Reporting the impact of inferior vena cava perforation by filters

Introduction

The use of inferior vena cava (IVC) filters has increased significantly in the last decade. This is due to expansion of the relative indications for filter placement such as prophylaxis in patients sustaining severe trauma or those with limited cardio-pulmonary reserve having deep vein thrombosis (DVT) requiring anticoagulation. ¹ Another reason is the introduction of retrievable IVC filters that made the use of such devices more feasible. ²

Reports of IVC filter placement suggest that the procedure is safe and effective in the vast majority of patients, but the guidelines for follow-up and surveillance still need further improvement. Short- and long-term complications related to IVC filters include migration or fracture of filter struts, IVC thrombosis, and perforation of the venous wall causing bleeding or penetration into surrounding structures; however, the incidence and impact of those complications are not widely studied in the general population.

Recently, a systematic review of symptomatic duodenal perforation by IVC filters found only 22 case reports suggesting that many other similar cases of IVC wall perforation in asymptomatic patients or those with less severe symptoms may pass unnoticed, contributing to a large number of complications that are not reported. ³ In an attempt to capture adverse events related to devices, the Food and Drug Administration (FDA) has launched a public website (MAUDE – MA nufacture and User Facility Device Experience) where manufacturers, institutions and healthcare providers can submit their reports.⁴ The purpose of our study was to assess the incidence of filter perforation through the IVC and surrounding structures and to evaluate the clinical outcomes.

Methods

We identified 3311 reported adverse events using ‘Filter, Intravascular, Cardiovascular’ product class search criteria from 1 January 2000 to 31 December 2011 in the MAUDE database. Outcomes of interest included incidence of IVC perforation, type of device utilized, clinical presentation, and IVC wall perforation management (including retrievability rates) throughout the studied years whenever available. We excluded adverse event descriptions that did not specifically involve IVC perforation such as device malfunction, failure of the device to deploy properly, and migration/tilting. Iatrogenic perforations or malfunction of the device noticed immediately following the deployment of the filter were also excluded.

A variety of devices are approved by the FDA for interruption of the IVC in the United States. Currently, the most common placed filters are retrievable which, by definition, allow the operators to remove them when the patient is no longer at risk of pulmonary embolism (PE). Seven permanent (Stainless and Titanium Greenfield®, Vena Tech®-LGM and LP, Simon Nitinol®, Bird’s Nest®, TrapEase® filter) and six retrievable filters (Gunther Tulip®, Bard G2/Eclipse®, OptEase®, ALN®, Option® and Celect® filter) are commercially available in the United States.

Briefly, the MAUDE dataset was launched by the FDA in June 1991 to gather reports of adverse events involving medical devices. All data collected in the website originates from user facility reports since 1991, voluntary and distributor reports since 1993, and manufacturer reports since 1996. Primarily, the intent of the web registry is to evaluate rates of adverse events or to compare adverse event occurrence rates across devices. An extensive online questionnaire gathering device specifics, including location of the event, date, device operator information, label and serial numbers, is provided to assist in identifying a particular product lot.

A variety of uniquely written reports are available through the database. A standard digital form does not exist for the purpose of reporting adverse events. Instead, the data is displayed in free text form. Reports often describe a brief patient history and, occasionally, any interventions employed. Some pre-established categories are available under event type and outcome. The type of events are broadly described as ‘death’, ‘injury’, ‘malfunction’, ‘other’ and ‘no answer provided’. Outcome and subsequent actions are listed as ‘hospitalization’, ‘disability’, ‘death’, ‘required intervention’ and ‘others’. Other outcomes may be listed as ‘invalid’, ‘unknown’ and ‘neither provided nor applicable’.

Malfunction of the device and failure to deploy are isolated, punctual events that occur at the time of filter placement. The purpose of this research was to investigate IVC wall perforation that has occurred over time and was not specifically related to any filter mechanical defect or issues during deployment. In addition, the term migration was broadly used to describe migration of the filter a few centimeters above the area of deployment or migration to the heart, and therefore was not included. In regards to tilting, it is inconsistently reported and difficult to gauge what were the criteria to define it.

Results

Epidemiology, clinical presentation and devices

There were 3311 events reported in the MAUDE database in the intra-vascular filter section during this study period. Of those, 391 (12%) were a description of IVC wall perforation. An annual distribution of 35 cases (9%), ranging from seven (2%) to 70 (18%) were found. Interestingly, a three-fold increase in adverse events was noticed from 2003 to 2004. The annual accrual rate of IVC perforation has not significantly increased, as depicted in Figure 1. The distribution of symptomatic patients who had IVC wall perforation by filters also remains stable over the years (Figure 2). OPEN IN VIEWER

Figure 1.

All reported adverse events and IVC perforation by filters in a national voluntary database over an 11-year period.

OPEN IN VIEWER

Figure 2.

Temporal distribution of symptomatic patients with a vena cava wall perforation caused by an IVC filter.

Nine devices, including the Bard G2, Recovery and Eclipse filters, Celect, Trapease, Greenfield, Gianturco-Roehm, Simon Nitinol and Gunther-Tulip, were reported causing IVC wall perforation. Remarkably, six out of nine were retrievable by design. Of these 391 reported IVC perforations, 168 incidents were associated with the Bard G2® filter, followed by the Recovery in 59 cases and the Celect in 46 cases. Only 46 adverse events (12%) were caused by permanent filters, with the Greenfield filter responsible for 19 cases (41%) and the Gianturco-Roehm for 14 cases (31%).

