Tissue plasminogen activator antigen is strongly associated with myocardial infarction in young women

Authors

  • P. M. MANNUCCI,

    1. Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Fondazione Luigi Villa and Department of Internal Medicine and Dermatology, IRCCS Maggiore Hospital and University of Milano; Department of Health Sciences, Section of Medical Statistics and Epidemiology, University of Pavia
    Search for more papers by this author
  • L. BERNARDINELLI,

    1. Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Fondazione Luigi Villa and Department of Internal Medicine and Dermatology, IRCCS Maggiore Hospital and University of Milano; Department of Health Sciences, Section of Medical Statistics and Epidemiology, University of Pavia
    Search for more papers by this author
  • L. FOCO,

    1. Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Fondazione Luigi Villa and Department of Internal Medicine and Dermatology, IRCCS Maggiore Hospital and University of Milano; Department of Health Sciences, Section of Medical Statistics and Epidemiology, University of Pavia
    Search for more papers by this author
  • M. GALLI,

    1. Division of Cardiology, ASL6 Hospital of Livorno
    Search for more papers by this author
  • F. RIBICHINI,

    1. Catheterization Laboratory, Maggiore Hospital and East Piedmont University, Novara
    Search for more papers by this author
  • M. TUBARO,

    1. Coronary Care Unit, Cardiovascular Department, San Filippo Neri Hospital, Rome, Italy
    Search for more papers by this author
  • F. PEYVANDI

    1. Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Fondazione Luigi Villa and Department of Internal Medicine and Dermatology, IRCCS Maggiore Hospital and University of Milano; Department of Health Sciences, Section of Medical Statistics and Epidemiology, University of Pavia
    Search for more papers by this author

P.M. Mannucci, Via Pace 9, 20122 Milan, Italy.
Tel.: +39 02 55035421; fax: +39 0250320723; e-mail: pmmannucci@libero.it

Abstract

Summary.  Women who develop acute myocardial infarction (AMI) at a young age have fewer classical risk factors and less coronary stenosis than older women. In this rare population, it is plausible that a heightened hemostatic system may play an important mechanistic role in thrombus formation and in the development of AMI. We chose to investigate whether or not there is an association between premature AMI and the plasma concentrations of five hemostatic measurements that had been previously established as risk factors for AMI, and of the inflammation marker C-reactive protein (CRP). Women who had survived AMI at the age of 45 years or less (n = 141) were drawn from those admitted to 125 Italian coronary care units over a 3-year period. In them, and in an equal number of controls, plasma levels of immunoreactive tissue plasminogen activator (tPA), plasminogen activation inhibitor 1 (PAI-1), von Willebrand factor (VWF), fibrinogen, D-dimer and CRP were measured. Higher levels of VWF, fibrinogen, CRP and tPA were associated with AMI. After adjustment for both classical and hemostatic risk factors, only tPA maintained an independent association with AMI: the odds ratios (taken as an index of relative risk) for tPA values in the middle and higher tertiles were 2.86 (CI 1.63–5.02) and 8.18 (CI 2.66–25.20), respectively. In conclusion, there is a strong association between non-fatal AMI and increased plasma levels of tPA antigen. This finding is thought to be the expression of a reduced rather than enhanced fibrinolytic activity.

Introduction

Acute myocardial infarction (AMI) is very rare in individuals younger than 40–45 years of age (2–6% of all incident cases), particularly in women (10–20% of the cases occurring in the young) [1,2]. In women of premenopausal age, coronary artery atherosclerosis and stenosis are usually less severe and extensive than in middle-aged and elderly women [3–7]. AMI results from the interplay of two mechanisms: the chronic formation of atherosclerotic plaques and the acute formation of thrombosis upon them. Hence, it is plausible that a heightened hemostatic system plays an important role in the increased tendency to thrombus formation and ultimately in the occurrence of AMI in young women with relatively little atherosclerosis and coronary artery stenosis. The ideal way to evaluate the role of this mechanism would be to measure hemostasis factors in healthy young women, to follow them prospectively and to compare the baseline results of those who subsequently developed AMI with those who remained healthy. However, it is difficult to implement such a study because of the prohibitively large number of young women at low risk that should be followed for a very prolonged period of time to have a sizable number of events.

