Benchmark for Time in Therapeutic Range in Venous Thromboembolism: A Systematic Review and Meta-Analysis

Introduction The percentage of time within the target INR range 2.0 to 3.0 (TTR) in patients treated with vitamin K antagonists varies considerably among efficacy-studies of novel anticoagulants. In order to properly asses the quality of anticoagulant control in upcoming cost-effectiveness studies and real life registries this systematic review reports a benchmark of TTR for different treatment durations in patients with venous thromboembolism and discusses ways to calculate TTR. Methods Medline and Embase were searched for studies published between January 1990 and May 2012. Randomized controlled trials and cohort studies reporting the TTR in patients with objectively confirmed venous thromboembolism treated with vitamin K antagonists (VKA) were eligible. Duplicate reports, studies only reporting INR during initial treatment or with VKA treatment less than 3 months were excluded. Three authors assessed trials for inclusion and extracted data independently. Discrepancies were resolved by discussion between the reviewers. A meta-analysis was performed by calculating a weighted mean, based on the number of participants in each included study, for each time-period in which the TTR was measured since the confirmation of the diagnosis of VTE. Results Forty studies were included (26064 patients). The weighted means of TTR were 54.0% in the first month since the start of treatment, 55.6% in months 1 to 3, 60.0% in months 2 to 3, 60.0% in the months1 to 6+ and 75.2% in months 4 to 12+. Five studies reported TTR in classes. The INR in these studies was ≥67% of time in therapeutic range in 72.0% of the patients. Conclusion Reported quality of VKA treatment is highly dependent on the time-period since the start of treatment, with TTR ranging from approximately 56% in studies including the 1st month to 75% in studies excluding the first 3 months.


Introduction
Traditionally, patients with venous thromboembolism (VTE) are treated with low molecular weight heparins (LMWH) and vitamin K antagonists (VKA) such as warfarin, acenocoumarol or phenprocoumon [1,2]. As with any medical treatment, the weighing of risks and benefits must be carefully balanced. The effect of VKA therapy depends on many factors including variation in dose response between patients, individual variation in pharmacokinetics and pharmacodynamic response, multiple interactions with food, co-medication and finally also by variation in adherence [3,4]. VKA have a narrow therapeutic index, which needs to be monitored carefully in order to reduce the risk of tromboembolic events as well as bleeding complications [5]. With the large scale clinical testing of novel, direct acting oral anticoagulants, including the thrombin and factor Xa inhibitors dabigatran and rivaroxaban, a new era has been heralded. The main advantage of these new anticoagulants is the lack of a need for laboratory monitoring and dose adjustment due to more stable pharmacokinetics [6]. Several recent large randomized controlled trials have shown non-inferiority in effectiveness and safety of the new anticoagulants compared to VKA treatment [7,8,9,10,11]. However, the percentage of time within therapeutic range in the VKA-group, representing the quality of the control group, appears to vary considerably among these studies.
The International Normalized Ratio (INR), the ratio of a patient's prothrombin time to a normal (control) sample, raised to the power of the International Sensitivity Index (ISI) value, is established by the World Health Organization (WHO) and the International Committee on Thrombosis and Hemostasis for monitoring the effects of VKA. A target INR range of 2.0 to 3.0 is recommended for the treatment of VTE [3]. The most recognized way to measure the therapeutic effectiveness of VKA over time is to measure the percentage of time in the therapeutic range (TTR). TTR has been shown to strongly correlate with the clinical outcomes of hemorrhage or thrombosis and, thus, TTR is a reliable measure of the quality of anticoagulation management [12].
Dabigatran and rivaroxaban have been recently approved in many countries including the USA, Canada and also in Europe. This development will cause major changes in thrombosis management in the near future. Cost-effectiveness studies and real life registries will be the next step in the implementation of new oral anticoagulants. In order to adequately compare all treatment options, including novel anticoagulants and VKA, and to interpret the relative efficacy and safety of these novel anticoagulants, it is important to properly assess the quality of anticoagulant control, i.e. TTR, in the VKA group. This systematic review tries to provide a benchmark of TTR in patients with VTE receiving VKA and discusses the pros and cons of various ways to calculate TTR. Finally, it emphasizes the need to standardize TTR reporting, thereby contributing to a meaningful comparison among treatment options in studies evaluating novel anticoagulants.

