The Peter Attia Drive - #270 ‒ Journal club with Andrew Huberman： metformin as a geroprotective drug, the power of belief, and how to read scientific papers

00:00:00.000 Hey, everyone. Welcome to the drive podcast. I'm your host, Peter Atiyah. This podcast,

00:00:16.580 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:21.580 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:26.780 wellness, and we've established a great team of analysts to make this happen. It is extremely

00:00:31.720 important to me to provide all of this content without relying on paid ads. To do this, our work

00:00:37.000 is made entirely possible by our members. And in return, we offer exclusive member only content

00:00:42.760 and benefits above and beyond what is available for free. If you want to take your knowledge of

00:00:47.980 this space to the next level, it's our goal to ensure members get back much more than the price

00:00:53.260 of the subscription. If you want to learn more about the benefits of our premium membership,

00:00:58.080 head over to peteratiyahmd.com forward slash subscribe. Welcome to a special episode of

00:01:06.620 the drive. This episode is actually a dual episode with Andrew Huberman, where we are going to be

00:01:11.040 releasing our conversation on both the Huberman lab podcast and on the drive. In this episode,

00:01:16.540 Andrew and I have a journal club where we each present and talk through a paper that we have

00:01:21.180 found interesting in the previous couple of months. Now, I hope this will help people not

00:01:25.600 only understand the results of the specific papers we go through, which is part of the exercise,

00:01:30.060 but also to give people an idea of how to read and interpret a paper that you might read. And really,

00:01:34.760 in some ways, I think that's equally, if not more important as part of this exercise.

00:01:38.680 For my paper, we looked at a study on metformin by Keyes et al., which looked back at the 2014

00:01:44.020 study by Bannister et al. that initially got everyone really interested in metformin as a possible

00:01:50.820 gyroprotective molecule. Through looking at this paper, we discussed metformin as a possible

00:01:56.620 gyroprotective drug, but also had a general discussion around gyroprotection and the current

00:02:01.620 lack of biomarkers of aging. Andrew then presented a paper that addressed how our beliefs of the drug

00:02:08.320 we take impacts the effect they have on us at a biological level. So not looking at placebo effects,

00:02:14.740 but actual belief effects, and what this could mean going forward. As a reminder, Andrew is an

00:02:20.960 associate professor of neurobiology at the Stanford University School of Medicine and the host of the

00:02:25.520 very popular Huberman Lab podcast. He's also a former podcast guest on episode 249. So without

00:02:32.460 further delay, please enjoy my conversation with Andrew Huberman.

00:02:36.040 Peter, so good to have you here. So great to be here, my friend. This is something that you and I have been

00:02:48.240 wanting to do for a while. And it's basically something that we do all the time, which is to

00:02:53.300 peruse the literature and find papers that we are excited about for whatever reason. And oftentimes that

00:03:00.260 will lead to a text dialogue or a phone call or both. But this time we've opted to try talking

00:03:07.940 about these papers that we find particularly exciting in real time for the first time as this

00:03:15.080 podcast format. First of all, so that people can get some sense of why we're so excited about these

00:03:19.380 papers. We do feel that people should know about these findings. And second of all, that it's an

00:03:25.440 opportunity for people to learn how to dissect information and think about the papers they hear

00:03:30.820 about in the news, the papers they might download from PubMed if they're inclined. Also just to start

00:03:35.780 thinking like scientists and clinicians and get a better sense of what it looks like to pick through

00:03:41.760 a paper, the good, the bad, and the ugly. So we're flying a little blind here, which is fun.

00:03:47.580 Um, I'm definitely excited for all the above reasons. Yeah, no, this is, uh, you and I've been

00:03:55.720 talking about this for some time and, and, um, you know, actually we used to run a journal club

00:03:59.920 inside the practice where once a month, one person would, um, just pick a paper and you would go

00:04:05.760 through it in kind of a formal journal club presentation. We'd gotten away from it for the last

00:04:09.420 year just because we've been a little stretched. Then I think it's something we need to resume because

00:04:14.500 it's, uh, it's a great way to learn and it's a skill, you know, people probably ask you all the

00:04:19.800 time. Cause I know I get asked all the time, Hey, what are the do's and don'ts of interpreting,

00:04:25.140 you know, scientific papers? Is it enough to just read the abstract? Um, and then, you know,

00:04:29.640 usually the answer is, well, no. Um, but the how to is, is tougher. And I think the two papers we've

00:04:35.220 chosen today illustrate two opposite ends of the spectrum. You know, you're going to obviously talk

00:04:39.540 about something that we're going to probably get into the technical nature of the assays, the

00:04:43.280 limitations, et cetera. And the paper ultimately I've chosen to present, although I apologize,

00:04:48.160 I'm surprising you with this up until a few minutes ago is, is actually a very straightforward,

00:04:53.260 simple epidemiologic paper that I think has important significance. I had originally gone

00:04:57.480 down the rabbit hole on a much more nuanced paper about ATP binding cassettes in cholesterol absorption.

00:05:04.280 But ultimately I thought this one might be more interesting to a broader audience.

00:05:07.700 By the way, I got to tell you a funny story. So I had a dream last night about you.

00:05:10.400 And, um, in this dream, you were obsessed with making this certain drink that was like your elixir

00:05:18.660 and it had all of these crazy ingredients in it. Supplements. Tons of supplements in it. But the

00:05:25.640 one thing I remembered when I woke up, cause I forgot most of them, I was really trying so hard to

00:05:29.640 remember them. One thing that you had in it was dew. Like you had to collect a certain amount of dew

00:05:36.020 off the leaves every morning to put into this drink. It was so, but it was like, just sounds

00:05:41.540 like something that I would do. And, and so, but here's the best part. You had, you had like a

00:05:47.420 thermos of this stuff that had to be with you everywhere. And all of your clothing had to be

00:05:52.160 tailored with a special pocket that you could put the thermos into so that you were never without

00:05:58.260 the special Andrew drink. And again, you know how dreams when you're having them seem so logical and

00:06:05.120 real. And then you wake up and you're like, that doesn't even make sense. Like, why would he want

00:06:09.840 the thermos in his shirt? Like that, that would warm it up. Like, you know, all these, but, but boy,

00:06:14.240 it was a realistic dream. And there were lots of things in it, including dew, special dew off the

00:06:20.640 leaves every morning.

00:06:21.460 I love it. Well, it's not that far from reality. I'm a big fan of yerba mate. I'm drinking it right

00:06:28.280 now. In fact, in its many forms, usually the loose leaf. I don't tend to drink it out of the

00:06:35.040 gourd. My dad's Argentine. So that's where I picked it up. I started drinking it when I was like five

00:06:39.080 years old or younger, which I don't recommend people do. It's heavily caffeinated. Don't drink

00:06:42.440 the smoked versions either folks. I think that was potentially carcinogenic, but this thing that you

00:06:47.500 describe of, of carrying around the thermos close to the body, if you are ever in Uruguay, or if you

00:06:54.300 ever spot grown men in a restaurant anywhere in the world, carrying a thermos with them and to their

00:06:59.840 meals and hugging it close, chances are they're Uruguayan and they're drinking yerba mate. They

00:07:06.480 drink it usually after their meals. It's supposed to be good for your digestion. So it's not that far

00:07:10.700 from, from reality. I don't carry the thermos, but I do drink mate every day. And I'm going to start

00:07:16.740 collecting dew off the leaves, uh, just a few drops every morning. Oh my. Um, some other time we can

00:07:24.940 talk about dreams recently. I've, I've been doing some dream exploration. I've had some absolutely

00:07:30.340 transformative dreams for the first time in my life. One dream in particular that has, that allowed me

00:07:35.520 to feel something I've never felt before and has catalyzed a large number of important decisions in a

00:07:42.000 way that no other experience waking or sleep has ever impacted me. And this was drug free,

00:07:48.080 et cetera. Um, and do you think you could have had that dream? We don't have to get into it if you

00:07:52.700 want to talk about it now, but was there a lot of work you had to do to prepare for that dream to

00:07:58.080 have taken place? Oh yes. Yeah. Um, at least, uh, 18 months of intensive, um, analysis type work,

00:08:06.660 um, with a very skilled psychiatrist, but I wasn't trying to seed the dream. It was just, I was at a

00:08:14.080 sticking point with a certain process in my life. And then I was taking a walk while waking and

00:08:20.040 realized that my brain, my subconscious was going to keep working on this. I just decided it's going

00:08:27.700 to keep working on it. And then two nights later I traveled to a meeting in Aspen and I had the most

00:08:32.500 profound dream ever, uh, where I was able to sense something and feel something I've always wanted

00:08:37.120 to feel as, uh, so real within the dream, woke up, knew it was a dream and realized this is what

00:08:44.620 people close to me that I respect have been talking about, but I was able to feel it. And therefore

00:08:49.740 I can actually access this in my waking life. It was, it was, it was absolutely transformative for me.

00:08:56.500 Um, anyway, sometime I can share more details with you or the audience, but for now we should talk

00:09:02.600 about these papers. Very well. Um, who should go first? I I'm happy to go first. This one's,

00:09:10.220 this one's, this is a pretty straightforward paper. So, so we're going to talk about a paper titled

00:09:14.260 reassessing the evidence of a survival advantage in type two diabetics treated with metformin compared

00:09:21.020 with controls without diabetes, a retrospective cohort study. This is by Matthew Thomas keys and

00:09:27.960 colleagues. This was published, uh, last fall. Um, why is this paper important? So this paper is

00:09:35.560 important because in 2014, uh, Bannister published a paper that I think in many ways kind of got the

00:09:44.620 world very excited about metformin. So this was almost 10 years ago. And I'm sure many people have

00:09:50.080 heard about this paper, even if they're not familiar with it, but they've heard the concept

00:09:53.460 of the paper. And in many ways, it's the paper that has led to the excitement around the potential

00:09:59.240 for zero protection with metformin. And I should probably just define for the audience what zero

00:10:04.740 protection means when we think probably also, sorry to interrupt what metformin is just for the

00:10:08.920 uninformed. That's a great point. So I'll start with the, with the latter. So metformin is a drug

00:10:14.560 that has been used for many years, uh, depends, you know, where it was first approved, I think was

00:10:20.980 in Europe. Um, but you know, call it directionally 50 plus years of use as a first line agent for

00:10:28.640 patients with type two diabetes, uh, in the U S maybe 40 plus years. So this is a drug that's been

00:10:34.700 around forever trade name, uh, glucophage, um, or brand name. And, uh, but, but again, it's,

00:10:40.420 you know, it's a generic drug today. The mechanism by which metformin works is debated hotly. Um,

00:10:48.940 but what I think is not debated is the immediate thing that metformin does, which is it inhibits

00:10:54.480 complex one of the mitochondria. So again, maybe just taking a step back. So the mitochondria is

00:11:00.420 everybody thinks of those as the cellular engine for making ATP. So the most efficient way that we

00:11:05.520 make ATP is through oxidative phosphorylation, where we take either fatty acid pieces or a

00:11:13.920 breakdown product of glucose. Once it's partially metabolized to pyruvate, we put that into an

00:11:19.460 electron transport chain and we basically trade chemical energy for electrons that can then be used

00:11:27.620 to make phosphates onto ADP. So it's, you know, you think of everything you do eating is taking the

00:11:33.480 chemical energy and food, taking the energy that's in those bonds, making electrical energy in the

00:11:38.560 mitochondria. Those electrons pump a gradient that allow you to make ATP. To give a sense of how

00:11:44.520 primal and important this is, if you block that process completely, you die. So everybody's probably

00:11:50.800 heard of cyanide, right? Cyanide is something that is incredibly toxic, even at the smallest doses.

00:11:56.280 Cyanide is a complete blocker of this process. And if my memory serves me correctly, I think it blocks

00:12:01.500 complex four of the mitochondria. I don't know if you recall if it complex three or complex four.

00:12:06.300 I know a lot about toxins that impact the nervous system, but I don't know a lot about

00:12:09.880 the mitochondria. But if ever you want to have some fun, we can talk about all the dangerous stuff that

00:12:14.700 animals make and insects make and how they kill you. Yeah. Like the trototoxin and all these things

00:12:19.920 that block sodium channels. I really geek out on this stuff because it allows me to talk about

00:12:24.840 neuroscience, animals, and scary stuff. It's like combines it. So we could do that sometime for fun.

00:12:30.380 Maybe at the end, if we have a few moments. So, you know, something like cyanide that is a very

00:12:35.060 potent inhibitor of this electron transport chain will kill you instantly. People understand that,

00:12:39.640 of course, a drop of cyanide and you would be dead literally instantaneously. So metformin works at

00:12:45.560 the first of those complexes. I believe there are four, if my memory serves correctly, four electron

00:12:50.040 transport chain complexes. But of course, it's not a complete inhibition of it. It's just kind of a weak

00:12:56.540 blocker of that. And the net effect of that is what? So the net effect of that is that it changes

00:13:02.100 the ratio of adenosine monophosphate to adenosine diphosphate. What's less clear is why does that

00:13:10.100 have a benefit in diabetics? Because what it unambiguously does is reduces the amount of

00:13:16.080 glucose that the liver puts out. So hepatic glucose output is one of the fundamental problems that's

00:13:22.080 happening in type 2 diabetes. You may recall, I think we talked about this even on a previous

00:13:26.640 podcast, you and I sitting here with normal blood sugar have about 5 grams of glucose in our total

00:13:33.140 circulation. That's it, 5 grams. Think about how quickly the brain will go through that within

00:13:37.980 minutes. So the only thing that keeps us alive is our liver's ability to titrate out glucose. And if

00:13:46.260 it puts out too much, for example, if the glucose level was consistently two teaspoons,

00:13:51.720 you would have type 2 diabetes. So the difference between being metabolically healthy and having

00:13:57.420 profound type 2 diabetes is one teaspoon of glucose in your bloodstream. So the ability of the liver

00:14:03.240 to tamp down on high glucose output is important. Metformin seems to do that.

00:14:08.480 Can I just ask one question? Is it fair to provide this overly simplified summary of the biochemistry,

00:14:16.620 which is that when we eat, the food is broken down, but the breaking of bonds creates energy that then

00:14:24.000 our cells can use in the form of ATP. And the mitochondria are central to that process. And that

00:14:28.900 metformin is partially short-circuiting the energy production process. And so even though we are eating

00:14:36.360 when we have metformin in our system, presumably there is going to be less net glucose. The bonds are

00:14:43.480 going to be broken down. We're chewing, we're digesting, but less of that has turned into blood

00:14:47.440 sugar, glucose. Well, sort of. I mean, it's not depriving you of ultimately storing that energy.

00:14:56.700 What it's doing is changing the way the body partitions fuel. That's probably a better way

00:15:03.160 to think about it to be a little bit more accurate. So for example, it's not depriving you

00:15:09.180 of the calories that are in that glucose. That would be, you know, fantastic. But that was the,

00:15:13.880 that was the, uh, Elestra approach. Remember the Elestra from the nineties? Elestra folks,

00:15:18.220 for those of you who don't remember, um, by the way, if you ever ate this stuff, you'd remember,

00:15:22.580 uh, because it was a fat that was, um, not easily digested. It had sort of in sort of analogous to

00:15:29.260 plant fiber or something like that. So it was being put into potato chips and whatnot. And

00:15:33.660 the idea is that people would, um, would simply excrete it. Um, and I don't know what happened

00:15:41.980 except that people got a lot of stomach aches and, um, everyone got fatter in the world. We know that

00:15:47.300 the anal seepage is what really did that product. Only a physician, because after all, Peter's a

00:15:54.140 clinician, a physician, an MD, and I'm not, um, could find it a, um, an appropriate term to

00:16:01.560 describe. Yeah. When you have that much, when you have that much fat malabsorption,

00:16:06.160 you start to have accidents. Wow. And so that, that did away with that product.

00:16:11.100 Right. It was either that or the diaper industry was going to really take off.

00:16:14.400 Okay. That's why you don't hear about Elestra. That's right. So, so we've got this drug,

00:16:18.200 we've got this drug metformin. It's considered a perfect first line agent for people with type

00:16:23.160 two diabetes. So again, what's happening when you have type two diabetes, uh, the primary insult

00:16:27.780 probably occurs in the muscles and it is insulin resistance. Everybody hears that term. What does

00:16:34.240 it mean? Uh, insulin is a peptide. It binds to a receptor on a cell. So let's just talk about it

00:16:39.640 through the lens of the muscle because the muscle is responsible for most glucose disposal. It gets

00:16:43.900 glucose out of the circulation. High glucose is toxic. We have to put it away and we want to put

00:16:49.020 most of it into our muscles. That's where we store 75 to 80% of it. When insulin binds to the

00:16:55.400 insulin receptor, a tyrosine kinase is triggered inside. So just ignore all that, but a chemical

00:17:01.440 reaction takes place inside the cell that leads to a phosphorylation. So ATP donates a phosphate group

00:17:08.080 and a transporter. Just think of like a little tunnel, like a little straw goes up through the

00:17:14.720 level of the cell and now glucose can freely flow in. So I'm sure you've talked a lot about this with

00:17:20.560 your audience. Things that move against gradients need pumps to move them. Things that move with

00:17:26.200 gradients don't. Glucose is moving with its gradient into the cell. It doesn't need active

00:17:30.800 transport, but it does need the transporter put there that requires the energy. That's the job

00:17:36.280 of insulin. By the way, I did not know that. I mean, I certainly know active and passive transport

00:17:41.540 as it relates to like neurotransmitter and ion flow. Um, but I'd never heard that when insulin

00:17:47.200 binds to a cell that literally a little straw is placed into the membrane of the cell doesn't need

00:17:52.260 a pump to move it in, um, because there's much more glucose outside the cell than inside. So it's

00:17:56.900 just, but the energy required is to move the straw up to the cell. So biology is so cool. Yeah, it is.

