Difficult retrieval of a recovery IVC filter.

Editor: The authors report the case of a 75-year-old man who was admitted for vertebroplasty for T5 and T6 compression fractures. During his hospitalization an episode of acute desaturation occurred and a computed tomographic (CT) pulmonary angiogram was obtained. This showed a pulmonary embolus within a segmental branch of the left lower lobe. Because the patient was to undergo vertebroplasty, he was deemed not to be a candidate for anticoagulation, and placement of an inferior vena cava (IVC) filter was requested. After extensive discussions with the referring team, it was decided to place a retrievable IVC filter. A Recovery retrievable IVC filter (C.R. Bard, Tempe, AZ) was placed via a right common femoral vein approach. The next day, the vertebroplasty procedure was performed without complication. Six days later the patient developed fever and bacteremia. In consultation with the primary physician and the hematology service, it was decided to remove the IVC filter and start anticoagulation treatment. The patient was brought to the angiography suite and right internal jugular vein access was obtained. An ELE straight multiple–side hole catheter (Cook, Bloomington, IN) was positioned in the IVC inferior to the Recovery filter and a cavogram in anteroposterior projection was obtained. This showed the IVC and filter to be free of thrombus. The catheter was then removed over a stiff Glidewire (Terumo/Boston Scientific, Natick, MA). The 10-F retrieval sheath (C.R. Bard) was placed after serial fascial dilation. The tip of the sheath was positioned superior to the filter tip. The Recovery Cone was then advanced through the sheath and attempts were made to dock the retrieval cone on the filter tip. Moderate filter tilt and the fact that the sheath was not aligned perfectly with the longitudinal axis of the IVC resulted in an angle of approximately 10° between the Recovery Cone and the long axis of the filter (Figure, part a). This misalignment resulted in eccentric positioning of the cone over the filter tip, and, as a consequence, the retrieval cone grabbed one of the filter arms rather than the filter tip. This was not noticeable on the anteroposterior view (Figure, part b), but upon attempting to retrieve the cone with the filter into the retrieval sheath, resistance was experienced and an oblique projection was obtained. This showed that the filter tip was outside the retrieval cone and the sheath (Figure, part c). Multiple further attempts to grab the filter tip were also unsuccessful. Inability to retrieve a filter was described in the initial series on this device (1). Although the retrieval of filters perfectly aligned to the long axis of the IVC generally posed no problem, the author reports that, in filters in which tilt was present, additional maneuvers had to be used. The key is to move the filter away from the caval wall and align it better with the filter cone. Two techniques were described. The first technique simply required advancement of a guide wire through the retrieval cone channel and through the filter, which improved straight-line access to the filter tip. If this was not successful, the retrieval cone was removed and a curved-tip catheter and hydrophilic guide wire were used to position the wire to the side of the filter, where it was tilted toward the caval wall. The guide wire was then exchanged for a stiff Amplatz wire, which pulled the filter tip away from the caval wall and allowed straight-line access. The retrieval cone was then successfully advanced over the filter tip with use of either technique (1). We attempted to use the retrieval technique with the Glidewire because this technique does not require removal of the retrieval cone from its sheath. However, despite multiple attempts, we were unable to position the wire toward the side of the filter that shows the shortest distance between the filter tip and the caval wall. Instead of removing the retrieval cone and directing the wire to the desired position with use of a curved-tip catheter, we opted to use a tipdeflecting wire (Reuter disposable tip deflecting wire, 10-mm radius; Cook, Bloomington, IN). The tip-deflecting wire was advanced through the central lumen of the retrieval cone and, with use of a combination of flection, deflection, torque, and release, we were able to place the wire through the filter struts on the desired side of the filter. We then performed maximum guide wire tip deflection, which moved the filter tip away from the caval wall and centered it within the IVC. This enabled us to slide the retrieval cone over the filter tip and capture it without capturing a filter arm (Figure part d and e). The filter was then removed uneventfully, and a cavogram showed a patent IVC without evidence of thrombus or injury to the caval wall. The Recovery filter is the first retrievable filter approved by the Food and Drug Administration for this purpose. The preliminary experience in humans was published in 2002, in which the device was implanted in 32 patients (1). The filter was successfully retrieved via a jugular approach in 24 of 24 patients. The mean implantation period in this series was 53 days, with a range of 5–134 days. Problems with filter removal were always related to suboptimal straight-line access to the filter tip. The author recommended two techniques facilitating retrieval, but modified his removal technique in the last four cases by always performing the initial wire placement through the side of the filter on which the tip was closer to the caval wall. Our technique is fairly simple and requires only insertion of a deflecting wire without removal of the retrieval cone. With use of a combination of tip deflection and gentle traction, it was possible to center the filter, advance the retrieval cone over the filter tip, and remove the filter successfully (1). Because it can be expected that these devices will be used more frequently in the future (2,3), interventional radiologists should keep this technique in mind when faced with the situation of retrieving a recovery filter in a tilted position. DOI: 10.1097/01.RVI.0000127887.62434.A8 Letters to the Editor