- Email: [email protected]
Therapeutic and Prophylactic Vena Caval Interruption for Pulmonary Embolism: Caval and Venous Insertion Site Patency
Therapeutic and Prophylactic Vena Caval Interruption for Pulmonary Embolism: Caval and Venous Insertion Site Patency Aft F. AbuRahma, MD, Patrick A. Robinson, MD, James P. Boland, MD, Robert C. Cochran, MD, Yale D. Conle~, MD, Kevin R. Snodgrass, MD, Todd A. Witsberger, MD, and Daniel J. Wood, MD, Charleston, West Virginia, and Groton, Connecticut
Although anticoagulation remains the treatment of choice for acute pulmonary embolism, vena caval interruption represents an alternative for patients with contraindications and complications or in whom anticoagulation fails. The purpose of this study was to evaluate the effectiveness and safety of two types of caval interruption devices: the original stainless steel Greenfield filter and the Adams-DeWeese clip. Emphasis has been placed on maintaining caval patency with filters and clips and the patency of the femoral vein vs. the jugular vein after filter insertion. We retrospectively reviewed 161 patients who underwent caval interruption (92 filters and 69 clips) for both therapeutic and prophylactic reasons. The operative mortality and morbidity rates were 0% and 3.3% for filter patients and 8.7% and 2,9% for clip patients; no procedure-related mortalities occurred. The late caval patency rate as documented by duplex ultrasonography/venography was 100% for filter patients and 88% for clip patients (p = 0.011). Seven percent of the filter patients and 20% of the clip patients experienced late limb swelling postoperatively (p = 0.05). The incidence of recurrent late pulmonary embolism was 2.5% in the filter group and 1.9% in the clip group. In the filter group, 10% of patients experienced postoperative thrombosis at the femoral vein insertion site and 0% at the jugular vein insertion site. We found that both devices were effective in preventing pulmonary embolism, the filter provided better caval patency than the clip, and the jugular vein had a better patency than the femoral vein after filter insertion. (Ann Vasc Surg 1993;7:561-568.)
Although anticoagulation remains the treatment of choice for acute pulmonary embolism, vena caval interruption represents an alternative for individuals with either contraindications or complications or in w h o m anticoagulation fails. Greenfield ~ has demonstrated the safety and effectiveness of vena caval filtration as a method of preventing death from pulmonary embolism. From the West Virginia University Health Sciences Center/Charleston Area Medical Center, Charleston, W.Va., and Pfizer Central Research (P.A.R.), Groton, Conn. Presented at the Sixth Annual Meeting of the Eastern Vascular Society, New York, N.Y., April 30-May 3, 1992. Reprint requests: Ali F. AbuRahma, MD, 3100 MacCorkle Ave., SE, Charleston, WV 25304.
Nevertheless, some are concerned about the liberal use of vena caval interruption. Perhaps this reflects previous experience with perioperative and postoperative complications, including caval occlusion associated with caval ligation, dipping, or insertion of the Mobin-Uddin umbrella. The Greenfield filter can be inserted under local anesthesia with minimal morbidity and has been demonstrated to allow continued caval patency in almost all cases. ~'2 Most of the transvenous devices used to date have been stainless steel Greenfield filters placed surgically via the internal jugular or common femoral vein. During the past 5 years caval filter placement has been accomplished percutaneously with increasing frequency. 3 561
562
AbuRahrna et al.
The purpose of this study was to evaluate the effectiveness and safety of two types of caval interruption devices used over a 5-year period. The original stainless steel Greenfield filter and the Adams-DeWeese clip were compared with regard to caval patency and the patency of the femoral vein vs. the jugular vein after filter insertion. PATIENTS AND METHODS The medical records of all patients who underwent inferior vena caval interruption using either the Greenfield filter or the Adams-DeWeese clip from 1985 to 1990 at Charleston Area Medical Center were reviewed. The indications for caval interruption included patients with (1) pulmonary embolism or deep vein thrombosis with contraindications to anticoagulation, (2) recurrent pulmonary embolism despite adequate anticoagulation, (3) pulmonary embolism and/or deep vein thrombosis with hemorrhagic complications of anticoagulation, and (4) incidental prophylactic caval interruption in high-risk patients during major abdominal operations. The prophylactically treated group included patients with minimal cardiopulmonary reserve who could not tolerate a pulmonary embolism. Each patient in this group had at least two major risk factors for deep vein thrombosis and/or pulmonary embolism, which included a past history of deep vein thrombosis and/or pulmonary embolism, age over 65 years, a major intra-abdominal procedure with a California relative value greater than a cholecystectomy,* obesity (weight >11/2 the ideal weight), class IV cardiac disease, and severe chronic obstructive pulmonary disease with a forced expiratory volume of 1 second less than 1.5 L.5"6 No other methods of prophylaxis were used in this group (i.e., no subcutaneous heparin, external pneumatic compression device, etc.). Anticoagulants were continued postoperatively in patients who had caval interruption due to failure of anticoagulation. The presence and character of associated diseases, operative morbidity and mortality, and late morbidity and mortality were documented for each patient. Each chart was specifically reviewed for clinical, autopsy, pulmonary scan, and/or pulmonary arteriographic evidence of embolism subsequent to caval interruption. All surviving patients were contacted and personally examined by us, which included specific inquiries regarding a history of pulmonary symptoms suggestive of
Annals of Vascular Surgery
recurrent pulmonary embolism and documentation of the presence and severity of leg edema, ankle ulceration, and stasis changes of the skin. Forty-three of the patients with caval filters underwent roentgenographic examination of the abdomen to verify the position of the filter. All surviving patients (54 with clips and 81 with filters) except those lost to follow-up were assessed by duplex ultrasonography or ascending venography and cavogram to determine the patency of the inferior vena cava below the caval interruption site, patency of the venous insertion site (internal jugular vein or common femoral vein), and the venous status of the lower extremities. The cava was judged to be patent if the spontaneous phasic augmented venous signal was detected with no thrombus identified on the color duplex ultrasound. Duplex ultrasound followed by cavogram was performed in eight patients, and the remaim'ng patients had duplex ultrasound only. In our early experience the cavogram confirmed the duplex ultrasound findings. The Greenfield filter was inserted via the internal jugular or c o m m o n femoral vein according to the technique described by Greenfield et al. 7 Local anesthesia was adequate in most cases and the procedure was performed in the operating room under fluoroscopic control. The insertion site veins were repaired with 6-0 prolene vascular sutures. The position of the filter was confirmed by abdominal plain x-ray films taken after the procedure. The fight internal jugular vein was the preferred filter insertion site for therapeutic indications. If this failed, the left internal jugular vein was preferred, followed by the right common femoral vein. The right c o m m o n femoral vein was the preferred filter insertion site for patients with prophylactic caval interruption if the groin incision was a part of the original vascular procedure. The Adams-DeWeese clip was inserted via a retroperitoneal lumbar muscle splitting incision below the renal vein for therapeutic apphcations (15 patients) or transperitoneally for incidental prophylactic caval interruption during an abdominal operation (54 patients). In this retrospective study only patients clinically suspected of having recurrent pulmonary emboli underwent further evaluation, including ventilation-perfusion lung scans and/or pulmonary arteriograms. Significant postoperative limb swelling was defined in our study as swelling of a previously normal limb or worsening of preoperative leg
Vol. 7, No. 6 1993
Venal caval interruption for pulmonary embolism 563
swelling requiring the use of elastic support stockings and/or administration of diuretic agents to control the swelling. Nominative variables such as caval patency, insertion site patency, the occurrence of limb swelling or postoperative pulmonary embolus, and morbidity or mortality were compared using a chi-square test with Yates' correction or Fisher's exact test. Continuous value variables such as age or duration of follow-up were compared using Student's t test. RESULTS
Ninety-two patients underwent filter procedures and 69 patients underwent clip procedures during the period of this study. Their ages ranged from 20 to 87 years (mean age of 61.8 years for filter patients and mean age of 68.3 years for clip patients). Sixty-one percent of the filter patients and 55% of the clip patients were men. Eightyone of the filter patients were followed for longer than the 30-day postoperative period (mean late follow-up of 19.4 months, range 2 to 54 months). Fifty-four of the clip patients were followed beyond the postoperative period (mean late followup of 19.1 months).
