Hello, and welcome to today's webcast titled, Novel Clinical Trial Designs. Today's webcast is brought to you by Discovery, the critical care research network at SCCM in collaboration with the clinical pharmacology and pharmacy section. My name is Brooke Barlow. I'm currently a PGY2 critical care pharmacy resident at University of Kentucky Healthcare in Lexington, Kentucky. I'm a member of the SCCM CPP Educational Committee, the SCCM Item Writing Committee, as well as serving as a visual abstract editor for critical care medicine. I will be moderating today's webcast, and I have no disclosures to disclose. This webcast is being recorded. There is no CE associated with this educational program. However, there will be an evaluation sent out at the conclusion of the webcast. Your opinions and feedback are important to us as we continue to plan and develop future educational offerings. We would greatly appreciate if you were to take the 5 to 10 minutes to complete the evaluation. A few housekeeping items before we get started. There will be a Q&A session at the conclusion of this presentation. To submit any questions for the speaker throughout the presentation, please type into the question box located on your control panel. And now I'm delighted to introduce our speaker for today, Dr. David Wong. Dr. Wong is a professor of critical care and emergency medicine, clinical and translational science, and director of the Multidisciplinary Acute Care Research Organization, or MACRO, at the University of Pittsburgh. He also serves as a co-PI of the Remap COVID-USA trial. And now I'll turn things over to our presenter, Dr. Wong. Thank you again for joining us today, and please take it away. Great. Thank you, Brooke. Okay. So it's my pleasure today to talk to you about novel clinical trial designs. These are my and the trials that I'm going to talk about, disclosures. So the next approximately 15 minutes, I'm going to talk about the background behind novel trial designs, delve deeper into two specific examples, and then end with some concluding thoughts. So for decades, every time there's a new intervention, a new potential novel treatment, there's this fundamental tension between learning versus doing. So on the left, in the clinical research world, there's this intense desire to randomize patients out of desire to generate future knowledge for future patients, while on the right, in clinical practice, there's often an intense desire to just give the novel treatment out of this very powerful desire to treat patients, particularly those patients you are personally responsible for. There's many reasons for this, for the separation between clinical research and clinical practice. The Belmont Report, first of all, fundamental text and research ethics actually recommends, stipulates even, that the two worlds be kept apart, and the two worlds have different duties and incentives. But having this type of parallel world structure of learning versus doing is really highly inefficient. If you think about it, every day, there's thousands of folks writing down the exact same data, one in the medical record, and one in the case report form, it's just one example of the inefficiencies. And this inefficiency has been basically tolerated for decades, but during a pandemic, and it's really rather unacceptable. So what if you could learn while doing? So for example, on the left, clinical researchers can make trials friendlier, they can have smoother logistics that don't get in the way of bedside care, and we can design trials that provide a higher probability that your patient will get a novel treatment and won't just be in the control. And then if they get a novel treatment, then provide a higher probability that it actually works. And then trialists can unify. Did the world need, last year, 115 or whatever the number was, individual clinical trials of hydroxychloroquine? Or would it have been better to have just a couple very rigorous, very large trials? Well, on the right, the clinical enterprise can do things such as leverage the electronical medical record. And then individual practitioners can reflect on equipoise, that really, there really isn't that much evidence to say, yes, we should give this treatment now. And we can explain this somewhat complex concepts to our patients. And in the middle, institutions can do things such as reducing the barriers between these two worlds, and then enable and incentivize them to cooperate. So REMAP is just one strategy to enact learning while doing. And it stands for Randomized Embedded Multi-Factorial Adaptive Platform. So I'll go through each letter. Randomized, we all know what that is. It's the most powerful way to generate causal inference by controlling for both measured and unmeasured confounders. Embedded refers to both a philosophy and an operational strategy to embed research processes into routine patient care processes. Multi-factorial just means you're testing multiple therapies at the same time. And it has the pleasant side effect that most patients will actually receive a novel treatment as the number of tested interventions grows. And then the A and P are perhaps the least familiar of this acronym. So I'll go over them in more detail. So adaptive. So a traditional trial, sometimes called a frequencies trial, you start off with making assumptions. What is the event rate in the control arm, for example? And generally starts with a fixed randomization