- Email: [email protected]
Pathogenesis of pancreatic sepsis
Pathogenesis of Pancreatic Sepsis David S. Medich, MO, Thomas K. Lee, Mo, Mona F. Melhem, MD, Marc I. Rowe, Mo, Wolfgang H. Schraut, MD, Kenneth K. W. Lee, MD, Pittsburgh,Pennsylvania
Although pancreatic sepsis is the most c o m m o n cause of major morbidity and mortality associated with acute pancreatitis, the pathogenesis of such infectious is u n k n o w n . Since intraperitoneal foci of inflammation are k n o w n to promote bacterial translocation, we hypothesized that acute pancreatitis promotes bacterial translocation that leads to infection of the inflamed pancreas and peripancreatic tissues. Non-lethal acute pancreatitis was induced in rats, and the trauslocation of live bacteria to the pancreas, mesenteric lymph nodes, liver, and spleen was determined. The presence of orally fed fluorescent beads, sensitive inert markers of translocation, was also determined in the pancreas and mesenterie lymph nodes. Live bacteria were recovered from 33% of the pancreata of rats with acute pancreatitis but from n o n e of the control rats. Beads were visualized in 9 I % of the pancreata of rats with acute pancreatitis but in n o n e of the pancreata from control rats. Beads were not visualized in the mesenteric lymph nodes of rats with acute pancreatitis, suggesting a transperitoneal route of migration. We conclude that acute panereatitis promotes bacterial transloeation leading to transperitoneal infection of the pancreas. These results support the use of selective decontamination of the gut and peritoneal lavage for the prevention of pancreatic infections in acute panereatitis.
ypically, acute pancreatitis is a mild disease that reT solves without major sequelae. However, in approximately 10% of patients, a more severe disease course develops and is associated with significant morbidity and mortality. As advances in critical care have greatly reduced the incidence of death caused by the early cardiopulmonary sequelae of fulminant pancreatitis, infections of the inflamed pancreas and surrounding tissues (collectively referred to as "peripancreatic infections") have emerged as the most common cause of major morbidity and mortality among these patients [1,2]. The pathophysiology of such infections is not clearly defined, and, therefore, the development and implementation of specific measures for their prevention have not been possible. The microbiology of peripanereatic infections reveals most pathogens to be common gastrointestinal flora and provides indirect evidence that the gut is a source of sepsis in acute pancreatitis [3-5]. Direct evidence of this phenomenon is lacking, however. Translocation of organisms from the gastrointestinal tract to extraintestinal sites is known to be promoted by factors that cause systemic insult or bowel injury [6]. Several studies have demonstrated that intra-abdominal inflammation, such as occurs during acute pancreatitis, promotes bacterial translocation in the absence of obvious histologic injury of the intestine. We, therefore, hypothesized that pancreatitis-related infections result from heightened translocation of enteric organisms into the region of the inflamed or injured pancreas. To test this hypothesis, we studied the translocation of bacteria to the pancreas, mesenteric lymph nodes, liver, and spleen of rats after the induction of non-lethal acute pancreatitis by means of intravenous infusion of cerulein. As a more sensitive marker of translocation whose origin from the gastrointestinal tract could be guaranteed, we also used orally fed inert fluorescent latex beads and studied their recovery from the pancreas and the mesenteric lymph nodes. MATERIALS A N D M E T H O D S
Specific pathogen-free male Sprague-Dawley rats (Harlan Sprague-Dawley, Indianapolis, IN) weighing 150 to 225 g were used in all experiments. The animals were housed and cared for at the University of Pittsburgh animal facility in accordance with the requirements for humane animal care as stipulated by the United States Departments of Agriculture and Health and Human Services. All protocols were approved by the institutional Fromthe Departmentsof Surgery(DSM,TKL,MIR, WHS,KKWL) animal care committee and complied with the requireand Pathology(MFM), Universityof PittsburghSchoolof Medicine, ments for animal care as stipulated by this committee. Pittsburgh,Pennsylvania. Animals were housed in the animal facility for at least 7 Requests for reprints shouldbe addressedto KennethK. W. Lee, days prior to use in order to stabilize their intestinal flora. MD, Departmentof Surgery,497 ScaifeHall,UniversityofPittsburgh, All procedures were performed with the rats receiving Pittsburgh,Pennsylvania15261. Presentedat the 33rdAnnualMeetingofthe Societyfor Surgeryof light methoxyflurane anesthesia. Twenty-two gauge inthe AlimentaryTract, San Francisco,California,May 11-13, 1992. travenous catheters were inserted aseptically into tail 46
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
veins and secured in place with adhesive tape and cylindrical metal tail vein holders, which allowed free movement. Phlebotomy was performed by cardiac puncture. Acute non-lethal pancreatitis was induced by a 12hour continuous intravenous infusion of cerulein (Sigma Chemical Co., St. Louis, MO) administered at a dose of 60 #g/kg and infused in 12 mL of saline at a rate of 1 mL/hr. This model predictably produces acute pancreatitis with characteristic histologic and biochemical changes [7] and avoids the possibility of peritoneal or pancreatic contamination attendant with other invasive models of acute pancreatitis, such as pancreatic duct infusion or ligation. Control rats received a comparable volume of sterile saline infused at the same rate. After the infusion of the rats was completed, phlebotomy was performed to measure serum amylase levels. Two experimental series were conducted. In the first, laparotomy was performed 12 hours after completion of the infusion, and the peritoneal cavity was swabbed to rule out contamination. Portal vein blood was then obtained and assayed for endotoxin, using a limulus lysate assay (Whittaker Products, Watersville, MD), since portal vein endotoxin is a sensitive indicator of portal vein bacteremia. The rats were killed, and portions of the pancreas, mesenteric lymph nodes, liver, spleen, and terminal ileum were harvested, weighed, and plated for the quantitative culture of aerobic and anaerobic organisms using methods previously described [8]. Briefly, gramnegative bacteria were identified using MacConkey agar supplemented with 10% lactose (Scott Co., West Warwick, RI). Gram-positive bacteria were identified using Columbia agar (BBL, Cockeysville, SC), and anaerobes were identified using Schaedlers agar (Difco, Detroit, MI). Tissue samples were prepared for routine light microscopy, using hematoxylin and eosin staining, and examined by a pathologist who did not know the source of the specimens. In the second experiment, fluorescent latex beads (0.9 urn) (Polyscience, Inc, Warrington, PA) were added to the drinking water of experimental animals at a concentration of 107 beads/mL beginning 48 hours prior to initiation of the cerulein or saline infusion and continuing until the completion of the experiments. These beads have previously been shown to be sensitive nonabsorbable markers of translocation [9,10]. Twelve hours after the completion of the infusion, the animals were killed, and the pancreas and mesenteric lymph nodes were excised and examined by a pathologist who did not know the source of the specimens using fluorescent microscopy with phase contrast to detect the presence or absence of beads. The incidence of bacterial and bead translocation in pancreatic and control animals was compared using x 2 analysis. Amylase and endotoxin levels were analyzed using the Student's t-test, p Values of <0.05 were considered significant.
RESULTS All rats survived until the completion of the experiments. Serum amylase levels after the 12-hour infusion of either cerulein or saline were 78,012 4- 10,405 U/mL
TABLE I Incidence of Bacterial Translocation
Control (n = 6)
Acute pancreatitis (n = 12)
0 0 0 0
4 (33%)* 2 (17%)* 0 0
Pancreas Mesenteric lymph nodes Liver Spleen *p <0,05 versus control.
TABLE II Organisms Cultured From Animals With Acute Pancreatitis Mesenteric Lymph Nodes
Pancreas Enterococcus* (~-Streptococcus (not Group D) c~-Streptococcus (not Group D) Lactobacillus
Enterococcus* Lactobacillus
*Positive culture for Enterococcus in pancreas and mesenteric lymph nodes of the same animal.
TABLE III Quantitative Terminal Ileum Flora ( l o g C F U / g )
Bacteria
Control (n = 6)
Acute pancreatitis (n = 12)
Gram-negative Gram-positive Anaerobic
7.3 _+ 0.8 6.5 _+ 1.2 4.6 _+ 1.0
7.1 _+ 1.0 6.1 _+ 0.8 5.1 +_ 1.5
CFU = colony-forming units.
