Introduction

Autoantibodies are produced under specific conditions, including autoimmune diseases and cancer. With advancements in detection techniques, antibodies against all proteins in the body have been identified, and they can serve as novel biomarkers1. We previously reported that antibodies against phosphoenolpyruvate carb kinase and proprotein convertase subtilisin/kexin type 9 are increased in patients with diabetes and associated with poor prognoses2,3.

Incretin hormones are secreted in response to nutrient intake, with glucose-dependent insulinotropic peptide (GIP) and glucagon-like peptide-1 (GLP-1) released from the upper and lower small intestines, respectively4. Both hormones enhance insulin secretion from pancreatic β-cells, and their receptors are expressed in the central nervous system, where activation contributes to appetite suppression5. Beyond these shared roles, their physiological functions diverge: GIP facilitates efficient energy storage in adipocytes and plays a crucial role in bone metabolism, whereas GLP-1 exerts pleiotropic effects on the heart, kidney, and liver6,7. These distinct physiological actions indicate that GIP and GLP-1 have non-identical biological roles. Moreover, large-scale clinical trials have demonstrated that the GLP-1 receptor agonist semaglutide reduces the risk of cardiovascular and renal events, underscoring the clinical importance of the GLP-1 pathway8,9.

In this context, incretin hormones—particularly GIP and GLP-1—are recognized as central regulators of glucose metabolism and have become important therapeutic targets. However, whether autoantibodies against these peptides exist in patients with diabetes and whether they influence prognosis has not been previously addressed. To the best of our knowledge, this is the first study to systematically examine incretin autoantibodies in patients with diabetes and evaluate their potential prognostic significance.

Based on these considerations, we hypothesized that anti-incretin autoantibodies might contribute to metabolic disturbances and poor prognosis. Therefore, this study aimed to investigate the clinical relevance of measuring these antibody levels.

Materials and methods

Collection of serum samples

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The Ethics Committees of the International University of Health and Welfare (Approval No. 21-Im-037) and Chiba University (Approval Nos. 2017–251, 2018–320, and 2020–1129) approved the study protocol. All participants provided written informed consent.

In the diabetes group, patients were included if they had a prior diagnosis of diabetes mellitus and were receiving antidiabetic pharmacotherapy.

In the healthy control (HC) group, participants were individuals undergoing routine medical examination who, based on medical interviews, had no history of diagnosis or treatment for diabetes mellitus, hypertension, or dyslipidemia.

Serum samples from patients with diabetes were obtained from Chiba University Hospital, whereas those of HC were collected at Port Square Kashiwado Clinic. The presence or absence of diabetic complications among patients with diabetes is reported based on information from electronic medical records documented by diabetes specialists. All serum samples were stored at −80℃ until use.

Preparation and purification of incretin proteins

The full-length GIP and GLP-1 cDNAs were recombined into the prokaryotic expression plasmid pGEX-4 T-1. ECOSTM competent Escherichia coli BL-21 cells (Nippon Gene; Tokyo, Japan) were transformed with pGEX-4 T-1, pGEX-4 T-1-GIP, and pGEX-4 T-1-GLP-1 and cultured for 3 h in 200 mL Luria broth containing 0.1 mM isopropyl β-D-thiogalactopyranoside (Wako Pure Chemicals, Osaka, Japan). The cells were lysed using sonication in BugBuster Protein Extraction Reagent (Merck Millipore, Darmstadt, Germany), and GST, GST-GIP, and GST-GLP-1 proteins were purified as previously described10.

Measurement of serum antibody levels

Serum levels of antibodies against GIP and GLP-1 (GIP-Abs and GLP-1-Abs, respectively) were measured using the amplified luminescent proximity homogeneous assay-linked immunosorbent assay (AlphaLISA) using 384-well microtiter plates as previously described11,12. Specific reactions were calculated by subtracting the alpha photon counts of the GST control from those of the GST fusion proteins.

