The Peter Attia Drive - #412 ‒ The evolving science of cancer： emerging approaches to prevention and early detection, and the advances reshaping treatment ｜ Siddhartha Mukherjee, 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 peteratiyamd.com forward slash subscribe. My guest this week is Dr. Siddhartha Mukherjee,

00:01:07.640 a returning guest to The Drive. Sid is a physician, oncologist, researcher, and a Pulitzer Prize

00:01:13.820 winning author of The Emperor of All Maladies, as well as other books, The Gene and The Song of the

00:01:20.080 cell. I wanted to have Sid back on because, frankly, there are few people who can move as

00:01:25.840 easily between the molecular details of a disease and the much bigger questions of why we age and

00:01:32.380 what we might be able to do about it. We begin our conversation with cancer, as this is the

00:01:37.060 15-year anniversary, it's hard to believe, of his remarkable book, An Emperor of All Maladies.

00:01:43.060 We talk about where it comes from, how much it is driven by genetics versus chance,

00:01:47.420 the challenges of detecting it early, and how treatment has advanced, before broadening our

00:01:52.980 discussion to what Sid calls regenerative failure, the declining ability of our tissues to repair

00:01:59.360 and replenish themselves as we age. Throughout this episode, we discuss how genetics, environment,

00:02:04.220 and random mutations contribute to cancer risk. We talk about the challenges of cancer screening

00:02:09.260 and early detection across all cancers, but especially pancreatic cancer, why aging isn't

00:02:14.480 simply the accumulation of diseases, but also the loss of our ability to regenerate, how declining

00:02:20.060 stem cell function may contribute to osteoarthritis and other age-related conditions, the relationship

00:02:25.180 between inflammation, the immune system, and regenerative failure, what it would take to

00:02:29.640 meaningfully extend human healthspan and lifespan, and how AI may help us better understand and

00:02:35.800 potentially address the biology of aging and cancer. So without further delay, please enjoy

00:02:41.740 my conversation with Dr. Sid Mukherjee.

00:02:44.480 sid awesome to see you as always and it's hard to believe it's been 15 years since emperor of

00:02:55.820 all maladies i remember reading it as soon as it came out and that i think makes me feel a little

00:03:00.440 old that 15 years has gone by so quickly if we take a step back what would you say

00:03:08.780 is true today in the field of oncology that was not true? Or maybe phrased another way,

00:03:16.000 what do we know today about cancer that we did not know when you published this remarkable book?

00:03:23.820 Well, Peter, first of all, thank you for having me. It's been a pleasure throughout. We've

00:03:27.980 spoken in the middle and learned more about cancer in the middle. Let me divide it very

00:03:34.320 simply into three large sections and ask the question, what do we know more about early

00:03:40.200 diagnosis or prevention, early detection, and treatment?

00:03:46.340 So three broad categories.

00:03:49.240 So prevention, what do you know about prevention?

00:03:51.560 Well, good news and bad news.

00:03:54.520 The good news is that we know a lot more about how cancer can arise not just out of being exposed to the traditional carcinogens, the cancer-causing agents that you and I know very much about, you know, x-rays and benzene and, you know, tobacco, etc.

00:04:21.540 So we now know that there is almost certainly inflammatory states that cause cancer.

00:04:30.280 We know a lot about the nature of these inflammatory states.

00:04:34.400 We know that there is a kind of a gun and trigger effect, the gun being the inflammatory

00:04:43.700 state, the trigger being the mutations that cause cancer, or a mix of some states like

00:04:50.100 that.

00:04:51.080 The bad news is that we are still early in that story.

00:04:55.040 In other words, I can't look at a person, a patient, my own patient, and say to them,

00:05:01.280 oh my goodness, I know that you're in an inflammatory state.

00:05:04.440 You have that inflammatory state, which is going to be pro-carcinogenic.

00:05:09.280 I don't have a biomarker in the same sense that I have a biomarker for, let's say, cardiovascular

00:05:14.520 disease being hypertension or high cholesterol levels.

00:05:17.980 I don't have a high cholesterol for cancer, almost certainly we will find it.

00:05:24.380 We will find that high cholesterol for cancer.

00:05:27.260 And we will find people who are at risk and put them in clinical trials to follow them

00:05:34.760 and figure out how we can prevent cancer.

00:05:37.460 The second piece of good news in prevention, we're still in prevention, is that we now

00:05:42.340 know much more about genetics.

00:05:45.500 We know much more about people who are at higher risk.

00:05:51.300 It's not, for instance, in breast cancer, we now know it's not just BRCA1-carrying patients,

00:05:57.140 mutation-carrying patients, or BRCA2-mutation-carrying patients, but actually many other genes that

00:06:03.380 are involved in breast and prostate and other cancers.

00:06:07.480 In fact, in some cases, not a single gene.

00:06:10.960 It's a multiplicity of genes.

00:06:12.560 We know what they are.

00:06:13.360 so I could actually quantify, really quantify by genetic sequencing your relative risk of having

00:06:21.840 some cancers, breast, ovarian, and more and more and more some other cancers.

00:06:27.900 We know almost with great certainty that obesity is a risk factor for certain kinds of cancers,

00:06:36.320 endometrial being one of them. We know for sure that I'm not speaking in the realm of prevention

00:06:43.520 again. We know almost certainly that most cancers are in mortality across the population

00:06:53.380 in the United States are going down. So we are actually making progress. The idea that cancer

00:07:01.640 is exploding and killing everyone in the population is nonsense. But, but, this is a very

00:07:07.940 important but, we know that colorectal cancer is on the rise in young people. We know that the gains

00:07:16.060 that we had made over the last 20 years in breast cancer have started to plateau. So something is

00:07:23.140 going on. Another way of saying that is that, you know, breast cancer mortality was coming down year

00:07:29.720 after year, after year, after year. And now it's beginning to sort of not come down as rapidly as

00:07:34.720 we had hoped in the years before. We know that a couple of cancers are on the rise. I told you

00:07:41.420 about colorectal cancer in younger people. We know that esophageal cancer, esophagus,

00:07:47.960 is on the rise. We know that a couple of other cancers are on the rise in younger people.

00:07:56.120 We don't know why. In the case of endometrial cancer in women, it's most likely because of

00:08:02.360 increasing rates of obesity. But the rest of it is a mystery, and the colorectal cancer business

00:08:08.920 is a complete mystery. The esophageal cancer is probably because of alcohol and tobacco mixed

00:08:15.200 together, but that's half a mystery. Especially given that the trend appears that people are

00:08:21.880 actually drinking less today, although maybe that hasn't caught up to where the cancer is actually

00:08:27.660 arising. Yeah, it's always a lag. It's always a five to 10-year lag. So that's what we know about

00:08:33.440 prevention. Big deals. We can talk about each of those individual pieces that I spoke about

00:08:38.580 separately, and each one of them is a full conversation. All right, let's talk about

00:08:42.900 early detection. We know that there are tests in patients, for instance, blood tests,

00:08:51.880 that will enable the early detection of some cancers.

00:08:58.200 We know that if you just, at least for now,

00:09:01.860 we know that if you just take those tests

00:09:03.560 and unloosen them on a general population,

00:09:08.680 so without selecting for high-risk people,

00:09:11.360 without selecting for people who have had cancer before,

00:09:14.960 we know that those tests are probably not going to work.

00:09:18.560 They're going to lead to false positives.

00:09:20.440 They're going to get to extra biopsies and extra, you know, tests, more tests, following

00:09:25.700 tests, et cetera.

00:09:27.260 But those are not going to decrease, are unlikely to decrease the mortality from cancers.

00:09:35.600 That's the bad news.

00:09:36.740 The good news is that probably if you had used those tests or if you were to use those

00:09:41.480 tests in patients who are at high risk, for instance, those with carrying genetic mutations

00:09:47.720 that put them at high risk or those who have had previous cancers. Those tests may actually yield

00:09:53.140 very important results. And we know also that we're inventing a whole host of new ways of

00:10:00.460 detecting cancer early using imaging technologies that are new. Whether they will work or not,

00:10:06.160 I don't know. So one example of this is what's called more continuous rather than discrete

00:10:11.060 imaging. So by discrete imaging, I mean, you know, people used to go and have their mammograms

00:10:16.180 once a year. By people, I mean women above a certain age. The mammograms themselves were of

00:10:22.480 relatively poor quality. The once a year was a discrete test. We know that potentially having

00:10:29.600 mechanisms to follow over time, less in episodic ways, but more in continuous ways is likely to

00:10:40.080 bring a different set of answers. So that's in the realm of early detection. In the realm of

00:10:46.580 treatment, there's lots of things. We know since the last 15 years that immunotherapy, activating

00:10:52.720 the immune system, is extremely productive against some cancers. Just to give you one example,

00:11:01.180 I had sort of written a personal note to myself saying that this was 15 years ago, that I would

00:11:08.820 not live to a time where there would be a medicine that would potentially extend the

00:11:17.120 lifetime of a smoker with lung cancer.

00:11:21.880 That's not true anymore.

00:11:23.560 Immune therapy is now known to extend the lifespan of smokers with lung cancer if given

00:11:30.700 to the right population of people.

00:11:33.180 These are drugs that unveil or unleash the immune system to recognize cancers as foreign

00:11:39.440 cells and enable treatment of cancers.

00:11:44.560 We know that in some cases, activated T cells, genetically engineered T cells, so-called

00:11:49.960 CAR T cells, can cure liquid cancers such as leukemias, blood cancers such as leukemias

00:11:58.320 and lymphomas.

00:11:59.100 and by cure, I mean not make you live five years, but make you live 15 years or 20 years

00:12:04.320 after your diagnosis. There are people walking around on earth that have had CAR T-cell therapy

00:12:10.900 for unresponsive leukemias and lymphomas who are 15 years post-therapy. We know, and we were

00:12:21.620 talking about it a little bit before, that there are early signs that metabolic therapies against

00:12:27.380 cancer. So using the idea that cancer metabolically is different from normal cells

00:12:33.140 is a potential way to treat cancers. We know that there are second-line and third-line therapies

00:12:43.040 in certain cancers that will truly expand your lifespan. So two good examples are the use of

00:12:51.300 a drug called NHER2 in breast cancer, which works even after many, many lines of therapies

00:12:59.760 have stopped working. In myeloma, there are three or four new medicines. In lymphoma, there are two

00:13:06.280 or three new medicines. So in prostate cancer, there are two or three new medicines. And we know

00:13:14.300 that there are now treatments, surgical treatments, for many of these breast, prostate, and other

00:13:20.620 cancers, which are much more sparing of normal tissue, so that you can regain functionality

00:13:27.660 more easily and better, and yet remove as much cancer as possible. And finally, in treatment,

00:13:36.200 we know that there are a radical new set of medicines on the horizon. You've seen cases

00:13:42.380 of pancreatic cancer treated with the Daxon Rasib, a new medicine, against one of those

00:13:51.260 killer genes called Ras. It's extended the life of patients. We've seen videos of these patients

00:13:58.220 on television. They're now living with pancreatic cancer months, if not years, after the cancer

00:14:07.500 or his spread, which would have been impossible 10, 15 years ago. So immunotherapy, novel biologics

00:14:15.460 and novel small molecule therapies are here. They're on the horizon. They're not just making

00:14:25.640 people live longer by one month or two months. They're sometimes actually making people live

00:14:32.540 longer into the five year range and extending survival meaningfully. So that's kind of what

00:14:43.440 the landscape looks like right now. I mean, Sid, that's fantastic because everything you've covered

00:14:47.760 there is sort of where we want to go with one exception. We didn't touch on AI, which we will

00:14:51.920 touch on, I'm sure. Yes, of course. But let's go back and look at something that we learned in the

00:14:58.280 last 15 years, courtesy of Brett Vogelstein and his colleagues. So in 2015, they published a paper

00:15:04.660 in Science. Both you and I remember this very well. The entire oncology community was sort of

00:15:10.240 divided a little bit on this. In 2017, they came out with another paper that kind of revisited it.

00:15:15.280 But the gist of the paper was that if you looked at all cancers by tissue type, if I'm remembering

00:15:22.300 the math correctly, Sid, about 5% of the mutations, because again, cancer is a genetic disease. It is

00:15:29.500 driven by genetic mutations. About 5% of these mutations were hereditary. And you already alluded

00:15:34.100 to some of these, you know, HER2NU and, pardon me, BRCA, BRCA1, et cetera, et cetera. But there

00:15:39.320 are many others that we don't necessarily talk about as frequently. On the order of 30% of the

00:15:44.840 mutations appeared to arise from environmental triggers or carcinogens. So when you smoke,

00:15:49.980 you're increasing your risk of a mutation. When you're exposed to lots of UV light, you're going

00:15:54.820 to have a mutation, et cetera. But the part that I think really was sobering for us was that two

00:16:02.840 thirds, 66-ish percent of mutations were purely random. And this is where the idea that stuck from

00:16:12.360 Vogelsen's paper was this idea of cancer is a bad luck disease, right? You can do everything right,

00:16:18.060 Sid, and you more than anybody else knows this, you're a practicing oncologist. You can have the

00:16:23.120 healthiest person who's done everything correct. And yet this random mutation occurs. And that

00:16:29.340 may explain up to two thirds of the mutations we see. And we don't see that in cardiovascular

00:16:34.540 disease, right? We don't see the perfect pillar of health, who's done everything right, who's

00:16:40.400 managed cholesterol and blood pressure, et cetera, et cetera. So how do we make sense of this

00:16:47.960 beyond the observation? In other words, how do we take that observation and do something about it?

00:16:54.660 So first of all, let's really take the observation seriously. Let's take it totally

00:16:59.800 seriously. 5% of mutations, hereditary, that number is most likely no longer 5%.

