Methylglyoxal (MGO), a major precursor for advanced glycation end products, is increased in diabetes. In diabetic rodents, inhibition of MGO prevents cardiovascular disease (CVD). Whether plasma MGO levels are associated with incident CVD in people with type 1 diabetes is unknown. We included 159 individuals with persistent normoalbuminuria and 162 individuals with diabetic nephropathy (DN) from the outpatient clinic at Steno Diabetes Center. We measured MGO at baseline and recorded fatal and nonfatal CVD over a median follow-up of 12.3 years (interquartile range 7.6–12.5 years). Data were analyzed by Cox regression, with adjustment for sex, age, HbA1c, DN, diabetes duration, smoking, systolic blood pressure, antihypertensive medication, and BMI. During follow-up, 73 individuals suffered at least one CVD event (36 fatal and 53 nonfatal). Higher MGO levels were associated with total, fatal, and nonfatal incident CVD (hazard ratios [HRs] 1.47 [95% CI 1.13–1.91], 1.42 [1.01–1.99], and 1.46 [1.08–1.98], respectively). We observed a similar trend for total mortality (HR 1.24 [0.99–1.56]). This study shows for the first time in our knowledge that plasma MGO levels are associated with cardiovascular events in individuals with type 1 diabetes. MGO may explain, at least in part, the increased risk for CVD in type 1 diabetes.
Introduction
Individuals with type 1 diabetes have an increased risk of cardiovascular disease (CVD) (1). A key mediator in the association between diabetes and CVD may be the formation of dicarbonyls, which are reactive glucose metabolites that interact with protein residues to form advanced glycation end products (AGEs) (2). Methylglyoxal (MGO), glyoxal (GO), and 3-deoxyglucosone (3-DG) have been identified as major dicarbonyl compounds (3). Of these, MGO has been identified as the most reactive dicarbonyl and has emerged as a key player in the development of diabetic complications (4,5).
MGO may contribute to diabetic CVD through several mechanisms, including the development of diabetic nephropathy (DN), endothelial dysfunction (ED) (6), and low-grade inflammation (LGI) (7). In plaques, MGO induces growth of the necrotic core, predisposing them toward rupture (8). We previously found that the MGO-derived AGE Nε(carboxyethyl)lysine (CEL) is associated with incident CVD in type 1 and 2 diabetes (9,10). However, whether plasma levels of MGO are associated with CVD in individuals with diabetes is unknown. Therefore, we investigated whether higher plasma levels of MGO, as well as the other major dicarbonyls GO and 3-DG, in type 1 diabetes are associated with incident CVD and whether the associations with incident CVD are explained by markers of DN, ED, and LGI.
Research Design and Methods
Study Population and Design
In 1993, 199 patients with type 1 diabetes and DN and 192 patients with type 1 diabetes and normoalbuminuria >18 years of age were enrolled in a prospective observational study at the outpatient clinic at Steno Diabetes Center. Patient selection and inclusion criteria are described elsewhere (11). The study was approved by the local ethics committee (Gentofte, Denmark) in accordance with the Declaration of Helsinki. All individuals provided informed consent. Measurements of biomarkers for ED and LGI; levels of the major AGEs Nε(carboxymethyl)lysine (CML), CEL, and pentosidine; and risk factors are described in detail elsewhere (9). Plasma levels of MGO, GO, and 3-DG were measured with ultra performance liquid chromatography tandem mass spectrometry (12). Measurements of other biomarkers and risk factors are described in detail elsewhere (9,13). Individuals with prior CVD (n = 24), end-stage renal disease at baseline (n = 10), and/or missing follow-up (n = 17) or plasma biomarker (n = 23) data were excluded, resulting in a study population of 321 (DN, n = 162; normoalbuminuria, n = 159).
Follow-up and Study End Points
All patients were followed until their last visit at Steno Diabetes Center, 1 September 2006, death (n = 81), or emigration (n = 3). All patients were traced through the national register during the autumn of 2006. When a patient died before 1 September 2006, the date of death was recorded and the cause of death obtained from the death certificate or necropsy reports when available. Nonfatal CVD information was retrieved from patient files at the Steno Diabetes Center or from other hospital records. In individuals who had more than one CVD event during follow-up, the first event was used in the analyses.
