OUP user menu

Tissue plasminogen activator antigen and coronary heart disease
Prospective study and meta-analysis

G.D.O. Lowe, J. Danesh, S. Lewington, M. Walker, L. Lennon, A. Thomson, A. Rumley, P.H. Whincup
DOI: http://dx.doi.org/10.1016/j.ehj.2003.11.004 252-259 First published online: 1 February 2004

Abstract

Aims To determine whether circulating tissue plasminogen activator (t-PA) antigen concentrations are prospectively related to risk of coronary heart disease (CHD) in the general population

Methods and results We measured baseline concentrations of t-PA antigen in the stored serum samples of 606 CHD cases and 1227 controls ‘nested’ in a prospective cohort of 5661 men monitored for 16 years, and conducted a meta-analysis of previous relevant studies to place our findings in context. Tissue plasminogen activator antigen values were strongly correlated with several vascular risk factors, including serum lipids, body mass index, alcohol consumption, and markers of systemic inflammation. In a comparison of men in the top third compared with those in the bottom third of baseline t-PA antigen values, the odds ratio for CHD was 2.20 (95% confidence interval (CI) 1.70–2.85) after adjustment for age and town only, but this fell to 1.48 (1.09–2.01) after further adjustment. Analysis of t-PA as a continuous variable gave similar results. Similarly, when published information on all seven available prospective cohort studies in general populations (2119 cases and 8832 controls in total) was synthesized, the combined odds ratio was 2.18 (1.77–2.69) after adjustment for age and sex only, and this fell to 1.47 (1.19–1.81) after further adjustment.

Conclusion Although there is a statistically significant association between circulating concentrations of t-PA antigen and subsequent CHD, additional studies are needed to determine to what extent this is independent from more established risk factors.

  • Coronary heart disease
  • Epidemiology
  • Fibrinolysis

1 Introduction

Tissue plasminogen activator (t-PA) is a glycoprotein produced mainly by vascular endothelial cells.1,2It activates clot dissolution in the presence of fibrin by conversion of plasminogen to plasmin, thereby cleaving cross-linked fibrin to D-dimer and other degradation products,3and it may also be involved in coronary plaque rupture.2As free active t-PA is difficult to measure in plasma (unless blood is collected into anticoagulants specific for this purpose),1most clinical studies have measured circulating t-PA antigen values.1,2The biological relevance of t-PA antigen is, however, less understood. As a marker of the up-regulation of endogenous fibrinolysis, it might be expected to be associated with a lower incidence of vascular disease. But, as it is mainly a marker of complex formation between t-PA and its major plasma inhibitor, plasminogen activator inhibitor-1 (PAI-1), rather than a measure of free t-PA, it might be expected to be associated with a higher incidence of vascular disease.1,2Tissue plasminogen activator and PAI-1 are also associated with lifestyle variables4and with the inflammatory response, markers of which are related to coronary disease.5

Several epidemiological studies have reported on the relations between circulating t-PA antigen values and coronary heart disease (CHD) in general populations6–11and in cohorts with existing vascular disease.12–21Most studies have involved relatively few CHD cases, and they have yielded apparently conflicting results.6–21To help clarify the epidemiological evidence, we report the largest and most prolonged community based prospective study of circulating t-PA antigen values and CHD thus far, as well as a meta-analysis of available prospective studies to place our findings in context. Given the proposed association between circulating concentrations of t-PA antigen and PAI-1 complexes, we have also conducted a subsidiary meta-analysis of available prospective studies of PAI-1 and CHD in general populations to help interpret our new data.6,10,11,22,23

