Predictors of retrieval and long-term mortality in patients treated with inferior vena cava filters

Abstract

Objective

Inferior vena cava filters (VCFs) are a therapeutic resource for the treatment of patients with thromboembolic disease who have a contraindication to full-dose anticoagulation. In the present study, we report the retrieval rate and long-term mortality of patients receiving optional inferior VCFs and identify the predictors for retrieval and all-cause mortality during follow-up.

Methods

We conducted a retrospective cohort study of 739 consecutive recipients of optional inferior VCFs from January 2002 to December 2021 in two hospitals. Different clinical characteristics and procedure-related variables were included in the analysis. The all-cause mortality rate and retrieval rate and the predictive factors were evaluated using multivariate analysis.

Results

Of the 739 patients, 393 (53%) were women. The mean patient age was 69 ± 15 years. Of the patients, 67% presented with pulmonary thromboembolism and 43% with deep vein thrombosis (DVT). A contraindication to anticoagulation was present for nearly 90% of the patients, mainly (47%) related to the surgical procedure. In addition, 44% of the patients had active cancer. Follow-up data were available for 94% of the patients, with an average follow-up time of 6.08 ± 5.83 years. Long-term mortality was 53%. Cancer (odds ratio [OR], 3.60; 95% confidence interval [CI], 2.22-5.83), age (OR, 1.03; 95% CI, 1.08-1.42), and DVT (OR, 2.01; 95% CI, 1.08-1.42) were identified as independent predictors of mortality. The retrieval rate at follow-up was 33%. The predictors for retrieval included the indication of the filter related to a surgical procedure (OR, 4.85; 95% CI, 2.54-9.59), the absence of cancer (OR, 2.89; 95% CI, 1.45-5.75), and younger age (OR, 0.98; 95% CI, 0.97-0.99).

Conclusions

High long-term mortality was observed. The predictors of mortality were cancer, older age, and DVT. One third of the filters implanted were retrieved. The predictors for retrieval were a contraindication to surgery-related anticoagulation, the absence of cancer, and younger age.

Article Highlights.

• •Type of Research: An observational, retrospective, multicenter study
• •Key Findings: We included 739 patients with thromboembolic disease and a contraindication for anticoagulation who received an inferior vena cava filter in two community hospitals in Argentina. The long-term mortality was 53%. Filter retrieval was performed for 33% of the patients.
• •Take Home Message: Patients with thromboembolic disease treated with optional inferior vena cava filters have high long-term mortality. The predictors of mortality are cancer, older age, and deep vein thrombosis. The retrieval rate is low (on average, one third of cases). The predictors for retrieval are a contraindication to surgery-related anticoagulation, the absence of cancer, and a younger age.

Deep vein thrombosis (DVT) is the most common cause of pulmonary thromboembolism (PE).¹ The incidence of DVT is reported to be 422 per 100,000 persons, 40% of whom will develop PE if left untreated.² More than 650,000 people develop PE each year in the United States, ∼20% of whom will have fatal complications.² PE is the third leading cause of cardiovascular death (after acute myocardial infarction and stroke) and the leading cause of preventable in-hospital death.³

The treatment of DVT is anticoagulation, which prevents PE in ≤95% of cases. However, 5% of the patients will have a permanent or transitory contraindication to anticoagulation therapy. Vena cava filters (VCFs) are an alternative treatment for patients with thromboembolic disease and permanent or temporary contraindications to full-dose anticoagulation. The use of VCFs is controversial, because, to the best of our knowledge, no randomized trials supporting the use of VCFS have been reported.4, 5, 6 No agreement has been reached between different scientific societies regarding the indications for VCFs.7, 8, 9 The only use for which a consensus has been reached (ie, thromboembolic disease with a contraindication to anticoagulation) has been excluded from all trials.

In 2003, the market launch of optional or retrievable VCFs occurred to reduce the burden of complications related to VCFs, including migration, fracture, penetration into the vein, caval thrombosis, and so forth.¹⁰ However, the retrieval rates reported in the literature are low and vary considerably among regions and worldwide.¹¹

The aims of the present study are to report the retrieval rate of previously placed optional VCFs and the long-term mortality in this group of patients and to identify the predictors for successful retrieval and all-cause mortality during follow-up in two hospitals from the city of Buenos Aires.

Methods

We performed a retrospective analysis of consecutive patients receiving VCFs in two hospitals from January 2002 to December 2021. The demographics, clinical features, reasons for VCF placement, contraindication to anticoagulation, procedure-related variables, and type of device implanted were analyzed. The primary outcome was all-cause mortality during follow-up. The secondary outcome was the retrieval rate. The predictors for the primary and secondary outcomes were both analyzed. Follow-up was determined from the information available in the medical records and databases. The institutional review board of both participating centers evaluated and approved the present study.

