Abstract
This review provides an overview of the relationship between ventilation/perfusion ratios and gas exchange in the lung, emphasising basic concepts and relating them to clinical scenarios. For each gas exchanging unit, the alveolar and effluent blood partial pressures of oxygen and carbon dioxide (PO2 and PCO2) are determined by the ratio of alveolar ventilation to blood flow (V′A/Q′) for each unit. Shunt and low V′A/Q′ regions are two examples of V′A/Q′ mismatch and are the most frequent causes of hypoxaemia. Diffusion limitation, hypoventilation and low inspired PO2 cause hypoxaemia, even in the absence of V′A/Q′ mismatch. In contrast to other causes, hypoxaemia due to shunt responds poorly to supplemental oxygen. Gas exchanging units with little or no blood flow (high V′A/Q′ regions) result in alveolar dead space and increased wasted ventilation, i.e. less efficient carbon dioxide removal. Because of the respiratory drive to maintain a normal arterial PCO2, the most frequent result of wasted ventilation is increased minute ventilation and work of breathing, not hypercapnia. Calculations of alveolar–arterial oxygen tension difference, venous admixture and wasted ventilation provide quantitative estimates of the effect of V′A/Q′ mismatch on gas exchange. The types of V′A/Q′ mismatch causing impaired gas exchange vary characteristically with different lung diseases.
Abstract
A review of ventilation–perfusion relationships and gas exchange, basic concepts and their relation to clinical cases http://ow.ly/wMUwq
Introduction
While the healthy lung efficiently exchanges respiratory gases, hypoxaemia and hypercapnia indicate pathophysiology and failure of the lung to provide adequate gas exchange. An understanding of how gases are transferred and the causes of inefficient gas exchange are central to caring for patients with lung diseases. In this article, we will review the normal and pathological mechanisms of gas exchange. Starting with simple and building to more complex, we will emphasise basic concepts and relate them to case scenarios familiar to clinicians.
Lung structure
The structure of the lung is well suited for efficient exchange of respiratory gases. Through the airway and vascular trees, fresh gases and venous blood are delivered to and removed from a large alveolar capillary surface area. In an adult, inhaled air enters the trachea with a cross-sectional area of ∼3 cm2 and is delivered to the alveoli with a surface area of ∼140 m2, roughly the size of a tennis court [1]. Similarly, the pulmonary vascular tree begins as the main pulmonary artery and repeatedly bifurcates into arterioles and capillaries that cover 85–95% of the alveolar surface [1]. An exceptionally thin membrane of only 1 μm [2, 3] separates the alveolar gas and blood compartments, allowing gases to diffuse rapidly between them. Due to the relatively large blood volume within the alveolar capillaries, blood flow slows and the transit time for blood increases, normally to 0.25–0.75 s, allowing more time for gas exchange. The remarkable design that allows this gas exchanging system to be constructed within the thoracic cavity has been highlighted by comparing this engineering feat to that of folding a letter so that it fits into a thimble [3].
A single lung unit
We start with a simple lung model consisting of just one gas exchanging unit (fig. 1a). The capillaries of the unit deliver mixed venous blood with a low partial pressure of O2 (Pv¯O2). The partial pressure of O2 (PO2) in the alveolar gas (PAO2) is much higher than in the capillary blood and O2 diffuses passively from the alveolar space into the blood during passage through the capillaries (fig. 2a). The membrane separating the alveolar gas and blood compartments causes little resistance to diffusion so the PO2 in end-capillary blood (PecO2) equilibrates with PAO2 well before the blood leaves the unit. Oxygenation of arterial blood is, therefore, primarily dependent on the PAO2. Note that in this idealised lung unit there is no difference between PAO2 and PO2 in arterial blood (PaO2). For reasons explained later, the alveolar–arterial O2 tension difference (PA–aO2) is very helpful when assessing the causes of gas exchange problems in clinical medicine and we will use it throughout this review. While the term “gradient” is also commonly used, we prefer to use “difference” to highlight that the difference between the PAO2 and PaO2 is not due to a pressure gradient between them. The partial pressure of CO2 (PCO2) is greater in mixed venous blood (Pv¯CO2) than in the alveolar gas (PACO2) and diffusion over the alveolar–capillary membrane, therefore, results in a net flow in a direction opposite to that of O2, from blood to alveolar gas (fig. 2b). The result is again an equal PCO2 in alveolar gas and end-capillary blood (PecCO2) because the resistance for diffusion is even less for CO2 than for O2. Due to differences in the relationship between partial pressures and blood content for O2 and CO2, there is approximately as much CO2 exchanged for a partial pressure difference between mixed venous and arterial blood of 5 mmHg (0.7 kPa) as there is O2 exchanged with a difference of 60 mmHg (6.7 kPa). The amount of O2 carried in the blood is determined by the haemoglobin concentration, the fraction of haemoglobin binding O2 and the PaO2 (fig. 1b). When we discuss multiple lung unit models below, it is important to understand that the end-capillary blood O2 content (CecO2), rather than the PecO2, from different units are additive.
Five causes of arterial hypoxaemia
A single-unit model (fig. 1a) is used to begin our discussion of five potential causes of hypoxaemia as well as hypercapnia. Our first example is one of diffusion limitation.
