Introducer curving technique to reduce tilting of transfemoral Günther Tulip IVC filter: in vitro study

Abstract

Background

Severe tilting of Günther Tulip filter (GTF) may be associated with difficulty in retrieval.

Purpose

To determine if an introducer curving technique of GTF can decrease the tilting degree of GTF in a caval model.

Material and Methods

The model was constructed by placing Dacron grafts in bifurcated glass tubes. The study included three groups: Right Straight Group (G₁), Left Straight Group (G₂), and Left Curved Group (G₃). In G₃, a 10–20° angle was curved on the metal introducer before insertion to decrease the angle between inferior vena cava axes and metal introducer (A_CM). Before GTF was released, the distance between the caval right wall and the apical hook (D_CH1), and A_CM were measured. The tilt angle of GTF (A_CF) was also measured.

Results

In G₁, GTF apex tended to center compared to G₂ (59% vs. 36%, P < 0.01). In G₃, GTF apex tended to center compared to G₂ (71% vs. 36%, P < 0.01). The differences of A_CF between G₁ and G₂ (2.66 ± 1.80 vs. 4.13 ± 2.07, P < 0.01) and between G₂ and G₃ (4.13 ± 2.07 vs. 2.39 ± 1.79, P < 0.01) were statistically significant. There were significant positive correlations between A_CM and A_CF, whereas significant negative correlations were detected between D_CH1 and A_CF in each group.

Conclusion

The oblique course of GTF delivery system relative to the axis of the cava causes filter tilt, and thus, curving the introducer prior to its introduction helps to reduce the filter tilt. We recommend a clinical study to determine whether the introducer curving technique improves filter centering and its retrievability.

Keywords

Tilt prevention inferior vena cava filters in vitro

It has been demonstrated that implantation of an inferior vena cava (IVC) filter is safe and effective in the prevention or reduction of fatal pulmonary embolism (PE) in numerous clinical studies (1–3). The retrieval of the IVC filter might decrease the possible long-term complications of filters, such as IVC occlusion, thrombosis, and recurrence of deep venous thrombosis (DVT). Filter retrieval has both potential benefits and risks. It has been impossible to accurately identify the period during which the patient might develop PE. It is not known how long the filter should be left in place and whether the risk of retrieval is less than the risk of a permanent filter. The Günther Tulip retrieval filter (GTF) (Cava MReye Filter Set; William Cook Europe, Bjaeverskov, Denmark) is designed for implantation through either a jugular or a femoral approach. It has four primary legs forming a half basket with anchors, the end of each leg attaching to the IVC wall and an apical hook on the filter apex facilitating filter retrieval (4). Although the incidence of significant filter tilting (>10°) is not high (13.2–16.6%) (5–7), severe tilting of the GTF may be associated with difficulty in retrieval (8–10).

It has been reported that the simple technique of keeping tension on the delivery system may prevent significant tilting of the jugular GTF in an in-vitro study (11). However, the technique of prevention for significant tilting of the femoral GTF has not been reported. We have observed that by the femoral approach, the degree of angle between the longitudinal axes of the superior segment of the GTF introducer and IVC is directly correlated with the extent of filter tilt. We hypothesize that by adjusting the orientation of the superior segment of the GTF introducer, we may be able to decrease the tilt of the filter. The purpose of this study was to determine whether adjustment of the GTF introducer decreases the degree of tilting of transfemoral GTF placement and to identify the mechanisms affecting tilting of transfemoral GTF in a caval model.

Material and Methods

A total of 300 femoral GTFs were deployed from a right graft without adjustment of the introducer (n = 100) and from a left graft with (n = 100) or without adjustment of the metal introducer (n = 100) by 10 radiology residents or fellows. Thirty filters and 30 introducers were used in the study. Ten introducers were bended and the angles of curvature were different. Every filter and introducer was re-used 10 times. In a Chinese article of IVC anatomy, 70 IVC were measured and the average diameter of IVC was found to be 26.0 ± 6.3 mm. The average frontal diameter of IVC 5 cm caudal to the left renal vein was 21.75 ± 3.01 mm and the angle between IVC and the approached iliac vein axis was 25.12° ± 9.46°measured on cavogram in one clinical study (12). The caval model was therefore constructed by placing one 25-mm diameter × 10-cm long and two 10-mm diameter × 10-cm long Dacron grafts inside a 27-mm internal diameter bifurcated glass tube in which the angles between the two iliac limbs and IVC were 175° and 155°, respectively (Fig. 1). The IVC model was not a circulatory flow model.

