Tissue Plasminogen Activator, Von Willebrand Factor, and Risk of Type 2 Diabetes in Older Men

Cardiovascular and Metabolic Risk ORIGINAL ARTICLE

Tissue Plasminogen Activator, von Willebrand Factor, and Risk of Type 2 Diabetes in Older Men

1 3 S. GOYA WANNAMETHEE, PHD PETER H. WHINCUP, FRCP predict incident diabetes (4–9), although 2 1 NAVEED SATTAR, MD LUCY LENNON, MSC the findings have not always been consis- 2 2 ANN RUMLEY, PHD GORDON D.O. LOWE, DSC tent. Some studies have shown the above associations to be independent of inflammation and insulin resistance (6,7). Adi- OBJECTIVE — The objective of this study was to assess the relationship between putative pokines, in particular low adiponectin, markers of endothelial dysfunction (tissue plasminogen activator [t-PA] antigen and von Wille- and hepatic function as measured by the brand factor [vWF] antigen) and development of type 2 diabetes, as well as the role of inflam- hepatic enzymes alanine aminotransfer- mation, adipokines, hepatic function, and insulin resistance in modifying these relationships. ase (ALT) and ␥-glutamyl transferase (GGT) have been associated with risk of RESEARCH DESIGN AND METHODS — This was a prospective study of 3,562 non- subsequent type 2 diabetes (10–13). Adi- diabetic men aged 60–79 years followed up for an average of 7 years during which there were 162 incident cases of type 2 diabetes. pokines and liver dysfunction have also been linked to PAI-1 and markers of en- RESULTS — Elevated t-PA (top third) was associated with a near threefold increase in risk of dothelial dysfunction (14–16). Few pop- diabetes compared with the risk in those in the bottom third after adjustment for lifestyle factors ulation studies have addressed the Ͻ and waist circumference (relative risk [RR] 2.98 [95%CI 1.79–5.00]; Ptrend 0.0001); weaker associations between liver dysfunction but significant (marginal) associations were seen with vWF (1.24 [0.83–1.85]; P ϭ 0.05 for and markers of endothelial dysfunction. trend). Both biomarkers of endothelial dysfunction correlated significantly with markers of Whether the association between markers ␥ inflammation (interleukin-6 [IL-6] and C-reactive protein [CRP]), hepatic function ( -glutamyl of endothelial dysfunction and diabetes is transferase [GGT]), and insulin resistance, with t-PA showing stronger associations with adipos- independent of adipokines and hepatic ity, hepatic function, and insulin resistance than vWF. t-PA was also significantly and inversely associated with adiponectin. Adjustment for IL-6, adiponectin, and GGT attenuated the associ- function has not been studied, but it is ation of incident diabetes with vWF (1.06 [0.71–1.60]), but the relationship seen with t-PA important to do so to assess possible re- remained significant (adjusted RR 2.19 [1.29–3.70]). Subsequent adjustment for insulin atten- sidual confounding. uated the association further, but t-PA was still associated with a significant increase in risk (1.66 Tissue plasminogen activator (t-PA) is ϭ [0.96–2.85]; Ptrend 0.02). released from vascular endothelium; hence, circulating levels of t-PA antigen CONCLUSION — t-PA antigen, but not vWF antigen, is independently associated with risk may be a marker of endothelial dysfunc- of type 2 diabetes. tion (17). However, a higher plasma t-PA antigen level represents largely inactive Diabetes Care 31:995–1000, 2008 circulating t-PA–PAI-1 complexes, which, in turn, may reflect both endothe- ndothelial dysfunction plays a piv- tance, although the relationship is almost lial disturbance (t-PA and PAI-1 release) otal role in the development of ath- certainly bidirectional (3). A number of and hepatic PAI-1 release (17). An ele- E erosclerosis (1) and may be one of prospective studies have shown that cir- vated t-PA antigen level is considered to the underlying causes (or “common soil”) culating biomarkers of endothelial dys- be an integral feature of the insulin resis- of both coronary heart disease (CHD) and function such as E-selectin, intercellular tance syndrome and is also related to the type 2 diabetes (2). There is substantial adhesion molecule 1, von Willebrand fac- inflammatory response (18). The results evidence suggesting a potential role for tor (vWF), and, in particular, plasmino- of one population study suggested that endothelial dysfunction in insulin resis- gen activator inhibitor type 1 (PAI-1) t-PA is predictive of future diabetes inde- ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● pendent of the metabolic syndrome (19), From the 1Department of Primary Care and Population Sciences, Royal Free and University College Medical although this study was limited in the School, London, U.K.; the 2Division of Cardiovascular and Medical Sciences, University of Glasgow, Royal small number of incident cases, and the 3 Infirmary, Glasgow, U.K.; and the Department of Community Health Sciences, St. George’s, University of findings were not statistically significant. London, London, U.K. Corresponding author: Dr. S. Goya Wannamethee, Department of Primary Care and Population Sciences, This question is worthy of study because Royal Free and University College Medical School, Rowland Hill St., London NW3 2PF, U.K. E-mail: t-PA is more stable than PAI-1 and also [email protected]. appears to be more strongly linked to in- Received for publication 9 August 2007 and accepted in revised form 20 January 2008. cident CHD events in population cohorts Published ahead of print at http://care.diabetesjournals.org on 30 January 2008. DOI: 10.2337/dc07- 1569. (17). We have therefore examined the re- Abbreviations: ALT, alanine transaminase; CHD, coronary heart disease; CRP, C-reactive protein; ELISA, lationships between t-PA and vWF (two enzyme-linked immunosorbent assay; GGT, ␥-glutamyl transferase; HOMA-IR, homeostasis model assess- circulating markers of endothelial dys- ment of insulin resistance; IL, interleukin; PAI-1; plasminogen activator inhibitor type 1; t-PA, tissue plas- function) and the risk of incident type 2 minogen activator antigen; vWF, von Willebrand factor. diabetes and have assessed whether these © 2008 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby associations are independent of adi- marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ponectin, hepatic function as measured

DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008 995 t-PA, vWF, and risk of type 2 diabetes in men by ALT and GGT levels, inflammation as and IV and V (manual groups). Those Health Service registers. Evidence regard- measured by levels of C-reactive protein whose longest occupation was in the ing diabetes was obtained by reports from (CRP) and interleukin-6 (IL-6), and insu- armed forces formed a separate group. general practitioners with biennial re- lin resistance. Heavy drinking was defined as Ͼ5 views of the patients’ notes (including drinks/day. On the basis of a physical ac- hospital and clinic correspondence) RESEARCH DESIGN AND tivity score (21), the men were classified through to the end of the study period. METHODS — The British Regional into four groups: inactive, light, moder- Cases of diabetes are based on self- Heart Study is a prospective study of car- ate, and moderately vigorous/vigorous. reported diagnoses confirmed by primary diovascular disease involving 7,735 men Plasma glucose was measured by a glu- care records, an approach that has been aged 40–59 years selected from the age- cose oxidase method using a Falcor 600 validated in the present study (25). sex registers of one general practice in automated analyzer. Serum insulin was each of 24 British towns, who were measured using an enzyme-linked immu- Statistical methods screened between 1978 and 1980 (20). In nosorbent assay (ELISA) that does not The men were divided by tertiles of the 1998–2000, all surviving men, now aged cross-react with proinsulin. Triglycerides, inflammatory markers. The Cox propor- 60–79 years, were invited for a 20th year blood glucose, and insulin concentrations tional hazards model was used to assess follow-up examination. All relevant local were adjusted for the effects of fasting du- the multivariate-adjusted relative risk for research ethics committees provided ap- ration and time of day (22). Insulin resis- each third compared with the reference proval. All men provided informed writ- tance was estimated according to group (lowest third). In the adjustment, ten consent for the investigation, which homeostasis model assessment (HOMA- smoking (never, long-term ex-smokers was performed in accordance with the IR) (the product of fasting glucose [milli- [Ͼ15 years], recent ex-smokers, and cur- Declaration of Helsinki. They completed moles per liter] and insulin [units per rent smokers), social class (seven groups), a questionnaire (Q20) that included ques- milliliter] divided by the constant 22.5) physical activity (four groups), alcohol in- tions on their medical history, lifestyle be- (23). Plasma levels of t-PA antigen were take (five groups), parental history of di- havior, and family history of diabetes. The measured with an ELISA (Biopool AB, abetes (yes or no), preexisting CHD (yes men were requested to fast for a minimum Umea, Sweden), as was vWF antigen or no) and stroke (yes or no), and use of of 6 h, during which time they were in- (DAKO, High Wycombe, U.K.) (21). CRP statins (yes or no) were fitted as categori- structed to drink only water and then to was assayed by ultrasensitive nephelome- cal variables. Waist circumference, present for measurement at a specified try (Dade Behring, Milton Keynes, U.K.). HOMA-IR, GGT, adiponectin, IL-6, and time between 0800 and 1800 h. All men Plasma adiponectin concentrations were CRP were fitted as continuous variables. were asked to provide a blood sample, determined using an ELISA (R&D Sys- collected using the Sarstedt Monovette tems, Oxford, U.K.) (10). IL-6 was as- RESULTS — During the mean fol- system. A total of 4,252 men (77% of sur- sayed using a high-sensitivity ELISA low-up period of 7 years (range 5–7 years) vivors) presented for examination. (R&D Systems) (10). Hepatic enzymes in- there were 162 incident cases of diabetes cluding GGT and ALT were measured us- in the 3,562 nondiabetic men. Table 1 Cardiovascular risk factors ing a Hitachi 747 automated analyzer. shows the baseline characteristics in the Details of measurement and classification men who developed diabetes and in men methods for smoking status, physical ac- Study subjects who remained free of diabetes. Men who tivity, BMI, social class, alcohol intake, Endothelial marker measurements (t-PA developed diabetes had higher BMI and blood pressure, and blood lipids in this or vWF) were available for 4,083 men at waist circumference than those who did cohort have been described (20–22). An- Q20. Men with a doctor’s diagnosis of di- not. They were also more likely to be thropometric measurements including abetes, men with a diagnosis of diabetes in physically inactive, to have a parental his- body weight, height, and waist circumfer- the year of reexamination, and those with tory of diabetes, to have a higher preva- ence were obtained. Subjects were mea- a fasting glucose of Ͼ7 mmol/l (World lence of CHD, to have higher levels of sured in light clothing without shoes in Health Organization criteria) were con- metabolic risk factors, and to have signif- the standing position. BMI (weight in ki- sidered to have prevalent diabetes and icantly higher levels of markers of inflam- lograms divided by height in meters were excluded (n ϭ 484). We further ex- mation (CRP and IL-6), endothelial squared) was calculated for each man. cluded men with missing data for IL-6 dysfunction markers (vWF and t-PA), From the combined information at the (n ϭ 37). Thus, analysis is based on 3,562 and hepatic enzymes (GGT and ALT) and initial screening and follow-up question- men. lower adiponectin levels than men who naires in 1996 (Q96) and at rescreening did not develop diabetes. (Q20), the men were classified into five Follow-up Table 2 shows the correlations be- cigarette-smoking groups: 1) those who All men were followed up for all-cause tween endothelial dysfunction markers had never smoked; 2) ex-smokers since mortality, cardiovascular morbidity, and and their correlations with metabolic risk screening; 3) smokers at baseline who development of type 2 diabetes from the factors, inflammation, adiponectin, and gave up between screening and Q96; 4) initial examination to July 2006 (24), and hepatic enzymes. t-PA antigen was smokers at baseline and at Q96 who gave follow-up was achieved for 99% of the strongly correlated with central adiposity up after 1996; and 5) current cigarette cohort. This analysis is based on fol- (waist circumference) and was signifi- smokers at Q20. The longest held occu- low-up from rescreening in 1998–2000, cantly associated with all components of pation of each man was recorded at a mean follow-up period of 7 years (6–8 the metabolic syndrome, insulin resis- screening, and the men were grouped into years). Information on deaths was col- tance, adiponectin, and hepatic enzymes. one of six social classes: I, II, or III non- lected through the established tagging These associations persisted even after ad- manual (nonmanual groups); III manual; procedures provided by the National justment for waist circumference (Table

