Key findings

  • 220newly diagnosed T2DM patients treated with metformin for 3 months, compared with 155 healthy controls
  • 0.74 vs 0.86baseline abundance of N-glycan peak 2, good versus poor response group (P = 0.029)
  • 0.635 vs 0.594AUROC for predicting good response: baseline peak 2 model versus baseline CP2h/FCP model
  • 0.719AUROC when all 12 baseline N-glycans were combined in one model to distinguish good from poor response

Abstract

Objective
To evaluate the predictive power of serum N-glycans for efficacy of metformin in patients with type 2 diabetes mellitus (T2DM).
Methods
A total of 155 healthy controls and 220 newly diagnosed T2DM patients treated with metformin for 3 months were enrolled. Serum N-glycan profile was determined by DNA sequencer-assisted fluorophore-assisted carbohydrate electrophoresis (DSA-FACE). Logistic regression analysis was used to establish the N-glycan model, and receiver operating characteristic curve (ROC) analysis was used to analyze the predictive power of the N-glycan model for metformin efficacy.
Results
The abundance of N-glycan peak 2 at baseline had a significant difference between the good (fasting plasma glucose < 7 mmol/L, glycated hemoglobin A1c < 6.5% and 2-h post-prandial plasma glucose < 11.1 mmol/L) and poor (other patients) response groups post-treatment, which may be a biomarker for predicting metformin efficacy. The area under ROC (AUROC) of peak 2 (0.635) for the predictive model of good response was greater than CP2h/FCP (C-peptide 2h/fasting C-peptide) (0.594). Peak 2 combined with other 11 N-glycans at baseline was used for establishing the predictive model to distinguish the good response from the poor response patients (AUROC = 0.719, sensitivity = 55.0% and specificity = 83.4%).
Conclusion
Serum N-glycans may have potential value as biomarkers for predicting efficacy of metformin in T2DM patients.

Introduction

Type 2 diabetes mellitus (T2DM) is a well-recognized cause of death and is experiencing a rapid increase in global prevalence1–3. In addition to hyperglycemia, numerous biochemical alterations in T2DM contribute to metabolic dysregulation4–6. The global cases with diabetes in 2021 were estimated to 536.6 million people, rising to 783.2 million in 20457. Metformin, a biguanide derivative, remains the first-line pharmacological therapy for T2DM due to its efficacy, safety, and low cost. However, a significant inter-individual variability in response to metformin exists, with approximately 30%-40% of patients failing to achieve adequate glycemic control8. Identifying biomarkers that can predict metformin response is crucial for optimizing treatment strategies and achieving personalized medicine in T2DM management. However, there are still no reliable biomarkers to predict effective treatments.

N-linked oligosaccharides of glycoproteins (N-glycans), complex carbohydrates attached to proteins, play essential roles in various biological processes, including protein folding, cell-cell communication, and immune response. N-glycans necessary to maintain an ordered cellular environment have great potential to serve as biomarkers of several diseases9,10. N-glycans are involved in pathological situations such as cancer and inflammation11–13. N-glycans could be related not only to inflammation but also to alteration of important metabolic pathways. Specifically, changes in N-glycan structures have been linked to insulin resistance, cell dysfunction, and chronic inflammation, all of which are key features of T2DM pathophysiology14. Based on DNA sequencer-assisted fluorophore-assisted carbohydrate electrophoresis (DSA-FACE) technology, N-glycan profiling of the human serum could be as biomarkers in clinical practice15.

The oral glucose tolerance test (OGTT) is the most commonly used for evaluating whole-body glucose tolerance in vivo. Many attempts have been made to assess insulin sensitivity from the OGTT which is measurements of post-prandial plasma glucose (PPG)16,17. The 2-h post-prandial plasma glucose (PPG2h) level during an OGTT is a strong predictor of future T2DM18. In the previous study, we found that the baseline serum N-glycan peak 9 have a potential power to distinguish non-responder (HbA1c level increased during 3 months treatment as compared with the baseline level) from responder (HbA1c level decreased) in T2DM patients receiving metformin therapy19. Given the close relationship between N-glycans and T2DM pathogenesis, we hypothesize that specific N-glycan profiles may serve as potential biomarkers for identifying good responders to metformin therapy in patients with T2DM. In order to further explore the clinical value of N-glycans for predicting treatment response to metformin in T2DM patients, we compared the changes in serum N-glycans and clinical indicators (mainly OGTT parameters) between healthy subjects and T2DM subjects at baseline and 3-month of treatment, and investigated the possibility of N-glycans as a potential biomarker to predict the good response to metformin therapy in T2DM patients. This study aims to investigate the association between N-glycans and metformin response in T2DM patients, with the ultimate goal of developing a predictive biomarker for personalized treatment strategies.

Materials and methods

Study design

This study was approved by the ethics committees of Peking University Health Science Center and Beijing Friendship Hospital (the approval number is IRB00001052-21108), and the informed consent was obtained from all patients and healthy controls. The workflow of this study is illustrated in Fig. 1. We recruited 220 treatment-naive patients with T2DM and 155 healthy individuals who matched with T2DM patients according to demographic characters such as age and sex. Patients with T2DM were recruited from the Department of Endocrinology at Beijing Friendship Hospital, while healthy controls were enrolled from the Physical Examination Center of the hospital. The T2DM patients met the following inclusion criteria: (1) patients with initial diagnosis of T2DM based on the international classification of disease 10 (ICD-10) criteria; (2) glycated hemoglobin A1c (HbA1c) ranged from 6.5% to 12% in baseline venous blood (3) between 18 and 70 years of age.; Individuals were excluded based on the following criteria: (1) mental illness patients receiving treatment or others who cannot sign the informed consent due to various reasons; (2) cancer patients undergoing treatment; (3) history of severe hepatic, nephritic and cardiogenic diseases; (4) patients with other types of diabetes mellitus. T2DM patients were administrated metformin therapy at a dose of 1500mg~2000mg per day and were followed up for 3 months. Serum samples and clinical data were collected from T2DM patients at baseline and 3-month, and the same procedure for healthy controls. All samples were stored in a refrigerator set at -80°C.

