Decreased fibrinolytic potential in South Asian women with ischaemic cerebrovascular disease

Authors


Dr Kirti Kain, Academic Unit of Vascular Medicine, G Floor, Martin Wing, Leeds General Infirmary, Leeds LS1 3EX, UK. E-mail: k.kain@leeds.ac.uk

Abstract

To investigate gender differences in conventional, coagulation and fibrinolytic factors in South Asian ischaemic stroke patients, we compared these variables in 50 South Asian females (SAFP) with 90 South Asian males (SAMP) with ischaemic stroke and in 52 females (SAFC) and 38 males (SAMC) without stroke. Plasminogen activator inhibitor-1 (PAI-1) antigen levels were significantly higher in SAFP compared with SAMP (18·2 vs. 13·3 U/ml, P = 0·04) even after adjustment for known covariates, but there was no difference in PAI-1 antigen levels between males and females in the control group. South Asian females exhibited higher levels of factor VII antigen and FVII:C activity in both stroke patients (114 vs. 99% in males, P = 0·01; 116 versus 104% in males, P = 0·04) and controls (116 vs. 97% in males, P = 0·004; 115 vs. 93% in males, P = 0·01). There were no significant differences in the levels of fibrinogen (3·8 vs. 3·7 g/l), FXIIa (2·2 vs. 2·4 ng/ml), von Willebrand factor (1·8 vs. 1·9 IU/ml) and tissue plasminogen activator (11·4 vs. 12·0 ng/ml) in SAMP and SAFP respectively. These results suggest that South Asian females have increased FVII levels and that females with a history of ischaemic stroke have a decreased fibrinolytic potential in comparison with males.

Stroke is a major public health issue, the frequency of which is increasing in white females more than in males because of increased life expectancy (Bousser, 1999). However, the mortality from cerebrovascular disease among migrants from south Asia (India, Pakistan and Bangladesh) is one and half times higher compared with the indigenous white population in the UK. South Asian males have a higher mortality than South Asian females (Wild & McKeigue, 1997).

Epidemiological studies demonstrate that there are gender differences in the prevalence of traditional risk factors that are strongly related to the development of atherosclerosis (Gregor et al, 1994; Wilhelmsen et al, 1997). However, the relative susceptibility to these vascular risk factors between the genders is not established, although there are some gender differences in sensitivity to traditional risk factors. For example, white diabetic females have an increased relative risk (RR) of ischaemic stroke (1·7) compared with white males with diabetes mellitus (1·4) (Kannel & McGee, 1979). The presence of atrial fibrillation increases the stroke risk to a greater extent in females than in males (RR 3·16 vs. 1·83) (Wolf et al, 1991). The risk of completed stroke after transient ischaemic attack is twice that in white females compared with white males (Whisnant et al, 1996).

However, not all the stroke risk can be accounted for by the conventional risk factors, which has lead to increased interest in the possible role of coagulation and fibrinolytic factors in the pathogenesis of occlusive vascular disease. The available evidence supports a positive association between elevated levels of haemostatic risk factors and the occurrence of vascular disease such as stroke (Wilhelmsen et al, 1984; Ridker et al, 1994) in some studies, but not in all (Heywood et al, 1997; Lowe et al, 1998).

There is no information about the gender differences in atherothrombotic risk factors in South Asian ischaemic stroke patients. Therefore, the aim of this study was to investigate differences in conventional, coagulation and fibrinolytic factors in South Asian males and females with and without ischaemic stroke.

Subjects and methods

Subjects One hundred and forty South Asian patients with ischaemic stroke and 90 subjects free from stroke were recruited from the West Yorkshire area. Each subject was originally from India, Pakistan or Bangladesh, and all four grandparents were from one of these three countries. All the subjects received a volunteer information sheet in English and different translations, namely Gujrati, Punjabi, Bengali, Hindi and Urdu, according to the subject's ethnic background. Patients were recruited 3 months after an acute ischaemic stroke to minimize the influence of the acute-phase response observed in levels of coagulation and fibrinolytic measures. The diagnosis of vascular ischaemic stroke was confirmed clinically according to the World Health Organization (WHO) definition and on computerized cranial tomography scan appearances.

