Keywords
IVC filter - venous thromboembolism - retrievable filters
Introduction
Inferior vena cava (IVC) filters are routinely placed in patients with venous thromboembolic
disease when systemic anticoagulation fails or is contraindicated. Potential complications
of retrievable IVC filters include penetration, tilt, migration, and fracture. While
the exact clinical significance of minor penetrance and tilt is not fully understood,
filter penetration can be associated with pain, gastrointestinal and retroperitoneal
bleeding, organ involvement by penetrating struts, hydronephrosis, aortic pseudoaneurysm,
and duodenocaval fistula formation.[1] The rate of such complications may vary depending on the IVC filter manufacturer.[1]
[2]
[3]
[4] Previously published data from our group demonstrated a 10% filter penetration rate
for the Option filter, the precursor to the Option-ELITE (Argon Medical Devices Inc.).[3] The Denali filter is a redesign of the Eclipse filter (Bard Peripheral Vascular)
that demonstrated low filter complication rates in its original prospective clinical
trial; however, follow-up studies have been limited.[5] The purpose of this study was to characterize the incidence of filter-related complications
observed at follow-up computed tomography (CT) for the Denali compared with the Option-ELITE
IVC filters.
Materials and Methods
Patient Population and Inferior Vena Cava Filters
This Health Insurance Portability and Accountability Act (HIPPA)-compliant retrospective
study was approved by our institutional review board, with waiver of informed consent.
All consecutive patients who underwent IVC filter placement at our institution between
March 2014 and March 2016 were included in the study. Patient demographics are displayed
in [Table 1]. During this time frame, the Denali and Option-ELITE IVC filters were the only two
devices used in our institution, and filter choice was based on operator preference.
Table 1
Patient demographics
Variable
|
Denali
|
Option
|
Total
|
P value
|
Abbreviations: N/A, not applicable; PE, pulmonary embolism; VTE, venous thromboembolic
disease.
|
Filters
|
21
|
72
|
93
|
N/A
|
Median age, years (range)
|
60 (22–76)
|
60 (14–86)
|
60 (14–86)
|
1.00
|
Male gender (%)
|
11 (52)
|
38 (53)
|
49 (53)
|
1.00
|
Placement via internal jugular vein (%)
|
20 (95)
|
60 (83)
|
80 (86)
|
0.28
|
Placement via common femoral vein (%)
|
1 (5)
|
12 (17)
|
13 (14)
|
|
Indication for filter placement (in addition to known VTE)
|
Perioperative prophylaxis, (%)
|
12 (57)
|
36 (50)
|
48 (52)
|
0.63
|
Known/suspected bleeding risk, (%)
|
8 (38)
|
31 (43)
|
39 (42)
|
0.80
|
Failed anticoagulation, (%)
|
1 (5)
|
1 (1.5)
|
2 (2)
|
0.40
|
High embolic risk during initiation of anticoagulation, (%)
|
0
|
3 (4)
|
3 (3)
|
1.00
|
PE with hemodynamic instability, (%)
|
0
|
1 (1.5)
|
1 (1)
|
1.00
|
Filters
|
21
|
72
|
93
|
N/A
|
Median age, years (range)
|
60 (22–76)
|
60 (14–86)
|
60 (14–86)
|
1.00
|
Male gender (%)
|
11 (52)
|
38 (53)
|
49 (53)
|
1.00
|
Placement via internal jugular vein (%)
|
20 (95)
|
60 (83)
|
80 (86)
|
0.28
|
Placement via common femoral vein (%)
|
1 (5)
|
12 (17)
|
13 (14)
|
|
Indication for filter placement (in addition to known VTE)
|
Perioperative prophylaxis, (%)
|
12 (57)
|
36 (50)
|
48 (52)
|
0.63
|
Known/suspected bleeding risk, (%)
|
8 (38)
|
31 (43)
|
39 (42)
|
0.80
|
Failed anticoagulation, (%)
|
1 (5)
|
1 (1.5)
|
2 (2)
|
0.40
|
High embolic risk during initiation of anticoagulation, (%)
|
0
|
3 (4)
|
3 (3)
|
1.00
|
PE with hemodynamic instability, (%)
|
0
|
1 (1.5)
|
1 (1)
|
1.00
|
A total of 245 filters (52 Denali and 193 Option-ELITE) were placed in 239 patients
during the study period. Patients with suprarenal and multiple filters were excluded
from analysis. An additional inclusion criterion was the availability of at least
one CT scan that completely visualized all filter components.
Filter placement was performed via the internal jugular or common femoral approach
using standard technique.
Computed Tomography Scans
Computed tomography images were acquired using General Electric scanners (HiSpeed/LightSpeed,
GE Medical Systems). As most follow-up scans were obtained for reasons unrelated to
filter placement, there was variability in the protocols, including use of intravenous
contrast agent and phase of opacification of the vasculature. Both groups were wellmatched
by study type and use of contrast agent ([Table 2]). Most studies were CT scans of the abdomen or abdomen/pelvis (150/200, 75%), followed
by positron emission tomography (PET)/CT scans (12/200, 6%), CT scans of the chest
(10/200, 5%), and CT scans of the lumbar spine or CT myelograms (10/200, 5%). The
remaining study types included CT-guided abscess drainages, CT-guided biopsies, CT-guided
lumbar punctures, and CT scans obtained for radiation treatment planning. Intravenous
contrast agent was used in 66% (131/200) of studies.