Indications for caval interruption are listed in Table I. A total of 73 patients (45.4%) underwent the procedure for " t r e a t m e n t " of an indication that required caval interruption, and 88 patients (54.6%) had the procedure performed for prophylaxis due to underlying conditions. The most common therapeutic indications were contraindications to anticoagulation (45 patients) and recurrent pulmonary embolism while on anticoagulation (22 patients). The most c o m m o n reason for prophylaxis was an anticipated complex procedure (more complex than a cholecystectomy) (87 patients) and advanced patient age (81 patients). These procedures included abdominal aortic aneurysm resections, aortoiliofemoral bypasses, c o m m o n bile duct exploration, colon resection, Whipple procedures, and gastrectomy. Forty-two of the filters were inserted via the right internal jugular vein, four via the left internal jugular vein, and 46 via the right common femoral vein. All of the left internal jugular vein and four of the fight common femoral vein insertions followed unsuccessful attempts at insertion in the right internal jugular vein. Thirty-four of the right common femoral vein insertions were done for prophylactic reasons. Local anesthesia was used in all filter cases except the prophylactic ones,
T a b l e I. Indications for caval interruption procedures No. of patients
Clip (N = 69) Therapeutic group Contraindications to anticoagulation Deep vein thrombosis Pulmonary embolism & deep vein thrombosis Failure of anticoagulation (recurrent pulmonary embolism) Complications of anticoagulation Prophylaxis group Age ( >65 years) Obesity Past history of deep vein thrombosis Past history of pulmonary embolism Complex procedurest No. of prophylaxis indications One indication Two indications >-Three indications
15 6 2 4
(22%) (9%) (3%) (6%)
Filter (N = 92) 58 39 9 30
(63%)
(42%) (10%) (33%)
Total (N = 161) 73 (45%) 45 (28%) t 1 (7%)
34 (21%)
13 (14%)
22 (14%)
37 (69%)* 19 (35%)* 54 (100%)*
6 34 34 14 30 14 33
6 88 81 15 67 33 87
0 14 40
0 0 34
9 (13%) 0 54 (78%) 47 (87%)* 1 (2%)*
*Percentage in relation to prophylaxisindication. tProcedures that are more complex than a cholecystectomy.
(7%) (37%) (100%)* (41%)* (88%)* (41%)* (97%)*
0 14 74
(4%) (55%) (92%)* (17%)* (76%)* (38%)* (99%)*
564
Annals of Vascular Surgery
AbuRahma et al.
T a b l e II. T h i r t y - d a y operative m o r t a l i t y and m o r b i d i t y according to caval i n t e r r u p t i o n procedure a n d indications
Whole series Operative mortality (chi square = 6.06, p = 0.014) Operative morbidity due to Caval procedure (chi square = 0.108, p = 0.743)* Caval & other procedure combined f (chi square = 7.72, p = 0.0055) Prophylaxis indications Operative mortality {chi square = 1.83, p = 0.176) Operative morbidity due to Caval procedure Caval & other procedure combined (chi square = 2.89, p = 0.089) Treatment indications Operative mortality (chi square = 0.54, p = 0.463) Operative morbidity due to Caval procedure (chi square = 0.294, p = 0.588) Caval & other procedure combined (chi square = 1.09, p = 0.296)
Clip
No. of patients Filter
N = 69
N = 92
6 (9%)
0
161 6 (4%)
2 (3%)
3 (3%)
5 (3%)
17 (25%)
7 (8%)
24 (15%)
N = 54 5 (9%)
N = 34 0
N = 88 5 (6%)
0 14 (26%)
o 3 (9%)
0 17 (19%)
N = 15
N = 58
N = 73
1 (7%)
0
1 (1%)
2 (13%)
3 (5%)
5 (7%)
3 (20%)
4 (7%)
7 (10%)
Total N=
*Refers to complications attributed to caval procedure. tRefers to complications due to other operative procedure.
w h i c h w e r e d o n e u n d e r general a n e s t h e s i a because of the associated major a b d o m i n a l operations. General or epidural a n e s t h e s i a was used for clip insertion. The filter was placed infrarenally in 91 patients a n d one was positioned suprarenally. In a late follow-up check, filter location was verified in 43 patients by plain a b d o m i n a l x-ray films t h a t s h o w e d n o evidence of migration. Operative m o r t a l i t y was significantly h i g h e r for patients u n d e r g o i n g d i p p i n g (8.7%) as c o m p a r e d w i t h those u n d e r g o i n g filter p l a c e m e n t (0%) (Table II); however, n o n e of the six d e a t h s was related to the caval clipping. One p a t i e n t died of m y o c a r d i a l infarction, one of cardiac a r r h y t h m i a , o n e of sepsis, a n d three of m u l t i p l e - s y s t e m failure. Operative m o r b i d i t y for the caval p r o c e d u r e was similar for the two groups (clip p r o c e d u r e group = 2.9%; filter p r o c e d u r e group = 3.3%). There w e r e t h r e e complications in the filter group, w h i c h included n e c k h e m a t o m a , p n e u m o n i a , and filter m i g r a t i o n to the h e a r t t h a t necessitated p l a c e m e n t o f an A d a m s - D e W e e s e clip. The cava in this p a t i e n t was larger t h a n r e c o m m e n d e d ( > 2 8 m m ) a n d this was believed to be the cause
of the filter migration. The overall filter insertion success rate was 99% (91/92). There were two complications in the clip group: p n e u m o n i a in one p a t i e n t a n d postoperative lower e x t r e m i t y deep vein t h r o m b o s i s in a n o t h e r patient. The m o r b i d i t y for the c o m b i n e d caval/other procedure was 24.6% for the clip group a n d 7.6% for the filter g r o u p (Table II). Late follow-up was available for 54 clip patients a n d 81 filter patients (135/161 = 84%). Follow-up r a n g e d f r o m 2 to 54 m o n t h s ( m e a n of 19.6 m o n t h s ) . There were 14 late deaths, n o n e of w h i c h was related to the caval i n t e r r u p t i o n or r e c u r r e n t p u l m o n a r y embolism. Two patients (2/81 -- 2.5%) h a d r e c u r r e n t n o n f a t a l p u l m o n a r y emboli after insertion of the Greenfield filter, and one p a t i e n t (1/54 -- 1.9%) h a d recurrent n o n f a t a l p u l m o n a r y e m b o l i s m after caval clipping. Venous duplex e x a m i n a t i o n / c a v o g r a p h y revealed the inferior v e n a cava to be p a t e n t in 87.8% (36/41) of the clip patients a n d 100% (72/72) of the filter patients. Significant late lower limb swelling was n o t e d in 20.4% (11/54) of the clip patients in c o n t r a s t to 7.4% (6/81) o f the filter patients. The
