The Peter Attia Drive - #409 ‒ Inside modern drug development： the science, economics, and regulatory hurdles behind bringing new medicines to patients ｜ Lloyd Klickstein, M.D., Ph.D.

00:00:00.000 Hey, everyone. Welcome to The Drive Podcast. I'm your host, Peter Atiyah. This podcast,

00:00:16.540 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:21.520 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:26.720 wellness, and we've established a great team of analysts to make this happen. It is extremely

00:00:31.660 important to me to provide all of this content without relying on paid ads. To do this, our work

00:00:36.960 is made entirely possible by our members, and in return, we offer exclusive member-only content and

00:00:42.980 benefits above and beyond what is available for free. If you want to take your knowledge of this

00:00:48.100 space to the next level, it's our goal to ensure members get back much more than the price of the

00:00:53.420 subscription. If you want to learn more about the benefits of our premium membership, head over to

00:00:58.600 peteratiamd.com forward slash subscribe. My guest this week is Dr. Lloyd Clickstein, a physician,

00:01:07.260 scientist, rheumatologist, and drug developer who has spent more than two decades helping

00:01:11.600 discover and develop new medicines. After beginning his career as a physician scientist at Harvard

00:01:16.800 Medical School and the Brigham and Women's Hospital, Lloyd joined Novartis, where he helped

00:01:21.800 pioneer translational medicine and led the company's new indication discovery unit,

00:01:26.660 identifying entirely new disease and therapeutic opportunities. He later held leadership roles at

00:01:32.760 several biotechnology companies, including Versantis Bio, where he led the development

00:01:37.380 of BEMA before the company was acquired by Eli Lilly. Today, he serves as the CEO of Coslap

00:01:45.240 Therapeutics. I wanted to have Lloyd on because very few people have had a front row seat to

00:01:50.140 every stage of modern drug development, from identifying the unmet medical need, to the

00:01:55.820 discovery of new therapeutic targets, to navigating clinical trials, regulatory approval, and even

00:02:01.840 commercialization. While we use BEMA as a case study throughout this conversation, the real goal

00:02:07.880 is to pull back the curtain on how new medicines are actually created, why the process takes so

00:02:13.400 long and costs so much, and how scientists decide which areas are worth pursuing in the first place.

00:02:18.920 In the episode, we talk about how new drugs are discovered, developed, and ultimately brought to patients, the science and economics behind choosing which diseases and therapeutic targets to pursue, the differences between small molecules, biologics, gene therapy, and other drug platforms, why drug development takes so long, costs billions of dollars, and so often fails.

00:02:40.460 The story behind this one particular drug from its discovery at Novartis to its development as a therapy for muscle loss and obesity, what clinical trials, FDA approval, and patent protection actually involve, and how the next generation of obesity and muscle-preserving therapies may reshape the treatment of metabolic disease.

00:03:02.560 So without further delay, please enjoy my conversation with Dr. Lloyd Klickstein.

00:03:10.460 Lloyd, thanks so much for coming out to Austin. Great to see you again in person. It's been

00:03:16.520 probably six, seven, maybe eight years since we were last in person.

00:03:20.720 It might well be that long, yes.

00:03:22.880 So look, for folks who don't know you as well as some of us do, tell us just a little bit about

00:03:30.800 your background. You're a physician and a scientist, but talk a little bit about how

00:03:35.140 those two paths came together. Right. So my original plan when I finished college was to go

00:03:42.800 to medical school, following is basically the family business. But between college and medical

00:03:47.660 school, I worked in a laboratory at Brigham and Women's Hospital in Boston and found two things.

00:03:53.420 One, that I love doing science and two, that I was pretty good at it. And I ended up doing an MD

00:03:59.980 PhD degree. And that is the training, not required, but helpful for becoming a physician

00:04:09.540 scientist. And from there, I did medicine training. I was a rheumatology training,

00:04:16.560 and I practiced rheumatology for maybe 10 years and also had an NIH-funded research lab

00:04:23.220 doing very basic science on adhesion molecules in the immune system.

00:04:27.640 And that was also at the Brigham? Did you stay there after you finished your training?

00:04:31.000 I have a high activation energy for moving and doing other things. So I stayed there my whole academic career.

00:04:38.560 Yeah, that's right. Medical school, MD, PhD, the whole thing.

00:04:40.560 The only job I ever had until I left and went to industry, which was a little over 20 years ago.

00:04:48.020 and I went to Novartis Institutes, which was founded by Mark Fishman when he was recruited

00:04:55.560 by the then CEO of Novartis, Dan DeSella, to reimagine how research and early clinical

00:05:02.280 development are done in industry. And they basically brought the whole concept of translational

00:05:08.820 medicine to industry. Now, that term, I think, was coined at Oxford, somewhere in the UK,

00:05:15.880 for sure. And this was the first industry manifestation of it.

00:05:21.600 So Lloyd, there are a lot of specific drugs and indications, pathways and targets that I want to

00:05:28.320 talk about today. But I think before I go there, there's a black box that exists that is the world

00:05:36.220 you occupy that I really think it's unknown by the public, but I think it would be very

00:05:41.780 insightful if people understood it. And it is the process by which a drug is discovered.

00:05:47.800 The first person who I ever read articulating this was Steve Rosenberg, wrote one of my favorite

00:05:53.320 books about science called The Transformed Cell. This book is over 30 years old. I probably read

00:05:59.220 it cover to cover 10 times during the course of my life. But he starts with a question, which was,

00:06:05.380 hey, he's at a party one day and someone says, hey, how do you discover new drugs for cancer?

00:06:08.780 Do you just go into the kitchen sink and kind of grab things and experiment?

00:06:12.320 So, again, it's a very intelligent person can still have an enormous blind spot to what it is people like you do.

00:06:20.840 So take that in any way you want, but I think it might be a great foundation.

00:06:24.720 Let's start from a very broad area of medicine, not just focus on cancer, because cancer is a special case of a more general concept.

00:06:33.940 So I personally, and this may be different for different people, but I personally start

00:06:38.740 with the patients and the clinical indications.

00:06:41.960 And essentially, you're looking for what's not there.

00:06:44.680 So you're looking for a drug that doesn't exist or a therapy that doesn't exist but

00:06:50.020 is needed.

00:06:51.220 And then there are two broad categories.

00:06:53.160 There are incremental improvements, and then there are quantum steps in the concept of

00:06:57.900 drug discovery.

00:06:59.180 Incremental improvements are making a drug that you have to take less frequently, or

00:07:03.920 as we've seen a lot in the press recently, an oral drug instead of an injectable drug.

00:07:08.580 Or the same kind of drug, but works better.

00:07:12.340 And there have been great examples of that.

00:07:14.480 Think atorvastatin and rosuvastatin in a mevacorzocor world.

00:07:19.700 And those have been enormous commercial successes.

00:07:21.740 and there's a strong bias throughout the drug development infrastructure to do those kinds of

00:07:28.780 things because they're relatively low risk and there's not so much uncertainty. On the other

00:07:36.280 hand, to see indications that may not even have been described yet and don't have ICD-10 or 11

00:07:42.240 codes and have to deal with the whole complexity of making physicians and patients and payers and

00:07:51.380 everybody else aware of this and then creating a regulatory path for it, it's a heavier lift.

00:07:59.620 But I think that's where the biggest value comes to society and individuals and patients

00:08:05.380 in creating new therapies. And that's kind of what I do. And there's a lot of failure involved

00:08:12.940 in there. And the things that we can talk about are how do you get a new regulatory pathway

00:08:20.020 established. Been part of that once, soup to nuts. And then how do you sort of create new indications?

00:08:28.240 That's also a challenge. Maybe it would be informative to think about how my team and I

00:08:36.660 started something called the New Indication Discovery Unit at Novartis. So Mark Fishman,

00:08:42.240 who was the president of Novartis Research at the time, had a few of us come into his office and

00:08:47.300 said, we need to make the most needed medicines and find out what we're not doing that we should

00:08:54.160 be doing and get it started. And we had a budget for that. So super. As I mentioned, I start with

00:09:01.700 patients and indications and medical need. So starting there, we made a list of about 7,000

00:09:09.740 clinical indications that were unmet. All clinical indications. We just made a list of all the

00:09:15.900 clinical indications. It didn't have everything there. It only had the ones that people had

00:09:20.080 recognized at the time. But it gave us a framework in which we could start thinking.

00:09:26.020 And then it sounds like a lot, but surprisingly it isn't. It's maybe 10 or 20 pages of paper

00:09:32.480 at the time. And we grouped them into things that we were already working on. Things that were

00:09:40.560 rare genetic developmental things that would be difficult to approach. And the remainder fell

00:09:49.040 into maybe seven or eight different buckets. And then we started working on those. One of them was

00:09:54.720 healthy aging, and one of them was ENT, and one of them at the time was renal diseases.

00:10:00.420 And we should talk about that some more because that's the regulatory endpoint.

00:10:05.320 Novartis wasn't at the time working on liver diseases. One of them was those,

00:10:08.760 sort of fibrotic diseases in general. And there were some others. And we started our program and

00:10:15.420 it eventually grew to have dozens of projects and became a bit of a monster. And then we had to sort

00:10:21.240 of whittle it down a little. But some really interesting studies came out of that. One of

00:10:28.200 them, which I know we're going to get to later, is the Magromab for muscle diseases. It's a long

00:10:33.920 story there, but again, it came out of a medical need of frail elderly people. And I was very

00:10:40.820 mindful as a rheumatologist, I saw some of these people. If you, at the time, I haven't looked at

00:10:46.940 the data recently, but at the time, if a patient had to go to a nursing home, not wanted to go or

00:10:54.400 whatever, but had to go, the three-year mortality rate approached 90%. So being a frail elderly 0.96

00:11:01.680 person who has to go to a nursing home was worse than cancer. And there are a lot of serious 0.78

00:11:07.520 diseases in medicine that are worse than cancer in terms of clinical outcomes, but we don't treat

00:11:12.660 in the same way. And I think we should, but so we, we really embraced the concept of being a frail

00:11:18.260 elderly person and what could we do to number one, treat it, but then prevent it down the road. 0.99

00:11:24.020 Yeah. I didn't know it was that. I didn't know the three-year mortality for frailty 0.99

00:11:27.760 in that specific indication was so high, but that's a staggering statistic.

00:11:32.680 It was terrible. Again, it might be a little better now, but it's not good.

00:11:36.620 Yeah. Okay. So let's now talk about that next leap. So let's assume you or the scientists and

00:11:43.840 the team have decided we have an indication. There's a target we want to go for, an unmet

00:11:49.400 clinical need, and there's not an incremental opportunity, right? The example you gave is a

00:11:55.220 great one. I have six statins already. I'm going to come up with the seventh. Let's take that one

00:12:01.020 off the table. How do you begin the thinking around what we're going to do? And maybe, by the

00:12:09.020 way, for the listener, can you explain some of the different classes of molecules? I also think we

00:12:14.540 talk very quickly and casually about monoclonal antibodies, small molecules, biologics, but I

00:12:21.140 I think that that nomenclature might not be clear to everybody and maybe provide a little bit of

00:12:26.140 that as well. Sure. So let's start with what we call small molecules, which is industry jargon.

00:12:32.620 It basically means a chemical. Historically, the companies that became our biggest drug companies

00:12:41.420 started over a hundred years ago as dye companies because the chemistry is very similar for making

00:12:49.440 a dye is for making a drug. So these are just chemicals. The other class are biologicals,

00:12:57.760 which is pretty much everything that's not a chemical. Within that broader group of biologicals,

00:13:05.480 and actually there's a third group, let's say devices. So for biologicals, that could be

00:13:12.780 an antibody. It could be some other protein like a peptide or a soluble receptor. And

00:13:21.300 I think we're now creating a separate category of gene therapies,

00:13:26.420 which themselves can be complex based on how they're delivered or targeted.

00:13:31.560 And then let's talk about devices because that's a completely different animal and it's regulated

00:13:42.080 by its own group called CDRH in the FDA. And that could include things like a gadget you make in a

00:13:51.880 workshop. It could be an app. It could be something simple like a syringe that you use to inject a

00:14:01.140 drug or an auto-injector, which are very common nowadays, or some combination of those things.

00:14:06.880 And include all the way up to implantable?

00:14:09.140 Yes, absolutely.

00:14:10.100 And we don't have to go into it, but are there differences in the IP treatment of small

00:14:16.380 molecules and biologics? Are there longer patent lives or anything like that?

00:14:21.100 So patent law, I think, is the same for anything, but there are other regulatory conditions that

00:14:29.600 apply to one kind of treatment or another. For example, there are different exclusivity periods

00:14:36.920 for a small molecule versus a biologic, and they change periodically. And all of this is something

00:14:43.420 that's considered as you're working on how do you protect the drug you're making. This is a really

00:14:51.040 important point, and it's very topical now with the concern about expensive new drugs and how do

00:14:58.700 we make them available for people. And I don't think it's well understood or adequately understood

00:15:05.560 how patent law works about drugs.

00:15:08.240 Maybe we could take a few minutes and talk about that.

00:15:10.920 I'm going to preface my remarks by saying I'm not a lawyer.

00:15:14.040 I don't play one on TV, and I wouldn't if I were asked to.

00:15:19.060 But you've been to the rodeo many, many times.

00:15:22.740 Yes. 0.94

00:15:23.660 I was the clown.

00:15:26.340 So the way this works is if you think about, at a very high level,

00:15:31.540 Making a new drug that's used by many people takes longer, is more expensive, and takes as many people as building the biggest skyscraper in the world.

00:15:49.600 Think, you know, the Burj Khalifa.

00:15:53.560 Just think about that for a minute.

00:15:55.220 That's thousands of people, many years, more than a billion dollars.

00:16:01.960 And unlike a building which can ultimately pay for itself and pay all the bondholders

00:16:09.520 and provide a return to the investors over decades,

00:16:14.680 patent law gives a very limited term in which all of the investment and potential profit can be recovered.

00:16:23.480 And at the end of the drug's patent life, it is freely available for anybody to make for the rest of eternity.

00:16:33.540 That's the deal you make with getting a patent.

00:16:36.520 A patent is essentially a monopoly on being able to make, use, and sell the drug in exchange for telling everybody how to do it.

00:16:46.860 That's basically what a patent is.

00:16:48.800 Now, the term of the patent from the time you file it is 20 years.

00:16:53.480 and there are a few little things about extending it for a small further period based on how much

00:16:59.120 time it took to work on it. Practically, you get 10 to 15 years of exclusivity from the time it can

00:17:05.680 be launched. So a couple of points I'll add just for you to expand on if you like. Some people

00:17:11.980 might also be aware of the idea that not everything has to be patented. So for example, the classic

00:17:16.840 example is Coca-Cola. Had they patented the formula for Coca-Cola hundreds of years or I

00:17:22.740 I don't remember when Coke started, but call it 150 years ago or something like that.

00:17:26.540 We wouldn't be drinking the same thing today.

00:17:28.440 So they chose a different route, which is we are never going to make this secret public.

00:17:33.420 And in exchange for that, we will have no protection.

00:17:36.460 That's correct.

00:17:37.180 So you're describing a trade secret.

00:17:39.460 Correct.

00:17:39.900 That's another form of intellectual property.

00:17:42.420 Now, I assume that is not an option in pharma.

00:17:46.440 Well, surprisingly, it is.

00:17:48.060 Interesting.

00:17:48.320 There are some specific drugs where they were protected by trade secrets.

00:17:52.740 So a couple of my favorite examples are Armour Thyroid, which was a thyroid place.

00:17:59.900 Desiccated thyroid hormone, yeah.

00:18:01.380 Yep.

00:18:01.680 It was the thyroid hormone for people who needed replacement therapy, and it was before we could make it synthetically.

00:18:08.440 And the process by which that was made was kept secret.

00:18:12.140 Now, it didn't stop other people from trying, but they had to copy exactly the composition of all the peptides as well as the impurities in the final preparation in order to be able to use the clinical and filing package that Armort used at the FDA to get FDA approval.

00:18:32.760 Turned out that was technically really hard.

00:18:35.880 Another one I really like is Acthar Gel.

00:18:38.120 So this was purified from pig pituitaries, and it was ACTH, basically.

00:18:45.400 That's the ACTH, A-C-T-H-A-R, gel.

00:18:48.600 I used to love it.

00:18:49.760 I used it in the emergency room a lot because it was a rheumatology smart missile.

00:18:54.740 Patients could come in with acute gout and be miserable.

00:18:57.980 You give them one shot of that, and it gives them endogenous steroid taper over a period of several days.

00:19:04.060 It was great.

00:19:05.820 Totally unaware of that, I see.

00:19:07.180 Yeah. It was a funny story behind that one too. It was cheap and widely available

00:19:13.900 until the BSE scare, the bovine spongiform encephalopathy scare happened. And then because

00:19:20.480 it was purified from pig pituitaries, the company was worried that it was going to be,

00:19:25.300 there's going to be something similar in pigs. And we know pigs have endogenous viruses.

