Journal Club with Dr. Peter Attia | Metformin for Longevity & The Power of Belief Effects

00:00:00.000 | - Welcome to the Huberman Lab Podcast,

00:00:02.280 | where we discuss science and science-based tools

00:00:04.880 | for everyday life.

00:00:05.900 | I'm Andrew Huberman,

00:00:10.040 | and I'm a professor of neurobiology and ophthalmology

00:00:13.160 | at Stanford School of Medicine.

00:00:15.120 | Today marks the first Journal Club episode

00:00:17.780 | between myself and Dr. Peter Attia.

00:00:20.440 | For any of you that are not familiar with Dr. Peter Attia,

00:00:22.840 | he is a medical doctor and MD

00:00:24.840 | who is an expert in all aspects of health and lifespan.

00:00:28.760 | He is the author of a bestselling book entitled "Outlive,"

00:00:31.680 | which is a phenomenal resource

00:00:33.300 | on all things healthspan and lifespan.

00:00:35.440 | And he is the host of the very popular podcast, "The Drive,"

00:00:38.960 | where he interviews various experts

00:00:41.160 | in all domains of medicine and scientists as well.

00:00:44.960 | Today, Peter and I hold

00:00:46.160 | our first online collaborative Journal Club.

00:00:48.920 | For those of you that aren't familiar

00:00:50.160 | with what a Journal Club is,

00:00:51.600 | a Journal Club is a common practice in graduate school

00:00:54.200 | and/or medical school,

00:00:55.520 | whereby students get together to discuss one or two papers

00:00:58.400 | to critique those papers

00:00:59.960 | and to really compare their own conclusions of those papers

00:01:03.120 | with the conclusions of the authors

00:01:04.740 | and to highlight any key takeaways.

00:01:06.820 | Peter and I have been wanting

00:01:07.880 | to do a Journal Club together for a very long time,

00:01:10.860 | and we decided to do that Journal Club

00:01:12.760 | and to record it for you.

00:01:14.640 | So today, you will be sitting in

00:01:16.100 | on the first Huberman-Attia Journal Club.

00:01:19.280 | By the way, it could just have easily been called

00:01:21.280 | the Attia-Huberman Journal Club.

00:01:23.280 | And we will discuss two papers.

00:01:25.120 | First, Peter is going to discuss a paper on metformin,

00:01:29.400 | which is a drug that many people are interested in

00:01:31.760 | for its potential role in longevity.

00:01:33.980 | I want to highlight potential there.

00:01:35.680 | He's going to compare that paper

00:01:37.720 | to previous findings on metformin.

00:01:39.680 | And by the end of that discussion,

00:01:41.360 | he will advise as to whether or not

00:01:43.080 | he himself would take metformin

00:01:45.180 | and whether or not other people might be well-advised

00:01:48.080 | or ill-advised to take metformin

00:01:50.080 | based on the data in that paper and at this time.

00:01:53.600 | Then I present a paper which is about the placebo effect.

00:01:56.860 | I have to imagine that most of you

00:01:58.040 | have heard of the placebo effect,

00:01:59.760 | but what's interesting about the paper

00:02:01.000 | that we discussed today is that it shows

00:02:03.320 | that the placebo effect can actually follow a dose response.

00:02:07.160 | So it's not just all or none.

00:02:08.960 | It actually is the case that you can scale

00:02:11.580 | the degree of placebo effect

00:02:13.360 | depending on whether or not you're thinking

00:02:14.780 | you're taking low doses, moderate doses,

00:02:17.420 | or high doses of a particular drug.

00:02:19.240 | And the particular drug that's discussed

00:02:21.200 | in the paper that I cover is nicotine.

00:02:23.960 | So for those of you that are interested

00:02:25.320 | in cognitive enhancement by way of pharmacology,

00:02:28.320 | or frankly, for people who are simply interested

00:02:30.660 | in how our beliefs can shape our physiology,

00:02:33.080 | I think you'll find that discussion to be very interesting.

00:02:35.960 | So by the end of today's episode,

00:02:37.400 | you will not only have learned

00:02:38.800 | about two novel sets of findings,

00:02:41.000 | one in the realm of longevity as it relates to metformin

00:02:44.000 | and another in the realm of neurobiology

00:02:46.360 | and placebos or placebo effects,

00:02:49.000 | but you will also learn how a journal club is conducted.

00:02:52.160 | I think you'll see in observing how we parse these papers

00:02:55.040 | and discuss them, even arguing in them at times,

00:02:57.540 | that what scientists and clinicians do

00:02:59.940 | is they take a look at the existing peer-reviewed research

00:03:03.680 | and they look at that peer-reviewed research

00:03:05.600 | with a fresh eye asking,

00:03:07.520 | does this paper really show what it claims to show or not?

00:03:11.080 | And in some cases, the answer is yes.

00:03:12.820 | And in other cases, the answer is no.

00:03:15.620 | What I know is for certain

00:03:16.640 | is that by the end of today's episode,

00:03:18.240 | you will learn a lot of science,

00:03:19.820 | you'll learn a lot about health practices,

00:03:21.440 | some of which you may want to apply or avoid,

00:03:24.000 | and you'll learn a lot

00:03:24.880 | about how science and medicine is carried out.

00:03:27.580 | Before we begin, I'd like to emphasize that this podcast

00:03:30.360 | is separate from my teaching and research roles at Stanford.

00:03:32.920 | It is, however, part of my desire and effort

00:03:34.840 | to bring zero cost to consumer information about science

00:03:37.400 | and science-related tools to the general public.

00:03:39.960 | In keeping with that theme,

00:03:41.000 | I'd like to thank the sponsors of today's podcast.

00:03:43.960 | Our first sponsor is Helix Sleep.

00:03:46.040 | Helix Sleep makes customized mattresses

00:03:48.460 | to give you the best possible night's sleep.

00:03:50.880 | Now, sleep is the foundation of mental health,

00:03:52.920 | physical health, and performance.

00:03:54.680 | When we are sleeping well and enough,

00:03:56.600 | mental health, physical health, and performance

00:03:58.620 | all stand to be at their best.

00:04:00.160 | One of the key things to getting a great night's sleep

00:04:02.020 | is to make sure that your mattress

00:04:03.640 | is tailored to your unique sleep needs.

00:04:05.840 | Helix Sleep has a brief two-minute quiz

00:04:07.920 | that if you go to their website,

00:04:09.460 | you take that quiz and answer questions

00:04:11.280 | such as, do you tend to sleep on your back,

00:04:12.920 | your side of your stomach?

00:04:13.800 | Do you tend to run hot or cold in the middle of the night?

00:04:16.280 | Maybe you don't know the answers to those questions

00:04:18.120 | and that's fine.

00:04:19.320 | At the end of that two-minute quiz,

00:04:20.800 | they will match you to a mattress

00:04:22.280 | that's ideal for your sleep needs.

00:04:24.240 | I sleep on the Dusk, a D-U-S-K, mattress.

00:04:26.800 | And when I started sleeping on a Dusk mattress

00:04:28.780 | about two years ago, my sleep immediately improved.

00:04:31.480 | So if you're interested in upgrading your mattress,

00:04:33.320 | go to helixsleep.com/huberman,

00:04:35.740 | take their two-minute sleep quiz,

00:04:37.000 | and they'll match you to a customized mattress for you.

00:04:39.280 | And you'll get up to $350 off any mattress order

00:04:42.340 | and two free pillows.

00:04:43.880 | Again, if interested, go to helixsleep.com/huberman

00:04:47.160 | for up to $350 off and two free pillows.

00:04:50.260 | Today's episode is also brought to us by Levels.

00:04:53.200 | Levels is a program that lets you see

00:04:54.800 | how different foods and behaviors affect your health

00:04:57.200 | by giving you real-time feedback

00:04:58.940 | using a continuous glucose monitor.

00:05:01.060 | One of the most important factors

00:05:02.400 | impacting your immediate and long-term health

00:05:04.860 | is the way that your body manages its blood glucose,

00:05:07.540 | or sometimes referred to as blood sugar levels.

00:05:10.200 | To maintain energy and focus throughout the day,

00:05:12.780 | you want to keep your blood glucose steady

00:05:14.540 | without big spikes or dips.

00:05:16.540 | Using Levels, you can monitor how different types of foods

00:05:19.320 | and different food combinations,

00:05:21.160 | as well as food timing and things like exercise,

00:05:23.760 | combine to impact your blood glucose levels.

00:05:26.360 | I started using Levels a little over a year ago,

00:05:28.800 | and it gave me a lot of insight

00:05:30.160 | into how specific foods were spiking my blood sugar

00:05:32.800 | and then leaving me feeling tired

00:05:34.380 | for several hours afterwards,

00:05:35.960 | as well as how the spacing of exercise and my meals

00:05:39.160 | was impacting my overall energy.

00:05:41.200 | And in doing so, it really allowed me to optimize

00:05:43.500 | how I eat, what I eat, when I exercise, and so on,

00:05:47.480 | such that my blood glucose levels and energy levels

00:05:50.180 | are stable throughout the day.

00:05:51.720 | If you're interested in learning more about Levels

00:05:53.560 | and trying a continuous glucose monitor yourself,

00:05:56.200 | go to levels.link/huberman.

00:05:59.020 | Right now, Levels is offering

00:06:00.320 | an additional two free months of membership.

00:06:02.220 | Again, that's levels.link, L-I-N-K/huberman,

00:06:06.100 | to get two free months of membership.

00:06:08.040 | And now for my Journal Club discussion

00:06:10.120 | with Dr. Peter Attia.

00:06:11.800 | Peter, so good to have you here.

00:06:14.160 | - So great to be here, my friend.

00:06:16.540 | - This is something that you and I

00:06:17.680 | have been wanting to do for a while.

00:06:20.760 | And it's basically something that we do all the time,

00:06:22.540 | which is to peruse the literature

00:06:25.160 | and find papers that we are excited about

00:06:27.920 | for whatever reason.

00:06:29.540 | And oftentimes that will lead to a text dialogue

00:06:32.600 | or a phone call or both.

00:06:34.620 | But this time we've opted to try talking about these papers

00:06:38.940 | that we find particularly exciting in real time

00:06:42.180 | for the first time as this podcast format.

00:06:46.220 | First of all, so that people can get some sense

00:06:48.240 | of why we're so excited about these papers.

00:06:49.820 | We do feel that people should know about these findings.

00:06:53.700 | And second of all, that it's an opportunity for people

00:06:56.440 | to learn how to dissect information

00:06:58.980 | and think about the papers they hear about in the news,

00:07:01.600 | the papers they might download from PubMed

00:07:03.460 | if they're inclined.

00:07:04.940 | Also just to start thinking like scientists and clinicians

00:07:08.140 | and get a better sense of what it looks like

00:07:10.440 | to pick through a paper, the good, the bad, and the ugly.

00:07:14.620 | So we're flying a little blind here, which is fun.

00:07:17.880 | I'm definitely excited for all the above reasons.

00:07:21.960 | - Yeah, no, this is...

00:07:25.060 | You and I have been talking about this for some time.

00:07:26.900 | And actually we used to run a journal club

00:07:30.020 | inside the practice where once a month,

00:07:32.340 | one person would just pick a paper

00:07:35.420 | and we would go through it in kind of

00:07:36.380 | a formal journal club presentation.