Briefly, recombinant full-length GIP or GLP-1 GST proteins conjugated to donor beads and anti-human IgG conjugated to acceptor beads were added to patient serum. The presence of autoantibodies brought the two beads into close proximity. Upon laser excitation, photosensitizers within the donor beads converted ambient oxygen into a singlet excited state. This singlet oxygen, with a lifetime of approximately 4 μs, diffused through the aqueous medium and reacted with chemiluminescent molecules embedded in the acceptor beads, thereby activating fluorescent molecules within the same beads. Subsequently, the fluorescent molecules emitted light at 615 nm (AlphaLISA). AlphaLISA provides performance comparable to or superior to that of conventional enzyme-linked immunosorbent assay (ELISA), while offering simpler handling and requiring smaller sample volumes. Moreover, it demonstrates higher sensitivity, enabling the detection of signals that would otherwise be obscured by background noise in ELISA, and provides a wider dynamic range.

Statistical analysis

Continuous data of the groups were compared using the Mann–Whitney U test. Correlations were examined using Spearman’s correlation analysis. The cutoff value for detecting diabetes was determined using Youden’s index derived from the receiver operating characteristic (ROC) curve analysis. X-tile software (Yale University, New Haven, CT)13 was used to determine the best cutoff level in distinguishing survival and mortality cases. The survival curves were represented using Kaplan–Meier plots. The log-rank test was used to compare the univariate analysis results. Statistical significance was defined as P < 0.05. Statistical analyses were performed using Python (version 3.11; Python Software Foundation, Wilmington, DE, USA) with the scikit-learn, pandas, and statsmodels libraries.

Results

Elevated incretin antibody levels in patients with diabetes

We examined 274 patients with diabetes (77.8% with type 2 diabetes, mean age ± SD: 63.1 ± 12.1 years) and 109 HC (mean age: 58.0 ± 5.8 years) (Table 1, upper panel). Characteristics of patients included in the analysis are presented in (Supplemental Table 1). Both anti-GIP-Ab and anti-GLP-1-Ab levels in patients with diabetes were significantly higher than those in HC (P = 0.0002 and P = 0.0036, respectively) (Fig. 1a,c; Table 1, lower panel). When the cutoff values were determined as the HC value + 2 SDs, the positive rates of GIP-Abs and GLP-1-Abs were 12.5 and 10.3%, respectively (Table 1, lower panel). The areas under the ROC curve of GIP-Abs and GLP-1-Abs were 0.623 and 0.595, respectively (Fig. 1b,d). Using the cutoff values determined by Youden’s index, the sensitivity and specificity for GIP-Abs were 46.3 and 79.8%, respectively, and those for GLP-1-Abs were 56.6 and 63.3%, respectively (Table 2). Given that both anti-GIP-Ab and anti-GLP-1-Ab levels were elevated in patients with diabetes compared with HC, we analyzed the associations between these antibodies and clinical laboratory findings specifically within those with diabetes. No association with glycated hemoglobin (HbA1c) levels was observed (Supplemental Table 2).

Table 1 Comparison of glucose-dependent insulinotropic peptide (GIP)-antibody (Ab) and glucagon-like peptide-1 (GLP-1)-Ab levels between healthy controls (HC) and patients with diabetes.
Full size table
Fig. 1
figure 1

Increased levels of serum anti-glucose-dependent insulinotropic peptide antibodies (GIP-Abs) and anti-glucagon-like peptide antibodies (GLP-1-Abs) in patients with diabetes compared with healthy controls (HC). Serum GIP-Abs and GLP-1-Abs were measured using an amplified luminescent proximity homogeneous assay-linked immunosorbent assay (AlphaLISA) and are shown in scatter dot plots (a,c). The bars represent the average and average ± SD. P values were calculated using the Mann–Whitney U test. **P < 0.01; ***P < 0.001 vs. HC specimens. The abilities of using GIP-Abs and GLP-1-Abs to detect diabetes were evaluated using receiver operating characteristic curve (ROC) analysis (b,d).

Full size image
Table 2 Results of receiver operating characteristic (ROC) analysis.
Full size table

Group comparisons

GIP-Ab and GLP-1-Ab levels were compared between men and women; patients with type 1 and type 2 diabetes mellitus; patients with and without complications; and patients who habitually smoked or consumed alcohol and those who abstained from these habits. None of the comparisons revealed significant differences in incretin levels, although GIP-Abs levels tended to correlate with dyslipidemia complications (P = 0.0721) (Fig. S1).

