Bench-to-bedside review: Carbon dioxide

Carbon dioxide is a waste product of aerobic cellular respiration in all aerobic life forms. PaCO2 represents the balance between the carbon dioxide produced and that eliminated. Hypocapnia remains a common - and generally underappreciated - component of many disease states, including early asthma, high-altitude pulmonary edema, and acute lung injury. Induction of hypocapnia remains a common, if controversial, practice in both adults and children with acute brain injury. In contrast, hypercapnia has traditionally been avoided in order to keep parameters normal. More recently, advances in our understanding of the role of excessive tidal volume has prompted clinicians to use ventilation strategies that result in hypercapnia. Consequently, hypercapnia has become increasingly prevalent in the critically ill patient. Hypercapnia may play a beneficial role in the pathogenesis of inflammation and tissue injury, but may hinder the host response to sepsis and reduce repair. In contrast, hypocapnia may be a pathogenic entity in the setting of critical illness. The present paper reviews the current clinical status of low and high PaCO2 in the critically ill patient, discusses the insights gained to date from studies of carbon dioxide, identifies key concerns regarding hypocapnia and hypercapnia, and considers the potential clinical implications for the management of patients with acute lung injury.


Hypocapnia
Hypocapnia is common in several diseases (Table 1; for example, early asthma, high-altitude pulmonary edema, lung injury), is a common acid-base disturbance and a criterion for systemic infl ammatory response syndrome [9], and is a prognostic marker of adverse outcome in diabetic ketoacidosis [10]. Hypocapnia -often prolongedremains common in the management of adult [11] and pediatric [12] acute brain injury.
Carbon dioxide transport, sensing and molecular response CO 2 is carried in the blood as HCO 3 -, in combination with hemoglobin and plasma proteins, and in solution. Inside the cell, CO 2 interacts with H 2 O to produce carbonic acid (H 2 CO 3 ), which is in equilibrium with H + and HCO 3 -, a reaction catalyzed by carbonic anhydrase. CO 2 transport into cells is complex, and passive diff usion, specifi c transporters and rhesus proteins may all be involved. CO 2 is sensed in central and peripheral neurons. Changes in CO 2 and H + are sensed in chemosensitive neurons in the carotid body and in the hindbrain [13,14]. Whether CO 2 or the pH are preferentially sensed is unclear, but the ventilatory response to hypercapnic acidosis (HCA) exceeds that of an equivalent degree of metabolic acidosis [15], suggesting specifi c CO 2 sensing. Bicarbonate directly activates adenylate cyclase [16], increasing cAMP and activating protein kinase A, opening Ca 2+ channels and permitting infl ux of Ca 2+ [14]. In the glomus cells of the carotid nucleus, increased CO 2 levels activate Ca 2+ channels independent of the pH [17].
A key molecular mechanism by which hypercapnia may exert its eff ects, both benefi cial and deleterious, is through the NF-κB transcription factor. NF-κB is a major transcription factor that regulates genes responsible for immunity and infl ammation, including proinfl ammatory cytokines. An in vitro study has demonstrated that elevated CO 2 levels suppress expression of TNF and other cytokines by pulmonary artery endothelial cells via suppression of NF-κB activation [18]. Furthermore, hyper capnia inhibits pulmonary epithelial wound repair, also via an NF-κB mechanism [19].

Physiologic eff ects of CO 2
Th e physiologic eff ects of CO 2 are diverse and incompletely understood, with direct eff ects often counterbalanced by indirect eff ects. In addition, the net eff ect of hypocapnia or hypercapnia may occur as a function of the pH or CO 2 per se.

Respiratory system
CO 2 is important in matching regional lung ventilation to perfusion. Alveolar CO 2 increases local alveolar ventilation [20] via inhibition of conducting airway tone. Hypercapnia increases pulmonary vascular tone, potentiating hypoxic pulmonary vasoconstriction and further augment ing V/Q matching, but in some cases exacerbating pulmonary hypertension.
Hypocapnia can worsen ventilation-perfusion match ing and gas exchange in the lung via a number of mechanisms, including bronchoconstriction [21], reduc tion in collateral ventilation [22], reduction in paren chymal compliance [23], and attenuation of hypoxic pulmonary vaso constriction and increased intra pulmo nary shunting [24].

