Article Text


Thrombophilic factors are not the leading cause of thrombosis in Behçet’s disease
  1. M Leiba1,
  2. U Seligsohn2,
  3. Y Sidi1,
  4. D Harats3,
  5. B A Sela4,
  6. J H Griffin5,
  7. A Livneh6,
  8. N Rosenberg2,
  9. I Gelernter7,
  10. H Gur1,
  11. M Ehrenfeld1
  1. 1Department of Internal Medicine C, Sheba Medical Centre,Tel-Hashomer, Israel
  2. 2Thrombosis and Haemostasis Research Institute, Sheba Medical Centre
  3. 3Institute of Lipid and Atherosclerosis Research, Sheba Medical Centre
  4. 4Institute of Chemical Pathology, Sheba Medical Centre
  5. 5Department of Molecular and Experimental Medicine, Scripps Research Institute, La Jolla, California, USA
  6. 6Department of Internal Medicine F, Sheba Medical Centre
  7. 7Statistics Laboratory, Tel-Aviv University, Ramat-Aviv, Israel
  1. Correspondence to:
    Dr M Ehrenfeld
    Department of Medicine C, Sheba Medical Centre, Tel-Hashomer 52621, Israel;


Background: Venous and arterial thromboses occur in patients with Behçet’s disease and are associated with significant morbidity and mortality. Studies on a possible association between the occurrence of thrombosis and thrombophilia in patients with this disease have been controversial.

Objective: To determine the prevalence of the most common thrombophilias and dyslipidaemia in patients with Behçet’s disease with and without thrombosis.

Methods: Blood samples from 107 patients with Behçet’s disease who had or did not have thrombosis were analysed for factor V Leiden, prothrombin G20210A polymorphism, methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism, factor VIII level, homocysteine and C reactive protein concentrations, dyslipidaemia, and plasma glucosylceramide.

Results: There was no difference between patients with and without thrombosis in the prevalence of prothrombin G20210A polymorphism, factor V Leiden, homozygous MTHFR C677T, or plasma concentrations of homocysteine, C reactive protein, or glucosylceramide. In contrast, patients with thrombosis were found to have significantly higher mean levels of factor VIII, total cholesterol, triglycerides, VLDL cholesterol, and apolipoproteins B-100, C-II, and C-III than those without thrombosis. Multistepwise logistic regression analysis showed that triglyceride concentration was the best marker associated with thrombosis (p = 0.008), with an estimated odds ratio of 1.58 (95% confidence interval, 1.09 to 2.30) for a difference of 40 mg/dl.

Conclusions: Thrombophilia does not seem to play a major role in the tendency to thrombosis in Behçet’s disease. However, dyslipidaemia, predominantly hypertriglyceridaemia, might be a risk factor.

  • DVT, deep vein thrombosis
  • Lp(a), lipoprotein (a)
  • MTHFR, methylenetetrahydrofolate reductase
  • SVT, superficial vein thrombosis
  • Behçet’s disease
  • thrombophilia
  • dyslipidaemia
  • factor VIII

Statistics from

Behçet’s disease is a chronic multisystem inflammatory disorder that can involve the skin, eyes, central nervous system, arteries, veins, lungs, and gastrointestinal tract. The common manifestations are recurrent oral and genital ulcers and ocular inflammation. Venous or arterial thromboses occur in 7–38% of patients.1 Venous thrombosis is more common than arterial thrombosis, with relative frequencies of 90% and 10%, respectively. Deep or superficial venous thromboses in the lower extremities are the common sites of thrombosis.2,3

Vasculitis underlies the thrombotic tendency in Behçet’s disease, but it is not clear why some patients present with thrombosis and others do not. Decreased levels of tissue plasminogen activator and thrombomodulin, and increased levels of plasminogen activator inhibitor-1, von Willebrand factor, thrombin-antithrombin complex, prothrombin fragment 1+2, and lipoprotein (a) (Lp(a)) have all been described in patients with Behçet’s disease and related to the thrombotic tendency.4–6 The possible role of acquired and inherited thrombophilias has also been evaluated. However, there are conflicting data in patients with and without thrombosis over the prevalence of factor V Leiden, methylenetetrahydrofolate reductase (MTHFR) C677T homozygosity, homocysteine level, prothrombin G20210A polymorphism, and levels of factor VIIa, factor VIII, and factor XII.4,6 A recently described thrombophilic index—decreased plasma glucosylceramide concentration—has not been examined yet in patients with Behçet’s disease.7

Our aim in this study was to determine, in a relatively large cohort of patients with Behçet’s disease with and without thrombosis, the prevalence of common thrombophilic factors such as factor V Leiden, prothrombin G20210A, homozygous MTHFR C677T, and increased levels of factor VIII and homocysteine. We also measured C reactive protein, plasma glucosylceramide, and indices of dyslipidaemia.


