- Email: [email protected]
Evaluation of a New Percutaneous Stainless Steel Greenfield Filter
Evaluation of a New Percutaneous Stainless Steel Greenfield F'ilterl Kyung J. Cho, MD Lazar J. Greenfield, MD Mary C. Proctor, MS Lisa A. Hausmann, BS Joseph Bonn, MD Bart L. Dolmatch, MD David J. Eschelman, MD Pamela A. Flick, MD Thomas B. Kinney, MD M. Victoria Marx, MD David R. McFarland, MD Stephen K. Ohki, MD S. Osher Pais, MD Steven K. Sussman, MD Arthur C. Waltman, MD Index terms: Embolism, pulmonary * Venae cavae, filters
JVIR 1997; 8:181-187 Abbreviations: FDA = Food and Drug Administration, SGF-AH = stainless steel alternating hook Greenfield filter, TGF-MH = modified-hook titanium Greenfield filter
From the Departments of Radiology (K.J.C., M.V.M.) and Surgery (L.J.G., M.C.P.), University of Michigan, 1500 E. Medical Center Dr, Ann Arbor, MI 48109; Boston Scientific Corporation, Natick, Massachusetts (L.A.H.); the Department of Radiology, Thomas Jefferson University Hospital, Philadelphia, Penn (J.B., D.J.E.); the Vascular and Intewentional Radiology, The Cleveland Clinic Foundation, Cleveland, Ohio (B.L.D.); the Department of Radiology, University of Tennessee Medical Group, Memphis, Term (P.A.F.); the Department of Radiology, UCSD Medical Center, San Diego, Calif (T.B.K.); the Department of~adioiogy,University of Arkansas for Medical Science, Little Rock, Ark (D.R.M.); the Department of Hartford Hospital, West Hartford, Conn (E.J.O., S.K.S.); the Department of Radiology, University of Maryland Hospital, Baltimore, Md (S.0.P.); and the Departmerit of Massachusetts General Hospital, Boston, Mass (A.C.W.). From the 1996 SCVIR annual meeting. Received ~~~~~t19, 1996; revision requested September 19; revision received and accepted November5. Address correspondence to K.J.C.
o SCVIR, 1997
PURPOSE: To evaluate a new percutaneous Greenfield filter with an alternating hook design and over-the-wiredelivery system. MATERIALS AND METHODS: The alternating hook stainless steel Greenfield filter was evaluated in a prospective clinical trial between March 10, 1994, and January 27,1995. Filters were placed in 75 patients in nine clinical centers and follow-up with radiographs and ultrasound scans was carried out at 30 days. RESULTS: Clinical trial results revealed successful placement in all patients. There were four cases of filter limb asymmetry (5.3%) without clinical sequelae, with one incidence of failure to span the cava. No significant migration was found. There were no clinically suspected pulmonary emboli, but one instance of probable caval penetration (1.7%)did occur. Caval occlusion was documented in three patients (5%). CONCLUSION: The percutaneous stainless steel Greenfield filter provides ease of insertion and improved deployment while maintaining the high standards of efficacy and safety associated with the standard and titanium Greenfield filters.
THE efficacy and safety of the modified-hook titanium Greenfield filter (TGF-MH) (Medi-techBoston Scientific, Watertown, Mass) have been objectively documented in clinical studies (1). However, the quest for the ideal filter led to the development of a percutaneous stainless steel Greenfield filter that would improve the ease of centering and placement. ~h~ new device, termed over-the-wire stainless steel alternating hook Greenfield vena cava filter with 12-F introducer system (SGF-AH; Medi-techBoston Scientific), underwent in vitro and experimental in vivo testing demonstrating that it was as effective as the standard and titanium filters (Greenfield LJ, Proctor MC, personal communication, 1993). The advantage of the over-the-wire delivery system includes passage through tortuous vessels, facilitating insertion from the left femoral vein and alignment of the carrier and vena cava to aid in centering the filter a t placement. The purpose of the study was to evaluate the
percutaneous Greenfield filter with a n alternating hook design and over-the-wire delivery system and to gain U.S. Food and Drug Administration (FDA) approval for the SGF-AH moacation of the TGF-MH.
MATERIALS AND METHODS Description of the Device The SGF-AH is manufactured from the same stainless steel wire as the original stainless steel Greenfield filter, but it differs in two major aspects. The angle of the leg-apex joint is 0" versus 17", which facilitates compressing the filter into a 12-F delivery system (Fig 1). The carrier capsule is delivered through the 15-F sheath. When initial animal testing in sheep demonstrated a tendency for the device to move distally, the hook configuration was redesigned. The hooks were re-curved to a n angle of 80" as with the TGF-MH and two of the six hooks were di-
181
182
Journal of Vascular and Interventional Radiology
March-April 1997
Figure 1. The new over-the-wire stainless steel Greenfield vena cava filter. (a)Lateral view. The filter is 4.9 cm in height and 3.2 cm in diameter at the base. (b) Axial view. The apical aperture of the filter allows passage of a 0.035-inch guide wire.
rected downward a t a n angle of 125" to further stabilize the device. Unlike the TGF-MH, there is an opening in the apex of the SGF-AH to allow passage of a guide wire. The legs are 4.9 cm long, in contrast to the 4.7-cm length of the TGF-MH and the 4.6-cm length of the standard Greenfield filter. The base diameter of the new filter is 3.2 cm as opposed to 3.8 cm for the TGF-MH and 3.0 cm for the standard Greenfield filter. The SGF-AH
Figure 2. Percutaneous placement of an over-the-wire stainless steel Greenfield vena cava filter. The Superstiff guide wire provides proper alignment and centering of the filter in the vena cava during release. (a)Posteroanterior views of filter insertion show inferior vena cavogram (left), carrier capsule exposed beyond the sheath tip over the Superstiff guide wire (middle), and filter in good position (right). (b) Lateral views of filter insertion show inferior vena cavogram (left), carrier capsule exposed beyond the sheath over the Superstiff guide wire (middle), and filter in good position (right).