Statistical Analyses
All analyses were performed with SPSS version 20 for Windows software (IBM Corporation, Chicago, IL). Variables that showed a skewed distribution were natural log (ln)-transformed before analyses. Pearson correlation coefficients were calculated between the plasma dicarbonyls and the ln-transformed plasma AGE levels. An LGI and ED score was calculated as described previously (9). We examined cross-sectional associations by using linear regression analyses. Cox proportional hazards regression models were applied to investigate associations between plasma z scores of MGO, GO, and 3-DG and study end points, with adjustment for sex, age, duration of diabetes, HbA1c, and DN status (model 1). Additional adjustments were performed for BMI, systolic blood pressure, total cholesterol, current smoking, and antihypertensive treatment (model 2). Adjustments for the associations for markers of DN (estimated glomerular filtration rate [eGFR], and ln–urinary albumin excretion [ln-UAE]; model 3), ED (model 4), and LGI (model 5) were performed to evaluate the extent to which these pathophysiological processes could explain the associations between dicarbonyls and study outcomes. We investigated whether associations differed between men and women or presence of DN by adding interaction terms to model 2 investigating total CVD and total mortality.
Results
Table 1 shows participant baseline characteristics. We found that individuals who suffered a CVD event had significantly higher MGO and GO levels at baseline. We also found significantly higher MGO levels in those who died during follow-up. Plasma levels of MGO correlated significantly with GO (r = 0.89; P < 0.01) and 3-DG (r = 0.51; P < 0.01). Plasma levels of MGO correlated significantly with plasma CML (r = 0.41; P < 0.01), CEL (r = 0.36; P < 0.01), and pentosidine (r = 0.43; P < 0.01). Plasma levels of GO correlated significantly with CML (r = 0.34; P < 0.01), CEL (r = 0.20; P < 01), and pentosidine (r = 0.25; P < 0.01). 3-DG correlated significantly with CML (r = 0.13; P = 0.02) but not with CEL (r = −0.02; P = 0.69) or pentosidine (r = 0.07; P = 0.21).
Baseline characteristics according to total incident CVD and survival status at 12 years of follow-up
| Characteristic . | No CVD event (n = 248) . | CVD event (n = 73) . | P value . | Alive (n = 240) . | Dead (n = 81) . | P value . |
|---|---|---|---|---|---|---|
| Age (years) | 40.4 ± 9.6 | 45.1 ± 9.1 | <0.01 | 40.1 ± 9.1 | 45.5 ± 10.0 | <0.01 |
