So this talk will focus on gas exchange physiology and its relevance in the peds ICU. I'm Ravi Kamani from Children's Hospital Los Angeles and here are my disclosures, none of which are really relevant for this talk. So here are the board principles that are important for gas exchange. Understand the mechanisms that match ventilation and perfusion. Understand the concept and calculation of pulmonary shunt and the venous admixture method. Calculation of dead space. Understand how PO2 influences carbon dioxide transport and how PCO2 influences oxygen transport by the blood. Know the factors that determine alveolar PO2. Understand the concept and estimation of ventilation perfusion mismatch and understand the mechanisms that increase the alveolar arterial PO2 difference. So as we think about gas exchange one of the most important physiologic principles is the idea of ventilation perfusion matching and that there are two real extreme ends of this spectrum that is shunt and dead space. As we think about interpulmonary shunt right those are areas that are receiving perfusion without any ventilation and the opposite side is alveolar dead space. Those are areas that are receiving ventilation without any perfusion and there are regions all in between regions where ventilation perfusion ratios are less than one and those live more on the side of interpulmonary shunt and where ventilation perfusion ratios are greater than one and those rely more on the area of dead space. As we think about the clinical symptoms with shunt or VQ regions less than one we think about hypoxemia and poor respiratory system compliance as the major clinical manifestation and for dead space it's hypercarbia or a large difference between the CO2 that we may measure at the end of the airway like an untitled CO2 and that which we would measure directly in the blood with an arterial CO2. And in fact ARDS or acute respiratory distress syndrome is really the classic disease for ventilation perfusion mismatch that we see in the pediatric ICU with both ends of the spectrum both shunt and dead space clearly seen in pediatric ARDS both of which have an independent relationship with mortality in pediatric ARDS. So our specific objectives with this talk are to review the physiology of shunt and dead space and VQ matching in general to characterize methods that we can use to estimate both shunt and dead space at the bedside and their limitations and then how we may use these measurements for prognosis to gauge response to therapy or even potentially to determine optimal treatment strategies. So one of the first things we need to think about as we're trying to understand VQ matching is to think about the VQ relationships on the alveolar PO2 and PCO2, i.e. what are the concentrations of each in the alveolus. And when we have a circumstance of near complete shunt, okay, on this example so these are regions with VQ ratios less than one over on this side, VQ ratios greater than one on this side, this is complete shunt, this is complete dead space over here. So in circumstances of complete shunt remember there is perfusion but there is no ventilation so no fresh gas is coming into the alveoli so that air in the alveoli resembles the content of mixed venous blood, i.e. has a low PAO2 and a relatively high PCO2. In contrast, the area of dead space where there is ventilation but no perfusion then the alveolar content really reflects what you would see in inspired air, a PAO2 of let's say 150 without any CO2. Now under normal conditions of health, remember that ventilation and perfusion are actually both highest at the basis, i.e. they're both gravity dependent. Perfusion certainly is gravity dependent as we think about pulmonary blood flow but ventilation is also gravity dependent. As we look at the relationship here in the pressure-volume curve, the dependent portion of the lung under normal health is actually what sits more, you know, in this zone of more optimal compliance but that does not necessarily mean that the ventilation-perfusion ratio is optimized at the base of the lung. As you sort of see here as I go from the top to the bottom of the lung, blood flow is increasing and ventilation is also increasing but the ratio between those two, i.e. the matching of ventilation and perfusion, is actually happening more towards the mid portion of the lung, let's say, or around the third rib. And in situations of disease or in situations like let's say sedation, for example, now there's a big switch in the awake versus the anesthetized patient in terms of the compliance relationship between the dependent and the non-dependent portion of the lung that now more of the lung sits, especially in this dependent region, falls below this lower inflection point. So that is not where compliance is in fact optimized once patients are sedated or as they develop more pulmonary disease. So really the hallmark symptoms of VQ mismatch that we see at the bedside are of course abnormalities in gas exchange, i.e. the patient is hypoxemic or the patient has an elevation in PCO2. And as we think about hypoxemia first, we can identify a multitude of potential causes for a patient to be hypoxemic and as we approach the hypoxemic respiratory failure patient, oftentimes we can think about this in terms of an algorithmic approach. The major mechanisms for hypoxemia are a decrease inspired PaO2, ventilation perfusion mismatch, the pure end of that of intrapulmonary shunt, or some combination of all of these mechanisms, right? So as we think about a patient that has hypoxemic respiratory failure, sometimes the first question that we ask is, is the PCO2 increased? If the PCO2 is increased, then we know that hypoventilation may be contributing to some degree in this relationship. And the way we differentiate that for sure is we learn about the AA gradient. If the patient has an elevated AA gradient, then hypoventilation alone is not the only mechanism for hypoxemia. In contrast, if the AA gradient isn't normal and the patient is hypoventilating with an elevated CO2, then that may alone be the etiology for the patient's hypoxemia. If the PCO2 is not increased, we go through the same mechanism here. We ask about the AA gradient, and if the