Physical Address

304 North Cardinal St.
Dorchester Center, MA 02124

Use of Vena Cava Filters and Venous Access Devices


Vena Cava Filters

Introduction

Venous thromboembolism (VTE), a disease entity that includes deep venous thrombosis (DVT) and pulmonary embolism (PE), is a common cause of morbidity and mortality in the United States. Administrative data indicate that 467,000 patients in this country are hospitalized annually with a DVT and 247,000 with a PE. Most patients can be managed safely with anticoagulation, but bleeding complications or presence of contraindications to anticoagulation (recent surgery, trauma) are not uncommon situations in clinical practice. In these instances, vena cava filters are effective tools in preventing PE. Their use has expanded steadily in the United States since the introduction of the Greenfield filter in 1972. In 1979, only 2000 filters were placed nationwide. By 1999, this number had grown to 49,000 filters, with almost 20% placed for prophylactic indications in patients without existing DVT or PE—a trend driven in part by use for extended indications and in part by improved filter technology that facilitated easy percutaneous insertion. An analysis of the Nationwide Inpatient Sample database noted that 132,049 vena cava filters were placed in 2009, a 234% increase over a decade. In light of the increasing use of vena cava filters, it is essential for practitioners to have a working knowledge of the characteristics of individual filters, as well as evidence of the risks and benefits of their use. The purpose of this chapter is to provide a comprehensive overview of the available literature on vena cava filters.

Historical Perspective

The intellectual foundation of vena cava filtration as a strategy for PE prevention can be traced to some of the nineteenth and twentieth centuries’ most renowned physicians. In 1846, Rudolf Virchow proposed the concept that pulmonary thrombi were in fact emboli that originated primarily in veins of the lower extremities. Two decades later in his lectures at the Hotel Dieu in Paris, Armand Trousseau suggested that a physical barrier might be an effective preventive measure against PE. In the 1940s, John Homans popularized femoral vein ligation for the prevention of PE. Although this procedure was effective, femoral vein ligation was associated with significant postthrombotic complications in the affected limb and did not provide protection from clots located in the contralateral extremity.

Consequently, an impressive array of surgical procedures and devices focusing on vena caval interruption was developed over the subsequent 4 decades. Surgical inferior vena cava (IVC) ligation was associated with an operative mortality of 12%, recurrent PE in 4%, and postthrombotic syndrome (PTS) in 22%. However, other series documented recurrent PE in 20% to 50% of patients with longer follow-up. To preserve caval blood flow, surgical plication was proposed to replace complete caval ligation. However, surgical plication proved to be technically demanding, so a variety of externally applied clips (Moretz, Miles, Adams, and Adams-DeWeese clips) were developed to reduce the procedure's complexity. Unfortunately, these devices were not associated with improved clinical outcomes (10% operative mortality, 4% recurrent PE, 16% PTS).

However, the concept of caval filtration served as a catalyst toward the development of intraluminal devices. The first widely used intraluminal device for caval interruption was the Mobin-Uddin umbrella filter. Although initial reports were favorable (0.6% operative mortality, 1.7% recurrent PE, 6.5% venous stasis), subsequent patient series noted higher rates of PE (7.8%), IVC occlusion (53%), venous stasis (75%), and occasional episodes of cardiopulmonary migration. These complication rates and the introduction of the stainless steel Greenfield filter precipitated eventual removal of the Mobin-Uddin filter from the US market in 1977.

The introduction of the stainless steel Kimray-Greenfield filter (SSGF [Boston Scientific, Marlborough, MA]) in 1972 revolutionized the field of vena caval interruption. It is a testament to its ground-breaking design that many of its features have been incorporated into many subsequent devices. The SSGF consists of a cone constructed of stainless steel wire affixed to a central apical cap. The area of filtration is maximized by having each of the six wire legs make alternating right- and left-hand bends. At the caudal terminus, each wire ends in a hook that serves to anchor the device within the vena cava ( Fig. 30.1 ). The funnel shape of the filter is designed such that the cross-sectional area of the IVC is reduced by only 50% when two-thirds of the filter are filled with thrombus, theoretically enhancing the body's fibrinolytic system the opportunity to lyse retained clots. Since its introduction in 1972, the SSGF has been implanted in more than 120,000 patients.

FIG 30.1, Vena Cava Filters.

Currently Available Vena Cava Filters

Permanent Filters

Although placed percutaneously beginning in 1984, the original SSGF was too inflexible for easy insertion. It required a 29-French (Fr) outer diameter (OD) introducer catheter, which may have contributed to the high incidence (45%) of insertion site thrombosis (IST) noted in some studies. Consequently, Greenfield and colleagues developed the titanium Greenfield filter (TGF [Boston Scientific]), which was introduced in 1988 and received US Food and Drug Administration (FDA) approval in 1989. Its conical design and the configuration of its six struts with terminal hooks were similar to those of the original SSGF (see Fig. 30.1 ). The titanium alloy (Beta II titanium alloy) allowed for greater flexibility, such that the filter could be introduced via a 12-Fr catheter (14.3-Fr OD). It was anticipated that reduction in catheter diameter would lead to a lower incidence of IST. Because titanium is nonferromagnetic, the filter also would be at lower risk for migration during magnetic resonance imaging (MRI) and may cause minimal MRI artifact. Preliminary clinical studies with the TGF revealed a high rate of distal slippage and penetration of the caval wall, so modifications of the hook design and base diameter were made, resulting in the modified-hook titanium Greenfield filter (MHTGF; Boston Scientific), which was introduced in 1991 ( Table 30.1 ; also see Fig. 30.1 ).

TABLE 30.1
Specifications of Currently Available Permanent Vena Cava Filters
Filter Company FDA Approval (year) Material MRI Compatibility Insertion Approach Recommended Caval Diameter (mm) Length (mm) Catheter ID (mm)
Titanium Greenfield Filter (TGF) Boston Scientific 1987 Beta III titanium alloy Yes Jugular/femoral 28 50 12
Modified-hook Titanium Greenfield Filter (MHTGF) Boston Scientific 1990 Titanium Yes Jugular/femoral 30 47 12
12-Fr Over-the-Wire Stainless Steel Greenfield Filter (PGF) Boston Scientific 1997 Surgical Stainless steel Yes Jugular/femoral 28 50 12
Bird's Nest Filter Cook Medical LLC 1989 Biocompatible Stainless Steel Yes Jugular/Femoral 40 80 12
Vena Tech LGM Filter B Braun Interventional 1989 Phynox plate Yes Jugular/Femoral 28 38 10
Vena Tech LP Filter B Braun Interventional 2001 Phynox wire Yes Jugular/femoral 28 43 7
Simon Nitinol Filter Bard Peripheral Vascular 1990 Nitinol Yes Jugular/femoral 28 38 7
TrapEase Filter Cordis 2000 Nitinol Yes Jugular/femoral/antecubital 30 50 6
SafeFlo Rafael Medical Technologies 2009 Nitinol Yes Jugular/femoral 25 mm NR 6
FDA, US Food and Drug Administration; ID, inner diameter; MRI, magnetic resonance imaging; NR , not reported.

