Thrombophilia testing: does practice follow guidelines?

In this issue of Blood Advances, using the large, multicenter international Computerized Registry of Patients with Venous Thromboembolism (RIETE) of >100 000 patients with venous thromboembolism (VTE), Cohen et al¹ have investigated the impact of inherited thrombophilia on anticoagulant management and clinical outcomes. Approximately 20% of all patients were tested for inherited thrombophilia, and not surprisingly, clinical characteristics and VTE risk factors differed markedly between tested and untested patients with VTE. Testing for inherited thrombophilia was done more frequently in younger patients, patients with VTE provoked by estrogen use, pregnancy or postpartum, unprovoked VTE, and prior VTE, and less commonly in those with immobilization (≥4 days) and major risk factors such as surgery or active cancer, a pattern that has been recognized over the past decades.^2,3 

Patients who were tested for inherited thrombophilia were treated with anticoagulation for a longer duration (median, 216 days; interquartile range [IQR], 133-398) than patients who were not tested (median, 164 days; IQR, 97-22), and patients with inherited thrombophilia were treated longer (median, 244; IQR, 146-467) than those without inherited thrombophilia (median, 204; IQR, 125-373). After discontinuation of anticoagulant treatment, patients with inherited thrombophilia had a risk of recurrent VTE of 7.72 (7.04-8.45) events per 100 patient-years, whereas this risk was 6.58 (6.11-7.08) in those without inherited thrombophilia. To further assess the impact of inherited thrombophilia testing on clinical outcomes, subgroup analyses were conducted based on the categorization of VTE events into unprovoked VTE, VTE provoked by specific risk factors, and VTE at unusual sites, with the aim to reflect the classification framework that was used in the 2023 American Society of Hematology (ASH) VTE thrombophilia testing guideline.⁴ The authors conclude that some of their findings align, whereas others do not align with the conditional recommendations in the guideline. For instance, they showed that almost half of women with estrogen-associated VTE and with pregnancy-related VTE had been tested, and almost half actually had inherited thrombophilia. This indeed aligns with the ASH guideline assumption that ∼40% of people with a VTE would test positive for thrombophilia. Women with pregnancy-related VTE who were tested negative for inherited thrombophilia had much lower recurrence rates after treatment discontinuation (0.23 per 100 patient-years; 95% confidence interval [CI], 0.01-1.11) than women with inherited thrombophilia (2.27 per 100 patient-years; 95% CI, 1.05-4.31). In contrast, patients with immobilization-related VTE had similar recurrence rates after treatment discontinuation (∼6 per 100 patient-years) among tested patients with and without inherited thrombophilia.

How do the observations of clinical outcomes of tested vs untested and thrombophilia-positive vs thrombophilia-negative patients with VTE in the RIETE registry fit into the context of the ASH VTE thrombophilia testing guideline?

The ASH guideline used a modeling approach to address 23 clinical questions.⁴ For patients with VTE, conditional recommendations for thrombophilia testing and subsequent continuation of anticoagulant treatment in those who had thrombophilia were given only in the following scenarios: (1) for patients with VTE associated with nonsurgical major transient or hormonal risk factors; and (2) for patients with cerebral or splanchnic venous thrombosis, in settings in which anticoagulation would otherwise be discontinued. For all other scenarios, that is, unprovoked VTE as well as VTE provoked by major surgery, the panel provided conditional recommendations against testing for thrombophilia. Modeling was performed using the prevalence of thrombophilia in a certain population (eg, patients with VTE) and risk association data (eg, the association of thrombophilia with recurrent VTE) to calculate the absolute risk of events in people with and without thrombophilia. From other ASH VTE guidelines,⁵ the effect of anticoagulant treatment was used to make the final step, on which recommendations were based: the absolute number of events in 1000 patients not undergoing thrombophilia testing and using anticoagulant treatment as per general VTE treatment guidelines, compared with the number of events in 1000 patients undergoing thrombophilia testing with continuing or stopping anticoagulant treatment based on the thrombophilia test result.

The RIETE registry is a rich resource of clinical practice data on patients with VTE from a large number of centers and countries that started collecting data in 2001 and thus span a time frame of >20 years. It could be a valuable source of information about how many patients would be affected by applying the ASH VTE guidelines. However, its observational design also comes with limitations, as is clear from the current analysis on impact of inherited thrombophilia testing. First, comparing rates of recurrent VTE between patients with and without inherited thrombophilia provide only 1 part of the model that was used in the ASH VTE thrombophilia guideline. The authors have provided crude recurrence rates in tested vs untested patients with various types of VTE after stopping anticoagulant treatment. In the example of women with pregnancy-related VTE, the rates of recurrence in the tested women (1.13 per 100 patient-years; 95% CI, 0.55-2.07) appeared indeed lower than that in women who had not been tested (2.53 per 100 patient-years; 95% CI, 1.23-4.63).¹ However, the data do not allow to analyze whether anticoagulant treatment was adjusted based on the test result, and many women with inherited thrombophilia had in fact discontinued anticoagulation. Second, comparing a tested vs an untested population in most clinical VTE scenarios is difficult. As the authors point out, clinical outcomes including mortality were poorer in untested patients with VTE than in those who were tested, which is likely due to the higher age, presence of comorbidities such as active cancer, and the need to undergo major surgery. Although many prognostic variables are captured in the registry, confounding of unknown or unmeasured characteristics that consciously or unconsciously affect the clinical decision-making of physicians treating the included patients plays a major role. Therefore, drawing conclusions on causal inference about the impact of thrombophilia testing (or for that matter, any other intervention) in the RIETE registry is methodologically problematic. Direct evidence on effects of thrombophilia testing would arise from comparing a testing strategy followed by treatment adjustment based on the test result vs not testing for thrombophilia and treating all patients according to current guidelines, which is difficult to achieve and still lacking.⁶ We previously showed that mere testing for thrombophilia in VTE patients did not appear to affect recurrence rates.⁷ This case-control study was lacking crucial information, including thrombophilia test results and, similar to the RIETE registry, information on how thrombophilia testing had affected clinical decisions. What is really needed is a randomized controlled trial comparing an anticoagulant strategy based on thrombophilia testing with or without other clinical or laboratory variables, such as the currently recruiting Leiden Thrombosis Recurrence Risk Prevention study.⁸ 

In conclusion, clinical practice data in patients with VTE confirm that testing for inherited thrombophilia is being performed widely. Guideline recommendations are yet based on indirect evidence, and whether and how thrombophilia testing should inform clinical decisions still requires further evidence to be generated.

Conflict-of-interest disclosure: S.M. and A.I. were chair and vice-chair of the American Society of Hematology 2023 Guidelines for Management of Venous Thromboembolism: Thrombophilia Testing. The authors declare no competing financial interests.