Fig. 1
Photo of IVC model (left) and metal introducer with filter (right)

The filter deployments were performed with fluoroscopy into an IVC model in a horizontal position on the angiographic table (the target-film distance was 92 cm and the target-object distance 87 cm). The operating procedures were as follows: the operator inserted an 8.5 F sheath of GTF set from a right or left graft, and the filter was preloaded into the metal end of the delivery catheter. The filter was then placed into the sheath and moved forward until the apical hook of the filter reached the confluence of the renal vein. The sheath was slowly withdrawn while the introducer was fixed, allowing the secondary legs of the filter to expand slightly and metal mount (metal filter holder at the end of the filter delivery catheter) to enter the lumen. Before the GTF was released, the introducer had to be revolved and adjusted to parallel with the axis of the IVC lumen as far as possible. Finally, the red hub was loosened and pulled toward the pin vise. Based on the types of technical maneuver and approach for the filter release, the study included three groups: right straight introducer Group (G₁), left straight introducer Group (G₂) and left curved metal introducer Group (G₃). Each maneuver was tested 10 times by each participating operator. In G₃, the introducer curving technique had been adapted. A 10–20° angle was curved on the metal introducer, and the distance between the introducer angle apex and the apical hook of the filter was 2 cm smaller than the distance between the lower edge of the renal vein opening and the bifurcation points of the IVC. The bent distal angle of the curved introducer wire was the same as the angle formed by the fixed end of the retrieval hook points and the free end of the retrieval hook. After being curved, the proximal and distal part of the curved introducer wire, and the fixed end and the free end of the retrieval hook were located in the same imaginary plane (Fig. 1).

Before the GTF was released, the distance between the right caval wall and the apical hook (D_CH1) and the angle between the longitudinal axis of IVC and the metal mount (A_CM) were measured. After release, the distance between the right caval wall and the apical hook (D_CH2), the angle between the longitudinal axes of the IVC and filter (A_CF), and the diameter of the IVC (D_IVC) were measured. The degree of tilting (DT) was measured and classified into five grades: grade I (severe right tilt), the apex was adherent to the right caval wall (D_CH2 less than 5 mm); grade II (right tilt), the apex was located with the right tilt at least 2.5 mm away from the center (D_CH2 between 5 mm and 10 mm); grade III (center), the apex was at or within 2.5 mm range of the cava center (D_CH2 between 10 mm and 15 mm); grade IV (left tilt), the apex was located with the left tilt at least 2.5 mm away from the center (D_CH2 between 15 mm and 20 mm); and grade V (severe left tilt), the apex was adherent to the left caval wall (D_CH2 more than 20 mm) (Fig. 2). A_CM and A_CF were considered positive when the metal mount or the apical hook of the filter rightwards tilted from the longitudinal axis of the IVC. Abbreviations and definitions of measured variables are described in Table 1. The measurement methods for measured variables are displayed in Fig. 3. In this study, the angle-measure and distance-measure tool of the picture archiving and communication system workstation were used. All the distance measurements were corrected by calibrating the diameter of Dacron graft (25mm).
Table 1
Abbreviations and definitions of measured variables

Abbreviation Definition

DT Degree of tilting

D_CH1 Distance between caval right wall and apical hook pre-releasing

D_CH2 Distance between caval right wall and apical hook post-releasing

A_CM Angle between axes of IVC and metal mount pre-releasing

A_CF Angle between the axes of IVC and filter

D_IVC Diameter of IVC

Fig. 2
Classification for degree of GTF tilting: severe right tilt (grade I) (a); right tilt (grade II) (b); centered (grade III) (c); left tilt (grade IV) (d); severe left tilt (grade V) (e)

Fig. 3
Example of measurement methods for the lengths and angles in study. Before filter release (a), ACM is defined as angle between axes of IVC and metal mount pre-releasing and DCH1 is defined as distance between caval right wall and apical hook pre-releasing. After filter release (b), ACF is defined as angle between the axes of IVC and filter and DCH2 is defined as distance between caval right wall and apical hook post-releasing and DIVC is defined as diameter of IVC model

The chi-square test was adopted to test the distribution difference of DT between G₁ and G₂, and G₂ and G_3, respectively. An independent sample t-test was performed to compare the difference of measured variables between G₁ and G₂, and between G₂ and G₃.