00:07:38.100 | We've gotten away from it for the last year

00:07:40.260 | just because we've been a little stretched then.

00:07:42.500 | I think it's something we need to resume

00:07:44.220 | because it's a great way to learn.

00:07:46.940 | And it's a skill.

00:07:47.980 | People probably ask you all the time,

00:07:50.140 | 'cause I know I get asked all the time,

00:07:51.780 | hey, what are the dos and don'ts

00:07:54.380 | of interpreting scientific papers?

00:07:56.560 | Is it enough to just read the abstract?

00:07:59.220 | And usually the answer is, well, no.

00:08:02.020 | But the how-to is tougher.

00:08:03.700 | And I think the two papers we've chosen today

00:08:06.660 | illustrate two opposite ends of the spectrum.

00:08:08.820 | You're gonna obviously talk about something

00:08:10.180 | that we're gonna probably get into

00:08:11.780 | the technical nature of the assays,

00:08:13.220 | the limitations, et cetera.

00:08:14.620 | And the paper ultimately I've chosen to present,

00:08:17.240 | although I apologize, I'm surprising you with this,

00:08:19.700 | up until a few minutes ago,

00:08:21.540 | is actually a very straightforward,

00:08:23.280 | simple epidemiologic paper

00:08:24.600 | that I think has important significance.

00:08:26.780 | I had originally gone down the rabbit hole

00:08:28.400 | on a much more nuanced paper about ATP binding cassettes

00:08:32.740 | in cholesterol absorption.

00:08:34.520 | But ultimately I thought this one

00:08:35.680 | might be more interesting to a broader audience.

00:08:38.060 | By the way, I gotta tell you a funny story.

00:08:39.280 | So I had a dream last night about you.

00:08:41.480 | And in this dream, you were obsessed

00:08:45.520 | with making this certain drink that was like your elixir.

00:08:49.440 | And it had all of these crazy ingredients in it.

00:08:53.080 | - Supplements.

00:08:54.160 | - Tons of supplements in it.

00:08:55.440 | But the one thing I remembered when I woke up,

00:08:57.400 | 'cause I forgot most of them,

00:08:58.400 | I was really trying so hard to remember them.

00:09:00.740 | One thing that you had in it was dew.

00:09:03.760 | Like you had to collect a certain amount of dew

00:09:06.160 | off the leaves every morning to put into this drink.

00:09:09.480 | It was so, but it was like just--

00:09:11.400 | - It sounds like something that I would do.

00:09:13.360 | - And so, but here's the best part.

00:09:15.840 | You had like a thermos of this stuff

00:09:18.360 | that had to be with you everywhere.

00:09:20.020 | And all of your clothing had to be tailored

00:09:23.080 | with a special pocket that you could put the thermos into

00:09:26.440 | so that you were never without the special Andrew drink.

00:09:30.440 | And again, you know how dreams, when you're having them,

00:09:32.800 | seem so logical and real,

00:09:35.840 | and then you wake up and you're like,

00:09:37.840 | that doesn't even make sense.

00:09:38.760 | Like, why would he want the thermos in his shirt?

00:09:41.240 | Like that would warm it up.

00:09:42.480 | Like, you know, all these,

00:09:43.560 | but boy, it was a realistic dream.

00:09:45.720 | And there were lots of things in it, including dew.

00:09:48.960 | Special dew off the leaves every morning.

00:09:51.640 | - I love it.

00:09:52.600 | Well, it's not that far from reality.

00:09:54.520 | I'm a big fan of yerba mate.

00:09:57.640 | I'm drinking it right now, in fact,

00:09:59.440 | in its many forms, usually the loose leaf.

00:10:03.820 | I don't tend to drink it out of the gourd.

00:10:05.360 | My dad's Argentine, so that's where I picked it up.

00:10:07.800 | I started drinking it when I was like five years old

00:10:09.600 | or younger, which I don't recommend people do.

00:10:11.240 | It's heavily caffeinated.

00:10:12.140 | Don't drink the smoked versions either, folks.

00:10:13.800 | I think that was potentially carcinogenic.

00:10:15.640 | But this thing that you describe

00:10:17.960 | of carrying around the thermos close to the body,

00:10:21.200 | if you are ever in Uruguay,

00:10:23.800 | or if you ever spot grown men in a restaurant

00:10:26.720 | anywhere in the world, carrying a thermos with them

00:10:29.120 | and to their meals and hugging it close,

00:10:32.320 | chances are they're Uruguay.

00:10:34.000 | And they're drinking yerba mate.

00:10:36.400 | They drink it usually after their meals.

00:10:38.200 | It's supposed to be good for your digestion.

00:10:39.960 | So it's not that far from reality.

00:10:42.360 | I don't carry the thermos, but I do drink mate every day.

00:10:45.040 | And I'm going to start collecting dew off the leaves.

00:10:48.960 | - Just a few drops every morning.

00:10:50.480 | [laughing]

00:10:52.740 | - Oh my.

00:10:54.300 | Some other time we can talk about dreams.

00:10:55.720 | Recently, I've been doing some dream exploration.

00:10:58.920 | I've had some absolutely transformative dreams

00:11:01.580 | for the first time in my life.

00:11:02.760 | One dream in particular that allowed me to feel something

00:11:06.400 | I've never felt before and has catalyzed a large number

00:11:10.780 | of important decisions in a way that no other experience,

00:11:14.120 | waking or sleep, has ever impacted me.

00:11:16.600 | And this was drug-free, et cetera.

00:11:19.680 | - And do you think you could have had that dream,

00:11:21.720 | we don't have to get into it if you don't want

00:11:22.900 | to talk about it now, but was there a lot of work

00:11:25.620 | you had to do to prepare for that dream to have taken place?

00:11:29.360 | - Oh yes, yeah.

00:11:31.000 | At least 18 months of intensive analysis type work

00:11:36.000 | with a very skilled psychiatrist.

00:11:39.820 | But I wasn't trying to seed the dream.

00:11:42.440 | It was just I was at a sticking point

00:11:44.720 | with a certain process in my life.

00:11:46.200 | And then I was taking a walk while waking

00:11:49.760 | and realized that my brain, my subconscious,

00:11:54.760 | was going to keep working on this.

00:11:56.680 | I just decided it's going to keep working on it.

00:11:58.480 | And then two nights later, I traveled to a meeting

00:12:00.860 | in Aspen and I had the most profound dream ever

00:12:04.240 | where I was able to sense something and feel something

00:12:06.500 | I've always wanted to feel as so real within the dream.

00:12:10.520 | Woke up, knew it was a dream and realized

00:12:14.000 | this is what people close to me that I respect

00:12:16.680 | have been talking about, but I was able to feel it

00:12:19.200 | and therefore I can actually access this in my waking life.

00:12:23.360 | It was absolutely transformative for me.

00:12:26.700 | Anyway, sometime I can share more details

00:12:29.840 | with you or the audience, but for now,

00:12:32.080 | maybe we should talk about these papers.

00:12:33.400 | - Very well.

00:12:34.240 | - Who should go first?

00:12:36.760 | - I'm happy to go first.

00:12:39.800 | - Great, yeah.

00:12:40.640 | - This is a pretty straightforward paper.

00:12:41.880 | So we're going to talk about a paper titled

00:12:44.600 | Reassessing the Evidence of a Survival Advantage

00:12:47.600 | in Type 2 Diabetics Treated with Metformin

00:12:50.740 | Compared with Controls Without Diabetes,

00:12:53.460 | a Retrospective Cohort Study.

00:12:55.260 | This is by Matthew Thomas Keyes and colleagues.

00:12:58.480 | This was published last fall.

00:13:00.620 | Why is this paper important?

00:13:04.200 | So this paper is important because in 2014,

00:13:08.720 | Bannister published a paper that I think in many ways

00:13:13.940 | kind of got the world very excited about metformin.

00:13:16.720 | So this was almost 10 years ago.

00:13:18.880 | And I'm sure many people have heard about this paper,

00:13:21.320 | even if they're not familiar with it,

00:13:22.440 | but they've heard the concept of the paper.

00:13:24.620 | And in many ways, it's the paper that has led

00:13:27.080 | to the excitement around the potential

00:13:29.820 | for giroprotection with metformin.

00:13:32.240 | And I should probably just define for the audience

00:13:34.280 | what giroprotection means.

00:13:35.780 | When we think of--

00:13:36.620 | - And probably also, sorry to interrupt,

00:13:37.540 | what metformin is, just for the uninformed.

00:13:39.680 | - That's a great point.

00:13:40.680 | So I'll start with the latter.

00:13:42.000 | So metformin is a drug that has been used

00:13:46.220 | for many years, depends where it was first approved,

00:13:50.600 | I think was in Europe.

00:13:52.440 | But call it directionally, 50 plus years of use

00:13:56.200 | as a first line agent for patients with type 2 diabetes.

00:14:01.200 | In the US, maybe 40 plus years.

00:14:03.560 | So this is a drug that's been around forever,

00:14:05.840 | trade name, Glucophage, or brand name.

00:14:09.800 | But again, it's a generic drug today.

00:14:13.000 | The mechanism by which metformin works is debated hotly.

00:14:18.000 | But what I think is not debated

00:14:20.900 | is the immediate thing that metformin does,

00:14:23.540 | which is it inhibits complex one of the mitochondria.

00:14:26.900 | So again, maybe just taking a step back.

00:14:29.320 | So the mitochondria, as everybody thinks of those

00:14:31.420 | as the cellular engine for making ATP.

00:14:33.940 | So the most efficient way that we make ATP

00:14:36.960 | is through oxidative phosphorylation,

00:14:39.260 | where we take either fatty acid pieces

00:14:43.140 | or a breakdown product of glucose

00:14:45.340 | once it's partially metabolized to pyruvate.

00:14:47.800 | We put that into an electron transport chain.

00:14:51.140 | And we basically trade chemical energy for electrons

00:14:56.140 | that can then be used to make phosphates onto ADP.

00:15:00.580 | So it's, you know, you think of everything you do.

00:15:02.580 | Eating is taking the chemical energy in food,

00:15:05.260 | taking the energy that's in those bonds,

00:15:07.000 | making electrical energy in the mitochondria.

00:15:09.580 | Those electrons pump a gradient that allow you to make ATP.

00:15:13.420 | To give a sense of how primal and important this is,

00:15:16.700 | if you block that process completely, you die.

00:15:20.120 | So everybody's probably heard of cyanide, right?

00:15:21.880 | Cyanide is something that is incredibly toxic,

00:15:24.320 | even at the smallest doses.

00:15:26.020 | Cyanide is a complete blocker of this process.

00:15:29.540 | And if my memory serves me correctly,

00:15:30.880 | I think it blocks complex four of the mitochondria.

00:15:33.320 | I don't know if you recall if it complex three

00:15:35.080 | or complex four.

00:15:36.240 | I know a lot about toxins that impact the nervous system,

00:15:39.080 | but I don't know a lot about poisons.

00:15:40.220 | - The mitochondria, yeah.

00:15:41.060 | - But if ever you want to have some fun,

00:15:42.280 | we can talk about all the dangerous stuff

00:15:44.480 | that animals make and insects make and how they kill you.

00:15:47.600 | - Yeah, like detrototoxin and all these things

00:15:49.960 | that block sodium channels, yeah, yeah, yeah.

