The brain stem and hypothalamus regulate the body's responses to stressful stimuli (2). There are several overlapping and interacting elements, each of which can have antiinflammatory actions. As noted in Table 1, central nervous system (CNS) regulation occurs through the hypothalamic–pituitary–adrenocortical (HPA) axis (producing cortisol and α-melanocyte-stimulating hormone [α-MSH]), the sympathetic-adrenomedullary axis (epinephrine and norepinephrine), the parasympathetic nervous system, and the thermoregulatory centers in the preoptic region of the hypothalamus (fever). Other elements of the systemic response include acute phase protein production, which occurs principally in the liver, and the elaboration of antiinflammatory cytokines (in particular, interleukins 4, -6, -10 and -13, IL-1 receptor antagonist, and soluble TNF receptors). Although there is interindividual and stimulus-dependent heterogeneity in their expression, for brevity we shall refer to these elements as the “systemic response.”

What triggers systemic responses? Central mediators that impinge directly on the hypothalamus are doubtless very important (2). Of the blood-borne activators, IL-6 may be the most significant; production and release of IL-6 are sensitive cellular reactions to injury or infection, and several authors have correlated the severity of tissue damage or infection with blood concentrations of this long-lived cytokine (see [3] and Table 1). Intravenous injections of low doses of TNF-α, IL-1β, and IL-12 can also trigger systemic responses (4); these cytokines, when released into the circulation from inflamed local sites, may be particularly important for mediating infection-related activation, which more often induces fever. On the other hand, systemic reactions to local infections or injury may not always be induced by mediators that act directly on the CNS. In rodents, subdiaphragmatic vagotomy reduces febrile and corticosterone responses to intraperitoneal (and, in some studies, intravenous) doses of lipopolysaccharide (LPS) or IL-1β, indicating that blood-borne mediators can act via the abdominal vagus to induce systemic responses (see references in Table 1). Other studies suggest that signaling the CNS from the periphery may not require blood-borne mediators at all. Levine and colleagues found that C-fiber (nociceptive) nerve afferents can inhibit inflammation at distant sites via a pathway that involves spinal cord afferents, the HPA axis, and corticosterone secretion. The same group found that noxious stimulation of a hindpaw decreased neutrophil movement into an inflamed distant site by inducing circulating neutrophils to shed l-selectin (5). Pain, a cardinal feature of local inflammation, thus may be linked to systemic responses that prevent inflammation in other tissues.

At the mild end of the severity spectrum, antiinflammatory components of the systemic response have been detected following cold exposure, strenuous exercise (6), and psychological stress (7). Much more extensive changes have been noted following major trauma (see below). There is interindividual and stimulus-related variability in how the elements of the response are expressed, however. For example, a proinflammatory cytokine response to anxiety has been reported (8), and there is strong evidence that fever and acute phase protein production are independently regulated. Although little is known about the systemic responses to early local infection in humans, we have found no evidence that they differ substantially from adaptations to other stresses, with the obvious exception that fever is more commonly a feature when infection is the stimulus. Variability in the systemic response was captured in the SIRS concept, which we believe was incorrectly termed “inflammatory” (1).

It is widely assumed that evolution shaped the body's immune defenses to contain and destroy microbes that invade through surface epithelia and enter host tissues. Both adaptive and nonadaptive (innate) defense mechanisms are important.

The adaptive defenses involve gene rearrangements, mature over many days, and produce highly specific T and B cell recognition of foreign antigens. The nonadaptive immune defenses, in contrast, are “hard-wired” in the genome, rapidly mobilized, and broadly microbicidal. Some nonadaptive mechanisms (natural antibodies, the alternative complement system) are preformed and ready to act without modification, whereas others require activation by non-self signals. In brief, molecular recognition of microbial carbohydrate or lipid moieties allows tissue macrophages to distinguish most invaders from self and engulf or kill them. When activated by microbial signals, these sentinel phagocytes also release mediators that induce local inflammation: they increase local blood flow, attract neutrophils to the site of infection, enhance leukocyte adhesion to (and diapedesis through) nearby vascular endothelium, promote microthrombosis in local vessels, and activate intracellular and extracellular mechanisms for killing microbes. Proinflammatory cytokines (TNF-α, IL-1β, IL-12, interferon-γ) and chemokines (IL-8 and others) play essential roles in local defense. Natural antibodies, the alternative complement pathway, and local inflammation are all innate mechanisms for preventing bacterial and fungal dissemination to uninfected tissues. In ways that at first may seem paradoxical, the systemic response also contributes to innate local defenses by preventing both microbial dissemination and self-inflicted injury to distant organs.

The systemic response raises the blood concentrations of several molecules (e.g., CRP, LBP, MBP) that assist in recognizing invading microbes, mobilizes leukocytes into the circulation, and increases blood flow to injured or infected sites. These actions favor the accumulation of effector molecules and leukocytes at inflamed local sites; in essence, they enhance local inflammation and antimicrobial defense. Concurrently, the systemic response prevents inflammation in uninvolved tissues by neutralizing inflammation-induced molecules (such as cytokines, proteases, and oxidants) that enter the bloodstream, by diminishing the proinflammatory responses of circulating leukocytes, and by forestalling endothelial activation in distant vessels; the latter action allows neutrophils to home to activated endothelium in the inflamed tissue. Although most previous discussions of innate immunity have not included these systemic mechanisms, we suggest that they are essential for both enhancing and containing local inflammatory and immune responses to invading microbes.

