Transcript Video Understanding Limits: Impact of Clinical TrialsProf. Benjamin Abella < Back to Boundaries of Temperature Session 4: Limits Understanding Limits: Impact of Clinical Trials Presented by Professor Benjamin Abella ladies and gentlemen, it's my great pleasure to introduce for you the next speaker which who we already know from yesterday. It's Professor Ben Abella. He is at the department of Emergency Medicine and at the Department of Internal Medicine and Critical Care of the University of pennsylvania. He is also the research director for the Center of Resuscitation Science. He's focusing on experimental and clinical studies. He has done a lot of publications over decades and all of them are very important and all of them are very good and most of them are about categories and re perfusion and T. T. M. So Ben. We are very much looking forward to your next talk. Well thank you and good morning, so you heard from me a little bit yesterday. So this will be my second talk and yesterday uh we didn't get so deeply into the notion of the clinical trials although certainly heard some about that this morning. The goal of this lecture. The charge I was given was to talk about clinical trials more generally and what are some of the limitations, Some what are some of the ways we need to think about clinical trials. And then in the second half of the talk will get a little bit more into how those concepts apply to the trials around T. T. M. And this is really a set up for my friend William Beringer is going to speak later today and patient selection which I think really follows from this notion of the clinical trial design and interpretation. Um so to begin I thought it would be interesting to spend just a few minutes talking about the history of clinical trials, where did they come from? Because you know, we, we think of them as really the gold standard in modern medical clinical research. The first clinical trial was in the 17 hundreds and then this guy James lind who was a Scottish physician, I was studying scurvy. So scurvy was a big deal uh in in older times because much of the commerce of the world was maritime. So people were in ships and they were in ships for long periods of time And it turns out they would develop scurvy so their hair would fall out, their bones would go bad, their gums would bleed and no one understood what this was from, but this had a major, major impact in the world. And so he did a clinical trial and equals 12. He literally enrolled 12 sailors and ran quasi randomized them in pairs of two different therapies. So two of these 12 people were put on a diet of limes, citrus and they were given citrus and he published on this. And so this is in a way, if you ever want to find the first ever publication of a randomized trial, this is it In 1750s and it's a treatise on scurvy and what he found was when these British sailors who were given lines, they dramatically improved and they're scurvy almost nearly resolved within a week. Um the other treatments were a little bizarre to modern ears. He put some people on a diet of seawater. He put two p if you can believe it, he put two people on other sorts of strange things and it didn't work. So, so that was the first, the first clinical trial. So it's been around this concept for a long time. Um but what follows from that Lynn trial where N equals 12 was one might legitimately ask, were those patients representative of all sailors? Well, he picked from amongst the sailors on the ship. But you know, there are thousands and thousands of sailors on the uh Atlantic Ocean back then he had 12 patients. And this question comes up very large and modern clinical trials. Are we enrolling people who are representative? And this is there's two-related concepts. One is generalize ability, that is are the people enrolled generalize able to the whole population of people with that condition? And then there's applique ability, which is for the patient in front of me. Does this trial apply? So they're subtly different but they're related. And to show it to you graphically. So we have we have a therapy here and we have a patient population, let's say in this case, all the people who get cardiac arrest. Well, obviously we're not enrolling all the people who get cardiac arrest in a trial. But we don't roll some of them. And so the question, the obvious question comes about are these patients representative of the universe of patients who we might need to treat in the future with this condition. Okay, so this is this notion of generalize ability and in a way applicability is the more important one for many of us as clinicians because we in a sense are doing a clinical study of n equals one right. A patient is in front of us that we have to treat and we have to interpret the evidence and how it applies now to point out the challenges and the fallacy of this. There's this is a bit of comedy here for you, but it's excellent. A number of authors in United States wrote this article, a randomized trial of parachutes for jumping out of airplanes. This was actually published in the british Medical Journal. Um, it's very tongue in cheek american english. That means sort of written fake serious, but it's actually sort of a joke. But they are clinical trial methods here to point out a major flaw with many clinical trials, which is this notion of generalize ability. So they asked the question, do you need a parachute and you jump out of an airplane? And they randomized subjects to either jump out of an airplane with or without a parachute. And they found that everyone survived and you didn't need a parachute. So parachutes were irrelevant. What's the catch? The catch is it turns out the only people who agreed where people where the plane was on the ground. So no