Gastroduodenal Stent Collapse Interventional Radiology Original Research Jin Hyoung Kim 1 Ho-Young Song 1 Ji Hoon Shin 1 Eugene Choi 1 Tae Won Kim 2 Sung Koo Lee 2 Byung Sik Kim 3 Kim JH, Song HY, Shin JH, et al. Keywords: gastrointestinal radiology, interventional radiology, stents DOI:10.2214/AJR.06.1385 Received October 18, 2006; accepted after revision January 17, 2007. 1 Department of Radiology, Asan Medical Center, 388-1 Pungnap-2dong, Songpa-gu, Seoul, Seoul, South Korea 138-736. Address correspondence to H. Y. Song. 2 Department of Internal Medicine, Asan Medical Center, Seoul, South Korea 138-736. 3 Department of Surgery, Asan Medical Center, Seoul, South Korea 138-736. AJR 2007; 188:1495 1499 0361 803X/07/1886 1495 American Roentgen Ray Society Stent Collapse as a Delayed Complication of Placement of a Covered Gastroduodenal Stent OBJECTIVE. The purpose of this study was retrospective evaluation of the incidence, predictive factors, and interventional management of stent collapse after placement of a covered metallic stent in patients with obstruction of the gastric outlet or duodenum due to malignant disease. MATERIALS AND METHODS. Among 259 patients with symptomatic malignant gastroduodenal obstruction successfully treated with stent placement, stent collapse occurred in 12 (4.6%) of the patients 34 270 days (mean, 101.8 days) after stent placement. Multivariate analysis was performed to evaluate factors predictive of stent collapse. Interventional management of stent collapse also was evaluated. RESULTS. Multivariate analysis showed that presence of the stent in the peripyloric region (odds ratio, 27.745; p = 0.036), longer survival time (odds ratio, 1.016; p < 0.001), and absence of chemotherapy after stent placement (odds ratio, 31.661; p = 0.048) were independent predictors of stent collapse. Eleven patients with stent collapse were successfully treated with placement of a second bare stent. The twelfth patient refused further treatment. CONCLUSION. Stent collapse is an uncommon delayed complication of placement of covered metallic stents in patients with malignant gastroduodenal obstruction. Collapse occurs most commonly in the peripyloric region, in patients with longer survival times, and in patients who do not undergo chemotherapy after stent placement. Stent collapse can be managed by coaxial placement of a second bare stent into the collapsed stent. lacement of bare or covered selfexpanding metallic stents has been P a promising noninvasive palliative method of treatment of patients with unresectable cancer causing gastric or duodenal obstruction [1 10]. It has a higher clinical success rate and is associated with a shorter hospital stay and less morbidity and mortality than palliative surgery [11, 12]. However, stent obstruction occurs in approximately 17% of cases of stent placement for malignant gastroduodenal obstruction and is the most frequent complication [13]. In bare stents, obstruction usually is caused by ingrowth of tumor or by tissue hyperplasia through the wire filaments. In covered stents, it is caused by overgrowth of tumor or by tissue hyperplasia above or below the ends of the stent. In theory, stent obstruction can occur from stent collapse due to extrinsic tumor growth compression. To our knowledge, the only case series [14] of stent collapses showed collapse in two (11%) of 18 patients 120 and 256 days after placement of covered metallic stents for the management of malignant gastroduodenal obstruction. The authors surmised that stent collapse resulted from centripetal growth of the tumor or fatigue fracture of the stent wires. However, because of the small size of the study population, conclusions cannot be drawn about the true incidence and factors predictive of stent collapse. In this study, we evaluated cases of stent obstruction due to stent collapse in patients with malignant gastroduodenal obstruction. We evaluated the incidence of, predictive factors for, and interventional management of stent collapse. Materials and Methods Informed consent was obtained from each patient, and our institutional review board approved this retrospective study. Patient Population The study population was 259 patients with symptomatic malignant gastroduodenal obstruction treated with fluoroscopic stent placement. The inclusion period was September 1998 February 2006. The characteristics of the patient population are summarized in Table 1. Because of the presence AJR:188, June 2007 1495
