Over the past 40 years, the prevalence of diabetes in Chinese adults has increased significantly from 0.67% in 1980 to as high as 10.4% in 2013 ^[1]. According to the latest report by the International Diabetes Federation, there are currently 121 million diabetic patients in China ^[2]. The heterogeneity of islet function and its dynamic changes are the key mechanisms for the natural occurrence of diabetes ^[3], and the pathogenesis of type 2 diabetes mellitus (T2DM) involves insulin resistance and relative insulin deficiency. In T2DM, the deterioration of islet function results from a variety of factors, such as glycolipid toxicity, islet autoimmunity, disease course, and metabolic syndrome ^[4-6].

Glucotoxicity refers to the condition where the islet function is impaired under hyperglycemia, causing a decreased insulin secretion and reduced capacity of blood glucose metabolism, which forms a vicious circle^[7]. Many studies of glucotoxicity examined the changes of islet function in the context of variation of blood glucose levels ^{[8, 9]}, but few reports have been available to document the effects of glycosylated hemoglobin on islet function. The pattern and speed of islet β-cell degeneration in T2DM are still unclear: whether β-cell function declines at a varying speed in different disease stages or at a relatively constant rate over the disease course needs to be investigated. Some studies suggested that in T2DM, the regression of islet β-cell function consisted of two phases: the initial period with a slow functional deterioration, and the acceleration period after the turning point ^[10]. However, due to the lack of consistent evidence, the exact picture how islet function declines in T2DM remains to be fully elucidated. In this study, we conducted a cross-sectional study and compared the islet β-cell functions in type 2 diabetic patients with different glycosylated hemoglobin (HbA1c) levels, and the findings may help to depict the pattern of islet β-cell function changes in the course of T2DM.

This cross-sectional study was conducted among the patients with T2DM admitted in the Department of Endocrinology of the First Affiliated Hospital of Nanjing Medical University between December, 2014 and April, 2016 with the approval by the Ethics Committee of the hospital. Written informed consent was obtained from all the participants. The patient's gender, age, duration of illness and medications were recorded, and their blood pressure, height, body weight and body mass index (MRI) were measured. After admission, all the patients discontinued oral hypoglycemic agents and received insulin hypoglycemic therapy. Inclusion criteria: the patients were diagnosed with T2DM in line with the 1999 WHO Diabetes Diagnostic Criteria and were eligible for the Steady State Model 2 (HOMA2) calculation requirements (with fasting blood glucose levels of 3.0-25 mmol/L and fasting C peptide levels of 200-3500 pmol/L); the patients were negative for islet autoantibodies. Exclusion criteria: other types of diabetes; positivity for islet autoantibodies; renal insufficiency (serum creatinine >110 μmol/L); liver dysfunction (ALT>150 U/L); heart failure; malignant tumors; fever, hyperosmotic coma, or diabetic ketoacidosis.

The patients were divided into groups A, B, and C according to HbA1c tertiles: group A with HbA1c levels ≤7.8% (278 cases), group B with HbA1c levels of 7.8%-9.8% (260 cases), and group C with HbA1c levels >9.8% (265 cases). The patients were also stratified into groups a, b, and c according to the disease course tertiles: group a with a diabetes duration ≤4.0 years (267 cases), group b with a duration of 4.0-10.9 years (270 cases), and group c with a duration >10.9 years (266 cases).

All the patients underwent examinations of liver and kidney functions, blood glucose, blood uric acid, blood lipids and other biochemical parameters using an automated biochemical analyzer (Beckman Coulter, USA). HbA1c level was detected using ion exchange high-performance liquid chromatography. Enzymelinked immunosorbent assay (ELISA) was used to detect islet autoantibodies (IAA), including glutamate decarboxylase antibody (GADA), protein tyrosine phosphatase antibody (IA-2A), islet cell antibody (ICA), and insulin autoantibody (IAs); the detection kits for GADA, ICA, and IA-2A were purchased from Oumeng Biotech Company (Germany), and the kit for IAs was a product of Biomerica (USA). The detection procedures followed the descriptions in a previous study^[11] and the instructions of the manufacturers. The patients were deemed "antibody-negative" for the negativity of all the 4 antibodies (GADA, ICA, IA-2A and IAs).