AA gradient is not elevated, here the problem is just that we're not getting enough inspired O2, i.e. something like altitude, for example. If it is increased, then it's one of the two ends of the spectrum, either complete shunt or ventilation perfusion mismatch. So let's start first here on the far left here and talk about hypoventilation. So what is that relationship between hypoventilation and hypoxemia? Well, this comes, of course, from the alveolar gas equation. Remember, the alveolar gas equation is given as the FiO2 times atmospheric pressure minus the partial pressure of water minus the PCO2 divided by the respiratory quotient, i.e. the PCO2 is a crucial component of the alveolar gas equation. So if we take the normal circumstance of a patient, let's say, that has normal health on room air, here's our normal PbAO2 with a PCO2 of 40 as we plug that into the equation. Now what happens is the patient may hypoventilate and has a rise in the PCO2, let's say, to 80, but it's still on 21% FiO2. Now PbAO2 drops, right, as we plug in 80 for the CO2 in this equation instead of 40, and now the PbAO2 is 50, i.e. this patient is hypoxemic. But for patients that have hypoventilation alone as their source of hypoxemia, we can really overcome that with just a very small amount of supplemental oxygen. So for example, if we increase the FiO2 from 21% to just 30%, i.e. put the patient on, let's say, two liters of nasal cannula, now you see that we can mask that increase in CO2. The patient, i.e., maintains a normal, if not higher, PaO2 and would not appear desaturated in any capacity. The punchline is that hypoxemia that's occurring from hypoventilation alone is very easy to overcome with even a mild amount of supplemental oxygen. Now what about high altitude as we switch over to this side? And again, here we go back to the alveolar gas equation. So now what is changing in this equation, right? And this is related to the atmospheric pressure. So if we would normally think atmospheric pressure is 500 millimeters of mercury at an altitude of 3,000 meters, i.e. this drops considerably from what we had originally put into that equation of 760. As we plug this in, now put 500 in here instead of 760, go through the alveolar gas equation again, now you see the PbAO2 is markedly decreased at 45, i.e. the patient is hypoxemic. But just like with hypoventilation, hypoxemia from high altitude will also easily be overcome with a small amount of oxygen. Again, I increase the oxygen to 30%, multiply this out here, and again the PaO2 is back up to a relatively normal range. So hence, the addition of even a small amount of oxygen can overcome hypoxemia from high altitude or low oxygen. Now both of those two previous circumstances that I talked about, hypoventilation or low-inspired PaO2, would have a normal AA gradient. But most of the conditions that we encounter in the ICU, in fact, have an elevated AA gradient. So what is the AA gradient? Well, the AA gradient, of course, is the difference between the alveolar PaO2 and the arterial PaO2 that you would measure with an arterial blood gas. As we've already shown, here's an example of the alveolar gas equation and how we would calculate PbAO2. And what would a normal PaO2 be? So PbAO2, let's say for this patient, is 99. A normal PaO2 ranges somewhere between 70 and 100, depending upon the age of the patient, etc. So a normal AA gradient is generally around 20 centimeters of water or less. Now if we have somebody that, let's say, has a PaO2 of 60 on 60% FiO2 with a PCO2 of 60, then we can go through this equation and calculate the AA gradient, i.e. we plug in normal values here for barometric pressure and partial pressure of water times the FiO2 of 0.6 minus the PCO2 divided by, let's say, the respiratory quotient, which is 0.8. And what we would see here is that PbAO2 should be 352. Now what do we measure on the arterial blood gas? The measured little PaO2 is 60. This results in a very elevated AA gradient of 292. Now the AA gradient does have a little bit of an age dependence. It is highest in the oldest patients as well as in the youngest patients, but generally this still lies somewhere in this range between 0 and 30. So you see an AA gradient above 30, you know that is definitely abnormal regardless of the patient's age. So what are the causes of an elevated AA gradient? Well, certainly one of the most important causes that we encounter in clinical practice is intrapulmonary shunt. And so the physiology associated with shunt, right, is this idea that you have a lung unit that has no ventilation but still has perfusion. So if we take this very simplified two lung unit model where we have, you know, half of the blood, you know, half of the blood going this way and half the blood going this way, then you have one normal area of ventilation where the PaO2 is 114 here and it diffuses across the alveolar capillary membrane to fully oxygenate that blood. But in this unit, right, there is no communication here between these two. There's no fresh oxygen that's coming in here, so there's no oxygen to diffuse across. So this PaO2 represents what you see in venous blood and the two then they're mixed together. And this is the concept behind the shunt equation or how we how we could think about determining shunt. That we have one component of pulmonary blood flow, right, which has a initial content of venous oxygen that then gets oxygenated after it goes to a normal alveolar capillary unit, and this results in a high capillary concentration of oxygen, okay. But you have another area where in fact there is no communication with the alveoli that goes at a certain amount of pulmonary blood flow, and this most closely resembles the mixed venous blood. These two then mix together to get what we see or we measure in the arterial content of blood. And so this is the idea behind shunt calculations, right. So we have the total oxygen content in the arterioles, which is given by the oxygen content, times the amount of flow that's going through the systemic circulation, i.e. the cardiac output. So this is the total volume of oxygen per unit time that's entering the systemic arteries. Now that's broken up into two components, those that have gone