Pearson correlation coefficients were used to judge the relationships between the measured variables and A_CF in each group. Spearman correlation coefficients were used to determine the relationships between the measured variables and DT in each group.

The level of significance was set to 0.01 in all statistical analyses. The Statistical Package for the Social Sciences (SPSS 13.0 for windows, SPSS Inc, Chicago, IL, USA) was used to perform the above statistical analyses.

Results

The deployment of the Günther Tulip filter was completed in all 300 tests by the 10 participating operators. The degree of GTF tilting in each group revealed a divergent tendency, as shown in Table 2. In G₁, the apex of the filter tended to be at the center or near the center of the caval model (grade III) compared with that in G₂ (59% vs. 36%, P < 0.01). In G₂, the apex tended to be close to (grade II and IV) (52% vs. 38%, P < 0.01) or in direct contact with the caval wall (grades I and V) (12% vs. 3%, P < 0.01), as opposed to G₁. In G₃, the apex of the filter tended to be at the center or near the center of the caval model (grade III) compared with that in G₂ (71% vs. 36%, P < 0.01). In G₂, the apex tended to be close to (grades II and IV) (52% vs. 27%, P < 0.01), or in direct contact with the caval wall (grades I and V) (12% vs. 2%, P < 0.01), as opposed to G₃.
Table 2
Degree of GTF tilting in three groups

Grade I Grade II Grade III Grade IV Grade V X² value P value

G₁ 1 15 59 23 2 15.902* <0.01*

G₂ 9 20 36 32 3 37.491^† <0.01^†

G₃ 2 21 71 6 0

Between G₁ and G₂

^†Between G₂ and G₃

G₁, Right Straight Group; G₂, Left Straight Group; G₃, Left Curved Group

The results of the measured variables in each group are shown in Table 3. The differences of A_CF between G₁ and G₂ (2.66 ± 1.80 vs. 4.13 ± 2.07, P < 0.01), and between G₂ and G₃ (4.13 ± 2.07 vs. 2.39 ± 1.79, P < 0.01) were statistically significant. The differences of A_CM and D_CH1 between G₁ and G₂, and between G₂ and G₃ were statistically significant as well.
Table 3
Results of delivery configurations in three groups

G₁ (n = 100) G₂ (n = 100) G₃ (n = 100)

Variables ± s ± s P value ± s P value^†

A_CM (°) –5.50 ± 1.72 16.93 ± 1.72 <0.01 3.30 ± 2.48 <0.01

D_CH1 (mm) 13.41 ± 4.21 2.98 ± 0.77 <0.01 10.40 ± 1.20 <0.01

A_CF (°) 2.66 ± 1.80 4.13 ± 2.07 <0.01 2.39 ± 1.79 <0.01

Between G₁ and G₂

^†Between G₂ and G₃

A_CF, angle between axes of IVC and filter; A_CM, angle between axes of IVC and metal mount pre-releasing; D_CH1, distance between caval right wall and apical hook pre-releasing; G₁, Right Straight Group; G₂, Left Straight Group; G₃, Left Curved Group; s, standard deviation; , average value

The relationships between A_CM and A_CF in each group are displayed in Fig. 4. The Figure shows that the absolute value of A_CF was almost positively correlated with A_CM in G₁ and G₂. As A_CF approaches 0°, A_CM approaches –5° and 17° in G₁ and G_2, respectively. A_CF was also positively correlated with A_CM in G₃. A_CF tended toward 0° as A_CM approached 0°. The absolute value of A_CF increased with the increase of A_CM.
Fig. 4
Relationship between ACM and ACF in the three groups. ACM is defined as angle between axes of IVC and metal mount pre-releasing and ACF is defined as angle between the axes of IVC and filter