The evidence that local inflammation is normally accompanied by systemic antiinflammation is suggestive but inconclusive. Although the local tissue response to injury or infection is predominantly proinflammatory, most antiinflammatory or modulatory molecules are also produced within inflamed local sites. Indeed, as has generally been found for proinflammatory mediators (excepting some chemokines), the concentrations of antiinflammatory molecules are often much higher in an inflamed local site than in the blood. The blood also contains a complex mixture of mediators, including antiinflammatory modulators, proinflammatory mediators released from an inflamed local site, and molecules, such as macrophage migration inhibitory factor (MIF), that may be derived from distant tissues and have strong proinflammatory actions (9). Regulation of inflammation obviously must occur in both local and systemic compartments.

Defining the pro- versus antiinflammatory balance within particular body fluids has not been straightforward, however. Simply measuring the relative concentrations of proinflammatory molecules and their inhibitors, for example, gives only a partial view of the overall mediator mix in a particular fluid or tissue (10), especially since the concentration–response relationships for individual cytokines may be nonlinear in vivo (11) and it may be impossible to predict the integrated sum of multiple cytokines acting together (12).

Interpreting the impact of measured mediator concentrations is further complicated when mediators can have different actions in different body compartments. Complement factor 3a and its degradation product, C3a desArg, are examples of this phenomenon; they enhance LPS-induced proinflammatory cytokine synthesis in adherent monocytes (which may mimic monocyte-macrophages in local tissue sites) yet they prevent these responses when monocytes are suspended in medium (as would be the case in the blood) (13). The impact of IL-10 on LPS-stimulated monocytes may also depend upon whether the cells are adherent or in suspension when exposed to the cytokine (14). Perhaps the most striking example of context-dependent action is that of corticotropin-releasing hormone (CRH), which triggers the secretion of ACTH and α-MSH from the pituitary and therefore may be considered an antiinflammatory mediator in the brain. When CRH is released from peripheral nerves, in contrast, its actions are strikingly proinflammatory (2). Although mediators obviously defy classification as “pro-” or “anti”-inflammatory when they have different actions in different physical settings or toward different cell types (15), each of these observations is consistent with the notion that the local response is proinflammatory whereas that in the circulation is not.

Confronted with these difficulties in assessing the impact of the diverse mediators detected in body compartments, investigators have addressed the net mediator balance in a fluid by measuring its ability to stimulate, or inhibit the stimulation of, target cells in vitro. When Pugin and coworkers compared the inflammatory potency of fluids from local (inflamed lung) and systemic (blood) compartments in this way, they found net proinflammatory activity in the local compartment but not in the blood (16). Although the outcome of this test is doubtless influenced by both the target cells chosen for the assay and the experimental agonist used to stimulate them, these results are again consistent with a predominantly antiinflammatory systemic response.

Surgeons and intensivists have recognized for many years that trauma can induce immunosuppression. We suggest that intense activation of the normal systemic response, as occurs posttrauma, can induce this state of immune paralysis (“endogenous immunosuppression”). In traumatized patients, the blood levels of proinflammatory mediators such as TNF-α, IL-1β, and IL-12 can be undetectable or unchanged while the concentrations of IL-4, IL-6, IL-10, prostaglandin E₂ (PGE₂), and acute phase reactants are strikingly elevated (17-19). Stress-induced IL-10 is especially prominent (20). Circulating monocytes decrease their expression of class II antigen-presenting molecules and reduce their proinflammatory responses to bacterial agonists, yet their ability to make antiinflammatory molecules is often retained. Trauma-induced immunosuppression can also impair lymphocyte function and inhibit delayed hypersensitivity reactions. In contrast, circulating neutrophils obtained from traumatized individuals may express increased numbers of adherence molecules (CD11b), have prolonged survival, have diminished ability to produce IL-8 in response to bacterial agonists (21), and yet be sensitized (“primed”) to react vigorously to stimulation by inflammatory mediators (3); these changes may enhance the ability of neutrophils to home to, and function within, inflamed local tissues.

Even minor surgical procedures can trigger elevations in blood cortisol, IL-6 and IL-10 levels, and induce immune paralysis in blood leukocytes (22). More severe trauma has generally been associated with more dramatic and persistent changes (particularly in blood IL-6 levels and monocyte class II molecule expression); the intensity and duration of these findings have often correlated with increased risks of infection, sepsis, and death (23, 24). At its most intense, the systemic response seems capable of increasing the risk of bacterial and fungal infections, particularly when epithelial barriers have been breeched by wounds, incisions, or catheters.

