Venous thrombosis of the upper extremity: effect of blood group and coagulation factor levels on risk
Frits R. Rosendaal, Department of Clinical Epidemiology C7-P, P.O. Box 9600, 2300 RC Leiden, The Netherlands. E-mail: firstname.lastname@example.org
Venous thrombosis of the upper extremity is a rare form of thrombosis, accounting for around 4% of all venous thromboses, and for which only a few risk factors are known. This case-control study investigated the effect of coagulation factors on risk of venous thrombosis of the upper extremity. Patients with venous thrombosis of the arm and partner controls were selected from the Multiple Environmental and Genetic Assessment study, a large population-based case-control study. Participants with a malignancy were excluded. Odds ratios (OR) were estimated for elevated levels of factor II, VII, VIII, IX, X, XI, von Willebrand Factor (VWF), and fibrinogen, low levels of protein C, protein S, and antithrombin, and for blood group non-O. Substantially increased risks of venous thrombosis of the upper extremity were found for patients with high levels (above 90th percentile versus below) of factor VIII (OR: 4·2, 95% confidence interval (CI): 2·2–7·9), VWF (OR: 4·0, 95% CI: 2·1–7·8), fibrinogen (OR: 2·9, 95% CI, 1·5–5·7), and for blood group non-O compared to O (OR: 2·1, 95% CI, 1·3–3·6). The other factors were not associated with an increased risk. Elevated levels of several procoagulant factors are associated with a strongly increased risk of venous thrombosis of the upper extremity.
Only 4% of all cases of venous thrombosis are located in the upper extremity (Hill & Berry, 1990; Kommareddy et al, 2002). Therefore, most studies on risk factors for venous thrombosis have focused on venous thrombosis of the leg or pulmonary embolism and only a few studies investigated the risk of venous thrombosis of the upper extremity, often restricted to a small number of risk factors. Malignancy, a central venous catheter (CVC), oral contraceptive use, surgery of the upper extremity, physical activity of the arm and F5 R506Q (factor V Leiden) (Bertina et al, 1994) and F2 (prothrombin) G20210A (Poort et al, 1996) mutation are known to increase the risk of venous thrombosis of the upper extremity in several studies, including the Multiple Environmental and Genetic Assessment (MEGA) study (Girolami et al, 1999; Fijnheer et al, 2002; Vaya et al, 2003; Martinelli et al, 2004; Tesselaar et al, 2004; Blom et al, 2005a; van Rooden et al, 2005; van Stralen et al, 2005).
Individuals with elevated levels of procoagulant factors II, VIII, IX, and XI have an increased risk of venous thrombosis of the leg, with two to threefold increased risks for those in the highest 10% of the population distribution (Koster et al, 1995; Poort et al, 1996; van Hylckama Vlieg et al, 2000; Meijers et al, 2000). Furthermore, blood group non-O is associated with a twofold increased risk of venous thrombosis of the leg (Wu et al, 2008). Only a few studies have investigated the effect of procoagulant factors in patients with venous thrombosis of the upper extremity. Two studies examined the effect of factor VIII and fibrinogen. High levels of factor VIII were found to be present in the same number of patients with venous thrombosis of the upper extremity as in patients with venous thrombosis of the leg (Lechner et al, 2008). Similarly a high prevalence of high levels of fibrinogen was found (Arnhjort et al, 2007). However, in both studies no comparisons with the general population were made and hence no risk estimates were reported.
Deficiencies of anticoagulants protein C, protein S, and antithrombin substantially increase the risk of venous thrombosis of the leg (Rosendaal, 1999). Studies on the association of these defects with upper extremity thrombosis are scarce and have yielded conflicting results (Martinelli et al, 1997; Héron et al, 2000; Leebeek et al, 2000, 2001).
The aim of this study was to investigate the levels of pro- and anticoagulant factors and blood group as risk factors for venous thrombosis of the upper extremity.
