Hypercapnia and Acidosis in Sepsis: A Double-edged Sword?

Severe sepsis associated with multiple organ failure, whether pulmonary or systemic in origin, is the leading cause of death in critically ill adults and children.21,22The incidence of sepsis-induced critical illness is 150 per 100,000 person-years in the United States alone.21Evidence suggests that approximately 40% of patients with severe sepsis develop ARDS,23and it has been estimated that sepsis-associated ARDS has an incidence of 45–63 cases per 100,000 person-years.24Furthermore, infection frequently complicates critical illness due to other causes, with an infection prevalence of more than 44% reported in this population.25Pneumonia and lower respiratory tract infection account for two thirds of nosocomial infections acquired in the intensive care unit.25Sepsis-induced ARDS is associated with the highest mortality rates.1,26,27 

Concerns have recently been raised regarding the safety of hypercapnia and/or acidosis in the context of sepsis.19,28The potent antiinflammatory effects of HCA underlie its protective effects in nonsepsis models of lung and systemic organ injury. However, immunocompetence is essential to an effective host response to microbial infection. Hypercapnia and/or acidosis may modulate the interaction between host and bacterial pathogen via  several mechanisms. HCA results in broad based suppression of many events that contribute to microbial killing, which could be detrimental to the host by facilitating bacterial spread and replication. Of further concern, the recent findings that HCA slows resealing of stretch-induced cell membrane injuries29and inhibits repair of pulmonary epithelial wounds30raise the potential that it could reduce the barrier to access of bacteria from the lung to the bloodstream.

The effects of hypercapnia and its associated acidosis may be a combination of the effects of hypercapnia and acidosis per se . Optimal anaerobic Escherichia coli  growth occurs at a carbon dioxide tension of 0.05 atm, which is similar to the Pco²in the gut, which is the usual environment in which E. coli  live. The aerobic growth rate of E. coli  was not inhibited by a Pco²of 0.2 atm but was inhibited at partial pressures of more than 0.6 atm.31,E. Coli  colony counts were halved when exposed to 350 atm of carbon dioxide for 20 min,32which is a massive dose. This inhibitory effect of high concentrations of carbon dioxide is mediated via  a direct effect,33although the exact mechanism is unclear. It does seem that carbon dioxide has broadly similar effects within the various families of microorganisms, but the sensitivity to carbon dioxide varies across the families, that is, yeasts are quite resistant to the inhibitory effects of carbon dioxide, gram-positive organisms are somewhat less resistant, and gram-negative organisms are the most vulnerable.34However, it is important to recognize that the levels of carbon dioxide used in these studies to inhibit bacterial growth are extremely high in the context of human physiology.

Pugin et al.  35have demonstrated that more clinically relevant degrees of metabolic acidosis can directly enhance bacterial proliferation in vitro . Cultured lung epithelial cells exposed to cyclic stretch similar to that seen with mechanical ventilation produced a lactic acidosis that markedly enhanced the growth of E. coli .35This was a direct effect of hydrogen ions, because direct acidification of the culture medium to a pH of 7.2 with hydrochloric acid enhanced E. coli  growth. In contrast, alkalinizing the pH of conditioned media from stretched lung cells abolished the enhancement of E. coli  growth effect. A range of gram-positive and gram-negative bacteria (including E. coli , Proteus mirabilis , Serratia rubidaea , Klebsiella pneumoniae , Enterococcus faecalis , and Pseudomonas aeruginosa ) isolated from patients with ventilator-associated pneumonia grew better in acidified media (fig. 1).35Interestingly, this effect was not seen with a methicillin-resistant Staphylococcus aureus  strain, which seemed to grow best at an alkaline pH (fig. 1).35 

The effects of HCA on bacterial proliferation at levels encountered in the context of permissive hypercapnia are unclear but may be a combination of the effects of the acidosis and of the hypercapnia. However, the inhibitory effects of carbon dioxide on bacterial growth have been demonstrated only at doses well beyond the clinically relevant range. In addition, the type of acidosis—respiratory versus  metabolic—may have differential effects on bacterial growth. However, the demonstration that clinically relevant levels of metabolic acidosis enhance bacterial growth is of concern.

