Crossing the Periprocedural Bridge in Patients with Atrial Fibrillation

A 76 year-old male with a past medical history of hypertension, diabetes, and atrial fibrillation on warfarin management is undergoing elective surgery that requires warfarin to be held. The patient’s CHADS₂ score is 3 with no history of stroke. Should this patient be bridged with low molecular weight heparin (LMWH)? For over two decades LMWHs have been routinely used to provide therapeutic coverage in patients who must temporarily stop warfarin.¹ Current guidelines suggest using injectable anticoagulants during warfarin interruption (aka bridging) in patients with atrial fibrillation based on patients’ risk of arterial thrombosis. Using the CHADS₂ score to assess risk, the guidelines recommend (grade 2C) bridge therapy if the CHADS₂ score is 5 or higher and not bridging if the CHADS₂ score is 2 or lower.² But what about patients with a CHADS₂ score of 3 or 4?

Observational and non-randomized studies have found a higher rate of bleeding but no reduction in the risk of thromboembolism when anticoagulation bridging has been used compared to no bridging.^3-5  The Outcomes Registry for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF) study reported a threefold higher risk of bleeding in patients who received bridging anticoagulation compared to those who did not.⁵ Moreover,  there was a higher incidence of myocardial infarction, stroke or systemic embolism, major bleeding, hospitalization, or death within 30 days in those patients who were bridged.  These results are obviously disconcerting but as a registry study, the ORBIT-AF data cannot prove causality.

The BRIDGE study is the first randomized, double blind, placebo controlled trial of sufficient power to address these important questions:  Is withholding warfarin perioperatively noninferior to bridging with LMWH for the prevention of arterial thromboembolism?² And does bridging increase major bleeding risk?  Enrolled patients had atrial fibrillation or atrial flutter, at least one CHADS₂ risk factor, an INR target 2-3, at least 3 months of treatment with warfarin, and were undergoing an invasive procedure. The primary outcomes included arterial thromboembolism, defined as stroke (ischemic or hemorrhagic), transient ischemic attack, systemic embolism at 30 days, and major bleeding. Secondary outcomes included acute myocardial infarction, deep-vein thrombosis, pulmonary embolism, death, and minor bleeding.

Patients were randomly assigned to receive either dalteparin 100 IU/kg twice daily or placebo. Warfarin was stopped five days before the procedure and dalteparin or placebo was initiated three days before the procedure. The last dose of dalteparin or placebo was given in the morning approximately 24 hours prior to the procedure. Warfarin was restarted at the patient’s usual dose on the evening of or day after the procedure. Dalteparin or placebo was resumed 12-24 hours following minor procedures and 48-72 hours after major procedures. A standardized classification was used to define major and minor procedures (see Table 1); however, the managing physician ultimately determined when the injectable anticoagulant was restarted after the procedure. Patients were contacted weekly for follow up until the final encounter 30-37 days after the procedure.

A total of 1884 patients were randomized between July 2009 and December 2014 with 950 assigned to placebo and 934 assigned to dalteparin.  The majority of patients were male, Caucasian, and the average CHADS₂ score was 2.3 (See Table 2). The majority of the patients had a CHADS₂ score less than 2 and underwent minor procedures. It should be noted that current clinical practice recommend forgoing bridging therapy in these patients.

Table 1: Type of surgery/procedure classification

Table 2: CHADS₂ scores of Patients Enrolled in BRIDGE study

Bridging with LMWH was noninferior to non-bridging for preventing arterial thromboembolism; however, the incidence of major bleeding was significantly higher in the bridging group (see Table 3). There were no significant differences in the secondary outcomes between groups, except for minor bleeding. The incidence of minor bleeding was 12% in the non-bridging group and 20.9% in the bridging group (P<0.001). The number needed to harm for major bleeding is 53 Therefore, it takes 53 patients to receive LMWH in order for one major bleeding event to occur. The calculated number needed to harm adds perspective to the increased risk of bleeding with the use of LMWH and emphasizes that the risk of bleeding outweighs the benefit .  The authors conclude that omitting bridge therapy is noninferior to bridging with dalteparin for the prevention of arterial thromboembolism and has a lower risk major bleeding risk.

Table 3: Primary Outcomes of the BRIDGE Study

The BRIDGE study has several limitations. High risk patients (CHADS₂ scores of 4-6) and high bleeding risk procedures were underrepresented. Over 500 patients were excluded before enrollment due to physician preference.  Many physicians might have been uncomfortable enrolling high risk patients in a placebo controlled study.  Thus, the results of this study may have been distorted by selection bias. Additionally, a majority of patients had CHADS₂ scores less than 2, representing a low-risk group that many clinicians would not bridge in clinical practice. Although the types of procedures included in the study are commonplace, some of these procedures might not warrant withholding warfarin. Another limitation is how medication adherence was assured; the authors do not report how they determined whether or not the study drug was actually taken. A final limitation is the risk of dosing error when self-administering a drug drawn from a multi-use vial. This creates some uncertainty as to the dose actually taken.

The study strengths include: specific criteria of the diagnosis of stroke, TIA, and systemic embolism that are consistent with previous studies and guidelines; clear definitions of high and low risk surgeries consistent with ACCP guidelines; and stopping the vitamin K antagonist prior to the procedure consistent with ACCP guidelines. Additionally, the trial was a randomized and placebo-controlled, thus reducing the risk for bias.

This study provides compelling evidence that periprocedural bridging with a LMWH may cause more harm than benefit in patients with arterial fibrillation who are at low to moderate risk for arterial thromboembolism. However, patients at high risk for arterial thromboembolism may still benefit from perioperative bridging and additional trials are need to evaluate the potential risks and benefits in this patient population. PERIOP 2 is an ongoing trial that includes patients with mechanical valves and atrial fibrillation that hopefully will provide us with more insights.⁶ Based on the BRIDGE study, crossing the periprocedural bridge with a LWMH in patients with atrial fibrillation who are at low to moderate risk of stroke undergoing low risk procedures is just not worth it.