00:17:09.920 During Vogelstein... Yeah, because that last paper, sorry to interrupt,

00:17:14.860 was actually nine years ago.

00:17:16.520 And I'm not, I don't know if there's been a follow-up.

00:17:18.400 So yes, yeah.

00:17:18.740 So, you know, Bert Vogelstein is a close friend,

00:17:21.220 one of the great scientists of the world,

00:17:22.720 one of the greatest cancer scientists of the world.

00:17:24.680 But obviously the field changes, the field moves.

00:17:27.180 Yep. 1.00

00:17:28.460 That 5% is going to change. 0.92

00:17:31.160 So, and how is it going to change? 0.85

00:17:32.560 Well, that 5% was a,

00:17:36.640 was usually focused on single gene mutations.

00:17:40.620 So BRCA1, BRCA2, you know, P53,

00:17:44.860 I can rattle them off. Your listeners won't know them. They'll know BRCA1, they'll know BRCA2.

00:17:52.600 So just to give you, let's just take breast cancer. So hereditary breast cancer, where you 0.99

00:17:58.060 know that there's a hereditary breast cancer. So you know that your mother had it, your mother's

00:18:04.940 sister had it, your aunt had it, your grandmother had breast cancer, and you sequence your genome,

00:18:12.960 And there's no BRCA1 mutation.

00:18:14.700 There's no BRCA2 mutation.

00:18:16.780 So now we've added many more mutations,

00:18:22.780 single mutations in single genes that increase risk.

00:18:27.820 But we still, we know that that only covers,

00:18:31.920 we're now talking only about hereditary breast cancer.

00:18:36.540 If you take all the single genes that we know

00:18:39.620 increase the risk of hereditary breast cancer,

00:18:41.600 that still only covers a fraction about a third maybe a fourth of hereditary breast cancer

00:18:48.960 so the question is what's the other so again remember we're talking about just hereditary

00:18:54.520 breast cancer so what's the what's the other whatever percentage and the answer now is that

00:19:02.980 that other fraction in hereditary breast cancer not breast cancer caused by random mutations

00:19:10.340 or by exposure to environments, et cetera, is a combination of multiple genes, not one,

00:19:17.060 not two, but probably hundreds of genes that are pushing your risk by a small percentage.

00:19:25.040 So the way I try to describe this to my patients is I say, in hereditary cancers,

00:19:30.560 there are two kinds of mutations. I call them shove mutations and nudge mutations.

00:19:37.500 BRCA1 is a shove mutation. It shoves you off the cliff and puts you at a ninefold, tenfold risk

00:19:46.180 above the normal population to get hereditary breast cancer at an early age. BRCA2 is a shove

00:19:53.400 mutation. It shoves you off the cliff. But there are also patients who have nudge mutations.

00:20:00.980 They're not shoved off the cliff. They're just sort of given little nudges. Each of these

00:20:07.140 changes gives you a little nudge and of course there's environmental factors that come into it

00:20:12.720 but they give you a little nudge along with the environmental factors until you fall off the

00:20:16.600 cliff and get cancers that's hereditary so that number five percent is no longer the number five

00:20:22.100 percent second big major point cancer as a disease of bad luck yes it is still a disease of bad luck

00:20:31.160 but it's not a disease of bad luck in the same sense that we imagined it 10, 15 years ago when

00:20:37.760 Bert wrote his paper. Yes, there's bad luck involved, but it's also bad luck plus something

00:20:44.220 else bad that's going on in the body. Such as inflammation.

00:20:48.860 Such as inflammation. And the best example of that is recent papers that have come out from

00:20:54.620 many laboratories. I specifically want to point out Charlie Swanton's lab in the UK.

00:21:01.160 But many laboratories that show that you can have cancer, that you and I right now, if we were to biopsy every cell in our lung, if we magically had the capacity to biopsy every cell in our lung, we would find almost certainly, and I don't smoke, you don't smoke.

00:21:26.400 we would almost certainly find cells that genetically speaking would bear all the

00:21:34.240 hallmarks of genetic mutations of cancer. But you and I don't have cancer.

00:21:41.400 So what's with that story? The answer to that story takes us back to the 1950s. And this is

00:21:48.960 a very important idea that I often emphasize. It's something that's gone out of the scientific

00:21:54.400 literature, it's sometimes very important, Peter, to publish an observation without explaining it.

00:22:03.640 You know, I'm going to take a philosophical aside here. I'll come back to the real point

00:22:07.740 in a second. I'm going to take a philosophical aside here. Our scientific publication strategy

00:22:16.740 has become a form of magic storytelling.

00:22:22.760 By magic storytelling, I mean you have to tell a full story

00:22:27.320 about what you observe, what the mechanism is,

00:22:32.220 what the explanation is, and how it's working at a molecular level.

00:22:39.080 That was not the case in the 1950s.

00:22:41.140 In the 1950s, 1940s, 1960s, you could say,

00:22:44.820 listen, I don't know the answer.

00:22:45.900 I don't know why this is happening.

00:22:47.800 I can just tell you it's happening.

00:22:50.160 You go figure it out five years later

00:22:52.120 when you have the technology to figure it out.

00:22:54.140 I can't emphasize how important that is.

00:22:58.180 You know, we have writers, I'm a writer myself,

00:23:00.780 you know this, we have writers like Oliver Sacks,

00:23:04.620 whose legacy has dependent on the idea

00:23:08.800 that they made very thorough, important observations.

00:23:14.200 William Osler,

00:23:15.900 Perhaps the greatest of physician scientists said, observe the patient first, explain later.

00:23:25.640 So this is a philosophical observation that I want to make.

00:23:28.680 So let's go back.

00:23:30.500 So in the 1950s, a team of scientists from Oxford published a paper which remained unexplained.

00:23:43.660 And it was a very simple paper.

00:23:45.900 They took a carcinogen, let's call it whatever you want to call it, let's call it DMBA,

00:23:53.740 chemical carcinogen, makes mutations, should cause cancer.

00:23:59.520 And they painted it on the back of a mouse.

00:24:03.920 And sure enough, a few cancers grew.

00:24:08.180 And then they went back and get this, this is a super interesting story.

00:24:13.040 Then they went back and said, okay, fine, just a few cancers grew.

00:24:15.900 But we painted it on the entire back of the mouse.

00:24:19.240 Why was it like four cancers grew?

00:24:24.020 So then, after painting the back with this chemical carcinogen,

00:24:29.960 they added an inflammatory agent, croton oil,

00:24:35.080 something that you can buy from Amazon.

00:24:38.180 And all of a sudden, 80 cancers grew.

00:24:41.540 years and years later in experiments by alan balmain in ucsf they genetically engineered a mouse

00:24:55.080 to get cancer so this mouse was genetically predisposed just like a human being with braco

00:25:03.160 one may be predisposed to get a cancer they made a so and sure enough when the mouse grew up

00:25:11.220 It had four or five or six cancers where, you know, obviously where the gene was making cancer cells.

00:25:21.740 And here's the interesting thing that happened.

00:25:23.860 They made a little slit or a cut, an inflammatory wound in the mouse.

00:25:30.320 And again, nothing very impressive happened.

00:25:34.540 The next thing they did was very interesting.

00:25:36.880 This is by chance.

00:25:38.000 And this is why science is very important.

00:25:39.540 It's very important to do experiments.

00:25:41.540 It's more to report experiments.

00:25:43.500 And it's more to report just observations.

00:25:46.540 They stapled where the wounds were.

00:25:48.800 And wherever there were staples, of course, you got an inflammatory response.

00:25:52.880 And exactly, exactly the place where they put the staples outgrew massive cancers.

00:26:02.040 Now, what does that tell us?

00:26:03.580 It tells us something very important.

00:26:05.040 It tells us that this whole idea that there's bad luck.

00:26:08.300 Yes, there's bad luck.

00:26:09.540 Let's say you, when I biopsy your lungs, I get, you know, you have several million,

00:26:16.820 maybe a billion cells in your lung.

00:26:19.220 I'll find, let's say, 2,000 cells that have already obtained or carrying cancer mutations.

00:26:31.160 Let's say I biopsy my cells and I find that that number is not 2,000, but 750.

00:26:38.920 You have some bad luck.

00:26:40.520 I have some bad luck.

00:26:41.540 Your bad luck is 2,000.

00:26:42.680 My bad luck is 750.

00:26:45.140 Those cells with mutations most likely, and I'm just saying most likely because we don't

00:26:50.700 know, most likely arose because you and I had bad luck.

00:26:54.120 And when the cells grew, they made mistakes.

00:26:57.620 They caught up.

00:26:58.940 They made genes that are cancerous and so forth.

00:27:02.460 But those cells are dormant.

00:27:04.180 I call them sleeper cells.

00:27:05.360 the real bad luck that happened between you and me is that you had some kind of inflammatory

00:27:15.000 response. And it's not, when we say inflammation, it's like using the word motion before Newton,

00:27:22.300 or using the word energy before Newton, or using the word atom before Dalton. When you say the

00:27:29.580 word inflammation, you know, it's a loose word. It's not like, oh, you know, I had a fever, I had

00:27:33.780 a cold, I had a virus. There's something very particular about the inflammation. We can talk

00:27:39.720 a little bit more about it. You had a particular kind of inflammatory response in your body

00:27:45.940 and one of your 2,000 cells, which already was carrying a mutation, turned from a sleeper cell,

00:27:54.980 a dormant cell, a doing nothing cell into a tumor, into an active cancer cell.

00:28:01.660 And from there on, you develop cancer.

00:28:05.840 I was lucky.

00:28:08.220 I didn't have that inflammatory stimulation.

00:28:12.060 My 750 cells remained dormant, remained quiescent, remained asleep, and I don't have cancer.

00:28:21.960 So now let's go back to Bert Vogelstein.

00:28:25.080 The number 5% is not 5% anymore.

00:28:28.480 It's a much larger number than 5%.

00:28:30.480 and we can detect that larger number now with greater and greater accuracy. Not in all cases,

00:28:37.020 not in all cancers. Breast cancer, yes. Ovarian cancer, yes. Maybe in some other cancers more and

00:28:43.240 more. And that number unlucky, a cancer of bad luck, is no longer a cancer of bad luck. Very,

00:28:52.860 very soon, I hope. I can't give you a date. Very soon, I will be able to tell

00:28:57.400 that I will be able to tell you

00:29:00.660 that you have bad luck

00:29:03.280 and I don't have bad luck.

00:29:04.400 And it's not bad luck,

00:29:05.240 it's just the nature

00:29:06.620 of our inflammatory response.

00:29:08.820 So this whole idea of dividing up,

00:29:11.800 creating a typology of cancer,

00:29:14.540 every typology of cancer evolves.

00:29:17.240 It's evolved since the beginning

00:29:18.980 of any typology of cancer.

00:29:21.320 And so those numbers

00:29:22.420 are all going to change

00:29:23.460 and going to change in a way

00:29:25.180 that is sensible.

00:29:27.400 So Burt is not wrong.

00:29:31.100 He is just early.

00:29:34.740 And maybe there's another point to just double click on this, Sid, which is the way that

00:29:41.060 that paper described it suggested that these are discrete types of mutations.

00:29:45.300 But as you pointed out, and I love, I'm going to borrow this, by the way, so thank you.

00:29:49.240 This idea of shove versus nudge mutations, they can both arise as hereditary mutations.

00:29:55.440 But here's where the environment overlaps. So instead of saying that hereditary mutations are

00:30:02.680 a discrete bucket, environmentally acquired or triggered mutations are a discrete bucket,

00:30:07.140 it could be no. If you have a bunch of shove mutations, or pardon me, a bunch of nudge

00:30:13.000 mutations that are acquired, or pardon me, that are passed through the germline,

00:30:18.300 the environmental mutation is really a push or a gentle nudge.

00:30:23.480 Yeah, exactly. So in other words, these are not discrete. They're going to start to overlap

00:30:27.720 a little bit more. Absolutely. And this is super important because

00:30:31.760 this is not just theorizing, right? These have very practical implications for patients.

00:30:39.120 Because if I could figure out where you are in the nudge versus shove cascade,

00:30:45.540 I could tell you, I could quantify your risk. I could tell you what your risk is, and you should

00:30:50.440 be maybe in a clinical study. You should maybe be in a prevention program or a trial prevention

00:30:57.580 program. So these are not just sort of abstractions. These are realities. The last thing we didn't

00:31:04.740 touch upon is metabolism. We talked about inflammation. We didn't talk about metabolism.

00:31:08.780 So metabolism is becoming very important. We know, I mean, just to give you the most obvious

00:31:14.620 example, you know, we know that there's an enormous risk between, and we don't know why

00:31:21.880 exactly, but an enormous risk that connects, for instance, endometrial cancer with obesity.

00:31:28.780 And endometrial cancer is one of those cancers which is either plateauing or maybe on the rise

00:31:33.540 in women, maybe because of obesity, but what's the connection? Is it through inflammation again?

00:31:42.160 Is it because there is a hormone that's making, you know, the sleeper cells alive?

00:31:49.780 Obesity itself is unlikely, I would say, to cause mutations.

00:31:56.720 Obesity is not a mutagen, but obesity is a physiological state.

00:32:02.620 And those physiological states, of course, are important because we've already seen,

00:32:06.440 in the case of inflammation, a physiological state that can take a dormant cell and convert

00:32:11.980 it into a cancerous or cancerous tumor cell. And Sid, how much of it do you think is obesity

00:32:19.300 per se with its adiposity versus insulin resistance, hyperinsulinemia, and the sort

00:32:26.920 of metabolic milieu that accompanies obesity often, but not always?