Statistical analysis

The association analysis was performed with the χ² test or Fisher exact test when the events were fewer than five for the qualitative variables and the Student t test for the quantitative variables. P ≤ .05 was considered statistically significant. Bivariate and multivariate analyses were performed to investigate possible associations. JASP, version 0.16, was used as the statistical analysis software.

Results

A total of 739 patients were analyzed. The mean age of the study population was 69 ± 15 years, 393 patients (53%) were women, and 67% had presented with PE and 43% with DVT. Of the 739 patients, most (90%) had a contraindication to anticoagulation. Also, for 9% of patients, the VCF had been placed to treat recurrent PE despite anticoagulation therapy. All the patients had documented thromboembolic disease. The indication was not prophylactic for any of the included patients. In addition, 44% of the patients had active cancer (Table I). The most frequent contraindication to anticoagulation was the need for a surgical procedure close to the thrombotic event (47% of patients; Table II).

Table I.

Baseline characteristics

Variable	Value
Age, years	69 ± 15
Female sex	393 (53)
PE	495 (67)
DVT	324 (43)
PE and DVT	85 (11)
Recurrent PE	39 (5)
Active cancer	325 (44)
Pregnancy	9 (1.2)

DVT, Deep vein thrombosis; PE, pulmonary embolism.

Data presented as mean ± standard deviation or number (%).

Table II.

Contraindications to anticoagulation

Reason for contraindication	No. of patients (%)
Active bleeding	119 (16)
Nonactive bleeding	72 (10)
Stroke	106 (14)
Coagulopathy	35 (4.7)
Surgery-related, preoperative	171 (23)
Surgery-related, postoperative	181 (24)
Brain tumor	66 (9)

For 97% of the patients, optional or retrievable VCFs were used. Femoral access was used for 95% of the procedures, and the VCFs were deployed in an infrarenal position in 98.5% of patients. Double-basket VCFs were implanted in 63% of the patients and conical VCFs in 37% (Fig 1). Follow-up was available for 94% of the patients, with a mean follow-up period of 6.08 ± 5.83 years. All the patients without a permanent contraindication to anticoagulation received anticoagulant therapy as soon as possible.

Fig 1.

Different devices used during the study period.

The long-term mortality (primary outcome) was 53%. The independent variables associated with the primary outcome were cancer (odds ratio [OR], 3.60; 95% CI, 2.22-5.83), age (OR, 1.03; 95% CI, 1.08-1.42), and DVT (OR, 2.01; 95% CI, 1.08-1.42; Table III). Cavography was performed just before VCF retrieval.

Table III.

Primary outcome

Variable	OR	P value	95% CI
Gender	0.989	1.00	0.710-1.378
PE	0.768	.173	0.537-1.099
Active bleeding	1.321	.286	0.825-2.115
Cancer	3.601	<.01	2.224-5.831
Age	1.030	<.01	1.018-1.042
DVT	2.015	<.01	1.244-3.266

CI, Confidence interval; DVT, deep vein thrombosis; OR, odds ratio; PE, pulmonary embolism.

The retrieval rate (secondary outcome) was 33%. If only the previous 10 years are considered, the retrieval rate was 41.8%. The interval to retrieval was a mean of 26 days (range, 1-315 days). An indication for VCF placement related to a surgical procedure, either preoperatively or postoperatively, close to the thrombotic event (OR, 4.85; 95% CI, 2.54-9.59), the absence of cancer (OR, 2.89; 95% CI, 1.45-5.75), age (OR, 0.98; 95% CI, 0.97-0.99; inversely proportional) were identified as independent predictors for retrieval (Table IV). These were the only variables that were significant on univariate and multivariate analyses among the other variables analyzed, some of which are also listed in Tables III and IV.

Table IV.

Secondary outcome

Variable	OR	P value	95% CI
Gender	1.015	.932	0.715-1.439
PE	1.013	1.00	0.70-1.064
Active bleeding	0.637	.08	0.368-1.064
Surgical-related	4.852	<.01	2.454-9.591
Age	0.988	<.033	0.977-0.999
Cancer-free	2.893	<.002	1.454-5.754

CI, Confidence interval; DVT, deep vein thrombosis; OR, odds ratio; PE, pulmonary embolism.

Retrieval failure occurred in 18% of cases. The most frequent cause was thrombosis associated with the device. An advance technique for retrieval was successfully used in two patients with an OptEase VCF (Cordis). The distribution of inferior VCF placement, retrieval, and retrievable index are shown in Fig 2.

Fig 2.

Trend of inferior vena cava filter (IVCF) placement and retrieval.