Figure 1. Workflow diagram

Serum N-glycan detection method

In general, the DNA sequencer–assisted fluorophore-assisted carbohydrate electrophoresis (DSA-FACE) method was used to detect and analyze the serum N-glycan profiles of individuals. The N-glycans in serum glycoprotein were separated and labeled with fluorescent substance using the N-glycomics analysis kit (developed by Jiangsu SysDiagno Biotechnology Co., Ltd.). The method was performed as described previously19. Finally, the N-glycan solution was used to detect N-glycan profiles by ABI3500dx genetic analyzer (Applied Biosystems, Foster City, CA). Twelve specific N-glycan peaks were obtained in each sample. The data of N-glycan profiles were analyzed using GeneMapper software version 4.1. The abundance of each peak was quantified by normalizing its height to the sum of the heights of twelve peaks.

Fiveitime point oral glucose tolerance test (OGTT)

A 75-g OGTT was performed after an overnight fast of at least 10 h at baseline and 3-month therapy. Blood samples were obtained before and at 30, 60, 120, and 180 min after ingesting a 75-g glucose solution. Blood collected during the studies was separated into serum. Glucose concentrations were measured using an enzymatic method in Department of Laboratory Medicine at Beijing Friendship Hospital.

Statistical analysis

Matched clinical data comparison was conducted through the Wilcoxon test, while the Mann-Whitney U test was applied in two unpaired groups. The relative abundance of N-glycans between two groups was compared by student's t test and paired t test between matched groups. The correlation between baseline serum N-glycans and demography characteristics in T2DM patients was assessed by Spearman's correlation analysis. N-glycans models were developed by logistic regression analysis, which were used to evaluate the serum N-glycan for predicting metformin treatment response, and for differentiating good response patients from poor response based on whether their FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L after 3-month therapy. The discriminant ability of the models was evaluated by receiver operating characteristic curve (ROC) analysis. Area under ROC (AUROC), sensitivity and specificity were calculated for assessing diagnostic power. All statistical analysis was conducted using SPSS (Statistical Package for Social Sciences) software 24.0. Two-tailed tests were used in all cases, and P-value<0.05 was considered statistically significant.

Results

Subject characteristics

We examined 220 treatment-naive patients with T2DM and 155 healthy individuals who matched with T2DM patients according to demographic characters such as age and sex (Supplementary Table 1). There were no significant differences in age and sex ratio between the healthy control subjects and the T2DM patients, Therefore, the confounding effects caused by factors such as age and sex were excluded.

Several clinical data were significantly changed in T2DM patients at baseline compared to those in healthy controls, as shown in Supplementary Table 1. For example, there were significantly higher fasting plasma glucose (FPG) (8.42 vs. 5.18 mmol/L), HbA1c (7.48% vs.5.47%), and triglycerides (TG) (2.51 vs.1.75 mmol/L) in T2DM patients compared with those in the control group. Supplementary Fig. 1 shows that T2DM patients exhibited significant improvements in OGTT results after 3 months of metformin treatment (Supplementary Table 1).

Differences in serum N-glycan profiling between type 2 diabetes mellitus (T2DM) patients and healthy control subjects

We sought to investigate differences in serum N-glycan profiles between T2DM patients and healthy controls. Fig. 2 shows representative serum N-glycan profiles of healthy control and T2DM patients at baseline and at 3 months of treatment.

Figure 2. Serum N-glycan profiles of healthy controls and type 2 diabetes mellitus (T2DM) patients at baseline and 3-month of treatment. The legend shows the specific structure of the detected glycans, and P/Peak indicates the N-glycan peak. Peak0: biantennary N-glycan (NGA2), peak1: agalacto core α1,6-fucosylated biantennary N-glycan (NGA2F), peak2: agalacto core α1,6-fucosylated bisecting biantennary N-glycan (NGA2FB), peaks3 and 4: monogalacto core α1,6-fucosylated biantennary N-glycan (NG1A2F), peak5: bigalacto biantennary N-glycan (NA2), peak6: bigalacto core α1,6-fucosylated biantennary N-glycan (NA2F), peak7: bigalacto core α1,6-fucosylated bisecting biantennary N-glycan (NA2FB), peak8: triantennary N-glycan (NA3), peak9: branching-1,3-fucosylated triantennary N-glycan (NA3Fb), peak10: tetragalacto tetraantennary N-glycan (NA4), and peak11: core α1,6-fucosylated tetraantennary N-glycan (NA4F). 🟡: galactose; 🟦: N-acetylglucosamine; 🟢: mannose; 🔺: fucose

The 12 measured data on the abundance of N-glycan peaks between healthy control and T2DM patients were shown in Supplementary Table 2. Compared with healthy controls, there were significant changes in the abundance of several N-glycans in the serum of T2DM patients, in which the abundance of peak 0, 1, 2, 3 and 4 were decreased significantly, while the abundance of peak 5 and 8 were increased. Interestingly, after three months of treatment, the abundance of peak 2 in T2DM patients was significantly improved compared to baseline, approaching that of the healthy control (Supplementary Table 2). Specifically, there was a decrease in the abundance of simple N-glycans in the baseline serum of T2DM patients, including diantennary N-glycans, core fucosylated or monogalactosylated modified diantennary N-glycans, whereas an increase in the abundance of complex non-fucosylated di- and tri-antennary N-glycans modified by di- and tri-galactosylation. Additionally, we observed a decrease in the abundance of core α-1, 6 fucosylated N-glycans and an increase in the abundance of non-core fucosylated N-glycans. Our findings suggest that the serum N-glycan profiles of T2DM patients differ significantly from those of healthy subjects and peak 2 may closely associate with treatment response in patients with T2DM.

