and we'll be moderating today's webcast. For some housekeeping information, a recording of this webcast will be available within five to seven business days, and you're welcome to log on to mysccm.org and navigate to the My Learning tab to access recording. As mentioned a few seconds ago, there will be a Q&A that we'll have at the end of the presentation, so feel free to submit your questions throughout the presentation, basically in the question box on the control panel. There will also be several interactive polls posed by our speakers. So when you see a poll, please click on the bubble next to your choice. And just a disclaimer, the content that we are presenting here today is for educational purposes only. And so without further ado, I'd love to introduce our speakers for today. Alexis Steinberg is an Assistant Professor of Neurology, Critical Care Medicine, and Emergency Medicine at UPMC Presbyterian in Pittsburgh, Pennsylvania. Bhikata Krishna Rajaji is a Professor of Neurosurgery and Neurology at the University of Michigan in Ann Arbor, Michigan. And Patrick Chen is an Assistant Clinical Professor of Neurology at the University of California, Irvine Medical Center in Orange, California. And without further ado, I'll turn things over to our first presenter, who is Dr. Chen. All right, thank you so much, Dr. Nelson, for that introduction, and really glad to be here and really excited to talk about the topic of neuroprognostication and traumatic brain injury, which is something that I'm very passionate about, very interested in. And these are our objectives for today. We're gonna go over a quick case and then really go over kind of what's the current landscape in the world of TBI for prognostication, and then really thinking about where future directions are going in this exciting field. So I direct and lead a program at my local university with Traumatic Brain Injury Program, which is dedicated basically to serving our region for care of TBI. We have a big, big team up here on the right. And we thought that this was important to implement over two to three years ago, because TBI is a growing epidemic. The rates of TBI are only increasing with our aging populations. And this is an ongoing epidemic that's gonna continue. And the big point here is that TBI really is a continuum of disease. It doesn't just stop at the intensive care level, which is what we are all specialists in, but really extends well, well beyond that, which is one of the big points that I wanna hit. Starting with a really quick case here, this is actually a real case that we had recently in our system, who we saw in the inpatient setting. And then I had the pleasure of being able to see six months out in my clinic. So this is a bad TBI. This is a 24-year-old non-white man who had a assault injury, was hit in the head with a pipe. And this is the initial information that we have. It's GCS3, very bad exam, bilateral blown non-reactive pupils, has multifocal hemorrhages everywhere, including an epidural, and has a sizable amount of midline shift. This is what the first image looked like in the emergency room. And the question, and this is for the audience, the audience will now weigh in, is what would you think the outcome of this patient would be projected in six months based off of what you just heard? We'll give a couple seconds to let this one come in. Looks like 48% are saying answer three. Awake, minimally aware, and disabled. So about half think that he's in a state of very, very severe disability and is not communicating. So let's see what the final answer is. So this is what happened in the hospital. So my team got brought in to prognosticate this patient pretty early on as they were moving through the ICU hospitalization. He had a couple complications. He had a surgery and then re-expansion of the epidural. An MRI was eventually done. He had a diffuse axonal injury, grade three, which is being shown here, and we'll talk a little bit about that later. So not a great MRI image, but on discharge, his GCS was actually 12. So he was awake and following some commands. He ended up having to get a tracheostomy and a gastrotomy tube to end up going to rehabilitation. This was the prognostication that I had given. This is my note from Epic, which gave an idea that I did not think that there was a devastating enough clinical picture here to say withdrawal of care. And at six months, this is him. This is him in the clinic just about a month ago. Here he is giving me the thumbs up, and he had a modified Rankin IV. So he is disabled. He lives at home. His mother is taking care of his daily needs, and they have an in-house nurse that comes to visit them. He has weakness, but he's talking, and he's communicating. He has some ongoing chronic issues that we're tinkering with, including epilepsy and a cranioplasty and BP shunt, but he has ongoing PTOT speech. They decannulated his tracheostomy, and we're planning on removing the peg, and we did some quality of life type of measures on him, and he's actually scoring quite high on this. And when you ask him if he's happy or not, he says he's happy with his life, and he's asking about kind of when can he go back, questions about when can he go back to work, when can he go back to drive, really high-level functional questions. So this is a really good outcome, and that's one of the things that I wanna impress upon people is that what we see in the hospital can be very different than what we see once the TBI leaves the hospital. And this is a million-dollar question. You have an image, a clinical picture, and everybody wants to know, family, team members, et cetera, what's going to happen. So we talk about outcomes a lot here, and in the world of traumatic brain injury, we use the Glasgow Outcome Extended Scale or the Glasgow Outcome Scale, the abbreviated version, and we drew a line in the sand right here that says that a four or below is bad, and anything higher than that is good. Again, these are arbitrary values that we created based off of dependence, saying that if somebody is completely or partially dependent and can't walk, that is the marker of disability. But again, there's a lot of literature that says that this is a moving target. And thinking about what sort of prognostic tools, what do we have at our arsenal right now to help answer these questions of what is going to happen with a patient when you have the initial story? So we do have risk calculators. These have been around since the early 2000s. The two biggest models are being shown here. You can go online and plug in data accordingly. The crash model on the left, and then the impact model on the right. These were created using large data sets of about 8,000 to 10,000 patients each, and were cross-validated externally onto each other. And they're very similar. One of them utilized, was done globally and integrated the state of where the country that the validation occurred. But long story short is that if you look at it, a lot of the key features are the same. It's largely driven by the patient's age. What was their initial Glasgow Coma Scale? Do the pupils react? And to some extent, there's some extended versions which include imaging, as well as some of the initial physiologic data from the ICU. And you can input this in for any patient. If you look at what the number was for this patient that I just presented, he was a predicted 70% mortality at six months. So these are all for just six month outcomes of mortality, and a 75 to 80% for morbidity. But there's clearly limitations because this patient to some extent didn't fit the mortality bucket on that number. And that's because these calculators, and they say this in all of the publications about this, are, you know, this is population data and it's never meant to be applied to the individual patient. These are large data sets and we know that there are flaws with data sets. And then these were created in a time in which ICU care, TBI care was very, very different. And we know now that intensive care and the care of TBI patients is a dynamic process which this doesn't account for. And in general, what studies have shown in more modern data sets is that it overall has a fairly pessimistic interpretation of outcome. It tends to