It's now my honor to welcome Dr. Jeff Manley to give the concluding talk, although I think that was a pretty thin excuse for getting the last word, Jeff, arranging to have that plane come in at the last minute. Dr. Manley is a professor of neurosurgery at University of California San Francisco. He's the chief of neurosurgery at San Francisco General Hospital, one of the legacy neurotrauma hospitals or trauma hospitals in the United States. He's world-renowned for his research, and he's going to talk to us about international initiatives. Jeff, good to see you, my friend. Well, thanks. I know that I'm the only thing standing between you now and cocktails on Bourbon Street, so I will move it right along here. Left or right here, David? There we go, super. So this is a slide that came out of a paper by Stein out of Penn, which I think sort of critically looked at where we are with traumatic brain injury. And so obviously there have been tremendous improvements from the 70s through really the 90s, and I think we can find a lot of reasons why that's been. We've had better pre-hospital care. We've had better antibiotics. Neurocritical care has gotten better. Surgery has gotten better. And then we see that things have sort of flattened out. And over this period of time of this flattening out, if you will, we've been doing a lot of randomized controlled studies. And I will be the first to tell you that the purest form of research is a randomized controlled study. You cannot get class data without it. But I think that we've had a real tough time doing these randomized controlled trials in our field. And so I think that more often than not, and we have some of the luminaries here like Peter and others who have worked on many of these, more often than not we're left wanting more. So, for example, with Peter's paper, someone just presented a slide that showed this outcome at six months. But then Peter, fortunately, then published what the outcome looked like at 12 months. And so the outcome at 12 months actually looked better than the outcome at six months. And who here would want a 12-month outcome versus a six-month outcome? I'm going to go for the 12-month outcome myself. And so, you know, I think we struggle a lot because we have a lot of legacy things. And these legacy things kind of drive us in different ways. So the notion that we must stratify on a coma score. So our colleague from Finland was just talking about, you know, the fact that you can't really stratify, you know, even in mild on the GCS and so on. We've talked a lot about that. And even the outcome measures that we look at, which is disability, and we use a measure that doesn't really capture a lot of the things that may be proximal to the disability. In fact, it's not even brain specific. So I think that, you know, one of the issues has been is we've all been struggling about this. And have decided that I think that there's different ways that one could look at this. And so, you know, people have already talked about these studies, so I'm not going to go through them now. But usually we're left wanting, you know, from these studies to figure out exactly. And so different groups tease out different parts of them. And so, you know, if you like monitoring ICP, then your interpretation is that this was a trial, you know, the best trip trial was a trial of two different ways of treating patients, and it wasn't a trial of ICP. But ICP was actually part of that. And so many of these things are left to interpretation. And what I've seen is people sort of interpret them the way that they want to. So these reductionistic approaches are really trying to isolate a single factor in something that's super complicated and very heterogeneous. And, you know, this idea of us not accepting the heterogeneity of the problem and still trying to lump these into large groups, it's my personal opinion, and I think others share this, that it's not going to help us out. When I was in medical school, I think we had five different kinds of blood cell cancers, and now there's about 200. So if there's 200 different kinds of blood cell cancers, I'm going to say that there's more than two or three different kinds of head injury. And so I think that the notion that, you know, it's impossible to solve. I mean, we just heard a beautiful talk about chronic subdural hematoma. That's a pretty well circumscribed type of lesion, and one should be able to study that because it's more homogeneous than others. So the big thing that I think that many of us thought about, too, is that we really don't understand the natural history of traumatic brain injury. So most of us in this room have been very focused on what is this early proximal event. We're trying to save lives. We've got people in the ICU. And then, you know, how many people routinely follow a TBI patient like you would follow a cancer patient, you know, or even follow a transphenoidal patient? I mean, we don't really follow these patients. And so I think that many of the things that we read about in the press, about whether Dan just mentioned about chronic traumatic encephalopathy, we don't know because we have these events and then we have these things out here, and there's nothing that ties it together. And so to really understand what is the natural progression of this, who recovers, who doesn't recover, there's a lot of uncertainty there. This is a slide that came from Andrew Moss's work from the IMPACT study. And, you know, this notion that we should embrace the heterogeneity and say, hey, let's just use this because the heterogeneity of the patients has led to heterogeneity of the treatment. Here we were just talking about are we going to use steroids for chronic subdurals or not use steroids, or are we doing this, are we doing that? So there's so much heterogeneity that's out there that Andrew, for a number of years, has had this idea, let's use this heterogeneity and try to make sense out of what's going on. And so in center TBI, which I'll talk about in a few minutes, they've been doing provider profiling, and why are they