Association of N-glycans with efficacy of metformin in type 2 diabetes mellitus (T2DM) patients

Many attempts have been made to assess insulin sensitivity from the OGTT which were measurements of post-prandial plasma glucose (PPG) (Bergman et al., 1985, Matsuda and DeFronzo, 1999). Supplementary Fig.1 shows that T2DM patients exhibited significant improvements in OGTT results after 3 months of metformin treatment. According to the World Health Organization guideline20, PPG2h value after 75 g OGTT ≥ 11.1 mmol/L, clinically diagnosed as T2DM.

Type 2 diabetes mellitus (T2DM) patients were divided into two groups based on whether their PPG2h decreased to normal level (PPG2h < 11.1 mmol/L) after 3 months of therapy. After 3 months of metformin treatment, 35 T2DM patients had their PPG2h < 11.1 mmol/L (74 missed). However, 111 patients had their PPG2h ≥ 11.1 (mmol/L), indicating these patients did not response well to metformin. In order to explore the correlation between the efficacy of metformin and serum N-glycans at baseline in T2DM patients, we analyzed differences of the serum N-glycans at baseline between PPG2h < 11.1 (mmol/L) and PPG2h ≥ 11.1 (mmol/L) after 3 months treatment. As shown in Table 1, we observed significant difference in the abundance of peak 2 at baseline (P<0.05) (Fig. 3A).

Table 1. Comparison of N-glycan profiles in T2DM baseline between the PPG2h < 11.1 (mmol/L) and ≥ 11.1 (mmol/L) (after 3 months) response groups

N-glycansHealthy controls
(A)
N=155
PPG2h < 11.1 (mmol/L)
(B)
N=35
PPG2h ≥ 11.1 (mmol/L)
(C)
N=111
PBC
Peak 01.01±0.460.93±0.330.90±0.350.633
Peak 14.82±1.454.23±1.454.47±1.590.411
Peak 21.10±0.320.75±0.250.88±0.280.017
Peak 34.00±0.823.50±0.823.54±0.820.810
Peak 43.67±0.673.25±0.723.30±0.650.730
Peak 550.12±3.3051.63±3.8351.42±3.380.765
Peak 615.27±2.2915.15±2.3714.71±2.250.320
Peak 74.21±1.014.08±1.034.08±0.940.993
Peak 89.27±2.099.93±2.1310.26±2.260.451
Peak 93.88±1.403.97±2.033.80±1.820.648
Peak 101.81±0.541.69±0.751.82±0.750.395
Peak 110.83±0.380.88±0.500.81±0.420.427

Note: 74 missed. PBC is the comparison between thePPG2h < 11.1 (mmol/L) and ≥ 11.1 (mmol/L) (after 3 months) response groups

Figure 3. Comparison of peak2 in T2DM baseline between the PPG2h < 11.1 (mmol/L) and ≥ 11.1 (mmol/L) (after 3 months) response groups. A, Violin plots showing the distribution of peak2 in T2DM baseline between the PPG2h < 11.1 (mmol/L) and ≥ 11.1 (mmol/L) (after 3 months) response groups. The dashed lines represent the median (large dashes) and the interquartile range (dark spot). Student’s t test was used. *p<0.05. B, ROC curve plots for prediction of metformin treatment response. Green line: the model constructed by 12 baseline serum N-glycans; blue line: the model constructed by baseline N-glycan peak2; yellow line: reference line

The binary logistic regression analysis was used to evaluate the baseline serum N-glycan to predict metformin treatment response. The abundance of N-glycan peak 2 at baseline was used to establish the prediction model (Y1), the AUROC was 0.629 (95%CI: 0.524-0.734), with a sensitivity of 66.7% and specificity of 57.1%. Furthermore, when enrolled total of 12 serum N-glycans at baseline in the model (Y2), the AUROC was 0.709 (95%CI: 0.611-0.806), with a sensitivity of 63.1% and specificity of 71.4% (Fig. 3B). The prediction models were shown as follows:

Y1=1.879*peak2-0.373

Y2=12.381*peak0+12.569*peak1+16.678*peak2+13.010*peak3+12.636*peak4+12.815*peak5+12.755*peak6+12.661*peak7+12.898*peak8+12.952*peak9+13.200*peak10+12.124*peak11-1282.050

N-glycans at baseline were associated with multiple measures of the efficacy of metformin

After 3 months of metformin treatment, 31 patients with T2DM had their FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L (1 missed and 57 no response). However, 131 patients had their FPG ≥ 7 mmol/L, or HbA1c ≥ 6.5% or PPG2h ≥ 11.1 mmol/L, indicating these patients did not response well to metformin. In order to explore the correlation between the efficacy of metformin and serum N-glycans at baseline in T2DM patients, we analyzed differences of the serum N-glycans at baseline between patients with FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L and other patients after 3 months of treatment. Patients with FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L were defined as the good responder, whereas other patients were defined as the poor responder.

As shown in Table 2, we observed significant difference in the abundance of peak 2 at baseline between the good and poor response groups (P<0.05) (Fig. 4A).

