Preclinical Sepsis Models

Abstract

Background:

A variety of sepsis models have been used to unravel pathophysiologic processes and to examine the effects of novel therapeutic interventions. The lack of therapeutic efficacy of numerous compounds in clinical sepsis trials, despite glorious results in animal models of sepsis, has raised doubt and debate about the usefulness of such models.

Methods:

Review of the pertinent literature.

Results:

Many sepsis models have been described, none of which is ideal. Clinical sepsis can originate from different sources, can be accompanied by many complicating conditions, and strikes human beings with strongly variable genetic backgrounds, co-morbidities, and drug usages. To provide answers to the three main objectives of research—insight into the regulation of normal host defense mechanisms in the early stages of infection; the mechanisms underlying dysregulation of the host response; and proof of principle for the mechanism of action of novel therapeutic agents and to establish their efficacy and potential harm—diverse models are required. The future of sepsis research lies in the systematic combination of models, together with in vitro studies and carefully designed and monitored Phase I/II clinical studies.

Conclusion:

This review discusses the nature of various animal sepsis models and the way their results should be interpreted.

In 1992, a Consensus Conference assembled by the American College of Chest Physicians (ACCP) and the Society of Critical Care Medicine (SCCM) introduced the term “systemic inflammatory response syndrome” or SIRS to define the systemic inflammation that accompanies severe infection and sterile trauma [1]. Although criticized by many, SIRS still is used for defining sepsis: Sepsis is considered to be present when patients with a confirmed or suspected infection have at least two of four SIRS criteria (body temperature >38°C or <36°C; heart rate >90 beats/min; hyperventilation, evidenced by a ventilatory rate of >20 breaths/min or a PaCO₂ <32 mmHg; and a white blood cell count of >12,000/mcL or <4,000/mcL). As such, sepsis is a syndrome caused by the host response to an infection [2,3].

The inflammatory reaction that eventually drives SIRS is part of the innate immune response instigated by a causative organism, aimed at eradicating the infection. The SIRS concept was derived in part from the theory that had emerged in the late 1980s dictating that death from sepsis is caused by an overstimulated immune system. This assumption was based largely on studies in animals that were infused with high doses of bacteria or bacterial products, in particular, lipopolysaccharide (LPS). Such models were associated with a brisk systemic release of an array of pro-inflammatory mediators, and many interventions targeting these host-derived proteins were found to be protective against death. In a hallmark manuscript resulting from a collaborative effort of the research groups led by Anthony Cerami and Stephen F. Lowry, elimination of the prototypic pro-inflammatory cytokine tumor necrosis factor (TNF)-α was reported to protect baboons against lethal gram-negative sepsis [4]. Inhibition of another pro-inflammatory cytokine, interleukin (IL)-1, also reduced lethality induced in baboons by intravenous infusion of viable Escherichia coli [5]. These landmark findings revolutionized the way of thinking about sepsis pathogenesis. Unfortunately, subsequent clinical sepsis trials with anti-TNF-α strategies and recombinant IL-1 receptor antagonist failed to show benefit, and many other anti-inflammatory therapies likewise failed to modify the outcome of patients with sepsis. Current knowledge indicates that the “systemic hyper-inflammation” theory is an oversimplification and does not hold true for most sepsis patients [2,3,6]. Indeed, the host response to infection is a dynamic, compartmentalized, and highly complex process, which involves not only enhancement of inflammatory reactions but also impairment of adequate immune responsiveness. This review summarizes the evolution of animal sepsis models, from the early systemic challenge models to more recent, more sophisticated models of initially localized infections and sepsis-associated immune suppression.

General Considerations and Limitations

Sepsis can be caused by many organisms, originate from various body sites, and be associated with a large variety of host responses, depending on the stage of the syndrome, the severity of the initial insult, the virulence of the pathogen, and the genetic composition, co-morbidity, and age of the host. The assumption that “sepsis” can be mimicked in an animal model shows little respect for the complex nature of this heterogeneous syndrome. At best, animal models can reproduce aspects of the consequences of severe infection, such as the early host response to a gradually growing bacterial load in the lungs on one end of the spectrum and fulminant septic shock on the other [7 –13]. Animal sepsis models need to be conducted with a clear objective in mind (Table 1). Indeed, some investigations seek primarily to examine sepsis pathogenesis, for example, the role of innate recognition of bacteria by pattern recognition receptors and the ensuing adequate response of the host to pathogens that try to invade normally sterile body sites. Other studies focus on evaluation of potentially new therapeutic strategies before going into clinical trials, either with regard to their presumed mechanism of action or to efficacy.

