Effect of Vasopressin and Methylprednisolone vs Placebo on Return of Spontaneous Circulation in Patients With In-Hospital Cardiac Arrest

This transcript is auto generated and unedited.

>> Lars Wiuff Andersen: Good evening, everyone. So, my name is Lars Andersen. I am a physician at Aarhus University Hospital here in Denmark and also a researcher at Aarhus University. It's quite our pleasure to present the Vasopressin and Methylprednisolone for In-Hospital Cardiac Arrest trial, also known as the VAM-IHCA or VAM In-Hospital Cardiac Arrest. Before we start with the actual presentation, I just want to say thanks to all the people who have been involved in these trials. These are the authors on the paper. I do want to mention one person, and that is Birthe Sindberg, who has been the project nurse on this project for the last 3 years. She has put in an extreme effort to make sure that this trial has run smoothly. So, thanks to her. I do also want to mention the funders before we get started. The trial is funded by three independent parties in Denmark. They're all public foundations. So, we have Aarhus University Research Foundation. We have the Department of Clinical Medicine at Aarhus University. And lastly, the Independent Research Fund in Denmark. The trial drug, specifically vasopressin, was given by Amomed, who makes the drug free of charge. None of the funders or the company had any say in the conduct of the trial, the manuscript, or the publication of the results. I want to now give the word to Asger, who will be presenting the background and the methods.

>> Asger Granfeldt: Thank you for the word, Lars. My name is Asger Granfeldt. I am also a physician at the Department of Intensive Care at Aarhus University Hospital. I'll have the honor to present the background and the method for this presentation for the VAM-IHCA trial. So, I am going to start by highlighting why it is important to study in-hospital cardiac arrest. In the United States, there is an annual incidence of 290,000 in-hospital cardiac arrests. And in Denmark, this annual incidence is around 2000. The return of spontaneous circulation rate is around 55% to 60% across registries, and survival rate is only 25% to 30%, illustrating the need to improve survival. In a recent paper published by Mathias Holmberg and with me and Lars, we demonstrated -- the y-axis is survival to hospital discharge, and the x-axis is year. We showed that there was an increase in-hospital cardiac-arrest survival until 2010. But since then, survival has plateaued, illustrating the need to improve survival following in-hospital cardiac arrest. And these data were produced using [inaudible] the American data on in-hospital cardiac arrest. So, when we talk about cardiac arrest, we most often talk about out-of-hospital cardiac arrest, but there are some differences and some similarities between in-hospital cardiac arrests and out-of-hospital cardiac arrests we need to highlight. The incidence is approximately the same. There's 290,000, as mentioned, in-hospital cardiac arrests each year and 350,000 cardiac arrests outside the hospital in the United States each year. The patient characteristics are also fairly similar, mean age 66 and mean age 65, and approximately 60% are men. Furthermore, in both populations, the patients often presented with a non-shockable rhythm in 80% of the cases. Finally, the most important differences we believe is the timing of basic life support. When you have an in-hospital cardiac arrest, you have a trained staff immediately delivering basic life support of a high quality. In contrast, this is varied depending on bystander involvement in the out-of-hospital cardiac-arrest setting whether it's witnessed, non-witnessed, private location or public location. Furthermore, especially in relation to this study, it's also important to notice timing of advanced life support drugs. This is normally reported to be within 5 to 10 minutes in in-hospital cardiac-arrest studies. And on average, this is approximately 20 minutes after onset of cardiac arrest in the out-of-hospital cardiac-arrest setting. And these differences might illustrate the difference in survival to hospital discharge being approximately 25% with in-hospital cardiac arrest and only 10% to 12% with out-of-hospital cardiac arrest. So, still having in mind the differences and the similarities between out-of-hospital cardiac arrest and in-hospital cardiac arrest, we know there is a lack of evidence for in-hospital cardiac-arrest interventions. We know that epinephrine and amiodarone, the two commonly used drugs, are only tested primarily in the out-of-hospital cardiac-arrest patient population and then transfer to patients with an in-hospital cardiac arrest. This is also illustrated by a review by Sinha, et al., from 2016 where they did a systematic review on all adult cardiac-arrest treatments, and this review demonstrated only 4% of all studies so far only involved in-hospital cardiac-arrest patients. We also did a systematic review of post-cardiac-arrest interventions, and we demonstrated that currently there's only three ongoing trials testing post-cardiac-arrest interventions in patients with an in-hospital cardiac arrest. So, this illustrates the need for new interventions to improve survival following in-hospital cardiac arrest. So, now, we're going to talk a little bit about vasopressin, which is one of the drugs used in the VAM-IHCA trial. We all know vasopressin as an anti-diuretic hormone. We know it's a very potent vasoconstrictor. But based on these vasoconstrictive properties, it's recommended for patients with septic shock as a second-in-line vasopressor agent. However, in patients with cardiac arrest, we know that our levels are lower, the vasopressin levels are lower in non-survivors compared to survivors. So, the question is, "Does vasopressin have an effect in out-of-hospital cardiac arrest?" And there was a study by Mathias Holmberg, et al., looking at the effect of vasopressin in patients with out-of-hospital cardiac arrest. In the first set of studies, vasopressin was compared to epinephrine, and there might be a small effect of vasopressin, but no significant difference in survival between patients. In the second group of patients, epinephrine was combined with vasopressin and then compared to epinephrine, and this meta-analysis favored epinephrine versus the combination. However, again, this is patients with out-of-hospital cardiac arrest with a very long time to drug administration. And again, with in-hospital cardiac arrest, there's only been one small study including 200 patients. The second drug used in this study was methylprednisolone, steroids or corticosteroids, and we all know these drugs have multiple effects on metabolism, inflammation, being a vasoactive agent. And we know they have genomic effect that works hours to days, but also some non-genomic effects that sets into action within minutes to hours. As for vasopressin, we know that there are lower levels of cortisol in non-survivors compared to survivors. And again, there's been a number of limited studies in patients with in-hospital cardiac arrest, but actually also cardiac arrest in general. There was a systematic review published in 2021 in June that only identified two studies looking at this, and one of them was from the '80s and the second study was in 2016. So, why combine vasopressin and steroids? We know that they both have vasopressor function, and steroids are very potent anti-inflammatory drugs, and we know that inflammation happens during cardiac arrest and post cardiac arrest. This interaction between vasopressin and corticosteroid treatment has also been acknowledged in the field of septic shock by these two publications. So, now, we're going to talk briefly about the two landmark publications by Mentzelopoulos published in 2019 and 2013. The first study from 2009 was a single-center, randomized, double-blind study, including 100 patients with in-hospital cardiac arrest with receiving at least one dose of adrenaline. The second study from 2013, multi-center, three-center study, randomized, double-blind again, including 268 patients with in-hospital cardiac arrest receiving at least one dose of adrenaline. So, what was the intervention in these two landmark studies? So, first, both studies included vasopressin 20 International Units plus methylprednisolone after the first adrenaline dose as soon as possible. Then, after each adrenaline dose, vasopressin was repeated 20 International Units again up to a maximum of 100 International Units. If the patient obtained ROSC and was in shock four hours after the cardiac arrest, hydrocortisone was administered to the patient. So, how did those -- what was the results of those two studies? So, very impressively, in study one, the first study from 2009, the ROSC grade increased from 52% to 81%. In the second study, the ROSC grade increased from 66% to 84%, so very consistent between the two trials. Most importantly and also most impressively, survival increased from 4% to 19% in the first study. And in the second study, it was survival with a favorable neurological outcome and increased from 5% to 14%. However, based on these two studies, the European Resuscitation Council stated that, "these studies are not generalizable to all cardiac arrests, and we suggest that steroids are not routinely used for cardiac arrest." The American Heart Association Guidelines stated that the combination of each drug "may be considered. However, further studies are needed before recommending the routine use of this therapeutic strategy." And these further studies are the VAM-IHCA trial. So, this trial was investigator-initiated by Lars Andersen. It's a multi-center, randomized, placebo-controlled, double-blind trial of vasopressin and methylprednisolone during adult in-hospital cardiac arrest, and the protocol was published in Resuscitation Plus. The study was conducted in Denmark and included 10 sites. Four of these sites were large university hospitals, and the last six were middle-sized hospitals. The inclusion criteria were fairly straightforward, in-hospital cardiac arrest, age above 18 years, and should have received at least one dose of adrenaline during CPR. Also, the exclusion criteria were straightforward, clearly documented do-not-resuscitate order prior to the cardiac arrest, prior enrollment in the trial, extracorporeal circulation at the time of cardiac arrest, and known or suspected pregnancy. So, this slide, I'll go through how the patients had the drugs administered. So, we had an in-hospital cardiac arrest, and bystander CPR was started by the person recognizing the cardiac arrest. We then had activation of the cardiac-arrest team. And with the cardiac-arrest team came this white box, the VAM-IHCA box. This box contained 20 International Units of vasopressin in four doses and 40 milligrams of methylprednisolone or placebo. As soon as the patients were deemed fit for inclusion in the trial based on the inclusion and exclusion criteria, the box was opened. In the box, there was instructions for how the medicine should be administered to the patient so any member of the clinical team could administer the drugs, whoever had the time during the cardiac arrest. So, as soon as adrenaline was administered, 20 International Units of vasopressin and 40 milligrams of methylprednisolone or placebo was administered. And then again, for each additional dose of adrenaline, 20 International Units of vasopressin or placebo was administered. And this was up to a maximum dose of 80 International Units or four doses. So, the primary outcome of this trial was return of spontaneous circulation without the need for chest compressions for 20 minutes. Key secondary outcomes included survival at 30 days and favorable neurological outcome at 30 days. Additional outcomes included quality of life, adverse events, and favorable neurological outcome and survival at 90 days, six months, and one year. We're still collecting data for the outcomes of six months and one year. So, the sample size calculation for the study was like this. So, return of spontaneous circulation, we estimated that 45% would achieve this in the placebo group. This was based on 2016 data from the participating sites where the study was initiated and also on data from the National Danish Cardiac Arrest Registry called DANARREST. The 58% return of spontaneous circulation intervention group yielding a risk rate of 1.29 and a risk difference of 13% was based on the effect size from the two original Greek studies. Using a power of 80% and 5% alpha, we should include 492 patients. So, the data was analyzed using a modified intention-to-treat analysis. This means that patients were included or analyzed if they received the first dose of the drug, if they fulfilled the inclusion criteria and no exclusion criteria, and they were analyzed according to their assigned intervention. The binary outcomes we presented were risk ratios and risk differences, and we performed a number of subgroup analysis, rhythm, witnessed status, age, time from cardiac arrest to trial drug administration, and time from epinephrine administration to administration of the trial drug. All analyses were pre-specified in the protocol. Now, I am happy to give the word to Lars to present the results and the discussion.

