Endoprostheses have been used for obstructing tumors of large and small bowel since the beginning of the millenium.^1,2 There has been a relatively rapid pace of development of new devices, these currently being in their 7th generation. Stents for use in the small and large bowel have slightly differing requirements from esophageal stents. The anatomical alignment of the esophagus is straight except for a variable mild angulation of the gastro-esophageal junction which rarely exceeds 60°. In contrast the “C”-shaped duodenum consists of two 90° flexures in its proximal portion and a potentially more acute angulation around the duodenojejunal flexure. Whereas movement in the esophagus is essentially limited to peristalsis, a much greater degree of motility is seen in the intraperitoneal segments of the small and large bowel.

An additional challenge for colonic stents is the necessity to pass solid feces of a considerable diameter through the stent. Patients have some control over the texture of bowel content passing through gastro-duodenal stents through modification of their oral intake, but less control is possible over the consistency of feces. Furthermore, constipation is common in patients with terminal cancer due to poor diet, dehydration, and opiate therapy. This may lead to fecal impaction or migration of the stent. Current guidelines recommend placement of colonic stents of at least 24 mm diameter with no preference for covered or bare stents.^3,4

Two main features have traditionally dominated the development of enteral stents: the need to conform to intestinal flexures and a preference for stents without a covering membrane to allow better mucosal fixation. However, uncovered stents have an increased occlusion rate by tumor growing through the bare metal skeleton. The trade-off between the two primary modes of stent failure of stent occlusion by tumor ingrowth into bare metal stents and migration of fully covered stents have traditionally been reported as similar, ranging between 30% and 40% and increasing with survival.^5–8 Newer stent designs have tried to combine the anchoring capability of a bare stent with an additional membrane applied to the inside of the stent. Designs pursued currently, include metal stents where the membrane is sandwiched between two layers of metal and stents inserted as separate coaxial modules.

An additional challenge to stents used in the gastric outlet and duodenum is the hostile chemical environment. Stents are exposed to a number of enzymes aimed at digesting protein and fat, bile and gastric acid. Stent degradation and fracture is a problem seen with increasing long-term survival after palliative stenting. Once structural integrity is lost, re-intervention is usually required (Fig. 1). Traditionally, stent insertion indicated the terminal stage of cancer palliation, with average survival of 2 to 3 months.^9,10 However, with the increasing addition of palliative chemotherapy and the improvement in tumor specific therapy, survival now commonly extends beyond six months,¹¹ often exceeding the life span of the stent.

The properties of the ideal enteral stent remain yet to be defined. A descriptive analysis of the properties of esophageal stents has shown a wide range of characteristics,¹² the two perceived most important features are the radial expansion force and the axial straightening force. Currently available stents with their specifications have been summarized comprehensively in the recent literature.^13,14

Radial force

The radial force determines the rate of stent expansion much more than the final diameter. The super-elastic properties of nitinol combined with the shape memory effect of the metal result in a structure that relentlessly attempts to reconfigure itself to its original uncompressed shape. The vast majority of current stents will reach at least 90% of their nominal diameter within 1 week. Stents with a stronger radial force will expand quicker, but not necessarily better. Inexperienced operators are often tempted to post-dilate a stent after it has just been deployed. In the vast majority of cases satisfactory stent expansion will occur within 48 hours. Allowing the stent to expand gradually avoids the pain and the risk of perforation associated with balloon dilatation.

Strictures which resist stent expansion are mostly those that have received previous oncological treatment, particularly radiotherapy. This induces a large area of fibrosis around the tumor and, although the stricture may look focal, the tumor is surrounded by a woody environment with very low compliance. Although occasionally seen in the colon, these are fortunately largely limited to the esophagus.

Generally, stent insertion is as effective for intrinsic as well as extrinsic tumors.¹⁵ An exception is extensive extracolonic tumor in the true pelvis. In this situation there may not be sufficient space for the stent to expand (“frozen pelvis”) and success rates are much reduced.¹⁶

While it may be more satisfying to the operator to see the stent expand quickly, this is also likely to cause more pain and needs to be carefully considered against the clinical need for rapid stent expansion. Stents should be chosen accordingly, but there are cases where a guaranteed and definitive early stent expansion is required such as in a critically distended colon at risk of rupture. Occasionally, stents do not expand sufficiently or collapse again.^8,17 In these cases, balloon dilatation may be considered, although it may be better to insert a second stent to strengthen the system and ensure an adequate lumen.

Longitudinal rigidity and conformability

The lower the forces are that return a stent into a straight configuration, the better the stent will sit coaxially in the bowel lumen. If a rigid stent is placed around a flexure, the ends of the stent will dig into the bowel wall, deforming the natural anatomy, so that the axis of the bowel and the axis of the stent are at an angle (Fig. 2). This will impair passage of bowel content through the stent, and in extreme cases, can lead to pressure necrosis of the bowel wall. An acute angle between the stricture and the axis of the bowel has shown to be a risk factor for perforation.¹⁸ Ideally, a stent should have low straightening forces, so it can conform well to the bowel anatomy. The main factor determining stent conformability is the way the stent is constructed.

Stent shortening

Compression of stents into their delivery system causes these to elongate. This is dependent on the degree of compression, defined by the diameter of the stent and its delivery system, as well as the stent construction.

