Potential Predictive Role of Lipid Peroxidation Markers for Type 2 Diabetes in the Adult Tunisian Population

ARTICLE IN PRESS Can J Diabetes xxx (2017) 1–9

Contents lists available at ScienceDirect

Canadian Journal of Diabetes journal homepage: w w w. c a n a d i a n j o u r n a l o f d i a b e t e s . c o m

Original Research

Potential Predictive Role of Lipid Peroxidation Markers for Type 2 Diabetes in the Tunisian Population Houda Bouhajja PhD a,*, Faten Hadj Kacem MD a, Rania Abdelhedi PhD b, Marwa Ncir PhD c, Jordan D. Dimitrov PhD d,e, Rim Marrakchi MD f, Kamel Jamoussi MD f, Ahmed Rebai PhD b, Abdelfattah El Feki PhD c, Mohamed Abid MD a, Hammadi Ayadi PhD b, Srini V. Kaveri PhD d,e, Mouna Mnif-Feki MD a, Noura Bougacha-Elleuch PhD g a

Unit of Obesity and Metabolic Syndrome, Department of Endocrinology, Hedi Chaker Hospital, Sfax, Tunisia Laboratory of Molecular and Cellular Screening Processes, Centre of Biotechnology of Sfax, Tunisia c Animal Eco-Physiology Laboratory, Faculty of Sciences of Sfax, Sfax, Tunisia d Centre de Recherche des Cordeliers, Sorbonne Universités, Paris, France e INSERM, Université Paris Descartes, Paris, France f Biochemistry Laboratory, CHU Hedi Chaker, Sfax, Tunisia g Laboratory of Molecular and Functional Genetics, Faculty of Sciences of Sfax, University of Sfax, Sfax, Tunisia b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 April 2017 Received in revised form 2 June 2017 Accepted 15 June 2017

Objectives: We evaluated the potential clinical relevance of malondialdehyde (MDA) and autoantibodies to copper oxidized low-density lipoprotein (CuOx-LDL) in type 2 diabetes occurrence. Methods: This cross-sectional study enrolled 69 normoglycemic subjects, 18 prediabetic patients and 108 type 2 diabetes patients. MDA concentration was assessed spectrophotometrically. Plasma IgG, IgA and IgM levels to CuOx-LDL were determined by ELISA. Results: Plasma MDA levels were considerably higher in obese, prediabetic and type 2 diabetes subjects compared to controls. In multiple linear regression analysis, both MDA and IgA to CuOx-LDL were significantly associated with glucose metabolism markers (p<0.05). Multiple logistic regression analyses showed that high plasma MDA and IgA to CuOx-LDL were independent risk factors for type 2 diabetes (OR 1.196, 95% CI: 1.058 to 1.353; p=0.004; OR 1.626, 95% CI: 1.066 to 2.481; p=0.024; respectively). Importantly, elevated IgA to CuOx-LDL predicted incident diabetes in patients with prediabetes (OR 2.321, 95% CI:1.063 to 5.066; p=0.035). From stratified analyses by body mass index (BMI), both MDA and IgA to CuOx-LDL remained independent predictors of type 2 diabetes occurrence in non-obese subjects (p<0.05). More interesting, elevated IgA to CuOx-LDL levels could be predictors of type 2 diabetes in obese prediabetic subjects (p=0.044). Conversely, neither IgG nor IgM to CuOx-LDL was associated with glucose metabolism markers, obesity or type 2 diabetes. Conclusions: Plasma MDA and IgA to CuOx-LDL were significantly associated with blood markers of glucose metabolism. High levels of MDA and IgA to CuOx-LDL could independently predict type 2 diabetes development in normoglycemia and prediabetic subjects. © 2017 Canadian Diabetes Association.

Keywords: anti Ox-LDL antibodies insulin resistance malondialdehyde obesity oxidative stress type 2 diabetes

r é s u m é Mots clés : anticorps anti-LDL oxydées insulinorésistance malondialdéhyde obésité stress oxydatif diabète de type 2

Objectifs : Nous avons évalué la pertinence clinique potentielle du malondialdéhyde (MDA) et des autoanticorps aux lipoprotéines de faible densité oxydées par le cuivre (CuOx-LDL) dans la survenue du diabète de type 2. Méthodes : La présente étude transversale comptait 69 sujets non diabétiques, 18 patients prédiabétiques et 108 patients atteints du diabète de type 2. La spectrophotométrie a permis de mesurer la concentration de MDA. La technique ELISA a permis de déterminer les concentrations plasmatiques d’IgG, d’IgA et IgM aux CuOx-LDL.

* Address for correspondence: Houda Bouhajja, PhD, Unit of Obesity and Metabolic Syndrome, Department of Endocrinology, University Hospital Hedi Chaker, El Ain Road, Sfax 3029, Tunisia. E-mail address: [email protected] 1499-2671 © 2017 Canadian Diabetes Association. The Canadian Diabetes Association is the registered owner of the name Diabetes Canada. http://dx.doi.org/10.1016/j.jcjd.2017.06.006

ARTICLE IN PRESS 2

H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

Résultats : Les concentrations plasmatiques de MDA étaient considérablement plus élevées chez les sujets souffrant d’obésité, de prédiabète ou de diabète de type 2 que chez les témoins. À l’analyse de régression linéaire multiple, le MDA et l’IgA aux CuOx-LDL étaient associés de manière significative aux marqueurs du métabolisme du glucose (p<0,05). Les analyses de régression logistique multiple montraient que les concentrations plasmatiques de MDA et d’IgA aux CuOx-LDL étaient des facteurs de risque indépendants du diabète de type 2 (RIA 1,196, IC à 95 % : 1,058 à 1,353; p=0,004; RIA 1,626, IC à 95 % : 1,066 à 2,481; p=0,024; respectivement). Notamment, des concentrations élevées d’IgA aux CuOx-LDL prédisaient l’incidence du diabète chez les patients prédiabétiques (RIA 2,321, IC à 95 % : 1,063 à 5,066; p=0,035). À partir des analyses stratifiées selon l’indice de masse corporelle (IMC), le MDA et l’IgA aux CuOx-LDL demeuraient des prédicteurs indépendants de la survenue du diabète de type 2 chez les sujets ayant un IMC<25 kg2 (p=0,003 et p=0,039, respectivement). Plus intéressant encore, des concentrations élevées d’IgA aux CuOx-LDL seraient des prédicteurs du diabète de type 2 chez les sujets prédiabétiques qui avaient un IMC≥25 kg2 (p=0,044). À l’inverse, ni l’IgG ni l’IgM aux CuOx-LDL n’était associé aux marqueurs du métabolisme du glucose, de l’obésité ou du diabète de type 2. Conclusions : Les concentrations plasmatiques de MDA et d’IgA aux CuOx-LDL étaient associées de manière significative aux marqueurs sanguins du métabolisme du glucose. Des concentrations plasmatiques élevées de MDA et d’IgA aux CuOx-LDL seraient respectivement des prédicteurs indépendants du développement du diabète de type 2 chez les sujets ayant une normoglycémie et chez les sujets ayant un prédiabète. © 2017 Canadian Diabetes Association.