00:15:50.800 | - Alpha-lactotoxin, bungroda.

00:15:52.000 | I really geek out on this stuff

00:15:53.680 | 'cause it allows me to talk about neuroscience,

00:15:56.120 | animals, and scary stuff.

00:15:58.380 | It's like combines it.

00:15:59.220 | So we could do that sometime for fun,

00:16:00.420 | maybe at the end if we have a few moments.

00:16:02.080 | - So, you know, something like cyanide

00:16:04.360 | that is a very potent inhibitor of this

00:16:06.800 | electron transport chain will kill you instantly.

00:16:08.720 | People understand that, of course, a drop of cyanide

00:16:10.760 | and you would be dead literally instantaneously.

00:16:14.000 | So Metformin works at the first of those complexes.

00:16:17.180 | I believe there are four, if my memory serves correctly,

00:16:19.400 | four electron transport chain complexes.

00:16:22.400 | But of course it's not a complete inhibition of it.

00:16:25.700 | It's just kind of a weak blocker of that.

00:16:28.160 | And the net effect of that is what?

00:16:30.500 | So the net effect of that is that it changes the ratio

00:16:32.960 | of adenosine monophosphate to adenosine diphosphate.

00:16:36.360 | What's less clear is why does that

00:16:40.480 | have a benefit in diabetics?

00:16:42.720 | Because what it unambiguously does is reduces

00:16:45.640 | the amount of glucose that the liver puts out.

00:16:48.060 | So hepatic glucose output is one of the fundamental problems

00:16:51.960 | that's happening in type two diabetes.

00:16:54.600 | You may recall, I think we talked about this

00:16:56.280 | even on a previous podcast,

00:16:58.000 | you and I sitting here with normal blood sugar

00:17:00.400 | have about five grams of glucose in our total circulation.

00:17:03.960 | That's it, five grams.

00:17:05.080 | Think about how quickly the brain will go through that

00:17:07.840 | within minutes.

00:17:09.200 | So the only thing that keeps us alive

00:17:12.240 | is our liver's ability to titrate out glucose.

00:17:16.100 | And if it puts out too much, for example,

00:17:18.120 | if the glucose level was consistently two teaspoons,

00:17:22.400 | you would have type two diabetes.

00:17:24.040 | So the difference between being metabolically healthy

00:17:26.880 | and having profound type two diabetes is one teaspoon

00:17:30.920 | of glucose in your bloodstream.

00:17:32.100 | So the ability of the liver to tamp down

00:17:34.560 | on high glucose output is important.

00:17:36.600 | Metformin seems to do that.

00:17:38.720 | - Can I just ask one question?

00:17:40.880 | Is it fair to provide this overly simplified summary

00:17:45.840 | of the biochemistry, which is that when we eat,

00:17:49.140 | the food is broken down, but the breaking of bonds

00:17:52.920 | creates energy that then our cells can use

00:17:55.040 | in the form of ATP.

00:17:56.200 | And the mitochondria are central to that process.

00:17:58.440 | And that metformin is partially short-circuiting

00:18:01.560 | the energy production process.

00:18:03.920 | And so even though we are eating,

00:18:06.780 | when we have metformin in our system, presumably,

00:18:09.640 | there is going to be less net glucose.

00:18:13.020 | The bonds are going to be broken down.

00:18:14.560 | We're chewing, we're digesting,

00:18:15.920 | but less of that is turned into blood sugar, glucose.

00:18:19.060 | - Well, sort of.

00:18:20.120 | I mean, it's not depriving you of ultimately

00:18:25.200 | storing that energy.

00:18:26.800 | What it's doing is changing the way

00:18:29.480 | the body partitions fuel.

00:18:32.480 | That's probably a better way to think about it

00:18:33.760 | to be a little bit more accurate.

00:18:35.000 | So for example, like it's not depriving you

00:18:39.240 | of the calories that are in that glucose.

00:18:41.540 | That would be fantastic, but that's not--

00:18:44.040 | - That was the elastro.

00:18:46.120 | Remember the elastro from the '90s?

00:18:47.560 | Elastro, folks, for those of you who don't remember,

00:18:50.560 | by the way, if you ever ate this stuff, you'd remember,

00:18:53.320 | because it was a fat that was not easily digested.

00:18:56.960 | It had sort of analogous to plant fiber

00:19:00.620 | or something like that.

00:19:01.460 | So it was being put into potato chips and whatnot.

00:19:03.520 | And the idea was that people would simply excrete it.

00:19:08.520 | And I don't know what happened,

00:19:12.120 | except that people got a lot of stomach aches.

00:19:14.400 | - Well, the anal seepage. - Everyone got fatter

00:19:16.000 | in the world.

00:19:16.820 | We know that.

00:19:17.660 | - The anal seepage is what really did that product in.

00:19:19.880 | - Anal seepage.

00:19:20.720 | - Only a physician, because after all,

00:19:23.640 | Peter's a clinician, a physician, an MD, and I'm not,

00:19:27.200 | could find it an appropriate term to describe

00:19:32.200 | what this is. - Yeah.

00:19:33.540 | When you have that much fat malabsorption,

00:19:36.440 | you start to have accidents.

00:19:38.100 | - Wow.

00:19:39.000 | And so--

00:19:39.840 | - So that did away with that product.

00:19:41.040 | - Right.

00:19:41.880 | It was either that or the diaper industry

00:19:42.800 | was going to really take off.

00:19:44.280 | Okay.

00:19:45.120 | That's why you don't hear about elastro.

00:19:46.440 | - That's right.

00:19:47.280 | So we've got this drug.

00:19:48.200 | We've got this drug metformin.

00:19:49.480 | It's considered a perfect first line agent

00:19:52.480 | for people with type two diabetes.

00:19:53.780 | So again, what's happening when you have type two diabetes,

00:19:56.600 | the primary insult probably occurs in the muscles

00:20:00.200 | and it is insulin resistance.

00:20:03.000 | Everybody hears that term.

00:20:04.040 | What does it mean?

00:20:05.360 | Insulin is a peptide.

00:20:07.000 | It binds to a receptor on a cell.

00:20:08.840 | So let's just talk about it through the lens of the muscle

00:20:10.680 | 'cause the muscle is responsible for most glucose disposal.

00:20:13.580 | It gets glucose out of the circulation.

00:20:15.560 | High glucose is toxic.

00:20:16.900 | We have to put it away.

00:20:18.320 | And we want to put most of it into our muscles.

00:20:20.720 | That's where we store 75 to 80% of it.

00:20:23.760 | When insulin binds to the insulin receptor,

00:20:26.960 | a tyrosine kinase is triggered inside.

00:20:29.800 | So just ignore all that.

00:20:31.020 | But a chemical reaction takes place inside the cell

00:20:33.880 | that leads to a phosphorylation.

00:20:35.920 | So ATP donates a phosphate group.

00:20:38.520 | And a transporter, just think of like a little tunnel,

00:20:42.360 | like a little straw goes up through the level of the cell.

00:20:46.160 | And now glucose can freely flow in.

00:20:48.560 | So I'm sure you've talked a lot about this

00:20:50.520 | with your audience.

00:20:51.840 | Things that move against gradients need pumps to move them.

00:20:55.240 | Things that move with gradients don't.

00:20:57.220 | Glucose is moving with its gradient into the cell.

00:21:00.080 | It doesn't need active transport,

00:21:01.400 | but it does need the transporter put there.

00:21:03.820 | That requires the energy.

00:21:05.760 | That's the job of insulin.

00:21:07.900 | - By the way, I did not know that.

00:21:09.160 | I mean, I certainly know active and passive transport

00:21:12.020 | as it relates to like neurotransmitter and ion flow.

00:21:15.320 | But I'd never heard that when insulin binds to a cell

00:21:18.200 | that literally a little straw

00:21:19.480 | is placed into the membrane of the cell.

00:21:21.140 | - Yeah, the glucose doesn't need a pump to move it in.

00:21:24.400 | Because there's much more glucose

00:21:25.560 | outside the cell than inside.

00:21:26.680 | So it's just, but the energy required

00:21:28.760 | is to move the straw up to the cell.

00:21:31.560 | - Cell biology is so cool.

00:21:32.940 | - Yeah, it is.

00:21:33.780 | So what happens is as, and Gerald Shulman at Yale

00:21:38.780 | did the best work on elucidating this,

00:21:41.600 | as the intramuscular fat increases,

00:21:45.340 | and by intramuscular, I mean intracellular fat,

00:21:49.040 | triacyl and diacylglycerides accumulate in a muscle cell,

00:21:53.000 | that signal gets interrupted.

00:21:55.360 | And all of a sudden, I'm making these numbers up,

00:21:57.660 | if you used to need two units of insulin

00:22:01.160 | to trigger the little transporter now, you need three,

00:22:04.480 | and then you need four, and then you need five.

00:22:07.020 | You need more and more insulin to get the thing up.

00:22:10.300 | That is the definition of insulin resistance.

00:22:13.840 | The cell is becoming resistant to the effect of insulin.

00:22:17.000 | And therefore, the early mark of insulin resistance,

00:22:20.560 | the canary in the coal mine is not an increase in glucose.

00:22:24.500 | It's an increase in insulin.

00:22:26.360 | So normal glycemia with hyperinsulinemia,

00:22:30.320 | especially postprandial,

00:22:31.740 | meaning after you eat hyperinsulinemia,

00:22:34.200 | is the thing that tells you,

00:22:35.600 | hey, you're five, 10 years away

00:22:37.180 | from this being a real problem.

00:22:39.160 | So fast forward many steps down the line,

00:22:41.300 | someone with type 2 diabetes has long passed that system.

00:22:44.800 | Now, not only are they insulin resistant,

00:22:47.000 | where they just need a boatload of insulin,

00:22:49.380 | which is made by the pancreas,

00:22:50.840 | to get glucose out of the circulation,

00:22:53.240 | but now that system's not even working well.

00:22:54.920 | And now they're not getting glucose into the cell.

00:22:57.420 | So now their glucose level is elevated.

00:22:59.920 | And even though it's continually being chewed up

00:23:03.060 | and used up, because again, the brain alone

00:23:05.180 | would account for most of that glucose disposal,

00:23:09.160 | the liver is now becoming insulin resistant as well.

00:23:11.660 | And now the liver isn't able to regulate

00:23:14.400 | how much glucose to put into circulation,

00:23:16.460 | and it's overdoing it.

00:23:17.580 | So now you have too much glucose

00:23:19.300 | being pumped into the circulation by the liver,

00:23:21.300 | and you have the muscles that can't dispose of it.

00:23:23.620 | And it's really a vicious, brutal cascade,

00:23:25.980 | because the same problem of fat accumulating in the muscle

00:23:29.180 | is now starting to happen in the pancreas.

00:23:31.260 | And now the relatively few cells in the pancreas

00:23:34.660 | called beta cells that make insulin

00:23:36.700 | are undergoing inflammation due to the fat accumulation

00:23:40.000 | within the pancreas itself.

00:23:41.620 | And so now the thing that you need to make more insulin

00:23:45.060 | is less effective at making insulin.

00:23:47.060 | So ultimately way, way, way down the line,

00:23:49.300 | a person with type 2 diabetes

00:23:50.900 | might actually even require insulin exogenously.