TABLE IV Translocation of Beads
Pancreas Mesenteric lymph nodes
Control (n = 7)
Acute Pancreatitis (n = 11)
0 0
10 (91%)* 0
*p < 0.05 versus control,
(SEM) and 1,576 4- 412 U/mL (p <0.01), respectively, confirming the reproducibility of this model of acute pancreatitis. Pancreas specimens that were examined with routine light microscopy during autopsy revealed the characteristic changes of acute pancreatitis, including pancreatic edema, cellular infiltration, single-cell necrosis, and acinar vacuolization. Tables I and II show the incidence of the recovery of live bacteria from the animals with acute pancreatitis and
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
47
MEDICH ET AL
Figure 1. Photomicrograph of pancreas specimen from animal with acute pancreatitis, utilizing fluorescence with phase contrast microscopy. Fluorescent beads are visualized within the peripancreatic edema fluid. Note on the far right that three beads are contained within a phagocytic cell.
the control animals and the specific organisms that were cultured. Bacterial translocation was found only in the rats with acute pancreatitis. Four rats with acute pancreatitis had pancreas cultures that were positive for organisms commonly found in the gastrointestinal tract of the rat; in one of these rats, a mesenteric lymph node culture was concurrently obtained that was positive for identical pathogens. One other rat with acute pancreatitis had a mesenteric lymph node culture that was positive but its pancreas culture was not. Bacterial translocation was not found in any of the six control animals. Portal vein endotoxin levels as measured by the limulus lysate assay were not significantly different in rats with acute pancreatitis or the control rats (0.266 4-0.200 U / m L versus 0.247 4- 0.120 U/mL), nor did the flora of the terminal ileum differ between rats with acute pancreatitis and the control rats (Table m ) . Light microscopic examination did not reveal any structural histologic abnormalities of the small intestine in rats with acute pancreatitis. The detection of the fluorescent beads in the pancreata and mesenteric lymph nodes of control rats and rats with acute pancreatitis issummarized in Table IV. In control animals, fluorescent beads were visualized only within and never outside of the gastrointestinal tract. In particular, no beads were identified in the pancreatic or peripancreatic tissues nor in mesenteric lymph nodes. In contrast, fluorescent beads were found in the pancreatic and, most predominantly, in the peripancreatic tissues in 10 of 11 rats with acute pancreatitis. Figure 1 demonstrates the fluorescent beads in the peripancreatic edema fluid. Some beads could be identified within phagocytic cells that resembled macrophages located in the edematous peripancreatic region. Beads were observed infrequently within the pancreatic parenchyma itself. No beads, however, were detected within the mesenteric lymph nodes of the rats with acute pancreatitis. 48
COMMENTS Peripancreatic infection occurs in only 1% to 5% of patients with acute pancreatitis but accounts for up to 80% of the mortality in such patients [11]. In recent years, modalities such as dynamic computed tomography and image-guided percutaneous aspiration have led to the earlier diagnosis of such infections, which permits more timely initiation of medical and surgical treatment [12,13]. Newer treatment approaches include early and repeated pancreatic necrosectomy, abscess drainage using open packing or closed sump drainage techniques, and percutaneous drainage of peripancreatic fluid collections. Despite these improvements, peripancreatic infections continue to be associated with a mortality rate of 10% to 59% and a major morbidity rate of 60% to 93% [11,14]. The microbiology of the peripancreatic sepsis that complicates acute pancreatitis has been well characterized as common gastrointestinal flora. In a series of 45 patients, Beger et al [3] found that gram-negative intestinal bacteria were most commonly cultured. Escherichia coli was cultured from 24 patients, Enterobacter aerogenes from 16, Enterococcus species from 6, and Pseudomonas aeruginosa and Proteus species from 5 patients each. Bacteroides species were isolated from five patients. Similarly, Holden et al [15] studied 28 patients with pancreatic abscesses and found E. coli, Klebsiella species, and Bacteroides species to be the most common isolates during culturing. Although these findings strongly suggest that the gastrointestinal tract is the source of these infections, few studies have attempted to define the pathogenesis of peripancreatic sepsis to confirm this conclusion. The results of our study confirm the hypothesis that acute pancreatitis promotes translocation of gastrointestinal organisms to the inflamed pancreas and peripancreatic region. Unlike the results that have been found in humans, no gram-
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
negative bacilli were recovered in either the pancreas or mesenteric lymph nodes of the rats with acute pancreatitis. However, the assessment of ileal flora demonstrated that the organisms isolated from the pancreata were common gastrointestinal flora present in high concentrations in the terminal ileum (at least 106 colony-forming units per g of ileum tissue). Our model does not induce pancreatitis of such severity so as to lead to the development of pancreatic and peripancreatic necrosis. Therefore, we can only assume that the bacteria that were translocated to the inflamed pancreas would be the harbinger of infection at this site. Further experimental studies are planned to support these assumptions by direct observations. Recently, Runkel et al [16] published data supportive of the hypothesis that peripancreatic infections are caused by translocated bacteria. In a duct ligation model of acute pancreatitis in the rat, they were able to demonstrate infection of extraintestinal sites in 4 of 12 rats (1 positive pancreas culture, 3 positive mesenteric lymph node cultures). However, it must be recognized that obstructive jaundice, such as results in this model of pancreatitis, may itself promote bacterial translocation [17]. More convincing evidence in support of the hypothesis that peripancreatic sepsis is derived from the gut came from Widdison and coworkers [18]. Using a feline model of acute necrotizing pancreatitis, they showed that radiolabeled E. coli placed into the colon was recovered from 67% (6 of 9) of pancreatic pancreata compared with 11% (2 of 19) of controls. They also noted that E. coli was not recovered from the pancreas when the colon was enclosed in an impermeable bag that prohibited transperitoneal spread of bacteria. Several studies support the hypothesis that intraperitoneal inflammation, such as that associated with acute pancreatitis, promotes bacterial translocation and that the site of inflammation is the preferred site of colonization by translocating organisms. Schweinburg et al [19] demonstrated that, after the administration of an irritant solution into the peritoneal cavity, radiolabeled E. coli administered by mouth appeared in the peritoneal cavity but not in the blood. Mainous et al [20] reported increased dose-dependent bacterial translocation after intraperitoneal injection of zymosan, a potent activator of neutrophils, macrophages, and the alternative complement pathway, and Wells et al [21] showed that enteric facultative bacteria are translocated to the intraperitoneal, but not subcutaneous, fibrinogen clots contaminated with Bacteroides species. More recently, Mora et al [22] have shown that enteric bacteria and inert fluorescent beads can be recovered from sterile endotoxin-free prosthetic materials placed intraperitoneally but not subcutaneously. They hypothesized that bacterial translocation is induced by the inflammatory response provoked by the prosthetic material. Complement consumption was observed in an experimental model of acute hemorrhagic pancreatic necrosis [23,24], suggesting that inflammation occurring in acute pancreatitis similarly induces bacterial translocation. The mechanism(s) by which intraperitoneal inflammation promotes bacterial translocation has not yet been defined but may involve the release of
