Great. So, we have heard lots of great data, wonderful slides. I've been really learning a lot. I'm going to talk a little bit in the first half of prognostication, sort of in the past, to see how we've gotten where we are and we're so confused. And then the second half of my talk is going to be in the future with neuromonitoring, because I know not many people are using it, but I think that this is potentially a path forward. So the data about survival after out-of-hospital cardiac arrest shows that over time, people are doing better in terms of survival to discharge. All these studies are based on wide population cohorts, epidemiologic studies. They're not really the same. Some focus on good neurological function. Others focus on survival. What is known, though, is that prognostication or survival with good neurological outcome probably has a lot to do with the initial rhythm or the etiology of the cardiac arrest, with the ventricular tachyarrhythmias with some perfusion having the best likelihood of neurological outcome, all comers being the same. So after this understanding how long a patient has been in circulatory arrest, what were the circumstances of the resuscitative efforts, and understanding the underlying comorbidities, we really see patients then in the neuro ICU or in an intensive care unit sharing that final common pathway of coma. So it's sort of a black box. People come to you with hypoxic ischemic brain injury or some potential for it, and they all look like they're in a coma, either from their injury or a combination, love to come right after the pharmacy talk, of deep sedation or paralysis. So it's really difficult to tell, and this is an opportunity for trying to improve secondary brain injury. So I'm going to talk about sort of the neuroprognostication story in sort of a timeline and sort of pepper them with sort of signs about hypothermia trials, because it really has influenced sort of our community's understanding of how we are neuroprognosticating based on where we were in time when these sort of guidelines and things came out. So the HACA and the Bernard trial came out the same year and showed, you know, suggested that hypothermia might be very helpful. So we're not going to dwell on these because this has been covered very well by previous speakers. Shortly thereafter was sort of the first update since the Levy criteria that had that nice flow sheet of saying very, you know, binary, if this, then this. It's very clear, and it must have been nice to live in that kind of an era where there was no uncertainty, you know, in terms of communicating with families about what one should do in this really devastating situation. And the evidence that was surveyed for this review really spanned a great deal of time up until that period of 2006, and the predictors that were analyzed, the literature that was surveyed for this really covered these nine types of features, from the exam all the way to EEG to biomarkers, imaging, et cetera. And the predictors that really kind of rose to the top for this review were exams, somatosensory evoked potential, biomarkers, and myoclonus or status epilepticus. And they even provided with a very nice flow chart, probably because there was evolution from the Levy flow chart. You know, people were really expecting and hoping and maintaining this hope for a nice flow chart of if this, then this, and this is what we could communicate to families. All of these studies were from before TTM, unfortunately, so the practice was moving forward with TTM, but the prognostication guidelines and the reviews were really focused on a pre-TTM period of time. And so, and if you look at the studies that comprise the guidelines, they all emphasize poor outcomes. What do I mean by that? It really was statistically aligned to say, well, what is the likelihood that this predictor can predict that someone's going to have a bad outcome? And then that bad outcome is defined by that research group. And these are the three sort of traditional common outcome criteria or scales that people use, and you'll notice they're not completely aligned. Just for one example, look at the modified Rankin scale for four and five, and it's really severe disability broken into two. And really, in a lot of the dichotomized studies, that severe disability, which is a moderate outcome for the other two, ends up being the same as a coma or a vegetative, persistent vegetative state. So why was this done? I mean, it's because at the time, people were thinking, well, the rationale behind building these kinds of flow charts is, well, most people would choose not to continue living if they reached some severe or severe state of disability or vegetative state or coma. The problem with this is that there's no real flow chart that really can be sustained by the literature that exists. You know, the goal here is to help families and surrogates try to avoid the withdrawal of life support, avoid that. If there's any kind of plausible chance of recovery, if there's even a zero point, whatever comfort level the surrogate might have with having that certainty. And so the tests need, then, if that's the goal, to have a near zero rate of false positives for determining poor prognosis, right? And that's just impossible. None of these studies had that. And also, all of the studies that went into this guideline did not avoid self-fulfilling prophecy, meaning that the feature that was being