Postthrombotic syndrome and quality of life after deep vein thrombosis in patients treated with edoxaban versus warfarin

Abstract Background Postthrombotic syndrome (PTS) is a long‐term complication after deep vein thrombosis (DVT) and can affect quality of life (QoL). Pathogenesis is not fully understood but inadequate anticoagulant therapy with vitamin K antagonists is a known risk factor for the development of PTS. Objectives To compare the prevalence of PTS after acute DVT and the long‐term QoL following DVT between patients treated with edoxaban or warfarin. Methods We performed a long‐term follow‐up study in a subset of patients with DVT who participated in the Hokusai‐VTE trial between 2010 and 2012 (NCT00986154). Primary outcome was the prevalence of PTS, defined by the Villalta score. The secondary outcome was QoL, assessed by validated disease‐specific (VEINES‐QOL) and generic health‐related (SF‐36) questionnaires. Results Between 2017 and 2020, 316 patients were enrolled in 26 centers in eight countries, of which 168 (53%) patients had been assigned to edoxaban and 148 (47%) to warfarin during the Hokusai‐VTE trial. Clinical, demographic, and thrombus‐specific characteristics were comparable for both groups. Mean (SD) time since randomization in the Hokusai‐VTE trial was 7.0 (1.0) years. PTS was diagnosed in 85 (51%) patients treated with edoxaban and 62 (42%) patients treated with warfarin (adjusted odds ratio 1.6, 95% CI 1.0–2.6). Mean differences in QoL scores between treatment groups were not clinically relevant. Conclusion Contrary to our hypothesis, the prevalence of PTS tended to be higher in patients treated with edoxaban compared with warfarin. No differences in QoL were observed. Further research is warranted to unravel the role of anticoagulant therapy on development of PTS.


| INTRODUC TI ON
Postthrombotic syndrome (PTS) is a long-term complication of deep vein thrombosis (DVT) that is reported by 20%-50% of patients. 1 PTS is a chronic condition that includes a variety of signs and discomforting symptoms of the leg that persists after the acute phase of DVT. Patients with PTS can report chronic pain, (nocturnal) cramps, tingling, and heaviness of the leg.
Furthermore, the appearance of the leg can be affected by swelling, hyperpigmentation, redness, and venous ectasia. Severity of signs and symptoms vary among patients and PTS is considered severe when venous insufficiency leads to ulcers and disability. 2,3 Generally, severe PTS comprises 5%-10% of the cases, but rates up to 22% have been reported. [4][5][6] Although the pathogenesis is not completely understood, PTS is thought to be the result of impaired thrombus resolution; persistent venous obstruction could cause venous hypertension and valvular reflux leading to impaired microcirculation. [7][8][9] The damage to the veins is irreversible in most cases. The burden of PTS has a significant impact on quality of life (QoL) and is associated with socioeconomic consequences. 3,[10][11][12][13] Self-reported physical QoL scores of patients with severe PTS are comparable to scores of patients suffering from angina, cancer, or congestive heart failure. 14 At present, therapeutic options are limited and, therefore, prevention of PTS is crucial. Studies on catheter-directed thrombolysis failed to consistently show a reduction on PTS [15][16][17] and evidence on the efficacy of elastic compression stockings (ECS) for PTS prevention is conflicting. Accordingly, current guidelines do not explicitly recommend the routine use of ECS. [18][19][20][21] Nevertheless, ECS may help to diminish PTS-related symptoms, leading to an improved QoL.
Given the relation of PTS to residual clot burden, preventive measures should focus on quality of anticoagulation. 22,23 Inadequate anticoagulant therapy with vitamin K antagonists (VKAs) (i.e., subtherapeutic international normalized ratio [INR]) is a known risk factor for the development of PTS. [24][25][26][27] In the past decade, VKAs have been replaced by direct oral anticoagulants (DOACs) as drug of first choice for most patients with an acute venous thromboembolism (VTE). 21 DOACs are pharmacologically more stable than VKAs and do not require dose adjustments or intensive therapeutic drug monitoring. Hence, quality of anticoagulation may be higher for DOACs than with VKA, particularly in the days to weeks after discontinuation of the heparin lead-in in patients with acute VTE.
Several studies, both randomized and observational, suggested a moderate reduction of PTS associated with rivaroxaban, [28][29][30][31][32][33][34][35] but not with dabigatran. 36 Edoxaban, a direct oral factor Xa inhibitor, is one of the DOACs that was proven as effective and safer with regard to serious bleeding in comparison with the VKA warfarin for treatment of acute VTE. 37 In this study, we aim to assess the long-term prevalence of PTS and health-related quality of life in patients treated for DVT with edoxaban or warfarin. We hypothesize that patients treated with edoxaban have a lower prevalence of PTS and a better health-related QoL compared with patients treated with warfarin for their index DVT.

