High factor VIII antigen levels increase the risk of venous thrombosis but are not associated with polymorphisms in the von Willebrand factor and factor VIII gene
Rogier M. Bertina, Haemostasis and Thrombosis Research Centre, Leiden University Medical Centre, C2-R, PO Box 9600, 2300 RC Leiden, The Netherlands. E-mail: r.m.bertina@ lumc.nl
High factor VIII levels increase the risk of venous thromboembolism, but the mechanisms that cause high factor VIII levels are unclear. In 301 thrombosis patients and 301 matched healthy controls, factor VIII antigen (VIII:Ag) levels ≥ 150 IU/dl increased the thrombosis risk more than fivefold. We investigated whether high factor VIII:Ag levels result from a genetic variation in the factor VIII or von Willebrand factor (VWF) genes. Six polymorphisms in the VWF gene and two CA-repeats in the factor VIII gene were not associated with plasma VWF levels, factor VIII:Ag levels, or thrombosis risk. Our data do not support the hypothesis that a single functional sequence variation in the factor VIII or VWF gene explains the majority of high factor VIII levels and thrombotic risk.
High factor VIII coagulant activity (factor VIII:C) levels are associated with an increased risk of deep vein thrombosis (Koster et al, 1995; Kraaijenhagen et al, 2000; Kyrle et al, 2000). The relative risk associated with levels ≥ 150 IU/dl was 4·8 (95% CI 2·3–10·0) compared with levels < 100 IU/dl (Koster et al, 1995). The prevalence of such high factor VIII:C levels is 25% in thrombosis patients and 11% in healthy control subjects (Koster et al, 1995). The mechanisms that lead to high plasma factor VIII levels are unclear.
Von Willebrand factor (VWF), as carrier protein of factor VIII, is an important determinant of the factor VIII level (Wise et al, 1991). Previously, we reported familial clustering for factor VIII:C and VWF antigen (VWF:Ag) levels, even after adjustment for blood group (Kamphuisen et al, 1998). This strongly suggests the existence of additional genetic components controlling both VWF and factor VIII levels.
In the present study, we investigated whether high factor VIII levels are associated with a single common variation in the factor VIII or VWF gene, and whether this variation is associated with an increased thrombosis risk.
Patients and methods
We studied 301 patients and 301 age-matched controls from the Leiden Thrombophilia Study (LETS) (Koster et al, 1995). Factor VIII:C and VWF:Ag levels were measured using a one-stage clotting assay (Koster et al, 1995) and an enzyme-linked immunosorbent assay (ELISA) (Blann et al, 1995) respectively. Factor VIII antigen (VIII:Ag) was measured using a sandwich-type ELISA with two monoclonal antibodies directed against different epitopes on the light chain of factor VIII.
Two CA-dinucleotide repeat polymorphisms within intron 13 (Lalloz et al, 1991) and intron 22 (Lalloz et al, 1994) of the factor VIII gene were analysed. We studied six single nucleotide polymorphisms in the VWF gene: four polymorphisms in the promoter region of the VWF gene (Zhang et al, 1994) (−1793C/G, −1234C/T, −1185 A/G and −1051G/A) and two polymorphisms in the region coding for the factor VIII-binding domain [2365A/G (Thr/Ala 789; exon 18) and 2555G/A (Arg/Gln 852; exon 20)].
Linear regression was used for comparisons between groups and multiple logistic regression for calculation of the thrombosis risk (odds ratio and 95% confidence interval).
The regression coefficient between factor VIII:C and factor VIII:Ag levels was 0·66 in thrombosis patients and 0·77 in the control group. Among all subjects (patients and control group) with factor VIII:Ag levels ≥ 150 IU/dl, 74% also had VWF:Ag levels ≥ 150 IU/dl (mean VWF:Ag level 183 IU/dl, range 150–265), whereas in the remaining 26%, the mean VWF:Ag level was 125 IU/dl (range 68–149). Factor VIII:Ag levels ≥ 150 IU/dl were associated with a five times higher risk compared with factor VIII:Ag levels < 100 IU/dl (OR 5·3, 95% CI 2·7–10·1) (Table I).
Table I. Relative risk of thrombosis for categories of factor VIII:Ag levels.
|< 100||82 (27)||164 (54)||1*||1*|
|100–125||67 (22)||47 (16)||3·7 (2·2–6·1)||3·2 (1·9–5·5)|
|125–150||65 (22)||51 (17)||3·4 (2·0–5·7)||2·9 (1·6–5·2)|
|≥ 150||87 (29)||39 (13)||6·0 (3·4–10·3)||5·3 (2·7–10·1)|
Analysis of the CA-repeats of the factor VIII gene was restricted to men, 121 patients and 120 control subjects, to enable the assignment of haplotypes. Among the control group, six alleles were found for intron 13, of which allele (CA)20 was the most common (53%). Five alleles were found for intron 22, with the (CA)19 repeat being the most prevalent (65%). The genotype and haplotype distribution of the intron 13 and 22 alleles was similar for patients and control group, and was not associated with factor VIII:Ag levels.
