WEBVTT NOTE This file was generated by Descript 00:00:00.029 --> 00:00:03.279 Judith: Welcome to Berry's In the Interim podcast, where we explore the 00:00:03.279 --> 00:00:07.289 cutting edge of innovative clinical trial design for the pharmaceutical and 00:00:07.289 --> 00:00:09.560 medical industries, and so much more. 00:00:10.180 --> 00:00:10.859 Let's dive in. 00:00:12.155 --> 00:00:12.725 Scott Berry: All right. 00:00:12.725 --> 00:00:13.925 Welcome everybody. 00:00:13.925 --> 00:00:21.395 Back to in the interim, I'm Scott Berry and I'm joined today by my common 00:00:21.395 --> 00:00:23.795 partner in crime here, Kurt Veley. 00:00:24.155 --> 00:00:24.575 Dr. 00:00:24.575 --> 00:00:27.695 Kurt Veley, welcome back to In the Interim. 00:00:28.105 --> 00:00:28.620 Kert Viele: Thanks, Scott. 00:00:29.615 --> 00:00:32.945 Scott Berry: Boy, you're showing up enough here on the interim, you probably 00:00:32.945 --> 00:00:34.835 don't have a whole lot of alpha left. 00:00:35.435 --> 00:00:36.575 Kert Viele: I don't, and I'm old too. 00:00:36.605 --> 00:00:38.015 I've done a lot of trials. 00:00:38.225 --> 00:00:38.540 Who knows? 00:00:39.455 --> 00:00:42.305 That's why we gotta hire new people more alpha. 00:00:42.790 --> 00:00:43.080 Scott Berry: yeah. 00:00:43.675 --> 00:00:44.095 Okay. 00:00:44.095 --> 00:00:47.970 That's not our topic for today, but that is, people get a sense of what 00:00:47.970 --> 00:00:49.620 our sense of humor is around here. 00:00:49.635 --> 00:00:56.265 in the interim, our topic today may sound dry, but it from its guidance 00:00:56.265 --> 00:01:04.305 documents, and it's specifically the ICH E20 draft guidance on adaptive designs. 00:01:05.650 --> 00:01:09.010 so my wife Tammy, listens to all of these. 00:01:09.260 --> 00:01:11.390 she's the one that listens to all of these. 00:01:11.600 --> 00:01:15.140 And so we, we have people like Tammy, who's brilliant. 00:01:15.190 --> 00:01:17.350 but she doesn't live this every day. 00:01:17.500 --> 00:01:24.730 So what is the ICH E20 the International Council for Harmonization. 00:01:24.870 --> 00:01:31.385 That's the ICH of technical requirements for pharmaceuticals, for human use. 00:01:32.325 --> 00:01:34.155 Kert Viele: That did make it really exciting right there, 00:01:34.295 --> 00:01:34.685 Scott Berry: yeah. 00:01:34.805 --> 00:01:35.495 Yes it did. 00:01:35.855 --> 00:01:36.575 Yes it did. 00:01:37.115 --> 00:01:39.860 So and the key word of this, I think is harmonization. 00:01:41.065 --> 00:01:47.725 And it talks about its mission statement, is bringing together regulatory 00:01:47.725 --> 00:01:53.845 authorities and pharmaceutical industry to discuss scientifical, scientific and 00:01:53.845 --> 00:01:56.965 technical aspects of drug registration. 00:01:58.390 --> 00:02:05.080 And this is really, can they come to a harmonization a, an agreement 00:02:05.350 --> 00:02:10.720 on guidelines for clinical trials interacting with regulators. 00:02:11.110 --> 00:02:15.480 And an important thing to think about this is harmonization across 00:02:15.480 --> 00:02:17.400 regulatory agencies, global. 00:02:18.090 --> 00:02:23.220 EMA, PMDA FDA global countries. 00:02:23.220 --> 00:02:27.840 And so for something to show up here, it's largely agreed upon 00:02:27.870 --> 00:02:30.600 globally about these guidances. 00:02:32.200 --> 00:02:37.390 what to add about ICH, the importance of this to some extent. 00:02:39.990 --> 00:02:41.160 Kert Viele: So I, you know, I think the. 00:02:41.340 --> 00:02:48.030 The important part here is that this is used for potential sponsors, 00:02:48.030 --> 00:02:49.860 people who are designing trials. 00:02:50.190 --> 00:02:55.720 This is a, essentially a document which is saying, this is, okay, 00:02:55.750 --> 00:02:57.670 this is mildly controversial. 00:02:57.670 --> 00:02:59.440 This may be more controversial. 00:02:59.750 --> 00:03:03.105 nothing in here says you can or can't do anything. 00:03:03.505 --> 00:03:04.495 but it's certainly giving. 00:03:04.495 --> 00:03:06.240 Here's the kind of issues we worry about. 00:03:06.915 --> 00:03:10.155 It provides a structure for how to talk to regulators. 00:03:10.395 --> 00:03:11.610 This is essentially it. 00:03:11.615 --> 00:03:15.555 It sets up the playing field and essentially says, these are the kind 00:03:15.555 --> 00:03:17.355 of things that are always in bounds. 00:03:17.355 --> 00:03:20.295 These are the things that are, you're gonna need more discussion. 00:03:20.565 --> 00:03:25.005 All of this is vital in when people first walk into the, I know I 00:03:25.005 --> 00:03:26.925 have a trial, what should I do? 00:03:27.165 --> 00:03:30.235 This kind of documents where it says, you know, here's my options. 00:03:30.265 --> 00:03:32.095 This is the set, the field I can play in. 00:03:33.975 --> 00:03:34.545 Scott Berry: Okay. 00:03:34.725 --> 00:03:38.990 And it says, harmonization is achieved through the development of 00:03:38.990 --> 00:03:44.030 ICH guidelines via our process of scientific consensus with regulatory and 00:03:44.030 --> 00:03:47.000 industry experts working side by side. 00:03:47.420 --> 00:03:48.800 So in, where is this? 00:03:48.800 --> 00:03:49.490 In the process? 00:03:49.490 --> 00:03:53.510 They wanted to put out a guidance document, an ICH guidance document 00:03:53.510 --> 00:03:56.270 on adaptive designs, something. 00:03:56.915 --> 00:04:01.145 Barry, you and I have been living in this space of adaptive designs. 00:04:01.145 --> 00:04:03.035 Now the FDA has a guidance. 00:04:03.405 --> 00:04:07.545 regulatory agencies may have guidance, and now this is an international 00:04:07.545 --> 00:04:12.955 harmonization of guidance, in the sense this is draft and they're looking for. 00:04:13.520 --> 00:04:17.000 Feedback from the industry on their draft guidance. 00:04:17.420 --> 00:04:22.700 So today we are going to dive into what this guidance, this draft guidance 00:04:22.700 --> 00:04:27.950 says, the opinion from people who live the life of adaptive designs. 00:04:27.950 --> 00:04:33.140 This is what we do and what our viewpoints on this draft guidances. 00:04:33.890 --> 00:04:35.480 So that's the topic today. 00:04:37.060 --> 00:04:41.740 So Kert what's your overall impression when you read this? 00:04:41.790 --> 00:04:44.820 and by, by the way, people can go out and get access to this. 00:04:44.820 --> 00:04:45.600 You can read this. 00:04:45.600 --> 00:04:49.950 You should, if you're listening to this podcast and you're not, Tammy, 00:04:50.190 --> 00:04:52.260 you should go read this draft guidance. 00:04:52.590 --> 00:04:54.180 Kert Viele: You should make Tammy read the guidance. 00:04:54.180 --> 00:04:55.290 That should be a requirement. 