Vol. 7, No. 6
Venal caval interruption for pulmonary embolism 565
1993
T a b l e III. Morbidity 30 days after surgery according to interruption indication for prophylaxis or treatment groups No. of patients Cfip Whole series Late caval thrombosis (chi square = 6.23,
p = 0.011) Late limb swelling (chi square = 3.84, p = 0.0501) Late pulmonary embolism (chi square = 0.13,
5/41 (12%) 11/54 (20%) 1/54 (2%)
Ffl~r 0/72 6/81 (7%) 2/81 (3%)
TotM 5/113 (4%) 17/135 (13%) 3/135 (2%)
p = 0.7207 Prophylaxis indications
Late cavaI thrombosis (chi square = 1.40, p = 0.238) Late limb swelling (chi square = 1.52, p = 0.218)
Late pulmonary embolism (chi square = 0.02, p = 0.8875) Treatment indications Late caval thrombosis (chi square = 5.63, p = 0.018) Late limb swelling (chi square = 2.29, p = 0.130) Late pulmonary embolism (chi square = 0.12, p = 0.7263)
11 clip patients who experienced late limb swelling included five with caval thrombosis associated with chronic deep vein thrombosis of the lower extremities and six with chronic deep vein thrombosis of the lower extremities with a patent inferior vena cava. One of these patients had lower limb venous ulceration. The six filter patients with late limb swelling had chronic deep vein thrombosis of the lower extremities and none was associated with caval thrombosis. One of these limbs had venous ulceration. The operative mortality of patients with prophylactic caval interruption was 0% in the filter group and 9.3% in the clip group, but none of the deaths was related to the clip insertion. The operative morbidity for the caval interruption procedure was 0% in both filter and clip patients, whereas the combined caval and other procedure morbidity was 8.8% for the filter patients and 25.9% for the clip procedures. This was comparable to patients who underwent the same procedure for therapeutic indications. The late caval thrombosis rate was 0% in the filter group and 8.8% for the clip group, which was somewhat better than the 28.6% in patients who had clips placed for therapeutic indications. Late limb swelling occurred in 5.9% of patients in the filter group in contrast to 17.8% for those in the clip group, which was also somewhat lower than in patients who had the procedure for therapeutic
3/34 (9%)
0/34
3/68 (4%)
8/45 (18%) 1/45 (2%)
2/34 (6%) 0/34
2/7 (29%)
0/38
2/45 (4%)
3/9 (33%) 0/9
4/47 (9%) 2/47 (4%)
7/56 (13%) 2/56 (4%)
10/79 (13%) 1/79 (1%)
indications. There were no cases of late recurrent pulmonary emboli in the filter patients and only one case in the clip patients (Table III and Fig. 1 ). Of the patients who were available for late follow-up, 32 had filters inserted via the internal jugular vein and 40 had filters inserted via the right c o m m o n femoral vein. The late patency rate of the internal jugular vein insertion site, as documented by duplex ultrasonography, was 100% (32/32), which was somewhat higher than the right common femoral vein insertion site, which was 90% (36/40), but the difference was not statistically significant (p = 0.1858).
DISCUSSION Pulmonary embolism is a potentially fatal disorder. In a recent prospective multicenter study Carson et al. s reported that 375 of 399 patients (94%) were treated for pulmonary embolism, usually by conventional anticoagulation. Only 10 patients (2.5%) died of pulmonary embolism; nine of them had clinically suspected recurrent pulmonary embolism. Clinically apparent pulmonary embolism recurred in 33 patients (8.3%), of w h o m 45% died during follow-up. A total of 95 patients with pulmonary embolism (23.8%) died within 1 year. The most frequent causes of death in these patients were cancer (37%), infection (22%), and cardiac disease (17%).
Annals of Vascular Surgery
566 AbuRahma et al.
[ ] FILTER
[]
CLIP
100 _ 80_
FZ 60_ LU 0 rr
LU 4 0 _ 0_
20_
'°
AI *p.,90.05
=
,
# p = not signif.
@ p = 0.0501
"%~o~o Insertion Site Patency
Fig, 1. The results of filter and clip caval procedures for prevention of pulmonary embolus.
Inferior vena cava interruption is recommended for prevention of pulmonary embolism w h e n anticoagulation therapy is either ineffective or contraindicated. The Adams-DeWeese clip and the stainless steel Greenfield filter have been used at our institution for the past 15 years. In the last 2 years we have used the percutaneous titanium Greenfield filter more often. No mortalities resulted from the operative insertion of the filter or Adams-DeWeese clip in our series. The operative morbidity was 3.3% and 2.9% for filter and clip patients, respectively, with late nonfatal pulmonary embolism of 2.5% and 2%, respectively. The late caval patency rate was 100% in filter patients, and only 7.4% experienced n e w significant late limb swelling following surgery. The late caval patency rate was 87.8% in the clip patients, and 20.4% experienced n e w significant late limb swelling. This swelling was probably partially related to caval thrombosis in some clip patients and to postphlebitic changes postoperatively in others. Of note, the majority of patients in w h o m the Adams-DeWeese clip was used were high-risk patients w h o underwent major abdominal operations. Previous studies reported a higher incidence of postoperative limb swelling and/or stasis ulceration, ~'9 perhaps because patients in these series w h o developed postphlebitic sequela manifested either significant leg swelling before filter insertion or subsequently developed deep vein thrombosis w h e n anticoagulation was withheld because of contraindications or hemorrhagic complications.