Table 1.

Objectives of Animal Sepsis Models

Physiology	To obtain insight into the regulation of normal host defense mechanisms in the early stages of infection
Pathophysiology	To obtain insight into the mechanisms underlying dysregulation of the host response during worsening infection
Therapy	To obtain proof of principle for the mechanism of action of novel therapeutic agents and to establish their efficacy and potential harm

An inherent limitation of animals for sepsis research is the fact that experimental groups consist almost invariably of young, previously healthy animals, which in the case of mice and rats also are identical genetically. The use of small animals, in particular mice, has major advantages that include low cost and the availability of many research tools, such as genetically modified animals and immunologic reagents. However, larger animals are required if invasive monitoring is deemed necessary for adequate readouts. What can be considered an advantage for preclinical research; i.e., constructing clearly defined experimental groups that are confronted with a strictly controlled challenge (e.g., injection of bacteria or bacterial products or a surgical procedure to initiate sepsis) is a major disadvantage for extrapolation to human sepsis. Indeed, in the clinical setting, sepsis is a highly variable syndrome, with many modes of onset and manifestations. This strengthens the notion that the ideal sepsis model does not exist. Clinical sepsis can be studied best in a preclinical setting by conducting multiple investigations, each seeking to reproduce specific appearances of sepsis as we face it in the intensive care unit.

Human sepsis often evolves over a period spanning several days, and intensive care treatment can last weeks; animal models typically are brief, ranging from hours to days. Indeed, whereas many clinical sepsis deaths occur late (i.e., after more than one week), death in preclinical sepsis models typically takes place within a maximum of several days. Besides differences in timing of disease development, most animal models do not involve supportive therapeutic interventions, such as hemodynamic support, mechanical ventilation, or source control, which sets them far apart from the clinical scenario.

Systemic Challenge Models

Early knowledge of the pathogenesis of sepsis has been derived almost exclusively from models in which bacteria or bacterial products were infused intravenously. Such challenges, given most commonly at high doses, induce a rapid response that resembles septic shock. Lipopolysaccharide (LPS), the toxic component of the outer membrane of gram-negative bacteria, has been used widely in this setting, not in the least because it elicits a reproducible reaction in inbred animals and does not require growing of bacterial inoculates for infection. In addition, intravenous administration of live bacteria has been utilized by many investigators to examine the acute response to a high intravascular bacterial load. A major limitation of systemic challenge models is that they lack a localized source from which the infection disseminates; the acute circulation of a huge burden of bacteria or products thereof is an unlikely event in the clinical setting.

Endotoxin

Lipopolysaccharide (endotoxin) has been implicated as a pathogenetic factor, not only in gram-negative sepsis but also in severe infection caused by other organisms. Circulating LPS has been detected in sepsis patients in intensive care settings irrespective of the causative organism and correlates with a poor outcome [14,15]. Intravenous injection of LPS elicits systemic inflammation that resembles some of the initial characteristics of sepsis. Notably, the sensitivity to LPS differs markedly among species; whereas in particular, mice, rats, and dogs are relatively resistant, humans, nonhuman primates, rabbits, and sheep are much more reactive. The use of high LPS doses in LPS-resistant species can be associated with toxicity that is not observed in LPS-sensitive species.

The response to bolus LPS injection is characterized by an overwhelming innate immune response. One of the typical responses to high-dose LPS is the early systemic release of pro-inflammatory cytokines, most notably TNF-α. Neutralization of TNF-α is associated with clear protection against LPS-induced death [16], which illustrates that this fulminant model shows the detrimental effect of aberrant stimulation of the innate immune system while obscuring the protective role of innate immunity in other infectious settings. Bolus injection of high-dose LPS produces an immediate hypodynamic cardiovascular state, whereas lower LPS doses promote a hyperdynamic reaction accompanied by increased cardiac output, although responses differ among species [7,11]. Some investigators have used continuous infusion of LPS in an attempt to more mimic human sepsis closely. Indeed, animals infused with LPS for long periods of time manifest features of compensated human sepsis, including hypermetabolism, mild hypotension, and elevated serum lactate concentrations [7].

