Thrombophilia Impact on Treatment Decisions, Subsequent Venous or Arterial Thrombosis and Pregnancy-Related Morbidity: A Retrospective Single-Center Cohort Study

(1) Background: Thrombophilia testing utility has remained controversial since its clinical introduction, because data on its influence on treatment decisions are limited. (2) Methods: We conducted a single-center retrospective cohort study of 3550 unselected patients referred for thrombophilia consultation at the Bern University Hospital in Switzerland from January 2010 to October 2020. We studied the influence of thrombophilia testing results on treatment decisions and evaluated the association between thrombophilia and thromboembolic and pregnancy-related morbidity events after testing up to 03/2021. (3) Results: In 1192/3550 patients (34%), at least one case of thrombophilia was found and 366 (10%) had high-risk thrombophilia. A total of 211/3550 (6%) work-ups (111/826 (13%) with low-risk thrombophilia and 100/366 (27%) with high-risk thrombophilia) led to an appropriate decision to extend or initiate anticoagulation, and 189 (5%) negative results led to the withholding of anticoagulation therapy inappropriately. A total of 2492 patients (69%) were followed up for >30 days, with a median follow-up of 49 months (range, 1–183 months). Patients with high-risk thrombophilia had a higher risk of subsequent venous thromboembolic events and pregnancy-related morbidity compared to those without thrombophilia. (4) Conclusions: Our study demonstrated the limited usefulness of thrombophilia work-up in clinical decision-making. High-risk thrombophilia was associated with subsequent venous thromboembolism and pregnancy-related morbidity.


Introduction
The clinical utility of thrombophilia testing has remained a subject of controversy since its introduction in clinical practice [1]. As guidelines of thrombophilia testing include only conditional recommendations, patterns of thrombophilia testing vary strongly across centers [2][3][4][5]. Considering the high cost of the work-up, testing is dependent on patients' and physicians' individual preference, financial status and the local healthcare system [6]. Moreover, the influence of test results on treatment decisions is still a matter of ongoing debate [1,4].
Data showing the clinical usefulness and benefits of positive thrombophilia testing results for further clinical decisions on treatment after venous thromboembolism (VTE) are limited. Garcia-Horton et al. showed the limited relevance of thrombophilia work-up in clinical decision-making after unprovoked VTE at a tertiary thrombosis center in Canada [7]. Given the growing body of evidence of the rather limited impact of thrombophilia upon the recurrence of VTE and overall mortality [8][9][10][11][12], current scientific data emphasize strict but variable selection criteria for thrombophilia testing [5,[13][14][15][16][17].
Since most studies on thrombophilia testing focus on accuracy and guideline interpretation [18][19][20], the quantification of its guidance on treatment decisions and impact on VTE outcome remains unclear. The potential negative effect of withholding anticoagulation or overtreating patients because of the work-up is being debated, whereas its significance is uncertain [1]. Thrombophilia work-up and its clinical consequences in patients with arterial thrombosis, pregnancy-related morbidities or asymptomatic thrombophilia carriers is even less defined [21][22][23][24] and clinical trials in this context are absent [25].
To investigate the impact of thrombophilia testing on the management and clinical course of thromboembolic disease and pregnancy-related morbidity, we conducted a large 10-year single-center retrospective cohort study. We analyzed the work-up patterns of thrombophilia and their impact on treatment decisions, and the subsequent occurrence of venous and arterial thrombosis and pregnancy-related morbidities.