Introduction The prevalence of type 2 diabetes is growing, so extensive attempts have been made to identify reliable and sensitive biomarkers for the early prediction and diagnosis of type 2 diabetes. Several studies have documented the predictive ability of the anthropometric measures, metabolic and inflammatory markers in discriminating risk for type 2 diabetes (1–3). However, data evaluating the potential clinical relevance of oxidative stress markers in the occurrence of diabetes are lacking. Elevated lipid peroxidation is well documented in subjects with metabolic syndrome (4), type 2 diabetes (5), impaired glucose tolerance (6) and obesity (7). Lipid oxidation leads to the formation of highly reactive aldehydes with several deleterious effects (8). Malondialdehyde (MDA) is among the most commonly used biomarkers of in vivo peroxidation of polyunsaturated fatty acids. Several previous studies have well documented increased MDA level in subjects with obesity (7) and type 2 diabetes (9,10). A recent data reported the deleterious signalling effect of high MDA level on insulin secretion (11). However, no data has evaluated the MDA value as an independent biomarker of type 2 diabetes. Oxidative LDL modification has been also shown to be highly immunogenic, leading to OxLDL antibodies formation (12), both in experimental animals and humans (13).These antibodies have shown to predict progression of atherosclerosis, myocardial infarction and coronary artery disease (14,15). However, the predictive role of circulating anti-OxLDL antibodies for type 2 diabetes risk has been less explored. Although IgG anti-OxLDL antibodies are the predominant isotype in humans (16), investigations of its circulating level for their protective or pathogenic role in type 2 diabetes are still controversial (17–20). In addition, very few studies assessed the relationship between plasma IgA to OxLDL and diabetes in humans, where an evidence on positive association of IgA to CuOx-LDL and IgA to malondialdehyde acetaldehyde (MAA)-LDL with type 2 diabetes were reported (17,21). Similarly, data are scarce on the association between circulating IgM antibodies to OxLDL and type 2 diabetes (17). On the other hand, the association of these antibodies with obesity has also been poorly evaluated, with clearly contradictory findings (20–23). Despite the well documented role of oxidative stress in the pathophysiology of diabetes and its complications, data evaluating the status of oxidative stress in patients at clinical level are scarce. Lipid peroxidation markers have not been regarded as traditional risk factors for type 2 diabetes. Here, we conducted a cross-sectional study to i) evaluate possible association of plasma MDA and OxLDL antibodies levels with obesity and blood markers of glucose

metabolism, ii) to investigate the clinical relevance of these lipid peroxidation markers as potential predictive biomarkers to discriminate subjects at higher risk for type 2 diabetes.

Methods Study population The study population comprised a sample of 21- to 84-yearold participants (n=195) enrolled in the department of Endocrinology at the Hospital Hedi Chaker (Sfax, Tunisia) in 2012–2014. The study population included 3 groups: group 1 comprised normoglycemic subjects (n=69), group 2 included subjects with prediabetes (n=18) and group 3 comprised type 2 diabetes patients (n=108).The World Health Organization (WHO) criteria were used to diagnose prediabetes and type 2 diabetes as previously detailed (24). The diagnosis of obesity was made according to the WHO standard recommended method (25,26). We excluded participants with cardiovascular diseases, liver and chronic kidney pathologies, thyroid diseases, malignancy, pregnancy, glucocorticoid therapy, acute or chronic inflammatory or infectious disease. The study was approved by the local ethics committee of the Hospital Hedi Chaker of Sfax and all patients were informed. Anthropometric and biochemical characterization All participants underwent anthropometric and biochemical characterization. In brief, clinical characteristics including body mass index (BMI), waist circumference (WC), waist to hip ratio and body fat content were recorded for each subject. Blood samples were collected from fasted subjects for the biochemical characterization. Plasma blood glucose, total cholesterol, high density lipoprotein cholesterol (HDL-C), triglycerides (TG), uric acid, creatinine and hepatic enzymes activities were measured using standard clinical methods and an automated chemistry analyzer (ADVIA Siemens 1800 system). Glycated hemoglobin (A1C) was measured using high performance liquid chromatography (HPLC) (Biorad D10). High sensitivity C-reactive protein (hs-CRP) was estimated by nephleometric assay. ELISA method was used to determine both fasting plasma leptin and insulin concentrations according to the commercial suppliers (cat no.: KAP2281 and cat no.: KAP1251, respectively, DIA source Europe S.A.; Nivelles, Belgium). The HOMA-IR index was calculated as HOMA-IR = FPG (mmol l) × FINS ( μIU ml) 22.5 . Insulin sensitivity was evaluated by quantitative insulin sensitivity check index (QUICKI) as follow:

ARTICLE IN PRESS H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

QUICK-index =[1 log FINS ( μIU ml) + log FPG (mg dl)] . HOMA β-cell function was estimated according to the following equation HOMAβ-cell function = 20 × FINS ( μIU ml ) ÷[FPG (mmol l) − 3.5] . LDL isolation Blood was drawn from three fasting normolipidemic blood donor volunteers (aged 24–33 years) into EDTA-containing Vacutainer tubes, and pooled EDTA-plasma was obtained after centrifugation at 4500 rpm, 4°C for 20 min. Native human LDL (nLDL: 1.019
3

subtracting the binding of native lipoproteins from the binding of Cu2+-lipoproteins. Determination of MDA concentration MDA concentrations were measured as an index of lipid peroxidation using the colorimetric method (28). Briefly, aliquots of plasma samples were mixed with 10% trichloroacetic acid (TCA) and centrifuged at 3000 rpm for 10 min. Thiobarbituric acid (TBA, 0.67%) was added to supernatant and the reaction mixture was heated at 95°C for 15 min. The mixture was then cooled and absorbance was measured at 532 nm. The MDA concentration was calculated using 1,1,3,3-tetramethoxypropane as standard under the same reaction conditions. The results were expressed as nmol MDA/ml plasma. Statistical analysis Statistical analyses were carried out with SPSS software (version 17.0, Chicago, Illinois, United States) and GraphPad Prism software (v. 6). The Shapiro-Wilk test was used to test the normality of distribution, and consequently variables with a skewed distribution were log transformed. Continuous variables were expressed as mean ± standard deviation or median with interquartile ranges. Categorical variables were presented as number of participants and percentage and are compared by Pearson chi-square tests. Unpaired student t-test, one way analysis of variance (ANOVA) and oneway analysis of covariance (ANCOVA) with Tukey’s corrections were used to compare means for continuous variables when appropriate. Two-way analysis of variance was used when the analyses included two grouping variables (3D graphs). Levene’s test was used to test the null hypothesis of the equality of variances. Pearson’s correlation analysis was performed to examine the relationship between markers of oxidative stress and metabolic parameters. A stepwise forward linear regression analysis was performed to evaluate the independent association of oxidative stress markers with diabetes-related traits. Using multivariate logistic regression analysis with forward conditional inclusion procedure, we calculated odds ratio (ORs) to evaluate the ability of oxidative stress markers levels to predict the occurrence of diabetes. A p value <0.05 was considered statistically significant.