00:04:56.205 --> 00:04:56.685 Scott Berry: Yeah. 00:04:56.925 --> 00:04:58.995 let me know if you're successful on that. 00:04:59.385 --> 00:04:59.535 yeah. 00:04:59.915 --> 00:05:00.485 okay. 00:05:00.665 --> 00:05:03.725 So the first time you read it, what did you think? 00:05:04.675 --> 00:05:06.505 Kert Viele: alright, so this is, first off, this is a 00:05:06.505 --> 00:05:07.765 really interesting place to be. 00:05:07.765 --> 00:05:10.315 'cause where we are is the public comment period. 00:05:10.495 --> 00:05:14.185 So, you know, the natural thing to do is immediately dive into issues. 00:05:14.185 --> 00:05:15.055 But, you know, I think the. 00:05:15.480 --> 00:05:19.740 The big picture is the first place to start, and by and 00:05:19.740 --> 00:05:21.570 large, this is a great document. 00:05:21.930 --> 00:05:27.090 this is evidence, you know, a continuation of the FDA, having adaptive guidance, 00:05:27.090 --> 00:05:29.880 the EMA, all of these other places. 00:05:30.190 --> 00:05:34.750 The fact that we've spent the last 20 years designing these trials, this is now 00:05:34.750 --> 00:05:37.300 something that all needs to be formalized. 00:05:37.640 --> 00:05:41.180 there's a wealth of experience, you know, hundreds of adaptive trials 00:05:41.180 --> 00:05:42.650 out there, probably thousands. 00:05:42.650 --> 00:05:45.380 I haven't done any kind of census on this. 00:05:45.670 --> 00:05:49.390 but this is a document that basically says adaptive trials are. 00:05:50.485 --> 00:05:53.335 Part of the armament, part of the things that you're meant to be coming 00:05:53.335 --> 00:05:54.775 with us and we need to discuss. 00:05:54.775 --> 00:05:55.915 So this is all great. 00:05:56.505 --> 00:06:00.645 I think the document does a wonderful job of laying out all the issues. 00:06:00.895 --> 00:06:03.655 all of the pros and cons of adaptive design. 00:06:04.055 --> 00:06:07.445 it talks about various kinds of designs, group sequentials, 00:06:07.445 --> 00:06:09.545 enrichment, seamless designs. 00:06:09.765 --> 00:06:13.605 it talks about platforms to some degree, although there are other 00:06:13.605 --> 00:06:15.045 guidance documents on that. 00:06:15.455 --> 00:06:20.225 So this is a thorough, complete document and I think everybody on the 00:06:20.225 --> 00:06:21.725 committee should be congratulated. 00:06:21.725 --> 00:06:25.805 So this is, you know, we're gonna talk about our, you know, points 00:06:25.805 --> 00:06:29.675 where we disagree on occasion, but by and large this is, you know, 00:06:29.705 --> 00:06:32.465 95%, you know, this is wonderful. 00:06:34.320 --> 00:06:40.970 Scott Berry: So 10 years ago, ICH was silent on adaptive designs and, it doesn't 00:06:40.970 --> 00:06:44.810 mean we weren't doing them, but to some extent it's not part of the harmonization. 00:06:45.050 --> 00:06:45.560 So I agree. 00:06:45.560 --> 00:06:47.720 The existence of this thing is great. 00:06:48.150 --> 00:06:51.900 and it says, and we will come to each one of these, but 00:06:51.900 --> 00:06:53.940 Bayesian can be confirmatory. 00:06:54.315 --> 00:06:56.145 Simulation is important. 00:06:56.265 --> 00:07:00.975 Seamless can be confirmatory, enrichment can be confirmatory. 00:07:01.575 --> 00:07:05.295 The, the, that, so that, I mean, almost full stop. 00:07:05.295 --> 00:07:09.265 We can end the, we can end the podcast and it says all of those things 00:07:09.820 --> 00:07:12.970 Kert Viele: Well, it, it says even more than that 'cause it says how to do it. 00:07:12.970 --> 00:07:13.960 I mean, there's layouts. 00:07:14.020 --> 00:07:17.050 If you do simulation, here's what needs to be in it. 00:07:17.050 --> 00:07:20.320 Here are the issues you need to discuss with regulators. 00:07:20.320 --> 00:07:25.990 It's not, it limits the surprises and de-risks the process for sponsors 00:07:25.990 --> 00:07:27.670 to adapt, adopt these methods. 00:07:28.030 --> 00:07:28.300 Scott Berry: Yep. 00:07:28.825 --> 00:07:29.035 Yeah. 00:07:29.485 --> 00:07:30.055 Okay. 00:07:30.175 --> 00:07:35.885 So let's go into, should we go into the various pieces of this? 00:07:36.075 --> 00:07:40.035 within it, what was the, what were potential, negatives? 00:07:40.035 --> 00:07:45.040 One of, one of the negatives for me when reading it is a little bit of the tone. 00:07:46.590 --> 00:07:47.760 to it a little bit. 00:07:48.180 --> 00:07:53.550 And, a really good example is back in 2010, the FDA came out with a 00:07:53.550 --> 00:07:59.580 draft guidance on adaptive designs, and it was felt to be, challenging. 00:07:59.730 --> 00:08:05.520 And they came out with the phrase that some designs are less well understood. 00:08:06.180 --> 00:08:11.760 And that phrase, that four word phrase for some people was. 00:08:12.225 --> 00:08:13.005 Negative. 00:08:13.455 --> 00:08:15.045 Oh, it's less well understood. 00:08:15.045 --> 00:08:16.365 You can't do this. 00:08:16.395 --> 00:08:17.325 It's a struggle. 00:08:18.255 --> 00:08:20.505 It didn't bother me when I read that. 00:08:20.565 --> 00:08:21.825 let's make it understood. 00:08:21.825 --> 00:08:24.515 You know, that's part of our job sort of thing, in it. 00:08:24.815 --> 00:08:32.585 So are there similar things here to perhaps the tone of it in your read that 00:08:32.585 --> 00:08:35.375 could be construed to be similar to that? 00:08:35.375 --> 00:08:36.695 Less well understood. 00:08:38.135 --> 00:08:40.325 Kert Viele: So on the plus side, I think there's less of that. 00:08:40.325 --> 00:08:43.835 I think they've, there some of that lesson has been learned. 00:08:44.115 --> 00:08:48.170 maybe that lesson is now well understood about how much a document influences. 00:08:49.185 --> 00:08:50.805 Practice indirectly. 00:08:51.405 --> 00:08:56.885 I think to me, in terms of the language, issues, like things 00:08:56.885 --> 00:08:59.225 are very hard or impossible. 00:08:59.515 --> 00:09:03.925 you know, things that I know that we've done and we've gotten accepted, I think 00:09:03.925 --> 00:09:08.275 are phrased as, oh, this is gonna be incredibly difficult and we know it's not. 00:09:08.695 --> 00:09:13.465 So I think that would be valuable to try to get rid of some of that 00:09:13.525 --> 00:09:15.895 and to more talk about the issues. 00:09:18.100 --> 00:09:18.400 Scott Berry: Yep. 00:09:19.060 --> 00:09:19.510 Okay. 00:09:19.570 --> 00:09:21.610 Alright, and the other really important thing, I should 00:09:21.610 --> 00:09:22.960 touch on the existence of it. 00:09:23.200 --> 00:09:29.470 Is there anything this document says you can't do when you're doing 00:09:29.470 --> 00:09:31.300 a design of an adaptive trial? 00:09:31.885 --> 00:09:35.025 Kert Viele: No, and I mean, we could keep in mind that there's nothing 00:09:35.175 --> 00:09:39.705 anywhere that says, I mean, I know more about the FDA's legal basis than 00:09:39.705 --> 00:09:43.635 other regulatory agencies, but there's nothing in the law that says you have 00:09:43.635 --> 00:09:45.525 to be frequentist versus Bayesian. 