00:18:03.380 So, so what happens is as, and Gerald Shulman at Yale did the best work on elucidating this

00:18:11.220 as the intramuscular fat increases. And I, by intramuscular, I mean, intracellular fat, uh, triacyl

00:18:19.660 and diacyl glycerides accumulate in a muscle cell, that signal gets interrupted. And all of a sudden

00:18:26.100 I'm making these numbers up. If you used to need two units of insulin to trigger the little

00:18:32.080 transporter, now you need three and then you need four and then you need five. You need more and more

00:18:38.220 insulin to get the thing up. That is the definition of insulin resistance. The cell is becoming

00:18:44.680 resistant to the effect of insulin. And therefore the early mark of insulin resistance, the canary

00:18:51.200 in the coal mine is not an increase in glucose. It's an increase in insulin. So normal glycemia

00:18:58.460 with hyperinsulinemia, especially postprandial, meaning after you eat hyperinsulinemia is the thing

00:19:05.000 that tells you, Hey, you're, you're five, 10 years away from this being a real problem.

00:19:08.900 So fast forward, many steps down the line, someone with type two diabetes has long past

00:19:13.560 that system. Now, not only are they insulin resistant where they just need a boatload of

00:19:18.900 insulin, which is made by the pancreas to get glucose out of the circulation, but now that

00:19:23.600 system's not even working well. And now they're not getting glucose into the cell. So now their glucose

00:19:28.540 level is elevated. And even though it's continually being chewed up and used up, because again,

00:19:34.180 the brain alone would account for most of that glucose disposal, the liver is now becoming insulin

00:19:40.620 resistant as well. And now the liver isn't able to regulate how much glucose to put into circulation

00:19:46.200 and it's overdoing it. So now you have too much glucose being pumped into the circulation by the

00:19:50.840 liver and you have the muscles that can't dispose of it. And it's really a vicious, brutal cascade

00:19:55.740 because the same problem of fat accumulating in the muscle is now starting to happen in the pancreas.

00:20:01.040 And now the relatively few cells in the pancreas called beta cells that make insulin are undergoing

00:20:07.480 inflammation due to the fat accumulation within the pancreas itself. And so now the thing that you

00:20:13.720 need to make more insulin is less effective at making insulin. So ultimately way, way, way down the

00:20:19.020 line, a person with type two diabetes might actually even require insulin exogenously.

00:20:23.860 Could you share with us a few of the causes of type two diabetes of insulin resistance? I mean,

00:20:29.180 one, it sounds like, is accumulating too much fat. Yeah. So energy imbalance would be an enormous

00:20:34.240 one. Inactivity or insufficient activity is probably the single most important. So when

00:20:39.920 Gerald Shulman was running clinical trials at Yale, they would be recruiting undergrads to study,

00:20:47.140 obviously, because you're typically recruiting young people. And they would, you know, be doing

00:20:50.580 these very detailed mechanistic studies where they would require actual tissue biopsies. So, you know,

00:20:54.660 you're going to biopsy somebody's quadriceps and actually look at what's happening in the muscle.

00:20:58.720 Well, I remember him telling me this when I interviewed him on my podcast, he said,

00:21:02.400 the most important criteria of the people we interviewed is because they were still lean.

00:21:06.600 These weren't people that were overweight, but they had to be inactive.

00:21:09.300 You couldn't have active people in these studies. So exercising is one of the most important things

00:21:14.820 you're going to do to ward off insulin resistance. But there are other things that can cause insulin

00:21:20.280 resistance. Sleep deprivation has a profound impact on insulin resistance. I think we probably talked

00:21:24.540 about this previously, but if you, you know, some very elegant mechanistic studies where you sleep

00:21:28.700 deprive people, you know, you let them only sleep for four hours for a week, you'll reduce their

00:21:33.520 glucose disposal by about half. Wow. Which is, I mean, that's a staggering amount of, you're basically

00:21:39.300 inducing profound insulin resistance in just a week of sleep deprivation. Hypercortisolemia is another

00:21:44.720 factor. And then obviously energy imbalance. So where, when you're, when you're accumulating excess

00:21:49.400 energy, when you're getting fatter, if you start spilling that fat outside of the subcutaneous fat

00:21:54.620 cells into the muscle, into the liver, into the pancreas, all those things are exacerbating it.

00:21:59.100 Got it. Okay. So enter metformin, first line drug. So most of the drugs, so every drug you give a

00:22:06.320 person with type two diabetes is trying to address part of this chain. So some of the drugs tell you to

00:22:12.700 make more insulin. That's, that's one of the strategies. So here are drugs like sulfonylureas.

00:22:17.620 They tell the body, make more insulin. Other drugs like insulin, just give you more of the

00:22:23.820 insulin thing. Metformin tackles the problem elsewhere. It tamps down glucose by addressing

00:22:29.560 the glucose, the hepatic glucose output channel. GLP-1 agonists are another drug. They increase

00:22:35.960 insulin sensitivity, initially causing you to also make more insulin. GLP-1 agonists.

00:22:40.160 So that's Ozempic.

00:22:41.320 Yes.

00:22:41.900 Yeah. And is it true that berberine is more or less the poor man's metformin?

00:22:47.040 Yeah. Okay. Yeah.

00:22:48.320 It's a, from a tree bark, it just happens to have the same properties of reducing mTOR and

00:22:53.280 reducing blood glucose.

00:22:54.760 Yeah. And metformin, by the way, occurs from a lilac plant in France. Like that's where it

00:22:58.560 was discovered. So it's also, metformin is also based on a substance found in nature.

00:23:02.600 So you, you need a prescription for metformin. You don't need a prescription for berberine.

00:23:06.920 Correct.

00:23:07.280 But yeah, we can talk about berberine a little bit later. I had a couple great experiences with

00:23:11.100 berberine and a couple bad experiences.

00:23:13.040 Interesting.

00:23:13.340 Yeah. So, um, maybe taking one step back from this in 2011, I became very interested in metformin

00:23:22.300 personally, just reading about it, obsessing over it, and just somehow decided like I should

00:23:27.980 be taking this. So I actually began taking metformin. I still remember exactly when I

00:23:32.240 started. I started it in May of 2011. And I realized that because I was on a trip with

00:23:36.800 a bunch of buddies, we went to the Berkshire Hathaway, uh, shareholder meeting, which is,

00:23:42.300 uh, you know, the Buffett, uh, shareholder meeting. And, uh, you know, it was kind of

00:23:46.860 like a fun thing to do. And I remember being so sick the whole time because I didn't titrate

00:23:51.620 up the dose of metformin. I just went straight to two grams a day, which is kind of like the full

00:23:56.260 dose. And we went to this.

00:23:58.760 Is that characteristic of your approach to things?

00:24:00.640 Yes. I think that's safe to say next time. I'll give you a thermos of this do that I collect in

00:24:05.860 the morning.

00:24:09.520 So I remember being so sick that the whole time we were in Nebraska or Omaha, I guess I couldn't,

00:24:15.840 we went to Dairy Queen cause you do all the Buffett things when you're there, right? Like I couldn't

00:24:19.640 have an ice cream at Dairy Queen. You couldn't, I mean, I couldn't, I'm so nauseous. Oh, cause I

00:24:23.360 would say if you've got metformin in your system, you're going to buffer glucose. You could have four

00:24:26.440 ice cream cones. Except I couldn't put, I couldn't keep anything down. I mean, it was so

00:24:29.860 nauseous. So, so clearly metformin has this side effect initially, which is a little bit of

00:24:34.380 appetite suppression, but regardless, that's the story on metformin. There were, there are a lot

00:24:38.680 of reasons I was interested in it. Um, I wasn't thinking true Gero protection. That term wasn't

00:24:44.480 in my vernacular at the time, but what I was thinking is, Hey, this is going to help you buffer

00:24:48.200 glucose better. It's got to be better. And this was sort of my first foray into, you know,

00:24:52.300 self-experimentation. Do you want to define Gero protection? Yeah. Yeah. It's a good term to

00:24:56.420 define. Geriatric Gero. Yeah. So, so yeah, Gero from, from geriatric old protection. So protection

00:25:03.120 from aging. And when we talk about a drug like metformin or rapamycin or even NAD, NR, these

00:25:11.560 things, the idea is we're talking about them as Gero protective to signal that they are drugs that

00:25:17.440 are not targeting a specific disease of aging. For example, a PCSK9 inhibitor is sort of Gero

00:25:23.900 protective, but it's targeting one specific pathway, which is cardiovascular disease and

00:25:29.440 dyslipidemia. Whereas the idea is a Gero protective agent would target hallmarks of aging. There are

00:25:36.180 nine hallmarks of aging. Please don't ask me to recite them. I've never been able to get all

00:25:40.200 nine straight, but people know what we're talking about, right? So decreased autophagy, increased

00:25:45.240 senescence, decreased nutrient sensing or defective nutrient sensing, uh, proteomic instability,

00:25:50.300 genomic instability, uh, methylation, all of these things, epigenetic changes. Those are all the nine

00:25:54.960 hallmarks of aging. Yeah. So a Gero protective agent would target those deep down biologic hallmarks

00:26:02.460 of aging. And in 2014, um, a paper came out by Bannister that basically got the world focused on

00:26:10.800 this problem by the world. I mean the world of, of anti-aging. So what, what Bannister and colleagues

00:26:16.700 did was they took a registry from the UK and they got a set of patients who were on metformin with type

00:26:27.040 two diabetes, but only metformin. So these were people who had just progressed to diabetes. They

00:26:32.500 were not put on any other drug, just metformin. And then they found from the same registry, a group

00:26:39.360 of matched controls. So this is a standard way that epidemiologic studies are done because again,

00:26:46.460 you don't have the luxury of doing the randomization. So you're trying to account for all the biases

00:26:51.540 that could exist by saying, we're going to take people who, uh, look just like that person with

00:26:57.420 diabetes. So can we match them for age, sex, socioeconomic status, um, blood pressure, BMI,

00:27:06.300 everything we can. And then let's look at what happened to them over time. Now, again, this is all

00:27:11.600 happening in the future. So you're looking into the past. It's retrospective in that sense.

00:27:16.240 And so let me just kind of pull up the, the sort of, uh, table here so I can kind of walk through,

00:27:21.460 and this is not in the paper we talked about, but I think this is an important background.

00:27:24.620 So they did something that at the time I didn't really notice. I didn't notice what they did. I

00:27:33.700 probably did. And I forgot, but I didn't notice this until about five years ago when I went back and

00:27:38.480 looked at the paper and they did something called, um, uh, informative censoring. So the way the study

00:27:45.700 worked is if you were put on metformin, we're going to follow you. If you're not on metformin,

00:27:50.240 we're going to follow you. And we're going to track the number of deaths from any cause that

00:27:54.500 occurred. This is called all cause mortality or ACM. And it's really the gold standard in a trial of

00:28:00.320 this nature or a study of this nature, or even a clinical trial. You want to know how much are people

00:28:04.520 dying from anything? Cause we're trying to prevent or delay death of all causes. Informative censoring

00:28:10.440 says if a person who's on metformin deviates from that inclusion criteria, we will not count them in

00:28:20.440 the final assessment. So how are the ways that that can happen? Well, one, the person can be lost to

00:28:26.360 follow up to, they can just stop taking their metformin three. And more commonly, they can progress

00:28:34.120 to needing a more, uh, significant drug. So all of those patients were excluded from the study.

00:28:42.120 So think about that for a moment. This is, in my opinion, a significant limitation of this study,

00:28:47.480 because what you're basically doing is saying, we're only going to consider the patients who were

00:28:54.360 on metformin, stayed on metformin and never progressed through it. And we're going to compare those

00:28:59.040 to people who were not having type two diabetes. So an analogy here would be, imagine we're

00:29:04.100 going to do a study of two groups that we think are almost identical. One of them are smokers and

00:29:09.780 the other are identical in every way, but they're not smokers. And we're going to follow them to see

00:29:13.140 which ones get lung cancer. But every time somebody dies in the smoking group, we stop counting them.

00:29:19.780 When you get to the end, you're going to have a less significant view of the health status of that

00:29:25.380 group. So with that caveat, the Bannister study found a very interesting result, which was the

00:29:34.500 crude death rate, um, was, and by the way, the way these are done, this is also one of the challenges

00:29:42.060 of epidemiology is the math gets much more complicated. You have to normalize death rate for

00:29:47.440 the amount of time you study the people. So everything is normalized to thousand person

00:29:52.800 years. So the crude death rate in the group of people with type two diabetes who were on metformin,

00:30:00.900 including the censoring was 14.4. So 14.4 deaths occurred per thousand patient years.

00:30:08.100 If you looked at the control group, it was 15.2. This was a startling result. And I remember reading

00:30:16.920 this in again, 2014 and being like, holy crap, this is really amazing. Is there, um, could you explain

00:30:23.960 why? Cause I, I, I hear those numbers and they don't seem that striking. It's a difference of about

00:30:28.920 a year and a half. Now, of course, um, a difference of about a year and a half and lifespan is, is,

00:30:34.500 well, it doesn't even translate to that. So, so taking a step back, diabetes, type two diabetes

00:30:39.720 on average will shorten your life by six years. I see. So that's the actuarial difference between

00:30:44.420 having type two diabetes and not all comers, but you're right. This is not a huge difference. It's

00:30:49.020 only a difference of a little less than one year of life per thousand patient years studied. Okay.

00:30:54.900 And by the way, up here, just point out my, my math was wrong when I said about a year and a half,

00:30:58.680 but the point here is you would expect the people in the metformin group to have a far

00:31:04.100 worse outcome, i.e. to have a far worse crude death rate. And the fact that it was statistically

00:31:11.660 significant in the other direction. And it turned out on the, what's called a Cox proportional hazard,

00:31:16.580 which is where you actually model the difference in lifespan. The people who took metformin and had

00:31:24.820 diabetes had a 15%, one five, 15% relative reduction in all cause death over 2.8 years, which was the

00:31:34.000 median duration of follow-up. Well, that seems to be the number that makes me go, wow. Yeah. Right.

00:31:39.520 That, um, because could you repeat those numbers again? Yeah. So 15% reduction in all cause mortality

00:31:47.200 over 2.8 years. That's a big deal. It is. And again, there's no clear explanation for it unless you

00:31:58.300 believe that metformin is doing something beyond helping you lower blood glucose. Because the

00:32:05.480 difference in blood glucose between these two people was still in favor of the non-diabetics.

00:32:12.100 So again, the proponents of metformin being a gyroprotective agent, and I put myself in this

00:32:18.460 category at one point, I would put myself today in the category of undecided. But at the time,

00:32:22.880 I very much believed this was a very good suggestion that metformin was doing other things.

00:32:29.680 You mentioned a couple already. Metformin is a weak inhibitor of mTOR. Metformin reduces

00:32:34.880 inflammation. Metformin potentially tamps down on senescent cells and their secretory products.

00:32:40.960 You know, there are lots of things metformin could be doing that are off target. And it might be that

00:32:46.540 those things are conferring the advantage. So fast forward until a year ago, and I think most people

00:32:53.640 took the banister study as kind of the best evidence we have for the benefits of metformin. And I'm sure

00:33:00.120 you've had lots of people come up to you and ask you, should I be on metformin? Should I be on

00:33:03.700 metformin? I mean, I probably get asked that question almost as much as I'm asked any question

00:33:08.620 outside of dew. I mean, people definitely want to know if you should be consuming dew, but after that,

00:33:12.980 it's metformin. Fresh off the leaves. Has to be. While viewing morning sunlight.

00:33:16.960 So, okay. So let's kind of fast forward to now the paper that I wanted to spend a few more minutes

00:33:22.000 on. Yeah. And thanks for that background. I'm still dazzled by the insertion of the straw by way

00:33:29.240 of insulin. I don't think I've ever heard that described. I need to go get a better textbook.

00:33:36.240 It's a pretty short straw in fairness. It's just a little transport.

00:33:39.380 Just to give people a sense of why I'm so dazzled by it, I am always fascinated by how quickly,

00:33:47.280 how efficiently, and how specifically biology can create these little protein complexes that

00:33:56.340 do something really important. I mean, you're talking about an on-demand creation of a portal,

00:34:01.560 right? I mean, these are cells engineering their own machinery in real time in response to chemical

00:34:05.820 signals. It's great. Yeah. But I'm sort of rusty on my neuroscience, but an action potential works

00:34:12.360 in reverse the same way. Like you need the ATP gradient to restore the gradient. But once the

00:34:19.760 action potential fires, it's passive outside, right? Yeah. So what Pierre's referring to is

00:34:24.240 the way that neurons become electrically active is by the flow of ions from the outside of the cell to

00:34:31.080 the inside of the cell. And we have both active conductances, meaning they're triggered by electrical

00:34:34.980 changes in the gradients, by changes in electrical potential. And then they're passive gradients

00:34:41.100 where things can just flow back and forth until there's a balance equal inside and outside the cell.