| Male sex (%) | 57.5 | 60.9 | 0.61 | 57.7 | 67.9 | 0.10 |
| Duration of diabetes (years) | 26.0 (21.0–32.0) | 29.0 (24.5–37.6) | <0.01 | 26.0 (21.0–32.0) | 28.0 (23.0–38.5) | <0.01 |
| HbA1c (%) | 8.0 ± 1.0 | 8.6 ± 0.9 | <0.01 | 8.1 ± 1.2 | 8.6 ± 0.8 | <0.01 |
| HbA1c (mmol/mol) | 64 ± 10.9 | 70 ± 9.8 | <0.01 | 65 ± 13.1 | 70 ± 8.7 | <0.01 |
| DN (%) | 42.7 | 76.7 | <0.01 | 40.8 | 79.0 | <0.01 |
| Retinopathy (%) | <0.01 | <0.01 | ||||
| None | 19.8 | 9.6 | 21.6 | 6.2 | ||
| Simple | 45.2 | 34.2 | 44.4 | 37.0 | ||
| Proliferative | 35.1 | 56.2 | 34.0 | 56.8 | ||
| BMI (kg/m2) | 23.9 ± 2.8 | 23.9 ± 3.5 | 1.00 | 23.9 ± 2.9 | 23.7 ± 3.2 | 0.62 |
| Total cholesterol (mmol/L) | 5.0 ± 1.1 | 5.8 ± 1.1 | <0.01 | 5.0 ± 1.1 | 5.8 ± 1.1 | <0.01 |
| HDL (mmol/L) | 1.6 ± 0.5 | 1.4 ± 0.4 | 0.07 | 1.5 ± 0.5 | 1.5 ± 0.5 | 0.98 |
| Triglycerides (mmol/L) | 0.8 (0.6–1.2) | 1.2 (0.9–1.7) | <0.01 | 0.8 (0.6–1.2) | 1.3 (0.9–1.7) | <0.01 |
| Serum creatinine (µmol/L) | 90.1 ± 43.8 | 132.2 ± 93.9 | <0.01 | 88.3 ± 36.9 | 133.4 ± 97.4 | <0.01 |
| eGFR (mL/min/1.73 m2) | 85.7 ± 22.0 | 63.9 ± 27.8 | <0.01 | 86.2 ± 22.0 | 65.1 ± 27.2 | <0.01 |
| UAE (mg/24 h) | 18.0 (7.0–553.5) | 644.0 (72.5–1,834.5) | <0.01 | 16.0 (6.5–451.0) | 761.0 (103.0–2,021.0) | <0.01 |
| Systolic blood pressure (mmHg) | 136.8 ± 20.7 | 156.9 ± 23.2 | <0.01 | 135.2 ± 19.0 | 159.6 ± 23.8 | <0.01 |
| Diastolic blood pressure (mmHg) | 79.9 ± 12.5 | 83.9 ± 11.6 | <0.01 | 78.8 ± 11.3 | 87.0 ± 13.5 | <0.01 |
| Antihypertensive medication (%) | 32.7 | 65.8 | <0.01 | 29.9 | 70.4 | <0.01 |
| ACE inhibitors (%) | 20.6 | 50.7 | <0.01 | 19.5 | 50.6 | <0.01 |
| Smoker (%) | 47.2 | 49.2 | 0.57 | 44.8 | 56.8 | 0.06 |
| MGO (nmol/L) | 661.3 ± 186.2 | 754.0 ± 230.4 | <0.01 | 662.3 ± 180.8 | 738.5 ± 243.6 | <0.01 |
| GO (nmol/L) | 2,397.3 ± 801.2 | 2,686.2 ± 1,100.6 | 0.01 | 2,453.2 ± 895.2 | 2,476.4 ± 864.1 | 0.86 |
| 3-DG (nmol/L) | 4,253.5 ± 1,957.3 | 4,345.2 ± 2,216.8 | 0.73 | 4,235.1 ± 2,006.3 | 4,342.6 ± 2,089.2 | 0.73 |
| CML (µmol/L) | 3.5 (3.0–4.0) | 3.5 (2.8–4.1) | 0.45 | 3.5 (3.0–4.0) | 3.5 (2.7–4.0) | 0.22 |
| CEL (µmol/L) | 0.9 (0.8–1.0) | 1.0 (0.8–1.2) | <0.01 | 0.9 (0.8–1.0) | 1.0 (0.8–1.1) | <0.01 |
| Pentosidine (pmol/L) | 40.8 (34.0–49.1) | 52.0 (34.8–71.9) | <0.01 | 41.0 (34.2–48.9) | 51.9 (33.5–48.9) | <0.01 |
| Characteristic . | No CVD event (n = 248) . | CVD event (n = 73) . | P value . | Alive (n = 240) . | Dead (n = 81) . | P value . |
|---|---|---|---|---|---|---|
