WEBVTT NOTE This file was generated by Descript 00:00:00.000 --> 00:00:03.050 Judith: Welcome to Berry's In the Interim podcast, where we explore the 00:00:03.050 --> 00:00:07.210 cutting edge of innovative clinical trial design for the pharmaceutical and 00:00:07.210 --> 00:00:09.800 medical industries, and so much more. 00:00:10.480 --> 00:00:11.170 Let's dive in. 00:00:12.844 --> 00:00:13.594 Scott Berry: All right. 00:00:13.594 --> 00:00:20.465 Welcome everybody to, in the interim, and Don Berry has joined me again and 00:00:20.465 --> 00:00:26.825 we, we have a, a really cool topic today to talk about in the interim. 00:00:27.185 --> 00:00:31.055 And that is the I SPY two trial. 00:00:32.255 --> 00:00:36.365 Many people may have heard of the I Spy two trial, so we're 00:00:36.365 --> 00:00:41.705 gonna go back and remember a little bit of, uh, of this trial. 00:00:42.065 --> 00:00:48.785 We'll, we'll walk through how it became, what it was, what happened in the trial. 00:00:49.385 --> 00:00:53.825 So Don, you ready to tell us the story of I Spy two? 00:00:54.830 --> 00:00:55.700 Don Berry: I sure am. 00:00:56.510 --> 00:00:57.590 Uh, thanks Scott. 00:00:59.270 --> 00:01:07.790 So I, SPY two has a, has a history, um, uh, and it revolves around Laura Essman. 00:01:07.790 --> 00:01:08.300 Of course. 00:01:08.300 --> 00:01:10.220 Laura Esserman was the Pi I. 00:01:10.910 --> 00:01:12.380 I was the co-PI 00:01:13.010 --> 00:01:13.940 of the trial. 00:01:14.090 --> 00:01:21.410 And this is back, we're talking, uh, like, uh, 2007 and eight. 00:01:22.340 --> 00:01:28.550 And she and I were part of a cooperative group called the C-A-L-G-B, 00:01:28.550 --> 00:01:32.330 the Cancer and Leukemia Group B, which itself has a long history. 00:01:32.840 --> 00:01:37.820 And we were in the breast committee, uh, of that, uh, cooperative group. 00:01:38.690 --> 00:01:44.390 And we had designed trials and commiserated with each other, 00:01:44.960 --> 00:01:50.330 uh, about the lack of innovation and about what we wanted to do. 00:01:52.790 --> 00:01:57.290 And, um, we designed a trial. 00:01:57.320 --> 00:01:58.670 I Spy one. 00:01:59.090 --> 00:02:02.810 You may wonder how the two came about, but I spy one. 00:02:03.650 --> 00:02:04.310 Um. 00:02:05.450 --> 00:02:10.250 Where we were looking at neoadjuvant disease. 00:02:10.280 --> 00:02:13.940 Now this is part of the story. 00:02:14.750 --> 00:02:23.240 Um, in breast cancer, there were two kinds of categories of disease at the time. 00:02:23.240 --> 00:02:32.781 One was adjuvant, which was, uh, patients who, uh, had typically had, uh, uh, 00:02:33.770 --> 00:02:40.100 uh, high risk disease, but were early patients, were newly diagnosed patients, 00:02:40.940 --> 00:02:49.610 uh, and they had, uh, bad tumors, you know, large tumors, um, uh, uh, 00:02:49.640 --> 00:02:57.680 tumors with, uh, positive lymph nodes and, uh, were receiving chemotherapy. 00:02:58.310 --> 00:03:02.990 Those with, um, estrogen receptor status positive were, we're receiving. 00:03:03.800 --> 00:03:07.430 Endocrine therapy, uh, Tamoxifen at that time. 00:03:08.270 --> 00:03:15.530 Um, and that, and, and what you would do is you would treat the 00:03:15.530 --> 00:03:24.620 patients, uh, after they had surgery and after they had surgery, means 00:03:24.620 --> 00:03:30.440 that for almost all of them, the tumor's gone because we took it out. 00:03:31.520 --> 00:03:34.190 Uh, and so what was the endpoint? 00:03:34.340 --> 00:03:37.490 The endpoint was when it comes back. 00:03:38.300 --> 00:03:43.670 Now, you know, if you're into clinical trials, that when it 00:03:43.670 --> 00:03:46.580 comes back is a dicey thing. 00:03:47.360 --> 00:03:54.560 Uh, it sometimes takes a long time and moreover, uh, which is of course good 00:03:54.560 --> 00:04:00.080 for the patient, uh, and moreover, it got better for the patient over 00:04:00.080 --> 00:04:03.110 time because these therapies worked. 00:04:03.515 --> 00:04:03.855 Scott Berry: and that 00:04:04.430 --> 00:04:07.640 Don Berry: Um, and that was a problem. 00:04:08.990 --> 00:04:14.300 The problem was that pharmaceutical companies and cooperative groups 00:04:14.300 --> 00:04:21.920 and the like, um, uh, sponsors of, uh, these things and patients, 00:04:22.220 --> 00:04:27.740 patient advocacy groups, it was difficult to run these trials. 00:04:28.580 --> 00:04:34.100 The other, just to touch the base, the other part, uh, is metastatic disease. 00:04:34.460 --> 00:04:40.250 Once it recurs, it becomes, um, uh, let's say it recurs distantly 00:04:40.250 --> 00:04:46.280 and, and, uh, uh, the lungs or the brain or the bone or something. 00:04:46.610 --> 00:04:48.860 Uh, then it becomes metastatic disease. 00:04:48.860 --> 00:04:50.660 So that's a different category. 00:04:50.780 --> 00:04:55.940 We're talking about the early, so-called early, uh, breast cancer. 00:04:56.765 --> 00:05:01.535 Scott Berry: So, so let me see if I, so the, the setting is typically when 00:05:01.535 --> 00:05:07.805 the diagnosis is made, the tumor's removed, and that's what adjuvant means. 00:05:07.805 --> 00:05:12.035 You go in, remove the tumor, and then you treat them with different therapies. 00:05:12.200 --> 00:05:14.035 Don Berry: the training is called the Avant. 