or the variable that was being studied was eventually released to the practitioners to actually inform their conversations that went into withdrawal of life-supporting therapy. So all of that ends up meaning that this flow chart that was presented ends up being a little bit overly pessimistic because the very variables that they're being promoted for helping you prognosticate are also the ones that have withdrawal of life support therapy not considered. So let's look at those four things, the clinical exam, the somatosensory evoked potentials, the biomarkers, and the myoconic status epilepticus, because there is some good things to come out of knowing this literature. So in the clinical exam, there are three things that you look at, sort of the brainstem things, like the pupils and the corneal reflex. And then you look at the motor response to painful stimuli, extensor posturing, or if you're doing nothing. So if you look deeper into this, you know, if you had the absence of epilepsy, then it would say, oh, it portends terrible outcome. And the reasoning for that is if you have such severe injury that you have no brainstem reflexes and your brainstem is actually more resistant to hypoxic damage than your cortex is, well, that must mean then you have widespread cortical damage. That means that you're not going to live a meaningful life. However, just because you have preserved brainstem reflexes, it doesn't actually imply that you have intact cortex. So it really only is very unidirectional in saying that. And more importantly, this number two motor response and corneal reflex, they're actually highly affected by neuromuscular blockade, drugs, and sedation. So the release of a guideline like that, just as people are starting to adopt hypothermia, putting people on drugs and sedation, and then doing a prognostication at 72 hours, you can see where that might become problematic. Somatosensory abode potentials are great. This is basically you're zapping someone at their median nerve in their arm, and then you're measuring the cortical signal at the N20 time, 20 millisecond time. And if you have this present, it's, again, binary. That means that there is a connection. There's enough functioning brain cells between that can elicit this response. But the absence of it, how do you interpret that? And so there are a couple of black swan case reports that have come out where you have an absent bilateral SACP, and the patient has a wonderful recovery. And I would argue if you have one of those, that's sort of enough to say, well, how many more are there? And especially when the SACP is the most prone to informing care and promoting that self-fulfilling prophecy, you can see where this could become problematic. It's also interesting that with any kind of electromyogram, the EMG that you're doing, they're all temperature sensitive. And so if you're doing this while you're being actively cooled, it's probably not a great idea. Will it totally remove the N20? Probably not, but it certainly can delay it in a linear fashion. It's also very prone to electrical interference. So in a very crowded ICU, maybe with ECMO, with other devices, you might not be able to get a very great signal. It's also very sensitive to sedation, which a lot of these patients are on. So the timing matters. Biomarkers also initially were super helpful. You can kind of use it on a threshold. If you were greater than this number, that probably meant you had so much widespread neuroglial disease or dysfunction that you're probably related to your possibility for cognitive outcome. But then hypothermia changed all that. It changed the threshold. And again, this was NSEs were being promoted as something you could use on a threshold-based way. You can actually see much higher NSEs with good outcome after hypothermia. And then finally, the status myoclonus. So interestingly, this myoclonus has been heralded since Levy, since the 2006 guidelines, that if you have status myoclonus, then you really kind of have no chance for good recovery. An interesting look into this with Dave Seder and his group looking at the NCAR data was that actually 20 percent of these patients who were diagnosed with myoclonus never got an EEG. Myoclonus is just a clinical observation of shaking, maybe with some time associated with it. But without an EEG, you're really doing a lumping, and you're doing a disservice to patients because not all myoclonus is the same. And so the limitations of status myoclonus, unfortunately, continue to be a little bit plagued by this inconsistent definition because of the historical problems that we have had and sort of propagated. And again, this can lead to a self-fulfilling prophecy. The EEG work that has been done, though, has actually tried, has sort of kept myoclonus, status myoclonus in the conversation, but now with this added benefit of knowing what the EEG is. I would point you to the second and third column where the myoclonus with poor outcome and myoclonus with good outcome. And I would show, of course, birth suppression, all of these signals have a much higher incidence in patients who have a poor outcome. But importantly, look at the second outcome. There are plenty of patients who have these bad predictors who have a good outcome. So this is the danger of using a single prognosticator and putting all of your eggs in that basket. The upshot is that 10 percent of patients who come in with myoclonus, not