Venous thromboembolism and ethnicity

Authors


Dr Lara N. Roberts, King’s Thrombosis Centre, King’s College Hospital, Denmark Hill, London SE5 9RS, UK. E-mail: lara.roberts@kch.nhs.uk

Summary

Venous thromboembolism (VTE) has long been considered a disease that affects predominantly white populations, a misconception resulting from a paucity of epidemiological data from non-Western countries, and the low incidence of hereditary thrombophilia in those of non-Caucasian background. Over the last decade, interest has grown in this area with the emergence of evidence that VTE is as prevalent, if not more so, in the black population and is also common in Asian groups. Much is still to be learned, as our current knowledge of hereditary thrombophilia and acquired risk factors do not fully explain the risk of VTE in non-Caucasian groups. This review summarises the current understanding of ethnic variation in VTE and highlights the need for further research in this area.

The first reported case of venous thrombosis occurred in a young man from Normandy in the 13th century (Dexter & Folch-Pi, 1974) (Fig 1). In a historical review of venous thromboembolism (VTE), the absence of reports compatible with venous thrombosis from antiquity worldwide was noted as unexpected given the prevalence of the disease today (Mannucci, 2001). Recently, VTE, encompassing deep vein thrombosis (DVT) and pulmonary embolism (PE), has gained recognition as a major public health issue (House of Commons Health Committee Report, 2005).

Figure 1.

 An extract from the illustrated representation of the first described case of venous thrombosis, affecting a young European male (Clichè Bibliothèque Nationale de France, Paris). It is now accepted that VTE has a worldwide distribution. Adapted from British Journal of Haematology, 114, Mannucci, P., Historical review: venous thrombosis and anticoagulant therapy, 258–70, Copyright (2001) with permission from Wiley-Blackwell.

Inherited prothrombotic mutations {e.g. Factor V Leiden [F5 R506Q] and prothrombin gene mutation [F2 G20210A]} are known to increase the risk of VTE. F5 R506Q is highly prevalent in Northern Europeans and results in up to a seven-fold increased risk of VTE (Rees et al, 1995). The low prevalence of hereditary thrombophilia in other ethnic groups contributed to the mistaken perception that VTE was uncommon in the non-white population (Rees et al, 1995, 1999; Ridker et al, 1997). Deficiencies of the natural anticoagulants have since been established as the predominant thrombophilia in South East Asia (Shen et al, 1997, 2000; Peng et al, 1998; Suihesa et al, 2001; Sakata et al, 2004a; Miyata et al, 2006). Currently recognised hereditary thrombophilia are rare in the black population, although elevated Factor VIII is an established independent risk factor for VTE (Patel et al, 2003).

Early epidemiological studies in predominantly white populations further compounded the misconception regarding the low prevalence of VTE in non-Caucasian populations (Anderson et al, 1991; Silverstein et al, 1998), as did early autopsy studies in Japanese and Ugandan populations (Thomas et al, 1960; Gore et al, 1964). Subsequent work in ethnically diverse populations has demonstrated that VTE is a significant problem in other ethnic groups including African Americans and the UK black population. Debate persists as to whether ethnicity influences the natural history of VTE with some evidence of higher rates of PE (Stein et al, 1999; White et al, 2005; Schneider et al, 2006) and also of increased early morbidity and mortality in African-Americans (Siddique et al, 1996; Stein et al, 2003, 2004a). This does not appear to reflect differences in access or quality of healthcare (Stein et al, 2003; Aujesky et al, 2007).

Our understanding of variation in outcomes following VTE includes differences in the response to pharmacological treatment. Ethnic variation in the prevalence of polymorphisms in the genes encoding Vitamin K epoxide reductase complex, subunit 1 (VKORC1) and cytochrome P450, family 2, subfamily C, polypeptide 9 (CYP2C9) is recognised (Goldstein et al, 2003; Takahashi et al, 2006). As these enzymes influence the efficacy and metabolism of warfarin respectively, it has been suggested that these polymorphisms may account for the observed variation in warfarin dose between ethnic groups (Dang et al, 2005; Schelleman et al, 2008). It is likely that these polymorphisms also contribute to intra-ethnic variation (Takahashi et al, 2006). The use of VKORC1 and CYP2C9 genotype within a pharmacogenetic algorithm for warfarin dosing has been shown to eliminate inter-ethnic variation in dose requirements of maintenance warfarin, with persistent intra-ethnic variation relating to specific genotypes (The International Warfarin Pharmacogenomics Consortium, 2009).

Previous reviews have examined ethnicity as a prothrombotic risk factor (Itakura, 2005; Keenan & White, 2007), with a single, global review of VTE and ethnicity (Patel & Arya, 2008). They emphasise the need for further prospective research to explore underlying causative factors and differences in clinical outcomes in non-white ethnic groups following VTE. For the purpose of this review, Medline was searched using the search criteria venous thrombosis, venous thromboembolism, deep vein thrombosis, pulmonary embolism, thrombophilia, Factor V Leiden, prothrombin gene mutation, protein C, protein S, antithrombin, warfarin, heparin and anticoagulation, in combination with race, and ethnicity to identify publications of interest up to December 2008. Further studies of interest were identified from the text of these publications and the authors’ own collections. In this review, we discuss the debate regarding the classification of ethnicity, the current understanding of epidemiology, thrombophilia, biomarkers of VTE and treatment response within different ethnic groups and highlight the need for further research in this area.

Defining ethnicity

Controversy surrounds the use of race and/or ethnicity to sub-classify individuals into groups for research purposes and classification remains un-standardised (Sanker & Cho, 2002). Conventionally, race may be considered to reflect the primary continent of origin, whilst ethnicity is regarded as a broader concept incorporating tradition in addition to geographic, social, cultural and religious influences (Risch et al, 2002; Burchard et al, 2003); however the terms are used inter-changeably by some. The United States Census, in 2000, used five broad racial categories of black or African American, white, Asian, native Hawaiian or other Pacific Islander, and American Indian or Alaskan (Keenan & White, 2007). In the UK, alternate categories of white, black or black British, Asian or Asian British, Mixed, and Chinese or other ethnic group are employed (Patel & Arya, 2008).

