Rapid elimination of CO through the lungs: coming full circle 100 years on

At the start of the 20th century, CO poisoning was treated by administering a combination of CO2 and O2 (carbogen) to stimulate ventilation. This treatment was reported to be highly effective, even reversing the deep coma of severe CO poisoning before patients arrived at the hospital. The efficacy of carbogen in treating CO poisoning was initially attributed to the absorption of CO2; however, it was eventually realized that the increase in pulmonary ventilation was the predominant factor accelerating clearance of CO from the blood. The inhaled CO2 in the carbogen stimulated ventilation but prevented hypocapnia and the resulting reductions in cerebral blood flow. By then, however, carbogen treatment for CO poisoning had been abandoned in favour of hyperbaric O2. Now, a half-century later, there is accumulating evidence that hyperbaric O2 is not efficacious, most probably because of delays in initiating treatment. We now also know that increases in pulmonary ventilation with O2-enriched gas can clear CO from the blood as fast, or very nearly as fast, as hyperbaric O2. Compared with hyperbaric O2, the technology for accelerating pulmonary clearance of CO with hyperoxic gas is not only portable and inexpensive, but also may be far more effective because treatment can be initiated sooner. In addition, the technology can be distributed more widely, especially in developing countries where the prevalence of CO poisoning is highest. Finally, early pulmonary CO clearance does not delay or preclude any other treatment, including subsequent treatment with hyperbaric O2.


Background
At the turn of the 20th century, CO poisoning was treated by administering high concentrations of O 2 to increase the O 2 carried in the blood and, if necessary, ventilation was stimulated by adding CO 2 . It was initially and mistakenly thought that patients asphyxiated to unconsciousness by CO had a total body deficit of CO 2 that was replenished by the inhaled CO 2 (Henderson et al. 1921). Furthermore, animal tests had shown that the addition of CO 2 to O 2 markedly increased the dissociation of carboxyhaemoglobin (COHb) and accelerated clearance of CO compared with using O 2 alone (Henderson & Haggard, 1920). Carbon dioxide was administered in concentrations of 5-10% in O 2 , known as 'carbogen' . by a deficit of CO 2 was rejected (Donald & Paton, 1955). Ventilatory stimulation by CO 2 was no longer required, because hypoventilation accompanying coma could be managed by endotracheal intubation and mechanical ventilation. It became feasible to increase CO dissociation from haemoglobin (Hb) by exploiting the mass action effect of O 2 on the equilibrium (Haldane, 1895) COHb + O 2 ← → O 2 Hb + CO by administering the O 2 at hyperbaric pressures (Pace et al. 1950). Hyperbaric O 2 replaced carbogen as the preferred treatment (Smith, 1962) because it was thought (mistakenly, as subsequently demonstrated; Fisher et al. 1999) to result in faster CO elimination (Norman & Ledingham, 1967) and, on theoretical grounds, to be effective at reversing the assumed toxic effects of CO in such extravascular tissues as the brain (Brown & Piantadosi, 1990;Stoller, 2007).

Time to treatment over type of treatment
The point cannot be too strongly emphasized that for treatment to be effective it must be applied at the earliest possible moment after the victim is discovered, and must remove the carbon monoxide from his blood as soon as possible. (Henderson & Haggard, 1922) Although the physics and chemistry underpinning the effectiveness of hyperbaric O 2 in clearing CO from the blood are unassailable, and some beneficial effects can be demonstrated in animals (Brown & Piantadosi, 1990Piantadosi et al. 1997), in practice it has been difficult to demonstrate its clinical efficacy. The poor response of most victims of CO poisoning to hyperbaric O 2 has been confirmed repeatedly by expert panels in Australia, Canada and the USA Juurlink et al. 2005;McMaster University Division of Emergency Medicine, 2006;Wolf et al. 2008), as well as large controlled trials in Australia (Scheinkestel et al. 1999) and France (Annane et al. 2010). The primary lesson to be learned from the discrepancies between animal and clinical studies is that for patients poisoned by CO, the time to treatment, rather than the method of treatment, is of major importance (Gorman et al. 1992;Scheinkestel et al. 1999). Even from the very beginning of hyperbaric O 2 treatment of CO poisoning in Glasgow, it was clear that delays between poisoning and treatment markedly reduced its effectiveness (Smith, 1962). Times to treatment as short as 3-6 h, which are all that can be expected for hyperbaric O 2 given the logistics of patient transport and chamber preparation, continue to show no benefit compared with normobaric O 2 (Scheinkestel et al. 1999;Annane et al. 2010).