00:09:45.525 --> 00:09:48.015 There's nothing in the law that says 0.025. 00:09:48.345 --> 00:09:53.865 These are all, there are standards that, you know, this is an easier 00:09:53.865 --> 00:09:58.065 path than others, but nowhere does anything say no to anything. 00:09:59.505 --> 00:09:59.775 Scott Berry: Yep. 00:09:59.965 --> 00:10:01.255 which is a really important thing. 00:10:01.255 --> 00:10:04.195 We, I don't find this restrictive of at all. 00:10:04.555 --> 00:10:08.605 I'm not going to be involved with the design of a trial and alter 00:10:08.605 --> 00:10:10.375 it because of what this says. 00:10:10.985 --> 00:10:11.675 overall, 00:10:12.905 --> 00:10:15.245 Kert Viele: Yeah, I think this is really about where are you on the 00:10:15.245 --> 00:10:17.855 adoption cycle for adaptive trials. 00:10:18.065 --> 00:10:22.685 If you're on the bleeding edge, this document's probably irrelevant to you. 00:10:22.985 --> 00:10:25.625 It doesn't say no, you're gonna stay on the bleeding edge. 00:10:25.925 --> 00:10:30.090 This document is an invitation for people a little farther back to go. 00:10:30.635 --> 00:10:35.885 Adaptive designs are now stable enough, acceptable enough that another 00:10:35.885 --> 00:10:37.065 group of people can hop on board. 00:10:38.895 --> 00:10:39.165 Scott Berry: Okay. 00:10:39.165 --> 00:10:43.680 So on the grand scheme, this is an intangible summary from you of this 00:10:43.680 --> 00:10:48.960 particular document, the original FDA draft guidance document, 2010, I 00:10:48.960 --> 00:10:55.450 think it was, I think generally was perceived to be a lot of, caution. 00:10:55.820 --> 00:10:56.870 and generally. 00:10:57.605 --> 00:11:00.275 the people on the edge of this doing adaptive designs. 00:11:00.275 --> 00:11:03.545 It wasn't gonna stop 'em, but it didn't feel like an invitation. 00:11:04.265 --> 00:11:11.465 The current full adaptive design guidance by the FDA is almost saying, 00:11:11.495 --> 00:11:13.355 you should be doing some of these things. 00:11:13.355 --> 00:11:17.615 Some of these things are better and it's gonna supply better ways for us 00:11:17.615 --> 00:11:19.685 to make decisions on your treatments. 00:11:19.685 --> 00:11:22.505 It was almost kind of a kick in the butt, if you will. 00:11:22.830 --> 00:11:26.360 So it was moving the swath of people to saying, boy, we better 00:11:26.360 --> 00:11:27.230 know what's going on here. 00:11:27.230 --> 00:11:28.610 We better do this kind of thing. 00:11:29.120 --> 00:11:33.810 That maybe that's the sort of scale of, it, it's very encouraging. 00:11:33.810 --> 00:11:34.830 You should be doing this. 00:11:34.830 --> 00:11:38.910 It's gonna, it's gonna change the industry to caution and 00:11:38.910 --> 00:11:40.620 people are gonna use it as well. 00:11:40.620 --> 00:11:44.490 If you read this diet got document, I don't think people want you 00:11:44.490 --> 00:11:47.610 to do this kind of thing, where do you find this on that scale? 00:11:48.900 --> 00:11:52.410 Kert Viele: This is probably a little closer to 2010, and, but it's 00:11:52.410 --> 00:11:56.575 also, you know, a few changes and it moves closer to the other document. 00:11:57.335 --> 00:12:00.035 so this is all, this isn't substance. 00:12:00.035 --> 00:12:01.655 This is, as you've said, tone. 00:12:02.215 --> 00:12:05.215 so, you know, I think in terms of, let's talk about, for 00:12:05.215 --> 00:12:06.895 example, a bias discussion. 00:12:07.195 --> 00:12:10.195 There's discussions of, oh, if you do this or that, you could 00:12:10.195 --> 00:12:12.235 run into biased estimates and on. 00:12:12.575 --> 00:12:17.345 and all this is fair to be, I mean, I certainly, unbiased I think is important. 00:12:17.525 --> 00:12:19.325 Quality of estimates is important. 00:12:19.325 --> 00:12:21.035 I'm not sure unbiased is. 00:12:21.350 --> 00:12:23.960 You know, the most important thing in every setting. 00:12:24.470 --> 00:12:28.010 But this is a set, this is a situation where, you know, you 00:12:28.010 --> 00:12:31.460 could have a tone where, oh, all these horrible things might happen. 00:12:31.790 --> 00:12:35.510 But, you know, we respond to FDA and EMA questions on this. 00:12:35.630 --> 00:12:37.550 We quantify what the bias is. 00:12:37.550 --> 00:12:41.570 It's not that much and things are acceptable, or we 00:12:41.570 --> 00:12:42.920 need to make small changes. 00:12:43.400 --> 00:12:47.030 So this isn't, it's not a roadblock, it's something to be aware of. 00:12:47.715 --> 00:12:48.045 Scott Berry: Yep. 00:12:48.435 --> 00:12:48.735 Yep. 00:12:49.485 --> 00:12:49.995 Okay. 00:12:50.115 --> 00:12:53.565 So let's sort of dive into the various pieces of this. 00:12:53.865 --> 00:12:58.615 in terms of adaptive designs, it goes through, different parts of, 00:12:58.665 --> 00:13:00.495 different kinds of adaptive trials. 00:13:01.095 --> 00:13:04.825 So, by the way, this does make a reference that this is 00:13:04.825 --> 00:13:07.045 largely for confirmatory trials. 00:13:07.315 --> 00:13:09.655 So we should be thinking this late stage. 00:13:09.805 --> 00:13:11.920 There's one, one comment in there that I do. 00:13:12.250 --> 00:13:13.330 really not like that. 00:13:13.330 --> 00:13:17.350 Says that this guidance is also relevant for early trials, and I think there's 00:13:17.350 --> 00:13:21.290 a lot of places that's completely misplaced, that this isn't really guidance 00:13:21.290 --> 00:13:22.670 and it doesn't make sense in that. 00:13:22.670 --> 00:13:27.810 But, so think about this as phase three trials for registration of this. 00:13:27.990 --> 00:13:30.150 So it talks about seamless trials. 00:13:30.315 --> 00:13:33.970 What do you think about the discussion it has about seamless trials? 00:13:35.070 --> 00:13:37.050 Kert Viele: So you gotta read that two or three times I think. 00:13:37.050 --> 00:13:42.030 So the seamless trial discussion, it talks about in some detailed trials 00:13:42.030 --> 00:13:46.230 that say, have two doses and you do a dose selection and narrow it down. 00:13:46.710 --> 00:13:50.730 It doesn't talk about trials where, you know, what we would 00:13:50.730 --> 00:13:52.135 consider the more general seamless. 00:13:52.805 --> 00:13:57.125 Where you're combining a phase two dose finding and a phase three confirmatory 00:13:57.125 --> 00:13:59.795 trial where you've got lots of doses. 00:14:00.355 --> 00:14:03.535 and it, it has a couple really negative statements in there. 00:14:03.725 --> 00:14:07.955 I think it says adaptive trial shouldn't be a replacement for 00:14:08.060 --> 00:14:10.310 a proper dose finding trial. 00:14:10.750 --> 00:14:15.250 you know, I think this is one of my places I really would like to see this changed. 00:14:15.620 --> 00:14:16.790 you know, a seamless trial. 