Extrapolation of animal data to human beings needs to be done with caution. The use of human beings to study the host response to sepsis can be valuable, although investigations using healthy human beings obviously never reproduce the severity of the sepsis insult in patients. The model of human endotoxemia in which healthy subjects are injected with a standardized preparation of LPS has some relevance for sepsis [17 –19]. The model can be used to study mechanisms that contribute to the activation of inflammatory cascades induced by a common bacterial antigen, to unravel interactions between distinct inflammatory systems (e.g., the cytokine network and the coagulation system), and to obtain proof of principle for the efficacy of novel anti-inflammatory compounds in man [17 –19].

Intravenous infusion of bacteria

The direct intravascular infusion of viable bacteria has been used as a model of sepsis in several animal species. Especially in the era in which pro-inflammatory cytokines were discovered as essential mediators of fulminant septic shock, the baboon model of E. coli-induced sepsis received much attention [20]. Indeed, hallmark investigations performed in Lowry's laboratory in New York pointed to TNF-α and IL-1 as crucial factors in death induced by intravenous infusion of E. coli. Escherichia coli-induced sepsis was associated with elevated circulating concentrations of TNF-α and IL-1β. Neutralizing monoclonal anti-TNF-α antibody fragments administered to baboons shortly before bacterial challenge protected against shock, vital organ dysfunction, persistent stress hormone release, and death [4]. Escherichia coli septic shock was associated with high plasma concentrations of IL-1β, and treatment with recombinant IL-1 receptor antagonist attenuated hypotensive shock and improved the survival rate from 43% to 100% [5]. These investigations were the basis for many subsequent preclinical studies demonstrating the involvement of pro-inflammatory cytokines in the pathology elicited by bolus infusions of live bacteria and eventually in the design and performance of clinical sepsis trials with anti-cytokine strategies, which unfortunately all were negative [21].

High-dose E. coli also results in disseminated intravascular coagulation with involvement of the microvascular endothelium and disturbed anticoagulant mechanisms, thereby mimicking the clinical features of severe human sepsis [20]. Interventions targeting the tissue factor pathway established not only that tissue factor is pivotal for activation of coagulation during septic shock but also that this pathway contributes to organ failure and death [22]. In addition, the anticoagulants antithrombin and activated protein C proved strongly protective against death in this model [23,24]. Similar to the disappointing findings with anti-cytokine interventions, neither tissue factor pathway inhibitor nor antithrombin showed benefit in patients with sepsis [22]. Although activated protein C originally was reported to reduce the mortality rate in patients with severe sepsis [25], this product was withdrawn from the market recently after a confirmatory trial in patients in septic shock demonstrated no benefit whatsoever [26,27].

Intravenous infusion of viable pathogens also has been done in smaller animals and using other bacteria [9,11]. It is beyond the scope of this review to discuss all of these studies in detail. Here it suffices to say that these models may have special relevance for studying the mechanisms contributing to septic shock. However, the course of the disease is far more fulminant than in most clinical sepsis cases and has insufficient value in predicting the clinical efficacy of sepsis interventions.

Infection Models with a Localized Source

The two main sources of infection in patients with sepsis are the respiratory tract and the abdominal cavity (25,28). As such, models of pneumonia and peritonitis bear the highest clinical relevance for sepsis research.

Pneumonia models

Lung infections can present as community-acquired pneumonia (CAP) and hospital-acquired pneumonia (HAP), both caused by partially overlapping and partially distinct pathogens [29]. The most frequent causative pathogen in CAP worldwide is Streptococcus pneumoniae, whereas Pseudomonas aeruginosa and Klebsiella pneumoniae are prominent causes of nosocomial pneumonia. The pneumonia models that have been used most frequently incorporate these clinically relevant pathogens [30]. Pneumonia can be induced by direct administration of bacteria into the trachea or the nose of anesthetized animals. The intratracheal infection usually is created with the help of a small catheter that is inserted into the trachea surgically or via the mouth followed by administration of a small volume of fluid containing the desired number of bacteria. Alternatively, mice are exposed to aerosolized bacteria in a whole-animal chamber.