through areas of shunt and those that have gone through areas of more normal ventilation. So the oxygen that's coming from normally functioning units are the total output, total cardiac output, minus the output that's going to areas of pure shunt, times the oxygen content in the pulmonary capillaries, right. And those that are coming from shunt, in fact the oxygen content of those units resemble the mixed venous content times the amount, right, or the amount of that cardiac output basically that is going through those shunted regions. And so you can see this is how we can calculate therefore the two components of this within the shunt equation. Now true shunt is not correctable with oxygen and what this translates to then is to get a measure of true shunt, you actually need to put the patient on 100% oxygen. Remember shunt is the fraction of cardiac output that does not perfuse ventilated lung units, i.e. these are areas with VQ ratios of zero. So you have to put that patient on 100% oxygen and the concept is no matter how much oxygen you put into the alveoli, it's not getting into the blood. That's the real measure of shunt. Now calculating the degree of shunt is therefore very cumbersome and it is not done in routine clinical practice for a variety of reasons, but it sometimes shows up on the boards. And it of course could be done, you know, in areas like the cardiac cath lab because it really requires, right, a measure, a true accurate measurement of the mixed venous sat, i.e., you know, with a pulmonary artery catheter. So let's say you had an eight-year-old child that was intubated with pneumonia and sepsis as an arterial line in a PA catheter. And here again for shunt fraction is calculated on 100% FiO2 with a hemoglobin of 10. Here's the arterial blood gas with a PaO2 of 75, an SaO2 of 97, and a PcO2 of 45. The venous blood gas, the mixed venous blood gas from the PA catheter, venous O2 of 33 with a venous saturation of 65. So how do we do this? First we have to calculate the oxygen content in the arterials, right, so that's our hemoglobin times our SaO2 times 1.34 plus 0.003 times the PaO2, right, all of that is available up above. And you have the amount that's coming from the arterial content here. Venous content, I'm plugging in my values from my mixed venous blood gas, so again the hemoglobin of 10 times the venous saturation of 0.65, and again the venous PaO2, and this gives us the venous content number. Now for CcO2 calculation technically you need the PbAO2, so there's a few assumptions that go into place there, but again we can calculate that if we know barometric pressure and the amount of FiO2 that's coming into the system, as well as the PcO2 going back to that alveolar gas equation. So here I get PbAO2 first, and then I use that as I plug that in now to this equation of CcO2, the hemoglobin times 1.34, again plus times 1 for my FiO2, plus 0.003 times the PbAO2 of 657, and then I plug all those into the shunt equation CcO2 minus Co2 divided by CcO2 minus CvO2 to get my shunt fraction of 0.33, i.e. there's 33% shunt for this patient. Now we did it, but of course that was very cumbersome and does require a PA catheter, which is why this is not really used in routine clinical practice. Now I made the point that to calculate shunt you need to be on 100% oxygen, but sometimes people use the shunt equation even when the patient's not on 100% oxygen, and just what that reflects is that is a combination of hypoxemia that's coming from VQ mismatch, diffusion limitation, and the presence of and true shunt, and this is sometimes referred to as the venous admixture method to estimate shunt, but what's important to remember there is it's not just pure shunt. Areas that are poorly ventilated with VQ ratios less than 1 as well as areas with diffusion limitation will also be represented when you use the shunt equation with an FiO2 of less than 1. Now there are some alternatives that don't require you to use a PA catheter to calculate shunt. Sometimes you can use an inert gas like SF6 as an example, and there have been some validation studies in mostly adults, but there have been some pediatric studies as well where you look at how this tracer is uptaked, and you can use that as a way to non-invasively estimate the degree of interpulmonary shunt, and many of these techniques in fact have very good agreement with what you would get with the true shunt equation with a PA catheter on 100% oxygen. Now it's important to put this in the context of what these metrics may tell us about the mechanisms of hypoxemia, so so far we've talked about the shunt equation on both 100% oxygen in the venous admixture method, i.e. the shunt equation on less than 100% oxygen, as well as the AA gradient. Now of course there are a couple other ones that we probably use more in clinical practice, things like the PaO2 to FiO2 ratio and the SpO2 to FiO2 ratio. So as we go back to these mechanisms, and if we do the shunt equation on 100% FiO2, what this is telling us is quantifying how much pure shunt there is, i.e. areas with no no ventilation. Now that's interesting, but that might not be all that clinically useful for many of the applications at the bedside, which is also part of why we don't use the shunt equation that much in real practice. Now the shunt equation on less than 100% FiO2 gives us a little bit more information in terms of quantifying both shunt as well as areas of Bq mismatch, but in fact that can also be estimated if you look directly at the AA gradient. And the last point that I think is important is the metrics that we use at the bedside routinely, things like the PF ratio, the SF ratio, the oxygenation index, or even the oxygen saturation index, are very global measures of hypoxemia, i.e. in almost all of these mechanisms those values can be abnormal. The PF ratio is probably, you know, one of the most widely used metrics because it can be applied both on invasive and non-invasive ventilation, and it is much less cumbersome than the AA gradient, and certainly there is a clear relationship between the PF ratio and the AA gradient, although this is not entirely linear as you see here. And I will say that the PF ratio is used for simplicity, and it's related