Study Design and Patients
A single-center retrospective cohort study was conducted at the Department of Hematology of the Bern University Hospital in Switzerland between January 2010 and October 2020. We systematically screened consecutive patients referred for testing of hereditary and/or acquired thrombophilia by general practitioners or non-hematologist medical specialists with the support of the hospital data management service using internal specified codes for thrombophilia work-up. Patients with general consent were included if thrombophilia testing was performed and they had a documented history of objectively confirmed VTE and/or arterial thrombosis in any location, a history of pregnancy-related morbidity or were referred for thrombophilia testing due to a positive family history for VTE or hereditary thrombophilia.
The objective diagnosis of deep vein thrombosis (DVT), superficial vein thrombosis and muscle vein thrombosis was defined by a positive compression ultrasonography or venography [26], and pulmonary embolism (PE) was defined by a new high-probability ventilation/perfusion lung scan or a new contrast filling defect on spiral computed tomography (CT) or pulmonary angiography [26]. Arterial thrombosis was defined by the presence of stroke on brain magnetic resonance (MR) imaging or CT or by a diagnostic coronary angiography in patients with myocardial infarction [27,28]. A critical limb ischemia of a peripheral artery disease was established by arterial Doppler ultrasound, CT angiography, MR angiography or catheter-based arteriography [29]. Renal artery or vein thrombosis, splanchnic vein thrombosis, cerebral vein thrombosis and thrombosis of aorta and vena cava were defined as filling defects in the corresponding vessel on CT or MR angiography or venography [30][31][32]. Other types of thromboembolism, such as retinal vein or artery thrombosis, penis vein thrombosis, osteonecrosis or chronic inflammatory disease, along with pregnancy-related morbidity, were defined by relevant specialists using a referral report. Pregnancy-related morbidities were defined as pregnancy loss at all gestational ages, placenta failure, preeclampsia [33] and HELLP syndrome (hemolysis, elevated liver enzyme levels, low platelet count) according to obstetrical diagnostic criteria [34].
Clinical data were collected from structured electronic forms using a standardized case report form and entered into a computerized database (REDCap software) by two persons. Data comprised demographic characteristics of patients and their family history of VTE in first-and second-degree relatives, details of all previous thrombotic events or pregnancy-related morbidity (date and location), risk factors for most recent VTE and arterial thrombosis event (namely heavy smoking (>20 pack years), immobilization > 4 h, infections requiring bedrest > 3 days, estrogen-based medications, pregnancy and peripar-tum, intravenous catheters, active cancer, obesity (body mass index [BMI] > 30 kg m −2 ), trauma, surgery, cancer medication, presence of extended varicose veins, patent foramen ovale or other septal defect) and co-morbidities (diabetes mellitus, arterial hypertension, liver cirrhosis, kidney failure, rheumatic disease, depression, chronic inflammatory disease, dyslipidemia, cardiovascular diseases, pulmonary diseases, neurological diseases).

Thrombophilia Testing
Thrombophilia testing was performed between 3 and 6 months following the index event, after the evaluation of the patient by a hematologist, taking into consideration age, risk factors, family history of VTE, co-morbidities and type of thrombosis or pregnancyrelated morbidity. A thrombophilia work-up was considered as "performed" if one or more of the following thrombophilia parameters were tested: factor V Leiden (FVL) mutation status, prothrombin gene 20210G>A mutation status, protein C (PC) and antithrombin (AT) activity as well as free protein S (PS) antigen, lupus anticoagulant (LA), anticardiolipin antibodies and anti-β2-glycoprotein I antibodies. Only results of accurate thrombophilia testing were considered, excluding PC and PS testing whilst on vitamin K antagonists (VKA) or PS level during pregnancy.

Classification of Thrombophilia and Risk Factors
Categorization as minor and major provoking risk factors was based on the guidance provided by the International Society on Thrombosis and Haemostasis (ISTH) [36]. In addition to the ISTH-based criteria, the presence of an intravenous catheter [37] and May-Thurner syndrome (>70% iliofemoral compression) [38] were categorized as major risk factors, whereas immobilization > 4 h [39] and heavy smoking (>20 pack years) [40] as minor risk factors. VTE in the presence of merely an environmental risk factor (male sex and older age) was categorized as unprovoked thromboembolism.
Minor thrombophilia was defined as isolated heterozygous FVL or prothrombin 20210G>A mutation according to institutional guidelines. AT activity < 70%, PC activity < 69% and PS free antigen < 59%, antiphospholipid antibody syndrome, homozygous FVL or prothrombin 20210G>A or any compound thrombophilias were considered as high-risk thrombophilia.

