Chapters Transcript Video Precision Medicine for Neurologic Trauma Robert D. Stephens, M.D. presents at the Johns Hopkins Department of PM&R’s Grand Rounds on February 21, 2017. I need to speak on the thing to be in there. So we need to, we need to speak in the microphone uh pro and some. So a couple of uh logistical things I want to say. Uh uh And then we uh start, but we like uh I think I mentioned in some meetings but not to, I'm gonna keep repeating. Uh We change the way that we are virtualizing these grand rounds from uh before we have a, a system that allowed for interaction. Now, we have a different system that is gonna be recording and people get a, a web site link that uh f has sent around. Uh so they can watch this from uh their uh computer desk um that I think works better. It's more resilient in terms of uh getting the, the actual lecture and so on, but it will not allow for communication or for questions. So if there's people uh listening to me right now in the computer screens, uh they want to ask questions to rob, they can go ahead and text me my cell phone or call me on my cell phone. I will pass the, the questions to to Robert. The, the thing is uh the, the people who are in the, here in the room who want to play in CME, they need to follow this procedure, the people who are on the computer, uh whatever you guys are, uh you also need to do the same uh mechanism. And lastly, all faculty should also sign uh when they walk into the room. And if you are listening to this, you should send an email to Peggy that you are listening to this. That's it. Thanks. All right. And also if you want to claim the CME credit, you need to, uh these are only for people who are doing it for CME. Um you need to text the 9720 to the number on the screen. So, um hopefully that system is working well. I just tried, I haven't heard back yet, but, um uh, so for today's grand rounds, we're thrilled to have Robert Stevens. Uh Doctor Stevens is an associate professor in almost every department in this hospital. Um He's an associate professor in neurology, neurosurgery, anesthesiology and radiology. Is that correct? Got him off. All right. Um And uh one of the attending physicians in the NCCU uh and uh very involved with that and um has a strong research interest at uh in uh effects of traumatic brain injury and he's gonna talk to you about that. And um, I think it'll be a great talk for all of us So with that, I'm gonna turn it over and, and go from there. So I wanted to thank uh doctor Mayor doctors for inviting me here today. I, at first when I walked into this room, I was having some second thoughts because I thought the department of G and I wasn't quite sure if I was in the right place, but it seems that I am because I see very many, many familiar faces And um in fact, many of the people that work with me on a daily basis in the NCC taking care of uh these um very um fragile vulnerable patients. So I wanted to talk about uh something today that is kind of a fashionable topic. Um And uh I think it's something that has generated a great deal of excitement and it really potentially represents one of the future faces of medicine uh in terms of how we practice it. Um And I, I'm clearly an enthusiast, but uh if you guys have questions or if you have skepticism, please feel free to interrupt me and ask questions. And I'm hoping that those who are looking uh through the simulcast are able to hear my voice and also ask questions perhaps by calling or texting uh doctor Sel. So I should acknowledge that um a lot of this uh work um is the result of a task force that I have the privilege of sharing for the Society of Critical Care Medicine uh for the past year and a half, we've been meeting on a very regular basis to try to delineate a research agenda regarding precision medicine in critically ill patients. Uh My focus today is gonna be on precision medicine as it applies to, to traumatic brain injury. But I did want to acknowledge a significant amount of um support that I received from the, from the SCCM uh in terms of this uh thinking uh through these different uh ideas. So I don't know how many of you are um fans of soccer. I, I certainly am. Uh And I don't know if you remember the FIFA World Cup uh just a few years ago um with a match that involved the team from Germany. Um And it was really a remarkable game actually because one of the German players was uh was actually knocked out uh during uh during the match, I think he received uh somebody's knee in his head and he became unconscious, was lying on the field as shown here. And for more than 10 seconds was really unconscious. Uh And then slowly came around and started sitting up. He was briefly seen by his teammates by the team doctors and they deemed that he was fine and he continued to play for 14 minutes. Actually, I don't know if you guys remember this, but for 14 minutes, he was staggering about the field, clearly disoriented, not able to coordinate his playing and his walking. And finally when it was obvious to many, many people that he was not doing well, he was taken off the field and he was, of course sent for further testing. Thankfully, uh This was just a severe concussion and no lasting consequences and this player has done very well since, but it just gives you a sense. This is a game that was broadcast to millions, maybe billions of people across the world. Uh And they let him play for 14 minutes after sustaining what was clearly a severe neurological insult. Nobody did anything about it. And I'm just, just sort of highlighting this anecdote as an example of something which is very, very common if this could occur in the World Cup in front of billions of people. I can guarantee you that it's happening every single day in soccer matches around the world and it's happening also in civilian life. It's happening in the military. It's something which is really, really common and we think about the epidemiology of TB I the best guess is, and it's a guess because we don't really know what the true magnitude of this problem is. But the best guess from the, you know, Center for Disease Control is approximately 2.2 million emergency department visits. About 252 180,000 people that are hospitalized each year with TB I, 50,000 deaths. And that death rate, the mortality of TB I is declining. But the real question is how many people actually sustain a TB I, um, and I, I'm sorry to say that we don't really know, but the best, you know, sort of guessimate is that about 3 to 4 million per year in the US alone, 1% of the population, uh, is involved in some kind of traumatic brain injury. And when I speak about traumatic brain injury, and I'm hoping that this is obvious for all of you. Um We're talking about not necessarily only a mechanical impact, we're talking about uh injury to the brain that is sustained by a very sudden acceleration, deceleration by a rotation or by a blast, right? So you don't necessarily have to hit a surface to sustain a traumatic brain injury. Um So 3 to 4 million in the US, if we extrapolate on a global scale, and these are estimates again from the World Health Organization, we're talking about maybe 60 to 80 million cases of TB each year. So this is not a rare condition. And it's notable that, you know, between 85 and 87% of T BS are concussions. And another notable fact and this is where the epidemiology becomes really challenging is that between 30 50% of T BS actually are not evaluated medically in the acute setting. Some of these patients eventually come back to for persisting symptoms. But the fact is that this is not a condition that immediately leads to medical consultation in many, many cases, 30 to 50% of the time. And so if we compare, for example, in the US, you know, the estimate of 3 million TD isa year, this