A more precise classification tool utilising human genomic sequence variation has been proposed as an alternative to race or ethnicity (Burchard et al, 2003; Rotimi, 2004). A relationship between genetic clustering and racial categorisation has been demonstrated with the major difference between the biological categorisation and self-reported race in categories used by the US Census being that South, Central and West Asians cluster with Europeans and are separate from East Asians (Burchard et al, 2003). However, using genetic clustering alone assumes variation in disease to be genetic and unrelated to environmental factors that may be influenced by race or ethnicity (Risch et al, 2002; Burchard et al, 2003).

For the purpose of this review, ethnicity will be used broadly to describe those from different geographic or ancestral backgrounds. The terminology utilised in original publications to describe study groups will be reproduced.

Epidemiology of VTE

Caucasian populations

The epidemiology of VTE in Caucasian populations has been well established with an incidence of between 80 and 117 cases per 100 000 per year (Gillum, 1987; Anderson et al, 1991; Bournemaux et al, 1996; Silverstein et al, 1998) and a dramatic rise in annual rates with increasing age up to 500 events per 100 000 in those over 80 years (Anderson et al, 1991). In those with VTE, identifiable predisposing factors are present in the majority, with idiopathic events accounting for approximately 26% (Heit et al, 2002). Common risk factors relating to VTE include recent surgery or trauma, hospitalisation with a medical illness, paralysis, cardiac failure and cancer (Heit et al, 2002). Additional risk factors in women include pregnancy (Heit et al, 2005) and the use of hormone replacement therapy or oral contraception (Rosendaal et al, 2003). One might argue that VTE aetiology is not yet fully characterised in Caucasians, given the large proportion experiencing an unprovoked and as yet, unexplained event.

Early reports suggesting a low incidence in non-Caucasian populations

Early studies reported low rates of VTE in Asian and African populations compared to the incidence in Caucasians. Post-operative DVT rates were found to be low (12%) in both Malaysian (Cunningham & Yong, 1974) and Sudanese patients (Hassan et al, 1973) using radio-isotope scanning. A low incidence of DVT was found in a black Caribbean population (11–17/100 000), however the authors acknowledged a significant associated risk of death at 6·2%, which was comparable with reported rates in Caucasian populations (Nossent & Egelie, 1993). Significantly lower rates of PE were documented at autopsy in Uganda compared to an age and sex matched cohort in St Louis (USA) (Thomas et al, 1960). A further autopsy-based study in Japan found a significantly lower frequency of PE compared to matched cases in Boston (USA) (Gore et al, 1964). In contrast, a small study (100 patients) undertaken in South Africa investigating post-operative DVT utilising both radio-isotope and Doppler ultrasound imaging in patients undergoing major surgery found high rates of asymptomatic DVT of up to 50% in non-Europeans, equal to the rate detected in Europeans (Joffe, 1974). This provided initial evidence to challenge the perception that VTE was rare in non-Caucasians and suggested that low rates reported in other ethnic groups were possibly due to under-diagnosis and retrospective study design. Further evidence has since emerged to support these findings.

Ethnic groups in America

Epidemiological studies incorporating non-white populations have increased over the last decade, although many of the studies have been undertaken in the Western world in ethnically diverse communities. In studies from the USA, data was predominantly retrospectively collected from Medicare claims or discharge data, in combination with data from the US Census for population estimates and mortality data, both of which include race/ethnic group as a collected variable (White et al, 1998, 2004, 2005; Stein et al, 2004b; Schneider et al, 2006). These studies are summarised in Table I.

Table I.   Population studies of VTE across different ethnic groups in the USA.
 Study populationWhite/ CaucasianAfrican AmericanHispanicAsian/Pacific Islander
  1. VTE, venous thromboembolism; DVT, deep vein thrombosis; PE, pulmonary embolism; 95% CI: 95% confidence interval; –, group not examined in study.

Stein et al (2004b): number of subjects2 490 0002 641 000440 00021 000
 All VTE (age adjusted rate/100 000  per year) 13013826, < 0·0005
 DVT (age adjusted rate/100 000 per year) 10410722, < 0·0005
 PE (age adjusted rate/100 000 per year)612 00036407, < 0·0005
White et al (2005): number of subjects21 00216 01517942216612
 Adjusted standardised incidence of all  VTE/100 000 104141, < 0·000155, < 0·000121, < 0·0001
 Idiopathic VTE: number of events54184209426588116
 Incidence of idiopathic VTE/100 000  (95% CI) 28 (27–29)32 (29–35), P = 0·0415 (13–16), P < 0·00016 (5–7), P < 0·0001
Schneider et al (2006)10 542    
 PE Male (age adjusted rate/100 000) 36·4753·64, < 0·001
 PE Female (age adjusted rate/100 000) 37·9761·53, P < 0·001

Estimated population incidence of VTE was calculated from the above data allowing comparison across ethnic groups. The estimated annual incidence of VTE in Hispanic, Asian/Pacific Islanders and American Indian/native Alaskans was reported as 55, 21 to 26 and 71 per 100 000 respectively (Stein et al, 2004b,c; White et al, 2005). These rates were significantly lower than those reported for Caucasians and African-Americans (Stein et al, 2004b,c; White et al, 2005). Interestingly, idiopathic events were significantly more common in females of Asian/Pacific Islander background compared to Caucasians with females, accounting for 64·1% and 55·5% of idiopathic events respectively (White et al, 2005). White et al (2005) reporteded a significantly higher incidence of VTE in African-Americans compared to Caucasians, Hispanics and Pacific Islanders (see Table I). Other studies reported an annual incidence for VTE in African Americans of 138–155 per 100 000 (Stein et al, 2004b,c), which is comparable to the incidence in Caucasians. The incidence of PE is reported to be higher in African-Americans (Stein et al, 1999; White et al, 2005; Schneider et al, 2006). There is a lack of consistency in some findings from these studies, which may be partly attributed to the lack of age adjustment across ethnic groups (White et al, 1998, 2004, 2005; Stein et al, 2004c).