Central nervous system
CO 2 stimulates ventilation (see above). Peripheral chemoreceptors respond more rapidly than the central neurons, but central chemosensors make a larger contribution to stimulating ventilation. CO 2 increases cerebral blood fl ow (CBF) by 1 to 2 ml/100 g/minute per 1 mmHg in PaCO 2 [25], an eff ect mediated by pH rather than by the partial pressure of CO 2 . Hypercapnia elevates both the partial pressure of O 2 in the blood and CBF, and reducing PaCO 2 to 20 to 25 mmHg decreases CBF by 40 to 50% [26]. Th e eff ect of CO 2 on CBF is far larger than its eff ect on the cerebral blood volume. During sustained hypocapnia, CBF recovers to within 10% baseline by 4 hours; and because lowered HCO 3 returns the pH towards normal, abrupt normalization of CO 2 results in (net) alkalemia and risks rebound hyperemia.
Hypocapnia increases both neuronal excitability and excitatory (glutamatergic) synaptic transmission, and suppresses GABA-A-mediated inhibition, resulting in increased O 2 consumption and uncoupling of metabolism to CBF [27].

Cardiovascular system
Hypercapnia directly inhibits cardiac and vascular muscle contractility, eff ects that are counterbalanced by sympathoadrenal increases in heart rate and contractility, increasing the cardiac output overall [28]. Th e partial pressure of O 2 in the blood is increased because increased cardiac output coupled with reduced intra pulmonary shunt increases O 2 delivery, and tissue release of O 2 is augmented because hypercapnia and acidemia shift the hemoglobin-O 2 curve rightward. In addition, HCA increases the O 2 tension in both subcutaneous tissues and in the intestinal wall [29]. Indeed, a large body of evidence now attests to the ability of hypercapnia to increase peripheral tissue oxygenation, independently of its eff ects on cardiac output [30,31]. In experimental polymicrobial sepsis in female sheep, HCA improved tissue oxygenation and reduced lung edema formation more than dobutamine administration [32]. A study in white rabbits ascertained that 150 mmHg was the permissive upper limit of acute hypercapnia with respect to improvement of tissue perfusion and oxygenation [33]. Finally, a further positive eff ect of HCA on bioenergetics is evident in the form of a reduction in cellular O 2 consumption [34] and hypercapnia induced mitigation of the fall in gut ATP during endotoxemia in rats, pointing to improvement in energy metabolism in this setting. Th e net impact is thus increased O 2 supply and less demand. In contrast, hypocapnia does the opposite.

Carbon dioxide -insights from the bench
Experimental studies provide important preclinical information on the eff ects and mechanisms of action of CO 2 .

Acute lung injury
HCA is protective in many models of ALI. Although hypercapnia reduces the severity of overwhelming experimental ventilator-induced lung injury [35], its eff ects in milder injury are modest [36] and it may not protect in extensive atelectasis [37]. Hypercapnia inhibits hypoxia-induced chronic pulmonary hypertension in adult and newborn rodents [38,39], and protects against chronic neonatal lung injury [40]. Th e benefi cial eff ects of HCA in such models are increasingly well understood, and include attenuation of lung neutrophil recruitment, pulmonary and systemic cytokine concentrations, cell apoptosis, and O 2 -derived and nitrogen-derived free radical injury.
Concern has been raised regarding the potential for the anti-infl ammatory eff ects of HCA to impair the host response to infection. In early pulmonary infection, this potential impairment does not appear to occur, with HCA reducing the severity of acute-severe Escherichia coli pneumonia-induced ALI [41]. In the setting of more established E. coli pneumonia, HCA is also protective [42]. Of concern, HCA worsens the severity of prolonged bacterial pneu monia by a mechanism involving reduced bacterial killing [43]. In contrast, HCA reduces the severity of lung injury and hemodynamic compromise caused by cecal ligation and puncture-induced polymicrobial systemic sepsis [44,45]. Th e eff ects of HCA in sepsis therefore appear to depend on the duration of infection and on the site of infection. Other potential adverse eff ects of hypercapnia may include impairment of alveolar epithelial function, leading to reduced edema clearance [46]. Lastly, HCA may delay cellular repair and wound healing [19], slowing recovery and healing following ALI/ARDS.
Hypocapnia increases microvascular permeability and impairs alveolar fl uid reabsorption in the isolated rat lung, due to an associated decrease in Na/K-ATPase activity [47]. Th ese eff ects may be important in the pathogenesis of pulmonary edema. Experimental hypocapnia causes profound acute parenchymal lung injury that may be ameliorated by normalization of alveolar CO 2 by adding inspired CO 2 [48]; it also worsens ischemiareperfusion-induced lung injury [49].

