- Top of page
- Disclosure of Conflict of Interests
See also Lowe GDO. Epidemiology of venous thromboembolism: the need for large (including prospective) studies and meta-analyses. This issue, pp 2186–8 and Rosendaal FR. Etiology of venous thrombosis: the need for small original studies. This issue, pp 2189–90.
The non-O blood type is a recognized, although often unattended, constitutional risk factor for venous thromboembolism (VTE). A meta-analysis reported that persons with the non-O blood type had an odds ratio for VTE of 1.79 (95% confidence interval [CI] 1.56–2.05) over persons with a blood type O . Evidence also suggests that combined exposure to the non-O blood type and the factor (F)V Leiden mutation have a positive interaction on the risk of thrombosis . However, interactions between blood type and lifestyle have previously not been studied in detail. This study investigates whether blood types modify the thrombotic risks associated with cigarette smoking, a high body mass index (BMI) and the FV Leiden mutation.
The study population was recruited from the Danish prospective study ‘Diet, Cancer and Health’. From December 1993 to May 1997, 80 996 men and 79 729 women, between the age of 50–64 years and with no previous diagnosis of cancer, were invited to participate in the study. Information on smoking habits, BMI measurements and blood samples were obtained at baseline. Participants with an episode of VTE before enrollment were excluded and incident VTE episodes were verified by review of medical records . Information on diagnosis of any cancer disease and myocardial infarction among participants during follow-up was collected from the Danish National Patient Registry. A RT-PCR method allowing identification of four common alleles (O1, O2, A, and B) was used to determine the blood type (see Appendix for details). A case–cohort design including incident VTE cases and a randomly selected subcohort of 1841 participants (including 23 VTE cases) was applied. Incidence rates (IR) were computed as if the full cohort was included, then modified with a weighting scheme with a robust variance estimate. Risk estimates were assessed using a Cox regression analysis with age for the time axis. The study was approved by the regional ethics committees in Copenhagen and Aarhus, and by The Danish Data Protection Agency (H-Kf-01-345/93).
A total of 56 014 participants were included in the study and 641 incident VTE episodes occurred during a median follow-up of 10.2 years. ABO genotyping was possible in 1733 subcohort members and 578 cases. Among subcohort members, 42% had blood type O, whereas the remaining were distributed as follows: A (42%), B (12%) and AB (4%). Participants with blood type A, B and AB had hazard ratios (HRs) for VTE of 1.94 (95% CI 1.56–2.42), 1.42 (95% CI 1.03–1.97) and 2.15 (95% CI 1.33–3.45), respectively, using blood type O as reference. The HR for VTE for non-O blood groups as a single group was 1.84 (95% CI 1.50–2.27). The effects of the non-O blood type on the thrombotic risks associated with a high BMI, smoking and the FV Leiden mutation are shown in Table 1. The adjusted HR (aHR) for VTE from the combined exposure to the non-O blood type and obesity was 3.22 (95% CI 2.14–4.86), but no consistent effect modification was observed. The aHR for VTE emerging from the combined exposure to the non-O blood type and heavy smoking was 2.98 (95% CI 1.89–4.69), which exceeded the sum of separate effects and indicated a positive interaction on an additive scale. The positive interaction between heavy smoking and the non-O blood type was responsible for an additional 30.5 VTE cases per 100 000 person years. The presence of the FV Leiden mutation conferred a HR for VTE of 2.84 (95% CI 2.15–3.76), but when present in persons with non-O blood type the aHR for VTE became 5.12 (95% CI 3.05–8.59) exceeding the sum of separate effects. The positive interaction between the non-O blood type and the FV Leiden mutation caused an additional 64 VTE cases per 100.000 person years. The Rothman  Synergy Index was 0.99 (95%CI 0.39–1.60) for severe obesity, 1.61 (95% CI 0.10–3.13) for heavy smoking and 1.39 (95% CI 0.14–2.64) for FV Leiden in combination with the non-O blood type.