ratio, most classically one-to-one. And you pick a sample size, an N, and then you enroll until you hit that N. And then in the adaptive trial, sometimes called Bayesian trials, you again start off with making assumptions. But then you use what's called response adaptive randomization, which I'll go over on the next slide or two. And then you enroll until you're 99% sure that the intervention works or not. So what RAR is, response adaptive randomization, is that every time a patient is randomized, you basically flip a coin. Patient gets randomized to arm A or B. Data is generated from that patient and their care. That data is then fed back into the model. And then that model then influences the randomization rule such that the next time the coin is flipped, it's a weighted coin with odds weighted towards the best treatment. And the RAR can also handle adding new arms as new ideas to treat the condition arise. It can also handle dropping arms as a given intervention is proven effective or not. You can also have different weights for different patient groups if you have this hypothesis that the intervention may be more beneficial in certain patient groups as well. So an adaptive trial is best described not as play the winner, but as you're probably playing what's probably the winner every time the randomization occurs. So in other words, patients are preferentially assigned to the best performing arm at that time, at the time of randomization. So in other words, not performing so well that you already declare victory for that intervention and recommend that the intervention be given to everybody. But at that time, which arm is looking the best? So P is for platform. So a traditional trial generally studies one patient population, tests one novel treatment, and you conduct one trial. And then once the trial is over, everything just goes away. So that would be like if every time you played a soccer game, you built a stadium, and then when the game was over, you tore it down. What if you could just leave the soccer game up and play game after game after game, or in this case, trial after trial after trial? And that's what a platform trial is. It's a type of adaptive trial, and it tests multiple treatments, and it's perpetual. It lasts forever, as long as there are ideas on how to fight the disease or condition, and as long as the disease or condition exists. So again, there's nothing special about any of the individual letters of REMAP. It's just that REMAP nicely packages it into one package. I sometimes compare it to CPR, and that Peter Saffer really did his research on the A and the B, but it was another scientist who worked on the C. But it was Saffer who put it together as a package. So if we looked at, for example, the PROACT-procastatonin trial or the ROSE neuromuscular blockade trial, those are basically just R-trials, maybe a touch of E for the procastatonin trial. Look at the PROCESS-sepsis and the FACT-ARDS trials, those are RM trials. They're randomized, and they tested multiple interventions, but that's it. And several years ago, the VA conducted a pretty cool point-of-care insulin trial, and this was an REA trial, basically. It was pretty cool. It was basically when a physician or ABP, advanced practice provider, went to enter insulin sliding scale in the electronic record, they would be prompted. Basically, it would say, hey, it looks like you're trying to order a sliding scale. Would you be willing to consider your patient for a trial comparing alternative insulin sliding scale regimens? And then if yes, then the coordinators would be notified, and the game would begin. And this was really cool because it was a great example, because it was one of the first pioneers of an embedded trial. And then I-SPY-2, which is sort of the grandparent of all adaptive platform trials, and this is a large breast cancer trial designed to basically do a continuous phase two trials of multiple candidate cancer-fighting agents to see which ones are worthy of investing millions of dollars into a large phase three. And I-SPY-2 is an RMAP trial. Because everything RMAP does except the E. That brings us to RMAP-CAP-COVID. So RMAP-CAP is a global adaptive platform trial founded in 2016 for severe pneumonia admitted to intensive care, and was launched in those countries with federal funding. It's structured as an international trial steering committee, and then with multiple committees for different functions. In the very center is the core protocol, and it sets out the overall structure and processes. Extremely, extremely small list of eligibility criteria, basically says you're an adult, you have pneumonia, and you're not going to die tomorrow. And then the most relevance for this past 18 months is this pandemic appendix. And what this is that the founders of RMAP-CAP, a priori said, you know, just in case a pandemic breaks out, we should have a pandemic appendix ready to go, such that we can just immediately activate it. And it's just an appendix that attaches to the core protocol. Similarly to how, if you look at the purple ovals on the right, every intervention is simply an appendix that gets attached to the core protocol. And by keeping the core protocol constant, and then just activating different appendices, this allows a modular flexibility. The reason why this pandemic appendix is important is because if you look at the history of H1N1, what happened was that when the pandemic started, by the