Table 2
CT data
Variables
|
Denali
|
Option
|
Total
|
P value
|
Abbreviations: CT, computed tomography; LP, lumbar puncture; N/A, not applicable.
|
Filters
|
21
|
72
|
93
|
N/A
|
CT scans
|
48
|
152
|
200
|
N/A
|
CT type
|
|
|
|
|
Abscess/biopsy/LP, (%)
|
1 (2)
|
5 (3)
|
4 (2)
|
|
Abdomen pelvis, (%)
|
29 (60)
|
108 (71)
|
137 (69)
|
|
Angiography, (%)
|
1 (2)
|
5 (3)
|
6 (3)
|
|
Chest, (%)
|
2 (4)
|
7 (5)
|
9 (5)
|
|
DynaCT, (%)
|
6 (13)
|
7 (5)
|
13 (7)
|
|
Spine (including myelogram), (%)
|
2 (4)
|
8 (5)
|
10 (5)
|
|
Positron emission, (%)
|
4 (8)
|
8 (5)
|
12 (6)
|
|
Radiation treatment planning, (%)
|
3 (6)
|
4 (3)
|
7 (3)
|
|
Contrast use, (%)
|
30 (63)
|
101 (66)
|
131 (66)
|
|
Imaging Interpretation
Computed tomography images were reviewed on a picture archiving and communication
system (PACS) workstation (Agfa Healthcare) by a single fellowship-trained abdominal
imaging attending physician (AW) who was blinded to the filter types being assessed,
was unfamiliar with the design features specific to both filters, and was not involved
in the placement or removal of the filters. Images were viewed in the axial plane
for assessment of penetration, defined as filter strut or anchor measuring ≥3 mm beyond
the outer caval wall, noting the number of penetrating struts and involvement of adjacent
structures ([Fig. 1]). Coronal and sagittal reformations were used to measure tilt, defined as a filter
axis >15 degrees separated from the longitudinal caval axis. These measurements are
in accordance with the Society of Interventional Radiology (SIR) Standards of Practice
Committee definitions.[3] Images were also assessed for fracture of filter components or dislocation, defined
in our study as movement cranially beyond the confluence of the renal veins or caudally
to the level of the common iliac veins.
Fig. 1 Axial, contrast-enhanced CT demonstrates penetration of multiple filter struts of
>3 mm without evidence of injury to adjacent tissues. CT, computed tomography.
Statistical Analysis
Statistical analysis was performed with SSPS software (version 16.0; SPSS Inc.). Patient
demographics and indications for filter placement were compared between filter types
by using a Fisher's exact test for categorical variables and a Student's t-test for continuous variables. The nonparametric Mann–Whitney test was used to compare
follow-up times between filter types, as the distributions were positive-skewed. A
Fisher's exact test was used to compare rates of penetration between filter types.
To compare rates of penetration over time between the two filter types, linear regression
models were used to calculate correlation coefficients, and the Fisher r-to-z transformation was applied to the correlation coefficients to determine statistical
significance.
Results
The study population consisted of 93 patients (53% male, median age: 60 years, range:
14–86) with 21 Denali and 72 Option-ELITE infrarenal IVC filters and at least one
follow-up CT examination. A total of 200 CT examinations (48 Denali and 152 Option-ELITE)
were available for analysis. The median number of follow-up studies per filter was
two for Denali (range: 1–7) and one for Option-ELITE (range: 1–9).
Technical success rate of filter deployment was 100%, and there were no reported immediate
complications. Venous access was via the right internal jugular vein in 79% (73/93)
of cases and via the right common femoral vein in 13% (12/93). Four filters (two Denali
and two Option-ELITE) were inserted via the left internal jugular vein and one (Option-ELITE)
via the left common femoral vein. There was no statistically significant difference
in the insertion via the jugular or common femoral approach between the two devices
(p = 0.28, [Table 1]).
The median time interval between filter insertion and the latest available CT study
for all filters was 49 days (mean: 97 days; range: 0–686 days) and was not significantly
different between groups (Mann–Whitney U test, p = 0.20). Penetration was observed in 12 of 93 total filters (13%), with the Denali
filters showing a penetration rate of 14% (3/21), compared with 13% (9/72) in the
Option-ELITE group, (p = 1.00) ([Table 3]). Two of the three penetrated Denali filters showed a single strut penetrating on
follow-up imaging. The third penetrated Denali filter showed three penetrating struts
at 36 days of follow-up, progressing to four penetrating struts at 49 days of follow-up,
and remained stable at four penetrating thereafter, observed through 199 total days
of follow-up. No filters demonstrated tilt of > 15 degrees. Neither the Denali nor
the Option-ELITE filters demonstrated filter migration or fracture. The Denali filter
demonstrated no statistically significant correlation between strut penetration and
filter indwell times (r: 0.065, 95% confidence interval [CI]: –0.225 to 0.346). The Option-ELITE filter did
demonstrate a statistically significant correlation between strut penetration and
filter indwell times (r: 0.598, 95% CI: 0.155–0.443). Using Fisher's r-to-z transformation for comparison of the Denali and Option-ELITE filters, this difference
was determined to be statistically significant (z = –3.67, p < 0.01) ([Fig. 2]).