00:23:53.740 | - Could you share with us a few of the causes

00:23:56.020 | of type 2 diabetes of insulin resistance?

00:23:58.980 | I mean, one it sounds like is accumulating too much fat.

00:24:01.820 | - Yeah, so energy imbalance would be an enormous one.

00:24:05.140 | Inactivity or insufficient activity

00:24:07.660 | is probably the single most important.

00:24:09.500 | So when Gerald Schulman was running clinical trials at Yale,

00:24:14.500 | they would be recruiting undergrads to study, obviously,

00:24:17.500 | 'cause you're typically recruiting young people.

00:24:19.380 | And they would be doing these very detailed

00:24:21.580 | mechanistic studies where they would require

00:24:23.140 | actual tissue biopsies.

00:24:24.300 | So you're gonna biopsy somebody's quadriceps

00:24:26.580 | and actually look at what's happening in the muscle.

00:24:28.860 | Well, I remember him telling me this

00:24:30.540 | when I interviewed him on my podcast.

00:24:31.940 | He said, "The most important criteria

00:24:34.580 | of the people we interviewed is that,"

00:24:35.740 | 'cause they were still lean.

00:24:36.620 | These weren't people that were overweight,

00:24:37.860 | but they had to be inactive.

00:24:39.460 | You couldn't have active people in these studies.

00:24:42.180 | So exercising is one of the most important things

00:24:45.100 | you're going to do to ward off insulin resistance.

00:24:48.360 | But there are other things

00:24:49.200 | that can cause insulin resistance.

00:24:50.820 | Sleep deprivation has a profound impact

00:24:52.860 | on insulin resistance.

00:24:53.820 | I think we probably talked about this previously,

00:24:55.440 | but if you, you know, some very elegant mechanistic studies

00:24:58.180 | where you sleep deprive people, you know,

00:24:59.860 | you let them only sleep for four hours for a week,

00:25:02.500 | you'll reduce their glucose disposal by about half.

00:25:05.660 | - Wow.

00:25:06.500 | - Which is, I mean, that's a staggering amount of,

00:25:08.720 | you're basically inducing profound insulin resistance

00:25:11.260 | in just a week of sleep deprivation.

00:25:13.000 | Hypercortisolemia is another factor,

00:25:15.320 | and then obviously energy imbalance.

00:25:16.680 | So where, when you're accumulating excess energy,

00:25:19.760 | when you're getting fatter,

00:25:20.840 | if you start spilling that fat

00:25:23.040 | outside of the subcutaneous fat cells

00:25:25.040 | into the muscle, into the liver, into the pancreas,

00:25:27.140 | all of those things are exacerbating it.

00:25:29.160 | - Got it.

00:25:30.280 | Okay, so enter metformin, first line drug.

00:25:33.960 | So most of the drugs,

00:25:35.380 | so every drug you give a person with type 2 diabetes

00:25:37.520 | is trying to address part of this chain.

00:25:40.200 | So some of the drugs tell you to make more insulin.

00:25:44.000 | That's one of the strategies.

00:25:45.340 | So here are drugs like sulfonylureas.

00:25:47.940 | They tell the body, make more insulin.

00:25:51.000 | Other drugs like insulin

00:25:52.880 | just give you more of the insulin thing.

00:25:55.120 | Metformin tackles the problem elsewhere.

00:25:57.420 | It tamps down glucose by addressing the glucose,

00:26:00.980 | the hepatic glucose output channel.

00:26:03.300 | GLP-1 agonists are another drug.

00:26:05.460 | They increase insulin sensitivity,

00:26:07.260 | initially causing you to also make more insulin.

00:26:09.980 | GLP-1 agonists- - So that's osempic.

00:26:11.380 | - Yes. - Yeah.

00:26:12.220 | - Yeah. - And is it true

00:26:13.540 | that berberine is more or less the poor man's metformin?

00:26:17.300 | - Yeah. - Okay.

00:26:18.140 | - Yeah. - It's from a tree bark.

00:26:19.500 | It just happens to have the same properties of-

00:26:21.460 | - Yeah, and by the way, metformin-

00:26:22.300 | - Reducing mTOR and reducing blood glucose.

00:26:24.780 | - Yeah, and metformin, by the way,

00:26:26.180 | occurs from a lilac plant in France.

00:26:28.140 | Like that's where it was discovered.

00:26:29.220 | So it's also, metformin is also based

00:26:30.900 | on a substance found in nature.

00:26:32.660 | - So you need a prescription for metformin.

00:26:35.180 | You don't need a prescription for berberine.

00:26:36.940 | - Correct. - But yeah,

00:26:37.780 | we can talk about berberine a little bit later.

00:26:39.180 | I had a couple great experiences with berberine

00:26:41.580 | and a couple bad experiences.

00:26:43.100 | - Interesting. - Berberine, yeah.

00:26:44.980 | - So maybe taking one step back from this.

00:26:48.340 | In 2011, I became very interested in metformin personally,

00:26:53.420 | just reading about it, obsessing over it,

00:26:56.460 | and just somehow decided like, I should be taking this.

00:26:59.180 | So I actually began taking metformin.

00:27:01.020 | I still remember exactly when I started.

00:27:02.620 | I started it in May of 2011, and I realized that

00:27:05.600 | because I was on a trip with a bunch of buddies.

00:27:08.260 | We went to the Berkshire Hathaway shareholder meeting,

00:27:11.860 | which is, you know, the Buffett shareholder meeting.

00:27:15.340 | And, you know, it was kind of like a fun thing to do.

00:27:17.740 | And I remember being so sick the whole time

00:27:20.480 | because I didn't titrate up the dose of metformin.

00:27:23.300 | I just went straight to two grams a day,

00:27:25.340 | which is kind of like the full dose.

00:27:27.240 | And we went to this-

00:27:28.740 | - Is that characteristic of your approach to things?

00:27:31.600 | - Yes, I think that's safe to say.

00:27:33.520 | - Next time, I'll give you a thermos of this dew

00:27:35.260 | that I collect in the morning.

00:27:36.260 | - Oh, really? (laughing)

00:27:39.740 | So I remember being so sick that the whole time

00:27:42.340 | we were in Nebraska or Omaha, I guess, I couldn't,

00:27:45.940 | we went to Dairy Queen 'cause you do all the Buffett things

00:27:48.260 | when you're there, right?

00:27:49.220 | Like I couldn't have an ice cream at Dairy Queen.

00:27:51.100 | - You couldn't? - I mean, I couldn't.

00:27:51.940 | I was so nauseous.

00:27:53.020 | - Oh, 'cause I would say if you've got metformin

00:27:54.620 | in your system, you're gonna buffer glucose.

00:27:56.020 | You could have four ice cream cones and probably-

00:27:57.180 | - Except I couldn't keep anything down.

00:27:59.140 | I mean, I was so nauseous.

00:28:00.860 | So clearly metformin has this side effect initially,

00:28:03.700 | which is a little bit of appetite suppression.

00:28:05.700 | But regardless, that's the story on metformin.

00:28:08.100 | There were a lot of reasons I was interested in it.

00:28:10.740 | I wasn't thinking true zero protection.

00:28:13.280 | That term wasn't in my vernacular at the time.

00:28:16.200 | But what I was thinking is, hey,

00:28:17.380 | this is gonna help you buffer glucose better.

00:28:18.860 | It's gotta be better.

00:28:19.700 | And this was sort of my first foray into, you know,

00:28:22.360 | self-experimentation.

00:28:23.460 | - Do you wanna define zero protection?

00:28:25.040 | - Yeah, yeah.

00:28:25.880 | It's a good term to define.

00:28:27.000 | - Geriatric zero?

00:28:28.460 | - Yeah, so yeah, zero from geriatric old protection.

00:28:32.660 | So protection from aging.

00:28:34.460 | And when we talk about a drug like metformin or rapamycin

00:28:39.460 | or even NAD, NR, these things,

00:28:42.700 | the idea is we're talking about them as zero protective

00:28:45.720 | to signal that they are drugs

00:28:47.340 | that are not targeting a specific disease of aging.

00:28:50.540 | For example, a PCSK9 inhibitor is sort of zero protective,

00:28:55.060 | but it's targeting one specific pathway,

00:28:57.740 | which is cardiovascular disease and dyslipidemia.

00:29:00.940 | Whereas the idea is a zero protective agent

00:29:04.180 | would target hallmarks of aging.

00:29:05.960 | There are nine hallmarks of aging.

00:29:07.220 | Please don't ask me to recite them.

00:29:08.760 | I've never been able to get all nine straight,

00:29:11.300 | but people know what we're talking about, right?

00:29:12.920 | So decreased autophagy, increased senescence,

00:29:15.900 | decreased nutrient sensing or defective nutrient sensing,

00:29:19.400 | proteomic instability, genomic instability, methylation,

00:29:22.740 | all of these things, epigenetic changes.

00:29:24.180 | Those are all the nine hallmarks of aging.

00:29:25.700 | - You got seven.

00:29:26.860 | - So a zero protective agent would target

00:29:29.780 | those deep down biological hallmarks of aging.

00:29:33.860 | And in 2014, a paper came out by Bannister

00:29:38.560 | that basically got the world focused on this problem.

00:29:41.240 | By the world, I mean the world of anti-aging.

00:29:43.620 | So what Bannister and colleagues did

00:29:47.020 | was they took a registry from the UK

00:29:50.660 | and they got a set of patients who were on metformin

00:29:55.660 | with type two diabetes, but only metformin.

00:29:59.100 | So these were people who had just progressed to diabetes.

00:30:02.440 | They were not put on any other drug, just metformin.

00:30:06.000 | And then they found from the same registry,

00:30:09.040 | a group of matched controls.

00:30:10.660 | So this is a standard way

00:30:13.200 | that epidemiologic studies are done.

00:30:16.100 | Because again, you don't have the luxury

00:30:17.500 | of doing the randomization.

00:30:18.920 | So you're trying to account for all the biases

00:30:21.740 | that could exist by saying, we're gonna take people

00:30:24.720 | who look just like that person with diabetes.

00:30:28.080 | So can we match them for age, sex, socioeconomic status,

00:30:33.080 | blood pressure, BMI, everything we can.

00:30:37.420 | And then let's look at what happened to them over time.

00:30:40.900 | Now again, this is all happening in the future.

00:30:42.920 | So you're looking into the past.

00:30:43.980 | It's retrospective in that sense.

00:30:46.380 | And so let me just kind of pull up the sort of table here

00:30:50.460 | so I can kind of walk through.

00:30:51.540 | And this is not in the paper we talked about,

00:30:52.900 | but I think this is an important background.

00:30:54.740 | So they did something that at the time

00:30:59.740 | I didn't really notice.

00:31:01.460 | I didn't notice what they did.

00:31:03.660 | I probably did and I forgot.

00:31:05.340 | But I didn't notice this until about five years ago

00:31:07.780 | when I went back and looked at the paper.

00:31:10.020 | And they did something called informative censoring.

00:31:14.980 | So the way the study worked is if you were put on metformin,

00:31:18.420 | we're gonna follow you.

00:31:19.440 | If you're not on metformin, we're gonna follow you.

00:31:21.180 | And we're gonna track the number of deaths

00:31:23.400 | from any cause that occur.

00:31:24.820 | This is called all-cause mortality or ACM.

