Autoantibodies Against Methylglyoxal-Modified Apolipoprotein B100 and ApoB100 Peptide Are Associated With Less Coronary Artery Atherosclerosis and Retinopathy in Long-Term Type 1 Diabetes

The Dialong study is a cross-sectional, controlled study on long-term survivors of type 1 diabetes conducted in 2015. Patients with type 1 diabetes from 1970 or earlier attending a specialized diabetes center were included (22). Out of 136 eligible patients, 103 joined the coronary artery disease substudy (76% of the total population). The control group without diabetes consisted of spouses/friends of patients with type 1 diabetes. First-degree relatives were excluded. The study was approved by the regional ethics committee, and all participants gave informed consent. HbA₁/HbA_1c measurements from 1980 to 2015 were collected, and the presence and degree of coronary artery disease were evaluated by clinical history and CAC imaging, and the degree of stenosis was determined by computed tomography coronary angiography (CTCA). The mean HbA_1c of 7.4 ± 0.8% (57.8 ± 8.6 mmol/mol) in the group with diabetes was similar to the average HbA_1c for all individuals with type 1 diabetes with 45 years’ duration registered in the Norwegian Diabetes Register of 7.6 ± 0.9% (59.3 ± 10.4 mmol/mol) (P = 0.08). Retinopathy was evaluated with a wide-angle camera based on SLO technique (Optos Daytona). Plasma samples were collected from all participants after an overnight fast and immediately frozen at −80°C (22).

We recruited a second population with long-term type 1 diabetes for a confirmatory analysis of the association between autoantibodies and CAD. This cohort was from the Oslo study originated in 1982, of whom a subpopulation of 27, not significantly different from the total group regarding age and glycemic control, had an invasive coronary angiography with intravascular ultrasound (IVUS) performed (23).

MGO-apoB100 and MGO-p5 were generated by incubation of apoB100 (Calbiochem, La Jolla, CA) and p5 (CSFIL KTSQC TLKEV YGFNP) (TAG Copenhagen, Copenhagen, Denmark), respectively, with 100 mmol/L MGO (Sigma-Aldrich, St. Louis, MO) in 0.2 mol/L PBS at 37°C for 24 h. The modified proteins or peptides were subsequently dialyzed against PBS (0.14 mol/L NaCl, 0.0027 mol/L KCl, and 0.010 mol/L phosphate, pH 7.4) before storage at −20°C. AGE modification was verified: 1) by demonstrating a 10.8-fold increase in AGE-specific fluorescence (excitation 370 nm and emission 440 nm) of MGO-apoB100 and a threefold increase in MGO-p5 relative to native protein or peptide, 2) by argpyrimidine fluorescence (excitation 320 nm and emission 380 nm) demonstrating an increase in MGO apoB100 and for MGO-p5 relative to native protein or peptide, and 3) by AGE carboxymethyllysine (CEL) epitopes on MGO-apoB100 and MGO-p5 measured by ELISA using a monoclonal anti-CEL antibody (KNH-30; Cosmo Bio, Tokyo, Japan) (absorbance 405 nm) (Supplementary Fig. 1A and B). Briefly, MGO-apoB100 or MGO-p5 was coated on microtiter plates, and wells were blocked with SuperBlock (Thermo Fisher Scientific, Rockford, IL) and incubated with anti-CEL (0.2 µg/mL), followed by biotinylated anti-mouse IgG (Vector Laboratories, Burlingame, CA) and alkaline phosphatase–conjugated streptavidin.

ELISAs measuring IgM and IgG in plasma against MGO-apoB100 and MGO-p5 were performed essentially according to a modified version used by Fredrikson et al. (17). Absorbance values were normalized to a control plasma pool (pooled from 11 individuals without diabetes) and are presented as arbitrary units (AU) (percentage absorbance compared with the plasma control pool). The intra-assay coefficients of variation were 3.6% for MGO-apoB100 IgM, 5.6% for MGO-p5 IgM, 11.7% for MGO-apoB100 IgG, and 10.8% for MGO-p5 IgG. The interassay coefficients of variation were 14% for MGO-apoB100 IgM, 6.5% for MGO-p5 IgM, 16% for MGO-apoB100 IgG, and 11.5% for MGO-p5 IgG. Antibody specificity was determined by a soluble-phase competitive ELISA (Supplementary Fig. 1C–F).