00:14:17.370 --> 00:14:21.640 The fundamental advantage that we see, and this goes back to, I should 00:14:21.640 --> 00:14:22.960 probably give the broader picture. 00:14:23.200 --> 00:14:27.340 If you separate these trials, what companies will do is they'll spend 00:14:27.340 --> 00:14:31.120 a little bit of time in phase two, so they can move to phase 00:14:31.120 --> 00:14:32.830 three as quickly as possible. 00:14:33.070 --> 00:14:36.400 When they do that, you're more likely to get the wrong dose 00:14:36.820 --> 00:14:38.680 when it's a seamless trial. 00:14:39.010 --> 00:14:43.060 You actually spend more time in phase two, you can do the math. 00:14:43.060 --> 00:14:46.690 You should spend more time in phase two because you get to use the 00:14:46.690 --> 00:14:49.060 data, you make better decisions. 00:14:49.240 --> 00:14:53.410 So I would say a seamless two, three trial, not always, we should have a 00:14:53.410 --> 00:14:59.980 discussion, but in many cases it's a more proper dose ranging trial and gets 00:15:00.130 --> 00:15:02.020 the information that regulators want. 00:15:02.020 --> 00:15:04.210 It has more information about what is the right dose. 00:15:05.390 --> 00:15:09.140 Scott Berry: Yeah, so they, there's snuck in there some phrases that just 00:15:09.140 --> 00:15:14.450 feel sort of cautionary, like saying if you do a seamless two three trial, you 00:15:14.450 --> 00:15:19.100 might have to do a proper dose ranging trial as though that's not proper. 00:15:19.380 --> 00:15:21.870 so a few words like that go away. 00:15:22.185 --> 00:15:25.355 And it's overly, you can do seamless two, three trials. 00:15:25.415 --> 00:15:29.495 And I agree that's the part where in, in some sense it also says that 00:15:29.495 --> 00:15:34.865 the goal of adaptive designs is to be more efficient and it almost 00:15:34.865 --> 00:15:37.535 sort of brings out this aspect that. 00:15:37.890 --> 00:15:40.380 It's not be, you're not gonna get better answers. 00:15:40.380 --> 00:15:42.630 You're just going to do it with less resources. 00:15:43.170 --> 00:15:47.010 And I think you gave a great example of where you actually 00:15:47.010 --> 00:15:48.300 get better answers with this. 00:15:48.430 --> 00:15:51.340 I'll throw a trial out there where I think this exactly happened. 00:15:51.340 --> 00:15:55.930 The Sepsis Act trial was a trial in sepsis. 00:15:55.930 --> 00:15:57.340 Faring was the sponsor. 00:15:57.340 --> 00:15:59.290 They published a paper on the design. 00:15:59.740 --> 00:16:01.695 It was a seamless two three trial with. 00:16:02.035 --> 00:16:07.795 Four doses in phase two going to phase three where it picks a single dose 00:16:08.455 --> 00:16:14.125 because the data in phase two was going to be included in the final analysis, 00:16:14.125 --> 00:16:16.225 and that's what a seamless trial is. 00:16:16.435 --> 00:16:21.985 It moves from multiple doses to a single dose, but you can use the data in the 00:16:21.985 --> 00:16:27.865 first part in the con confirmation because that data was used in the confirmation. 00:16:27.865 --> 00:16:30.115 They made the phase two part larger. 00:16:31.275 --> 00:16:36.555 it's actually a trial that because it was larger, it stopped for futility 00:16:36.555 --> 00:16:37.995 and didn't go to phase three. 00:16:39.015 --> 00:16:39.615 Interesting. 00:16:39.615 --> 00:16:43.275 If you go into the interim of where that trial was and you go back 00:16:43.275 --> 00:16:46.605 to a small trial, they might have run, they might have jumped and 00:16:46.605 --> 00:16:48.705 run a very large phase three trial. 00:16:49.230 --> 00:16:52.320 Inappropriately and failed this large phase three trial. 00:16:52.350 --> 00:16:55.500 They got better answers by being adaptive in that circumstance. 00:16:55.500 --> 00:16:59.460 This isn't, you know, it, there, there are efficiencies, but that's a, that's 00:16:59.460 --> 00:17:04.050 an example of a trial where I think you get better answers by being adaptive. 00:17:04.975 --> 00:17:06.305 Kert Viele: Well, in a lot of trials like this. 00:17:06.605 --> 00:17:11.195 We go, Hey, if I do this one, if I do this in a standard way, I can only 00:17:11.195 --> 00:17:14.465 explore three doses if those efficiencies. 00:17:14.465 --> 00:17:16.535 We don't always trade them for sample size. 00:17:16.775 --> 00:17:20.645 We trade them to answer a more complete question, explore five doses. 00:17:20.825 --> 00:17:21.930 Whatever you need to do here. 00:17:22.755 --> 00:17:23.115 Scott Berry: Yep. 00:17:23.235 --> 00:17:28.345 And if I have to do fixed randomization and I add a dose, it linearly 00:17:28.345 --> 00:17:34.465 increases the size of the trial if I do adaptations over there, seamless. 00:17:34.465 --> 00:17:35.575 Two three response. 00:17:35.575 --> 00:17:36.400 Adaptive randomization. 00:17:36.810 --> 00:17:38.460 I might explore more doses. 00:17:38.670 --> 00:17:43.350 And we have examples of that as well where Eli Lilly did seven doses 00:17:43.350 --> 00:17:47.310 exploring Trulicity, where the original design was three in a fixed trial. 00:17:47.310 --> 00:17:50.310 But because of response, adaptive randomization, they could explore 00:17:50.310 --> 00:17:52.450 seven, within that circumstance. 00:17:52.450 --> 00:17:55.070 So, multiple examples of that. 00:17:55.170 --> 00:17:58.380 so I think it's a case where, first of all, they talk about 00:17:58.380 --> 00:17:59.760 seamless being confirmatory. 00:17:59.760 --> 00:18:03.685 That's great, but there's a few hidden adjectives we would love to see removed. 00:18:06.160 --> 00:18:06.760 Okay. 00:18:06.940 --> 00:18:09.250 what about Bayesian? 00:18:09.430 --> 00:18:12.130 Does it talk about Bayesian in the guidance 00:18:13.370 --> 00:18:14.450 Kert Viele: so yes and no. 00:18:14.520 --> 00:18:16.860 so I, you know, I have strong opinions about the Bayesian. 00:18:17.050 --> 00:18:18.580 I noticed you didn't lead with that. 00:18:18.830 --> 00:18:24.020 but anyway, so the Bayesian trial, it is there, there's a section on it. 00:18:24.310 --> 00:18:29.170 there's a brief introduction to Bays in general, defines it as 00:18:29.200 --> 00:18:34.060 basically updating data with a prior and so on, and then it moves to. 00:18:34.655 --> 00:18:39.045 borrowing of external trial data, which going back to the 00:18:39.045 --> 00:18:41.085 seamless trial, it's kind of odd. 00:18:41.425 --> 00:18:45.415 you know, if you say you can't do seamless, you're saying you can't borrow 00:18:45.415 --> 00:18:49.945 in that context, but you could borrow external data into your phase three 00:18:49.945 --> 00:18:54.445 trial, but you can't borrow from phase two that was randomized in the same anyway. 00:18:54.445 --> 00:18:56.775 It's an odd, it doesn't flow right. 00:18:57.345 --> 00:19:00.570 So, but the Bayesian, I just, one thing that's important to 00:19:00.570 --> 00:19:05.430 me when you get into borrowing Bayesian is not equal to borrowing. 