The operative mortality of 8% for partial interruption reported by Adams et al) ° and the 12.2% to 14.2% mortality in the collected series reviewed by Bernstein" stand in sharp contrast to the experience with Greenfield filter placement. Many of the operative deaths associated with a direct approach to the inferior vena cava are undoubtedly a consequence of the progression of underlying diseases or associated major operations. Nevertheless, the obvious differences in the extent of operative trauma and potential anesthetic hazards make insertion of the filter the preferred approach for prevention of pulmonary embolism w h e n caval interruption is necessary. In a review of 260 patients, Greenfield ~ documented the safety and effectiveness of caval filtration. There were no deaths related to filter placement, and only 1 of 35 postoperative deaths was possibly related to recurrent pulmonary embolism. The only significant complication of filter insertion that occurred in his series was a nonfatal air embolism. He reported a long-term cavat patency rate of 98% and a recurrent pulmonary embolism rate of 5%. Others have confirmed Greenfield's experience with caval filtration. 2'9'~2 Despite these favorable results, caval filters should be placed for appropriate indications since significant complications have been reported using the stainless steel Greenfield filter (i.e., significant proximal migration ~3 and perforation of the limb of the filter through the wall of the inferior vena cava~4). The frequency of postprocedural vein thrombo-
Vol. 7, No. 6 1993
sis in our series was 0% if the filter was inserted via the internal jugular vein and 10% if the filter was inserted via the common femoral vein. To our knowledge the patency rate in filter patients after repair of the placement vein has never been reported. The literature on vein patency following repair for other reasons documents patency in about 50% to 60% of cases if flow is not augmented by creation of an arteriovenous fistula? Four patients had common femoral vein thrombosis at the site of filter insertion in our series: two were associated with symptoms and signs of deep vein thrombosis and two were asymptomatic. The incidence of femoral vein thrombosis following percutaneous stainless steel Greenfield filter insertion in which a 24 F sheath was needed has been reported to range from 0% to 41%. 15 Pais et al. ~5 reported that the incidence of documented femoral vein thrombosis that could be related to percutaneous placement via the femoral veins was 33%; however, none of these patients had permanent venous stasis sequela. When objective measures of femoral vein thrombosis are used (i.e., duplex ultrasonography and/or venography), the incidence of femoral vein thrombosis is found to be as high as 30% to 40% in patients in w h o m percutaneous stainless steel Greenfield filters were inserted. However, it is notable that even in those studies that have objectively identified femoral vein thrombosis, the clinical manifestations of this thrombosis are evident in only 12% of patients) ~ This postprocedural femoral vein thrombosis may decrease the use of the new titanium Greenfield filter that requires a 12 F sheath. In most surgical series the percentage of filters placed prophylactically is between 10% and 2 4 % . 9'12A5 An exception is the report by Cantelmo et al. '6 in which 47 of 70 patients (67%) had filters placed for prophylaxis. However, this entire group consisted of cancer patients w h o had both a high incidence of hypercoagulability and problems with anticoagulation. In our series, 54% of all patients had caval interruption for prophylaxis (37% of filter patients and 78% of clip patients). Our criteria for prophylaxis were reported previously2~~ The high percentage of prophylactic interruptions in our series reflects our concern as well as that of others about the significant complications of anticoagulation therapy in the elderly. ~'~7"2~ Although anticoagulation is the appropriate first-line therapy for pulmonary embolism, its use is associated with a finite number of complications. Certain patient groups such as the elderly have a higher risk of significant morbidity
Venal caval interruption for pulmonary embolism
567
from anticoagulant therapy. We believe, as do others, that early caval interruption in this group of patients with a higher risk of anticoagulation complications is probably safer than long-term anticoagulation therapy. H'12'22-24 Patients with a history of venous thromboembolic disease have a higher risk of recurrent thromboembolism after major surgery and are not as well protected with low-dose anticoagulant prophylaxisY Based on the success with the Greenfield filter, we have felt justified in broadening the indication for its use in this group of patients. Our results were favorable, particularly in the filter patients with 0% mortality and caval-related morbidity rates and no late recurrent pulmonary embolism or caval thrombosis. Recently Fink and Jones 22 recommended the insertion of Greenfield filters prophylactically in association with abdominal, pelvic, or orthopedic procedures in patients at high risk for postoperative pulmonary embolism. Even with the favorable results in our series the question of whether caval interruption should be used prophylactically will probably remain controversial and can only be resolved by randomized trials. REFERENCES 1. Greenfield LH. Current indications for and results of Greenfield filter placement. J Vasc Surg 1984;1:502-504. 2. Cimochowski GE, Evans RH, Zarins CK, et al. Greenfield filter versus Mobin-Uddin umbrella. J Thorac Cardiovasc Surg 1980;79:358-365. 3. Dorfman GS. Percutaneous inferior vena caval filters. Radiology 1990;174:987-992. 4. Center for the Analysis of Health Practices Newsletter. Adapted from Hsiao WC, Stason WB. Toward developing a relative value scale for medical and surgical services. Health Care Financing Rev 1979;1:23-39. 5. AbuRahma AF, Boland JP. Selection of high-risk patients for incidental prophylactic IVC clipping. Curt Surg 1978;35:143147. 6. AbuRahma AF, Boland J, Lawton WE Jr, et al. Long term follow-up of prophylactic caval clipping. J Cardiovasc Surg 1981;22:550-554. 7. Greenfield IA, Steward JR, Crute S. Improved technique for insertion of Greenfield vena cava filter. Surg Gynecol Obstet 1983;156:217-219. 8. Carson JL, Kelley MA, Duff A, et al. The clinical course of pulmonary embolism. N Engl J Med 1992;326:1240-1245. 9. Gomez GA, Cutler BS, Wheeler HB. Transvenous interruption of the inferior vena cava. Surgery 1983;93:612-619. 10. Adams JT, Feingold BE, DeWeese JA. Comparative evaluation of ligation and partial interruption of the inferior vena cava. Arch Surg 1971;103:272-276. 11. Bemstein E. The place of venous interruption m the treatment of pulmonary thromboembolism. In Moser K, Stein M, eds. Pulmonary Thromboembolism. Chicago: Year Book, 1973, pp 312-323. 12. Goleuke PJ, Garrett WV~ Thompson JE, et al. Interruption of the vena cava by means of the Greenfield filter: Expanding the indications. Surgery 1988; 103:111-117.
568
AImRahma et aL
Annals of Vascular Surgery
13. C a s t a n e d a E Herrera M, Cragg AH, et al. Migration of a Kilnray-Greenfield filter to t h e right ventricle. Radiology 1983; 149:690-693. 14. Carabasi RA, Moritz M J, Jarrell BE. Complications encountered w i t h the u s e of the Greenfield filter. A m J Surg 1987; 154:163- t 68. 15. Pals SO, "robin KD, A u s t i n CB, et al. Percutaneous insertion of t h e Greenfield inferior vena cava filter: Experience with ninety-six patients. J Vasc Surg 1988;8:460-464. 16. Cantelmo NL, Menzoian JO, LoGerfo FW, et al. Clinical experience with vena caval filters in high-risk cancer patients. Cancer 1982;50:341-344. 17. Coon WW, Willis PW III. Hemorrhagic complications of anticoagulant therapy. Arch Intern Med 1974;133:386-392. 18. Jick J, Slone D, Borda IT, et al. Efficacy a n d toxicity of heparin in relation to age a n d sex. N Engl J Med 1968;279: 284-286.
19. Vieweg WVR, Piscatelli RL, Houser J J, et al. Complications of intravenous administration of heparin in elderly women. JAMA 1970;213:1303-1306. 20. Walker AM, Jick H. Predictors of bleeding during heparin therapy. JAMA 1980;244:1209-1212. 21. Hull RD, Raskib GE, Hirsh J. Prophylaxis of venous thromb e . e m b o l i s m - - a n overview. Chest 1986;89:374S-383S. 22. Fink JA, J o n e s BT. The Greenfield filter as the primary m e a n s of therapy in venous thromboembolic disease. Surg Gynecol Obstet 1991;172:253-256. 23. Rosenthal D, C o s s m a n D, M a t s u m o t o G, et al. Prophylactic interruption of the inferior vena c a v a - a retrospective evaluation. A m J Surg 1979;137:389-393. 24. Rao G, Zikria EA, Miller WH, et al. Concomitant prophylactic inferior vena caval clipping. Int Surg 1976;61:14% 150.
Commentary
passes, common bile duct exploration, colon resection, Whipple procedures, and gastrectomy. Although the results of vena caval interruption in this series are generally good, the data presented should not be construed to support routine prophylactic placement of vena cava filters or clips for "high-risk" patients. Careful use of other means of thromboembolic prophylaxis (e.g., lowdose anticoagulation with either heparin or warfarin, lower extremity pneumatic compression devices, and early ambulation) remains highly effective in most patients at high risk.
