Lipid and protein oxidation contribute to a prothrombotic state in patients with type 2 diabetes mellitus


Dr Raimondo De Cristofaro, Hemostasis Research Center, Catholic University School of Medicine, Largo F. Vito 1, 00168 Rome, Italy.Tel.: +39 06 30154438; fax: +39 06 30155915; e-mail:


Summary.  Diabetes mellitus (DM) is associated with enhanced lipid oxidation and persistent platelet activation. We investigated whether oxidant stress (OS) also affects circulating proteins and is associated with an abnormal coagulative pattern. In 72 type 2 DM (T2DM) patients, urinary 8-iso-prostaglandin (PG) F and 11-dehydro-thromboxane B2 (TXM) were measured as markers of lipid peroxidation and platelet activation, respectively. The carbonyl content of plasma proteins (PCARB) was measured as global index of protein oxidation. 8-Iso-PGF and PCARB levels were higher in DM patients than in controls (P < 0.05). Likewise, both TXM and prothrombin F1+2 levels were higher in diabetics (P < 0.05). By contrast, anticoagulant markers, such as activated protein C, protein C activation peptide, and soluble thrombomodulin (TM) were depressed in T2DM (P < 0.05). In conclusion, OS in T2DM involves circulating proteins and is associated with an unbalanced promotion of procoagulant reactions. These effects in concert with platelet activation may contribute to atherothrombotic complications in T2DM.

Oxidant stress (OS) has been reported to be increased in diabetes mellitus (DM) and hypothesized to be involved in the pathogenesis of vascular damage and thromboembolic complications [1,2]. Diabetic patients are characterized by increased urinary excretion of F2-isoprostanes [2], which are produced from arachidonic acid through a nonenzymatic process of lipid peroxidation, catalyzed by oxygen-free radicals on cell membranes and low density lipoprotein (LDL) particles [3,4]. Enhanced lipid peroxidation and the generation of bioactive isoprostanes may represent an important biochemical link between impaired glycemic control and persistent platelet activation [2–4]. In addition, the formation of advanced glycation end (AGE) products, such as pentosidine, acrolein, carboxymethyl-lysine, may lead to oxidant damage of proteins, which in turn may accelerate AGE formation [1,5,6].

Moreover, oxidant damage may promote coagulation and alter the structure and the function of coagulative proteins, as previously shown both in vitro and in vivo[7–13]. In fact, oxidized lipid micelles accelerate the activity of the prothrombinase complex in vitro, thus enhancing thrombin production [7,8]. In addition, upon exposure to oxidizing stimuli, fibrinogen, thrombin, and physiological anticoagulants, such as protein S and antithrombin III, undergo major structural and functional changes [9–12]. Furthermore, OS may affect thrombomodulin (TM), an endothelial protein which enables thrombin to activate the protein C (PC) pathway [14]. For instance, oxidation of TM methionine 388 by either H2O2 or activated PMN reduces its cofactor capacity by 90% [15]. Finally, oxidant agents such as homocysteine and oxidized LDL inhibit both the synthesis and surface expression of TM in cultured endothelial cells [16,17]. Therefore, OS may not only favor coagulative activation [7,8], but also impair anticoagulant pathway.

We have previously reported biochemical evidence of enhanced generation of F2-isoprostanes [2] and persistent platelet activation, as reflected by enhanced excretion of 11-dehydro-thromboxane B2, a major TXA2 enzymatic metabolite excretion, in patients with T2DM [18]. We speculated that increased OS associated with DM could also induce oxidative changes in proteins, such as those involved in coagulation, thus altering their function. Therefore, in the present study we investigated whether the carbonyl content of plasma proteins is altered in T2DM patients when compared with age-matched non-diabetic subjects and whether it correlates with the rate of F2-isoprostane formation. We also measured both the activation peptide released from PC by the thrombin–TM complex, and the activation peptides released from procoagulant enzymes/substrates (prothrombin F1+2 fragment and fibrinopeptide A) along with plasma level of soluble TM. The association analysis of both pro- and anticoagulant markers with oxidation indexes provided a multidimensional picture of how oxidant stress affects the hemostatic balance in DM.