| Age (years) | 40.4 ± 9.6 | 45.1 ± 9.1 | <0.01 | 40.1 ± 9.1 | 45.5 ± 10.0 | <0.01 |
| Male sex (%) | 57.5 | 60.9 | 0.61 | 57.7 | 67.9 | 0.10 |
| Duration of diabetes (years) | 26.0 (21.0–32.0) | 29.0 (24.5–37.6) | <0.01 | 26.0 (21.0–32.0) | 28.0 (23.0–38.5) | <0.01 |
| HbA1c (%) | 8.0 ± 1.0 | 8.6 ± 0.9 | <0.01 | 8.1 ± 1.2 | 8.6 ± 0.8 | <0.01 |
| HbA1c (mmol/mol) | 64 ± 10.9 | 70 ± 9.8 | <0.01 | 65 ± 13.1 | 70 ± 8.7 | <0.01 |
| DN (%) | 42.7 | 76.7 | <0.01 | 40.8 | 79.0 | <0.01 |
| Retinopathy (%) | <0.01 | <0.01 | ||||
| None | 19.8 | 9.6 | 21.6 | 6.2 | ||
| Simple | 45.2 | 34.2 | 44.4 | 37.0 | ||
| Proliferative | 35.1 | 56.2 | 34.0 | 56.8 | ||
| BMI (kg/m2) | 23.9 ± 2.8 | 23.9 ± 3.5 | 1.00 | 23.9 ± 2.9 | 23.7 ± 3.2 | 0.62 |
| Total cholesterol (mmol/L) | 5.0 ± 1.1 | 5.8 ± 1.1 | <0.01 | 5.0 ± 1.1 | 5.8 ± 1.1 | <0.01 |
| HDL (mmol/L) | 1.6 ± 0.5 | 1.4 ± 0.4 | 0.07 | 1.5 ± 0.5 | 1.5 ± 0.5 | 0.98 |
| Triglycerides (mmol/L) | 0.8 (0.6–1.2) | 1.2 (0.9–1.7) | <0.01 | 0.8 (0.6–1.2) | 1.3 (0.9–1.7) | <0.01 |
| Serum creatinine (µmol/L) | 90.1 ± 43.8 | 132.2 ± 93.9 | <0.01 | 88.3 ± 36.9 | 133.4 ± 97.4 | <0.01 |
| eGFR (mL/min/1.73 m2) | 85.7 ± 22.0 | 63.9 ± 27.8 | <0.01 | 86.2 ± 22.0 | 65.1 ± 27.2 | <0.01 |
| UAE (mg/24 h) | 18.0 (7.0–553.5) | 644.0 (72.5–1,834.5) | <0.01 | 16.0 (6.5–451.0) | 761.0 (103.0–2,021.0) | <0.01 |
| Systolic blood pressure (mmHg) | 136.8 ± 20.7 | 156.9 ± 23.2 | <0.01 | 135.2 ± 19.0 | 159.6 ± 23.8 | <0.01 |
| Diastolic blood pressure (mmHg) | 79.9 ± 12.5 | 83.9 ± 11.6 | <0.01 | 78.8 ± 11.3 | 87.0 ± 13.5 | <0.01 |
| Antihypertensive medication (%) | 32.7 | 65.8 | <0.01 | 29.9 | 70.4 | <0.01 |
| ACE inhibitors (%) | 20.6 | 50.7 | <0.01 | 19.5 | 50.6 | <0.01 |
| Smoker (%) | 47.2 | 49.2 | 0.57 | 44.8 | 56.8 | 0.06 |
| MGO (nmol/L) | 661.3 ± 186.2 | 754.0 ± 230.4 | <0.01 | 662.3 ± 180.8 | 738.5 ± 243.6 | <0.01 |
| GO (nmol/L) | 2,397.3 ± 801.2 | 2,686.2 ± 1,100.6 | 0.01 | 2,453.2 ± 895.2 | 2,476.4 ± 864.1 | 0.86 |
| 3-DG (nmol/L) | 4,253.5 ± 1,957.3 | 4,345.2 ± 2,216.8 | 0.73 | 4,235.1 ± 2,006.3 | 4,342.6 ± 2,089.2 | 0.73 |
| CML (µmol/L) | 3.5 (3.0–4.0) | 3.5 (2.8–4.1) | 0.45 | 3.5 (3.0–4.0) | 3.5 (2.7–4.0) | 0.22 |
| CEL (µmol/L) | 0.9 (0.8–1.0) | 1.0 (0.8–1.2) | <0.01 | 0.9 (0.8–1.0) | 1.0 (0.8–1.1) | <0.01 |
| Pentosidine (pmol/L) | 40.8 (34.0–49.1) | 52.0 (34.8–71.9) | <0.01 | 41.0 (34.2–48.9) | 51.9 (33.5–48.9) | <0.01 |
Data are mean ± SD or median (interquartile range) unless otherwise indicated. Differences were tested with a χ2 or Student t test, as appropriate.