inflammatory mediators or cytokines; alternatively, although we did not observe histologic changes in the intestine, intraperitoneal inflammation may directly injure the intestine and disrupt intestinal barrier function. The possible role of other known promoters of bacterial translocation has not been fully studied in acute pancreatitis. However, our data indicated that increased translocation does not result from either bacterial overgrowth or endotoxemia. Tumor necrosis factor levels were not found to be elevated (data not shown). Our finding that fluorescent beads were recovered from the pancreas but not mesenteric lymph nodes of rats with acute pancreatitis supported a transperitoneal, rather than lymphatic, route of infection by gut-derived organisms. This route of infection is also supported by the observation by Widdison et al [18] that pancreatic infections can be decreased if the colon is enclosed in an impermeable bag [18]. Our failure to detect portal vein endotoxin, an indicator of bacteremia, indicates that there is not a hematogenous route of infection. The mechanism by which bacteria or inert particles reach and localize to the pancreas is unknown; however, extensive evidence exists that supports the idea that bacteria can translocate as viable organisms within macrophages [25-28]. The localization of phagocytic cells containing bacteria or inert beads to specific sites of inflammation may be explained by the establishment of chemotactic gradients around the sites of inflammation. Mora et al [22], for example, reported that intraperitoneal biomaterials triggered the alternative pathway of the complement cascade in vitro, thereby liberating the chemotactic substances, C3a and C5a. Since acute pancreatitis also causes activation of complement pathways and produces complement degradation products, this mechanism may account for the development of peripancreatic sepsis in acute pancreatitis and our finding that inert beads lacking intrinsic motility localize to the pancreas after induction of acute pancreatitis. Our finding of fluorescent beads contained within phagocytic cells in peripancreatic fluid is also consistent with such a mechanism. Nevertheless, further studies are needed to confirm this route of infection and to determine !f hematogenous or lymphatic pathways also contribute to peripancreatic infections in acute pancreatitis as has been suggested by other investigators [18,29,30]. The proposed concept of heightened translocation being the mechanism leading to peripancreatic infection suggests that two therapies, selective bowel decontamination and peritoneal lavage, may be useful for the prevention of such infections complicating acute pancreatitis. Selective bowel decontamination has gained acceptance as a means of preventing nosocomial infections among seriously ill patients, although its efficacy in reducing their mortality has not been demonstrated in clinical trials [31,32]. However, few patients in these trials had severe acute pancreatitis or developed peripancreatic sepsis, and, consequently, no data are available pertaining to the influence of bowel decontamination on the incidence of peripancreatic infections. Clinical trials of short courses of peritoneal lavage
THE AMERICANJOURNALOF SURGERY VOLUME165 JANUARY1993 49
MEDICH ET AL
have yielded conflicting results [33,34]. Recently, however, Ranson and B e r m a n [35] reported that patients with severe pancreatitis who received long courses of peritoneal lavage have a decreased frequency of peripancreatic sepsis c o m p a r e d with patients who received either a short course of peritoneal lavage or none [35]. Although the efficacy of peritoneal lavage in preventing peripancreatic sepsis has generally been attributed to removal of substances ("toxins") such as activated proteolytic enzymes present in the peritoneal exudate, the removal of phagocytic cells containing gut-derived bacteria, chemotactic factors, or unidentified inflammatory mediators that promote bacterial translocation m a y be additional effects consistent with the observations of this study. The greater effectiveness of prolonged peritoneal lavage is readily understandable, since translocation m a y be an ongoing process driven by continuing pancreatic inflammation. In s u m m a r y , our results demonstrate that acute pancreatitis leads to heightened translocation of gastrointestinal organisms and that these organisms subsequently infect pancreatic and peripancreatic tissues. Our results also support a transperitoneal route o f peripancreatic infection by gut-derived organisms. These findings suggest that selective decontamination of the bowel or peritoneal lavage m a y be useful for the prevention of such infections. REFERENCES 1. Prey CF, Bradley EL, Beger HG. Progress in acute pancreatitis. Surg Gynecol Obstet 1988; 167: 282-6. 2. Renner IG, Savage WT, Pantoja JL, Rennar VJ. Death due to acute pancreatitis: a retrospective analysis of 405 autopsy cases. Dig Dis Sci 1985; 30: 1005-18. 3. Beger HG, Bittner R, Block S, Buchler M. Bacterial contamination of pancreatic necrosis. Gastroenterology 1986; 91: 433-8. 4. Becker JM, Pemberton JH, DiMagio EP, et al. Diagnostic factors in pancreatic abscess. Surgery 1984; 96: 455-60. 5. Bradley EL, Aller K. A prospective longitudinal study of obstruction versus surgical intervention in the management of necrotizing pancreatitis. Am J Surg 1991; 161: 19-25. 6. Wells CL, Maddaus MA, Simmons RL. Proposed mechanisms for the translocation of intestinal bacteria. Rev Infect Dis 1988; 10: 958-78. 7. Lampel M, Kern H. Acute interstitial pancreatitis in the rat induced by excessive doses of a pancreatic secretagogue. Virchows Arch 1977; 373: 97-117. 8. Jackson R J, Smith SD, Rowe MI. Selective bowel decontamination results in gram positive translocation. J Surg Res 1990; 48: 444-7. 9. Mora EM, Cardona MA, Simmons R L Enteric bacteria and ingested inert particles translocate to intraperitoneal prosthetic material. Arch Surg 1991; 126: 157-63. 10. Wells CL, Maddaus MA, Erlandsen SL, Simmons RL. Evidence for the phagocytic transport of intestinal particles in dogs and rats. Infect Immun 1988; 56: 278-82. 11. Lumson A, Bradley EL. Secondary pancreatic infections. Surg Gynecol Obstet 1990; 170:459-67 12. Bradley EL, Murphy F, Ferguson C. Prediction of necrosis by dynamic pancreatography. Ann Surg 1989; 210: 485-504. 13. Banics PA, Geezof SG, Chong FK, et al. Bacteriologic status of necrotic tissue in necrotizing pancreatitis. Pancreas i 990; 5: 330-3. 14. Bradley EL, Fulenwider JT. Open treatment of pancreatic
50
abscess. Surg Gynecol Obstet 1984; 159: 509-13. 15. Holden JL, Berne TV, Rosoff L. Pancreatic abscess following acute pancreatitis. Arch Surg 1976; 111: 858-61. 16. Runkel NSF, Moody FG, Smith GS. et al. The role of the gut in the development of sepsis in acute pancreatitis. J Surg Res 1991; 51:18-23 17. Deitch EA, Sittig K, Li M, et al. Obstructive jaundice promotes bacterial translocation from the gut. Am J Surg 1990; 159: 79-84. 18. Widdison AL. Karanjia ND, Reber HA. Route(s) of spread of bacteria in acute necrotizing pancreatitis [abstract]. Pancreas 1990; 5: 736. 19. Schweinburg FB, Sdigman AM, Fine J. Transmural migration of intestinal bacteria: a study based on the use of radioactiveE. coll. N Engl J Med 1980; 242: 747-51. 20. Mainous MR, Tso P, Berg RD, Deitch EA. Studies of the route, magnitude, and time course of bacterial translocation in a model of systemic inflammation. Arch Surg 1991; 126: 33-7. 21. Wells CL, Rotstein OD, Pruett TL, Simmons RL. Intestinal bacteria translocate into experimental intra-abdominal abscesses. Arch Surg 1986; 121: 102-7. 22. Mora EM, Cardona MA, Simmons RL. Enteric bacteria and ingested inert particles translocate to intraperitoneal prosthetic material. Arch Surg 1991; 126: 157-63. 23. Kelly RH, Rao KN, Harvey VS, Lombardi B. Acute hemorrhagic pancreatic necrosis In mice: lack of a pathogenetic role for complement. Acta Hep Gastro 1979; 26: 302-9. 24. Kelly RH, Rao K N . Harvey VS. Lombardi B. Acute hemorrhagic pancreatic necrosis in mice: alterations of serum complement. Digestion 1981: 22: 1-7. 25. Wells CL, Maddaus MA, Jechorek RP, Simmons RL. Ability of intestinal Escherichia coli to survive within mesenteric lymph nodes. Infect Immun 1987; 55: 2834-7. 26. Wells CL, Maddaus MA, Simmons RL. Role of the macrophage in the translocation of intestinal bacteria. Arch Surg 1987; 122: 48-53. 27. Wells CL, Maddaus MA, Simmons RL. Bacterial translocation. In: Marston A, Bulkley G, Fiddian-Green RG, Hagbind VH, editors. Splanchnic ischemia and multiple organ failure. London: Edward Arnold, 1989: 195-204. 28. Wells CL, Maddaus MA, Simmons RL. Proposed mechanisms for the translocation of intestinal bacteria. Rev Infect Dis 1988; 10: 958-79. 29. Widdison AL, Karanjia ND, Alvarez C, Reber HA. Sources of pancreatic pathogens in acute necrotizing pancreatitis [abstract]. Gastroenterology 199i; 100: A304. 30. Webster MW, Pasculle AW, Myerowitz RL, et al. Postinduction bacteremia in experimental acute pancreatitis. Am J Surg 1979; 138: 418-20. 31. Cerra FB, Maddaus MA, Dunn DL, et al. Selective gut decontarnination reduces nosocomial infections and length of stay but not mortality or organ failure in surgical intensive care unit patients. Arch Surg 1992; 127: 163-9. 32. Gastinne H, Wolff M, Delatour F, et al. A controlled trial in intensive care units of selective decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Med 1992; 326: 594-9. 33. Mayer DA, McMahon M J, Corfield AP, et al. Controlled clinical trial of peritoneal lavage for the treatment of severe acute pancreatitis. N Engl J Meal 1985; 312: 399-404. 34. Stone HH, Fabian TC. Peritoneal dialysis in the treatment of acute alcoholic pancreatitis. Surg Gynecol Obstet 1980; 150: 878-82. 35. Ranson JHC, Berman RS. Long peritoneal lavage decreases pancreatic sepsis in acute pancreatitis. Ann Surg 1990; 211: 708-18.