00:19:29.220 So it was pulled off the market. And then it was bought by a very small company

00:19:33.720 who eventually re-commercialized it for infantile seizures and they jacked the price up a hundred

00:19:41.200 or a thousand fold. And so it hasn't been available for rheumatologists to use since then.

00:19:48.280 It's stupid, bad pharma trick. Yeah. Well, I'd like to actually talk about a few more examples 1.00

00:19:53.980 of that because there are several. Going back to the patent, in classic pharma, are patents

00:20:02.400 primarily issued only for decomposition of matter? Or are companies trying to get patents

00:20:10.460 for process manufacturing and other things? Or is the playbook that, hey, if there's a really

00:20:16.280 complicated process required to make this drug, I'm going to file my patent on composition of

00:20:21.300 matter. I'm going to keep the process as trade secret. So even when this thing runs off patent,

00:20:26.380 you might know what the finished product looks like. You'll never figure out how to make it.

00:20:30.120 So both of those things happen depending on the drug.

00:20:32.880 Exactly.

00:20:33.800 So one good example would be, let's say, AbbVie and Humira.

00:20:38.440 They patented every single little thing around that drug that they could.

00:20:42.880 It was a huge winner for them, and they wanted to keep it protected as long as they possibly could.

00:20:47.820 And they would stagger it.

00:20:48.820 So they would first patent composition of matter, wait 10 years, patent this step-in process, wait five years, and you just keep doing it, doing it,

00:20:57.020 and you sort of effectively extend the patent life of the molecule and process.

00:21:01.480 Exactly. And that can be abused, I think, personally.

00:21:07.740 Okay.

00:21:07.940 But exactly. They'll patent the drug. They'll patent the formulation. They'll patent the salts.

00:21:14.000 They'll patent the auto-injector. They'll patent, of course, the indication right at the beginning

00:21:19.680 to the extent that they can, which is sort of method of use. And they'll patent the dose.

00:21:25.400 They'll patent the route of administration, everything.

00:21:28.340 Yeah. Combining it with something else.

00:21:29.500 Okay. Going back-

00:21:30.880 And the method of manufacturing.

00:21:32.320 Yes. Going back to the clinical team, the scientific team that is beginning the exploration of what to do,

00:21:42.760 are you at the outset completely agnostic to whether you're looking for or entertaining small molecules or biologics as targets?

00:21:51.140 Do certain disease states lend themselves for you to go looking in one path versus the other?

00:21:57.080 I think there's two considerations.

00:21:58.780 One is that there are some conditions that might suggest one route or the other, but

00:22:06.520 there are also sometimes companies or infrastructure that would make it better to use one format

00:22:15.240 or the other.

00:22:15.780 I think big companies are format agnostic because all the big drug companies are now doing small molecules and biologics and many gene therapies and even some now cell therapies, which are sort of the frontier of complex medical therapeutics.

00:22:33.480 For example, if you're treating a childhood disease, then you need to make something that is oral and tastes good.

00:22:41.200 All the parents out there are going to remember, um, you know, the, the grape and the cherry

00:22:46.360 flavored Tylenol or Advil or grape Sceptra was, was a particular favorite of mine when I was

00:22:52.240 given my kids drugs. Yeah. So that's one example. Uh, others are, there are inhaled drugs for

00:22:59.680 specific lung conditions. And so some formats will, will make sense. And there's a medical

00:23:05.160 rationale for that. So then how do you begin the screening process? How do you begin to identify

00:23:10.680 molecules. And again, I think we should assume that our listeners who are otherwise well-informed

00:23:16.080 won't know the details of what an IND is, what we use phase one, two, three, four, and what has to

00:23:22.560 happen prior to the IND. Like, let's just start from the very beginning. Right. So with the

00:23:27.040 preamble that this would be about a year-long course. Yeah. If we did this as a seminar at

00:23:33.660 Harvard, this would take you a year. Yeah. But to cover it at a high level, starting from the idea

00:23:40.040 and the medical indication. And it might be useful to think about one specific example. And

00:23:46.040 we can, let's think about muscle weakness, which can be described as sarcopenia. And the definition

00:23:53.660 of that is still evolving, frankly. I've gone to some of these specialty meetings, like the

00:23:59.360 Cachexia Consensus Conference, and it's a topic of discussion every year. But I think one, we seem

00:24:06.160 to be converging on the concept of decreased muscle mass with impaired muscle function

00:24:11.600 as a good definition of sarcopenia, and whether the function is grip strength or gait speed

00:24:18.260 or stair climb, or people use different ones.

00:24:22.480 Okay, so we'll talk about that as a specific indication.

00:24:25.240 So impaired muscle function with low muscle volume.

00:24:30.000 Which to your point, by the way, nobody listening to this doesn't care about this.

00:24:33.200 So this is a very topical consideration.

00:24:35.740 and it's not esoteric.

00:24:36.840 It's something I've personally been working on

00:24:38.980 for 20 years, as you know.

00:24:40.800 So what should be the drug format?

00:24:43.400 Well, the patient population are likely to be older adults.

00:24:46.840 So it needs to be a drug format that'll be suitable to them.

00:24:51.900 And shouldn't underestimate the importance of this.

00:24:56.780 It's what's most likely to work.

00:24:59.000 Because in a drug development program,

00:25:01.600 especially with a new target and a new indication,

00:25:04.120 There are so many unknowns, and all of the risks multiply across a drug development program.

00:25:10.680 So if you have, say, 20 risks you're taking in a drug development program, whether it's target and format and bioavailability.

00:25:19.460 Toxicity, the whole thing.

00:25:20.860 Everything.

00:25:21.880 And all of them have maybe a 90% chance of success.

00:25:25.300 You multiply all that out 10 or 20 times, it's zero.

00:25:27.980 It's a huge failure.

00:25:28.960 Yes.

00:25:29.200 So you have to minimize risks at every step.

00:25:34.120 other than the ones that you sort of identify and accept, this is the big question that we don't

00:25:40.240 know and we have to answer early. So we want to- And you want to fail fast.

00:25:45.060 If you can, yeah, absolutely. The worst outcome in drug development is failing in phase three.

00:25:52.380 Oh, yeah.

00:25:53.000 Actually, that's probably not true. The worst outcome is succeeding in phase three and failing

00:25:56.820 commercially. But failing early is super important. Now, let me just ask a question

00:26:01.760 about this specific indication. So you've already made a decision, which is we're going after

00:26:07.420 sarcopenia, and you've decided to do that as opposed to say, we're going to go after muscular

00:26:13.980 dystrophy. So are you doing that because sarcopenia is a much, much, much bigger market,

00:26:19.640 which is the obvious choice? Or are you saying it's easier to get approval there, and then

00:26:25.120 ultimately we can also demonstrate that this will work in Duchesne's muscular dystrophy?

00:26:29.160 How would you think about those two ways to proceed?

00:26:33.740 So my personal bias is I have a limited amount of time on this earth to develop medicines and help people.

00:26:42.960 I want to help as many people as I can.

00:26:45.040 So I mostly do large indications, and that's my personal choice.

00:26:50.500 Now, in that situation, regulatory pathway might be harder for sarcopenia if it is not yet identified clearly as a disease.

00:26:57.780 No question.

00:26:57.900 No question.

00:26:58.820 Whereas Duchenne's muscular dystrophy, that's an orphan disease, you would probably get a quicker

00:27:03.180 regulatory pathway to approval. Yes. Yes. And so a different strategy might be we go after Duchenne's

00:27:11.520 because we can get there faster. And then once we have demonstrated a drug that works in that

00:27:16.000 category, we then chase the approval pathway to help as many people as possible. Would those just

00:27:21.000 be two different strategies? Yes. And frankly, I've done both depending on the circumstance and

00:27:27.140 mechanism of the drug and so forth. And frankly, I'm doing that right now with one of the two

00:27:33.160 companies that I work with. Okay. So let's go back to your example, though. Let's get back to

00:27:37.180 sarcopenia. The very first thing that I tried to do in sarcopenia was prevent falls. Now,

00:27:47.000 how do you measure a fall, Peter? Well, I was going to ask a harder question, which is,

00:27:53.460 which are the muscles that are most responsible for a fall. And maybe that's part of what you

00:27:58.160 could test. It turns out, I think it's very complicated, right? I've had a guest on this

00:28:02.360 podcast who presented data that suggested something that's seemingly as innocuous as

00:28:08.080 great toe strength is an enormous predictor of falls. Now there's a very objective way to measure

00:28:13.120 the force that a person can generate with their great toe. And as that force gets below certain

00:28:18.080 thresholds relative to their body weight, the probability of falls just starts to go straight

00:28:22.000 out. Balance, obviously. I would argue that they don't really know that. Okay. Because I asked the

00:28:30.440 question, what causes falls? There's about 11 different things that cause falls. And people

00:28:36.660 can think through themselves. There are the obvious ones like weakness and dizziness, which

00:28:41.740 dizziness itself is very complicated, but also vision, also attention. Lots of things can cause

00:28:49.160 falls. Yeah. I mean, I think another big one that we see clinically, Lloyd, is loss of reactivity.

00:28:56.120 So foot speed and reactivity. So you and I, if we went for a walk today around Lake Austin,

00:29:02.660 so there's a 10 mile beautiful loop around Lake Austin, the probability that on that 10 mile loop,

00:29:08.660 you and I wouldn't stumble once, at least once and miss our footing is zero. But I would venture

00:29:15.560 that neither of us would fall. And so the question is why, why wouldn't we fall despite being

00:29:21.720 challenged at a great level, you know, stepping on a twig, missing a branch or a root or something

00:29:27.340 like that. And the reason is we have the reactive speed of our feet to catch ourselves. And that to

00:29:33.380 me is one of the things that's missing. And that tends to come down to the type two, a muscle

00:29:37.420 fiber, right? Like that's a very explosive, but also proprioception. Yes, absolutely. Vision,

00:29:42.600 all the things that we're talking about. But maybe the broader point is this is multifaceted.

00:29:47.500 There is an atrophy that is beginning of various systems in the body, and it's creating this

00:29:53.160 perfect storm where when you watch an elderly person fall, you realize that that's a situation 0.98

00:30:00.580 where they would have saved that fall. It's not the insult that's the problem, whatever caused

00:30:05.340 the perturbation in the step. It's the inability to catch it. I think that's exactly right. So

00:30:10.880 let's get back to the issue of measuring falls. But that's hard. Yeah, but this becomes a very

00:30:16.560 hard problem. So I tried to do that because in a clinical environment, the only falls

00:30:24.200 that are ascertained are those that cause injury, right? From our perspective as physicians,

00:30:30.940 those are the patients we see. And patients are reticent about reporting falls because they know

00:30:37.880 they could potentially be taken out of their home environment if they were felt to be unsafe.

00:30:44.380 And frankly, I would do exactly the same thing if I were in that circumstance. I want to stay home.

00:30:49.920 So we developed a study to try to measure falls. And so we worked with a large company that

00:31:00.060 manufactures triaxial accelerometers. And we made a, we made a research device for this,

00:31:07.080 for this effort. And I designed a wonderful study. So the study was, we were first going to put the

00:31:13.600 device on bad ice skaters in Boston in the winter and videotape the rink. And so the videotape

00:31:21.120 results are the positive controls for actual falls. And then we would look at the device

00:31:27.000 telemetry and look at the sensitivity and specificity of the device.

00:31:31.460 It's worn on the ankle, the wrist, where would you wear it?

00:31:33.800 This was going to be a pendant.

00:31:35.140 Okay.

00:31:36.020 And then the second part of the study is if the first part worked and the device worked

00:31:40.880 on falls that were real, we were going to put it on elderly nursing home residents.

00:31:46.220 And there the positive control was little old lady found on the floor because the people

00:31:52.280 were old and frail and unable to get up themselves.

00:31:54.620 So when the nurses or the staff found them on the floor, that was a fall, however they got there.

00:32:00.720 And then we look at the device telemetry and see.

00:32:03.620 And sorry, was the purpose of that exercise, Lloyd, to see if the ice skating telemetry could predict a fall on ice?

00:32:10.020 And was it offering the same insight that you were seeing in the actual field with the elderly people?

00:32:15.860 Well, it's a little bit that falls are like the Supreme Court in pornography.

00:32:20.100 You know, you say you know it when you see it.

00:32:21.460 And so we wanted to see it on the ice skating rink.

00:32:24.620 And then look at the device telemetry and see, was it reporting falls when they actually happened?

00:32:29.940 Was it reporting falls when they didn't?

00:32:32.340 Was it missing things?

00:32:34.160 Basically to see if the device worked.

00:32:37.060 So I put that through the institutional processes for funding, and I was told, cut the first part out.

00:32:44.720 And so just put it on the older adults.

00:32:46.600 So I worked with a really good geriatrician named Lou Lipsitz, and we did that study.

00:32:52.280 And Kieran Dole was the clinical operations person, and it was really a good study execution

00:33:00.800 because working with patients and research subjects, participants in their 80s up to

00:33:06.940 a little over 100 is challenging.

00:33:09.980 And these were all individuals in one living environment or across multiple?

00:33:14.120 Yes, they were in one living environment.

00:33:15.440 We wanted to minimize the variables.

00:33:19.260 How many subjects were?

00:33:20.340 We had 60 subjects.

00:33:21.640 And you falled them for how many months?

00:33:22.940 We falled them for six months.

00:33:24.580 And these are people who had fallen at least once in the prior six months.

00:33:29.180 So we knew they're at a high risk for more falls.

00:33:32.060 And remember, this institution has a lot of protocols and procedures in place to try to prevent falls, and they're still falling now and then.

00:33:41.380 So Lloyd, 60 subjects followed by six months, how many falls did you capture?

00:33:45.400 Well, we ended up 117, I think, was the number of events that actually happened based on someone found on the floor.

00:33:53.540 However, the device was awful in that I think it had about, it detected 17% of the real falls.

00:34:04.280 And only 17% of the device actuations where it said somebody fell were false.

00:34:10.680 So it was just not useful.

00:34:13.080 And it was the best we could come up with.

00:34:16.120 And just to be clear, you weren't asking it to predict antecedent movement pattern of fall.

00:34:23.040 No, no.

00:34:23.640 You were just asking, once a person has fallen, do you know?

00:34:27.740 Yes.

00:34:28.500 Very simple.

00:34:29.360 I would have guessed that you would have been higher.

00:34:31.320 So did I.

00:34:32.300 But we were wrong.

00:34:33.900 This study is actually published now, but then we couldn't use it for measuring falls.

00:34:38.560 Yeah.

00:34:38.860 We tried one more thing. 0.99

00:34:40.560 There is a Massachusetts Institute of Technology professor named Dina Khatabi who was using Wi-Fi type devices to measure people's movements in their homes.

00:34:54.040 And it is a little scary.

00:34:55.760 It could measure wherever you were and whatever you were doing.

00:35:00.460 And so we thought that would be a great way to assess falls.

00:35:03.360 But ultimately there, I don't remember the issue, but we ended up not being able to use it.

00:35:08.380 So we had to stop our whole program from making drugs to prevent falls because we couldn't measure them.

00:35:15.600 Now, let me play devil's advocate for a moment. Couldn't you have just said, we want everyone involved in a study to report if they fell because it's going to help us get better data for the study. And I mean, given that you're, you're going to capture all the reported falls and really all you're trying to do is capture the signal of unreported falls.

00:35:40.200 But if you ask the people, say, look, we're not taking away your driver's license, you know, whatever it is that you're afraid of losing, we are just trying to report this as an AE or as an outcome.

00:35:50.760 I mean, it seems to me you'd have a much higher chance of capturing it than any sort of device at this point.

00:35:56.460 That would be true.

00:35:57.620 We were just worried about the data quality. 0.91

00:36:00.280 Because people, you know, remember, these are older adults who were frail.

00:36:05.320 and we were worried about the, you know, recall bias and we were worried about whether they would

00:36:12.380 have something to write it down with. And people generally weren't device savvy at the time. I

00:36:18.440 think that might be better now. Anyway, we elected not to do it. I have to think there's some AI way

00:36:21.920 to do it. Like there must be, I mean, were these accelerometers using any AI? This was a long time

00:36:27.880 ago. This was 10 years ago or more. So there was no AI in those days. Well, some people would take

00:36:33.820 issue with that. Not the way we see it today. Anyway, it was an interesting example of a lot

00:36:40.360 of ideas are conceptualized and we have to be able to objectively measure things ideally

00:36:46.020 in drug development. And we tried and we couldn't and we moved on. So this kind of brings us to

00:36:52.500 sarcopenia and muscle mass and strength, which was the genesis of Bimagramab and active and

00:36:59.640 receptor antagonists in general. Now, we knew at the time about myostatin. So Seijin Lee is sort of

00:37:08.100 the father of myostatin. He discovered the biology. And in rodents, myostatin is, you know,

00:37:18.180 kind of amazing. You can turn a mouse into an Arnold Schwarzenegger mouse by blocking myostatin.

