Deployment Problems with the Titanium Greenfield Filter

Deployment Problems with the Titanium Greenfield Filter1 Timothy J. Sweeney, MD2 Michael E. Van Aman, MD Index terms: Embolism, pulmonary, 60.72 Venae cavae, filters, 982.1267 Venae cavae, thrombosis, 982.751

JVIR 1993; 4:691-694 ~ b b ~ ~ SGF ~ =istainless ~ ~ steel i ~ Greenfield filter, TGF = titanium Greenfield filter

PURPOSE: The authors retrospectively reviewed their initial experience with deployment of the modified hook titanium Greenfield filter. PATIENTS AND METHODS: Twenty-threepatients underwent filter placements over a 1-yearperiod. Radiographs were obtained immediately after placement to confirm filter position in all cases. Follow-up images were available in 15patients (65%). RESULTS: Twenty-fourfilters were placed in 23 patients. Tilting of the filter (> 15")was evaluated in 22 placements without complications and was present in five (23%).In 17 of 24 placements (71%),distribution of ~filter ~ :legs was poor, with wide gaps between clustered legs. Manipulation of the filter legs with an angiographic catheter resulted in improved distribution in three of six attempts but also resulted in a caudal displacement, which necessitated placement of a second filter. At follow-up (range, 4-16 months; mean, 9 months), three cases of asymptomatic inferior vena caval thrombosis and one recurrent pulmonary embolism were discovered. CONCLUSION: No untoward event resulting from filter placement was demonstrated. Further study and review of the deployment mechanism may be necessary.

percutaneous placement of the stainless steel Greenfield filter (SGF) was accomplished through a 24-F sheath and was frequently complicated by femoral vein thrombosis (14). Since that time, a number of new filtering devices have been introduced. Each offers the advantage of a smaller carrier system in hopes of reducing the complication rate of percutaneous insertion. The titanium Greenfield filter (TGF) was developed a 12-F carrier 'ystem and a 14-Fintr~ducer.Theoretically, the TGF would reduce the incidence of femoral vein thrombosis, yet retain the proved safety and efficacy achieved with the original stainless steel model. In addition, the TGF included a simplified release mechanism devised to improve the confidence in placement at the appropriate levels. Clinical trials of the original TGF demonstrated an improved release mechanism, virtual elimination of procedural bleeding, and a signifiT H E

From the Division of Cardiovascular and Interventional Radiology, Ohio State University Hospitals, 410 W 10th Ave, Columbus, Ohio.ReceivedNovember25,1992; revision requested January 12, 1993; revision received June 3; accepted June 21. Address reprint requests to M.E.v.A., D ~ partment of Radiology, Mount Carmel Medical Center, 777 W State St, Columbus, OH 43222. T u r r e n t address: Department of Radiology, University of Florida Health Sciences Center, Jacksonville, Fla. 1

"

SCVIR, 1993

See also the article by Moore et al (pp 687690) and the editorial by Dorfman (pp 6 1 7 620) in this issue.

cantly reduced incidence of femoral vein thrombosis (5). However, problems with migration and caval penetration necessitated further design modification and testing prior to release for general use in November 1990 (6-8). We report our experience with percutaneous insertion of the modified hook TGF in 23 patients. PATIENTS AND METHODS

Twenty-three patients referred to the angiography division at Ohio State University Hospitals (Columbus, Ohio) between January 1991 and January 1992 for interruption of the inferior vena cava received a modified hook TGF (Medi-tech/Boston Scientific, Watertown, Mass). The mean age of these patients was 59.4 years, with a range of 29-90 years. Nine were men and 14 were women. Fifteen of the filters were placed from a right femoral vein approach, five were placed from a left