Effect of time to treatment on pathology of CO poisoning
It has been long understood that 'asphyxia is not immediately terminated when the victim is removed from the gassing chamber. . .although his body may be surrounded and his lungs filled with fresh air, his brain continues to be asphyxiated' (Henderson & Haggard, 1922). Eventually, there is a redistribution of CO from blood to extravascular tissues (Coburn, 1970), drawn there by the high affinity of some cellular molecules for CO [e.g. myoglobin in heart muscle (Coburn, 1970;Dolan, 1985) and cytochromes in the brain (Cronje et al. 2004)], even at low [COHb], and particularly with hypoxaemia (Dolan, 1985).
One instructive model of CO distribution kinetics to an extravascular compartment is CO in the fetus, as studied by Longo and colleagues  in pregnant sheep. Fetal Hb has a higher affinity for both O 2 and CO than maternal Hb. After an initial maternal exposure to CO, there is a delay in the transfer of CO to the fetus of about 1 h , which is characteristic of many tissues (Cronje et al. 2004). This delay is due to the low partial pressure of CO (P CO ) in the plasma, because it is tightly bound to Hb (Bruce et al. 2008). Eventually, at higher [COHb], P CO rises and CO begins to diffuse into the tissues. At equilibrium, fetal [COHb] will exceed maternal [COHb] (dotted lines in Fig. 1). If rescue occurs prior to equilibration of CO, maternal [COHb] will follow the time course illustrated in Fig. 1. If normobaric O 2 is administered, the maternal halftime of CO elimination will be ∼80 min (Dolan, 1985). However, because of the greater affinity of fetal Hb for CO, fetal [COHb] will continue to rise and so exceed that of the mother, even as her [COHb] is falling. If CO clearance from the mother is accelerated, the P CO gradient between the fetus and mother increases , thereby also increasing the rate of elimination from the fetus. A computer simulation of CO kinetics between mother and fetus using the model proposed by  is available as a supplemental file entitled CO Model.zip.
These principles of CO kinetics have long been acknowledged (Henderson & Haggard, 1922;Smith, 1962;Scheinkestel et al. 1999); yet somehow, by consensus, a treatment that was highly effective because it could be administered with the least delay (carbogen) was abandoned for another (hyperbaric O 2 ) despite its associated delay in treatment. The (presumed) greater rate Figure 2. Effect of poikilocapnic and isocapnic normobaric O 2 treatment on cerebral O 2 delivery (DO 2 ) in CO-exposed human subjects Fourteen human subjects were exposed on two separate occasions to CO until their [COHb] reached 10-12%. They were administered 100% O 2 with, and without, maintaining isocapnia. Blood O 2 delivery was calculated from arterial PO 2 , arterial P CO 2 , haemoglobin saturation (11-[COHb]), plasma O 2 content, and changes in middle cerebral artery flow velocity (MCAV) as measured by transcranial Doppler (as a surrogate of change in cerebral blood flow). Poikilocapnic hyperoxia resulted in a significantly lower DO 2 . (From Rucker et al. 2002, with permission from the publisher.) of CO elimination and the potential of reversing COrelated pathology (Sharp et al. 1962) with hyperbaric O 2 was considered an acceptable trade-off for the difficult logistics, increased expense and added delay in treatment. Despite little evidence of its value, hyperbaric O 2 has remained the mainstay of treatment for the last half a century.

