[Music] I am Francisco Flores and I have the pleasure to introduce today professor dr. Patrick burden my good friend and professor Porton obtained his bachelor's in science from Harvard College in biomedical engineering he then moved to his PhD studies to the Massachusetts Institute of Technology where he obtained also PhD in biomedical engineering currently he is a system professor at Harvard a system right associate associate professor sorry associate professor at Harvard Harvard Medical School and it's also researcher at the Massachusetts General Hospital so his research integrates neuroimaging by America's signal processing and the system neuroscience of general anesthesia he's also done a lot of work in developing like all the technologies that the paper was saying that will help further the understanding and study and helping the general population into all these topics of anesthesia and unconsciousness and how to monitor the state of consciousness during clinical interventions so without further ado professor Gordon please yeah I often get mistaken for an assistant professor because I you know look so young but but I'm actually old you know it's just you gotta eat right exercise and sleep well let me um can I take a little poll of the audience how many people here are our anesthesiologists okay we've got a few gynecologists cool thank you thank you for coming and how about how many are neuroscientists okay okay mostly neurosciences all right and then how many are engineers okay to it awesome yeah me too yeah okay so we're in the minority but okay so you know when we talked about the session today we thought we would make it a more clinical talk so I actually I think that can be interesting for the neuroscientists tubes you can literally see how the brain activity in the neuroscience a sort of plays out or could play out in in the operating room so we're gonna be looking at some kind of kind of newer I guess neuroscience principles that we're trying to introduce into the clinical education and then we'll look at case studies actual real-life cases from the or and hopefully that'll be interesting so just some disclosures so we have a licensed patent of Massimo from Mass General I've also receive speakers honoraria from them I'm also an inventor on patents that have been assigned to Mass General that are related to brain monitoring but most of the actually all the support for this workers really come from my department at Mass General as well as the NIH so we've been very fortunate to be able to receive that support and that allowed a lot of us to develop this new framework for monitoring so let me start off with this question so it's gonna be obvious to neuroscience engineers but I think it hasn't been obvious to anesthesiologists you know the question why use the EEG okay so first of all the EEG has a form that relates fundamentally to the mechanisms and brain states that make up anesthesia induced unconsciousness so it's a direct readout of the brain space so that's part one so the second is that actually the form of that EEG signal changes as a the dough so you change the dose and the signal changes so if you put those one and two together then it's clear that you can actually use the eg two in real time read a patient's brain state and just adjust the doses accordingly now as simple and straightforward is that idea sounds it's actually a little bit contrary to the existing framework so Pepe mentioned that that currently anesthesiologists used you know primarily you know cardiovascular and autonomic signals to to monitor anesthesia and then in addition they use these other things called pharmacokinetic and pharmacodynamic models so and this is very typical within medicine so the idea is that okay we can have a model for how the drug goes in and how it sort of distributes across these different compartments you know and especially into the brain and then given some drug concentration in the brain you can measure the response and then characterise across a population of people you know what the response is and then you can later use that as a way of guiding the dosage for the drug so so this eg 50 is the the point in which 50 percent of people say lose their response to say an external stimulus so that's great that's a nice rough guideline but that really only applies for a population you know you have an individual patient that comes in you don't really know how they respond to the drug so you know you're more than likely to be off and actually anesthesiologist put a lot of faith in these models because that's all they've had for most of the history of the field but in fact as you can imagine those models can be off so this is an example of how a study where you know state of the art modeling of propofol an intravenous anesthetic drug was administered icked when patients lost consciousness with increasing levels of propofol and on average they lost consciousness where the model predicted but of course the spread was quite large for individual patients sometimes they lost clusters consciousness at a higher dose than predicted and others at a lower concentration that predicted and that can vary factor of two above or below that the the kind of median dose so you can imagine that any given subject you're likely to be under or over dosed so then bringing this back to the question you know why use the eg well the eg makes it possible to provide like a personalized anesthetic so everyone gets the right dose essentially if we can read the eg so let's just let's just build our intuition as to whether we could actually read the eg because for a long time that maybe wasn't clear so we're gonna look at a case we'll start off with a case study so this is um a 60 year old woman who is coming in for thyroid surgery so she's gonna receive a 115 milligram bolus of propofol so just intravenous infusion very rapidly and um and this is very very typical so let's just watch this video of the EEG as she's receiving this propofol okay so here you can see like little eye blinks there's a little muscle artifact there here's an idling either like I blink I should have turned on the audio because you can hear that her heart is beating and everything so so there's heartbeat heartbeat are starting a heartbeat I believe excuse me oh yeah maybe just turn on the audio you just click that audio there there we go yeah that'll work you can hear a little bit okay so you can see this fuzz here that's electromyogram signaling probably from the burning sensation it's a little uncomfortable when the propofol goes in so so now you can start to see you know the beginnings of an organized oscillation here and every one of these marks is about one second so look at that then something changed really abruptly and then that got a little bigger and so something happened here something differents happening here okay how flat that is that's fine da da wasn't that poor okay so so this is a very typical pattern of events during induction of anesthesia and we're just measuring for leads VG on the front of the scalp it's nothing special okay so let's take a look at UM no it's good it's gotten even more thank you thank you them yeah I should have just remembered to turn that on so what did we see here so we saw to start off with at this point some beta and alpha oscillations so beta being 12 to 25 Hertz alpha 8 to 12 you can see this is kind of slowing through the course of this for second period we next saw superimposed on the Alpha oscillations this slow oscillation and it was kind of abrupt it was very instantaneous you know sort of happened from one screen to the next so that's kind of interesting and then a little while after that we saw these flat periods here alternating with this a little bit of activity so that's a pattern that we would refer to as birth suppression so there's a suppression of activity here where brain activity is completely silenced and then this bursts of activity here which actually is just kind of in a way a version of this okay and actually to go backwards um so in this state we could predict that the patient would be sedated meaning that they could maybe respond to external stimuli in this state we would feel confident saying they were unconscious so they could not respond and regain consciousness after some external stimulus and here this burst depression period is definitely an unconscious state but one that's also associated with coma so we might say that this is kind of beyond what you needed to remain unconscious for general anesthesia so the questions I want to use to frame the rest of the talk are you know how do we how do we do these ways to relate to sedation and unconsciousness I mean