Porcine Models of Severe Sepsis: Emphasis on Porcine Peritonitis

INTRODUCTION Human sepsis is usually caused by a nidus of infection with replicating bacteria (pneumonia or intra-abdominal abscess) that persists for an extended time. Investigation of human sepsis is complicated by numerous factors such as patient selection, infecting agent, time of onset, time of clinical assessment when a patient has developed sepsis, and other factors. To investigate sepsis in using rigorous criteria, a wide variety of animal models has been developed to simulate human sepsis to reduce covariate factors inherent in human investigations. This review presents data derived from a porcine model of human sepsis we have developed (1-3). Pigs were chosen because they have very similar endotoxin sensitivity and tissue antigenicity to humans. In addition, their size allows extensive instrumentation for serial and ultimately continuous monitoring and the capacity for serial blood sampling. Because sequential measurements and samples can be obtained, statistical comparison of results using analysis of variance for repeated measures is very powerful, yielding significance with low numbers of experiments. Porcine cardiovascular and renal physiology is very similar to humans as well (4). Implantation of chronically indwelling instruments allows for recording of cardiovascular parameters in basal states unaffected by anesthetics (1). In this preparation, we have reported that basal heart rate ranges between 90 and 120 beats per minute, arterial systolic blood pressure ranges between 100 and 130 mm Hg and diastolic blood pressure ranges between 50 and 75 mm Hg, lung ventilation is 37 L/min, cardiac output is 2.5 to 3.5 L/min or 100 to 125 mL/kg/min, hemoglobin is 13.7 g/dL, the platelet count is 300,000, and the leukocyte count is 7,000 to 20,000. These results were obtained using rather young pigs weighing between 30 and 40 kg. At adult size, this strain can reach 150 to 200 kg, a size that makes adult animals unworkable. Thus, using pigs at such a young age, although appearing mature, may represent a genetically immature animal. Pigs are much better suited for studies of sepsis and shock compared with canine models because dogs have a sphincter around the hepatic vein that is sensitive to adrenergic stimulation, whereas the pig does not (5, 6). In response to adrenergic stimulation, such as that after the onset of shock, in canine models this sphincter constricts, raising intestinal venous pressure, injuring the mucosal barrier and promoting translation of gut flora (5). The pig is very sensitive to sepsis-induced capillary leak, promoting the development of pulmonary edema, and therefore volume resuscitation must be done with caution in this model. A fluid resuscitation quantity of approximately 2 L on day 1 and 1 L on subsequent days can be administered with minimal risk for clinically significant pulmonary edema. The pulmonary leak is also exacerbated by a significant rise in pulmonary vascular resistance in the pig. The coagulation changes in porcine sepsis are similar to those seen in human sepsis (Fig. 1). The typical hemodynamic profile of lethal porcine septic shock is a decrease in cardiac output concomitant with tachycardia (Fig. 2). Mean arterial pressure, stroke volume, and ejection fraction all decrease, while mean pulmonary artery pressure and pulmonary vascular resistance increases.FIG. 1: Saetre T, Lindgaard AK, Lyberg T: Systematic activation of coagulation and fibrinolysis in a porcine model of serogroup A streptococcal shock. Blood Coagul Fibrinolysis 11:433-438, 2000.FIG. 2: Continuous recording of cardiac output (top) and heart rate (bottom) in a conscious pig before and after E. coli peritonitis. On basal day, left side, these variables were measured continuously over 6 h. In the latter one-half of the day, the animal became restless and was fed. A typical pattern of increased cardiovascular activity was noted. The next day, it was returned to the laboratory and implanted with an E. coli-infected clot at the time indicated by the sharp drop in cardiac output. This animal expired 2 h after implant after heart rate steadily increased. Cardiac output demonstrated a typical response, recovery after surgery, leading to a terminal hypodynamic phase.PORCINE MODELS OF SEVERE SEPSIS/SEPTIC SHOCK Models of porcine sepsis have used endotoxin infusion, bacteria infusion, and peritonitis as infection stimulus. Table 1 demonstrates endotoxin infusion amounts and time ranges in previous experimentation (7-12). Murphey and Traber (13) and Vassilev et al. (14) used a novel approach for endotoxin infusion titration, titrating it to mean pulmonary artery pressure. Murphey and Traber held endotoxin infusion after the initial rise of mean pulmonary artery pressure above 45 mm Hg and then restarted the infusion at a fixed dose when the mean pulmonary artery pressure dropped to less than 40 mm Hg (13). Vassilev titrated endotoxin from the start of the experimentation to maintain mean pulmonary artery pressure at 35 to 45 mm Hg (14). The potential benefit