John W. Hallett, Jr., MD
The authors present a relatively large n u m b e r of patients who underwent caval interruption with either a filter or clip. Over 50% of these patients had vena caval interruption for prophylaxis of thromboembolism. The most c o m m o n reason for prophylaxis was an anticipated complex procedure in 87 patients and advanced age in 81 patients. These procedures included abdominal aortic aneurysm resections, aortoiliofemoral by-
Although anticoagulation remains the treatment of choice for acute pulmonary embolism, vena caval interruption represents an alternative for patients with contraindications and complications or in whom anticoagulation fails. The purpose of this study was to evaluate the effectiveness and safety of two types of caval interruption devices: the original stainless steel Greenfield filter and the Adams-DeWeese clip. Emphasis has been placed on maintaining caval patency with filters and clips and the patency of the femoral vein vs. the jugular vein after filter insertion. We retrospectively reviewed 161 patients who underwent caval interruption (92 filters and 69 clips) for both therapeutic and prophylactic reasons. The operative mortality and morbidity rates were 0% and 3.3% for filter patients and 8.7% and 2,9% for clip patients; no procedure-related mortalities occurred. The late caval patency rate as documented by duplex ultrasonography/venography was 100% for filter patients and 88% for clip patients (p = 0.011). Seven percent of the filter patients and 20% of the clip patients experienced late limb swelling postoperatively (p = 0.05). The incidence of recurrent late pulmonary embolism was 2.5% in the filter group and 1.9% in the clip group. In the filter group, 10% of patients experienced postoperative thrombosis at the femoral vein insertion site and 0% at the jugular vein insertion site. We found that both devices were effective in preventing pulmonary embolism, the filter provided better caval patency than the clip, and the jugular vein had a better patency than the femoral vein after filter insertion. (Ann Vasc Surg 1993;7:561-568.)
Although anticoagulation remains the treatment of choice for acute pulmonary embolism, vena caval interruption represents an alternative for individuals with either contraindications or complications or in w h o m anticoagulation fails. Greenfield ~ has demonstrated the safety and effectiveness of vena caval filtration as a method of preventing death from pulmonary embolism. From the West Virginia University Health Sciences Center/Charleston Area Medical Center, Charleston, W.Va., and Pfizer Central Research (P.A.R.), Groton, Conn. Presented at the Sixth Annual Meeting of the Eastern Vascular Society, New York, N.Y., April 30-May 3, 1992. Reprint requests: Ali F. AbuRahma, MD, 3100 MacCorkle Ave., SE, Charleston, WV 25304.
Nevertheless, some are concerned about the liberal use of vena caval interruption. Perhaps this reflects previous experience with perioperative and postoperative complications, including caval occlusion associated with caval ligation, dipping, or insertion of the Mobin-Uddin umbrella. The Greenfield filter can be inserted under local anesthesia with minimal morbidity and has been demonstrated to allow continued caval patency in almost all cases. ~'2 Most of the transvenous devices used to date have been stainless steel Greenfield filters placed surgically via the internal jugular or common femoral vein. During the past 5 years caval filter placement has been accomplished percutaneously with increasing frequency. 3 561
562
AbuRahrna et al.
The purpose of this study was to evaluate the effectiveness and safety of two types of caval interruption devices used over a 5-year period. The original stainless steel Greenfield filter and the Adams-DeWeese clip were compared with regard to caval patency and the patency of the femoral vein vs. the jugular vein after filter insertion. PATIENTS AND METHODS The medical records of all patients who underwent inferior vena caval interruption using either the Greenfield filter or the Adams-DeWeese clip from 1985 to 1990 at Charleston Area Medical Center were reviewed. The indications for caval interruption included patients with (1) pulmonary embolism or deep vein thrombosis with contraindications to anticoagulation, (2) recurrent pulmonary embolism despite adequate anticoagulation, (3) pulmonary embolism and/or deep vein thrombosis with hemorrhagic complications of anticoagulation, and (4) incidental prophylactic caval interruption in high-risk patients during major abdominal operations. The prophylactically treated group included patients with minimal cardiopulmonary reserve who could not tolerate a pulmonary embolism. Each patient in this group had at least two major risk factors for deep vein thrombosis and/or pulmonary embolism, which included a past history of deep vein thrombosis and/or pulmonary embolism, age over 65 years, a major intra-abdominal procedure with a California relative value greater than a cholecystectomy,* obesity (weight >11/2 the ideal weight), class IV cardiac disease, and severe chronic obstructive pulmonary disease with a forced expiratory volume of 1 second less than 1.5 L.5"6 No other methods of prophylaxis were used in this group (i.e., no subcutaneous heparin, external pneumatic compression device, etc.). Anticoagulants were continued postoperatively in patients who had caval interruption due to failure of anticoagulation. The presence and character of associated diseases, operative morbidity and mortality, and late morbidity and mortality were documented for each patient. Each chart was specifically reviewed for clinical, autopsy, pulmonary scan, and/or pulmonary arteriographic evidence of embolism subsequent to caval interruption. All surviving patients were contacted and personally examined by us, which included specific inquiries regarding a history of pulmonary symptoms suggestive of
Annals of Vascular Surgery
recurrent pulmonary embolism and documentation of the presence and severity of leg edema, ankle ulceration, and stasis changes of the skin. Forty-three of the patients with caval filters underwent roentgenographic examination of the abdomen to verify the position of the filter. All surviving patients (54 with clips and 81 with filters) except those lost to follow-up were assessed by duplex ultrasonography or ascending venography and cavogram to determine the patency of the inferior vena cava below the caval interruption site, patency of the venous insertion site (internal jugular vein or common femoral vein), and the venous status of the lower extremities. The cava was judged to be patent if the spontaneous phasic augmented venous signal was detected with no thrombus identified on the color duplex ultrasound. Duplex ultrasound followed by cavogram was performed in eight patients, and the remaim'ng patients had duplex ultrasound only. In our early experience the cavogram confirmed the duplex ultrasound findings. The Greenfield filter was inserted via the internal jugular or c o m m o n femoral vein according to the technique described by Greenfield et al. 7 Local anesthesia was adequate in most cases and the procedure was performed in the operating room under fluoroscopic control. The insertion site veins were repaired with 6-0 prolene vascular sutures. The position of the filter was confirmed by abdominal plain x-ray films taken after the procedure. The fight internal jugular vein was the preferred filter insertion site for therapeutic indications. If this failed, the left internal jugular vein was preferred, followed by the right common femoral vein. The right c o m m o n femoral vein was the preferred filter insertion site for patients with prophylactic caval interruption if the groin incision was a part of the original vascular procedure. The Adams-DeWeese clip was inserted via a retroperitoneal lumbar muscle splitting incision below the renal vein for therapeutic apphcations (15 patients) or transperitoneally for incidental prophylactic caval interruption during an abdominal operation (54 patients). In this retrospective study only patients clinically suspected of having recurrent pulmonary emboli underwent further evaluation, including ventilation-perfusion lung scans and/or pulmonary arteriograms. Significant postoperative limb swelling was defined in our study as swelling of a previously normal limb or worsening of preoperative leg
Vol. 7, No. 6 1993
Venal caval interruption for pulmonary embolism 563
swelling requiring the use of elastic support stockings and/or administration of diuretic agents to control the swelling. Nominative variables such as caval patency, insertion site patency, the occurrence of limb swelling or postoperative pulmonary embolus, and morbidity or mortality were compared using a chi-square test with Yates' correction or Fisher's exact test. Continuous value variables such as age or duration of follow-up were compared using Student's t test. RESULTS
Ninety-two patients underwent filter procedures and 69 patients underwent clip procedures during the period of this study. Their ages ranged from 20 to 87 years (mean age of 61.8 years for filter patients and mean age of 68.3 years for clip patients). Sixty-one percent of the filter patients and 55% of the clip patients were men. Eightyone of the filter patients were followed for longer than the 30-day postoperative period (mean late follow-up of 19.4 months, range 2 to 54 months). Fifty-four of the clip patients were followed beyond the postoperative period (mean late followup of 19.1 months).