Plasma anti CuOx-LDL antibodies measurement Plasma IgG, IgA and IgM titers to CuOx-LDL were determined by enzyme linked immunosorbent assay (ELISA) as described below: Maxisorp plates (Nunc, Rochester, New York, United States) were coated with either native or oxidized lipoproteins, both at 5μgml−1 in PBS, overnight at 4°C. To prevent oxidation of native LDL, 0.27 mmol/L EDTA and 20 μmol/L BHT were added to PBS. Each well was washed three times with PBS containing 0.05% Tween 20. Plates were then blocked with 2% BSA (Sigma, St. Louis, Missouri, United States) in PBS containing 0.27 mmol/l EDTA and 20 μmol/l BHT for 2h at 4°C. Plates were washed as above. Plasma samples were diluted 1:100 in 0.05% Tween-PBS and incubated for 2h at room temperature. After washing 5 times, a Horseradish peroxidase-conjugated anti human IgG (1:3000), IgA (1:4000) or IgM (1:4000) were added to the wells in PBS containing 0.05% Tween20, 0.27 mmol/l EDTA and 20 μmol/l BHT. Plates were incubated for 1h30 at room temperature followed by a wash step to remove unbound conjugate. O-phenylene-diamine (OPD) was added to each well, and plates were incubated for 5 min in dark. The reaction was then stopped by adding 2N H2SO4 and the absorbance was measured at 492 nm with a microplate reader. The results were expressed as an optical density (OD). The binding of antibodies to oxidized lipoproteins was calculated by

Results Baseline characteristics of the study population Baseline characteristics of the study participants according to their glucose status and body weight are summarized in Table 1. Irrespective of diabetes status, the obese groups displayed significantly increased levels of fasting insulin, HOMA-IR and HOMA β-cell function but markedly lower Quick-index compared with nonobese groups (p<0.001). Plasma levels of glucose and A1C were higher in the obese group, especially in patients with diabetes (p=0.031 and p<0.001, respectively). With regard to obese groups, we observed a gradual significant increase in levels of glucose metabolism parameters with progressing from non-diabetes to prediabetes and later to diabetes, with the group of diabetes showing the highest concentrations of glucose (p<0.001), A1C (p<0.001), HOMA-IR (p<0.001) but the lowest Quickindex (p<0.001) and HOMA β-cell function (p<0.001). No difference was observed in level of insulin among the groups (p>0.05). Considering obesity parameters, the subjects with prediabetes were significantly abdominally obese, as indicated by WC, compared to diabetes (p=0.019) and diabetes-free groups (p=0.004). In term of

4

Table 1 Baseline characteristics of the study participants according to diabetes status and body weight (n=195) Non-diabetes Baseline characteristics

37 (21/16) 44.51±9.34*** 13 (18.8) 3 (4.6) 0 17 (27.9) 23.30 87.50 0.90 14.40

(18.70–25.40) (73–102)** (0.81–1.01)*** (6.40–23.80)

5.00 5.30 5.3 5.73 1.28 0.37 72.50

(4.40–6.00)*** (3.50–7.40) (4.4–6.0)*** (1.65–15.53)** (0.35–3.24)*** (0.32–0.46)*** (22.90–258.78)***

64.24 (39.00–91.88) 244.68 (100.67–444.56)** 19.0 16.0 11.0 55.0

(12.0–27.0)** (6.0–24.0) (3.0–48.0)** (30.0–112.0)***

0.68 (0.17–4.74)** 4.80 (0.24–26.86)

32 (6/26) 39.44±11.24 3 (4.3) 0 4 (6.2) 24 (39.3) 35.75 111.00 0.94 42.00

(29.90–45.50) (98.00–135.00) (0.81–1.80) (20.40–63.80)

4.72±0.77 1.23 (0.73–1.84) 2.86±0.69 1.32 (0.42–4.64) 5.25 5.89 5.4 16.95 3.90 0.31 204.80

(4.60–6.00) (3.10–7.70) (4.6–6.0) (10.08–33.09) (2.42–8.41) (0.28–0.33) (92.64–392.04)

54.62 (34.83–98.76) 273.03 (152.00–485.94) 19.5 18.0 16.5 63.0

(13.0–41.0) (11.0–43.0) (6.0–50.0) (44.0–97.0)

3.16 (0.34–8.65) 37.90 (3.50–82.31)

0.044 0.001 0.011 0.018 0.005 <0.001 <0.001 0.122 <0.001 0.041 0.912 0.182 <0.001 0.224 0.071 0.082 <0.001 <0.001 <0.001 <0.001 0.040 0.015 0.072 0.004 0.002 0.030 <0.001 <0.001

Prediabetes

Diabetes

Obese

Non-Obese

18 (6/12) 51.33±10.14 a 1 (5.6) 1 (5.6) 4 (22.2) 12 (66.7)

52 (30/22) 58.35±9.84 15 (14.2) 5 (4.8) 16 (15.4) 48 (45.3)

38.80 120.0 0.98 44.30

23.11 92.50 0.98 14.40

(30.30–52.00) (108.00–143.00) a (0.87–1.79) (27.70–73.40)

4.76±0.95 1.35 (0.88–2.03) 2.68±0.67 1.53 (0.68–3.45) 6.1 8.3 5.7 18.64 5.24 0.30 126.27

(5.0–6.9) (4.7–10.8) a (5.2–6.5) (11.92–45.68) (3.00–11.57) (0.27–0.32) (88.69–415.28)

64.41 (38.75–105.74) 340.96 (214.00–488.18) 21.00 21.00 18.00 80.50

(15.00–30.00) (7.00–45.00) (9.00–27.00) (40.00–128.00)