00:32:31.080 So in humans, we don't know. But about five years ago, Luke Cantley and I wrote a paper in Nature, which actually you were you inspired in some ways, where we know that in at least in animal models, you don't need to be obese in order to unleash or at least in order to have a cascade that sustains a cancer.

00:32:58.900 I have to be careful with my words because that paper was really a paper about treatment.

00:33:05.740 But what we can surmise from that paper, and I'm being careful with my words surmise,

00:33:14.440 is that it's not being obese.

00:33:19.720 It is the insulin and it's the cascade of insulin and what follows the insulin cascade

00:33:28.000 that is responsible at least

00:33:31.280 for the resistance of cancer cells to therapy.

00:33:36.140 Now, from that, you could extrapolate,

00:33:38.200 and I'm using again, being careful with my words,

00:33:40.420 you can extrapolate the idea

00:33:43.420 that insulin and insulin resistance

00:33:49.420 and the presence of chronic insulin response

00:33:55.580 may have something to do with releasing cancer cells in the body, but if that's an extrapolation,

00:34:05.600 it would be a complicated experiment, but a doable experiment in which you chronically exposed

00:34:13.340 animals to hyperinsulinemia in the way that obese people are exposed to hyperinsulinemia without

00:34:21.000 out the appropriate insulin response, which normal non-obese people have, and figure out

00:34:29.220 whether it was truly that that was causing the cancerous cells to develop. The general hypothesis

00:34:36.600 I have, and again, I have to say this is a hypothesis, is that that's the case.

00:34:41.500 Yeah. Mitch Lazar at Penn wrote a paper. It's been about 15 years as well. And he looked at

00:34:51.560 the incidence of cardiometabolic disease and he sort of created a two by two, right? So you were

00:34:57.900 either obese and or metabolically unhealthy. But there were states, as you know, where you can have

00:35:06.580 lean individuals who are still metabolically unhealthy. You can actually have obese individuals

00:35:11.700 who are metabolically healthy. And when you do the two by two of that, what Mitch's paper found,

00:35:18.100 again, this was focused on cardiovascular disease, not oncology, but it showed that the prevalence

00:35:23.940 of disease didn't track with the adiposity. It tracked with the hyperinsulinemia and everything

00:35:30.880 you just described that accompanies it. So it's, I mean, that would at least suggest plausibility

00:35:36.060 that the same thing could be true in oncology, although, of course, the diseases are different.

00:35:40.120 Yeah, the diseases are different. But again, as I said, our work has been, and it's now a full-fledged

00:35:47.180 clinical trial. We can talk about the trial. Our work has focused on treatment, and we know that

00:35:55.300 chronic insulin tone, I use the word tone like the tone of muscles, but chronic insulin

00:36:06.040 tone has the capacity to make cancers grow and become or be or remain resistant to some

00:36:16.280 forms of therapy, and that you need to remove that chronic insulinemic tone for those cancers

00:36:22.800 to become vulnerable.

00:36:25.560 These are facts.

00:36:27.260 From those facts, you may extrapolate that chronic insulinemic tone is also responsible

00:36:34.820 for the sustenance and maintenance

00:36:38.220 of cancerous cells that have mutations.

00:36:42.260 And in the same way that inflammation

00:36:43.720 is making those cells sustain, maintain, and grow,

00:36:48.680 that chronic insulinectone

00:36:50.300 maintains those cells in a chronic state.

00:36:54.320 Now, I would be so happy

00:36:56.400 if someone did that experiment.

00:36:59.040 It's not that complicated an experiment to do.

00:37:03.780 It just takes a lot of very careful controls and careful—it's a fussy experiment to do.

00:37:12.680 And good experimental people can do fussy experimental science.

00:37:17.260 Sid, the GLP-1 agonists, while potentially in the very near term, will transiently raise

00:37:26.040 insulin levels to aid with glucose disposal, ultimately tend to reduce hyperinsulinemia.

00:37:31.260 It's not entirely clear why.

00:37:32.660 Of course, it could be that with enough weight loss and, you know, a reduction in energy

00:37:39.920 balance that hyperinsulemia goes down.

00:37:42.440 But has anyone proposed doing a study of using a GLP-1 agonist as an adjuvant to conventional

00:37:50.860 cancer therapy where you would just have a simple randomization of standard of care versus

00:37:55.900 standard of care plus GLP-1 agonist?

00:37:58.800 No, many people have thought about it.

00:38:00.400 I don't know if there's been a formal proposal to either NOVO or Lilly or the NIH.

00:38:08.780 This is a fundable and thoughtful study.

00:38:12.060 You'd have to do it.

00:38:13.060 Again, it's a fussy study.

00:38:14.060 It's to pick the right cancer.

00:38:15.060 You'd have to pick the right cancer.

00:38:17.060 I would pick endometrial or ovarian or some obesity.

00:38:20.600 Even breast, potentially.

00:38:21.900 A breast? 0.99

00:38:22.900 Yeah. 0.95

00:38:23.900 And really think through it.

00:38:26.160 You want to do this thoughtfully.

00:38:27.160 don't want to just run into this and just do it in a silly way. You want to do this thoughtfully.

00:38:32.300 I think it's a beautiful study. It would be a beautiful study. You would randomize to patients

00:38:37.760 with on GLP-1 versus with not GLP-1. You would study their insulinemic response versus their

00:38:47.400 weight loss response. And with AI, which we'll talk about in a second, you could probably

00:38:52.260 dissociate those two and disconnect them and ask the question, which one is leading and which one's

00:38:57.460 following. But it's a beautiful study. And look, we could do a very broad, sweeping epidemiological

00:39:05.140 study, which you could do in five years or 10 years. It's a little bit too early. And just

00:39:10.820 ask the question, listen, if people who are in GLP-1 versus not GLP-1, is there a lower risk

00:39:18.340 of endometrial cancer. Now, see, I prefer the experimental approach here, Sid. Yeah,

00:39:24.180 not just because of time. I think there might be too many confounding factors if we just do it

00:39:28.540 observationally. Of course. I mean, of course. But the observational study is much easier to do.

00:39:34.460 It's, you know, you don't have to... Yeah, you don't have to intervene.

00:39:37.600 You don't have to intervene, no. So yeah, of course, we all prefer the real proper study.

00:39:42.700 And I think someone should fund this study.

00:39:45.560 I mean, you know, if it's not the NIH, if it's not the Eli Lilly or Novo or whoever else is now making, you know, any of the healthy ones, you know, any of the people, anyone in the audience who's, you know, interested in longevity.

00:40:01.520 Because I'm interested ultimately in, you know, what these drugs do, not just to cancer, but to the overall physio-inflammatory state, which is certainly pushed by insulin.

00:40:19.900 And I would love to ask the question, well, what happens to longevity markers?

00:40:23.940 What happens to, you know, so this is a study that one of the many, very many people who have the means and capacity to do so can fund through, you know, a remarkable institute like UCSF or Columbia or Stanford or Harvard.

00:40:45.220 I'm just going to name some names, but there are many.

00:40:47.440 Yeah, yeah.

00:40:48.260 And by the way, there is a bit of a precedent on some of this.

00:40:50.720 I don't know if you've been following the literature on osteoarthritis.

00:40:53.940 Oh, this is one of my favorite topics.

00:40:57.800 Yeah.

00:40:58.220 Again, maybe just explain the observation to people and how it points or shines a light

00:41:04.920 on inflammation independent of weight loss because it's quite fascinating.

00:41:09.660 Well, so osteoarthritis is my neck of the woods because my laboratory discovered the

00:41:17.020 first osteoarthritic stem cells.

00:41:19.520 We published that in Cell and then subsequently in Nature and Nature Medicine.

00:41:26.960 I have never believed from the beginning of the time I began to study osteoarthritis,

00:41:33.500 I never believed that osteoarthritis was an obesity disease.

00:41:38.060 There are non-obese people who have osteoarthritis.

00:41:41.880 There are obese people who have osteoarthritis.

00:41:43.340 It's much larger.

00:41:44.460 but the idea there was something kind of like it didn't make sense and also it was sort of like

00:41:50.080 let's blame the victim let's say fat people get osteoarthritis and let's just say oh you know

00:41:56.560 that's a fat people problem and it just didn't make sense it doesn't make sense for many reasons

00:42:02.480 look i'll tell you some of the reasons so yes of course the weight-bearing limbs the knees the hips

00:42:10.180 get osteoarthritis and people complain about knees and hips and they get hips replaced,

00:42:16.000 knees replaced, etc. Great. Fine. Makes sense. Usually people who are obese, usually women who

00:42:21.760 are obese. Point number one, it seems to be like many things that we've done in medicine for women, 0.96

00:42:31.040 a backwards problem, which is to say, oh, you know, they are obese and therefore they must have 0.99

00:42:37.300 weight-bearing problems, and therefore it's a weight problem. But then you meet a person who's 0.71

00:42:44.460 just completely has no obesity, a woman, and their osteoarthritis is in their hands.

00:42:51.840 They have these massive nodes, inflammatory nodes, inflamed nodes in their hands.

00:42:58.120 The hand is not a weight-bearing joint. Some of them have degeneration of their

00:43:05.040 upper neck joints. Their complaint is upper neck pain, shoulder pain, and they're not obese.

00:43:14.680 And you say, well, this observation can't be true. It just doesn't make any sense.

00:43:18.700 So now we know that osteoarthritis is almost certainly a combination of two problems.

00:43:29.340 One that we've, that my lab has been a big proponent of,

00:43:33.620 which is that the stem cells that grow cartilage in the joints over time,

00:43:40.220 those stem cells, there's a balance.

00:43:44.620 So during our lifetime, cartilage is being generated and is being destroyed,

00:43:50.820 generated, being destroyed, generated, being destroyed.

00:43:53.000 if that genesis and destruction is an adequate balance then we maintain cartilage and we don't

00:44:02.640 get osteoarthritis if the genesis gets stopped with the destruction remains the same then of

00:44:10.900 course you know you won't get new cartilage and over time you'll eat up the cartilage that you

00:44:15.300 have and you'll get osteoarthritis so the question then becomes well what causes this imbalance so

00:44:21.580 So if you imagine them as weights on a scale, remember, normal is equal weight, genesis and destruction, genesis, destruction, genesis, destruction.

00:44:36.060 In women who have obesity, and in very older men and women, genesis goes down, destruction remains the same or goes up, and all of a sudden you have an imbalance. 0.75

00:44:49.080 So the question becomes, well, what's driven genesis down?

00:44:53.320 Well, we now know, at least in animals, what's driven genesis down.

00:44:57.000 Then the stem cells that form cartilage, the cells that create real cartilage, have gone down.

00:45:03.680 We've measured it.

00:45:04.720 We know this is a fact.

00:45:08.060 Now the question is, why have they gone down?

00:45:11.280 Well, some of it is age.

00:45:13.340 Age because, you know, in overage, all of our stem cells go down.

00:45:17.480 skin stem cells go down, your bone marrow stem cells that make blood go down. So some of it is age.

00:45:24.120 If you inject in a young animal, young mouse, if you inject an inflammatory response into a joint

00:45:34.880 and it's a young animal, has no reason to have any other problem, we know that that inflammatory

00:45:43.420 response will cause the stem cells to stop making cartilage. And sorry, when you say

00:45:49.100 inflammatory response, do you mean like LPS or bacteria? Yeah, it could be like a bacterial

00:45:55.740 mimic. Yep. Yes. So something to activate the immune system. Yeah, something to activate the

00:46:01.400 immune system, exactly. And we now know that if you activate the immune system either using a

00:46:07.840 small molecule or using some kind of stimulation, the inflammatory response, you'll start an

00:46:15.640 inflammatory response and it will create, it will mimic, almost exactly mimic osteoarthritis in the

00:46:23.780 animal. The balance, remember again, the weight scales, genesis, destruction. The destruction

00:46:29.800 goes up, the genesis goes down and you have osteoarthritis. And so let's think through

00:46:37.060 why that might be happening, because that's also not entirely intuitive that an immune response,

00:46:43.860 which serves the purpose of trying to eliminate a foreign pathogen, would also in its wake have

00:46:51.680 its host's stem cells reduce their output. Well, now you're asking a really hard question,

00:46:59.240 And the answer is, I don't know. I just, I don't know. I think that, I think that it's actually

00:47:07.440 not uncommon in the stem cell world across different kinds of stem cells, skin, bone marrow,

00:47:16.460 blood stem cells, and in this case, cartilage forming stem cells, that certain immunological

00:47:23.180 responses, tend to make the stem cells reduce their capacity to regenerate.

00:47:33.060 And it's a common theme.

00:47:35.580 This is not an uncommon.

00:47:36.920 I mean, there must be some at least teleologic explanation that I'm just missing.

00:47:40.660 But the problem, as you know, Peter, teleology is not biology.

00:47:50.100 No, it's rationale.

00:47:51.220 It's a way we tell our story to ourselves that we like.

00:47:53.520 It's just so story.

00:47:54.260 It's like Rudyard Kipling's just so story.

00:47:56.400 Yeah.

00:47:57.220 Teleology is like saying, oh, I think it's because in ancient times.

00:48:02.680 Right, right, exactly.

00:48:03.940 Yep, yep.

00:48:04.360 When people were hunting and gathering and fighting lions and running from woolly mammoths or whatever, you know, crazy nonsense.

00:48:14.040 Yeah, yeah.

00:48:15.160 You know, stem cells had to shut down because they had.

00:48:19.080 That's right.

00:48:19.620 Right. They were too energy demanding. And when the immune system... Yeah, yeah, exactly.

00:48:23.900 I mean, you know, some of this can be observed. Some of this can be experimentally tested.

00:48:29.920 But you can't put a human being in, you know, prehistoric times, you know, 100,000 years ago,

00:48:36.520 or a pre-human being or a monkey or a, God knows, ancestral, common ancestor in some

00:48:43.260 stress condition and ask the question, well, why do their stem cells not stop working?

