Assessment of beta-cell function in young patients with type 2 diabetes: arginine-stimulated insulin secretion may reflect beta-cell reserve




Simple methods for the evaluation of dynamic β-cell function in epidemiological and clinical studies of patients with type 2 diabetes (T2D) are needed. The aim of this study was to evaluate the dynamic beta-cell function in young patients with T2D with different disease durations and treatments.


Overall, 54 subjects with T2D from the Diabetes Incidence Study in Sweden (DISS) and 23 healthy control participants were included in this cross-sectional study. Beta-cell function was assessed by intravenous (i.v.) administration of arginine followed by i.v. glucose. The acute insulin and C-peptide responses to arginine (AIRarg and Ac-pepRarg, respectively) and to glucose (AIRglu and Ac-pepRglu, respectively) were estimated. Homeostasis model assessment of β-cell function (HOMA-β) and C-peptide assessments were also used for comparisons between patients with T2D and control participants.


AIRarg and Ac-pepRarg, but not AIRglu and Ac-pepRglu, could differentiate between patients with different disease durations. AIRglu values were 89% (< 0.001) lower and AIRarg values were 29% (< 0.01) lower in patients with T2D compared with control participants. HOMA-β and fasting plasma C-peptide levels did not differ between the T2D and control groups.


In young patients with T2D, the insulin secretory response to i.v. glucose is markedly attenuated, whereas i.v. arginine-stimulated insulin release is better preserved and can distinguish between patients with different disease duration and antidiabetic therapies. This suggests that the i.v. arginine stimulation test may provide an estimate of functional beta-cell reserve.


Type 2 diabetes mellitus (T2D) is a progressive disease, which becomes manifest when endogenous insulin secretion is no longer sufficient to compensate for insulin resistance. β-Cell failure then gradually progresses in a majority of patients [1] who eventually become insulin dependent. Even though insulin secretion generally decreases over time in patients with T2D, the pattern and speed of decline differ considerably amongst individuals [2]. Up until now, no pharmacological agent that prevents or restores beta-cell dysfunction in humans has been found. Peroxisome proliferator-activated receptor (PPAR) agonists, glucagon-like peptide-1 (GLP-1) analogues and DPP-IV (dipeptidyl peptidase IV) inhibitors may have such effects [3-6], but this remains to be further demonstrated.

There is a need to establish methods that are appropriate for an outpatient setting to evaluate the long-term course of functional β-cell mass in clinical and epidemiological studies. Homeostasis model assessment of β-cell function (HOMA-β) is often used to estimate β-cell function, but the method is limited because it is based on fasting glucose and insulin levels, only measures β-cell function under fasting nondynamic conditions and has not been validated in patients with T2D treated with insulin [7].

The standardized glucose-potentiated arginine-induced insulin secretion method has been used for more than 25 years to measure β-cell function in T2D and is often considered to be the gold standard measure [8]. However, a simpler, less labour-intensive method is needed for use in large studies and clinical trials. This standardized method has been shown to correlate with a simpler test using an intravenous (i.v.) bolus of glucose or arginine in normoglycaemic healthy participants [9] and in patients who have undergone pancreas transplantation [10]; however, to our knowledge, this methodology has not been tested in patients with T2D.

Therefore, the aim of this study was to evaluate dynamic β-cell function in young patients with T2D with different disease durations and treatments and in healthy control participants using i.v. arginine and glucose stimulation. β-Cell function was also compared between patients and healthy participants using the HOMA-β index, proinsulin/insulin ratio and measurement of fasting plasma C-peptide levels.