Although migration and penetration were largely improved in the MHTGF, tilting remained an occasional problem. To remedy this situation, the stainless steel percutaneous Greenfield filter (PGF [Boston Scientific]) was developed. Similar to the SSGF in design, its apical hub differs in that it has a central hole that allows placement over a guidewire to prevent filter tilting and asymmetry. Its design gives it increased flexibility, allowing insertion via a small 12-Fr (15-Fr OD) catheter. The filter comprises six stainless steel struts fitted into a cylindrical hub (see Fig. 30.1 and Table 30.1 ). Once deployed, the filter is 49 mm in length and 32 mm in base diameter. The PGF can be used in patients with IVC diameters of 28 mm or less. To prevent migration, the anchor hooks are positioned bidirectionally, with four directed superiorly and two inferiorly. Although the filter appears to be MRI safe, its stainless steel composition causes significant image artifact. In 2005, all PGFs manufactured before March 10, 2004, were recalled from the market because of eight reports of detachment of the filter carrier capsule at the bond with the insertion catheter, which in two instances led to serious patient injury, including one death (see Table 30.1 ).

The Bird's Nest filter (Cook Medical LLC, Bloomington, IN), introduced in 1982, has a unique structure and comprises four 25 cm long stainless steel wires that are folded over several times and attached to two V -shaped struts. These struts have hooks that affix the filter within the vena cava (see Fig. 30.1 and Table 30.1 ). It can be inserted via a 14-Fr OD catheter. Because the Bird's Nest filter expands up to a diameter of 60 mm, it is the filter of choice for patients with caval diameters greater than 30 mm. The original version of the Bird's Nest filter was associated with several fatal episodes of migration, a flaw remedied by introduction of a new version with stiffer anchor struts. The Bird's Nest filter generates the largest MRI artifact of any filter because of its stainless steel construction, although it is stable in magnetic fields up to 1.5 T (see Table 30.1 ).

The LGM or VenaTech filter (B. Braun Interventional, Bethlehem, PA) has six struts arranged in a conical fashion (similar shape as the Greenfield filter) made of Phynox, a cobalt-chromium-nickel alloy used to make pacemaker leads. The side rails attached to the filter struts anchor the filter within the vena cava (see Fig. 30.1 and Table 30.1 ). The VenaTech filter received FDA approval in 1989 and is MRI compatible. It is inserted with a 12-Fr sheath (14-Fr OD) and can be placed in venae cavae up to 28 mm in diameter. The original VenaTech filter is no longer available on the US market and has been replaced by the VenaTech Low-Profile (LP) Filter (B. Braun/VenaTech), which was approved by the FDA in 2001. In contrast to its predecessor, the VenaTech LP filter uses eight Phynox wires shaped in a conical fashion that fuse caudally in pairs to form side rails that secure the filter to the vena caval wall. Each side rail has a hook oriented superiorly or inferiorly (see Fig. 30.1 ). It can be introduced by a 7-Fr sheath, and once deployed, it is 43 mm in height and 40 mm in diameter. The VenaTech LP filter is limited to IVC diameters of 28 mm or less, and it is MRI compatible (see Table 30.1 ).

The Simon Nitinol filter (Bard Peripheral Vascular, Tempe, AZ) is composed of a nickel/titanium alloy (Nitinol) that possesses thermal memory properties. At 4°C, the filter exists as a set of straight wires that automatically unfold at body temperature to form an umbrella filter with seven petals. Six hooked struts anchor the filter within the vena cava (see Fig. 30.1 and Table 30.1 ). The FDA approved the filter in 1990, and it is designed for IVC diameters of 28 mm or less (see Table 30.1 ).

The G2 filter system (Bard Peripheral Vascular) is a 12-legged conical filter constructed of Nitinol wires that gained FDA approval for permanent placement in November 2005. The conical array of six short and six long curved Nitinol wires creates two planes of caval filtration and has a maximal deployment length of 41 mm. The maximal caval diameter for placement is 28 mm. The flexibility afforded by its Nitinol construction allows placement with a 7-Fr (inner diameter [ID]) catheter system from a jugular or femoral venous access site ( Table 30.2 ; also see Fig. 30.1 ).

TABLE 30.2
Specifications of Currently Available Retrievable Filters
Filter Company FDA Approval (Year) Material MRI Compatibility Insertion Approach Recommended Caval Diameter (mm) Length (mm) Catheter ID (mm)
Günther Tulip Filter Cook Medical LLC 2000 (2003 for retrieval) Conichrome Yes Jugular/femoral 30 50 Jugular: 7 Femoral: 8.5
OptEase Filter Cordis 2002 Nitinol Yes Jugular/femoral/antecubital 30 54 6
Recovery Filter Bard Peripheral Vascular 2002 Nitinol Yes Femoral 28 41 7
G2 Bard Peripheral Vascular 2005 Nitinol Yes Jugular/femoral 28 42 7
Celect Cook Medical LLC 2007 Conichrome Yes Jugular/femoral 30 48 Jugular: 7 Femoral: 8.5
ALN ALN Implants Chirurgicaux 2008 Stainless steel Yes Jugular/femoral/brachial 32 55 7
G2 Express/G2x Bard Peripheral Vascular 2008 Nitinol Yes Jugular/femoral 28 42 7
Option Argon Medical Devices 2009 Nitinol Yes Jugular/femoral 32 54.5 6
Eclipse Bard Peripheral Vascular 2010 Nitinol Yes Jugular/femoral 28 42 7
Meridian Bard Peripheral Vascular 2011 Nitinol Yes Jugular/femoral 28 42 8
Denali Bard Peripheral Vascular 2013 Nitinol Yes Jugular/femoral 28 50 8
FDA, US Food and Drug Administration; ID, inner diameter; MRI, magnetic resonance imaging.

The TrapEase filter (Cordis Corp., Milpitas, California) is also composed of Nitinol and consists of two conical filter baskets—one facing cranially, the other caudally—that are connected to create two levels of caval filtration (see Fig. 30.1 and Table 30.1 ). When viewed in cross section, the filter baskets form a filtration plane composed of six diamond-shaped flow corridors. Six struts with proximal and distal hooks anchor the baskets within the vena cava. The FDA approved the TrapEase filter for clinical use in 2000. It can be placed through a 6-Fr sheath in patients with IVC diameters of up to 30 mm. The filter's nickel titanium alloy ensures MRI compatibility and produces minimal image artifact (see Table 30.1 ).

The Günther Tulip filter (Cook Medical LLC) is constructed of Conichrome, an MRI-compatible alloy similar to Phynox, composed of cobalt, chromium, nickel, iron, molybdenum, and manganese. It has been in use in Europe since 1992 and first became available for use in the United States in 2001. In 2003, it received FDA approval for use as a retrievable filter. The filter basket is formed by four struts stabilized by wire loops that extend three-quarters down the length of each strut. Hooks on the end of each strut attach the filter to the vena caval wall. A hook on the filter apex facilitates retrieval (see Fig. 30.1 and Table 30.2 ). It can be introduced via an 8.5-Fr sheath (11-Fr OD) through a femoral or jugular approach and can be removed via a jugular approach.

The SafeFlo filter (Rafael Medical Technologies, Dover, DE) is a two-stage device constructed of Nitinol wire that consists of an outer ring platform and an attached filter element that is composed of five petal loops surrounding a central flow corridor. The filter is anchored in the IVC without hooks by two wire rings angled toward each other that exert external radial pressure on the IVC wall. This design allows the opportunity for postinsertion filter repositioning in the event of misplacement (see Fig. 30.1 ). Three different filter sizes are available for small (diameter 16 to 19 mm), intermediate (19 to 22 mm), and large IVCs (22 to 25 mm). The SafeFlo filter can be inserted through a 5- or 6-Fr delivery catheter. The FDA approved it in May 2009 as a permanent inferior vena cava filter (IVCF); in Europe it was approved for permanent and retrievable indications in 2004.