>> Lars Wiuff Andersen: So, I will continue with the results. Here, we have the CONSORT diagram. The trial started in October 2018, and we enrolled the last patient on January 21, 2021. As you can see in this time period in the participating sites, there was approximately 2360 in-hospital cardiac arrests. We did exclude a number of patients either because they did not meet inclusion criteria, a large number of patients never received epinephrine. This could be because they had termination of resuscitation very early or because they had a shockable rhythm and obtained return of spontaneous circulation before epinephrine was needed. We also, unfortunately, excluded a number of patients for other unrelated reasons. The most important ones were the patient received return of spontaneous circulation prior to the trial drug. So, this was after they received epinephrine, but before the clinical team was able to provide the trial drug. It did take a little time to mix the drugs and get them administered. We also had a number of patients that in that time period had early termination of resuscitation. Lastly, we had a group of patients where the clinical team forgot in this hectic situation of a cardiac arrest or there was some clinical decision that there were other priorities during the cardiac arrest such as airway handling or rapid transport to the OR. There was also some logistical reasons including that the trial drug was not available and for a few patients that they were isolated due to COVID-19. We ended up randomizing 512 patients who all received the first dose of the drug. As Asger mentioned, we only analyzed those patients who did not meet inclusion -- all inclusion criteria or had some exclusion criteria. So, we ended up with 237 patients in the intervention group and 264 in the placebo group. On the right side here, you see the Table 1 describing the baseline characteristics of the included patients according to the treatment assignment. I'll just highlight a few things. So, we included a total of 501 patients. The mean age was 71 years old, and about two-thirds of the patients were male. This is quite similar to a general in-hospital cardiac-arrest population. About two-thirds of the patients had their in-hospital cardiac arrest on a ward or medical or surgical floor, and most of the patient had an initial non-shockable rhythm with 90% having asystole or pulseless electrical activity. The time from the recognition of the cardiac arrest to the administration of the first dose of epinephrine was a median of 5 minutes. The time from the recognition of the cardiac arrest to delivery of the first trial drug, which was vasopressin, was 8 minutes. We had no loss to follow-up and have complete outcome data for all patients. I am now going to move on to the primary results. Here, our return of spontaneous circulation illustrated. As you can see, in the vasopressin and methylprednisolone group, 42% of the patients had return of spontaneous circulation. In the placebo group, this number was 33%. This yielded a risk difference of almost 10% and a risk ratio of 1.30. As can be seen from the confidence intervals and from the p-value, this was statistically significant with a p-value of 0.03. For the secondary outcomes, we looked at survival and favorable neurological outcome at 30 days. Unfortunately, we found no difference in these outcomes for the patients. In the vasopressin and methylprednisolone group, approximately 10% of the patients survived to 30 days, and this number was 12% in the placebo group, yielding a risk difference of minus 2% and a risk ratio of 0.83. These findings were not significant with a p-value of 0.48. Results were quite similar for favorable neurological outcome. This was assessed by the Cerebral Performance Category score of 1 or 2. Actually, the [inaudible] was similar in the two groups, 7.6, yielding a risk difference of 0 and a risk ratio of 1. And again, this was obviously not statistically significant. We also looked at the same outcomes after 90 days, and these results were very similar to those at 30 days with no clear difference in survival and no clear difference in a favorable neurological outcome. At both 90 days and 30 days, we looked at quality of life. We use the EQ-5D-5L. This is a score ranging from 0 to 100 with 100 indicating a perfect quality of life. As you can again see here, there was no difference between the groups with the placebo group in the blue and the intervention group in red. There was some improvement over time in both groups. In this figure, it's a Kaplan-Meier curve. We illustrate survival over time. As you can see, most patients die early. Those are the patients not receiving return of spontaneous circulation. But otherwise, the curves are very similar out to 90 days again with no difference between groups. We did look at a number of potential adverse events. These patients have multiple things happening to them, multiple potential adverse events. So, we had pre-specified a number of adverse events that we wanted to focus on and collect detailed data on. These were tailored to the intervention, so things that we might expect could happen if you received either steroids or vasopressin. For those adverse events that might be related to steroid use like hyperglycemia, hypernatremia, and infections, there was no clear difference between groups, although all of those were quite common in this patient population. Those adverse events that could be related to vasopressin, such as mesenteric ischemia or peripheral ischemia, on the other hand, was quite rare, and there was no clear difference between the two groups. As Asger mentioned, we had pre-specified a number of subgroup analyses. We focused here on return of spontaneous circulation, although the results are quite similar for 30-day survival and 30-day favorable neurological outcome. On the left side, you can see the overall effect and then the subgroups, as Asger mentioned. Then, we have the raw numbers. And then, we have the risk ratio and the risk difference. The only subgroup that perhaps had an interesting finding was time from arrest to trial drug where it looked like that those who received the trial drug earlier perhaps had a better effect. But again, the confidence intervals were overlapping, and there was no statistical significant difference in these numbers. Before I want to go on to the discussion, I want to share some additional data with you. This is not part of the manuscript, but it's a separate manuscript that will be coming out or get submitted hopefully within the next few weeks. This was led by Mathias Holmberg who's a colleague of mine and Asger's. So, this was a systematic review and individual participant meta-analysis. We did a systematic search of PubMed, Embase, and the Cochrane Register. We were looking for trials that assessed the combination of vasopressin and steroids during cardiac arrest. The plan was then to conduct an individual participant data meta-analysis using both traditional frequentist approaches, but also Bayesian analysis. The analyses were adjusted for the trial, which for the primary analysis included as a fixed effect, also age, location, time of the day, whether or not the cardiac arrest was witnessed, and the initial rhythm. Many of these factors were included because we know they are strongly associated with many of the outcomes we were interested in. We identified two trials, the two Greek trials by Spyros Mentzelopoulos, et al., and then, of course, our own trial. All of these trials only included patients with in-hospital cardiac arrest and had similar interventions. There were some differences that I'll get back to later. And we want to thank Spyros Mentzelopoulos, of course, for sharing his data so we were able to conduct this analysis with 869 patients. Here, we illustrate the results. So, first, we have the three outcomes, return of spontaneous circulation, survival at hospital discharge, and also CPC 1 or 2 at hospital discharge. Notice that we here use hospital discharge because those are the outcomes the previous trials reported. Then, we have the individual trials and then the combined individual participant data meta-analysis. And there are two important things here. First, there was some heterogeneity in the odds ratio from the different trials as illustrated by a diminishing effect from the earlier to the later trials. The other important thing is that there was a clear and strong effect on return of spontaneous circulation with an odds ratio of approximately 2. However, for survival and for a favorable neurological outcome, the results were more unclear. The results certainly suggested a benefit, but the confidence intervals were wider and did not exclude 1. Given that we now had additional power with the additional patients, we also again conducted some pre-specified subgroup analysis. The results are again similar with no clear difference according to subgroups, but perhaps a suggestion that those who received the trial drug earlier had better outcomes. As I mentioned, we also conducted a number of Bayesian analysis. So, in Bayesian analysis, we specify a prior belief, then add on the data we have from the trials, and then get a posterior belief. We conducted this using a number of pre-specified priors to sort of emulate different beliefs, so some that were pessimistic, skeptical, or optimistic, and also some that were weak or strong. For return of spontaneous circulation, the results were very similar given the strong effect on return of spontaneous circulation, so I'll instead focus on 30-day survival and just mention a few of the analyses we did for this manuscript. If we started out with a strong optimistic prior, so we believed that there was an effect already, with an odds ratio of 1.2 and a fairly narrow confidence interval or credible interval in this case, then, of course, the results are in line with this prior and we get a posterior that is very similar. The nice thing about Bayesian analysis is that we can also calculate or estimate different