Do Pancreatic Duct Stents Cause or Prevent Pancreatic Sepsis?

Do pancreatic duct stents cause or prevent pancreatic sepsis? Richard Kozarek, MD, Oistein Hovde, MD, Fouad Attia, MBBch, Renee France, BHK Seattle, Washington

Background: Pancreatic sepsis can occur after contrast injection into an obstructed or disrupted pancreatic duct. Whether stents cause or prevent pancreatic sepsis is unknown. Accordingly, the pancreatic duct bacteriology in patients with pancreatic duct stents was retrospectively reviewed and contrasted with biliary cultures taken from patients at the time of bile duct stent retrieval and/or exchange. Methods: Of 61 patients (29 men, 32 women; 72 stents; mean age 51 [16] years, range 14-88 years), 36 with pancreatic duct stents had pancreatic duct cultures obtained at the time of stent exchange and/or retrieval. The results of these cultures were compared with bile duct cultures taken from 36 patients at the time of biliary stent exchange/retrieval. Eleven of the 36 patients with pancreatic duct stents also had bile duct stents. Data collected included stent patency, clinical sepsis at initial stent placement or retrieval, administration of antibiotics before the procedure, indication for stent placement, stent duration, and culture results. Results: At stent retrieval and/or exchange, all 61 patients with pancreatic and/or biliary stents had contamination of the respective ducts with multiple enteric bacteria (mean 3.4 organisms in patients with pancreatic duct stents vs. 3.3 in those with bile duct stents). Clostridium perfringens was found in 17% and 0% of patients with, respectively, bile duct and pancreatic duct stents. Among the most common indications for pancreatic duct stent placement were stricture (28), sphincterotomy (9), leak (7), stones (3), and dilated pancreatic duct (1). Indications for a biliary stent included benign stricture (29), malignancy (6), stones (2), cholangitis (1), chronic pancreatitis (1), and dilated common bile duct (1). Pancreatic cultures were taken at a median of 85 days (interquartile range 60-126; range 13-273) and biliary cultures at a median of 87 days (interquartile range 45-149; range 19-927) after stent placement. Eleven patients, 6 with a bile duct stent, 4 with a pancreatic duct stent, and one with dual stents, developed pre-exchange/retrieval clinical sepsis; 3 had pancreatic sepsis. All had received antibiotics at initial placement. In the 11 patients with sepsis (12 stents), 8 stents were completely occluded at exchange/retrieval, 3 were partially occluded, and one was patent. In 50 patients (60 stents), no clinical sepsis developed; 7 stents were patent, 31 partially occluded, and 22 completely obstructed. Conclusions: (1) Comparable to patients with biliary stents, all patients with pancreatic stents had contamination of the pancreatic ductal system by enteric flora. (2) In contrast to the 17% of patients with bile duct stents who had intraductal Clostridium perfringens, there were no instances of contamination with this organism in patients with pancreatic stent (p = 0.025), although, after adjusting for multiple comparisons, statistical significance was lost. (3) There was a tendency for stent occlusion to predispose to pancreatic sepsis, but occlusion by itself was insufficient (p = 0.106). (4) Further investigation is required to define the additional variables that are associated with the development of pancreatic sepsis. (Gastrointest Endosc 2003;58:505-9.)

Polyethylene stents are both a cause and a cure for cholangitis. On one hand, stent insertion has been shown to be as effective as nasobiliary drainage for the treatment of biliary sepsis associated with choledocholithiasis, even in the absence of Received March 17, 2003. Accepted June 5, 2003. Current affiliations: Section of Gastroenterology, Virginia Mason Medical Center, Seattle, Washington. Reprint requests: Richard A. Kozarek, MD, Chief of Gastroenterology, Virginia Mason Medical Center, 1100 Ninth Ave., Seattle, WA 98101. Copyright © 2003 by the American Society for Gastrointestinal Endoscopy 0016-5107/2003/$30.00 + 0 PII:S0016-5107(03)01891-1 VOLUME 58, NO. 4, 2003

clearance of ductal calculi.1 On the other hand, plastic stents develop biofilms by virtue of complex interactions between host proteins to include fibrin, immunoglobulin A, fibronectin, and collagen, which promote bacterial adherence and subsequent elaboration of glycoproteins.2-10 These biofilms result in concentric luminal occlusion that has the potential to cause not only jaundice but frank cholangitis with biliary sepsis. Infectious complications of pancreatitis are well described, and pancreatic sepsis has been associated with morbidity and mortality rates that range from 5% to 10% for incompletely treated, infected, pancreatic pseudocysts to 100% for undrained pancreatGASTROINTESTINAL ENDOSCOPY