Cho et a1
183
Volume 8 Number 2
ing radiologists and surgeons held prior to final IDE approval. The focus of the standards was the Patients Enrolled potential for clinical sequelae resulting from either filter movement 23 or penetration. The consensus panel 14 determined that movement greater 16 than 20 mm or placement that 6 resulted in an ineffective final 6 position had potential for clinical 4 3 sequelae. The ineffective final posi2 tion was defined as significant 1 tilting (>15"),incomplete opening, or both. Penetration was defined as 75 extension of one or more filter limbs by 5 mm or more outside of the cava as confirmed with computed turned clockwise for hemostasis and tomographic (CT) scanning. An event rate of greater than 10% in counterclockwise immediately before introduction of the preloaded either of these standards would be deemed unacceptable. This event filter carrier capsule. As the carrier rate limit was based on the results capsule is advanced into the vena of a multicenter study of the TGFcava, the guide wire should be held stationary. After the carrier capsule MH (2). The upper limit 95% confidence interval on the movement has reached the desired implantarate (3.8%) and the penetration rate tion site, the sheath is retracted (3.8%) of that study was 9.1%. This and locked to the handle, exposing the carrier capsule in the vena cava expected range and the clinically acceptable range (10%) were in beyond the tip of the sheath. While the handle is held stationarv. the close agreement. Based on these event rates, a sample size of 50 release tab is unlocked and :lid toward the operator, which retracts patients was selected. It was hythe carrier capsule and exposes the pothesized that penetration or filter. After the filter has been migration would occur in five or deployed, the guide wire is slowly fewer patients enrolled in the study. withdrawn under fluoroscopic guidA total of 75 sequential patients ance because a buckled guide wire from the nine clinical centers, who may occasionally catch in the apex met the study criteria and agreed to or the hook. The carrier is withparticipate, were enrolled. There drawn into the sheath and removed. were 48 men (64%)and 27 women Gentle pressure is applied to the (36%),with a mean age of 57 years puncture site for hemostasis. 2 17 (range, 20-89 years). Patient demographics and indications for placement were deemed comparable Clinical Trial to those in the TGF-MH clinical After approval of the Investigastudy (Table 2). The exclusion tional Device Exemption (IDE) by criteria were ( I ) patients with a the FDA, and of the institutional caval diameter greater than 28 mm; review board of each institution (2) presence of thrombus at the (Table I), this prospective nonranpuncture site, in the iliac vein, or in domized, observational study was the inferior vena cava; ( 3 )patients conducted between March 10, 1994, in whom preplacement cavography and January 27, 1995, at nine could not be performed; (4) patients institutions. The primary outcome in whom pregnancy had been convariables to be evaluated a t 30 days firmed; (5)patients for whom inafter placement included filter formed consent could not be obmovement and caval penetration. tained; (6) patients in whom a The performance standards used in coexisting medical disorder prethis study were developed during cluded them from filter placement; the consensus meeting of participat- and (7) patients who were unwilling
Table 1 Patient Enrollment at Nine Institutions Institution University of Michigan Hospitals University of Maryland Hospital Massachusetts General Hospital Hartford Hospital University of California San Diego University of Arkansas Hospital Cleveland Clinic University Tennessee Yale University Total
is preloaded into the carrier capsule and introduced through a 12-F catheter. There are separate femoral and jugular insertion systems that differ in the orientation of the apex of the filter with respect to the introducer.
Recommended Insertion Procedure The technique for percutaneous insertion of the SGF-AH is similar to that used for the TGF-MH, but the filter is deployed over a guide wire. After percutaneous catheterization of the vena cava from either the femoral or jugular vein with use of the Seldinger technique, a 5-F pigtail catheter is advanced over the 0.035-inch guide wire into the vena cava. After a vena cavogram is obtained, the catheter is removed over a 0.035-inch Arnplatz Superstiff guide wire (Medi-techBoston Scientific), and a 3-4-mm skin incision is made along the guide wire. It is important to advance the floppy portion of the guide wire well beyond the intended implantation site because the stiffer portion of the guide wire is necessary to center the filter in the vena cava (Fig 2). Sequential dilation is then performed, and the 15-F sheath1 dilator system with attached Rotovalve (Medi-techBoston Scientific) for hemostasis is advanced over the guide wire into the vena cava beyond the desired implantation site. The dilator is then withdrawn, leaving the sheath and the guide wire in place. The Rotovalve is
184
Journal of Vascular and Interventional Radiology
March-April 1997
to comply with follow-up requirements as indicated in the study protocol. Vena cavography was performed in all patients before filter placement for the evaluation of caval patency, caval anomaly, and location of the renal vein. After placement, biplane abdominal radiographs were taken in expiration (Fig 2). All measurements of filter base diameter after implantation and at 30-day follow-up, as well as change in filter position a t 30-day followup, were corrected for magnification. The known length of the SGFAH was used to correct for mamification. The distance from the Gp of the apex to the bottom of the hook was 5.2 cm. Filter movement was assessed by measurement of the change in position of the filter apex in relation to the top of the nearest vertebral body for the anteroposterior radiograph on both postimplantation and follow-up films.