to shunt of course, but it is not the same thing at shunt, especially on FiO2s that are less than 100%, because the relationship between the FiO2 and the PF ratio can be altered by things such as the hemoglobin dissociation curve, the arterio-to-venous oxygen difference, the PCO2 and the hemoglobin. So the PF ratio is really measuring a combination of both true shunt and VQ mismatch, and as you see here on this graph that looks at the relationship between FiO2 and PF ratio, even if you keep the AvO2 difference the same, right, as the degree of shunt changes and increases as you go across here, this relationship between FiO2 and PF ratio also changes, and at certain degrees of shunt there's in fact even more of a U-shaped relationship here. So how can we use these hypoxemia-based metrics at the bedside, and how can they help us? Well, certainly they're useful for risk stratification, i.e. we think about things like ARDS to define disease severity, even to look at prognosis, but it may also be helpful for us to gauge response to therapy, i.e. does this patient have persistent disease of some type, or what has the adequacy been of our recruitment strategies, for example, in ARDS on oxygenation or even dead space metrics, but we'll talk about dead space later. And this is of course why hypoxemia metrics such as the PF ratio have been a part of defining ARDS really since its inception and has been included in the most recent Berlin definition of ARDS, the use of the PF ratio. And in pediatric ARDS we use the PALICC definition, which has maintained the PF or the SpO2 to FiO2 ratio for those in non-invasive ventilation, but uses the oxygenation index or the oxygen saturation index for those on invasive ventilation to account for potential confounders in the PF ratio, i.e. the degree of ventilator support that might be applied. And in truth, there is likely a lot of overlap between patients that would be gauged as, let's say, having severe ARDS based upon Berlin criteria versus those based upon the oxygenation index. But as you see here, this is a party study which was conducted in 145 international pediatric intensive care units. And this is looking at ARDS severity grouped by the Berlin PF-SF ratio groups as compared to the PALICC OI-OSI groups. And at ARDS onset, what you'll notice is that there's quite a bit of difference in terms of the severity stratification. The Berlin PF-SF groups would put more patients into this moderate and severe group. The PALICC definition would put more patients into the mild group, for example. Now, interestingly, there is very little discrimination in terms of mortality risk when using initial values for either PF ratio or OI-OSI, although certainly those that are severe have a relatively large difference. Now, this starts to equalize a little bit more over the first 24 hours, i.e. the groups converge a little bit more on each other, which highlights this idea that the way we apply therapies in those first several hours of ARDS management may alter the PF and SF ratio relatively considerably. And here's just an illustrative example of it. So this is a patient that was freshly intubated, is on a PIPA-5. PO2 is 60 with an FiO2 of 0.6 and a mean airway pressure of 8. So the PF ratio here is 100. That would be gauged as severe by Berlin criteria, although the oxygenation index is 8, and that would fall in that moderate range. Now, somebody comes in and tries to recruit the lung, does, let's say, a recruitment maneuver, increases the PEEP, and what you see is that the PF ratio has now gone up, i.e. there's been a response to the recruitment strategy, and the OI has, in fact, stayed the same. So did this patient really go from having severe ARDS to mild ARDS if we were to use the Berlin criteria in a three-hour period? And this highlights that maybe the PF ratio is really that marker of response to therapy, whereas the oxygenation index is capturing a little bit more of the cost associated with achieving that degree of oxygenation by incorporating the mean airway pressure. And this slide simply reinforces that point, that the OI really reflects the cost needed to achieve a given PF ratio, and it may better reflect overall lung disease severity, and the PF ratio may be more a response to therapies that we're seeing. Now, I've spoken mostly about PaO2-based metrics, but one of the real limitations, at least in pediatric practice, is that we don't use arterial blood gases on most of our ventilated patients. We use pulse oximetry on everybody. And, of course, there is a clear relationship between SpO2 and PaO2-based markers of lung disease severity in children, and there is a linear relationship here between the oxygen saturation index, which substitutes the SpO2 for PaO2 in that equation, or the SF ratio and the PF ratio, although a one-over transformation is needed to make this fully linear. And this is applicable to use these pulse oximetry-based metrics as long as the saturation is less than or equal to 97%. And using these SpO2-based criteria are important because most of our patients, i.e. over half of our patients that are mechanically ventilated, right, do not have an arterial line. And so we're likely really underestimating the number of cases of ARDS because patients don't have an arterial blood gas. And we also might be selecting for a sicker patient population if we're only relying upon those with PaO2. And this point has been illustrated in the PARDI study. If we look at those that meet full Berlin criteria at PARDS diagnosis, and this diagnosis is based upon meeting the PALIC-PARDS criteria, right, with oxygenation index or oxygen saturation index, or even SpO2 to FiO2 ratio, then we see that only a small proportion have a PF ratio available with bilateral infiltrates, and these are the patients with the highest mortality. There is a very large number of patients that also still have bilateral infiltrates, but don't have a PF ratio that's available. Now, when you go out to three days, there is still a substantial number of patients that never get an arterial line, right? There's more that have an arterial line here, but there's still a substantial number here that never get the arterial line. So one major reason to