Factor VII and fibrinogen levels as risk factors for venous thrombosis: a case-control study of plasma levels and DNA polymorphisms - The Leiden Thrombophilia Study (LETS)

Thrombosis and Haemostasis © F K Schattauei Verlagsgesellschaft mbH (Stuttgart) 71 (6) 719-22 (1994)

Factor VII and Fibrinogen Levels äs Risk Factors

for Venous Thrombosis

A CaseControl Study of Plasma Levels and DNA Polymorphisms

-The Leiden Thrombophilia Study (LETS)

T, Koster

, F. R Rosendaal

, P. H Reitsma

, P. A van der Velden

, E Briet

,

J. P, Vandenbroucke

From the Departments of 'Clinical Epidemiology and the 2Haemostasis and Thrombosis Research Center, University Hospital Leiden The Netherlands

Summary

The plasma levels of coagulation factor VII and fibrmogen aie well known nsk factors for artenal thrombosis We tested the hypothesis that this assoctaüon also exists for venous thrombosis Additionally, Mspl and Haelü polymoφhlsms m the factor VII and fibrmogen genes

have recently been reported to be associated with the concentration of both protems m the plasma However, no conclusion could be drawn with respect to an increase or decrease m thrombosis nsk We under-took a population-based case control study, m which 199 patients with a first, objecüvely confirmed episode of deep vem thrombosis, aged

less than 70, and without a known mahgnant disorder were compared to 199 age- and sex-matched healthy controls, to evaluate the clmical impoitance of these reported fmdmgs

For fibrmogen we found a positive level-ielated association between the plasma fibrmogen level and thrombotic nsk Subjects with a plasma fibrmogen greater than 5 g/l had an almost 4 fold increase of thrombosis nsk The frequencies of the different Haelll genotypes were out of balance only for the thiombosis patients, with a deficit of the H1H2 genotype Possession of an H1H2 genotype was associated with a 40% reduction m thrombosis risk

For factor VII, neithei the plasma level nor the Mspl genotypes were related to deep vem thrombosis, although possession of a M2 allele was clearly associated with sigmficantly lower factor VII levels The frequencies of the Mspl-genotypes weie the same for patients and control subjects and exhibited Hardy-Weinberg equilibnum

Our results support that plasma fibrmogen, a determmant of aitenal thrombosis is also a risk factor for venous thrombosis, while factor VII plasma concentration is unrelated to deep vem thrombosis, which is supported by the data from the DNA analysis of polymorphisms Introduction

Deep vem thrombosis is a common disordei (1) The numerous causes of thrombosis mclude both hereditary and acquired conditions (2) However, despite our knowledge of these risk factors, most

epi-This study was supported by the Netherlands Heart Foundation (Grant

no 89 063)

Correspondence to Dr T Koster, Dept of Clmical Epidemiology, Buildmg l, CO-P, University Hospital, P 0 Box 9600, 2300 RC Leiden, The Netherlands - Fax Number +31 71 24 8122

sodes of deep vem thrombosis seem to occui spontaneously This suggests that theie are additional mechamsms for the development of venous thrombosis, yet to be discovered We investigated two key com-ponents of the extnnsic and common clottmg pathway, factor VII and fibrmogen Patients deficient foi factor VII or fibrmogen suffer from a tendency to bleed (3,4, 5) On the othei hand, higher mean factor VII levels were found in 25 patients with deep-vem thrombosis than m 38 control subjects, suggestmg that there might be an association (6) Elevated plasma levels of both factor VII and fibrmogen aie well estabhshed nsk factois for artenal thiombosis (7-9) We tested the hypothesis that these associations also exist for venous thrombosis

Many envuonmental factois mfluence the plasma levels of both fac-toi VII and fibrmogen (10-12) They do not explam all the vanations among individuals, and genes contnbute äs well (13) Several reports have been pubhshed which showed a clear relation between plasma levels and genetic factors exammed through restlichen fragment length polymorphism (RFLP) techniques (14-17) Since these studies were hmited to healthy individuals, they offered no Information on a lelation between a ceitam genotype or plasma levels and the nsk of thrombosis

