This study shows that elevated levels of tPA, PAI-1, and tPA/PAI-1 complex precede incident diabetes after adjusting for metabolic and cardiovascular risk factors. The association of PAI-1 with incident T2DM was also detected in participants with normal blood glucose levels at baseline, indicating that fibrinolytic impairment occurs prior to increased glucose levels.
An earlier study of individuals with normal OGTT who were followed for 9 years revealed that the 15 participants who developed T2DM had significantly higher tPA antigen levels compared to referents after adjusting for factors associated with metabolic syndrome. In this larger study, we found an association between tPA and incident T2DM after adjusting for metabolic and cardiovascular risk factors. This association disappeared after additional adjustments were made for FPG and 2hPG. Diabetic patients have active coagulation and hypofibrinolysis induced by both chronic and acute hyperglycaemia[6, 7]. An experimental investigation has shown that high glucose levels increase the production of tPA antigen from human mesengial cells, suggesting that the tPA increase found in this study may be stimulated by glucose.
Both PAI-1 antigen and the tPA/PAI-1 complex were significantly associated with incident T2DM in all three multivariate models. This result is in agreement with other reports of a relationship between PAI-1 and incident diabetes[18, 19] independent of metabolic syndrome. Visceral adipose tissue has been shown to increase PAI-1 secretion. In a cross-sectional study, both tPA and PAI-1 activities and antigens were associated with metabolic syndrome parameters in patients with or without T2DM. However, the association found in this study was independent of BMI and other components of the metabolic syndrome.
In participants with normal glucose levels, only PAI-1 remained significantly associated with incident diabetes in the multivariate analysis. In participants with elevated glucose levels, only tPA and tPA/PAI-1 complex remained significantly associated with incident diabetes. This finding implies that tPA and PAI-1 play different roles in the pathophysiology of T2DM, with PAI-1 levels increasing early in the disease. As pre-diabetic changes develop and blood glucose levels start to increase, tPA and tPA/PAI-1 complex increase.
The influence of PAI-1 on the development of T2DM was discussed previously in a variety of contexts. Obesity has been shown to be associated with chronic inflammation. CRP has been shown to up-regulate the gene expression of PAI-1 in human aortic endothelial cells. Resident macrophages in the adipose tissue produce cytokines, such as tumour growth factor (TGF)-β and tumour necrosis factor (TNF)-α, that also up-regulate PAI-1[37, 38], indicating that the increase in PAI-1 occurs parallel to the obesity-induced development of T2DM. Therefore, the up-regulation of PAI-1 early in the development of T2DM may be seen as a result of the metabolic syndrome. Mouse models have raised the possibility that PAI-1 is not merely a product of obesity, but has a causal role in obesity and insulin resistance, as PAI-1 inhibition has the potential to reduce obesity and improve insulin sensitivity[39, 40], possibly via PAI-1 functioning as an integrator of cell signalling. In T2DM patients, PAI-1 antigen levels have been shown to decrease following metformin treatment. Weight reduction in obese individuals was also associated with a decrease in PAI-1 antigen levels, which is explained, in part, by the loss of PAI-1-secreting visceral adipose tissue but may also be affected by other PAI-1 functions.
A previous investigation reported no association between VWF and incident T2DM after adjusting for IL-6, adiponectin, and γ-glutamyl transferase. In our study, VWF was associated with T2DM in the univariate analysis and model 1, but this association was lost after further adjustments in model 2 for CRP, systolic blood pressure, and triglycerides. This result is in line with previous research showing that inflammatory markers can activate endothelial cells that produce VWF.
Strengths and limitations
The major advantage of this study is the prospective nested case-referent study design within a well-defined population-based intervention programme, the VIP. Men and women age 30 to 60 years underwent an extensive baseline health examination regarding cardiovascular risk factors at their local health centre. Diabetic patients were defined by FPG and OGTT. Participants who developed T2DM were examined years before the manifest diagnosis.
The number of missing cases is high, as 80 of the original 237 cases did not have blood samples available for laboratory analysis of the haemostatic variables. Baseline data for 60 cases without blood samples available showed that no significant difference existed in diastolic and systolic blood pressure, FPG levels, or BMI compared to the cases in the study. As data on the cases without blood samples was similar to the data on those included in the study, selection bias is unlikely. Lifestyle changes recommended in the initial health examination, such as weight loss and smoking cessation, may have influenced the results. In a 10-year follow-up study, VIP participants with baseline glucose intolerance or T2DM were the most likely to avoid weight gain, but this occurrence most likely affected the incidence of T2DM and not the relationship between baseline fibrinolytic data and the risk of developing diabetes in a given individual. The vast majority of participants were Caucasian, limiting the possibility of generalising the results to other ethnicities. A risk of regression dilution bias exists because only a single baseline blood sample was used, which may decrease the estimations of the observed associations.