Well, good morning, everyone. It's Wednesday, the last day of the conference. Thanks again for joining and making it on the last day. Our numbers are starting to kind of dwindle, but I really appreciate everyone coming today. I think, I hope we will have a great session for the next two hours for everyone. My name is George Lopez. I am a neurologist, a neurointensivist, inpatient neurologist. And we've got a great lineup this morning. A couple of reminders. Let's silence our cell phones. And we're going to start today. So the goal of this topic is really to kind of give a pretty large overview of the neurologic complications of ECMO. And we're going to kind of try to cover everything from little adults, so little babies. So we've got two pediatricians talking today. And then kind of cover the entire gamut of what can and sometimes does occur to our patient's brains in those undergoing ECMO. ECMO, as you know, is really burgeoning over the last several years, becoming more widely available. And a lot of the big centers are obviously running ECMO programs, but even other community-based and smaller hospitals are starting to get more involved. So I'd like to start with Tom Black, who, if you don't know, is probably one of the grandfathers of neurocritical care, has a lot of experience, obviously, in this realm, and actually has recently published a nice review article on the neurologic complications. Then we will move on to some of our other speakers looking at neuromonitoring. Is it useful? It's certainly available. We'll have our pediatrician up here talking about what happens in our little ones that undergo ECMO. And then we'll also talk about some of the obvious complications, that is brain death. Some of our patients, unfortunately, don't make it through the entire process. So how do you determine someone is brain dead while they're on ECMO? Everything else is being oxygenated and perfused, except one vitally important organ. And then we'll also have a great speaker who will talk about the long-term outcomes and how to measure them, and actually seeing these patients in the clinic, and how to actually set up a clinic-based, in a sense, outcomes clinic-based practice. So without further ado, let's start with, and we're going to hold the questions to the very, very end. So we'll have about 20 minutes at the very end. So save up all your questions at the very end. We want to try to keep it smooth going. We won't take questions after each speaker. So let's have Tom Bleck as our first speaker, talking about the complications in adult patients. All right. Well, thank you, George. Glad to see a few people in the audience. Who does hands-on real work with ECMO patients? Just a good, quite an assemblage there. Who does just consultation? And who just wants to learn? All right. So we'll perhaps tailor it a little bit that way. None of these disclosures actually have anything to do with ECMO. We had looked at what was available in terms of either guidelines or ways of trying to suggest management for the patients who are on ECMO who need neurologic critical care. By the way, instead of bothering to take notes or take pictures of all the slides, if you just put your camera on this, you'll get a PDF of all the slides. It'll also be available as the last slide. OK. So, I'm only showing this part not for the management but for what we found in terms of epidemiology. So, these were the issues that we uncovered. So, patients of course have global ischemia. Most of the global ischemia occurs before or during cannulation. So, we're looking at the consequence of why they're being put on ECMO usually rather than a problem with the ECMO itself. Although it is possible for patients to become hypoxic on ECMO, usually it's controlling pretty well. The patients who get ischemic stroke represent a relatively large proportion of the neurologic complications. We'll talk about some mechanisms there. You might expect that patients on VA ECMO would be more likely to have stroke than the VV ECMO patients. Turns out that the VV ECMO patients actually have a substantial percentage of the strokes, ischemic strokes that occur. And it's not all PFOs. There are a lot of things going on that we don't precisely understand. Intracerebral hemorrhage of course can occur perhaps related to anticoagulation, related to thrombocytopenia which occasionally arises. But interestingly, one of the major triggers for intracerebral hemorrhage is a rapid drop in the PCO2. And why that happens, you know, usually patients who are going on to ECMO have much more of an oxygenation problem than a CO2 retention problem. But if you have become hypercarbic over time and have compensated renally, if you immediately then take somebody, put them on ECMO or even put them on regular mechanical ventilation, and rapidly raise their pH by blowing off that accumulated CO2, it can produce a tremendous amount of vasoconstriction. And that brain shrinkage in an older patient may trigger the usual sorts of extra axial hematomas. Why it would trigger intraaxial bleeding is another issue that remains unresolved. Air embolism can occur if the oxygenator is not being managed properly, but it's quite rare. Seizures and status epilepticus are sometimes hard to recognize in that the patients are often sedated or even receiving neuromuscular junction blockade. ICP management is not all that different here, except that you're unlikely to get an ICP monitor since most of the patients are anticoagulated. And what to do with sedation and analgesia, I'll just say nobody's got any particular magic formula for doing that. So the group at Basel did a very nice review looking at the world's literature on the complications that occur. If you look at the amount of ECMO that occurs based on relative size of these circles, you can see that over time it's increased a fair amount. This line represents the A1N1 epidemic that led to a much larger number of patients getting put on ECMO because of ARDS. The reddish circles are VA ECMO, and the bluish ones are VV ECMO. So you can see that a lot of the early complications didn't either distinguish or were reported with VA, but as time has gone on, we're seeing more and more complications related to VV ECMO as well. Overall, if you look at the world's literature when they did, 13% of the reported patients had a neurologic complication. But if you excluded the ones that usually occurred either before or during cannulation, it was down to about 9%. Both ischemic stroke and intracerebral hemorrhage make up about 5%. About 2% of the reported patients had seizures, but most of these patients have not had EEG monitoring. We're going to hear more about monitoring from one of the later speakers, so I won't say too much about it right now. So again, looking at the VA versus VV, we've got more complications related to VA, but still two-thirds as many related to VV. The types of complications that we're seeing, the odds ratio goes up, as you might imagine, a little bit with age. The use of inotropes did it a little bit. Now, of course, most of the blood pressure control you're doing is related to flow, but some of the patients do need inotropes, depending on the reason for their ECMO. Getting hypoglycemic actually was a substantial problem. Since many of the patients are sedated, it may be difficult to recognize that they're hypoglycemic unless you're checking frequently. And also, since many of them are not going to make any kind of a blood pressure response to becoming hypoglycemic, there won't be a clue that they're having a hypersympathetic response. And having had a cardiac arrest, so usually having ECLS by itself, is almost a three-fold rise in the risk of complications. Hemorrhaging not surprisingly relating to drops in fibrinogen or to problems with platelets down here are substantial. You can see thrombocytopenia is a major risk factor for intracranial hemorrhage. The change in the PCO2 doesn't show up in most of the epidemiologic work because there's not that many patients who are that hypercarbic. But the ones who are hypercarbic are the ones who were at highest risk. Need for dialysis. Women are at more risk probably because vessels are smaller. The creatinine rise is important. Most of the patients are receiving heparin, but you can run ECMO without heparin. You'd have to change your oxygenator more frequently. In terms of ischemic stroke, the one thing they uncovered was an elevation in lactate beforehand. So people who were in cardiogenic shock, which usually don't go on to, well they'll all be on VA ECMO, sorry. So again, I'm not going to talk about the neurophysiology, but this is another source of interesting information. So in this study, looking at our patients, there were no intracranial hemorrhages in this group. But some of the patients by scan developed cerebral edema, which seemed to be global, not related to hemorrhage, the mechanism for which isn't clear, although hypoxic ischemic encephalopathy may be related to that. You can see that some of them had fairly impaired outcomes. And where they went, of course many of them had withdrawal of life-sustaining therapy because they weren't improving. Some of them died related to the arrest that got them on to ECMO. One of the things to keep in mind is the balance between what VA ECMO is delivering to the patient and what the patient's own heart is doing. So if you have someone whose heart is not pumping or is not pumping very much, then even though you're delivering oxygenated blood into the femoral artery up the aorta, you're still perfusing the cranial vessels with good oxygenated blood. But as the heart starts to improve its contractility, it then starts to pump whatever blood came out of the lungs, basically fighting with what's coming up from the oxygenator, sometimes called the north-south phenomenon. As you can see, if the cardiac contractility improves, but the lungs are still bad either because of pulmonary edema or because of ARDS, you're going to end up getting deoxygenated blood going out into the cranial vessels. Now, we typically require that there be a right radial arterial line so that we're able to sample this blood as it comes out to get the best measure of how poorly oxygenated that blood is. If you take the PO2 off the oxygenator, you'll find that it's astronomical. But it's the balance that is getting to the brain that's going to be important in terms of many of the neurologic complications. Looking at the study from France of ischemic and hemorrhagic lesions in VA ECMO, again, in this group, there were some patients who did have hypercarbia. And that, even in the VA patients, was a risk factor for the development of intracerebral hemorrhage. They looked at a number of different parameters to try and tease out what was different between the patients who did not have any brain damage, so in a very large population compared to those with either ischemic stroke or intracerebral hemorrhage. And for pH, there was really no difference. The mean for PCO2 was higher here, and the range was higher. Of course, these are interquartile ranges. So the few patients that were really high were up in that highest quartile. Again, not that much difference in the lactate. Fibrinogen wasn't that much lower. But if you had the really low fibrinogen, then you were more at risk for hemorrhage. Looking now at risk factors for ischemic stroke, if you were younger, you were protected. In this study, women were protected. If you weren't quite as sick, that was good. Just going on RRT, though, started to raise the odds ratio here. And you can see, if you had a coincident placement of an atriotic balloon pump. So the balloon pump will give you maybe an extra liter and a half of cardiac output. The ECMO could give you four or five liters. But if you're on the balloon pump beforehand and then went on to ECMO, that by itself raised the risk for stroke a little bit. You can see these other things. The ones that were statistically significant were central cannulation. So there are two reasons for central cannulation. Most central cannulation is the patient couldn't come off of cardiopulmonary bypass in the operating room and just got put on to ECMO and brought out to the intensive care unit. But there are patients whose peripheral vessels either couldn't be cannulated or the flow wasn't great enough and they eventually had to open the chest and do central cannulation. And that does raise the risk of stroke