- Top of page
- Patients and methods
Summary. Elevated plasma levels of interleukin 8 (IL-8) were previously shown to be associated with recurrent venous thrombosis. To assess the risk of venous thrombosis, IL-8 plasma concentrations were measured in patients and control subjects of the Leiden Thrombophilia Study (LETS). This population based case–control study included 474 patients with a first deep-vein thrombosis and 474 age- and sex-matched controls. The risk of venous thrombosis for subjects with elevated IL-8 levels (above 90th percentile of controls) compared with subjects with IL-8 levels below the 90th percentile was increased 1·8-fold (95%CI 1·2–2·8). Adjusted for age and sex, the odds ratio was 1·9 (95%CI 1·3–2·8). IL-8 concentrations were weakly correlated with age, male sex, and concentrations of C-reactive protein, factor VIII coagulation activity and homocysteine, but adjustment for these factors did not substantially affect the association between IL-8 and venous thrombosis. Our results suggest that IL-8 is a risk factor for venous thrombosis.
Classic risk factors for venous thrombosis include acquired factors, such as immobilization, surgery and malignancies, and genetic risk factors, i.e. activated protein C resistance (factor V Leiden), and deficiencies of protein C (PC), protein S (PS) or antithrombin (AT). Recently, elevated plasma levels of factor VIII, hyperhomocysteinaemia and prothrombin 20210A have been added to the series of risk factors (reviewed by Rosendaal, 1999). A previous study in patients with recurrent venous thrombosis suggested interleukin 8 (IL-8) to be a risk factor for venous thrombosis (Van Aken et al, 2000).
Interleukin 8, a C-X-C chemokine, is produced by several cell types including endothelial cells, peripheral blood monocytes, neutrophils, epithelial cells and fibroblasts (reviewed by Hoch et al, 1996 and Luster, 1998). The main function of IL-8 is the activation of integrin-mediated adhesion of neutrophils (Luster, 1998) and it was recently shown that IL-8 is a powerful trigger for adhesion of monocytes to vascular endothelium (Gerszten et al, 1999). Several studies have identified IL−8 in association with various acute and chronic inflammatory conditions including sepsis, psoriasis, rheumatoid arthritis and asthma (reviewed by Hoch et al, 1996). Proinflammatory cytokines IL-1β and tumour necrosis factor α (TNF-α), and viral and bacterial stimuli can induce IL-8 production.
The importance of inflammatory responses in venous thrombosis has been shown in in vitro studies (Johnson et al, 1996; Senden et al, 1998) and in animal models (Wakefield et al, 1993, 1995, 1997), but little is known about the association of inflammatory mediators and venous thrombosis in humans. Studies using a human experimental endotoxaemia model have suggested an interaction between inflammatory mediators and blood coagulation (Van Deventer et al, 1990). Infusion of endotoxin in healthy volunteers resulted in increased plasma levels of IL-1β, IL-6, IL-8 and TNF-α as well as of coagulation activation markers. Recently, elevated concentrations of C-reactive protein (CRP) have been detected in patients with a first episode of venous thrombosis, suggesting an association with systemic inflammation (Kamphuisen et al, 1999).
Based on these considerations, we theorized that Il-8 levels may play a causative role in venous thrombosis. In order to obtain evidence for this supposition, the association between elevated plasma levels of IL-8 and venous thrombosis was analysed in a case–control study of patients with a first episode of venous thrombosis.
- Top of page
- Patients and methods
The mean age of the whole population was 45 years (range 15–72), 57% were women. Thirty-six per cent of both patients and control subjects were current smokers, while the mean body mass index (BMI) was 26 (range 14–46) kg/m2. The mean IL-8 concentration was higher in the patients than the controls [8·0 pg/ml (95%CI 6·0–10·0) versus 4·4 pg/ml (95%CI 4·0–4·8)]. Eleven per cent of the controls and 9% of the patients had plasma levels below the detection limit (2·0 pg/ml). The IL-8 plasma concentrations of individual patients and controls are shown in Fig 1.
Figure 1. Plasma levels of IL-8 are elevated in patients with venous thrombosis compared with control subjects. The solid lines represent the mean, the dotted lines the detection limit. The numbers at the top of the cases column indicate the IL-8 levels that were above 100 pg/ml in five patients.
Download figure to PowerPoint
To evaluate the association between IL-8 plasma concentrations and venous thrombosis, a cut-off value of 7·1 pg/ml (90th percentile of the control population) was used. Seventy-seven (17%) patients were detected with elevated IL-8 concentrations (above 90th percentile) compared with 45 (10% by definition) of the controls [odds ratio 1·9 (95%CI 1·2–2·8)]. Using the cut-off of 16·2 pg/ml (99th percentile), 23 (5%) patients with elevated IL-8 were detected in comparison with four (1% by definition) controls [odds ratio 6·0 (95%CI 2·0–17)]. Odds ratios using other cut-off values are shown in Table I. By dividing the IL-8 concentrations into quintiles and comparing each quintile with the lowest, we found a concentration-dependent association pointing towards a threshold effect. The risk was not increased in the second [odds ratio 1·0 (95%CI 0·7–1·6)] and third quintile [odds ratio 1·0 (95%CI 0·6–1·5)], while in the fourth quintile [odds ratio 1·2 (95%CI 0·8–1·8)] and in the fifth quintile [odds ratio 1·8 (95%CI 1·2–2·8)], the odds ratio was increased.
