Objectives.—To review the state of the art relating to elevated hemostatic factor levels as a potential risk factor for thrombosis, as reflected by the medical literature and the consensus opinion of recognized experts in the field, and to make recommendations for the use of specific measurements of hemostatic factor levels in the assessment of thrombotic risk in individual patients.
Data Sources.—Review of the medical literature, primarily from the last 10 years.
Data Extraction and Synthesis.—After an initial assessment of the literature, key points were identified. Experts were assigned to do an in-depth review of the literature and to prepare a summary of their findings and recommendations. A draft manuscript was prepared and circulated to every participant in the College of American Pathologists Conference XXXVI: Diagnostic Issues in Thrombophilia prior to the conference. Each of the key points and associated recommendations was then presented for discussion at the conference. Recommendations were accepted if a consensus of the 27 experts attending the conference was reached. The results of the discussion were used to revise the manuscript into its final form.
Conclusions.—Consensus was reached on 8 recommendations concerning the use of hemostatic factor levels in the assessment of thrombotic risk in individual patients. Detailed discussion of the rationale for each of these recommendations is presented in the article. This is an evolving area of research. While routine use of factor level measurements is not recommended, improvements in assay methodology and further clinical studies may change these recommendations in the future.
Recent studies have found that elevated levels of coagulation factors are associated with an increased risk of thrombosis. Elevated levels of factors II, VIII, IX, and XI have primarily been associated with an increased risk of venous thrombosis, while elevated levels of fibrinogen, factor V, factor VII, and von Willebrand factor have primarily been associated with an increased risk of arterial thrombosis. Whether the elevated factor level is itself the cause of the increased thrombotic risk or is changing along with or in response to some other process is not known. Furthermore, for most factors it is not known which elevations are due to polymorphisms in the gene for the coagulation factor itself or some other regulatory factor, or why some factors relate most closely to venous thrombosis and others to arterial thrombosis. When specific gene polymorphisms are discovered, like the prothrombin G20210A transition, they can be evaluated directly. Until that happens, the level of the factor itself will need to be measured to assess risk, as is done now for antithrombin, protein C, and protein S. The assays currently used to measure most coagulation factors were designed to detect factor deficiencies, not modest elevations of the factors. Use of coagulation factor levels to assess risk in individual patients may require new assay development, better standardization and calibration of assays between institutions, and criteria to assess whether the level measured is indicative of the typical average factor level in the patient and potentially the long-term risk of thrombosis, or it only represents a transient change in the level of the factor.
The remainder of this review is separated into sections briefly discussing what is currently known about the epidemiologic relationship of each individual factor to thrombotic risk, followed by several sections discussing common problems in the measurement and interpretation of coagulation factor levels with regard to assessing risk in individual patients. Note that 100% for a coagulation factor level is equivalent to 100 IU/dL.
Fibrinogen is a 340-kd soluble glycoprotein, the plasma component of which is synthesized exclusively in the liver. Fibrinogen is modified by thrombin to produce fibrin monomers that are the primary constituent of the fibrin clot. Decreased levels of fibrinogen are associated with an increased risk of bleeding. Increased levels of fibrinogen have been proposed as potentially contributing to thrombophilia, and fibrinogen polymorphisms resulting in increased fibrinogen levels have been described.1,2 Increased fibrinogen levels are postulated to enhance thrombus formation by altering the kinetics of the coagulation cascade, resulting in increased fibrin formation, augmenting platelet interaction by increased binding to the glycoprotein IIb/IIIa receptor, and increasing plasma viscosity. Additionally, fibrinogen has been demonstrated to have a role in the stimulation of smooth muscle migration, and fibrinogen functions as an acute-phase reactant.3
Reduced fibrinogen most commonly results from acquired conditions, such as decreased synthesis in liver insufficiency and increased consumption in disseminated intravascular coagulation. Rarely, genetic defects in fibrinogen synthesis occur and result in a wide variety of fibrinogen disorders, ranging from severe hemorrhagic afibrinogenemia to the often asymptomatic dysfibrinogenemias, which may lead to an increased tendency to bleed or clot.
Increased levels of fibrinogen have a strong and consistent association with an increased risk of atherosclerotic vascular disease. Hyperfibrinogenemia has been consistently associated with many other cardiovascular risk factors, such as smoking and diabetes mellitus. In these conditions, fibrinogen may be a direct mediator of arterial thrombosis or it may simply be a marker of an underlying inflammatory process. It is thus unclear if fibrinogen is an independent risk factor for arterial thrombosis, a synergistic risk factor, or merely a marker for the presence of atherosclerosis, which is well recognized as a chronic inflammatory condition.4
Studies have consistently demonstrated a relationship between elevated fibrinogen levels and arterial thrombosis. The European Concerted Action on Thrombosis and Disabilities (ECAT) Angina Pectoris Study5 evaluated 3043 patients who underwent coronary angiography because of suspected coronary artery disease. In this study, increased fibrinogen was significantly associated (P = .01) with risk of myocardial infarction or sudden cardiac death after adjustment for age, body mass index, smoking status, and other risk factors. In comparison with the lowest quintile for fibrinogen, the risk of subsequent cardiac events was tripled among those in the highest quintile. The association of fibrinogen with the risk of coronary events was statistically significant (P = .01) after adjustment for C-reactive protein concentrations. Additionally, the risk of cardiac events in patients with hypercholesterolemia was markedly increased in patients with elevated fibrinogen and C-reactive protein, in contrast to patients with low concentrations of these 2 mediators, who had low risk of subsequent coronary events even in the presence of increased total cholesterol.
