The Peter Attia Drive - #62 - Keith Flaherty, M.D.： Deep dive into cancer—History of oncology, novel approaches to treatment, and the exciting and hopeful future

00:00:00.000 Hey everyone, welcome to the Peter Atiyah drive. I'm your host, Peter Atiyah. The drive

00:00:10.880 is a result of my hunger for optimizing performance, health, longevity, critical thinking, along

00:00:15.940 with a few other obsessions along the way. I've spent the last several years working

00:00:19.660 with some of the most successful top performing individuals in the world. And this podcast

00:00:23.620 is my attempt to synthesize what I've learned along the way to help you live a higher quality,

00:00:28.360 more fulfilling life. If you enjoy this podcast, you can find more information on today's episode

00:00:33.000 and other topics at peteratiyahmd.com. Hey everybody, welcome to this week's episode

00:00:43.360 of the drive. I'd like to take a couple of minutes to talk about why we don't run ads on this podcast

00:00:48.580 and why instead we've chosen to rely entirely on listener support. If you're listening to this,

00:00:53.820 you probably already know, but the two things I care most about professionally are how to live

00:00:59.060 longer and how to live better. I have a complete fascination and obsession with this topic. I

00:01:04.760 practice it professionally and I've seen firsthand how access to information is basically all people

00:01:09.920 need to make better decisions and improve the quality of their lives. Curating and sharing this

00:01:15.100 knowledge is not easy. And even before starting the podcast, that became clear to me. The sheer volume

00:01:20.380 of material published in this space is overwhelming. I'm fortunate to have a great team that helps me

00:01:25.800 continue learning and sharing this information with you. To take one example, our show notes are in a

00:01:31.600 league of their own. In fact, we now have a full-time person that is dedicated to producing those

00:01:36.260 and the feedback has mirrored this. So all of this raises a natural question. How will we continue

00:01:42.440 to fund the work necessary to support this? As you probably know, the tried and true way to do this

00:01:47.980 is to sell ads. But after a lot of contemplation, that model just doesn't feel right to me for a few

00:01:54.200 reasons. Now, the first and most important of these is trust. I'm not sure how you could trust me if I'm

00:02:00.400 telling you about something when you know I'm being paid by the company that makes it to tell you about

00:02:05.080 it. Another reason selling ads doesn't feel right to me is because I just know myself. I have a really

00:02:11.280 hard time advocating for something that I'm not absolutely nuts for. So if I don't feel that way

00:02:16.460 about something, I don't know how I can talk about it enthusiastically. So instead of selling ads,

00:02:21.580 I've chosen to do what a handful of others have proved can work over time. And that is to create

00:02:27.420 a subscriber support model for my audience. This keeps my relationship with you both simple and

00:02:33.480 honest. If you value what I'm doing, you can become a member and support us at whatever level

00:02:39.460 works for you. In exchange, you'll get the benefits above and beyond what's available for free.

00:02:44.300 It's that simple. It's my goal to ensure that no matter what level you choose to support us at,

00:02:50.040 you will get back more than you give. So for example, members will receive full access to the

00:02:57.520 exclusive show notes, including other things that we plan to build upon, such as the downloadable

00:03:04.240 transcripts for each episode. These are useful beyond just the podcast, especially given the technical

00:03:09.180 nature of many of our shows. Members also get exclusive access to listen to and participate

00:03:15.940 in the regular ask me anything episodes. That means asking questions directly into the AMA portal

00:03:22.660 and also getting to hear these podcasts when they come out. Lastly, and this is something I'm really

00:03:28.000 excited about. I want my supporters to get the best deals possible on the products that I love.

00:03:32.800 And as I said, we're not taking ad dollars from anyone, but instead, what I'd like to do is work

00:03:37.360 with companies who make the products that I already love and would already talk about for free and have

00:03:43.220 them pass savings on to you. Again, the podcast will remain free to all, but my hope is that many of

00:03:51.000 you will find enough value in one, the podcast itself, and two, the additional content exclusive

00:03:57.880 for members to support us at a level that makes sense for you. I want to thank you for taking a moment

00:04:02.960 to listen to this. If you learn from and find value in the content I produce, please consider

00:04:08.480 supporting us directly by signing up for a monthly subscription. My guest this week is Dr. Keith

00:04:13.940 Flaherty. Keith is a physician scientist at Massachusetts General Hospital in Boston,

00:04:17.980 where he's the director of clinical research, as well as targeted therapies. His research focuses

00:04:23.000 on the understanding of targeted therapies in cancer. And if that term is a bit nebulous to you,

00:04:27.600 don't worry about it. We define it quite clearly. And he focuses on the development of responses and

00:04:32.460 predictive biomarkers to define the mechanisms of action and resistance of novel therapies.

00:04:37.060 He's researched a lot of stuff in immunotherapy. So this is really the first podcast that I discuss

00:04:43.220 immunotherapy in, which for me is super exciting because I've been looking forward to discussing this

00:04:47.740 topic for quite some time. And you'll see why when we get into it, because immunobased therapies

00:04:53.220 are basically the most exciting recent development in cancer therapy. And we talk quite a bit about

00:04:59.120 these checkpoint inhibitors for which actually the Nobel prize in medicine and physiology was awarded

00:05:04.020 last year. He's been a PI in a too numerous account first in human clinical trials using novel

00:05:09.580 therapies. What else can I say? He's a professor of medicine at Harvard medical school and serves as

00:05:14.560 editor in chief of the journal clinical cancer research. Overall, Keith is really a wealth of knowledge

00:05:19.880 in cancer. We talk about a bunch of stuff. Now, honestly, the first 20 minutes, we're just talking

00:05:25.960 general life medicine. We don't even touch on cancer. So if you're short on time and you really just want

00:05:33.500 to get into the stuff on cancer, definitely jump ahead to 20 minutes into this stuff. We talk quite a bit

00:05:40.460 about the history of chemotherapy, what its successes were, why they asymptoted, same with radiation

00:05:48.520 therapy, surgical therapy, and ultimately what took place and what didn't take place, maybe more to the

00:05:55.140 point in the period of time between when the war on cancer was declared in 1974 and the sequencing of

00:06:00.880 the entire human genome about 25 years later. And then we talk about what took place in the two

00:06:05.140 decades since that time. And you'll see that, well, at least even for me, I think you'll tell, I was

00:06:10.180 actually learning quite a lot here vis-a-vis how some of those changes took place. And again, for that

00:06:15.160 reason, this was a highly enjoyable experience. Even if not one person listens to this podcast, I got a ton

00:06:19.880 out of it. We basically got into this notion of what's a different approach to cancer. And again,

00:06:26.020 clinically, I found this very helpful because this is a problem I think a lot about. So I do think

00:06:30.480 people will enjoy this. We talk a little bit about liquid biopsies, even touch on roles of potentially

00:06:36.180 CRISPR and overhyped with respect to cancer therapies, potentially even at the end, talk about stem cell

00:06:43.360 therapy, vitamin D, melanoma, sun exposure, the list goes on. This is a pretty long episode. So

00:06:49.940 hopefully we talk slow enough that you can listen to it at a slightly higher speed. And the show notes,

00:06:55.100 as always, will contain a ton of information, not just links to the studies we talk about, but a lot

00:07:00.420 of the semantics. I wouldn't be discouraged if you find this topic and particularly this episode to be

00:07:06.820 somewhat challenging based on how technical it gets at times. I think we both do a pretty good job

00:07:12.140 of remembering that we're not just talking with each other. And I try to ask questions to bring

00:07:17.080 us back 30,000 feet and focus on the big stuff. But in many ways, I think this will be probably the

00:07:22.840 deepest podcast I've done to date on cancer, though certainly not the last. So without further delay,

00:07:27.980 please enjoy my conversation with Dr. Keith Flaherty.

00:07:33.740 Keith, thank you so much for letting me intrude in the middle of your living room on a rainy Monday

00:07:40.040 afternoon. My pleasure, Peter. Is this normal Boston April weather?

00:07:43.660 This is the wet season, mud season as they call it, but it's a welcome break from winter. Hasn't

00:07:49.240 been a particularly harsh one, but yeah, it's a good transition time. It was brown last week,

00:07:54.600 so everything's just finally waking up. Now you've pretty much spent your whole life

00:07:59.840 from basically Connecticut to, or actually Connecticut south of Boston, right? So

00:08:05.720 sort of basically Boston to Baltimore has been most of your life.

00:08:09.520 Yeah.

00:08:09.540 That's right. Born and raised in Baltimore, broke away to boarding school in Massachusetts for four

00:08:14.480 years, Connecticut for college, four years, then back to Baltimore, four years in medical school,

00:08:19.420 Boston for the first time, three years of residency, and then Philadelphia for nine years,

00:08:24.940 which was both medical oncology fellowship and my first faculty stint. And now 10 years ago,

00:08:30.840 moved back up to Boston to Mass General. So this fall will be 10 years, but as you say,

00:08:35.960 Northeast quarter through and through.

00:08:37.460 Yeah. Well, you get your seasons, right?

00:08:39.600 Yeah. That's always my defense of the Northeast to Californians is I need the seasons. And certainly

00:08:44.940 during my educational years, I always thought if I had San Francisco, LA, or God forbid, San Diego,

00:08:50.580 weather, I'm not sure I really would have been able to keep my nose to the grindstone. Seasons are good

00:08:55.140 to kind of force you inside and take your breaks when you need them.

00:08:59.020 Well, I think there's something else about it that I sort of lament. Both my wife grew up in

00:09:02.460 Baltimore. I grew up in Toronto. And there is an intestinal fortitude that comes from being in a

00:09:06.380 climate that is not particularly hospitable. You get a little tougher as a kid, I think,

00:09:11.740 when it's, you have to pay attention. Like if you forget your gloves, like you're hosed.

00:09:15.880 Yeah. You're going to pay for it.

00:09:16.720 Yeah.

00:09:16.980 I suppose that's true. I think resilience comes from a few parts of one's upbringing, but

00:09:21.480 climate is probably, you're right. It's at least a bias of mine that it's a good thing to have.

00:09:26.280 Even as the years keep passing and I think about, well, where would I want to spend the next 10

00:09:30.880 years, 20 years beyond? I love it up here. I mean, for me, it's really the, it gets just brutal

00:09:36.220 enough in the winter, but not Canadian brutal.

00:09:38.520 Although Boston is its own little, Boston and Toronto are not that different.

00:09:42.460 Oh, okay.

00:09:42.940 Yeah. Montreal might be worse.

00:09:44.460 Yeah. I've just had more figuring latitude drives harshness of winter. I have more respect for

00:09:49.220 Canadian winters, but maybe it isn't that different. I feel like, again, just gets harsh enough.

00:09:53.200 And then in the summertime, get the payoff for the flip side, which is just so pleasant,

00:09:57.300 but not like Baltimore burdensome in terms of the heat and humidity.

00:10:00.500 So you're kind of one of these guys who was exposed to medicine throughout, right? But I know

00:10:04.880 your father is a physician. Your mother was as well, right?

00:10:06.880 Yeah, that's right. My father was an academic cardiologist for 25 years and then almost an

00:10:12.040 equal stint in industry, pharma and biotech. And my mom was an academic psychiatrist, actually

00:10:16.920 really still is at the age of now 70, will be 76 end of this month. Very different career in

00:10:24.080 psychiatry versus cardiology, but an academic career nonetheless. So a bit of research, a lot of

00:10:29.740 teaching, mentoring, and that bit continues. She has incredible stamina for staying connected with

00:10:34.700 her field. She's retired from patient care a good long time ago, but couldn't give up the rest. And so

00:10:39.980 continues to haul herself off to conferences and oral examinations that are still required for,

00:10:45.380 for psychiatric boarding and so on. So kind of a inspiration in terms of longevity in the field.

00:10:51.280 You knew pretty early that you wanted to go into medicine or was that not even clear when you went

00:10:54.800 off to college? Yeah, no, it wasn't clear to me at all. I mean, I honestly would have to admit that I

00:11:00.260 didn't really understand how my parents' worlds turned. They were perfectly transparent and happy to

00:11:05.280 talk about their careers. I just didn't know to ask the questions. The two older brothers who then and

00:11:10.340 now had no interest in science or medicine. I guess all that I could say from a young age,

00:11:17.740 maybe by the time I was in boarding school and going to college, was that some way, somehow I

00:11:24.000 wanted to help people. This is a phrase that I recall saying, but I had no idea what that might mean

00:11:29.640 and what the different versions of it were. As years passed, I got more and more of the sense that

00:11:34.980 a lot of how the world turned was kind of transactional business transactions, I suppose,

00:11:40.680 people making careers on some version of transaction. I had this aversion, or developing

00:11:45.380 this aversion to that idea that I didn't want to be involved in what I broadly construed to be

00:11:50.820 business. Now, we'll come back to this later. I was about to say the irony, but okay.

00:11:54.860 Absolutely. I'm laying that right out there for you. There's so many ways in which I turned on that.

00:11:59.900 I mean, in a good way, I've turned on those initial principles. But yeah, in any case,

00:12:04.680 I thought this idea that I needed to find a path where I felt like I was very directly helping people

00:12:09.960 was going to be the most sort of satisfying career and something that would get me out of

00:12:13.980 bed in the morning without having to really try. That part, I'd say, really did work out.

00:12:18.540 So I think that's probably the one piece that I think I did get from my parents. So I didn't know

00:12:24.320 exactly what they did when they got to the hospital.

00:12:26.900 But you had some sense that it involved people and they were in probably a more direct way of

00:12:32.160 helping them. I mean, you could argue most people are helping people in some way, but you knew that

00:12:35.760 there was fewer degrees of separation. That's right. And then that final point that they were

00:12:40.320 super motivated by their careers, it was never a complaint. There was one thing about their marriage

00:12:44.860 they would talk about just inside baseball of division department, hospital level politics and

00:12:51.740 issues that were kind of headwind for them. And you hear them talk about their real mechanistic

00:12:59.360 elements of their careers. And it just was never a complaint about it. They were very kind of focused

00:13:04.480 on results and group dynamic and so on. But I guess my point is that you never had a notion that they

00:13:10.780 did not love what they did. And end of every vacation, end of every weekend, I mean, they were just

00:13:15.700 like hard charging back at work. My parents, because they were both building academic careers

00:13:20.220 when we were young, they were oftentimes alternating evening duty. One of them, even for a whole week

00:13:25.880 at a time, tied up either on service or something that was keeping them in the hospital. So watching

00:13:30.780 that conviction and how much they just loved doing that thing that I didn't understand, that was

00:13:37.020 probably the very nebulous notion that I kept with me. So that as I kept kind of crossing off like

00:13:43.640 vast segments of American economy had to offer in terms of careers, medicine's kind of the thing

00:13:48.940 that remained. Yeah. It's an interesting point you raise about the fine line for parents in terms of

00:13:54.680 the example setting versus the time there. I mean, both are really important, right? I mean, I know

00:13:59.080 this now. I have kids, met your daughter earlier. There's in some ways no substitute for this term

00:14:03.620 quality time sort of always struck me as a little bit of a hoity-toity term. I mean, there's time and

00:14:07.740 there's no time, right? But call it quality time. And there's no substitute for that, but there's

00:14:11.960 something to be said when a child sees their mom or dad feeling incredibly passionate about what

00:14:17.360 they're doing and how without saying a word about that, it sets an exam. Yeah. I hear my daughters

00:14:23.940 talk about this in relation to my wife. She's an internist at Mass General and she's part-time,

00:14:29.600 but because of Epic, the electronic medical record, she's full-time and then some because for every

00:14:34.600 minute she takes care of a patient directly, she's taking care of their medical record for three to 1.00

00:14:38.480 five minutes easily. That sounds like an excellent ratio. Yeah. And of course the kids see that side,

00:14:43.400 right? Because they're not with her in the hospital, but they see the constant overflow.

00:14:47.420 They know what she does in principle and internists, I guess, maps to a pediatrician. They have direct 0.92

00:14:53.420 experience with that, but they watch her pour herself into the indirect care of patients through the 0.99

00:15:00.000 electronic medical record and constantly reflect that she's the hardest working person they know.

00:15:04.220 And that example, like this, that kind of willingness to like give, give, give outside of

00:15:10.120 one's personal time, family time, it's related to what I was trying to put words to, but this very

00:15:16.520 nebulous notion of what it is that qualified as a satisfying career. I've spent my adult life,

00:15:22.300 particularly raising kids, making this comment that in my imagination, a number that I often comes back

00:15:29.340 to is like 95% of people don't like what they do. And they're just holding their breath, trying to get

00:15:33.760 to the weekend or vacations. And maybe I'm vastly off base, but I guess I'm partly using a high number

00:15:39.340 like that because I just feel incredibly fortunate to have been brought up in a household and educational

00:15:45.840 environment serially that my parents made available to me that allowed me to be in that, what I think is

00:15:51.040 a relatively small group where I just always loved what I did and just, and felt like I had absolute

00:15:57.140 choice and could engineer my schedule in every respect in a totally sort of suicidal, not respectful

00:16:06.120 of balance way. I mean, I just developed a career that was ridiculously off the rails in terms of

00:16:11.340 being overcommitted and trying to juggle way too many things in some respects. For me, I would defend

00:16:17.580 it also and say it was exactly the optimal amount of chaos that one needs to really feel like you're

00:16:24.300 kind of pushing on all fronts simultaneously. But, but how do you both find the, the outlet and

00:16:29.740 define the scope of a career that allows you to do that? Man, what a challenge and what a different

00:16:33.880 challenge I think that our children will have versus what it was like to try to build a career

00:16:38.700 in this case in medicine starting decades ago versus what they're looking at.

00:16:43.120 And medicine still may be one of the slightly easier places to do it because the path to quote

00:16:48.760 unquote academic success, non-academic clinical success, it's still relatively unperturbed from

00:16:54.800 so many other paths, right? Outside of medicine. Or is that not even true anymore?

00:16:58.200 I don't see it that way, but I'd be interested in your perspective as a surgeon, because I think

00:17:01.920 maybe we have a different angle on that being in medicine versus surgery, which obviously culturally

00:17:06.480 we've always had some real differences, not just culturally, sort of skill sets. So the thing that

00:17:11.860 I've been deliberating on, let's say this, I've had a 20 year oncology career, almost 20 years,

00:17:16.700 a few months away from that. And so I divide my career up into these two decades and I've watched

00:17:21.960 technology advance and the field-wide body of knowledge accelerate in a 2000 to 2010 interval

00:17:29.160 and then these past 10 years. And this is true in so many areas of technology advance, but in

00:17:34.180 biomedical research and in cancer, which is my only area of any expertise, I thought 2000, 2010 was

00:17:40.980 mind-blowing and I thought we were going to spend time catching up with the mind-blowing advances.

00:17:45.820 That was my talking point coming into the field in 2000, that I thought we've got this huge wave

00:17:51.180 of molecular insights, mostly genetic insights in terms of cancer that had built up, but hadn't

00:17:56.140 been transformed into medicine. That was my unbelievably naive, simple-minded pitch as I

00:18:00.140 was entering the field. And I said, I want to make myself useful in that translation of science

00:18:04.180 to medicine. Anyway, 2000, 2010, then you could say we're the first chapter of translating molecular

00:18:09.420 insights into medicine and oncology. Well, 2010 to 2020 makes that first decade look unbelievably

00:18:16.140 slow. And where it was every two years, there was maybe a monumental event, whereas now it's like six

00:18:23.160 at a time. So that's in terms of like crossing the finish line, but in terms of data generation,

00:18:28.400 ability to produce high-dimensional data, analyze it, try to make sense of it, raise hypotheses,

00:18:32.900 test hypotheses. Yeah, the cycle time is changing a lot. And the constituents that we now need to

00:18:37.840 interact with are totally different. The point I'm coming to, to answer your question, is I see

00:18:42.680 medicine now as this terrifying arena in which to try to assemble a multi-decade long career. Like,

00:18:49.120 how do you keep yourself relevant in a, let's say, 40-year arc? Here I am at 20 years and feel like I'm

00:18:55.800 an absolute dinosaur. Mentoring is the thing that I think I can still do with some relevance to the

00:19:01.640 younger generation. And in mentoring them, I tell them, look, you need to learn how to talk to a

00:19:07.120 bioinformatician, computational biologist. I couldn't have even imagined 10 years ago that I'd

00:19:13.600 be telling my mentees that, much less that I would have this notion that I need to understand

00:19:18.020 the world from that degree of mathematical modeling complexity. But it's having now mentees

00:19:24.460 who were in computational biology myself, I've come to realize that it's, I didn't have the skill set at

00:19:30.320 the dawn of my career. Is it possible for people, broadly speaking, to actually retrain themselves and

00:19:36.800 grow a new lobe of their brain? It's possible, but it's going to force a totally different

00:19:41.700 paradigm in terms of how one thinks about taking breaks, taking sabbaticals, gaining that knowledge

00:19:47.440 base. So I was actually thinking of something different, but what you're bringing up is more

00:19:50.760 interesting. So I want to double click on that before we jump into the meat of what we're going to

00:19:54.960 talk about. Based on the way you're describing it, Keith, I would say doctors coming out of training

00:19:59.620 are hosed because medical school, and I'm going to really make a lot of enemies by saying this,

00:20:04.240 but I guess it's my podcast I'm allowed to say this. I think medical school might be one of the

00:20:07.920 most anti-intellectual forms of higher education that exists. I mean, when you contrast it with even

00:20:14.220 the experience that many people have in college, where you really get to think creatively, you really

00:20:18.980 get to problem solve, you really get to explore the limits of what is known and what is not known and ask

00:20:24.180 questions, not having gone to law school, but my brother did. I got to see my brother do that in

00:20:29.080 law school, having more friends than I can count who did PhDs in everything from the humanities to

00:20:34.700 the sciences. They got to do that. Well, I went to a good medical school, but that's not the way my

00:20:38.560 education was. And I'm not sure it's that much more like it today. In other words, I'm not sure

00:20:42.360 medical school teaches you how to be a thinker or a problem solver, or even, and maybe again, I hope

00:20:48.600 I'm wrong. But if you don't learn how to learn and God forbid the people who go into medical school

00:20:54.920 as pre-meds. So then they've missed out on at least taking an engineering degree or taking a

00:20:59.520 humanities degree where you would have got some of that stuff. So, so yeah, then I think you're in

00:21:03.560 real trouble. I totally agree with you. And the word that you didn't use that I'll use is

00:21:07.920 investigation. That's not taught. And at best it's taught in a retrospective field-wide

00:21:15.280 introspective way. And by that, I mean like diagnostics. I mean, thinking about how to

00:21:19.280 diagnose illness in medical school is the beginnings of teaching that concept.

00:21:24.620 I actually tell people the one type of statistics that physicians get very well trained in without

00:21:29.860 explicitly being told what they're doing is Bayesian statistic, because that really is

00:21:34.120 clinical medicine. Clinical medicine is pure Bayesian statistics. It is learning how to update your

00:21:40.840 pre-test probability with new information over and over again. And the reality of it is that doctors get

00:21:44.660 very good at that. It's never codified and formalized such that they understand that that's

00:21:48.660 actually what's being taught. Yeah. Although contaminated by bias as in recency bias, right?

00:21:52.640 So this is, I mean, there's a huge issue in the practice of medicine that you're so clouded by the

00:21:57.240 last few outcomes. Your participability is not as accurate as it could be if you were able to

00:22:01.680 eliminate it. Going back to that point that medical school, it teaches information as fact. It teaches

00:22:08.080 the known, that known unknown divide is exactly, it's my pitch in academic medicine, at least talking to

00:22:14.340 people about cancer as a career. I tell them, look, it's so simple to find the boundary between what's

00:22:20.340 known and not known. There's so much out there we don't know. And in any area of trying to understand

00:22:24.680 physiology, pathophysiology in the mode of cancer, you will so quickly get to that frontier. And then

00:22:30.720 how do you operate in that frontier? Like once you're there, how do you help the patient in front

00:22:35.160 of you, the next wave of patients that are coming in the not too distant future, and then like a full

00:22:39.280 generation out, how do you help across that gradient? Well, investigation is the only way you can pose

00:22:45.280 yourself as being helpful. And that, I didn't get that in medical school, sadly, even though I was at a

00:22:50.680 fantastic environment for learning the archival version of medicine.

00:22:54.720 That's a great way to distinguish, right? I mean, you go to a place like Hopkins, you're going to get the

00:22:58.340 best archival education imaginable. You're walking through the halls where the actual profession came to

00:23:05.000 North America. And I actually, it was one of the things I cherished when I was there. And the few

00:23:09.840 books I brought with me out of medical school, surrounding myself with books that are not

00:23:14.120 medicine as a general sport, were the history of medicine. I mean, Osler particularly, I thought was

00:23:19.680 absolutely an Aristotelian figure. So I thought those are the roots and we have to figure out how to

00:23:25.120 reimagine ourselves like those great builders of the medical discipline. So it was very retrospective.