HCA modulates diverse components of the host immune response, including cytokine and chemokine signaling, neutrophil and macrophage function, complement activation, and the adaptive immune response. The effects of HCA on the immune response may predominantly be a function of the acidosis rather than the hypercapnia per se .36However, the type of acidosis, that is, the fact that this is a hypercapnic rather than a metabolic acidosis seems to be important. As a specific example, although HCA inhibits neutrophil function in vitro ,37metabolic acidosis seems to inhibit function in some studies,38,39while activating neutrophils in other studies.40Methodologic differences may explain the differences in these findings, because incubation of cells in strong acids may have caused injury.40Similarly, the infusion of hydrochloric acid into the bloodstream in vivo  has been demonstrated to cause direct injury and activation of the inflammatory response.41For these reasons, we will focus on studies in which the effects of HCA have been determined and include studies of metabolic acidosis where these studies yielded important insights not otherwise available and/or where the acidosis was not generated using hyperosmolar strong acids.

HCA interferes with coordination of the immune response by reducing cytokine signaling between immune effector cells.16,18,42HCA reduces neutrophil16and macrophage18,42production of proinflammatory cytokines such as TNF-α, IL-1β, IL-8, and IL-6. HCA reduced endotoxin-stimulated macrophage release of TNF-α and IL-1βin vitro .18,42Peritoneal macrophages incubated under hypercapnic conditions demonstrated reduced endotoxin-stimulated TNF-α and IL-1β release. TNF-α inhibition was seen despite normal levels of TNF message, required more than 30 min of carbon dioxide exposure, and persisted after withdrawal of carbon dioxide.42This effect of HCA on cytokine production seems to be sustained. Intraperitoneal macrophages demonstrated impaired TNF-α production for up to 3 days after exposure to hypercapnia.43,44The mechanism underlying this inhibition of cytokine and chemokine production seems to be mediated at least in part via  the inhibition of nuclear factor-κB, a pivotal transcriptional regulator in the setting of inflammation, injury, and repair.45 

The complement system is an essential component of the innate immune response, and its activation targets pathogens for phagocytosis or lysis. Acidosis, whether hypercapnic or metabolic, seems to activate the complement system. Emeis et al.  46demonstrated that HCA and lactic and hydrochloric acid-induced acidosis all activate C3 and C5, two key components of the complement system. This activation of complement seems to be a direct effect of the acidosis per se .46The implications of HCA-mediated activation of complement, whether beneficial through enhanced complement-mediated bacterial killing or detrimental due to complement depletion or nonspecific actions of activated complement components, is not clear.

Neutrophils and macrophages are important effectors of the innate immune response in the setting of bacterial infection. Neutrophils rapidly migrate from the bloodstream to areas of infection throughout the body. Once in contact with bacteria, neutrophils rapidly phagocytose them. Neutrophils contain granules in the cytoplasm, which contain a range of destructive enzymes, including elastases and proteases, nicotinamide adenine dinucleotide phosphate oxidase, which generates superoxide and hydrogen peroxide, and myeloperoxide, which generates hypochlorous acid. These granules rapidly fuse with phagosomes containing bacteria to lyse and destroy them. Reduction of neutrophil migration to infection sites is associated with a poor outcome in sepsis.47 

Tissue macrophages and their blood-borne monocyte counterparts are of particular importance in coordinating the activation of the lymphocyte system in the setting of infection by presenting foreign antigen to the appropriate cells and secreting chemokines. Macrophages are activated by bacteria or by molecules such as endotoxin or complement components. Both monocytes and macrophages phagocytose and kill pathogens by similar mechanisms but at a slower rate than neutrophils.

HCA may impact on the cellular immune response via  several mechanisms. HCA may indirectly impair the cellular immune response by impairing the cytokine signaling between immune effector cells necessary to the coordination of the response to pathogens. Direct effects on the cellular immune response include (1) reduced phagocyte migration, chemotaxis, and adhesion; (2) decreased phagocytic ability; (3) reduced oxidant-mediated bacterial killing; and (4) altered mechanisms of neutrophil death. These effects may be mediated at least in part by disruption of phagocyte intracellular pH regulation.

Maintenance of intracellular pH in the physiologic range (6.8–7.3) is necessary for normal immune function and also for other cellular functions, such as proliferation, differentiation, apoptosis, migration, cytoskeletal organization, and maintenance of cell volume.48,49Two main transport systems regulate intracellular pH in immune cells, namely the Na/H exchanger and the plasmalemmal V-type H ATPase.48,49Intracellular neutrophil pH decreases when neutrophils are activated.17,50Where milieu pH is normal, there tends to be a rapid recovery in neutrophil cytosol pH back toward neutral or even alkaline pH.51Decreases in pH and increases in carbon dioxide in the local milieu, as might be encountered with HCA, result in a rapid decrease in neutrophil cytosolic pH.16,52Studies of the effects of metabolic acidosis demonstrate the potential for intracellular acid loading to decrease neutrophil spreading,52,53motility and migration,51and chemotaxis.54,55HCA may, therefore, reduce neutrophil and macrophage recruitment to the site of a septic insult via  this mechanism. Furthermore, the inhibition of carbon dioxide induced changes in neutrophil cytosol pH by acetazolamide16and buffering of intracellular pH abolish neutrophil inhibition.36 

