sign in sign up
<<
Home , Antithrombin, Fibrinogen

J Am Soc Nephrol 12: 1255–1263, 2001 Circulating Levels of Inhibitor Type-1, Tissue Plasminogen Activator, and Thrombomodulin in Hemodialysis Patients: Biochemical Correlations and Role as Independent Predictors of Coronary Artery Stenosis

ALFONS SEGARRA,* PILAR CHACON,´ † CRISTINA MARTINEZ-EYARRE,‡ XAVIER ARGELAGUER,* JOSEFA VILA,* PILAR RUIZ,* JOAN FORT,* JORGE BARTOLOME,*´ JOAQUIN CAMPS,* ERNESTO MOLINER,§ ANTONI PELEGR´I,࿣ FERNANDO MARCO,¶ ANTONIO OLMOS,* and LLUIS PIERA* *Servicios de Nefrologı´a and †Bioquı´mica, Hospital Valle Hebro´n, Barcelona; ‡CDR Monolab, §Unidad de Hemodia´lisis, Hospital Sant Gervasi; ࿣Centro de Nefrologı´a, Virgen de Montserrat; and ¶Centro de Dia´lisis Nephros, Barcelona, Spain.

Abstract. This study investigated the relationship between the extent with major classic vascular risk factors and to a lesser circulating levels of the endothelial cell - extent with CRP and triglycerides. Forty-six patients ogen activator inhibitor type 1 (PAI-1), tissue plasminogen (23%) had evidence of CAD. Variables associated with CAD activator (TPA), and thrombomodulin (TM) and the major in the univariate analysis included age, time on dialysis, male vascular risk factors described in dialysis patients. In addition, gender, number of packs of cigarettes per year, high BP, the role of these endothelial cell products as independent , apolipoprotein B, , PAI-1 activity, CRP, predictors of coronary artery (CAD) was analyzed. - complexes, and fibrinopeptide A. Lo- Levels of TM, TPA antigen (Ag), TPA activity, PAI-1 Ag, gistic regression analysis found age, high-density PAI-1 activity, TPA/PAI complexes, thrombin-antithrombin cholesterol, gender, high BP, CRP, time on dialysis, and PAI-1 complexes, fibrinopeptide A, C-reactive (CRP), inter- activity to be independent predictors of CAD. This model leukin-1␤ and tumor necrosis factor-␣, lipids, apoproteins A1 classified correctly 85% of patients as having CAD and and B, and albumin were measured in a group of 200 nondi- showed adequate goodness of fit for all risk categories. Our abetic dialysis patients and 100 healthy matched volunteers. data support a pathogenic link among activated inflammatory When compared with healthy controls, dialysis patients response, endothelial injury, and CAD in hemodialysis patients showed increased levels of CRP, TM, TPA, and PAI-1 and and suggest that assessment of circulating PAI-1 levels could evidence of increased thrombin-dependent formation. be an additional tool to identify dialysis patients who are at risk Increased levels of active PAI-1 were associated to a great for developing atheromatous .