Follow-Up and Outcomes
The primary study outcome was the influence of thrombophilia testing on anticoagulation management decisions. Secondary outcomes included the occurrence of first or recurrent VTE, arterial thrombosis or pregnancy-related morbidity after thrombophilia testing.
The influence of thrombophilia testing on management decisions was assessed from a structured medical report of each patient by two persons and classified as follows: (1) no influence on management; (2) appropriate management decision and (3) results not considered/overlooked, meaning a positive or potential positive influence; (4) decision to overtreat and (5) decision to undertreat, showing a negative influence ( Table 1). Because of the implementation of a structured reporting form at our center before 2010, considering clinical and laboratory factors leading to the choice of management, we were able to document clinical decisions on treatment that were merely based on thrombophilia. Therefore, management decisions based on clinical characteristics of thrombotic event (e.g., unprovoked or recurrent VTE), patient family history for VTE, preferences or high-bleeding risk were categorized as non-influential, irrespective of the thrombophilia result. We exclusively assessed the thrombophilia-based management decisions on prophylactic and therapeutic anticoagulation treatment, and did not consider other decisions or patient education, such as avoidance of estrogen-based treatment, change in type of anticoagulant or lifestyle modification. Guidance for the appropriateness of management decisions regarding thrombotic events, pregnancy-related morbidity or asymptomatic carriership of thrombophilia was based on international guidelines [16,17,41,42]. Data from all complete hospital records, including other disciplines, for the identification of subsequent VTE, arterial thrombosis or pregnancy-related morbidity after the thrombophilia consultation were investigated until March 2021. Only objectively confirmed events, according to the previously mentioned inclusion criteria, were considered. As 36 patients had a partial thrombophilia work-up before January 2010 in our clinic and were referred for a second time, the start of the follow-up time was defined by the first performed work-up from December 2004 to March 2021. For time-to-event analyses, we considered only patients with a follow-up >30 days and censored patients at the time of last hospital record or time of event.

Statistical Analysis
Continuous variables are presented as mean ± standard deviation (SD) and were compared using one-way ANOVA. Categorical variables are shown as percentages and compared with x 2 test. Univariable logistic regression models were used to assess the influence of thrombophilia on treatment decisions. As the negative result did not lead to any influence on treatment, heterozygous FVL mutation was used as a reference due to its lowest impact on treatment decision. Associations between low-and high-risk thrombophilia and the time to new thrombotic event or pregnancy-related morbidity were assessed using Cox proportional hazard models, yielding hazard ratios (HR) with their corresponding 95% confidence intervals (CI) and p-values for the failure event of primary interest. We adjusted the models for previously published predictors of venous and arterial thrombosis, including age > 50 years, male sex and risk factors and co-morbidities, such as smoking, diabetes mellitus, obesity, arterial hypertension, kidney failure, dyslipidemia, depression, chronic inflammatory disease and active cancer [43][44][45]. The Kaplan-Meier method was used to plot time from work-up for thrombophilia to recurrence of new thrombotic event or pregnancy-related morbidity. Only complete case analysis was performed, without an attempt to replace missing values with imputation methods. A value of p < 0.05 was considered statistically significant. All analyses were performed with R 4.1.1 and figures were edited with GraphPad Prism v9.1.2.

Prevalence of Thrombophilia in the Cohort Study
A total of 1260 thrombophilias were found in 1192 (34%) patients. The most common type of thrombophilia was heterozygous FVL mutation (714 patients, 20%), followed by heterozygous prothrombin 20210G>A mutation (193 patients, 5%) and antiphospholipid antibody syndrome (119 patients, 3%) ( Figure 2). One hundred and seven patients (3%) had more than one thrombophilia. Patients with thrombophilia were younger and were less likely to have co-morbidities or risk factors. More patients with thrombophilia were referred because of a recurrent VTE. Patients with arterial thrombosis had significantly less positive thrombophilia results compared to other patient groups, whereas women with pregnancy-related morbidities and asymptomatic patients with a positive family history were more likely to test positive (Table 2). PT, prothrombin. Testing was not performed or missing for presence of FVL mutation (6%), PT G20210A mutation (13%), antithrombin deficiency (20%), protein C deficiency (30%), protein S deficiency (29%) and antiphospholipid antibody syndrome (11%). Low-risk thrombophilia comprises heterozygous factor V Leiden or heterozygous prothrombin 20210G>A mutation. Highrisk thrombophilia comprises homozygous factor V Leiden mutation, homozygous prothrombin 20210G>A mutation, antithrombin < 70%, protein C < 69% and protein S < 59%, antiphospholipid antibody syndrome and compound thrombophilias.