Optional Filters

Given the long-term complications of permanent filters, development of a safe and effective optional filter would offer great benefit to patients who have experienced an acute episode of VTE and have a short-term contraindication to anticoagulation. Optional filters can be divided into three categories: temporary, retrievable, and convertible. As their name suggests, temporary filters are designed only for short-term insertion and are held in place by a catheter or wire anchored to the insertion site. One temporary filter, the Tempo filter (B. Braun Interventional), was investigated in a clinical trial in the United States. Although initial results were promising, subsequent episodes of catheter buckling and atrial migration halted further US development. A new version of the Tempo filter is under clinical investigation.

The Angel Catheter (BiO 2 Medical, Golden, CO) is the only temporary filter on the US market. It consists of a triple-lumen central venous catheter that is tipped with a self-expanding nitinol filter. It is placed via a femoral vein approach using a 35 cm 9-Fr OD sheath over a 0.035 inch guidewire (see Fig. 30.1 ). It was FDA approved in August 2016.

Retrievable filters resemble permanent filters in that they are anchored within the IVC by small tethering hooks. These anchoring hooks gradually become incorporated into the wall of the IVC, so most retrievable filters can be left in place for only a limited time before they must be removed. However, if the need for prolonged, indefinite caval interruption arises, these filters may be left in place to function as permanent filters.

Seven filters have been approved for retrieval by the FDA: the Günther Tulip filter, the OptEase filter (Cordis Corp.), the Denali filter (and earlier versions including Recovery, G2/G2X, Eclipse, and Meridian filters; Bard Peripheral Vascular), the ALN filter (ALN Implants Chirurgicaux, Ghisonaccia, France), the Celect filter (Cook Medical LLC), and the Option filter (Argon Medical Devices, Plano, TX; see Table 30.2 ). The Günther Tulip filter was approved for use as a retrievable filter in 2003 and was described earlier. Although the manufacturer recommends the Günther Tulip filter be removed within 10 to 14 days of placement, in individual cases, filters have been retrieved as long as 8 years after placement. In some cases, the caval lifespan has been prolonged by periodic filter repositioning. A more complete discussion of Günther Tulip outcomes, including retrieval, can be found in a later section.

In 2002, the FDA approved the OptEase vena cava filter for retrieval up to 23 days after insertion. The OptEase filter is composed of Nitinol and has a double basket design similar to the TrapEase permanent IVC filter. Its six struts have superior barbs to provide resistance to migration. It can be placed through a 6-Fr sheath in patients with IVC diameters of up to 30 mm and is MRI compatible (see Fig. 30.1 and Table 30.2 ).

The Bard Recovery filter was approved by the FDA in 2003 for permanent and optional use (see Table 30.2 ). The Recovery filter is a cone-shaped filter consisting of six radiating arms and legs composed of 0.13-inch Nitinol wire (see Fig. 30.1 ). It is MRI compatible. The filter legs have hooks that tether the device at the insertion site. The filter is 40 mm long and expands up to 32 mm in diameter. It is inserted via the femoral vein using a 7-Fr ID/9-Fr OD delivery catheter and can be deployed in cavae up to 28 mm in diameter. The Recovery filter is retrieved by the jugular route using a 12-Fr retrieval cone catheter set. Initial clinical experience with the Recovery filter in 32 patients reported by Asch was promising. The mean implantation period was 53 days, and the longest duration was 134 days. Filter retrieval was successful in all 24 patients in whom it was attempted. No symptomatic episodes of PE, IVCT, or IST were noted. One filter with trapped clot migrated 4 cm from its implantation site. Subsequent studies documented an increased incidence of filter migration, IVC penetration, and filter leg fracture. Consequently, the Recovery filter was removed from the US market and replaced with the G2 filter.

The Bard G2 filter is a redesigned model of the original Recovery filter. It was FDA approved for use as a permanent filter in 2005 and then subsequently as a retrievable filter in 2008 (see Fig. 30.1 and Table 30.2 ). The G2X filter, a modified form of the G2, was approved by the FDA in 2009. The G2 filter has six arms of 19.8 mm in length (increase from 11.3 mm in Recovery filter) and six legs with a 40-mm span (increase from 32 mm) and anchor hooks with increased thickness (from 0.216 to 0.267 mm) that provide it greater resistance to fracture and migration, along with a reduced tendency to tilt compared with the original Recovery filter design. Maximal recommended caval diameter for insertion is unchanged at 28 mm. The G2 filter can be inserted percutaneously via the femoral or jugular routes using 7- or 10-Fr introducer catheters, respectively. Retrieval is accomplished from a jugular approach using the 10-Fr Recovery cone retrieval catheter system. The G2 filter has also been plagued by filter strut fracture and migration, prompting release of the G2X and recently the Eclipse filter, which has an apical hook for retrieval. The latest Bard filter is the Denali filter. It consists of 12 nitinol struts that are engineered to form two levels of filtration. The six filter legs (lower struts) have cranial and caudal hooks to prevent migration and limit caval penetration, while the six arms (upper struts) provide a second layer of filtration. It can be placed in cava as large as 28 mm in diameter and is deployed via an 8.4 Fr (inner diameter) sheath from either a jugular or femoral vein approach.

The Celect filter is based on the Günther Tulip filter (see Table 30.2 ). The FDA approved it for use in the United States in 2008. It is constructed of Conichrome, a cobalt, nickel, chromium, molybdenum, and iron-containing nonferromagnetic alloy. The filter has four anchoring struts and eight secondary struts to reduce tilt and improve clot capture. It can be placed by the femoral or internal jugular veins (IJVs) using 8.5- or 7-Fr introducer catheters. Retrieval can be achieved using the Günther Tulip retrieval system via a right jugular approach.

The ALN filter is a cone-shaped nonferromagnetic stainless steel filter that consists of six arms with hooks for anchoring and three longer legs that center the device in the vena cava (see Fig. 30.1 and Table 30.2 ). It was approved by the FDA in 2008. It is designed for deployment in cavae 32 mm or less in diameter. It can be inserted via the internal jugular, femoral, or brachial veins using a 7-Fr introducer catheter. Maximal deployment length is 55 mm. It can be retrieved only from the internal jugular route, using a 7-Fr extraction catheter housed in a 9-Fr introducer sheath.

The Option filter is a cone-shaped filter cut out of Nitinol tubing that was approved by the FDA for use as a retrievable filter in 2009 (see Table 30.2 ). It has six collapsible legs tipped with retention struts that anchor it in the vena cava and an apical hook for retrieval. The device is deployed through a 6.5-Fr OD (5-Fr ID) sheath via the femoral or IJV. It has been cleared for insertion into cavae up to 30 mm in diameter. The filter measures 54.5 mm in height after placement. The filter is retrieved through the jugular vein with an 8- or 10-Fr catheter.

The Crux filter (Crux Biomedical Inc., Menlo Park, CA) consists of two spiral Nitinol wire loops fashioned to form a helical structure. One loop suspends a web of polytetrafluoroethylene (PTFE) that forms a 6 × 8 mm filtration grid flow channel mechanism; the other loop contains three anchors that secure the device in the IVC. It is delivered via a 6-Fr ID/8-Fr OD catheter into cavae up to 28 mm in diameter. Insertion or retrieval can be performed via the jugular or femoral veins. It was approved for prevention of PE by the FDA in July 2012.