here. And also, the platelet count was an issue. Looking for intracranial bleeding, the ones that were statistically significant, now women were more likely to have that, whereas they were somewhat less likely to have ischemic stroke. Again, central cannulation and thrombocytopenia. Looking at venovenous ECMO, they had 135 EV patients. They said they had five patients who had CT scans performed without clinical signs. Now, it's sometimes suggested that once you get the patient on ECMO, you scan them, especially if it was ECLS, to figure out whether there's an intracranial lesion at the start. It's gotten a lot easier to do that as the ECMO devices have gotten smaller, it's easier to transport them. So of those patients, one had edema present supposedly without clinical signs. But again, clinical signs are sometimes not as apparent in patients who are heavily sedated or under neuromuscular junction blockade. Of these patients who were found to have clinical brain events, three of them went on fairly quickly to brain death and had no imaging. Of the patients who were scanned, some of them had nothing visible even though they had clinical phenomena suggesting an intracranial process. Of the ones who had something visible on CT scan, 10 of those 16 had intracranial hemorrhage. Three were ischemic strokes. Again, this is VV ECMO. Two of them had cerebral microbleeds, and one just had global edema. Looking at the physical findings that suggested that there was a problem, you know, blown pupils, patients unarousable, apparently brain dead, hemiplegia. It's important to recognize, by the way, that someone who's getting VA ECMO with a line in the femoral artery might not be moving their leg because the leg is ischemic, not because they've had a stroke. And typically that's treated by doing a bypass from that arterial return cannula to a little bit lower in the femoral artery. But if you're not attending to the possibility that the leg weakness or the leg immobility is due to ischemia rather than to something in the head, you'll miss treating the leg properly. Seizures were pretty rare, but again, seizures are often detected only by EEG monitoring if you're doing it. This is looking at that CO2 phenomenon again. So the patients who had a big boost in pH and a drop in the PCO2 were the ones who ended up with the intracerebral hemorrhage. Although those who were having oxygenation problems, and you can see there's a pretty wide range of PO2s that were seen in the patients who had ischemic stroke, I'm sorry, intracerebral hemorrhage. Interestingly, one of the ways of doing VA ECMO that is perhaps the fastest is to put a dual lumen cannula in through the jugular vein, ideally having the jet of the oxygenated return blood going straight in through the tricuspid valve so that you get good oxygenated blood going to the left side of the heart. But it seems to make a difference as to whether you're using a large or a really large cannula. They're all large. But the reason for this is somewhat unclear. The authors have suggested that perhaps there's obstruction to venous return producing intracranial hypertension. Of course, we all put a lot of lines in jugular veins, but these are really big. And it may be that there is some problem with jugular venous return producing intracranial hypertension. Is this producing something equivalent to a venous infarct with hemorrhage? It's really not clear from the published literature what's going on there. Now for ECPR, so people who are crashed onto ECMO because they can't be resuscitated, you have already likely a period of hypoxemia. Often the CPR wasn't as good as you wanted in the first place. So you expect more neurologic complications to occur as a consequence of ECPR. The fact that the heart isn't contributing to flow at all does raise the likelihood of a stroke substantially. So these are people in whom, because the heart is doing very little, you've got a constant blood pressure coming out of the ECMO device because it's a constant flow, not pulsatile. As the heart begins to add flow, then you start to see a pulse pressure. So if the heart doesn't work at all, there's no pulse pressure at all. As the heart gets better, there's more pulse pressure. But the people who had very little cardiac contributions were the ones who had the most neurologic complications as a consequence of going on to ECPR. And as you might imagine, the older you are and the longer that it takes to establish flow again, the more likely you are to have a neurologic complication. So the likelihood of a poor neurologic outcome for the younger patients who got on to ECMO pretty quickly, here was about almost 50% in this group, the neurologic complications are really the main problem with ECPR. But as the pump time increased, or the time to get on the pump, even in the younger patients, they were somewhat worse. The older patients, the older being over 65, had a much worse prognosis. So you might wonder, is it ever worthwhile doing it in this circumstance? And I think that's something we need to reconsider. So this is the suggested workup algorithm from Souter and his colleagues in Basel. They suggest that you do a daily interruption of sedation and neuromuscular junction blockers in order to figure out whether there is something on physical exam that would clue you in to the possibility of a neurologic problem. When you've done that interruption, then see what the GCS is. And if consciousness is altered, make sure that you're not looking at residual effects of the drugs you've been giving, or the drugs the patient was taking beforehand. There are, of course, a lot of possibilities for metabolic complications. If you've done those things and the cause remains unclear, then they're looking at some very specific physical exam findings, mostly now looking for either brain stem or cerebral lateralized pathology that might clue you into looking for structural problems. Meyer's reflex is not commonly done in the US, but it's another way of looking at corticospinal function. And consider an EEG if there's no explanation. And finally, go on to scan. Well, that's one way to look at it. I think more likely what happens is we get to this point, and then they get rushed off to the scan right away. And I'm not really opposed to that. I think when you reach this point, you should decide you're going to do a scan and an EEG. Just because you find something structural doesn't mean the patient's not having seizures. And if you don't look, you won't find and treat them. But nevertheless, this is, I think, probably the most useful guide to how to approach these patients, making clear that you really need to stop the things that will obscure your exam before you just say, well, yeah, the patient's been on ECMO for two months, and now we're going to find out if they're alive or not. So with that, I'd like to thank you. If you either email me or just scan this QR code, I'd be happy to get the references. And thank you very much. Thanks, Dr. Bleck. Our next speaker, we're going now from adults to the pediatric age population, Mark Wainwright from the University of Washington, who is section chief of the Pediatric Neurocritical Care, is going to talk to us about what happens in the little ones who undergo ECMO. They have a lot more data because the kids have been undergoing ECMO a lot longer than adults. So it'll be nice to kind of hear, compare and contrast between the pediatric age population and the adult age population. So Mark. Good morning. I don't have any relevant disclosures. So I'm going to cover briefly an overview of VA and VV ECMO and the associated neurologic complications. Then I'm going to talk about what the definitions of neurologic injury are in children because that becomes relevant, particularly relevant when one's thinking about what the morbidity of ECMO is in pediatrics. From there, I'll go and review the neuromonitoring modalities, primarily EEG and TCDs with a brief comment on NEERS, which has not really been adopted in the neuromonitoring with ECMO in children, and the potential of biomarkers. And then particularly important for children is what the long-term functional consequences are of a vascular injury or a seizure. We'll look at the literature there, and then I'll finish with some cautious but practical recommendations. So the ELSA registry has been used as a tool for looking at neurologic morbidity in ECMO in children. These are data from a study published by the Boston Children's Group. The ELSA registry data from between 2006 and 2008. So about 2,700, 2,800 ECMO cases, the majority of which were neonates. And a couple of key points here. The first, that the majority of the morbidity occurred in neonates. The majority was associated with carotid cannulation, and the overall incidence of seizures were 21 percent. So the data from the study. The caveats are important, too. It's retrospective. This is from a registry. When one talks about seizures here, this is based on reported clinical seizures. Use of EEG monitoring was not standard in this population at this time. So we don't know anything here about morbidity associated with electrographic seizures. And the imaging was based on clinical indications. Imaging was not a standard here. And importantly, none of this neurologic morbidity was related to long-term outcome. Similar theme from a recent retrospective study from the experience at Texas Children's. So here they took 160 patients, the VA and VV ECMO, the PICU and the NICU. One observation you can see here, this is the frequency of the use of VV ECMO and the NICU and the PICU over about a 10-year period. As you can see, it's increasing. This increase is not associated with any difference in survival. This is 30-day survival. No difference between VV in blue and VA ECMO in green. So no difference in survival. And the mortality is pretty consistent across most of these pediatric studies, somewhere between 25% to 50%, 26% in this review. But the vascular complications, that is intracranial hemorrhage, was lower in the patients with VV ECMO. And the other neurologic complication is also lower with VV ECMO. So one might conclude from this that VV ECMO is associated with lower neurologic morbidity. That is a theme in this pediatric literature. But again, the caveat is what's defined here as a neurologic complication is a seizure or brain death. And these are outcomes measured at the time of hospital discharge. So when we're talking about neurologic morbidity in pediatric ECMO, we have to keep in mind what's the timeline of recovery. And in the majority of these studies, it's survival to hospital discharge or 30-day recovery. So the key issues, then, as we think about this literature, we have variable definitions of neurologic injury with a composite often of a vascular injury, seizures, a neurologic exam, and brain death. Very importantly, as you can see from these retrospective studies, there's a lack of a standardized approach to neuromonitoring. The retrospective study is looking at the experience with ECMO from 10 or 15 years ago. And we don't know the link between these acute neurologic insults and long-term functional outcome. So turning now to the use of neuromonitoring. Transcranial Dopplers are an attractive monitoring modality in children. They're non-invasive. They measure cerebral velocity and therefore flow, brain perfusion. Particularly attractive in pediatric neurocritical care, where the availability or indications for use of intracranial pressure monitoring are really limited to traumatic brain injury. This was a survey we did from the Pediatric Neurocritical Care Research Group. We asked 43 centers in the PNCRG, how many are using TCDs? 27 groups said they were. And of those, 8 said they were using them as part of their standard neurologic surveillance in children on ECMO. This is still an emerging field. What's not known is what each group uses as an indication for TCD, how they interpret those data, and what impact, if any, the adjustments in care have on outcome. A couple of centers have published their experience. This is from Nicole O'Brien, who led a study. This was a prospective study from nationwide Boston Children's and Phoenix Children's using daily transcranial Dopplers in children on ECMO. So this is systolic velocity, mean flow velocity, diastolic, and pulsatility index. So they had 55 children in their series on VA ECMO, and the other imaging was just clinically indicated. The first thing they saw was that the velocities and the pulsatility index were lower on each day compared to published comparison controls. Now for controls here, they used not age-dependent and gender-matched controls, of which there are some published data. But these are children, again, from the O'Brien group, critically ill children on a ventilator. So even compared to other critically ill ventilator children, these velocities and pulsatility index were low. Amongst this cohort of 55, 8 had an acute neurologic injury. That was defined as a vascular injury or brain death. And amongst this 8, 6 were neonates and all had global ischemia. So when you look at this slide, what they're showing here are the infants who had it. This is the time course of systolic velocity, mean velocities, and diastolic velocities in the children, really neonates who had an acute neurologic injury. And the observation from this is the only statistical difference was that those infants who had an acute neurologic injury had a higher pulsatility index. Again, no association with outcome. We did another prospective study. Jay Rillinger, when he was a fellow with us, we looked at 27 infants and children on VA ECMO. And Jay did, every other day, transcranial Doppler's. And this is the mean velocities, systolic and diastolic velocities, before here and after ECMO. And so the lesson from this was first that, again, on ECMO, velocities are lower compared to, this is the black line here, is a comparison to the published values for other critically ill children of the same age. Velocities are low on ECMO and they increase after ECMO. In this group, 7, there were 15 survivors. 