00:37:25.400 this was in the mid nineties, right? I think so. Yeah. Yeah. I mean, I remember this was when I was

00:37:30.980 in medical school, I remember in 97 seeing the images of the mice, the chickens, the dogs,

00:37:39.040 cows. I mean, we, we couldn't get enough of these myostatin knockout animals. We thought it was the

00:37:45.180 greatest thing we'd ever seen as students. There were no people who had that though. No. So,

00:37:51.380 So at the time, the people who led the discovery project, so this is the laboratory research, were Chris Liu and his team and David Glass and his team.

00:38:04.020 And that's where the original code of Bimagramab was BYM338.

00:38:09.380 And that's where the VAT came from.

00:38:11.740 I was part of that team on the clinical side.

00:38:16.740 And if they made the drug, how would we test it?

00:38:20.460 Do you want to explain how myostatin inhibition would lead to enormous muscles?

00:38:28.440 Yes.

00:38:28.900 So the broader question is what governs the size of your muscles?

00:38:34.160 And we know it's nutrition and we know it's use.

00:38:39.100 And then there are biochemical things that regulate it.

00:38:42.860 And myostatin is an inhibitor of muscle growth.

00:38:46.380 So if you inhibit the inhibitor or block myostatin, muscles get larger up to another point where there's something that regulates them, and we don't know what that is.

00:38:56.300 And that's most of the inhibitory effect biochemically in animal species.

00:39:02.640 In humans, it turns out that it's more complex.

00:39:05.980 It's myostatin plus activins, mostly active in A.

00:39:09.300 And this is the advantage of inhibiting myostatin and activin A together or blocking the receptor, which is what the Magrumab does.

00:39:22.160 What was the evolutionary, I mean, I'm asking you as though you were there during the design phase, but what is your best guess for what the evolutionary reason was to limit muscle growth? Was it simply a nutrient management system, which was like, hey, we grew up in a nutrient sparse environment. We can't just have muscles demanding all of this protein.

00:39:41.160 I would have to guess that based on everything else about people, there were times during our

00:39:47.540 evolution when resources were very scarce and we had to conserve. But I don't know. Your guess is

00:39:53.520 as good as well. But it is interesting because the other way that nature could have solved that

00:39:57.860 problem was just to make it completely supply limited and just say, yeah. But anyway, okay,

00:40:04.800 so we don't really have a great teleologic reason. That's kind of the way it works with

00:40:07.520 adipose tissue. Yeah, yeah, exactly. That's right. So we don't have a great explanation for why,

00:40:14.220 but regardless, in humans, myostatin plays a smaller role than it does in these less,

00:40:22.400 presumably slightly less complex mammals. And what is the actual mechanism by which

00:40:27.660 myostatin is inhibiting? Is it doing something in actin myosin filaments? What is it doing to

00:40:33.600 prevent hypertrophy complicated but to summarize it briefly the receptors are part of this larger

00:40:40.960 tgf beta super family of receptors and there's dozens of them there's type one type two type

00:40:46.580 threes and they all signal via mostly a common pathway of called smads which was named by those

00:40:56.180 whimsical drosophila geneticists it stands for similar to mothers against decapentaplegic

00:41:01.340 which people don't need to know. And those are transcription factors, and they

00:41:08.320 govern a whole lot of gene programs. Some of the more important ones are muscle. So muscle size

00:41:16.260 is regulated by nutritional availability and then muscle protein synthesis versus muscle protein

00:41:25.320 turnover. And the proteins that turn over muscle are MRF1 and MAFBOX or atrogen. And David Glass

00:41:36.060 was one of the discoverers of this pathway, who I mentioned earlier. And myostatin signaling via

00:41:44.940 the active in receptors suppresses the proteins that are involved in targeting muscle proteins

00:41:53.440 for degradation. So before we leave myostatin to talk about BEMA, do you want to say anything

00:42:00.280 about folistatin? And I don't even know if you're aware, but folistatin became a very popular

00:42:05.080 recreational sort of gray market agent that was sold for research purposes only in quotes.

00:42:13.220 And the marketing material suggested, look, if you take folistatin, folistatin inhibits myostatin,

00:42:20.280 you're going to get really big muscles. And so people were pumping themselves full of folistatin.

00:42:25.800 Actually, that's not really true. I think the folistatin was so expensive. They weren't doing

00:42:29.600 that. They were doing some attempt at folistatin gene therapy. So do you want to just explain

00:42:35.300 what folistatin is or why it may not be as holy grail as it was made out to be?

00:42:41.180 So folistatin and then there's folistatin-like proteins are endogenous inhibitors of this

00:42:46.920 pathway we've been talking about. And it does, if you do a gene therapy in rodents, result in larger

00:42:54.140 muscles. But it's a small protein with a relatively short half-life. And I don't think dosing it

00:43:00.640 systemically now and then would be successful. We did the math on this and you would need to

00:43:05.620 give it several times a day. And given the price of folistatin, you'd be spending above, I don't

00:43:14.600 You know, you'd be spending a million dollars a month on folistatin, but also it wasn't clear that

00:43:20.120 it would do anything in an adult. In other words, it seemed that there might've been a critical

00:43:24.460 window during which administration of folistatin or folistatin gene therapy would have an impact,

00:43:30.260 but it had to be pretty young. It'd be during the development of the muscle more so than a mature

00:43:35.980 phase of the muscle. I think it would work in adults if you could solve the half-life. And

00:43:41.160 I know there are companies working on this with FC fusion proteins and other half-life

00:43:46.020 extended versions.

00:43:47.540 That's a good point.

00:43:48.180 Can you tell people why an FC fusion, I mean, I think we have to talk about some of this

00:43:51.740 technical stuff, unfortunately.

00:43:53.120 I almost caught myself as I was asking the question, but explain what an FC fusion protein

00:43:57.940 is and why that might be able to keep it in place longer.

00:44:00.640 Sure.

00:44:00.880 You can edit this out later if it gets too technical.

00:44:03.420 I'll try to speak in plain English.

00:44:05.880 Well, yeah.

00:44:06.420 Just think about it through the lens of how you manipulate drugs.

00:44:09.940 Like, I think of this as part of the story.

00:44:11.820 Right.

00:44:12.140 So if I remember correctly, the technology that we use now for FC fusion proteins was

00:44:19.360 originally developed by Brian Seed at the Mass General.

00:44:23.580 And the very first drug that used it was Enbrel, which is etanercept, which is itself a really

00:44:31.780 interesting story because this is a TNF inhibitor that's now approved in rheumatoid arthritis

00:44:36.600 and psoriasis and some other things.

00:44:39.160 but it was first tested in sepsis. I didn't know that. And it made people worse. Well,

00:44:45.800 because I think physicians and scientists at the time knew that sepsis was an exuberant

00:44:51.580 inflammatory reaction to an infectious stimulus. And they thought by tamping down the inflammatory

00:44:58.280 component, you might be able to have better outcomes. And it turned out it made people worse.

00:45:02.320 But ultimately it was then tested in rheumatoid arthritis and it was amazing.

00:45:06.500 And you know the story after that.

00:45:09.340 But a company called Immunex at the time licensed the technology from Mass General to make a tannercept.

00:45:16.280 And essentially what it does, it does two things.

00:45:18.620 So the soluble receptor, which was the low affinity regulatory TNF receptor, I think, at the time.

00:45:27.900 Got to double check that.

00:45:28.920 I haven't thought about a tannercept in like 20 years.

00:45:30.880 had a relatively short half-life just as a protein injected all by itself. And doing the

00:45:39.780 recombinant DNA technology to attach it to this part of an antibody called the FC region

00:45:46.980 did two things. One is it extended the half-life of the protein in the circulation by allowing it

00:45:54.000 to recirculate the way some blood proteins are recirculated normally. So you're sort of

00:46:00.620 hijacking an endogenous mechanism to preserve proteins in the bloodstream and doing it the

00:46:05.300 same way. And the second thing it did is it put a hook on the protein to allow you to purify it

00:46:12.080 easily because the protein G and protein A columns were well-established at the time and easy to use

00:46:20.780 to purify proteins. So that's what it did for the drug developers. And since then, the technology,

00:46:27.740 this is a great example of going back to our patent discussion. Originally, that was patented

00:46:31.900 technology and only ImmuneX could use it or some other licensee of the Mass General. But the patents

00:46:37.600 expired and now the technology is freely available to the rest of the world for the rest of the

00:46:44.120 eternity. And many, many, many companies use this technology to make drugs and we're all better off

00:46:50.400 for it. Okay. So how did you guys discover BIMA? How did you create it? So again, the earliest

00:47:03.660 biology was done by Chris Liu's team in the Pathways group at Novartis Institutes. Jeff

00:47:10.120 Porter was the leader of that group. And the idea was we wanted to inhibit the receptors,

00:47:16.820 not go after all the possible ligands because we knew that myostatin wasn't the whole story in

00:47:22.460 humans. We didn't know which activins it was, thought it was probably activin A, but it could

00:47:27.620 have been others too. And that story and that thinking has panned out subsequently, as we can

00:47:31.920 get into later. And at the time, therapeutic antibodies were the best technology to do this.

00:47:40.940 Remember, the affinity of the ligands, myostatin and activins for the receptors was nanomolar,

00:47:50.020 low nanomolar, maybe high picomolar.

00:47:52.480 So we needed an inhibitor that could bind down in the low picomolar range in order to

00:47:59.520 effectively prevent ligands from binding.

00:48:02.000 Can't really do that easily with small molecules.

00:48:05.620 Okay, this is a great example of the importance of understanding that distinction.

00:48:10.040 Yes. I'm going to explain what you just said, and then I want to have you state that last point

00:48:14.980 again. So people, when they're talking about pharmacokinetics and they're talking about

00:48:20.160 affinity, you'd have to talk about a concentration. So what concentration of this hormone or this

00:48:26.760 ligand is necessary to get into this? And the, when you start talking about, you know, we can

00:48:33.200 talk about millimoles, micromoles, nanomoles, as we get smaller and smaller and smaller, as those

00:48:38.400 numbers get smaller and smaller and smaller, it means you don't need very much of the thing to get

00:48:42.700 in the receptor. And therefore, if you're trying to develop something to block that, it becomes a

00:48:48.120 harder problem because you better figure out a way to usurp this guy getting in and this guy gets in

00:48:55.160 very easily. Yes, exactly. Exactly. And you're saying to get a small molecule to have that degree

00:49:02.720 of sensitivity is very challenging.

00:49:05.200 Yes.

00:49:05.600 And therefore, something that's biologic makes more sense.

00:49:08.560 Yes.

00:49:08.960 And is that due to just the physics of the conformational fit?

00:49:14.260 Yeah, I think you could describe it that way.

00:49:15.920 It's essentially this, it's more complicated than this, but think about it as the surface

00:49:21.360 of interaction of the two molecules binding together.

00:49:26.180 Essentially, we need, if the ligand, myostatin and activin for the receptor is sticky, you

00:49:32.700 need an inhibitor that's even stickier. Stickier, yeah. And that would be very challenging to do

00:49:37.160 with small molecules. So in this case, we went with the biologic route, sort of ran a therapeutic

00:49:42.420 antibody project in collaboration with Morphosis, with whom we had a collaboration at the time,

00:49:50.180 and had a bunch of candidate antibodies, and then ran through the usual developability,

00:49:57.260 maturation, improvement of what we get in the screen until we have what we think could be

00:50:02.560 a therapeutic drug. How many molecules enter the top of that funnel? Oh, there might be thousands

00:50:08.820 of antibodies that get tested initially. This is done with something called phage display technology.

00:50:14.820 So I would say for the past 15 to 20 years, we don't make antibodies in mice anymore.

00:50:22.500 It's all done by recombinant DNA using these viruses. Okay. And I think back, I've been in

00:50:28.940 the business long enough. I've made antibodies by immunizing mice and fusing cells and growing

00:50:35.440 them up and putting them back into mice as a CITES tumors to get enough antibody to do experiments

00:50:42.480 with. I mean, it was terrible. It's much better now. So you're literally running a screen with

00:50:48.940 all of these antibodies and you're screening for two things, but well, basically one thing.

00:50:55.060 what is going to give me the lowest concentration that binds to this ligand?

00:50:59.440 That binds, sorry, to this receptor.

00:51:01.220 Yes.

00:51:01.960 So knowing that you have to be below a certain concentration.

00:51:05.140 We knew we needed really high affinity antibodies.

00:51:07.920 There were some things we didn't know.

00:51:10.520 So in the laboratory, most of the time you can measure a cell's surface receptor on the surface of the cell

00:51:17.540 pretty easily using things like flow cytometry.

00:51:21.600 The active and type 2 receptors are actually expressed at such low levels, you can't see

00:51:26.120 them by flow cytometry, unless you somehow very artificially manipulate the cell.

00:51:31.500 So we had to create a screening assay that was essentially a reporter assay.

00:51:37.360 So we couldn't measure the receptor on the surface of the cell.

00:51:40.280 We could have developed an assay to do that, but it would have been very laborious, radioactivity,

00:51:45.520 not necessary.

00:51:46.760 So we put in a reporter gene, which is, and basically made the cells glow with firefly

00:51:54.060 luciferase if the ligand bound myostatin or activin.

00:51:59.180 And then we were looking for decrease in the glowing with therapeutic intervention.

00:52:05.460 And we also didn't know whether we needed to do inhibit activin receptor type 2A or

00:52:11.780 B or both.

00:52:12.860 Most of the work in vitro suggested for muscle hypertrophy, most of it was driven by 2B, but we weren't sure.

00:52:23.240 And we ended up getting a 2B preferential drug, but it also hits 2A.

00:52:31.080 And this has nothing to do with muscle fiber type?

00:52:34.380 No, this works on all fiber types.

00:52:36.900 Oh, okay.

00:52:37.300 And then we looked for antibodies that worked in that cellular assay, where we wanted to prevent

00:52:45.080 cells from glowing when we added myostatin and activin. It had to work on both of them.

00:52:50.900 In other words, it shouldn't matter which ligand you put in. The antibodies should block

00:52:54.420 their activity. And the affinity of the antibodies had to be really good. So it had to be much

00:53:01.640 stickier than the ligands for the receptor. And so we had that. And then the real important

00:53:07.160 experiment is could we block the activity in an animal? But just that first step, Lloyd,

00:53:13.480 until you could identify candidates, how many months was that? Years. That was years. So again,

00:53:20.120 just going back to the analogy of building a skyscraper, that's the planning phase of the

00:53:26.660 skyscraper. That's the excavation of the hole. That's probably the laying the foundation. You

00:53:31.420 haven't actually put any of the big pillars up yet. It's the permitting. It's securing the

00:53:35.940 funding. It's the, it's sort of the site planning. It's all of that from your building perspective.

00:53:42.760 And by the way, I assume that the standard estimate 20 years ago was every approved drug

00:53:49.700 is approximately 10 years and a billion dollars. That's got to be pretty low today. What's the,

00:53:55.720 do you have a sense of what the more accurate dollar figure is? It's got to be more than a

00:53:59.800 billion today per approved drug. Oh yeah. There's a, there's a Tufts organization for the study of

00:54:05.760 drug development, and they think it's, I don't know, two, three, four billion, something like

00:54:10.080 that. Yeah. The part you're doing now is time-consuming. Luckily, it's not that expensive,

00:54:14.320 correct? Right. You're spending millions of dollars, but not necessarily, you know,

00:54:19.040 tens of millions at this point. Yes. Okay. So you finally identify a candidate or several

00:54:25.940 candidates that you now want to take to the next step, which is, hey, in vivo, does this thing work?

00:54:30.680 Yeah. So we would typically have two to 10 at this stage. And it's taken years because we have

00:54:39.360 to work through the biology and make sure that we understand what's going to happen.

00:54:45.660 Not because we want to save the cells in the Petri dish, but because

00:54:49.060 we don't want to start something unless we have confidence it'll succeed if we pass each step.

00:54:54.620 So we have to work through the biology. We have to make all the tools we need,

00:54:58.380 which are these glowing cells in response to myostatin, for example, and there's plenty of

00:55:04.240 other tools that we need too. We need to make the reagents. We need to make the myostatin and

00:55:09.220 the active NA and all the tools we need to do these experiments. So, and then run the experiments

00:55:16.620 and we have to make the antibodies and that takes quite a bit of time as well as

00:55:21.860 developability on the antibodies, which means we need to, at the very earliest stage, have

00:55:28.020 high confidence that we're going to make an antibody that we could give to people and it'll

00:55:32.120 be stable and it'll have predictable physical properties and it'll have a shelf life. And all

00:55:38.780 of these things are what we build into the antibodies at this very early stage. How

00:55:44.180 confident are you at that stage, Lloyd, that the antibody won't elicit an immune response in a

00:55:50.100 human? It's still one of the biggest unknowns. And the best way to do it is to use fully human

00:55:56.520 Humanized, yeah.

00:55:57.400 Not humanized, fully human.

00:55:59.560 Full, straight up human.

00:56:00.120 Yeah.

00:56:00.680 Can you explain to folks the distinction there?

00:56:02.320 Yeah, well, humanized is used to describe taking a mouse or other species antibody and replacing as much of it as you can with human sequences based on what we know about human antibody codon usage and amino acid preference and so forth.

00:56:20.280 whereas fully human means you're starting

00:56:23.980 with human genetic material antibodies.