The Atherosclerosis Risk in Communities (ARIC) study evaluated 14 477 adults who were initially free of coronary heart disease and found a positive correlation (P ≤ .05) with fibrinogen that was moderately strong in both men (relative risk [RR] = 1.76) and women (RR = 1.54) after adjustment of other risk factors.6 In the same patient population, evaluation of 14 700 participants demonstrated an adjusted relative risk of 1.26 for ischemic stroke in comparison of the highest and lowest quartiles of fibrinogen concentration. Determinations of fibrinogen were made by a single measurement, and no correlation with markers of acute-phase response were performed.7
The Framingham Study8 evaluated 2632 subjects using both Clauss and immunoprecipitation assays. A statistically significant association (P < .001) with fibrinogen levels and traditional cardiovascular risk factors was found, and there were higher fibrinogen levels among men and women with prevalent cardiovascular disease. After adjustment for covariates, the immunologic method remained significantly correlated with the prevalence of cardiovascular disease in men and women, while the Clauss fibrinogen assay did not, thus demonstrating the potential for variations in study outcome based on the assay used for fibrinogen determination.
The Edinburgh Artery Study evaluated 1592 patients with peripheral artery disease who were followed for 5 years. Plasma fibrinogen levels were significantly higher (P ≤ .01) in those patients who lacked baseline peripheral arterial disease, but who developed peripheral arterial disease in the course of the study. The significant relative risk (RR = 1.35; P ≤ .05) of development of peripheral arterial disease with a fibrinogen increase of 0.69 g/L persisted after correction for cardiovascular risk factors and baseline ischemic heart disease. Additionally, the results demonstrated that individuals within the highest tertile of fibrinogen distribution had a significantly greater risk (P < .01) of developing peripheral arterial disease compared with those in the lower tertile. In patients with baseline peripheral arterial disease, fibrinogen was not associated with progression of disease.9 In this same patient population, a fibrinogen increase of 0.69 g/L was found to be independently related to the risk of stroke (RR = 1.52; P < .01) in a multivariate analysis. Significant relationships of elevated fibrinogen with angina and myocardial infarct failed to persist with multivariate analysis.10 In neither of these studies were evaluations of C-reactive protein or another marker performed to evaluate the effect of an acute-phase response on the fibrinogen elevations observed.
Fibrinogen has been proposed to have a role in venous thrombosis, although the supporting data are currently limited. In a case control study, a positive relationship between an elevated plasma fibrinogen level >5 g/L and thrombosis in patients who had a first, objectively confirmed episode of deep vein thrombosis was demonstrated.11 A subsequent study on a similar population found the fibrinogen increase to be independent of an acute-phase response.12 Fibrinogen was not significantly associated with venous thrombosis in other studies and was found to not be an additional risk factor for thrombosis in patients with factor V Leiden mutations.13,14
Numerous methods exist for the measurement of plasma fibrinogen. The most frequently used methods are the modified thrombin time method (Clauss assay), in which clotting time is inversely proportional to the fibrinogen concentration; immunologic methods using anti-fibrinogen antibodies; and determination of clot turbidity by a photometric prothrombin time assay. There is significant interassay variability in the determination of fibrinogen, suggesting the need for better standardization of methods and calibration with an international standard.15
Criteria for Diagnosis
No threshold measurement for the determination of hyperfibrinogenemia has been determined. Studies that have evaluated fibrinogen as a risk factor have typically performed quartile comparisons, in which increased risks of atherosclerosis within the upper and lower quartiles were compared, and a significant difference was determined. The comparison of the intermediate levels of fibrinogen fail to have significance when compared with either the high or low quartiles, thus suggesting a broad category of “indeterminate” results that will limit the usefulness of the assay. Additionally, as fibrinogen is an acute-phase reactant, the coexistent determination of another acute-phase reactant, such as C-reactive protein, may warrant evaluation to exclude the elevation of fibrinogen purely as an acute-phase response. Finally, significant intraindividual temporal variance in fibrinogen levels has been demonstrated, suggesting the need for serial determinations prior to definitive patient classification.
Routine measurement of fibrinogen to assess individual thrombotic risk is not recommended for 3 reasons: (1) as Folsom et al6 indicated in the ARIC study, currently there is no universal standardization system for the fibrinogen assay; (2) the independent contribution of fibrinogen to prediction of risk appears to be modest; and (3) insufficient clinical trial data are available to demonstrate that lowering fibrinogen will prevent ischemic heart disease (Tables 1 and 2).
Elevated factor II (prothrombin) levels and the prothrombin gene G20210A transition are associated with an increased risk of venous thrombosis. This disorder is covered in detail elsewhere in this issue (pp 1319–1325).