is almost twice as many cases as the incidence of cancer. Of course, the the mortality, especially with cancer is higher, but it just gives you a sense of the magnitude of this, of this problem. And the CDC again believes that between five and 5.5 million people are living with the sort of chronic disabilities, long term functional disabilities associated with TBI. This is 1.6% of the population and again, difficult to estimate entirely. But the costs associated with TB are in the region of 100 billion indirect and indirect expenditures each year. So given the importance of this problem, uh you would expect that there would be, you know, a huge amount of funding for TB I research. In fact, um the funding for um TB I is actually quite miserable and abysmal compared to uh the funding for many other conditions. This is just an example of all the funding for the uh National Institute for Neurological Disease and Stroke. So not just TB I but TB I is just a tiny fraction less than 5% compared to funding for Cancer National Cancer Institute. And you can see that it's a 2.5, 3 times more money is being allocated by the NIH for Cancer Research than to neurological disorders. And yet the magnitude again of TB I is which we could consider potentially one of the most, if not the most common neurological disorder is uh very, very low at this time. Why is it important to fund TB I research to support? Because as you guys know, uh TD I is uh not an isolated event, it typically leads to a long term uh neurologic physic physical emotional um consequences. So, uh many of you are familiar, of course, with the sort of post concussion syndrome that can be uh quite disabling. A subset of patients with TB I can develop seizures. There's a known association between TBI I and movement disorders like Parkinson's disease, motor neuron disease. A LS is linked to traumatic brain injury. We know that traumatic brain injury can set into play a neurodegenerative process which can lead to accelerated dementia. Um There's this well defined Tey, which is called chronic traumatic encephalopathy that has been described in uh you know, retired NFL players and other retired athletes, um which seems to have a much delayed onset with the development of a neuropsychiatric syndrome, often leading to uh you know, severe depression and suicide. Many patients with um TB I go on to uh have uh psychological problems including anxiety disorders and mood disorders. So, this is not a benign um uh condition. And the question is, you know, can we do a better job of evaluating TB I and treating it. This is just a one slide as an example of uh the potential cumulative effect of aging and TB I on the risk of dementia. This is a cohort of, um, us veterans. And you can see that, uh when you compare those veterans who have no history of TB I, with those who sustained the TB I during their time, uh in the, in the service, the ones who did the CD I have an incrementally higher risk of developing dementia over time. Um And this is being described, a very similar relationship has been described also, for example, in players of contact sports, soccer players, football players, those who have sustained one or several TD I are at higher risk, um you know, to develop dementia. So one of the biggest problems with the TV I is it's, it's not a disease or it's a disorder, it's kind of a syndrome. Um and it's extremely variable in presentation and also extremely variable in terms of its um its trajectory, right? So uh this slide just illustrates this, you, you have um on the, on the Y scale, you have the Glasgow uh outcome scale extended, which is you all know is the functional scale used often used to measure outcomes in TB I. And you can see that this report followed over time. Each one of these lines represents an individual subject. You can see that there's an extreme amount of variability you could start for example, initially, with the G OS C of, uh let's say three and any range of outcomes is possible, there's a huge amount of uh diversity or heterogeneity in the individual trajectories of TD I patients. Um And the fact is that we're not very good. In fact, we're quite bad at the outset when the TD I just happens at discriminating uh which patients will have a favorable trajectory versus a less favorable one. And so the reason I'm bringing this up is because I think that precision medicine um uh sort of thought of globally could potentially represent a solution to this problem in terms of predictive models. Prognostic models uh used in TV I uh in moderate and severe TB I, the best known uh are the impact score and the crash. Uh These are two Multivariate models uh that essentially combine clinical characteristics and characteristics observed on CT scans. So clinical and CT sometimes uh some biochemical or physiologic uh variables as well. Um And these models, you know, are currently the state of the art, these are the sort of industry standard standard in terms of clinical research on T and B I. The fact is that, that they, they do very poorly. So if we wanted to measure how well they discriminate between, you know, favorable and unfavorable trajectories using an ROC analysis, um you know, the, the the area under the ROC curve that we're getting is maybe 70% 80% which means that 2030% of the time when we use these uh these multi variable models, uh we actually are completely wrong. We, we are not able to accurately predict or discriminate between outcome categories. So clearly, we can, you know, we need to think about better ways to do this. Uh This is also um a depressing slide because it details um a number of very well known large scale multi center million dollar randomized controlled trials, looking at different interventions, neuroprotection, hypothermia to potentially ameliorate the outcome in moderate and severe TBI. And uh you know, this is research carried out over 10 years. But and there's many, many other studies that I could show you uh the fact of the matter is that there is actually no effective biologically targeted therapy for TB I, there is no treatment for this condition, right? So we're talking about something that affects maybe 1% of the population. We're talking about a high potential for long term neurological and psychological and functional and problems. And yet there is no treatment, right? We provide supportive care. We are quite good. For example, in the intensive care unit, we treat brain swelling, we can do a lot of different things, we can resuscitate, but really, we really don't know how to fix the traumatized brain. And I think this is a major challenge for us which brings me to this question of precision medicine. And you know, what is precision medicine I think many of you have already thought about this. I read about it. But what we're talking about essentially is data. We're talking about integrating multiple levels of data, molecular data, imaging data, phenotype data. And using these in um sort of iterative computational models to arrive at a better classification uh of uh of the disease. Um So um another way to sort of state this maybe in simpler terms is, you know, precision medicine is really to ensure that the right treatment is delivered to the right patient at the right time. And um you know, II I just want to spend a few minutes on this slide because it kind of details how we've arrived at this, this stage now in precision medicine. So, and we're gonna work from this box and go up here and then over there. But essentially what we have here on the sort of vertical scale is uh you know, interventions or uh or detection or diagnoses that are made on an individual scale. And then here on this part of the uh uh of the vertical scale, we have population based interventions or measurements. And then