Transcript Video ProCon Debate: Should We Just React in Post Cardiac Arrest? < Back to Summit23 Cracks in the Ice ProCon Debate: Should We Just React in Post Cardiac Arrest? Presented by Professor Christian Hassager (Pro) and Professor Wilhelm Behringer (Con) Moderated by Professor Fabio Silvio Taccone Uh So I had the pleasure first to welcome uh Christian Asger from Copenhagen. So Christian work in the University Hospital of Copenhagen responsible for the card board member of the European Society of Cardiology being chaired at any Sharp Foundation and you have seen his name a few times in New England by the way recently. So in the box trial, Chris will talk, he will be the pro in this debate. And the title is, should we just react in post arrest, the fro yours and then it will be the come from William Ben. Thank you so much for this kind introduction. Honestly, giving this talk here makes me feel a little bit like a cuckoo cup in a foreign nest. But I'll try to go through the same trials that Professor Brier did this morning, perhaps to a more with a more academic approach or at least a more trial approach. These are my conflicts of interest and I've also received honorary for standing here today, but I have a confession to make. I was once a believer I really was, but I am now much in doubt and I'm also a believer in randomized clinical trials. I really am not observatory data. I'm here today to uh be a little provocative also to, to, to tight up the, the idea that we should just prevent fever. Having said that I have no doubt that marked marked hypothermia works. It works during cardiac surgery. We use it every day transplant, organ preservation when they have to go from one hospital to another or accidental hypothermia. We heard case stories just a few minutes ago. But remember here, cooling has always been applied before the circulatory arrest. The question is, does it also work afterwards? Just to uh give you a case story from my country that accidental hypothermia works. Uh about 10 years ago, a boat like that one capsized in Denmark in February and it's very cold there. So these seven of these young students, they were around 15 years old fellow in the water and didn't make it to the to the shore and they were found floating with head down in the water with a core temperature between 16 and 20 degrees. You can see the time before there were any help was more than an hour hard lung resuscitation was 1 to 2 hours and time to ECMO was three hours. But they all survived before they were cold before they got the cardiac arrest. They all, some of them are going into high school and get an education but some of them also severely damaged, but they all survived. This is what changed our guidelines 20 years ago, we all know these curves. I don't have to spend too much time on them. There was the Haka trial, which was a real randomized trial. Uh, and you can see the numbers there about 100 and a little less than 100 and 40 in each group. And then on the other side, there was the beard trial which was quasi randomized, not really a randomized trial. I would call it two case series which shows something similar in the similar direction. So does this end the discussion? The trial was of course not a blinded study. It was stopped early because of funding and, and, and slow enrollment. Actually, when you divide the time and the number of senders, they only enrolled six patients a year per sender. So it's very slow enrollment back then and they compare a brand new method towards more or less leaving the patient among the other ones in the ward. That's my postulation. And you have to remember this is our C lab when we start a new therapy that new things really get more focus, you concentrate a lot. Could that be part of the effect? We don't know, I was part of the TTM or the so called TTM. One group that tried to set up a trial where we compared 33 and 36 and here both patients received treatment. They were, they looked very much similar and we saw absolutely no difference either in survival or in these very rough measures of, of cerebral function. And I would say many of these trials have just as an end point C PC when going home from hospital. I think that's a very poor outcome. It's very subjective, it's very early in the face of the patient. I think we have to look at the survival curves honestly or invite the patient in for details neurological assessment. Later on now, if cooling should work, then we all started to discuss maybe rapid cooling is better then. So there were different uh different trials here. I've just mentioned the four early trials. You can see with numbers almost as high as in, in, in the hacker trial. There were absolutely no difference whether you could pre hospitals or in hospital even though you saved an hour or so. So five years from now, we had the situation that we had the haka trial showing an effect. We had the uh Bernard trial showing a SIM similar effect but not significant. This is survival only survival but we did not see any dose response. We've seen some of these trials, we just saw the TTM trial. So 33 versus 36 no dose response. You see the T th trial earlier today for for 24 hours or for 48 hours, no difference. In fact, some of the trials that showed the largest difference only had four hours, it was an ice cap. So no dose response. And if you go out here are some of the larger prest trials you can see. Now we have, uh, now we have or more than 1000 in each trial, no effect of free hospital versus in hospital. And then came