ELEVATED FACTOR V
Factor V is a 330-kd glycoprotein made in megakaryocytes, vascular endothelial cells, and the liver. Factor V is activated by thrombin. Activated factor V accelerates the conversion of prothrombin to thrombin by activated factor X. Reduced factor V activity slows the clotting process, leading to an increased risk of bleeding. Presumably, increased factor V activity results in an acceleration of clotting, leading to an increased risk of thrombus formation.
Hereditary deficiency of factor V is a rare autosomal recessive disorder. Acquired factor V deficiency can be caused by acute and chronic liver disease, consumptive syndromes, and development of specific factor V immunoglobulin inhibitors. Factor V Leiden is a hereditary abnormality of the activated protein C cleavage site, leading to a increased risk of venous thromboembolism; this disorder is discussed in detail elsewhere in this issue.
Two studies have evaluated the relationship between the level of factor V in blood and the risk of thrombosis. Redondo et al16 reported that elevated factor V activity is associated with an increased risk of myocardial infarction. They compared factor V activity levels in 200 survivors of myocardial infarction with 100 age- and sex-matched healthy control subjects. Median factor V clotting activity was 103% in control subjects versus 111% in survivors of myocardial infarction (P < .001). Subjects with a factor V activity >109% had a 3.3-fold (95% confidence interval [CI] 1.8–6.6) increased risk compared to patients with ≤96% factor V activity. The association remained significant after correction for age, sex, smoking, diabetes, lipid disorders, hypertension, and fibrinogen.
Kamphuisen et al17 measured factor V antigen levels in 474 patients with a history of venous thrombosis unrelated to malignancy versus 474 age- and sex-matched healthy control subjects. Mean factor V antigen levels were 134% in patients and 132% in control subjects (difference not significant). There was no relationship between factor V antigen and venous thrombosis. Factor V and factor VIII antigen levels were correlated, but factor V did not modify the thrombotic risk associated with high factor VIII levels. The normalized activated protein C ratio was not influenced by the factor V antigen level in subjects with or without factor V Leiden.
Factor V is typically measured using a prothrombin time–based clotting assay. Kamphuisen et al17 measured factor V antigen using an in-house enzyme immunoassay.
Criteria for Diagnosis
Currently, there is no evidence indicating that factor V antigen levels are associated with a risk of venous thrombosis. Based on the single study that is available, the separation between low risk (≤96% of factor V activity) and high risk (109%) for myocardial infarction is only 13%. The lack of a widely used international standard for factor V, variations in methods used to measure factor V, and imprecision of current methods make accurate identification of elevated factor V activity difficult.
Routine measurement of factor V to assess individual thrombotic risk is not recommended for 4 reasons: (1) factor V antigen levels are not associated with the risk of venous thrombosis, (2) there is only a single study relating factor V activity and ischemic heart disease, (3) current methodology is inadequate for assessing individual levels, and (4) no studies have shown that prospective reduction of factor V activity or specific treatment directed at patients with elevated levels results in a reduction in risk of ischemic heart disease or stroke.
ELEVATED FACTOR VII
Factor VII is a 50-kd, vitamin K–dependent zymogen synthesized by the liver that is critical for initiation of tissue factor–induced coagulation (extrinsic pathway). Factor VII is converted to a coagulation protease, factor VIIa, in the presence of tissue factor. Factor VIIa converts factor X to factor Xa and factor IX to factor IXa. The conversion of factor VII to factor VIIa is mediated by thrombin or other coagulation proteases, or by an autocatalytic mechanism. The tissue factor–factor VIIa pathway of initiating coagulation is thought to be the major mechanism for initiating thrombin generation.
Factor VII levels are diminished in conditions associated with vitamin K deficiency, including malnutrition, warfarin use, and hepatic or biliary disease. Inherited factor VII deficiency is an uncommon autosomal recessive disorder. Increases in factor VII levels are seen in patients using certain types of oral contraceptives,18 pregnancy,19 and hyperlipidemia,20 as well as in obesity and aging.
Interest in factor VII as a risk factor for thrombosis emerged with the Northwick Park Heart Study.21 Elevated factor VII levels present at study entry were significantly associated with death from ischemic heart disease within 5 years of entry and for the total follow-up period. The number of fatal events was 2 to 3 times greater in the highest tertile for factor VII (level >120% of normal) than for the lowest third. However, more recent epidemiologic studies have not identified elevated factor VII levels as an independent risk factor for either arterial6,9,16 or venous thrombosis.11 A factor VII polymorphism associated with increased factor VII levels22 is not associated with an increased risk of cardiovascular disease.11,23
Factor VII levels are typically measured using a 1-stage, prothrombin-based assay. Assay results can be confounded by the phenomenon of cold activation of factor VII and by the sensitivity of different thromboplastin reagents to activity of factor VII versus VIIa. Factor VII antigen can also be measured using a commercially available enzyme immunoassay.
Criteria for Diagnosis
Not applicable. Despite numerous epidemiologic studies, there is no decisive evidence that an elevated factor VII level is an independent risk factor for venous or arterial thrombosis. There is a relationship between factor VII genotype and factor VII levels, but the data do not link either clearly to disease.