00:56:28.760 But ultimately, it still has something in it that's foreign.

00:56:35.040 As little as possible. 0.97

00:56:36.000 Part of the sequence.

00:56:36.860 As little as possible.

00:56:37.840 There's going to be something new

00:56:38.980 because antibodies have this intrinsic ability

00:56:41.980 to recombine and create new sequences.

00:56:44.100 Yeah.

00:56:44.820 Now, in vivo, in people,

00:56:47.480 so if you make some brand new antibody

00:56:49.280 and it's not suitable for some reason, it gets selected out.

00:56:53.640 Whereas when we do this in vitro in a test tube, that doesn't happen.

00:56:58.020 We do the best we can and try to end up with sort of common codon usages,

00:57:03.920 common antibody sequences, even pair-wise and so forth.

00:57:07.580 But you don't really know until you put it into people.

00:57:12.260 There are some in silico screens you can do by looking at what peptides are likely to be generated

00:57:18.420 in a lysosome and are they going to have high affinity binding to an MHC molecule and so forth

00:57:24.680 and we do all that stuff. But you still don't know until you do it. Now it's an interesting

00:57:29.960 historical perspective because I'm getting old enough to be interested in history now.

00:57:35.360 The very first antibodies tested in humans were mouse antibodies. And there's still one that's

00:57:41.120 used to this day as a therapeutic. So it's OKT3. So this is an antibody used to prevent

00:57:47.920 transplant rejection. It's an anti-T cell antibody, human T cell antibody, and we still use it to this

00:57:55.280 day. And what is it targeting? Is it literally targeting CD3? Yep. Wow. So it's broad. Yes.

00:58:02.280 It's going after every T cell. Yes. Right. And there's rabbit antithymocyte globulin too that's

00:58:08.500 still used clinically. That's a blast from the past. It's still used. Yeah. And these were the

00:58:13.820 first antibodies used in people. And of course, uh, the other antibody, the other foreign antibodies

00:58:19.120 that are still used are antivenoms. Most of them are horse serum. As you can imagine, you only want

00:58:26.260 to get bitten by a snake once and need that because when you need it the second time, you have kind of

00:58:30.960 a ferocious serum sickness response. Great point. So you identify BEMA plus a few others and you

00:58:40.500 start now running these into the mice. Yes. Is there another chance that because you've gone to

00:58:47.060 all the trouble to make sure you have a human antibody, it won't work in the mice when it

00:58:53.520 otherwise would have worked in humans? We test that beforehand. Okay. Got it. So this is, this

00:58:59.200 is again part, there's an enormous amount of detail in drug development. And one of the things

00:59:08.220 is that your drug has to cross-react, I mean, work also in at least one of the two species

00:59:16.260 we're going to use later for toxicology. And it has to work in a species we're going to use for

00:59:21.220 pharmacology. If it doesn't, then you need to make surrogate drugs to do that. But in our case,

00:59:28.900 we made one that worked in rodents. Now, it turns out that bimagrimab is pretty immunogenic in mice.

00:59:38.220 but not in rats. Although- How common is that? It's kind of idiosyncratic.

00:59:45.280 So the construct we ended up using in mouse experiments was something called CDD866,

00:59:52.960 which was, you talked about humanizing antibodies. We murinized the Magromab so that we could use

01:00:00.020 it in mice. So you could go back and use it. Yeah. That makes sense. And to make a long story short,

01:00:05.940 we were able to give these antibodies to mice and see if it caused muscle hypertrophy or not.

01:00:10.660 And how much did it do so relative to the pure myostatin knockouts that were enormous?

01:00:16.580 Probably more so than the myostatin knockouts.

01:00:19.660 Wow.

01:00:20.000 It was really impressive.

01:00:22.680 Again, we're going to link in the show notes to what these images look like,

01:00:27.500 but it is truly a caricature.

01:00:30.680 It's impressive in the mice.

01:00:32.020 Look at also the whippet dogs, the Belgian blue cattle that are double muscled.

01:00:37.720 It's impressive.

01:00:39.580 Did you do anything else, Lloyd, at that time?

01:00:42.560 So when you demonstrate that BEMA is making bodybuilding mice, was there any assessment of muscle function?

01:00:51.820 Yes.

01:00:52.860 Okay.

01:00:53.240 And what did you find?

01:00:54.460 The mice were stronger and can run faster.

01:00:57.020 Okay.

01:00:57.320 But remember, they had perhaps a 30% increase in their muscle mass.

01:01:02.020 I mean, which is a huge, and people can look at the pictures online and see it's quite obvious.

01:01:11.340 It doesn't work this well in humans, because just skipping all the way ahead, and we'll come back,

01:01:17.560 humans get about 4% to 8% increase in muscle mass.

01:01:22.340 Most of the people we've tested have been older people, which is one caveat, whereas...

01:01:27.940 Did you do it in old mice?

01:01:29.760 Not in old mice.

01:01:31.900 I think David Glass did some experiments in older rats.

01:01:35.220 And it still worked, but not quite, not as well as in younger ones.

01:01:39.420 And remember, when people do these muscle experiments in rodents,

01:01:43.280 they almost always do males because it works better in males than females.

01:01:47.640 Cool thyself. 0.60

01:01:50.500 It would be interesting to know, as you ran it across a continuum of escalating age,

01:01:57.100 what accounts for the reduction in efficacy, right? Is it substrate limited? Is it a muscle

01:02:03.620 protein synthesis problem? I mean, what is, you know-

01:02:06.640 I don't think that's been studied formally by scientists. It might be known in the bodybuilding

01:02:13.720 community. And just a heads up to the listeners who might be bodybuilders, the first thing that

01:02:21.520 any company does when they're working on a drug that has the potential for abuse

01:02:25.860 is we work with WADA, the World Anti-Doping Agency,

01:02:29.320 to make sure that they can screen for these things.

01:02:32.340 So where in the pathway of BEMA did you begin notifying WADA

01:02:37.480 that we're working on this thing?

01:02:39.280 As soon as we had a therapeutic antibody, we started that process.

01:02:46.380 And basically, they've had an assay for more than 10 years.

01:02:53.080 Have they ever caught it?

01:02:54.320 Have they ever screened it?

01:02:54.800 I have no idea.

01:02:55.860 That's funny. Okay. So after you get the home run in mice, do you want to go into a primate?

01:03:03.920 Where do you typically go from mice?

01:03:06.080 So for therapeutic antibodies, often it includes a primate simply because we want to have one of the two toxicology species in whom the antibody has the expected pharmacology.

01:03:22.520 so we can look at sort of on-pathway and off-pathway toxicity of the therapeutic.

01:03:28.280 So we typically use non-human primates for this.

01:03:32.820 When you're at the mouse stage, Lloyd, how are you screening for tox besides the most obvious,

01:03:37.820 right? Obviously, mortality or something catastrophic is obvious. But for non-apparent

01:03:43.240 or non-mortality-based toxicity, what are you looking for? And is any toxicity at the mouse

01:03:48.320 level disqualifying to go forward? Or are you evaluating it case by case and saying, look,

01:03:52.520 okay, this ended up being pretty bad for the mice despite the efficacy. We don't think that's going

01:03:57.900 to be an issue, or we think we got the dose wrong, or do you basically keep going back and perfecting

01:04:02.580 it in the mice until you get the dose response right before you move up? Or do you just sometimes

01:04:08.200 say, no, we're going to go to the primate or whatever other model we're going to look at

01:04:11.600 and reassess tox as we get closer to our species of interest? Yeah. So with the caveat that I'm not

01:04:18.900 a toxicologist, the fundamental principles are you use a weight of evidence approach

01:04:24.820 based on all the data that accumulates. From a clinical perspective, we try to balance risk

01:04:32.140 and benefit with new medicines in general. So if it turned out that bimagrimab had some,

01:04:38.520 and bimagrimab has some tox that we'll get into, but if it were unsuitable for use in a big

01:04:44.780 population that we then think about higher medical need patients.

01:04:49.520 That's when you would go from maybe sarcopenia to Duchenne's, muscular dystrophy.

01:04:53.280 For example, yes.

01:04:53.720 If we, yep.

01:04:54.760 And where whatever the adverse effects are would be outmatched by the potential benefits.

01:05:01.740 Yep.

01:05:02.440 But secondly, we want to know precisely what the toxicology is.

01:05:08.460 And the two big things we look for are whether it's monitorable, whether it's reversible.

01:05:14.780 So if we have irreversible cardiac toxicity with a therapeutic, that's usually the end, for example, or neurologic. If it's serious organ toxicity.

01:05:26.800 But there's enough of a prodrome. So for example, if it's, well, you look at a drug like Lamisil, right? Something as supposedly benign as Lamisil. I mean, that can destroy your liver.

01:05:41.100 but there you can stop the drug. So that's your point. It's moderatable and reversible

01:05:46.580 if you stop it and you get a long enough warning. Because otherwise, if it would just

01:05:51.900 automatically destroy a person's liver at a frequency of one in a hundred people using it,

01:05:57.120 you can never justify it. So you're describing idiosyncratic liver toxicity,

01:06:01.280 which is the most common reason drugs get pulled off the market still. And I personally have killed

01:06:09.020 drug programs for that. And it's hard to, can't predict it preclinically. Yeah. Yeah. So, so let's

01:06:16.420 get back. You get BEMA into the primates. Yes. And how does it, how does it perform? 0.99

01:06:22.220 Well, in order to do toxicology studies, you have to know how much to give them and how long it's

01:06:29.720 going to last and what it's doing. So there are, there are preliminary studies in a small number

01:06:35.140 of animals. And so we did those, but we did them long enough so that we could see the muscle

01:06:39.500 hypertrophy of what were going to happen. And it did work, not as well as in the rodents,

01:06:45.000 but it did work. So it gave us confidence that we could move ahead with the rest of the activities.

01:06:51.420 Now, manufacturing enough antibodies for use in larger animals is time and expense. And so

01:06:59.840 all of that was going on in parallel. And each of these decisions, so you're doing this all

01:07:04.260 side of Novartis. Is there an, I see, is there an investment committee that basically revisits

01:07:10.220 every time there is a new, you know, allocation of capital to move from one thing to the other,

01:07:16.200 where everybody presents and how does that typically work in a large company?

01:07:19.960 All big drug companies work kind of the same way in that there are, there are typically two or

01:07:24.540 three or four, depending on the company, major checkpoints where all the data are assembled and,

01:07:30.520 made into slide decks and presented and feedbacks obtained and programs, courses adjusted. And so

01:07:39.020 that happens. And I think it's more frequent, but faster at small companies. But things are

01:07:46.800 constantly being reevaluated and reassessed. Yeah. And what's interesting for folks listening

01:07:51.340 to us is we're talking about this in the context of a large company that doesn't have to go out

01:07:56.240 raise capital every time it does this. But the exact same idea that you just described,

01:08:01.220 everything you just said could have been done by a startup, but now it would have a totally

01:08:05.220 different look and feel in that, Hey, we're going to go raise some seed funding to go test this

01:08:10.540 idea. Okay. Guess what? We were able to find the antibody. We're going to have to go raise another,

01:08:14.820 you know, $20 million and boom, boom. And now they'd be at the stage where they'd be probably

01:08:19.920 raising a series B or no, probably this would be a, still be an A I think as they go into the

01:08:25.820 primate. Whatever you call it. But give folks a sense of how much you'd have to raise for this

01:08:31.980 next stage, which is basically your pre-IND. So you have to manufacture toxicology and clinical

01:08:42.140 material. You have to run the IND enabling studies, which is toxicology and some other things.

01:08:49.860 and frankly, investors want value creation for their money and reasonably so. So I would think

01:09:00.120 for Bimagramab, if this were in a small company, the value creation step would be showing muscle

01:09:07.140 hypertrophy in the very first clinical study. So the funding that I would raise would be IND

01:09:13.500 enabling plus phase one plus phase one plus a runway to raise the next round yeah and you would

01:09:20.900 structure your phase one to demonstrate efficacy even though technically you only need to do talks

01:09:26.260 yes you would have it long enough big enough not not talks but safety that's right intolerability

01:09:31.740 yeah so okay and then just for using bema as an example how many dollars would that be from where

01:09:37.520 we are now to give you that runway into 2a in today's dollars probably 20 million dollars 0.99

01:09:43.760 Okay.

01:09:44.240 Yeah.

01:09:44.660 So Series A.

01:09:45.420 Thereabout.

01:09:45.960 Yeah.

01:09:46.340 Okay.

01:09:47.880 That's lower than I would have guessed, by the way.

01:09:50.400 Maybe a little more.

01:09:51.480 Okay.

01:09:51.800 It depends on where you do the manufacturing.

01:09:54.420 So for Novartis, this is nothing.

01:09:57.140 For a startup, this is everything.

01:09:59.300 You're betting the farm.

01:10:00.260 Well, not enough.

01:10:01.840 I don't want to make light of it inside of Novartis, but the point is Novartis doesn't

01:10:04.240 have to go back to the public market to say, I need to raise another $25 million to fund

01:10:07.780 this.

01:10:08.000 They're doing a lot of these in parallel.

01:10:09.240 Big companies, though, their resources are stretched, too. It's kind of funny thinking about it, looking from the outside, but having been inside a big company, people are competing for a fixed amount of research dollars, the people within the company, and resources are allocated based on company strategy.

01:10:30.060 and ironically sometimes there's more project capital available in a small company than in a

01:10:37.660 big company because the company the small company's got one or two or three right they're taking fewer

01:10:41.520 shots on gold all the money's going there yeah whereas there are hundreds in the big companies

01:10:46.720 so and i've seen it both ways i've seen some big company projects get high profile high importance

01:10:54.840 well-funded. So yeah, so we'll say $20 to $30 million maybe at this stage. And

01:11:04.360 the Magromab at that point made it through rodent and non-rodent toxicology studies and

01:11:12.840 what we call DMPK, which is distribution metabolism pharmacokinetics. It's

01:11:19.260 knowing that when you give a participant or a patient, a subject, a medicine, does it get into

01:11:26.480 their body? Does it go to where you want it to be? Does it do what you expect it to do?

01:11:32.040 You have to know all that stuff before you go into patients for the first time. And we assess that

01:11:37.360 in animals. You asked earlier, what do you actually do to measure the toxic effects of a medicine?

01:11:44.000 and animals receive courses of therapy and we do blood tests just like we do in people.

01:11:51.540 Sometimes we would do x-rays if it was warranted and then they get autopsied to look at all the

01:11:57.200 organs and look for microscopic changes that you might not perceive clinically. And we know,

01:12:03.660 we have to know all of that before we give people an experimental medicine for the first time.

01:12:08.240 So any red flags whatsoever as you, or anything that is of concern, not necessarily a red flag, but anything that's, that's still an unknown as you're going into the phase one?

01:12:22.980 Lots, lots of unknowns going in.

01:12:25.200 And there are always things of concern.

01:12:28.500 I've never seen a drug development program that couldn't be stopped for some reason.

01:12:33.200 and you have to balance the unknowns and the uncertainties and the risks with the potential

01:12:40.760 benefits and make a decision about whether you move forward or not. It's kind of a joke in the

01:12:48.280 industry that every really successful program has been almost killed or killed several times

01:12:55.160 before it eventually makes it out into humans and then eventually commercialization.

01:13:01.600 What's the approximate attrition from that first candidate drug discovery to the IND filing?

01:13:10.660 That's a winnowing down of what to one.

01:13:14.740 It's hard to put in an aggregate because it depends based on the format of the drug.

01:13:21.340 And then there's other factors like strategy and funding and everything else.

01:13:25.280 But for biologics like bimigromab and a therapeutic antibody, it's pretty low, actually.

01:13:34.040 Five to one? Six to one?

01:13:36.340 I would say maybe 30% of them actually could get into humans.

01:13:41.920 Okay. Yeah, more than I would have thought.

01:13:43.940 Yeah. It's simply because there are no off-target adverse effects with antibodies in general.

01:13:50.800 There are some specific counterexamples to that, but in general, an antibody is not like a small molecule that could have liver tox or some other tox that you can't predict for reasons that you don't understand.

01:14:03.220 That's a great point. Yeah. So maybe I'll restate that so folks get it because the antibody is so specific by definition, it can't bind to many other things.

01:14:14.460 And in fact, we screen to make sure it doesn't.

01:14:16.540 Yeah. Yeah. Whereas the chemical can do lots of things off target. You know, I had on recently,

01:14:22.700 we had a podcast talking about a CTEP inhibition and you know, the very first version of that drug

01:14:29.500 lowered LDL cholesterol, but raised blood pressure. And that was a completely off target

01:14:35.260 complication of the drug. Yeah. Exactly. So that doesn't generally happen with biologists.

01:14:40.880 Got it. So that's why you have the higher throughput. Yeah. Okay. Let's skip. So you

01:14:45.340 file the IND and you're now ready to start a phase one. So an IND is requesting regulatory

01:14:51.920 permission to administer the drug to people. And you've already filed your patent at this point?

01:14:58.600 Yes. Yeah. Where in that process did you file it? Patents typically get filed.