Study design

Overall, 54 subjects with T2D (diabetes duration 2–10 years) and 23 healthy control participants were enrolled in this study D0280M00003. Patients were recruited from the national Diabetes Incidence Study in Sweden (DISS) register, which covers incident cases of diabetes in individuals between 15 and 34 years of age and collects data in collaboration with departments of internal medicine, endocrinology and paediatrics in Sweden. The classification into type 1, type 2 and unclassified diabetes in the DISS register is based on the treating physicians' clinical diagnosis, according to criteria of the World Health Organization [11]. In this study, patients with T2D were recruited at five university hospitals: Gothenburg, Umeå, Linköping, Lund and Huddinge, and the control group of 23 healthy participants were recruited at the Gothenburg hospital. All patients were recruited from a cohort of individuals in the DISS register diagnosed with diabetes between 1998 and 2006 (= 397). All participants (male or female) were invited to take part in the study and were eligible if they met the following criteria: (i) aged 15–34 years at the time of diabetes diagnosis and inclusion in the DISS register; (ii) classified as having T2D by the reporting physician; (iii) plasma samples negative for islet-cell antibodies at or within 3–4 months of diagnosis; (iv) diabetes duration of 2–10 years; and (v) HbA1c <10% (Swedish Mono-S standard; normal reference 3.0–5.3%). In total, 66 subjects with T2D were enrolled, and 54 subjects were included in all analyses and safety. The main reason for exclusion of enrolled patients was unacceptably high plasma glucose levels before the dynamic i.v. glucose and arginine tests.

Male and female healthy (as judged by the investigator) control participants between 25 and 50 years of age were recruited via advertisement and included if they had a body mass index (BMI) of 19–40 kg m−2 and did not have diabetes, impaired fasting glucose or impaired glucose tolerance.

All participants gave written informed consent prior to enrolment, and the study protocols were approved by the ethics committee in Gothenburg.

The subjects with T2D were treated with diet and exercise alone n = 10, oral hypoglycaemic agents (OHAs) (metformin n = 33, thiazolidinediones n = 3, sulphonylurea n = 2, alpha-glucosidase inhibitors n = 1) and insulin treatment n = 22 (10 treated with insulin only and 12 with insulin in combination with OHA). Fig. 1 shows the distributions of treatment in patients included in the DISS register <5 years, 5–<8 years and 8–10 years, respectively, before the start of the study.

Figure 1.

Treatments in patients included in the DISS register <5, 5 to <8 and 8–10 years before the start of the study; OHA, oral hypoglycaemic agents.

At the enrolment visit (2–4 weeks before the investigational day), all participants underwent an examination including recording of demographic characteristics (sex and age), medical history, family history of diabetes, current medication, weight and height (to calculate BMI), waist and hip circumference, physical examination, blood pressure, pulse and a laboratory screen. In addition, in the control group, an oral glucose tolerance test (OGTT) was performed to exclude diabetes or impaired fasting glucose/glucose intolerance. Patients with T2D and healthy control participants were matched for age, sex and BMI.

On the investigational day, subjects who fulfilled all the eligibility criteria for inclusion in the study underwent testing in the morning after fasting overnight. No medication, including insulin, was taken on the morning of the investigational day. A venous cannula was inserted into one arm for blood sampling. Arginine and glucose (i.v.) for the β-cell function test were administered in the other arm. Blood samples were collected before the start of the β-cell function test for analysis of fasting plasma biomarkers including glucose, insulin, proinsulin, C-peptide and HbA1c.

In addition, we obtained nonfasting plasma samples that had been collected close to diagnosis of diabetes and inclusion in the DISS register and stored at −80 °C. These samples were analysed for insulin, proinsulin and C-peptide levels at AstraZeneca R&D (Mölndal, Sweden).

Dynamic testing of beta-cell function

The i.v. β-cell function test was performed on the investigational day according to the method of Robertson [10]. Subjects with T2D were asked not to take their usual pharmacological treatment (including insulin) on the morning of the study day. Before the test, glucose levels were measured to ensure that they were between 4 and 12 mmol L−1. A predose blood sample was collected immediately before the start of the test. Arginine-induced insulin secretion was assessed by injecting an i.v. bolus of 5 g arginine at time 0, and samples were drawn 2, 3, 4, 5, 7, 10, 25 and 30 min following the injection. Immediately thereafter, glucose-induced insulin secretion was assessed by injecting an i.v. bolus of 0.3 g kg−1 glucose, and samples were collected at 3, 4, 5, 7, 10, 15, 20, 25, 30, 60, 115 and 120 min.