The relationship between D_CH1 and A_CF in each group is displayed in Fig. 5. The Figure shows that the absolute value of A_CF is negatively correlated with D_CH1 in G₁ and G₂. When D_CH1 approaches 13 mm (G₁) and 3 mm (G₂), A_CF will approach 0°. A_CF was negatively correlated with D_CH1 in G₃. A_CF tended toward 0° as D_CH1 approached 12 mm. The absolute value of A_CF increased with the decrease of D_CH1.
Fig. 5
Relationship between DCH1 and ACF in the three groups. DCH1 is defined as distance between caval right wall and apical hook pre-releasing and ACF is defined as angle between the axes of IVC and filter

Discussion

The location of the GTF mainly depended on the positions of the four anchors of the primary legs. When the longitudinal axis of the filter paralleled the longitudinal axis of the IVC, the tip and the pre-distended parts of the filter would not touch the vascular wall. When the metal mount was placed in the center of the IVC, each pair of the anchors would touch the IVC wall at the same velocity synchronously. The longitudinal axis of the filter would keep parallel with IVC. Seo et al. reported that when the four anchoring hooks of the GTF are on an imaginary line perpendicular to the longitudinal axis of IVC, the GTF would position at the center of IVC without tilt in his in vitro* study (13). When the metal mount was not placed at the center of the IVC, every pair of anchors would not touch the IVC wall simultaneously. If the pre-distended parts of GTF could not touch the IVC wall due to a large caval diameter, the filter tip would tilt in the direction of the anchor that first touched the IVC wall. When the caval diameter are small, the pre-distended parts of GTF might touch the IVC wall, and the force from the vascular wall would push the filter toward the IVC center, then the filter tip would slightly tilt. Wicky et al. (5) reported that tilt magnitudes increased with wider infrarenal caval diameters measured before GTF placement. In Sag's study (7), by using more of a caudal measurement, a weak positive correlation was detected between the caval diameter before GTF placement and the magnitude of GTF tilt at the first retrieval attempt.

When the axis of the filter was oblique to the axis of the IVC before being released and the filter tip did not touch the vascular wall, the factors that influenced the positions of the four anchors included the position of the metal mount and the angle and direction of the filter during releasing. The angle and direction of the filter depended on the vascular approach and anatomic tortuosity. Generally, when the filter was implanted through the right femoral vein, the tilting angle of the filter would be smaller and the filter would lean to the left wall of the IVC. When the filter was implanted through the left femoral vein, the tilting angle of filter was larger, and the filter leaned toward the right wall of the IVC.

When the axis of the filter was oblique to the axis of the IVC before being released and the filter tip touched the vascular wall, the factors influencing tilt included the position of the metal mount, the angle of the filter and the force from the vascular wall during the releasing procedure. Due to IVC generally located to the right side of the spine except in left sided IVC and caval duplication, the angle between the IVC and the left iliac vein axis was usually larger than the angle between IVC and the right iliac vein axis. From a left femoral approach, the longitudinal axis of the filter before being released tilted severely toward the right of the IVC axis. The filter tip touched the right wall of the IVC, and the position of the metal mount was deviated to the right of the IVC center line. Because the filter tip touched the right wall, it would be pushed leftward by a force from the vascular wall after being released. At the same time, the right primary leg would be pushed leftward by a force from the vascular right wall due to its earlier contact with the wall. The forces from the vascular wall conjointly rotated the filter tip to the left. Finally, the filter might tilt to the left or stay centered.

A_CM was one of the most important factors for GTF tilting. In the three groups, A_CF was approximately positively correlated with A_CM. However, it did not indicate that A_CF was close to 0° as A_CM approached 0°. Actually, A_CF was near 10° as A_CM approached 0° in G₁. The main reason is that if A_CM approaches 0° in G₁ and G₂, the filter and introducer must touch the vascular wall. Under this condition, the apical hook of the filter is close to the wall of the IVC. In addition, the force exerted by the vascular wall will produce apparent deflexion and together cause the filter to tilt. Therefore, this result did not conflict with the hypothesis that the technique of changing the curvature of the introducer decreases the tilt of filter. In G₃, the factor of the force from the IVC wall before releasing did not exist. Because A_CF tended toward 0° as A_CM approached 0° in G₃ and A_CF was near 10° as A_CM approached 0° in G₁, there are other important factors for GTF tilting in addition to A_CM.