The analyses were performed as part of a large population-based case-control study. Patients and control subjects were selected from the MEGA study of risk factors for venous thrombosis. Details of this study have been described elsewhere (Blom et al, 2005b). In the overall study over 5000 consecutive patients with deep venous thrombosis of the leg, a pulmonary embolism, or a venous thrombosis of the upper extremity between the age of 18 and 70 years, and their partners (>3000) were included. Patients were treated in six anticoagulation clinics in the Netherlands.
A total of 329 consecutive patients with a venous thrombosis of the upper extremity were invited to participate in the study between March 1999 and September 2004 and 224 took part (68%). There were no differences in sex and age between those who did and did not participate. Blood was obtained from patients and control subjects included up to July 2002. As the risk factors in the current study involve plasma factor levels, a blood sample was necessary. Consequently, 107 patients and 1825 controls were eligible for this study. Venous thrombosis was objectively confirmed with either ultrasonography, duplex sonography or computed tomography. Partners of patients with venous thrombosis in the overall study were included as control subjects. These were therefore partners of patients with different types of venous thrombosis, including deep-vein thrombosis of the leg and pulmonary embolism. Control subjects did not have a history of venous thrombosis and were aged between 18 and 70 years. Individuals with a malignancy (31 cases and 57 controls) and individuals without a blood sample (seven cases and 329 controls) were excluded from the present study. In total, 69 patients with a venous thrombosis of the upper extremity and 1439 control subjects were included.
All participants were asked to fill in a questionnaire on acquired risk factors for venous thrombosis within a few weeks after the index date. For patients the index date was date of the thrombotic event. For control subjects the index date was the date of diagnosis of thrombosis of their partner. Oral contraceptive use was classed as use within 3 months before the index date. Idiopathic venous thrombosis was defined as present when occurring in the absence of F5 R506Q, F2 G20210A mutation, oral contraceptive use, surgery, plaster cast and, injury within 3 months before thrombosis, CVC within a month before thrombosis and pregnancy or puerperium (<5 weeks after delivery) at time of thrombosis. Both patients and control subjects were invited for blood draw and interview at least 3 months after discontinuation of vitamin K antagonist treatment of the patient. Infection was present when patients reported a febrile disorder at the time of blood draw. When duration of treatment was more than 1 year patients were invited for a blood draw at least 12 months after the thrombotic event. The mean time between index date and blood draw was 10 months for cases (range 4–24 months) and 7 months for controls (range 1–34 months).
All participants filled in an informed consent form and gave permission to obtain information about their medical history. This study was approved by the Ethics Committee of the Leiden University Medical Centre, the Netherlands.
Blood samples were drawn into vacuum tubes containing 0·1-volume 0·106-mol/l trisodium citrate as anticoagulant. Activity of factor II, VII, VIII, X, and XI was measured with a mechanical clot detection method on a STA-R coagulation analyzer. All measurements were performed following the instructions of the manufacturer (Diagnostica Stago, Asnières, France). Levels of factor IX antigen were determined by enzyme-linked immunosorbent essay (ELISA). von Willebrand factor antigen was measured with the immuno-turbidimetric method, using the STA liatest kit (rabbit anti human von Willebrand factor (VWF) antibodies). Fibrinogen activity was measured on the STA-R analyzer according to the method of Clauss. The 20146G/- (rs8176719), 21463C/G (rs7853989), 21867A/G (rs8176749), and 21996C/- (rs8176750) blood group polymorphisms were determined by a 5′nuclease assay (Taqman; Applied Biosystems, Foster City, CA, USA) using a standard PCR reaction mix (Eurogentec, Seraing, Belgium) and an allele-specific fluorescent probe equipped with a minor groove binding moiety (Applied Biosystems).