TABLE 1: Characteristics of Patient Population (n =259) Characteristic No. % Age (y) Mean 63 Range 15 86 Sex Men 165 63.7 Women 94 36.3 Site of obstruction Peripyloric region 160 61.8 Nonperipyloric region, 73 28.2 duodenum alone Anastomosis (gastrojejunostomy, 26 10 gastroduodenostomy) Source of malignancy Primary Gastric cancer 141 54.4 Duodenal cancer 8 3.1 Nonprimary Secondary Pancreatic cancer 33 12.8 Gallbladder cancer 17 6.6 Cholangiocarcinoma 21 8.1 Metastasis 13 5 Anastomosis 26 10 of advanced or metastatic disease or a debilitated condition, none of the 259 patients was considered a candidate for surgery. The diagnosis was established by means of endoscopic biopsy, percutaneous needle aspiration biopsy, or forceps biopsy during percutaneous transhepatic or endonasal biliary drainage. Among 259 patients successfully treated with stent placement, stent collapse occurred in 12 (4.6%) of the patients 34 270 days (mean, 101.8 days) after stent placement (Table 2). Nine of the patients were men, and three were women (mean age, 53.2 years; age range, 27 65 years). Stent collapse was identified because it caused recurrent obstructive symptoms in all 12 patients. Placement of the Stent and Follow-Up Three types of stent were used in this study. A type A stent (Song covered stent) was knitted from a single thread of 0.2-mm nitinol wire in a tubular configuration and was fully covered with 12% polyurethane solution (Chronoflex, Cardiotech International). The stent was 18 mm in diameter, and both ends of the stent were flared up to 28 mm (Fig. 1A). The type A stent was used in 18 patients. A type B stent consisted of two stents a bare stent and a fully covered type A stent of the same TABLE 2: Comparison of Group Without Stent Collapse and Group With Stent Collapse Characteristic No Stent Collapse Stent Collapse p No. of patients 247 12 Mean age ± SD (y) 61.8 ± 11.5 53.2 ± 14.2 0.01 Sex 0.545 Men 156 (63.2) 9 (75.0) Women 91 (36.8) 3 (25.0) Site of obstruction 0.033 Peripyloric region 149 (60.3) 11 (91.7) Nonperipyloric region 98 (39.7) 1 (8.3) Source of malignancy 0.247 Primary tumor 140 (56.7) 9 (75.0) Nonprimary tumor 107 (43.3) 3 (25.0) Length of obstruction (cm) 6.3 ± 2.6 5.7 ± 2.5 0.327 Covering membrane 0.075 Nylon mesh 191 (77.3) 12 (100) Polyurethane 56 (22.7) 0 (0) Chemotherapy 0.195 No 181 (73.3) 11 (91.7) Yes 66 (26.7) 1 (8.3) Survival time (d) 75.5 ± 74.1 141.3 ± 62.8 < 0.001 Note Unless otherwise noted, data are numbers of patients. Values in parentheses are percentages. diameter (18 mm). The bare stent was knitted from a single thread of 0.2-mm nitinol wire in a tubular configuration (Fig. 1B). These two stents were designed for placement of the bare stent and then coaxial placement of the type A stent into the bare stent. The type B stent was used in 38 patients. A type C stent (Hercules SP Pyloric stent) [15] consisted of an outer stent and an inner stent. The outer stent had three parts: a leading bare part 28 mm in diameter, a nylon mesh 18 mm in diameter, and a trailing bare part 28 mm in diameter (Fig. 1C). The leading and trailing bare parts were knitted from a single thread of 0.22-mm nitinol wire in a tubular configuration in an interlocking diamond-shaped pattern. The inner stent was the outer bare part of a type B stent. Like the type B stent, the type C stent was designed for coaxial placement. The type C stent was used in 203 patients. The details of stent placement techniques are described elsewhere [8, 12, 15, 16]. All patients underwent a barium radiographic study and endoscopy 1 3 days after stent placement to confirm the position