CardioVascular and Interventional Radiology

, Volume 36, Issue 4, pp 998–1005

Safety and Efficacy of the Gunther Tulip Retrievable Vena Cava Filter: Midterm Outcomes

Authors

    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
  • Rebecca J. Mueller
    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
  • Marcus R. Luciano
    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
  • Nicole N. Lee
    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
  • Anne T. Michaels
    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
  • John M. Gemery
    • Department of Radiology, Section of Vascular and Interventional RadiologyDartmouth-Hitchcock Medical Center
Clinical Investigation

DOI: 10.1007/s00270-012-0517-7

Cite this article as:
Hoffer, E.K., Mueller, R.J., Luciano, M.R. et al. Cardiovasc Intervent Radiol (2013) 36: 998. doi:10.1007/s00270-012-0517-7

Abstract

Purpose

To evaluate of the medium-term integrity, efficacy, and complication rate associated with the Gunther Tulip vena cava filter.

Methods

A retrospective study was performed of 369 consecutive patients who had infrarenal Gunther Tulip inferior vena cava filters placed over a 5-year period. The mean patient age was 61.8 years, and 59 % were men. Venous thromboembolic disease and a contraindication to or complication of anticoagulation were the indications for filter placement in 86 % of patients; 14 % were placed for prophylaxis in patients with a mean of 2.3 risk factors. Follow-up was obtained by review of medical and radiologic records.

Results

Mean clinical follow-up was 780 days. New or recurrent pulmonary embolus occurred in 12 patients (3.3 %). New or recurrent deep-vein thrombosis occurred in 53 patients (14.4 %). There were no symptomatic fractures, migrations, or caval perforations. Imaging follow-up in 287 patients (77.8 %) at a mean of 731 days revealed a single (0.3 %) asymptomatic fracture, migration greater than 2 cm in 36 patients (12.5 %), and no case of embolization. Of 122 patients with CT scans, asymptomatic perforations were identified in 53 patients (43.4 %) at a mean 757 days.

Conclusion

The Gunther Tulip filter was safe and effective at 2-year follow-up. Complication rates were similar to those reported for permanent inferior vena cava filters.

Keywords

Deep-vein thrombosisInferior vena cava filterPulmonary embolismVenous intervention

Introduction

Indications for placement of inferior vena cava (IVC) filters are well established, but the indications for retrieval are not [1]. Whether a filter should be removed depends on whether the risk of a filter-related complication exceeds the benefit [2, 3].

Although the long-term benefit of a filter in the prevention of pulmonary embolus has been documented [4, 5], long-term risk data on retrievable filters are limited, as research has largely focused on aspects of their retrieval [6]. Publications that assess the risk associated with the Gunther Tulip vena cava filter (Cook Medical Inc., Bloomington, IL) are limited by the inclusion of various devices, a small number of patients followed beyond retrieval, or short duration of follow-up [719] (Table 1). Although there is a real fracture risk for any filter, whether this is a clinically significant concern in patients who receive the Gunther Tulip filter is unknown [2, 20, 21].
Table 1

Gunther Tulip filter published outcomes

Study

Years

No. of patients

Retrieval

 

Complication (%)

Follow-up (month)

 %

At day

Death

PE

DVT

IVC occlusion

Migration

Perforation

Fracture

Neuerburg [13]

1997

83

6.0

NR

20.5

3.6

0.0

3.6

3.6

2.4

0

4.5

Millward [14]

2001

90

57.8

9

7.8

0.0

1.1

2.2

0.0

0.0

0

3.3

De Gregorio [15]

2003

87

78.2

13

0.0

0.0

0.0

0.0

0.0

0.0

NR

NR

Wicky [16]

2003

71

46.5

8

25.4

0.0

1.4

0.0

2.8

0.0

NR

4.1

Terhaar [17]

2004

53

9.4

43.6

11.3

1.9

0.0

0.0

0.0

0.0

NR

12.8

Hoppe [10]

