Applied Radiology

Pulmonary thromboembolism (PE) is a common condition and represents a major cause of morbidity and mortality. The true incidence has not been exactly established, but it is estimated that between 570,000 and 750,000 cases of PE occur annually in the United States. The fatality rate resulting solely from PE is estimated to be approximately 50,000 patients, and PE is thought to be a contributing factor of mortality in 20,000 to 150,000 patients.1,2

Systemic anticoagulation is the therapeutic mainstay for patients with PE.3,4 However, anticoagulation therapy has been reported to fail in some patients. In a metaanalysis of patients treated with anticoagulation, the rate of fatal PE among patients presenting with DVT was 0.4%, and the rate of fatal PE for patients presenting with PE was 1.5%.5 The complication rate associated with anticoagulation also has been reported to be as high as 26%, with a fatality rate between 5 and 12%.6,7 In addition, contraindications for anticoagulation exist in certain patients (table 1).

Approximately 80 to 90% of acute PE cases originate from thrombus migrating from the pelvic or lower extremity veins. Therefore, inferior vena cava (IVC) filters have been designed to prevent the migration of thrombus into the lung. It is estimated that, in the United States, 30,000 to 40,000 IVC filters are being placed annually.7

Indications for filter placement

Inferior vena cava filters are effective in preventing PE. However, their implantation also results in significant costs, as well as the potential for serious complications. Therefore, it is mandatory to establish the presence of deep venous thrombosis (DVT) or PE and to document the need for an IVC filter. The presence of DVT in the absence of a contraindication to anticoagulation is not an indication to place a filter. A recent, randomized trial compared the usefulness of IVC filtration versus anticoagulation in patients with DVT who were at risk for PE. This study showed that patients who had filters placed had an initial beneficial effect in the prevention of PE, but this was counterbalanced by an excess of recurrent DVT. In addition, there was no difference in mortality between the patients who received a filter and those patients who were anticoagulated.8

Absolute indications-The absolute indications for IVC filter placement are: 1) contraindication to conventional anticoagulation (table 1); 2) recurrent PE despite adequate anticoagulation; 3) significant complication of anticoagulation treatment; and 4) massive PE resulting in hemodynamic compromise.

Relative indications-Relative indications for IVC filter placement include: 1) large, free-floating thrombus in the IVC or iliac veins; 2) prophylactic placement in high risk patients; (i.e. patients status post-pulmonary embolectomy, DVT in patients with underlying severe cardiopulmonary disease9); 3) presence of any relative contraindication to anticoagulation; 4) non-compliance with anticoagulation treatment.

Controversial indications-Some indications for placement of an IVC filter currently are controversial, including the prophylactic use of filters in patients without DVT or PE but who are at high risk due to neurosurgical or orthopedic procedures or trauma,10-13 and patients with incurable malignancy.14,15

Contraindications

Contraindications for placing an IVC filter are not well-defined. Certainly the risk of filter insertion has to be balanced with the morbidity and mortality associated with PE. Because up to 30% of patients with acute PE who do not receive appropriate treatment die secondary to a complication from recurrent PE, no absolute contraindication for insertion of an IVC filter exists. However, potential risks and benefits always have to be thoroughly considered, especially in the following situations:

Uncorrectable coagulopathy-This can be an occasional finding in patients after extensive trauma or with liver or multisystem failure. We generally correct preexisting coagulopathies whenever possible. In patients where the coagulopathy is not correctable, the Simon Nitinol filter, which uses the smallest introducer system (9 F) and allows placement from an antecubital fossa approach, is the filter that we prefer to use.

Uncontrolled, untreated bacteremia-Although not a preferred practice, IVC filters can be inserted in patients with uncontrolled bacteremia. However, the assistance of an infectious disease team and the immediate institution of antibiotic treatment are mandatory in this setting. It has been shown in animal experiments that appropriate antibiotic therapy will sterilize septic emboli caught in IVC filters.16

Pediatric patients-Indications for filter placement in young patients should be extremely strict, as the long-term effects of filter placement, as well as the mechanical durability of the devices, are not well known.