Is normobaric oxygen a standard of care?
Even normobaric O 2 treatment of CO poisoning is problematic. The effect of P O 2 on the half-time of [COHb] reduction in patients treated in hospital (as opposed to laboratory volunteers) is highly unreliable (r 2 = 0.19), ranging from 26 to 148 min (Weaver et al. 2000). Furthermore, normobaric O 2 treatment may even contribute to the morbidity of CO poisoning. Apart from the potential for free radical generation by hyperoxia (Thom, 1990), there is also the underappreciated effect of hyperoxia as a ventilatory stimulant. Hyperoxia-induced hyperventilation results in some degree of hypocapnia (Becker et al. 1996), which is associated with a reduction of blood flow in such CO 2 -responsive vascular beds as the coronary (Case et al. 1975) and cerebral circulations. The reduction in cerebral (Kety & Schmidt, 1948) blood flow with hypocapnia occurs even in the presence of increased levels of CO in the blood (Rucker et al. 2002). In normoxic individuals, as well as those with high [COHb] (Henderson & Haggard, 1922), normobaric O 2 produces only a very small increase in blood O 2 content that is carried in the plasma, where it is poorly soluble. If this small increase in blood O 2 content is accompanied by even a small reduction in tissue blood flow, the result can be a net reduction in organ O 2 delivery (Case et al. 1975;Rucker et al. 2002). Figure 2 illustrates that the administration of normobaric O 2 , an undisputed treatment for CO poisoning since the time of Haldane (Haldane, 1895), may even exacerbate the brain ischaemia resulting from CO poisoning.

Back to the future
If there are problems with carbogen, hyperbaric and normobaric O 2 , where do we go from here?

Increased alveolar ventilation can be as effective as hyperbaric O 2
About a decade ago, the trade-offs between rate of CO elimination and time to treatment were re-examined. The initial studies compared the half-times of reduction of [COHb] induced by increases in alveolar ventilation with those resulting from hyperbaric O 2 . Previous studies (Henderson & Haggard, 1920) had concentrated on the relative efficacies of various mixtures of CO 2 in O 2 for reducing [COHb] in spontaneously breathing animals (Walton et al. 1925) and humans (Henderson & Haggard, 1922). In the early 1960s, it became apparent that the elimination of rebreathing during assisted ventilation (Douglas et al. 1961) and the magnitude of the minute ventilation (Killick & Marchant, 1959), i.e. the net alveolar ventilation, rather than the concentration of CO 2 in the carbogen, was the main factor determining the halftime of elimination. Indeed, with controlled ventilation Fisher et al. (1999) demonstrated, in dogs, that isocapnic increases in alveolar ventilation result in the same halftimes of CO elimination as those for hyperbaric O 2 (Fig. 3). Takeuchi et al. (2000) then investigated CO elimination half-times in spontaneously breathing human volunteers exposed to CO. Subjects breathed O 2 using a circuit that maintained normocapnia. Several findings from this study are of interest. First, the ventilatory response to normobaric O 2 (open symbols in Fig. 4) varied between subjects. Second, the relationship between elimination half-times and minute ventilation is a rectangular hyperbola. This shape means that initial graded increases in minute ventilation above resting values result in the

. Elimination half-times for [COHb]
Five anaesthetized, intubated, spontaneously breathing dogs were exposed to CO until [COHb] reached ∼70%. They were then administered, sequentially, room air (Air), normobaric O 2 (NBO 2 ) and then vigorously mechanically ventilated with O 2 while maintaining normocapnia (IHO 2 ). Blood was drawn every 5 min and analysed for [COHb]. Plots of log [COHb] versus time were used to calculate the half-times of reduction in [COHb]. Values are compared with dogs prepared in a similar manner and treated with normocapnic ventilation with O 2 at 3 atm (304 kPa). Isocapnic hyperpnoea resulted in a similar rate of [COHb] reduction to hyperbaric O 2 (HBO 2 ). Reprinted with permission of the American Thoracic Society. Copyright C American Thoracic Society. Hyperbaric data from the original study reported in the text was added to the figure by the authors. greatest reductions in half-times. For example, a 70 kg patient ventilating at about 15-20 l min -1 (levels easily tolerated by patients without severe lung disease) can reduce the half-time to a value similar to that reported for hyperbaric O 2 (Takeuchi et al. 2000). Finally, the relationship between minute ventilation and elimination half-time is scalable to body size and sex (Tesler, 2000).