VWF promoter polymorphisms segregated as follows: allele −1185A was always linked to allele −1051G and −1185G to −1051A. The other two promoter polymorphisms −1793C/G and −1234C/T were in strong linkage disequibilibrium with the −1185A/G and −1051G/A polymorphisms, and were therefore excluded from the analysis. The frequencies of the −1185A/−1051G allele was 0·36 in thrombosis patients and 0·35 in the control group, whereas the frequency of the −1185G/−1051A allele was 0·64 in thrombosis patients and 0·65 in the control group. The allele frequencies of the VWF polymorphisms in the factor VIII-binding region were similar in patients and the control group: 2365A/G – patients 0·64/0·36, control group 0·66/0·34; 2555G/A – patients 0·93/0·07, control group 0·92/0·08. Overall, no association was found in either patients or the control group between any of these polymorphisms and VWF:Ag levels or factor VIII:Ag levels (Table II). In 70% of the patients and 72% of the control group, haplotypes could be inferred for the promoter and factor VIII-binding region polymorphisms, but the haplotype distributions were similar among patients and the control group.
Table II. Polymorphisms in the VWF promoter region and in the region coding for the factor VIII-binding domain and VWF:Ag and factor VIII:Ag levels.
|Promoter polymorphism −1185A/G/−1051G/A|
| AA/GG||133 (38)||124 (39)||127 (43)||104 (40)|
| AG/GA||140 (35)||120 (39)||134 (40)||114 (41)|
| GG/AA||138 (43)||117 (32)||134 (55)||99 (30)|
|Factor VIII-binding domain 2365A/G, Thr/Ala789|
| AA||132 (36)||122 (38)||127 (44)||106 (40)|
| AG||138 (31)||121 (37)||132 (44)||111 (40)|
| GG||148 (41)||119 (47)||140 (40)||104 (39)|
|Factor VIII-binding domain 2555G/A, Arg/Gln852|
| GG||139 (38)||122 (39)||132 (43)||108 (39)|
| GA||123 (31)||116 (37)||121 (50)||102 (42)|
| AA||142 (2)||141 (47)||129 (24)||135 (64)|
We found a strong association between factor VIII:Ag and factor VIII:C levels in patients with thrombosis and in the control group. In a study limited to patients with thrombosis, O'Donnell et al (1997) made similar observations. Thus, in general high factor VIII:C levels are based on high factor VIII protein. Subjects with factor VIII:Ag level ≥ 150 IU/dl have a five- to sixfold increased risk of thrombosis, similar to the relative risk of factor VIII:C levels ≥ 150 IU/dl (Koster et al, 1995).
Based on our previous data on familial clustering of factor VIII and VWF factor (Kamphuisen et al, 1998), we hypothesized that increased factor VIII may result from a genetic variation in the factor VIII or VWF gene. If one single variation was the cause of the majority of high factor VIII levels (analogous to the factor V Leiden mutation causing the majority of activated protein C resistance), then this would be a frequent variant considering the high prevalence (25%) of factor VIII levels ≥ 150 IU/dl among thrombosis patients. We aimed at identifying such an unknown variant through linkage to known polymorphisms in the factor VIII and VWF gene, which would be feasible considering the expected high prevalence of that variant. Therefore, we investigated whether polymorphisms in the factor VIII and VWF genes were associated with plasma factor VIII levels or with an increased thrombosis risk.
We selected the CA repeats in intron 13 and 22 of the factor VIII gene (Lalloz et al, 1991, 1994) because these are common and highly informative. In addition, functional single nucleotide polymorphisms in the factor VIII gene are rare. These polymorphisms showed similar distributions for each genotype and haplotype in male thrombosis patients and control subjects and were not associated with the plasma levels of factor VIII:Ag. However, because CA-repeats are subject to change over time this may have reduced possible linkage with a functional variant. Recently, Mansvelt et al (1998) investigated the promoter and 3′ terminus of the factor VIII gene for variations associated with high factor VIII:C levels and found none.
High VWF levels may be the main determinant of high factor VIII levels. Therefore, we also analysed polymorphisms in the VWF gene to identify haplotypes that might be linked to a functional variation. We have selected VWF polymorphisms in the VWF gene promoter region and factor VIII-binding domain because of the possibility that these polymorphisms might also be directly responsible for high factor VIII levels by increasing the amount of carrier protein or by improving the stability of factor VIII. No relationship with VWF levels, factor VIII:Ag levels or an increased risk of venous thrombosis was found. These results are in agreement with recent data of Bowen et al (2001) who did not find gain-of-function mutations in the factor VIII-binding domain of VWF that could explain elevated factor VIII levels.
The absence of an association between VWF and factor VIII gene haplotypes and factor VIII levels may indicate that high factor VIII is not caused by one single common variant in the investigated genes (either the analysed polymorphism itself or a sequence variation linked to this polymorphism). Of course, this does not exclude the possibility of the existence of several different functional variations located within different founder haplotypes.
In conclusion, high factor VIII:Ag levels are, like factor VIII:C, associated with an increased risk of venous thrombosis. Our data do not support the hypothesis that one of the common haplotypes of the factor VIII or VWF gene contains a single functional sequence variation that explains the majority of high factor VIII levels and thrombotic risk.
This study was supported by The Netherlands Organization for Scientific Research (grant no. 950-10-629) and by The Netherlands Heart Foundation (grant no. 89.63). We thank N.H. van Tilburg, Mrs T. Visser, Mrs C.C. Spaargaren and M.A. van Werkhoven for their excellent technical assistance.