Types of inferior vena cava filters

There are currently 6 IVC filters that are FDA approved: the standard 24-F stainless steel Greenfield filters (SGF); the newer, 12-F, over-the-wire stainless steel Greenfield Filter (SGF-OTW); the titanium Greenfield filter (TGF); the Simon Nitinol filter (SNF), the Vena Tech filter (VTF) and the bird's nest filter (BNF). All these devices are designed for permanent implantation. Naturally, all have individual strengths and weaknesses.

Standard stainless steel Greenfield IVC filter (SGF)- The stainless steel 24-F Greenfield filter (Meditech, Boston Scientific Corp., Natick, MA) was introduced in 1973. It was primarily designed for surgical implantation through a venotomy, although percutaneous implantation is possible.17 The filter has to be delivered through a 24-F (29.5-F outer diameter) sheath using a stiff delivery capsule. The filter is made from stainless steel, and thus can generate significant artifacts during magnetic resonance imaging (MRI).

This filter has the longest track record and has been shown to be fairly reliable and durable. Recurrent PE and IVC patency rates for this filter range from 2 to 5% and 90 to 98%, respectively.9,17-19 However, the papers reporting these data generally lack adequate follow-up, so that the true incidence for recurrent PE or IVC occlusion remains unknown. Complications have frequently been reported with this device (figure 1). There is a high incidence (up to 41%) of insertion site thrombosis, with one reported case of jugular vein thrombosis resulting in fatal sigmoid sinus thrombosis and death.17,20 Other complications reported with this filter include vena cava perforation (in up to 70% of patients followed with CT) and filter migration (seen in 30 to 49% of patients, and usually in the caudad direction).21-23 This filter also has been shown (in vitro) to have decreased clot trapping capabilities if tilted more than 15 degrees in relation to the long axis of the IVC.24

Over-the-wire stainless steel Greenfield filters (SGF-OTW)- A newer version of the SGF, the over-the-wire filter, was designed for percutaneous implantation (figure 2). It requires a 12-F introducer system (15.6-F outer diameter) and is intended for placement over a .035" super stiff guidewire. The deployment mechanism is simple and allows very accurate positioning of the filter. In order to minimize the chance for clot formation on the legs of the filter, we infuse saline through the delivery device during filter insertion. The device can not be used in patients with an IVC larger than 28 mm in diameter. It is MRI safe, but causes moderately large artifacts. In vitro studies have shown that magnetic forces and torque do not cause filter migration. As with the larger system, tilting beyond 15 degrees adversely affects the efficacy of the filter (figure 3). Patency and effectiveness rates are similar to the SGF.25 Compared to the TGF, the SGF-OTW has been shown to have similar degrees of tilting, but the pattern of the struts was more uniformly symmetric with the SGF-OTW.26

Titanium Greenfield filter (TGF)- The TGF (Meditech/Boston Scientific Corp.) device is constructed of a titanium alloy which is MRI compatible. It has a conical design that is similar to both of the stainless steel Greenfield filters (figures 4,5). The TGF is designed for percutaneous insertion and requires a 12-F (14.3-F outer diameter) introducer sheath. Its deployment mechanism is simple and allows for precise positioning. The filter carrying capsule is rigid, and kinking of the introducer sheath during its insertion in tortuous iliac veins has been reported (figure 6). As the device is not inserted over a guidewire, significant tilting within the IVC can occur (figure 5); this occurs more often when the device is placed from a left common femoral vein approach. Earlier versions of the device had a high caval perforation rate (figure 7), but modifications in the design of the leg hooks have reduced the frequency of this complication.27-29 Other problems reported in conjunction with this device include incomplete opening of the filter with or without crossing of the filter legs at the time of deployment (figure 8A). When this occurs, careful manipulation of the filter legs can be attempted using steerable guidewires or catheters in order to correct this problem (figure 8B). However, care should be taken not to dislodge the filter. Alternatively, a second filter can be placed in a more cephalad position. The incidence of incomplete opening can be reduced if heparinized saline is continuously infused through the delivery device during filter deployment.

