Factors that predict thrombosis in relatives of patients with venous thromboembolism

Venous thromboembolism (VTE) is a multifactorial disease caused by hereditary and acquired risk factors.^1-3  Because there is often a hereditary component to the occurrence of VTE, thrombosis occurs more often in the first-degree relatives (ie, parents, siblings, children) of patients with VTE than in the general population.^4-9  The increased risk for thrombosis in relatives is incompletely explained by the presence of known thrombophilias, as the risk for thrombosis in first-degree relatives is increased even if patients do not have a detectable defect.^8,9 

We previously showed that the risk for thrombosis in first-degree relatives of patients with a first unprovoked VTE tends to be higher if the patient had factor V Leiden or the prothrombin 20210A gene variant compared with neither abnormality.⁹  However, a more striking finding was that relatives of younger index cases (<45 years) had a much higher risk for thrombosis than relatives of older index cases. We suspect that patients who have unprovoked VTE at a young age often have undetected hereditary thrombophilias and that these defects increase the risk for thrombosis in their relatives.

The risk for thrombosis is uncertain in relatives of patients with provoked VTE (eg, recent surgery or cancer). Hereditary thrombophilias are associated with provoked VTE, but less commonly than with unprovoked VTE.^10-12  Therefore, it is reasonable to suspect that the risk for thrombosis is lower in first-degree relatives of patients with provoked than those with unprovoked VTE. To test this hypothesis, we assessed the risk for thrombosis in first-degree relatives of patients with provoked VTE in the current study and compared that risk with the risk we previously observed in the first-degree relatives of patients with unprovoked VTE. We also wanted to determine whether, similar to in first-degree relatives of patients with unprovoked VTE,⁹  there was a higher risk for thrombosis in family members of patients who had provoked VTE at a young age. We then combined individual patient data from the current and previous study⁹  to increase our ability to identify characteristics of patients and first-degree relatives that predicted the risk for thrombosis in relatives. This information is important for counseling of first-degree relatives of patients with thrombosis.

Using a cross-sectional design and predefined criteria (see following), we determined whether there was a history of previous VTE in the first-degree relatives of consecutive patients (index cases) with a first episode of provoked symptomatic proximal deep vein thrombosis or pulmonary embolism. Index cases were enrolled prospectively at 6 university hospitals (5 in Canada and 1 in France) when they were diagnosed with a first episode of acute symptomatic VTE. Index cases had to satisfy all of the following criteria: objectively confirmed proximal deep vein thrombosis (ie, ultrasonography) or pulmonary embolism (ie, lung scanning),¹³  willingness to provide blood for testing for factor V Leiden and the prothrombin 20210A gene variant, have at least a single first-degree relative who could be evaluated for previous VTE, and willingness to provide written informed consent to participate in the study and to allow at least 1 of their first-degree relatives to be approached for the study. For thrombosis to be considered provoked, it had to have occurred either within 3 months of having major surgery, major leg trauma, or immobilization (confined to bed for at least 3 consecutive days) or in association with an active malignancy (ie, within the past 2 years).⁹  Thrombosis that occurred in the absence of these risk factors was considered unprovoked.

First-degree relatives were eligible as study subjects if they were a biological child, full sibling, or biological parent of an index case and were at least 16 years of age, and if they provided informed consent. First-degree relatives who were dead could be included as study subjects provided the index case agreed and information about previous VTE was available. Research ethic boards of Brest Hospital Center (France) and Henderson General Hospital, McMaster University (Hamilton, ON, Canada) approved the study.