Central to the response-to-injury hypothesis is the proposal that vidual patients, some of them are elevated significantly in the different vascular risk factors somehow lead to endothelial disorders with acute endothelial damage and may provide cell injury, which can elicit a series of cellular interactions that reliable correlations in large population studies (3–8). In recent culminate in the lesions of atherosclerosis (1). Several endo- years, it has been shown clearly that circulating levels of PAI-1 thelial products have been proposed as possible in vivo markers and other endothelial cell glycoproteins are increased in hemo- of the endothelial cell injury, such as , dialysis patients (9–13). Although this increase has been con- thrombomodulin (TM), tissue plasminogen activator (TPA), sidered as a subclinical sign of endothelial cell injury (9,12), plasminogen activator inhibitor (PAI), and soluble p-selectin this association is merely a hypothesis that remains to be (2–5). Although all of these markers lack sensitivity and/or proved. Before the circulating levels of PAI-1 and other endo- specificity for assessment of endothelial dysfunction in indi- thelial glycoproteins are considered as indicators of a chronic activated state, other potential mechanisms that contribute to the increased levels of these endothelial products Received July 19, 2000. Accepted November 1, 2000. should be taken into account. First, it is widely known that Correspondence to Dr. Alfons Segarra Medrano, Unidad de Investigacio´n, PAI-1 may behave as an acute-phase reactant (14). Second, Servicio de Nefrologı´a, Hospital Valle Hebro´n, Passeig Vall d’Hebro´n, 119- certain studies suggest that PAI-1 activity in chronic renal 129, E-08035 Barcelona, Spain. Phone: 34-93-274 61 52; Fax: 34-93-274 62 04; E-mail: [email protected] failure and dialysis patients is associated strongly with the 1046-6673/1206-1255 common metabolic abnormalities of obesity and hyperlipid- Journal of the American Society of Nephrology emia (15). Moreover, if circulating endothelial glycoproteins Copyright © 2001 by the American Society of Nephrology such as PAI-1, TPA, and TM are subclinical markers of endo- 1256 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1255–1263, 2001 thelial cell injury, then the levels of these molecules should be PAI-1 activities were determined by chromogenic assays (Biopool, statistically associated with major classical vascular risk fac- Umea, Sweden). Fibrinopeptide A (FPA) was determined by ELISA tors and/or with the presence of atheromatous cardiovascular (Boehringer Mannheim GmbH; Diagnostica Stago). C-reactive pro- disease. tein (CRP) was determined by the Behring Nephelometry immuno- This study was designed (1) to investigate the relationship assay (NA latex CRO, Behring Institute, Galway, Ireland). Serum -1␤ (IL-1␤) and tumor necrosis factor-␣ (TNF␣) were between the circulating levels of the endothelial cell glycop- determined by ELISA (Medgenix Diagnostics, Brussels, Belgium). roteins PAI 1, TPA, and TM and the major vascular risk factors Albumin concentration was determined by the green bromcresol described in dialysis patients and (2) to determine the role of method. these endothelial cell products as independent predictors of All assays were performed in duplicate and calibrated with purified CAD in a large group of nondiabetic dialysis patients. standards and reference plasmas from the manufacturers. Intra- and interassay coefficients of variation for all tests were determined with Materials and Methods the use of 20 different plasma samples. Patients To avoid the influence of acute intercurrent on the bio- chemical parameters, we carried out analyses at least 6 mo after the We studied 200 nondiabetic patients who were receiving dialysis clinical event in all patients who required hospitalization for acute treatment in three outpatient dialysis centers affiliated with our ne- illness or infections (n ϭ 6) or who had surgical procedures (n ϭ 4), phrology department. The study group consisted of 120 men and 80 episodes of kidney allograft rejection (n ϭ 2), or acute vascular women, 31 to 80 yr old, dialyzed three times a week for 4 to 200 mo. (n ϭ 1) before being entered in the study. The hemodialysis prescription was 9 to 13.5 h/wk, with the use of a A prerequisite for including the levels of endothelial cell glycop- 1.2- to 1.6-m cuprophane hollow-fiber filter and bicarbonate dialysate roteins, CRP, and inflammatory cytokines in statistical analyses was containing 2 g/L glucose. The dialyzers were not reused. All patients to demonstrate that the predialysis levels of all of these variables received a multivitamin supplement after the dialysis sessions, includ- showed little variation over time when measured in the same group of ing vitamins C and B and folic acid. A total of 140 patients (70%) patients. For this purpose, we analyzed the variation coefficients for received erythropoietin therapy, 70 patients (35%) received antiplate- the different variables in 30 random patients before the start of four let drugs ( or ), 65 patients (32.5%) were treated with consecutive dialysis sessions. They were as follows: TPA Ag, 12%; -converting inhibitors (ACEI), and 34 patients PAI-1 Ag, 16%; PAI-1 activity, 14%; FPA, 8.4%; TAT complexes, (17%) received therapy with simvastatin or pravastatin. 12.3%; CRP, 11%; IL1-␤, 26%; TNF-␣, 20%; and soluble TM, 9.8%. The cause of chronic renal failure was glomerular disease in 70 patients (35%), interstitial nephritis in 30 patients (15%), polycystic kidney disease in 35 patients (17.5%), vascular disease in 20 patients Clinical Data Collection (10%), and unknown in 45 patients (22.5%). Information on risk factors was obtained by medical record review, The control group comprised 100 healthy, age- and gender- personal interview, and physical examination. Coronary angiography matched volunteers who were recruited from among the population of was performed on all patients who had no previous episodes of healthy people living in the geographic area of our center. well-documented acute myocardial and who experienced The study protocol was accepted by the ethics committee of our clinical symptoms that suggested ischemic heart disease associated hospital, and all patients gave their written informed consent before with ST changes in the ECG. participation. Definitions Laboratory Procedures CAD was diagnosed in the presence of one of the following: (1) For lipid and apoprotein assays, was collected from the definitive episode of myocardial infarction with appropriate rise in antecubital in glass tubes with no additives after an overnight fast serum creatinine phosphokinase associated with typical ECG signs of of 12 h before the start of a dialysis session. For the hemostasis and necrosis (Q waves) or confirmed with technetium pyrophosphate assays, a 19-gauge needle with no tourniquet was used to nuclear cardiac scan, (2) previous coronary bypass or coro- Ͼ collect 10 ml of blood into silicone tubes that contained 0.3 ml of nary angioplasty, or (3) evidence of 50% coronary artery stenosis as 3.8% sodium citrate. Samples were cooled in ice water and centri- determined by selective coronary angiography. fuged immediately at 1000 ϫ g for 45 min at 4°C. The -poor Smoking was assessed by personal interview and entered in the plasma so obtained was aliquoted and stored at Ϫ80°C until assayed analysis in two ways: (1) as current smokers versus others and (2)as within a 1-mo period. a quantitative variable: packs per year (number of packs per day High-density lipoprotein (HDL), low-density lipoprotein, and very- multiplied by years of