the uh the TT M two trial. And here we have almost 2000 patients randomized to 33 or stay below 37.5. We have heard today that perhaps if it helps some patients, we should do it because there's only minor, only minor side effects a little, I agree. Only minor side effects, a little arrhythmia, electrolytes, perhaps infections. It's very little. But if you see the two curves here and then I like that and then you have a significant subgroup where it works, then it must mean that it also kills just as many in another subgroup. You just can't decide those conditions. Now, what we, the question is, are these curves really so different and yes, they are different. You can see I've drawn a very narrow line through 37. The, the, the TT M curve show 95% confidence intervals. The other one shows standard errors. So they are not that different regarding how well the temperature was controlled, but it's only a very little difference. And it may be that the, the fact that they go above 37 is a problem we have that is left still to find out we combined the TTM trial that I was part of and the TTM TTM two which I was not part of combined all the individual data, every single data point, every single patient, we had all the data and you can see the curves there. Now we have nearly 3000 patients randomized. And on the other side of the screen, we have these uh uh MA RS and there's absolutely no difference between the two curves. What we have discussed earlier today too is, and I believe this is true, but there may be somebody who is simply like Christian Erickson. That picture was by the way, taking up my ward uh that are too good and we don't need to cool them down. There are patients who will die anyway, whatever we do. So what we tried with these 3000 patient was actually to divide them by miracle score. This is the scores whether they have AC T or not witnessed cardiac massage and so forth a score for that ph lac a all that stuff. And you can see if we take those with a low risk, both normal tia and hypothermia stays below 20% that die medium. About 50 high risk patients have about 80% metal, but it's completely similar in the two groups. So there's no sign whatsoever. The TTM works in the middle intermediate group in a randomized trial where you have patient specific data. It's not a meta analysis made from a mean value. All the raw data perhaps we should try to. And I still believe in that, that if you can really somehow make it, that you cool the patient down before any of the neurons feel that they are suffering, then I think we could work. Remember the kits that were called, if you can apply it extremely early or even before then, of course, it works. We had this uh uh princess trial. It's very good to know that it's a princess to running because the first one, this is their primary end point was, was neutral. And then I would like to say, when we say this, this is the temperature curves for the, for the hacker trial, you've seen this figure many times I believe. But if I draw a line here, please note it took 12 hours, 12 hours to reach 33 degree in the hacker trial. The only trial that has really proven that it works. So putting this together means that we do today, do not have anything that, that really scientifically in human shows that early treatment works. I'm sorry to say that that's the way it is. I hope we have it five years from now. Should we go deeper? There's a Canadian trial larger than the hacker trial, 34 versus 31 no effect at all. What about the Non Shaggs? Most of these trials has been dominated by Shal patients and we have the French hyperion trial about 500 patients. And then you can, you can see that most of these patients die. But there is a small significant difference among those who survive a very small one. And it's uh it's actually a very fragile statistical test that they have done. And uh it's not the same test in the journal as it was in the, in the document stating what they were going to do. So it's very fragile, but there seems to be more people doing well. There later on, they published their neuron specific inner lasers and there were really no difference between the two groups there, which is kind of strange because if this works in the hacker trial, you can actually see NSE is lowered by the either the temperature treatment. Furthermore, the effects in this was not in the out of hospital cardiac arrest. This is a mixed trial. It was only among the patients who had in hospital cardiac arrest and we actually have a small trial just in those patients. Now, the German trial, which is has this very nice name the in hospitals categories. Hacket trial where the randomized 249 paces to either 32 degrees or keeping a normal theia about 25 were shock. This is the usual in hospitals, cardi arrest group and they didn't see a signal again, perhaps too few patients to see a relevant signal. But there's no sign that it works. Furthermore, hypothermia may not work because there's been other trials may not work during uh surgery for intracerebral hemorrhage after brain injury. After a stroke, in severe bacterial meningitis, it may even, it may even hurt status, epilepticus or sepsis. There's a lot of other cases where the large randomized trials fail to show something. I'd like to spend the last uh few minutes on discussing what after the TT M and uh Ilco as well as well as this uh uh nice paper uh which is E IC and ES IC MS guideline on temperature control. They say we recommend actively preventing fever for at least 72 hours in post arrest patient who remain comatose. It's good practice statement. There has until recently been no randomized trial whatsoever and trying to find out when this started, it started. First time you can see it is in 2005. In the statement that 72 hours is recommended, we did a uh a trial in 800 patients. Uh the box trial where our primary focus was oxygen and blood pressure targets, but we also randomized and this both of these interventions were neutral. We also randomized the patients to device based fever prevention after c arrest duration of that. The way it was set up was it that everybody had 36 degrees. That was what we did when we started the trial in 2017, 18, we uh they started for 24 hours. 