Type 1 Diabetes Is Associated With Increased Cyclooxygenase- and Cytokine-Mediated Inflammation

Subjects with type 1 diabetes, all attending the clinic of diabetes and endocrinology at the Uppsala University Hospital, were asked to participate in the study. A total of 38 diabetic subjects accepted, of whom 36 were using meal-related rapid-acting insulin combined with bedtime NPH insulin, and two were treated with insulin pumps. Their mean duration of diabetes was 16.1 ± 10.3 years. A total of 41 apparently healthy nondiabetic control subjects were recruited from the staff of the Uppsala University Hospital and among medical students. The participants were between 20 and 45 years of age. The two groups were well matched for sex, age, and BMI. Clinical characteristics are shown in Table 1. None of the included subjects was pregnant or had a history of malignancy, liver disease, untreated hypothyroidism, alcohol abuse, intercurrent diseases, or use of lipid-lowering drugs. According to a self-administrated questionnaire, eight of the diabetic subjects and four of the control subjects reported daily smoking of cigarettes. Three diabetic and five control subjects reported that they took multivitamin supplements regularly.

Macrovascular disease was defined as clinical history of angina or myocardial infarction. Microalbuminuria was defined as a urinary albumin excretion rate of 30–300 mg/24 h and macroalbuminuria as >300 mg/24 h, respectively, in more than one urine collection. Proliferative retinopathy was considered present if observed by fundus photography, which all diabetic subjects had undergone, or if there was a history of laser treatment. Peripheral neuropathy was defined as the loss of ankle reflexes and/or the loss of lower-limb sensation to light touch (monofilament) or vibration on clinical testing. Four subjects had signs of multiple microangiopathic complications, here defined as a clinical history of retinopathy, neuropathy, and/or micro- or macroalbuminuria. One type 1 diabetic subject had a history of microalbuminuria and one patient had a history of macroalbuminuria. Three of the subjects had a history of proliferative retinopathy. Six diabetic subjects were using either ACE inhibitors or an angiotensin receptor antagonist. Informed consent was obtained from all participating subjects, and the study was approved by the ethical committee at the Medical Faculty of Uppsala University.

Blood samples from all 79 participants were collected, after a 12-h overnight fast, into Vacutainer tubes containing Na-heparin, EDTA, or buffered sodium-citrate. HbA_1c was measured in fresh blood samples taken on the morning of the study.

The blood was stored on ice and thereafter centrifuged at 3,000 rpm for 10 min at 4 C°, and the plasma was separated. For determination of cholesterol and triglycerides, blood was collected into Vacutainer tubes without any addition and left to coagulate for 2 h. After low-speed centrifugation for 10 min, the serum was separated. One aliquot was stored in 4 C° before lipid analysis. A 24-h urine collection was delivered by 76 of the 79 participants at the time of the blood sampling. Urine samples were stored at −70°C until analysis for 15-keto-dihydro-PGF_2α and albumin were performed.

The 24-h urinary and plasma samples were analyzed for 15-keto-dihydro-PGF_2α by a radioimmunoassay at our laboratory, as previously described by Basu (14). In brief, the cross-reactivity of the antibody with PGF_2α; 15-keto-PGF_2α; PGE₂; 15-keto-13,14-dihydro-PGE₂; 8-iso-15-keto-13,14-dihydro-PGF_2α; 11β-PGF_2α; 9β-PGF_2α; thromboxane B₂; and 8-iso-PGF_3α was 0.02, 0.43, <0.001, 0.5, 1.7, <0.001, <0.001, <0.001, and 0.01%, respectively. The detection limit was 45 pmol/l. The levels of 15-keto-dihydro-PGF_2α in urine were adjusted for creatinine by division with the urinary creatinine concentration.

High-sensitivity CRP and SAA measurements were performed by latex-enhanced reagent (Dade Behring, Deerfield, IL), using a BN ProSpec analyzer (Dade Behring). The intra-assay coefficient of variation (CV) of the CRP method was 1.4% at both 1.23 and 5.49 mg/l, and the intra-assay CV of the SAA method was 5.9% at 12.8 mg/l and 3.2% at 81.7 mg/l.

High-sensitivity IL-6 was analyzed by an enzyme-linked immunosorbent assay kit (IL-6 HS; R&D Systems, Minneapolis, MN). Samples and standards were pipetted in a microtiter plate coated with monoclonal antibody against IL-6. After incubation and washing, the enzyme substrate solution was pipetted and followed by the addition of anti–IL-6 antibody. The color reaction was proportional to the bound IL-6. The total CV of the method was 7%, and the interassay CV was 5%.