Table I. Thrombosis risk for different cut-off points of IL-8 concentrations.
|IL-8 plasma concentrations||Patients (466)||Controls (462)||OR (95%CI)|
|< 80th percentile (5·6 pg/ml)||336||373|| |
|> 80th percentile||130||89||1·7 (1·2–2·3)|
|< 90th percentile (7·1 pg/ml)||389||417|| |
|> 90th percentile||77||45||1·9 (1·2–2·8)|
|< 95th percentile (8·2 pg/ml)||401||440|| |
|> 95th percentile||65||22||3·3 (2·0–5·5)|
|< 99th percentile (16·2 pg/ml)||443||458|| |
|> 99th percentile||23||4||6·0 (2·0–17)|
In control subjects we examined which factors influenced the IL-8 concentration. The mean IL-8 concentration was higher in men [5·3 pg/ml (95%CI 4·4–6·1)] than in women [3·8 pg/ml (95%CI 3·4–4·1)]. IL-8 concentrations increased by 0·2 pg/ml (95%CI−0·5–1·0) per 10 years of age. Distribution of the control subjects in age groups showed that 30 (10%) of the subjects between 40 and 51 years, and 70 (23%) above the age of 51 years, had an elevated IL-8 concentration (> 7·1 pg/ml, 90th percentile) compared with 22 (7%) of the subjects under the age of 40 years. The mean IL-8 concentration was higher in subjects withelevated (> 5·1 µg/ml, 90th percentile) CRP plasma levels [5·5 pg/ml (95%CI 3·0–8·0)] than in other subjects [4·3 pg/ml (95%CI 3·9–4·7)].
Risk factors for venous thrombosis include high levels of factor VIII coagulant activity (FVIII:C) and homocysteine. Among healthy subjects with elevated FVIII:C concentrations (> 100 IU/dl) the IL-8 concentration was slightly increased [n = 344, 4·6 pg/ml (95% CI 4·0–5·1)] compared with subjects with low FVIII:C (< 100 IU/dl)[n = 130, 4·0 pg/ml (95%CI 3·6–4·4)]. Homocysteine was analysed in the subjects recruited in the Leiden area only (control subjects n = 309). The IL-8 concentration was increased in subjects with elevated homocysteine concentrations(> 90th percentile, 16·7 µmol/l) [n = 30, 4·8 pg/ml (95%CI 3·9–5·7) versus n = 279, 4·2 pg/ml (95%CI 4·0–4·5)]. Analysis of life-style factors BMI and smoking showed that among overweight subjects (BMI > 90th percentile, 31 kg/m2) the mean IL-8 was higher [4·6 mg/l (95%CI 2·8–6·3)] than among other subjects [4·4 mg/l (95%CI 4·0–4·8)]. In smokers the IL-8 concentration was not different from non-smokers [respectively 4·2 mg/l (95%CI 3·8–4·7) versus 4·5 (95%CI 3·9–5·1)].
Next, we investigated the role of the above-mentioned factors on the association between elevated IL-8 (> 90th percentile) and venous thrombosis. Adjustment for FVIII:C [odds ratio 1·7 (95%CI 1·1–2·6)], homocysteine [odds ratio 1·7 (95%CI 1·1–2·6)], CRP [odds ratio 1·8 (95%CI 1·2–2·6)] or BMI [odds ratio 1·7 (95%CI 1·1–2·6)] only marginally affected the association between venous thrombosis and elevated IL-8 plasma concentrations.
The presence of other thrombosis risk factors (factor V Leiden, prothrombin G20210A, factor VIII, factor IX, factor XI, and antithrombin, protein C or protein S deficiencies) did not affect the association between elevated IL-8 (> 90th percentile) and venous thrombosis [odds ratio 1·7 (95%CI 1·1–2·7)], and neither did exclusion of these patients [odds ratio 1·9 (95%CI 1·2–2·9)].
Patients were included in the study at different time intervals after the thrombotic event [mean 21 (range 6–68) months]. IL-8 concentration was not related to this time interval. Patients who were seen more than 28 months after the thrombotic event had a similar frequency (16%) of elevated IL-8 concentration as those seen within 12 months after the thrombosis (18%). Additionally, analysis after exclusion of the subjects with an elevated CRP concentration (> 5·1 µg/ml, 90th percentile) did not affect the association between IL-8 and venous thrombosis, although 51 (13%) patients with elevated IL-8 concentrations were present, compared with 37 (9%) of the controls [odds ratio 1·6 (95%CI 1·0–2·5)].
