Predicting the Safety and Effectiveness of Inferior Vena Cava Filters (PRESERVE): Outcomes at 12 months.

OBJECTIVE: To determine the safety and effectiveness of vena cava filters (VCFs). METHODS: A total of 1429 participants (62.7 ± 14.7 years old; 762 [53.3% male]) consented to enroll in this prospective, nonrandomized study at 54 sites in the United States between October 10, 2015, and March 31, 2019. They were evaluated at baseline and at 3, 6, 12, 18, and 24 months following VCF implantation. Participants whose VCFs were removed were followed for 1 month after retrieval. Follow-up was performed at 3, 12, and 24 months. Predetermined composite primary safety (freedom from perioperative serious adverse events [AEs] and from clinically significant perforation, VCF embolization, caval thrombotic occlusion, and/or new deep vein thrombosis [DVT] within 12-months) and effectiveness (composite comprising procedural and technical success and freedom from new symptomatic pulmonary embolism [PE] confirmed by imaging at 12-months in situ or 1 month postretrieval) end points were assessed. RESULTS: VCFs were implanted in 1421 patients. Of these, 1019 (71.7%) had current DVT and/or PE. Anticoagulation therapy was contraindicated or had failed in 1159 (81.6%). One hundred twenty-six (8.9%) VCFs were prophylactic. Mean and median follow-up for the entire population and for those whose VCFs were not removed was 243.5 ± 243.3 days and 138 days and 332.6 ± 290 days and 235 days, respectively. VCFs were removed from 632 (44.5%) patients at a mean of 101.5 ± 72.2 days and median 86.3 days following implantation. The primary safety end point and primary effectiveness end point were both achieved. Procedural AEs were uncommon and usually minor, but one patient died during attempted VCF removal. Excluding strut perforation greater than 5 mm, which was demonstrated on 31 of 201 (15.4%) patients' computed tomography scans available to the core laboratory, and of which only 3 (0.2%) were deemed clinically significant by the site investigators, VCF-related AEs were rare (7 of 1421, 0.5%). Postfilter, venous thromboembolic events (none fatal) occurred in 93 patients (6.5%), including DVT (80 events in 74 patients [5.2%]), PE (23 events in 23 patients [1.6%]), and/or caval thrombotic occlusions (15 events in 15 patients [1.1%]). No PE occurred in patients following prophylactic placement. CONCLUSIONS: Implantation of VCFs in patients with venous thromboembolism was associated with few AEs and with a low incidence of clinically significant PEs.