RESULTS Placement
The 12-F SGF-AH was successfully deployed in all 75 patients enrolled in the study, for a 100% success rate. All filters were placed with use of the percutaneous insertion technique. The sites for insertion were the right femoral vein in 55 patients (73%), the left femoral vein in 12 patients (16%), and the right internal jugular vein in eight patients (11%). Overall, there were 67 femoral insertions (89%) and The eight jugular insertions (11%). filter locations were infrarenal (between the lowest renal vein and the junction of the right and left common iliac veins) in 74 patients (99%)and suprarenal (between the right hepatic vein and the highest renal vein) in one patient (1%). The procedure-related complications were limited to two femoral insertion site hematomas without clinical sequelae (2.6%). There were four reported cases of filter limb asymmetry in 75 patients (5.3%). In one patient, the
Table 2 Comparability of Patients Characteristics for 12-F SGF-AH versus TGF Category
12-F SGF-AH
TGF*
No. of Placements Age (y) Mean Range Males (%) Indications for Use Contraindication to anticoagulation Recurrent thromboembolism despite anticoagulation Complications of anticoagulation Adjunctive prophylaxis Other Trauma patients (new indication) Filter Location Infrarenal Suprarenal At level of renal vein Superior vena cava
75
181
57 20 - 89 48 (64%)
58 17 - 92 115 (64%)
49 (65%)
136 (75%)
7 (9%)
17 (9%)
3 (4%) 8 (11%) 6 (8%) 6 (8%)
19 (10%) 20 (11%) 15 (8%) 0
74 (99%) 1(1%) 0 0
173 (94%) 8 (4%) 2 (1%) 1(1%)
* Titanium Greenfield Vena Cava Filter (K901659). filter leg did not fully span the inferior vena cava, which was noted on the lateral radiograph. This may have been related to the flattened shape of the cava. In all four cases, the investigators chose to manipulate the legs with use of a guide wire and a catheter to provide better leg spacing and coverage of the vena cava. In one case, caudal movement (9.6 mm) of the filter occurred during withdrawal of the catheter after manipulation. There were no clinical sequelae related to manipulation in these patients. Three cases were reported in which a small amount of thrombus (2 cm x 2 mm) was seen on the top of the filter on the postplacement vena cavogram. The cause of thrombus formation remains unknown but may have originated in the sheath because of delayed deployment or failure to flush the sheath during insertion. No clinical sequelae were observed in these three patients. In one patient, a postplacement cavogram demonstrated an air bubble in the inferior vena cava, which may have been due to incomplete flushing prior to insertion or to introduction of air into the sheath during insertion of the sheath. The patient recovered uneventfully.
Follow-up
The stability of the SGF-AH (migration and penetration) was evaluated with anteroposterior and lateral radiographs obtained at 30 days after placement. The follow-up films were compared to the radiographs obtained immediately after insertion, and changes in filter position in relation to the near vertebrae and filter base diameter were determined. In cases of increased filter base diameter of at least 5 mm, CT scans were obtained to assess caval penetration by the filter hooks. To reduce bias, all radiographs were read a t a core reading laboratory. Reports from the core laboratory were available for all 60 patients who underwent follow-up (mean, 40 days after placement; range, 19-90 days). Eight patients died prior to 30-day follow-up, whereas three patients were not physically able to return. Three of the eight patients who died, underwent early radiographic assessment (two patients a t 24 hours and one patient a t 7 days). Early follow-up radiographs revealed no significant filter movement from the postplacement position in any of the three patients. One of these patientsdied from urethral cancer
Cho et a1
.
185
Volume 8 Number 2
Table 3 Incidence of Movement
Table 4 Outcomes of Long-term Follow-up in Nine Patients
Movement (mm)
No. of Patients
Category
Yes
No
Indeterminant
0-4 5-9 10 - 20 > 20
34 (57) 24 (40) 22 (3) 0
Insertion site patency Caval patency New DVT Filter movement 3 20 m m Filter base increase 2 5 m m
9
0 0 8 9 8
0 1 0 0 0
Note.-Numbers centages.
8 1 0 l*
in parentheses are per-
and one patient died from tension pneumothorax. An autopsy was not performed in either of these two instances. The third patient died from multiple trauma injuries sustained prior to filter placement. Only a preliminary autopsy result was available, which described the trauma-related injuries. Of the remaining five patients who died prior to 30-day follow-up, one died from brain death related to multiple trauma. Although an autopsy was performed on this patient, the filter could not be examined because the vena cava had been removed during organ harvesting for donation. The causes of death in the other four patients were brain tumor, metastatic lung cancer, metastatic thyroid cancer, and aspiration pneumonia. The investigators have determined that the causes of the eight deaths were due to coexisting medical conditions. Four patients had incomplete follow-up. Movement
Compared to previous experience with the TGF-MH study, observed movement of greater than 20 mm or movement resulting in an unacceptable filter position was considered to have potential for clinical sequelae and was the primary outcome measure. Based on this definition, all filters were stable a t 30-day follow-up. The actual measurements are listed in Table 3. Penetration
Three patients had an increase in filter base diameter that was equal to or greater than 5 mm. CT scans were obtained on all three
Note.-Average follow-up, 13.5 months. DVT * Patient refused CT.
patients. Penetration, defined as leg extension of at least 5 mm outside of the caval wall, was ruled out in two patients who had an increase of base diameter by 9.5 mm and 5.7 mm, respectively, and was suspected in one patient with an increase of 5.8 mm. In the one patient who had suspected penetration, there was disagreement in interpretation of the CT scan among the investigators and the core laboratory. This is most likely related to the poor quality of the CT scan and the elliptical shape of the cava, which complicated evaluation. There was no evidence of pericaval hematoma or organ injury in any of these CT studies. The incidence of suspected penetration in the 60 patients with follow-up was 1.7%, as opposed to the 3.8% incidence of penetration of the titanium Greenfield filter. Caval Occlusion
Three patients presented with bilateral lower extremity edema at follow-up. In all three patients, caval occlusion was documented with duplex ultrasonography (US) (5%). No vena cavograms were obtained because of the patient's poor condition and prognosis. This was not unexpected because all three patients had end-stage metastatic disease, which often leads to embolic filter occlusion with a rate as high as 38.5% (3). One of the three patients was receiving warfarin therapy, whereas the other two patients could not be given anticoagulants. Indications for filter placement in these three patients
=
deep vein thrombosis.
were thromboembolic disease, with a contraindication to anticoagulation in two patients and progression of deep vein thrombosis despite anticoagulation in the third patient. Although long-term follow-up was not part of the FDA study design, some sites routinely see patients on an annual basis. Twenty-three patients had been enrolled from these study centers, with 15 surviving an average of 13.5 months after placement. Nine of the patients had both US studies of the inferior vena cava and lower extremities, and anteroposterior and lateral radiographs. Results are found in Table 4. Although the numbers are small, the outcomes are encouraging.