Study group

Records of patients with Behçet’s disease who attended the Sheba Medical Centre between 1984 and 1999 were reviewed. The initial cohort with a diagnosis of Behçet’s disease according to the computerised survey included 175 patients, 10 of whom had died over the years and were thus not included in the study. Of the remaining 165 patients, 58 were either non-cooperative, did not fit the inclusion criteria, or were lost to follow up.

All living patients were invited for an interview and a detailed history was taken. The institutional review board of the Sheba Medical Centre approved the study, and informed consent was obtained from all participants. The diagnostic criteria of the International Study Group for Behçet’s disease were used for inclusion.8 Age, sex, body mass index (BMI), disease duration, symptoms, drugs, and past thrombotic events were recorded. Based on the history and the medical records, patients were categorised into those with a history of venous or arterial thrombosis and those without such events. Venous thromboembolic events were regarded as such when imaging examinations—that is, Doppler and compression ultrasonography, angiography, or ventilation–perfusion lung scans—were consistent with the diagnoses of deep vein thrombosis or pulmonary embolism, respectively. Cerebral artery thrombosis was diagnosed by computed tomography or magnetic resonance imaging, and transient ischaemic attacks by a typical clinical presentation. Myocardial infarction was diagnosed by typical clinical presentation, ECG changes, and raised levels of troponin or other cardiac enzymes.

After 12 hours of fasting, venous blood was drawn from all patients. From those who had thrombosis, blood was taken at least three months after the event. Blood was collected in EDTA for DNA extraction and measurements of homocysteine and C reactive protein, and in buffered citrate for assays of factor VIII and plasma glucosylceramide. Plasma was separated soon after taking blood. Serum was separated from clotted blood for analysis of the lipid profile.

Analysis of prothrombotic polymorphisms

DNA extraction was carried out by a standard method. Factor V Leiden (G1691A) was detected by polymerase chain reaction (PCR) and MnlI digestion, MTHFR C677T by PCR and HinfI digestion, and prothrombin G20210A as previously described.9

Factor VIII level was measured by an aPTT based assay using factor VIII immunodepleted plasma. Commercial human freeze dried reference plasma was used to construct a standard curve. The normal range of factor VIII level was 60 to 160 U/dl (±2 SD).

Homocysteine concentration

Homocysteine concentration was measured following labelling of plasma with monobromobimane for fluorescence detection after separation by high performance liquid chromatography (HPLC).10 This method measures total homocysteine (free and protein bound) and mixed disulphides. The normal values in adults ranged between 5 and 15 μmol/l (±2 SD).

C reactive protein

C reactive protein level was measured by a highly sensitive turbidimetric method using Olympus AU 2700 apparatus. The upper limit of the normal range (+2 SD) was 5 mg/l.

Lipid profile

Serum total cholesterol, high density lipoprotein (HDL) cholesterol, low density lipoprotein (LDL) cholesterol, very low density lipoprotein (VLDL) cholesterol, and triglycerides were determined by colorimetric enzymatic procedures with an Olympus autoanalyser which was standardised against reference samples kindly provided by the standardisation programme of the Centers for Disease Control, USA. The Olympus system for direct determination was used for HDL measurement. Serum concentrations of apoliproteins A-I, B-100, C-II, C-III, and Lp(a) were measured by an immunoturbidometric technique. Apo C-II was measured using Kamia Biomedical Company kit. The within-run and between-run assay coefficients of variation were 6% and 7%, respectively. These results were similar to those obtained in other studies.11,12


Glucosylceramide concentration in plasma was determined as described elsewhere, using HPLC analysis of organic solvent lipid extracts.7 The normal range (±2 SD) was 4.85 to 9.59 μg/ml.

Statistical analyses

Univariate analyses were carried out to compare subjects with and without thrombosis. Continuous variables were compared using t tests. Discrete variables were compared using χ2 tests. Multivariate stepwise logistic regression was used to estimate the odds ratio for occurrence of thrombosis associated with potential risk factors among subjects with Behçet’s disease. Age was included as a covariate in the logistic regression models to adjust for any age differences among subjects with and without thrombosis. One analysis considered only readily available clinical measurements as potential risk factors. A separate analysis also included specific measurements of apolipoproteins.