smoking). low-density lipoprotein were isolated by sequential ultracentrifuga- Patients were considered hypertensive when they were receiving tion. In the three lipoprotein subfractions, cholesterol and triglycerides antihypertensive treatment or when three resting predialysis measure- Ͼ Ͼ were determined by enzymatic methods (cholesterol, CHOD-PAP; ments showed diastolic BP 90 mmHg or systolic BP 140 mmHg. triglycerides, glycerol-3-phosphate-oxidase-peroxidase; Boehringer Body mass index was calculated as weight divided by height, Mannheim, Mannheim, Germany). Apoprotein A1 (apo A1) and apo squared. B were determined by nephelometry (Beckman Array System; Beck- man, Fullerton, CA; interassay CV 7% for apo A1 and 4.5% for apo Statistical Analyses B). Results are given as the mean Ϯ SD. Differences in clinical and Serum levels of PAI-1 Ag, TPA Ag, D-dimer, thrombin-antithrom- biochemical risk factors between groups were calculated by unpaired bin (TAT) complexes, TPA/PAI complexes, and TM were determined t test. Qualitative variables were compared with the use of the ␹2 test. by enzyme-linked immunosorbent assay (ELISA; Asserachrom Diag- Correlation analyses among quantitative variables were carried out nostica Boehringer Mannheim, Mannheim, Germany). TPA and with Pearson’s correlation test. A P value of less than 0.05 was J Am Soc Nephrol 12: 1255–1263, 2001 Endothelial Dysfunction and CAD in Hemodialysis Patients 1257 considered statistically significant. To analyze the factors that deter- Table 1. Clinical and biochemical variables and univariate mined the plasma concentration of endothelial glycoproteins, we analysis in dialysis patients and healthy control carried out single regression analyses introducing the dependent vari- subjectsa ables (FPA, PAI, CRP, TM, and D-dimer) after logarithmic transfor- mation. All of the variables with P values less than 0.1 in single Variable Study Group Control P regression analysis were entered into a stepwise multilinear regression Group Value analysis. We determined the most parsimonious model by removing single variables. N 200 100 Ϯ Ϯ To determine which variables were associated independently with Age (yr) 56.0 14.1 52 18 NS CAD, we carried out a univariate analysis that compared patients with Male gender (%) 60 60 NS 2 and without CAD. All of the variables with P values less than 0.1 in BMI (kg/m ) 23.4 Ϯ 3.6 22.8 Ϯ 4.1 NS the univariate analysis were entered into stepwise multiple logistic Hypertension (%) 57 15 0.0001 regression analysis with a forward selection method. PAI activity, Systolic BP (mmHg) 145 Ϯ 32 125 Ϯ 25 0.05 CRP, and fibrinogen were tested in the model both as continuous Diastolic BP (mmHg) 88 Ϯ 10 66 Ϯ 27 0.05 variables and after being categorized into four groups, each deter- Smokers (%) 30 36 NS mined by the cutoff quartile points. Odds ratios (OR) were calculated HDL cholesterol (mg/dl) 39.5 Ϯ 13.6 51.0 Ϯ 4.1 0.01 from the regression coefficients as an approximation of the relative LDL cholesterol (mg/dl) 99.95 Ϯ 33.0 112 Ϯ 67 NS risk. The linearity of OR was analyzed to determine whether there was Triglycerides (mg/dl) 173 Ϯ 72 85 Ϯ 42 0.001 a continuous relation between the risk factors and clinical event. To Ϯ Ϯ examine a possible effect modification, we tested the interaction terms Apo A1 (mg/dl) 94 96 124 31 0.001 Ϯ Ϯ of PAI-1 activity with fibrinogen and CRP. Apo B (mg/dl) 113.0 32.6 119 46 NS Once obtained, the predictive logistic model was tested blindly on Hb (g/L) 10.17 Ϯ 1.37 14.0 Ϯ 1.9 0.001 an independent group of 60 dialysis patients. This group was obtained Albumin (g/L) 40.8 Ϯ 3.06 44.0 Ϯ 0.8 0.001 randomly from among 150 patients who had been dialyzed in two Fibrinogen (mg/dl) 3.98 Ϯ 3.4 1.1 Ϯ 0.3 0.001 outpatient dialysis centers that did not participate in the first phase of CRP (mg/L) 22.5 (14.0–35.5) 3 (1.2–16.0) 0.001 the study and consisted of 40 men and 20 women, aged 40 to 75 yr IL-1␤ (pg/ml) 34.9 Ϯ 10.9 12.0 Ϯ 10.6 0.001 (62.5 Ϯ 24 yr), dialyzed three times a week for 10 to 125 mo. The TNF␣ (pg/ml) 22.9 Ϯ 3.6 17.0 Ϯ 16.8 0.001 hemodialysis prescription was 9 to 13.5 h/wk with the use of a 1.2- to PAI-1 Ag (ng/ml) 35.4 Ϯ 17.25 9.4 Ϯ 10.0 0.001 1.6-m cuprophane hollow-fiber filter and bicarbonate dialysate con- PAI-1 activity (U/ml) 35.45 Ϯ 7.2 10.2 Ϯ 8.5 0.001 taining 2 g/L glucose. Forty-two patients (70%) received erythropoi- TPA Ag (ng/ml) 6.56 Ϯ 1.78 3.7 Ϯ 2.5 0.001 etin therapy. Information on risk factors was obtained by medical Ϯ Ϯ record review, and blood samples were obtained before a hemodial- TPA activity (U/ml) 3.4 3.02 3.1 0.7 NS Ϯ Ϯ ysis session. An independent group of researchers who were blinded TPA/PAI complex 3.34 3.04 1.1 0.9 0.001 to biochemical results defined the presence of CAD according to the (ng/ml) criteria defined previously. TAT complex (ng/ml) 49.9 Ϯ 18.6 12.0 Ϯ 10.8 0.001 Hosmer-Lemeshow’s test (16) was used to calculate the discrimi- FPA (ng/ml) 60.28 Ϯ 18.9 16.0 Ϯ 13.2 0.001 nation power and goodness of fit of the logistic model. Statistical D-dimer (ng/ml) 999.5 Ϯ 330.0 216 Ϯ 75 0.0001 analyses were performed with the Statistical Package for the Social S-TM (ng/ml) 186 Ϯ 182 32.0 Ϯ 7.6 0.001 Sciences for Windows, 6.1.2 (SPSS, Inc., Cary, NC). a Results are mean Ϯ SD or median (percentiles 25 to 75). BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density Results lipoprotein; Apo, apolipoprotein; Hb, ; CRP, C-reactive When compared with healthy controls, dialysis patients had protein; IL-1␤, interleukin-1␤; TNF␣, tumor necrosis factor-␣; significantly higher levels of triglycerides, TPA Ag, PAI-1 Ag, PAI-1, plasminogen activator inhibitor type 1; Ag, antigen; TPA, PAI-1 activity, FPA, TAT, CRP, IL1-␤, TNF-␣, and soluble tissue plasminogen activator; TAT, thrombin-antithrombin; FPA, fibrinopeptide A; S-TM, serum thrombomodulin. TM. In contrast, serum levels of HDL cholesterol, apo A1, and albumin were significantly lower (Table 1). When the patients were classified according to the disease that led to end-stage renal failure, patients with renal disease of both vascular and prospectively, CRP showed lower variability than IL-1␤ and unknown origin were older than the remaining groups (P Ͻ TNF␣. Therefore, we selected CRP as the most representative 0.01). There were, however, no significant differences among variable that indicated an activated acute-phase response. CRP groups for any of the biochemical variables analyzed. correlated significantly with age (r, 0.23; P Ͻ 0.01), fibrinogen Table 2 summarizes the matrix of Pearson correlations (r, 0.52; P Ͻ 0.01), and certain fibrinolysis variables (see Table among the endothelial cell-derived molecules, fibrinolysis, and 2) and correlated negatively with both HDL cholesterol (r, hemostasis parameters. Table 3 summarizes the results of Ϫ0.19; P Ͻ 0.05) and albumin levels (r, Ϫ0.44; P Ͻ 0.01). stepwise multiple regression analyses carried out to determine When the logarithm of CRP concentration was considered as a the independent predictors of circulating levels of endothelial dependent variable in a forward stepwise multiple regression glycoproteins and certain hemostasis parameters. analysis, albumin, age, fibrinogen, and PAI-1 activity were the We found a close correlation among CRP, IL-1␤, and TNF␣ only variables accepted in the final equation and accounted for (P Ͻ 0.01), indicating that these three variables probably were 18% of the variability of CRP levels (r2, 0.18; P Ͻ 0.001). measuring the same process. In addition, when determined Forty-six patients (23%) had evidence of CAD as diagnosed 1258 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1255–1263, 2001