I DISCUSSION Recent advances in both percutaneous catheterization techniques and filter design allow for percutaneous placement of filter devices for prevention of pulmonary embolism. Changing the metal alloy of the 24-F Greenfield filter to a titanium alloy, as well as a design change, has reduced the carrier size to 12 F, but a t the expense of the aperture of the filter nose for passage of a guide wire. With these innovations, however, the Greenfield filter can be implanted percutaneously with significant ease of insertion, decreasing insertion-site complications. Since its introduction for clinical use in 1990, the long-term safety and efficacy of the titanium Greenfield filter have been demonstrated in clinical studies. Occasion-
186
Journal of Vascular and Interventional Radiology
March-April 1997
ally, procedural problems are encountered with the TGF-MH, including sheath kinking, leg asymmetry, and incomplete opening (4-6). Further efforts to improve percutaneous placement of the Greenfield filter led to the development of the 12-F SGF-AH by combining the conical design of the Greenfield filter, the 12-F profile of the titanium model, and the capability of delivering the stainless steel Greenfield filter over a guide wire. The results of a multicenter study of the TGF-MH were chosen as an appropriate control in the present clinical trial to use in establishing acceptable frequency of event rates. The previous clinical trial for the TGF-MH and the consensus meeting provided the basis for choosing the sample size (50 patients) and the primary outcome variables (filter movement and penetration) evaluated using radiographic imaging technique a t 30 days. Performance of the newly designed SGF-AH is comparable to that of the TGF-MH with respect to the major outcome variables of movement and penetration, and the new device surpassed the expected performance. The outcomes observed in the present SGF-AH study (0% movement and 1.7% suspected penetration) were below the rates documented in the TGFMH study when these rates are adjusted using the criteria for the SGF-AH study (3.8% movement and 3.8% penetration). The design of the present study allowed for comparison with the event rates observed in the study with the titanium filter. It remains unknown how the results of the present study compare with other filter designs currently in use. Asymmetric opening of the SGFAH was reported in four of the 75 patients (5.3%). The 5.3% reported incidence in this study is less than the 10% rate reported for the TGFMH (2). Although asymmetric leg spacing of the titanium filter has not been shown to be associated with any adverse clinical consequence, the investigators in one of
the clinical centers chose to manipulate the asymmetric filter legs in four patients with an angiographic catheter and a guide wire immediately after deployment. No negative clinical event resulting from the manipulation has been reported, but it resulted in a caudal displacement of one of the filters. Although manipulation has been shown to be effective in improving leg asymmetry of titanium filters, its clinical benefits remain unknown. Further study of the mechanism and clinical significance of asymmetric opening of the Greenfield filters may be necessary. Procedural adverse events in the present study were limited to insertion site hematomas that did not require evacuation, and the incidence of insertion site venous thrombosis documented with US was 5.5%. Failure to adhere to the manufacturer's directions regarding flushing of the delivery system and manipulation of the filter with a catheter seems to have been related to the observed air embolism and caudal displacement of the filter. Both of these events were o~erator related and preventable. The major modification in this new filter allows insertion of the sheathldilator, advancement of the carrier capsule, and release of the filter over a guide wire. Although the study was not specifically designed to evaluate the function of the over-the-wire placement, the clinical results of the study have demonstrated the benefits of use of the guide wire. The guide wire seems to have two basic functions. First, it provides support for passage of the carrier capsule through the soft tissues at the puncture site and through the tortuous iliac vein into the vena cava. Second, it passes from the puncture site into the right atrium, providing centering of the filter in the caval lumen. Both of these functions are important in preventing sheath kinking and filter tilting, both of which may occasionally occur when inserting the titanium Greenfield filter. The Superstiff guide wire that has been recom-
mended for use for insertion of the SGF-AH is straight and has a soft, flexible end, measuring 10 cm. It is important to remove the wire under fluoroscopic guidance because a wire with a bend at its tip may catch on the apex or hooks and drag the filter as it is withdrawn. Literature reports of vena caval occlusion after filter placement range from 3% to 20%, compared to 5.5% in this series. This is slightly higher than in some of our earlier reports (1,7) but is probably related to the fact that nearly 50% of the patients with follow-up had malignancy as coexisting disease and these patients are known to be highly thrombophilic. Many could not be given anticoagulants, while occlusion occurred in one despite therapeutic anticoagulation. Additionally, these evaluations were done 30 days after placement without sufficient time for lysis. Among our patients with malignancy, the rate of occlusion was 12%, which is far superior to other reports of 38%. Although postfilter pulmonary embolism was not objectively evaluated, there were no clinical reports of pulmonary embolism. The new filter was successfully deployed in all 75 patients enrolled in the study for a 100% success rate, including filter insertion from the left femoral approach in 16% of cases. The right femoral approach is the most preferable for filter insertion. If the right femoral vein cannot be used because of the presence of iliofemoral thrombosis, either a left femoral or right jugular approach is used. The new SGF-AH is inserted percutaneously over a guide wire, providing ease of insertion and improved deployment while maintaining the high standards of efficacy and safety associated with the standard and titanium Greenfield filters. The alternating hook design has reduced the incidence of migration and caval penetration. Based on the results of this study, the SGF-AH received FDA approval in June 1995. Although the numbers of long-term follow-up are small, the outcomes of
Cho et a1
187
Volume 8 Number 2
15 surviving patients followed up with both US studies of the inferior vena cava and lower extremities and radiographs for an average of 13.5 months are encouraging. However, longer follow-up with a large series is needed to ensure late caval patency and filter stability comparable to the standard and titanium Greenfield filters. References 1. Greenfield LJ, Proctor MC, Cho KJ, et al. Extended evaluation of the
titanium Greenfield vena caval filter. J Vasc Surg 1994; 20:458-465. 2. Greenfield LJ. Cho KJ, Proctor MC, et al. Results of a multicenter study of the modified hook titanium Greenfield filter. J Vasc Surg 1991; 14: 253-257. 3. Grassi CJ, Matsumoto A, Teitelbaum G. Vena caval occlusion after Simon nitinol filter placement: identification with MR imaging in patients with malignancy. JVIR 1991; 3:535-539. 4. Dorfman GS. Risks and benefits of manipulation of the titanium Greenfield inferior vena cava filter after
deployment: filter facts and filter fantasies. JVIR 1993; 4:617-620. 5. Sweeny TJ, Van Arnan ME. Deployment problems with the titanium Greenfield filter. JVIR 1993; 4:691694. 6. Greenfield LJ. Assessment of asymmetric opening of the titanium Greenfield filter. JVIR 1994; 5:528532. 7. Greenfield LJ, Proctor MC. Twentyyear clinical experience with the Greenfield filter. Cardiovasc Surg 1995: 3:199-205.