Associations of MGO With Markers of DN, ED, and LGI Scores
Higher plasma levels of MGO were associated with lower eGFR (model 1) (Table 2), which remained borderline significant after additional adjustment (model 2). We observed a trend between higher plasma MGO levels and higher UAE levels. Moreover, higher MGO levels were associated with significantly higher ED and LGI scores. We observed similar but weaker associations for GO levels, whereas 3-DG levels were not associated with markers of DN, ED, or LGI scores.
Multivariable standardized β-values (95% CIs) for the cross-sectional associations among plasma dicarbonyls plasma creatinine clearance, ln-UAE, ED, and LGI scores
| Model . | eGFR . | ln-UAE . | ED . | LGI . |
|---|---|---|---|---|
| MGO | ||||
| 1 | −0.11 (−0.21 to −0.02) | −0.01 (−0.06 to 0.04) | 0.14 (0.03 to 0.25) | 0.17 (0.06 to 0.27) |
| 2 | −0.08 (−0.18 to 0.01) | −0.04 (−0.08 to 0.01) | 0.17 (0.06 to 0.28) | 0.22 (0.11 to 0.33) |
| GO | ||||
| 1 | 0.02 (−0.07 to 0.11) | −0.02 (−0.06 to 0.03) | −0.01 (−0.11 to 0.10) | 0.07 (−0.04 to 0.17) |
| 2 | 0.05 (−0.05 to 0.14) | −0.04 (−0.08 to 0.00) | 0.04 (−0.07 to 0.15) | 0.13 (0.02 to 0.23) |
| 3-DG | ||||
| 1 | 0.03 (−0.06 to 0.12) | −0.04 (−0.09 to 0.00) | −0.04 (−0.15 to 0.07) | −0.01 (−0.12 to 0.10) |
| 2 | 0.04 (−0.05 to 0.13) | −0.03 (−0.08 to 0.01) | −0.01 (−0.11 to 0.10) | 0.03 (−0.07 to 0.14) |
| Model . | eGFR . | ln-UAE . | ED . | LGI . |
|---|---|---|---|---|
| MGO | ||||
| 1 | −0.11 (−0.21 to −0.02) | −0.01 (−0.06 to 0.04) | 0.14 (0.03 to 0.25) | 0.17 (0.06 to 0.27) |
| 2 | −0.08 (−0.18 to 0.01) | −0.04 (−0.08 to 0.01) | 0.17 (0.06 to 0.28) | 0.22 (0.11 to 0.33) |
| GO | ||||
| 1 | 0.02 (−0.07 to 0.11) | −0.02 (−0.06 to 0.03) | −0.01 (−0.11 to 0.10) | 0.07 (−0.04 to 0.17) |
| 2 | 0.05 (−0.05 to 0.14) | −0.04 (−0.08 to 0.00) | 0.04 (−0.07 to 0.15) | 0.13 (0.02 to 0.23) |
| 3-DG | ||||
| 1 | 0.03 (−0.06 to 0.12) | −0.04 (−0.09 to 0.00) | −0.04 (−0.15 to 0.07) | −0.01 (−0.12 to 0.10) |
| 2 | 0.04 (−0.05 to 0.13) | −0.03 (−0.08 to 0.01) | −0.01 (−0.11 to 0.10) | 0.03 (−0.07 to 0.14) |
Data were analyzed by using linear regression analyses. Regression coefficients are expressed as a 1 SD difference in eGFR, ln-UAE, ED, or LGI score per SD increase of plasma dicarbonyl. Model 1: adjusted for age, sex, diabetes duration, HbA1c, and presence of DN. Model 2: model 1 + adjustment for total cholesterol, current smoking, and systolic blood pressure and blood pressure–lowering treatment.
Associations of MGO With Total, Fatal, and Nonfatal Incident CVD
After adjustment for sex, age, duration of diabetes, HbA1c, and presence of DN, higher plasma MGO levels were associated with total, fatal, and nonfatal CVD as well as with total mortality (model 1) (Table 3). Additional adjustment for cardiovascular risk factors did not change the point estimates of these associations in a major way (model 2). Associations between MGO levels and incident CVD and total mortality did not differ between men and women or for individuals with and without DN (all Pinteraction > 0.1). All associations between plasma MGO levels and incident total, fatal, and nonfatal incident CVD remained largely unchanged overall after adjustment for markers of DN and ED scores (models 3 and 4). Overall, the associations attenuated slightly after we adjusted for LGI scores (model 2 vs. 5).