Definitions of low levels of antithrombin, protein S, and protein C have been described elsewhere (Bezemer et al, 2009). In brief, protein C, protein S, or antithrombin deficiency was defined as a value below the mean minus two standard deviations in the control subjects. Due to unreliable test results, individuals with a blood sample at time of pregnancy were excluded from the analysis of protein C and protein S. The same applied to individuals using oral contraceptives at time of the blood sample in the determination of protein S. Sixty-six out of 69 patients and 1430 out of 1439 controls were evaluable for the analysis of protein C and 57 patients versus 1322 controls for protein S. The risk of vitamin K-dependent factors was determined for 57 patients whose blood sample was drawn after discontinuation of vitamin K antagonist treatment. There were no missing data for the pro- and anticoagulant factors.
Odds Ratios (OR) and 95% confidence intervals (CI) were calculated as an estimate of the risk of venous thrombosis associated with coagulation factor levels and blood group non-O. Odds Ratios were adjusted for age and sex with a logistic regression model. The 90th percentile in the control subjects was used as cut-off point for factors II, VII, VIII, IX, X, and XI, VWF, and fibrinogen. For those factors with an increased risk for patients above the 90th percentile, tertiles were defined on the basis of the distribution among control subjects to investigate a dose response relationship using the lowest tertile as a reference. Individuals with blood group non-O were compared with individuals with blood group O. The thrombotic risks associated with elevated factor VIII and VWF levels and blood group were mutually adjusted. As previous research found a joint effect of factor VIII and oral contraceptive use in patients with a venous thrombosis of the leg, the joint effect of factor VIII above the 90th percentile with oral contraceptive use among women was assessed (Bloemenkamp et al, 1999; Legnani et al, 2004; Kuipers et al, 2009). Participants with low factor VIII and who were not using an oral contraceptive at the time of index date were used as a reference group. The joint effect was adjusted for age. All analyses were performed using the Statistical Package for the Social Sciences (spss) software, version 14 (SPSS Inc, Chicago, IL).
There were slightly more males in the control group than in the patient group: 49% vs. 44%. Median age of the 69 patients was 39 (5–95th percentile 21–65) years compared with 51 (5–95th percentile 29–66) years in the 1439 controls (Table I). The prevalence of infection was similar between cases and control subjects. White blood cell count was 5·6 vs. 5·7 respectively.
Table I. Baseline characteristics of the study population, including 69 patients with a venous thrombosis of the upper extremity and 1439 control subjects.
|Sex (% men)||30 (44)||706 (49)|
|Median age (years)||39||51|
|CVC*||4 (6)||0 (0)|
|Infection†||19 (27)||321 (23)|
|WBC* (mean, SD)|| 5·6 (1·4)|| 5·7 (1·7)|
|Idiopathic thrombosis||22 (32)||–|
|Oral contraceptive use‡||24 (62)||131 (18)|
Mean levels of procoagulant factors in patients with venous thrombosis of the upper extremity and control subjects are shown in Table II. Mean levels of factor VIII, VWF and fibrinogen were higher among the cases. All other procoagulant factors were similar in cases and controls or even higher in the control than the case group. The distribution of blood group differed for cases and controls. After dichotomization by the 90th percentile, elevated levels of factors II, VII, IX, and X were not associated with an increased risk of an upper extremity venous thrombosis (Table III). Mean levels of factor VIII, VWF, and fibrinogen were higher in patients than in controls (mean difference: factor VIII (21·7 iu/dl, 95%CI, 12·7–30·7), VWF (24·8 u/dl, 95%CI, 12·7–36·9), and fibrinogen (0·1 g/l, 95%CI, 0·01–0·3)). Elevated levels of fibrinogen were found in 19% of patients i.e. over 4·15 g/l. An elevated level of fibrinogen (i.e. above the 90th percentile) was associated with a 2·9-fold (95% CI: 1·5–5·7) increased risk of thrombosis compared with low levels (below 90th percentile), after adjustment for age and sex.