00:34:45.600 I think what's different is that there's some movement of a lot of stuff inside of neurons when

00:34:52.020 neurotransmitters like dopamine binds to its receptor and then a bunch of, you know, it's like a

00:34:55.800 bucket brigade that gets kicked off internally. But it's not often that you hear about receptors

00:35:00.540 getting inserted into cells very quickly. Normally, you have to go through a process of,

00:35:04.200 you know, transcribing genes and making sure that the specific proteins are made. And then those are

00:35:08.940 long, slow things that take place over the course of many hours or days. What you're talking about is

00:35:12.820 a real on-demand insertion of a channel. And it makes sense as to why that would be required. But it's

00:35:19.740 just so very cool. It's cool. Yeah. So Keys and colleagues came along and said,

00:35:23.880 we would like to redo the entire banister analysis. And I think their motivation for it was the

00:35:32.180 interest in this topic is through the roof. There is a clinical trial called the TAME trial that is,

00:35:40.000 I think, pretty much funded now and may be getting underway soon. The TAME trial,

00:35:44.760 which is an important trial, is going to try to ask this question prospectively and through random

00:35:49.480 assignments. So this is the targeting aging with metformin trial.

00:35:53.100 That's correct. Near Barzilai is probably the senior PI on that. And I think in many ways,

00:36:02.240 the banister study, along with some other studies, but of lesser significance, probably provided some

00:36:08.520 of the motivation for the TAME trial. So they said, okay, look, we're going to do this. We're going to

00:36:12.460 use a different cohort of people. So the first study that we just talked about, the banister study

00:36:18.100 used, I believe it was like roughly, they sampled like 95,000 subjects from a UK biobank. Here,

00:36:25.940 they used a larger sample. They did about half a million people sampled from a Danish health registry.

00:36:32.840 And they did something pretty elegant. They created two groups to study. So the first was just a standard

00:36:38.520 replication of what banister did, which was just a group of people with and without diabetic that they

00:36:44.560 tried to match as perfectly as possible. But then they did a second analysis in parallel with

00:36:49.840 discordant twins. So same-sex twins that only differed in that one had diabetes and one didn't.

00:36:57.580 I thought this was very elegant because here you have a degree of genetic similarity and you have

00:37:03.360 similar environmental factors during childhood that might give you, you know, allow you to see if

00:37:09.540 there's any sort of difference in signal. So now turning this back into a little bit of a journal

00:37:13.940 club, virtually any clinical paper you're going to read, table one is the characteristics of the

00:37:21.900 people in the study. You always want to take a look at that. So when I look at table one here,

00:37:26.860 you can see it's, and by the way, just for people watching this, we're going to make all these papers

00:37:30.720 and figures available. So if you're, you know, don't, you know, we'll have nice show notes that'll make

00:37:35.580 all this clear. So table one in the keys paper shows the baseline characteristics. And again,

00:37:42.020 it's almost always going to be the first table in a paper. Usually the first figure in the paper is a

00:37:47.080 study design. It's usually a flow chart that says these were the inclusion criteria. These are all

00:37:52.200 the people that got excluded. This is how we randomized, et cetera. And you can see here that

00:37:56.660 there are four columns. So the first two are the singletons. These are people who are not related.

00:38:01.740 And then the second two are the twins who are matched. And you can see, remember how I said

00:38:06.380 they sampled about 500,000 people? You can see the numbers. So they got, you know, 7,842 singletons

00:38:13.720 on metformin, the same number. Then they pulled out matched without diabetes on the twins. They got

00:38:18.480 976 on metformin with diabetes. And then by definition, 976 co-twins without them. And you look

00:38:27.800 at all these characteristics. What was their age upon entry? How many were men? What was the year

00:38:32.840 of indexing when we got them? What medications were they on? What was their highest level of education,

00:38:38.260 marital status, et cetera? The one thing I want to call out here that really cannot be matched in a

00:38:44.020 study like this, so this is a very important limitation, is the medication. So look at that

00:38:49.020 column, Andrew. Notice how pretty much everything else is perfectly matched until you get to the

00:38:53.540 medication list. Yeah, it's all over the place. Yeah, it's just, it's not even close. They're

00:38:58.960 nowhere near matched, right? In other words, just to give you a couple of examples, right?

00:39:03.460 On the, and let's just talk about the singletons, because it's basically the same story on the twins.

00:39:07.140 If you look at what fraction of the people with type 2 diabetes are on lipid-lowering medication,

00:39:13.080 it's 45.6% versus 15.4% in the matched without diabetes. It's a 3x difference.

00:39:19.880 What about antiplatelet therapy? That's 30% versus 14%. Antihypertensive, 65% or 63% versus 31%.

00:39:28.660 Because people who have one health issue and are taking metformin are likely to have other health

00:39:32.640 issues. Exactly. So this is, again, a fundamental flaw of epidemiology. You can never remove all the

00:39:40.860 confounders. This is why I became an experimental scientist, so that we could control variables.

00:39:45.980 That's right. Because without random assignment, you cannot control every variable. Now, you'll see

00:39:50.900 in a moment when we get into the analysis, they go through three levels of corrections,

00:39:56.500 but they can never correct this medication one. So just keep that in the back of your mind.

00:40:00.580 Okay. So the two big things that were done in this experiment or in this survey or study to

00:40:07.760 differentiate it from Bannister was one, the twin trick, which I think is pretty cool.

00:40:11.860 The second thing that they did was they did a sensitivity analysis with and without informative

00:40:19.360 censoring. So one of the other things they wanted to know was, hey, does it really matter if we don't

00:40:25.380 count the metformin patients who progress? So let's see kind of what transcribed. So the next figure,

00:40:34.220 figure two, pardon me, the next table, table two, walks you through the crude mortality rate

00:40:41.660 in each of the groups. So the most important row, I think, in this table is the one that says crude

00:40:49.480 mortality per thousand person years. Now, you recall that in the previous study, in the Bannister

00:40:55.620 study, those were on the ballpark of about 15 per. Okay. So let's look at each of these. So in the

00:41:04.260 single, the singletons with, without, so the non-twins who were not diabetic, it was 16.86.

00:41:13.000 And could you put a little more contour on what this thousand person years?

00:41:17.800 What it is?

00:41:18.540 Are you talking about pooling the lifespans of a bunch of different people until you get to the

00:41:23.640 number 1,000? Because you're normalizing not, so it's not who's going to live 1,000 years because

00:41:28.900 no one's expecting that. You're essentially taking, so you've got some people that are going to live

00:41:33.780 76 years, 52 years, 91 years, and you're pooling all of those until you hit 1,000. And then that

00:41:41.080 becomes kind of a, it's like a normalized division. You're basically like, so let's say the control

00:41:48.280 group, you're asking if there were 1,000 person years available to live, how likely is it that

00:41:55.480 this person would live another 15? Yeah. So a couple of ways to think about it. So taking a step

00:42:00.280 back, we always have to have some way of normalizing. So when we talk about the mortality from a disease

00:42:05.280 like cancer in the population, we would, we report it as what's the mortality rate per, and it's

00:42:12.520 typically per 100,000 persons. Okay. That's a much more intuitive way to express it.

00:42:17.740 It is. But the reason we can do it that way is because we're literally looking at how many people

00:42:24.300 died this calendar year, and we divide it by the number of people in that age group. So it's

00:42:29.720 typically what you're doing when you look at aged groups in buckets of like decades. So that's why we

00:42:37.340 can say the highest mortality is like people 90 and up. Even though the absolute number of deaths is

00:42:44.900 small, it's because there's not that many people there, right? The majority of deaths in absolute

00:42:49.780 terms probably occur in the seventh decade. But as you go up, because the denominator is shrinking,

00:42:56.980 you have to normalize to it. So we just normalize to the number of people. Here are all the people

00:43:01.020 that started the year. Here are all the people that ended the year. What's the death rate? Why are

00:43:04.780 these done in a slightly more complicated way? Because we, we, we don't follow these people for their

00:43:10.380 whole lives. We're only following them for a period of observation. In this case, roughly three years.

00:43:15.160 So to say something like, you know, we have a crude death rate of five deaths per thousand person

00:43:22.340 years. One way to think about that is if you had a thousand people and you followed them for one year,

00:43:29.860 you'd expect five to die. If you had 500 people and you followed them for two years, you expect five

00:43:36.320 to die. If you have a thousand people and you follow them for one year, you expect five to die.

00:43:42.020 Those would all be considered equivalent mortalities. Great. Thank you for clarifying that.

00:43:46.540 No, no, this, this stuff is, I mean, like I find, I find epidemiology when you get in the weeds is

00:43:52.640 way more complicated than following the basics of, um, experimental stuff where you just, you get to

00:43:59.180 push all this stuff into the garbage bin and just say, Hey, we're going to take this number of

00:44:03.880 people. We're going to exclude this group. We're going to randomize. We're going to see what happens.

00:44:07.300 Yeah. That's what like the paper we'll talk about next. So when you adjust for age and they don't

00:44:15.640 show it in this table, it's only in the text. When you adjust for age, a very important check to do

00:44:21.920 is what is the crude death rate of the people on metformin who are not twins versus who are twins.

00:44:28.660 Now in this table, they look different because it's 24.93 for the metformin group and 21.68 for

00:44:36.580 the twin group in that's on metformin. When you adjust for age, they're almost identical. It's,

00:44:41.580 it goes from 29 point, 24.93 to 24.7. One other point I'll make here for people who are going to be

00:44:48.880 looking at this table is, um, you'll notice there are parentheses after every one of these numbers.

00:44:54.340 What does that, what does that offer in there? Those parentheses are offering the 95% confidence

00:44:59.960 interval. So for example, to take the number, you know, 24.93 is the crude death rate of how many

00:45:07.100 people are dying who take metformin. What it's telling you is we're 95% confident that the actual

00:45:13.360 number is between 23.23 and 26.64. If a 95% confidence interval does not cross the number

00:45:23.160 zero, it's statistically significant. Okay. So the first thing that just jumps out at you,

00:45:30.980 I think when you look at this is there's clearly a difference here between the people who have

00:45:35.960 diabetes and those who don't. It complicates the study a little bit because it's basically two studies

00:45:40.460 in one, but you're comparing, um, 95, pardon me, uh, 24.93 to 16.86, which by the way,

00:45:50.300 remains after age adjustment. When you go to the twin group, it's 24.73 to 12.94.

00:45:56.700 So maybe just to zoom out for that, what you're describing, if I understand correctly is this, um,

00:46:01.860 uh, crude deaths per 1000 person years. Let's just talk about the singletons, the non-twins

00:46:07.040 is 16.86. So 16.86 people die. And some people are probably thinking, how can 0.86 of a person die?

00:46:14.900 Well, it's not always whole numbers, but, um, there's a, there's a bad joke to be made here,

00:46:19.960 but, um, yeah, just call it 17 versus 25, right? 17 deaths per thousand versus 25 deaths. Yep. And the

00:46:28.400 25 is in the folks that took metformin. Now that to the naive listener and to me means, oh, you know,

00:46:37.000 metformin basically kills you, right? Um, not a faster, or you, you know, you're more likely to

00:46:42.420 die, but we have to remember that these people have another, they have a major health issue that

00:46:47.460 the other group does not have. That's right.

00:46:49.260 Because people weren't assigned drug or not assigned drug. It wasn't placebo drug. It's let's

00:46:54.780 look at people taking this drug for a bad health issue and compare to everyone else.

00:47:00.160 That's right. So now you have to go into, and I'll just sort of skip the next figure,

00:47:06.620 but the next figure is a Kaplan-Meier curve. I think it's actually worth looking at it because

00:47:10.900 they show up in all sorts of studies. So if you look at figure one, it's a Kaplan-Meier curve,

00:47:15.740 which is a mortality curve. So you'll see these in any study that is looking at death.

00:47:23.660 And this can be prospective randomized. This can be retrospective, but these are always going to

00:47:28.440 show up. And I think it's really worth understanding what a Kaplan-Meier curve shows you. So on the x

00:47:33.120 axis is always time. And on the y axis is always the cumulative survival. So it's a curve that always

00:47:40.380 goes from zero to one, one or 100%. And it's always decreasing monotonically, meaning it can only go

00:47:49.040 down or stay flat. It can never go back up. So that's what a cumulative mortality curve looks like.

00:47:55.480 Now we're looking at, you're starting at alive and you're looking at how many people die for every

00:48:01.460 year that passes. That's right. And in each curve, there's one on the left, which is the matched

00:48:07.880 singletons. And there's the one on the right, which are the discordant twins. You have two lines. You

00:48:12.800 have those that were on metformin with type two diabetes and you have their matched controls.

00:48:18.140 And in this figure, the matched controls are the darker lines and the people with type two diabetes

00:48:24.560 on metformin, that's the lighter line. You'll also notice, and I like the way they've done it here,

00:48:29.800 they've got shading around each one. And we should mention for those that are just listening that

00:48:34.400 in both of these graphs, the, um, downward trending line from the controls. So again,

00:48:41.180 non-diabetic, not taking metformin is above the line, uh, corresponding to the diabetics who are

00:48:48.740 taking metformin. Um, put crudely, um, the people who are taking metformin that have diabetes are dying

00:48:57.820 at a faster rate for every single year examined. The two lines do not overlap except at the beginning

00:49:02.960 when everyone's alive. It's like a foot race where basically the people with metformin and diabetes

00:49:07.960 are falling behind and dying as they fall. That's right. And I'm glad you brought up a good point.

00:49:13.480 It's not uncommon in treatments, uh, to see Kaplan-Meier curves cross. They don't have to,

00:49:20.340 it's not a requirement that they never cross. It's only a, uh, a requirement that they're

00:49:24.600 monotonically decreasing or staying flat. So I've seen cancer treatment drugs where they have like

00:49:30.540 two drugs going head to head in a cancer treatment. And like one starts out looking really, really bad,

00:49:35.920 but then all of a sudden it kind of flattens while the other one goes bad. And then it actually

00:49:39.700 crosses and goes underneath, but that's not the case here. So to your point, the people with diabetes

00:49:46.260 taking metformin in both the match singletons and the discordance are dropping much faster and they

00:49:52.800 always stay below. And I was just going to say that the shading is just showing you a 95% confidence

00:49:58.120 interval. So you're just putting basically error bars along this. So if this were experimental data,

00:50:03.440 if you were doing an experiment with a group of mice and you were watching their survival and you

00:50:09.660 were, you know, what you'd have error bars on this, which you're actually measuring. So this is

00:50:14.160 because you have much more data here, you're just showing this in this fashion.

00:50:17.160 For those that haven't, um, been familiar as to statistics, no problem. Um, error bars correspond to

00:50:22.080 like if you were just going to measure the heights of a, of a room full of 10th graders,

00:50:25.240 there's going to be a range, right? You'll have the very tall kid and the, and the very, uh,

00:50:30.000 shorter kid and you'll have the short kid and the medium kid. And you'll, and so there's a range,

00:50:33.400 there's going to be an average, a mean, and then there'll be standard deviations and standard errors.

00:50:37.640 And, um, uh, so these confidence intervals just give a sense of how much range, you know,

00:50:43.580 some people, um, die, die early. Some people die late within a given year. They're going to be

00:50:49.520 different ages. Um, so it, these error bars can account for a lot of different forms of variability

00:50:54.660 here. You're talking about the variability is how many people in each group die. We're not

00:51:00.500 tracking one diabetic taking metformin versus, um, a control. I should have asked this earlier,

00:51:06.080 but, um, well, and it's also a mathematical model at this point too, that's smoothing it out

00:51:11.260 because notice it's running for the full eight years, even though they're only following people

00:51:16.000 for, you know, typically, I think the median was like three or four years at a time. So they're

00:51:20.620 using this quite complicated type of mathematics called a Cox proportional hazard, which is what

00:51:25.880 generates hazard ratios. And basically any model has to have some error in it. And so they're

00:51:31.960 basically saying, this is the error. So you could argue when you look at that figure, we don't know

00:51:37.340 exactly where the line is in there, but we know it's in that shaded area. Sorry to make one other

00:51:43.620 point. If those shaded areas overlapped, you couldn't really make the conclusion. You wouldn't

00:51:50.160 know for sure that one is different from the other. Yeah. That's actually a good opportunity to, um,

00:51:55.740 to, uh, raise a common myth, which is a lot of people, when they look at a paper, let's say it's

00:52:02.120 a bar graph, you know, um, and they see these error bars and they will say, people often think,

00:52:08.920 oh, if the error bars overlap, it's not a significant difference. But if the error bars

00:52:14.440 don't overlap, meaning there's enough separation, then that's a real and meaningful difference.

00:52:18.360 And that's not always the case. It depends a lot on the form of the experiment. Um, I often see some

00:52:24.000 of the more robust Twitter battles over, you know, how people are reading graphs. And I think it's

00:52:29.060 important to remember that, um, you run the statistics, hopefully the correct statistics for

00:52:34.040 the, for the sample. Um, but determining significance, whether or not that the result

00:52:38.880 could be due to something other than chance, of course, your confidence in that increases as it

00:52:44.640 becomes typically P values, P less than 0.00001% chance that it's due, um, to chance, right? So very

00:52:52.040 low, probably less than 0.05 tends to be the kind of gold standard cutoff. Um, but when you're talking

00:52:58.280 about data like these, which are repeated measures over time, people are dropping out literally,

00:53:03.420 um, over time, you're saying they've modeled it to make predictions as to what would happen.

00:53:08.620 We're not necessarily looking at, you know, raw data points here.