A more feasible approach is a retrospective case–control study, based upon a comparison of the behavior of hemostatic factors in women who developed AMI before the age of 45 years with those in healthy women. The young women included in this study are a subset of a larger study carried out in Italy to assess the predictive value of genetic markers in men and women with premature AMI [8]. Haemostatic factors were selected from among those that in previous prospective and retrospective studies had been associated more frequently and consistently with the development of AMI, such as plasma fibrinogen, von Willebrand factor (VWF), plasminogen activator inhibitor type 1 (PAI-1), tissue plasminogen activator (tPA) and D-dimer [9,10]. Serum C-reactive protein (CRP), a marker of systemic inflammation and an independent predictor of AMI in men and women, was also evaluated [11–14].

Patients and methods

Population

Between January 1998 and January 2001, 149 women were admitted to 125 coronary care units (CCU) in Italy after a first AMI that occurred before the age of 45 years. They all underwent coronary arteriography at the time of hospitalization. Venous blood was obtained during a period spanning from 3 to 12 months after AMI. Plasma samples were not available for eight women so only 141 cases were investigated. Controls (n = 141) were healthy women unrelated to the patients but individually matched with them for age and geographic origin and recruited from among the staff of the same participating hospitals. They had no history of arterial or venous thromboembolism as assessed using a validated questionnaire [15]. Another questionnaire was used to obtain information on classical, non-hemostatic risk factors for AMI.

Definitions

We have previously reported the criteria used to define AMI, the presence or absence of such classical non-hemostatic risk factors as a family history of ischemic heart disease, smoking, overweight, hypercholesterolemia, diabetes, hypertension, oral contraceptive intake, cocaine use, alcohol consumption and physical inactivity [8]. The absence of any narrowing in diameter was considered evidence of a normal coronary artery, a narrowing of < 70% (< 50% for the left main coronary artery) was considered-non-significant stenosis, and a narrowing > 70% (> 50% for the left main coronary artery) was considered significant stenosis. No specific information was obtained on the drugs used by patients after their AMI at the time of blood sampling because the study was originally based on genotyping [8]. However, it is common practice in Italian CCUs to give low-dose aspirin, statins, β-blockers and angiotensin-converting enzyme inhibitors to all patients who had AMI, whereas it is very unlikely that control women used these medications systematically. Information was obtained for cases at the time of AMI, and at the time of hospital evaluation for inclusion in the study for controls. The Institutional Review Boards of the participating hospitals approved the study protocol and the informed consent form.

Blood collection and laboratory measurements

Blood was drawn from an antecubital vein into evacuated tubes containing 0.129 mol L−1 trisodium citrate and separated into platelet-poor plasma, buffy coat for DNA extraction and red cells by centrifugation at 2000 g for 20 min. Plasma and DNA were divided into small subsamples and kept at − 70 °C until the relevant assays were carried out. A normal plasma pool obtained from 40 blood donors was taken as reference for all the assays. For fibrinogen, tPA, VWF and CRP the reference plasma was calibrated against the corresponding International Standards provided by the National Institute for Biological Standards and Control (NIBCS).

Plasma fibrinogen was measured as immunoreactive protein by enzyme immunoassay (EIA) using a commercial rabbit anti-human fibrinogen antiserum as a source of both first and second antibody (Dako, Glostrup, Denmark). Plasma tPA and PAI-1 antigens were measured by EIA obtained from Chromogenix (COALIZA t-PA and PAI-1; Chromogenix, Instrumentation Laboratory, Milan, Italy), CRP was measured using a highly sensitive EIA (ZYMUTEST CRP Hyphen; BioMed, Andresy, France), VWF antigen was measured using the IMUBIND VWF EIA (American Diagnostica, Greenwich, CT USA) and D-dimer was measured using the Dimertest Gold EIA (Agen Biomedical, Brisbane, Australia). The within- and between-assay coefficients of variation of these methods were consistently below 10% in the quality control exercises that were regularly performed in the laboratory.