The islet β-cell function was assessed using a standard steamed bun test. The test was performed after the initiation of insulin therapy and when the patients' fasting blood glucose was controlled below 10 mmol/L. The blood glucose levels at 30, 60, 120, and 180 min after fasting and the postprandial blood glucose levels at 30, 60, 120, and 180 min were measured using the glucose oxidase method. Chemiluminescence (Roche Diagnostics, Switzerland) was used to detect serum C-peptide and insulin levels at 0, 30, 60, 120, and 180 min after fasting and at 0, 30, 60, 120, and 180 min after meal. The area under curve (AUC) was calculated using the trapezoidal formula. The homeostasis model assessment 2-%β (HOMA2-%β) and HOMA2 insulin resistance index (HOMA2-IR) were calculated using HOMA2 software (http://www.dtu.ox.ac.uk).

To evaluate insulin secretion of the patients, InsAUC30/GluAUC30 (INSR30) was calculated as a surrogate index for insulin secretion in the early phase, and InsAUC180/GluAUC180 (INSR180) was calculated as a surrogate index for total insulin secretion ^[12]. To evaluate insulin sensitivity, Matsuda insulin sensitivity index (MatsudaISI) was calculated as $10000 / \sqrt{(G 0 \times I 0) \times(G \times I)} $, where G and I are the average levels of plasma glucose (mg/dL) and insulin (mIU/L) during oral glucose tolerance test (OGTT) ^[13]. Glucose disposition indices were used to assess the β-cell function, including both insulin secretion and insulin sensitivity ^{[12, 14]}. The disposition index 30 (DI30) was calculated as Matsuda ISI × INSR30, and the disposition index180 (DI180) as Matsuda ISI×INSR180.

Statistical analysis of the data was performed using SPSS 22.0 software. The data were tested for normality, and the normally distributed data were presented as Mean ± SD, and the non-normally distributed data (ALT, AST, TG, HOMA2-IR, HOMA2-%β, DI30, and DI180) were transformed into natural logarithmic (ln) for analysis. Correlation analysis between the variables was performed using Pearson correlation, and the correlation among multiple variables was analyzed using a general linear regression model. ANOVA was used to compare the data from multiple groups. As the baseline data did not match between the groups, the statistical results were corrected using covariance. The multivariate linear regression model was used to analyze the correlations between β-cell function parameters (HOMA2-IR, HOMA2-%β, DI30 and DI180) and the other variables. A P value less than 0.05 was considered to indicate a statistically significant difference.

The results of Pearson correlation analysis showed that HbA1c level was negatively correlated with HOMA2-IR, HOMA2-% β, DI30, and DI180 (P=0.000; Tab. 1). As HbA1c level increased, islet β-cell secretory function and insulin resistance after bun meal gradually reduced. The disease course was also negatively correlated with HOMA2-IR, HOMA2-%β, and DI180 (P < 0.05) but was not correlated with DI30 (P=0.209). At the disease course extended, the secretion function (HOMA2-%β and DI180) and insulin resistance of islet β-cells decreased gradually.

表 1(Tab.1)

Tab.1 Pearson correlation coefficients (r) of HbAlc and diabetes duration with islet β-cell function index Variable HOMA2-IR HOMA2-%β DI30 DI180 HbA1c   r -0.194 -0.408 -0.210 -0.268   P 0.000 0.000 0.000 0.000 Diabetes duration   r -0.121 -0.146 -0.044 -0.093   P 0.000 0.000 0.209 0.008

Tab.1 Pearson correlation coefficients (r) of HbAlc and diabetes duration with islet β-cell function index

The 3 groups of patients with different HbA1c levels showed no significant differences in gender distribution, systolic blood pressure (SBP), diastolic blood pressure (DBP), aspartate aminotransferase (AST), urea nitrogen (UREA), triglyceride (TG) or high-density cholesterol (HDL-C); but they differed significantly in age, disease duration, body mass index (BMI), alanine aminotransferase (ALT), creatinine (Cr), uric acid (UA), total cholesterol (TC), and low-density cholesterol (LDL-C) (all P < 0.05;Tab. 2).