2006

41

56.1

11.1

0.0

2.4

12.2

2.4

2.4

0.0

NR

3.0

Ray [7]

2006

143

42.7

15.6

0.0

0.7

4.2

1.4

0.0

Common

NR

NR

Keller [9]

2007

93

50.0

11

14.1

2.2

4.3

5.4

1.1

0.0

0

8.7

Looby [12]

2007

147

24.5

33.6

25.9

0.7

0.0

0.0

0.0

0.7

0

8.1

Kim [27]

2008

50

28.0

9.7

NR

4.0

18.0

2.0

NR

NR

NR

10.4

Given [11]

2008

317

59.3

77

4.1

0.3

0.0

0.3

0.0

0.0

NR

2.6

Ota [18]

2008

118

50.8

10.1

9.3

0.8

NR

2.5

NR

19.5

0

12.3

Seshadri [19]

2008

42

28.6

14

31.0

4.8

0.0

0.0

0.0

0.0

NR

24.6

Total/mean

 

1284

45.9

22.4

10.6

1.0

1.8

2.5

0.9

3.8+

0

14.1

PE pulmonary embolism, DVT deep-vein thrombosis, IVC inferior vena cava, NR not reported

The purpose of this investigation was to evaluate the medium-term (2–4 year) prevalence of recurrent venous thromboembolism (VTE), fracture, and other complications associated with the Gunther Tulip retrievable filter when placed as a permanent device.

Materials and Methods

This retrospective, single-center study was conducted after institutional review board approval was obtained. The study was compliant with the Health Insurance Portability and Accountability Act. The requirement for informed consent for participation in this retrospective study was waived.

Study Design

Consecutive patients who had undergone Gunther Tulip IVC filter placement were identified from hospital billing data at a 400-bed level 1 trauma and tertiary academic medical center from January 2004 through December 2008.

Patient and procedural information was obtained from entries in the electronic medical record (EPIC Systems Corporation, Verona, WI) and picture archiving and communications system (Centricity; GE Healthcare, Waukesha, WI). Information abstracted included demographics, details of the incident VTE event, indication for filter placement (vs. anticoagulation), risk factors for venous thromboembolic disease, retrieval status, the use of pharmacologic anticoagulation, and follow-up clinical data.

Assessment for complications was performed through review of the electronic hospital records, clinic notes, and primary care physician records. Evaluation for asymptomatic IVC filter dysfunction was performed by the investigators’ independent review of all CT studies and radiographic images that included the filter. Other than those cases of filter retrieval or chest CT studies in patients with suspected new or recurrent pulmonary embolism (PE), the imaging studies were invariably obtained for reasons unrelated to the filter. CT images were obtained at a variety of institutions via a variety of techniques. Section collimation was 5 mm or less. If an abnormality was identified, the study with its earliest presence was considered the date of the event, and the medical record was further reviewed to see if there were associated symptoms. Survival data were obtained from hospital records, state death certificates, and the Social Security Death Index. Data were collected in an Excel spreadsheet (Microsoft, Redmond, WA). Review was performed in accordance with the Society of Interventional Radiology reporting standards [22, 23].

The primary outcome measures were effectiveness and safety. Effectiveness was assessed by the incidence of new objectively documented PE (i.e., positive CT pulmonary angiogram or ventilation–perfusion scan). Safety was measured by the incidence of symptomatic complications of caval thrombosis, filter perforation, migration, and device fracture and/or embolization. The diagnosis of caval thrombosis or filter thrombus required contrast-enhanced CT or venographic documentation. Filter migration was diagnosed if a radiographic image showed a change of at least 2 cm in the position of the filter from the immediate postdeployment position, assessed in relation to an adjacent bony landmark. Filter perforation was defined as axial CT evidence of a filter leg extending at least 3 mm from the caval wall, and identified beyond the caval wall on at least two axial images [24]. Fracture assessment required the availability of a radiograph or CT of the entire filter of sufficient quality to assess the integrity of each of the four primary legs.