01:15:04.520 Again, there's a... You want to do it as late as possible, but while still protecting.

01:15:09.780 Exactly. So typically around the point where you have a group of candidates from which your final drug will be selected, that's typically when we would do it.

01:15:21.260 Okay, wow.

01:15:22.060 Because you want the patent to last as long as possible, but once information about what you're doing is getting out, you want to have it protected.

01:15:30.760 Okay. So from the time you file the IND with the FDA until... And you have to show them

01:15:39.260 everything that we've talked about. Do you also have to, at the IND, show them that you can

01:15:44.060 manufacture in GMP? Yes. So the manufacturing is a core element of the OMN application that you do

01:15:51.180 for an IND. And this is US-specific nomenclature. IND stands for Investigational New Drug. In Europe,

01:16:00.320 It's called a clinical trial application, CTA.

01:16:05.040 There are other countries that have different nomenclature, and companies can do the first

01:16:10.820 in human study anywhere in the world that's got a proper regulatory environment and suitable

01:16:16.740 investigators and clinical sites with adequate quality and so forth.

01:16:22.000 But maybe we'll be U.S.-centric for this discussion.

01:16:25.980 Sure. Can you explain to folks what the hurdle is to GMP or good manufacturing processes and why

01:16:33.860 it's so important? And again, I call this out to listeners because we live in an era now where

01:16:39.600 these peptide therapeutics are very prevalent, these sort of gray market peptides. And there

01:16:45.700 are people out there that think, hey, I'm buying retitrutide. Yeah, no, they're not.

01:16:49.360 Yeah, exactly. Maybe use the GMP process as a way to explain why when you think you're buying retitrutide peptide for research purposes only, you are definitely not buying what Eli Lilly is going to eventually sell if they get FDA approval.

01:17:07.120 Right. So GMP stands for Good Manufacturing Process, and it's basically a commitment by the manufacturer to use high-quality standards with extensive documentation to be able to prove what they've made so that everybody can have confidence that this is a good quality material

01:17:35.140 and they know that what's on the label is what's in the bottle, basically.

01:17:38.680 And that there's nothing in the bottle that's not on the label.

01:17:41.640 Exactly.

01:17:42.240 It's purity, it's activity, it's contamination or lack thereof.

01:17:48.700 It's sterility, if you will.

01:17:50.940 It's all of those things.

01:17:53.240 It's that the material that's being purchased eventually commercially

01:17:58.280 is the same material as what was tested clinically.

01:18:02.840 and we can have confidence in it basically the factories are inspected and the factories that

01:18:08.860 make it have and typically there are many manufacturers involved when you buy a with

01:18:14.600 the exception of saying buying a bottle of terzapatide from eli lily but typically when

01:18:20.520 you buy a drug from somebody the drug's substance the chemical is manufactured by one company

01:18:28.000 And then it is formulated or put into a mixture that makes it predictably absorbed or administered, is done by another company, and then it's put into a package by a third company, and then it's distributed by a fourth company.

01:18:46.600 So there's a lot of people involved. And this whole manufacturing infrastructure and pipeline is well-controlled and well-documented. And you can buy online peptides that may be the same as redditrutide. They might not. You have no way of knowing.

01:19:03.220 Yeah. And in many ways, that's the premium you're paying when you're buying the drug from Novo Nordisk or Eli Lilly or Novartis or whatever. Part of the premium is it's very expensive to manufacture under GMP conditions. So it's a bit of a buyer beware when you decide not to.

01:19:21.940 I think it's a big mistake to buy these peptides from fly-by-night manufacturers.

01:19:28.120 Conceptually, it's no different than going in a drug user, going and buying some opioid from a street corner drug dealer.

01:19:36.220 You have no idea what's in that.

01:19:37.580 Could it have fentanyl in it?

01:19:38.840 Could it have car fentanyl in it, which is even worse than fentanyl?

01:19:42.420 Baking soda, you have no idea what's in there.

01:19:44.500 It's the same thing with these peptides.

01:19:45.840 You have no idea.

01:19:47.120 I think it's a mistake.

01:19:47.960 And plus, even if you were to have confidence that those peptides were what they're saying they were, the data to support what they do are almost non-existent.

01:19:59.360 I've been reading in the popular literature about this one that's, I think it's called BP-197.

01:20:04.080 BPC-157.

01:20:05.560 157, yeah.

01:20:06.780 All of the data for that peptide come from one investigator who's the only person published on it.

01:20:13.220 And remember, the fundamental tenet of science is if it's real, it's reproducible.

01:20:17.960 this has not been reproduced. What the hell is it? It's not encoded in the human genome,

01:20:24.620 so it's not a human peptide. And it has no known receptor.

01:20:30.500 Yeah. So we don't even know how it works. There's so many red flags for this.

01:20:35.440 No, it's the poster child for what I would argue is the absolute greatest grift of the entire

01:20:42.920 health and wellness industry. There's a lot of, I'm sorry if I'm insulting you, Peter,

01:20:47.440 There's a lot of grift in the health and wellness industry.

01:20:49.640 Oh, you're not insulting me.

01:20:51.360 But that's my point.

01:20:52.540 Despite how much grift there is in the health and wellness industry, I'm putting BPC-157 on the podium, at least.

01:20:59.880 I would, too.

01:21:00.580 Yeah.

01:21:01.580 I would, too.

01:21:02.980 Okay.

01:21:03.320 So we've made Bimagrimab, and it's made it through all of the IND-enabling study activities.

01:21:10.880 And we're ready to give it to people.

01:21:13.540 Who do we give it to?

01:21:14.220 And this depends on what we need to measure and what the expected safety and tolerability issues are in people.

01:21:23.860 Again, we talk about toxicology in animal species.

01:21:27.720 We talk about safety and tolerability in humans.

01:21:31.960 And again, most antibodies, including Bimagramab, won't have safety and tolerability issues that are off the pathway that it's working on.

01:21:41.680 and didn't really see any safety and any toxicology to speak of in the animals.

01:21:48.460 The only thing that I was a little worried about that we saw was in the rats, they had cardiac hypertrophy.

01:21:58.000 However, remember the animals had enormous change in their body size because of the muscle hypertrophy.

01:22:03.940 And if you normalize the heart size to the body size, it was normal.

01:22:09.460 So does that mean that you didn't know if the cardiac hypertrophy was in response to more resistance that the heart had to work against, or whether the antibody was working directly on the cardiac myocytes and increasing hypertrophy there as it was in the skeletal muscle or both?

01:22:28.960 Exactly. We didn't know. But we could make an argument that rats of the size that they became-

01:22:37.020 Should have bigger hearts. 0.77

01:22:38.020 Should have bigger hearts.

01:22:39.460 And that was the argument we made to regulators, and so we didn't think there was any specific cardiac toxicity.

01:22:45.740 And typically in toxicology studies, you have something we call a recovery period, where the drug is withdrawn and some of the animals are followed to look and see whether any toxic effects that did occur are reversible.

01:22:59.480 And in fact, when you stop giving the animals Vimagramab, the muscles got smaller and the heart got a little smaller.

01:23:06.360 How often was it dosed?

01:23:07.560 So the way you dose in the toxicology studies is you want the exposure, which means the amount of drug in the blood, to be ideally higher than what we ever expect to get in humans.

01:23:22.000 And then when you got to the humans?

01:23:24.060 So that was the preamble to the answer to your question, which was weekly.

01:23:29.420 So we gave the animals the drug weekly.

01:23:31.960 And the half-life of the drug presumably is short, but...

01:23:35.700 It's shorter in animals, but again, you drive the dosing in the toxicology studies to make

01:23:41.680 the amount of drug in their blood ideally higher than we will get in people so that

01:23:47.920 we have what we call a safety margin of exposure.

01:23:51.100 Now, how do you know at that point, Lloyd, if toxicology is driven by peak or trough?

01:23:55.460 Because some drugs are-

01:23:57.420 You don't.

01:23:58.020 You don't.

01:23:58.720 You don't.

01:23:59.100 You make a best judgment, but-

01:24:01.220 Is there a general rule of thumb?

01:24:02.820 You measure both and you want both of them to be higher.

01:24:05.360 higher in the animals than what you get in people. Now, this is in general medicine

01:24:10.240 therapeutic indications. In some nasty oncology drugs, toxic effects and therapeutic effects are

01:24:17.800 at the same exposure or even lower sometimes. But the medical need is so great that you accept

01:24:23.660 the toxicity. Well, I was going to actually use that as an example. We sort of skipped ahead a

01:24:28.600 little bit on the phase one patient selection. We rushed through that. Or actually, we didn't

01:24:33.780 answer it. We went off topic. We're going to come back to which patients do you select for the phase

01:24:38.400 one? And that really depends on the drug. Because in an oncology drug, you're going to test it on

01:24:43.840 the most recalcitrant cancer patient, right? You're going to test it on a patient who's

01:24:49.100 progressed through every therapeutic. They have stage four version of whatever cancer you're

01:24:53.920 testing. And this is the Hail Mary. And you're not just testing for tolerance and side effects.

01:25:00.020 you're hoping to get a sliver of efficacy through dose escalation.

01:25:04.080 That's the most common scenario in Kassar.

01:25:06.480 But here, what are you doing? 0.85

01:25:07.620 Are you going out to the most frail, sarcopenic, elderly person,

01:25:11.040 or are you going to test it in?

01:25:12.520 So what I do personally,

01:25:15.340 and a very experienced drug developer named Bob Schmouter taught me this,

01:25:19.980 and I think he's right,

01:25:21.900 is ideally you'd like to test this, a new medicine,

01:25:27.760 in the cleanest population you possibly can where anything you measure is related to the drug and

01:25:35.580 not some underlying disease or other thing. However, we do not want to expose healthy

01:25:41.780 volunteers to risks if we possibly can. So we use our best clinical judgment to say that

01:25:50.700 I don't want to expose people to a risk greater than that of a lightning strike in a year.

01:25:57.760 Is that literally a probabilistic formula you use?

01:26:01.440 That's what I use.

01:26:02.240 Interesting.

01:26:02.580 Yeah.

01:26:02.840 So the risk of being struck by lightning in the U.S. in a year is about one in 100,000.

01:26:10.160 That's actually higher than I would have thought.

01:26:11.680 Me too.

01:26:12.460 That's a little scary.

01:26:13.320 But that's what it is.

01:26:14.180 Okay.

01:26:15.720 And so I don't want the risk of something bad happening to one of my volunteers to be greater than that.

01:26:24.000 And that's frankly how I explain it to him.

01:26:27.020 So if I can have some confidence that that's true, we will test drugs in healthy volunteers.

01:26:33.260 If we're worried about a toxicity or a risk, then we will go into people who have a potential benefit from the therapy so you can make a risk-benefit argument.

01:26:44.840 And this is all laid out in plain English in the consent forms.

01:26:48.560 I mean, so first of all, that's a great framework, Lloyd, which is if the risk of adverse event is greater than 1 in 100,000, we must move to a population that is going to potentially get benefits to justify it.

01:26:59.920 Yes, serious adverse events.

01:27:00.520 Yeah.

01:27:00.980 Is that a Lloydism or is that a truism across the entire industry?

01:27:04.880 Is that something the FDA would ask of every company?

01:27:07.000 It's a schmouterism, Bob, if you're listening, thank you.

01:27:11.960 But the FDA doesn't force that?

01:27:14.620 The FDA doesn't force that, but the principle is still there.

01:27:19.300 Okay.

01:27:19.880 But it's a great standard.

01:27:20.980 Yeah.

01:27:21.740 Again, and if you think about the industry as a whole, how do we do in bringing new medicines into healthy volunteer populations?

01:27:31.820 I've been in this business partially in academia, wholly in industry for maybe 30 years total where I've been watching this.

01:27:40.020 And about once every 10 years, we see something serious happen to healthy volunteers.

01:27:47.980 Once every 10 years in a study.

01:27:49.900 So that's pretty good.

01:27:52.500 And we learn something when those happen.

01:27:54.700 So we're talking about therapeutic antibodies.

01:27:57.160 The one that comes to my mind and maybe to others is the Tegeneron incident.

01:28:02.000 Say more about that.

01:28:02.840 I don't remember that.

01:28:03.700 This was a therapeutic antibody that was directed against CD28.

01:28:07.900 It was an agonist antibody.

01:28:08.920 Now, CD28 is an inhibitory receptor on T cells, and the idea was, I'm sorry, is an activating receptor on T cells.

01:28:16.600 And I forget the actual therapeutic indication they were going for, but they tested the antibody preclinically.

01:28:23.140 Everything was fine.

01:28:25.000 And then they started at a very low dose in humans, and what we think happened is they cross-linked the CD28 receptor,

01:28:32.660 and they had extremely strong T-cell activation and an acute cytokine release syndrome in healthy

01:28:40.040 volunteers, some of them died. I mean, it was terrible. I mean, why did more than one of them

01:28:46.740 die? In other words, why didn't they figure this out the very first time they administered this?

01:28:50.640 That's super important. So that study, which that happened, boy, more than 20 years ago,

01:28:58.300 That experience is why ever since we typically have sentinel patients in dosing cohorts when

01:29:05.020 we're bringing something brand new into people.

01:29:07.860 So they dosed, I think, six people at once with the active drug.

01:29:12.300 Oh my God.

01:29:13.020 We don't do that anymore.

01:29:14.840 Yeah, wow.

01:29:16.280 The other one, there was a example with a small molecule, Bial, I think was the example.

01:29:25.440 B-I-A-L, people can look it up.

01:29:27.160 But it's, it's extremely uncommon to have healthy volunteers have anything bad happen

01:29:32.440 to them in a, in a drug study.

01:29:34.720 Extremely uncommon.

01:29:35.600 If you think of the.

01:29:36.920 I remember there was one at Hopkins when I was there.

01:29:39.620 It was, it was an, oh no, no, no.

01:29:43.360 You know what it was?

01:29:43.920 I'm sorry. 0.99

01:29:44.500 That was not a, it was a woman that a healthy volunteer that underwent a bronchoscopy and

01:29:48.420 I think had a horrible bronchospasm.

01:29:51.800 So it was, if I'm remembering it correctly, it wasn't a drug that caused the issue, but

01:29:56.660 it was a horrible adverse event to procedures but anyways but she she died procedures can have

01:30:02.760 yeah adverse consequences which which are known and disclosed in the in the consent this gave me

01:30:07.980 a lot of when i was in medical school i was you know i was so broke and doing anything i could

01:30:16.440 to generate a buck i vol i was probably one of the most volunteered people for studies at stanford

01:30:23.240 and like if, if there was a study that paid a thousand dollars, it didn't matter what it asked

01:30:28.640 of me, I would do it. And I, I remember coming away from that. I mean, I had radial lines in my,

01:30:34.060 I had, you know, as you know, what a radial line is, but arterial lines into my radial arteries

01:30:39.140 that to this day, I still have scars over my wrists. Can you, can you imagine that I subjected

01:30:44.580 myself to that? That, that seems a little much. My, I think many of us as medical students

01:30:50.460 volunteered for this stuff. My personal favorite was, there was, there was a study called brain

01:30:56.220 electrical activity mapping that children's hospital was running when I was a medical student.

01:31:01.260 And essentially they, um, they attach electrodes to your head and then you go sleep in the lab

01:31:07.180 and they monitor that and video, video you while, while you're sleeping. They loved me as a subject

01:31:13.100 because I was bald as a medical student. It was really easy to put electrodes on and off.

01:31:17.320 I loved it because all I had to do is go in and go to sleep.

01:31:20.940 But there were some others like inhaling radioactive microspheres.

01:31:25.280 I used to donate plasma via plasmapheresis as often as I could when I was at the NIH.

01:31:30.940 And on one occasion, I was in there, and this was like a lymphocyte plasmapheresis.

01:31:35.380 It's a four-hour procedure.

01:31:37.540 And again, it probably paid 200 bucks, which seemed like-

01:31:41.480 It was a lot of money in those days.

01:31:42.460 When you're a medical student, that's infinite money.

01:31:45.160 and then at one point somehow the nurse stepped out and the lab locked and i was stuck in their

01:31:51.640 lock and they could not find a key so they couldn't get back in they were losing their

01:31:56.720 minds but i didn't know it i was just in there watching whatever movie was on the thing it

01:32:01.860 turned out to be like one of the most stressful moments in the in the nci history you know trying

01:32:07.440 to figure out a way to get a spare key to get into the lab and and think and i was completely

01:32:12.940 oblivious to it. Oh, yeah. Okay. So back to patient selection. So ultimately for BEMA,

01:32:19.480 ultimately for BEMA, a healthy volunteer study. You did go with healthy volunteers?

01:32:24.600 Older volunteers. Okay. So people- What was your criteria specifically in terms of muscle mass?