The acute insulin response to arginine (AIRarg) was calculated as the mean of the three highest plasma insulin levels obtained within 5 min after the arginine bolus minus the prestimulus plasma insulin level. The acute insulin response to glucose (AIRglu) was calculated as the mean of the plasma insulin levels obtained 3, 4 and 5 min after glucose injection minus the prestimulus level. In addition, the acute C-peptide response to glucose (Ac-pepRglu) and the acute C-peptide response to arginine (Ac-pepRarg) were calculated as described for insulin.

HOMA was calculated using the formula HOMA-IR = (FPI × FPG)/22.5 and HOMA-β = (20 × FPI)/(FPG- 3.5) for insulin resistance and β-cell function, respectively, where FPI is fasting plasma insulin concentration(mU L−1) and FPG is fasting plasma glucose (mmol L−1) .

Analytical methods

Glucose concentrations in plasma were analysed using a Cobas Mira Plus analyser (Hoffman-La Roche, Basel, Switzerland) and an enzymatic (hexokinase) colorimetric method (ABX Pentra, Montpellier, France). HbA1c was assessed by high-performance liquid chromatography using Mono-S ion-exchange chromatography (Pharmacia-Amersham Bioscience, Uppsala, Sweden). The reference interval was 3.0% to 5.3%. Plasma concentrations of insulin and C-peptide were measured using commercial radioimmunoassays (Millipore, St. Charles, MO, USA). Proinsulin levels were measured using a commercial enzyme-linked immunosorbent assay (Mercodia, Uppsala Sweden). The total (intra- and interassay) coefficient of variation was <3% for glucose, <15% for insulin and <8% for C-peptide. Cross-reactivity for the C-peptide and insulin assays was <0.2% and <4% for human proinsulin (HPI), respectively. Cross-reactivity of the HPI assay was 84% for HPI Des (64–65), 90% for HPI Split (65–66), 95% for HPI Des (31–32) and 90% for HPI Split (32–33), respectively, and <0.03% and <0.006% for insulin and for C-peptide, respectively.

Statistical analysis

Data are presented descriptively as means ± SD. Comparisons between groups at baseline were made using a two-sample t-test, and comparisons between different lengths of diabetes duration and different treatment groups were made using the Kruskal–Wallis test and subsequent post hoc analysis with the Dunn's test. An exact Wilcoxon rank-sum test was used for comparisons between healthy control participants and patients with T2D. Relationships were explored by Pearson's correlation including confidence intervals based on Fisher's z-transformation.


Clinical characteristics

Subjects with T2D were slightly younger than those in the control group (34.6 ± 5.4 vs. 39.6 ± 5.8 years); the two groups were well matched for sex (male/female: 33/21 vs. 12/11), BMI (33.2 ± 5.7 vs. 32.6 ± 4.5 kg m−2) and waist circumference (110 ± 14.0 vs. 111 ± 12.9 cm). HbA1c was 6.2 ± 1.3% (54 mmol mol−1) and 4.1 ± 0.3% (32 mmol mol−1) in the T2D and control groups, respectively (< 0.001). Clinical characteristics in subgroups based on diabetes duration and antidiabetic treatment are shown in Tables S1 and S2.

Insulin and C-peptide responses after i.v. glucose and arginine

Fasting pretest glucose concentrations were 7.3 ± 1.8 and 5.5 ± 0.9 mmol L−1 in the T2D and control groups, respectively (< 0.001). Mean fasting plasma insulin was almost twofold higher in the T2D patient cohort than in the control group (217 ± 161 vs. 119 ± 77 pmol L−1, < 0.01). The AIRarg and Ac-pepRarg values were 29% lower (< 0.01) and AIRglu and Ac-pepRglu values were 89% lower (< 0.001) in patients with T2D compared with control participants (Table 1 and Fig. 2a).

Table 1. Acute insulin (AIR) and C-peptide (Ac-pepR) responses after i.v. glucose and arginine administration and HOMA-β, HOMA-IR, fasting C-peptide and proinsulin/insulin ratio in study groups
 Control (= 23)Type 2 diabetes (= 54)P-value
  1. Data are mean ± SD. AIR and Ac-pepR values are the mean of the three maximum values minus the prestimulus value.