36 degrees. And then we left the ca the cooling catheter inside and then they either had an additional 12 hours or 48 hours or, and we did wake up calls and if they woke up we just stopped. So we, they basically, we test 36 and 72 hours of uh active device based fever prevention strategy. Keep them on 37. This is our population. As you see, they are a fairly good survivor population, fairly benign disease. If we say so many bystanders, we've been talking about that lately. Most of them shagg about rhythms and only 20 for 21 minutes to rusk on as a median and uh patient who were in coma. This is just the subgroup that were in coma. About 40% had the fever in one group and about 20 in the other. Actually the red one there is very super is just about the same as what happens in the, in the TT M two trial. You can see that in the supplementary figure. And uh here we have our primary end point for that paper which is CBC three or four at discharge or death within 90 days. It's in my mind five years later here, I think it's a crazy end point. It should just have been survival which actually shows the same thing completely neutral. The survival curve looks very much like this. So then I have two minutes about about mesa analysis. I think it's a good thing if you have individual patient data meta analysis, I think that's worth doing combining raw data from trial, but just to take the mean and SD from tables and combining it, that makes me think that just as analyzer and meta analysis and it's the same as physics and metaphysics. Let me show you an example that is uh written by Professor Beringer who will come right after me uh showing that uh cooling is better. You can see it out here. What I think it is important to note is that when you just take these groups and you do a random effect modeling, then small trials have a much higher weight than larger trials. Let me focus in on these three trials. What you see there, the percentage written out here, it actually the weight that this data has in the analysis. So if you combine the hacker and the beard trail, then you have, let's say about about uh 300 patients, 350 patients compared to the TT M two trial, which has 2000 patients almost, but you can see they account for 40% and this one only 30 it makes no sense. So I acknowledge that tt might work, we just haven't found the way to do it, but at least we have not been able to show a dose response at all, not on temperature target, not on time to target ha did 12 hours before they reach 33 degrees and not at time at Target when our trial was 12 hours. Many trials 2024 and uh the one from Denmark, TT H 48 no difference. So in conclusion, TTM at lower temperatures has really no documented beneficial effect. I'm putting this to a little point. I know fever prevention for 36 hours is enough if needed. A t all, I think we need more trials and occupations, perhaps especially those with a history and noncardiac courses and TT M maybe even fever prevention seems to belong more in trials than in daily clinic. Thank you Christian for making the points very clear. Now, it's my pleasure to welcome William Behringer. We will have the discussion just after William Behringer is the head of the emergency department starting from Vienna with the trial. Then to Vienna, then back to Vienna, of course has been in us having a lot of experimental research in very severe condition has been fellow with Peter Safar. So no better than him to take the car is being reactive, please. Well, uh Fabio, thank you very much and I drive if the mouse is working, unfortunately, the mouse is not working doesn't matter. Yeah, you spoke about uh Peter Safar and that's a good moment to say what you always said because we see a lot of slides repeating and you always said retention through repetition. So the more the often you see the slides that the better your retention will be. He always said, he also said very often inspiration by titration. Titration meant red wine every once every week, we had red wine for the lab meeting which was forbidden university, but for titration, it was allowed for us. So uh I have now the pleasure as Christian here to take the other part. And these are my conflicts of interest. I get honoraria from these two companies and um a lot was said today and I think which was not mentioned so far when we speak about the temperature is time to start cooling. This was missing in your slides. I think this is a very important issue. And if you look at the hacker study, the time to start cooling was 100 minutes in a million. And the Panna study, they started cooling already out of the hospital. So maybe this might be a little bit different to the TT M studies where cooling was started four or five hours after RC. However, we heard a lot of limitations of the TT M studies were burned. We heard limitations by the