and patency of the stent. A barium study and endoscopy were performed 1 month after stent placement to detect the presence of delayed complications, such as stent migration and obstruction. The status of oral food intake was monitored at 1-month intervals on an outpatient basis. Further follow-up barium studies and endoscopy were performed only for patients with recurrent symptoms. Patients with recurrent symptoms due to stent collapse were treated with interventional management. A second bare stent made of 0.229-mm nitinol wire was chosen for secondary intervention. The technique is shown in Figure 2. For coaxial placement of a second bare stent into the collapsed first stent, a stiff angled, 260-cm-long, 0.035-inch exchange guidewire (Radifocus M, Terumo) was inserted under fluoroscopic guidance across the collapsed stent into the third portion of the duodenum or proximal jejunum. In patients who had difficulty passing the exchange guidewire through the collapsed stent, use of a sizing gastric coil catheter (SongLim gastric catheter) helped negotiation. The exchange guidewire was replaced with a superstiff 260-cm-long guidewire (Medi-tech/Boston Scientific) with use of a sizing gastric coil catheter (S&G Biotech) [15]. When it was extremely difficult to advance the stent delivery system and load the second bare stent through the collapsed portion of the initial stent, we dilated the tight collapsed portion with the use of an 8- or 10-mm balloon catheter. Definition and Analysis of Data Stent collapse was defined as delayed luminal narrowing of a fully expanded stent owing to extrinsic compression from tumor growth. Univariate 1496 AJR:188, June 2007
Gastroduodenal Stent Collapse Fig. 1 Stents. A, Photograph shows polyurethane-covered nitinol stent (type A). B, Photograph shows type B stent, which consists of type A stent inside bare stent. C, Photograph shows type C stent, which consists of bare stent and partially covered stent designed for coaxial placement. Fig. 2 54-year-old man with stent collapse. A, Upper gastrointestinal radiograph after passage of sizing gastric coil catheter over guidewire to distal part of collapsed stent and injection of water-soluble nonionic contrast medium through catheter shows collapsed stent in peripyloric region. B, Upper gastrointestinal radiograph shows successful insertion of stent delivery system and loading of 0.229-mm-wire inner bare stent (arrowheads) through collapsed stent. C, Upper gastrointestinal barium radiograph obtained with water-soluble nonionic contrast medium immediately after coaxial placement of second bare stent into collapsed stent shows good downstream passage of contrast medium. Fig. 3 62-year-old man with gastric cancer. A, Upper gastrointestinal barium radiograph 3 days before stent placement shows malignant stricture (arrow) in peripyloric region. B, Upper gastrointestinal barium radiograph 1 day after placement of type C stent shows fully expanded stent (arrow) and good barium passage. C, Fluoroscopic image 116 days after stent placement shows collapsed stent (arrow). analysis was performed to compare variables between the group with stent collapse and the group without stent collapse. The Mann-Whitney U test was used to compare continuous variables, and the Fisher s exact test was used for categoric variables. A multivariate logistic regression model with forward stepwise selection was planned for finding independent predictive factors associated with stent collapse. Only variables with p <0.2 at univariate analysis (Mann-Whitney U test or Fisher s exact test) were entered into the multivariate logistic regression model. This lenient p value was chosen to avoid rejection of a variable that might have contributed to prediction of outcome (stent collapse) in the multivariable analysis. Model fit was evaluated by means of the Hosmer- Lemeshow goodness-of-fit test [17]. Two-sided p < 0.05 was considered to indicate statistical significance. All statistical analysis was performed with the SPSS program (version 11.5, SPSS). After interventional management for stent collapse, clinical outcome was evaluated retrospectively. Results Predictive Factors of Stent Collapse All episodes of stent collapse occurred in the barrel portion of type C stents (Fig. 3) but not in type A or type B stents, which were covered with polyurethane membrane. Univariate analysis showed stent collapse was associated with younger age (53.2 ± 14.2 vs 61.8 ± 11.5 years for patients without stent collapse, p = 0.01), location in the peripyloric region (p = AJR:188, June 2007 1497