time studies were designed, ethics applications were submitted and approved, contracts, funding, et cetera, by the time the study was open, the pandemic was almost over, and very few H1N1 patients actually got enrolled in the study. So therefore, in February 2020, RMAP-CAP activated the pandemic appendix, and essentially became RMAP-COVID. And this did a number of things. It added a moderate state, because RMAP-CAP was founded as a pure ICU trial. This moderate state allowed enrollment of hospitalized patients, but who are not yet receiving ICU-level care. Also added specific COVID-19 domains, and then importantly enabled new sites to join the network. So UPMC Health System, where I work at, launched UPMC RMAP-COVID as the first U.S. site. Sponsors the Global Coalition for Adaptive Research, GCAR, which is also the sponsor for ICE-V2. And we launched with those IRBs, and with all UPMC hospitals on the Cerner EMR, because most of our 40 hospitals use Cerner. And we also serve as the U.S. Regional Coordinating Center. So you may be thinking, as I did, the first time I heard of the concept of RMAP several years ago from Dr. Derek Angus, that's terrific, but what about a whole host of practical factors? Commitments, overlay with other clinical research programs, consent, a huge list. So I won't pretend that at UPMC we've solved it all, but over the next 10 minutes or so, I'm going to share with you how we went about implementing this trial in our health system. And we published our experience early this year in the journal Trials. So I'd be lying to you if I didn't admit that it was very much building the plane while flying it back in March 2020, but as we took a breather to write the paper, we realized that implementation really fell across these six domains, which I'm going to run through. So leadership engagement, we did have some advantages. There actually was a prior UPMC RMAP trial of something completely different, metformin to improve postoperative outcomes in frail patients, but the concept of RMAP was not new. We also were lucky to have an existing PIT, Clinical and Translational Institute, which are NIH hubs designed to try and provide some centralization across the campus. We did have some disadvantages, though. The number of proposed trials and opinions at a shop like ours, at least back in March, really significantly exceeded the number of patients that we had back in March. And then we had multiple discussions with both administrative and direct patient care leaders as well as key groups such as informatics, the blood bank, back when convalescent plasma was a hot topic of investigation, and many others. And this allowed unity as well as access to key infrastructure resources, such that when myself or Dr. Brian McVeary, Dr. Chris Seymour, or any of the implementation leaders here asked folks at UPMC or PIT to help, it wasn't just us asking, it was on behalf of the institutions. So UPMC committed to remap COVID as a priority trial, and the University of CTSI centralized COVID-19 research. As for trial embedment, this was really essential. Again, an operational and philosophical commitment to embed the trial into routine care processes. So we engage system-wide administrative groups, as well as information vehicles such as online newsletters, and then accessing key, even something as simple as key outlook groups where we had the leaders of pharmacy, respiratory therapy, whatever, communicate the same message to their cohorts. And then the P&T committee really made a strong stand and said that remap COVID will be the only route for novel treatment, at least in patients that were trial eligible. So in other words, they took a firm stand and said physicians would not be allowed, for example, to just prescribe hydroxychloroquine, but they would have to go through the trial. And as you know, in many places, this type of call engendered strong feelings, strong negative feelings. And then the clinical pharmacies committed to dispense drugs exempt from IND, and actually, as we'll discuss later on, evolved to also manage, in some circumstances, INDs. Obviously, telemedicine had an explosion last year, and we did everything we could to leverage for remap COVID new telemedicine resources. So the vision for trial embedment in the EHR was a one-stop shop. So there's a COVID management portal that when a practitioner clicks on it, brings up a whole host of stuff for daily care, power plant orders, infinite resources, links to up-to-date, but then crucially for the trial, a launchpad into the remap COVID trial. This launchpad was an intake form, and it basically prompted the physician to add in basic symptoms and comorbidities to help feed public health reporting mandates. And then for the trial itself, we just asked the physician, or actually any member of the treating team, to ask the patient or their family, there are no known treatments, remember this was back in March, would the patient like to hear about potential additional therapies? And if they said yes, then this gave us the permission to reach out to the patient. And this is a bit complicated, but this was essentially our remote consent process. And for obvious reasons, it had to be remote, at least in the beginning. So after the intake form was completed, this would generate an automatic alert to research staff who would then review the patient's eligibility, sometimes with the help of a study physician. Staff would then call