Frits R. Rosendaal, Department of Clinical Epidemiology, Leiden University Medical Centre, Postbus 9600, 2300 RC Leiden, The Netherlands. E-mail: email@example.com
Reduced overall fibrinolytic capacity increases the risk of myocardial infarction (MI), as demonstrated in studies with predominantly male participants. We determined the influence of altered fibrinolysis on the risk of MI and ischaemic stroke (IS) in young women. The RATIO (Risk of Arterial Thrombosis In relation to Oral contraceptives) study is a population-based case-control study including young women with MI (n = 203), IS (N = 175) and 638 matched healthy controls. Fibrinolytic potential was determined with a tissue factor/tissue plasminogen activator induced clot-lysis assay. Odds ratios (OR) adjusted for cardiovascular risk factors were obtained with logistic regression. Clot-lysis time (CLT) was divided into tertiles based on the control group (T1–T3), with T2 as reference. Hypofibrinolysis (prolonged CLT) was associated with an increase in risk of MI (T3 vs. T2, OR 2·8; 95%confidence interval [CI] 1·7–4·7). Hyperfibrinolysis (decreased CLT) had no clear effect (T1 vs. T2, OR 1·6; 95% CI 0·9–2·9). Hypofibrinolysis did not affect the risk of IS (T3 vs. T2, OR 1·5; 95% CI 0·7–3·0), whereas hyperfibrinolysis increased this risk (T1 vs. T2, OR 4·1; 95% CI 2·1–8·0). Oral contraceptive use and smoking further increased these risks. Hypofibrinolysis increases the risk for MI in young women, a finding similar to previous studies. Counter-intuitively, hyperfibrinolysis increased the risk of IS four-fold, which suggests that MI and IS have different aetiologies.
Cardiovascular disease places a large burden on western societies, both on the healthcare systems and quality of life. The two major forms of arterial thrombosis are myocardial infarction and ischaemic stroke (Lloyd-Jones et al, 2010). Arterial thrombosis is most prevalent in the elderly; risk factors, such as diabetes, hypertension and hypercholesterolaemia are also more common with progressing age. Because prevalence of these ‘traditional’ risk factors is less pronounced in the young, new risk factors can more easily be identified in young participants (Lloyd-Jones et al, 2010).
An increased capacity to form blood clots leading to a procoagulant state could increase the risk of arterial thrombosis (Rosendaal, 2003). A procoagulant state may not solely be the result of an increased propensity to initiate or propagate clot formation, but can also result from a decreased ability to dissolve these newly formed clots (Chan et al, 2008). The main fibrinolytic factor is plasmin, which degrades fibrin and thereby dissolves the clot. The zymogen of plasmin, plasminogen, can be activated by tissue plasminogen activator (tPA) or urokinase (Colman & Schmaier, 1997). Plasmin itself can be inhibited directly by α2-antiplasmin. Furthermore, plasminogen activator inhibitor 1 (PAI-1) decreases the formation of plasmin by direct inhibition of tPA and urokinase. Thrombin Activatable Fibrinolysis Inhibitor (TAFI) also hampers fibrinolysis by altering the fibrin structure and thereby reducing binding and activation of plasminogen. Individual factors of the fibrinolytic system, especially PAI-1 and TAFI, have been linked to myocardial infarction and ischaemic stroke although results have been inconclusive and contradictory, as was reviewed recently (Meltzer et al, 2007).
Besides individual factors, clot lysis assays can be used to assess the fibrinolytic capacity. Some studies, but not all, indicate an association between the euglobin clot lysis time (CLT) or dilute whole blood CLT and myocardial infarction. This association might be most pronounced in patients with minimal atherosclerosis (Prins & Hirsh, 1991; Meltzer et al, 2007). Studies on ischaemic stroke are limited (Sharma et al, 1978; Kilpatrick et al, 1993). However, these global tests in these studies do not reflect all the appropriate components of the fibrinolytic system. A previously described test takes all primary components of the fibrinolytic system (plasminogen, alpha2-antiplasmin, PAI-1 and TAFI) into account and thus reflects the true global plasma fibrinolytic potential better (Lisman et al, 2001; Meltzer et al, 2010). The main determinant of this test is the level of PAI-1, followed by plasminogen TAFI, prothrombin and α2-antiplasmin levels (Meltzer et al, 2010).