The effect of ethnicity as a risk factor for outpatient VTE, postoperative VTE and idiopathic VTE has also been investigated (Klatsky et al, 2000; White et al, 2004, 2005). These studies demonstrated a significantly higher risk of VTE in those of African-American race, with Hispanic and Asian/Pacific Islander race conferring a significantly lower risk (see Table II).

Table II.   Ethnicity as a risk factor for VTE in the USA.
 Study designStudy population White/CaucasianAfrican AmericanHispanic Asian/Pacific Islander
  1. VTE, venous thromboembolism; DVT, deep vein thrombosis; PE, pulmonary embolism; 95% CI, 95% confidence interval; –, group not examined in study.

  2. *Patient number estimated based on investigation of VTE following 40 procedures which had been undertaken in at least 4500 patients each.

Klatsky et al (2000): number of subjectsProspective128 93472 01934 661577713 593
VTE age-adjusted relative risk (95% CI) cohort Reference group1·4 (1·1–1·8), P < 0·050·8 (0·4–1·6)0·1 (0·04–0·4), < 0·005
White et al (2004): number of subjectsRetrospectivec. 180 000*    
 Odds ratio for postoperative DVT (95% CI) cohort Reference group1·2 (1·1–1·3), P < 0·00010·9 (0·8–0·9), P = 0·00030·4 (0·4–0·5), P < 0·0001
White et al (2005)Retrospective     
 Odds ratio for idiopathic PE vs. DVT (95% CI) cohort5418Reference group1·2 (1·1–1·3), P = 0·00010·8 (0·7–0·9), P < 0·0001

Whilst the above studies found no difference in short term mortality between ethnic groups, evidence from other studies is conflicting, suggesting a higher short-term mortality in African-Americans following a PE (Siddique et al, 1996; Stein et al, 2003, 2004a) (Table III). These studies raise a number of confounding factors, such as access to treatment, time to presentation, mode of treatment and various co-morbidities, which may all contribute to this observed discrepancy. However, there is no evidence to suggest that diagnostic procedures or difficulty in establishing a diagnosis of VTE differ between black and white patients (Stein et al, 2003). Aujesky et al (2007) demonstrated African-American race was associated with increased risk of adverse outcome (defined as death, bleeding or recurrent VTE at 90 d) compared to white subjects and that this was not related to quality of care whilst an inpatient (odds ratio 5·2, 95% confidence interval 1·3–21·6). It is possible that differences in compliance and monitoring of anticoagulation following discharge are responsible for this higher complication rate but further studies are required to confirm these findings and define underlying contributing factors.

Table III.   Variation in short term mortality by ethnicity in studies from USA.
 Study populationWhite/ CaucasianAfrican AmericanHispanicAsian/Pacific Islander
  1. VTE, venous thromboembolism; DVT, deep vein thrombosis; PE, pulmonary embolism; –, group not examined in study.

  2. *In comparison with Caucasian group.

Siddique et al (1996) 30-d mortality>65 years    
 PE primary discharge diagnosis192 90512·9%16·1%, P < 0·001
 PE secondary discharge diagnosis199 08630·2%34·7%, P < 0·001
Stein et al (2004a) 66 87914 033
Age-adjusted mortality/100 000 per year – PE 3·46·9, P < 0·0005
White et al (2005)
 28-d fatality rate all idiopathic VTE54181·8%3·8%, P = 0·0007*2·3%2·6%
 28-d fatality rate idiopathic PEc. 16363·6%8·0%4·1%6·1%
Schneider et al (2006)10 542    
 Age-adjusted fatality rate/million – male PE 35·265·6, P < 0·001
 Age-adjusted fatality rate/million – female PE 30·366·5, P < 0·001

Rates of VTE in black and white hospitalised patients with cancer have been reported as similar (Stein et al, 2006). These findings were essentially confirmed by a further study in California, investigating VTE in cancer patients (Chew et al, 2006). In this study, significantly higher rates of VTE were seen in African Americans with uterine cancer and lower rates were seen in those with non-Hodgkin lymphoma and lung cancer. This group also found significantly lower rates of VTE in Asian and Pacific Islanders with stomach, breast, lung, colorectal cancer, prostate cancer, and lymphoma and pancreatic cancer compared to black and white patients with the same diagnosis.

One of the main limitations of the studies discussed above is their retrospective nature and reliance on discharge data, which may have resulted in coding errors and misclassification. The majority of incidence rates for VTE are calculated estimates based on population numbers derived from Census data, which may also give rise to errors. Whilst these studies were undertaken in ethnically diverse communities, Caucasians remained the majority with smaller proportions of other ethnic groups. For example, a number of studies were undertaken in California (White et al, 1998, 2004, 2005; Klatsky et al, 2000; Chew et al, 2006) where the adult population consists of 55% Caucasians, 6% African-Americans, 26% Hispanics and 10% Asian/Pacific Islanders (Keenan & White, 2007). Additionally, environmental factors due to the adoption of Western lifestyle may play a role in the development of VTE, which is not accounted for in these studies.

A prospective case–control study is ongoing in Atlanta examining racial differences in the epidemiology of VTE between Caucasians and African Americans (Dowling et al, 2003). They found the average age of African Americans presenting with VTE to be significantly lower compared to Caucasians (47·5 years compared to 50·7 years). Family history of VTE was frequently positive in both groups (28% in African Americans and 29% in Caucasians), although identifiable genetic thrombophilia was far more common in Caucasians. This suggests there may be an as yet undiscovered genetic predisposition to VTE in African Americans.

