Southern Association for Vascular surgery

October 17, 2008

Dynamic Geometry and Spatial Orientation of the IVC: Risk Factors for Migration of Vena Caval Filters

Erin H Murphy*, Frank R Arko, III, Clayton K Trimmer*, Christopher K Zarins
University of Texas Southwestern Medical Center, Dallas, TX

Introduction:
Each year more than 140,000 percutaneous vena caval filters (VCFs) are placed worldwide. While VCFs effectively decrease the rate of fatal pulmonary embolism, filter related complications including device migration have been reported with all currently available permanent and retrievable filters. Clinically significant filter migration occurs in 3-12% of patients and when defined more strictly, up to 63% of patients experience migration of greater than 1 cm from the original deployment site.
Potential causes of migration include inaccurate sizing or deployment, poor filter alignment, j-wire entrapment during central line placement, clot pushing the filter distally towards the heart, and an increase in IVC diameter after resuscitation or with normal respiratory variation and Valsalva maneuvers. However, the underlying causes of filter migration are unknown and prevention of this complication may rely a more thorough understanding of the dynamic movement of the IVC. Currently, vena caval dynamics are poorly understood and are largely based on qualitative image assessment. Despite advances in imaging capabilities, thus far, there have been no quantitative descriptions of the 3D spatial orientation or dynamic geometry of the IVC.
Herein we describe the 3D spatial orientation and dynamic geometry of the infrarenal IVC in response to changes in intravascular volume. Serial contrasted computed tomographic (CT) scans obtained in severely injured trauma patients during hypovolemic and fluid resuscitated states were evaluated and compared. Trauma patients were ideal for this study as they experience the extremes of volume loss and resuscitation. Implications for VCF placement and design, as well as specific risk factors for VCF migration are discussed.
Methods:
Between March 2005 and March 2006, 559 patients underwent VCF placement using venacavography image guidance. Of these, thirty trauma patients had abdominal CT scans at different stages of fluid resuscitation. The first CT was obtained within the first hour of admission, at the time of marked hypovolemia. The second scan was obtained after fluid resuscitation.
Vena cava dimensions were measured in each patient at both time points at 1cm and 5cm below the renal veins, corresponding to the most common location of IVC filter placement, using a Vista Imaging Viewer. Diameters were recorded from the center of the lumen in the major and minor axis. A CT representation of the diameter of the IVC as visualized on standard anterior-posterior view in venography was determined by projecting the images in their true lateral plane. Lumen contour was traced on axial CT scan images and cross-sectional area was calculated. Volume of the infrarenal IVC was calculated from area and diameters according to Simian’s formula for calculating volume of a tapered log.
Changes in diameter and volume were evaluated using a Student’s t test for paired data. Linear regression was used to correlate changes in diameter and volume. Analyses of measurement method comparison data according to Bland and Altman were performed to analyze repeatability. Intraobserver and interobserver variability showed no significant differences within or between observers.
Results:
Mean patient age was 32 +/- 11.5 (17-57) years. Mean trauma injury severity score was 40.2 +/- 13.1(34-75). All 30 patients had evidence of hypovolemia on initial CT, determined by the presence of a collapsed IVC (<15mm minor axis). Mean time between admission (hypovolemic) and follow-up (fluid resuscitated) CT scan was 49.5 days (range:1-202). All follow-up scans demonstrated expansion of the IVC with a corresponding increase in volume of the infrarenal segment (Image I). The volume of the IVC segment increased from 6.9 +/- 2.2 (range: 3.1-12.4) to 15.7 +/- 5.0 (range: 9.2 - 28.5) cm3 (p<0.01), representing more than a two-fold increase in intravascular volume between the hypovolemic and fluid resuscitated state.
Table I demonstrates the dynamic changes of the major and minor axes of the IVC that occurred with fluid resuscitation. With fluid resuscitation, the minor axis of the IVC increased by 6.9-7.6 mm on average, while the major axis only increased 1.2 to 1.8 mm(p<0.001)(Table I). Overall, 84% of the increase in volume of the infrarenal segment between scans was accounted for by an increase in the diameter of the short axis (r2=0.84), while the long axis accounted for only a 0.04% increase in IVC volume of the infrarenal segment (r2=0.04).