00:19:05.700 --> 00:19:08.640 There are lots of ways to be a frequentist and do borrowing. 00:19:08.830 --> 00:19:12.610 we don't tend to do those, so I'm not expert in them, but they exist. 00:19:12.915 --> 00:19:17.335 and they certainly do nice things, Bayesians do borrowing, and there's 00:19:17.335 --> 00:19:22.225 lots to Bayesian that is not borrowing and that's given short shrift there. 00:19:22.225 --> 00:19:25.795 I think there ought to be discussions on hierarchical models for modeling 00:19:25.795 --> 00:19:30.325 subgroups, predictive probabilities there, those kind of things. 00:19:30.325 --> 00:19:34.495 Should, Bayesian should not only be informative priors, there 00:19:34.495 --> 00:19:35.905 are lots of other applications. 00:19:36.990 --> 00:19:37.230 Scott Berry: Yep. 00:19:37.500 --> 00:19:37.710 Yep. 00:19:38.080 --> 00:19:42.150 in that, so gen generally, when you read that section, and by the 00:19:42.150 --> 00:19:46.590 way, that's done by others, a lot of times people contact us and say, 00:19:46.590 --> 00:19:48.630 we want to do a Bayesian trial. 00:19:48.900 --> 00:19:53.970 And what they mean is an adaptive trial in that it's not necessarily driven by. 00:19:54.325 --> 00:20:00.585 Bayesian statistical models, in that many of them are, which is why that tying those 00:20:00.585 --> 00:20:03.095 things together happened and many times. 00:20:03.185 --> 00:20:08.915 I su, I suspect when people come to the agencies about potentially 00:20:08.915 --> 00:20:12.305 borrowing, they're probably bringing a Bayesian with them. 00:20:12.575 --> 00:20:15.995 And so there's ways in which these things kind of get tied 00:20:15.995 --> 00:20:17.755 together, but agree completely. 00:20:17.945 --> 00:20:18.965 they're separate topics. 00:20:19.115 --> 00:20:23.075 Overall, what was your view of their discussion? 00:20:23.075 --> 00:20:23.825 Of base? 00:20:24.685 --> 00:20:27.385 Kert Viele: so I thought with the idea that it needs to be expanded for 00:20:27.385 --> 00:20:31.485 the bays part, the borrowing part I liked, it has a reasonable discussion. 00:20:31.975 --> 00:20:33.835 Of what you need to worry about. 00:20:34.135 --> 00:20:40.045 I am not anti type one error, so I am okay with the discussion of that as well. 00:20:40.045 --> 00:20:41.305 I think it ought to be there. 00:20:41.455 --> 00:20:42.955 I think there's a discussion. 00:20:43.195 --> 00:20:46.405 You know, when we go to the agency with these trials, a lot of times 00:20:46.405 --> 00:20:47.965 we don't even argue the statistics. 00:20:47.965 --> 00:20:51.325 We argue is the data we want to borrow from. 00:20:51.620 --> 00:20:54.650 Relevant, and that's certainly in this document. 00:20:54.650 --> 00:20:57.675 So I think it says, you know, here are the issues you ought to 00:20:57.675 --> 00:21:00.615 discuss, and these are the issues that we've discussed in the past. 00:21:00.945 --> 00:21:04.005 And borrowing is sometimes appropriate and sometimes not. 00:21:04.375 --> 00:21:07.805 Scott Berry: I thought it was, in some ways I think it's relatively weak. 00:21:07.805 --> 00:21:09.245 There's not a lot there. 00:21:09.465 --> 00:21:10.125 that's not bad. 00:21:10.865 --> 00:21:16.985 in it, again, it's not restrictive in it, it exists, and this is a harmonization. 00:21:16.985 --> 00:21:21.785 So some ways this is the factors that largely agreed upon regulatory agencies 00:21:21.785 --> 00:21:24.095 globally is that Bess is confirmatory. 00:21:24.500 --> 00:21:29.250 I, I mean it's existence in that, that's a really important thing in that, and 00:21:29.250 --> 00:21:36.390 it being not overly prescriptive I think is actually pretty good in there. 00:21:36.870 --> 00:21:42.750 I thought it was opening, the idea of borrowing is a very. 00:21:43.420 --> 00:21:47.170 A reasonable thing, and I think by the way, forecasting forward, that's something 00:21:47.170 --> 00:21:49.240 that's gonna be done much more now. 00:21:49.240 --> 00:21:52.960 It's not always borrowing things we already know about. 00:21:52.960 --> 00:21:58.720 Historically, it's a trial of four baskets that I'm prospectively going 00:21:58.720 --> 00:22:04.480 into four subsets of disease and I wanna borrow from one subset to another. 00:22:04.810 --> 00:22:09.780 I think this is pretty open to that idea as a very reasonable thing to do. 00:22:10.110 --> 00:22:10.170 It. 00:22:11.280 --> 00:22:13.845 Kert Viele: So I read it as well, and I think here's where it's 00:22:13.845 --> 00:22:15.660 worth a little bit more language. 00:22:16.020 --> 00:22:20.340 Because part of the language in there on say, bias and estimation 00:22:20.820 --> 00:22:24.150 is a little bit at odds with bays. 00:22:24.510 --> 00:22:28.740 I mean, to me, working in adaptive trials for the last 15 years or 00:22:28.770 --> 00:22:30.180 however it has been, 15 years. 00:22:30.180 --> 00:22:32.130 Oh, you're, anyway, it's been 15 years. 00:22:32.430 --> 00:22:36.740 The, But anyway, the greatest advantage to bays when you're doing adaptive 00:22:36.740 --> 00:22:41.210 trials is the posterior distribution is the posterior distribution. 00:22:41.600 --> 00:22:45.080 You don't have to worry about correcting it. 00:22:45.080 --> 00:22:47.270 You don't have to worry about what the design is. 00:22:47.270 --> 00:22:51.080 You don't need these different, corrections all over the place. 00:22:51.420 --> 00:22:53.220 the posterior is the posterior. 00:22:53.220 --> 00:22:55.500 The inferences are the inferences, and you are done. 00:22:55.950 --> 00:23:00.450 However, from a strict frequentist standpoint, those are. 00:23:00.735 --> 00:23:05.295 Biased in some respect, and I think there ought to be some discussion that we 00:23:05.295 --> 00:23:10.485 ought to be trying to get good estimates, and especially in adaptive trials, that 00:23:10.485 --> 00:23:16.005 should be a focus on mean square error and not necessarily bias is the only thing 00:23:16.245 --> 00:23:18.255 or the pre-eminent thing in the world. 00:23:18.870 --> 00:23:19.290 Scott Berry: Yeah. 00:23:19.560 --> 00:23:19.980 Yep. 00:23:20.640 --> 00:23:20.970 Yep. 00:23:20.970 --> 00:23:21.480 agree. 00:23:21.885 --> 00:23:22.485 Okay. 00:23:23.065 --> 00:23:26.035 you talked about enrichment trials being in there. 00:23:26.285 --> 00:23:29.585 anything specifically about enrichment trials? 00:23:30.195 --> 00:23:31.455 Kert Viele: I like what's there. 00:23:31.455 --> 00:23:34.755 I do think it auto, this is the same with, seamless. 00:23:34.815 --> 00:23:39.195 It talked about two doses and not the full seamless, the enrichment 00:23:39.195 --> 00:23:40.815 section, it talks about essentially. 00:23:41.545 --> 00:23:43.615 Biomarker positive and negative. 00:23:43.615 --> 00:23:45.235 And that's the focus. 