here on the sort of uh horizontal scale, we have low dimensional data and on uh on the other extreme, we have high dimensional data or precision medicine. And so in a way, you can say that, you know, uh the the medicine as it is practiced, that has been practiced until uh you know, maybe the past 2030 years has been one of trial and error uh of uh making inferences based on single patients on anecdotes. Um And of course, this method, which is the sort of tried and tested method of classic medicine is a huge risk for bias and confounding. Right? Just because you observe something in one patient doesn't mean that it's true for a population. And so this is what has led, of course to, you know, epidemiology to evidence based medicine, which I think has certainly contributed to significantly advancing the way we think about disease and the way we treat disease. But it has many, many pitfalls, it has many risks, right? So one of the risks of uh you know, evidence based medicine is that we tend to cluster patients together that maybe are quite different. Uh One, we have this idea of one size fits all right. So we we're gonna treat all patients um with uh say hypertension in a similar fashion. And you have hypertension which affects maybe 50% of the population. Is this hugely heterogeneous disorder that has many, many biological subsets. Um you know, so this this approach which you know, has led to standardization to protocols, to algorithms, to guidelines. This is really the way we practice medicine. In 2017. I would argue that this approach which is hugely valuable is really reaching its limitations because we realize that one size does not fit all. And in many cases such as in TB I, uh it hasn't worked, right. We don't have a treatment for TB I, we don't, we're not really good at predicting uh TB I outcomes. And so this is what has led to uh newer approaches thinking about not just uh the phenotype or some uh gross imaging marker, but thinking in terms of the entire realm of data that is available on a single patient, right. So that means looking potentially at gene expression or protein expression, that means obtaining quantitative images. It means very detailed genotyping props using uh you know, wearable sensors. And this is I think where we're potentially moving very in a very short period of time right now, it also means potentially using smart or adaptive clinical trial designs to better select those patients that we think are most likely to respond to the intervention. And the hope is that uh by applying these methodologies and populations, we could then this this will trickle down to um a uh an approach where we could characterize the biological uh uh specifics of individual patients. And and using that uh that information to potentially tailor therapy, individualized therapy for, for care. So I hope this this makes some sense, but I'm going to illustrate this a little bit more. How do we, so the question then you could ask is, do we really need precision medicine, precision medicine? I would argue emphatically that yes, we do because current phenotype based descriptions, clinical descriptions that we make at the bedside are good, but they're not great, they're not great in terms of really telling us about the underlying biology of our patients. Uh they lack specificity and the same applies for treatments, right, the treatments that we implement and many of our patients are nonspecific and they have significant toxicities and we're not very good at prognostication. And so this is the uh the sort of the basic idea of precision medicine. It's to integrate data at multiple scales, going from the molecular to uh to the entire organism to the actual uh uh environment, social uh and uh and physical environments of our patients and thinking globally um uh about these different layers of data, potentially we can arrive at a more precise idea of how a single patient is doing and how that patient will respond to treatment. So what is driving? Uh Yes. So this makes a lot of sense and maybe by the by the integration data across different levels, we will understand other things. But just by doing that, it's really a shooting on the duck because you also have to inform these by like you said, biological processes that may be so not just by integrating data will come up with solutions, the part of the problem because you couldn't see before because it was a non integration, certain situation may happen but others may not. I think, I think you're touching upon a fundamental question which also arises in regards to precision medicine, which is that um we're talking not only about a potential shift in the way that we might think about disease or practice medicine, but we're talking also about a shift in the way that we might conduct um you know, research on disease. Because uh you know, the classical approach of course is to formulate a question, a hypothesis and model and then work your way through, you know, either animal experimentation or on humans to try to demonstrate whether or not your hypothesis or your model is correct. And in many ways, what you know, precision medicine is proposing is a very different non model model free approach where you would essentially collect data. And that by, you know, applying specific statistical techniques, you know, computational approaches, artificial intelligence, machine learning approaches to these complex data sets that actually some significance will emerge that hypotheses might emerge from, from that, from that, from that approach. But I think that there is a risk of course that in many cases, you know, we can go on a fishing expedition and we can pour through huge amounts of and many different layers of data without arriving at anything that is significant. So I think what we need ultimately is a good balance between hypothesis model driven research and data driven research. And I'm not sure if I'm answering your question, but one question that also, you know, arises arises is, you know, why now, why are we talking about precision medicine? Now? I mean, we've been doing, you know, genomics for, for more than uh several decades, we've been doing protein expression arrays. We've been doing advanced imaging for many, many years. Why is everybody so excited about precision medicine right now? And I think one of the big reasons is economic, it's financial and the fact is that you, you all are familiar with the uh Moore's law, which essentially states that uh the cost of computing is uh is uh going down uh over time. And that uh the the power of computing is going up uh in an almost reciprocal fashion. Um And yet, when you look at the cost of sequencing the genome, it actually is uh going down. But at a at a rate that is even faster than would be predicted by Moore's Law. So probably part of it has to do with, you know, better computing. But part of it may also have to do with uh more sophisticated technologies in terms of molecular sequencing. And I wanted to show you this, this is a paper from the New England Journal Medicine where a review on Precision Medicine. But I want you to look at this blue line. OK. This blue line represents sort of hypothetical cost of phenotyping. It represents, for example, what it would cost if you have a patient, you know who is admitted to the hospital for evaluation of congestive heart failure. You bring this patient into the medicine service and you do you know an EKG and a chest x-ray, you examine this patient, maybe you get an echo, maybe you get a cardiac cath. Uh and then a day or two later you release this patient after modifying their, their treatment. But what I'm trying to show you is that it's quite clear that the cost of sequencing the genome is now potentially has reached this this inflection point where