AIRglu, pmol L−1558 ± 36060 ± 156<0.001
Ac-pepRglu, nmol L−11.40 ± 0.790.23 ± 0.50<0.001
AIRarg, pmol L−1426 ± 258306 ± 264<0.01
Ac-pepRarg, nmol L−11.10 ± 0.460.93 ± 0.60<0.01
HOMA-β179 ± 86163 ± 114Ns
HOMA-IR4.4 ± 3.211.0 ± 8.2<0.001
C-peptide, nmol L−11.10 ± 0.401.10 ± 0.56Ns
Proinsulin/insulin1.06 ± 0.651.17 ± 0.85Ns
Figure 2.

Acute insulin (a) and glucose (b) responses after i.v. glucose and arginine in healthy participants and patients with type 2 diabetes (T2D).

The sensitivity of the β-cells to respond to changes in glucose (the glucose sensitivity) was also estimated by dividing the insulin responses to glucose (AIRglu) by the acute glucose response after i.v. glucose challenge. AIRglu adjusted for glucose response was 90% lower in patients with T2D compared with control participants. This difference between groups was similar to the results without adjustment for glucose levels.

The plasma insulin and glucose levels had returned to baseline values in both groups after the arginine challenge and before administration of glucose (Fig. 2a, b). In addition, the C-peptide concentration had returned to basal levels between the arginine and glucose challenges (data not shown). HOMA-β, proinsulin/insulin ratio and fasting plasma C-peptide concentration did not differ between the two groups. HOMA-IR was significantly higher in patients with T2D than in control participants (Table 1).

Correlations between fasting plasma glucose, proinsulin/insulin ratio and C-peptide levels and insulin and C-peptide responses to i.v. glucose and arginine

AIRglu and Ac-pepRglu, but not AIRarg and Ac-pepRarg, responses were negatively correlated with fasting p-glucose (r = −0.52 and −0.48, < 0.05 and r = 0.10 and 0.15, ns, respectively) in the T2D group. The proinsulin/insulin ratio showed no significant correlation with AIRglu and Ac-pepRglu (r = −0.24 and −0.22) or with AIRarg and Ac-pepRarg (r = 0.26 and 0.32). Fasting C-peptide was not significantly correlated with AIRglu and Ac-pepRglu (r = −0.11 and −0.14), but was positively correlated with AIRarg and Ac-pepRarg responses (r = 0.78 and −0.72, < 0.05) in the T2D group. C-peptide, insulin and proinsulin levels sampled at the time of diabetes diagnosis were significantly and positively correlated with AIRarg (r = 0.67, < 0.05; r = 0.59, < 0.05; and r = 0.52, < 0.05, respectively), but not with AIRglu (r = −0.12, −0.13 and −0.12, respectively). There was no significant correlation between proinsulin/insulin ratio at diagnosis and either AIRarg (r = 0.03) or AIRglu (r = −0.25). The concentration of C-peptide at diagnosis was significantly correlated with the fasting C-peptide level on the investigational day (r = 0.67, < 0.0001).

Insulin and C-peptide responses after i.v. glucose and arginine, HOMA-β and fasting C-peptide in subjects with different duration of T2D

The insulin and C-peptide responses after arginine administration were gradually reduced in the groups with longer duration of T2D, whereas the glucose-induced insulin and C-peptide responses were similar irrespective of disease duration (Table 2). These findings of the impact of diabetes duration on arginine-stimulated insulin secretion, as presented in Table 2, are further supported by linear correlation analysis (< 0.05). AIRarg and Ac-pepRarg were significantly negatively correlated with diabetes duration (r = −0.41 and −0.47, < 0.05), whereas AIRglu and Ac-pepRglu showed no significant correlation (r = −0.13 and −0.14). Pretest plasma glucose concentration was 7.8 ± 1.8, 7.6 ± 1.7 and 7.3 ± 1.8 mmol L−1 (ns) in patients with a diabetes duration of <5 years, 5 to <8 years and 8–10 years, respectively. HOMA-β did not differ significantly between the three patient subgroups (Table 2). Patients with a disease duration of <5 years had significantly higher pretest C-peptide levels than those in the two other groups with longer duration of diabetes (< 0.05) (Table 2).

Table 2. Acute insulin (AIR) and C-peptide (Ac-pepR) responses after i.v. glucose and arginine, and HOMA-β, HOMA-IR and fasting C-peptide, in subgroups based on diabetes duration
 Diabetes duration (years)P-value
<5 (= 11)5 to <8 (= 20)8–10 (= 23)
  1. Data are mean ± SD.