Statistical analysis

The frequency distribution for classical risk factors was evaluated in cases and controls. For each factor the unadjusted relative risk for AMI (estimated as odds ratio, OR) was calculated by fitting a conditional logistic regression model including one factor only. The adjusted ORs were then were calculated, including in the model all the factors.

The behavior of the hemostasis factors was described using medians and 10th and 90th percentiles of the observed values, because the distribution of values was skewed in cases and controls. Levels of hemostatic factors in cases and controls were graphically visualized by means of box plots. A test on the departure from the null hypothesis of no difference was performed via a Wilcoxon test. Before performing the test, values of hemostasis factors were logarithmically transformed to make symmetric the distribution of the differences computed within each matched pair.

To analyze the association between each hemostasis factor and AMI, a conditional logistic regression model was fitted including the results of each hemostasis factor divided into tertiles. The use of tertiles is motivated by the observation that the distribution of all factors was strongly skewed with the presence of several outliers and by the need to interpret clinically the association of any measured level of each factor with AMI. Each hemostasis factor was included in the model as a categorical variable if the test of departure from linearity was statistically significant, otherwise it was a continuous variable.

To assess the possible influence of classical risk factors on the relationship between each hemostasis factor and AMI, adjusted ORs were obtained by conditional logistic regression including the classical risk factors found to be associated with AMI. To identify those hemostasis factors that were associated with AMI independently on each other a conditional logistic regression model that included all the studied hemostasis factors was fitted. ORs were also adjusted for classical risk factors by including in the model both the hemostasis factors and the classical risk factors found to be associated with AMI. Results were presented as ORs and their corresponding 95% confidence intervals (CIs).

Results

The mean ages (± standard deviation) of the 141 cases with premature AMI and of the 141 controls were 39 ± 5 and 39 ± 4 years, respectively. Table 1 shows the frequency distribution of the classical, non-hemostatic risk factors and the medians with 10th and 90th percentile values of the hemostasis factors for the matched sets of cases and controls.

Table 1.  Classical risk factors and hemostasis factors in young women who survived acute myocardial infarction
Classical risk factorCases
(n = 141)
Controls
(n = 141)
Haemostatic factorsCases
(n = 141)
Controls
(n = 141)
P-value
  1. Results of classical, non-hemostatic factors are expressed as frequencies (%) in cases and controls. Results of hemostatic factors are expressed as median values and, between parenthesis, 10th and 90th percentile values.

Smoking
 No25.551.1Tissue plasminogen activator (ng mL−1)4.6 (2.3–9.0)2.9 (2.1–5.6)< 0.0001
 Yes74.548.9    
Oral contraceptives
 No58.974.1Plasminogen activator inhibitor (ng mL−1)56.6 (20.7–83.6)51.2 (18.3–87.5)0.715
 Yes41.125.9    
Hypertension
 No74.192.2von Willebrand factor (U dL−1)122 (74–184)99 (63–153)< 0.0001
 Yes25.97.8    
Family history
 No38.357.4Fibrinogen (mg dL−1)389 (309–568)364 (283–514)0.0089
 Yes61.742.6    
Body mass index
 Normal weight68.878.7D-dimer (ng mL−1)29 (13–108)28 (13–74)0.821
 Preobese and
 obese
31.221.3    
 (WHO classes I–III)
Hypercholesterolemia
 No54.157.0C-reactive protein
  (μg mL−1)
1.1 (0.2–17.1)0.6 (0.1–4.6)0.0014
 Yes45.943.0    
Alcohol
 No72.375.0    
 Yes27.725.0    
Physical exercise
 No64.551.1    
 Yes35.548.9    

Table 2 reports the ORs and the corresponding 95% CIs unadjusted and adjusted for all the classical risk factors. Diabetes (seven cases and one control) and cocaine use (four cases and zero controls) were not included in the analysis because of their low frequency. Smoking and hypertension were the classical risk factors with the strongest effects, even after adjustment. Body mass index, oral contraceptive intake, family history and physical exercise had weaker effects, particularly after adjustment. No effect was detected for alcohol consumption and hypercholesterolemia.