00:23:45.305 --> 00:23:48.335 we do a lot of trials, you know, rare oncology, rare. 00:23:48.895 --> 00:23:53.425 Tumor types in oncology where we may look at 10 different cancers 00:23:53.425 --> 00:23:55.765 and we wanna borrow among them. 00:23:55.985 --> 00:24:02.105 the ROAR trial, for example, was the first pan tumor approval for FDA at least, 00:24:02.205 --> 00:24:06.345 where you got a drug that was targeted to anything that had the right mechanism. 00:24:06.715 --> 00:24:06.765 the. 00:24:07.200 --> 00:24:09.600 The right target for the mechanism to act on. 00:24:09.940 --> 00:24:14.950 I do think the enrichment section should talk about those kind of trials, and 00:24:15.070 --> 00:24:20.020 this gets into the estimation again, we know if we look at multiple groups, the 00:24:20.020 --> 00:24:24.580 highest of the high is biased high, and, you know, hierarchal models to correct 00:24:24.580 --> 00:24:26.950 that, that's an important part of this. 00:24:26.950 --> 00:24:31.670 Again, getting good estimates, not necessarily unbiased, isn't 00:24:31.670 --> 00:24:32.870 necessarily the best estimate. 00:24:33.610 --> 00:24:33.940 Scott Berry: Yep. 00:24:34.300 --> 00:24:34.510 Yep. 00:24:35.050 --> 00:24:36.850 and it tied in this idea. 00:24:36.850 --> 00:24:43.510 So we do a lot of adaptive designs where we've been told no by a 00:24:43.510 --> 00:24:47.000 regulator because we haven't yet convinced them of type one error. 00:24:47.575 --> 00:24:51.910 I mean, this is a red line to some extent that we have had trials 00:24:51.910 --> 00:24:56.110 where we know type one error is inflated, and that's been okay too. 00:24:56.260 --> 00:24:59.110 So I'm not saying that every single trial is that way, but 00:24:59.110 --> 00:25:00.460 that's a very strong thing. 00:25:01.490 --> 00:25:08.480 maybe this is a bad thing to say, but I've never, and we supply estimation summaries. 00:25:09.035 --> 00:25:13.985 In almost every one of our simulation reports, we look at the behavior of 00:25:13.985 --> 00:25:19.175 the estimation, and I don't think we've ever been told no because 00:25:19.175 --> 00:25:20.975 of the behavior of the estimation. 00:25:23.165 --> 00:25:25.535 Kert Viele: Well, I think in general we want this to, the 00:25:25.535 --> 00:25:27.425 shrinkage does reasonable things. 00:25:27.755 --> 00:25:31.145 We actually run into the opposite problem where we're, we run a 00:25:31.145 --> 00:25:35.795 basket trial, one of the baskets, the drug performs incredibly well. 00:25:36.005 --> 00:25:40.715 We shrink it down and we can run into debates over, oh, I don't want the 00:25:40.715 --> 00:25:44.635 estimate to be less, Hey, I want to be able to put the big number on there. 00:25:44.965 --> 00:25:47.425 But we go, Hey, you know, this is the highest of the high. 00:25:47.425 --> 00:25:48.295 It's biased high. 00:25:48.295 --> 00:25:49.105 It should come down. 00:25:50.575 --> 00:25:50.995 Scott Berry: Yep. 00:25:51.245 --> 00:25:54.665 and in some scenarios we know that the estimate of the 00:25:54.665 --> 00:25:55.985 highest group's gonna be biased. 00:25:55.985 --> 00:25:59.135 I mean, that's the whole role of the particular shrinkage of that. 00:25:59.675 --> 00:26:03.485 It does spend a good bit of time on, on two other topics 00:26:03.485 --> 00:26:04.895 in the FDA it does as well. 00:26:05.125 --> 00:26:09.605 Operational bias and the role of that, the role of blinding and all of that. 00:26:09.605 --> 00:26:11.165 What do you think it's dis about? 00:26:11.165 --> 00:26:16.145 Its discussion about the implementation, about adaptive designs, ensuring 00:26:16.145 --> 00:26:18.845 blinding in those care, those aspects. 00:26:19.230 --> 00:26:20.550 Kert Viele: I think those are great. 00:26:20.650 --> 00:26:23.920 I think, you know, if I have comments, they're all minor. 00:26:24.170 --> 00:26:25.460 it lays out issues. 00:26:25.460 --> 00:26:29.510 It talks about communication flows, emphasizes you don't want 00:26:29.510 --> 00:26:31.730 data to get to the wrong places. 00:26:32.160 --> 00:26:38.230 it discusses what I would consider maybe esoteric topics for Tammy, in terms of. 00:26:38.860 --> 00:26:42.550 If a sponsor needs access to certain information, there needs 00:26:42.550 --> 00:26:46.240 to be blinded firewalls in place, you know, to make sure it doesn't, 00:26:46.400 --> 00:26:48.320 doesn't negatively affect the study. 00:26:48.590 --> 00:26:51.470 There's discussions of interim analysis committees. 00:26:51.470 --> 00:26:52.760 I think it's very thorough. 00:26:52.760 --> 00:26:53.930 I like it a lot. 00:26:54.605 --> 00:26:55.085 Scott Berry: Yeah. 00:26:55.175 --> 00:26:55.385 Yeah. 00:26:55.385 --> 00:26:55.835 I do too. 00:26:55.890 --> 00:26:57.210 I, and largely agree with it. 00:26:57.550 --> 00:27:01.780 and making sure that's done really well, I think is important for adaptive designs. 00:27:02.170 --> 00:27:07.440 So they, it also discusses a, what we call an adaptive design report. 00:27:07.950 --> 00:27:12.630 A report that summarizes and describes in complete detail the 00:27:12.630 --> 00:27:17.780 adaptive design and the simulations conducted, around the design. 00:27:18.080 --> 00:27:21.560 What did you think about its description of the need for 00:27:21.560 --> 00:27:24.055 this, the components of this? 00:27:24.795 --> 00:27:26.335 In the ICH. 00:27:27.120 --> 00:27:30.450 Kert Viele: I think so this is another one of my sections 00:27:30.450 --> 00:27:32.100 I think has really done well. 00:27:32.410 --> 00:27:36.800 I think it lays out tons of issues, talks about different scenarios, 00:27:36.980 --> 00:27:39.230 has important points like, you know. 00:27:39.570 --> 00:27:43.080 Plausible scenarios rather than all scenarios. 00:27:43.290 --> 00:27:46.770 So there's the famous example we've talked about in other podcast where 00:27:46.770 --> 00:27:50.700 we were asked to simulate when people become immortal under certain 00:27:50.700 --> 00:27:55.320 conditions, which I think is, you know, largely a time of time passed. 00:27:55.600 --> 00:27:59.245 but this document is in the present, so it does that well. 00:27:59.705 --> 00:28:03.755 it has a few throwaway statements in it, I'd say. 00:28:03.755 --> 00:28:06.605 I don't know if they're throwaway or not, but it's essentially. 00:28:06.950 --> 00:28:12.860 This is impossible to do under certain conditions, and I would prefer that 00:28:13.160 --> 00:28:15.470 this gets back to less well understood. 00:28:15.470 --> 00:28:20.120 I'm worried about somebody just pointing to that and going, oh, they wrote 10 pages 00:28:20.120 --> 00:28:25.040 on this, but they really only meant the one sentence that says this is impossible. 