505

R Kozarek, O Hovde, F Attia, et al.

Do pancreatic duct stents cause or prevent pancreatic sepsis?

Table 1. Indications (more than one possible) PD stent Stricture Sphincterotomy Leak Stones Dilated PD

BD stent 28 9 7 3 1

Benign stricture Malignancy Stones Cholangitis Dilated BD Chronic pancreatitis

29 6 2 1 1 1

PD, Pancreatic duct; BD, bile duct.

ic abscess.11,12 Infected pancreatic necrosis, in turn, is associated with mortality rates of 10% to 50% as a consequence of sepsis and multisystem organ failure.13-19 Although these infectious complications are usually the consequence of bacterial translocation of gut flora, they also may occur after endoscopic manipulation of an obstructed or disrupted pancreatic duct (PD). Pancreatic duct stents are routinely placed after such manipulation to preclude obstruction, improve ductal flow, and prevent iatrogenic infection and pancreatitis.20 Although it is known that pancreatic stents also occlude over time by virtue of biofilm encrustation,21 it is not known whether stent placement precludes or predisposes to infection as a result of stent obstruction or reflux of duodenal contents into the pancreas. Accordingly, a study was conducted of the bacteriology of the PD at the time of stent exchange or retrieval in patients in whom a PD stent had been placed to facilitate a sphincterotomy or treat strictures, stones, or ductal disruptions. These results, to include infectious complications, were compared with those for bacteriologic assessment of the bile duct (BD) in a group of patients with BD stents scheduled for exchange or retrieval. PATIENTS AND METHODS A retrospective study was conducted of 61 consecutive patients (29 men, 32 women; mean age 51 [16] years, range 14-88 years) between January and June of 2001. The study was approved by the institutional review board of our medical center. No patient who underwent endoscopic stent retrieval without ERCP was included. The 61 patients had 36 pancreatic and 36 biliary stents; 11 had stents in both ducts. The original indications for stent insertion are outlined in Table 1. In addition to the indications for stent placement, clinical data collected included stent patency, clinical sepsis at the time of initial stent placement and retrieval (spiking fever, leukocytosis, abdominal pain with or without positive blood cultures), administration of antibiotics before the procedure, duration of stent placement, stent diameter, and culture results. After stent retrieval, patency was determined by a saline solution injection through the proximal end of the stent by using a blunt tip needle connected to a 3-mL 506

GASTROINTESTINAL ENDOSCOPY

syringe. Patency was defined as immediate flow through the distal end and side flaps; partial patency, as occlusion of the distal end but with flow through the more distal flaps; and occlusion, as no flow through either the end or flaps without excessive injection pressure and delivery of significant luminal debris with injection. Care was taken to prevent significant kinking of the stent at time of removal with a polypectomy snare. Cultures were obtained after stent removal and before contrast injection. Two milliliters of pancreatic juice and/or 2 mL of bile were aspirated into a sterile ERCP catheter. Previous studies in our unit have demonstrated either sterile cultures in the BD (9/10 patients) and PD (8/10 patients), respectively, with isolation of Staphylococcus or yeast in the remainder. Specimens were plated on chocolate agar and cultured aerobically and anaerobically as previously described.22 Colonies were isolated and identified at 24, 48, and 72 hours before discarding the plate. Statistical comparisons of patient groups were made by using a two-sided Fisher exact test. It was assumed that observations in patients with both PD and BD stents were independent. A p value less than 0.05 was considered statistically significant. Bonferroni’s method was used to correct significant p values to recognize that multiple statistical tests of hypotheses were performed on data from individual patients. Summary statistics are presented with mean, standard deviation, and range. Highly skewed variables are presented as median, interquartile range (IQR) (25th and 75th percentile), and range.