01:32:31.000 So I actually didn't run these studies. The clinicians involved were Dan Rooks and Ronan

01:32:35.500 Rubinoff at the time. But the principle here is healthy volunteers doesn't necessarily mean

01:32:42.860 your 20 something year old with no problems. It means people without typically diagnosable

01:32:51.260 disease or concomitant medications that could confuse any, any assessments. In the case of

01:32:57.620 BEMA, because we were thinking older adults, these were older healthy volunteers and people

01:33:04.760 in whom we would be able to measure some of the effects of BEMA, we hope. And again, what you

01:33:10.920 measure in a healthy volunteer study depends on what the drug is expected to do and what adverse

01:33:16.720 effects you might expect. So there's some things you always do, like a set of standard blood tests.

01:33:22.060 But in the case of BEMA, we were assessing people's muscle mass.

01:33:26.160 Via DEXA?

01:33:27.320 And strength. This is almost archaeology at this point, thinking of what we measured, but

01:33:33.800 we would have measured muscle mass, and at different times we used MRI and we used DEXA.

01:33:38.660 I don't remember what that study had, but we would have measured muscle mass. We measure

01:33:44.900 soluble muscle proteins in the blood like CK and aldolase and LDH and so forth.

01:33:50.680 Did you see any adverse effects in the phase one?

01:33:54.040 Yes. So the three adverse effects that are evident with BEMA that we think are on target

01:34:00.980 that were assessed in that study, where muscle spasms were cramps.

01:34:09.260 Acne is rare in older adults, and it was rare in this study because those were older adults.

01:34:14.620 But skipping ahead, when we've tested younger people, acne is more common.

01:34:19.000 We don't understand why.

01:34:20.640 And then there are GI symptoms of diarrhea that happen.

01:34:24.540 They tend to be first-dose-related and less common subsequently,

01:34:28.180 but they're reproducible and we think they're real. And we saw that stuff.

01:34:32.880 And how many steps of dose escalation did you do in that study? By the way, if that's too much

01:34:37.300 detail to remember, don't worry about it. But I'm wondering if you remember how high you got

01:34:41.280 relative to what was an efficacious dose. So there's some principles here. I personally

01:34:48.280 like to dose as high as we can in the first in human study to understand if there is going to

01:34:53.660 be any safety and tolerability issues in people, while at the same time never exceeding the

01:35:00.120 exposures we've tested in animals. So the study designs include the opportunity to go as high as

01:35:06.160 we can. Typically in antibodies, that ends up being as high as is feasible. And there are a lot

01:35:13.680 of technical details here we don't need to get into, but when your antibodies are made from cell

01:35:19.120 culture, and they're highly purified, but there's still some measurable contaminants in them. And

01:35:25.580 the amount of contaminants, of course, they're also tested as part of the toxicology studies

01:35:31.500 because they're in the drug we give the animals. But we can't exceed the exposure to the contaminants

01:35:37.440 either in the clinical study. So sometimes it's the volume we can administer, the mass we can

01:35:43.580 administer and so forth. So I think the highest dose that we ended up doing in Bimagramab was

01:35:49.240 something like 50 to 100 milligrams per kilogram, but I don't remember.

01:35:55.940 And in that study in humans, you're administering once a month?

01:35:59.100 Initially, you'd administer once, then based on the emerging results for how long that lasts.

01:36:05.780 And we knew what exposures we needed to reach in order to get maximal efficacy based on the

01:36:11.400 culture data. There were a lot of cell culture experiments we did that we haven't talked about.

01:36:16.580 You can culture muscle cells in a dish, and we did that and looked at the ability of the drug

01:36:22.040 to cause hypertrophy of those cells. So we knew what exposures we needed to get to and how long

01:36:28.180 we wanted to do it. But again, we don't exceed the exposures that we get in animals. So I think

01:36:34.560 After the single-dose study, we probably did three doses, and that was it for the first

01:36:40.960 in-human study.

01:36:42.720 And typically, you need multiple doses in order to be able to see the technical term

01:36:48.920 is pharmacodynamic effect, so the effects on the body that the drug causes.

01:36:54.280 So the end point for the phase one, before you move to phase two, where you're really

01:36:59.480 going to actually look as your primary, and you're always looking, obviously, for safety,

01:37:04.060 but now you're really pivoting to efficacy being the thing that you're trying to chase.

01:37:08.880 What did you need to submit to the FDA to say, okay, have we checked our phase one box?

01:37:15.220 Typically, you're in reasonable communication with regulators, whether it's the FDA or whether

01:37:20.160 you're overseas elsewhere, and you provide them with a report.

01:37:24.900 And Novartis is a European company, right?

01:37:27.160 Yes.

01:37:27.480 Based in Switzerland.

01:37:28.200 But this work was being done in the U.S.?

01:37:30.360 I think we did do the first in-human study in the U.S., yeah.

01:37:33.540 Any reason for that? Are European and U.S. regulators so comparable on this point that

01:37:38.780 it's really just a question of where your teams are? So every country is a little different.

01:37:44.400 Europe is somewhat homogeneous, but not completely so. Every country is a little different.

01:37:50.620 And I think it's true both for large companies and small companies that you go wherever makes

01:37:55.380 the most sense. It's where you can, remember the three biggest challenges of any clinical study

01:38:02.020 are recruitment, recruitment, and recruitment. So you have to be able to get the subjects or

01:38:07.380 the patients or the participants. You need qualified, experienced, reliable clinical

01:38:13.820 investigators. You need a regulatory environment that's supportive for what you're trying to do.

01:38:20.340 And then you think about cost of the study. They're different in different countries.

01:38:24.580 And so you integrate all of that stuff. And that chooses where, at least personally,

01:38:29.980 where I would go to do a first in human study. Countries that are often used nowadays are

01:38:36.820 Germany, Australia is pretty popular, New Zealand. Australia, New Zealand's very popular now.

01:38:43.540 Things have really changed, I guess, maybe in the past year or two about China being really popular

01:38:48.920 because China has a regulatory environment that's become more favorable and they can do

01:38:57.960 investigator-initiated studies with less supporting data than we require for a typical

01:39:04.320 IND. So it can often be a faster way to test something. And of course, China has a lot of

01:39:10.560 patients. So that's something that's being done now too. I personally love doing studies in the

01:39:18.400 U.S. and Taiwan. Taiwan has, they have wonderful investigators. They speak English better than we

01:39:25.020 do. They have a very centralized clinical environment, so they have many patients at

01:39:29.820 a limited number of clinical sites. The regulatory environment is very similar to the U.S.

01:39:37.540 Australia and New Zealand is favorable because they have a, especially this is Australia now,

01:39:42.540 they have a different regulatory construct where safety is assessed by the ethics committee and

01:39:49.820 And drug quality is assessed by the regulators.

01:39:53.220 So they have a clinical trial notification process rather than an approval process.

01:39:59.500 And plus the exchange rate's favorable now, so too.

01:40:02.720 So if we get back to what is the trial going to cost, that's useful.

01:40:06.640 And how much reciprocity is there between agencies?

01:40:09.460 So if you, well, I should clarify the question.

01:40:13.180 You can conduct the trial in Australia, but under the auspices of the FDA where they're

01:40:18.000 issuing, or does it have to be in the U.S. if the FDA is overseeing? If the FDA is overseeing,

01:40:23.460 the study's done in the U.S. Okay. So if you do a study in Europe and get European approval,

01:40:28.260 or you do a study in Australia and get Australian approval, how much of an additional hurdle is

01:40:34.220 there for the FDA to typically approve a drug? So let's talk about running a study versus

01:40:41.280 marketing approval. Very different. So for running a study, if you're doing it in, say, Australia,

01:40:46.540 You apply to the Australian regulatory authorities and the ethics committee for the study, and they do the review and request modifications and eventually approve.

01:41:00.020 And then the studies run in Australia.

01:41:02.260 If you want to then do a study in the U.S., you have to apply for an IND, just as you would if you were doing it any other time.

01:41:10.920 But you include all the data that you got in Australia as well.

01:41:14.260 And if you were doing it the other way around, it would be the same thing.

01:41:17.580 And it's true for any two countries.

01:41:19.500 Meaning, if you had a drug that went all the way to the equivalent of a phase three ready

01:41:24.300 for approval in Australia, and you come back to the U.S. from scratch and say, we want

01:41:29.260 to be able to sell this drug in the United States, they're going to say, submit an IND.

01:41:33.260 The FDA.

01:41:33.560 I think so, yes.

01:41:34.580 Yeah, wow.

01:41:35.420 But how much do you get the shortcut?

01:41:37.720 Would they still make you do a phase one and a phase two, or would they let you go straight

01:41:41.180 to phase three?

01:41:41.820 I would think you could go right to a regulatory study. I mean, to a registration study. And in fact, this kind of thing is often done because typically you do the phase one study somewhere, but rarely in more than one, two, or three countries. And then you can use that data to go to many countries for a phase two and then use that data to go globally.

01:42:04.100 There are a few specific examples where you do have to run a phase one study before you go into that country.

01:42:11.160 Best examples, the best defined examples are Japan. 0.95

01:42:16.560 So to run a large study in Japan, you need to have run a phase one study in Japanese people.

01:42:23.420 And there is a formal regulatory definition of who is Japanese from the Japanese regulators.

01:42:29.760 and you have to provide that data before you can do a larger study in Japan.

01:42:35.320 They're called ethnic sensitivity studies.

01:42:38.260 And scientific rationale for this is that the genetic background of Japanese people

01:42:43.860 can be a little different.

01:42:45.380 Average body size is often different from people in the West

01:42:48.420 and you want to make sure that the dosing and exposure will be safe and tolerable.

01:42:53.260 But because the Japanese are so well organized and specific about what a Japanese person is, you can do these ethnic sensitivity studies in Hawaii, for example, or even California, or you can do them in Japan.

01:43:10.260 And I like to do them in Hawaii.

01:43:13.460 China generally requires an ethnic sensitivity study also for the same reasons.

01:43:18.260 And there are some specific examples of where there's toxicity of drugs in people on Chinese

01:43:24.700 ethnicity, but they have a less specific definition of who's Chinese. Easiest way to do it is in China. 0.98

01:43:34.960 All right. So let's go back to BIMA. 1.00

01:43:37.060 BIMA.

01:43:37.900 You go into phase two now. By the way, at some point, doesn't Novartis sell this asset?

01:43:43.440 Yes. So Novartis had strong confidence in Vimagramab. It was first in class, had really obvious biology in humans. And basically, Novartis ran maybe, I think, 16 phase two studies of one sort or another, or phase one, phase two study, different indications. Tried very hard.

01:44:11.960 So the drug reliably and predictably increases muscle size, but not performance assessments in a major way.

01:44:24.340 And I think that's because, remember in the rodents, in whom we saw both, size increase and performance increase, the mass increase was large, 20 to 30 percent or more.

01:44:35.240 In humans, it's 4 to 8 percent.

01:44:38.600 And 8 is the absolute max.

01:44:40.060 Were those differences based on dose or starting mass?

01:44:45.120 Biology. People are just not mice.

01:44:47.960 Oh, sorry. I mean, the difference between the four and the eight, how much of that is dose

01:44:52.500 dependent versus other demographic dependent on the pay? Like, you know, do you get more muscle

01:44:58.380 mass in younger people, more muscle mass in people starting with more muscle mass?

01:45:02.480 Yeah. There's a trend to more in males versus females, a trend to more in younger versus

01:45:09.340 older, but there's a lot of variability. Do we know if other variables such as resistance training,

01:45:17.540 nutrition, protein consumption would have augmented these findings? And how much were

01:45:22.200 those variables controlled in these studies? So we know some of that. We try to control as

01:45:25.900 much as we can. There was one study that has not been published in peer-reviewed form yet,

01:45:32.660 but there is an abstract for it. If people want to find it, they can look at it. So the BELIEVE

01:45:38.040 study of Bimagramab in obesity was just published a few months ago in Nature Medicine. If you look

01:45:44.440 in there, this nutrition study is referenced. We'll link to it in the show notes.

01:45:49.500 There was a study of Bimagramab in patients who were dosed at three different levels of

01:45:59.140 protein calorie nutrition. And the bottom line is the more protein, and it was the recommended

01:46:07.180 a daily amount, half of that or one and a half times that, I think. And basically within those

01:46:14.560 boundaries, the more protein you ate, the more muscle you built. Probably shouldn't surprise

01:46:19.240 anybody. The other really interesting finding- And by the way, twice the RDA is only 1.2 grams

01:46:24.580 per kilogram. Yeah. Maybe that's what we used. It was 1.2. Yeah. I would argue had you gone to 1.6

01:46:30.120 or 2, you probably would have seen more hypertrophy. It's not been tested. I think you're probably

01:46:34.920 be right, but it hasn't been tested. It's interesting. So it suggests that in humans,

01:46:39.420 you might've been substrate limited. Amino acid limited or protein synthesis limited.

01:46:45.120 It's a possibility. And not drug limited.

01:46:47.500 I'll tell you a funny story about that in just a minute. But just to finish that study,

01:46:51.040 the other very cool thing we found is that as you would expect, if you have

01:46:54.900 half the recommended daily amount of protein calorie nutrients, you lost muscle mass. But

01:47:01.760 Bimagrimab prevented that. So there was some biology working there for sure. No question.

01:47:07.940 No question. So the really interesting story is that when the Bimagrimab project, when it was

01:47:13.620 still in the research stage, moved from Chris Liu's lab to David Glass's department, which I was part

01:47:21.360 of as the clinical side of that, one of the things we really wanted to do was co-develop a nutritional

01:47:26.860 component to this therapy for the exact reasons that you brought up. And at the time, Novartis

01:47:32.880 had a nutrition arm. And so we were working with them to develop a specific nutritional supplement

01:47:41.340 for what became Vimagramab. At the time, it didn't even have a code yet. But then Novartis

01:47:46.680 sold their nutrition unit, I think to Nestle. And so got the rug pulled out from under us on that

01:47:52.380 side. And we were never able to fully pursue that. But in retrospect, I really wish we had.

01:47:58.100 Do you think this is a blind spot for big pharma? Just the role of nutrition and other behaviors

01:48:05.900 that can potentiate drugs? Big pharma tries to control it, but they don't see that as their

01:48:12.320 core mission. Yeah. But I'm saying a blind spot, I appreciate that they want to control it and that

01:48:17.340 makes sense, but I'm saying it's an opportunity lost. Here's a great example. Bima could have

01:48:26.900 been more of a hit if maybe, and maybe not, but a drug like that could have been a hit 0.87

01:48:33.400 had it been appreciated that, oh, by the way, you actually have to do something to reap the

01:48:38.760 benefits of this. We would have figured this out many years sooner if we had kept that nutrition

01:48:43.180 element. But again, one of the challenges of big companies is there's so many people involved. They

01:48:48.560 don't all know what the others are doing despite everyone's best efforts. It was a missed

01:48:54.980 opportunity. And so did Novartis then spin it out after the less than expected findings in humans?

01:49:02.960 Yeah. So what happened was we saw muscle mass get larger, but not stronger. And parenthetically,

01:49:09.240 that's the same as was seen with IGF-1 agonists and with androgen agonists. Remember, the SARMs 0.82

01:49:15.440 were extensively studied, is that you can make muscles larger, they don't get stronger in the

01:49:20.000 absence of resistance training. So it's not unique to the active and receptor antagonist pathway.

01:49:26.560 And in a meta-analysis of the Novartis studies, where they looked at muscle hypertrophy and

01:49:34.600 sarcopenia, the meta-analysis showed an increased six-minute walk distance,

01:49:42.840 nine meters, so a small effect. That's my least favorite test in the world.

01:49:49.540 It's surprisingly hard to standardize. Why wouldn't they just do something like,

01:49:54.100 you know, a wall sit or, you know, something that really tests strength?

01:50:00.100 Many other things were done. The timed up and go test, the short physical performance batteries. But six-minute walk was included in multiple studies, so you were able to do a meta-analysis of that.

01:50:12.980 got it so the an academic group did this and they published it and so the the four to i don't know

01:50:20.840 eight percent increase in muscle mass that these older adults got yielded a nine minute in nine

01:50:27.680 meter meter increased in six minute walk distance i'm not convinced that's going to help anybody not

01:50:33.140 fall no i don't think it will either and and and novartis didn't think so either is i guess i don't

01:50:38.600 I wasn't an insider at the time.

01:50:40.160 I don't know why they out-licensed it.

01:50:41.760 I was the recipient of that.

01:50:43.180 Yep.

01:50:43.760 We were on the outside pulling.

01:50:45.280 But the very last study Novartis did was a study in type 2 diabetics

01:50:50.640 because we had had data that hemoglobin A1Cs decreased in patients given bimagromab.

01:51:00.700 And that study, which ran for 48 weeks,

01:51:04.900 So it was 10 mgs per kg monthly for 12 doses.

01:51:10.760 Okay.

01:51:11.200 So a lower dose than you were giving for hypertrophy?

01:51:14.520 No, this maximized the-

01:51:16.700 Oh, why did I think you said earlier 50 mgs per kg?

01:51:19.640 In the phase one, we went up higher.

01:51:21.300 You went up that high.

01:51:22.000 Got it.

01:51:22.200 We went up as high as we could because we wanted to know what would happen if people were overdosed later.

01:51:26.200 Got it.

01:51:26.660 Okay.

01:51:27.140 And the answer was nothing.

01:51:28.580 Okay.

