Dr. Nevitt, Dr. Conti, Dr. Hauptmann, and Dr. Cummings are in the Department of Radiology and Dr. Kozarek is in the Section of Gastroenterology at Virginia Mason Medical Center in Seattle, WA. This article was presented at the American Roentgen Ray Society Meeting, 1997, and the Radiological Society of North America Annual Meeting, 1997.

F ood and Drug Administration (FDA) approval of expandable metallic stents for use in malignant esophageal stricture has brought forth widespread demand and use of the device in the medical community. While conventional (silastic or plastic) esophageal prosthesis placement has been associated with a 6 to 8% perforation rate and a high incidence of post-procedural complications, ¹ expandable metal prostheses provide theoretical advantages in the treatment of malignant strictures, fistulas, and perforations ² because of their overall decreased risk of major bleeding and/or perforation, because there is no need for esophageal dilation to 48- to 5l-French (as is necessary for rigid conventional prostheses), and because of their easier placement in acutely angulated stenoses. ³

Despite the 10-fold cost difference between expandable and conventional prostheses, expansile metal stents have been shown to be a cost-effective and safe alternative in the treatment of malignant esophageal stricture. Recent prospective studies of conventional versus metallic prostheses by Knyrim et al (Wilson Cook silicone prostheses vs. uncovered Wallstent ^® ) ⁴ and Siersema et al (latex Celestin prosthesis vs. European Z-stent) ⁵ were conducted on 39 and 75 patients, respectively. These studies demonstrated fewer complications, decreased hospitalization time, and comparable results of palliation with the expandable metallic prostheses.

It is therefore important for the radiologist to be able to identify each of these stents and the potential problems which may arise, either during deployment or later. At our institution, we have placed more than 70 expandable metallic stents in more than 57 patients to date. Twenty-nine of these patients required additional interventions, as a consequence of either procedure-related complications, prosthesis migration, or tumor progression. The following report describes and illustrates many of the complications we encountered with the use of the Gianturco Z-stent ^® (Wilson-Cook Inc., Winston-Salem, NC), the Wallstent ^® (Schneider Inc., Minneapolis, MN), the Ultraflex (Microvasive Inc., Natick, MA), and the EsophaCoil ^® (Instent Inc., Eden Prairie, MN) which may direct reintervention (table 1).

Background

Esophageal cancer carries a poor prognosis, with up to 80% of patients presenting with advanced disease at diagnosis. While there have been no significant advances in curative treatment of esophageal cancer, treatment of the related dysphagia, specifically with prostheses, has continued to evolve over the last 150 years. The first esophageal prosthesis was created in 1845 by a French surgeon, Leroy d'Etoiles, who carved a short tube of decalcified ivory to make his device. ⁶ In 1885, Sir Charles Symmonds introduced the first indwelling tube, made of boxwood and German silver, which was held in place by retrograde strings fed through the nostrils and either tied behind the ears or to a mustache. ⁷

The evolution of prostheses led to the use of rigid tubes which were defined as pulsion tubes (placed antegrade or through a cervical or thoracic esophagotomy) or traction tubes which were pulled through the tumor from below (requiring laparotomy and gastrotomy). Various modifications of these rigid prostheses are still being used today. However, given the lower morbidity rates of expandable metallic stent insertion, rigid prostheses may eventually go the way of boxwood and German silver. ^3,8

The stents

Mounted on or within a delivery catheter, stents can be inserted over a guidewire after delineation of tumor length radiographically or endoscopically. Correct positioning of the stent at the time of placement can be evaluated fluoroscopically and/or endoscopically and is often dependent upon the regional expertise of the radiologist and/or gastroenterologist.

Expandable metallic prostheses are placed from the mid and distal esophagus. Generally, prostheses are not placed above the region that is 2 cm distal to the cricopharyngeus because of the sensation of foreign body aspiration and narrowing to the trachea.

The original Ultraflex ^® nitinol mesh stent (figure lA) is delivered within a gelatin mold, deploying slowly when heated to body temperature. It is advantageous when placed across acutely angulated stenoses because of its excellent longitudinal flexibility. Constructed of a single knitted strand of elastic alloy wire (0.15 mm nitinol alloy), the Ultraflex possess the least radial force of all of the self-expanding stents available and is more prone to incomplete expansion. Accordingly, a newer variety of the prosthesis uses a larger gauge of wire to produce greater radial force and is available in uncovered and partially covered forms which are delivered by virtue of a thread release.

The self-expanding Wallstent ^® (figure 1B) is formed by a double layer of intertwining stainless steel mesh with or without silicone sandwiched between the layers. It is effective in sealing tracheoesophageal and tracheopleural fistulas (figure 2). The flared wire edges, or "double dogbone" ends, enable it to maintain position in the absence of a distinct tumor shelf. Wallstent II ^® (figure 1E) is a variation of a theme, constructed of a single layer of stainless steel mesh but without the distal semi-rigid flaring. It is available with or without a silicone coating.

The EsophaCoil ^® (Figure 1C) is formed from super-elastic nickel titanium flat wire shaped into a coil spring design which is wrapped tightly onto a deployment catheter. Upon release, the stent springs into place, losing 50% of its original length. Of the available stents, the EsophaCoil exerts the most external radial force and is potentially advantageous for especially tight and angulated strictures.

The Z-stent ^® (Figure lD) is a self-expanding wire cage stent which may or may not be covered with urethane. Placed under fluoroscopic guidance, the covered Z-stent can be used to seal tracheoesophageal and tracheopleural fistulas.

All of the self-expansile metal stents, with the exception of the EsophaCoil, are available with and without an external covering. The coating helps prevent tumor and hyperplastic ingrowth, which may result in subsequent luminal stenosis or occlusion. A disadvantage to adding a coating is the increased propensity of prosthesis migration.