RESULTS At the time of stent retrieval and/or exchange, all 61 patients with pancreaticobiliary stents had contamination of the respective ducts (Table 2). The mean number of organisms was 3.4 (range 2-6) in the PD and 3.3 (range 2-5) in the BD. Clostridium perfringens was found only in BD cultures (6/36 [17%] vs. 0/36; p = 0.025). However, comparisons were made between the BD and PD for all organisms (n = 20), and, because of this multiple comparison issue, the statistical significance for the detection of C perfringens in the BD but not the PD must be discounted, and there may not truly be differences. After using Bonferroni’s method to adjust for multiple statistical tests, this p value is no longer significant. Pancreatic duct stents had been in place for a median of 85 days (IQR 60-126; range 13-273) and biliary stents for a median of 87 days (IQR 45149; range 19-927). The duration of stent placement did not appear to play a role because there was no difference in the contamination rate and number of different organisms whether the patient had a stent in place for days or months. All patients had received either cefotetan (1 g) or levofloxacin (500 mg) at the time of initial stent placement and at the time of retrieval or exchange. Eleven patients, 6 with biliary, and 4 with PD VOLUME 58, NO. 4, 2003

Do pancreatic duct stents cause or prevent pancreatic sepsis?

R Kozarek, O Hovde, F Attia, et al.

Table 2. Bacteriology of aspirated bile/pancreatic juice

Organism Streptococcus Enterococcus Coliforms* Candida albicans/yeast Gram-negative rods* Staphylococcus coag –ve Pseudomonas Escherichia coli Anaerobes* Klebsiella Total

BD (n = 36 stents)

PD (n = 36 stents)

26 27 21 16 3 5 1 0 3 2

34 26 18 12 8 4 4 3 3 3

Organism Bacteroides fragilis Diphtheroids Enterobacter cloacae Neisseria Staphylococcus aureus Capnocytophaga Lactobacillus Eikenella Haemophilus sp Clostridium perfringens

BD (n = 36 stents)

PD (n = 36 stents)

3 1 1 0 0 0 2 0 2 6 119

2 1 1 1 1 1 1 1 0 0 124

BD, Bile duct; PD, pancreatic duct. *Unable to further characterize.

stents, and one with dual stents, presented with fever, leukocytosis, and clinical evidence of sepsis, although positive blood cultures were obtained in only 8. All of these patients had comparable organisms cultured from blood and ductal systems, with the exception of a patient with pancreatic enterococcal sepsis in whom this organism was isolated from a peripancreatic fluid collection. For the 11 patients with sepsis (including two of the 3 patients with pancreatic infections), 8 stents were completely occluded, 3 partially occluded, and one stent was patent. Neither stent diameter nor indication for placement predisposed to ductal contamination or sepsis. In the 50 patients (60 stents) in whom there were no features of sepsis, 7 stents were patent, 31 partly occluded, and 22 completely obstructed. Statistical analysis demonstrated that stent occlusion predisposed to sepsis (p = 0.106), but occlusion by itself was insufficient to cause clinical infection. There was no statistical association between indications for stent placement, duration of placement, age, gender, stent diameter, or the administration before the procedure of antibiotics and the development of preprocedure sepsis. DISCUSSION Much has been written about cholangitis as a consequence of biliary stent occlusion. The latter is most commonly the consequence of bacterial biofilm development with subsequent biliary stasis and overgrowth of bacteria translocated from the gut.3-5,7-10 Gram-negative rods and gram-positive cocci predominate, both in the contaminated bile23 and within the lumen of the stent, although studies suggest that anaerobic bacteria may play a role related to early adherence to the stent.10 Neither antibiotics nor use of ursodeoxycholic acid consistently has been shown to improve stent patency, and VOLUME 58, NO. 4, 2003