IVC Dimension	Volume Status		Dynamic Change
Minor Axis	Hypovolemic IVC Diameter(mm)	Volume Resuscitated IVC Diameter (mm)	Diameter Change (mm)	p-value
1 cm Below Renals	9.2 +/- 3.0 (4.5-15)	16.8 +/- 4.1 (10.2-20.7)	7.6 +/-4.4 (-0.2-17.4)	p<0.001
5 cm Below Renals	9.8 +/- 3.2 (3.3-18.1)	16.8+/-3.7 (9-20.5)	6.9 +/- 4.4 (0.6-20.15)	p<0.001
Major Axis
1 cm Below Renals	24.9+/- 3.9 (14-30.5)	26.7+/-3.8 (21-32)	1.78 +/-2.7 (-4.7 - 7.8)	p<0.001
5 cm Below Renals	23.5+/-2.5 (20-29.3)	24.7 +/-2.1 (21.4-29.1)	1.2+/-2.3( -2.3 - 6.4)	p<0.001
Venographic Diameter
1 cm Below Renals	22.8 +/- 3.85 (11.5-29.1)	25.4 +/- 3.78 (19-35.7)	2.6 +/- 3.2 (-2.1 - 8.65)	p<0.001
5 cm Below Renals	21.1 +/- 2.8 (19-25.8)	23.3 +/- 1.98 (20.6-25.5)	2.1 +/- 2.9 (-1.7-9)	p<0.001

Table I
The major and minor axes of the IVC were oblique. This resulted in significantly shorter measurements in the projected plane of the venacavographic diameter when compared to the major axis dimension. (Table II).



	Volume Status
Major Axis - Venographic Diameter	Hypovolemic IVC Diameter (mm)	Volume Resuscitated IVC Diameter (mm)	p-value
1 cm Below Renals	2.16 +/- 1.4 (0-4.75)	1.37 +/- 1.24 (0-5.75)	p<0.001
5 cm Below Renals	2.39+/- 1.75 (0-6.75)	1.5 +/- 1.36 (0-5.95)	p<0.001

Table II
All patients in this study had IVC filters placed using standard venogram. This resulted in significant underestimation of the true maximal IVC diameter. In 3 patients who had retrievable filters placed, measuring the larger oblique axis 5 cm below the renals (site of VCF fixation) instead of the projected diameter seen on venogram, would have made a difference in filter choice. In these patients the oblique maximal diameter was 3-4 mm larger than the venographic diameter and exceeded 28mm, the cutoff for safe placement of these filters, while the projected venographic measurement did not (p<0.01).
Conclusions:
In the supine position, the cross-sectional contour of the inferior vena cava is elliptical with the major and minor axes lying obliquely. This is contrasting to current descriptions of the cava which depicts the major and minor axes as lateral and anterior-posterior respectively. We have demonstrated that the long axis may be up to 45% larger than the lateral-lateral diameter obtained on standard venogram, which is often used for filter sizing and placement. Current filter placement by venogram may therefore result in filter undersizing in up to 10% of patients.
Furthermore, the vena cava deforms anisotropically in response to intravascular volume changes, with greatest movement seen in the minor axis. However, small movements in the major axis, which should be used for filter sizing, are also statistically significant in the infrarenal cava.
Filter sizing should be based on the major axis which is not visible in most patients on standard venogram. Further, device design must accommodate the elliptical nature of the cava and allow for significant diameter changes in both the major and minor caval axes. Failure to take these factors into consideration predisposes patients to device failure and filter migration.
27.0 mm

Back to Annual Meeting
Back to Program