7 of those who had acute neurologic injuries had brain imaging based on an abnormal neurologic exam. And we found no association between an abnormal TCD velocity and the presence of a vascular injury on imaging. And we looked at 6-month outcome and found no association between abnormal TCD velocities during ECMO and 6-month outcome either. But one thing that did strike us was, here are those 7 patients with an acute neurologic injury that vary in age. In every case, the TCD velocity was increased. Not enough that one would make a clinical decision based on that, but between the rise in the pulsatility index data from O'Brien, this observation, there's a suggestion here that the increase in TCD velocities may be a marker for some kind of vascular injury. Certainly nothing that's achieved a level at which one would make clinical decisions based on it. So the issues with the TCD data are we don't really know what's a pathological velocity in children on ECMO. We don't have a clear relationship between an abnormal TCD velocity and a vascular injury, nor do we have a link between short and long-term neurodevelopmental outcome. And this is a work in progress. The standards for the use of TCDs in this field are still a work in progress. As for NEERS, there's a couple of interesting papers. This is from on the top here from Texas Children's and on the bottom from University College London. This is an attempt, or this is a series of about 25 children where they took regional cerebral oxygenation and a maximal tissue coherence with mean arterial pressure. The University College group used 12-channel NEERS measuring oxygenated hemoglobin and correlated that with mean arterial pressure. Nothing that's risen to the standard of a clinical use or other detection neurologic injury yet. Using the possibility of serum biomarkers as a means for non-invasive detection of brain injury is work that's being led by Dr. Bembeya, who's speaking later here. So this is work from Johns Hopkins. And here they used daily measurements of a number of biomarkers for both astrocyte injury, glial fibrillary acidic protein, as well as neuronal injury, neuron-specific enolase, astrocyte injury again, S100B, intracellular adhesion molecule, BDNF, and monocyte chemotractive protein. And as you find, the primary outcome was a pediatric cerebral performance score at the time of discharge. For those of you not pediatricians, a PCPC of one to two is either a normal outcome or mild disability. Beyond that is moderate disability through coma and death. So what's shown here is that in each case, each of these biomarkers was elevated in those with an unfavorable outcome, that is, spectrum of moderate disability to death. This is a summary of this cohort. This was about 90% was VA ECMO in these 80 patients. So these are the patients with a favorable outcome. One other point to make from this is even with those patients with a favorable outcome, a number of them had a stroke, cerebral hemorrhage, or had abnormal EEGs. So again, this is a relatively short-term outcome study, and even these patients who can look relatively well at the time of discharge or within 30 days after discharge still may have some kind of vascular injury. In pediatrics, short-term outcome is not necessarily predictive of long-term outcome. What about seizures? So I made the point earlier that the data from the ELSO registry about seizures as a burden was based on clinical seizures, not electrographic. Juan Piantino, when he was a fellow with us in Chicago, looked at our experience with continuous EEG monitoring in children on ECMO. This was 19 patients, mostly, again, VA ECMO. They'd had continuous monitoring for clinical suspicions for some kind of neurologic insult, so already a biased example. What we found was electrographic seizures in 21% of this cohort, the majority of which occurred within about the first six hours of monitoring, that is, within the first day of being placed on EEG. There was no association between seizures and outcome, but what was striking to us was that, again, looking at the clinical course of these patients, in every case of the patients who had electrographic seizures, all of them had some kind of vascular injury as well. Some of the patients with vascular injury did not have seizures, but again, a theme emerging that seizures may be a surrogate for some other ischemic or hemorrhagic insult on ECMO. The CHOP group extended this concern and evaluation of the potential for continuous EEG monitoring to detect acute neurologic injury in children on ECMO and performed a prospective study. They had about 112 eligible subjects, of whom they consented 99. So this is continuous EEG monitoring prospectively performed in all consented subjects on ECMO over about a four-year period. Consistent with our data, they had electrographic seizures in 18%, and in most of those patients, those seizures were electrographic only. Without continuous monitoring, they would not have been detected. The main key risk factor for this was low cardiac output syndrome. Again, another theme. They asked whether seizures were associated with outcome. Indeed, they were associated with poorer scores on a pediatric cerebral deformis category at hospital discharge. Again, in this case, also an abnormal background, so a discontinuous or a burst suppression EEG, not surprisingly, was associated with worse outcome. So one of the struggles with this is that you have these promising technologies like EEG, NIRS, which are available to varying extent that they often looked at in isolation from cardiorespiratory monitoring. And in the case of EEG, it's subject to a relatively subjective interpretation by an epileptologist, which I'm not. So one of the things we've been trying to do in Seattle is to try to combine signals from cardiorespiratory EEG and NIRS monitoring so that we can have something other than just a detection for seizures and that we can look at the patterns that are emerging from the signals from brain and cardiorespiratory monitoring. So one of the other challenges is what's abnormal? Not only do we have a challenge of does something that appears abnormal predict long-term outcome? Is that tractable? Is it treatable? Does it make a difference? And what actually is an abnormal EEG or abnormal NIRS, abnormal blood flow velocity? Usually the comparison is not to healthy children or other critically ill children in general. It's to other critically ill children on ECMO with a comparable insult. So this is some of the work we've been doing with the post-op heart population. So these are five neonates with hyperplastic left heart. One of them, the patient in red, had an aortic tear intraoperatively and came back from the OR on ECMO. The others are off bypass. All the patients at our institution, as in many others now, undergo continuous EEG monitoring as a standard on ECMO or for two days after congenital heart surgery. So this is a rate pressure product, shock index. The patient in red is the one to attend to really. Yellow is when this patient after the aortic tear on ECMO had a bradycardic arrest. So shock index, rate pressure product. You can see a change in the alpha-delta ratio, the normalized alpha-delta ratio for this patient, for left and right. This is heart rate variability by frequency analysis, the ratio of low frequency to high frequency. And heart rate frequency here by heart rate variability by frequency. This is a normal patient. And this is cerebral autoregulation using HVX, the autocorrelation of NEARs and mean arterial pressure. Most positive values are less autoregulation. I'm not showing you anything here to make a claim that we can detect. The key areas we're interested in here for this kind of patient is what's happening here in the run-up to a cardiac arrest. This is not to show you we have a lesson from this. This is technically possible, although quite interesting. But this, I think, is one of the futures of trying to have more effective non-invasive neuromonitoring for detection of CNS complications in these patients. Another way of displaying the same data. These are scatter plots of the same points. These are the individual data points. This is in red here is the patient with the arrest. This is the alpha delta ratio for that patient. Just looking at these contour plots. So not to say we have worked out what the actual predictive measures are. But this is something I think has some power to be valuable in the long term. I made the point about the limitations of studies looking at long-term function. So there are a couple of important studies here. This is a population-based study for neonates from the Netherlands. Two centers where all ECMO is performed. All VA ECMO in this particular series. And this is a five-year outcome from children over a nine-year period. The majority of these children had an IQ of around 50, so mostly in the normal range. But many did not. Many had deficits in three different motor or cognitive areas. And 13% of those were severe. This is a retrospective series from Baylor looking at five-year outcome in VA ECMO. Again about one-third in mortality. These are both PICU and NICU combined. About a nine-year period, again, a five-year follow-up. And one of the key things from this paper is overall, this is a relatively good outcome. When they separated out the patients with an intracranial injury, ischemia or hemorrhage, those were the – and this is intracranial hemorrhaging at the time of ECMO. Those were the patients at greatest risk for five-year neurologic compromise. And last, a recent paper from Dr. Bembeya and the Hopkins group, they looked at – this is actually two centers, Hopkins and Children's National. They looked at 90 patients at both their centers over a five-year period of ECMO, outcome at up to 12 months, and 20 percent of them already had a neurologic deficit. And one of the important findings from this paper is they correlated the neurologic exam at the time of discharge. This case, this is the PCPC. Again, one and two is a normal or mild disability. They correlated that with a Vineland or a pediatric stroke outcome measure at 12-month recovery. And you can see the association here. So this is the suggestion that an abnormal standardized exam at the time of hospital discharge is a predictor of potential for neurologic compromise at a 12-month outcome. So now we're getting to the point where we have evidence that these – a composite of acute neurologic injury measured at the time of discharge from the hospital is stratifying patients into those who have a risk for long-term neurodevelopmental outcome. So I made the point here that neurologic injury may be associated with poor long-term outcome. That can be variously from these studies, a vascular injury, ischemia, or