Fax: (713) 792-4098
Use of inferior vena caval filters and survival in patients with malignancy
Version of Record online: 9 SEP 2004
Copyright © 2004 American Cancer Society
Volume 101, Issue 8, pages 1902–1907, 15 October 2004
How to Cite
Wallace, M. J., Jean, J. L., Gupta, S., Eapen, G. A., Johnson, M. M., Ahrar, K., Madoff, D. C., Morello, F. A., Murthy, R. and Hicks, M. E. (2004), Use of inferior vena caval filters and survival in patients with malignancy. Cancer, 101: 1902–1907. doi: 10.1002/cncr.20578
- Issue online: 1 OCT 2004
- Version of Record online: 9 SEP 2004
- Manuscript Accepted: 13 JUL 2004
- Manuscript Revised: 2 JUL 2004
- Manuscript Received: 5 MAY 2004
- inferior vena cava filters;
- pulmonary emboli;
- venous thromboembolic disease;
- patient survival
Inferior vena cava (IVC) filters have proven to be a viable alternative to anticoagulation therapy for the prevention of life-threatening pulmonary emboli (PE) for patients who have contraindications to anticoagulation therapy. The clinical benefit of placing IVC filters in patients with advanced-stage cancer is controversial. The current study reported the authors' experience with IVC filters in patients with cancer.
Between January 2000 and May 2003, IVC filters were placed in 308 patients with venous thromboembolic (VTE) disease. Of these patients, 267 had solid tumors and 41 had liquid tumors. Outcome was reviewed retrospectively with regards to patient survival as well as procedural and filter-related complications. Patients with solid and liquid tumors were used to generate Kaplan–Meier estimates for survival and the probability of surviving 30, 90, and 365 days was also calculated. The prognostic effect of age, primary malignancy, gender, extent of disease, indication, admission to the intensive care unit, and IVC thrombus on overall survival was also evaluated using univariate and multivariate Cox models for patients with solid tumors.
The median survival periods for patients with solid and liquid tumors were 145 days and 207 days, respectively. The probability of survival at 30, 90, and 365 days was 0.81, 0.60, and 0.35, respectively, for patients with solid tumors and 0.85, 0.67, and 0.48, respectively, for patients with liquid tumors. There was no statistically significant difference in survival based on primary malignancy for solid tumors (P = 0.628) or between solid and liquid tumors (P = 0.16). For patients with solid tumors, a statistically significant difference in survival was found by extent of disease (P = 0.002). Patients with solid tumors classified as local disease (n = 15), locally advanced disease (n = 95), and widely metastatic or disseminated disease (n = 153) had a probability of survival at 30 days of 0.93, 0.87, and 0.76, respectively. Compared with patients with local disease, patients with metastatic or disseminated disease were 3.7 times more likely to die (P = 0.013). Patients with a history of deep venous thrombosis (DVT) and hemorrhage were 2 times more likely to die than patients with DVT and no history of hemorrhage (P = 0.0057). Documented complications occurred in 22 of 308 (7.1%) patients and included PE (n = 4), new caval thrombosis (n = 14), retroperitoneal hemorrhage (n = 2), and maldeployed filters (n = 2).
IVC filters were shown to be safe and highly effective in preventing PE-related deaths in patients with cancer with VTE disease. Patients with a history of DVT and bleeding or metastatic/disseminated stage of disease had the lowest survival after IVC filter placement. Cancer 2004. © 2004 American Cancer Society.
The association between venous thromboembolism (VTE) and malignancy was first described by Trousseau in 1865.1 Patients with cancer have a higher risk of developing VTE and having a recurrent thromboembolic event than patients with nonmalignant diseases.2–4 In addition, patients with VTE have an increased risk of being diagnosed with a cancer in the first 6 months after the initial thromboembolic event.5
Survival is also negatively impacted when VTE and malignant disease coexist. The probability of death within 6 months after the initial thromboembolic event can be twice as high compared with patients with malignancy in the absence of VTE.2
Over the past two decades, the placement of inferior vena cava (IVC) filters has increased dramatically.6 Filters have proven to be a viable alternative to anticoagulation therapy for the prevention of life-threatening pulmonary emboli (PE) for patients who have contraindications to anticoagulation therapy. The clinical benefit of placing IVC filters in patients with advanced-stage cancer was challenged in an article by Jarrett et al.7 The goal of our study is to review our experience with IVC filters in patients with cancer with regards to survival as a measure of potential clinical benefit.