Myocardial and vascular injury
HCA protects the heart following ischemia-reperfusion injury. Reperfusion with a hypercapnia acidotic perfusate enhances recovery of myocardial function following prolonged ischemia ex vivo as well as in vivo [50]. In experimental polymicrobial sepsis in female sheep, HCA improved tissue oxygenation and reduced lung edema formation more than dobutamine administration [32]. Hypocapnia worsens ischemic injury in the neonatal lamb myocardium [51] and abolishes the protective eff ects of preconditioning.

Central nervous system
Hypercapnia attenuates hypoxic-ischemic brain injury in the immature rat [52] and protects the porcine brain from reoxygenation injury by attenuation of free radical action. Hypercapnia increases the size of the region at risk of infarction in experimental acute focal ischemia; in hypoxic-ischemic injury in the immature rat brain, hypocapnia worsens the histologic magnitude of stroke [52] and is associated with a decrease in CBF to the hypoxia-injured brain as well as disturbance of glucose utilization and phosphate reserves. Hypocapnia during resuscitation increases functional and histologic evidence of brain injury following experimental cardiac arrest in dogs [53]. Hypocapnia further exacerbates the cerebral O 2 supply:demand imbalance by increasing neuronal excitability, increasing excitatory synaptic transmission, and via a direct eff ect on the neuronal membrane itself [54]. Severe hypocapnia increases N-methyl-d-aspartic acid receptor-mediated neurotoxicity in the newborn piglet and increases neuronal dopamine, particularly in the striatum, which may worsen reperfusion injury, especially in the immature brain [55]. Indeed, hypocapnia may be directly neurotoxic, through increased incorporation of choline into membrane phospholipids [56].

Potential benefi ts
Th ere are some potential benefi ts of acute hypocapnic alkalosis in specifi c critically ill patients [57]. For patients who have life-threatening elevations in intracranial pressure, rapid induction of hypocapnia for short durations may prevent brainstem herniation, allowing defi nitive diagnosis and therapy to be instituted. Hypocapnia may also be indicated in neonates with pulmonary hypertensive crises. Th e knowledge of the dramatic deleterious eff ects of hypocapnia on the neonatal brain, together with the known adverse eff ects of excessive lung stretch, however, have led to the use of alternative measures in this regard. Th ere are very few other situations where hypocapnia is of benefi t.