表 2(Tab.2)

Tab.2 Demographic and clinical characteristics of patients grouped by HbA1c levels (Mean±SD) Parameter Group A Group B Group C P n 278 260 265 Gender (male/female) 156/122 157/103 173/92 0.092 Age (year) 59.64±11.74 58.76±12.48 56.10±13.79 0.004 Diabetes duration (year) 8.70±7.11 9.58±7.08 6.86±7.37 0.000 BMI (kg/m2) 25.14±3.63 26.00±3.67 24.86±3.18 0.001 SBP (mmHg) 133.16±18.11 134.67±18.40 134.50±19.60 0.594 DBP (mmHg) 77.61±11.26 78.52±11.80 80.06±12.20 0.050 HbA1c (%) 6.78±0.65 8.82±0.59 11.51±1.49 0.000 ALT (U/L) 24.46±18.21 29.03±22.39 25.28±20.39 0.005 AST (U/L) 22.00±9.96 23.87±15.11 22.27±13.76 0.119 Cr (μmol/L) 65.33±13.78 62.63±14.68 60.61±14.78 0.001 UREA (mmol/L) 5.48±1.43 5.43±1.40 5.37±1.77 0.708 UA (μmol/L) 308.71±76.45 296.04±76.76 283.04±78.35 0.001 TG (mmol/L) 1.45±1.15 1.59±1.39 1.58±1.48 0.225 TC (mmol/L) 4.42±0.96 4.63±1.24 4.92±1.14 0.000 HDL-C (mmol/L) 1.12±0.27 1.08±0.30 1.10±0.27 0.175 LDL-C (mmol/L) 2.84±0.77 3.05±0.91 3.27±0.88 0.000 BMI: Body mass index; SBP: Systolic blood pressure; DBP: diastolic blood pressure; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; Cr: Creatinine; UREA: Urea nitrogen; UA: Uric acid; TG: Triglyceride; TC: Total cholesterol; HDL-C: High-density cholesterol; LDL-C: Low-density cholesterol.

Tab.2 Demographic and clinical characteristics of patients grouped by HbA1c levels (Mean±SD)

As shown in Tab. 3, the patients with different HbA1c levels showed significant differences in HOMA2-IR, HOMA2-% β, DI30 and DI180 (P=0.000), even after covariance correction (P=0.000). HOMA2-%β, DI30 and DI180 in group A were significantly higher than those in the other two groups. With the increase of HbA1c level, HOMA2-%β gradually decreased, and the decrement was more obvious in group C.

表 3(Tab.3)

Tab.3 β-cell functions of the patients grouped by HbA1c levels (Mean±SD) Variable Group A Group B Group C P Adjusted P lnHOMA2-IR 0.21±0.97 0.15±1.17 -0.27±1.27 0.000 0.000 lnHOMA2-%β 3.81±0.84 3.39±0.97 2.89±0.91 0.000 0.000 lnDI30 2.25±0.52 1.92±0.49 1.94±0.65 0.000 0.000 lnDI180 2.68±0.56 2.31±0.55 2.25±0.66 0.000 0.000

Tab.3 β-cell functions of the patients grouped by HbA1c levels (Mean±SD)

There was no obvious difference in SBP or UA level among the 3 groups (P>0.05), but they differed significantly in gender distribution, age, BMI, DBP, HbA1c, ALT, AST, Cr, UREA, TG, TC, HDL-C and LDL-C levels (all P < 0.05; Tab. 4).

表 4(Tab.4)