00:23:30.720 And it wasn't really honoring the notion that everybody in the field, particularly these big

00:23:36.320 academic centers that are given tons of resources to investigate, can and should see themselves as

00:23:41.120 being absolutely on the frontier. It's a more dynamic frontier now than ever before, mostly because

00:23:45.940 the sensing technology, to be able to actually understand what's wrong in the system. And when

00:23:50.980 you perturb the system with a therapy, how the system responds, cancer is such a great example of this.

00:23:55.320 Our sensing technologies have just gone berserk in terms of the acceleration. So there's just so much

00:24:02.000 more that one can learn. But again, this is the concept that only, even in my oncology training,

00:24:07.180 I didn't get that. My oncology training was the same notion of learning about how we got here over

00:24:11.900 decades. Well, my talking point going on oncology was that I was right at this end of the conventional

00:24:17.720 chemotherapy era, as I was calling it, and the beginning of the so-called molecularly targeted therapy

00:24:23.040 era. But it hadn't happened yet. And so I wasn't interested in learning about the past.

00:24:27.260 So in other words, you could see that the futility of chemo, which basically had made minimal progress

00:24:31.680 over 40 years.

00:24:32.480 Would have reached an asymptote, right? So you basically, if you blindly develop drugs against

00:24:37.200 cancer cells to find out what's poisonous to them, and then you filter those in mice to see what kills

00:24:42.840 the mice at equal concentrations versus what doesn't quite kill them, and you call that your possible

00:24:48.360 therapeutic, what will you get out of that? This is one of my first lectures that I used to give to

00:24:52.540 medical students about cancer therapeutics as I was trying to outline a path forward,

00:24:57.300 is that basically what you get out of that is all these agents that tangled the DNA

00:25:00.000 and some microtubules stabilizing and destabilizing drugs. And the blind approach only gave those types

00:25:06.280 of therapies. Those types of therapies then thrown at cancer cure a good fraction of testicular cancer,

00:25:12.440 a respectable fraction of lymphomas and leukemias, and cure almost nothing else.

00:25:16.220 Let's pause there for a moment because we're going to just dive right into cancer. But in the spirit of

00:25:20.480 being true to the history, let's just reproduce what you said. So let's take a step back. Many

00:25:25.640 people listening to this certainly hear the terms chemotherapy, radiation therapy. It might be a bit

00:25:31.760 of a fog beyond that. So what you just described was chemotherapy, and what you just described was

00:25:37.360 an arbitrage that must be true. Chemotherapy, it's not hard to kill a cancer cell. It's actually very

00:25:43.260 easy to kill a cancer cell.

00:25:44.260 Yeah, bleach does a fantastic job.

00:25:45.400 Bleach, formaldehyde, go to Home Depot. Most things on the shelf kill cancer perfectly. Problem

00:25:50.080 is they kill everything perfectly. So the arbitrage you described eloquently is it has to kill cancer

00:25:56.800 and almost kill, but not kill, not cancer. So that's a narrow subset of things. And as you described it,

00:26:04.680 they're targeting DNA for the most part, which is why they're targeting things that grow, which is why

00:26:10.180 when most people think of a patient on chemotherapy, the first thing they think about is their hair

00:26:14.360 has fallen out. They've got sores in their mouth. Their nails are brittle. And if you look a little

00:26:19.320 deeper, you'll understand that their gut is really falling apart. What's common to all those things,

00:26:23.540 right?

00:26:23.760 And their bone marrow cells.

00:26:24.640 Yeah.

00:26:24.760 Yeah. Rapidly dividing bone marrow cells. That's exactly right. So the analogy I oftentimes use

00:26:29.080 with patients about talking about chemotherapy is it's bicycle going down the road slowly,

00:26:34.340 one bike going very quickly, one going very slowly. So the cyclist in this case being the

00:26:38.900 determinant of that. And you're standing between two cars with your broomstick and you come out to

00:26:45.120 throw your broomstick in the front wheel of those two bicycles. One of them goes ass over tea kettle 0.99

00:26:50.260 and the other one looks at you and says, what are you trying to do here, punk? And it's that fast 0.99

00:26:56.220 growing cell concept. It's that filtering system of what can kill a fast growing cell. That's what

00:27:02.800 generated those hits. We could have come up with other hits potentially. It's a chemical diversity

00:27:08.040 of probes that were used were broader. Maybe other hits could have been found.

00:27:11.540 For example, we didn't really look at metabolic distinction between them beyond DNA. That Warburg's 0.63

00:27:16.840 stuff left behind, we'll come back to it later, but you're right.

00:27:20.000 How to manipulate it particularly was completely lacking. This concept of you take cancer cells,

00:27:24.400 you establish this mind-blowing ability to grow them outside of a patient and create so-called

00:27:29.220 immortalized cancer cell lines. Henrietta lacks cervical cancer being a paradigm example out of Hopkins, 0.96

00:27:33.820 but eventually dozens to hundreds to now thousands and hundreds of thousands of these cell lines.

00:27:38.980 Which ones take in plastic versus which ones don't? Well, there's definitely a fast versus

00:27:42.940 slow cell growing just by cancer type. And then within the cancer cell population, which is always

00:27:47.640 heterogeneous, you've got faster and slower growing varieties there too. And the ones that take are

00:27:53.060 these hot rod cells. And if you then use that as your filtering mechanism to ask, well, what will kill

00:27:58.800 them? What you come up with are these toxins to rapidly dividing cells. That's exactly right. So

00:28:02.960 that caps. The term chemo, when I use it, is always to describe these things. Now, important piece of

00:28:09.860 nomenclature is that people will still always say that, well, a cancer therapeutic is a chemotherapy.

00:28:14.940 Right. But no, I favor the way you think about it. I like to think of chemo as that type of chemical.

00:28:20.680 Immunotherapy is used chemicals, but they're totally different, right? Metabolic therapies, PI3K inhibitor,

00:28:25.160 I'm not really considering that chemo, but I want to put a point on what you said, because it's sad,

00:28:29.860 but it's just the reality of it. So when Nixon declares a war on cancer, it's what? It's about 72,

00:28:35.460 74. 74. Okay. If you look at 10 years survival for metastatic solid organ tumors, so I'm going to

00:28:44.360 just take out leukemia and lymphoma. And as you pointed out, we have seen success in some of those,

00:28:50.020 and we'll come back to it. From a solid organ metastatic perspective, from 74 until 10 years

00:28:57.380 ago, isn't testicular the only success? Maybe, and GIST? Yeah. Yeah. GIST was an early targeted

00:29:03.120 therapy success. So testicular, basically seminobinous testicular tumor. That's right.

00:29:06.980 You could cure with doublet chemotherapy. You could cure the large majority, greater than 90% even.

00:29:11.800 But if a woman had metastatic breast cancer in 1974, and a woman had metastatic breast cancer

00:29:16.020 40 years later. Shockingly a little difference. I mean, we had median survival extensions of maybe

00:29:20.920 six months, 12 months. Yeah. Per new therapy that might come along at best. And then you could stack

00:29:25.420 a few of those in a disease like breast cancer. So what actually kills the population of cancer,

00:29:30.600 breast, colon, prostate, and lung, and then pick up some other notables that are still refractory to

00:29:35.300 this day, like pancreatic cancer and so on, in terms of number of deaths. But the common cancers,

00:29:39.800 they are perturbed at best a little bit by these agents. So some subpopulation of those cells will

00:29:46.360 slow down and die with conventional chemotherapy, but many of them are pre-wired. They were hardy to

00:29:51.720 begin with. They got there through a hard-earned evolution under selective pressure of the immune

00:29:56.620 system and adverse metabolic environment. And they used all the tricks up the sleeve of a cell,

00:30:01.280 of a normal cell, to reprogram themselves to be able to survive in these harsh environments.

00:30:06.340 You throw in another harsh environment reagent in the form of chemo. And not surprisingly that

00:30:12.280 these things were already basically hardwired to be able to survive yet another insult. But that's

00:30:17.820 chemo. That's an interesting way you explain that. I think that that's important, I think,

00:30:21.080 for a listener to understand is a cancer cell has to overcome a heck of a lot to get to be

00:30:25.640 clinically relevant. So how many, I used to remember all the numbers and the growth rates,

00:30:29.740 but a clinically relevant mass has how many billion cells in it approximately?

00:30:33.360 Yeah. Clinically relevant that you can find on a scan.

00:30:35.580 Something that you could see, like a one by one by one centimeter cancer.

00:30:38.280 No, you're in a billion cells already.

00:30:39.780 You're in a billion cells. So just to get a billion rogue actors, you have to evade T-cells. You have

00:30:45.760 to grow your own vascular supply. You have to overcome, as you said, adverse metabolic conditions.

00:30:52.080 These things are evolutionary warriors by this point.

00:30:56.100 Yeah. I mean, I usually start, you're absolutely right.

00:30:58.340 I mean, it starts much smaller.

00:30:59.640 The filters. Yeah, no, no, but I start with the analogy. There's a random spinning of the

00:31:03.840 combination lock where mutations are being acquired over time through a variety of insults or not even

00:31:09.440 necessarily insults, just bad replication technology, if you will, where the detection of errors and the

00:31:15.620 correction of errors is not perfect. Anyway, we accumulate mutations over time, some of which we

00:31:19.940 could limit and control, some of which we can't. Combination lock is just spinning, spinning.

00:31:24.080 The combination has to be dialed in the right sequence, just like when you're opening your gym locker

00:31:27.760 or you don't get cancer. You've got to get your tumor suppressor early and in the right order

00:31:32.300 before your activated oncogene comes along. You're randomly spinning the lock. You're picking up lots

00:31:36.740 of past true mutations, not just true drivers. So these incidental mutations that happen here and

00:31:41.760 there. So as that's happening, you finally click important components of the program. But along the

00:31:47.700 way, some of those are actually mutations that are seen as foreign by the immune system. And if they're

00:31:51.700 too visible, if they're too far out there, then they're gone. So it has to be the right kind of

00:31:57.480 genetic alteration that will give the cell what it needs to be able to proliferate abnormally,

00:32:02.060 to be able to sustain a lot of DNA damage as it accumulates and not commit suicide as a consequence,

00:32:07.700 and be able to handle all the other adverse features and filters of the tumor microenvironment

00:32:11.660 that you enumerated. It has to be able to do all that, but it has to be below the radar,

00:32:15.320 as in not detectable, as overly far into the immune system. That's a powerful insight that we only

00:32:19.900 really picked up in the field within the past five years. This notion that this is not just a cancer

00:32:24.940 cell co-opting its normal cells in the microenvironment in some kind of cocooned way.

00:32:30.920 There are a few cancers that do develop in so-called sanctuary sites that aren't subject

00:32:35.200 to immune surveillance, but the vast majority are. So that is a lot to have to overcome.

00:32:41.040 So there's a chance element on the combination lock spinning, but then there's these, you have

00:32:44.620 to survive the selective pressures that are being applied both within the cell and outside the cell.

00:32:49.920 The relevance of that thought process is highly relevant to how we think about developing therapies

00:32:54.040 right now. But it's also highly relevant to understanding this asymptotic limit of chemotherapy,

00:32:58.220 like how it is that you try to poison this DNA replication process and like just sort of shred

00:33:04.000 the DNA a bit more. Radiation, the same concept.

00:33:07.220 Yeah. I mean, radiation even more limited in that it's not systemic.

00:33:10.060 Yeah.

00:33:10.240 So basically radiation became a great tool to locally control cancer, but now you've basically reached

00:33:15.880 an asymptote at the two of the three pillars. And you could argue the third pillar being surgery,

00:33:21.860 also asymptotes at your alma mater. I mean, basically when you look at one of the most

00:33:26.680 complicated surgical procedures for cancer, which is removing the head of the pancreas,

00:33:30.680 that was an operation that used to carry a 50% 30 day mortality. Meaning the surgery was,

00:33:36.020 you had a 50% chance of living through the surgery and the post-operative course. Today that's 0.1%

00:33:42.320 mortality at 30 days, but long-term survival is still abysmal. So cutting tumors out matters a lot,

00:33:48.840 especially for colon cancer, I would say is the poster child for where cutting cancer out

00:33:53.820 really makes the biggest difference.

00:33:55.640 Even in metastatic disease.

00:33:56.880 Yeah. Yeah. It's a huge benefit. Other cancers, frankly, I'm still not even convinced,

00:34:01.320 not to say we shouldn't be cutting them out. We absolutely should, but it's not clear to me,

00:34:05.660 for example, when you talk about breast cancer, if the dye is cast long before the mastectomy is done

00:34:09.840 and that local control, I don't know anymore.

00:34:12.120 Yeah. We're learning so much more. Again, as our sensing technologies,

00:34:14.900 you can continue to elaborate because this issue of how many patients already have

00:34:18.440 metastatic disease at the time of their initial surgery.

00:34:20.740 And you take 10 women who all have one by one by one centimeter breast lesions that are all going 0.99

00:34:26.280 to be, let's assume, make them all the same stage. Some of them are going to make it,

00:34:30.060 some of them are not. And it's not clear to me how much of that we could have, or Monday might know

00:34:35.800 at the time of surgery and alter our course.

00:34:37.780 Not all hope is lost. I mean, remember the cure rate from all cancers, solid tumors that are

00:34:43.820 amenable to surgery is in my career has never been less than 40% and is probably more like 50%.

00:34:49.840 As we develop more diagnostic technology, it allows us to understand that there's circulating

00:34:54.340 cells in the blood. Those have a hard time surviving, right? People think about a circulating

00:34:58.320 cancer cell and assume that it's going to succeed. No, no, it's super hard to launch a cancer cell

00:35:03.280 into the bloodstream and actually have it find soil, get out of the bloodstream and find fertile

00:35:07.440 soil. That's a whole different problem. Lung cells aren't supposed to live in the liver. 0.97

00:35:11.400 It's not a happy environment for them. The growth factors there aren't. Their native ones,

00:35:15.080 super, super challenging bit of biology to overcome. But anyway, the point being that there

00:35:19.620 are a large number of patients who have curative outcomes from surgery who had seeding of distant

00:35:26.060 organs, even successful microscopic deposits, at least in animal model data, you'd say that that

00:35:31.960 can happen. And yet they still outlive the diagnosis, which is to say at least it lived decades

00:35:36.600 thereafter and never have a metastatic occurrence.

00:35:39.060 You alluded to this earlier. What do you think it is about the removing the mothership

00:35:43.420 that harms the satellite ships?

00:35:45.900 Well, I think they are fragile, the micrometastases. I don't know how much they're being fed. That's

00:35:50.480 not so much, say we have the evidence to support that, but they have a significant task ahead of

00:35:54.800 them to really take and survive in this so-called dormant state and then ultimately awake out of

00:36:00.900 that dormant state. So what I'm getting at is the idea that there's a larger fraction than we would

00:36:05.800 have thought decades ago when surgery was curing patients, but we just assume that they got it

00:36:11.040 early before it had ever traveled. More and more evidence suggests, no, you didn't get it that early,

00:36:16.620 but you got it early enough because those micrometastases that were established, micrometastases,

00:36:21.580 they didn't have all that it takes. In another wave, more evolution was going to be needed in the

00:36:27.140 primary tumor to be able to launch now micrometastases that had all the skills to be able

00:36:32.240 to both set up shop in the first place, live in a dormant state for not just months, but years and

00:36:37.860 even decades in some cases, and then make their way out again. So there's those additional challenges

00:36:42.740 that cancer cells have to overcome or tricks that they have to pick up that clearly just become

00:36:49.180 incrementally more likely the more time and evolution gets to happen in the primary tumor. So I think you

00:36:54.260 just stop the clock in terms of that evolution when you come along and either a patient in their bodies

00:37:00.140 tell them that they have a tumor or it's an only found or whatever the diagnostic trick is. So that's

00:37:05.560 the great hope in an area of cancer research that I watch very avidly, but being a therapeutics oriented

00:37:11.640 researcher, and that's where my mind always goes. So let's take stock of where the world was 25 years

00:37:18.580 after the war on cancer was declared. It's the late nineties. The scientific community is on the cusp

00:37:25.500 of fully sequencing the human genome. At this point, we know a number of things, right? I mean,

00:37:31.600 that's when I was in medical school. We certainly understood that cancers are initiated by genetic

00:37:37.340 mutations and everything you've alluded to with respect to genes that suppress tumors and genes that

00:37:43.480 promote tumors. The relationship there is pretty well understood. If you refresh my memory, I also

00:37:48.460 think people understood that virtually all of those mutations were somatic. Did anybody have a hope

00:37:52.740 that the human genome project was going to give us a whole bunch of germline mutations that cause

00:37:57.600 cancer? No, I don't think we believe that, right? No, that's right. You're right. The spinning of the

00:38:01.940 combination lock concept, the right sequence, Burt Vogelstein innovation in colorectal cancer

00:38:07.220 initially, that was in hand by the late nineties, that basic concept. Now, what was left to be

00:38:13.700 sorted through and a justification for doing whole genome sequencing in cancer was that we didn't know

00:38:20.420 what that sequence might look like across cancer. We didn't know what the spectrum of genetically simple

00:38:25.740 to genetically complex cancers was. And I mean that at the individual cell level, not talking

00:38:30.200 necessarily about across all cancer types. So what I'm getting at is, do you need three mutations to get

00:38:35.240 cancer? Is eight somehow better a number? By that, I guess, I mean, just look at a distribution of

00:38:39.940 quote unquote true drivers. So wait, because I sort of know the answer today, but I never actually

00:38:43.800 thought about this then. In 1999, did we know if the average breast cancer had three versus 30 versus

00:38:50.520 300 mutations? No. And most importantly, how many- How many were drivers? True drivers. I mean,

00:38:56.340 I'd say true drivers because that term driver is thrown around a bit. And true drivers, what are the

00:39:00.440 essential building blocks? And we'll come back to what essential ultimately means because I think that

00:39:05.740 has to do with therapeutic vulnerability. But any case, yeah, this idea of how many hits did you

00:39:10.580 really need? Remember that Knudsen described this, you know, the two-hit hypothesis, right? And it seems

00:39:15.440 reasonably clear that in very simple tumors, two hits actually might be enough. Now, probably then,

00:39:23.280 I think it's fair to hypothesize- Explain for the listener what two hits means in that situation,

00:39:27.940 because that's a great teaching example. Yeah. So two hits, I mean, as it was initially construed,

00:39:31.680 really was the combination of the inactivation of a tumor suppressor gene, tumor suppressor gene

00:39:36.520 being by definition, the gene, when its function is disabled, cancer becomes more likely. Nearly all

00:39:42.980 of the genetically inherited cancer types that people are familiar with come from having a inherited

00:39:49.500 alteration that partially or completely disables a tumor suppressor gene. We know very few of the

00:39:55.640 opposite, which is the activation of another gene. Oncogene could just mean cancer gene, but in the field,

00:40:01.080 most people think of oncogene as the thing that gets activated, that tumor suppressor gene is the

00:40:05.080 thing that gets inactivated. That's the simplistic version of the two-hit hypothesis.

00:40:08.320 So you're born with two copies of each gene, and one of them is not working, of a tumor suppressor

00:40:12.260 gene, and later on in life, just from a stochastic process, you only have to hit it once, not twice,

00:40:18.340 which is far less likely. And that's why these are patients that are getting cancer in childhood and

00:40:22.720 adolescence. And multiple of them over time, new unrelated cancers, sometimes in the same organ.

00:40:27.240 I mean, a patient with- Yeah, multiple colon cancers.

00:40:29.240 Exactly. They keep starting new cancers over and over again. Now, even in the most genetically

00:40:34.320 simple cases, I would just quickly assert that we have plenty of evidence now that suggests that

00:40:39.060 you get these genetic alterations that can dial the combination lock in the right way. But the

00:40:44.920 simplest genetic-defined cancers almost certainly have then a quote-unquote epigenetic, as in not

00:40:51.260 hardwired mutation alteration change as well. So a so-called state change, where they occupy a different

00:40:58.960 state of development, basically, compared to their normal cell counterparts. So if we're talking

00:41:04.040 about a pancreas cell, endocrine cell in the pancreas becoming a cancer, it moves away from

00:41:10.160 its differentiated, fully mature neighbors and does so in a way that's important in terms of its cancer

00:41:16.160 biology. It makes it more hardy. It makes it be able to survive insults. It makes it be able to adopt

00:41:20.640 programs that it's not supposed to have, like traveling, as in becoming metastatic.

00:41:24.660 More and more evidence suggests that, yes, a lot of that work is done by mutations, but then the

00:41:29.520 mutations, in some cases, actually make the cells more plastic, make them more able to read parts

00:41:37.000 of the genome and use parts of the genome that that cell type isn't normally supposed to have access to.

00:41:42.880 This is an insight that I think has only come...

00:41:45.740 Yeah, we didn't know that 20 years ago.

00:41:47.260 No, I didn't know in the late 90s. This is my argument, in retrospect, about what it is that the

00:41:53.220 Cancer Genome Atlas really uncovered for us. A criticism is that we just relearned a lot of the

00:41:58.240 same things we already knew. We already knew that RAS mutations occurred in 25% of all cancers,

00:42:02.420 and we knew then and pretty much know now that we can't drug them directly, so that's an inconvenient

00:42:07.560 reality. The point being that the catalog of activated oncogenes had actually been largely

00:42:12.720 resolved one by one by one from the mid-80s going up to the late 90s. This was the catalog I was

00:42:18.760 looking at when I was saying, I'm going to go into the field and try to figure out how to take this

00:42:23.120 information and turn it into some version of first therapy. I was a student at Hopkins when HIV was

00:42:29.380 absolutely uncountered with any therapies at all. I did a second sub-internship on the service where

00:42:36.660 the HIV patients were cared for as inpatients, usually...

00:42:39.340 That was on the Osler service, wasn't it?

00:42:41.040 Exactly, right. And AZT was in humans in clinical trials for the first time. It was talked about

00:42:45.780 actively on that service because it was definitely a research-minded service. So by the time I got

00:42:50.460 finally in through my medical training into oncology, a few more years have passed...

00:42:54.480 So wait, you nadered. I mean, the nadir, meaning the nadir of despair, the peak of hope,

00:42:58.720 would have been 96 when heart was introduced, right?

00:43:01.580 That's right.

00:43:01.900 And that's when you were starting medicine.

00:43:03.480 So I was graduating from medical school at that time.

00:43:06.040 Yeah, I mean, starting your medicine residency right as heart was introduced in 96.

00:43:09.400 And so my talking point was we need to find in cancer as many AZTs as we need as the initial

00:43:15.960 foot in the door. We know we're going to see resistance in this much more complex entity

00:43:20.080 that is a human cell that's gone rogue versus relatively simple virus. But it's when we poke

00:43:25.780 it with the AZT equivalents, we're going to learn about how it is that it tries to adapt. And that's

00:43:31.440 going to allow us to develop rational combination therapy exactly the way the HIV field does.

00:43:35.300 I still say that in 2019 and it still hasn't really happened.

00:43:39.860 I was about to say, when you were saying that a second ago, I was thinking that doesn't sound

00:43:43.460 that naive. Earlier you said you had this very naive point of view.

00:43:47.320 It was naive in the sense that with no single example, being able to drug an oncogene,

00:43:53.480 so an activated gene, nearly all of the drugs that people take are inhibitors of something

00:43:58.860 as opposed to activators. We have some activators in medicine, but not that many.

00:44:02.480 So if you think about cancer and its genetic assembly, then and now, we don't know how

00:44:07.560 to restore the function of something that's lost, particularly if it's wildly genetically

00:44:11.440 disabled, to the point of even missing from the DNA of a tumor cell in some cases. We don't 1.00

00:44:17.580 know how to replace that. This was the hope of gene therapy, the dawn of my career, that

00:44:20.980 we might be able to figure that problem out. We haven't figured it out by a long shot.

00:44:25.160 Inhibiting things that are activated, that has been where there has been success, A, in all

00:44:29.160 of medicine, and B, in oncology. So could we catalog activated events and target them with

00:44:34.520 drugs successfully and have these AZT moments? The headwind against that, and the reason why

00:44:38.960 people kind of smirked or smiled at my pitch, this is when I was applying for medical oncology

00:44:43.900 fellowships and saying these very, I think, truly simple-minded concepts, was, kid, do you

00:44:49.120 understand how complex cancer is? I mean, you poke it with a stick, it's not going to care

00:44:53.260 about blocking one thing. I mean, these things have an inordinate number of tricks up their

00:44:57.900 sleeve. Like, no way.