00:31:27.260 | And it's really the gold standard in a trial of this nature

00:31:30.940 | or a study of this nature or even a clinical trial.

00:31:32.840 | You wanna know how much are people dying from anything

00:31:35.700 | 'cause we're trying to prevent or delay death of all causes.

00:31:39.500 | Informative censoring says if a person

00:31:44.500 | who's on metformin deviates from that inclusion criteria,

00:31:49.140 | we will not count them in the final assessment.

00:31:52.100 | So how are the ways that that can happen?

00:31:54.280 | Well, one, the person can be lost to follow up.

00:31:57.720 | Two, they can just stop taking their metformin.

00:32:01.180 | Three, and more commonly, they can progress

00:32:04.300 | to needing a more significant drug.

00:32:08.480 | So all of those patients were excluded from the study.

00:32:11.660 | So think about that for a moment.

00:32:13.840 | This is, in my opinion,

00:32:15.420 | a significant limitation of this study.

00:32:18.240 | Because what you're basically doing is saying

00:32:21.580 | we're only gonna consider the patients

00:32:23.900 | who were on metformin, stayed on metformin,

00:32:26.320 | and never progressed through it.

00:32:27.880 | And we're gonna compare those to people

00:32:29.980 | who were not having type 2 diabetes.

00:32:31.740 | So an analogy here would be imagine we're gonna do a study

00:32:35.300 | of two groups that we think are almost identical.

00:32:38.240 | One of them are smokers,

00:32:39.740 | and the other are identical in every way,

00:32:41.260 | but they're not smokers.

00:32:42.140 | And we're gonna follow them

00:32:42.980 | to see which ones get lung cancer.

00:32:44.740 | But every time somebody dies in the smoking group,

00:32:48.260 | we stop counting them.

00:32:49.980 | When you get to the end,

00:32:51.100 | you're going to have a less significant view

00:32:54.060 | of the health status of that group.

00:32:56.660 | So with that caveat,

00:32:59.060 | the Bannister study found a very interesting result,

00:33:02.780 | which was the crude death rate

00:33:08.060 | was, and by the way, the way these are done,

00:33:10.740 | this is also one of the challenges of epidemiology

00:33:13.020 | is the math gets much more complicated.

00:33:15.580 | You have to normalize death rate

00:33:17.380 | for the amount of time you study the people.

00:33:19.920 | So everything is normalized to 1,000 person years.

00:33:24.000 | So the crude death rate in the group of people

00:33:27.540 | with type 2 diabetes who were on metformin,

00:33:30.980 | including the censoring, was 14.4.

00:33:33.760 | So 14.4 deaths occurred per 1,000 patient years.

00:33:38.760 | If you look at the control group, it was 15.2.

00:33:42.640 | This was a startling result.

00:33:46.060 | And I remember reading this in, again, 2014

00:33:48.800 | and being like, holy crap, this is really amazing.

00:33:52.080 | - Is there, could you explain why?

00:33:54.320 | 'Cause I hear those numbers

00:33:56.880 | and they don't seem that striking.

00:33:58.200 | It's a difference of about a year and a half.

00:33:59.620 | Now, of course, a difference of about a year and a half

00:34:03.020 | and lifespan is remarkable.

00:34:06.060 | - It doesn't even translate to that.

00:34:07.180 | So taking a step back, type 2 diabetes on average

00:34:10.320 | will shorten your life by six years.

00:34:12.720 | So that's the actuarial difference

00:34:14.240 | between having type 2 diabetes and not all comers.

00:34:17.260 | But you're right, this is not a huge difference.

00:34:18.900 | It's only a difference of a little less

00:34:21.000 | than one year of life per 1,000 patient years studied.

00:34:24.540 | - Okay, and by the way, up here, just point out,

00:34:26.620 | my math was wrong when I said about a year and a half.

00:34:28.980 | - But the point here is you would expect

00:34:31.720 | the people in the metformin group

00:34:33.080 | to have a far worse outcome,

00:34:35.800 | i.e. to have a far worse crude death rate.

00:34:39.720 | And the fact that it was statistically significant

00:34:42.400 | in the other direction,

00:34:43.680 | and it turned out on what's called

00:34:45.320 | the Cox proportional hazard,

00:34:46.920 | which is where you actually model

00:34:49.320 | the difference in lifespan.

00:34:51.320 | The people who took metformin and had diabetes

00:34:56.040 | had a 15%, one five, 15% relative reduction

00:35:00.640 | in all-cause death over 2.8 years,

00:35:03.560 | which was the median duration of follow-up

00:35:05.600 | compared to the non-group. - Well, that seems to be

00:35:06.440 | the number that makes me go, wow.

00:35:08.520 | - Yeah.

00:35:09.360 | - Because, could you repeat those numbers again?

00:35:13.640 | - Yeah, so 15% reduction in all-cause mortality

00:35:17.360 | over 2.8 years.

00:35:18.640 | - That's a big deal.

00:35:21.000 | - It is, and again, there's no clear explanation for it

00:35:27.720 | unless you believe that metformin is doing something

00:35:31.720 | beyond helping you lower blood glucose,

00:35:35.040 | because the difference in blood glucose

00:35:37.640 | between these two people was still in favor

00:35:40.200 | of the non-diabetics.

00:35:42.480 | So again, the proponents of metformin

00:35:45.600 | being a giroprotective agent,

00:35:47.200 | and I put myself in this category at one point,

00:35:49.560 | I would put myself today in the category of undecided,

00:35:52.360 | but at the time, I very much believed

00:35:54.900 | this was a very good suggestion

00:35:57.640 | that metformin was doing other things.

00:35:59.740 | You mentioned a couple already.

00:36:01.400 | Metformin is a weak inhibitor of mTOR.

00:36:03.840 | Metformin reduces inflammation.

00:36:06.080 | Metformin potentially tamps down on senescent cells

00:36:09.420 | and their secretory products.

00:36:11.320 | There are lots of things metformin could be doing

00:36:13.840 | that are off target,

00:36:14.940 | and it might be that those things

00:36:17.320 | are conferring the advantage.

00:36:19.440 | - As many of you know, I've been taking AG1 daily since 2012,

00:36:23.040 | so I'm delighted that they're sponsoring the podcast.

00:36:25.480 | AG1 is a vitamin mineral probiotic drink

00:36:27.800 | that's designed to meet

00:36:28.640 | all of your foundational nutrition needs.

00:36:30.860 | Now, of course, I try to get enough servings

00:36:32.480 | of vitamins and minerals through whole food sources

00:36:34.780 | that include vegetables and fruits every day,

00:36:37.040 | but oftentimes I simply can't get enough servings.

00:36:39.720 | But with AG1, I'm sure to get enough vitamins and minerals

00:36:42.680 | and the probiotics that I need,

00:36:44.440 | and it also contains adaptogens to help buffer stress.

00:36:47.600 | Simply put, I always feel better when I take AG1.

00:36:50.400 | I have more focus and energy and I sleep better,

00:36:52.940 | and it also happens to taste great.

00:36:55.160 | For all these reasons, whenever I'm asked,

00:36:56.960 | "If you could take just one supplement, what would it be?"

00:36:59.740 | I answer, "AG1."

00:37:01.420 | If you'd like to try AG1,

00:37:03.120 | go to drinkag1.com/huberman to claim a special offer.

00:37:07.600 | They'll give you five free travel packs

00:37:09.360 | plus a year supply of vitamin D3K2.

00:37:12.000 | Again, that's drinkag1.com/huberman.

00:37:16.460 | - So fast forward until a year ago,

00:37:19.260 | and I think most people took the Bannister study

00:37:21.640 | as kind of the best evidence we have for the benefits

00:37:25.640 | of metformin, and I'm sure you've had lots of people

00:37:28.160 | come up to you and ask you, "Should I be on metformin?

00:37:29.840 | "Should I be on metformin?"

00:37:30.680 | I mean, I probably get asked that question

00:37:33.520 | almost as much as I'm asked any question, outside of dew.

00:37:36.220 | I mean, people definitely wanna know

00:37:37.360 | if you should be consuming dew,

00:37:38.620 | but after that, it's metformin.

00:37:40.160 | - Fresh off the leaves.

00:37:41.040 | - Has to be.

00:37:41.880 | - While viewing morning sunlight.

00:37:43.540 | - So okay, so let's kind of fast forward to now the paper

00:37:47.200 | that I wanted to spend a few more minutes on.

00:37:48.800 | - Yeah, and thanks for that background.

00:37:49.880 | I'm still dazzled by the insertion of the straw

00:37:54.880 | by way of insulin.

00:37:57.560 | I don't think I've ever heard that described.

00:37:59.320 | I need to go get a better textbook.

00:38:02.920 | - It's a pretty short straw, in fairness.

00:38:04.720 | You know, it's just a little transport.

00:38:06.760 | - Just to give people a sense of why I'm so dazzled by it,

00:38:09.960 | I am always fascinated by how quickly, how efficiently,

00:38:15.600 | and how specifically biology can create

00:38:20.600 | these little protein complexes

00:38:22.880 | that do something really important.

00:38:24.840 | I mean, you're talking about an on-demand creation

00:38:26.680 | of a portal, right?

00:38:28.480 | I mean, these are cells engineering their own machinery

00:38:30.980 | in real time in response to chemical signals.

00:38:32.920 | - But remind me.

00:38:34.120 | - It's great.

00:38:34.960 | - Yeah, but I'm sort of rusty on my neuroscience,

00:38:37.280 | but an action potential works in reverse the same way.

00:38:40.240 | Like you need the ATP gradient to restore the gradient.

00:38:45.960 | But once the action potential fires,

00:38:47.380 | it's passive outside, right?

00:38:49.180 | - Yeah, so what Pierre's referring to is

00:38:51.740 | the way that neurons become electrically active

00:38:53.940 | is by the flow of ions from the outside of the cell

00:38:57.700 | to the inside of the cell.

00:38:58.540 | And we have both active conductances,

00:39:00.580 | meaning they're triggered by electrical changes

00:39:02.260 | in the gradients, by changes in electrical potential.

00:39:05.740 | And then there are passive gradients

00:39:07.820 | where things can just flow back and forth

00:39:09.220 | until there's a balance equal inside and outside the cell.

00:39:12.280 | I think what's different is that there's some movement

00:39:16.320 | of a lot of stuff inside of neurons

00:39:18.520 | when neurotransmitters like dopamine binds to its receptor

00:39:20.800 | and then a bunch of, it's like a bucket brigade

00:39:23.060 | that gets kicked off internally.

00:39:24.740 | But it's not often that you hear about receptors

00:39:27.340 | getting inserted into cells very quickly.

00:39:29.260 | Normally you have to go through a process

00:39:30.640 | of transcribing genes and making sure

00:39:33.600 | that the specific proteins are made.

00:39:34.980 | And then those are long, slow things that take place

00:39:36.900 | over the course of many hours or days.

00:39:38.620 | What you're talking about is a real on-demand insertion

00:39:41.760 | of a channel. - Yeah, it works in minutes.

00:39:43.180 | - And it makes sense as to why that would be required,

00:39:46.140 | but it's just, oh, so very cool.

00:39:47.660 | - It's cool, yeah.

00:39:48.620 | So Keys and colleagues came along and said,

00:39:50.500 | "We would like to redo the entire banister analysis."

00:39:54.220 | And I think their motivation for it was

00:39:58.660 | the interest in this topic is through the roof.