overhaul poor outcome, whether that's mortality or morbidity. This is work done by H.E. Hinson recently that was just published where they applied it to a recent modern data set and has AUC rocks of 0.6. And there are teams that are continuing to try to improve these models, whether that's externally validating them on different data sets or utilizing various statistical techniques to try to improve them. So these two large calculators are in some ways flawed. And so then the question then is what about just clinical gestalt? Shouldn't we as providers, clinicians, people that see TBI and sick patients all the time be able to look at a patient and tell a family yes or no, this person's gonna do great or not. And there's actually been a study on this exact scenario asking this exact question about clinical gestalt. This is a study by Izzy et al in 2013 where they basically surveyed a bunch of different providers who take care of TBI, many of them specialists in neurology or neurosurgery and gave them multiple cases. This is just one example of a 22 year old that had a really bad exam like the one I showed you. And they showed them the images and then they asked a bunch of questions about what would you do clinically and how confident do you feel about your decision-making and where do you think the outcome's going to be? And you can see here on the right, there's just a widespread of comfort with regards to the prognostication of these various cases here. And it's across the spectrum of specialty and attending level and specialty level. So it's regardless of the level of training. And one of the big things they found was that the patient cohort, that withdrawal of care was one of the biggest drivers of TBI mortality when they surveyed these people. But also the vast differences between perspective between people who are in surgical specialties versus non-surgical specialties, et cetera. And many people got the eventual outcome wrong. Many people thought that this patient, for example, this young patient was going to die when the patient in fact had a ghost of a three, which is severe disability, similar to my case. And this all speaks to the point that a lot of these studies and a lot of these calculators, et cetera, are very, very much biased by the concept of self-fulfilling prophecy, which is that if we as providers or team members think that a patient's going to do poorly, and we say that, that in itself starts a cascade of events that might lead to withdrawal of care and overall adds to this idea of nihilism, which I think a couple of my co-speakers will speak on more. So what about radiographic images? We have great images, MRIs, various forms of fancy MRIs. Shouldn't this be able to give us a sense of how a TBI patient is going to do? And the answer is, as of now, no. This is speaking to MRI specifically. This is looking at diffuse axonal injury, which is a very common finding with moderate to severe traumatic brain injury. When you image their heads, axonal injury being represented by micro hemorrhages or dark spots on this specialized scan here. And diffuse axonal injury can be rated on a scale of one to three, three being very bad, it's in the brainstem, and one being in the cortex. For a long time, people thought that if patients had a grade three brainstem damage, that they would have very, very poor abilities to wake up because it affects the brainstem, and so they would not have an ability to awake or arouse. This has since been disproven through large, large studies. Many, many meta-analyses have been done to show that multiple cohorts of patients can have very, very good outcomes, even as short as one month out from discharge, despite these radiographic findings. So there's a disconnect between how the image looks and how a patient does, and this is just another example of multiple reports being about this sort of rapid turnaround, even in these worst of the worst images. Direct hemorrhage is a very, very bad finding, historically, to see. You can see this on a head CT, you can see here the blood is in the brainstem, in the sort of midbrain pontine area, and it's thought to be caused by a very severe trans-tentorial herniation downward. And there have been at least 10 different published reports of patients that have this finding and actually end up awakening and having, you know, ghosts at six months of about four or better. So again, even this, what was thought to be because of death is actually has, in select cases, can have a good outcome. Our team has done similar thematic research on this, as well, looking at DAI-3, finding, you know, that DAI-3 patients, that there's no correlation with GCS discharge, as well as mental status findings on discharge, with non-reactive pupils. Similarly, 25% of those that we looked at were not comatose on discharge. And then we're now looking at methods of early quantitative EEG to see if there's any other markers that could predict outcome, in addition to the clinical features that we talked about. So getting to where's TBI prognostication going then? So the idea then is, can we detect covert consciousness? Maybe the exam doesn't represent actually what's going on in the brain. This is ongoing, exciting work being done by Edlow and group. This was a really landmark paper done in 2017, where they basically did fMRI images on patients that seemed to be either in a coma or a vegetative state. And they stimulated them with music and commands and tried to see if they could see any radiographic changes on fMRI. And they call that cognitive motor dissociation, if they saw changes in response to their stimulus. And a decent amount of patients actually had this, even though they looked comatose on exam. And all the patients that showed these findings were not confused and had a decent mental status exam at six months. Similar themes here. This was very popularized in the New York Times in 2019, where Clauston et al did a similar study with quantitative EEG, looking for reactivity of quantitative EEG to stimulus. And they found a very similar finding, too, as well. Thinking long, long-term, we stretched out outcomes well, well beyond that six-month, one-month mark. Patients do well. And this has been shown in the rehabilitation data, you know, where 50% of severe TBIs are to some extent independent at one year. Similar findings, even if you stretch that out well beyond one year. 10 years, recovery still occurs. And this is a cohort that was, you know, not following commands at discharge. And about 50% of them had full recovery at one year. So really, really reassuring news. And we know that there's a heterogeneous recovery track for TBI when you think about that 10-year mark. Some of them do really great from the beginning. Some of them do very poor from the beginning. And some of them have fluctuations, whether that's they improve over long periods of time, or maybe get a little bit worse over long periods of time. So we shouldn't prognosticate too early is the big punchline here based off of that literature. And that's where the field is really pushing for. And the current state of not only just trying to find covert levels of consciousness with traumatic brain injuries ongoing, but also asking, are there different sort of markers that we can do, whether that's connectome, whether that's these advanced methods, to try to figure out who would or would not be responsive to pharmacotherapy, et cetera, to help accelerate the process of awakening and improvement to push forward their positive prognostication. And this is still in the pipeline from research, which is very exciting. But until then, this is what we have right now in the world of neuroprognostication for TBI. This is the most recent guidelines that were published in the Neurocritical Care Society very, very recently. And this is the state, this is, they basically integrated everything that I have said, and said, this is the state of the literature that we have here. And as you see, they did 32 peacock questions. Overall, this is, they had to say that this was, everything is weak recommendation, low quality of evidence. And really most factors, clinical factors, imaging factors, et cetera, in isolation are not reliable. And the ones that are moderately