Effect of Lower Tidal Volume Ventilation Facilitated by Extracorporeal Carbon Dioxide Removal vs Standard Care Ventilation on 90-Day Mortality in Patients With Acute Hypoxemic Respiratory Failure

This transcript is auto generated and unedited.

Rob Mac Sweeney: REST is the largest yet multi-centered respiratory support trial. It's been led by two of my colleagues, Danny McAuley and James MacNamee both who work with me in the regional intensive care unit at the Royal Victoria Hospital.

Rob Mac Sweeney: We are joined by an expert panel as well as also Eddy Fan from Toronto who will be delivering the editorial. Our panel consists of Kathy Rowan from London, who we met yesterday, Liz Wilcox from Toronto, her colleague in Toronto, Art Slutsky, and from Melbourne, Carol Hodgson.

Rob Mac Sweeney: So without further ado, I will hand over to my colleague, James McNamee, who will deliver the results of the REST trial.

James McNamee: Many thank you, Rob. Can you hear me okay?

Rob Mac Sweeney: Yep. That looks good, James.

James McNamee: Okay. Well, I'll just go into screen share and you let me know.

Rob Mac Sweeney: That's perfect, James. Thank you.

James McNamee: Okay. Many thanks, Rob. Great meeting as per usual. Feels a bit like a coffee break really in our normal day's work. I'm presenting this study as clinical lead and on behalf of Danny McAuley, who was the chief investigator for REST. But also on behalf of a huge group of people and organizations. In particular, the NIHR, which funded the trial and Alung who generously supported the trial with equipment and support.

James McNamee: This was a 10-year endeavor. So, there was a lot of organizations that helped support us and endorse this. I just realized [EcNorc 00:01:58] is missing from that. So, thanks Kathy and EcNorc for their support as well. There's the disclaimer at the bottom of the slide. The study was sponsored by the Belfast Trust and supported by Queens University in Belfast.

James McNamee: The trial was managed by the Northern Ireland Clinical Trials Unit. In particular, I want to thank Collette Jackson on our team. I was relatively new to the trial management process, but you soon learn that a good trials unit is key. And we've got a good one on Belfast. So, thanks a lot to Collette and our team. Also to Jeanette, who was our REST trial nurse, who very diligently chased sites whenever they were struggling with recruitment. So, thanks a lot there.

James McNamee: And I also just want to thank the steering committee and the data monitoring committee for their sage advice and guidance, particularly at the end of the trial, and all our co-investigators who are listed here.

James McNamee: So, just to give some context to the trial, acute hypoxic respiratory failure is very common. The best and most recent data that we have is the lung safe observational study which was looking at patients admitted to intensive care who had respiratory support. And this was over four consecutive winter weeks in both hemisphere. And they found up to a third of these patients had acute hypoxic respiratory failure and two thirds of those had ARDS. So, we manage a lot of these patients, it's responsible for a lot of our work, no matter where you work in the world. It's a significant burden on both patients and healthcare settings as these patients often have longterm physical and cognitive impairment. So, in the UK alone, this is over 15,000 patients a year.

James McNamee: So, in trying to treat these patients, intubation and mechanical ventilation, although often the life saving intervention in critical care when patients are breathing spontaneously but struggling with an adequate gas exchange, ventilation comes at a cost. And our understanding has improved in that we now understand that the distribution of the injury to the lungs is often uneven. And ventilation, likewise, is often heterogeneous. And this is graphically illustrated by the earlier lab work by [Drevis 00:04:39] on animal lungs that demonstrated that sustained pressures and volumes causes lungs to become injured [inaudible 00:04:46].

James McNamee: And then as our understanding improved, we've begun to see that ventilation causes further structural consequences and biological alterations. We know now that the release of inflammatory mediators from damaged lungs causes systemic inflammation leading to multiorgan failure and death. And these adverse effects on patients can have profound effects years down the line. The prognosis has improved on how we mitigate ventilator induced lung injury. And this is the premise of our trial, is this seminal work by the ARDSnet group, the arm trial from 2000 that demonstrated a 22% reduction in mortality by the use of lower tidal volume ventilation.

James McNamee: In this case, moving from 12 mills per kilo predicted body weight to 6 mills per kilo. So, reducing tidal volume, limiting damage and improving outcomes in patients. The secondary analysis of that trial suggested when they plotted plateau pressure with mortality, that there may be no safe threshold for ventilation. So, like radiation, there may be no safe dose and that less may be more. And there may be further benefit in ventilating patient below what we would currently consider as safe.

James McNamee: So, important work by the Italian group of [Marco Manieri 00:06:25] and [Gavin Noni 00:06:29] took CT images of patients who were ventilated what we would consider as lung protective ventilation. As you can see in the top image in the red, up to 30% of these patients with ARDS still had hyperinflation. They're also able to demonstrate that with the introduction of simple extracorporeal CO2 removal device, which allowed them to turn the tidal volumes down to fours mils per kilo, they were able to reduce the hyperinflation but also, importantly, reduce the biological levels of lung injury.

James McNamee: So, extracorporeal CO2 removal or ECCOR is a form of extracorporeal life support that has really got three components. It's got a catheter, it removes venous blood through a pump past a lung membrane that exposes the blood to a large surface area and allows gas exchange where carbon dioxide is removed, and then the blood is returned back to the body. And particularly after our last pandemic in 2007, 2008, we notice that there was an increase in the use of these devices throughout the UK and worldwide.

James McNamee: So, we wanted to take a look to see what evidence there was on the use of these devices. When we carried out a systematic review in 2014, we found there wasn't a lot of evidence. Mostly observational studies. We found two randomized controlled studies. There was one in 1994 with the use of really redundant technology. But a more recent one from a German group, the extravent study, they looked at patients with ARDS as a cause of hypoxemic respiratory failure. They applied CO2 removal via the use of a simple pumpless device and turned the tidal volumes down, aiming for three to four mils per kilo.

James McNamee: They struggled with recruitment and the trial was stopped early, of 79 patients. Didn't show a difference in the primary outcome of ventilator free days at 60 days. But on post-HOC analysis, there was a suggestion that there was an improvement in respiratory outcomes in the more hypoxemic group. We use this to inform the inclusion criteria for our study.

James McNamee: So, we put a suggestion into the research prioritization exercise, which was scored highly in the UK. And subsequently there was a commission call by the NAHR to evaluate the clinical and cost effectiveness of ECCOR. And we were successful in not funding call. And that led to the REST study. So, I apologize for the convoluted acronym. I'll blame Danny for that one.

James McNamee: The REST study is protective ventilation with veno-venous lung assist in respiratory failure. And the hypothesis was in adult patients who require invasive mechanical ventilation for acute hypoxemic respiratory failure, veno-venous extracorporeal CO2 removal and lower tidal volume ventilation reduces mortality at 90 days.

James McNamee: So, this was a pragmatic study, multi-centered in NHS hospitals, open label, as the intervention would prove to be very challenging to blind. Although we were confident that the primary outcome was non-objective so that wouldn't be a problem. It was randomized in variable blocks with a one-to-one ratio and stratified for site.

James McNamee: So, what patients were we looking for? Patients in intensive care on invasive mechanical ventilation deemed to have potentially reversible cause of acute respiratory failure by the treating clinician of the day, and within 48 hours of the onset of hypoxemia defined by PF ratios of less than 20 kPa or 150 millimeters of mercury for our transatlantic colleagues. And that was trying to standardize it on a PEEP of five. So, if these patients have ARDS, it's in the moderate to severe group.

James McNamee: But importantly, this wasn't specifically an ARDS trial although we expected a large portion of patients to have it, we wanted to include patients with unilateral pneumonia, acute severe asthma as well. As data had suggested that the benefits of lung protective ventilation could be extended to patients that didn't have ARDS.

James McNamee: So, I'm not sure all the exclusions here, some of them are on the [inaudible 00:11:26] diagram as well. But importantly, patients that had been ventilated for more than seven days were excluded just to prevent patients being included that villi had already taken place, where the clinician thought that the tidal volumes could be turned down to three mils per kilo without the use of the device, anybody with hypoxic respiratory failure fully explained by fluid overload or left ventricular failure or that had an untreated primary cause such as pneumothorax or plural effusion.

James McNamee: Also patients that there was a decision to start ECMO, patients that had unconventional mechanical ventilation such as high frequency oscillator or APRV although these patients could be converted to more conventional forms and then recruited. And then patients not expected to survive. And importantly, heparin infusion was required due to an extracorporeal circuit. So, patients that were contraindicated to anticoagulation couldn't enter the trial.