AIRglu, pmol L−151.6 ± 228.062.4 ± 156.027.6 ± 50.4Ns
Ac-pepRglu, nmol L−10.13 ± 0.660.23 ± 0.500.10 ± 0.23Ns
AIRarg, pmol L−1498 ± 312342 ± 306186 ± 120


8–10 years vs. <5 years

Ac-pepRarg, nmol L−11.20 ± 0.630.93 ± 0.600.53 ± 0.23


8–10 years vs. 5 to <8 and <5 years

HOMA-β162 ± 122224 ± 305200 ± 198Ns
HOMA-IR10.5 ± 8.211.1 ± 8.810.2 ± 7.9Ns
C-peptide, nmol L−11.50 ± 0.561.10 ± 0.600.86 ± 0.36


<5 years vs. 5 to <8 and 8–10 years

Insulin and C-peptide response after i.v. glucose and arginine, HOMA-β and fasting C-peptide in subjects with T2D treated with diet/exercise alone, OHAs and/or insulin

AIRglu and Ac-pepRglu were higher amongst patients treated with diet/exercise alone compared with those treated with OHAs, whereas the AIRarg and Ac-pepRarg responses were approximately 60–70% lower in patients treated with insulin compared with the other treatment groups (Table 3). The results were essentially the same in a subanalysis in which patients receiving OHAs other than metformin were excluded (data not shown). Pretest plasma glucose concentration was 6.2 ± 1.2, 7.8 ± 1.7, 7.8 ± 1.9 and 7.8 ± 1.7 mmol L−1 in the groups treated with diet/exercise alone, OHAs and OHAs in combination with insulin and insulin alone, respectively (significantly lower in the diet/exercise-treated group versus all other treatment groups, < 0.05). Subjects treated with insulin alone had significantly lower C-peptide levels than those treated with OHAs (< 0.05) or diet/exercise alone (< 0.05) (Table 3).

Table 3. Acute insulin (AIR) and C-peptide (Ac-pepR) responses after i.v. glucose and arginine, and HOMA-β, HOMA-IR and fasting C-peptide, in subgroups based on treatment
 Diabetes treatmentP-value
Diet/exercise = 10OHA = 22OHA + insulin = 12Insulin alone = 10
  1. Data are mean ± SD.

AIRglu, pmol L−1159.0 ± 252.025.8 ± 108.013.8 ± 90.012.0 ± 22.8


diet/exercise vs. tablets

Ac-pepRglu, nmol L−10.46 ± 0.790.07 ± 0.330.15 ± 0.300.05 ± 0.12


diet/exercise vs. tablets

AIRarg, pmol L−1402 ± 372354 ± 288300 ± 162108 ± 39.6


insulin vs. other treatments

Ac-pepRarg, nmol L−10.96 ± 0.660.93 ± 0.600.83 ± 0.400.40 ± 0.17


insulin vs. other treatments

HOMA-β173 ± 126121 ± 78391 ± 402177 ± 138


tablets + insulin vs. tablets

HOMA-IR6.2 ± 5.48.4 ± 6.219.1 ± 9.99.6 ± 5.0


tablets + insulin vs. diet/exercise and tablets

C-peptide, nmol L−11.20 ± 0.731.26 ± 0.461.00 ± 0.530.60 ± 0.17


insulin vs. diet/exercise and tablets


The results of the present study indicate that the i.v. arginine test can differentiate between subjects with different duration and severity (as reflected by different antidiabetic treatments) of disease. In addition, the present data confirm that the arginine-stimulated insulin secretion response is less influenced by the pretest plasma glucose level than the insulin secretory response after administration of glucose. Moreover, diabetes duration was negatively correlated with AIRarg and Ac-pepRarg, but not with AIRglu and Ac-pepRglu.