Table 2.  Association of classical risk factors with acute myocardial in young women
VariableUnadjusted OR
(95% CI)
P-valuesAdjusted OR
(95% CI)
P-values
  1. Associations are expressed as odds ratios (ORs) as indices of relative risk. ref denotes reference group.

Smoking
 No1.0 (ref) 1.0 (ref.) 
 Yes3.12 (1.80–5.38)P < 0.00013.52 (1.70–7.26)P = 0.001
Hypertension
 No1.0 (ref) 1.0 (ref.) 
 Yes5.33 (2.23–12.75)P < 0.00015.41 (1.78–16.41)P = 0.003
Body mass index
 Normal weight    
 Preobese and obese1.0 (ref) 1.0 (ref) 
 (WHO classes I–III)1.64 (0.96–2.78)P = 0.0692.04 (0.99–4.21)P = 0.055
Oral contraceptives
 No1.0 (ref) 1.0 (ref) 
 Yes2.31 (1.28–4.16)P = 0.0051.87 (0.91–3.85)P = 0.087
Family history
 No1.0 (ref) 1.0 (ref) 
 Yes2.50 (1.45–4.32)P = 0.0011.81 (0.91–3.59)P = 0.091
Physical exercise
 No1.0 (ref) 1.0 (ref) 
 Yes0.54 (0.32–0.90)P = 0.0180.75 (0.37–1.51)P = 0.423
Hypercholesterolemia
 No1.0 (ref) 1.0 (ref) 
 Yes1.17 (0.71–1.92)P = 0.5290.85 (0.44–1.64)P = 0.627
Alcohol
 No1.0 (ref) 1.0 (ref) 
 Yes1.15 (0.68–1.95)P = 0.5931.22 (0.56–2.68)P = 0.614

Figure 1 shows the box plots of the results of hemostasis factors in cases and controls. There were statistically significant differences for levels of tPA and VWF between cases and controls (P < 0.0001). This was also true, but to a lesser degree, for CRP (P = 0.0014) and fibrinogen (P = 0.0089), whereas D-dimer and PAI-1 levels were not significantly different.

Figure 1.

Haemostatic factors and C-reactive protein in controls and cases. Boxes represent the quartile range, with the medians represented by the solid horizontal lines. Whiskers at the top and bottom of the boxes show the highest and lowest values. (A) denotes controls, (B) denotes cases. The values in the vertical axis are in a logarithmic scale.

To evaluate whether or not there was an association between AMI and the degree of abnormality of hemostatic factors, the values of each factor in cases were divided into tertiles and the lowest tertile was taken as a reference to calculate the ORs. Table 3 shows crude ORs and ORs adjusted for smoking, hypertension, body mass index, physical exercise, family history and oral contraceptive intake, i.e. those non-hemostatic factors that had shown some association with AMI in the univariate analysis. ORs for each hemostatic factor (and CRP) increased progressively from the first to the third tertile. In particular, all the hemostatic factors with the exception of D-dimer and PAI-1 showed a positive association with AMI. After adjustment for traditional risk factors statistically significant ORs were still detectable for tPA and VWF antigens, while CRP preserved borderline significance and fibrinogen lost it.

Table 3.  Association of hemostatic factors (and C-reactive protein) with acute myocardial infarction in young women according to the distribution of values in tertiles
VariableUnadjusted OR
(95% CI)
P-valueAdjusted ORs*
(95% CI)
P-value
  • *

    Odds ratios adjusted for smoking, hypertension, body mass index, physical exercise, family history and oral contraceptive intake.