00:48:50.080 Yeah, the more relevant question is how, right? If we knew the how,

00:48:55.340 then we could intervene, because that's what we have to figure out.

00:48:58.900 Well, again, so let's talk philosophically. Biology, and maybe I should say this as a,

00:49:05.420 I'm now taking off my hat

00:49:08.100 as an immediate scientist

00:49:10.820 and talking to you

00:49:12.320 as a science philosopher

00:49:13.680 why questions

00:49:17.740 are very hard to answer in science

00:49:20.340 how questions

00:49:22.040 are what science

00:49:24.900 science can do

00:49:26.700 I can tell you how something happens

00:49:29.120 I can tell you

00:49:30.380 I can tell you how

00:49:32.640 an apple falls from a tree

00:49:35.260 and with what velocity and how.

00:49:38.140 Now, if you were to ask me why does the apple fall from the tree

00:49:42.640 and go towards the ground, I would say I have no idea.

00:49:46.180 Who knows?

00:49:48.540 Why are planets attracted to each other?

00:49:51.260 I don't know.

00:49:52.200 I mean, you know, this is like lives in the depths of very abstruse physics,

00:49:59.500 and biology is so many degrees removed.

00:50:02.220 If you ask me, you know, why does a stress response in humans shut down blood-forming

00:50:10.500 stem cells?

00:50:11.140 I'd say I have no clue, but I can tell you how.

00:50:14.360 I can tell you exactly the molecules.

00:50:16.500 I can tell you the molecules that are shut down.

00:50:18.700 I can tell you who's moving.

00:50:20.440 But why?

00:50:21.720 Why is not a science question?

00:50:23.320 Why is a metaphysical question?

00:50:26.800 So let's talk about the how, especially as it pertains potentially to the most important question, which is, well, before I go there, I want to ask you a question that might set the stage for this.

00:50:43.560 So we have a remarkable immune system.

00:50:47.300 You've already alluded to it a number of times through the lens of immune-based therapies.

00:50:52.280 And while we didn't go into great detail, we have a remarkable immune system.

00:50:56.800 Yeah. So we have this incredible system and we've danced around the success of this class of drug

00:51:03.120 called checkpoint inhibitors, which come along and remove the brakes on the immune system and

00:51:09.640 unleash the immune system on the cancer, suggesting that our immune cells at least

00:51:16.880 have the potential to recognize cancer as non-self. And in fact, when I had Steve Rosenberg

00:51:23.180 on the podcast a couple of years ago, he shared some unbelievable data, blew my mind, which was

00:51:29.760 that 80, eight zero percent of solid organ tumors had novel neoantigens that could be recognized

00:51:39.100 by self. Now think about that for a moment if you're the average person trying to make sense

00:51:44.180 of this. What that means is 80 percent of the cancers that are most lethal, the ones we think

00:51:49.340 of, breast, colon, lung, pancreas, these are the really bad cancers. 80% of these are immune cells.

00:51:58.440 Our T cells know that that is bad and it's not self. Then the question is, why can't they mount

00:52:06.100 on their own enough of a response to do the job? Sometimes they can. That's very rare. We now have

00:52:12.200 a drug that can take a subset of those. And this is the checkpoint. But we're still sort of

00:52:19.360 wandering in the desert with cancers that are non-self and an immune system that just can't

00:52:26.540 make it happen. So I want to spend some time talking about T-cell biology. And what do we

00:52:32.160 think it's going to take? Is it simply the age of the T-cells? And is it the case that if you're 10

00:52:39.000 years old, your T cells are doing this automatically. That's why 10-year-olds rarely

00:52:43.180 get cancer. But by the time you're 70, they've been around the block too many times.

00:52:49.300 So the answer is likely all of those things. But let's talk about neoantigens for a second, 1.00

00:52:54.100 because we need to really think about this. So Steve Rosenberg, who I know very well,

00:53:00.380 was a pioneer in the field. And years ago, Steve Rosenberg and I had a very public argument on

00:53:08.420 stage at a very prominent cancer conference. And he said, well, there are neoantigens in cancer.

00:53:15.920 And I said, but where are they coming from? And he said, well, some of them are coming from

00:53:19.580 mutations. And I said, well, come on, there are like 30 mutations, 50, 100.

00:53:24.920 A hundred. Yep. Call it a hundred. And it doesn't explain it all.

00:53:28.220 It doesn't explain anything. Now there's beautiful work from Tyler Jack's lab. And

00:53:34.080 I would say it's something from my lab that says that that's not the real new antigen.

00:53:39.120 The real new antigens are coming because when cancer cells start deregulating,

00:53:44.800 dysregulating, the way genes are, their genetic apparatus is decoded,

00:53:51.200 they make mistakes, they go wacko.

00:53:55.840 And all sorts of cryptic antigens, cryptic pieces of genes

00:54:01.800 that should not have been expressed, made into proteins,

00:54:06.400 are now coming out and making it into proteins.

00:54:12.480 Tyler's lab at MIT has done an incredible job at decoding what this is.

00:54:21.800 It's an incredible story.

00:54:23.120 I would encourage people to read about it.

00:54:25.560 But we now know what those are.

00:54:27.320 We can map them.

00:54:29.960 That's the first part of the problem.

00:54:31.260 So as you know, there are two parts of the problem.

00:54:32.980 One part of the problem is, what are the new antigens that are coming?

00:54:36.300 And number two is, why aren't T cells recognizing them?

00:54:40.220 The second part of the problem is more interesting, or equally interesting.

00:54:44.340 The T cells don't recognize them because of all the reasons that we've talked about.

00:54:48.160 There's not one checkpoint.

00:54:50.160 There are dozens of checkpoints in cells.

00:54:54.400 And what's a checkpoint?

00:54:55.840 A checkpoint is a gene or a flag that a normal cell floats on its surface, saying, I'm normal, leave me alone, to a T cell.

00:55:09.780 And that's why we don't have desperate, terrifying autoimmunity.

00:55:15.600 when we know that patients, people who have mutations in some of these checkpoints

00:55:23.120 are consumed by T, that T cells eat them alive. We have enough T cells in our body to eat

00:55:30.300 ourselves alive. Think of a disease like, you know, a T cell autoimmune disease like diabetes

00:55:36.940 type 1. It eats the pancreas alive. Think of a B cell, again, another immune cell disease,

00:55:47.200 like lupus. It eats our body alive. Think of scleroderma. Think of a patient. That patient

00:55:54.020 is eaten alive by their own autoimmune cells. But you and I are not. Why not? And that's because

00:56:00.040 all of our cells are carrying not one flag, not two flags, but dozens of flags saying

00:56:07.200 to a B cell and T cell, I'm normal, don't eat me, don't kill me, don't create an autoimmune

00:56:13.000 response against me.

00:56:15.480 So cancer cells, of course, figure this out by themselves, and I'm anthropomorphizing

00:56:22.120 a cancer cell.

00:56:23.120 Right, but that's okay.

00:56:24.440 It's a safe spot for that.

00:56:25.440 This is an evolutionary process, right?

00:56:28.100 The cancer cells that survive are the cancer cells that are able to put up these flags.

00:56:35.500 So that is the non-anthropomorphic variant of what I'm just saying.

00:56:40.700 Cancer cells that survive create flags that enable them to tell the T cells and B cells not to attack them.

00:56:50.820 And there's not one, there's not two, but a dozen such response elements.

00:56:56.400 and these response elements are ultimately responsible for the fact that even though

00:57:01.680 cancers have now, we now know, not one, not two, but dozens of neoantigens, that they're still

00:57:08.620 not attacked by the immune system. And so do you think that along the lines of the

00:57:17.560 inflammation angle, which I want to spend more time talking about, do you think there's a world

00:57:23.140 in which T-cell regeneration, just as people are starting to talk about how do we regenerate

00:57:29.380 cartilage, how do we regenerate corneas, things of that nature, are we going to get to the

00:57:34.440 place where we have serious considerations of medications, drugs, therapies that say,

00:57:40.060 look, if an 80-year-old could be armed with the 20-year-old version of their immune system,

00:57:48.120 their mortality is going to plummet.

00:57:49.980 Because not only are they going to be able to combat cancer, but also infections, right?

00:57:54.280 I mean, think about it.

00:57:55.480 Yeah, yeah.

00:57:56.140 So how much is that sci-fi and how much is that something that we might see in our lifetime?

00:58:02.440 Well, now we come to AI.

00:58:03.720 One of the most interesting talks I heard recently was from someone at Harvard who was looking through CAT scans of patients and asking the question about wellness, longevity, disease, cancer, etc.

00:58:28.480 using AI, and asking the question, was there anything in the scan of their bodies that

00:58:36.060 was correlative, correlative, not causative, that correlated with any of these markers?

00:58:43.600 And they found one marker, and it was the presence or absence of an organ called the

00:58:51.360 thymus.

00:58:51.820 So the thymus, T, the thymus is the reason T cells are called T cells, because the thymus is the, you can imagine it in metaphorical ways, as the baby nursery or the kindergarten where T cells grow up.

00:59:09.040 and people across thousands of people across thousands of disease states,

00:59:20.240 people who had an intact thymus in this study,

00:59:25.400 it's been well published, well noted,

00:59:28.460 people who had an intact thymus tended to have less diseases

00:59:34.120 such as autoimmune disease and cancer.

00:59:37.060 people who lost their thymus as they grew older because the thymus naturally involutes and dies

00:59:44.340 tended to have more now this is a correlative study it's not a causal study there you can

00:59:50.900 propose i right now you can think of 10 000 ways that you can probe this but it's super interesting

00:59:58.120 because all of a sudden it says oh maybe just like lipitor for cardiovascular disease or

01:00:07.040 hypertension for cardiovascular disease, maybe have a marker. Maybe the marker is the size of

01:00:13.720 the thymus or the presence of the thymus. The real test, of course, is starting with animals.

01:00:20.780 Take an older animal whose thymus has gone away, put a new thymus in, and see if you can restore

01:00:25.960 their capacity. These are difficult but important studies, but you're on the right track. Could we

01:00:33.940 make a T-cell booster, but what better T-cell booster to make than re-implant the kindergarten

01:00:39.820 where T-cells grow up and learn to become T-cells. And this study very clearly points to the idea

01:00:46.260 that that's one of the places to start. You know, there was a paper that came out,

01:00:52.180 I don't think it was quite 10 years ago, but probably around then. And it was,

01:00:57.860 and I say this with no disrespect to the authors or anything, it was a bit of a clumsy paper

01:01:02.080 aimed at thymic regeneration. And it was a little bit of a kitchen sink approach, right? They gave

01:01:06.660 these patients growth hormone and metformin and a whole cocktail of things. And on imaging,

01:01:13.180 they showed, hey, look, in the patients who got this, and by the way, it was even worse because

01:01:17.820 I think it was an open label study, so there wasn't even a randomization. But they said,

01:01:21.480 look, the thymus grew. And this sort of idea of thymic regeneration became very popular.

01:01:26.620 I haven't actually followed it closely enough to say, hey, has the field moved to a place where there are legitimate inquiries?

01:01:33.820 Because I don't even recall the rationale for metformin growth hormone and DHEA or whatever they were giving.

01:01:41.680 Have you followed this space at all?

01:01:43.380 And are there molecular attempts at this?

01:01:45.720 I have, I have. You know, again, the good news is we're going from a human, generally speaking,

01:01:58.080 the kind of study that I like is what I call a closed circular study. You start with a human

01:02:05.780 epidemiological observation. And it's an observation, this goes back to my old philosophy,

01:02:10.760 publish observations. You don't need to publish explanations. The observation is humans who

01:02:17.840 retained their thymuses tended to have fewer various diseases. This was picked up by AI

01:02:28.280 reading of multiple tens of thousands of CAT scans. Great. Now take that into an animal model

01:02:36.300 in which you can do an experimental implantation of the thymus.

01:02:41.780 This is possible to do.

01:02:42.820 You can actually open the chest, put a new thymus in,

01:02:46.660 and ask questions.

01:02:47.740 You can ask very detailed questions.

01:02:49.160 It'll take some time, but it will happen.

01:02:52.220 And then slowly work yourself back up to the human and close the loop

01:02:55.400 and find out, you know, what are the thymic growth factors?

01:03:00.240 I'm sure there's some.

01:03:01.680 Otherwise, you know, why would, let's say, your thymus,

01:03:06.200 You and I, let's say, are the same age.

01:03:07.880 I don't know if you are.

01:03:08.880 But your thymus is gone.

01:03:10.720 My thymus is not gone.

01:03:12.540 Can I find the growth factor that made your thymus involute and mine not involute?

01:03:17.440 Great.

01:03:17.840 I'll replace yours with the growth factor.

01:03:19.880 But, you know, putting in random stuff, metformin, that's just a mess.

01:03:27.280 So, look, science is systematic.

01:03:30.820 And we need to follow systems.

01:03:33.360 It's great when we follow systems.

01:03:35.320 and when we understand systems.

01:03:37.180 Sometimes we have to make a few leaps of faith.

01:03:39.280 It's okay.

01:03:40.560 Sometimes there's a lot of insight,

01:03:42.100 there's some intuition, blah, blah, blah.

01:03:43.680 It's okay.

01:03:44.820 But ultimately, it's a systematic way of thinking.

01:03:48.340 And there's nothing more to it.

01:03:49.340 It's a systematic way of thinking.

01:03:50.960 So yes, the answer is we'll find the hormones.

01:03:53.820 There's, I'm sure, look, I'm sure there's a hormone.

01:03:59.480 There's some hormone that sustains your thymus

01:04:01.480 and doesn't sustain mine.