MATERIALS AND METHODS
Between January 2000 and May 2003, IVC filters were placed in 308 consecutive patients (190 men [61.7%] and 118 women [38.3]) with thromboembolic disease at The University of Texas M. D. Anderson Cancer Center (MDACC; Houston, TX). Of these patients, 267 had solid tumors and 41 had liquid tumors (lymphoma or hematopoietic malignancies) at the time of filter placement. All filter placements were performed at MDACC. We retrospectively reviewed the hospital records and recorded the appropriate data associated with IVC filter placement and its follow-up assessment. Additional survival data were obtained from the U.S. Social Security Death Index.8 Our study was approved by the MDACC institutional review board, and a waiver was granted for informed consent.
Outcome was reviewed retrospectively with regards to patient survival as well as procedural and filter-related complications. Using Kaplan–Meier methodology, median survival and the corresponding 95% confidence interval (95% CI) values for survival were calculated for patients with solid and liquid tumors. The probability of surviving 30, 90, and 365 days as well as the corresponding 95% CIs were also computed. Log-rank tests were used to compare survival between groups of patients. Of the 267 patients with solid tumors, 4 patients did not have adequate follow-up and, thus, survival assessment was based on a cohort of 263 patients. This group included 167 men (63.5%) and 96 women (36.5%). The median age of the patients was 60 years (range, 24–81 years). Of these 263 patients, 255 had 1 primary malignancy and 8 patients had 2 primary malignancies. One patient had a history of both a solid and liquid tumor at the time of filter placement (Table 1).
|Survival group||No.||Probability of survival|
|30 Days (95% CI)||90 Days (95% CI)||365 Days (95% CI)|
|Overall||263||0.81 (0.76–0.86)||0.6 (0.54–0.67)||0.35 (0.29–0.43)|
|Breast||15||0.8 (0.62–1.00)||0.5 (0.29–0.86)||0.4 (0.20–0.81)|
|CNS||49||0.87 (0.78–0.97)||0.61 (0.48–0.78)||0.33 (0.20–0.54)|
|GI||53||0.79 (0.68–0.91)||0.54 (0.42–0.71)||0.29 (0.18–0.48)|
|GU||49||0.86 (0.76–0.96)||0.79 (0.68–0.91)||0.47 (0.32–0.68)|
|GYN||22||0.82 (0.67–1.00)||0.6 (0.42–0.87)||0.36 (0.18–0.72)|
|Lung||32||0.75 (0.61–0.92)||0.55 (0.39–0.76)||0.29 (0.16–0.55)|
|Sarcoma||19||0.79 (0.63–1.00)||0.47 (0.30–0.76)||0.38 (0.20–0.72)|
|Skin||21||0.66 (0.48–0.90)||0.54 (0.36–0.82)||0.28 (0.12–0.67)|
|Extent of disease|
|Local||15||0.93 (0.80–1.00)||0.84 (0.65–1.00)||0.63 (0.39–1.00)|
|Locally advanced||95||0.87 (0.81–0.94)||0.64 (0.55–0.75)||0.4 (0.30–0.53)|
|Metastatic/dissemin||153||0.76 (0.69–0.83)||0.55 (0.48–0.64)||0.29 (0.22–0.40)|
|DVT w/o hemorrhage||112||0.88 (0.83–0.95)||0.67 (0.58–0.77)||0.38 (0.29–0.50)|
|DVT and hemorrhage||39||0.67 (0.54–0.84)||0.4 (0.27–0.61)||0.18 (0.07–0.43)|
|PE w/o hemorrhage||95||0.78 (0.70–0.87)||0.6 (0.51–0.71)||0.37 (0.27–0.50)|
|PE and hemorrhage||17||0.77 (0.59–1.00)||0.56 (0.36–0.87)||0.42 (0.21–0.86)|
|No||236||0.81 (0.76–0.86)||0.59 (0.53–0.66)||0.34 (0.28–0.42)|
|Yes||27||0.82 (0.68–0.98)||0.66 (0.50–0.87)||0.47 (0.29–0.75)|
|No||248||0.8 (0.76–0.86)||0.59 (0.53–0.65)||0.34 (0.28–0.42)|
|Yes||15||0.86 (0.69–1.00)||0.86 (0.69–1.00)||0.48 (0.23–1.00)|
Overall survival was computed as the number of days from the procedure date to the date of death for the patients who died or the date of last contact for patients still alive. For each patient who had two malignancies, data on the primary malignancy currently under therapy were utilized for the survival analyses.