Fig. 2 Number of penetrating struts over time. There was a time-dependent increase in strut
penetration with the Option-ELITE filter, which was not seen with the Denali filter.
Table 3
Primary outcomes
Outcome
|
Denali
|
Option
|
Total
|
P value
|
Abbreviations: CT, computed tomography; IVC, inferior vena cava.
|
Filters with IVC penetration ≥3 mm by hook or strut on ≥1 study, (%)
|
3 (14)
|
9 (13)
|
12 (13)
|
1.00
|
All CT studies with IVC penetration ≥3 mm, (%)
|
9 (19)
|
20 (13)
|
29 (15)
|
0.35
|
Filters with tilt ≥15 degrees on ≥1 study, (%)
|
0 (0)
|
0 (0)
|
0 (0)
|
N/A
|
All CT studies with tilted filters ≥15 degrees, (%)
|
0 (0)
|
0 (0)
|
0 (0)
|
N/A
|
Fractured filters
|
0
|
0
|
0
|
N/A
|
Migrated filters
|
0
|
0
|
0
|
N/A
|
Discussion
This is the first trial comparing the outcomes of the Denali and Option-ELITE IVC
filters. Based on the findings, both filters are associated with a small and similar
rate of complication, in particular IVC penetration. However, there is a statistically
significant increase in strut penetration with longer indwell times with the Option-ELITE
filter when compared with the Denali filter.
The DENALI trial, a prospective, multicenter study, demonstrated no evidence of tilt
or fracture of the Denali IVC filter (0/200 and 0/184, respectively), which corresponds
to similar absence of these complications on the current study.[5] However, the DENALI trial demonstrated only a 1.5% incidence of caval penetration
at placement and retrieval, which is markedly lower than the 13% penetration rate
demonstrated in our study. This is likely due to the use of CT for evaluation of strut
penetration in our study as opposed to venography used in the DENALI trial. Our results
more closely align with those found in the large meta-analysis of IVC filter complications
by Jia et al, which demonstrated a 19% prevalence of overall IVC filter caval penetration.[1] Tsui et al described comparable outcomes for the Option filter, with low rates of
filter tilt and migration (2/323: 0.6% and 1/323: 0.3%, respectively), but higher
rates of strut penetration (57/221: 26%) compared with the current study.[6] In both trials, imaging assessment used mixed examination methods—such as plain
film, CT, and retrieval cavagram—possibly lending to measurement inaccuracy.[7]
Published data regarding complication rates of the Option-ELITE, a variation of the
Option device that was redesigned for better retrieval success in January 2014, are
limited. In addition to the study by Tsui et al, other previous studies have demonstrated
a 10% rate of strut penetration with the Option filter and no significant tilt or
filter fracture,[3] similar to the 13% rate of caval penetration and lack of additional filter complications
seen with the Option-ELITE filter in our study, suggesting that the Option-ELITE has
similar complication rates compared with its precursor. Unlike the Option filter,
however, the Option-ELITE did demonstrate a statistically significant increase in
time-dependent strut penetration.[3]
There are multiple limitations to the current study. The retrospective nature prohibited
the evaluation of complications at defined time points; as such, the wide variability
in time intervals between placement and follow-up imaging could result in lead-time
bias, where longer time intervals would be more likely to show filter complication
than studies performed within shorter time intervals. A single radiologist reviewed
all CT examinations for evidence of filter complication, preventing measurement of
interobserver variability; the impact of this is considered minimal given the relatively
objective use of measurement thresholds to define complications. There was variation
in type and coverage of the CT examinations, and many of the CTs did not include the
chest. Given the possibility of IVC filter fracture and proximal embolization, the
frequency of hardware fracture may be underestimated. The small sample size could
create a Type II statistical error where a true difference in complication frequency
between filter types is missed due to lack of sufficient power.
There are multiple options for retrievable IVC filters, and the selection of one versus
another is often solely dependent on user preference. Knowledge regarding filter complications
is helpful in further determining appropriate filter selection.[8]
[9]
[10]
[11]
[12]
[13] Although the clinical consequences of asymptomatic caval strut penetration is unknown,
it seems reasonable that progressive filter penetration could lead to symptomatic
penetration over time. The Denali and Option-ELITE IVC filters demonstrate similar
rates of strut penetration, migration, and fracture; however, increased strut penetration
over time with the Option-ELITE suggests that longer indwelling times may result in
greater complication frequency for certain IVC filters.[12] This underscores the importance of patient follow-up after filter placement to detect
complications and to remove the filters as soon as they are no longer clinically indicated.
A great deal will be learned about the outcomes of the different available filters
following completion of the Predicting the Safety and Effectiveness of Inferior Vena
Cava Filters (PRESERVE) trial.