01:04:03.300 yours is there, mine is gone. I need to keep mine alive. I'll keep mine alive. And then I'll have a

01:04:09.160 new way of preventing many diseases, including cancer. Before we leave the immune system,

01:04:15.680 I want to go back to something you brought up, which has been one of the great successes of the

01:04:19.880 past 15 years, which has been the success of CAR-T therapy, especially around the B-cell

01:04:25.460 lymphomas and leukemias, where, to your point, we were talking about durable cure. We're not

01:04:30.640 talking about extension of median survival. Maybe just spend a little moment explaining

01:04:35.700 what the CAR-T is, like what is a chimeric T-cell, et cetera. But where I really want to go with this,

01:04:42.220 Sid, is will we ever, or do you believe that there's going to be a future where the same

01:04:48.360 technology that has given us a CAR-T cell to treat CD19 B-cell disease is going to allow us

01:04:56.740 to treat a solid cancer like colon cancer, breast cancer, pancreatic cancer?

01:05:02.260 Okay, so let's start with what a CAR-T is. A CAR-T is, so remember we said T cells are able

01:05:08.180 to recognize cancer cells and kill them, provided that the cancer cell doesn't have enough flags

01:05:17.000 on its surface that prevents it from being recognized. Okay, so we get that. That's

01:05:22.440 standard immunotherapy, Keytruda is one example of that.

01:05:28.300 There are other ones.

01:05:29.680 Okay, a CAR-T is a slightly different variant of that.

01:05:32.820 A CAR-T is a variant in which I say,

01:05:35.260 okay, I'm gonna override all of those signals.

01:05:39.040 I'm gonna take cells from your body, 0.60

01:05:42.220 genetically modify them to recognize cancer cells

01:05:48.000 using a genetic modification system based on a virus.

01:05:52.440 The virus just basically inserts a gene into the T cells.

01:05:57.760 I'm going to grow those T cells either in the body or outside the body,

01:06:03.300 re-inject them in the body or change the T cells in the body.

01:06:08.440 And now those weaponized, very important word, weaponized T cells,

01:06:14.880 will go and kill your cancer.

01:06:17.600 This has been a great success in leukemias, B-cell leukemias, and B-cell lymphomas, 0.88

01:06:25.680 producing not just small cures, but durable cures.

01:06:29.620 Just as a plug for one of the things I do, I created one of the first CAR T-cell companies

01:06:36.340 in India, where we sell the therapy for 1 20th the price that is available in the United States.

01:06:44.120 We've treated almost 150-odd patients, maybe coming up to 200 soon.

01:06:50.460 So that's what a CAR-T cell is.

01:06:54.040 Now, your question is, why can't we do that for colorectal cancer and other tumors, like

01:07:01.020 brain tumors?

01:07:02.920 The answer is quite subtle.

01:07:06.040 And if you're a listener to this podcast, I want you to pay very close attention to

01:07:12.560 the subtlety of the answer, because the answer is extremely subtle. Why could we cure B cell

01:07:20.520 and B cell lymphomas and B cell leukemias? And why have we struggled to cure brain cancer and

01:07:28.180 others with these CAR T cells, which are weaponized T cells, which ignore, by the way,

01:07:32.940 ignore all these red flags from cancer cells. They just go kill them. The answer lies in the 0.92

01:07:40.100 fact that no one, and I'm sure that no one in your program has really talked about and very

01:07:46.560 few people talk about. So the CAR T cells that kill B cell leukemias and lymphomas

01:07:56.860 don't just kill B cell leukemias and lymphomas. They kill all B cells. I'm going to repeat that

01:08:05.700 fact. They don't kill B-cell leukemias and lymphomas. They kill all B-cells. And it just

01:08:14.780 happens to be the case, for reasons that we don't know, evolutionary reasons and others,

01:08:21.500 that human beings can live with B-cells killed. The problem, and that's great,

01:08:29.760 for B-cell leukemias, B-cell lymphomas, and other cancers. 0.99

01:08:33.740 But we can, a woman can live without any mammary tissue. 1.00

01:08:37.380 So why can't we just, yeah, let's kill every breast tissue in her body when she has breast 1.00

01:08:42.360 cancer. 1.00

01:08:43.120 A woman could live without her ovaries, if, I mean, a pleasant thought, but a woman could

01:08:49.420 live without her endometrium, not a pleasant thought, but yes.

01:08:53.640 So CAR T cells really should be thought about as lineage, not cancer specific, but lineage

01:09:02.220 and tissue-specific ways of killing cells.

01:09:05.560 Now, you can't live with, obviously,

01:09:08.780 one wonders, 0.88

01:09:10.360 but obviously you can't live without a brain.

01:09:12.940 So you can't create a CAR T cell

01:09:14.940 to a brain cancer

01:09:16.940 because your brain cancer

01:09:18.000 will almost certainly be

01:09:19.200 almost similar to a normal brain cell.

01:09:23.180 But you can live without an ovary.

01:09:25.120 You can live without an endometrium.

01:09:26.400 I mean, you can live without a pancreas

01:09:28.200 if push comes to shove. 0.65

01:09:29.280 You can live without a pancreas.

01:09:30.420 You can live without a prostate. 0.90

01:09:31.920 Yep. 0.92

01:09:32.220 You can live without.

01:09:33.480 So there are a bunch of tissues that you can live without.

01:09:36.840 We just need to find CAR T cells to be able to kill those cells.

01:09:40.940 And what's limiting us?

01:09:43.240 Do you believe that there are unique antigens on mammary cells, pancreatic cells?

01:09:49.100 Why do there's no belief here?

01:09:50.100 We've studied this to the nth point.

01:09:53.600 And the answer is, yes, there are.

01:09:56.120 So then there's no physical reason why we should not have a CAR T cell against breast

01:10:01.100 cancer then?

01:10:02.220 breast cancer is a hard example you've chosen a particularly hard one

01:10:09.500 pancreas is a hard one because there's you know there's not there are not very many unique antigens

01:10:16.840 in a pancreas cell that is not present in normal cells but i'll give you an example that's not so

01:10:22.180 hard ovary ovarian cells which re ovarian cancer cells which really are fallopian tube cells we

01:10:31.940 now know that ovarian cancer is not an ovarian cancer it's really a fallopian tube cancer long

01:10:37.040 story we can talk about it later but um ovarian cancer cells have an antigen it's very well known

01:10:42.620 have more than one antigen and there are multiple attempts at making car t cells against them one

01:10:48.060 of them is going to be successful.

01:10:50.920 What percentage of pancreatic cancers have KRAS mutations?

01:10:55.340 Aren't most of them?

01:10:56.580 Most of them have KRAS mutations.

01:10:57.880 And does the KRAS mutation produce an antigen that we could target with a CAR-T?

01:11:03.980 Not really.

01:11:04.640 It produces an antigen in some patients, but most of KRAS is hidden inside.

01:11:11.440 Most of KRAS is not a CAR-T target.

01:11:14.600 Now, Keras, I'm talking about Keras, Keras is a great target for small molecules that shut off its function.

01:11:23.660 So now we're not talking about the immune system.

01:11:25.500 We're talking about small molecules, locks and keys, keys that go and shut off the lock, Daxon Rassib, nasty side effects.

01:11:33.080 You've seen the senator with those lesions in the face.

01:11:37.660 I'm sorry about those nasty side effects.

01:11:39.500 I think we'll get over most of them.

01:11:41.340 not get over, but as in find nude medicines that no longer have those effects. But yes,

01:11:49.420 I think the way to attack pancreatic cancer is either to find a unique pancreatic antigen

01:11:56.680 that a CAR-T can attack. And if it takes all of the pancreas away, great. People with type 1

01:12:02.700 diabetes have no pancreas or they have no- Effectively. No endocrine pancreas, yes.

01:12:09.760 People with a Whipple surgery.

01:12:11.800 Yeah, they lose enough pancreas.

01:12:13.340 I mean, plenty of patients have Whipples with a significant...

01:12:16.700 I have no pancreas.

01:12:17.900 So yes, if we could eat up every pancreatic cell and people would still live, yes.

01:12:23.460 But we haven't found one yet.

01:12:25.500 That's one direction we could go in.

01:12:27.820 Another direction we could go in and say, listen, we can't eat up the whole pancreas,

01:12:31.740 including pancreatic cancer, but we can shut off the gene that drives pancreatic cancer

01:12:38.200 Keras. Dapson Rasib does that. Nasty side effects, but maybe we can live with that.

01:12:45.020 And that's another direction to go. What do you think it is about... Sorry, go ahead.

01:12:51.080 That's non-immunotherapy. Yes, yes, correct. What is it you think

01:12:54.360 about pancreatic cancer that makes it so devastating, right? So there's an argument

01:12:59.780 that's been made, and I'm not sure I agree with it, by the way, but I've heard it made by some,

01:13:03.960 which is any patient who is alive five years after they were diagnosed with an adenocarcinoma of the

01:13:11.980 pancreas didn't have an adenocarcinoma of the pancreas. So the argument that those people

01:13:19.080 are making, which again, I don't personally agree with, and that's probably my bias coming from a

01:13:24.260 surgical program that treated this, was that their argument is there is no such thing as early

01:13:29.980 detection of pancreatic cancer. The moment that cancer is present, it is a fait accompli. It has

01:13:35.240 spread. You are dead. The other argument would be, the whole steadying argument would be, no, 0.99

01:13:39.800 if you catch this at an early, early, early stage one and you do the appropriate Whipple procedure,

01:13:44.700 a subset of these patients can live. And yes, it's probably only 5% that will be surgically cured of

01:13:50.280 all comers. But if we could detect of those 95% who were caught too late, could we catch them

01:13:57.540 earlier where do you sit on this spectrum so look i mean so i agree with you on this i think

01:14:04.980 so so as you know the pancreas is a long organ it looks like a big leaf it's sitting sort of in your

01:14:11.540 in your gut they're fundamentally speaking there are two kinds of pancreatic cancer head

01:14:17.060 the ones that in the head one in the tail so if you think about it as a leaf the ones that

01:14:21.380 in the head are closer to the parts of your intestines the ones that in your tail are further

01:14:26.340 away. The tail ones tend to be detected incidentally. The head ones tend to be detected

01:14:32.880 with pain and jaundice and other things. I think that, surgically speaking, if you can find

01:14:40.940 the ones in the tail, they're less involved. They're less involved with other organs, and

01:14:47.640 it's an anatomically distinct kind of cancer. So I think, surgically speaking, some of those

01:14:54.220 patients can actually live longer. And I agree with you. The ones that are in the head,

01:14:58.900 the problem is that it's all, there's a lot of anatomy there. Your intestines are going down,

01:15:05.440 your blood vessels are crossing. It's in a place, it's sort of like in the, it's like, you know,

01:15:10.780 it's like a, it's like a traffic accident. It's probably the most complicated real estate

01:15:15.900 in the entire abdomen. Below the neck, below the neck, I would say it's the most complicated.

01:15:21.040 There are three truly complicated real estates in the body.

01:15:28.240 One is right here, one's right here, and the third one is right there.

01:15:33.120 So real estate-wise, it's like the traffic intersection of everything going everywhere,

01:15:39.660 and you have an accident.

01:15:41.960 And that accident ain't going to resolve easily.

01:15:47.040 So yes, you're absolutely right.

01:15:48.440 there's surgical cures of patients with, I have patients who lived 10 years with a

01:15:53.820 tail pancreatic cancer, not in the traffic accident area, far away from it, had a whipple.

01:16:01.180 And it was incidentally found and...

01:16:03.060 Incidentally found. And so the answer is, yes, it depends on the anatomy. Now, by the way,

01:16:10.700 what we don't know, and here's what's important. So cancer therapy almost always moves in the

01:16:18.280 following way. And you need, again, your listeners need to know about this. We can talk about

01:16:23.980 longevity later or in a different podcast. But listen, cancer therapy almost always begins with

01:16:33.680 therapy directed against diffused metastatic cancer. So the senator has diffused metastatic

01:16:42.600 prostate cancer, is on the drug, is living great. The next step in cancer therapy is to take

01:16:51.480 patients, take most of their cancer out surgically, and then give them the therapy after most of the

01:17:00.480 cancer has been treated surgically or with radiation to now kill the remaining cancer

01:17:07.140 cells, and that's called adjuvant or neoadjuvant therapy. Adjuvant because it is an accompaniment

01:17:16.760 to the main therapy, which in this case was the removal of the bulk of the tumor.

01:17:24.220 The real interesting question is, what's Daxon-Rasif going to do to adjuvant therapy? So

01:17:29.180 patient comes in, not the senator, the patient comes in, has painless jaundice, classic example

01:17:37.000 of pancreatic cancers, found to have a big lump goomba in the traffic intersection of

01:17:45.220 all of this, goes through Whipple.

01:17:47.540 Now, most of the visible cancer is out.

01:17:51.160 And now they have KRAS mutation.

01:17:53.960 We put them on Daxon RAS and then ask the question, are they better or worse off?

01:17:58.580 Is that trial being done?

01:18:01.200 It'll be done if it's not in the works.

01:18:03.540 I mean, is it in the works is what I mean?

01:18:04.920 It is in the works.

01:18:06.600 Absolutely. In a thousand ways.

01:18:09.720 And by the way, that's such an important example of a principle in oncology that has

01:18:17.940 some exceptions, but not many, which is the burden of the tumor matters. A lower burden of tumor is a

01:18:27.080 higher probability of success. Again, we can think of a few exceptions. Papillary thyroid of the

01:18:31.880 cancer doesn't seem to matter. These handful of really remarkable checkpoint inhibitors often

01:18:36.900 don't matter. But for the most part, fewer tumor cells is a better outcome. And it's a greater...