it's probably much cheaper. In fact, it is uh it costs about $600 to sequence your genome to admit a patient with heart failure for one or two days is I don't know, maybe 10 $15,000. So it's cheaper today to actually sequence all the information that is really kind of the, the um you know, the the cellular hub of all the information that is going to happen in your body than it is to perform uh you know, a routine sort of history and physical and get a few additional lab tests. And so why is this important? It's important because it means that uh it's becoming economically feasible actually to perform precision medicine in large numbers of patients. Uh And this is what is behind um the, this concept of uh the of advancing uh today with with precision medicine. The basic idea is that, you know, we collect data, we collect data at multiple scales, we collect um um you know uh detailed quantitative data, this generates information then using computational approaches and other statistical methods, we potentially arrive at some new knowledge or some new insights on disease, potentially, this will lead to modifications in in health outcomes. And the idea is that these, these changes in health outcomes will then inform uh the data stream of your origin and will help us in a sort of recursive or iterative way to uh um potentially advance uh uh science and biology. Uh And I think that this, this idea is very powerful and it's really what has been behind. Um the Precision Medicine Initiative, doesn't this seem kind of old? It seems like it was ages ago that he was president and yet it's been only a few weeks. But uh anyway, um so our dearly regretted, um you know, former President Barack Obama actually was a big proponent of precision medicine, even in his days as a senator back in the early 2000 period, he was advocating for precision medicine as a new approach to thinking about disease. And during his tenure, he launched this initiative two years ago, which is essentially is an alliance between the NIH and the dod the DARPA uh to uh enroll 1 million individuals, um healthy individuals, uh ill individuals across the country and perform very, very detailed phenotyping, imaging genome expression analysis, proteome uh and sensor based measurements and follow these people over time, right. So this uh Precision Medicine initiative is actually under way. Uh There are a number of centers, not Johns Hopkins, but a number of centers have been funded to carry out this uh this enrollment. And this, this uh this research and the expectation is that this huge data collection effort, 1 million individuals is going to potentially transform the way that we think not only about disease but also about health. And how can we potentially reform our ideas about, about classification? So let me talk a little bit about the remaining time that I have about um a traumatic brain injury. And I've kind of put this down as six tasks, six herculean tasks uh to uh potentially uh transform the way that we deal with traumatic brain injury patients. Um You know, so um I think first we, we, we do need to understand more about the biology of TB I TB I is a very complex disorder. Um And it's notwithstanding a great deal of interesting basic science and translational science in this field. We really haven't arrived at a sort of unifying hypothesis about how TB I um uh occurs, how it um you know how it damages structures in the brain and how over time it can result in different types of deficits, but also it can result in regeneration and repair. We need to uh I would like to talk a little bit about imaging because this is an area that I do my, my, my research in primarily we'll talk about um uh the electronic health records, um biosensors, clinical trials and uh and knowledge networks. So I think perhaps the most important point is that we don't know enough about the biology of TB I TB I is a very complex disorder which affects not only neurons but glial cells. It affects the function of synapses. It disrupts neural circuits at the micro circuit level and also at the network level. Um And it also disrupts the vasculature, both the macrovascular and the microvascular. So it's a very complex and heterogeneous disorder that affects the brain in many different ways. And the specific preponderance of these different mechanisms may vary considerably between individuals and also depending on the underlying mechanism, whether we're talking about an impact or a blast or whether we're talking about an acceleration or deceleration type injury. And so another point just to sort of be um you know, emphatic here is that T I is really not a localized disease you should never think about even though you might see a patient with a focal Pran confusion. Uh and say, OK, this is just a localized process. It is not, in fact, TB I is a disease really of systems of circuits. Uh And this has been shown abundantly uh in a number of studies in animals and humans. One of the, the ways that we've been able to really understand the fact that TB is a disease of neural systems is that uh we are able now to map out the white matter. Uh for example, using diffusion tensor imaging and we can see very clearly that even in the mildest forms in concussion patients who apparently had a very light or mild injury, you can see that the white matter is disrupted and it's disrupted in a multifocal or disseminated way in almost every case. So this diffuse Saxon injury is really one of the hallmarks of TB of all severity. Yes, yes, this was done. Uh This was done in the acute setting, I think it was like five days after, after TV. And so, you know, we arrive at an interesting kind of model where we know that the connectivity or the networks of the brain, you know, over time because of aging, perhaps because of vascular processes, perhaps because of amyloid deposition, this network uh loses some of its integrity. Um But what we know as well is that TB I uh actually provides a uh an interesting scenario where that's a loss of connectivity occurs much more rapidly. And so what you might see in some individuals at least is a cumulative effect of both age related cognitive impairment due to disconnect or loss of connectivity compounded by the disconnect that occurred because of TB I. And so what TB I may do in some cases is it may reduce the um cognitive reserve that individuals have that allow them to potentially, you know, suffer a brain insult or an encephalopathy without, without major symptoms. If a patient has had TB before, it's possible that that amount of reserve is diminished and they will be much more symptomatic. In fact, we have some data to suggest this. This is just a slide demonstrating the very dramatic effects of uh this uh chronic traumatic encephalopathy. The um the uh brown stains here are actually selected binding of TAU protein. Uh You can see that there's significant cortical atrophy that is occurring. This is a 23 year old individual who committed suicide in the setting of uh traumatic encephalopathy. And you can see the brain is shriveled up and there's a very deposition of uh tau protein in the cortical uh layers. And so the question is, you know, can we understand more about this process through precision medicine? And of course, this is the what is familiar to all of you, this basic dog in biology of, you know, genes that are, of course leading to transcripts that are leading to translation to proteins that then lead to um uh the um specification of specific metabolites. And the question is, can we exploit this type of uh paradigm to better understand what is happening in TB I? And I think the answer is probably yes. What we need to do though is maybe to get rid of uh the very simplistic um linear associations that many people have tried to make