00:53:11.480 Yeah. The raw data was in the previous table. That's now taken and run through this Cox model

00:53:17.180 and it's smoothed out. Got it.

00:53:19.960 And to your point about the bar graphs, yeah, I think the other thing you always want to understand

00:53:24.280 is just because something doesn't achieve statistical significance, the only way you can say it's not

00:53:30.940 significant is you have to know what it was powered to detect. Um, and statistical power

00:53:37.060 is, uh, a very important concept that probably doesn't get discussed enough. Uh, but before you

00:53:42.900 do an experiment, you have to have an expectation of what you believe the difference is between the

00:53:49.100 groups and you have to determine the number of samples you will need to assess whether or not

00:53:56.960 that difference is there or not. So you use something it's, it's called a power table and

00:54:02.760 you, you would go to the power table. So if you, if you're doing treatment a versus treatment B

00:54:06.440 and you say, well, look, I think treatment a is going to have a 50% response. And I think treatment

00:54:11.840 B will have a 65% response. You literally go to a power table that says 50% response, 15% difference.

00:54:20.900 That gives you a place on the grid. And I want to be 90% sure that I'm right. So 90% power. I'm being

00:54:27.760 a little bit, so there's going to be a statistician listening to this. Who's going to want to kill me,

00:54:30.980 but this is directionally the way we would describe it. And that tells you, this is how many animals or

00:54:36.600 people you would need in this study. You're going to need 147 in each group. And by the way, if you now

00:54:42.960 do the experiment with 147 and you fail to find significance, you can comfortably say there is

00:54:49.780 no statistical difference, at least up to that 15%. There may be a difference at 10%, but you

00:54:55.740 weren't powered to look at 10%. Yeah. And very important point that you're making. Another

00:55:00.720 point that's just a more general one about statistics in general, the way to reduce variability in a data

00:55:07.020 set is to increase sample size. And that kind of makes sense, right? If you, if I just walk into a

00:55:11.540 10th grade class and go, Hey, I'm going to measure height. And I look up by the first three kids that I

00:55:16.240 see. And I happen to look over there and it's the three that all play on the volleyball team

00:55:20.240 together. I, my sample size is small and I'm likely to get a skewed representation in this case,

00:55:26.940 taller than average. So increasing sample size tends to decrease variation. So that's why when you hear

00:55:33.060 about a study from the UK biobank or from, you know, um, half a million Danish citizens, like for instance,

00:55:39.760 in this study, that's, those are enormous sample sizes. So even though this is not an experimental

00:55:45.860 study, it's an epidemiological observational study. Um, there's tremendous power by way of

00:55:51.840 the enormous number of subjects in the study. And that's the way that epidemiology will make up for

00:55:57.140 its deficit. So you could never do a randomized assignment study on half a million people. Um, you

00:56:04.560 know, so, so epidemiology makes up for its biggest limitation, which is it can never compensate for

00:56:13.120 inherent biases by saying we can do infinite duration if we want. Like we could, we could

00:56:18.620 survey people over the course of their lives and we can have the biggest sample size possible

00:56:22.500 because this is relatively cheap. The cost of actually doing an experiment where you have tens of

00:56:28.500 thousands of people is prohibitive. I mean, if you look at the woman's health initiative, which was a

00:56:31.640 five year study on, I don't know, what was it? 50,000 women. I mean, that was a billion dollar

00:56:36.600 study. So this is, this is the balancing act between epidemiology and randomized prospective

00:56:44.420 experiments. And, uh, they, so they both offer something, but you just have to know their blind

00:56:49.220 spots of each one. Um, so let's just kind of wrap this up. I mean, I think, uh, let's just go to table

00:56:55.080 four, which I think is the most important table, um, in, in here, which now lays out the, the,

00:57:01.060 the final results in terms of the hazard ratio. So this is, this is the way we want to really be

00:57:05.540 thinking about this. So again, hazard ratios, um, these are important things to understand.

00:57:10.500 A hazard ratio is a number and you always subtract one from the hazard ratio. And that tells you if

00:57:18.160 it's a positive number, if it's a number, sorry, if it's a number greater than one, you subtract one

00:57:21.700 and that tells you the relative harm. So if the hazard ratio is 1.5, you subtract 1.5 is a 50%

00:57:28.420 increase in risk. Um, if the number is negative, you may recall on the banister paper, the hazard

00:57:34.200 ratio was 0.85. So when it's nothing, so that means it's a 15% reduction in relative risk.

00:57:39.920 And here you can see all the hazard ratios are positive. So what it's telling you here is,

00:57:45.140 and I'm going to walk through this cause it's, there's a lot of information packed here.

00:57:48.240 You've got singletons, you've got twins. They're showing you three different ways that they do it.

00:57:53.440 They do an unadjusted model. If you just look at the singletons with and without metformin and you

00:57:59.660 make no adjustments, the hazard ratio is 1.48. Meaning the people on metformin had a 48% greater

00:58:08.420 chance of dying in any given year than their non-diabetic counterpart. The only reason I'm

00:58:13.720 smiling. It's not because I enjoy people dying quite, uh, quite to the contrary is that, um,

00:58:18.720 this is novel for me in that I've read some epidemiological studies before, but it's not

00:58:23.080 normally where I spend the majority of my time. But up until now I was thinking, okay, people taking

00:58:28.160 metformin are, are dying more than those that aren't. I just, and I, I'm just relieved to know

00:58:33.000 that I wasn't, um, looking at all this backwards. Okay. So they're dying more, but of course we don't

00:58:38.820 have a group that's taking metformin who doesn't have diabetes and we don't have a group, um, who,

00:58:44.080 uh, has diabetes and, uh, you know, is taking metformin plus something else. So again, we're,

00:58:49.900 we're only dealing with these constrained. Yeah. Now there's another arm to this study that I'm not

00:58:54.900 getting into because it adds more complexity, which is they also have another group that's got

00:59:00.280 diabetes, takes metformin and take sulfonylureas, which is a bigger drug. And those people die even more.

00:59:06.600 Whoa. So, which again speaks to the point, right? The more you need these medications,

00:59:13.180 they're never able to erase the effect of diabetes.

00:59:17.360 But in this case, it seems that they might be accelerating,

00:59:20.920 possibly accelerating death due to diabetes. Possibly.

00:59:23.860 We, we could never know that from this because we're, we don't see, we would need to see diabetics

00:59:29.560 who don't take metformin, who take nothing. And I would bet that they would do even worse.

00:59:33.360 Mm-hmm. So my intuition is that the metformin is helping, but not helping nearly as much as we

00:59:40.400 thought before. So my point is they make another set of adjustments. They say, okay, well, look in

00:59:45.320 the first one, in the unadjusted model, we only matched for age and gender. Okay. That's pretty

00:59:51.300 crude. What if we adjust for the medications they're on the cardiovascular, psychiatric, pulmonary,

00:59:57.600 dementia meds, and marital status? I don't know why they threw marital status in there,

01:00:01.280 but they did. I don't know. Maybe being married or unmarried can accelerate.

01:00:04.120 I'm sure it can, but it just seems like a random thing to throw in with all their meds. I would

01:00:07.360 have personally done that adjustment higher up. But nevertheless, if you do that, all of a sudden,

01:00:12.740 the hazard ratio drops from 1.48 to 1.32, which means, yep, you still have a 32% greater chance

01:00:20.560 of dying in any given year. All right. What if we also adjust for the highest level of education

01:00:28.500 along with any of the other covariance? Well, that doesn't really change it at all. It ends up at

01:00:32.380 1.33 or a 33% chance increase in death. Okay. I always knew that more school wasn't going to save

01:00:37.880 me. It's not doing jack. So now let's do it for the twins. If you do the twin study, which you could

01:00:43.340 argue is a slightly purer study because you at least have one genetic and environmental thing that

01:00:49.500 you've attached, the unadjusted model is brutal. 2.15. That's 115%. Think about this. These are twins

01:00:58.260 who in theory are the same in every way, except one has diabetes and one doesn't. And the one with

01:01:03.680 diabetes on metformin still has 115% greater chance of dying than the non-diabetic co-twin.

01:01:10.380 When you make that first adjustment of all the meds and marital status, you bring it down to a 70%

01:01:15.460 increase in risk. And when you throw education in, it goes up to an 80% chance of risk.

01:01:21.220 Now they did this really cool thing, which was they did the analysis on with and without censoring.

01:01:27.880 So everything I just said here was based on no censoring. Tell me about censoring.

01:01:33.940 Censoring is when you stop counting the metformin people who have died.

01:01:38.240 Okay. So in the singleton group, when you unadjust it, and the reason I'm doing the unadjusted is that's

01:01:45.380 where they did the sensitivity analysis. I don't think it really matters that much. It's you just

01:01:49.420 have to draw a line in the sand somewhere. You'll recall that that was a 48% chance of increased

01:01:55.740 mortality, all cause mortality. If you stop counting, if you, if you, pardon me, if you don't

01:02:01.160 censor, meaning if you include everybody, including when people on metformin with diabetes die, if you

01:02:07.260 censor them, it comes down to 1.39. In other words, this is a very important finding. It did not

01:02:14.420 undo the benefits that we saw in the Bannister study. Bannister saw a 15% reduction in mortality

01:02:22.720 when they censored. When Keyes censored, it got better, but not that much better. It went from

01:02:31.140 48 to 39%. In the twins, it went from 115% down to only 97%. So in some ways, this presents a little

01:02:43.040 bit of an enigma because it's not entirely clear to me, having read these papers many times, exactly

01:02:49.700 why Bannister found such an outlined, like such a different response. There's another, there's

01:02:55.880 another technical detail of this paper, which is they, you can see on the right side of table four,

01:03:01.040 they did something called the nested case control. But you'll see, and I was going to go into a long

01:03:06.700 explanation of what nested case controls are. It's another pretty elegant way to do case control

01:03:12.660 studies where you sample by year and you sort of normalize, you don't count all the cases at the

01:03:19.780 end, you count them one by one. I don't think it's worth getting into, Andrew, because it doesn't

01:03:23.880 change the answer. You can see it changes it just slightly, but it doesn't change the point.

01:03:27.840 The point here is the Keyes paper makes it undeniably clear that in that population, there was no

01:03:35.480 advantage offered by metformin that undid the disadvantage of having type 2 diabetes. This does

01:03:42.640 not mean that metformin wasn't helping them because we don't know what these people would have been

01:03:47.420 like without metformin. It could be that this bought them a 50% reduction in relative mortality to

01:03:53.480 where they'd been. But what it says is, in a way, this is what you would have expected.

01:03:59.480 This is what you would have expected 10 years ago before the banister paper came out.

01:04:03.700 Or maybe even before metformin was used, because in some ways it's saying,

01:04:07.260 what is the likelihood that sick people who are on a lot of medication are going to die compared to

01:04:11.860 not sick people who aren't on a lot of medication? It's not quite that simple in the sense that,

01:04:18.480 as you said, there are ways to try and isolate the metformin contribution somewhat because they're

01:04:26.780 on a bunch of other meds. And presumably that was done and analyzed in other figures where they can

01:04:32.860 sort of try and – they can never attach the results specifically to metformin, right? But there

01:04:39.620 must be some way of weighting the percentage that are on psychiatric meds or not on psychiatric meds as

01:04:45.400 some way to tease out whether or not there's actually some contribution in metformin to this

01:04:49.980 result. Well, that's what they're doing in the partial adjustment is they're actually doing their

01:04:57.040 best to say – Oh, right. Married or not married. They're going variable by variable.

01:05:00.960 They're going drug by drug all the way through. High blood pressure, non-high blood pressure,

01:05:04.060 smoking, non-smoking, et cetera. Right. And the way they would do that presumably is

01:05:07.180 by saying, okay, married, not married. That's a simple one.

01:05:10.740 Are you on lipid-lowering meds? Yes or no? Okay. You are not. You are not.

01:05:18.160 And then comparing those groups. Yeah. Yeah. Okay. So no differences jumping out that can be purely

01:05:24.240 explained by these other variables. Yes. Although, again, this is a great opportunity to talk about

01:05:29.600 why no matter how slick you are, no matter how slick your model is, you can't control for everything.

01:05:33.900 There's a reason that, to my knowledge, virtually every study that compares meat eaters to non-meat

01:05:39.560 eaters finds an advantage amongst the non-meat eaters. And we can talk about all the – Lifespan

01:05:45.460 advantage. Yes. And we can – or disease, you know, incidence studies. And yeah, it might be

01:05:51.680 tempting to say, well, therefore, eating meat is bad until you realize that it takes a lot of work

01:05:57.700 to not eat meat. That's a very, very significant decision that a person – for most people,

01:06:02.600 that's a very significant decision a person makes. And for a person to make that decision,

01:06:06.320 they probably have a very high conviction about the benefit of that to their health.

01:06:10.660 And it is probably the case that they're making other changes with respect to their health as well

01:06:16.020 that are a little more difficult to measure. Now, there's a million other problems with that.

01:06:20.160 I picked a silly example because the whole meat discussion then gets into, well, you know,

01:06:25.040 when we say eating meat, what do we mean? Like –

01:06:27.220 You're not talking about it. It's like deli meat versus grass-fed.

01:06:29.500 Exactly.

01:06:29.940 Or a deer that you hunted with your ball.

01:06:32.680 That's right. So how do we get into all those things? But my point is it's very difficult

01:06:37.160 to quantify some of the intangible differences. And I think that even a study that goes to great

01:06:42.080 lengths, as this one does, epidemiologically to make these corrections can never make the

01:06:45.880 corrections. And so for me, the big takeaway of this study is, one, this makes much more sense to

01:06:52.980 me than the Bannister paper, which never really made sense to me. And again, I was first critical of

01:06:57.640 the Bannister paper in 2018, about four years after it came out. That's about the time I stopped

01:07:01.000 taking metformin, by the way. I stopped taking it for a different reason, which we can talk about

01:07:04.140 in a sec. But that was the first time I went back and said, wait a minute, this information – this

01:07:09.420 informative censoring thing is – that's a little fishy. And I think we weren't looking at a true

01:07:15.800 group of real type 2 diabetics. Now, that said, maybe it doesn't matter. In other words, maybe – and

01:07:22.280 even the Keyes paper doesn't tell us that metformin wouldn't be beneficial, because it could be

01:07:27.640 that those people, if they were on nothing, as their matched cohorts were on nothing, would have

01:07:33.860 been dying at, you know, a hazard ratio of 3. And this brought it down to 1.5, in which case you

01:07:39.860 would say there is some Giro protection there. It is putting the brakes on this process. All of this

01:07:45.420 is to say, absent a randomized control trial, we will never know the answer.

01:07:50.020 Has there been a randomized control trial in metformin?

01:07:52.780 Not when it comes to a hard outcome. Now, there has been in the ITP. So, the interventions

01:07:57.820 testing program, which is kind of the gold standard for animal studies, which is run out

01:08:04.980 of three labs. So, it's an NIH-funded program that's run out of three labs. They basically

01:08:11.160 test molecules for Giro protection. The ITP was the first study that really put rapamycin on

01:08:17.100 a map in 2009. That was the study that's fortuitously demonstrated that even when rapamycin was given

01:08:23.380 very, very late in life, it was given to 60-month-old mice, it still afforded them a 15% lifespan extension.

01:08:31.740 Has a similar study been done in humans? I mean, it's hard. I mean, it's hard to...

01:08:35.140 No, I mean, you can't really control with rapamycin.

01:08:36.980 No. But when the ITP studied metformin, it did not succeed. So, there have not been that

01:08:44.260 many drugs that have worked in the ITP. The ITP is very rigorous, right? It doesn't use an

01:08:50.920 inbred strain of mice. It is done concurrently in three labs with very large sample sizing.

01:08:57.320 And so, when something works in the ITP, it's pretty exciting. Rapamycin has been studied

01:09:02.040 several times. It's always worked. Another one we should talk about at a subsequent time is 17-alpha

01:09:08.600 estradiol. This continues to work in male mice. And it produces comparable effects

01:09:14.120 to rapamycin. Estrogen.

01:09:15.460 Doesn't work in female rice. But this is alpha, not beta. So, this is 17-alpha estradiol,

01:09:21.280 not beta estradiol, which is the estradiol that is bioavailable in all of us.

01:09:25.520 And just as a brief aside, I think you and I basically agree that unless it's a problem,

01:09:35.200 males, we're talking post-puberty, should try and have their estrogen as high as possible without

01:09:41.660 having negative symptomology because of the importance of estrogen for libido, for brain

01:09:45.580 function, tissue, bone health, bone health. This idea of crushing estrogen and raising

01:09:50.240 testosterone is just silly, right? Let's just leave raising testosterone out of it. But many

01:09:57.060 of the approaches to raising testosterone that are pharmacologic in nature also raise estrogen. A lot

01:10:01.460 of people try and push down on estrogen. And that is just, again, unless people are getting

01:10:06.860 hyperestrogenic effects like gynecomastia or other issues, is the exact wrong direction to go. You

01:10:13.220 want estrogen high. Estrogen is a very important hormone for men and women. Yeah, that's it.

01:10:20.180 Kinagaflozin, an SGLT2 inhibitor, also very successful in the ITP. But again, interestingly,

01:10:25.440 metformin not. So, metformin has failed in the ITP. So, you no longer take metformin?