691

692

Journal of Vascular and Interventional Radiology

September-October 1993

femoral vein approach, and three were placed from a right jugular vein approach. The indications for caval interruption in the 23 patients were as follows: Therapy with anticoagulation was unsuccessful in six patients, 15 patients were considered to have contraindications to anticoagulation therapy, and two patients had freefloating clots in the inferior vena cava or iliac veins. All filters were placed in accordance with the recommended procedure described in the package insert that accompanies the TGF. Prior to placement, all patients underwent single-plane caiography and a right femoral approach was used when possible. All filters were placed percutaneously with assistance of fluoroscopy. Postprocedure radiographs were obtained to confirm and document filter placement in all cases. Patient records were obtained when possible, and retrospective analysis of potential untoward effects of caval interruption due to filter placement was attempted. Postplacement images demonstrating the TGF were available in 15 patients (65%). These images were obtained by clinicians for reasons not related to the current study. Available images were evaluated for changes in angulation (tilt), changes in the distribution of the filter feet, gross changes in position relative to the spine and renal veins, evidence of thrombosis or caval penetration suggested by changes in width and distribution of the filter legs, and evidence of filter fracture. Magnification differences were corrected by comparing skeletal landmarks on the L-2 vertebrae in corresponding images.

RESULTS Twenty-four modified hook TGFs were placed in the 23 patients. One patient received a second filter after the first failed to open completely and attempts to manipulate the limbs with an angiographic catheter resulted in entanglement. This filter

was dislodged and drawn to the iliac vein confluence before the catheter could be freed. Subsequently, a second filter was placed through the first in the appropriate position in the inferior vena cava (Fig 1). The distribution of filter legs and tilting were evaluated on postplacement radiographs. Entanglement of two or more filter legs or grossly uneven distribution of legs without evidence of entanglement was considered a n unsatisfactory release and was seen in 17 of 24 placements (71%). During the later part of the study period, it was learned that catheter manipulation of the filter legs could be without significant risk of cephalic filter migration (Marx MV, personal communication, 1991). Six of the last nine filters demonstrated grossly asymmetric leg distribution, and an attempt to improve leg placement was made with a n angiographic catheter in all six. Significant improvement was seen in three cases, and the one complication of distal filter displacement, described above, occurred. The plain radiographic finding of poor circumferential leg distribution was confirmed in five patients in whom follow-up cross-sectional imaging was available (Fig 2). The end result was radiographically poor circumferential filter leg distribution in 58% of cases (14 placements). The longitudinal axis of the inferior vena cava was related to that of the filter for determination of filter angulation. Tilting of more than 15" occurred with five of 22 placements (23%).Two filter placements were not evaluated for tilting. In one case, the filter was placed such that two legs rested in the post-transplant hepatic vein. The filter drawn to the iliac confluence was also not considered. Follow-up radiographs were obtained in 15 cases at 4-16 months (mean, 9 months) after placement and were compared with postplacement radiographs. Evaluations for gross changes in filter span and in tilt relative to the longitudinal axis of the spinal column were conducted. The only significant changes in filter

I Figure 1. An attempt to improve the leg distribution of the first TGF with an

angiographic catheter resulted in displacement to the iliac venous confluence. A second TGF was then placed through the first in the appropriate position. Note the poor leg distribution of the second filter, with three legs along the medial wall of the IVC.

width ( 2 7 mm) were associated with asymptomatic inferior vena cava thrombosis seen in three cases (20%). One case of recurrent pulmonary embolus was documented angiographically. An inferior vena cavogram in this patient demonstrated thrombus in the apex of the filter with propagation above the filter. This filter was not significantly tilted and had relatively good leg distribution. Recurrent embolic disease was not clinically suspected in the other 14 patients. Follow-up examinations in this review did not reveal filter leg fractures. There was no evidence of significant strut penetration of the caval wall. Further, no incidence of femoral vein thrombosis, extraretroperitoneal hematoma, or untoward event resulting from percutaneous placement of the inferior vena cava filters was demonstrated. This was reviewed retrospectively, however, and it is possible that such events did occur asymptomatically (4). Four patients

Sweeney and Van Aman

693

Volume 4 Number 5

b. Figure 2. (a)Initial postplacement radiograph of TGF shows markedly uneven leg distribution and moderate canting. (b) Computed tomographic scan obtained 9 months after filter placement confirms uneven leg distribution, with wide gaps between clustered legs.

were known to have died prior to the writing of these results, all having succumbed to their underlying illness. No autopsies were performed. Eight patient histories after filter placement were not available.