Multivariable adjusted HRs (95% CIs) for the associations among plasma dicarbonyls, incident cardiovascular outcomes, and mortality
| Model . | Total CVD (n = 73) . | Fatal CVD (n = 36) . | Nonfatal CVD (n = 53) . | Total mortality (n = 81) . |
|---|---|---|---|---|
| MGO | ||||
| 1 | 1.51 (1.19–1.93) | 1.37 (1.00–1.87) | 1.54 (1.17–2.04) | 1.25 (1.01–1.54) |
| 2 | 1.47 (1.13–1.91) | 1.42 (1.01–1.99) | 1.46 (1.08–1.98) | 1.24 (0.99–1.55) |
| 3 | 1.49 (1.14–1.93) | 1.49 (1.04–2.15) | 1.53 (1.13–2.08) | 1.26 (0.98–1.60) |
| 4 | 1.47 (1.13–1.90) | 1.41 (1.02–1.97) | 1.46 (1.08–1.97) | 1.24 (0.99–1.55) |
| 5 | 1.41 (1.08–1.83) | 1.36 (0.96–1.92) | 1.40 (1.04–1.90) | 1.18 (0.94–1.48) |
| GO | ||||
| 1 | 1.34 (1.08–1.67) | 1.16 (0.88–1.54) | 1.40 (1.09–1.79) | 0.96 (0.78–1.19) |
| 2 | 1.36 (1.07–1.73) | 1.29 (0.95–1.74) | 1.35 (1.04–1.76) | 1.00 (0.80–1.24) |
| 3 | 1.47 (1.16–1.87) | 1.35 (0.99–1.86) | 1.49 (1.15–1.95) | 1.03 (0.82–1.29) |
| 4 | 1.37 (1.08–1.73) | 1.30 (0.97–1.76) | 1.36 (1.04–1.77) | 1.01 (0.81–1.26) |
| 5 | 1.32 (1.05–1.67) | 1.24 (0.92–1.67) | 1.32 (1.02–1.71) | 0.97 (0.78–1.20) |
| 3-DG | ||||
| 1 | 0.98 (0.76–1.25) | 1.20 (0.85–1.69) | 0.79 (0.59–1.06) | 1.01 (0.81–1.26) |
| 2 | 1.08 (0.84–1.38) | 1.37 (0.97–1.93) | 0.88 (0.65–1.18) | 1.16 (0.93–1.45) |
| 3 | 1.13 (0.87–1.45) | 1.39 (0.98–1.98) | 0.92 (0.68–1.24) | 1.19 (0.95–1.49) |
| 4 | 1.08 (0.84–1.39) | 1.39 (0.98–1.95) | 0.88 (0.65–1.18) | 1.18 (0.94–1.47) |
| 5 | 1.08 (0.84–1.37) | 1.34 (0.96–1.88) | 0.88 (0.66–1.18) | 1.15 (0.92–1.42) |
| Model . | Total CVD (n = 73) . | Fatal CVD (n = 36) . | Nonfatal CVD (n = 53) . | Total mortality (n = 81) . |
|---|---|---|---|---|
| MGO | ||||
| 1 | 1.51 (1.19–1.93) | 1.37 (1.00–1.87) | 1.54 (1.17–2.04) | 1.25 (1.01–1.54) |
| 2 | 1.47 (1.13–1.91) | 1.42 (1.01–1.99) | 1.46 (1.08–1.98) | 1.24 (0.99–1.55) |
| 3 | 1.49 (1.14–1.93) | 1.49 (1.04–2.15) | 1.53 (1.13–2.08) | 1.26 (0.98–1.60) |
| 4 | 1.47 (1.13–1.90) | 1.41 (1.02–1.97) | 1.46 (1.08–1.97) | 1.24 (0.99–1.55) |
| 5 | 1.41 (1.08–1.83) | 1.36 (0.96–1.92) | 1.40 (1.04–1.90) | 1.18 (0.94–1.48) |
| GO | ||||
| 1 | 1.34 (1.08–1.67) | 1.16 (0.88–1.54) | 1.40 (1.09–1.79) | 0.96 (0.78–1.19) |
| 2 | 1.36 (1.07–1.73) | 1.29 (0.95–1.74) | 1.35 (1.04–1.76) | 1.00 (0.80–1.24) |
| 3 | 1.47 (1.16–1.87) | 1.35 (0.99–1.86) | 1.49 (1.15–1.95) | 1.03 (0.82–1.29) |
| 4 | 1.37 (1.08–1.73) | 1.30 (0.97–1.76) | 1.36 (1.04–1.77) | 1.01 (0.81–1.26) |
| 5 | 1.32 (1.05–1.67) | 1.24 (0.92–1.67) | 1.32 (1.02–1.71) | 0.97 (0.78–1.20) |
| 3-DG | ||||
| 1 | 0.98 (0.76–1.25) | 1.20 (0.85–1.69) | 0.79 (0.59–1.06) | 1.01 (0.81–1.26) |