01:10:31.420 I stopped five years ago. I mean, you're not a diabetic. So, presumably,

01:10:34.140 you were taking it for a gyroprotection. To buffer blood glucose in order to potentially

01:10:39.120 live longer. Yes, exactly. And the reason I stopped, and this will be the last thing before

01:10:42.940 we move on. Well, because you couldn't go to the dairy queen at the buffet of that.

01:10:46.320 No, finally, the nausea went away after a few weeks or a month maybe. But once I got really

01:10:52.880 into lactate testing, I noticed how high my lactate was at rest. So, a resting fasted lactate

01:11:01.560 in a healthy person should be below one, like somewhere between 0.3, 0.6 millimole. And only

01:11:07.960 when you start to exercise should lactate go up. And in 2018 was when I started blood testing for my

01:11:14.660 zone two. So, previously, when I was doing zone two testing, I was just going off my power meter

01:11:18.920 and heart rate. But this is after I'd met Inigo San Milan, and I started wanting to use the lactate

01:11:26.080 threshold of two millimole as my determinant of where to put my wattage on the bike. And I'm like

01:11:32.840 doing finger pricks before I start, and I'm like 1.6 millimole. And I'm like, what the hell is going

01:11:37.640 on? I can't be 1.6. So, if you ran a flight of stairs up the back of the Empire State Building.

01:11:42.400 Well, no, that would put me a lot higher, right? I was being generous to your fitness.

01:11:47.720 No, but that's when I started doing a little digging and realized, oh, you know what?

01:11:52.040 But this totally makes sense. If you have a weak mitochondrial toxin, what are you going to do?

01:11:58.740 You're going to shunt more glucose into pyruvate and more pyruvate into lactate. I'm anaerobic at a

01:12:06.140 baseline. Yeah, you need an alternative fuel source. That's right. And then my zone two numbers just

01:12:10.620 seemed off. My lactate seemed great. Could you feel it? Sorry to interrupt, but could you feel it in

01:12:14.340 your body? Because maybe now I'll just briefly describe. I took berberine. During the period of maybe

01:12:21.660 somewhere in the 2012 to 2015 stretch. I don't recall exactly. And what were you taking it for?

01:12:26.140 Well, I'll tell you. So, I was and I still am a big fan of Tim Ferriss' slow carbohydrate diet

01:12:32.500 because I like to eat meat and vegetables and starches. I'm an omnivore. And I found that it

01:12:38.660 worked very quickly, got me very lean. I could exercise. I could think. I could sleep. A lot of

01:12:45.520 my rationale for following one eating regimen or another, what I eat is to enjoy myself,

01:12:51.500 but also have mental energy. I mean, because if I can't sleep at night, I'm not going to replenish.

01:12:55.040 I'm not, if I don't replenish, I'm going to feel like garbage. I don't care how lean I am or what,

01:12:58.300 you know. So, I found the slow carb diet to be, which was in the four-hour body, to be a very good

01:13:04.000 plan for me. It was pretty easy. You drop some things like bread, et cetera. You don't drink calories

01:13:08.720 except after a resistance training session, et cetera. But one day a week, you have this so-called

01:13:14.680 cheat day. And on the cheat day, anything goes. And so, I would eat, you know, eight croissants and

01:13:20.040 then I'd alternate to sweet stuff. And then I'd go to a piece. And by the end of the day, you don't

01:13:23.220 want to look at an item of food at all. So, the only modification I made to the slow carb diet

01:13:27.600 for our body thing was the day after the cheat day, I wouldn't eat. I would just fast. And I had no

01:13:33.600 problem doing that because it was just basically, well, since you said, what was it? Anal-

01:13:39.500 Anal seepage.

01:13:40.120 Yeah. I did not have that. But since you said that, I won't up the ante here, but I'll at least

01:13:45.380 match your anal seepage comment by saying I had, let's just call it profound gastric distress after

01:13:50.940 eating like that the next day. So, the last thing you want to do is eat any food. I would just hydrate

01:13:53.960 and oftentimes to try and get some exercise. And what I read was that berberine, poor man's

01:14:01.300 metformin, could buffer blood glucose and in some ways make me feel less sick when ingesting all these

01:14:08.080 calories and in many cases spiking my blood sugar and insulin because you're having ice cream and

01:14:14.500 you know, et cetera. And indeed, it worked. So, if I took berberine and I don't recall the milligram

01:14:19.840 count and then I ate, you know, 12 donuts, I felt fine. It was as if I had eaten one donut.

01:14:26.280 Wow.

01:14:26.460 I felt sort of okay in my body and I felt much, much better. Now, presumably because it's buffering

01:14:32.360 the spikes in blood sugar, I wasn't crashing in the afternoon nap and that whole thing.

01:14:35.640 And do you remember how much you were taking?

01:14:38.000 I think it was a couple hundred milligrams. Does that sound about right?

01:14:41.220 It was a bright yellow capsule. I forget the source. But in any case, one thing I noticed was

01:14:47.580 that if I took berberine and I did not ingest a profound number of carbohydrates very soon

01:14:54.400 afterwards, I got brutal headaches. I think I was hypoglycemic. I didn't measure it, but I just felt I

01:14:59.340 had headaches. I didn't feel good. And then I would eat a pizza or two and feel fine. And so,

01:15:05.200 I realized that berberine was putting me on this lower blood sugar state. That was the logic anyway.

01:15:10.780 And it allowed me to eat these cheap foods. But when I cycled off of the four out, because I don't

01:15:17.940 follow the slow carb diet anymore, although I might again at some point, when I stopped doing those

01:15:21.920 cheat days, I didn't have any reason to take the berberine. And I feared that I wasn't ingesting

01:15:27.700 enough carbohydrates in order to really justify trying to buffer my blood glucose. Also,

01:15:31.380 my blood glucose tends to be fairly low. Did you ever try acarbose?

01:15:35.060 No. What is that?

01:15:36.360 So acarbose is-

01:15:36.960 Another glucose disposal.

01:15:38.480 Yeah. It's actually a drug that, but it works more in the gut and it just prevents glucose

01:15:42.840 absorption. Acarbose is another one of those drugs that actually found a survival benefit

01:15:48.280 in the ITP. And it was a very interesting finding because the thesis for testing it,

01:15:55.640 the ITP is a very clever system. Anybody can nominate a candidate to be tested. And then the panel over

01:16:01.180 there reviews it and they decide, yep, this is interesting. We'll go ahead and study it.

01:16:03.640 So when I think David Allison nominated acarbose to be studied, the rationale was it would be a

01:16:11.020 caloric restriction mimetic because you would literally just fail to absorb, I don't know,

01:16:16.360 make up some number, right? 15 to 20% of your carbohydrates would not be absorbed.

01:16:20.400 And therefore you would, the mice would effectively be calorically restricted.

01:16:24.000 They would just pass them out.

01:16:24.920 That's right. And what happened was really interesting. One, the mice lived longer on acarbose,

01:16:31.000 but two, they didn't weigh any less. So it, what it, they lived longer, but not through calorie

01:16:37.440 restriction. That's interesting. Yes. And it, the, the, the speculation is they lived longer

01:16:41.940 because they had lower glucose and lower insulin. And I don't want to send us down some rabbit holes

01:16:47.340 here, but there are all sorts of interesting ideas about, um, for instance, that some forms of dementia

01:16:53.520 might be so-called type three diabetes, the diabetes of the brain. And so things like berberine,

01:16:58.080 metformin, lowering blood glucose, ketogenic diets, et cetera, might be beneficial there. I mean,

01:17:03.020 there's a lot to explore here. And I know you've explored a lot of that on your podcast. I've done

01:17:06.700 far less of that, but well, at least it seems that we know the following things for sure. One,

01:17:12.100 you don't want insulin too high nor too low. You don't want blood glucose too high nor too low.

01:17:18.100 If the buffering systems for that are disrupted, clearly exercise, meaning regular exercise is the

01:17:24.520 best way to keep that system in check. But in the absence of that tool, or I would say in addition

01:17:31.240 to that tool, is there any glucose disposal agent, because that's what we're talking about here,

01:17:36.780 metformin, berberine, acarbose, et cetera, that you take on a regular basis because you have that much

01:17:42.900 confidence in it? The only one that I take is an SGLT2 inhibitor. Um, so this is a class of drug

01:17:52.140 that is used by people with type two diabetes, but which I don't have, but because of my faith

01:17:58.300 in the mechanistic studies of this drug, coupled with its results in the ITP, coupled with the human

01:18:03.540 trial results that show profound benefit in non-diabetics taking it even for heart failure.

01:18:09.580 I think there's something very special about that drug. I've actually, that was another paper I was

01:18:13.120 thinking about presenting this time. Maybe we'll do that the next time.

01:18:15.360 But do you believe in caloric restriction as a way to extend life? Or are you more of the,

01:18:23.820 do the right behaviors? Um, and that's covered in your book, Outlive and elsewhere on your podcast,

01:18:29.300 um, and buffer blood glucose is, do you still, obviously you, you believe in buffering blood

01:18:36.380 glucose in addition to just doing all the right behaviors.

01:18:38.760 Yeah. I think you can uncouple a little bit, the buffering of blood glucose from the caloric

01:18:42.840 deficit. So, um, I think you can be in a reasonable energy balance and buffer glucose with good sleep

01:18:49.080 hygiene, lots of exercise and just thoughtful eating, uh, without having to go into a calorie

01:18:55.340 deficit. So, you know, it's not entirely clear if profound caloric restriction would offer a survival

01:19:01.740 advantage to humans, even if it were tolerable to most, which it's not right. So for most people,

01:19:06.160 it's just kind of off the table, right? Like if I said, Andrew, you need to eat 30% fewer calories

01:19:11.000 for the rest of your life. I'll, I'll live 30% fewer years. Thank you. Yeah. Like there's just

01:19:15.020 not many people who are willing to sign up for that. So it's kind of a moot point. Um, but the

01:19:20.620 question is, you know, do you need to be fasting all the time? Do you need to be doing all of these

01:19:25.760 other things? And the answer appears to be outside of using them as tools to manage energy balance.

01:19:32.480 It's not clear, right? And energy balance probably plays a greater role in glucose homeostasis than,

01:19:41.560 uh, from a nutrition standpoint, than the individual constituents of the meal. Um,

01:19:47.320 now that's not entirely true. Like I can imagine a scenario where a person could be in a negative

01:19:51.700 energy balance, eating Twix bars all day and drinking, you know, big gulps. But I also don't

01:19:57.760 think that's a very sustainable thing to do because if by definition, I'm going to put you in negative

01:20:01.940 energy balance, consuming that much crap, I'm going to destroy you. Like you're going to feel

01:20:07.260 so miserable. You're going to be starving, right? You're not going to be satiated eating pure garbage

01:20:13.560 and being in caloric deficit. You're going to end up having to go into caloric excess.

01:20:19.200 So that's why it's interesting thought experiment. I don't think it's a very practical experiment

01:20:22.960 for a person to be generally satiated and an energy balance. They're probably eating about the right

01:20:27.480 stuff, but I don't think that the specific macros matter as much as I used to think.

01:20:33.120 I'm a believer in getting most of my nutrients from unprocessed or minimally processed sources

01:20:39.720 simply because it allows me to eat foods I like and more of them. And I just love to eat. I so

01:20:48.760 physically enjoy the sensation of chewing that, you know, I'll just eat cucumber slices for,

01:20:53.520 for fun. Yeah. Right. You know, that's, I mean, that's not my only form of fun, fortunately.

01:21:01.360 This is an amazing paper for the simple reason that it provides a wonderful tutorial of the

01:21:09.820 benefits and drawbacks of this type of work. And I think it's also wonderful because we hear a lot

01:21:16.080 about metformin, rapamycin, and these anti-aging approaches, but I was not aware that there was

01:21:24.420 any study of such a large population of people. So it's pretty interesting.

01:21:28.300 Yeah. So I think it remains to be seen if, and my patients often ask me, hey, should I be on

01:21:32.840 metformin? And I give them a much, much, much, much shorter version of what we just talked about.

01:21:37.100 And I say, look, if the TAME study, which should answer this question more definitively, right? This

01:21:42.940 is taking a group of non-diabetics and randomizing them to placebo versus metformin and studying for

01:21:49.720 specific disease outcomes. If the TAME study ends up demonstrating that there is a gyroprotective

01:21:57.260 benefit of metformin, I'll reconsider everything, right? So I think that's, you know, we just have to,

01:22:02.420 I think all walk around with an appropriate degree of humility around what we know and what we don't

01:22:06.940 know. But I would say right now, the epidemiology, the animal data, my own personal experience with

01:22:13.300 its impact on my lactate production and exercise performance, we could, there's a whole other

01:22:17.940 rabbit hole we could go down another time, which is the impact on hypertrophy and strength, which

01:22:21.640 appears to be attenuated as well by metformin. You know, I'll, I'll, I'll, I still prescribe it to

01:22:28.280 patients all the time if they're insulin resistant, for sure. It's still a valuable drug, but I don't think

01:22:32.700 of it as a great tool for the person who's insulin sensitive and exercising a lot.

01:22:37.720 I can't help but ask this question. Do you think there's any longevity benefit to short periods of

01:22:46.460 caloric restriction? You know, so for instance, I decide to, by the way, I haven't done this,

01:22:53.520 but let's say I were to decide to, you know, fast and do a one meal a day type thing where I'm going

01:22:59.800 to be in a slight caloric deficit, you know, 500 to a thousand calories for a couple of days and then

01:23:05.040 go back to eating the way that I ate before that short caloric restriction slash fast. Is there any

01:23:12.220 benefit to it in terms of cellular health? Can you, you know, sort of reset the system? Is there any

01:23:16.980 idea that the, the changes, the clearing of senescent cells that we hear about autophagy that we,

01:23:22.560 you know, that in the short term, you can glean a lot of benefits and then go back to the, to your

01:23:27.100 regular pattern of eating. And then periodically, you know, once every couple of weeks or once a month,

01:23:32.160 just, you know, fast for a day or two, is there any benefit to that? That's, that's purely in the

01:23:36.980 domain of longevity, not because there's all discipline function there. There's a flexibility

01:23:43.320 function. There's probably an insulin sensitivity function, but is there any evidence that it can

01:23:47.120 help us live longer? I think the short answer is no. Um, for two reasons. One, I don't think that

01:23:53.880 that duration would be sufficient if, if one is going to take that approach, but two, um, even if you

01:23:59.780 went with something longer, like what I used to do, right? I used to do seven days of water only per

01:24:04.500 quarter, three days per month. So I was, but I was basically always like, it'd be three day fast,

01:24:10.220 three day fast, seven day fast. Just imagine doing that all year, rotating, rotating, running for many

01:24:15.180 years. I did that. Um, now I certainly believed. And to this day, I would say I have no idea if that

01:24:20.800 provided a benefit. Um, but my thesis was, uh, the downside of this is relatively circumscribed,

01:24:27.620 which is profound misery for a few days. And, um, what I didn't appreciate the time, which I

01:24:34.120 obviously now look back at and realize is muscle mass lost. You're just, it's very difficult to

01:24:38.860 gain back the muscle cumulatively after all of that loss. Um, but my thought was exactly, as you said,

01:24:45.100 like there's got to be a resetting of the system here. This must be sufficiently long enough to trigger

01:24:50.360 all of those systems, but you're getting at a bigger problem with gyroscience, which I'm really

01:24:58.160 hoping the epigenetic field comes to the rescue on. It has not come close to it to date, which is we

01:25:04.720 don't have biomarkers around true metrics of aging. Everything we have to date stinks. So we're really

01:25:13.780 good at using molecules or interventions for which we have biomarkers, right? Like when you lift weights,

01:25:22.280 you can look at how much weight you're lifting. You can look at your DEXA scan and see how much muscle

01:25:27.100 mass you're generating. Like that. Those are biomarkers. Those are giving you outputs that say

01:25:31.960 my input is good, or my input needs to be modified. Um, when you take a sleep supplement, you can look at

01:25:38.340 your eight sleep and go, Oh, my sleep is getting better. Like there's a biomarker. Um,

01:25:43.780 when you take metformin, when you take rapamycin, when you fast, we don't have a biomarker that gives

01:25:51.520 us any insight into whether or not we're moving in the right direction. And if we are, are we taking

01:25:56.480 enough? We just don't know. So I, I often get asked like, what's the single most important topic you

01:26:04.160 would want to see more research dollars put to in terms of this space. And it's unquestionably this

01:26:10.740 as unsexy as it is, like who cares about biomarkers, but like without them, I don't

01:26:16.600 think we're going to get great answers because you can't do most of the experiments you and I would

01:26:21.540 dream up. Got it. Well, I'm grateful that you're sitting across the table for me telling me all this

01:26:28.940 and that, um, everyone can hear this. Uh, but again, we will put a link to the papers plural that,

01:26:36.120 uh, Peter just described. And for those of you that are listening and not watching,

01:26:40.220 um, hopefully you were able to track the general, um, themes and takeaways. And, um, it is fun to go

01:26:45.500 to these papers. You see these big stacks of numbers and it can be a little bit overwhelming,

01:26:49.480 but, um, my, uh, additional suggestion on parsing papers is notice that Peter said that he spent,

01:26:56.600 you know, he's read it several times. Unlike a newspaper article or, or a Instagram post with a paper,

01:27:04.000 you're not necessarily going to get it the first time. You certainly won't get everything so that

01:27:09.120 I, I think spending some time with papers for me means reading it and then reading it again a little

01:27:13.200 bit later or, you know, one figure at a time. Yeah. I was just about to say, what's your,

01:27:16.960 cause, cause I kind of have a way that I do it, but I'm curious as to how you do it. Like if you're,

01:27:20.780 if you're encountering a paper for the first time, what do you have an order in which you like to go

01:27:24.560 through the, do you, do you want it? Do you read it sequentially or do you look at the figures first? I

01:27:29.080 mean, how do you, how do you go through it? Yeah. Unless it's an area that I know very,

01:27:31.940 very well where I can, you know, skip to some things before reading it the whole way through.