JVIR 1997; 8:181-187 Abbreviations: FDA = Food and Drug Administration, SGF-AH = stainless steel alternating hook Greenfield filter, TGF-MH = modified-hook titanium Greenfield filter
From the Departments of Radiology (K.J.C., M.V.M.) and Surgery (L.J.G., M.C.P.), University of Michigan, 1500 E. Medical Center Dr, Ann Arbor, MI 48109; Boston Scientific Corporation, Natick, Massachusetts (L.A.H.); the Department of Radiology, Thomas Jefferson University Hospital, Philadelphia, Penn (J.B., D.J.E.); the Vascular and Intewentional Radiology, The Cleveland Clinic Foundation, Cleveland, Ohio (B.L.D.); the Department of Radiology, University of Tennessee Medical Group, Memphis, Term (P.A.F.); the Department of Radiology, UCSD Medical Center, San Diego, Calif (T.B.K.); the Department of~adioiogy,University of Arkansas for Medical Science, Little Rock, Ark (D.R.M.); the Department of Hartford Hospital, West Hartford, Conn (E.J.O., S.K.S.); the Department of Radiology, University of Maryland Hospital, Baltimore, Md (S.0.P.); and the Departmerit of Massachusetts General Hospital, Boston, Mass (A.C.W.). From the 1996 SCVIR annual meeting. Received ~~~~~t19, 1996; revision requested September 19; revision received and accepted November5. Address correspondence to K.J.C.
o SCVIR, 1997
PURPOSE: To evaluate a new percutaneous Greenfield filter with an alternating hook design and over-the-wiredelivery system. MATERIALS AND METHODS: The alternating hook stainless steel Greenfield filter was evaluated in a prospective clinical trial between March 10, 1994, and January 27,1995. Filters were placed in 75 patients in nine clinical centers and follow-up with radiographs and ultrasound scans was carried out at 30 days. RESULTS: Clinical trial results revealed successful placement in all patients. There were four cases of filter limb asymmetry (5.3%) without clinical sequelae, with one incidence of failure to span the cava. No significant migration was found. There were no clinically suspected pulmonary emboli, but one instance of probable caval penetration (1.7%)did occur. Caval occlusion was documented in three patients (5%). CONCLUSION: The percutaneous stainless steel Greenfield filter provides ease of insertion and improved deployment while maintaining the high standards of efficacy and safety associated with the standard and titanium Greenfield filters.
THE efficacy and safety of the modified-hook titanium Greenfield filter (TGF-MH) (Medi-techBoston Scientific, Watertown, Mass) have been objectively documented in clinical studies (1). However, the quest for the ideal filter led to the development of a percutaneous stainless steel Greenfield filter that would improve the ease of centering and placement. ~h~ new device, termed over-the-wire stainless steel alternating hook Greenfield vena cava filter with 12-F introducer system (SGF-AH; Medi-techBoston Scientific), underwent in vitro and experimental in vivo testing demonstrating that it was as effective as the standard and titanium filters (Greenfield LJ, Proctor MC, personal communication, 1993). The advantage of the over-the-wire delivery system includes passage through tortuous vessels, facilitating insertion from the left femoral vein and alignment of the carrier and vena cava to aid in centering the filter a t placement. The purpose of the study was to evaluate the
percutaneous Greenfield filter with a n alternating hook design and over-the-wire delivery system and to gain U.S. Food and Drug Administration (FDA) approval for the SGF-AH moacation of the TGF-MH.
MATERIALS AND METHODS Description of the Device The SGF-AH is manufactured from the same stainless steel wire as the original stainless steel Greenfield filter, but it differs in two major aspects. The angle of the leg-apex joint is 0" versus 17", which facilitates compressing the filter into a 12-F delivery system (Fig 1). The carrier capsule is delivered through the 15-F sheath. When initial animal testing in sheep demonstrated a tendency for the device to move distally, the hook configuration was redesigned. The hooks were re-curved to a n angle of 80" as with the TGF-MH and two of the six hooks were di-
181
182
Journal of Vascular and Interventional Radiology
March-April 1997
Figure 1. The new over-the-wire stainless steel Greenfield vena cava filter. (a)Lateral view. The filter is 4.9 cm in height and 3.2 cm in diameter at the base. (b) Axial view. The apical aperture of the filter allows passage of a 0.035-inch guide wire.
rected downward a t a n angle of 125" to further stabilize the device. Unlike the TGF-MH, there is an opening in the apex of the SGF-AH to allow passage of a guide wire. The legs are 4.9 cm long, in contrast to the 4.7-cm length of the TGF-MH and the 4.6-cm length of the standard Greenfield filter. The base diameter of the new filter is 3.2 cm as opposed to 3.8 cm for the TGF-MH and 3.0 cm for the standard Greenfield filter. The SGF-AH
Figure 2. Percutaneous placement of an over-the-wire stainless steel Greenfield vena cava filter. The Superstiff guide wire provides proper alignment and centering of the filter in the vena cava during release. (a)Posteroanterior views of filter insertion show inferior vena cavogram (left), carrier capsule exposed beyond the sheath tip over the Superstiff guide wire (middle), and filter in good position (right). (b) Lateral views of filter insertion show inferior vena cavogram (left), carrier capsule exposed beyond the sheath over the Superstiff guide wire (middle), and filter in good position (right).
