Vena Cava Filters: A Call to Action

The concept of venous interruption can be traced back to Armand Trousseau’s lectures at the Hotel Dieu in Paris in the mid-19th century (Trousseau A. Phlegmasia alba dolens. In: Trousseau A, ed. Clinique Medicale de l’Hotel-Dieu de Paris. Paris, France: Balliere, 1865; 654). Although surgical venous interruption was occasionally performed, its invasive nature and attendant morbidity and mortality prevented widespread application. This situation dramatically changed with the introduction of the stainless steel Greenfield filter in 1973, which made percutaneous venous interruption feasible (Greenfield et al. Surgery. 1973;73[4]:599). Between 1979 and 1999, the number of vena cava filters (VCF) placed annually in the United States rose 25-fold, from 2,000 to 49,000 (Stein et al. Arch Intern Med. 2004;164[14]:1541). With the introduction of optional/retrievable VCF that are easy to insert, the expansion in clinical use of these devices has continued to increase. In 2007, nearly 167,000 VCF were placed in the United States, and it is estimated that annual use will top 259,000 in 2012. This expansion in VCF use has been largely driven by the increasing availability of optional filters and increased use for prophylactic rather than treatment indications (Smouse and Johar. Endovascular Today. 2010:74-77; Athanasoulis et al. Radiology. 2000;216[1]:54; Kim et al. J Vasc Interv Radiol. 2008;19[3]:393). Despite the popularity of these devices, there is surprisingly little scholarship documenting their efficacy. Among 2,503 publications on VCF, only two randomized controlled trials (RCT) have been conducted examining outcomes. In comparison, 252 RCT have been reported for the use of low-molecular-weight heparin for venous thromboembolism (VTE) among 2,265 publications (PubMed search conducted Nov 14, 2010).

Consequently, the appropriate indications for VCF insertion continue to be a subject of considerable debate, in large part due to the limited evidence supporting their utility in the treatment of VTE (Table 1). There is broad support for their application in patients who have acute VTE and contraindications to anticoagulation. Although it did not test the efficacy of VCF in the absence of anticoagulation, the PREPIC study did show that filters reduce the incidence of pulmonary embolism (PE) by 63% (6.2% vs 15.1%, hazard ratio 0.37 [95% CI 0.17- 0.79], P = .008) in patients who received at least 3 months of anticoagulation therapy (PREPIC Study Group. Circulation. 2005;112[3]:416). However, skeptics are much less likely to support filter insertion for other indications, such as failure of anticoagulation, as these patients were excluded from participation in PREPIC (Decousus et al. N Engl J Med. 1998;338[7]:409). In these instances, hematologists are more apt to intensify or alter anticoagulant therapy (eg, increase the target INR or switch to an alternative anticoagulant, such as a low-molecular-weight heparin), and look for a potentially correctable etiology for anticoagulation failure (eg, heparin-induced thrombocytopenia, antiphospholipid syndrome, Trousseau syndrome, vascular compression— May-Thurner syndrome, thoracic outlet syndrome, and others), rather than resort to filter placement. In many instances, a VCF may impair rather than enhance local thrombotic control by reducing blood flow proximal to the site of thrombosis. Patients with cancer and idiopathic VTE appear to be at particularly high risk for filter-associated thrombotic complications (PREPIC Study Group. Circulation. 2005; 112[3]:416).

Indications for Vena Cava Filters

Broad Agreement
   Acute VTE with contraindication to anticoagulation
Less Agreement
   Failure of anticoagulation
   Chronic thromboembolic pulmonary hypertension
   Limited cardiopulmonary reserve and acute VTE
   Iliocaval thrombus
   Proximal free-floating thrombus
   Thrombolysis of iliocaval DVT
   Treatment of VTE in cancer patients
   Treatment of VTE in pregnant patients
   VTE prophylaxis in high-risk trauma patients
   VTE prophylaxis in high-risk surgery patients

Although vena cava filters have traditionally been placed in the inferior vena cava (IVC), several recently published papers have examined filter placement in the superior vena cava (SVC) (Usoh et al. Ann Vasc Surg. 2009;23[3]:350). While pulmonary embolism does occur from the upper extremity and SVC, the incidence is significantly less than the IVC (Muñoz et al and the RIETE Investigators. Chest. 2008;133[1]:143). In addition, the complications of filters in the SVC location can be devastating (Owens et al. J Vasc Interv Radiol. 2010;21[6]:779). Placement of filters in the SVC should be considered only in extenuating circumstances.

