Predictors of endobronchial forceps utilization for inferior vena cava filter retrieval: when snare retrieval fails

Background: Endobronchial forceps are commonly used for complex IVC filter removal and after initial attempts at IVC filter retrieval with a snare have failed. Currently, there are no clear guidelines to help distinguish cases where primary removal should be attempted with standard snare technique or whether attempts at removal should directly be started with forceps. This study is aimed to identify clinical and imaging predictors of snare failure which necessitate conversion to endobronchial forceps.

Methods: Retrospective analysis of 543 patients who underwent IVC filter retrievals were performed at three large quaternary care centers from Jan 2015 to Jan 2022. Patient demographics and IVC filter characteristics on cross-sectional images (degree of tilt, hook embedment, and strut penetration, etc.) were reviewed. Binary multivariate logistic regression was used to identify predictors of IVC filter retrieval where snare retrieval would fail.

Results: Thirty seven percent of the patients (n = 203) necessitated utilization of endobronchial forceps. IVC filter hook embedment (OR:4.55; 95%CI: 1.74-11.87; p = 0.002) and strut penetration (OR: 56.46; 95% CI 20.2-157.7; p = 0.001) were predictors of snare failure. In contrast, total dwell time, BMI, and degree of filter tilt were not associated with snare failure. Intraprocedural conversion from snare to endobronchial forceps was significantly associated with increased contrast volume, radiation dose, and total procedure times (p < 0.05).

Conclusion: IVC filter hook embedment and strut penetration were predictors of snare retrieval failure. Intraprocedural conversion from snare to endobronchial forceps increased contrast volume, radiation dose, and total procedure time. When either hook embedment or strut penetration is present on pre-procedural cross-sectional images, IVC filter retrieval should be initiated using endobronchial forceps.

Level of evidence: Level 3, large multicenter retrospective cohort.