Body mass index (BMI) was expressed as weight (kg) divided by height (m) squared. Blood pressure was calculated from a mean of three resting systolic and three diastolic blood pressures. Current smoking habit was recorded and current drug therapy noted. Subjects gave their informed consent to participate. The study was approved by the Leeds Teaching Hospitals and Bradford Teaching Hospitals (NHS Trust) Research Ethics Committee.

Blood sampling A venous blood sample was taken from each subject between 08.00 and 10.00 using a 19-gauge butterfly needle from an antecubital vein. All patients were in a fasting state and had abstained from smoking for the preceding 12 h. Blood was drawn into 0·9% citrate (pH 8·8) on ice (in a ratio of nine parts blood to one of citrate) for assay of plasminogen activator inhibitor-1 (PAI-1) and tissue plasminogen activator (t-PA), and lithium heparin on ice for analysis of insulin levels. Samples were centrifuged at 1560 gand 4°C for 30 min. Blood was also taken into 0·9% citrate at room temperature for analysis of FVII:C, von Willebrand factor (VWF), FXIIa and fibrinogen and spun at room temperature for 20 min. Plasma was aliquoted, snap frozen and stored at −40°C until assay. Blood was also collected into lithium fluoride for plasma glucose estimation, lithium heparin for plasma lipid analysis and EDTA for HbA1c and DNA analysis.

Laboratory assays PAI-1 activity was measured by chromogenic assay (Spectrolyse, Biopool) and PAI-1 antigen, t-PA antigen by enzyme-linked immunosorbent assay (ELISA; Imulyse, Biopool). Plasma FVII:C was assayed using the ACL 3000 plus (Instrumentation Laboratory), with factor VII-deficient plasma and rabbit thromboplastin (Instrumentation Laboratory) as reagents. FVII:C was expressed as a percentage of the activity given by calibration plasma. Fibrinogen was measured by the Clauss method (Clauss, 1957) and VWF and plasma insulin by an ELISA (Dako). FXIIa was measured by a direct enzyme immunoassay, which detects both µ-XIIa and β-XIIa in human plasma (Shield Diagnostics, Dundee, UK) (Ford et al, 1996). Plasma glucose was measured by the glucose oxidase method. Cholesterol and triglyceride levels were determined using the Hitachi 747 autoanalyser (Boehringer Mannheim). HbA1c was measured by Glycomat autoanalyser (Ciba Corning), with a reference range of 4·5% to 6·5%. Interassay and intra-assay coefficients of variation were: PAI-1 antigen 9% and 5%; PAI-1 activity 9% and 6·3%; t-PA 10% and 8%; fibrinogen 3·5% and 2%; FVII:C 4·3% and 3·2%; FXIIa 9·29% and 9·9%; VWF 4·7% and 2·8%; insulin 5·6% and 5·3% at 18 µIU/ml and 9·8% and 3·0% at 84 µIU/ml respectively.

Genotype analysis of FVII Arg/Gln-353 and PAI-1 4G/5G promoter polymorphisms A coding (arg-gln) polymorphism in the eighth exon of the FVII gene was genotyped by polymerase chain reaction (PCR) amplification of the genomic DNA and digestion with MspI enzyme (Green et al, 1991). The PAI-1 4G/5G promoter polymorphism genotype was determined by PCR endonuclease digestion (Margaglione et al, 1997).

Statistical analysis Patients' ages were not normally distributed and are shown as the median (range). The Mann–Whitney U-test was used to compare values of these between the two genders. A χ2-test was used to compare the frequency of categorical variables such as diabetes mellitus, hypertension, coronary artery disease, previous ischaemic stroke and history of smoking by gender. Non-parametric tests were used to compare PAI-1 antigen/activity levels between the two genders. Values for FXIIa, FVII:C, FVII antigen and VWF were normally distributed.