James McNamee: So, what do we do as an intervention? We inserted a single dual lumen catheter under ultrasound guidance ECCOR was applied and maximized to allow tidal volumes to be turned down to aiming for three to four mills per kilo as long as the pH remained about 7.2. And 7.2 was found to be an acceptable level following a survey of UK intensive care physicians. Then ECCOR was continued for a minimum of two days to try and guarantee an effective dose, up to a maximum of seven days as we were limited due to regulations with the device. And then the device was waned when there was seen to be clinical improvement.

James McNamee: So, the device we used was the Alung Hemolung. The protocol was designed using generic technology in that you could have used any ECCOR device as there was no comparative data at that time between devices. But we wanted to potentially extend the benefits of lung protective ventilation to all centers throughout the UK. So, we didn't want this just done in specialized centers. So, we wanted a device where it could potentially be used in centers that were familiar with extracorporeal circuits such as renal replacement therapy.

James McNamee: And we found the Along device had the smallest cannula. It was a fairly simple device to try and use. And there was a well-developed support network throughout the UK. These centers were naive, we wanted support for the first two applications of the device.

James McNamee: So, it was a large training program rolled out. The Alung trained and supported sites. We had a full-time REST nurse who did additional training. There was also a manual produced with the algorithms and how to apply the device and ween it off and troubleshoot. We also put together an E-learning program. This is a screenshot from it here. ICUs often have a high turnover of staff. So, this allowed staff to come on, took part in these modules to get familiar with the trial and how to manage ECCOR and also how to report events if they were part of the research team. There was also a smartphone app so you can have all this information in your hand as well.

James McNamee: So, what was the intervention compared to? It was compared to a standard care arm. As I said, this was a pragmatic study. So, best care is six mils per kilo predicted body weight. So, we fed back pre-randomization tidal volumes to sites, just as a quality control measure to ensure best practice was the comparative arm. The rest of the care was usual care of acute hypoxemic respiratory failure, including neuromuscular blocking drugs, propositioning and ECMO was allowed on both arms, but ECCOR not allowed obviously in the standard care arm.

James McNamee: So, our outcomes were all caused mortality at 90 days for the primary outcome. And then for secondary outcomes, tidal volume reduction, ventilator free days at day 28 and duration of ventilation overall, ICU and hospital length of stay, the need for ECMO up to day seven. And for safety, we're looking at adverse event rates

James McNamee: On calculating sample size, the trial had an 86% par to detect a 9% absolute risk reduction. And this postulated effect was attributable to the arm trial that I talked about earlier. And this was assuming a control group mortality of 41%. And that number was taking from what was the control arm of the largest trial in hypoxemic respiratory failure in the UK to date, the Oscar trial, at that time, and also data from the EcNorc database on patients with equivalent levels of hypoxemia. And we came to us a sample size of 1120 patients.

James McNamee: So, we published our protocol nice and early for transparency. And the statistical analysis plan was made available before any analysis was done. And we began recruiting to the trial in May in 2016, started off as an internal pilot to test recruitment and then followed on with the trial.

James McNamee: We recruited for 44 months, recruited 412 patients from May 2016 to December 2019 in 51 adult intensive care units throughout the UK. And you can see there was a good spread of sites throughout the devolved nations. And just at this point, I want to thank all the investigators at all those sites and the research staff for the amount of work that you would've put in. Particularly in our own hospital, I want to thank our research staff in the Royal Victoria Hospital. It meant a lot of long nights.

James McNamee: I'll get in trouble if I don't thank my wife also, who was part of the research staff, Leah, she explained a lot of evenings coming home late. And at this point also, I want to thank patients and relatives who are involved which, without them, we wouldn't be able to answer these questions.

James McNamee: So, a good mix of sites. The plan was to move down to the south of Ireland as well, but unfortunately we didn't get to that stage. Recruitment was challenging. We had predicted 0.7 patients per site per month, and ended up with about half that. That was despite fairly aggressive site management, as PI's will know. They've probably my phone number blocked at this stage from calling them. But we had monthly teleconferences, six monthly get-togethers at conferences, trying to share ideas on how best to recruit.

James McNamee: There was repetitive Alung training and they went out to sites to support them if they hadn't recruited for a while. Jeanette, the REST nurse, went out to sites as well and online E-learning. So, we thought we did what we could in trying to recruit, but it was a challenge.

James McNamee: So, the trial continued till December of 2019 until it was stopped. Initially, it was paused due to an SAE after a third intracranial hemorrhage. So, there was an ad hoc DMEC meeting called. And that recommended changes to the informed consent document. There was also discussions with the trial steering committee and lay members on further advice suggested to include a CT brain pre-randomization.

James McNamee: And while that advice was being actioned, it was agreed to conduct an expedite planned interim analysis planned prior to 560 patients. Then accordingly, the DMEC recommended that the study should be stopped for futility and feasibility.

James McNamee: So, we screened over 7,000 patients. This shows most of the exclusions here, big hitters are contraindication to heparin, other patients not expected to survive. Most of the other reasons there are physicians were uncomfortable with patients potentially getting recruited to the intervention, as they thought the patients were fairly rapidly improving and would get extubated within 48 hours. That left us with a sample of 412 patients, which were analyzed on intention to treat.

James McNamee: So, we had four patients that withdrew, two patients that were lost to follow-up, and one was re-randomized in error. And altogether that left 205 patients for the standard care group, and 200 for intervention. So, when we look at baseline characteristics, so the table one, you would see in manuscript, these patients were typical of ICU trial patients, predominantly male, average age of about 60, fairly active patients, so able to live without assistance, but sick.

James McNamee: So, an APACHE scores of 20, and SOFA score [inaudible 00:22:07] 10. The predominant diagnostic admission category was respiratory and/or sepsis. And these were mutually exclusive. You could satisfy a couple of categories, hence the numbers. And these are mean and percentages. Not unexpectedly from the lung CF data, over 60% of these patients had ARDS as a cause of hypoxemic respiratory failure. The predominant cause being pneumonia, again, plus or minus sepsis. And again, you could have satisfied two criteria here.

James McNamee: So, when we look at the ventilation perimeters, again, this just reflects sick lungs that most of the patients were mandatory ventilated. Mean PEEP of 11. Driving pressure of 16. This is pre-randomization. And tidal volumes reassuringly at 6.6 mils. So, the standard care in the UK was good. These patients also had a mild respiratory acidosis, but mean qualifying ratio of 15. So, sick hypoxic patients. Exactly the group we wanted.

James McNamee: So, when we looked at intervention fidelity, we were happy. 92% of these patients received ECCOR when randomized to the intervention. Of the 17 patients that didn't get the intervention, eight improved, two had deteriorated, and there was device issues with six. For those patients that got ECCOR, the main duration was 4.2 days. That's just over a hundred hours of ECCOR. And it was successfully waned in 28% of patients, stopped due to the seven-day treatment in 18%, and discontinued due to medical withdrawal in 16%.

James McNamee: So, the ECCOR was applied and tidal volumes were waned. And you can see here, the separation between the two groups. And these are means plot with standard deviation. And significant difference from day two to day seven. Day one on the device often was a small day because that encompassed randomization, the intervention being applied, tidal volumes being waned. So, for the purposes of this data, we took from day two. And the main difference was two to two and a half each day.

James McNamee: To two and a half each day. When we looked at a per protocol analysis, the main difference was 2.4 mills on day two to 2.9 mills on day three. So good separation. And with that, there was a reduction in plateau pressure between day two and day four. The main difference was about 2.3 centimeters of water. We also noticed these patients were a little bit more hypoxemic. And that wasn't surprising, from anecdotal data, that when you use lower tidal ventilation you often see tidal derecruitment. And as a result, the peak was in increased on that intervention group. We found that there's significance difference on day three to day five.

James McNamee: A main difference of, peak, of about one and a half centimeters. So not a huge amount, but difference nonetheless. And with those ventilatory changes, with a significant difference in driving pressure, main difference of between 2.4 and 3.3 centimeters of water. Those changes then resulted in a respiratory acidosis despite the use of a CO2 removal device. So, small changes raised PaCO2 in the intervention arm and lowered pH, although small. So the intervention was applied. Tidal volumes were separated. So, we're onto the money slide here, so drum roll. The difference between the two groups, no difference. So the mortality in the standard care group was 39.5.

James McNamee: And the mortality in the intervention group was 41.5. Risk ratio of 1.1. And when we looked at a sensitivity analysis, which was performed for the primary outcome, excluding the first two intervention arms to try and address any potential learning effects, there was also no difference in the primary outcome. And when we looked at a per protocol analysis, which excluded the 17 patients that didn't get ECCO2R, there was also no difference in the primary outcome. That's the Kaplan-Meier curves there. You can see the two curves coming together.