01:18:42.580 And this is where, again, our friend Bert Vogelstein says it very succinctly. He says

01:18:48.660 that there are three zeros between 1 million and 1 billion. And that means that if you have 1

01:18:56.680 billion cancer cells and you take them out and there's only 1 million left, that means that

01:19:03.340 there's 1 in 1,000, you've reduced the chances of 1 in 1,000 that you'll have a cancer cell that's

01:19:10.440 resistant to your therapy by 1,000. It's huge. Like by 0.001 degrees, whatever you want to call it.

01:19:21.060 Yeah, yeah. You touched on something at the outset, Sid, that I am not seeing enough hoopla

01:19:30.020 about this, and that is the staggering and upsetting rise of colorectal cancer in young

01:19:35.280 people. Now, I've taken an unbelievable amount of shit for this personally over the past 10 years, 0.98

01:19:42.480 which is driving towards earlier and earlier initial colonoscopy. And it's been very pleasant 0.98

01:19:47.960 to see that the formal recommendations have now moved from 50 to 45. But if you do the math,

01:19:54.840 you'll realize there are still many people dying of colorectal cancer under the age of 45. And

01:20:00.120 therefore, I believe that really we should start screening even average risk adults at 40 and

01:20:05.280 obviously earlier if they have any of the known hereditary drivers. I know why questions are

01:20:12.260 a fool's errand at times. But as an epidemiologist, if I ask you to put on your epidemiology hat, 0.61

01:20:19.040 what are some of the possible explanations for why this particularly lethal cancer,

01:20:24.820 it is the third leading cause of cancer death. Why are we seeing this go up? Is it antibiotic

01:20:30.300 use, gut health, obesity? Where do you think the culprit is?

01:20:35.440 So if I was to be totally honest, I would say I don't know the answer.

01:20:39.480 If I was to be a little bit more speculative, I would say it has to do, our colorectal milieu

01:20:48.300 is rapidly changed and has rapidly changed in the last few decades.

01:20:55.120 We are ingesting more and more forever chemicals.

01:21:00.980 And there is strong evidence from animal models that some of these forever chemicals,

01:21:06.000 including the ones that are ingested in food are very pro-inflammatory, so it brings us back to

01:21:15.580 the inflammation theory of cancer. We have strong evidence that these are multiplied

01:21:23.260 by changes in the microbiology of the gut, the gut microbiome, which is being disrupted by all

01:21:30.400 sorts of things, antibiotics and other foods included. And this is, I'm sorry, this is again,

01:21:38.840 maybe for your audience, uncomfortable territory, but Burkitt, Dennis Burkitt, who went to Africa,

01:21:47.180 the person who discovered Burkitt's lymphoma, went to Africa in the 1950s, 60s, I don't remember when,

01:21:54.140 but from oxford and he said that he virtually found no evidence of colorectal cancer especially

01:22:01.600 in young men and women because the diet was so fiber rich that virtually everyone

01:22:10.600 was essentially evacuating their colorectal content content twice a day in the united

01:22:22.220 States, we've reached a place where we're doing that once every two days or three days.

01:22:29.080 So my recommendation for people who want to avoid colorectal cancer is triple or double your fiber

01:22:36.180 intake. Obviously, reduce exposure to these forever chemicals. And based on no evidence

01:22:44.680 whatsoever. Use probiotics to, and probiotics are cheap. I don't mean go buy very fancy. I just

01:22:53.200 mean have live yogurt. Yogurt, yeah. Yeah, at least once a week so that you can replace your

01:22:58.840 probiotics. So we've touched a little bit on AI, and I know that it's something that we are both

01:23:05.240 thinking a lot about. I want to talk about it through the lens of maybe drug discovery. So

01:23:10.660 you've already given a great example of how AI can do things on a CT scan read that no human can

01:23:18.220 do. And of course, there are lots of reasons for why CT scans are a beautiful environment for AI

01:23:23.440 to shine. But let's talk a little bit on the drug side. Now, AlphaFold, the company that has

01:23:29.980 garnered a lot of much-deserved credit for using AI to be able to predict the sequences of amino

01:23:37.020 acids, which has allowed them to more rapidly develop drugs. This is a very exciting area.

01:23:44.320 So if we would say this is one area in which AI is going to fundamentally change drug discovery,

01:23:51.200 you could say, okay, another area would be AI could help us propose better hypotheses and or

01:23:57.020 design better experiments. That would be fantastic. Different idea.

01:24:00.760 Different idea. And then a third idea, and this is the one that I think personally is the biggest

01:24:05.940 unlock is having real simulated biology that would dramatically shorten the preclinical period,

01:24:13.860 but more importantly, increase the probability of success as we go from preclinical to clinical.

01:24:19.240 And then potentially a fourth layer would be extracting better insights from the actual

01:24:25.800 clinical trials to speed things up. So are there other areas you're thinking about this in and

01:24:31.860 then which of these do you think is the biggest unlock? I think all of them are big unlocks. I

01:24:38.020 don't think any of them is any less unlocked than the other. I think we need the entire ecosystem

01:24:42.660 to unlock because every one of these is a potential unlock. I'll just give you a couple

01:24:48.460 more unlocks. I'll give you one example in the beginning, which is using AI as an epidemiological

01:24:56.280 scanning tool. So that's how researchers found the thymus correlation, right? AI for the

01:25:05.200 recruitment of clinical trials. We haven't talked about that. AI for the monitoring of clinical

01:25:10.580 trials. We haven't talked about that. So it's AI all the way. My personal arena has been AI in

01:25:18.520 drug discovery. So as you very well know, every medicine, every drug, aspirin, Lipitor,

01:25:28.240 you know, Prozac, works by binding to something. It's a key and a lock, lock and a key. The protein

01:25:38.360 that it binds to is the lock, and the drug is the key. It opens or shuts down some function

01:25:47.040 of the protein. Lock and key is a little bit misleading because the proteins are actually

01:25:56.300 moving in space. They're wriggling around. So it's a wriggling lock. And so the key has to be

01:26:01.400 a little bit of a wriggle of a key that goes in. But basically, it's all aspirin is a key.

01:26:05.860 I know we know the lock. Lipitor is a key. We know the lock. Prozac is a key. We know the lock

01:26:11.600 and so forth. All right. So in the past, we had two ways of looking for keys.

01:26:20.040 Either we could design them from scratch. So we could look at the protein and say,

01:26:24.700 what is likely to look like a key? And we needed an enormous amount of knowledge of chemistry

01:26:33.840 and chemical biology to do that.

01:26:38.260 This is why the smartest chemists in the world

01:26:42.560 were required to create these keys to the lock.

01:26:47.380 And another way to do this is to do it a little bit more

01:26:51.680 like a search function.

01:26:52.720 So let's make a library of a billion keys

01:26:56.400 and try to see which one fits.

01:26:59.820 Now, both of those, the first one I'll call

01:27:03.340 synthetic molecular design, generative molecular design. And the second one I'll call

01:27:09.660 search-and-fit molecular design. Both of them are very amenable to AI and computation.

01:27:17.040 So search-and-fit is a little bit harder because the number that I quoted to you,

01:27:23.700 a billion, is not a billion. It's actually 10 to the power of 20.

01:27:29.120 Which is not, that's technically not computational, is it?

01:27:31.720 So that is technically beyond the level of any computational.

01:27:35.140 So there must be an inference that allows this to happen.

01:27:38.460 That's right.

01:27:38.880 You have to do some inference that allows it to happen.

01:27:40.900 That's right.

01:27:42.160 So you have to have an inferential way of taking.

01:27:46.680 So in other words, let's boil it down to simple language.

01:27:49.820 You start with some keys and you begin to infer the properties of the key based on what fits and doesn't fit.

01:28:00.180 it's turned out that that is computationally too expensive right now.

01:28:06.500 And it probably require, I mean, this is my guess,

01:28:09.360 it will require quantum computers able to do that level of inferential reasoning.

01:28:15.600 Because you probably need to fit at least a billion keys

01:28:19.080 to learn enough about the kind of key that you need to really open the lock.

01:28:25.420 Which is amazing because it's more than a billion billion, so that's a huge shortcut.

01:28:33.860 It's more than a billion billion.

01:28:36.120 But that's still an enormous undertaking.

01:28:39.600 The second way, which is the usual way that we are doing drug discovery, is to look at the inside of the lock and ask the question, what kind of key will fit into this lock?

01:28:55.420 how do I design a key? And synthetic molecular chemists working in laboratories, in pharmaceutical

01:29:07.660 companies, etc., they have rules. They know that if there's a crevice here, I'm going to put a

01:29:16.880 certain kind of molecule there. If there's a hollow surface there, I'm going to put a certain

01:29:21.060 kind of molecule there, et cetera, et cetera. So this is called, this is generally speaking,

01:29:26.460 it's generative chemistry, generative because you're making it from scratch. You're not just

01:29:31.800 searching and fitting, but you need to know the rules of chemistry and physics. This is not

01:29:38.420 biology anymore. This is pure chemistry and actually really is pure physics. The good news

01:29:44.420 is that you can teach an AI all those rules.

01:29:51.020 And that's what we've done at Manus.

01:29:53.280 We've taught AI all the rules of synthetic molecular chemists

01:29:57.420 and said, let's take one synthetic molecular chemist

01:30:01.220 and make them into 50,000. 1.00

01:30:04.800 Because that's what AI does.

01:30:06.600 AI can scale.

01:30:08.200 And if you have one, you can make 50,000.

01:30:10.640 And so the answer to the question is, that is what we do.

01:30:17.600 We look at the shape, the geometry, the electrostatics,

01:30:23.640 all the elements that form the pocket where you want to put your key.

01:30:28.940 And we design a key from scratch using basic molecular principles.

01:30:34.620 And if you can do that, you can essentially recreate.

01:30:37.900 you can make the smartest generative molecular chemist that ever lived.

01:30:44.740 And the last plug I'll make for this is that this is the best, to me, the best use case,

01:30:50.340 proximal use case of good AI in the world. This is not a chat bot. It is not a image-making bot.

01:30:59.880 It is making medicines for human diseases.

01:31:04.780 How difficult do you think it will be to teach AI the rules of biology to thereby simulate

01:31:14.560 actual experiments? I mean, you very briefly touched on how much more complicated

01:31:22.340 biology is than physics or chemistry. The rules are not as neat.

01:31:28.940 There are barely any rules.

01:31:30.920 Yeah. So is this actually something we can aspire to?

01:31:34.780 Yeah, I think it will come. I'm a little, look, I'm personally skeptical of how useful it's going to be because, look, the biology of a kidney cell, a kidney cell has a very particular biology. Its job is to filter out salts.

01:31:55.560 Right. It's quite well known. It's very discreet.

01:31:58.920 A cell in your brain has no resemblance. It has some resemblance, of course,

01:32:05.320 metabolically is the same, blah, blah, blah. But its job is completely different. Its job is to

01:32:10.520 transmit information, sequester information, storage, whatever it might be, right? A cell

01:32:16.180 in your skin is a completely different kind of cell. So I'm a little skeptical of...

01:32:22.520 And it's also, just to add one more layer there, Sid, sorry to interrupt, it's that there are

01:32:26.080 certain cells in the body like a muscle cell where it's there's a very mechanical and relatively

01:32:32.160 straightforward functional determination of what it can do whereas the neuron just to use that

01:32:38.640 example again completely different like what's happening in the neuron is not as mechanical as

01:32:43.580 what's happening in the myocyte or what's happening in the nephron where you can actually just see oh

01:32:49.500 this ion moves through this channel against this gradient and when this gradient switches it does

01:32:53.640 this? I think there's some use to building biological models of cells. Do I think it's

01:33:04.100 going to be easy? No. Do I think it's going to be until you get down to every cell-by-cell level,

01:33:09.440 is it going to work? No, I'm a skeptic, but skeptics are often wrong.

01:33:15.780 Sid, if we're sitting here in 15 years, which I hope we are going to be, both you and I

01:33:21.600 personally sitting here having another discussion in 15 years, looking back and saying, God,

01:33:25.700 I can't believe it was 30 years ago you wrote Emperor of All Maladies.

01:33:29.700 If you had to predict what we're going to say in 15 years that is true about cancer,

01:33:35.580 that is not yet true today, what do you realistically think that might be?

01:33:41.480 Oh, I think that, you know, we've covered a lot of it. New antigens, new theories of how cancers

01:33:48.400 arise because of inflammatory states or other states, the role of obesity, all of this will

01:33:56.320 become clear. Where do you think it will translate to the most life saved? Do you think there's a

01:34:07.120 chance in 15 years? The answer is always going to be prevention. Now, the question that's a real

01:34:16.340 question is, can we do general prevention? Is prevention some kind of like habit change,

01:34:22.640 behavioral change, lifestyle change, or is it chemoprevention?

01:34:26.560 I was just about to say, will there be a drug? Just as I can give you Lipitor today,

01:34:31.720 long before you have a heart attack, and I can give it to you with an ACE inhibitor if your

01:34:36.320 blood pressure is high, these two molecules will prevent you from having a heart attack.

01:34:42.340 it's chemo prevention. Do you actually think there will be a day when we have molecules that

01:34:49.220 will prevent cancer? Yes, there will be chemo prevention. It is not clear to me that the

01:34:57.700 chemo prevention is going to be as simple. So remember, Lipitor blocks decreases your LDL levels

01:35:08.880 and Losartan or whatever ACE inhibitor or ARB

01:35:14.680 inhibits your capacity to have high blood pressure

01:35:18.640 by acting on your kidney's hormonal system.

01:35:24.500 It's not clear to me that it's going to be

01:35:26.320 sort of a relationship that's like that.