The participants without previous coronary heart disease (n = 88) in the group with diabetes and control subjects (n = 60) of the Dialong study were referred to CTCA (24). CAC and CTCA imaging were performed using a Dual Source CT scanner (Somatom Definition Flash; Siemens, Erlangen, Germany). CAC levels are given in Agatston units. The CTCA imaging protocol has been previously described (25). Normal coronary arteries were defined as no detected plaque in any of the coronary arteries on CTCA, nonsignificant stenosis as 1–50% lumen diameter stenosis, and significant artery stenosis as having >50% lumen diameter stenosis in any coronary segment.

An IVUS system with an automatic pullback device was used in the confirmatory Oslo study (iLab; Boston Scientific Corporation, Sunnyvale, CA). IVUS images were acquired at a rate of 30 frames/s and pullback speed of 0.5 mm/s. Technically satisfactorily IVUS images of all three coronary segments were performed. By IVUS, an intima thickness >0.3 mm was considered significant. Percent atheroma volume (PAV) was calculated using the equation PAV = (EEMarea − lumen area)/EEMarea × 100 (EEM, external elastic membrane).

Retinopathy was categorized as none, background, or proliferative (pan-retinal photocoagulation scars or proliferative findings) retinopathy based on retina photos (wide-angle camera based on SLO technique [Optos Daytona]) taken of the group with type 1 diabetes during the study and analyzed by one certified ophthalmologist.

The distributions of continuous variables were assessed using histograms, Q-Q plots, skewness, and kurtosis. Skewed variables were log-transformed before statistical tests. Clinical characteristics, autoantibody levels, and CTCA outcome measurements were compared between the groups using two-tailed Student t test or Mann-Whitney U test for continuous and χ² for categorical data. Spearman correlation analyses were performed to assess correlations between autoantibody levels and the continuous CTCA, CAC score, or IVUS markers. Differences among more than two groups were analyzed by ANOVA for normally distributed and Kruskal-Wallis test for nonnormally distributed variables. Odds ratios (ORs) were calculated using binary logistic regression analyses, which were performed both in univariate analyses and multivariate analyses adjusting for possible confounders. Variables for the model were chosen based on significant univariable associations with anti–MGO-apoB100 or anti–MGO-p5 or previously known to be associated with the outcome measures. There were very few smokers among the participants; therefore, this variable was not included in the model even if smoking is a known risk factor for CVD. Variables found to be best fitted for the model included were age, sex, HbA_1c, LDL, systolic blood pressure, estimated glomerular filtration rate (eGFR), and proliferative retinopathy. Persistent albuminuria was not included in the model as we did not find any significant association between persistent albuminuria and anti-apoB100 or p5 and due to the high association with eGFR. Values are presented as mean (SD), median (interquartile range [IQR]), or number of individuals (percentages). P values <0.05 were considered significant. All calculations were performed using SPSS Statistics version 25 (IBM Corp., Armonk, NY).

Clinical and demographic characteristics of the study population and previously reported CTCA data (24,25) are summarized in Supplementary Table 1. The group with type 1 diabetes had significantly higher CAC score, and women with diabetes had significant lower CAC score than men (24). The group with type 1 diabetes had a significantly lower rate of normal coronary arteries than the control group (14.1% vs. 47.6%; P < 0.001) and a higher rate of obstructive CAD measured by CTCA and previous coronary events. CTCA also showed significantly higher rates of total plaque volume and calcified plaque volume in the group with type 1 diabetes compared with the control group.

Clinical and demographic characteristics of the subjects in the group with type 1 diabetes with no coronary artery stenosis compared with those with significant stenosis (>50%) measured by CTCA are presented in Table 1. There were significantly higher levels of IgM autoantibodies against MGO-apoB100 and MGO-p5 in subjects with no coronary artery stenosis compared with those with significant stenoses (>50%) measured by CTCA: 96.2 AU (71–126.7) versus 54 AU (36.1–85.2) (P = 0.003) for anti–MGO-apoB100; and 77.4 AU (58–106) versus 36.9 AU (28.9–57.4) for anti–MGO-p5 (P = 0.005), respectively (Fig. 1). The intermediate group with nonsignificant stenoses also had lower antibody levels than the group with normal coronary arteries (P < 0.05) (Fig. 1). There were no significant findings when comparing IgG autoantibody levels and degree of stenosis on CTCA imaging (data not shown). Levels of IgM autoantibodies against MGO-apoB100 and MGO-p5 were negatively associated with the degree of coronary artery stenosis (r = −0.18; P = 0.04) and (r = −0.27; P < 0.001), respectively, and MGO-p5 with calcium score (r = −0.27; P = 0.006) in subjects with type 1 diabetes, whereas IgG for MGO-apoB100 or MGO-p5 did not associate to either degree of coronary stenosis or calcium score (Table 2).