01:51:29.340 So at 10 mgs per kg monthly over 48 weeks?

01:51:32.480 Yeah.

01:51:32.620 It was a maximal dose.

01:51:34.900 in terms of effect size. And they saw the expected muscle mass increase.

01:51:42.500 Interestingly, they saw a substantial fat mass decrease. And hemoglobin A1c in these type 2

01:51:50.160 diabetics decreased by about 0.7% or 0.8%, absolute, which is a pretty good effect.

01:51:55.960 Yeah. And do you think that that was on account of just more insulin sensitivity,

01:52:00.720 or was it a larger reservoir for glucose disposal?

01:52:04.460 Both of those things, I think.

01:52:07.540 And these patients didn't have to do anything else.

01:52:10.820 It wasn't like, in addition to that,

01:52:12.740 they changed the way they ate or they exercised more.

01:52:15.000 You gave them a drug that added muscle mass,

01:52:17.800 took off fat mass, and lowered A1C by 0.7%.

01:52:20.860 Yep, and they had standardized dietary advice to...

01:52:25.620 Yeah, both groups.

01:52:26.340 The 500...

01:52:27.360 So who...

01:52:27.840 Did Novartis run that study?

01:52:28.960 Novartis did the whole study.

01:52:30.720 And then they made a strategic decision that the effect size wasn't big enough.

01:52:35.600 I mean, actually, I don't know what their strategic decision was, but they decided...

01:52:38.820 The output was we're going to spin it out.

01:52:40.400 Yep.

01:52:40.700 The output was to spin it out.

01:52:42.200 And at the time, I was working with Joe Jimenez and Mark Fishman and Praveena Kandula at Adidam

01:52:49.540 Bio as an advisor, and we really wanted Bimagrumab.

01:52:56.320 What year is this approximately?

01:52:57.600 That it spun out was 2021.

01:53:03.060 Okay.

01:53:04.800 I think it was 2021.

01:53:06.720 Yeah.

01:53:07.240 Discussions had been ongoing in 2020, but I think it finally happened in 2021.

01:53:11.540 Okay.

01:53:11.940 So you guys acquired the asset, obviously, for a lot less than you could have produced it.

01:53:16.700 Yeah.

01:53:17.740 Still wasn't cheap because it was a phase two ready program.

01:53:21.380 Yeah.

01:53:22.120 But we acquired it.

01:53:24.340 Did you guys raise money for that acquisition?

01:53:25.980 No, they did and did it all themselves, to their credit.

01:53:31.760 And the plan was we were going to develop it in older adults with low muscle mass and impaired muscle function,

01:53:41.380 because we thought, who were also obese, because we thought this was the patient population most likely to benefit,

01:53:47.780 Losing fat and building muscle and maintaining muscle in the context of weight loss, we thought

01:53:55.740 would be super important for those people. And remember, all of this happened in the context of

01:54:02.100 nobody being interested in obesity. Everybody thought it was a wasteland for drug development.

01:54:07.960 Every drug that had been developed in obesity had failed commercially. I mean, there are some that

01:54:12.900 it had been registered, right? But by 21, you're saying pre-21? This was before NOVA's semaglutide

01:54:20.200 data came out. Yeah. Okay. That's right. Yep. And I'll tell you- Because that was 21, wasn't it?

01:54:26.360 Later in 21. Okay. So it's November of 21. Yeah. I think February 21, we started Versanus Bio,

01:54:32.200 which is the company that licensed Bimagrimab from Novartis. So at that point,

01:54:38.720 So I was the founding CEO. I was working with Elon Zipkin, and we went out to raise money from investors because now, all right, great, we had this asset. We needed to run a big phase two study, and we went out to raise money. I think we talked to 53 investors. Almost none were interested.

01:54:58.220 Because you told them indication, sarcopenia still.

01:55:02.160 Well, it was sarcopenic obesity.

01:55:04.100 Yep.

01:55:04.300 and obesity was just not a successful area for drug development so people weren't interested

01:55:12.160 how much did you need to raise we ended up how much did we need versus how much we got

01:55:17.400 are different issues yeah i know that's why i asked but we ended up raising 70 million

01:55:22.040 and what was your way if you could have had your wish list what would you have raised

01:55:26.200 About 100.

01:55:27.040 Okay.

01:55:28.480 However, Atlas Venture and Medici liked the story.

01:55:33.240 And I had worked with Atlas before.

01:55:36.220 And Michael Gladstone was the partner.

01:55:39.020 And it was Vani Marigi and Nick at Medici, Nick Williams.

01:55:46.600 And we then built an investment syndicate and they funded the company.

01:55:52.080 And then everything changed when Novo's data came out with semaglutide, which was amazing.

01:56:00.120 It was the first really effective obesity medical therapeutic.

01:56:05.120 But then when you're doing drug development, you skate to where the puck is going to be,

01:56:10.240 as Jay Bradner's favorite saying.

01:56:12.540 But the puck was going someplace else now.

01:56:16.360 Semaglutide was going to become the standard of Clare or some incretin agonist.

01:56:20.240 We knew it.

01:56:20.660 And so what we did is we quickly repositioned the company to think about what is bimagrimab

01:56:30.320 going to do on top of that, because that's going to be the standard of care.

01:56:34.240 So I quickly ran a bunch of mouse studies, and the efficacy was additive when you took

01:56:43.020 bimagrimab with semaglutide or terzepatide or liraglutide.

01:56:48.280 did them all. And the efficacy was sort of unprecedented. Never seen anything.

01:56:53.540 For both fat loss and obviously for preservation of lean mass.

01:56:56.740 Exactly. For weight loss, especially fat loss and preservation of lean mass. It was amazing.

01:57:04.200 So the opportunity became much larger. And I was part of the board. The board then brought in

01:57:12.440 a super experienced CEO, this was Mark Brzezansky, to lead the company then because we had a really

01:57:20.360 big opportunity and we knew it. And we brought in a CMO, Ken Addy, because I had been serving

01:57:26.000 as the CMO also. And then I stepped into president and CSO role just in terms of company organization.

01:57:33.140 But we all kept working on the program. We ultimately ran what became the Believe Study.

01:57:39.480 And the story here, we spent a lot of time thinking about what we would name our studies.

01:57:43.820 The plan was believe was going to be phase two, become was going to be phase three, and behold was going to be post-registration studies.

01:57:53.320 And then you actually have to come up with what those things stand for, knowing only what the B stands for when you start.

01:57:58.520 I mean, this is so funny how drug name studies work.

01:58:00.720 Yeah, you know how it goes.

01:58:01.620 I know the drill.

01:58:02.320 But they all became with B for B-Magrumab.

01:58:05.140 Yeah.

01:58:05.260 Now, you haven't asked me where Bimagramab came from.

01:58:09.860 So this is another interesting, wonky drug development thing.

01:58:14.520 So the generic name is called the INN name for, I think, International Nomenclature.

01:58:20.520 I'm not sure what it's an acronym for.

01:58:22.320 Which is why it ends in MAB, obviously.

01:58:24.100 So the suffix of a drug generic name is pre-specified based on the class.

01:58:30.900 If you're the first in class, pick a new one.

01:58:33.940 but the company gets to recommend the prefix and sometimes the infix. So Bima is the Indian god 0.81

01:58:46.040 who's as strong as 10,000 elephants. And that's why Bimagrimab is Bimagrimab. 0.88

01:58:52.540 And then ends in Mab, monoclonal antibody.

01:58:55.680 The Grumab is a monoclonal antibody suffix.

01:58:58.280 So any other drug that comes along in that class, what would be the nomenclature naming options?

01:59:04.660 Antibodies are just complicated. There's a lot of different criteria for naming antibodies and

01:59:09.360 you have options for infixes and suffixes. And what was the GLP-1 before liraglutide?

01:59:17.260 There's exenotide.

01:59:18.520 Exenotide, right. So the tide became the thing that everybody had to link to going forward?

01:59:24.680 Tide is peptide.

01:59:26.020 So were they forced into liraglutide, semaglutide, terzepatide?

01:59:33.200 The Tide is used for that class, but there are other peptides that end in Tide.

01:59:37.900 Yeah.

01:59:38.740 Okay.

01:59:39.220 So you guys ran Believe.

01:59:41.560 Yes.

01:59:41.920 So Believe, it was originally planned to be a 24-week study in patients with sarcopenic

01:59:50.640 obesity.

01:59:51.180 That was what we were going to do.

01:59:52.440 But then when Novo's data came out and we got super excited about obesity, we said, oh, my God, we've got to do a bigger study.

01:59:59.640 We're going to do it in combination with semaglutide.

02:00:02.300 And this was during the pandemic.

02:00:04.880 So there was a lot of complexity and supply chain disruptions.

02:00:09.020 And remember, with semaglutide, it was a proprietary drug of Novo.

02:00:14.620 We couldn't get the drug substance.

02:00:17.620 so we had to use the commercial presentation of semaglutide, which was expensive, and it's an

02:00:25.360 auto-injector. We couldn't make a placebo for that, so the study design included semaglutide

02:00:31.280 as open label, but we did placebo-controlled Bimagromab because we had control of that.

02:00:39.140 We used Bimagromab intravenously just because it was the fastest, most straightforward way to get

02:00:45.920 into the clinic, not public, was that we were working hard on an auto-injector and we would

02:00:51.840 have been ready in the next study for an auto-injector, but it was intravenous for

02:00:56.500 bimagromab. And then what combination should we use? We ended up doing something called a

02:01:03.540 full factorial design. So we did all possible combinations of low-dose semaglutide, high-dose

02:01:11.180 semaglutide, low-dose spemagromab, high-dose spemagromab, and placebo. So it's a nine-arm

02:01:17.020 study. Why? Because we didn't know in humans what would happen with those different combinations.

02:01:24.900 We didn't know if there would be adverse effects of the drug combinations or not.

02:01:28.720 They did have a couple of adverse effects in common, diarrhea, for example. And this is in

02:01:35.420 part why I did that pharmacology study in rodents, because if there was anything unexpected that

02:01:39.220 what happened with the drug combination. I wanted to know about it. Technically,

02:01:43.520 actually, we, and we had to do this because we weren't using the semaglutide in its

02:01:50.520 intended population. I think it, it wasn't registered yet, I guess was the issue.

02:01:56.460 We had the data, but it wasn't registered. So it wasn't indicated in obesity.

02:02:00.540 So if you're using it. And so, yeah, you were using Ozempic, not Wagovi. You were still using.

02:02:04.660 We used them both. Oh, you did.

02:02:06.080 Whatever we could get. Remember they were in short supply.

02:02:08.160 And to the regulators' credit, they recognized all of this and were willing to allow us to substitute interchangeably Ozepic and Mugovie.

02:02:22.680 So what were the findings of this six-month, nine-arm study?

02:02:27.680 So it was originally going to be six months, but we changed it from the original plan with

02:02:33.480 Bimagramab alone to the combination, and it eventually became 72 weeks of treatment,

02:02:39.820 48 weeks was the primary endpoint, and then we had a six-month follow-up period. So 104 weeks

02:02:46.200 total study, two years. Did you have to raise more money? We did. Yeah. I was going to say,

02:02:52.240 that's a hard study to do for 70 million bucks. Yes, we did. And it was 500 people

02:02:57.200 enrolled roughly 507 was the exact number so

02:03:02.120 we found so i guess number one and one of the things i'm kind of proud of is i got the doses

02:03:11.700 right because you wanted to see a partial response with the low dose a full response with the high

02:03:18.380 dose for both of the drugs and see those kinds of dose responses in the combination arms remember

02:03:24.000 there's four combination arms, right? There's low-dose bema, high-dose bema, low-dose sema,

02:03:29.040 high-dose sema, four combinations, and then placebo. And we saw the dose effects in all

02:03:34.080 the arms. So that was good. And the primary endpoint was body weight. It wasn't what I

02:03:40.620 wanted for a primary endpoint. I wanted waist circumference.

02:03:44.660 Why couldn't you get DEXA? Too expensive.

02:03:47.100 And we thought that since the registration decision is made on the basis of body weight loss, we wanted that as the primary endpoint.

02:03:57.840 We included DEXA in every single patient.

02:04:00.940 Yeah, it's crazy to me that you would be held to the standard of weight loss when in reality a better outcome might be less weight loss.

02:04:11.360 That is true.

02:04:11.960 If you're preserving muscle.

02:04:13.060 We were acutely aware of that.

02:04:14.500 Yeah, that's awful.

02:04:15.420 But it is not what the field was thinking at the time.

02:04:19.160 No, I know, but it's just, I mean, it's good biology abuts regulatory simplicity for Lutman.

02:04:27.000 Yeah, yeah.

02:04:27.800 I personally wanted waist circumference, and I wrote a long white paper about this because waist circumference is more closely linked to important clinical outcomes than is BMI or body.

02:04:40.920 Yeah, for sure.

02:04:42.220 Yeah.

02:04:42.340 So body mass index is, if you're following longitudinally, is essentially the same as body weight because height doesn't change over the short term, just for listeners.

02:04:54.820 So we ended up using body weight as the primary endpoint.

02:04:58.560 And it wasn't just regulatory intransigence.

02:05:01.660 It was also what do investors and potential acquirers think.

02:05:06.800 Everybody cares about what the approval endpoint is going to be.

02:05:09.680 We wanted that to be the primary endpoint.

02:05:12.340 but we measured all these other things. And to sort of zip ahead to the end, in the high-dose

02:05:18.220 combination group, the body weight loss at 72 weeks was 22, 23% of starting body weight.

02:05:25.220 High, high.

02:05:26.440 Yeah.

02:05:26.640 The double positive.

02:05:27.220 The high, high combination.

02:05:28.480 And what was the high?

02:05:29.600 But the fat loss was 45.7% of starting body fat. Now, that's what you get with bariatric surgery.

02:05:38.400 So this, to me, this is the first medical therapy that gives fat loss equivalent to or superior than bariatric surgery.

02:05:50.420 That's amazing.

02:05:51.600 Did you do any functional testing in that study?

02:05:54.800 We did.

02:05:55.440 We did.

02:05:56.200 And we even did a preliminary observational study in overweight adults at different age cohorts.

02:06:04.260 And we tested a few different things.

02:06:06.120 We tested essentially timed up and go for short physical performance battery.

02:06:12.860 We tested the 30-second chair stand test, which is my personal favorite.

02:06:18.100 And we tested grip strength.

02:06:19.780 But ultimately, we went with grip strength in the BELIEVE study because it was the one most closely linked to clinical outcomes.

02:06:26.840 And did you see an improvement in strength?

02:06:28.820 Small, but also it was a variable assessment.

02:06:32.260 Okay.

02:06:32.520 it's and it's in the published study that came out a few months ago

02:06:36.860 now novo bought this asset from you guys right no lily did oh lily did yeah okay so we were

02:06:45.240 super excited about the study it was ongoing we were enthusiastic

02:06:49.880 we closed a series b in two tranches and we called the first tranche

02:06:58.640 And then the company got bought by Eli Lilly.

02:07:01.240 And so they have Bimagrimab now.

02:07:04.160 What are they doing with it?

02:07:05.900 You have to ask Lilly.

02:07:07.740 But they made some noise last fall that they were pausing the program or-

02:07:12.060 No, they just, they paused one study, but they still have other studies in clinicaltrials.gov.

02:07:18.560 Okay.

02:07:19.300 But you got to ask them.

02:07:20.920 I see.

02:07:21.820 So publicly, the only thing we know is they're still doing something with it, presumably

02:07:27.260 testing it with terzepatide, I'm guessing, or retitrutide.

02:07:30.340 The study that's in clinicaltrials.gov is a complex combination study with terzepatide.

02:07:36.060 I do think it's fair to mention the one adverse outcome that happened in the BELIEVE study

02:07:42.120 that we weren't really expecting, which is an increase in LDL.

02:07:46.740 Yeah.

02:07:46.960 How much I remember that.

02:07:48.000 Now, I thought that was actually something that Lily saw, but that was in your study.

02:07:53.760 Yeah.

02:07:54.000 And how much of an increase was it?

02:07:55.500 It was about 20%.

02:07:56.520 Why do you think that was biologically?

02:08:00.480 It's a direct effect of the drug in the liver.

02:08:03.080 So interesting.

02:08:04.280 Again, you wouldn't expect this off target, would you?

02:08:07.740 This is on target, because remember, there's active in receptors everywhere.

02:08:12.480 Oh, I didn't realize that.

02:08:13.400 Yeah.

02:08:13.560 Ah. So it's doing something to interfere with LDL clearance, presumably.

02:08:21.580 I guess, but I don't know what the biology is. It hasn't been studied to my...

02:08:25.340 Or no, maybe not. I mean, what would be a more... I mean, that would be studyable, right? Is it

02:08:29.880 impeding LDL clearance or is it increasing LDL synthesis?