doing this? If you look at this box here on the bottom right here, you can see that in certain parts of Europe that there were many-fold differences in your unfavorable outcome depending on where you were treated, suggesting that there may be better ways of treatment even in the midst of us not having a quote-unquote drug to be able to treat this disease. So the other thing that we've really, that are sort of forming the core of this international TBI research initiative is not only comparative effectiveness research, but also this notion of precision medicine. And cancer has really been the poster child of that. So when I say that we now have 200 different kinds of blood cell cancers, it's because we continue to split and to get more and more homogeneous groups as we find a new gene or a new pathway or a new surface marker. And as we start to look at these, we find out, well, these things actually behave very much the same way and these things behave the same way. And so that really the first step in all of this is to try to have a better diagnosis. So I think most people would argue that mild is not a good diagnosis. Concussion is probably not a good diagnosis, as our colleague just said. There's over 40 definitions of that. And so without an accurate diagnosis, then you can't really have a targeted treatment. And I think what people have pushed back on this, like, well, this is just so complicated. We'll never do this. And I would argue against that because we already have precision diagnoses in traumatic brain injury. We just heard about one, chronic subdural hematoma, epidural hematoma. These are things that we know precisely the mechanism with the epidural hematoma in particular. There's usually a skull fracture and an injury to an artery. I mean, we've all taken care of these patients. And the targeted treatment for that is not ICP monitoring. It's not drugs. I mean, it's neurosurgery. That's how we fix epidural hematomas. There are discrete groups, and maybe, you know, that's another subset that we can start to find. And so this notion of finding these different groups of patients and figuring out what they look like and how they sort of aggregate together to be able to target them is ultimately the way that we get to improved outcome. So the idea, then, is can we start capturing large amounts of data on a vast array of heterogeneous patients to start doing the splitting that will be necessary to be able to do this? So what came about this is that these ideas were really starting to ferment about 10, maybe I said probably about 10 years ago. And so there was discussions that Andrew and others were having with folks in the European Commission. David and many of us were starting to talk to people at the NIH to say, hey, you know, we really probably need a natural history study here. You know, we're calling it precision medicine. But it is a natural history study with just lots of new tools. And to try to better understand this and maybe to come together and work together. So it's now been probably, I would say, about six, seven years that we had this first meeting where we got together and we started talking about this notion of an international TBI research initiative to bring together groups in Europe and also in the United States and Canada. And subsequently, there's been more and more folks that have been added on to this. So the idea here is that during this time, we were developing common data elements that were ways to structure data such that if we collected data in Europe or we collected data in South America or in, you know, Pittsburgh, that all of this would be collected in the same way and how we would define age would be the same way. Same thing with blood-based biomarkers. So I know Andras is here who's been doing work with the Center TBI group. And so we've tried to harmonize how we're collecting our blood-based biomarkers. Same thing with imaging. When you see all these imaging studies, usually they're just a one-off because somebody will run a certain sequence on their magnet at their hospital and they'll do it on 75 patients, and that stuff really doesn't replicate. So this notion of standardizing things came about. Also, again, strength in numbers is that if you're going to start splitting and splitting and splitting, you can't really start off with 75 subjects. You actually need hundreds, if not thousands, of subjects to be able to do this. So this whole effort has been based on, you know, big data, if you will, in terms of trying to get a larger sample size to be able to do this and the power of actually working together. So really, if you think about Entbier today, there's two very large studies that have come out of this, which are the Center TBI study, which I'll just briefly mention, and then I'll tell you a little bit more about TRAC and leave you with a couple of interesting things that we've seen. ADAPT was a study which was comparative effectiveness, looking at how we treat children with traumatic brain injury, and that also had international engagement and papers are just being written on that. And then we worked also with the Canadian group, where there were more smaller studies that were done primarily with the concussion population in kids, and there's been several important papers that have come out of that work since Entbier started. So I want to just touch on Center TBI just briefly, and I think, you know, what's happening, as I'll show you, is that these studies are all coming to fruition, and you're going to see tons and tons and tons of papers coming out over the next five years from these studies. But Center TBI is in 21 countries. They're in over 70 sites, and it's everything from, you know, Addenbrookes and Cambridge, where Peter works with David Menon, you know, which I think would be, you know, sort of like one of the shining stars of academic neurotrauma, all the way to very small hospitals where there's been very little research that's ever been done in one of those hospitals. So there's a lot of variation, but that was by design, because they wanted the