Materials and methods


Seventy-two patients with T2DM (39 F, 33 M, aged 46–86 years), as defined in accordance with the criteria of the American Diabetes Association, and 72 age- and gender-matched healthy subjects were enrolled in the study. The baseline characteristics of patients and controls are detailed in Table 1. Diabetic patients were examined for the presence of macrovascular and microvascular (retinal) complications. Of the 72 patients, 10 patients had a past history or physical examination positive for evidence of macrovascular complications. Six patients had stable angina pectoris or had had a myocardial infarction, two had cerebrovascular disease (one transient ischemic attack and 1 previous stroke), and two had peripheral vascular disease. Patients with coronary heart disease were in a stable phase. Patients with peripheral vascular disease were in Fontaine stage II (intermittent claudication, ankle-arm pressure index of less than 0.85, and no resting pain) with a constant level of pain while walking. In none of the patients had vascular disease undergone detectable progression during the previous 6 months, as judged by clinical examination during outpatient visits. Moreover, all had mild-to-moderate symptoms compatible with a virtually normal lifestyle. At the time of the study, diabetic patients were being treated by diet alone (six patients), insulin alone (two patients) or by diet plus oral hypoglycemic agents (metformin and/or sulfonylureas) (61 patients); in three patients insulin was added to the oral hypoglycemic agents. Thirty-eight patients had arterial hypertension, defined as current systolic/diastolic blood pressure higher than 140/90 mmHg. Thirteen patients were hypercholesterolemic (blood cholesterol level >240 mg dL−1).

Table 1.  Clinical characteristics of diabetic patients and control
(n = 72)
(n = 72)
Sex (F/M)39/3339/33
Age (years)63 ± 1257 ± 13
Cigarette smoking00
Diabetes duration (years)12 ± 10
Fasting blood glucose (mg dL−1)211 ± 14988 ± 14
HbA1c (%)7.7 ± 1.2ND
Hypertension (n)380
Hypercholesterolemia (n)130
Macrovascular complications (n)100
Microvascular complications (n)100

Patients with renal insufficiency or proteinuria (by serum creatinine levels and urinalysis), altered hepatic function (by liver enzymes), or alcohol abuse were excluded. Patients with cancer, recent infectious diseases, autoimmune disorders, pregnancy, presence of activated protein C resistance, ongoing antioxidant (vitamin C and vitamin E),-non-steroidal anti-inflammatory drugs (NSAIDs), iron or contraceptive treatments were also excluded. Both diabetic patients and healthy subjects were selected for being non-smokers at the time of study, in order to eliminate a potential confounder. Informed consent was obtained from each participating subject, and the protocol was approved by the ethical committee.

Blood sample collection

Fasting blood samples were collected by venipuncture in 0.38% sodium citrate (final concentration), with addition of either 10 µm PPACK for samples used to measure FpA, or 1 mm EDTA for measurement of the protein carbonyl content. After collection of blood samples, plasma fractions were obtained by centrifugation at 4 °C for 15 min at 1800 × g, quickly frozen in liquid nitrogen, and stored at −80 °C until use.

Coagulation assays

Determinations of prothrombin time (measured as International Normalized Ratio [INR]), aPTT, fibrinogen (Clauss method), Protein C (chromogenic method), factor II (coagulative method), and activated PC-resistance were carried out with automatic coagulometers (Electra MLA 1800C and Futura-Plus, Instrumentation Laboratory, Milan, Italy) using recombinant tissue factor (RecombiPlastin), synthetic phospholipids and colloidal silica (SynthAsil), the Fibrinogen-Clauss kit, the ProChrom kit, lyophilized factor II-deficient plasma, and APC-resistance kit purchased from Instrumentation Laboratory (Milan, Italy).