Indications for caval interruption are listed in Table I. A total of 73 patients (45.4%) underwent the procedure for " t r e a t m e n t " of an indication that required caval interruption, and 88 patients (54.6%) had the procedure performed for prophylaxis due to underlying conditions. The most common therapeutic indications were contraindications to anticoagulation (45 patients) and recurrent pulmonary embolism while on anticoagulation (22 patients). The most c o m m o n reason for prophylaxis was an anticipated complex procedure (more complex than a cholecystectomy) (87 patients) and advanced patient age (81 patients). These procedures included abdominal aortic aneurysm resections, aortoiliofemoral bypasses, c o m m o n bile duct exploration, colon resection, Whipple procedures, and gastrectomy. Forty-two of the filters were inserted via the right internal jugular vein, four via the left internal jugular vein, and 46 via the right common femoral vein. All of the left internal jugular vein and four of the fight common femoral vein insertions followed unsuccessful attempts at insertion in the right internal jugular vein. Thirty-four of the right common femoral vein insertions were done for prophylactic reasons. Local anesthesia was used in all filter cases except the prophylactic ones,
T a b l e I. Indications for caval interruption procedures No. of patients
Clip (N = 69) Therapeutic group Contraindications to anticoagulation Deep vein thrombosis Pulmonary embolism & deep vein thrombosis Failure of anticoagulation (recurrent pulmonary embolism) Complications of anticoagulation Prophylaxis group Age ( >65 years) Obesity Past history of deep vein thrombosis Past history of pulmonary embolism Complex procedurest No. of prophylaxis indications One indication Two indications >-Three indications
15 6 2 4
(22%) (9%) (3%) (6%)
Filter (N = 92) 58 39 9 30
(63%)
(42%) (10%) (33%)
Total (N = 161) 73 (45%) 45 (28%) t 1 (7%)
34 (21%)
13 (14%)
22 (14%)
37 (69%)* 19 (35%)* 54 (100%)*
6 34 34 14 30 14 33
6 88 81 15 67 33 87
0 14 40
0 0 34
9 (13%) 0 54 (78%) 47 (87%)* 1 (2%)*
*Percentage in relation to prophylaxisindication. tProcedures that are more complex than a cholecystectomy.
(7%) (37%) (100%)* (41%)* (88%)* (41%)* (97%)*
0 14 74
(4%) (55%) (92%)* (17%)* (76%)* (38%)* (99%)*
564
Annals of Vascular Surgery
AbuRahma et al.
T a b l e II. T h i r t y - d a y operative m o r t a l i t y and m o r b i d i t y according to caval i n t e r r u p t i o n procedure a n d indications
Whole series Operative mortality (chi square = 6.06, p = 0.014) Operative morbidity due to Caval procedure (chi square = 0.108, p = 0.743)* Caval & other procedure combined f (chi square = 7.72, p = 0.0055) Prophylaxis indications Operative mortality {chi square = 1.83, p = 0.176) Operative morbidity due to Caval procedure Caval & other procedure combined (chi square = 2.89, p = 0.089) Treatment indications Operative mortality (chi square = 0.54, p = 0.463) Operative morbidity due to Caval procedure (chi square = 0.294, p = 0.588) Caval & other procedure combined (chi square = 1.09, p = 0.296)
Clip
No. of patients Filter
N = 69
N = 92
6 (9%)
0
161 6 (4%)
2 (3%)
3 (3%)
5 (3%)
17 (25%)
7 (8%)
24 (15%)
N = 54 5 (9%)
N = 34 0
N = 88 5 (6%)
0 14 (26%)
o 3 (9%)
0 17 (19%)
N = 15
N = 58
N = 73
1 (7%)
0
1 (1%)
2 (13%)
3 (5%)
5 (7%)
3 (20%)
4 (7%)
7 (10%)
Total N=
*Refers to complications attributed to caval procedure. tRefers to complications due to other operative procedure.
w h i c h w e r e d o n e u n d e r general a n e s t h e s i a because of the associated major a b d o m i n a l operations. General or epidural a n e s t h e s i a was used for clip insertion. The filter was placed infrarenally in 91 patients a n d one was positioned suprarenally. In a late follow-up check, filter location was verified in 43 patients by plain a b d o m i n a l x-ray films t h a t s h o w e d n o evidence of migration. Operative m o r t a l i t y was significantly h i g h e r for patients u n d e r g o i n g d i p p i n g (8.7%) as c o m p a r e d w i t h those u n d e r g o i n g filter p l a c e m e n t (0%) (Table II); however, n o n e of the six d e a t h s was related to the caval clipping. One p a t i e n t died of m y o c a r d i a l infarction, one of cardiac a r r h y t h m i a , o n e of sepsis, a n d three of m u l t i p l e - s y s t e m failure. Operative m o r b i d i t y for the caval p r o c e d u r e was similar for the two groups (clip p r o c e d u r e group = 2.9%; filter p r o c e d u r e group = 3.3%). There w e r e t h r e e complications in the filter group, w h i c h included n e c k h e m a t o m a , p n e u m o n i a , and filter m i g r a t i o n to the h e a r t t h a t necessitated p l a c e m e n t o f an A d a m s - D e W e e s e clip. The cava in this p a t i e n t was larger t h a n r e c o m m e n d e d ( > 2 8 m m ) a n d this was believed to be the cause
of the filter migration. The overall filter insertion success rate was 99% (91/92). There were two complications in the clip group: p n e u m o n i a in one p a t i e n t a n d postoperative lower e x t r e m i t y deep vein t h r o m b o s i s in a n o t h e r patient. The m o r b i d i t y for the c o m b i n e d caval/other procedure was 24.6% for the clip group a n d 7.6% for the filter g r o u p (Table II). Late follow-up was available for 54 clip patients a n d 81 filter patients (135/161 = 84%). Follow-up r a n g e d f r o m 2 to 54 m o n t h s ( m e a n of 19.6 m o n t h s ) . There were 14 late deaths, n o n e of w h i c h was related to the caval i n t e r r u p t i o n or r e c u r r e n t p u l m o n a r y embolism. Two patients (2/81 -- 2.5%) h a d r e c u r r e n t n o n f a t a l p u l m o n a r y emboli after insertion of the Greenfield filter, and one p a t i e n t (1/54 -- 1.9%) h a d recurrent n o n f a t a l p u l m o n a r y e m b o l i s m after caval clipping. Venous duplex e x a m i n a t i o n / c a v o g r a p h y revealed the inferior v e n a cava to be p a t e n t in 87.8% (36/41) of the clip patients a n d 100% (72/72) of the filter patients. Significant late lower limb swelling was n o t e d in 20.4% (11/54) of the clip patients in c o n t r a s t to 7.4% (6/81) o f the filter patients. The