probabilities. So, below is the probability that the odds ratio is above 1, so what we would consider an effect, and that was 99% given that you had a strong optimistic prior. However, if we instead used a strong skeptical prior, and we use skeptical in the sense here meaning that there was no effect. So, an odds ratio of 0.1 again with a fairly narrow credible interval, 95 credible interval. Then, we see that the posterior estimate is slightly different. And this time, we get a result that is more unclear, and the probability that the odds ratio is above 1 is closer to 84%. If we, on the other hand, were quite pessimistic and thought that this treatment would harm patients -- so, this means that the prior odds ratio was 0.83 -- well, then, the posterior is, of course, affected by this prior. And here, we can see that the probability that there is an effect -- so, the odds ratio is above 1 -- is only 22%. I think this clearly illustrates that the way we interpret the results does depend on the way we look at it from a prior belief, and I'll get back to that later. So, going to the discussion, I want to briefly summarize our results from the VAM-IHCA trial. We found an improvement in return of spontaneous circulation, 42% in the intervention group and 33% in the placebo group. This was a clinically relevant difference, almost 10%, and also statistically significant. However, we did not find a difference in survival, 10% versus 12%, and no difference in a favorable neurological outcome where both groups had approximately 8%. So, of course, there's some important differences between these results and the previous Greek studies. So, I want to go back to this figure just briefly where we can illustrate that both -- there are two important differences. First, there's a difference in the effect on the odds ratio scale where we can see that the effect gets smaller from the 2009 trial to the 2013 trial and to our current 2021 trial. And this is true for all the outcomes. However, I also want to point your attention towards the proportion with the outcome in the placebo group. So, for example, if we look at our trial, a third of the patients had return of spontaneous circulation in the placebo group, and this was almost twice as high in the Greek studies. However, when we look at survival to discharge, we had 10% to 12% survival and 7% favorable neurological outcome in the placebo group, and this was much higher than the previous Greek trials. So, this clearly illustrates that there is something different going on in these different trials. I want to try to briefly discuss some of the similarities and differences between the trials. First, some of the similarities. In both trials, we use a vasopressin dose of 20 International Units. The maximum number of doses was five in the Greek trials and four in our trial for some logistical reasons. The methylprednisolone dose was 40 milligrams in both trials. A very important difference was that the Greek trials provided post-cardiac-arrest steroids in the intervention group and 300 milligrams of hydrocortisone if the patients, four hours after return of spontaneous circulation, had cardiovascular shock. We did not provide this in our trial. Another important difference is the time to trial drug administration. This was very fast in the early trials, 3 minutes, 5 minutes in the 2013 trial, and 8 minutes, as we mentioned previously, in our trial. I think there could be multiple reasons for these differences. Importantly, the first trial was a single-center trial. Perhaps the environment was a bit more controlled. Then, there was three centers. And then, there was 10 centers. So, in some way, our trial is perhaps more pragmatic. That's also illustrated by the fact that the two earlier trials had dedicated research personnel who ran to the cardiac arrest and provided the trial drug as soon as possible. In our trial, that was not possible because we had 10 sites. At least we didn't have funding for that. So, we had the clinical cardiac-arrest team provide the intervention, as Asger mentioned briefly. This, of course, means that the clinical team also have other responsibilities and they're there for some delay in the administration of the trial drug. While that might be a limitation, it also reflects what we think is real life clinical practice. When we look at the patient characteristics of the included trials, there are also some important differences. As has been mentioned before, the sample size increased and the number of centers increased with the trials. The patient age and the gender composition were quite similar with an age of approximately seven years and about two-thirds of the patients being male. However, there was quite a lot of differences in the location of the cardiac arrest. So, in the Greek trials, approximately one-third of the patients had their cardiac arrest in an intensive care unit, and that number was only 8% in our trial. Corresponding, we had many more patients having a cardiac arrest in a normal medical or surgical floor. Another important difference is the number of patients with asystole as the presenting rhythm. That was about two-thirds in the Greek trials, but only one-third in our trial. And of course, corresponding to that, we had more patients with pulseless electrical activity. The number of patients that were witnessed were a bit higher in the Greek trials, but otherwise fairly similar, and the number of patients who were monitored were quite similar between the trials. So, there are some important strengths of our trial. Considering the topic or the patient population of in-hospital cardiac arrest, it's a relatively large trial and was also multi-centered. We do think, despite the limitations I just mentioned, that the time to drug delivery of 8 minutes is a strength, especially when we compare it to out-of-hospital cardiac-arrest trials where, as Asger mentioned, the time to drug delivery might be 20 minutes or more. This might illustrate that the in-hospital cardiac-arrest population is a good population to test advanced interventions such as trial drugs or drugs during cardiac arrest. We also have long-term outcomes with some of it is still ongoing with no loss to follow-up. There are also some important limitations. As I mentioned, there were many potentially eligible patients who did meet the inclusion criteria and did not have formal exclusion criteria that were not included for various reasons. We do not believe that this has a strong influence on the internal validity of the results, but it could affect generalizability. And although I list the time of drug delivery as a strength, it is also a limitation, especially when we compare it to the previous trials where it was much, much shorter. Some might look at the survival rate, perhaps both in the Greek trials, but also in our trial, as low. Asger mentioned in the beginning that the survival was 25% to 30%, but it's very important here to look at the patient population. These were also patients who required at least one dose of adrenaline. And there are many patients, therefore, that are not included. The patient who has a ventricular fibrillation and gets one or two shocks, not included. The patient who maybe has PEA and gets one dose of adrenaline and very quickly has return of spontaneous circulation is not included either. So, of course, when we consider that patient population, the outcome will be worse. This is very similar to the PARAMEDIC trial where they had a very low survival rate because the population was defined by receiving adrenaline. Lastly and of course perhaps the most important limitation is that this was, of course, not powered for 30-day or 90-day outcomes. So, as you can see, the results were inconclusive for some of these outcomes, and the confidence intervals were wide, including both potential benefit and potential harm. So, what is the conclusion from this? Should we recommend for or against vasopressin or steroids? I think it depends on a lot of things. It depends on how much you value ROSC as a primary outcome. It also depends on your prior beliefs, as we've shown from the Bayesian analysis. I think my own and Asger's opinion is that with the current data, it's still not strong enough to go out and recommend this treatment for all patients with in-hospital cardiac arrests. This is a complex intervention that has to be added on to an already complex set of guidelines that includes multiple drugs. However, we do believe that the results both from our trial, but also from the individual participant data meta-analysis, do call for larger and bigger trials to be conducted. However, as we know, if we want to look at small differences in outcomes, we need a lot of patients. And I just added two examples here to illustrate that concept. Our results from the meta-analysis could indicate perhaps that there could be a difference of 8% to 12% in favorable neurological outcome or survival. This would require a little bit over 2000 patients. However, the effect could be even smaller, 8% perhaps in the placebo group and 10% in the intervention group. Although this seems small, I think it would still be a clinically relevant difference and it would result in thousands of lives saved each year globally. However, to detect such a small difference, we would need more than 8000 patients. That's at least not possible in Denmark, but could be possible with international collaboration. I want to once again thank all the authors and co-investigators on this study and this manuscript. There's a lot of people who put in a lot of effort at the different sites. I also want to thank the clinical personnel at all the individual sites because, as we mentioned, it was the frontline clinician, there were hundreds of them across these sites, who enrolled the patients and provided the trial drug. I also want to say a big thank you to Rob and Critical Care Reviews for allowing us to present these results tonight. And lastly, I am happy to share that the results are now published in JAMA and the manuscript is here. Thank you.

