The relationship between the systemic delivery and consumption of oxygen in critical illness … the concept of delivery-dependence

In research into critical care, there has been considerable interest in the impact of reduced delivery of oxygen on its consumption in metabolising tissues.^{3, 4} It appears that, in intact organisms, modest reductions in the delivery of oxygen do not significantly impact on its consumption at a global, regional or cellular level. In health, this balance can generally be maintained because the extraction of oxygen from the circulating blood can be increased where necessary, thus allowing normal metabolism to continue. If the delivery of oxygen falls below a “critical” level, however, then this is accompanied by a concomitant fall in the consumption of oxygen, as the ability of the tissues to increase their extraction of oxygen is overwhelmed.⁵ It appears that this critical level of delivery occurs when the systemic extraction of oxygen reaches 50 to 60 percent.^{6, 7} At levels of delivery below this, anaerobic metabolism and lactic acidosis may develop (Fig. 1).

Figure 1. The graph shows the relationship between the systemic delivery of oxygen, with its systemic consumption and extraction as delivery is progressively reduced (solid lines). During modest reductions in delivery, consumption remains unchanged, as the systemic extraction of oxygen increases. Further reductions in delivery below a critical level (point 1), progressively reduce consumption, as the ability to increase the systemic extraction of oxygen is overcome. The critical level of delivery may be altered by changes in the overall consumption of oxygen (point 2), or in the ability of the tissues to regulate their microcirculation or their extraction of oxygen (point 3).

A number of mechanisms may enable metabolising tissues to maintain their consumption of oxygen during modest reductions in its delivery. Under normal circumstances, the flow of blood to individual organs is matched to prevailing needs, so that each receives its required fraction of the total systemic delivery of oxygen and other substrates. When the latter is compromised, it is redistributed so that those organs able to increase their local extraction of oxygen receive a progressively smaller fraction of the total cardiac output. Conversely, some organs in which the ability to increase the extraction of oxygen is less developed have the ability to “autoregulate” their flow, and thus obtain a greater fraction of cardiac output. It thus appears that an increase in the extraction of oxygen and autoregulation are both fundamental to the ability of the organism to maintain its consumption of oxygen as delivery declines to towards the critical level.^8–10 Interestingly, irrespective of which of these particular mechanisms predominates in an individual organ, all appear to become delivery-dependent at around the same levels of total systemic delivery of oxygen.¹¹ A number of mechanisms may increase this critical level, including impaired delivery of oxygen at the level of the microcirculation, in other words impaired autoregulation, altered mitochondrial function, representing impaired extraction, or increases in total levels of consumption due to either increased metabolic activity or uncoupling¹⁰ (Fig. 1).

The relationship between the systemic delivery of oxygen and its consumption early after cardiac surgery

The changes which occur in the systemic delivery and consumption of oxygen have been well-described in the adult early after cardiac surgery. Typically, the delivery of oxygen is reduced, largely reflecting a diminished cardiac output, while paradoxically, its consumption may be increased. The latter has been attributed to elevations in temperature and to a systemic inflammatory response, such that an association between elevations in the levels of proinflammatory mediators and increases in the systemic consumption of oxygen has been described.¹²

Li et al.^{13, 14} have used respiratory mass-spectrometry to examine whether similar factors modulate these relationships in children after cardiac surgery. In twenty children who were studied during the first 4 hours after cardiac surgery, they observed that the initial consumption of oxygen was unrelated either to the duration of cardiopulmonary bypass or aortic cross-clamping; or to the depth of hypothermia. During the subsequent 4 hours, changes in the consumption of oxygen closely followed changes in temperature. As a consequence, for every incremental increase in body temperature of one degree centigrade, the consumption of oxygen increased by 11 percent. During this period, the progressive increase in the consumption of oxygen was closely paralleled by changes in its delivery.¹³ It thus appears that as in adults, temperature is a powerful determinant of the systemic consumption of oxygen in children, early after cardiac surgery.