00:44:59.220 Right. To think that you're going to poke it once and everything will remain unperturbed

00:45:03.660 except that one thing, and then... The idea that that would actually help a patient and

00:45:07.960 that you could learn from it, both things, I think, were absolutely not accepted. You remember

00:45:12.560 the first targeted therapy success in cancer.

00:45:14.300 But the concept, the concept translationally makes sense. I think, I guess what people were

00:45:19.160 bristling against was the notion that, oh, and by the way, in the next three years, this

00:45:23.900 is what I'm going to do.

00:45:24.800 Yeah. Right. And doing this in humans, as opposed to the laboratory, wet lab, relying on reduced

00:45:29.760 model systems like immortalized cancer cell lines or some other newfangled approach of

00:45:34.760 a reduced system in the lab, that saying that doing this in humans was going to be the way

00:45:39.400 to make fastest progress, that job description hadn't been created yet. So saying that I wanted

00:45:45.480 to do it, A, and I wanted to do it in people, that's what made people for their brows, which

00:45:50.700 was fine. And that really didn't slow me down because I just, I kind of intuited that at

00:45:54.760 least this is going to be a thing to get out of bed in the morning to do.

00:45:57.680 So you did your residency at the Brigham and then your med-onc was at Penn.

00:46:02.220 At Penn.

00:46:03.000 Yeah.

00:46:03.220 Was Carl June there at the time?

00:46:04.360 He was.

00:46:04.720 So let's take a detour into immunotherapy for a moment because there's part of me that

00:46:08.020 is like wants to do this temporally because of how much we know today. And I don't want

00:46:13.060 to lose stuff that we now take for granted, but at the time was so important. So there

00:46:19.160 were a couple of cancers. Well, let's go even further than that. Based on everything

00:46:23.320 you said a moment ago, it's pretty clear that we need systemic therapy. We've reached

00:46:27.740 the limits of local therapy. So surgery and radiation work pretty darn well when they

00:46:33.260 work, but there ain't a lot of ways to make it better.

00:46:37.220 Yeah. You can minimize their adverse effects.

00:46:39.140 Exactly right. You can make them less harmful, but they're about as efficacious as they can

00:46:42.960 be. We take our first type of systemic treatment, which is chemical chemotherapy. And as we basically

00:46:48.840 saw from 25 years of 74 to 99, we basically cured one additional cancer. Somewhere along

00:46:55.080 the way, there's another idea for a systemic type of therapy. And you alluded to it earlier,

00:47:00.000 which is look, I mean, you didn't say it explicitly, but I'm just going to put words in your mouth.

00:47:04.020 Once the mutations happen to cancer, they cease to be purely self. They start to display a little

00:47:09.260 bit of non-self characteristic. And we have this branch of the immune system that is, I mean,

00:47:16.260 staggeringly effective at eradicating non-self things, namely viruses. So we don't have to go

00:47:22.340 back to Cooley's toxins, but basically if you just go into the nineties, you've got guys like

00:47:27.760 Allison who are working on things called checkpoint inhibitors, which we're going to come back to.

00:47:32.140 You've got Steve Rosenberg at the NIH, who's having limited success in melanoma and renal cell

00:47:37.360 carcinoma. You've got Carl at Penn and a number of people around the country that are starting to

00:47:43.740 show little cracks in the armor of cancer. And these results, well, I want to talk about the

00:47:48.600 durability of them, but just for a moment, explain what cancers were we seeing this in and what did

00:47:55.200 we know by the late nineties about immunotherapy? Yeah. So what you just described, one way of

00:48:00.400 rephrasing it is that you had people recognizing this idea that there's a, let me back up a step.

00:48:06.300 What had been recognized pathologically decades before was that there are some tumors that at

00:48:12.500 the time of surgery are evidently visible to the immune system because you can find a ton of immune

00:48:16.920 cells infiltrating into them. I mean, I like to go back to melanoma. Many instances is a good entry

00:48:22.620 in there because melanoma is the cancer type that I have focused on throughout my career. Fortuitous for

00:48:27.320 a couple of reasons, that choice, but in any case, it definitely useful for this discussion. So what's

00:48:31.860 useful about melanoma in this discussion? Let me just remind listeners that melanoma harbors just

00:48:37.420 about the largest number of mutations per cell of a tumor that makes it, if you will,

00:48:41.860 to succeed in becoming a cancer. It flies as high as possible beneath the radar. That's right.

00:48:47.340 There's a few other cancers out there, but it's a simple reality that they're picking up so many

00:48:50.860 mutations, these melanocytes, the precursors to melanoma from ultraviolet radiation, the vast majority

00:48:55.980 of which we think are useless to the formation of a cancer. What's the typical number of mutations

00:48:59.420 in a metastatic melanoma? It's high thousands. So you can have in the tens of thousands, but high

00:49:04.820 thousands. There's a distribution and a limit to that distribution. Aren't there only about 20,000

00:49:08.960 genes? Oh, I mean individual. Oh yeah. You'll find multiple mutations per gene in a melanoma. So

00:49:13.660 on mutations per megabase and then scale that out to the... And how many of those do we think are

00:49:17.680 playing a functional role? We don't know. Five, six is a rough estimate based on real functional

00:49:22.820 evidence now. So you have this huge onslaught of mutations, which are useless to the cancer's

00:49:29.080 purpose, probably adverse for the purpose of immune recognition. And so it's an outlier. It's

00:49:34.860 not a completely separate, but it's at the far end of the spectrum in terms of cancers that aren't

00:49:39.140 cleared and eliminated by the immune system and survive with all of this mutational abnormality

00:49:44.600 in them, a lot of non-self. So that's a cancer that at the time of initial diagnosis, a superficial

00:49:51.400 melanoma on the skin, you will find, not in all cases, but in the vast majority, a robust amount of

00:49:56.500 immune recognition. Infiltrating immune cells and T cells that are the variety that can clear a virally

00:50:01.040 infected cell and can kill a cancer cell. When they see that the internal contents being presented on

00:50:05.800 the cell surface by the so-called MHC complex have enough difference from self, then that's the cell

00:50:12.440 population that you will witness in those tumors. And then it had been described in the 60s that a

00:50:18.680 melanoma, notably other cancers by the 80s, you looked at a spectrum of melanomas. Those that had the

00:50:24.340 most robust immune cell infiltrate were going to be least likely to be life-threatening to the

00:50:28.780 patient after adjusting for other factors. In other words, was the experiment ever done

00:50:32.680 when you took a hundred patients who had a local resection, normalize them for depth, so they're all

00:50:38.940 Breslow 5, and you add up the TIL, tumor infiltrating lymphocytes, and you get a prediction of who's going

00:50:46.840 to be alive in 10 years? That's right. Powerful, very powerful, that prediction. And then that was just

00:50:51.600 replicated across the rest of cancer. That finding, colorectal cancer has its own distribution in

00:50:56.860 that regard, and some that are wildly mutated, immunogenic. Has anyone commercialized this today

00:51:02.340 as a diagnostic, at the time of therapy, diagnostic or predictive tool? To a degree. So immunoscore is a

00:51:10.200 commercial assay that's been developed that is about more complex version of immune recognition than

00:51:15.900 just these T cells. But we can unpack this more deeply now, cataloging the success, the huge

00:51:23.080 waterfall event in the immunotherapy development era when PD-1, PD-L1 interactions were discovered and

00:51:29.700 then leveraged as a therapeutic. That fraction of cancer patients who are one drug away from clearing

00:51:34.320 their tumor with a PD-1 antibody, which is the largest impact we've had with an immunotherapy of any

00:51:38.320 kind by a long shot in terms of numbers of patients helped. These are those patients. Yes, their cancer

00:51:43.900 succeeded, but they had the immune system nibbling at their heels at every step of the way. So yes,

00:51:48.440 they developed a true bona fide cancer. We just had to block this one checkpoint inhibitor. This

00:51:53.940 one break on the immune system where the foot was being expressed by the tumor cell and literally

00:51:58.460 reaching across the divide and repelling or making quiescent a T cell that had succeeded otherwise and

00:52:04.200 making it into the environment. That's a swath of cancer. Now it's interesting too when you go,

00:52:08.580 like I'm much more familiar with CTLA-4 because that was the work, what I was doing when I was doing my

00:52:12.860 time in the lab, both in medical school and later in residency. It was interesting the amount of

00:52:17.920 autoimmunity you saw as well, suggesting that boy, when you took the brakes off the immune system,

00:52:23.420 it didn't just want to kill cancer. It actually wanted to kill a bunch of things. Yeah. So this

00:52:28.040 reminds people these systems were not created purely to survey for cancer and eliminate it, right? So

00:52:33.540 the purpose of dampening effects on the immune system are that you don't just mount an immune

00:52:38.340 response and have it just take over your body, which is what a CTLA-4 blocking or transgenic

00:52:43.740 experiment will produce is lymph proliferative overdrive that kills the mouse. So the idea that

00:52:49.520 these brakes exist for a reason, I think is intuitive. If you think about the fact that basically we live

00:52:54.980 in a complex system, we're exposed to pathogens. We were talking about viruses, but we're being exposed

00:53:00.420 to microbial pathogens that are trying to infiltrate all the time. You wipe out someone's immune system

00:53:04.840 with chemotherapy or a bone marrow transplant and those bacteria that are living in the gut and,

00:53:09.860 you know. Seemingly harmless in symbiosis. Exactly. They will then infiltrate. They're

00:53:14.640 right at the margin to begin with. They'll infiltrate and they'll take over. This is a really active

00:53:18.740 border zone that's being policed all the time. You give a CTLA-4 antibody and where does the most

00:53:24.840 life-threatening chaos erupt? It's at the gut. You've unleashed this pre-existing force that's at work

00:53:31.380 all day every day in this very immunologically active environment and it starts attacking normal

00:53:37.440 colonic tissue and can perforate the colon from which patient then dies. So leading cause of death

00:53:43.180 from that therapy is rare as those events are, but it's a very powerful sign of how this sort of gas

00:53:48.960 pedal and brake component is at play. Playback the conversation to talking about chemotherapy and

00:53:54.120 this notion of therapeutic index. Well, here's the issue with unleashing the immune system systemically

00:53:59.180 with these types of therapies. Yeah. The early days of this in the 80s when Steve Rosenberg and

00:54:04.360 his colleagues were using mega doses of interleukin-2, you saw that flip side, right? Which

00:54:09.760 was the systemic inflammatory response syndrome was as life-threatening as the cancer. You were

00:54:16.280 teetering between the patient dies of what looks like sepsis versus the patient dies of cancer and you

00:54:23.100 have to thread that needle. Right. Sepsis in a bottle. You start dripping in interleukin-2 and the only

00:54:27.920 other time a human being sees that cytokine at levels like that is when the immune system says,

00:54:33.700 screw it, we've got to go all guns. We're going all nuclear. All guns blazing or else the host is

00:54:38.660 going down here. And there's a 50% chance we're going to kill ourselves in the process, but so be

00:54:42.580 it. But we, but this is our last shot and with interleukin-2, thank God you can turn it off and

00:54:46.720 within certainly hours, but even minutes in some cases like that whole storm settles down. But the

00:54:51.680 cytokine era, that was the first wave. So coming to your point about how do we get- 0.67

00:54:55.820 Yeah, there's a great proof of concept.

00:54:56.980 Yep. But narrow therapeutic index, as in the asymptote was reached awfully early. What were

00:55:02.120 the cancers that responded? Melanoma?

00:55:03.580 RCC and renal. Yeah, melanoma.

00:55:05.440 Melanoma that has the highest mutation burden. I didn't cover kidney cancer. Kidney cancer,

00:55:09.460 for reasons that we're still trying to unpack, is very immunogenic. It is highly visible to the

00:55:14.200 immune system. When you diagnose a kidney cancer at the time of surgery, the amount of infiltrating

00:55:17.480 immune cells, active immune cells, so-called cytolytic T-cells that are like churning out

00:55:22.520 the enzymes that will kill their neighboring cancer cells, that actually scores at the top.

00:55:26.600 I understand why we see it in melanoma. Why in RCC?

00:55:29.440 Not known because it's not mutation burden driven. If you look at mutation burden in relation to this

00:55:35.780 immune recognition, there's a very strong correlation across all of cancer with a couple of outliers.

00:55:41.060 And RCC is the renal cell carcinoma. It's an outlier on that curve.

00:55:44.820 Yeah, it's immunogenic without a high mutational burden.

00:55:47.360 Very precisely. And how much of this is epigenetic then, as opposed to genetic? That's the piece that

00:55:52.440 investigators are currently trying to uncover. Admittedly, renal cell carcinoma, because of

00:55:55.920 its relative rarity, doesn't draw that much attention, whereas so much more of this work

00:56:01.020 is being done in the more common cancers. But there's a story here across cancer that there's

00:56:05.420 this whole issue of what determines visibility of the immune system versus invisibility, right?

00:56:09.920 These successful cancers have to find cloaking mechanisms. They will not succeed otherwise.

00:56:14.000 One cloaking mechanism is just stop presenting antigens. Don't make MHC complexes that present

00:56:20.360 these mutated antigens. And if you go high enough up the mutation scale...

00:56:23.960 We'll go so far in this that I want people to be comfortable with the lingo. So what's MHC class

00:56:29.580 one to... What is antigen presentation? Maybe do... Pretend this is a group of first year college

00:56:35.340 kids and you've got a few minutes to explain how T cells work, basically.

00:56:39.060 So the major histocompatibility complex, MHC, is this machinery that exists. These are cell

00:56:45.600 surface proteins. Well, they become cell surface proteins. When they're first expressed endoplasmic

00:56:51.580 reticulum as all proteins, when they're initially translated into proteins, they have the opportunity

00:56:56.400 to sample internal contents, protein fragments, basically. So they're sampling all the time the

00:57:01.880 protein fragment repertoire. And MHC class one and class two, and we each have a portfolio of them.

00:57:09.560 We inherit the diversity of these ultimately from our parents, but we have a massive diversity of

00:57:15.060 them that we're born with. So class one and class two can hold different size protein fragments.

00:57:19.700 So they're fundamental difference. And there's a flexibility or a nimbleness in terms of class two,

00:57:25.220 which can hold longer peptides. It can see potentially bigger repertoire than class one.

00:57:30.060 So the analogy is your job is making sure a house is okay. There's a house party.

00:57:34.980 The internal house.

00:57:35.680 Yeah. The internal house is okay. And you're kind of the guy that got hired to help make sure the

00:57:39.820 house party doesn't get out of control. And you're roaming around the house looking to try to figure

00:57:45.320 out, do I need to take any of these guys outside to show the police?

00:57:48.300 That's right. So the immune cells that are not only activated, not only T cells,

00:57:52.460 other immune cells are interviewing normal cells constantly and effectively all organs.

00:57:58.520 So this surveillance process is happening. We think viral pathogens are what created the

00:58:02.800 evolutionary pressure to evolve this system. Cancer couldn't have been the excuse, right?

00:58:07.060 If cancer is distributed across ages 40, 60 to 80, it's not relevant because you've reproduced and

00:58:13.360 you've served your purpose. So we benefit from having this system that evolution handed to us from

00:58:18.700 viral selected pressures. But in any case, so the system was tuned for that, for picking up viral

00:58:23.460 pathogens inside of cells, being able to present them on the cell surface.

00:58:26.840 So we'll use an example, sorry, just that everyone will get. When you get a virus that

00:58:31.440 gives you a sore throat, for example, the pain you're feeling, the soreness of your throat

00:58:36.520 is the inflammation. It's the actual endothelial cells within your throat that are hurting because

00:58:42.220 a virus has gotten there. The virus has hijacked your DNA replication system. It's doing its nonsense

00:58:48.280 because that's what it does. It wants to survive. It can't make its own DNA. In the process, 0.88

00:58:52.800 new proteins are showing up inside a cell. And these little antigen presenters are saying,

00:58:58.140 I don't recognize this. I don't think this belongs in here. I'm going to take it to the surface

00:59:01.860 and let these guys that come by who are... We think the interviewing happens even with

00:59:05.960 normal proteins. So normal proteins will be shown as well as abnormal proteins, not only...

00:59:09.740 That's right. And then it's the cop who's coming by who has the ability to go,

00:59:12.900 that gun's okay, that one's okay, that one's okay. Whoa, that one, we actually need to go in the

00:59:16.300 house and rip it up. Exactly right. And it turns out that when MHC complexes are presenting antigen,

00:59:21.300 it turns out that differences even in a peptide fragment or protein fragment,

00:59:25.300 the position of the abnormality matters. And if it's in the middle, that's able to be seen

00:59:29.520 more efficiently. Some of these principles now have been reasonably well elucidated.

00:59:33.520 So it's really nuanced interview technology. So a ton of normal self that's being seen and excused.

00:59:39.220 And then there's the chance then that some of the abnormal proteins, viral pathogens,

00:59:43.480 and then mutated proteins as well can be presented and seen as non-self.

00:59:47.280 But I do want to just jump quickly to mention a point, which is that, well, if cancer is developing

00:59:53.820 all these mutations and quote, unquote, trying to become a cancer, what might be a trick that you

00:59:59.040 could use to try to evade the immune system? How about if you disable this machinery? How about

01:00:03.700 if you just don't allow MHC complexes to be made?

01:00:06.640 Right. So a cancer cell is, instead of a cell that gets infected with a virus, although notwithstanding,

01:00:11.420 that's how some cancers start. If now the cancer cell takes over the entire DNA replication system,

01:00:17.720 the smartest thing to do is say, of course, I'm going to make proteins that are foreign.

01:00:22.300 I'm just not going to get them presented on the MHC molecules outside. I'm not going to let anybody

01:00:27.600 outside of me know what's going on inside.

01:00:29.900 Right. So you might think a virus would have figured this out a long time ago to suit their 0.67

01:00:34.440 purposes. Maybe they would figure out how to wipe out MHC complex presentation. It turns out

01:00:39.000 natural killer cells don't like that very much. This is very primordial branch of the immune system

01:00:43.560 that we think about innate immunity, adaptive immunity as being primordial versus more modern,

01:00:48.380 quote unquote, higher organisms having them. Any case, these natural killer cells, which are part of

01:00:52.500 the more primordial immune surveillance machinery, if they see a cell not showing MHC complexes,

01:00:59.160 that targets that cell for destruction. So it's not okay in a fully civilized ecosystem of all cells

01:01:06.340 playing their role and being willing to be interviewed. It's not okay not to make MHC

01:01:11.120 complexes. So it's thought to be intolerable, if you will, at least in the face of natural killer

01:01:15.680 cells to do that. Otherwise, all cancers would have figured that trick out long ago. 0.97

01:01:21.220 There's this really fascinating story that's emerged in trying to understand the features of

01:01:25.620 cells that will survive the onslaught of an activated immune system after checkpoint therapy,

01:01:30.060 PD-1 or CTLA-4 therapy. Some patients will clear their tumor and they're cured. There are those who

01:01:34.500 don't clear their tumor and they're not cured. And under selective pressure of a-

01:01:38.080 Meaning they undergo a complete response, but then they relapse or they only undergo a PR

01:01:43.160 and never CR.

01:01:44.280 Yeah. Amazingly, if you look at the data with the PD-1 antibodies, which are, again,

01:01:48.620 that's the thing that's really helped humanity in terms of large numbers, I usually summarize that

01:01:53.160 to say that there's 10% of cancer patients currently are getting a heroic benefit from that therapy.

01:01:58.280 There's a higher percentage who get some benefit, like actually double the number, 20%,

01:02:02.360 who do respond. Enough tumor cells are killed by the activated immune system that the tumors will

01:02:07.180 shrink. Under that selective pressure of now heightened immunity, remember there was already

01:02:12.080 baseline immunity to a degree, variably, yes, across cancer, all cancer types, but still now this

01:02:18.500 activated immune system has just been raging. Those that survive, what do they do? Well, it turns out

01:02:24.560 they start to dial down their MHC complex expression through genetic and epigenetic means.

01:02:30.240 Just enough to satisfy the NK cell, but no more.

01:02:33.700 Precisely. Exactly. So they find this new homeostatic set point, and this is true in

01:02:37.220 metabolism and other programs in cancer, that they find new set points under selective pressure

01:02:42.620 of the onslaught, in this case of activated immune systems, but it could be other oncogene

01:02:46.480 targeted therapy does a similar thing. Actually, more and more convergence as opposed to divergence

01:02:51.300 has been emerging in the cancer therapy resistance world, where even melanoma, for example,

01:02:56.600 where we have effective oncogene targeted therapy and we have effective immunotherapy,

01:03:00.800 there's more and more evidence that actually what survives both is the same sort of phenotype,

01:03:06.080 if you will. Cancer is adopting similar programs to be able to try to survive the onslaught,

01:03:11.180 even from those two very different modalities.

01:03:13.000 Do you think it's safe to say that the lethality of cancer is directly proportional to that evasion?

01:03:18.420 For example, why is it that when you take the ratio of people diagnosed with pancreatic

01:03:24.620 adenocarcinoma to people who die of it, contrasting that with something like prostate cancer? Now,

01:03:31.100 prostate's a tricky one because it's sort of immune protected, but maybe we can pick an example

01:03:34.760 that's less...

01:03:35.700 Glioblastoma is right there with pancreas.

01:03:37.760 Although for a totally different reason, right? It doesn't really metastasize. I mean,

01:03:40.420 it's causing most of its local destruction.

01:03:42.420 And it's in a bad spot.

01:03:43.160 Yeah, yeah. It's in a space-restricted...

01:03:44.420 You're right. But it is fair to say, though, that both pancreatic cancer and glioblastoma

01:03:48.760 are really not immunogenic on the spectrum. Prostate, as you say, is out also in that end

01:03:54.520 of the spectrum. So if you just look at this baseline at the time of diagnosis, how much

01:03:58.320 immune recognition is there? How much does that relate to how long someone lives and how likely

01:04:03.040 they are actually to respond even to conventional chemotherapy? You can splay cancer out, all cancer

01:04:08.980 types, and there's heterogeneity within cancer types as well. So not all lung cancers are the same.

01:04:13.540 And you can come up with some of these that are, as you say, they are aggressive tumors.

01:04:18.340 They're not being seen adequately. So not being slowed or being nibbled at by an immune system in

01:04:23.820 their evolution. And then at the time that we diagnose them to the time of death, basically

01:04:27.840 almost nothing happening there either.

01:04:29.520 Do you teleologically have a rationale for it? I mean, we didn't come up with a good explanation

01:04:33.160 on why RCC, renal cell carcinoma, would be so immunogenic despite the relative positive

01:04:38.600 mutations. At the other end of that spectrum, what is it about a pancreatic endocrine cell

01:04:45.560 that would allow it or have it be so protected?

01:04:50.560 Yeah. This is a black box at the moment. What you're asking is the fundamental comparative

01:04:56.040 biology question in cancer. That's this decade.

01:04:59.340 Yeah, I leapfrogged us past 2010.

01:05:00.960 We are now wrestling with this opportunity to do comparative biology, not just experiments,

01:05:07.420 but analyses. I mean, what are the building blocks, genetic, epigenetic, the metabolic features?

01:05:12.700 How did each of these tumors assemble themselves to be able to accomplish the various behaviors

01:05:17.440 of cancer that make them lethal? I would say this is where we have more black boxes in diseases

01:05:23.440 like pancreatic cancer and glioblastoma anyway. The hormonally driven cancers are a unique entity,

01:05:29.100 right? So prostate cancer is almost all hormonally driven, at least at its outset. It can evolve away

01:05:33.960 from hormonal drive slash dependence during its subsequent evolution. And then a significant

01:05:39.620 portion of breast cancer, but not all, is hormonally driven. Those cancers need to be considered a little

01:05:45.020 bit separately because they're still using this lineage-dependent tissue of origin-dependent

01:05:50.980 program of using the feeder, the hormone, to help achieve its purpose, if you will, always going

01:05:57.160 back to anthropomorphic concepts in cancer. It's just how I've always thought about it.

01:06:00.580 Cancer and immunology are so geared towards that way of thinking.

01:06:03.680 Yeah, that's right. Yeah, it's immune cells, exactly. You readily put yourself in the driver's

01:06:07.300 seat, if you will, in terms of various immune cell types. Everybody likes to be a CD8-positive

01:06:11.140 T-cell, I think.

01:06:11.980 I'd rather be CT4.

01:06:13.200 A helper.

01:06:14.020 A CD4, yeah.

01:06:14.660 A helper, yeah.

01:06:15.020 I'm more of a helper cell.