between specific genes being increased in expression or decrease in expression and specific uh phenotypes. What we know is that genes never, and proteins never act in isolation. They're part of networks, right? And so we really need to understand is how the system of genes is potentially modified in specific individuals. How the system of proteins is modified. And uh using again uh methods from systems biology, we can um develop a very, very interesting hypotheses about perturbations of these systems and how these perturbations lead to specific um clinical phenotypes. So, if we talk about genes and genes that have been associated with traumatic brain injury outcomes, uh A number of, you know, associations have been made and I'm just showing you some examples, there's many others. But um the question is, you know, what is the significance of these associations there, it appears that some um you know, um single nucleotide polymorphisms. So some variants in gene expression are associated with specific recovery phenotypes after TB I. Um But you know, the question is, can we actually using these types of associative uh you know, descriptions, can we actually arrive at some biological hypotheses about what is driving recovery after TB I? So it's a very interesting one of the genes that has been shown to be significantly uh depressed after TB I is disc one disrupted in schizophrenia. One, this is a uh kind of an interesting scaffold protein that really orchestrates the movement of many proteins in the cell and seems to play a hugely important role in two things. One is in the division of differentiation and migration of neuronal stem cells. And the second is in the expression of glutamate receptors at the synapse. And, and therefore, it plays a role in synaptic plasticity of both a role in neuro genesis and in synaptic plasticity. So this is quite possibly the single gene that you wouldn't want to depress or after a TB I, you really want this gene to be actually increased in expression to allow potentially some regeneration to occur after this, this damage. In fact, it turns out that this one is dramatically used. Interestingly. This, this one has been described by investigators here. Um and Johns Hopkins, the group of Guo Li Ming and have done some really, really nice work demonstrating how uh this gene mutations in this gene which have been associated with severe mental disorders. Like schizophrenia and depression. Actually, uh you can uh buy um deriving somatic cells from individuals who have this mutation and de differentiating them into induced prepotent stem cells. You can then have a model system where you can induce a neuronal fate. And you can observe in vitro the synaptic characteristics of these of these cells. And this has led to some very, very interesting hypotheses about how this one plays a role in determining our variance or abnormal synaptic function very, very early on possibly during childhood and individuals who are at risk for schizophrenia and for depression. But I guess the point I'm trying to make here is that it's possible that this one plays a very important role in determining post injury, for example, post stroke or post TBI I regeneration or in addition of regeneration. And this is something that we're beginning to look at in our lab. There's also quite obvious evidence of epigenetic changes. So, epigenetic changes mean those changes that do not involve the actual sequence of base pairs in the DNA. But involve, for example, modifications in histon or methylation acetyl that occur after TBI I. And it's likely that these modifications play a very important role in determining gene expression and protein expression. And we know that they are quite important and dramatic after TBI. Another form of epigenetic change is the the expression of these micro RNAs that can be actually released from cells and can be measured in the extracellular fluid and also in the bloodstream. And increasingly now these micro RNAs are being recognized as potential biomarkers of neuronal damage as it occurs in TB I. But also providing clues as to the underlying neuronal changes that occur both in terms of the injury and also in terms of the regeneration. And this is a another example of the molecular analysis that can be applied. This is a lipid um analysis again uh in, in TD I, looking at the expression of different phospho lipids uh in an experimental model of TB I. This is some of our own data and this is uh was done with uh my colleague, a pen from the uh Proteomics Discovery Center. Uh We have looked at a small number of 25 patients with a concussion. We took their blood and we've been analyzing lipid expression and we found that selected lipids seem to be increasing in expression in concussion subjects compared to healthy controls. And the question is, what is the significance of these lipid changes in the blood? And can they inform us not only as biomarkers to detect injury, but could they potentially help us understand something about the biology of neuronal damage? And then this is some very interesting work that we've been doing um in my lab. This is actually not TB I it's a, it's a mouse stroke model, but we've been interested in the uh like many others in the composition of the microbiome in the gut of these mice. Um And um what we did was we uh gave uh some of the mice, uh antibiotics and some of them, we gave them no antibiotics and then we gave them all a stroke. And we found that um you know, so when you give antibiotics and this is just essentially a cluster gram of all the different uh uh types of, of, of um microbes or bacteria that are found in the gut of these mice. And you can see that when you get antibiotics, you deplete, of course, the the bacteria. But more interestingly, we found that when you give antibiotics before the stroke to mice, you actually decrease the size of the infarct uh quite significantly. Um And we're just trying to work out what the significance of this, but I bring this to your attention because it's likely that um alterations in the microbiome, not only in the gut, potentially in the skin, potentially in the in the lungs, um could have a major effect in terms of modulating immunological responses and determining the degree of recovery that occurs after TB I. So just a few words about mapping the brain, it's very clear that non quantitative qualitative approaches CT scan, for example, does not allow us to discriminate. So these are two of my patients in the neuro clinical care unit, they both had identical, you know, clinical characteristics, they were both identical in terms of their CT scan or almost identical. One of them. Uh This patient actually survived and had a great uh functional outcome. This patient died and yet at the outset, these patients were indistinguishable. So can we do better? And the idea that I that we have and others is that maybe since we postulate the TB I is this disease of neuronal systems of circuits, maybe we can interrogate those systems in vivo using advanced uh MRI. So the technique that we've been using over the past uh three or four years is uh diffusion tensor imaging. And this is an example of tract gray combining that with uh resting functional MRI to look at how different parts of the brain are activated at rest and determining the functional coherence between these different regions. So this is just an example of um how TB I this is a control subject. And you can see this kind of spaghetti of white matter tracks uh that is quite robust. And here, by comparison is a subject about A B I six months earlier with severe TB I. And you can see this rarefaction of the, the white matter track as reconstructed here in three dimensions. So very, very dramatic effect. And this is what I was talking to you about