Can counteracting one or more elements of the normal systemic response prevent endogenous immunosuppression without causing harm? Mannick and colleagues found that administering IL-12 to traumatized mice increased their ability to resist a bacterial challenge (25). Similar reasoning suggests that prophylactic usage of agents that promote phagocyte antimicrobial function should reduce the severity of immunosuppression, and therefore the incidence of nosocomial infection, in patients who have sustained major trauma. Unfortunately, administering interferon-γ did not prevent posttrauma infection in clinical trials. Despite these and many other efforts (see the review by Windsor and coworkers [26]), there is little convincing evidence that boosting human immune defenses can prevent nosocomial infection. Perhaps combining immunostimulatory agents (such as interferon-γ, G-CSF, or interleukin-12) with passive or active immunization (27) would reduce the incidence of nosocomial infections by enhancing both phagocyte function and opsonin availability. Much remains to be learned about the pathogenesis of endogenous immunosuppression (how does the systemic response interfere with local defenses? [28, 29]) and how to prevent it without doing harm.

The concepts discussed here do not offer a ready explanation for the occurrence of multiple organ dysfunction or septic shock in patients who develop nosocomial infection while they are endogenously immunosuppressed. It is worth noting, however, that antiinflammatory forces seem to dominate in the blood even when patients are very sick and have high blood levels of TNF-α and other proinflammatory mediators (20, 30). In addition, the normal systemic response engenders net procoagulant and antifibrinolytic changes in the circulating blood. In patients who have sustained minor trauma, these changes are only detectable using sensitive tests, whereas in those who experience severe sepsis, disseminated intravascular coagulation and thrombosis in small vessels likely contribute to multiple organ dysfunction (31). Enhanced blood coagulation is a highly conserved systemic response that is thought to provide a primitive antimicrobial defense mechanism (32); just as other elements of the systemic response can induce endogenous immunosuppression, extreme activation of the clotting arm of the systemic response may contribute to organ injury.

We have focused here on the pathogenesis of nosocomial infection in trauma patients, yet it is important to note that a large literature has addressed the relationship between various stresses and infection risk (reviewed in [33, 34]). Almost without exception, the results of these studies point to the ability of systemic mechanisms (in particular, catecholamines and glucocorticoids) to reduce resistance to viral and/or bacterial infection. Suppressive effects on both acquired (e.g., antibody responses to T-dependent antigens) and innate (susceptibility to infection with E. coli or Salmonella) immunity have been noted. Most of the studies have been done in vitro or in experimental animals, however. Of the few prospective clinical studies in humans, those documenting a correlation between psychological stress or strenuous exercise and the risk of respiratory virus infections have been the most convincing (6, 35). Although a causal relationship has not been established, the literature reviewed here suggests strongly that normal systemic responses can increase susceptibility to infection by both viral and bacterial agents. Further research is needed to define the precise host–pathogen interactions at play in different clinical settings.

Why should animals use cortisol, catecholamines, and cytokines to regulate processes as seemingly different as blood pressure, cellular metabolism, and host defense? One obvious answer is that these functions are highly interrelated. Since systemic inflammation would disrupt normal operation of the circulation and prevent successful adaptations to exercise, cold, and other stresses, the mechanisms that mediate these adaptations also prevent inflammatory reactions within the circulating blood. Preventing systemic inflammation is also necessary for effective host defense, in part because it helps leukocytes home to the local sites where they are needed to confront invading microbes. Unfortunately, stresses such as major trauma can so strongly activate the systemic response that it suppresses the local defenses that normally destroy invading commensal bacteria and fungi. Understanding the genetic contribution to individual variability in the systemic response may help identify patients at risk for immunosuppression and infection.

If the antiinflammatory systemic response plays such an important role in the pathogenesis of nosocomial infection in posttrauma patients, does it also contribute to susceptibility when the invading microbe is a pathogen, not a commensal, and when the patient is otherwise healthy? How important is endogenous immunosuppression for patients with cancer, arthritis, or other illnesses? These questions cannot be answered at this time. Many underlying diseases will doubtless modify the systemic response in significant ways, as may features of patient management. So the concepts proposed here will probably not apply to every individual, microbe, or predisposing condition. If it succeeds, this article will encourage experiments to test the applicability of these ideas to diverse clinical settings.

Inflammation. A response to injury, infection, or other stimuli that includes activation of leukocytes and vascular endothelium, transudation of fluid into tissue spaces, and homing of leukocytes (particularly neutrophils) to the affected site; it results in local hyperemia, warmth, edema, and pain and promotes antimicrobial host defenses. Proinflammatory mediators include cytokines (TNF-α, IL-1β, IL-12, IL-15, IL-18, interferon-γ), chemokines, bioactive lipids (leukotrienes, thromboxane, platelet-activating factor), and reactive oxygen and nitrogen metabolites. The actions of some mediators are context dependent (see text).

Antiinflammation. A response that inhibits the production or action of proinflammatory mediators, prevents or reduces phagocyte activation, and neutralizes the potentially toxic enzymes and metabolites produced during inflammation. Mediators with antiinflammatory actions include cytokines (IL-4, IL-6, IL-10, IL-11, IL-13, transforming growth factor-β), catecholamines, prostaglandin E₂, glucocorticoids, α-MSH, interleukin-1 receptor antagonist (IL-1Ra), and soluble TNF receptors (also see Table 1).