Li et al.¹⁴ subsequently examined the relationships between the systemic consumption and delivery of oxygen in a different group of infants, for the first 12 hours after cardiac surgery. In these patients, intra-operative management included modified ultrafiltration, a technique known to modulate the inflammatory response to cardiopulmonary bypass. In addition, their temperature was actively controlled during the period of observation. In this study, the systemic consumption and delivery of oxygen both gradually fell until they reached a nadir after eight hours. A loose relationship was demonstrated between the changes in the systemic consumption of oxygen and the concentrations of inflammatory mediators in arterial blood.¹⁴

The relationship between the systemic delivery of oxygen and its consumption as an indicator of outcome in children after cardiac surgery

Given the central role for the metabolism of oxygen in the preservation of tissue function, one might hypothesize that changes in the balance between the delivery and consumption of oxygen could provide us with predictors of outcome after cardiac surgery. To this end, in retrospective review of a group of infants undergoing cardiac surgery, a systemic extraction of oxygen of greater than 0.5 at 6 hours after surgery was associated with an increased risk of mortality.¹⁵ Interestingly however, in a group of children early after cardiac surgery, Duke et al.¹⁶ found no association between adverse outcome and cardiac output, oxygen content of arterial blood or of mixed venous blood. It was an elevated arterial lactate, either at the time of admission to the intensive care unit or four hours later, which appeared to be the best indicator of a subsequent adverse event.¹⁶

In a group of infants after the Norwood procedure, in whom the relationship between the systemic delivery and consumption of oxygen was inferred from the oxygen saturation of mixed venous blood, Hoffman et al.¹⁷ demonstrated that a saturation of less than 30 percent was a predicted anaerobic metabolism. These investigators also demonstrated that a mixed venous oxygen saturation of less than 40 percent was a predictor of adverse neurodevelopmental outcome in survivors of the Norwood operation; and that the severity of the abnormality was proportional to the length of time that the saturation remained at this level.¹⁸

The relationship between the systemic delivery of oxygen and its consumption in circulations with one functional ventricle

Patients with functionally univentricular circulations are particularly susceptible to the development of a syndrome of low cardiac output in the early postoperative period. Unfortunately, in these patients, in whom accurate assessment of systemic perfusion would be particularly useful while on the intensive care unit, an additional level of complexity is presented by the parallel pulmonary and systemic circulations.¹⁹

A unique feature of infants with a functionally univentricular circulation after construction of a superior cavopulmonary shunt is the dependence of flow of blood to the lungs and systemic oxygenation on cerebral haemodynamics. An increase in flow of blood to the brain in patients with a cavopulmonary shunt would be expected to increase the flow into the lungs. An elevation of arterial carbon dioxide, a cerebrovascular vasodilator, might therefore augment the pulmonary flow in these patients, and produce beneficial effects on systemic oxygenation. Carbon dioxide, however, is also a pulmonary vasoconstrictor. Important studies^{20, 21} recently sought to determine the relative importance of these two, potentially opposing influences, in patients early after the cavopulmonary shunt. In these studies, ventilation was adjusted from normocapnia to achieve a mild respiratory acidosis, with an arterial carbon dioxide level of 50 millimetres of mercury. This manoeuvre increased the flow of blood to the lungs, and the systemic delivery of oxygen, and also appeared to have an additional beneficial action of reducing the consumption of oxygen.