hacker study by you. And now I will try to make a little bit of sum up my point of view. So my rational for 33 degrees and um there is another way to show the complex part of physiology was what happens after successful resuscitation. Uh it starts immediately with resuscitation, all these cascades leading to cell death to apoptosis, necrosis. And there are tons of cell culture studies showing that hypothermia a range of 33 to 32 to 34 degrees influence all these cascades and uh beneficial for the sales. Then the next step in evidence are the animal studies and a colleague of mine, we together made a meta analysis, meta analysis of all the animal studies. Approximately 60 animal studies and what we took, we took only c arrest studies. So no animal studies where you can ligate the brain arteries, no only categories studies and and not only looking at uh histology but also to neurologic outcome. As only studies with neurological outcome, it's difficult to read for you. I give you the summary, the summaries overall in animals mortality is reduced by almost 70%. So very powerful treatment in animals. Then we heard that the guidelines after the TTM study changed and many hospitals also changed their temperature management from 33 to 36 degrees. And uh again, a meta analysis and for the ones who are not so used to read me analysis, why is the mouse not working? However, I try to do it like this. How do you read a meta analysis? The ones who are not familiar with that? I will go through it very slowly. You can see here the singing studies and the blue is the effect of low temperature or high temperature. And the line here is the 95% confidence interval. And if this line crosses one, which is no effect, then this effect which is shown here is not statistically significant. If you look at the first study, the blue point, there is an effect for hypothermia, however, not statistically significant. If you look at the 12, 3/4 study, you see an effect. And since the 95% confidence interval does not cross one, it's statistically significant. And the diamond here is the summary effect of all studies. And the ones who know statistics know the more end in the study, the greater the power of the study, the 95 confidence will become smaller. And if the diamond here, this is the 95% confidence interval does not cross one, the summary effect is statistically significant. So that's how you read the meta analysis. And this is a meta analysis of retrospective studies comparing 33 degrees before then changing to 36 degree each study. Thanks. I tried it like this. So uh what I wanted to say is so the summary effect of all this retrospective study was if you change to 36 degree neurologic outcome is statistically significant worse but only retrospective studies. This is also retrospective studies. It was shown today shortly. I don't know who showed it. I think burn. You showed it. I don't know who showed it. However, ah no Raymond you showed it right? I think it's a very good study. It's a large study from Australia, New Zealand, almost 16,000 categorized patients and it works great but very slow. However, I come here but you can see on the YX is the temperature. And over the years from 2005 to 2013, the temperature slowly, slowly, slowly decreased. If you look at the Y axis, the difference is only one degree over time. And then when the new guidelines were published, changed to 36 degrees temperature changed. But if you look at the va again, only very little from 34.5 to 35 degrees, if you look here, it's 3435 all only one degree difference, only one degree. And if you look at the mortality which went down over the years, new temperature, only one degree difference, only one degree, not only that the slope stopped, it also the mortality went up again. So these are real life data from 16,000 categories. Patients, what happens in real life different from randomized trials. But I think both is important to see randomized trials and to see what happens in the real world without randomization. So this should be something to think about. And now I come to the to the guidelines and I agree with you meta analysis are difficult. However, the guidelines are based on a meta analysis by Graf which was published in 2021. So this is the basis on the recommendations. And now you already learned how to read meta analysis. Ah Grenfell divided the studies by outcome by outcome reported at discharge or 30 days is the upper graph and the lower graph is outcome reported at 90 or 180 days. And there are few studies in each group and the diamond in both sub meta analysis, they show in effect favoring hypothermia. However, this effect was not statistically significant because the diamond crosses the one but there was an effect, it was not statistically significant. There are only two studies reporting the outcome at two time points, but you have two met analysis with very few studies included. And the statistician among you know, if you decrease the number of n either in a study or in a meta analysis, the power decreases, this means the less N you have in your study, the greater the effect has to be to be statistically significant. And we already showed before that in critical care studies, the outcome over the over the time does not change. And if you look specifically in categorized patients, it was also shown from our group, that outcome might change between C PC one C PC two changing. But overall the outcome over time does not change. So it's not necessary to divide the meta analysis into sub met analysis