01:27:38.660 My process is always the same. And actually this is fun because I used to teach a class when I was

01:27:43.480 a professor at UC San Diego called neural circuits and health and disease. And it was an evening course

01:27:48.380 that grew very quickly from 50 students to 400 plus students. And we would do exactly this. We would

01:27:54.220 parse papers. And, um, and I had everyone ask what I called the four questions. Um, and it wasn't

01:28:01.440 exactly four questions, but I have a little three by five card next to me or a piece of a main half by

01:28:07.280 11 paper typically. And when I sit down with a paper, I want to figure out what is the question

01:28:12.820 they're asking? What's the general question? What's the specific question? And I write down the

01:28:16.940 question. Then what was the approach? You know, how did they test that question? And sometimes that can

01:28:21.960 get a bit detailed, you can get into immunohistochemistry and they did a, you know, PCR for

01:28:25.760 this. It's not so important for most people that they understand every method, but it is worthwhile

01:28:32.580 that if you encounter a method like PCR or, um, you know, chromatography or fMRI that you at least

01:28:40.720 look up on the internet, what its purpose is. Okay. That will help a lot. And then it was what they

01:28:45.440 found. And there, um, you can usually figure out what they believe they found anyway, by reading

01:28:50.600 the figure headers, right? What are, you know, figure one, here's the header that typically if

01:28:56.040 it's an experimental paper, it will tell you what they want you to think they found. And then I tend

01:29:01.580 to want to know the conclusion of the study. And then this is really the key one. And this is the

01:29:05.820 one that, um, would really distinguish the high performing students from the others. You have to go

01:29:12.200 back at the end and ask whether or not the conclusions, the major conclusions drawn in the paper

01:29:16.360 are really substantiated by what they found and what they did. And that involves some thinking.

01:29:21.400 It involves really, you know, spending some time thinking about what, what they identified.

01:29:24.980 Now, this isn't something that anyone can do straight off the bat. It's a skill that you develop

01:29:28.140 over time and different papers require different formats. But those four questions really form the

01:29:32.740 cornerstone of a, of teaching undergraduates. And I think graduate students as well of how to

01:29:36.540 read a paper. And, um, again, it's something that can be cultivated. Um, and it's still how I approach

01:29:45.920 papers. So what I do typically is I'll read title abstract. I usually then will skip to the figures

01:29:52.400 and see how much of it I can digest without reading the text and then go back and read the text.

01:29:57.780 But in fairness, journals, great journals like science, like nature's oftentimes will pack so much

01:30:04.000 information in the cell press journals to into each figure. And it's coded with no definition of

01:30:09.160 the acronyms that almost always I'm into the introduction and results within a couple of

01:30:13.880 minutes, wondering what the hell this acronym is or that acronym is. And it's, um, it's just,

01:30:18.560 yeah, it's just wild how much, um, how much nomenclature there really is. I can't remember,

01:30:24.380 was it you or was it our friend, Paul Conti, when he was here, um, who said that, oh no, I'm sorry.

01:30:29.940 It was neither. It was chair of ophthalmology at Stanford. Uh, Dr. Jeffrey Goldberg, who was a

01:30:34.380 guest on the podcast recently who off camera, I think it was told us that if you look at the total

01:30:40.640 number of words and terms that a physician leaving medical school owns in their mind and their

01:30:47.560 vocabulary, it's the equivalent of like two additional full languages of fluency beyond

01:30:53.420 their native language. So you're trilingual at least. I don't know. Do you speak a language other

01:30:57.980 than English poorly? Okay. So you're, you're, you're at least trilingual and probably more. So

01:31:03.920 no one is expected to be able to parse these papers the first time through without, you know,

01:31:08.960 substantial training. Yeah, no, I, I, I think that's a, that's a great format and you're

01:31:13.960 absolutely right. I have a different way that I do it when I'm familiar with the subject matter

01:31:18.440 versus when I'm not. Uh, well, again, if I'm reading papers that are something that I know really

01:31:23.820 well, I can basically glean everything I need to know from the figures. Um, and then sometimes I'll

01:31:29.340 just do a quick skim on methods. Um, but I don't need to read the discussion. I don't need to read

01:31:33.860 the intro. I don't need to read anything else. Uh, if it's something that I know less about,

01:31:37.920 then I usually do exactly what you say. I try to start with the figures. I usually end up generating

01:31:45.460 more questions like what, what, what do you mean? What, what is this? How did they do that? Uh,

01:31:50.760 and then I got to go back and read methods typically. And one of the other thing that's

01:31:55.100 probably worth mentioning is a lot of papers these days have supplemental information that are not

01:31:59.220 attached to the paper. So, um, you're amazed at how much stuff gets put in the supplemental section.

01:32:04.880 And the reason for that, of course, is that the journals are very, uh, specific on the format and

01:32:10.180 length of a paper. So a lot of the times when you're submitting something, you know, like

01:32:14.880 if you want to put any additional information in there, it can't go in the main article. It has

01:32:18.540 to go in the supplemental figure. So even for this paper, there were a couple of the numbers I spouted

01:32:22.780 off that I had to pull out of the supplemental paper. For example, when they did the sensitivity

01:32:27.620 analysis on the, um, censoring versus non-censoring, that, that was in the supplemental figure. That was

01:32:35.340 actually not even in the paper we presented. Well, should we pivot to this other paper? Yeah. It's a

01:32:41.980 very different sort of paper. It's an experimental paper where there's a manipulation. I must say,

01:32:47.040 I love, love, love this paper. And I don't often say that about papers. I'm so excited about this

01:32:53.540 paper for so many reasons, but I want to give a couple of caveats up front. First of all, the paper

01:32:59.400 is not published yet. The only reason I was able to get this paper is because it's on bio archive.

01:33:05.540 There's a new trend over the last, I would say five, six years of people posting the papers that

01:33:10.940 they've submitted to journals for peer review online so that people can look at them prior to

01:33:16.060 those papers being peer reviewed. So there is a strong possibility that the final version of this

01:33:21.060 paper, which again, we will provide a link to is going to look different, maybe even quite a bit

01:33:25.340 different than the one that we're going to discuss. Nonetheless, there are a couple of things that make

01:33:30.260 me confident in the data that we're about to talk about. First of all, the group that published this

01:33:36.100 paper is really playing in their wheelhouse. This is what they do. And they publish a lot of really

01:33:40.880 nice papers in this area. I'm going to mispronounce her first name, but I think it's

01:33:46.860 Chao Si Gu, who's at the Econ School of Medicine in Mount Sinai, runs a laboratory there studying

01:33:53.960 addiction in humans. And the first author of the paper is Ofer Pearl. This paper is wild. And I'll

01:34:02.940 just give you a couple of the takeaways first as a bit of a hook to hopefully entice people into

01:34:07.620 listening further, because this is an important paper. This paper basically addresses how our

01:34:13.880 beliefs about the drugs we take impacts how they affect us at a real level, not just at a subjective

01:34:24.020 level, but at a biological level. So just to back up a little bit, a former guest on this podcast,

01:34:29.040 Dr. Ali Crum, whose name is actually Aaliyah Crum, but she goes by Ali Crum, talked about belief

01:34:35.980 effects. Belief effects are different than placebo effects. Placebo effects are really just

01:34:41.540 category effects. It's, okay, I'm going to give you this pill, Peter, and I'm going to tell you that

01:34:47.460 this pill is molecule X5952, and that it's going to make your memory better. And then I give you a

01:34:54.680 memory test, right? And your group performs better than the people in the control group who I give a

01:34:59.380 pill to and I say, this is just a placebo. Or there are other variants on this where people will get a

01:35:06.340 drug and you tell them it's placebo. They'll get a placebo, you tell them it's drug. It's a binary

01:35:11.840 thing. It's an on or an off thing. You're either in the drug group or the placebo group, and you're

01:35:15.600 either told that you're getting drug or placebo. And we know that placebo effects exist. In fact,

01:35:20.740 one of the crueler ones, I was never the subject of this, but there was kind of lore in high school

01:35:24.460 that kids would do this mean thing. It's a form of bullying. I really don't like it where they get

01:35:29.180 some kid at a party to drink alcohol-free beer, and then that kid would start acting drunk, and

01:35:35.580 then they'd go, gotcha. It doesn't even have alcohol in it. Now, that's a mean joke and just

01:35:41.720 reminds me of some of the horrors of high school. Maybe that's why I didn't go very often, which I

01:35:46.120 also don't suggest. But no, it's a mean joke, but it speaks to the placebo effect, right? And there's

01:35:50.980 also a social context effect. So placebo effects are real. We know this. Belief effects are different.

01:35:58.500 Belief effects are not A or B, placebo or non-placebo. Belief effects have a lot of

01:36:04.460 knowledge to enrich one's belief about a certain something that can shift their psychology and

01:36:11.560 physiology one way or the other. And I think the best examples of these belief effects really do

01:36:16.920 come from Allie Crump's lab in the psychology department at Stanford, although some of this work

01:36:20.660 she did prior to getting to Stanford. For instance, if people are put into a group where they watch a

01:36:26.540 brief video, just a few minutes of video about how stress really limits our performance, let's say

01:36:31.720 at archery or at mathematics or at music or at public speaking, and then you test them in any of those

01:36:38.120 domains or other domains in a stressful circumstance, they perform less well. And we know they perform less

01:36:45.520 well because by virtue of a heightened stress response. You can measure heart rate. You can

01:36:51.200 measure stroke volume of the heart. You can measure peripheral blood flow, which goes down when people

01:36:55.360 are stressed, narrowing a vision, et cetera. You take a different group of people and randomly assign

01:37:01.000 them to another group where now they're being told that stress enhances performance. It mobilizes

01:37:07.680 resources. It narrows your vision such that you can perform tasks better, et cetera, et cetera.

01:37:11.580 And their performance increases above a control group that receives just useless information or

01:37:16.600 at least useless as it relates to the task. So in both cases, by the way, the groups are being told

01:37:21.320 the truth. Stress can be depleting or it can enhance performance. But this is different than

01:37:27.200 placebo because now it's scaling according to the amount and the type of information that they're

01:37:32.220 getting. And can you give me a sense of magnitude of benefit or detriment that one could experience in

01:37:37.500 a situation like the one you just described? Yeah. So it's striking. They're opposite

01:37:41.440 in direction. So the stress gets us worse, makes you, let's say, I think that if we were to just put

01:37:47.240 a rough percentage on this, it would be somewhere between 10 and 30% worse at performance than the

01:37:52.160 control group. And stress is enhancing is approximately equivalent improvement. So they're

01:37:57.200 in opposite directions. Even more striking is the studies that Ali's lab did and others looking at,

01:38:04.920 for instance, you give people a milkshake, you tell them it's a high calorie milkshake,

01:38:08.140 has a lot of nutrients, and then you measure ghrelin secretion in the blood. And ghrelin is a marker

01:38:13.760 of hunger that increases the longer it's been since you've eaten. And what you notice is that

01:38:17.720 it suppresses ghrelin to a great degree and for a long period of time. You give another group a shake,

01:38:23.180 you tell them it's a low calorie shake, that it's got some nutrients in it, but that doesn't have

01:38:27.360 much fat, not much sugar, et cetera. They drink the shake, less ghrelin suppression.

01:38:32.340 And it's the same shake.

01:38:33.420 And it's the same shake. And satiety lines up with that also in that study. And then the third one,

01:38:39.360 which is also pretty striking is they took hotel workers, they gave them a short tutorial or not,

01:38:44.020 informing them that moving around during the day and vacuuming and doing all that kind of thing is

01:38:47.340 great. It helps you lower your BMI, which is great for your health. You incentivize them.

01:38:51.660 And then you let them out into the wild of their everyday job. You measure their activity levels.

01:38:56.500 The two groups don't differ. They're doing roughly the same task, leaning down,

01:38:59.400 cleaning out trash cans, et cetera. Guess what? The group that was informed about the health

01:39:03.440 benefits of exercise lose 12% more weight compared to the other group.

01:39:10.000 And no difference in actual movement?

01:39:12.360 Apparently not. Now, how could that be? I mean, literally this was sparked by, in Ali's words,

01:39:19.000 this was sparked by her graduate advisor saying, what if all the effects of exercise are placebo?

01:39:25.560 Right? Like, which is, which is not what anyone really believes, but it's just such a,

01:39:29.780 you know, I love that anecdote that Ali told us because it just really speaks to how like really

01:39:35.080 smart people think. They sit back and they go, yeah, like exercise obviously has benefits,

01:39:38.820 but like, what if a lot of the benefits are that you tell yourself it's good for you and the brain

01:39:41.860 can actually activate these, these mechanisms in the body? And why wouldn't that be the case?

01:39:46.300 Because the nervous system extends through both.

01:39:47.820 So, so interesting. So interesting. Okay. So fast forward to this study, which is really about

01:39:55.080 belief effects, not placebo effects. And to make a long story short, we know that nicotine,

01:40:02.860 vaped, smoked, dipped, or snuffed, or these little ZIN pouches or taken in capsule form

01:40:07.700 does improve cognitive performance. I'm not suggesting people run out and start doing any

01:40:11.920 of those things. I did a whole episode on nicotine. The delivery device often will kill you

01:40:15.260 some other way or is bad for you, but it causes vasoconstriction, which is also not good for

01:40:19.840 certain people, but nicotine is cognitive enhancing. Why? Well, you have a couple of sites in the brain,

01:40:25.360 namely in the basal forebrain, nucleus basalis, in the back of the brain structures like locus

01:40:32.340 coeruleus, but also this, what's called, it's got a funny name, the pedunculopontine nucleus,

01:40:36.860 which is this nucleus in the, in the, the pons, in the back of the brain, in the brainstem that sends

01:40:42.280 those little axon wires into the thalamus. The thalamus is a gateway for sensory information.

01:40:46.680 And in the thalamus, the visual information, the auditory information, it has nicotinic receptors.

01:40:52.820 And when the pedunculopontine nucleus releases nicotine, or when you ingest nicotine, what it

01:40:57.780 does is it increases the signal to noise of information coming in through your senses.

01:41:02.740 So the fidelity of the signal that gets up to your cortex, which is your conscious perception of

01:41:07.280 those senses is increased. And how much endogenous nicotine do we produce?

01:41:11.960 Ooh. Well, it's going to be acetylcholine binding to nicotinic receptors.

01:41:16.260 I see. We're not making nicotine. We're not making nicotine.

01:41:18.340 So this is a, this is a nicotinic acetylcholine receptor.

01:41:21.600 Right. Of which there are at least seven and probably like 14 subtypes. But so, right. They're

01:41:28.280 called nicotinic receptors in an annoying way, in the same way that cannabinoid receptors are called

01:41:32.680 cannabinoid receptors. But then everyone thinks, oh, you know, those receptors are there

01:41:36.240 because we're supposed to smoke pot or those receptors are there because we're supposed to

01:41:39.700 ingest nicotine. No, the drugs that we use to study them.

01:41:42.580 The drug is named after the receptor.

01:41:43.200 That's right. Exactly. Receptor is named after the drug. And so the important thing to know is

01:41:48.400 that whether or not it's basal forebrain or pedunculopontine nucleus or a locus coeruleus

01:41:52.520 that at least in the brain, because we're not talking about muscle where acetylcholine does

01:41:56.180 something else via nicotinic receptors, there in general, it just tends to be a signal to noise

01:42:01.260 enhancer. And so for the non-engineering types out there, no problem. Signal to noise,

01:42:05.980 just imagine I'm talking right now and there's a lot of static in the background.

01:42:09.440 There are two ways for you to be able to hear me more clearly. We can reduce the static or I can

01:42:13.340 increase the fidelity, the volume and the clarity of what I'm saying. Okay. For instance, and that's

01:42:21.460 really what acetylcholine does. That's why when people smoke a cigarette, they get that boost of

01:42:24.980 nicotine and they just feel clearer. It really works. The other thing that happens is the thalamus

01:42:31.280 sends information to a couple of places. First of all, it sends information to the reward centers

01:42:36.520 of the brain, the mesolimbic reward pathway that releases dopamine. And typically when nicotine

01:42:40.980 is increased in our system, dopamine goes up. That's one of the reasons why nicotine is reinforcing.

01:42:45.940 We just like it. We seek it out. I've done beautiful experiments with honeybees even where

01:42:50.880 you put nicotine on certain plants or it comes from certain plants and they'll forage there more.

01:42:55.160 You get them kind of like buzzed. That was a pun, bad pun. In any event, there's also an output from

01:43:01.720 this thing, the thalamus to the ventromedial prefrontal cortex, which is an area of the

01:43:06.480 forebrain that really allows us to limit our focus and our attention for sake of learning. It allows us

01:43:11.920 to pay attention. This is the circuit. You talked about this in your fantastic podcast on stimulants.