Impact of Thrombophilia Testing on Treatment Decisions
In 3050 patients (86%), a thrombophilia work-up did not have any influence on the treatment decision, mostly when patients tested negative (2171 patients, 71%) or confirming low-risk thrombophilia (671 patients, 22%) ( Table 3). A total of 211 positive work-ups (6%) led to an appropriate decision to extend or initiate anticoagulation; 82 positive workups (2.2%) were inappropriately overlooked-21 (26%) with antiphospholipid antibody syndrome, 20 (24%) with heterozygous factor V Leiden mutation and 14 (17%) with heterozygous prothrombin 20210G>A mutation. Of 195 patients (5.4%) with an inappropriate treatment decision, 181 patients (93%) had a negative thrombophilia work-up. Only 11 patients (0.3%) had a positive work-up that led to overtreatment. Categorical variables are compared by x 2 test. Twelve work-ups could not be categorized due to unclear statement on treatment decision in the clinical report. Low-risk thrombophilia is defined by the presence of heterozygous factor V Leiden, heterozygous prothrombin 20210G>A mutation; high-risk thrombophilia comprises homozygous factor V Leiden mutation, homozygous prothrombin 20210G>A mutation, antithrombin < 70%, protein C < 69%, protein S < 59% and compound thrombophilias.
The presence of antiphospholipid antibody syndrome had the highest positive influence on treatment decision (71/119, 60%), followed by high-risk hereditary thrombophilia (86/247, 35%) (Table 3). However, only 17% positive results (136/826) for low-risk thrombophilia led to a change in treatment, and merely 116 out of 714 (16%) carrierships of a heterozygous FVL mutation provided further guidance (Supplemental Table S2  Abbreviations: OR, odds ratio. ORs are calculated by univariable logistic regression, using heterozygous factor V Leiden mutation as a reference due to its lowest influence on treatment decision.
None of the thrombophilias was associated with arterial thrombosis during follow-up ( Figure 3, Table 5). Notably, hereditary high-risk thrombophilia and antiphospholipid antibody syndrome were not only associated with a subsequent VTE during follow-up in the entire study cohort, but also with recurrent VTE, after the exclusion of patients with arterial thrombosis and pregnancy-related morbidity and asymptomatic patients (HR, 2.00; 95% CI, 1.12-3.57; and HR, 2.82; 95% CI, 1.28-6.29, respectively). Due to small sample size, a modification of y-axis scale was applied for presentation purposes. Lowrisk thrombophilia comprises heterozygous factor V Leiden or heterozygous prothrombin 20210G>A mutation. High-risk thrombophilia comprises homozygous factor V Leiden mutation, homozygous prothrombin 20210G>A mutation, antithrombin < 70%, protein C < 69% and protein S < 59%, antiphospholipid antibody syndrome and compound thrombophilias.

Discussion
In this large retrospective cohort study of 3550 patients performed at a single tertiary hematology department, we evaluated the impact of thrombophilia work-up and its result Due to small sample size, a modification of y-axis scale was applied for presentation purposes. Low-risk thrombophilia comprises heterozygous factor V Leiden or heterozygous prothrombin 20210G>A mutation. High-risk thrombophilia comprises homozygous factor V Leiden mutation, homozygous prothrombin 20210G>A mutation, antithrombin < 70%, protein C < 69% and protein S < 59%, antiphospholipid antibody syndrome and compound thrombophilias.