Tissue plasminogen activator (ng mL−1)
 < 2.81.0 (ref) 1.0 (ref) 
 2.8–4.62.46 (1.74–3.47) 2.51 (1.62–3.89) 
 > 4.66.05 (3.03–12.04)P < 0.00016.30 (2.62–15.13)P < 0.0001
Plasminogen activator inhibitor 1(ng mL−1)
 < 37.31.0 (ref) 1.0 (ref) 
 37.3–641.29 (0.95–1.74) 1.14 (0.76–1.70) 
 > 641.66 (0.90–3.03)P = 0.101.30 (0.58–2.89)P = 0.529
von Willebrand factor (U dL−1)
 < 901.0 (ref) 1.0 (ref) 
 90–1291.87 (1.37–2.56) 1.62 (1.11–2.37) 
 > 1293.50 (1.88–6.55)P < 0.00012.62 (1.23–5.62)P = 0.013
C-reactive protein (μg mL−1)
 < 0.41.0 (ref) 1.0 (ref) 
 0.4–1.61.62 (1.19–2.21) 1.48 (0.99–2.20) 
 > 1.62.62 (1.42–4.88)P = 0.0022.19 (0.98–4.84)P = 0.054
Fibrinogen (mg dL−1)
 < 3451.0 (ref) 1.0 (ref) 
 345–4151.39 (1.03–1.89) 1.35 (0.91–2.00) 
 > 4151.93 (1.06–3.57)P = 0.0321.82 (0.83–4.00)P = 0.137
D-dimer (ng mL−1)
 < 221.0 (ref) 1.0 (ref) 
 22–381.12 (0.85–1.48) 1.17 (0.82–1.67) 
 > 381.25 (0.72–2.19)P = 0.4021.37 (0.67–2.79)P = 0.378

Table 4 shows the unadjusted and adjusted ORs obtained fitting a model including all the hemostatic factors at the same time. Tissue PA antigen levels remained strongly associated with AMI. Association for VWF was detected at borderline levels and was limited to unadjusted ORs.

Table 4.  Association of hemostatic factors (and C-reactive protein) with acute myocardial infarction in young women according to the distribution of values in tertiles, with all hemostatic factors included in the model
VariableAdjusted ORs*
(95% CI)
P-valueAdjusted ORs
(95% CI)
P-value
  • *

    Odd ratios adjusted for all the hemostatic factors at the same time.

  • †Odd ratios adjusted for smoking, hypertension, body mass index, physical exercise, family history, oral contraceptive intake and for all the hemostatic factors at the same time.

Tissue plasminogen activator antigen (ng mL−1)
 < 2.81.0 (ref) 1.0 (ref) 
 2.8–4.62.37 (1.52–3.70)P < 0.00012.86 (1.63–5.02)P < 0.0001
 > 4.65.62 (2.31–13.69) 8.18 (2.66–25.20) 
Plasminogen activator inhibitor 1 (ng mL−1)
 < 37.31.0 (ref) 1.0 (ref) 
 37.3–640.81 (0.53–1.22)P = 0.3130.65 (0.38–1.13)P = 0.128
 > 640.66 (0.28–1.49) 0.42 (0.14–1.28) 
von Willebrand factor (U dL−1)
 < 901.0 (ref) 1.0 (ref) 
 90–1291.43 (0.99–2.05)P = 0.0501.19 (0.76–1.87)P = 0.443
 > 1292.04 (0.98–4.20) 1.42 (0.58–3.50) 
Fibrinogen (mg mL−1)
 < 3451.0 (ref) 1.0 (ref) 
 345–4151.08 (0.71–1.62)P = 0.7271.13 (0.67–1.90)P = 0.649
 > 4151.17 (0.50–2.62) 1.28 (0.45–3.61) 
D-dimer (ng mL−1)
 < 221.0 (ref) 1.0 (ref) 
 22–380.94 (0.67–1.33)P = 0.7461.08 (0.71–1.62)P = 0.725
 > 380.88 (0.45–1.77) 1.17 (0.50–2.62) 
C-reactive protein (μg mL−1)
 < 0.41.0 (ref) 1.0 (ref) 
 0.4–1.61.03 (0.66–1.60)P = 0.8900.92 (0.53–1.60)P = 0.774
 > 1.61.06 (0.44–2.56) 0.85 (0.28–2.56) 

Using the criteria given earlier, 49 of the 141 patients (39%) had no or non-significant stenosis on coronary arteriography. In them, the frequency distributions of classical risk factors were not significantly different from those of the whole group of women with premature AMI (for hypertension, 20% vs. 26%; smoking, 63% vs. 74%; family history, 53% vs. 62%; obesity 23% vs. 31%, oral contraceptive intake, 49% vs. 41%, physical activity, 47% vs. 35%). After adjustment, only smoking remained a statistically significant risk factor (OR 3.54, 95% CI 1.07–11.74, P = 0.039). In this subset, the pattern of changes of classical and hemostatic factors tended to behave as in the whole cohort (data not shown), but because of the smaller sample size the P-values of the effects increased. Only tPA antigen preserved an effect of borderline statistical significance (adjusted OR 2.04, 95% CI 0.94–4.41, P = 0.071) after fitting a model containing all the hemostatic and classical risk factors.