Secondary safety measures were imaging evidence of nonsymptomatic IVC occlusion, fracture or embolization, filter migration, or filter perforation. Other recorded events possibly related to the filter or the treatment of venous thromboembolic disease included worsening or new symptomatic deep-vein thrombosis (DVT), and bleeding associated with anticoagulation. DVT was defined as any new episode of objectively documented venous thrombosis in the pelvis or lower extremities [22, 23].

Clinical Presentation

Over the 5-year period, 5,310 patients were diagnosed with PE and/or DVT. Of these, 462 received therapeutic IVC filters. An additional 61 patients had an IVC filter placed for prophylaxis. Of the 523 IVC filters placed, 454 (89.1 %) were retrievable, reflecting an institutional decision to stock only retrievable filters rather than a specific intention to retrieve. The choice of filter was based on operator preference or the available inventory. A total of 371 Gunther Tulip filters were placed in 369 patients, which comprise the study group. One patient had both superior and IVC filters placed; another had a filter misplaced and replaced at the initial procedure. The mean patient age was 61.8 ± 16.5 (range 19–93) years, and 59.1 % were men (Table 2).
Table 2

Patient demographics

Characteristic

Value

No. of patients

369

Gender (male)

59 %

Age (year), mean (range)

61.8 (19–93)

Race (white)

97 %

Hypercoagulable state

32 %

 Tumor

69 %

 Hematologic

29 %

 Postpartum

1 %

Trauma

31 %

Patients had filters placed for DVT (50.7 %), DVT and PE (13.2 %), PE (22.0 %), or prophylactically (14.1 %). Indications are detailed in Table 3. PE was usually documented by contrast-enhanced CT and in few instances ventilation–perfusion scan. DVT diagnosis was documented by ultrasound or rarely, CT venography.
Table 3

Indications for filter placement

Indication

Prophylaxis

DVT

PE with or without DVT

Total

 %

Contraindication to anticoagulation

    

67.8

 Intracranial disease

19

66

23

108

29.3

 Perioperative

19

34

27

80

21.7

 Acute trauma

6

8

4

18

4.9

 Active hemorrhage

5

18

13

36

9.8

 Poor compliance

 

1

1

2

0.5

 Thrombocytopenia

 

3

3

6

1.6

Complication to anticoagulation

    

26.6

 Hemorrhage on anticoagulation

3

29

17

49

13.3

 New PE/DVT

 

16

28

44

11.9

 Heparin-induced thrombocytopenia

 

4

1

5

1.4

In addition to anticoagulation

    

5.7

 Low cardiopulmonary reserve

  

10

10

2.7

 Iliocaval thrombus

 

5

2

7

1.9

 Transcatheter therapy

 

3

1

4

1.1

 Total, n (%)

52 (14.1)

187 (50.7)

130 (35.2)

369 (100)

 

DVT deep-vein thrombosis, PE pulmonary embolism

Filters placed in the absence of VTE (n = 52), were placed perioperatively (48.1 %) or after head (38.5 %) or long bone (23.1 %) trauma. These patients had a mean of 2.3 risk factors each, including prior VTE (34.6 %), age over 65 years (32.7 %), immobility (17.2 %), or malignancy (9.6 %).

The Gunther Tulip filter has been used in our institution as a permanent filter since its approval in 2000, and its use expanded once approved as a removable device in 2003. The filter is constructed of conichrome (an MRI-compatible cobalt, nickel, and chromium alloy). The four primary struts are barbed at the ends to secure it to the caval wall, and there are eight secondary limbs that bridge the primary struts, creating a petal configuration.

Interventional radiologists and vascular surgeons placed all filters in dedicated angiographic suites using standard techniques: initial 21-gauge needle access, venacavography, and a final image to document infrarenal deployment. Filters were placed via the right common femoral vein (n = 251, 68.0 %), unless there was local thrombus or wound, in which case the left common femoral vein was used (n = 45, 12.2 %). In some patients, particularly those who were coagulopathic or obese, the right internal jugular vein was used (n = 73, 19.8 %).