James McNamee: So, apologies it's a busy slide, but there was a number of pre-specified subgroups. And you can see that while the patients in ARDS or not, their degree of hypoxemia, plateau pressures... How familiar they were with ECCO2R. So depending on the amount of participants of site, or importantly driving pressure, or how sick they were, all had showed no difference in the primary outcome. For the secondary outcomes, we did show a significant difference in ventilator free days of 2.1 days. So, less ventilator free days on the intervention. So ventilator free days is a composite outcome of mortality and ventilation. Death having zero ventilator free days.

James McNamee: And we think this difference may have been driven by a nonsignificant difference in mortality between the two groups. So slightly more died in the intervention arm, although be it nonsignificant. That may have driven down the ventilator free days. But we looked at the other secondary outcomes, such as duration of ventilation, need for ECMO between the two groups, 20 a day mortality, ICU length of stay or hospital length of stay. There was no difference between the two groups, no significant difference, although small numbers for some of those.

James McNamee: So when we looked at adverse events for safety, we found that adverse events were more common in the intervention group. And that was not unexpected due to the open label nature of the trial. You do tend to find under reporting in the standard care arm. When you look at the SAEs, which are intervention related, there was 15 compared to zero in the standard care arm. Most of these were hemorrhagic. Five were device related intracranial hypertension, and three of those fatal. But importantly, this is well within the rates that you would see with extracorporeal lung support in patients in hypoxemic respiratory failure, and often what you would see in control groups in those studies as well.

James McNamee: So when we combine the AEs and AEs in topics that may be of interest, we see that there was three episodes of haemolysis in the intervention group on four episodes of heparin induced thrombocytopenia. We also looked at the daily use of adjunctive therapies for that week. Now the table showing this was fairly busy, so I'll just sort of summarize in text. In that if you were in the intervention group, you're more likely to be ventilated with a mandatory mode and more likely to receive neuromuscular blockers. And I suppose that was just reflecting physicians just trying to control ventilation and trying their best to keep tidal volumes down low. But, in the intervention group also found to be ventilated less frequently in the prone position in days one and day two.

James McNamee: And these rates were in keeping with what we've seen in the lung safe data. I suppose this just reflected, again, the intervention while that's applied on a sick patient. A reluctance to prone, but then after day two becoming familiar. And then the grips were the same. So just in conclusion, our summary was for lower tidal volume ventilation facilitated by extracorporeal carbon dioxide removal. It did not reduce mortality of 90 days compared to standard care for patient and on mechanical ventilation for acute hypoxemic respiratory failure. And many thanks for listening. Over to you, Rob.

Rob Mac Sweeney: Thanks very much, James. That is fantastic. I have to say a huge congratulations to yourself and to Danny, Andrew Boyle, and another member of the research team, [Jenest 00:31:59], the research nurses, and the wider team as well. This appears to be an excellently executed study. You've got the right population. You've both, the control and the intervention group, received protective mechanical ventilation, which really allowed the intervention to be closely examined. I think it's a very impressive setting to test extracorporeal CO2 removal as you have described. So I'll now hand over to Eddy Fan in Toronto, and we'll get the editorial please.

Eddy Fan: Okay, great. Thanks Rob. And thank you James for the fantastic presentation. I think we've all been waiting, anxiously, to hear the results of this fantastic trial. So I'm going to speak briefly and give some thoughts about the REST study. Here are my disclosures. Maybe two amongst the important ones are that I provide consulting for ALung Technologies which provided devices for this trial, but I won't be speaking specifically about their devices. And the trial was endorsed by the International ECMO Network, and I'm a member of the executive committee. So I've heard about this trial a number of times previously. So, I think it's important to understand the context in which this trial, the REST study was performed. There's been lots of rationale that James has nicely presented in his talk regarding the fact that mortality from both ARDS and acute hypoxemic respiratory failure remains high despite the use of protective mechanical ventilation.

Eddy Fan: And there's been lots of preclinical and some early clinical data suggesting that going lower with tidal volume to mitigate ventilator induced lung injury further could lead to further benefits. And at least one study recently that many of us on the panel participated, the supernova study, was a pilot say that showed the proof of concept, safety and feasibility, of a strategy of extracorporeal CO2 or ECCO2R facilitated protective mechanical ventilation where tidal volumes were lowered further from six mills per kilo to fours mills per kilo could be done. And that this could be done safely and feasibly. And that obviously advocacy would be subsequently needed to be shown in a larger clinical trial, such as that undertaken by James, Danny, and their entire team. And importantly, a trial was needed because there are very important costs or risks associated with this technology.

Eddy Fan: And again, very nicely demonstrated by the systematic review led by Danny and James showing that, in the available data, that the use of these kinds of devices, not just full blown ECMO, but ECCO2R, is associated with significant local complications like limb ischemia and especially those that he had mentioned in the trial. Things that related to bleeding, particularly since the use of these devices required the use of anticoagulation. So now we're going to talk about the REST study. And I really want to start by congratulating James and Danny for this fantastic result, and their entire team. It's very difficult, as many of you are aware, to enroll very sick and complex, critically old patients into these kinds of trials. This is a complex intervention using a device and a strategy of mechanical ventilation that involved a huge interprofessional team. They rolled this out of across more than 50 ICUs in the UK, and more than 7,000 patients screens.

Eddy Fan: So congratulations for this fantastic effort. As you heard, this was a response to a health technology assessment call from their national funding agency. They really completed a very nice pragmatic open label clinical trial. I guess we'll hear more about the cost data later on that they've collected. And even without some of the requirements that were present in the call for this trial, this would've been a challenging clinical trial to design and implement across many centers, even without some of those requirements. And Danny maybe summarized best yesterday in a tweet saying that, in follow up to this or lead up to this talk today, that just 10 years from initial submission to prioritization, five years from first patient recruited. So a vast amount of work that needs to go into this kind of a trial, design and execution. So congratulations. This slide is just to remind you that, again, that although we like to label these trials as device trials, like it's an ECCO2R trial, it's an ECMO trail, it's some kind of sort of focused on the device, that really it's a bundled intervention.

Eddy Fan: And so we're looking here on the right, that the ECCO2R group or the ECCO2R facilitated group, it's not just the device, but it's also the ventilation strategy that gets paired with the device. The device facilitates, again, a reduction in the intensity of mechanical ventilation here, primarily, by targeting lower tidal volume. And it's the bundle of that intervention together that really is being tested in this clinical trial, and it's really not just the device in and of itself. So just some quick important notes. And again, James has done a great job in being transparent about all the results in his presentation. The groups are very well balanced. I think it was just interesting to note that there were low rates of prone positioning in this group. At baseline, just 11%. At least this group, at least from a P/F ratio standpoint, would certainly qualify perhaps for prone positioning.

Eddy Fan: Again, and just to highlight that it's a complex bundled intervention of ECCO2R and lower tidal volume ventilation targeting three or less males per kilo predicted body weight. And importantly, as when we get to the results, the requirement of systemic anticoagulation for ECCO2R. And a huge number, 1500. Just over of the 7,000 patients screened were excluded because of contraindications to systemic anticoagulation. They have good adherence to the ECCO2R intervention. I think is important to note that, again, they rolled this out across both expert and many non-expert centers in the provision of extracorporeal support. But despite sort of the rollout of this, James has already highlighted, patients that were randomized to the intervention group were more likely to be on controlled mechanical ventilation and on neuromuscular blockade up until day seven. And importantly, they also had less prone positioning mostly on day one and slightly less on day two, and then more comparable following that.

Eddy Fan: And that might have had some effect on the results or the differences in the results between groups as well. They seemed to achieve their goals. So with the relatively good levels of adherence to ECCO2R, it did its job. And what you could see in the red lines, and these few sample graphs that I provide here, is that they did have a reduction in mechanical ventilation intensity. And that was evidenced, especially by panel a here, where, again, their target was to reduce tidal volume. And that happened with the use of the ECCO2R device. And interestingly and importantly, that led to a concomitant reduction in plateau pressure, driving pressure, which again suggests that intensity of mechanical ventilation was reduced, which is the rationale for this bundled intervention. Interestingly, they had higher respiratory rates in the ECCO2R group, and maybe we'll talk about perhaps the importance of that on their results.