Neutrophil margination within the bloodstream, adhesion to the capillary wall, and migration and recruitment to the site of injury is an important step in the host response.56The ability of cells to migrate is dependent on several well-defined processes, such as cytoskeleton remodeling, membrane recycling by endocytosis and exocytosis, detachment and reformation of focal adhesions mediated by integrins, and cell volume regulation. HCA has been demonstrated to inhibit the neutrophil expression of the chemokines, selectins, and intercellular adhesion molecules16,45that facilitate neutrophil binding to the endothelium and migration out of the vascular system. The potential for HCA to inhibit neutrophil chemotaxis and migration to the site of injury has been confirmed in vivo , where HCA inhibits pulmonary neutrophil infiltration in response to endotoxin instillation.11 

Macrophages and neutrophils phagocytose bacteria and internalize them into the phagosomes. The phagosome subsequently fuses with endosomes and lysosomes containing lytic enzymes that digest the bacteria. Acidosis, whether hypercapnic or metabolic, impairs neutrophil and macrophage phagocytic ability. HCA has been demonstrated to decrease cytokine production by both stimulated and unstimulated macrophages.18,42,43Metabolic acidosis depresses macrophage phagocytosis,57slows internalization of bacteria, and decreases intracellular bacterial killing.58HCA has also been demonstrated to directly impair neutrophil phagocytosis in vitro .37This inhibitory effect seems to be a function of the acidosis, with buffering restoring neutrophil phagocytosis.36 

Neutrophil and macrophage production of free radicals, such as superoxide, hydrogen peroxide, and hypochlorous acid, produced during the “respiratory burst” in response to immune activation is a major mechanism by which phagocytes kill bacteria.59In these cells, the enzyme nicotinamide adenine dinucleotide phosphate oxidase, which produces superoxide in response to microbial invasion, is markedly pH sensitive and displays a pH optimum of 7.0–7.5.60Reduction of cytosolic pH impairs intracellular enzyme function and decreases free radical production. Macrophage superoxide radical release decreases linearly at an intracellular pH less than 6.8.61HCA inhibits the generation of oxidants such as superoxide by unstimulated neutrophils and by neutrophils stimulated with opsonized E. coli  or with phorbol esters.16In contrast, hypocapnic alkalosis stimulated neutrophil oxidant generation.16Carbon dioxide modulation of neutrophil oxidant generation seems to be pH mediated, because inhibition in intracellular pH changes with acetazolamide attenuated this effect. HCA has also been demonstrated to reduce macrophage superoxide production.62 

Neutrophils have a short lifespan and tend to die within 48 h of release into the circulation, by undergoing programmed cell death, that is, apoptosis. Apoptosis seems to be a normal and appropriate fate of a neutrophil after phagocytic activity, whereas neutrophil death via  necrosis causes release of intracellular contents, including harmful enzymes, which can cause tissue destruction. Neutrophils seem to have an increased probability of dying by necrosis after intracellular acidification during phagocytosis.17HCA may, therefore, increase the probability of neutrophil cell death occurring via  necrosis rather than apoptosis.

Much of the focus of the effect of acidosis on the adaptive immune response has been on the implications for cancer, because the tumor microenvironment is characterized by poor vascularization, tissue hypoxia, and acidosis.63In a situation analogous to sepsis, acidosis in this setting may hamper the host immune response to tumor cells, potentially leading to increased tumor growth and spread. The cytotoxic activity of human lymphokine activated killer cells64and natural killer cells65is diminished at acidic pH. Metabolic acidosis reduces lysis of various tumor cell lines by cytotoxic T lymphocytes66and also inhibits IL-2-stimulated lymphocyte proliferation.67In contrast, the motility of IL-2-stimulated lymphocytes seems to be stimulated in the presence of an acidified extracellular matrix.68Severe extracellular acidosis (pH 6.5) also seems to enhance the antigen-presenting capacity of dendritic cells.69The net effect of these contrasting actions of metabolic acidosis on the adaptive immune response is unclear. However, the demonstration that HCA enhances systemic tumor spread in a murine model70raises clear concerns regarding the potential for HCA to suppress cell-mediated immunity.