talking about EEG here, will have a good functional outcome in that in-car study. So take away, get an EEG. And understand that it's not going to tell you definitively, definitely going to have a great outcome, definitely going to have a bad outcome. And same with seizures. Patients who do well have seizures, patients who do poorly have seizures. But among patients with seizures, the great majority of them died because of withdrawal of life-supporting therapy. So this is a very useful study. I think it points to four very highly malignant EEG patterns. So this you can take into account. Myoclonus is one thing, but whether you have myoclonus plus or minus this EEG, this is what's probably most important. A, on the upper left-hand corner, is completely suppressed. B, on the upper right, is suppressed, but with G-pads, generalized periodic epileptiform discharge. You see them across all of the leads. C, on the lower left, is burst suppressed with the bursts that come out, look very sort of, you know, normalish, not so sharp. And then D, burst suppression with epileptiform discharges. You have those bursts interspersed with those very spiky, generalized outputs, just like you see in the upper right-hand corner. So these four are very extreme EEGs that portend likely bad outcome. But again, this can also be seen in 14% of patients. I'm sorry, this can also be seen in some patients who end up doing okay. I don't know if I can go back, but it was in that chart as well. Okay, so like even burst suppression, you see this in 13% of patients with myoclonus who have a good outcome. You're probably familiar with this study that was released this year in the New England Journal. And you know, it's unfortunate that the title is very sort of, if you were just to be a casual reader of this abstract, you would think, oh, what's the point in treating an EEG? Patients who seize, they don't seize, they have a bad EEG. Seizure medication doesn't do anything. So this is just pathognomonic for that outcome that we're not going to do anything differently with these patients. It's very important is this table on the right, which is that because of the power calculation, the number of patients, while large-ish, was not large enough to look at each of those different kinds of backgrounds. So they lumped together those patients with those terribly malignant patterns, as well as electrogaphic seizures or ictal patterns. And so we already know that those are two different groups of patients. So if you're merging all those patients in both groups, and you're treating one with anti-seizure medications and the other, it's not surprising that you're not going to see a big difference. But it doesn't mean that treating seizures or trying to prognosticate based on treatable seizures shouldn't make a difference. I would argue it absolutely should. And if you look at the subgroup analysis here, the group, if you separate out the ones with GPEDs, which is the highly malignant upper right-hand corner background, those patients, again, did not do any better than control with seizure treatment. But the ones who are non-GPED, which end up being these other patients with electrographic seizures or revolving patterns that look electrographic seizure, soon-to-be electrographic seizures, those patients did do better. Then along this time course comes, well, maybe hypothermia is not all that we thought it was going to be. Well, now this confuses things. In this study, they did assessment of 72 hours after normothermia was achieved, and then they recommend a withdrawal of life support. And then we've sort of talked in our panel and looking at how there's a diversity even in the audience about who's doing hypothermia, who's doing normothermia. So, you know, what I would say is the jury is still out, right? There's still ice cap. There's other trials continuing. Maybe there'll be a subset. But the take-home about neuroprognostication is, think about the sedation the patient is receiving and think about the timing of when you're trying to make some of these conversations with family members. The AHA, not the AAN, came out with guidelines in 2015. And this one, interestingly, had the absence of pupillary response, again, brain stem, status myoclonus you see there again, the absence of EEG reactivity, this persistent malignant background in EEG, or intractable status epilepticus. They talked about the bilateral absence of N20s. And then they added this MRI. It's now in neuroimaging. No longer, as opposed to the AAN guidelines from 2006, no, they don't recommend absence of motor or extensor posturing because actually, if you have a patient who's localizing, that is a good prognostic sign. And so we can use it in a different way rather than portending bad outcome, portend good outcome. They no longer recommend just using myoclonus. They have to do it with the EEG. And then the NSCs was removed in the sense that you should limit this to confirmation. So we're now moving more towards a multimodal approach to prognostication. MRI. So all MRIs are not the same. The timing actually matters. So if you do an MRI too soon, so the top row is two hours, same patient as 55 hours after arrest, you think about MRIs being like a roadmap of all ischemic injury. Well, in cardiac arrest, it turns out that's not true. So hypothermia can mask these changes that come out 55 hours later. So the timing is very important. So a very important take