Eddy Fan: And this sort of altogether, despite having relatively similar physiologic parameters, then again James has already highlighted such as pH, PCO2 and relatively similar P/F ratios. You already saw the results. There was really no significant difference in mortality. Although, the intervention group had fewer ventilator free days at day 28 in this group and more adverse events. Interestingly, the trial was stopped early for futility and feasibility. Maybe we'll hear more about the feasibility aspect of this decision from James and Danny in the discussion. But certainly I think one of the things that might have driven this is that it was powered, the study was powered for a very large absolute risk difference of 9% that they took from ARMA. And it's interesting to note because, again, you might recall that the 9% absolute risk reduction from ARMA was reduced mortality from 40% to 31% in a group that was randomized to 12 mills per kilo predicted body weight tidal volume.

Eddy Fan: So reduction of six mills per kilo led to this large change. And targeting a smaller reduction in tidal volume was I think maybe not likely to result in such a large difference in mortality in these patients who are already receiving about six and a half mills per kilo tidal volume. At baseline, sort of adhering to current evidence based practices. I think a strength of this study, as James has already presented, is the number of secondary and sensitivity analyses they performed, and their results were robust to most of these analyses. They had a pre-specified adjusted analysis that showed no those very similar. They performed a per protocol analysis eliminating about 18 patients. Those who didn't didn't actually receive ECCO2R or the one patient that did actually receive ECCO2R in the control group, and that really was similar. A sensitivity analysis to address, again, the potential learning effects, which may have been important in this study given that many non-expert centers had been enrolling patients.

Eddy Fan: And by eliminating the first two patients, the results were still similar. And there was no effect modification as you saw in the graph that James showed in many predefined subgroups. I think an important highlight of this trial is that there were significantly more adverse events in the intervention group, particularly the dreaded complication of extracorporeal support, which is intercranial hemorrhage. Eight of them in the ECCO2R group and none in the standard group. And again, this is just to, James already shown you there's really no evidence, clear of that evidence of the effect modification across subgroups that they examined in this study. So what now? What do we do now with these results we've been anticipating for some time in the field? And again, a congratulations to James, Danny, and the entire team. No difference in their primary outcomes, some worse secondary outcomes. And more adverse events despite this reduction of about two mills per kilo predicted body weight and tidal volume, and concomitant reductions in driving pressure and plateau pressure.

Eddy Fan: So clearly the REST trial at this point does not support the routine use of ECCO2R in the clinical setting for patients with moderate to severe, again, ARDS, which is the majority of patients or acute hypoxemic respiratory failure patients with a P/F ratio less than 150. There is more data to come I'm sure from the cost effectiveness analysis. And also the fact that the authors' sort of, the investigators' sort of suggest that they might also use novel methods to identify potential heterogeneity of treatment effect that may be forthcoming. So some questions that we're left with now are: Given the potential adverse events associated it seems with ECCO2R and extracorporeal support, do we really need to use this device or this technology to reduce mechanical ventilation intensity and hopefully improve outcomes further? Or do we really need to go back to the drawing board and rethink how we design this intervention to hopefully improve outcomes in patients further?

Eddy Fan: I think it gives us pause that, again, there's a target for a modest reduction. In most trials to date of extracorporeal CO2 removal have targeted this sort of reduction to three to four mills per kilo from six meals per kilo. So a relatively small change in tidal volume to see if that would lead to a benefit or translate to a benefit in clinically important outcomes. This nice study recently published by John Christoff Rashard shows that actually, in a reasonable proportion of patients, we can actually reduce tidal volume to about four mills per kilo without using ECCO2R. So if we have all these dreaded complications from the device, particularly the need for anticoagulation, at least in this very small study of 35 moderate to severe ARDS patients, tidal volume could be reduced to four mills per kilo, increasing respiratory rate to maintain a relatively normal pH.

Eddy Fan: And again, a very similar target to that used in REST. However, the trade off here weren't complications such as bleeding or intracranial hemorrhage like we saw in the REST trial, but other complications patient such as cor pulmonale, transient severe acidosis. So these things are important. But just to say that there is a potentially the, when targeting such a modest reduction in tidal volume, there could be a substantial portion of patients where we could do this without the need for extracorporeal support. And again, to highlight that the trial really didn't just test the device, but really is the device plus some ventilation strategy. And of course, we need to find the right balance of how much to ramp up extracorporeal support so we can ramp down on mechanical ventilation, again, which we think is the mediator of much of the ventilator induced lung injury.

Eddy Fan: And then thus the worst clinical outcomes that we want to modify with these kinds of strategies. And again, so the idea maybe that we do need to go back to the drawing board. Perhaps targeting such a modest reduction in tidal volume is not enough to be translated into a clinically important reduction in mortality, or improvement in ventilator free days, or organ failure, these sorts of things. And again, that in thinking of going back to the drawing board, we now have some more preclinical data suggesting that more dramatic reductions in the intensity of mechanical ventilation. And this is just one such study, recently published in the Blue Journal, showing that models of near apneic ventilation might be the way to go. And so, again, that perhaps one of the reasons that the REST study really didn't show a change in a difference in outcomes is that we need a more dramatic reduction or more aggressive reduction in the intensity of mechanical ventilations to see a difference.

Eddy Fan: So, should we abandon this idea of using ECCO2R and more protective ventilation for ARDS? Maybe we need to target different patient subgroups and forget about the P/F ratio as a crude way of figuring out what patients to use this in. Do we need a different dose? And I think the trial tried to use a minimum of two days and up to seven days sort of limited by the regulatory limitations around the device. It was actually discontinued at seven days, and about a fifth of patients due to these device regulations. Maybe we need to continue it in these patients to have a more sustained reduction in mechanical ventilation intensity. And again, that might help show a difference in results. Maybe we need to define different targets. Again, sort of a modest reduction in tidal volume led to a modest reduction in driving pressure. But maybe we need to target some more holistic markers like mechanical power or some non-traditional targets such as respiratory rate.

Eddy Fan: And then again, importantly, that was increased in the ECCO2R group in this study, that might have led to some differences. And again, this idea that most studies have, to date, have targeted three to four mills per kilo. But maybe a greater magnitude of reduction in mechanical ventilation intensity is needed to see a signal in clinic and patient important outcomes. And so how the ventilation strategy works is going to be very important in these studies.

Eddy Fan: So some ways to perhaps tackle that, and maybe Art will speak about this as a part of the panel. Is that, again, rather than using P/F ratio, perhaps we need to target those patients who might have the best chance of showing improvement with this kind of a strategy. And again, this is some modeling work from Ewan Gallagher and Art Slutsky showing that perhaps those with the lowest respiratory system compliance are those with the highest alveolar dead space fraction. We might enrich trial populations in the future, learning what we have from REST, that using perhaps P/F ratio less than 150 will not perhaps get the target population that will be successful in a feasible, again another important result from the REST, a feasible amount of time when use of resources to answer the question of whether extracorporeal CO2 removal could be useful.

Eddy Fan: So perhaps enriching populations could be one strategy. Again, the idea that perhaps using a more sustained approach and a target of driving pressure might be more important. We've shown that, again, that it seems that high driving pressure seems to be associated with the worst outcomes across the spectrum of patients with acute hypoxemic respiratory failure. And that effect seems to be quite durable across the whole duration of their ICU and mechanical ventilation exposure. And so perhaps using the device to lower tidal volume or driving pressure for a more stained period of time might lead to more important benefit. And then finally, perhaps using a target like mechanical power that uses some non-traditional targets like respiratory rate.

Eddy Fan: And again, just to highlight that perhaps there was a small contribution of the fact that in the intervention group, respiratory rate was actually increased. And that's an important component of mechanical power. This editorial by Michael Quintel and [Luchiano Ganoni 00:47:54] showing that in various cohorts, and this was the near apneic ventilation study, that not only reducing tidal volume can lead to a reduction, an important reduction in mechanical power, but in reducing respiratory rate. It might be another target to add that might lead to the maximum reduction in mechanical power. And perhaps reducing mechanical power and finding that trade off between the reduction in mechanical power and best alveola ventilation might lead to that holy grail or that Goldilocks just right sort of situation to reduce [inaudible 00:48:23] and then improve outcomes best. So in terms of where we go from here, and again, the REST trial's going to be important not only for the clinical results it's provided, but it's important impact on the future of either current or ongoing trials of ECCO2R extracorporeal support in patients with acute respiratory failure.

Eddy Fan: So maybe we could learn to enhance the feasibility of future trials, or maybe we just need a different extracorporeal strategy altogether given what we've learned from the REST study. And again, I think this is very important because device trials, and I'm sure James and Danny will tell us, are extremely difficult to run. Only 6% of their screened patients were enrolled. And even if there was a positive result, there would be issues with generalizability perhaps of their results. And so, again, feasibility is so important because there are harms from uninformative clinical trials that are ultimately not feasible because a lot of resources are diverted, time is spent, and opportunity costs of studying something else or something more feasible where we could have an answer. This is a very important, especially in the device space. And feasibility as we've learned from the REST trial, just like others trials of extracorporeal support, is very important.