When filter retrieval was elected, a right internal jugular vein access was used to perform venacavography to exclude thrombus within the filter. If there was no or minimal thrombus, the filter hook was snared with an Amplatz loop snare (Cook Medical) and the filter retrieved as per the manufacturer’s guidelines. Postretrieval venography was performed.

Routine clinical follow-up entailed evaluation of the insertion site and general health status of the patient within 24 h of the procedure (placement or retrieval). There was no systematic approach to follow patients who were potentially eligible for retrieval.

Statistical Analysis

Statistical analyses were performed by SPSS software, version 19.0 (IBM, Armonk, NY) for Windows (Microsoft, Redmond, WA). Descriptive statistics were used to summarize data. Categorical data were summarized as counts and percentages. Continuous data were summarized as means with standard deviation, with median and range where appropriate. The primary end points of this study were reported as proportions of patients at specific follow-up time points. The actuarial method of Kaplan–Meier was used to calculate the probability of an event as a function of length of follow-up after filter insertion. Cox regression analysis was used to perform multivariate analysis of variables associated with events and are presented as odds ratios with 95 % confidence intervals. Patients were censored from time-to-event analysis when lost to follow-up, when they died, or if the filter was successfully removed. A p value of less than 0.05 was considered significant.

Results

Filter Implantation

Technical success was achieved in 369 (99.5 %) of 371 Gunther Tulip filter deployments. Immediate complications were encountered in five (1.3 %) of 371 placements. These included two episodes of technical failure as a result of malposition; one filter was recovered and repositioned, the other was removed and a second filter placed. Three minor complications consisted of one arrhythmia treated medically, and two hematomas (one groin, one neck), which were stable after manual compression. Few anatomic variants were encountered during the study. There were three (0.81 %) circumaortic renal veins, eight (2.2 %) left iliac vein compressions of greater than 50 % luminal diameter, and 11 (2.9 %) left and two (0.5 %) right bifid common iliac veins.

At the time of insertion, 21 patients (5.7 %) were receiving anticoagulation therapy. Anticoagulation was generally started as soon as clinically indicated after the placement of the IVC filter, such that 53.7 % of patients were receiving anticoagulation therapy at 30 days (Table 4).
Table 4

Patient survival, follow-up, and anticoagulation status at various time points

Characteristic

1 month

3 month

6 month

12 month

24 month

36 month

48 month

60 month

Survival

310 (84.3 %)

279 (76.1 %)

256 (69.8 %)

237 (64.6 %)

211 (57.6 %)

195 (53.2 %)

145 (50.0 %)

97 (48.9 %)

With follow-up

285 (91.9 %)

240 (87.0 %)

220 (85.9 %)

202 (85.2 %)

174 (82.5 %)

136 (69.7 %)

80 (55.2 %)

47 (48.5 %)

Receiving anticoagulation therapy

153 (53.7 %)

149 (62.1 %)

129 (58.6 %)

104 (51.5 %)

86 (49.4 %)

65 (47.8 %)

39 (48.8 %)

15 (31.9 %)

Filter removal was attempted in 10 cases and was successful in nine (2.4 % of all patients). These procedures occurred at a mean of 66 (range 5–587, median 30) days after initial placement. Other than one failure, there were no complications associated with the retrievals.

Follow-up

The mean clinical follow-up was 780 (±779, median 616, range 1–2843) days. Survival at 2 and 4 years was 57.2 and 39.6 %, respectively (Table 4; Fig. 1). Cause-of-death information was available in 47.6 % (90 of 189) of patients. Between 1 month’s and 4 years’ follow-up, approximately half of the patients were receiving anticoagulation therapy (Table 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00270-012-0517-7/MediaObjects/270_2012_517_Fig1_HTML.gif
Fig. 1

Kaplan–Meier analysis of time to pulmonary embolism, deep-vein thrombosis, and death over a period of 7 years after filter placement