home is if you are going to do an MRI as a multimodal feature of your informing your prognostication, wait till at least three days after your normothermic. So in the context of TTM or not TTM, wait for 72 hours after ROSC or after return to normothermia. With awareness of your context or metabolism of the medications that you chose on your individual patient, does it make sense that maybe this patient has cleared all of the midazolam, for example? You want to exclude the major confounders of your reliable exam, like persistent with doing train of force, persistent paralysis, et cetera. Polyneuropathy, ICU neuropathy, myopathy is a patient they were not able to exhibit that they can follow your command because they're actually totaled too weak to do so. And then definitely going towards more of a multimodal approach to prediction and taking the weight off of single modalities. And most importantly, accepting uncertainty. And I'm going to talk a little bit more about that. Again, another potential nail in the coffin for hypothermia, but we know that that's not true yet. Then we have ice cap that continues. And now we're doing this Bayesian adaptive design to try maybe a dose response of hypothermia. So it's clear that some patients will be cooled, some patients won't be. And so this is a good take home slide. If you want to take one picture, this might be a good one for this part of the talk. This multimodal algorithm is sort of time wise as well. So you do the exam, you note things about brainstem, about motor response, remembering that motor response localization is a good prognostication sign. You do an EEG with those first 24 hours. I would argue whether or not you have myoclonus. It can tell you if your lack of, if your encephalopathy or your coma is related to a non-convulsive status epilepticus, so it gives you that information early. You do a daily biomarker, the most common one right now being an NSC. And then you try to do, of course, this is a generic recommendation. We do what we can for individual patients, but if you can, you try to use short acting sedation and discontinue early. And then if your patient is still not waking up, what do you do? You continue to do the biomarker because the trajectory seems to be informative and the directionality. You could continue the EEG because more EEG can give you more information. At this point, you could consider a somatosensory vote potential, but again, not to use it as a single modality, and it should really just be more confirmatory or puts you in a direction another data point, and then you can consider doing the MRI. The earliest prognostication, again, I can't stress this enough, shouldn't be earlier than 72 hours, and in many cases could be later. And if you have any prognosticators that mismatch, I would argue, accept the uncertainty and continue to give time, if that is in line with what the surrogate decision makers want for their family member. How long should you wait? There is no flowchart for this, you know, it's all individualized. I think Nick Johnson said really well about your first, you know, day zero conversation and your day four or five conversation, your data gathering about what your surrogate decision makers and the providers' sort of goals for quality of life, there's always going to be a mismatch. There's always going to be an opportunity to learn about that. It's kind of like if you think about a cancer diagnosis, the first thing that you come to if you have a cancer diagnosis is what are my chances, what are my chances for dying, when am I going to die? These are answers that you know that cannot be answered, but that is always the first question that comes up in mind. But when you think about the statistics, it's given to you as you have a 5% chance of being alive in five years, and some people will say, well, I'm young, I'm going to fight this, I'm going to go for that, right? So it's really just the way that you look at these statistics. Only in cardiac arrest do we seem to strive for this false positive rate of 0% for bad outcome. Nowhere else in neuro-ICU, or we're also dealing with acute brain injury, severe acute brain injury, do we demand this of the literature and do we demand this of providers? And do families demand this? And this is, I think, historical because we went from a flow chart where it was very easy to do that to now where we accept uncertainty, and that the biggest contributor to those studies were early withdrawal of life support, which may have been inappropriate. This is a very, I mean, somewhat arguable study, but I think is a really impactful one for me, which is that looking at a very large dataset and doing an extrapolation of the American population, early withdrawal of life-supporting therapy, less than 72 hours in the past, may have been associated with a mortality in 2,300 Americans a year, in whom 64% might have had a functional recovery. That's really crazy, right? So I think we need to start to shift away from a flow chart mentality of being able to tell somebody, a family member, what your chances are for bad or good outcome and accept that we don't really have that in much of acute brain injury. So in the future, there are lots of hopefully interesting things coming out. There's using ice cap data, they're working on a deep learning method to try to be more precise on an individual level about individual chance for good recovery. This is a study that