Using a previously described algorithm,⁹  first-degree relatives were classified as “have had VTE” if they satisfied either of the following 2 criteria: First, results of diagnostic testing were available that documented previous deep vein thrombosis (including thrombosis confined to the distal deep veins) or pulmonary embolism;¹³  and second, the relatives had, in addition to a history of symptoms suggestive of VTE, at least 1 of the following¹⁴ : a history of having been treated with anticoagulant therapy for at least 2 months without another indication,¹⁴  a current ultrasound examination that showed that the proximal deep veins were not fully compressible or that there was reflux in a popliteal vein,¹⁵  or current symptoms and signs suggestive of the postthrombotic syndrome (defined as a score of 5 or higher on the Villalta scale).¹⁶  Relatives were classified as “have not had VTE” if they satisfied all of the following criteria: no known or suspected previous diagnosis of VTE, no unexplained anticoagulation in the past, and not currently having symptoms or signs suggestive of the postthrombotic syndrome (ie, had a score lower than 5 on the Villalta scale). Relatives were classified as “uncertain for previous VTE” if they did not satisfy the criteria for either previous or no previous VTE.

After first-degree relatives had completed assessments for previous VTE, their index cases were categorized as positive for factor V Leiden or the prothrombin 20210A gene variant or negative for both.⁹  Personnel who were unaware of the index case’s family history of VTE or the participant’s past history of VTE performed these assays in a central laboratory in either France or Canada.

The number of first-degree relatives of index cases with a provoked VTE who needed to be studied was based on 80% power to detect a 50% lower prevalence of thrombosis in relatives of patients with provoked compared with unprovoked VTE,^10-12  while accepting a 2-sided α error of 0.025, assuming a prevalence of previous thrombosis of 5.7% in the 1752 relatives of patients with unprovoked VTE.⁹  This yielded a sample size of 875 for the number of relatives of patients with provoked VTE who needed to be evaluated (including those classified as “uncertain for previous VTE”).

The average annual incidence of VTE was calculated by dividing the number of first episodes of VTE by the total number of years of observation (expressed as events per 100 person-years). For each study participant, the period of observation was from 16 years of age until their age at the time of the first VTE, death, or when they were assessed at enrolment (whichever occurred latest). The influence of the index case’s characteristics (factor V Leiden or prothrombin 20210A gene variant vs neither, age at diagnosis, female vs male, deep vein thrombosis vs pulmonary embolism) and the first-degree relative's characteristics (age during the observation period [assessed as a time-dependent variable], female vs male, parent vs sibling vs child, dead vs alive when assessed) on the risk for VTE in the first-degree relatives was investigated using a logistic regression. If the study subject had more than 1 first-degree relative with a history of VTE (ie, other than the index case), we introduced a random intercept to account for the clustering effect within families (intrafamily correlation) into a generalized linear mixed model.¹⁷  Associations between the characteristics of the index case or the first-degree relative, and the risk for thrombosis in the first-degree relative, were assessed as follows: first, single characteristics were assessed using univariable analysis, and second, a multivariable model was fitted that contained only those characteristics associated with a significance level of P < .1 in the univariable analysis.¹⁸  Statistical analyses were performed using SPSS software (version 20.0; IBM Corporation, Armonk, NY) with the exception of the random coefficient logistic modeling, which was done using SAS PROC GLIMMIX (version 9.1; SAS, Inc., Cary, NC).

To compare findings and combine individual patient data, we used the same study design, definitions of VTE (proximal deep vein thrombosis alone and pulmonary embolism; provoked and unprovoked thrombosis), and statistical methods in the current study as we used in our previous study that assessed VTE in first-degree relatives of patients with a first unprovoked VTE.⁹ 

Between January 2006 and January 2012, 197 patients with a provoked VTE were assessed, of whom 40 did not consent to be tested for factor V Leiden and the prothrombin 20210A gene variant, 9 did not have any first-degree relatives, and 10 did not consent to having their relatives contacted (Figure 1). The remaining 138 index cases with provoked VTE had 915 first-degree relatives who were enrolled (Figure 1). Among these 915 relatives, the assessment of previous VTE was uncertain in 50 (Figure 1). Therefore, analyses included 865 first-degree relatives who could be categorized as either “have had VTE” or “have not had VTE.”