Tab.4 Demographic and biochemical profiles of the patients with different diabetes durations (Mean±SD) Parameter Group A Group B Group C P n 267 270 266 Gender (male/female) 178/89 165/105 143/123 0.009 Age (year) 53.22±13.91 56.98±10.97 64.39±10.55 0.000 Diabetes duration (year) 1.17±1.21 7.16±2.17 16.85±5.26 0.000 BMI (kg/m2) 25.63±3.65 25.46±3.63 24.90±3.28 0.044 SBP (mmHg) 133.35±19.23 133.24±18.82 135.70±17.98 0.231 DBP (mmHg) 80.46±11.78 79.18±11.80 76.48±11.44 0.000 HbA1c (%) 9.52±2.64 8.67±1.90 8.82±1.83 0.000 ALT (U/L) 31.08±25.36 25.37±18.73 22.18±14.84 0.000 AST (U/L) 25.01±16.75 21.61±10.53 21.45±10.72 0.004 Cr (μmol/L) 61.48±12.91 62.42±13.91 64.83±16.39 0.023 UREA (mmol/L) 5.03±1.39 5.47±1.48 5.80±1.65 0.000 UA (μmol/L) 298.20±73.62 297.74±78.89 292.53±80.89 0.648 TG (mmol/L) 1.64±1.15 1.52±1.47 1.46±1.39 0.005 TC (mmol/L) 4.81±1.14 4.64±1.10 4.52±1.16 0.012 HDL-C (mmol/L) 1.05±0.24 1.08±0.25 1.16±0.31 0.000 LDL-C (mmol/L) 3.21±0.90 3.04±0.84 2.89±0.85 0.000

Tab.4 Demographic and biochemical profiles of the patients with different diabetes durations (Mean±SD)

Tab. 5 showed that HOMA2-IR, HOMA2-%β, and DI180 differed significantly among the 3 groups of patients with different diabetes durations (P < 0.05), but DI30 was comparable among the 3 groups (P=0.123). After covariance correction, the differences were significant for all the 4 parameters (HOMA2-IR, HOMA2-%β, DI30, and DI180; P=0.000). The data suggested that DI30 and DI180 decreased progressively with the extension of the disease course.

表 5(Tab.5)

Tab.5 β-cell functions of the patients with different diabetes durations (Mean±SD) Variable Group A Group B Group C P Adjusted P lnHOMA2-IR 0.16±1.12 0.02±1.10 -0.09±1.25 0.045 0.000 lnHOMA2-%β 3.51±0.85 3.38±0.95 3.22±1.12 0.003 0.000 lnDI30 2.07±0.60 2.07±0.56 1.98±0.56 0.123 0.000 lnDI180 2.51±0.62 2.43±0.64 2.32±0.60 0.002 0.000

Tab.5 β-cell functions of the patients with different diabetes durations (Mean±SD)

When using lnHOMA2-IR as the dependent variable, HbA1c, diabetes duration, and BMI were all independent factors affecting HOMA2-IR in T2DM patients; HbA1c and diabetes duration were negatively correlated with lnHOMA2-IR, and BMI was positively correlated with lnHOMA2-IR. Similarly, when using lnHOMA2-% β as the dependent variable, HbA1c, diabetes duration, and BMI were all independent factors for affecting HOMA2-%β; HbA1c and diabetes duration were negatively correlated and BMI was positively correlated with lnHOMA2-% β. When using lnDI30 as the dependent variable, HbA1c, diabetes duration, and BMI all independently affected DI30; HbA1c and diabetes duration were negatively correlated with lnDI30. When using lnDI180 as the dependent variable, HbA1c, diabetes duration, and BMI all independently affected DI180, and HbA1c and diabetes duration were negatively correlated with lnDI180 (Tab. 6).

表 6(Tab.6)

Tab.6 Multivariate linear regression analysis with different islet β- cell function parameters as the dependent variable Variable β t P Model 1 (lnHOMA2-IR)   HbA1c -0.100 -5.773 0.000   Diabetes duration -0.018 -3.484 0.001   BMI 0.106 9.880 0.000 Model 2 (lnHOMA2-%β)   HbA1c -0.187 -13.600 0.000   Diabetes duration -0.024 -5.746 0.000   BMI 0.072 8.506 0.000 Model 3 (lnDI30)   HbA1c -0.060 -6.531 0.000   Diabetes duration -0.007 -2.409 0.016   BMI 0.016 -2.771 0.006 Model 4 (lnDI180)   HbA1c -0.084 -8.652 0.000   Diabetes duration -0.012 -4.186 0.000   BMI 0.018 -3.058 0.002

Tab.6 Multivariate linear regression analysis with different islet β- cell function parameters as the dependent variable