2 Methods

2.1 Participants

In 1978–1980, 7735 males aged 40–59 years were randomly selected from general practice registers in each of 24 British towns, and invited to take part in the British Regional Heart Study (response rate 78%). Nurses administered questionnaires, made physical measurements, recorded an ECG, and, in 5661 men in 18 of the towns, collected non-fasting venous blood samples, from which serum was stored at −20°C for subsequent analysis.24Additional questionnaires on car ownership and childhood social circumstances (father's social class and childhood household amenities) were mailed 5 years (98% response among survivors) and 12 years (90% response among survivors) after entry, respectively. All men have been monitored subsequently for all-cause mortality and for cardiovascular morbidity, with a follow up loss of <1% to date.25A prospective, nested, case control study, matched for age and town, was established within the cohort. Eligible cases were 279 men who died from CHD and 364 men who had non-fatal myocardial infarction before 1996.4Fatal cases were ascertained through National Health Service Central Registers on the basis of a death certificate with ICD-9 codes 410–414. Non-fatal myocardial infarction was based on reports from general practitioners, supplemented with hospital reports confirming the diagnosis in accordance with World Health Organisation criteria;26a validation study has beenreported.27Cases were ‘frequency matched’ with 1278 controls, on town of residence and age in 5-year bands, of GP record reviews. These were randomly selected from among men surviving to the end of the study period free from incident CHD. Due to limited sample availability, t-PA antigen measurements were available for only 606 of these cases and for only 1227 of these controls.

2.2 Laboratory and statistical analyses

Laboratory workers, blinded to the case-control status of participants, measured serum concentrations of t-PA antigen with an enzyme immunoassay (Biopool AB, Umea, Sweden) previously used in several other prospective cohort studies.4,8,10,12To validate use of serum, we assayed t-PA antigen in paired plasma and serum samples from 56 healthy individuals and observed a correlation coefficient of 0.95 between the values, as well as very similar mean and SD values (Fig. 1). Because of fluctuations of t-PA antigen concentrations within individuals over time, case-control comparisons of measured baseline values tend tounderestimate any association with CHD risk.27–29Measurements of t-PA antigen were made in pairs of samples collected at an interval of 5 years apart in 892 controls in a separate study,8yielding a self-correlation coefficient of 0.52 (Lowe et al.unpublished data from the Edinburgh Artery Study). This was used to estimate the magnitude of regression dilution and to correct for it (see Results).28–30C-reactive protein,31serum amyloid A,31albumin,32white cell count,33Chlamydia pneumoniae IgG and IgA titres,34Helicobacter pylori seropositivity,35fibrin D-dimer,36von Willebrand factor37and homocysteine38were measured as previously described.

Fig. 1

Comparison of serum and plasma t-PA antigen levels (ng/ml).

We pre-specified case-control analyses by thirds of t-PA antigen values in controls, involving unmatched stratified logistic regression fitted by unconditional maximum likelihood (SAS Corporation).31,34,35,37For associations between t-PA antigen and a variety of known and suspected risk factors, emphasis was given to differences more extreme than 2.6SD (2P≈0.01) to make some allowance for multiple comparisons. Evidence of ischaemia on baseline electrocardiogram was according to published criteria.39Using methods that have been described previously,5a meta-analysis was conducted of prospective studies of CHD with greater than one year of follow-up published before mid-2001 reporting on t-PA antigen6–21or PAI-1.6,10,11,22,23Four studies of t-PA antigen could not be included because they did not report separate results for CHD, but they involved fewer than 5% of all the cases in our meta-analysis.18–21Cases were compared only with controls within the same study to avoid potential biases.

3 Results

3.1 Present study

There were highly significant differences between cases and controls with respect to various known vascular risk factors and t-PA antigen (Table 1). Table 2shows that baseline t-PA antigen values in the control population were highly significantly associated with several classical risk factors, including age (2P<0.0001), alcohol consumption (2P<0.00001), body mass index (2P<0.00001), blood pressure (2P<0.0001), total cholesterol (2P<0.00001) and triglyceride (2P<0.00001). There were strong adjusted associations of t-PA antigen values with forced expiratory volume in 1s (2P<0.00001), haematocrit (2P<0.00001), leucocyte count (2P<0.001), and serum concentrations of von Willebrand factor (2P<0.00001), C-reactive protein (2P<0.00001), insulin (2P<0.0001) and urea (2P<0.00001). No strong associations were observed of t-PA antigen values with serum concentrations of homocysteine or amyloid A protein, markers of persistent infection, or indicators of socioeconomic status (data not shown).