Diffusion limitation
In case 1, a 25-year-old male elite cyclist who undergoes a cardiopulmonary exercise study is noted to have progressively worsening arterial hypoxaemia with increasing workload (a response not seen in normal individuals). Arterial blood gases (ABGs) at end-exercise are: pH 7.18, PCO2 in arterial blood (PaCO2) 30 mmHg (4.0 kPa), PaO2 81 mmHg (10.8 kPa) and arterial haemoglobin O2 saturation 88%.
Exercise increases the amount of O2 extracted from arterial blood in the systemic circulation, which tends to reduce Pv¯O2. Therefore, more O2 needs to be taken up in the lung to reach normal oxygenation of arterial blood. Exercise also increases pulmonary blood flow, which shortens the time that the blood is exposed to alveolar gas. The combined effect is that more O2 needs to be taken up in less time (fig. 3). At very high cardiac outputs, the transit time might be too short for complete equilibration between PAO2 and PecO2. This represents diffusion limitation as a cause of hypoxaemia in athletes achieving extremely high cardiac outputs [4–6]. Hypoxaemia due to diffusion limitation can also be seen in normal individuals during exercise at altitude. In this setting, the driving pressure for O2 diffusion is decreased due to the lower PO2 in inspired air (PIO2) at altitude and the transit time of blood through the alveolar capillaries is shorter due to higher cardiac outputs. A diffusion limitation can also occur in patients with interstitial lung diseases. Patients with these diseases might have a normal PaO2 at rest but develop hypoxaemia during exercise, which can be explained by the combined effect of increased resistance to diffusion through the thickened alveolar–capillary membrane, reduced Pv¯O2 and a shortened transit time [7].
Diffusion limitation is one of five causes of hypoxaemia (table 2). The lack of equilibration between PAO2 and PecO2 creates an increased PA–aO2. Hypoxaemia in these patients, as a rule, responds nicely to supplemental O2. While this does not correct the diffusion limitation, it raises the PAO2 and the driving pressure for diffusion of O2 into the blood. Due to the low resistance to diffusion of CO2, PaCO2 is normal in patients with O2 diffusion limitations. Although diffusion limitation is our first cause listed for hypoxaemia in table 2, from a clinical perspective, it is an uncommon reason for a low PaO2.
Hypoventilation
In case 2, an 82-year-old female with chronic renal failure on continuous ambulatory peritoneal dialysis (CAPD) had recently contracted whooping cough from her grandchildren. Her physician prescribed an opioid-containing antitussive to reduce the cough. She is now drowsy and unable to manage the CAPD routines. An ABG analysis reveals pH 7.17, PaO2 45 mmHg (6.0 kPa) and PaCO2 77 mmHg (10.3 kPa).
This case highlights hypoventilation as a cause of hypoxaemia. Although the quantitative relationship is complex [8, 9], it can be intuitively understood that PAO2 in our single alveolar unit (fig. 1a) is influenced by the ratio between O2 delivery (ventilation, V′A) and O2 removal (blood flow, Q′). Hence, one of the determinants of PAO2 is the balance or ratio between ventilation and blood flow (V′A/Q′). In the patient from case 2, hypoventilation and normal blood flow results in a low V′A/Q′ ratio (less delivery and unchanged removal of O2), which lowers PAO2 and consequently PecO2. Conversely, hyperventilation (increased ventilation relative to blood flow) increases PAO2 and PecO2. PACO2 is also determined by the V′A/Q′ ratio but CO2 is delivered through blood flow and CO2 is removed through ventilation, the converse of O2. Hence, a decreased V′A/Q′ ratio results in higher PACO2 and PaCO2, the hallmark of hypoventilation, while an increased ratio lowers PACO2. These relationships are illustrated in figure 4. The balance between O2 delivery and removal results in a PAO2 that can be quantified using the alveolar gas equation (box 1).
Hypoxaemia secondary to hypoventilation is, thus, due to a low PAO2 and is the second cause of a low PaO2 listed in table 2. Because there is no impairment of gas exchange across the alveolar–capillary membrane, PAO2 and PaO2 are similarly decreased and the PA–aO2 is normal in hypoxaemia due to hypoventilation alone. In case 2, the alveolar gas equation (box 1) results in an estimated PAO2 of 53 mmHg (7.1 kPa) and PA–aO2 ∼7 mmHg (0.9 kPa), a normal value. Hypoxaemia due to hypoventilation can be resolved with supplemental O2, increasing PAO2 even if hypoventilation remains uncorrected. One important aspect of O2 treatment in this situation is that hypoventilation, while breathing air, will result in severe hypoxaemia before PaCO2 increases to dangerously high levels. In contrast, providing supplemental O2 can result in PAO2 and PaO2 being maintained above 90 mmHg (12 kPa) even with hypoventilation causing severe hypercapnia. A clinically important corollary is that a normal pulse oximetry reading in a patient breathing air is a good indication of adequate ventilation (normal PaCO2). However, when a patient is on supplemental O2, a normal O2 saturation cannot be used to judge the adequacy of ventilation.
