Hamostaseologie
DOI: 10.1055/a-2614-8526
Original Article

The Impact of Genetic and Acquired Risk Factors on Thromboembolic Events: A Retrospective Study

1   Rheumatology, Department of Clinical Internal, Anaesthesiologic and Cardiovascular Sciences, Sapienza University of Rome, Rome, Italy
,
Ana-Luisa Stefanski
2   Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Department of Rheumatology and Clinical Immunology, Berlin, Germany
3   Deutsches Rheumaforschungszentrum Berlin, Berlin, Germany
,
Thomas Dörner
2   Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Department of Rheumatology and Clinical Immunology, Berlin, Germany
3   Deutsches Rheumaforschungszentrum Berlin, Berlin, Germany
› Institutsangaben
 

Abstract

This study, involving a cohort of 980 patients with arterial and/or venous events, evaluated the relative importance of genetic and traditional risk factors using a machine learning model approach. The analysis revealed that conventional risk factors—such as age, gender, tobacco use, obesity—and acquired thrombophilia (most prominently antiphospholipid syndrome [APS]) outperformed any genetic risk variants analyzed. Of note, the presence of heterozygous FVL mutations was associated with a reduced arterial risk, whereas carriers of heterozygous factor VII activation protease (FSAP) mutations had a lower risk of venous events.

Thus, by employing diverse statistical approaches, the study showed that acquired risk factors exert the most substantial impact on the development of thrombotic events, except for the well-known role of FVL mutations in venous events.


Introduction

Arterial thrombotic events and venous thromboembolism (VTE), which include deep vein thrombosis (DVT) and pulmonary embolism (PE), represent a significant cause of morbidity and mortality worldwide.[1] [2] [3] Various risk factors, both acquired and inherited, are known to heighten the risk of thromboembolic events, typically through mechanisms like blood flow stasis, endothelial injury or increased hypercoagulability.[4] [5] [6] Traditional and acquired risk factors include tobacco use, obesity, arterial hypertension, diabetes mellitus, hypercholesterolemia, oral contraceptives, hormonal replacement therapy, cancer, and antiphospholipid syndrome.[7] Furthermore, an increased risk of thrombosis has been attributed to specific gene mutations. After the discovery of the prothrombotic role of factor V Leiden mutation,[8] other gene polymorphisms have emerged as potential risk factors, such as prothrombin G20210A, plasminogen activator inhibitor-1 4G/4G, and methylenetetrahydrofolate reductase C677T.[9] [10] [11] [12] However, while most patients with thromboembolic events have at least one identifiable risk factor, the predictive strength of these factors varies significantly.[13] For effective management of these patients, it is essential for clinicians to understand the relative impact of the specific risk factors.

Nevertheless, in most studies genetic polymorphisms associated with thrombophilia are evaluated independently from traditional risk factors including a distinction of arterial versus venous risk factors.

The purpose of this study was to examine the occurrence of thrombotic events in a retrospective cohort of individuals who had undergone thrombophilia testing, focusing on traditional and genetic risk factors. Additionally, the study aimed to assess the relative importance of individual genetic and traditional risk factors on arterial and venous thrombotic events.


Materials and Patients

We retrospectively analyzed data from a total of 980 patients ([Table 1]), who underwent thrombophilia assessment including various clinical and laboratory parameters as well as genetic testing of the following polymorphisms or mutations: prothrombin (FII) 20210G/A mutation, factor V Leiden (FVL) 1691G/A mutation, plasminogen activator inhibitor-1 (PAI1) 5G/4G mutation, tissue-type plasminogen activator (t-PA) intron h deletion/insertion, factor VII activation protease (FSAP) G511E Marburg I polymorphism, and methylenetetrahydrofolate reductase (MTHFR) mutation C677T.

