Elevated factor VIII levels and risk of venous thrombosis
Correspondence: Professor James O'Donnell, Haemostasis Research Group, Institute of Molecular Medicine, St James's Hospital, Dublin 8, Ireland. E-mail email@example.com
Modern thrombophilia testing fails to identify any underlying prothrombotic tendency in a significant number of patients presenting with objectively confirmed venous thromboemboembolism (VTE). This observation has led to a search for other novel inherited or acquired human thrombophilias. Although a number of putative mechanisms have been described, the evidence behind many of these candidates remains weak. In contrast, an increasing body of work supports the hypothesis that increased plasma factor VIII (FVIII) levels may be important in this context. An association between elevated plasma FVIII levels and VTE was first described in the Leiden Thrombophilia Study (LETS). Subsequently, these conclusions have been supported by an increasing number of independent case–control studies. Cumulatively, these studies have clearly demonstrated that high FVIII levels constitute a prevalent, dose-dependent risk factor for VTE. Furthermore, more recent studies have shown that the risk of recurrent venous thrombosis is also significantly increased in patients with high FVIII levels. In this review, we present the evidence supporting the hypothesis that elevated FVIII levels constitute a clinically important thrombophilia. In addition, we examine the biological mechanisms that may underlie persistently elevated FVIII levels, and the pathways through which high FVIII may serve to increase thrombotic risk.
Despite significant advances in the field, current state-of-the-art thrombophilia testing still fails to identify an underlying inherited or acquired prothrombotic tendency in up to 50% of patients presenting with an objectively confirmed venous thromboembolism (VTE) (Dahlbäck, 2008). This observation has prompted the search for other novel pathological prothrombotic mechanisms. Bearing in mind the complexity of the coagulation and fibrinolytic systems, it is clear that many other candidate proteins could be important in this context. Moreover, it may seem self-evident that elevation of specific procoagulant factors could increase risk of thrombosis. However, although this relationship appears to be true for some coagulation factors (e.g. factor XI and prothrombin) (Meijers et al, 2000; Poort et al, 1996), it does not apply for all factors (e.g. factor V) (Kamphuisen et al, 2000a).
Factor VIII (FVIII) is a plasma sialoglycoprotein that plays an essential role in normal haemostasis by acting as a critical cofactor for the serine protease, activated factor IX (FIXa). For many years, it has been recognized that FVIII deficiency in patients with haemophilia A results in a significant bleeding diathesis. Over more recent years, increasing evidence suggests that the converse is also true, in that high plasma levels of FVIII may constitute a clinically important risk factor for thrombosis. In this review, we examine the evidence supporting the hypothesis that elevated plasma FVIII levels constitute a common and dose-dependent risk factor for venous thrombosis.
Elevated FVIII levels and venous thrombosis
The Leiden Thrombophilia Study (LETS) was the first to report an association between high plasma levels of FVIII and VTE (Koster et al, 1995). This case–control study enrolled 301 patients with first objectively confirmed VTE, together with a similar number of healthy controls matched for age and sex. Although the median time interval between acute thrombosis and FVIII coagulant activity (FVIII:C) measurement was 18 months (minimum 6 months) elevated FVIII:C levels (>150 iu/dl) were present in 25% of the patient cohort, compared to 11% of controls. Univariate analysis demonstrated that FVIII:C, von Willebranf factor antigen (VWF:Ag) and non-O blood group were all associated with increased risk for VTE. However, on subsequent multivariate analysis, only FVIII:C levels remained a significant independent risk factor. These data suggest that plasma VWF:Ag and ABO blood group are only risk factors for VTE by virtue of their influence on plasma FVIII:C levels. When adjusted for ABO blood group, an odds ratio of 4·8 [95% confidence interval (CI) 2·3–10·0] was determined for individuals with FVIII:C levels >150 iu/dl compared to those with FVIII:C levels <100 iu/dl.
This high prevalence of elevated FVIII:C levels in patients with deep vein thrombosis (DVT) or pulmonary embolism (PE) has subsequently been confirmed in a number of other cohort and case–control studies (O'Donnell et al, 1997; Bloemenkamp et al, 1999; Kamphuisen et al, 1999; De Mitrio et al, 1999; Kraaijenhagen et al, 2000; Patel et al, 2003; Oger et al, 2003; Ota et al, 2011; Ryland et al, 2012). In addition, the increased risk of VTE associated with high plasma FVIII levels has consistently been shown to be a dose-dependent phenomenon (Koster et al, 1995; Kamphuisen et al, 1999; Kraaijenhagen et al, 2000; Patel et al, 2003; Ota et al, 2011).