Serum triglycerides and cholesterol were measured by enzymatic methods using IL Test Cholesterol Triander’s Method 181618-80 and IL Test Enzymatic-Colorimetric Method 181709-00 for use in the Monarch apparatus (Instrumentation Laboratory, Lexington, MA). Apolipoprotein B–containing lipoproteins were precipitated with a sodium phosphotungstate, and magnesium chloride solution (17) LDLs were calculated according to Friedewald et al. (18). Albumin in urine was analyzed by rate nephelometry on an Array protein system (Beckman Instruments, Brea, CA). The assay was performed according to the recommendations by the manufacturer. HbA_1c level was measured by automated high-performance liquid chromatography (reference range 3.5–5.0%) (19).

Unpaired Student’s t test was used to compare the two groups. Logarithmic transformations were applied on non–normally distributed parameters. The nonparametric Wilcoxon test was used on parameters, which were still not normally distributed after transformation. Linear relationships between variables were evaluated by Pearson’s correlation test. Spearman correlation was used with parameters not normally distributed.

The urinary albumin excretion rate (means ± SD) was significantly higher among the diabetic subjects than the control subjects (2.23 ± 5.83 vs. 0.32 ± 0.49 mg/h, P < 0.05) (Table 1). In a subgroup analysis excluding patients with blood pressure–lowering drugs (n = 6), type 1 diabetic patients had significantly higher urinary albumin excretion rates than control subjects (10.8 ± 11.5 vs. 0.56 ± 1.2 mg/h, P < 0.01). No differences were noted in serum lipid and lipoprotein levels between diabetic and control subjects, except for serum triglycerides, which were higher in the control group than in the type 1 diabetic group (0.82 ± 0.61 vs. 1.07 ± 0.54 mmol/l, P < 0.05) (Table 2). Exclusion of subjects who were smokers and/or subjects who were taking antioxidant supplements from both groups did not change any of the relationships between diabetic and control subjects, except for serum triglycerides, which showed no significant difference after the exclusion (0.82 ± 0.68 vs. 0.99 ± 0.53 mmol/l, NS).

Type 1 diabetic patients had significantly higher levels (P < 0.01) of urinary 15-keto-dihydro-PGF_2α than control subjects (Table 3). In a subgroup analysis excluding patients with blood pressure–lowering drugs (n = 6), type 1 diabetic patients had significantly higher levels of urinary 15-keto-dihydro-PGF_2α than control subjects (P = 0.02). Type 1 diabetic patients with blood pressure–lowering drugs had no difference in urinary 15-keto-dihydro-PGF_2α compared with the normotensive diabetic patients (310 ± 90 vs. 310 ± 110 pmol/mmol creatinine). There was no difference in the levels of 15-keto-dihydro-PGF_2α between the subjects who have or do not have microvascular complications (microalbuminuria) (data not shown).

Type 1 diabetic patients had significantly (P < 0.04) higher plasma levels of IL-6 than control subjects (Table 3). Neither CRP nor SAA differed significantly between the diabetic patients and the control subjects (Table 3). In a subgroup analysis excluding patients with blood pressure–lowering drugs (n = 6), type 1 diabetic patients had significantly higher levels of IL-6 than control subjects (P = 0.04). Type 1 diabetic patients with blood pressure–lowering drugs had no difference in the levels of IL-6, CRP, and SAA compared with the normotensive diabetic patients, and there was no difference in the levels of IL-6, CRP, and SAA between the subjects who have or do not have microvascular complications (microalbuminuria) (data not shown).

Urinary levels of 15-keto-dihydro-PGF_2α correlated with the degree of glycemic control, as reflected by HbA_1c (r = 0.42, P < 0.0005) (Fig. 1). Plasma levels of 15-keto-dihydro-PGF_2α were positively correlated with IL-6 (r = 0.24, P < 0.04). No correlation was found between the duration of diabetes and any measured inflammatory or metabolic parameters.

In this case-control study, we compared type 1 diabetic patients with nondiabetic control subjects. The group was relatively heterogeneous with regard to duration of diabetes and diabetes-related complications, representing a random sample of an average Swedish diabetic population. With few exceptions type 1 diabetic subjects attend hospital clinics, whereby the sample can be considered to be population based in the hospital catchment area (population 220,000).

This study shows that COX-mediated urinary 15-keto-dihydro-PGF_2α and plasma IL-6 levels were significantly higher in patients with type 1 diabetes compared with healthy control subjects. Plasma levels of 15-keto-dihydro-PGF_2α and IL-6 were positively correlated (r = 0.24, <0.04). Although this correlation is not very strong, it is of potential biological importance because the pathways of formation of PGF_2α and IL-6 are novel and different. Furthermore, there was a fair degree of correlation (r = 0.42, P < 0.0005) between urinary levels of 15-keto-dihydro-PGF_2α and HbA_1c. Previously, we did not find any significant association between the inflammatory biomarkers and HbA_1c levels in elderly men with type 2 diabetes (16). These subjects often suffer from long-lasting atherosclerotic macrovascular damage, in contrast to relatively young, well-controlled type 1 diabetic patients in this study without signs of any macrovascular complications (Table 1). Type 1 diabetic patients in this study do not show any signs of macrovascular disease, which is often related to inflammation, and it is reasonable to assume that it is easier to show a significant relationship between hyperglycemia and the levels of inflammatory biomarker PGF_2α metabolite in this group.