| 2 | 1.08 (0.84–1.38) | 1.37 (0.97–1.93) | 0.88 (0.65–1.18) | 1.16 (0.93–1.45) |
| 3 | 1.13 (0.87–1.45) | 1.39 (0.98–1.98) | 0.92 (0.68–1.24) | 1.19 (0.95–1.49) |
| 4 | 1.08 (0.84–1.39) | 1.39 (0.98–1.95) | 0.88 (0.65–1.18) | 1.18 (0.94–1.47) |
| 5 | 1.08 (0.84–1.37) | 1.34 (0.96–1.88) | 0.88 (0.66–1.18) | 1.15 (0.92–1.42) |
Data were analyzed by using Cox proportional hazard regression analyses. HR is expressed per SD increase of plasma dicarbonyl. Model 1: adjusted for age, sex, diabetes duration, HbA1c, and presence of DN. Model 2: model 1 + adjustment for total cholesterol, BMI, current smoking, and systolic blood pressure and blood pressure–lowering treatment. Model 3: model 2 + eGFR and ln-UAE. Model 4: model 2 + ED (soluble intracellular adhesion molecule [sICAM] + soluble vascular cell adhesion molecule). Model 5: model 2 + LGI (sICAM + ln-C-reactive protein + ln-interleukin-6 + ln-secretory phospholipase A2).
Although glucose is a major precursor for MGO, additional adjustment of model 2 for fasting plasma glucose did not weaken the association between MGO and total CVD (hazard ratio [HR] 1.59 [95% CI 1.18–2.15]). Similarly, the omission of HbA1c from model 2 did not materially change the association between plasma MGO levels and total CVD (HR 1.48 [95% CI 1.14–1.92]).
Associations of GO With Total, Fatal, and Nonfatal Incident CVD
After adjustment for sex, age, duration of diabetes, HbA1c, and presence of DN, higher GO levels were associated with total and nonfatal CVD (model 1) (Table 3). We observed a borderline significant association with fatal CVD (model 1). Additional adjustment for cardiovascular risk factors did not majorly change the associations (model 2). We did not observe any associations between GO levels and total mortality (models 1–5). Associations between GO levels and incident CVD and total mortality did not differ between men and women or for individuals with and without DN (all Pinteraction > 0.1). Adjustment for markers of DN seemed to strengthen associations between GO and total and nonfatal CVD (model 3). All associations between plasma GO levels and incident total, fatal, and nonfatal incident CVD remained similar overall after adjustment for ED or LGI scores (models 3–5).
Associations of 3-DG With Total, Fatal, and Nonfatal Incident CVD
Either in crude analyses (Table 1) or after adjustment for potential confounding (Table 3, models 1 and 2) or potential mediating (Table 3, models 3–5) factors, we did not observe any significant associations between plasma 3-DG levels and total and nonfatal incident CVD or total mortality. We did observe a nonsignificant trend between 3-DG levels and fatal CVD.