996 DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008 Wannamethee and Associates

Table 1—Distribution of risk factors and inflammatory/hemostatic markers in 3,562 nondi- GGT ϩ HOMA-IR), elevated t-PA was abetic subjects aged 60–79 years at reexamination according to diabetes status at follow-up still associated with a 60% increase in risk. Developed diabetes The increased risk of diabetes associ- P ated with elevated t-PA was seen in both No Yes for difference obese and centrally obese (waist circum- Ͼ Ͼ 2 n 3,400 162 ference 102 cm or BMI 30 kg/m ) and nonobese (waist Յ102 cm) men and in Age (years) 68.7 (5.6) 68.4 (5.4) 0.56 Ͻ BMI (kg/m2) 26.6 (3.5) 29.7 (4.1) Ͻ0.0001 those with normal glucose levels ( 6.1 Ͻ mmol/l) and impaired glucose levels Waist circumference (cm) 96.2 (9.9) 104.1 (10.4) 0.0001 Ͼ Current smokers (%) 13.3 9.3 0.14 ( 6.1 mmol/l) (data not shown). No sig- Inactive (%) 32.4 45.2 0.0008 nificant interaction was seen between obesity or glucose levels and t-PA and risk Manual (%) 53.1 64.6 0.004 ϭ ϭ Heavy drinkers (Ͼ5 drinks/day) (%) 3.7 2.5 0.42 of diabetes (P 0.41 and P 0.50 for Parental history of diabetes (%) 5.4 9.9 Ͻ0.0001 obesity and high glucose, respectively). CHD (%) 17.6 32.1 Ͻ0.0001 Use of statins (%) 6.5 13.6 0.0005 CONCLUSIONS — In this large ran- Stroke (%) 5.0 6.2 0.57 dom sample of nondiabetic British men SBP (mmHg) 148.2 Ϯ 23.9 153.0 Ϯ 22.7 0.01 aged 60–79 years, we have confirmed the Triglyceride (mmol/l) 1.57 (1.12–2.13) 2.10 (1.54–2.81) Ͻ0.0001 findings of several previous studies (4–9) HDL (mmol/l) 1.34 Ϯ 0.34 1.17 Ϯ 0.29 Ͻ0.0001 that circulating markers of endothelial Glucose (mmol/l) 5.52 (5.21–5.89) 5.99 (5.59–6.52) Ͻ0.0001 dysfunction are associated with incident Log HOMA-IR 0.65 Ϯ 0.58 1.27 Ϯ 0.57 Ͻ0.0001 diabetes. We have shown that t-PA anti- CRP (mg/l) 1.67 (0.80–3.30) 2.59 (1.27–4.42) Ͻ0.0001 gen, but not vWF antigen, is indepen- IL-6 (pg/ml) 2.41 (1.55–3.42) 2.92 (1.93–4.23) Ͻ0.0001 dently associated with the development of Adiponectin (␮g/ml) 7.11 (4.52–11.46) 4.96 (3.11–7.39) Ͻ0.0001 diabetes in men. Our findings extend GGT (IU/l) 27.4 (18–37) 37.0 (24–52.5) Ͻ0.0001 those of other studies on circulating en- ALT (IU/l) 15.2 (12–20) 19.5 (13.5–28.0) Ͻ0.0001 dothelial markers and risk of type 2 dia- vWF (IU/dl) 137.8 Ϯ 44.4 146.4 Ϯ 48.1 0.02 betes by assessing a wider range of risk t-PA (ng/ml) 10.74 Ϯ 4.21 13.37 Ϯ 4.79 Ͻ0.0001 factors and adjusting for correlates reflec- tive of perturbances in other relevant Data are means Ϯ SD or geometric means (interquartile range) for skewed variables. SBP, systolic blood pressure. pathways including markers of inflammation and insulin resistance as well as adi- 2). vWF showed weaker but significant Ͼ2-fold increase in risk of diabetes even ponectin and hepatic enzymes, which associations with waist circumference, after adjustment for demographic factors have not previously been assessed or hepatic enzymes, and insulin resistance. and waist circumference (model 2). Waist comprehensively adjusted for. The