the patient and then briefly assess interest. And if they were interested, would switch to a video call, often with the help of the bedside nurse. And then if the patient agreed to take part and or had questions for the physician, then a study physician would be zoomed in as well. And then if the patient wished to proceed, then using a separate electronic platform would sign an electronic consent form, as would the physician. The COC, the coordinator, would then activate and complete an enrollment form, which would activate the study protocol, and then most importantly, provide pop-up alerts for the treating physician, him or herself, to sign the order as part of routine care. So after enrollment, response-adapted randomization was used with pop-up order sets, and then for the most part, automated data capture for primary outcome demographics, et cetera. And then the regulatory compliance and oversight, that study launched, we launched with the University of Pittsburgh IRB, and focused on repurposed FDA-approved drugs under an IND exemption, simply because this was the fastest and smoothest way to launch. As we expanded, we switched to WERB, in close partnership with GCAR, and then also switched to add in experimental drugs for immunomodulation, erythroin, and apremilas, for which an IND is required, and therefore, which requires a significant amount of additional administrative oversight. And then, at least for somebody who grew up with traditional trials, we mapped COVID as an adaptive novel trial, required quite a bit of modification versus traditional trial management. So there's a lot of adapting with adaptive trials. As domains drop or add, you're not quite launching a brand new trial every single time, but almost. It's a new intervention, with new eligibility criteria, et cetera. Data is tremendously important. Dr. Chris Horvath took the lead here, and did a huge amount of work, not just with collection, but between different platforms, and then how to export the data safely and securely, and in the right format to analytical teams. And monitoring, for obvious reasons, had to be remote to UPMC hospitals that I've never physically been in. And then adverse events also had to be collected in a different way, semi-automated through the EHR, as well as some manual monitoring and clinical feedback. And then alignment, and apologies for the typo here, alignment with other COVID-19 studies. So importantly, UPMC-specific studies, meaning studies only being conducted, sorry, not only, but studies not being conducted in the overall RemapCAP international platform, but just at UPMC, per se, can be added on top, but can use the same intake COVID-19 portal and remap infrastructure, but tailored to the specific PI's interests. And then ideally can share some of the extracted data elements. So basically, as we implemented in UPMC, RemapCOVID provided the core infrastructure, and then on the left are the different domains for RemapCOVID, while on the right are what we call mini-domains for different studies, such as when the site PI was participating in a large NIH trial, smaller phase two, observational studies, et cetera. So as of last week, about 400 patients have been enrolled across the system. And what we're particularly pleased about is that this enrollment happens, it happened and is happening across the system, particularly in non-academic sites, which is really, really quite good, we thought, because at least in the past, pretty much all clinical trials happened only, at least for acute care, in the ivory tower shops, which is not good for many reasons. Currently enrolling in three domains, anti-platelets in conjunction with ACTIV-4 and the ATT&CK, vitamin C in conjunction with the Canadian LUV-IT trial, and immunomodulation with industry sponsors. We're definitely looking to expand. So if your site you think would be interested, please feel free to contact me. We're deeply in the process of expanding the number of US sites, and notably, with an eye towards the future. We all hope that in the next several months, COVID will just become another diagnosis and not the center of our existence. But we definitely wish to continue the network and flip over to studying non-COVID conditions. Also going to expand the number of domains that we test here. And next will be the ACE domain and anti-coagulation 2.0 to follow up the first round of anti-coagulation studies. Impact, I think most notably are these three. ReMAP-CAP helped to cement along with, of course, the massive recovery trial that corticosteroids are become standard of care now for COVID-19. As well as demonstrating that an IL-6 agent, tocilizumab is also highly efficacious. And then the results for the first anti-coagulation studies, that was a ReMAP-CAP active for an ATT&CK collaboration. Those results are under review at a large journal, and the results have been released publicly in MedRx. So there's definitely been challenges, which I'm happy to share. And I think these experienced adaptive protocol trialists published a series of papers in 2019 in trials. And I thought there was quite clever, changing platforms without stopping the train. Yes, there's quite a bit of challenges with data, data management for the first paper. And then second, just the general operational aspects. And this is really stuff like that. Yes, every time you start a new domain, it's incredibly cool, but you have to start it very fast. And unlike with a