Table 2. Matrix of Pearson correlations among plasma endothelial glycoproteins, hemostasis parameters, and inflammatory markers

TPA/ Parameter TPA TPA PAI1 PAI1 PAI TAT FPA D-Dimer S-TM CRP IL-1␤ Ag Activity Ag Activity Complex Complex

TPA activity 0.85b PAI-1 Ag 0.66b 0.26a PAI-1 activity 0.68b 0.33b 0.89b TPA/PAI complex 0.70b 0.18a 0.78b 0.76b TAT complex 0.12a 0.10a 0.32b 0.48b 0.76b FPA 0.21a 0.11a 0.35b 0.38b 0.55b 0.81b D-dimer 0.11a 0.09 Ϫ0.31b Ϫ0.45b Ϫ0.42b 0.65b 0.43b S-TM 0.68b 0.26a 0.35b 0.38b 0.33b 0.42b 0.35b 0.12a CRP 0.55b 0.32a 0.65b 0.68b 0.65b 0.07 0.03 0.22a 0.11a IL-1␤ 0.23a 0.12a 0.45b 0.30a 0.24a 0.05 0.01 0.31b 0.03 0.75b TNF 0.18a 0.08 0.22a 0.23a 0.11a 0.09 0.01 0.25a 0.04 0.76b 0.75b

a P Ͻ 0.05. b P Ͻ 0.01. by previous coronary bypass or coronary angioplasty (n ϭ 11), obtained the main effects model, we tested the interaction definitive myocardial infarction (n ϭ 18), or coronary angiog- terms of PAI-1 activity with fibrinogen, CRP, and FPA, which raphy (n ϭ 17). Table 4 shows the clinical and biochemical were consistently nonsignificant. characteristics of the patients with and without CAD. The The prevalence of CAD in the independent group of patients clinical variables associated with CAD in the univariate anal- selected to validate the model was 28.3% (n ϭ 17). There were ysis were age, time on dialysis, male gender, number of packs no significant differences in any of the clinical or biochemical of cigarettes per year, and high BP. Fibrinogen, apo B, PAI-1 parameters between this group of patients and the 200 patients Ag, PAI-1 activity, CRP, TAT, and FPA all were significantly studied in the previous phase. The logistic model classified higher in the patients with CAD, whereas serum HDL choles- correctly 82.35% of patients in this group for having CAD, and terol, apo A1, and albumin all were significantly lower. Solu- goodness of fit also was adequate for all of the risk categories ble TM was not associated with CAD in the univariate (Table 6). analysis. The final logistic regression model obtained after a forward Discussion selection of the variables included age, HDL cholesterol, gen- This study was conducted to obtain information on the der, high BP, CRP, PAI-1 activity, and time on dialysis as factors that determine and the clinical significance of the independent predictors of patient status (Table 5). This model increased circulating levels of endothelial cell-derived glyco- correctly classified 85% of patients as having CAD and described in hemodialysis patients. showed an adequate goodness of fit for all risk categories (␹2, In agreement with previous reports (9–13), our dialysis 1.21). patients showed increased circulating levels of endothelial Although the degree of association between CRP and CAD glycoproteins. Although we cannot exclude the possibility that in the univariate analysis was very similar to that observed individual patients who exhibited increased levels of these between albumin and CAD, in the multivariate analysis albu- endothelial products might have experienced subclinical infec- min concentration lost its predictor significance when CRP was tions or imperceptible vascular complications, our method was introduced in the model. Moreover, among the variables de- designed to guarantee that the parameters studied were not termined to analyze activation of the inflammatory response, influenced by acute intercurrent events in the vast majority of CRP showed the highest statistical significance as predictor of patients. Moreover, our patients showed features consistent coronary stenosis. When IL-1 or TNF␣ were introduced in- with an activated acute-phase response. The plasma levels of stead of CRP, the predictive model, although significant, lost CRP, IL-1, and TNF tended to remain constant over time, discriminative capacity and showed an inadequate goodness of suggesting that the activation of acute-phase response was not fit. That was particularly evident for the low-risk categories due to acute intercurrent diseases. (area under the curve, 0.77 [P Ͻ 0.01]; ␹2, 10.32 [P Ͻ 0.01]). The first relevant finding of our study was the evidence of a The goodness of fit of the model was significantly better when significant statistical association between levels of both PAI-1 values for CRP, PAI-1, and FPA were introduced as continu- Ag and TPA Ag and major vascular risk factors. This associ- ous quantitative variables rather than as categorical variables ation supports the hypothesis that the increase in circulating defined by their corresponding quartiles (␹2, 1.21 versus 6.12, levels of these molecules was due, at least in part, to direct 7.1, and 9.45, respectively; P Ͻ 0.001). Finally, after having injury of endothelial cells. Moreover, to a lesser extent, levels J Am Soc Nephrol 12: 1255–1263, 2001 Endothelial Dysfunction and CAD in Hemodialysis Patients 1259

Table 3. Independent determinants of endothelial cell molecules and hemostasis factors in multiple regression models