High FVIII:C levels and risk of recurrent VTE
In addition to representing a prevalent risk factor for first VTE, elevated plasma FVIII:C levels are also associated with significant increased risk for developing recurrent thrombotic episodes (Kraaijenhagen et al, 2000; Kyrle et al, 2000; Cristina et al, 2004). In a retrospective study, Kraaijenhagen et al (2000) studied 60 patients with recurrent VTE, 65 patients with a single episode of thrombosis, and 60 age- and sex-matched controls. In all patients with VTE, the time elapsed between acute thrombosis and blood sampling was at least 6 months. Nevertheless, FVIII:C levels > 175 iu/dl were observed in 10% (95% CI: 4–21) of controls, 19% (95% CI: 10–30) of patients with single VTE, and 33% (95% CI: 22–47) of the recurrent VTE cohort. The authors calculated that for each 10 iu/dl increment in plasma FVIII:C level, the risk for a single and recurrent episode of VTE increased by 10% and 24% respectively. Furthermore, for patients with FVIII:C levels above 200 iu/dl, the odds ratio for a recurrent thrombotic episode was markedly elevated at 45 (95% CI: 6–370).
The hypothesis that high plasma FVIII:C levels influence risk of recurrent VTE was further addressed in the larger prospective Austrian Study on Recurrent VTE (Kyrle et al, 2000). This study included 360 patients with a first episode of objectively confirmed VTE. Following discontinuation of oral anticoagulation, patients were followed for an average period of 30 months. In total, symptomatic recurrent VTE was observed in 38 (10·6%) patients. Plasma FVIII:C levels were significantly increased in the cohort of patients who developed recurrent VTE (182 ± 66 vs. 157 ± 54 iu/dl; P = 0·009). Moreover, in keeping with the findings of Kraaijenhagen et al (2000), the risk of recurrent VTE associated with elevated FVIII:C was shown to be dose-dependent. However the relationship between increased FVIII levels and risk of recurrent thrombosis was non-linear. Consequently, patients with plasma FVIII:C levels above the 90th percentile (>234 iu/dl), were at particular high risk for developing recurrent VTE. Amongst this high risk group, the overall relative risk of recurrence was 6·7 (95% CI: 3·0–14·8). Consequently, for this cohort of patients, the likelihood of developing recurrent VTE within 2 years of discontinuing anticoagulation therapy was estimated at 37% (compared to only 5% for patients with normal FVIII:C levels; P < 0·001).
More recently, another prospective cohort study also observed that the risk of recurrent venous thrombosis was significantly increased in patients with elevated FVIII:C levels (Cristina et al, 2004). A total of 564 patients were enrolled after a first episode of VTE, and were followed for a median observational period of 19·7 months. 53 patients developed objectively confirmed recurrent venous thrombosis during this period. Overall, the rate of recurrence was higher in patients with a first idiopathic VTE compared to those whose first thrombotic event was secondary to a known risk factor (12·6% vs. 5·5%; P = 0·007). In keeping with the findings of Kyrle et al (2000), high plasma FVIII:C levels represented an independent risk factor for recurrence in patients with idiopathic VTE. For patients with FVIII:C levels above the 90th percentile (294 iu/dl), univariate analysis demonstrated a fourfold increased relative risk of recurrence. In contrast, in patients whose first VTE occurred secondary to a recognized transient risk factor, elevated FVIII:C levels were not associated with increased risk for recurrent thrombosis.
On the basis of these three independent studies, it is clear that, unlike many of the other established thrombophilias, elevated FVIII levels appear to identify a subgroup of patients at increased risk for developing recurrent thrombosis. This important finding has clear translational relevance.
High FVIII levels and VTE – age and ethnicity
It is well recognized that plasma FVIII:C levels increase progressively with age (Conlan et al, 1993). In this context, it is interesting that elevated FVIII levels have been associated with increased risk of VTE across a range of age groups. Although thrombosis is relatively rare in childhood, elevated FVIII:C levels (>90th percentile) were observed in 19·5% of children with venous thrombosis compared to only 4·4% of healthy controls (Kreuz et al, 2006). Moreover, high FVIII levels in this study were associated with a fivefold increased risk for childhood VTE. Persistently elevated FVIII levels in children presenting with VTE have also been identified as a predictor of poor outcomes (Goldenberg et al, 2004). In particular, high FVIII:C levels were associated with increased risk for both recurrent thrombosis and the post-thrombotic syndrome. At the other end of the age spectrum, elevated plasma FVIII levels have also been reported to constitute a dose-dependent VTE risk factor in the elderly. In patients over 70 years, Oger et al (2003) reported that patients with FVIII:C levels above 225 iu/dl had a 2·4-fold (95% CI: 1·1–5·3) increased risk of VTE compared to age-matched individuals with FVIII levels below 130 iu/dl.