| Study design and population
The source population for this cohort study consisted of the participants of the Hokusai-VTE trial (NCT00986154). 37 The Hokusai-VTE trial was an international, randomized, double-blind, noninferiority trial evaluating the efficacy and safety of edoxaban (30 or  An overview of the study timeline is presented in the Appendix S1, Figure S1.

| Study outcomes
The primary outcome of this study was prevalence of PTS. The secondary outcomes included long-term disease-specific and generic health-related QoL. The item that is not included in either score concerns intensity of symptoms over the day. Responses for each item are rated on a Likert response scale (2-point to 7-point), higher scores indicate better QoL. To be able to easily compare scores of different scales, raw scores were first transformed to Z-score equivalents (mean, 0; standard deviation [SD], 1), and then transformed to T-scores (mean, 50; SD, 10). 41 We considered a mean difference of 4 points in the VEINES-QOL T-scores between the two treatment groups clinically relevant. 46,47 To evaluate the generic health-related QoL regardless of any underlying disease, version 1 of the Short Form Health Survey questionnaire (SF-36) was used. 43,44 This questionnaire comprises 36 items in which the general well-being during the previous 30 days was assessed. It contains eight subscales: physical functioning, social functioning, physical role functioning, emotional role functioning, mental health, vitality, bodily pain, and general health. In addition to these eight domains, physical and mental summary scores were calculated. Scores are expressed on a 0-100 scale, with higher values indicating better general well-being. We considered a mean difference of 10 points in the SF-36 score between the two treatment groups clinically relevant. 48

| Statistical analysis
Statistical analysis was conducted according to the statistical analysis plan for the Hokusai post-VTE studies and therefore similar to the Hokusai post-PE study. 38 Differences in baseline characteristics between both groups were evaluated by means of the 2-sample t tests for normally distributed variables and the Mann-Whitney test for skewed distributions; χ 2 tests were applied for categorical data.
The association between the occurrence of PTS (yes/no) and treatment group (edoxaban, warfarin) was assessed using logistic regression analysis. We evaluated the association between QoL We performed a sensitivity analysis in the subgroups of patients who were unaware of treatment allocation during the Hokusai-VTE trial when participating in the Hokusai PTS study, and patients who stopped versus continued using anticoagulants after discontinuation of the Hokusai-VTE trial until inclusion in the Hokusai PTS study.

| Data sharing statements
For original data, please contact i.m.bistervels@amsterdamumc.nl.

| RE SULTS
Between April 2017 and September 2020, 316 patients were included in the Hokusai PTS study. Figure 1 represents the flowchart of patient inclusions in this study. The Hokusai-VTE trial included 8292 patients in 439 centers in 37 countries. Of these randomized patients, 5735 were diagnosed with a DVT and used at least one dose of assigned study treatment. 37 During site selection for the current study, 78 centers (1530 Hokusai-VTE trial patients) were approached. Of these centers, 52 centers (740 Hokusai-VTE trial patients) were not eligible for participation for logistical reasons (no ethical approval obtained, unable to approach participants, no study staff available). In the remaining centers, 474 of potentially eligibly 790 patients were excluded due to refusal of or inability to ask for consent for participation (n = 415), lost to follow-up (n = 38), or death (n = 21), leaving 316 (5.5%) patients from 26 centers in eight countries for inclusion. The included countries were Australia, Belgium, Canada, France, Germany, Italy, Norway, and the Netherlands.

| Patients and treatment
The mean time (SD) from randomization in the Hokusai-VTE trial to inclusion in the Hokusai PTS study was 7 (1)   Appendix S1 shows the corresponding baseline characteristics for all patients with DVT included in the Hokusai-VTE trial, for patients from study sites that were not invited for the participation    Table 3). The adjusted mean Mental Component Score difference was comparable.

| Quality of life
In Figure 2A, the scores of the eight dimensions of the SF-36 are displayed for patients treated with edoxaban and patients treated with warfarin. Crude mean differences were smaller than the minimal clinically important difference for all domains and were not statistically significant ( Figure 2B). The adjusted mean difference for the physical functioning domain was −5.6 and statistically significant (95% CI −10.7 to −0.6). A detailed overview of the crude and adjusted differences in SF-36 QoL domains are depicted in the Appendix S1 (Table S1; Table S1 for included variables in the adjusted models).
Sensitivity analyses in patients who were unaware of the allocation of treatment during the Hokusai-VTE trial (n = 217) (Appendix S1, Table S1), and in patients who stopped (n = 161) and who continued (n = 127) using anticoagulants after discontinuation of the Hokusai-VTE trial until inclusion in the Hokusai PTS study (Appendix S1, Table S1), did not show any large differences in comparison to the analysis in all 316 patients. Villalta score of 5 or higher or presence of venous ulcer, at least 6 months after DVT. c Adjusted by variables from model 1 (clinical reasoning) as described in the methods section, model 1 included age, sex, body mass index, thrombotic history, cardiovascular disease, concomitant medication use, awareness of randomized treatment, thrombus location, and duration of anticoagulant use. Details on the included variables per model are presented in the Appendix S1. d Adjusted by variables derived by model 2 (p value <0.25) as described in the methods section, model 2 included sex, cardiovascular disease, chronic analgesic use, and concomitant antiplatelet therapy. Details on the included variables per model are presented in the Appendix S1.