One hundred and seven patients (65 female and 42 male) met the inclusion criteria of Behçet’s disease. Their mean age was 42.5 years at the time of diagnosis (range 12 to 78). The clinical manifestations of the patients are summarised in table 1. Seventy eight per cent of the patients were of Sephardic-Jewish origin, 13% were Ashkenazi (European) Jews, and 9% were of mixed origin. A mean delay of 10 years was noted from onset of manifestations until the diagnosis was made. Symptoms and signs did not differ between patients of different sex or ethnic origin, or between HLA-B5 positive and negative patients.

Table 1

 Differences in clinical manifestations in patients with Behçet’s disease with and without thrombosis

Thirty three patients (30.8%) had a history of at least one thrombotic event. Eighteen (16.8%) had venous thromboembolism, of whom 10 had deep vein thrombosis (DVT), four had superficial vein thrombosis (SVT) of a lower extremity, three had pulmonary embolism (in one this was associated with DVT), and one had hepatic vein thrombosis. Eleven patients (10%) had arterial occlusions which involved the cerebral arteries in two, the coronary arteries in one, a peripheral artery of a lower extremity in two, and the central retinal artery in two. In addition, three patients had typical transient ischaemic attacks, and one had an aortic aneurysm. Four patients (3.7%) had both arterial and venous thrombosis.

The mean time lag between diagnosis of Behçet’s disease and blood sampling of the non-thrombotic group was 14.9 years (range 0 to 55), with a median of 13 years. The mean time lag between the thrombotic event and blood sampling was 7.0 (4.3) years, with a median of 6 years (range 6 months to 15 years).

The mean (SD) age of patients with thrombosis was 45.6 (12.5) years and of patients without thrombosis, 41.0 (14.9) years. The occurrence of thrombosis was not significantly related to age, sex, ethnic origin, BMI, HLA-B5 status, or other diseases such as diabetes mellitus and hypertension. Furthermore thrombosis was not associated with any of the drugs used in Behçet’s disease, including corticosteroids, colchicine, methotrexate, azathioprine, and cyclosporin A (p = 0.3–1). Patients who had a thrombotic event had a significantly higher frequency of recurrent genital ulcers and ophthalmic, gastrointestinal, and central nervous system manifestations (table 1).

Table 2 shows that there were no statistically significant differences between patients with and without thrombosis in the prevalence of the three common prothrombotic polymorphisms and concentrations of homocysteine, C reactive protein, and glucosylceramide. Similar non-significant differences were found when patients with arterial thrombosis alone were compared with those without thrombosis, as well as when the patients with venous thrombosis alone were compared with those without thrombosis. The only difference found was in the mean factor VIII level which was significantly higher in patients with thrombosis.

Table 2

 Levels and prevalence of common thrombophilic factors and C reactive protein in Behçet’s disease patients with and without thrombosis

Mean concentrations of total cholesterol, triglycerides, VLDL, apolipoproteins A-I, B-100, C-II, and C-III were significantly higher in patients with thrombosis than in those without thrombosis (table 3). A separate analysis confined to patients with venous thrombosis showed that most concentrations of the measured lipid variables were still significantly higher than in patients without thrombosis.

Table 3

 Lipid concentrations, body mass index, and factor VIII activity levels in patients with Behçet’s disease

Table 4 shows lipid profile variables and factor VIII activity according to their relative risk for thrombosis, comparing patients with Behçet’s disease with and without thrombosis after adjustment for body mass index, age, and sex.

Table 4

 Lipid profile variables and factor VIII activity and their relative risk for thrombosis: comparison between Behçet’s disease patients with and without thrombosis

Among the standard clinical measurements, triglyceride concentration proved to be the best marker for the occurrence of thrombosis (p = 0.008). Mean (SD) triglyceride concentration in the patients without thrombosis was 1.16 (0.57) mmol/l (the average value for triglycerides in our CDC controlled laboratory for an Israeli population matched for age and sex is 1.20 (0.71) mmol/l), and it was 1.73 (1.04) mmol/l in the patients with thrombosis. The estimated odds ratio for a difference of 0.45 mmol/l (40 mg/dl) triglyceride was 1.58 (95% confidence interval (CI), 1.09 to 2.30). Although age and BMI were included in the logistic regression model, they had only a minor effect, reflecting the fact that patients with and without thrombosis had similar age and BMI distributions.