Notably, different respiratory pathogens produce distinct infection courses in mice [30]. For example, in the case of Pseudomonas, a relatively high dose must be instilled into the airways to induce pneumonia; smaller inocula are cleared from the airways and do not cause disease. Second, compared with the respiratory tract infection models using S. pneumoniae or K. pneumoniae, in which the bacterial load increases gradually over several days, eventually resulting in dissemination of the infection to distant body sites, the Pseudomonas model is associated with acute pneumonia that develops within 6–24 h. The inoculum size and the character of the inflammation that develops in response to the bacterial challenge may not reflect clinical pneumonia adequately.

Some investigators have used “two-hit” models, seeking to mimic the clinical scenario of HAP. These sequential challenge models most commonly use cecal ligation and puncture (CLP) to induce sublethal abdominal infection and sepsis followed one to several days later by induction of bacterial pneumonia, especially making use of pathogens relevant for HAP such as P. aeruginosa. Clearly, mice subjected to CLP have a strongly impaired host defense against subsequent bacterial infection of the respiratory tract [31,32]. Other “first hits” that make mice more vulnerable to bacterial pneumonia include stroke [33], a sterile acute-phase protein response [34], and pancreatitis [35].

Peritonitis models

Several models of peritonitis associated with sepsis have been used to study pathophysiology and potential new treatment options. Two models that seek to mirror the clinical scenario of abdominal sepsis resulting from a perforated colon leading to polymicrobial infection are CLP and colon ascendens stent peritonitis (CASP) [36]. The CLP method has been considered the gold standard for sepsis research by many investigators [10,37], although it is challenged by others [38]. The procedure is done by laparotomy, with ligation of the cecum distal to the ileocecal valve and puncture of the ligated cecum with or without massaging fecal contents of the gut into the peritoneal cavity. This procedure results in two types of injury: Infection of the peritoneum by mixed gut-derived bacterial flora and an inflammatory source of necrotic tissue. The severity of disease associated with CLP can be varied by using different needle sizes to puncture the cecum, by the number of punctures, and varying the length of the ligated cecum. In addition, in some studies, animals are treated with antibiotics after the surgical procedure, whereas in others, they are not. Clearly, for a proper understanding of scientific publications in which CLP is used as a sepsis model, a clear and detailed description of the methods used is mandatory. Although CLP can re-create several aspects of human sepsis, including hemodynamic and metabolic phases, the model involves responses that are not part of many clinical sepsis cases, such as surgery, abscess formation, and the presence of necrotic bowel, which all may impact the outcome.

The CASP technique is accomplished by implanting a stent into the ascending colon, resulting in persistent spill of fecal contents into the abdominal cavity. The stent can be removed after induction of disease, resembling therapeutic intervention by the surgeon. Like CLP, the severity of CASP can be varied: Stents of larger diameter cause higher mortality rates.

Direct comparison of CLP and CASP has indicated that these models are complementary and may reflect different types of syndromes and host responses, wherein CASP is more associated with diffuse peritonitis and systemic inflammation, whereas CLP is a model primarily of intra-abdominal abscess formation associated with less profound systemic inflammation [36]. Notably, however, in both models, the severity and the acuity of the induced disease depend strongly on the specific details of the surgical techniques, as described above.

The direct intraperitoneal injection of viable bacteria has been used to induce abdominal sepsis in small animals. Although depending on the virulence of the bacterial strain, the direct inoculation models of peritonitis resemble more the hyperacute syndrome produced by intravenous challenge than do CLP or CASP. Nonetheless, peritonitis induced by a single bacterial strain relevant for abdominal infection (most commonly E. coli) may provide information on antibacterial host defense mechanisms in the absence of an injured or necrotic bowel (39). Another method to induce abdominal sepsis in animals is implantation in the peritoneum of a fibrin clot impregnated with a pathogen. This model resembles human sepsis in that it includes a persistent nidus of infection with systemic dissemination [40,41].