Low inspiratory oxygen pressure
In case 3, a 21-year-old very fit, mountaineer notices, while ascending Mont Blanc (4810 m), that her fingers are blue and that her haemoglobin O2 saturation is 65%.
Although the fraction of O2 in air is always 0.21, decreasing barometric pressure at high altitude causes a proportional decrease in the PIO2. Using the alveolar gas equation (box 1), an appropriate barometric pressure for her altitude (420 mmHg (56 kPa)), no change in ventilation (PaCO2 40 mmHg (5.3 kPa)) and an R of 0.8 results in an expected PAO2 at the top of Mont Blanc of ∼28 mmHg (3.7 kPa) and severe hypoxaemia. Low PIO2, at high altitude or when breathing an inspiratory O2 fraction (FIO2) <0.21, is therefore one additional cause of hypoxaemia. Again, gas exchange at the alveolar–capillary membrane is normal. The cause of hypoxaemia is reduced PAO2 and consequently the PA–aO2 is normal. How is it then possible to ascend Mont Blanc? Hypoxaemia increases the respiratory drive, thereby increasing the total V′A/Q′ ratio and O2 delivery to the alveoli, which partially corrects PAO2 and PaO2 (fig. 3). Increased ventilation results in hypocapnia, defined as PaCO2 <40 mmHg (<5.3 kPa). Recent ABG samples from climbers breathing air at the summit of Mount Everest revealed a mean PaO2 of 25 mmHg (3.3 kPa) and PaCO2 of 13 mmHg (1.7 kPa) [11]. A second method to raise PAO2 and improve PaO2 is to breathe supplemental O2. Most mountaineers reaching the summit of Mount Everest use supplemental O2.
Two-unit lung models
For the two remaining causes of hypoxaemia, both examples of V′A/Q′ mismatch, we must shift to multi-unit lung models. While fascinating in its design and gas exchange capacity, the anatomical structure of the lung with multiple parallel gas exchanging units sets the stage for non-uniform and different distributions of regional ventilation and blood flow. If regional ventilation and perfusion are not perfectly matched, i.e. each unit is not receiving equal proportions of total blood flow and total ventilation, V′A/Q′ ratios will vary between lung regions, each of them will have different PAO2 and PACO2 and gas exchange will be less efficient. Figure 5 shows the range of PAO2 and PACO2 seen in units as a function of their V′A/Q′ ratio. To demonstrate how gas exchange becomes less efficient with units of differing V′/Q′ ratios, we will begin with a simple two-unit lung model.
Low V′A/Q′ units
In case 4, a 67-year-old female with severe emphysema has an ABG of pH 7.35, PaO2 55 mmHg (7.3 kPa) and PaCO2 55 mmHg (7.3 kPa) when breathing air. With supplemental O2 at 3 L·min−1, haemoglobin O2 saturation improves from 87% to 93% on pulse oximetry.
Figure 6 illustrates uniform distribution of blood flow but uneven distribution of ventilation to two units, such that one unit has a V′A/Q′ ratio of 2.0 while the other unit has a ratio of 0.1. The low V′A/Q′ ratio results in low PAO2 in this unit and, thus, lower O2 content and PecO2 for blood leaving this unit. Note that the effect is analogous to that in the patient who was hypoventilating in case 2, but in this case it relates to only a part of the lung. The nearly complete oxygenation of haemoglobin at a normal V′A/Q′ ratio means that the relative over-ventilation (high V′A/Q′ ratio) of the other unit results in only a slightly increased O2 content for blood leaving this unit. Also remember that the O2 content of arterial blood (CaO2) is determined by the flow-weighted mean of the O2 contents, not partial pressures, for blood from different units. The overall effect of low V′A/Q′ units is, therefore, a reduced PaO2 and an increased PA–aO2. Low V′A/Q′ units are, thus, the fourth cause of hypoxaemia (table 2). The effect of low V′A/Q′ units on PaCO2 is less compared with the effect on PaO2, because CO2 elimination is increased in the unit with a high V′A/Q′ ratio. In patients with low V′A/Q′ regions, hypercapnia is also prevented by the ventilatory response to increased PaCO2. Because low V′A/Q′ regions are in fact ventilated, the detrimental effect on arterial oxygenation can be counteracted by increasing the FIO2, which increases PAO2, even in low V′A/Q′ regions. Low V′A/Q′ units, therefore, cannot be the sole cause of hypoxaemia in patients that fail to respond to increased FIO2.
Shunt
In case 5, a 32-year-old male with severe acute respiratory distress syndrome (ARDS) secondary to bacterial pneumonia has an ABG on mechanical ventilation with an FIO2 of 0.80 that reveals pH 7.28, PaO2 67 mmHg (8.9 kPa) and PaCO2 61 mmHg (8.1 kPa).