Discussion
In this large retrospective cohort study of 3550 patients performed at a single tertiary hematology department, we evaluated the impact of thrombophilia work-up and its result on treatment decisions, as well as the association between thrombophilia and thromboembolic events and pregnancy-related morbidity, in a real-world setting. In 86% of patients in this cohort, the result of the thrombophilia work-up did not lead to a change in treatment decision, while, in 5% of the cohort, patients' anticoagulation was withheld inappropriately because of a negative thrombophilia work-up. Positive work-ups for hereditary high-risk thrombophilia and antiphospholipid antibody syndrome had significantly more influence on treatment decisions compared to low-risk thrombophilia, and patients with these thrombophilias were more likely to develop subsequent VTE and pregnancy-related morbidity during follow-up.
Considering that the aim of our study was to demonstrate the utility of thrombophilia work-up in a real-world setting in a tertiary care center, the cohort reflected the expected clinical characteristics of patients pre-selected by general practitioners or other specialists. Therefore, the cohort comprised young patients with few co-morbidities or major risk factors for thromboembolism. The prevalence of thrombophilia was in line with previously reported cohorts of thrombophilia work-up in Europe [21,[46][47][48]. Data from our center show the limited clinical utility of thrombophilia work-up in reference to the length of anticoagulation in already pre-selected younger patients, which is in line with the findings of a few other studies in smaller, combined inpatient and outpatient VTE or arterial thrombosis cohorts from a tertiary non-hematologic care center [49,50] and tertiary hematology center [6]. Furthermore, an adverse outcome of the work-up could be demonstrated in terms of the withholding of indicated anticoagulant treatment in 5% of the cohort patients due to a negative thrombophilia work-up. Although a high-risk thrombophilia was only found in 10% of the cohort patients, it contributed to the treatment decision in 4.3% of cohort patients, providing a greater impact compared to low-risk hereditary thrombophilia. Besides continuous training to interpret the results of thrombophilia work-up, better-defined predictive factors for hereditary and acquired high-risk thrombophilia are needed to increase the utility of the work-up.
The presence of high-risk hereditary thrombophilia and antiphospholipid antibody syndrome was independently associated with subsequent and recurrent VTE and an-tiphospholipid antibody syndrome with pregnancy-related morbidity after the work-up. Although the thrombophilia risk for first VTE is well established [51], data on the recurrence of VTE are less clear. Multiple reports indicate no association of a positive thrombophilia result with the recurrence of VTE, mostly in low-risk hereditary thrombophilia [9,10,12], and heterogeneous results on recurrent pregnancy morbidity [52][53][54] with the exception of antiphospholipid antibody syndrome [55]. However, data on hereditary high-risk thrombophilia are limited and studies including the whole panel of thrombophilia are sparse. To our knowledge, this is the first study reporting on new and recurrent thrombotic events and pregnancy-related morbidity after the work-up, including all types of main thrombophilia and indications of testing from real-world practice. Therefore, it seems that despite the treatment change due to high-risk thrombophilia, it still has an impact on the clinical course of VTE and pregnancy-related morbidities and its clinical management should be further investigated.
Our study has several limitations. Firstly, the retrospective study design may have introduced selection and information bias due to the possible misclassification of primary outcomes. Nevertheless, because of the introduction of a structured reporting system and testing pattern in our center before 2010, and cross-evaluation of the data by two individuals, missing values and random errors were limited. Secondly, we could not consistently retrieve information on death, major bleeding and anticoagulation status at the time of subsequent thrombotic events and pregnancy-related morbidity due to insufficient documentation of these events and regulatory restrictions to search civil registries for death. Therefore, our findings do not allow the balancing of the risks and benefits of anticoagulant interventions and treatment decisions, and, moreover, analyses could not be adjusted for death as a competing event, which may lead to biased outcome rates. Thirdly, the definition of high-risk thrombophilia is not well established and may be debatable, especially in the context of pregnancy-related morbidity. Nevertheless, the study gives a new, comprehensive insight into the impact and outcome of thrombophilia work-up and its result. Fourthly, a very small proportion of patients with PS type II deficiency might have been missed, because no systemic measurement of PS activity was performed. Lastly, we evaluated the impact of thrombophilia on prophylactic and therapeutic anticoagulation treatment and did not consider other positive aspects of the work-up, such as a change in anticoagulant in reference to antiphospholipid antibody syndrome or any other high-risk thrombophilia, stronger motivation for lifestyle modification, avoidance of estrogen-based treatments or positive psychological effects of testing negative. Therefore, no conclusions can be made about the overall benefit-risk balance of the testing.
In conclusion, we found that the benefit of thrombophilia testing is limited in already pre-selected outpatients and has some adverse effects on the clinician's management of anticoagulation in all types of index thrombotic events and pregnancy-related morbidity. Better selection criteria to identify patients who may benefit from testing for hereditary and acquired high-risk thrombophilia are needed to improve the diagnostic and therapeutic yield of thrombophilia work-up and reduce the risk of inappropriate management decisions based on negative tests, the high costs of the investigations and the unfavorable impact on the psychological status of patients due to the results of unnecessary tests. Therefore, the clinical utility of the current selection criteria and the strongest factors associated with treatment should be investigated in order to establish better clinically oriented testing guidelines for thrombophilia work-up.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm11144188/s1, Table S1: Clinical characteristics of cohort patients in accordance with thrombophilia work-up result; Table S2: Influence of different thrombophilias on treatment decisions; Table S3: Clinical characteristics and prevalence of thrombophilia in cohort patients according to follow-up status. Data Availability Statement: There were no publicly archived datasets analyzed or generated during the study.

Conflicts of Interest:
T.T. is an investigator of the CanVECTOR Network; the Network receives grant funding from the Canadian Institutes of Health Research (CDT-142654). The remaining authors declare no competing financial interests. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.