and decreasing power of each analysis. Very recently. This was shown today twice already. Uh The Cochrane meta analysis met Cochrane is known to have a very thorough review process. It's very difficult to publish things there. The reviews are very, very picky and however, very recent a new me analysis on all randomized trials was published. And uh in the A A you see the comparison conventional cooling versus no cooling or fever control and you have the single studies, some of them show a benefits, statistically significant, some of them are neutral. However, none of these blue boxes are on the left side showing a tendency of better outcome in no theia. So the effect, all the studies show an effect except the TT M study, there's no effect but no study shows harm, not in the direction of harm. And if you look at the summary effect, then you have the group with cooling versus 36 degree. There are only three studies, one is neutral, one shows an effect, not statistically significant. One shows a statistically significant effect and the overall effect is not statistically significant uh many meta analysis, but the summary effect, all the evidence together for the years and maybe there's really no effect. Now, I don't know. But if, if I take together all the evidence together. Now cell culture, animal studies randomized trial showing a benefit, randomized ST didn't show a benefit. If I'm in doubt, I will give the patient the things I think they can benefit from. Even if it's only half of the patients benefit. As long, it does not harm. I will give them the chance because there's no study showing harm in terms of neurologic outcome. And as long as we don't have a study showing harm in terms of neurologic outcome and I have evidence might help. So I give the patients a chance, listen, that situation you're describing. There is the perfect one for doing more science. That's exactly where we should not lean back and do clinical work. We should randomize the patient in a more correct trial. The way you would like to have it actually the time to reach target temperature from, from cardiac arrest until you reach, that was about half as long in the TT M trial as it was in the hacker trial just too. I mean, so it's not just time to time to have targets. Furthermore, we tried actually in the uh in the uh TT M trial and we found a population that was very similar to the hacker and took it out of the trial, sort of make a hacker within the trial and there was no effect. So I can, I can really not explain how that could be. But if you're correct, it's 20 years ago, it's two different environments. Perhaps, perhaps mortality was different, different population that we're studying. We don't know today things have changed somehow and we discussed it in the, in the break that it's also the study population in, in, in, in Vienna. We have 50% of the patients have we or less than 50% which we admit and we have more tm is 70 80% of. But that's because the inclusion criteria, we excluded unwitnessed as so. So there, there was different inclusion and exclusion criteria. So I think, I think we both believe that if we could cool the patient down immediately, it would work. Definitely, it would work. We're just not sure that the method we're using today are sufficient to give this effect. And uh I think also we agree that we should at least avoid fever. Just I would like to do dare a great question to Willem. Would you agree to combine the data from a trial to the data from DTM to the looks like a trial for individual patient data meta analysis. This is what Christian was alluding to, you know, having one database, having the same selection criteria. Would you think make, you know, help the community to better understand, you know how to interpret the data because there are patients in TT M two that looks like, you know, with the shock, a rhythm, witness arrest times from 5 to 15 minutes, uh arrival of ems and all the selection criteria you made. Will you think it might make sense if we help the community? If, if we could match also the time to start cooling, which is like 100 minutes in, in the hacker trial might be worth to look at it. But I, I'm not sure if we find something because it's too different of study populations. Um I'm not sure if you find something but to look at it doesn't harm. But we should only spend time of it if we agree that it gives benefits. But I remember that we once asked Nicholas Nielsen to get the data for, for a patient meta analysis, but he refused to give the data. This is bad. We should, we should share, we should share. I would say the reason I don't like this me analysis is also also in the Cochrane study, the Haka trial and the beard trial, which is perhaps 1/4 of the TTM trial in numbers still have a higher weight than the TTM trial. This is really strange that you really in your calculations, you give them much higher weight because there's a problem with numbers. When you do, you get rid of that, if you combine them, of course, the real data. So any question from the audience? I'm surprised you have no questions for the two here. OK, please. It seems like you both agree that when cooling is initiated prior to global reperfusion, there's benefit. Do you think that it is if not neuro protective but cardio protective to have started the cooling prior to focal reperfusion in the cath lab, there has been the is still out there. There's a 222 to 3 that have tried that because you can easily cool down to, for instance, 35 degree and then open the artery. And uh