In the present study, we have investigated the clmical importance of the association between plasma levels and RFLPs m 199 unselected consecutive patients, aged less than 70, with a first, objectively con-firmed episode of deep vem thrombosis and without a known mahgnant disease We compared the plasma concentrations of factor VII and fibrmogen and the allele frequencies of the RFLPs m patients and age-and sex matched healthy control subjects This work was performed äs pait of an ongomg population-based case-control study on hereditaiy venous thrombosis called the Leiden Thrombophilia Study (LETS)

Materials and Methods

Selectwn of Patients and Control Subjects

Patients were selected fiom the Computer flies of the Anticoagulation Clmic m Leiden In the Netherlands, Anticoagulation Clmics momtor coumarm treatment m virtually all patients with venous thrombosis and this treatment is invanably monitored by a legional Anticoagulation Clmic (18,19) Each Anti coagulation Clmic serves a well defmed geographical area, which for the Leiden Anticoagulation Clmic has about 450,000 mhabitants We mcluded the first 220 consecutive out-patients of this well defmed geographical area, youn-gei than 70 yeais, who were lefened foi anticoagulant treatment after a first, objectively diagnosed, episode of deep vem thrombosis that occurred between January l, 1988 and December 31, 1992 Patients with known mahgnant dis

Orders were excluded Information about the mclusion and exclusion cntena was obtamed from general practitioners and discharge records from the hospi tals Patients were seen only after anticoagulant treatment had been discon-tinued for at least thiee months The median Urne between the occurrence of deep vein thrombosis and venepuncture for this study was 15 (ränge 6-44) months The response of the ehgible patients was high (91 percent)

Each thiombosis patient was asked to find his own healthy control subject accordmg to the followmg cntena same sex, about the same age (± five years), no biologic relative and no history of venous thiomboembohsm no use of coumanns for at least three months not known with a mahgnant disorder and mhabitant of the same geographical area Partners of patients were also mvited to serve äs control subjects When a patient was unable to find a control subject, the first mdividual from this hst of partneis who matched for age and sex was asked to jom the study 91 (46%) control subjects were partners of other pa-tients

crease or decrease m factor VII01 fibnnogen plasma levels per umt mcrease m the factor studied, adjusted for the effect of other variables in the model and the matchmg factors age and sex

Smce plasma factoi VII and fibnnogen levels are contmuous variables we calculated matched odds ratios, äs estimates of the relative nsk, over several strata of exposure, with 95% confidence limits The odds ratio reflects the thrombosis nsk when a factor is present relative to the nsk when it is absent, ad-justed for the matchmg factors When the 95% confidence interval of the odds ratio does not mclude unity, the odds ratio is different from umty at a signifi cance level of 5% or less As possible confounders we mcluded smokmg, alco-hol mtake oral contraceptive use, menopausal Status, pregnancy and body mass mdex Adjustment for these possible confounders was accomplished by entenng all mto a (conditional) logistic regression model (Egret® Software) (21) Smce genes cannotbe mfluenced by any of these confoundeis, only crude odds ratios are provided for the Mspl and Haelll RFLPs

Collection of the Data

All subjects completed a Standard questionnaire, which contamed questions about variables which are known or suspected to mfluence factor VII or fibrmogen levels smokmg, alcohol mtake, oral contraceptive use, menopausal state pregnancy history of artenal thrombosis (11, 20) Blood pressure and body mass mdex (weight/height squared) were measured under Standard condi üons for every participant at the study center Blood was collected from the antecubital vem m Saistedt Monovette0 tubes, contammg 0 106 mM tnsodium

citrate Plasma was prepared by centnfugation for 10 mm at 2000 X g at loom temperature and stored at -70° C High moleculai weight DNA was isolated from leucocytes and stored at -4° C Factor VII was measured usmg Thrombo rel° S reagent (Behnngwerke AG, Warburg, Germany) and Factor VII defi cient plasma (Organen Tekmca, Durham, USA) The fibnnogen concentration was determmed accordmg to method of Clauss usmg Dade° thrombin reagent (Baxter, Miami, USA) Both tests were performed on an Electra 1000 (MLA Pleasantville, USA) For the factor VII and fibrmogen gene we used Mspl and Haelll restnction enzymes äs reported by Green and Thomas, respectively (15-16) The alleles with the restnction site were designated Ml (VII) and Hl (fibrmogen) and the non cleaveable alleles were designated M2 (VII) and H2 (fibrmogen)