earlier, which is how the fact that TB I is this multifocal diffuse uh disease involving many, many different regions of the brain and in particular the white matter. So we did this study where we looked at uh the integrity of white matter tracts in 100 patients with severe TB I. And then we followed the survivors over one year and we got uh their functional outcome scores. And what we found was that when we compared uh the impact score, which is what I showed you earlier. It's the CT and Clinical classification. And we compare that with a new score that incorporated uh CT clinical but also a composite DT I image of the white matter. We were able to increase discrimination significantly. So the area of the receiver operating factors curve went from about 0.6 to 0.9 right. So this is practically a doubling of our discrimination. Um and this was a very impressive result. And, you know, since then, we've done some additional studies, this was a study where we looked at um um a a very large number of 10 1000 patients who had undergone functional MRI after brain injury. And we tried to determine which were those regions of the brain where the signal was most profoundly depressed in comatose patients. So we found that the area that seems to correspond best was an area that involved the thalamus, bilateral retros and uh and uh and also some media frontal structures. Um And these are structures that form part of a network, which is called the default mode network, um which is postulated to be significantly associated with loss of consciousness. And as it is also associated associated with many other um uh uh functions such as uh introspection. And so, based on these preliminary observations, we conducted a study of patients with severe TB I. Uh this is uh uh in 28 patients with severe TB I who were admitted to the SCCU, we took them to the MRI scanner and we uh did diffusion tensor imaging, but we also did resting functional MRI to identify the uh functional architecture of these uh of these severely injured patients. Um And what we found was that uh when we focused on specific networks, this is that default mode that I was mentioning to you earlier. This is the attention network, this is executive control network and this is a salient network which is involved with um uh identifying salient features in our environment. When we compared control subjects with those TD I subjects who had a favorable or unfavorable outcome, there were significant differences, but then again, there seemed to be major differences between those who had a favorable and those who had an unfavorable outcome in terms of their connectivity. Um And uh the most dramatic effect was actually not so much the coherence or connectivity within specific systems or networks, but the discon connectivity that occurred between networks. So for example, uh the uh anti correlation that is normally seen between uh the default mode and the salience network, which is something that is uh uh characteristic of all normal individuals. When you open your eyes and engage in a task, your default mode tends to shut down your uh salience network tends to light up. Uh that anti correlation um actually is lost in patients with TB I and in particular, in those patients with TB I who had a poor outcome. So what we think is that maybe this is a represents an interesting kind of bio marker uh of uh the types of functional changes that might result from this massive damage that is occurring to the white matter tracts in the brain. And there is some biological plausibility to this. This is actually a very, very interesting study that was published, looking at uh these different networks that I described to default mode salience and also determining uh the uh gene expression. So they had access to uh autopsy specimens in a subset of patients. And they looked at gene expression and they found that there was correlated gene expression that seemed to correspond almost exactly to these functional maps of networks that we are obtaining using resting functional MRI. Uh So there is some sort of uh we're not just talking about blobs in the brain, we're talking potentially about something that could be biologically possible. Uh And there is of course, increasing focus now on uh resting as opposed to task based on F MRI techniques to assess uh the integrity of the functional architecture in the brain. But just to tell you that, you know, there are so many other ways that we can map the brain. This is mapping of amyloid and we know that amyloid of course, is associated most significantly with Alzheimer's dementia. But it turns out that in the setting of TB amyloid deposition is dramatically increased and this amyloid deposition never quite goes away. So this is a subject at 48 hours, 54 days and nearly one year after TD I. Uh And you can see it's moderate and severe TD, you can see the significant amyloid deposition, which is present, which is totally abnormal. This is a 34 year old subject. Um So something that might suggest a potential link between, again, the acute setting where you have uh an insult, that diffuse axonal damage. And these long term sequela that we're observing in many patients with TB I. This is another pet based um radio ligas that is selective, not for amyloid, but for uh to protein not more recently. And you can see this is a, these are retired uh NFL players. Uh And you can see that when you compare controls those subjects with those subjects with sustained TB I during their tenure in the NFL has significantly increased uh TAO uptake using this technique. And the uptake was actually comparable to what was seen in subjects with Alzheimer's disease. Um And this is uh data from here from Hopkins uh with another a third type of um radio Ligon. This one is for TSPO. So the TSPO is essentially expressed in activated microglia. So it's used as a marker or a Serg for inflammation. And you can see that in this particular former NFL player um years years after he had sustained uh B I, there was still evidence of significant microglial activation throughout the brain. I should say that uh this TSPO is not specific for microglia and also B uh as activated astrocytes. So just a few words about biosensors because I know we're running short on time. But I think this is something we can't overestimate the importance of wearable wifi enabled biosensors. These are there's an increasing focus now on sensors that can measure not just motion or motor outputs but also the composition of the chemistry in your skin. Uh You can measure as well the um the um uh different electrical circuits that are generated at the surface of the, of the skin. Uh So I've been working for the past uh year actually with a group from uh the Department of Biomedical Engineering in Homewood to develop a multi segment um sensor that uh sensor system that allows us to essentially uh measure uh and to track in real time uh the motor outputs of uh of subjects. And so we're currently using this system uh to classify different types of movements that we see. For example, in coma patients in the NCCU, we, we are, we've obtained a high degree of accuracy and classification with healthy subjects. And now we're going to implement this in the, in the IC U to determine whether we can in a sort of automated um unbiased fashion, obtain uh classifications of motor outputs. And then of course, this is going to lead to a number of other interesting projects, but just to sort of give you a sense that this is also what we're talking about when we talk about precision medicine in terms of clinical trials. Well, it's it's quite clear that, you know, in the realm of oncology and cancer. The um approach to treating cancers has really been revolutionized by the identification of specific molecular targets. Uh