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
DISCUSSION Edward L. Bradley (Atlanta, GA): Dr. Medich, the determination of the actual pathway of the beads to the pancreas is critical to any attempts at prophylaxis against infection. I believe that you are suggesting that the pathway is through the bowel wall into the peritoneum, and then, in some way, to the peripancreatic tissue, perhaps by macrophages. Do you have any evidence that these beads were in the peritoneal cavity in the ascitic fluid or that you could recover them from the peritoneal cavity? Another potential mechanism could be through the transverse colon mesentery to the surface of the pancreas and then into the surrounding peripancreatic fat. Since you administered the beads orally, the beads could also have migrated in a retrograde manner through the pancreatic duct. Were any beads found in the ductal system, or were they were all in the parenchyma of the pancreas? Did you make any attempt to differentiate the two sites? David S. Medich: To address your first question, we have done additional work since this study was completed in which we used cerulein to induce pancreatitis in animals that were fed beads. We then lavaged the abdomens of the rats and analyzed the peritoneal macrophages. Beads were found within the peritoneal macrophages. These are preliminary studies, and confirmatory studies are pending. Regarding the location of the beads, it was very difficult to find beads within the pancreatic parenchyma, and beads were never visualized within the pancreatic duct. The highest concentration of beads was in the peripancreatic edema fluid. Subsequent studies using antibodies to macrophages on this same preparation may confirm that the beads and macrophages are in the same location. John L. Cameron (Baltimore, MD): Dr. Medich, did you find beads in ascitic fluid from locations other than near the pancreas o r if you washed out the peritoneal cavity? David S. Medieh: Our study wasn't designed to answer that question. Robert E. Condon (Milwaukee, WI): Since you have already demonstrated that the route of migration is not through the lymphatic system, I think it's incumbent upon you to confirm that the route of migration is transperitoneal, rather than assuming that it is. The overgrowth of Candida species within the gut lumen is regularly associated with transmural migration of Candida organisms into the peritoneal cavity as well as into lymphatics and mesenteric tissues. Although migration does occur, it does not produce disease. It only produces disease when Candida species enter the bloodstream. Finally, is there evidence to indicate that gut sterilization alters this process or would have an impact on the infective complications of pancreatitis? David S. Medich: The use of Candida species overgrowth would be an ideal model to study. Concerning the use of oral decontamination, multiple
trials have shown a decrease in nosocomial infection in critically ill patients whose bowel was decontaminated, but no study has confirmed the impact of bowel decontamination on survival. In patients with pancreatitis, since survival is closely related to infection of the pancreas and peripancreatic tissue, I think that there is a better chance of demonstrating a benefit from oral decontamination. However, we have not yet undertaken such a study. Michael S. Nussbaum (Cincinnati, OH): In a recent study at our institution, a significant number of infections of patients with Candida species were documented both initially and on subsequent explorations in patients who underwent surgery for either pancreatic necrosis or pancreatic infection. These patients had no evidence of candidemia and, only rarely, other concomitant sites of Candida species infection. Thus, we hypothesized a role for translocation as the source of these infecting organisms. Do you have any information regarding the bacteriology of the positive cultures in your study? Did you identify any fungal species in any of these cultures? David S. Medich: None of our cultures were positive for the presence of fungus. The organisms that were cultured from the pancreas were enterococcus and nonGroup D a-streptococcus in two rats. Interestingly, no gram-negative organisms were cultured. However, the organisms that were cultured were all shown to be present in the gastrointestinal tract in high concentrations. H. R. Fretmd (Jerusalem, Israel): When your results are compared with those of other models of translocation, you had a very low rate of bacterial translocation. How many colony-forming units did you consider significant? How do you reconcile the discrepancy between the rate of bacterial translocation and the rate of translocation of a few beads? Your real objective was to show the bacterial translocation or infection rate of the pancreas. David S. Medich: Because our model of acute pancreatitis is not perfect, we were somewhat limited in our research. Beger has shown that the infectivity of the pancreas is most closely related with the degree of necrosis. In our model of acute pancreatitis using cerulein infusion, only isolated single-cell necrosis develops, despite the observed inflammation. We postulate that the reason for a higher incidence of translocation of the beads is because beads are a more sensitive marker and are easily recovered. We were unable to demonstrate a similar incidence of translocation of bacteria perhaps because of the culture techniques that are available and because the pancreatitis produced in our model is not sufficiently severe. R. S. Bennion (Los Angeles, CA): Did you consider altering the feed of the rats, specifically, the use of meat when feeding the study rats? We have found in several studies that, to increase the number of gram-negative and anaerobic organisms that are recovered from either translocation or infection due to ischemia, the use of meat as food is successful. Have you increased the length of time before killing
THE A M E R I C A N J O U R N A L OF SURGERY
VOLUME 165
J A N U A R Y 1993
51
MEDICH ET AL
the animals? You've indicated that the organisms recovered from the infected pancreata were gram-positive cocci only. Have you extended the length of the study to observe if other organisms are present? We found that gram-positive cocci, both aerobic and anaerobic, are often the first organisms to translocate and that the facultative and aerobic gram-negative rods and the anaerobic organisms will often grow later. David S. Medieh: The animals were fed a standard rat chow without meat throughout the experiment; therefore, I cannot comment on the possible effects of a meatcontaining diet on the recovery of gram-negative or anaerobic organisms. We are interested in the time course of these infections. We have also performed our studies immediately after, or 6 or 18 hours after, the completion of cerulein infusion and have observed the same results. Studies are planned in which the duration is changed, for example, to 6 hours. Andrew L. Warshaw (Boston, MA): If we accept your hypothesis of translocation and migration across the peritoneal cavity, can you speculate on why the beads migrated to the pancreas? Did you find beads or attempt to look for them in any other organs? Why did the beads go to the pancreas, not in macrophages but into the interstitial spaces of the gland? David S. Medieh: I don't have any data. Ranson's
52
T H E A M E R I C A N J O U R N A L OF SURGERY
study showed that lavage decreased the incidence of secondary peripancreatic infection. The use of lavage in patients with pancreatitis is often thought to be beneficial because it removes proteases. I would hypothesize that perhaps pancreatic inflammation produces agents that promote translocation or even chemotactic agents. Infusing the lavage from these animals into other animals would test that hypothesis. R. S. Jones: You studied two groups Of animals, experimental and control. The control group contained six animals. The experimental group contained 12 animals. You had two occurrences of translocation to lymph nodes in the experimental group and three or four occurrences of translocation to the pancreas. What statistical method did you employ to ensure that those occurrences were not random? Was this a statistically significant occurrence? If there were equal numbers in both groups, it would have been highly likely that your findings were significant. But since there were only half the number of control animals as experimental animals, the question of statistical significance is raised. David S. Medich: We used X 2 analysis to determine the significance. In other experiments in our laboratory, the recovery of live bacteria from mesenteric lymph nodes occurs in about 10% to 15% of normal animals. We've never had a positive culture from the pancreas of a control animal.