Analysis and Statistics

Twenty-one patients were on long term coumarm treatment and were ex-cluded from this analysis, which therefore mex-cluded 199 matched patient-con-trol pairs A chi-square test was used to compare the observed numbers of each genotype with those expected for a population m Hardy-Wemberg equilibnum

We analyzed the effect of the polymorphic genotype on plasma factor VII and fibrmogen levels for patients and control subjects by multivanate linear regression techmques Factor VII and fibrmogen levels were entered äs conti-nuous variables The Mspl and Haelll genotype and the Status of an mdividual (patient or control subject) were entered äs discrete variables (0 for 1-1, l for 1-2 and 2 for 2-2 genotype, and 0 for control subjects l for patients, respec-tively) The regression coefficient obtamed by this method indicates the

m-Results

The male/female ratio among patients and control subjects was l/l 5 and mean age was 44 years (ränge 17-70 patients, 17-71 controls) The mean body mass mdex was slightly higher m patients than in the con-trol subjects (26 3 versus 25 3 kg/m2) Table l shows mean factor VII

and fibrmogen levels for other charactenstics of the patients and control subjects Oral contraceptive use and post-menopausal states were more frequent among the control subjects and tended to be related to higher levels of factor VII and fibnnogen compared to non-users and pre menopausal states, respectively

Table 2 gives the results on factor VII and fibnnogen plasma levels and thrombosis nsk The odds ratios for the plasma factor VII strata dtd not differ from umty, but the nsk of venous thrombosis was related to the fibnnogen level Subjects with a fibnnogen level between 4 and 5 g/l had a twofold higher nsk than those m the refeience category (<3 0 g/l) Beyond 5 g/l of fibrmogen an even stionger association was present, but äs a consequence of a low number of subjects the confi-dence interval remamed wide The association persisted when control-1mg for the possible confounders The uncategonzed odds ratios, which apply over the whole ränge, were l 02 (95% confidence interval 0 99-1 03) for (each percent change in) factor VII, and l 4 (95% con-fidence interval l 02-1 95) for (each gram/l change in) fibrmogen

Whether or not a certain genotype is associated with thrombosis, the allele distnbution of control subjects should be m Hardy Weinberg equi-hbnum, smce they are selected from the general population The pa-tients, however, were mcluded because of their episode of deep vem thrombosis If a relation exists between deep vem thrombosis and a cer-tain genotype, the allele frequencies will be out of balance, with over-representation of the nsk enhancmg allele In Table 3 the frequencies of the different alleles are presented As expected, the control group was m equilibnum for the Mspl and Haelll polymorphisms In the

throm-Table l Mean factor VII and fibrmogen levels for several charactenstics of the patients and control subjectsf

Smokers

Females usmg oral contraceptives yes no Females m menopause yes no Factor VII (%) patients 111(37%) 114(15%) 110(85%) 116(18%) 108 (82 %) controls 109(36%) 121 (24 %) 108(76%) 120 (24 %) 107 (76 %) Fibrmogen (g/l) patients controls 34 35 3 3 37 3 3 3 3 32 32 34 3 2

* At the time of the venepuncture while visitmg the study center

Table 2 Thrombosis nsk for factor VII and fibrinogen plasma levels Table 4 Thrombosis nsk for the differenl Mspl and Haelll genotypes Risk factoi Factor VII (%) Fibrinogen (g/l) <85 85-109 >110 < 3 0 30-39 40^19 >50 Gases 27 78 94 64 105 24 6 Contiols 23 79 97 78 105 14 2 ORc m d c(95%CI)' 1 02 3 08(04-16) 0 8 (0 4-1 5) 1 09 3 4 1 3 (0 8-2 0) 2 1 (1 (M 5) 37(07-19) OR denotes age and sex-matched odds raüo 2Reference category 3Adjust

ment for current pill use (yes/no), body mass mdex, in menopause (yes/no), alcohol use (non/low/high) and smoking (yes/no) did not affect these results None of the female participants were pregnant at the time of the venepuncture "Test for trend p <0 05