Discussion

The young women investigated in this study were partially different from the corresponding young men previously investigated in the study of genetic markers [8]. They were less often overweight (31% women vs. 68% men were preobese or obese), less hypercholesterolemic (46 vs. 65%), less physically active (35 vs. 50%) and smoked less (74 vs. 90%). On the other hand, such risk factors as hypertension (26 vs. 25%) and family history (62 vs. 62%) were similarly distributed in young men and women. An important difference was the frequency of normal coronary arteries or non-significant stenosis: 39% in women vs. 21% in men. Hence, the characteristics of this cohort are consistent with the original rationale of the study, i.e. to investigate the behavior of hemostatic factors in a group of women with premature AMI but who were relatively free from atherosclerosis, to dissect optimally any role of heightened hemostasis in the occurrence of coronary thrombosis.

After adjusting for such covariates as classical non-hemostatic risk factors, only tPA antigen and VWF showed statistically significant associations with AMI. Women with tPA antigen in the highest tertile of the distribution (more than 4.6 ng mL−1) had a risk eight times greater than women in the lowest tertile, while women in the highest tertile of VWF had a risk nearly three times greater. The association of VWF (a marker of endothelial cell activation) was not independent, because it was lost after adjusting for both non-hemostatic and hemostatic factors. This is consistent with the views that classical risk factors are confounders or partly operate through their effect on hemostatic factors, and with the existence of a strong interrelationship between hemostatic factors themselves.

Hence only tPA antigen remained strongly and independently associated with AMI in young women. It may appear paradoxical and biologically untenable that the presence of high plasma levels of the main activator of the fibrinolytic system is an adverse factor for the occurrence of AMI in young women. However, increased concentrations of tPA antigen indicate a reduced rather than heightened fibrinolytic activity, because the immunoassay of tPA measures, to a large extent, the circulating complexes between tPA and the main fibrinolysis inhibitor PAI-1 [16]. Tissue PA antigen correlates better than PAI-1 with AMI, both in patients with pre-existing coronary artery disease [17–20] and in healthy persons [21], as also seen in this study. Intraindividual levels of tPA antigen are known to be relatively consistent outside the acute phase, while PAI-1 fluctuates markedly over time [22], perhaps explaining why the former but not the latter fibrinolysis measurement was strongly associated with premature AMI in this study.

This study has limitations. Even though the cases were studied and blood samples were obtained at a distance from the acute episode, when the patients could be considered to be in a relatively stable condition, it cannot be ruled out that high tPA antigen is the expression of a reaction to disease, and hence the consequence rather than the cause of AMI. On the other hand, the magnitude of tPA increase is likely to be underestimated in this study, because at the time of blood sampling the great majority of young women surviving AMI were on long-term treatment with aspirin and statins, which reduce tPA antigen levels [23–25]. Since statins are known to lower CRP [26, 27], the almost consistent intake of these drugs after AMI might explain why this measurement was not independently associated with AMI. Another limit of the study is that, because of the rarity of AMI in young women, this is a highly selected population, however representative it may be of those who developed the disease, at least in the Italian population. The tPA antigen had some association with AMI even in a subgroup of women with no or non-significant coronary stenosis, supporting the view that hypofibrinolysis is not secondary to atherosclerosis. Finally, an obvious limit of the study is that women who died after AMI could not be investigated, so that the possibility that these patients may have had a specially high proportion of high or low tPA antigen values cannot be excluded. On the other hand, it is unlikely that a prospective study of sufficient size and duration will ever be performed in young women to establish prospectively the value of tPA antigen as a predictive marker.

Ancillary