VOLUME 165
J A N U A R Y 1993
Although pancreatic sepsis is the most c o m m o n cause of major morbidity and mortality associated with acute pancreatitis, the pathogenesis of such infectious is u n k n o w n . Since intraperitoneal foci of inflammation are k n o w n to promote bacterial translocation, we hypothesized that acute pancreatitis promotes bacterial translocation that leads to infection of the inflamed pancreas and peripancreatic tissues. Non-lethal acute pancreatitis was induced in rats, and the trauslocation of live bacteria to the pancreas, mesenteric lymph nodes, liver, and spleen was determined. The presence of orally fed fluorescent beads, sensitive inert markers of translocation, was also determined in the pancreas and mesenterie lymph nodes. Live bacteria were recovered from 33% of the pancreata of rats with acute pancreatitis but from n o n e of the control rats. Beads were visualized in 9 I % of the pancreata of rats with acute pancreatitis but in n o n e of the pancreata from control rats. Beads were not visualized in the mesenteric lymph nodes of rats with acute pancreatitis, suggesting a transperitoneal route of migration. We conclude that acute panereatitis promotes bacterial transloeation leading to transperitoneal infection of the pancreas. These results support the use of selective decontamination of the gut and peritoneal lavage for the prevention of pancreatic infections in acute panereatitis.
ypically, acute pancreatitis is a mild disease that reT solves without major sequelae. However, in approximately 10% of patients, a more severe disease course develops and is associated with significant morbidity and mortality. As advances in critical care have greatly reduced the incidence of death caused by the early cardiopulmonary sequelae of fulminant pancreatitis, infections of the inflamed pancreas and surrounding tissues (collectively referred to as "peripancreatic infections") have emerged as the most common cause of major morbidity and mortality among these patients [1,2]. The pathophysiology of such infections is not clearly defined, and, therefore, the development and implementation of specific measures for their prevention have not been possible. The microbiology of peripanereatic infections reveals most pathogens to be common gastrointestinal flora and provides indirect evidence that the gut is a source of sepsis in acute pancreatitis [3-5]. Direct evidence of this phenomenon is lacking, however. Translocation of organisms from the gastrointestinal tract to extraintestinal sites is known to be promoted by factors that cause systemic insult or bowel injury [6]. Several studies have demonstrated that intra-abdominal inflammation, such as occurs during acute pancreatitis, promotes bacterial translocation in the absence of obvious histologic injury of the intestine. We, therefore, hypothesized that pancreatitis-related infections result from heightened translocation of enteric organisms into the region of the inflamed or injured pancreas. To test this hypothesis, we studied the translocation of bacteria to the pancreas, mesenteric lymph nodes, liver, and spleen of rats after the induction of non-lethal acute pancreatitis by means of intravenous infusion of cerulein. As a more sensitive marker of translocation whose origin from the gastrointestinal tract could be guaranteed, we also used orally fed inert fluorescent latex beads and studied their recovery from the pancreas and the mesenteric lymph nodes. MATERIALS A N D M E T H O D S
Specific pathogen-free male Sprague-Dawley rats (Harlan Sprague-Dawley, Indianapolis, IN) weighing 150 to 225 g were used in all experiments. The animals were housed and cared for at the University of Pittsburgh animal facility in accordance with the requirements for humane animal care as stipulated by the United States Departments of Agriculture and Health and Human Services. All protocols were approved by the institutional Fromthe Departmentsof Surgery(DSM,TKL,MIR, WHS,KKWL) animal care committee and complied with the requireand Pathology(MFM), Universityof PittsburghSchoolof Medicine, ments for animal care as stipulated by this committee. Pittsburgh,Pennsylvania. Animals were housed in the animal facility for at least 7 Requests for reprints shouldbe addressedto KennethK. W. Lee, days prior to use in order to stabilize their intestinal flora. MD, Departmentof Surgery,497 ScaifeHall,UniversityofPittsburgh, All procedures were performed with the rats receiving Pittsburgh,Pennsylvania15261. Presentedat the 33rdAnnualMeetingofthe Societyfor Surgeryof light methoxyflurane anesthesia. Twenty-two gauge inthe AlimentaryTract, San Francisco,California,May 11-13, 1992. travenous catheters were inserted aseptically into tail 46
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
veins and secured in place with adhesive tape and cylindrical metal tail vein holders, which allowed free movement. Phlebotomy was performed by cardiac puncture. Acute non-lethal pancreatitis was induced by a 12hour continuous intravenous infusion of cerulein (Sigma Chemical Co., St. Louis, MO) administered at a dose of 60 #g/kg and infused in 12 mL of saline at a rate of 1 mL/hr. This model predictably produces acute pancreatitis with characteristic histologic and biochemical changes [7] and avoids the possibility of peritoneal or pancreatic contamination attendant with other invasive models of acute pancreatitis, such as pancreatic duct infusion or ligation. Control rats received a comparable volume of sterile saline infused at the same rate. After the infusion of the rats was completed, phlebotomy was performed to measure serum amylase levels. Two experimental series were conducted. In the first, laparotomy was performed 12 hours after completion of the infusion, and the peritoneal cavity was swabbed to rule out contamination. Portal vein blood was then obtained and assayed for endotoxin, using a limulus lysate assay (Whittaker Products, Watersville, MD), since portal vein endotoxin is a sensitive indicator of portal vein bacteremia. The rats were killed, and portions of the pancreas, mesenteric lymph nodes, liver, spleen, and terminal ileum were harvested, weighed, and plated for the quantitative culture of aerobic and anaerobic organisms using methods previously described [8]. Briefly, gramnegative bacteria were identified using MacConkey agar supplemented with 10% lactose (Scott Co., West Warwick, RI). Gram-positive bacteria were identified using Columbia agar (BBL, Cockeysville, SC), and anaerobes were identified using Schaedlers agar (Difco, Detroit, MI). Tissue samples were prepared for routine light microscopy, using hematoxylin and eosin staining, and examined by a pathologist who did not know the source of the specimens. In the second experiment, fluorescent latex beads (0.9 urn) (Polyscience, Inc, Warrington, PA) were added to the drinking water of experimental animals at a concentration of 107 beads/mL beginning 48 hours prior to initiation of the cerulein or saline infusion and continuing until the completion of the experiments. These beads have previously been shown to be sensitive nonabsorbable markers of translocation [9,10]. Twelve hours after the completion of the infusion, the animals were killed, and the pancreas and mesenteric lymph nodes were excised and examined by a pathologist who did not know the source of the specimens using fluorescent microscopy with phase contrast to detect the presence or absence of beads. The incidence of bacterial and bead translocation in pancreatic and control animals was compared using x 2 analysis. Amylase and endotoxin levels were analyzed using the Student's t-test, p Values of <0.05 were considered significant.
RESULTS All rats survived until the completion of the experiments. Serum amylase levels after the 12-hour infusion of either cerulein or saline were 78,012 4- 10,405 U/mL
TABLE I Incidence of Bacterial Translocation
Control (n = 6)
Acute pancreatitis (n = 12)
0 0 0 0
4 (33%)* 2 (17%)* 0 0
Pancreas Mesenteric lymph nodes Liver Spleen *p <0,05 versus control.
TABLE II Organisms Cultured From Animals With Acute Pancreatitis Mesenteric Lymph Nodes
Pancreas Enterococcus* (~-Streptococcus (not Group D) c~-Streptococcus (not Group D) Lactobacillus
Enterococcus* Lactobacillus
*Positive culture for Enterococcus in pancreas and mesenteric lymph nodes of the same animal.
TABLE III Quantitative Terminal Ileum Flora ( l o g C F U / g )
Bacteria
Control (n = 6)
Acute pancreatitis (n = 12)
Gram-negative Gram-positive Anaerobic
7.3 _+ 0.8 6.5 _+ 1.2 4.6 _+ 1.0
7.1 _+ 1.0 6.1 _+ 0.8 5.1 +_ 1.5
CFU = colony-forming units.