Asia

Studies undertaken in Asian countries have demonstrated higher rates of VTE compared to that found within mixed population studies such as those arising from the USA. Chau et al (1997) reported PE rates from hospital post-mortem findings over a 5-year period and documented a rate of 4·7%, which is similar to rates in Caucasian populations at autopsy. A small number of studies undertaken in Asians document a lower incidence of VTE in comparison with Caucasians (Cunningham & Yong, 1974; Atichartakarn et al, 1988; Cheuk et al, 2004; Jain et al, 2004). However, inpatient mortality remained significant, with rates of 7·3% and 23·8% reported for DVT and PE respectively (Cheuk et al, 2004). These reports of low incidence of post-operative VTE may have arisen at least in part due to under-diagnosis and have subsequently led to the infrequent use of surgical thromboprophylaxis.

A comprehensive review and analysis of published data on post-operative VTE in Asians demonstrated that VTE is not rare (Leizorovicz et al, 2004). They determined rates of objectively documented VTE (both symptomatic and asymptomatic) in patients not receiving thromboprophylaxis from studies undertaken in Hong Kong, Korea, Japan, Malaysia, Thailand and Singapore. The overall adjusted rate of VTE following general surgery was 13%, 16% after total hip and 50% after total knee replacement, and 18% following hip fracture surgery (Leizorovicz et al, 2004). Further prospective studies in post-operative patients have since demonstrated VTE rates (both symptomatic and asymptomatic) in Asians to be significant and comparable with that seen in Caucasian populations (see Table IV) (Geerts et al, 2004; Leizorovicz et al, 2005;Piovella et al, 2005; Sakon et al, 2006). Low molecular weight heparin in Asians is effective in the prevention of VTE in patients undergoing major orthopaedic surgery (Yoo et al, 1997; Fong et al, 2000). The above studies suggest the site of venous thromboembolism differs in Asians with less proximal DVT and symptomatic PE reported (Chan et al, 2001; Piovella et al, 2005; Sakon et al, 2006). Aetiological and pathogenic factors to account for this observation are yet to be elucidated.

Table IV.   VTE rates in recent prospective studies of Asian surgical patients compared with studies in Western populations without thromboprophylaxis.
 Country of studyNumber of patientsAsymptomaticSymptomatic events
All VTEProximal DVTDVTPE
  1. VTE, venous thromboembolism; DVT, deep vein thrombosis; PE, pulmonary embolism; THR, total hip replacement, TKR, total knee replacement, HFS, hip fracture surgery; –, not examined/reported.

  2. *DVT detected by mandatory venography in controls receiving placebo or no thromboprophylaxis in prospective trials from 1980 to 2002 (Geerts et al, 2004).

  3. †VTE rates until time of discharge.

  4. ‡Asymptomatic DVT detected by objective diagnostic screening in patients not on prophylaxis.

Major orthopaedic surgery
 Geerts et al (2004)*Europe, USA41–85%5–36%
  THR  42–57%18–36%
  TKR  41–85%5–22%
  HFS  45–60%23–30%
 Leizorovicz et al (2005)Bangladesh, China, India,2420222 (0·9%)6 (0·2%)
  THRIndonesia, Malaysia, Pakistan,4084 (1%)0
  TKRPhillipines, Singapore, Taiwan,94413 (1·4%)3 (0·3%)
  HFSThailand10685 (0·5%)3 (0·3%)
 Piovella et al (2005)China, Indonesia, Malaysia,407121/295 (41%)30/295 (10·2%)2 (0·5%)
  THRPhillipines, South Korea,17525·6%5·8%
  TKRTaiwan, Thailand13658·1%17·1%
  HFS 9642%7·2%
Abdominal/pelvic surgery
 Sakon et al (2006)Japan17341 (23·7%)5 (2·9%)1 (0·6%)
 Geerts et al (2004)Europe, USA15–40%

Pregnancy-associated VTE rates are significant in the Chinese population, with 1·88 events occurring per 1000 deliveries (95% CI 1·18–2·52; Chan et al, 2001). The incidence of VTE in Chinese children has been reported as 0·74 per 100 000 (Lee et al, 2003). Both rates are comparable with reported rates in Caucasian populations (Andrew et al, 1994; van Ommen et al, 2001; Heit et al, 2005). DVT occurs commonly in Asian patients following ischaemic stroke with asymptomatic events detected in 45% by day 30 (De Silva et al, 2006). In contrast, rates of VTE in Asian patients with pancreatic cancer are reported to be lower at 5·6% (Oh et al, 2008) compared to up to 20% in Western populations (Chew et al, 2006). A review of patient characteristics from a Japanese PE registry study found a slight female preponderance (female:male ratio 1·26) and a high proportion of idiopathic events (58% of 307 patients) (Nakamura et al, 2001). Retrospective studies in Taiwan and Singapore, investigating inpatient diagnoses of DVT, found malignancy, immobility and known thrombophilia to be important risk factors (Peng et al, 1998; Tan et al, 2007). Leung et al (2006) undertook a prospective study in 1754 Chinese medical inpatients (not receiving thromboprophylaxis) investigating those who developed clinical symptoms of DVT or a rise in d-dimer during their medical admission with lower limb Doppler ultrasound. They document an overall incidence of at least 10 per 1000 in those admitted for more than 2 d. The rates in association with known risk factors, such as malignancy, stroke and cardiac failure were found to be higher at 6·3%, 4·1% and 3·8% respectively. These findings are similar to those documented in Western patients. Thus, this ethnic group should be considered for thromboprophylaxis during periods of risk.