And we know that the outcomes in many types of cancer um are dramatically different now than they were even 10 or 15 years ago. And I'm thinking for example of melanoma or breast cancer. So the question is, can we apply similar methodologies or approaches to think about treatments for conditions like TB I. And here we need to talk about two different types of enrichment. And when we talk about enrichment, we're essentially talking about selecting a subpopulation um within a specific disease category. So, prognostic enrichment essentially means identifying those individuals that we think are most likely to encounter a specific outcome. Um And this can be based again on this multilayered data set uh combined with clinical and imaging characteristics. Um And this has to be contrasted with predictive enrichment, which is really a clinical trial strategy. It means essentially to identify those individuals using specific markers to identify those individuals that we think are most likely to respond to a specific treatment. Um and enrolling preferentially those patients into our clinical trials to maximize the likelihood of demonstrating a treatment effect. Um And even though this seems very simple, it's interesting that the vast majority of clinical trials that are being done in medicine today are not uh are not using these enrichment strategies. It's clear that we do need to do this and uh a a major emphasis in clinical trials moving forward is going to be to use biomarkers, right, either biomarkers obtained at the outset to select patients for specific treatments or biomarkers that are obtained uh along the way after a specific treatment, get the biomarkers and then continue to treat only those who are positive or negative of the biomarker. So this um adaptive clinical trial design is something which I think is, is hugely important in the last few minutes. Um Of this talk, I'd like to just briefly mention to you that right now, there are a number of ongoing uh data collection efforts that are hugely important in the realm of TB I. So, one of them is uh just the, the basis for many of these uh efforts is the what's called the common data elements, which essentially is an uh a group of experts um that uh met in the NIH to determine specific clinical research, imaging uh definitions for um uh uh specific endpoints uh with respect to TB I. Um And then based on this common data elements project which really was put into place about 78 years ago. A number of studies have evolved. There's the track TB I which is an NH sponsored multi center um efforts. The target enrollment is 3300 patients. Um And the goal is essentially to recruit patients with TB I across the severity spectrum and obtain very detailed imaging phenotype genome and proteome and then a subset of patients to follow them over time to determine how we can build better prediction models. There's a very similar effort that is being run in Europe. It's the center TB I project. Now the target enrollment is 5000. Um and this effort is ongoing and I have the privilege of being a part of both of these. Um There is for pediatric patients, there's this approaches and decisions in a key pediatric TB I adapt trial. There is a very interesting effort, the chronic effects of neurotrauma Consortium, which is led out of the VCU and then there is the concussion assessment of research and education. All of these are accessible online and all of these efforts I'm putting them out to because all of these efforts are funded federally, which means that we as being part of the scientific community have potentially access to these data sets. In fact, I can tell you that for example, the track TBI, it's not very complicated to make a request and obtain access to very detailed data sets in TBI patients. And this could be a huge advance for people, especially who are beginning their research careers and who don't have necessarily the means to institute on their own their multi center trial. So there's also um these uh data sharing platforms. There's the, if you look it up online, the Fit Beer, which is the Federal Trade Agency, uh traumatic brain injury research um consortium, which essentially is a clearing house or a platform that allows people with TB data, whether it's experimental or human to deposit their data sets into this repository. And this data can then be accessible to the scientific community. And there's a similar effort that is being done internationally. The in I just wanted to also point out that uh these two large scale initiatives, the brain initiative which was launched maybe uh four or five years ago, initially called the Brain Map and then the Human Brain Project. So the Brain initiative here in the US funded by NIH and the Human Brain Map um funded by the European Union. These are huge opportunities in terms of funding. And I think that the I'm bringing them up because we also represent opportunities to use TB as a model system to interrogate the neuronal architectures and circuits in the brain. And while a lot of this is focused on basic science and potentially non disease related neuroscience. In fact, there are opportunities to tap into these funding streams for TB related research. And I can point to you if you're interested to specific projects that are doing this. Um As far as precision medicine is concerned here in uh Johns Hopkins, I think you're familiar with the in health or individualized health initiative, which is run by Scott Zieger and uh Anthony uh Rosen. And essentially what it is is an alliance between the health system the applied physics lab and the university to really push the agenda forward in terms of precision medicine um both uh in, in, in, in the realm of biology and also in medicine. And this initiative, which currently is still trying to find its way and is not sufficiently funded. It's likely that this will become a much bigger deal and that we're going to see in the near future, the emergence of a true Precision Medicine Institute as part of the Johns Hopkins system. So in conclusion, I just wanted to sort of give you a flavor of TB I TB I is, I think, I hope you've been able to appreciate it is a major public health concern. We really don't have any accurate um you know, markers for detection, for prediction for treatment selection and there's no effective therapies, right? So the opportunity here I think is is precision medicine and here we're talking about capturing the biological sort of signatures of individual, you know TD I patients or subgroups uh using the methods that I described to you. Um And of course, working in a collaborative fashion uh with uh our colleagues in engineering and statistics to develop uh the appropriate models that can help us understand these complex data sets. Um And of course uh getting engagement from investors and stakeholders because while this is becoming a very, very important topic right now, this is not the case before and perhaps one of the reasons again, it's economic, the cost of sequencing the genome in 2000, the year 2000 was about $15 million. And as I mentioned to you today, it costs about 400 or $500 to sequence an individual's genome. So here is kind of like the the the sort of road map. Um you know, we have individuals that arrive in the hospital when they acute illness and surgery and injury, they undergo these trajectories in the hospital and they have long term um outcome trajectories. And of course, one of the tasks for precision medicine is to generate uh better biomarkers again for detection and classification, but more importantly for um prevention and for treatment. So I'd like to acknowledge the support from um from these different institutions and also recognize a lot of my collaborators that we're currently working with uh both here