Table 1

Clinical and demographic characteristics of the cohort

Characteristics

N (%)

Female sex

637 (65)

Median age, years (IQR)

45 (34–59)

Thrombotic risk factors assessed

617 (63)

Tobacco use

93 (9.4)

Diabetes

53 (5.4)

Hyperlipidemia

220 (22.4)

Hypertension

278 (28.3)

Hormonal estrogen therapy[a]

66 (6.7)

Obesity

153 (15.6)

Antiphospholipid syndrome

64 (6.5)

Active cancer

65 (6.6)

Autoimmune/inflammatory disease (other than APS)[b]

125 (12.8)

Thrombotic events

582 (59.3)

Arterial thrombosis

173 (17.6)

Ischemic stroke

109 (11.1)

Myocardial infarction

37 (3.7)

Other arterial thrombosis[c]

43 (4.3)

Venous thrombosis

459 (46.8)

Deep vein thrombosis

284 (28.9)

Pulmonary embolism

185 (18.8)

Other venous thrombosis[c]

122 (12.4)

Arterial + venous thrombosis

50 (5.1)

Recurrent arterial events

46 (4.6)

Recurrent venous events

145 (14.7)

Abbreviations: APS, antiphospholipid syndrome; IQR, interquartile range.


Notes: aCombined oral contraceptives or hormone replacement therapy.


b Further details in [Supplementary Table S4] (online only).


c Thrombosis occurring in unclassified locations.


We included all individuals who underwent thrombophilia genetic screening, regardless of their history of thrombotic events. In the case of asymptomatic patients, the indication for testing was based on clinical judgment, with the intention to stratify the thrombotic risk in specific individuals, including subjects with a strong family history of thrombosis, especially if these events occurred at a young age (<50 years) and/or in multiple family members, before starting therapies that increase thrombotic risk, such as combined oral contraceptives or hormone replacement therapy, and patients with underlying clinical conditions that predispose to thrombosis (i.e., autoimmune diseases, cancer).

To further evaluate the relative importance of genetic and traditional risk factors, we employed a machine learning approach, comparing different predictive models, including logistic regression, random forest, and classification tree. The outcome variable was the occurrence of any thrombotic event. Model training and evaluation were conducted using the “Orange data mining” software (version 3.35), applying 10-fold cross-validation to ensure robustness. Predictor variables were selected using the Gini index as a ranking criterion, and the top 10 features were retained for model development ([Supplementary Table S1] (online only)). Predictive performance was primarily assessed using receiver operating characteristic (ROC) curves and the corresponding area under the curve (AUC) values ([Fig. 1A, B]). Finally, we applied both binary and multinomial logistic regression models to assess risk factors for any vascular event including different arterial and venous events.

Zoom
Fig. 1 (A) Receiver operating characteristic (ROC) curves for the prediction of any thrombotic event. (B) Area under the ROC curve (AUC) of the prediction models analyzed. (C) Relative importance of variables in predicting thrombosis, estimated by decrease in AUC of the logistic regression model.

Results

In total, 980 patients were included in the study, with a female to male ratio of 2:1 and a median age of 45 years ([Table 1]). Of this, 156 patients carried FVL mutation (1691G/A mutation; 9 homozygous, 147 heterozygous), 66 had factor II mutation (20210G/A; 1 homozygous, 65 heterozygous), 763 had PAI1 mutation (5G/4G; 298 homozygous, 465 heterozygous), 770 carried t-PA mutations (intron h deletion/insertion; 260 homozygous, 510 heterozygous), 74 were found to have FSAP mutations (G511E Marburg I; 2 homozygous, 72 heterozygous), and finally 517 had MTHFR mutations (C677T; 91 homozygous, 426 heterozygous) ([Table 2]).

Table 2

Prevalence of the genetic mutations

Gene

Homozygous mutated

N (%)

Heterozygous mutated

N (%)

Any mutation

N (%)

Factor V

9 (0.9)

147 (15)

156 (15.9)

Factor II

1 (0.1)

65 (6.6)

66 (6.7)

MTHFR

91 (9.2)

426 (43.4)

517 (52.7)

PAI1

298 (30.4)

465 (47.4)

763 (77.8)

t-PA

260 (26.5)

510 (52)

770 (78.6)

FSAP

2 (0.2)

72 (7.3)

74 (7.6)

Abbreviations: FSAP, factor VII activation protease; MTHFR, methylenetetrahydrofolate reductase; PAI1, plasminogen activator inhibitor-1; t-PA, tissue-type plasminogen activator.