View this table:
Table 1

Baseline characteristics of men with coronary heart disease and of age- and town-matched male controls. Values are means±SD or numbers (%)

CharacteristicCases (n=606)Controls (n=1227)P-value
Questionnaire
Age (years)52.6±5.252.5±5.3matched
Current smoker315 (52%)522 (43%)0.0007
Evidence of coronary diseasea216 (36%)248 (20%)<0.0001
Treated diabetic18 (3%)20 (2%)ns
>2 drinks alcohol/day121 (20%)271 (22%)ns
Occupation in social classes I–II139 (23%)343 (28%)0.03
Homeownerb327 (54%)784 (64%)<0.0001
Physical measurements
Body mass index (kg/m2)26.0±3.425.4±3.30.001
Systolic blood pressure (mmHg)152±22147±21<0.0001
Diastolic blood pressure (mmHg)86±1483±13<0.0001
FEV1(l)306±71323±77<0.0001
Blood sample
Total cholesterol (mmol/l)6.62±1.056.21±0.99<0.0001
HDL cholesterol (mmol/l)1.10±0.271.16±0.28<0.0001
Triglyceride (mmol/l)c1.95±1.241.68±1.09<0.0001
t-PA (ng/ml)12.10±6.5010.40±5.76<0.0001
Top third286 (47%)419 (34%)
Middle third192 (32%)409 (33%)
Bottom third128 (21%)399 (33%)
Loget-PA2.49±0.472.34±0.48<0.0001
Median t-PA (interquartile range)12.5 (8.9–17.1)10.5 (7.6–14.7)
  • Comparisons involved student’s t-tests.

  • a Evidence of ischaemia on baseline electrocardiogram or reported history of angina or myocardial infarction.

  • b Information on home ownership was available for only 428 cases and 943 controls.

  • c Geometric mean±approximate SD.

View this table:
Table 2

Comparisons of the levels of risk factors and other characteristics in controls by thirds of t-PA. Values are means±SD or numbers (%)

Top (n=419)Middle (n=409)Bottom (n=399)t+t++
Classical risk factors
Age (years)53.1±4.953.0±5.251.3±5.64.94.2c
Current smoker198 (47%)177 (43%)147 (37%)2.22.8a
>2 alcohol drinks/day125 (30%)93 (23%)53 (13%)5.85.8d
Evidence of CHD at baseline105 (25%)72 (18%)71 (18%)2.22.0
Total cholesterol (mmol/l)6.40±1.006.31±0.985.91±0.937.15.4d
HDL cholesterol (mmol/l)1.14±0.311.16±0.281.17±0.250.30.0
Triglyceride (mmol/l)1.97±1.321.72±1.111.40±0.769.76.5d
Body mass index (kg/m2)26.5±3.525.7±3.023.9±2.811.612.5d
Systolic blood pressure (mmHg)150±20147±21142±215.43.8b
Diastolic blood pressure (mmHg)86±1383±1379±138.75.5d
Markers of inflammation and infection
C-reactive protein (mg/l)2.23±4.721.45±3.400.89±2.6610.07.4d
Serum amyloid A protein (mg/l)7.35±6.637.41±6.036.12±6.433.31.7
Albumin (g/l)44.5±2.544.6±2.344.5±2.61.10.1
White cell count (×109/l)7.5±1.87.2±1.86.9±1.74.23.2a
Chlamydia pneumonia IgG titres (×106)1.87±0.561.79±0.601.77±0.632.42.2
Chlamydia pneumonia IgA titres (×106)2.37±1.862.20±1.782.12±1.621.41.8
Helicobacter pylori seropositivity254 (61%)222 (54%)228 (57%)0.40.5
FEV1(l)306±75324±74338±793.74.9d
Factors related to haemostasis and blood flow
Fibrin D-dimer (ng/ml)69.2±82.278.8±112.578.9±115.82.71.8
Von Willebrand factor (IU/dl)113.7±49.7102.0±42.898.3±43.44.85.7d
Haematocrit (%)44.2±8.241.9±9.841.0±9.75.54.6d
Other blood factors
Homocysteine (μmol/l)16.5±11.915.4±10.013.4±5.12.12.1
Log10insulin2.67±0.752.49±0.742.32±0.776.84.2c
Glucose (mmol/l)5.77±1.555.58±1.345.50±1.142.41.8
Urate (μmol/l)374±72349±63330±6210.78.7d
Creatinine (μmol/l)99±1499±1499±240.50.2
Urea (mmol/l)5.2±1.35.5±1.45.7±1.75.15.8d
  • Adjustments for social class were omitted in the regressions involving markers of socioeconomic status.