Routine measurement of factor VII to assess individual thrombotic risk is not recommended because factor VII is not an independent risk factor for thrombosis.
ELEVATED FACTOR VIII
The factor VIII gene is large, 186 kb, and has 26 exons. Protein structure is homologous with factor V.24 It is predominantly synthesized in the liver, although there may be a component from the endothelium. It circulates at an average concentration of 0.1 ng/mL (0.6nM). The normal range of factor VIII levels varies about 3-fold; no 5′ sequence variants were noted that correlate with levels.25 Its large size accounts for essentially complete intravascular localization. Its half-life is short (8–10 hours). In circulation, it is stabilized by binding to von Willebrand factor. Thrombin (or factor Xa) activates factor VIII. Factor VIIIa is a calcium-stabilized heterotrimer with reduced von Willebrand factor binding. Factor VIIIa is the cofactor for intrinsic system activation of factor X by factor IXa in the presence of phospholipids and calcium. Platelet or endothelial cell surfaces support intrinsic factor X activation.
Hereditary deficiency of factor VIII leads to hemophilia A, an X-linked bleeding tendency of varying severity depending on the mutation.26 The severe form is associated with frequent spontaneous bleeding episodes; less severe forms include delayed bleeding or prolonged oozing only after significant trauma, including surgery. Deficiency of combined factors V and VIII can also occur and is most commonly due to autosomal recessive deficiency of an intracellular chaperone protein, ERGIC 53.27
Elevated factor VIII levels have been associated with thrombosis. Penick et al28 used partial purification to characterize the elevated activity as the same as in subjects with normal levels, making a nonspecific acceleration of intrinsic system clotting unlikely. Furthermore, specific factor VIII alloantibodies inhibited the elevated factor VIII clotting activities. Because elevation of factor VIII activity can be secondary to acute stress, chronic inflammation, or estrogen effects, or even artifacts from partial clotting during blood drawing, these acquired effects need to be considered. Factor VIII levels also vary with inherited traits, such as one's ABO blood type; those with type O average about 15% lower than those with types A or B. Only recently have epidemiologic studies demonstrated that factor VIII level is an independent risk factor for either venous or arterial thrombosis.
Koster et al29 found factor VIII clotting activities were associated as a continuous variable with venous thrombotic risk. To avoid acquired elevations from acute illness, assays were performed on samples drawn several months after the acute event. Activities above 150% occurred in 25% of 301 consecutive Dutch patients with initial thrombotic episodes, but in <10% of control subjects. Compared to control subjects, the adjusted odds ratio for factor VIII >150% was 4.8-fold, comparable to the risk of factor V Leiden heterozygosity. They also found that von Willebrand factor antigen level and non-O blood group were related on univariate analysis; on multivariate analyses, however, the odds ratio for elevated von Willebrand factor was 1.0 and for non-O blood group, 1.5. Thus, factor VIII was an independent risk factor, non-O may be a mild additive factor, but von Willebrand factor provided no additional risk.
In additional studies, the association of elevated factor VIII clotting activities with venous thrombotic risk was shown to be independent of elevations secondary to inflammation or other known thrombotic risk factors.12,30–32 The association held for factor VIII protein levels ascertained by enzyme-linked immunosorbent assays (ELISAs)17,25 and in chromogenic assays.33 It was an additive risk factor when co-occurring with other genetic risk factors.30,34 Indeed, a familial component does account for at least a significant proportion of one's baseline factor VIII level.32,35,36 The level of plasma von Willebrand factor and ABO blood type (the latter as an indirect effect on von Willebrand factor mediated by similar glycosylation) account for about one third of the population variability of factor VIII.37 Thus, there are quite likely additional genetic factors that influence one's baseline factor VIII level and, once identified, these factors should be examined to determine if they are a direct effect responsible for the factor VIII elevations that are associated with venous thrombotic risk. Although these effects may be within or linked to the factor VIII gene, none were found in screening a 5′ promoter region, including 0.7-kb sequencing.25 Factor VIII levels could readily be influenced by changes in genes at several loci whose proteins are involved in intracellular processing of factor VIII.38 Kyrle et al33 found that the risk of recurrent venous thrombosis among 340 Austrian patients correlated with their factor VIII clotting activity; however, it was not specified when the factor VIII activities were obtained, which would allow determination of any confounding effects of an inflammatory response. Kraaijenhagen et al32 found baseline elevations were more frequent in subjects who had had recurrent venous thrombotic events than in subjects who had experienced a single event, and that elevations were more frequent in both these groups than in matched control subjects. Thus, one's baseline factor VIII level as assessed by a clotting activity determined several months after an acute thrombotic event is a definite mild risk factor for venous thrombosis. It remains to be determined if that risk extends to recurrence. Although one may assume that it is the constitutional, life-long elevation that confers the risk, a single case report of renal vein thrombosis occurring in a hemophilic patient while on factor VIII replacement therapy39 suggests that acute elevations should be studied to determine if they contribute independently to venous thrombotic risk.