Dependent Variable ␤ SEM T Significance

PAI-1 activitya age 0.072 0.021 3.54 0.001 smoking (packs/yr) 0.031 0.040 2.99 0.001 BP (mmHg) 0.082 0.020 4.08 0.001 triglycerides (mg/dl) 0.089 0.001 3.99 0.001 HDL cholesterol (mg/dl) Ϫ0.079 0.018 Ϫ4.03 0.001 fibrinogen (mg/dl) 0.099 0.085 3.50 0.001 CRP (mg/dl) 0.099 0.026 3.41 0.001 constant 17.33 8.95 TPA Agb age 0.12 0.076 8.36 0.001 smoking (packs/yr) 0.17 0.005 5.08 0.002 BP (mmHg) 0.18 0.004 4.19 0.001 CRP (mg/dl) 0.21 0.234 9.6 0.001 constant 12.34 3.45 FPAc time on dialysis (mo) 0.06 0.021 3.009 0.003 TAT complex 0.03 0.004 6.836 0.001 fibrinogen (mg/dl) 0.1 0.020 5.786 0.001 PAI-1 activity (U/ml) 0.05 0.018 3.181 0.002 constant 30.88 4.38 TAT complexd time on dialysis (mo) 0.22 0.03 7.01 0.001 fibrinogen (mg/dl) 2.5 0.43 5.65 0.001 constant 36.69 6.71 D-dimere FPA 0.026 0.06 4.16 0.001 PAI-1 activity (U/ml) Ϫ2.18 0.41 Ϫ5.23 0.001 CRP (mg/dl) 0.33 0.03 9.62 0.001 TPA/PAI complex Ϫ0.35 0.07 Ϫ4.94 0.001 constant 16.15 7.96 S-TMf CRP (mg/dl) 0.105 0.018 5.98 0.001 PAI-1 activity (U/ml) 0.294 0.005 6.38 0.001 FPA 0.005 0.002 2.7 0.007 constant 26.85 4.3

a ANOVA F, 21.97; P Ͻ 0.0001; R2, 0.387. b ANOVA F, 23.25; P Ͻ 0.001; R2, 0.217. c ANOVA F, 26.28; P Ͻ 0.0001; R2, 0.337. d ANOVA F, 53.94; P Ͻ 0.001; R2, 0.21. e ANOVA F, 36.77; P Ͻ 0.001; R2, 0.418. f ANOVA F, 23.23; P Ͻ 0.001; R2, 0.183. of both PAI-1 Ag and TPA Ag—but not TM—correlated our group of patients is in agreement with data reported pre- significantly with CRP and other inflammatory markers. This viously in both dialysis and nondialysis patients (12,15,21). association has been described in other studies and indicates This association is supported by recent experimental data in- that PAI-1 is an acute-phase protein that can rise in response to dicating that very-low-density induce PAI-1 syn- several stimuli, including cytokines such as IL-1 and TNF. An thesis through a signaling pathway involving protein kinase alternative explanation for the relationship between PAI-1 and C–mediated mitogen-activated protein kinase activation (22). CRP is that the increase in circulating PAI-1 reflected a cyto- This pathway could be of relevance in dialysis patients because kine-mediated or -facilitated endothelial cell injury as sup- the increase in very-low-density triglycerides is a common ported by certain clinical and experimental data (17–20). The feature of uremic dyslipidemia (23). association between PAI-1 and serum triglycerides observed in Although the increase in soluble-free TM has been consid- 1260 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1255–1263, 2001