came out of our group where, you know, and this is, I want to stress this is not in cardiac arrest patients, but in other types of acute brain injury, where you have a patient who's in a coma, maybe an ICH or a subarachnoid hemorrhage, and the family wants to know, are they going to wake up? And using an EEG and doing complex analysis of the EEG using a command following framework, you have them open and close your hand, and 15 to 20% of these patients, they were able to have the same EEG signature offer on that they were able to understand your command. And we call that cognitive motor dissociation, and that's highly associated with recovery with a prolonged time for recovery. And so this has not been studied yet in cardiac arrest patients, and I don't want you to extrapolate this to that yet because it's different diseases, but certainly this is the kind of work in the future that maybe we can try to inform that conversation. There are better biomarkers that hopefully will become available, and specifically the NFL, which we've been talking about for many, many years, but it's very highly sensitive and it doesn't seem to be affected by hypothermia, and its prognostic performance seems to be better than for all of the other variables that we have. Summary, this is another one to take a picture if you like. Remember allow for clearance of sedation, do multimodal evaluation. If you have a bad, if you don't have pupillary reflex or corneal reflex at 72 hours, that's probably a bad prognostic sign. If you have localizing movements, grab onto that. That's a good prognostic marker. The extremes of EEG are more informative than just myoconus. People with myoconus can be awake and talking to you. SSCPs may become useful, but more for good prognosis, so if it's present, but if it's absent, don't use it. Low NSEs are always reassuring. Okay, and MRI, three to five days. I'm going to shift now to that second part. This will be a lot shorter because there's a lot less literature on this, and it's about the future. Multimodality and neuromonitoring, multimodality monitoring. And I'm going to start with this sort of a very short, for people who don't know what multimodality and neuromonitoring is, which there might be a couple of people in the audience, but really focusing on that. In cardiac arrest, it's been shown that invasive PbTO2 or brain tissue oxygen could be useful to avoid secondary brain injury. So multimodality monitoring is a small, so this looks big, but it's just the bolt that anchors it to the skull. It's a small hole, like a burr hole, that's drilled into the skull, and fine catheters are placed into the white matter of the brain. And we're able, through these invasive monitors, to be able to monitor lots of things in the neuro-ICU. But in the context of cardiac arrest, what's been studied is intracranial pressure and brain tissue oxygen. It goes through the same port, and it's right next to each other. There are lots of non-invasive monitors that we do use. And importantly, all of this with exam, with biomarkers, all of it, imaging, we have to integrate that, and we do data analysis to try to give you real-time information about trying to avoid secondary brain injury. And by integrating this, this is a subarachnoid hemorrhage patient. You can see, before a clinician is able to see it physically, that you see metabolites, the glucose-lactate-pyruvate-lactate-pyruvate ratio, your brain tissue oxygenation, your cerebral blood flow, and able to see that a patient is starting to suffer from brain tissue hypoxia and ischemia, and that your metabolism at your cellular level is able to tell you in real-time that this patient is developing a stroke, a delayed cerebral ischemia. I'm actually giving a talk just on multimodality monitoring in coma on Monday, on Tuesday, so if you're interested, it's at 10 a.m. But that's all we're really going to talk about. Multimodality monitoring, we think it's important because it really is a physiological possibility of that we're trying to avoid real-time injury, secondary brain injury, in minutes' time. If you figure out six hours later that you didn't have the optimal mean arterial pressure, you didn't have the optimal cerebral perfusion pressure or temperature or PCO2, to find that out six hours later or the next day is not as helpful as finding out right when you're detecting it to try to do an intervention in real-time. And there's a lot we can do if we understand sort of not just what's a number, but actually what does that number mean in concert with other numbers. So here's that study. So this is a really cool study, I think. It's a very small study, right, so it's all in the future. But this is a group in Canada. They've done primarily all of the work in invasive monitoring and post-cardiac arrest. But what they did was they took 18 patients who had post-cardiac arrest with hypoxic ischemic brain injury. Ten of them ended up having low brain tissue oxygen from various systemic occurrences. Those patients, so they had an arterial and a jugular vein sampling of these biomarkers. And in real-time, they were able to show that when these patients were experiencing the brain tissue oxygen dipping below this number that we kind of accept as at the upper limit of ischemia, brain tissue oxygen can be dropped from several things, but one of the main drivers is ischemia. These patients were having active, time-temporally detectable cerebral