More than half of the patients included (59.3%) experienced a thrombotic event. Among those, 21.1% displayed arterial thrombosis, 70.3% had venous thrombosis, while 8.6% patients presented with both arterial and venous events ([Table 1]). A total of 64 subjects (6.5%) were diagnosed with the autoimmune acquired thrombophilia antiphospholipid syndrome (APS). The remaining individuals underwent screening testing related to familial, obstetric, or other clinical risk evaluation.

The machine learning model revealed several notable findings. Conventional risk factors—such as age, gender, tobacco use, obesity—and acquired thrombophilia (here notably APS) consistently outcompeted genetic risk variants. In particular, age appeared as the most prominent variable, followed by sex, APS, hormonal estrogen therapy, tobacco use, and obesity ([Fig. 1C]).

As an initial step, we conducted an adjusted binary logistic regression analysis, which identified several significant predictors of an increased risk for any thrombosis including age, male sex, hypertension, hormonal estrogen therapy, and obesity. Of note, APS exhibited the highest odds ratio (5.244; 95% CI 2.369–11.605; p < 0.001). None of the genetic mutations assessed were significant predictors in this model ([Supplementary Table S2] (online only)).

Using the multinomial logistic regression model ([Table 3]), we found age, male sex, and APS associated with an increased risk for all three outcomes: arterial thrombosis, venous thrombosis, and both arterial and venous events. Of particular note, APS was associated with the highest risk for arterial manifestations (odds ratio 8.069; p < 0.001). Furthermore, arterial thrombosis was significantly linked to tobacco use, hyperlipidemia, and hypertension ([Fig. 2A]). For venous events, key risk factors included hormonal estrogen therapy, obesity, and homozygous FVL, the latter yielding an odds ratio of 10.467 (p = 0.042) ([Fig. 2B]).

Table 3

Adjusted multinominal logistic regression for thrombotic outcomes

Variable

Arterial thrombosis

Venous thrombosis

Arterial + Venous thrombosis

Exp (B) (95% confidence interval)

p-value

Exp (B) (95% confidence interval)

p-value

Exp (B) (95% confidence interval)

p-value

Age

1.042 (1.023–1.061)

<0.001

1.059 (1.046–1.073)

<0.001

1.052 (1.027–1.078)

<0.001

Male sex

2.985 (1.765–5.050)

<0.001

3.641 (2.500–5.304)

<0.001

3.423 (1.716–6.831)

<0.001

Tobacco use

4.783 (2.341–9.774)

<0.001

1.424 (0.734–2.764)

0.296

1.502 (0.491–4.592)

0.476

Diabetes

0.716 (0.239–2.142)

0.550

1.088 (0.446–2.652)

0.853

0.614 (0.141–2.667)

0.515

Hyperlipidemia

4.032 (2.328–6.985)

<0.001

0.888 (0.557–1.417)

0.619

3.045 (1.480–6.262)

0.002

Hypertension

2.513 (1.388–4.550)

0.002

1.346 (0.853–2.124)

0.202

2.745 (1.264–5.963)

0.011

Hormonal estrogen therapy

2.236 (0.766–6.530)

0.141

5.240 (2.773–9.900)

<0.001

1.212 (0.148–9.944)

0.858

Obesity

0.978 (0.471–2.032)

0.953

2.554 (1.580–4.128)

<0.001

1.348 (0.537–3.382)

0.525

Antiphospholipid syndrome

8.069 (3.067–21.230)

<0.001

4.383 (1.918–10.013)

<0.001

9.105 (2.924–28.352)

<0.001

Autoimmune/inflammatory disease

1.510 (0.731–3.118)

0.265

1.467 (0.882–2.442)

0.140

2.384 (0.994–5.719)

0.052

Active cancer

1.459 (0.574–3.707)

0.428

0.875 (0.446–1.719)

0.699

1.283 (0.380–4.330)

0.688

Factor V

Heterozygous

0.377 (0.169–0.842)

0.017

0.802 (0.514–1.251)

0.331

0.672 (0.256–1.762)

0.419

Homozygous

0.994

10.467 (1.086–100.837)