the use of antimucin drugs such as aspirin also has been disappointing.24-31 Pancreatic infections have long been recognized to result in considerable morbidity and mortality if untreated and undrained.11-19 The mortality rate ranges from 10% to 50% for infected pancreatic necrosis to 100% for undrained pancreatic abscess. Pancreatic stents, which have been used to prevent edematous closure of a PD sphincterotomy and to treat a variety of ductal obstructions and leaks, are being used increasingly, both to treat underlying conditions, as well as to preclude iatrogenic pancreatitis related to manipulation. As with biliary stents, bacterial biofilm formation results in eventual obstruction and the potential for obstructive pancreatitis and/or iatrogenic infection.21 Despite a belief that pancreatic enzymes inhibit bacterial growth within the PD, all 36 patients undergoing pancreatic stent removal or exchange in the present series had colonization of the PD, primarily with enteric flora. This contamination occurred early because there was no statistical difference in the number of organisms cultured from patients in whom stents were retrieved within 2 weeks of placement as opposed to those in whom stents had been left in place for several months. Moreover, polymicrobial growth was associated with clinical sepsis in 3 patients (intraductal infection in two and subsequent infection of an associated fluid collection in one patient). All of these infections responded to treatment with antibiotics plus stent exchange (2) or removal (1), although, in one patient, concomitant percutaneous drainage of a residual fluid collection was required. Our technique of duct culture, although imperfect because of passage of a sterile catheter through a potentially contaminated elevator channel, has proven to be reproducible, and previous studies in our unit have demonstrated proper endoscope disinfection as perGASTROINTESTINAL ENDOSCOPY

507

R Kozarek, O Hovde, F Attia, et al.

haps the most important factor in preventing contamination of endoscopic accessories.32-34 When individual variables were reviewed, neither patient demographics nor indication for stent placement or stent diameter appeared to predispose to development of polymicrobial contamination of the PD or pancreatic sepsis. Stent patency, however, may have played a role because two of the 3 patients with pancreatic infections had complete stent occlusion, and the third had a partial stent occlusion in the setting of an infected pancreatic fluid collection. The fact that all but one of the 36 pancreatic stents were of 5F to 7F diameter compared with the 10F to 11.5F diameter for biliary stents would seem to be important, because stent occlusion appears to be inversely related to stent diameter.9 However, neither stent diameter nor occlusion by itself demonstrated statistical significance in this study, probably because of low statistical power to detect differences given the relatively small number of observations. Stent occlusion by itself, however, was clearly insufficient to cause clinical infection, because all but 7 of the pancreaticobiliary stents were partly (n = 31) or completely (n = 22) occluded in the 50 patients who did not manifest signs or symptoms of sepsis or local infection. Our methodology for defining stent patency has worked well in our unit for many years, although stent compression by the polypectomy snare loop can result in an overestimation of occlusion. Care was taken to loosely grasp the stents at the level of an external flap to prevent this. Other, differing methodologies for defining patency have been described, but these also appear to have inherent limitations.35 There are data to suggest that adherence of anaerobic bacteria, particularly C perfringens, may be an early component of bacterial biofilm development in biliary stents.10 This organism was identified in 17% of patients in whom biliary stents were retrieved or exchanged, but it was not found in the pancreatic juice of any patient who had a PD stent exchanged or retrieved. The significance of this observation remains uncertain because other anaerobes were occasionally found (Table 2). These data are in contrast to those of Rerknimitr et al.,36 who found an extremely low frequency of anaerobic bacteria in the bile of patients with stents, although there may have been methodologic problems with specimen handling in their study. In summary, PD stent placement is associated with contamination of the PD by multiple organisms. Although most patients with such contamination were asymptomatic, the contamination may lead to infectious complications that include the development of pus within the duct or infection of 508

GASTROINTESTINAL ENDOSCOPY

Do pancreatic duct stents cause or prevent pancreatic sepsis?