4.47 (0.85–9.83) 41.94 (11.83–67.95)

(17.80–25.70) (69.00–106.00) (0.81–1.14) (5.10–23.10)

4.33±0.92 1.07 (0.76–1.89) 2.60±0.72 1.23 (0.44–3.53) 11.45 NP 10.3 8.60 4.55 0.31 20.85

(5.10–18.80) (7.0–17.5) (1.26–39.71) (0.71–17.82) (0.26–0.41) (2.77–131.36)

62.26 (22.86–124.44) 217.22 (113.00–492.58) 16.0 15.0 14.0 79.0

(11.0–34.0) (7.0–58.0) (7.0–53.0) (43.0–135.0)

1.21 (0.17–7.55) 5.34 (0.53–16.29)

p¶

Obese 56 (18/38) 56.49±10.19 a 6 (5.7) 0 (0) 32 (30.8) 43 (40.6) 33.30 114.00 1.02 32.80

(29.38–46.20) b (99.00–143.00) b (0.82–2.04) (21.80–60.50) a,b

4.56±0.90 1.08 (0.60–2.12) 2.45±0.79 a 1.55 (0.81–4.65) a 9.30 NP 8.6 17.78 7.48 0.29 53.79

(5.30–18.60) a,b (5.7–15.8) a,b (1.60–44.40) (0.57–26.84) a (0.25–0.42) a (5.32–274.98) a,b

62.47 (33.40–195.16) 271.17 (106.76–681.43) 21.0 23.0 25.5 78.0

(12.0–36.0) (13.0–53.0) (8.0–78.0) a,b (34.0–130.0) a

3.19 (0.54–9.80) 31.24 (2.10–83.20)

0.341 0.011 0.023 0.005 0.360 <0.001 <0.001 0.028 <0.001 0.192 0.463 0.295 0.002 0.031 <0.001 <0.001 <0.001 <0.001 <0.001 0.268 <0.001 <0.001 <0.001 <0.001 0.809 <0.001 <0.001

Data are means ± SD, medians (interquartile ranges) for skewed variables, or numbers (percentages) for categorical variables. † ANOVA with Tukey’s test or unpaired student t-test when appropriate (for continuous variables); ‡ Chi-square test (for categorical variables). p§: p value when obese compared with non-obese in the group without diabetes; p¶: p value when obese compared with non-obese in the group with diabetes. a: p<0.05 when prediabetes and type 2 diabetes compared to non diabetes in obese group; b: p<0.05 when type 2 diabetes compared to prediabetes in obese group. *** p<0.001 when comparing non-obese subjects of groups with and without diabetes, ** p<0.05 when comparing non-obese subjects of groups with and without diabetes. 2hPG, 2h plasma glucose after oral glucose tolerance test; A1C, glycated hemoglobin; ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; BMI, body mass index; FINS, fasting plasma insulin; FPG, fasting plasma glucose; HDL-C, high density lipoprotein-cholesterol; HOMA β-Cell function, homeostasis model assessment of beta cell function; HOMA-IR, homeostasis model assessment of insulin resistance; hs-CRP, highly sensitive C-reactive protein; LDL C, low density lipoprotein-cholesterol; QUICK-index, quantitative insulin sensitivity check index; TC, total cholesterol; TG, triglycerides; WC, waist circumference; WHR, waist to hip ratio; γG, gamma-glutamyltransferase.

ARTICLE IN PRESS

4.30±0.91 1.30 (0.79–2.12)** 2.62±0.74 0.68 (0.18–2.76)***

p§

Obese

H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

Clinical parameters n (Male/Female) Age † (year) Current smoking ‡, n (%) Current drinking ‡, n (%) Hypertension ‡, n(%) Family history of diabetes ‡, n (%) Anthropometric parameters BMI † (kg/m2) Waist circumference † (cm) Waist to hip ratio † Body Fat † (kg) Lipidic parameters TC † (mmol/l) HDL-C † (mmol/l) LDL-C † (mmol/l) TG † (mmol/l) Glycemic parameters FPG † (mmol/l) 2hPG † (mmol/l) A1C † (%) FINS † (μIU/ml) HOMA-IR † QUICK-index † HOMA β-cell function † Renal parameters Creatinine † (μmol/l) Uric Acid † (μmol/l) Hepatic parameters AST † (IU/l) ALT † (IU/l) γGT † (IU/l) AP † (IU/l) Other biochemical parameters hs-CRP † (mg/l) Fasting plasma leptin † (ng/ml)

Non-Obese

ARTICLE IN PRESS H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

5

Figure 1. Lipid peroxidation markers according to the glycemic conditions and obesity status of the study participants (N=195). Plasma levels of MDA (A) and IgA to CuOxLDL (B) by glycemic condition. The data are adjusted for age, sex and BMI. Plasma levels of MDA (C), IgA to CuOx-LDL (D), IgG to CuOx-LDL (E) and IgM to CuOx-LDL (F) by groups of BMI. The data are adjusted for age and sex. The data shown are mean values (SEM). **p<0.05; ***p<0.001. CuOx-LDL, LDL oxidized by exposure to copper-ions; MDA, malondialdehyde; OD, optical density; T2D, type 2 diabetes.

Figure 2. Associations between oxidative stress markers, body weight and glucose metabolism parameters in the whole study population. (A) HOMA-IR (n=188) according to quartiles of MDA and groups of body weight; (B) HOMA-IR (n=188) according to quartiles of plasma IgA to CuOx-LDL levels and groups of body weight. The data shown are mean values (SEM). CuOx-LDL, LDL oxidized by exposure to copper-ions; HOMA-IR, homeostatic model assessment and insulin resistance; MDA, malondialdehyde.

biochemical parameters of obese subjects, the group with diabetes displayed higher levels of TG (p=0.006) and AP (p=0.030) and lower levels of LDL-cholesterol (p=0.041) compared to the groups without diabetes. Type 2 diabetes obese subjects showed also a significantly higher level of γGT than prediabetes (p=0.020) and nontype 2 diabetes subjects (p=0.013).

However, this difference did not reach statistical significance in adjusted model (p=0.104, Figure 1B). There were no differences in plasma IgG (p=0.753) and IgM (p=0.930) levels to CuOx-LDL between studied groups (data not shown).

