Vena Caval Filters
Introduction
Vena caval filters (VCFs) are devices prophylactically placed inside patients to
prevent deep vein thrombosis (DVT) from becoming pulmonary emboli (PE). The
number of VCFs placed in 2003 was about 100,000 1. And the industry estimated that due
to percutaneous placement, expanded indications and new “optional” filters, the number
would grow at 16% per year 1. Despite this, the quality of research on the true indications
and efficacy of VCFs is woefully lacking. This article will give an overview of VCFs,
their indications and complications.
History
The history of PE prevention is very unlike our current treatment methods and is
fraught with failed surgical attempts to stop DVT embolization. The idea of caval
interruption goes back to the 1700’s but it was not until 1893 that the first surgical
ligation of the vena cava was successfully performed to prevent PE 3. The procedure was
naturally associated with a high mortality 3. Femoral vein ligation was later popularized
in the US in the 1930’s by Homans 3. With single femoral vein ligation a patient was still
susceptible to PE from the contralateral leg, and so Ochsiner, Debakey, and O’Neil
moved the point of ligation upwards to the IVC 5. By the 1960’s, vena caval ligation
carried a mortality of 14% and PE still recurred at a rate of 6% primarily due to the
formation of collaterals 3. Harp string grids and IVC plication with sutures were both
unsuccessful attempts to overcome the failures of IVC ligation 5. In the 1950’s though,
the emphasis had switched from surgical management to medical management with the
advent of anticoagulation 3. In the 1970’s the first percutaneous device for vena caval
interruption, the Mobin-Uddin Umbrella filter came onto the market 5. It was removed
due to thrombotic complications in 1977 but by then the Greenfield filter had already
been developed and released in 1973 5. VCFs grew in popularity after the development of
the Greenfield filter and many other filters were developed to refine it and overcome its
shortcomings. In 2003, the “retrievable” filter was approved, even further expanding the
acceptability of these prophylactic devices 4.
Vena Caval Filter Design
Structural Basis
VCFs have seemly chaotic forms (see illustration), but the genius of their design
lies in their ability to capture as much thrombus as possible without compromising the
blood flow through the vena cava. The original Kimray-Greenfield filter only decreased
the cross-sectional area of the IVC by 50% when it was 2/3 filled with thrombus 5. Not
only does this allow for better venous return but also allows the fibrinolytic system
greater exposure to the emboli and therefore a greater chance to lyse them 5.
VCFs have hooks which act to anchor the filter into the vessel wall. Because of
this, VCFs have an upper limit on the size of the vessel that they can be placed into. Most
VCFs cannot be placed into sections of vessels > 28mm 5. The Bird’s Nest filter is the
only filter approved for IVC locations > 30mm 5.
Figure 1 Vena caval filters: (a) Stainless steel Greenfield filter, (b) percutaneous stainless steel Greenfield filter, (c)
titanium Greenfield filter, (d) Bird’s Nest filter, (e) Simon Nitinol filter, (f) VenaTech filter, (g) Nitinol TrapEase filter,
(h) Gu¨nther Tulip filter 5.
Types of Filters
There are three types of filters: permanent, temporary, and retrievable. Permanent
filters are placed with no intention or option of removal at a late date. These were the
standard until 2003. Temporary filters are attached to lines outside the body. They are
easily placed and easily removed. But they are potentially less effective in preventing PE
(I did not read data that directly supported this statement) and because they are attached
to outside lines they have a higher infection rate 5. Most interest has focused not so much
on temporary filters as retrievable filters 4.
Retrievable filters are VCFs that are designed to be both entirely internal and to
be removed at a later date barring endothelialization or other complications 3. Retrievable
filters have been called by many names including removable, convertible and optional 4.
Optional is a name that has stuck as some have proposed that it is harmless to leave these
filters in place if they cannot be removed although this has not been sufficiently studied
3,4
. Because these filters must be removed before they become endothelialized they can
only be left in for a finite time, usually a maximum of 3 months, depending on the
specific filter placed 3. An additional benefit is that these filters can be repositioned if
they are misplaced or found to have excessive tilt as long as they have not yet
endothelialized 3.
The different filter types follow the same basic indications for use 4.
Insertion
The standard placement of a VCF is at the L3 vertebral body, caudal to the renal
veins . This is done to prevent occlusion of the renal veins in the event of complete VCF
obstruction or thrombosis 5. The suprarenal IVC also has a shorter length and wider
diameter making it more prone to VCF complications 5.
5
Risk Factors of Venous Thromboembolism (VTE)
Before discussing the indications of VCF, it is appropriate to discuss the
epidemiology of the condition that they are used to prevent. General risk factors for PE
include prolonged immobility, recent surgery, pregnancy, estrogen therapy, cancer, and
an inherited hypercoagulable tendency 3. But the principle factor that determines a
patient’s risk for VTE is their reason for hospitalization 1. Hospitalization for major
trauma carries the highest risk for VTE 1. And trauma that includes spinal cord injury
(SCI), lower extremity fractures, pelvic fractures, or surgical repair carries an even higher
risk within the category of trauma 1. Patients who are critically ill also usually have a
moderate to high risk of VTE 1.