Vol. 7, No. 6
Venal caval interruption for pulmonary embolism 565
1993
T a b l e III. Morbidity 30 days after surgery according to interruption indication for prophylaxis or treatment groups No. of patients Cfip Whole series Late caval thrombosis (chi square = 6.23,
p = 0.011) Late limb swelling (chi square = 3.84, p = 0.0501) Late pulmonary embolism (chi square = 0.13,
5/41 (12%) 11/54 (20%) 1/54 (2%)
Ffl~r 0/72 6/81 (7%) 2/81 (3%)
TotM 5/113 (4%) 17/135 (13%) 3/135 (2%)
p = 0.7207 Prophylaxis indications
Late cavaI thrombosis (chi square = 1.40, p = 0.238) Late limb swelling (chi square = 1.52, p = 0.218)
Late pulmonary embolism (chi square = 0.02, p = 0.8875) Treatment indications Late caval thrombosis (chi square = 5.63, p = 0.018) Late limb swelling (chi square = 2.29, p = 0.130) Late pulmonary embolism (chi square = 0.12, p = 0.7263)
11 clip patients who experienced late limb swelling included five with caval thrombosis associated with chronic deep vein thrombosis of the lower extremities and six with chronic deep vein thrombosis of the lower extremities with a patent inferior vena cava. One of these patients had lower limb venous ulceration. The six filter patients with late limb swelling had chronic deep vein thrombosis of the lower extremities and none was associated with caval thrombosis. One of these limbs had venous ulceration. The operative mortality of patients with prophylactic caval interruption was 0% in the filter group and 9.3% in the clip group, but none of the deaths was related to the clip insertion. The operative morbidity for the caval interruption procedure was 0% in both filter and clip patients, whereas the combined caval and other procedure morbidity was 8.8% for the filter patients and 25.9% for the clip procedures. This was comparable to patients who underwent the same procedure for therapeutic indications. The late caval thrombosis rate was 0% in the filter group and 8.8% for the clip group, which was somewhat better than the 28.6% in patients who had clips placed for therapeutic indications. Late limb swelling occurred in 5.9% of patients in the filter group in contrast to 17.8% for those in the clip group, which was also somewhat lower than in patients who had the procedure for therapeutic
3/34 (9%)
0/34
3/68 (4%)
8/45 (18%) 1/45 (2%)
2/34 (6%) 0/34
2/7 (29%)
0/38
2/45 (4%)
3/9 (33%) 0/9
4/47 (9%) 2/47 (4%)
7/56 (13%) 2/56 (4%)
10/79 (13%) 1/79 (1%)
indications. There were no cases of late recurrent pulmonary emboli in the filter patients and only one case in the clip patients (Table III and Fig. 1 ). Of the patients who were available for late follow-up, 32 had filters inserted via the internal jugular vein and 40 had filters inserted via the right c o m m o n femoral vein. The late patency rate of the internal jugular vein insertion site, as documented by duplex ultrasonography, was 100% (32/32), which was somewhat higher than the right common femoral vein insertion site, which was 90% (36/40), but the difference was not statistically significant (p = 0.1858).
DISCUSSION Pulmonary embolism is a potentially fatal disorder. In a recent prospective multicenter study Carson et al. s reported that 375 of 399 patients (94%) were treated for pulmonary embolism, usually by conventional anticoagulation. Only 10 patients (2.5%) died of pulmonary embolism; nine of them had clinically suspected recurrent pulmonary embolism. Clinically apparent pulmonary embolism recurred in 33 patients (8.3%), of w h o m 45% died during follow-up. A total of 95 patients with pulmonary embolism (23.8%) died within 1 year. The most frequent causes of death in these patients were cancer (37%), infection (22%), and cardiac disease (17%).
Annals of Vascular Surgery
566 AbuRahma et al.
[ ] FILTER
[]
CLIP
100 _ 80_
FZ 60_ LU 0 rr
LU 4 0 _ 0_
20_
'°
AI *p.,90.05
=
,
# p = not signif.
@ p = 0.0501
"%~o~o Insertion Site Patency
Fig, 1. The results of filter and clip caval procedures for prevention of pulmonary embolus.
Inferior vena cava interruption is recommended for prevention of pulmonary embolism w h e n anticoagulation therapy is either ineffective or contraindicated. The Adams-DeWeese clip and the stainless steel Greenfield filter have been used at our institution for the past 15 years. In the last 2 years we have used the percutaneous titanium Greenfield filter more often. No mortalities resulted from the operative insertion of the filter or Adams-DeWeese clip in our series. The operative morbidity was 3.3% and 2.9% for filter and clip patients, respectively, with late nonfatal pulmonary embolism of 2.5% and 2%, respectively. The late caval patency rate was 100% in filter patients, and only 7.4% experienced n e w significant late limb swelling following surgery. The late caval patency rate was 87.8% in the clip patients, and 20.4% experienced n e w significant late limb swelling. This swelling was probably partially related to caval thrombosis in some clip patients and to postphlebitic changes postoperatively in others. Of note, the majority of patients in w h o m the Adams-DeWeese clip was used were high-risk patients w h o underwent major abdominal operations. Previous studies reported a higher incidence of postoperative limb swelling and/or stasis ulceration, ~'9 perhaps because patients in these series w h o developed postphlebitic sequela manifested either significant leg swelling before filter insertion or subsequently developed deep vein thrombosis w h e n anticoagulation was withheld because of contraindications or hemorrhagic complications.