  • a P<0.01

  • b P<0.001.

  • c P< 0.001.

  • d P<0.0001.+t tests derived from regression of t-PA values on each characteristic separately adjusting for age and town only.++t tests derived from regression of t-PA values adjusting for age, town, smoking, body mass index, physical activity, and markers of socioeconomic status (including height).

The change in the chi-squared statistic for the strength of association provides a quantitative indication of the impact of stepwise adjustment for potentialconfounding factors.40In a comparison of men in the top third compared with those in the bottom third of baseline t-PA antigen values (tertile cutoffs, >13.0 vs <8.4ng/ml), the odds ratio for CHD was 2.20 (95% CI 1.70–2.85; χ21=36) after adjustment for age and town only (Table 3). Additional adjustment for smoking status reduced the odds ratio to 2.09 (1.62–2.72) and the corresponding chi-squared values fell from 36 to 31. Further adjustment for baseline values of serum lipids, blood pressure, body mass index, and physical activity reduced the odds ratio to 1.60 (1.20–2.13) and the chi-squared value to 10, and when additional correction was made for markers of lifetime socioeconomic status the odds ratio was 1.48 (1.09–2.01: Table 3) and the chi-squared value was only 6. In an analysis restricted to the 404 cases and the 1007 controls without evidence of CHD at baseline, the odds ratio was 1.49 (1.03–2.14; χ21=5) after adjustment for baseline values of all these factors. The findings were unaffected by varying the pre-specified cut-off level of t-PA antigen for analysis (such as analyses by quarters, fifths, or by one standard deviation increases in t-PA or a continuous variable: data available upon request).

3.2 Meta-analysis of previous prospective studies of t-PA antigen

Twelve relevant prospective studies of t-PA antigen published by mid-2001 were identified, including 6 in general populations,6–11one in a cohort defined on the basis of peripheral vascular disease,12two in cohorts defined on the basis of angina,13,17and three in cohorts defined on the basis of myocardial infarction (Fig. 2).14–16Including the present study, there were a total of 2119 cases of fatal CHD or non-fatal myocardial infarction and 8832 controls, with a mean weighted age at entry of 54 years and mean weighted follow-up of 8 years. All used enzyme-linked immunoassays, and all used plasma samples. Seven of these studies were conducted in general populations, involving a total of 1669 CHD cases and 5635 controls (all of which reported associations adjusted for age and sex only, as well as associations adjusted for age, sex, smoking, blood pressure, and blood lipids, and, in four studies (including the present one)7,10,11body mass index). A combined analysis yielded an odds ratio of 2.18 (1.77–2.69) in individuals in the top third versus those in the bottom third of baseline t-PA antigen values when associations reported in these seven studies were adjusted for age and sex only (corresponding to mean ‘usual’ t-PA antigen values of 13.5 vs 8.0ng/ml, using a ‘self-correlation coefficient’ of 0.52 to correct for regression-dilution). When adjusted for age, sex, and baseline values of classical vascular risk factors, however, a combined analysis yielded an odds ratio of only 1.47 (1.19–1.81: Fig. 3), and the corresponding chi-squared values fell from 55 to 13. Similarly, among the six studies in cohorts defined on the basis of previous vascular disease (450 CHD cases and 3197 controls in total), a combined analysis yielded an odds ratio of 3.23 (1.71–6.09) adjusted for age and sex only, but only 1.32 (0.70–2.50) after additional adjustment, and the corresponding chi-squared values fell from 13 to 0.8. There was only marginal evidence of heterogeneity among the 13 separate studies (χ212=23; P=0.03).

View this table:
Table 3

Odds of coronary heart disease and 95% confidence intervals in men who were in the top third (>13.0ng/ml) compared with those in the bottom third of t-PA concentrations (<8.4ng/ml), with increasing degree of adjustment for baseline values of potential confounding factors

Age and townonlyAge, town, andsmokingAll preceding plus other risk factorsaAll preceding plus adult SESbAll preceding plus lifetime SESc
All cases and all controls
2.20 (1.70–2.85)2.09 (1.62–2.72)1.60 (1.20–2.13)1.57 (1.17–2.10)1.48 (1.09–2.01)
χ21=36χ21=31χ21=10χ21=9χ21=6
Only participants without evidence of CHD at baseline
2.03 (1.49–2.77)1.95 (1.42–2.66)1.57 (1.11–2.23)1.57 (1.10–2.23)1.49 (1.03–2.14)
χ21=20χ21=17χ21=6χ21=6χ21=5
  • a Risk factors=total cholesterol, HDL cholesterol, triglycerides, body mass index, blood pressure, physical activity.