DISCUSSION The design of the TGF, as that of its predecessor the SGF, allows for

maximal blood filtration when the ment (15). Greenfield et a1 reported filter is positioned parallel to the lononly a 5.4%rate of leg asymmetry gitudinal axis of the inferior vena and a 2%incidence of incomplete cava. Katsamouris et al demonopening a t discharge of the TGF. All strated in vitro that a SGF tilted at incompletely opened TGFs were de14"trapped only 20%of challenge scribed as completely deployed after clots and allowed clots as large as 6 x manipulation (10).We were unable to 90 mm to pass through (9). While reproduce these results. some suggest that the risk of deleteriIn vitro experiments by Greenfield ous effect of tilting is only significant and Proctor showed a significant deif the filter apex is adjacent to the crease in clot-trapping ability of the wall of the inferior vena cava (lo), TGF when the legs were asymmetriothers argue that any tilting filter cally deployed in venae cavae 22 mm should be considered a potential or less in diameter. There may have problem (9).Messmer and Greenfield been a similar effect in venae cavae suggested that optimal filtration oclarger than 22 mm, although this was curs when the filter is angled less masked by other factors (16). Greenthan 15"(11). field et a1 attribute asymmetry of the In our study, 23%of the TGFs filter legs in the inferior vena cava to demonstrated tilts of 15"or more on the introduction of the carrier system postplacement radiographs. This against the caval wall as opposed to compares unfavorably with our centrally within the vena cava. In our institution's experience with percuta- experience, centering the carrier sysneous placement of SGFs (1) and tem within the inferior vena cava apwith the experience of others with pears to be technically difficult to the SGF (121, which demonstrates an evaluate and not controllable by the approximate 3%rate of significant operator. Despite the experimental tilting. Consistent positioning of the data (9,13,16),Greenfield believes SGF was accomplished by placing the that tilt and asymmetry have never filter over a guide wire. The TGF rebeen proved to have an adverse effect lease mechanism does not involve the on the clinical effectiveness of the use of a guide wire to direct positionfilter. However, careful long-term ing. Although thrombus-trapping follow-up of the TGF with its apparefficiency of the TGF was initially ent increased incidence of these probconsidered similar to that of the SGF, lems has not been performed. a recent in vivo study has suggested Messmer and Greenfield demonthat the TGF does not trap thrombi strated the value of the abdominal as effectively as the SGF (13). We beradiograph when used to evaluate lieve the high incidence of tilting fur- Greenfield filters after placement ther compromises the efficacy of the (11).Filter migration, span, and TGF design. angle can all be noted with reasonWe were further concerned by the able accuracy on follow-up plain rafrequent difficulties encountered diographs. Ferris et a1 recommend with poor leg distribution and leg enboth anteroposterior and lateral protanglement upon deployment of the jections for postplacement imaging, TGF from the introducer. Initially, with follow-up scheduled at 1 and 6 71%of filters demonstrated less than months and annually thereafter (17). satisfactory distribution and/or enEconomic restrictions may preclude tanglement. Improvement was noted applying this criteria. in several instances when an angioOne of the significant advantages graphic catheter was used to enhance of the TGF is the ease of deployment. limb positions, but this was not uniThe filter comes preloaded in its capversally successful and manipulation sule from the manufacturer, and its was not without complications. release is a simple one-step maneuWhile crossing of the legs with the ver. There is no learning curve, and SGF was reported (14), early studies there was no difference in the inciwith the TGF reported no entangledence of tilting or clustered legs