Two studies indicated that a prolonged CLT in this test is associated with venous thrombosis, especially in combination with other risk factors such as oral contraceptives (Lisman et al, 2005; Meltzer et al, 2008).
Hypofibrinolysis is also associated with arterial thrombosis: prolonged CLT (fourth quartile versus first quartile) was associated with a two-fold increase in risk of myocardial infarction for men aged <50 years, whereas no such association was found for men aged ≥50 years (Lisman et al, 2005; Meltzer et al, 2009). Another case-control study with 330 young patients with coronary heart disease, ischaemic stroke (including transient ischaemic attack) and peripheral arterial disease also showed an increased risk for the combined endpoint arterial thrombosis with increasing CLT (Guimaraes et al, 2009). The inclusion of both sexes (37% men) and the broad case definitions make it difficult to draw conclusions for some subanalyses, such as an interaction analysis focussing on oral contraceptive use and smoking. We therefore set out to determine whether abnormal fibrinolysis, both hyper- and hypofibrinolysis, is associated with both myocardial infarction and ischaemic stroke in young women and whether other risk factors influence this risk.
The RATIO (Risk of Arterial Thrombosis In relation to Oral contraceptives) study is a multicentre case control study set up to identify risk factors for arterial thrombosis in young women, and has previously been described in detail (Tanis et al, 2001; Kemmeren et al, 2002; van den Bosch et al, 2003). Briefly, young women (18–50 years) diagnosed with a form of arterial thrombosis in the 16 participating hospitals including eight academic medical centres in the Netherlands, were approached to participate. Diagnosis of myocardial infarction was based on the presence of symptoms, elevated cardiac-enzyme levels, and electrocardiographic changes, whereas ischaemic stroke was diagnosed on the basis of medical history, neurological examination and computerized tomography or magnetic resonance imaging scan by experienced neurologists in the participating centres. Women without a history of arterial thrombosis were approached by random digit dialling and frequency matched with the case groups on age (in 5 year categories), year of event and area of residence. All participants were asked to fill out a standardized questionnaire on several topics, such as demographic characteristics and medical history, which included oral contraceptive use in the year prior to event or comparable time frame for healthy controls.
The women were reapproached to provide either a blood sample or DNA by means of a buccal swab. Because not all women suffering from ischaemic stroke were able to participate in this second phase of the study, 50 additional cases were approached to participate. Citrated plasma was aliquoted in multiple tubes and stored at −80°C until use. Informed consent was obtained from all participants and the study protocol was approved by the institutional review boards of the participating hospitals.
Citrated plasma used for these measurements was not thawed previously. CLT was assessed by measuring the changes in plasma turbidity during tissue-factor induced clot formation and subsequent lysis by exogenous t-PA (Lisman et al, 2001). Briefly, 50 μl of mixture containing phospholipid vesicles (40%l-α-dioleoylphosphatidylcholine, 40%l-α-dioleoylphosphatidylethanolamine and 20%l-α-dioleoylphosphatidylserine in a final concentration of 10 μmol/l), t-PA (final concentration 56 ng/ml), tissue factor (final dilution 1/1000) and CaCl2 (final concentration 17 mmol/l) diluted in HEPES buffer [25 mmol/l HEPES (N-2-hydroxytethylpiperazine-N’2ethanesulfonic acid) 137 mmol/l NaCl, 3,5 mmol/l KCl, 3 mmol/l CaCl2, 0,1% bovine serum albumin, pH 7.4] was added to 50 μl of citrated plasma, all in a 96-well microtitre plate. After thorough mixing, the plate was placed in a Spectramax 340 kinetic microplate reader at 37°C (Molecular Devices Corporation, Menlo Park, CA, USA). Optical density (OD) was measured every 20 s at 405 nm, resulting in a clot-lysis turbidity profile. CLT was defined as the time from the midpoint of the clear to the maximal turbid transition, representing clot formation, to the midpoint of the maximum turbid to clear transition, representing clot lysis. The laboratory technician was unaware of the case-control status of the samples.