traditional trial, where generally the middle couple of years of enrollment are sort of more or less on a cruise control. Here, it's quite different, and even small things like you almost don't even have a chance to celebrate because you're constantly launching new domains, which is the whole purpose, but definitely provides challenges. I'd be remiss if I didn't address the elephant in the room of COVID fatigue. Obviously, patient care folks are fatigued, but so too are those in research. And most importantly, the patients. I think many of us outside and inside of EMAP-CAP have noticed that patients seem less interested in participating in COVID trials than they were a year ago. Then obviously other diseases need lots of attention too. As for embedment, there are other key health system priorities. Not only at UMC, but everywhere in the world. EMR bill freezes, kind of put a freeze on asking for additional EMR asks. And then I think we're all inspired to move towards a UK recovery style consent by the treating practitioners themselves. That'll be much more, that'll obviously be more efficient. But there are multiple barriers to that in the United States, not the least of which is a lot of folks in prior practice or who don't live and breathe research. You really don't wanna spend five, six hours doing city modules. And of course, when everything's urgent, well, I think you know the expression. And then in terms of design, interpretation is really an important thing. Bayesian stats are less familiar to both physicians and regulatory bodies. So you have to be very careful in describing the design and the output. And then as I alluded to, internationally, there is a definite goal of flipping back to remap CAP, community-acquired pneumonia. The question is when and exactly how. And I think it'd be very interesting to think through keeping certain aspects of remap COVID, such as, so remap COVID's primary outcome is 21-day organ support free days, or remap CAPs is 90 days for death alone. So there's gonna be a lot of thinking about what is the right time point and should we stick with composite versus pure death? And then should and can remap CAP continue to enroll non-ICU patients? So that's remap CAP slash COVID. And then for the next 15 minutes, I'd like to describe a new trial that we're conducting only at UPMC, launched just two months ago, called Optimize C19. So optimizing treatment and impact of monoclonal antibodies through evaluation for COVID-19. At the bottom left, you can see the trials.gov number, and the methods paper was just published online, actually, this morning. So this is similar in philosophy to remap, but also somewhat different. So briefly, monoclonal antibodies, they provide passive immunity and they reduce viral burden and improve outcomes. There are multiple monoclonal antibodies, I'm gonna call them MABs for COVID-19, have been and are available under FDA EUA. And these EUAs, emergency use authorizations, they specify in great detail how the MAB should be given. Eligibility criteria, dosing, consent, monitoring, et cetera. However, there's been limited uptake of MABs for a variety of reasons. You have to go to an infusion center and then you have to hang out and get this infused over an hour. And then how is the patient supposed to get from home? Who's gonna, to this infusion center, who's gonna take them? They obviously can't, or shouldn't, just hire a taxi. And then there's also unknown comparative effectiveness. All the trials have been done MAB versus no MAB. They've never been compared head to head. And then recently, the National Academies of Sciences, Engineering, and Medicine said it's critical to collect the data and evaluate whether they are working as predicted, which we completely agree with. So we set up Optimize C19 as a so-called inside the EUA study. So the objective is simple, to evaluate the comparative effectiveness of COVID-19 MABs with EUA approval. And it is an open-label, pragmatic platform trial with a spot adapter randomization and EUA-eligible patients ordered MAB by a physician. So only the MAB allocation is randomized. In other words, which MAB type they get is random. Everything else, all their care, including the treatment consent, dose monitoring, all that is as per the EUA. And variant testing and 28-day follow-up is just part of routine care. Primary endpoint is a pragmatic one, used in other trials, hospital-free days by day 28. So on the operation side, there was just a massive rapid ramp up at the health systems level. Done in partnership with the Federal COVID Response Team, formerly called Operation Warp Speed, and this tremendous increase in infusion center capacity, as well as expansion to emergency departments. I'm sorry, I should have mentioned that the medical antibody EUA essentially says that you have to be, you have to have mild to moderate you have to be, you have to have mild to moderate illness defined as you are not yet hospitalized for COVID-19. So essentially outpatients. And then there's a massive lift by UPMC clinical analytics, pharmacy supply chain, underserved communities, advisory board, et cetera. There's also extensive outreach to patients, to physicians, and then lots of collaboration with local and state public health offices, as well as other healthcare systems. So the whole design is built on this somewhat unique regulatory structure, in that the trial is approved by both the UPMC QI Committee and