Black population in the United Kingdom

The Camberwell Thrombophilia Study studied a prospective case–control cohort to identify the epidemiology and risk factors for VTE in the black population (Patel et al, 2003). VTE was found to be as prevalent in black subjects as in white, with 22·5% of patients presenting with VTE being black (consistent with the makeup of the local population in which 25% were of African or Caribbean origin). This study also noted a younger age of presentation in blacks compared to white subjects and a trend to more proximal events in blacks presenting with DVT compared to whites (Patel & Arya, 2008). It was also noted that obesity and pregnancy were more common risk factors in black subjects, with travel being a less frequent risk factor in comparison to white subjects.

Incidence of recurrent VTE

Ethnic origin has also been found to modulate risk of recurrent VTE. It is well established in Caucasian populations that the risk of recurrent VTE following an idiopathic event is higher in males (Heit et al, 2000; Baglin et al, 2004; Kyrle et al, 2004). Epidemiological studies undertaken in California, USA documented increased rates of recurrent idiopathic VTE in Hispanic women (5·5% compared to 3·6% in Caucasians; White et al, 2005) (see Table V). The authors acknowledged that the recurrence rates found in this study were lower than previously reported rates of 7% at 1 year (Hansson et al, 2000) and 17·5% at 2 years (Prandoni et al, 1996), and thus further investigation is required. This group proceeded to examine a larger retrospective group over a 5 year period and established that risk of recurrence following a first idiopathic VTE in African-American, Asian/Pacific Islander and Hispanic men was comparable to rates in Caucasian men (White et al, 2006) (Table V). They confirmed earlier findings that rates of recurrent VTE following a first idiopathic event are higher in Hispanic women following a DVT and further demonstrated an increased risk of recurrence following an idiopathic PE in African-American women, which was comparable with recurrence rates in men (see Table V).

Table V.   Recurrence rates of VTE by ethnicity in studies from USA.
 Study population White/CaucasianAfrican AmericanHispanicAsian/Pacific Islander
  1. VTE, venous thromboembolism; DVT, deep vein thrombosis; PE, pulmonary embolism; RR, relative risk; 95% CI, 95% confidence interval; –, group not examined in study.

White et al (1998)
 RR of recurrence at 3   months (95% CI)17 9911·01·03 (0·91–1·15)0·95 (0·84–1·06)0·65 (0·49–0·85)
White et al (2005)21 00216 01517942216612
 Incidence of recurrence     
 After any VTE at 6   months (% of VTE)3·4% of 21 0022·9%4·0%4·0%2·0%
 After idiopathic VTE  at 6 months  (% recurrent)3·4% of 54183·1%5·5%, P = 0·007
White et al (2006)
 Number of subjects11 514889010721552
 RR of recurrence after idiopathic DVT (95% CI)
   Male 1·7 (1·4–2·1), P < 0·00012·0 (1·3–2·8), P = 0·00052·0 (1·4–2·7), P < 0·0001
   Female 1·0 (reference)1·2, (0·8–1·9)1·7 (1·2–2·4), P = 0·001
RR of recurrence after  idiopathic PE  (95% CI)
   Male 1·6 (1·1–2·1), P = 0·0021·4, (0·8–2·5)2·0 (1·2–3·3), P = 0·001
   Female 1·0 (reference)2·1 (1·3–3·3) P = 0·0031·6 (1·0–2·7)

Hereditary thrombophilia

Hereditary thrombophilia are found in up to a third of unselected Caucasian patients with VTE (Murin et al, 1998; Bombeli et al, 2001), increasing to 50% in those with a family history of VTE (Murin et al, 1998). Whilst hereditary thrombophilia is an established risk factor for VTE, many carriers remain asymptomatic. Currently recognised hereditary thrombophilia resulting in an increased risk of VTE are F5 R506Q, F2 G20210A, and deficiencies of the natural anticoagulants: antithrombin (AT), protein C (PC) and protein S (PS). The prevalence of these anomalies varies and is most widely documented in Caucasian populations; the data is summarised in Table VI (healthy control population) and Table VII (subjects treated for VTE).

Table VI.   Prevalence of hereditary thrombophilia in the healthy population by ethnicity. Please refer to text for references.
 F5 R506QF2 G20210AAT deficiencyPS deficiencyPC deficiency
  1. –, Data not available.

  2. *Note reference range not specific to ethnic group.

Europeans8·8–15%1·7–3%0·02–0·15%0·03–0·13%0·2–0·4%
SE Asian000·15%1·12%0·13%
UK blacks002%*4%*
African Americans1·1–1·23%<0·001%
Table VII.   Prevalence of hereditary thrombophilia in unselected subjects with VTE by ethnicity. Please refer to text for references.
 F5 R506QG20210AAT deficiencyPS deficiencyPC deficiency
  1. –, Data not available.

  2. *Patients with unprovoked DVT.

  3. †Note reference range not specific to ethnic group.

Europeans20%6·2%1–3%1–5%3–5%
SE Asian005·6%18%*8%
UK blacks1·4%00·7%†2·8%†4·2%†
African Americans2·9%1·1%

Factor V Leiden (F5 R506Q)

F5 R506Q is the most commonly detected hereditary thrombophilia in Europeans with carrier rates of 8·8%, increasing up to 15% in Greeks (Rees et al, 1995). It is the most frequently detected hereditary thrombophilia within Caucasians with VTE and is present in up to 20% of these patients (Ridker et al, 1995; Rosendaal et al, 1995). It is also common in the Middle East with population rates of 14·2% in Lebanon (Irani-Hakime et al, 2000) and 12% in Jordan (Awidi et al, 1999). F5 R506Q is widespread in North America and Australia with a carrier frequency of almost 6% and 3% respectively (Ridker et al, 1997; Said et al, 2006). F5 R506Q is additionally seen in 12% of DVT patients from north India (Garewal et al, 2003). In contrast, F5 R506Q is exceedingly rare in African, South East Asian, Chinese and indigenous Australasian populations (Rees et al, 1995; Pepe et al, 1997).