heterogeneity in order to be able to do the comparative effectiveness work. And so as part of that, they've already started doing provider profile surveys. There's been several papers that have been written about this already, and it's interesting to see the variability in just how everybody's doing this within these countries and the European Commission community. So this is an observational study, just like TRACT-TBI is an observational study. They're basically powered to have a core study, which has 5,400 subjects, and I don't know exactly what their final enrollment was, because they just stopped enrollment recently, and their idea was that they were going to get about 1,800 people in a cohort that was in the emergency department and discharged, another 1,800 that was in the hospital and then discharged, another 1,800 in the ICU. So they're much more heavily ICU-oriented than we are, and so by the nature of that, probably the severity of injuries that they're looking at is a little bit more significant. So for TRACT, and again, as I said, you're going to see lots of papers coming out about CENTER-TBI. If you want to know a little bit more about CENTER, there was a paper that was published in the Red Journal laying out the whole protocol and so on. So with TRACT-TBI, this is a U.S.-based study, which is slotted to enroll 3,000 subjects, including controls. So that's very important. So the CENTER-TBI study does not have control groups. We actually have two control groups. We have trauma controls, which are basically orthopedic injuries who presumably don't have head injury, and then we have friend controls because, as we were hearing about, you know, you injure the brain that you started with. And so this notion of having friend controls to control for some of the socioeconomic factors that could also be leading to unfavorable outcome, we felt it was important to have those, and we're already now seeing the value of having these controls. And I think it's just natural to want to study TBI if that's what you're doing, but without controls, I think it's going to be difficult to make definitive conclusions. And we'll be sharing some of this control data with the CENTER-TBI folks because we've been collecting the data in the same way so we can help them out. We're going across the spectrum from concussion to coma. We're very agnostic to the terms mild, moderate, and severe. We're looking at anybody that meets what's a broad definition, external force injury to the head, some sort of alteration of consciousness. You don't have to lose consciousness. That means that we get everything from concussion all the way out to coma. And the goals of this have been to improve diagnosis. As I just said, without a better diagnostic model, I don't think we're going to have better treatments moving along. We're also trying to look at outcome assessment because I think that part of this effort was to understand why the GOSC might work and why the GOSC might not work. And are there other measures that load into this that are more proximal to the injury? So, for example, when people talk about doing rodent studies, Most people that do rodent studies are looking at rotarod and they're looking at things like a Morse water maze which are a test of memory and learning. The GOSC is not a test of memory and learning. But there are tests of memory and learning and so we're looking to see how those impact the GOSC and could they actually be a better memory test or a better test for outcome and so we're looking at all this now. There was just this very nice talk on mild TBI. We don't think that mild TBI is mild and we've been learning a lot about what goes on particularly with the kind of mild TBI patients that we take care of. These are the people with GCS 13 to 15 that come into a level 1 trauma center. They're not necessarily your run of the mill mild TBI patients and what we've seen and one of David's trainees wrote a paper that's been highly cited over the last couple of years from the pilot study showing that these people actually look a lot worse than we were ever taught as residents as to how they were going to look long term. And then also building this big, fat, juicy precision medicine data set that we can go back and mine for many years. So the reason why we believe that this is going to be helpful is that if we look at the pilot study which David was part of and we conducted in between 2009-2011, so now that's been seven years since then, since that study's been done I think we've now published over 35 papers just out of that data set and we've still got papers that are being written. So this newer data set that I'm telling you about is even larger and should have much more value long term and so it will be something I think that will be valuable to the community for years to come. So what are we doing? Well this is sort of where we work today. We're collecting a lot of acute clinical data and we've been doing this for a while but now we're doing it in a structured way. We also know that imaging is very important and so we're now collecting CT scans, we're collecting tons of MRI scans. So we have the largest collection of serial MRIs that's been done to date and I'll tell you a little bit about what we're doing with those that make them different. We're harnessing a lot of clinical data, we're collecting real time clinical data, working with our colleagues in center TBI, particularly those from Cambridge. We're trying to use the same data file format so that we can merge up some of these data sets and I think we now have over 250 subjects where we've got real time clinical data coming out of the ICU so that's going to be very exciting. We have lots of information on the proteome and I'll share some new data with you in a moment but primarily we've been looking at S100 as we heard about a few talks ago, UCHL1, GFAP and some of the newer analytes as well and the genetics. So we are basically slotting to finish up this study in the next month and a half, two months and then we'll be doing GWAS, genome wide association type deep