There was an inverse relationship between the age of index cases and the risk for VTE in family members. For each year that the index case was younger, the incidence of VTE increased by 3.0% in the first-degree relatives (corresponding to an adjusted OR of 0.97 per year that the index case was older [95% CI, 0.96-0.999; P = .04]). When the age of index cases was divided into quartiles, the incidence of VTE was almost 5 times higher in the first-degree relatives of index cases in the youngest (index cases < 47 years) compared with the oldest (index cases > 72 years) quartile (adjusted OR, 4.97; 95% CI, 1.24-19.99; P = .024).

Among the 138 index cases with provoked VTE, 20 (14.5%) had factor V Leiden, the prothrombin 20210A gene variant, or both abnormalities, and 118 (85.5%) had neither abnormality (Table 1). The 20 index cases with 1 or both abnormalities had 123 first-degree relatives, and among these individuals, there were 2 episodes of VTE, corresponding to an incidence of 0.04 events per 100 person-years (95% CI, 0.01-0.15 events) (Table 2). The 118 index cases with neither abnormality had 742 first-degree relatives; among these, there were 19 episodes of VTE, corresponding to an incidence of 0.06 events per 100 person-years (95% CI, 0.04-0.09 events) (Table 2). The adjusted OR for VTE in the relatives of index cases with 1 or both abnormalities compared with the relatives of index cases with neither abnormality was 0.61 (95% CI, 0.14-2.66) (Table 2).

The 20 index cases with provoked VTE who had factor V Leiden, the prothrombin 20210A gene variant, or both abnormalities had 60 relatives who were tested for the same abnormality as in the index case. One of the 31 relatives who had the same abnormality as their index case had a history of VTE, but there were no episodes of VTE in the 29 relatives who did not have the same abnormality as their index case (supplemental Table 1).

The incidence of VTE in relatives of index cases with provoked VTE was lower than in relatives of index cases with unprovoked VTE (adjusted OR, 0.42; 95% CI, 0.26-0.70) (Tables 2 and 3).

In multivariable analyses, both unprovoked vs provoked VTE in the index case (adjusted OR, 2.38; 95% CI, 1.43-3.85) and VTE at a younger age in the index case (adjusted OR, 0.97 per year older; 95% CI, 0.96-0.99 per year older) were independent predictors of the risk for thrombosis in the patient’s first-degree relatives (Table 3; Figure 2). Presence of factor V Leiden or prothrombin 20210A gene variant in index cases was associated with a higher risk for VTE in first-degree relatives in univariable analysis, but this association was no longer statistically significant in the multivariable analysis (Table 3). Sex of the index case, and whether the index event was a deep vein thrombosis or a pulmonary embolism, were not predictive of thrombosis in the index case’s relatives.

Among the 507 index cases, 405 (120 provoked; 285 unprovoked) had no other first-degree relative with a history of VTE, 82 (15 provoked; 67 unprovoked) had 1 other first-degree relative with VTE, 19 (3 provoked; 16 unprovoked) had 2 other first-degree relatives with VTE, and 1 (unprovoked) had 3 other first-degree relatives with VTE (Table 4). The risk for VTE was higher if, in addition to the index case, study participants had another first-degree relative (ie, 2 or more relatives vs 1 relative) with a history of VTE (adjusted OR, 2.71; 95% CI, 2.22-3.31) (Table 4). The increase in the risk for VTE associated with having 2 or more first-degree relatives with VTE, as opposed to just the index case, did not appear to differ according to whether the index case had provoked (adjusted OR, 2.78; 95% CI, 1.79-4.33) or unprovoked (adjusted OR, 2.53; 95% CI, 2.02-3.18) VTE, had factor V Leiden or prothrombin 20210A gene variant (adjusted OR, 2.73; 95% CI, 1.89-3.94), or did not have 1 or both of these abnormalities (OR, 2.68; 95% CI, 2.11-3.41).