hemorrhage, or seizures. Seizures in a couple of studies seem – there's a suggestion now that they may be a therapeutic target. That is to say seizures are an indicator, at least in some cases, of a vascular injury and may be providing a secondary neurologic insult, not surprisingly, on top of a vascular injury. So treating the patients at whom you see seizures for whom really – which appear to occur in about 20% of this population and for which you need continuous EEG monitoring, patients at risk for a long – again, poor long-term developmental outcome. So what does that mean in practical terms? Well, these patients really should have continuous EEG monitoring. Continuous EEG is not sufficient, neither is the detection of clinical seizures. In those patients that do have seizures, you need to have a high index of suspicion for a vascular injury. All these patients, once they're off ECMO and can get a scan, should have at least an MRI because that will help you, together with an abnormal neurologic exam, the presence of seizures, the presence of vascular injury. Those are the patients at higher risk for poor neurodevelopmental outcome. And importantly, just as a contrast of VA and VV ECMO, this is not absolute. All these children are at risk for long-term neurodevelopmental compromise, and all of them need long-term neurodevelopmental follow-up, which really needs to be done in an organized fashion. Thank you very much. Thanks, Mark. Hopefully, you're getting a sense that monitoring is very important in this patient population. So let's talk about neuromonitoring. So we have it available. It's not always at all of our institutions. Christos Lazaradis from the University of Chicago, who spent a great deal of time with transcranial Doppler and other types of neuromonitoring, is going to talk to us a little bit about, is neuromonitoring actually helpful in these patients during their time on ECMO? Thank you, George. It's a privilege to be here. I have to admit that when George asked me to talk about neuromonitoring, I really liked the idea. I started reading. But then the title he gave me had a question mark if it's useful, and then I realized that I would have to change the title because I have no idea if it's useful or not. So I decided not to answer the question directly, but I'm hoping to just offer maybe some thoughts and remarks about how to think about ECMO. So we've heard several points about mechanisms of injury. This is just a slide from a pediatric review paper. However, several of these mechanisms are relevant. So the point, I think, of neuromonitoring is to try to understand flow in quantitative terms and potentially adequacy of flow. And we also want to look at impairment of pressure flow to regulation. So what are the tools that are available? And of course, all these are going to be non-invasive. I think it's an interesting idea to think about putting invasive probes in these patients, but probably not a very practical approach. So my remarks are going to be focused around monitors that can look at cerebral blood flow, so transcranial Doppler and near-infrared spectroscopy. In terms of imaging, of course, has a very important role. However, per se, it's not a monitoring tool. EEG, we already heard several comments about EEG. As Tom said, if you don't do it, you're not going to potentially find seizures, since these are going to be mostly non-clinical seizures in these sedated, potentially paralyzed patients. Then biomarkers, again, not exactly a monitoring tool. And then finally, in terms of pupillometry, there are two recent papers, one in chest and one in intensive care medicine, suggesting that an NPI less than 3 in patients within 48 hours of initiation of VA ECMO has a very high predictive value for poor neurologic outcome and death. So potentially, in terms of prognostication, that's a useful tool as well. So let me focus and make a few comments on TCD and NEARS for today. So this is flow patterns from two patients. I'm sorry. I know it's kind of maybe hard to see. So let's say this is an exam. You take your boards. You're given these two patients. And the question is, who is the VV ECMO and who is the VA ECMO? So you see the pattern here is actually a fairly normal TCD waveform. If you look at MCA mean flow velocities, they are within range, between 40 and 80 centimeters per second. And PIs are anywhere from 0.56 to 1, which is normal waveforms. This is the VV ECMO patient with preserved myocardial contractility. And the TCD waveform, usually in the uncomplicated VV ECMO, follows a normal pattern. This patient here, you notice that the waveform is a lot more attenuated, much less pulsatile. And this is a patient on VA ECMO. If you look at the numbers, mean flow velocities tend to be lower. And the pulsatility in this patient is actually within normal range, but in the low border of that. This is an example of one. These are, by the way, all patients that we monitored in Baylor a few years ago during a surveillance study. We're trying to understand flow patterns in these patients. So we're getting serial TCDs. And this is a patient that you see almost there's no pulsatility. Something to be expected, right? Because in VA ECMO, flow is mostly constant. And so you don't see pulsatility. So you notice that the flow velocities are quite low. And pulsatility indices are extremely low. So this is almost constant flow in a patient who has no native myocardial contractility. A point to make is actually that pulsatility tends to be important. There is a reason why we have pulsatile flow. It turns out, and this is literature from cardiopulmonary bypass, lack of pulsatility is associated with loss of endothelial integrity. Apparently, this cyclic stretching of the endothelium is important for nitrous oxide production. And so there is literature suggesting that lack of pulsatility associates with gastrointestinal hemorrhage and also cerebral microblades, which is a mechanism of injury in these patients. This is another paper from the Columbia Group in New York, about 30 to 40 VA ECMO patients. I found it interesting because, again, it discusses this issue of flow patterns in these patients. They compared people who had two different types of peripheral cannulation, femoral and axillary. Apparently, they do use axillary cannulation in patients who are going to be on VA ECMO for a prolonged period of time. And I know this is, again, hard to read. This is femoral cannulation. This is axillary. And I just want to point out a couple of things. One is, if you look at the flow, at the bump for patients on the femoral cannulation, flows are less, significantly less. And the authors discuss that they do that almost intentionally to decrease afterload in these patients. In the same time, in arterial pressure is the same in both groups. So what happens with TCD flow patterns in PIs? It turns out, this is just right and left MCA, I just want to point out that with the axillary cannulation that has a higher flow, pump flow, mean flow velocity is higher and the pulsatility index is lower. Potentially, showing you this link of higher flow on the pump potentially leads to higher cerebral blood flow and maybe a circulation that has lesser brobascular resistance. Again, this is an assumption made based on what actually is measured by the flow velocity in the PI. So this is the mathematical formula that takes into account all the physiologic determinants of PI. And I'm putting it here because I think it's kind of important to, you know, we make these assumptions of high PI potentially correlates to high distal cerebral vascular resistance or high ICP. However, there are several other parameters here that may make this assumption not to hold. Notice that this A1 is the first harmonic of the arterial blood pressure waveform. And so in VA ECMO, since this is very blunted, this is going to be very low. And that's why PI in these patients is going to be low as well in general. The other thing to note is that if you have a very sudden rise in pulsatility index, somebody has to think that, you know, there is an event causing high distal cerebral vascular resistance or potentially an ICP issue in these patients. For example, you know, a hematoma, a space-occupying lesion has developed, or there is a large area of the brain that is ischemic. OK. So a couple of comments about near-infrared spectroscopy. This is easy to use, non-invasive. Two optodes basically are placed in the forehead, right and left. And what we are sampling is about 2 and 1 half centimeters from the skull surface. When you try to understand what is actually measured, the advertisement is that you're getting a mixed venous blood of the brain. I don't know if that's really true. It's a very catchy way to think about it. But certainly, you're getting an admixture of venous and arterial blood 2 centimeters from the skull. When you actually look at the physiological determinants and the mathematical formula for tissue oxygen index or RSO in some of the other devices, by the way, I have no disclosures with any devices, unfortunately. I just want to make sure that I have said that in the beginning. So RSO or residual saturation of oxygen or tissue oxygenation index, again, we want to think about this as a surrogate cerebral blood flow. And it may be if all these other parameters are stable. OK. So one has to think about saturation of hemoglobin, cerebral blood volume in veins and arteries at the tissue place where it is sampled, and cerebral metabolic rate of oxygen, which, by the way, the only way to know if flow is adequate and oxygen delivery is adequate is to know cerebral metabolic rate of oxygen and demand, which we cannot get with any of these devices. OK. So just a couple of studies. This is a group from University of Maryland just using NEERS. And again, there's a lot of experience with NEERS, particularly in cardiopulmonary bypass and the cardiothoracic ICU. And there is more and more literature now emerging in patients in ECMO and VA ECMO, again, because it's very, very easy to use non-invasive. So these authors basically found that if you have cerebral desaturations, and those are defined in different ways in the literature, less than 60% of an absolute number. Some authors have used the rapid drops from a baseline more than 25%. Other people look at difference between right and left. So this group basically found that the high burden of cerebral desaturation correlates with neurologic outcomes. This is another Italian group, but this is one of the larger studies that I could find, probably over 100 patients on VA ECMO. And again, they showed that a low RSO2 or a sudden drop or difference between the two sides correlate in neurologic injury. You can appreciate this Katskan with this large left hemispheric stroke. Actually, this patient looks like they have bilateral occipital strokes. And this was signaled by a large difference between right and left sides in RSO2. OK, so in the last two minutes, I just want to make a couple of points about autoregulation of flow. I think that that's a natural feature of the physiology to care about. So what we would like at the bedside is to know, number one, does this patient have intact mechanisms of blood flow autoregulation? And if they do, is there a plateau? Can we actually find the plateau of this Lassen curve and titrate our blood pressure and flow goals to stay within this plateau? And again, the idea is that when you are in a plateau, you are actually in a... This is a neuroprotective place to be. OK, so one way to do this at the bedside, and I'm only putting this paper here because of the nice demonstration, but this is in a patient with traumatic brain injury and uses ICP. But apart from that, the principle is exactly the same if you're using TCD flow velocity or RSO2 from near infrared. So what you want is basically the arterial blood pressure waveform, and you need some surrogate of cerebral blood flow, again, with the physiological limitations we just mentioned. So you could use mean flow velocity here as a surrogate, or you can use a reginal saturation of oxygen. And what you're looking at is cross-correlation between the two waveforms. And the theory goes that a normally autoregulating vascular system should not... The two waveforms should not cross-correlate. So if you take an index of... A Pearson index of correlation, it should be zero or negative, suggesting that, in fact, the circulation