TABLE 3: Results of Multivariate Logistic Regression Analysis for Evaluation of Factors Predictive of Stent Collapse Variable No. (%) Odds Ratio 95% CI p Site of obstruction Peripyloric region 160 (61.8) 27.745 1.251 615.551 0.036 Nonperipyloric region 99 (38.2) 1.000 Survival time 259 (100) 1.016 1.008 1.024 < 0.001 Chemotherapy No 192 (74.1) 31.661 1.033 970.213 0.048 Yes 67 (25.9) 1.000 0.033), and longer survival time (141.3 ± 62.8 vs 75.5 ± 74.1 days for patients without stent collapse, p < 0.001) (Table 2). Other variables showed no significant difference between the group with and the group without stent collapse. The following variables of p <0.2 at univariate analysis were entered into the multiple logistic regression model: age, site of obstruction, covering membrane, chemotherapy after stent placement, and survival time. Multivariate analysis confirmed that peripyloric region (odds ratio [OR], 27.745; 95% CI, 1.251 615.551; p = 0.036), longer survival time (OR, 1.016; 95% CI, 1.008 1.024; p < 0.001), and absence of chemotherapy after stent placement (OR, 31.661; 95% CI, 1.033 970.213; p = 0.048) were independent predictors of stent collapse (Table 3). The Hosmer-Lemeshow goodness-of-fit test showed a nonsignificant p value (0.998) for the model, which indicated good fitness of the model. Interventional Management for Stent Collapse Eleven of the 12 patients with stent collapse underwent interventional management. One patient refused further treatment and underwent total parenteral nutrition until death. Before secondary stent placement, dilation of the collapsed portion with an 8- or 10-mm balloon catheter was needed in all 11 patients. In two patients, a sizing gastric coil catheter was needed for negotiation of an exchange guidewire through the collapsed stent. All 11 patients had relief of recurrent obstructive symptoms after placement of the second bare stent and had no complications. After management of stent collapse, the 11 patients maintained oral intake until death or the end of the follow-up period (mean, 46.4 days; range, 10 104 days) without recurrence of obstructive symptoms. Discussion The results showed the rarity of stent collapse, which occurred with a frequency of 4.6% (12/259) and which tended toward late occurrence (mean, 101.8 days after placement; range, 34 270 days). In our multiple logistic regression analysis, stent placement in the peripyloric region (OR, 27.745; p =0.036), long survival time (OR, 1.016; p <0.001), and absence of chemotherapy after stent placement (OR, 31.661; p = 0.048) were significant predictors of stent collapse. Therefore, we can predict this uncommon complication after placement of covered metallic stents in patients with malignant peripyloric gastroduodenal obstruction, longer survival times, and no chemotherapy after stent placement. To our knowledge, there have been no reports of studies of stent collapse after placement of covered stents in the esophagus or colorectal area [18 23], but there have been a few reports [14] of this complication in the gastroduodenal area. There are several explanations for the lack of reporting of this complication. First, in cases of polyurethane-covered stents, not stent collapse but tumor ingrowth through the disruption of the covering membrane occurs [16, 20]. In our study, stent collapse occurred only in the type C stents covered with nylon mesh, which may be more durable than polyurethane membrane. Both the outer and inner components of type B stents had a