Previous population studies have demonstrated that plasma FVIII:C levels are influenced by ethnicity, with significantly higher levels in African Americans compared to Caucasians (Conlan et al, 1993; Miller et al, 2001). The biological mechanisms underlying this observation have not been clearly defined, but probably relate in part to the higher prevalence of blood group O in the Caucasian population. In the Camberwell case–control study, Patel et al (2003) recruited 100 patients with objectively confirmed DVT from south London, together with 100 contols matched for age, sex and ethnicity. All subjects enrolled were of either African (29%) or Caribbean origin (71%). In the patient group, 34% had plasma FVIII:C levels above the 90th percentile (228 iu/dl for ethnic-specific control group). Furthermore, FVIII:C levels >228 iu/dl were associated with an 11-fold increased risk for VTE compared to individuals with FVIII levels below 150 iu/dl (95% CI 4·3–29·4). Finally, notwithstanding the fact that baseline FVIII:C levels were increased compared to normal Caucasians, plasma FVIII:C levels were again shown to constitute a dose-dependent risk factor for VTE. Given the rarity of both the F5 R506Q (Factor V Leiden) and the F2 G20210A polymorphisms in non-Caucasian individuals, it is perhaps unsurprising that high FVIII levels also emerged as the single most prevalent risk factor for VTE observed in this study. Similarly, another recent study demonstrated that mean FVIII:C levels were significantly higher in Japanese patients with VTE compared to healthy controls (154 vs. 114 iu/dl) (Ota et al, 2011). Moreover, 33% of the patient group had plasma FVIII:C levels above the 75th percentile. Cumulatively therefore, it is clear that elevated plasma FVIII:C levels represent a prevalent and independent risk factor for venous thrombosis (Table 1). Moreover, this association appears dose-dependent, and is consistently observed across patients of different ages and ethnic groups.
Table 1. Case–control studies of high FVIII levels in patients with objectively confirmed venous thromboembolism
|Koster et al (1995)||Case–control (LETS)||301||46 years (17–70 years)||301 (Age- and sex-matched)||Single DVT||18 months (Range 6–48)||25%||11%||6·2||Yes|
|Bloemenkamp et al (1999)||Case–control (LETS)||155||15–49 years||169 (Sex-matched)||Single DVT||≥3 months after anticoagulation||36%||17%||4·0||Yes|
|Kamphuisen et al (1999)||Case–control (LETS)||474||47 years (16–70 years)||474 (Age- and sex-matched)||Single DVT||18 months (Range 6–48)||24%||10%||6·7||Yes|
|De Mitrio et al (1999)||Case–control||30||45 years (28–63 years)||32 (Age-, sex- and FVL-matched)||DVT/PE||≥3 months||47%||6%||2·5||NR|
|Kraaijenhagen et al (2000)||Case–control||65||55 years (39 – 71 years)||60 (Age- and sex-matched)||Single DVT or PE||≥6 months||19% (>175 iu/dl)||10% (>175 iu/dl)||5·4a||Yes|
|Patel et al (2003)||Case–control||100||44 years (21–80 years)||100 (Age-, sex- and ethnic-matched)||DVT||≥6 weeks after anticoagulation||81%||38%||7·0||Yes|
|Oger et al (2003)||Case–control||161||66 years||239||DVT/PE||At initial presentation||NR||NR||2·4||Yes|
|Ota et al (2011)||Case–control||68||56 years||40||DVT/PE||≥3 months||34%||2%||32·7||Yes|
|Ryland et al (2012)||Case–control||91||42 years (21–79 years)||52||DVT/PE||≥4 weeks after anticoagulation||40%||NR||NR||NR|
Elevated FVIII levels and the acute phase response
It is well established that plasma FVIII:C and VWF:Ag levels increase significantly during the acute phase response. In most of the studies that have examined the prevalence of high FVIII levels in patients with venous thrombosis, plasma FVIII:C levels were measured following the acute thrombotic episode (Table 1). Consequently, it is not clear whether the increase in FVIII represents a congenital thrombophilia, an acquired prothrombotic tendency, or is rather a marker of ongoing post-thrombotic acute phase response. This critical question cannot be addressed definitively without a large prospective trial. Nevertheless, accumulating evidence suggests that elevated FVIII:C levels cannot be explained simply as a post-VTE reactive phenomenon.