00:05:14.360 --> 00:05:14.800 Scott Berry: adjuvant. 00:05:14.800 --> 00:05:15.000 Is 00:05:15.000 --> 00:05:15.120 called 00:05:15.885 --> 00:05:17.005 Don Berry: a adjuvant to 00:05:17.210 --> 00:05:17.390 Scott Berry: To 00:05:17.720 --> 00:05:18.290 surgery. 00:05:18.620 --> 00:05:24.440 To the surgery, and that trials ended up being extremely large because 00:05:24.440 --> 00:05:26.330 the endpoints took a long time. 00:05:26.330 --> 00:05:30.740 The treatments were reasonably effective, so clinical trials became 00:05:30.740 --> 00:05:37.730 incredibly uh, onerous to run in this patient population in breast cancer. 00:05:37.850 --> 00:05:38.120 Don Berry: Yeah. 00:05:38.210 --> 00:05:40.970 E expensive, uh, onerous. 00:05:41.330 --> 00:05:42.950 Nobody wanted to do it. 00:05:43.610 --> 00:05:49.490 Uh, what could we do and what the CLGB said? 00:05:50.450 --> 00:05:55.130 Uh, and I was the faculty statistician and working with the committee and. 00:05:55.715 --> 00:05:58.235 Uh, par parcel of these decisions. 00:05:58.895 --> 00:06:02.825 Um, we should try the neoadjuvant approach. 00:06:04.025 --> 00:06:06.155 Uh, now this was novel. 00:06:06.155 --> 00:06:15.065 It was led by, uh, other ki uh, something called the, uh, N-S-A-B-P, another one of 00:06:15.065 --> 00:06:17.825 the cooperative groups that had led this. 00:06:17.825 --> 00:06:26.525 Because the neoadjuvant approach is all you do is you exchange the surgery and the 00:06:26.525 --> 00:06:33.695 treatment, but that's a big deal because you're leaving the tumor in the body for 00:06:33.695 --> 00:06:43.175 like six months, but you're pummeling it with, um, uh, you know, lots of, 00:06:43.775 --> 00:06:50.285 uh, toxic, uh, therapies that kill lots of cells, including, of course, cancer cells. 00:06:51.095 --> 00:06:54.995 Um, so it was a big deal, but. 00:06:55.070 --> 00:06:55.290 Scott Berry: But 00:06:56.000 --> 00:06:57.140 Don Berry: It had a benefit. 00:06:58.100 --> 00:07:05.720 The benefit was that you got to see whether the tumor responded to the therapy 00:07:05.720 --> 00:07:07.550 that you gave them before you took it out. 00:07:08.060 --> 00:07:13.190 So you left it in, you watched it carefully, um, that you saw 00:07:13.280 --> 00:07:17.270 whether the treatment that you gave, the treatment you gave could 00:07:17.270 --> 00:07:19.100 be different in a clinical trial. 00:07:19.130 --> 00:07:21.560 Of course it should be because you're learning. 00:07:22.430 --> 00:07:31.880 And, uh, when you go in and do surgery after, uh, six months of therapy, you see 00:07:31.880 --> 00:07:34.340 whether or not the tumor is still there. 00:07:35.480 --> 00:07:41.090 And that in the neoadjuvant approach became the endpoint, uh, called 00:07:41.090 --> 00:07:42.830 pathologic complete response. 00:07:43.100 --> 00:07:47.720 You send the tissue to the pathologist and pathologists can't find any, uh, tumor. 00:07:48.290 --> 00:07:48.680 Okay? 00:07:49.010 --> 00:07:50.750 So that was sort of a risk. 00:07:50.750 --> 00:07:52.220 I described it to. 00:07:53.120 --> 00:08:01.070 Uh, the, my colleagues at the CLGB as betting the farm, uh, this had 00:08:01.640 --> 00:08:10.010 this, this was novel, um, and we did studies in different, um, uh, 00:08:10.040 --> 00:08:14.570 categories of the disease, different biomarker categories of the disease. 00:08:15.320 --> 00:08:21.200 Uh, so I designed, uh, some of these trials and meanwhile Laura and I keep 00:08:21.200 --> 00:08:31.160 talking about what we're going to do, uh, and, um, we both wanted to do, uh, what I 00:08:31.160 --> 00:08:39.470 called the bandit approach, uh, where you have multi-arm bandit, where you do lots 00:08:39.470 --> 00:08:43.640 of therapies and you try 'em on different patients and you see which patients work. 00:08:44.330 --> 00:08:48.140 And this, uh, you know, I, I had tried to do some of this 00:08:48.140 --> 00:08:50.330 back at, uh, MD Anderson, where. 00:08:51.290 --> 00:08:59.990 Uh, I was in am, uh, in the biostatistics department and we designed trials, uh, 00:09:00.110 --> 00:09:06.950 in various diseases and we, uh, Laos Push Eye, a faculty member there now at Yale. 00:09:07.640 --> 00:09:14.030 Uh, and I, uh, tried to do the, uh, this kind of thing where we would look at 00:09:14.030 --> 00:09:18.650 various therapies and go to pharmaceutical companies and try to sell it. 00:09:19.610 --> 00:09:31.010 And it was a hard sell, uh, to have, uh, therapies from Eli Lilly, from Pfizer, 00:09:31.070 --> 00:09:38.450 from Merck that you're comparing to a control, but in the same trial as you're 00:09:38.450 --> 00:09:40.730 comparing to each other, you know, like. 00:09:41.465 --> 00:09:45.245 I, because I had the data, I could look to see how the Eli 00:09:45.245 --> 00:09:47.855 Lilly was doing versus the Merck. 00:09:48.755 --> 00:09:52.655 Uh, and that's, uh, a, a dicey thing. 00:09:53.405 --> 00:09:55.265 So we failed. 00:09:55.295 --> 00:09:56.525 Lache and I failed. 00:09:57.365 --> 00:10:05.225 Um, uh, but we, in the context of the neoadjuvant approach, um, Laura 00:10:05.225 --> 00:10:09.515 and I mused about maybe we can do this in the neoadjuvant approach. 00:10:09.965 --> 00:10:15.755 And, uh, she is an amazing person. 