Table 4. Clinical, biochemical parameters and univariate TAT complexes, and D-dimer. In agreement with other studies analysis of patients with and without CADa (26–28), our patients showed higher levels of TAT complexes and FPA than healthy control subjects. FPA, a biochemical P Parameter With CAD Without CAD Value marker of intravascular fibrin formation (29), was associated significantly with fibrinogen and TAT complexes, indicating N 46 154 an increased intravascular thrombin-dependent fibrin forma- Age (yr) 60 Ϯ 11 53 Ϯ 15 0.05 tion. Moreover, D-dimer was associated negatively with PAI-1 Male gender (%) 73.9 50.6 0.01 activity, which is concordant with a certain degree of hypofi- 2 BMI (kg/m ) 23.2 Ϯ 4.0 23.6 Ϯ 3.5 NS brinolysis that was evident even in the absence of clinical Hypertension (%) 60.8 55.8 NS manifestations of intravascular thrombosis. Ϯ Ϯ Systolic BP (mmHg) 142 21 138 31 NS To determine the role of endothelial cell products as inde- Ϯ Ϯ Diastolic BP (mmHg) 89.0 4.3 78 12 0.1 pendent predictors of CAD, it was necessary to establish ob- Antiplatelet drugs (%) 32.6 35.7 NS jective criteria to define CAD. In dialysis patients, the preva- Statins (%) 17.3 16.8 NS lence of ischemic heart disease resulting from coronary artery ACEI (%) 32.6 32.5 NS stenosis is difficult to determine because clinical angina with Smokers (%) 34.7 28.5 NS no CAD occurs in 30 to 50% of cases and a number of patients Ϯ Ϯ HDL cholesterol (mg/dl) 39.0 13.7 44.0 13.77 0.05 may have silent coronary stenoses (30,31). To avoid false- LDL cholesterol (mg/dl) 94.4 Ϯ 32.0 101.6 Ϯ 36.0 NS positive results, we established a restrictive operational defini- Triglycerides (mg/dl) 136 Ϯ 79 159.0 Ϯ 71.6 NS tion of ischemic heart disease, requiring unequivocal evidence Apo A1 (mg/dl) 93.9 Ϯ 13.0 91.2 Ϯ 16.0 NS of coronary stenosis. According to these definition criteria, the Apo B (mg/dl) 116.0 Ϯ 24.9 108.0 Ϯ 34.5 0.046 prevalence of CAD observed in our patients was similar to that Hb (g/L) 9.8 Ϯ 1.5 10.2 Ϯ 1.28 NS described in previous studies (30–35). Albumin (g/L) 38.9 Ϯ 3.04 43.3 Ϯ 3.24 0.001 The third and more relevant finding of our study was that an Fibrinogen (mg/dl) 4.15 Ϯ 1.25 3.9 Ϯ 4.28 0.01 increased circulating level of active PAI-1 was an independent CRP (mg/l) 24 (20.0–51.2) 17 (14–30) 0.026 predictor of coronary artery stenosis in nondiabetic dialysis IL-1␤ (pg/ml) 35.3 Ϯ 10.31 33.1 Ϯ 11.06 0.07 patients after adjusting for CRP and other major vascular risk TNF␣ (pg/ml) 23.08 Ϯ 4.15 22.8 Ϯ 3.4 NS factors. It should be emphasized that the multivariate logistic PAI-1 Ag (ng/ml) 39.3 Ϯ 7.1 32.4 Ϯ 3.8 0.002 PAI-1 activity (U/ml) 40.32 Ϯ 9.8 33.6 Ϯ 5.9 0.002 model obtained had a high predictor capacity and goodness of TPA Ag (ng/ml) 6.89 Ϯ 0.45 6.4 Ϯ 2.01 NS fit, both in the large series studied and in the validation series, TPA activity (U/ml) 3.28 Ϯ 3.1 32.3 Ϯ 3.0 NS indicating that it could be of potential applicability in clinical TPA/PAI complex 3.5 Ϯ 2.9 3.3 Ϯ 2.4 NS practice for estimating the risk of CAD. (ng/ml) The independent association between CRP levels and vas- TAT complex (ng/ml) 58.6 Ϯ 12.0 47.2 Ϯ 19.4 0.0001 cular disease observed in our patients concurs with data from FPA (ng/ml) 69.1 Ϯ 19.8 57.5 Ϯ 17.8 0.0001 several authors (36–42). Of all of the inflammatory markers D-dimer (ng/ml) 944 Ϯ 305 1016 Ϯ 336 NS analyzed, CRP showed the highest statistical significance as a S-TM (ng/ml) 180 Ϯ 194 187 Ϯ 179 NS predictor of coronary artery stenosis. When albumin, IL-1, or TNF␣ were introduced instead of CRP, the model, although a Ϯ Results are mean SD or median (percentiles 25 to 75). CAD, significant, lost discriminative capacity and showed an inade- coronary artery disease; ACEI, angiotensin-converting enzyme inhibitors. quate goodness of fit that was particularly evident for the low-risk categories. The independent association between circulating PAI-1 and ered classically as a biochemical marker of endothelial injury, the presence of coronary artery stenosis had not been described our findings are not in concordance with this hypothesis be- previously in dialysis patients. In the general population, ele- cause TM was associated neither with major vascular risk vated PAI-1 has been shown to be unvariably predictive of factors nor with coronary heart disease. This lack of associa- unstable angina and myocardial infarction only in patients with tion probably indicates that TM is not a reliable marker of preexisting CAD (43,44). In comparison with these observa- endothelial cell injury in chronic renal failure because a sig- tions, population-based studies have demonstrated an inconsis- nificant amount of the increase in circulating TM is due to the tent association between fibrinolytic parameters and the devel- loss of renal function itself (24,25). opment of CAD. The Caerphilly Study (45) found that PAI-1 The second interesting point of our study was the evidence activity was not predictive of incident CAD, whereas the Prime that most of PAI-1 was circulating in its active form with the Study (46), a prospective cohort study of 10,500 men who potential ability to bind and inactivate TPA. This increased initially were free of cardiovascular disease, noted a significant PAI-1 activity might explain the discrepancy that we observed association between the PAI-1 level and cardiovascular disease between TPA concentration and TPA activity. To analyze the after adjusting for other cardiovascular risk factors. Theoreti- functional significance of increased PAI-1 activity on intravas- cally, PAI-1 could be related with atheromatous disease either cular fibrin breakdown, we measured plasma levels of FPA, as a passive marker of an endothelial cell injury of multifac- J Am Soc Nephrol 12: 1255–1263, 2001 Endothelial Dysfunction and CAD in Hemodialysis Patients 1261

Table 5. Logistic regression analysis to predict presence of CAD in the study groupa

␤ 95% Confidence Variable SEM Odds Ratio Interval Significance

Age 0.0981 0.0244 1.10 1.01–1.23 0.0001 HDL cholesterol Ϫ0.3455 0.0962 0.70 0.55–0.93 0.0003 PAI-1 0.1645 0.0487 1.17 1.09–3.22 0.0007 Gender (m) 0.6013 0.0610 1.91 1.05–3.04 0.0010 CRP 0.2059 0.0612 1.22 1.12–6.01 0.0008 Months on HD 0.3816 0.1366 1.16 1.02–2.12 0.0052 BP (1) 0.0918 0.0257 1.09 1.05–3.19 0.0003

a Age (10-yr increment), HDL cholesterol (increment, 10 mg/dl), PAI-1 activity (increment, 5 U/ml), gender (male versus female), CRP (increment, 1 mg/dl), months on HD (increment, 10 mo), BP (1, hypertensive; 0, nonhypertensive). Model ␹2, 155.3; df, 10; significance, 0.000.

Table 6. Logistic regression analysis to predict the presence of CAD in an independent sample of 60 patientsa

␤ 95% Confidence Variable SEM Odds Ratio Interval Significance

Age 0.2263 0.6012 1.32 1.09–1.99 0.0414 HDL cholesterol Ϫ0.1800 0.0685 0.83 0.73–0.95 0.0086 PAI-1 0.3214 0.2174 1.03 1.01–4.46 0.0472 Gender (m) 0.4996 0.0129 1.60 1.04–4.54 0.0001 CRP 0.7210 0.5876 2.27 1.44–14.4 0.0096 Months on HD 0.1954 0.6243 1.02 1.009–2.13 0.0455 BP (1) 0.3573 0.2696 1.57 1.29–6.59 0.0210

