My topic, our topic, is to talk about curing coma or coma patients and the role of multimodality monitoring. So my specialty is really in comatose subarachnoid hemorrhage, but I think for this audience I also want to dip into a little bit about multimodality neuromonitoring as well. So subarachnoid hemorrhage is aneurysmal rupture. This is a small but important type of a stroke that occurs. It tends to happen in the younger population, the average age of about 50 to 60. And so it represents a pretty substantial burden on healthcare resources because of the accumulated morbidity and long-term functional outcomes that are affected. This is the type of patient who comes to us in coma, a 73-year-old woman. She actually went to a hospital first. They tried to secure the aneurysm, they had to abort. She came to us in coma with her bone flap back on. She went immediately back to surgery, secured the aneurysm, and then she was in our unit for about 21 days. And that's a pretty typical course, 14 to 21 days of a severe subarachnoid hemorrhage. Because during that course, patients are at risk for something called delayed cerebral ischemia and multiple other systemic syndromes. Delayed cerebral ischemia is mediated by multiple things, but most prominently because of vasospasm, where the blood products that are now in a place where they're not supposed to be is quite irritating, various pathologies or pathomechanisms, and then they develop actually second strokes. And in patients who come in a coma, we don't really have great signs that this is happening, no impending signs of neurological injury. So these are the patients that really warrant invasive and non-invasive neuromonitoring. This patient was able to be detected on having delayed cerebral ischemia around day 10, and that was first diagnosed in this comatose patient with things like brain tissue oxygen, cerebral blood flow, microdialysis, and surface EEG. Neuromonitoring really is the crux of what we're able to bring to a specialized care that we can bring to these comatose patients, because what are we doing? Patients are coming in with their primary injury. We're able to define that, the scope of that, maybe even try to even start to have conversations with families about prognostication in some limited way. And then our job is really to try to detect secondary brain injury in real time as it's happening, because if we can do that, then there's potential to intervene early enough that we can save brain cells in time as brain. So what is it? So this is the invasive portion. It's called a bundle or a bolt, because simply because of this white piece here that looks like that bolt that you screw in, that's not as big as it is inside the brain. That's truly just to secure it at the surface of the skull. It's a burr hole and placing fine catheters a couple of centimeters or millimeters into the brain parenchyma. And you can measure things like pressure, brain tissue oxygen, metabolic parameters such as glucose, lactate, pyruvate, glycerol. You can look at electrical activity with a depth EEG. You can use a thermodilution catheter to look at cerebral blood flow. And then on the surface with non-invasive modalities, you have another realm of bilateral or more global physiology that you can detect that Dr. Zafar is going to actually go into a little bit more. And why is it important and why do we feel justified in using it? Well, again, we think time is brain. It's physiologically plausible that if you can detect things in real time as they're happening, you can halt the process by understanding subclinical or silent clinical syndrome that's occurring and then intervene to improve outcome. We're looking for actionable knowledge from all of that data, and there's really few signs of external injury. And so there are lots of things we can do, and they really center around perfusion, detecting electrical abnormalities, understanding whether or not you have metabolic or mitochondrial dysfunctions so you take a different tack in your approach so it's not really just a perfusion problem but an energy problem. You look for influences of damage or primary or secondary brain injury impending upon normal brain structures. And through this complex interaction of systemic and brain physiology, you really get more information from that data when you can integrate all of those pieces of data and time sync it together because not only can you look at that information in a threshold-based way, but you could potentially do things to that data in real time to try to understand the physiology of your patient in real time. This is just threshold-based way, right? So the next section I want to talk about is just one of those things that you might potentially do. So in our field, we have understood for a while that autoregulation is another parameter to understand about physiology could influence in that sort of multimodal protocol your understanding of how will this patient respond if I were to do action X, Y, or Z? Is this patient going to be safe? Do they need it? Can they tolerate bringing your blood pressure up? Should I actually work on your ICP? I really like this study. It's sort of a natural history study, if you will, where they compared these authors from Sweden compared two hospitals, one in Scotland and one in Sweden, and they