Table II. Mean levels of procoagulant factors and the prevalence of blood group in 69 patients and 1439 control subjects.
|FII†||98 (35)||111 (18)|
|FVII†||91 (35)||112 (25)|
|FVIII||131 (44)||109 (37)|
|VWF||138 (68)||113 (49)|
|FIX†||102 (20)||106 (19)|
|FX†||100 (42)||117 (20)|
|FXI||100 (19)||101 (20)|
Table III. The risk of venous thrombosis of the upper extremity for procoagulant factors.
|FII†||129·0 iu/dl||5(9)||1·0 (0·4–2·5)||0·9 (0·4–2·3)|
|FVII†||145·0 iu/dl||2(4)||0·3 (0·1–1·4)||0·5 (0·1–1·9)|
|FVIII||155·0 iu/dl||15(22)||2·5 (1·4–4·6)||4·2 (2·2–7·9)|
|VWF||170·0 u/dl||14(20)||2·5 (1·4–4·6)||4·0 (2·1–7·8)|
|FIX†||131·1 u/dl||5(9)||0·9 (0·3–2·2)||1·1 (0·4–2·7)|
|FX†||141·0 iu/dl||5(9)||0·9 (0·4–2·3)||0·9 (0·3–2·3)|
|FXI||126·0 u/dl||8(12)||1·3 (0·6–2·7)||1·5 (0·7–3·2)|
|Fibrinogen||4·15 g/l||13(19)||2·2 (1·2–4·2)||2·9 (1·5–5·7)|
|Blood group|| ||N of patients blood group non-O||OR (95% CI)||Adjusted OR (95% CI)*|
|Non-O versus O||–||47(68)||1·9 (1·1–3·2)||2·1 (1·3–3·6)|
Twenty-two percent of the patients had an elevated level of factor VIII, i.e. over 155 iu/dl (Table III). Individuals with an elevated level of factor VIII had a 4·2-fold (95% CI: 2·2–7·9) increased risk compared with those with low levels after adjustment for age and sex.
Twenty percent of patients had an elevated level of VWF, i.e. over 170 u/dl. Individuals with elevated levels of VWF had a 4·0-fold (95% CI: 2·1–7·8) increased risk of venous thrombosis compared to those with low levels after adjustment for age and sex.
Of all patients, 68% were blood group non-O compared with 53% of all controls. Blood group non-O was associated with a 2·1-fold (95% CI: 1·3-3·6) increased risk of venous thrombosis compared with blood group O, after adjustment for age and sex.
Associations between factor VIII, VWF and blood group non-O have been described previously (Koster et al, 1995). In control subjects, levels of factor VIII and VWF differed between blood group O and non-O. The mean difference for individuals with blood group O compared with non-O for factor VIII was 34 iu/dl (95% CI: 30–39), for VWF the mean difference was 37 u/dl (95% CI: 32–41). In a model with blood group, von Willebrand factor and factor VIII levels, all effects were attenuated, particularly those of factor VIII (OR 1·7, 95%CI (0·7–3·6) and VWF (OR 1·9, 95%CI (0·8–4·7)), while blood group non-O remained associated with an 1·7-fold risk (95% CI: 1·0–2·8).
None of the patients had a protein S deficiency. Only one patient had low protein C levels and only one patient low antithrombin levels. Therefore, the thrombotic risk for these anticoagulant factors could not be estimated.
Increased risks of venous thrombosis were seen for elevated levels of factor VIII, VWF, and fibrinogen. To explore a dose response relationship between the levels of these factors and the risk of venous thrombosis, factor levels were divided in tertiles. Cut-off values for tertiles (in the controls) and risk estimates are shown in Table IV. With the lowest tertile as a reference category, there was a clear dose response relationship for both VWF and factor VIII levels. For fibrinogen levels both the second and third tertile showed similarly increased risks.