Changes in the systemic delivery and consumption of oxygen during infusions of catecholamine

In the critically-ill patient in whom the relationship between the systemic delivery and the consumption of oxygen may be deranged, an important goal of therapy must be to re-set a balance between the two. In adults, catecholamines such as dobutamine and dopamine augment the systemic delivery of oxygen by way of direct increases in cardiac output.^{22, 23} In some species, catecholamines can produce additional augmentation through splenic contraction, which results in an increase in haemoglobin and therefore the content of oxygen in arterial blood.²⁴ Catecholamines, nonetheless, also stimulate the consumption of oxygen through their effects on myocardial work, requirements for energy for the breakdown and cycling of stored fuels such as triglyceride and glycogen, and elevations of the metabolic rate. In healthy adults, this augmentation in the consumption of oxygen is far-exceeded by a concomitant increase in delivery, so that as a result, the systemic extraction of oxygen is reduced.^{22, 23}In the neonatal period, however, catecholamines have additional thermogenic actions through their effects on brown adipose tissue.²⁵ We have previously observed exaggerated increases in the systemic consumption of oxygen during infusions of both dobutamine and dopamine in neonatal lambs, and this was particularly evident at high rates of infusion.²⁶ Further studies will be required to examine whether this phenomenon exists in sick neonates receiving inotropes early after cardiac surgery. If so, the extent to which this negates any beneficial effects on systemic delivery of oxygen.

The role of therapies specifically reducing the systemic consumption of oxygen

While the primary goal of inotropic therapy has been to increase the systemic delivery of oxygen to meet the metabolic demands of the tissues, an alternative approach to the treatment of the sick infant after cardiac surgery could be to reduce the systemic consumption of oxygen. Until now, there have been few studies which have examined the effects of treatments which have this as a primary aim. In a laboratory study, we demonstrated that infusions of insulin-like growth factor 1 appeared to reduce the systemic consumption of oxygen after cardiopulmonary bypass.²⁷ Further studies will be required to examine whether this type of intervention will be of therapeutic benefit in the sick infant after cardiac surgery.

Measuring the systemic consumption and delivery of oxygen

The systemic consumption of oxygen equals the quantity of oxygen taken up by the lungs over a set time, and its measurement requires the careful collection and analysis of expired gas. One might ask whether we need to measure this at all, as “assumed” or “calculated” values are used frequently. There is considerable variability in the consumption of oxygen in health and disease, and we, and others, have established that assumed values, or those derived from nomograms are of no real utility in our population of infants and children with congenital heart disease.^{28, 29}

From first principles, measurement of the consumption of oxygen requires the analysis of inspired gas, and the collection and analysis of expired gas. The Douglas bag method applies the hypothesis that the consumption of oxygen is equivalent to the difference between inspired and expired oxygen over a given period of time. This method is now rarely used in the clinical setting as it is time-consuming, and relies on long periods of cardiorespiratory stability, and importantly assumes a constant respiratory quotient of 1.

An alternative approach to measuring the consumption of oxygen would be to measure the concentration of oxygen in inspired and expired gases in tandem with the Minute Ventilation, using the method of indirect calorimetry. Because of the relative ease with which indirect calorimetry can be adapted for ventilated patients, there is increasing interest in the use of for both clinical and research purposes.

We have used two indirect calorimetric methods for estimating the systemic consumption of oxygen in ventilated patients. The first, which employs the Deltatrac^® Metabolic monitor (Datex Division Instrumentarium, Helsinki)²⁷ can be used in spontaneously breathing or mechanically ventilated children. The concentration of oxygen in inspired and expiredgas is measured using paramagnetic sensors. Minute ventilation is measured using an air-dilution method, in which ambient air is mixed at a constant and known rate with the expired gas. The amount of gas expired per minute can be derived from the difference in the concentration of carbon dioxide in the expired gas and the resulting gas mixture (Fig. 2).

Figure 2. Measurement of the systemic consumption of oxygen using the Deltatrac® metabolic monitor. The concentrations of oxygen in the inspired and expired gases are measured with paramagnetic sensors. The minute volume is derived by measuring the concentration of carbon dioxide in the expired gas, before and after dilution with ambient air, at a known rate of flow.

The gold-standard for indirect calorimetric measurement of oxygen consumption is respiratory mass spectrometry.^{28, 30} Mass spectrometry involves the separation of components of a respiratory gas mixture according to their atomic masses, and the subsequent quantification of each component. The consumption of oxygen can be accurately measured using the mixed expirate method. This requires analysis of inspired gas, the collection of expired gas, the addition of a tracer inert gas to expired gas at a known rate, and the subsequent continuous analysis of the mixed expirate.