00:28:25.130 --> 00:28:29.390 That's the secret code that tells us not to read the other 10 pages 00:28:29.390 --> 00:28:30.590 and pay any attention to it. 00:28:31.130 --> 00:28:32.480 Sometimes this is hard. 00:28:32.480 --> 00:28:36.560 Sometimes the null space is hard to explore and you know, we need 00:28:36.560 --> 00:28:38.900 to be transparent about that. 00:28:39.140 --> 00:28:42.380 But in a number of cases, you know, a dichotomous response. 00:28:42.710 --> 00:28:46.430 You know, look at a grid of probabilities and you've done what you needed to do. 00:28:47.685 --> 00:28:52.485 Scott Berry: This is another place if we go back to, there's an ICHE 00:28:52.485 --> 00:28:57.135 20 guidance that says simulation for demonstration of type one error is. 00:28:58.575 --> 00:29:00.075 re is a thing you can do. 00:29:00.285 --> 00:29:04.425 It's a scientifically reasonable thing, and it says it in the ICHE 20. 00:29:04.515 --> 00:29:05.265 It's draft. 00:29:05.625 --> 00:29:10.065 but that's another place that, looking back over the years, that's 00:29:10.065 --> 00:29:14.415 been a controversial thing for years and it was never in the FDA. 00:29:14.415 --> 00:29:17.565 The question is this, and now it's in the ICHE 20. 00:29:17.955 --> 00:29:21.135 The progress that's been made in this, I, that's another place 00:29:21.135 --> 00:29:22.095 we don't want to ignore that. 00:29:22.095 --> 00:29:24.795 This is a super positive thing that's in them. 00:29:28.215 --> 00:29:28.905 Okay. 00:29:29.175 --> 00:29:35.385 there are a couple things that we kind of disagree with that I, and I'll 00:29:35.385 --> 00:29:40.065 call them little myths, for example, that I'm not sure are grounded. 00:29:40.605 --> 00:29:42.675 Don't do a lot of interims. 00:29:42.705 --> 00:29:45.805 You know, there, there are people that have this viewpoint that, oh, 00:29:45.805 --> 00:29:47.425 you should only do one interim. 00:29:47.545 --> 00:29:50.155 And we get people that tell us, oh, you shouldn't do two interims. 00:29:50.305 --> 00:29:51.745 Regulators don't like that. 00:29:52.015 --> 00:29:56.815 Now, it doesn't tell you can't do six interims in it. 00:29:56.815 --> 00:30:01.615 But it gives this feeling that generally interims are bad. 00:30:02.455 --> 00:30:03.715 If you read that statement, 00:30:06.875 --> 00:30:08.015 in, in that, and we. 00:30:09.530 --> 00:30:15.470 I understand the notion that I, and it's interesting for a long time, and 00:30:15.680 --> 00:30:21.920 Don has said this, that if you do one interim, it might actually be something 00:30:21.920 --> 00:30:25.160 that supplies a certain amount of information about what's going on. 00:30:25.370 --> 00:30:29.960 If you were to, at the other extreme, do interims every day, there's almost 00:30:29.960 --> 00:30:31.820 nothing that's ever learned by anybody. 00:30:32.240 --> 00:30:36.020 Because it's just continually going on within this. 00:30:36.080 --> 00:30:40.460 So it's always a statement I've found odd, and I'm not sure it's actually 00:30:40.460 --> 00:30:43.460 real, but it's repeated by people. 00:30:43.460 --> 00:30:47.170 So there's a couple things like that, that show up in the document as well. 00:30:47.735 --> 00:30:52.105 Kert Viele: and one thing I think we ought to, so things that aren't missed that are 00:30:52.105 --> 00:30:55.610 cautionary in this document, I think we ought to be, we ought to make sure we. 00:30:55.945 --> 00:30:56.395 Echo. 00:30:56.395 --> 00:31:03.505 In this podcast, there's a discussion about there is some limit on how many 00:31:03.805 --> 00:31:06.295 adaptations and so on you can do. 00:31:06.625 --> 00:31:10.825 Now, we certainly aren't on the side of, you know, hey, three interims as 00:31:10.825 --> 00:31:13.135 opposed to one is exceeding the limit. 00:31:13.315 --> 00:31:17.095 That's not what we would call the budget, but certainly the 00:31:17.095 --> 00:31:18.955 document raises a good point. 00:31:19.195 --> 00:31:23.635 If you're talking about a platform trial, the model's changing. 00:31:23.830 --> 00:31:26.590 there's different protocols at different sites. 00:31:26.810 --> 00:31:29.330 you know, lots of things can be very complex. 00:31:29.330 --> 00:31:29.630 We've. 00:31:30.085 --> 00:31:33.775 We've at some level experimented with experimenting, and I 00:31:33.775 --> 00:31:34.675 think it should be fair. 00:31:34.675 --> 00:31:38.935 We've walked at the end of the trial and said, Hey, that was maybe too complex. 00:31:38.935 --> 00:31:41.305 It didn't do exactly what we wanted it to do. 00:31:41.645 --> 00:31:43.595 but this should be part of the discussion. 00:31:43.595 --> 00:31:47.795 I don't think, I don't think anyone should take this as, you know, don't 00:31:47.795 --> 00:31:50.075 do two interims or don't do two things. 00:31:50.075 --> 00:31:55.655 You know, do a factorial design with early stopping, but don't throw in everything 00:31:55.655 --> 00:31:58.055 in the toolbox at once, is fair. 00:31:58.820 --> 00:31:59.180 Scott Berry: Yep. 00:31:59.240 --> 00:32:03.175 I, and I do think that's fair that touches on another topic that I'm 00:32:03.175 --> 00:32:05.065 interested in your viewpoint on this. 00:32:05.365 --> 00:32:10.315 It does give the viewpoint that you should justify why you're 00:32:10.315 --> 00:32:12.705 doing, say, an adaptive design. 00:32:14.015 --> 00:32:18.220 Now, I think some people might take the viewpoint of what 00:32:18.280 --> 00:32:20.110 does somebody have to justify? 00:32:20.110 --> 00:32:21.310 They're doing a fixed trial. 00:32:21.310 --> 00:32:24.130 Why do I have to justify a scientifically credible thing? 00:32:24.640 --> 00:32:29.140 What's your take on that general sentiment in the document? 00:32:30.220 --> 00:32:33.430 Kert Viele: So I am, I'm in favor of that sentiment in the document. 00:32:33.790 --> 00:32:37.390 I do think, You know, if we're talking about non-adaptive trials, 00:32:37.450 --> 00:32:38.830 they're easier to implement. 00:32:39.310 --> 00:32:43.240 They're usually easier for people to write up to understand. 00:32:43.420 --> 00:32:47.150 If you get into corrections for estimation from a frequentest 00:32:47.170 --> 00:32:52.000 standpoint, there are complexities that come with an adaptive trial if 00:32:52.000 --> 00:32:55.000 you are going to get exactly the same. 00:32:55.000 --> 00:32:55.990 So at the end of the day. 00:32:56.680 --> 00:33:00.210 Patients are gonna be treated essentially the same chance they 00:33:00.210 --> 00:33:04.050 get a good drug, the same chance they mistakenly get a bad drug. 00:33:04.050 --> 00:33:09.390 If all of that's the same, and that includes time, correct, choice of 00:33:09.390 --> 00:33:12.570 dose, I mean there's a lot more into it than power and type 1 error. 00:33:12.810 --> 00:33:15.990 If those are all the same, definitely go simple. 00:33:16.170 --> 00:33:20.190 You need to say, here is the advantage that justifies the 00:33:20.190 --> 00:33:22.260 cost of doing an adaptive design. 