Table 2. Comparison of N-glycan profiles in T2DM baseline between good and poor response groups (after 3 months)

N-glycansGroup with good response
N=31
Group with poor response
N=131
P
Peak 00.94±0.340.91±0.300.539
Peak 14.08±1.354.48±1.750.233
Peak 20.74±0.230.86±0.300.029
Peak 33.43±0.773.47±0.840.786
Peak 43.24±0.693.21±0.670.864
Peak 551.65±3.6751.31±3.610.642
Peak 615.14±2.3914.75±2.580.455
Peak 74.03±0.944.09±0.950.756
Peak 810.00±2.1710.50±2.190.259
peak94.17±2.043.68±1.640.157
peak101.68±0.781.83±0.740.317
peak110.90±0.490.89±0.530.928

Note: The good response is defined as patients with FPG < 7 (mmol/L), HbA1c < 6.5% and PPG2h < 11.1 mmol/L after 3 months of treatment, the poor response is defined as others patients after 3 months of treatment. 1 miss and 57 no response

Figure 4. Comparison of peak2 in T2DM baseline between good and poor response groups (after 3 months). A, Violin plots showing the distribution of peak2 in T2DM baseline between good and poor response groups (after 3 months). The good response is defined as patients with FPG < 7 (mmol/L), HbA1c < 6.5% and PPG2h < 11.1 mmol/L after 3 months of treatment, the poor response is defined as others patients after 3 months of treatment. The dashed lines represent the median (large dashes) and the interquartile range (dark spot). Student’s t test was used. *p<0.05. B, ROC curve plots for prediction of metformin treatment response. Green line: the model constructed by 12 baseline serum N-glycans; blue line: the model constructed by baseline N-glycan peak2; yellow line: reference line

The binary logistic regression analysis was used to evaluate the serum N-glycan to predict the good treatment response of metformin. The abundance of N-glycan peak 2 at baseline was used to establish the prediction model (Y3), the AUROC was 0.635 (95%CI: 0.530-0.740), with a sensitivity of 64.9% and specificity of 61.3%. Furthermore, when enrolled total of 12 serum N-glycans at baseline in the model (Y4), the AUROC was 0.719 (95%CI: 0.618-0.821), with a sensitivity of 55.0% and specificity of 83.9% (Fig. 4B). The prediction models were shown as follows:

Y3=1.873*peak2-0.060

Y4=14.815*peak0+15.407*peak1+18.64*peak2+15.66*peak3+14.56*peak4+15.423*peak5+15.422*peak6+15.274*peak7+15.429*peak8+15.283*peak9+15.864*peak10+15.433*peak11-1540.531

Comparison of predicting response of the N-glycan model with baseline C-peptide 2h/fasting C-peptide (CP2h/FCP)

To further define the potential of N-glycans as marker for predicting therapeutic efficacy, we also determined another potential marker compared to N-glycans. The ratio of C-peptide release in CP2h/FCP (C-peptide 2h/fasting C-peptide) during OGTT test was found the strongest predictor of response to metformin therapy21. T2DM patients with C-peptide release data after 3 months of metformin treatment, 30 patients with their FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L (1 missed) were defined as the good responder. However, 115 patients with their FPG ≥ 7 mmol/L, or HbA1c ≥ 6.5% or PPG2h ≥ 11.1 mmol/L (16 missed) were defined as the poor responder. In order to explore the correlation between the efficacy of metformin and CP2h/FCP at baseline in T2DM patients, we compared the CP2h/FCP in T2DM baseline between the good and poor response groups after 3 months of treatment. As shown in Supplementary Table 3, we found no significant difference in baseline CP2h/FCP observed between the good and poor response groups after 3 months of treatment.

The binary logistic regression analysis was used to evaluate the CP2h/FCP to predict metformin treatment response. The CP2h/FCP at baseline was used to establish the prediction model (Ya), the AUROC was 0.594 (95%CI: 0.487-0.702), with a sensitivity of 80.0% and specificity of 52.0% (Supplementary Fig. 2). The prediction model was shown as follow:

Ya=-0.123*CP2h/FCP+1.783

The AUROC were much lower for the CP2h/FCP (AUROC = 0.594) than for the N-glycan peak 2 (AUROC = 0.635). In a word, the differences in N-glycans indicating the potential value to predict the efficacy of metformin treatment in T2DM patients.

Discussion

In this study, 220 patients with T2DM were treated with metformin and followed up for 3 months, and the relative abundance of 12 main serum N-glycan peaks at baseline and 3 months of treatment were detected by DSA-FACE method. Compared to high performance liquid chromatography and mass spectrometry, DSA-FACE has the advantages of easy operation, simple experimental pretreatment, and fast procedure22,23. Several studies have reported that DSA-FACE was used to assist in the diagnosis of various diseases24–27. We compared serum N-glycan profiles between baseline and 3-month of metformin treatment in T2DM patients, and found the dynamic changes in N-glycan profiles and the tendency toward the N-glycan profiling of the healthy control after metformin treatment. We defined the efficacy of metformin according to PPG2h, FPG, HbA1c levels post-treatment and compared N-glycans profiles between the good and poor response groups. The baseline N-glycan profiling in T2DM patients associated with the efficacy of metformin was revealed.

We found that the abundance of peak 2 in T2DM patients was significantly improved compared to baseline, approaching that of the healthy control (Supplementary Table 2). This suggests that peak 2 may associate with the treatment response in T2DM patients. Peak 2 (NGA2FB) belongs to core α-1, 6 fucosylated N-glycans. In this study, the proportion of α-1, 6 core fucosylation was decreased in T2DM patients at baseline, and the relative abundance of peak 2 was significantly reduced, which is consistent with other studies28,29. T2DM patients were stratified by whether their PPG2h decreased to normal level (PPG2h < 11.1 mmol/L) after 3 months of treatment in order to further investigate the correlation between the efficacy of the drug and changes of serum N-glycan. We found that the abundance of peak 2 at baseline was significant difference between PPG2h < 11.1 mmol/L and ≥ 11.1 mmol/L groups after the drug treatment (Table 1). In addition, the abundance of peak 2 at baseline was significant difference between good (patients with FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L after 3 months of treatment) and poor response groups (Table 2). It suggested that peak 2 may have the potential value to predict the efficacy of metformin treatment in T2DM patients. Peak 2 at baseline was used to establish the predictive model to distinguish the good response from the poor response with an AUROC of 0.635 (95%CI: 0.530-0.740), with a sensitivity of 64.9% and specificity of 61.3%. Moreover, peak 2 combined with other 11 N-glycan at baseline was used to establish the predictive model to distinguish the good response from the poor response patients (AUROC=0.719, sensitivity= 55.0% and specificity=83.4%).

