of disease. The difference might have been result from using of a more sensitive immunoassay in the present and Oram’s studies, but it is not clear why the prevalence of a detectable C-peptide level after mixed- meal stimulus in the Joslin Medalist study, was higher than that in the DCCT study .

The present study has some limitations. It is possible that some patients with non-type 1 diabetes were inadvertently included in the study although a tight type 1diabetes definition was used. Family history for MODY was recorded in each of the patients in the present study. However the patients were not genotyped for risk polymorphisms in MODY genes.

In conclusion, the present study shows that many of the type 1 diabetics had detectable fasting and stimulated C-peptide levels. Moreover, 44.4 % of T1DPs had a non-diabetic-like C-peptide level in response to mixed- meal stimulus. As, the presence of functioning β cells in type 1 diabetics can facilitate their long-term survival and reduce the incidence of macro and microvascular complications and hypoglycemia, future studies are needed to discover the source of insulin-producing cells in some T1DPs as observed in the present study. If the source can be identified, subsequent research can focus on how to enhance this source of insulin-producing cells.

 Rydén A Ludvigsson J Fredrikson M Faresjö M General immune dampening is associated with disturbed metabolism at diagnosis of type 1diabetes.Doi: 10.1038/pr.2013.167.PediatrRes.2014Jan,(Epub ahead of print),75(1-1): 45 50 10.1038/pr.2013.167.PediatrRes.2014Jan,(Epub Mortensen H B Swift P G Holl R W Multinational sudy in children and adolescents with newly diagnosed type 1diabetes: association of age , ketoacidosis, HLA status, and autoantibodies on residual beta cell function and glycemic control 12 months after diagnosis. Pediatr Diabetes 2010 11 218 226 Pipeleers D In’t Veld P Pipeleers-Marichal M Gorus F The β cell population in type 1 diabetes.discussion.24-31,122-129,202-203.Novartis Found Symp2008;292: 24 Keenan H A Sun J K LevineJ Residual Insulin Production and Pancreatic β cell Turnover After 50 years of Diabetes: 2010 Joslin Medalist Study. Diabetes 59 11 2846 2853 Lohr M Kloppel G Residual insulin positivity and pancreatic atrophy in relation to duration of type 1 (insulin-dependent) diabetes mellitus and microangiopathy 1987 Diabetologia 30 757 762 Oram R A Besser RE JonesAG The majority of patients with long-duration type 1 diabetes are insulin microsecretors and have functioning beta cells 2014 Diabetologia 57 1 187 191 Thorel F Nepote V Avril I Conversion of adult pancreatic α-cells to β-cells after extreme β-cells loss 2010 Nature 464 1149 1154 Desgraz R Bonal C Herrera P L β-cell regeneration: the pancreatic instrintic faculty 2011 Trends Endocrinol Metab 22 34 43 Chera S Baronnier D Ghila L Diabetes recovery by age dependent conversion of pancreatic δ-cells into insulin producers.doi: 2014 10 1038 10.1038/nature13633.Nature Bramswing N C Everett L J Schug J Epigenomic plasticity enables human pancreatic α to β-cell to cell reprogramming 2013 J Clin Invest 123 3 1275 1284 Butler A E Campbell-Thompson M Gurlo T Marked expansion of exocrine and endocrine pancreas with incretin therapy in humans with increased exocrine pancreas displasia and potential for glucagon-producing neuroendocrine tumors 2013 Diabetes 62 2595 2604 Yoneda S Uno S Iwahashi H Predominance of β-cells neogenesis rather then replication in humans with an impaired glucose tolerance and newly diagnosed diabetes 2013 J Clin Endocrinol. Metab 98 5 2053 2061 Zhou Q Brown J Kanarek A DA Rajagopal J &Melton In vivo reprogramming of adult pancreatic exocrine cells to beta-cells 2008 Nature 455 627 632 Gepts W Pathologic anatomy of the pancreas in juvenil diabetes mellitus Diabetes.1965 14 619 633 Foulis A K Liddle C N Ricmond JA FarquharsonMA Weir R S The histopathology of the pancreas in type 1(insulin-dependent) diabetes mellitus: a 25 year rewiev of deaths in patients under 20 years of age in the United Kingdom 1986 Diabetologia1986; 29 267 274 Avolio F Pfeifer A Courtney M From pancreas morphogenesis is to β-cell regeneration 2013 Curr Top Dev Biol 106 217 218 10.10.1016/B978-0-12-416021-7.00006-7 The Diabetes Control and Complications Trial Research Group. Effect of intensive therapy on residual beta-cell function in patients in the diabetes contol and complications trial Ann Intern Med.1998; 128 7 517 523 McGee P Steffes M Nowicki M DCCT/EDIC Research Group. Insulin secretion measured by stimulated c-peptide 2014 in long-established Type 1 diabetes in the Diabetes control and Complications Trial (DCCT/ Epidemiology of Diabetes interventions and Complications (EDIC) cohort: a pilot study. Diabet Med 31 10 1264 1268 10.1111/dme.12504 Greenbaum C J Mandrup-Poulsen T McGee P F Type 1 Diabetes Trial Net Research Group; European C-Peptide Trial Study Group Mixed-meal tolerance test versus glucagon stimulation test fort he assessment of β-cell function in therapeutic trials in type 1diabetes 2008 Diabetes Care 31 1996 1971