While our own as yet unpublished observations suggest excellent agreement between the metabolic monitor and mass spectrometry, the latter has some advantages, including the capacity to make valid estimates of consumption at higher levels of oxygen supplementation, which is of particular relevance in patients early after surgery. Irrespective of which method is employed, it is important that fluctuations in the concentration of oxygen in the inspired gas, obstructions to the flow of gas through the inlets of the instrument, and leaks of gas are meticulously eliminated. As a result, a cuffed endotracheal tube is recommended for all patients.

An alternative approach to the calculation of the systemic consumption of oxygen is through the so-called reverse Fick method. This technique involves the direct measurement of cardiac output as the primary variable, typically achieved by thermodilution, dye dilution, through echocardiographic methods, or through the analysis of the contour of the arterial pulse (PICCO^®, Pulsion Medical Systems, Germany). The consumption of oxygen is then derived from the cardiac output and the content of oxygen in arterial and mixed venous samples of blood. The advantages and limitations of each of these techniques is beyond the scope of this article, but have been considered elsewhere.³¹

The problem of mathematical coupling and shared variables

It is common practice to investigate the relationship between the consumption and delivery of oxygen by deriving measures of each, and using statistical methods to investigate the strength of the association between them. As we have already discussed, while it is possible to measure both the delivery and consumption of oxygen as primary variables, in practice, either delivery is measured and consumption derived, or the other way round (Fig. 3). The presence of the so-called shared variables may result in an apparent statistical relationship, where in fact no physiological relationship exists. Thus, a close relationship between the systemic delivery of oxygen and its consumption was demonstrated in critically-ill patients, when delivery was calculated as the primary variable, from which consumption was derived. In the same patients, it was observed that the relationship was poorer when consumption was measured independently using indirect calorimetry.³² This is one aspect of the phenomenon of mathematical coupling.³³

Figure 3. Derivation of the systemic delivery and consumption of oxygen in intensive care. In the left-hand panel the cardiac output is measured as the primary variable and the systemic delivery and consumption of oxygen are derived by relating the changes in cardiac output to those in the concentrations of oxygen in the arterial and venous blood (indirect Fick principle). In this method, the cardiac output is a ‘shared variable’. In the right-hand panel, the systemic consumption of oxygen is measured as the primary variable and cardiac output and the systemic oxygen delivery are derived using the Fick principle. In this method, the systemic consumption of oxygen is a ‘shared variable’.

As an example in Figure 4, I obtained two independent variables x and y by interrogating a random number generator for 20 numbers between 1 and 100. Not surprisingly comparison of the relationship between x and y demonstrated a poor association between the two, with a correlation coefficient of 0.008, and a P value of 0.3. When a secondary variable was derived by multiplying x and y, and the relationship between this variable and x was examined, a statistically-significant relationship was revealed, with a correlation coefficient of 0.41 and a P value of 0.002. Thus, the introduction of a shared variable, in this case x, led to the suggestion of an apparent relationship in two variables, originally derived from two series of random numbers (Fig. 4).

Figure 4. Graph of twenty pairs of random numbers (x and y) between 1 and 100 (left-hand panel). When a third variable is derived from the same random numbers (the product of x and y), the derived variable is positively correlated with x, because of the phenomenon of mathematical coupling, related to the presence of the shared variable x.

Conclusions

The examination of the relationship between the systemic delivery of oxygen and its consumption provides a useful framework around which the care of the critically-ill may be developed. Until now, only a few groups have used this framework to examine the infant after cardiac surgery. Greater insights into optimising the care of this group of patients in the future will come from further studies of these relationships.

Acknowledgement

Figure 3 was modified and used with the permission of Jia Li, from her doctoral thesis, as submitted successfully to the University of London. I thank her for permitting me to use the figure.