01:06:15.800 Right. Any case, now to answer your question, if we had better insights into this, we would have

01:06:20.180 more hypotheses moving towards therapeutics in these cancer types. But the have and have

01:06:25.960 not spectrum in cancer is only getting wider. The advances that have been made in certain

01:06:31.400 areas where we very clearly fingerprinted the top of the pyramid vulnerability, it's not

01:06:37.100 just about mutations. I mean, it really is about other cancer programs that are at the

01:06:41.300 forefront of that cancer slash cancer type. The AZT equivalent has been worked out to dismantle 0.82

01:06:47.120 that one program. It doesn't cure everybody, but it helps patients directly, perturbs the

01:06:52.360 hell out of that cell. And I still maintain the hope and, I guess, belief even, if I'm

01:06:57.660 allowed to use that word in a scientific discussion, that secondary vulnerabilities are going to

01:07:02.540 come from being able to have enough AZT-like primary interventions or points of vulnerability.

01:07:07.920 But you said something earlier that I want to repeat because you've alluded to it twice,

01:07:11.580 I believe. It's a bit of a scary concept, right? Which is some of these mutations may have

01:07:16.380 no obvious, immediate, targetable quality, but they enable epigenetic change that itself

01:07:27.120 is the problem and much more difficult to target. Is that a fair assessment?

01:07:31.300 That's right. I dropped that thread from before. I meant to finish a thought previously that the

01:07:35.340 Cancer Genome Atlas Project, like whole genome sequencing of large numbers of cancers, what did

01:07:40.960 it teach us? Well, it retaught us what we already knew in terms of the common tumor suppressor

01:07:45.100 genes and activated oncogenes. The big discovery was how commonly you will find across cancer

01:07:51.500 driver genetic events, so like causative genetic alterations in genes whose protein product

01:07:58.700 regulates chromosomal well-being. So epigenetic regulators, as they're oftentimes thought of.

01:08:04.880 So chromosomes are complicated. They're dynamic in normal cells, and they're definitely dynamic

01:08:10.420 in cancer cells. Think of it as a folding and unfolding of the blueprints. Basically,

01:08:14.980 certain cells in the body only need to read one segment of the blueprint to be able to do their

01:08:19.220 job. Cancers, generally speaking, like to open up the blueprints. That's a fair generalization, 0.83

01:08:24.560 and you can reflect that even at just the entire chromosome level. Literally, chromatin is open and

01:08:30.060 being read more actively. It's the only way that a lung cell can figure out how to travel like a white 0.99

01:08:35.400 blood cell, ultimately, is you have to open up the blueprint and see that part of it. It's these

01:08:39.760 epigenetic regulators that are being, that now were really discovered by the Cancer Genome Atlas

01:08:44.540 Project. So going back now, I guess 10 years ago is when it really kind of these insights first began

01:08:49.200 about five years ago. That era taught us how widespread these types of genetic alterations were.

01:08:56.140 That was not known before that campaign was launched. So yes, you have these activated events,

01:09:02.100 these tumor suppressor genes that are eliminated. Many of them have to do with DNA repair mechanisms

01:09:06.740 and talk about other common tumor suppressor genes. But anyway, in the middle are these epigenetic

01:09:12.280 regulators that themselves are activated or inhibited through genetic alteration. That was just a big

01:09:18.600 aha moment in the field because it showed how essential, well, A, how cancers do it, that they

01:09:22.980 create this so-called plasticity, this ability to basically go from being a differentiated lung cell into

01:09:28.400 a less differentiated lung cell that's not now able to do things that it's normal lung cells

01:09:32.080 It gives up some of its lung superpowers in exchange for greater pliability. And yeah, it's 0.94

01:09:38.180 just, it's like a reality TV show you couldn't make up.

01:09:42.640 Yeah, it's diabolical, but it's using the whole playbook, right? I mean, I use that blueprint

01:09:46.620 analogy, but it really is, you know, every cell in the body has the entire chromosomal content in it.

01:09:53.900 It just doesn't use it. Like that discussion in neuroscience about how much your brain do you use

01:09:58.140 at any given time? Well, the normal cell isn't using so much of the blueprint to do its job.

01:10:03.140 If it's going to become a cancer, which is a cell doing many jobs all at once and to the detriment

01:10:08.840 of the host, ultimately, it really has to open up and maintain this open blueprint, not just wildly

01:10:14.400 open, but very, you know, kind of strategically and purposely. So that happens. Again, relatively recent

01:10:19.540 insight. We are just at the beginning of actually developing tools as in like chemical tools to actually

01:10:25.800 alter the function of these proteins to see, well, then what would happen? Can you actually

01:10:29.620 restrict the blueprint reading? So go back to melanoma one more time. Melanocytes derive from

01:10:35.760 the so-called neural crest. So the brain tissue and these pigment producing cells, like super weird

01:10:41.060 developmental fact, come from the same tissue type. So, and melanocytes is people probably well know

01:10:46.680 distribute generally just throughout the sun exposed skin, although there's a few stragglers here and

01:10:51.780 there, which can form melanomas. Any case, these guys are sparsely distributed in the skin, creating

01:10:56.700 a little bit of a shield of sorts, but not a very effective one. Any case, they derive from the

01:11:01.720 neural crest. That's where they come from. So what happens when you blast the hell out of a melanoma

01:11:06.980 with BRAF targeted therapy and a PD-1 antibody and you're shrinking tumors down, but not eradicating

01:11:11.960 the cells and things come back out? What do they start to look like? They look like neural crest

01:11:16.740 cells. So they take a step back in the evolutionary developmental path and in doing so, find hardiness

01:11:24.780 mechanisms that allow them to survive an activated T-cell repertoire, allow them to have their dominant 0.78

01:11:30.440 driver activated oncogene, largely disabled, and be able to manage, not only survive, but ultimately

01:11:36.820 then begin again growing and becoming life-threatening. This is a theme across cancers. I mean, this has been

01:11:42.360 seen now lung cancer and breast cancer and other tumor types where there's been substantial advance

01:11:46.900 in terms of therapies and real outcomes being achieved for patients. So these leapfrog moments

01:11:52.160 that have been happening, common, common principle is this so-called epigenetic state change thing.

01:11:57.660 And the genetic determinants of it, as I said, are really, it's within this past 10 years that we've

01:12:01.680 gotten any insights into it. When we talk about targeted therapy, it's a bit of a buzzword now. And I

01:12:06.960 think for someone like you, it's worth maybe clarifying, how do you define targeted therapy?

01:12:13.440 And what do you think is really our first great example of it?

01:12:17.200 Just to get right through the jargon and be able to keep chemo to one side and targeted therapy to

01:12:22.620 the other. What I've always said is targeted therapy is simply that we knew what we were doing

01:12:27.560 from the beginning. So we knew what we wanted. We knew what the specs were.

01:12:33.440 It's Babe Ruth actually pointing at the wall. Everybody can hit a home run.

01:12:38.880 Not just closing your eyes.

01:12:39.620 That's right. But if you can actually point to where the ball is going to go before you hit it.

01:12:43.520 So it's basically none of the conventional chemotherapy drugs were known to be DNA binding

01:12:48.660 and designed and engineered to be them. But the target product profile, if you will,

01:12:52.820 was specced out for everything that we call targeted therapy. Now, some people might take exception

01:12:57.560 with that, including the first monumental success. So we actually had two versions of targeted

01:13:02.740 therapy at work in the 90s, somewhat quietly. Epidermal growth factor receptor and HER2,

01:13:08.300 which is related in the same family, close-knit family, actually, of epidermal growth factor

01:13:12.540 receptors, plural. These two surface growth factor receptors had been kind of discovered and cataloged

01:13:19.320 in terms of their biologic function in many cancers in the 80s. And then antibody that could reach

01:13:24.400 the cell surface component of these receptors were engineered and were being investigated in clinical

01:13:29.320 trials. They weren't causing heroic effects, as had been hoped. But those were definitely targeted

01:13:34.600 therapies. They did end up moving the needle and largely in combination with conventional

01:13:39.380 chemotherapy, notably in the case of HER2-targeted therapies in breast cancer and EGFR antibodies,

01:13:44.720 at least in colorectal cancer.

01:13:46.200 And before we leave HER2-new, what's the success rate? So if a woman... HER2-new is used pretty commonly 1.00

01:13:51.520 as an adjuvant now, right?

01:13:52.560 Yeah. But the naked antibodies produced a 10% to 20% response rate. So tumor shrinkage by the

01:13:58.420 criteria that we use in clinical trials.

01:13:59.760 Yeah. So more than a 50% reduction.

01:14:01.940 Yeah. So notable benefit in a pretty small fraction of patients when given as monotherapy.

01:14:06.580 And given in patients who still had disease.

01:14:08.860 That's right. Initially, patients with overt metastatic disease, overt versus covert metastatic

01:14:13.240 disease. So that's right. So is that really an AZT-like moment to see that type of tumor shrinkage

01:14:19.800 depth slash reliability or unreliability of tumor shrinkage in that low range. So to be debated

01:14:25.980 with conventional chemotherapy, which itself had a 30, 40% likelihood of causing the same

01:14:31.140 amount of regression. And then you are now in business because you're producing 60 plus

01:14:35.320 percent response rates.

01:14:36.280 Now, these are mostly PRs, not CRs?

01:14:38.360 Correct.

01:14:38.760 PR is partial response. CR, complete response.

01:14:41.240 And in every cancer type, it's true in leukemias or hematologic blinkancies, as is true in solid

01:14:46.220 tumors. The more you can beat down the tumor, the more likely that's going to last for a

01:14:50.600 while. So putting people, quote unquote, into remission, which is a leukemia term, actually

01:14:54.980 has its solid tumor equivalent. So complete responses are the most durable. Surprise,

01:14:59.460 surprise. And amazingly, even our kind of stupid CAT scan technology, as much as we think, oh, 0.99

01:15:04.560 I can only pick up a smaller than billion, but many, many millions of cells needed to be 0.98

01:15:08.820 visible on a CAT scan. It turns out that actually that threshold, like clearing below that threshold,

01:15:12.720 a so-called complete response, actually does mean something very powerful in terms of patient's

01:15:18.100 longish term outcome, maybe not cure, but still longish term outcome, even with certain monotherapies.

01:15:23.220 So HER2 was an example where, yeah, it moved the needle to a degree in a subpopulation. 0.88

01:15:27.220 With conventional, stupid chemotherapy, it seemed to actually collaborate reasonably well and improve 0.82

01:15:33.020 survival in big phase three trials when HER2 antibody was given with chemotherapy versus 0.99

01:15:38.160 chemotherapy alone. That was a slow motion aha moment. What was the fast motion one was the,

01:15:45.040 to me, kind of validating moment, which was in BCR-able translocated chronic myelogenous leukemia.

01:15:52.520 And then the same drug amazingly worked in gastrointestinal stromal tumor. But any case,

01:15:56.400 that came a couple of years later.

01:15:57.540 So what, explain what's the tyrosine kinase? What is that whole thing that you just said? And

01:16:01.380 why does that matter?

01:16:02.240 Right. So the chronic myelogenous leukemia.

01:16:04.660 So there's four leukemias, broadly speaking, right? Acute and chronic, myeloid and lymphoid

01:16:09.320 leukemias.

01:16:09.780 The kids are more typically getting the acute ones, both lymphocytic and myeloblastic, correct?

01:16:15.260 Yeah. Turn that around. Pediatric cancers are rare, generally speaking, but it turns out-

01:16:20.240 Sorry, when a kid gets one, it's more likely-

01:16:22.260 Yeah, right, right. Bone marrow drive cancers are a common problem for kids. We think that actually

01:16:26.560 relates to the number of hits, by the way. Like if you're a kid, you can't accumulate that many hits.

01:16:30.440 And so cancers that can form with few hits are the ones reflected in the pediatric population.

01:16:35.880 So exactly as you say, those things, they're built on few hits. They're rapidly, acute leukemias,

01:16:40.540 rapidly dividing cells, and you can cure them. Kids, greater than 90% with multi-agent chemotherapy

01:16:45.440 cocktails. Yes, that make them sick, but you can cure them. And then there can be long-term

01:16:48.860 consequences. So I'm not trivializing the room for improvement there. But anyway, chemo definitely

01:16:53.660 in pediatric acute leukemia is a big deal. Chronic leukemias happen very uncommonly in kids,

01:16:59.560 but can happen. Adults more commonly get chronic leukemias, but can also get acute leukemias that

01:17:05.580 are more genetically complex than the kids' versions, notably, and therefore harder to cure with the

01:17:10.140 same exact chemotherapy regimens. So one form of chronic leukemia, so-called chronic myelogous

01:17:15.640 leukemia, so in the quadrants as you were depicting them, this is one. People could live with it for

01:17:20.020 five to seven years. If you replace their bone marrow, so-called bone marrow transplant, it could cure

01:17:23.700 a minority population. On a good day, maybe 40% of CML patients can be cured with bone marrow transplant.

01:17:29.560 But a molecular insight came in the 1970s that basically there was a very common

01:17:34.200 kind of macroscopic, if you will, chromosomal change, where one part of a chromosome would

01:17:39.460 very stereotypically, 95% likelihood, that a CML patient would have the repositioning of one

01:17:45.760 portion of a chromosome to another. Which is kind of hard to imagine when you think of how big that

01:17:51.580 is. Yeah. Right? It's a big migration.

01:17:53.700 Everything we've been talking about is this base pair. Maybe it's worth it. I think most listeners

01:17:59.520 know this, but it never hurts to be reminded. Can you walk from the scale of base pair to gene

01:18:06.380 to chromosome? Just give people a sense of, you're on a little spaceship, you get shrunk down, you are

01:18:11.640 entering the nucleus of a cell. What do you start to see as the plane's landing?

01:18:16.940 Right. So we talk about 23 pairs of chromosomes, and that really is a biologic reality. Decades and

01:18:22.340 go the practice of being able to kind of spread out the chromosomes and fix them in a way that you

01:18:26.460 could stare at them outside of intact cell is the picture that people have been shown in elementary

01:18:32.060 school. In cells, they really are. They do exist as coherent, separated entities, but much more

01:18:37.780 nebulous than how they're splayed out as 23 pairs. 23 chromosome pairs, we get one of each from each of

01:18:44.120 our parents, and they range in size. So you got one to 22 looks sort of the way they do, and then XX

01:18:49.160 or XY, round out the... Exactly. And so their size difference relates then to the amount of genetic 0.65

01:18:54.560 content in each of them. It's an important point. I usually go straight to genes, but we can come

01:18:59.140 back to base pairs. So there's 30,000 genes. That's the current number that people... I've got to update

01:19:03.960 my estimate. I've been saying 2025. Yeah, so this is... Coding versus non-coding. Coding versus non-coding.

01:19:09.160 So, and that's still, even if you include non-coding, true genes, you still have a lot of in-between

01:19:14.880 material. A ton, a ton. The vast majority of base pairs are function not determined, essentially.

01:19:21.180 Are they important scaffolds? Probably, at a minimum. They're at least scaffolds. This whole

01:19:25.540 notion about opening the blueprint, closing the blueprint, they probably play an important role.

01:19:29.140 Much of the genome probably plays some role in that opening-closing process, normal and

01:19:33.500 pathophysiological. Any case, so you think about the number of total genes that exist and the size of

01:19:39.680 them and the amount of genetic code that's in them versus the dark, quote-unquote, dark matter. At least

01:19:44.200 scaffold, maybe smarter scaffold than we give it credit for. There's relatively small portions of

01:19:50.460 it that we actually understand. Really, coding genes are what we at best understand.

01:19:54.560 Which is why when somebody does a 23andMe sequence where for $100, I mean, you can pretty much... I mean,

01:20:00.940 you're not going to get a complete sequence, but you can get a complete sequence anywhere on the street

01:20:04.320 today. Right. And so usually those tests will hone in on the component of the genetic sequence that we

01:20:09.140 know something about. So you can at least present fingerprint to someone in terms of

01:20:14.080 their inherent or their ancestry and then maybe something about disease, but not a lot of insight

01:20:19.400 there. The genes, of course, vary, individual genes amongst the tens of thousands, vary enormously in

01:20:25.180 their size and therefore the amount of genetic code in them. And the mutation opportunity, of course,

01:20:30.340 we think is to a degree equal opportunity. I mean, you can develop a mutation either because of a

01:20:35.180 replication error or because of an insult.

01:20:36.760 So the larger the gene, in theory, the greater the probability it can acquire a mutation.

01:20:41.840 Sure. And P53, the most famous tumor suppressor gene of them all.

01:20:45.120 How many base pairs?

01:20:46.400 P53, I don't know off the top of my head. I should Google that someday.

01:20:50.100 We'll do it for you. It'll be in the show notes.

01:20:51.880 Yeah. Right. So let's go with ballpark it at a couple hundred thousand base pairs. Big gene, 0.84

01:20:59.040 multiple exons that are separated by introns that are stitched together when the gene is

01:21:03.160 transcribed into RNA. And then...

01:21:05.540 And extrons and introns for folks is coding. I mean, it's part that actually gets turned into

01:21:10.460 RNA versus that that doesn't.

01:21:11.980 Exactly. But they still play an important regulating role. And again, at least scaffolding,

01:21:15.980 kind of stitching together.

01:21:17.020 And that's something that, again, 20 years ago, people thought those introns don't matter.

01:21:21.040 Yeah, they're dead. Right. Exactly. But that's where all the...

01:21:23.140 Junk DNA.

01:21:23.600 That's where all the regulating elements really largely reside. So it's a huge... In the past 20 years,

01:21:28.480 it's been a parallel track, big area of innovation. So in any case,

01:21:32.920 so then you have a big gene like P53. So 50% of cancers have genetic aberration P53. Well,

01:21:38.700 partly it's just a really large gene, but so you can pick up mutations in it seemingly relatively

01:21:43.700 easily. But it's a master regulator.

01:21:46.180 What percentage of cancers do not have a P53 mutation?

01:21:49.600 50%. And the other 50 do. 0.81

01:21:50.960 Oh, so it's 50-50.

01:21:51.900 Yeah, yeah, yeah.

01:21:52.220 I would have guessed that fewer people with cancer do not have a P53 mutation.

01:21:56.880 Do not.

01:21:57.360 Yeah, I would have thought it was almost essential.

01:21:58.420 Okay. So P53 is such a complex central node in a very complex network. So there's tons

01:22:04.780 of translated proteins that interact with the protein product P53. And then P53 has a lot

01:22:09.860 of outputs, like literally networks of outputs. It's in cancer, in any case, I would say it

01:22:14.140 is the most magnificent and thus far well mapped out of these networks. What is the function

01:22:19.360 of P53? Very simply put, it's basically a sensing apparatus. It's trying to understand

01:22:23.400 how bad things are in a cell. And usually people would say, well, first, if there was a singular

01:22:28.440 function, it's how poorly is the genetic code doing? Which is to say the chromosomal architecture,

01:22:34.760 how intact versus not intact is it? How many mistakes, mutations or mistakes, so insults versus

01:22:40.920 mistakes, how many of those exist and are being repaired at any one time? Like, so that machinery

01:22:44.940 is at work, the DNA repair machinery that feeds then into P53, which basically is just sensing

01:22:50.120 overall, how well are we doing in terms of kind of... And P53 possesses the power to command

01:22:57.300 apoptosis directly. Yeah. So it senses damage in a normal cell, that's its function, in a cell that's

01:23:02.380 becoming abnormal, continues to keep its finger on that pulse. And if there's catastrophic damage,

01:23:07.220 then it's P53 that says, let's not try to repair this anymore. Let's fold up shop and undergo a civilized

01:23:12.660 cell death and let our neighboring, let's say, liver cell just take over and do our job if we're a liver 0.99

01:23:18.560 cell. So P53 is the master regulator in all cells in the body. I love doing these podcasts because I

01:23:23.100 still, even if it's on a topic like this where I know a little bit, like I'm amazed at how much I

01:23:27.100 keep learning. So 50% of people who get cancer have a perfectly intact P53. Yeah, that's right.

01:23:33.140 Yeah. But the point about describing the network is the likelihood that they will have a genetic

01:23:37.600 aberration in one of the inputs or one of the outputs of P53 is enormously high. Got it. So we say

01:23:43.080 the gene P53 only 50% of the time is mutated, but the pathway is virtually always hosed.

01:23:48.800 So retinoblastoma gene, you're familiar with that one. That's another very famous,

01:23:52.140 long ago described tumor suppressor gene. What was not known decades ago, but is known now is that

01:23:57.420 they actually relate to each other and you will find cancers that if they don't have a P53 mutation,

01:24:02.080 they will have an RB retinoblastoma gene mutation. It's another big tumor suppressor gene that actually

01:24:07.280 is quote unquote downstream or regulated by P53 and its alteration will do much of what a P53

01:24:13.560 mutation will do. So there's these kinds of functional substitutes in that axis. On the

01:24:18.460 activated side, that's where I spent most of my career. So KRAS. Exactly. Talk to me about KRAS.

01:24:24.280 What is this clown doing? So I was just going to say that growth factor receptors, which I touched 0.94

01:24:27.400 on before because epidermal growth factor receptor and HER2 were these kind of early discovery slash

01:24:32.720 therapeutic translation exercises. Well, that was the beginning of a theme that to this day we think

01:24:38.820 is kind of the biggest unit in terms of where the activating events happen and where we have drugs

01:24:44.820 now is in the growth factor receptor RAS and RAS pathway system. So growth factor receptors,

01:24:52.420 literally their normal function is to receive growth factors. So in a cancer cell, if you can figure out

01:24:56.940 how to grow in a growth factor independent way, then you've accomplished a good trick because now

01:25:01.880 you'll be able to survive very harsh environments. You'll be able to replicate kind of an autonomous,

01:25:06.140 not governed by not just environment in terms of nutrient availability, but even like what your

01:25:10.360 neighbors are telling you, you should and shouldn't do. You can ignore that.

01:25:13.060 One of the explanations for why we see more aggressive prostate cancers in men with lower

01:25:18.180 testosterone levels than higher testosterone levels. If your prostate cancer can grow without

01:25:22.700 testosterone, beware. It's a bad problem. And then same with breast cancer, with hormone receptor,

01:25:27.220 positive breast cancer, quote unquote, versus negative, the prognosis and treatment response.

01:25:31.140 Yeah, the negatives are wildly different.

01:25:32.520 Much more different.

01:25:33.000 A different disease, basically. So that's exactly right. So cancers fundamentally need to accomplish

01:25:38.420 this task. So growth factor receptor genetic alterations, HER2 is genetically amplified,

01:25:44.780 then like massively increasing the number of surface receptors and allowing them to actually complex

01:25:49.400 together and signal in the absence of needing the growth factor itself. That was the seminal discovery

01:25:54.640 going back even into the 80s, but certainly picked up a lot of steam in the 90s. And then was a validated

01:25:59.760 target, ultimately, as I said, in a relatively slow motion way compared to BCRA-able, which we're

01:26:04.360 going to come back to. So anyway, these growth factor receptors feed RAS, R-A-S, and that comes

01:26:10.900 in three forms, KRAS, HRAS, and NRAS. And 25% of all cancers have a RAS mutation. Right downstream of

01:26:18.360 these growth factor receptors is where RAS sits just on the inside of the cell surface. We can't drug it

01:26:23.120 directly, certainly with an antibody. Small molecule inhibitors have been hard. That's its own discussion

01:26:27.100 terms of what's assailable and what's not assailable in terms of cancer drug targets, but let's park

01:26:31.520 that one and say that RAS is a big deal. And then RAS has its so-called effector pathways. And the

01:26:37.560 numbers, to a degree, I think, how many people agree on the number of cancer-relevant RAS effector

01:26:43.200 pathways? I can't find a lot of consensus there, but six is a reasonable number of described cancer-related

01:26:49.260 RAS effector pathways. So downstream of RAS, RAS will activate other cascades. The two most famous are

01:26:55.080 the MAP kinase pathway, where I've focused my career, PI3 kinase pathway. Arguably, there's

01:26:59.540 more mutations that activate the PI3 kinase pathway than the MAP kinase pathway. But

01:27:02.580 Lou and I have been trying to sit down for, we cohabitate parts of New York that are 10 feet away

01:27:07.460 from each other. So at some point, Lou and I are going to sit down and have a lengthy discussion on

01:27:11.140 PI3K. Yeah. This is where I wanted to take a little deep dive into Lou and Craig Thompson's-

01:27:16.520 By all means. Travels in the PI3 kinase pathway. Here's a pathway in this growth factor

01:27:20.980 receptor signaling apparatus. Metabolic regulation comes, we think, largely, but not completely,

01:27:27.900 through the PI3 kinase pathway. If you link up all the discoveries of Lou and others regarding the

01:27:33.300 importance of PI3 kinase in cancer, the fact that 20% of cancers have PI3 kinase, intrinsic mutations

01:27:39.060 in one of the isoforms of PI3 kinase, most commonly alpha, it's a nasty little trick. I mean,

01:27:44.680 it's co-opting this kind of metabolic regulation pathway. It largely explains growth factor independence

01:27:50.380 of cancers that have those mutations. But these are metabolic pathways that are fundamental to

01:27:54.840 normal cells as well. And so where's the therapeutic index in terms of leveraging that

01:27:58.660 observation? This has been a major challenge and not yet really adequately tackled. But one positive

01:28:04.400 result, at least in PI3 kinase mutant breast cancer that finally came across in slow motion,

01:28:09.320 phase three result. When it takes a phase three clinical trial to be the aha moment, that's a slow

01:28:13.240 motion result versus 20 patients get treated and you know you're in different territory, which is

01:28:17.440 the BRAF example in melanoma and then other cancers thereafter. So the MAP kinase pathway is this

01:28:23.560 proliferative pathway as it's canonically described, but it does other things, at least in cancer when

01:28:27.880 it's co-opted by mutation. Common theme, by the way, when you activate an oncogene in cancer,

01:28:33.820 very commonly you will see that the downstream wiring diagram changes and that pathway is now able to do

01:28:40.720 more things than we would have given it credit for or said that it had those same abilities in a

01:28:46.180 normal cell. It's what we mean by oncogene addiction. If you turn it around in terms of

01:28:50.300 building blocks, oncogene addiction means that cancer really needs that activated oncogene to

01:28:56.800 be able to do not just one thing like dry proliferation, but actually to alter other essential

01:29:01.620 programs for cancer. The flip side is that normal cells in the body, they have other ways. They can

01:29:07.340 break up the work between proliferation and metabolism between, for example, the MAP kinase pathway and the

01:29:11.860 PI3 kinase pathway. But in melanoma, for example, where we see, we call it kind of this paradigmatic

01:29:17.280 MAP kinase pathway activated tumor. Yeah, it cares about the PI3 kinase pathway, but in quite a secondary

01:29:22.620 way. Breast cancer, flip that around. So this idea that we could actually come up with an understanding

01:29:28.920 of these normal cell processes that are co-opted by cancer, drug them, and do more harm to the cancer

01:29:34.180 cell than the normal cell, it wasn't intuitive to people. I mean, this goes back to this watershed era of

01:29:39.980 the early 2000s when the concept of targeted therapy finally kind of had its aha moment.