To assess CLT in relation to arterial thrombosis we calculated the mean difference and corresponding 95% confidence interval (CI) of CLT in the control and case groups with the t-test for independent samples. To further study the effect of both hypo- and hyperfibrinolysis CLT was divided into three categories based on the tertiles of the control group. Logistic regression models were used to obtain odds ratios and corresponding 95% CIs as measures of relative risk with normofibrinolysis, defined as the middle tertile, as the reference category. The covariates in these models included as a minimum the stratification factors, age (on a continuous scale), year of event and area of residence. Furthermore, smoking status, body mass index, hypertension, diabetes mellitus and hypercholesterolaemia were considered as confounders and included in subsequent models. As triglyceride levels could also affect CLT, we also additionally included log transformed trigyceride levels in the model (Meltzer et al, 2008, 2009). Triglyceride data were only available for the myocardial infarction and control subgroup.
In total, 248 cases with myocardial infarction, 203 cases with ischaemic stroke and 925 controls were recruited to participate in the first phase of the RATIO study. Thirty cases with myocardial infarction, 60 cases with ischaemic stroke and 158 controls could not be traced, died or refused to participate when re-approached for the second phase. Fifteen cases with myocardial infarction, 15 with ischaemic stroke and 129 controls only provided buccal swabs for DNA extraction but no blood was drawn. Three blood draws were unsuccessful in ischaemic stroke patients. Therefore, we ultimately obtained blood samples from 203 myocardial infarction cases, 175 ischaemic stroke cases and 638 controls.
As expected, the ‘traditional’ risk factors, such as smoking and diabetes mellitus, were more prevalent in the two case groups when compared with the controls (Table I). As age was a matching variable, the three groups were comparable on age. Clot-lysis time was higher among myocardial infarction cases when compared with controls (mean difference 10·8 min, 95% CI 7·2–14·5). CLT in ischaemic stroke cases was shorter (mean difference 3·7 min, 95% CI −2·0–9·4). Figure 1 shows the clot-lysis time for each participant with the horizontal lines indicating the mean, all stratified according to case group.
Table I. Characteristics of participants stratified by case and control status.
Myocardial infarction N = 205
Ischaemic stroke N = 175
Control N = 638
BMI, body mass index; SD, standard deviation; Q1–Q3, 25th and 75th percentile; CLT, Clot-lysis time; NA, not applicable.
Clot-lysis time measurements were unavailable or missing for eight controls, three cases in the myocardial infarction group, and 10 patients in the ischaemic stroke group.
*At moment of event (cases) or index date (controls).
†In the year prior to event (cases) or index year (controls).
‡At time of blood draw.
Mean age*, years
Oral contraceptives use†
Median BMI‡, (Q1–Q3) kg/m2
Median triglycerides‡, (Q1–Q3) mmol/l
Clot-lysis time‡, min
Median (Q1 - Q3)
With the middle category as reference, hypofibrinolysis (or the third tertile [T3] and longest CLT) was associated with a three-fold increase in risk of myocardial infarction (OR 3·15, 95% CI 2·04–4·86; Table II). Adjustment for confounders, including triglyceride levels, decreased this risk slightly (OR 2·84, 95% CI 1·69–4·76). Hyperfibrinolysis, or the first tertile (T1) and shortest CLT, was not associated with myocardial infarction. The effects were more pronounced in women younger than 40 years (Table III). Hypofibrinolysis (T3) increased the risk of ischaemic stroke about twofold (OR 2·07, 95% CI 1·24–3·48), but this risk decreased after adjustment for confounders (OR 1·50, 95%CI 0·74–3·05). Hyperfibrinolysis (T1) also increased the risk of ischaemic stroke (OR 4·07, 95%CI 2·07–8·03 after adjustment). The risks were more pronounced in the young (see Table III). Upon exclusion of oral anticoagulant users (16 ischaemic stroke cases and 19 myocardial infarction cases) these results were largely the same (data not shown).
Table II. Risks of myocardial infarction and ischaemic stroke; tertiles.
Odds ratio (95% confidence interval)
Odds ratio (95% confidence interval)
NA, not applicable; CLT, Clot-lysis time; REF, reference group; T1, 1st tertile of clot-lysis time or hyperfibrinolysis; T2, 2nd tertile of clot-lysis time, or normofibrinolysis; T3, 3rd tertile of clot-lysis time, or hypofibrinolysis.
All tertiles are based on the CLT of the control group. Odds ratios are calculated with the middle tertile as reference. Regression model 1 includes stratification factors age, year of event and area of residence as covariates. Model 2 includes the covariates of model 1 plus smoking, body mass index, hypertension, diabetes mellitus and hypercholesterolaemia. Model 3 includes the covariates of model 2 plus log transformed triglycerides levels.