>> Rob Mac Sweeney: Lars, thank you so much. Congratulations to yourself, Asger and the rest of the trial team. That is a fantastic achievement. And I'll echo your final comment. Thanks very much to JAMA, to Chris Seymour, and to Stacy Christiansen for synchronizing the publication of the paper with tonight's livestream. So, for a huge amount of fascinating data, who better to turn to now than Spyros Mentzelopoulos, who was the chief investigator for the original two trials that Asger has referred to and Lars. So, we will now cross to Athens and to Spyros for your editorial, please.

>> Spyros Mentzelopoulos: So, thank you very much, and thank you very much for this honorary indication. We are colleagues. I am Spyros Mentzelopoulos. I am a clinician and researcher at the University of Athens and the First Department of Intensive Care Medicine at Evaggelismos Hospital. This is a relatively large hospital in the center of Athens. And these are my disclosures. I have been the lead investigator of the VSE I and II trails that were published in 2009 and 2013 by the JAMA Network Journal. So, yes, I would also like to congratulate Lars and his associates. It was an excellent trial, very well conducted, and it was a pragmatic trial. Now, let's go to the rationale about the drug combination of vasopressin, epinephrine, and steroids. What we want in CPR is an immediate and maximum vasopressor effect. Why? Because it may be beneficial. It may be beneficial because -- especially if we have a simultaneous adrenergic and vasopressin V1a receptor stimulation, we can maximize pharmacologic vasoconstriction, increase diastolic pressure and coronary perfusion pressure, and expedite ROSC. But if we expedite ROSC, we minimize low-flow time. And thus, we may also improve patient outcomes. Now, if we co-administer steroids we may potentiate the pressor effects of the vasopressors, we may prevent early post-resuscitation hypotension because of the longer action of vasopressin and the methylprednisolone. I just wanted to remind that the half-life, the elimination half-life of vasopressin is about 24 minutes. And we may also achieve an attenuation of the systemic inflammatory response, which has been associated with worse outcomes in cardiac arrest. Now, if we continue with the post-resuscitation stress-dose hydrocortisone, we may still be able to improve patient hemodynamics, and we may also maintain a possible attenuation of the systemic inflammatory response. On the last, I previously presented the major differences between our studies. So, I will just return to this. The first major difference is that in our study, we administered post-resuscitation stress-dose hydrocortisone 300 milligrams per day. This is a summary of our protocol. So, during the first CPR cycle post randomization, we administered actually the VAM combination, the same combination as Lars and his associates administered. And then, we went on with a simultaneous administration of epinephrine and 20 units of vasopressin in the intervention group for the subsequent four CPR cycles. And then, if we did not have ROSC, we continued ALS according to the guidelines 2005. And for patients who achieved resuscitation, we evaluated them at four hours following ROSC. And if they fulfilled the criteria, the pre-specified criteria for post-resuscitation, then they were treated with stress-dose hydrocortisone. Now, about post-resuscitation steroids, there are not much data, and this is the main difference between our trials and the VAM-IHCA trial with respect to post-resuscitation interventions. So, one randomized, controlled trial with 50 patients conducted by Donnino, et al. This was a three-center trial. Administered post-resuscitation stress-dose hydrocortisone to 25 patients and placebo to 25 control patients with either out-of-hospital or in-hospital cardiac arrest. It was a mixed population. And the stress-dose hydrocortisone was initiated at about 10 hours post-resuscitation. And the results were impressively neutral, apart from a very, very small subgroup of possible responders to steroids. But overall, in this very small, of course, patient population, the results were impressively neutral. We have completed actually about 2 and 1/2 years ago a trial of -- a trial aiming to evaluate the physiological effects of intra-arrest and post-arrest steroids in my hospital. This trial was conducted actually during a period when vasopressin was not available in Greece. Otherwise, we would have had some problems in getting this protocol approved by our IRB. And we started about -- we started 100 patients, and our intervention was 40 milligrams of methylprednisolone during CPR, and this was followed by stress-dose hydrocortisone for some patients. There were actually 46 patients in the intervention group and 54 patients in this other group. And our results were consistent, very consistent, with those of Donnino, et al. So, there was no benefit of post-resuscitation steroids alone in two small trials. So, there is a question whether there is actually post-ROSC resistance to stress-dose steroids, and this may be partly explained by an ischemia and reperfusion-related oxidation of the glucocorticoid receptor that leads to its degradation by the ubiquitin-proteasome system, and there is also data suggesting that the protein oxidation due to the ischemia and reperfusion increases with ischemia duration. Now, the median low-flow time, that is -- actually ALS time, ALS duration in the study of post-resuscitation steroids was 18 minutes, but in the VSE studies it was just 12 minutes. So, this has led to a possibly explanatory hypothesis that the longer ALS duration in the CORTICA study may have predisposed to resistance to steroids, but this is only hypothetical. Now, we do know about the significance of preventing hypotension early post-resuscitation. Early post-resuscitation hypotension is associated with increased mortality. There are multiple data of this. It's not only this study I am presenting right here. And also, any post-resuscitation hypotension is associated with poorer neurological outcome. And well, either resistance to steroids or delays in post-ICU admission -- post-ROSC ICU admission and initiation of invasive monitoring may predispose to the occurrence of hypotensive episodes. So, my first question is whether they have recorded -- they have data on ICU admission type. Now, regarding additional differences between our studies, we did have limited sample sizes in both our studies. So, when our results -- even by the putting of our results, we have some randomization-related imbalances that may have slightly -- hopefully, slightly -- favored the VSE group. So, you see here that, for example, myocardial ischemia was less frequent in our combined VSE control groups. And that high potential was more frequent in the intervention groups. Also, a respiratory etiology of cardiac arrest was more frequent in the control group, and this is known to have a poorer prognosis. So, there were some randomization-related imbalances that, of course, could not be controlled. Now, an additional important difference between our studies and the VAM-IHCA study was the time to drug administration, to vasopressor administration. We do know from the literature that the time to epi is associated with differences in survival, and the earlier epi is administered, the better it is for the long-term patient outcomes. Now, this study, which was published in Circulation, illustrates this difference. And this Q1 corresponds to a very low frequency of delays in epinephrine administration, and the functional outcome is, of course, better than when the frequency of delays in epinephrine administration is higher. The other study that supports this is a study published by Donnino, et al., which shows that as epinephrine administration is delayed, we have a lower probability of achieving return of spontaneous circulation and, of course, a lower survival to hospital discharge. Now, regarding the time to drug administration, we did have some differences between our trials and the VAM-IHCA trial. The time to epi was remarkably similar, and I have to acknowledge that, despite the fact that the vasopressors were not administered by a research team in Denmark, they did a remarkable job in having very, very fast times to vasopressor administration. We may have a little lower times in our studies -- these are pooled data -- because about one-third of our population was actually in the intensive care unit. And in the intensive care unit, when a cardiac arrest occurs, the time to drug administration is very, very short, especially if it's an unshockable rhythm. And if we look at the asystole/PEA patients, then our times to drug administration are even a little lower than the whole population because the time to drug administration increased a little bit by adding the patients with VF/VT. You, of course, do the application of the ALS algorithm. So, another important difference is that we use pre-loaded syringes, so we were able to actually co-administer the study drugs with epinephrine. And the Donnino study did not use pre-loaded syringes, so they did have some slight delay in the administration of the study drugs. And this just might have some impact on the outcomes. Now, regarding the previously-described rationale for the administration of the drugs during CPR, I was just wondering whether the actual targets had been achieved. In our study, post-hoc analysis of the VSE I study showed that a rapid and successful resuscitation was more frequent in the static group than in the control group. Its frequency was approximately double. And also, in the VSE II study, the frequency of rapid and successful resuscitation was again significantly more frequent, 16% more frequent, than in the control group. So, I would like to ask whether fast and successful resuscitation differed between any VAM/control subgroups, especially when t-drug was less than 8 minutes. And just an opinion from Lars, whether the achievement of a hyperacute pressor effects is easier with pre-loaded syringes versus vials. Now, regarding the time lag between dose one of epi and dose one of study drugs, there was a 3 to 4-minute delay. There was a 3-minute separation between epi dose one and study drugs. However, the EPI pressor effect may start waning within 3 minutes or at least 3 minutes and more. So, they administered vasopressin at that time point. And the maximum pressor effect may have not actually been achieved. Of course, I believe that there were certainly a very substantial pressor effects. So, were there a possible consequence of this time lag? Was there a rush delay with possible impact on long-term outcomes as low-flow time may have not been effectively reduced in VAM group compared to control. Now, let's see. Let's have a look at the data consistent with the time-lag interpretation. When there is no real time lag or when there is a very small time lag from epinephrine to trial drug, we seem to have a better effect estimate in the long-term outcomes, though not significant, of course. And the same seems true when the time from arrest to trial drug is less than 8 minutes, less or equal than 8 minutes. And now, there is a striking difference. But of course, this is not a randomized comparison. But anyway, I couldn't help observing it. There's a striking difference between the VAM subgroups. So, you see here when the time lags are significant, then the good outcome percentages, the frequency of a good outcome, is very different. And I would think that the characteristics -- I believe that the characteristics of our studies actually fit better to all these three subgroups with more favorable outcomes in the VAM-IHCA study, and that is a younger age and shorter time to trial drug and, of course, no time lag between epinephrine and trial drug administration. Now, if one performs just a simple exploratory, post-hoc, unadjusted within-group comparison, then regarding time lag epinephrine and study drug, you will see that there is a difference between the subgroups only in the VAM group, but not in the control group. And the same -- and there is also a heterogeneous response of the two groups. Of course, this is a post-hoc, non-randomized, and unadjusted within-group comparison, so it has many limitations. And also, these results are more striking when you have time to study drug of less than 8 minutes versus time to study drug for more than 8 minutes. Now, this comparison should be adjusted for age, preadmission frailty, rhythm, time of arrest, epi dose, use of hypothermia, and perhaps the use of ECMO, which observed that it was more frequent in the control group, the post-resuscitation and ECMO, of course, I think. And there was some report for post-arrest steroids, which were more frequent in the control group. Was there any protocol as to administration of steroids? Were they prescribed by attending physicians in a protocolized manner? Or was it random treatment of, for example, an allergic reaction? I don't know. I would like to ask that question to Lars. And then, I returned to the VAM-IHCA subgroup analysis. I can see that when the time from arrest to trial drug was short, then we have an almost identical result with the result of the VSE II study. So, here, the difference in ROSC is about 16%, and we had a difference of about 18% in successful ROSC reported in the VSE II study. Now, some of the benefit, some of the ROSC benefit, came from unwitnessed arrests. This can partly explain, I think, the loss of the benefit early post-resuscitation in the VAM group. Now, an additional potential problem when we have an extended time to study drug is the fact that we are losing the beneficial effect of vasopressin, which seems to inhibit or desensitize the beta-1 adrenergic receptors and inhibit some of the detrimental effects of epinephrine. This is what was shown in this study. And this, for this to happen, we need to have ATP. So, the longer the time of ischemia and low flow and the longer -- and the larger the ATP depletion, the less the probability of vasopressin-induced beta-1 adrenergic receptor desensitization. This could partly explain the difference between the effect of vasopressin when it's administered later versus being administered earlier. And another difference, of course, was the frequency in the presenting rhythm, the frequency of asystole as presenting rhythm versus pulseless electrical activity as presenting rhythm. I just would like to remind that our major benefit actually in our study was mainly an asystole study, and the major benefit came from the asystole subgroup both in regards to return of spontaneous circulation and long-term outcome. And I would just like to remind an old study, which has specified a frequency of pseudo-PEA of about 40%. Pseudo-PEA corresponds to -- -- very little maintenance of cardiac mechanical activity, which of course may be related to a difference, to a better outcome, and this is illustrated by a higher coronary perfusion pressure. And these patients, of course, ideally, should be -- such patients should be excluded because they could be easier -- whenever the blood pressure is so low that you cannot palpate the pulse, you can manage the patient just with epinephrine or with a very high infusion rate temporarily of norepinephrine, but not with CPR and all these drugs, but it's difficult to exclude such patients. So, in conclusion, I think that the VAM-IHCA has been a very well-conducted pragmatic trial. Our knowledge on the effects of VAM/VSE on ROSC has become more definitive. But unfortunately, the uncertainty about neurological outcome remains, and this strongly prompts for larger trials. Now, another valuable thing about the VAM-IHCA results is that they seem to define that there is a quite narrow temporal window for a meaningful use of vasopressor administration in combination with steroids. And I think that to achieve the best results, during CPR, these drugs should be administered as soon as possible. And lastly, I would like to thank everyone for their attention, and I also would like to ask you to evaluate these statements I made in light of my disclosure. So, I would like to thank you very much, and this is a summary questions for Lars.