how do we know that the the relationship of the waves and the states and then what are the mechanisms underlying these waves so we'll go through that so again to this audience I mean you guys know a lot of this but but just to mention it the EEG is of course generated by a postsynaptic from the cortex but of course the cortex and thalamus are richly interconnected so when we see activity the cortex we're we can make inferences about what connected structures might be doing particularly we know a priori you know how what the connections are and how those connections includes the dynamics and again to this audience and primary neuroscience audience it's not surprising that these oscillations are sort of really crucial or fundamental to organizing brain function so and just to illustrate kind of the relationship between oscillations and and neuronal firing here's an example from crew Nell Ian Hughes showing neuronal and local field potential recordings from the from the thalamus and I think in a cat and you can see here clearly that the neurons are firing periodically at roughly 10 Hertz roughly an alpha frequency and then you can see that the overlying oscillations are you know tightly synchronized to those neurons so when we you know of course see the oscillations either locally or at the scalp we can make the inference that underlying populations of neurons are are firing with some underlying periodicity and of course these oscillations again organize brain function within individual circuits and a class you know larger networks so you can imagine that if we were to introduce drugs such as propofol that has a you know powerful GABA agonist effect so it amplifies inhibitory GABA signaling so we're to do that you could imagine that it would throw off the time constants for these circuits and actually you know impose or impose oscillations or impair oscillations that that might otherwise happen at different frequencies so let's just take a quick look to you know place these things in scale so here's a typical eyes closed alpha wave so you know if we close our eyes we don't need so much the visual system and so that alpha and some respects reflects a idling state for the visual system if we were to continue closing our eyes and fall asleep we might see these slow waves or slowly asleep and of course we know that the slow wave sort of represents alternation between transient down States and up states where the neurons can fire so that and and roughly you know maybe these might be like 50 micro volts in size we like a propofol the oscillations are much bigger this is one of the most surprising things you know when we started doing this research you know back in the early to mid 2000s we were expecting to see you know subtle little changes in gamma frequencies when the when the anaesthetic wouldn't went in and no it was like not subtle we're like whoa what is that big thing we weren't expecting that but but that's what you see you know every time just as you saw in the video and and and it's much larger than sort of the endogenous oscillation so you can imagine the level of impairment you know compared to these sort of endogenous states of brain quiescence are quite significant with the anesthetic drugs and actually in anesthesiology people sort of question you know whether the waves really reflect anything fundamental I think you know but it's becoming increasingly clear to them and I think certainly the neuro scientists that that when you see these big waves it's a big deal okay so so to get at this question of how do the the waves relate to the States you know back back in the day myself Emory Brown and others from our department did this experiment and volunteers where we slowly induced unconsciousness with increasing doses of propofol and then just slowly allowed the propofol concentration to come down and and we asked the volunteers to press a button in response to these auditory stimuli so we're just trying to measure when they lost a responsiveness so there are two types of stimuli verbal stimuli as well as just some you know kind of a simple click train so that was actually it turned out to be pretty important because true that actually the patients are the subjects lost consciousness at very different drug concentrations as a whole so we needed to really measure when an individual at each individual subject lost consciousness so we could take the responses to the difference with the stimuli and then estimate a probability response for each class of stimuli and then we could tell well when the subjects started to perform poorly on the task and then also when they they finally stopped up responding altogether and that allowed us to identify a point of loss of consciousness for that subject and then later a return of consciousness when they started responding again so after that then for every single subject we could line up the data essentially register it in time around these points and then figure out what was going on so I'm lucky I'm showing up you know midway through the conference because you already know how to do what I'm going to talk about next we just spectral analysis so of course the traditional way of looking at the EEG clinically is to just look at those waves and as you can see from the video that's highly informative it's actually really easy in some respects but to quantify what's in the signal and in some sense to have more precision even in the clinical context you know we need to do a little more so so um we need to do spectral analysis and I'm just gonna repeat this just for the benefit of folks who might not have been here on a day one or two of the meeting so him so if we look at this raw EEG trace we can see you know there's some definitely some organized waves there but imagine if we were to break it down into components so we could see here if we knew off priority that these were the key frequencies right we could break it down and we can see one component that has about one cycle per second here okay here's one second and then we might count out these different components here to get ten cycles per second on this other wave but of course this is quite cumbersome it's cumbersome to count the waves and as more and more wave sort of combined it becomes really difficult so it's easier than just to analyze the frequency content across all frequencies using the spectrum so the spectrum is essentially estimating Danah energy or power in the signal so sort of imagine taking the approximate amplitude of the signal squaring and then taking the log that's essentially what we're plotting here in a decibel scale and as so you can see then across all frequencies there's a big peak around one Hertz which corresponds to this rhythm and a big peak here around ten Hertz which corresponds to this thing and that's just that one instant of time so of course if we want to track how this changes through the course of an actual clinical anesthetic then we can you know track this in time now kind of stacking these up in 3d but 3d is kind of annoying so let's put it into 2d with the color scale where now the height of these Peaks corresponds to this color scale where the hotter colors represent Peaks and then the cooler colors represent the smaller values and then we'll put frequency on the y-axis here time and minutes on the x-axis and so now you can see this band here at around you know ten or a little more than ten Hertz corresponds to this guy which of course corresponds to that guy and then this band here to around 1 Hertz corresponds to this peak here in the spectrum and then this this part of the waveform so it all fits together everything we're seeing here relates back fundamentally to the underlying signal but it just kind of gives us this quick snapshot and allows us to see changes you know through time so so we took every individual subjects behavior lined it all up and and sure enough it has this you know really I guess in a way predictable pattern that the less salient stimuli you know sort of are less likely to provoke a response with increasing doses of the drug and then if we look at the EEG the spectrogram lined up around these loss of consciousness and recovery of consciousness points we can see then that prior to loss of consciousness essentially in a sedated state where the subjects can still respond we see this increase in beta and kind of low gamma activity but after loss of consciousness we saw this you know quite strong a slow oscillation and kind of high alpha band oscillation and then at recovery of consciousness we saw sort of things go more in Reverse and Francisco will have a lot more to say about that later in the meeting so so we zoom in on kind of