01:29:45.360 It really relied on this concept. And term oncogene addiction, when did I first hear that? I suppose it

01:29:52.040 was in the early 2000s. This idea that by a cancer co-opting this one molecule, it's actually now

01:29:58.760 essentially using it to drive kind of more components of cancer biology than you would have thought in

01:30:05.060 terms of the normal physiologic role of that molecule and where there's compensatory mechanisms

01:30:09.520 and parallel processing that can happen in normal cells that make it not so dependent on the function

01:30:14.240 of that one molecule. So anyway, RAS, 25% of cancers have a RAS mutation. 20% of cancers have a PI3

01:30:19.820 kinase mutation. There's a little bit of overlap there. 8% of cancers have a BRAF mutation, which is

01:30:24.600 intrinsic in the MAP kinase pathway. It's the most popular point of mutation in that pathway.

01:30:29.300 What percentage of melanomas have a BRAF mutation?

01:30:31.400 About 50%. And it was that simple alignment of facts. I was finishing my fellowship in 2002,

01:30:38.640 June of 2002, becoming faculty at Penn in July of 2002. And June of 2002 in Nature was described the

01:30:46.220 research project that the Sanger Center in the UK, one of the sequencing powerhouses then and now,

01:30:53.160 they had launched this campaign specifically to sequence the RAF genes, not RAS, R-A-S, the one that's

01:30:58.340 undruggable, unfortunately, still to this day, but rather the RAF genes. Why did they do this?

01:31:03.100 Because the MAP kinase pathway had been implicated as being relevant in cancer for a couple decades.

01:31:07.680 And no one knew other than RAS mutations that can activate it. Outside of that, no one understood

01:31:12.840 how and why the pathway could be co-opted by cancer cells. So a logical thing to do would be just go one

01:31:18.780 bucket down in the bucket brigade from RAS to the molecule that RAS regulates, which is RAF,

01:31:23.720 three isoforms of RAF, C-RAF, B-RAF, and A-RAF, understood lesser and lesser degree across that

01:31:30.020 sequence. So C-RAF was the first discovered, called RAF1 at the time. Then it was renamed C-RAF

01:31:34.640 later. People had studied C-RAF to a large degree. B-RAF, not so much. There were only really a couple

01:31:40.440 slash few B-RAF mavens in the world, and still to this day, not very many A-RAF mavens.

01:31:45.280 The hypothesis was that these RAF genes are probably going to have some mutations,

01:31:49.800 and C-RAF would be the one most likely because it'd been the best studied up to that date.

01:31:54.720 So the big headline from that nature paper was rarely ever do you see a C-RAF.

01:31:59.860 B-RAF was the big discovery. 8% of all cancer was the estimate when they sequenced 484 tumors.

01:32:05.880 A-RAF rarely, if ever, mutated. So B-RAF was the one. The why of that is its own fascinating little

01:32:11.900 bit of kind of molecular evolutionary history. B-RAF and C-RAF are different molecules.

01:32:16.240 Does RAS actually play a causal role in the mutation, or is it more a function of

01:32:21.840 the things that bug RAS bug B-RAF? Yeah, that's true. They're related. They talk to each other.

01:32:29.000 But let me answer it this way. In a cancer that has a RAS mutation, you will not find a B-RAF

01:32:34.800 mutation and vice versa. They are mutually exclusive. So you don't need to skin that cat twice. You do it

01:32:40.220 one way. RAS is sufficient or B-RAF is sufficient to get activation of the pathway. Melanomas,

01:32:46.320 for example, 50% of them have B-RAF mutation. Very similar to the P53 problem.

01:32:50.100 Yeah, that's right. These tumors, anyway, need it. And many cancers need this pathway on.

01:32:54.980 How they accomplish it varies. And we don't understand all the determinants of why certain

01:33:00.300 cell types are more prone to picking up certain mutations as their, if you will, their means of

01:33:05.620 activating certain pathways. But there's certainly a teleology argument there. Any case, so B-RAF

01:33:11.060 mutations discovered an 8% of all cancers. I had decided melanoma was a cancer of terrible unmet need.

01:33:16.740 Interesting biochemistry insights coming from the previous couple decades. Very strong lab-based

01:33:21.720 science in melanoma at Penn, where I was choosing to put myself on the clinical frontier, clinical

01:33:27.000 research frontier. This aha moment in this paper was that 8% of cancers had B-RAF mutations,

01:33:32.060 but the cancer type that most commonly had them in that paper, later paper described that there's

01:33:37.620 one rare cancer that more commonly has B-RAF mutations, was melanoma at 50%. And the vast,

01:33:44.080 vast majority of those mutations affected a single point in the gene.

01:33:47.900 So let me pause now to go back to the question so that people understand scale. A point in a gene,

01:33:53.500 right? So you have 23 chromosomes, call it 30,000 genes. Let's make the math easy. You could have

01:33:57.660 about 1,000 genes wrapping around each chromosome, order of magnitude. Okay. So one chromosome's got

01:34:03.500 about 1,000 genes wrapped around it. You got 23 pairs of those. And now each gene could be anywhere

01:34:10.540 from a few hundred to a few hundred thousand base pairs. And what you're talking about is one of those

01:34:17.760 could be mutated. You could get one letter wrong out of 100,000 and you change the function of a gene.

01:34:24.020 Exactly. You asked before about kinases and this is going to-

01:34:26.960 All of this is in service of CML.

01:34:28.880 Right. Exactly. We're going to link those two concepts. So basically we have, I use that phrase

01:34:34.160 bucket brigades. We have these pathways where one molecule alters another, alters another. We

01:34:39.640 too commonly think of these as linear, one augmenting another followed by another and not

01:34:44.180 as systems that where there's side inputs into these pathways, which has been well described,

01:34:49.480 including in the MAP kinase pathway. In any case, follow me here that basically

01:34:53.260 RAS activates RAF. When RAS itself is activated, it pushes RAF into a pair of molecules, so-called

01:34:59.580 dimers, and will facilitate their phosphorylation or activate a molecular feature change that allows

01:35:05.700 them to be active. So kinases are a form of enzyme whose job is to add a phosphate group,

01:35:13.320 a single fairly small molecular entity to specific amino acid residues on its target. And usually it's

01:35:20.900 more than one target, but the point is that there is a fidelity in terms of that relationship where

01:35:25.120 you've got a kinase and its substrates, plural. So RAS will activate RAF, RAF will activate MEK,

01:35:31.180 MEK will activate ERK, all through phosphorylation events. So they glom onto each other and find the

01:35:35.800 right domain and stick a phosphate group on there. People are used to hearing me say that I get

01:35:39.760 phosphorylated and I think they understand exactly what you're-

01:35:43.220 I can sometimes- my five-year-old can get me more phosphorylated than any tyrosine kinase in the

01:35:49.980 history of our known universe. So phosphate residue additions are- they're not the most important,

01:35:56.200 but they're a key facet of how the molecular machinery works inside of cells in terms of how

01:36:01.080 to activate or inactivate these networks. And here we're talking about growth factor receptor

01:36:05.540 related networks. So right in the middle of the so-called kinase domain, the part of BRAF

01:36:11.460 that actually is basically responsible for latching a phosphate residue on MEK, its downstream

01:36:16.620 substrate, right in the middle of that domain, I mean literally right in the middle, is where

01:36:20.560 these mutations happen in the vast, vast majority of cases. That discovery alone was just shocking,

01:36:26.480 right? Because by chance alone, you're not going to find- stumble upon these mutations piling up in

01:36:30.840 8% of cancer. And there's one point right in the middle of the kinase domain.

01:36:34.780 That wasn't also in the same paper, was it?

01:36:36.760 No, that point was.

01:36:37.960 It was.

01:36:38.420 Oh yeah. The distribution, if you will, of them being V600 position mutations, a valine at the

01:36:43.660 600 position in the amino acid sequence, right in the heart of exon 15, which is in the middle of

01:36:47.920 the kinase domain, that was in the paper. The additional killer experiment that they did was to

01:36:52.700 basically transfect that into a fibroblast, so a normal cell, and show that that could transform

01:36:56.980 them and make them proliferate to a degree. And that was it, kind of end of paper. So the phenomenology

01:37:02.260 that these mutations exist, this wild distribution, or if you will, non-distribution, this like piling up

01:37:07.840 at this one position, it was just this huge aha moment. And when you link that in melanoma with

01:37:13.680 this couple decades worth of insights that the map kinase pathway really seemed to be important in

01:37:18.280 this tumor, and now you find these mutations sitting here right in the middle of the pathway,

01:37:21.380 like that was just like the most drop-dead, obviously important thing in my view, and a

01:37:26.260 couple other people's view, but shockingly few other people at the time decided to take a complete

01:37:30.600 left or right turn and focus on it. But that was the dawn of my career. So this is a problem with

01:37:35.160 biomedical research, is that every time a discovery is made. It impacts the people who are in search

01:37:40.380 of substrate. Precisely. Who've said, I want to get to the frontier of known and not known,

01:37:44.060 and I want to investigate. Everybody else is busy. They're already doing their thing. And if they're

01:37:47.920 grant-funded, then forget about it. Like they're already mining away. I think of the lab I was in,

01:37:51.860 in 02, 03, 04, it was all immunotherapy. Like we never talked about, I mean, maybe that paper came up

01:37:59.060 at Journal Club, as like one of 50 papers discussed that year, is interesting. But I mean, it was

01:38:05.480 anti-CTLA-4. It was TIL. It was, because everybody there was super seasoned, super senior, and they

01:38:11.240 were already on this path. That's a great point. I never really thought of that bias that can exist

01:38:16.180 temporally through a person's career.

01:38:18.580 Yeah. And melanoma, for all the reasons we discussed before, had been dominated by cancer

01:38:22.140 immunologists, because they recognized that there was this robust evidence of tumor-immune

01:38:26.080 interaction. You had this huge opportunity.

01:38:27.980 Yeah. And so the idea that you could just make one more maneuver, be it a cell therapy or a

01:38:32.980 checkpoint antibody or a cytokine, and tip the balance and clear the tumor, the field was very

01:38:37.740 focused on that. And if you were interested in the general notion of cancer immunity, well,

01:38:41.840 melanoma seemed like a very natural home for that work. And that really was the subtext. I came walking

01:38:47.200 in with this very different idea.

01:38:49.100 The last thing you wanted to do was do what everybody else was already doing.

01:38:52.040 Exactly. I'll just finish this point by saying that this is the whole coaching point to the

01:38:57.460 petrified or even paralyzed young trainee who's thinking, well, how am I going to find my entry

01:39:03.320 point? Where am I going to find something to work on? What I say is, okay, you're not waiting

01:39:07.440 for something top secret that someone comes and whispers in your ear. What you're waiting

01:39:11.000 for is the next Nature Science and Cell paper to be published that describes something that's

01:39:15.000 very important in an area that you've said you've just, for some intuitive reason, that you have

01:39:18.640 an interest in. And trust me, the field is not going to drop everything that they're doing

01:39:22.720 to go pursue that. But if you're at the dawn of your career, you're in the perfect moment

01:39:26.600 to now actually build the knowledge base, meet the people who are the relevant players, assemble

01:39:31.720 the knowledge, pull together the tools to actually start testing this hypothesis, whether it's

01:39:35.240 you're a wet lab investigator or a clinical investigator, the same. And I've never witnessed

01:39:39.220 a case where someone's been basically crowded out or hasn't been able to make a career

01:39:43.360 in investigation by using that approach. I'm trying to come back to the watershed moment

01:39:47.360 of BCRA, but I'm going to do it this way. So BRAF mutations were discovered in 2002.

01:39:52.300 People ignored it, not only because they thought tumor-immune interactions were a cooler thing

01:39:57.340 in melanoma, but remember, melanoma is jacked up with mutations. It's got an inordinate number

01:40:02.700 of them. So you find this one. So what? If there's a cancer on Earth where we're poking

01:40:09.340 that beast with a single drug approach, just antagonizing BRAF, is going to do nothing.

01:40:14.960 This is the example. That was the headwind argument.

01:40:17.940 Absolutely. In fact, I would say that that should be the null, second, and third, fourth

01:40:23.880 alternative hypothesis. That is a capital so what?

01:40:27.760 Yeah, exactly. Maybe we'll circle back later.

01:40:30.820 Maybe that's why it wasn't even mentioned in journal club.

01:40:33.120 And then when the first putative RAF inhibitor didn't work, this was the other big, ah, we

01:40:37.700 told you so moment, which let's table that for a moment. So chronic myelitis leukemia

01:40:42.860 is fundamentally different. I mean, this was known even in the seventies, that if you look

01:40:46.400 at the splay of chromosomes and you see this one migration of one segment.

01:40:50.140 Right. And the listener now understands why I got phosphorylated when you said that so

01:40:53.700 they could understand it. These chromosomes are huge. Like one piece of one of them could

01:41:00.300 literally, it's like picking up a building in a cell and moving it and attaching it to

01:41:05.640 a condo somewhere else.

01:41:07.400 So this was observed by Peter Knoll decades ago. And as a characteristic event in CML, that

01:41:13.400 you would see this in at least 95% of cases, this massive chromosomal shift, but not others.

01:41:17.660 In other words, it wasn't like the DNA was shredded at the chromosomal level. It was this

01:41:21.620 one migratory move and it was like characteristic pathognomonic of the disease. That was striking,

01:41:28.100 right? Just that fact.

01:41:29.540 Striking. It's also important for the listener to understand. You could see it under a microscope.

01:41:33.440 Like there's nothing genetically, like all this stuff you and I are talking about right

01:41:38.760 now, you don't look at your microscope and see that.

01:41:41.200 No, you got to sequence individual base pairs and large numbers of them.

01:41:43.760 This might be the only genetic mutation in cancer that is visible under a microscope.

01:41:48.980 Well, we'll come back to that.

01:41:50.280 There's another example?

01:41:51.180 Yeah. Well, fusions in general.

01:41:52.480 Okay. Yeah, yeah, yeah. No, that's a fair point.

01:41:53.840 So the term translocation was the term that was used then. Now we use the term fusion.

01:41:58.400 Yeah. My point being anything outside of a piece of a chromosome moving.

01:42:01.640 So fusions as a theme in cancer have their own kind of substrate of discovery, but this was the

01:42:06.940 first one. So basically you had what appeared to be at the chromosomal level, genetically simple

01:42:11.540 cancer, where nearly always one big fragment migrated to join another big fragment. What was

01:42:17.020 at that juncture? Like why were those two coming together? So fascinating. So this is a white blood 1.00

01:42:23.120 cell cancer, right? The myeloid cells are white blood cells, a branch of them anyway. And so to get

01:42:27.800 chronic myelitis leukemia, you needed this genetic migration thing to happen. On one side of it is

01:42:34.100 the BCR gene. And that gene is basically responsible for immunoglobulin kind of reshuffling in white

01:42:41.980 blood cells. It's a very dynamic, very active gene in white blood cells. If they're going to be able

01:42:46.840 to do their immunologic job and create the relevant kind of repertoire of foreign recognition,

01:42:52.200 then the BCR gene needs to be active to facilitate that program. Let me put it that way.

01:42:57.500 So any case, the BCR gene is on one end of this migration event. It's not a doer in and of itself.

01:43:04.860 It's just a very active gene locus that's just on, on, on in white blood cells. On the other end

01:43:11.980 is the abel kinase. So abel kinase is a signaling molecule. It's inside of cells. It's not important in

01:43:18.940 all cell types. It's important to a degree in white blood cells. That's, I think, a fair

01:43:23.940 summary statement now, looking back 20 years in retrospect with a lot more information.

01:43:28.380 But this is a signaling molecule that does a lot of work inside of cells. It's a kinase,

01:43:31.660 so it phosphorylates substrates. And when BCR-ABLE, this very active regulating domain,

01:43:38.080 is stitched onto the abel kinase.

01:43:39.940 BCR is stitched to abel.

01:43:41.180 Yes, correct. So it creates the BCR-ABLE translocation or fusion. So you now have a new

01:43:44.940 gene product of these two genetic components being stitched together that otherwise

01:43:48.160 wouldn't have been. They're quite far apart from each other on chromosome 9 and 22.

01:43:52.100 So in any case, when they come together, you crank up the expression of abel kinase. So abel kinase

01:43:57.240 is normal abel kinase. It's not mutated.

01:44:00.100 Yeah. It's just all of a sudden, instead of firing once every minute, it fires 40 times a minute.

01:44:05.540 It's jacked up. It's an expression to preposterous degrees. You have an enormous amount of abel kinase

01:44:09.680 being made in these white blood cells because the BCR-ABLE locus is just being driven all the time in

01:44:14.580 normal white blood cells. So now you've got ABLE on the other end of that.

01:44:18.380 Which is really aptly named.

01:44:20.040 Abelson was the discoverer of it.

01:44:22.360 But in this context, it's aptly named. It's very ABLE. It's an ABLE kinase that has become much

01:44:26.540 more ABLE.

01:44:27.280 Right. But unlike BRAF, which has this activating mutation sitting right in the middle of it,

01:44:31.660 nucleotide substitution resulting in an amino acid substitution that alters its kinase function,

01:44:35.520 this is just normal ABLE.

01:44:36.780 Yeah. It's just more on.

01:44:38.040 Yeah. And HER2, amplification and breast cancer, also just there's more of it. And then they

01:44:42.560 complex together and signal more. So when we talk about mutation and genetic alteration,

01:44:47.280 just understand that there's amplification, there's point mutation, and there's translocation

01:44:51.420 or fusion. These are the three canonical ways that you can have it.

01:44:54.520 And do most of the fusions have this phenotype?

01:44:57.780 So this is other fusion cancers outside of-

01:44:59.960 Yeah. Meaning do they have this phenotype of normal protein just doing more?

01:45:05.520 Precisely. Yeah. And kinases are the ones that have been discovered, described, and now drugged

01:45:10.120 multiple times. So what's fascinating about now, this is a 2019-ish insight, or at least the past few

01:45:16.820 years, is these fusion or translocation, quote unquote, driven cancers, they tend to be genetically

01:45:22.860 simple. Not all of them. Point mutations happen in the sea of genetic complexity, melanoma being an

01:45:30.360 example, but all BRAF mutations distribute across cancer types that are actually quite genetically

01:45:34.820 complex. There's a lot of other genetic aberrations turning on and turning off other things.

01:45:39.380 These fusion-driven cancers, they seem to get a lot of juice out of that one genetic alteration.

01:45:45.960 So BCR-ABLE was the initial example. But remember, it's chronic myelogenous leukemia. It would kill

01:45:50.280 patients over five to seven years. And it was super genetically simple, at least at the level that

01:45:55.160 one could make such comments in the 90s, which is when therapeutically attention started to be

01:46:00.180 turned to this able kinase phenomenon. So it turned out that Siva-Gyge ultimately subsumed into Novartis

01:46:06.860 over serial acquisitions, had a kinase inhibitor program broadly for cardiovascular disease. I knew

01:46:13.080 Siva-Gyge from my father's academic cardiology pursuits. So they had these kinase inhibitors. They

01:46:18.620 were pretty crude instruments. But in the mix, my understanding initially was all endothelial

01:46:23.620 proliferative cardiovascular disease. It was the kind of phenotypic screening, if you will, that was

01:46:28.740 being done with the kinase inhibitor library at the time. Again, this predates the cancer

01:46:32.880 investigations. So they created this library, not massive, of kinase inhibitors. It's a small

01:46:38.000 molecule tool compounds. So there's this guy, Brian Druker, at Dana-Farber in Boston at the time,

01:46:43.080 who uncovers that now Novartis has this kinase inhibitor library. And in it is a apparent

01:46:49.560 able kinase inhibitor. Not perfectly selective for able kinase. Kinases exist in the hundreds.

01:46:55.340 So about 600 or so kinases have been described in the family tree. They're highly related,

01:47:00.880 but then there's some that are more distant cousins than others. And if you try to inhibit

01:47:05.520 a kinase, it's pretty easy to pick up inhibitory activity against another kinase because of their

01:47:11.220 relatedness, their structural relatedness. So it turns out imatinib, the first able kinase inhibitor,

01:47:15.980 it was not just an able kinase inhibitor. Fortuitously, it was also a C-kit inhibitor,

01:47:20.940 which we'll come back to because that's the gastrointestinal stromal tumor, dual purpose of

01:47:24.440 that molecule. But in any case, he wanted an able kinase inhibitor. The more perfect he could have

01:47:28.820 had, the better, more selective and only targeting able. But back in these days, certainly in the 90s,

01:47:35.500 it was a difficult sport to actually profile how promiscuous or selective a kinase inhibitor was.

01:47:41.680 So it wasn't really known what its full selectivity spectrum was. But able kinase is what it was

01:47:46.640 labeled as an inhibitor of. So it's been cataloged in many papers, but just briefly put that basically

01:47:52.940 Brian started this campaign to try to get Novartis to liberate this molecule that had been to a degree

01:48:00.600 investigated in cardiovascular disease models, but to like kind of get it out of the company and have

01:48:06.960 it actually made available experimentally initially in model systems, not even in humans yet.

01:48:12.040 Any case, fast forward now. So the drug enters phase one clinical trials in cancer patients with

01:48:18.640 knowledge that there's this cancer out there.

01:48:20.840 And this is like mid 90s?

01:48:22.660 No, no, no. This is now late 90s. And I'm about to show up in fellowship in July of 2000.

01:48:27.380 And it's 2000-2001 that the ongoing phase one trial cleared the first few dose levels where

01:48:34.640 somewhat homeopathic doses were given. Didn't have to go on for very long as a phase one study.

01:48:39.660 Recruiting these patients, these patients who didn't need to be molecularly tested because you

01:48:43.860 just knew by their diagnosis that they were at least 95% likely to have this alteration in those

01:48:48.420 cells. And so it was in the midst of my first year of fellowship.

01:48:52.340 Yeah, this is the phase one trial where there's efficacy.

01:48:54.420 Yeah. First time ever that basically consecutive patients were responding

01:48:58.980 to therapy in a phase one trial as the dose was still being escalated.

01:49:03.100 So talk about aha moment. I mean, you needed three patients, six patients,

01:49:07.180 not a big phase three trial to know that there was a transformational event happening.

01:49:11.680 So that happened in the first year of my fellowship. And my naive talking points, I was saying,

01:49:17.540 this is the future of cancer. This is what we're going to do.