Table 3 Frequencies of the Mspl and Haelll genotypes m patients and

controls Genotype m RFLP Patients 1-1 1-2 2-2 Controls 1-1 1-2 2-2 Mspl polytnorphism observed expected1 164 1646 34 328 1 16 X2 = 027 (ns)2 162 161 34 36 3 2 A2 = 0 59 (m)2 Haelll polymorphism observed expected1 144 1377 43 557 12 56 X' = 105(p<001) 127 1254 62 651 10 85 X2 = 0 43 (ns)2

'For a population m Hardy Weinberg equihbnum with /; A alleles and q a alleles the expected genotype distibutation was assessed usmg the equation

p2 AA + 2pq Aa + q2 aa = l

2ns denotes not sigmficant at the 5 % level

FVII activity (%) fibrinogen (g/j)

130 120- 110- 100- 90-80 - 3 8 - 3 6 - 3 4 - 3 2 2 6 1-1 1-2 2-2 Polymorphic genotype

Fig l Mean factor VII activity and fibrinogen plasma level by polymorphic

genotype and mdividual Status (patient or control) in 199 thrombosis patients and 199 age and sex matched healthy controls ·—· denotes factor VII levels m thrombosis patients ·—· denotes factor VII levels in healthy controls A—A denotes fibrinogen levels m thrombosis patients A—A denotes fibrinogen levels m healthy controls The ränge of each Y-axis reflects approximately the mean ± l Standard deviation for the group äs a whole (199 patients and 199 controls)

Risk factor

Factoi VII Mspl RFLP

Fibrinogen Haelll RFLP

OddsRatiomalched 95%CI

1-1 1-2 2-2 1-1 1-2 2-2 101 10 03 101 06 1 1 06-17 0 03-3 2 04-10 05-25 'Referencecategory

bosis patients, the observed frequencies of factor VIIMl andM2 alleles were m equihbnum In addition, the frequency of the Mspl genotypes was the same for patients and control subjects, supporting that they ongmated from the same source population The observed frequencies of fibrinogen Hl and H2 alleles were sigmficantly different from the expected ones, with less counts of the H1H2 genotype in the patients

Figure l shows the mean levels of factor VII activity and fibrinogen for each genotype for patients and contiol subjects For factor VII a clear association was present between the Mspl genotype and the plas-ma levels of factor VII, the M2 allele being associated with lower levels (legression coefficient -13 9, 95% CI -19 3 to -8 6) We found no obvious relation between genotype and plasma fibrinogen level (re-gression coefficient 0 04,95% CI -0 07 to 0 15), without any subgroup effect for smokeis (data not shown) Even when only Haelll 1-2 and 2-2 genotypes were considered, no sigmficant association could be observed

Table 4 gives the thrombosis nsk for the several polymorphic geno-types Foi the factoi VII Mspl genogeno-types we did not find any relation with deep vem thiombosis, although the Mspl genotypes correlated well with the plasma levels For the fibrinogen Haelll genotypes we found that subjects with one H2 allele, had a sigmficant 40% reduction m the nsk of venous thrombosis compared to HIHI mdividuals

Discussion

Among 199 patients, younger than 70 years of age, with a first ob-jectively confirmed episode of venous thrombosis and without a known mahgnant disorder, and 199 age- and sex-matched healthy control sub-jects, we found an association between the Mspl genotype and factor VII levels, but neither Mspl genotypes nor factor VII levels were relat-ed to an mcrease or decrease in thrombosis nsk For fibrinogen it seems the öfter way round, we found a clear association between nsk of thrombosis and fibrinogen levels, which nsk mcreased with higher levels of fibrinogen Subjects with H1H2 genotype seemed to have a 40% reduction m the nsk of thrombosis, but no association was present between plasma fibrinogen and polymorphic genotype

The fmding of a positive trend between fibrinogen levels and the enhanced nsk of developmg venous thrombosis seems analogous to the role of fibrinogen at the artenal site An explanation might be that elevation of the fibrinogen level increases blood viscosity, which may further enhance the nsk of thrombus formation (22) Furthermore, m-creasing levels of fibrinogen withm the physiological ränge mfluences platelet aggregation m vitro (23) In our study design we cannot discn-mmate whether a high fibrinogen level is a nsk factor per se or a post-thrombotic phenomenon or the result of another, true nsk factor The venepuncture for this study was performed at least six months (on aver-age 18 months) after the acute thrombotic event Therefore, a post-hoc phenomenon seems less likely an explanation