Using Cox proportional hazard regression modeling, the prognostic effect of age, primary malignancy, gender, extent of disease, indication, admission to the intensive care unit (ICU), and IVC thrombus on overall survival was evaluated for patients with solid tumors. The extent of disease was defined as local disease, locally advanced disease, or disseminated/metastatic disease. Indications for filter placement or contraindication for anticoagulation therapy utilized for prognostic assessment were divided into four groups: deep venous thrombosis (DVT) with a history of hemorrhage proximate to the time of filter placement; DVT with nonhemorrhage-related indication; PE; and PE with history of hemorrhage proximate to the time of filter placement. Of the 308 patients, 68 had a history of bleeding as a primary indication for IVC filter placement. In the absence of hemorrhage, other primary indications for filter placement included central nervous system malignancy (n = 75), proximate surgical intervention (n = 71), anticoagulation failure (n = 28), thrombocytopenia (n = 20), limited cardiopulmonary reserve (n = 15), IVC thrombus (n = 9), substantial PE burden (n = 8), peptic ulcer disease (n = 1), and syncope (n = 1). No specified contraindication to anticoagulation therapy was identified through chart review in an additional 12 patients. The diagnosis of PE and DVT was documented by imaging findings on computed tomography (CT) scan, ultrasound, or ventilation/perfusion scan. ICU admissions included patients residing in the ICU immediately before filter placement and the presence or absence of IVC thrombus as documented by cavography during filter placement. For the continuous variable age, Martingale residual plots were used to assess the appropriate functional form as well as goodness of fit. Univariate Cox models were fit to evaluate the predictive effect of each factor alone. Furthermore, a multivariable Cox model including all variables assessed in the univariate analysis was investigated.
The overall median survival for patients with solid tumors was 145 days (95% CI, 112–183 days) with a total of 155 of 263 deaths during the follow-up period. The overall medial survival for patients with liquid tumors was 207 days (95% CI, 173, not attained). The probability of overall survival at 30, 90, and 365 days was 0.81 (95% CI, 0.76–0.86), 0.60 (95% CI, 0.54–0.67), and 0.35 (95% CI, 0.29–0.43), respectively, for patients with solid tumors and 0.85 (95% CI, 0.75–0.97), 0.67 (95% CI, 0.54–0.84), and 0.48 (95% CI, 0.33–0.69), respectively, for patients with liquid tumors (Table 1).
There was no statistically significant difference in survival based on primary malignancy for solid tumors (P = 0.628) or between solid and liquid tumors (P = 0.16). For patients with solid tumors, a statistically significant difference in survival was found for extent of disease (P = 0.002). Patients with solid tumors classified as local disease (n = 15), locally advanced disease (n = 95), and widely metastatic or disseminated disease (n = 153) had a probability of survival at 30 days of 0.93 (95% CI, 0.80–1.00), 0.87 (95% CI, 0.81–0.94), and 0.76 (95% CI, 0.69–0.83), respectively (Table 1).
In the univariable models, extent of disease and indication were significantly predictive of survival, whereas age was marginally predictive. When all factors were evaluated simultaneously in a multivariable model, age, extent of disease, and indication remained predictive of survival. Specifically, compared with patients with local disease, patients with metastatic or disseminated disease were 3.7 times more likely to die (95% CI, 1.32–10.46; P = 0.013; Table 2). Patients with a history of DVT and hemorrhage were two times more likely to die than patients with DVT and no history of hemorrhage (95% CI, 1.23–3.38; P = 0.0057; Table 2).
|RR (95% CI)||P value||RR (95% CI)||P value|
|Age||260||0.99 (0.98–1.00)||0.053||0.99 (0.97–1.00)||0.029|
|Male||165||1.19 (0.85–1.68)||0.31||1.26 (0.82–1.94)||0.30|
|CNS||49||1.01 (0.47–2.15)||0.98||0.92 (0.36–2.38)||0.87|
|GI||53||1.22 (0.58–2.55)||0.60||0.85 (0.37–1.96)||0.70|
|GU||49||0.81 (0.37–1.77)||0.60||0.51 (0.21–1.24)||0.14|
|GYN||22||1.01 (0.42–2.44)||0.99||1.17 (0.46–2.99)||0.75|
|Lung||32||1.46 (0.67–3.18)||0.34||0.94 (0.39–2.28)||0.89|
|Sarcoma||19||1.09 (0.46–2.55)||0.84||1.14 (0.44–2.95)||0.79|
|Skin||21||1.35 (0.57–3.22)||0.50||0.97 (0.38–2.50)||0.96|
|Extent of disease|
|Locally advanced||94||2.32 (0.84–6.41)||0.10||1.72 (0.58–5.06)||0.33|
|Metastatic/disseminated||151||3.59 (1.32–9.81)||0.013||3.71 (1.32–10.46)||0.013|
|DVT w/o hemorrhage||111||1.00||1.00|
|DVT and hemorrhage||39||1.76 (1.11–2.78)||0.016||2.04 (1.23–3.38)||0.0057|
|PE w/o hemorrhage||95||1.13 (0.79–1.62)||0.50||1.05 (0.71–1.54)||0.82|
|PE and hemorrhage||15||1.29 (0.62–2.70)||0.50||1.43 (0.64–3.17)||0.38|
|Yes||27||0.88 (0.50–1.56)||0.66||0.69 (0.38–1.28)||0.24|
|Yes||15||0.73 (0.34–1.55)||0.41||0.59 (0.26–1.33)||0.20|
A total of 311 filters were placed in 308 patients with cancer. Table 3 lists the type and number of permanent filters implanted. The reasons for replacement filters in three patients included duplication of the IVC, incomplete filter opening, and deployment of an inverted filter.