Braided stents of 22 mm or more compressed into a 10 to 11 Fr delivery system are approximately twice as long within the delivery sheath as in their unconstrained state (Fig. 6A). Consequently, as the stents are released and expand, they shorten. This is to be expected and should not come as a surprise. However, the degree of shortening is dependent on the degree of expansion (Fig. 6B) and in tough strictures the stent may end up much longer than expected. While in most cases this is not an issue, it can be a problem in critical areas, for example where access to the duodenal papilla needs to be preserved or stenting into the lower rectum needs to be avoided.

Knitted stents shorten less, typically in the region of 30%, however, this still reflects an elongation of 50% (e.g., a 10 cm stent becoming 15 cm long when loaded into the sheath).

Stent expansion

Some relief of obstructive symptoms should be expected immediately after stent deployment. However stent expansion requires time (Fig. 7), as the shape memory of nitinol is temperature dependent. To what extent tissue necrosis from direct pressure plays a role is unclear. Experience with “stent-in-stent” technique for removal of uncovered stents^24,25 suggests that tumor necrosis can be induced within a few days; however, it is safe to assume, that the majority of the expansion occurs by simple mechanical dilatation.

Braided stents tend to expand more rapidly, due to their relatively high radial force. The trade-off is a higher incidence of pain in sensitive strictures.

While stent expansion may look unsatisfactory on a control radiograph, this may vary with changes in luminal pressure and peristalsis. It is important to consider the functional outcome, regardless of the radiographic appearance.

Disappointingly many operators and published studies do not appreciate the differences in behavior offered by different stent constructions and studies reporting outcomes from a cohort of patients treated with different stents become less meaningful. Equally very few studies have compared different stent constructions and therefore limited data exists on the comparative performance.

Stent migration

Displacement of bare stents is uncommon, occurring in 0% to 6%.^7,9,26 Migration is much higher in fully covered braided stents, reported in up to 38% to 56%^7,26 and historically bare stents have been preferred. The membrane of double knitted or modular construction does not extend to the stent ends, allowing the bare metal to impress into the bowel mucosa. Subsequent tissue hyperplasia as a reaction to the mechanical irritation results in firm fixation of these stents as long as they stay in place in the first few weeks. Migration rates are improved to < 3% for modular stents.^21–23 Results regarding double covered stents are less clear. Limited data has only been published to date with a migration rate of 10% and 20% in the small⁸ and large²⁷ bowel respectively. While Moon et al²⁷ reported a relatively high migration rate in the colon, the re-intervention rate was disproportionately low. This raises the question about the role of tumor bulk reduction by further chemotherapy. Shrinking the tumor reduces stent fixation,²⁸ and while this may result in displacement, it may equally obviate the need for further intervention, if the lumen has been restored. While the majority of migrated stents are passed via the rectum without harm, they can impact (Fig. 8) and perforate.^28,29

Tumor ingrowth

Occlusion of bare stents by tumor progression will eventually occur (Fig. 8), if given enough time and is reported as a cause for stent failure and re-intervention in over 50% after 6 months.³⁰ Reports regarding the effect on patency by additional chemotherapy are controversial.^30,31 Improved survival may paradoxically result in a higher re-intervention rate, as there is more time for tumor to grow into the interstices or over the ends of the stent.^32,33 The use of covered stents reduced stent occlusion as low as 0% in some studies.²⁶ Nonetheless, this benefit seems to be balanced by an increased migration rate. As indicated above, however, this varies significantly between braided and knitted stents and emerging new stent designs seem to start to combine the benefits of increased fixation by bare ends with the central membrane reducing tumor ingrowth.

Stent fracture

A complication increasingly observed is fracturing of the nitinol wire of the stent. This may be observed incidentally on computed tomography (Fig. 8), but may progress to separation of stent segments. As a consequence obstruction may recur, but the displaced stent fragments may impact in the bowel downstream with the risk of acute obstruction or perforation (Fig. 9). To what extent mechanical stress or the hostile biochemical environment play a role in stent fracture remains to be identified.

One of the biggest challenges to stent development is the continued improvement of palliative chemotherapy. As a positive effect, the reduction in tumor bulk can result in asymptomatic expulsion of a covered stent from a colonic stricture, in a fashion “shedding it” when it is no longer required. More commonly, however, increased long-term complications are the rule due to extended patient survival. The longer the stent is in situ, the more likely it is to obstruct or fracture and re-intervention rates rise above 50% at 6 months.^5,30 Development of materials needs to be aimed at improving long-term biocompatibility and resistance. Equally the operator may have to take an increasingly critical approach to stent selection in order to maximize the options for future re-intervention. In cases of likely tumor response, the use of temporary and bio-degradable stents may have an increasing role to play.

Construction of enteral stents has begun do diverge into different types. These are characterized by different stent behavior in terms of stent elongation, radial force and conformability. An understanding of stent properties is essential in order to choose the most suitable device for each patient. The benefits of partially covered stents in offering adequate stent fixation with reduced ingrowth still require confirmation.

Probably the greatest challenge to the interventionist is the increasing survival of palliative patients. This requires an increasingly critical and forward-looking approach to stent selection and procedure planning.