02:08:32.900 I know who knows the answer to this. So Chris Liu, when he left Novartis,

02:08:37.140 eventually founded a biotech called Lakena in China. And he's made therapeutic antibodies to

02:08:44.040 active and receptor type 2A, type 2B, and the combination, and he studied them. So he knows

02:08:50.340 the answer to this, and I assume he'll publish it at some point. And Lloyd, were there any adverse

02:08:55.120 effects on glucose in the other direction? Did glucose ever go up? No. Okay. And did you continue

02:09:01.100 in the belief to see glucose go down the way you did in the diabetic studies? Yes. Independent of

02:09:06.220 what you would have seen from SEMA, I mean. Yes. So we have all of that data and it's published

02:09:11.260 in the Nature Medicine paper. At EASD in September, which is a European meeting,

02:09:17.820 we're going to publish the results of the six-month off-drug results. So that's-

02:09:24.940 That's off both drugs.

02:09:26.640 Yes. And so the real issue is what's going to happen when you withdraw the drugs?

02:09:31.940 and we know what happens when you withdraw semaglutide. Everything goes back towards

02:09:37.600 where it was. It doesn't quite get there. And we're going to find out with bimagrumab.

02:09:42.380 I expect some things are going to reverse. We know that muscle mass with every muscle anabolic

02:09:49.040 agent reverts towards baseline when you withdraw the therapy. I expect that's going to happen

02:09:53.440 in the humans. It happens in the rodents. In BELIEVE, we deliberately included patients

02:09:59.660 with metabolic syndrome. So these are people who are pre-diabetic. So we can measure diabetic

02:10:06.460 endpoints in these people. And it's going to be super interesting to see what happens there.

02:10:11.180 Personally, if we had kept Vimagrimab in Bersanus and had been a standalone entity,

02:10:17.000 we would be well advanced into phase three by now. And the reason is because I believe,

02:10:23.360 even with those LDL effects, which are not favorable, LDL predicts adverse cardiovascular

02:10:29.220 endpoints. But I believe that- Because you can monitor it and you can treat it.

02:10:33.520 Yes. As an aside, since my personal interest is making drugs to prevent the most common

02:10:43.800 causes of morbidity and mortality in older adults, side effect of that is healthy longevity.

02:10:50.080 That's what I do. I am not making cardiovascular drugs, even though it is the number one cause of

02:10:56.600 morbidity and mortality in adults in the U.S. and in many developed countries around the world.

02:11:01.320 The reason is we've already got a lot of good drugs. We're just not using them for primary

02:11:05.660 prevention, which we need to be doing more of. So with that aside, I would be well advanced in

02:11:12.320 developing Vimagrimab in phase three, but I think the paradigm for managing obesity is going to be

02:11:20.100 induction and maintenance of remission, probably combination and injectable therapies,

02:11:26.600 to get people to move them categorically from obese to non-obese. And then they need something

02:11:34.400 for maintenance, which might be something like Orforglopron or some oral GLP-1 agonist to

02:11:40.420 maintain appetite and satiety. And you don't think just a lower dose of the injectable could do?

02:11:46.740 Absolutely, it could. Yeah. You're just saying economically,

02:11:48.940 it might be easier to make it orally or something. Exactly. Yeah. Exactly.

02:11:51.460 Okay. I want to pivot and talk about one other thing, which is also an area where you know a lot about it, which is selective mTOR inhibition. We're not going to spend as much time on it, of course, but again, talk to me about where your head is at these days on that pathway in general.

02:12:10.380 Do you believe that there are, that this is geoprotective in humans? I mean, it's been well

02:12:18.400 established how geoprotective this is in mice. Almost assuredly, I think it'll end up being

02:12:24.620 geoprotective in dogs. So it might be safe to say that inhibiting mTOR in everything from yeast to

02:12:32.100 dogs and maybe even primates extends life. We don't have a clue if it's going to in humans.

02:12:37.860 we'll never probably get to directly test it. There are really good, compelling arguments on

02:12:43.460 both sides of why it may or may not be the case in humans, including the longevity quotient

02:12:49.600 argument and things like that. What are your thoughts? I think it probably will. It's highly

02:12:55.500 conserved biology across evolution. So I think so. Reductively, if you envision mTORC1 as a master

02:13:05.120 regulator of sensing, sort of integrating nutritional inputs and then deciding to grow

02:13:11.920 or not grow. And not grow means circling the wagon, upregulating autophagy and recycling

02:13:17.780 pathways. I think it probably would. I think the effect size is going to be modest.

02:13:23.720 Is it just as it has been preclinically? And I think it's TORC1. I haven't seen a lot of new data

02:13:30.720 on that. Well, there was that study somewhat recently suggesting that rapamycin impaired,

02:13:38.700 I don't know if it was impairing MPS or some other metric of physical performance or something.

02:13:44.700 Obviously, you're familiar with the agents you've tested. We're still at sort of the infancy of

02:13:52.200 these drugs, right? What do you think is standing in the way of more drug development on more and

02:13:57.140 more selective, potentially higher efficacy, but potentially lower side effect burden versions of

02:14:03.720 drugs that can inhibit mTOR complex 1? The selectivity is the big challenge,

02:14:11.260 because with rapalogs, as you know, their TORC 1 is selective, but there's a downregulation of

02:14:20.600 TORQ2 with sustained exposure. And I don't know that we, you know, so in RestoreBio, we tried to

02:14:28.840 manage that by a combination of a catalytic and an allosteric inhibitor, which seemed to do it.

02:14:34.880 And I know there are other companies that are working on other ways to get TORQ1 selective

02:14:38.820 inhibition. And I think that's what we need is a real TORQ1 selective inhibitor. And then we can

02:14:43.760 test the biology. And do you think that just intermittent dosing of Everolimus or Sirolimus

02:14:49.820 gets that? Maybe. Again, it's hard to tell in healthy people because, you know, in cancer,

02:14:58.300 when you study mTOR inhibitors, the cancers have a highly upregulated pathway and it's easy to see

02:15:04.080 the biology. You can't really see the active biology in humans measuring blood. It is not

02:15:09.460 necessarily the tissue you want anyway. When we do this in rodents, we measure their liver activity.

02:15:14.100 And I think we mentioned, you and I discussed this before, that in young rodents with fasting, they down-regulate mTOR, as you would expect.

02:15:24.180 In old rodents, they didn't.

02:15:27.000 So it makes me call into question the whole concept of intermittent fasting in older people, because I don't know if it'll do the same thing.

02:15:36.940 Yeah.

02:15:37.720 Again, imminently testable.

02:15:40.280 Nobody's lining up for liver biopsies, though.

02:15:42.740 No.

02:15:42.960 And we can't get that with MRS or anything else?

02:15:47.120 I don't think so.

02:15:48.060 It's just...

02:15:48.680 Yeah.

02:15:49.300 Again, the price of admission is so great on that.

02:15:53.000 Okay.

02:15:53.600 Final question slash topic.

02:15:56.360 As you think about drug discovery over the next decade, I'm not going to ask you the

02:16:01.160 question everybody's thinking is, how is AI going to help?

02:16:03.580 We'll punt that for now.

02:16:04.780 Thank you.

02:16:05.280 Yeah.

02:16:05.440 What are you most optimistic about in terms of pathway, disease?

02:16:14.700 Where are you most excited?

02:16:16.360 Where do you think we're going to be in 10 years where there's been a step function change?

02:16:20.280 I think we're starting to wake up to the concept of real medicine, real medical prevention.

02:16:26.500 I mean, this is something I've been saying for years.

02:16:29.500 you've talked about it a lot, is that we need to get away from being a sick care system to a

02:16:35.780 health care system. And the way you do that is with preventive medicine. And the way to implement

02:16:40.920 it is you need better primary care and you need codes for preventive visits. Because right now

02:16:48.060 if I wanted to see a patient for prevention of cancer, for example, so my new company is

02:16:54.300 cancer prevention. There's not codes for that. So you can't bill for it. So there's, there's a lot

02:17:02.100 of institutional hurdles that we need to get through. But I think people are waking up to

02:17:10.000 the concept of, I want to stay healthy rather than get sick and get treated.

02:17:14.620 Say a bit more about your current company and how, how could one develop a drug for cancer

02:17:19.540 prevention. Yeah. So this is conceptually difficult to wrap your head around. But again,

02:17:26.540 where do new drugs come from? They come from reading the literature and thinking, which is

02:17:31.000 sort of what I did after Versanus ended for me. It's still ongoing in Lilly. And there are some

02:17:39.180 papers published over the past five to 10 years about drugs that cause cancer. So if a drug causes

02:17:48.220 cancer, it must be most likely inhibiting a cancer-protective pathway. Most drugs are

02:17:54.020 inhibitors of things. The prototype for this is serafinib, which is a multi-kinase inhibitor

02:18:00.660 that's used to treat renal cell carcinoma and hepatocellular carcinoma primarily. If you give

02:18:06.820 that drug to people, about 10% of the older patients get skin cancers. Why is that?

02:18:13.160 And are these melanomas or are these squamous or basal cells?

02:18:17.180 The cancers they get seem to be the prevalence that's reflected in the normal population.

02:18:23.100 So almost everything that's ascertained is basal cell and squamous cell.

02:18:27.440 And more recently, we understand the pathway biology of that.

02:18:32.300 Serafinib is a multi-kinase inhibitor.

02:18:34.800 It inhibits a lot of kinases.

02:18:37.300 But one of the ones that it inhibits is the sensing kinase that triggers something called

02:18:42.300 ribotoxic stress. And this is a pathway that causes cell death. That pathway, if you turn it

02:18:49.820 on irreversibly and covalently, is the target of some of the nastiest toxins that you know about,

02:18:56.360 like diphtheria toxin, sarsin, ricin. It's a very, very potent pathway. My innovation is putting

02:19:05.840 together different parts of the literature. I came up with a way to turn it on in a gentle

02:19:12.000 in controlled fashion. Remember, it's on constitutively in people because if you turn

02:19:17.060 it off with these multi-kinase inhibitors, you get cancer. So the hypothesis of the company is

02:19:24.500 if we turn that pathway on a little more, we'll prevent cancers. Not all of them, but maybe 50%

02:19:31.200 is what I'm hoping. And since skin cancer is almost as common as all other cancers put together,

02:19:37.760 We got to start there. But in a phase two study of older adults and older adults in this context

02:19:45.700 means 50 and up. Sorry, Peter. I'm in that category squarely. Don't worry. I have been for a while

02:19:51.680 and who have had at least five skin cancers in the past. Those people have a 50% chance of having

02:19:57.800 another skin cancer within a year. So if we recruit a cohort of 100 or 120 of those, we would

02:20:03.820 be able to test the low dose, high dose, placebo, and actually measure cancer prevention in a phase

02:20:10.320 two study. Now, is there a risk that it will only work in preventing squamous cell and basal cell

02:20:16.160 carcinoma, but will not progress an epithelial tumor? Or prevent, I'm sorry. That's possible

02:20:23.080 because the only data we have are for skin cancer. But even if it only prevented skin cancer,

02:20:28.780 That's a really big medical need, but I think it will work on multiple cancers, but it's

02:20:34.740 going to be almost impossible to test that before approval, just because cancer incidence

02:20:41.340 is a really rare event.

02:20:42.740 Yeah.

02:20:43.180 Yeah.

02:20:43.620 I guess the next thing that would be an interesting question, Lloyd, would be you take a bunch

02:20:50.520 of patients who have successfully undergone adjuvant therapy for a stage three epithelial

02:20:57.960 cancer. So I would think colon cancer or breast cancer. They're NED, no evidence of disease for

02:21:06.940 the listener, but there's a 50% chance they're going to have a recurrence. You know, you stratify

02:21:13.640 it in a way that you, you basically find people who have a very high risk of a cancer recurrence

02:21:17.580 and then you, you treat them. I think that's one way to do the other study.

02:21:22.660 What I've done in COSLAPE therapeutics is a collaboration with the Broad Institute where we took our tool compound.

02:21:31.540 Now, we don't quite have a development candidate yet, but we took the tool compound, which is good enough, and run it through their panel of a thousand cancer cell lines to see what tumors are sensitive to it.

02:21:44.920 So this would be a treatment mode rather than a prevention mode.

02:21:48.680 And you think it could have efficacy in treatment as well?

02:21:52.320 That's the question we were asking.

02:21:54.220 Okay.

02:21:54.840 And melanomas emerged as by far the most sensitive tumor.

02:21:59.760 Now, I'm not sure why, because my hypothesis, the thing about skin cancer is it's got a heavy mutational burden because of all the UV exposure.

02:22:08.880 Yep.

02:22:09.880 It's the highest mutational burden organ we have in normal people.

02:22:14.140 And I thought that was going to be it.

02:22:16.440 And there was a correlation between mutational burden of the cell lines and susceptibility

02:22:22.300 to this mechanism, but it wasn't great enough to explain the tumor susceptibility.

02:22:27.180 So it's something else.

02:22:27.820 That would be good news.

02:22:29.640 Yeah.

02:22:30.220 It's there, but it's not good enough.

02:22:32.400 So I don't know why melanomas are so sensitive, but they're enormously sensitive.

02:22:37.180 So I'm very confident this will be a therapeutic for melanoma as well.

02:22:41.040 And it augurs well to the idea of preventing melanoma, which is also testable and has been

02:22:48.200 proven with a therapeutic intervention in a wonderful study conducted in Australia.

02:22:53.000 The intervention was intensive sunscreen use compared to usual practice.

02:22:58.360 So it's going to be testable in a large phase three study, but not before that.

02:23:04.680 So, yeah.

02:23:06.360 So that's what I'm doing.

02:23:08.180 And I think preventing cancers...

02:23:09.320 That's a very interesting idea.

02:23:10.620 I mean, talk about a whole new playing field, right?

02:23:15.920 Yep.

02:23:16.040 Pharmacologically.

02:23:16.880 Yep.

02:23:17.220 Nobody's made a drug for this mechanism.

02:23:19.360 Because we usually think avoiding cancer or preventing cancer comes down to avoiding carcinogens.

02:23:26.520 Which we should definitely do.

02:23:28.040 Yes, yes, absolutely.

02:23:28.920 So don't drink alcohol much.

02:23:30.840 Or don't smoke.

02:23:32.900 Don't smoke.

02:23:33.440 Be as insulin sensitive as possible.

02:23:35.660 Yes.

02:23:35.980 Yeah, all of these things.

02:23:37.040 All of those things.

02:23:37.580 Lose weight if you're overweight.

02:23:38.780 We know that successfully treating obesity prevents a bunch of cancers, and we know that

02:23:44.980 from the Swedish obesity study, which is an observational cohort of Swedish patients who've

02:23:52.320 undergone bariatric surgery, and these people are being followed for decades.

02:23:58.860 It's doing all the good things you'd expect of successfully managing obesity.

02:24:04.580 Well, Lloyd, this has been great.

02:24:06.340 this has been kind of a wonderful education on drug discovery using, I think, a very interesting

02:24:13.240 drug in BEMA as a case study for the complexity and the nuance of the process. And by the way,

02:24:20.460 I don't think I realized that the BEMA story is still ongoing. So that's great. So we're going

02:24:24.540 to continue to follow this biology and it'll be interesting to see where Eli Lilly goes with this

02:24:29.120 drug. But it sounds like based on what's showing up on clinicaltrials.gov, they're following in

02:24:34.560 your footsteps in that they're probably testing this in parallel with the newer generation GLP-1

02:24:40.040 agonists and the... As best I can tell, that's what they're doing. And it's not now just Eli Lilly

02:24:44.880 because many other companies, you know, who have seen the Believe data because we've been presenting

02:24:50.680 it at national meetings and international meetings for that matter. There's a lot of

02:24:57.480 other pathway inhibitors that are under development. Yeah. Well, really appreciate

02:25:03.520 your time, Lloyd. It's been a pleasure to chat again. Thank you. Thank you. Thank you for

02:25:09.160 listening to this week's episode of The Drive. Head over to peteratiamd.com forward slash show

02:25:15.960 notes if you want to dig deeper into this episode. You can also find me on YouTube, Instagram, and

02:25:21.840 Twitter, all with the handle peteratiamd. You can also leave us a review on Apple Podcasts or

02:25:28.080 whatever podcast player you use. This podcast is for general informational purposes only and does

02:25:34.120 not constitute the practice of medicine, nursing, or other professional healthcare services,

02:25:38.260 including the giving of medical advice. No doctor-patient relationship is formed.

02:25:43.720 The use of this information and the materials linked to this podcast is at the user's own risk.

02:25:49.440 The content on this podcast is not intended to be a substitute for professional medical advice,

02:25:53.680 diagnosis, or treatment. Users should not disregard or delay in obtaining medical advice from any

02:25:59.580 medical condition they have, and they should seek the assistance of their healthcare professionals

02:26:04.060 for any such conditions. Finally, I take all conflicts of interest very seriously. For all

02:26:09.860 of my disclosures and the companies I invest in or advise, please visit peteratiamd.com forward

02:26:16.420 slash about where I keep an up-to-date and active list of all disclosures.

02:26:23.680 Thank you.

The Peter Attia Drive - September 28, 2026

#409 ‒ Inside modern drug development： the science, economics, and regulatory hurdles behind bringing new medicines to patients ｜ Lloyd Klickstein, M.D., Ph.D.

Episode Stats

Transcript