00:10:16.415 --> 00:10:19.565 She, first of all, she never takes no for an answer. 00:10:20.195 --> 00:10:21.905 Uh, you say you can't do that, Laura. 00:10:21.935 --> 00:10:23.315 Oh, yes, we can. 00:10:24.635 --> 00:10:29.615 Uh, and so who's gonna fund this? 00:10:30.845 --> 00:10:40.025 Uh, we were told by various people, uh, Anna Barker is a good friend of ours, 00:10:40.025 --> 00:10:46.325 and she was once the deputy director of the NCI and she was the head of, 00:10:46.325 --> 00:10:52.265 um, uh, committee in something called the FNIH, the foundation for the, 00:10:52.895 --> 00:10:55.085 uh, national Institutes of Health. 00:10:55.655 --> 00:11:01.385 Uh, that's a, uh, that it works with, with pharmaceutical 00:11:01.385 --> 00:11:03.095 companies, including the government. 00:11:03.095 --> 00:11:05.525 So it works across this. 00:11:05.585 --> 00:11:12.365 And, uh, she said, you'll never get this approved by the NCI. 00:11:12.485 --> 00:11:14.015 The CLGB studies were. 00:11:14.300 --> 00:11:18.200 CLGB was funded by the NCI National Cancer Institute. 00:11:18.920 --> 00:11:25.820 Um, and we would have to get the, the, the funding from the NCI 00:11:25.820 --> 00:11:27.710 if we were to do this at CLGB. 00:11:28.610 --> 00:11:32.840 Um, and she told us, and others told us, and we knew actually 00:11:33.590 --> 00:11:34.940 this would never happen. 00:11:35.750 --> 00:11:41.960 Uh, I had personally spent years trying to get innovations into the NCI with 00:11:41.960 --> 00:11:47.990 a little bit of success, uh, but this would've been well beyond their, their, 00:11:48.050 --> 00:11:51.500 uh, ability to imagine this could happen. 00:11:52.520 --> 00:11:59.930 Uh, so we, we worked with the FDA, I remember the FDA, I remember, uh, Rick 00:11:59.930 --> 00:12:07.910 Paster, who's the, was still is the head of, uh, the oncology, uh, telling me. 00:12:08.450 --> 00:12:09.470 But these are 00:12:09.800 --> 00:12:11.330 early patients. 00:12:11.840 --> 00:12:13.040 These are. 00:12:14.240 --> 00:12:20.480 Curable and you're experimenting with them, how can you do that? 00:12:20.505 --> 00:12:20.625 Scott Berry: that? 00:12:21.680 --> 00:12:25.760 Don Berry: And so I said, uh, well, 00:12:26.060 --> 00:12:26.330 we're 00:12:26.330 --> 00:12:36.050 going to, uh, have a, a data monitoring committee made up of, uh, a couple of 00:12:36.050 --> 00:12:41.930 great statisticians and a couple of great clinicians, and they're gonna be meeting 00:12:41.930 --> 00:12:46.850 monthly and looking at the data, seeing how things are going, and making sure that 00:12:46.850 --> 00:12:49.940 nothing, and he was satisfied with that. 00:12:50.750 --> 00:12:56.420 So they allowed us to do this study, but then how to get the funding. 00:12:57.170 --> 00:13:05.570 And the funding was through the foundation for the NIH, uh, the initial funder. 00:13:06.110 --> 00:13:08.600 Um, but, uh, actually, Laura. 00:13:09.035 --> 00:13:11.915 Was responsible for getting most of the funding. 00:13:11.915 --> 00:13:18.005 You know, uh, she, she was a surgeon who still is, um, and would 00:13:18.125 --> 00:13:19.865 treat patients and they would, 00:13:20.555 --> 00:13:29.195 uh, tell her that, uh, uh, she made them, she, that they were alive because of her. 00:13:29.705 --> 00:13:36.755 And therefore, since I'm the CEO of this big corporation will fund you. 00:13:37.685 --> 00:13:39.965 Uh, she got lots of funding that way. 00:13:39.965 --> 00:13:47.045 It was originally funded by, um, uh, the donations and philanthropy. 00:13:47.855 --> 00:13:54.365 Uh, as time went along and as we got more and more, uh, companies 00:13:54.365 --> 00:14:00.755 involved with their drugs, uh, we passed the funding off onto them. 00:14:00.755 --> 00:14:04.205 So they had to, uh, pay to play. 00:14:05.780 --> 00:14:08.060 Scott Berry: So, so let's back up a little bit, but Sure. 00:14:08.120 --> 00:14:14.810 So you, I is, is I spy one funded or I Spy one? 00:14:14.810 --> 00:14:19.760 You, you described this as a bit of a pilot and from there you went out and 00:14:19.760 --> 00:14:25.430 were able to get this additional funding to bring in the first investigation arm. 00:14:26.255 --> 00:14:26.525 Don Berry: Uh, 00:14:26.915 --> 00:14:27.575 yes. 00:14:27.845 --> 00:14:30.065 Um, uh, exactly. 00:14:30.065 --> 00:14:32.945 I Spy one was funded by the government. 00:14:33.065 --> 00:14:35.585 It was funded by the NCI. 00:14:36.545 --> 00:14:39.665 Uh, and it, it was not randomized 00:14:40.025 --> 00:14:41.735 or adaptively randomized. 00:14:42.185 --> 00:14:46.985 It was looking at patients, uh, in the neoadjuvant setting because, you 00:14:46.985 --> 00:14:53.435 know, the surgeons had to learn how to do, uh, the, this, how this thing, 00:14:53.465 --> 00:14:57.845 you know, not do surgery right away, but then come in and do surgery later. 00:14:57.845 --> 00:14:59.705 It was a new thing to surgeons. 00:15:00.755 --> 00:15:00.785 I. 00:15:00.785 --> 00:15:06.140 And, um, so it, I spy one had two endpoints. 00:15:06.260 --> 00:15:10.340 Uh, I mean we, we were still looking at path pathologic, complete response, 00:15:10.340 --> 00:15:20.210 PCR, uh, but we were interested in could we predict PCR from an 00:15:20.210 --> 00:15:23.690 MRI that we give, uh, intermediate 00:15:23.960 --> 00:15:30.440 in time between the initial presentation and the initial therapies. 