the University of Pittsburgh IRB. And the design centers on the fact that prior to Optimize Existing, the initial just patient care rollout of MABs back in the fall of 2020, UPMC offered MABs under what's called therapeutic interchange, meaning that, and the way it works is that a physician would order a MAB as a generic referral order, and the pharmacy would fill it under a therapeutic interchange, meaning that the pharmacy would supply whichever MAB happened to be available. And the MAB availability is overseen by the pharmacy, by the P&T Committee. So then patients then would provide verbal consent for the MAB treatment, and then as per the EUA instructions, physicians would provide and review the EUA fact sheets, and under UPMC policy, because of the therapeutic interchange policy, they would in fact review the EUA fact sheets for all the available MABs, because it's possible that the patient could get one of three, sorry, well, one of whichever was available at that time. So Optimize C19, which launched in March of this year, it simply provides the therapeutic interchange via a formal random allocation. So before it was random, and now it is randomized. Everything else is the same. And as many of you probably know, there've been a lot of EUA changes. So from November 2020 to February 2021, EUAs were issued, and I'm just gonna call them by their first letter, issued for B, B plus E, and C plus I. On March 24th, the US government halted the distribution of B alone, followed a few weeks later with just complete revocation of the EUA for B alone. And then just last, just two weeks ago, the EUA criteria were quite largely expanded, such that more patients could qualify. So every time the EUA changed policies, so too did UPMC P&T policies, and then Optimize C19 simply followed along. As again, Optimize C19, the only thing that is different is changing a random allocation to a randomized allocation in order to do a comparative effectiveness trial. So as of a few days ago, there've been more than 1,300 randomized allocations with mostly automated data collection by the EMR, and variant testing is ongoing. So I think this is my final slide. So launching Optimize C19, and actually it was 21 days, was very much like building the plane while flying it all over again. Similar challenges as with RemapCAP COVID, or really as with any adaptive trial conducted anywhere for any condition, what has to be done fast. So we're definitely still learning while doing, that's sort of our team's inside joke all the time. And we were very much inspired by the UK, not only recovery, and how impressively it has done these huge patient-centered trials with treating team consent and processes. It's like the ultimate embedment, but also the national health system. So what they've done over the last, I think just only about six, seven years ago, was that NHS, every year or so, they identify NHS priority trials, and then they actually incentivize and promote those trials across the entire country, and say, okay, if a hospital in Liverpool or whatever, actively engages and enrolls into this priority trial, then they qualify for certain incentives. So you can imagine at some point in the future, CMS, for example, doing something similar. There was like a national priority trial or trials, something similar happening. Also inspired by the many EUA changes in the monoclonal antibodies, have to very carefully draw the consort, create and operationalize the Statistical Analysis Plan, or SAP, and be quite careful with the analyses. The vision, though, I think remains incredibly compelling for all of us, which is simple. It's bigger, faster, and stronger trials to yield faster answers for our patients. And I think, yes, that's it. So I'm happy to take questions. Thank you, Dr. Wang, for that excellent presentation. We do have one question for you, and I just wanna encourage the attendees to include any questions you may have for our speaker today into the question box. So how are the permutations of inclusion to multiple or varying interventions handled from a statistical perspective? Sure, so it's Scott Berry and his colleagues at Berry Consultants, who are the masters and creators of the model. My simple non-statistician, I'll give you my simple non-statistician understanding. So probably the biggest variable with any domain is to consider upfront, do you have reason to believe that there is an interaction? So for example, corticosteroids and immunomodulator. So if you do a prior declare that it's plausible that there's an interaction, then that can be built into the model. And appropriately, there is a penalty, so to speak, in terms of power for that interaction. If you don't stipulate the a priori, then the interactions can be queried post hoc, but obviously that's not quite as strong. But it's a very complex novel model that factors in a large list of variables, including time, since obviously care hopefully improves over time. So I believe the analysis is in every four-week epoch, along with other variables such as age, gender, et cetera. He noted that answer to his question, so thank you very much. Another question came through, and asking if you could discuss the ethical concerns, particularly regarding an adaptive trial design. The ethical concerns, can you elaborate on that a bit? Introducing bias into the potential study with randomizing to specific treatments in an adaptive trial design, particularly in patients