While therapeutic anticoagulation remains the first-line treatment for DVT and PE, patients with proximal DVT who have a contraindication to or proven failure of anticoagulation require an IVC filter [17, 18]. For this group of patients, IVC filter placement provides an effective method of PE prophylaxis that is well supported by level 1 evidence and recent consensus guidelines [18]. Appropriate patient selection that strictly adheres to well-defined insertion criteria will ensure that IVC filters are used responsibly with a favorable risk-benefit ratio [19–21] (Tables 13.1, 13.2, 13.3, 13.4, and 13.5).

The ease of use and retrievability of modern IVC filters have blurred the indications for treatment and lowered the threshold for IVC filter insertion. In response to the dramatic rise in the use of IVC filters, the American College of Chest Physicians (AACP) and the Society of Interventional Radiology (SIR) published indication guidelines specific to retrievable IVC filters [20, 21]. These retrievable filters can be removed when the risk of VTE is reduced and/or anticoagulation therapy is no longer contraindicated.

Debate continues on expanding the use of IVC filters to a wider population of patients outside the absolute indications described above (Tables 13.2 and 13.3). Clinicians in favor of placing IVC filters for relative and prophylactic indications cite the fatal consequences of a potentially preventable PE and the low morbidity of filter placement. Despite its intuitive appeal, the argument for lowering the threshold for IVC filter placement relies on nonrandomized data and observational studies. Ingber et al. reviewed the current clinical evidence and concluded that the only inarguable indication for IVC filter use is in patients with proximal DVT and a contraindication to anticoagulation [22]. In this specific context, a retrievable IVC filter should be placed and then removed as soon as therapeutic anticoagulation is reestablished. According to Ingebar et al., major complication rates for IVC filters equaled or exceeded the incidence of PE in patients without an IVC filter. Objective clinical evidence therefore does not support the liberal use of IVC filters for relative or prophylactic indications.

Absolute contraindications for IVC filter placement involve clinical conditions which preclude safe deployment of the device at its intended position. Examples of absolute contraindications include chronic IVC occlusion, IVC anomalies, inability to access the IVC, and IVC compression. Relative contraindications include a variety of clinical situations which increase the risk for IVC filter-associated morbidity. A large diameter IVC which approaches or exceeds the maximum diameter of the filter constitutes a relative contraindication because of the risk of device migration or dislodgement. IVC filters should also not be used in patients for whom placement of the filter is unlikely to improve their clinical course. A noteworthy example of this contraindication is the placement of a filter in the superior vena cava (SVC). The rationale for placing a filter in the SVC is to prevent PE in a patient with upper extremity DVT and a contraindication to anticoagulation. Not only does SVC filter placement have an unacceptably high periprocedural morbidity rate, patients who receive SVC filters also have limited survival due to the severity of their underlying disease [23]. Although IVC filters have been used in the SVC, no commercially available filter is specifically designed for this location [24]. Reported complications of IVC filters placed in the SVC include SVC perforation, cardiac tamponade, SVC thrombosis, and pneumothorax [25].

Despite the exponential rise in IVC filter use for an expanding list of indications, data on device efficacy remain limited [4]. Even the core principle and absolute indication for IVC placement has not been critically evaluated. To date, no randomized trial or prospective observational study has been performed to evaluate IVC filters as PE prophylaxis in patients with DVT who cannot receive anticoagulation.

The best available data come from a single, large, randomized controlled European trial examining the use of anticoagulation with or without IVC filters as the initial treatment of DVT. The PREPIC (Prevention du Risque d’Embolie Pulmonaire par Interruption Cave) study has been influential in shaping the current recommendations for IVC filter use. PREPIC randomized patients with a DVT to receive permanent IVC filters and anticoagulation vs. anticoagulation alone for at least 3 months. Patients were discharged on oral anticoagulation if not contraindicated. Pulmonary angiograms performed on day 12 showed a new PE in only 1.1 % of filtered patients compared to 4.8 % of anticoagulated patients who did not have a filter (P = 0.03). This difference in PE rate did not translate into a decrease in mortality as survival curves were identical for the two groups of patients. Although 2-year data showed no difference in the incidences of symptomatic PE, major bleeding, and overall mortality between the two treatment arms, PE-related death was 10 times higher (0.6 vs. 6.3 %) in the unfiltered group. Patients who received an IVC filter had an increased rate of recurrent DVT (21 %) compared to patients in the no IVC filter group (11.6 %, p = 0.02) [26]. Of the patients with recurrent DVT in the IVC filter group, 16 patients (9 %) had symptomatic IVC thrombosis.