New VTE events occurred in 63 (17.1 %) of the 369 patients, 27 (42.8 %) of whom were anticoagulated at the time. New or recurrent PE occurred in 12 patients (3.3 %) (Table 5, Fig. 1) at a mean of 256 (range 10–899) days. Five patients (41.7 %) were anticoagulated at the time. Two patient deaths were related to filter failure—one massive PE, and one large PE with stroke in the presence of an intracardiac shunt. These two patients were both hypercoagulable and were anticoagulated at the time of the fatal PE. Six of the 12 patients with PE after filter placement had contrast-enhanced CT that permitted evaluation of filter patency; one had a filling defect within the filter, and four had IVC occlusion. By Kaplan–Meier life table analysis, the cumulative incidence of new or recurrent PE was 3 % at 1 year, 4.6 % at 2 years, and leveled off at 5.2 % from years 3 to 5. Cox logistic regression was performed to assess the impact of six independent variables on the development of PE: gender, age, level of initial thrombus, hypercoagulability, lower extremity access, and anticoagulation. A protective effect of anticoagulation was found, with an odds ratio of 0.195 (95 % confidence interval 0.056–0.673, p = 0.010).
Table 5

Details of patients with filter failure and new or recurrent PE

Age (year)

Gender

Diagnosis

VTE

Indication

Specific

Risk factors

Anticoagulation therapy at time of PE

Time to PE (d)

IVC occlusion

41

F

Head trauma

Prophylaxis

Contraindication

Intracranial bleeding

Immobilization

No

10

Yes

37

M

Head trauma

Iliac DVT

Contraindication

Intracranial bleeding

Immobilization

Yes

12

Yes

72

M

Thigh hemorrhage

Popliteal DVT + PE

Complication

Intramuscular bleeding

Coronary artery disease

No

24

Yes

72

F

Stroke

Prophylaxis

Contraindication

Perioperative

Lupus anticoagulant, stroke, prior VTE

No

62

No

53

M

Glioblastoma

PE

Contraindication

Noncompliant

Cancer, prior VTE

No

64

No

49

F

PE

Iliac DVT + PE

Complication

PE/DVT despite anticoagulation

Cancer, stroke, HIT, prior VTE

Yes

81

Yes

64

M

Leukemia

Calf DVT + PE

Complication

Thrombocytopenia

Cancer

No

83

Yes

81

M

GI bleeding

Femoral DVT

Contraindication

Active hemorrhage

COPD

Yes

92

No

84

F

Head trauma

Femoral DVT

Complication

Intracranial bleeding

None known

No

428

No

44

M

Perioperative

Prophylaxis

Contraindication

Perioperative

Obesity, prior VTE

No

657

No

69

M

Intracranial hemorrhage

Calf DVT

Complication

Intracranial bleeding

None known

No

661

No

66

M

Stroke

Femoral DVT + PE

Complication

Intracranial bleeding

Cancer

No

899

No

PE pulmonary embolism, VTE venous thromboembolism, IVC inferior vena cava, DVT deep-vein thrombosis, COPD chronic obstructive pulmonary disease, HIT heparin-induced thrombocytopenia

New or progression of existing DVT was identified in 53 (14.4 %) of 369 patients at a mean of 329 (range 6–2045) days; 23 (43.4 %) of them were anticoagulated at the time (Fig. 1). By Kaplan–Meier life table analysis, the rate of new or recurrent symptomatic DVT was 11.6 % at 1 year, 15.7 % at 2 years, and 21.2 % at 4 years (Fig. 1).