01:49:19.920 This is it. This is our first ACT moment. Drugs working extremely reliably, killing lots of cancer

01:49:26.220 cells, admittedly not eradicating all of them in most patients. But still, there were cures even

01:49:31.020 in the early days, or at least durable, complete responses. So anyway, the point being that this was

01:49:36.400 the big moment. The objection to that big moment relates to the comment I made a little while ago,

01:49:41.500 which is, this is chronic myelitis leukemia. This is a genetically simple thing. This is barely cancer

01:49:48.180 compared to pancreatic cancer. This is barely cancer compared to non-small cell lung cancer.

01:49:52.660 Okay, fine. It worked here. But why on earth would you think this is going to work in real cancers?

01:49:58.380 And again, let's make sure people understand why you're saying CML is barely cancer. You have a

01:50:03.220 translocation fusion that does not even mutate the kinase involved. So is it safe to say that

01:50:11.420 someone with CML doesn't actually contain true genetic mutation?

01:50:16.920 That's fair. Stable phase CML, as it's called, which is this long period of time where people will

01:50:23.020 have large numbers of these abnormal white blood cells circulating, but not impairing their health in

01:50:28.080 any significant way. Stable phase CML is this genetically simple thing. Given enough time to

01:50:34.740 evolve, just like we talked about with the small solid tumor, give it enough time to evolve, it will

01:50:39.600 pick up more alterations. And the rate at which alterations are picked up accelerates also. So

01:50:44.440 evolution begins to happen in a more substantial way, leading to accelerated phase and then blast crisis

01:50:50.300 and patients die of an acute leukemia-like death. So CML isn't quote-unquote real cancer in this

01:50:57.780 objection rendering frame here for a good long time, but it certainly becomes real cancer

01:51:03.320 ultimately. So the New England Journal of Medicine publishes the back-to-back papers in 2001 now of the

01:51:10.180 phase one clinical trial describing these heroic and quite reliably observed responses to single

01:51:15.220 agent abel kinase inhibition with a non-perfect abel kinase inhibitor in the form of imatinib.

01:51:20.600 So this felt certainly like a big deal. The next paper in this back-to-back, same authors,

01:51:26.140 were reporting on the activity of imatinib, the anti-tumor effects of imatinib in patients with

01:51:31.160 accelerated phase and blast crisis CML. So same disease, now allowed to genetically evolve and

01:51:37.020 become more complex and certainly to pick up additional mutations beyond this foundational

01:51:41.080 translocation event. What happened in those patients? It wasn't even discussed. People for

01:51:46.360 years were celebrating this CML, stable phase CML result, and never would you see a talk describing,

01:51:53.060 well, what happened with the same drug now in a more genetically complex environment where the same

01:51:57.340 truncal as an original alteration existed, but now surrounded by these other partners that were

01:52:02.520 clearly important because now you're transforming into a truly aggressive life-threatening disease.

01:52:07.580 You got responses.

01:52:09.060 They weren't durable.

01:52:09.700 They were transient. They lasted for months at best, weeks in some cases. So yeah, you could poke it with

01:52:15.100 a stick, but it would just laugh its way right around within a very short time and patients would still

01:52:19.740 come to their disease. So this is where the debate was. Okay. So you have CML in its stable phase where

01:52:26.040 you get these kind of heroic, deep and durable responses. Okay. Map that out for me and the rest

01:52:30.940 of cancer.

01:52:31.660 Right. So you realize that CML is the exception. It's not the rule.

01:52:34.860 Well, this is the argument, right? This was the argument that never again are we going to find such

01:52:39.240 genetically simple cancers where you can get deep and durable responses from a single agent targeted

01:52:43.020 therapy. Now, again, I didn't feel so defeated by that argument in the sense that I,

01:52:46.760 my talking point, even before we had the cancer equivalent of AZT was, well, this isn't going to

01:52:51.840 be about one drug. This is cocktails. Look, look, it took cocktails to wrestle HIV down. And that's a

01:52:57.060 laughably simple organism compared to a human cell that's now been co-opted by the blueprint being

01:53:03.460 widely opened. So of course it's going to take combinations. How high order a combination do we

01:53:08.160 need? Well, that goes back to the building block argument. We still don't know any case, but we still

01:53:13.200 need to find our individual AZT moments and hope that they would actually do something for an

01:53:17.080 individual patient and not just serve as a biologic building block. So this was the subtext. Fast

01:53:22.760 forward and just connect one dot, which is that one thing that we have learned is that even in

01:53:28.640 so-called solid tumors, so leaving aside the leukemias like chronic myelitis leukemia,

01:53:33.880 even in solid tumors, you will find subsets of them that are driven by these translocation

01:53:39.700 fusion events. And they tend to be genetically simple. And now time and time again, that has

01:53:44.940 proven to be a population sparsely distributed, if you will, across cancer types where you get deep

01:53:51.440 and durable responses. That was the next big aha moment. So time matters, meaning? Evolution.

01:53:57.260 Well, yes, but also like if you catch CML and treat it early enough, in theory, you're going to have a

01:54:04.260 better response than waiting until the tail starts coming out of the dragon.

01:54:08.900 That's right. So the compensation that a cancer cell will be able to leverage by having now these

01:54:14.620 built-in accelerants or more disabled tumor suppressors, like the ability to adapt and work

01:54:19.800 their way around with de novo resistance or rapidly acquired resistance is absolutely a huge risk the

01:54:26.220 longer you wait. That CML example, well, we've begun to map it out in quote-unquote common solid tumors.

01:54:32.400 So in breast cancer, if you look at HER2, I was beating up HER2 for its not aha moment efficacy.

01:54:39.480 This is trastuzumab, the first naked antibody. But modified forms of HER2-targeted therapies have

01:54:43.920 come along since. These more armed antibodies, if you will, that definitely have greater effect.

01:54:48.760 But even just take trastuzumab, the naked antibody, its ability to help a metastatic breast

01:54:54.740 cancer patient live longer for a period of time, well-cataloged, but measured in months,

01:54:58.960 several months, let's say. Now you take that into the so-called adjuvant setting. So this is covert

01:55:05.920 metastatic disease. So for those who have heard enough jargon and had family members deal with

01:55:10.620 cancer, they know of two situations where the surgeon says they got it all, but the medical

01:55:16.060 oncologist is telling them, we're still worried there's still some cancer cells around. We're too

01:55:20.020 stupid to know if they're there or not. CAT scans can't tell us because they're too low resolution. 1.00

01:55:24.280 We're not there yet in terms of actually measuring circulating tumor cells or circulating tumor

01:55:28.320 DNA in a diagnostic and high-resolution way. But we're just worried because we know that looking

01:55:34.100 back a decade, if we had 1,000 cancer patients like you, we know that half of them are going to show

01:55:40.280 up with overt metastatic disease over the first few to several years of follow-up. So you know a time

01:55:46.920 that someone had microscopic metastatic disease after their surgery with curative intent.

01:55:51.740 So what I'm getting at is now that's the adjuvant setting. So the possibility but not certainty that

01:55:57.040 microscopic metastatic disease exists, you give therapy in that situation, systemic therapy,

01:56:01.800 to seek and destroy microscopic deposits. That's so-called adjuvant therapy. So with that jargon

01:56:06.360 stated, adjuvant use of HER2 antibody, this was the first example of this is why I'm giving it its due

01:56:12.380 credit, cures patients. Cures patients. You don't cure patients in the overt metastatic setting with

01:56:17.300 HER2 antibody therapy. Like that's unheard of. So that was the first

01:56:21.100 paralog in a common cancer that kills lots of women that this CML evolution concept.

01:56:26.840 And remind me what the numbers were because we sort of take it for granted today.

01:56:30.020 You take two groups of women that have HER2 new positive tumors that are NED, meaning surgically 0.95

01:56:36.820 resected down to having no evidence of disease as you described. Half of them get a placebo. Half of

01:56:41.880 them get the antibody in 10 years. What percentage are alive in each group?

01:56:46.640 Usually the way we think about it is like how many patients are saved, if you will,

01:56:50.080 because it all depends on the level of risk of recurrence to start with.

01:56:54.140 Yeah. So what's the AR and RR?

01:56:55.600 Yeah, right, right. So whatever you start with, so it's depending on the trial,

01:56:59.120 a third to a half of the recurrences that would have occurred don't happen by the addition. This is,

01:57:05.380 you're right to mention placebo, but by historical fact, in this case, everybody got chemo.

01:57:08.740 Everybody gets the same. Yeah, yeah.

01:57:10.140 Chemo plus minus, if you will. So the addition of the HER2 antibody or not. And this is how many

01:57:14.300 subsequent studies and other cancer types have been done to try to show the same kind of benefit of

01:57:18.940 treating only microscopic or digital disease as opposed to overt metastatic disease. So a third

01:57:23.640 to a half reduction in risk of relapse over long periods of time now, and stably so in the case of

01:57:29.400 HER2, we have lots of data.

01:57:30.240 Do women take anti-HER2 new for life now in adjuvant or how many, it's five years? 1.00

01:57:33.820 No, no, no. The hormonal therapy is the long duration therapy in the adjuvant setting. So

01:57:38.460 hormonally driven breast cancers, you're now complicating this, my attempt to make adjuvant

01:57:43.080 therapy, disease eradication sounds simple. Hormonally driven cancers like breast cancer

01:57:48.020 and prostate cancer in the so-called adjuvant setting, it is clear that the original data was

01:57:55.080 a year of therapy, then three years of therapy, five years of therapy. It's clear even that 10 years

01:57:58.600 of therapy is better than five years of therapy in a population. And hormonal therapy

01:58:03.260 prevents relapse at least to a third to half standard analogous to HER2. But when you see

01:58:10.340 an effect like that in series of clinical trials where longer durations of therapy to treat covert

01:58:15.960 metastatic disease, longer is better than shorter, it tells you you're not eradicating in everybody,

01:58:21.380 you're suppressing in some people. And that's the story over and over again. And gastrointestinal

01:58:26.000 stromal tumor, which was the first solid tumor big targeted therapy success where imatinib was being

01:58:30.940 repurposed for the fact that it's a C-kit inhibitor and C-kit is mutated in two-thirds of gastrointestinal

01:58:35.760 tremor tumors. That drug first showed benefit in metastatic patients, overt metastatic patients.

01:58:40.080 Then in a series of adjuvant trials, one year, three years now ongoing therapy beyond that,

01:58:46.920 incrementally better. Is there a reason to believe that some patients are being cured? Absolutely.

01:58:51.100 But there's some patients where disease is just being suppressed and maintained in a

01:58:55.720 micro-metastatic state from which they will not die, at least in the foreseeable future.

01:59:00.300 But this just goes to my point about, okay, is single agent targeted therapy a tumor clearing

01:59:06.860 treatment for very many cancer patients? Even when we use optimal next generation early detection

01:59:12.340 methodology, that is going to be realized in a real fraction of cancer patients. But we're still

01:59:18.400 going to need combination regimens to dismantle tumors in their fully complex way. The BRAF example

01:59:23.120 has already now played out. So melanoma, BRAF inhibitor monotherapy, improved survival to the

01:59:29.680 tune of a nine-month improvement in overall survival, which in melanoma, which is median survival used to

01:59:34.920 be six to nine months. So on average, six to nine months. So like getting a bump like that.

01:59:38.500 You're basically doubling the survival of patients with metastatic melanoma, but not curing.

01:59:42.620 Right. That was a big deal. Interestingly, BRAF inhibitor monotherapy was tried in an adjuvant

01:59:47.560 trial. Didn't meet its endpoint. So it apparently numerically reduced risk of recurrence, but

01:59:52.960 marginally so, and not enough to achieve statistical significance. So that trial was called negative.

01:59:58.120 In parallel, I and others had been pushing hard in the metastatic setting to go from AZT to doublet,

02:00:04.940 what was the first doublet HIV regimen? I'm blanking now, but in any case, a proteasome inhibitor.

02:00:10.460 And so in any case, in melanoma, we were treating with BRAF inhibitor monotherapy. We're seeing

02:00:14.100 responses. Those responses would last on average six months. It's an aggressive disease. Again,

02:00:18.640 patients would typically die within six months in the untreated state. So patients would respond,

02:00:23.420 and their response would be maintained for a huge range of times, but the average was six months.

02:00:30.000 We saw that basically the tumors were working their way right around the drug in the MAP kinase

02:00:34.060 pathway. They were bypassing BRAF through CRAF, most commonly. That was the easiest trick for them

02:00:39.500 to use. They didn't need to develop mutations that resisted the drug itself.

02:00:44.100 Which is a theme in PCR-able CML and other oncogene-driven tumors that are treated effectively

02:00:49.340 with targeted therapy, and fusion-driven cancers in particular. Remember those genetically simple

02:00:53.600 tumors? A very, very common theme is they need to mutate the actual gene that's being targeted

02:00:59.140 because they need that thing back on. They don't have very many tools in the toolkit. They need that

02:01:03.700 guy back on, and the only way they can evolve resistance is to basically repel the drug in the

02:01:09.380 first place. Those are so-called gatekeeper mutations. So in BRAF, mutant cancer,

02:01:13.600 BRAF mutant melanoma, you don't see those mutations emerge because it's too easy for

02:01:18.860 these cells to rewire their way past BRAF through CRAF. So we saw this happening in humans. So yes,

02:01:26.000 in parallel and laboratory systems, but more importantly, we were seeing it in humans within

02:01:29.940 the same year that we first documented responses in those patients because we were biopsying them

02:01:35.280 serially for research purposes, which was then thought to be a crazy concept, but now it's not so crazy.

02:01:41.120 In any case, we saw this bypass happening. We knew that there were available inhibitors of

02:01:45.840 downstream MEK, M-E-K, the guy that BRAF turns on. Those drugs already existed. They weren't shown to

02:01:51.420 be useful on their own yet anywhere in cancer. There were signs of life here and there in clinical

02:01:57.780 trials. But MEK is right below BRAF and C-RAF. And we said, let's just put these two together,

02:02:03.460 try to intercept this bypass. That worked. Like we went very quickly from BRAF inhibitor monotherapy to

02:02:08.440 BRAF-MEK combination therapy, including overall survival improvements that were as big as the

02:02:13.200 overall survival improvement. BRAF inhibitor monotherapy. I'm fast forwarding through my

02:02:16.640 entire career here. In any case, BRAF-MEK combination in the adjuvant setting prevents

02:02:21.600 relapse by 50% and patients receive a urotherapy, stop treatment. And there's a persistent gap there

02:02:28.040 in terms of cured patients. This is melanoma we're talking about. Melanoma that will work its way

02:02:32.780 around BRAF inhibitor monotherapy and on average six months. You can wipe it out in the covert

02:02:38.300 metastatic state, microscopic residual disease, aka adjuvant therapy, with that same regimen. So the

02:02:43.700 point is there is some real kind of early detection, early treatment theme that is absolutely yet to be

02:02:49.820 fully leveraged and realized because we're still working on the early detection technology, blood-based

02:02:55.500 mostly. But it's coming. And then adjacent to that, even in some cases, when we find cancers very early,

02:03:02.700 they are already genetically complex. We have to have a toolkit that allows us to dismantle

02:03:07.220 at more than one point. And that's its own long conversation. But the issue of adjacent to AZT,

02:03:14.020 what does that armamentarium look like in terms of the next agents that we're discussing?

02:03:17.820 Right. Because with HIV, we think of it by class of drug. You basically have a toolkit

02:03:22.760 of drug classes. And it seems to me that you've done a very eloquent job explaining this.

02:03:29.480 There are basically some fundamental pillars in growth and some fundamental pillars in protection.

02:03:37.340 And when the day comes that we have a toolkit that knows, okay, I've got these three things that can hit

02:03:42.760 this pillar, three things that can hit that pillar, four things that hit this pillar, six things that hit

02:03:47.460 this pillar, nothing that hits this pillar, and two things that hits it. I mean, you're in the golden

02:03:52.080 zone when you can start stacking, because that's what HART, highly active antiretroviral therapy

02:03:56.360 did. It basically put whatever, three or four pillars together. And at that point, HIV was not

02:04:02.280 cured, but it was chronic. You didn't have to die. You didn't have to get AIDS.

02:04:06.320 We can't keep hitting the same pillar and expect that we're going to cure cancer. So we got away with

02:04:11.860 it in melanoma because we kept hitting the same pathway, got away with it by hitting it twice and

02:04:16.780 improving outcomes and actually improving side effects, which is its own weird story, but really cool

02:04:21.560 story biochemically. The bigger point, just hitting some of the talking points that we touched on

02:04:26.420 before, is we need the activators of the immune system. We need the inhibitors of the activated

02:04:31.400 oncogenes. We need the drugs that target these epigenetic regulators. We need the metabolic switch

02:04:37.080 regulators, which are emerging, I would say, just as we speak, very early days. Epigenetics and metabolism

02:04:42.860 generally are what I point to to say that these are the pillars where the tools are coming,

02:04:47.680 but it's early. And I generally make the point that if we just flesh out those toolboxes and we're

02:04:55.720 lacking still, um, some famous gaps would be like understanding how to wrestle down telomerase.

02:05:01.600 This kind of like is a clock that exists in cells that allows them to basically measure their age.

02:05:08.820 Did you see the science paper last week? It was the science paper about the,

02:05:11.920 the astronaut Kelly's time in space, the year in space. I read a lay article summary of it,

02:05:17.320 but yeah, go ahead. You know, it's interesting. So twin brother, right? So one,

02:05:21.000 the, I only bring this up because you mentioned, uh, telomerase. So the, I could be having this

02:05:25.280 backwards. I think the telomeres of the astronaut that was in space elongated significantly during his

02:05:31.960 year in space, but within three days back on earth, completely reverted to normal, which of course

02:05:36.620 just made me question the importance of telomere length. It's an interesting point. So your pillars

02:05:41.480 then, cause I want to share with you my framework, which is purely a clinical framework. It's not a

02:05:46.060 research framework. It's a, you're on the front lines, you're a primary care doctor, you're a

02:05:51.140 patient. How do you think about cancer? Because while I think most patients, when you talk about

02:05:57.280 the big diseases, most people are afraid of Alzheimer's disease above all else because of

02:06:02.240 the phenotype. But when you think about probabilistically, most people are afraid of cancer

02:06:07.760 because the likelihood of getting it is somewhere between a third and a half, depending on your

02:06:12.100 gender. So you said epigenetics, metabolism, immune, would those be your three fundamental

02:06:17.360 pillars of cancer? Yeah. I mean, growth factor receptors, which again,

02:06:20.560 that was the original pillar because those are the first discoveries made, frankly, in cancer

02:06:25.180 biology and cell signaling. But these others are clearly the other major pillars of normal cell

02:06:32.080 programs that have to be co-opted by cancer for cancer.

02:06:34.160 Yeah. So the epigenetic modulation, the immune stuff, the metabolic stuff, and of course the

02:06:39.500 growth factors. So I usually tell patients, I think cancer is really hard. Like atherosclerosis

02:06:45.860 is inevitable, but we also know so much about it. Not everything, but we know enough about it. And we

02:06:51.440 have enough tools that look, if you really want to be aggressive, you can delay it so that you're not

02:06:57.320 going to have your first heart attack till you're a hundred. That's doable. That's totally doable.

02:07:01.040 If you start early enough, I'll save my soliloquy on Alzheimer's disease. But I say, look, cancer is

02:07:07.560 the hardest one. And it's the one that I think most about in the sense of it's the one that I am

02:07:14.140 least confident at our ability to reduce risk in. So I say, look, here's my take on it. Step one,

02:07:19.720 try not to get cancer. Sounds like a dumb thing to say, but we know a handful of things that are 0.73

02:07:24.240 increasing our risk for cancer. So let's keep those to a minimum. Luckily, most people have figured on

02:07:29.000 that smoking is not a good idea, but right behind it is insulin resistance and obesity. And so there's

02:07:33.980 something about that probably down. So if smoking was probably acting more on the mass genetic level,

02:07:39.800 something about insulin resistance was acting on this growth sort of pathway. Okay. So we could talk

02:07:44.640 about all the things we can do to not get cancer. Step two, which again, creates a lot of enemies,

02:07:50.780 especially depending on which side of the Twitter sphere you live on, is let's look for cancer early.

02:07:56.000 If burden of disease matters, you can take the approach of women should never have a mammogram.

02:08:03.140 Just if a lump shows up in your breast, go see your doctor, but otherwise don't do anything as

02:08:07.540 one end of the spectrum. And then you've got coconuts like me on the other end of the spectrum 1.00

02:08:11.340 that say, no, I understand why you might come to that conclusion if you're trying to do it at a cost

02:08:17.060 basis. And if you're only limiting yourself to mammography, which has such horrible sensitivity and

02:08:22.300 specificity, but if at least theoretically you could say, well, in a world where costs become

02:08:28.000 less ridiculous, i.e. not in the United States, if you took it out of the equation and you were willing

02:08:33.920 to layer mammography, which will always be important to catch a calcified lesion with diffusion-weighted

02:08:39.880 imaging, MRI, as an even superior technology to ultrasound, coupled with molecular screening,

02:08:46.780 well, you can make the case that no woman should ever present with breast cancer. And in that

02:08:53.900 situation, can we do better? And then you talk about looking at therapies that would go after

02:09:01.140 multiple pillars simultaneously, not in serial, not waiting for one to fail and the other to go on.

02:09:07.140 So that's my sort of Neanderthal approach to cancer. How would you make that more robust?

02:09:11.380 Okay. So what we're missing is still components of the toolbox to be able to actually knock out

02:09:16.720 pillars. So that's, I often say we still need to diversify our toolbox. Then it's the issue of

02:09:22.020 marrying diagnostics or therapeutics. So understanding the assembly process of cancer in a patient-specific

02:09:28.280 way and being able to deploy those therapies. That is an emerging hard task. I spend a lot of my

02:09:34.580 academic time railing against the impediments that keep us from pushing drugs together in

02:09:40.440 supervised settings, both experimentally and clinically in clinical trials. So this idea of

02:09:45.180 getting to following the HIV example, we are chronically facing headwind in terms of getting

02:09:50.500 there.

02:09:51.020 Is that because HIV killed patients so much quicker and there was more desperation that clinicians,

02:09:57.500 IRBs were more willing to move quickly to stacked therapies?

02:10:01.800 Yeah. We have all the same regulations. The HIV advocates created- 0.99

02:10:06.120 It was the advocacy.

02:10:07.100 Yeah. They've created regulations to cover life-threatening disease and cancer is captured

02:10:11.240 right in that. We have an amazing forward-thinking regulatory environment in the United States and

02:10:16.620 increasingly in Europe. Because of the lead blocking that was done by the HIV advocacy community,

02:10:21.940 we have the same benefits.

02:10:22.960 So the heavy lifting on that front was done in the 80s and early 90s.

02:10:26.000 Absolutely. And I've spent tons of time exploring whether there are remaining impediments there,

02:10:31.860 talking with FDA leadership in particular. And I usually quickly sum up to say they are on our

02:10:37.600 side. They are our friends. They have self-organized in a way that actually will be the accelerator,

02:10:42.080 not decelerator progress. The problem, if I were to put a finger on it, is the way in which companies

02:10:48.480 that decided they could see a business model in HIV and basically decided they were going to pursue it,

02:10:52.240 could create the toolbox within their one company. Had it not been for that, we would not have seen

02:10:58.260 doublet and triplet therapy.

02:10:59.560 And is that purely a simplicity of therapy and therefore an economic issue?

02:11:04.400 Precisely. And simplicity of the organism that you're-

02:11:06.260 That's what I mean. Yes. It's simplicity of the opponent.

02:11:08.460 That's right. The number of enzymes that the thing has in its genome, you could then

02:11:11.540 postulate as potential targets, massively smaller. And so within one umbrella of one company,

02:11:16.340 you could see combination therapies-

02:11:18.240 Right. So even though Merck or Pfizer could have a program under each pillar,

02:11:22.840 that's typically not the way it goes, is it?

02:11:24.940 Here's how I generally draw up the math. So 60% of drugs coming in the cancer pipeline now come from

02:11:31.340 small biotech companies. It used to be that Big Pharma was the driver. And the phrase I often use is

02:11:37.520 that Big Pharma outsourced its R&D to risk-taking, venture-backed, highly specialized small biotech companies,

02:11:44.120 not just in cancer, but here I'm sticking with cancer.

02:11:46.380 So you've got this distribution of where drugs are coming from that is a huge swath of different

02:11:51.780 firms, many of which have a single asset in the small biotech space. And at most,

02:11:56.040 they get to have two or three.

02:11:57.200 What's the sweet spot for Big Pharma? Is between phase two and phase three or between

02:12:01.040 phase one and phase two? If you exclude the big aha moments.

02:12:03.900 Yeah. Okay. So if it's an aha moment, then it's phase two. That's the acquisition moment. And if it's

02:12:07.940 not, then it's randomized phase two slash phase three. Optimal moment, as you said, kind of where's

02:12:12.380 enough risk been taken off the table and where now you've got a commercialization opportunity,

02:12:16.740 which is what Big Pharma is for us in oncology, at least. Yes, there's still innovation and incubation

02:12:22.180 new therapies coming out of Big Pharma. But as I said, it's shrunk down to a small component. The basic

02:12:27.160 science, as you well know, across all of public private domain is in the public sector, right?