Lipid peroxidation markers according to diabetes status

Obese subjects displayed higher MDA concentrations than normal weight subjects, even after adjusting for age and sex (p<0.001) (Figure 1C). However, none of the studied antibodies to CuOx-LDL were associated with obesity (Figure 1D-F). Figure 2 (panels A-B) illustrated three dimensional bar graphs of HOMA-IR index by groups of BMI and quartiles of plasma lipid peroxidation markers. HOMA-IR index was positively associated with plasma MDA (p=0.001) and obesity (p<0.001) (Figure 2A). Moreover, HOMA-IR index was highest among the obese groups having the highest plasma MDA levels. We showed also that HOMA-IR index

The mean MDA level was found to be significantly higher in subjects with prediabetes (p=0.011) and diabetes (p<0.001) when compared to those with normal glucose regulation, after adjustment for age, sex and BMI (Figure 1A). The mean value of the IgA antibody to CuOx-LDL, expressed in optic density, was higher in subjects with type 2 diabetes when compared to normoglycemic and prediabetes subjects (p=0.010 and p=0.034, respectively in unadjusted model, data not shown).

Lipid peroxidation markers according to obesity status

ARTICLE IN PRESS 6

H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

Table 2 Univariate relationship between lipid peroxidation markers and selected anthropometric and metabolic parameters in the whole study subjects Variable

Log BMI Log WC Log Body Fat Log Leptin Log Glucose Log A1c Log Insulin Log HOMA-IR Log HOMA β-Cell function Cholesterol Log HDL-C LDL-C Log TG Log CRP Log MDA

Log MDA

Log IgG to CuOx-LDL

Log IgA to CuOx-LDL

Log IgM to CuOx-LDL

r value

p value

r value

p value

r value

p value

r value

p value

0.263 0.286 0.238 0.295 0.219 0.263 0.399 0.466 0.104 0.084 −0.010 0.003 0.293 0.327 -

<0.001 <0.001 0.001 <0.001 0.002 <0.001 <0.001 <0.001 0.162 0.248 0.889 0.971 <0.001 <0.001 -

0.090 0.107 0.085 0.076 0.070 0.051 0.179 0.190 0.055 0.215 0.072 0.192 0.097 0.063 0.064

0.239 0.167 0.270 0.312 0.351 0.502 0.018 0.012 0.472 0.004 0.337 0.010 0.194 0.404 0.398

−0.022 0.006 −0.038 0.037 0.213 0.202 0.118 0.209 −0.074 0.009 −0.026 0.021 0.026 0.057 0.161

0.774 0.934 0.621 0.620 0.004 0.006 0.116 0.005 0.326 0.900 0.731 0.779 0.727 0.449 0.030

0.036 −0.021 0.023 0.053 −0.135 −0.192 0.043 −0.032 0.138 −0.066 0.090 −0.061 −0.068 −0.035 −0.070

0.629 0.783 0.761 0.470 0.062 0.008 0.564 0.670 0.062 0.365 0.216 0.400 0.351 0.638 0.341

Pearson correlation coefficients are indicated, p<0.05 is considered as statistically significant. A1C, glycated hemoglobin; BMI, body mass index; CRP, C-reactive protein; CuOx-LDL, LDL oxidized by exposure to copper-ions; HDL-C, high density lipoprotein-cholesterol; HOMA β-cell function, homeostasis model assessment of beta cell function; HOMA-IR, homeostasis model assessment of insulin resistance; LDL-C, low density lipoproteincholesterol; MDA, malondialdehyde; TG, triglycerides; WC, waist circumference.

was significantly associated with plasma IgA to CuOx-LDL (p=0.002) (Figure 2B). Additionally, the highest HOMA-IR index was observed among obese subjects with the highest IgA levels to CuOx-LDL. The statistical interactions between all tested factors were nonsignificant, suggesting that the above observed associations were consistent.

Table 3 Stepwise multiple linear regression analysis of association between lipid peroxidation markers and diabetes-related traits Dependent variables

Independent variables

β

SE

P value

Log Glucose R2=0.469 R2adjusted=0.439

age Log TG Log BMI Log CRP LDL-C Sex Log IgA to CuOx-LDL Hypertension age Log BMI Log TG Log MDA Cholestrol Sex Log BMI Family history of diabetes Log MDA Log TG Smoking Log MDA Log IgA to CuOx-LDL Log BMI Family history of diabetes

0.209 0.386 −0.471 0.179 −0.196 0.193 0.136 0.150 0.342 −0.422 0.382 0.185 −0.195 0.158 0.550 0.189 0.138 0.365 −0.201 0.169 0.171 0.191 0.170

0.001 0.062 0.134 0.032 0.016 0.025 0.031 0.029 0.001 0.096 0.051 0.062 0.011 0.021 0.195 0.050 0.139 0.111 0.032 0.160 0.060 0.240 0.056

0.005 <0.001 <0.001 0.023 0.003 0.005 0.033 0.037 <0.001 <0.001 <0.001 0.007 0.005 0.022 <0.001 0.005 0.049 <0.001 0.005 0.018 0.007 0.009 0.011

Relation of lipid peroxidation markers with anthropometric and metabolic variables by univariate analysis Correlation coefficients for the associations between oxidative stress markers with anthropometric and biochemical variables are shown in Table 2. We observed significant positive association of MDA levels with BMI (p<0.001), WC (p<0.001), body fat content (p=0.001) and leptin (p<0.001); meanwhile, none of all three antibodies to CuOx-LDL were correlated with variables related to obesity. Regarding glucose metabolism parameters, we found that MDA levels were positively correlated with fasting plasma glucose (p=0.002), A1C (p<0.001), fasting plasma insulin (p<0.001) and HOMA-IR (p<0.001). Moreover, both IgG and IgA to CuOx-LDL were positively correlated with HOMA-IR (p=0.012 and p=0.005, respectively). Furthermore, IgG to CuOx-LDL displayed significant positive correlation with fasting plasma insulin (p=0.018), while IgA to CuOx-LDL showed positive association with fasting plasma glucose (p=0.004) and A1C (p=0.006). IgM to CuOx-LDL was negatively correlated with A1C (p=0.008). Among lipid peroxidation markers studied, only IgG was positively correlated with plasma total cholesterol (p=0.004) and LDLcholesterol (p=0.010), whereas MDA was positively associated with TG (p<0.001) and CRP (p<0.001). We also reported positive correlation between MDA and IgA to CuOx-LDL levels (p=0.030). Association of lipid peroxidation markers with diabetes parameters by multivariate analysis Multiple stepwise regression analyses were further performed to evaluate independent association between lipid peroxidation markers and diabetes-related traits in a model including age, sex, BMI, total cholesterol, TG, LDL-C, CRP, smoking, alcoholism, hypertension and family history of diabetes as additional covariates (Table 3). MDA was found to be independently associated with A1C (p=0.007), fasting plasma insulin (p=0.049) and HOMA-IR (p=0.018) in the overall study cohort. MDA did not emerge as independent