New IVC thrombus was identified in 15 patients (4.1 %) on contrast CT or venography, five of whom had PE (described above). Of the 10 patients without PE, two had symptoms in association with complete occlusion of the IVC at the filter; both occurred within 2 weeks of filter placement, and one patient was anticoagulated. Initial thrombus in those two patients was at the calf and popliteal levels, respectively. Half of the eight asymptomatic patients had complete IVC occlusion; the others had filling defects within the filters identified on contrast CT or venography. These patients were all anticoagulated except one of those with an occlusion. Cox regression analysis of factors related to any recurrent VTE found therapeutic anticoagulation to be associated with a reduced risk (odds ratio 0.427, 95 % confidence interval 0.254–0.718, p = 0.001), and a hypercoagulable state (hematologic factor or cancer) was associated with increased risk (odds ratio 1.787, 95 % confidence interval 1.032–3.093, p = 0.038).

Postthrombotic syndrome was identified in 15 patients at a mean of 1.9 (range 0.4–5.5) years. Major hemorrhage was a complication in four patients, all of whom were anticoagulated, at a mean 1134 (range 300–1677) days.

There were no symptomatic filter migrations, fractures, or caval perforations. Of 287 patients with radiographic image follow-up at mean of 730 (±730, median 483, range 1–2,843) days, asymptomatic filter migration of greater than 2 cm was documented in 36 patients (12.5 %) and filter fracture was identified in one patient (0.36 %). The fracture was identified at day 1108, without interval imaging or a potentially causative event. Two opposite legs of the filter were broken, one 1.2 cm and the other 4.2 cm from the apex. Fragments remained adjacent to the filter with no evidence of distal embolization on chest radiograph.

Perforation assessment by CT scan was available in 129 patients at mean imaging follow-up of 774 (±758, median 561, range 1–2,843) days. Asymptomatic IVC perforation of greater than 3 mm was found in 59 patients (43.4 %) evaluated by venography (n = 2) or CT scan. Of those without evidence of perforation, 27 had images more than 1 (range 1–5.8) year after filter placement.

Discussion

This retrospective review of our institutional experience with the Gunther Tulip filter revealed a low incidence of symptomatic complications at 2-year follow-up, extending the evidence that the risk profile for this filter is similar to that of permanent filters [6, 9, 10, 12, 25, 26]. The low fracture rate in this study suggests that performance of the various retrievable IVC filters is not homogeneous.

Despite 20 years of Gunther Tulip filter use, there is scant long-term information. The literature is largely limited to reports that focus on retrieval; and with mean dwell times of 2 months, follow-up is consequently quite limited (Table 1). There are no prospective randomized studies and the few prospective or observational studies with longer than 2-year follow-up report on cohorts of fewer than 30 patients [9, 16, 19]. This is problematic when attempting to decide whether a filter should be removed because of potential long-term complications.

In this study, the midterm morbidity and mortality risk attributable to the Gunther Tulip filter as a permanent implant was low and in line with reported risks for retrievable and permanent filters. The 2.2 % 1-year incidence of postfilter PE found in the current study lies within the 1.0–3.6 % incidence reported in the Gunther Tulip literature [719, 27]. At 2-year follow-up, the cumulative risk of PE was 4.6 %, rising to 5.2 % at 4 years. These results are consistent with reported PE rates in the filter literature, in particular the long-term studies, where the rate of increase of filter-failure-related PE diminishes after the initial 2 years [2, 46, 25, 26].

The remaining filter complications of perforation, migration, and fracture were asymptomatic in this study and did not require intervention. Despite the case reports in the literature and in the Manufacturers and User Device Experience (MAUDE) database, the available evidence would suggest a less than 1 % incidence of symptomatic fracture, migration, or perforation complications for the Gunther Tulip filter [6]. Asymptomatic perforations were common, as has been previously described [7, 24]. Durack et al. [24] reported that all of 23 Gunther Tulip filters with more than 71 days’ follow-up had some degree of caval perforation; a strut extending 3 mm beyond the expected contour of the IVC was found in 78 % of patients.