In sequence to an mcreasmg number of reports about certam restnc-tion length polymorphisms and plasma levels of factor VII and fibrmo-gen, we were able to link our RFLP-data to the nsk of venous throm-botic disease We found that VII levels or Mspl genotype are not related to venous thrombosis, but we could confirm the results of Green et al (15) that possession of a M2 allele is associated with lower factor VII levels In contrast to the fmdmgs of others (16,24), we found no relation between fibnnogen levels and the Haelll genotypes, though subjects who are homozygous for the H2 allele had higher levels Possibly because of a low number of H2H2-subjects this could not reach statistical sigmficance The unclear association between Healll genotype and fibnnogen levels (mteraction with srnokmg) m yet another study m myocardial infarction survivors (25), makes one wonder about possible differences between populations We did not find any subgroup effect for smokers

We cannot easily explam why Haelll 1-2 mdividuals seemed to have a 40% reduction m the nsk of venous thrombosis, especially when it is assumed that H2-mdividuals have higher fibnnogen levels Inter-estingly, a similar effect, albeit not significant, was recently reported for myocaidial mfarcüon, i e a 10-20% reduction of risk for those who carned an H2 allele (24) Therefore, we do not thmk this was a chance occurrence

Our control group was m perfect Hardy-Weinberg equilibnum for the Mspl and Haelll alleles This was to be expected since the control subjects were chosen from the general population The patient group showed a disequilibnum for the Haelll alleles Smce we selected these patients from the same geographical area and solely on the basis of their deep vem thrombosis, it is plausible that we selected on a certam geno-type which is associated with venous thrombosis This finding is of m-terest since it also pomts to an effect of fibnnogen on the risk of venous thrombosis which cannot be explamed by the plasma level

In addition, another mterestmg pomt raised by the Hardy-Weinberg equilibnum seems the possibihty to check the genetic comparabihty of patients and control subjects by usmg a parameter which is irrelevant for the disease of mterest For mstance, since factor VII plasma levels and polymorphic Mspl genotypes are unrelated to deep vem thrombo-sis, one will expect both patients and control subjects to exhibit equili-bnum, and also that the frequencies of each genotype are equal for pa-tients and control subjects Disequihbnum among cases suggests an as-sociation with disease, whereas equilibnum in cases and control sub-jects, but different allele frequencies suggests mcomparability of cases and control subjects We found stnkmgly similar frequencies for each genotype (Table 3) This suggests that patients and control subjects ongmated from the same source population, and further supports the absence of an association between factor VII levels and venous throm-bosis

Our results show that plasma fibrmogen, a determmant of artenal thrombosis is also a nsk factor for venous thrombosis, while the factor VII plasma concentration is unrelated to deep vem thrombosis, which is supported by the data from the DNA analysis of polymorphisms

Acknowledgements

We thank Dr F J M van der Meer (Anticoagulation Chnic Leiden) for bis kmd cooperation, Mrs T Visser for laboratory assistance, Mrs A van Beek for secretanal and administrative support

References

l Koster T, Rosendaal FR, van der Meer FJM, Briet E, Vandenbroucke JP More objective diagnosis of venous thromboembohsm' Neth J Med 1991,38 246-8

2 Hirsh J, Hüll R, Raskob GE Epidemiology and pathogenesis of venous throm bosis J Am Coll Cardiol 1986, 8 104B-13B

3 Marder VJ, Shulman NR Chmcal aspects of congenital factor VII deliciency Am Jof Med 1964, 37 182-94

4 Manam G, Marzuccom MG Factor VII congenital deficiency Haemostasis 1983,13 169-77

5 Beck EA Congenital abnormalities of fibnnogen Clmics m Haematology 1979,8 169-81

6 Boyer C Wolf M, Rothschild C, Migaud M, Amiral J, Mannucci PM, Meyer D, Larrieu MJ An enzyme immunoassay (ELISA) for the quantitation of human factor VII Thromb Haemost 1986, 56 250-5

7 Wilhelmsen L, Svardsudd K, Korsan-Bengtsen K, Larsson B, Tibblm G Fibnnogen äs a nsk factor for stroke and myocardial infarction N Engl J Med 1984, 311 501-5

