Sep 11, 1995 - Treatment with either cefotaxime or levamisole (an immunostimulant) were effective. However, the anti- inflammatory drug dopamine, which ...
Ann R Coll Surg Engl 1996; 78: 350-353
Pathogenesis of pancreatic infection Adam L Widdison
DM FRCS
Senior Registrar
Bristol Royal Infirmary, Bristol
Key words: Pancreas; Pancreatic diseases; Pancreatitis; Infection
John Hunter studied comparative anatomy of the pancreas but was unaware of pancreatic infection which is now the leading cause of mortality in pancreatitis. This was investigated using a feline model of pancreatitis. Pathogens spread to the healthy and inflamed gland from many sources including colon, gallbladder, or a septic focus and by various routes including the circulationa reflux into the pancreatic duct or by transmural migration from the colon. Colonisation risk was proportional to necrosis and inflammation, confirming clinical observations. These studies showed that pathogens frequently colonised the pancreas, but infection developed only in animals with pancreatitis. In cats with pancreatitis, phagocytic function was reduced by 28%. This was problably owing to phagocytic capacity being overwhelmed by protease-antiprotease complexes because, in humans, granulocyte and lymphocyte function was normal. These experiments suggested that it would be difficult to prevent pancreatic colonisation, but indicated some types of therapy may have potential. These were investigated using this animal model of pancreatic infection. Treatment with either cefotaxime or levamisole (an immunostimulant) were effective. However, the antiinflammatory drug dopamine, which reduced inflammation, did not eradicate all pathogens.
Hunter's lasting contribution, however, was to combine astute clinical observations with animal experiments to improve treatment in patients. The scientific principles he established are as relevant today as they were then. These principles are particularly appropriate when applied to investigation of the pathogenesis of pancreatic infection. In recent years pancreatic infection has become the most important life-threatening complication of acute pancreatitis (1). It occurs in 30% to 40% of patients with severe pancreatitis, but is a rare complication of mild disease. Escherichia coli (E. coli) and other enteric aerobes are the dominant pathogens suggesting the colon is a possible source. However, pancreatic pathogens are also normal commensals or frequent pathogens elsewhere and anaerobes, which dominate infections caused by colonic pathology, rarely cause pancreatic infection. A series of experiments was therefore performed to determine likely sources of pancreatic pathogens, routes of spread to the pancreas and host factors associated with the development of sepsis. With increased knowledge about the pathogenesis of pancreatic infection, novel treatments may be tested.
Experimental model John Hunter was born in 1728, less than 100 years after Georg Wirsung described the main pancreatic duct (1642) and Greissellius reported a patient who had died from gallstone pancreatitis (1681). Unfortunately, Hunter did not devote much of his time to the pancreas. His only direct contribution was to compare the gland in various animals including fish, geese, elephants, and snakes. Based on a Hunterian lecture delivered during a meeting of the Surgical Club of South West England, held at the Bristol Royal Infirmary, on 27 October 1995 Correspondence to: Mr A L Widdison, Consultant Surgeon, Treliske Hospital, Truro TRI 3LJ
The feline low-pressure duct perfusion model of acute biliary pancreatitis was used (2). Briefly, acute pancreatitis was created in anaesthetised animals by the antegrade infusion into the main pancreatic duct of first glycodeoxycholic acid and then pancreatic juice in which the enzymes had been activated by incubation with enterokinase. Simultaneously, 16,16-dimethyl prostaglandin E2 was infused intravenously. In control animals, the main pancreatic duct was perfused with physiological salt solution. Aseptic precautions were taken throughout. After this cats were allowed to recover. Animals developed the typical features of acute necrotising pancreatitis within 24 h. The severity of pancreatitis was assessed using a previously validated histological score
Pathogenesis of pancreatic infection (2). In all experiments a distinctive clinical strain of E. coli was used as a marker pathogen. Standard microbiological techniques were used to culture and identify E. coli. All results were compared using non-parametric statistics.
Sources of pancreatic pathogens To investigate the likely origin of pancreatic pathogens, E. coli were placed in the proximal colon, gallbladder, or obstructed renal pelvis (to represent a distant-septic focus) after induction of acute pancreatitis or sham operation in control animals (3). Within 72 h most (67-83%) pancreases were colonised, regardless of the source. This confirmed that pathogens can spread to the pancreas from the colon, as suspected. Alternative sources, however, include infected bile or a distant septic focus. The high colonisation rate suggested that, clinically, bacteria may spread to the pancreas more frequently than is currently thought, but not all cases progress to infection.
Routes of spread to the pancreas The only route of spread common to each of the possible sources is via the circulation. Indeed, bacteria can only reach the pancreas from the kidney by this route. To investigate this, cats with and without pancreatitis were given a bolus of E. coli intravenously to duplicate a bacteraemic episode (4). After 24 h, 67% of inflamed glands and 30% of control glands were colonised by E. coli (P= 0.1). This experiment revealed that pathogens can spread to the pancreas via the bloodstream. Bacteraemias may arise spontaneously from the biliary and gastrointestinal tract, from urinary or respiratory tract infections, by direct shedding from venous cannulas or during invasive procedures. Bacteraemias have a tendency to infect dead and devitalised tissue and are a recognised cause of many diverse infections, including pyelonephritis, brain abscesses, endocarditis, septic arthritis and osteomyelitis. They must now be implicated in the pathogenesis of
pancreatic infection. An alternative route of access to the pancreas is by reflux of infected bile or duodenal contents along the main pancreatic duct. To investigate this, E. coli were placed inthe pancreatic duct in cats with and without pancreatitis (4). After 24 h, E. coli could be cultured from 83% of glands, whether or not they were inflamed. This has important implications for patients with biliary pancreatitis. Choledocholithiasis is associated with bactobilia and biliary-pancreatic reflux may occur when a calculus is lodged in the common channel. This may explain why some patients with biliary acute pancreatitis develop infection early in the course of the disease (1). Similarly, patients are at risk of becoming infected
during endoscopic retrograde pancreatography. Some patients with biliary pancreatitis have associated cholangitis and bacteraemia as a result of cholangiovenous
351
reflux. This suggested that pathogens may also spread from infected bile via the circulation. To investigate this, E. coli were placed in the gallbladder after induction of pancreatitis (4). In half of the cats, the common bile duct (CBD) was ligated to prevent biliary-pancreatic reflux. The CBD was patent in the remainder. After 24 h the rate of pancreatic colonisation in cats with a patent CBD was 83% compared with 50% in the duct ligation group (difference not significant). This observation suggested that cholangiovenous reflux may also lead to pancreatic infection. In support of this the bacteraemia rate among cats with pancreatic colonisation caused by E. coli from the gallbladder, whether or not the CBD was patent, was greater (56% and 100%, respectively) than that among cats colonised directly from the main pancreatic duct (10%, P3) pancreatitis. In patients with severe pancreatitis, C reactive protein was increased, 152 (19) mg/l compared with 67 (20) mg/l in mild pancreatitis (P = 0.05), APACHE II scores higher, 14 (2) compared with 6 (1) in mild pancreatitis (P=0.01), and organ failure more frequent, cyte function in 20
*
NECROSIS
*
PMN INFILTRATE
TOTAL INFLAMMATION
2. Effect of persisting coliform infection on pancreatic inflammation 1 week after induction of pancreatitis. (Mean ± SEM, PMN = polymorphonuclear neutrophil infiltrate, *P