The mechanisms by which bacterial infection produces lung and systemic organ injury are complex and involve contributions from the host response, injury due to bacterial exotoxins, endotoxins, and other components of the bacterial wall, and direct bacterial-induced injury caused by spread and tissue invasion. The host response to sepsis, while central to immunocompetence, may also play an important role in the pathogenesis of organ injury. In addition, the relative contributions of different processes involved in causing tissue injury may vary depending on the stage of the infective process.

The effects of hypercapnia in the context of infection may result in part from its modulation of the host immune response. As discussed earlier, HCA results in broad-based suppression of many events necessary for effective microbial killing. Central to the effects of HCA in the context of bacterial infection are its effects on the cellular immune system, particularly its effects on phagocytes, and on the coordination of the immune response. The phagocytic activity of neutrophils and macrophages is necessary for an effective host response to invading bacteria. However, the neutrophil armory of cytotoxic agents also has the potential to escape into the extracellular space and damage host tissue71and contribute to the development of lung injury. The protective effects of HCA in nonseptic injury models seems to be mediated at least in part via  inhibition of neutrophil recruitment and/or neutrophil function.11These findings are consistent with previous observations that there is decreased severity of ALI in neutrophil depleted animal models,72–74and the finding that the severity of ALI worsens in neutropenic humans on recovery of the neutrophil count.75The potential deleterious effect of this HCA-induced alteration in neutrophil function in the presence of live bacterial sepsis is highlighted by the fact that defects in neutrophil function are associated with increased sepsis severity76and worsened outcome.47 

The effects of HCA on bacterial injury may vary—from benefit to harm—depending on the stage of injury process, that is, depending on whether it is an early or established infection or whether the infection is prolonged. Early bacterial pneumonia is accompanied by a vigorous host inflammatory response. The antiinflammatory effects of HCA may reduce the magnitude of the host inflammatory response, thereby ameliorating host-induced tissue damage. HCA has also been demonstrated to reduce injury produced by bacterial endotoxin.11Therefore, in the setting of an early bacterial infection, HCA might reduce lung and systemic organ injury (fig. 2). In contrast, in late or prolonged bacterial infection, direct bacterial tissue invasion and spread may play a greater role. The immunosuppressive effects of HCA, particularly neutrophil inhibition, might impair bactericidal host responses. Overall, this would cause greater tissue damage in the setting of an established injury where a large bacterial load may exist. These effects would negate any protective effects of reduced host mediated tissue damage. This potentially deleterious effect of HCA has important clinical implications given that many critically ill patients already have well-established infections at presentation, and the fact that they may be exposed to prolonged hypercapnia in the context of protective lung ventilation strategies (fig. 2).

The effect of hypercapnia in early pneumonia seems to depend on the severity of the injury. HCA did not modulate the development of evolving lung injury induced by E. coli  pneumonia of moderate severity.77However, in the setting of more severe evolving pneumonia, produced via  a higher intrapulmonary E. coli  inoculate, HCA reduced the severity of the lung injury compared with normocapnic conditions.78Of interest, the protective effects of HCA in evolving pneumonia-induced lung injury seem to be independent of any effect on neutrophil function.78Of importance, HCA did not increase pulmonary bacterial load in early pneumonia, allaying concerns regarding the potential for HCA to retard killing and/or increase proliferation of bacteria.77,78 

HCA also reduces the severity of the lung injury caused by a more established bacterial-induced pneumonia,79a model more closely analogous to that seen in the clinical setting. When HCA was introduced several hours after the induction of an E. coli -induced lung injury, it reduced the severity of subsequent lung damage.79Of importance, although these protective effects of HCA were greater in the presence of appropriate antibiotic therapy (fig. 3A), HCA also reduced pneumonia-induced injury in the absence of antibiotic therapy (fig. 3B). Importantly, there was no evidence to suggest that HCA increased bacterial load, as evidenced by similar levels of bacteria recovered from the lungs of animals exposed to both HCA and normocapnia.79These findings again offer reassurance regarding the use of HCA in situations of established infection.