  • b Adult SES=occupation, housing tenure, marital status, car ownership.

  • c Lifetime SES=adult SES plus father's social class, family car ownership, bathroom in house, hot water tap in house, bedroom sharing, height.

Fig. 2

Prospective studies of tissue plasminogen activator and coronary heart disease published before mid-2001. Odds ratio compare top and bottom thirds of baseline measurements. Black squares indicate the odds ratio in each study, with the square size proportional to the number of cases and the horizontal lines representing 99% confidence intervals—no adjustment reported for possible confounders; +, adjustment for age and sex only; ++ for these plus smoking; +++, for these plus some other classical vascular risk factors.

Fig. 3

Prospective studies of tissue plasminogen activator antigen and coronary heart disease including the present study. Odds ratios compare top and bottom thirds of baseline measurements. Squares indicate the odds ratio in each study, with the square size proportional to the number of cases and the horizontal lines representing 95% confidence intervals.

3.3 Meta-analysis of previous prospective studies of PAI-1

Five prospective studies of PAI-1 in general populations were identified involving a total of 833 CHD cases and 3122 controls.6,10,11,22,23They had a mean weighted age at entry of 58 years and mean weighted follow-up of 5 years. One study measured free, active PAI-1,23two measured PAI-1 antigen,6,11one used a chromogenic assay of PAI-1 activity,10and one did not specify the assay method used.22There was no significant heterogeneity among the five separate studies (χ24=5.5; P>0.1), and a combined analysis yielded an odds ratio of 0.98 (0.53–1.81) in individuals in the top third versus those in the bottom third of baseline PAI-1 values.

4 Discussion

Previous studies have generally involved too few CHD cases to determine reliably whether there is a statistically significant association between circulating concentrations of t-PA antigen and subsequent CHD, independent of known cardiovascular risk factors. Whereas the previous largest published prospective study included 326 patients with CHD and 720 controls,6the present report involves almost twice as much new data, involving 606 CHD cases and 1223 controls. It also includes a meta-analysis of available prospective studies, increasing the numbers available for analysis to 2119 CHD cases and to 8832 controls. A combined analysis of these studies, based on published odds ratios that wereadjusted for some risk factors, suggests that CHD risk is about 50% greater in those in the top third compared with those in the bottom third of baseline t-PA antigen values. The relationship of t-PA antigen to CHD, however, still remains uncertain. A tendency to more extreme odds ratios reported in the smaller studies in Fig. 1suggests the likelihood of some exaggeration in the overall estimate due to ‘publication bias’.41Moreover, because incomplete adjustment for some risk factors reduced the odds ratio substantially, it is not known how much, if any, of the residual association between t-PA antigen and CHD would persist with more complete adjustment for these (and other) factors. On the other hand, we have demonstrated that the predictive ability of t-PA antigen for CHD has been under-estimated in previous studies due to lack of correction for its within-individual variation over time.

In the present study, a comparison of those in the top third with those in the bottom third of baseline t-PA antigen values yielded an odds ratio for CHD of 2.20 (1.70–2.85; χ21=36) after adjustment for age and town only, and of 1.48 (1.09–2.01;χ21=6) after further adjustment for classical CHD risk factors. A combined analysis of all six available prospective studies in approximately general populations yielded an odds ratio of 2.18 (1.77–2.69; χ21=55) based on published associations that were adjusted for age and sex only; and an odds ratio of 1.47 (1.19–1.81; χ21=13: Fig. 3) based on published associations that were adjusted for age, sex, and baseline values of some risk factors. This pattern of attenuation in the odds ratio for CHD with increasing degree of adjustment was also seen in the six prospective studies of patients with previous vascular disease. The fact that adjustments based just on baseline measurements of these risk factors reduced the chi-squared values so substantially suggests that exact adjustment for these (and other) confounders would produce even greater reductions. The estimates of effect in patients with previous vascular disease, although rather weaker than those in patients without previous vascular disease, are not markedly different.