who may not benefit from a certain arm. This may not be answering your question, but I think one thing that has come up in terms of the practical part of consenting is that when you have six, seven domains that the patient is eligible for, there's certainly an ethical concern of how can you properly get informed consent for six different domains? You basically have to describe six different novel interventions to a patient who's ill, and that can be challenging. Is that what you were referring to in terms of ethics? Yeah, I assume most, you know, particularly, you know, you mentioned that in patients who may not benefit from a certain arm, you know, how are patients, I guess, thoroughly educated and what ethical concerns exist, you know, that you guys have had to overcome, if any? Well, you don't, I mean, there's equipoise for each of the domains, so there's no way of knowing if a patient would benefit from that arm or the other. Thank you for answering that question. I'll wait to see if any other questions come through, but, you know, in, yeah, she said that didn't answer the question, so thank you for that. So just out of curiosity, you know, what sorts of, I know you mentioned some of the pitfalls in terms of interpreting from adaptive trial designs, but, you know, what type of limitations or potential pitfalls do you see in interpreting them, and then how do you go about that, I guess, in reporting the results? Sure, so if you look at the published papers, the primary readout is an adjusted odds ratio, and I think a lot of folks appropriately ask, what does an adjusted odds ratio of, say, 1.4 for organ support-free days by day 21, like, what does that translate to in terms of, like, less death and or less time on organ support? But it's not a clean, like, you know, just arithmetic translation between that odds ratio and X reduced percentage mortality or X number of days less on organ support. But nonetheless, I think a lot of us are starting to think that we should at least give an estimate of that, even though that's not how the model is primarily powered for. So that's one aspect. But I think the big thing is that, you know, I certainly grew up reading traditional trials with, you know, p-values and nice, hard, simple endpoint, like, 20-day mortality. Now we're talking about a novel combined endpoint using novel stats. And, you know, research doesn't matter if it doesn't change practice. So it's definitely an issue that has come to fore. And also research doesn't matter if FDA or EMA doesn't like your protocol. FDA, as many of you may know, just, I think, last week issued its final guidance for master platform trials for COVID-19, which I think is at least helpful to set a benchmark where basically FDA pretty much embraced the concept and just said, if you're going to do one, we want you to do it in these certain ways. But sometimes there's, you know, FDA is not, like, completely homogenous. Sometimes FDA at the top will say X. But then in your individual protocol talks with your FDA staff person, the operationalization of that vision may differ and can get quite tricky at the fine print level. Absolutely. Thank you for your detailed response to that question. So just one last question for you for the podcast today would be, what sorts of, any reason as to why an adaptive trial design should not be used as a clinical trial design in your mind? That's a nice question. That's an interesting question. Well, adaptive trials are best used when you really have multiple treatments to test. So if all you want to ask is one question and you're happy with that, like, if you want to do a traditional trial, one patient population, one question, one treatment, then yes, it would be much faster to do a traditional trial. Like in the FDA's guidance, I think on the first page, it basically says that overall, an adaptive trial testing multiple therapies will be more efficient. But it requires a lot more prep up front. You know, so like RemapCap spent in its founding, a lot of time on that, quote unquote, simple core protocol. But that was a huge amount of investment up front versus just doing, say, antibiotic A versus antibiotic B. So I think if you're happy with just testing one question, which most of us have been quite happy doing for many years, then that's the way to go. But I think clearly the future will be in adaptive trials because I think certainly the world, the country and the world has seen that having a jillion one question at a time trials is like really bad during a pandemic. And it is really bad, just not quite as obvious during non-pandemic times. Absolutely. Well, thank you again for your presentation today. And that will conclude our Q&A session. Thank you again, Dr. Wang. Thank you to our presenter and to the audience for attending today. Again, you will receive a follow up email with a link to complete an evaluation. Just as a reminder, there is no CE with this educational program. However, your opinions and feedback are important to us as we plan and develop future educational offerings. We would greatly appreciate your time in completing the brief five to 10 minute evaluation. That concludes our presentation today. Thank you again for attending. Thank you. Bye-bye.

Dr. David Wong

critical care

emergency medicine

University of Pittsburgh

novel clinical trial designs

REMAP

randomized embedded multi-factorial adaptive platform

clinical research

clinical practice