there was a pilot trial first where they showed afterwards that if they only took the L ad the big vessel in front of the heart in the calculation afterwards, then it seemed to help. And they did a slightly larger trial where they did not find a significant but still a trend in the same direction. And they are still uh there's still research going on in that area, but it's not been proven yet. But it's definitely something that could be followed. Can we have the microphone on the front line? This is another question, please. Until it comes there. I would like to add something to the trials you mentioned for early cooling during categories and except the princess trials in the other trials, cooling was with cold IV. Yes. And we know from animal studies that if you give cold IV fluid during categories, you decrease coronary perfusion pressure and you will prevent risk by giving too much food during categories. So these, I don't know why these studies were done in patients since before. In animals, we learned that you should not give fluid during categories. So maybe this was the reason why the early cooling didn't work because you didn't even have risk. But in humans, you had equal amount of risk in the two groups with 2500 patients. Now you killed the good ones, Kia. Well, so first of all, thank you for your both presentations and the debate. Uh Well, being a clinician, um I'm asking you what are in your opinion, uh the biggest obstacles to create uh a big randomized trial on normal patient. I mean, what happens in the daily base life? Because all the data of the most recent trial absolutely do not uh um uh are mm are not transparent to what happened, not similar to what are our usual patients? I mean, uh uh the percentage of bystander CPR the percentage of a witness, cardiac arrest, uh the time to risk the survival rate. This is not similar to what happens. For instance, in Italy, I mean, uh our registry do not uh um give us this kind of uh statistics. So what are the limits to try to perform a big trial where the people who are randomized are the combine the, the usual patient, the normal patients that uh happens to Ems and to um emergency department laziness. I, I think it's very important when you look at these that you have a lot of a s to leave. You had a lot of unwitnessed and so forth. You have to, that's out in the scene at the hospital, the people who actually survived to the hospitals there, you have a higher rate of uh people who have had uh CPR and people who have been witnessed and so forth because the other one never made it to the hospital. So, so, so you Yeah. So, so you have to be aware of that. But apart from that, I don't see why we couldn't make a trial with everyone that is alive in the ICU. Who do you think? Yes. Well, since we are just putting a trial together, um, I can tell you we have to be very careful with the new study not to throw out a child with a bath. If we have another neutral trial, I think hypothermia is uh dead. Despite most of us believe that there are subpopulations who benefit from hypothermia. So now to prove this, we have to include only patients, we believe they are in the subpopulation. Yes. But you don't, you believe that uh if you continue to work, I mean, I, I can part your opinion and I, I do, but I don't you believe that if we continue to try to find what we are looking for, we are in a certain sense of modifying the results in order to uh get uh what uh what is our gral? OK. And on the other end, if we get out of our laziness and we try to uh have all these uh clinicians uh having on the paper, the same patient that they get uh uh on at the emergency department. So I'm getting out to those who do not get to the hospital. So probably if the, the cohort of patients is sufficiently big, the Multivariate analysis might give us the subpopulation we are looking for. I, I think the most important thing is to, to give good hypothermia and good hypothermia for me means to start as soon as feasible. Like in our study, we are planning now, the inclusion criteria is ros to start cooling within one hour. So I don't know how many centers which are not involved in research can start cooling within one hour after I think we first we have to go back two steps. We have to prove that hypothermia works in. We think it's a good hypothermia. It's a well controlled hypothermia for 48 hours starting within one hour. Once we have proven this, hopefully again, then we can start with, with a more practical trial after that. But I think we have to go due to TTM one and two. We have to go two steps back, I think and prove again, hypothermia works if it's done well in the correct population. That's my before closing the debate just for a matter of time. Sorry, because I was sure it was going to take hours and hours, but we are already a little bit delayed. Can we conclude the both of you? But in all cases, whatever his position, we are still in the idea to actively manipulate temperature for normothermia or hypothermia. We are still doing something active. You don't abandon any temperature control. Do we agree on that? A t the moment we still do it, still do it. So we still a rein the area where reactive. I think it's a good message because people have considered TM two as being the end, as William said, the end of any temperature control, which is not the idea. Of course, we still prevent fever, still preventing fever. Ok. So honestly, thanks for this. I'm sure that some people have a lot of questions for you after the break. Created by