were very well matched in terms of socioeconomic makeup of the population near the hospital, the rural, urban, suburban makeup of that area. The sort of catchment area was very similar. They practiced very similar types of medicine, used the same kinds of tools, but the one place where they really structurally differed was in their protocol of management of traumatic brain injury. So in Uppsala in Sweden, they're very precise about if there's no evidence to push your cerebral perfusion up with induced hypertension or trying to get anything above CPP of 60, then we don't do it. We keep our ICP nicely well controlled, and that's how we manage all patients. In Edinburgh, it was more reflective of, I think, the general community where we believe that some patients benefit from a more enhanced CPP, that they're suffering from that. You might allow a little bit of an ICP rise a little bit higher because there's not really great evidence that a strict 20 number is really the goal. And so just from that one difference you were able to see, and I want to orient you to these graphs here, where on the y-axis is the probability of favorable outcome, right? So the higher number means you did well, and the lower means you didn't do well. Along the x-axis is MAP over ICP slope. So basically you plot out all of the MAP and ICP data points as they match together across the patient's stay, group them all together, and you did sort of a best fit curve. And that slope, if it was positive, meant that with a little bit of increased blood pressure, your ICP reflected an increase. And what does that mean? And you have a closed system like a skull, and there's brain, there's blood, and there's CSF, or cerebral spinal fluid. For a large portion of that, if you keep on adding volume from edema, from bleeding, from a mass, you keep on adding volume, that system really compensates pretty well for a large amount of volume with a very small increase in pressure. However, on that right-hand side, as you enter a poorly compliant, pretty full space where you can no longer squeeze out the CSF, you can no longer enhance the venous drainage, what happens is at that right side of the curve, even a slight increase in volume gets expressed as a high increase in pressure. And that relationship is taken advantage of when you plot these curves out and you see a positive slope. And if you have a positive slope, that's bad. That means that you're poorly auto-regulating. You're no longer able to accommodate for that increasing volume by vasoconstricting your vessels. You can no longer accommodate to maintain that cerebral blood flow, and you're just seeing this increase in pressure. So it is one aspect, or an indirect measurement, of auto-regulation. So using that, you can see on the right side, that's poorly auto-regulating. So in Upsila, those patients who were poorly auto-regulating and they had a very restrictive CPP protocol, they did better. Whereas the patients who were auto-regulating well, they could have tolerated a higher CPP, but they did not see it, and they did not do well. And the very exact opposite was seen in Scotland. I find this study really elegant, even though it's sort of looking at all the data points, because it nicely demonstrates that there might be potential for individualizing perfusion goals if you can select the right patient. So a group in Cambridge capitalized on this kind of idea. They were sort of doing this in parallel. If you can generate this information in real time, can you then use it to try to individualize your goal for perfusion for patients? So in an index that's widely known as PRX, it's pretty classic, you're basically doing that MAP-ICP slope. You're doing a cross-correlation or Pearson correlation coefficient between the data points of MAP and the data points of ICP in short periods of time that represent the period of time when auto-regulation occurs, the slow wave components of auto-regulation. And you find these gradients. You find these slopes. And then if it's positive range, that means you're poorly auto-regulating, and if you're zero or below, that means that you're doing okay. So potentially, if you're poorly auto-regulating, maybe those are the patients who should not be able to tolerate extremes of CPP, and you should try to stay within the more narrow range. And so what was shown in TBI patients was that PRX, unlike even cerebral perfusion pressure or arterial pressure alone, the PRX was what correlated with their outcome or mortality. This group with Cambridge moved that forward to say, okay, can we do this in a continuous fashion, meaning more to the implementation at the bedside? So they were able to show that that PRX number could now be visualized in a different method. So if you imagine you have a PRX available to you at the bedside, that tells you that in this moment of time, in the last 10 minutes, the patient's data tells me that the patient may not be auto-regulating, or they are, but that might be fraught with a lot of artifact. You have to look at the big picture, so you look at it over time. What this information is telling you is now on the X-axis, you're binning all the CPP values for that patient for some period of time, let's say an hour, let's say two hours. So looking over that period of time, you map out on the X-axis, every