Data sources and searches
A systematic search was performed to identify randomized controlled trials and cohort studies reporting the TTR in patients treated with VKA for deep vein thrombosis (DVT) confirmed by a non-compressible venous segment on an ultrasound of the extremities, or pulmonary embolism (PE) confirmed by an arterial filling defect on Computed Tomographic Pulmonary Angiography (CTPA) or a high probability ventilation/perfusion (V/Q) scan, or both (VTE). We searched Medline and Embase for articles in English, French, German, Dutch, Polish, Swedish, Danish, Italian and Spanish. Since the World Health Organization introduced the INR in 1983 [13] and the first studies reporting TTR in VKA in patients with VTE were published in the nineties, we searched for publications between January 1990 and May 2012. See Appendix 1 for detailed information about the search strategy and key words.

Study selection
To be eligible for inclusion, studies had to fulfill the following criteria: 1. Study population consisted of consecutive adult patients with objectively confirmed DVT or PE. 2. Patients were treated with VKA for a minimum of three months.
Studies were excluded if they only reported the TTR in the initial treatment period while patients were still on parental medication such as low molecular weight heparin and unfractionated heparin.

Data extraction and management
Three reviewers (PE, HTC, MP) operating in pairs of two extracted independently the following characteristics from each included study: study design, type of study (e.g. evaluation of a new drug, dose-finding, evaluation of duration of anticoagulation), characteristics of the study population (e.g. number of patients treated with VKA, country, inclusion criteria, proportion of patients with a malignancy), initial treatment, type of VKA (e.g. warfarin, acenocoumarol, phenprocoumon or other), initial dose of VKA, treatment duration, INR-monitoring by thrombosis

Data synthesis and analysis
A meta-analysis was performed by calculating a weighted mean, based on the number of participants in each included study, for each time-period in which the TTR was measured since the confirmation of the diagnosis of VTE.

Results of the search
The systematic search yielded 3636 citations. The results were screened and after reading titles and abstracts 3154 articles were excluded. Of the remaining 482 publications the full text was assessed. (Figure 1).