Table 1. Number of venous thromboembolism (VTE) cases, incidence rates per 100.000 person years (IR), crude and adjusted hazard ratios (HR)
| ||Blood Type O||Non-O Blood Type|
|Body mass index (BMI)|
| BMI < 25 kg m−2|
| Number of cases (n)||53||138|
| IR||55.7 [41.9–75.3]||96.1 [79.6–116.5]|
| Crude HR||1.0 [reference]||1.72 [1.21–2.44]|
| Adjusted HR*||1.0 [reference]||1.63 [1.14–2.32]|
| 25 kg m−2≤ BMI < 30 kg m−2|
| Number of cases (n)||60||190|
| IR||61.7 [47.1–81.9]||150.3 [126.8–178.6]|
| Crude HR||1.01 [0.67–1.51]||2.63 [1.87–3.71]|
| Adjusted HR*||0.92 [0.61–1.39]||2.40 [1.69–3.41]|
| BMI ≥ 30 kg m−2|
| Number of cases (n)||50||84|
| IR||162.1 [116.0–228.7]||209.1 [159.4–275.3]|
| Crude HR||2.87 [1.82–4.52]||3.46 [2.30–5.19]|
| Adjusted HR*||2.61 [1.65–4.13]||3.22 [2.14–4.86]|
| None smoker|
| Number of cases (n)||96||245|
| IR||67.6 [54.4–84.9]||126.7 [109.5–147.0]|
| Crude HR||1.0 [reference]||1.88 [1.44–2.47]|
| Adjusted HR*||1.0 [reference]||1.85 [1.40–2.44]|
| Moderate smokers (1–24.9 g day−1)|
| Number of cases (n)||53||123|
| IR||76.1 [56.8–103.6]||129.7 [105.5–160.3]|
| Crude HR||1.09 [0.75–1.59]||1.95 [1.43–2.65]|
| Adjusted HR*||1.13 [0.77–1.66]||2.03 [1.48–2.78]|
| Heavy smokers (≥ 25 g day−1)|
| Number of cases (n)||16||44|
| IR||109.0 [62.9–197.4]||198.6 [137.7–289.2]|
| Crude HR||1.66 [0.90–3.05]||3.12 [2.01–4.85]|
| Adjusted HR*||1.38 [0.74–2.57]||2.98 [1.89–4.69]|
|FVL mutation status|
| Wild type|
| Number of cases (n)||134||332|
| IR||63.5 [52.9–76.8]||116.4 [102.8–132.0]|
| Crude HR||1.0 [reference]||1.86 [1.48–2.33]|
| Adjusted HR*||1.0 [reference]||1.85 [1.37–2.50]|
| Factor V Leiden (G1691A)|
| Number of cases (n)||31||81|
| IR||200.7 [129.1–316.2]||317.6 [233.2–432.4]|
| Crude HR||3.12 [1.90–5.10]||4.91 [3.39–7.09]|
| Adjusted HR*||3.03 [1.55–5.94]||5.12 [3.05–8.59]|
The mechanisms by which blood type modifies VTE risk are not fully elucidated, but may be related to differences in von Willebrand factor (VWF) and Factor (F)VIII levels. Persons with a blood type O have lower concentrations of VWF, which is probably caused by binding of unmodified H-antigen to circulating VWF resulting in an accelerated hepatic clearance . VWF protects FVIII from instantaneous degradation and lower level of FVIII in persons with blood type O at least in part explain their lower VTE risk [6,7]. Carrying the less frequent genotypes of A2A2 or A2O (phenotype A) is associated with H-antigen secretion in levels similar to that of blood type O. Consistently, these persons have equally reduced levels of FVIII and VWF and a low VTE risk [1,5]. We did not distinguish between the A1 and the less frequent A2 alleles in the genotyping process, which could have underestimated the association between blood type A1 and VTE.
Obesity is a risk factor for VTE . In agreement with two previous studies, we found no consistent interaction between BMI and blood type on VTE risk [6,9]. We observed a positive interaction between the non-O blood type and heavy smoking with the joint effects being responsible for an additional 30.5 VTE cases episodes per 100.000 person-years. This is consistent with the results obtained in a retrospective study of 499 incident cases of pulmonary embolism showing a positive interaction between the non-O blood and past/current smoking . Smoking is an independent risk factor for VTE and the systemic effects of smoking are widespread causing changes to both hemostatic and inflammatory plasma proteins [3,10]. The positive interaction between the non-O blood type and heavy smoking could be mediated through multiple procoagulant changes. As expected, we observed a positive interaction between the non-O blood type and FV Leiden mutation in relation to VTE risk, but not beyond multiplicativity as reported in a previous study .
We analyzed data among participants who did and did not develop acute coronary syndrome or cancer before the diagnosis of VTE. In all subgroups, the association between blood type and VTE was observed. Thus, the association between blood type and VTE could not be explained by acute coronary syndrome or cancer being potential intermediate variables.
The Danish National Patients Registry captures nearly complete information on nationwide admissions and allows accurate VTE identification, although some deaths from VTE could have been missed because of a low frequency of autopsy in Denmark. The reliability of lifestyle data over time is controversial, but the validity of our lifestyle data was corroborated by the fact that 80% of smokers maintained the same doses of tobacco at 5 years follow-up . Furthermore, changes in lifestyle would bias our results only if they were systematically associated with blood type. Persons with a first episode occurring before the age of 50 years (youngest age for inclusion) were not represented in the study population and therefore our results may not generalize to a younger population. Although the observed interaction between heavy smoking and the non-O blood type calls for further validation, it emphasizes the potential hazards of an unhealthy lifestyle when combined with a genetic predisposition to VTE.