Figure 7 shows a situation in which one unit is without any ventilation, hence it cannot participate in gas exchange. Blood flow through this unit, therefore, constitutes an intrapulmonary shunt. Shunt can be viewed as one extreme of the range of V′A/Q′ ratios with a ratio of 0. End-capillary blood flow from this unit has the same PecO2 and CecO2 as mixed venous blood and when combined with blood from the other unit decreases oxygenation of arterial blood. Shunt is, therefore, the fifth and last cause of hypoxaemia (table 2). The CaO2 is again the flow-weighted average of O2 content from the gas exchanging unit (CecO2) and the shunted blood (Cec′O2) (fig. 7). The amount of shunt is quantified as the fraction of cardiac output (Q′s/Q′t) distributed to nonventilated units (box 2). The effect on arterial oxygenation depends on this fraction and the O2 content of mixed venous blood (Cv¯O2). For a shunt of a certain magnitude, lower Cv¯O2 will result in worse arterial oxygenation. Interventions aimed at increasing mixed venous O2 saturation (Sv¯O2) (and consequently Cv¯O2), such as raising the CaO2 through increased haemoglobin concentration, decreasing O2 consumption or increasing cardiac output, might therefore improve hypoxaemia caused by a large shunt. The shunt fraction has, however, been shown to increase with increasing cardiac output [13, 14]; such interventions might therefore be less effective than expected. Increased Pv¯O2 and increased pulmonary arterial pressure, which may decrease and counteract hypoxic vasoconstriction (discussed later), are speculative explanations for this effect of cardiac output on shunt.
In contrast to the situation with low V′A/Q′ regions, with shunt increasing FIO2 is much less effective at improving arterial oxygenation. With increasing shunt fraction, the effect of raising FIO2 is progressively less and for very large shunts, even an FIO2 of 1.0 has little effect on PaO2. The reasons for the small effect are that raised FIO2 fails to improve PAO2 in unventilated units and that the little extra O2 that can be added to blood flow through ventilated units is not enough to offset the impact of the shunted blood. The relationships between the variables shunt size, FIO2 and PaO2 are illustrated by the iso-shunt diagram (fig. 8). Examples of clinical conditions with hypoxaemia due to large intrapulmonary shunts are extensive atelectasis, severe pneumonia and ARDS. In these conditions, the shunt might exceed 50% of total lung blood flow [13, 16–19]. An intracardiac right-to-left shunt has the same effect on PaO2 as an intrapulmonary shunt. Even in normal subjects, PaO2 is less than PAO2 due primarily to a right-to-left shunt of 2–3% of cardiac output. Most of this is not a strict intrapulmonary shunt but represents venous blood added to arterial blood from bronchial veins and the Thebesian veins of the left ventricle [12]. The effect of shunt on CO2 exchange is discussed in box 3.
Dead space
In a continuation of case 4, the 67-year-old female with severe emphysema is hospitalised with an acute exacerbation of chronic obstructive pulmonary disease (COPD). ABG reveals pH 7.21, PaO2 67 mmHg (8.9 kPa) and PaCO2 85 mmHg (11.3 kPa). Measurements reveal she is breathing 12 L·min−1.
Despite a minute ventilation (V′E) that is about twice what is normal at rest, this patient presents with an elevated PaCO2, indicating hypoventilation. How can this be explained? With each inspiration, in all subjects, part of the tidal volume (VT) is retained in the conducting airways and, therefore, fails to contribute to gas exchange. This volume corresponds to the anatomical dead space, and is ∼150 mL (2–3 mL·kg−1) or one third of a normal VT [12, 21]. A normal dead space fraction of the tidal volume (VD/VT) is therefore ∼0.3. Figure 9 illustrates two units that are equally ventilated but one of the units is without blood flow and, therefore, constitutes alveolar dead space. One clinical cause might be pulmonary embolism. Alveolar dead space refers to gas exchanging units that are ventilated but without blood flow and the V′A/Q′ ratio is infinite. Apparatus dead space refers to the volume of any face mask, tubing, etc. that functions as an extension of the anatomical dead space. Total dead space (VD), therefore, consists of summed anatomical dead space, alveolar dead space and any apparatus dead space.
Minute ventilation refers to the total volume of expired gas per minute, while effective alveolar ventilation refers to the volume of gas per minute that participates in gas exchange. In contrast, dead space ventilation, or wasted ventilation, is the fraction of total ventilation that does not contribute to gas exchange. This fraction of wasted ventilation is often expressed as the ratio VD/VT. Effective alveolar ventilation can be calculated as V′A=RR×(VT–VD), where RR is respiratory rate, or as V′A=V′E×(1–VD/VT). Note that effective alveolar ventilation is determined by both minute ventilation and VD/VT. Increased VD inherently means that a greater minute ventilation is needed to maintain effective alveolar ventilation adequate for CO2 removal. An increase in VD/VT without a compensatory increase in minute ventilation results in decreased PaO2 and increased PaCO2. In the patient in case 4, the high minute ventilation is not enough to compensate for the increased VD/VT, which results in hypercapnia. Increased PaCO2 might also be seen in those sedated and mechanically ventilated patients who do not increase their minute ventilation in the setting of a new pulmonary embolism.