Measurements of Fibrinopeptide A, F1+2, soluble TM, von Willebrand factor, and protein concentration

Plasma FpA concentration was measured by RIA (IMCO Co., Stockholm, Sweden), according to the manufacturer's instructions. F1+2 levels were measured by an ELISA method from Dade-Behring (Enzygnost F1+2 micro). Soluble TM and von Willebrand factor (VFW) antigen concentrations were measured by an ELISA method from American Diagnostica Inc. (Greenwich, CT, USA) and an immunoturbidimetric assay (STA-LIA VFW) using the STA Compact automatic coagulometer (Diagnostica Stago, Asniere, France), respectively. Plasma protein concentration was estimated by a colorimetric micromethod using the Bio-Rad Protein assay, using bovine serum albumin (BSA) for the reference curve.

Radioimmunoassay for PCP detection in plasma

The RIA method for PCP quantification, was performed on citrated plasma samples, that were deproteinized and concentrated, according to the method reported by Bauer et al.[19,20]. Radiolabeling of Tyr-PCP was accomplished by the Iodogen method [21], using 10 µg of peptide and 0.5 mCi of carrier free Na125I (Amersham Co., Arlington Heights, IL, USA). Detection of plasma PCP was accomplished by a double antibody method, as recently described [22]. The unknown concentration of the competing PCP was calculated by the grafit software (Erithacus software Ltd, Staines, UK). The measurements were always performed in duplicate.

Quantification of circulating activated protein C in plasma

Detection of circulating activated PC (APC) was accomplished with a previously reported technique [23], modified as recently reported [22].

Quantification of plasma protein carbonyl groups

Carbonyl groups formed in plasma proteins were quantified by means of 2,4-dinitrophenylhydrazine (DNPH), as previously described [22,24]. Absorbance of the carbonyl-DNPH adduct was measured at 365 nm, and the results were expressed as weight of 2,4-dinitrophenylhydrazine/weight of protein, based on an average absorptivity of 21 mmol L−1 cm−1 for the adduct and a molecular weight of 28 for the carbonyl group.

Urinary eicosanoid assays

Urinary 8-iso-PGF and 11-dehydro-TXB2 were measured by previously described radioimmunoassay methods [25,26]. These radioimmunoassays have been validated using different antisera and by comparison with gas chromatography/mass spectrometry, as detailed elsewhere [25,26].

Clinical laboratory measurements

Fasting plasma glucose was measured by the glucose oxidase method. HbA1c level was measured by automated HPLC (Menarini, Italy). All blood samples for lipid studies were drawn into EDTA (1 mg mL−1) and separated within 1 h after sampling. Total cholesterol and triglycerides were determined by an enzymatic method and HDL cholesterol was measured after phosphotungstic acid/MgCl2 precipitation on fresh plasma. LDL cholesterol was calculated using Friedewald's formula [27].

Statistical analysis

Statistical analysis was carried out using the spss for Windows software (spss for Windows, version 6.0). The data were analyzed by non-parametric methods to avoid assumptions about the distribution of the measured variables. Two-tailed significance tests were routinely performed. The associations between the biochemical and hemostatic parameters were assessed by the Spearman rank correlation test. In all cases, a value of P ≤ 0.05 was considered statistically significant. Values are expressed as means ± SD.


Eicosanoid and isoeicosanoid excretion

Consistent with previous findings [2,18], urinary 8-iso-PGF and 11-dehydro-TXB2 excretion were significantly higher in diabetic patients than in control subjects (323 ± 179 vs. 197 ± 69 pg mg−1 creatinine, P < 0.05 and 1084 ± 1082 vs. 208 ± 92 pg mg−1 creatinine, P < 0.05, respectively).