one jumped out of an airplane at height. And you see here very sort of dry humor. I'll just read it. They said, you know, the the people enrolled at a significantly lower altitude, a mean of 10.6 m as opposed to the other participants in the air at 30,000 ft. So, and here's an example of one of the subjects testing the parachute jumping. She jumped out of an airplane, right? She's jumping out of an airplane. And it turns out even without the parachute, she did fine. She did very well jumping from one ft. So you see the point, the point is the devil is in the details of who you enroll and whether their representative if you're now in the in the air. And there's a problem, does this trial apply to you if you jump out of the airplane at 10,000 m. Um, and I think you can all understand that the answer is not. And and there's another really good, I'm gonna kind of find it to read it. Therefore, although we can confidently recommend that individuals jumping from small stationary aircraft on the ground do not require parachutes. Individual judgment should be exercised when applying these findings at higher altitudes. So I highly recommend that it's it's a great read and it's a lot of fun, but these are very serious investigators who are making a point here. Um, Now, the other major issue that we have to think about when we think about clinical trials is the timing. And I think you've heard intermittently through this conference some suggestions that in some of the cardiac arrest clinical trials and I suspect others in T. T. M. T. B. I. This has come up as well. Is are we hitting the therapy in the right window? That is does the biology match up with the intervention such that you can actually intervene at the right time. So you might imagine a condition there is a window of opportunity and then after that it's just too late. Just nothing we can do will impact outcomes. And so again, just to play this out, let's say you enrolled people within this circle here outside of the therapeutic window and maybe you don't know after all many situations, we don't actually know what the therapeutic window is not well defined. That creates a problem that when we actually enroll maybe we're missing that window and it really has to do with the biology of the underlying condition. And so to show you an example of this. Now let's say we did Now this next trial doesn't exist. Okay, so we're gonna I think that the american word from the german Duncan experiment, a thought experiment. Um Now we're going to fix the generalize ability. We're gonna do a trial of people jumping out of that high altitude. Great. So we fixed the generalize ability. We're not gonna do this ground stuff. And now we're going to randomize people to parachutes or no parachutes jumping from a great height. The only catch the randomization happens when they hit the ground. So now we're gonna have the parachute either open or not open once the patients this altitude sensor says they've actually hit the ground, okay? And what we're gonna find is 100% of people die in the parachute group and 100% of people die in the night. I know there's a bit of a sick experiment. So forgive me. Um people die in the parachute group and people who die in the not parachute group. The conclusion parachutes don't work right because we've set the timing way too late for the therapeutic window. It just it just won't work. So you see this could very well apply in many of the clinical trials. We read if you see no treatment effect if the timing is wrong. And I think this has been a major issue historically in a number of trials of T. T. M. For a variety of conditions by the way. Um Not just the one that I I study which is cardiac arrest. Okay, so now let's turn with these principles in mind. Now let's turn to some of the studies around cardiac arrest. And I showed you versions of the slide yesterday. This is essentially a summary of the major randomized clinical trials and T. T. M. For cardiac arrest. Three. Clinical trials show benefit of T. T. M. 2 33 and two studies. And again like yesterday I'll apologize. The capture is not quite right. It's hard on PowerPoint to summarize it. Well one of the two neutral studies showed that 33 36 were equivalent and one showed that 33 therapeutic norma ther mia were equivalent. Um So 23 studies showing improvement to studies are neutral. Okay so that's that's essentially the sum total evidence from major randomized trials in postgres T. T. M. Now when we look at the biology. So so again this question comes about that timing issue. The biology underlying it. The animal laboratory literature is quite clear. It's quite clear that cooling to 33 or lower is beneficial. So there's a strong dose effect. And of course the advantage of laboratory evidence as you can get at mechanisms so they can do brain biopsies, hurt biopsies uh look at blood markers so you can actually see the effect. And there's good mechanistic biology for Y. T. T. M. Would work again too long to treat in a 20 minute talk but things involving for example mitochondrial dysfunction, a membrane stabilization, inflammation, glutamate release and neurologic side of toxicity. A number of mechanisms whereby T. Tm. Can improve outcome. But one of the other important findings from the animal literature is that the earlier T. T. M. Started the better and there's clearly therapeutic Windows and we'll get to that in a little bit Before we do though. I did want to show you a slide I showed you yesterday which is what I like to call the scoreboard. Um so there are no studies to date showing that cooling the 33 worsens outcomes in cardiac arrest. Now again like yesterday I apologize. This is a unfair scientific summary because all these studies are different sample sizes and different qualities. But