694

Journal of Vascular and Interventional Radiology

September-October 1993

among the three physicians deploying the filters, despite marked differences i n their experience. Sheath perforation did not occur. When iliac vein angulation made filt e r advancement difficult, i t could be overcome by advancing the sheath a n d filter a s a unit for a short distance. Insertion site made n o statistically significant difference i n t h e incidence of asymmetric leg deployment. In summary, o u r experience with percutaneous placement of t h e modified hook T G F suggests that t h e curr e n t method of introduction of this device into t h e inferior vena cava is suboptimal. We encountered a substantial incidence of filter tilting a n d clustered asymmetric limb distribution. These occurrences may reduce t h e effectiveness of this filter design. Because of t h e small numbers a n d limited follow-up, we cannot docum e n t t h e deleterious effects of t h e described deployment problems. We believe further study is necessary.

References 1. Rose BS, Simon DC, Hess ML, Van Aman ME. Percutaneous transfemoral placement of the KimrayGreenfield vena cava filter. Radiology 1987; 165:373-376. 2. Pais SO, Tobin KD, Austin CB, Queral L. Percutaneous insertion

of the Greenfield inferior vena cava filter: experience with ninety-six patients. J Vasc Surg 1988; 8:460-464. Mewissen MW, Erickson SJ, Foley DW, et al. Thrombosis at venous insertion sites after vena cava filter placement. Radiology 1989; 173: 155-157. Dorfman GS, Cronan J J , Paolla LP, et al. Iatrogenic changes at the venotomy site after percutaneous placement of the Greenfield filter. Radiology 1989; 173:159-162. Greenfield LJ, Cho KJ, Pais SO, Van Aman ME. Preliminary clinical experience with the titanium Greenfield vena cava filter. Arch Surg 1989; 124:657-659. Ramchandani P, Koolpe HA, Zeit RM. Splaying of titanium Greenfield inferior vena cava filter. AJR 1990; 155:1103-1104. Greenfield LJ,Cho KJ, Tauscher JR. Evolution of hook design for fixation of the titanium Greenfield filter. J Vasc Surg 1990; 12:345353. Teitelbaum GP, Jones DL, van Breda A, et al. Vena caval filter splaying: potential complication of use of the titanium Greenfield filter. Radiology 1989; 173:809-814. Katsamouris AA, Waltman AC, Delichatsios MA, Athanasoulis CA. Inferior vena cava filter: in vitro comparison of clot trapping and flow dynamics. Radiology 1988; 166:361366. Greenfield LJ, Cho KJ, Proctor M, et

11.

12.

13.

14. 15.

16.

17.

al. Results of a multicenter study of the modified hook-titanium Greenfield filter. J Vasc Surg 1991; 14253-257. Messmer JM, Greenfield W. Greenfield caval filters: long-term radiographic follow-up study. Radiology 1985; 156:613-618. Denny DF, Dorfman GS, Cronan J J , Greenwood LH, Morse SS, Yoselevitz M. Greenfield filter: percutaneous placement in 50 patients. AJR 1988; 150:427-429. Millward SF, Marsh JI, Pon C, Moher D. Thrombus-trapping efficiency of the LGM (Vena Tech) and titanium Greenfield filters in vivo. JVIR 1992; 3:103-106. Otchy DP, Elliot BM. The malpositioned Greenfield filter: lessons learned. Am Surg 1987; 53:580-583. Burke PE, Michna BA, Harvey CF, Crute SL, Sobel M, Greenfield LJ. Experimental comparison of percutaneous vena caval devices: titanium Greenfield filter versus Bird's Nest filter. J Vasc Surg 1987; 6:66-70. Greenfield LC, Proctor MC. Experimental embolic capture by asymmetric Greenfield filters. J Vasc Surg 1992; 16:436-444. Ferris EJ, Carver DK, McCowan TC. Inferior vena cava filters: technical aspects and follow-up. In: Mueller PR, vansonnenberg E, Becker GJ, eds. Syllabus: a categorical course in diagnostic radiology. Oak Brook, Ill: Radiological Society of North America, 1991; 169-178.