00:40:02.140 | There is a clinical trial called the TAME trial

00:40:05.860 | that is, I think, pretty much funded now

00:40:08.420 | and may be getting underway soon.

00:40:09.820 | The TAME trial, which is an important trial,

00:40:12.620 | is going to try to ask this question prospectively

00:40:15.340 | and through random assignment.

00:40:17.060 | - So this is the Targeting Aging with Metformin trial?

00:40:19.980 | - That's correct.

00:40:20.820 | Near Barzilai is probably the senior PI on that.

00:40:25.820 | And I think in many ways, the banister study,

00:40:30.120 | along with some other studies,

00:40:32.880 | but of lesser significance,

00:40:34.060 | probably provided some of the motivation for the TAME trial.

00:40:37.040 | So they said, "Okay, look, we're gonna do this.

00:40:38.820 | We're gonna use a different cohort of people."

00:40:41.580 | So the first study that we just talked about,

00:40:43.580 | the banister study used, I believe it was,

00:40:47.780 | roughly they sampled 95,000 subjects from a UK biobank.

00:40:52.060 | Here, they used a larger sample.

00:40:53.760 | They did about half a million people sampled

00:40:55.700 | from a Danish health registry.

00:40:59.500 | And they did something pretty elegant.

00:41:01.340 | They created two groups to study.

00:41:03.380 | So the first was just a standard replication

00:41:06.100 | of what banister did, which was just a group of people

00:41:09.880 | with and without diabetic that they tried to match

00:41:11.860 | as perfectly as possible.

00:41:13.280 | But then they did a second analysis in parallel

00:41:16.300 | with discordant twins.

00:41:18.260 | So same-sex twins that only differed

00:41:21.900 | in that one had diabetes and one didn't.

00:41:24.500 | I thought this was very elegant

00:41:25.740 | because here you have a degree of genetic similarity

00:41:29.460 | and you have similar environmental factors during childhood

00:41:34.180 | that might give you, allow you to see

00:41:36.140 | if there's any sort of difference in signal.

00:41:37.820 | So now turning this back into a little bit

00:41:40.180 | of a journal club, virtually any clinical paper

00:41:43.420 | you're gonna read, table one is the characteristics

00:41:47.860 | of the people in the study.

00:41:49.860 | You always wanna take a look at that.

00:41:51.160 | So when I look at table one here, you can see it's,

00:41:54.480 | and by the way, just for people watching this,

00:41:56.400 | we're gonna make all these papers and figures available.

00:41:58.440 | So if you're, don't, we'll have nice show notes

00:42:01.900 | that'll make all this clear.

00:42:02.820 | So table one in the keys paper

00:42:05.180 | shows the baseline characteristics.

00:42:08.220 | And again, it's almost always gonna be the first table

00:42:10.960 | in a paper.

00:42:11.800 | Usually the first figure in the paper is a study design.

00:42:14.540 | It's usually a flow chart that says,

00:42:16.600 | these were the inclusion criteria.

00:42:18.500 | These are all the people that got excluded.

00:42:20.140 | This is how we randomized, et cetera.

00:42:21.900 | And you can see here that there are four columns.

00:42:24.380 | So the first two are the singletons.

00:42:26.860 | These are people who are not related.

00:42:28.520 | And then the second two are the twins who are matched.

00:42:31.380 | And you can see, remember how I said they sampled

00:42:33.580 | about 500,000 people?

00:42:35.320 | You can see the numbers.

00:42:36.500 | So they got, you know, 7,842 singletons on metformin,

00:42:41.240 | the same number then they pulled out,

00:42:42.780 | matched without diabetes.

00:42:44.220 | On the twins, they got 976 on metformin with diabetes.

00:42:48.860 | And then by definition, 976 co-twins without them.

00:42:53.500 | And you look at all these characteristics.

00:42:55.740 | What was their age upon entry?

00:42:57.540 | How many were men?

00:42:58.860 | What was the year of indexing when we got them?

00:43:01.500 | What medications were they on?

00:43:03.200 | What was their highest level of education,

00:43:04.860 | marital status, et cetera?

00:43:06.680 | The one thing I want to call out here

00:43:08.460 | that really cannot be matched in a study like this,

00:43:11.540 | so this is a very important limitation, is the medication.

00:43:14.740 | So look at that column, Andrew.

00:43:16.780 | Notice how pretty much everything else is perfectly matched

00:43:19.500 | until you get to the medication list.

00:43:21.860 | - Yeah, it's all over the place.

00:43:22.980 | - Yeah, it's just, it's not even close.

00:43:25.420 | They're nowhere near matched, right?

00:43:27.700 | In other words, just to give you a couple of examples,

00:43:29.420 | right, on the, and let's just talk about the singletons,

00:43:32.160 | 'cause it's basically the same story on the twins.

00:43:33.860 | If you look at what fraction of the people

00:43:36.460 | with type 2 diabetes are on lipid-lowering medication,

00:43:39.860 | it's 45.6% versus 15.4% in the matched without diabetes.

00:43:44.860 | It's a 3X difference.

00:43:47.180 | What about antiplatelet therapy?

00:43:48.800 | That's 30% versus 14%.

00:43:51.220 | Antihypertensive, 65% versus 63% versus 31%.

00:43:55.380 | - Because people who have one health issue

00:43:57.220 | and are taking metformin are likely

00:43:58.620 | to have other health issues.

00:43:59.660 | - Exactly.

00:44:00.620 | So this is, again, a fundamental flaw of epidemiology.

00:44:04.980 | You can never remove all the confounders.

00:44:08.820 | - This is why I became an experimental scientist,

00:44:11.300 | so that we could control variables.

00:44:13.060 | - That's right, because without random assignment,

00:44:15.260 | you cannot control every variable.

00:44:17.100 | Now you'll see in a moment when we get into the analysis,

00:44:19.860 | they go through three levels of corrections,

00:44:23.200 | but they can never correct this medication one.

00:44:25.740 | So just keep that in the back of your mind.

00:44:27.220 | Okay, so the two big things that were done

00:44:30.740 | in this experiment, or in this survey, or study,

00:44:34.260 | to differentiate it from Bannister was, one,

00:44:36.500 | the twin trick, which I think is pretty cool.

00:44:39.100 | The second thing that they did was they did

00:44:42.960 | a sensitivity analysis with and without

00:44:45.560 | informative censoring.

00:44:47.340 | So one of the other things they wanted to know was,

00:44:48.860 | hey, does it really matter if we don't count

00:44:52.420 | the metformin patients who progress?

00:44:55.260 | So let's see kind of what transcribed.

00:44:59.620 | So the next figure, figure two,

00:45:02.340 | pardon me, the next table, table two,

00:45:04.180 | walks you through the crude mortality rate

00:45:08.460 | in each of the groups.

00:45:10.160 | So the most important row, I think, in this table

00:45:14.740 | is the one that says crude mortality

00:45:16.800 | per thousand person years.

00:45:18.940 | Now you recall that in the previous study,

00:45:21.780 | in the Bannister study, those were on the ballpark

00:45:25.060 | about 15 per.

00:45:27.100 | Okay, so let's look at each of these.

00:45:29.740 | So in the single, the singletons, without,

00:45:34.580 | so the non-twins who were not diabetic,

00:45:37.660 | it was 16.86.

00:45:39.740 | - And could you put a little more contour

00:45:41.700 | on what this thousand person years?

00:45:44.460 | - What it is?

00:45:45.300 | - Are you talking about pooling the lifespans

00:45:47.300 | of a bunch of different people until you get

00:45:50.100 | to the number 1,000?

00:45:51.540 | - Yeah, because this-- - Because you're normalizing,

00:45:53.540 | so it's not who's gonna live a thousand years,

00:45:55.380 | 'cause no one's expecting that.

00:45:56.940 | You're essentially taking, so you've got some people

00:45:59.720 | that are gonna live 76 years, 52 years, 91 years,

00:46:04.660 | and you're pooling all of those until you hit a thousand.

00:46:07.300 | And then that becomes kind of a,

00:46:10.540 | it's like a normalized division.

00:46:12.060 | You're basically like, so let's say the control group,

00:46:16.220 | you're asking if there were a thousand person years

00:46:19.700 | available to live, how likely is it that this person

00:46:22.740 | would live another 15 years?

00:46:24.700 | - Yeah, so a couple of ways to think about it.

00:46:26.220 | So taking a step back, we always have to have

00:46:28.320 | some way of normalizing.

00:46:29.420 | So when we talk about the mortality from a disease

00:46:32.060 | like cancer in the population, we report it as

00:46:36.620 | what's the mortality rate per,

00:46:38.940 | and it's typically per 100,000 persons.

00:46:41.860 | - Okay, that's a much more intuitive way to express it.

00:46:44.500 | - It is, but the reason we can do it that way

00:46:47.460 | is because we're literally looking at how many people died

00:46:51.420 | this calendar year, and we divide it by the number

00:46:54.660 | of people in that age group.

00:46:55.820 | So it's typically what you're doing when you look

00:46:59.100 | at aged groups in buckets of like decades.

00:47:03.160 | So that's why we can say the highest mortality

00:47:06.600 | is like people 90 and up.

00:47:09.580 | Even though the absolute number of deaths is small,

00:47:12.100 | it's because there's not that many people there, right?

00:47:14.560 | The majority of deaths in absolute terms probably occur

00:47:18.020 | in the seventh decade, but as you go up,

00:47:22.140 | because the denominator is shrinking,

00:47:23.860 | you have to normalize to it.

00:47:25.100 | So we just normalize to the number of people.

00:47:26.960 | Here are all the people that started the year.

00:47:28.760 | Here are all the people that ended the year.

00:47:30.180 | What's the death rate?

00:47:31.140 | Why are these done in a slightly more complicated way?

00:47:34.200 | Because we don't follow these people for their whole lives.

00:47:37.620 | We're only following them for a period of observation,

00:47:39.820 | in this case, roughly three years.

00:47:42.060 | So to say something like we have a crude death rate

00:47:46.500 | of five deaths per thousand person years,

00:47:49.980 | one way to think about that is if you had a thousand people

00:47:53.960 | and you followed them for one year,

00:47:56.580 | you'd expect five to die.

00:47:58.560 | If you had 500 people and you followed them for two years,

00:48:02.220 | you expect five to die.

00:48:04.420 | If you have a thousand people and you follow them

00:48:06.700 | for one year, you expect five to die.

00:48:08.740 | Those would all be considered equivalent mortalities.

00:48:11.940 | - Great, thank you for clarifying that.

00:48:13.300 | - No, no, this stuff is, I mean,

00:48:16.220 | like I find epidemiology, when you get in the weeds,

00:48:19.180 | is way more complicated than following the basics

00:48:22.120 | of experimental stuff, where you just,

00:48:25.460 | you get to push all this stuff into the garbage bin

00:48:28.660 | and just say, hey, we're gonna take this number of people,

00:48:30.900 | we're gonna exclude this group, we're gonna randomize,

00:48:33.100 | we're gonna see what happens.

00:48:33.980 | - Yeah, that's what, like the paper we'll talk about next.

00:48:36.200 | - Yep, yep.