Another important factor for GTF tilting was D_CH1. In G₂, the value of D_CH1 was smaller than that in the other two groups due to the relatively larger angle between the longitudinal axis of IVC and the axis of the introducer. Moreover, A_CF did not decrease with the decrease of the distance between the tip of GTF and the IVC center (1/2D_IVC – D_CH1) in G₂. This was also caused by the contact between GTF and the IVC wall before being released. In G₁ and G₃, A_CF was close to 0° when D_CH1 ranged from 12 mm to 13 mm, and the distance of 1/2D_IVC – D_CH1 was close to 0 mm. On the other hand, the absolute value of A_CF was less than 5° when D_CH1 was between 10 mm and 15 mm.

In this study, the incidence and extent of filter tilting was successfully minimized by adjusting the orientation of the superior segment of the introducer. By this method, the filter axis was parallel to the IVC axis (A_CM = 0°), while the tip of GTF was placed at the IVC center (1/2D_IVC – D_CH1 = 0 mm).

How to determine the optimal curving angle was the most important single factor in the curving technique and was the objective of this study. After the analysis of the data in this study and clinical practical experience, the optimal curving angle of the introducer was 5–15° smaller than the angle between infrarenal IVC and the approached iliac vein axis (12).

Comparing the results of each group, we found the tilting degree of each group to be different, with the largest in G₂ and the smallest in G₃. In the absence of the adjustment of the introducer, the angle between the approach vein and IVC was positively correlated with A_CF, whereas adjusting suitable curvature of the introducer reduced the influence of the approach vein and decreased the incidence rate of GTF tilting.

With the use of a suspended Dacron graft, it was possible to test the degree of tilting without filter migration by securely fixing the filter anchors. This model avoided the limitations of other in-vitro models constructed with plastic tubes, in which the caval wall was incapable of adaptive deformation, and the anchors of the filter were not securely attached on the caval wall (14).

In this study, we examined several factors related to GTF tilting in details. However, the results should be more significant after overcoming the model limitations used in this study. First, the evaluating methods for filter tilt included only anteroposterior X-ray photograph, not a lateral one. Therefore, the anteroposterior direction of tilting was not evaluated. Second, it was hard to evaluate the impact of the IVC size on tilting of GTF due to a fixed diameter of the caval model. Moreover, the transverse section of the graft had a round shape, differing with the elliptical shape of the human IVC. Third, the distance between the lower margin of the lowest renal vein and the bifurcation of IVC was 80 mm in our model, while the distance between the lower margin of the lowest renal vein and the bifurcation of IVC was 126.4 ± 23.2 mm in one clinical study (12). As a longer caval distance will help to reduce the oblique degree of the introducer, the extent of filter tilt in this study would be exaggerated slightly. Finally, it was unclear whether forces acting on the filter such as gravity, respiratory motion, cardiac pulsation, or increase in intra-abdominal pressure affected filter tilting during deployment and follow-up period before retrieval. Despite these limitations, our studies described a simplified and significant model system to understand GTF tilting.

In conclusion, the oblique GTF axis before being released is a major cause of transfemoral GTF tilting. In our in-vitro study, the incidence and extent of the GTF tilting was reduced by adjusting the orientation of the filter introducer. The in-vitro Dacron/glass model of the vena cava is different from the human iliocaval system, so the result from this model cannot be directly translated to humans. The introducer curving technique is applicable by infrarenal placement but not by the jugular route or in the presence of circumaortic or retroaortic left renal vein where a shorter caval length may preclude infrarenal filter placement. We recommend a clinical study to determine whether the introducer curving technique improves filter centering and its retrievability.