reliable are bilateral pupils non-reactive initially, as well as maybe some of the impact crash scores that I showed you initially. But again, they say that none of this should be done in isolation. You have to do this in a multi-modal way. And you can't just hinge it on one variable or another. So this is how I do it at a practical level, taking this all in when I talk to patients and families, and I'm looking at cases, is really looking at the big picture. What is it that brought them in? What were they like before? Is there any pre-morbid stuff here that makes the patient severely ill and or ongoing ICU illness that's going to drive a mortality? What does the objective image and exam look like? Like, where do I see the damage? And where do I think that damage would lead to a functional deficit? And then do I see anything catastrophic here, whether that there was a huge herniation event and the stigmata of that on imaging, or is there a loss of pupil that fits an image or exam, which would drive me towards a more poor outcome. But overall, I really try to hold back on giving poor prognosis based off of all the literature that I had shown you. And it really comes down to family and really asking family to what degree are they okay with a at least short period of time with disability, which often these patients will have to go through before getting to that rehabilitation state that I showed you. So these are the big points that I want to make. We're not good at prognostication. We're getting a little bit better, but historically we've been overly pessimistic with traumatic brain injury. We're looking into covert forms of detecting consciousness or covert consciousness detection, and TBI patients can do really great long-term. And I've shown you multiple cases of that. And then the final, final point that I want to make here is that in the world of TBI, we're also very, very interested not just in that one year mark, but asking that question of what happens way, way beyond? Because a lot of patients are living now because we're so good at taking care of TBI patients in the ICU, et cetera. And traumatic brain injury is a chronic disease. This is what we as a field are coming to the conclusion of. And we've done work on this most recently, looking at the sort of risks of TBI patients, people that have TBI and compare them to controls and showing that they have long-term risk of chronic diseases, things like hypertension, hyperlipidemia, obesity, stroke, and of course dementia, which is what everybody's very, very interested in. But there is an increased risk of chronic disease in the traumatic brain injury patients that live well, well beyond that one year mark. And this is something important that we're looking into because it probably to some extent is driven to some extent by other factors other than the biologic event of getting hit in the head. For example, we've shown that there's socioeconomic factors and long, that can drive long-term TBI. The area that you're from can drive some of the disability that we see long-term at one year. And so this is all very exciting work that's being developed by our field. And so the final conclusions are that there's no perfect prognostication model. Many moderate severe TBI patients like the ones I showed you can do really well. The current state is that we have these impact and crash models, but know that they were population data and that it is limited to six month predictions. And you really have to talk to the families and get through the nuances of how much disability a family would be okay with, with their loved one. And TBI is a chronic disease, and we're really excited to see what's going to happen in the future with this theme. All right, thank you so much. And I am happy to now hand it over now to start talking about cardiac arrest and ECMO. All right, great. Thank you so much, Patrick. So also thanks to Sarah and SCCM for organizing this webinar. My topic today is going to be an update in neuroprognostication after cardiac arrest. So unfortunately, this is something that is all too common for intensivists to deal with. So this is a patient that we consulted on not that long ago. This is a 55-year-old man who suffered an out-of-hospital cardiac arrest. His initial rhythm was pulseless electrical activity, PEA. He achieved a return of spontaneous circulation, or ROSC, after about 40 minutes, so significant period of time. The emergency department initiates therapeutic hypothermia to a goal of 33 Celsius. He is examined in the ED about an hour after ROSC, and the nurse uses a pen light and reports that the bilateral pupils are about three to four millimeters and non-reactive. He gets a CT from the emergency department that's reported as normal. And then over the next several hours, there are reports of irregular muscular jerking of individual limbs. With three hours after ROSC, he is connected to continuous EEG, which demonstrates a severely suppressed pattern. So question number one. His family's at the bedside. They are tearful. They want to know the prognosis. His wife says he would not want to be kept alive on tubes and machines. So what is the most appropriate response at this point? Number one, the prognosis for neurological recovery is four. Number two, a brain-death evaluation should be initiated. Number three, we should wait at least 72 hours from ROSC to prognosticate. And number four, we should wait at least 72 hours from re-warming to normothermia to prognosticate. Okay, excellent. Yep, and that's exactly right. So we should wait 72 hours from re-warming to normothermia, and I will come to why in just a minute. So let's just move on. So the entire content of this lecture is gonna be based off the guidelines for neuroprognostication and comatose of all survivors of cardiac arrest from the Neurocritical Care Society and the German Society of Neurointensive Care Medicine, which was published in 2023. Sort of the landmark publication, the landmark guidelines for neuroprognostication were in 2006 from the American Academy of Neurology. There was a consensus statement that came out in 2014 from the European Society of Intensive Care Medicine, but a whole lot has changed in the last decade or decade and a half. And so this update was actually really pretty urgently required. I strongly urge all of you to access this open access document. There's a QR code on the slide. You can use the link. But there's only a small part of the document that I can actually cover in this presentation. So when we talk about good or bad, just like Patrick was just talking about, what do we mean? What is good and what is bad after you have a cardiac arrest? First and most important thing to note is that the idea of what is good or bad really comes down to the individual. And so we really make an effort, speaking with families to determine that. And you will be very surprised as to what individuals consider to be good or bad. Particularly if they're already disabled, you'd be surprised at how many folks actually think that they're doing okay or that their quality of life is good. However, in the literature, there's clear cutoffs that have been used for what's good or what's bad in the vast majority of studies. So after cardiac arrest, we typically use the cerebral performance category of the CPC scale. And with the CPC scale, what is most often considered a good outcome after cardiac arrest is that they are independent and basic activities of daily living but able to work in a sheltered environment. So that or anything better is good. Whereas severe disability, a persistent vegetative state or an unresponsive wakeful state as we call it and death are bad. And I can't stress enough that outcomes should really be assessed at least three months after discharge. Outcomes that are assessed at discharge are really not as valid because there's multiple reports that patients can and do improve in the months after discharge. So the first and most important recommendation from the guidelines is that the assessment of neurological prognosis be deferred for at least 72 hours after ROSC in patients who are not treated with therapeutic hypothermia and at least 72 hours following re-warming in patients who are treated with hypothermia. So