| DISCUSS ION
Contrary to our hypothesis, we observed that 7 years after enrollment in the Hokusai-VTE trial patients with acute, symptomatic DVT who had been treated with edoxaban tended to have a higher risk for PTS compared with patients treated with warfarin. This was driven by a larger proportion of mild and moderate PTS because there was no difference in the prevalence of severe PTS between both treatment groups. No clinically relevant differences were observed in long-term disease-specific and generic health-related QoL between treatment groups. This is the first study to evaluate the PTS prevalence in patients treated with edoxaban versus warfarin. In two similarly designed studies (i.e., follow-up of patients who previously participated in randomized controlled trials comparing DOACs with VKA), no statistical significant differences were found between treatment groups. 35 49 It has to be mentioned that predominantly nonrandomized cohort studies were included in this meta-analysis and that older historical cohorts were used as control in two studies. 31,33 Furthermore, follow-up duration often was <5 years and varied be- There are several potential explanations for our findings.
Because our study population consisted of a subgroup of the original trial population, confounding owing to between-group differences could have obscured an association between treatment and PTS. To assess the robustness of the association we decided to adjust for possible (available) confounders in different models; one model for which confounders were selected based on clinical reasoning and one model for which confounders were selected based on p values.
Both models showed an adjusted OR of 1.6; one was statistically significant, whereas the other was not. It is possible that in a wellcontrolled clinical trial setting in which patients adhered to protocol that included drug accountability, INRs were more often in the therapeutic range than they would have been in clinical practice, and advantages of the stable anticoagulant effect of edoxaban over warfarin may be obscured. We also evaluated disease-specific and generic health-related QoL in this study. Even though we did find a trend for a higher frequency of PTS in the edoxaban treatment arm, we did not observe clinically relevant differences in disease specific or in generic health-related QoL between groups. This may be explained as a larger proportion of moderate PTS drove the observed PTS difference, whereas the proportion of severe PTS was similar in both groups.
Strengths of our study are the randomized, blinded design of the treatment of acute VTE and the thorough assessment of different health aspects including original Villalta score assessed by trained physicians and both disease-specific and generic health-related QoL assessment. However, we acknowledge that the small sample of included patients in this follow-up study is the most important limitation leading to potential selection bias. Selection bias might cause both over-and underestimation of the prevalence of PTS. Especially after 7 years since randomization, patients with PTS, leg symptoms, and/or other comorbidities would be more likely to have up-to-date contact details in their medical charts and might be more willing to visit the hospital. This might also explain the somewhat high prevalence of PTS as compared with 20%-50% reported in literature 1 and the percentages reported in a study with shorter duration of follow-up. 35 On the other hand, patients with extensive DVT might have suffered from underlying disease that was, at the time of randomization in the Hokusai-VTE trial still unknown, and are therefore not represented in our study population. This could explain the lower proportion of common femoral and iliac DVT in our sample compared with the original Hokusai-VTE trial. Differences between our sample and the Hokusai-VTE trial population might lead to limitations in generalizability. Moreover, no information on treatment allocation of patients who declined participation was available and we were unable to account for the competing risk bias of mortality, which poses a risk for attrition bias. Furthermore, because of the observational nature of this follow-up study, unmeasured confounding (e.g,. history of ipsilateral DVT or a Villalta score at baseline) could have affected the findings. We attempted to control for confounding by adjusting for a wide range of potential confounders and explored the associations by means of two different multivariable models.
However, unmeasured confounding cannot be fully excluded. In addition, we evaluated the association between the occurrence of PTS and treatment group by means of logistic regression analyses. Last, the high drug adherence rates in this study could limit the generalizability of our results for daily clinical practice.
Clinical implications of our findings are limited. The lack of difference in PTS prevalence and QoL outcomes between treatment arms indicates that DOAC therapy does not necessarily improve long-term sequelae compared with well-managed VKA therapy. PTS remains a common complication after DVT, regardless of type of anticoagulant used to treat DVT, and our study suggests that choosing edoxaban over warfarin to treat acute DVT should not be based on long-term outcomes. Future studies should focus on assessing PTS prevalences in patients included in clinical practice data studies with associated drug adherence rates, such as data from the ETNA-VTE-Europa study. 50

| CON CLUS ION
Contrary to our hypothesis, the observed prevalence of PTS tended to be higher in patients with acute DVT treated with edoxaban compared with warfarin. Health-related QoL did not differ between treatment groups. Further research is warranted to unravel the role of anticoagulant therapy on thrombus resolution and development of PTS.