Potential risks
Hypocapnia may be a pathogenic entity in the setting of critical illness, particularly in ARDS patients. Edmunds and Holm demonstrated more than 30 years ago that alveolar hypocapnia produces hemorrhagic consolidation in the lung, and that attenuation of such adverse eff ects could be achieved by addition of inhaled CO 2 [48]. In the clinical context, Trimble and colleagues in 1971 demonstrated that hypocapnia increased airway resistance, increased work of breathing, worsened ventilation/ perfusion matching, increased the alveolar-arterial O 2 gradient and decreased the partial pressure of O 2 in the blood in ARDS patients, and that administration of CO 2 (that is, therapeutic hypercapnia) improved systemic oxygenation and reduced the shunt fraction [58]. Both hyperventilation and hypocapnia have been identifi ed as independent determinants of long-term pulmonary dysfunction in patients with bronchopulmonary dysplasia, as well as being implicated in the pathogenesis of asthma.
In traumatic brain injury, sustained hypocapnia may exacerbate cerebral hypoperfusion. Hypocapnia-induced cerebral vasoconstriction may worsen cerebral vasospasm in these patients, and has been demonstrated to exacerbate pre-existing impairment of CBF and metabolism in patients with traumatic brain injury. Coles and colleagues demonstrated that moderate hypocapnia (<34 mmHg) can signifi cantly reduce global CBF and result in signifi cant increases in the volume of critically hypoperfused tissue in the injured brain [27]. Prophylactic hyperventilation of head-injured patients, formerly employed in order to reduce intracranial pressure, worsens outcome [59]. In children with severe head injury, hypocapnia-induced critical cerebral ischemia has also been demonstrated. In the setting of resuscitation following cardiac arrest, hypocapnia induces cerebral ischemia in the post-resuscitation period [60]. Hypocapnic aklalosis has also been associated with poor prognosis in patients with acute cerebrovascular accidents.
In the critically ill patient, hypocapnia has been clearly linked to the development of arrhythmias, adversely aff ects the myocardial supply:demand balance [61], increases the risk of coronary vasospasm [62], and may worsen the progression of acute coronary syndromes.

Potential benefi ts
Th ere are no clinical studies of hypercapnia independent of V T in mechanical ventilation strategies. Indirect evidence exists in humans, however, for a protective eff ect of HCA independent of V T . Kregenow and colleagues examined mortality as a function of permissive hypercapnia in patients enrolled in the ARDS Network V T study [4,8]. Using multivariate logistic regression analysis, and controlling for other co-morbidities and severity of lung injury, they demonstrated that permissive hypercapnia reduced mortality in patients randomized to the higher V T (12 ml/kg) [8]. Th ere was, however, no additional protective eff ect of permissive hypercapnia in patients randomized to receive the lower V T (6 ml/kg) [8].
In children with congenital heart disease undergoing cardiopulmonary bypass, blood gas management with a pH-stat strategy, which results in higher PaCO 2 levels, reduces postoperative morbidity [63].

Potential risks
Rapid induction of hypercapnia in the critically ill patient may have adverse eff ects. Acute hypercapnia impairs myocardial function. HCA may induce vasoconstriction of the pulmonary vascular bed, leading to right ventricular systolic overload; and in patients with ARDS, HCA may exacerbate hypoxic vasoconstriction. In patients with severe ARDS, HCA -induced by V T reduction and increases in positive end-expiratory pressure -impaired right ventricular function and hemodynamics despite positive eff ects on oxygenation and alveolar recruitment [64]. An adverse eff ect of hypercapnia on skeletal muscle function, and diaphragmatic function in particular, is a concern.
Hypercapnia-induced increases in CBF and cerebral blood volume may adversely impact on intracranial pressure in patients with traumatic brain injury. An association between intraventricular hemorrhage and severe, but not mild, hypercapnia has been reported in retrospective studies.
Defi ning a safe and effi cacious threshold of hypercapnia remains an elusive goal. Although benefi cial eff ects, including tissue oxygenation, may have a ceiling level in animals [33], similar fi ndings have not been reported in humans. Evidence of a temporal limit to benefi cial eff ects undermines this concept further -timing (that is, adaptation) may be as important as the degree of severity. Attempts to specify such a value are problematic outside the clinical context. Clinicians must be mindful of the tradeoff between the benefi cial and deleterious eff ects of hypercapnia as outlined, and must tailor treatment in each individual case; for example, in the case of combined lung and head injury, regional monitors of cerebral oxygenation and intracranial pressure may be used to guide therapy.