01:43:18.180 Yeah. So typically ADHD drugs. So things like Adderall, Vyvanse, methamphetamine for that matter,

01:43:26.400 Ritalin. Yeah. Why it's counterintuitive that a stimulant would be a treatment for someone with

01:43:32.800 difficulty focusing. Yeah. In young kids who have difficulty focusing, if you give them something

01:43:37.700 they love, they're like a laser. And the reason is that ventromedial prefrontal cortex circuit can

01:43:44.880 engages when the kid is interested and engaged. But kids with ADD, ADHD tend to have a hard time

01:43:51.660 engaging their mind for other types of tasks and other types of tasks are important for getting

01:43:55.700 through life. And it turns out that giving those stimulant drugs in many cases can enhance the

01:44:00.940 function of that circuit and it can strengthen so that ideally the kids don't need the drugs in the

01:44:06.300 long run, although that's not often the way that it plays out. And there are other ways to get at

01:44:10.960 this. There's now a big battle out there. Is ADHD real? Is it not real? Of course it's real. Does

01:44:15.960 every kid need ADHD meds? No. Are there other things like nutrition, more playtime outside,

01:44:21.520 et cetera, that can help improve their symptoms without drugs? Yes. Is the combination of all

01:44:26.140 those things together known to be most beneficial? Yes. Are the dosages given too high and generally

01:44:32.420 should be titrated down? Maybe. Some kids need a lot, some kids need a little. I probably just

01:44:37.720 you know, gained and lost a few enemies there. So the point is that these circuits are hardwired

01:44:44.480 circuits. Sorry, one other question, Andrew. If my memory serves correctly, doesn't nicotine

01:44:51.680 potentially have a calming effect as well? And that seems a bit counterintuitive to the focusing

01:44:57.320 one. Is it a dose effect or a timing effect? How does that work? Yeah, it's a dosing effect. So the

01:45:02.580 interesting thing about nicotine is that it can enhance focus in the brain, but in the periphery,

01:45:07.500 it actually provides some muscle relaxation. So it's kind of the perfect drug if you think about

01:45:12.400 it. Again, it was reflecting on this, how when we were growing up, people would smoke on plane,

01:45:18.500 they had a smoking section on the plane. You know, people smoked all the time and now hardly anyone

01:45:22.660 smokes for all the obvious reasons. But yeah, it provides that really ideal balance between being

01:45:28.500 alert, but being mellow and relaxed in the body. So hence it's reinforcing properties. Okay. This

01:45:35.340 study is remarkable because what they did is they had people come into the laboratory. They gave them

01:45:42.780 a vape pen. These are smokers. So these are experienced smokers. Typically there's a washout

01:45:50.680 before they come in. So they're not smoking for a bit so they can clear their system of nicotine and

01:45:54.380 they measure. How long is that needed? Typically it's a couple of days. Okay. Yeah. Which must be

01:45:59.060 miserable for those people. Because they can't have Nicorette gum or anything. No, nothing. They

01:46:02.780 must be dying. And I wonder how many cheat. But they can measure carbon monoxide, right? Yeah. They

01:46:07.340 measure carbon monoxide and they're measuring nicotine in the blood as well. So they do a good

01:46:10.580 job there. So then what they do is they have them vape and they're vaping either a low, medium, or high

01:46:18.000 dose of nicotine. The doses just don't really matter because tolerance varies, et cetera. And

01:46:23.360 then they are putting them into a functional magnetic resonance imaging machine. So where they

01:46:29.440 can look at, it's really blood flow. It's really hemodynamic response. For those of you who want

01:46:33.580 to know, it's the ratio of the oxygenated to deoxygenated blood because when blood, blood will

01:46:39.380 flow to neurons that are active to give it oxygen and then it's deoxygenated. And then there's a

01:46:44.040 change in what's called the bold signal. So fMRI, when you see these like hotspots in the brain

01:46:48.880 is really just looking at blood flow. And then there's some interesting physics around and I'll

01:46:54.280 probably get this wrong, but I'll take an attempt at it so that I get beat up a little bit by the

01:46:57.260 physicists and engineers. Do you remember the right hand rule? Yep. Right. Okay. So do I have

01:47:01.200 this right? Correct. The right hand rule. If you put your thumb out with your first, with your index

01:47:05.500 finger, your middle finger, your thumb facing up, I think that the thumb represents the charge,

01:47:09.260 the direction of the charge. Right. And then isn't the electromagnetic field is the

01:47:13.120 downward facing figure. And then it's, do I have that right? I have to look this up. I actually

01:47:18.740 don't. Okay. Well, someone will look it up. But what you do is when you put a person's head in

01:47:22.280 this big magnet and then you pulse the magnet, what happens is the oxygenated and deoxygenated blood,

01:47:27.980 it interacts with the magnetic field differently. And that difference in signal can be detected.

01:47:33.580 And you can see that in the form of activated brain areas. Yeah. I mean, MRI all works by proton

01:47:39.080 detection. So presumably there's a difference in the proton signal when you have high oxygen versus

01:47:45.160 low oxygen concentration. Yeah. That's right. And what they'll do is they'll pulse with the magnet

01:47:49.820 because my understanding is that, and this is definitely getting beyond my expertise, but that

01:47:54.980 the spin orientation of the protons, then it's going to relax back at a different rate as well. So

01:48:00.780 by the relaxation at a different rate, you can also get not just resting state activation, like,

01:48:06.700 look at a banana, what areas of the brain light up, but you can look at connectivity between areas and

01:48:12.400 how one area is driving the activity of another area. So very, very powerful technique. Um, so what

01:48:18.420 they do is they, they put people in a scanner and then you'll like this. Cause what are the,

01:48:21.540 what are the limitations of, of fMRI in terms of, I mean, how fine is the resolution? I mean,

01:48:27.300 where are the blind spots of the technique? So resolution, you can get down to sub centimeter.

01:48:32.600 They talk about it always in these paper as a voxels, which are these little cubic pixels,

01:48:36.700 um, things, um, uh, you know, sub sub centimeter, but you're not going to get down to millimeter.

01:48:42.660 Okay. Um, there are a number of little confounds that maybe we won't go into now

01:48:47.620 that have been basically worked out over the last 10 years by doing the following. You can't just give

01:48:52.800 somebody a stimulus compared to nothing. I'll just tell you the experiment. It was discovered for

01:48:57.620 instance, that when someone would move their right hand, cause you, when you're in the MRI and just went

01:49:02.140 for one of these recently for clinical, not a problem, but just for a diagnostic scan, you're

01:49:06.120 leaning back and you, and you can move your right hand a bit and they would see an area in motor

01:49:10.660 cortex lighting up. But what they noticed was that the area corresponding to the left hand was also

01:49:15.040 lighting up. So what you really have to do is you have to look at resting state. How much are they

01:49:20.720 lighting up just at rest and then subtract that out. So now you'll always see resting state versus

01:49:26.680 activation state. Yeah. Wasn't there a really funny study done as a spoof, maybe a decade ago that

01:49:33.400 put a dead salmon into an MRI machine and did an F like they did an fMRI of a dead salmon that

01:49:40.220 demonstrated like some interesting signal. No, I didn't know that, but, but we got to find this

01:49:45.700 one for the, for the show notes. We should do one of these wild papers ones. There's, there are papers

01:49:50.980 of, you know, people putting, don't do this folks, putting elephants on LSD that were published in

01:49:55.040 science and things like, like crazy experiments. We should definitely do a crazy experiments journal

01:49:58.960 club. Um, in any event, you can get a sense of which brain areas are active and when with fairly

01:50:05.400 high spatial resolution, fairly high and pretty good temporal resolution on the order of hundreds

01:50:10.900 of milliseconds. Not, but it's not ultra, ultra fast because a lot of neural transmission is

01:50:16.980 happening on the tens of milliseconds. Um, especially when you're in talking about auditory

01:50:21.460 processing. Okay. So they put people into the scanner and then they give them a, essentially

01:50:27.340 a task that's designed to engage the thalamus known to engage the thalamus reward centers and

01:50:34.360 the ventromedial prefrontal cortex. And it's a very simple game. You'll like this because, um,

01:50:38.460 you have a background in finance. You let people watch a market, you know, okay, here's the stock

01:50:44.000 market, or you could say that, or the price of peas, it doesn't really matter. It goes up,

01:50:47.880 it goes down and they're looking at a squiggle line. Then it stops and then they have the option,

01:50:51.700 but they have to pick one option. They're either going to invest a certain number of the hundred

01:50:55.200 units that you've given them, or they can short it. They can say, oh, it's going to go down and

01:51:00.460 try and make money on the, on the prediction. It's going to go down. You could explain shorting

01:51:04.180 better than I could for sure. So depending on whether or not they get the prediction right or

01:51:08.600 wrong, they get more points or they lose points and they're going to be rewarded in real

01:51:12.640 money at the end of the experiment. So this is going to engage this type of circuitry. Now,

01:51:16.860 remember these groups were given a vape pen prior to this, where they've vaped. What they were told

01:51:24.320 is either a low medium or high dose of nicotine. And they do this task. The goal is not to get them

01:51:31.920 to perform better on the task. The goal is to engage the specific brain areas that are relevant to this

01:51:36.420 kind of error and reward type circuits. And we know that this task does that. So that includes the

01:51:42.040 thalamus that includes the mesolimbic reward pathway and dopamine. It includes the ventral

01:51:46.800 medial prefrontal cortex. First of all, they measure nicotine in the blood. They are measuring

01:51:52.960 how much people vaped. They were very careful about this. One of the nice things about the

01:51:56.320 vape pen for the sake of experiment and not recommending people vape, but they can measure

01:52:00.520 how much nicotine is left in the vape pen before, after they can measure how long they inhaled,

01:52:04.540 how long they held it in. There's a lot that you can do that's harder to do with a cigarette.

01:52:07.820 Okay. They measured people's belief as to whether or not they got low, medium, or high amounts of

01:52:15.340 nicotine. And it's- They were told.

01:52:17.660 They were told they got either this is a low amount, a medium amount, or a high amount.

01:52:21.940 And then, of course, they looked at brain area activation during this task. And what they found

01:52:27.300 was very straightforward. Sorry, they were all given the same amount.

01:52:29.860 Yes. This is the sneak. I was going to offer it as a punchline, but that's okay. No, I think that

01:52:34.260 the cool thing about this experiment is that the subjects are unaware that they all got the exact

01:52:39.720 same amount of relatively low nicotine-containing vape pen. So they basically, and they're measuring

01:52:46.480 it from their bloodstream. So they all have fairly low levels of nicotine, but one group was told you

01:52:51.360 got a lot. One group was told you got a medium amount, and the other was told you got a little bit.

01:52:55.300 Now, a number of things happen, but the most interesting things are the following. First

01:53:01.300 of all, people's subjective feeling of being on the drug matches what they were told. So if they

01:53:08.540 were told, hey, this is a high amount of nicotine, like, yeah, it feels like a high amount of nicotine.

01:53:12.640 And these are experienced smokers. If it was a medium amount, they're like, yeah, that feels like

01:53:16.400 a medium amount. If it's a low amount, they think it was a low amount. Now that's perhaps not so

01:53:22.320 surprising. That's you're just- That's the placebo in a sense.

01:53:24.720 Placebo. But if you look at the activation of the thalamus in the exact regions where you would

01:53:33.140 predict acetylcholine transmission to impact the function of the thalamus, so these include areas

01:53:38.220 like what's called the centromedian nucleus, the ventroposterior nucleus, the names that really don't

01:53:41.860 matter, but these are areas involved in attention. It scales with what they thought they got in the

01:53:49.540 vape pen. Meaning if you were told that you got a low amount of nicotine, you got a little bit of

01:53:53.320 activation in these areas. If you were told that you got a medium amount of nicotine, and that's

01:53:57.720 what you vaped, then you had medium amounts or moderate amounts of activation. And if you were

01:54:04.540 told you got high amounts of nicotine, you got a high degree of activation. And the performance on

01:54:09.580 the task, believe it or not, scales with it somewhat. So keep in mind, everyone got the exact same amount

01:54:17.100 of nicotine in reality. So here, the belief effect isn't just changing what one subjectively

01:54:23.300 experiences. Oh, this is the effect of high nicotine or low nicotine. It actually is changing

01:54:28.400 the way that the brain responds to the belief. And that to me is absolutely wild. Now, there are a

01:54:35.580 couple of other things that could have confounded this. First of all, it could have been that if you

01:54:40.460 believe you got a lot of nicotine, you're just faster, or you're reading the lines better,

01:54:45.000 or your response time to hit the button is quicker. I tell you, you have a drug that's

01:54:48.860 going to improve reaction time. You might believe that about nicotine. And so you're quicker on the

01:54:52.580 trigger and you're getting, they have a destination. More activation. More activation. They rule that

01:54:58.260 out. They also rule out the possibility. How did they rule that out? By looking at rates of pressing.

01:55:04.060 And there was no difference? Nothing. And in sensory areas of the brain that would represent that kind

01:55:09.140 of difference, they don't see that. The other thing that is very clear is that the connection

01:55:14.480 between the thalamus and the ventral medial prefrontal cortex, that pathway scales in the

01:55:21.120 most beautiful way such that people that were told they had smoked a low or vaped a low amount of

01:55:26.840 nicotine got a subtle activation of that pathway. People that were told that they got a moderate

01:55:31.820 amount of nicotine got a more robust activation of that pathway. And the people that were told that

01:55:36.240 they got a high amount of nicotine in the vape pen saw a very robust activation of the thalamus to

01:55:41.480 this ventral prefrontal cortical pathway. Now, of course, this is all happening under the hood of

01:55:46.140 the skull simply on the basis of what they were told and what they believe. And technically the

01:55:52.180 fMRI is showing the activation of those two areas and that's how you can infer the strength of that

01:55:58.800 connection. That's right. There's a separate method called diffuser tensor imaging, which was

01:56:03.360 developed, I believe, out of the group in Minnesota. Minnesota has a very robust group in terms of

01:56:08.020 neuroimaging that can measure activation in fiber pathways. This is not that, but you can look at

01:56:14.240 the timing of activation and it's a known what we call monosynaptic pathway. So we haven't talked so

01:56:18.760 much about figures here, but I guess if we were going to look at any one figure and I can just describe

01:56:26.100 it for the audience that's not paying, doesn't have the figure in front of them. The, let's see,

01:56:31.600 the most, probably the most important figure is figure two. Remember I said I like to read the

01:56:39.460 titles of figures, which is that the belief about nicotine strength induced a dose dependent response

01:56:43.920 in the thalamus. Basically, if you and figure two B can tell you if they believe that they got more

01:56:50.680 nicotine, that's essentially the response that they saw. So if you look, or sorry, panel E,

01:56:57.720 if you look at the belief rating as a function of the estimate in the thalamus of what, how much

01:57:06.580 activation there was, it's a mess when you look at all the dots at once, but if you just separate it

01:57:10.860 out by high, medium, and low, you run the statistics, what you find is that there's a gradual increase,

01:57:16.160 but a legitimate one from low to medium to high. In other words, if I tell you this is a high dose

01:57:22.660 of nicotine, your brain will react as if it's a high dose of nicotine. Now, what they didn't do was

01:57:27.460 give people zero nicotine. Yeah, I was about to say there's a control that's missing here, right?

01:57:32.180 Yeah. So what they didn't do is give people zero nicotine and then tell them this is a high amount

01:57:37.380 of nicotine. It's sort of the equivalent of the cruel high school experiment. No alcohol, but then

01:57:41.820 the kid acts drunk. Now, in the high school example, it's unclear whether or not the kid actually

01:57:49.260 felt drunk or not. It's unclear whether or not they had been drunk previously, if they even knew

01:57:55.680 what it would be like to feel drunk, et cetera. And there's the social context. What I find just

01:58:00.360 outrageous and outrageously interesting about this study is simply that what we are told about the

01:58:07.620 dose of a drug changes the way that our physiology responds to the dose of the drug. And in my

01:58:13.840 understanding, this is the first study to ever look at dose dependence of belief effects, right?

01:58:20.340 And why would that be important? Well, for almost every study of drugs, you look at a dose dependent

01:58:25.620 curve. You look at zero, low dose, medium dose, high dose. And here they clearly are seeing a dose

01:58:33.340 dependent response simply to the understanding of what they expect the drug ought to do. In other words,

01:58:42.600 you can bypass pharmacology somewhat, right? Now look at figure 2B. Am I reading this correctly?

01:58:49.680 So it's got four bars on there. You've got the group who were told they got a low dose,

01:58:56.020 the group who was told they got a medium dose, the group that was told they had a high dose,

01:58:59.940 and then these healthy controls who presumably were non-smokers who were just put in the machine.

01:59:06.500 That's right. This is measuring parameter estimate. Is that referring to their ability

01:59:14.720 to play the trading game? The parameter estimate is the activation, reward-related activities from

01:59:23.780 independent thalamus mask, right? So what they're doing is they're just saying, if we just look at

01:59:27.480 the thalamus, what is the level of activation? I see. So this suggests that the only statistical

01:59:32.860 difference was between the low and the high. That's right.

01:59:38.520 And nobody else was statistically different. That's right.

01:59:40.840 But that's not the whole story? No, that's not the whole story. So when you look at the output

01:59:45.080 from the thalamus to the ventromedial prefrontal cortex, that's where you start to identify the-

01:59:51.920 Is that figure 4? That is, yes. So this is where you see, so figure 4B,

01:59:58.040 if you look at parameter estimates, so this is the degree of activation between

02:00:01.480 the thalamus and the ventromedial prefrontal cortex. And it's called the instructed belief.