00:15:31.100 --> 00:15:35.720 Uh, you know, after, uh, three weeks, for example, on a 00:15:35.720 --> 00:15:37.820 therapy, maybe there's an effect. 00:15:38.525 --> 00:15:45.005 And maybe that effect could predict not only path cr, but also, uh, survival. 00:15:46.235 --> 00:15:48.815 So that's what I SPY one was about. 00:15:48.965 --> 00:15:56.196 It was, uh, uh, uh, kind of a registry if you like, or, uh, it, it, it 00:15:56.201 --> 00:16:00.035 wasn't, uh, a randomized, we weren't trying to learn about therapies. 00:16:00.035 --> 00:16:06.395 There was a specific therapy that patients got who, who met the eligibility criteria. 00:16:07.660 --> 00:16:10.055 Scott Berry: So, so interesting people listening to this. 00:16:10.295 --> 00:16:16.235 You've described the neoadjuvant as a huge advantage to clinical trials and 00:16:16.235 --> 00:16:18.605 the ability to learn about the treatment. 00:16:19.265 --> 00:16:22.235 What do we know now about it? 00:16:22.235 --> 00:16:23.885 For the patient benefit? 00:16:24.185 --> 00:16:29.195 Um, there's no sense that this is better or worse for the patient, 00:16:29.735 --> 00:16:33.635 that that goes through neoadjuvant as opposed to adjuvant care. 00:16:34.970 --> 00:16:38.270 Don Berry: Uh, boy, I'm, I'm gonna answer the question, but the only way I can 00:16:38.270 --> 00:16:42.230 answer the question is, uh, with, um, 00:16:45.170 --> 00:16:46.730 with, with hindsight. 00:16:47.480 --> 00:16:57.680 Uh, so at the time we thought that the benefit for the patient was this business 00:16:57.710 --> 00:17:02.270 about learning what benefits the tumor. 00:17:02.870 --> 00:17:09.500 When we would give therapies different experimental therapies, and we do 00:17:09.530 --> 00:17:15.260 the MRI, sometimes we looked after three weeks and the tumor was gone. 00:17:15.760 --> 00:17:16.050 Scott Berry: Yeah. 00:17:17.450 --> 00:17:24.140 Don Berry: Um, and that, I mean, is an obvious advantage to the patient 00:17:24.200 --> 00:17:26.900 that, uh, you can do something. 00:17:27.620 --> 00:17:28.160 Uh. 00:17:29.120 --> 00:17:33.080 Uh, else if it, if it hasn't gone, I mean, you could stop. 00:17:33.080 --> 00:17:36.680 If it's not defected at all, you might want to do something else. 00:17:37.190 --> 00:17:43.220 So it was getting the information that would help the individual patient, but 00:17:43.220 --> 00:17:47.180 it was also getting the information that would help in understanding which 00:17:47.180 --> 00:17:51.800 therapies are benning, benefiting which patients, and then we could, you know, 00:17:51.800 --> 00:17:58.700 emphasize those, those pairings, the right patient for the, or the right 00:17:58.700 --> 00:18:01.310 therapy for, for the individual patient. 00:18:02.330 --> 00:18:11.270 Um, and so it was this, uh, PCR that was the attraction for the clinical trialists 00:18:11.570 --> 00:18:14.270 that perhaps you could get approval. 00:18:15.140 --> 00:18:19.310 Um, and in fact, this has happened since perhaps you could 00:18:19.310 --> 00:18:21.260 get approval for a therapy. 00:18:21.920 --> 00:18:26.450 Uh, that had a, uh, a, a great benefit on the path. 00:18:26.450 --> 00:18:35.690 CR rate, you know, improves a 30% rate to 50%, uh, because patients who got 00:18:36.530 --> 00:18:39.800 complete responses did extremely well. 00:18:39.800 --> 00:18:44.360 And it didn't matter what the therapy was, it didn't matter what 00:18:44.540 --> 00:18:46.760 the biomarker characteristics were. 00:18:47.450 --> 00:18:52.970 If you got a, if you were disease free as far as they could tell at the time 00:18:52.970 --> 00:18:58.940 of surgery, uh, and then follow them, they did extremely well for overall 00:18:58.940 --> 00:19:01.790 survival and event-free survival. 00:19:02.360 --> 00:19:06.140 Now, I mentioned fast forward, uh, we're gonna have to go 00:19:06.140 --> 00:19:07.910 back to tell you about ipy two. 00:19:07.940 --> 00:19:09.080 You know, what it did. 00:19:09.530 --> 00:19:13.130 But, but, but fast forward with the neo adjunct approach. 00:19:13.580 --> 00:19:14.720 What happened? 00:19:15.380 --> 00:19:22.370 Was there are patients who come out of the surgery, and as I said, when 00:19:22.370 --> 00:19:24.770 they're Pat cr they do extremely well. 00:19:25.340 --> 00:19:29.810 But if they don't have a pat cr, if they still have residual 00:19:29.810 --> 00:19:32.240 disease, they did poorly. 00:19:33.680 --> 00:19:39.530 The pharmaceutical companies latched onto this possibility and said, 00:19:39.530 --> 00:19:47.450 maybe we should, uh, treat those patients who have residual disease. 00:19:48.020 --> 00:19:55.460 And to your point earlier, Scott, about the size of the trials, when we have, uh, 00:19:55.820 --> 00:20:03.470 residual disease, um, the, uh, event rate is a lot greater 00:20:03.740 --> 00:20:05.870 because they, you know, they recur. 00:20:06.590 --> 00:20:07.340 Um. 00:20:07.985 --> 00:20:14.525 Uh, much sooner than, and more likely than patients who get, uh, 00:20:14.585 --> 00:20:19.325 PCR patients, who gets get a PCR don't need additional therapy. 00:20:20.015 --> 00:20:25.085 Uh, and moreover, in a clinical trial, it would be difficult to see that 00:20:25.085 --> 00:20:26.615 there's a benefit for the therapy. 