First, sample collection for FVIII:C analysis in most studies has typically been delayed until at least 3–6 months following the acute thrombotic event (Table 1). Nevertheless, and notwithstanding the variations in different study designs, elevated FVIII:C levels appear to be relatively consistently observed in more than 20% of patients after first VTE. Furthermore, longitudinal follow-up studies have clearly demonstrated that the increased plasma FVIII:C levels observed in these patients with VTE tend to remain persistently elevated for many years (O'Donnell et al, 2000; Kraaijenhagen et al, 2000).
Second, several groups have sought to quantify the extent and duration of the post-VTE acute phase response by measuring other known plasma indices [fibrinogen, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR)] in parallel with FVIII:C (O'Donnell et al, 1997; Kamphuisen et al, 1999; O'Donnell et al, 2000; Kraaijenhagen et al, 2000). Using these indices, overt inflammation was observed in <10% of patients with elevated FVIII:C levels (Kamphuisen et al, 1999; O'Donnell et al, 2000). Moreover, plasma FVIII:C levels failed to demonstrate any correlation with ESR, CRP or fibrinogen.
The nature of the acute phase reaction in patients after VTE has been further addressed in a recent prospective cohort study. In contrast with previous studies, Tichelaar et al (2012) investigated the respective time courses of FVIII:C, CRP and fibrinogen levels in patients immediately following acute venous thrombosis. Unsurprisingly, plasma FVIII:C, CRP and fibrinogen levels were increased in the majority of patients (88%, 76% and 64% respectively) at their initial presentation. However, although FVIII:C levels remained persistently elevated in 72% of subjects for at least 6 months, plasma CRP and fibrinogen levels had both corrected to within the normal range by 3 months in the majority of patients. These findings clearly suggest that although the post-thrombotic acute phase response contributes to the elevation in FVIII:C observed immediately after the thrombotic episode, other more important unidentified factors are responsible for maintaining high FVIII levels in the longer term. Importantly, Kamphuisen et al (1999) have further demonstrated that even after adjustment for the potentially confounding effects of CRP, high FVIII:C levels still increased the risk of VTE sixfold. In summary therefore, it seems likely that in the overwhelming majority of patients with elevated plasma FVIII levels, the post-thrombotic acute phase response is of minor importance.
Normal variation in plasma FVIII-VWF levels
In the normal population, plasma FVIII:C levels demonstrate significant inter-individual variation. Previous studies have shown that this wide variation is due to a number of important genetic and environmental determinants. In the Genetic Analysis of Idiopathic Thrombophilia (GAIT) study (Souto et al, 2000), the relative importance of these different determinants in regulating normal plasma FVIII:C were assessed in 397 individuals from 21 extended Spanish kindred. In this cohort, Souto et al (2000) estimated the overall heritability of FVIII:C variation at 40%. In keeping with these findings, another large twin study calculated that 57% of the total variation in plasma FVIII, together with 66% of variation in plasma VWF:Ag levels, was genetically determined (Orstavik et al, 1985).
In normal plasma, FVIII binds with high affinity (Kd = 0·5 nmol/l) to its carrier protein VWF (Vlot et al, 1996). Consequently under physiological conditions, approximately 94% of FVIII molecules are bound to VWF, whilst the remaining 6% circulates in free-form (Schambeck et al, 2004). This interaction with VWF serves to stabilize heterodimeric FVIII, and also protects FVIII from premature proteolytic degradation. In the absence of VWF, the half-life of free FVIII is markedly reduced (Lenting et al, 2010). Therefore, it is not surprising that plasma VWF plays a critical role in regulating plasma FVIII levels. Interestingly, Schambeck et al (2004) have demonstrated that in patients with a history of VTE associated with elevated FVIII:C levels, the proportion of free FVIII in the plasma is reduced. Consequently the FVIII:VWF ratio is significantly increased in this cohort of patients compared to normal controls. The biochemical and clinical significance of this novel observation remain to be determined.
Plasma levels of the FVIII-VWF complex are markedly influenced by ABO blood group (Jenkins & O'Donnell, 2006). Normal individuals with blood group O have plasma VWF:Ag and FVIII:C levels that are 25–30% lower compared to non-O individuals. Amongst the non-O groups, Gill et al (1987) demonstrated that AB individuals had the highest plasma VWF:Ag levels, followed by group B and then group A. This blood group effect is modulated by ABO(H) blood group carbohydrate structures, which are expressed on the glycans of both FVIII and VWF (O'Donnell et al, 2002). Recent studies suggest that these glycans may influence the in vivo clearance of VWF, and thereby determine plasma levels of the VWF-FVIII complex (Gallinaro et al, 2008). In addition, VWF glycans also play a critical role in modulating susceptibility to ADAMTS13 proteolysis (McKinnon et al, 2008; McGrath et al, 2010), Other factors associated with increased plasma levels of the FVIII-VWF complex include increasing age, sex (women higher than men), exercise, stress, pregnancy, surgery and any other causes of an acute phase response (e.g. chronic inflammation or malignancy) (Terraube et al, 2010).