framework made of 0.2- mm nitinol wire. On the other hand, the barrel portion of type C stents was weaker than that of type B stents because the outer, partially covered component of a type C stent does not have a framework in the nylon mesh between the proximal and distal stent bodies. We speculate that this structural weakness caused collapse of the type C stents alone. Second, given the rarity and delayed presentation of this complication in this study, stent collapse may be difficult to detect, especially in patients with short-term survival and follow-up periods. Third, stent collapse appears to have a preferable location. In our study and that of Jeong et al. [14], almost all cases of collapse occurred in the peripyloric region. The observation that peripyloric lesions are related to the incidence of stent collapse can be explained by the presence of the pyloric channel, which is a physiologic stenotic and resistant structure. The pyloric channel is formed by duplication of mucous membrane and an abundant muscle layer [24]. Therefore, stents in this area can be subjected to high levels of stress, which causes collapse, especially with peristaltic movement in the area around the stent. We found that absence of chemotherapy after stent placement was a risk factor for stent collapse. Chemotherapy, by reducing the tumor burden, might have decreased the risk of stent collapse due to compression from tumor growth. Jeong et al. [14] successfully managed a case of stent collapse by placing a second covered stent through the collapsed portion of the initial covered stent. Although they obtained positive clinical outcome in the one case, coaxial placement of a covered stent because of stent collapse seemed inappropriate for the following reasons: Placement of the covered stent through the collapsed portion can be difficult because of the requirement for a rigid 6-mm delivery system, and an inner covered stent is at high risk of migration [16]. In all 11 patients who underwent placement of a second bare stent, recurrent obstructive symptoms were relieved and new symptoms did not occur. Hoop strength, a parameter that describes the ability of a stent to withstand external forces, is directly influenced by stent wire thickness [25]. Because of its greater hoop strength, a 0.229-mm-wire inner bare stent as opposed to a 0.2-mm-wire inner bare stent was used in all 11 patients. The 0.2-mmwire inner bare stent appeared to be too weak in patients with bulky masses. Although stent collapse was uncommon after placement of type C stents, which had 0.2-mm-wire inner bare stents, the incidence of stent collapse is more likely to decrease when a 0.229-mmwire inner bare stent is used initially. The optimal hoop strength value, however, remains unclear. A stent with high hoop strength may have the potential for increasing the perforation rate or decreasing stent flexibility. Further study regarding the optimal hoop strength of gastroduodenal stents is needed. The principal limitation of this study was its nonrandomized and retrospective design, 1498 AJR:188, June 2007
Gastroduodenal Stent Collapse which inherently decreased the statistical strength. Another limitation was the small number of patients who underwent placement of polyurethane-covered (types A and B) stents compared with the number who underwent placement of nylon-mesh-covered (type C) stents. In summary, stent collapse is an uncommon delayed complication of placement of covered metallic stents in patients with malignant gastroduodenal obstruction. Factors predictive of stent collapse are the presence of the stent in the peripyloric region, longer survival time, and absence of chemotherapy after stent placement. Stent collapse can be managed by coaxial placement of a second bare stent into the collapsed stent. References 1. Binkert CA, Jost R, Steiner A, Zollikofer CL. Benign and malignant stenoses of the stomach and duodenum: treatment with self-expanding metallic endoprostheses. Radiology 1996; 199:335 338 2. Baron TH, Harewood GC. Enteral