is properly autoregulating. Vascular reactivity is intact. On the other hand, if you are getting a positive index or a one-to-one, if you like, at the extreme correlation between these two variables, then that suggests that there is no... The arteriolar system in the brain is passive. The reason why this is interesting is because it's one of the very few ways to get continuous real-time autoregulation at the bedside, to actually get an autoregulation loss in care in real time. One way to do it is to use specialized software to basically get 10-second means of arterial blood pressure in whatever variable you're using as a surrogate of CBF. And then you're getting a five-minute window of these 10-second means, and that generates an index that gets constantly updated. It's a noisy signal, so you need several hours of monitoring to actually understand what the autoregulation curve of the individual patient looks like. This is work from the Cambridge group that has pioneered, basically, these methods. And here you see they are using... And again, the vast majority of this literature is in traumatic brain injury, but some of it now is spilling over to other conditions where autoregulation may be important. And so here you see arterial blood pressure, you see flow velocity from TCD. And again, this has some limitations. You need to have a TCD system mounted for a couple of hours continuously to get this data. And so obviously, in the ICU, in a VA ECMO patient, that will have technical challenges. You can also have NIRS monitoring, and you get continuous... This is tissue hemoglobin index, which is a surrogate of cerebral blood volume. This is TOE or RSO2, which is a surrogate of cerebral blood flow. And then the whole point of doing this is to try to find out the so-called sweet spot for your hemodynamic titration. And so for example, an optimal mean arterial pressure is defined as the arterial pressure that corresponds to the most negative index. Why? Because the more negative the index it is, it suggests that you are within this plateau range of autoregulation. And so you can do this at the bedside and dynamically kind of monitor patients. There is experience with this kind of work, particularly the group from Hopkins and Charles Hogg, doing this in patients in the OR, for example, during cardiopulmonary bypass. And there's some emerging data for ECMO patients as well. It's also a way to actually titrate mean arterial pressure in patients after cardiac arrest. And just a last slide. Again, this is from the cardiothoracic group, Hopkins and Ono, how they suggested to have actually an integrated system of both TCD and near-infrared spectroscopy in order to get indices from both monitors and try to figure out what potentially is the most appropriate hemodynamic target. I didn't talk about EEG, but there are people here who know a lot more about this than me. And then also, pupillometry may be an interesting tool to use, particularly for prognostication. Thank you very much. Thank you, Christos. Our next speaker, again, we're talking about complications, heard a little bit about monitoring. And now we're going to look at, unfortunately, one of the feared complications that we typically or sometimes we'll see in our patients, both pediatric and adult age population, is when patients unfortunately progress on to brain death. So that opens up another can of worms in terms of finding out that, indeed, they have progressed, and then how to make that diagnosis. It's not always easy in a patient hooked up to ECMO, both in the pediatric, which has its own kind of set of protocols to follow in terms of diagnosing brain death. So we have Mila Bembia from Johns Hopkins, who's going to talk to us about brain death in this patient population. Good morning. Thank you for that introduction. I have no commercial disclosures. All my funding comes from the NIH and independent investigator-sponsored research grants. So to delve right in, acute neurologic injury, as we heard, occurs in 15% to as many as 50% of ECMO cases in the form of hypoxic ischemic injury, intracranial hemorrhage, and ischemic stroke. Clinical and electrographic seizures have been described in 10% to 23% of ECMO patients. And brain death is frequently listed as a neurologic complication in ECMO patients, primarily because it's captured as such in the Extracorporeal Life Support Organization, or the ELSO registry, where it's reported as occurring in 0.2% to as many as 10% of patients, depending on age group and the primary indication for ECMO. In an ELSO registry review of almost 5,000 adults supported on VV-ECMO for respiratory failure between 1992 and 2015, that's the graph on the left, there were 426 neurologic complications reported in over 300 patients, or 7% overall. And of these 300-plus patients, 100, or 28%, were brain deaths. And brain death represented 2% of these neurologic complications overall. And then in a similar ELSO registry review of 4,500 adult ECMO patients supported on VA-ECMO, similar time frame 1993 to 2013, of whom 19% were ECPR cases, neurological complications were reported in 15% of all patients. And of those patients with neurologic complications, 52% were reported as brain deaths. And that percent is 7.9% overall. And then not displayed here, in the most recent ELSO registry review of neonatal and pediatric ECMO cases, the rates of brain death between 2009 and 2015 range from 2% in neonatal ECPR to 10% in pediatric ECPR, 0.4% to 4% in neonatal and pediatric respiratory ECMO, and 1% to 3% in neonatal and pediatric cardiac ECMO. But how were these brain death determinations done? There are no standardized practice guidelines for brain death determination on ECMO. The apnea test can be challenging. And no guidelines exist for selection of ancillary studies if an apnea test is not feasible or not safe. And the consensus guidelines, both for adult and for pediatric determination of brain death, published in 2010 and 2011, do not offer a process for patients on extracorporeal life support. At the time, there was complete absence of supporting literature to make such recommendations. Since then, and in fact, a lot of the articles cited here that I'm going to talk about were published really in the last two years, 2018 and 2019, especially for adult ECMO. That's a conversation we were having before starting the session. In pediatrics, there's a much longer history and a lot of continuity in taking care of ECMO patients. But of course, a lot of adult ECMO programs have been built just over the last 10 years. In any case, the usual prerequisites prior to brain death testing must be met, just like in a non-ECMO patient, to ensure that variables that could interfere with testing have been addressed and treated. And these prerequisites include establishing irreversible and proximate cause of coma by history examination, neuroimaging, or laboratory tests, achieving normal core temperature, normal blood pressure for age, and a clinical examination that documents coma, absence of brainstem reflexes, and apnea. For ECMO, challenges start when the apnea test is being done. This is a schema published by Bion et al. in 2019. And the algorithm proposed here is generally aligned with practices reported by most centers. The patient is pre-oxygenated, and an arterial blood gas confirms normal capnea. The pre-oxygenation is achieved by increasing the sweep gas inlet oxygen fraction to 100%. And then the sweep gas flow is decreased, typically to about 0.1 liters per minute in infants, and 0.5 to 1 liters per minute in older children, adolescents, and adults. Just a caveat, when the sweep gas flow is decreased a lot, the delivery of oxygen may also become compromised. So that's just something to keep in mind as these procedures are done. And the patient is removed from mechanical ventilation. Typically patients are connected to an inflating bag system with PEEP. But alternative forms of oxygen delivery during the apnea test include the use of TPs, passing an oxygen tube down the endotracheal tube with insufflation. But our center does not do that, because you risk inducing barotrauma. Or some have reported foregoing the addition of supplemental oxygen altogether, again, in our center. And most that I know of, that's not done, because you do risk to have a pretty rapid onset of hypoxia. And all these patients are at risk for having that anyway. An inline PCO2 monitor is typically used in order to trend PCO2. Serial blood gases still need to be obtained to confirm hypercapnia, again, just like in a non-ECMO patient of at least 60 millimeters mercury, and more than 20 millimeters mercury above the starting PCO2. And then the apnea test is terminated if there's evidence of breathing, or if the patient becomes hemodynamically unstable or hypoxic. And the largest reported series of brain death determination for adult ECMO was published also in 2018 in 25 patients who underwent apnea testing by decreasing the sweep gas flow. And hemodynamics and the heart rhythm were relatively stable during the assessment. But hypoxia with a PO2 less than 40 millimeters mercury still occurred in two, which is 8% of patients. So an alternative to that, drawing from experience in cardiopulmonary bypass, is to conduct the apnea testing by adding CO2 to the ECMO circuit. And that's displayed here very nicely in the same Bain et al. paper. This method, in principle, provides a reliable increase in CO2. And because the ECMO sweep gas flow is not altered, oxygenation should be maintained. The authors recommend adding 8% CO2 volume to the ECMO circuit, which should, again, in principle, lead to an arterial CO2 above 60 millimeters mercury pretty reliably and independent of patient size, ECMO circuit volume, or oxygenator surface area. The one caveat with this method, and something that is very different from any of us doing brain death determinations and conducting the apnea test, is that that switch, that increase in PCO2, will happen very quickly, typically as quickly as two to three minutes. All these different approaches to determination of brain death in ECMO patients are nicely summarized in a systematic review also published last year. And this review highlights some of the differences among centers in how the apnea test is conducted, both in terms of CO2 manipulation through the ECMO circuit, as well as in terms of providing oxygenation. But it also highlights the limited published data available. When you check the sample sizes here, really, most of them are case reports or case series, with the largest one being 25 patients, and the grand total is 40 patients. Something that's not mentioned in the review, but again, in 2019, the Mayo Group, no, this is an Australian group, published their findings, is the location for monitoring PCO2. And why is this important? During peripheral VA ECMO, as we heard from Dr. Black earlier today, as the myocardial function improves, deoxygenated blood flows to the proximal aorta from the LV, as native cardiac output. And this blood mixes with oxygenated blood returning from the ECMO circuit at a point in the aorta that's usually described as the mixing point. And this mixing point can be proximal or more distal. And it's definitely dynamic. Most of the times, it's difficult for us clinicians to determine exactly where it is. And it's dynamic because it's a function of the competing relative blood flows of the native circulation and the extracorporeal circulation. So what this group is proposing is that samples from both the radial artery as well as post-oxygenator should be obtained to reliably determine the cerebral CO2 gas tension. And this is regardless of where the mixing point may be. So during an apnea test, the presence or absence of respiratory effort would only be definitely evaluated when both the radial and the post-oxygenator PCO2 are above 60, providing proof of cerebral circulation PCO2 above goal. And as illustration, they provide the comparison of the radial PCO2 in red versus post-oxygenator PCO2 with the black line representing the threshold of 60 in one patient in whom both were above goal and in a different patient in whom the post-oxygenator PCO2 was lagging behind. So these were all adult data. How about children? There are only two published series of determination of brain death in pediatric ECMO patients. One is in three children, and one is in eight children. This is an algorithm for determination of brain death