nonetheless at its simplest conception. It's important to remember that some studies show benefit of cooling for post cardiac arrest care and no studies show a harm a outcomes related harm a clinical outcomes related. Um Now now why might this be the case? Why is there this discrepancy that is much debated between these two columns between these studies and some of it has to do with this generalize ability and applicability issue. I show you here just a few of the clinical parameters from the studies and I recognize that Green may not be showing up as well as I'd like. So the three studies in Green are the studies that show that cooling to 33 has worked and the two studies have shown that it's it's neutral. And when you just scan across you can see for example in the T. T. M. One and two studies much higher rated bystander cpr than the price is much higher. A much higher rate of stem E. Requiring P. C. I. Which I'm still struggling to really understand. Well honestly because what we see in the US cohort of postgres patients that I see in philadelphia is maybe about a 5 to 10% stem E rate. So how they got a phone, 40% stem e rate suggests to me it's a very different patient population with with a variety of other confound urz. And then importantly, when you look at the shock rate, the rate of shock and each study defines shock a little differently. So again, I'm being a little uh simple here, but the rate of shock was very different in the studies. So the prior studies much higher rate of shock than T. T. M. One and two. So again, not necessarily representative of all patients with cardiac generalize ability issue. And actually as I'm up here today, um right now I'm getting texts from my team, we're managing a cardiac arrest patient that I'm gonna call them right get off the stage. Um We're applicability is an issue. This patient is in shock with allocated 15 after about 30 minutes of cpr and ongoing intermittent rask. So the question is, can I apply T. T. M. One and two data to that patient and I struggle as whether I can. So this is a very important question that you have to ask when you treat a patient after cardiac arrest, do they look like the T. T. M. One and two patient population? And I would suggest if you have a patient in front of you who had by center CpR witnessed arrest, likely stem e. Um maybe therapeutic normal for me is fine. Maybe. I mean they fit the T. T. M. One to patient population. Unfortunately I rarely see that. Um I mostly see people who either did not get bystander CpR got very little. Maybe they had none witnessed arrest. And also by the way, I might point out that when we say bystander Cpr it's not a binary variable. We think of it as such. Right? They got bystander CpR or they did not. Well, I'll tell you that much of my career has been studying CpR quality and and the one thing I can say that I know to beach true is not all by center CpR is created equal and just the fact that is delivered doesn't mean it's delivered well. And so a further question that none of us measure well enough is in that community. How well is CpR delivered? Um And I can say unfortunately us even when delivered is often delivered quite poorly relating therefore to lower flow, more skinny repression injury and so forth. Now when we talk about timing, it's important to think about the literature from the laboratory a little bit. I'd mentioned this briefly but now a bit more detail the laboratory studies of T T M. And post cardiac arrest. Clearly show that timing is very important And indeed Peter Safir one of the founders of this field uh believe that you really had about a 15 to 30 minute window, that was his opinion. But just to put it out, there is a stark comma terry on this. He felt having studied this for decades and animals that if you started cooling an hour or two later, you may as well not do it Now. I don't know if that's actually true, but that was his perspective and certainly it's backed up by some of the science. If you look on the right here, this is a laboratory study looking at skinny re perfusion injury. And don't worry so much about what the markers are and what the specific size. I'll just tell you that a number of studies including this one have found that a number of the biochemical markers of the triggers of ischemia provisions injury are set off within minutes, within minutes after resuscitation. So there clearly are changes occurring at the cellular and moral level within the 1st 30 minutes after arrest. That may not certain may bake in fix in some of the injury that may be less less easy to treat. Now, clearly we know that's not completely true because we do see some benefit with later treatment, but it does speak to this issue that there is a trigger that occurs early and that timing may be important. So this is this window in this one study in cardio Maya sites from one of our laboratories. Now to take this one step further. Laboratory investigators including our team looked at whether if this was true. If it's true that timing is so important. Could we cool during arrest itself so that the re perfusion phenomena occurred when cold. Now this is very difficult to do. And back when we did this work. The technology was not there for instant cooling. We still struggle with this in clinical care how to cool very rapidly. But the good news is in mice it's very easy. Mice are 30 g cooling. A mouse is a very simple matter. And so we could cool mice during so CpR and what we found in other groups on this as well that if you cool during arrest before CpR you could have dramatic improvements in survival. Very qualitatively different than post Roscoe cooling. So we call this cold re perfusion. For those of you know about cold fusion. It's sort of a planned words. So if you re