Log A1C R2=0.473 R2adjusted=0.451

Log Insulin R2=0.446 R2adjusted=0.434 Log HOMA-IR R2=0.476 R2adjusted=0.453

Values are beta coefficients with SE. Independent variables are: age, sex, Log BMI, Cholesterol, Log TG, LDL-C, Log CRP, Log IgG to CuOx-LDL, Log IgA to Cu-OxLDL, Log IgM to CuOx-LDL, Log MDA, smoking, alcoholism, family history of diabetes and hypertension. Only variables that had a P<0.05 were considered in the final fitted model. β: standardized coefficient, SE: standard error. β represents the change in the dependent variable according to 1-unit increase of Log IgA to CuOx-LDL, Log MDA and other covariates. A1C, glycated hemoglobin; BMI, body mass index; CRP, C-reactive protein; CuOxLDL, LDL oxidized by exposure to copper-ions; HOMA-IR, homeostasis model assessment of insulin resistance; LDL-C, low density lipoprotein-cholesterol; MDA, malondialdehyde; TG, triglycerides.

determinant of both fasting plasma glucose and HOMA β-cell function. IgA to CuOx-LDL but not IgG or IgM to CuOx-LDL was an independent predictor of fasting plasma glucose (p=0.033), and HOMA-IR (p=0.007). The univariate relationship previously described between plasma IgA to CuOx-LDL and A1C disappeared after multivariate adjustments for the independent effects of covariates (data not shown). The association between plasma IgA to CuOx-LDL and both fasting plasma insulin and HOMA β-cell function remained nonsignificant in multivariate analysis (data not shown).

ARTICLE IN PRESS H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

7

Figure 3. Binary logistic regression analyses of lipid peroxidation markers and the occurrence of type 2 diabetes, according to model 1 (A and B), model 2 (C) or both models (D). In model 1, MDA and antibodies to CuOx-LDL were analyzed separately. In model 2, MDA and antibodies to CuOx-LDL were analyzed simultaneously. Additional covariates included in the model were age, sex, smoking, waist circumference, triglycerides and low-density lipoprotein-cholesterol. CI, confidence interval; CuOx-LDL, LDL oxidized by exposure to copper-ions; MDA, malondialdehyde; T2D, type 2 diabetes; WC, waist circumference.

The final fitted models explained 43.9%, 45.1%, 43.4% and 45.3% of fasting plasma glucose, A1C, Insulin levels and HOMA-IR variance, respectively.

Predictive Role of Lipid Peroxidation Markers for Glucose Metabolism Disorders Logistic regression analyses were performed to evaluate the potential independent role of oxidative stress markers in the prediction of glucose metabolism disorders. MDA and antibodies to CuOx-LDL were first analyzed separately in a model 1, adjusted for the following confounding factors: age, sex, WC, TG, LDL and smoking. We demonstrated that higher MDA levels were significantly associated with increased risk of carbohydrate metabolism disorders (OR: 1.226, 95%CI 1.089–1.379, p=0.001), particularly with incident type 2 diabetes (OR: 1.196, 95%CI 1.058–1.353, p=0.004) (Figure 3A). The whole model (including MDA, age, TG and LDL levels) explained 61.5% of the risk of having diabetes, with MDA alone explaining 8.3%. We also showed that only IgA to CuOx-LDL but not IgG and IgM to CuOx-LDL was an independent risk factor for type 2 diabetes (OR: 1.626, 95%CI 1.066– 2.481, p=0.024) in a model adjusted for the same covariates (Figure 3B). The whole model (including IgA to CuOx-LDL, age, TG and LDL levels) explained 63% of the risk of having type 2 diabetes, with IgA to CuOx-LDL alone explaining 2.7%. In addition, we performed a second model (model 2), adjusted for the same covariates and by considering the simultaneous role of lipid peroxidation markers, in order to determine the best and important marker in the occurrence of type 2 diabetes. We showed that high MDA levels remained an independent risk factor for type 2 diabetes in normoglycemic subjects (OR: 1.285, 95%CI 1.101– 1.500, p=0.001) (Figure 3C). An exciting result was that plasma IgA to CuOx-LDL, when analyzed separately (model 1), was an independent determinant of type 2 diabetes in patients with prediabetes (OR: 2.380, 95%CI 1.141–4.965, p=0.021) (Figure 3D). More interestingly, similar independent role of IgA to CuOx-LDL in risk of having type 2 diabetes

in the group with prediabetes was obtained when analyzed according to model 2 (OR: 2.321, 95%CI 1.063–5.066, p=0.035) (Figure 3D). We further assessed the effect of obesity on the association of lipid peroxidation markers with the risk of glucose metabolism disorders. For patients with a BMI<25 kg m−2, both MDA and IgA to CuOx-LDL were independent predictors of occurence of type 2 diabetes (OR: 1.475, 95%CI 1.144–1.903, p=0.003 and OR: 2.011, 95%CI 1.035–3.909, p=0.039, respectively) (data not shown). For those with a BMI≥25 kg m-2, only elevated IgA to CuOx-LDL level predicted the development of type 2 diabetes in the group with prediabetes (OR: 2.235, 95%CI 1.023–4.882, p=0.044) (data not shown).

Discussion Given the well-described deleterious role of increased oxidative stress in diabetes (29–31), the present cross-sectional study has addressed the potential importance of two different lipid peroxidation markers (MDA and antibodies titers to CuOx-LDL) in the discrimination of subjects at high risk for diabetes. Earlier studies have documented increased lipid oxidation in patients with carbohydrate metabolism disorders (9,32). MDA has been recognized as a relevant lipid peroxidation end product marker. Several investigations have reported elevated MDA levels in type 2 diabetes subjects (9), a finding confirmed by our study. Although high MDA levels may be involved in the development of diabetic complications (33), no previous data have addressed its potential ability to predict diabetes. An important observation in the present study was that elevated MDA level may independently predict the occurrence of type 2 diabetes by approximately 1.3 times. In addition, we showed significant relationship of MDA with obesity and its related parameters (Figure 2A and univariate analysis) as previously reported (34). These findings suggest that obesity gave rise to increased oxidative stress in type 2 diabetes patients. Thus, we expected that obesity may strengthen the association between plasma MDA and the risk for type 2 diabetes. However, from our stratified analyses by BMI, we reported that elevated MDA may predict the development of type 2 diabetes especially in normal