Eddy Fan: And we heard that James say that they had estimated to enroll 0.07 patients per unit, per month, but enrolled half of that across their sites. Again, this is a common theme that we've seen in other trials of extracorporeal support where there's very low enrollment. Feasibility is a major challenge, and this is something that we need to work on. And again, just to highlight, again, Danny's point that it took 10 years from initial submission, five years from first patient enrollment, to get to where we are today. It's sort of a huge and important undertaking. We're so thankful.

Eddy Fan: ... today is a huge and important undertaking. We're so thankful for James and Danny and their team for having undertaken this trial. Money is the key, as it is for many things in medicine and in life. And this is some sage advice from Rinaldo Bellomo from dasSMACC a few years ago, that when you are thinking about these difficult trials, there's only two questions that you need to ask yourself. Do you have the money? And do you have the patients? Of the two, money is the most important. If you have the money, people will get you the patients.

Eddy Fan: And I think, again, in the setting of device trials, we're talking about an expensive and resource intensive intervention, collaboration, liken the REST trial with industry funders and a broad group of stakeholders will be important to get these trials done, because trials are becoming more and more expensive. These are data from the United States showing that pivotal trials for therapeutic agents are now reaching the 5 to 20 to 50 million dollars to complete, depending on what you're studying.

Eddy Fan: So money is going to continue to be important. And so finding ways to improve the use of these limited resources of time, to make the things more efficient are important. And again, perhaps the next ECCO2R trial might use innovative trial designs, like embedding into registries, using adaptive platform designs, perhaps Bayesian designs, would be very important to enhance both feasibility and efficiency of these trials in the future. Hopefully, now with the REST result in hand, that this will actually have, I think, a very important impact into a number of planned ECCO2R trials that are coming down the road and the sun, or the follow-on, of the SUPERNOVA trial we discussed at the beginning.

Eddy Fan: Maybe it will make us rethink both the intervention and the strategy that we are using around these trials. So it is so important for the REST trial to have provided these results. Important amongst these trials that might be performed in a number of jurisdictions is to harmonize data collection and use similar core outcomes and plan for an IPDMA of all three trials at the end, much along the lines of something that Kathy Rowan, maybe she will speak about, that was done, for instance, for early goal-directed therapy trials in sepsis. This might help us to also examine heterogeneity of treatment effect and explore important subgroups in a larger harmonized cohort than we could in any smaller individual trial.

Eddy Fan: So in the end, maybe it is just a different extracorporeal strategy. And again, this idea and the allure of using ECCO2R, because it uses smaller cannulas. Maybe it leads less anticoagulation, maybe it's less risky in patients. I think the REST trial maybe puts some of that idea or thinking to rest, no pun intended. Given that many risks are fixed for the use of extracorporeal support, I mean, really all we're talking about is a different sized pump, perhaps a different size membrane, ultrasound-guided seldinger insertion of a cannula that might be slightly bigger. And indeed similar, if not maybe more need for anticoagulation when using lower blood flow in an ECCO2R device.

Eddy Fan: So again, if the rationale is to reduce mechanical ventilation intensity as much as possible to translate to lower VILI, and to translate that into an improvement in mortality, maybe we just need to use full flow VV ECMO to maximize the potential efficacy of this therapy. And indeed, using high flow VV ECMO may also allow us to reduce the amount of anticoagulation that we need. And in some studies now, none at all, and sort of abrogate the need for any of these risks that we're talking about. So with that-

Rob Mac Sweeney: Eddie, I'll need to cut in here, I'm afraid.

Eddy Fan: Yep. Finished.

Rob Mac Sweeney: Yeah. Thank you very much for that.

Eddy Fan: [crosstalk 00:53:27] panel discussion.

Rob Mac Sweeney: Perfect. That was a great editorial, lots of very interesting points. I will turn to James and Danny now to maybe reply briefly, if possible, guys, to the points that Eddy has made.

James McNamee: I will let Danny jump in afterwards. I suppose just a couple of points. Thanks Eddy for that really, really interesting breakdown. It is important that this was a ventilation trial. We really wanted every site, not specifically extracorporeal sites, to run it. So it's almost we actually sort of played down the use of the device. I think these devices do what they say on the 10, it's just how you use them, and how you use them to manipulate the ventilation. And if someone does not have VILI and you put on the device, the device isn't going to reduce VILI, so I think how you manage the ventilation is key.

James McNamee: [inaudible 00:54:20] was a bundle of care. It's really what people would do normally in their day-to-day practice. We were very keen that patients were proned, if the physician thought they were proned and we tried to drive that home as much as possible. I think they done what they could. When you use these devices, things are a bit chaotic once you put them in at the start and you want to see the patient iron out a bit before they flip them over. There's definitely a comfort level once you find the first few patients who are randomized, you find sites become more comfortable with proning patients initially.

James McNamee: So I think they eventually bounced out over a couple of days. The neuromuscular blocking drugs, I am not sure that had a huge effect on the outcome, with more recent data that we've seen. We didn't want to ask physicians to keep the device on for longer than they felt the patient should have been intubated. So it was very hard to operationalize how the device was weaned, actually. But if the physician felt the patient was improving and particularly the gas exchange was improving, we had to allow them to wake the patient up and try and get extubated. And as these sites were naive in extracorporeal lung support, asking them to keep on the device for an awake, agitated patient to facilitate extubation, would have been a challenge.

James McNamee: I think that is a different question that needs answered in another trial and an interesting one, whether it facilitates extubation. But it would have been a huge ask, because feasibility is already tricky. I suppose, just about the absolute risk reduction, we were going by the Hager study. It seemed to be a linear relationship. Maybe it was a big ask to expect 9% absolute risk reduction, but that's all we had. You are right about the heterogeneity of treatment effect. These weren't just ARDS patients, these were all comers. We tried to remove as many of the patients as possible that we thought wouldn't benefit from lung protective ventilation. But there's definitely a range of patients there.

James McNamee: I suppose they were a more hypoxic group than [inaudible 00:56:38], so we were trying to enrich that population, use predictive enrichment. But certainly enrichment of these patients is probably the way to go for further trials. But there is only so many patients out there. And I think international collaboration is going to be necessary to try and recruit the patients. We screened every one of the patients coming into the unit. We have a big unit and all types of patients coming in. And there was months where we did not find a patient either and that's with 24-hour phone on the ready, willing to come in to recruit them.

James McNamee: I think it is going to require some sort of enrichment stroke adaptive type trial if you are going to look into this further. One other point, just about the [Richard 00:57:26] paper. Yes, they were able to turn the tidal volumes down without the use of the device. They did say they had a couple of severe episodes of acidosis and cor pulmonale. We have carried out echos on a subgroup of these patients for more exploratory analysis. So it will be interesting to see the differences, particularly on the right heart, on how the intervention affected that, because it definitely is a potential risk. You did find there was an inflection point as you turned down the tidal volume, the PCO2 shot up. So that was something that we experienced, as well. So I may have done enough talking. I will let maybe Danny jump in.

Danny McAuley: I'll be brief, because you have covered I think all of the important points. I will have some inflammatory biomarker work in a small subset, as well, which might get at the question of, did we have a biological effect, even if we didn't see it translated to a clinical effect? So that is coming, as well. And I guess one theme that seems to have come up, just looking at the feed, is the question around mechanical power. And I think that is a really interesting question. It's interesting, whenever we presented these results to the DMEK, Taylor Thompson commented that the three negative lung protection studies, [inaudible 00:58:57], all had a respiratory rate in the twenties, whereas the ARMA study had a respiratory rate in the thirties.

Danny McAuley: He thought that that might actually be the other way. So I think we don't know that question, but it is really important to try and explore it further. We certainly will do. Rob, with that I think I will stop because I would like to hear the panel, rather than me.

Rob Mac Sweeney: Excellent, Danny. Good idea. So at this point, we will move to Phil [Gillan 00:59:34] and some questions from our chat function and from Twitter. Phil, if you could just take a few minutes to bring in some questions.

Phil Gillan: Certainly. Thanks Danny. Thanks James. A huge amount of work, certainly on both the chat function and on Twitter, there is a lot of praise for the work involved. So there are a number of quite technical questions that I would like to get some answers to, essentially. Some of them revolve around subgroup analyses and others revolve around adverse events. So I will start with the adverse events.