Back to carbogen?
Is it therefore appropriate to resurrect carbogen as a readily deployable means to increase alveolar ventilation without reducing arterial P CO 2 ? Unfortunately, it is not. As early as 1955, an official report to the Medical Research Council (UK) (Donald & Paton, 1955) warned about the risk of exacerbating acidosis by administering carbogen to patients who are already retaining CO 2 due to ventilatory depression from severe CO poisoning or previously ingested drugs. As for those patients with an intact ventilatory response to CO 2 , administration of CO 2 up to a concentration of 4% increases the minute ventilation only by a factor of two (Soley et al. 1941), thereby limiting its effectiveness in CO elimination. Moreover, large individual variations in ventilatory responses to inhaled CO 2 (Solely et al. 1941;Prisman et al. 2007) mean that

. Half-time of COHb reduction versus minute ventilation in humans
Seven men were exposed to CO until [COHb] reached 10-12% on two separate occasions. On one occasion, subjects breathed 100% O 2 ('resting ventilation'). On the other occasion, subjects were administered 100% O 2 and asked to increase their minute ventilation; on that occasion, isocapnia was maintained. Venous blood was drawn every 5 min and analysed for [COHb]. Open symbols represent values during resting ventilation (normobaric O 2 ); filled symbols during normocapnic hyperpnoea. Half-times of elimination were calculated from plots of log [COHb] versus time. Most of the increase in [COHb] reduction was reached at a relatively modest 200 ml min -1 kg -1 , or 14 1 min -1 for a 70 kg person. (From Takeuchi et al. 2000; reprinted with permission of the American Thoracic Society. Copyright C American Thoracic Society.)

Figure 5. A self-inflating bag circuit suitable for spontaneous and controlled ventilation
Reserve gas enters circuit through the inspiratory relief valve of the self inflating bag (modified from Fig. 1 of Sasano et al. 2001; figure reproduced with permission of the publisher.) one cannot guarantee an increased rate of CO elimination, or even that hypocapnia will be prevented (Baddeley et al. 2000;Prisman et al. 2007). Above an inspired CO 2 concentration of 4%, minute ventilation markedly increases, but so does respiratory distress (Baddeley et al. 2000); these investigators found that 30% of patients and healthy subjects were unable to tolerate 5% CO 2 . It is therefore unlikely that a single premixed carbogen dose will fit all.

Hyperpnoea without carbogen
It follows from the preceding discussion that exploiting an increase in alveolar ventilation to clear the blood of CO will require a different approach. The method used must maintain normocapnia in order to allow patients to sustain increased ventilation comfortably for two to three half-times of CO elimination, thereby achieving more complete elimination of CO. Rather than administering a fixed concentration of CO 2 in an attempt to maintain normocapnia with hyperpnoea, one can administer CO 2 in direct proportion to increases in minute ventilation above basal levels (Sommer et al. 1998). Ideally, the apparatus that would be used to maintain normocapnia would be safe, easy to use, portable and, if at all possible, inexpensive.