why wait at all? Really the basic reason is that if you wait 72 hours, then the majority of patients who are going to recover to a good outcome will in fact have awoken and followed commands during this period. Thereby rendering neuroprognostication unnecessary. That's the primary reason. But it's really important to note that if your patient is comatose beyond the 72 hours, that does not necessarily mean that the prognosis is poor. Even after cardiac arrest, there are multiple, multiple reports of patients who wake up and go on to have a good outcome, you know, one week, two weeks, four weeks, even after cardiac arrest. Why wait longer in patients who are cooled? There's multiple studies that show that if you use therapeutic hypothermia, patients will wake up later than if you don't use therapeutic hypothermia. And we think that it's got something to do with the impact of hypothermia on the sedatives that we use. The guidelines recommend against the use of baseline demographic and cardiac arrest variables as reliable predictors of outcome after cardiac arrest. So older age, a non-shockable rhythm. So PEA or asystole, for example, a longer time to return to spontaneous circulation, which is what we saw in this patient. These are all independent predictors of poor outcome, but they're not reliable because a large minority of these patients will, in fact, go on to have a good outcome. So the false positive rate with each of these variables is as high as 30% to 40%. What is the false positive rate? This is sort of an important concept in the predictors that you use after cardiac arrest and really any sort of brain injury condition. The formula, and I'll let you read it off the slide, but it's really an assessment of how often the predictor goes wrong. In these guidelines, a reliable predictor was one that had a false positive rate less than 3%, whereas a moderately reliable predictor was one that had a false positive rate of less than 5%. So let's come back to our patient. He is now 72 hours after re-warming to normal thermia. He's been off propofol for 6 hours. You repeat a physical exam. This time you use a pupillometer and not a pen light, and both pupils are reactive. He has an extensor response to pain. A repeat CT, 72 hours from RASC, is again normal. You get somatosensory evoked potentials, 72 hours from RASC, and this demonstrates the presence of cortical responses on both sides. His EEG at this point demonstrates a continuous background with reactivity to external stimuli. So based on all the data you obtained, your assessment of the prognosis now is, number one, his prognosis is poor, number two, his prognosis is actually promising for recovery to functional independence, and number three, his prognosis is indeterminate. Oh, there we go. Yeah, so 65% indeterminate, 31% promising. So the correct answer, unfortunately, is that the prognosis is indeterminate. Let's just talk about these various predictors. So let's start with the physical exam. This is the oldest and most important of the predictors of outcome that we have used, and this is, you know, really where we've seen the most change from the original 2006 practice parameter. The motor response, which is an important component of the Glasgow Coma Scale, when the motor response is absent or extensor, this was once thought to be a reliable predictor of poor outcome and really served as the basis for neuroprognostication in a large number of patients. Now, since then, there's been a large, you know, volume of data that's shown that the false positive rate for this finding is as high as 30% in the TTM clinical trial, for example, in patients who were treated to hypothermia. The false positive rate was about 20%, so it really is not considered a reliable predictor anymore. The corneal reflex has a lower false positive rate than the motor response, but it's still a little bit too high for it to be considered either a reliable or a moderately reliable predictor. Early myoclonus, again, previously considered a reliable predictor, and this patient suddenly demonstrated what might have been early myoclonus, but there have been multiple reports of good outcome despite the presence of early myoclonus. Status myoclonus is a distinct form of early myoclonus. This is really unremitting, generalized, whole-body myoclonus. Some studies will specify a time frame of 30 minutes. Now, this might actually be a predictor of poor outcome. We just don't have enough high-quality data to support the idea that it is. There's also studies that suggest that even neurologists are not fantastic at distinguishing the various sort of key features of myoclonus, including whether it's localized or generalized. So what is reliable in the physical exam? It's the presence of bilateral, non-reactive pupils at 72 hours. And when we say reliable, again, we should stress that we must always use multi-modal prognostication. We should always be aware, you know, conscious of the possibility of confounders. Someone used patriotic eye drops, for example. And the guidelines do suggest that if you have a pupilometer, you should use it. And the reason for that is, just as an example, in one study, about a third of pupils that were judged to be non-reactive by neurocritical care nurses were actually found to be reactive when they were reassessed with a pupilometer. So given that the stakes are so high, you really want to use the best tools at your disposal. Now, how did those prediction studies over the years go so badly wrong? And we think that the answer is the self-fulfilling prophecy. How this...and Patrick talked about this a little bit, but basically how this works is I see that the extensive response is a patient has an extensive response at 72 hours. And as I have done many, many times in the past, the years past, I walk up to the family and say, I think the prognosis is poor. So the family says, well, okay. You know, he would not want to be that way. They withdraw life support. And I have just made my prediction come true. Whereas if we had continued life-sustaining therapy, there's a reasonable chance the patient would have actually done okay. Now, this is an important source of bias in studies of prognostication after all forms of brain injury and really sort of contaminates the literature that way. What about imaging? Now, the guidelines consider imaging to be moderately reliable for the prediction of poor outcome. What are we looking for? On a CT, we're looking for the diffused loss of gray-white differentiation with cell-color basement where the CT performed at least 48 hours from ROSC. On an MRI, we're looking at diffused restricted diffusion. On an MRI, that's performed two to seven days from ROSC. What do we mean by the term diffuse? We mean across muscular distributions, across anterior and posterior circulation with involvement of the bilateral cerebral cortex and white matter. What about EEG? Again, moderately reliable for the prediction of poor outcome. What are we looking for? Suppression, which the ACNS defines as a background voltage of less than 10 microvolts for over 99% of the record. Burst suppression, which is the same thing for 15 and 99% of the record. We really look at this at 72 hours after ROSC and should be aware that there's many potential confounders, deep sedation, for example. These rhythms are often seen early on in the early hours of a cardiac arrest. Very often go away and some of these patients do in fact make a good recovery. How about the absence of reactivity on the EEG? It's really not considered reliable. It has a false positive rate of up to 40% or more. Status epilepticus, there's several patients, reports of patients with status epilepticus who did in fact make a decent recovery. And it might be that the background that patients revert to after you treat seizures is more important. Somatosensory evoked potentials, this is when you stimulate the median nerve in the hand and then you follow the conduction all the way to the cortex. And of those cortical responses, the N20s are absent on both sides. That's indicative of a poor outcome. This is considered to be reliable as long as you're using multimodal prognostication. You do need to confirm conduction through the other points in the