Discussion

In our study, all-cause mortality during long-term follow-up was 53%. Cancer, older age, and DVT were identified as independent predictors of the primary outcome. It has been reported that long-term mortality for VCF recipients results from underlying disease, with cancer, pneumonia, and respiratory failure the most frequent, and is not associated with new PE episodes or the device itself.12, 13, 14 Our cohort had a high percentage of patients with active cancer (≤44%), with primary or secondary involvement of the central nervous system in 9%. In the randomized PREPIC (Prévention de Risque d Émbolie Pulmonarie par Interruption Cave) trial with 8 years of follow-up, the reported mortality was 48.1%, slightly lower than that reported in our study, although the percentage of patients with cancer in their cohort was also less (44% vs 16%).⁴ In a study by Rivas et al,¹³ with an average follow-up of 6 years, long-term mortality was 45.5%, with cancer the main cause of death for 49%. In another study, Everhart et al¹⁵ reported a mortality rate of 44% at 2 years of follow-up.

The PREPIC study found cancer, heart failure, respiratory failure, and age were predictors of mortality during follow-up.⁴ Their findings are similar to those in our study. Buso et al¹⁶ reported 19 years of experience with the use of VCFs and reported solid cancer, cirrhosis, anemia, and the presence of a permanent VCF as predictors of mortality at 1 year of follow-up. Some of these variables were not included in the present analysis. In a study by Chow et al,¹⁷ with 1 year of follow-up, cancer, a permanent contraindication to anticoagulation, and age were reported as independent predictors of mortality. In our report, DVT was also included among the predictors, which can be explained by its close relationship with cancer.

In relation to the secondary outcome, optional VCF retrieval was achieved in 33% of the patients, and the absence of cancer, contraindication to anticoagulation related to a surgical procedure, and younger age were identified as predictors of success. The VCF removal rates vary in the literature from 10% to 96%^18,19 and, in most series, are 30% to 50%.20, 21, 22 The findings observed in the present study are similar to those of a systematic review that included 37 studies with 6834 patients and a retrieval rate of 34%.¹¹ In the study by Rivas et al,¹³ the retrieval rate was 66%. The main reported cause of nonretrieval was the lack of planning and loss to follow-up (57% of cases), both preventable causes.¹³ In our study, in the beginning, the lack of a culture for retrieval was the main reason for nonretrieval. Another reason was the short period available for removal of the OptEase filter, which accounted for almost 60% of VCFs used in our study. A dedicated protocol was developed during the previous 10 years by a nurse, based on close surveillance, which allowed for improvement of the VCF retrieval rate to 41.8%.

In accordance with these results, in terms of the predictors for retrieval, Chong et al²³ reported age <70 years, an absence of cancer history, prophylactic insertion, and a contraindication to anticoagulation related to a postoperative procedure as independent predictors of retrieval. Brown et al,¹⁸ in a large cohort of 55,000 patients from the United States, identified young age, residence in an urban area, and the absence of comorbidities related to prothrombotic disease (including cancer) as independent predictors of retrieval.

The study by Buso et al¹⁶ found the presence of solid cancer and age as predictors of nonretrieval and the contraindication to anticoagulation related to a surgical procedure as a predictor of retrieval. These findings are very similar to those obtained in our study. Finally, Inonescu et al²⁰ reported the possibility of restarting anticoagulation at any time as a predictor of retrieval, with advanced age and the presence of cancer as predictors of nonretrieval.

Study limitations

Several limitations of this study should be acknowledged. The biases inherent to the retrospective and observational design of our study cannot be ruled out. Also, because our study was performed at two community hospitals with similar characteristics and by the same work group at both centers, extrapolation of the results to the general population should be interpreted with caution. During the long-term follow-up period, a change occurred in the paradigm of anticoagulation therapy, with the advent of new oral anticoagulant agents. These pharmacologic agents could change the indications for optional VCF placement and retrieval.

Conclusions

In our group of patients with thromboembolic disease and contraindications to anticoagulation who underwent placement of optional, retrievable VCFs, the long-term mortality was high. The predictors of all-cause mortality were the presence of cancer, older age, and DVT. One third of the VCFs were retrieved. The identified predictors of retrieval were a contraindication to anticoagulation related to a surgical procedure, the absence of cancer, and younger age.

Author Contributions

Conception and design: GM, JN, OT, BD, JA, GL, JÁ

Analysis and interpretation: GM, OT, JA, JÁ

Data collection: GM, JN, JA

Writing the article: GM, JN, OT, JÁ

Critical revision of the article: GM, OT, BD, JA, GL, JÁ

Final approval of the article: GM, JN, OT, BD, JA, GL, JÁ

Statistical analysis: GM, OT, JA, JÁ

Obtained funding: Not applicable

Overall responsibility: GM