>> Rob Mac Sweeney: Spyros, thank you so much. Fantastic editorial. Lots of things to think about. What we will do now is -- Lars, if you're able to close your screen share, please -- what we will do is we will move back to Denmark and Lars and Asger now. We're a little bit behind time, guys, so if you're able to do it succinctly, feel free to answer some of the points that Spyros has raised. Thank you.

>> Lars Wiuff Andersen: Thank you very much, Rob. And of course, first of all, thanks to Spyros for this editorial, and I also want to say thanks to him. When I started planning the trial back in 2017, he was very helpful in providing additional data and help and sort of setting everything up, so thanks for that. I think we want to focus on two things. I want to just briefly touch on the time thing, and then I'll let Asger answer some of the questions related to what happens after cardiac arrest. We couldn't agree more that, of course, we should try to give drugs as fast as possible. That's also what's suggested in the guidelines. And there's also no doubt that our trial is more pragmatic. I think it's important when we look back at the guidelines that that's exactly some of the concerns they had about the Greek trials, that they were potentially not generalizable to other settings. So, that was really what we were trying to test. We did everything we could to try to get drugs that were in pre-filled syringes, but in Denmark that was not possible. The pharmacy wouldn't allow it. There were some issues. And we knew very early on that we could not have clinical personnel at 10 different hospitals 24/7 running out and giving this medication. We also think that would not reflect clinical care. So, yes, it's a more pragmatic trial. And yes, it took longer to give the drugs. And I agree that we should definitely aim for, if there's another big trial, to get it in as soon as possible, but I do think we have to come up with some solutions that will also work in a real clinical setting. And then, I'll let Asger answer the ones -- the questions related to what happens sort of after the cardiac arrest.

>> Asger Granfeldt: So, the first question was about ICU admission time. The simple answer is we don't have any data on time from ROSC to when the patients were admitted. It's our impression that when you have in-hospital cardiac-arrest patients, they are admitted to the ICU immediately. And also, we know that there is an anesthesiologist often connected to the intensive care department as a part of the cardiac-arrest team. I will they make sure that the patient is admitted as soon as possible. They also have the opportunity to put in an arterial line and measure blood pressure on the patient in the ward. But sometimes, admission to ICU can be delayed for other things such as surgery, CT scans, etc., to examine or identify the cause of the arrest. The second question was regarding the administration of corticosteroids following post ROSC. Again, a similar question. The post-ROSC treatments were at the discretion of the treating clinician, so we have no data or we didn't have a protocol for corticosteroid treatment. It's very interesting that despite this, it was used in 24% in the [inaudible] vasopressin, methylprednisolone group and actually 46% in the placebo group, but we don't have any data on those, etc., or indication. However, it's important to notify that when we look at the SOFA score up to also 72 hours, there were no differences between groups. And also, when looking at the cardiovascular SOFA score, there were no difference between groups, although this is a very rough estimate of cardiovascular function with very wide intervals in the different cardiovascular SOFA scores. I think --

[ Inaudible ]

>> Rob Mac Sweeney: Guys, thank you very much. What I'll do now is I will bring in my colleague, Chris Nutt, who has questions from Twitter and from our chat function. Chris, over to you.

>> Chris Nutt: Thanks, Rob. So, yeah, we've got -- I have lots of questions. And first of all, congratulations on a terrific trial. I think, you know, we have enough problems trying to recruit patients in the intensive care unit, never mind recruiting them all around the hospital. So, a lot of people are thinking that this trial is really well conducted. So, congratulations. And I suppose there's a lot of focus on the timing of the drugs, and we've already discussed it a little bit. And previous out-of-hospital trials have compared vasopressin and on adrenaline, as you mentioned before, and some of those trials suggested maybe a better outcome in asystolic patients with vasopressin. So, there's a few questions around why give the epinephrine first? Why not just go in with the vasopressin at the same -- from the beginning or at the same time?

>> Asger Granfeldt: So, first of all, as this is an add-on to standard treatment, we didn't want a study drug to interfere with standard treatment and thereby potentially get -- make sure the patient didn't get poor care just because they were randomized into a study. So, they needed to get standard of care first. And then, as this was clinical personnel, it took time to take the medication from the ampules to syringes and administer it, and we think that would be the same if let's say it was recommended by guidelines, and it may also be the case that it would take a couple of minutes more following administration of epinephrine. So, simply based on we didn't want to give poor treatment by amplifying or augmenting that they should focus on the research protocol instead of the standard of treatment.

>> Chris Nutt: Yeah. And if there's limited evidence for the standard of care and limited evidence for using adrenaline, is that as much of a factor? Do you need the adrenaline at all?

>> Lars Wiuff Andersen: Yeah, well, that's an excellent question. It's a different trial, of course. And maybe someone should do that trial. We felt that the evidence for adrenaline is quite strong in the out-of-hospital cardiac-arrest setting or at least strong enough to it to be recommended in cardiac-arrest guidelines across the globe. We don't have any large trials assessing adrenaline versus placebo. Actually, we have none in the in-hospital cardiac-arrest setting. So, we don't know. But I think as long as it's recommended in guidelines, we were looking for a different research question, as Asger mentioned, an add-on. That doesn't mean that another trial is not valuable. It's just a different trial than the one we conducted.

>> Chris Nutt: Sure. And can you talk a little bit about the dose of vasopressin? That was a similar dose to the previous [inaudible] trials. Other out-of-hospital trials have used larger doses, and there are some resuscitationists who talk about perfusion pressures and whether the dose should be titrated to diastolic or mean arterial pressures.

>> Lars Wiuff Andersen: Yeah, that's an excellent question. The very simple answer is we picked the dose because Spyros picked that dose, and that might sound silly, but it's exactly what happened, and we wanted to try to do a trial that could validate the findings that Spyros had done. Of course, there were some differences. We thought it would be too complex to have a post-cardiac-arrest intervention and maybe also muddy the water a bit, and so we focused on the intra-cardiac-arrest interventions. So, it's very simple. We picked that dose because that was the one that has been used previously. I think the literature on sort of titrating your cardiac-arrest therapy, being it the chest compression or the vasopressors, based on invasive measurements such as diastolic pressure, systolic pressure, or other things, end-tidal CO2, is very interesting, but I think it's still very preliminary. Most of it is based on observational studies or animal studies, and I do think we need much more data before that can be commonplace. It will require a different setting. It might be possible in a well-equipped emergency department or an ICU, but as we showed here, many of the patients, especially in Europe, have a cardiac arrest on a medical or surgical ward, and before you put in that in-line and get everything set up, it takes quite a long time. So, we think it's important to focus on the things that are already recommended in the guidelines. And then, we try to see if we could add this still complex, but still straightforward intervention.

>> Chris Nutt: Okay, thanks. And I have a question I suppose about the other side of the intervention from Jerry Nolan, and he sort of stated that he felt that the return of spontaneous circulation was due to the vasopressin, and what do you think the methylprednisolone adds in that circumstance?

>> Asger Granfeldt: So, it's a combination of the two drugs, so we can't really separate the effects of these two drugs. We know that the methylprednisolone has some genomic effects that work right away that might affect receptor -- -- how many vasopressin receptors you have. We know that it has a vasopressor effect from septic studies that works right away, but we can't stop or separate the effects of these two drugs. So, that's impossible for us to say based on our data.