single time you had some CPP value, what was my PRX? So then you're able to see on the X-axis the PRX, and if the middle of that line here is zero, anything above it is poorly auto-regulating, and anything below it is you're doing okay. So looking in a purist fashion, you can say the nadir of this U-shaped curve, which happens about 60% of the time if you have adequate data in that period of time that you selected, could potentially be called your optimal range or your optimal point of auto-regulation. And so this was aptly called CPP opt, or optimal CPP. Whether or not that is really true, your CPP opt is something that people are really working out because it's just a number, and the range is probably more important, but you can even start to wonder about the lower limit of auto-regulation and perhaps even the upper limit of auto-regulation. So even moving this implementation science a little bit further at the bedside, they were able to show, with Marcel Aries from the Netherlands, that you could do this in real time. So you can have a moving window of all the information, and it turns out about four hours is about much data you need, and a lot of fluctuation and movement, your ABP will go up or down, and then you'll be able to see what happens to your ICP. And so about four hours above which you have sort of diminishing returns, below which you really continue to get more information, and in a moving window you can get an absolute value of like a CPP opt, move it forward five minutes, move it forward five minutes, moving that whole period of time that you're analyzing the data. What was really compelling here was that looking retrospectively at a population of severe TBI that was managed very much the same, they were able to show that if you have an optimal CPP that's designated at a moment in time, it doesn't matter what the actual number is, and then you look at what your actual CPP was, as opposed to what was calculated to be your optimal, you could subtract the two. And if your real CPP was below, you can imagine you're hypo-perfusing, you're approaching the lower limit of auto-regulation. Above it, maybe you're at risk of hyperemia, at risk of increasing your ICP. So the closer they were to the patients at time, any given time across the entire stay, the closer they were to their optimal CPP, the better they did. The further away the patients were, the more likely they were to die. And the closer they were, closer to the zero, your morbidity actually decreased. Again, this is a retrospective study, not clear if this is just pathognomonic of what patients are able to achieve, what their CPP op and their CPPs are able to be, so not proof positive that this is something that we should do in a prospective way. And so then a feasibility trial was planned, where really the point of this was it's pretty complicated to get to that little red box of integration. You can look at threshold-based parameters, we're used to doing that in multiple boxes around the patient, but integrating that together is really complicated. And so this feasibility was to do two things. One was, can we actually deliver that information at the bedside in an interpretable way to physicians and nurses so they can act upon it with information rather than data? Number two, what were some of the safety features? What kind of interventions were given? Were there more incidents of things like ERDS, et cetera? So but the primary outcome that they were seeking was, can we achieve that goal of plus or minus five of that CPP actual minus your CPP opt? And this is an interventional trial. And they were able to show that they were able to achieve that, and obviously was powered for that. So COGITATE is the name of the trial, COGITATE-2, COGITATE-3 is well underway now in planning. So backing up a little bit, now I've talked to you about threshold-based multimodality monitoring, I've talked about what you could possibly do if you were able to integrate the data at the bedside and look at things, not just the PRX values, but maybe how does my intracranial pressure vary with time as my arterial blood pressure has been going up and down? So this is sort of my approach to multimodality monitoring. So this is a really team-based approach. You need a lot of documentation, you need some feedback, some annotations to look at time series, and sort of looking at the patient in a cycle. So first I look for, is there ischemia? I might use any of those adjunctive invasive monitors, look at cerebral blood flow or brain tissue oxygen or in the microdialysis, glucose-lactate-protein ratio, and try to understand, is this metabolic dysfunction or is this ischemia? And then I might look at, what are the causes of that? Is my ventilation adequate, patient too anemic? What's going on with the patient's systemic glucose? Is there vasospasm, depending on if your patient has subarachnoid hemorrhage versus if they have TBI or if they have some other kind of disease like intracerebral hemorrhage? I also look at, and probably in parallel, is there intracranial hypertension? Is my ICP very high? And then I can kind of cleverly look at the data and look at the patient and say, what's co-varying with that intracranial pressure? Is there something that we're iatrogenically doing in our treatment that is causing harm to this