In contrast, hypercapnia increases the severity of lung injury caused by prolonged E. coli  pneumonia.37Animals exposed to environmental hypercapnia for 48 h after intrapulmonary E. coli  inoculation developed more severe lung damage, as evidenced by a greater reduction in pulmonary compliance, increased histologic injury, and greater alveolar neutrophil infiltration in comparison with similarly treated animals maintained under normocapnic conditions37(figs. 4A and B). Of particular concern, prolonged exposure to environmental hypercapnia increased bacterial load, as evidenced by increased levels of bacteria in the lungs of animals exposed to hypercapnia compared with normocapnia (fig. 4C). The mechanism underlying this effect seemed to involve the inhibition of neutrophil function, because neutrophils isolated from hypercapnic rats demonstrated impaired phagocytosis ability (fig. 4D).37Of importance to the clinical context, the use of appropriate antibiotic therapy abolished these deleterious effects of hypercapnia, reducing lung damage and lung bacterial load to levels comparable with that seen with normocapnia.

The deleterious effects of HCA in the setting of prolonged pneumonia are mediated, at least in part, via  impairment of neutrophil phagocytosis.37In contrast, this inhibitory effect of HCA on neutrophils seems less important in the setting of more acute pneumonia,78and it can be abrogated in the setting of prolonged pneumonia by treatment with appropriate antibiotic therapy. However, these findings raise concerns regarding the safety of prolonged exposure to hypercapnia in critically ill patients with pneumonia.

HCA reduces the severity of early septic shock and lung injury in the first 3 h after cecal ligation and puncture-induced fecal peritonitis.80HCA slowed the development of hypotension, preserved central venous oxygen saturations, and attenuated the increase in serum lactate when compared with normocapnia.80Central venous pressures did not change throughout the protocol, reducing the likelihood that differences in fluid volume status contributed to the hypotension in either group. HCA reduced the severity of lung injury, as evidenced by a reduction in the alveolar-arterial oxygen gradient, and reduced lung permeability, compared with normocapnia.80HCA reduced alveolar neutrophil infiltration but did not alter bronchoalveolar lavage IL-6 or TNF-α levels compared with normocapnia. There was no effect of HCA on bacterial load in the lung or in the bloodstream, whereas peritoneal fluid bacterial loads were also similar in both groups.

Wang et al.  81demonstrated that HCA improved the hemodynamic profile in a manner comparable with that seen with dobutamine in an ovine model of established fecal peritonitis. Fecal peritonitis was induced in anesthetized, invasively monitored, mechanically ventilated female sheep. Two hours after fecal spillage, animals were randomized to HCA, dobutamine infusion or control conditions, and followed up until animal death. Both HCA and dobutamine increased heart rate, cardiac index, and oxygen delivery, and reduced lactate concentrations compared with normocapnic animals. HCA, but not dobutamine, also reduced certain indices of lung injury, as assessed by lung wet–dry ratio, alveolar-arterial oxygen partial pressure difference and shunt fraction.81However, HCA did not increase survival time compared with normocapnia in this setting.

In contrast to the findings in prolonged pulmonary sepsis, environmental hypercapnia reduced the severity of prolonged systemic sepsis-induced lung damage.80Prolonged hypercapnia reduced histologic indices of lung injury in comparison with normocapnia. Of interest, hypercapnia did not reduce alveolar neutrophil infiltration or lung IL-6 or TNF-α concentrations. Furthermore, HCA did not alter survival from prolonged systemic sepsis. Reassuringly, there was no effect of HCA on bacterial load in the lung, the bloodstream or the peritoneal cavity.80 

A growing body of literature attests to the beneficial effects of direct intraabdominal administration of carbon dioxide—by means of a pneumoperitoneum—further supporting the safety and efficacy of HCA in abdominal sepsis. Carbon dioxide pneumoperitoneum improved survival compared with helium pneumoperitoneum in animals subjected to combined laparotomy and endotoxemia injury.82Insufflation of carbon dioxide into the peritoneal cavity before laparotomy for endotoxin contamination increases animal survival.83Most recently, carbon dioxide pneumoperitoneum has been demonstrated to increase survival in both mice and rabbits with polymicrobial peritonitis induced by cecal ligation and puncture (fig. 5).84,85These protective effects of intraperitoneal carbon dioxide insufflation seem to be due to the immunomodulatory effects of HCA,86which include an IL-10-mediated down-regulation of TNF-α.83Importantly, these effects seem to be mediated by the localized peritoneal acidosis produced rather than by any systemic effect.87,88 

The effects of hypercapnia in sepsis may be a function of the hypercapnia or the acidosis per se . As discussed, the effects of HCA on the immune response seem to be predominantly a function of the acidosis, rather than the hypercapnia per se , but the fact that the acidosis is hypercapnic rather than metabolic is of importance. The potential exists for hypercapnia to exert direct effects, independent of pH changes. A specific example is the binding of carbon dioxide to free amine groups on proteins to form carbamates, which can alter certain protein behavior or activity. The classic example is hemoglobin in which carbamino formation alters HbO²affinity. In addition, the potential for buffering of a HCA to modulate its effects in sepsis is also of importance.