ARTICLE IN PRESS 8

H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

weight subjects but not in obese ones. This finding suggests that diabetes, even in the absence of obesity, can promote lipid peroxidation through hyperglycemia. In fact, as numerous mechanisms may promote the increase of oxidative stress in diabetes, hyperglycemia per se may promote the generation of free radicals which in turn may catalyze lipid peroxidation (35,36). To our knowledge, our findings provide new insights into the predictive value of MDA for assessing the risk of type 2 diabetes. On the other hand, our study, using multivariate analyses, showed an independent association of MDA with hyperinsulinemia and insulin resistance, suggesting an etiological role of MDA on glucose metabolism alteration. Indeed, a recent study has shown that elevated MDA level inhibited glucose stimulating insulin secretion (11). However, the exact signalling role of MDA in pathological processes of type 2 diabetes is not completely understood. With respect to Ox-LDL antibodies, there are only few data assessing their role in the prediction of diabetes in humans, with conflicting results. Our study reported that higher IgA titers to CuOx-LDL increased the risk of type 2 diabetes in normoglycemic subjects by approximately 1.6 times regardless of classical risk factors, as it was previously reported (17). However, no available data have addressed the effect of obesity on the magnitude of this association. The most relevant finding from this study was the independent role of IgA to CuOx-LDL in the occurrence of type 2 diabetes in obese prediabetes patients (non-obese prediabetes patients are lacking in our cohort), which could have a clinical relevance in discriminating subjects with high risk for type 2 diabetes. More interestingly, increased IgA to CuOx-LDL was associated with insulin resistance in obese subjects as previously reported (17,21), suggesting that IgA to CuOx-LDL may be merely a marker of disturbances in glucose metabolism in predisposed obese subjects. Therefore, specifically designed studies are needed to confirm the exact role of IgA antibodies to OxLDL in glucose metabolism. Otherwise, we showed no association of IgM to CuOx-LDL levels and the risk of diabetes as reported by (17). The lack of association of IgG to CuOx-LDL and diabetes risk was discordant with the findings of Garrido-Sanchez et al and may be related to the nature of the studied sample (hospitalized patients with cardiovascular diseases which were an exclusion criterion in our study) (19). Why only IgA class of autoantibodies was affected in patients with diabetes? The underlying mechanism is yet unclear, but could be related to gut microbiota. In fact, the IgA is the predominant immunoglobulin class in mucosal secretions and it has been suggested that it may provide a second line of immunoprotection against bacteria invading via intestinal surfaces and portal vein (37). Considering the simultaneous role of MDA and IgA to CuOxLDL in the occurrence of type 2 diabetes, it is not fully understood which one was better indicator of type 2 diabetes. When the analysis was expanded to consider combinations of all lipid peroxidation markers, our findings emphasized the independent role and the importance of IgA to CuOx-LDL in the occurrence of type 2 diabetes, particularly in prediabetes subjects. Thus, IgA to CuOx-LDL appeared to be the most sensitive biomarker of the disease progression which could have implications for clinical practice to discriminate subjects at higher risk for type 2 diabetes. Nevertheless, the positive association between MDA and IgA to CuOx-LDL (Table 2) allowed us to suggest that MDA generation could potentially trigger the production of auto-antibodies cross-reacting with this aldehyde (21). The exact pathophysiological mechanisms underlying the association between lipid peroxidation and type 2 diabetes are not fully understood. Some plausible pathways are: i) Hyperglycemia that promotes the generation of free radicals by several mechanisms including glucose auto-oxidation, which in turn may catalyze lipid peroxidation (35,36). ii) Hypertriglyceridemia leading to increase

of reactive oxygen species (ROS) production (38) and lipid peroxides levels (39). iii) Hyperleptinemia, as it was correlated with MDA in our cohort, may increase oxidative stress and lipid peroxidation (40). Our study has several strengths: first, the diagnosis of diabetes was based on oral glucose tolerance test. Second, we evaluated the association of lipid peroxidation level with a large panel of blood markers of glucose homeostasis. However, our study has some limitations: first, it consists on cross-sectional study design that addresses associations between lipid peroxidation markers, obesity and diabetes which restrains the interpretation of cause and effect relations. Second, the sample size for the prediabetes group was relatively small and all of them are obese.

Conclusions We demonstrated the association of lipid peroxidation markers with blood parameters of glucose metabolism and obesity. Interestingly, we proposed MDA and IgA to CuOx-LDL levels as new useful biomarkers of glucose metabolism disorders and the disease progression through deterioration in insulin sensitivity. We believe that our results could have potentially important clinical relevance since lipid peroxidation markers may be additionally useful in discriminating individuals at higher risk for type 2 diabetes. However, the mechanisms by which lipid peroxidation, particularly MDA and IgA to CuOx-LDL, can mediate such dysregulation have not been elucidated. Thus, further investigations are needed to explore the related signaling pathway and consequently to establish appropriate clinical treatment strategy.

Acknowledgments We thank the staff at the department of Endocrinology at the hospital of Hedi Chaker of Sfax for excellent research assistance. We are grateful to members of Biochemistry department at the hospital of Hedi Chaker of Sfax and members of immunology department at the hospital of Habib Bourguiba of Sfax for biochemical analyses. We thank Dr. Anatol Kontush (INSERM Unit 1166, Paris, France) for critical reading of our manuscript. We would like to express our gratitude to all participants for their contribution.

Author Contributions All authors contributed to data analysis, data interpretation of the manuscript and approved the final draft for submission.

References 1. Hardy DS, Stallings DT, Garvin JT, et al. Anthropometric discriminators of type 2 diabetes among White and Black American adults. J Diabetes 2017;9:296– 307. 2. Wang Y-L, Koh W-P, Talaei M, et al. Association between the ratio of triglyceride to high-density lipoprotein cholesterol and incident type 2 diabetes in Singapore Chinese men and women. J Diabetes 2017;9:689–98. 3. Doi Y, Kiyohara Y, Kubo M, et al. Elevated C-reactive protein is a predictor of the development of diabetes in a general Japanese population: The Hisayama study. Diabetes Care 2005;28:2497–500. 4. Holvoet P, Lee D-H, Steffes M, et al. Association between circulating oxidized low-density lipoprotein and incidence of the metabolic syndrome. JAMA 2008;299:2287–93. 5. Rösen P, Nawroth PP, King G, et al. The role of oxidative stress in the onset and progression of diabetes and its complications: A summary of a Congress Series sponsored by UNESCO-MCBN, the American Diabetes Association and the German Diabetes Society. Diabetes Metab Res Rev 2001;17:189–212.