this top part here just like about two Hertz and above and we look at that a little more closely what we see is that with increasing propofol infusion rates there's a sort of this parametric change in the center frequency of this of this band here it kind of just slows down it drifts down from like beta frequencies to alpha and there's a salt also this sort of smooth like kind of narrowing in the oscillation too and then the reverse happens when we reduce the dose and of course we can identify these states these behavioral states so basically this tells us that that you know we have like a really smooth dose response pattern that allows us to adjust the drug to whatever you know state we need so if we needed the patient to be able to respond say if we're doing a short a procedure like an endoscopy or something you could place them in this state I think they need you to be under general anesthesia where they had to be unconscious you could put them into this state and you could kind of regulate that very very precisely just from looking at this from the EEG and from the spectrogram so I'm not going to get too much into mechanisms now because we're gonna cover different aspects of a drink talking I think later today Francisco is going to be talking about thalamocortical mechanisms and I'll be getting a little bit more into the slow oscillation mechanisms but but just to give it a quick overview you know Falls acting everywhere within the brain amplifying GABAergic signaling okay but it does produce these two oscillations in the EEG and through you know a significant body of work we've there's there's a lot of evidence now this is essentially a a propofol induced fronto thalamocortical oscillation that sort of reflects a frontal functional disruption of this tamil cortical circuit and the the slow oscillation represents sort of an exaggerated version of the up/down states that you can see during sleep slow oscillations and we were able to actually get this from human recordings where we could get microelectrode lay arrays and griddle electrode recordings in patients who were undergoing epilepsy surgery so after having electrodes implanted they stayed in the hospital for a week or two to localize where there are peptic seizures were coming from and then they had to go back to the operating room to have the electrodes removed so that's just a you know that's an experiment for free so we just had to show up and we were like little commandos you know they they they wouldn't tell us what was happening but we'd show up anyways and and hook up the electrodes record the data and so we were able to get this very really useful information and and what we saw is that when the slow oscillation appeared which which was right at loss of consciousness you can see you know the oscillations at different scales here from L fps to ACOG and and just very briefly the the neurons were of course coupled to the slow oscillations and you can see that they're firing briefly for a few hundred milliseconds but over that two-second period they tend to be mostly silent so so our reasoning was that you know if the cortex is is silenced in this way during the slow oscillation and then in addition to that you know you have this thalamocortical disruption then that means that when you see this pattern of slow and alpha you're really legitimately seeing a very profound brain disruption so if so behaviorally they're there we know they're unconscious but also at some fundamental neurophysiologic level you know we can be confident they're unconscious - okay cool so I'm gonna spend a few minutes going through some other drugs so you can get a feel for those patterns and we're gonna get into cases which I think will be interesting okay so this is a general anesthesia maintained with sevoflurane so steve-o fluorine is a derivative of ether so it's a halogen ated ether so so really a cousin of the the very first anesthetic from you know back in the 1840s and it has a you know an accent a variety of sights but it definitely has a really strong gaba-a mechanism so not surprisingly it has a very similar eg profile as as propofol so you can see the combination of slow and alpha waves there during this maintenance of general anesthesia and this is just clinical data this is from patients we record from and the interesting that we see those that at four for some reason we're not sure why when you increase the dose of the sevoflurane the alpha oscillation will kind of continue to dip down to into ten Hertz or slightly below and and then we'll see this kind of increasing power and theta frequencies so we call that a fader fill in just just to give it a name and it happens at higher concentrations of the drug above the point where you know they patients tend not to remove too painful stimuli so again we don't know the mechanism of this but but at least it's useful kind of practically in that if you need to go to those higher levels of anesthesia you can you can maybe titrate to that that effect so I'll quickly look at dex medet amma dean so Dex - Homma Dean has a an alpha - mechanism wherein alpha 2 receptor I so sorry zan it's a presynaptic alpha-2 agonist that sort of prevents release of norepinephrine and so as such it sort of allows the preoptic area to become active inhibiting brain stem arousal centers and also their direct effects on the cortex as well so it has this effect that sort of mimics non REM sleep so when we look at the EEG then you know we expect something to look at look like non REM sleep and and it kind of does resemble it so so here's an example of a sedative kind of a lightly sedative dose of Dex meta Tom Adeem you know very typical the thing that you would see clinically and you can see that there are these kind of spindle oscillations here that that are you know transient so they come on and then they go away and then they come on and of course that's different from what we saw a propofol where that you know alpha beta oscillation was quite sustained and you can see in the spectrogram - where the where the spin oscillation sort of has a more staccato form and it's actually sitting at a slightly higher frequency you know around 12 or 13 Hertz so more in the traditional like spindle van Sigma band of activity so the sleep EG person saw that they might you know think that that looks like a non REM - and then in some cases when we increase the dose of propofol this has a slightly higher maintenance dose we can sometimes see these spindle oscillations go away and what we have is just a a slow oscillation so that you know maybe it looks like slow wave sleep so we really think that that deck cement etomidate sedation closely mimics true physiologic sleep and and but the interesting thing is that if we look at it quantitatively using the spectrum if we were to compare say Dex might have met etomidate in blue right - propofol and red we can see that although the EEG has a similar form I mean except for maybe some details with the spindles big the the Dex mettaton manin EG pad waves are much smaller so it's sort of like that first slide I showed you with the different you know endogenous oscillations in the propofol you know propofol induced oscillations are much larger so we think that it has a much more profound disruption of brain function and sure enough that if you give Dex Amanda Tama Dean clinically at those doses you can actually arouse the patient to consciousness so they they can you know respond if you talk to them they'll be very comfortable and they'll be sedated but you can actually wake them up so so we think that difference in the you know ability to rouse the patient is related to the to the size of the oscillations and the level the underlying disruption okay so one more drug and we're gonna see these again in the case studies that's why I'm presenting them so ketamine you know as a of course an NMDA antagonist interestingly at low doses it seems to have an effect preferentially on on inhibitory interneurons so so the the effect is that that by blocking the inputs to these inhibitory interneurons you know you might predict them that that structures that are being inhibited like cortex or the limbic system might then become disinhibited and then it might create sort of an excited state and certainly behaviorally at low doses you you do kind of see that people you know develop loosen ations and so you can imagine that if that sort of inhibitory control were removed they they might activity might be dis coordinated in a way that they could actually loose innate and then in the EEG we see a manifestation of that as well so this is a patient who's coming in for a vacuum dressing change so a dressing change for