8 Meade TW, Mellows S, Brozovic M Miller GJ, Chakrabarti RR, North WRS, Haines AP, Stnling Y, Imeson JD, Thompson SO Haemostatic function and ischaemic heart disease, pnncipal results of the Northwick Part Heart Study Lancetl986,n 533-7

9 Kännel WB, Wolf PA, Castelli WP, D'Agostmo RB Fibrmogen and nsk of cardiovascular disease, the Frammgham Study J ΑΜΑ 1987, 258 1183-6

10 Meade TW, North WRS Population based distnbutions of haemostatic variables Br Med Bull 1977, 33 283-8

11 Balleisen L, Bailey J, Eppmg PH, Schulte H, van de Loo J Epidemiological stu-dy on factor VII, factor VIII and fibnnogen m an mdustrial popula tion I Baselme data on the relation to age, gender, body-weight, smokmg, al cohol, pill usmg and menopause Thromb Haemost 1985, 54 475-9 12 Thompson S, Maitm J Meade T Sources of variability m coagulation

factor assays Thromb Haemost 1987, 58 1073-7

13 Berg K Genetics of atherosclerosis In Atherosclerosis Biology and climcal science Olsson AG, ed Edinburgh Churchill-Livmgstone 1987, pp 323-37

14 Humphnes SE, Cook M, Dubowitz M, Stirlmg Y, Meade TW Role of genetic Variation at the fibrmogen locus m determmation of plasma fibrmogen concentrations Lancet 1987, i 1452-5

15 Green F, Kelleher C, Wilkes H, Temple A, Meade T, Humphnes S A common genetic polymorphism associated with lower coagulation factor VII levels m healthy mdividuals Arterioscler Thromb 1991, 11 540-6

16 Thomas AE, Green FR, Kelleher CE, Wilkes HC, Brennan PJ, Meade TW, Humphnes SE Variation in the piomotor region of the β fibrmogen gene is äs

sociated with plasma fibrmogen levels m smokers and non smokers Thromb Haemost 1991, 65 (5) 487-90

17 Humphnes SE, Lane A, Dawson S, Green FR The study of gene-environment interactions that influences thrombosis and fibrmolysis Genetic Variation at the loci for factor VII and plasmmogen activator mhibitor-1 Arch Pathol Lab Med 1992,116 1322-9

18 Meer van der FJM, Rosendaal FR, Vandenbroucke JP, Bnet E Bleedmg com-plications m oral anticoagulant therapy An analysis of risk factors Arch Intern Med 1993,153 1557-62

19 Loehger EA, van Dyk-Wierda CA, van den Besselaar AMHP, Broekmans AW, Roos J Anticoagulant control and the risk of bleedmg, orgamzational mfra-structure In Anticoagulants and myocardial infarction a reappraisal Meade TW, ed Chichester, UK John Wiley and sons 1984, pp 157-62

20 Meade TW, Chakrabarty, Haines AP, North WRS, Stirlmg Y Charactenstics affectmg fibrmolytic activity and plasma fibrmogen concentrations Br Med J 1979,1 153-6

21 Schlesselman JJ Cases control studies design, conduct, analysis New York, Oxford Umversity Press 1982

22 Löwe GDO, Drummond MM, Lonmer AR, et al Relation between extent of coronary heart disease and blood viscosity Br Med J 1980, i 673^1 23 Meade TW, Vickers MV, Thompson SG, Seghatchian MJ The effect of phy

siological levels of fibrmogen on platelet aggregation Thromb Res 1985, 38 527-34

24 Scarabm PY, Bara L, Ricard S, Poiner 0, Cambou JP, Arveilei D, Luc G, Evans AE, Samama MM, Cambien F Genetic Variation at the ß-fibnnogen locus in re lation to plasma fibnnogen concentrations and nsk of myocardial infarction The ECTIM Study Arterioscler Thromb 1993, 13 886-891

25 Green F, Hamsten A, Blomback M, Humphnes S The role of β fibrmogen ge

notype m determmmg plasma levels in young suivivors of myocardial mfartion and healthy controls from Sweden Thromb Haemost 1993, 70 915-20