Familial clustering of elevated FVIII levels
The relative contributions from genetic and inherited factors in the aetiology of persistently elevated plasma FVIII:C levels remain unknown. Nevertheless, data from several independent studies suggest that inherited factors may be important in at least some patients (Kamphuisen et al, 1998, 2000b; Kraaijenhagen et al, 2000; Schambeck et al, 2001; Bank et al, 2005; Kreuz et al, 2006). In particular, familial clustering of high FVIII:C levels has been reported. For example, in a retrospective study of 177 patients with high FVIII:C levels, Bank et al (2005) observed that 40% of their first-degree relatives (n = 584) also had elevated plasma FVIII:C levels above the 75th percentile of the normal population. Moreover, these first-degree relatives with high plasma FVIII:C levels also demonstrated increased risk for both VTE and arterial thromboembolic disease (myocardial infarction and peripheral arterial thrombosis). In relatives with elevated FVIII:C, the absolute annual incidence for VTE was 0·34% (95% CI: 0·22–0·49), compared to only 0·13% (95% CI: 0·07–0·22) in relatives with normal FVIIII levels. Furthermore, the risk of VTE was highest in those first-degree relatives with highest plasma FVIII:C levels.
FVIII and VWF gene analysis in patients with high FVIII:C
The evidence that high FVIII levels may be in part genetically determined has prompted investigation of the F8 gene locus on Xq24. However a study of 62 patients with VTE and elevated FVIII:C levels failed to identify any candidate polymorphisms in the promoter region (−542 to +165) of the F8 gene, nor in the 3′ polyadenylation signal region (Mansvelt et al, 1998). As part of LETS, Kamphuisen et al (2001a) also analysed two informative CA-dinucleotide repeat polymorphisms within introns 13 and 22 of the F8 gene. Neither of these polymorphisms was associated with plasma FVIII:Ag levels, nor risk of thrombosis. Nevertheless, more recent studies have suggested that specific FVIII haplotypes can influence plasma levels of FVIII and risk of VTE (Nossent et al, 2006a,b). Therefore, although no specific F8 gene polymorphisms have been associated with elevated plasma FVIII levels, this does not preclude the possibility that as yet unidentified regulatory F8 gene sequences may have important roles to play.
Although the physiological relevance remains unclear, activated FVIII (FVIIIa) can undergo proteolytic degradation by activated protein C at amino acid positions Arg336 and Arg562. Consequently, the presence of possible substitutions at these two cleavage sites in FVIII have been analysed in patients with VTE. However in a study of 125 patients with VTE, no significant F8 gene mutations were identified (Roelse et al, 1996).
As previously detailed, the majority of plasma FVIII circulates in high-affinity complex with VWF. This VWF-binding is a critical determinant of FVIII half-life. Moreover, elevated plasma VWF levels have been reported in some patients with high FVIII:C levels after VTE (Koster et al, 1995; O'Donnell et al, 1997; Nossent et al, 2006a,b). In this context, several groups have sought to investigate whether VWF gene variation may be important in some patients with elevated FVIII levels. In a study of 13 patients with VTE and high FVIII:C, Bowen et al (2001) failed to identify any sequence mutations within exons 18–24 of the VWF gene that encode the FVIII-binding domain. Moreover, two polymorphisms in this same region of the VWF gene (2365 A/G; exon 18 and 2555 G/A; exon 20) were not associated with any variation in plasma FVIII:Ag levels, or thrombotic risk (Kamphuisen et al, 2001a).
There is also evidence that increased endothelial cell secretion of VWF can lead to increased plasma FVIII:C levels, and thereby increased thrombotic risk. Nossent et al (2006a,b) measured plasma levels of mature VWF:Ag together with VWF propeptide in LETS. In keeping with previous reports, the ratio of VWF:Ag compared to VWF propeptide was then used to assess the respective rates of VWF synthesis and clearance. Interestingly, VWF propeptide levels were significantly increased in patients with VTE compared to age-matched controls. VWF propeptide levels also correlated with VWF:Ag and FVIII:C levels. Finally, and most importantly, plasma VWF propeptide levels constituted a dose-dependent risk factor for VTE. However, after adjustment for FVIII levels, the risk largely disappeared. Cumulatively these data support the hypothesis that increased VWF secretion may contribute to the pathophysiology in at least some cases of high FVIII. Nevertheless, given that <50% patients with high FVIII:C after VTE demonstrate concurrent elevated plasma VWF:Ag levels, it is clear that additional factors must also be implicated.