of such an approach is to counter the profound rise in pulmonary vascular resistance that may occur in the pig after Escherichia coli infection and may be the determining factor for survival. Lee et al. (15) used an osmotic pump to infuse S. enteritidis endotoxin continuously intraperitoneally, which allowed up to 18 h to elapse before infusion of endotoxin. This allowed for complete recovery from surgical stress and its accompanying stress hormones and anesthesia. Intravenous bacterial infusion models have used E. coli (O127:bb, 011:84, and 5044552), Pseudomonas aeruginosa, or Group A Streptococcus. Peritonitis models have used peritoneal injection of diluted cecal content or diluted cecal content + E. coli, direct installation of E. coli, open gastrotomy with subsequent closure, or a bacteria-impregnated autologous clot.Table 1: Endotoxin Infusion Protocols in Porcine SepsisFluid resuscitation protocols in these porcine models have varied considerably. Table 2 demonstrates various fluid resuscitation protocols that have been used in porcine models of severe sepsis and septic shock. (7-9, 13-19). Strate et al. (20) used hemodynamic targets for fluid resuscitation. Ringer's solution was administered to a maximum of 1500 mL/h to maintain central venous pressure ≥3 mm Hg, mean atrial pressure at ≥60 mm Hg, and heart rate at ≤140 beats/min.Table 2: Fluid Resuscitation in Porcine Severe Sepsis ModelsREGIONAL PERFUSION The porcine model has been used to measure indices of regional and tissue bed perfusion (transcutaneous oxygen tension, subcutaneous oxygen, tension and gastrointestinal intramucosal pH) (21). Antonsson et al. (22) validated tonometric measurement of gut intramucosal pH (pHI) during endotoxemia and mesenteric occlusion in pigs. In this study, tissue pH was calculated by using the Hendersson-Hasselbalch equation and measurements of arterial (bicarbonate) and intramucosal PCO2, the latter obtained from a saline-filled balloon placed within the stomach lumen. Gut intramucosal pH estimation by this method was validated with direct measurement of ileal pH with a microelectrode. Endotoxemia and mesenteric occlusion were used to demonstrate excellent correlation between direct measurement and tonometric measurement. Montgomery et al. (23) also used a porcine septic model (intravenous infusion of live E. coli) to provide tonometric estimation of gastric, small intestinal, and sigmoid colonic pHI, all of which decrease gradually during the 4-h observation. Like humans, abnormally low gastrointestinal intramucosal pH was found early in porcine septicemia, preceding microscopically detectable damage by several hours. Tonometric technique in the porcine model does provide early detection of gastrointestinal ischemia and septic shock and mirrors events that occur in humans. Strate et al. (20) showed that endotoxin levels and a decrease in intragastric mucosal pH have the highest predictive value for mortality in a porcine peritoneal sepsis model. The authors postulated that the source of endotoxin was potentially dual, peritoneum and gut. SEPSIS-INDUCED ENCEPHALOPATHY Human severe sepsis is often associated with encephalopathy. Because of the multifactorial aspect of altered mental status in human sepsis, the human model is problematic as to adequately isolating any effect of sepsis-induced encephalopathy. Ohnesorge et al. (24) used pancreatitis in a porcine model to create severe systemic inflammatory response syndrome and followed brain somatosensory evoked potentials as a marker of encephalopathy (24). Attenuation of the amplitude of the somatosensory-evoked potentials were observed at least 4 h before defined hemodynamic systemic inflammatory response syndrome criteria. COMBINED HEMORRHAGE AND SEPSIS The combination of hemorrhage and sepsis has been used in the porcine model in two different ways. Parker et al. (25) delivered the two insults simultaneously in an anesthetized porcine model of fixed-volume hemorrhage combined with intraperitoneal sepsis. E. coli was used for the peritoneal implants. The authors used variable bacterial loads and autotransfusion to create scenarios compatible with human intra-abdominal trauma associated with blood loss and bacterial soilage and infection, i.e., a model of combined hemorrhage and peritonitis. In nonresuscitative animals in this study, it was difficult to differentiate the relative contributions of hypovolemia and sepsis. However, in the autotransfused animals, a period of recovery from the early episode from hypovolemic shock was followed by subsequent deterioration and production of a hyperdynamic circulation with reduced vascular resistance characteristics of severe sepsis. The porcine model has also been used to reproduce the two-hit model of human sepsis. The one-hit model of sepsis postulates that massive tissue injury and shock produce systemic inflammation resulting in early multiple organ dysfunction. However, in human sepsis, the typical systemic dysfunction is often associated with the priming insults such as a previous episode of trauma. Therefore, the two-hit model