use the SpO2 base criteria is simply to capture more patients in the diagnosis of conditions like ARDS, or even to risk stratify them, you know, in mortality prediction models like PIM or PRISM, for example. But it might also be that pulse oximetry-based metrics can better represent the concept of intrapulmonary shunt. Why? Well, if we go back to the shunt equation, remember that the SaO2, right, has in fact the higher contribution to the oxygen-carrying content compared to what we have with PaO2-based, with PaO2. And in fact, in the study that we did where we looked at PF ratio, SF ratio, oxygenation index, and oxygen saturation index on all mechanically ventilated patients, and we looked at the risk of overall mortality, described it with the area under the curve of the receiver operating characteristic plot. Certainly what we saw is that there were a lot more patients that could be diagnosed with pulse oximetry-based criteria, and that the SF ratio had a stronger ability to discriminate mortality even than PF ratio or some of the other metrics. Now, this is of course confounded by the fact that we have a lot more patients in the SpO2-based group. So if we limit it to only the 400 or so patients that had all of the metrics available for computation, we still find this same idea that the SpO2 to FiO2 ratio, in fact, it has the strongest association with mortality more so even than the PF ratio or even the oxygenation index. And this improves slightly as we look at the overall average over the first day of ventilation. Now, another important use for these hypoxemia-based strategies is to help us think about therapies. And certainly as we look to the adult ARDS literature, for example, and as we think about PEEP management approaches, stratifying based upon PF ratio might be really important that those that had a PF ratio less than 200 were more likely to benefit from higher PEEP strategies, whereas those with more modest PF ratios between two and 300, in fact, maybe were harmed from very high PEEP strategies. And in fact, this is the concept behind why there are mild, moderate, and severe stratifications in the Berlin definition. And the same holds in the Palak definition that the therapies that we're going to consider will differ based upon the severity of the patient and the potential likelihood of benefit or response to those therapies based upon their initial degree of hypoxemia. Now let's switch gears away from shunt and hypoxemia to the other side of the coin here as we think about alveolar dead space. And remember dead space is areas that are receiving ventilation without perfusion and the hallmark symptom is hypercarbia or a large difference between the CO2 we measured at the airway, like the end tidal CO2 is markedly lower than what we may see in the arterial CO2. So remember that adequate CO2 removal is dependent upon our alveolar minute ventilation. And what we measure and see at the bedside often is our total minute ventilation, which is the respiratory rate, of course, times the tidal volume. But remember that tidal volume has two components. It's got a component of dead space as well as a component of alveolar ventilation. And what really determines CO2 removal is the alveolar minute ventilation, right? Which is our rate times our alveolar ventilation, the total tidal volume minus the volume that's going to dead space. So we set the tidal volume, overall tidal volume, and maybe the rate, let's say on a patient that's mechanically ventilated, but it's the underlying disease state or maybe how we're managing things on the ventilator that ultimately control the dead space component. And remember there are two components of dead space, right? There's anatomic dead space and alveolar dead space. And under conditions of health, normally anatomic dead space is fixed. We may introduce more dead space, for example, when we intubate a patient, if we add a lot of ventilator tubing, et cetera. And then there's alveolar dead space, which under normal circumstances is minimal. But as patients start to develop disease, we may see increases in alveolar dead space related to lung disease severity and BQ mismatch, inadequacies of pulmonary perfusion, like pulmonary hypertension or poor cardiac output, or maybe even how we manage the ventilator with over distention, for example, in some areas. And the relationship between CO2, PCO2, and minute ventilation is affected, of course, by the amount of dead space ventilation. In circumstances with low dead space ventilation, here we see this relationship that at a normal CO2 and a normal minute ventilation, let's say five liters per minute, here's what this curve looks like, that we can lower the CO2 further, right? With modest increases in the overall minute ventilation as long as dead space is low. In contrast, when dead space is very high, right, then it is very hard to achieve that even a normal, achieve a normal PCO2 without really significantly increasing the total minute ventilation. And if we wanna further reduce that CO2, we really exponentially have to increase that minute ventilation. But of course, it's not just the total minute ventilation, right, it's how that minute ventilation is achieved. Because the same minute ventilation can result in a different amount of alveolar ventilation depending upon the tidal volume that's chosen. So in this example, right, where I have a, in all of these situations, we have a total tidal volume of, or a total minute ventilation, excuse me, of eight liters. In this circumstance, that's achieved with low tidal volume ventilation and a fast rate. And in this circumstance, it's high tidal volume ventilation and a low rate. And certainly the higher the tidal volume, generally the more alveolar ventilation we will receive because you overcome that dead space or anatomic dead space component relatively easily. And then you get to more alveolar ventilation. Now, I'm not advocating that strategy, right, of high tidal volume ventilation. That certainly is what causes harm, especially from lung injury, but it's to illustrate that point. So how can we measure dead space? Well, there are multiple ways that we can think about measuring dead space at the bedside. The more, most