Discussion
The main finding of this study was that plasma MGO levels are associated with incident total, fatal, and nonfatal CVD in individuals with type 1 diabetes. We observed similar associations between GO levels and total and nonfatal CVD. Although plasma MGO levels were associated with plasma markers of DN, ED, and LGI, these markers did not seem to mediate large parts of the associations between plasma MGO levels and incident CVD. The associations between plasma MGO levels and markers of DN are in line with a previous report in individuals with type 2 diabetes (14). Many studies have implicated MGO and AGEs in the development of DN (6,7,15,16). Indeed, in an animal model of type 1 diabetes, we found that a reduction of MGO prevents vascular LGI, ED, DN, and retinopathy (15,17). MGO may hamper function of the vasculature through several mechanisms, such as impairment of vasodilation (6), inhibition of the proteasome (18), and generation of oxidative stress (6).
Moreover, this study is the first to our knowledge that shows an association between plasma MGO levels and incident CVD. One previous study has linked MGO levels with changes in carotid intima-medial thickness in type 2 diabetes (19). The current study also is in line with previous findings demonstrating that MGO-derived AGEs are associated with incident CVD (9,10) and rupture-prone plaques (8,20). Similarly, skin levels of AGEs have been linked to macrovascular disease in the Diabetes Control and Complications Trial (DCCT) (21). The current finding that MGO is associated with incident CVD is in line with AGE intervention studies performed in rodents. Aminoguanidine, an inhibitor of MGO and glycation, reduced atherosclerosis in an experimental model of diabetes (22). Similarly, the AGE-breaker alagebrium, which has MGO-quenching properties, reduces atherosclerotic burden (22). Other MGO/AGE-quenching compounds, such as pyridoxamine, reduce atherosclerotic burden in mice and should be evaluated in a human study on cardiovascular outcomes (23). Whether the trans-resveratrol and hesperetin coformulation, which acts as a glyoxalase inducer, reduces vascular disease by inhibiting MGO should be determined (24).
In addition to the association between plasma MGO levels and incident CVD, this study reveals that plasma GO levels display similar associations with incident CVD as MGO. We found that adjustment for LGI scores slightly attenuated the associations between MGO and incident CVD. Therefore, MGO may lead to CVD at least partly through LGI. Whether drugs with anti-inflammatory properties mitigate detrimental effects of MGO is unknown. The fact that DN and ED scores did not mediate the association between MGO and CVD should be interpreted with care because we only assessed a limited set of biomarkers.
Because MGO fluctuates with plasma glucose levels (25), novel glucose-lowering compounds may be effective treatments to target dicarbonyl stress. Furthermore, MGO has been shown to increase after a mixed meal in diabetes (25) and that obesity and dyslipidemia may also influence MGO levels. Indeed, we identified caloric restriction as a method to greatly reduce MGO levels in obese individuals with type 2 diabetes (25).
A major strength of this study is that we used a relatively large sample of individuals with type 1 diabetes with and without microvascular complications. We measured MGO and other markers of dicarbonyl stress with state-of-the-art techniques. However, this study also has several limitations. It did not address where plasma dicarbonyl levels originate. We hypothesize that erythrocytes, endothelial cells, and perhaps plaque macrophages are major sources. Furthermore, the current study was completed in 2006; therefore, we could not investigate whether adjustment for risk modifying drugs, such as statins, at baseline influenced the results. We also cannot exclude that we underestimated the associations because of prolonged storage times of plasma samples and changes in medication use during the follow-up period. Nonetheless, the current study provides an important proof of principle that confirms a large body of experimental evidence identifying MGO as a key player in diabetic complications and CVD.
In conclusion, the current study underlines the importance of MGO in type 1 diabetic CVD in humans. Future studies will address whether these findings also hold true for individuals with type 2 diabetes. Furthermore, this study underlines the potential interest of identified inhibitors of MGO formation in curtailing diabetic complications.
Article Information
Funding. This research was supported by the European Foundation for the Study of Diabetes (EFSD/Lilly program 2016).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. N.M.J.H. analyzed data and wrote the manuscript. J.L.J.M.S. measured plasma dicarbonyls and edited the manuscript. A.J. and H.-H.P. collected follow-up data on patients and edited the manuscript. H.-H.P., L.T., and P.R. designed the study and edited the manuscript. C.D.A.S. wrote and edited the manuscript. C.G.S. was the principal investigator and wrote and edited the manuscript. C.G.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