IgM autoantibodies against MGO-apoB100 were significantly associated with the presence of >50% coronary artery stenosis (no stenosis vs. >50% stenosis) nonadjusted (odds ratio unadjusted [OR_unadj] 0.5 [95% CI 0.3–0.9]; P = 0.03) in type 1 diabetes. This association became stronger when adjusting for age, sex, HbA_1c, blood pressure, presence of proliferative retinopathy, LDL-c, and eGFR (OR adjusted [OR_adj] 0.21 [95% CI 0.06–0.6]; P = 0.01).

Also, MGO-p5 IgM was significantly associated with coronary stenosis (no stenosis vs. >50% stenosis) nonadjusted (OR_unadj 0.6 [95% CI 0.3–0.99]; P = 0.04) as well as after adjustment for age, sex, HbA_1c, presence of proliferative retinopathy, LDL-c, and eGFR (OR_adj 0.22 [95% CI 0.06–0.75]; P = 0.02) (Supplementary Table 2).

Clinical and demographic characteristics of the Oslo study population have previously been reported (26).

By IVUS, 14.8% of all arterial segments (out of 3,263 total) had a normal intima thickness <0.3 mm. Mean PAV was 33.3% (10.9%). There was a significant association between IgM levels of autoantibodies against MGO-modified apoB100 p5 and PAV (Spearman correlation: r_s = −0.49; P = 0.038). This association was still significant after adjustments for age, sex, and HbA_1c (adjusted r² = 0.58 in linear regression analysis) (Supplementary Table 3).

Individuals with type 1 diabetes with PDR had lower levels of IgM MGO-apoB100 (median [IQR] 53.7 [33.8–82.3] vs. 66 [49.6–114.4] AU; P = 0.03) and IgM MGO-p5 (36.2 [26.6–73.7] vs. 55.9 [34.2–91.4] AU; P = 0.04) than individuals without retinopathy or background retinopathy in the Dialong study (Fig. 2). In the Oslo study, 18 individuals were classified with proliferative retinopathy and 9 with background or no retinopathy. Those with proliferative retinopathy had lower numerical values of IgM MGO-apoB100 (89.7 [39] vs. 102.3 [27] AU; P = 0.4) and MGO-p5 (74.9 [21] vs. 92.5 [31] AU; P = 0.17), but the differences between the groups were not significant. In the Dialong study, IgM autoantibodies against MGO-apoB100 were significantly associated with the presence of proliferative retinopathy (background/no retinopathy vs. proliferative retinopathy) (OR_unadj 0.98 [95% CI 0.97–0.99]; P = 0.04) in type 1 diabetes. This association remained when adjusted for blood pressure, LDL-c, eGFR, HbA_1c, age, and sex (OR_adj 0.98 [95% CI 0.97–0.99]; P = 0.028). Also, MGO-p5 IgM was significantly associated with proliferative retinopathy (background/no retinopathy vs. proliferative retinopathy) (OR_unadj 0.16 [95% CI 0.03–0.95]; P = 0.04) nonadjusted and when adjusted for blood pressure, LDL-c, eGFR, age, and sex (OR_adj 0.11 [95% CI 0.01–0.85]; P = 0.034). Supplementary Table 4 shows risk factors associated with stages of retinopathy.

IgM MGO-apoB100 autoantibodies were significantly associated with systolic blood pressure and HbA_1c. IgM MGO-apoB100 autoantibodies did not associate with plasma lipid levels (Supplementary Table 5). The levels of anti–MGO-p5 were significantly higher in females than in males: median 62.4 AU (IQR 38.4–96.4) versus 47.9 AU (IQR 25.9–79.7) (P = 0.034). IgM MGO-p5 was significantly associated with eGFR and LDL, but did not show significant associations with HbA_1c (Supplementary Table 6) in the group with type 1 diabetes. In the control group, significant associations were found between IgM MGO-apoB100 and systolic blood pressure and between levels of IgM against MGO-p5 and age in the control group, but no associations were observed for LDL or HbA_1c (Supplementary Tables 5 and 6). We did not find any significant differences in autoantibody levels between statin and nonstatin users (data not shown). None of the participants used other drugs with immunomodulatory effects that could affect the autoantibody levels.