02:12:33.040 The basic science, the stuff where you can't guarantee any kind of return in terms of when you're

02:12:37.380 going to get an insight that you could turn into a potential therapy. So you've got this fantastic

02:12:41.380 chaos of publicly funded biomedical research, the world's greatest bio. I was just about to say,

02:12:45.620 by scale, when you look at the public domain, everything from most of it, of course, being NIH,

02:12:51.440 but also Howard Hughes and others, what percentage of the world's pure exploratory science is funded

02:12:58.700 in the United States? 90%. It's that much of an advantage. Yeah. It's narrowing in Europe a bit.

02:13:04.020 With now central investment in cancer research in Europe, that number is dropping percent by

02:13:08.600 percent. Truthfully, if you said 50, I would have still thought that was impressive. If half the

02:13:12.840 world's basic exploratory research was happening. It's where the money is. The public investment

02:13:17.540 in research and then add the private. Well, I'm adding the philanthropic on top of that, but yes.

02:13:21.460 Yeah. Well, we live in a philanthropic environment that's not known in much of the world. So that is a

02:13:25.640 meaningful additional layer. So no, huge, huge, huge engine here is the public domain. Those

02:13:31.940 discoveries then are out-licensed to small firms over large firms by a huge margin. Of course,

02:13:37.600 I here live in one of the, well, not one of, the world's biggest biomedical research engine vis-a-vis

02:13:43.760 taking those discoveries into small firms and the Bay Area for sure and New York and San Diego are the

02:13:50.040 kind of other major pillars of that. But so you have these hubs of activity of taking new discoveries,

02:13:54.640 some of which come from outside the United States even, but then are incubated into companies.

02:13:58.140 What I'm describing here is this very dynamic, very exciting, very purpose-driven, expertise-heavy

02:14:05.480 biotech sector, which is great for so many reasons. But I'm bringing it up as a complaint,

02:14:11.520 even though I personally have benefited from the ability to step into entrepreneurial roles and

02:14:16.540 co-found companies and so on and so forth. That's been enormously gratifying for so many reasons.

02:14:20.940 I registered as a complaint because it's distributed our toolbox so widely. And we live in a world right now

02:14:27.860 where 0.37 of rational combinations of two cancer therapeutics that are still in investigational

02:14:33.780 territory are finding each other in clinical trials. I published a paper on this topic a

02:14:37.420 year and a half ago. It is a terrible sampling mingling rate, terrible. So we make new discoveries

02:14:44.060 all the time in the academic domain that would suggest a new combination. You've got patients dying

02:14:47.960 every day of that addressable cancer by molecular subtype or whatever. The likelihood that you're going to be

02:14:54.360 able to launch a clinical trial to marry those two drugs when they exist in two firms is that small.

02:14:59.660 That is a terrible problem. HIV didn't have that problem. But because of the complexity of human cells

02:15:05.440 and therefore human cancer cells, the toolbox is being chaotically distributed. We need a way to change

02:15:11.700 that. And we need a way for drugs to be able to be married in life-threatening cancer in a rapid, 0.82

02:15:16.360 rapid fashion. So it's not an inertia problem in the culture. It's not doctors. It's not patients.

02:15:21.700 It's not academic medical centers. It's not the regulatory environment. It's not.

02:15:25.400 And it's a shame because some of those are a heck of a lot easier to solve.

02:15:28.140 You'd think. This is the least rocket science thing that I've put my shoulder into the past

02:15:32.360 number of years now. Novel, novel cancer therapeutic investigation. But it's the hardest to change.

02:15:37.780 And to me, there's an obviousness in terms of focusing on this problem. But the obviousness in the

02:15:44.140 entire ecosystem in terms of actually getting alignment and incentives aligned to allow this

02:15:50.280 dabbling to happen, it is a major, major impediment. And I've been making the point that I think we need

02:15:56.140 quadruplet therapy to dismantle most complex cancers if early detection is only going to get

02:16:01.420 us so far. And ideally across four pillars. Yes, exactly. And different ones, right? So going back

02:16:06.760 to the PCP3. Yeah, you have a huge combination. Right. You might want to always be targeting tumor

02:16:10.640 suppressors. Yeah, sure. But that's not one drug. That's right. That's up and down the chain.

02:16:13.940 Right. And unique to certain patients' tumors. Now on the tumor suppressor side,

02:16:17.820 does CRISPR offer any role? Because if you fix a tumor suppressor gene, that seems more beneficial

02:16:24.040 than just fixing an oncogene. This is gene therapy. You need to be able to deliver that to every last

02:16:29.740 cancer cell. And how in the world do we do that? Is there a virus that can do this? No, no, not

02:16:34.280 currently. And I, in the beginning of my career, because if I've communicated nothing in talking with

02:16:39.400 you, I am an optimist. The beginning of my career, I absolutely thought like this is, on one hand,

02:16:44.280 we have these things that we need to inhibit. And on the other hand, thank God, we're going to have

02:16:47.800 gene therapy. 20 years later, we still don't have gene therapy. We have gene therapy that can correct

02:16:52.560 if you need only a little bit. We're a little at Penn when Jesse-

02:16:55.460 I got there right after. Okay. Yeah.

02:16:57.080 Yeah. So I was still to Brigham at the time when I was postulating that gene therapy was going to help

02:17:01.740 us on the tumor suppressor side. And yet friends of mine who were in the lab doing work on genetic

02:17:07.300 delivery methods were saying, no, no, we're not. We're not there. These adenoviruses are getting wiped

02:17:11.680 out. You can't get persistent expression. Hemophilia then to a degree now was kind of

02:17:16.080 paradigm case where people were saying, well, we just need to get these clotting factors expressed

02:17:19.700 to a little degree in a small fraction of cells and we'll be okay. As people were trying to just

02:17:24.440 climb that hill, I was like, well, that's not the cancer hill. The cancer hill is every single cell

02:17:28.820 widely distributed. And some of these cells are dormant. These micromats, they've lodged in distant

02:17:33.760 sites and they are truly quiescent dormant cells. How are you going to get integration into that cell?

02:17:38.420 So it's a major, major problem. So I don't see that trick coming in the foreseeable future. So I

02:17:43.780 don't focus on it. I do think we can understand the pillars up and down, as you said, and knock them

02:17:49.500 out in ways that potentially at least turn off what's turned on as a consequence. I mean, this is

02:17:55.100 a thing about tumor suppressor genes that are eliminated is that you can always find a downstream

02:17:59.680 thing that's turned on as a consequence. So the issue is, can you find that point of

02:18:03.480 drug ability intervention to counter that? So that's at least until I retire, I think that's

02:18:08.740 going to be the relevant approach. You've alluded to it already, but are you pretty bullish on liquid

02:18:12.440 biopsies? For sure. There's several reasons why. The early detection piece, I think we've touched

02:18:17.100 a little bit on. Let's tell folks what they are. I use the term from time to time. Right. So tumor

02:18:21.780 cells, a primary tumor with no, let's go with the notion that we can have knowledge that there is no

02:18:27.360 microscopic metastatic disease. It's just a primary tumor. Right. You have a tumor in your

02:18:32.320 colon. You have an adenocarcinoma in your colon, and we know that it's nowhere else.

02:18:36.600 And that thing will shed its genetic contents. DNA and RNA, which mutated genes will have their 0.99

02:18:43.560 mutated RNA versions. You have mutated DNA and RNA kind of detection opportunities, if you will.

02:18:49.320 So DNA is shed, we think, from lots of cell types, but definitely cancer cells seem to

02:18:54.720 disproportionately shed their DNA, which is chewed up a bit as it circulates in the blood,

02:18:59.300 but you can find it. And with higher and higher resolution technologies, if you find a single

02:19:04.160 copy even and can amplify that up and detect it as a signal, since we know what the genetic map is

02:19:09.780 of all cancer, we can have those probes reasonably well in hand. And a lot of acceleration has happened

02:19:15.540 in this space the past just few years now. I'll come back to my complaint on the impediment side

02:19:20.680 when it comes to reimbursement of diagnostic tests, because that's the problem I think we face there.

02:19:24.480 In reality. But from a public good perspective and from a science perspective.

02:19:27.380 Yeah, just, I mean, I'm just curious right now from a purely, like I'm thinking about this through the

02:19:32.140 lens of cancer screening at the population level. And again, I'm always trying to make the problem

02:19:37.400 simpler by taking away one constraint, which is cost. Because in the short term, I think you have to

02:19:42.020 start, if you try to solve this problem at a quality adjusted life year perspective, one, you're

02:19:47.960 inserting your morality into the discussion. And two, I don't know what the number is, and it's too hard.

02:19:52.140 So let's make the problem simpler. All we're doing is trying to reduce the risk of physical

02:19:56.600 harm to a patient and psychological harm through false positives. If you could layer liquid biopsy

02:20:02.660 on top of diffusion-weighted imaging, MRI, the best colonoscopy, the best mammography,

02:20:08.860 boop, boop, boop, boop, boop, boop, boop. I mean, you can theoretically construct a point

02:20:12.080 where you can catch every cancer prior to its clinical presentation.

02:20:17.480 Yeah. Let's just go with breast, colon, prostate, and lung, since that's 80% of

02:20:20.940 solid tumor cancer deaths. So that would be a good goal just to go after those.

02:20:25.940 So the challenges are, so again, the DNA is shed.

02:20:28.480 Do we have a sense of how big a tumor needs to be to even start shedding?

02:20:32.780 That has not been well mapped out, but certainly less than a centimeter. And this,

02:20:36.080 you mentioned centimeter before, is it radiographically findable and so on, nodule.

02:20:39.800 So if you go map out cure rates with surgery alone across common cancers, breast, colon,

02:20:45.620 prostate, and lung, at centimeter...

02:20:47.660 Right. At centimeter below, the exceptions are few. I mean, pancreatic would still be one of the 0.75

02:20:53.000 few where even sub-centimeter, your odds of survival are less than 50-50.

02:20:56.760 Melanoma happens to be measured in literally one millimeter increments, so huge metastatic

02:21:00.560 potential. But in any case, we have other strategies for that superficial tumor to find

02:21:04.580 it early. But in any case, so coming back to your point...

02:21:07.320 I see your point, which is, if you could decide no one ever shows up with a tumor bigger than

02:21:11.680 one centimeter, it's potentially doubled cancer survival, right?

02:21:16.000 So those guys are definitely shedding, so it should be detectable. If we need to use

02:21:20.700 microvesicles or exosomes, which have RNA and some DNA in them as a way of being able to find

02:21:25.800 scarce entities, these mutated gene products that are now, in this case, in the case of exosomes,

02:21:31.560 why they're being transmitted around the body, we don't know, but it happens.

02:21:34.660 Not just tumors, of course. Normal cells do this too. So circulating tumor DNA, exosome or

02:21:39.860 microvesicle packages of nucleotides, these opportunities for detection, as well as circulating

02:21:45.820 tumor cells, but circulating tumor cells are based on available technologies. They're the

02:21:49.300 hardest to find in a patient who has a one centimeter tumor nodule, using that as a threshold.

02:21:54.440 So already it's in sight that we're going to have high resolution methods for shed DNA and probably

02:22:00.060 same for exosomes and circulating tumor cells, I think could get there. The issue, to come back

02:22:05.280 to your framing here of like false positives and anxiety provoking and so on, remember P53 mutations,

02:22:12.200 50% of cancer. Finding a P53 mutation in blood, what does that tell you? A, you might have cancer,

02:22:17.780 you may not have cancer. And where is it? What part of the body is it coming from? Well, P53,

02:22:21.960 50% of cancer, it could be virtually.

02:22:23.960 Yeah. So how much of the effort in the liquid biopsy is going to be histology specific?

02:22:28.540 Exactly. This is where circulating tumor cells would be the favored approach.

02:22:32.080 The DNA will not offer tissue specificity.

02:22:34.540 You can get the whole fingerprint out of a circulating tumor cell and know exactly what

02:22:37.820 the entity is that you're hunting for now in terms of organ type. And so you just survey that

02:22:42.400 one organ with every imaging methodology yet to come. So there's reason to be hopeful there going

02:22:47.600 from CTCs forward. What's really cool in the circulating tumor DNA and exosomal RNA field

02:22:55.260 is lineage mapping. So it turns out basically that the genetic blueprint, when it's folded up

02:23:00.420 in a colorectal villus cell of the colon, the epigenetic alterations that occur in that cell

02:23:07.820 are characteristic. So you can fingerprint cell of origin by looking at epigenetic marks. This is a

02:23:13.040 realization that's now being ported into the blood detection world so that you can actually map

02:23:18.640 where did that fragment of DNA or RNA come from. So this is work in progress as we speak. There's

02:23:25.260 an ongoing collaboration between our group and the Broad Institute at MIT on this topic, but I know of

02:23:30.180 other groups who are in this space. So there's a convergence of sort of cell of origin fingerprinting

02:23:35.260 with genetic detection methodology that you could readily see how we could get there

02:23:40.060 in the foreseeable future. This idea of being able to say, okay, our imaging, it still isn't picking

02:23:44.760 it up, but we know that you've got a breast cancer brewing. And now- 1.00

02:23:47.940 And in some cancers, for example, if you took a woman that was high risk to begin with,

02:23:52.520 and now you had DNA proof that she had cancer, I suspect there are some women who would actually 1.00

02:23:57.480 just say without additional imaging, I might be willing to just undergo a mastectomy. Or a guy would 0.54

02:24:01.960 say, look, I'm willing to undergo a prostatectomy or whatever. Now it gets harder with lung cancer.

02:24:06.100 You can't have just bilateral lobectomies all day long or pancreatic even, but that's a step in

02:24:11.220 the right direction. Because as you said, maybe at that point we could justify, even at the societal

02:24:15.860 level, because certainly at the individual level, you'll do anything. But at the societal level,

02:24:19.340 we'd say, look, once you have that DNA test that points to that tissue, we're going all out on

02:24:24.340 scrutinizing the tissue.

02:24:25.220 And your first instinct is to think down a potential upcoming surgical path. If not, if you can't see it

02:24:30.580 now, you don't know where to cut, but eventually you will. But I'm actually, as a medical oncologist,

02:24:34.420 just leaping to the systemic therapy concept, can we motivate an immune response against that thing

02:24:39.140 when we know enough about its genetic fingerprint or an oncogene-targeted therapy and monitor the

02:24:44.420 response using the same method, right? So when you were talking about blood biopsy, a huge near-term

02:24:49.620 opportunity is to use it as a therapeutic monitoring tool to understand, can you actually

02:24:53.480 get people down to a minimal residual disease, no detectable evidence anymore by this blood method?

02:24:59.680 So yes, it's an early detection tool, but it's also a therapeutic monitoring tool. So you can

02:25:04.000 take a patient, you don't know where their breast cancer is, but you don't wait for it to emerge.

02:25:08.240 You attack it when it's still genetically simple and monitor that effect in blood. So you know when

02:25:12.820 to stop treatment or you know when treatment's not proving effective and switch gears to something

02:25:17.380 else because you've got a toolbox with different regimens in it.

02:25:21.120 I want to switch gears for a minute. There's so many just sort of cancer questions that I get asked all

02:25:26.300 the time that I don't know the answer to. So I have at least two patients who, I guess,

02:25:32.820 somewhat against my recommendation are adamant that stem cells, intravenous stem cell therapy has

02:25:39.100 played an enormous role in the improvement of their health. One patient in particular, there's no

02:25:45.160 denying that his symptoms improved dramatically with IV stem cell therapy. And I don't want to

02:25:50.000 represent that I have the hubris to suggest that it's not working. I just don't know. My bigger concern

02:25:54.860 because putting the cost aside and the immediate risk like infections and sort of risks of the

02:26:02.560 therapy, which are non-trivial, of course, the risk on the other side, which is it's too successful

02:26:07.400 in a sense, and those stem cells acquire a life of their own. So is this something you spent much

02:26:11.440 time thinking about? I've thought about it. I just think that if they're wild type stem cells,

02:26:16.580 I think they'll follow the rules. So this whole discussion has been about cells that basically,

02:26:21.660 not through anthropomorphic spiritual means, but by random genetic alteration basically are able to

02:26:29.400 not follow the rules. I mean, this is this revolution that happens inside of our bodies

02:26:33.160 that is cancer. It's actually, I usually use the term evolution. This is individual cells following

02:26:38.560 the preservation and success, organismal success instincts that allowed us to crawl out of the

02:26:44.520 swamp in the first place and become the complex organisms that we are. This drive to continue

02:26:49.040 evolving for selfish purpose. So using that mindset to come back to your question,

02:26:55.060 I think a truly wild type stem cell will find its home and find its niche and its proper influences

02:27:00.700 and behave accordingly is my assumption. Unless it were to become non-wild type through external

02:27:06.800 manipulation outside the body when it's potentially fragile and capable of picking up aberrations that

02:27:11.700 aren't corrected. But stem cells are notably hardy cells that can survive insults remarkably well and

02:27:19.520 correct. So a cell that survives a genetic insult, a stem cell is particularly capable of detecting the

02:27:27.000 damage, repairing the damage, taking its sweet time in doing so, and not spawning daughter cells until

02:27:31.960 it's got its house back in order. So its P53 detection program and DNA repair machinery is particularly

02:27:38.780 robust. So this is why when people talk about cancer stem cells, notably, which no one would want

02:27:43.700 those, cancer stem cells are thought to be kind of the worst of the worst. They're these primordial

02:27:48.800 cells in the cancer cell population within a given tumor that have found their way back in development

02:27:54.620 towards a stem cell and in doing so have adopted these really hardy skills, survival skills. No,

02:28:02.700 they can't proliferate and divide very quickly, so they don't have that, but they can survive almost

02:28:07.100 anything. Whereas the avant-garde, highly proliferative, very dynamic population that will

02:28:12.100 multiply and actually be the life-threatening leading edge of a cancer, those things are more

02:28:16.520 vulnerable. So actually, if you look at what chemotherapy has done for us, it's fairly conventional

02:28:22.200 chemotherapy. It's fairly clear that actually what it can do is it can prune this highly proliferative

02:28:27.880 avant-garde population, leaving behind this more quote-unquote stem cell-like pool. And there's a whole

02:28:33.560 field, it's a contentious field to a degree in cancer biology slash therapeutics of how much true

02:28:39.700 stem cell biology really exists there. But you brought it up more from the perspective of whether

02:28:43.540 stem cells themselves could go rogue. And I would say, unless they're perturbed in some significant

02:28:49.060 and ultimately genetic way, I don't see how. Now, let's talk about another fundamental cancer

02:28:52.940 question that serves no real purpose other than unless there's some clinical application of

02:28:57.660 prevention, I suppose. It's impossible to deny the age-related association between age and cancer.

02:29:05.760 Certainly, kids can get cancer, but for the most part, cancer rates rise monotonically until the

02:29:11.840 ninth decade or so. I mean, there's a bimodal distribution because you've got the childhood 1.00

02:29:15.040 cancers. If you're 50 versus 60 versus 70, you leapfrog up in risk, right?

02:29:20.920 Almost exponentially, or at least quadratically. It's not just linear.

02:29:23.960 Consider two hypotheses. One is, as we get older, our genome is more susceptible to injury. Consider

02:29:31.200 three. Two, as we get older and we've accumulated more of these, the probability that they can start

02:29:38.880 to layer and stack and you get the phenotype that's not advantaged. Three, the immune system

02:29:46.220 loses some of its steam and therefore the radar window in which you can detect cancer

02:29:53.540 narrows, basically. Or the ability to detect all of the above, one of the above.

02:29:59.720 All of the above. You've just described very nicely what I call the inevitability of cancer. We've got

02:30:05.620 too many stochastic events accumulating and surveillance systems that are breaking down. DNA

02:30:11.020 damage repair and immunologic. Those are the two fundamental, most important components. And if

02:30:17.620 you're allowed to go one log shift in mutation burden per cell and the immune system not see it

02:30:24.500 and not clear it, that's it. I mean, there's no human that...

02:30:27.220 There's no immune system that's going to...

02:30:28.620 That's going to overcome that. So when I say inevitability, you connect the dots in terms of

02:30:33.340 the per decade acceleration and appearance of cancers across the population. And I oftentimes say,

02:30:39.220 if we all lived to 130, we'd all have a cancer, quote unquote, real cancer. Let's not get hung up

02:30:43.620 on benign growths that aren't cancerous. The definition of cancer for a totally lay person

02:30:48.840 who doesn't think about cancer much is ability to travel and actually wreak havoc and kill. Yes,

02:30:53.840 glioblastoma kills in its local site. But benign tumors, which are almost cancers, yes,

02:30:59.340 those happen broadly across the population. There's autopsy studies that show at least 50% of people die

02:31:03.640 with at least a benign tumor. So almost getting there, sure, that happens already. But I'm talking

02:31:09.400 about fully getting there, a full-blown and will be eventually life-threatening cancer.

02:31:15.280 It's interesting. Actuarially, by your 10th decade, your risk of cancer starts to go down. Although,

02:31:21.320 and I discuss this with my patients a lot, my explanation for that is that atherosclerosis

02:31:26.200 ratchets up faster than cancer. And it's not that cancer is going down, it's that heart disease

02:31:31.540 is dominating. And that's my, I say the same thing to patients that you do, which is whenever I get

02:31:36.520 asked these questions that frankly kind of annoy me, which is like, aren't we just going to figure

02:31:41.980 out a way to all live to be 200 one day? And I'm sort of like, no, what are you talking about?

02:31:46.980 You have to figure out a way to prevent age-related disease. And yes, there's lots of cool ideas. And

02:31:53.780 oh, what if you could just maintain telomere length is like one I love hearing. And it's like,

02:31:58.340 that's totally irrelevant because even if there were some benefit that came from that,

02:32:03.900 you have to thwart atherosclerosis and cancer.

02:32:07.280 Yeah. You don't want a proto-cancer cell to be getting more telomere length.

02:32:10.840 Yeah, yeah, yeah. Not even withstanding.

02:32:12.160 Going back to your stem cell comment, that's not a fix-all by any stretch. No, you're right. These

02:32:16.040 are intersecting risk issues. So you're right that this competing risk issue makes it look as though

02:32:21.660 if you make it long enough and you don't have cancer, then you're just not cancer prone.

02:32:25.800 It's a bad problem to have when your risk of atherosclerosis is so dominant over your cancer

02:32:30.800 risk. Now, one last thing I want to talk about is melanoma. And again, I don't actually know,

02:32:36.120 I haven't written anything you've read on the topic we're about to discuss. So feel free to

02:32:39.580 just dismiss it out of hand and say, I don't pay attention to that literature at all. But

02:32:42.700 have you been following any of the vitamin D melanoma sun exposure discussion lately?

02:32:48.140 Yeah, sure.

02:32:48.740 Okay, so I'll do my best to give a relatively brief synopsis and then feel free to just correct it

02:32:55.620 because I'm pretty sure I'm bastardizing it. There's not a huge body of descent to the notion

02:33:01.240 that sun exposure increases the risk of melanoma. We know that...

02:33:04.800 Not linear, but there's no question that there's an association there.

02:33:07.960 Okay, so we won't call it, it's not an axiomatic statement, but it's the evidence that sun exposure

02:33:13.060 and melanoma are associated in a causal way from sun to melanoma is strong. Okay. We also know that

02:33:20.080 sun exposure increases the de novo synthesis of vitamin D in the human body. We know that from

02:33:27.480 an association perspective, higher levels of vitamin D port with good things happening and low levels of

02:33:34.600 vitamin D port with bad things happening. This has led people to suggest that we should be supplementing

02:33:39.900 with vitamin D because yes, you can get it by being out in the sun, but the risk of melanoma is going

02:33:46.260 up, but we can just take it. It's a fat soluble vitamin. It's easy to administer. You can clearly

02:33:51.600 achieve levels in the plasma that mirror that of the sun. That should solve the problems. So many

02:33:57.100 doctors, myself included, have historically normalized patients' vitamin D levels. Now, again,

02:34:02.460 there's a whole spectrum in there. There's a bunch of weirdos that think you have to have super

02:34:05.900 normal levels, but most of us walk around thinking, God, if somebody shows up in their vitamin

02:34:09.900 D level is 20, we want it to be 40 or 50 or 60. The problem is these clinical trials, these pesky

02:34:15.880 little things keep showing up demonstrating that supplemental vitamin D doesn't really seem to help

02:34:21.100 that much. And I've scrutinized many of these trials and I can poke holes in all of them, but

02:34:27.080 the body of evidence is becoming hard to ignore. So any one of these trials, I can say, look, they didn't

02:34:33.660 use enough vitamin D or they didn't look long enough or they didn't have the right patient

02:34:39.660 population or, but when you have, I don't know what X is, but when you have X studies that are

02:34:44.780 basically all saying the same thing, which is give everybody vitamin D, it doesn't seem to matter.