The TGF should be placed when the vena cava is smaller than 28 mm in diameter. The rate of PE recurrence has been reported to be 3.3%. In one series, the caval patency rate at 30 days of follow-up was reported to be 100%. However, we have seen a number of patients with IVC thrombosis in the presence of a TGF. Insertion site thrombosis rate has been reported to be 8.7%.28,30

LGM or Vena Tech filter -The Vena Tech or LGM filter (B. Braun/Vena Tech, Evanston, IL) was introduced in France in 1986. The filter is a self-centering device which is conically-shaped, with stabilizing side rails. The filter is made of phynox, a non-ferromagnetic alloy, and is MRI compatible.31 The total filter length is 38 mm and its base diameter is 30 mm, which restricts its use to caval sizes of 28 mm or less.32 Placement of this device requires a 12-F introducer sheath (14.6-F outer diameter). The deployment mechanism is simple and allows for fairly precise filter placement (figure 9). Incomplete opening of the device can be seen; however, this problem can be minimized by constantly infusing heparinized saline through the lumen of the pusher rod. The hooks on the device prevent cephalad migration. However, caudal migration can occur, though this is usually limited to less than one vertebral body height. The caval patency rate is 92% at one year post-filter placement,33 and 70% at 6 years.34 The recurrent PE rate is 0 to 3.8%.33,34 Tilting up to 15 degrees has been described in 2 to 9% of filters placed, but does not seem to be clinically significant; tilting greater than 15 degrees has not been reported. Incomplete opening of the device was initially a problem when it was deployed from a jugular vein approach. However, modifications in the filter design have made this problem less frequent.23,35 Access site thrombosis has been reported in 7 to 23% of patients.35 Guidewire entrapment between the flat-wire struts of this filter can lead to filter displacement. Therefore, fluoroscopy should be used whenever placing a central venous line in the presence of this filter.

Bird's nest filter (BNF)- The Gianturco-Roehm bird's nest vena cava filter (Cook Inc., Bloomington, IN) is made from stainless steel and generates the largest magnetic susceptibility artifact of all the filter devices. It also has the potential to torque in a high strength magnetic field. However, the BNF will not get displaced in a 1.5 T magnet.36

The BNF was introduced for clinical trials in 1982, and is currently the only FDA approved device for placement in patients with caval diameters larger than 28 mm (the so-called mega-cavae) (figure 10).37 It consists of 2 v-shaped struts which are connected with four 25-cm

long wire strands.38 The device is inserted through a 12-F (14-F outer diameter) introducer sheath. The delivery mechanism of the BNF is probably the most complex of all the IVC filters, but it is still straightforward once the operator is familiar with the device. Because of the length of the anchoring struts, the filter requires the longest segment of IVC (approximately 7 cm) for appropriate insertion. Therefore, it is more difficult to place in patients who have a short distance between the renal veins and the iliac vein confluence.

The BNF can be inserted from both a jugular or femoral vein approach, with the 2 kits only differing in the length of the delivery catheters (75 vs 40 cm). In its current form, the device has a push button mechanism attached to a 3-finger handle which allows separation of the guidewire pusher from the filter. This filter delivery system also is prone to kinking in patients with very tortuous iliac veins, especially if placement from the left common femoral approach is attempted. Phlegmasia cerulea dolens resulting in death has been reported in association with the BNF.39 The rate of symptomatic IVC occlusion has been described to be increased in cases where the wires prolapse beyond the level of the cranial anchoring struts into a suprarenal location.40 Rotating the delivery sheath 360 degrees 2 or 3 times during deployment of the filter wires results in a reduction in wire prolapse.41 Several cases of caval perforation have been described, and access site thrombosis has been found to occur in about 2% of cases.42,43 In one study, the IVC occlusion rate was reported to be 2.9%, and the rate of recurrent PE was 2.7%. However, this study lacks systematic follow-up and, therefore, the real numbers are unknown.38

Simon nitinol filter (SNF)- The SNF (Bard Radiology, Covington, GA) was introduced in 1988 and received FDA-approval in 1990. It is made of a nickel-titanium alloy that has a thermal memory. At temperatures below 27 degrees Celsius, this alloy material becomes malleable, thus facilitating its insertion through tortuous vessels.44,45 The filter is MRI compatible and causes only minimal artifacts. It consists of a cloverleaf dome and six conically arranged legs. Both the legs and the dome provide filtration.