At 8 years follow-up, less than 50 % of patients in the PREPIC study had received anticoagulation for more than 1 year. There were lower rates of symptomatic PE in the IVC filter group compared to the no filter group (6.2 % vs. 15.1 %, P = 0.008), and a lower rate of fatal PE with a filter (P = 0.014). Despite these differences in PE incidence, mortality was the same for both groups, and a filter was not protective relatively to long-term survival according to multivariate analysis. The long-term clinical cost of an IVC filter was a higher incidence of symptomatic DVT (35.7 % vs. 27.5 %, P = 0.042). Overall IVC filter placement with anticoagulation reduced the risk of PE, but increased the risk of DVT without improving long-term survival. The authors concluded that IVC filter placement should be reserved for patients at high risk of recurrent PE [27].

Although PREPIC stands alone as the only randomized, prospective study on IVC filters, it had several limitations. The study included different types of permanent filters and randomly assigned patients to anticoagulation with unfractionated or low-molecular weight heparin. Long-term anticoagulation was not standardized in terms of drug, dosage, or duration of therapy. Statistical adjustments for multiple comparisons were not performed making the PREPIC data underpowered for subgroup analyses. Although it failed to show long-term survival benefit for IVC filters, PREPIC did show that IVC filters significantly decrease PE. These data provide the best evidence supporting the comparative efficacy of filters while providing an objective analysis of filter complications. Smaller studies have corroborated IVC filter efficacy for PE prophylaxis, as another review also reported a 1.3 % incidence of symptomatic PE after IVC filter placement [28].

Prophylactic indications for IVC filter placement apply to patients who pose a high risk for VTE and cannot receive standard DVT chemoprophylaxis or anticoagulation. Preemptively placing an IVC filter is an attempt to reduce the PE risk among patients who have no other form of protection from VTE. In theory, a prophylactically placed retrievable IVC filter could be removed when the patient’s VTE risk decreased or when the patient could safely receive chemoprophylaxis for DVT. The practice of prophylactic IVC filter placement has gained popularity for several patient populations with varying amounts of supporting evidence.

IVC filters are placed most often with the aid of fluoroscopy. Many patients, especially in the trauma population, have a recent CT scan of the abdomen and pelvis available for preprocedural review of renal vein landmarks and anomalies. Contrast venography performed as part of the procedure provides visualization of anatomic landmarks and accurate, real-time positioning of the IVC filter. Venous access usually involves a femoral or internal jugular approach using a vein uninvolved with clot. Previous studies suggest an increased risk of subsequent femoral vein thrombosis related to femoral vein access, particularly in patients receiving prophylactic filters [70, 71]. The use of an ultrasound guidance to evaluate for clot, rule out abnormal anatomy, and assist in venous cannulation can minimize access-related complications. A micropuncture kit is typically used to cannulate the vein. If there is any suspicion for clot between the puncture site and the infrarenal cava, a venogram should be obtained before proceeding any further to decrease the risk of embolizing clot. Once patency has been confirmed, the micropuncture catheter is exchanged over a wire for a larger sheath, and a long starter wire is advanced into the IVC under fluoroscopy (Fig. 13.2a). For a femoral approach, a catheter is positioned near the L4 level under fluoroscopy. An inferior cavagram is obtained with the patient in Valsalva to maximize visualization of the renal veins and to obtain an accurate diameter measurement of the IVC at this level (Fig. 13.2b). In some cases, an adequate venocavogram requires a power injection of contrast (20 cc/s rate; 40 cc total volume). A marker wire or catheter should be used as a reference to measure caval diameter immediately below the renal veins. Caval diameter smaller than 18 mm [72] should preclude filter placement unless the SafeFlo IVC filter designed for smaller vena cavas is used (16–19 mm). In this clinical scenario, the indications for filter placement should be compelling, if not lifesaving, as the radial force of the filter in a small cava can lead to perforation of the struts. Caval diameter >28 mm precludes the use of most standard retrievable filters and mandates the use of the permanent Bird’s Nest filter (Cook Medical, Bloomington, IN) if filter placement is essential.

Renal vein anomalies, IVC transposition, IVC duplication, and IVC agenesis are anatomic variants important to identify before IVC filter placement. Empiric placement of IVC filters by bony landmarks alone should be discouraged due to the relatively high prevalence of venous anomalies. Contrast venography performed during filter placement provides a roadmap of venous anatomy and helps identify the renal veins to guide placement of the device. The IVC filter is inserted into the sheath and positioned with the superior tip of its cone between the renal veins. Under direct fluoroscopic guidance, the sheath is slowly withdrawn to release and deploy the IVC filter (Fig. 13.2c, d). A completion venogram is performed to ensure proper placement, confirm IVC patency, and evaluate for significant filter tilt (Fig. 13.2f). The clinical scenarios which warrant suprarenal filter placement are listed in Table 13.6. As previously discussed, the larger diameter of the suprarenal cava usually precludes the use of standard retrievable IVC filters.