classic initial description was based upon the Douglas bag. Since then, volumetric capnography has emerged as probably the new sort of standard way to measure dead space or gold standard way to dead space and to differentiate alveolar dead space from anatomic or airway dead space. But oftentimes volumetric capnography is not available widely at the bedside. And so we can use simpler but more crude approximations from estimates from the alveolar gas equation, minute ventilation corrected for PCO2 or the ventilation ratio, or things like time-based capnography. So the fundamental component with measuring dead space is trying to quantify the difference between the CO2 that's measured in the bloodstream compared to the CO2 that's exhaled. So in this circumstance where there's no significant dead space, right, you have a venous CO2, let's say a 46, that diffuses across the alveolar capillary membrane and you have exhalation of that CO2. And now the new CO2 on the other side, let's say is 40, is also 40 centimeters of water. So there's minimal difference between the end tidal CO2 that could be measured at 35 to 40 and the PCO2 that's measured, let's say a 40. Now, as you develop dead space, i.e. there's no perfusion now to this unit, now there's no CO2 that's coming here. So there's no CO2 in this lung unit that's exhaled. There may be a normal amount of CO2 in this lung unit, right? And so these two mix together and then what you see is a lower amount of CO2 that's exhaled relative to what you see in the bloodstream. So one of the first methods that was done to calculate dead space was based upon the Douglas-Fagg with a modification of the Borengelhoff equation where total VDBT, right, which is the dead space to tidal volume ratio is the PCO2 in the bloodstream minus the mean expired CO2 divided by the PCO2. So how was mean expired CO2 measured? You would simply collect all of the CO2, you know, with a mouthpiece and they put a nose clip over the patient, exhaled into this bag and you measure the CO2 concentration in that bag and compared it to what you got in the bloodstream. And this is one of the early methods that was used to demonstrate that dead space is highly relevant as we think about ARDS as an example, that in this adult trial, that the mortality increases as dead space fraction increases in these adult ARDS patients. And in fact, that dead space was independently associated with higher risk of mortality. There are limitations with the Douglas-Fagg, it's clunky, it's somewhat difficult to use in actual practice, you have to put it into the ventilator circuit and there may be some inaccuracies when it's put into the ventilator circuit, especially with humidification, et cetera. And so other techniques have been developed including volumetric capnography, either a single breath method or a method to just measure mean expired CO2 more in real time like the Douglas-Fagg technique. But the single breath method basically simultaneously is measuring tidal volume, so there's a flow sensor in volumetric capnography as well as the CO2 sensor and it's plotting continuously the tidal volume against the CO2 that's measured. And if you put in a value for the arterial blood gas or arterial PCO2, now you can compartmentalize this curve into the component that's due to anatomic dead space, which is given here in Z, the alveolar ventilation, which is in X and the alveolar dead space, which is given here in Y and these are compartmentalized into these equations that are shown above. Now, what the algorithms will do is they'll take this curve that's drawn and they'll divide it into the three components of the curve that's coming from anatomic dead space, the mixture between dead space ventilation and alveolar ventilation and then the alveolar ventilation and put half of it on this side in the alveolar ventilation and half of it on this side in the dead space ventilation. Now, as you can imagine, there may be some problems with some of that analysis, especially if it's trying to cut this phase two curve in half, if you've got a slope of that line that's different, as an example, it may be difficult to accurately differentiate phase two and phase three. So there are also some waveform independent techniques that go back to the Bohr-Engelhoff equation where you simply use this CO2 sensor to give an estimate of the mean expired CO2 and then calculate the overall VDBT. And there has been several investigations, mostly in adults that have shown that the Douglas-Bagg technique as compared to the volumetric ethnography technique yield relatively similar values, especially for overall VDBT. Now, there are some other techniques that have been predominantly validated in adults. So that you don't have to use any specialized equipment. And conceptually, you can think of any patient that needs to generate a very high minute ventilation to achieve a relatively normal CO2 as a patient that must have elevations in dead space. And that's what some of these more non-invasive estimates are based upon. This concept that the PCO2 is equal to the CO2 production times a constant, right? Divided by your overall alveolar ventilation where remember the alveolar ventilation is your total minute ventilation minus your dead space ventilation. So you can therefore try to solve this equation to calculate what the dead space fraction is if you can measure the PCO2 and you get an estimate of what CO2 production would be. And the CO2 production estimate is coming from equations like the Harris-Benedict equation. So as you can see, there are some potential inaccuracies in these estimates. Nevertheless, certainly this has been shown in adult ARDS patients that using these kinds of estimates show that there's a sort of stepwise increase in risk of mortality with higher estimated dead space fractions that have been done using these methods. Now, I have spoken of shunt and dead space as if they are not tied to one another, but of course they're very highly related to one another. And remember, CO2 elimination is typically much more efficient than diffusion of oxygen. But as the shunt worsens, the dead space fraction will also increase and that's dependent upon the overall cardiac output as an example, i.e. here's