pathway to make sure you don't have a technical error or for example, cervical spine injury from hanging. It's not as affected by sedation or paralysis, but it can be obscured by muscular artifact, by noise. So we do suggest that you use one dose of neuromuscular blockade when you do the test to maximize the quality of the test. Neuron-specific enolase biomarkers no longer considered reliable. The original 2006 practice parameter did suggest this 33 microgram threshold as reliable. We now know that has a false positive rate of 40% or more. Now a higher threshold, 70, 80 micrograms might be reliable. We just don't have enough high quality data to support the idea that it is. So very quickly. So another patient, 71-year-old man, cardiac arrest following STEMI, gets therapeutic normothermia. This patient has a return of a continuous background within 12 hours. You do an MRI at 48 hours and it is normal. At 72 hours after RASC, he has weak localization on both sides. What is your assessment of the prognosis? Poor, promising for recovery, or indeterminate? Excellent. So 61% say that it's promising, and indeed, it is promising. So these guidelines talk about predictors of good outcome after cardiac arrest, and this is the first time that any guidelines have really addressed this issue. It's sort of a new idea and a new concept. In the past, we've really focused on the prediction of poor outcome. Now, what were the predictors that were identified as being something you can use for prediction of good outcome? Keep in mind that the threshold for the prediction of good outcome is lower than the threshold we set for predictors of bad outcome. So a motor-responsive withdrawal or better, a clean MRI two to seven days from RASC, no DWI lesions, the return of a continuous background less than 12 hours from RASC, and the return of a reactive background less than 24 hours from RASC, and SSCP N20 amplitude of greater than 4 microwatts. These are the predictors of a good outcome. This is the algorithm within the guidelines. Strongly recommend that you take a look at it. Patients are following commands at any time. They've met the definition for awakening, and they are likely to have a good prognosis. Patients who don't have brain stem reflexes should probably undergo a brain death evaluation. 24 hours or more from rewarming, and you then look for predictors of good outcome as we talked about. If they have these predictors, a good outcome is not assured, but it's more likely than not. It's then recommended that you look for the more reliable predictors of poor outcome, the absent pupillary light response and the absent cortical responses on SSCP. If one of these is present, then the suggested language is that a poor outcome is very likely, but that some uncertainty should be acknowledged. If you don't have any reliable predictors, look for moderately reliable predictors. That's basically imaging and EEG. If you have any of these, then the suggested language is that a poor outcome is likely, but that significant uncertainty exists. If you have none of these, you've got an indeterminate outcome, in which case, an extended period of observation is recommended, as long as that's consistent with the individual's goals of care, the family's goals of care. The duration of observation should be individualized. Two to four weeks is reasonable. Really, you know, depends on the situation. Patients who wake up late are, in fact, somewhat less likely to recover functional independence, but can do so. In the setting of an indeterminate outcome, it's important to acknowledge that the likelihood of recovery, the timeframe of recovery, and the extent of recovery is uncertain. So I'm going to stop there, and, you know, if there's any questions, my email is up there. All right, so I got the lucky task of doing a prognostication in the setting of ECMO. My interest is very much cardiac arrest, so I thought I would mostly speak about patients who have undergone, or will potentially undergo, eCPR, or extracorporeal cardiopulmonary resuscitation. So this is a case, I work with our post-cardiac arrest group at Pitt, and this is a case we took care of in the last year. So this is a 17-year-old boy, no past medical history. He had an out-of-hospital cardiac arrest. Family witnessed him to become unconscious, and they called EMS. No clear bystander CPR. He was in the ICU for about a week, and he was in the ICU for about a week and a half. He became unconscious, and they called EMS. No clear bystander CPR. When EMS got there, he was in polymorphic UT. He had over an hour of CPR in the field, and he ended up getting defibrillated times six, epi times five, a bunch of amnio. His final rhythm in the field was actually asystole, so he comes into the ED on a Lucas device that was ongoing, so did not have ROSC yet. In the ED, he required a significant amount more of resuscitation of ACLS. He had multiple different rhythms, administered three milligrams of epi, and then his pupils were actually in the middle of, you know, going on and off, arresting. They were five millimeters briskly reactive, but he had no other brainstem reflexes except for maybe over-breathing the ventilator, hard to say, and no motor exam at the moment when we saw him down there. EMS states maybe he was posturing at some point, unclear what that really meant. Labs were drawn. His lactate was greater than 20. His pH was less than seven. A PCO2 of 45, a PaO2 was okay. Some elevated LFTs in his tropona, not surprising, very high. So my first question is, what criteria should be used to decide about eCPR? Perfect, yeah, and I think that is the right answer, and so I'm going to talk about that a little bit and kind of what's the impact of not having clear guidelines or a clear understanding of who should be cannulated and who should be not. You know, what Dr. Rajiji went through was a post-cardiac arrest patient and how prognostication dramatically influences outcomes, so if we feel that a patient's going to have a poor outcome, we usually say that to family members and we do some type of shared decision-making. Many times the decision is to withdraw life-sustaining therapy and the patient ends up dying, and if we're wrong in that scenario, then there are potentially deaths of patients who could have potentially had good outcome, but we don't necessarily, and this can be true kind of all throughout medicine, we think about it in, we think about the effect of prognostication in sepsis, in stroke, but also in ECMO, because how we prognosticate initially also impacts our decisions to perform life-saving interventions such as eCPR, and if we upfront are not able to have accurate predictions about how patients are going to do, we may be making, maybe not have the best judgments or best decisions and worsening deaths of patients who potentially could have had good outcome. And ECMO to me is very interesting, or eCPR is very interesting, you know, I go down there and it's very chaotic in the ED, and especially when it's a young patient who's arresting, it's a very high stakes decision and it's very chaotic, and I have to advocate for the patient potentially to get cannulated, but at the end of the day, I'm not the one who's cannulating, it's a lot of my colleagues, my CTICU colleagues who are, and it's this very fine balance, you know, it's this balance of they potentially, this patient will die without treatment, if we do go on eCPR, potentially it can improve cardiac outcomes or neurologic outcomes, it allows time for other life-saving interventions like getting them to the cath lab, at the same time, we don't love in medicine to do things to patients that would have no effect on their overall outcome, something called medical futility, it is a very high resource intensive procedure, and it can cause a lot of emotional distress, especially for patients who would never have had a good outcome, one to the clinician, you know, to my intensive, my colleagues who are in the ICU have to take these patients off of ECMO because they have a poor prognosis, this is very difficult, and it can put surrogates and patients through a lot of emotional distress, and so what factors can help us decide about one to cannulate, this is a really nice meta-analysis in