Values for BMI, triglycerides, fasting blood glucose, fibrinogen, insulin, t-PA and HbA1c were log transformed to normalize their distribution and permit use of parametric statistical analysis. Unpaired Student's t-test was used to compare mean values of these variables by gender. Factorial anova was performed for PAI-1 antigen and FVII:C with age, BMI, HbA1c, plasma insulin triglyceride and cholesterol as covariates. For FVII levels, genotype at the coding polymorphism was entered into a factorial anova model as a code, in which those patients with the most common polymorphism Arg/Arg genotype were coded as 2 and those carrying the Gln allele as 1.

All the statistical analysis was performed with SPSS for Windows version 8·0 (SPSS, Chicago, IL, USA).

Results

Country of origin

A total of 56% of the 140 ischaemic stroke patients were from Pakistan, 8% from Bangladesh and the remaining patients were migrants from India. In the control group, 41% were from Pakistan, 2% from Bangladesh and the rest were from India. BMI and waist–hip ratio (WHR) were similar in subjects from India, Pakistan and Bangladesh within both the stroke group and the control group. The number of subjects was matched for gender, and South Asian males and females were matched for age within both the stroke and control groups.

Stroke patients

The clinical and metabolic characteristics of the 90 South Asian male patients (SAMP) and 50 South Asian female patients (SAFP) with ischaemic stroke are given in Table I. The groups were of similar ages and BMI with no significant differences in the prevalence of diabetes mellitus, hypertension and coronary artery disease. The percentage of subjects with previous strokes was high in both genders at 91% and 84% for SAMP and SAFP respectively. Some 94% of all SAFP were post-menopausal, and 2% were on hormone replacement therapy. Patients with type 2 diabetes mellitus were commenced on oral therapy (biguanides and/or sulphonyl ureas) when first diagnosed but, at the time of recruitment, 13% of male patients and 12% of female patients were on insulin treatment. Altogether, 18% of SAMP were current smokers, and there were no SAFP smokers. There were no differences in fasting blood sugar, cholesterol or triglycerides. High-density lipoprotein (HDL)-cholesterol was significantly increased in SAFP.

Table I.  Clinical features of 140 South Asian stroke subjects.
CharacteristicMales (n = 90)Females (n = 50)P
  1. NS, not significant. Age is median (range); BMI is geometric mean (standard deviation).

Age (years)65 (44–96)67 (43–88)NS
Diabetes mellitus (%)5850NS
Hypertension (%)5162NS
Smokers (%)1800·001
Ischaemic heart disease (%)2414NS
Previous stroke (%)9184NS
BMI (kg/m2)24·3 (5·7)24·1 (4·7)NS
Waist–hip ratio1·0 (1·1)0·92 (1·2)< 0·0001
Aspirin/warfarin (% ratio)67/360/4NS

As shown in Table II, PAI-1 antigen and FVII:C levels were significantly higher in SAFP. There was also a non-significant trend towards higher levels of t-PA in SAFP. Fibrinogen, VWF and FXIIa levels were similar in both groups. Insulin resistance, as determined by homeostasis model assessment (HOMA) (Matthews et al, 1985) showed no significant differences by gender.

Table II.  Biochemical, coagulation and fibrinolytic measurements in South Asian stroke patients.
MeasurementMales (n = 90)Females (n = 50)P
  1. Values are mean (standard deviation); NS, not significant.

  2.   *Insulin resistance is fasting blood sugar × fasting insulin/22·5.

  3.   †Median (interquartile range).