Studies have reported that baseline CP2h/FCP as a strongest predictor of response to metformin therapy in T2DM patients21. The CP2h/FCP at baseline was used to establish the prediction model to distinguish the good response from the poor response with the AUROC of 0.594 (95%CI: 0.487-0.702), with a sensitivity of 80.0% and specificity of 52.0% (Supplementary Fig. 2). The AUROC were much lower for the CP2h/FCP than for the N-glycan peak 2, indicated that serum N-glycans were a relative strong predictor of response to metformin therapy in T2DM patients.

N-glycosylation is mediated by glycosyl transferases and glycosyl hydrolases that add and remove monosaccharides, respectively, from the maturing N-glycan structures within the endoplasmic reticulum and Golgi apparatus30. N-glycosylation has a close relationship with diabetes. For example, several studies have found that multiple glycosyl synthesis-related genes are associated with diabetes31. Different expression levels of genes lead to differentiation of the glycan synthesis process, which in turn manifests in N-glycan structures or quantitative change. Moreover, T2DM patients are in a low-grade chronic inflammatory state for a long time32, while N-glycosylation plays an important role in the occurrence and development of inflammation33,34. This indicates that the characteristics of N-glycosylation are closely related to the occurrence and development of diabetes.

In addition, N-glycans are an emerging source of specific biomarkers for diseases including cancer35, inflammatory36, autoimmune diseases37 and more. In the past decade, the continuous development of glycomics databases and the development of glyco-analytic methods have led to improvements in throughput, resolution, and sensitivity of glycan analysis. Characterizing N-glycans in disease states cohorts can accurately reflect the host status compared to healthy control38. Furthermore, longitudinal analysis of patient N-glycosylation following diagnosis and during treatment can track disease predisposition and monitoring progression39. Changes in N-glycans provide indications of health status on the spectrum from health to disease, which aim to diagnose early, delay further development of disease, monitor treatment efficacy, and improve health status through prevention and tailored interventions. Therefore, the detection of N-glycan changes can be of great significance for early diagnosis, delaying disease progrssion and monitoring the therapeutic effect.

Meanwhile, we found that compared to health control, N-glycan profile changes in T2DM patients are characteristic changes in response to treatment. Our previous study found that baseline peak 9 predicted both response (HbA1c decreased group) and no response (HbA1c increased group)19, and in this study, we found that peak 2 predicted good response (FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L) and poor response (other patients). This study is a prospective study to further explore the correlation between the therapeutic efficacy of metformin and serum N-glycans change in T2DM patients. The observed association between specific N-glycan and metformin response may be attributed to several mechanisms. For example, N-glycans are known to modulate insulin signaling pathways, and alterations in their structures may influence insulin sensitivity, a key determinant of metformin efficacy. Moreover, N-glycans play a role in regulating inflammation, and metformin's anti-inflammatory effects may be more pronounced in individuals with specific N-glycan profiles.

The identification of N-glycan biomarkers for metformin response holds significant clinical implications. By stratifying patients based on their N-glycan profiles, clinicians can identify individuals who are more likely to benefit from metformin therapy, thereby optimizing treatment strategies and improving patient outcomes. N-glycan biomarkers are expected to become an indispensable component of personalized and precision medicine. Furthermore, this approach may help reduce the risk of adverse effects and unnecessary treatment in non-responders. However, there are some limitations in our study. This study is the results of 3 months follow-up of metformin therapy in newly diagnosed T2DM patients, the OGTT trial performed after 3 months treatment in many patients was not obtained for some reasons, resulting in a small sample size. We did not explore the underlying molecular mechanism of characteristics of N glycans with diabetes progression or drug treatment. Mechanism studies will help us to better understand the association of N-glycans with disease. Future studies should explore the functional roles of specific N-glycan structures in metformin's action and investigate potential interactions with other genetic and environmental factors.

Conclusion

In conclusion, our data clearly showed the prospective study on the dynamic changes of serum N-glycans in T2DM patients during the treatment of metformin. We found significant differences in the abundance of serum N-glycans between T2DM patients baseline and healthy controls or three months of metformin treatment. The abundance of N-glycan peak 2 at baseline had a significant difference between PPG2h < 11.1 mmol/L and PPG2h ≥ 11.1 mmol/L groups after three months of metformin treatment. Moreover, the abundance of N-glycan peak 2 at baseline had a significant difference between the good (FPG < 7 mmol/L, HbA1c < 6.5% and PPG2h < 11.1 mmol/L) and poor response groups after three months of metformin treatment. Serum N-glycans may have potential value as biomarkers to predict efficacy of metformin in T2DM patients. Our findings provide scientific basis for the development of personalized treatment strategies, ultimately improving patient care and outcomes in T2DM management.

Declarations

Acknowledgments
Authors thank the participants in the present study.
Author Disclosure Statement
The authors declare they have no conflicting financial interests.
Funding Information
This work was supported by the Major Science and Technology Special Project of China Thirteenth Five-year Plan (2018ZX10732401-003-015) and the project of Peking University - SysDiagno Biotech Joint Laboratory of Glycomics.
Ethical declaration
This study was approved by the ethics committees of Peking University Health Science Center and Beijing Friendship Hospital (the approval number is IRB00001052-21108), and the informed consent was obtained from all patients and healthy controls.
Data availability
Source data collected in this study are available from the corresponding author upon request.