a Age (10-yr increment), HDL cholesterol (increment, 10 mg/dl), PAI-1 activity (increment, 5 U/ml), gender (male versus female), CRP (increment, 1 mg/dl), mo on HD (increment, 10 mo), BP (1, hypertensive; 0, nonhypertensive). Model ␹2, 78,621; df, 8; significance, 0.0000. torial origin or as a potential inductor of an hypofibrinolysis References state, which may permit fibrin to persist on the surface of the 1. Ross R: The pathogenesis of atherosclerosis—An update. N Engl injured vessel. Our data account only for a small percentage of J Med 314: 488–500, 1986 circulating PAI-1 variability but support the hypothesis that in 2. Blann AD, Taberner DA: A reliable marker of endothelial cell dialysis patients, circulating levels of PAI-1 increase in part as dysfunction: Does it exist? Br J Haematol 90: 244–248, 1995 a result of endothelial cell injury and may contribute to the 3. Blann AD, Dobrotova M, Kubisz P, McCollum CN: Von Wil- atheromatous disease promoting a certain degree of lebrand factor, soluble P-selectin, tissue plasminogen activator and plasminogen activator inhibitor in atherosclerosis. Thromb hypofibrinolysis. Haemost 74: 626–630, 1995 In summary, in this study we provided evidence that, in 4. Blann AD, McCollum CN: Circulating endothelial cell/leukocyte dialysis patients, the circulating levels of the endothelial cell adhesion molecules in atherosclerosis. Thromb Haemost 72: glycoproteins PAI-1 and TPA were statistically associated with 151–154, 1994 major vascular risk factors and, to a lesser degree, with an 5. Verhaar MC, Beutler JJ, Gaillard CA, Koomans HA, Fijnheer R, activated acute-phase response and serum triglycerides. More- Rabelink TJ: Progressive vascular damage in hypertension is over, circulating PAI-1 was an independent predictor of coro- associated with increased levels of circulating P-selectin. J Hy- nary artery stenosis after adjusting for the major vascular risk pertens 16: 45–50, 1998 factors and CRP. Taken together, our data suggest that in- 6. Nakamura Y, Chida Y, Tomura S: Enhanced -fibrin- creased circulating PAI-1 could indicate a chronic endothelium olysis in patients on regular hemodialysis treatment. Nephron 58: activated state and could be an additional tool to identify 201–204, 1991 7. Tomura S, Nakamura Y, Deguchi F, Chida Y, Ohno Y, Kodama dialysis patients who are at risk for developing atheromatous S, Hayashi TK, Suzuki K, Marumo F: Plasma von Willebrand cardiovascular disease. Future studies designed to define fur- factor and thrombomodulin as markers of vascular disorders in ther the underlying pathways that account for increased circu- patients undergoing hemodialysis therapy. Thromb Res 58: 413– lating PAI-1 levels would be of great interest to delineate the 419, 1990 pathogenic mechanisms that lead to ischemic cardiovascular 8. Takagi M, Wada H, Mukai K, Kihira H, Yano S, Minamikawa K, disease in this group of patients. Wakita Y, Nakase T, Nagaya S, Deguchi K, et al.: Increased 1262 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 1255–1263, 2001