wound and it's very painful but they don't need to be unconscious for that procedure so low dose ketamine was administered in bolus is here and then you can see that the EEG kind of has this fuzzy excited look and if you do the spectral analysis you can see that that's actually kind of an organized oscillation around thirty Hertz so we see that very typically with ketamine they're also interesting effects when you combine ketamine with the other drugs I don't have examples of that actually but but I'll just mention it okay so so on the clinical side of things you know you know there was this and still is its construct that that will EG it's too complicated you know people can't understand that and and and so decisions were made in the 90s to take all that information and try to process it in some you know empirical way and reduce it down to a single number between 0 and 100 so that made it easy to use in some sense but the problem is that was a drastic oversimplification you know you see that the the drugs are doing you know very different things in ways that make sense so it's it's it's hard to imagine that you could pull all that information just into one one number and I'll show you some other things later that that also kind of go against this concept but I think in a way in addition to simplicity the device developers were sort of responding to the this notion within anesthesiology that there was kind of just one mechanism and it's funny because that that idea of just a single unitary mechanism in a way goes back to you know over a hundred years ago to just the observation that the potency of anesthetic drugs was related to their solubility and lipids so I just mentioned that parenthetically I mean but but I think it goes back to that idea and but in fact I think you know in today's neuro scientific point of view we know that the different drugs first of all clinically have different you know behavioral profiles right so propofol ezio fluorine can produce deep states of unconsciousness ketamine can produce states of loosen Asian decks med etomidate produces this state of sedation from which you can be aroused right without knowing anything about the brain we just know behaviorally to do different things they have different molecular mechanisms and Co Florence probably more gaba than anything else that are working through different brain circuits and so you know we just view the EEG in terms of these squiggles it would be hard to discern the differences but if we analyze that we see the structure of the oscillations really corresponds to the drugs and to the drug classes and and drug mechanisms action so the new approach that we're trying to communicate to clinicians and which i think is probably really exciting to neuroscientists is that the different drugs have different ug signatures that relate back to their fundamental mechanisms and and i think that'll be powerful i'm going forward so one other thing i want to get into before we go into case studies and i actually can ask how long do i have when do I need to okay perfect yeah I'll come in well under well plenty of time for questions so that's one advantage to talking too fast right so so with let's take a look at how this changes with age and if I have time tomorrow I'll try to work into this material more because it's actually really interesting I think especially for the the neuroscientists here so here we have slow and alpha oscillations with propofol for this 30 year old patient okay let's take a look at a 57 year old patient okay so this patient has slow and alpha oscillations also now let's take a look at an 81 year old patient so here's this 81 year old okay not interesting now you have to squint you can see it it's there and I'll I'll prove that to you later the the structures there but it's a lot smaller now this is one's really kind of interesting so this is a 50-something year-old patient who kind of looks more like the 81 year old so that's interesting and then if we go to the other side of the age range we have a three year old slowing alpha 14 year old so often you can see that essentially the size of the signal it seems to just to scale with with age so I guess this points to another perhaps shortcoming of the existing you know one number single number zero to 100 paradigm which is that you know really old people can kind of fool the monitor and so can really young children so for instance in young children this this kind of high-power in the beta range can fool the monitor that was trained on adults to think that actually the patient's conscious so in this state where the child is clearly unconscious where they have clearly the slow and off oscillations that you know we know in adults at least are associated with the consciousness the the numbers will tend to read high so it would tell the anesthesiology oh you should give more drug but in fact you shouldn't right and then the opposite is actually true in these old folks the the absence of a signal that tells the monitor to bring the number down causes the number to read high and then the anesthesiologists are compelled to to give more drug so so it's interesting but if you just look at the spectrogram and then you change the scale the relevant information and features come right back so so this spectrogram is actually you know very useful for accounting for differences in age and then just to go back one more you know and then you might imagine hey there's got to be some underlying explanation for all this it's you know related to say neuro degeneration and development and we definitely believe that and we have some papers on that I don't have time to get in that today but but again I'll try to work it in tomorrow I think you guys will probably enjoy that okay cool so now it's time for quiz okay so I'm gonna show you some spectrograms you've seen them before and I'll give you a second to kind of examine them and then I'll just ask you guys you know to to call them out okay so all right let's start with the easy one which drug do you think this one corresponds to propofol great okay cool how about this one yeah x-men's how many so yeah Dex aren't perfect okay and then how about this one DISA people say ketamine okay cool and then how about this one see you Lauren that's right good okay cool great a plus all right all right so let's take a look at this we saw this earlier in the video okay so which EG oscillations do we see here yeah so so we do see that alpha and then and then what else slow slow Delta yeah that's great so so and then what is most likely the patient state of consciousness unconscious yeah good cool all right all right and then here's another one oh yeah we're gonna get into this one later I didn't talk about this one but actually this is unfair I think I only mentioned it once but yeah what which pattern is this first oppression right okay cool and then what is most likely the patient state of consciousness in a coma or unconscious yeah cool and then the patient is receiving propofol so does this state occur at higher or lower concentrations for whole high okay cool yeah these are just basic things but awesome this is great so we'll see this again okay so now let's get into some case studies so good job guys thank you so let's take a look at this 60 year old patient she's coming in for a mastectomy and she's being maintained with 1.3% sevoflurane so this corresponds to 0.7 Mac so Mac is the the median dose the eg 50 the median dose at which the patients are expected to lose responsiveness to painful stimulus okay so she's at at point seven of that so it's kind of a low dose you know for the population and this is age adjusted - so this is like thought to be appropriate for her age she's hemodynamically stable okay so let's take a look at the monitors so okay Oh point seven Mack age adjusted she's getting sevoflurane oxygen so okay that looks pretty good I'm just an engineer it looks fine to me you know here's okay let's see heart rates at around 52 okay that that seems good you know 128 over 58 blood pressure yeah that's the map of 85 mean arterial pressure that's good Espio 299 so she's well oxygenated you know it's got a good heart rate good blood pressure so okay that's great how hey cool everything's great so let's move along okay so if you look at the EEG though that tells you a different story okay so this patients actually you know just take a look at this so it's kind of first of all the EEG is kind of small not surprisingly because she's little older but we look at this and look there's this kind of flat period here this is kind of flat period here and then this is a slow and alpha period here so she's actually in burst suppression