Protein oxidation and hemostatic markers

Plasma protein carbonyls (PCARB) were significantly higher in patients than in controls: 6.1 ± 1.4 × 10−6 vs. 4.6 ± 1 × 10−6 w/w, P < 0.05. Haemostatic markers, such as INR, aPTT, and fibrinogen were not significantly different in diabetic patients and controls. However, prothrombin fragment 1 + 2 (F1+2), which reflects active thrombin production, was significantly higher in diabetic than in control subjects (1.7 ± 0.8 vs. 1.1 ± 0.8 nmol L−1, P < 0.05), as shown in Fig. 1. FpA levels were numerically but not significantly higher in patients than controls (1.5 ± 1.1 vs. 1.2 ± 0.8 nmol L−1). VFW levels were not significantly different in patients and controls (12.2 ± 7.1 vs. 11.6 ± 2.6 µg mL−1).

Figure 1.

Plasma levels of F1+2 in 72 T2DM and 72 age and sex healthy subjects. Data represent individual measurements; horizontal bars are mean value for each group.

On the other hand, levels of APC, PCP and soluble TM were significantly reduced in diabetic patients compared to control subjects, as shown in Figs 2, 3, and 4. In fact, plasma APC levels were 0.3 ± 0.8 and 1.29 ± 0.65 ng mL−1 in patients and controls, respectively (P < 0.001, Fig. 2). Likewise, levels of PC activation peptide, PCP, were lower in patients than in healthy subjects (0.7 ± 0.6 pm vs. 1.3 ± 0.7 pmol L−1, P < 0.05, Fig. 3). By contrast, zymogen PC levels were even higher in diabetic patients than in controls (118 ± 30% vs. 98 ± 15%, respectively, P < 0.05). Plasma levels of soluble TM were lower in diabetic patients than in control subjects (1.63 ± 0.6 ng mL−1 vs. 3.6 ± 1 ng mL−1, respectively; P < 0.01), as shown in Fig. 4.

Figure 2.

Plasma levels of APC in 72 T2DM and 72 age and sex healthy subjects. Data represent individual measurements; horizontal bars are mean value for each group.

Figure 3.

Plasma levels of PCP in 72 T2DM and 72 age and sex healthy subjects. Data represent individual measurements; horizontal bars are mean value for each group.

Figure 4.

Plasma levels of TM in 72 T2DM and 72 age and sex healthy subjects. Data represent individual measurements; horizontal bars are mean value for each group.

Association analyses

No statistically significant difference was found for both coagulation and oxidation parameters between the patient subgroups suffering from either past micro- (seven patients), or macro-vascular complications (10 patients) as well as between these groups and the patients without apparent vascular disease. Likewise, the different therapies (metformin and/or sulfonylureas) did not affect the values of coagulative and oxidation parameters.

Bivariate analysis, adjusted for age, was performed to investigate associations between the biochemical parameters. A statistically significant association was found between PCP and APC (R = 0.788, P < 0.001), and between F1+2 and FpA (R = 0.234, P = 0.046). Urinary 8-iso-PGF and 11-dehydro-TXB2 were positively correlated (R = 0.665, P < 0.001), as previously reported in T2DM [2]. These associations are detailed in Table 2. It is noteworthy that urinary 8-iso-PGF and PCARB were positively correlated (R = 0.242, P = 0.039) suggesting an overall increase of OS, affecting both lipids and proteins, in diabetic patients. Moreover, both oxidation markers were positively correlated to F1+2 (Table 2 and Fig. 5). Interestingly, a positive correlation was found between blood glucose, HbA1c and 8-iso-PGF2α, whereas PCARB were associated to HbA1c only (Table 2).

Table 2.  Correlation coefficients and p values * of bivariate analysis of various parameters measured in diabetic patients (n = 72)
 8-isoPGFCarbonyls F1+211-Dehydro-TXB2HbA1cGlucose
  • *

    Two-tailed significance, adjusted for age.