Risk factors for arterial thrombosis are generally more difficult to establish because of their lower frequency. Severe factor VIII deficiency may protect against arterial thrombosis,40 although these data are biased toward younger subjects and the number of incidents are low. The case for an association between elevated factor VIII levels and arterial thrombotic risk has been summarized41,42 and is supported by epidemiologic studies.43 The association holds in a 5-year longitudinal study in which relative risks of 1.4 to 1.8 were noted in elderly women with stroke or transient ischemic attacks or mortality from coronary heart events in elderly men, respectively.44 The risk appears at least in part to be due to concurrent elevation of von Willebrand factor.45 Because inflammatory conditions contribute to thrombotic risk,46 it will be important to determine the degree to which factor VIII levels constitute an independent risk factor.
Factor VIII clotting activity is determined by 1-stage clotting assays based on activated partial thromboplastin time reagents. Severe factor VIII–deficient plasma (either from a patient or immunoabsorbed) is used as substrate, and the time for dilutions of an individual subject's plasma to clot is compared with times from a pool of normal human donors' plasmas. Although a wide variety of reagents and instrumentation are used, 1-stage assays are available in most clinical laboratories. Currently, a significant problem in the routine use of 1-stage clotting assays for factor VIII activity is the tremendous interlaboratory variation in results. For example, on a recent College of American Pathologists survey, 587 laboratories reported factor VIII activities; the mean and median values were 130% with a standard deviation of 18.7%. The range of values reported varied from 79% to 190%. This is unacceptable performance for evaluating risk of thrombosis due to elevated factor VIII activity. Two-stage assays were sometimes used in the past with reagents prepared by the investigator's own laboratory; the stages were initial generation of factor IXa and VIIIa (intrinsic system “X-ase”) in a first stage with dilutions of that incubation added to a source of factor X–determining factor Xa by the clotting time. A commercially manufactured chromogenic assay (Coatest; Pharmacia) is available and is similar to the 2-stage procedure, except that factor Xa generation is determined by cleavage of a specific peptide chromogen and the color is read spectrophotometrically; it is used in many European countries. Neither of these kinetic assays are standardized using calibrators for high levels of factor VIII.
Factor VIII antigen levels have been determined by an ELISA using monoclonal antibodies.17 A kit with these antibodies has been commercially available (Immuno-Baxter, Vienna, Austria), although availability is limited. In general, ELISAs show narrower coefficients of variation than clotting assays and are not influenced by spurious factors, such as partial clotting activation during blood drawing. Most other commercially available immunoassays are limited in their ability to bind to factor VIII in plasma.
Criteria for Diagnosis
Based on the study by Koster et al,29 a baseline factor VIII clotting activity >150% is a risk factor for venous thrombosis. To establish the diagnosis of persistently elevated factor VIII activity, the factor VIII level must be shown to be high on more than one occasion and in the absence of an acute-phase response.
A statistical association between one's baseline factor VIII clotting activity and venous thrombotic risk is clear. Providing that the sample is drawn when there is no inflammation or acute clinical illness, no aerobic exercise in the past several to 24 hours, and no use of estrogen effects (including pregnancy), a baseline value >150% is a significant risk factor for individual subjects. This finding is very common (occurs in approximately 20% to 25% of patients with venous thrombosis or thromboembolism) and accounts for about 16% of its population's attributable risk.37 Routine measurement of baseline factor VIII is not recommended to assess individual venous or arterial thrombotic risk for 3 reasons: (1) sample and assay variables, such as the lack of an established standard for elevated levels, (2) tremendous interlaboratory variation in testing, and (3) the yet-to-be-established direct effects or causal relationships. Testing for elevated baseline factor VIII may be useful in some individuals after testing for established genetic components, when the clinical presentation is more severe than explained by the other results.
ELEVATED FACTOR IX
Factor IX is a 57-kd, vitamin K–dependent protein made in the liver. Factor IX is activated by factor VIIa or factor XIa. In turn, activated factor IX along with factor VIIIa activates factor X. Reduced factor IX activity slows the clotting process, leading to an increased risk of bleeding. Presumably, increased factor IX activity results in an acceleration of clotting, leading to an increased risk of thrombus formation.
Hereditary deficiency of factor IX (hemophilia B) is an X-linked recessive bleeding disorder. Acquired factor IX deficiency can be caused by vitamin K deficiency, liver disease, warfarin therapy, nephrotic syndrome, and development of specific factor IX immunoglobulin inhibitors. Factor IX levels increase with age and with the use of oral contraceptives.47
Two recent studies have reported that high levels of factor IX are associated with an increased risk of venous thrombosis. In the first study, factor IX antigen measured by ELISA was compared in 426 patients with a first deep venous thrombosis versus 473 age- and sex-matched healthy control subjects.48 Ten percent of control subjects versus 20% of patients with venous thrombosis had factor IX antigen levels above 129%, for a relative risk of 2.3 (95% CI 1.6–3.5). After correction for age, sex, oral contraceptive use, factor VIII, factor XI, and other vitamin K–dependent clotting factors, the relative risk of venous thrombosis associated with elevated factor IX was 2.0 (95% CI 1.3–3.2). Combined high factor IX and high factor VIII was associated with an 8-fold relative risk of venous thrombosis.