Table III. Risks of myocardial infarction and ischaemic stroke; younger versus older
OR (95%CI) Model 2
OR (95%CI) Model 2
OR, odds ratio; 95%CI, 95% confidence interval; T1, 1st tertile of clot-lysis time, or hyperfibrinolysis; T2, 2nd tertile of clot-lysis time, or normofibrinolysis; T3, 3rd tertile of clot-lysis time, or hypofibrinolysis.
Tertiles are based on clot-lysis time in control group. Odds ratios and corresponding confidence intervals are calculated with the middle category as reference and with smoking, body mass index, hypertension, diabetes mellitus and hypercholesterolemia as covariables (‘model 2’).
Oral contraceptive use (OC) within the subgroup of participants with normofibrinolysis increased the risk of both myocardial infarction and ischaemic stroke (Table IV) about two to fourfold (T2/− OC vs. T2/+ OC OR 2·4, 95%CI 1·0–5·6 for myocardial infarction and OR 3·7, 95%CI 1·3–11 for ischaemic stroke). This was also true for smoking (T2/− smoking vs. T2/+ smoking OR 3·3, 95%CI 1·4–7·7 for myocardial infarction and OR 1·9, 95%CI 0·7–5·3 for ischaemic stroke). The risk of smoking was highest in women with abnormal CLT when compared with women with normofibrinolysis and who did not smoke (T3/+ smoking OR 14, 95% CI 6·2–31 for myocardial infarction and T1/+ smoking OR 7·2, 95%CI 2·7–19 for ischaemic stroke).
Table IV. Interaction analyses; oral contraceptive use and smoking behaviour
T1, 1st tertile of clot-lysis time, or hyperfibrinolysis; T2, 2nd tertile of clot-lysis time, or normofibrinolysis; T3, 3rd tertile of clot-lysis time, or hypofibrinolysis; OC use, use of oral contraceptives in the year prior to event (cases) or index year (controls); OR, Odds Ratio; 95% CI, 95% confidence interval; REF, reference group.
Tertiles are based on clot-lysis time in control groups. Odds ratios and corresponding confidence intervals are calculated with the middle category as reference and with body mass index, hypertension, diabetes mellitus and hypercholesterolaemia as covariates (‘Model 2’). Smoking was only included in the oral contraceptive use interaction analyses.
Our study indicates that hypofibrinolysis increases the risk of myocardial infarction in young women. Counter intuitively, we found that hyperfibrinolysis is associated with an increased risk of ischaemic stroke. Furthermore, the use of oral contraceptives increases the risk in all strata of CLT.
A decreased fibrinolytic capacity is a plausible causal risk factor for arterial thrombosis and the results for myocardial infarction are in correspondence with current knowledge and the results in earlier studies (Guimaraes et al, 2009; Meltzer et al, 2009). Our results, unexpectedly, indicate also that hyperfibrinolysis increases the risk of ischaemic stroke. This finding could be due to chance, but also raises the question whether the observed association reflects a causal mechanism in which an altered fibrinolytic capacity plays a role in the aetiology of ischaemic stroke or whether it can be explained by other mechanisms. For instance, hyperfibrinolysis in patients suffering from atrial fibrillation could increase the risk of cardioembolic stroke as theoretically an increased potential to dissolve blood clots can lead to instable clots and subsequent embolization. However, the RATIO study was aimed at strokes of arterial origin (i.e. non cardioembolic) and therefore we feel that this mechanism cannot explain our results.
Our counter-intuitive findings on hyperfibrinolysis and ischaemic stroke are not necessarily in contrast with previous findings. The 4G/5G polymorphism of the PAI-1 promotor, which is associated with low levels of circulating PAI-1 (with carriers of the 5G allele having lower levels), has been suggested to be associated with an increased risk of ischaemic stroke (Tsantes et al, 2007, 2008). PAI-1 is one of the main determinants of CLT and is known to have functions, other than decreasing fibrinolytic potential, which could explain our findings (Meltzer et al, 2010). One of these functions is the so-called tPA-serpin axis, which is involved in neuronal damage after cerebral ischaemia (Benchenane et al, 2004). Animal studies suggest that endogenous tPA or rtPA used as treatment after ischaemic stroke could lead to an N-Methyl-d-aspartic acid-mediated Ca2+ influx, which enhances neuronal damage. PAI-1, a serpin that inhibits tPA, could counteract this mechanism. Therefore, low levels of PAI-1, reflected by hyperfibrinolysis in our study, could lead to neuronal damage. In addition, high PAI-1 has been shown to be related to a decreased tendency of plaque rupture (Benchenane et al, 2004). Although atherosclerosis is not abundantly present in these young women, low levels of PAI-1 could be related to increased tendency of plaque rupture and subsequent thrombus formation and thus increase the risk of ischaemic stroke of atherosclerotic origin. So, the observed association between low CLT and increased risk of ischaemic stroke in our study does not necessarily reflect a causal role of increased fibrinolytic tendency but can also indicate other causal mechanisms that include low levels of PAI-1. Given this uncertainty, we can still conclude that our results suggest a clear difference between ischaemic stroke and myocardial infarction.