No randomized controlled trials between different IVC filter models have been performed, and comparative observational studies failed to show superiority of one filter over another. The interventionalist placing the filter should consider the features unique to each type of filter as well as the specific clinical scenario in deciding which type of filter to use. IVC filters are designed in the following categories:

The stainless steel conical Greenfield filter was introduced in 1973 and has been extensively evaluated. The Greenfield filter spans a diameter of 30 mm and traps 3 mm or larger emboli [2, 14, 15, 80]. Initially, there was a high rate of venous access site thrombosis (up to 41 %), due to the 24 Fr delivery system [81]. In 1989, Greenfield filters were constructed out of titanium alloy (nonthrombogenic, nonferromagnetic, resistant to fatigue and corrosion) and delivered via a 12 Fr sheath. Subsequent modifications have minimized technical complications including tilt (<5 %), migration (1 %), recurrent PE (2 %), and access site thrombosis (4 %). Caval patency after Greenfield filter placement is approximately 98 % [15].

The Bird’s Nest filter was introduced in 1982. This permanent filter is delivered by a 12 Fr system. At deployment, two struts are released to attach to the caval wall, and four wires extend in a random “Bird’s Nest” configuration. The unique feature of the Bird’s Nest filter is the struts which span a diameter of up to 60 mm allowing the filter to accommodate large caliber cavas. Unlike other filters, the Bird’s Nest filters rarely perforate the IVC because the devices do not have radially oriented struts [82]. The tighter configuration of the Bird’s Nest filter traps smaller emboli resulting in a low recurrent PE rate of 1–7 %. The downside of the Bird’s Nest geometry is a higher rate vena caval thrombosis (3–19 %) [2, 83, 84].

The Vena Tech LGM (B Braun, Bethlehem, PA) permanent IVC filter is made from biocompatible and nonferromagnetic Phynox (cobalt, chromium, iron, nickel, molybdenum). It maintains a conical shape with a central apex and six radiating struts containing flat side rails with hooks. The Vena Tech LGM accommodates caval diameters as large as 28 mm and is deployed through a 10 Fr delivery from either a jugular or femoral approach. The Vena Tech Low-Profile (LP) (B Braun, Bethlehem, PA) IVC filter is similar to the LGM but uses a 7 Fr delivery system. An experience with both Vena Tech LGM and LP IVC filters has exposed technical problems and complications including incomplete opening of the filter, recurrent PE rates as high as 6 %, and access site thrombosis rates as high as 23 % [85]. Caval patency rates for Vena Tech LGM filters are 92 % at 2 years, but only 67 % at 9 years [86].

The Simon nitinol (Bard Peripheral Vascular Inc., Tempe, AZ) permanent IVC filter was approved in 1990 and is made from nonferromagnetic nitinol. The bi-level conical configuration includes a superior cone with eight overlapping loops in radial array from the apex, while the inferior cone has six limbs in radial array from its apex, each with a hook to engage the caval wall. The bi-level configuration allows entrapment of large emboli in the inferior limbs and of small emboli in the superior dome [87, 88]. It can accommodate caval diameters as large as 28 mm. Long-term studies report recurrent PE rate of 4 % and vena caval thrombosis of 4 %. The 7 Fr delivery system is appropriate for femoral, jugular, and antecubital approaches with a reported 4 % rate of access site thrombosis [89].

The TrapEase (Cordis, Miami, FL) permanent IVC filter was introduced in 2000. Its unique symmetric trapezoidal double-basket designs connects a cephalic conical basket to its mirror image pointing caudally through six side struts, each with hooks providing filter attachment to the caval wall. A 6 Fr delivery system deploys this 50 mm long and 35 mm wide IVC filter through antecubital, jugular, and femoral approaches. The OptEase (Cordis, Miami, FL) IVC filter is similar to the TrapEase design with the addition of a hook placed on the caudal apex and unidirectional anchoring barbs to allow retrievability. Based on initial safety and efficacy data on retrievals, the FDA approved OptEase IVC filter retrieval up to 14 days after placement [90].