Phil Gillan: First of all, some discussion as Eddie pointed out, as well, on the use of heparin to achieve anticoagulation. Do you think there was alternatives to that or actually do we need anticoagulation as one of the major side effects was intracranial bleeding? Other questions regarding the propositioning and the propensity to do that in the various trial centers and CO2 removal. The other questions revolve around the level of what we feel would be an acceptable respiratory acidosis, by which we want to intervene, either with CO2 removal or not. So a number of different questions.

Danny McAuley: James, do you want me to kick off or ... okay. So all good questions. The pH one is interesting and we didn't know at the outset, so we did a national survey and a pH of 7.2 seemed to be what most clinicians felt comfortable with tolerating before intervening. So that was the sort of level that nationally we came to it, at 7.2. The need for anticoagulation is important at present, because these devices are typically low flow. There is, if anything, a greater need for anticoagulation. Some of the more experienced ECMO centers, who use this technology, talked about using alternatives, such as antiplatelet therapy rather than actually full anticoagulation, but we felt that the device particularly the advice was that there was a need for systemic anticoagulation. And certainly, particularly at the outset, that there was a risk of clotting. So I think that the device, as it currently is, will continue to need anticoagulation.

Danny McAuley: I suspect as the technology improves, we will move away from anticoagulation. I mean, there are devices now that can be integrated into renal replacement therapy and use citrate. So there are additional options coming in terms of the technology. In terms of the SAEs, and I guess Eddy mentioned that the catastrophic concern about intercranial hemorrhage. And certainly, we saw five intercranial hemorrhages that were felt to be related to the device, three of which were fatal, unfortunately. That gives an incidence that's fairly similar to what has been reported in other extracorporeal circuits. It's typical of what we see, but still clearly not something that we would like to see. And that was the major adverse event. We saw other bleeding complications in other body sites. There were some other issues around thrombosis on a couple of patients, as well. So they were the more detailed on the SAEs. Phil, does that cover, do you think?

Phil Gillan: Yes, that's very clear and was dealt with with Eddy, as well. Another couple of questions that have come up, is there a potential role for CO2 removal in other specific subgroups, for instance, traumatic brain injury and ARDS patients?

Danny McAuley: I think that's challenging again, in the setting of the anticoagulation requirement. Perhaps if the technology improves, it might be an option. Certainly, we have had patients with traumatic brain injury requiring ECMO, but that's in the setting where anticoagulation is not used. So I think for traumatic brain injury, it probably isn't where I would typically go. I think it's also important to remember that this was ECMO or ECCO2R applied in a very broad population of patients. And as Eddy really elegantly said, it's not beyond the realms of possibility that if you enrich for a particular type of patient or subset of patient, that there may be benefit. We will leave someone else to do that trial.

Phil Gillan: That's fantastic. I am going to hand you back over to Rob. Thank you.

Rob Mac Sweeney: Phil, thanks very much. So we'll enter our panel discussion now. And we'll start, initially, with getting the thoughts of our panelists. So Liz, maybe we'll start with you. If you could just give us you're initial thoughts on hearing the trial.

Elizabeth Wilcox: Thank you so much, Rob. And congratulations to James and Danny and the other investigators, and a great presentation by all. As you've heard in the comments, as well as yourself, this is a very well-executed study, both very robust in terms of their study protocol, how it was conducted. And then, as Eddie said, in its afterwards analyses. I think that, as Eddie said, it's definitely pushed the field forward a little bit in terms of better understanding indications for these devices.

Elizabeth Wilcox: Danny actually stole one of my questions, which was asking about whether or not Danny collected systemic biomarkers so we would have a little bit more of an underlying kind of physiologic understanding of potentially what the intervention was or was not successful in delivering. I think what's interesting to me, and it will be interesting to see as we move forward in the area of COVID, is that we kind of underestimate how good we are at providing standard of care. And the pronic rates were still really quite low in this study, and these were ventilation expert centers.

Elizabeth Wilcox: And so, not only the device itself, as James alluded to kind of made people apprehensive about proning, but even in the standard of care group, it was still quite low at about 11%. And those results are on par with the [Row 01:06:42] study that was recently published. And so maybe the only benefit, in terms of us being intensivists of COVID will be that will be proning more of our patients, or be more comfortable with proning. But I was wondering if James and Danny he had any thoughts in terms of why are we not seeing creep of these proven mortality beneficial interventions, but we are focusing on these very complex and expensive kind of modes of providing less injurious mechanical ventilation. Those are my thoughts.

Rob Mac Sweeney: James, Danny?

James McNamee: There's no doubt as intensivists a shiny new machine at the end of the bed has some allure, because I think we're all a little bit scientific. I think you are absolutely right about the proning. Interestingly on sites that took on proning, they tend to stop recruiting, because you'll often have proned after the first blood gas and then the numbers look a bit better and it stretches out the time for the second qualifying blood gas. And I think proning is definitely underutilized. But as I said earlier, there's certainly no problem with proning with the device. It's purely our comfort level. And yeah, absolutely, COVID, let's hope we prone a little bit more. It has shown a to have a little bit of a benefit. But thanks for the kind words on the study, as well.

Rob Mac Sweeney: Fantastic. Thanks James. We'll keep moving. Try and get round everybody. Carol, your first thoughts?

Carol Hodgson: Thanks so much, Rob. Thanks for inviting me. Danny and James, congratulations to you and your group. You've done a fabulous job. I have to admit that I had slight palpitations looking at the complex intervention that you've rolled out across the number of centers that you have. You've done an amazing job. I think most people don't understand how difficult that is to perform well and you have done it brilliantly.

Carol Hodgson: I think that Australia has been waiting with bated breath for the results of the REST trial, to be honest. Because of our over 30 ECMO centers, we have really only got two or three that have ever used ECCO2R. It's not something that's used standard of care in Australia. And it's certainly not something that's been used in any of our major ECMO centers. Even our largest ECMO centers tend not to use it. And there's several reasons for that, I think. I mean, it's certainly considered more expensive, but it's also difficult to titrate and difficult to swap to ECMO, or considered in our opinion to be difficult to swap to ECMO if the patient requires it. So I guess my first question to you was, you got 51 centers, which is absolutely brilliant. Did you have a lot of pushback from centers who were worried about difficulty if the patients deteriorated and required ECMO from ECCO2R or was that not really a problem across the board?

James McNamee: Thanks Carol for those words. We wanted ECMO centers bought in right from the start. We didn't want to create the sort of them and us. There's enough culture wars going on. So we wanted to bring them in on the design of the trial. On sites feeling as if ECMO wasn't being excluded, it was actually part of the care that ECMO was allowed on both arms. And it was a really important outcome, I think, to look at, to see does this prevent patients progressing to ECMO, because it is a hugely complex and expensive intervention.

James McNamee: The numbers weren't huge and weren't able to show a difference. ECMO centers were included as sites, as well. And some of them were some of the top recruiters. So I think sites felt comfortable. Maybe they'll tell me differently as the months go by, but I think they felt comfortable to refer. And I have often got a phone call from sites to say, "Look, we've considered this patient, but now they've got sick. We're referring them on ECMO." They would often call us just to see was that okay? And by all means, it's okay, because whatever the patient needs the patient gets. So I don't think that was a huge barrier, to be honest, Carol.

Carol Hodgson: Well, unfortunately, I don't think the results are going to change Australian's opinions for the use of ECCO2R. But the last thing that I just wanted to comment on, which is really quite different to a lot of our [inaudible 01:11:09] trials is they are stopping for futility or feasibility. So most of the times we have stopping rules in our trials, which are really only for harm or for efficacy. And I think it is brilliant that you have the option to stop for futility or feasibility, because certainly in this case I can imagine it would have been difficult to continue. And as Eddy rightly pointed out in his editorial, potentially even a waste of research money to continue to try and recruit patients. But, I guess in Australia we have been a bit concerned about missing a signal, even if it is not in the primary outcome in some of the secondary outcomes. And I just wondered if you wanted to comment on that.

James McNamee: Danny, do you want to maybe-

Rob Mac Sweeney: Dan, you're muted.

Danny McAuley: Oh, am I?

Rob Mac Sweeney: You're back now.

Danny McAuley: All right. So I think it is a really important point and I would say two things. We had a fantastic oversight DMEC committee, and I am sure Marion is hopefully watching. And she chaired it and that made decisions a whole lot easier. We had Taylor Thompson and Neil Ferguson. I think that is the first thing I would say, make sure you choose your DMEC carefully.