8-isoPGFR = 0.242R = 0.472R = 0.66R = 0.231R = 0.359
P = 0.039P < 0.001P < 0.001P = 0.049P = 0.002
CarbonylsR = 0.242R = 0.281NSR = 0.235NS
P = 0.039P = 0.016 P = 0.045 
F1+2R = 0.472R = 0.281R = 0.251NSNS
P < 0.001P = 0.016P = 0.033  
HbA1cR = 0.231R = 0.235NSR = 0.231R = 0.597
P = 0.049P = 0.045 P = 0.049P < 0.001
GlucoseR = 0.359NSNSR = 0.287R = 0.597
P = 0.002  P = 0.014P < 0.001
Figure 5.

Correlation between urinary excretion of 8-isoPGF2a and plasma F1+2 levels in T2DM patients (n = 72). The straight line was drawn according to the best-fit linear regression parameters obtained with a bivariate analysis adjusted for age. The correlation coefficient and the P-value are reported in the plot.

Multiple regression analysis

On the basis of the results obtained in the bivariate analysis, a multiple regression analysis was also performed. In particular, a simple model was cast whereby the tendency to thrombin generation in DM patients, expressed by F1+2 levels, was analyzed as a function of (i) oxidation markers, that are 8-iso-PGF, and PCARB; (ii) metabolic control, expressed by HbA1c levels; and (iii) 11-dehydro-TXB2 generation by activated platelets.

Only age (P = 0.008) and 8-iso-PGF (P = 0.0013) were positively correlated to F1+2 levels (Table 3). A similar trend was found between F1+2 and PCARB, although the association was not statistically significant (P = 0.106).

Table 3.  Partial correlation coefficient (β) and p values of multiple regression analysis
XiYi → F1+2
Protein carbonyls0.171
P = 0.1060
P = 0.0013
P = 0.644
P = 0.459
P = 0.008


Enhanced lipid peroxidation is associated with persistent platelet activation in diabetic patients [2]. This study confirms and extends our previous findings. In fact, both lipid and protein oxidation were found significantly elevated in patients with T2DM. Plasma protein carbonyls provide a global index of protein oxidation involving the side chains of several amino acid residues. They represent the stable end-product generated upon formation of transient radical species, such as chloramines and nitrogen/carbon radicals, which are induced by oxidant stimuli [28,29]. However, direct oxidation of amino acid side chain is not the only way through which carbonyl groups can be formed in proteins. In fact, glycation may induce formation of protein carbonyls, such as ketoamine derivatives, thus generating reactive radicals and perpetuating a vicious cycle [30]. Thus, protein oxidation may be generated in part by nonenzymatic glycation processes, consistently with the significant association between HbA1c and PCARB found in the present study (Table 2). Moreover, a potential mechanism linking lipid and protein oxidation has been recently reported [31]. Thus, oxidized polyunsaturated fatty acids may form α,β-unsaturated aldehydes, that can attack lysyl side chains, contributing to the generation of stable protein carbonyl groups [31]. Such a mechanism is consistent with our finding of a significant relationship between protein and lipid oxidation in vivo in the setting of T2DM. Although both urinary 8-iso-PGF and PCARB were associated with F1+2 levels in univariate analysis, only the former was significantly associated with F1+2 in a multiple regression analysis. The strong association of F2-isoprostane formation with both 11-dehydro-TXB2 and F1+2 levels suggests that lipid peroxidation can affect platelet as well as coagulative activation, thus contributing to a pro-thrombotic state in this setting. This observation is consistent with the in vitro demonstration that oxidized lipid micelles enhance the prothrombinase activity, thus enhancing thrombin generation [7,8]. The present results confirm and extend these in vitro findings to the clinical setting of T2DM.