The VenaTech Convertible IVC filter is an optional cone filter based on the VenaTech LP design with eight legs attached to lateral stabilizers with anchors to affix the device in the IVC. The filter is constructed of a cobalt-chromium alloy and is MRI compatible. It can be inserted into cavae up to 32 mm in diameter using a 10-Fr/12.5-Fr (ID/OD) catheter. The filter can be converted into an IVC stent by removing the hub at the filter tip, which allows the filter legs to fold against the IVC wall. It was approved for use in the United States by the FDA in February 2017.

Technical Aspects

Most IVCF procedures are performed in the angiography suite under fluoroscopy. The most common approach is through the femoral vein or the right IJV. Deployment of IVCFs in right-sided vessels versus left-sided vessels and via internal jugular access versus femoral vein is associated with a lower rate of IVCF tilt. With the growth of endovascular interventions, bedside placement of IVCFs by intravascular ultrasound (IVUS) guidance has been shown to be safe, effective, and reliable. This technique is especially advantageous in patients who cannot tolerate intravenous contrast dye load or ionizing radiation (e.g. renal insufficiency, pregnancy) or who are too critically ill for transportation to the angiography suite. When using IVUS, the femoral vein approach is preferred.

Traditionally, prior to filter placement, cavography was performed to rule out anatomic variants of the IVC, thrombus in the IVC or iliac veins (the latter only in cases of femoral access), and to identify the origin of the renal veins. IVCFs are placed infrarenally to avoid renal vein thrombosis and subsequent renal vein outflow obstruction in the event that the filter becomes occluded with thrombus either from embolization or de novo. Most IVCFs have a conical shape and are designed to be removed via gooseneck snare techniques or propriety retrieval cone device via the IJV. This endovascular snaring technique captures the filter by hook and allows for collapse and disengagement from the caval wall. Although retrievable IVCFs should be removed as soon as no longer indicated, advanced endovascular retrieval techniques have virtually removed an “expiration” on retrieval time periods. Dwell time is strongly associated with the need for advanced retrieval procedures; however, these techniques are associated with a 97% success rate and low complication rate.

Studies on Efficacy of Inferior Vena Cava Filters

The only purpose for placing an IVCF is to prevent PE, so it is essential for every physician to become familiar with data supporting filter effectiveness in this task and with their adverse effects. Unfortunately, the vast majority of data on vena cava filters has been derived from nonrandomized case series. Substantial differences exist between studies regarding subject populations (and baseline risk of recurrent VTE), concomitant anticoagulation use, and the intensity, comprehensiveness, and duration of follow-up. Therefore combined results of these studies ( Tables 30.3 and 30.4 ) should be interpreted in light of these limitations. Nevertheless, in the absence of randomized comparisons of filters, these data are the principal means available by which filter efficacy and safety can be assessed. As is shown in Tables 30.3 and 30.4 , most available filters are roughly equivalent. Next generation filter models (VenaTech LP, OptEase, Recovery, G2, and Celect) appear to have lower PE rates. In contrast, the Celect Platinum and Option filters have a relatively higher rate of symptomatic PE after placemen. Since each of these newer devices has been studied in relatively few patients and for short intervals of time, we suggest these event rates be interpreted cautiously. Randomized controlled trials (RCTs) of different filter models will be necessary to determine the relative efficacy of different filters.

TABLE 30.3
Clinical Studies of Permanent Inferior Vena Cava Filters
Filter Patients (Studies) F/U Duration (Range [months]) Pulmonary Embolism Deep Venous Thrombosis IVC Thrombosis
Stainless Steel Greenfield 3636 (42) 18 (1–60) 104/3038 (3.4%) 96/1634 (5.9%) 89/2529 (3.5%)
Range (0%–9%)
Fatal 38 (1.3%)		 (Range 0%–18%) (Range 0%–18%)
Titanium Greenfield 649 (10) 5.8 (0–81) 19/556 (3.4%) 5/22 (22.7%) 16/364 (4.4%)
(Range 0%–4.4%)
Fatal 10 (1.8%)		 (Range 0%–36%) (Range 1%–31%)
Percutaneous Stainless Steel Greenfield 740 (5) 20.6 (8.5–26) 9/372 (2.4%)
Fatal 1 (0.3%)		 41/305 (13.4%) 8/267 (3%)
Bird's Nest 1742 (18) 14.2 (0–60) 49/1441 (3.4%) 27/448 (6%) 38/1334 (2.8%)
(Range 0%–7.1%)
Fatal 22 (1.5%)		 (Range 0%–20%) (Range 0%–15%)
Vena Tech 1353 (16) 17.3 (0–65) 45/1266 (3.6%) 8/25 (32%) 102/1074 (9.5%)
(Range 0%–6.3%)
Fatal 12 (0.9%)		 (Range 32%) (Range 0%–28%)
Simon-Nitinol 1022 (12) 15 (0–62) 30/967 (3.1%) 11/123 (8.9%) 48/945 (5.1%)
(Range 0%–5.3%)
Fatal 17 (1.8%)		 (Range 8%–11%) (Range 0%–50%)
VenaTech LP 1306 (2) 53 (3–72) 10/1020 0.98% (Range 0–1.04%) Fatal 10 (1.04%)0 110/1020 (10.8%)
(Range 10.8%–14%)		 124/1020 (12.2%)
TrapEase 1791 (11) 11.2 (4–65) 28/1791 (1.6%) 114/1791 (6.5%) 28/1791 (1.6%)
(Range 0%–3.3%)
Fatal 1 (0.06%)		 (Range 0%–10.7%) (Range 0%–6.9%)
F/U, Follow-up; IVC, inferior vena cava.

TABLE 30.4
Clinical Results of Studies of Retrievable/Temporary Inferior Vena Cava Filters
Filter Type Patients (Studies) Mean Age Follow-Up (Months) PE DVT IVCT
Günther Tulip 2634 (20) 50.5 11.3 30/2634 (1.1%) 76/2634 (2.8%) 38/2634 (1.4%)
Range 3–41 Range 0%–3.6%
Fatal 3 (0.1%)		 Range 0%–14.4% Range 0%–9.3%
Optease 1200 (9) 47.5 5.3 19/1036 (1.8%) 47/1036 (4.5%) 17/1036 (1.6%)
Range 4–20 Range 0%–4.2% Range 0.5%–10.4% Range 0%–6.3%
Recovery 1054 (10) 49.9 16 13/691 (1.8%) 28/637 (4.4%) 9/637 (1.4%)
Range 5.3–20 Range 0%–3.8% Range 0%–18% Range 0%–8.8%
G2 1156 (8) 48.4 6.2 11/569 (1.9%) 11/569 (1.9%)
Range (0%–14.3%)		 5/961 (0.5%)
Range 4.1–13.2 Range 0%–6.5% Range 0%–2.2%
ALN 912 (6) 69.9 9.3 11/912 (1.2%) 36/912 (3.9%) 16/912 (1.8%)
Range 3–21 Range 0%–3.1% Range 0%–15.2% Range 0%–6%
Celect 2308 (13) 58.7 5.6 15/655 (2.3%) 8/493 (1.6%) 8/655 (1.6%)
Range 2.3–10.1 Range 0.9%–4.2% Range 0%–14.9% Range 0%–2.5%
Celect Platinum 556 (1) 68 3 3/39 (7.7%) NR 11/335 (3.3%)
Option 824(4) 64.6 7.5 35/781 (4.7%) 65/781 (8.7%) 15/781 (2%)
Range 6.1–9.5 Range 3.3%–8% Range 6.6%–18% Range 1.2%–8.3%
Crux 125 (1) 59.1 6 3 (2.9%) 12 (9.6%) 5 (4%)
Denali 394 (3) 56.6 24 6/287 (2.1%)
Range 0%–3%		 26/200 (13%) 1/87 (1.1%)
Angel catheter a 277 (2) 45.2 0.22
Range 0.21–0.24		 01/277 (0.36%)
Range 0%–0.9%		 30/277 (10.8%)
Range 0%–18%		 31 /277 (11.2%)
Range 0%–19%
DVT, Deep venous thrombosis; IVCT, inferior vena caval thrombosis; PE, pulmonary embolism; NR , none reported.

a Temporary filter.