00:20:26.645 --> 00:20:27.875 'cause everybody lives, 00:20:28.755 --> 00:20:28.835 Scott Berry: Hmm. 00:20:29.165 --> 00:20:32.495 Don Berry: uh, not everybody, but, uh, most, most women. 00:20:33.095 --> 00:20:42.155 And, um, so they did this and they found, uh, depending on the, uh, 00:20:42.395 --> 00:20:47.435 biomarker characteristics that some therapies, immunotherapy for 00:20:47.435 --> 00:20:50.135 example, worked extremely well for 00:20:50.555 --> 00:20:54.095 patients who were so-called her two negative. 00:20:55.145 --> 00:21:03.905 Um, and the, um, the, the company would then go to the FDA and say, 00:21:03.905 --> 00:21:05.675 we've got this great therapy. 00:21:06.455 --> 00:21:11.135 Moreover, that's changing the course. 00:21:11.865 --> 00:21:12.155 Scott Berry: yeah. 00:21:12.155 --> 00:21:16.325 Don Berry: When, and, and, and, and it means that patient, you could, if 00:21:16.325 --> 00:21:19.835 you've got a pharmaceutical, if you're a pharmaceutical company and you take using 00:21:19.835 --> 00:21:25.445 the neoadjuvant approach and you randomize patients to get your therapy versus the 00:21:25.445 --> 00:21:32.915 standard therapy, and you saw that there's a PCR benefit, you're still not going 00:21:32.915 --> 00:21:41.315 to get approved by the FDA because what happens is the patients who don't have 00:21:41.315 --> 00:21:46.865 a benefit, who have residual disease are getting other therapy that's effective. 00:21:46.955 --> 00:21:47.525 Scott Berry: Yeah. 00:21:48.425 --> 00:21:50.285 Don Berry: So it's confounding. 00:21:51.125 --> 00:21:51.995 And so the neoadjuvant 00:21:52.835 --> 00:21:53.735 approach has 00:21:54.065 --> 00:22:01.745 completely changed, um, uh, breast cancer treatment and, uh, and approach. 00:22:02.435 --> 00:22:08.255 And it, it means that the original hope, it's sort of ironic, the 00:22:08.255 --> 00:22:10.835 original hope was to get a PCR. 00:22:11.855 --> 00:22:16.565 And now the, um, the, the benefit for the neoadjuvant approach 00:22:17.075 --> 00:22:18.785 is it really doesn't matter. 00:22:18.785 --> 00:22:21.755 You're just getting information about what therapy works and then 00:22:21.755 --> 00:22:28.475 you give some other therapy and they call it adjuvant after the surgery. 00:22:28.565 --> 00:22:28.785 Scott Berry: Mm. 00:22:29.615 --> 00:22:34.115 Don Berry: So that's a whole different story that's not really part of 00:22:34.115 --> 00:22:39.410 IFI two, but it's part of, uh, the breast cancer, uh, uh, story RA. 00:22:40.580 --> 00:22:41.000 Scott Berry: Okay. 00:22:41.000 --> 00:22:47.510 So, so I, in the timeline of this, I Spy one is proof of concept that 00:22:47.510 --> 00:22:51.080 you can do neoadjuvant, there's no investigational therapies. 00:22:51.350 --> 00:22:54.800 You can get MRIs, you're getting data on PCR. 00:22:55.250 --> 00:23:01.100 So meanwhile, if, if a company wants to do a phase two trial before I 00:23:01.100 --> 00:23:04.370 SPY two in the adjuvant setting. 00:23:04.775 --> 00:23:10.175 It would be incredibly hard because disease-free survival 00:23:10.175 --> 00:23:12.935 after surgery is so very long. 00:23:13.205 --> 00:23:15.665 These trials would be huge and long. 00:23:15.815 --> 00:23:20.375 It's hard to do drug development, so you turn it upside down 00:23:20.375 --> 00:23:21.995 and you're doing neoadjuvant. 00:23:22.415 --> 00:23:27.455 So you're ready for I SPY two, which is intervention in the 00:23:27.455 --> 00:23:31.055 neoadjuvant space and the funding. 00:23:31.055 --> 00:23:34.115 Are we ready to describe the I SPY two trial? 00:23:34.115 --> 00:23:34.205 Yes. 00:23:34.365 --> 00:23:34.585 Don Berry: Yes. 00:23:35.440 --> 00:23:35.730 Scott Berry: Okay. 00:23:36.095 --> 00:23:36.605 Okay. 00:23:37.085 --> 00:23:37.535 Okay. 00:23:37.955 --> 00:23:42.725 So, um, do you want to describe what this trial looks like? 00:23:42.725 --> 00:23:46.055 You, you want to do, you described you want to do the bandit approach, 00:23:46.055 --> 00:23:51.815 you want multiple therapies simultaneously, but this is also 00:23:51.815 --> 00:23:55.175 about precision medicine in a sense. 00:23:56.825 --> 00:23:57.515 Don Berry: Exactly. 00:23:59.105 --> 00:24:08.255 So the precision medicine aspect is, uh, breast cancer is a, probably the 00:24:08.315 --> 00:24:11.885 poster child of precision medicine. 00:24:12.665 --> 00:24:16.865 Um, it has biomarkers that are extremely important. 00:24:17.345 --> 00:24:22.745 Estrogen receptor status, progesterone receptor status, her two. 00:24:23.510 --> 00:24:31.280 Uh, status, um, of, uh, something called MammaPrint, or, which is very similar 00:24:31.280 --> 00:24:35.870 to Oncotype dx, which is a 21 gene. 00:24:36.440 --> 00:24:45.710 MammaPrint is a 70 gene, uh, uh, uh, biomarker multi, you know, poly marker. 00:24:46.550 --> 00:24:48.440 Um, uh, 00:24:49.185 --> 00:24:51.485 Scott Berry: So, so I-SPY 2 classifies. 00:24:51.505 --> 00:24:55.525 You described these, there's HER2 status, hormone receptor 00:24:55.625 --> 00:24:57.405 status and MammaPrint status. 