>> Lars Wiuff Andersen: And of course, some might argue that we should have done a completely different trial where we had randomized maybe only to vasopressin or only to steroids or done some form of factorial design where we had randomized to both, and I think that would be ideal. It would require many, many, many more patients, of course, so we felt like we had to start somewhere. And I think it will be a puzzle trying to come up with this next large trial. Should we randomize to exactly what Spyros did? Exactly what we did? Some combination? Should we give it -- how should we give it early? All these things. So, I think there's still a lot of work to try to figure out exactly how that trial will be designed and conducted.

>> Spyros Mentzelopoulos: May I essentially say something about methylprednisolone?

>> Rob Mac Sweeney: Of course. Yes, please.

>> Spyros Mentzelopoulos: Thank you. It's not only the potentiation of the pressor effects during CPR, but methylprednisolone may help the maintenance of ROSC and stabilize the patient after the return of spontaneous circulation by potentiating, again, mainly the effects of vasopressin, which has a longer duration of action.

>> Rob Mac Sweeney: Chris, we maybe have time for one more question.

>> Chris Nutt: Okay. Well, there were a lot more questions, Rob. Perhaps I'll just ask about the patient population and the type of arrest. So, there seemed to be a higher proportion of unwitnessed cardiac arrest in comparison to previous trials, and they seem to have a fairly poor outcome. Do you think this impacted on the overall results of the trial in your 30-day or your neurological outcomes? And should they be excluded or analyzed separately?

>> Lars Wiuff Andersen: So, I think -- I think the patient population reflects this patient population, and I think we tried to make inclusion criteria that were inclusive so that we could have as many patients with an in-hospital cardiac arrest as possible. And this, of course, is sort of a long discussion within critical care and many other things. Should we try to select a very small group of patients where a benefit might be or should be test it in a broader population? Right now, we do the same cardiac-arrest treatment for all patients, so that's why we thought that if we should test whether this add-on worked, we should test it on all patients. Whether or not there are more or less witnessed, I am not sure. I think it depends on what country you look at, and some countries have more cardiac arrests in the ICU and less in the ward. And of course, if they are in the ICU, they will be witnessed, they will be monitored, whereas we have many in the ward. The overall in-hospital cardiac-arrest incidents in Denmark is probably lower than other places, at least lower than the U.S. So, it's not like I think we have that many more on the wards. It's just that the proportion is different.

>> Asger Granfeldt: And it's also important to notice that, as Spyros mentioned, he had the effect in the asystole group and not the PEA group and not the shockable group. So, that's actually -- most often, if you have an asystole, you have an unwitnessed cardiac arrest before it actually develops into asystole from either PEA or VF/VT. So, those patients would normally have an arrest in the ward unwitnessed, and that's where the effect was seen in previous trials.

>> Chris Nutt: Thank you.

>> Rob Mac Sweeney: Fantastic, guys. Thank you so much for that. What we're going to do now is move into the panel discussion. Our panelists have been patiently waiting for quite a while. What I'll do is I will cross to Boston and to Kate Berg first. Kate, could we have your initial impressions on this trial, please?

>> Katherine Berg: Sure. First, like everybody is saying, I am very impressed by the trial, and I've taken much longer to achieve much smaller trials, so that's a congratulations to both of you and your whole group. So, a lot of the thoughts that I have had while reading this and hearing the presentation and similar to what some people have raised is that do we think this is the combination of the Solu-Medrol and vasopressin or is there really something about vasopressin that's important in this patient population? And I think, you know, as I think you guys both mentioned in your presentation, vasopressin historically is not nearly as well studied as adrenaline, and that's why adrenaline has ended up in the guidelines. There's a lot more data. But we haven't really looked at anything in the in-hospital cardiac-arrest population where the time to drug is so much shorter. So, I am wondering what you both think is the most logical next trial to do, if you would, you know, look again at the vasopressin/steroids combination or if you think that we should be comparing epi to vasopressin plus epi again.

>> Lars Wiuff Andersen: Well, that's an excellent question, and the honest answer is, "I really don't know." I really don't know. I think the problem is that if we first test the trial of vasopressin versus -- or as an add-on and that doesn't work, we might compare it to epinephrine, and if that doesn't work, and then we'd do what Spyros has just done, steroids versus placebo, and if that doesn't work, then we -- we have some lighting issues here. Then, we might end up in a situation where we have three negative trials, but it was actually the combination that worked. So, we decided to do it sort of the other way around, start with the combination, see if that works. And then, of course, you could later on then do trials where we look at the individual components. And that's maybe not the traditional way of doing it, but we thought it was the most pragmatic way of doing it. Instead of having to do five trials, we would start with one, because if the combination didn't work, we felt like then the individual components probably didn't work. Of course, we can't conclude that firmly, but that was sort of our belief. And again, we were just limited by what was feasible, both with funding and the number of patients we have in Denmark, so we could not include thousands and thousands of patients. And I don't know, before we started this presentation, Asger asked me, "Did you regret that we didn't do the post-cardiac-arrest hydrocortisone?" And I don't know, maybe. Maybe that's the difference. I don't think so based on our data, based on the hemodynamic profile of the patients, based on what they died of, but that's also a possible explanation. So, I think we have to really sit down, look at these trials, look at the individual patient meta-analysis, and try to figure out, "What is the next trial? Is it feasible? And how and where are we going to do it?"

>> Asger Granfeldt: I also think it's important to notice when we talk about vasopressors, we often think about epinephrine, as well as cardiac arrest. There's so much discussion about epinephrine vasoconstricting and causing brain injury, but the cause of death in these in-hospital cardiac-arrest patients is slightly different. Within our study, only one-third dying of neurological injury, and other causes are the remaining. So, there might be different starting population, and vasopressors may have had a very different effect on this compared to the out-of-hospital cardiac-arrest studies.

>> Rob Mac Sweeney: Great. We will cross over to Cologne. Bernd, your comments, please?

>> Bernd Böttiger: First of all, good evening. I am currently on a congress in Austria. So, congratulations for this great study. Everyone who is doing clinical studies knows what amount of work this is, and it's no surprise that we have more questions than answers after this study, as we have already faced. I have four brief issues that I would like to focus on first, and it's probably a question. I mean, in the vasopressin group, you just have more vasopressors or more adrenergic activity, so this reminds me to the high-dose epinephrine trials some many years ago where return of spontaneous circulation rate was also higher because you just had more vasopressors in the patients, but overall outcome was not different. So, this is my question number one. Question number two, and I am playing a little bit devil's advocate, 30-day outcome is not significantly, but absolutely 2% worse in the treatment group, and if you are looking to the Modified Rankin Scale, which was not visible in the presentation, but it's in the manuscript, also at 30 days, neurological outcome is significantly worse or at least the trend being worse in the treatment group. So, therefore, I think this trial does not support to use it already routinely, as you pointed out. Number three is we have probably 30% unwitnessed arrests, and time from collapse to study drug administration is a key issue, and I am wondering how you can calculate the time in unwitnessed cardiac arrest patients. And number four is, if I have seen it correctly in the manuscript, I mean, outcome after cardiac arrest is most dominantly related to the outcome of the brain. And therefore, the question is also what was the post-cardiac-arrest treatment? Have the patients received targeted temperature management? If I have seen it correctly in the manuscript, it was only 30% of the patients that have received targeted temperature management, and it would be great if you can briefly comment on this. Thank you very much.

>> Lars Wiuff Andersen: Thanks, Bernd. So, I think I'll try to at least comment on some of the first points, and then I'll let Asger on the last one. So, yes, adding vasopressin is just adding more vasopressor. We do think it has another pharmacokinetic and dynamic profile. It has a much longer half-life, for example. So, perhaps you get, as Spyros also mentioned, this prolonged effect, and you might avoid some of those dips in blood pressure, maybe even after return spontaneous circulation. Of course, it also works on other receptors than epinephrine, so we do think there is at least a theoretical added benefit of vasopressin. Then, yes, the point estimates do suggest harm in this trial. That is, of course, unfortunate, and it also is unfortunate if we want to do another big trial because some people might look at that. I do think it's important -- two things. The trial was not powered for this, and when you look at the confidence intervals, they include both relevant harm and relevant benefit, and I also do think we should not look at this trial in isolation, and that's why we try to do the patient-level meta-analysis as soon as possible. And hopefully, we'll get it submitted very soon so you guys can see that as well. So, I do think it's challenging with these results, but I don't think they mean that we should not do a larger trial. And then, I'll let -- oh, there was one more thing about the recognition of the cardiac arrest, of course. Yes. So, the time to something is the time from the recognition of the cardiac arrest, similar to how we normally define it in the out-of-hospital cardiac-arrest setting because the actual time is, of course, unknown in those with unwitnessed cardiac arrests.

>> Asger Granfeldt: Yes, and relating to the final question about the post-cardiac-arrest care, it's correct that only about 30% received TTM therapy in this trial, and it can be discussed whether that could have influenced outcomes. But based on the TTM2 trial, we think this is unlikely, although the HYPERION trial showed there might be a benefit in patients with in-hospital cardiac arrest, although it's important to recognize, as Bernd said, that the most common cause of cardiac arrest is brain injury. However, that is primarily true for out-of-hospital cardiac arrest and not in-hospital cardiac arrest. In our population, only one-third died of a neurological cause. So, I think [inaudible]. And I would also say the in-hospital cardiac-arrest population may be different with sepsis, infection, [inaudible], etc., and we have the old CAST trial demonstrating in sepsis patients that the cooling actually may be harmful. And also, the group tree is the same, so I don't think it relates to the difference between the groups.