Our study has some limitations and strengths: although we included patients with primary ischaemic strokes, we cannot stratify our results according to the subtype of stroke (e.g. TOAST [Trial of Org 10172 in Acute Stroke Treatment] criteria) (Adams et al, 1993), which hampers the interpretation of the results. It is possible that our results reflect the causal mechanism of only one stroke subtype, which would imply that the effect for that subtype would even be stronger. Furthermore, due to the case control-design of this study it cannot be ruled out that the differences in clot-lysis time between the case groups are a consequence of the disease or disease treatment, instead of a cause (i.e. ‘reverse causation’). A major source of this bias is the use of blood samples taken during the acute phase of the disease, directly after the event. However, because blood was drawn after the acute phase (mean 82 months, at least 23 months after the event), it is unlikely that the results are explained by this phenomenon. Other sources of this bias, such as changes in cardiovascular risk factors after the event, would have led to an underestimation of the true effect and are not likely to exert a different effect between patients suffering from myocardial infarction and ischaemic stroke. Also, treatment initiated after the event could have introduced a bias: as antihypertensive drugs and statins might increase clot lysis and hence decrease CLT, the effect of CLT on myocardial infarction might be underestimated whereas the effect on ischaemic stroke might be overestimated (Undas et al, 2006). As treatment strategies largely overlap for these two diseases it is not likely that our main finding, the difference in effect of CLT between myocardial infarction cases and ischaemic stroke cases, is to be explained by this potential source of bias. Our study only includes patients that survived a first event, which may affect the results. Therefore our results only apply to those who survived their first event. If the effect of CLT leads to a more severe form of disease, our results are an underestimation of the true effect.
The secondary analyses regarding oral contraceptive and smoking yield six strata and this results in small numbers of participants per stratum, as reflected in the wide confidence intervals. Therefore, these analyses provide some idea into the associations with the combination of risk factors, but do not allow strong conclusions on the presence of interactions.
Increased levels of triglycerides could confound the relationship between clot-lysis time and arterial thrombosis (Meltzer et al, 2009). Unfortunately, triglyceride levels were not available for the ischaemic stroke analyses. However, adjustment for triglyceride levels in the myocardial infarction analyses only minimally changed the associations. Therefore, it is unlikely that effects observed in the ischaemic stroke analyses can be attributed entirely to triglycerides levels.
In summary, we found that in young women hypofibrinolysis increased the risk of myocardial infarction whereas hyperfibrinolysis increased the risk of ischaemic stroke. Although the results obtained from the ischaemic stroke analyses are not easily interpreted and causal conclusions on the mechanism by which hyperfibrinolysis increases the risk of ischaemic stroke cannot be drawn, these results indicate that myocardial infarction and ischaemic stroke may have different causal mechanisms.
The authors further wish to thank the participants and earlier contributors of the RATIO study. Special thank goes out to Jelle Adelmeijer, for the help with the laboratory measurements.
BS: Data analysis, data interpretation and manuscript drafting. MM: Hypothesis formulation, data interpretation and manuscript drafting. PdG, TL: Hypothesis formulation, data interpretation, supervision of CLT analyses and critical revision of manuscript. AA, FRR: Study design of RATIO, data interpretation and critical revision of manuscript.
Conflicts of Interest
The authors declare no conflict of interest.
This study was supported by grants from the Netherlands Heart Foundation, The Netherlands (grant 1997·063, 2001·069 and 2005B060), the Prevention Fund, The Netherlands (no. 28-2879) and the Leducq Foundation, France for the development of Transatlantic Networks of Excellence in Cardiovascular Research (grant 04 CVD 02).