Conflicting Results from Different Sepsis Models

Different preclinical sepsis studies may reveal opposite roles for a single mediator, depending on the animal model. As an example, the impact of TNF-α inhibition strongly depends on the infection model [16]. Anti-TNF-α strategies especially were efficacious with regard to mortality reduction in systemic challenge models, whereas they had no influence or even caused harm in models of initially localized infections. Along the same line, whereas TNF-α and IL-1β contribute to death after intravenous infusion of E. coli into nonhuman primates [4,5], these prototypic pro-inflammatory cytokines act together to mount an adequate innate immune response during pneumococcal pneumonia [42]. Similarly, whereas administration of the anti-inflammatory cytokine IL-10 is protective after a systemic LPS challenge [43], the same intervention impairs host defense during gram-positive pneumonia [44]. Comparable discrepancies have been found regarding the effects of anticoagulants used in clinical sepsis trials; e.g., recombinant activated protein C exerts anticoagulant and anti-inflammatory effects in several preclinical sepsis models but not in human endotoxemia [45,46]. These findings, which are provided here merely as examples, further emphasize the necessity to evaluate interventions in multiple sepsis models in order to comprehend their effects fully, or, in the case of pathogenetic studies, the biological roles of the endogenous mediators eliminated. These results accentuate the need to define clearly the objective of each specific model. As such, studies that seek to obtain insight into the normal function of innate immunity preferentially should make use of low bacterial inocula without antibiotics, whereas investigations that aim to examine the potential efficacy of a novel therapeutic should focus primarily on models in which postponed treatment is provided in the context of concurrent antibiotic therapy.

Conclusion

Many sepsis models have been described in the literature. None of these preclinical simulations can be considered the “ideal” sepsis model. This should not come as a surprise considering the tremendous variability in which sepsis patients are admitted to the hospital. Clinical sepsis can originate from different sources, including pneumonia, peritonitis, and soft tissue and urinary tract infection; can be accompanied by many complicating conditions, such as necrotic tissue or the need for surgical intervention, acute lung or kidney injury or both, disseminated intravascular coagulation, and cardiovascular collapse; and strikes humans with strongly variable genetic background, co-morbidities, and drug use. As such, interpretation of preclinical sepsis research involving animals should be done with great modesty. In seeking to provide answers to the three main objectives for animal sepsis research (Table 1), many preclinical studies involving diverse models are required. Whereas preclinical sepsis research was dominated by systemic challenge models in the 1980s, current investigational models also involve localized infections, most notably pneumonia and peritonitis. The future of sepsis research seeking to obtain insight into pathogenetic mechanisms and evidence for the therapeutic efficacy of new drugs lies in the systematic combination of different animal models, together with in vitro studies and carefully designed and monitored Phase I/II clinical studies.

Footnotes

Acknowledgment

This article is dedicated to my mentor and friend Stephen F. Lowry, who provided me with a solid scientific basis for my understanding of sepsis pathophysiology (see ).

Author Disclosure Statement

No competing financial interests exist.

Appendix. Stephen F. Lowry As My Mentor and Friend

This article is dedicated to my mentor and friend Stephen F. Lowry, who died June 4, 2011. The topic of this manuscript fits nicely with my professional relationship with Steve. He taught me the essence of translational science, introducing me to the field of preclinical models of inflammation and infection. I met Steve for the first time in 1991, when he was a member of my PhD thesis committee in Amsterdam, The Netherlands. I remember looking up to him. He had established his name and fame with the animal models described herein, being an eminent member of the research team that identified tumor necrosis factor-α as a crucial mediator of fulminant experimental sepsis in hallmark papers in Nature and similar journals. It struck me how kind and gentle a person he was. He treated me (a young and inexperienced investigator) with much respect, showing genuine interest in my opinion on TNF biology. From 1993 to 1995, I was postdoctoral fellow in Steve's laboratory (the Laboratory of Surgical Metabolism) at Cornell University Medical College, New York. The way Steve worked with me is how I now try to work with my students and fellows. He was an enthusiastic mentor, who gave me much responsibility and freedom, let me learn from my mistakes, and always was there for feedback and advice. My postdoctoral period with Steve without doubt has been the solid base for my professional career. In addition, my family is lucky to have shared many personal and family events with Steve and his wife Susette Coyle. In spite of the geographic distance, our families have numerous memorable moments to cherish. As a mentor, Steve Lowry is my role model; as a friend, he meant the world to me.