Therapeutic hypercapnia in the critically ill patient
Despite extensive eff ort over the past decade, particularly in the experimental setting, the ideal target population for trials of therapeutic hypercapnia (that is, administration of CO 2 to the ventilator breathing circuit) remains somewhat ill-defi ned. Given the immunosuppressive eff ects of HCA, and its potential to retard reparative processes, HCA may ultimately prove its utility as a temporary and brief measure, designed to limit predictable and transient organ injury. In that respect, therapeutic hypercapnia during or immediately post cardiopulmonary bypass would appear to hold some promise. HCA is protective in numerous experimental models of organ ischemia-reperfusion, and recent clinical studies have shown improved systemic oxygenation with the use of thera peutic hypercapnia after bidirectional superior cavo pulmonary anastomosis in children [65].

Hypercapnia in the critically ill patient -role of buff ering
In patients managed with protective ventilation strategies, buff ering of the acidosis induced by hypercapnia remains a common -albeit controversial -clinical practice. Buff ering with sodium bicarbonate was permitted in the ARDS Network V T study [4]. Th e need to consider the eff ects of buff ering HCA is emphasized by the fact that both hypercapnia and acidosis per se may exert distinct biologic eff ects. Th ere is no evidence to support buff ering, however, and a number of specifi c concerns exist regarding this practice. Th e protective eff ects of HCA in experimental lung injury are a function of the acidosis, rather than the elevated CO 2 per se [66], and therefore buff ering may simply ablate any protective eff ects. In experimental lung injury induced by E. coli or endotoxin, renal buff ering of hypercapnia signifi cantly worsened physiological and histological measurements of injury [67].
Specifi c concerns exist regarding sodium bicarbonate, the buff er used most frequently in the clinical setting. Th e eff ectiveness of bicarbonate infusion as a buff er is dependent on the ability to excrete CO 2 , rendering it less eff ective in buff ering HCA. In fact, bicarbonate may further raise PaCO 2 where alveolar ventilation is limited, such as in ARDS. While bicarbonate may correct the arterial pH, it may worsen an intracellular acidosis because the CO 2 produced when bicarbonate reacts with metabolic acids diff uses readily across cell membranes, whereas bicarbonate cannot.
Th ere may be a role for alternative buff ers, such as the amino alcohol tromethamine (tris-hydroxymethyl aminomethane (THAM)). THAM penetrates cells easily and can buff er pH changes and simultaneously reduce the partial pressure of CO 2 [68]. Unlike bicarbonate, which requires an open system for CO 2 elimination in order to exert its buff ering eff ect, THAM is eff ective in a closed or semi-closed system [68]. THAM rapidly restores pH and acid-base regulation in acidemia caused by CO 2 retention [68]. In ARDS patients, THAM attenu ates the hemodynamic consequences of a rapidly induced HCA.
In summary, if a clinician elects to buff er HCA, the rationale for this practice should be clear -for example, to ameliorate potentially deleterious hemodynamic conse quences of acidosis -and THAM should be considered rather than bicarbonate.

Summary and conclusions
Th e importance and complexity of inter-relationships between alterations in systemic CO 2 tension and critical illness states is increasingly appreciated. Ventilator strategies involving hypercapnia are widely utilized in the critically ill adult and child, with the aim of realizing the benefi ts of reduced lung stretch. Th e potential for hypercapnia to directly contribute to the benefi cial eff ects of protective lung ventilatory strategies is clear from experimental studies demonstrating protective eff ects in models of acute lung and systemic organ injury. Concerns persist, however, regarding the potential for hypercapnia and/or acidosis to exert deleterious eff ects, and the need for caution prior to extrapolation to the clinical context must be emphasized.
Hypocapnia is an underappreciated phenomenon in the critically ill patient, and is potentially deleterious, particularly when severe or prolonged. Hypocapnia should be avoided except in specifi c clinical situations; when induced, hypercapnic acidosis should be for specifi c indications while defi nitive measures are undertaken.
A clearer understanding of the eff ects and mechanisms of action of CO 2 is central to determining its safety and therapeutic utility in the critically ill patient. In the coming years, research eff orts should focus on determining the potential mechanisms by which alterations in CO 2 tension contribute to the pathogenesis of acute organ injury states. Such insights should advance our understanding of the situations in which hypercapnia or hypocapnia may be helpful or dangerous, and should guide clinicians with regard to the rational use of CO 2 in the critically ill patient.