We analyzed the associations of HbA1c and disease course with islet β-cell function in T2DM patients. To eliminate the confounding effect of islet autoantibodies on the evaluation of β-cell function, we excluded all the patients positive for islet autoantibodies. We calculated the islet β-cell function (HOMA2-% β) and insulin resistance index (HOMA2-IR) using the glucose-C peptide model^[15-17], and used the disposition index (DI30 and DI180) for assessing islet function after insulin resistance correction. The validity of all these indexes has been confirmed previously^[18]. Our analysis suggests that HbA1c is negatively correlated with insulin resistance and islet β-cell function in T2DM patients. With the increase of HbA1c level, the degree of insulin resistance and islet β-cell secretion function gradually decreased. The disease course was negatively correlated with HOMA2-IR, HOMA2-%β, and DI180, but was not correlated with DI30. The definite relationship between disease course and the secretion function of islet β-cells was not seen in early-stage diabetes, possibly due to impaired glucose tolerance and reduced insulin secretion in the early stage of T2DM ^[3].

HbA1c can reflect the average blood glucose level in the recent 2-3 months with a good stability. Although there are guidelines recommending HbA1c for diabetes diagnosis^[19], it is not mentioned in the latest guidelines for T2DM in China^[20]. HbA1c has become an important indicator for clinicians to initiate drug therapy and adjust hypoglycemic regimens: an HbA1C level ≥7.0% is an important indicator for initiating drug therapy in T2DM patients, and HbA1c≥9.0% is recommended as a criterion for short-term intensive insulin therapy in newly diagnosed T2DM patients ^[20]. In this study, we found that the patients with a low HbA1c level (≤7.8%) had significantly higher mean HOMA2-% β than those with moderate (7.8%-9.8%) and high (>9.8%) HbA1c levels. Similarly, DI30 and DI180 were significantly higher in the low HbA1c group than in the moderate and high HbA1c groups. Insulin secretion by islet β-cells at the basal level (HOMA2-%β), in the early stage (DI30), and across the whole course (DI180) were better in the groups with well-controlled blood glucose, as compared to the groups with poorly controlled blood glucose; also, a HbA1c level >9.8% indicates significant impairment of islet β-cell function.

Hyperglycemia can cause apoptosis and functional impairment of islet β-cells through multiple mechanisms, such as oxidative stress, hypoxia, endoplasmic reticulum stress, decreased insulin gene expression, and increased production of inflammatory factors ^{[7, 21-23]}. Our results suggest that active glycemic control in T2DM patients can improve islet β-cell function, which is consistent with most studies^{[24, 25]}. Moreover, some studies suggested that in patients with newly diagnosed T2DM, short-term intensive insulin therapy could significantly improve islet β-cell function and clinically relieve diabetes^[26]. It is possible that the islet cells are in varying states at the same time: desensitization, exhaustion or apoptosis. Therefore, after blood glucose level becomes normal, some β cells can recover their normal function of insulin secretion ^[23]. We also noted significant differences in HOMA2-IR among the patients with different HbA1c levels, as was consistent with the findings by Li et al^[25]. However, the effect of HbA1c on HOMA2-IR is still unclear and needs further investigation.

We found that an extended disease course of T2DM was associated with progressive deterioration of β-cell secretion function, although the decline, based on our data, appeared not statistically significant. The United Kingdom Prospective Diabetes Study (UKPDS) ^[27] reported a β cell function decline rate of 5% per year in T2DM. But in a recent study, researchers followed T2DM patients negative for islet autoantibodies for 20 years and found no signs of significant decline in islet β-cell function in these patients^[28]. A study conducted among Chinese patients^[29] suggested that within 1 year after diagnosis, the islet function of T2DM patients decreased significantly compared with the control group; but after 10 years, the islet function decreased at a much lower rate, possibly due to repair of islet dedifferentiation after glycemic control ^[30].

In summary, we found that in T2DM patients, the secretion function of islet β-cells decreased with the increase of HbA1c level and the extension of the disease course, but the disease course did not produce as strong an effect as HbA1c level on islet β-cell function. These results suggest that T2DM patients with long-term treatment but poorly controlled blood glucose can still benefit from active and intensive insulin therapy in terms of blood glucose control and islet β-cell function.