00:48:37.700 | So when you adjust for age,

00:48:41.940 | and they don't show it in this table, it's only in the text,

00:48:44.600 | when you adjust for age, a very important check to do

00:48:48.800 | is what is the crude death rate of the people on metformin

00:48:53.280 | who are not twins versus who are twins.

00:48:55.460 | Now in this table, they look different,

00:48:58.000 | because it's 24.93 for the metformin group

00:49:01.320 | and 21.68 for the twin group that's on metformin.

00:49:05.900 | When you adjust for age, they're almost identical.

00:49:08.320 | It goes from 24.93 to 24.7.

00:49:13.560 | One other point I'll make here for people

00:49:15.220 | who are gonna be looking at this table is,

00:49:17.740 | you'll notice there are parentheses

00:49:19.100 | after every one of these numbers.

00:49:21.540 | What does that offer in there?

00:49:23.460 | Those parentheses are offering the 95% confidence interval.

00:49:27.820 | So for example, to take the number 24.93

00:49:32.180 | is the crude death rate of how many people are dying

00:49:35.060 | who take metformin.

00:49:36.420 | What it's telling you is we're 95% confident

00:49:39.300 | that the actual number is between 23.23 and 26.64.

00:49:43.700 | If a 95% confidence interval does not cross the number zero,

00:49:49.600 | it's statistically significant.

00:49:52.920 | Okay, so the first thing that just jumps out at you,

00:49:57.960 | I think when you look at this,

00:49:58.960 | is there's clearly a difference here

00:50:01.000 | between the people who have diabetes and those who don't.

00:50:04.640 | It complicates the study a little bit,

00:50:06.000 | 'cause it's basically two studies in one,

00:50:08.240 | but you're comparing 95, pardon me,

00:50:13.200 | 24.93 to 16.86, which by the way remains after age adjustment

00:50:18.200 | when you go to the twin group, it's 24.73 to 12.94.

00:50:23.280 | - So maybe just to zoom out for that,

00:50:24.780 | what you're describing, if I understand correctly,

00:50:26.760 | is this crude deaths per 1000 person years,

00:50:31.760 | let's just talk about the singletons, the non-twins,

00:50:34.340 | is 16.86, so 16.86 people die,

00:50:38.660 | and some people are probably thinking,

00:50:39.500 | "How can 0.86 of a person die?"

00:50:41.520 | Well, it's not always whole numbers,

00:50:43.100 | but there's a bad joke to be made here, but--

00:50:47.380 | - Yeah, just call it 17 versus 25.

00:50:49.460 | - Right, 17 deaths per 1000 versus 25 deaths.

00:50:54.460 | And the 25 is in the folks that took metformin.

00:50:58.340 | Now, that to the naive listener and to me means,

00:51:03.140 | oh, metformin basically kills you, right?

00:51:06.460 | Not faster, or you're more likely to die,

00:51:09.340 | but we have to remember that these people have another,

00:51:12.840 | they have a major health issue

00:51:14.020 | that the other group does not have.

00:51:15.640 | - That's right.

00:51:16.480 | - Because people weren't assigned drug or not assigned drug.

00:51:19.140 | It wasn't placebo drug.

00:51:20.780 | It's let's look at people taking this drug

00:51:22.720 | for a bad health issue and compare to everyone else.

00:51:26.880 | - That's right.

00:51:28.220 | So now you have to go into,

00:51:31.540 | and I'll just sort of skip the next figure,

00:51:33.380 | but the next figure is a Kaplan-Meier curve.

00:51:35.900 | I think it's actually worth looking at it

00:51:37.460 | 'cause they show up in all sorts of studies.

00:51:40.020 | So if you look at figure one, it's a Kaplan-Meier curve,

00:51:43.260 | which is a mortality curve.

00:51:45.500 | So you'll see these in any study that is looking at death.

00:51:50.420 | And this can be prospective randomized,

00:51:52.380 | this can be retrospective,

00:51:53.660 | but these are always gonna show up.

00:51:55.740 | And I think it's really worth understanding

00:51:57.340 | what a Kaplan-Meier curve shows you.

00:51:59.220 | So on the X-axis is always time and on the Y-axis

00:52:02.540 | is always the cumulative survival.

00:52:05.060 | So it's a curve that always goes from zero to one,

00:52:08.980 | one or 100%, and it's always decreasing monotonically,

00:52:13.920 | meaning it can only go down or stay flat.

00:52:16.820 | It can never go back up.

00:52:18.660 | So that's what a cumulative mortality curve looks like.

00:52:22.340 | - Now we're looking at, you're starting it alive

00:52:25.560 | and you're looking at how many people die

00:52:27.700 | for every year that passes.

00:52:29.020 | - That's right.

00:52:30.340 | And in each curve, there's one on the left,

00:52:33.660 | which is the matched singletons,

00:52:35.380 | and there's the one on the right,

00:52:36.260 | which are the discordant twins.

00:52:38.140 | You have two lines.

00:52:39.360 | You have those that were on metformin with type 2 diabetes

00:52:42.780 | and you have their matched controls.

00:52:45.180 | And in this figure, the matched controls

00:52:47.580 | are the darker lines and the people with type 2 diabetes

00:52:51.540 | on metformin, that's the lighter line.

00:52:54.060 | You'll also notice, and I like the way they've done it here,

00:52:56.460 | they've got shading around each one.

00:52:58.500 | And we should mention for those that are just listening

00:53:00.820 | that in both of these graphs,

00:53:02.980 | the downward trending line from the controls,

00:53:07.460 | so again, non-diabetic, not taking metformin,

00:53:10.220 | is above the line corresponding to the diabetics

00:53:15.220 | who are taking metformin.

00:53:16.500 | Put crudely, the people who are taking metformin

00:53:22.620 | that have diabetes are dying at a faster rate

00:53:25.860 | for every single year examined.

00:53:27.320 | The two lines do not overlap except at the beginning

00:53:29.740 | when everyone's alive.

00:53:30.860 | It's like a foot race where basically the people

00:53:32.420 | with metformin and diabetes are falling behind

00:53:35.900 | and dying as they fall.

00:53:37.740 | - That's right, and I'm glad you brought up a good point.

00:53:40.460 | It's not uncommon in treatments

00:53:43.440 | to see Kaplan-Meier curbs cross.

00:53:46.140 | They don't have to.

00:53:46.980 | It's not a requirement that they never cross.

00:53:49.140 | It's only a requirement

00:53:51.020 | that they're monotonically decreasing or staying flat.

00:53:54.140 | So I've seen cancer treatment drugs

00:53:56.540 | where they have two drugs going head-to-head

00:53:58.540 | in a cancer treatment,

00:53:59.500 | and one starts out looking really, really bad,

00:54:02.960 | but then all of a sudden it kind of flattens

00:54:04.580 | while the other one goes bad,

00:54:05.900 | and then it actually crosses and goes underneath.

00:54:08.240 | But that's not the case here.

00:54:09.400 | So to your point, the people with diabetes taking metformin

00:54:14.240 | in both the matched singletons and the discordants

00:54:17.100 | are dropping much faster, and they always stay below.

00:54:21.060 | And I was just gonna say that the shading

00:54:22.840 | is just showing you a 95% confidence interval.

00:54:25.320 | So you're just putting basically error bars along this.

00:54:28.540 | So if this were experimental data,

00:54:30.420 | if you were doing an experiment with a group of mice

00:54:33.980 | and you were watching their survival,

00:54:36.180 | and you'd have error bars on this

00:54:38.880 | which you're actually measuring.

00:54:40.060 | So this is, because you have much more data here,

00:54:42.500 | you're just showing this in this fashion.

00:54:43.800 | - For those that haven't been familiar as to statistics,

00:54:46.600 | no problem, error bars correspond to,

00:54:48.800 | like if you were just gonna measure the heights

00:54:50.340 | of a room full of 10th graders,

00:54:52.460 | there's gonna be a range, right?

00:54:53.600 | You have the very tall kid and the very shorter kid,

00:54:57.460 | and you have the short kid and the medium kid.

00:54:59.140 | And so there's a range.

00:55:00.080 | There's gonna be an average, a mean,

00:55:01.460 | and then there'll be standard deviations and standard errors.

00:55:04.400 | And so these confidence intervals

00:55:08.180 | just give a sense of how much range.

00:55:10.240 | Some people die early, some people die late.

00:55:14.580 | Within a given year, they're gonna be different ages.

00:55:17.740 | So these error bars can account

00:55:20.020 | for a lot of different forms of variability.

00:55:21.560 | Here, you're talking about the variability

00:55:23.700 | is how many people in each group die.

00:55:26.860 | We're not tracking one diabetic taken metformin

00:55:29.440 | versus a control.

00:55:31.440 | I should have asked this earlier, but--

00:55:33.900 | - Well, and it's also a mathematical model

00:55:35.660 | at this point too that's smoothing it out.

00:55:38.220 | 'Cause notice it's running for the full eight years,

00:55:40.400 | even though they're only following people for,

00:55:43.160 | you know, typically, I think the median

00:55:44.900 | was like three or four years at a time.

00:55:46.980 | So they're using this quite complicated type of mathematics

00:55:50.580 | called a Cox proportional hazard,

00:55:52.140 | which is what generates hazard ratios.

00:55:54.420 | And basically, any model has to have some error in it.

00:55:58.180 | And so they're basically saying this is the error.

00:56:00.420 | So you could argue when you look at that figure,

00:56:03.420 | we don't know exactly where the line is in there,

00:56:05.820 | but we know it's in that shaded area.

00:56:07.900 | Sorry, one other point.

00:56:10.800 | If those shaded areas overlapped,

00:56:13.900 | you couldn't really make the conclusion.

00:56:16.400 | You wouldn't know for sure

00:56:18.140 | that one is different from the other.

00:56:19.660 | - Yeah, that's actually a good opportunity

00:56:21.180 | to raise a common myth,

00:56:25.340 | which is a lot of people, when they look at a paper,

00:56:28.380 | let's say it's a bar graph, you know,

00:56:31.380 | and they see these error bars,

00:56:33.460 | and they will say, people often think,

00:56:35.660 | oh, if the error bars overlap,

00:56:37.980 | it's not a significant difference.

00:56:40.020 | But if the error bars don't overlap,

00:56:41.940 | meaning there's enough separation,

00:56:43.260 | then that's a real and meaningful difference.

00:56:44.900 | And that's not always the case.

00:56:46.580 | It depends a lot on the form of the experiment.

00:56:49.880 | I often see some of the more robust Twitter battles

00:56:52.760 | over how people are reading graphs.

00:56:54.880 | And I think it's important to remember

00:56:56.480 | that you run the statistics,

00:56:59.360 | hopefully the correct statistics for the sample,

00:57:02.160 | but determining significance,

00:57:04.140 | whether or not the result could be due

00:57:05.960 | to something other than chance.

00:57:07.760 | Of course, your confidence in that increases

00:57:11.040 | as it becomes typically p-values,

00:57:12.960 | p less than 0.00001% chance that it's due to chance, right?

00:57:18.440 | So very low probably, p less than 0.05

00:57:20.680 | tends to be the kind of gold standard cutoff.

00:57:23.500 | But when you're talking about data like these,

00:57:26.600 | which are repeated measures over time,

00:57:28.480 | people are dropping out literally over time,

00:57:32.120 | you're saying they've modeled it to make predictions

00:57:34.440 | as to what would happen.

00:57:35.420 | We're not necessarily looking at raw data points here.