Regarding types of anti-hypoglycemic drugs, GIP-Ab and GLP-1-Ab levels were not correlated with insulin, GLP-1 receptor agonists, DPP IV inhibitors, metformin, or thiazolidinediones. Anti-GIP Ab levels were higher in patients using glinides (anti-GIP Abs with [n = 21] vs. without [n = 251] glinide: 4345.9 ± 3776.9 vs. 3324.8 ± 2077; P = 0.01) or α-glucosidase inhibitors (αGIs) (anti-GIP Abs: with [n = 66] vs. without [n = 206] αGI: 3751.7 ± 2941.1 vs. 3292.2 ± 1989.1; P = 0.013). Anti-GLP-1 Ab levels were higher in patients using only glinides (3751.7 ± 2941.1 vs. 3292.2 ± 1989.1) (Supplemental Table 3).

Survival analysis

We subsequently examined whether GIP-Ab and GLP-1-Ab levels were related to survival in patients with diabetes with a mean follow-up of 4.9 years (maximum 10 years). GIP-Ab and GLP-1-Ab levels were categorized into positive and negative groups using the cutoff values obtained using the X‑tile software13. The GIP‑Ab‑positive group presented a significantly more unfavorable prognosis than the GIP‑Ab‑negative group (Fig. 2a; P = 0.0072). The GLP-1-Ab-positive and -negative groups showed a similar tendency, with no statistical significance (Fig. 2b; P = 0.0640). We also categorized participants into two groups according to age (under 65 and 65 or older), HbA1c level (HbA1c < 7 and ≥ 7%), presence or absence of complications such as retinopathy and nephropathy, and smoking to draw survival curves for each group. Consequently, significant differences were obtained for age and smoking. This implies that incretin antibodies can predict a prognosis where HbA1c or the presence or absence of complications cannot, as shown in Fig. S2.

Fig. 2
figure 2

Survival analysis of patients with diabetes. Comparison of overall survival of the patients with diabetes according to glucose-dependent insulinotropic peptide antibody (GIP-Ab)-positive (GIP-Ab +) and -negative (GIP-Ab-) groups (a) and GLP-1-Ab-positive (GLP-1-Ab +) and -negative (GLP-1-Ab-) groups (b) are shown in Kaplan–Meier plots. Cutoff values were determined using X-tile software. Statistical analyses were performed using the log-rank test.

Full size image

Furthermore, after adjusting for age, the GIP antibody-positive group showed significantly worse survival than the antibody-negative group (P = 0.0057) (Fig. 3a,c), and the GLP-1 antibody-positive group also exhibited poorer survival than the negative group (P = 0.0487) (Fig. 3b,c).

Fig. 3
figure 3

Age-adjusted survival curves according to incretin antibody status. (a) Kaplan–Meier analysis demonstrated significantly poorer survival in the GIP antibody-positive group than in the antibody-negative group (P = 0.0057). (b) Similarly, the GLP-1 antibody-positive group exhibited worse survival than the antibody-negative group (P = 0.0487). (c) Baseline characteristics of the age-adjusted sample, including the number of subjects, sex distribution, and mean age (± SD) in each antibody-positive and -negative group. GIP glucose-dependent insulinotropic peptide, GLP-1 glucagon-like peptide-1, SD standard deviation.

Full size image

Discussion

To the best of our knowledge, this is the first study to examine anti-incretin antibodies in the sera of patients with diabetes. GIP-Abs and GLP-1-Abs were detected in both HC and patients with diabetes; however, their levels were significantly higher in patients with diabetes. Furthermore, a mean 4.9-year (maximum 10-year) follow-up study revealed that patients who tested positive for GIP-Abs had a significantly worse prognosis than those who tested negative for GIP-Abs. Those who tested positive for GLP-1-Abs also tended to have a worse prognosis.