Also, DVTs that extend into the thigh or pelvis have a higher chance of
embolizing to become PEs than do those that do not extend beyond the calf 3.
Research
Rutherford reported the sad state of VCF research after doing a Medline search
covering the years from 1975 to 2000 6. Of the 568 studies found, 65% were retrospective
studies or case reports, 12.9% were animal or in vitro studies, and 7.4% were prospective
trials 6. Only 16 of the 568 studies involved more than 100 participants 6. Of the 568
studies, only one was a randomized trial 6. Since most data is derived from unrandomized
case series with differences in populations, outcomes, duration and follow-up there is
little chance for good meta-analysis 5. And by definition, since studies validating the
efficacy of VCFs are lacking, there is no good comparison data between VCFs and other
prophylactic methods of preventing PE 6.
Efficacy
The only randomized controlled trial of VCFs has been the PREPIC trial 3. This
was an open trial of 400 participants with documented proximal DVT or PE 3. All
participants were receiving concurrent anticoagulation 3. The patients underwent a V/Q
scan at some time between 8 to 12 days after having a VCF placed 7. At day 12 there
were significantly fewer PEs in the VCF group (1.1% versus 4.8%) 5. But the statistical
significance only holds if asymptomatic and symptomatic PE are taken together as a
group 5. If symptomatic PE are considered alone then the significance is lost (1.1% versus
2.6%) 5. At 2 year follow-up the difference in symptomatic PEs was not significant
between the two groups and the recurrence of DVT became significant in the filter group
5,7
. No survival advantage was shown in the filter group, although the study was
underpowered to detect a long-term difference in mortality due to PE because the mean
age of the study population was 73 years 5.
The next best study has been based on the California Patient Discharge Data Set 7.
This was a retrospective study of discharge data from all non-federally funded hospitals
in California 7. This data showed that the risk of readmission with one year after filter
placement was not lower than in the control group 7. Also, a 2-fold increase in the risk of
subsequent DVT was noted in patients with VCFs who had had a prior episode of PE 5.
Retrievability
There has been an increased trend towards retrievable filters due to the possibility
of increased incidence of DVT reported in the PREPIC trial 3. But the retrieval of these
filters is not always possible. Retrieval rates have been reported to range from 88 to
100% but the true rate is not known 3. The true rate is probably affected by the clinical
scenario and the experience of the interventionist 3. There is no good data on the longterm consequence of leaving retrievable filters in place 6.
VCF Indications, Use and Alternatives
Indications
The Indications for IVC filter placement are as follows: 5
A contraindication to anticoagulation and a failure to adequately anticoagulated to
prevent recurrent PE are the two strongest indications for VCF placement and yet there is
a lack of data for even these two scenarios 3. There is a paucity of evidence and much
literature controversy for all other indication 3.
Absolute indications for VCF have been defined as thromboembolic
complications associated with a contraindication to anticoagulation, documented failure
of anticoagulation to prevent PE, or complications of anticoagulation. Most patients
undergoing filter placement meet none of these criteria 4.
Indications for filter placement have also been based upon the clinical condition
of the patient. It has been suggested that filters be placed in patients who could not
withstand a PE (such as those with severely compromised cardiopulmonary status) even
though they have not yet had a PE. Again, this is an unsubstantiated use.
Matching use to Evidence
Data from the California Patient Discharge Data Set showed just how much of a
mismatch exists between the accepted indications for filter placement and who they are
being placed in. The data showed that 45% of the VCFs were placed in patients with
DVT alone, 36% were placed for PE, and 19% were placed patients without either DVT
or pulmonary embolism 4. The routine use of VCFs has been strongly discouraged
because their benefit has not been established and so has neither been compared to their
cost 1. The placement of VCFs in trauma patients increased from 3% in the period
between 1991 and 1996 to 57% between 1996 and 2001 leading to the Eastern
Association for the Surgery of Trauma (EAST) to issue guidelines to limit their use 5.
Filter placement also does not treat the patient for PE. The patient must still be
prophylactically treated and the underlying cause of increase risk for PE managed
appropriately 4. It has also not been shown that VCF themselves are not thrombogenic.
Alternatives
What are the alternatives to putting in a VCF? The first option is simply to not put
one in. There is no evidence that the devices reduce symptomatic recurrent PE or
decreases mortality, so not using them is a reasonable option. Another thing that can be
done is to confirm that the patient is getting basic mechanical prophylaxis such as IPCs,
graduated compression stockings, and venous foot pumps 1.
And a patient who develops a recurrent PE while being anticoagulated still has
not necessarily entirely failed anticoagulation. An investigation should be made to
determined whether or not the patient has been therapeutic on their regimen 5. The patient
may also been screened for hypercoagulable syndromes such as antiphospholipid
antibody syndrome which may require a higher INR 5.