In a second study, factor IX activity was compared in 66 women with a history of venous thrombosis versus 163 healthy control subjects.49 Mean factor IX activity was 150% in the women with venous thrombosis versus 138% in control subjects (P = .07). When a cutoff of 150% was used, elevated factor IX activity was associated with a relative risk of venous thrombosis of 2.34 (95% CI 1.26–4.35; P = .007) after adjustment for age, location, date of admission, and hormone replacement therapy use.
In the first study, factor IX antigen was measured using an ELISA. In the second study, an automated 1-stage clotting assay was used to measure factor IX activity.49
Criteria for Diagnosis
A factor IX antigen level above 129% compared to a reference level of 100% in healthy control subjects (29% increase) was associated with a relative risk of 2.0. A factor IX activity of 150% compared to a mean of 138% in healthy women aged 45 to 64 years (9% increase) was associated with a relative risk of 2.3. Based on current information, no single standard cutoff value is available to assess an increased risk of venous thrombosis associated with elevated factor IX levels.
Routine measurement of factor IX to assess individual venous thrombotic risk is not recommended for 2 reasons: (1) the limited amount of clinical data relating elevated levels to risk of thrombosis and (2) the limited availability of the ELISA used for the measurement of factor IX antigen in one study.
ELEVATED FACTOR X
Factor X is a 59-kd, vitamin K–dependent zymogen involved in the common pathway of blood coagulation. Factor X is synthesized by the liver and secreted into the plasma as a precursor to a coagulation protease. Factor X is activated by the tissue factor–factor VIIa (extrinsic) pathway or by the factor IXa–factor VIIIa (intrinsic) pathway. In conjunction with the cofactor, factor Va, factor Xa catalyzes the conversion of prothrombin to thrombin.
Factor X levels are diminished in conditions associated with vitamin K deficiency, including malnutrition, warfarin use, and hepatic or biliary disease. Acquired factor X deficiency has also been reported with amyloidosis. Inherited factor X deficiency is a very uncommon autosomal recessive disorder for which fewer than 100 cases have been reported.
Increased factor X levels have been reported in users of oral contraceptives50 and in pregnancy.19 However, elevations in factor X during pregnancy are not as substantial as those that occur with other hemostatic factors, such as fibrinogen, factor VII, and factor VIII. One study investigated factor X levels in patients with coronary artery disease. Although factor X activity levels were significantly higher in affected patients, there was no independent association with an increased risk of myocardial infarction.16 The Leiden Thrombophilia Study found that while elevated factor X levels were associated with a 1.6-fold increased risk of venous thrombosis, the risk resolved after adjusting for other vitamin K–dependent coagulation factor levels.51
Factor X activity is generally measured with a prothrombin time–based clotting assay. Factor X antigen can be measured with a commercially available enzyme immunoassay.
Criteria for Diagnosis
Routine measurement of factor X to assess individual thrombotic risk is not recommended because factor X is not an independent risk factor for arterial or venous thrombosis.
ELEVATED FACTOR XI
The factor XI gene is 23 kb and is located on 4q35. Synthesis occurs in both the liver and megakaryocytes. Plasma levels average 5 μg/mL (30nM). Due to its large size, factor XI gene distribution is intravascular. Its half-life is about 3 days in the circulation.
Thrombin activates factor XI, creating an active serine protease, factor XIa. In vitro, contact activation generates factor XIIa, which activates factor XI (neither phospholipid nor calcium are required). This process forms factor XIa, a contact activation product capable of activating factor IX with calcium to begin intrinsic system clotting in activated partial thromboplastin time–based assays. Intrinsic factor IX activation can occur on the surface of activated platelets and possibly on the endothelium,52 although in vivo, extrinsic factor IX activation appears more important to hemostasis.
Factor XIa is both formed by thrombin and creates positive feedback by intrinsic system activation, leading to even higher levels of thrombin, approaching those required to activate thrombin-activatable fibrinolysis inhibitor.53 Inhibition of fibrinolysis could account for an association of factor XI levels with venous thrombosis. It is unclear if infusion of factor XI concentrates leading to thrombosis in some factor XI–deficient patients or elevation of markers of thrombin generation seen after patient infusions54 relates to rising levels of the zymogen or trace amounts of activated factor XI or other proteins in the concentrates.
Hereditary deficiency can cause a bleeding tendency that is usually of mild severity. Two common mutations account for a high frequency among persons of Ashkenazi Jewish ancestry; a variety of other mutations have been found in patients with sporadic occurrence. Bleeding patterns are variable among deficient individuals and with different hemostatic challenges. Even individuals who are homozygous for a nonsense mutation may have a mild bleeding tendency. Mild deficiency has been observed in some patients with Noonan syndrome, although levels do not correlate with the severity of the cardiac abnormalities.