References

Bone

, Sibbald

, Sprung

. The ACCP–SCCM Consensus Conference on sepsis and organ failure. Chest, 1992; 101:1481–1483.

van der Poll

, Opal

. Host–pathogen interactions in sepsis. Lancet Infect Dis, 2008; 8:32–43.

Fry

. Sepsis, systemic inflammatory response, and multiple organ dysfunction: The mystery continues. Am Surg, 2012; 78:1–8.

Tracey

, Fong

, Hesse

et al. Anti-cachectin/TNF monoclonal antibodies prevent septic shock during lethal bacteraemia. Nature, 1987; 330:662–664.

Fischer

, Marano

, Van Zee

et al. Interleukin-1 receptor blockade improves survival and hemodynamic performance in Escherichia coli septic shock, but fails to alter host responses to sublethal endotoxemia. J Clin Invest, 1992; 89:1551–1557.

Skrupky

, Kerby

, Hotchkiss

. Advances in the management of sepsis and the understanding of key immunologic defects. Anesthesiology, 2011; 115:1349–1362.

Fink

, Heard

. Laboratory models of sepsis and septic shock. J Surg Res, 1990; 49:186–196.

Deitch

. Animal models of sepsis and shock: A review and lessons learned. Shock, 1998; 9:1–11.

Schultz

, van der Poll

. Animal and human models for sepsis. Ann Med, 2002; 34:573–581.

10.

Buras

, Holzmann

, Sitkovsky

. Animal models of sepsis: Setting the stage. Nat Rev Drug Discov, 2005; 4:854–865.

11.

Poli-de-Figueiredo

, Garrido

, Nakagawa

, Sannomiya

. Experimental models of sepsis and their clinical relevance. Shock, 2008; 30,Suppl 1:S53–S59.

12.

Dyson

, Singer

. Animal models of sepsis: Why does preclinical efficacy fail to translate to the clinical setting? Crit Care Med, 2009; 37,Suppl 1:S30–S37.

13.

Doi

, Leelahavanichkul

, Yuen

, Star

. Animal models of sepsis and sepsis-induced kidney injury. J Clin Invest, 2009; 119:2868–2878.

14.

Opal

, Scannon

, Vincent

et al. Relationship between plasma levels of lipopolysaccharide (LPS) and LPS-binding protein in patients with severe sepsis and septic shock. J Infect Dis, 1999; 180:1584–1589.

15.

Marshall

, Foster

, Vincent

et al. Diagnostic and prognostic implications of endotoxemia in critical illness: Results of the MEDIC study. J Infect Dis, 2004; 190:527–534.

16.

Lorente

, Marshall

. Neutralization of tumor necrosis factor in preclinical models of sepsis. Shock, 2005; 24,Suppl 1:S107–S119.

17.

Fiuza

, Suffredini

. Human models of innate immunity: Local and systemic inflammatory responses. J Endotoxin Res, 2001; 7:385–388.

18.

Lowry

. Human endotoxemia: A model for mechanistic insight and therapeutic targeting. Shock, 2005; 24,Suppl 1:S94–S100.

19.

Van der Poll

. Experimental human sepsis models. Drug Discovery Today: Disease Models, 2012; 9:e3–e9.

20.

Taylor

Jr . Staging of the pathophysiologic responses of the primate microvasculature to Escherichia coli and endotoxin: Examination of the elements of the compensated response and their links to the corresponding uncompensated lethal variants. Crit Care Med, 2001; 29,Suppl 7:S78–S89.

21.

van der Poll

. Immunotherapy of sepsis. Lancet Infect Dis, 2001; 1:165–174.

22.

Levi

, van der Poll

. Inflammation and coagulation. Crit Care Med, 2010; 38,Suppl 2:S26–S34.

23.

Taylor

Jr , Emerson

Jr , Jordan

et al. Antithrombin-III prevents the lethal effects of Escherichia coli infusion in baboons. Circ Shock, 1988; 26:227–235.

24.

Taylor

Jr , Chang

, Esmon

et al. Protein C prevents the coagulopathic and lethal effects of Escherichia coli infusion in the baboon. J Clin Invest, 1987; 79:918–925.