00:24:58.310 --> 00:25:01.970 Each one is dichotomous, uh, for this. 00:25:02.360 --> 00:25:11.300 So women who come into I-SPY 2 fit into a subgroup, and they're, or a subtype, 00:25:11.300 --> 00:25:16.610 sorry, I SPY two created all kinds of new terminology, and there are eight 00:25:16.610 --> 00:25:22.580 subtypes by the classification of that when a woman comes in that their, 00:25:22.580 --> 00:25:29.390 their, their, uh, tumors classified in eight, eight different subtypes. 00:25:30.845 --> 00:25:32.015 Don Berry: Uh, right. 00:25:32.885 --> 00:25:40.385 And then we want to see which subtypes benefit from which therapies. 00:25:40.745 --> 00:25:47.190 But now if you, uh, we will, what we want to do is, is, uh, tell the company. 00:25:48.500 --> 00:25:53.360 Your therapy is beneficial in this, the subtype, but not in that subtype. 00:25:54.020 --> 00:25:55.550 Now, how can you do that? 00:25:55.580 --> 00:26:01.820 There are, if there are eight subtypes, there are 255 different combinations 00:26:02.270 --> 00:26:11.120 of those subtypes and we can't do 255 different partitions of of cancer. 00:26:11.120 --> 00:26:17.570 So we looked at what we call signatures, which are subsets 00:26:17.570 --> 00:26:24.080 of subtypes, and we looked at 10 that were the primary analysis. 00:26:24.080 --> 00:26:31.700 So every month we analyzed how well the various therapies in the trial were 00:26:31.700 --> 00:26:36.290 doing in these 10 different signatures. 00:26:36.770 --> 00:26:44.570 So one signature, uh, would be HER two positive, that as you described 00:26:44.570 --> 00:26:46.550 it, Scott has four subtypes. 00:26:48.170 --> 00:26:53.570 Uh, but we're looking not at those individual subtypes, but at the, 00:26:53.570 --> 00:26:56.120 at the, uh, all of them together. 00:26:57.260 --> 00:26:59.510 Uh, we also look at the HER two negative. 00:26:59.510 --> 00:27:02.480 We look at the estrogen receptor positive, the negative. 00:27:02.870 --> 00:27:10.490 We look at the, uh, estrogen receptor negative, HER two negative, which 00:27:10.490 --> 00:27:12.680 is called triple negative disease. 00:27:13.100 --> 00:27:18.410 And that has two subtypes, namely MammaPrint positive, MammaPrint 00:27:18.410 --> 00:27:21.590 negative, um, that we don't, 00:27:22.160 --> 00:27:28.550 we're not advertising those subtypes as being an indication for your drug. 00:27:28.940 --> 00:27:32.120 It's the signatures which are indications for your drug. 00:27:33.170 --> 00:27:41.510 So that was, uh, uh, a, a big deal to go to the FDA and say, we have these, 00:27:42.410 --> 00:27:46.010 uh, 10 different possible indications. 00:27:47.165 --> 00:27:54.155 Uh, and that in itself is a huge innovation, which by the way, uh, I SPY 00:27:54.155 --> 00:28:01.115 two is a phase two trial, and we're, we're trying to ready it for phase three. 00:28:01.685 --> 00:28:09.095 Uh, we've since then designed similar trials in, uh, GBM glioblastoma, uh, 00:28:09.185 --> 00:28:13.865 pancreatic cancer, uh, that have a smaller number of subtypes because they 00:28:13.865 --> 00:28:19.235 have, you know, different biomarker, uh, uh, science associated with them. 00:28:20.105 --> 00:28:27.545 Um, and, but that the FDA has agreed with looking at these multiple 00:28:27.545 --> 00:28:31.325 signatures in a registration trial. 00:28:32.195 --> 00:28:37.985 Uh, so it's, it's, it's pie in the sky, but we're eating the pie. 00:28:38.570 --> 00:28:38.960 Scott Berry: mm. 00:28:39.440 --> 00:28:44.720 So, uh, a woman comes in and she belongs to one of these eight subtypes and is 00:28:44.720 --> 00:28:54.770 randomized among control and, uh, call it 20% chance she goes to control. 00:28:55.010 --> 00:29:00.620 And the other 80% she could go to the various investigational 00:29:00.620 --> 00:29:02.450 therapies that are there at a time. 00:29:02.480 --> 00:29:05.540 So this is your randomized bandits. 00:29:05.930 --> 00:29:10.760 In a way that if there are three therapies in the trial right now, she 00:29:10.760 --> 00:29:13.490 gets randomized among the three therapies. 00:29:14.120 --> 00:29:18.770 And so, uh, you probably need to describe the response, adaptive 00:29:18.770 --> 00:29:22.640 randomization aspect and which is so critical to icey two. 00:29:23.840 --> 00:29:24.050 Don Berry: Yeah. 00:29:24.110 --> 00:29:24.620 So, 00:29:24.920 --> 00:29:32.480 um, the bandit problem is that you have these, uh, multi arms and, and, uh, it's 00:29:32.480 --> 00:29:36.890 usually posed in the context of a single. 00:29:37.580 --> 00:29:48.560 Uh, subtype, um, where you want to treat patients effectively, uh, in the 00:29:48.560 --> 00:29:55.040 trial, which is, uh, you know, trying to blow, I've been trying to blow up 00:29:55.100 --> 00:30:00.230 the, uh, notion that you can't learn. 00:30:00.230 --> 00:30:05.240 You, you, you, you're not treating patients in a clinical trial. 00:30:05.660 --> 00:30:07.850 You're learning about patients. 00:30:08.450 --> 00:30:10.340 And I say, why can't we do both? 00:30:11.210 --> 00:30:13.370 Um, and we had done that. 00:30:13.580 --> 00:30:17.900 Um, uh, I've been writing about it for many years. 00:30:18.500 --> 00:30:22.910 Uh, and, and when I went to MD Anderson, we actually started to do it. 00:30:23.600 --> 00:30:28.280 So we had trials where we, uh, adaptively, randomized. 