Increasing alveolar ventilation while maintaining normocapnia
Historically, the advances in treatment of CO poisoning were also linked to the fabrication of devices required to implement them. Henderson and Haggard in New York devised their H-H Infusor to administer carbogen (Henderson & Haggard, 1922). Smith and Sharp (1960) built the first fixed and then portable hyperbaric chambers (Norman et al. 1970) in the Aberdeen Royal Infirmary, in Scotland. Recently, researchers in our laboratory (Sommer et al. 1998) described a method that passively maintains normocapnia regardless of minute ventilation and pattern of breathing. In that circuit, a constant O 2 flow is provided to a standard self-inflating bag, and the inspiratory relief valve of the self-inflating bag is attached to a demand regulator supplying 6% CO 2 in O 2 (Fig. 5). Any increase in minute ventilation above the O 2 flow is therefore supplied by the demand regulator (6% CO 2 in O 2 ). The O 2 flow is  Fig. 6, on end-tidal P CO 2 Lower panels show minute ventilation (V E ) and upper panels show continuous capnograph traces. Peaks are end-tidal P CO 2 and troughs represent inspired P CO 2 . Note proportional increases in inspired P CO 2 asV E increases; end-tidal P CO 2 remains unchanged, regardless ofV E . adjusted to match the patient's metabolic CO 2 production and controls the alveolar ventilation for CO 2 . Arterial P CO 2 is therefore unchanged by any increase in ventilation, because any ventilation exceeding the O 2 flow is composed of 6% CO 2 in O 2 , a mixture that does not contribute to a CO 2 diffusion gradient between capillary blood and the alveoli (Sommer et al. 1998;Somogyi et al. 2005 ; Fig. 6). However, it is the combined flow of O 2 and 6% CO 2 in O 2 that serves to wash out CO from the lungs, thereby clearing it from the blood.
The system is designed to be used in the field, but it cannot be readily improvised and requires deliberate preparation. It requires a customized breathing circuit or modification of available self-inflating bags, compressed CO 2 -containing gas with specific pressure regulator and flow controller. Such tanks require care in storage or use in extreme cold because CO 2 liquefies readily when cold. Use of the system requires some clinical expertise or monitoring of end-tidal gas in order to set the fresh gas flow (O 2 or air) appropriately to attain an appropriate end-tidal P CO 2 . However, due to the benign nature of acute hypercapnia in adults (Potkin & Swenson, 1992;Ayas et al. 1998), as well as in children (Goldstein et al. 1990), when oxygenation is maintained, the fresh gas flow need not be exact and can be safely titrated to comfort or ventilatory response, or can be set according to guidelines based on approximate body weight.

Isocapnic hyperpnoea in practice
We suggest that the availability of a portable device to increase CO clearance would be a useful adjunct to current treatment of CO poisoning. It can be brought to the field to begin treatment immediately at the time of rescue and continue treatment during transportation to hospital. The same device can be applied to patients breathing spontaneously, as well as those requiring ventilatory assistance. Prior CO clearance at the site of rescue would make emergency air transport safer, should it be required. As normocapnia is maintained and there are no foreseeable risks, this treatment can be administered on the suspicion of CO poisoning. It would therefore provide the earliest possible treatment if CO poisoning is later confirmed, and nothing is lost if it is not. Carbon monoxide poisoning often occurs in clusters, and this treatment approach can be inexpensively and safely applied to all victims. Finally, early pulmonary CO clearance does not delay or preclude any other treatment, including subsequent treatment with hyperbaric O 2 , if deemed necessary (Piantadosi, 2002;Weaver et al. 2002).

Summary
We believe we have now come full circle in the treatment of CO poisoning. At the beginning of the 20th century, carbogen proved to be an effective means of treating CO poisoning. Only relatively recently was it realized that it was not the CO 2 in carbogen but the increase in alveolar ventilation induced by the CO 2 that accelerated the clearance of CO. By then, however, rapid advances in the technology of positive-pressure ventilation and hyperbaric chambers overshadowed the old-fashioned approach using carbogen. Despite the initial enthusiasm for hyperbaric O 2 as the treatment for CO poisoning, the fact remains that hyperbaric O 2 facilities are expensive and their distribution around the world is poorly matched to the incidence and prevalence of CO poisoning. Even in wealthier urban areas, the inherent delays to initiate treatment make them clinically ineffective. The technical barriers to safely enable lung clearance of CO are low, making it feasible to provide for widespread availability of the means for early and rapid CO elimination. In any case, early pulmonary CO clearance does not delay or preclude any other treatment, including subsequent treatment with hyperbaric O 2 .