High V′A/Q′ units
High V′A/Q′ units are units that are overventilated in proportion to blood flow, for example, a unit that receives a ventilation of 1 L·min−1 but a blood flow of only 0.1 L·min−1 resulting in a V′A/Q′ ratio of 10. Although this unit does not represent alveolar dead space, ventilation in excess of blood flow causes a “dead space effect”, because less CO2 will be removed by ventilation of a high V′A/Q′ unit than by ventilation of a better perfused region. Thus, similar to alveolar dead space, addition of high V′A/Q′ units means that a greater minute ventilation is needed to maintain adequate gas exchange. High V′A/Q′ regions are primarily due to areas of normal ventilation with low blood flow. The impact of these regions on ABGs is small, because the contribution of any region to ABG composition is flow-weighted.
More than two-unit lung models
So far we have focused on models with one or two lung units. In reality, gas exchange within the entire lung and the resulting PaO2 and PaCO2 are determined by the effluent blood flow from thousands of units with different V′A/Q′ ratios (fig. 10). Riley and Cournand [9] suggested that the effect of V′A/Q′ mismatch on gas exchange can be more simply thought of and quantified as a lung consisting of only three compartments with different V′A/Q′ ratios: one compartment with ideal V′A/Q′ matching (V′A/Q′=1.0), one totally without ventilation (V′A/Q′=0, shunt) and one without blood flow (V′A/Q′=∞, dead space).
The model is practical because it is relatively easy to obtain the measurements needed to quantify shunt blood flow and dead space ventilation (discussed below). It is important to understand that the calculations assume that the effects of V′A/Q′ mismatch on PaO2 and PaCO2 are entirely due to shunt and dead space ventilation, and that all gas exchange occurs in units with ideal V′A/Q′ matching. The model does not reflect the true situation as it ignores gas exchange in units with other V′A/Q′ ratios.
Venous admixture
With the three-compartment model, shunt blood flow is calculated using the shunt equation (box 2). The calculated shunt, alternatively and more accurately described as venous admixture, corresponds to the amount of right-to-left shunt of mixed venous blood that would result in the observed arterial oxygenation in the absence of low V′A/Q′ regions. Venous admixture might, therefore, be increased even in the absence of true shunt.
Using values obtained while breathing 100% O2 can differentiate hypoxaemia due to low V′A/Q′ regions from the effect of a true shunt, because the venous admixture from low V′A/Q′ regions is abolished in this situation [12]. One problem with this approach is that breathing 100% O2 tends to increase the amount of shunt due to absorption atelectasis so the degree of shunt while breathing a lower FIO2 might be overestimated. Shunt can also be measured using inert gases with low solubility in blood [17, 22]. It should be noted that this methodology also differs from the calculation of venous admixture as it excludes the shunt effect of venous blood added to arterial blood after the passage through the lung, from Thebesian and bronchial veins.
Wasted ventilation (physiological dead space)
According to the three-compartment model of Riley and Cournand [9], physiological dead space or wasted ventilation corresponds to the total dead space that would cause the observed impairment of CO2 elimination (box 3). We prefer the term “wasted ventilation”, since it better describes the meaning of this quantity. In normal lungs, the wasted ventilation closely corresponds to the anatomical dead space. In addition to anatomical dead space, wasted ventilation also results from any apparatus, alveolar dead space and high V′A/Q′ units. Blood from shunt regions will have a PecCO2 equal to the mixed venous value, which raises PaCO2 and thereby the calculated wasted ventilation; this has been coined shunt dead space [21]. This is an effect of using PaCO2 as a surrogate for PACO2; hence, while not an effect of increased dead space or high V′A/Q′ regions, it does represent impaired CO2 elimination [23]. As discussed above, effective alveolar ventilation equals minute ventilation minus wasted ventilation. A minute ventilation of 7.5 L·min−1 and a normal VD/VT of 0.3 results in an effective alveolar ventilation of 5.25 L·min−1. Chronic lung disease might increase VD/VT to ≥0.8; in this case, a minute ventilation of ≥25 L·min−1 is required for an effective alveolar ventilation of 5 L·min−1. Measures of wasted ventilation, thus, estimate the effect of V′A/Q′ mismatch on the minute ventilation needed to maintain adequate gas exchange. Although not measured, for the patient in case 4, the increased PaCO2 despite a high minute ventilation is most likely explained by a greater component of wasted ventilation.
Because the control of breathing aims to maintain a normal PaCO2, an increase in wasted ventilation requires an increased minute ventilation to maintain unchanged effective alveolar ventilation. The first consequence of increased wasted ventilation in most situations is, therefore, not an increase in PaCO2 but a change in minute ventilation and, therefore, the work of breathing. Chronic hypercapnia can be viewed as an adaptive response as it means that CO2 removal can be maintained with a lower minute ventilation and, thus, lessens the work of breathing.
