Direct Oral Anticoagulants Versus Vitamin K Antagonists in Cerebral Venous Thrombosis: A Systematic Review and Meta-Analysis

We included interventional (randomized and nonrandomized) and observational (prospective and retrospective) studies of patients 18 years or older, published between January 1, 2007 and February 22, 2022. We excluded studies of fewer than 5 patients in each group and those with full texts not in English.

The intervention was DOAC treatment, defined as anti-Xa inhibitors, such as apixaban, rivaroxaban, and edoxaban, and direct thrombin inhibitors, such as dabigatran.

The control was VKA treatment.

Outcomes as defined in individual studies included: an efficacy outcome of subsequent VTE; safety outcomes of major hemorrhage (intracranial hemorrhage or extracranial hemorrhage), intracranial hemorrhage, and death; and a radiological outcome of full recanalization on follow-up venous imaging.

A comprehensive search was conducted by a health sciences librarian (C.M.) on May 26, 2021, in MEDLINE (via Ovid), EMBASE (via Embase.com), CINAHL (via EBSCOhost), and the following databases in the Web of Science Core Collection: Science Citation Index Expanded (SCI-EXPANDED), Social Sciences Citation Index (SSCI), Arts & Humanities Citation Index (A&HCI), and Emerging Sources Citation Index (ESCI). The MEDLINE search was peer-reviewed by another health sciences librarian using the PRESS checklist²¹ before being translated into the other databases. Search filters were used to limit the results to articles published since when DOACs were available, that is, between January 1, 2007 and December 31, 2020. Search hedges were used to exclude animal studies. Search terms for all databases include variations of CVT and warfarin. In addition, the search also includes terms for DOACs mentioned in a prior review²² to identify single-arm studies. The search was updated on February 22, 2022 to include all studies from January 1, 2021 to February 22, 2022.²³ The complete search strategy for all databases can be found in the Supplemental Material.

The unique records were imported into Abstrackr (http://abstrackr.cebm.brown.edu). Each abstract was independently screened by 2 of 7 investigators (B.P., S.Y., I.J.S., B.Z., A.S., K.J., and F.A.), with disagreements resolved by a third investigator. Potentially eligible abstracts were similarly screened using their full texts.

We extracted data into a standardized spreadsheet (in Microsoft Excel) to include participant age, sex, study follow-up duration, and results for each outcome for each study group. We did not contact study authors for missing information. Study data were extracted independently by 2 of 6 investigators (E.B., E.A.M., B.M., E.D.G., G.T., and A.D.), with disagreements resolved by a third investigator.

Each study was evaluated for risk of bias. The Cochrane Risk of Bias Tool 2.0 (ROB 2)²⁴ was used to evaluate RCTs and the Risk of Bias in Nonrandomized Studies (ROBINS-I) Tool²⁵ was used for observational studies. The ROB 2 tool assesses several domains including adequacy of the randomization process, deviations from intended interventions, missingness of outcomes data, measurement of the outcome, and selection of the reported result. The ROBINS-I tool assesses confounding, selection of participants, classification of intervention, deviations from intended intervention, missing data, measurement of outcomes, and selection of the reported result. Risk of bias was assessed by 2 investigators (B.M.G. or A.T.) who worked independently.

Publication bias across individual studies was graphically assessed for the outcomes of interest, using both funnel plot inspection and the Egger’s test.²⁶ We did this for outcomes reported by at least 10 studies; these tests are considered unreliable when there are fewer than 10 studies.²⁷

We summarized the evidence both narratively and, when feasible, quantitatively. Each study included in the systematic review was described in summary and evidence tables presenting study design features, study participant characteristics, descriptions of interventions, outcome results, and risk of bias/methodological quality. We used these characteristics and the I² values to help determine the appropriateness of meta-analysis.

We estimated a pooled treatment effect of DOACs (when compared with VKAs) primarily with risk ratios (RRs) and odds ratios (ORs) as data allowed. Where there are at least 3 studies reporting sufficiently similar results, we conducted pairwise meta-analyses (using random-effects models) comparing DOACs with VKAs. We explored and considered all forms of heterogeneity (clinical, methodological, and statistical) when considering pooling results. Regarding statistical heterogeneity, we avoided meta-analysis in the context of substantial statistical heterogeneity (ie, where I² values exceed ≈60%). We conducted subgroup analyses by type of study design (RCTs versus nonrandomized studies).

We identified 10 665 records and screened 254 of them as potentially eligible. We excluded 235 of these articles; the most common reasons for exclusion were that studies did not compare our treatments of interest (n=122 studies), full texts were not available (n=31 studies), and studies enrolled fewer than 5 patients per group (n=26 studies). We therefore included 19 studies—3 RCTs^17,28,29 and 16 observational studies^{12–16,23,30–39}—in the systematic review (Figure 1).

One of the 3 RCTs had some methodological concerns, whereas 2 of them are judged to be at high risk of bias, mainly due to bias from the randomization process but also due to deviations from intended interventions and bias in measurement of the outcome (Figure 2A).

All 16 observational studies had at least moderate risk bias; 5 had critical risk and 10 had serious risk of bias (Figure 2B). Observational studies were generally rated poorly for each of the 7 domains.

We evaluated publication bias only for the full recanalization outcome because it was reported by 12 studies. There was no evidence of asymmetry (Egger’s P=0.84; Figure S1).

The Table shows a summary of the included studies. We identified 3 RCTs^17,28,29 and 16 observational studies.^{12–16,23,30–39} Fifteen studies reported on recurrent VTE, 16 studies reported on major hemorrhage, 16 studies reported on intracranial hemorrhage, 14 studies reported on death, and 12 studies reported on recanalization. In all studies, all outcomes reported were not significantly different between the 2 groups, except for 1 study²³ where the risks of major hemorrhage and intracranial hemorrhage were lower with DOAC treatment. Furthermore, there were no events observed with both treatments in several studies, and these results are shown in Figure 3.