Diabetic patients also showed an impairment of anticoagulant mechanisms. Plasma soluble TM and APC/PCP levels were in fact decreased in diabetic subjects. It has to be remarked that the finding of reduced soluble TM levels was in apparent disagreement with previous reports concerning DM [32]. However, the mechanism(s) generating soluble TM in blood is unclear, and different levels of soluble TM have been reported in different clinical settings [33,34]. Generally, soluble TM level is thought to reflect the endothelial damage in peripheral atherosclerotic disorders [34]. However, in primary pulmonary hypertension, despite of a severe damage of pulmonary endothelium, a net decrease of soluble TM levels has been reported [35,36]. Infusion of prostacyclin was reported indeed to enhance TM levels in these patients [36]. Thus, TM levels may be considered the net resultant of multiple and complex phenomena, whose evolution may generate very different patterns of soluble TM levels. In our opinion, when the PC anticoagulant pathway has to be investigated, measurement of both soluble TM and APC/PCP levels should be performed at the same time to provide a coherent scenario. Moreover, the set of DM patients investigated in this study did not present signs of severe endothelial destruction, as revealed by normal values of VWF and fibrinogen levels as well. This may contribute to explain the apparent disagreement with previous results.

It has to be remarked that oxidizing agents, such as H2O2, oxidized LDL or homocysteine can down-regulate the activity and/or expression of TM on the surface of endothelial cells [17]. Furthermore oxidation of methionine 388 in the TM molecule, inactivates almost completely the anticoagulant function of the endothelial protein [15]. Thus, increased oxidation of circulating proteins may involve and impair TM function as well. Based on these considerations, oxidant stimuli might inhibit APC production by reducing TM concentrations on the endothelial surface, where PC activation takes place, and/or by chemical inactivation of the TM molecule. It may be of interest that two recent studies showed that regression of atherosclerosis in monkeys reduces also the vascular superoxide production, increases pulmonary TM activity and increases the anticoagulant response to thrombin in vivo[37,38]. The results obtained in the present study, showing a significant decrease of APC/PCP levels in T2DM patients under oxidative stress, may just represent the reversal of the medal, hinging on similar mechanisms.

An antioxidant pharmacological intervention aimed at investigating its effects on the anticoagulant PC pathway in this setting might be theoretically of clinical interest. However, recent studies raised many doubts on vitamin E supplementation, alone or in combination with other antioxidant agents like vitamin C and β-carotene, as protective agents against thrombotic cardiovascular diseases [39]. Accordingly, a recent pilot study on the effects of vitamin E supplementation on both protein oxidation and coagulation parameters in normal subjects showed a modest effect on F1+2 production only, while markers of the TM and APC/PCP pathway and protein oxidation remained unchanged [22]. The chemistry of biological oxidations is very complex and likely specific for a particular kind of chemical species. A molecule with antioxidant properties for lipids not necessarily shows the same degree of activity for protein oxidation. For instance, it is known that vitamin E is an inappropriate antioxidant against the superoxide anion, which can react directly and rapidly with proteins or nitric oxide, while it needs metal ions to peroxidize lipids [40]. Further studies are thus needed before designing potential intervention studies with antioxidant agents in the T2DM setting.

In conclusion, the present study demonstrates increased oxidation of both lipids and proteins in T2DM. The capacity of oxidized lipids to activate platelets may constitute a thrombotic trigger per se. Furthermore, oxidized lipids present on activated platelets can provide a better surface for the assembly and activation of prothrombinase complex. This may trigger a vicious cycle that, together with the oxidation-linked depression of the anticoagulant pathway, may predispose to a pro-thrombotic state in this setting. This complex scenario may explain the limited efficacy of conventional antiplatelet therapy with low-dose aspirin in diabetics [41], providing a rationale for trials of more aggressive antithrombotic therapy in this setting.


This study was supported by grants from MURST (ex-60% to RDC, ex-60% to GD, and to the Center of Excellence on Aging). We wish to thank Selene Fulvi, Marina Marinopiccoli, Stefania Maione, Erminia Di Bartolomeo, and Antonella Fragassi, for assistance with the clinical studies.

Note added in proof

After acceptance of the present manuscript, a paper published in Eur J Biochem showed the rate of oxidative deamination of lysine residue in plasma protein of diabetic rats, in agreement with the present findings on humans. (See Akagawa M, Sasaki T, Suyama K. Eur J Biochem 2002; 269: 5451–8.)