release of GFAP, NFL-Tau, all the biomarkers that show neuroglial injury. And the eight patients who did not had brain normoxia did not have this occur. And so is this outcome? No. This is brain tissue oxygen showing you're actually having cellular damage in real-time. It's a prospective study. I find this a beautiful, elegant study that really is very supportive of the idea we need to see if we avoid brain hypoxia in patients, can we actually reduce, in a secondary brain injury period of time, the amount of accumulated neuroglial injury that might actually be associated with cognitive outcome. So the same group tried to do this goal-directed therapy versus standard care. So the goal-directed therapy was really threshold-based, looking at intracranial pressure. If it was greater than 25, treat. Brain tissue oxygen, if it's less than 20, treat. This is the standard care, which is MAP greater than 65, normothermia, keeping your PaCO2 right in this normal range in the neuro-ICU we do, PaO2 80 to 100. And all of those parameters were the same, except for maybe anything beyond this. If your ICP or Pb2O2 exceeded these parameters, the clinician was allowed to do whatever they felt that was right to write those numbers. And what they found was very, I mean, interesting, right? They found a very favorable neuro-outcome using CPC at six months in the goal-directed care. Forty-three percent of that group versus 10 percent of the group that was in the standard care. Very compelling. One criticism of this study, and it is a small study, is that the group that had the goal-directed care also were a group, this is sample size, also had lower temperature. So potentially maybe the standard care maybe had fever, more fever burden, that could actually change your outcome as well. In a recent RCT, it was already mentioned about the box one, the high versus low MAP targets that did not affect mortality or neurological outcome. So this takes another event, right? This is not about ICP or Pb2O2, but this is another systemic characteristic or variable that you could potentially manipulate to maybe influence your outcome. So this is a very large study. It was a two-by-two factorial design, and they showed with 77 versus 63, there was no difference in mortality, severe disability, or coma. But I would argue that potentially it's not just even doing studies where you're doing MAP of 80 to 100 versus a MAP of 66, but rather, could we identify individually a patient what their individual MAP goal should be? Because no matter how you do it, without that kind of individual identification of a goal for a patient, you're always going to be diluting your response, you're going to be hurting people who are not going to benefit from an increased MAP and actually be hurt by it, and you might be not benefiting the patients who really need it. So this is a study that was done at Penn, and they looked at, retrospectively, post-cardiac arrest and non-post-cardiac. So these are all opioid overdose patients, the four that were not post-cardiac arrests, and they justified grouping them together because of the shared sort of pathway. That was what they explained. And all these patients used TTM, or most of them used TTM, and they looked at your intracranial pressure, your brain tissue oxygen, and then something called PRX. And so probably many people have not heard of PRX, but simply put, PRX is a real-time index where you're basically taking the knowledge of the skull is a closed system, and you have volume. And volume is equivalent to pressure, especially as you get to the upper limit of that volume. And so for a large portion of that, if you add fluid or substance to a fixed space, you can accommodate that volume and not have too much of an increase in pressure, but when you get to that right-hand side, you get this abrupt small change in volume is experienced as a high pressure, when you no longer have the ability to egress things like venous blood or cerebrospinal fluid, et cetera. So on that pressure-volume curve, especially on that second half of it, as your compliance is sort of decreasing, you have this ability to look at changes in blood flow, changes in cerebral perfusion pressure. If you exceed the amount that your blood volume can be accommodated for, you're going to see a concomitant increase in pressure. So if you have volume change, increased blood pressure, and a concomitant increase in pressure, volume pressure, that's a positive correlation, and that's bad. That's showing that you're no longer auto-regulating enough by vasoconstriction to keep the same amount of volume in your brain, right, to maintain a cerebral blood flow requirement. And so capitalizing on that, that's what a PRX is. It's a moving correlation coefficient between tiny changes of blood pressure and tiny changes of ICP. And so this has been shown in TBI and subarachnoid hemorrhage, other types of acute brain injury, to show that the association of this auto-regulation, you can imagine, I have four hours of information or even an hour. The patient has been fluctuating for normal nursing care, turning, suctioning, just living. And with those fluctuations, you can capture this passive information and give you actually every time my CPP, my cerebral perfusion pressure, was X, this was where my auto-regulation curve was. And in so doing, you can actually draw out a U-shaped curve that very much aligns with a auto-regulation curve. So