Although randomized comparisons of various filter models have not been performed, two randomized trials of vena cava filters in the management of VTE have been published. In PREPIC, investigators randomly assigned 400 patients with proximal DVT with or without PE who were judged to be at high risk for PE to receive either a permanent IVCF or no filter. All patients received therapeutic anticoagulation for known VTE with either unfractionated heparin (UFH) or enoxaparin followed by vitamin K antagonists (VKAs) for at least 3 months. The subject population was clearly at high risk for adverse outcomes: 60% had unprovoked VTE, 35% had a previous history of VTE, 40% had iliocaval thrombosis, and 14% had active cancer. Permanent filters were placed within 48 hours of random assignment. Ventilation/perfusion (V/Q) scans were performed at baseline and after 8 to 12 days of anticoagulation. Ninety-nine percent of patients were discharged while on anticoagulation, and 94% received anticoagulant therapy for at least 3 months. Only 62% of filter patients and 64% of the group without filters received anticoagulation beyond 3 months. At 2 years, 38% of patients were still on oral anticoagulation, and 35% received anticoagulation over the entire 8-year follow-up, with no significant differences noted between groups. After 12 days of treatment, IVCFs were associated with a significant decrease in the incidence of symptomatic and asymptomatic PE, compared with anticoagulation alone (1.1% vs. 4.8%; P = .03). When only symptomatic PEs were considered, differences between the filter- and no-filter groups were no longer significant (1% vs. 3%). At 2 years, symptomatic PE tended to be less frequent among filter recipients than among those who had received anticoagulation alone (3% vs. 6%), although this difference was not significant ( P = .16).

Fatal emboli were also more common in patients treated solely with anticoagulation (0.5% vs. 2.5%). However, IVCFs were associated with significantly more recurrent DVT than was observed with anticoagulation alone (21% vs. 12%; P = .02). No difference in bleeding or mortality was documented. Sixteen of 37 patients (43%) with IVCFs who had recurrent DVT also had IVC thrombosis. At 8 years, outcome data on 99% of patients became available, and 198 patients (50%) were found to be still alive. The occurrence of symptomatic PE was less frequent in filter recipients than in those treated with anticoagulation alone (6% vs. 15%; P = .008); 50% of PEs in the no-filter group occurred during the first 2 years of follow-up. DVT was more frequent among filter patients (36% vs. 28%; P = .042); 65% of DVTs occurred among filter patients within the first 2 years of follow-up. Symptomatic filter thrombosis occurred in 13% after 8 years. PTS was observed in 70% of patients in either the filter and no filter groups. No difference in overall survival was reported.

The objective of PREPIC 2 was to assess the efficacy and safety of a retrievable filter, the ALN filter (ALN Implants Chirurgicaux, France), in patients with acute symptomatic PE associated with a lower-extremity DVT or superficial vein thrombosis and at least one additional criterion for severity: advanced age (>75 years old), active cancer, chronic cardiac or respiratory insufficiency, ischemic stroke with leg paralysis within the last 6 months, iliocaval DVT, bilateral DVT, or signs of right ventricular dysfunction or myocardial injury. Similar to PREPIC, patients with a contraindication to anticoagulation were excluded from the study. A total of 399 patients were recruited in 6 years. Overall patients were at high risk for recurrence, as evidenced by 35% with personal history of VTE, 15% with active cancer, 9% with iliocaval DVT, and 77% with unprovoked VTE. There was no difference in administration of thrombolytic therapy between groups (1.5%). Patients received at least 6 months of therapeutic anticoagulation. Median duration of initial parenteral therapy and subsequent VKA was similar between the filter and non-filter groups. Time in therapeutic range with VKA was approximately 60% in both groups.

At 3-month follow-up, recurrent PE occurred in 3.0 % of the filter group compared to 1.5 % in the nonfilter group (RR 2.0; 95% CI 0.51 to 0.79). No difference was found in rates of DVT (0.5% in both groups). Overall, 15 deaths (7.5%) occurred in the filter group and 12 (6.0%) in the nonfilter group. Filter removal was attempted at 3 months in 91% of patients and was successful in 93% of those patients. Among those that received a filter, access site hematoma occurred in 2.6%, filter thrombosis in 1.6%, and technical retrieval failure in 5.7%.

Taken together, the results of PREPIC and PREPIC 2 do not provide justification for routine placement of IVCF in patients with PE. In PREPIC there was short-term reduction in the total number of PEs at the cost of a long-term increase in recurrent DVT, with no reduction in mortality. In addition, although it was not routine at the time PREPIC was performed, extended-duration anticoagulant therapy is commonly recommended for high-risk patients similar to the PREPIC population. Therefore it is likely that the benefits associated with filter placement in PREPIC would be considerably less with currently recommended durations of anticoagulation for VTE. In PREPIC 2 the low rate of PE observed in the control group of PREPIC 2 (1.5%) is consistent with successful prevention of recurrent PE in contemporary therapeutic anticoagulation trials.

Additional valuable information about the efficacy of IVCFs can be gained from a small randomized study evaluating the benefits of an IVCF in addition to fondaparinux in patients with cancer and acute DVT with or without PE. In this study, Barginear et al. randomly assigned patients to a weight-based dose of fondaparinux alone or fondaparinux and a permanent IVCF (Vena Tech LP, B. Braun Medical). Patients with severe renal dysfunction (CrCl <30 mL/min), hereditary thrombophilia, intracranial hemorrhage, or brain metastasis secondary to melanoma, choriocarcinoma, renal cell carcinoma, or medullary thyroid carcinoma were excluded. The study period was 90 days and patients were followed for 3 years. Baseline characteristics with regard to the underlying cancer and treatment were similar between the two groups. In patients with a confirmed DVT at baseline, a bilateral ultrasound of the lower extremities was systematically performed on days 14, 30, and 56 to evaluate for recurrence or progression. In patients with a confirmed PE at baseline, a CTA was systematically performed on day 56 to evaluate the clot burden. A total of 64 patients were enrolled, and 107 DVT and 43 PE sites were confirmed by imaging. Fifty-one percent of all patients enrolled had DVT resolution by study day 56, with those in the nonfilter arm demonstrating a higher rate of resolution (61% vs. 37.5%, P = .02). A similar percentage of patients (47%) enrolled with a PE had resolution of the PE by study day 56. No patient had a recurrent DVT. Two patients (3%) had new asymptomatic PEs, one in each treatment arm. Eighty-six percent of patients completed the planned 90 day period of fondaparinux treatment. Major bleeding complications were <5% in both groups. No difference in survival was observed between the two groups. These study results do not support the routine use of IVCFs in cancer patients with VTE being treated with fondaparinux. A subsequent meta-analysis combining the results from all three RCTs (PREPIC PREPIC 2, and Barginear et al.) concluded that an IVCF in addition to anticoagulation in patients with acute VTE did not affect mortality, recurrent or fatal PE, or recurrent DVT. Unfortunately, because all patients in these randomized trials received therapeutic anticoagulation, these data offer no insight into the outcome of the typical patient who has had an IVCF placed—namely, those who have contraindications to anticoagulation. Nevertheless, in the absence of a randomized comparison between IVCFs and anticoagulation for VTE, these data are the principal available means by which filter efficacy and safety can be assessed.