in three patients at Wake Forest using this protocol. And apnea testing demonstrated no respiratory effort in two cases. One child experienced severe hemodynamic instability and hypoxia, which invalidated the apnea test. And then in the second series of eight patients at Children's National, these eight patients underwent a total of 14 brain death examinations that included apnea testing while supported by VI-ECMO. And this resulted in six cases of clinical determination of brain death and two children with withdrawal of support after the first examination. During 13 of those 14 apnea tests, the ECMO circuit was modified to achieve hypercarbia while maintaining oxygenation and hemodynamic instability by decreasing the sweep flow prior to apnea testing in 10 examinations, adding CO2 during three examinations, and increasing the ECMO pump flow in response to hypotension during two examinations. So in this series, there was a sequential approach to first decreasing the sweep flow to allow for CO2 to rise passively, followed then by addition of CO2 when the process was too slow. So now we're moving to a situation where if the apnea test fails or is not feasible, ancillary testing needs to be performed, similarly to non-ECMO patients. There are a few practical considerations, including institutional availability of specific ancillary study, patient stability for transport, and test-specific advantages and disadvantages. Hemodynamic instability may make transport of ECMO patients unsafe. And with ECMO centers that are very new, there might be also some stress over transporting these patients. In experienced ECMO centers, almost 100% of ECMO transports occur without any type of mechanical complications. And then also, perturbations in cerebral blood flow during ECMO have not been well investigated and can vary depending on the ECMO cancellation strategy. For example, reduced or asymmetric concentrations of contrast in the cerebral vessels have been described during CT and geo studies on ECMO, leading to poor quality or non-diagnostic tests. We've just heard about transcranial Doppler on ECMO. And TCD on ECMO has also been used as part of brain death determination as an ancillary study. This is a figure from a study conducted in Europe that showed that TCD patterns seem to be a reliable test for cerebral circulatory arrest in VA ECMO, only when a pulsatile cardiac flow was still maintained. And what's really important for TCD use for this particular situation is that a prior determination of adequate acoustic windows is mandatory before employing TCD. A first study that's conducted at the time of brain death determination is not appropriate, so adequate trans-temporal windows have to be documented. And then in the context of determination of brain death on ECMO, TCDs have been reported as being used both as ancillary studies or as a tool to select timing of a definitive or an accepted ancillary study, with a caveat that not all countries accept TCDs as an ancillary study. And TCDs are also not recommended for children in the context of brain death determination. So now, after conducting this entire process and brain death determination, the next step is evaluation for potential organ donation. There is very little data published. Again, all of it is from the last two years. These are data from a report from the National French Registry of Donors after brain death that showed that 40% of 161 ECMO patients reported between 2007 and 2013 were organ donors. And donors with ECMO were significantly younger. They had more organ dysfunction than donors without ECMO. And they had more medical incidents that led to exclusion of a brain death subject for organ procurement. And medical incidents were seen in 38% of patients, and they were defined as multiple organ failure, hemodynamic instability, terminal cardiac arrest, laparotomy findings, and other complications that precluded organ donation. On the flip side, there were significantly fewer donors with ECMO compared to donors without ECMO, in whom the reason for organs not being retrieved was refusal. And then in terms of recipient outcomes, there was no significant differences in one-year kidney graft survival, which is depicted here, but also no significant differences in one-year liver recipient survivor comparing donors with versus without ECMO. And at least in this series, the authors concluded that donation after brain death with ongoing ECMO should be at least considered more systematically by ICU organ donor coordinators and transplant teams. And with that, I conclude. And I think we have one more talk before we move to questions. Thank you. I forgot to mention that Dr. Brimby is a pediatric intensivist out of Johns Hopkins. So thanks again for sticking around. We've got our last speaker of the session. Carla Seven is coming to us from Nashville at Vanderbilt, where she publishes extensively on outcomes after critical care illness. Carla is a adult pulmonary critical care physician. One of the articles I would just like to mention from her, I'd love the title, Post-ICU Clinics Should Be Staffed by ICU Clinicians. This was an editorial published in Critical Care Medicine just last year. So now we get to see and hear about what happens to these patients when they actually leave the hospital, obviously didn't progress to brain death, and are seen in the clinic. So hopefully developing kind of a algorithmic approach to seeing these patients and capturing their long-term outcomes. Carla? Thank you, George, for that nice introduction. And I guess I also need to disclose that I have some opinions about survival after critical illness in general and ECMO specifically. I do have some modest funding from the Society of Critical Care Medicine, which I will talk about that work today a little bit, and from the Department of Defense to study long-term outcomes after critical illness in survivors of acute lung injury. I should further disclose that I am not a neurologist and also not a pediatrician. So I'm only going to be talking about adults today. And there's one picture of a brain, which I hope no neurologist will criticize me for. But I do want to talk a little bit about the long-term impairments after ECMO, what little we know, which is pretty little, and then describe some possible clinical approaches to managing these effects in patients who've received ECMO. And at the very end, I'm going to squeeze in a plea for prioritizing some research outcomes and interventions in ECMO survivors. And just to lay the groundwork, I want to remind you all that as a field of critical care, we're getting better and better at our job. And every year we're depositing more and more survivors of critical illness into the community. And especially when we've pulled out all the stops like ECMO, we sort of feel like we've done our jobs if somebody survives to hospital discharge. But patients and families feel a little bit differently about it. Survival is not an endpoint for them. And I'm a fan of inadvertently amusing signs. This one made me think of ICU survivorship because it said, this is in Germany, there's danger of life. And if we succeed and people survive, there are, in fact, quite a few things that we need to do at the end stage of recovery from critical illness. ECMO is especially difficult to study because it is a treatment or the cure, not the disease. So many of our research priorities are disease-specific, sepsis, ARDS, for example, cardiac arrest. And we do ECMO for various reasons in various populations and using various modes, VV, VA, kids, adults, civilian, military. This is a young man who developed ARDS while he was serving in the military. And you can see that he's been prepared for transport here. And yes, he's on ECMO. And he's got, as we say in the South, a couple, three other things going on. And these underlying comorbidities may be, in fact, the cause of some of the impairments that we see after critical illness. So there's not a lot of research in long-term outcomes after ECMO, especially in adults. They do a little bit better in the pediatric population. And this is, in fact, a little bit older study from 2006. But I like it because when we talk about long-term outcomes, as a critical care group, we previously talked about long-term outcomes as 30-day survival, which is not very long. Now we're getting a little bit better. We're up to 12-month outcomes, usually starting with 12-month survival. In this study, they had an average median duration of follow-up of five years at the time of this assessment. And one patient was 12 years out from their ECMO. And you can see that if you have received VA ECMO, you're much more likely to have abnormal neuroradiologic findings here in the hatched bar, less common in VV. And same thing for neuropsychological findings. These are a total of 28 patients. If you had VA ECMO, you were more likely to have pathological findings neuropsychologically. And the black bars are normal. Not surprisingly, these radiologic findings and neuropsychological findings correlate. So if you had an abnormal brain imaging, you probably had abnormal neuropsychologic testing. And the only other thing that's really been even starting to be studied in terms of long-term outcomes is health-related quality of life. Interestingly, Liz Wilcox just published this systematic review this year, 2020, which found that there were greater decrements among ECMO survivors in health-related quality of life compared to patients who had been on conventional mechanical ventilation alone. But significantly less psychological morbidity, which is interesting, sort of leading us to question, again, what is the underlying cause of long-term cognitive impairment? If it's hypoxia and you get on ECMO quickly and you're not as hypoxic, that perhaps will predict better psychological outcomes, although we haven't seen that in cognitive impairment. Likewise, at least at Vanderbilt, one of the things that we're able to do with our ECMO patients, if it's a choice between being paralyzed and prone for ARDS, you're going to be deeply sedated for a longer period than if you're on ECMO. Often on ECMO, we can wake you up and walk you around even when your lungs are not better. And that may be a sign of better psychological recovery. A study by Roll et al. in heart, lung, and circulation in 2019, I think, well, illustrates the problem with studying ECMO outcomes. So they looked at 78 ECMO admissions, and this was over a five-year period. And 30 of those patients died. Of their 47 survivors, 14 did not participate in follow-up. And in the end, they ended up with only 12 patients with analysis at hospital discharge and at 12 months. Interestingly, this is an Australian study of the 19 patients that they had return-to-work data on. I'm sorry, 20 patients they had return-to-work data on, 19 had return-to-work, confirming my bias that Australians are somewhat hardier than the rest of us, but also perhaps pointing to some of the societal predictors of return-to-work and survivors of critical illness. So since we don't have a lot of data that's ECMO-specific, we do look to other diagnoses-specific studies. This is a really interesting study by Jack Awashina in JAMA in 2010, which was great because it used a long-term outcomes database. These are healthy population community living adults for whom we had pre-illness cognitive outcomes. And you can look at their prevalence of cognitive impairment three years and one year before an episode of sepsis, and then one year and three years after an episode of sepsis. And you can see that there's a huge jump on more than fourfold increased risk of moderate to severe cognitive impairment after just an episode of sepsis. So these are not patients who've been on ECMO, but perhaps analogous critical care population. And the specific deficits that we see after critical illness in general and sepsis in specific are deficits in verbal learning and memory, decreased hippocampal volume and more low-frequency EEG activity, indicating brain dysfunction. Again, these are, except for the decreased hippocampal volume, fairly normal-looking brains. We don't see infarcts or bleeds, but they still are very abnormal in their ability to perform cognitive testing. And in this seminal paper, which was done by my colleagues at Vanderbilt and others, they looked at a general critical illness survivor population, so not specifically sepsis but including sepsis, and looked