TABLE IV Translocation of Beads
Pancreas Mesenteric lymph nodes
Control (n = 7)
Acute Pancreatitis (n = 11)
0 0
10 (91%)* 0
*p < 0.05 versus control,
(SEM) and 1,576 4- 412 U/mL (p <0.01), respectively, confirming the reproducibility of this model of acute pancreatitis. Pancreas specimens that were examined with routine light microscopy during autopsy revealed the characteristic changes of acute pancreatitis, including pancreatic edema, cellular infiltration, single-cell necrosis, and acinar vacuolization. Tables I and II show the incidence of the recovery of live bacteria from the animals with acute pancreatitis and
THE AMERICAN JOURNAL OF SURGERY
VOLUME 165
JANUARY 1993
47
MEDICH ET AL
Figure 1. Photomicrograph of pancreas specimen from animal with acute pancreatitis, utilizing fluorescence with phase contrast microscopy. Fluorescent beads are visualized within the peripancreatic edema fluid. Note on the far right that three beads are contained within a phagocytic cell.
the control animals and the specific organisms that were cultured. Bacterial translocation was found only in the rats with acute pancreatitis. Four rats with acute pancreatitis had pancreas cultures that were positive for organisms commonly found in the gastrointestinal tract of the rat; in one of these rats, a mesenteric lymph node culture was concurrently obtained that was positive for identical pathogens. One other rat with acute pancreatitis had a mesenteric lymph node culture that was positive but its pancreas culture was not. Bacterial translocation was not found in any of the six control animals. Portal vein endotoxin levels as measured by the limulus lysate assay were not significantly different in rats with acute pancreatitis or the control rats (0.266 4-0.200 U / m L versus 0.247 4- 0.120 U/mL), nor did the flora of the terminal ileum differ between rats with acute pancreatitis and the control rats (Table m ) . Light microscopic examination did not reveal any structural histologic abnormalities of the small intestine in rats with acute pancreatitis. The detection of the fluorescent beads in the pancreata and mesenteric lymph nodes of control rats and rats with acute pancreatitis issummarized in Table IV. In control animals, fluorescent beads were visualized only within and never outside of the gastrointestinal tract. In particular, no beads were identified in the pancreatic or peripancreatic tissues nor in mesenteric lymph nodes. In contrast, fluorescent beads were found in the pancreatic and, most predominantly, in the peripancreatic tissues in 10 of 11 rats with acute pancreatitis. Figure 1 demonstrates the fluorescent beads in the peripancreatic edema fluid. Some beads could be identified within phagocytic cells that resembled macrophages located in the edematous peripancreatic region. Beads were observed infrequently within the pancreatic parenchyma itself. No beads, however, were detected within the mesenteric lymph nodes of the rats with acute pancreatitis. 48
COMMENTS Peripancreatic infection occurs in only 1% to 5% of patients with acute pancreatitis but accounts for up to 80% of the mortality in such patients [11]. In recent years, modalities such as dynamic computed tomography and image-guided percutaneous aspiration have led to the earlier diagnosis of such infections, which permits more timely initiation of medical and surgical treatment [12,13]. Newer treatment approaches include early and repeated pancreatic necrosectomy, abscess drainage using open packing or closed sump drainage techniques, and percutaneous drainage of peripancreatic fluid collections. Despite these improvements, peripancreatic infections continue to be associated with a mortality rate of 10% to 59% and a major morbidity rate of 60% to 93% [11,14]. The microbiology of the peripancreatic sepsis that complicates acute pancreatitis has been well characterized as common gastrointestinal flora. In a series of 45 patients, Beger et al [3] found that gram-negative intestinal bacteria were most commonly cultured. Escherichia coli was cultured from 24 patients, Enterobacter aerogenes from 16, Enterococcus species from 6, and Pseudomonas aeruginosa and Proteus species from 5 patients each. Bacteroides species were isolated from five patients. Similarly, Holden et al [15] studied 28 patients with pancreatic abscesses and found E. coli, Klebsiella species, and Bacteroides species to be the most common isolates during culturing. Although these findings strongly suggest that the gastrointestinal tract is the source of these infections, few studies have attempted to define the pathogenesis of peripancreatic sepsis to confirm this conclusion. The results of our study confirm the hypothesis that acute pancreatitis promotes translocation of gastrointestinal organisms to the inflamed pancreas and peripancreatic region. Unlike the results that have been found in humans, no gram-
THE AMERICAN JOURNAL OF SURGERY VOLUME165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
negative bacilli were recovered in either the pancreas or mesenteric lymph nodes of the rats with acute pancreatitis. However, the assessment of ileal flora demonstrated that the organisms isolated from the pancreata were common gastrointestinal flora present in high concentrations in the terminal ileum (at least 106 colony-forming units per g of ileum tissue). Our model does not induce pancreatitis of such severity so as to lead to the development of pancreatic and peripancreatic necrosis. Therefore, we can only assume that the bacteria that were translocated to the inflamed pancreas would be the harbinger of infection at this site. Further experimental studies are planned to support these assumptions by direct observations. Recently, Runkel et al [16] published data supportive of the hypothesis that peripancreatic infections are caused by translocated bacteria. In a duct ligation model of acute pancreatitis in the rat, they were able to demonstrate infection of extraintestinal sites in 4 of 12 rats (1 positive pancreas culture, 3 positive mesenteric lymph node cultures). However, it must be recognized that obstructive jaundice, such as results in this model of pancreatitis, may itself promote bacterial translocation [17]. More convincing evidence in support of the hypothesis that peripancreatic sepsis is derived from the gut came from Widdison and coworkers [18]. Using a feline model of acute necrotizing pancreatitis, they showed that radiolabeled E. coli placed into the colon was recovered from 67% (6 of 9) of pancreatic pancreata compared with 11% (2 of 19) of controls. They also noted that E. coli was not recovered from the pancreas when the colon was enclosed in an impermeable bag that prohibited transperitoneal spread of bacteria. Several studies support the hypothesis that intraperitoneal inflammation, such as that associated with acute pancreatitis, promotes bacterial translocation and that the site of inflammation is the preferred site of colonization by translocating organisms. Schweinburg et al [19] demonstrated that, after the administration of an irritant solution into the peritoneal cavity, radiolabeled E. coli administered by mouth appeared in the peritoneal cavity but not in the blood. Mainous et al [20] reported increased dose-dependent bacterial translocation after intraperitoneal injection of zymosan, a potent activator of neutrophils, macrophages, and the alternative complement pathway, and Wells et al [21] showed that enteric facultative bacteria are translocated to the intraperitoneal, but not subcutaneous, fibrinogen clots contaminated with Bacteroides species. More recently, Mora et al [22] have shown that enteric bacteria and inert fluorescent beads can be recovered from sterile endotoxin-free prosthetic materials placed intraperitoneally but not subcutaneously. They hypothesized that bacterial translocation is induced by the inflammatory response provoked by the prosthetic material. Complement consumption was observed in an experimental model of acute hemorrhagic pancreatic necrosis [23,24], suggesting that inflammation occurring in acute pancreatitis similarly induces bacterial translocation. The mechanism(s) by which intraperitoneal inflammation promotes bacterial translocation has not yet been defined but may involve the release of