The F5 R506Q mutation is thought to have arisen as a single mutation 21–34 000 years ago based on haplotype analysis (Zivelin et al, 1997), which post-dates the probable divergence of the modern Europe group from other human groups (estimated to be 40 000 years ago; Cavalli-Sforza et al, 1994). Lucotte and Mercier (2001) reviewed the reported allelic frequency of F5 R506Q from 21 geographic populations and propose the mutation arose in Anatolia (Turkey), and subsequently expanded toward Western Europe, in parallel with the movement of migrating Neolithic farmers (see Fig 2). The low prevalence of F5 R506Q in UK blacks with DVT (1·4%) (Patel et al, 2003) and in healthy non-Caucasian USA controls, including African Americans (up to 1·23%), Asian Americans (0·45%) and Native Americans (1·25%) (Ridker et al, 1997; Dowling et al, 2003) may have arisen via genetic admixture within ethnically diverse communities.

Figure 2.

 Prevalence of F5 R506Q in Europe and southeast and North Africa. Large arrows indicate Neolithic expansion of the mutation toward western and northern Europe from the Anatolian centre of origin; small arrows indicate the secondary expansion of the mutation toward southeast Mediterranean regions. Reprinted from Blood Cells, Molecules and Diseases, 27(2), Lucotte & Mercier, Population genetics of Factor V Leiden in Europe, 362–367, Copyright (2001) with permission from Elsevier.

F2 mutations

The F2 G20210A polymorphism is common within European countries with a prevalence of 1·7–3% amongst healthy individuals (Poort et al, 1996; Rosendaal et al, 1998). Its prevalence is increased in those with DVT at 6·2%, rising further to 18% in the presence of a positive family history (Poort et al, 1996). A recent study in Iran determined a population prevalence of 1·6% (Rahimi et al, 2008). It is rare in other ethnic groups and is absent from indigenous populations of Africa, America, Asia, Australasia and Middle Eastern Arabs (Franco et al, 1998; Rees et al, 1999). F2 G20210A remains undetected in the UK black population (Patel et al, 2003) and its prevalence is low in healthy African-Americans, at less than 1 in 1000 (Dilley et al, 1998; Dowling et al, 2003), with a small number of cases identified in African-Americans with VTE (Table VI; Dilley et al, 1998; Dowling et al, 2003).

A further mutation (C to T) in F2 at position 20209 has been described in a small number of African-Americans, two of whom had a prior venous thrombotic event (Warshawsky et al, 2002), in addition to a single black VTE patient in the UK. Soo et al (2005) identified this patient from 61 black subjects with DVT and no other detectable thrombophilia, with no cases identified amongst 183 controls (UK blacks, Asian Indians and South East Asians). This suggests a carrier frequency of 1·6% within the UK black population. A further study found this mutation in 0·5% of Caucasian women (of Jewish-Moroccan background) undergoing investigation for recurrent miscarriage (Danckwardt et al, 2006). They were further able to demonstrate that the mutation leads to upregulation of prothrombin expression. Further investigation of a larger population cohort is required to clarify the prevalence of the mutation in different ethnic groups and its contribution to risk of thrombosis.

Antithrombin, protein C and protein S deficiency

The prevalence of AT, PC and PS deficiencies is much lower in comparison with both F5 R506Q and F2 G20210A within European populations. Population studies have been undertaken in Caucasian populations, where the prevalence of AT, PC and PS deficiencies were reported to be 0·02–0·15% (Tait et al, 1994), 0·2–0·4% and 0·03–0·13% respectively (Franco & Reitsma, 2001) (Table VI). The prevalence in patients with VTE is higher at 1–3% for AT deficiency; 3–5% for PC deficiency and 1–5% for PS deficiency (Franco & Reitsma, 2001) (Table VII). Overall, these deficiencies are seen at a combined rate of 5–15% in patients presenting with VTE. Similar rates of these deficiencies are seen in DVT patients in Jordan (Eid, 2002).

There are few studies investigating AT, PC and PS deficiency in other ethnic groups. Patel et al (2003) measured the above during thrombophilia screening of 142 black patients following completion of treatment for VTE and found PC deficiency in 4·2%, PS deficiency in 2·8% and AT deficiency in 0·7%. Another UK group sought to establish normal ranges for these natural anticoagulants in healthy black subjects compared to stroke patients and found that PC and PS levels were significantly lower than the reference range derived from a white population, with a trend towards lower antithrombin levels (Jerrard-Dunne et al, 2003). Using the laboratory reference range, 4% of healthy black controls met the criteria for PC deficiency and a further 2% the criteria for PS deficiency, exceeding the proportion of thrombophilia detected in white controls. This suggests that blacks could be erroneously diagnosed with PC, PS or AT deficiency if ethnic-specific reference ranges are not employed. Clearly, further studies are required to confirm this finding but it would seem appropriate to suggest that reference ranges be derived from healthy controls of the same ethnic origin.

A large study in Japanese blood donors found the population prevalence of AT and PC deficiency to be similar to Western populations, with a prevalence of 0·15% and 0·13% respectively (Sakata et al, 2004a). PS deficiency was found to have a higher prevalence, affecting an estimated 1·12% of the healthy population (Sakata et al, 2004b). Japanese patients with a DVT have a higher prevalence compared with that reported for European DVT cohorts; PC deficiency has been detected in up to 8% (Suihesa et al, 2001; Sakata et al, 2004a; Miyata et al, 2006), PS deficiency in almost 18% with unprovoked DVT (Suihesa et al, 2001) and AT deficiency in up to 5·6% (Sakata et al, 2004a; Miyata et al, 2006). Normal ranges were derived from within the tested population and therefore inappropriate reference ranges do not account for the increased prevalence. Studies in Chinese patients have also found deficiencies of these natural anticoagulants to be more prevalent compared to the Western population with up to 50% of patients with a DVT found to have a deficiency of PC, PS or AT (Shen et al, 1997, 2000; Peng et al, 1998). Thus PC and PS deficiencies are the more important thrombophilia associated with VTE in South East Asian groups.