at their global cognition scores at three months and 12 months. And the population mean is up here in the green line and standard deviation in the green bar. And then here in the orange line is about where mild cognitive impairment is. In the blue bar down here, this is Alzheimer's disease. In the middle, traumatic brain injury. And these are patients whose only brain insult was critical illness. They didn't have a stroke. They didn't have a bleed. Some of these patients have been on ECMO, but not all of them. And the most disturbing part of this figure, in my mind, is that being young, less than 49 years old, which is seeming younger and younger all the time, did not protect you from cognitive impairment after critical illness. So this is not a problem that's specific to a geriatric population. Teasing out what is the ECMO and what is the critical illness and what is the sepsis is difficult and is not made easier by this figure, which looks at the trajectories of cognitive decline in a global population. And the black bar is just people who were never in a hospital. And I guess none of us are getting smarter as we age. But if you got admitted to a hospital for whatever reason, this red line shows a significant acceleration in the rate of cognitive decline. So there are clearly things that are contributing to cognitive decline that we're doing to people in the hospital, not just the underlying problem for which they were hospitalized. So how do we predict who is going to have cognitive impairment? There's certainly a number of risk factors, but delirium is the strongest independent risk factor for cognitive dysfunction. So if you think back to all your patients that you've taken care of in an ICU, the ones who were delirious are the ones who were at highest risk of having cognitive impairment later in their recovery course. And the longer they're delirious, the more impaired they're going to be. One of the strongest risk factors for developing delirium is the use of benzodiazepines. And one of the scary numbers that I learned from Joanna Stallings, who's one of our ICU pharmacists, and has given a number of talks here, is that a cumulative dose of 20 milligrams of lorazepam equates to a 100% incidence of delirium. So if you think of anyone that you've given 20 milligrams of Ativan to in your life, they're over here. So the question is, should these patients have some sort of ICU-specific follow-up? And I think George gave away my position on this. And by the way, that editorial was supposed to be a pro-con, but I don't know what happened to the con. So now it just sounds like a big rant, but still has all of my arguments for why we should be doing ICU follow-up. We certainly believe that at Vanderbilt, and we started a clinic specifically for following up survivors of critical illness in 2012. And it's a multidisciplinary team clinic. Patients are selected based on clinical characteristics. Either they've been on mechanical ventilation, they've had shock of any etiology, and or delirium. Usually they have one or more of those. And our ECMO survivors are really the poster children for these survivors of critical illness. So they have all those things plus. And they tend to have longer hospital courses, more paralysis, more sedation, more critical illness myopathy. And these are the things that we're trying to identify and target relatively shortly after hospital discharge. So this is our clinic flow. We have patients come in and do a short spirometry. About a quarter of the Tennessee population smokes, and about 50% of our ICU survivors are smokers or former smokers. So we uncover airway issues related to intubation and also underlying obstructive lung diseases. They get checked in. They do a six-minute walk test, which is a measure of their exercise capacity. And it's about usually around 50% predicted in the first month after hospital discharge. These are patients who've been discharged to home and are supposedly ambulatory. When we described our first 62 patients that we saw in the clinic, 27% could not walk to do the six-minute walk test. So that was not a good indicator that they were having strong physical recovery. And then we do, Joanna Stallings does our med reconciliation. She's our ICU pharmacist. There's a medical history and exam by either our MP or myself and a kind of debrief of the ICU stay. A neuropsych evaluation, which I'm going to show you an example of in a minute. Case management evaluation, a lot of these patients don't have an established relationship with a PCP, for example, or they don't have insurance, or they don't have a PCP that's covered by their insurance. And then we sort of discuss with the patient and the family the assessment and plan and update their med list. We send a report to their primary care provider and give some recommendations about what needs to be done from there. And this is a typical cognitive evaluation. This was actually one of the first patients, maybe the first patient that we saw post-ECMO in our clinic. And we've since started to do a lot more ECMO in the medical ICU where I primarily practice. And this was a woman who had had flu ARDS, was on ECMO and had kind of all the complications. She had a DVT at her cannulation site. And then she was put on anticoagulation. And then she had a bleed. And then she got an IVC filter. And she was basically paralyzed from critical illness myopathy and couldn't move. And so the inpatient rehab wouldn't take her because she couldn't do three hours of inpatient rehab a day because she couldn't move because she had critical illness myopathy, for which the only treatment is inpatient rehab anyway. She eventually went home without any inpatient rehab because I was told that she could walk. So she comes back to clinic about five weeks after discharge to home. She still can't walk. And this was her cognitive evaluation. Her Montreal Cognitive Assessment score was 21 out of 30, which was significantly impaired. She had problems organizing and attending and made errors during a clock, which when I first started learning about cognitive evaluations, this didn't seem that bad to me. But it is, in fact, grossly abnormal if you can't line up the 12 and the 6 and you can't put the hands at 11 and 2. She made multiple errors on a serial sevens task and had concrete thinking on a similarities test, leading Jim Jackson, who's our neuropsychologist who does these assessments in clinic, to conclude that although her job is not particularly cognitively demanding, she was a cashier in a high school cafeteria, her cognitive problems are so great that they would likely interfere with her performance. Unless you think that this is a little old lady who had preexisting cognitive impairments, this woman was 37 at the time of her evaluation. So unsurprisingly, if you can't walk and you can't think, you're probably not able to drive well and you're not back at work, which leads to profound implications for our patients. And although we're doing a pretty good job of figuring out how to do critical care better and better, these are all critical care citations in the solid bar and all randomized controlled trials in the dotted bar here. When you look at ICU survivor outcomes, there's a lot of interest, but not a lot of evidence about what we should be doing. So some of us have started doing these multidisciplinary team clinics, post-ICU clinics. This is usually at a tertiary center like Vanderbilt, but we're seeing more and more in community hospitals. Most patients who survive a critical illness will get their care from a PCP, usually a PCP that we give no guidance to about what happened in the ICU or what complications the patient might be experiencing from their critical illness. There's some emerging evidence that peer support can be helpful in the prolonged recovery that these patients experience. And we're starting to utilize technology such as telehealth a little bit more to try to reach out to this very debilitated population. We just started a telehealth study at Vanderbilt. But most patients who survive a critical illness will not get any of those things and they end up in this kind of no man's land between the hospital and the outpatient setting and they can never quite get stable to here, so they end up being readmitted quite a bit. This is a study that we published last year looking at whether we could impact readmissions by implementing this ICU recovery program, which included information about ICU recovery on the inpatient side, a 24-hour hotline that patients could call after discharge, an email that they could email for help and advice, and a clinic visit. And almost nobody called the 24-hour hotline, very few people emailed, very few people came to clinic, and yet we were able to show a decrease in readmissions, especially in this early post-hospital discharge period, suggesting that even just providing some anticipatory guidance about the prolonged ICU recovery that people are going to experience can impact outcomes. So I certainly feel that there's a lot that we need to do for patients after they've, you know, after we've put all our money and time and energy into saving their life. There are things that we need to do for the patient, but there are also, there's also a lot that we need to learn and we're only going to learn about those things if we see patients after the ICU. This is a paper that we published out of the Cairns Study, Collaborative Assessment of ICU Recovery Needs. This was a qualitative study in three countries looking at what sort of things we're getting back from seeing patients after the ICU. And one of those things was improving our care in the ICU. Certainly that can include finding out things that we need to improve in our clinical care. Of course, also benefit to the clinicians who are doing this kind of work that they perceive that they became better clinicians and had less burnout when they were able to see, you know, their success stories and sort of bring them over the finish line. Risk mitigation, which is something that we need to explore further as well as ICU culture, education and awareness. So even in the absence of a huge body of data telling us what to do for patients who survive ECMO and critical illness in general, there are a number of places that have started to see these patients and are learning by doing. You can find a list of the ones that participated and thrive at this website down here. But now that we're actually seeing these patients, we need to be a little bit better about researching these outcomes and what interventions we can do to impact them. This is from a Delphi process that Carol Hodgson just published a couple months ago looking at core outcome measures that are needed specifically in ECMO survivors. And these are closely correlated with the ones that we've sort of settled on for survivors of critical illness in general, but also including intracranial hemorrhage and major bleeding. And Liz Wilcox and Wes Ely just published another paper about studying long-term outcomes after critical illness. And they stated quite bluntly one of the reasons why people don't study this is because it's very difficult to get patients months and years after critical illness to study them and their outcomes. But we can do hard things. And I think if we can get a fixed-wing aircraft at our university and fly out and cannulate people for ECMO in the field and bring them back hundreds of miles to have a 24-hour perfusionist at the bedside, we can probably set up some sort of follow-up clinic system. And I'll end there. I'd like to thank my team. This is only part of them at the Vanderbilt ICU Recovery Team. If this is something that interests you, if you have follow-up for your ECMO survivors at your center or you think you want to have some, I would encourage you to follow us at Cairo. This is the critical and acute illness recovery organization that we just started to continue the work of the Thrive Collaboratives. And all are welcome. I'll end there. Thank you. So I'm going to invite my speakers to come up and take any and all of your questions. Hopefully you guys have a few questions. And then we'll let you guys get out of here and enjoy the cloudy day out there. Question. Go ahead. Thank you all for the excellent talks. I had a question about the systolic blood pressure goal versus using MAP in patients on mechanical support. I was in a situation where I did a brain death exam