|Filter name||No. of implantsa||Manufacturer|
|TrapEase||130||Cordis Europa, Roden, The Netherlands|
|Vena Tech-LP||63||B, Braun Medical Inc., Bethlehem, PA|
|LGM Vena Tech (original)||15||B, Braun Medical Inc., Bethlehem, PA|
|Simon Nitinol||47||Bard, Tempe, AZ|
|Stainless Steel Greenfield||41||Boston Scientific, Natick, MA|
|Bird's Nest||14||Cook Inc., Bloomington, IN|
|Gunther Tulip||1||Cook Inc., Bloomington, IN|
Documented complications occurred in 22 of 308 (7.1%) patients and included new caval thrombosis (n = 14), PE (n = 4), retroperitoneal hemorrhage (n = 2), and maldeployed filter (n = 2). Follow-up imaging data were available for more than one-half (157 of 308) of the patients with a median of 73 days to imaging (range, 1–1148 days). Imaging consisted of contrast-enhanced CT scans of the abdomen (n = 124) and chest (n = 30), magnetic resonance imaging scans of the abdomen (n = 7), and ventilation/perfusion scans (n = 1). Of 308 patients, 14 evaluable patients (4.5%) presented with new caval thrombus. Thrombus was occlusive in nine patients, seven of whom were symptomatic. Five of the seven patients experienced substantial lower extremity pitting edema, one developed phlegmasia cerulea dolens, and another developed acute renal failure and respiratory failure. Thrombus was nonocclusive in five patients in whom thrombus was identified incidentally on follow-up imaging. Caval thrombus was identified in patients who received the TrapEase filter (10 of 130 [7.7%]), the Vena Tech-LP filter (2 of 61 −[3.3%]), the Stainless Steel Greenfield filter (1 of 39[2.5%]), and the Simon Nitinol filter (1 of 47 [2.1%]). TrapEase filters were present in six of seven patients with symptomatic caval occlusion. Symptomatic retroperitoneal hemorrhage that did not require therapy occurred in two patients in whom the TrapEase filter was inserted.
Premortem PE was documented in 4 (1.3%) patients. Based on clinical suspicion, PE was responsible for four deaths among 308 patients and was considered to be a potential contributing factor in the death of an additional 10 of 308 patients who had other concomitant respiratory disease or multiple compounding disorders at the time of death. Respiratory compromise without suspicion of PE was considered the primary cause of death in 6 patients and nonrespiratory causes of death were considered in 18 patients. The cause of death was secondary to progression of disease in 66 patients and was unknown in 69 patients.
Despite the increasing utility of IVC filters over the past 30 years,6 controversy regarding the appropriateness of filter placement continues for patients with advanced-stage cancer.7, 9–11 Jarrett et al.7 recently reported 30-day, 90-day, and 365-day overall survival rates of 68.8%, 49.9%, and 26.8% in a cohort of 116 patients. Based on the dismal 6-week (48%) and 1-year (13.7%) survival rates of patients with Stage IV disease, Jarrett et al. concluded that IVC filter placement may be of little clinical benefit and a poor utilization of resources. In the current study, the overall survival rates were higher than those reported by Jarrett et al., i.e., our 30-day, 90-day, and 365-day overall survival rates were 81%, 60%, and 35%. This was, in part, due to a higher absolute number, but lower percentage, of patients with advanced disease (58.2% vs. 78.4%). However, in our series, patients with metastatic or disseminated solid tumors (n = 153 of 263 [58%]) demonstrated substantially better survival at 30 days (76%), 90 days (55%), and 365 days (29%) compared with 6-week (48%) and 1-year (13.7%) survival rates for comparable patients (n = 91 [78.4%]) in the series by Jarrett et al. Our higher overall survival rate was supported by a large single-center, 26-year retrospective review by Athanasoulis et al.6 They reported the placement of IVC filters in 1739 patients, of whom 926 patients had an underlying primary malignancy. Their overall 30-day survival rate was 83% (1436 of 1731). Patients with a malignancy only had a 5% lower 30-day survival rate (737 of 915 [80.5%]) compared with those without a malignancy (699 of 816 [85.7%]). These authors did not stratify patients with cancer according to stage and did not assess the statistical significance in survival between patients with and without cancer.