Buffered hypercapnia, that is, hypercapnia in the presence of normal pH, seems to worsen lung injury induced by intrapulmonary bacterial instillation.36To avoid the confounding effects of the administration of exogenous acid and/or alkali, animals were first exposed to environmental hypercapnia until renal buffering had restored pH to the normal range. These animals were then subjected to intrapulmonary inoculation of E. coli , and the severity of lung injury produced during a 6-h period was compared with that seen in similarly inoculated animals exposed to normocapnia. Buffered hypercapnia significantly increased E. coli -induced lung injury when compared with normocapnic controls, as assessed by arterial oxygenation, lung compliance, proinflammatory pulmonary cytokine concentrations, and measurements of structural lung damage. Of interest, buffered hypercapnia did not reduce the phagocytic capacity of neutrophils and did not increase lung bacterial load. These findings contrast markedly with the protective effects of HCA in evolving bacterial pneumonia-induced lung injury.78The effect of buffered hypercapnia in the setting of more established or prolonged pneumonia has not been determined.

Buffered hypercapnia exerts differential effects compared with that seen with HCA in the setting of systemic sepsis. In these experiments, animals were again exposed to environmental hypercapnia until renal buffering had restored pH to the normal range. These animals were then subjected to cecal ligation and puncture, and the severity of hemodynamic compromise and lung injury was assessed over a 6-h period. Buffered hypercapnia attenuated the hemodynamic effects of evolving systemic sepsis induced by cecal ligation and puncture.89The extent of attenuation of shock was comparable with, if not greater than that seen with HCA.89However, buffered hypercapnia did not protect the lung from systemic sepsis-induced lung injury. Reassuringly, there was no evidence to suggest that buffered hypercapnia worsened the degree of lung injury compared with normocapnia, and buffered hypercapnia did not increase the bacterial load in the lungs or the bloodstream.89 

Hypercapnia constitutes an important component of protective lung ventilatory strategies. However, the generally beneficial effects of HCA in the setting of experimental nonseptic inflammatory injury contrast with a more complex spectrum of effects in the setting of live bacterial infection. Hypercapnia and/or acidosis exert diverse—and potentially conflicting—effects on the immune response, both humoral and cellular. Overall, HCA seems to suppress the immune response, although the net effect of its multiple actions seems to vary depending on the site of infection and also on whether the acidosis produced by the hypercapnia is buffered or not. HCA seems to protect the lung from injury induced by evolving or more established lung and systemic bacterial sepsis in relevant preclinical models. In contrast, the effects of HCA in prolonged untreated bacterial sepsis seem to differ depending on the source of the infection, with the immunosuppressive effects of HCA worsening lung injury in the setting of prolonged pneumonia. This deleterious effect is abrogated by effective antibiotic therapy. In contrast, HCA reduced lung damage caused by prolonged systemic sepsis, again highlighting the potential importance of the source of infection. Buffering of HCA seems to confer little benefit in the setting of experimental sepsis and worsens pneumonia induced lung injury.

Taken together, recent experimental findings in relevant preclinical models provide some reassurance regarding the safety of hypercapnia in sepsis, particularly in evolving and established pneumonia and in the setting of abdominal sepsis. However, in the setting of prolonged pneumonia, the immunosuppressive effects of hypercapnia remain a concern. Although the use of ventilation strategies resulting in hypercapnia is clearly justified in patients with ALI/ARDS, care is warranted in the setting of sepsis. The finding that deleterious effects of hypercapnia in the setting of prolonged pneumonia are abrogated by appropriate antibiotic therapy is of importance. Clinicians should, therefore, carefully consider the use of early empiric antibiotic therapy in hypercapnic ALI/ARDS patients in whom sepsis is suspected. However, concerns persist, particularly where antibiotic cover may be suboptimal, or the bacteria are more resistant to antibiotic therapy. The finding that hypercapnia may increase septic lung injury in the setting of prolonged pneumonia is also of relevance to other patient groups such as patients with infective exacerbations of chronic obstructive airways disease. Finally, buffering of the acidosis induced by hypercapnia does not confer any benefit and may worsen injury in the setting of pneumonia.