The arterial to end-tidal PCO2 difference
The physiological dead space equation (box 3) uses the difference between PaCO2 and the PCO2 of mixed expired gas (PĒCO2) to calculate the total fraction of wasted ventilation. By contrast, the difference between PaCO2 and the PCO2 of mixed expired alveolar gas excludes the effect of anatomical and apparatus dead space and, therefore, estimates the extent of high V′A/Q′ and alveolar dead space regions. Because the last part of the exhaled VT consists of alveolar gas (box 4), the end-tidal PCO2 (PETCO2) can be used as a surrogate measure of mixed PACO2. The PaCO2–PETCO2 difference, therefore, correlates with dead space and wasted ventilation [28–30]. It is noteworthy that, remembering the physiological dead space equation, modelling studies have shown that (PaCO2–PETCO2)/PaCO2 has a more linear correlation with alveolar dead space fraction than the simpler PaCO2–PETCO2 [29, 30]. Although dependent on, for example, age, the inhaled/exhaled gas volume and mode of ventilation, a PaCO2–PETCO2 <5 mmHg (0.7 kPa) is considered normal [31, 32] and is supported by PETCO2 being close to 36 mmHg (4.8 kPa) in a small number of normal subjects [33]. PaCO2–PETCO2 has been shown to correlate with alveolar dead space fraction and alveolar recruitment in animal models of ARDS [34]. PETCO2 is illustrated in figure 11 and further discussed in box 4.
The combined effect of low and high V′A/Q′ regions
In comparison with optimal V′A/Q′ matching, redistribution of either regional ventilation or blood flow must result in both low and high V′A/Q′ regions. For example, a shift in blood flow from one unit to another will cause one unit to be less perfused in proportion to ventilation (high V′A/Q′) and the other unit to be more perfused in proportion to ventilation (low V′A/Q′). Compared with the effect on PaCO2, the effect of V′A/Q′ mismatch on PaO2 is often more obvious because 1) the numerical changes in PaO2 are larger than those in PaCO2 and 2) impaired CO2 elimination is often counteracted by increased minute ventilation which offsets the effect of V′A/Q′ mismatch on PaCO2 to a much greater degree than the effect on PaO2. The net effect of V′A/Q′ mismatch is always reduced gas exchange efficiency and for a given FIO2 and minute ventilation results in lower PaO2 and higher PaCO2 than would have occurred if V′A/Q′ matching were optimal.
V′A/Q′ mismatch in the normal lung
Regional ventilation and blood flow are not uniform in the normal lung. Heterogeneities of ventilation and perfusion have been attributed to the influence of airway and vascular tree geometries as well as regional differences in blood flow and lung compliance due to gravity [35–39]. The traditional zone model [38] of perfusion and ventilation predicts that both perfusion and ventilation increase in the gravitational direction, from apex to lung base in the upright posture. However, they do not increase at the same rate, and V′/Q′ tends to decrease from apex to base. Animal studies using higher spatial resolution measurements have suggested a fractal model [40, 41], in which asymmetries in the vascular tree result in perfusion heterogeneities that cannot be explained by the hydrostatic gradient due to gravity. Regional ventilation and blood flow are further influenced by posture, lung volume and application of positive airway pressures [36, 37, 42], to name just a few factors. Despite these non-uniformities, efficient gas exchange is possible through the close matching of ventilation and blood flow, attributed to the shared influences of structure and gravity [35]. The correlation is, however, not perfect, which causes V′A/Q′ ratios and, therefore, PAO2 and PACO2, to vary between different lung regions [35].
In young healthy individuals, most ventilation and blood flow are distributed to units with V′A/Q′ ratios between 0.3 and 2.0 [22], which tends to result in a range of PAO2 from about 90 mmHg to 130 mmHg (fig. 5) [43]. With increasing age, ventilation and blood flow are distributed to units with a wider range of V′A/Q′ ratios, with blood flow to low V′A/Q′ regions being attributed to reduced ventilation of dependent lung regions due to airway closure [12, 44]. Consequently, PA–aO2 increases with age [12, 44].
Assessing arterial oxygenation
Several indices are used to evaluate PaO2 in relation to FIO2. When breathing air, the PA–aO2 provides a sensitive measure of gas exchange efficiency but is dependent on age. A number of different studies have sampled relatively small numbers of individuals in different populations to determine the distribution of PA–aO2 in normal individuals [45–48]. In all studies, the mean PA–aO2 increases with age, as do the confidence intervals around the mean. table 3 incorporates data from two of the larger populations [45, 46]. A simplified formula for the normal PA–aO2 when breathing air is (age in years/4)+4 mmHg (divide result by 7.5 for kPa). Because the normal PA–aO2 is primarily due to venous admixture, increasing FIO2 raises PAO2 more than PaO2 and the normal value for PA–aO2 increases with increasing FIO2. The PaO2/FIO2 ratio is more frequently used in clinical practice and in the definition of ARDS. A lower limit for normal PaO2 when breathing air of 80 mmHg (10.7 kPa) corresponds to a normal PaO2/FIO2 ratio being >400 mmHg (53 kPa), while diagnosing ARDS requires a PaO2/FIO2 <300 mmHg (40 kPa). The PaO2/FIO2 is also dependent on the FIO2, level of positive end-expiratory pressure (PEEP) and the arterial venous O2 extraction [49, 50] and is, therefore, a difficult measure to interpret. The PaO2/FIO2 is not helpful in differentiating between different causes of hypoxaemia.