Cholesterol (mmol/l)5·2 (1·2)5·4 (1·0)NS
Triglycerides (mmol/l)1·9 (1·6)1·9 (1·7)NS
HDL (mmol/l)0·96 (0·3)1·1 (0·3)0·01
Glucose (mmol/l)7·2 (1·6)6·6 (1·5)NS
HbA1C (%)6·6 (1·3)6·0 (1·2)0·04
Insulin (mU/l)11·1 (2·4)12·2 (2·0)NS
HOMA*7·4 (12·1)6·5 (14·9)NS
FVII:C (%)104 (30)116 (30)0·04
FVII:ag (%)99 (31)114 (34)0·01
VWF (IU/ml)1·8 (0·8)1·9 (0·8)NS
FXIIa (ng/ml)2·2 (1·0)2·4 (1·1)NS
Fibrinogen (g/l)3·8 (1·3)3·7 (1·3)NS
PAI-1 activity (U/ml)16·1 (11·7)19·3 (10·8)0·04
PAI-1 antigen (U/ml)13·3 (11·0)18·2 (13·5)0·04
tPA antigen (ng/ml)11·4 (1·6)12·0 (1·7)NS

Factor VII:C correlated with WHR, triglycerides, fasting glucose and fasting insulin in SAFP. There were no conventional correlates in the SAMP (Table III). PAI-1 antigen correlated with age, WHR, serum triglycerides, fasting glucose and fasting insulin in SAMP and with insulin in SAFP (Table III). However, after adjustment for age, the only significant correlates of PAI-1 antigen were WHR and triglycerides in SAMP.

Table III.  Correlation coefficients between plasminogen activator inhibitor-1 antigen, factor VII:C and variables associated with the metabolic syndrome in males (m) and females (f) in South Asian stroke patients.
 Factor VII: C (%)PAI-1ag (U/ml)
 mfmf
  1. Values are correlation coefficients, r-value (P-value).

  2.   Bold values refer to P-value < 0·05.

n90509050
Age (years)−0·08 (0·44)−0·1 (0·52)0·22 (0·04)0·19 (0·23)
Waist–hip ratio0·27 (0·08)0·30 (0·05)0·32 (0·003)0·17 (0·29)
Body mass index0·05 (0·65)0·26 (0·13)0·22 (0·054)0·26 (0·14)
Systolic BP (mmHg)0·15 (0·18)−0·012 (0·93)0·03 (0·80)−0·16 (0·37)
Diastolic BP (mmHg)0·05 (0·61)−0·18 (0·22)0·19 (0·10)−0·22 (0·19)
Serum triglycerides (mmol/l)0·15 (0·17)0·502 (0·000)0·44 (0·000)0·09 (0·56)
HDL cholesterol (mmol/l)−0·02 (0·86)0·05 (0·76)−0·15 (0·18)−0·10 (0·54)
LDL cholesterol (mmol/l)0·12 (0·27)0·12 (0·44)0·041 (0·71)0·03 (0·85)
Fasting glucose (mmol/l)−0·02 (0·84)0·302 (0·04)0·30 (0·005)0·10 (0·54)
Fasting insulin (µU/ml)0·09 (0·42)0·42 (0·004)0·30 (0·005)0·33 (0·03)
HOMA-IR0·06 (0·61)0·38 (0·01)0·36 (0·001)0·26 (0·1)

There was no difference in the frequency of PAI-1 4G/5G genotypes between the two groups and no significant association of the genotype with PAI-1 antigen levels (data not shown). At the factor VII Arg/Gln-353 polymorphism, 25% of SAMP and 27% of SAFP carried the Gln allele. However, FVII:C levels were significantly associated with factor VII Arg/Gln-353 polymorphism only in SAMP (P = 0·02).

In a factorial anova model, gender was a significant and independent predictor of PAI-1 antigen levels. Levels of PAI-1 antigen expressed as the geometric mean adjusted for age, BMI, smoking habit, HbA1c and levels of insulin, triglycerides and cholesterol were 9·3 U/ml for SAMP and 15·7 U/ml for SAFP (P = 0·003). Gender accounted for 8% of the interindividual variation in PAI-1 antigen levels in the model, which had an R2 value of 0·25. In a linear regression analysis with age, gender, BMI, triglycerides, blood glucose, insulin, smoking and PAI-1 4G/5G genotype in the model, only age, gender and triglycerides remained significant (R2 = 0·28, P = < 0·0001).