Incretin promotes insulin secretion through receptors on pancreatic β cells14,15. Clinical trials have shown the glucose-lowering, cardioprotective, and renoprotective effects of GLP-19,16. Tirzepatide, a dual GIP and GLP-1R activator, has both glucose-lowering and weight-reducing effects17. Our findings suggest that incretin Ab titers are higher in patients with diabetes than in healthy individuals and that the inhibition of incretin effects may contribute to the development of diabetes. However, this cross-sectional study is limited in establishing causality. The lack of correlation between antibody titers and fasting blood glucose levels or glycated hemoglobin reflects these limitations. Additionally, the ROC analysis yielded area under the curve values below 0.65 for both GIP-Abs and GLP-1-Abs, indicating that their diagnostic capacity is limited. These results suggest that incretin autoantibodies are unlikely to be sufficient as independent diagnostic biomarkers; however, their association with prognosis highlights an alternative and potentially more relevant clinical role. Nevertheless, their consistent elevation in patients with diabetes and their association with prognosis underscore their potential value as novel prognostic rather than diagnostic markers.

Regarding the relationship with drugs, αGIs are known to increase incretin expression by suppressing glucose absorption. Increased incretin levels may have enhanced antibody production. In contrast, αGIs are known to prevent the development of diabetes; therefore, the significance of high GIP antibody titers in individuals taking αGIs remains unclear, along with the relationship between glinide and incretin antibody titers.

The mechanism by which the body produces antibodies against incretin is unclear. It is speculated that these antibodies, generated in response to previous infections, may have cross-reacted with incretin, leading to the formation of autoantibodies. This indicates that individuals who develop such autoantibodies may be prone to diabetes owing to the reduced effectiveness of incretin. Although the reason for elevated incretin antibody levels in patients with diabetes remains unclear in this cross-sectional study, incretins play a pivotal role in postprandial glucose lowering. Therefore, the presence of incretin antibodies may lead to higher postprandial glucose levels. However, because this analysis was based on fasting blood samples, we could not examine associations with postprandial glucose, postprandial insulin, or incretin levels themselves. Additionally, recent studies have shown that GIP facilitates efficient triglyceride storage in the postprandial state18. Thus, the inhibition of GIP function by antibodies may induce postprandial dyslipidemia19, which could contribute to insulin resistance and thereby be related to diabetes.

Furthermore, the relationship with prognosis is interesting. This study revealed that the overall mortality rate was associated with high levels of incretin antibodies; however, the frequency of malignant tumors and cardiovascular events as causes of death was too low to perform statistical analyses. It has been clinically demonstrated that the activation of the GLP-1R suppresses cardiovascular and renal events and that the suppression of the endogenous action of incretin by antibodies may have increased the occurrence of these events, as well as the associated mortality.

Although this remains speculative, it is conceivable that postprandial hyperglycemia and hyperlipidemia induced by incretin antibodies could also contribute to adverse survival outcomes. Indeed, numerous studies have reported that both postprandial hyperglycemia and hyperlipidemia are linked to cardiovascular events20,21,22. Additionally, the presence of incretin antibodies may interfere with the clinical efficacy of incretin-based therapies, an issue that warrants further investigation. To further explore these possibilities, long-term follow-up of the HC group is warranted to assess the development of impaired glucose tolerance, postprandial dyslipidemia, and their underlying mechanisms. This represents an important issue for future research.

The study had some limitations. Selection bias may have arisen due to participants’ recruitment based on informed consent. This study did not evaluate the relationships among postprandial glucose, blood incretin concentrations, and incretin antibodies or the association between incretin antibody levels and blood insulin and glucagon levels. Additionally, a robust multivariate analysis was not performed, which limits the ability to determine the independent contribution of incretin antibodies to prognosis. Moreover, the small sample size and cross-sectional design further limited the study’s ability to establish causality. Further studies with larger sample sizes and longer follow-up periods are needed.

In conclusion, this study provides evidence that autoantibodies against incretin hormones are elevated in patients with diabetes and associated with prognosis. While further validation is required, these observations raise the possibility that incretin autoantibodies could serve as novel prognostic biomarkers and highlight an underexplored immunological aspect of diabetes.