Alternate possible determinants of elevated FVIII:C
On the basis of the familial clustering data, it seems likely that high FVIII levels represent at least in part a constitutional phenomenon, and thereby a novel genetic thrombophilia. In the absence of any identifiable polymorphisms at either the F8 or VWF gene loci, studies have focused on other candidate genetic loci that may be involved in regulating plasma FVIII levels.
Although the physiological basis underlying FVIII clearance is not well defined, recent studies suggest that the low-density lipoprotein receptor-related protein (LRP) may be important (Lenting et al, 1999; Saenko et al, 1999; Bovenschen et al, 2003). LRP is a cell surface endocytic receptor expressed on a variety of human tissues including hepatocytes. In vitro studies have shown that LRP can mediate cellular uptake and degradation of FVIII (Lenting et al, 1999; Saenko et al, 1999). Moreover, plasma FVIII levels were significantly increased in LRP-deficient mice (Bovenschen et al, 2003). In this context, several groups have investigated a number of known polymorphisms of this putative clearance receptor. Both the C663T and C766T polymorphisms of LRP1 gene have been associated with elevated FVIII:C and VTE (Cunningham et al, 2005; Vormittag et al, 2007). In particular, LRP1 C663T has been shown to influence plasma FVIII:C levels independently of blood group, CRP and VWF levels (Vormittag et al, 2007). In addition, heterozygosity for the 663T allele was associated with a threefold increased odds ratio for VTE. Interestingly, this same polymorphism has also been previously reported to increase the risk of ischaemic heart disease (Pocathikorn et al, 2003).
It is well recognized that plasma FVIII levels increase significantly in response to either epinephrine or vasopressin infusions (Ingram et al, 1968). This effect can be attenuated by prior administration of a β-adrenoreceptor blocker (Ingram & Jones, 1966). In a study of 10 patients with previous DVT and persistently elevated FVIII:C levels (above the 90th percentile for healthy volunteers: 175 iu/dl), Hoppener et al (2004) demonstrated that treatment with the β-blocker propranolol (40 mg tds) for 2 weeks resulted in a significant reduction in plasma FVIII:C levels (223–171 iu/dl; P < 0·0001). Interestingly, following discontinuation of propranolol therapy, plasma FVIII:C levels reverted to their previously high baseline levels. In contrast, β-blocker administration had no effect on plasma FVIII:C levels in a cohort of healthy volunteers. These findings clearly suggest that the sympathetic nervous system may be important in determining high plasma FVIII:C levels after VTE. Furthermore, in view of the increased risk of recurrent VTE in patients with elevated FVIII:C levels, these findings may have important therapeutic implications. However a subsequent small study, which included seven male patients with VTE and high FVIII:C levels failed to identify any significant beneficial FVIII-lowering effect following 2 weeks of propanolol administration (Schönauer et al, 2003). Moreover, two independent studies concluded that coding polymorphisms at the β1 and β2 adrenergic receptors were not associated with elevated FVIII:C levels in patients with VTE (O'Donnell et al, 2003; Nossent et al, 2005).
Aquaporin 2 receptor (AQP2) is a water channel receptor expressed in the kidney. Given that vasopressin administration is associated with an increase in plasma VWF and FVIII levels, Nossent et al (2008) hypothesized that variations in AQP2 function might be important in patients with elevated FVIII levels. Consequently, they performed sequence analysis of the AQP2 gene in 25 patients from LETS. In total, 18 single nucleotide polymorphisms (SNPs) were identified. Although several of these AQP2 SNPs were associated with increased risk of VTE, they were not associated with any significant increases in plasma FVIII or VWF levels.
In addition to these studies examining specific candidate genes, a number of large genome-wide association studies (GWAS) have sought to identify novel genetic loci that contribute to the heritability of FVIII and/or VWF levels (Smith et al, 2010; Antoni et al, 2011). For example, the large CHARGE (Cohorts for Heart and Aging Research in Genome Epidemiology) consortium study, which included more than 23 000 participants of European ancestry, identified six novel candidate genes associated with plasma VWF levels (SCARA5, STX2, STXBP5, CLEC4M, STAB 2 and TC2N) (Smith et al, 2010). Interestingly, three of these genetic loci (SCARA5, STXBP5, and STAB 2) were also associated with plasma FVIII levels. The molecular mechanisms through which these genes influence plasma VWF-FVIII levels have not been elucidated. Nevertheless, an independent case–control study subsequently demonstrated that a SNP in STXBP5 also constituted a novel candidate risk factor for VTE (Smith et al, 2011). Furthermore, Campos et al (2012) recently reported that four intronic SNPs within the F8 gene also significantly associated with plasma FVIII activity levels. Further studies will be required in order to define how such non-coding F8 gene polymorphisms influence FVIII levels.