the relationship between shunt fraction. The shunt fraction increases, the dead space fraction will also increase, but that the slope of this line is much less for those that have a better cardiac output. But as cardiac output worsens, now there's a more of a sort of stronger linear relationship with the higher slope. And cardiac output is only one of the variables that affects this relationship between dead space here and shunt on this axis, right? That as an example, acid base status and hemoglobin also have an effect on changing this relationship as well as metabolic versus respiratory acidosis and where they fall on this relationship. And this relationship between the PaO2 and the pCO2 is regulated by what's called the Haldane effect, right? And this is based upon the idea that deoxygenated hemoglobin is a higher affinity for CO2 because it can more readily be a proton acceptor than oxygenated hemoglobin. So when deoxygenated hemoglobin, i.e. if you look at the tissues, increases the amount of CO2, which can be carried back to the lungs to be exhaled, i.e. it's going to pick up that CO2 to bring it back to the lungs. But then when it hits the oxygenation at the lungs, right? Then CO2 more readily dissociates from the hemoglobin so it can be exhaled. And so this also affects that shunt and dead space relationship. And in fact, so therefore one of the important pieces as you look at the relationship of dead space and mortality in ARDS is to control for the degree of shunt that's present there. And so that in fact has been validated here in this adult study, looking at these markers of VDBT after controlling for things like the PF ratio or the oxygenation index, highlighting that there does retain an independent association between dead space and mortality in adult ARDS. Now, another alternative is to use time-based capnography. Time-based capnography, other than volumetric capnography, right? Is just looking at CO2 as a function of time. And here's where we measure often the end tidal CO2, i.e. this point at point A. And so you can calculate the alveolar dead space fraction or ABDSF, which is the PCO2 minus the end tidal CO2 divided by the PCO2. Now, what's important to remember is that the shape of this curve is important. That you really are looking at the end tidal CO2 here at the end as a reflection of the mean expired CO2 that's measured here. So if you've got a square wave that can be readily applied, if you've got a triangle like this, you would see an obstructed airways. Then in fact, the areas are in fact not quite represented because it's not really a rectangle. So you should not apply ABDSF in situations of significant lower airway obstruction as it's unlikely to be representative. But now there've been several pediatric studies. Here's one that came out of the PALICC secondary analysis from data that Anup Bala had gathered here at Children's Hospital Los Angeles. That you see that there is in fact an independent relationship between dead space and mortality in pediatric ARDS, even after controlling for the degree of hypoxemia. That not surprisingly mortality increases as PF ratio decreases, i.e. shunt is associated with mortality in ARDS. And that those with significant elevations in dead space, have a threefold higher mortality than those with normal amounts of dead space, as you see here. But that importantly, even when you stratify by the degree of hypoxemia, within each of these sub-stratifications, the risk of mortality with elevated dead space is nearly two to threefold higher. And this has been recreated in all ventilated patients that have an arterial line, the same basic concept of elevated dead space is independent of oxygenation index. And the same thing in another cohort of ARDS patients. And this has been independently validated in another data set from the Children's Hospital of Philadelphia, which shows that ABDSF at PARDS onset has a strong relationship with mortality. Now, interestingly, it becomes less apparent at the 24 hour mark. Here is that PARDS onset, there's a clear relationship with very high ABDSF and higher mortality. But that the 24 hour mark, this is less predictive, and it might be modulated by use of therapies that affect pulmonary perfusion, such as the use of nitric oxide. In addition, in pediatric patients, the use of this time-based capnography with ABDSF has been validated against volumetric capnography, showing very strong associations between overall VDBT and alveolar dead space fraction, and in particular, the strongest association between ABDSF and the alveolar VDBT as measured by volumetric capnography. Now, the last area I'm gonna talk about is dead space and therapy. I think it's become clear that there's prognostic relevance with dead space. But as we think about the mechanisms, especially as in ARDS with elevated dead space, it can be threefold. It could be reflective of microvascular injury, i.e. endothelial dysfunction. It may relate to overdistension and potentially be useful for ventilator management. But it might also be a marker that the patient has inadequate blood flow, impaired oxygen delivery, or impaired cardiac output. So can we use this to help with therapy decisions? Now, as we think about the pathobiology of ARDS and the risks of harm and mortality, it has been clear in both the adult and pediatric ARDS literature that abnormalities with coagulation and microvascular thrombosis increase the risk of mortality, i.e. patients with the lowest amount of protein C and the highest amount of pi1, this is a prothrombotic milieu, have the highest risk of mortality. And this has been replicated in pediatric patients that non-survivors are more likely to have larger elevations in pi1 as an example, and that is also associated with ventilator-free days. And so therapies targeted at coagulation have been an area of interest in both sepsis and ARDS for many years, including the use of activated protein C in adult patients with ARDS. And what's interesting is while there was no clear benefit overall, what we see is that maybe the use of a dead space fraction can be helpful to identify patients that are responding potentially to therapy. And so here what we see is in the placebo