vascular endothelial cell markers in patients with chronic renal 25. Ishii H, Uchiyama H, Hiraishi S, Nakano M, Tsubouchi J, failure on maintenance haemodialysis. Blood Coagul Fibrinoly- Kazama M: Clinical significance of the measurement of plasma sis 5: 713–717, 1994 thrombomodulin. Rinsho Byori 37: 266–271, 1989 9. Gris JC, Branger B, Ve´cina F, Al Sabadani B, Fourcade J, 26. Ishii Y, Yano S, Kanai H, Maezawa A, Tsuchida A, Wakamatsu Schved JF: Increased cardiovascular risk factors and features of R, Naruse T: Evaluation of blood coagulation-fibrinolysis system endothelial activation and dysfunction in dialyzed uremic pa- in patients receiving chronic hemodialysis. Nephron 73: 407– tients. Kidney Int 46: 807–813, 1994 412, 1996 10. Haaber AB, Eidemak I, Jensen T, Feldt-Rasmussen B, 27. Ambu¨hl PM, Wu¨thrich RP, Korte W, Schmid L, Krapf R: Plasma Strandgaard S: Vascular endothelial cell function and cardiovas- hypercoagulability in haemodialysis patients: Impact of dialysis cular risk factors in patients with chronic renal failure. JAmSoc and anticoagulation. Nephrol Dial Transplant 12: 2355–2364, Nephrol 5: 1581–1584, 1995 1997 11. Ishii Y, Yano S, Kanai H, Maezawa A, Tsuchida A, Wakamatsu 28. Sagripanti A, Cupisti A, Baicchi U, Ferdeghini M, Morelli E, R, Naruse T: Evaluation of blood coagulation-fibrinolysis system Barsotti G: Plasma parameters of the prothrombotic state in chronic uremia. Nephron 63: 273–278, 1993 in patients receiving chronic hemodialysis. Nephron 73: 407– 29. Nossel HL, Ti M, Kaplan KL, Spandonis K, Solnad T, Butler VP 412, 1996 Jr: The generation of fibrinopeptide A in clinical blood samples: 12. Tomura S, Nakamura Y, Mayumi D, Ryoichi A, Takashi I, Chida Evidence for thrombin activity. J Clin Invest 58: 1136–1144, Y, Ootsuka S, Shinoda T, Yanagi H, Tsuchiya S, Marumo F: 1976 Fibrinogen, coagulation factor VII, tissue plasminogen activator, 30. Rostand SG, Kirk KA, Rutsky EA: Dialysis-associated ischemic plasminogen activator inhibitor-1, and lipid as a cardiovascular heart disease: Insights from coronary angiography. Kidney Int risk factors in chronic hemodialysis and continuous ambulatory 25: 653–659, 1984 peritoneal dialysis. Am J Kidney Dis 27: 848–854, 1996 31. Rostand SG, Kirk KA, Rutskey EA: The epidemiology of coro- 13. Mezzano D, Tagle R, Pais E, Panes O, Perez M, Downey P, nary artery disease in patients on maintenance hemodialysis: Mun˜oz B, Aranda E, Barja P, Thambo S, Gonzalez F, Mezzano Implications for management. Contrib Nephrol 52: 34–41, 1986 S, Pereira J: Endothelial cell markers in chronic uremia: Rela- 32. Parfrey PS, Foley RN, Harnett JD, Kent GM, Murray D, Barre tionship with hemostatic defects and severity of renal failure. PE: Outcome and risk factors of ischemic heart disease in Thromb Res 88: 465–472, 1997 chronic uremia. Kidney Int 49: 1428–1434, 1996 14. Brozovic M: Physiological mechanisms in coagulation and fibri- 33. Koch M, Kutkuhn B, Trenkwalder E, Bach D, Grabensee B, nolysis. Br Med Bull 33: 231–238, 1977 Dieplinger H, Konenberg F: Apolipoprotein B, fibrinogen, HDL 15. Irish AB: Plasminogen activator inhibitor-1 activity in chronic cholesterol, and apolipoprotein (a) phenotypes predict coronary renal disease and dialysis. Metabolism 46: 36–40, 1997 artery disease in hemodialysis patients. J Am Soc Nephrol 8: 16. Hosmer DW, Lemeshow S: Confidence interval estimates of an 1889–1898, 1997 index of quality performance based on logistic regression mod- 34. Harnett JD, Foley RN, Kent GM, Barre PE, Murray D, Parfrey PS: els. Stat Med 106: 565–570, 1995 Congestive heart failure in dialysis patients: Prevalence, incidence, 17. Mussoni L, Baldasarre D, Mannucci L, Sirtori CR, Tremoli E: prognosis and risk factors. Kidney Int 47: 884–890, 1995 Relationship between fibrinolytic variables: A study in patients 35. Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE: attending a lipid clinic. Ann Med 32: 134–141, 2000 Impact of hypertension on cardiomyopathy, morbidity and mortality 18. Juhan-Vague I, Alessi MC, Joly P, Thirion X, Vaghe P, Declerck in end-stage renal disease. Kidney Int 49: 1379–1385, 1996 PJ, Seradimigni A, Collen D: Plasma plasminogen activator- 36. Docci D, Bilancioni R, Baldrati L, Capponcini C, Turci F, Feletti inhibitor in angina pectoris. Influence of plasma and acute C: Elevated acute phase reactants in hemodialysis patients. Clin phase response. Arteriosclerosis 9: 362–367, 1989 Nephrol 34: 88–91, 1990 19. Vasse M, Mirshahi SS, Soria J, Mirshahi M, Borg JY, Montcon- 37. Bergstro¨m J, Heimbu¨rger O, Lindholm B, Qureshi AR: Elevated duit M, Soria C: Potent activity of peripheral blood monocytes in serum C-reactive protein is a strong predictor of increased mor- tality and low in hemodialysis (HD) patients inducing stimulating factor and in mono- [Abstract]. J Am Soc Nephrol 6: 573, 1995 cytes. Blood Coag Fibrinolysis 4: 143–147, 1993 38. Owen FO, Lowrie EG: C-reactive protein as an outcome predic- 20. Cotran RS, Pober JS: Effects of cytokines on vascular endothe- tor for maintenance hemodialysis patients. Kidney Int 54: 627– lium: Their role in vascular and immune injury. Kidney Int 35: 636, 1998 969–975, 1989 39. Kaysen GA, Stevenson FT, Depner TA: Determinants of albu- 21. Chadarevian R, Bruckert E, Dejager S, Preseberg P, Turoin G: min concentration in hemodialysis patients. Am J Kidney Dis 29: Relationship between triglycerides and factor VIIc and plasmin- 659–668, 1997 ogen activator inhibitor type-1: Lack of threshold value. Thromb 40. Zimmermann J, Herrlinger S, Pruy A, Metzger T, Wanner C: Res 96: 175–182, 1999 enhances cardiovascular risk and mortality in he- 22. Banfi C, Mussoni L, Rise P, Cattaneo MG, Vicentini L, Battaini modialysis patients. Kidney Int 55: 648–658, 1999 F, Galli C, Tremoli E: Very low density lipoprotein-mediated 41. Stenvinkel P, Heimbu¨rger O, Paultre F, Diczfalusy U, Wang T, signal transduction and plasminogen activator inhibitor type-1 in Berglund L, Jogestrand T: Strong association between malnutri- cultured HepG2 cells. Circ Res 85: 208–217, 1999 tion, inflammation and atherosclerosis in chronic renal failure. 23. Attman PO, Alaupovic P: Lipid and apolipoprotein profiles of Kidney Int 55: 1899–1911, 1999 uremic dyslipoproteinemia—Relation to renal function and dial- 42. Ikizler TA, Wingard RL, Harvell J, Shyr Y, Hakim RM: Asso- ysis. Nephron 57: 401–410, 1991 ciation of morbidity with markers of nutrition and inflammation 24. Ishii H, Kazama M: Thrombomodulin. Blood Vessels 20: 496– in chronic hemodialysis patients: A prospective study. Kidney Int 505, 1989 55: 1945–1951, 1999 J Am Soc Nephrol 12: 1255–1263, 2001 Endothelial Dysfunction and CAD in Hemodialysis Patients 1263

43. Jansson JH, Olofsson BO, Nilsson TK: Predictive value of tissue HF, Elwood PC: Fibrin D-dimer, tissue plasminogen activator, plasminogen activator mass concentration of long-term mortality plasminogen activator inhibitor, and the risk of major ischemic in patients with coronary artery disease. A 7-year follow-up. heart disease in the Caerphilly Study. Thromb Haemost 79: Circulation 88: 2030–2034, 1993 129–133, 1998 44. Juhan-Vague I, Pyke S, Alessi MC, Jespersen J, Haverkate F, 46. Scarabin PY, Aillaud MF, Amouyel P, Evans A, Luc G, Ferrie`res Thompson SG, ECAT Study Group: Fibrinolytic factors and the J, Arveiler D, Juhan-Vague I: Associations of fibrinogen, factor risk of myocardial infarction or sudden death in patients with VII and PAI-1 with baseline findings among 10,500 male par- angina pectoris. Circulation 94: 2057–2063, 1996 ticipants in a prospective study of myocardial infarction—The 45. Lowe GDO, Yarnell JWG, Sweetnam PM, Rumley A, Thomas Prime Study. Thromb Haemost 89: 749–756, 1998