Complications
There are many possible complications to VCF placement. These can generally be
thought of in two classes, short-term and long-term complications. VCF rates of
complication may be as common as massive PE, and are extremely costly 1.
Short-term complications include:
1. Recurrent PE 1
2. Inappropriate delay in instituting effective anticoagulation prophylaxis 1
3. Thrombosis at vascular access site 1
4. Filter misplacement
5. Pneumothorax 2
6. Hemorrhage 2
7. Filter misplacement 2
8. Vascular injury or perforation 2
Long-term complications include: 2
1. Recurrent PE
2. Caval occlusion
3. Filter migration
4. Excessive tilt
5. Filter and caval thrombosis
6. Strut erosion
7. Complications related to venous access procedures
Filter migration and embolism occur in 2.3% of VCF in retrospective studies 2. Hooks
placed on the ends of filter struts have successfully reduced the rates of filter migration,
but have subsequently increased IVC penetration 5.
Excessive filter tilt occurs in 0 to 56% of filter depending on the definition and type
of filter used. Excessive tilt is important because it is a risk factor of PE 2. Filter tilt can
be caused by placing a filter in the vicinity of entering veins 2. Entering veins can
promote strut instability, which in turn leads to filter tilt 2.
Caval thrombosis rates are currently substantially less than in the era of the MobinUddin model, but thrombosis remains a common event after VCF placement and has
been reported at rates of 1 to 24% 2. Filter and caval thrombosis can be viewed in two
different lights. Either it can be thought of as protective and a sign that the filter is
protecting the lung from emboli. Or it can be thought of as the VCF itself being
thrombogenic and clotting off 2. Sequelae of IVC thrombosis includes phlegmasia cerulea
dolens, recurrent DVT, post-thrombotic syndrome (PTS) and recurrent PE due to thrombi
that extend proximal to the thrombosed filter 5.
Strut erosion can be shown on a CT scan with contrast as struts penetrating outside of
the vessel 2. Strut erosion occurs at high rates and although the majority of them are
asymptomatic, it may also lead to ulceration, hemorrhage, and AV fistula 2.
Complications may occur in vascular access procedures (such as central line
placement and pulmonary wedge catheters) in patients with VCF 2. The guidewires used
to place these devices can become entrapped in a VCF 5. And attempts at forceful
removal have led to cases of filter displacement or removal 5. Use of straight wires
instead of J-tipped reduces the risk of entrapment as does limiting the insertion of
guidewires to 18 cm 5.
Future Trials
A search of clinicaltrials.gov finds two VCF studies currently recruiting
participants. The first study is called “Anticoagulation and Inferior Vena Cava Filters in
Cancer Patient with a Venous Thromboembolism” and is cosponsored by
GlaxoSmithKine. This study will compare anticoagulation with and without an IVC filter
in cancer patients with PE or DVT. Patients will be monitored for 90 days for death, PE
recurrence or complications of the VCF.
The second study is call “PREPIC 2: Prevention of Recurrent Pulmonary
Embolism by Vena Cava Interruption”. This study will enroll 400 participants with a
symptomatic PE and one risk factor. The arms of the study are anticoagulation alone
versus anticoagulation and a retrievable VCF. The follow-up will be for 90 days after
filter placement. Outcomes will be death, recurrent PE and complications of the VCF.
Any important inadequacy in these current trials is that they do not address the
most pressing question about vena caval filters, which is their use in patients who can not
be anticoagulated.
Summary
I may be wrong in my opinion, but I am adamantly against the use of VCFs. I feel
that with no data to show reduced recurrence of symptomatic PE and decreased mortality
they are unsupported by any evidence. Even if they are proven to be efficacious, they
currently have no role in evidence-based medicine. With the enormous number of VCFs
placed each year I see no reason that good randomized trials aren’t done to clarify their
basic indications and long term efficacy. Instead, the published data on VCFs is flooded
with case series supporting the reduced complication rate of one particular VCF over
another particular VCF, leaving us with little reliable data at all.
References
1. Geerts, William. Prevention of Venous Thromboembolism in High-Risk Patients.
Hematology 2006.
2. Joels, Charles. Complications of Inferior Vena Cava Filters. The American
Surgeon 2003; 69: 654-659
3. Young, T. Vena Caval Filters for the Prevention of Pulmonary Embolism
(Review). The Cochrane Library 2007; Issue 4
4. Comerota, Anthony. Retrievable IVC Filters: A Decision Matrix for Appropriate
Utilization. Perspectives in Vascular Surgery and Endovascular Therapy 2006;
18; 1: 11-17
5. Hann, Christine. The Role of Vena Caval Filters in the Management of Venous
Thromboembolism. Blood Reviews 2005; 19: 179-202
6. Rutherford, Robert. Prophylactic Indications for Vena Cava Filters: A Critical
Appraisal. Seminars in Vascular Surgery 2005; 18: 158-165
7. Streiff, Michael. Vena Caval Filters: A Review for Intensive Specialists. Journal
of Intensive Care Medicine 2003; 18: 59-79