group, there was minimal change in dead space fraction either on day one or day four, but those with activated protein C had a much larger reduction in dead space. And so this might be related to that improvement in microvascular thrombosis that therefore results in better pulmonary perfusion and a reduction in dead space fraction. It may also be helpful to help titrate PEEP. And certainly as we think about PEEP titration, it's often a balance between oxygen delivery, lung compliance, CO2 elimination, over distention, and the dead space component, here this is the mean expired CO2, gives us a reflection of that balance between pulmonary recruitment and pulmonary blood flow. And you can use this as an example with this decremental PEEP titration, where the esophageal pressure, airway pressure, and mean expired CO2, perhaps to find that point at which mean expired CO2 is highest and that balance point. And here's just another illustration of that here, that if you use compliance, you may get one value of sort of optimal compliance. If I look at VDVT, I may get a slightly different value for example. And so the relationship might be close to one another for many patients, but it might be that there is significant heterogeneity in this response in other patients and that you can use both of these metrics to potentially find that optimal PEEP. It also might be relevant as we think about whether the lung disease severity or the pulmonary perfusion abnormalities have improved enough that the patient is ready to let's say be extubated. And this is a study that was done in a lot of cardiac patients and children back in 2000, which highlighted that the risk of extubation failure almost had a linear relationship with the VDVT and that a relatively high VDVT, a VDVT of 0.5, had this balance cut, balance between sensitivity and specificity to help with extubation decisions. Now, others have tried to replicate this and mixed results. Some have shown no difference in those with successful versus unsuccessful extubation. Others have shown modest differences. Now, this is a complicated area I would say because there are many potential mechanisms for elevated VDVT. And in the first Hubble study, there were a lot more cardiac patients. And so maybe this is more a reflection of the pulmonary perfusion and cardiac output. And that might not be the same in patients that have more pulmonary parenchymal disease, for example. The other issue of course, is that there are many reasons that patients fail extubation. And now we use a lot more noninvasive ventilation, all of which can confound this relationship. So again, here is the board outline for pulmonary gas exchange. I think we've covered all of these topics and hopefully this has been helpful. All right, so the first question, which of the following statements about shunt and dead space are true? Alveolar PbAO2 is typically greater than 100 in areas of shunt. Two, alveolar PCO2 is typically 40 in areas of dead space. Three, in areas of shunt, alveolar PCO2 is similar to PaCO2 in the mixed venous blood. Four, dead space can be estimated from arterial blood gas, data and fraction of inspired oxygen or none of the above. So the correct answer here is three, in areas of shunt, alveolar PCO2 is similar to the PCO2 in the mixed venous blood. Remember the other ones here are wrong. Alveolar PbAO2 is not high in areas of shunt. Remember areas of shunt are areas that are perfused but are not receiving ventilation. So no O2 is coming into that alveoli in the safe sort of contrast of what you see for number two, where PCO2 is typically 40 in areas of dead space. Remember in areas of dead space, there's no ventilation. So PCO2 would be expected to be zero, the alveolar PCO2 in those areas. And then number four, dead space can be estimated from arterial blood gas data and not FiO2. Here, we look at the CO2 value. Remember dead space is about CO2 and oxygenation metrics are about PaO2. Shunt is about PaO2. Second question, two-year-old with ARDS and sepsis is intubated with a 4-O-cup tube. Ventilator settings are assist control, pressure control, peak inspiratory pressure 28 over a PEEP of 10, FiO2 is 0.6, tidal volume is 8 mLs per kilo. Respiratory rate is 45. Blood pressure is 65 over 45 with a heart rate of 180. Cap refill is five seconds. pH is 725 with a CO2 of 65, a PaO2 of 68, end tidal is 37, which the following statements is true. The patient has a normal dead space fraction. Two, dead space cannot be estimated because the end tidal CO2 is not correlating with the PCO2. Three, the primary mechanism of dead space is related to an increase in anatomic dead space from an endotracheal tube. Four, fluid bolus will likely decrease the dead space fraction. Or five, reducing the tidal volume to 6 mLs per kilo will likely decrease the dead space fraction. Okay, so the correct answer here is four, fluid bolus will decrease the dead space fraction. So this is a patient that has inadequate cardiac output or low blood pressure, high heart rate, diminished capillary refill. So inadequate blood flow is probably the main mechanism for elevated dead space here or one of the mechanisms here. So improving cardiac output, fluid bolus should decrease the dead space fraction. Dead space fraction is certainly elevated. As you can see, the end tidal CO2 is 37 and the PCO2 in the blood is 65. So this tells you dead space fraction is high. And in fact, two is wrong because that's the definition of how we estimate dead space is to show that the end tidal CO2 and the PCO2 are different from one another. The end tidal CO2 are markedly lower than the PCO2. This is an anatomically appropriately sized endotracheal tube, so it would not be expected for there to be a significant increase in anatomic dead space. And reducing the tidal volume will unlikely decrease the dead space fraction in any significant amount. It would only be if there were significant overdistension that was occurring with the tidal volume of 8 mLs per kilo, which is unlikely to be true, especially with the peak inspiratory pressure of 28.

gas exchange physiology

pediatric intensive care unit

shunt

dead space

hypoxemia

hypercarbia

volumetric capnography

time-based capnography

PF ratio