okay and then this number reads really low which generally corresponds to a deeper state right and we can see through the history of this that that it's been low for about you know it's 948 right now nine nine a.m. here so it's been low for a while it's been low for almost an hour 50 minutes or so and then there's other thing indicated here on this monitor called the suppression ratio where which is literally the ratio of the suppressed periods to the total time interval so if this oppression ratio is a hundred that means they're fully flatlined okay so this this patients had some level of birth suppression throughout this whole hour long period so it turns out burst suppression with the according to the latest studies is associated with post-operative delirium it's certainly associated longer emergence time because the patient has more drug onboard than they need so this patients kind of in a way overdosed on the drug even though they're getting less than they're supposed to be getting so and this is something you know that that it's been reported in the literature so this is a paper thousand levin showing that patients who are maintained you know at a kind of appropriate EEG number they actually will still be in birth suppression and that risk for burst suppression increases with increasing age and then two recent studies have shown that the time that you spend in burst suppression during operation relates to the likelihood of having delirium and post-operative cognitive problems so delirium as you may know is this state of sort of a confusion that develops very often after surgery and delirium is related to again later cognitive problems and it's also related to actually increased level incidence and mortality so so it's essentially some kind of you know acute confusional state you know and we don't really know why it happens or what what the the pathophysiology is but it's certainly you know not good and and it's it's of itself as a bad outcome and again it's related to post-operative cognitive problems so you want to avoid this state and you can imagine that if you weren't brain monitoring that anesthesiologist would kind of use the the standard approaches right heart rate blood pressure and and they could could probably contribute to this just by way of being putting the patient's in a burst suppression without knowing it okay so this is a oh this is another case this is interesting one kind of on the other side of the scale this is a situation where actually you would want to kind of maintain the lightest anesthesia possible okay so this is a patient with severe cystic fibrosis so and she's waiting for a lung transplant the transplant so if you may recall cystic fibrosis is this disease where I think there's a disruption in I think maybe a chloride Tran Porter although I I'm a little fuzzy on it but in essence what happens is that that there's a you know a really a buildup of mucus within the the lungs which disrupts breathing and it's actually you know usually fatal at an early age so the fact that she's actually made it to 43 year old years old is pretty impressive so and and it's so bad of course she's waiting for a lung transplant she's so she has problems breathing so in her defaults you know normal state at home she has to have supplemental oxygen okay so that's how bad her breathing is and and uses daily uses bronchodilators to keep her airway open she also has history of chronic pain and has high narcotic requirements so she's she's used to getting lots of anesthetic drugs and probably has a tolerance the drugs and is able to clear them out faster so one thing we didn't mention earlier is that anesthetic drugs all produce respiratory depression so you get above a certain dose of the drug patients will stop breathing and that's the last thing you want for this patient here okay and yet they clear the drug really rapidly so you gotta somehow titrate it and she has the history of anxiety so these are all you know potentially potentially challenging so um I was over at Beth Israel Hospital in Boston and I had the pleasure of observing this case with Fred Shapiro a professor at Harvard who practices at Beth Israel and so he's kind of developed this intuitive skill for forgiving kind of low doses of anesthetics you know he calls it office based anesthesia to maintain respiratory function and not disrupt hemodynamic stability and so his approach involves essentially giving you know small amounts of of ketamine dex medic Tommy might okay and propofol usually on top of a background of propofol and Dex minatom Adeem infusions so he's doing this really intuitively so we went to the OL art adjust you know see what is actually half the brain during this process so this is about a 25-minute period so he starts off the case with a propofol infusion about 50 micrograms sorry 50 microns for 4 milligrams sorry micrograms per kilogram per minute and then he's reducing it slowly during that time period now along the way he's giving small little bonuses of lidocaine propofol in red here ketamine in purple I can't really see the purple oh there's one little purple in there maybe and then Dex Amedda tama d in there and blue so Dex Mentos in here here here and here and and they're really small doses so I'll just tell you right now we're not gonna see the ketamine in fact is I think the dose is too small but what do we see let's take a look okay so actually so I'll give you a hint that we're not seeing the ketamine but but take a look at that well what does this look like yeah like a low-dose propofol right okay so that's cool and then and what does that look like yeah exactly that's exactly what's going on so so essentially what what what Fred figured out to do is start the case with propofol initially because you you want to get going and and Dex minute Tommy takes you know tens of minutes to get started right and then later switch over to Dex minute Tommy that was his his idea and you can see he keeps it in the sedative range where the patient is sedated but but unlikely to get into respiratory problems so so that was kind of his idea and so it's interesting to be able to kind of in a way decode what's happening just from looking at the EEG and now once we know that in some sense we can replicate this you know more easily more precisely okay here's another example so want to make sure I'm on time here yeah we're good so this is a case that emery brown did and it's sort of following a new approach that he's he's doing that he refers to as a multi modal anesthetic approach where he puts a lot more attention on on maintaining anti nos deception so essentially controlling pain so he's giving three different pain medications is sort of influence the pain system through three different mechanisms so Remi fentanyl block blocks of MU opioid receptors ketamine of course is an NMDA antagonist that that also works at a spinal cord level to to block nos septic stimuli and then dex med otama dean has this alpha-2 mechanism which also has an ID to anti gnosis if it effects so he's kind of blocking pain in three different ways and then what he figured out is that well he could actually run the propofol at half to a third of the dose that he normally would have to run to maintain unconsciousness so so okay so the vitals look good I won't go through that receiving no and inhalation last anesthesia and and respirations fine I guess and then if you look at the EEG so you see this really nice alpha wave and you can see like kind of a slow oscillation coming and going there and if you believe the number she's been just in this in the state you know really really comfortably for almost an hour and probably before that so with this combination of drugs by by you know controlling the pain well you can actually reduce the amount of anesthetic you need to maintain unconsciousness propofol and and and kind of balance out the anesthesia a lot better and and by reducing the amount of drugs you give you can actually allow the patient to wake up a lot faster at the end and more comfortably so here's here's one more so so it's a it's a question in the field right now as to whether you can use these monitoring technologies and children and especially you know III think actually the the consensus is that you know the monitoring doesn't apply to children this is a slightly older child but but I think it illustrates the point so this is a 19 year old male patient with severe autism so he had a history of aggression and behavioral issues he's large also zip so he's a big boy he routinely takes antiepileptic medications so maybe the thought would be you know he's he's able to clear the drugs faster than you know a typical