Cho et a1
183
Volume 8 Number 2
ing radiologists and surgeons held prior to final IDE approval. The focus of the standards was the Patients Enrolled potential for clinical sequelae resulting from either filter movement 23 or penetration. The consensus panel 14 determined that movement greater 16 than 20 mm or placement that 6 resulted in an ineffective final 6 position had potential for clinical 4 3 sequelae. The ineffective final posi2 tion was defined as significant 1 tilting (>15"),incomplete opening, or both. Penetration was defined as 75 extension of one or more filter limbs by 5 mm or more outside of the cava as confirmed with computed turned clockwise for hemostasis and tomographic (CT) scanning. An event rate of greater than 10% in counterclockwise immediately before introduction of the preloaded either of these standards would be deemed unacceptable. This event filter carrier capsule. As the carrier rate limit was based on the results capsule is advanced into the vena of a multicenter study of the TGFcava, the guide wire should be held stationary. After the carrier capsule MH (2). The upper limit 95% confidence interval on the movement has reached the desired implantarate (3.8%) and the penetration rate tion site, the sheath is retracted (3.8%) of that study was 9.1%. This and locked to the handle, exposing the carrier capsule in the vena cava expected range and the clinically acceptable range (10%) were in beyond the tip of the sheath. While the handle is held stationarv. the close agreement. Based on these event rates, a sample size of 50 release tab is unlocked and :lid toward the operator, which retracts patients was selected. It was hythe carrier capsule and exposes the pothesized that penetration or filter. After the filter has been migration would occur in five or deployed, the guide wire is slowly fewer patients enrolled in the study. withdrawn under fluoroscopic guidA total of 75 sequential patients ance because a buckled guide wire from the nine clinical centers, who may occasionally catch in the apex met the study criteria and agreed to or the hook. The carrier is withparticipate, were enrolled. There drawn into the sheath and removed. were 48 men (64%)and 27 women Gentle pressure is applied to the (36%),with a mean age of 57 years puncture site for hemostasis. 2 17 (range, 20-89 years). Patient demographics and indications for placement were deemed comparable Clinical Trial to those in the TGF-MH clinical After approval of the Investigastudy (Table 2). The exclusion tional Device Exemption (IDE) by criteria were ( I ) patients with a the FDA, and of the institutional caval diameter greater than 28 mm; review board of each institution (2) presence of thrombus at the (Table I), this prospective nonranpuncture site, in the iliac vein, or in domized, observational study was the inferior vena cava; ( 3 )patients conducted between March 10, 1994, in whom preplacement cavography and January 27, 1995, at nine could not be performed; (4) patients institutions. The primary outcome in whom pregnancy had been convariables to be evaluated a t 30 days firmed; (5)patients for whom inafter placement included filter formed consent could not be obmovement and caval penetration. tained; (6) patients in whom a The performance standards used in coexisting medical disorder prethis study were developed during cluded them from filter placement; the consensus meeting of participat- and (7) patients who were unwilling
Table 1 Patient Enrollment at Nine Institutions Institution University of Michigan Hospitals University of Maryland Hospital Massachusetts General Hospital Hartford Hospital University of California San Diego University of Arkansas Hospital Cleveland Clinic University Tennessee Yale University Total
is preloaded into the carrier capsule and introduced through a 12-F catheter. There are separate femoral and jugular insertion systems that differ in the orientation of the apex of the filter with respect to the introducer.
Recommended Insertion Procedure The technique for percutaneous insertion of the SGF-AH is similar to that used for the TGF-MH, but the filter is deployed over a guide wire. After percutaneous catheterization of the vena cava from either the femoral or jugular vein with use of the Seldinger technique, a 5-F pigtail catheter is advanced over the 0.035-inch guide wire into the vena cava. After a vena cavogram is obtained, the catheter is removed over a 0.035-inch Arnplatz Superstiff guide wire (Medi-techBoston Scientific), and a 3-4-mm skin incision is made along the guide wire. It is important to advance the floppy portion of the guide wire well beyond the intended implantation site because the stiffer portion of the guide wire is necessary to center the filter in the vena cava (Fig 2). Sequential dilation is then performed, and the 15-F sheath1 dilator system with attached Rotovalve (Medi-techBoston Scientific) for hemostasis is advanced over the guide wire into the vena cava beyond the desired implantation site. The dilator is then withdrawn, leaving the sheath and the guide wire in place. The Rotovalve is
184
Journal of Vascular and Interventional Radiology
March-April 1997
to comply with follow-up requirements as indicated in the study protocol. Vena cavography was performed in all patients before filter placement for the evaluation of caval patency, caval anomaly, and location of the renal vein. After placement, biplane abdominal radiographs were taken in expiration (Fig 2). All measurements of filter base diameter after implantation and at 30-day follow-up, as well as change in filter position a t 30-day followup, were corrected for magnification. The known length of the SGFAH was used to correct for mamification. The distance from the Gp of the apex to the bottom of the hook was 5.2 cm. Filter movement was assessed by measurement of the change in position of the filter apex in relation to the top of the nearest vertebral body for the anteroposterior radiograph on both postimplantation and follow-up films.
RESULTS Placement
The 12-F SGF-AH was successfully deployed in all 75 patients enrolled in the study, for a 100% success rate. All filters were placed with use of the percutaneous insertion technique. The sites for insertion were the right femoral vein in 55 patients (73%), the left femoral vein in 12 patients (16%), and the right internal jugular vein in eight patients (11%). Overall, there were 67 femoral insertions (89%) and The eight jugular insertions (11%). filter locations were infrarenal (between the lowest renal vein and the junction of the right and left common iliac veins) in 74 patients (99%)and suprarenal (between the right hepatic vein and the highest renal vein) in one patient (1%). The procedure-related complications were limited to two femoral insertion site hematomas without clinical sequelae (2.6%). There were four reported cases of filter limb asymmetry in 75 patients (5.3%). In one patient, the
Table 2 Comparability of Patients Characteristics for 12-F SGF-AH versus TGF Category
12-F SGF-AH
TGF*
No. of Placements Age (y) Mean Range Males (%) Indications for Use Contraindication to anticoagulation Recurrent thromboembolism despite anticoagulation Complications of anticoagulation Adjunctive prophylaxis Other Trauma patients (new indication) Filter Location Infrarenal Suprarenal At level of renal vein Superior vena cava
75
181
57 20 - 89 48 (64%)
58 17 - 92 115 (64%)
49 (65%)
136 (75%)
7 (9%)
17 (9%)
3 (4%) 8 (11%) 6 (8%) 6 (8%)
19 (10%) 20 (11%) 15 (8%) 0
74 (99%) 1(1%) 0 0
173 (94%) 8 (4%) 2 (1%) 1(1%)
* Titanium Greenfield Vena Cava Filter (K901659). filter leg did not fully span the inferior vena cava, which was noted on the lateral radiograph. This may have been related to the flattened shape of the cava. In all four cases, the investigators chose to manipulate the legs with use of a guide wire and a catheter to provide better leg spacing and coverage of the vena cava. In one case, caudal movement (9.6 mm) of the filter occurred during withdrawal of the catheter after manipulation. There were no clinical sequelae related to manipulation in these patients. Three cases were reported in which a small amount of thrombus (2 cm x 2 mm) was seen on the top of the filter on the postplacement vena cavogram. The cause of thrombus formation remains unknown but may have originated in the sheath because of delayed deployment or failure to flush the sheath during insertion. No clinical sequelae were observed in these three patients. In one patient, a postplacement cavogram demonstrated an air bubble in the inferior vena cava, which may have been due to incomplete flushing prior to insertion or to introduction of air into the sheath during insertion of the sheath. The patient recovered uneventfully.