resuscitation recently, that looked at association of pre-eCPR prognostic factors and the likelihood of good outcome, they looked at over 29 observational and randomized studies, over 7,000 patients were included, and not surprisingly, I mean, I don't think this is very surprising at all, that factors that were had either moderate or high association with increased survival with a favorable outcome, more stuff like pre-arrest factors, so young age, female sex, and then intra-arrest factors, shock, low rhythm, witness, bystander CPR, ROS, shorter time to cannulation, interestingly, even though lactate is sometimes used, it has a very low association with whether the patient will have a good outcome or not. And so then, how do centers decide who should get cannulated or not? This is a really lovely survey from Dr. Tona's group, what they did is they surveyed over 70 centers across the U.S., only about 30-something have ED ECMO, so are able to do eCPR, which is not a whole lot in the U.S., almost all of them are academic or teaching hospitals, so was the patient lucky enough to end up in that hospital to get eCPR? 65% of those centers are less than five years old, and 60% of those programs have less than three cases a year, so not only is this a high-stakes chaotic situation, it's something that most centers don't do at a high volume. To me, the most interesting fact was the fact most of these programs do not have formal inclusion and exclusion criteria, so when they surveyed these centers, they used all different criteria, but overall there was no formal criteria that they implemented, it was really up to the decision of whoever the physician who was cannulating in the ED to make that decision to cannulate or not. As many of you know, or may not know, there are two recent landmark RCTs of eCPR refractory out-of-hospital cardiac arrest, well, three, but two showed two different results, we have arrest and inception, and there's many different reasons why that is thought to be the case, but one major factor is potentially patient selection, so who we choose to cannulate on ECMO directly influences outcomes. On the left-hand screen, I have, not from our center, but from another very high-volume center, the inclusion criteria of patients who would undergo eCPR, and actually my patient would not, this patient that we saw, would not meet this criteria, would not have been cannulated. And so, okay, so that's great, we try to make, maybe we have protocols, maybe we don't, but also with Dr. Tona's research, he did a really lovely study, this is a qualitative study of 12 semi-structured interviews of physicians who already performed eCPR, and what he found using these qualitative studies was that physicians had very strong beliefs, so overall they want to follow protocols, but they, over time and time again, there's multiple really lovely quotes in this paper that show that they will break them to cannulate younger, healthy patients, or patients with immediate pre-hospital CPR shockable rhythms, and that physicians really lacked echo poise to randomize these types of patients. And so the question is, should we really be enrolling these young patients with these, like, immediate pre-hospital factors, or should we be looking at older patients, younger patients without immediate pre-hospital CPR, or patients with high comorbidities, because physicians really don't have echo poise in these scenarios. And one issue when we don't have protocols, and we leave it up to physician decision-making, is that potentially we add in a bunch of biases to who we cannulate and who we do not, and who will have better outcomes or not. And so, a similar researcher, I really love his work on this, but he performed an observational analysis looking at adult inpatient cardiac arrest using the AHA's Get With The Guidelines for the Resuscitation Registry. Overall, eCPR was used in less than 1% of cases, but when you look at the cases, there was a huge influence on both patient-level factors and institutional-level factors. So, age, sex, race, comorbidities, time of day, so time of day and arrest location were dramatically, were definitely associated with outcome. Patients were more likely to get eCPR if they were young, healthy, white males who arrested during the day. And so, are we setting up a scenario of disparities because we don't have good guidelines or a good sense of really who's going to benefit or not, and who we should really do patients, how we should select these patients in a thoughtful way. So, for my case to continue, we actually decided, we advocated very greatly to cannulate this patient, even though he did not meet our local inclusion criteria. He had, once cannulated, he ended up going for a CD head that had some mild edema on it. His gray-white ratio was 1.4. Initially, he was in burst suppression, not identical burst, though. In over 48 hours, they transformed to continuous and reactive background. He did have generalized periodic discharges on top of that background at 1 Hz. On post-arrest day 3, he's not awake. His pupils are reactive on pupilometry. He does gain some brainstem reflexes back and some motor back. He's flexing in the bilateral upper, triple flexing in bilateral lower. We perform evoked potentials at 72 hours, and he's got positive N20s. We can't obtain an MRI because he's on ECMO. On post-arrest day 7, he is not awake, but continues to be in quite significant cardiogenic shock, still requiring peripheral ECMO support. And so, what's the patient's prognosis now? So, yeah, so 75% say indeterminate outcome, which I agree with. I'm just going to kind of take you through very quickly, you know, Dr. Rajiji went through cardiac arrest prognostication, and I just wanted to talk you a little bit through how tests in eCPR are different. So, there are a lot of issues with using our classic post-arrest prognostication practices, and I think that essentially my take-home message is you take that uncertainty in the most recent NCS guidelines, and you make that even more because of so many different factors. The main one is sedation. So, we classically, especially here at Pitt, we use a continuous Dilaudid drip that we don't always use in a lot of our other post-arrest patients. We also are using, because of the circuit and how the circuit deals with metabolism medications, we're sometimes using medications with longer half-lives, so we use a lot of phenobarb, so we're really confounding a lot of our prognostic tests with our sedation. We do set the temperature with our circuit, so even if they go undergo targeted hyperthermia or not, there's differences in clearance of medications that they're given. Many tests we can't actually get on ECMO, so MRI is the main one, they're cannulated with metals, we can't put them in an MRI machine. There's also limited studies in eCPR, so most of our studies that are looking at these prognostic tests in eCPR are really single-center studies, and so really, can you use the same interpretation of those studies with the same false positive rates? They're probably the largest registries out of Japan, and they're slowly publishing more and more. There was a paper that just came out the other day about, we use a lot of the gray-white ratio for CT head, and their paper showed that probably their AOC curves different on ECMO, so there's probably differences in tests, and that needs to be sussed out, and we're probably withdrawing even more frequently in these patients compared to our overall post-duress, and so is a self-fulfilling prophecy even worse? And then many of these patients, because of the significance of cardiogenic shock, of multi-system organ failure, and that will decrease your metabolism of drugs even more, and to me, you know, when I get asked to prognosticate these patients, sometimes it's not even the brain that's going to be the main factor, it's really going to be a prognostic, helping them prognosticate the whole body, and it could be that their cardiogenic shock or their multi-system organ failure is actually going to be the rate-limiting step compared to the brain, and so, you know, there's many different tools that look at refractory cardiogenic shock that people have used over the years, and it's been shown time