Elevated FVIII and thrombogenicity
Despite the increasing number of studies describing an association between high FVIII levels and both venous and arterial thrombosis, the mechanism(s) through which elevated FVIII:C levels serve to increase thrombogenicity have not been defined. However, Kawasaki et al (1999) have shown that elevated FVIII:C levels in a mouse model are associated with significantly enhanced thrombus formation. In brief, human recombinant FVIII was administered to transiently increase plasma levels in FVB mice. Subsequently, rate of thrombus formation was quantified following exposure of the carotid artery to photochemical injury. A significant dose-dependent effect of elevated plasma FVIII levels on increasing arterial thrombus size was observed in this model. More recently, Machlus et al (2011) demonstrated that increased FVIII levels in C57Bl6 mice were also associated with a significantly reduced time to occlusion following ferric chloride saphenous vein injury. Together, these in vivo studies confirm that elevated plasma FVIII:C levels are directly involved in the aetiology of venous thrombosis.
Following initiation of coagulation, plasma FVIII is cleaved and activated by thrombin. The resulting FVIIIa heterotrimer is released from its complex with VWF, and is then able to interact with exposed anionic phospholipids, where it functions as a non-enzymatic cofactor for FIXa in the activation of factor X. In view of the critical cofactor function of FVIIIa, it is not surprising that FVIII levels have been predicted to influence the rate of thrombin generation (Jones & Mann, 1994; Butenas et al, 1999). In keeping with the hypothesis that high FVIII levels may serve to increase basal thrombin generation, increased levels of thrombin-antithrombin (TAT) and prothrombin fragment 1 + 2 (F1 + 2) were observed in approximately 80% of patients with confirmed venous thrombosis and increased plasma FVIII:C levels (O'Donnell et al, 2001). Interestingly, this prevalence is much higher than has been reported in association with other prothrombotic states. For example, only one-third of patients with VTE and heterozygous protein C or protein S deficiency have F1 + 2 levels greater than the upper limit of the normal range (Bauer et al, 1988; Mannucci et al, 1992). Moreover, in patients with the F5 R506Q mutation, elevated TAT and F1 + 2 levels were observed in only 23% and 32% respectively (Martinelli et al, 1996).
The in vivo data suggesting that elevated FVIII may exert its thrombotic effect in part through enhancing thrombin generation have been supported by more recent in vitro studies. Using Calibrated Automated Thrombography (CAT) to quantify thrombin generation, two independent groups have demonstrated that elevated plasma FVIII directly influences thrombin generation parameters (including reduced lag time, increased peak thrombin formation and increased endogenous thrombin potential) (Machlus et al, 2009; Ryland et al, 2012). In combination, these in vitro and in vivo experimental data infer that the increased risk of thrombosis associated with elevated FVIII levels results in large part through a direct FVIII-mediated enhancement of thrombin generation. Nevertheless, elevated FVIII may exert other prothrombotic mechanisms by diminishing the influence of the anticoagulant pathway. In particular, a direct inverse relationship between plasma FVIII:C levels and activated protein C resistance has also been reported (Laffan & Manning, 1996; De Mitrio et al, 1999), and may further contribute to thrombotic phenotype observed in patients with high FVIII:C levels.
High FVIII in combination with other thrombophilias
From the epidemiological findings, it is clear that elevated plasma FVIII levels constitute a common independent risk factor for VTE. Consequently, several studies have examined the clinical significance for those individuals who have high FVIII levels in combination with another inherited or acquired thrombophilia. Oestrogen-containing oral contraceptive pills (OCPs) are a well- recognized acquired risk factor for VTE. In a subset of 155 premenopausal women with DVT enrolled in LETS, Bloemenkamp et al (1999) observed that use of OCPs was associated with an odds ratio of 3·8 (95% CI: 2·4–6·0). In this same cohort, women with high FVIII:C levels were also at significant increased risk of venous thrombosis, with an odds ratio of 4·0 (95% CI: 2·0–8·0). Importantly however, the combination of both OCP and high FVIII resulted in an odds ratio of 10·3 (95% CI: 3·7 – 28·9). Similarly, in another case–control study, Legnani et al (2004) reported an even more marked synergistic increase in VTE risk (odds ratio 13·0; 95% CI: 4·9–34·3) for women with combined high FVIII levels and OCP usage. The combination of high FVIII levels and heterozygosity for the F5 R506Q mutation has also been associated with increased risk of VTE (Kamphuisen et al, 2000a; Lensen et al, 2001). Finally, elevated FVIII:C levels have also be identified as an independent risk factor for VTE in patients with underlying malignancy (Vormittag et al, 2009). Cumulatively, these additive effects serve to highlight the potential clinical significance of interactions between the different prevalent prothrombotic tendencies, and emphasize the multi-factorial aetiology of VTE.