Follow-up
The stability of the SGF-AH (migration and penetration) was evaluated with anteroposterior and lateral radiographs obtained at 30 days after placement. The follow-up films were compared to the radiographs obtained immediately after insertion, and changes in filter position in relation to the near vertebrae and filter base diameter were determined. In cases of increased filter base diameter of at least 5 mm, CT scans were obtained to assess caval penetration by the filter hooks. To reduce bias, all radiographs were read a t a core reading laboratory. Reports from the core laboratory were available for all 60 patients who underwent follow-up (mean, 40 days after placement; range, 19-90 days). Eight patients died prior to 30-day follow-up, whereas three patients were not physically able to return. Three of the eight patients who died, underwent early radiographic assessment (two patients a t 24 hours and one patient a t 7 days). Early follow-up radiographs revealed no significant filter movement from the postplacement position in any of the three patients. One of these patientsdied from urethral cancer
Cho et a1
.
185
Volume 8 Number 2
Table 3 Incidence of Movement
Table 4 Outcomes of Long-term Follow-up in Nine Patients
Movement (mm)
No. of Patients
Category
Yes
No
Indeterminant
0-4 5-9 10 - 20 > 20
34 (57) 24 (40) 22 (3) 0
Insertion site patency Caval patency New DVT Filter movement 3 20 m m Filter base increase 2 5 m m
9
0 0 8 9 8
0 1 0 0 0
Note.-Numbers centages.
8 1 0 l*
in parentheses are per-
and one patient died from tension pneumothorax. An autopsy was not performed in either of these two instances. The third patient died from multiple trauma injuries sustained prior to filter placement. Only a preliminary autopsy result was available, which described the trauma-related injuries. Of the remaining five patients who died prior to 30-day follow-up, one died from brain death related to multiple trauma. Although an autopsy was performed on this patient, the filter could not be examined because the vena cava had been removed during organ harvesting for donation. The causes of death in the other four patients were brain tumor, metastatic lung cancer, metastatic thyroid cancer, and aspiration pneumonia. The investigators have determined that the causes of the eight deaths were due to coexisting medical conditions. Four patients had incomplete follow-up. Movement
Compared to previous experience with the TGF-MH study, observed movement of greater than 20 mm or movement resulting in an unacceptable filter position was considered to have potential for clinical sequelae and was the primary outcome measure. Based on this definition, all filters were stable a t 30-day follow-up. The actual measurements are listed in Table 3. Penetration
Three patients had an increase in filter base diameter that was equal to or greater than 5 mm. CT scans were obtained on all three
Note.-Average follow-up, 13.5 months. DVT * Patient refused CT.
patients. Penetration, defined as leg extension of at least 5 mm outside of the caval wall, was ruled out in two patients who had an increase of base diameter by 9.5 mm and 5.7 mm, respectively, and was suspected in one patient with an increase of 5.8 mm. In the one patient who had suspected penetration, there was disagreement in interpretation of the CT scan among the investigators and the core laboratory. This is most likely related to the poor quality of the CT scan and the elliptical shape of the cava, which complicated evaluation. There was no evidence of pericaval hematoma or organ injury in any of these CT studies. The incidence of suspected penetration in the 60 patients with follow-up was 1.7%, as opposed to the 3.8% incidence of penetration of the titanium Greenfield filter. Caval Occlusion
Three patients presented with bilateral lower extremity edema at follow-up. In all three patients, caval occlusion was documented with duplex ultrasonography (US) (5%). No vena cavograms were obtained because of the patient's poor condition and prognosis. This was not unexpected because all three patients had end-stage metastatic disease, which often leads to embolic filter occlusion with a rate as high as 38.5% (3). One of the three patients was receiving warfarin therapy, whereas the other two patients could not be given anticoagulants. Indications for filter placement in these three patients
=
deep vein thrombosis.
were thromboembolic disease, with a contraindication to anticoagulation in two patients and progression of deep vein thrombosis despite anticoagulation in the third patient. Although long-term follow-up was not part of the FDA study design, some sites routinely see patients on an annual basis. Twenty-three patients had been enrolled from these study centers, with 15 surviving an average of 13.5 months after placement. Nine of the patients had both US studies of the inferior vena cava and lower extremities, and anteroposterior and lateral radiographs. Results are found in Table 4. Although the numbers are small, the outcomes are encouraging.