on a patient on a VAD. And the AN guidelines would require a systolic blood pressure of 100 or above to just start testing these patients. And the patient had a MAP of 80, which we all know is more than sufficient to perfuse the brain. But a lot of my colleagues were not comfortable doing the brain death exam on a patient with a systolic blood pressure less than 100 due to non-pulsatile blood flow versus just using the MAP of 80 and going ahead. One of the family members was a nurse. I discussed that with her, told her, well, these guidelines came out before ECMO was prevalent. And I think this is sufficient to perfuse the brain. And with your permission, I would like to go ahead and perform the brain death exam. And she was OK with that. I just was wondering how you would approach that and if we should have a more collective effort where this issue would be clearer to providers at the bedside. Thank you. Well, I'll start. With the exception of aortic dissection, every guideline that uses systolic pressure is bullshit. Right? There's no circumstance in which systolic pressure is more meaningful than the MAP other than tearing your aorta apart. So I would say our guidelines have to be rewritten. The same thing with the other parts of the brain death exam. It's the pH, not the PCO2, that's important. And we have to stand together and get rid of ancient nonsense. Having said that, I think a MAP of 80 has got to be plenty of perfusion. I don't know a way to prove that directly, but I would be very happy to support you on that. And a similar approach is taken in children too, where these discussions are made with the family. And by the time you get to that point, they have a pretty good understanding of the physiology because you've been at the bedside. Most of the times, these situations occur when there is no recovery, right? And they understand by the time you go and have these discussions with them. It can become very tricky in children because there are age-based limits that are also not very clear. So we oftentimes do these things on a case-by-case basis. Two questions. Any and all are happy to try to attack these. How do we separate the complications between the actual primary disease and those complications from the ECMO machine itself? And then should we be imaging these patients regardless of whether or not we find changes on neurologic examination in the ICU? So should they all be scanned in a sense serially, either at the end of ECMO and before hospital discharge and then after follow-up, so we can actually at least radiographically see what's going on with these patients and try to correlate kind of with our long-term outcome measures, whichever one you're using. I'll start on that one. So I think I'm wary of any edicts that says that we have to scan everyone before they go on. I don't know that it would actually predict anything. If I'm trying to make a decision about long-term anticoagulation, yes, I'd want to know if somebody had something in their head. But in many cases, you know, especially it sounds like in the pediatric population, having an abnormal brain scan may not correlate with your outcome very well. And likewise, you can have a very normal-looking brain, like my lady who I talked about, her cognitive evaluation, and have a very impaired cognition. One of the interesting things that came up when we started seeing these patients after the ICU, I think I mentioned the six-minute walk test, which seems like a very benign test. And we really only started doing it for intellectual curiosity to see if our patients were as weak as have been described in the literature. But we've had a couple patients come in who were really, I mean, it seems like a strong word, but they were crestfallen, devastated to learn that their six-minute walk test was 47% predicted. And, you know, we were like, well, that doesn't necessarily predict what your recovery is going to be. And I can only imagine that having an abnormal brain scan would be even more upsetting for somebody, especially if it didn't predict what their outcomes were going to be. Michelle Katani, I'm from Newmorse Children's Hospital. Dr. Blake, you mentioned that the intracranial hemorrhage with patients who had a rapid correction in their pH. I just wonder if this is a phenomenon that should be worried about in our pediatric population. We actually published an article with the single center series back in 2012, I believe, that showed similar findings of a rapid decrease in PCO2 with ECMO cannulation in neonates being associated with increased mortality and intracranial hemorrhage. So based on that, us and a lot of pediatric ECMO programs, I don't know for sure in adults, have changed their ECMO initiation strategies so that the correction of the pH and the PCO2 in patients who have profound respiratory acidosis before going on ECMO is done very slowly and gradually over the first two to three hours of ECMO. A lot of the going back 15 to 20 years, respiratory alkalosis used to be a therapy that used to be applied in neonates with PPHN. And very aggressive correction with ECMO was not necessarily considered deleterious. So this practice has definitely kind of progressed over time. Again, I'm not very familiar with the adult literature in this respect, but it's definitely something that's considered in neonates. Just one other thing to consider is that alkalosis is a risk factor for seizures. So that in addition to trying to prevent intracerebral hemorrhage from occurring, it's probably good not to drive somebody to severe alkalosis anyway. Hi, I'm Yi Chen, I'm from Texas Children's. I enjoy your talks. I'm often sort of struggling between when to monitor and, you know, what modality to use on the patient on ECMO. Like for instance, our institution are fairly aggressive trying to sort of wake them up and basically cut off sedations and trying to sort of manage them as adults. And the thought process behind that, obviously, is you get a much cleaner neurological exam. So for those patients, would you still recommend to sort of put them on EEGs, hoping that or expecting that you may be able to pick up subtle changes before they are clinically evident? And then the other question is, a lot of look, it seems like a lot of sort of change, there's a lot of changes in cerebral hemodynamics during the, you know, at the time of cannulations. And yet, most of the times, we don't really monitor them while they were trying to, you know, while they're trying to cannulate. And so the question is, would that be something that can be helpful? I'll start that. It's hard to, as a neurologist, it's hard to argue with the primacy of the neurologic exam above neuromonitoring. On the other hand, as you saw, as you know, about consistently between 15 to 20% of these patients have nonconvulsive seizures. That to some extent begs the question as to whether those seizures are contributing to secondary neurologic injury and whether that secondary neurologic injury has some kind of impact on long-term neurodevelopment. I tried to show in the data that I showed, which I think what the literature conveys is that indeed, the seizures are a secondary insult, and they likely do contribute to long-term burden. So I think the rationale for continuous EEG monitoring is strong in this population. When should it be started? Well, probably as soon as possible after cannulation. That's what all these studies have done, really. The time varies. There are practical constraints about getting access to the scalp during cannulation. I just don't see that changing. The question, though, is really, I think, in a patient in whom you have an exam, should you still obtain continuous monitoring, say, as that patient wakes up? In most of these scenarios, the EEG is applied soon after the cannulation. The patients generally are sedated at that point. You're letting them wake up. In most of these studies, I think, certainly in the data in ours, the seizures occurred within about the first four to six hours after the beginning of monitoring. So as a practical matter, if continuous monitoring begins shortly after cannulation is completed, the patient's stabilized, so a couple of hours later, and the monitoring goes on for around six to 12 hours, if you pick up seizures then, which you're most likely to do, then by that time, if the patient's amenable to being woken up and getting an exam, I would trust the exam over the EEG. And in terms of daily practice, by that point, by the time you have an exam you feel you can trust, you really have at least 12 hours of monitoring anyway. In the neonates, infants really, where the sensitivity of the neurologic exam for vascular injury is much lower, then even with an exam, I would continue to monitor, because you certainly then may miss seizures, even with a reassuring neurologic exam. I would say that somebody who has monitored or severely increased intracranial pressure should not be awakened every day. It's pretty well shown that the attempt to do this awake breathing stuff will markedly spike their intracranial pressure. Where I used to work, we would monitor all of the neonates, because we paralyzed all of them to keep the carotid cannula in place. Yeah, I think you brought up some excellent points, and I think that, you know, you started by saying that you struggle with some of these decisions, and I think part of the problem is there is no guidelines on how to do this, right? What we suggested, that was at Baylor, actually, in Houston, which is probably familiar, Large Cardiothoracic Surgery Service, we suggested a neurosurveillance protocol. Everybody who goes on ECMO gets serial PCDs, gets a CAT scan, again, when these things are feasible, you know, particularly the comatose people. For a number of reasons. One is we don't have normative values, and especially in VA, ECMO, we don't fully understand cerebral hemodynamics, right? And then you can do things like, you know, emboli monitoring and auto-regulation monitoring. And so, I think, and the other thing is that changes, rather than absolute values, may be more important. So, the earlier you establish a baseline for these patients, maybe the more you will learn. Thank you for these several interesting talks. I'm Ibrahim McDaddy from the Cleveland Clinic. My question is about brain death testing on ECMO. You know, I'm wondering, in the absence of any available guidelines, what is the adequate observation period before actually documenting an irreversible, you know, coma, especially in the presence of some available literature that talks about that, you know, ECMO circuit affects some sedatives and pharmacodynamics of some medications. The other part is about the reliability of apnea testing on ECMO. You know, there is some literature that suggests that extracorporeal CO2 removal does not always reliably affect the respiratory drive. I just want to see what you guys think of that. Thank you. In children, generally, the observation period is a minimum of 48 hours from ECMO start. And all precautions are done to ensure that a drug screen is negative, especially in all the kids, that levels of antiepileptics that the patient may have been on or other medications that can be measured are measured. But yeah, that's the minimum period of observation is 48 hours. And then, as opposed to adults where the second examination can be done after a short period of observation in children, we extend that period to a minimum of 12 hours for older children and a minimum of 24 hours for the second exam in neonates and infants. If you don't breathe with a pH under 7.2 in the absence of sedation or neuromuscular junction blockers, you don't breathe. It doesn't matter what your PCO2 is. It's not the driver. What's the pH of the CSF? Well, the pH of the CSF is about 0.1 under what the systemic pH is anyway because of the albumin and a few other factors. But the sensing for acidosis is in an area that's not affected by that. So it's a pH under 7.2 in the peripheral blood. Any other questions from the audience? Well, thanks again for hanging out for the last day. Let's thank our great group of faculty and speakers.

ECMO

Extracorporeal Membrane Oxygenation

neurologic complications

adult patients

pediatric patients

strokes

intracerebral hemorrhages

seizures

cognitive impairments

monitoring techniques

continuous EEG monitoring

transcranial Doppler

near-infrared spectroscopy

cerebral blood flow

autoregulation