00:33:22.630 --> 00:33:27.310 Often that's in adaptive's favor, but we know, you know, if accruals fast, 00:33:27.730 --> 00:33:32.320 not time to be adaptive, for example, there are exceptions to adaptive wins. 00:33:33.985 --> 00:33:36.955 Scott Berry: The other part of that, and I'm completely fine with that as 00:33:36.955 --> 00:33:40.200 well, there, there's one challenge to it, but one other part is. 00:33:42.250 --> 00:33:44.380 we don't decide on a design. 00:33:44.380 --> 00:33:47.830 We don't throw in adaptive things because, ooh, it would be cool 00:33:47.830 --> 00:33:50.440 to do that, or we're showing off. 00:33:50.500 --> 00:33:54.250 We've gone through a sequence of comparisons with the design team. 00:33:54.250 --> 00:33:56.050 We've said, here's what you could do with this. 00:33:56.170 --> 00:33:57.610 We always start with a fixed design. 00:33:57.610 --> 00:33:58.720 Here's a fixed design. 00:33:58.900 --> 00:34:00.550 Now we start to add adaptations. 00:34:00.700 --> 00:34:04.060 We've done all those comparisons and we've demonstrated the 00:34:04.060 --> 00:34:05.920 advantages that this design. 00:34:06.530 --> 00:34:08.070 Supplies, in it. 00:34:08.070 --> 00:34:09.780 So I think it's a very reasonable thing. 00:34:09.960 --> 00:34:14.940 One of the challenges that I've always found with this is we go through this 00:34:14.940 --> 00:34:19.350 iteration of the design where we do lots of comparisons to other designs. 00:34:19.650 --> 00:34:22.770 And what if we do RAR with different frequency? 00:34:22.770 --> 00:34:23.650 What if we don't do it? 00:34:24.600 --> 00:34:30.570 And it's hard to present that entire story of comparisons 00:34:30.570 --> 00:34:32.910 to regulators of the designs. 00:34:32.910 --> 00:34:33.960 We're not running. 00:34:34.630 --> 00:34:38.500 So we walk in and we say, this is the design here. 00:34:39.100 --> 00:34:45.400 I know it relies upon us to explain some, the value of the things we're 00:34:45.400 --> 00:34:49.720 doing on that, but it's a little bit harder at the end of the story 00:34:49.720 --> 00:34:51.190 to provide all of that story. 00:34:52.400 --> 00:34:55.195 Kert Viele: And we've had confusions where regulators have said back. 00:34:55.870 --> 00:34:59.980 essentially they have brought up reasonable issues about designs 00:34:59.980 --> 00:35:03.490 we're not proposing because they're often those comparisons. 00:35:03.490 --> 00:35:04.720 It can be confusing. 00:35:04.990 --> 00:35:09.400 So, you know, it's a cautionary tale to lay those out carefully and make 00:35:09.400 --> 00:35:10.990 sure we know what the default is. 00:35:11.350 --> 00:35:15.910 Now, I will say on that issue, my pet peeve is the one where it says 00:35:15.910 --> 00:35:18.880 Bayesian has to be justified adaptive. 00:35:19.150 --> 00:35:22.325 I think there's a cost to being adaptive that needs to be. 00:35:23.020 --> 00:35:27.610 Justified in terms of the extra effort, operational bays and 00:35:27.640 --> 00:35:30.280 frequentist, that's not the same thing. 00:35:30.370 --> 00:35:33.160 This is just two different ways of being inference. 00:35:33.160 --> 00:35:38.470 I always call this the animal farm argument of statistical paradigms. 00:35:38.830 --> 00:35:42.940 Frequentist is equal, but frequentist is more equal than base. 00:35:43.300 --> 00:35:45.590 So this, I don't, I object to that one. 00:35:46.595 --> 00:35:47.165 Scott Berry: it's good. 00:35:47.165 --> 00:35:49.085 I should have asked you your pet peeve. 00:35:49.085 --> 00:35:52.970 You have a pet peeve of ev each episode, so I, okay, good one. 00:35:55.140 --> 00:35:57.930 other topics I missed on the guidance 00:36:00.090 --> 00:36:01.050 draft guidance. 00:36:03.250 --> 00:36:06.250 Kert Viele: I think we've actually done pretty well going through it. 00:36:06.250 --> 00:36:11.150 As I said, you know, I think we should come back to the positives here and just 00:36:11.150 --> 00:36:13.700 how much of this document we agree with. 00:36:13.950 --> 00:36:16.830 I think we agree with virtually everything they say. 00:36:16.830 --> 00:36:19.770 You know, you should justify the following aspects. 00:36:19.770 --> 00:36:22.620 We agree, you know, these are the pros and cons. 00:36:22.620 --> 00:36:23.430 We agree. 00:36:23.710 --> 00:36:26.260 it's a lot of the adjectives around it. 00:36:26.550 --> 00:36:31.920 That we're, we are reticent of essentially, you know, people who 00:36:31.920 --> 00:36:36.210 are involved in regulatory consulting going, this is what this means. 00:36:36.210 --> 00:36:39.600 It's, oh, this is the sentence that invalidates the rest of the document. 00:36:40.020 --> 00:36:42.400 And we've been through that often enough. 00:36:42.550 --> 00:36:45.220 We'd like it to focus on the issues. 00:36:46.070 --> 00:36:46.460 Scott Berry: Yep. 00:36:46.610 --> 00:36:46.880 Yep. 00:36:47.035 --> 00:36:51.325 that the regulators don't think a seamless two three trial is proper. 00:36:52.145 --> 00:36:52.345 Kert Viele: Yeah. 00:36:52.390 --> 00:36:53.780 Scott Berry: you know, and weird things like that. 00:36:53.780 --> 00:36:57.470 So largely, I agree that I think this is, I read this and it was 00:36:57.470 --> 00:36:59.805 a little bit of uhoh, you know, what are we gonna get here? 00:37:00.075 --> 00:37:04.075 And I thought it was largely a very positive document, in that. 00:37:04.255 --> 00:37:07.025 So, congratulations to everybody that worked on it. 00:37:07.075 --> 00:37:07.465 I know. 00:37:07.465 --> 00:37:09.775 Consensus and harmonization. 00:37:10.045 --> 00:37:10.885 Yikes. 00:37:10.965 --> 00:37:13.665 in this, and so congratulations to them. 00:37:13.915 --> 00:37:19.945 are, is Barry planning to provide a response to this? 00:37:20.125 --> 00:37:21.715 They've invited responses. 00:37:21.745 --> 00:37:23.395 Are we planning to provide a response? 00:37:23.880 --> 00:37:27.070 Kert Viele: Yeah, we're definitely going to provide our own response. 00:37:27.130 --> 00:37:31.270 we're also interested, I know other groups are working on responses as well. 00:37:31.460 --> 00:37:34.310 we're happy to collaborate on those responses. 00:37:34.530 --> 00:37:37.920 so certainly, you know, we are, we're gonna make sure that we, we express 00:37:37.920 --> 00:37:39.630 where we would like to make changes. 00:37:39.860 --> 00:37:42.860 and I think we've covered a lot of the, our big topics today. 00:37:43.050 --> 00:37:43.360 Scott Berry: Yeah. 00:37:43.660 --> 00:37:43.990 Yeah. 00:37:44.530 --> 00:37:45.310 Brilliant. 00:37:45.400 --> 00:37:49.880 Alright, so, thank you Kurt for joining. 00:37:49.880 --> 00:37:54.980 Thanks for weighing in on this and here in the interim. 00:37:54.980 --> 00:37:59.960 And remember, we're always in the interim of science and life, so here we are. 00:38:00.560 --> 00:38:01.460 Thanks Kurt. 00:38:01.935 --> 00:38:02.385 Kert Viele: Thank you.