person so he's in for a sigmoidoscopy for chronic constipation so they actually have to just help clear his constipation and he has to go under general anesthesia for this so so he has to be unconscious and and he has to you know have his pain control so before inducing general anesthesia they gave him oral ketamine I actually don't know kind of if that's a higher low dose because I don't have intuition for how the oral doses go but he also received him midazolam point2 so I think that's that's yeah that's pretty typical kind of on the high side maybe and and then for induction of anesthesia he received one milligram per kilogram of propofol in a bolus and then and then he had a purple hole infusion of 350 micrograms per kilogram per minute so this is actually really high this is maybe you know more than I would say double the typical dose for for a adult maybe for a young person it would be double the dose I don't know so what I'm happening is that he had a delayed emergence for two hours so in these clinical cases you know except for the ones where anesthesiologists on our team err or manage them we typically just record the case especially in the pediatric cases and we don't intervene because there's no you know kind of guideline for how you use the EEG so we just record and we're using it as a way of studying again we can't influence the management in these recordings but I kind of wish we could have because this is what we saw so this is time and seconds here so we're looking at you know maybe about half an hour of data here so at the outset after inducing so he's coming in here with this background of midazolam and ketamine right around here the propofol bolus and infusion starts the colors all saturated here but you can basically see a strong alpha star alpha and slow and there's a little bit of theta in there too I think we're just saturating the screen here and and so then the infusion of course is started right around here and what you can see is is that the screen turns blue okay and then you zoom in and so there's blue you know so silence a burst of activity bloob respectively blue bursts of activity and so this and then blue they're on out for about 20 minutes so this patients actually isoelectronic flatline on the EEG you know totally totally suppressed for about 20 minutes so the clinicians didn't know this because they weren't looking at the monitor but you know when we went back and looked at it you know that's what we saw and actually one of our med students was there and so they actually you know usually she's just kind of a fly on the wall but they're like oh my god you know come back in and monitor this patients brain because we don't know why he's not waking up well it's not waking up because they gave way too much drug during the case so it's just an example of how the simple introduction of this EEG monitor could make a lot of difference for a lot of patients we think so in summary you know you know we really believe the form of the EEG relates fundamentally to the mechanisms to the drugs and to the underlying brain states that comprise general anesthesia and sedation you know these oscillations aren't subtle they're really huge quite easy to see the spectrum lets you easily visualize the EEG but also lets you analyze it as well for the neuroscientists here what we found is that the different drugs have different signatures that relate back to the underlying drug classes and mechanisms with aging and with development the signal changes in size but the underlying structure remains so that we think that you can use this across a broad range from you know quite young children and say about one year of age or older all the way up through of course elderly adults if you just look at the signal and then finally the EEG provides a real-time readout of the individual patients drug response you can titrate it for every individual subject you don't have patient you don't have to rely on a pharmacologic model and you know this allows us to give personalized anesthesia care to patients so I want to thank my collaborators you know I've worked with Emory Brown for a long long time on these studies I've also worked with Shannon okay jus to characterize the EEG in these different brain States and I want to thank all of you and and the organizers a Francisco Pepe Karolina for for putting this all together and for inviting me you know especially there were some travel delays we had to contend with so I really appreciate you know making adjustments and and thank you all again for for coming at this time we have time for questions yeah I'm happy to take [Applause] most of the people here are not an essential register they may not know that anesthesia is basically a combination of a lot of drugs or not a lot but usually more than one you have shown that you can monitor combination but do you have a have you ever tried to figure out there's a splitting pattern of certain combinations because okay propofol has a clear pattern ketamine also but in the world they're combination of groups that all affects the brain in some way so I don't know if you're trying to deal with that in your studies yeah so so and actually in part in in part of this online program in the there are two modules right now that are online the second module there are lots of examples that but I'll just tell you verbally that yeah in most cases actually the combinations sort of makes sense so if ketamine you know has this you know 30 ish Hertz component and SIBO has slow and alpha then when you put the two together they actually kind of have almost some hybrid of the two and then interestingly with both SIBO and with propofol there's and when you combine it with ketamine there's this interesting effect where you know the the the the propofol you know gaba induced of alpha oscillation you know requires gaba to actually make that oscillation right so when you block the inputs of the gaba interneurons with the ketamine you would expect that that would go away and it does actually at a high enough dose you can actually see the alpha oscillations go away and get replaced by the bike by gamma when you're combining propofol and and ketamine and the slow oscillation remains underneath a similar thing seems to happen with ketamine in combination with dex and then for the for the the drugs that tend to produce like a slow oscillation whether that's you know the opioids propofol Dex so when they're combined they seem to all just add up and make more and more slow oscillations and then what happens the frequency's is you know either not visible or a little unclear so in most cases it actually kind of makes sense as a hybrid and and they're examples of that on the website and we're going on to characterize that in a little more detail in the future oh thank you I have two questions related to your work with propofol the first one is you've you've mentioned at the beginning that there is a high variability between subjects with that relates those and and the state of consciousness is the unconsciousness state that the individual is yeah so my first question is what about the EEG the relation the individual because you shown group data yeah so I was wondering how is it if you also have a high degree of variability individual variability that relates the EEG signal and the state of unconsciousness that that individual is in and the second question is you've been interpreting your data saying that you have an increase of a frontal alpha but alpha is not a rhythm that we do see in the awake state in frontal electrodes so can we interpret that as an increase alpha that I mean we don't detect frontal alpha or can we just look at this data and and and and think that it might be just a decrease in the frequency of beta oscillations of frontal beta oscillations so it's the same mechanism but it's just reducing the frequency of the beta oscillations instead of having an alpha a frontal alpha yeah I hear you're saying yeah so just to answer the first part it's you know although the dose response relationship for individual subjects is different the nice thing is that the EEG patterns are so stereotyped they're so sorry baby we have like you know thousands of cases now of EEG you can you know we could go to an or now put on the electrodes we'll see it I mean it's it's very very very stereotyped the the individual frequency like and and these are these are things were teasing out like though like exactly where that you know what I'm calling fertile alphys it's you know may vary from patient to patient the exact sizes the signals you saw will vary according to age and and and development but but the form for propofol it's so stereotyped it's