25.

Bernard

, Vincent

, Laterre

et al. Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med, 2001; 344:699–709.

26.

Angus

. Drotrecogin alfa (activated)…A sad final fizzle to a roller-coaster party. Crit Care, 2012; 16:107.

27.

Ranieri

, Thompson

, Barie

et al. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med, 2012; 366:2055–2064.

28.

Vincent

, Sakr

, Sprung

et al. Sepsis in European intensive care units: Results of the SOAP study. Crit Care Med, 2006; 34:344–353.

29.

Torres

, Rello

. Update in community-acquired and nosocomial pneumonia 2009. Am J Respir Crit Care Med, 2010; 181:782–787.

30.

Knapp

, Schultz

, van der Poll

. Pneumonia models and innate immunity to respiratory bacterial pathogens. Shock, 2005; 24,Suppl 1:S12–S18.

31.

Steinhauser

, Hogaboam

, Kunkel

et al. IL-10 is a major mediator of sepsis-induced impairment in lung antibacterial host defense. J Immunol, 1999; 162:392–399.

32.

Deng

, Cheng

, Newstead

et al. Sepsis-induced suppression of lung innate immunity is mediated by IRAK-M. J Clin Invest, 2006; 116:2532–2542.

33.

Prass

, Meisel

, Hoflich

et al. Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke T helper cell type 1-like immunostimulation. J Exp Med, 2003; 198:725–736.

34.

Renckens

, van Westerloo

, Roelofs

et al. Acute phase response impairs host defense against Pseudomonas aeruginosa pneumonia in mice. Crit Care Med, 2008; 36:580–587.

35.

van Westerloo

, Schultz

, Bruno

et al. Acute pancreatitis in mice impairs bacterial clearance from the lungs, whereas concurrent pneumonia prolongs the course of pancreatitis. Crit Care Med, 2004; 32:1997–2001.

36.

Maier

, Traeger

, Entleutner

et al. Cecal ligation and puncture versus colon ascendens stent peritonitis: Two distinct animal models for polymicrobial sepsis. Shock, 2004; 21:505–511.

37.

Parker

, Watkins

. Experimental models of gram-negative sepsis. Br J Surg, 2001; 88:22–30.

38.

Dejager

, Pinheiro

, Dejonckheere

, Libert

. Cecal ligation and puncture: The gold standard model for polymicrobial sepsis? Trends Microbiol, 2011; 19:198–208.

39.

van Zoelen

, Achouiti

, Schmidt

et al. Ligands of the receptor for advanced glycation end products, including high-mobility group box 1, limit bacterial dissemination during Escherichia coli peritonitis. Crit Care Med, 2010; 38:1414–1422.

40.

Mathiak

, Szewczyk

, Abdullah

et al. An improved clinically relevant sepsis model in the conscious rat. Crit Care Med, 2000; 28:1947–1952.

41.

Goldfarb

, Dellinger

, Parrillo

. Porcine models of severe sepsis: Emphasis on porcine peritonitis. Shock, 2005; 24,Suppl 1:S75–S81.

42.

Rijneveld

, Florquin

, Branger

et al. TNF-alpha compensates for the impaired host defense of IL-1 type I receptor-deficient mice during pneumococcal pneumonia. J Immunol, 2001; 167:5240–5246.

43.

Gerard

, Bruyns

, Marchant

et al. Interleukin 10 reduces the release of tumor necrosis factor and prevents lethality in experimental endotoxemia. J Exp Med, 1993; 177:547–550.

44.

van der Poll

, Marchant

, Keogh

et al. Interleukin-10 impairs host defense in murine pneumococcal pneumonia. J Infect Dis, 1996; 174:994–1000.

45.

Levi

, van der Poll

. Recombinant human activated protein C: Current insights into its mechanism of action. Crit Care, 2007; 11,Suppl 5:S3.

46.

Derhaschnig

, Reiter

, Knobl

et al. Recombinant human activated protein C (rhAPC; drotrecogin alfa [activated]) has minimal effect on markers of coagulation, fibrinolysis, and inflammation in acute human endotoxemia. Blood, 2003; 102:2093–2098.