Danny McAuley: The second thing, then, is we had really important input from our patient representative. And, again, that is one of the things in the UK that I think is very front and center to all of what we do. Our lay member on the trial steering committee was really helpful in informing and supporting the decision not to continue whenever it was not going to be feasible to do it. And particularly in the setting of the planned interim, which showed the futility, as well. So I think patient input and a carefully chosen oversight committee.

Rob Mac Sweeney: Great. Thanks very much. Art, we'll turn to you. Your reflections on the trial that you have just heard.

James McNamee: Thanks. First of all, congratulations. A very difficult trial to do and very well done. So to me, the key question was why was the study negative? And there are two big possibilities. One is, ECCO2R is a great hypothesis, but because of the complications, et cetera, it's just not correct. That might certainly be one. But the more interesting one might be, is there something about the trial that suggests we might continue investigating ECCO2R.

James McNamee: So I'd like to get at sort of three possibilities. The one is ... and I know this is, in the UK, pragmatic effectiveness trials are sort of what's done a lot. And I wonder about that with non-expert centers. There was a sensitivity analysis that you did, which showed that when you removed the first two intervention patients from each study, it wasn't a significant difference, but the mortality rate was 3.8% absolute decrease. That's half the size of the mortality rate you're looking for. So that's a pretty big effect, 3.8%. So the question is, should effectiveness studies be done before efficacy studies?

James McNamee: I'm actually in belief that you want to do an efficacy study first, rather than effectiveness study. And that, as I said, just to point out, that's a pretty big effect, even though it wasn't significant, of course. The second thing is related to the patients. Included patients with moderately severe hypoxemia. That makes a lot of sense, irrespective of the underlying diagnosis. It does make sense, but these are patients with pretty high mortality rates. But does it make sense with respect to the underlying mechanism of action of ECCO2R. ECCO2R is focused on eliminating CO2, not on improving hypoxemia. So what you want is patients who are going to benefit from getting rid of CO2. Eddie mentioned, they-

James McNamee: Going to benefit from getting rid of CO2. Eddy mentioned the [Gallagher 01:15:04] study, applying precision medicine and the concept there is to enrich patients enrolled in any study, with patients likely to respond. And the two key factors there are, are debt space and compliance. And so, we estimated for a reasonable size study about a thousand patients, you'd have to have a decrease in driving pressure of about five. You had a driving decrease in driving pressure of less than three. And in that, in the Gallagher study, the PF wasn't particularly helpful at identifying these patients. So, to Eddy's point, should a future trial look at this kind of enrichment, which I think would make a lot of sense. And the third point is Eddy talked about dose in terms of days on the device.

James McNamee: I wonder about dose in terms of how much CO2 elimination there is. If you look at the Gallagher analysis, the expected change in driving pressure and therefore mortality depends greatly on the CO2 elimination from the device. And you could see that if you like in your study, you aimed for, to get tidal volume less than three, but you sort of averaged about a tidal volume of about four. So, and that was at the cost of still our PCO2 was about seven millimeters mercury higher, respiratory rate was a little bit higher and pH was lower. So the question, I guess, is, do you think you could get... Another approach is to use devices where you get greater CO2 elimination than you had, and that... So those are the three points, pragmatic trial. It can be problematic, the kind of patients should you have enrichment, with the kind of patients you'd expect to benefit and then the dose of CO2 eliminate.

Danny McAuley: Thanks Art. I mean, all fundamentally important questions. Here in the UK, I look after a funding program that just funds efficacy studies. So I'm completely bought in that efficacy studies should be done first, if possible. I think the challenge that we faced with this and in fact why the funder commissioned was that this was coming into routine practice throughout the UK. So before we started, 30% of sites had used this type of technology and there was a worry that it could become embedded without any evidence. So, that is the rationale for going to effectiveness. That as you rightly say, doesn't preclude doing an efficacy study in a different population. So I think you're absolutely right. The next study should probably be efficacy rather than clinical effectiveness. In terms of the surrogate, I completely agree that enrichment is probably now the way forward given this broad population hasn't worked. I guess the rationale for the PF and obviously a co-author in the extra vent study.

Danny McAuley: So, you'll know the [postdoc 01:18:13], it wasn't really using PF as a measure of... will this make your hypoxia better? It was a measure of these are sick lungs and therefore at risk of [VILI 01:18:25]. So it was really a surrogate, not really expecting... Because it doesn't improve oxygenation. So, that was just an attempt to enrich, as you rightly say, [urine's 01:18:35] work suggest there may be other ways to enrich and I think that work needs to be done to test that hypothesis. And then I think you're absolutely right that the dose, both in terms of getting the tidal volume down further, that Eddy touched on and for longer, I think is an important question. And as Liz mentioned, if we can see a biomarker signal that suggests that we had an impact, that would be useful to help inform that discussion. The problem with the biomarkers is that it was a very small subset of only about 50 patients out of the overall cohort. So it may not be appropriately [inaudible 01:19:16]. James [crosstalk 01:19:18].

James McNamee: Sorry, in turn, you looked at priority versus [inaudible 01:19:21] as that which you get at a bit of the obviously in terms of stiffness of the lung, did you look at compliance as... If you look at patient with low compliance versus higher compliance, but that would be a cheap and dirty way of sort of getting at that.

Danny McAuley: Absolutely. So we didn't look at compliance. We used plateau pressure as our surrogate for compliance and certainly plateau pressure doesn't modify the effect. And I guess the one other point I meant to say is you talked about the change according to removing the sort of first two intervention patients at each site. It's interesting again, and one of the subgroup analysis, we looked at sites who recruited more than 10 versus those who had less than 10, and that didn't seem to make an awful lot of difference. So I'm not as convinced that the learning effect was a major factor. I think the other points around enrichment and dose are really important.

James McNamee: By the way, in terms of the [inaudible 01:20:18] dropping two patients, you may want to also do a comparison where you drop two patients in the control group, because that may have been a time effect rather than a device effect. But...

Danny McAuley: Yeah, that's a reason-

Elizabeth Wilcox: Also, Danny, do you think, given the fact that your guys' educational kind of intervention that you applied at the beginning of the study might have gotten away those center effects of clinical competency? Because it seems like it's more than any other trials have really done in the past. And so maybe you and James could comment on that.

Danny McAuley: So the education package was huge and there was [Along 01:20:57] went out for the first one to two patients and helped the site and stayed with the site for the first. So maybe the third and fourth patients are the ones to take out. We had [Genara 01:21:12], our REST nurse went to sites regularly for recurrent training, also went to help if a site hadn't recruited a patient for a while, there was an e-learning platform, there was an app, but there was a lot of education. So I think you're right. I am less convinced by the learning effect because of that as well. I don't know, James, if you want to add to that.

James McNamee: Certainly just from... Because you would get calls from sites if they're having a problem, if they're halfway through [cannulation 01:21:44], I'd get a phone call. I could hear alarms going on, the background going what do we do? But you definitely find a comfort level and we find it ourselves. We weren't a hugely experienced center before we started. Comfort level after one or two, just to adjust one or two points a bit, Art. Regarding the pragmatic study, [Ronaldo Blom 01:22:05] was right. We got the money first and the trial was based somewhat on the commission call. So they give you the [pickle 01:22:13] and the [inaudible 01:22:13] you have to use. So you sort of have to follow that to get the cash. And that sort of explain some of the design of the trial.

Rob Mac Sweeney: Okay, great. Eddy. I wanted to come back to you. Could you comment just... What we're going to do now is look at some more specifics of the trial. Can you comment particularly about the device and its use and the settings on it? Was that as good as it could be?

Eddy Fan: I'd be happy to, Rob, I'm not sure if you wanted to have Kathy offer any... [crosstalk 01:22:54]

Rob Mac Sweeney: Oh, I have indeed forgotten Kathy, [crosstalk 01:22:57] but in my mind, Sorry.

James McNamee: How is it possible to forget Kathy? That's not possible.

Rob Mac Sweeney: Been waiting and waiting and waiting. Last, but by no means least Kathy, your thoughts on [RESTs 01:23:08]. My apologies.

Kathy Rowan: Okay. Well, clearly a labor of love that probably felt like a labor of desperation all the way through. Great leaders, great team and I think people need to know that in the UK, we have the most amazing research test bed. We have our NIHR, the funder. We have the clinical research network. So this was done insights that would've been given financial resources and support to deliver the study. So I guess, first of all, I really want to congratulate you, the control group mortality that you put in your power calculation, was actually the one seen in the trial, we rarely see that in critical care trials. So congratulations for that. I suppose one of the things that had the professional support, it had the resources support, [inaudible 01:23:57], I'm left thinking, how do we do these trials? How are we going to do these trials and actually reach the sample sizes?