Once the effects of triglyceride and age were accounted for, no other variables made a statistically significant contribution to being a marker associated with the occurrence of thrombosis. The potential effect of factor VIII was examined more closely by fitting a logistic regression model with triglyceride, factor VIII, and age. The estimated odds ratio for a 20 U/dl increase in factor VIII was 1.39 (95% CI, 1.002 to 1.92).

When apolipoproteins were also considered as potential risk factors, the strongest marker found to be associated with the occurrence of thrombosis was apolipoprotein C-III (p = 0.002). The estimated odds ratio for thrombosis associated with an increase of 4 mg/dl in apolipoprotein C-III concentration was 1.76 (95% CI, 1.17 to 2.65). There was, however, a highly positive correlation between apolipoprotein C-III and triglyceride concentration, so that once apolipoprotein C-III was accounted for, there was no additional predictive value associated with triglyceride concentration. These two variables were almost equally effective as markers associated with the occurrence of thrombosis.


The pathogenesis of arterial and venous thrombosis in Behçet’s disease is not completely understood. It is generally accepted that vasculitis, a hallmark of Behçet’s disease, partially explains the initiation of thrombosis in small as well as large blood vessels.13 Attempts to identify additional prothrombotic factors have so far been conflicting. For example, in Turkey, where the prevalence of Behçet’s disease is probably the highest in the world, factor V Leiden was associated with the occurrence of thrombosis in several but not all studies,6,14 and recent studies from Italy and Spain also did not reveal such an association.15,16 In the present study none of the common prothrombotic polymorphisms nor homocysteine concentration was significantly associated with thrombosis. Increased factor VIII level, recently shown to be a risk factor for idiopathic venous thrombosis,17 was the only prothrombotic variable which appeared to be associated with the occurrence of thrombosis.

We are aware of the rather long lag period between the thrombotic events and the analyses performed, related to the retrospective nature of our study. However, such possible bias would have affected both the thrombotic and the non-thrombotic group.

The main finding in this study was the lack of significant association between any of the thrombophilic factors and thrombosis in Behçet’s disease patients. Interestingly, an association between thrombosis and dyslipidaemia was found. Traditionally, dyslipidaemia has been associated with atherothrombotic diseases. In recent years, idiopathic venous thrombosis was thought to be associated with hypertriglyceridaemia; however, after adjustment for BMI, triglycerides were not found to be an independent risk factor for venous thrombosis.18 Nevertheless, hypercholesterolaemia,19 increased serum concentration of lipoprotein (a),20 and decreased plasma concentration of glucosylceramide7 are associated with a thrombotic tendency. The mechanisms whereby dyslipidaemic changes affect blood coagulability vary, as recently reviewed elsewhere.21 For example, triglyceride-rich lipoproteins such as VLDL enhance the conversion of prothrombin to thrombin by factor Xa22 and augment the activation of factor VII by factor Xa,23 whereas glucosylceramide, a neutral glycosphingolipid, augments the inactivation of factor Va by activated protein C.7 Consequently, increased concentrations of VLDL and decreased concentrations of glucosylceramide augment thrombin generation, the former by increased procoagulant activity and the latter by decreased anticoagulant activity. Lipoprotein (a) apparently competes with plasminogen owing to structural homology and thus an increased concentration of this component may lead to decreased fibrinolysis.20

In the current study neither increased mean concentration of lipoprotein (a) nor decreased mean concentration of glucosylceramide was observed in Behçet’s disease patients with thrombosis compared with patients without thrombosis. In contrast, the main statistically significant differences between these groups were in mean concentrations of total cholesterol, triglycerides, VLDL, and apolipoproteins B-100, C-II, and C-III (table 3). The multistep regression analysis showed that the concentration of either triglycerides or apo C-III were the best markers associated with the occurrence of thrombosis.

Why some patients with Behçet’s disease manifest changes in their lipid profile and others do not remains to be established. It is also unclear at the present time whether the use of lipid lowering agents would decrease the morbidity associated with major thrombotic events affecting patients with Behçet’s disease. However, such an option should be borne in mind in view of a recent study which suggested that the use of statins reduces the risk of idiopathic venous thrombosis.24


View Abstract


  • We thank Professor David Steinberg from the Statistics Laboratory, Tel-Aviv University, for his advice in the statistical analyses, Dr Hiroshi Deguchi and Ms Young Mee Lee (The Scripps Research Institute) for measuring plasma glucosylceramide, and Dr Adi Leiba, Department of Internal Medicine D, Sheba Medical Centre, for invaluable assistance and support.

    Grant support: Katznelbogen grant, Sackler Faculty of Medicine, Tel-Aviv University; and National Institutes of Health grant HLZ1544.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.