inflammatory mediators or cytokines; alternatively, although we did not observe histologic changes in the intestine, intraperitoneal inflammation may directly injure the intestine and disrupt intestinal barrier function. The possible role of other known promoters of bacterial translocation has not been fully studied in acute pancreatitis. However, our data indicated that increased translocation does not result from either bacterial overgrowth or endotoxemia. Tumor necrosis factor levels were not found to be elevated (data not shown). Our finding that fluorescent beads were recovered from the pancreas but not mesenteric lymph nodes of rats with acute pancreatitis supported a transperitoneal, rather than lymphatic, route of infection by gut-derived organisms. This route of infection is also supported by the observation by Widdison et al [18] that pancreatic infections can be decreased if the colon is enclosed in an impermeable bag [18]. Our failure to detect portal vein endotoxin, an indicator of bacteremia, indicates that there is not a hematogenous route of infection. The mechanism by which bacteria or inert particles reach and localize to the pancreas is unknown; however, extensive evidence exists that supports the idea that bacteria can translocate as viable organisms within macrophages [25-28]. The localization of phagocytic cells containing bacteria or inert beads to specific sites of inflammation may be explained by the establishment of chemotactic gradients around the sites of inflammation. Mora et al [22], for example, reported that intraperitoneal biomaterials triggered the alternative pathway of the complement cascade in vitro, thereby liberating the chemotactic substances, C3a and C5a. Since acute pancreatitis also causes activation of complement pathways and produces complement degradation products, this mechanism may account for the development of peripancreatic sepsis in acute pancreatitis and our finding that inert beads lacking intrinsic motility localize to the pancreas after induction of acute pancreatitis. Our finding of fluorescent beads contained within phagocytic cells in peripancreatic fluid is also consistent with such a mechanism. Nevertheless, further studies are needed to confirm this route of infection and to determine !f hematogenous or lymphatic pathways also contribute to peripancreatic infections in acute pancreatitis as has been suggested by other investigators [18,29,30]. The proposed concept of heightened translocation being the mechanism leading to peripancreatic infection suggests that two therapies, selective bowel decontamination and peritoneal lavage, may be useful for the prevention of such infections complicating acute pancreatitis. Selective bowel decontamination has gained acceptance as a means of preventing nosocomial infections among seriously ill patients, although its efficacy in reducing their mortality has not been demonstrated in clinical trials [31,32]. However, few patients in these trials had severe acute pancreatitis or developed peripancreatic sepsis, and, consequently, no data are available pertaining to the influence of bowel decontamination on the incidence of peripancreatic infections. Clinical trials of short courses of peritoneal lavage
THE AMERICANJOURNALOF SURGERY VOLUME165 JANUARY1993 49
MEDICH ET AL
have yielded conflicting results [33,34]. Recently, however, Ranson and B e r m a n [35] reported that patients with severe pancreatitis who received long courses of peritoneal lavage have a decreased frequency of peripancreatic sepsis c o m p a r e d with patients who received either a short course of peritoneal lavage or none [35]. Although the efficacy of peritoneal lavage in preventing peripancreatic sepsis has generally been attributed to removal of substances ("toxins") such as activated proteolytic enzymes present in the peritoneal exudate, the removal of phagocytic cells containing gut-derived bacteria, chemotactic factors, or unidentified inflammatory mediators that promote bacterial translocation m a y be additional effects consistent with the observations of this study. The greater effectiveness of prolonged peritoneal lavage is readily understandable, since translocation m a y be an ongoing process driven by continuing pancreatic inflammation. In s u m m a r y , our results demonstrate that acute pancreatitis leads to heightened translocation of gastrointestinal organisms and that these organisms subsequently infect pancreatic and peripancreatic tissues. Our results also support a transperitoneal route o f peripancreatic infection by gut-derived organisms. These findings suggest that selective decontamination of the bowel or peritoneal lavage m a y be useful for the prevention of such infections. REFERENCES 1. Prey CF, Bradley EL, Beger HG. Progress in acute pancreatitis. Surg Gynecol Obstet 1988; 167: 282-6. 2. Renner IG, Savage WT, Pantoja JL, Rennar VJ. Death due to acute pancreatitis: a retrospective analysis of 405 autopsy cases. Dig Dis Sci 1985; 30: 1005-18. 3. Beger HG, Bittner R, Block S, Buchler M. Bacterial contamination of pancreatic necrosis. Gastroenterology 1986; 91: 433-8. 4. Becker JM, Pemberton JH, DiMagio EP, et al. Diagnostic factors in pancreatic abscess. Surgery 1984; 96: 455-60. 5. Bradley EL, Aller K. A prospective longitudinal study of obstruction versus surgical intervention in the management of necrotizing pancreatitis. Am J Surg 1991; 161: 19-25. 6. Wells CL, Maddaus MA, Simmons RL. Proposed mechanisms for the translocation of intestinal bacteria. Rev Infect Dis 1988; 10: 958-78. 7. Lampel M, Kern H. Acute interstitial pancreatitis in the rat induced by excessive doses of a pancreatic secretagogue. Virchows Arch 1977; 373: 97-117. 8. Jackson R J, Smith SD, Rowe MI. Selective bowel decontamination results in gram positive translocation. J Surg Res 1990; 48: 444-7. 9. Mora EM, Cardona MA, Simmons R L Enteric bacteria and ingested inert particles translocate to intraperitoneal prosthetic material. Arch Surg 1991; 126: 157-63. 10. Wells CL, Maddaus MA, Erlandsen SL, Simmons RL. Evidence for the phagocytic transport of intestinal particles in dogs and rats. Infect Immun 1988; 56: 278-82. 11. Lumson A, Bradley EL. Secondary pancreatic infections. Surg Gynecol Obstet 1990; 170:459-67 12. Bradley EL, Murphy F, Ferguson C. Prediction of necrosis by dynamic pancreatography. Ann Surg 1989; 210: 485-504. 13. Banics PA, Geezof SG, Chong FK, et al. Bacteriologic status of necrotic tissue in necrotizing pancreatitis. Pancreas i 990; 5: 330-3. 14. Bradley EL, Fulenwider JT. Open treatment of pancreatic
50
abscess. Surg Gynecol Obstet 1984; 159: 509-13. 15. Holden JL, Berne TV, Rosoff L. Pancreatic abscess following acute pancreatitis. Arch Surg 1976; 111: 858-61. 16. Runkel NSF, Moody FG, Smith GS. et al. The role of the gut in the development of sepsis in acute pancreatitis. J Surg Res 1991; 51:18-23 17. Deitch EA, Sittig K, Li M, et al. Obstructive jaundice promotes bacterial translocation from the gut. Am J Surg 1990; 159: 79-84. 18. Widdison AL. Karanjia ND, Reber HA. Route(s) of spread of bacteria in acute necrotizing pancreatitis [abstract]. Pancreas 1990; 5: 736. 19. Schweinburg FB, Sdigman AM, Fine J. Transmural migration of intestinal bacteria: a study based on the use of radioactiveE. coll. N Engl J Med 1980; 242: 747-51. 20. Mainous MR, Tso P, Berg RD, Deitch EA. Studies of the route, magnitude, and time course of bacterial translocation in a model of systemic inflammation. Arch Surg 1991; 126: 33-7. 21. Wells CL, Rotstein OD, Pruett TL, Simmons RL. Intestinal bacteria translocate into experimental intra-abdominal abscesses. Arch Surg 1986; 121: 102-7. 22. Mora EM, Cardona MA, Simmons RL. Enteric bacteria and ingested inert particles translocate to intraperitoneal prosthetic material. Arch Surg 1991; 126: 157-63. 23. Kelly RH, Rao KN, Harvey VS, Lombardi B. Acute hemorrhagic pancreatic necrosis In mice: lack of a pathogenetic role for complement. Acta Hep Gastro 1979; 26: 302-9. 24. Kelly RH, Rao K N . Harvey VS. Lombardi B. Acute hemorrhagic pancreatic necrosis in mice: alterations of serum complement. Digestion 1981: 22: 1-7. 25. Wells CL, Maddaus MA, Jechorek RP, Simmons RL. Ability of intestinal Escherichia coli to survive within mesenteric lymph nodes. Infect Immun 1987; 55: 2834-7. 26. Wells CL, Maddaus MA, Simmons RL. Role of the macrophage in the translocation of intestinal bacteria. Arch Surg 1987; 122: 48-53. 27. Wells CL, Maddaus MA, Simmons RL. Bacterial translocation. In: Marston A, Bulkley G, Fiddian-Green RG, Hagbind VH, editors. Splanchnic ischemia and multiple organ failure. London: Edward Arnold, 1989: 195-204. 28. Wells CL, Maddaus MA, Simmons RL. Proposed mechanisms for the translocation of intestinal bacteria. Rev Infect Dis 1988; 10: 958-79. 29. Widdison AL, Karanjia ND, Alvarez C, Reber HA. Sources of pancreatic pathogens in acute necrotizing pancreatitis [abstract]. Gastroenterology 199i; 100: A304. 30. Webster MW, Pasculle AW, Myerowitz RL, et al. Postinduction bacteremia in experimental acute pancreatitis. Am J Surg 1979; 138: 418-20. 31. Cerra FB, Maddaus MA, Dunn DL, et al. Selective gut decontarnination reduces nosocomial infections and length of stay but not mortality or organ failure in surgical intensive care unit patients. Arch Surg 1992; 127: 163-9. 32. Gastinne H, Wolff M, Delatour F, et al. A controlled trial in intensive care units of selective decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Med 1992; 326: 594-9. 33. Mayer DA, McMahon M J, Corfield AP, et al. Controlled clinical trial of peritoneal lavage for the treatment of severe acute pancreatitis. N Engl J Meal 1985; 312: 399-404. 34. Stone HH, Fabian TC. Peritoneal dialysis in the treatment of acute alcoholic pancreatitis. Surg Gynecol Obstet 1980; 150: 878-82. 35. Ranson JHC, Berman RS. Long peritoneal lavage decreases pancreatic sepsis in acute pancreatitis. Ann Surg 1990; 211: 708-18.