Elevated FVIII in clinical practice
Although the clinical utility of thrombophilia screening in patients presenting with VTE remains contentious, such testing is nevertheless performed on a common basis (Baglin et al, 2010). Notwithstanding the pros and cons associated with thrombophilia testing, it seems reasonable that FVIII levels should also be measured when such testing is undertaken. For example, in terms of the prevalence, and degree of associated VTE risk, elevated FVIII levels are at least as important as either F5 R506Q or F2 G20210A mutation in the Caucasian population. Moreover in the African-American population, high plasma FVIII levels represent an even more important prothrombotic risk factor. Of course, one of the principal limitations of current inherited thrombophilia testing is that it fails to predict risk of recurrent thrombosis. In this context, the strong evidence that high plasma FVIII:C levels also constitute an independent and dose-dependent risk factor for recurrent venous thrombosis is of particular importance.
If plasma FVIII levels are to be included in thrombophilia testing, several practical questions need to be considered. In particular, the timing of sample collection and the type of FVIII assay must be addressed. First, in order to minimize the confounding influence of any post-thrombotic acute phase response, FVIII levels should be measured a minimum of at least 3–6 months following the acute VTE (Barcat et al, 2006; Tichelaar et al, 2012), and ideally following completion of therapy (Passamonti et al, 2010). In terms of optimal FVIII assay, plasma FVIII levels can be determined in a functional clotting assay (FVIII:C); in a chromogenic assay (FVIII:Ch); or via a standard enzyme-linked immunosorbent assay methodology (FVIII:Ag). To date, most studies in patients with VTE have assessed FVIII procoagulant activity using a one-stage clotting assay. Previous studies have shown that the one-stage clotting assay is sensitive to the presence of even small amounts of FVIIIa, and thus can result in an over-estimate of plasma FVIII levels if there is ongoing activation of the coagulation cascade. However, several studies have clearly demonstrated that plasma FVIII:C, FVIII:Ch and FVIII:Ag levels are all elevated in patients with VTE (O'Donnell et al, 1997; Kamphuisen et al, 2000a, 2001a; Cristina et al, 2004; Ota et al, 2011). Consequently, if FVIII levels are to be included in the thrombophilia screen, it seems reasonable that the FVIII:C assay is utilized.
The duration of anticoagulant therapy for patients with VTE is determined based upon the risk of recurrent venous thrombosis, together with the risk of bleeding complications associated with anticoagulation. As previously discussed, elevated plasma FVIII levels are associated with significant increased risk of developing recurrent VTE. To date however, only one small prospective study has investigated whether patients with VTE and high FVIII:C levels may benefit from extended duration anticoagulant therapy (Eischer et al, 2009). This study confirmed that the risk of recurrence was increased in patients with high FVIII levels. Furthermore, Eischer et al (2009) also concluded that prolonged anticoagulation (30 months rather than 6 months) was effective in reducing the risk of recurrent VTE. Further adequately-powered randomized trials are required in order to definitively address this key therapeutic management issue.
In conclusion, elevated FVIII levels are a common finding in patients with VTE. Moreover, high plasma FVIII:C levels constitute a dose-dependent risk factor for VTE. These findings have been consistently replicated across a series of independent case–control studies. Current data suggest that the high FVIII levels do not simply reflect a post-thrombotic acute phase response. In particular, accumulating data from animal models support the hypothesis that elevated FVIII can play an important role in the pathogenesis of the thrombosis. Further studies will be required to elucidate the molecular mechanism(s) underlying high plasma FVIII levels in these patients. Nevertheless, in view of the high prevalence and associated risk, it is perhaps unsurprising that studies have estimated that high FVIII levels may be responsible for up to 16% of all symptomatic VTE (Kamphuisen et al, 2001b). Importantly, further clinical trials will also be essential in order to understand the critical observation that elevated FVIII levels may also be useful in identifying a specific subset of patients who are at increased risk for developing recurrent thrombotic events.
This work was supported by a Bayer Hemophilia Special Project Award, and a Science Foundation Ireland Principal Investigator Award (JSOD) SFI 11/PI/1066.