asso- Less consistent associations were seen for circumference and BMI are highly corre- ciation between vWF and diabetes was the metabolic risk factors. All showed lated (r ϭ 0.86). The findings were virtu- largely dependent on inflammation. Ad- similar significant associations with in- ally unchanged if BMI was included in the justment for IL-6 attenuated the associa- flammatory markers (CRP and IL-6). adjustment instead of waist circumfer- tion between vWF and risk of diabetes. In Table 3 shows the incidence rates for ence. A weaker association was seen be- contrast, the association between t-PA an- diabetes and the relative risk of diabetes tween vWF and diabetes. To further tigen and diabetes was independent of by tertiles of the biomarkers of endothe- assess whether the relationship between markers of adiposity or inflammation, lial dysfunction with adjustment for de- diabetes and markers of endothelial dys- adiponectin, and hepatic enzymes, demographic factors (age, social class, function was independent of inflamma- spite being correlated with markers in smoking, physical activity, alcohol in- tion, adipocytes, and hepatic enzymes, we each pathway. The increased risk associ- take, parental history of diabetes, preex- adjusted in turn for adiponectin, IL-6, ated with elevated t-PA partially reflected isting CHD, and use of statins) and the and GGT. Simultaneous adjustment for insulin resistance, but there still remained effects of adjustment for waist circumfer- these factors attenuated the associations, a 60% increase in risk even after adjust- ence, inflammation (IL-6), adiponectin, but the increased risk of diabetes associ- ment for HOMA-IR. We considered vWF hepatic enzymes (GGT), and insulin resis- ated with elevated t-PA remained signifi- and t-PA to reflect endothelial dysfunc- tance (HOMA-IR). IL-6 was used in the cant. No association was seen with vWF tion; however, the modest correlation be- adjustment, as IL-6 and CRP are strongly after adjustment. tween these two biomarkers (r ϭ 0.21) correlated (r ϭ 0.57), and CRP showed Insulin resistance could mediate the and the differing associations seen be- no independent association with diabetes association between t-PA and risk of dia- tween t-PA and vWF with insulin resis- after adjustment for IL-6. Similarly GGT betes. Adjustment for HOMA-IR in addi- tance and features of the metabolic and ALT are strongly correlated (r ϭ tion to IL-6 and adiponectin (model 3) syndrome support the view that these 0.43), and GGT was a stronger predictor considerably attenuated the relationships endothelial biomarkers are regulated dif- of diabetes than ALT (17) and showed for t-PA, but it remained significantly as- ferently. Therefore, one should be cau- stronger associations with the endothelial sociated with increased risk of diabetes. tious in labeling such markers equally or dysfunction markers. When GGT and HOMA-IR were both in- purely under the endothelial function Elevated t-PA was associated with a cluded in the adjustment (model 3 ϩ umbrella.

DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008 997 t-PA, vWF, and risk of type 2 diabetes in men

Table 2—Spearman correlation coefﬁcients of endothelial markers with metabolic and anthropometric variables, inﬂammatory markers, liver enzymes, and adiponectin

t-PA vWF Unadjusted Adjusted for WC Unadjusted Adjusted for WC rPrPrP r P t-PA 1.00 0.21 Ͻ0.0001 0.20 Ͻ0.0001 vWF 0.21 Ͻ0.0001 0.20 Ͻ0.0001 1.00 Metabolic factors WC 0.31 Ͻ0.0001 0.06 Ͻ0.001 HDL cholesterol Ϫ0.20 Ͻ0.0001 Ϫ0.12 Ͻ0.0001 Ϫ0.07 Ͻ0.001 Ϫ0.04 Ͻ0.01 Triglycerides 0.34 Ͻ0.0001 0.26 Ͻ0.0001 0.03 NS 0.02 NS SBP 0.09 Ͻ0.0001 0.07 Ͻ0.0001 Ϫ0.01 NS Ϫ0.02 NS Glucose 0.13 Ͻ0.0001 0.11 Ͻ0.0001 0.04 NS 0.03 NS HOMA-IR 0.35 Ͻ0.0001 0.25 Ͻ0.0001 0.11 Ͻ0.0001 0.10 Ͻ0.01 Inﬂammatory markers and adiponectin CRP 0.25 Ͻ0.0001 0.20 Ͻ0.0001 0.26 Ͻ0.0001 0.25 Ͻ0.0001 IL-6 0.20 Ͻ0.0001 0.16 Ͻ0.0001 0.25 Ͻ0.0001 0.24 Ͻ0.0001 Adiponectin Ϫ0.15 Ͻ0.0001 Ϫ0.12 Ͻ0.0001 0.06 Ͻ0.001 0.07 Ͻ0.001 Liver enzymes ALT 0.17 Ͻ0.0001 0.11 Ͻ0.0001 0.005 NS Ϫ0.005 NS GGT 0.31 Ͻ0.0001 0.27 Ͻ0.0001 0.13 Ͻ0.0001 0.12 Ͻ0.0001 SBP, systolic blood pressure; WC, waist circumference.

vWF is produced mainly by vascular men, suggest that the association between t-PA but also circulating complexes of endothelial cells activated by proinflam- vWF and diabetes is explained by levels of t-PA with its major inhibitor, PAI-1 (17), matory cytokines (26). Although vWF the proinflammatory cytokine IL-6, which is synthesized by adipocytes, hepa- was shown to be predictive of diabetes which is known to be associated with in- tocytes, and endothelial cells and is also independent of a marker of inflammation cident diabetes (10) and which stimulates an acute-phase reactant. t-PA (and PAI-1) (CRP) and insulin in the Framingham release of vWF from vascular endothe- levels increase as part of the inflammatory Offspring Study (7), others have reported lium (28). Thus, any link between vWF response (29), and in the present study no association (8) or no independent as- and incident diabetes may be indirect t-PA correlated with several inflammatory sociation between vWF and diabetes (27). rather than direct. markers including CRP and IL-6. Numer- Our results, generated in a more homoge- Circulating t-PA antigen reflects not ous studies have shown PAI-1 to be pre- neous population of predominantly white only endothelial synthesis and release of dictive of type 2 diabetes (4,6–8) and, in

Table 3—Incidence rates and adjusted relative risk of type 2 diabetes according to thirds of inﬂammatory and endothelial markers

Thirds

Endothelial markers Lowest Middle Top Ptrend t-PA (ng/ml) Ͻ8.7 8.7 12.1 Rate/1,000 person-years 2.6 6.3 11.6 Model 1 1.00 2.36 (1.40–3.99) 4.54 (2.76–7.47) Ͻ0.0001 Model 2 1.00 1.89 (1.12–3.20) 2.98 (1.79–5.00) Ͻ0.0001 Model 3 1.00 1.81 (1.07–3.07) 2.65 (1.58–4.45) Ͻ0.0001 Model 3 ϩ GGT 1.00 1.65 (0.97–2.81) 2.19 (1.29–3.70) 0.0003 Model 3 ϩ HOMA-IR 1.00 1.35 (0.78–2.32) 1.92 (1.12–3.27) 0.001 Model 3 ϩ GGT ϩ HOMA-IR 1.00 1.25 (0.72–2.16) 1.66 (0.96–2.85) 0.02 vWF (IU/dl) Ͻ115 115 155 Rate/1,000 person-years 5.8 6.1 8.4 Model 1 1.00 1.04 (0.70–1.56) 1.39 (0.94–2.03) 0.007 Model 2 1.00 0.94 (0.62–1.41) 1.24 (0.83–1.85) 0.05 Model 3 1.00 0.91 (0.61–1.38) 1.18 (0.79–1.78) 0.11 Model 3 ϩ GGT 1.00 0.87 (0.57–1.31) 1.06 (0.71–1.60) 0.31 Model 3 ϩ HOMA-IR 1.00 0.87 (0.57–1.32) 1.10 (0.73–1.66) 0.27 Model 3 ϩ GGT ϩ HOMA-IR 1.00 0.84 (0.55–1.28) 1.01 (0.67–1.53) 0.49 Data are relative risk (95% CI). Model 1, adjusted for age, social class, smoking, alcohol intake, physical activity, parental history of diabetes, preexisting CHD, stroke, and use of statins; model 2, model 1 plus waist circumference; and model 3, model 2 plus adiponectin and IL-6.