Epidemiological studies of t-PA antigen consistently show strong correlations with PAI-1 activity or antigen.42This association may reflect simultaneous release from endothelial cells, delayed clearance of t-PA-PAI-1 complexes, acute-phase reactions, or mutual correlations with measures of insulin resistance.43Our subsidiary meta-analysis of available prospective studies of CHD and PAI-1 was largely inconclusive, yielding a null odds ratio with a wide confidence interval (0.98, 0.53–1.81). Larger observational studies involving measurement of both t-PA antigen and of PAI-1 in the same participants (enabling assessment of the relevance of each factor to the other and to CHD) and involving serial measurements of t-PA antigen and of several potential confounding factors (enabling correction for fluctuations of values of risk factors within individuals over time)29,30should help to elucidate the strength of the association seen between t-PA antigen and CHD.

Potential limitations of the present study include measurement of t-PA antigen in serum instead of plasma, and prolonged storage at −20°C (11–16 years). However, our validation study showed high correlation (r=0.95) between serum and plasma t-PA antigen levels (with very similar mean and SD values), as in a previous report.44We recently performed a study of repeat t-PA antigen assays after 9 years’ storage at −50°C in 255 samples. There was no significant degradation in t-PA values (mean and SD) during this time (Rumley et al., unpublished). Moreover, values for serum t-PA antigen levels in CHD cases and controls in the present study were very similar to those reported for plasma t-PA antigen (usually stored for a shorter period of time) in previous prospective studies,6–21and showed the expected correlations with some classical vascular risk factors.4

What is the potential biological significance of the positive association between circulating t-PA antigen and risk of CHD? The possible influence of PAI-1 on t-PA antigen has already been noted. Tissue plasminogen activator is released from vascular endothelium, hence increased circulating levels may be a marker of endothelial disturbance: in the present study, circulating t-PA antigen correlated strongly with circulating levels of von Willebrand factor (vWF), another endothelial release product,37as well as with risk markers associated with endothelial dysfunction (Table 2). Tissue plasminogen activator (and PAI-1) levels also increase as part of the inflammatory response:43in the present study, circulating t-PA antigen correlated with several inflammation-related measures (e.g., C-reactive protein, leucocyte count). Hence it is possible that the association of t-PA with CHD may partly reflect their mutual associations with the inflammatory response. While increased circulating free t-PA might increase fibrin lysis, no significant association of t-PA antigen was observed with fibrin D-dimer levels in the present study. Finally, it has been proposed that t-PA may play a role in coronary plaque rupture.2In a prospective study of patients with stable angina, both t-PA antigen and leucocyte elastase (another protease which may play a role in plaque rupture) were predictive of myocardial infarction (which is usually preceded by plaque rupture).45However, at present the pathophysiological relevance of circulating levels of t-PA to plaque rupture remains uncertain.

5 Conclusions

There is a statistically significant association between circulating concentrations of t-PA antigen and subsequent CHD, but additional studies are needed to determine to what extent this association is independent from more established risk factors, and to determine the biological significance of t-PA antigen.

Acknowledgments

The views expressed in this paper are those of the authors and not necessarily the funding agencies. Professor A. G. Shaper established the British Regional Heart Study, which is a British Heart Foundation Research Group and also receives support form the Department of Health. Paul Appleby plotted the figures. K. Craig, F. Key, and L. Oxford provided t-PA, von Willebrand factor and fibrin D-dimer assays; Professor M. B. Pepys and J. R. Gallimore provided C-reactive protein and serum amyloid A assays; H. Refsum and P. Ueland provided homocysteine assays; M. Thomas, Yuk-ki Wong and Professor M. Ward provided C pneumoniae serology; J. Atherton and Professor C. Hawkey provided H pylori serology; and J. John provided valuable assistance. J.D. was supported by a programme grant from the British Heart Foundation and by the Raymond and Beverly Sackler Award in the Medical Sciences. A.R. and G.D.O.L. are supported by project and programme grants from the British Heart Foundation.

References

View Abstract