Data are limited to support the use of VCF for other indications, including chronic thromboembolic pulmonary hypertension ( Jamieson and Nomura. Semin Vasc Surg. 2000;13[3]:236) and patients with free-floating thrombi (Norris et al. Arch Surg. 1985;120[7]:806; Pacouret et al. Arch Intern Med. 1997;157[3]:305). Prophylactic VCF are extensively used in surgical patients (trauma, bariatric surgery, and others), despite an absence of high-quality data supporting their utility (Girard et al. Thromb Res. 2003;112[5-6]:261; Cherry et al. J Trauma. 2008;65[3]:544; Antevil et al. J Trauma. 2006;60[1]:35; Rodriguez et al. J Trauma. 1996; 40[5]:797; Birkmeyer et al. Arch Surg. 2010;252[2]:313; Longitudinal Assessment of Bariatric Surgery (LABS) Consortium, Flum et al. N Engl J Med. 2009;361[5]:445).

In addition to concerns about the limited data documenting the effectiveness of filters for many of their indications, there are legitimate reasons to be concerned about the safety profile of VCF. While fatal periprocedural complications are rare (0.12%) (Athanasoulis et al. Radiology. 2000;216[1]:54), deep venous thrombosis (35.7% vs 27.5%, HR 1.52 [95% CI 1.02-2.27], P = .042) and IVC thrombosis (13% vs 1%; these two patients received IVC filters during follow-up) are common adverse events associated with filter placement (PREPIC Study Group. Circulation. 2005;112[3]: 416). Although previous reports have suggested that migration and mechanical failures are infrequent events after filter placement (Hann and Streiff. Blood Rev. 2005;19[4]:179; Owens et al. J Vasc Interv Radiol. 2010;21[6]:779; Athanasoulis et al. Radiology. 2000;216[1]:54), a recent study indicates that this conclusion may need to be revised. Nicholson and colleagues conducted a radiographic surveillance study of strut fracture in 80 recipients of Bard Recovery and Bard G2 filters (Bard Peripheral Vascular; Tempe AZ) at their institution (Nicholson et al. Arch Intern Med. 2010;170[20]: 1827). They noted a high rate of filter leg fracture for both devices (13/80, 16%). Seven of 28 (25%) Recovery filters and 6 of 52 (12%) G2 filters suffered strut fracture. In seven patients, these fractures were associated with clinical symptoms, including one sudden death and one episode of cardiac tamponade due to a hemorrhagic pericardial effusion that required emergent cardiac surgery (Nicholson et al. Arch Intern Med. 2010;170[20]:1827).

Thus, we are left with a number of questions. Given the risks of thromboembolism posed by the presence of a filter, should all patients with IVC filters receive indefinite anticoagulation? The answer to this question is of significant clinical importance, since the vast majority of optional/retrievable filters remains unretrieved (Mission et al. J Gen Intern Med. 2010;25[4]:321; Dabbagh et al. Thromb Res. 2010;126[6]:493), and many are placed for controversial or questionable indications (Spencer et al. Arch Intern Med. 2010;170[16]:1456). Are the problems noted by Nicholson and colleagues a result of selection bias, or do they represent the tip of a previously unrecognized iceberg? Is this complication unique to the Bard Recovery and Bard G2 filters, or is this experience generalizable to all filters? Are optional/retrievable filters more susceptible to mechanical failures than permanent filter models as a consequence of their design? These and many other outstanding questions regarding VCF warrant further rigorous investigation. In some instances (eg, PE prevention in trauma patients), randomized controlled trials are feasible, while in other circumstances (eg, complications of VCF), prospective cohort studies are a more realistic approach. The important goal is to increase the number and quality of clinical studies of VCF.

Two years ago, the surgeon general, Admiral Steven Galston, MD, issued a call to action to improve efforts on prevention of VTE (www.surgeongeneral.gov/topics/deepvein. Accessed Jan 20, 2011). In this spirit, we believe these recent studies sound a clarion call to action to intensively study VCF to better assess their benefits and risks. They represent a valuable tool in the treatment of VTE, but further information is needed as to when and in whom we should use these devices and what are the short-and long-term complications of their use. We think this research will need to be sponsored and supported at the federal level, as there is little incentive for individual manufacturers to conduct these studies. Although almost 40 years have passed since the introduction of the stainless steel Greenfield filter, many questions remain about the safety and efficacy of these devices. The report of Nicholson and colleagues indicates that a careful reexamination of our clinical use of VCF is urgently needed.

Michael B. Streiff, MD
Division of Hematology, Department ofMedicine
Johns Hopkins Medical Institutions
Baltimore, MD

Kevin Kim, MD
Division of Interventional Radiology and Image Guided Medicine
Department of Radiology
Emory University, Atlanta, GA

Kelvin Hong, MD
Division of Vascular and Interventional Radiology
Department of Radiology
Johns Hopkins Medical Institutions
Baltimore, MD