01:35:29.300 Although what you described with inflammation,

01:35:32.660 there's actually an analogy in cardiovascular disease.

01:35:35.360 As I'm sure you know,

01:35:36.220 we are now starting to check for CHIP in patients. And while our therapies are still limited,

01:35:43.520 we have colchicine. Now it's a crude drug, but there's evidence that if you have, you know,

01:35:50.240 if you're- Oh, we have better as an otacine, by the way.

01:35:52.660 Yes. I'm saying on the cardiovascular side, but I think that this is an opportunity in

01:35:57.400 cardiovascular medicine to go after residual risk where you identify clonal hematopoiesis

01:36:03.440 in a patient that's posing risk for CVD, we treat that preemptively. And that's really our first

01:36:09.500 immune-based therapy for heart disease. So there's a model, I think, that I would be very excited if

01:36:15.340 we're doing that in 15 years for oncology. Yeah. I mean, as I said, we have some tools.

01:36:20.700 We have the GLP-1s, for instance, in obesity and potentially in inflammation. So that's a trial

01:36:26.780 that I would like to watch very carefully. I hope someone listening to us today, by the way,

01:36:32.920 has taken, has become motivated by that idea.

01:36:36.120 Yeah.

01:36:36.560 And, you know, you can use, you don't have to go all the way.

01:36:39.560 You can use breast and endometrial cancer as your sort of two simple, not simple, but

01:36:45.780 two models.

01:36:48.000 Yeah.

01:36:48.520 Models. 0.98

01:36:49.080 Yeah.

01:36:49.800 There are a variety of anti-inflammatories of which we know some.

01:36:54.420 But again, you know, Peter, I know you care about things like longevity and I do too.

01:36:59.780 You're not the only person who cares, but longevity, obviously.

01:37:04.020 It may turn out that the answer is orthogonal.

01:37:07.500 It may turn out that the answer is that, and we know to some extent this is,

01:37:13.560 we already know the answer to this question, that, again, I'm going to use my words carefully here.

01:37:20.060 And the prevention of pathological disease, cancer, heart disease, the two biggest killers, is not the answer to the prolongation of longevity.

01:37:39.040 So I'm going to say that sentence again.

01:37:40.920 I'm going to say it very slowly, and I'm happy to have to be pushed back on it.

01:37:46.240 the current model of longevity

01:37:50.280 is the prevention of diseases that kill us.

01:37:57.060 Cancer and cardiovascular,

01:37:58.920 or one current model,

01:38:00.020 not the only current model.

01:38:01.680 But one current model of longevity

01:38:03.260 is the prevention or the elimination

01:38:06.300 or the treatment or the cure

01:38:08.040 of diseases that kill us,

01:38:10.340 of which obviously the two big ones

01:38:11.840 are cardiovascular disease and cancer.

01:38:13.460 But the orthogonal way to look at longevity is to say, I'm actually not interested in

01:38:21.700 the diseases, cancer and heart disease, the two most common killers.

01:38:26.920 I'm actually interested in a physiological state that promotes the elimination or the

01:38:37.280 decrease of not just those diseases, but all diseases, including diseases of degeneration,

01:38:44.040 osteoarthritis, immune failure, bone marrow failure, chronic unopposed inflammation, etc.

01:38:54.500 That's a question which is a deep, deep, deep question. And when I say orthogonal, I mean,

01:39:03.760 And maybe if we could change the physiological state of the body, reduce the inflammatory

01:39:14.920 state, reduce the chronic insulinemic tone, and there's a big maybe there, because otherwise

01:39:22.840 everyone's going to come and attack me and tell me I'm a crackpot.

01:39:25.940 But maybe it is that the reduction of the chronic inflammatory insulinemic tone and the chronic tone that comes from thymic reduction, I mean, all the things we talked about before, wellness instead of illness, maybe that's the path to longevity.

01:39:52.080 You know, Sid, I think about this all the time.

01:39:54.100 I know you think more than I do.

01:39:56.120 No, no, no.

01:39:56.700 Yeah, I'll share my thoughts.

01:39:58.640 So I agree with the broad statement for sure.

01:40:03.620 My concern is that different diseases behave differently.

01:40:07.680 So when it comes to atherosclerotic cardiovascular disease, which is indeed the number one killer, not just in the U.S., but globally, there is actually a world in which we can turn that into an orphan disease.

01:40:20.400 because the causal drivers are well-known

01:40:22.920 and we have treatments for all of them.

01:40:24.920 You can take aspirin, Lipitor, and PCSK9.

01:40:27.700 You can take like 17 drugs now.

01:40:29.840 That's right.

01:40:30.480 So if you don't smoke,

01:40:32.440 if you take whatever drugs are necessary

01:40:34.420 to stay insulin sensitive,

01:40:35.900 if you obliterate your ApoB and LDL-C

01:40:38.560 down into the physiologic levels you were born with,

01:40:41.760 and if you ever develop hypertension,

01:40:43.680 you immediately nip it in the bud with a drug,

01:40:46.180 you can't get heart disease.

01:40:47.920 You just can't.

01:40:48.660 The pathophysiology inside the artery wall can't happen.

01:40:53.420 So you're going to die of something, but it won't be the number one cause of death.

01:40:57.940 I don't have that degree of confidence that there's anything we could do that will eliminate

01:41:04.420 all cancers.

01:41:05.440 So even if we were completely insulin sensitive and we never smoked and et cetera, et cetera,

01:41:10.880 et cetera, we still have enough genes.

01:41:14.700 We still have random mutations.

01:41:16.280 we still need to rely heavily on early detection and treatment.

01:41:22.740 All fair, all fair, except let me give you the only counterfactual to that,

01:41:27.260 which is that people who are super, super generians,

01:41:35.020 people who, you know, more than 110, they don't die of cancer.

01:41:40.120 What they're dying of is failure of skin regeneration, their skin falls off, they're

01:41:49.240 dying of bones being broken, basically they're dying of regenerative failure.

01:41:56.080 But believe it or not, they still die of atherosclerosis and they still die of infections, like

01:42:02.060 their immune cells are breaking down.

01:42:04.500 Infection is immune failure.

01:42:05.840 Yes, yes, yes, yes.

01:42:06.620 So, but what they're not dying of, I mean, assume for a second that they took, you know, every anti-arthesterotic drug, which they don't, by the way, because they're in Greece or some island in Corfu or somewhere else.

01:42:24.420 God knows.

01:42:25.600 But what they're not dying of is cancer.

01:42:28.300 they're dying of bone breaks and skin failure and immune failure and blood failure and

01:42:37.140 you know all of these other things that's the only that's the important counterfactual i'll

01:42:42.780 present to you which is to say that there are two worlds it's almost like there are two worlds

01:42:47.780 there's one world in which humans die at age 70 80 90 of

01:42:54.620 the usual suspects, the three major usual suspects, neurological disease, cancer,

01:43:03.820 heart disease, in opposite order. Yeah, whatever. And there's another world in which once you're

01:43:09.420 not dying of those diseases, you can sail into your hundreds, not sail, get into your hundreds,

01:43:17.520 And there, your killers are not those diseases.

01:43:23.060 There, your killers are immune failure, blood system failure, bone breakage, basically regenerative

01:43:30.840 failure.

01:43:32.480 You know, I've been writing a lot about, thinking a lot about Eli Mechnikov, who the

01:43:38.500 Russian Parisian scientist who discovered the innate immune system.

01:43:45.020 And Mechnikov in the 1890s, get this, in the 1890s, had a theory.

01:43:52.540 He was the first person to propose that we study death systematically.

01:43:58.680 He created the word thanatology, the study of death, which, of course, the opposite of which is a study of longevity.

01:44:06.000 Mechtikov argued that if you could remove the proximal causes, the pathological causes

01:44:17.980 of death, cancer, heart disease, neurological disease, and of course-

01:44:21.680 Accidents and infections, yeah.

01:44:23.160 Accidents, infections, et cetera, et cetera, that human beings could live to 100, 102,

01:44:28.880 and they would, at some point of time, would naturally want to die,

01:44:35.240 just like trees at a certain age, hundreds of years, finish their sap and die.

01:44:45.820 It was a very radical idea in its time.

01:44:50.260 And I often come back to this thought thinking, maybe that's true.

01:44:54.640 Maybe that's where we are.

01:44:57.360 maybe eventually we need to think about, instead of thinking about, oh, let's pick off the buds

01:45:07.440 one by one, atherosclerosis, cancer, et cetera, or maybe keep picking off the buds. It's okay.

01:45:13.760 Right. But we need something that goes after the foundation of aging.

01:45:17.720 We need to go to something. That's the point.

01:45:19.900 And so what is your degree of optimism? What does the sort of AI buzz say? Everybody's saying, look, AI is going to cure this. We're going to be immortal. We just have to not die for the next 10 or 15 years and we're going to become immortal. And I'm struggling to come up with a steel man for that argument.

01:45:40.760 No, no, no, no, no. You know, I think, as I said, regenerative failure is a real phenomenon. And treating regenerative failure is not easy. It's going to be another, that is a huge new frontier in biology.

01:46:01.120 We haven't even faced that frontier in biology.

01:46:03.820 Why do, what is, you know, what is regenerative failure of cartilage?

01:46:11.520 I don't know.

01:46:13.260 But what is regenerative failure of the brain?

01:46:15.020 I don't know.

01:46:16.240 What is regenerative failure of the immune system?

01:46:17.940 Okay, fine, there's no thymus.

01:46:19.480 How can I make a new one?

01:46:20.460 I don't know.

01:46:21.900 So the new frontiers of longevity will have to deal with not just sort of, oh, you know,

01:46:29.260 Let's change our diet and make ourselves, I don't know, somehow dietary perfections

01:46:38.760 or inject ourselves with some random peptides and make ourselves, by the way, everyone knows

01:46:45.260 this, but let me just remind you, there are 20 amino acids per individual molecule.

01:46:53.400 you, so the combinatorics of injecting yourself with a random group of peptides, the chances

01:47:00.840 you'll find that the right peptide at the right concentration are basically zero, like

01:47:05.680 zero, not, it's like there's no, this is a mathematical truth.

01:47:12.180 Well, it goes back to what you said about the billion-billionth chance, like a billion-billion

01:47:17.300 chances, yeah.

01:47:18.640 But on the other hand, if you were to find the right peptide, I'd be rooting for you, but not you, but rooting for the person who found the right peptide.

01:47:27.560 Listen, the problem is that, yes, I think there is a hack, as the Silicon Valley people want to call it, to longevity.

01:47:40.800 The hack, but the hack is not just a simple hack.

01:47:44.960 It's not going to be peptides.

01:47:46.460 it's going to be finding out systematically how to address these regenerative failures

01:47:53.620 and separate it out from the diseases that kill you at 70 and 80. Let's separate those out.

01:48:02.120 Yeah, we need both of these happening in parallel, for sure.

01:48:05.300 Exactly, yeah, they're both in parallel, right? So you remove the diseases, and you have a good

01:48:09.940 example, atherosclerosis is a good example, right? You remove atherosclerotic heart disease

01:48:16.740 using all the mechanisms, all the medicines that we have now, including behavioral change,

01:48:21.600 most importantly, behavioral change. And then you're left with the question of regenerative

01:48:27.140 failure. And that's not a hack. That is a powerful, systematic study of how you will remove

01:48:36.260 the regenerative failure of people who are in their 70s, 80s, 90s, including the regenerative

01:48:43.900 failure of the brain, which we know the least about. And yes, then I can conceive of physical

01:48:53.120 longevity. Now, there's a whole other world in which you can talk about virtual longevity.

01:48:58.400 Sure, sure. Yeah.

01:48:59.520 That's a different, you know, that's a, you know, matrix, red pill, blue pill.

01:49:03.060 Yeah, beyond my pay grade.

01:49:04.180 it's beyond my figure 0.98

01:49:06.140 but it's true

01:49:08.500 fair enough you can imagine

01:49:10.040 there's a wet pill

01:49:11.600 matrix longevity universe

01:49:14.480 that you no longer

01:49:16.040 need your physical body

01:49:18.500 or maybe robotics

01:49:19.700 takes over most of it god knows

01:49:22.120 maybe your brain in a box or whatever

01:49:23.880 let's just put those aside

01:49:26.620 for a second I'm just talking about the physical body

01:49:28.280 and I think that those

01:49:30.560 are not hacks these are

01:49:32.360 they need systematic

01:49:33.920 And we're beginning to find out.

01:49:36.160 AI is helping us find out.

01:49:37.360 I'll give you one example.

01:49:39.220 There are many.

01:49:40.600 But this is not really a priority of the NIH.

01:49:43.840 I mean, the NIH is not putting the resources into studying the hallmarks of aging and gyroscience.

01:49:52.080 Because to your point earlier, Sid, if you want to start answering these fundamental questions, you must do them systematically.

01:49:58.700 Right?

01:49:59.060 And you started with a human observation.

01:50:00.980 It's a very important human observation.

01:50:02.160 Now we need to go back and do the animal models that begin to look at the ways that one could

01:50:08.200 actually perturb that system.

01:50:11.420 And then, of course, you bring that full circle back into humans.

01:50:14.600 But this is very big science.

01:50:16.820 This is a Manhattan Project of science.

01:50:18.800 I would argue that it nearly lies outside the current NIH mandate.

01:50:27.760 And we need a new NIH.

01:50:29.000 if we were to do this, which we should or could, it needs a new NIH mandate. The problem with

01:50:35.720 geroscience is, remember, again, history is very important here. What changed, what really changed

01:50:43.700 cardiovascular science? It was the fact that we found biomarkers for... They were proxies for

01:50:50.580 disease. You can measure blood pressure, you can measure cholesterol. You can measure blood

01:50:54.440 pressure can measure cholesterol. And very soon, we said, we don't need to wait for heart attacks.