Adverse Outcomes After Filter Placement

Permanent Filters

Adverse outcomes after IVCF placement may occur at the time of insertion or months to years later. Acute procedure-related complications include misplacement (1.3% of insertions), pneumothorax (0.02%), hematoma (0.6%), air embolism (0.2%), inadvertent carotid artery puncture (0.04%), and arteriovenous fistula (AVF; 0.02%). According to the published case series, fatal complications of placement are rare, occurring in only 0.13% of insertions. Among various individual filter models, the published fatal complication rate is highest for the Bird's Nest filter (0.34%), compared with the original stainless steel Greenfield filter (0.11%), the titanium Greenfield filter (0.15%), the TrapEase filter (0.13%), and the VenaTech filter (0.07%). No fatal complications have been reported in the published literature for the PGF, the Simon Nitinol filter, the Günther Tulip, OptEase filters, or the Bard G2. The higher fatal complication rate associated with the Bird's Nest filter primarily reflects results of a single study that documented four episodes of fatal IVC thrombosis. If one considers this experience isolated, the fatal complication rate associated with the Bird's Nest filter is comparable with other filter models. Bleeding complications related to IVC filter placement have been reported to occur in 6% to 15% of insertions.

Other adverse outcomes after filter placement include DVT, IVC thrombosis, filter migration, IVC penetration, and filter disruption. Similar to the data on PE, the frequency of other filter complication rates must be interpreted cautiously in light of the limitations of the existing literature. Because filters are often placed in patients with a VTE who cannot at least initially receive anticoagulation, DVT is not a totally unexpected event in this patient population, a conclusion underscored by the 27.5% cumulative incidence of DVT noted in patients receiving anticoagulation alone in the PREPIC study during 8 years of follow-up. The frequency of DVT from case series of permanent filters varies widely, reflecting differences in the VTE risk profile of the enrolled patients; the use of anticoagulation, including its duration and quality; and the duration and intensity patient follow-up (see Table 30.3 ). Among individual studies of different permanent filters, the highest recorded frequencies of DVT were reported for the titanium Greenfield filter (36%) and the Vena Tech filter (32%). Randomized trials are needed to determine whether important differences exist between different permanent filter models.

IVC thrombosis, although substantially less problematic for contemporary filter models than for the Mobin-Uddin umbrella filter and IVC clips, remains a common complication among filter recipients. Possible sequelae of IVC thrombosis include phlegmasia cerulea dolens, venous gangrene, recurrent DVT, PTS, and recurrent PE due to thrombi that extend proximally to the thrombosed filter. In case series of permanent filters, the observed summary frequency of IVC thrombosis varies from 1.6% to 12.2% (see Table 30.3 ). The highest summary frequencies were noted for the Vena Tech LP (12.2%) and Vena Tech filter (9.5%), Angel catheter (11.2%), and VenaTech filters (9.5%). The highest rates from individual studies were noted for the Simon Nitinol filter (up to 50%), the Titanium Greenfield filter (31%), and the Vena Tech LP filter (up to 28%; see Table 30.3 ). Thirteen percent of filter recipients in the PREPIC study experienced symptomatic IVC thrombosis over 8 years of follow-up. However, it is important to note that 35% of filter recipients in the PREPIC study received anticoagulation over the entire duration of follow-up. Therefore it is likely that the rate of IVC thrombosis would have been higher in PREPIC in the absence of anticoagulation. This conclusion is supported by the study of Crochet and associates. They conducted routine radiographic surveillance, including abdominal radiography, duplex scanning, and venacavography in 142 VenaTech filter recipients. IVC occlusion was identified in 22% and 33% of patients after 5 and 9 years of follow-up, respectively. Among the subgroup of patients with PE and anticoagulation failure, the caval occlusion rate was 65%. This study strongly suggests that the frequency of IVC thrombosis is underestimated in many of the published case series.

Filter migration was a significant problem with the Mobin-Uddin filter—a complication that resulted in several deaths. The Society of Interventional Radiology (SIR) Guidelines define migration as movement of the filter from the deployment location by more than 2 cm. Improved anchoring technology has made this event uncommon among recipients of contemporary permanent filters (0.3%). One unavoidable consequence of using hooks to reduce filter mobility is IVC penetration/perforation. Perforation occurs when filter components traverse the IVC wall and enter the pericaval space. Limited penetration of the IVC wall is desired and necessary if the filter is to be anchored at its intended location. Rarely, however, filter components penetrate into adjacent structures and produce clinical consequences (0.3%). The SIR has defined perforation as penetration of filter components greater than 3 mm beyond the caval wall. Small bowel obstruction, duodenal perforation, and retroperitoneal hemorrhage due to penetration of the abdominal aorta or iliac artery are among the reported complications of IVC penetration by filter components. Concomitant anticoagulation has been associated with several instances of bleeding and conceivably may enhance this risk. Therefore IVCFs should be placed carefully during anticoagulation, and the development of abdominal pain in this setting warrants prompt abdominal imaging.

Filter tilting and strut fracture theoretically may contribute to impaired filtration efficiency and thus reduced filter performance in PE prevention. Rogers et al. noted an increase in PE in filter recipients whose filter was tilted more than 14 degrees. The SIR has defined filter tilting as deviation of the filter from the central axis of the IVC by more than 15 degrees. Only a fraction of permanent filter case series have documented tilting (5.3%) and strut fracture (2.7%).

Incomplete opening of the filter after deployment could impair filter clot trapping efficiency and increase the risk of migration. Operator as well as filter defects and the presence of clot in the filter have been implicated as causes of this complication. The frequency of incomplete opening has been estimated in filter case series to be 0.7% to 13.9%.

A recently recognized potential long-term source of complications associated with IVCFs is entrapment of guidewires used to place vascular access catheters. Forceful attempts to remove guidewires have led to a number of filter displacements that in some instances have required filter removal or placement of a second device. One patient died from a cerebrovascular accident following anticoagulation after such an event. Case reports and one experimental study suggest that VenaTech and Greenfield filters are more likely to result in guidewire entrapment. In vitro tests indicate that the Trap­Ease filter may also entrap 3-mm and 1.5-mm J-tipped wires. Design modifications in the VenaTech LP make it much less likely to entrap guidewires and catheters. Simple precautionary measures that may reduce the frequency of guidewire entrapment include prominent documentation of IVCF placement in the medical record and provision of identification bracelets or wallet cards to patients with filters. Use of straight-tipped guidewires, which are less likely to become entrapped than J-tipped wires, and limiting insertion of guidewires to 18 cm (mean distance to the SVC/right atrial junction) minimizes the chances of this complication. If a guidewire does become entrapped, physicians should immediately contact an interventional radiologist or vascular surgeon for safe removal of the wire under fluoroscopic guidance.