at Johns Hopkins Department of Radiology, uh Department of Neurology, Department of uh Biostatistics, uh Proteomics, and then also a number of colleagues in other institutions um uh in uh in Europe and also in the US. So with that I will stop and I'm happy to take any questions. Forgive me if I had to rush through that a little bit. Yes, you mentioned uh antibiotics in size and we got bacteria. Is there any indication that going the other way and giving things like probiotics? Why have any positive effects in that? Yeah. In fact, that's um it's, it's a specific hypothesis that is being tested by one of my colleagues in Berlin. Um And so far, the results have been not very conclusive. Um but there, there is a great deal of interest in this kind of brain gut access as you probably know. And on the one hand, it seems like an acute brain inside like a stroke can have profound effects on, for example, gut permeability translocation of bacteria. On the other hand, it seems like modifications in the flora of the gut can profoundly modulate the immune system and lead to changes in the way that the brain responds to an injury. And so the question is, can we interfere in that cycle, you know, potentially by giving antibiotics or by probiotics? Um There was, I should mention there was a large randomized controlled trial, I think it was published just a few months ago on lancets looking at the effect of poststroke antibiotics um in uh in humans. Uh and this was done in Europe. Unfortunately, it didn't show any benefit. So I think there's still a lot of things that we need to learn. But um the the point is that the the microbiome represents potentially a very, very important opportunity to uh to modulate the way that the brain is, is uh is responding to an injury. Yes, the kind of inter I was too much because with any small change carry on, more us, kind of pointing out at this point, this morning this matter. I so you know what I mean? Is this going to stop in between this time there? So no. So I would respond by saying that you're not alone. I also have numerous difficulties trying to think about the enormity of the task. I think this is probably the most important question because it's in a way you could say that it, it's not so hard to collect data, it's not so hard to obtain imaging, to obtain to sequence the genome. But making sense of that in a way that is relevant clinically and can change the outcome of a patient is a much more difficult task. So that analytical task and the construction of appropriate models is really where the heart of this problem lies. And I don't think that anybody has the answer. So believe me, you're not the only one who feels that this is a daunting task. Yes. So in rehab for all the progress is something and that of me is psychosocial fire. I'm wondering how you're able to after that hit. Yeah. Yeah. Yeah. I think, I think most theoreticians of the this precision medicine process believe that the environments including the psychosocial context is part of that multi layer data set. So that means that you can't pretend to be comprehensively collecting data and analyzing your patients without actually integrating that type of information. So I guess I would answer by saying that that is part of the precision medicine construct, it's sometimes difficult, I guess because you know, how do you quantify a lot of these things? How do you quantify, for example, the the richness of somebody's social network, how do you, you know, and really what precision medicine is about in many ways is about moving away from qualitative approaches towards more measurable quantitative approaches to all these different data sets, right? And so, you know, so I would challenge people to think about, you know, what are useful metrics that we can use to integrate things like the social psychosocial environment into precision medicine models. And I think that's something that we should do going on that there are people who have to do research and science or psychosocial. So they have quantitative metrics. I think you can, you can argue with how sensitive they are and how they are and so on. But, but they use me to psychosocial. And I think that that the behavior is a contribution that the the specialist can, can do to contribute to the spectrum of the preserve medicine. So that's why I don't wanna say that that's important and it used to be paired with other elements of data that you can. So just of comments. So I think it's collaborating point um that you had that slide about um TB I and veterans uh and dementia outcome, for example, was rank a covariant in that an idea of, of their or their socioeconomic status, for example, and their risk of, yeah, I seem to recall that they, they did a multi variable model and they adjusted for different things. I'm not sure if they adjusted for rank, but they found that exposure to TB, I was independently associated with a higher risk of dementia. That graph that I showed you was just one out of many presented. But sir, you had a question. I think one of the points that I think about you need to think about the definition of precision medicine a little bit broader. That is the perfect decision is to try to identify some characteristic about a person that should judge their responsiveness or prognosis, prognosis as well as their responsiveness to treatment. The current thinking about the decision last year is that the genome is the place to look for that. And I would argue that, but it will take some time for that to bear some fruit and that it would be useful to. But the concept though is very useful for all of us that is trying to identify unique characteristics of this person, their genome, their social situation, their cognitive status, their other statuses that may indicate their responsiveness or appropriateness kinds of treatment. So I think the model is a very powerful model, but it's not just the genome to look at. There, there are other classes of variables to look at, but the model is very, very important. And I think it's easy to say, well, you know, the genome is not relate to rehab. But I think the model is very relevant to rehab. Although we may be looking at other kinds of factors that determine outcome and in many ways precision medicine, in the broadest sense of precision is the future of medicine. I I couldn't agree more with you. And I think at the beginning of my talk, I did emphasize that, you know, the genome is just one layer out of many different layers of uh of data that we need to try to think about in developing this this individual signature of of people. And and it it relates not just to disease, but it relates to health as well. So I agree completely with you. I think is very, very useful for us to be thinking about in our department because I can't give it a big grant to look at data. Well, in fact, there are large databases out there that can be queried with appropriate permissions and support and I would just add to the ones that he did. Of course, the model systems DB database is the largest brain database in the world 14,755 people as of yesterday. Well, over 25 years, nothing is remotely as far as I know we are that size although it's older, so somebody is not so good, but that one and also can be queried for free. And I think our residents and a lot of our faculty should be looking into these large data sets to begin to investigate questions, particularly when funding is so tight and there are important questions to be answered. Ok. Well, thank you very much. Created by Related Presenters Robert Stephens, MD Dr. Stevens is an intensive care physician who treats patients with critical surgical and neurological illnesses and injuries. View full profile
Facebook
Twitter
YouTube
LinkedIn
Instagram
E-News
BROADCASTMED