Mechanisms of hypoxemia

Introduction

A fundamental aspect of cardiopulmonary homeostasis is the adequate delivery of oxygen to meet the metabolic demands of the body. Cardiac output, O₂-carrying capacity (i.e., hemoglobin concentration and quality), and arterial PO₂ (PaO₂) determine O₂ transport. Relevant to this discussion, arterial hypoxemia commonly occurs in patients with acute respiratory failure (ARF). If arterial hypoxemia is severe enough, it is not compatible with life. The two primary causes of ARF are acute lung injury (ALI)/acute respiratory distress syndrome (ARDS) and chronic obstructive lung disease (COPD). Although the treatment for arterial hypoxemia always includes increases in the fractional inspired O₂ concentration (FIO₂), the degree to which patients’ PaO₂ improves and the need for adjuvant therapies differ markedly between these two groups of disease processes. The mechanisms by which arterial hypoxemia occurs in ALI/ARDS and COPD have been characterized using the multiple inert gas elimination technique (MIGET) approach [1]. MIGET provides precise estimates of the distributions of alveolar ventilation and pulmonary perfusion (V_A/Q) and their relationships, there is no need to change the FIO₂ during measurements, hence avoiding variations in the pulmonary vascular tone, and it facilitates the unraveling of the active interplay between intrapulmonary, namely V_A/Q imbalance, intrapulmonary shunt and limitation of alveolar to end-capillary O₂ diffusion, and extrapulmonary (i.e., FIO₂, total ventilation, cardiac output and oxygen consumption) factors governing hypoxemia [2].

The cardinal gas exchange features under which the lung operates that uniquely determine the PO₂ and PCO₂ in each gas exchange unit of the lung are the V_A/Q ratio, the composition of the inspired gas, and the mixed venous blood gas composition [3]. Each of these three factors may play key role influencing oxygenation. For example, the major mechanism of arterial hypoxemia in ALI/ARDS is intrapulmonary shunt (zero V_A/Q ratios) induced by the presence of collapsed or flooded alveolar units, whereas in COPD the primary mechanism of hypoxemia is V_A/Q mismatching.

Effect of breathing oxygen on oxygenation

In ALI/ARDS, as FIO₂ increases, PaO₂ increases as long as the amount of shunt is limited. The greater the degree of shunt, the less PaO₂ increases. In contrast, in COPD, in which the prime mechanism of hypoxemia is V_A/Q mismatching, the response to high FIO₂ levels is broadly similar irrespective of disease severity. With moderate V_A/Q imbalance PaO₂ increases almost linearly as FIO₂ is increased. In severely acute COPD the degree of very low V_A/Q ratios resembles shunt; the increase in PaO₂ in response to increasing FIO₂ is only slightly limited, becoming less responsive to increases FIO₂.

Importantly, FIO₂ can also alter V_A/Q balance through two additional mechanisms: hypoxic pulmonary vasoconstriction (HPV) and reabsorption atelectasis (RA). One of the main means by which the normal lung adjusts to low regional V_A is to induce vasoconstriction of the associated pulmonary vasculature to redirect perfusion away from nonventilated or under ventilated alveolar units. Thus HPV minimizes V_A/Q inequality, limiting the decrease in PaO₂ that would have occurred if such redistribution of blood flow had not occurred. One of the best V_A/Q indicators of the presence of HPV, as measured by MIGET, is the behavior of the area with normal and low V_A/Q ratios, reflected in the dispersion of pulmonary blood flow. In sequential measures one sees a significant increase in the latter V_A/Q descriptor while breathing 100% O₂. By contrast, shunt and the dispersion of alveolar ventilation that incorporates areas with normal and high V_A/Q ratios remain unchanged during HPV release.

Breathing 100% O₂ (FIO₂=1.0) can induce intrapulmonary shunt because lung units with low inspired V_A/Q ratios, termed “critical” V_A/Q ratios, can result in absent expired ventilation because all the inflated gas is absorbed. This results in alveolar denitrogenation, allowing complete gas resorption with atelectasis (RA) to develop spontaneously [4]. These critical V_A/Q units are dependent on the FIO₂, increasing both their potential area of collapse and rate of collapse considerably as FIO₂ approaches 1.0. Alternatively, these critical units may remain open despite increasing FIO₂ levels if functional residual capacity and tidal volume are increased, owing to alveolar interdependence. This is the rationale for using positive end-expiratory pressure (PEEP) and larger tidal volumes in patients with ALI/ARDS to prevent RA.

Both RA and HPV and can be observed, respectively, in the responses of patients with ALI/ARDS and COPD needing mechanical support who are given an FIO₂ of 1.0 [5] (Fig. 1). Intrapulmonary shunt increases moderately then remains stable for at least 30 min in ALI/ARDS patients given an FIO₂ of 1.0. In contrast, in COPD patients the dispersion of pulmonary blood flow, one of the most common V_A/Q indicators in COPD, further increases to an FIO₂ of 1.0 while the modest levels of intrapulmonary shunt remain unchanged, a response that strongly suggests HPV release. Both responses to pure O₂ breathing are accompanied with increases in PaO₂, which are much more prominent in patients with COPD.