peripancreatic fluid collections. Stent occlusion may be necessary but is insufficient as a sole cause of these infectious complications. Additional variables, to possibly include the concentration of pancreatic enzymes with intraductal activation, may play a protective role. Studies addressing these, as well as additional variables, are currently under way. REFERENCES 1. Lee DW, Chan AC, Lam YH, Ng EK, Lau JY, Law BK, et al. Biliary decompression by nasobiliary catheter or biliary stent in acute suppurative cholangitis: a prospective randomized trial. Gastrointest Endosc 2002;56:361-5. 2. Groen AK, Out T, Huibregtse K, Delzenne B, Hoek FJ, Tytgat GN. Characterization of the content of occluded biliary endoprostheses. Endoscopy 1987;19:57-9. 3. Speer A, Cotton P, Rode J, Seddon AM, Neal CR, Holton J, et al. Biliary stent blockage with bacterial biofilm. Ann Intern Med 1988;108:546-53. 4. Dowidar N, Kolmos HJ, Lyon H, Matzen P. Clogging of biliary endoprostheses: a morphologic and bacteriologic study. Scand J Gastroenterol 1991;26:1137-44. 5. Leung JWC, Liu YL, Desta T, Libby ED, Incardi JF, Lam K. Is there a synergistic effect between mixed bacterial infection in biofilm formation in biliary stents? Gastrointest Endosc 1998;48:250-7. 6. Yasuda H, Koga T, Fukuoka T. In vitro and in vivo models of bacterial biofilms. Methods Enzymol 1999;310:579-95. 7. Faigel DO. Preventing biliary stent occlusion. Gastrointest Endosc 2000;51:104-7. 8. van Berkel AM, van Marle J, van Veen H, Groen AK, Huibregtse K. A scanning electron microscopic study of biliary stent materials. Gastrointest Endosc 2000;51:19-22. 9. Maillot M, Aucher P, Robert S, Richer JP, Bon B, Moesch C, et al. Polyethylene stent blockage: a porcine model. Gastrointest Endosc 2000;51:12-8. 10. Leung JW, Liu YL, Chan RCY, Tang Y, Mina Y, Cheng AF, Silva J Jr. Early attachment of anaerobic bacteria may play an important role in biliary stent blockage. Gastrointest Endosc 2000;52:725-9. 11. Howell DA, Elton E, Parsons WG. Endoscopic management of pseudocysts of the pancreas. Gastrointest Endosc Clin N Am 1998;8:143-62. 12. Isenmann R, Schoenberg MH, Rau B, Beger HG. Natural course of acute pancreatitis: pancreatic abscess. In: The pancreas. HG Beger, AL Warshaw, MW Büchler, DL Carr-Locke, JP Neoptolemos, C Russell, MG Scarr, editors. Vol. 1. London: Blackwell Science; 1998. p. 461-5. 13. Tsiotos GG, Luque-de Leon E, Sarr MG. Long-term outcome of necrotizing pancreatitis treated by necrosectomy. Br J Surg 1998;85:1650-3. 14. Büchler P, Reber HA. Surgical approach in patients with acute pancreatitis. Is infected or sterile necrosis an indication: in whom should this be done, when, and why? Gastrointest Endosc Clin N Am 1999;28:661-71. 15. Slavin J, Ghaneh P, Sutton R, Hartley M, Rowland SP, Garvey C, et al. Management of necrotizing pancreatitis. World J Gastroenterol 2001;7:476-81. 16. Isenmann R, Rau B, Beger HG. Bacterial infection and extent of necrosis are determinants of organ failure in patients with acute necrotizing pancreatitis. Br J Surg 1999;86:1020-4. 17. Büchler MW, Gloor B, Müller CA, Friess H, Seiler CA, Uhl W. Acute necrotizing pancreatitis: treatment strategy according to the status of infection. Ann Surg 2000;232:619-26. VOLUME 58, NO. 4, 2003

Do pancreatic duct stents cause or prevent pancreatic sepsis?