IgG MGO-apoB100 was significantly lower in type 1 diabetes compared with control subjects, whereas there were no differences in IgM autoantibody levels between the group with type 1 diabetes and the control group (Supplementary Table 1). When combining subjects with type 1 diabetes and control subjects without diabetes, we observed lower autoantibodies against MGO-apoB100 for both IgM and IgG for those who had significant stenosis versus no stenosis on CTCA: MGO-apoB100 IgM values, median 57.5 (IQR 44.2–88.9) versus 72.9 (IQR 51.1–103.9) AU (P = 0.03) and for MGO-apoB100 IgG, 53.2.3 (IQR 36.1–84.5) vs. 76.1 (IQR 47.9–91.8) AU (P = 0.005).

In the current study, we show that individuals with long-term type 1 diabetes and high plasma levels of IgM against MGO-modified LDL protein apoB100 and apoB100 p5 are characterized by 88% and 78%, respectively, reduced risk of having coronary artery disease and also that high autoantibody levels were associated with less proliferative retinopathy in the Dialong study. There was a nonsignificant tendency to lower levels of anti–MGO-apoB100 and MGO-p5 in the group with diabetes versus the control group. Our findings suggest a possible protective role of MGO-apoB100 IgM and MGO-p5 IgM against vascular complications in autoimmune diabetes.

There are several possibilities for how these autoantibodies could play a protective role in the vasculature of type 1 diabetes. An increased atherogenicity of the lipids due to MGO modification may explain some of the elevated cardiovascular risk in diabetes (9). Previous studies have shown that MGO-induced modification of LDL alters the three-dimensional structure of the apoB100 protein, making the LDL particles more prone to bind to the vessel wall matrix (9). The p5 peptide sequence of apoB100 is suggested to be involved in the binding of the matrix of the vessel wall (R. Khamis, personal communication). Furthermore, modified LDL particles are highly proinflammatory and can act as a ligand for macrophage pattern recognition receptors, including Toll-like receptors, and modified LDLs can also be engulfed by macrophages, causing cellular cholesterol accumulation amplifying Toll-like receptor signaling when stimulated (27). Autoantibodies against AGE-modified proteins and molecules may have a function by facilitating the removal of antigens or by blocking the binding of AGE epitopes to the AGE receptor or binding of MGO–LDL to matrix molecules in the vessel wall. They may also induce impairment of inflammasome-mediated innate immune response in macrophages (28,29). Lastly, IgM antibodies may retard atherosclerosis by blocking the scavenger receptor–mediated uptake of modified LDL into macrophages (30).

In addition, we found lower levels of MGO-apoB100 IgM and MGO-p5 IgM in individuals with proliferative retinopathy. However, most individuals with diabetes over 45 years’ disease duration have some background retinopathy. Nevertheless, our findings may implicate a protective role of autoantibodies in severe microvascular complications represented as proliferative retinopathy that clinically matters the most. This is in accordance with data from the Joslin study, which found that absence of proliferative retinopathy in subjects with type 1 diabetes and chronic kidney disease were associated with a decreased prevalence of CVD, suggesting common protective factors for PDR and CVD (31). Furthermore, data from the ACCORD trial and Finnish Diabetic Nephropathy Study (FinnDiane) suggest tight interconnections between micro- and macrovascular disease progression in both type 1 and type 2 diabetes (32,33).

The levels of MGO-p5 IgM were found to be significantly higher in female than in male subjects. Women were also characterized by less severe coronary disease as assessed by the CAC score (24). This is in accordance with previous findings of IgM against MGO-apoB100 in type 2 diabetes (14) and may reflect differences in the immune response during atherosclerosis in men and women, resulting in different disease phenotypes (34).

We did observe a significant inverse correlation between HbA_1c and IgM anti–MGO-apoB100 in the group with diabetes, but not in the control group. This relates to similar findings in a study of patients with type 2 diabetes and control subjects in whom glycated apoB100 was positively associated with HbA_1c and fasting glucose concentration (35). We did not find any correlation between LDL-c and anti–MGO-apoB100 IgM or IgG, but this was observed for anti–MGO-p5 IgM in type 1 diabetes, supporting a link between LDL-c and anti–MGO-p5 in diabetes (36).