Supplementary material

Table S1. Comparison of demographic characteristics and clinical data between the T2DM and healthy control

indexHealthy controls
N=155
T2DM baseline
N=220
Post treatment
N=220
PABPBC
Age (year)51.2 (22-71)50.9 (19-84)0.827
sex ratio (male/female)110/45134/860.057
BMI (Kg/m2)26.64 (19.83-38.06)27.88 (19.27-46.30)0.001
TG (mmol/L)1.75 (0.44-12.70)2.51 (0.52-27.93)0.001
CHOL (mmol/L)4.97 (1.33-7.60)5.23 (2.31-10.15)0.029
HDLC (mmol/L)1.28 (0.63-3.52)1.20 (0.68-2.39)0.018
LDLC (mmol/L)3.13 (1.63-5.18)3.28 (0.75-6.83)0.077
HbA1c (%)5.47 (4.5-6.2)7.48 (5.50-12.41)<0.001
FPG (mmol/L)5.18 (4.11-6.15)8.42 (5.40-19.62)6.87 (4.50-13.88)<0.001<0.001
PPG30min (mmol/L)13.19 (8.48-21.59)11.83 (7.57-21.64)<0.001
PPG1h (mmol/L)16.51 (10.40-23.60)14.85 (5.00-25.55)<0.001
PPG2h (mmol/L)15.64 (5.90-27.42)13.54 (6.47-23.33)<0.001
PPG3h (mmol/L)11.07 (3.63-24.91)8.73 (2.80-18.43)<0.001
FCP (µg/L)2.84 (0.27-9.30)2.51 (0.79-6.07)<0.001
CP30min (µg/L)4.77 (0.32-14.76)5.25 (1.61-18.95)<0.001
CP1h (µg/L)7.16 (0.45-16.85)8.17 (2.45-18.88)<0.001
CP2h (µg/L)9.22 (0.57-21.12)10.52 (2.82-24.32)<0.001
CP3h (µg/L)7.53 (0.50-16.91)8.04 (2.74-21.10)0.039

Note: PAB is the comparison between healthy controls and T2DM baseline, and PBC is the comparison between T2DM at baseline and at 3 months of metformin. BMI: body mass index; TG: triglyceride; CHOL: total cholesterol; HDLC: high-density lipoprotein; LDLC: low-density lipoprotein; HbA1c: glycated hemoglobin; FPG: fasting plasma glucose; PPG: postprandial glucose; FCP: fasting C-peptide content; CP 30min: C-peptide Release Test 30min results; T2DM: type 2 diabetes mellitus

Figure S1. Changes in plasma glucose concentrations during the 5-time oral glucose tolerance test. ***p<0.001

Table S2. Comparison of N-glycan profiles between healthy control and T2DM patients in baseline and post treatment

N-glycansHealthy controls (A)
N=155
T2DM baseline (B)
N=220
Post treatment (C)
N=220
PABPBC
Peak 01.01±0.460.91±0.330.85±0.280.0150.033
Peak 14.82±1.454.40±1.634.37±1.550.0110.523
Peak 21.10±0.320.84±0.291.02±0.33<0.001<0.001
Peak 34.00±0.823.53±0.823.55±0.79<0.0010.586
Peak 43.67±0.673.26±0.663.31±0.63<0.0010.118
Peak 550.12±3.3051.32±3.4851.44±3.420.0010.518
Peak 615.27±2.2914.92±2.4714.61±2.370.1590.003
Peak 74.21±1.014.09±0.984.01±0.920.2240.025
Peak 89.27±2.0910.29±2.2410.15±2.21<0.0010.064
Peak 93.88±1.403.77±1.773.94±1.790.4820.005
Peak 101.81±0.541.82±0.741.70±0.780.8830.089
Peak 110.83±0.380.85±0.481.06±0.610.674<0.001

Note: PAB is the comparison between healthy controls and T2DM baseline, and PBC is the comparison between T2DM at baseline and at 3 months of metformin

Table S3. Comparison of CP2h/FCP in T2DM baseline between good and poor response groups (after 3 months)

N-glycansGroup with good response
N=30
Group with poor response
N=115
P
CP2h/FCP3.77±1.343.43±1.630.292

Note: 17 missed in the response group

Figure S2. ROC curve plots for prediction of metformin treatment response. blue line: the model constructed by baseline CP2h/CP0h; green line: reference line