Jarrett et al. reported a statistically significant difference in survival (P < 0.001) when patients with Stage IV disease were compared with patients with Stage 1, 2, and 3 disease. Survival comparison among patients with Stage1–3 disease was not, however, reported. We also found a statistically significant reduction in survival based on extent of disease at the time of filter insertion (P = 0.002).
The cost-effectiveness of caval filter placement for the prevention of PE in 24 patients with brain tumors was analyzed by Chau et al.12 Based on the few patients in their series, they concluded that filter placement was effective in reducing the rate of PE but was not cost-effective for patients with brain tumors. They also suggested that IVC filter placement may be cost-effective for patients with longer life expectancies. Aside from the few patients in their series, one major limitation was the lack of comparison between patients undergoing filter placement and those receiving no therapy. IVC filter placement is currently used when there is a contraindication to, or a failure of, anticoagulation therapy. The standard of therapy is not to deny a medically proven therapy based on cost when the risks of the planned intervention are low. This is especially true when there is a contraindication to first-line therapy. This is supported by the results of an early randomized controlled prospective study reported by Barritt and Jordan.13 In that study, 26% (5 of 19) of untreated patients with a clinical diagnosis of PE died of PE during a follow-up period of approximately 2 weeks, and another 26% of patients experienced nonfatal disease recurrences. The difference in PE mortality was statistically significant when compared with the PE-related deaths among patients (54 patients [0%]) treated with anticoagulation therapy (P = 0.0007).
IVC thrombosis is a well documented complication of caval filter placement, ranging from 1.5% to approximately 11% depending on the patient population and the type of filter utilized.6, 11, 14–17 Clinical follow-up data are limited on three newer filter designs (TrapEase, Vena Tech-LP, and Gunther Tulip) recently made commercially available in the United States. Our caval thrombosis rate of 4.5% is well within the accepted range for all filter types and is consistent with a caval thrombosis rate of 4.4% (40 of 915) among patients with cancer versus 1.8% (15 of 816) among patients without cancer who received filters in the series by Athanasoulis et al.6 Given that patients with malignancy are at a higher risk for recurrent VTE events compared with patients without malignancy,2–4 higher rates of caval thrombosis are expected. We encountered a higher caval occlusion rate of 7.7% for the TrapEase filter compared with approximately 3% for the remaining filter types in this series. In addition, the TrapEase filter was used in six of the seven patients who developed complete symptomatic caval obstruction. One patient subsequently developed phlegmasia cerulea dolens and a second developed acute renal failure from IVC and renal vein thrombosis. The reported range of caval thrombosis for the TrapEase filter is limited to date, but is reported to be ≤ 3%.16, 17 In addition, two patients experienced back pain and had evidence of a retroperitoneal hematoma after placement of the TrapEase filter. Although the majority of complications were recorded for the TrapEase filter, no statistically significant difference in complication rates was found between the filters used in the current study.
One limitation of our investigation primarily revolves around the retrospective and noncontrolled nature of the study when attempting to identify accurate rates of filter complications. Systematic follow-up and prospective data collection after filter placement would be ideal but not always attainable. In our population of patients with cancer, follow-up imaging is more likely to occur than in populations of patients without cancer because routine imaging follow-up is one of the most utilized methods of determining the effects of cancer therapy or documenting the progression of disease.
In conclusion, IVC filters are highly effective in preventing PE-related deaths in patients with VTE. Filters can be placed in patients with VTE and cancer without higher rates of procedure or filter-related complications compared with historical controls. However, patients with a history of DVT and bleeding or metastatic/disseminated stage of disease have the lowest survival.
- 1Phlegmasia alba dolens. Clin Med Hotel Dieu de Paris 1865; 3: 94–96..
- 8U.S. Social Security Death Index Web Site. In: Available from URL: http://www.ancestry.com/ssdi/advanced.htm [accessed 9 January 2004].