self-expandable stents. Gastrointest Endosc 2003; 58:421 433 3. Lopera JE, Alvarez O, Castano R, Castaneda- Zuniga W. Initial experience with Song s covered duodenal stent in the treatment of malignant gastroduodenal obstruction. J Vasc Interv Radiol 2001; 12:1297 1303 4. Adler DG, Baron TH. Endoscopic palliation of malignant gastric outlet obstruction using self-expanding metal stents: experience in 36 patients. Am J Gastroenterol 2002; 97:72 78 5. Yoon CJ, Song HY, Shin JH, et al. Malignant duodenal obstructions: palliative treatment using self-expandable nitinol stents. J Vasc Interv Radiol 2006; 17:319 326 6. Baron TH. Expandable metal stents for the treatment of cancerous obstruction of the gastrointestinal tract. N Engl J Med 2001; 344:1681 1687 7. Mauro MA, Koehler RE, Baron TH. Advances in gastrointestinal intervention: the treatment of gastroduodenal and colorectal obstructions with metallic stents. Radiology 2000; 215:659 669 8. Jung GS, Song HY, Kang SG, et al. Malignant gastroduodenal obstructions: treatment by means of a covered expandable metallic stent initial experience. Radiology 2000; 216:758 763 9. Park KB, Do YS, Kang WK, et al. Malignant obstruction of gastric outlet and duodenum: palliation with flexible covered metallic stents. Radiology 2001; 219:679 683 10. Bessoud B, de Baere T, Denys A, et al. Malignant gastroduodenal obstruction: palliation with self-expanding metallic stents. J Vasc Interv Radiol 2005; 16:247 253 11. Del Piano M, Ballare M, Montino F, et al. Endoscopy or surgery for malignant GI outlet obstruction? Gastrointest Endosc 2005; 61:421 426 12. Lopera JE, Brazzini A, Gonzales A, Castaneda- Zuniga WR. Gastroduodenal stent placement: current status. RadioGraphics 2004; 24:1561 1573 13. Dormann A, Meisner S, Verin N, Wenk Lang A. Self-expanding metal stents for gastroduodenal malignancies: systematic review of their clinical effectiveness. Endoscopy 2004; 36:543 550 14. Jeong JY, Han JK, Kim AY, et al. Fluoroscopically guided placement of a covered self-expandable metallic stent for malignant antroduodenal obstructions: preliminary results in 18 patients. AJR 2002; 178:847 852 15. Song HY, Shin JH, Yoon CJ, et al. A dual expandable nitinol stent: experience in 102 patients with malignant gastroduodenal strictures. J Vasc Interv Radiol 2004; 15:1443 1449 16. Jung GS, Song HY, Seo TS, et al. Malignant gastric outlet obstructions: treatment by means of coaxial placement of uncovered and covered expandable nitinol stents. J Vasc Interv Radiol 2002; 13:275 283 17. Hosmer D, Lemeshow S. Applied logistic regression. New York, NY: Wiley, 1989 18. Sabharwal T, Morales JP, Salter R, Adam A. Esophageal cancer: self-expanding metallic stents. Abdom Imaging 2005; 30:456 464 19. Homs MY, Steyerberg EW, Kuipers EJ, et al. Causes and treatment of recurrent dysphagia after self-expanding metal stent placement for palliation of esophageal carcinoma. Endoscopy 2004; 36:880 886 20. Saranovic DJ, Djuric-Stefanovic A, Ivanovic A, Masulovic D, Pesko P. Fluoroscopically guided insertion of self-expandable metal esophageal stents for palliative treatment of patients with malignant stenosis of esophagus and cardia: comparison of uncovered and covered stent types. Dis Esophagus 2005; 18:230 238 21. Meisner S, Hensler M, Knop FK, West F, Wille- Jorgensen P. Self-expanding metal stents for colonic obstruction: experiences from 104 procedures in a single center. Dis Colon Rectum 2004; 47:444 450 22. Sebastian S, Johnston S, Geoghegan T, Torreggiani W, Buckley M. Pooled analysis of the efficacy and safety of self-expanding metal stenting in malignant colorectal obstruction. Am J Gastroenterol 2004; 99:2051 2057 23. Keymling M. Colorectal stenting. Endoscopy 2003; 35:234 238 24. Gray H. Anatomy of the human body, 20th ed. New York, NY: Bartleby.com, 2000. Available at: www.bartleby.com/107. Accessed April 6, 2006 25. Duda SH, Wiskirchen J, Tepe G, et al. Physical properties of endovascular stents: an experimental comparison. J Vasc Interv Radiol 2000; 11:645 654 AJR:188, June 2007 1499