0.042

0.996

Wild type

Factor II

Heterozygous

0.431 (0.136–1.365)

0.152

1.085 (0.566–2.082)

0.806

0.681 (0.171–2.717)

0.587

Homozygous

0.998

0.997

Wild type

MHTFR

Heterozygous

0.803 (0.485–1.330)

0.394

0.984 (0.698–1.386)

0.925

0.541 (0.271–1.082)

0.082

Homozygous

0.76 (0.323–1.788)

0.530

0.907 (0.495–1.665)

0.753

0.537 (0.163–1.766)

0.306

Wild type

PAI1

Heterozygous

1.008 (0.533–1.907)

0.980

0.809 (0.532–1.230)

0.321

0.833 (0.372–1.863)

0.656

Homozygous

1.370 (0.703–2.669)

0.355

0.813 (0.515–1.283)

0.374

0.836 (0.35–1.997)

0.686

Wild type

t-PA

Heterozygous

0.958 (0.527–1.741)

0.887

0.860 (0.566–1.307)

0.480

0.697 (0.297–1.640)

0.409

Homozygous

0.528 (0.257–1.083)

0.082

0.883 (0.549–1.422)

0.610

1.022 (0.418–2.498)

0.961

Wild type

FSAP

Heterozygous

1.205 (0.535–2.714)

0.653

0.463 (0.235–0.915)

0.027

1.152 (0.396–3.353)

0.795

Homozygous

0.997

0.995

0.998

Wild type

Abbreviations: FSAP, factor VII activation protease; MTHFR, methylenetetrahydrofolate reductase; PAI1, plasminogen activator inhibitor-1; t-PA, tissue-type plasminogen activator.


Zoom
Fig. 2 Odds ratios for arterial thrombosis (A), venous thrombosis (B), and both arterial and venous thrombosis occurring in the same patient (C).

Interestingly, heterozygous FVL (1691G/A) mutations appeared to have a protective effect against arterial events (odds ratio 0.377; p = 0.017). Similarly, carriers of heterozygous Marburg I FSAP mutations had a lower risk of venous events (odds ratio of 0.463; p = 0.027).

In patients who experienced both arterial and venous thrombosis, male sex, hyperlipidemia, hypertension, and APS appeared as key risk factors ([Fig. 2C]).


Discussion

This study evaluated hereditary and acquired risk factors among patients undergoing thrombophilia assessment. Notably, it provides compelling evidence that traditional cardiovascular risk factors are the dominant contributors to the occurrence of not only arterial but also venous events, with APS emerging as a shared risk factor for venous and arterial events. In contrast, the role of hereditary (genetic) risk factors, although still relevant, was found to have a comparatively smaller impact. Beyond the well-established significance of the FVL mutation,[8] other genetic factors were not identified as substantial risk contributors.

Of note, this study found that heterozygous carriers of the FVL mutation (1691G/A) were associated with a reduced risk of arterial thrombosis. Similarly, the Copenhagen City Heart Study[14] showed a reduction of arterial events (ischemic stroke, but not myocardial infarction) among both heterozygous and homozygous FVL carriers. The high prevalence of FVL in the general Caucasian population, combined with studies indicating a reduced risk of severe sepsis[15] [16] in individuals with this mutation, has also led to the hypothesis that FVL polymorphism may represent an evolutionary “survival advantage” mutation.[17] However, the association of FVL and arterial events remains controversial. For instance, an increased risk of ischemic stroke has been described for both heterozygous and homozygous FVL.[18] Moreover, a recent meta-analysis[19] found no association between FVL and increased risk of subsequent atherothrombotic events or mortality in high-risk participants with established and treated coronary heart disease. Overall, the current evidence suggests no clear consensus on the role of FVL in arterial events.

The current study also highlighted a novel association between the G534E single nucleotide polymorphism (Marburg I) of FSAP and a reduced risk of venous events. Initially, this mutation has been linked to accelerated atherosclerosis and a low proteolytic activity.[20] Moreover, some authors found an association between Marburg I FSAP mutation and venous thrombosis.[21] [22] However, this result was not confirmed by other studies.[23] [24] FSAP seems to play a dual role in thrombus regulation by activating fibrinolysis while also promoting coagulation through activation of tissue factor (TF) and inhibition of tissue factor pathway inhibitor (TFPI).[25] The Marburg I FSAP variant, associated with reduced proteolytic activity,[26] may lead to lower “procoagulant activity” of FSAP, acting as a protective factor toward venous thrombosis.