References

  1. Zimmet P, Alberti KG, Shaw J. Global and societal implications of the diabetes epidemic. Nature. 2001; 414(6865):782-7. doi: 10.1038/414782a.
  2. Tao Z, Shi A, Zhao J. Epidemiological Perspectives of Diabetes. Cell Biochem Biophys. 2015;73(1):181-5. doi: 10.1007/s12013-015-0598-4.
  3. Hu FB. Globalization of diabetes: the role of diet, lifestyle, and genes. Diabetes Care. 2011;34(6):1249-57. doi: 10.2337/dc11-0442.
  4. Reaven GM. Banting lecture 1988. Role of insulin resistance in human disease. Diabetes. 1988;37(12):1595-607. doi: 10.2337/diabetes.37.12.1595.
  5. Wajchenberg BL, Malerbi DA, Rocha MS, Lerario AC, Santomauro AT. Syndrome X: a syndrome of insulin resistance. Epidemiological and clinical evidence. Diabetes Metab Rev. 1994;10(1):19-29. doi: 10.1002/dmr.5610100103.
  6. Pickup JC, Crook MA. Is type II diabetes mellitus a disease of the innate immune system? Diabetologia. 1998;41(10):1241-8. doi: 10.1007/s001250051058.
  7. Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, Stein C, Basit A, Chan JCN, Mbanya JC, Pavkov ME, Ramachandaran A, Wild SH, James S, Herman WH, Zhang P, Bommer C, Kuo S, Boyko EJ, Magliano DJ. IDF Diabetes Atlas: Global, regional and country level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119. doi: 10.1016/j.diabres.2021.109119.
  8. Damanhouri ZA, Alkreathy HM, Alharbi FA, Abualhamail H, Ahmad MS. A review of the impact of pharmacogenetics and metabolomics on the efficacy of metformin in type 2 diabetes. Int J Med Sci. 2023; 20(1):142-150. doi: 10.7150/ijms.77206.
  9. Dall'Olio F, Malagolini N, Trinchera M, Chiricolo M. Mechanisms of cancer-associated glycosylation changes. Front Biosci (Landmark Ed). 2012;17(2):670-99. doi: 10.1016/j.arr.2012.02.002.
  10. Lan Y, Hao C, Zeng X, He Y, Zeng P, Guo Z, Zhang L. Serum glycoprotein-derived N- and O-linked glycans as cancer biomarkers. Am J Cancer Res. 2016;6(11):2390-2415.
  11. Jenkins N, Parekh RB, James DC. Getting the glycosylation right: implications for the biotechnology industry. Nat Biotechnol. 1996;14(8):975-81. doi: 10.1038/nbt0896-975.
  12. Ohtsubo K, Marth JD. Glycosylation in cellular mechanisms of health and disease. Cell. 2006;126(5):855-67. doi: 10.1016/j.cell.2006.08.019.
  13. Taniguchi N, Korekane H. Branched N-glycans and their implications for cell adhesion, signaling and clinical applications for cancer biomarkers and in therapeutics. BMB Rep. 2011;44(12):772-81. doi: 10.5483/BMBRep.2011.44.12.772.
  14. Dall'Olio F, Vanhooren V, Chen CC, Slagboom PE, Wuhrer M, Franceschi C. N-glycomic biomarkers of biological aging and longevity: a link with inflammaging. Ageing Res Rev. 2013;12(2):685-98. doi: 10.1016/j.arr.2012.02.002.
  15. Vanhooren V, Laroy W, Libert C, Chen C. N-glycan profiling in the study of human aging. Biogerontology. 2008;9(5):351-6. doi: 10.1007/s10522-008-9140-z.
  16. Bergman RN, Finegood DT, Ader M. Assessment of insulin sensitivity in vivo. Endocr Rev. 1985;6(1):45-86. doi: 10.1210/edrv-6-1-45.
  17. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care. 1999;22(9):1462-70. doi: 10.2337/diacare.22.9.1462.
  18. Unwin N, Shaw J, Zimmet P, Alberti KG. Impaired glucose tolerance and impaired fasting glycaemia: the current status on definition and intervention. Diabet Med. 2002;19:708-23. doi: 10.1046/j.1464-5491.2002.00835.x.
  19. Dong HJ, Li XH, Gu QX, Ma CF, Yuan MX, Wang ZZ, Su JR, Xu L, Chen CY, Ebule Q, Zhuang H, Liu XE.. N-glycan as new potential biomarker for predicting treatment response in patients with type 2 diabetes mellitus. Biomark Med. 2024;18(24):1113-1122. doi: 10.1080/17520363.2024.2432309.
  20. Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med. 1998;15(7):539-53. doi: 10.1002/(sici)1096-9136(199807)15:7<539::Aid-dia668>3.0.Co;2-s.
  21. Abdelgani S, Puckett C, Adams J, Triplitt C, DeFronzo RA, Abdul-Ghani M. Insulin Secretion Predicts the Response to Antidiabetic Therapy in Patients With New-onset Diabetes. J Clin Endocrinol Metab. 2021;106 (12):3497-3504. doi: 10.1210/clinem/dgab403.
  22. Lageveen-Kammeijer GSM, Kuster B, Reusch D, Wuhrer M. High sensitivity glycomics in biomedicine. Mass Spectrom Rev. 2022;41(6):1014-1039. doi: 10.1002/mas.21730.
  23. Makrydaki E, Kotidis P, Polizzi KM, Kontoravdi C. Hitting the sweet spot with capillary electrophoresis: advances in N-glycomics and glycoproteomics. Curr Opin Biotechnol. 2021;71:182-190. doi: 10.1016/j.copbio.2021.07.013.
  24. Zhao YP, Ruan CP, Wang H, Hu ZQ, Fang M, Gu X, Ji J, Zhao JY, Gao CF. Identification and assessment of new biomarkers for colorectal cancer with serum N-glycan profiling. Cancer. 2012;118(3):639-50. doi: 10.1002/cncr.26342.
  25. Fang M, Zhao YP, Zhou FG, Lu LG, Qi P, Wang H, Zhou K, Sun SH, Chen CY, Gao CF. N-glycan based models improve diagnostic efficacies in hepatitis B virus-related hepatocellular carcinoma. Int J Cancer. 2010;127(1):148-59. doi: 10.1002/ijc.25030.
  26. Chen J, Fang M, Zhao YP, Yi CH, Ji J, Cheng C, Wang MM, Gu X, Sun QS, Chen XL, Gao CF. Serum N-Glycans: A new diagnostic biomarker for light chain multiple myeloma. PLoS One. 2015;10(6):e0127022. doi: 10.1371/journal.pone.0127022.
  27. Cao X, Shang QH, Chi XL, Zhang W, Xiao HM, Sun MM, Chen G, An Y, Lv CL, Wang L, Nan YM, Chen CY, Tan ZN, Liu XE, Zhuang H. Serum N-glycan markers for diagnosing liver fibrosis induced by hepatitis B virus. World J Gastroenterol. 2020;26(10):1067-1079. doi: 10.3748/wjg.v26.i10.1067.
  28. Lemmers R F H, Vilaj M, Urda D, Agakov F, Šimurina M, Klaric L, Rudan I, Campbell H, Hayward C, Wilson JF, Lieverse AG, Gornik O, Sijbrands EJG, Lauc G, van Hoek M. IgG glycan patterns are associated with type 2 diabetes in independent European populations. Biochim Biophys Acta Gen Subj. 2017;1861(9): 2240-2249. doi: 10.1016/j.bbagen.2017.06.020.
  29. Adua E, Memarian E, Russell A, Trbojević-Akmačić I, Gudelj I, Jurić J, Roberts P, Lauc G, Wang W. High throughput profiling of whole plasma N-glycans in type II diabetes mellitus patients and healthy individuals: A perspective from a Ghanaian population. Arch Biochem Biophys. 2019;661:10-21. doi: 10.1016/j.abb.2018.10.015.
  30. Russell A, Adua E, Ugrina I, Laws S, Wang W. Unravelling Immunoglobulin G Fc N-Glycosylation: A Dynamic Marker Potentiating Predictive, Preventive and Personalised Medicine. Int J Mol Sci. 2018;19(2):390. doi: 10.3390/ijms19020390.
  31. Onengut-Gumuscu S, Chen WM, Burren O, Cooper NJ, Quinlan AR, Mychaleckyj JC, Farber E, Bonnie JK, Szpak M, Schofield E, Achuthan P, Guo H, Fortune MD, Stevens H, Walker NM, Ward LD, Kundaje A, Kellis M, Daly MJ, Barrett JC, Cooper JD, Deloukas P; Type 1 Diabetes Genetics Consortium; Todd JA, Wallace C, Concannon P, Rich SS. Fine mapping of type 1 diabetes susceptibility loci and evidence for colocalization of causal variants with lymphoid gene enhancers. Nat Genet. 2015;47(4): 381-386. doi: 10.1038/ng.3245.
  32. Calle MC, Fernandez ML. Inflammation and type 2 diabetes. Diabetes Metab. 2012;38(3):183-191. doi: 10.1016/j.diabet.2011.11.006.
  33. Irons EE, Lau J. Systemic ST6Gal-1 is a pro-survival factor for murine transitional B cells. Front Immunol. 2018;9:2150. doi: 10.3389/fimmu.2018.02150.
  34. Lontchi-Yimagou E, Sobngwi E, Matsha TE, Kengne AP. Diabetes mellitus and inflammation. Curr Diab Rep. 2013;13(3):435-444. doi: 10.1007/s11892-013-0375-y.
  35. Lin Y, Lubman DM. The role of N-glycosylation in cancer. Acta Pharm Sin B. 2024;14(3):1098-1110. doi: 10.1016/j.apsb.2023.10.014.
  36. Tijardović M, Marijančević D, Bok D, Kifer D, Lauc G, Gornik O, Keser T. Intense Physical Exercise Induces an Anti-inflammatory Change in IgG N-Glycosylation Profile. Front Physiol. 2019;10:1522. doi: 10.3389/fphys.2019.01522.
  37. Zhou X, Motta F, Selmi C, Ridgway WM, Gershwin ME, Zhang W. Antibody glycosylation in autoimmune diseases. Autoimmun Rev. 2021;20(5):102804. doi: 10.1016/j.autrev.2021.102804.
  38. Gudelja I, Lauc G, Pezer M. Immunoglobulin G glycosylation in aging and diseases. Cell Immunol. 2018; 333:65-79. doi: 10.1016/j.cellimm.2018.07.009.
  39. Flevaris K, Kontoravdi C. Immunoglobulin G N-glycan Biomarkers for Autoimmune Diseases: Current State and a Glycoinformatics Perspective. Int J Mol Sci. 2022;23(9):5180. doi: 10.3390/ijms23095180.