18. Bosscha K, Hulstaert PF, Hennipman A, Visser MR, Gooszen HG, van Vroonhoven JMV, van Werken C. Fulminant acute pancreatitis and infected necrosis: results of open management of the abdomen and “planned” reoperations. J Am Coll Surg 1998;187:255-62. 19. Sharma VK, Howden CW. Prophylactic antibiotic administration reduces sepsis and mortality in acute necrotizing pancreatitis. Pancreas 2001;22:28-31. 20. Kozarek RA. Endoscopic therapy of complete and partial pancreatic duct disruptions. Gastrointest Endosc Clin North Am 1998;8:39-53. 21. Leung JW, Liu YL, Herrera J, Jang S, Schutz S, Grant D. Bacteriological and scanning electron microscopy (SEM) analyses of pancreatic stents [abstract]. Gastrointest Endosc 2001; 53:AB138. 22. Kozarek RA, Raltz SL, Brandabur JJ, Patterson DJ, Ball TJ, Gluck MG, et al. In vitro and in vivo application of a reusable double channel sphincterotome. Endoscopy 2001;33:401-4. 23. Bjornsson ES, Kilander AF, Olsson RG. Bile duct bacteria isolates in primary sclerosing cholangitis and certain other forms of cholestasis: a study of bile cultures from ERCP. Hepatogastroenterology 2000;47:1504-8. 24. van den Hazel SJ, Speelman P, Dankert J, Huibregtse K, Tytgat GNJ, Leeuwen DJ. Piperacillin to prevent cholangitis after endoscopic retrograde cholangiopancreatography: a randomized controlled trial. Ann Intern Med 1996;125:442-7. 25. Libby ED, Leung JW. Prevention of biliary stent clogging: a clinical review. Am J Gastroenterol 1996;91:1301-8. 26. Barrioz T, Ingrand P, Besson I, de Ledinghen V, Silvain C, Beauchant M. Randomized trial of prevention of biliary stent occlusion by ursodeoxycholic acid plus norfloxacin. Lancet 1994;344:581-2. 27. Leung JW, Liu YL, Desta TD, Libby ED, Inciardi JF, Lam K. In

VOLUME 58, NO. 4, 2003

R Kozarek, O Hovde, F Attia, et al.

28.

29.

30.

31.

32.

33.

34.

35.

36.

vitro evaluation of antibiotic prophylaxis in the prevention of biliary stent blockage. Gastrointest Endosc 2000;51:296-303. Ghosh S, Palmer KR. Prevention of biliary stent occlusion using cyclical antibiotics and ursodeoxycholic acid. Gut 1994;35:1757-9. Luman W, Ghosh S, Palmer KR. A combination of ciprofloxacin and Rowachol does not prevent biliary stent occlusion. Gastrointest Endosc 1999;49:316-21. Smit JM, Out MMJ, Groen AK, Huibregtse K, Jansen PLM, van Marce J, et al. A placebo-controlled study on the efficacy of aspirin and doxycycline in preventing clogging of biliary endoprostheses. Gastrointest Endosc 1989;35:485-9. Tsang TK, Pollack J, Chodash HB. Inhibition of biliary endoprostheses occlusion by ampicillin-sulbactam in an in vitro model. J Lab Clin Meth 1997;130:643-8. Lee RM, Kozarek RA, Sumida SE, Raltz SL. Risk of contamination of sterile biopsy forceps in disinfected endoscopes. Gastrointest Endosc 1998;47:377-81. Kozarek RA, Attia FM, Sumida SE, Raltz SL, Roach SK, Schembre DB, et al. Reusable biopsy forceps: a prospective evaluation of cleaning capabilities, function, adequacy of tissue specimen, and durability. Gastrointest Endosc 2001;53:747-50. Kozarek RA, Raltz SL, Ball TJ, Patterson DJ, Brandabur JJ. Reuse of disposable sphincterotomes for diagnostic and therapeutic ERCP: a one-year prospective study. Gastrointest Endosc 1999;49:39-42. Ikenberry SO, Sherman S, Hawes RH, Smith M, Lehman GA. The occlusion rate of pancreatic stents. Gastrointest Endosc 1994;40:611-3. Rerknimitr R, Fogel EL, Kalayci C, Esber E, Lehman GA, Sherman S. Microbiology of bile in patients with cholangitis or cholestasis with and without plastic biliary endoprosthesis. Gastrointest Endosc 2002;56:885-9.

GASTROINTESTINAL ENDOSCOPY

509