V′A/Q′ mismatch and vascular tone
Hypoxic pulmonary vasoconstriction
Hypoxic pulmonary vasoconstriction (HPV) causes pre-capillary vasoconstriction in units with low PAO2 (<60 mmHg (8.0 kPa)), such as those with low V′A/Q′ or shunt [51–54]. The effect of HPV is to divert blood flow away from hypoxic units toward better ventilated units, improving V′A/Q′ matching and arterial oxygenation (fig. 12). For example in humans, HPV reduces blood flow to atelectatic regions by ∼50% [55, 56]; the beneficial effect on arterial oxygenation can be deduced from moving from one iso-shunt line to another at a fixed FIO2 in the iso-shunt diagram (fig. 8). The efficacy of HPV in correcting V′A/Q′ mismatch decreases with the extent of the hypoxic lung region [53, 57]. One example is low PAO2 within the entire lung (e.g. due to severe hypoventilation or high altitude). In this case, generalised vasoconstriction raises pulmonary arterial pressure without improving V′A/Q′ matching. Animal and human experiments indicate that HPV is of little importance for V′A/Q′ matching in normal lungs [53, 58, 59]. By contrast, HPV can be of critical importance in patients with disease causing regional alveolar hypoxia [53], improving PaO2 by up to 20 mmHg (2.7 kPa) in patients with COPD or ARDS [54]. An illustration of the importance of HPV is that systemically administered vasodilators, inhibiting HPV, might reduce PaO2 in such patients [60].
Supplemental oxygen and carbon dioxide retention in COPD patients
One further illustration of the clinical importance of HPV is its role in O2-induced hypercapnia in patients with COPD. Although often attributed to a decreased drive to breathe, it has been shown that administration of supplemental O2 to these patients can increase PaCO2 despite constant minute ventilation [61, 62]. The major explanation is that supplemental O2 raises the PAO2 in low V′A/Q′ units, inhibiting regional HPV and increasing blood flow to these units. Consequently, blood is diverted away from better ventilated regions, converting them to high V′A/Q′ units, which increases wasted ventilation [61, 62]. This is an illustration of how increased wasted ventilation decreases effective alveolar ventilation (discussed above), despite unchanged minute ventilation. In addition, because the low V′A/Q′ units have a higher PACO2, the restored blood flow to these units results in more CO2 being delivered to the arterial blood. Increased V′A/Q′ mismatch as the dominant mechanism causing O2-induced hypercapnia remains, however, controversial: other work has shown that decreased minute ventilation can contribute to increased PaCO2 [63]. One further explanation is the Haldane effect, in which increased PaO2 decreases binding of both H+ and CO2 to haemoglobin, thereby increasing the amount of physically dissolved CO2 and, thus, PCO2 [12, 64]. In other words, for any given CO2 content, PCO2 will be higher with greater haemoglobin O2 saturation. The Haldane effect has been estimated to explain 6–78% of the increase in PaCO2 in COPD patients after administration of supplemental O2 [61, 65, 66].
Inhaled vasodilators
Inhaled vasodilators can be used to improve V′A/Q′ mismatch. The concept is that inhaled vasodilators are distributed to each region in proportion to regional ventilation and well ventilated regions are, therefore, more vasodilated. Locally reduced vascular tone causes a shift in blood flow away from unventilated regions toward the better ventilated regions, improving V′A/Q′ matching and gas exchange. Inhaled nitric oxide has been shown to reduce shunt [60, 67] and improve arterial oxygenation in patients with ARDS [60, 68, 69]. A similar effect has been demonstrated with inhaled prostacyclin [70, 71]. Interestingly, the effect is less consistent in patients with COPD. In this case, inhaled nitric oxide can result in both improved and worsened gas exchange; the latter can be explained by the vasodilator also reaching low V′A/Q′ regions. Inhibition of HPV increases blood flow to these low V′A/Q′ regions, which causes further arterial desaturation [67, 72]. The beneficial effect of inhaled nitric oxide might, therefore, be less consistent in hypoxaemia caused by low V′A/Q′ regions than in conditions characterised by shunt regions.
V′A/Q′ mismatch in lung diseases
COPD
Patients with COPD might present with hypoxaemia, hypercapnia or both. Hypoxaemia is primarily explained by perfusion of underventilated (low V′A/Q′) regions, which has been attributed to airway disease [73–76]. Shunt is of little importance in the absence of complicating conditions such as pneumonia or atelectasis [7, 74] and hypoxaemia is, therefore, expected to respond to increased FIO2. Hypercapnia is explained by decreased effective alveolar ventilation due to increased wasted ventilation secondary to high V′A/Q′ regions and alveolar dead space caused by emphysematous destruction of the lung parenchyma [73–75]. Hypercapnia becomes evident only when the patient is unable to maintain the added work of breathing required for adequate alveolar ventilation. Changes in cardiac output, O2 consumption and minute ventilation have been shown to modulate the effect of increased V′A/Q′ mismatch in patients with acute COPD exacerbations [74].