IL-8 concentrations were higher among oral anticoagulant (OAC) users than among non-users. The reasons for continuing OAC treatment were as follows: 15 of these 48 patients had a recurrence, six had a history of myocardial infarction (MI) and seven were diagnosed with PC/PS/AT deficiency. Thus 28 out of 48 individuals had a clear medical reason for continuing OAC treatment. These conditions (recurrence, MI, second thrombophilic defect) were present in only 19 of the other 399 patients.
Twelve out of 46 (26%) patients still using OAC had elevated IL-8 concentrations in contrast to 65 (15%) of the 420 patients without OAC. Exclusion of OAC users resulted in 6 (16%) patients with elevated IL-8 concentrations (> 90th percentile) versus 45 (10%) controls [1·7 (95%CI 1·1–2·5)]. Using the cut-off of 7·1 pg/ml (90th percentile) the adjusted odds ratio was 2·9 (95%CI 21·8–4·9), and using the cut-off of 16·2 pg/ml (99th percentile) the adjusted odds ratio was 6·0 (95%CI 2·0–17·7).
Separate analysis showed similar odds ratios for men and women with elevated IL-8 concentrations [for men 1·5 (95%CI 0·8–2·6) and for women 2·2 (95%CI 1·3–3·9)]. Stratification by age showed that the effect of IL-8 was limited to middle-aged subjects [age 30–51 years, odds ratio 3·4 (95%CI 1·6–7·2)], while in younger [age < 30, odds ratio 2·9 (95%CI 0·6–14·7)] or elderly subjects [age > 51 years, odds ratio 1·3 (95%CI 0·8–2·2)] an increased risk of venous thrombosis was less obvious.
- Top of page
- Patients and methods
Elevated concentrations of IL-8 are associated with venous thrombosis. Plasma concentrations of IL-8 (above the 90th percentile) lead to a 1·9-fold (adjusted for age and sex) increased risk of venous thrombosis. The association between venous thrombosis and IL-8 is most pronounced between the age of 40 and 51 years, and is not affected by levels of FVIII:C or homocysteine.
Several mechanisms may explain the association between elevated IL-8 concentrations and venous thrombosis. First, IL-8 has been shown to induce tissue factor, an important inducer of blood coagulation, on monocytes (Neumann et al, 1997). Second, leucocyte recruitment is induced by IL-8 (Luster, 1998). Leucocytes are the first cells to adhere to venous endothelium in a model of stasis-induced deep venous thrombosis and these leucocytes then stimulate thrombus formation (Schaub et al, 1984). Furthermore,IL-8 could contribute to the induction of a procoagulant surface by triggering the adhesion of monocytes to the endothelium (Gerszten et al, 1999).
The main stimuli for IL-8 production are proinflammatory cytokines such as IL-1 and TNF-α, bacterial products and viral infection (Baggiolini et al, 1994). It can only be speculated which stimuli are responsible for the increased IL-8 concentrations associated with venous thrombosis. The potential role of infectious agents in the pathogenesis of vascular disease has been studied in relation to atherosclerosis. These studies have suggested that infections with, for example, Chlamydia pneumoniae, Helicobacter pylori and cytomegalovirus are associated with atherosclerosis (Ross, 1999). These infectious agents may induce the production of IL-8 in the endothelium and monocytes (Kragsbjerg et al, 1995; Bliss et al, 1998; Murayama et al, 1998). Therefore it is worthwhile to explore the potential role of these infections in the induction of IL-8 in relation to venous thrombosis.
Studies in baboons have shown elevations in IL-8 levels after induction of a thrombus, suggesting that the thrombotic event could contribute to an increased IL-8 concentration (Wakefield et al, 1993). Nevertheless, it is unlikely that the thrombosis itself was responsible for the increased IL-8 concentrations, as the subjects were included at least 6 months after the thrombotic event. IL-8 concentration was not related to the time interval between thrombotic event and measurement of IL-8, and no decrease in IL-8 level was detected in patients who were included more than 28 months after their thrombotic event compared with the patients included within 12 months. Although plasma levels of CRP, a marker for inflammation, are high in patients, adjustment for CRP did not affect the thrombosis risk of elevated acute phase proteins FVIII:C and fibrinogen (Kamphuisen et al, 1999) or IL-8, suggesting that these associations are not owing to a post-thrombotic acute phase response. However, we cannot exclude the possibility that post-thrombotic syndrome or underlying malignancies contributed to the elevated IL-8 concentration in some patients, which may in part or completely be responsible for the observed association.
Another possibility is that the increased concentrations of IL-8 reflect occult cancer because many cancers produce IL-8. It is also well-known that patients with idiopathic venous thrombosis have a higher frequency of underlying cancer (Monreal et al, 1997). In fact, venous thrombosis is often the first sign of malignancy in most patients with prostatic and pancreatic carcinoma. The design of the LETS study excludes patients with known cancer, and overt cancer does not therefore explain the results. The LETS protocol, however, did not include an exhaustive search for occult cancer, and non-symptomatic cancer may therefore explain all or part of the increased IL-8 concentrations. Onlylarge prospective studies focusing on cancer during follow-up will provide a definite answer.
In conclusion, the association between elevated IL-8 concentrations and a first thrombotic event provides evidence for the notion that IL-8 increases the risk of venous thrombosis.