involves multiple sequential insults, i.e., trauma followed by infection. The initial hit increases the organism's vulnerability to a second insult such as a bacterial infection. This priming is because of alterations in neutrophil as well as macrophage immunologic function, and likely includes maladaptive and protective responses. Eissner et al. (26) used hemorrhagic shock as the initial insult, and after recovery, exposed the pigs to P. aeruginosa by infusion. When compared with a control group, Pseudomonas-induced sepsis was characterized by persisting mean pulmonary artery pressure and pulmonary vascular resistance increase, whereas systemic vascular resistance was reduced, thus paralleling typical human sepsis hemodynamics. METABOLIC STUDIES One of the metabolic features of hyperdynamic sepsis is marked catabolism and hypermetabolism (27, 28). After sepsis, the body mobilizes substrates from the periphery to be used by visceral tissues and immune cells, resulting in loss of lean body mass. There is accelerated protein degradation and increased oxidation of amino acids and skeletal muscle. Gluconeogenesis is a mechanism to create energy for the host immune response. By this route, skeletons from amino acids and lactate are converted into glucose, while amino groups are used in ureagenesis, resulting in net nitrogen loss excreted as urea. Bruins et al. (29) used fasting pigs receiving intravenous E. coli and volume resuscitation over 24 h to reproduce hyperdynamic sepsis. Glucose amino acid and protein metabolism were measured across the portal-drained viscera, liver, and hind quarters. Endotoxemia was shown to increase hindquarter and visceral glycolysis as well as branch-chain amino acid transamination. Portal-drained viscera provided significant gluconeogenetic amino acids and lactate to the liver. This model was demonstrated to be a very appropriate model for human hypermetabolic sepsis. ACUTE LUNG INJURY Acute lung injury is the most common organ dysfunction associated with sepsis. Porcine sepsis models using infusion of P. aeruginosa have been used to reliably produce pulmonary shunting, noncardiac pulmonary edema, and changes typical of human sepsis-induced acute lung injury. Dehring (30) used Pseudomonas, S. aureus, and E. coli infusions to produce systemic hypotension, pulmonary hypertension, increased pulmonary vascular resistance, hypoxemia, and pulmonary edema typical for acute lung injury-induced acute respiratory failure in sepsis. THE PORCINE PERITONITIS MODEL Porcine peritonitis induced by implantation of an organism-impregnated fibrin clot into the peritoneum offers a feasible and appropriate model of human septic shock. This model imitates human sepsis in many areas to include persistent nidus of infection, systemic bacteremia over several days, dose-dependent survival, and cardiovascular responses that mimic multiple aspects of human sepsis. Other similarities to human clinical sepsis include a defined nidus of infection, live bacteria, a conscious animal, and continuous monitoring. In addition, circulating myocardial depressant substances can be demonstrated with reversal of effect in survivors. Nonsurviving animals are euthanized based on achieving thresholds of increased heart rate, decreased cardiac output, or decreased stroke volume. Various authors have identified significant variability among porcine peritonitis models of sepsis (31, 32). This likely relates to the lack of standardization of the model. In general, peritoneal sepsis has been induced by peritoneal fecal soilage or by introduction of laboratory-grown bacteria into the peritoneum. Peritoneal soilage models have included cecal ligation and puncture, intraperitoneal introduction of fecal-soaked gauze, fecal pellet implantation, and direct fecal soilage. The problem associated with soilage models is difficulty in controlling the quantity and quality of bacteria introduced into the peritoneum. Alternatively, a known quantity and type of bacteria can be introduced into the peritoneum to produce sepsis. However, introduction of pure cultures of bacteria into the peritoneal cavity is artificial. Virulence may be altered by culture growth before implantation. Both types of models have potential benefits and disadvantages. Kazarian et al. (33) contrasted two types of intraperitoneal soilage, autologous fecal inoculum versus pure culture of E. coli. Early (1-4 h) and late (24-72 h) changes were identified in both groups. The E. coli group was characterized early by hypotension, low cardiac output, and increased systemic and pulmonary vascular resistances along with leukopenia, hypoglycemia, and increased lactic acid. With a pure culture of E. coli, physiology returned fairly rapidly to baseline in those animals that survived. In the autologous fecal inoculum pigs, the early hypotension was characterized by increased cardiac output and reduced systemic vascular resistance. Unlike the E. coli group, the changes in the FEC group persisted in later days and were associated with leukocytosis as opposed to leukopenia. Endotoxin levels were higher in the