The SNF filter has the smallest introducer sheath (9-F outer diameter) which, in combination with its flexibility, allows placement via an antecubital vein approach (figure 11). The filter measures more than 8 cm in length in the delivery system, but foreshortens to about 4.5 cm upon deployment. This foreshortening causes difficulty in precisely positioning this device, as compared to the other filters. Occasionally, it can take several hours for the filter to form to its final shape (figure 12). Because the IVC can enlarge during induction of general anesthesia, it is recommended that insertion of the SNF be limited to IVCs with a maximum diameter of 24 mm if the patient is to undergo general anesthesia within 2 weeks of filter placement. Otherwise, the SNF can be used if the IVC is 28 mm or less in diameter.

In-vitro studies have shown that this filter has superior clot trapping performance when compared to the SGF. Tilting does not affect the efficiency of the SNF.22,45,46 The IVC patency rate is reported to be between 50 and 91%.23,47 Recurrent PE occurs in 0 to 4.8% of patients with a SNF;48,49 asymptomatic caval penetration by filter legs has been reported in up to 25% of patients.48,49 Filter migration has also been encountered, but seems to be rare (1.7%).23

Filter placement

Patient preparation -Informed consent should be obtained for all IVC filter placements. A platelet count, activated partial thromboplastin time (PTT) and prothrombin time (PT) should be available. Correction of any coagulation disorders and discontinuation of any heparin use at the time of the procedure is recommended. The platelet count should be at least 50,000, and preferably 100,000. Patients should be NPO or on clear liquids for at least 4 hours prior to the procedure. We generally perform the procedure using conscious sedation. An IVC gram also should be obtained prior to the procedure in order to evaluate for IVC patency, extent of thrombus, diameter of the IVC, location of the renal veins and iliac confluence, and venous anomalies. If the patient is allergic to iodinated contrast agents or has impaired renal function, carbon dioxide can be used as an alternative contrast agent. If the inflow from the renal veins is not well delineated or there is concern for an accessory renal vein (figures 2,9), the renal veins can be selected with a catheter. Spot films to document the level of the renal veins can then be taken without the need to perform venography. Without an IVC gram, either ultrasound or MRI is mandatory in order to document caval patency or to exclude a venous anomaly.50

The IVC is an elliptical vessel, with its greatest diameter being in the left to right direction. Therefore, we measure the IVC diameter on an AP IVC gram. This can be facilitated by the use of a calibrated pigtail catheter designed for IVC filter placement. These catheters (PIG-CAVA, Cook Inc., Bloomington, IN) have metal markers that are 28 mm apart (figure 12A). Alternatively, a radiopaque ruler can be positioned under the patient. The diameter of the IVC can vary depending on the hydration status of the patient and can increase during a Valsalva maneuver. IVC diameters greater than 28 mm (corrected for magnification) require placement of a BNF.

Ideally, the filter should be placed immediately below the level of the renal veins, with the tip of the conical filters (SGF, TGF, and VTF) extending to the level of the inflow of the renal veins. Suprarenal placement of an IVC filter has been described in patients with renal vein thrombosis, those with extensive thrombus in the infrarenal IVC (figure 4), renal transplant recipients, pregnant women, and women of child bearing age. In patients with recurrent PE from upper extremity or internal jugular vein thrombosis, placement of filters in the superior vena cava can become necessary and has been described.51

The route preferred by many interventionalists for vena cava filter insertion is the right common femoral vein, followed by the right internal jugular vein. The latter offers the advantage of the straightest access to the IVC and results in the least amount of tilting of the conical filter devices. If the femoral access is chosen, care should be taken to enter the vein at the level of the mid-femoral head, as lower punctures can result in transarterial puncture of the vein and the development of arteriovenous fistulae. This is due to the fact that the proximal superficial femoral artery is frequently positioned anterior to the vein below the level of the femoral head (figure 13).52 This complication can be minimized by the use of a single wall puncture needle.