at the upper limit of auto-regulation, you're pressure passive. If you give more volume, you're going to really increase your pressure. And below that, you have a lower limit of auto-regulation. When you surpass that on the other side, you're going to start having misery, perfusion, and ischemia. That's the story of PRX, as simply put as I can, for this purpose. So they calculated PRX in these patients. They also looked at recovery of consciousness, which they defined as a GCS of six. And they defined that as their favorable outcome. And what they found was ICP and PRX were associated with an unfavorable outcome. But interestingly, with PB202, they were not. But again, this was not a prospective study. This was not powered actually to look for a difference in PB202, which I think still is very compelling based on that original study that I showed you of the 18 patients with the real-time biomarkers of brain tissue hypoxia. The Canada group, again, did another study where they looked at brain tissue hypoxia events. They found that they're very frequent. And now that you know, like, oh, I have a U-shaped curve, and instead of cerebral perfusion pressure, they looked at mean arterial blood pressure. And so looking at that invasive marker of PRX, they were able to show that within this narrow window of mean arterial pressure, my PRX was in a good range, right? So it's an optimal range. And by identifying that nadir in the U-shaped curve, they labeled that as a MAP-opt, an optimal MAP. And when a patient's actual MAP was closer to zero, those patients in those moments were associated with better PB202, better brain tissue oxygenation. So this is, for me, very, you know, not surprising that PRX can yield this optimal curve. It's been shown in other types of brain injury. You can define a lower and upper limit of autoregulation. But there's this threshold. Not everyone here is going to be able to go and put in a catheter in the head. So these invasive measures have been validated in so many different kinds of brain injury for now 20 years. But I think in this point in time, in 2022, we really, in parallel, people are understanding there has to justify the invasive nature, again, because we're not used to doing it yet, even though I think it may eventually show a lot of value. Until we get there, people are also looking at non-invasive monitors to derive the same indices of autoregulation to give the same kind of information. Unfortunately, I think in a very small study, it seems that when you derive a MAP-opt or that U-shaped curve using your intracranial pressure monitor, it doesn't really agree with the one derived from something like near-infrared spectroscopy, which is the next obvious one has been looked at in many different, especially neonates in cardiac surgery, adults in cardiac surgery, but it doesn't seem to really agree. And if we agree, if we accept that PRX is the gold standard for this in an intensive care setting, then this indices from a near-infrared spectroscopy only detects that impaired autoregulation 3% of the time. And this may be a limitation of actually the frequency of the information you can get from the NEARs. There's actually very high frequency information you can get. You can actually start to differentiate deoxy from oxyhemoglobin in your actual brain that you're measuring if you have access to that high frequency information. But these device companies do not yet release that to everybody. Also, I think it's important to know that maybe this study or these types of studies, when they're redone with larger numbers, need to be done with understanding the influence of CO2, O2, and ICP. Not all patients have high ICP, and those are maybe the patients who don't need this kind of MAP-opt. If a patient has ICP, those are the patients who are the most vulnerable. So if you group all those patients in without knowledge, even non-invasively, if a patient has high ICP or not, that might be the way you need to study it. You dichotomize an arm for high or low ICP and think about individualizing your goal. This is the last slide. So people are very hopeful for this in the future. A group of three centers in Canada have gotten together to try to do a feasibility study. This was the first step in TBI that they did as well. We said, well, okay, we have this very complex thing. You have this invasive monitor or non-invasive monitor, in this case it was NEARS. Can you even get that data and do the index at bedside, and can you actually deliver that information in some cogent way to try to direct care? Can you do it? And that's a very important step, and it's not 100%, and no one's surprised by that. So there's some work to do about data visualization and how to, you know, communicate this very complicated thing. I did it in five minutes, but you can't do that for everybody. And so there's some exploratory analysis in this study that showed that maybe there's an association of the index with unfavorable outcome. So these are the takeaways that I have. Very much in the future, but I hope to sort of not close the door on the possibility that one day we could have an individualized therapy, goal-directed therapy, for post-cardiac arrest patients that could do a lot to reduce the amount of secondary brain injury. Thank you.

neuroprognostication

neuromonitoring

post-cardiac arrest

prognostication

resuscitative efforts

secondary brain injury

goal-directed therapy