After adjustment for age, BMI, smoking habit, HbA1c and the levels of insulin, triglycerides and cholesterol, gender did not predict FVII antigen or FVII:C levels (P = 0·06). Fibrinogen, FXIIa and VWF levels remained similar when allowing for these other covariates.

To determine significant differences in thrombotic risk factors between SAFP and SAMP, a logistic regression analysis was performed with gender as the dependent variable and smoking, WHR, HDL-cholesterol, HbA1C, factor VII antigen, PAI-1 antigen, total cholesterol, triglycerides and insulin as covariates. Only PAI-1 antigen [odds ratio (OR) 1·05; confidence interval (CI) 1·002–1·095), P = 0·006] was significantly associated with female gender.

Controls

The clinical, metabolic and haemostatic characteristics of the 38 South Asian male controls (SAMC) and 52 South Asian female controls (SAFC) are given in Table IV. The WHR was increased in SAMC. Some 93% of the SAFC were post-menopausal. There were no differences in PAI-1 antigen/activity levels by gender. Age-adjusted correlates of PAI-1 antigen in males and females were similar, the P-values in males being: WHR, P = 0·03; triglycerides P = 0·002; HDL, P = 0·05; insulin, P = 0·01; and insulin resistance, P = 0·02 and, in females, WHR, P = 0·04; triglycerides, P = 0·03; HDL, P = 0·01; insulin, P = 0·001; and insulin resistance, P = < 0·0001.

Table IV.  Biochemical, coagulation and fibrinolytic measurements in South Asian control subjects.
MeasurementMales (n = 38)Females (n = 52)P
  1. Values are mean (SD).

  2.   *Insulin resistance is fasting blood sugar × fasting insulin/22·5.

  3.   †Median (interquartile range); age is median (range)

Age (years)59 (35–46)60 (49–75)NS
Diabetes mellitus (%)2133NS
Hypertension (%)2431NS
Smokers (%)32NS
Ischaemic heart disease (%)104NS
Waist–hip ratio0·97 (0·062)0·91 (0·057)0·001
Cholesterol (mmol/l)5·65 (0·89)5·65 (1·03)NS
Triglycerides (mmol/l)1·68 (1·7)1·92 (1·7)NS
HDL (mmol/l)1·14 (1·4)1·10 (1·4)NS
Glucose (mmol/l)6·20 (2·0)6·69 (3·4)NS
HbA1C (%)5·94 (1·3)6·20 (1·8)NS
Insulin (mU/l)9·83 (2·2)1·02 (1·8)NS
HOMA*2·69 (2·5)2·32 (2·2)NS
FVII:C (%)93 (1·5)115 (1·4)0·01
FVII:ag (%)97 (24)116 (32)0·004
VWF (IU/ml)1·45 (0·7)1·51 (0·4)NS
FXIIa (ng/ml)2·22 (1·6)2·19 (1·7)NS
Fibrinogen (g/l)3·25 (1·4)3·58 (1·2)NS
PAI-1 activity (U/ml)16·8 (16)20·5 (24)NS
PAI-1 antigen (U/ml)13·5 (18)22·5 (27)NS
tPA antigen (ng/ml)10·99 (3·4)11·87 (5·3)NS

FVII:C and FVII antigen levels were elevated in females compared with males. There was no significant difference in the distribution of factor VII Arg/Gln-353 genotype in males and females (data not shown). The FVII antigen or activity was not significantly related to the Arg/Gln-353 genotype in males or females (data not shown).

Mean FVII antigen levels after adjustment for age, BMI, smoking habit, HbA1c, insulin triglycerides and cholesterol were 95% for males and 118% for females, P = < 0·0001. There were no significant age-adjusted correlates of FVII antigen in either males or females. In logistic regression analysis with gender as the dependent variable and the same covariates as used for stroke patients, FVII antigen was associated with female gender, OR = 1·031 (CI 1·007–1·056), P = 0·012.