The prevalence of factor V Leiden, the prothrombin 20210A gene variant, or both abnormalities was higher in index cases of unprovoked compared with provoked VTE (28.4% vs 14.5%; P < .001) (Table 1). A total of 125 index cases (20 provoked and 105 unprovoked) had 1 or both abnormalities, and among their first-degree relatives tested for the same abnormality, that abnormality was present in 149 and absent in 136 (supplemental Table 1). VTE occurred more often in relatives who had the same abnormality as their index case compared with those who did not have the abnormality that was present in their index case (adjusted OR, 4.42; 95% CI, 1.35-14.38).

First, we found that the risk for VTE in the first-degree relatives of patients with a provoked VTE is less than half the risk in first-degree relatives of patients with unprovoked VTE. Second, we found that thrombosis at a younger age was associated with a higher risk for VTE in patients’ first-degree relatives compared with thrombosis that occurred at an older age: the risk for VTE in a patient’s first-degree relatives was about 3 times as high if thrombosis occurred before 45 years compared with after 72 years. The influence of these 2 factors on the risk for VTE in first-degree relatives was additive and occurred independent of the presence of factor V Leiden or the prothrombin 20210A gene in index cases. Third, the risk for thrombosis in first-degree relatives was higher if, in addition to the index case, VTE had occurred in other family members, with more than a doubling with 1 additional and more than a tripling with 2 additional effected relatives. Last, as is well recognized, if the index case had factor V Leiden or the prothrombin 20210A gene, the risk for thrombosis in first-degree relatives was markedly influenced by whether they had the same mutation.^5-9 

We are not aware of previous studies that compared the risk for VTE between first-degree relatives of patients with provoked vs unprovoked VTE. We previously reported that unprovoked VTE at a young age is associated with a heightened risk for VTE in patient’s relatives.⁹  We now show that this is also true for the first-degree relatives of patients with provoked VTE at a young age. Consistent with our analyses, the Swedish Multigenerational Registry reported a higher risk for VTE in siblings of younger compared with older patients with a spectrum of VTE types, and that the risk for VTE also increased with the number of family members with thrombosis.¹⁹ 

The strengths of our study include the prospective enrolment of unselected cases of provoked and unprovoked VTE, use of a standardized approach to determine whether first-degree relatives had a history of VTE, adjustment for potential confounding factors (eg, age of relatives during the observation period and within-family clustering), and blinding of assessors to first-degree relatives’ or index cases’ factor V Leiden and prothrombin 20210A gene variant status.²⁰  Limitations of our study include that index cases were not evaluated for hereditary thrombophilias other than factor V Leiden and the prothrombin G20210A gene variant (eg, deficiencies of protein C, protein S, and antithrombin); that follow-up of first-degree relatives was retrospective rather than prospective, which increases the potential for misclassification of VTE events and the occurrence of survivor bias; and that we were unable to adjust for factors that may vary over time, such as obesity. A sensitivity analysis in which all “uncertain for VTE” first-degree relatives of both provoked and unprovoked index cases were assumed to either “have had VTE” or “not have had VTE” had little effect on the comparison of VTE risk between the first-degree relatives of provoked vs unprovoked index cases (data not shown). It is also possible that inclusion of estrogen-associated VTE among the unprovoked index cases could have diluted the comparison of VTE risk between first-degree relatives of unprovoked vs provoked cases. We were unable to assess this directly, but we think this is unlikely because findings for this comparison in the relatives of younger women were similar to those for the overall study (data not shown).