and time again that because, once again, this is really done at a population level, but there's a lot of biases with these prediction models, and my personal belief is that, you know, we're not great at prognosticating overall, and we're probably even worse in terms of the brain, we're probably even worse when trying to prognosticate who's going to recover from cardiogenic shock or not. So to continue this case very quickly, we have a post-duress day 7, the patient develops hypoxia from pneumonia while still on peripheral VA ECMO. He actually develops a peripheral oxygen difference on the gases between the two arms, and he's actually on NEARS, and we see a drop in his brain tissue oxygen on the NEARS from baseline. So my last question, what are the complications of ECMO that influence prognosis? Oops, so this is supposed to be what is not, what are not, what, which of these are not the, which of these is not a complication? I messed that up. So yes, this, what we're talking about here is North-South syndrome, and so if we continue, I messed up that question, did not realize I missed a word, but there are many different things that can happen while a patient's on ECMO that can actually influence your neurologic prognostication. So this patient ended up developing differential hypoxia in the peripheral VA ECMO, and essentially what happens here is when they develop systemic hypoxia because of pneumonia, the LV ends up having deoxygenated blood that competes with the oxygenated blood from the circuit since it's peripheral, and you end up getting a differential between the two, and so you actually can get hypoxia to the brain, which potentially in theory can worsen outcomes, and there are different ways to monitor for that and fix that, but that is one of the big complications that can happen on VA ECMO that you have to think about in terms of secondary brain injury that can influence patient outcomes and prognostication. And there are all other complications. There is aortic dissection, early versus late. They can get spinal cord infarcts. We don't necessarily think about that pretty frequently, but that can influence their neurologic outcome. Limb ischemia, so how well are they going to be able to deal with limb ischemia, how well they're going to be able to deal with rehab. We know that 15% of patients on VA ECMO do have other subsequent neurologic complications such as stroke, intracranial hemorrhage, seizure, and then there's all other secondary brain injury issues that we see with our normal post-cardiac arrest patients. And then something that I've been consulted a lot over the years down in our cardiothoracic ICU is ICU acquired weakness. So many times we can see patients awaken, but they still can't move, and thinking about ICU acquired weakness is really important. And we can monitor for some of these, and most of this is done in our general ECMO. There's a bunch of studies done in our general VV, VA ECMO, but not ECPR. And so we potentially can monitor for some secondary brain injury with NEARS, EEG, pupillometry, but most of these studies, once again, are single-centered and done in adult ECMO. So a huge, currently, limitation to what we're able to monitor for. And so to continue our case, this patient was ended up switching to VAV ECMO, and their brain tissue hypoxia improved on NEARS. Over the course of a few days, their cardiogenic shock improved, and they ended up with decanalated from ECMO. They actually awaken on post-arrest day 21, and then are discharged to rehab. And this is not surprising. If you look at the group, it's a really lovely paper from 2013 that shows that, you know, majority of our post-arrest patients do awaken in the first few days, but probably even majority of them are the seven days, and then there's a full slew of cohort that wakes up at 14 days. And with all the sedation, the metabolic disarray that this patient underwent, it is not surprising that this patient took 21 days to awaken. Thank you. I'm gonna hand it back over to Dr. Nelson. Thank you all so much for the terrific presentation. So I know we're almost at time, but I was hoping you might be able to stay on to answer a few quick questions that came up in the chat. So we'll start with Dr. Chen. One of the first questions that came in is, would you consider the case you presented an outlier, or would you constantly tell family that this is your usual expectation? And sort of related to that is, would you consider age as playing a role in prognosis? Yeah, that's a great point. I mean, I would say it's sort of a moderate example of what we see. You're right that all these models and everything are largely driven by age. There was a study by TrackTBI that was done recently that says, you know, 50% of severe TBI ends up doing great at one year or something like that. And you're right that those populations are all very young in general, and TBI has historically been a young disease. And so certainly if the patient is in this like non-elderly category, then we will heavily probably use examples like this as possible outcome. I think you're right that once you start entering the elderly population with multiple comorbidities, it's a very big question mark, I think. And that's, we're trying to figure out better prognostic models for the elderly, and that's where the nuance of really hammering the point that like the elderly that get into that situation that I described, they will at best end up with a tracheostomy and a PEG at least for a couple months. And that in itself is enough often for families to make a decision off of. So that's usually what I hinge on until we have better backing with like models and stuff. Thank you so much. And then there was an interesting question that came in for Dr. Jhaji. I worry about the metrics by which we define a good neurologic outcome. Are there any data to look at whether patients' families would make the same decision about withdrawal of life-sustaining therapies if knowing what they knew years after the decisions were made? Yeah, no. So that's a great question, and it's incredibly complicated. The short answer is yes. There's been studies in a variety of brain injury settings that have looked, you know, we've gone back and asked people, would you have made the same decision? And it's, you know, there's no single easy answer to that. I'd say it's complicated. It really ranges. The responses range from I didn't necessarily understand what the choices were at the time, all the way to the disability paradox, which is just, you know, I think sort of like the first patient that Patrick was talking about, who many of us may consider very disabled, but if you ask him how he thinks he's doing, he thinks he's doing great and has no regrets at all about all of the choices that were made on his behalf. So those studies have been done. I don't know that the answers that we've received from those studies have really been straightforward. I think they're complicated. Thank you, thank you. And then a question I think may be applicable to both Dr. Steinberg and Rajaji, if you want to start, Dr. Steinberg. Thoughts on using invasive brain monitoring in cardiac arrest patients, like monitoring for ICP, PVT02, things like that. Yeah, I think it's very useful. I think though the caveat is the patient population here, right? So it's a little different than TBI and Ceberax. We actually have an agreement with our neurosurgical group to do intraparenchymal monitoring, if indicated, but remember a lot of our patients end up undergoing cath, and so they're on DAP, so it's not, it is a higher risk thing, and then once they're on VA ECMO, they're actually anticoagulated. So we really need to be thoughtful about who we put intraparenchymal monitors in, and I think it would be really useful because there's a lot we just don't know from a secondary brain injury standpoint in these patients, but there is some risk associated, which is a little different than the rest of our patient population in our ICU. I was just going to say, I totally concur. I don't know that it would be considered the standard of care, but I think there's certain, you know, some centers, I think Alexis Center in particular, that's kind of cutting-edge of this, but I don't know that it