Kathy Rowan: Danny said, this was [ECCOR 01:24:07] in a broad population of patients. And we're now talking about precision medicine and enrichment and that usually lowers the numbers of patients available. So, and I guess I think we really almost want to sort of have a bit of a moratorium where we say, well, how are we going to this? And Eddy, you sort of touched on this in your editorial. And I think we do need to be global. We do need to be collaborative. We are engaging in [Bayesian 01:24:37] adaptive global trials. I think we just need to think, how are we going to do these trials to ensure that we don't end up having to stop a trial for feasibility or utility or the combination of the two. I know that we had an internal pilot in REST and you managed to tick those boxes.

Kathy Rowan: I'm wondering whether reflecting now with the benefit of hindsight, whether you set those targets for traversing into the full trial too low, wasn't where the recruitment targets in that and might that have been a time, I'm not saying that not going on was the right thing to do, but it's just that notion of how do we get these trials to succeed to their full sort of sample size, because we are left with a situation where we're underpowered and asking the questions, I wonder if, I wonder whether et cetera. So huge congratulations, you know that you guys are great leaders of trials and research in the UK. And I can't imagine the trial in better hands, but somehow we didn't manage to pull it off. And I guess it's just about stopping before... I noticed you said Danny, you'd leave the next trials to other people which suggests the labor of desperation and maybe just sharing that learning.

Danny McAuley: Thanks, Kathy. I mean, I maybe come back. So in terms of the control event, well, it was clearly always going to be correct because it was informed by [ECNA 01:26:22] data. So that's why we got that right. So that'll be the first thing. I think your point about internal pilots is really important and given that we're between friends and no one else is listening, the internal pilot, we succeeded beautifully but, I think we probably picked our best potential sites and we set the bar probably a bit lower. And I think if I was doing this again, I would set maybe harsher targets because it is desperate to try and recruit to a study where you're torturing your friends to keep on recruiting. So I think it's an important point that we probably need to think more closely about our internal pilots. I think that's a learning point for me.

Rob Mac Sweeney: Super. Eddy, we'll pick up with that point that I had asked you just about the device itself and the implementation of that.

Eddy Fan: Yeah. I think from the details that are available and again, thanks to James and Danny for sharing the details of this study. It seems that for this device, they want to maximize CO2 removal within the technical capabilities of the device. And so the idea was to maximize CO2 removal, to afford the greatest reduction in tidal volume while managing the consequential hypercapnia and acidosis that came from that. So I think at least from the details that were provided, and I'm not sure that there's anything extra that James or Danny could shed light on, but it seems like they had a very practical approach to the use of the device. The device runs at about half a liter per minute of blood flow if you maximize gas flow on this device to remove as much CO2 as it can.

Eddy Fan: And then go to reduce tidal volume as much as you can within the physiologic target they set, I think that makes a lot of sense. As Art sort of alluded to, there are now other devices available that might have different efficiencies at removing CO2. So you might have some devices that could remove slightly more CO2 and then maybe afford a greater reduction in tidal volume or intensity to mechanical ventilation but certainly within the confines of the study device, as they deployed it, I think it was employed to the best of what it could accomplish for the targets of the trial.

Rob Mac Sweeney: Fantastic. And Carol, moving on to you, as Eddy said, this is a bundled intervention. It's not just a device that comes with other interventions as well, anticoagulation, sedation and the effects they can have on mobility, which would be another component to this. Can you comment on how you try and tease all of those different aspects of a bundle out in a study like this?

Carol Hodgson: Yeah, I think it's really difficult. In my experience and in my opinion, I don't think that you can separate them out and that's part of the problem. And it's why Eddy was very quick to say that there are separate components to this intervention. And as James rightly pointed out, more people received controlled mechanical ventilation and more neuromuscular blockers and that's part and parcel of needing to manage somebody on ECCOR, as is probably increased sedation and reduced mobilization. And there's no two ways around that until we have better technology where we are more mobile and perhaps that's coming, because when you look at how we can move people on hemofilters now, and how we can move people on ventilators now, I certainly think that the tide is changing, but it moves slowly. And we are certainly not there yet with ECCOR or [ECMO 01:30:05].

Rob Mac Sweeney: Sure. And moving on, Art, the populations seem to be the right population. As James has said, they were protectively ventilated, they were sick. The intervention appears to be very well executed in this group. What more could be done to try and show efficacy in this setting?

James McNamee: Well, I think it's the enrichment part that we talked about. Hypoxemia per se, you can imagine, for example, patients with a pulmonary embolism, right. It might be very hypoxemic, they're not necessarily going to benefit much from CO2 removal so it's really focusing on which patients you think are likely to benefit if [inaudible 01:30:45] decreasing ventilator induced lung injury, decrease power, decrease driving pressure. So that's the focus there. And there was a couple comments that, well, it's hard enough to get this many patients. If you use an enrichment strategy, you have fewer patients, that's correct. But you also have increased power with fewer patients. If you need a thousand patients looking at all comers, you might only need 500 patients if you're enrich appropriately. So it's not like you still need the same thousand patients. So you probably end up with the same time, quite frankly, but at least you're focusing on the patients that are most likely to benefit.

Rob Mac Sweeney: Sure. And one thing that struck my mind when you presented the results, James, was the feedback to the centers about tidal volume the whole way through the trial. Kathy, methodologically, how does that play when you're constantly feeding back about an intervention or about a component of the trial, does that introduce a contamination?

Kathy Rowan: I think transparent reporting about what goes on , you can see, I presume it was happening in intervention and control. I think there's a lot of sort of things that could be seen as contamination, most trials have feedback going on all the way through them in terms of... about recruitment of patients, about data as data's amassing and that kind of type stuff. So I think it's part of the interaction, if that makes sense, of how trials are conducted. I think in this study, it was very, very key to the study that they wanted low tidal volume, ventilation to be being done so that the trial wasn't actually sort of seen in the light of a harmed control group, as opposed to a beneficial intervention group, if that makes sense. So it seems to me, as long as you report transparently that that was happening, that it's part of the conduct of this trial.

Rob Mac Sweeney: Sure, sure. And Liz, a final question. There have been comments certainly online about the inflammation that might be induced by the extracorporeal circuit and we limit the tidal volume to try and reduce the inflammation in the lung. Yet we potentially have a second source of inflammation with the extracorporeal circuit and there are maybe biomarker studies coming down the line. How do you think that plays out. You're muted at the moment.

Elizabeth Wilcox: Sorry, Rob. That was bound to happen. So I think maybe in Danny's next study, what you would want to do is potentially a three arm study, as Eddy alluded to, that maybe clinically you can achieve lower tidal volume ventilation in the absence of ECCOR. So that would be one arm and Danny would measure systemic inflammatory mediators there, then with the ECCOR arm, then that would get at your question of whether or not it's the device, if you're able to achieve similar tidal volumes. And then you would compare that to the six mills per kilo of standard care, if that's what you decided to use in terms of what was going to be your mechanical ventilation target. Because as Eddy said, there's probably a lot of physiology at play and it's probably not just the one thing that we're targeting [crosstalk 01:34:08]

James McNamee: I have a feeling that Danny's next study is a zero arm study in this field.

Rob Mac Sweeney: Okay. We're over time once again, there's just too much to talk about. I'll finish with a question quickly to Danny and James, really a joint question. So what does a clinician do? We now have the results of REST in our hands, we have a patient with acute respiratory failure, acute hypoxemic respiratory failure, they meet your criteria, should they use this device at the moment as per the trial?

Danny McAuley: James, do you want to go first?

James McNamee: Yeah. One thing from doing trials is you begin to look at the patient population that you've investigated. And so we have looked at [hypoxic 01:34:55] respiratory failure patients that were at six. We tried to bring them down to four. So, we didn't look at patients that you can't get to six. We didn't look at trying to [extubate 01:35:09] patients with the device. So, specifically look at our hypothesis. So currently if I wanted, if I had a patient and I could get them comfortably onto six mills per kilo, I wouldn't be looking to turn down the tidal volumes any further facilitated by this device. But that's just for that indication.

Danny McAuley: I think that's right, Rob. I would agree with that. So I think in a broad population of patients, in a broad population of intensive care units, if you can get the six mills, I don't think there's any additional benefit at this stage.

Rob Mac Sweeney: Okay. Thanks very much guys. With that, I'll draw this session to a close. That has been absolutely fantastic. Congratulations again to the REST trial team and my thanks to our panel from across the world where it is early morning, late at night, early morning in Canada, late at night in Melbourne and mid afternoon here in Ireland and in London. So thanks very much for that.