sequencing on all these subjects at the Broad Institute which will give us some further insight into the genetics or as we heard earlier the brain that you injure and what are the genetic constituents of that. So we're at 18 sites so it's not as big a study as in terms of the range of sites that they have in center but we particularly picked level one trauma centers that had a history of doing TBI research so we were not so interested in comparative effectiveness, we were interested in getting really high end biomarkers so that we had really high quality MRIs, we had high quality biospecimens and we tried to do things at the best level that we possibly could in order to be able to make sure that this would standardize, generalize and replicate and you can recognize many of the sites that are on there. This is also a big team science effort like center TBI so most of the people who you read papers by and so on that are part of this study, it's a very big tent and we have a lot of investigators that are underneath this and all working together, many people who didn't work together previously and I think that's probably if I were to look back over the last five years and to say what's been the most rewarding part about this is really the relationships that we've built and what we've learned from each other working together. So, as of this morning when I got up we have now enrolled 2,850 subjects so we're 150 subjects away from our target enrollment and this is predominantly a mild TBI study because mild TBI is the predominant form of traumatic brain injury. 57% of these subjects have a GCS of 15 so you might even call them concussion patients. So, with this said though there's actually quite a bit of morbidity in this population. So, I don't have time today to go over all these things as I said papers will be coming out. I just want to just touch on a couple of things that we've been doing here and really as sort of teasers for what's to come. So, one of the things that we've done is that we've standardized the imaging. So, people have been writing papers for 10 years about diffusion weighted imaging, resting state fMRI, connectomics, all this other types of really exciting things. The problem is that they haven't been standardized. So, what we did is we used these physical phantoms for what I would call ground truth for both diffusion, for structural imaging, and also for functional imaging. So, for the first time not only will we be able to look at these images but we have quantitative assessment of this. So, numerically we can look at this data and so when you start putting this into these large big data algorithms it sort of bypasses, if you will, the need to extract features like well this is an injury to the anterior corona radiata. I mean just numerically you can represent that and that should allow for some big time number crunching. And it's also helped us to understand a little bit more about diffusion weighted imaging because now we have it standardized and I think we have over 800 baseline scans with our phase one protocol and then David is running another protocol using high definition fiber tracking which we have I think close to about 300 scans on now. So, all of this is mounting in a very large data set. You know why MRI? This is some earlier work that we had that just showed that you can see all kinds of things on an MRI that you can't see on a head CT. The TBI is the last group that's using head CT as the workhorse. Everything else has moved to MRI and I think what we'll see is that there will likely be selected sequences that we'll be running in the future for this because we see things that you can't see otherwise. So, I want to note that this data is from patients being brought to level one trauma centers. So, I don't want you to generalize this you know for somebody that hits their head against a door jam or something. But what we found was and we've now replicated this in the current studies. So, this was from the pilot that David and Alex Velotka and I put together as a prelude to this larger study where we saw that of these people that were coming into level one trauma centers a quarter of them that had a negative head CT had pathology on an MRI scan. And we've now reproduced this with close to 800 MRIs at this point. So, again we've seen the same thing and you know twice in a larger more generalizable study design. Again, I won't go into this but right now we're processing data for the human connectome. We're processing data for resting state fMRI. We're processing the tractography. And we're processing all the volumetric work. So, we should probably know a little bit more about what happens to people's brains at least between two weeks and six months. And we're getting some farther out imaging now at a year and further on. The other thing importantly that we're doing is we're going back and correlating this with the pathology. So, we're collecting brains from grateful patients' families who are donating those to us. We now have eight brains in the study and you can see some of the interesting things you can do here. So, this was one of my patients actually who had seen this thing in the brain stem. But then when you're able to take the brain out and don't have to worry about the cardiac pulsations and you can run it on seven Tesla scans and things like this, we can go back and really do high resolution tractography and make sure that the pathology is matching up with what we see on these images. Because they are an extracted value and this is the ground truth if you will which is the pathology which we've gotten away from. I'll sort of wrap up here with the blood based biomarkers. So, there's a whole raft of these things out there now. We've been primarily focusing on GFAP and UCHL1. But tau is out there. Phospho-tau, we had a paper on that last year in JAMA Neurology. CRP which is just a great marker of inflammation seems to be very important as does this neurofilament light that's being seen in a lot of other neurodiseases as well. So, this is a paper that actually that David wrote back