E. coli group and persisted longer in the FEC group. The FEC group, despite lower levels of LPS, developed abdominal abscesses with continued sepsis over a longer period of time. INTRAPERITONEAL IMPLANT MODEL The model uses Yorkshire pigs of either sex between 30 and 35 kg who are instrumented with catheters and cardiac transducers under general anesthesia and mechanical ventilation. Figure 3 illustrates location of implanted transducers in our model. Through a left thoracotomy, the pericardium is severed (not closed) and instruments are implanted to measure pulmonary artery blood flow, left ventricular pressure, and short axis diameter, and catheters are placed within the pulmonary artery and aorta through which arterial and pulmonary artery pressures are measured. From these measured variables, ejection fraction, cardiac output, stroke volume, vascular resistances, and many other variables are calculated. The experiment begins after 5 to 7 days of postsurgical recovery. Bacteria are cultured overnight and are incorporated into a fibrin clot. Implantation of a bacteria-laden fibrin clot into the peritoneum is made under brief general anesthesia through a small (5-7 cm) midline abdominal incision. Animals recover in approximately 10 min. Bacterial strains used in our laboratory have included E. coli 011.B4 and Staphylococcal aureus to produce gram-negative and gram-positive septic models. Figure 4 demonstrates the experimental protocol used by our laboratory for porcine peritonitis. Concentrations vary based on virulence of the organism, with from 1 × 109 to 1 × 1011 cfu/kg producing mortality rates of 70% to 85% in untreated animals. Our group has previously published data on the expected survival decrease associated with clot implant (Fig. 5). Sixty percent of survivors and 80% of nonsurvivors can be demonstrated to be bacteremic at 30 min after implantation, with that number decreasing to 25% in survivors and increasing to 100% in nonsurvivors at 2 h. A significant pulmonary artery hypertension response occurs in the septic pig with the magnitude correlating with mortality. The nonsurviving pig enters a hypodynamic state similar to that of under-resuscitated human septic shock, and is characterized by hypotension with decreased cardiac output and elevated systemic vascular resistance. This manifestation of septic shock was commonly reported in humans between 1965 and 1973, and was then see infrequently thereafter as more aggressive early fluid resuscitation led to the high cardiac output-low systemic vascular resistance model now commonly associated with human sepsis. With recent publication of early goal-directed resuscitation of septic shock in the emergency department, this profile has been rediscovered. In nonsurviving pigs, the hemodynamic profile is characterized by tachycardia, hypotension, low cardiac output, and high pulmonary and systemic vascular resistance. In survivors, the tachycardia response is less severe and, after an initial drop, the blood pressure returns to near normal levels (similar to sham animals). Cardiac output initially decreases in survivors but to a much less extent than in nonsurvivors. In contrast to nonsurvivors, systemic vascular resistance in survivors goes down and pulmonary vascular resistance increases significantly, but to a lower level than in nonsurvivors. Contractility decreases initially in survivors and nonsurvivors, with a much greater decrease in nonsurvivors. Recovery occurs after 2 to 3 days in survivors. Myocardial depressant substance activity can be demonstrated to be present in vitro during the first 3 days after infected clot implantation. After of several factors may a pig from which include of infection and based on blood and of clot at the time of for as includes any of the heart rate greater than 200 cardiac output less than of cardiac output less than L/min, or stroke volume less than 10 or of transducers transducers and A pressure was implanted in the left from which pressure was measured measured short axis from which cardiac output was measured. catheters were in the pulmonary artery to measure pulmonary pressure and to blood in the aorta to measure and blood and in the left to measure pressure and protocol of experimental porcine E. coli porcine peritonitis with bacterial A porcine model of peritonitis and bacteremia human septic shock. animals implanted with a clot survived. Animals implanted with a variety of E. coli expired in a time from 1 to h see for of sepsis have been and to be in our these has to the and was in decreasing mortality in our porcine model and in the human clinical (Fig. have shown in our model and are in various of to clinical 7 and These results that this porcine model is well suited to the of in human sepsis and the of experimental factor on survival in E. coli pigs. The animals were administered × 109 cfu/kg E. coli on levels and survival rates in E. coli pigs. The animals were administered bacterial loads of × of E. coli rate of porcine E. coli peritonitis in control and animals. The animals were administered a bacterial of 2 × cfu/kg 2).