Discussion

There are differences in the epidemiology, presentation and prognosis of cerebrovascular disease in white males and females (Mosca et al, 1997; Mendelsohn & Karas, 1999). The classical risk factors hypertension, diabetes and cigarette smoking play a role in ischaemic stroke in whites. However, these conventional risk factors do not fully explain the risk of ischaemic stroke in whites. Elevated levels of fibrinogen, FVII:C, PAI-1, t-PA, VWF and FXIIa appear to play an important role in the pathogenesis of vascular events in whites (Wilhelmsen et al, 1984; Meade et al, 1986; Hamsten et al, 1987; Kelleher et al, 1992; Ridker et al, 1994; Thompson et al, 1995), and gender differences in these factors have been demonstrated in white subjects with coronary artery disease (Ossei-Gerning et al, 1998) and with type 2 diabetes mellitus (Mansfield et al, 1996).

The South Asian population resident in the UK is at increased risk of ischaemic stroke compared with whites. Between 1989 and 1992, the standardized mortality ratio for cerebrovascular disease for South Asian subjects in the UK was 155 for males and 141 for females compared with whites taken as 100. Gender differences in conventional risk factors exist in South Asians, in a manner similar to whites. There is an increased prevalence of type 2 diabetes and hypertension in South Asians, particularly in South Asian females (Cappuccio et al, 1997). In contrast, the prevalence of current smokers is low in South Asian females compared with South Asian males (Cappuccio et al, 1997). Hughes et al (1998) demonstrated that South Asian males in Singapore in both diabetic and non-diabetic groups had increased prevalence of smoking, higher WHRs and lower HDL-cholesterol levels compared with South Asian females.

In the present study, there was a higher prevalence of smoking, higher WHR, lower levels of HDL-cholesterol and higher HbA1c in South Asian males compared with South Asian females in the ischaemic stroke group. Levels of HDL-cholesterol are gender specific, and higher levels may contribute to the relative rarity of coronary artery disease in younger females. The contribution of HDL-cholesterol to stroke in whites or South Asians is not known. WHR was higher in males in the control group. In contrast, the prevalence of diabetes mellitus, hypertension and previous coronary artery disease was similar in both genders, as were levels of cholesterol and triglyceride. Although there were no gender differences in the prevalence of diabetes, it is notable that, in both genders and in both groups, there was a remarkably high prevalence of type 2 diabetes mellitus. Both groups had evidence of marked insulin resistance, but more so in stroke patients, underpinning this observation that may contribute to the pathogenesis of stroke in South Asian subjects.

PAI-1 levels

Previous studies of gender differences in the fibrinolytic system in white populations have found conflicting results. PAI-1 levels have been found to be similar in both genders in the healthy white population (Eliasson et al, 1993), which concurs with our finding in the control group. Studies on subjects with coronary stenosis have found higher PAI-1 in white females than in males (Haverkate et al, 1995; Ossei-Gerning et al, 1998). Similarly, higher PAI-1 levels in white diabetic females compared with white diabetic males have been reported by our group (Mansfield et al, 1996). In this study, PAI-1 antigen levels were significantly elevated in South Asian female stroke patients, and the association with gender persisted in the logistic regression model.

In white populations, PAI-1 levels correlate strongly with estimates of insulin resistance and are considered to contribute to vascular risk. This relationship was stronger in males in one study (Henry et al, 1998) and females in another study (Kario et al, 1995). In our study, a similar relationship existed in males and females in the control group and in males with stroke. However, there was no relationship in South Asian female stroke patients, and high levels of PAI-1 antigen were independent of smoking habit, WHR and HDL-cholesterol. This may suggest alternative mechanisms to account for raised PAI-1 levels in South Asian females with stroke.