I DISCUSSION Recent advances in both percutaneous catheterization techniques and filter design allow for percutaneous placement of filter devices for prevention of pulmonary embolism. Changing the metal alloy of the 24-F Greenfield filter to a titanium alloy, as well as a design change, has reduced the carrier size to 12 F, but a t the expense of the aperture of the filter nose for passage of a guide wire. With these innovations, however, the Greenfield filter can be implanted percutaneously with significant ease of insertion, decreasing insertion-site complications. Since its introduction for clinical use in 1990, the long-term safety and efficacy of the titanium Greenfield filter have been demonstrated in clinical studies. Occasion-
186
Journal of Vascular and Interventional Radiology
March-April 1997
ally, procedural problems are encountered with the TGF-MH, including sheath kinking, leg asymmetry, and incomplete opening (4-6). Further efforts to improve percutaneous placement of the Greenfield filter led to the development of the 12-F SGF-AH by combining the conical design of the Greenfield filter, the 12-F profile of the titanium model, and the capability of delivering the stainless steel Greenfield filter over a guide wire. The results of a multicenter study of the TGF-MH were chosen as an appropriate control in the present clinical trial to use in establishing acceptable frequency of event rates. The previous clinical trial for the TGF-MH and the consensus meeting provided the basis for choosing the sample size (50 patients) and the primary outcome variables (filter movement and penetration) evaluated using radiographic imaging technique a t 30 days. Performance of the newly designed SGF-AH is comparable to that of the TGF-MH with respect to the major outcome variables of movement and penetration, and the new device surpassed the expected performance. The outcomes observed in the present SGF-AH study (0% movement and 1.7% suspected penetration) were below the rates documented in the TGFMH study when these rates are adjusted using the criteria for the SGF-AH study (3.8% movement and 3.8% penetration). The design of the present study allowed for comparison with the event rates observed in the study with the titanium filter. It remains unknown how the results of the present study compare with other filter designs currently in use. Asymmetric opening of the SGFAH was reported in four of the 75 patients (5.3%). The 5.3% reported incidence in this study is less than the 10% rate reported for the TGFMH (2). Although asymmetric leg spacing of the titanium filter has not been shown to be associated with any adverse clinical consequence, the investigators in one of
the clinical centers chose to manipulate the asymmetric filter legs in four patients with an angiographic catheter and a guide wire immediately after deployment. No negative clinical event resulting from the manipulation has been reported, but it resulted in a caudal displacement of one of the filters. Although manipulation has been shown to be effective in improving leg asymmetry of titanium filters, its clinical benefits remain unknown. Further study of the mechanism and clinical significance of asymmetric opening of the Greenfield filters may be necessary. Procedural adverse events in the present study were limited to insertion site hematomas that did not require evacuation, and the incidence of insertion site venous thrombosis documented with US was 5.5%. Failure to adhere to the manufacturer's directions regarding flushing of the delivery system and manipulation of the filter with a catheter seems to have been related to the observed air embolism and caudal displacement of the filter. Both of these events were o~erator related and preventable. The major modification in this new filter allows insertion of the sheathldilator, advancement of the carrier capsule, and release of the filter over a guide wire. Although the study was not specifically designed to evaluate the function of the over-the-wire placement, the clinical results of the study have demonstrated the benefits of use of the guide wire. The guide wire seems to have two basic functions. First, it provides support for passage of the carrier capsule through the soft tissues at the puncture site and through the tortuous iliac vein into the vena cava. Second, it passes from the puncture site into the right atrium, providing centering of the filter in the caval lumen. Both of these functions are important in preventing sheath kinking and filter tilting, both of which may occasionally occur when inserting the titanium Greenfield filter. The Superstiff guide wire that has been recom-
mended for use for insertion of the SGF-AH is straight and has a soft, flexible end, measuring 10 cm. It is important to remove the wire under fluoroscopic guidance because a wire with a bend at its tip may catch on the apex or hooks and drag the filter as it is withdrawn. Literature reports of vena caval occlusion after filter placement range from 3% to 20%, compared to 5.5% in this series. This is slightly higher than in some of our earlier reports (1,7) but is probably related to the fact that nearly 50% of the patients with follow-up had malignancy as coexisting disease and these patients are known to be highly thrombophilic. Many could not be given anticoagulants, while occlusion occurred in one despite therapeutic anticoagulation. Additionally, these evaluations were done 30 days after placement without sufficient time for lysis. Among our patients with malignancy, the rate of occlusion was 12%, which is far superior to other reports of 38%. Although postfilter pulmonary embolism was not objectively evaluated, there were no clinical reports of pulmonary embolism. The new filter was successfully deployed in all 75 patients enrolled in the study for a 100% success rate, including filter insertion from the left femoral approach in 16% of cases. The right femoral approach is the most preferable for filter insertion. If the right femoral vein cannot be used because of the presence of iliofemoral thrombosis, either a left femoral or right jugular approach is used. The new SGF-AH is inserted percutaneously over a guide wire, providing ease of insertion and improved deployment while maintaining the high standards of efficacy and safety associated with the standard and titanium Greenfield filters. The alternating hook design has reduced the incidence of migration and caval penetration. Based on the results of this study, the SGF-AH received FDA approval in June 1995. Although the numbers of long-term follow-up are small, the outcomes of
Cho et a1
187
Volume 8 Number 2
15 surviving patients followed up with both US studies of the inferior vena cava and lower extremities and radiographs for an average of 13.5 months are encouraging. However, longer follow-up with a large series is needed to ensure late caval patency and filter stability comparable to the standard and titanium Greenfield filters. References 1. Greenfield LJ, Proctor MC, Cho KJ, et al. Extended evaluation of the
titanium Greenfield vena caval filter. J Vasc Surg 1994; 20:458-465. 2. Greenfield LJ. Cho KJ, Proctor MC, et al. Results of a multicenter study of the modified hook titanium Greenfield filter. J Vasc Surg 1991; 14: 253-257. 3. Grassi CJ, Matsumoto A, Teitelbaum G. Vena caval occlusion after Simon nitinol filter placement: identification with MR imaging in patients with malignancy. JVIR 1991; 3:535-539. 4. Dorfman GS. Risks and benefits of manipulation of the titanium Greenfield inferior vena cava filter after
deployment: filter facts and filter fantasies. JVIR 1993; 4:617-620. 5. Sweeny TJ, Van Arnan ME. Deployment problems with the titanium Greenfield filter. JVIR 1993; 4:691694. 6. Greenfield LJ. Assessment of asymmetric opening of the titanium Greenfield filter. JVIR 1994; 5:528532. 7. Greenfield LJ, Proctor MC. Twentyyear clinical experience with the Greenfield filter. Cardiovasc Surg 1995: 3:199-205.