Wire struts designed to hold stents in position also are variable components of the different stents. While providing anchoring stability, they also increase the risk of bleeding, the formation of granulation tissue, and may predispose to chronic ulceration and perforation.

In comparing stents to each other, May and coworkers reviewed 87 patients with 96 expandable prostheses placed (Wallstent: 31; Ultraflex: 35, Z-stent: 30). Early technical success in placement was universal. Twenty-two percent, 37%, and 10% of patients in which a Wallstent, Ultraflex, or Z-stent was placed, respectively, required early reintervention, while 43%, 38%, and 21% of patient, respectively, required subsequent interventions. ⁹ In a comparison of the uncoved Wallstent to the Ultraflex in a prospective study of 82 patients with malignant dysphagia by Schmassmann et al, there was increased procedure-related mortality (16% vs 1%), early complications (32% vs 8%), and severe persistant pain (23% vs 0%) in the Wallstent patients. However, subsequent stent dysfunction (7% vs 32%) and need for reintervention (9% vs 34%) were significantly lower in those patients receiving the Wallstent. ¹⁰

MR imaging considerations

Reports describing the effects of MR on gastrointestinal prostheses are scarce. In a recent publication, Taal et al studied the magnetic effects exhibited on Z-stents, Wallstents and Ultraflex stents when placed in a MR bore. ¹¹ The European model of the Wallstent which was used in this study is constructed of a titanium based alloy which differs from the stainless steel construction of the U.S. model. The researchers found that no significant torque was exhibited on the titanium based Z-stents and Wallstents. However, the stainless steel ferromagnetic Z-stents demonstrated appreciable attractive forces and torque. The authors were uncertain whether this force would be enough to dislodge a stent, but they go on to recommend similar guidelines to stainless steel stents placed into the vascular system, which is to wait a minimum of 6 weeks after deployment to assure firm implantation. ¹¹ Of note, most MR magnets used today generate static fields which are aligned parallel to the patient, and hence parallel to the esophageal prosthesis, which limits effects of torque. Some open air magnets, however, align static fields perpendicular to the patient and hence the stent. This may accentuate torque caused by ferromagnetic stents in some cases.

In addition to dislodgement concerns, significant ferromagnetic artifact will be evident with stainless steel stents which can degrade image quality. No problems with image quality in patients with titanium based stents have been reported. Review of the patient's stent construction, time interval from initial placement, and properties of your institution's magnet will help avoid potential complications and guide imaging choices.

Complications

An increased incidence of device related complications in patients with esophageal stents has been noted following radiation and/or chemotherapy. Radiation has been shown to cause esophageal injury which can manifest as esophagitis, ulceration, and submucosal fibrosis. These effects may be further potentiated in patients receiving chemotherapy, and can lead to perforation and fistula formation secondary to transmural ischemia. ¹²

Complications are arbitrarily divided into two categories: those that require reintervention at the time of or immediately following the procedure, and delayed problems occurring more than 24 hours following the insertion. As patients at our institution stay overnight (or less than 24 hours) after receiving IV sedation, a contrast swallow is obtained before discharge to detect early problems which may require additional interventions.

Procedure-related difficulties which require immediate reintervention include migration, incomplete expansion, and malposition. Immediate migration of a completely covered prosthesis is not uncommon. For instance, recent multicenter Z-stent trial data show a propensity for the stent to migrate into the stomach when placed near or at the EG junction (figure 3). ¹ At our institution, migrated stents are removed endoscopically with a polyp snare, although in some centers surgical removal has been required.

Infolding of the stent will reduce the effective lumen of the stent. Following deployment, infolding commonly is seen with the Ultraflex, and to a lesser degree with the Z-stent. This infolding is remedied by balloon dilation and is encountered less often with the newer stent models. However, even with the strong radial force of the EsophaCoil, infolding, which is really side folding, can be seen (figure 4).

Stent foreshortening is a potential pitfall leading to malposition. Because of the expansile properties of metallic stents, all will shorten following release. In particular, the spring-like deployment of the EsophaCoil, along with a 50% shortening of the pre-deployment length, are factors leading to potential malposition.

Other immediate complications which may require emergent surgical intervention include perforation (seen as pneumomediastinum, pneumothorax (figure 5) and/or contrast extravasation) and uncontrolled bleeding. Perforation, which may occur during dilation, can commonly be sealed with covered prostheses without additional problems. Finally, depending upon the severity of symptoms, tracheal compression may require immediate stent retrieval (if possible), intubation, or stent placement into the pulmonary tree.

As noted, we define delayed complications as those occurring more than 24 hours following deployment and often occur following discharge. Delayed complications may therefore present to a community radiologist at some distance from our tertiary care center. Complications requiring reintervention include migration, tube occlusion, and chronic erosions from the anchoring struts, which can cause bleeding and/or perforation.

Tube occlusion may be divided into several categories: tumor ingrowth through the prosthesis (uncovered stents); tumor overgrowth or elicitation of granulation tissue at the ends of the prosthesis (figure 6); food impaction; and hiatal hernia intussusception (figure 7). We have found an increased incidence of tumor ingrowth with Ultraflex prostheses (figure 8) and early uncovered Wallstents (figure 9) weeks to months after insertion. In general, endoscopic evaluation of tube obstruction best characterizes the cause of the blockage.

Migration of the covered prosthesis was particularly prevalent with the
Z-stent. Migration, a complication seen during deployment, also may present several weeks after insertion (figure 10).

Conclusion

In summary, because of the recent introduction of a variety of expansile metallic stents, more prostheses and their attendant problems will be seen in a general practice. Early and accurate diagnosis of complications by the radiologist with plain films and contrast swallow will help to decrease morbidity and mortality and to guide timely reintervention. AR