Optional (Retrievable) Filters

Over the last decade, there has been a dramatic shift from use of permanent filters to optional vena cava filters. The clinical flexibility afforded by retrievability and the perception of roughly equivalent performance between permanent and retrievable filters are thought to be major factors in this transition. However, it is important to recognize that the data comparing permanent and optional filters derive from retrospective case series, not randomized studies. A compilation of the clinical outcomes derived from case series (including 25 or more subjects) with the Günther Tulip filter, the Optease filter, the Recovery filter, the G2 filter, the ALN filter, the Celect filter, the Option filter, the Crux filter, the Celect Platinum filter, and the Angel catheter can be found in Table 30.4 . Similar to permanent filters, 67 of the 95 optional filter studies (71%; range 0% to 100%) were retrospective in design. In contrast to permanent filters, the mean follow-up was less than 1 year for most filter models, which may contribute to the lower frequency of PE, DVT, and IVC thrombosis reported for many of these devices. Exceptions include the Celect Platinum and Option filters, which were associated with PE frequencies of 7.7% and 4.7%. However, these filters have been studied in a limited number of patients in only a few studies. Similarly, the frequency of DVT tends to be lower with optional filters, except for the Denali filter (13%), Crux filter (9.6%), and the Option filter (8.7%). However, except for the Option filter, these data were derived only from one study. Similar trends are seen in the frequency of IVC thrombosis, reflecting the shorter follow-up and dwell times in the optional filter studies. The Angel catheter, a temporary filter, had higher frequencies of DVT and IVC thrombosis, although these outcomes likely reflect the design of the device, which consists of a conical filter attached to an intravenous catheter that results in a much greater surface area of prosthetic material in the vasculature serving as a nidus for clot formation. There is significant variation in the reason for filter placement among the optional filters, with Optease (76.2%), Recovery (64.3%), and G2 filters (54.7%) being placed primarily for prophylaxis (no evidence of VTE at the time of placement). In contrast, only a minority of ALN (9.8%), Option (12.5%), and Celect Platinum (8.5%) filters were placed for prophylactic indications ( Table 30.5 ). In the published case series, retrieval is attempted in 44% of patients, with the attempted retrieval rate varying from 23% for the Option filter to 67% for the Denali filter. The average in situ time prior to retrieval varied from 18.5 days for the Optease filter to 189 days for the Celect filter. Retrieval was successful in 82% (Optease) to 99% (Celect Platinum) of patients.

TABLE 30.5
Retrieval Results in Studies of Retrievable/Temporary Inferior Vena Cava Filters
Filter Type Patients (Studies) Prophylactic Indication Dwell Time (days) Attempted Retrieval Successful Retrieval Migration IVC Perforation Fracture
Günther Tulip 2772 (23) 1095 (45.8%) 45 1361/2772 (49.1%) 1205/1361 (88.5%) 42/1319 (3.1%) 86/1475 (5.8%) 2/895 (0.2%)
Range 10–133 Range 6%–100% Range 58%–100% Range 0%–9.8% Range 0%–19.5% Range 0%–0.3%
Optease 1216 (10) 716 (76.2%) 18.5 556/1050 (53%) 456/556 (82%) 2/634 (0.32%) 0 3/650 (0.5%)
Range 11–60 Range 13%–100% Range 56%–100% Range 0%–2.7% Range 0%–2%
Recovery 1054 (10) 384 (64.3%) 123 335/1020 (32.8%) 290/335 (86.6%) 28/918 (3.1%) 54/597 (9.0%) 37/906 (4.1%)
Range 20–228 Range 32%–100% Range 65%–94.3% Range 0%–31% Range 0%–75% Range 0%–8.5%
G2 1156 (8) 619/1079 (54.7%) 145 571/1156 (49.4%) 527/571 (92.3%) 60/1004 (6.0%) 134/1004 (13.3%) 9/1004 (0.9%)
Range 1%–100% Range 53–230 Range 19%–100% Range 85%–100% Range 0%–18% Range 0%–44% Range 0%–3.4%
ALN 912 (6) 70/912 (9.8%) 97 451/912 (49.5%) 422/451 (93.6%) 4/912 (0.4%) 0 0
Range 0%–24% Range 51–179 Range 25%–100% Range 78%–99% Range 0%–3.3%
Celect 2308 (13) 247/751 (32.9%) 189 779/1919 (40.6%) 672/779 (86.3%) 28/1396 (2.0%) 515/1567 (32.9%) 4/1651 (0.24%)
Range 0%–100% Range 60–554 Range 19%–91% Range 63%–100% Range 0%–4.2% Range 0%–79% Range 0%–2.1%
Celect Platinum 556 (1) 47/556 (8.5%) 101 (Range 1–445) 155/562 (27.6%) 154/155 (99.4%) 1/335 (0.3%) 65/335 (19.4%) 0
Option 824 (4) 99/824(12.5%) 126 192/824 (23.3%) 174/192 (90.6%) 18/781 (2.3%) 60/781 (7.7%) 0
Crux 125 (1) 52 (41.6%) 168 54 (43.2%) 53 (98.1%) 0 0 0
Denali 394 (3) 87 (22.1%) 154 263 (66.8%) 258 (98.1) 0 9/394 (2.3%) 0
Angel Catheter a 277 (2) 221/277 (79.8) 6.8 277 (100%) 277 (100%) 5/277 (1.8%) 0 0
IVC, Inferior vena cava.

a Temporary filter.

Migration of the filter or filter components has become a serious safety concern since publication of a retrospective case series of strut fracture and migration among 80 patients with Recovery and G2 filters by Nicholson and colleagues in 2010. Using fluoroscopy, they noted strut fracture in 13 patients (16%). Seven of 28 patients (25%) with Recovery filters had at least one strut fracture and migration from the site of filter placement. In five cases (71%), at least one strut migrated to the heart, and three patients suffered ventricular tachycardia and/or cardiac perforation and tamponade. One patient suffered sudden cardiac arrest. Among 52 G2 filter recipients, six patients (12%) had strut fracture and two had asymptomatic strut embolization. Although the fracture rate appeared to be 50% less with the G2 filter compared with the Recovery filter, the authors noted that the mean follow-up period was twofold longer among Recovery filter recipients (50 months vs. 24 months). Therefore the fracture rate was similar for either device. It is important to note that this was a retrospective study of only 80 patients; 109 patients (58%) were lost to follow-up.

However, Vijay and colleagues noted 63 fractured Recovery, G2, and G2 Express filters (11.5%) among 548 patients presenting for filter retrieval between April 2004 and November 2010. The fracture rate increased from 1.9% (4/212) at 6 months to 30.8% (4/13) at 36+ months. Nine of 74 Recovery filters (12.2%) fracture compared with 19 of 350 G2 filters (5.4%) and 4 of 124 G2 Express filters (3.2%). The mean dwell time for fracture filters was 692 days and thus similar to the dwell time noted in the Nicholson et al. study for G2 filters. This study suggests fractured struts are less common with the G2 than the Recovery filter. Tam et al. reported on 266 Recovery filters followed for an average of 18.4 months; 26 limb fractures (9.8%) were noted in 20 patients, and 8 limbs migrated to the pulmonary artery and 1 into the right ventricle. They estimated the fracture rate to be 40% at 5.5 years. No patients suffered symptomatic consequences of filter limb fracture. In case series, strut fracture and migration rates have been substantially lower (0% to 4.1%), although this result may be due to less intense investigation for this complication and shorter follow-up durations (see Table 30.5 ). A literature-based review of filter complications noted a fracture rate of 5.5% to 25% for the Recovery filter, which increased to 39.5% at 60 months of follow-up. They also noted reported fracture rates up to 50% for the Optease filter.