Despite these findings, the study underscores that the influence of genetic risk factors, such as FVL and FSAP mutations, remains modest compared with the dominant role of acquired risk factors, which consistently demonstrated significant impact across all statistical models and methodologies.

It is important to note that this study has certain limitations that should be considered when interpreting the results.

First, it is crucial to note that these findings were obtained from a specific at-risk population referred for thrombophilia workup and as such preselected, which may limit the generalizability of the results.

Second, sociodemographic differences between asymptomatic subject and patients with thrombosis could affect the analysis. Indeed, the thrombosis group displayed a higher age and a greater proportion of males ([Supplementary Table S3] (online only)), both known to be risk factors of thrombosis.[27] [28] [29] To mitigate their confounding effect, both variables were included in the adjusted multivariate analysis. Other potential influencing factors, such as education level, income, and access to healthcare, were not available due to the retrospective design of the study. Ideally, prospective cohort studies with broader age ranges and longer follow-up periods would yield more robust and generalizable data.

Finally, the low number of subjects with certain genetic risk alleles, such as homozygous mutations for FVL (1691G/A), FII (20210G/A), and FSAP (G511E Marburg I) as well as the small number of patients with malignancies, restricted the statistical power to draw definitive conclusions for these subgroups.


Conclusions

This retrospective study provides evidence that acquired risk factors predominantly contribute to the development of thrombotic events, except for the well-established role of homozygous FVL mutation on VTE. Interestingly, individuals with heterozygous FVL mutations in our cohort demonstrated a reduced risk of arterial events, while carriers of heterozygous FSAP mutations exhibited a decreased risk of venous events. Thus, while genetic factors cannot be entirely disregarded, their influence on arterial and venous risks appears less significant compared with the predominant influence of partially modifiable cardiovascular risk factors and the presence of APS.

What is Known About this Topic?

  • Various risk factors, both acquired and inherited, are known to heighten the risk of thromboembolic events:

    • - Traditional risk factors (i.e., smoke, obesity, arterial hypertension, diabetes mellitus, hypercholesterolemia, oral contraceptives, hormonal replacement therapy, cancer, and antiphospholipid syndrome).

    • - Specific genetic mutations: prothrombin (FII) 20210G/A; factor V Leiden (FVL) 1691G/A; plasminogen activator inhibitor-1 (PAI1) 5G/4G; tissue-type plasminogen activator (t-PA) intron h deletion/insertion; factor VII activation protease (FSAP) G511E Marburg I polymorphism; methylenetetrahydrofolate reductase (MTHFR) mutation C677T.5

What Does this Paper Add?

  • It gives a comprehensive view on the interaction between genetic and acquired risk factors in the development of arterial and venous events in a large population.

  • It uses a machine learning model approach to evaluate the individual risk contribution of each variable.



Conflict of Interest

The authors declare that they have no conflict of interest.

Supplementary Material


Address for correspondence

Luca Rapino, MD
Department of Clinical Internal, Anaesthesiologic and Cardiovascular Sciences, Sapienza University of Rome
VII Padiglione, Viale del Policlinico, 155, 00191, Rome
Italy   

Publikationsverlauf

Eingereicht: 14. Februar 2025

Angenommen: 16. Mai 2025

Artikel online veröffentlicht:
08. Juli 2025

© 2025. Thieme. All rights reserved.

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany


Zoom
Fig. 1 (A) Receiver operating characteristic (ROC) curves for the prediction of any thrombotic event. (B) Area under the ROC curve (AUC) of the prediction models analyzed. (C) Relative importance of variables in predicting thrombosis, estimated by decrease in AUC of the logistic regression model.
Zoom
Fig. 2 Odds ratios for arterial thrombosis (A), venous thrombosis (B), and both arterial and venous thrombosis occurring in the same patient (C).