Biomarkers associated with VTE

Fibrinogen

Ethnic differences have been reported in fibrinogen levels, with significantly lower levels found in blacks, compared to whites and South Asian Indians residing in South London, UK (Cook et al, 2001). This difference increased when adjusted for potential confounders, such as age, smoking and obesity, with fibrinogen levels significantly lower by 0·11 g/l in black women and 0·21 g/l lower in black men. This contrasts with previous studies in which African-Americans were found to have significantly higher fibrinogen levels (by 0·1–0·6 g/l) compared to US whites (Folsom et al, 1991; Folsom et al, 1993). Given the discrepancies in these findings, it is difficult to attribute the higher risk of VTE (or cardiovascular disease) described in black populations to a possible difference in baseline fibrinogen level. A small study in African-Americans with VTE demonstrated a non-significant trend to higher fibrinogen levels (>4 g/l) compared with controls (Austin et al, 2000), although the authors were unable to demonstrate a link between common polymorphisms within FGB (β-fibrinogen gene) and risk of venous thrombosis. It remains unclear as to whether fibrinogen is a significant risk factor or marker for VTE risk.

d-dimer

d-dimer is a marker of fibrinolysis and commonly utilised screening test in the risk stratification of patients with a possible VTE. A normal result in combination with a low clinical probability of DVT essentially excludes the diagnosis (Kearon et al, 2001; Wells et al, 2003). It has recently been shown in European populations that persistent elevation of d-dimer following completion of anticoagulation for a first VTE identifies patients at an increased risk of recurrent event (Palareti et al, 2002; Eichinger et al, 2003). d-dimer levels are known to be higher in the black population. Pieper et al (2000) tested d-dimer in older patients and found blacks to have a 40% higher d-dimer compared to age-matched white subjects. A study in the UK black population found significantly higher d-dimer levels in those with DVT compared to healthy controls (Patel et al, 2004). Furthermore, d-dimer levels above the 90th centile of healthy controls were associated with a more than four-fold increased risk of DVT. Elevated d-dimer most likely reflects underlying activation of coagulation and fibrinolysis, rather than a direct causal link to VTE. Clustering of d-dimer, fibrinogen and factor VIII (FVIII), reported in healthy Europeans (Van Hylckama Vleig et al, 2003), was not demonstrable within the UK black population (Patel et al, 2004). Unless an ethnically appropriate reference range is determined, using d-dimer to stratify VTE recurrence risk could result in many black DVT patients being erroneously considered at high risk of recurrence and potentially treated with extended anticoagulation. It may also result in over-investigation of blacks referred with a low clinical probability of DVT.

Factor VIII

Elevated FVIII levels are recognised as a risk factor for VTE (Koster et al, 1995; Kraaijenhagen et al, 2000) and possible risk factor for recurrent VTE (Kyrle et al, 2000) in the Caucasian population. Elevated FVIII has also been described as a common, independent risk factor for VTE in the UK black population (Patel et al, 2003). Furthermore, a dose-dependent effect was established with a near five-fold increase in VTE risk with FVIII levels above the 90th centile (228 i/u per dl). Thus far, this is the only marker found to be strongly associated with VTE in the black population and is more important than conventional thrombophilia in this group. A genetic cause for elevated FVIII is plausible and may explain the frequency of a positive family history in those presenting with VTE, however this is yet to be elucidated.

Response to treatment

Ethnic variation in response to drug therapy has been observed for some time. A recent review of 42 genetic variants associated with altered pharmacological response to drug therapy demonstrated ethnic variation in two thirds of these variants, including CYP2C9, which plays a key role in warfarin metabolism (Goldstein et al, 2003). There is wide inter-patient variability in response to warfarin, to which ethnic origin contributes, with African-Americans requiring higher doses than Caucasians, and Asians requiring lower doses still (Dang et al, 2005). This difference may be in part accounted for by genetic variation.

The two genes associated with ethnic variation in response to warfarin are CYP2C9 and VKORC1. CYP2C9 is one of three enzymes involved in the metabolism of warfarin leading to its inactivation. Numerous variants of CYP2C9 have been described with reduced enzymatic activity compared to CYP2C9*1 resulting in increased sensitivity to warfarin, reduced dose requirement during induction and increased time to stable maintenance dosage (Peyvandi et al, 2004). The most common variants (*2, *3) have an allelic frequency of up to 25% in Caucasian, 5% in African-American and 1·6% in South East Asian populations (Table VIII; Goldstein et al, 2003; Takahashi et al, 2006). VKORC1 encodes the Vitamin K epoxide reductase, which is inhibited by warfarin leading to reduced γ-carboxylation (and thereby activity) of Vitamin-K dependent coagulation factors. VKORC1 variants are associated with warfarin resistance and are variably prevalent, with the most common variant (C1173T) present at rates of 8·6% in African-Americans, 42% in Caucasians and 89% in Japanese (Table VIII; Takahashi et al, 2006). The presence of intra-ethnic variation of genotypes for both of these enzymes limits the application of dose adjustment based on ethnicity.