patient? And we can try some interventions and look at the feedback. Then I look at, where is the patient in their autoregulatory curve? I look at sort of the PRX trends in the last four hours, 24 hours, but I also look at that CPP op curve. I don't look at the number. I don't try to seek it out. I just try to see, are they above that line of zero the whole time? Are they nicely below it or only at certain points creeping up? So I'll use that to inform my idea of where they are in their autoregulatory curve to know from my understanding of the physiology what might be a safe intervention to do. And I might intervene. I might do some multimodal intervention, some bundle of care, and then I'll measure my effect and kind of go back and repeat and repeat again. So it doesn't tell me what to do. I don't act upon it in a threshold-based way necessarily, but I use it to inform my understanding of the patient's physiology in real time. Now I'm going to talk to you really specifically about subarachnoid hemorrhage, which is my real passion. So I call this the Goldilocks perfusion problem. I think everybody knows, you know, you have the bears and all of the porridge and some's too hot, it's too cold, and it's just right. I love this picture because it's sort of shaped like a U-shaped curve. But so in subarachnoid hemorrhage, unlike traumatic brain injury, it's not just your lower limit of autoregulation where you want to try to keep your perfusion up. You also really want to know about your upper limit of autoregulation. I mean, I should say you probably worry about that in TBI as well. But in this case, we actually can see where that causes damage. So in this study we did, we looked at CPP, CPP-opt, which is the, you know, a calculated CPP, and PRX. In subarachnoid hemorrhage patients, it was a small group of patients. All these patients had the multimodality monitor placed, so they were a GCS of nine or less. And then we looked at the rates of DCI. We looked at DCI diagnosed clinically or by, meaning clinically by the brain tissue oxygen or their adjunctive monitors or by CT perfusion. What we found was that the physiology prior to delayed cerebral ischemia or that secondary stroke, the CPPs were lower in the patients who had developed CPP. And that was driven mostly by MAP, not by their ICP. Interestingly, while their CPP was lower, the actual numbers, 80 and 90, were plenty in the range of normal. So individually, in front of the patient, we never would have known that that was not enough for them. Their CPP-opt, interestingly, increased. So their demand or their need, you can interpret that, a full day before the delayed cerebral ischemia occurred and even was further enhanced three hours before DCI was diagnosed. This means that the subtraction of the two, the CPP, the delta CPP was much lower in the patients who developed DCI. What was behind this? So I think the most interesting takeaway from this study was that you saw the MAPs were in the normal range. At the bedside, we would not have known these patients needed to be boosted higher. Their CPPs were in the normal range. We didn't know that they needed to be boosted higher. But the patients who ended up having DCI were similar in all ways, except that the group that had DCI had a higher proportion of infection. So you can theorize that maybe they had an overall negative fluid balance that was not detected because all these patients were aggressively treated to be euvolemic. So moving on to adjunctive monitors. So you can look at adjunctive monitors, if they're placed correctly, in a clairvoyant way in the proper side of the brain where the vasospasm or the DCI is going to occur. You can see, it's been demonstrated over and over again, that you can use microdialysis. You can use your brain tissue oxygen to detect when you have sort of ischemia or brain tissue hypoxic events, especially on this lower range over all patients of CPP. And it can be done in a way that predicts your outcome. It happens as you're actually missing the clinical event. I want to draw attention to the fact that while PRX is a little bit more of a global measure because you're based on ICP, these types of adjunctive monitors are really you have to be clairvoyant. You have to have been lucky to have been knowing that your probe is going to be in the exact location to tell you that perfusion isn't sufficient or your brain tissue oxygen isn't sufficient or your metabolism isn't sufficient. And this study by Claudia Robertson Group in Houston shows this very much. This is a TBI study. And all this is showing that even in diffuse axonal injury, which one would consider brain injury, if you lump that in with normal looking brain, the only time brain tissue oxygen catheter seems to be very informative is when it's perilesional. And what does that mean? So in our patients, we grouped it together with a different group at Aachen, have now moved on to Aarhau in Switzerland, and we have a very large set of neuromonitoring data sets in subarachnoid hemorrhage. And we looked at these patients and calculated their optimal CPP. We looked at their CPP opt and their delta CPP. And we looked at and we normalized it to CPP opt. And we were able to show that their incidence of brain tissue hypoxia really jumped up on the lower level, lower limit of autoregulation