The operative mortality of 8% for partial interruption reported by Adams et al) ° and the 12.2% to 14.2% mortality in the collected series reviewed by Bernstein" stand in sharp contrast to the experience with Greenfield filter placement. Many of the operative deaths associated with a direct approach to the inferior vena cava are undoubtedly a consequence of the progression of underlying diseases or associated major operations. Nevertheless, the obvious differences in the extent of operative trauma and potential anesthetic hazards make insertion of the filter the preferred approach for prevention of pulmonary embolism w h e n caval interruption is necessary. In a review of 260 patients, Greenfield ~ documented the safety and effectiveness of caval filtration. There were no deaths related to filter placement, and only 1 of 35 postoperative deaths was possibly related to recurrent pulmonary embolism. The only significant complication of filter insertion that occurred in his series was a nonfatal air embolism. He reported a long-term cavat patency rate of 98% and a recurrent pulmonary embolism rate of 5%. Others have confirmed Greenfield's experience with caval filtration. 2'9'~2 Despite these favorable results, caval filters should be placed for appropriate indications since significant complications have been reported using the stainless steel Greenfield filter (i.e., significant proximal migration ~3 and perforation of the limb of the filter through the wall of the inferior vena cava~4). The frequency of postprocedural vein thrombo-
Vol. 7, No. 6 1993
sis in our series was 0% if the filter was inserted via the internal jugular vein and 10% if the filter was inserted via the common femoral vein. To our knowledge the patency rate in filter patients after repair of the placement vein has never been reported. The literature on vein patency following repair for other reasons documents patency in about 50% to 60% of cases if flow is not augmented by creation of an arteriovenous fistula? Four patients had common femoral vein thrombosis at the site of filter insertion in our series: two were associated with symptoms and signs of deep vein thrombosis and two were asymptomatic. The incidence of femoral vein thrombosis following percutaneous stainless steel Greenfield filter insertion in which a 24 F sheath was needed has been reported to range from 0% to 41%. 15 Pais et al. ~5 reported that the incidence of documented femoral vein thrombosis that could be related to percutaneous placement via the femoral veins was 33%; however, none of these patients had permanent venous stasis sequela. When objective measures of femoral vein thrombosis are used (i.e., duplex ultrasonography and/or venography), the incidence of femoral vein thrombosis is found to be as high as 30% to 40% in patients in w h o m percutaneous stainless steel Greenfield filters were inserted. However, it is notable that even in those studies that have objectively identified femoral vein thrombosis, the clinical manifestations of this thrombosis are evident in only 12% of patients) ~ This postprocedural femoral vein thrombosis may decrease the use of the new titanium Greenfield filter that requires a 12 F sheath. In most surgical series the percentage of filters placed prophylactically is between 10% and 2 4 % . 9'12A5 An exception is the report by Cantelmo et al. '6 in which 47 of 70 patients (67%) had filters placed for prophylaxis. However, this entire group consisted of cancer patients w h o had both a high incidence of hypercoagulability and problems with anticoagulation. In our series, 54% of all patients had caval interruption for prophylaxis (37% of filter patients and 78% of clip patients). Our criteria for prophylaxis were reported previously2~~ The high percentage of prophylactic interruptions in our series reflects our concern as well as that of others about the significant complications of anticoagulation therapy in the elderly. ~'~7"2~ Although anticoagulation is the appropriate first-line therapy for pulmonary embolism, its use is associated with a finite number of complications. Certain patient groups such as the elderly have a higher risk of significant morbidity
Venal caval interruption for pulmonary embolism
567
from anticoagulant therapy. We believe, as do others, that early caval interruption in this group of patients with a higher risk of anticoagulation complications is probably safer than long-term anticoagulation therapy. H'12'22-24 Patients with a history of venous thromboembolic disease have a higher risk of recurrent thromboembolism after major surgery and are not as well protected with low-dose anticoagulant prophylaxisY Based on the success with the Greenfield filter, we have felt justified in broadening the indication for its use in this group of patients. Our results were favorable, particularly in the filter patients with 0% mortality and caval-related morbidity rates and no late recurrent pulmonary embolism or caval thrombosis. Recently Fink and Jones 22 recommended the insertion of Greenfield filters prophylactically in association with abdominal, pelvic, or orthopedic procedures in patients at high risk for postoperative pulmonary embolism. Even with the favorable results in our series the question of whether caval interruption should be used prophylactically will probably remain controversial and can only be resolved by randomized trials. REFERENCES 1. Greenfield LH. Current indications for and results of Greenfield filter placement. J Vasc Surg 1984;1:502-504. 2. Cimochowski GE, Evans RH, Zarins CK, et al. Greenfield filter versus Mobin-Uddin umbrella. J Thorac Cardiovasc Surg 1980;79:358-365. 3. Dorfman GS. Percutaneous inferior vena caval filters. Radiology 1990;174:987-992. 4. Center for the Analysis of Health Practices Newsletter. Adapted from Hsiao WC, Stason WB. Toward developing a relative value scale for medical and surgical services. Health Care Financing Rev 1979;1:23-39. 5. AbuRahma AF, Boland JP. Selection of high-risk patients for incidental prophylactic IVC clipping. Curt Surg 1978;35:143147. 6. AbuRahma AF, Boland J, Lawton WE Jr, et al. Long term follow-up of prophylactic caval clipping. J Cardiovasc Surg 1981;22:550-554. 7. Greenfield IA, Steward JR, Crute S. Improved technique for insertion of Greenfield vena cava filter. Surg Gynecol Obstet 1983;156:217-219. 8. Carson JL, Kelley MA, Duff A, et al. The clinical course of pulmonary embolism. N Engl J Med 1992;326:1240-1245. 9. Gomez GA, Cutler BS, Wheeler HB. Transvenous interruption of the inferior vena cava. Surgery 1983;93:612-619. 10. Adams JT, Feingold BE, DeWeese JA. Comparative evaluation of ligation and partial interruption of the inferior vena cava. Arch Surg 1971;103:272-276. 11. Bemstein E. The place of venous interruption m the treatment of pulmonary thromboembolism. In Moser K, Stein M, eds. Pulmonary Thromboembolism. Chicago: Year Book, 1973, pp 312-323. 12. Goleuke PJ, Garrett WV~ Thompson JE, et al. Interruption of the vena cava by means of the Greenfield filter: Expanding the indications. Surgery 1988; 103:111-117.
568
AImRahma et aL
Annals of Vascular Surgery
13. C a s t a n e d a E Herrera M, Cragg AH, et al. Migration of a Kilnray-Greenfield filter to t h e right ventricle. Radiology 1983; 149:690-693. 14. Carabasi RA, Moritz M J, Jarrell BE. Complications encountered w i t h the u s e of the Greenfield filter. A m J Surg 1987; 154:163- t 68. 15. Pals SO, "robin KD, A u s t i n CB, et al. Percutaneous insertion of t h e Greenfield inferior vena cava filter: Experience with ninety-six patients. J Vasc Surg 1988;8:460-464. 16. Cantelmo NL, Menzoian JO, LoGerfo FW, et al. Clinical experience with vena caval filters in high-risk cancer patients. Cancer 1982;50:341-344. 17. Coon WW, Willis PW III. Hemorrhagic complications of anticoagulant therapy. Arch Intern Med 1974;133:386-392. 18. Jick J, Slone D, Borda IT, et al. Efficacy a n d toxicity of heparin in relation to age a n d sex. N Engl J Med 1968;279: 284-286.
19. Vieweg WVR, Piscatelli RL, Houser J J, et al. Complications of intravenous administration of heparin in elderly women. JAMA 1970;213:1303-1306. 20. Walker AM, Jick H. Predictors of bleeding during heparin therapy. JAMA 1980;244:1209-1212. 21. Hull RD, Raskib GE, Hirsh J. Prophylaxis of venous thromb e . e m b o l i s m - - a n overview. Chest 1986;89:374S-383S. 22. Fink JA, J o n e s BT. The Greenfield filter as the primary m e a n s of therapy in venous thromboembolic disease. Surg Gynecol Obstet 1991;172:253-256. 23. Rosenthal D, C o s s m a n D, M a t s u m o t o G, et al. Prophylactic interruption of the inferior vena c a v a - a retrospective evaluation. A m J Surg 1979;137:389-393. 24. Rao G, Zikria EA, Miller WH, et al. Concomitant prophylactic inferior vena caval clipping. Int Surg 1976;61:14% 150.
Commentary
passes, common bile duct exploration, colon resection, Whipple procedures, and gastrectomy. Although the results of vena caval interruption in this series are generally good, the data presented should not be construed to support routine prophylactic placement of vena cava filters or clips for "high-risk" patients. Careful use of other means of thromboembolic prophylaxis (e.g., lowdose anticoagulation with either heparin or warfarin, lower extremity pneumatic compression devices, and early ambulation) remains highly effective in most patients at high risk.
John W. Hallett, Jr., MD
The authors present a relatively large n u m b e r of patients who underwent caval interruption with either a filter or clip. Over 50% of these patients had vena caval interruption for prophylaxis of thromboembolism. The most c o m m o n reason for prophylaxis was an anticipated complex procedure in 87 patients and advanced age in 81 patients. These procedures included abdominal aortic aneurysm resections, aortoiliofemoral by-