Abbreviation	Definition
DT	Degree of tilting
D_CH1	Distance between caval right wall and apical hook pre-releasing
D_CH2	Distance between caval right wall and apical hook post-releasing
A_CM	Angle between axes of IVC and metal mount pre-releasing
A_CF	Angle between the axes of IVC and filter
D_IVC	Diameter of IVC

	Grade I	Grade II	Grade III	Grade IV	Grade V	X² value	P value
G₁	1	15	59	23	2	15.902*	<0.01*
G₂	9	20	36	32	3	37.491^†	<0.01^†
G₃	2	21	71	6	0

	G₁ (n = 100)	G₂ (n = 100)		G₃ (n = 100)
A_CM (°)	–5.50 ± 1.72	16.93 ± 1.72	<0.01	3.30 ± 2.48	<0.01
D_CH1 (mm)	13.41 ± 4.21	2.98 ± 0.77	<0.01	10.40 ± 1.20	<0.01
A_CF (°)	2.66 ± 1.80	4.13 ± 2.07	<0.01	2.39 ± 1.79	<0.01

Footnotes

ACKNOWLEDGMENTS

This article was supported by research grants from the Scientific Research Fund of Liaoning Science and Technology Agency, China (No. 2008225010-5) and the Scientific Research Fund of Liaoning Education Agency, China (No. 2007T183) and the Scientific Research Fund of First Hospital of CMU (No. FSFH1006).

Conflict of interest: None.

References

Yamagami

, Yoshimatsu

, Matsumoto

Prophylactic implantation of inferior vena cava filter during endovascular therapies for deep venous thrombosis of the lower extremity: is it necessary?

Acta Radiol 2008;4:391–3

Yamada

, Ishikura

, Ota

, Pulse-spray pharmacomechanical thrombolysis for proximal deep vein thrombosis. Eur J Vasc Endovasc Surg 2006;2:204–11

Schleich

, Morla

, Laurent

, Long-term follow-up of percutaneous vena cava filters: a prospective study in 100 consecutive patients. Eur J Vasc Endovasc Surg 2001;5:450–7

Neuerburg

, Gunther

, Vorwerk

, Results of a multicenter study of the retrievable Tulip Vena Cava Filter. Cardiovasc Intervent Radiol 1997;1:10–6

Wicky

, Doenz

, Meuwly

, Clinical experience with retrievable Günther Tulip vena cava filters. J Endovasc Ther 2003;5:994–1000

Terhaar

, Lyon

, Given

, Extended interval for retrieval of Günther Tulip filters. J Vasc Interv Radiol 2004;11:1257–62

Sag

, Stavas

, Burke

, Analysis of tilt of the Günther Tulip filter. J Vasc Interv Radiol 2008;5:669–76

Looby

, Given

, Geoghegan

, Gunther Tulip retrievable inferior vena caval filters: indications, efficacy, retrieval, and complications. Cardiovasc Intervent Radiol 2007;1:59–65

Rosenthal

, Wellons

, Hancock

, Retrievability of the Günther Tulip vena cava filter after dwell times longer than 180 days in patients with multiple trauma. J Endovasc Ther 2007;3:406–10

10.

Hoppe

, Nutting

, Smouse

, Günther Tulip filter retrievability multicenter study including CT follow-up: final report. J Vasc Interv Radiol 2006;6:1017–23

11.

Lopera

, Araki

, Kirsch

, A modified technique to minimize filter tilting during deployment of the Günther Tulip filter: in vitro study. J Vasc Interv Radiol 2005;11:1539–44

12.

Xiao

, Huang

, Shen

, Introducer Curving Technique for the prevention of Tilting of Transfemoral Günther Tulip IVC Filter. Korean J Radiol 2012;13:483–91

13.

Seo

, Cha

, Park

, Detection and correction of anterior or posterior tilting of the Günther-Tulip filter in the inferior vena cava and correction by repositioning: a phantom study and preliminary clinical experiences. J Vasc Interv Radiol 2007;3:427–36

14.

Katsamouris

, Waltman

, Delichatsios

, Inferior vena cava filters: in vitro comparison of clot trapping and flow dynamics. Radiology 1988;2:361–6

	G₁ (n = 100)	G₂ (n = 100)		G₃ (n = 100)
Variables	± s	± s	P value*	± s	P value^†
A_CM (°)	–5.50 ± 1.72	16.93 ± 1.72	<0.01	3.30 ± 2.48	<0.01
D_CH1 (mm)	13.41 ± 4.21	2.98 ± 0.77	<0.01	10.40 ± 1.20	<0.01
A_CF (°)	2.66 ± 1.80	4.13 ± 2.07	<0.01	2.39 ± 1.79	<0.01