Table VIII.   Allelic frequencies of common CYP2C9 and VKORC1 variants (Goldstein et al, 2003; Takahashi et al, 2006).
 CYP2C9*1 (wild type)CYP2C9*2CYP2C9*3VKORC1 C1173T
Caucasian74·3%11·3–14·3%8·4–10·9%42·2%
Asian98·4% (Japanese)01·6%89·1% (Japanese)
African-American95·3% 0–2·9%0·8–1·8% 8·6%

A recent review acknowledged the significant effect of both CYP2C9*2 and *3 and VKORC1 polymorphisms on warfarin dose requirements in Caucasians and Asians, but this effect was less clear in those of African descent (Schelleman et al, 2008). The authors further commented on the lack of studies investigating the effect of these variants in those of African descent. A recent international study demonstrated the superiority of a pharmacogenetic model over a clinical algorithm for warfarin dosing. Clinical and genetic (CYP2C9 and VKORC1 genotype) data were collected from a large cohort of patients (4043 subjects), from which dosing algorithms were derived and then validated in a further cohort (1009 subjects; International Warfarin Pharmacogenetics Consortium, 2009). Subjects were of different ethnic background with approximately 55% Caucasians, 30% Asians and 9% Black in each cohort. The pharmacogenetic model was effective in predicting patients with low (<21 mg/week) and high (more than 49 mg/week) warfarin requirements and furthermore eliminated the ethnic variation (seen in the clinical model) in response to warfarin. These patients accounted for 46% of the cohort and are those potentially at greatest risk of over- or under-anticoagulation with conventional warfarin dosing. Currently genotyping for the above variants is not standard practice, however if an improvement in safety outcome measures were demonstrated to offset the increased costs of testing, this practice could be adopted. The advent of new anticoagulants, such as dabigatran and rivaroxaban, with predictable pharmacokinetics is likely to substantially reduce the use of warfarin in the near future. To date there are no reports of varied efficacy of the new anticoagulants across different ethnic groups.

There are no studies comparing efficacy of thromboprophylaxis across different ethnic groups. Two small studies undertaken in South East Asia demonstrated efficacy of low molecular weight heparin (LMWH) thromboprophylaxis with a significant reduction in DVT following total knee replacement from 14% without thromboprophylaxis to 0 (Fong et al, 2000), and from 16% to 2% following total hip replacement (Yoo et al, 1997), with no increase in major bleeding in either study. Given that LMWH has high bioavailability and is renally excreted unchanged, it seems unlikely that its efficacy would be adversely influenced by ethnicity.

Discussion

VTE is now acknowledged as a significant problem in both black and Asian populations, with a probable higher incidence and complication rate in blacks. Family history is a common feature and is, to some extent, explained by known hereditary thrombophilia within both Caucasian and Asian populations. Idiopathic VTE is prevalent in all groups and may reflect an as yet uncharacterised underlying thrombophilia. The genetic predisposition among black subjects is yet to be determined. Conventional thrombophilia screening in the black population is rarely informative and should thus be avoided. It is essential that reference ranges from ethnically-matched controls are established for the natural anticoagulants and markers of fibrinolysis, such as d-dimer, to enable appropriate interpretation.

The lack of consistency in the classification of ethnic/racial groups greatly limits the application and extension of our knowledge of ethnic variation in VTE. The terminology employed by the studies reported above varies greatly and demonstrates the lack of uniformity in this area. As previously mentioned, the US Census classification of Asian race includes East Asians, which represent a distinct group in studies of genetic clustering (Burchard et al, 2003). The categories employed in the UK reflect genetic clustering more closely with the separate category for Chinese, but this does not encompass others from East Asia. Until there is universal use of a single classification system, comparing results from different countries will remain problematic. A further consideration is that race/ethnicity is self-assigned, and may therefore reflect cultural and social influences rather than biological or geographic origin. It has been proposed that these problems be overcome with the use of genomic sequencing to classify research subjects into groups that will reflect ancestral origin more closely (Burchard et al, 2003; Rotimi, 2004). This approach raises further ethical issues and certainly would not be practical for the purpose of population-based studies.

Whilst there is a body of evidence slowly emerging in the field of ethnicity and VTE, there remains a lack of high quality prospective research. Much of the epidemiological work investigating ethnicity and VTE has arisen from USA and, unfortunately, the vast majority of this involved analysis of retrospective data (Siddique et al, 1996; White et al, 1998, 2004, 2005; Stein et al, 2004a,b,c; Schneider et al, 2006), with a single ongoing prospective case–control study reported (Dowling et al, 2003). In the UK, only a single prospective case–control study has been published in this field (Patel et al, 2003). The studies in the African-American population have identified the increased incidence of VTE (Stein et al, 2004b; White et al, 2005; Schneider et al, 2006) and associated adverse outcome in comparison with other American ethnic groups (Siddique et al, 1996; Stein et al, 2004a; Aujesky et al, 2007). Whilst the lack of heritable thrombophilia (Dowling et al, 2003; Patel et al, 2003) and importance of raised FVIII levels (Patel et al, 2003) has been demonstrated, the pathogenesis and natural history of VTE remains poorly understood in African-Americans and the UK black population. Despite the occurrence of a positive family history as frequently in African-Americans as in Caucasians (Dowling et al, 2003) a heritable predisposition is yet to be identified. There is a dire need for quality prospective research to further elucidate underlying genetic and environmental contributors to the increased risk of VTE within the black population. Further investigation of the apparent protection afforded by Asian/Pacific Islander and Hispanic race in the USA is also warranted.

There is a virtual absence of VTE research in Africa and other developing regions. The impact of the adoption of a Western lifestyle on the risk of VTE within black populations remains unclear. Healthcare systems are poorly established in the majority of developing countries and illnesses rarely seen in the Western world, such as malnutrition and preventable communicable diseases, attract a higher international profile. Given VTE has only recently gained recognition as a serious public health issue in the UK (House of Commons Health Committee Report, 2005), it is unsurprising that it is not yet acknowledged as such in developing countries. Furthermore, early studies suggesting a low incidence of VTE in these regions (Thomas et al, 1960) may have provided false reassurance and led to a subsequent lack of interest in this area. Without investment into VTE research in these countries, our understanding of the role of ethnicity as a risk factor for VTE will remain limited.

Whilst there is a plethora of well-conducted research exploring epidemiology, genetic predisposition and outcomes following VTE in Caucasian populations, this is sadly still lacking in other ethnic groups. Despite growing recognition, there remains an urgent need for prospective research to further elucidate the pathogenesis and natural history of VTE in non-Caucasian ethnic groups.

Ancillary