998 DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008 Wannamethee and Associates some, to be independent of adiposity and predicted diabetes is consistent with a Acknowledgments— The British Regional markers of inflammation (6,7). Similar re- much smaller previous report of t-PA in Heart Study is a research group supported by lationships were seen for t-PA in this the Northern Sweden Monitoring of the British Heart Foundation. The measure- study, as one might expect, given its Trends and Determinants in Cardiovas- ments and laboratory analyses reported here strong correlation with PA1-I (26). In- cular Disease (MONICA) Study (19), were supported by British Heart Foundation flammation and adiponectin appeared to which also considered only a limited Project grants PG97012 and PG97027. have only modest attenuating effects on number of potential confounders. Thus, We thank Karen Craig, Estelle Poorhang, the relationship between t-PA and diabe- Paul Welsh, and Fiona Key for technical our work extends the available data on the support. tes. In our previous report we have shown link between t-PA and incident diabetes. that IL-6 (positively) and adiponectin (in- Our study is not without some lim- versely) are predictive of diabetes (11). itations. It was performed in an older, Because t-PA levels correlated with IL-6 References predominantly white Caucasian male and adiponectin, t-PA may explain the 1. Ross R: Atherosclerosis—an inflamma- positive relationship between IL-6 and di- population, and we cannot generalize tory disease. N Engl J Med 340:115–126, abetes seen in this and other studies and our findings to women, younger men, 1999 the significant inverse association seen or other ethnic groups, although, as 2. Caballero AE: Endothelial dysfunction in between adiponectin and diabetes in noted above, t-PA has been shown to be obesity and insulin resistance: a road to diabetes and heart disease. Obes Res 11: obese men (11). However, these relation- strongly correlated with measured insulin resistance in women with polycystic 1278–1289, 2003 ships remained significant after adjusting 3. Sjoholm A, Nystrom T: Endothelial in- for t-PA. Thus, the associations between ovary syndrome and control subjects flammation in insulin resistance. Lancet t-PA, adipokines (IL-6 and adiponectin), (31), suggesting that the t-PA–diabetes 365:610–612, 2005 and diabetes appear to be independent of relationship is likely to be seen in 4. Festa A, D’Agostino R, Tracy RP, Haffner each other. women as well. Diabetes incidence in SM: Elevated levels of acute-phase pro- Insulin has been shown to stimulate this study was determined by docu- teins and plasminogen activator inhibi- expression of PAI-1 and t-PA (30). t-PA mented doctor-diagnosed cases of dia- tor-1 predict the development of type 2 was associated with insulin resistance and betes, which would inevitably result in diabetes: the Insulin Resistance Athero- all components of the metabolic syn- sclerosis Study. Diabetes 51:1131–1137, underascertainment of cases. It may be 2002 drome, and the relationship between t-PA argued that those with elevated markers 5. Meigs J, Hu FB, Rifai N, Manson JE: Bi- and diabetes was to some extent ex- of endothelial dysfunction are more omarkers of endothelial dysfunction and plained by insulin resistance, but there likely to be obese and to have had con- risk of type 2 diabetes mellitus. JAMA remained a significant independent asso- tact with their general practitioner and 291:1978–1986, 2004 ciation after adjustment for HOMA-IR. thus are more likely to