Table IV. Factor VIII, VWF, and fibrinogen and the risk of venous thrombosis of the upper extremity.
|Factor VIII (89, 120)||1†||2·2 (1·3–3·9)||8·8 (3·9–19·8)|
|VWF (87, 122)||1†||2·0 (1·1–3·5)||5·9 (2·8–12·2)|
|Fibrinogen (3·0, 3·5)||1†||2·5 (1·3–4·9)||2·3 (1·2–4·1)|
Since many female patients were using oral contraceptives at time of the thrombotic event and previous studies showed a high joint effect of oral contraceptive use and elevated factor VIII levels in patients with a venous thrombosis of the leg, we considered the possibility of a synergistic effect between elevated levels of factor VIII and oral contraceptive use. Results are shown in Table V. An elevated level factor VIII alone as well as use of oral contraceptive use only, were associated with an increased risk of venous thrombosis of the upper extremity (OR 5·3, 95%CI(1·3–20·5) and 7·5, 95%CI(2·2–25·5) respectively). The thrombotic risk was highest for women with both elevated levels of factor VIII and oral contraceptive use (OR 30·5, 95%CI (8·8–105·7)).
Table V. Joint effect of factor VIII and oral contraceptive use and the risk of thrombosis in the upper extremity.
In this study we assessed the association between levels of pro- and anticoagulant factors and blood group and the risk of venous thrombosis of the upper extremity. Elevated levels of factor VIII, VWF, fibrinogen, and blood group non-O increased the risk of a venous thrombosis of the upper extremity. Elevated levels of factor II, factor VII, factor IX, factor X, and factor XI were not associated with an increased risk. We were not able to assess the risk of venous thrombosis of the upper extremity associated with protein C, protein S, or antithrombin deficiency.
The overall prevalence of deficiencies of anticoagulant factors protein C, protein S, and antithrombin among patients in several studies varies between 2–12% (Prandoni et al, 1997; Héron et al, 2000; Leebeek et al, 2001). In the current study no patients with protein S deficiency were identified; only one patient had a protein C deficiency, and only one patient was antithrombin deficient (both 3%). Given that the population prevalence of these abnormalities varies between 0·2% (protein C deficiency) and 0·02% (antithrombin deficiency), our findings are compatible with highly increased risks.
After dichotomization at the 90th percentile as measured in the control subjects, elevated levels of factor VIII, VWF, and fibrinogen were found to increase the risk of a first venous thrombosis of the upper extremity. There was also a clear dose response relationship for factor VIII and VWF after stratification of the levels in tertiles. For fibrinogen no gradual increase in risk over the tertiles was observed.
Factor VIII, VWF, and fibrinogen are acute phase proteins. Therefore the question arises whether the elevated levels are a cause or consequence of venous thrombosis of the upper extremity. Blood was drawn at least 4 months after venous thrombosis. When comparing patients with a blood draw within one year with those whose blood sample was drawn more than 1 year after the thrombotic event, levels of factor VIII, VWF, and fibrinogen were similar (mean differences: factor VIII (2·4 iu/dl, 95%CI, −21·8 to 26·6), VWF (0·9 u/dl, 95%CI, −36·4 to 38·2), and fibrinogen (−0·2 g/l, 95%CI, −0·6 to 0·3)). Furthermore, the prevalence of reported febrile disorders at time of blood draw and mean white blood cell count were similar between patients and controls. Finally, the effect of blood group is likely to be mediated via VWF and factor VIII levels, and of course invariant. These results indicate that it is unlikely that the elevated levels of factor VIII, VWF, and fibrinogen are due to an acute phase reaction, which corroborates what has been proposed before for patients with venous thrombosis of the leg (Kamphuisen et al, 1999). Thus, elevated levels of factor VIII, VWF, and fibrinogen are likely to be a cause of venous thrombosis of the upper extremity rather then a consequence of the disease.