Discussion

Strong indications for IVC filter placement are contraindication to anticoagulation and failure of anticoagulation to prevent PE. Lately, filters have been used prophylactically in high-risk patients who cannot be on chemical DVT prophylaxis, such as those sustaining severe trauma, including spinal cord injury or severe brain injury. In addition, patients with DVT and a severely reduced cardiopulmonary reserve, and those who are at risk of life-threatening bleeding secondary to falls are currently potential candidates for IVC filter placement. Controversy remains regarding indications for IVC filter placement. A large experience from a community-based database assessed by a panel of specialists demonstrated that IVC filter placement was inappropriate or debatable in up to 50% of patients. ⁵ Further, data from an eight-year follow-up study indicate no effect on mortality from PE after IVC filter placement; however, rates of PE were significantly lower in patients with IVC filters. ⁶

The accrual number of IVC filters placed in the study period years is unknown. One can make an argument that the number of IVC filters commercialized in a certain year can reflect the number of filters placed. However, some filters may expire, be returned to the manufacturer and will never be used. Also, some filters may be intended for use but are discarded before being placed for safety or technical issues. Other filters may never make it to the consumers due to logistic issues. At this time, while an estimate of the number of commercialized filters may be provided by the manufacturers and an extensive market search, an estimate of number of IVC filters actually placed in the US during the study period would be still grossly inaccurate. In an effort to assist in estimating the number of filters placed annually in the US, the American Venous Forum has launched a database to gather this information.

A rapid increase in the number of IVC filter complications between 2003 and 2004 demonstrated in the MAUDE database may be explained by the commercial availability of the first two retrievable filters approved by the FDA in 2003: the Gunther Tulip and Recovery filters. Notably, the latter filter was replaced by the G2 filter in 2005 due to fractures and migration; however, adverse events are still being reported with the new G2 filter, as well as with other retrievable filters. Single-center data on 400 filters showed that studies of all five filter types had similar complication rates except for the Bard G2, which had more migration/tilting events. ⁷ Conversely, a series of 604 IVC filters demonstrated that permanent filters are more prone to complications than retrievable ones. ⁸ Some retrieval filters seem to be more prone to cause IVC perforation due to the configuration of their struts (i.e. Recovery). Nonetheless, it still has to be further investigated whether a specific retrievable filter design was responsible for the increasing number of adverse events since 2004.

Also concerning is the paucity of data available to generate a widely accepted consensus on mid- and long-term surveillance standards in patients possessing an IVC filter. The rate of perforation of the IVC wall remains low and proportionally distributed during the time span of our study. The incidence of perforation reported in the literature is highly debatable. A single-center, prospective randomized trial comparing the TrapEase and Greenfield filters evaluated 156 patients demonstrating no cases of IVC wall perforation assessed by duplex ultrasound (DU) during a two-year study. ⁹ Conversely, another prospective study analyzing retrievability rates of 35 Gunther Tulip filters using computed tomography (CT) 30 days from placement showed IVC wall perforation in 17 cases (48%). ¹⁰ The caveat of the studies on IVC complications is the lack of long-term follow-up, which limits our conclusions on IVC wall perforation rate. In addition, the follow-up provided in some reports is based upon DU, an operator-dependent tool with which one may encounter difficulties to demonstrating venous wall perforation in patients with large amount of bowel gas or those with an obese body habitus, as opposed to CT. Thus, the actual incidence of adverse events related to IVC filters can only be estimated.

Our study shows an interesting and worrisome scenario pertaining to IVC filters, but the data available in the database is far from ideal. The voluntary data provided by healthcare providers is not uniform and lacks important details pertaining to clinical presentation, risk factors, and short- and long-term outcomes. A voluntary database clearly underestimates the number of complications encountered on an annual basis. Furthermore, complications reported in the MAUDE may be merely incidental findings with no serious clinical implications. Manufacturer reports also have flaws due to the fact that they contain more focused, device-oriented information. Nonetheless, the FDA recognizes the magnitude of issues with IVC filters, as reflected in its own website database. This agency issued a warning letter in August, 2010 about filter complications, mainly focusing on those cases in which a retrievable filter had not been removed when the risk of PE subsided. ¹¹

Some of the shortcomings encountered while searching the MAUDE database could be remedied by a multi-center, prospective, mandatory IVC filter registry. A national registry would generate structured data that could be used to address uncertainties about the safety of IVC filters and their clinical impact. Single-center results on filter retrieval rates have been published with excellent results.^12–14 An initiative has been conducted by the American Venous Forum to gather information on filters at the national level using a dedicated database. First large-scale, multispecialty prospective study, the PRESERVE (PREdicting the Safety and Effectiveness of InferioR VEna Cava Filters), has also been launched to evaluate the use of IVC filters and related follow-up treatment.

The role of filters in modern medicine continues to be of critical importance in the prevention of PE in select patients. However, IVC filter are not exempt from complications, regardless of ease of handling and deployment. Thus, indications for filter placement must be carefully reviewed prior to any procedure in order to avoid detrimental results.

The estimated rate of IVC wall perforation was 10% of all reported complications in a national voluntary database and this was frequently found when retrievable filters were utilized. Therefore, filter retrieval has to be performed whenever possible in order to decrease the number of filter-related complications such as IVC perforation.