02:34:48.000 Now there's exceptions. You know, JAMA had a paper a week ago about supplemental vitamin D in patients

02:34:52.720 with colon cancer, and there actually seemed to be a legitimate benefit. I'd love to hear your thoughts

02:34:55.940 on that. Where did this leave us all? This left us all, I don't know, for many of us, about six

02:34:59.920 months ago, there was a huge blitzkriek of just nonstop information being put out about this that

02:35:06.400 came to the suggestion as follows. The vitamin D that you take supplementally ain't fixing the problem.

02:35:12.360 The vitamin D that's getting fixed in the sun is a proxy for things that are happening good in the sun.

02:35:17.600 So it's sort of like saying gray hair is a proxy of aging, dyeing your hair black or blonde,

02:35:25.160 isn't fixing your aging. So get out in the sun. And oh, by the way, yes, your risk of melanoma will

02:35:33.280 go up slightly. We're not denying that, but it goes up at one eighth, I believe is the number,

02:35:38.760 your rate of decline of all of these other cardiometabolic diseases. So should we, or should

02:35:46.000 we not take vitamin D question one, or more importantly, I think for you question two, given

02:35:50.520 that you're one of the world's experts on melanoma, would you advocate we spend more

02:35:55.100 time in the sun as a way to get our vitamin D? No, I don't go that far. I'm recognizing all of

02:36:00.100 the cross currents in data. I thought where you were going to finish up, which is what you just said

02:36:05.000 to say, exercise, exercise, and yet again, exercise. Okay. Are there people currently who exercise a ton

02:36:13.640 get a lot of cardiovascular benefit from that and do all that indoors? Yeah, that exists. So shouldn't

02:36:20.100 we be able to pick that up in the data? No, I don't think this is a classic example of a confounder.

02:36:24.080 I don't think that's extricated from the data. What I'm hearing you say is you think that the

02:36:28.820 huge confounder is all these people outside are exercising. That's what's been giving them the

02:36:33.420 benefit. It's not the sun per se or the vitamin D per se. I'm going with the vitamin D being

02:36:37.900 beneficial. I'm going to come back to a reinforcing element of that argument, at least as I see it in

02:36:43.600 scientific evidence. What I'm getting at is that the most powerful benefit, you said good things are

02:36:47.920 happening in the sun, or I think I'm paraphrasing what you said. And what I'm saying is that good thing

02:36:52.040 that's happening is exercise. I face this with my patients, melanoma survivors, all the time.

02:36:56.980 I love to- Ride my bike.

02:36:59.040 Yeah. I'm a swimmer. I'm a runner. This is what I love to do. Is that okay? I've had a melanoma

02:37:04.540 diagnosis. I have a 10% lifelong risk of developing a new primary melanoma. So you, Dr. Flaherty,

02:37:09.580 have told me, I want to minimize that, but this is how I get my exercise. This is the conversation we

02:37:14.760 have. I say, I get it. If I can talk you into a more sun-safe version of exercise, as in early

02:37:20.980 morning slash evening running, swimming, tennis, I'm going to try to talk you into that. Stay out

02:37:26.120 of the sun in the middle of the day. When now you're pursuing the cardiovascular benefit of

02:37:29.840 exercise, you're taking on more vitamin D than you need, and you're taking on an inordinate risk in

02:37:34.260 terms of skin cancer risk. Obviously, other skin cancers are more common than melanoma, but not as

02:37:38.140 life-threatening. So we usually frame this discussion around global skin cancer risk, where it's more

02:37:42.700 linear with squamous cell and basal cell, the relationship between sun exposure. In melanoma,

02:37:46.920 it's not as linear, but it's clearly causally related. So I synthesize all of the available

02:37:51.300 evidence to say there's confounders in the behaviors that relate to sun exposure. Vitamin D is still,

02:37:57.460 I think, there's enough strength to say vitamin D. There's obviously strong associations, which is

02:38:01.920 where you were starting with. I mean, the epidemiology regarding vitamin D levels and risk,

02:38:05.740 colorectal cancer was one of the first to come, but then other cancers undeniably related to low

02:38:11.060 vitamin D levels. So I think there is a causal relationship. I think you want to serve higher

02:38:15.400 up on the vitamin D curve. The issue is how you get there. My synthesis is you exercise. And if

02:38:20.720 your exercise means that you get a low enough UV exposure that you need supplementation, then

02:38:26.300 supplement. And until proven otherwise, it's those two factors that are why pure vitamin D supplementation

02:38:33.220 alone is never going to do it in a population where you're not controlling for their exercise.

02:38:39.000 Now you're dealing with a highly motivated population, I think, in general. So where

02:38:42.920 they're inclined to be doing positive health behaviors. So they're probably already tackling

02:38:47.000 the exercise piece. When you look at these studies, you're looking at a different population of people,

02:38:51.860 in my view. So let me just throw one little bit of science at you. It's a nature paper published by

02:38:56.880 my very close colleague at Mass General, David Fisher, who's one of the preeminent melanoma biologists

02:39:01.560 in the field. So he came across this link in terms of the addiction potential of sun exposure.

02:39:09.000 And it comes from the fact that melanocortin, which is the, if you will, the growth factor for

02:39:13.960 melanocytes in terms of their daily function as well as their development, comes from propiomelanocortin

02:39:20.500 produced in the anterior pituitary. And that basically a component of that gene product or this

02:39:26.800 hormone as the pro-hormone is cleaved into its hormones is endorphin. Endorphin is produced and

02:39:32.060 released in addition to melanocortin at equal stoichiometric quantities.

02:39:36.020 So it was that simple bit of endocrinology that led him to wonder, is there something about sun

02:39:42.280 seeking behavior that's wired into us to try to get us to go out and get vitamin D exposure?

02:39:48.360 You have to go back to evolutionary pressure times. Like, was there a reason why human beings

02:39:52.240 would rather have not gone out and expose themselves to the sun because predatory risk,

02:39:57.520 presumably? So that's a cute argument, but the wiring is very clear. You can show that mice will

02:40:04.000 basically prefer sun exposure. We'll go through narcotic-like withdrawal when you deprive them

02:40:09.820 of their sun exposure. So it's meeting the criteria of addiction the way we would see it with cocaine,

02:40:16.460 for example, in a mouse. Exactly. So it's of lower grade to a degree, but it's anyway,

02:40:21.300 this fascinating initial paper in Nature and a couple follow-on papers since then that kind of

02:40:25.820 reinforces this kind of circuit. This idea that basically this is even in a lower species like 0.63

02:40:31.100 mice, you can even detect this sun-seeking behavior that humans presumably have as well

02:40:36.380 toward the end of getting vitamin D. So here's an argument to say, well, we're engineered to get

02:40:40.760 out in the sun and there must be benefits from it. And the issue I think that you posed is how much of

02:40:46.140 the benefit is vitamin D? Then the shakiness of that argument is that, well, you supplement vitamin D,

02:40:51.120 you don't seem to get all of the benefit one would think. Anyway, what I'm getting at is that,

02:40:55.240 no, no, I think this is a behavior that is maladapted for those who are going to live to be 60,

02:41:01.420 70, 80, a hundred years old. Not maladaptive for people who are going to live to be 30 and

02:41:06.040 die from predatory death or whatever the issue back in evolutionary times.

02:41:10.260 Yeah, evolution would have placed absolutely zero pressure on preventing you from getting melanoma.

02:41:15.460 That's right.

02:41:15.800 So from a thought experiment perspective, if you could exercise indoors, never see the sunlight except

02:41:24.040 through your window and supplement vitamin D, do you think you are doing as well health-wise

02:41:30.180 as someone who does not supplement vitamin D, exercises outdoors, and attains the same vitamin

02:41:36.340 D level?

02:41:37.160 That's my hypothesis.

02:41:37.880 That's your hypothesis. Those are the same.

02:41:39.100 Yeah. I look at all available evidence in each study as it comes out and I take that,

02:41:43.340 I have that long-standing hypothesis, put that filter on and I say-

02:41:46.320 Does this study contradict that hypothesis?

02:41:48.560 So the idea that there's some other magical component of UV exposure, presumably UV being that

02:41:54.920 basically that there's some other magical property, I'd say we lack the evidence to support that.

02:41:59.320 So I guess last thing I want to ask you about, because in many ways this interview, Keith,

02:42:03.860 has been great because I think there are tons of scientists, scientists in training that I know

02:42:08.840 enjoy this podcast and I think you've offered incredible advice to people across different

02:42:12.800 levels of that spectrum. In particular, those people reaching out who are in MDA, PhD programs

02:42:17.780 and they're trying to straddle how do I balance research and clinical work? And though you are an MD,

02:42:24.300 PhD, not an MD, PhD, you're from a functional standpoint, you are an MD, PhD. You are a

02:42:28.320 clinician who predominantly does research. I want to ask you another question along that thread is

02:42:33.160 you also do something that a lot of people haven't figured out how to do. And I'm curious as to what

02:42:38.320 you've learned along the way, which is you have a very rigorous academic, completely standalone

02:42:46.200 credentialed existence. And at the same time, you really understand industry. You've been a huge part

02:42:52.500 of industry and you seem to run these two parallel tracks that virtually, I don't want to say nobody,

02:42:59.120 but most people really struggle to live on both sides of that. They usually tend to be very good

02:43:05.300 at one and kind of mediocre at the other. I would add to that. I think people are cautious to believe

02:43:12.380 that it would be a good idea to do it, that there'd be synergy in trying to live on both sides.

02:43:18.080 That's been the lesson I've learned over the past, let's say six years anyway, in earnest,

02:43:22.960 which is the dates back to when I co-founded my first company of a series now of five and this

02:43:28.540 year working on a couple more. Is that because of what you said at the outset, which is you were

02:43:33.840 immediately focused on translational stuff. And if you were focused on basic science, for example,

02:43:38.960 that synergy is a lot harder. Is it because of where the intersection you sit?

02:43:42.600 I mean, I should back up at step and say that you can't do the work that I do and not interact

02:43:46.520 with industry. So my entire career, the drugs come from industry. We do not live in a world yet

02:43:51.320 where academic entities or the National Institutes of Health produce investigational agents, take them

02:43:56.820 through their measures and then make them widely available. That could happen. Actually, there's a

02:44:01.560 part of me that has pushed and advocated for that day to come because I think it would help bend the

02:44:05.640 cost curve in a major way. But park that one for a moment. That hasn't been the world I've operated in.

02:44:09.780 I depend on pharma, which means big companies and biotechs to be aligned partners, at least for much

02:44:16.740 of the time that we work together. The misalignment totally comes. And when it comes, we shake hands and

02:44:22.560 part company. But my world has revolved around the fact that therapies come from companies,

02:44:28.580 proto therapies, and then true therapies. Diagnostics kind of come from us and diagnostics are 50% of the

02:44:34.820 equation. They're woefully undervalued. Yet a lot of our research and NIH funded research is about

02:44:40.460 quote unquote biomarkers that are in their best form going to become diagnostics.

02:44:44.340 I mean, this is the world's worst industry.

02:44:46.460 I get that.

02:44:46.760 But that's the point.

02:44:47.280 But you can't direct a patient to therapy and use the phrase precision medicine.

02:44:51.520 Without pairing it with the best diagnostic.

02:44:53.400 So this isn't about diagnosing colon cancer. This is about diagnosing the molecular type and what

02:44:57.900 its therapeutic vulnerabilities are going to be. And that could be ex vivo functional diagnostics,

02:45:02.320 not just static diagnostics.

02:45:03.380 Shouldn't that just be absorbed into the therapeutic world?

02:45:05.760 Right. Not just the cost borne by the therapeutic world, but also the development and deployment

02:45:10.320 as well. And that's where it works. Companion diagnostics, quote unquote, are the success examples.

02:45:15.580 It will still be the case that that therapy developing company marketed diagnostic and therapeutic

02:45:20.580 world will make 10 cents off the diagnostic, a billion dollars off the therapeutic. But if that's

02:45:24.720 what it took to get there, then they will do it. They will also give the test away rather than

02:45:29.820 bothering with the 10 cents. They will say, well, we readily recognize that we can't protect this

02:45:34.320 diagnostic space. Yes, we have IP in this very specific diagnostic method, but we're going to

02:45:38.820 get undercut by 10 to 100 others who are just going to come in and compete us. That's fine because we

02:45:44.120 still own the therapy. And especially in a regulated environment, if certain companion diagnostics

02:45:49.260 only link to that therapy, then the therapy developer still benefits.

02:45:52.760 The problem that I faced in my career, this is now taking a little bit of a detour from your first

02:45:58.000 question, but I'm going to come back to it. The problem that I faced as an academic is the issue

02:46:02.220 of alignment slash misalignment around diagnostic therapy pairs. If you get to market in a broad

02:46:08.080 population that is not biomarker selected, and you can get there, you know you're treating some

02:46:12.320 patients who aren't getting benefit from therapy. You've seen it in phase one slash two. You're still

02:46:16.340 seeing it in phase three, but you're having enough of an effect in enough of a subpopulation that

02:46:20.580 you're able to drive a result in an unselected population. If you can do that, that remains in

02:46:25.920 the current environment, the best version of success there is. It decreases the degree of

02:46:30.220 difficulty in terms of having two moving parts in your development plan, both a diagnostic and a

02:46:34.400 therapeutic, and you get the broadest population. Admittedly, patients who aren't getting benefit from

02:46:38.680 therapy aren't going to make a lot of money for you because in a cancer context, they're going to get

02:46:42.300 two months of treatment and stop. But still, broad population, broad use. A clinician doesn't have to

02:46:47.920 think about doing more testing than they've already done. They just give this colon cancer patient the

02:46:52.800 EGFR antibody, go back to the mid-2000s when that was the reality. A light bulb went off over hundreds

02:46:59.580 of people's heads saying downstream of the epidermal growth factor receptor is RAS, and in colorectal

02:47:05.440 cancer, RAS is mutated in a substantial fraction of patients. I bet that if you have a RAS mutation

02:47:12.480 downstream of the receptor, that tumor is not going to care about having the receptor blocked with the

02:47:17.560 antibody. Brief run through the lab confirmed that hypothesis. An attempt then to carve out

02:47:23.660 the RAS mutant population from an approved epidermal growth factor receptor antibody was a ground war

02:47:28.880 that took years and years. The number of constituents, first the drug developer themselves and regulators

02:47:35.680 joining them, who stood in the way of the most obvious scientific hypothesis I've ever seen in my

02:47:40.840 academic career, it was terrible. So this is what I'm getting about misalignment, is that basically,

02:47:45.720 if you do this too late and you're now carving out populations from an approved therapy, disaster.

02:47:51.860 Because then you're relying on the diagnostic to be the tool, the business incentive generating tool

02:47:57.360 to carve that population out, and it doesn't exist. Now, if we had single payer healthcare and

02:48:02.140 we had one ecosystem constituents talking about this, we could talk about the savings and not having

02:48:08.380 patients get ineffective therapy, and we could align on who takes ownership and responsibility. But that's

02:48:13.220 not the world we live in. So in the meantime, it's a mad panic as an academic to try to translate the

02:48:18.620 science to medicine with a diagnostic and therapeutic pair prospectively. So you get there at the finish

02:48:24.000 line with a heroically effective therapy that's benefiting 80 plus percent of patients. You're always

02:48:28.160 going to be wrong in some fraction, let's say. But you're almost nailing it in terms of assigning

02:48:34.120 patient in a true precision medicine way. If you can do it prospectively, which is a mad panic,

02:48:38.760 because cancer drug development has moved more quickly in recent years, you've got to now develop

02:48:43.660 that diagnostic, show that your biomarker positive, biomarker negative population do and don't

02:48:48.500 benefit to the satisfaction of FDA in time to then have that diagnostic locked down, ready to roll with

02:48:55.020 its approval supporting data set. So this is a long-winded way of saying that I've traveled this path

02:49:01.160 enough times as a collaborator with industry and seen it go well and go poorly. I've also seen lots of

02:49:08.320 decisions made in drug development in terms of compromises in chemistry, shortcuts that are

02:49:13.620 driven by just artificial deadlines. Like, look, this program's got to advance or we're going to

02:49:18.560 kill it either in a young company with venture capital backing or in a big pharma company that

02:49:22.580 says, look, we've got a huge pipeline. We've got teams competing against each other. We've got to

02:49:27.220 win this pipeline down to best candidates. So we're going to make some decisions next quarter.

02:49:31.900 So you show me what you've got in terms of chemistry advances towards your development candidate

02:49:36.600 and we're just going to make a decision. I don't care if you think you've in another quarter you

02:49:40.660 could do better. We're just drawing a line and making a deadline. That's the big company version.

02:49:44.780 The venture investors draw different timelines in the small company context. I watch these things

02:49:50.480 happen as an external collaborator for enough times and different compromises and what I thought

02:49:57.000 bad decisions being made to say, I think there's an opportunity here to do what I say the best of

02:50:02.380 academic medical work can be, which is to lead by example. Go into companies as a founder, meaning

02:50:08.540 with an idea, and try to do it the way that I think has the greatest integrity to the hypothesis

02:50:14.100 that you're trying to test. Not everything's going to work. That's not the point. It's just don't make

02:50:18.880 decisions for the wrong reasons. Make them for the right reasons that the hypothesis is basically not

02:50:23.600 held up and needs to be abandoned. You need to spend more time in optimization on the chemistry side.

02:50:29.300 Your regulatory path needs to be innovative because it turns out that next generation sequencing is or

02:50:35.440 isn't being adopted in the way that you thought it would be. I'm touching on some real examples here.

02:50:40.760 Being in the boardroom, being around the table, the smallest table where these decisions are made,

02:50:47.600 I had no idea how satisfying that would be. Do I suggest it for everybody? Well, but I would suggest

02:50:54.080 for everybody who cares about therapeutics-related research. Lab investigators, but particularly

02:51:00.080 clinical investigators, is you need to understand to the best of your ability how people's worlds turn.

02:51:07.040 You really need to develop that understanding. Otherwise, you cannot be an effective

02:51:11.420 constituent. You can't be an advocate within your field for your current patients, for the future

02:51:16.660 patients, next generation patients, patients worldwide who you're trying to advocate for.

02:51:20.420 You can't do that if you don't understand what's constraining on the other side. That has been

02:51:27.700 the lesson that I've learned. So the fact that I was approached and offered serially to take light

02:51:33.340 bulb moments and park them into companies, I couldn't have seen that coming eight years ago.

02:51:37.620 Clinical investigators didn't have those opportunities. Should they have those opportunities? Well,

02:51:41.540 you can imagine from what I've just said, absolutely. I mean, this is to me, I've got mentees

02:51:47.200 in my own group and mentees elsewhere in the country who I look at and say, oh, this field needs more

02:51:53.320 people like you around a board table. The thing that I've learned, and I hope that all of my co-board

02:51:58.300 members, if they heard this, would understand how much I'm saying this with a lot of respect and even

02:52:04.080 affection. These are super bright people who have accomplished in multiple dimensions. But if they are

02:52:10.100 not an MD or MD-PhD or straight PhD in biomedical research, let's say in cancer, there's a component

02:52:17.560 that they don't see, where if you're able to provide that as a single person around a board of

02:52:23.680 eight, that contribution can be disproportionate. Not on the financing side, not on domains that I'm

02:52:29.740 not very clever about and certainly not very experienced about. Learn to appreciate those

02:52:35.220 things. This is a seat that is currently not filled. I can only speak to cancer biomedical

02:52:40.240 research in the private sector. This is a seat that's not filled. It's empty. There have been

02:52:45.660 board seats held for basic investigators historically, but not the translational clinical investigator.

02:52:53.620 And to me, I think this is, I'm not looking for more work. So this isn't at all about me,

02:52:58.420 but I've seen that as cancer science has become so much more translational, the opportunity unfolded

02:53:06.140 for me. And I think it would be a major benefit to the field for the private sector basically to engage

02:53:12.860 more voices at the table of those who are involved in the truly applied science down to the patient

02:53:20.120 level. So you've got stakeholder representation in those discussions that includes the people who

02:53:25.700 actually, in the case of cancer, hold the hand of a dying cancer patient, understand what the

02:53:31.800 ramifications are of clinical trial decisions that are the most expedient path to approval versus the

02:53:37.520 let's make this really stick and matter for patients. Doing this through scientific advisory boards,

02:53:43.800 advisory boards, ad hoc consulting, which is the whole rest of my travels before I got into the

02:53:49.640 entrepreneurial mode of actually launching companies. Been there, done that. Those meetings are held.

02:53:55.200 They conclude and life goes on. You don't have a fiduciary responsibility.

02:54:00.260 That's exactly it. All of these companies, every single one of them, I mean, every company I've

02:54:05.120 interacted with in biotech slash pharma, they will all say that they're trying to move the needle for

02:54:10.280 patients, right? This is their stated goal and I believe them. How do you do that if you don't have

02:54:16.880 in the room, exactly as you say, invested responsible leaders of the company in the case of the board

02:54:22.960 who actually have career-long skin in that game? In retrospect, I don't get it, even though it wasn't

02:54:29.260 there 10 years ago when I wouldn't have imagined that these doors would begin to open.

02:54:33.680 Well, that's a great way to close this discussion, Keith. I hadn't actually thought of it through the

02:54:38.060 lens of the gap, the translational gap per se. Well, I want to thank you very much. You've been

02:54:42.660 incredibly gracious with your time and even more gracious with your insights. This is an episode where

02:54:48.020 I learned a lot along the way, which always makes it fun for me as well. Thank you very much.

02:54:52.360 Well, I appreciate the opportunity. Wouldn't be able to talk this long were it not for the fact

02:54:57.100 that you're pulling out of me all the talking points that I've used in different venues at

02:55:02.460 different times, but in one conversation. I think this is an incredibly exciting time in terms of

02:55:08.900 understanding the molecular underpinnings of health and disease. Cancer is incredibly anxiety-provoking.

02:55:15.260 It's actually why I got into the field, to be honest. It was the translating science to medicine

02:55:19.060 and the most disproportionately havoc-wreaking entity. And so I really appreciate opportunities,

02:55:26.740 this being a really unique one, to try to help people understand this seemingly impossible to

02:55:33.240 understand entity that is this kind of revolution within our bodies or betrayal within our bodies.

02:55:38.300 But we'll get there. The pace is quickening of progress. So I hope I've communicated that,

02:55:44.340 but I hope you circle back to revisit this topic with others in the field because you will see

02:55:49.660 year over year that the pace of progress will continue to advance.

02:55:52.980 That'd be great. Or maybe I'll just come back to Boston once a year and

02:55:55.460 I'll always time it in a nice time of year.

02:55:57.640 Sounds great.

02:55:58.200 All right. Thanks, Keith.

02:55:58.920 You can find all of this information and more at peteratiamd.com forward slash podcast.

02:56:06.460 There you'll find the show notes, readings, and links related to this episode.

02:56:10.560 You can also find my blog at peteratiamd.com. Maybe the simplest thing to do is to sign up for

02:56:16.040 my subjectively non-lame once a week email, where I'll update you on what I've been up to,

02:56:20.380 the most interesting papers I've read, and all things related to longevity, science, performance,

02:56:25.320 sleep, et cetera. On social, you can find me on Twitter, Instagram, and Facebook,

02:56:30.060 all with the ID peteratiamd. But usually Twitter is the best way to reach me to share your questions

02:56:35.080 and comments. Now for the obligatory disclaimer, this podcast is for general informational purposes

02:56:40.040 only and does not constitute the practice of medicine, nursing, or other professional healthcare

02:56:44.540 services, including the giving of medical advice. And note, no doctor-patient relationship is

02:56:50.020 formed. The use of this information and the materials linked to the podcast is at the user's

02:56:55.180 own risk. The content of this podcast is not intended to be a substitute for professional

02:56:59.320 medical advice, diagnoses, or treatment. Users should not disregard or delay in obtaining medical

02:57:04.820 advice for any medical condition they have and should seek the assistance of their healthcare

02:57:08.980 professionals for any such conditions. Lastly, and perhaps most importantly, I take conflicts of

02:57:14.760 interest very seriously. For all of my disclosures, the companies I invest in and or advise,

02:57:19.540 please visit peteratiamd.com forward slash about.

The Peter Attia Drive - July 15, 2019

#62 - Keith Flaherty, M.D.： Deep dive into cancer—History of oncology, novel approaches to treatment, and the exciting and hopeful future

Episode Stats

Summary

Transcript