00:30:29.210 --> 00:30:36.590 Uh, you get a, a randomization, but if a, if that therapy is doing well for your 00:30:36.590 --> 00:30:43.250 subtype, if you're the patient, you get that therapy with a higher probability. 00:30:44.780 --> 00:30:49.100 And what it means is that, uh, patients in the trial, 00:30:49.305 --> 00:30:49.745 Scott Berry: the trial, 00:30:50.420 --> 00:30:51.080 Don Berry: um, 00:30:51.265 --> 00:30:51.485 Scott Berry: um, 00:30:52.190 --> 00:30:55.850 Don Berry: on average have a higher, 00:30:56.450 --> 00:30:58.190 uh, PA PCR rate 00:30:58.910 --> 00:30:59.570 than 00:31:00.500 --> 00:31:02.360 uh, a, a standard trial 00:31:02.630 --> 00:31:06.350 because they're more likely to get a therapy that's doing well. 00:31:06.680 --> 00:31:12.200 And moreover, you're not exposing those patients to a therapy that's 00:31:12.200 --> 00:31:14.240 not doing well for your subtype. 00:31:15.590 --> 00:31:21.320 Um, and it it means that patients in, in the trial do better. 00:31:22.130 --> 00:31:27.170 Um, it also, uh, means that the, uh, overall patient. 00:31:27.905 --> 00:31:29.855 Uh, population does better. 00:31:30.545 --> 00:31:31.085 Um, 00:31:31.415 --> 00:31:39.065 and you get more information about the therapies that are doing well 00:31:39.125 --> 00:31:46.325 in the subtype signature because they're getting it more often. 00:31:46.475 --> 00:31:48.815 So you get a bigger sample size. 00:31:49.715 --> 00:31:58.115 Um, because in, in and faster you learn faster, you learn better about 00:31:58.115 --> 00:31:59.855 where you really wanted to learn. 00:31:59.855 --> 00:32:04.775 You don't care how well it does in a therapy, in, in a subtype that 00:32:05.075 --> 00:32:06.515 doesn't do very well with it. 00:32:07.295 --> 00:32:10.775 Uh, you don't use that for, for those patients. 00:32:11.135 --> 00:32:14.285 You use it only for those patients who it is doing well. 00:32:15.285 --> 00:32:15.575 Scott Berry: okay. 00:32:15.680 --> 00:32:20.240 So, uh, when women come in, they're classified in their eight groups and 00:32:20.240 --> 00:32:26.300 the randomization, uh, probabilities for one subtype are different 00:32:26.300 --> 00:32:30.770 than another because the drugs are modeled as potentially having 00:32:30.770 --> 00:32:32.930 differential effect for those women. 00:32:33.020 --> 00:32:33.620 Don Berry: Exactly. 00:32:33.860 --> 00:32:34.670 Scott Berry: And it might 00:32:34.670 --> 00:32:39.110 even be that they have no chance to get a particular therapy 00:32:39.110 --> 00:32:40.250 'cause it's not doing well. 00:32:40.250 --> 00:32:44.210 And a very high chance to get another one that's doing well for women like them 00:32:44.930 --> 00:32:47.270 from, from, the drug side. 00:32:47.690 --> 00:32:52.490 The drug can focus on the women where it's having an effect and not 00:32:52.490 --> 00:32:54.800 randomize to those, that they're not. 00:32:55.550 --> 00:32:59.150 What's amazing about this is you can't really do this. 00:32:59.240 --> 00:33:06.080 You can't do it very well if it's a single drug trial, because now you can't change 00:33:06.080 --> 00:33:10.820 the prevalence of various subsets without stopping and rolling them all together. 00:33:11.180 --> 00:33:17.060 It's almost undoable to do this precision unless you've got multiple things to give. 00:33:18.950 --> 00:33:22.340 Now, now you're doing these, I I, I know you wanna jump at this, 00:33:22.340 --> 00:33:26.990 but you're doing these adaptations and you're analyzing the data. 00:33:27.290 --> 00:33:30.410 Uh, the algorithms are being run and resetting. 00:33:30.410 --> 00:33:34.070 The randomization weekly during the, i I think it varied 00:33:34.070 --> 00:33:35.660 during the time of Ipy two. 00:33:35.810 --> 00:33:38.360 I think initially it was actually daily, then weekly. 00:33:38.960 --> 00:33:42.020 Now your endpoint is six months. 00:33:43.055 --> 00:33:48.905 And you're trying to learn about what works and you don't wanna wait six months. 00:33:48.905 --> 00:33:52.625 And so you're trying to accelerate the learning to allow all of these 00:33:52.625 --> 00:33:54.635 things you just described to do better. 00:33:54.785 --> 00:33:56.075 So how do you do that? 00:33:58.160 --> 00:34:00.080 Don Berry: Uh, thanks for the setup. 00:34:00.590 --> 00:34:06.680 Uh, uh, I mentioned earlier the MRI, we did MRIs in I spy one to learn 00:34:06.680 --> 00:34:11.900 about that and we, we learned that it is predictive what we di like to do. 00:34:12.485 --> 00:34:15.425 Is to use all of the patients. 00:34:15.905 --> 00:34:17.765 Uh, we don't wanna wait six months. 00:34:18.635 --> 00:34:23.975 Um, in the context of waiting for disease-free survival or overall 00:34:23.975 --> 00:34:26.255 survival, six months is a short time. 00:34:26.915 --> 00:34:32.075 But in the context of, uh, uh, I spy two, six months is a long time. 00:34:33.065 --> 00:34:41.105 Um, we get MRIs, we got MRIs, uh, MRIs at three weeks and at, uh, 12 weeks. 00:34:41.705 --> 00:34:47.045 Uh, and we use that information to predict is it gonna be a PCR. 00:34:47.795 --> 00:34:54.815 Now here is where, uh, I reported to the, uh, data Safety Monitoring Board. 00:34:55.605 --> 00:34:56.535 Uh, every month. 00:34:57.225 --> 00:35:02.865 Um, and, uh, uh, I explained to them the trial and they 00:35:02.865 --> 00:35:04.485 kept learning about the trial. 