01:50:59.660 If there's a drug that reduces blood pressure, if a drug that reduces cholesterol, we'll make

01:51:04.120 that drug available. And in post-talk studies, we found that those medicines actually reduce

01:51:10.660 heart attacks. You don't have that for cancer. And that's a separate question.

01:51:16.320 You don't have it for aging.

01:51:17.360 You don't have it for longevity. Yeah. So there's talk of biological flocks.

01:51:22.340 So here's my question back to people who study longevity science, an important science,

01:51:29.780 very important science.

01:51:31.680 Find me a reliable biomarker.

01:51:34.760 The problem with finding me a reliable biomarker is that the biomarker study itself will take

01:51:40.080 15, 20, 30 years.

01:51:43.320 Because you can't do it retrospectively.

01:51:44.760 You can do it sort of retrospectively.

01:51:46.160 You can say, what is there in the blood of people who live at 100?

01:51:49.740 Yeah. And we can start with, I mean, and I've talked about this with others on the podcast. I mean, we could go back and look at the interventions testing program, which is that great NIH program that looks at all the molecules. You could just begin by looking in the metabolome and the proteome of all the ITP winners versus losers, and we could at least get some clues. But yes, just finding the biomarker is big science, is very big science.

01:52:16.320 It's very big science. It's very big science. And it's a reliable science. The only problem

01:52:21.160 with it, or the question with it that I would argue, is that we can't screw it up because...

01:52:28.080 Right. It sets the trajectory of where we're going.

01:52:30.280 It sets the trajectory indefinitely. So we need to find it, confirm it, find it. And

01:52:37.520 once we've found it, again, remember Lipitor, remember cholesterol, remember hypertension.

01:52:43.740 You can do dozens of studies to think about its systematity.

01:52:51.160 But until that's found, we're sort of stuck.

01:52:55.920 You know, it's funny.

01:52:56.420 Just a couple of weeks ago, I was talking to a friend of mine who I'm sure you know,

01:52:59.580 David Sabatini, good friend of Lou's.

01:53:02.060 Yes, and a friend of mine.

01:53:03.480 And so David was lamenting this.

01:53:05.740 He said, look, what we really need is a Manhattan project around this.

01:53:10.900 And his view is it's not going to be the NIH stepping up. It's going to be a group of philanthropists that decide that we need to put money. But his point, which I think is absolutely spot on, is the effort must have a significant engine in the basic science. It cannot go straight to drug discovery, which is the temptation.

01:53:31.660 When these things are started as for-profit companies, you have to return to shareholders quickly.

01:53:38.020 You've got to go straight to drug discovery and whack-a-mole, whereas David's view is, and I think he is correct, you have to have a continuum, and you must have basic science.

01:53:47.860 He's absolutely correct, and not only that, he's absolutely correct because, again, the loop is quite clear.

01:53:53.040 You start with human observation.

01:53:54.460 You go backwards.

01:53:56.100 You understand the basics, and David's absolutely correct.

01:54:00.060 And, you know, you know, he's been studying this for years with Lou, with me, with others.

01:54:05.780 There's no short, unfortunately, there's no shortcut.

01:54:08.800 And this is not a drug discovery.

01:54:11.040 That's right.

01:54:11.840 This is a cholesterol discovery project.

01:54:14.700 This is a hypertension discovery project.

01:54:17.300 Yeah.

01:54:17.680 And unless we discover cholesterol, hypertension, et cetera, for aging, we're going to be lost.

01:54:22.820 And it requires a 30 year Manhattan project, maybe 15 years if we're good with AI,

01:54:29.080 Maybe you can shorten that down. And the NIH has no mandate to, right now, maybe a new NIH will, but the current NIH doesn't have a mandate for doing this.

01:54:40.340 I don't think the NIH ever will. And I think, yeah, I think there's many reasons why this is not going to be within the purview of NIH.

01:54:50.140 age. Unfortunately, it should be, but it won't be. And therefore, I think you, me, Lou, David,

01:54:55.780 and a bunch of us are long overdue for a dinner to brainstorm some ideas here.

01:55:00.560 Well, I'm happy to have the dinner. And as I said, if there are people who are

01:55:04.220 truly interested in this without the temptation of-

01:55:09.760 We need a drug. We need a drug within six years.

01:55:12.960 If there are people, then we should really launch this, launch it systematically. There is a

01:55:18.700 National Institute of Aging, as you know, they could, if they wanted to, they could

01:55:25.460 change some of their focus. And there is a focus on well aging. The problem is the NIH has banned

01:55:35.020 it as illness. It's not wellness. It's not. And that is a great thing. We've done great things

01:55:42.020 with it, and it has produced incredible things from vaccines to, you know, Lipitor, great.

01:55:50.840 That's wonderful. It's not the current mandate. Well, Sid, sounds like I need to get up to Boston

01:55:58.860 or New York, and we all need to grab dinner. Well, as you know, I mean, this metabolic trial...

01:56:04.980 Yeah, you want to just... I don't know how much you can talk about it. Can you give us...

01:56:07.980 Is there anything you can give us? Well, it's a public company now,

01:56:11.140 So it's all good. But listen, you know, the idea was initially proposed by you to Lou and Lou

01:56:17.700 Candley and to me. And then Lou Candley and I sat down for dinner in a restaurant in the Upper East

01:56:23.420 Side and on the back of a napkin sketched out a potential strategy, which is basically the idea

01:56:29.740 that cancer cells must depend on particular metabolic inputs, which make them different

01:56:40.100 from normal cells. I mean, know that from Otto Warburg's work in the 1900s, the cancer cell,

01:56:46.980 the metabolism of a cancer cell is not the metabolism of a normal cell. And every time

01:56:51.500 you say something is different about a cancer cell than a normal cell, that makes it possible.

01:56:55.580 It's an opportunity.

01:56:56.440 It's an opportunity. So in this case, we use that opportunity to make new kinds of drugs

01:57:02.820 that would particularly target cancer metabolism

01:57:06.600 and less so normal metabolism.

01:57:10.800 And that led to a company called Fayeth,

01:57:13.520 Fayeth Therapeutics.

01:57:15.240 It went public maybe two weeks ago,

01:57:17.920 four weeks ago, God knows when.

01:57:19.660 But it's in the middle of trials

01:57:21.000 and early trials are very promising.

01:57:22.720 And we'll see, we'll have to see.

01:57:24.840 It's a public company,

01:57:25.980 so I'm actually no longer privy

01:57:28.220 to the internal information.

01:57:30.840 I'm a founder, but Lou is a founder.

01:57:33.880 We no longer know anything about the internal information about the company.

01:57:37.380 But the last we heard, the last public disclosures that we heard, that the idea that targeting

01:57:44.240 cancer metabolism in endometrial and breast cancer was working.

01:57:49.300 So we're happy about that.

01:57:50.300 Well, we will continue to follow that story.

01:57:52.860 And if it's a success, we owe you a dinner in Manhattan.

01:57:56.780 Yeah, I remember that dinner very well.

01:57:59.300 it was sushi and it was a great sushi dinner. Are you working on another book? You mentioned

01:58:04.620 your writing. Yeah, I'm working. Well, this is why I talked about a lot of this. I'm working

01:58:08.820 on a book on death and dying and longevity, but not this kind of usual, oh, you know, let's get

01:58:15.120 to, it's the kind of stuff that we talked about. How do we really systematically think about

01:58:21.600 longevity, systematically think about death? What is your process for writing? I'm just

01:58:27.660 going to have to embarrass you a little bit, um, because it's possible there are people listening

01:58:31.440 to this podcast who haven't read any of your books. I've read all of them multiple times.

01:58:36.200 And I, um, I'm not just saying this, Sid, I think you are, uh, not only a generational talent,

01:58:42.680 you are the single greatest writer I have ever read when it comes to the discipline that I find

01:58:48.720 myself in. And I, and I really mean that Sid you, uh, and I think I've told you this before

01:58:53.060 Or when I was struggling to write my book, I would curse you constantly and think, God 0.98

01:58:58.300 damn it, he does it so well. 0.99

01:59:00.960 My book is going to be garbage. 1.00

01:59:03.060 I'll never be able to do what he does.

01:59:05.220 And I can't, but that's OK.

01:59:07.440 And I don't want to suggest that you're just gifted and it doesn't come down to how hard

01:59:11.140 you work.

01:59:11.600 But my guess is it's a bit of both.

01:59:14.160 I think you have a way with words and you obviously work very hard at it.

01:59:17.720 So tell me a little bit about the process by which you're writing this book.

01:59:21.320 Look, you know, the first part of the process, and this is what I generally tell people who ask me this question, it's going to sound like a generic answer, it's not generic, it's real.

01:59:35.320 Well, I think it's very important before starting to put the first word on the page is to spend a lot of time.

01:59:44.120 I call it dream time.

01:59:46.140 People can have different words for it.

01:59:48.380 Thinking about what the book will sound like.

01:59:52.420 Who's driving the narrative?

01:59:56.680 Because it has to be a who.

01:59:57.980 Where is it placed? 0.99

01:59:59.800 Where does it go?

02:00:01.540 Where is it geographically?

02:00:03.540 What time is it in?

02:00:05.320 What mood does it convey?

02:00:07.960 Who's the reader and why is he or she being moved by it?

02:00:12.900 What's the structure?

02:00:15.040 What's the anatomy?

02:00:17.240 Is it compartmentalized?

02:00:20.840 What do the chapters seem like?

02:00:22.440 What do the big sections look like?

02:00:25.960 And I think that before you put the first word down for a book,

02:00:30.600 You should spend months, weeks, years thinking about all of that, because that's what's the

02:00:40.200 basis, that's the center of a book.

02:00:43.520 And once you've gotten that piece straight, it turns out that writing a book, I mean,

02:00:49.100 it's complicated, you know, if you have a lot of explanation, blah, blah, blah, blah.

02:00:53.020 But then the book just kind of moves along much more quickly.

02:00:56.560 so in other words you know musicians often say this and you have to listen to the musicians

02:01:03.020 because of course there's a lot of closeness between prosody and music musicians often say

02:01:09.900 you need to know the mood of the piece you're playing before you play the piece long before

02:01:15.460 you play the piece it's the mood of the piece that's important the the notes are just a mechanism

02:01:21.860 To drive the mood

02:01:24.200 By the same logic

02:01:26.340 You need to know the mood of the book

02:01:27.820 And the skeleton

02:01:30.120 Once you have the skeleton

02:01:31.600 You can hang meat on the skeleton

02:01:32.780 You can hang pages on words

02:01:36.040 Etc, etc

02:01:36.900 So my

02:01:39.100 Advice, process, whatever it might be

02:01:42.080 Don't write the first word

02:01:44.540 Before knowing anything

02:01:46.220 About all of this

02:01:47.120 Once you know all of this

02:01:49.000 You can write the first word to the last word

02:01:51.680 for 600 pages, 700 pages, it'll just come to you.

02:01:56.840 So your writing, the actual period of pen to paper,

02:02:00.560 is short, it sounds like, relative.

02:02:02.580 My writing is actually short.

02:02:06.100 My editing is not.

02:02:08.080 I edit very great.

02:02:09.360 I give time.

02:02:10.480 I have great editors.

02:02:11.580 Nan Graham has been my editor from day one.

02:02:14.560 She's an incredible editor, one of the story editors.

02:02:16.960 you know, she's edited Pynchon and Roy and Stephen King and God knows what. She's a great

02:02:24.760 editor. We work together. Editing is not a slow process. It's not a fast process. It's a slow

02:02:28.320 process. Writing a book is a fast process. So lots of thinking, speeding through the writing,

02:02:37.400 slow editing. Yes. Yeah. Fast on the straightaway, slow on the curbs.

02:02:41.840 Yeah.

02:02:43.520 That's what we think about it, yes.

02:02:45.640 Sid, it's always a pleasure.

02:02:48.060 Always a pleasure.

02:02:48.660 Thanks for having me.

02:02:49.540 Thank you so much for everything.

02:02:50.980 This has always been a great discussion.

02:02:52.840 Great.

02:02:53.220 My pleasure and good luck and I owe you a dinner.

02:02:56.600 Talk soon.

02:02:57.940 Thank you for listening to this week's episode of The Drive.

02:03:01.040 Head over to peteratiyamd.com forward slash show notes if you want to dig deeper into

02:03:08.060 this episode.

02:03:08.680 You can also find me on YouTube, Instagram, and Twitter, all with the handle Peter AtiyahMD.

02:03:14.780 You can also leave us review on Apple Podcasts or whatever podcast player you use.

02:03:20.340 This podcast is for general informational purposes only and does not constitute the

02:03:24.540 practice of medicine, nursing, or other professional healthcare services, including the giving

02:03:28.860 of medical advice.

02:03:30.500 No doctor-patient relationship is formed.

02:03:32.920 The use of this information and the materials linked to this podcast is at the user's own

02:03:38.300 risk. The content on this podcast is not intended to be a substitute for professional medical advice,

02:03:43.720 diagnosis, or treatment. Users should not disregard or delay in obtaining medical advice from any

02:03:49.200 medical condition they have, and they should seek the assistance of their healthcare professionals

02:03:53.680 for any such conditions. Finally, I take all conflicts of interest very seriously. For all

02:03:59.460 of my disclosures and the companies I invest in or advise, please visit peteratiamd.com forward

02:04:06.040 slash about where I keep an up-to-date and active list of all disclosures.

The Peter Attia Drive - October 19, 2026

#412 ‒ The evolving science of cancer： emerging approaches to prevention and early detection, and the advances reshaping treatment ｜ Siddhartha Mukherjee, M.D., Ph.D.

Episode Stats

Transcript