ARDS
Gas exchange in patients with ARDS, often caused by pneumonia or sepsis, is characterised by severe hypoxaemia due to loss of aeration of large lung regions. Several studies have shown that arterial hypoxaemia in these patients is mostly caused by shunt blood flow that might exceed 50% of cardiac output [13, 16–19]. Low V′A/Q′ regions are either absent or a lesser cause of low PaO2 [16, 17]. Impaired diffusion does not seem to contribute to arterial hypoxaemia [16]. The effect of PEEP on V′A/Q′ mismatch in these patients is especially interesting. PEEP has been shown to reduce the amount of nonaerated lung tissue through alveolar recruitment and to improve arterial oxygenation through a decrease in shunt blood flow [13, 16, 17, 19, 77, 78]. Increased PEEP has also been shown to increase total dead space, which is interpreted as an increased number of unperfused alveoli secondary to increased airway pressure with compression of alveolar capillaries in nondependent lung regions [16, 77–79]. The effect of PEEP on dead space is attenuated if the negative effect of PEEP on cardiac output is counteracted by fluid loading to increase cardiac preload, which also illustrates that reduced cardiac output might increase VD/VT [79]. By contrast, PEEP in combination with alveolar recruitment has been shown to decrease wasted ventilation [80]. When using PEEP to optimise gas exchange in these patients it is, thus, necessary to balance the potential beneficial and detrimental effects on oxygenation, wasted ventilation and cardiac output. Another tool to improve gas exchange in ARDS patients is treatment in the prone posture. Animal experiments and clinical studies have demonstrated that the prone posture decreases shunt and V′A/Q′ mismatch, improving arterial oxygenation [81–83]. A recent clinical trial demonstrated that the use of the prone posture early in the course of ARDS was associated with improved survival [84]. The primary mechanism postulated to increase survival is not improved gas exchange, but a more uniform distribution of regional ventilation resulting in less ventilator-induced lung injury [83, 85].
Pulmonary embolism
Pulmonary embolism results in lung regions with no or reduced blood flow (high V′A/Q′ regions) and, thus, increased wasted ventilation. Less obviously, redistribution of blood flow to other lung regions causes these regions to become relatively overperfused, and transformed into low V′A/Q′ regions, creating hypoxaemia and an increased PA–aO2. The reason for hypoxaemia is, thus, increased V′A/Q′ mismatch but not increased shunt [86, 87]. The effect on ABGs is modified by the respiratory response to increased PaCO2, the end result often being a dyspnoeic patient with hypoxaemia and normal or even low PaCO2. Hypoxaemia might be aggravated by circulatory failure causing low Cv¯O2 or by intracardiac right-to-left shunting, both provoked by acute pulmonary hypertension. Importantly, the absence of hypoxaemia or increased PA–aO2 does not rule out pulmonary embolism [88, 89]. Similarly, using the PaCO2–PETCO2 to gauge wasted ventilation is informative, but in isolation is not definitive for diagnosing pulmonary embolism [90]. While a PETCO2 measurement indicating increased wasted ventilation increases the likelihood of pulmonary embolism [91] and a normal amount of wasted ventilation diminishes the likelihood [33, 92], the utility of these measures is very dependent on the a priori clinical probability of pulmonary embolism in the individual patient [90] and the fraction of the pulmonary vascular tree that is involved [91].
Summary
1) There are five cause of hypoxaemia. Decreased PaO2 can be caused by hypoventilation, low PIO2, diffusion limitation, low V′A/Q′ regions or shunt. In contrast with other causes, hypoxaemia due to shunt characteristically responds poorly to increased FIO2. Low V′A/Q′ regions and shunt are by far the most common causes of clinically encountered hypoxaemia.
2) Hypoventilation and low V′A/Q′ regions also impair CO2 removal, but the magnitude of the effect on PaCO2 is less and modified by the ventilatory response to hypercapnia.
3) High V′A/Q′ regions and alveolar and apparatus dead space cause increased wasted ventilation and, thus, impaired CO2 elimination. The primary response to increased wasted ventilation is, in most situations, increased minute ventilation and work of breathing, not increased PaCO2.
4) The effect of V′A/Q′ mismatch on gas exchange efficiency can be quantified using calculations of PA–aO2, venous admixture and wasted ventilation.
5) Low and high V′A/Q′ regions cause hypoxaemia, impaired CO2 elimination and increased work of breathing in COPD patients.
6) Shunt is the most important cause of hypoxaemia in patients with ARDS and pneumonia.
Footnotes
Previous articles in this series: No 1: Naeije R, Vachiery J-L, Yerly P, et al. The transpulmonary pressure gradient for the diagnosis of pulmonary vascular diseases. Eur Respir J 2013; 41: 217–223. No. 2: Hughes JMB, van der Lee I. The TL,NO/TL,CO ratio in pulmonary function test interpretation. Eur Respir J 2013; 41: 453–461. No. 3: Vonk-Noordegraaf A, Westerhof N. Describing right ventricular function. Eur Respir J 2013; 41: 1419–1423. No. 4: Hamzaoui O, Monnet X, Teboul J-L. Pulsus paradoxus. Eur Respir J 2013; 42: 1696–1705. No. 5: Prisk GK. Microgravity and the respiratory system. Eur Respir J 2014; 43: 1459–1471. No. 6: Dempsey JA, Smith CA. Pathophysiology of human ventilatory control. Eur Respir J 2014; 44: 495–512.
Conflict of interest: None declared
- Received February 24, 2014.
- Accepted May 4, 2014.
- ©ERS 2014