on the left-hand side, with not really an appreciable increase or information gain on the upper limit of autoregulation. That's what's showing here. Again, you must be clairvoyant. So if your probe is placed in that hypoperfused area where your DCI is going to occur, could be informative. If it's not, so really, if you think about it, it should be around 50% of the time that this is informative. So what we were able to show was exactly that. So in 50% of the patients, it's about informative. And ORX was something we calculated. So ORX is a correlation coefficient of your brain tissue oxygen with your CPP. Interestingly, people used to say ORX is just another measure of autoregulation. There's always been some confusion or contesting of this supposition. And we are one of those groups that never thought that. So we compared ORX with PRX. And certainly, there was no agreement between what the actual value was, the numbers itself. And there's no correlation between the PRX and ORX numbers. But why would ORX be actually interesting? So if you think about the cogitate TBI population, their ICPs tend to be in a much higher scale than your subarachnoid hemorrhage patients. Your subarachnoid hemorrhage patients who are in coma have this ICP monitor. And certainly, some of them have intracranial pressure issues. But on the whole, they have much lower, more normal ICPs than your traumatic brain injury patients. If you remember what I talked about with the pressure-volume curve or volume-pressure curve, you need sort of that information gain on that right-hand side where slight changes in volume correspond with higher changes in pressure. So if you don't have that relationship, your PRX value may not be very informative. I've shown you that your CPP OPT value is informative, but is your PRX informative? And what we found was it wasn't. So your PRX value as a time-varying measure for trying to tell you this patient is going into DCI, and this is all anchored to DCI, was not very informative. And this is separated out by where the probes were. So these probes were the red in the hyperperfused area, again, clairvoyantly determined. The orange was in the not-hyperperfused area in patients who ended up having DCI. And the dashed ones are patients who did not have DCI but had the probe. Interestingly, the ORX did give you this information about half of the time in patients when it was in the right location. It didn't give you wrong information. It was no different than the dotted line, but it gave you additional information. So what is this? Is this autoregulation? Probably not. I think what it's showing you is that, oops, what it's showing you is that the, if you think about it, your brain tissue oxygen, how does it vary as your CPP varies, is really kind of a perfusion or a flow measure, if anything, so it gives you valuable information. Now on the upper limit of autoregulation, what we were able to show in our own patient cohort with monitors, we looked at service CEG, which 100% of our patients, high-grade patients with subarachnoid hemorrhage, receive for the entire time they are at risk for DCI. So we were able to show that when you calculate your delta CPP, the patients who had a higher hyperemic, if you will call it, delta CPP, a positive eight or higher, these patients were at higher risk of developing ictal-interictal continual, continuum phenomenon or seizures. So this is the data for the seizures where your delta CPP was higher and your PRX values were the same. It's a different information and same thing seen with ictal-interictal continuum. I think the biggest takeaway here, though, is that the patients who had those hyperemic events and had seizures and IACs stayed a much longer period of time, but interestingly, they were the ones that did better. So they were more likely to go home or to acute rehab as opposed to a skilled nursing facility from the hospital. So what does that tell me? It tells me that the first work that we did or the later work that we did that showed about the negative delta or the low CPPs is actually really impactful. So if you have a really higher than optimal CPP, you probably do better, but probably too high as maybe you could maybe have an intervenable point here where you can maybe reduce the length of stay if, in fact, those are causative. So this is the basis for a multicenter. This is all the people who are involved in the studies and where we've planned a feasibility trial now for an auto-regulation guided therapy study in subarachnoid hemorrhage, utilizing your auto-regulation index, calculating both PRX and CPP OPT, and trying to inform care in a bundled way with all of our adjunctive information from our multimodality monitors. This is my last slide I want you to take away from here. So why do we do multimodality monitoring? Because time is brain and we don't have anything better. We don't have a great exam. I think this is probably amenable for anybody who is at risk for secondary brain injury. That means all the severe primary brain injuries. Potentially we can, if we can integrate the data, we can try to find understandings of physiology and maybe even do personalized perfusion targeting.

multimodality monitoring

coma patients

subarachnoid hemorrhage

delayed cerebral ischemia

cerebral perfusion pressure

autoregulation