Blood group non-O was associated with a twofold increased risk of venous thrombosis of the upper extremity. After adjustment for factor VIII and VWF the risk of venous thrombosis of the upper extremity for blood group non-O compared with O was similar to the crude risk estimate, indicating that blood group non-O increases the risk of venous thrombosis via a different pathway than factor VIII and VWF levels. After additional adjustment for blood group and VWF or factor VIII, the risk of venous thrombosis associated with elevated factor VIII levels decreased from a 4·2- to a 1·7-fold increased risk, whereas the risk associated with elevated VWF levels decreased from a 4·0 to a 1·9-fold increased risk. This shows that the effect of factor VIII and VWF is partly due to the effects of blood group non-O. There was a synergistic effect seen for high levels of factor VIII and oral contraceptive use, which was supra-additive.
This is the first study to assess the risk of venous thrombosis of the upper extremity associated with elevated procoagulant factors. Although the number of patients was low due to the exclusion of individuals with a malignancy, we found an increased risk of upper extremity thrombosis for patients with an elevated level of factor VIII, VWF, and fibrinogen. The results for VIII, VWF, fibrinogen, and blood group non-O were similar to results found for venous thrombosis of the leg. Due to the small number of patients with anticoagulant protein C, protein S, and antithrombin deficiency, we were unable to estimate the risk of venous thrombosis of the upper extremity associated with low levels of anticoagulant factors, although the finding of two patients among 69 with such a defect is highly suggestive of an increased risk.
This study showed that elevated levels of several of the procoagulant factors are risk factors for both deep venous thrombosis of the leg as well as venous thrombosis of the upper extremity. However, there are also differences with regard to coagulation factors as risk factors for a venous thrombosis of the leg or of the upper extremity. Factors II, IX and XI have been described as risk factors for venous thrombosis of the leg but do not appear to increase the risk of venous thrombosis of the upper extremity in our study. This might be explained by the number of patients in our study. Although our study is the largest study investigating the effect of coagulation factors on venous thrombosis in the upper extremity among individuals without a malignancy, it only included 69 patients. A hypothetical explanation for the different coagulation factors that give an increased risk for deep venous thrombosis of the leg and or the upper extremity could be that venous thrombosis of the leg is likely to be more often induced by stasis (plaster cast, travel, hospital admission) and venous thrombosis of the upper extremity more often by injury of the vessel wall (CVCs, physical effort). In the presence of an injury of the vessel wall it might be important that factors related to the acute phase are elevated compared to other factors. If venous thrombosis is due to stasis, it might be less important as to which procoagulant abnormality is present. However, this hypothesis should be tested in a different study.
In conclusion, we found an increased risk of venous thrombosis of the upper extremity in patients with an elevated level of fibrinogen, factor VIII, VWF, and in patients with blood group non-O. However, due to the low number of patients the confidence intervals are wide.
The authors wish to thank the directors of the Anticoagulation Clinics of Amersfoort (Dr M.H.H. Kramer), Amsterdam (Dr M. Remkes), Leiden (Dr F.J.M. van der Meer), The Hague (Dr E. van Meegen), Rotterdam (Dr A.A.H. Kasbergen) and Utrecht (Dr J. de Vries-Goldschmeding) who made the recruitment of patients possible. The interviewers Ms. J. C. M. van den Berg, Ms. B. Berbee, Ms S. van der Leden, Ms M. Roosen and Ms E. C. Willems of Brilman also performed the blood draws. Ms I. de Jonge, Ms R. Roelofsen, Ms M. Streevelaar, Ms L. M. J. Timmers and Ms J. J. Schreijer are thanked for their secretarial, administrative support and data management. The fellows Ms I. D. Bezemer, Ms MD J. W. Blom, Ms E. R. Pomp, Ms MD L. W. Tick and Ms K. J. van Stralen took part in every step of the data collection. Ms C. J. M. van Dijk, R. van Eck, J. van der Meijden, Ms P. J. Noordijk, N. van Tilburg, Ms I. van de Linden and Ms Th. Visser performed the laboratory measurements. Dr H. L. Vos supervised the technical aspects of DNA analysis, Prof. Dr. R. M. Bertina supervised the technical aspects of the factor VIII, IX and, protein S analysis. We express our gratitude to all individuals who participated in the MEGA study.