in 2013 out of the pilot. And it's a simple idea that if you smash your brain, you may release GFAP into the bloodstream. And so, you can see there's a nice separation between the CT negative and the CT positive. Here's what the receiver operator curve characteristics look like. So, already this has better performance for CT scan than BNP does for a lot of the things that that's used for. So, it's actually a very good biomarker. And just several months ago here on February the 14th, GFAP and UCHL1 were approved by the Food and Drug Administration here in the United States for blood-based biomarkers for traumatic brain injury. This is a monumental achievement. The fact that this runs on a platform that takes three and a half hours is not going to really be useful to most of the people in this room. But the fact that the biomarkers got approved is really, in my opinion, transformational because they don't have a blood-based biomarker for Parkinson's. They don't have a blood-based biomarker for epilepsy. They don't have it for CTE. And yet, we have one for traumatic brain injury. Now, what's happening is that now that this has happened, there's a lot of companies that have been working in various areas that are really jumping into this. And so, David and I and other track investigators have really been starting to see, okay, well, how would you actually use this? Because, I mean, if we think about it today, if you've got chest pain, they're not asking you, you know, trying to come up with some definition of your chest pain. Is it mild chest pain? Is it moderate chest pain? Is it severe chest pain? They're going to draw blood and see what your troponin looks like, right? They're going to stick an EKG on you. I don't even know how often they're using a stethoscope these days, you know, because their objective biomarkers, EKGs and troponins are so good that that really just gets rid of sitting there talking to somebody for 15 minutes. The troponin's either elevated or it's not. You've got a T-segment elevation or you don't. So then, this idea of being able to actually have a point-of-care device that you can measure these proteins that have already been approved by the FDA really provides us with, I think, what will be the next generation of diagnosing these and managing these patients. So, I'll show you, this is just from, okay, so I think up to now, the three sites that we're using this device on are Pittsburgh, Baylor, and UCSF. And we're just about ready to expand out to another seven sites in the network. And so, the idea here is, is that this was a patient that I had right before Christmas, a 63-year-old person. What you can see here on the right is that there's a drop of blood there. This is a microfluidics cartridge. This cartridge already has been approved by the FDA and also has CE marks for many of these analytes. So, you can do troponin on this. You can do electrolytes. There's many things that you can run that are already approved. So then, what you do then is on this little cartridge, you put a drop of blood there, that little plastic thing at the bottom, you slide it across, and you essentially then stick it into this handheld point-of-care device. And so, what you can see here, this particular patient has a GFAP that's greater than 30,000. Let me just say that nobody in this room wants a GFAP greater than 30,000 if you come into this. Now, for those of us who have been doing this forever, if you look at the CT scan, you say, well, Jesus, you know, you don't really need a GFAP of 30,000 to look at that scan and say this patient's not going to do well. The problem is, is that this patient, in particular, was waiting behind two other patients who got CT scans before this patient got a CT scan. And it would have been nice that even in, you know, a very well-known level one trauma center, if the emergency medicine physicians had a way to screen, not who doesn't need a CT, but maybe who does need a CT. So, what we're beginning to see now is that we can actually start to stratify patients based upon these blood-based biomarkers. The UCHL1 is also very important. I'm just showing you GFAP, but UCHL1 comes on very early in the process, within the first, you know, four hours. And the GFAP tends to peak a little bit later, but in concert, they're actually more powerful than either one of them working independently. So, I think that, you know, what you're going to see is Abbott, just now, based upon using some of our data, has gotten what's called an expedited approval pathway at the FDA, and they will be launching a phase one, they'll be launching a pivotal study soon. So, where are we today? I think we've got a lot of people working together, whether it's the folks in Center TBI, Track TBI, our Canadian colleagues. Andrew has been an amazing person going all over the world and traveling. He now has China on board and is getting data from the Chinese. The Indians are trying to collect data. The folks in Jamie Cooper and some of our friends in Australia have collected data. So, he's truly made this an international effort, and we're just now starting to talk about how can we share data together. I think this is, you know, this is not different than any other big time problems in science. Nobody was ever going to figure out particle physics working alone, and if you look at something like CERN, you get 608 institutions, 113 nationalities. You know, big ideas and big fundamental change in how we do things actually happens when people work together, not when they compete with one another. So, I do think that the future is international collaboration, and I'm glad that you allowed me to reschedule this at the end of the day. I do not want to keep you guys from having a great evening. If you can't have fun in New Orleans, you can't have fun. So, I, but be careful, because we know what the consequences of that are, too. And since all the trauma neurosurgeons are here, you know, be safe. All right. Thank you very much. Thank you.

traumatic brain injury

GFAP

UCHL1

diagnosis

MRI scans

international collaboration