The insulin and C-peptide responses after glucose administration were decreased by approximately 90% in the T2D group, whereas the responses to arginine were decreased by approximately 30% compared with controls. It should be noted, however, that without controlling for baseline plasma glucose levels or adjusting for apparent differences in insulin sensitivity (i.e. HOMA-IR) between the two study groups, the presently estimated relative impairment of β-cell secretion in patients with T2D may have been underestimated. The difference in insulin secretory responses to i.v. arginine and glucose stimulation in patients with T2D, on the other hand, cannot be attributed to variations in glucose levels, as plasma glucose was comparable before the two challenge tests. It is well known that the glucose-dependent first-phase insulin response is impaired in T2D and in states of increased plasma glucose levels [10, 12, 13] due to glucose toxicity. Indeed, in the present study, fasting plasma glucose was negatively correlated with the acute insulin and C-peptide response after a glucose load (even though fasting glucose level was required to be in the range 4–12 mmol L−1), whereas no correlation was found between fasting plasma glucose and the acute insulin and C-peptide response to arginine. Thus, our data may indicate that AIRarg and Ac-pepRarg are better tests of the functional β-cell reserve than AIRglu and Ac-pepRglu in patients with T2D with moderately increased fasting plasma glucose levels. This is in agreement with earlier findings [9], in patients who have undergone pancreas transplantation, that AIRarg and Ac-pepRarg are less dependent on baseline plasma glucose levels. In addition, AIRarg has been shown to be a better predictor than AIRglu of β-cell secretory capacity in subjects found to be positive for islet-cell antibodies during progression to manifest type 1 diabetes and in those with type 1 diabetes who received pancreatic islet-cell transplantation [14, 15]. Rodent experimental models have also demonstrated that AIRglu is reduced to a larger extent than AIRarg following partial pancreatectomy [16]. The glucose-stimulated insulin secretion was higher in the diet/exercise-treated patients with T2D than in the other antidiabetic treatment groups. This may be partly explained by the lower fasting glucose levels (and a lesser degree of glucose toxicity) before the stimulation test compared with the groups receiving pharmacological antidiabetic treatment (who had similar pretest fasting glucose levels). On the other hand, the arginine-stimulated insulin secretion decreased more gradually in parallel with escalating therapies and was most clearly reduced in patients treated with insulin alone. This may indicate that the arginine test is more sensitive to different degrees of β-cell dysfunction at later stages of the disease than the i.v glucose tolerance test (IVGTT).

In the present study, the often-used index of β-cell function, HOMA-β, did not differ between individuals with T2D and matched healthy control participants, although HOMA-IR was significantly higher in the T2D group. This may be explained by the fact that fasting insulin concentration (included in the estimation of both HOMA-β and HOMA-IR) is strongly influenced by both β-cell function/mass and peripheral insulin resistance. This is supported by the recent finding that β-cell area was predicted by a β-cell function test (C-peptide/glucose ratio after OGTT), but not by HOMA-β index in patients with chronic pancreatitis or pancreatic tumours undergoing surgery [17]. The use of HOMA-β as a reliable measure of β-cell function has previously been questioned [18] and, in particular, its use in cohorts with TD2 treated with insulin and/or insulin secretagogues (as the relationship between fasting glucose and insulin is altered in the diseases) is controversial [7]. Pharmacological glucose-lowering treatments affecting insulin secretion, for example sulphonylureas, will also skew the HOMA measures. In the current study, there was a large variation in both HOMA-β and HOMA-IR, especially in the T2D group, possibly because approximately 40% of the subjects with T2D in this study were treated with insulin, which could affect the fasting insulin levels. Taken together, our findings confirm that HOMA-β is not a reliable measure of β-cell function in most patients with manifest T2D. Also, the lack of difference in proinsulin/insulin ratio between patients with T2D and control participants was probably due to the fact that a high proportion of the patients were treated with insulin.

Plasma C-peptide levels both at diagnosis of diabetes and on the day of the investigation were correlated with arginine-stimulated insulin release. In addition, fasting C-peptide declined with longer duration of T2D and so may have the potential to serve as a surrogate marker of β-cell function. However, fasting C-peptide levels did not differ between the T2D and healthy control groups. Therefore, overall, the role of C-peptide as a marker for functional β-cell mass remains unclear. It has previously been reported that β-cell function (measured as C-peptide concentration) decreases over time in patients with T2D, but the rate of decline varies between individuals [2]. Similar to HOMA indices and insulin levels, C-peptide will be related to insulin resistance, and again, on its own not likely to reflect β-cell function, as the degree of insulin resistance needs to be taken into account as well [19].