Elevated levels of factor XI were noted among 474 patients (younger than 70 years) with their first episode of venous thrombosis compared to an equal number of matched case control subjects.55 Factor XI antigen levels above the 90th percentile (>121%) were associated with a 2.2-fold elevated risk of thrombosis (95% CI 1.5–3.2). A dose-response relationship between the factor XI level and the risk of thrombosis strengthened the case for a significant association, and the relationship held after adjustment for confounding variables, including other known inherited risk factors, age, sex, and use of oral contraceptives.
Routine factor XI levels are determined by a 1-stage clotting activity assay, comparing dilutions of subject plasma to clotting times of dilutions of a plasma pool from normal donors. Typically, substrate plasma is either from individuals with known severe deficiency (<1% activity) or immunodepleted normal plasma, as activated in an activated partial thromboplastin time–based assay. A chromogenic substrate has also been used, detecting levels >3% of normal after adding inhibitors to inactivate factor XIIa and kallikrein. Among normal subjects, correlation coefficients between results from the chromogenic and either the 1-stage functional assay or an immunoassay were 0.95 and 0.96, respectively.56 Correlation between a radioimmunoassay with polyclonal antibodies and the 1-stage clotting activity was less strong (r2 = 0.68).57 In the ELISA used by Meijers et al,55 intra-assay and interassay coefficients of variation were 4.6% and 7.0%, respectively.
Criteria for Diagnosis
Based on the study by Meijers et al,55 factor XI antigen levels above 121% are associated with a significantly elevated risk of venous thrombosis.
Although the statistical association between a high factor XI level and venous thrombosis does apply to 10% of a Dutch population, routine measurement of factor XI to assess individual venous thrombotic risk is not recommended for 4 reasons: (1) the association between factor XI levels and thrombosis remains to be verified by other studies, and the degree of generalization to other populations has not been determined; (2) the factor XI antigen assays used in these studies are not generally available in clinical laboratories in the United States; (3) the 1-stage clotting assay may be sufficiently more variable to weaken the relationship, and activated partial thromboplastin time–based specific clotting factor assays can vary significantly depending on the reagents and instrumentation used; and (4) clotting factor activities can be influenced by sample variables, such as partial activation that may occur with difficult venipunctures.
ELEVATED VON WILLEBRAND FACTOR
von Willebrand factor is a large multimeric protein synthesized in endothelial cells and megakaryocytes that serves 2 important functions in hemostasis: carrier protein for factor VIII and the ligand that initiates adhesion of platelets to the damaged vessel. von Willebrand factor binds to the glycoprotein Ib receptor on platelets and to collagen in the vessel wall. von Willebrand factor is stored in Weibel-Palade bodies in endothelial cells and platelet α granules. von Willebrand factor levels vary with blood type. Individuals with type O blood have the lowest levels, those with type AB have the highest levels, and those with types A and B have intermediate levels. Reduced functional von Willebrand factor slows the binding of platelets to the wound, leading to an increased risk of bleeding. Presumably, elevated levels of von Willebrand factor could lead to enhanced platelet binding to damaged vessel walls, including ruptured atherosclerotic plaques, accelerating the formation of thrombi and potentially increasing the risk of thrombosis.
Reduced levels and abnormal forms of von Willebrand factor lead to an autosomal bleeding disorder known as von Willebrand disease. von Willebrand factor can be acutely released in a matter of minutes from endothelial cells in response to a variety of vasoactive substances, including epinephrine, bradykinin, and vasopressin and its analogues. von Willebrand factor is also an acute-phase reactant. The concentration in blood can increase chronically in response to inflammation, infection, cancer, trauma, female hormones, sepsis, and other stimuli. von Willebrand factor levels increase with age and may be higher in women as compared to men.58,59
There is growing evidence that some individuals have relatively elevated basal levels of von Willebrand factor in the absence of known acute or chronic stimuli and that these individuals may be at increased risk of arterial thrombosis.60 A number of prospective studies have shown that elevated von Willebrand factor antigen is associated with an increased risk of ischemic heart disease and stroke. The 2 largest trials, the ARIC study6,7 and the ECAT Angina Pectoris Study5 evaluated 14 700 and 3043 subjects, respectively. In the ARIC study, the von Willebrand factor antigen level was significantly associated (P ≤ .05) with an increased risk of coronary heart disease in patients free of cardiovascular disease at the start of the study, after adjustment for age, race, and field center (men, RR = 1.20; women, RR = 1.18). However, the association between coronary heart disease and von Willebrand factor antigen was no longer significant after adjustment for other risk factors, including smoking, hypertension, diabetes, and lipid profile. In the same ARIC study, the von Willebrand factor antigen level was significantly associated with an increased risk of ischemic stroke both after adjustment for age, sex, race, and field center (P < .001) and after adjustment for other risk factors (P ≤ .05; men, RR = 1.29; women, RR = 1.24).