Article information

Authors

  • Hui-jun DongDepartment of Microbiology and Center of Infectious Diseases, School of Basic Medical Sciences, Peking University Health Science Center, Beijing 100191
  • Xiao-Hui LiDepartment of Endocrinology, Beijing Friendship Hospital, Capital Medical University, Beijing
  • Qi-Xin GuDepartment of Microbiology and Center of Infectious Diseases, School of Basic Medical Sciences, Peking University Health Science Center, Beijing 100191
  • Qiqige EbuleDepartment of Microbiology and Center of Infectious Diseases, School of Basic Medical Sciences, Peking University Health Science Center, Beijing 100191
  • Chi-Fa MaDepartment of Endocrinology, Beijing Friendship Hospital, Capital Medical University, Beijing
  • Ming-Xia YuanDepartment of Endocrinology, Beijing Friendship Hospital, Capital Medical University, BeijingCorresponding author · [email protected]
  • Zhen-Zi WangDepartment of laboratory medicine, Beijing Friendship Hospital, Capital Medical University, Beijing
  • Jian-Rong SuDepartment of laboratory medicine, Beijing Friendship Hospital, Capital Medical University, Beijing
  • Lei XuSysDiagno Biotechnology Co., Ltd, Nanjing
  • Cui-Ying Chitty ChenSysDiagno Biotechnology Co., Ltd, Nanjing
  • Chao ZhangSysDiagno Biotechnology Co., Ltd, Nanjing
  • Hui ZhuangDepartment of Microbiology and Center of Infectious Diseases, School of Basic Medical Sciences, Peking University Health Science Center, Beijing 100191
  • Xue-en LiuDepartment of Microbiology and Center of Infectious Diseases, School of Basic Medical Sciences, Peking University Health Science Center, Beijing 100191Corresponding author · [email protected]

How to cite

Dong, H.-J., Li, X.-H., et al., 2026. Baseline serum N-glycans as a predictor for the efficacy of metformin in newly diagnosed patients with type 2 diabetes mellitus. Medical Research Archives, [online] 14(10). https://doi.org/10.18103/mra.2026.0624

Record

Journal
Medical Research Archives
Section
Research Articles
Published
31 October 2026
ISSN
2375-1924
Downloads
69

Copyright and licence

Published by the European Society of Medicine under a Creative Commons Attribution 4.0 International licence (CC BY 4.0), which permits unrestricted use, distribution and reproduction in any medium, provided the original authors and source are credited.