00:35:05.025 --> 00:35:10.755 And after some period of time, uh, they said to me, said, Don, 00:35:10.815 --> 00:35:18.075 you told us about MRI and how you use MRI, uh, at least eight times. 00:35:18.525 --> 00:35:19.935 Could you tell us again? 00:35:20.715 --> 00:35:25.575 So it's not an easy concept for those people that are used to, 00:35:26.055 --> 00:35:29.595 you know, looking at an endpoint and focusing on the endpoint. 00:35:29.595 --> 00:35:30.195 If you say 00:35:30.585 --> 00:35:35.985 we're gonna, we focus on the endpoint, being MRI, uh, first of 00:35:35.985 --> 00:35:38.175 all, nobody would, uh, accept that. 00:35:38.175 --> 00:35:41.385 It's not a surrogate for, uh, anything. 00:35:41.775 --> 00:35:43.065 But we looked at MRI, 00:35:43.905 --> 00:35:47.085 not with the notion that what we're gonna do is. 00:35:47.980 --> 00:35:57.095 Uh, uh, focus on MRI as the endpoint, but as an auxiliary endpoint, as a marker of 00:35:57.125 --> 00:35:59.945 how likely is it that it's gonna be a PCR? 00:36:00.365 --> 00:36:01.115 So it, 00:36:01.805 --> 00:36:02.585 if 00:36:02.615 --> 00:36:09.065 if, there's no tumor, uh, that you can see on the MRI, that means there's a 00:36:09.065 --> 00:36:11.405 high probability is gonna be a PCR. 00:36:12.095 --> 00:36:15.305 So what we do is we do multiple imputation. 00:36:15.965 --> 00:36:21.215 We have this probability, and when we are doing the multiple imputation for all of 00:36:21.215 --> 00:36:24.605 the women in the trial who don't have. 00:36:24.965 --> 00:36:29.585 Six months, you know, who don't have the result of surgery. 00:36:29.975 --> 00:36:33.545 We're predicting the result of surgery, but it's probabilistic. 00:36:34.595 --> 00:36:41.945 So, um, uh, there was a drug pembrolizumab, which you've seen 00:36:41.975 --> 00:36:50.105 advertisements for, for Keytruda, uh, from Merck, uh, which, uh, graduated. 00:36:50.105 --> 00:36:54.935 We call it graduate when it, we've learned what it's, what it's, 00:36:54.965 --> 00:37:04.955 uh, uh, signature is, uh, when no patient had results at six months. 00:37:06.425 --> 00:37:12.155 Um, and the, the way we could do that, 00:37:12.455 --> 00:37:16.655 we will talk about the time machine, uh, in a minute, I'm sure. 00:37:17.105 --> 00:37:21.035 Uh, the, the, the way we could do that is. 00:37:21.860 --> 00:37:27.200 Uh, we had the algorithm that was making the patient assignment and 00:37:27.200 --> 00:37:29.540 making decisions about graduation. 00:37:29.810 --> 00:37:35.600 The drug would graduate, uh, no longer get, uh, patients, uh, but we would 00:37:35.600 --> 00:37:44.450 continue follow up, uh, when only one patient had the result of surgery 00:37:45.260 --> 00:37:47.120 in triple negative breast cancer. 00:37:47.960 --> 00:37:51.050 And, uh, how could we do that? 00:37:51.320 --> 00:37:55.820 It was because the algorithm was seeing that the therapy 00:37:56.060 --> 00:37:58.280 was melting the tumor away. 00:37:58.550 --> 00:37:59.690 And I went in and I 00:38:00.110 --> 00:38:02.390 looked at the data, I said, how can it do this? 00:38:02.930 --> 00:38:09.860 And the answer was, you know, in the 12 patients who had, uh, results at 12 weeks. 00:38:10.715 --> 00:38:20.285 Um, 11 of them eventually had a PCR, uh, and 11 of them, uh, 00:38:20.345 --> 00:38:25.205 were, uh, and it was clear that they were gonna do extremely well. 00:38:26.285 --> 00:38:28.145 Um, so 00:38:29.190 --> 00:38:32.765 Scott Berry: so, so, and at the time I think one patient had been through 00:38:32.765 --> 00:38:40.745 six months and the model predicted something like a 63% PCR rate for 00:38:40.745 --> 00:38:45.275 the arm and, uh, with one patient. 00:38:45.545 --> 00:38:50.165 And by the time all the patients got through, it stopped for 00:38:50.255 --> 00:38:52.895 graduation, uh, at some point. 00:38:53.480 --> 00:38:58.585 It almost nailed exactly the PCR rate based on the MRIs. 00:38:58.835 --> 00:39:02.045 Don Berry: it was exactly the sa the, the same as the, 00:39:02.375 --> 00:39:04.325 as the MRI, but now. 00:39:05.030 --> 00:39:07.940 Um, there's another aspect to this. 00:39:07.940 --> 00:39:13.340 How could it do this 60% and get it right with only one patient? 00:39:14.090 --> 00:39:17.570 Uh, the answer is the time machine and the controls. 00:39:18.290 --> 00:39:18.620 Now, you, 00:39:18.620 --> 00:39:19.640 Scott Berry: so hang on, hang on. 00:39:19.640 --> 00:39:19.970 Let's, 00:39:19.970 --> 00:39:25.040 let's, we're gonna be adaptive here, so for the first time, we're gonna call this 00:39:25.040 --> 00:39:31.940 part A of the podcast and we're gonna make people have to tune in for part B of the 00:39:31.940 --> 00:39:36.170 podcast and learn about the time machine. 00:39:36.320 --> 00:39:40.760 And thank you all for joining and join our next episode and 00:39:40.760 --> 00:39:42.320 learn about the time machine 00:39:42.365 --> 00:39:45.575 Don Berry: And, and other things as well, predictive probabilities, 00:39:45.575 --> 00:39:49.380 for example, which are usually important in, uh, uh, being 00:39:49.660 --> 00:39:49.780 adaptive. 00:39:50.925 --> 00:39:53.720 Scott Berry: Alright, so thank you. 00:39:53.830 --> 00:39:54.300 Don Berry: Thank you.