Once the femoral vein is entered, we place a 5-French dilator and perform a limited iliac venogram to assess venous patency. An AP IVC gram is then performed using a calibrated pigtail catheter. Placement of a ruler under the patient facilitates accurate placement of the device. Alternatively, bony landmarks can be used as reference points. After defining the level of the renal veins and the iliac confluence, the appropriate device is chosen and placed according to the manufacturer's description.

In patients with tortuous iliac veins or a "deep pelvis", it should be kept in mind that the delivery capsules for the TGF, SGF, VTF, and BNF are relatively rigid and prone to kinking the delivery sheath (figure 6). If difficulty is encountered while trying to advance a filter through a tortuous iliac vein, the operator should stop. Once the introducer sheath is kinked, the device can not be delivered. Several options exist at this point: 1) place a bolster under the patient's buttocks to reduce the curvature of the pelvis; 2) use an Amplatz super stiff wire (Meditech/Boston Scientific Corp.) adjacent to the introducer sheath to straighten the vein; and/or 3) advance the sheath and the filter as a unit across the tortuous bend in the iliac vein (being careful not to force the sheath through the wall of the IVC). Otherwise, the SNF, which is the most flexible device, can be used.

A post-placement IVC gram should be performed through the delivery sheath in order to document filter position. All the filters allow passage of catheters and guidewires through their interstices, but dislodgement or displacement of the filtration device remains a risk. Of the currently available IVC filters, the VTF and the SGF-OTW seem to carry the greatest risk for guidewire entrapment and possible displacement when performing transfilter procedures.53

Timing of filter placement -Once the diagnosis of PE or DVT is made, timing of filter insertion often is debated. Jones et al have proposed guidelines and introduced three urgency classes. Emergent placement (within the next 12 hours) is indicated in patients with acute PE or free-floating iliofemoral DVT in whom anticoagulation is contraindicated or has failed, in patients with cor pulmonale or other severe cardiorespiratory compromise, and in patients with septic emboli. Urgent filter placement (the next available slot) is recommended for patients with a contraindication to anticoagulation and DVT (not free-floating). Elective placement is reserved for cases where filters are placed prior to surgical procedures in high-risk patients.54 Despite these recommendations, except for elective situations, we attempt to place IVC filters as soon as the need occurs, as the morbidity from recurrent PE appears to be greatest within the period immediately after the diagnosis.

Comparison of filters

The SNF has the smallest introducer sheath. It is the most flexible device and it allows insertion from many access sites, including the antecubital fossa. The delivery system should be flushed with heparinized saline during the procedure. However, the newer model does not require chilled saline, as did the older system. If the patient is to undergo general anesthesia within the next 2 weeks, the filter should be restricted to an IVC not larger than 24 mm in diameter.

The BNF is the only filter that can be used in the so-called mega-cavae (IVCs larger than 28 mm).37 The alternative to using a BNF in mega-cavae would be biliac filter placement. The maximum recommended diameter for the BNF is 40 mm. This device also has been used as a temporary filter.55

The Greenfield and the VTF filters are the easiest devices to deploy and they also allow very accurate positioning. These filters should have saline infused during deployment in order to minimize clot formation on the legs of the filter (which could result in incomplete opening of the devices).

Patients with IVC filters can undergo MRI. However, the BNF and stainless steel Greenfield filters create large artifacts. The least amount of magnetic susceptibility artifacts can be found with the TGF, SNF, and VTF filters (table 2).