Recurrent VTE was assessed in 15 studies (2 RCTs and 13 observational studies; one study³⁰ only reported recurrent CVT). Meta-analysis suggested that when compared with VKA, DOAC was associated with a similar risk of VTE (relative risk [RR], 0.85 [95% CI, 0.52–1.37], P value for Cochran Q=0.81, I²=0%; Figure 3A).

Major hemorrhage during follow-up was reported in 16 studies (2 RCTs and 14 observational studies). When compared with VKA, DOAC was associated with a comparable risk of major hemorrhage (RR, 0.70 [95% CI, 0.40–1.21], P for Cochran Q=0.46, I²=0%; Figure 3B). The definition of major bleeding varied slightly across studies. For instance, 2 studies defined it as bleeding resulting in a reduction in hemoglobin of at least 2 g/dL or requiring 2 or more PRBC transfusion or bleeding in a critical organ,^17,33 one as nontraumatic symptomatic intracranial hemorrhage or bleeding resulting in the need for transfusion of 2 or more units of blood, the need for hospitalization or prolongation of an existing hospitalization, or death,¹³ and one as any intracranial hemorrhage or extracranial bleeding resulting in the need for transfusion or 2 or more units of blood.²³ In the other studies,^12,32,34,38 there was no definition provided.

Intracranial hemorrhage during follow-up was reported in 16 studies (3 RCTs and 13 observational studies). When compared with VKA, DOAC was associated with a comparable risk of intracranial hemorrhage (RR, 0.58 [95% CI, 0.30–1.12], P for Cochran Q=0.72, I²=0%; Figure 3C).

Death was reported in 14 studies (3 RCTs and 11 observational). The risk of death was comparable in DOAC and VKA groups (RR, 1.14 [95% CI, 0.54–2.43], P for Cochran Q=0.41, I²=1%; Figure 3D).

Venous recanalization on a follow-up imaging study was assessed in 12 studies (2 RCTs and 10 observational studies). Meta-analysis of these studies showed that the rate of complete recanalization was similar between the 2 groups (RR, 0.98 [95% CI, 0.87–1.11], P for Cochran Q=0.95, I²=0%; Figure 3E).

We performed subgroup analyses by study type (RCTs versus observational studies). As summarized in Figure 3A through 3E, the results did not seem to differ by study type.

In this systematic review and meta-analysis of studies of patients with CVT, DOAC, and VKA treatment seem to have a similar efficacy and safety profiles in terms of subsequent venous thromboembolism, major hemorrhage, intracranial hemorrhage, death, and complete recanalization.

These findings are consistent with data from studies comparing DOAC versus VKA in patients with peripheral VTE.^8–11 It is not surprising that DOAC therapy had a similar efficacy in terms of reducing VTE risk, which mirrors findings from RCTs in patients with deep venous thrombosis and pulmonary embolism.^8–11 The goal of anticoagulation in patients with CVT is to reduce the risk of subsequent venous thrombosis, reduce the risk of thrombus propagation, and allow the body’s endogenous fibrinolytic mechanisms to facilitate recanalization.

Advantages of DOAC treatment over VKA^40–42 are generally lower rates of intracranial hemorrhage and ease of use without the need for monitoring and frequent dose adjustment. In this systematic review, the rates of major hemorrhage were not significantly different between the 2 groups. One explanation of this finding is the inclusion of systematic hemorrhage in the major hemorrhage outcome, which did not differ between DOACs and VKAs in prior studies.^40–42 Another possibility is that intracranial hemorrhage in some patients may have been due to venous congestion as opposed to anticoagulation-related coagulopathy. Last, CVT occurs in younger patients than those with atrial fibrillation or deep venous thrombosis, and the risk of intracranial hemorrhage tends to be relatively low in younger patients regardless of treatment choice.

The evidence summarized in this systematic review has several limitations. First, and most importantly, our assessment of the studies suggested a high risk of bias, which could have impacted our findings. But given the rare occurrence of the disease, large and well powered RCTs are very challenging to perform, and thus observational studies are crucial; meta-analyses of these data can shed light on the comparative effectiveness across the 2 treatments. Second, the major hemorrhage outcome was not defined in the same manner across studies, which is a source of heterogeneity and limits the major hemorrhage analysis.

Our systematic review also has some limitations. First, we restricted to English-language articles and may have missed some studies published in other languages. Second, we excluded 31 of the 240 potentially eligible abstracts because we could not find the full-text articles for these studies. Furthermore, important limitations of this study include the low number of patients in RCTs, zero numbers of events in some studies, different DOACs used, and confounding by indication as well as unmeasured confounding in the observational studies.

This systematic review and meta-analysis suggest that in patients with CVT, DOACs, and warfarin may be comparable in terms of efficacy and safety. Given the limitations of the identified evidence, our findings should be interpreted with caution pending confirmation by ongoing RCTs and large, prospective, observational studies with adequate sample size and statistical power. The ongoing randomized SECRET trial (Study of Rivaroxaban for Cerebral Venous Thrombosis; https://clinicaltrials.gov; Unique identifier: NCT03178864) and the DOAC-CVT observational study (Direct Oral Anticoagulants in the Treatment of Cerebral Venous Thrombosis; https://clinicaltrials.gov; NCT04660747) are expected to provide additional high-quality evidence about the comparative efficacy of DOACs and VKA in patients with CVT.

Supplemental Materials and Methods

Figure S1

STROBE Checklist