have had diag- 6. Festa A, Williams K, Tracy RP, Wagen- knecht LE, Haffner SM: Progression of Clearly, more direct measures of insulin nostic tests for diabetes. However, resistance, e.g., clamping, may have at- plasminogen activator inhibitor-1 and fi- among the prevalent diabetic subjects at tenuated the relationship between t-PA brinogen levels in relation to incident type rescreening who had been excluded, and incident diabetes further. Indeed, we 2 diabetes. Circulation 113:1753–1759, have previously reported strong associa- t-PA and vWF were positively associ- 2006 7. Meigs JB, O’Donnell CJ, Tofler GH, Ben- tions between clamp-measured insulin ated with both diagnosed diabetes (phy- sician diagnosis) and undiagnosed jamin EJ, Fox CS, Lipinska I, Nathan DM, resistance and the t-PA antigen concen- Sullivan LM, D’Agostino RB, Wilson PW: tration in women with polycystic ovary diabetes (using fasting glucose concen- Ͼ Hemostatic markers of endothelial dys- syndrome (r ϭϪ0.59; P Ͻ 0.05) and tration criteria of 7.0 mmol/l). Thus, function and risk of incident type 2 dia- healthy control subjects (r ϭϪ0.62; P Ͻ the increased risk associated with ele- betes: the Framingham Offspring Study. 0.05) (31). Thus, t-PA antigen concentra- vated markers of endothelial dysfunc- Diabetes 2:530–537, 2006 tions, partly by virtue of strong correlation is unlikely to be due to bias in 8. Thorand B, Baumert J, Chambless L, tions with insulin resistance, may be ascertainment. Finally we did not mea- Meisinger C, Kolb H, Doring A, Lowel H, useful as part of a multiple biomarker sure PAI-1 directly, but note that t-PA is Koenig W, MONICA/KORA Study portfolio to predict future diabetes risk. the more stable marker (33). Group: Elevated markers of endothelial dysfunction predict type 2 diabetes mel- The association of t-PA and diabetes Circulating levels of t-PA (17) and litus in middle-aged men and women may also partially occur via common as- vWF (26) are associated with risk of from the general population. Arterioscler sociations with liver disturbance. Indeed, CHD, and their elevations in individuals Thromb Vasc Biol 26:398–405, 2006 metformin, which is thought to work with type 2 diabetes (34) might therefore 9. Song Y, Manson JE, Tinker L, Rifai N, principally at the liver, has been consis- promote their atherothrombotic compli- Cook NR, Hu FB, Hotamisligil GS, Ridker tently shown to lower t-PA concentra- cations. The present study adds to the lit- PM, Rodriguez BL, Margolis KL, Oberman A, Liu S: circulating levels of endothelial tions in different groups and to do so in erature on the potential importance of correlation with improvements in insulin adhesion molecules and risk of diabetes t-PA in the development of diabetes. We mellitus in an ethnically diverse cohort of resistance measures (32). The association suggest that further detailed studies of en- between t-PA and diabetes was attenuated women. Diabetes 56:1898–1904, 2007 dothelial markers (in particular t-PA) and 10. Spranger J, Kroke A, Mohlig M, Bergmann further but not eliminated after inclusion risk of diabetes be performed to clarify MM, Ristow M, Boeing H, Pfeiffer AF: of both insulin resistance and hepatic Adiponectin and protection against type 2 function in the adjustment. Prospective their value in prediction of diabetes, ide- ally in combination with other markers, diabetes mellitus. Lancet 361:226–228, studies relating t-PA to diabetes are ex- 2003 tremely limited. Our finding that t-PA and their possible roles in pathogenesis. 11. Wannamethee SG, Lowe GDO, Rumley

DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008 999 t-PA, vWF, and risk of type 2 diabetes in men

A, Cherry L, Whincup PH, Sattar N: Adi- tor-1, inflammation, obesity, insulin tigators: Factor VIII and other hemostasis pokines and risk of type 2 diabetes in resistance and vascular risk. J Thromb variables are related to incident diabetes older men. Diabetes Care 30:1200–1205, Haemost 1:1575–1579, 2003 in adults: The Atherosclerosis Risk in 2007 19. Eliasson MCE, Jansson JH, Lindahl B, Communities (ARIC) Study. Diabetes 12. Sattar N, Scherbakova O, Ford I, O’Reilly Stegmayr B: High levels of tissue plasmin- Care 22:767–772, 1999 D, Stanley A, Forrest E, Macfarlane PW, ogen activator (tPA) antigen precede the 28. Kerr R, Stirling D, Ludlam CA: Interleu- Packard CJ, Cobbe SM, Shepherd J; West development of type 2 diabetes in a lon- kin 6 and haemostasis. Br J Haematol 115: of Scotland Coronary Prevention Study: gitudinal population study: the Northern 3–12, 2001 Elevated ALT predicts new onset type 2 Sweden Monica Study. Cardiovasc Diabe- 29. Juhan-Vague I, Alessi MC: Fibrinolysis diabetes independently of classical risk tol 2:19–26, 2003 and risk of coronary artery disease. Fibri- factors, metabolic syndrome and C-reac- 20. Shaper AG, Pocock SJ, Walker M, Cohen nolysis 10:127–137, 1996 tive protein in the West of Scotland Cor- NM, Wale CJ: British Regional Heart 30. Carmassi F, Morale M, Ferrini L, onary Prevention Study. Diabetes 53: Study: cardiovascular risk factors in mid- Dell’Omo G, Ferdeghini M, Pedrinelli R, 2855–2860, 2004 dle-aged men in 24 towns. BMJ 283:179– de Negri F: Local insulin infusion stimu- 13. Wannamethee SG, Shaper AG, Whincup 186, 1981 lates expression of plasminogen activator PH, Lennon L: Hepatic enzymes, the met- 21. Wannamethee SG, Lowe GDO, Whincup inhibitor-1 and tissue-type plasminogen abolic syndrome and the risk of type 2 PH, Rumley A, Walker M, Lennon L: activator in normal subjects. Am J Med diabetes in older men. Diabetes Care 28: Physical activity and hemostatic and in- 107:344–350, 1999 2913–2918, 2005 flammatory variables in elderly men. Cir- 31. Kelly CJ, Lyall H, Petrie JR, Gould GW, 14. Schindheim RK, Diamant M, Bakker SJL, culation 105:1785–1790, 2002 Connell JM, Rumley A, Lowe GD, Sattar Van Dijk RAJM, Scheffer PG, Teerlink T, 22. Emberson J, Whincup PH, Walker M, N: A specific elevation in tissue plasmin- Kostense PJ, Heine RJ: Liver alanine ami- Thomas M, Alberti KGMM: Biochemical ogen activator antigen in women with notransferase, insulin resistance and en- measures in a population based study: the polycystic ovarian syndrome. J Clin Endo- dothelial dysfunction in normotrigly- effect of fasting duration and time of day. crinol Metab 87:3287–3290, 2002 ceridaemic subjects with type 2 diabetes Ann Clin Biochem 39:493–501, 2002 32. De Jager J, Kooy A, Lehert P, Bets D, mellitus. Eur J Clin Invest 35:369–374, 23. Matthews DR, Hosker JP, Rudenski AS, Wulffele MG, Teerlink T, Scheffer PG, 2005 Naylor BA, Preacher DF, Turner RC: Ho- Schalkwijk CG, Donker AJ, Stehouwer 15. Leiper K, Croll A, Booth NA, Moore NR, meostasis model assessment: insulin re- CD: De Jager J, Kooy A, Lehert P, Bets D, Sinclair T, Bennett B: Tissue plasminogen sistance and ␤-cell function from fasting Wulffele MG, Teerlink T, Scheffer PG, activator, plasminogen activator inhibi- plasma glucose and insulin concentra- Schalkwijk CG, Donker AJ, Stehouwer tors and activator-inhibitor complex in tions in man. Diabetologia 28:412–419, CD: Effects of short-term treatment with liver disease. J Clin Pathol 47:214–217, 1985. metformin on markers of endothelial 1994 24. Walker M, Shaper AG, Lennon L, Whin- function and inflammatory activity in 16. Shimabukuro M, Higa N, Asahi T, Oshiro cup PH: Twenty year follow-up of a co- type 2 diabetes mellitus: a randomized, Y, Takasu N, Tagawa T, Ueda S, Shimo- hort study based in general practices in 24 placebo-controlled trial. J Intern Med 257: mura I, Funahashi T, Matsuzawa Y: Hy- British towns. J Publ Health Med 22:479– 100–109, 2005 poadiponectinemia is closely linked to 485, 2000 33. Rumley A: Fibrinolytic and Endothelial endothelial dysfunction in man. J Clin En- 25. Perry IJ: Non-Insulin-Dependent Diabetes Markers in Cardiovascular Disease and Di- docrinol Metab 88:3236–3240, 2003 and Coronary Heart Disease in Middle-Aged abetes Mellitus. PhD thesis, Glasgow, Uni- 17. Lowe GDO, Danesh J, Lewington S, British Men. PhD thesis. London, Univer- versity of Glasgow, 1996 Walker M, Lennon L, Thomson A, Rum- sity of London, 1995 34. Wannamethee SG, Lowe GDO, Shaper ley A, Whincup PH: Tissue plasminogen 26. Lowe GDO, Rumley A, Whincup PH, AG, Rumley A, Lennon L, Whincup PH: activator antigen and coronary heart dis- Danesh J: Hemostatic and rheological Insulin resistance, haemostatic and in- ease: prospective study and meta-analy- variables and risk of cardiovascular dis- flammatory markers and coronary heart sis. Eur Heart J 25:252–259, 2004 ease. Semin Vasc Med 2: 429–439, 2002 disease risk factors in type 2 diabetes with 18. Juhan-Vague I, Alessi MC, Mavri A, Mor- 27. Duncan BB, Schmidt MI, Offenbacher S, and without coronary heart disease. Dia- ange PE: Plasminogen activator inhibi- Wu KK, Savage PJ, Heiss G, ARIC Inves- betologia 47:1557–1565, 2004

1000 DIABETES CARE, VOLUME 31, NUMBER 5, MAY 2008