really written oh and it relates to the state of consciousness know it's its commits it's tightly locked to the state of consciousness yeah and then to answer your other question know that that's totally a valid perspective completely and I think that it's really just a a matter of nomenclature we were you we chose the the term you know frontal alpha because you know alpha usually refers to the the frequency band right we were thinking of the frequency band you know 8 to 12 Hertz and its frontal but clearly it's a different in a way a physiologic mechanism than say the the posterior alpha so so that that's totally clear and I guess Francisco is gonna talk about there's a lot more in his talk which later today right yeah so so actually I won't ruin that but but but essentially you're you're partly right and then and then he's gonna he's going to show you the details that yeah thank you buddy that was really brilliant um I have a question so the patterns that you're showing us are pretty distinct right and this is that you're showing us are pretty distinct for the spectrogram right so this is a wonderful case where you can use a classifier to give you one number to tell you where the patients in which drug the patient needs and probably which doses that patient's having right and that would be fairly good for the doctors because he says one number right so why you see for us is fairly is fairly easy to look at those waters right but doctors you know may not have time to do that so why not try classifiers and just give one number yeah I think the reason is you know ultimately they're there I think there could be a classifier that you could use to kind of do the job properly so first of all at present the job is not being done properly so you know I think the when they started off trying to build these devices in the 90s they didn't have this insight that you know age drug were the key covariates that the governed the eg and so as such they didn't know how to structure and build the if you will algorithms to to account for those variables and now certainly having presented it it's clear and one could construct an algorithm but but I actually think the you know the primary reason for for not doing that is that it's to anti-intellectual I mean really like like the doctors they they they're experts in physiology like they can look at the EKG you know see you know small subtle things interpret that they can look at your respiratory waveforms interpret that they can have you know pulmonary catheters and every point they could just look at the waveform oh yeah I meant this part of the you know pulmonary circulation I mean they have this exquisite knowledge and it comes into play it it ends up being you know clinically useful so I think for for for the profession of the the specialty of anesthesiology it's really important just to have that kind of installed as part of the knowledge base I think it it gives for that feel that gives rise to too many things that you can do to improve the anesthetic down down the road insights that you wouldn't have if you didn't think about the mechanisms and and the dynamics and then the other you know super practical thing thing is hey you know machines they're they're prone to error you know they sometimes they sometimes fail and and when they give something that that's odd or not quite right you need to go back to the fundamental so if the autopilot in the aircraft isn't working you know the pilots gotta be able to take over if the you know instrument guided landing system isn't working pilots gotta take over so they still have to be able to have that facility of the signal I agree with you what I was trying to think of is you know like you might think that you have a case that you don't know before right you can take a an EEG and try to predict what kind of you know drugs you should get you should give to that person so now do you actually get into the diagram of prognosis yes you can actually do things that otherwise you can't because you know like now you once the subjects is under proper fold you can tell okay he's under this amount of propofol but what you would really wanna achieve is I don't know who you're gonna be but you know based on your eat Barney gee I think that this is who you want to be yeah and we're we're definitely working on that I didn't have time to show data on that but that's especially with regard to aging and development we definitely believe that so there there should be it needs to be personalized on that level and there's you know essentially fundamental you know brain physiology pathophysiology that should be able to tell us that we're working on that I'll try to work on that in tomorrow because I think you guys will find that interesting and I think you're also pointing to the notion that that you know you could really personalize care for an individual patient if in some sense this dose response information we're at part of their medical record that's a whole other can of worms to deal with but yeah these are great thoughts and and and clearly yeah really important yeah yeah so the thing about post-op delirium you know it's bad and we're all well talking about that a lot lately so but he said that faces with post-op delirium will have more complications you see the chicken the patient was in a poor state and he he did bad he had the lis room because he was he had complications he was in very bad shape or do you believe that because he had the lyric he will have these complications right or the chicken yeah yeah I think right right so essentially I think both cases are probably you know possible so so in one instance it's possible that no matter what you did the patient was going to do bad because they had a pre-existing you know sort of if you will you know brain problems so so no amount of like just the the the surgical inflammation would have been enough to cause further compromise and then result in their in their delirium so so I guess they're definitely a categories of patients who are so fragile that that would be true on the other hand if you know it's also possible that that things that happen during surgery including the anesthesia and the amount of anesthesia could contribute to that to the to the delirium so so I guess the idea is that in the in the beginning you have this brain that has you know a degree of frailty and you don't know you know what it is and then the combination of factors that are happening you know surgical inflammation anesthetic the the ongoing you know kind of deep level sedation I see all those things can kind of chip away at the brain reserve I think and then at some point results in in in delirium I'll point to also in the in the fruits paper that that there they do show this kind of relationship between increasing time and verse suppression and then the the delirium postoperatively so suggesting that you could turn down the dose and at least reduce that that influence so in a sense both the chicken and egg are true I think okay thank you and also you have any you know hope you heard about the the use of general anesthesia and children under two on problem with development and that's been a huge topic in us they had a warning installing generals this yeah I don't know if you know that yeah no I do that so can you tell us something about this oh yeah I definitely know that because on Sunday I was at the IRS international anesthesia Research Society and and they picked us to give a panel in the morning the FDA guys were there and and and so yeah I definitely know this so and we've been thinking about it let's see so I'm gonna try to work this into tomorrow's thing too but I wonder if I don't know if I should I don't know who's gonna be there tomorrow so let me just I'll just give you the punchline now so check this out so so the the existing framework for that field is that okay it's neurotoxicity right and that works with the FDA's approach right so hey we're gonna you know in an animal model you know if you give a certain amount of a drug it it you know kills neurons you know it produces apoptosis and then that is associated with you know cognitive performance issues and that would later affect the child that's kind of the hypothesis and least in the animal models there's lots of evidence for that in the clinical studies there's mixed evidence and way I break it down is to say that in in the clinical studies where they're looking at specific if you will neurophysiological principal measures of brain function you know something specific like recognition memory or or or maybe even learning disabilities on specific modalities you see in effect if you get a little higher like more amorphous with