We did not find any difference in the inflammatory biomarkers in patients with microvascular complications compared with those who do not have microvascular complications. Furthermore, we did not find any difference in the inflammatory biomarkers in type 1 diabetic patients who were treated with blood pressure–lowering drugs compared with the normotensive patients. It is known that angiotensin II may stimulate IL-6 and other inflammatory biomarkers (20,21). Thus, it is possible that administration of angiotensin II blockers, as given to six patients, would reduce the levels of inflammatory biomarkers. This might explain why there is no significant difference in inflammatory biomarkers among blood pressure–treated and nontreated patients.

This study is the first to report increased levels of COX-mediated PGF_2α as measured by urinary 15-keto-dihydro-PGF_2α in type 1 diabetic patients compared with control subjects. Prostaglandins are well-known mediators of inflammation, and 15-keto-dihydro-PGF_2α, a major metabolite of bioactive PGF_2α, is a potent indicator of in vivo inflammatory processes (14,15,22,23). Thus, the results from this study suggest ongoing COX-mediated low-grade inflammatory processes among patients with type 1 diabetes. PGF_2α is shown to be a potent vasoconstrictive compound (11), and this might have further contributed to the pathogenesis of diabetes complications via decreased blood flow (24) and endothelium dysfunction (25). Recently, it has been shown that 11-dehydro-thromboxane B₂, a metabolite of thromboxane A₂ and a COX-mediated product primarily formed in the thrombocytes, is associated with type 1 diabetes (26). This might reflect that PGF_2α, being a major product of arachidonic acid through COX, could possibly induce platelet adhesion and aggregation, as has been demonstrated for an isomer of PGF_2α, 8-iso-PGF_2α (27,28). Furthermore, we have also shown that a significant correlation exists between the glycemic status as reflected by HbA_1c and the urinary levels of 15-keto-dihydro-PGF_2α.

This study also demonstrates enhanced levels of IL-6 in type 1 diabetic patients, which further indicates that type 1 diabetes is associated with an increased cytokine-mediated inflammatory response, which is corroborated by other studies (6,8). Furthermore, the levels of IL-6 correlated with COX-mediated PGF_2α metabolite. Cytokines have earlier been shown to stimulate prostaglandin formation in fibroblasts through induction of COXs (29). Both the cytokines and COXs are activated in the cells through the nuclear factor-κB pathway under inflammatory conditions elicited by various stress factors, such as reactive oxygen species and viral and bacterial infection (30). Together, these results indicate that the inflammatory mechanism might be responsible for enhanced atherogenesis and is possibly associated with common vascular complications among type 1 diabetic patients. Although these patients have an acceptable average glycemic status, it did not account for a lower inflammatory response, as reflected by increased levels of cytokines and cytokine- and COX-mediated products.

Acetylsalicylic acid is known to reduce cardiovascular risk through affecting inflammatory processes (2). Intake of low-dose aspirin or other anti-inflammatory drugs is considered to be effective as secondary prophylaxis of cardiovascular diseases and primary prophylaxsis of myocardial infarction (31). Aspirin inhibits COXs and thus the biosynthesis of parent prostaglandins and thromboxanes that are related to inflammation (12,32,33). Recently, it has also been shown that high doses of salicylates specifically inhibit the function of I-κ-kinase-β, a key protein involved in the regulation of inflammation that interferes with insulin signaling and contributes to both insulin resistance and diabetes (34). The current study shows that basal levels of PGF_2α and IL-6 are increased among type 1 diabetic patients, involving both COX and cytokine pathways. Thus, it is tempting to speculate that acetylsalicylic acid might have some effect in reducing the inflammatory condition in this patient group, but this has yet to be determined.

In conclusion, this study demonstrates an involvement of low-grade cytokine- and COX-mediated inflammation in type 1 diabetes. Furthermore, we found a significant correlation between the levels of PGF_2α metabolite and the degree of metabolic control, which suggests a potential link to fatty acid oxidation and insulin resistance. Thus, hyperglycemia is associated with increased production of these important inflammatory compounds. The association of these inflammatory biomarkers with type 1 diabetes indicates that an important association exists between type 1 diabetes and the future possibility of development of atherosclerosis.

This study received support from the Swedish Society of Medicine, the Foundation for Geriatric Research, the Thuréus Foundation, the Henning and Gösta Ankarstrand Foundation, and the Swedish Nutrition Foundation.

We are grateful to Barbro Simu, Siv Tengblad, and Eva Sejby for excellent technical assistance. Rawya Mohsen is acknowledged for statistical analyses.