All filters have shown high efficiency rates in preventing recurrent PE.24,46,48,56 However, access site and IVC thrombosis can occur with all of these devices. The true incidence of these events is unknown due to the lack of well-controlled studies and follow-up. Wide variations are reported in the literature, but it is likely safe to say that significant thrombosis at the venous access site occurs in approximately 5 to 10% for all filters. IVC thrombus formation seems to occur in approximately 10 to 20% of all patients; however, only a fraction of these patients become symptomatic.23,57

Other complications such as filter migration, IVC penetration, and filter fracture have been described for all devices, but their true incidences are unknown. The extent of tilting seems to affect the clot-trapping capability of the Greenfield filters in vitro. Tilting

has been shown to occur with both the over-the-wire and not over-the-wire systems.25,26,58 However, the clinical significance of tilting in vivo is poorly defined.24 Tilting of a conical-shaped filter can be minimized by placing the filter from a right internal jugular approach.59 The left common femoral and left internal jugular vein approaches are known to result in the largest degree of tilting and should therefore be avoided whenever possible. Alternatively, other filters should be used if these approaches become necessary.

New developments

Temporary (removable) IVC filters are currently being used in Europe and are undergoing clinical trials in the United States. They have been found to be beneficial in patients who are in need of short term caval filtration due to a temporary risk for anticoagulation, such as the postoperative or post-trauma state. Occasionally, these devices have been used in conjunction with thrombolysis of the IVC or iliac veins in order to reduce the risk of embolizing large clot volumes into the lungs (figure 14). Temporary filters need to be removed within 2 to 3 weeks because after that time neointima formation causes them to adhere to the vessel wall. Removing a temporary filter may be difficult if it is filled with thrombus or if a significant risk for PE persists.

Another class of filters currently under development are retrievable filters. These filters can be left in place permanently if necessary, but are retrievable during the first 2 or 3 weeks.60-62

Conclusion

Each filter has its own advantages and disadvantages. Therefore, radiologists involved with the placement of IVC filters should be familiar with all of these devices in order to be able to customize the use of each device for individual patients. For instance, in patients with a large caliber IVC, only the BNF should be utilized. In cases where only a short segment of infrarenal IVC is available or where precise positioning is crucial, the SGF-OTW or VTF are preferable. For an IVC of less than 18 mm in diameter, the SNF may not form immediately or may "kink" the IVC during formation (figure 12). For tortuous veins or limited access, the flexibility and small caliber sheath of the SNF are major advantages.

The systematic use of filters in patients with DVT in the absence of a contraindication for anticoagulation can not be recommended. AR

References

1. Schuman LM: The epidemiology of thromboembolic disorders: a review. J Chronic Dis 18:815-845, 1965.

2. Freiman DG, Suvemoto J, Wessler S: Frequency of pulmonary thromboembolism in man. N Engl J Med 290:1278-1286, 1965.

3. Coon WW, Willis PW: Deep venous thrombosis and pulmonary embolism: prediction, prevention, and treatment. Am J Cardiol 4:611-621, 1959.

4. Goldhaber SZ: Strategies for management. In: Goldhaber SZ (ed): Pulmonary Embolism and Deep Venous Thrombosis, pp 121-133. Philadelphia, WB Saunders, 1985.

5. Douketis JD, Kearon C, Bates S, et al: Risk of fatal pulmonary embolism in patients with treated venous thromboembolism. JAMA 279:458-462, 1998.

6. Petett DB, Strom BL, Melnic KL: Duration of warfarin anticoagulation therapy and the probabilities of recurrent thromboembolism. Am J Med 81:255-259, 1986.

7. Magnant JG, Walsh DB, Juravsky LI, Cronewett JL: Current use of inferior vena cava filters. J Vasc Surg 16:701-706, 1992.

8. Decousus H, Leizorovicz A, Parent F, et al: A clinical trial of vena caval filters in the prevention of pulmonary embolism in patients with proximal deep-vein thrombosis. N Engl J Med 338:409-415, 1998.

9. Greenfield LJ, Michna BA: Twelve-year clinical experience with the Greenfield vena caval filter. Surgery 104(4):706-712, 1988.