say like Bailey scales or or standardized test scores the effect doesn't really show up and interestingly one positive news is that that children who receive anesthesia for less than an hour at less than one year of age don't seem to have an effect when they measure say Bailey scales at two years of age so that's good news but but what what I think's happening is that a lot of the clinical studies that show no effect are looking at measures that are too coarse to pick up true developmental effects that can be sort of masked by the coarseness of the measurement or other factors of plasticity that that can you know make up for whatever deficits are introduced early on and here's the key thing this is the bombshell this is the bottom gonna drop so these you know fundamental I guess mechanism of brain development and children are these critical periods of plasticity so you may remember like say in the 30s there is that experiment where like a goose hatches from an egg and then within 24 hours if they see a human they will imprint on that human and just follow the human around as if it were its mother that's the first example of a critical period a window of plasticity that opens transiently and then closes other examples would be in the first year of life you know you can acquire like language phonemes right and then at about nine ten months of age you lose that so so for instance in in in Chinese you know Cantonese they're like nine tones right so if you grew up listening to Cantonese you could hear those tones and then but if you didn't you would have a hard time in adulthood or even after 10 months picking them up in Spanish I think V&B have a very similar sound so you you know in that first ten months if you don't hear V and B you'll have a harder time later on hearing it right and then in Japanese like l and r you here LNR in the first 10 months of life you can't distinguish it later on right so that's in that first 10 months so that critical period those critical periods all of them guess what they're opened by GABA signaling so I'll just stop right there so that's our hypothesis so we're trying to chase that it's been hard getting funding for it but we're trying to get after it Hey thank you for the talk very nice I wonder which is the comparative importance of the reduction in the a chi the height oscillation frequency oscillations in comparison in comparison to this Lobos relation and I wonder why you're asking this because in our experiments in animals are trading cuts a under a cholinergic muscarinic and that one is like a scopolamine you have an ana slow wave a oscillation very similar to non REM sleep slow asleep and the animal is fully awake but the animal maintain in any way they make the high-frequency oscillations and I wonder which is the portance comparative of the reduction of Chi frequency in comparison to the increase in slow waves for the unconscious process yeah I think the increase in the soul oscillation is way more important I mean and I haven't studied you know I hope you can tell me like the details of it for the scopolamine case but I think the Salah solution is way more important because essentially with increasing doses of a drug you know you can make that slow oscillation larger and larger and essentially make the the down states more and more prolonged and I think when those downstate states are really prolonged you know the cortex just can't be active and that's you know a more powerful mechanism than say the the alpha one which seems to only you know involve like the medial dorsal thalamus and frontal Mito prefrontal cortex yeah but the slow oscillations at least with propofol it seems to be everywhere and so you have a much broader you know kind of cortical disruption yeah so I think Schloss H is more important thing thank you very much Patrick I I am also Nesta series and you know you and now I am regarding eg also in during the surgery and for me it's very when I habitation with perspiration I old I always think that oh what happened with the brain with the patient so has a negative consequence immediately in the during the perspiration or not and and also I know that after that the patient has more probability to to have delivery for so perfectly but between the lyrium and the suppression they are I don't know there are many change that can occur in the brain for me is a really a to know what happened with the brain during perspiration and after what happened to recover again the the normal rhythms I don't know what do you think about this period when the patient has a perspiration and the consequence with the neural function honor and connection currents and even though right yeah what what's the mechanism how do you get from there to delirium right yeah I don't know I mean I guess one you know one thought could be that I guess there's a lot of evidence that the that the anesthetics themselves are just inherently neurotoxic so it could be that we're just crossing over some threshold or or or you know that with the increasing dose you we start to see more of the neurotoxicity which which then you know manifests itself down the road as as brain function aka delirium so that's one possibility so another possibility though is that you know you do see you do see birth suppression in stroke as well so if you you know in animal models if you you know provoke ischemia you actually see birth suppression as a result of that so it's possible that that you know we may be on masking actually underlying you know neurovascular problems and that that what we may be seeing is sort of a you know some deficit in either local or kind of diffuse brain perfusion which then has consequences down the road as well so yeah it's not really clear we're trying to study it we just got a grant that will allow us to do some neural imaging so we can kind of correlate you know Nerja january if you neurodegeneration or maybe neuro vascular problems with you know what we see with per suppressing or so hopefully we can get into that a little bit more but it's it's a great question and you know i think a big question for the field i forgot to thank you perfect very nice presentation but you have described the transition from consciousness with loss of consciousness and at least in proper fall very clearly in it happens very fast that's something even as an assessee ologist we didn't know basically most of the time we use a huge bottles and the patient's go down very quickly but if you even if you do it slowly like you did in in a very short time you get from conscious to unconscious but in adding with antony what antonella's asking have you ever have you have an idea or in an actual electrophysiological parent of the transition from i would say good and a seashell good deep group level of unconsciousness dude to burst depression is fast slow or that's best depression appears in a transition way or or in a in a quick way like loss of consciousness in an idea to to predict that in during the the anesthesia were going too much deep and we're may be able to to I don't know good less drug or something yeah sure okay so so um I think in general it the transition is is his pup is probably gradual because in certainly a 2013 study we had a few patient that actually did go into subjects did you go into perspiration and when we saw as before they went into recession they had this particular phase amplitude modulation pattern which was different from what we were seeing during induction so we saw this essentially if you will almost like a an upstate a pattern where where the Alpha oscillations are highest at the peaks of the salaah solution and it was you can actually see it by eye you have to kind of get the montage right but you can see by eye and the waveforms and you can certainly see it you know with some analysis so so that that seems to come before birth suppression now the question is you know does that happen for all patients I think it's probably pretty consistent for for younger healthier patients but the impression I get from you know looking at data and from talking to Emery from his cases is that for the elderly patients it's it's not so so gradual they seem to kind of just like dip into birth suppression you know really easily so and again I think that may relate to some underlying kind of frailty that they have so so I think for them it could be could be more difficult but but I think in general yeah it would be possible to identify a pre burst suppression state and and yeah so so I think it could be done and it would help you know again guide titration alright hey thank you so much nobody's here all right great thank you all thanks so much [Applause] [Music] [Applause] [Music]