Fig. 1

Index of oxygenation (PaO₂/FIO₂), intrapulmonary shunt (expressed as percentage of cardiac output), and dispersion of pulmonary blood flow (log SDQ, dimensionless) while breathing 100% O₂. In ALI/ARDS (open circles) both PaO₂/FIO₂ and log SDQ remain essentially unchanged while shunt increases significantly, indicating RA; note that after reinstatement of maintenance FIO₂ shunt still remains increased. In COPD exacerbation (closed squares) PaO₂/FIO₂ and log SDQ substantially increase while the very modest shunt unvaried, indicating HPV release (by permission from [5])

The increase in intrapulmonary shunt in ALI/ARDS is likely due to RA. If cardiac output increases as part of the sympathetic response to arterial hypoxemia, one may also see a parallel increase in mixed venous PO₂ owing to increased O₂ delivery. This can offset the increased shunt fraction minimizing the decrease in PaO₂. The deleterious effects of RA on pulmonary gas exchange may be enhanced by the mechanical trigger imposed on peripheral airways by ventilator support. Indeed, the repeated opening and closing of distal airways and/or the overexpansion of closed alveolar units with abnormally high shear stresses may result in more inflammatory lung changes, aggravating the initial mechanical stress injury. On the other hand, the changes observed in COPD during hyperoxia suggest that inhibition of HPV is the primary process. Interestingly, gas exchange abnormalities in both entities take place in the absence of measurable changes in pulmonary hemodynamics, suggesting that regional blood flow redistribution can have relevant effects on gas exchange despite minimal changes in pulmonary arterial pressure and blood flow.

If V_A were to decrease or dead space to increase, arterial PCO₂ (PaCO₂) would increase. Hyperoxia-induced increases in PaCO₂ in response to FIO₂ 1.0 breathing are more notable in ALI/ARDS than in COPD and can be attributed almost completely to the parallel increases in dead space, with a marginal role of the Haldane effect (i.e., decreasing PaO₂ increases PaCO₂ off-loading from hemoglobin). Conceivably, the increased dead space indicates redistribution of pulmonary blood flow from high V_A/Q areas either to regions with no ventilated (shunt) alveolar units in ALI/ARDS or to those poorly ventilated with low V_A/Q areas in COPD. This process, however, has not been definitely characterized. An alternative and/or complementary mechanism in ALI/ARDS for the observed increase in PaCO₂ could be overexpansion of remaining normal lung zones provoked by RA, but in COPD by bronchodilation secondary to the hypercapnia.

Protective ventilator support

Protective ventilator support with low tidal volume and high PEEP levels has become the preferred approach to decrease the impact of ventilator-associated lung injury [6]. This protective support causes a substantial improvement in gas exchange by increasing PaO₂ and decreasing intrapulmonary shunt. However, this strategy of increased PEEP and small tidal volumes is often accompanied by hypercapnia. Hypercapnia induces both vasodilatation and increased cardiac output, both of which increase intrapulmonary shunt and potentially impair arterial oxygenation. The principal factor to explain the observed reduction in shunt in protective lung ventilation is the recruitment of previously collapsed alveoli, as shown by the close correlation between the decreased intrapulmonary shunt and the amount of PEEP-induced lung volume recruitment. Furthermore, the parallel increase in cardiac output caused by the hypercapnia-induced vasodilatation does not induce any proportional injurious increases in intrapulmonary shunt. Conceivably, the alveolar recruitment induced by recruitment efficiently redistributes pulmonary blood flow to regions with alveolar units with normal V_A/Q balance. A parallel finding in protective lung ventilation is the significant increase in physiological dead space, possibly related to the combined effects of a decreased alveolar ventilation and increased functional residual capacity. Thus the application of a protective ventilator support combining low tidal volumes and high PEEP levels represent a beneficial ventilator strategy in ALI/ARDS both in terms of minimizing lung stress and augmenting gas exchange.

Summary

The primary mechanisms leading to arterial hypoxemia in ARF secondary to COPD exacerbations and ALI/ARDS are V_A/Q imbalance and intrapulmonary shunt while the conditions that uniquely determine the PO₂ and PCO₂ in gas exchange units of the lung are the V_A/Q ratio and the composition of inspired gas and mixed venous blood. This is why the extrapulmonary factors governing hypoxemia, i.e., FIO₂, total ventilation, cardiac output, and O₂ consumption, always need to be considered. The increase in intrapulmonary shunt characteristically shown in ALI/ARDS patients breathing high FIO₂ levels is secondary to the development of RA, whereas in COPD patients it is usually due to withdrawal of HPV, reflected by further V_A/Q worsening only without parallel increases in intrapulmonary shunt.