Time for a Relevant Randomized Controlled Trial of Vena Cava Filters

To the Editor:

Members of the Prévention du Risque d'Embolie Pulmonaire par Interruption Cave (PREPIC) Study Group originally published their results in 1998¹ and reported their 8-year follow-up in 2005,² both of which have been widely cited. The PREPIC study demonstrated that although vena cava filters (VCFs) facilitated a decrease in pulmonary embolism (PE), patients with filters had a concurrent increase in deep venous thrombosis (DVT) without a survival benefit. A journal citation report through July 2012 on Web of Science indicates that the PREPIC study, as the only randomized controlled trial of VCFs, has been cited 601 times in the literature and thus has likely impacted the use of VCFs for venous thromboembolism (VTE) in the United States. However, it is critically important for physicians to understand the shortcomings of the study.

The PREPIC study was composed of patients who would not have received a VCF in the United States. The investigators inserted VCFs in patients with active VTE who were receiving therapeutic anticoagulation. At the start of the PREPIC study, the third edition of the American College of Chest Physicians consensus statement, Antithrombotic Therapy for Therapy of Venous Thromboembolic Disease, recommended VCFs for patients with DVT or PE with contraindications to therapeutic anticoagulation or recurrent emboli despite anticoagulation, not in addition to anticoagulation.³ The ninth edition of the consensus statement clearly recommends against VCF use in patients who can instead receive therapeutic anticoagulation; however, VCF use may be appropriate for patients with recurrent PE while on therapeutic anticoagulation.⁴ As such, recurrent PE and, especially, fatal PE in patients receiving therapeutic anticoagulation are uncommon. Therefore, a decrease in mortality would not be expected.

Recurrent DVTs are expected to be more prevalent in patients with chronic thrombophilic conditions, such as advanced malignancies, compared with patients with reversible (or provoked) conditions, such as traumatic injuries. The PREPIC study did not stratify patients in this manner, and its data diverge from that of other investigations, albeit not randomized controlled trials. For example, although VCF use in patients with cancer is not clearly defined, Schunn et al⁵ documented that 134 (2.2%) of 5970 patients with cancer, contraindications to anticoagulation therapy, and subsequent VCFs developed VTE. Researchers at MD Anderson Cancer Center reported a 1.3% PE rate and a 4.5% caval thrombus rate among 308 patients with a VCF.⁶ A Memorial Sloan-Kettering Cancer Center report⁷ showed a 2% rate of recurrent PE and a 6% rate of recurrent DVT in cancer patients with a VCF. Patients with all stages of malignancy were included in these analyses. Notably, recurrent DVT and PE rates in these patient groups are much lower than reported in PREPIC.

It is well known that subtherapeutic warfarin puts patients at an increased risk for recurrent VTE. The PREPIC investigators reported that warfarin was prescribed for 91% of their patients at discharge with 94% receiving warfarin at 3 months.¹ At 2 years, 38% of both groups were still receiving warfarin. Despite receiving anticoagulation therapy, 20.8% of patients with VCFs and 11.6% without VCFs had developed recurrent DVT at 2 years.¹ At 8 years, 35.7% of patients with VCFs and 27.5% without VCFs developed recurrent DVT.² This finding contrasts with findings in a study by Billet et al,⁸ in which no statistically significant differences in DVT rates were found among patients with and patients without VCFs. Furthermore, the documented percentage of time that patients were in the therapeutic range for the international normalized ratio was just over 50%.⁸ There was no evaluation of time in therapeutic range among patients enrolled in the PREPIC study. Could the differences in the DVT rates among PREPIC study patients be potentially related to subtherapeutic warfarin therapy?

The most important shortfall of the PREPIC study is extrapolating the use of permanent devices to newer retrievable or “optional” VCFs. Of the filters used in the PREPIC study, the Cardial filter (C.R. Bard Inc, Saint-Etienne, France) has not been approved for use in the United States. The VenaTech LGM filter (B. Braun Melsungen AG, Boulogne, France) was used in 56% of PREPIC patients but has shown a long-term progressive decrease in caval patency to 66.8% at 9 years; this finding was unaffected by age, sex, level of DVT, risk factors, or anticoagulant use.⁹ Post-VCF DVT was not analyzed by filter. Was the increase in DVT the result of using “inferior” VCFs? Would the retrieval of an optional device after the high-risk period for PE passes reduce long-term DVT rates?

For patients with contraindications to therapeutic anticoagulation, use of VCFs in the setting of an acute VTE would be appropriate. Importantly, all VCFs are prophylactic because the purpose of insertion is to prevent further PEs and not to manage an already present PE or DVT. Patient accrual for the PREPIC study occurred before the advent of the study and widespread use of retrievable VCFs. However, the rate of use of optional VCFs has dramatically increased. As technology has improved outcomes with the use of optional VCFs, there are potential advantages of these devices compared with permanent filters regarding inferior vena caval patency and long-term complications.

The cost of managing a pulmonary embolism is significant,¹⁰ as is the cost of inserting (and removing) a VCF. Now is the time for a multicenter prospective examination of current optional VCFs. The challenge is to characterize and identify the patient group that will benefit from a VCF, as well as to identify patients who will ultimately benefit from removal of the devices when the PE risk has resolved. Because such a study would be managed in a controlled fashion and clear guidelines would be created for placement and removal, findings could result in a “best practices model” for all.