Several limitations of this study should be considered. First, because of the relatively small number of patients in the T2D subgroups, type II errors cannot be excluded, and thus, the power to detect true differences between subgroups is limited. Notably, the T2D patients were confirmed by the absence of islet-cell antibodies. A second limitation is the cross-sectional design; therefore, longitudinal studies to follow the change in functional β-cell mass (e.g. measured as AIRarg and Ac-pepRarg) over time are needed. Thirdly, the T2D group in the current study was relatively young with an early onset of diabetes; thus, our finding of β-cell dysfunction may not be generalizable and should be confirmed in a patient population with a broader, more typical age range.

Another limitation of this study is the short washout period of ongoing antidiabetic treatments, so that pharmacological effects could remain and affect the measures of β-cell function. However, agents that directly stimulate insulin secretion, that is sulfonylureas, were short acting, and exposures should be below therapeutic levels at the time of the investigation. In this study, differences between the groups remained essentially the same after excluding patients treated with OADs other than metformin. In addition, 40% of the patients were treated with insulin, and residual insulin from long-acting preparations taken the night before, which could still remain in the circulation at the time of testing, may have influenced the insulin secretory responses. However, when estimating the dynamic β-cell function with the i.v. arginine and glucose test, the changes in insulin and C-peptide levels are used, which may be assumed to be less affected than the fasting levels.

Hyperglycaemic and euglycaemic clamps are the gold standard method to measure β-cell function and insulin sensitivity, and fasting glucose levels should be normalized by an overnight infusion of insulin. However, in the clinical setting, less time-consuming and labour-intensive methods are warranted. In theory, the insulin secretory response after i.v. glucose administration may also have been affected by the preceding arginine stimulus, as arginine may modify the responsiveness of the β-cells to subsequent stimulation (the ‘memory’ of the β-cell) [20]. As reviewed previously [21], experimental studies using the perfused isolated rat pancreas have shown that arginine time-dependently reduces the insulin response to subsequent stimulation. Following combined challenge with arginine and glucose, however, there is no inhibition of the second response. Accordingly, studies in vivo in healthy participants and patients with T2D have shown that the priming effect of arginine is small at normal plasma glucose concentrations and absent at moderate levels of hyperglycaemia [22, 23].

In conclusion, β-cell secretory capacity measured with the arginine stimulation test (AIRarg and Ac-pepRarg) was less attenuated in patients with T2D, compared with healthy control participants, and was gradually reduced with longer disease duration and intensified antidiabetic treatment. We confirmed that the first phase of insulin secretion after a glucose load was markedly attenuated in young patients with T2D compared with healthy control participants, but did not seem to differ between patients with different stages of disease. In addition, in contrast to arginine-stimulated insulin secretion, insulin secretion after a glucose load was highly dependent on and inversely correlated with the baseline plasma glucose level. Furthermore, HOMA-β was an unreliable measure of β-cell function in this study population, and plasma C-peptide was not able to differentiate between patients with T2D and healthy control participants.

Although further validation is required, the arginine test may prove to be useful in epidemiological and clinical studies for identification of T2D patients with different degrees of β-cell dysfunction. It might also be valuable for selection and follow-up of patients in trials of interventions to preserve or restore β-cell function and mass.

Author contributions

M.S. wrote the manuscript, designed the study and participated in the interpretation of data. J.B and J.W.E designed the study, participated in the interpretation of data, contributed to the discussion and edited/reviewed the manuscript. K.C did the statistical analysis of the data. S.L participated in acquisition of data. M.K.S, M. L-O, S.G, H.J.A and L.N contributed to the design of the study, participated in acquisition and interpretation of data and reviewed the manuscript.

Conflict of interest statement

JB has received consulting and/or lecture fees from AstraZeneca. MS, JWE and KC are full-time employees of AstraZeneca. MKS, ML-O SG, SL, HJA and LN have no conflict of interests to declare. The study was funded by AstraZeneca.


Professor Björn C Carlsson at AstraZeneca is the guarantor of the study.