Permissive hypercapnia (acceptance of raised concentrations of carbon dioxide in mechanically ventilated patients) may be associated with increased survival as a result of less ventilator-associated lung injury.

Accumulating clinical and basic scientific evidence points to an active role for carbon dioxide in organ injury, in which raised concentrations of carbon dioxide are protective, and low concentrations are injurious.

Although hypercapnic acidosis may indicate tissue dysoxia and predict adverse outcome, it is not necessarily harmful per se. In fact, it may be beneficial. There is increasing evidence that respiratory (and metabolic) acidosis can exert protective effects on tissue injury, and furthermore, that hypocapnia may be deleterious.

If hypoventilation is allowed in an effort to limit lung stretch, carbon dioxide tension increases. Such “permissive hypercapnia” may be associated with increased survival in acute respiratory distress syndrome (ARDS);2 this association is supported by outcome data from a 10-year study.3

Furthermore, hypocapnia shifts the oxyhaemoglobin dissociation curve leftwards, restricting oxygen off-loading at the tissue level; local oxygen delivery may be further impaired by hypocapnia-induced vasoconstriction.

Brain homogenates develop far fewer free radicals and less lipid peroxidation when pH is lowered by carbon dioxide than when it is lowered by hydrochloric acid.19

Finally, greater inhibition of tissue lactate production occurs when lowered pH is due to carbon dioxide than when it is due to hydrochloric acid.20

An association between hypoventilation, hypercapnia, and improved outcome has been established in human beings.2, 5, 21 In lambs, ischaemic myocardium recovers better in the presence of hypercapnic acidosis than metabolic acidosis.22 Hypercapnic acidosis has also been shown to protect ferret hearts against ischaemia,23 rat brain against ischaemic stroke,16 and rabbit lung against ischaemia-reperfusion injury.24 Hypercapnia attenuates oxygen-induced retinal vascularisation,25 and improves retinal cellular oxygenation in rats.26 “pH-stat” management of blood gases during cardiopulmonary bypass, involving administration of large amounts of additional carbon dioxide for maintenance of temperature-corrected PaCO2, results in better neurological and cardiac outcome.27

Hypercapnia results in a complex interaction between altered cardiac output, hypoxic pulmonary vasoconstriction, and intrapulmonary shunt, with a net increase in PaO2 (figure).28 Because hypercapnia increases cardiac output, oxygen delivery is increased throughout the body.28 Regional, including mesenteric, blood flow is also increased,29 thereby increasing oxygen delivery to organs. Because hypercapnia (and acidosis) shifts the haemoglobin-oxygen dissociation curve rightwards, and may increase packed-cell volume,30 oxygen delivery to tissues is further increased. Acidosis may reduce cellular respiration and oxygen consumption,31 which may further benefit an imbalance between supply and demand, in addition to greater oxygen delivery. One hypothesis32 is that acidosis protects against continued production of further organic acids (by a negative feedback loop) in tissues, providing a mechanism of cellular metabolic shutdown at times of nutrient shortage—eg, ischaemia.

Acidosis attenuates the following inflammatory processes (figure): leucocyte superoxide formation,33 neuronal apoptosis,34phospholipase A2 activity,35 expression of cell adhesion molecules,36 and neutrophil Na+/H+ exchange.37 In addition, xanthine oxidase (which has a key role in reperfusion injury) is inhibited by hypercapnic acidosis.24 Furthermore, hypercapnia upregulates pulmonary nitric oxide38 and neuronal cyclic nucleotide production,39 both of which are protective in organ injury. Oxygen-derived free radicals are central to the pathogenesis of many types of acute lung injury, and in tissue homogenates, hypercapnia attenuates production of free radicals and decreases lipid peroxidation.19 Thus, during inflammatory responses, hypercapnia or acidosis may tilt the balance towards cell salvage at the tissue level.

However, we know from several case series that human beings, and animals, can tolerate exceptionally high concentrations of carbon dioxide, and when adequately ventilated, can recover rapidly and completely. Therefore, high concentrations (if tolerated) may not necessarily cause harm.

From the published studies reviewed, and from the pathological mechanisms assessed, we postulate that changes in carbon dioxide concentration might affect acute inflammation,33—36 tissue ischaemia,16 ischaemia-reperfusion,20, 24 and other metabolic,12, 21,32 or developmental14 processes.

We argue that the recent shift in thinking about hypercapnia must now be extended to therapeutic use of carbon dioxide. Our understanding of the biology of disorders in which hypocapnia is a cardinal element would require fundamental reappraisal if hypocapnia is shown to be independently harmful.

In summary, in critically ill patients, future therapeutic goals involving PaCO2 might be expressed as:“keep the PaCO2 high; if necessary, make it high; and above all, prevent it from being low”.