IVC perforation by filter components has been reported to cause retroperitoneal bleeds, gastrointestinal (GI) obstruction, and painful neuropathies. In case series of the different retrievable filters, average IVC perforation frequencies varied from no instances with the Optease or ALN filters up to 32.9%, 19.4% and 13.3% for the Celect, Celect Platinum, and G2 filters, respectively (see Table 30.5 ). A literature-based review of filter complications noted frequencies of IVC perforation that varied from 27% to 100% for the Recovery filter, 26% to 44% for the G2/G2X/Eclipse filters, 25% to 95% for the Simon-Nitinol filter, 22% to 78% for the Gunther Tulip filter, and 22% to 93% for the Celect filter. Prospective cohort studies with routine screening for filter complications are needed to provide better estimates of complication rates for different filters.

What Is the Safety and Efficacy of Vena Cava Filters in Alternative Insertion Sites (Suprarenal Inferior Vena Cava and Superior Vena Cava)?

Rarely, patients with contraindications to anticoagulation develop thrombus in the suprarenal portion of the IVC or in the superior vena cava (SVC). In these instances, vena cava filters have been placed in the suprarenal IVC or the SVC to prevent PE. Stewart and colleagues published the first series of 10 patients who were treated with suprarenal IVCFs. After 16.9 months of follow-up, one patient had suffered a PE. No IVC thrombosis or episodes of thrombosis-associated renal failure were noted. Since then, a number of case reports and case series have been published. The six case series report 277 patients who had suprarenal IVCFs placed. Greenfield filter models were used in 215 patients (77.6%). TrapEase (22, 7.9%) and ALN filters (13, 4.7%) were used less often. A variety of other filter models were each used in a handful of patients. Clinical follow-up was reported for 193 patients (69.7%) for an average of 26.3 months. Sixty-six patients (45.5%) were treated with anticoagulation. Filter-associated complications were unusual. Eight patients suffered PE (4.1%), one of which was fatal. IVC thrombosis developed in four (2%), and two patients developed renal failure (1%). Cancer patients appear to be at particularly high risk for filter thrombosis and subsequent renal failure. Therefore suprarenal filters should not be placed without careful consideration for their risks and benefits.

Proximal upper extremity (UE) DVT is associated with a significantly lower risk of PE than proximal lower extremity (LE) DVT. Nevertheless, SVC filters have been used on occasion to prevent PE in patients with upper extremity thrombosis. A recent review of upper extremity deep vein thrombosis (UEDVT) examined the role of SVC filters in its management. Among 3747 patients with UEDVT, 5.6% suffered a PE, and PE-related mortality was 0.7%. Twenty-one publications reporting 209 patients with SVC filters were identified. During mean follow-up of 29 weeks, eight patients developed filter-related complications, including SVC perforation (6, 2.9%), cardiac tamponade (4, 1.9%), SVC thrombosis (2, 1%), aortic perforation (2, 1%), and pneumothorax (1, 0.5%). The relative infrequency of PE and fatal PE associated with upper extremity thrombosis and the morbidity and mortality associated with SVC filters indicate that filters should not be used for upper extremity thrombosis except in the most extreme circumstances.

Should Patients With Permanent Vena Cava Filters Receive Prophylactic Anticoagulation?

As outlined in previous sections, vena cava filters have been associated with multiple thrombotic complications, including DVT and IVC thrombosis. Thus should all patients with an IVCF receive long-term prophylactic anticoagulation to prevent these thrombotic complications? Anticoagulation is known to be effective in preventing recurrent thromboembolism, but this protection comes at the cost of increased hemorrhagic morbidity and mortality. Systematic reviews of RCTs comparing vitamin K antagonists (VKA) such as warfarin and direct oral anticoagulants (DOAC) have noted that the incidence of major bleeding with DOACs was 1.8 per 100 patient years (95% CI: 1.3 to 2.5) compared with 3.1 per 100 patient-years (95% CI 2.4 to 3.9) for VKA. The incidence of fatal bleeding with DOACs was 0.14 per 100 patient years (95% CI: 0.07 to 0.24) compared with 0.33 per 100 patient-years (95% CI 0.21 to 0.48) for VKA. In a prospective observational study conducted in Canada, Jun et al. noted major bleeding in 3% of DOAC-treated patients and 3.4% of warfarin treated patients over the first 3 months of therapy. If all patients with a vena cava filter in situ were treated with anticoagulation, this practice would place patients at a risk for bleeding over the long term.

In a prospective observational cohort study, Hajduk et al. followed 121 patients with an IVCF treated with therapeutic anticoagulation (INR 2 to 3) for a mean of 3.1 years. The time spent in the therapeutic range was 70%. The incidences of symptomatic PE and DVT were 1.6 and 7 per 100 patient-years, respectively. Two patients (1.7%) developed complete IVC occlusion. Major bleeding occurred in 8 patients (6.6%), comparable to a cohort of patients on anticoagulation without filters (6%). In comparison to the results of the PREPIC study in which only 40% of patients received anticoagulation for the duration of follow-up, Hajduk's study participants had a lower incidence of IVC thrombosis (1.7% vs. 13% in PREPIC) and DVT (20% vs. 35.7%), although follow-up was shorter (3.1 years vs. 8 years). In addition, it is not clear that the VTE risk profile was comparable between the studies. These data suggest anticoagulation may reduce but not eliminate the thrombotic complications associated with IVCFs. Damascelli et al. reported a similar experience in cancer patients with VTE and optional IVCFs. They followed 106 cancer patients who received an IVCF as part of their treatment for VTE. Patients were treated with 40 to 60 mg of enoxaparin twice daily, followed by a VKA (acenocoumarol) adjusted to an INR of 1.5 to 2. After mean follow-up of 320 days, four patients developed IVC occlusion (3.7%), and three developed a PE (2.8%). This study demonstrates that IVCF may be safely used in cancer patients in conjunction with low-intensity anticoagulation. More recently, Mahmood et al. reported outcomes on 154 cancer patients with an IVCF. In those receiving anticoagulation, filter-related complications were lower (OR 0.3, 95% CI 0.1 to 0.7).

The case fatality rate of major bleeding in patients with VTE on anticoagulation with a VKA has been estimated to be 11%; the case fatality rate for recurrent VTE is approximately 11% during the first 3 months of therapy, after which it declines to 3.6%. This hemorrhagic mortality risk translates to an absolute fatal bleeding risk of 0.2% to 0.9% per patient-year, depending on the mode of anticoagulation management used (individual practitioner vs. anticoagulant therapy management clinic). As noted in the DOAC trials, these risks are likely lower with DOACs. Therefore it is important for providers to carefully assess the risk of VTE and bleeding in patients with retained IVC filters. In patients at low risk for bleeding, anticoagulation should be continued particularly in patients with a high risk for recurrent VTE. In contrast, anticoagulation should be discontinued in patients who have risk factors for bleeding or a history of anticoagulation-associated bleeding. In all patients, filter retrieval should be strongly considered if they can tolerate anticoagulation. Until additional data are available, the duration of anticoagulation for patients with IVCFs should be guided by a case-by-case assessment of the risk of bleeding and thrombosis in each patient, and not solely by the presence of a filter.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here

Related Posts