Discussion
A strong relationship between TTR and bleeding or thromboembolic rates has been observed across a large number of studies with different patient populations [53]. Since under-anticoagula-  tion gives inadequate protection against thromboembolic events and over-anticoagulation increases the bleeding risk, it is important to report the quality of VKA treatment by using the TTR [54]. The evidence for non-inferiority of new anticoagulants depends on the quality of the VKA control group. The present review provides a benchmark of TTR in patients with VTE receiving VKA and discusses the pros and cons of various ways to calculate TTR. We included 40 studies with more than 26000 participants and the results indicate that the achieved TTR ranges from approximately 56% to 75%.
The reported quality of VKA treatment was highly dependent on the time-period since the start of treatment. A statistically significant lower TTR was seen in studies reporting a TTR that covers all INRs, including the first month, compared to studies reporting the TTR without the first month. This difference is to be expected because of the difficulty to reach the therapeutic range in the initial treatment period and improvement in TTR during continuation of VKA treatment. Another explanation of the high TTR during longterm treatments is a selection-to-continue bias. Patients with stable INRs are more likely to continue their treatment with VKA than patients who experience problems in reaching the therapeutic range [55]. However, even after 4 to 12 months of treatment with VKA, patients spent 25% of their time outside of the therapeutic range.
Our review has some limitations that have to be mentioned. First, methods used to calculate TTR differed across the included studies. Fifty percent of the studies used linear interpolation, a few studies reported the percentage of time in a certain TTR class and 30% of the studies did not report the method of TTR calculation at all. Due to missing information about the exact calculation of TTR, we were unable to compare the different methods in a meaningful way. In literature, several methods to assess therapeutic control are described: e.g. the assessment of the number of INR measurements within the target range expressed as a percentage of the total number of INRs obtained, the cross-section-of the-files technique (the fraction of patients in range at one point in time compared to the total number of patients who had an INR at that point in time), equidivision, linear interpolation and the hybrid method [52,56,57]. Each approach has its advantages and disadvantages. A disadvantage of the first two methods is that they do not incorporate time and therefore cannot be used to calculate incidence rates of recurrences at different INR levels [54]. Time is incorporated in the method of equidivision, which assumes that the change between two consecutive INR measurements occurred halfway the interval [56]. The time spent in INR ranges can also be estimated by linear interpolation, which assumes that the INR between two measurements varies linearly from the first INR to the second INR [52]. A disadvantage of these last two methods is that extreme out of range INR values may bias overall results [58]. The hybrid method, in addition, takes effects of dosage modifications into account [54]. The results of all of these methods depend on whether an exact (INR 2.0-3.0) or an expanded therapeutic range is used, whether VKA-naïve patients (those just beginning therapy) are included or only patients already on established therapy, whether INRs obtained during invasive procedures when VKA therapy might be interrupted are excluded, and whether different oral anticoagulant preparations (e.g. warfarin or acenocoumarol) are allowed [53]. In a comparison of the equidivision, linear interpolation and hybrid methods, linear interpolation has been suggested as the preferred method as it shows a high validity and reproducibility [54]. We suggest that drug trials and real life registries with a VKA control group report the TTR in a uniform manner, to allow adequate comparison of data. Since linear interpolation has a high validity and was the most common method used to calculate TTR in the present review, we recommend to use linear interpolation in future studies covering the INRs from each patient from the discontinuation of heparin until the end of treatment. In order to avoid complex calculations, we believe that including time-periods with interruptions in VKA treatment in the TTR are acceptable. However, for calculating the relationship between TTR and adverse events, such as major bleeding episodes and thromboembolic events, we would suggest to exclude bridging periods, since the TTR will not represent the quality of anticoagulant treatment during these periods when most patients receive LMWH (Figure 3).
A second important limitation of the present review is that we were not able to investigate the association between TTR and clinical endpoints. Several studies in literature show a strong relationship between TTR and bleeding or thromboembolic events [53]. Unfortunately, data on such clinical endpoints related to TTR was not provided in the included studies.
Additionally, some other interesting sub-analyses were difficult due to small subgroups and the absence of detailed data. Hutten et al. indicated that the therapeutic quality of treatment was decreased when patients were treated with acenocoumarol rather than with warfarin [59]. This might implicate that the use of warfarin is preferable. However, since it is not clear whether these results might be influenced by factors such as frequency of monitoring and comorbidities, we need to be careful with drawing a conclusion. Furthermore, Hutten et al. showed that TTR was decreased in the presence of cancer and in the presence of a pulmonary embolism [59]. The same subgroup analyses in the present review did not show statistically significant results (data not shown). This might be explained by the fact that we did not have individual patient data (IPD). An IPD meta-analysis may give more detailed information for investigating such associations and may be interesting. Hutten et al. also showed a decrease in the therapeutic quality of VKA treatment when more than four changes in co-medication occurred [59]. Unfortunately such data was not available for our review.
The main conclusion of our systematic review is that the reported quality of VKA treatment is highly dependent on the time-period since the start of treatment, with the TTR ranging from approximately 56% in studies including the first month to 75% in studies excluding the first 3 months. The clinical consequences of our findings are not straightforward. However, it needs to be emphasized that the reported quality of VKA treatment should be taken into consideration while interpreting results from trials with new anticoagulants. Assuming an average treatment duration of 6 months, the mean TTR is approximately 60%. We recommend to calculate the TTR by using linear interpolation covering the INRs from each patient from discon-tinuation of heparin until the end of treatment. Furthermore, TTR is predictive of thromboembolic and bleeding complications for patients on VKA [53]; therefore a proper calculation of TTR in the VKA group is of importance in assessing the adequacy and quality of novel anticoagulants.
Oral anticoagulants are also effective in preventing stroke [60, 61,62,63,64] and prolonging survival rates in patients with atrial fibrillation (AF) [65]. It may be interesting to investigate a benchmark of the TTR in patients treated with VKA in AF in the near future. However, since patients with AF are usually on longterm VKA treatment, selection-to-continue bias will be more evident than in patients with VTE and should be taken into consideration in an analysis in AF patients [55].