profuse cold it may be a real game changer. And this has now been taken to the clinical environment. And some very brave investigators have pursued this in europe. If you don't know about that, Prince and Princess studies, I encourage you to check them out. This is using a technology. This is the first time I mention a device by name but there's only one like at rhino chill. It's not commercially available in the United States is still very much under study. I don't use it. I have no relationship with the company that makes it, but I'll tell you that this is a device that can provide very, very rapid nasal based cooling. Um, and so the concept is, could we use this to cool during Cpr or immediately after resuscitation And in the princess trial that she looked at intra arrest T T. M very brave and very difficult to do in the field. The results were not statistically significant, much to the chagrin, I'm sure of the investigators. But there was a big trend and noticeable trend towards better outcomes in the group that got cooling during CPR. So a tough thing when you're, when you see a trend. So essentially you're an underpowered study with all the confound ear's in out of hospital cardiac cross care. But that's where we're at now. A sub analysis was done of this and they found that even cooling during arrest, the timing mattered. So if the cooling was started early during the resuscitation episode, there seemed to be a much bigger signal for benefit then later, which again stands to reason if we believe this timing concept. So, uh, this hammers home the point that if we're starting cooling hours later and we see a benefit, we may have just gotten really lucky because the more likely outcome is we're just too late. So this is a really important thing and and my prediction is that intra arrest cooling is going to be an important topic of research and conversation in years to come as technology improves, including for example, the use of ECMO as you may know, a number of places, paris is as a real leader in this. A couple of places in the U. S. And Asia. They're doing this inter arrest ECMO in the field and this may open up whole new technologic opportunities to improve rapid TPM implementation. Now, one of the big issues with clinical trials also is in this applicability is just the real world experience. And so I think it's important that you're aware that there have been a number of real world experiences. I wouldn't even call them studies but cohort evaluations that that give us pause as to whether we can really fully have full confidence in these trials as they apply. So this is a study out of Melbourne Australia where after the T. T. M. One study, they switched their protocol to 36. And much to their surprise and I suspect alarm survival worsened after they switched to 36. So now was it because of the temperature? Was it because of other aspects of care? It's hard to know. But all I know is in the real world when they switched from 33 to 36 they hurt people, more people have bad outcomes. Now if you'd like a confirmation of that. This was done in the United States. The exact same concept hospital switched to 36 survival worsened. They switched back to 33 and they're now devout believers in 33 because they saw in front of their own hands with their patients that this worsened outcomes. And so what I think it speaks to is under randomized trial conditions, things are different. You know, things are more protocol eyes, you have coordinators following up, uh, you have rules that might violate. So they're taken out of the data analysis in a modified intention to treat type analysis. Whereas in the real world, it's two in the morning with clinical care providers, they're doing what they're doing. It's just a very, very different environment. And so these are some of the real world things that give me a little bit of hesitation to fully adopt or endorse these newer T. T. M. One and two studies, because I've seen what's happened in the real world, And I'm just gonna give you a preview of what will is gonna talk about later, which are some of the newer works have come out suggesting that it has to do with injury severity. This is one such study out of Pittsburgh, I won't go into detail, but they found that more sick patients, more injured patients did better at 33 less injured patients did okay, at 36. And so this speaks to this, this big picture where, you know, a trial comes studies come out, we think something's great t t m works for everyone. Then the studies come out saying, oh it doesn't work at all. And then the reality always is a little bit more complicated. It's the road less traveled. This more complicated world of well, maybe it works, maybe it doesn't work uh which people don't like because we all like simple answers, but biology and certainly critical care are not so simple. And so I mentioned this yesterday. What? I didn't know that there's this implementation gap. What I didn't mention is we've started a program at Penn is a shameless plug, although it's not a business, I make no money from it. It's just an educational program. We started something called the Pen T T. M Academy, which runs courses on cardiac arrest implementation and TPM implementation. I think that your code even works on the screen. If you want to take a Picture of it, you're welcome to do so. But we have podcasts and we do online virtual courses and have other resources available. And indeed our next course will be October seven if any of you, it's online. So you don't have to travel to Philadelphia. Of course, time zones may be an issue, but you're welcome to check that out as well. And we go into great depth on how to implement high quality post arrest care. And with that, I will stop. Thank you very much Created by