02:00:06.520 You can see that there's a low, medium, and high scatter of dots for each, and that each one of

02:00:12.760 those is significant. So isn't it interesting that at the thalamus, which is, and you'll immediately

02:00:20.800 appreciate my stupidity when it comes to neuroscience, which is more proximate to the nicotinamide,

02:00:26.680 or nicotinamide, what do you call it, the nicotine acetylcholine receptor, you have a lower difference

02:00:34.500 of signal strength. And somehow that got amplified as it made its way forward in the brain?

02:00:39.960 Yeah.

02:00:40.320 Does that surprise you?

02:00:41.120 It is surprising. And it surprised them as well. The interpretation they give, again,

02:00:47.240 as we were talking about before, important to match their conclusions against what they actually

02:00:50.880 found, which is what we're doing here, the interpretation that they give is that it doesn't

02:00:55.620 take much nicotinic receptor occupancy in the thalamus to activate this pathway. But they too

02:01:01.700 were surprised that they could not detect a raw difference in the activation of the thalamus. But

02:01:05.920 in terms of its output to the prefrontal cortex, that's when the difference showed up.

02:01:10.740 That figure for B is more convincing than figure two, because even figure two E, if you read the

02:01:17.840 fine print, the R, the correlation coefficient is 0.27. It's not that strong.

02:01:24.580 It's weak.

02:01:25.200 So at the thalamus, it's kind of like, yeah, there might be a signal. By the way,

02:01:28.780 this goes back to our earlier discussion. There could be a huge signal here and we're underpowered.

02:01:32.940 How many subjects were in this? You wouldn't have a lot of subjects in this experiment.

02:01:36.300 Yeah. No. And this just speaks to the general challenge of doing this kind of work. It's hard

02:01:42.420 to get a lot of people in and through the scanner.

02:01:44.460 Yeah. And it's expensive.

02:01:45.280 And it's expensive. I should know this, but we can go back to the methods.

02:01:50.220 But you can sort of just look at the number of dots on here. I mean, it's in the low tens,

02:01:54.100 right? It's like 40, 30, something like that. So it's possible you do this-

02:01:58.160 It's not your Danish study.

02:01:59.200 Yeah. You do this with a thousand people. This could all be statistically significant.

02:02:03.500 Right. So they talk about this. Based on this, we estimate that an N of 20 N is sample size. In

02:02:09.360 each belief condition, the final sample would provide 90% power to detect an effect of this

02:02:13.520 magnitude at an alpha of 0.5 in a two-tailed test. Okay. So that's them referring to what we just

02:02:21.100 talked about, which is we believe at 90% confidence to get an alpha of 0.05, which means we'll want to

02:02:27.200 be 95% confidence. We need 60 people, 20 per group. Right. Yeah.

02:02:31.520 But if the difference is smaller than what they expected, they'll miss out on some of the

02:02:36.440 significance, which that looks like they're missing between the medium and high group.

02:02:39.800 Yep. And I too was surprised that they did not see a difference between the medium and the high

02:02:45.080 group, but they did in the output of the thalamus. I was also surprised that they didn't see a

02:02:49.900 difference. This is kind of interesting in its own right. In figure three talks about their belief

02:02:54.300 about nicotine strength did not modulate the reward response, the dopamine response.

02:02:58.920 How was that measured? Also just in FMRI? Yeah, exactly. So if you look at figure three B,

02:03:04.280 other people can't see it, but basically what you'll see is that there's no difference between

02:03:09.220 these different groups in terms of the amount of activation in these reward pathways, if people

02:03:14.620 got a low, medium, or high amount of nicotine. Now that actually could be leveraged, I believe,

02:03:20.660 if somebody were trying to quit nicotine, for instance, and they were going to do that by

02:03:25.480 progressively reducing the amount of nicotine that they were taking, but you told them that it was

02:03:30.340 the same amount from one day to the next, you could whittle it down to, presumably to a low

02:03:36.360 amount before taking it to zero. And if they believed it to be a greater amount, then it might

02:03:41.440 actually not disrupt their reward pathways, meaning they would feel, presumably they'd feel rewarded by

02:03:48.420 whatever nicotine they were bringing in.

02:03:49.880 What would be your prediction if this experiment were repeated, but it was done exactly the same

02:03:55.300 way with non-smokers?

02:03:58.400 Well, one thing that's sort of interesting, you asked about potential sources of artifact,

02:04:05.620 problems with FMRI. One of the challenges that they note in this study was you have to stay very

02:04:09.760 still in the machine, but the subjects were constantly coughing because they're smokers.

02:04:15.420 So, okay. So presumably the data would be higher fidelity. I started chuckling at that one, but I

02:04:20.640 was like, I had to read that one twice. I was like, oh, that makes sense. They're smokers,

02:04:24.140 they're coughing, they can't stay still. So movement artifact. But in all seriousness,

02:04:28.800 I think that for people that are naive to nicotine, even a small amount of nicotine is likely to get this

02:04:36.720 pathway activated to such a great degree, sort of like the first time effect of pretty much any drug.

02:04:41.980 But I wonder if they would be more or less susceptible to the belief system.

02:04:48.700 Yeah, that's a really good question. Right. Because they have no prior to compare it to.

02:04:51.820 They have no pleasant, they have no experience to compare it to with respect to

02:04:55.900 the obviously beneficial effects of nicotine that the smokers are well used to.

02:05:01.320 So this is the poor kid that got duped into thinking the non-alcoholic beer

02:05:05.380 was at alcohol, though they're actually the winner we know because they didn't have so on alcohol.

02:05:09.400 Alcohol is bad for you. So in the end, that kid wins and the other ones lose. Poetic justice.

02:05:14.080 But that kid, having never been actually drunk before, presumably would experience it more-

02:05:20.040 I would feel like they'd be more susceptible potentially.

02:05:22.880 That's my guess as well.

02:05:24.700 So my glee for this experiment is not, or this paper rather, is not because I think it's the be-all,

02:05:31.720 end-all or it's a perfect experiment.

02:05:33.260 I just think it's so very cool that they're starting to explore dose dependence of belief

02:05:38.720 because that has all sorts of implications. I mean, use your imagination, folks. Whether or not

02:05:45.460 we're talking about a drug, we're talking about a behavioral intervention, we're talking about

02:05:51.580 a vaccine, and I'm not referring to any one specific vaccine. I'm just talking to vaccines

02:05:56.720 generally. I'm talking about psychoactive drugs. I'm talking about illicit drugs. I'm talking about

02:06:04.220 antidepressants. I'm talking about all the sorts of drugs we were talking about before,

02:06:09.080 metformin, et cetera. Just throw our arms around all of it. What we believe about the effects of a drug,

02:06:17.320 presumably, in addition to what we believe about how much we're taking and what those effects ought to be,

02:06:22.760 clearly are impacting at least the way that our brain reacts to those drugs.

02:06:28.520 Yeah. It's very interesting. I mean, when you consider how many drugs that have peripheral effects

02:06:34.700 or peripheral outputs begin with central issues. So again, I think the GLP-1 agonists are such a

02:06:42.020 great example of this.

02:06:43.220 Ozempic.

02:06:43.540 Yeah. I don't think anybody fully understands exactly how they're working,

02:06:49.940 but it's hard to argue that they're impacting, that the GLP-1 analog is having a central impact.

02:06:57.500 It's doing something in the brain that is leading to a reduction of appetite.

02:07:02.060 We believe that.

02:07:02.840 Yeah.

02:07:03.120 Yeah. And I think the mouse data point to different areas of the hypothalamus that are

02:07:07.220 related to satiety, that it's at least possible.

02:07:11.240 Yeah. I mean, there's no quicker way to make a mouse overeat or undereat than by

02:07:18.220 lesioning its hypothalamus, depending on where you do so. So presumably these drugs work there.

02:07:23.200 But again, it speaks to what do you need to believe in order for that to be the case?

02:07:27.920 Have they done placebo trials there where people get something and they're told-

02:07:32.600 Oh, they do. I mean, of course, those drugs have all been tested via placebo and the placebo groups

02:07:37.420 don't do anywhere near as well. That's how we know that there's activity of the drug. But again,

02:07:41.760 that's a little bit different than being told you are absolutely getting it, right? Because in the

02:07:49.700 RCTs, you're just told you might be getting it, you might not be getting it. So it's not quite the

02:07:55.620 same as this experiment. This experiment is one level up where you're being told, no,

02:08:00.480 you're absolutely getting it. You're just getting different doses of it.

02:08:03.620 Yeah. To take this to maybe the ADHD realm, let's say a kid has been on ADHD meds for a while and

02:08:08.820 the parents, for whatever reason, the physician decide they want to cut back on the dosage.

02:08:13.780 But if they were to tell the kid it's the same dosage they've always been taking and it's had a

02:08:18.080 certain positive effect for them, according to the results, at least in this paper, which are not

02:08:24.100 definitive but are interesting, the lower dose may be as effective simply on the basis of belief.

02:08:30.120 And, and this is the part that makes it so cool to me is that, and it's not a kid tricking

02:08:35.920 themselves or the parents tricking the kid so much as the brain activation is corresponding to

02:08:42.520 the belief, right? So that's where this, this is why, because it's done in the brain, I think we can,

02:08:48.280 you know, it gets to these kind of abstract, nearly mystical, but not quite mystical aspects of

02:08:54.000 belief effects, which is that, you know, your brain is a prediction making machine. It's a

02:08:58.760 data interpretation machine, but it's clear that one of the more important pieces of data are your

02:09:04.480 beliefs about how these things impact you. So it's not that this bypasses physiology. People

02:09:11.500 aren't deluding themselves. The thalamus is behaving as if it's a high dose when it's the same dose as

02:09:15.940 the low dose group. Wild.

02:09:18.720 Yeah. I mean, I think of the implications, for example, of blood pressure, right? Like we don't

02:09:22.600 really understand essential hypertension, which is the majority of people walking around with high

02:09:26.860 blood pressure. It's unclear etiology. Um, so lots of people being treated, how do we know that the

02:09:33.460 belief system about it can't be changed? And, um, yeah, this is, this is, I don't know, this is

02:09:41.420 eyeopening. Yeah. It's cool stuff. And Allie Crumb is onto some other really cool stuff. Like for

02:09:47.440 instance, um, just to highlight where these belief effects are starting to show up. If you tell a group

02:09:53.780 that the side effects of a drug that they're taking are evidence that the drug really works

02:09:59.300 for the purpose that they're taking it, even though those side effects are kind of annoying,

02:10:03.480 people report the experiences less awful and they report more relief from the primary symptoms that

02:10:09.940 they're trying to target. So our belief about what side effects are can really impact how quickly and

02:10:16.340 how, um, compatible, uh, we feel about how quickly a drug works, excuse me, and how compatible we feel

02:10:23.080 that drug is with our entire life. So maybe if we call them something else, like not side effects,

02:10:27.340 but like additional benefits or something, it's kind of crazy. And you don't want to lie to people,

02:10:31.860 obviously, but you also don't want to send yourself in the opposite direction, which is reading the list

02:10:37.780 of side effects of a drug and then developing all of those side effects. Um, when, and then maybe later

02:10:45.360 coming to the understanding that some of those were raised through belief effects. Um, we definitely

02:10:49.820 see that that's the nocebo effect, right? That's, that's the one we see a lot, uh, you know, with,

02:10:55.280 with all sorts of drugs. Uh, and it's tough because, you know, how do you, how do you know which is which?

02:11:00.620 And, uh, I think there are some people who are really impacted by that and it makes it very difficult

02:11:04.980 for them to take any sort of pharmacologic agent because they basically, they can't help,

02:11:11.920 but incur every possible side effect. Um, is it, is it true that medical students often will

02:11:17.920 start developing the symptoms of the different diseases that they're learning about? Is that

02:11:21.460 true? Well, you know, I'll tell you, I do think that in medical school, you start to,

02:11:25.180 you start to think of the zebras more than the horses all the time, you know, like, you know what

02:11:32.500 I'm referring to, right? Uh, you know, you see footprints, you see hoof prints, you should think

02:11:36.300 of horses, but of course, medical students, you only think of the zebras. There are some really

02:11:40.060 funny things in medical school. Like there are certain conditions that you spend so much time

02:11:44.120 thinking about that you have a very warped sense of their prevalence. Uh, you know, like in medical

02:11:48.760 school, there's this condition called sarcoidosis. Like we, I feel like we never stopped talking about

02:11:53.460 sarcoidosis. I've seen like three cases in my life, right? Like it's just not that common. Um,

02:11:59.860 does it provide a great teaching tool or something? I don't know. Like I just, some of these things I

02:12:04.120 don't know. Uh, how much time did we spend talking about situs inverses? This is when people

02:12:09.680 embryologically have a reversed rotation and everything in their body is flipped. Literally

02:12:15.180 everything is flipped. So their heart is on the right side. Their liver is on the left side.

02:12:20.900 Their appendix is on the left side. Like, and so I'm not making this up. How common is this?

02:12:25.940 I've never seen it. Okay. I was thinking about boxing in the liver shot. Like you could easily be

02:12:31.280 going for the wrong side of the body. No, I swear to God, like as a medical student, if you were told

02:12:35.580 someone had left sided lower quadrant pain, to which the answer is almost assuredly that they

02:12:40.620 have diverticulitis, you'd think they could have appendicitis in the context of situs inverses.

02:12:46.220 Like the fact that that would even register in the top 10 things that it could possibly be,

02:12:51.860 but yes, you just have a totally warped sense of what's out there. Oh man. Well, um,

02:12:58.140 this has been pure pleasure for me. I don't know about you. I don't

02:13:01.260 know about our listeners, but for me, this is among the things that I just delight in and,

02:13:06.720 and even more so because you're the one across the table for me, teaching me about these incredible

02:13:11.720 findings and the gaps in those findings, which are equally incredible because they're equally

02:13:17.120 important to know about. Yeah. So let's do this again in Austin.

02:13:20.300 Absolutely. Next time on your home court. Very well. And bring a little bit of that do

02:13:24.400 if you've got it. Oh yeah. Yeah. Yeah. I'll bring a low, medium and high.

02:13:28.100 Low, medium and high. Thanks Peter. You're the best.

02:13:33.560 Thank you for listening to this week's episode of the drive. It's extremely important to me to

02:13:38.680 provide all of this content without relying on paid ads to do this. Our work is made entirely

02:13:43.660 possible by our members. And in return, we offer exclusive member only content and benefits above

02:13:50.080 and beyond what is available for free. So if you want to take your knowledge of this space to the

02:13:54.420 next level, it's our goal to ensure members get back much more than the price of the subscription.

02:13:59.800 Premium membership includes several benefits. First, comprehensive podcast show notes that detail

02:14:05.900 every topic, paper, person, and thing that we discuss in each episode. And the word on the street

02:14:11.500 is nobody's show notes rival ours. Second, monthly ask me anything or AMA episodes. These episodes are

02:14:19.600 comprised of detailed responses to subscriber questions typically focused on a single topic

02:14:24.480 and are designed to offer a great deal of clarity and detail on topics of special interest to our

02:14:29.840 members. You'll also get access to the show notes for these episodes, of course. Third, delivery of our

02:14:35.720 premium newsletter, which is put together by our dedicated team of research analysts. This newsletter

02:14:41.120 covers a wide range of topics related to longevity and provides much more detail than our free weekly

02:14:47.340 newsletter. Fourth, access to our private podcast feed that provides you with access to every episode,

02:14:54.220 including AMA's sans the spiel you're listening to now and in your regular podcast feed. Fifth,

02:15:01.160 the qualies, an additional member only podcast we put together that serves as a highlight reel featuring

02:15:07.340 the best excerpts from previous episodes of the drive. This is a great way to catch up on previous episodes

02:15:13.080 without having to go back and listen to each one of them. And finally, other benefits that are added

02:15:17.900 along the way. If you want to learn more and access these member only benefits, you can head over to

02:15:23.440 peteratiamd.com forward slash subscribe. You can also find me on YouTube, Instagram, and Twitter,

02:15:30.400 all with the handle peteratiamd. You can also leave us review on Apple podcasts or whatever podcast

02:15:36.960 player you use. This podcast is for general informational purposes only and does not

02:15:42.160 constitute the practice of medicine, nursing, or other professional healthcare services,

02:15:46.120 including the giving of medical advice. No doctor patient relationship is formed. The use of this

02:15:52.140 information and the materials linked to this podcast is at the user's own risk. The content on

02:15:58.120 this podcast is not intended to be a substitute for professional medical advice, diagnosis, or treatment.

02:16:03.160 Users should not disregard or delay in obtaining medical advice from any medical condition they

02:16:08.340 have, and they should seek the assistance of their healthcare professionals for any such conditions.

02:16:13.760 Finally, I take all conflicts of interest very seriously. For all of my disclosures and the companies

02:16:19.200 I invest in or advise, please visit peteratiamd.com forward slash about where I keep an up-to-date

02:16:26.820 and active list of all disclosures.

02:16:33.160 For all of my disclosures, please visit peteratiamd.com.

The Peter Attia Drive - September 11, 2023

#270 ‒ Journal club with Andrew Huberman： metformin as a geroprotective drug, the power of belief, and how to read scientific papers

Episode Stats

Length

Words per Minute

Word Count

Sentence Count

Hate Speech Sentences

Summary

Transcript