>> Lars Wiuff Andersen: Yeah. And lastly, I think this is not unique for in-hospital cardiac arrest in Denmark. I think this very well reflects how patients with in-hospital cardiac arrest are treated around the world, and I think in the U.S. some observational data suggests that even fewer patients receive hypothermia or TTM after in-hospital cardiac arrest.

>> Rob Mac Sweeney: Fantastic. What I am going to do now is move to Ranjit. Ranjit will have to leave shortly, so I am very keen to get her involved. Ranjit, from a stats and methods point of view, how does this trial look?

>> Ranjit Lall: Thank you, Rob. So, I'd just like to congratulate Lars and his colleagues on a really well-conducted trial. I scrutinized the methods. And, you know, I was very impressed with what was -- how it was conducted. It's interesting to note that you recognize that there is a need for a bigger trial. And as you've said, your study was not completely conclusive inasmuch that the p-value was very, very close to 0.05. So, it's good to see that you're kind of thinking about a bigger study and also the fact that you've done a meta-analysis, which has interesting results as well. I just had one or two questions really in terms of the math as I noticed you did a modified intention-to-treat analysis, and 11 patients were excluded from that, and I wondered why not a straightforward intention-to-treat? And if you'd included those 11 patients, which way the conclusions might go?

>> Lars Wiuff Andersen: Yeah. So, thank you for your comments and the question. We had planned in the protocol to conduct it the way we did, as a modified intention-to-treat. Importantly, the trial is double-blind, and so there was no exclusion of patients based on what intervention they received. Patients were only excluded based on information that occurred prior to the cardiac arrest. So, these are basically patients who were included as a mistake. So, it was because the clinician in this hectic situation of a cardiac arrest perhaps included a patient who came in with an out-of-hospital cardiac arrest or included someone who had previously been included in the trial. So, since the exclusion is not dependent on whether or not they received treatment, this type of analysis does not result in any bias. But actually, because we removed some patients who were just noise in some sense, we get more precise results. So, I think that's why we did it, and you could argue you should just do a normal intention-to-treat, but that's how we decided, and I think it's reasonable given that it's a double-blind trial. It would be very different if it was not.

>> Ranjit Lall: And I wonder whether -- so, did you, as part of a sensitivity analysis, look at intention-to-treat just to see -- because the p-value is very -- you know, it's very borderline, isn't it? So, I wonder whether the inclusion of those patients would tip it over in either direction.

>> Lars Wiuff Andersen: Actually, we did not do that because we did not collect full outcome data on these patients because very early on, it was decided that they would not be analyzed. Going forward, that's probably something we would do in a future trial, make sure that we at least have the outcome data such that we can do such sensitivity analysis as you're suggesting.

>> Ranjit Lall: I was just going to also ask you -- your randomization was one-to-one I think, wasn't it? That's how it's specified. But when we look at the numbers, there's a little bit of imbalance there.

>> Lars Wiuff Andersen: Yeah, I think it's just by chance. We aimed to do this both. We used both block randomization and stratified randomization. So, the randomization was stratified by the site. And then within sites, it was block randomized. But of course, when we have many sites, even when you use block randomization, by chance, you can have a different number of patients in each group. And of course, this was not one patient comes in and gets randomized, then another patient comes in. This was a little bit of a -- not a mess. I wouldn't say that. But we had, you know, different code cards, different things all around. So, the block randomization was not 100% in the sense that all patients sort of filled out all blocks. And sometimes, the drugs were open, but not used. So, we lost a little bit of the advantage of having block randomization, and that's probably one of the explanations. But there is -- again, it was a double-blind trial, so I am not -- the only possible explanation is that it was a chance imbalance.

>> Rob Mac Sweeney: Ranjit, one quick question for you before you have to leave. It was mentioned -- I think Lars mentioned it -- that a trial to identify a 2 percentage point difference would require 8000 patients. PARAMEDIC-2 was 8000 patients, but that was in an out-of-hospital setting. Would it make a difference to try and recruit 8000 patients in-hospital versus an out-of-hospital setting? Would that be easier or more difficult do you think?

>> Ranjit Lall: I think that's more of a logistic question. I mean, with PARAMEDIC-2 -- -- I mean, I don't know whether Gavin is listening to this, you know, but I'd say, you know, that we had logistical problems. And, you know, kind of we were recruiting right across the U.K. over four to five trusts and ambulance services. And I think your question is more logistics. I don't quite know how in-hospital cardiac-arrest studies would be able to recruit 8000 patients. It's more a logistical question I think whether it's something that can be done in-hospital as opposed to out-of cardiac arrest.

>> Lars Wiuff Andersen: And maybe I can briefly comment on that, Rob, as well. We have just finished a trial in the out-of-hospital cardiac-arrest setting that we will hopefully be presenting and publishing later this year. I think it's hard to compare, and I wouldn't say one is easier or more difficult. I will say the logistics is more complex in the in-hospital cardiac-arrest setting. There is a different identification of patients and setting in the out-of-hospital and a smaller group actually of providers, whereas in the in-hospital cardiac-arrest setting, there are many more providers who have to train and who have to sort of be involved. In Denmark, for example, we also have 5000 out-of-hospital cardiac arrests a year. We have only 2000 in-hospital cardiac arrests. So, it's not -- of course, enrolling 6000 patients or 8000 patients in the out-of-hospital cardiac-arrest setting is much more difficult than 500 patients in the in-hospital setting, but there are different challenges. So, it will definitely require cross-country and collaboration to get that amount of patients.

>> Rob Mac Sweeney: Super. Thanks. Thank you very much, Ranjit, for joining us. We'll move now to Toronto. Laurie, thank you so much for being with us this afternoon your time. I am sure you'll have some initial thoughts and comments for Lars and Asger.

>> Laurie Morrison: Yes. So, first of all, they look like brothers in their matched attire and the blackboard background. But I just wanted to say, you know, just kudos to you both and to the Danish investigator team because I think anyone that conducted a cardiac-arrest, non-COVID-19 trial during the last 3 years should be commended. You were fearlessly courageous, and a well-done trial. Thank you. I did have a comment about it. As I read it through, I was worried that maybe we didn't achieve a significant difference or it's just close. But looking at their survival outcomes and your functional survival outcomes, I was just wondering if you had measured CPR metrics. Like, we found in out-of-hospital cardiac arrests that as soon as we measured CPR metrics and got both control and intervention group really hot, that a lot of our drug and device differences disappeared. And the same occurred when we finally got the ICU working with post-arrest care. And when it became uniform, like the TTM investigators, that a lot of our drug and device and processes of care measures became similar in both groups. So, I just wondered if you measured CPR metrics and post-arrest care metrics, if you could comment that they are not affecting your trial.

>> Asger Granfeldt: Thank you so much for this comment, Laurie. It's a very important question. Again, as this was a pragmatic trial, we didn't collect CPR metrics for the trial, so that's a very simple answer, and we can't show any data on this.

>> Laurie Morrison: And the post-arrest metrics, were any of the groups involved with the TTM trial? I guess they couldn't enroll in the TTM trial, but I would assume that the TTM protocol by the investigators would be the same applied to eligible or non-eligible TTM patients.

>> Asger Granfeldt: Yes, some of the sites. For instance, always where I work was participating in TTM2 trial, but that was out-of-hospital cardiac-arrest patients, so there was no crossover or influence on each other from those two studies. And the post-cardiac-arrest care was set to the discretion of the clinician treating the patient, and we published all the data on post-TTM care in the supplement with how many patients got PCI, etc., dialysis or therapy.

>> Laurie Morrison: So, it's just like the [inaudible]. Obviously, I heard Lars very clearly, Lars, when you said that you thought the in-hospital control trial in cardiac arrest was even more difficult than out-of-hospital. But eventually, we did get, you know, up to 90% of CPR metrics, and it may be worthwhile if you're considering something larger. I also had two other just suggestions, and that was this duration to ROSC, we find in out-of-hospital cardiac arrest that since we don't have any sort of genotypic bedside thing to separate an epinephrine responder from non-responder, that duration to ROSC can be used as a surrogate. So, you might want to go back in and see in in-hospital cardiac -- if you haven't already done this. Perhaps you've already done it. But duration to ROSC may help separate a group that actually benefit from the intervention or the control approach.

>> Lars Wiuff Andersen: Yeah, it's an interesting -- it's an interesting comment. Of course, we know that the shorter time to return of spontaneous circulation, the better you do either way. The problem is, of course, that that would have to be sort of a post-intervention subgroup, and then we have issues where the results might be biased because the definition of the subgroup could be influenced by the intervention. Also, it's a kind of difficult metric to look at, "time to something," because it sort of requires that you have it. So, we can't measure time to return of spontaneous circulation in those who do not have it. We reported in the manuscript what the duration is. I think it's 16 minutes in one group and 18 in the other. I think there is a lot of data here, and a lot of it we haven't analyzed yet. So, of course, it's always good with suggestions for going forward.