10. Vaughn BK, Knezevich S, Lombardi AVJ, Mallory TH: Use of the Greenfield filter to prevent fatal pulmonary embolism associated with total hip and knee arthroplasty. J Bone Joint Surg Am 71:1542-1548, 1989.

11. Darcy MD, Smith TP, Hunter DW, et al: Short-term prophylaxis of pulmonary embolism by using a retrievable filter. AJR 147:836, 1986.

12. Maynar M, Martin M, Cabrera P, et al: Percutaneous introduction of inferior vena cava filters as prophylaxis for pulmonary embolus. Semin Intervent Radiol 2:156-160, 1988.

13. Khansarinia S, Dennis JW, Veldenz HC, et al: Prophylactic Greenfield filter placement in selected high-risk trauma patients. J Vasc Surg 22:231-236, 1995.

14. Hubbard KP, Roehm JO Jr., Abbruzzese JL: The bird's nest filter, an alternative to long-term oral anticoagulation in patients with advanced malignancies. Am J Clin Oncol 17(2):115-117, 1994.

15. Cohen JR, Grella L, Citron M: Greenfield filter instead of heparin as primary treatment for deep venous thrombosis of pulmonary embolism in patients with cancer. Cancer 70:1993-1996, 1992.

16. Peyton JW, Hylemon MB, Greenfield LJ, et al: Comparison of Greenfield filter and vena caval ligation for experimental septic thromboembolism. Surgery 93:533-537, 1983.

17. Rose BS, Simon DC, Hess ML, Van Aman ME: Percutaneous transfemoral placement of the Kimray-Greenfield vena cava filter. Radiology 165:373-376, 1987.

18. Golueke PJ, Garrett WV, Thompson JE, et al: Interruption of the inferior vena cava by means of the Greenfield filter: expanding the indications. Surgery 103:111-117, 1988.

19. Greenfield LJ, Peyton R, Crute S, Barnes R: Greenfield vena cava filter experience, late results in 156 patients. Arch Surg 116:1451-1456, 1981.

20. Kantor A, Glanz S, Gordon DH, Scalafani SJ: Percutaneous insertion of the Kimray-Greenfield filter: incidence of femoral vein thrombosis. AJR 149(5):1065-1066, 1987.

21. Berland LL, Maddison FE, Bernhard VM: Radiologic follow-up of vena cava devices. AJR 134:1047-1052, 1980.

22. Friedell MC, Goldenkranz RJ, Parsonnet V, et al: Migration of a Greenfield filter to the pulmonary artery: A case report. J Vasc Surg 3:929-931, 1986.

23. Savader SJ: Inferior vena cava filters. In: Savader SJ, Trerotola SO (eds): Venous Interventional Radiology With Clinical Perspectives, p 367. New York, Thieme Medical Publishers, Inc., 1996.

24. Thompson BH, Cragg AH, Smith TP, et al: Thrombus-trapping efficiency of the Greenfield filter in vivo. Radiology 172:979-981, 1989.

25. Cho KJ, Greenfield LJ, Proctor MC, et al: Evaluation of a new percutaneous stainless steel Greenfield filter. J Vasc Interv Radiol 8(2):181-187, 1997.

26. Kinney TB, Rose SC, Weingarten KE, et al: IVC filter tilt and asymmetry: Comparison of the over-the-wire stainless-steel and titanium Greenfield IVC filters. J Vasc Interv Radiol 8(6):1080-1082, 1997.

27. Teitelbaum GP, Jones DI, van Breda A, et al: Vena caval filter splaying: Potential complication of use of the titanium Greenfield filter. Radiology 173:809-814, 1989.

28. Greenfield LJ, Cho KJ, Proctor M, Bonn J, et al: Results of a multicenter study of the modified hook-titanium Greenfield filter. J Vasc surg 14(3):253-257, 1991.

29. Greenfield LJ, Cho KJ, Tauscher JR: Evolution of hook design for fixation of the titanium Greenfield filter. J Vasc Surg 12(3):345-353, 1990.

30. Greenfield LJ,