The findings of this study have potentially important clinical and pathophysiological implications. First, all patients with VTE, but particularly those with unprovoked VTE, unprovoked or provoked VTE at a young age, or a family history of VTE, should inform their first-degree relatives that they have a heightened risk for thrombosis. Relatives can then remind healthcare providers that they should receive VTE prophylaxis when they are in high-risk situations (eg, after surgery), and they can factor this heightened risk into decision making around the use of estrogens, which will further increase their risk for thrombosis.²¹  Second, our findings support that known and unknown hereditary thrombophilias contribute to both provoked and unprovoked VTE, particularly in younger patients and in those with unprovoked VTE and a family history of thrombosis.

In conclusion, the risk for VTE in the first-degree relatives of patients with a first VTE is strongly influenced by whether the VTE was provoked or unprovoked, the patient’s age when the VTE occurred, and the number of relatives who have had thrombosis. The risk for VTE in first-degree relatives is about twice as high if the index case had an unprovoked compared with a provoked VTE, is about 3 times as high if the index case had VTE before about 50 years compared with later in life, and is at least twice as high if 2 rather than 1 family members have had VTE. This information compliments the results of thrombophilia testing when counseling family members about their risk for thrombosis.

The following institutions and nurses participated in this study: G. Marhic, M.L. Guennegues, L. Gourhant, S. Melac, Brest University Hospital Centre, Brest, France (n = 1742); T. Winkworth, Hamilton Health Sciences–Henderson Hospital, Hamilton, ON, Canada (n = 373); L. Sardo, P. Stevens, J. Joval, Hamilton Health Sciences–McMaster Medical Centre, Hamilton, ON, Canada (n = 202); P. Waddell, Ottawa Hospitals–Civic Campus, Ottawa, ON, Canada (n = 161); C. Strulovitch, Jewish General Hospital, Montreal, QC, Canada (n = 114); and K. Woods, T. Schnurr, St. Joseph’s Hospital, Hamilton, ON, Canada (n = 25).

Coordinating and Methods Center: Groupe d’Etude de la Thrombose de Bretagne Occidentale, EA 3878, CIC INSERM 0502, University Hospital Centre, Brest, France, and the Henderson Research Centre, Hamilton, ON, Canada.

Central Laboratories: P. Joshua, G. Mattina, M. Lauzon, M. Johnston, Hemostasis Reference Laboratory, Hamilton, ON, Canada; C. Ferec, C. Le Marechal, B. Mercier, INSERM Unité Mixte de Recherche 1078, Laboratory Molecular Genetic, Brest, France.

S.R.K. is a recipient of a National Clinical Investigator Award from the Fonds de la Recherche en Santé du Québec; J.S.G. holds the David Braley and Nancy Gordon Chair in Thromboembolic Disease; P.S.W. was supported by a Canada Research Chair and a Heart and Stroke of Ontario Program Grant; and C.K. is supported by the Jack Hirsh Professorship in Thromboembolism and by the Heart and Stroke Foundation of Ontario.

This study was supported by grants from Programme Hospitalier de Recherche Clinique, France, and the Heart and Stroke Foundation of Ontario.

Contribution: F.C. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis; F.C. and C.K. conceived and designed the study; F.C., C.K., C.L., C.T., D.M., and J.A.J. analyzed and interpreted the data and drafted the article; F.C., C.L., C.K., J.A.J., S.R.K., J.S.G., P.S.W., J.D.D., C.T., and D.M. provided critical review of intellectual content; F.C., C.L., C.T., S.R.K., J.S.G., P.S.W., J.D.D., D.M., and C.K. provided study material or patients; F.C. and C.K. obtained funding; F.C., C.K., C.L., and D.M. provided administrative, technical, and logistic support; F.C., C.K., J.A.J., and C.T. collected and assembled data; and all authors approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Francis Couturaud, Groupe d’Etude de la Thrombose de Bretagne Occidentale, EA 3878, Centre d'Investigation Clinique INSERM O502, Institut de Formation et de Recherche 148, Department of Internal Medicine and Chest Diseases, University Hospital Centre La Cavale Blanche, European Brittany University, 29609 Brest, France; e-mail: francis.couturaud@chu-brest.fr.