While IgM antibodies against both MGO-apoB100 and MGO-p5 were associated with coronary artery stenosis in subjects with type 1 diabetes, similar associations were not present for IgG against the same antigens in the group with type 1 diabetes. This is in accordance with our previous studies, in which IgM, but not IgG, against MGO-apoB100 was associated with lower CAC in type 2 diabetes (14). In addition, IgM, but not IgG, against the apoB100-derived peptide MGO-p220 was associated with increased risk for cardiovascular events (15). Previous studies by other groups have also found differences between IgM and IgG antibodies against oxidized LDL and an association with CVD (20,37). The DCCT/EDIC study found high levels of IgG immune complexes for AGE–LDL to be associated with CVD (19). Previous data from mouse models have reported decreased IgG titer, increased formation of immune complex formation with modified LDL, and larger atherosclerotic lesions in the absence of circulating IgM (38). These studies support that both circulating IgM and IgM immune complexes may be players in the complex development of atherosclerosis. Another possible explanation of the differences in IgM versus IgG associations with CVD could be that IgM antibodies against MGO-apoB100 and MGO-p5 are T-cell independent, whereas most IgG subtypes are T-cell dependent and may reflect different pathways in atherosclerosis development (39).

The limitations of this study are: 1) the cross-sectional design, which makes this study hypothesis-generating and, thus, conclusions of causality cannot be drawn; and 2) the sample size, which increases the risk for false-negative findings and limits the possibility to adjust for all possible confounders. Multiple testing rounds in a relatively small population may weaken the statistical significance, so the results must be interpreted with caution. However, the lack of association of IgG against MGO-apoB100 or MGO-p5 to CVD presented in this study is in accordance with previous studies from our group, which have investigated apoB100 autoantibodies in larger cohorts of type 2 diabetes. The strengths of the current study are: 1) the control group, consisting of spouses/friends of the patients; 2) the strong associations found for IgM MGO-apoB100 and macro- and microvascular disease in type 1 diabetes even after adjusting for confounders; and 3) a confirmation of the results in a replication cohort of type 1 diabetes showing the same association between anti–IgM-MGO-p5 and coronary atherosclerosis, the latter assessed by invasive angiography with IVUS as a gold-standard measurement.

In summary, we demonstrate that high levels of IgM antibodies against MGO-apoB100 and MGO-p5 are associated with absence of coronary artery disease and proliferative retinopathy in patients with long-term type 1 diabetes. These observations propose a potential role of an immune response against AGE-modified self-antigens in vascular complications of autoimmune diabetes and suggest that IgM antibodies against MGO-apoB100 and MGO-p5 may have a protective function and are potential biological treatments of macro- and microvascular complications in type 1 diabetes.

Acknowledgments. The authors thank Jenifer Vallejo and Fong To at Clinical Research Centre, Malmö, Sweden, for laboratory assistance; Anne Karin Molvær, research nurse, and the staff at Norwegian Diabetes Centre for administrative help; Dag Fosmark (Oslo University Hospital) for analyzing the fundus photos; Professor Knut Dahl-Jørgensen (University of Oslo, Oslo University Hospital, Oslo Diabetes Research Center) for enthusiastic support during the years; and all of the participants in the study.

Funding. The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-PEOPLE-2013-COFUND) under grant agreement 609020 (Scientia Fellows), Oslo Diabetes Research Centre, the Norwegian Diabetes Centre, the Swedish Heart-Lung Foundation, Lund University Diabetes Center (Swedish Research Council, Strategic Research Area EXODIAB Dnr 2009-1039, Linnaeus grant Dnr 349-2006-23, and the Swedish Foundation for Strategic Research Dnr IRC15-006), Crafoordska Stiftelsen, and Direktör Albert Påhlssons Stiftelse.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.A.S., E.B., and T.J.B. designed this substudy of the Dialong study. K.A.S. wrote the initial draft of the manuscript with assistance from E.B. and T.J.B. and conducted all analyses. K.B.H., M.S., K.F.H., and J.N. critically reviewed and edited the manuscript. K.A.S. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in abstract form at the 55th Annual Meeting of the European Association for the Study of Diabetes, Barcelona, Spain, 16–20 September 2019.