ARTICLE IN PRESS H. Bouhajja et al. / Can J Diabetes xxx (2017) 1–9

6. Kopprasch S, Pietzsch J, Kuhlisch E, et al. In vivo evidence for increased oxidation of circulating LDL in impaired glucose tolerance. Diabetes 2002;51:3102–6. 7. Olusi SO. Obesity is an independent risk factor for plasma lipid peroxidation and depletion of erythrocyte cytoprotectic enzymes in humans. Int J Obes 2002;26:1159–64. 8. Cabré A, Girona J, Vallvé JC, Masana L. Aldehydes mediate tissue factor induction: A possible mechanism linking lipid peroxidation to thrombotic events. J Cell Physiol 2004;198:230–6. 9. Bandeira SDM, Guedes GDS, Da Fonseca LJS, et al. Characterization of blood oxidative stress in type 2 diabetes mellitus patients: Increase in lipid peroxidation and SOD activity. Oxid Med Cell Longev 2012;2012:819310. 10. Davì G, Falco A, Patrono C. Lipid peroxidation in diabetes mellitus. Antioxid Redox Signal 2005;7:256–68. 11. Wang X, Lei XG, Wang J. Malondialdehyde regulates glucose-stimulated insulin secretion in murine islets via TCF7L2-dependent Wnt signaling pathway. Mol Cell Endocrinol 2014;382:8–16. 12. Lopes-Virella MF, Virella G. Pathogenic role of modified LDL antibodies and immune complexes in atherosclerosis. J Atheroscler Thromb 2013;20:743– 54. 13. Ylä-Herttuala S, Palinski W, Butler SW, et al. Rabbit and human atherosclerotic lesions contain IgG that recognizes epitopes of oxidized LDL. Arterioscler Thromb 1994;14:32–40. 14. Shimada K, Mokuno H, Matsunaga E, et al. Predictive value of circulating oxidized LDL for cardiac events in type 2 diabetic patients with coronary artery disease. Diabetes Care 2004;27:843–4. 15. Wegner M, Piorunska-Stolzmann M, Araszkiewicz A, et al. Does oxidized LDL contribute to atherosclerotic plaque formation and microvascular complications in patients with type 1 diabetes? Clin Biochem 2012;45:1620–3. 16. Virella G, Koskinen S, Krings G, et al. Immunochemical characterization of purified human oxidized low-density lipoprotein antibodies. Clin Immunol 2000;95:135–44. 17. Sämpi M, Veneskoski M, Ukkola O, et al. High plasma immunoglobulin (Ig) A and low IgG antibody titers to oxidized low-density lipoprotein are associated with markers of glucose metabolism. J Clin Endocrinol Metab 2010;95:2467– 75. 18. Garrido-Sánchez L, Cardona F, García-Fuentes E, et al. Anti-oxidized lowdensity lipoprotein antibody levels are associated with the development of type 2 diabetes mellitus. Eur J Clin Invest 2008;38:615–21. 19. Garrido-Sánchez L, García-Pinilla JM, Jimen´ez-Navarro M, et al. Reduced levels of anti-MDA LDL antibodies in patients with carbohydrate metabolism disorders. Clin Lab 2011;57:901–7. 20. Babakr AT, Elsheikh OM, Almarzouki AA, et al. Relationship between oxidized low-density lipoprotein antibodies and obesity in different glycemic situations. Diabetes Metab Syndr Obes 2014;7:513–20. 21. Vehkala L, Ukkola O, Kes Ä, et al. Plasma IgA antibody levels to malondialdehyde acetaldehyde-adducts are associated with infl ammatory mediators, obesity and type 2 diabetes. Ann Med 2013;45:501–10.

9

22. Suzuki K, Ito Y, Ochiai J, et al. Relationship between obesity and serum markers of oxidative stress and inflammation in Japanese. Asian Pac J Cancer Prev 2003;4:259–66. 23. Wonisch W, Falk A, Sundl I, et al. Oxidative stress increases continuously with BMI and age with unfavourable profiles in males. Aging Male 2012;15:159– 65. 24. World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: Diagnosis and classification of diabetes mellitus provisional report of a WHO consultation, vol. 15. 1999. 25. Khaodhiar L, Blackburn GL. Obesity assessment. Am Heart J 2001;142:1095– 101. 26. OMS. Obesity: Preventing and managing the global epidemic 1998:1–158. 27. Choi K, Lee HS, Chung HK. Production and characterization of monoclonal antibodies to oxidized LDL. Exp Mol Med 1998;30:41–5. 28. Drapper HH, Hadley M. Malondialdehyde determination as index of lipid peroxidation. Methods Enzymol 1990;186:421–31. 29. Furukawa S, Fujita T, Shumabukuro M, et al. Increased oxidative stress in obesity and its impact on metabolic syndrome. J Clin Invest 2004;114:1752–61. 30. Rudich A, Tlrosh A, Potashnik R, et al. Prolonged oxidative stress impairs insulininduced GLUT4 translocation in 3T3-L1 adipocytes. Diabetes 1998;47:1562–9. 31. Matsuoka TA, Kajimoto Y, Watada H, et al. Glycation-dependent, reactive oxygen species-mediated suppression of the insulin gene promoter activity in HIT cells. J Clin Invest 1997;99:144–50. 32. Huang M, Que Y, Shen X. Correlation of the plasma levels of soluble RAGE and endogenous secretory RAGE with oxidative stress in pre-diabetic patients. J Diabetes Complications 2015;29:422–6. 33. Tiwari BK, Pandey KB, Abidi B, Rizvi SI. Markers of oxidative stress during diabetes mellitus. J Biomark 2013;2013:378790. 34. Weinbrenner T, Schröder H, Escurriol V, et al. Circulating oxidized LDL is associated with increased waist circumference independent of body mass index in men and women. Am J Clin Nutr 2006;83:30–5. 35. Lopes-Virella MF, Klein RL, Virella G. Modification of lipoproteins in diabetes. Diabetes Metab Rev 1996;12:69–90. 36. Mullarkey CJ, Edelstein D, Brownlee M. Free radical generation by early glycation products: A mechanism for accelerated atherogenesis in diabetes. Biochem Biophys Res Commun 1990;173:932–9. 37. van Egmond M, van Garderen E, van Spriel B, et al. FcαRI-positive liver Kupffer cells: Reappraisal of the function of immunoglobulin A in immunity. Nat Med 2000;6:680–5. 38. Fridlyand LE, Philipson LH. Oxidative reactive species in cell injury: Mechanisms in diabetes mellitus and therapeutic approaches. Ann N Y Acad Sci 2005;1066:136–51. 39. Efe H, Deger O, Kirci D, et al. Decreased neutrophil antioxidative enzyme activities and increased lipid peroxidation in hyperlipoproteinemic human subjects. Clin Chim Acta 1999;279:155–65. 40. Pandey G, Shihabudeen MS, David HP, et al. Association between hyperleptinemia and oxidative stress in obese diabetic subjects. J Diabetes Metab Disord 2015;14:24.