would be considered standard of care across centers. That does make sense. Not all centers have access to all those types of therapies, I recognize as well, and then a question that also came in, Dr. Chen, for you during your presentation, but I think really applies to all of the speakers' presentations as well, why I recognize that withdrawal of care is a common term of art. Isn't it important in prognostication and otherwise to refer instead to withdrawal of life support? After all, if we refer to not caring for patients after prognostication occurs, doesn't that also influence how prognostication influences outcomes? Yeah, I'll take a stab at that. I mean, it's a good point, and I think, you know, I think, especially in the TBI literature, that's very clear in the latest guidelines, is that all of the prognostication things that we have out there are tainted by this self-fulfilling prophecy and withdrawal of life support. Like, none of those studies accounted for withdrawal of life support. A lot of them didn't even say what the rate of withdrawal of life support, even though we know that this is a very, very common situation, and I think, you know, in general, any future model has to account for, and that was actually one of the guidelines. Within the guideline, it says next steps. This is, like, one of the most important things that we have to address in any sort of future prognostication type of model. I'll let the other speakers also speak on about this conundrum. Yes, I mean, I think there's some recommendations on how to minimize the impact of the self-fulfilling prophecy in studies of prognostication, but even those really just mitigate the impact of the self-fulfilling prophecy. You can't prohibit people from withdrawing life support, you know, so as long as people have the freedom to do that, they should have. I think that there's no way to entirely remove that source of bias, but I think there are definitely ways to mitigate it. I'd encourage people to look at that section of the guidelines that Patrick refers to. You know, they really go over. If you are planning a study of prognostication of any form of brain injury, it's useful to look at those suggested ways to mitigate the bias from the self-fulfilling prophecy. Yeah, I think from a research standpoint, I think that calling it out in the papers, what your rates of withdrawal are, are super important. I see, you know, as a reviewer of many papers, that that's frequently missed, and even in a lot of our big clinical trials, like for post-arrest, we don't actually have good rates of what withdrawal by sustaining therapy is, and so how did that influence outcomes of different interventions? I think, you know, we're never going to be able to not do withdrawal by sustaining therapy, but calling it out in our papers, I think, is quite important. Thank you so much, and then let's see here, so a few other questions kind of came in after the presentations. After you finished presenting, let's see here, when a patient comes in with absolutely no exam after a TBI, how do you decide to push ahead or declare a patient brain dead? And I think you answered this a little bit here, Dr. Chen. Did you have anything else to add to that? No, I mean, the exam can't be trusted. I think what we've shown is that the initial GCS exam is very highly unreliable, and they said it's in the guidelines too, as well, as to what the ultimate outcome is, so no decision should ever be based off of that. Often, brain death declarations don't happen that fast within the first couple hours because there's so many confounders, and this is a whole other topic that can be unwrapped in the future, but brain death in itself has to fit all of these certain criterias. You know, no sedation in five half-lives is probably a classic one for TBI. Like, there's a lot of sedation in the first couple hours of TBI, and so we never make that call that early, and if a family can't decide, I mean, which is often the case, we do full care and full treatment under the paradigm that a lot of TBI can theoretically do well based off of all these studies that I've shown. And then some questions. Thank you, Dr. Chen. A couple questions about cardiac arrest. I know we're still going over, so maybe these might be the last two questions. I apologize that we're not going to be able to answer all of them in this time frame, but this question asks, when would Dr. Rajaji recommend doing EEG? Would it be more meaningful to do it at the 72-hour mark? And then potentially a question for both of our presenters who discussed cardiac arrest. A discussion about gray-white differentiation. Seeing it early, even less than 48 hours on CT, has invariably for this particular person been associated with a poor outcome? Have there been cases where this actually resolves or improves? Is it worth it to, you know, have a long observation period, essentially? So in terms of EEG, I think it really comes down to what resources you have. I think at my center and many other centers, we connect the patients up to continuous EEG as soon as the patient is stable and the suscitation is complete. The idea is that you can, you know, detect and manage seizures and satitis epilepticus if they show up, and it's useful to be able to detect, for example, the return of a continuous background or reactive background early, but that's simply not the resources that are available at a lot of centers. So, you know, the only thing we are trying to do is to get a snapshot for prognostication that it's very reasonable to not do continuous EEG and to use just, you know, a single EEG at 72 hours once you're off ulceration and you're beyond all confounders. So my take on the CT question is, when I see, you know, truly diffuse loss of gray-white differentiation, the way that they've been defined by the guidelines, when I see that early, I'm telling myself, I don't like this, I think this looks good, but I still wait the 72 hours, and I do this specifically because I've been burned in situations where I think often the issue is technical. Sometimes our EEG leads, even if they are CT compatible, they're able to generate just about enough artifact to kind of confound gray-white differentiation. There's a variety of factors, so I'm, you know, it's true that when you see diffuse loss of gray-white differentiation early, you want to worry about that, and it's certainly concerning, but I would still encourage giving the patient sufficient time, incorporating, you know, other factors as well. Yeah, I agree. I think that it also is like what, you know, we give them a gray-white ratio number, we numerically do it, which to try to quantify it, but you know, when you see the read or when you're looking at it, it's actually very qualitative, and so are we talking about like severe, you know, loss of gray-white differentiation, or are we talking about mild edema, and I think giving, you know, so I think giving people, trying to get time and getting more tests is actually very important. Thank you, and I lied, just one other quick question because I'm curious as well. Would you still wait 72 hours if the patient underwent therapeutic normothermia at 36 degrees? So the guidelines suggest that you wait 72 hours from ROSC, so 36 degrees would be considered, you know, therapeutic normothermia, so it's, so then the guidelines are 76 hours from ROSC rather than 72 hours from pre-warming. Okay, okay. Thank you so much, because I was curious about that as well. Well, I know we've run over time, and I really appreciate the speakers' time and all of their, all the questions that they answer for us. I hope that this was really educational, and I really appreciate the audience for attending, and once again, this webcast is being recorded, and the recording will be available about five to seven business days on mysccm.org under the My Learning tab, and that concludes our presentation for today. Thank you so much.

Webcast

Neuroscience

Traumatic Brain Injury TBI

Neurotrauma

Spinal Cord Injury

2024

Professional

Select

neuroprognostication

traumatic brain injury

cardiac arrest

eCPR

TBI recovery

prognostic tools

GCS limitations

bilateral non-reactive pupils

ECMO protocols

multi-modal approach

individualized care