In the ECAT study, the von Willebrand factor antigen level was significantly associated (P = .05) with an increased risk of myocardial infarction or sudden cardiac death in patients with a history of angina pectoris after adjustment for age, field center, and other risk factors (RR = 1.24). Other studies have shown a significant association between von Willebrand factor antigen level and ischemic heart disease after adjustment for age, sex, and other risk factors45; a significant association with acute myocardial infarction,61 recurrent myocardial infarction,62 stroke,63 and peripheral arterial disease10 after adjustment for age and sex, but not after adjustment for other risk factors; an association with fatal ischemic heart disease but not all ischemic heart disease after adjustment for blood group43; and no association between von Willebrand factor levels and stroke.9 Elevated von Willebrand factor levels are associated with an increased risk of venous thrombosis on univariate analysis, but this association is eliminated when factor VIII activity levels are taken into account.29 von Willebrand factor levels are closely correlated with factor VIII levels, blood group, and other acute-phase proteins like fibrinogen. Therefore, an elevation of von Willebrand factor does not appear to be an isolated, independent predictor of thrombotic risk, but part of a larger complex of hemostatic alterations associated with inflammation, atherosclerosis, and arterial thrombosis.
The most common method for measuring von Willebrand factor antigen is enzyme immunoassay. Other studies have used quantitative immunoelectrophoresis (rocket electrophoresis), but it is less precise.64 A newer, faster method that has not been studied prospectively, but correlates closely with the enzyme immunoassay method, is automated latex immunoassay.65
Criteria for Diagnosis
Most prospective studies of von Willebrand factor and ischemic heart disease or stroke have used a cutoff for determining relative risk of approximately 1 SD or 50% above the mean level in the control group. To show that the von Willebrand factor level is not being transiently increased by an acute-phase response, another acute-phase factor like C-reactive protein can be run and if normal indicates no evidence of an acute-phase response.
While elevated von Willebrand factor has been associated with an increased risk of ischemic heart disease, stroke, and peripheral arterial disease in univariate analysis, after correction for other risk factors, elevated von Willebrand factor was not significantly associated with an increased risk in most studies. Routine measurement of von Willebrand factor antigen or activity to assess individual thrombotic risk is not recommended for 2 reasons: (1) it is not an independent risk factor for venous or arterial thrombosis and (2) there are no studies showing that prospective reduction of von Willebrand factor levels or specific treatment directed at patients with elevated levels results in a reduction in risk of ischemic heart disease, stroke, or peripheral arterial disease.
The underlying premise for measuring elevated coagulation factor levels is that if the average level of the factor is increased in the patient long-term, then the patient may be at increased risk of thrombosis long-term. A number of different questions arise when we attempt to apply the findings from epidemiologic studies to individual patients. Do age, sex, or race affect the level of the factor and is the risk associated with an absolute factor level, or does the risk profile change in different groups? Both risk of thrombosis and certain factors increase with age (eg, fibrinogen, factor VII, factor VIII, factor IX, and von Willebrand factor). Are these effects linked or do we need age-specific ranges? Do acquired effects like other diseases or medications affect factor levels, and do the same risk thresholds apply in these instances? How do we assure that the level we are measuring is a true indication of the patient's average baseline level and not a transient change? For example, fibrinogen, factor VIII, and von Willebrand factor are all strong acute-phase reactants that may rise and fall within hours to days. Factors II, VII, IX, and X, along with protein C and protein S, are vitamin K–dependent factors. How do we assure that the levels of these factors represent the average long-term level in the patient and not a transient rise or fall?
The current generation of assays used to measure coagulation factor levels was designed to detect factor deficiencies. Risk of bleeding associated with coagulation factor levels increases with roughly log decreases in factor levels. Compared to normal (100%), 60% to 90% decreases in a coagulation factor may be associated with excess bleeding with major trauma, 95% to 98% decreases with minor trauma, and >99% decrease with spontaneous hemorrhage. In contrast, the difference between low risk and high risk for thrombosis may be separated by as little as 15% above normal. While many factors such as fibrinogen have international standards, others do not. Even for the assays with international standards, they are not widely available or routinely used to calibrate most assays. In a recent College of American Pathologists survey (CED-A) plasma, laboratories were asked to assay many of the coagulation factor levels mentioned in this article. The between-laboratory coefficient of variation for a sample with an approximately normal level of factor ranged from 11.6% for factor X activity to 20.4% for factor IX activity. Owing to the lack of widely used calibrators for some assays and the imprecision of the individual assay methods, it is difficult to define a single cutoff value for risk that will apply to all methods. It may be possible to define relative cutoffs for specific factors, for example, 50% above the mean level determined locally in healthy subjects for a certain factor. Before coagulation factor levels can be routinely used to assess individual risk, work must be done to better standardize and calibrate the assays used.
Criteria for diagnosis of increased risk of thrombosis in individual patients associated with elevated factor levels will need to take into account all of the problems mentioned. For example, for factor VIII it could be that the criteria for diagnosis of increased risk would be a factor level a certain percentage above the normal level in the local population and another measure of acute-phase response, such as a C-reactive protein value that is normal, indicating no evidence of an acute-phase response.
Presented at the College of American Pathologists Consensus Conference XXXVI: Diagnostic Issues in Thrombophilia, Atlanta, Ga, November 9–11, 2001.
Reprints: Wayne L. Chandler, MD, Department of Laboratory Medicine, Box 357110, University of Washington, Seattle, WA 98195 (firstname.lastname@example.org)