Post-menopausal females and males of similar age had similar PAI-1 levels in the Framingham Offspring Study (Gebara et al, 1995). However, healthy females with higher oestrogen levels or those on hormone replacement therapy (HRT) had lower PAI-1 levels (Gebara et al, 1995). In this study group, the majority of South Asian females with stroke (94%) and without stroke (93%) were post-menopausal, and only 2% were on HRT in the stroke group. Therefore, oestrogen and HRT are unlikely to influence the results.

Although levels of PAI-1 antigen are elevated in South Asian females with stroke, the mortality from stroke is increased in South Asian males. The apparent inverse association between mortality, gender and impaired fibrinolytic activity was unexpected. Indeed, we might well have expected that male subjects would have exhibited higher PAI-1 levels, which would be in keeping with higher mortality rates from vascular disease in Asian males. However, this study was not designed to establish the relationship of haemostatic factors with vascular mortality. Furthermore, it is difficult to comment on any relationship with mortality, as the cases were recruited at least 3 months after the stroke and, in this sense, had ‘avoided’ early mortality from stroke. Had this group been included at the time of the acute stroke, then a relationship of fibrinolytic factors with male gender might well have emerged. In addition, it is plausible that there are as yet unknown genetic and environmental factors that contribute to increased mortality in males.

FVII levels

Our finding of raised FVII:C in South Asian females in both stroke patients and controls accords with studies in whites, in the general population (Balleisen et al, 1985), with angina pectoris (Benamer et al, 1996) and in females with significant coronary stenosis (Haverkate et al, 1995). Factor VII antigen was associated with gender in controls. However, in the regression model, gender was not an independent predictor of FVII antigen/FVII:C in patients, and this is similar to that described in a study of white acute ischaemic stroke patients (Heywood et al, 1997). It is plausible that FVII:C is associated with other risk factors in stroke.

Fibrinogen, VWF, FXII and t-PA

In contrast to FVII:C and PAI-1, we found no difference in fibrinogen, VWF, FXIIa and t-PA levels between South Asian males and females with ischaemic stroke or in controls. Fibrinogen has been reported to be higher in white healthy females (Ernst, 1994). However, studies on VWF in healthy white subjects (Conlan et al, 1993) and subjects with angina (Haverkate et al, 1995) have not demonstrated any gender differences. Similarly, studies on t-PA antigen have shown conflicting results in healthy subjects (Eliasson et al, 1993), some finding similar levels and others finding higher levels in males (Siegert et al, 1994). Diabetes mellitus, low HDL-cholesterol and increased triglycerides exert a stronger effect on white females than on white males (Kannel & McGee, 1979; Rifkind & Gordon, 1998). Similarly, although there were no differences in the levels of VWF, XIIa and t-PA in South Asian males and females, it is not known whether these risk factors exert their effect more strongly in one gender than in the other. To elucidate the importance of such differences, a prospective study is required.

We have selected South Asian patients 3 months after stroke to lessen the effect of the acute-phase reaction increase in haemostatic variables; however, this may have led to selection bias and may have influenced our results. The apparently high incidence of recurrent stroke is difficult to explain. Although selection bias may be a factor, the cases were recruited from both hospital and community sources to reduce possible bias resulting from a hospital-based stroke study. Furthermore, the diagnosis of stroke and recurrent stroke was a clinical one, and the diagnosis was made by an experienced clinician with the necessary language skills to elicit a history of recurrent stroke accurately. We could perhaps have improved diagnostic certainty in the recurrent stroke cases using cranial magnetic resonance imaging (MRI) scanning, but this was not available to us. Furthermore, it is well recognized that the South Asian population can be regarded as ‘vasculopaths’, and the disproportionately high rates of recurrent ischaemic stroke reflect the greater burden of atheroembolic disease.

In conclusion, South Asian females with ischaemic stroke exhibit higher PAI-1 antigen levels. Activation of the fibrinolytic system in South Asian ischaemic stroke patients may be gender related. Whether these gender differences in PAI-1 levels influence the onset and outcome of ischaemic stroke remains to be determined.

Acknowledgment

This work was supported by the Stroke Association UK.

Ancillary