THE AMERICAN JOURNAL OF SURGERY VOLUME 165 JANUARY 1993
PATHOGENESIS OF PANCREATIC SEPSIS
DISCUSSION Edward L. Bradley (Atlanta, GA): Dr. Medich, the determination of the actual pathway of the beads to the pancreas is critical to any attempts at prophylaxis against infection. I believe that you are suggesting that the pathway is through the bowel wall into the peritoneum, and then, in some way, to the peripancreatic tissue, perhaps by macrophages. Do you have any evidence that these beads were in the peritoneal cavity in the ascitic fluid or that you could recover them from the peritoneal cavity? Another potential mechanism could be through the transverse colon mesentery to the surface of the pancreas and then into the surrounding peripancreatic fat. Since you administered the beads orally, the beads could also have migrated in a retrograde manner through the pancreatic duct. Were any beads found in the ductal system, or were they were all in the parenchyma of the pancreas? Did you make any attempt to differentiate the two sites? David S. Medich: To address your first question, we have done additional work since this study was completed in which we used cerulein to induce pancreatitis in animals that were fed beads. We then lavaged the abdomens of the rats and analyzed the peritoneal macrophages. Beads were found within the peritoneal macrophages. These are preliminary studies, and confirmatory studies are pending. Regarding the location of the beads, it was very difficult to find beads within the pancreatic parenchyma, and beads were never visualized within the pancreatic duct. The highest concentration of beads was in the peripancreatic edema fluid. Subsequent studies using antibodies to macrophages on this same preparation may confirm that the beads and macrophages are in the same location. John L. Cameron (Baltimore, MD): Dr. Medich, did you find beads in ascitic fluid from locations other than near the pancreas o r if you washed out the peritoneal cavity? David S. Medieh: Our study wasn't designed to answer that question. Robert E. Condon (Milwaukee, WI): Since you have already demonstrated that the route of migration is not through the lymphatic system, I think it's incumbent upon you to confirm that the route of migration is transperitoneal, rather than assuming that it is. The overgrowth of Candida species within the gut lumen is regularly associated with transmural migration of Candida organisms into the peritoneal cavity as well as into lymphatics and mesenteric tissues. Although migration does occur, it does not produce disease. It only produces disease when Candida species enter the bloodstream. Finally, is there evidence to indicate that gut sterilization alters this process or would have an impact on the infective complications of pancreatitis? David S. Medich: The use of Candida species overgrowth would be an ideal model to study. Concerning the use of oral decontamination, multiple
trials have shown a decrease in nosocomial infection in critically ill patients whose bowel was decontaminated, but no study has confirmed the impact of bowel decontamination on survival. In patients with pancreatitis, since survival is closely related to infection of the pancreas and peripancreatic tissue, I think that there is a better chance of demonstrating a benefit from oral decontamination. However, we have not yet undertaken such a study. Michael S. Nussbaum (Cincinnati, OH): In a recent study at our institution, a significant number of infections of patients with Candida species were documented both initially and on subsequent explorations in patients who underwent surgery for either pancreatic necrosis or pancreatic infection. These patients had no evidence of candidemia and, only rarely, other concomitant sites of Candida species infection. Thus, we hypothesized a role for translocation as the source of these infecting organisms. Do you have any information regarding the bacteriology of the positive cultures in your study? Did you identify any fungal species in any of these cultures? David S. Medich: None of our cultures were positive for the presence of fungus. The organisms that were cultured from the pancreas were enterococcus and nonGroup D a-streptococcus in two rats. Interestingly, no gram-negative organisms were cultured. However, the organisms that were cultured were all shown to be present in the gastrointestinal tract in high concentrations. H. R. Fretmd (Jerusalem, Israel): When your results are compared with those of other models of translocation, you had a very low rate of bacterial translocation. How many colony-forming units did you consider significant? How do you reconcile the discrepancy between the rate of bacterial translocation and the rate of translocation of a few beads? Your real objective was to show the bacterial translocation or infection rate of the pancreas. David S. Medich: Because our model of acute pancreatitis is not perfect, we were somewhat limited in our research. Beger has shown that the infectivity of the pancreas is most closely related with the degree of necrosis. In our model of acute pancreatitis using cerulein infusion, only isolated single-cell necrosis develops, despite the observed inflammation. We postulate that the reason for a higher incidence of translocation of the beads is because beads are a more sensitive marker and are easily recovered. We were unable to demonstrate a similar incidence of translocation of bacteria perhaps because of the culture techniques that are available and because the pancreatitis produced in our model is not sufficiently severe. R. S. Bennion (Los Angeles, CA): Did you consider altering the feed of the rats, specifically, the use of meat when feeding the study rats? We have found in several studies that, to increase the number of gram-negative and anaerobic organisms that are recovered from either translocation or infection due to ischemia, the use of meat as food is successful. Have you increased the length of time before killing
THE A M E R I C A N J O U R N A L OF SURGERY
VOLUME 165
J A N U A R Y 1993
51
MEDICH ET AL
the animals? You've indicated that the organisms recovered from the infected pancreata were gram-positive cocci only. Have you extended the length of the study to observe if other organisms are present? We found that gram-positive cocci, both aerobic and anaerobic, are often the first organisms to translocate and that the facultative and aerobic gram-negative rods and the anaerobic organisms will often grow later. David S. Medieh: The animals were fed a standard rat chow without meat throughout the experiment; therefore, I cannot comment on the possible effects of a meatcontaining diet on the recovery of gram-negative or anaerobic organisms. We are interested in the time course of these infections. We have also performed our studies immediately after, or 6 or 18 hours after, the completion of cerulein infusion and have observed the same results. Studies are planned in which the duration is changed, for example, to 6 hours. Andrew L. Warshaw (Boston, MA): If we accept your hypothesis of translocation and migration across the peritoneal cavity, can you speculate on why the beads migrated to the pancreas? Did you find beads or attempt to look for them in any other organs? Why did the beads go to the pancreas, not in macrophages but into the interstitial spaces of the gland? David S. Medieh: I don't have any data. Ranson's
52
T H E A M E R I C A N J O U R N A L OF SURGERY
study showed that lavage decreased the incidence of secondary peripancreatic infection. The use of lavage in patients with pancreatitis is often thought to be beneficial because it removes proteases. I would hypothesize that perhaps pancreatic inflammation produces agents that promote translocation or even chemotactic agents. Infusing the lavage from these animals into other animals would test that hypothesis. R. S. Jones: You studied two groups Of animals, experimental and control. The control group contained six animals. The experimental group contained 12 animals. You had two occurrences of translocation to lymph nodes in the experimental group and three or four occurrences of translocation to the pancreas. What statistical method did you employ to ensure that those occurrences were not random? Was this a statistically significant occurrence? If there were equal numbers in both groups, it would have been highly likely that your findings were significant. But since there were only half the number of control animals as experimental animals, the question of statistical significance is raised. David S. Medich: We used X 2 analysis to determine the significance. In other experiments in our laboratory, the recovery of live bacteria from mesenteric lymph nodes occurs in about 10% to 15% of normal animals. We've never had a positive culture from the pancreas of a control animal.
VOLUME 165
J A N U A R Y 1993