>> Laurie Morrison: Yeah, especially if you can pre-define it for the larger trial to say that maybe there is a group here that has a shorter duration to ROSC, for example, or a longer duration to ROSC, which benefits from the physiological effects that both Asger and Spyros elucidated so nicely. And the last was, I just wondered if you have or could publish any data you have on organ donation with vasopressin and steroid because there is an undercurrent where the concern is post cardiac arrest that, you know, the organs aren't useful, and we know they are, and I just wondered if using steroids so early in their cardiac arrest might actually lead to more salvageable organs for the non-survivors in your trial. I know it's a stretch, but there's just so little published out there, Lars and Asger, that if you have data on it, get it out there.

>> Lars Wiuff Andersen: It is a stretch, Laurie. Of course, unfortunately, we don't have that data. I think organ donation after in-hospital cardiac arrest, at least in Denmark, is quite rare. Sort of anecdotally, I don't remember any of the patients having been organ donors. I do think it's a very interesting question because, for example, in the PARAMEDIC-2 trial, we saw that there was an increased number of organ donors in the epinephrine group compared to the placebo group. And this sort of ties into a whole other discussion on whether return of spontaneous circulation is a valuable outcome in itself because it gives you more organ donors. There's also other reasons why return of spontaneous circulation in itself might be a valuable outcome. It could be that, for example, it's valued by patients and families. There's been some discussion about that, giving people time to say goodbye. Another issue is, of course, as you mentioned, post-cardiac-arrest treatment. There is a lot we don't know about post-cardiac-arrest treatment. And perhaps -- and this is a big perhaps -- going forward, we will find some treatment that can optimize outcomes also for some of these patients who had return of spontaneous circulation, but didn't end up surviving to 30 days. So, I do think it's a complex question where we can't just say, "Oh, it didn't work on survival, throw it out." I think we have to consider whether return of spontaneous circulation and some of these related issues such as organ donation might also be important.

>> Rob Mac Sweeney: Okay. We will move to Pittsburgh. Laurie, thank you very much. Chris, we've been saving the best to last, of course. Can we have your comments? And then, I'll come back with a question after that.

>> Christopher Seymour: Yes, sure. So, good afternoon or good evening to those of you joining us. So, of course, like all, I'd like to say congratulations to the Danish investigator team, to Lars, to Asger and others. A beautiful trial. It's good to hear this discussion. I know I've learned quite a bit from all of you as trialists and content experts. From a JAMA perspective, which is why I am joining today, we were very excited to get a trial like this as a submission during COVID, which was great and was -- when the adjectives up front are multi-center, double-blind, placebo-controlled, parallel-group trial, it always catches our interest. And on top of that, cardiac arrest, of course, remains an incredibly difficult problem and common, and it's fantastic for our readership. So, that's part of the reasons that we were excited to move forward, and this study team has been fantastic to work with. Our reviews and critiques have really -- that led to the final version we see today, you know, have touched on many of the topics that have come up in the discussion, which is, "Well, what's the best way to analyze this dataset? Is it, you know, accounting for, you know, site in models? Should we be using this intention-to-treat versus the as-treated population?" And then, as often is the case, reviewers will say, "Well, this isn't the trial I wanted." And then, you hear from the authors, "Well, that's the trial that we conducted." And true to form, this team stayed very true to their pre-specified protocol, which was published, and it was quite well done. So, thanks for having me.

>> Rob Mac Sweeney: Fantastic. Chris, a quick question. We've had three trials now in this field. We've seen the IPDMA that was presented earlier. Is there room for a fourth trial? Do we need a fourth trial?

>> Christopher Seymour: So, it is actually really nice to anchor this data with the previous papers by Spyros and others in the JAMA Network family. Do we need a fourth trial? Well, actually, we spent quite a bit of time thinking about what's the right interpretation from this one, which is there may be this difference in ROSC, but does the absence of benefit or harm in the longer-term outcome, you know, necessitate us going forward? That's not a decision for JAMA, of course, or even me as an intensivist, but for folks like Laurie and others on the call. So, if there is a trial, you know, we'd love to see it. But I think it's really interesting the sample size calculations and others that were shown during the initial presentation. It's quite a few patients. Perhaps there's an opportunity for a novel design that may use RAR or get us to an answer sooner.

>> Rob Mac Sweeney: Fantastic. And our panelists have been waiting a long time. We're over time, but I want to get around to each person just once more for a quick question. Kate, I am going to come back to you, and it's the impossible question. Why is the return in ROSC not translating into improved mortality?

>> Katherine Berg: That, as you said, is an impossible question, so I don't really know the answer, but I had written down the same question, actually, as how do we move forward when we have a difference in ROSC, but not in any other outcome? And I think, you know, often, similar questions have come up with epi or adrenaline, right? That the differences in ROSC are much bigger than the survival differences. And there are even some questions of neurologic outcome after epinephrine not being better potentially. So, I think either it's because -- you know, either we focus on getting more ROSC and then we work on improving our post-arrest care to improve their longer-term outcomes. I think the more negative interpretation of that is that when we are very aggressive at getting return of spontaneous circulation, we might be getting some patients back who already have some irreversible damage and that will then have -- will die later, you know, in the first 30 days or 90 days. But I don't think there's -- right now, we don't have any way to differentiate from those patients, you know, who's likely to have an okay outcome after ROSC and who has no chance of a good outcome. So, I think it's -- yeah, it's impossible to answer, but I think an intervention that improves ROSC and we then have room to intervene in the post-arrest side is worth at least pursuing.

>> Rob Mac Sweeney: Bernd, we heard earlier that possibly there were improved outcomes in those in the ICU who received this combination of therapy. Is there room for a trial looking at vasopressin purely in the ICU, as opposed to in a setting where it might be delivered slightly later? Could this be a place-specific therapy?

>> Bernd Böttiger: No one knows. There's always room for many further trials, and the question is, "It just the dose of vasopressors regardless whether it is epinephrine or vasopressin or what is it?" We don't know, and there is a lot to do and a lot to study in the future. And unfortunately, we probably do not have enough patients -- or luckily we do not have enough patients to have all these questions answered. So, to be honest, I cannot really -- I cannot really answer this question. But the main question is also, as Kate pointed out, is an increase in ROSC, which is most probably due to the fact that we can get the heart back to beat. Does that really help the whole body or the whole individual or is it only the heart that is then started again and later on dying again or, you know, while the brain is already severely affected? That's, for me, the most important question. That is also the reason why I think post-resuscitation therapy is important and TTM is also important. And I would like to repeat that I was really surprised that in Denmark, only 30% of patients received TTM. And even if you are saying TTM2 has already shown us that 36 is as effective as 33, nevertheless, in both groups, they have used TTM, isn't it?

>> Rob Mac Sweeney: Absolutely. Spyros, we've had three trials now in this area. Is it worth looking at alternatives? Perhaps terlipressin and prednisolone or hydrocortisone, rather than vasopressin and methylprednisolone?

>> Spyros Mentzelopoulos: Well, I would really recommend to clarify our knowledge with existing data and perhaps conduct a larger trial. I would power it to 5% difference, but I would also try to administer the drugs as early as possible to get the best result. This is what the VAM-IHCA results tell me as well. And then, perhaps, move on. I would be very -- I would be very cautious about any refinement of the combination because the combination is complex anyway. And may I point out that we have -- we have left out the post-resuscitation steroids, which are an intervention actually tested only in my trials. So, the first question is, "What are we going to test now?" Are we going to test the VSE again and try to administer the vasopressors as early as possible? These vasopressors have no -- this combination has no relation to high-dose epinephrine. It's vasopressin, which that does not have the detrimental effects of epinephrine on cardiac myocyte metabolism and the ATP depletion and ischemic contracture and arrhythmias. So, I think that we first must concentrate on how to optimize this combination. I think this is the best alternative. Not to try to modify or use a drug that will have a similar activity, but maybe more powerful or maybe more short-acting and so on.

>> Rob Mac Sweeney: Laurie, final question to you. Your focus is on time-sensitive interventions in emergencies. How do you think it would be possible to speed up the administration of these study drugs? In this trial, they were delivered very quickly, adrenaline in 5, study drug in 8, but can we get -- can we administer them more quickly than this?

>> Laurie Morrison: The answer is, "Yes." But I think Lars and Asger addressed that beautifully when they said it logistically was not possible as a study drug because the preparation has to be regulatory approved in order to put it out there to use. So, it becomes logistically difficult, but Spyros has already shown that if you put it in a single bolus, it's fast. So, we know that. I just want to commend that these investigators have reduced the time to drug that the ROC investigators couldn't do in the out-of-hospital setting. We were at a median of 19 minutes and hammered them for two straight years, and they've got it down to 8 minutes. So, let's -- I wouldn't beat them up for that. I commend them.

>> Rob Mac Sweeney: Fantastic. Thank you all so much in the panel. So, we'll finish it with a very quick final word from Lars and Asger.

>> Lars Wiuff Andersen: Yeah, I think we just want to end this by saying thank you to everyone, both the team who conducted the trial, but of course also to the panel here tonight. And of course, to you, Rob, and to Critical Care Reviews for hosting us. And lastly, for JAMA for publishing the trial. Thank you so much.