TrapEase and OptEase have high technical success rates with limited filter migration, insertion site thrombosis, and filter fracture rates. In a large single-center experience, symptoms of PE developed in 8 % of patients with TrapEase filters, but only 0.1 % developed into fatal PEs. The double-basket filter design has generated concern that TrapEase and OptEase filters have a higher risk of vena cava occlusion. In animal models, trapped emboli in the TrapEase filter slow blood flow between the inferior basket of the filter and the caval wall. This stagnant blood flow promotes further thrombosis. Schutzer et al. reported symptomatic vena caval thrombosis of 1.5 %, and Corriere et al. reported TrapEase filters had an increased risk for vena caval thrombosis compared to other filter designs [91, 92]. In a large study, venography before retrieval confirmed intrafilter thrombus in 22 % of TrapEase filters; however, the study failed to make a clear distinction between captured emboli and thrombus formation [93].

The Recovery G-2 (Bard Peripheral Vascular Inc., Tempe, AZ) permanent IVC filter was modified from the original Recovery nitinol IVC filter. Approved in 2006, the 41 mm length filter is conical in design with six radiating stabilizing arms and six separate radiating fixation legs with hooks for anchoring. In 2008, the Recovery G-2 IVC filter and G-2 Express filter incorporated a hook at the apex of the conical filter to allow for retrievability. These filters are delivered through a 7 Fr low-profile system and can accommodate caval diameters up to 28 mm. Early reports cite PE rates of 2.8 % and no cases of caval thrombosis [94]. The EVEREST trial recorded a 95 % successful retrieval rate of Recovery G-2 filters with a mean indwelling time of 140 days and a maximum of 300 days [95]. In 2010, the FDA approved the Eclipse, an electropolished finish version of the G-2 Recovery IVC filter (Bard Peripheral Vascular Inc., Tempe, AZ). The electropolish creates an ultrasmooth finish to prevent micro-imperfections.

The nonferromagnetic Gunther Tulip (Cook Medical, Bloomington, IN) IVC filter was approved in 2000 for permanent use and in 2003 for retrieval. The conichrome filter is conical with four filters legs with barbed anchoring hooks radiating from the apex and four additional wires wrapped around these legs. The Gunther Tulip filter is 50 mm long and 30 mm wide and delivered with an 8.5 Fr system from the femoral approach and a 7 Fr system from the jugular approach. A hook at the apex allows removal even after long dwell times of greater than 180 days [96]. Rates of recurrent PE rates are 1–2 %, filter tilt 4–11 %, IVC penetration 2–4 %, and caval thrombosis 5–7 % [97–101].

An updated model of the Tulip IVC filter is the Celect (Cook Medical, Bloomington, IN), approved in 2007 for permanent use and in 2008 for retrievable use. Changes to the tulip structure included eight secondary legs to improve stabilization and centering and a wider base for use in caval diameters up to 30 mm. Although the Celect IVC filter provides PE protection and retrievability (symptomatic PE rate of 2.8 %; successful retrieval 89 % at 52 weeks), it also demonstrated a high incidence of caval penetration by a filter leg [102, 103].

Recently, approved IVC filters include ALN (ALN Implants Chirurgicaux Ghisonaccia, France), SafeFlo (Rafael Medical Technologies, Caesarea, Israel), Crux (Crux Biomedical, Menlo Park, CA), and Option (Rex Medical/Argon Medical Devices, Athens, TX). The ALN filter is designed from nonferromagnetic stainless steel in a conical shape with three additional long curved legs radiating from the apex to facilitate self-centering. After a median follow-up of 21 months in a small group of patients receiving the ALN filter, no recurrent PE or vena caval thrombosis rates were reported [104]. Successful ALN filter retrieval rates dramatically decrease after 3 months; only 50 % of IVC filters were retrieved with failures due to adherence to the wall and filter tilt [105]. SafeFlo IVC filter has a non-tilt spiral shape oversized to a double-ring platform. The available filter sizes correlate with the caval diameter: small, 16–19 mm; medium, 19–22 mm; and large, 22–25 mm. Data on the long-term safety and efficacy of the SafeFlo filter remains limited. The novel design of the Crux filter includes two nitinol spiral parts crimped at the ends to form a double-looped helical structure. A filter mesh is attached to one loop, while three fixation anchors are attached to the opposite loop. Delivered through a 6 Fr system, two sizes are available: one for small cavas (17–22 mm) and one for large cavas (22–28 mm). The symmetric design of the Crux filter allows retrieval using a jugular or femoral approach [106]. The Option IVC filter is laser cut from one single piece of nitinol. Its conical shape has six struts radiating from the apex and includes a caudal hook for retrieval. Although it only requires a 5 Fr delivery system, the Option filter can be used in a maximal caval diameter of 32 mm. In a multicenter, single-arm clinical trial, 8 % of patients had recurrent PE, 2 % of patients had filter migration, and 3 % had caval thrombosis [106].