Fractures are rarely reported in the Gunther Tulip filter literature [6]. However, as discussed, the long-term follow-up, which is precisely where metal fatigue-related fracture would be expected to manifest, is quite limited. The cause of the fracture in this series was unclear and was identified during an evaluation for back pain. There was no attempt to retrieve the filter or fragments, and the patient’s pain resolved with symptomatic treatment alone. Although fractures of this filter have been reported in the MAUDE database [6], the data from this study and the limited Gunther Tulip filter follow-up data in the literature suggest that the high (3.7–8.2 % 2-year) fracture risk reported for Bard retrievable filters should not be generalized to the Gunther Tulip retrievable filter [20, 28, 29].

A common concern in the literature is a reported increased incidence of DVT when a filter is present. A relatively high DVT rate (in filter vs. nonfilter populations) may be expected, as the PREPIC and Worchester population studies have demonstrated: filters do not decrease the overall VTE rate, and so where filters successfully reduce the incidence of those presenting with PE, there will be an increase in the number of VTE that manifest as DVT. In the absence of an IVC filter, the risk of new or recurrent DVT after anticoagulation is 9.6–11.8 % at 2 years and 19–29 % at 5 years [2, 4, 5, 3033]. The 2-year risk for symptomatic DVT in this study was 15.7 %, somewhat higher than the 10–11.6 % rate reported for patients without filters, but comparable to the 17–20.8 % reported for those with filters [4, 5]. Although there does appear to be an absolute increase in DVT risk in patients who have an IVC filter, rather than a complication of the filter, it could be argued that this is a reflection of the filter functioning properly, the increase in DVT rate being roughly that of the decreased PE rate.

IVC thrombus or occlusion was noted in 4.1 % of patients, though it was symptomatic in only 0.8 % of all patients. This compares with a 2.3 % reported incidence in a review of retrievable filter literature and a 3.2 % incidence in permanent filters [6, 25]. It remains unclear whether these occlusions were due to the filter functioning appropriately by arresting emboli, or whether thrombogenicity of the filter led to local thrombosis. A recent evaluation of filter thrombi found decreasing thrombus burden over time, arguing against filter thrombogenicity as the cause [34]. The frequency of large filter thrombus in association with recurrent PE (five of six imaged) in this study suggests that thrombus extending from an occluded filter, rather than missed emboli, may be the explanation for IVC filter failure.

There were a number of limitations to this study. The retrospective nature resulted in limited data availability, and use of existing images and medical records introduced a selection bias. The availability of follow-up imaging and clinical data was generally dependent on a patient presenting for unrelated medical care. The impact of this bias toward a population that had additional or ongoing medical issues is unclear, although the selection bias may be offset by the large percentage (82 %) of the total cohort affected. In addition, the tendency for patients to get medical care from several sources led to incomplete retrieval of medical records. The high mortality rate, with half of the patient deaths occurring outside of a medical facility, resulted in limited data with regard to cause of death. Even for those who had a death record available, the rarity of an autopsy made it difficult to assess whether PE was contributory. The complication rates we report are therefore likely lower limits and my not be representative of the natural history of the Gunther Tulip filter in all patients. Although the patient sample size was limited as a result of the high mortality rate, this is not an uncommon finding in long-term studies of IVC filter populations [5, 26].

The findings of this observational single-center study suggest that the documented safety and efficacy of the Gunther Tulip filter may be extended to 2-year follow-up; this filter is associated with a low rate of symptomatic pulmonary emboli and minimal filter-related complications. The filter appears to remain intact at 2-year follow-up, consistent with a safety profile different from those devices addressed in the U.S. Food and Drug Administration alert. With regard to indications for retrieval, a Society of Interventional Radiology consensus statement recommended that “discontinuation of filtration should only occur when the risk of clinically significant PE is estimated to be less than the risk of leaving the filter in situ” [3]. The effectiveness of IVC filters in the long-term prevention of PE [4, 5], coupled with the current findings of minimal filter-related symptomatic complications at midterm follow-up, support leaving the filter in place. However, additional long-term evaluation is necessary to ensure that the Gunther Tulip IVC filter is safe as a permanent device.

Conflict of Interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer Science+Business Media New York and the Cardiovascular and Interventional Radiological Society of Europe (CIRSE) 2012