00:57:38.200 | - Yeah, the raw data was in the previous table.

00:57:40.520 | That's now taken and run through this Cox model,

00:57:44.340 | and it's smoothed out.

00:57:46.620 | And to your point about the bar graphs,

00:57:48.720 | yeah, I think the other thing you always wanna understand

00:57:51.500 | is just because something

00:57:52.880 | doesn't achieve statistical significance,

00:57:55.840 | the only way you can say it's not significant

00:57:58.360 | is you have to know what it was powered to detect.

00:58:01.060 | And statistical power is a very important concept

00:58:06.740 | that probably doesn't get discussed enough.

00:58:08.880 | But before you do an experiment,

00:58:11.160 | you have to have an expectation

00:58:13.300 | of what you believe the difference is between the groups.

00:58:16.600 | And you have to determine the number of samples

00:58:20.420 | you will need to assess whether or not

00:58:23.720 | that difference is there or not.

00:58:25.300 | So you use something, it's called a power table,

00:58:29.300 | and you would go to the power table.

00:58:30.840 | So if you're doing treatment A versus treatment B,

00:58:33.600 | and you say, well, look, I think treatment A

00:58:35.260 | is going to have a 50% response,

00:58:37.880 | and I think treatment B will have a 65% response.

00:58:41.620 | You literally go to a power table that says 50% response,

00:58:46.160 | 15% difference, that gives you a place on the grid,

00:58:50.080 | and I wanna be 90% sure that I'm right, so 90% power.

00:58:54.120 | I'm being a little bit,

00:58:55.080 | so there's gonna be a statistician listening to this

00:58:56.680 | who's gonna wanna kill me,

00:58:57.880 | but this is directionally the way we would describe it.

00:59:00.440 | And that tells you this is how many animals or people

00:59:03.680 | you would need in this study.

00:59:05.220 | You're gonna need 147 in each group.

00:59:08.640 | And by the way, if you now do the experiment with 147

00:59:12.320 | and you fail to find significance,

00:59:14.440 | you can comfortably say there is no statistical difference

00:59:17.900 | at least up to that 15%.

00:59:20.080 | There may be a difference at 10%,

00:59:22.120 | but you weren't powered to look at 10%.

00:59:24.460 | - Yeah, and very important point that you're making.

00:59:27.120 | Another point that's just a more general one

00:59:29.340 | about statistics, in general,

00:59:31.680 | the way to reduce variability in a data set

00:59:34.280 | is to increase sample size.

00:59:36.040 | And that kinda makes sense, right?

00:59:37.000 | If I just walk into a 10th grade class and go,

00:59:39.040 | hey, I'm gonna measure height,

00:59:40.480 | and I look up by the first three kids that I see,

00:59:43.400 | and I happen to look over there,

00:59:44.880 | and it's the three that all play

00:59:46.120 | on the volleyball team together.

00:59:48.240 | My sample size is small,

00:59:49.960 | and I'm likely to get a skewed representation,

00:59:53.080 | in this case, taller than average.

00:59:55.040 | So increasing sample size tends to decrease variation.

00:59:58.320 | So that's why when you hear about a study

01:00:01.000 | from the UK Biobank or from half a million Danish citizens,

01:00:05.920 | like for instance, in this study,

01:00:08.000 | those are enormous sample sizes.

01:00:09.880 | So even though this is not an experimental study,

01:00:13.060 | it's an epidemiological observational study,

01:00:15.460 | there's tremendous power by way of the enormous number

01:00:19.920 | of subjects in this study.

01:00:20.760 | - And that's the way that epidemiology

01:00:23.060 | will make up for its deficit.

01:00:24.760 | So you could never do a randomized assignment study

01:00:29.160 | on half a million people.

01:00:30.540 | So epidemiology makes up for its biggest limitation,

01:00:36.740 | which is it can never compensate for inherent biases

01:00:41.460 | by saying we can do infinite duration if we want,

01:00:44.300 | like we could survey people over the course of their lives,

01:00:47.320 | and we can have the biggest sample size possible,

01:00:49.520 | 'cause this is relatively cheap.

01:00:51.560 | The cost of actually doing an experiment

01:00:54.420 | where you have tens of thousands of people is prohibitive.

01:00:56.400 | I mean, if you look at the Women's Health Initiative,

01:00:57.920 | which was a five-year study on, I don't know,

01:01:00.640 | what was it, 50,000 women,

01:01:02.140 | I mean, that was a billion-dollar study.

01:01:04.360 | So this is the balancing act between epidemiology

01:01:09.320 | and randomized prospective experiments.

01:01:12.160 | And so they both offer something,

01:01:14.740 | but you just have to know the blind spots of each one.

01:01:17.720 | - I'd like to take a quick break

01:01:18.840 | and acknowledge our sponsor, InsideTracker.

01:01:21.520 | InsideTracker is a personalized nutrition platform

01:01:23.880 | that analyzes data from your blood and DNA

01:01:26.380 | to help you better understand your body

01:01:27.920 | and help you meet your health goals.

01:01:29.820 | I'm a big believer in getting regular blood work done

01:01:32.140 | for the simple reason that many of the factors

01:01:34.460 | that impact your immediate and long-term health

01:01:36.680 | can only be analyzed from a quality blood test.

01:01:39.260 | However, with a lot of blood tests out there,

01:01:41.300 | you get information back about blood lipids,

01:01:43.180 | about hormones, and so on,

01:01:44.320 | but you don't know what to do with that information.

01:01:46.340 | With InsideTracker, they have a personalized platform

01:01:48.640 | that makes it very easy to understand your data,

01:01:51.400 | that is, to understand what those lipids,

01:01:53.320 | what those hormone levels, et cetera, mean,

01:01:55.300 | and behavioral, supplement, nutrition, and other protocols

01:01:58.480 | to adjust those numbers to bring them into the ranges

01:02:01.280 | that are ideal for your immediate and long-term health.

01:02:03.560 | InsideTracker's ultimate plan now includes measures

01:02:05.840 | of both ApoB and of insulin, which are key indicators

01:02:09.480 | of cardiovascular health and energy regulation.

01:02:12.280 | If you'd like to try InsideTracker,

01:02:13.720 | you can visit insidetracker.com/huberman

01:02:16.520 | to get 20% off any of InsideTracker's plans.

01:02:19.300 | Again, that's insidetracker.com/huberman to get 20% off.

01:02:24.300 | - So let's just kind of wrap this up.

01:02:25.640 | I mean, I think let's just go to table four,

01:02:27.560 | which I think is the most important table in here,

01:02:31.280 | which now lays out the final results

01:02:34.320 | in terms of the hazard ratio.

01:02:35.840 | So this is the way we wanna really be thinking about this.

01:02:38.360 | So again, hazard ratios,

01:02:41.000 | these are important things to understand.

01:02:42.600 | A hazard ratio is a number,

01:02:44.440 | and you always subtract one from the hazard ratio,

01:02:48.500 | and that tells you, if it's a positive number,

01:02:51.200 | if it's a number, sorry,

01:02:52.040 | if it's a number greater than one, you subtract one,

01:02:53.960 | and that tells you the relative harm.

01:02:56.400 | So if the hazard ratio is 1.5,

01:02:58.240 | you subtract 1.5 is a 50% increase in risk.

01:03:01.700 | If the number is negative,

01:03:04.160 | you may recall on the Bannister paper,

01:03:06.000 | the hazard ratio was 0.85.

01:03:08.080 | So that means it's a 15% reduction in relative risk,

01:03:12.200 | and here you can see all the hazard ratios are positive.

01:03:15.340 | So what it's telling you here is,

01:03:17.340 | and I'm gonna walk through this

01:03:18.260 | 'cause there's a lot of information packed here.

01:03:21.080 | You've got singletons, you've got twins.

01:03:22.880 | They're showing you three different ways that they do it.

01:03:25.520 | They do an unadjusted model.

01:03:27.600 | If you just look at the singletons

01:03:30.040 | with and without metformin and you make no adjustments,

01:03:33.720 | the hazard ratio is 1.48,

01:03:36.620 | meaning the people on metformin had a 48% greater chance

01:03:40.960 | of dying in any given year

01:03:43.200 | than their non-diabetic counterpart.

01:03:45.200 | - The only reason I'm smiling,

01:03:46.280 | it's not 'cause I enjoy people dying quite to the contrary,

01:03:49.600 | is that this is novel for me,

01:03:52.640 | and that I've read some epidemiological studies before,

01:03:54.720 | but it's not normally where I spend the majority of my time.

01:03:58.160 | But up until now, I was thinking,

01:03:59.400 | okay, people taking metformin are dying more

01:04:02.280 | than those that aren't,

01:04:03.120 | and I'm just relieved to know

01:04:05.120 | that I wasn't looking at all this backwards, okay?

01:04:08.700 | So they're dying more,

01:04:09.800 | but of course, we don't have a group

01:04:11.740 | that's taking metformin who doesn't have diabetes,

01:04:13.760 | and we don't have a group who has diabetes

01:04:18.760 | and is taking metformin plus something else.

01:04:21.480 | So again, we're only dealing with these

01:04:23.040 | constrained populations.

01:04:24.240 | - Yeah, now there's another arm to this study

01:04:26.260 | that I'm not getting into 'cause it adds more complexity,

01:04:29.680 | which is they also have another group

01:04:31.580 | that's got diabetes, takes metformin,

01:04:33.880 | and takes sulfonylureas, which is a bigger drug,

01:04:37.120 | and those people die even more.

01:04:39.120 | - Whoa, so how?

01:04:40.440 | - Which again, speaks to the point, right?

01:04:43.280 | The more you need these medications,

01:04:45.920 | they're never able to erase the effect of diabetes.

01:04:49.700 | - But in this case, it seems that they might be accelerating,

01:04:53.120 | possibly accelerating death due to diabetes, possibly.

01:04:55.920 | - We could never know that from this

01:04:57.880 | because we don't see,

01:05:00.100 | we would need to see diabetics who don't take metformin,

01:05:02.760 | who take nothing,

01:05:03.840 | and I would bet that they would do even worse.

01:05:06.200 | So my intuition is that the metformin is helping,

01:05:08.960 | but not helping nearly as much as we thought before.

01:05:13.780 | So my point is they make another set of adjustments.

01:05:16.240 | They say, okay, well look, in the first one,

01:05:18.160 | in the unadjusted model, we only matched for age and gender.

01:05:22.720 | Okay, that's pretty crude.

01:05:24.280 | What if we adjust for the medications they're on,

01:05:27.200 | cardiovascular, psychiatric, pulmonary, dementia, meds,

01:05:30.820 | and marital status?

01:05:31.900 | I don't know why they threw marital status in there,

01:05:33.400 | but they did.

01:05:34.240 | - I don't know, maybe being married or unmarried can--

01:05:36.200 | - I'm sure it can, but it just seems like a random thing

01:05:38.200 | to throw in with all their meds.

01:05:39.220 | I would have personally done that adjustment higher up,

01:05:41.340 | but nevertheless, if you do that,

01:05:43.580 | all of a sudden the hazard ratio drops from 1.48 to 1.32,

01:05:48.580 | which means, yep, you still have a 32% greater chance

01:05:52.700 | of dying in any given year.

01:05:54.920 | All right, what if we also adjust for the highest level

01:05:58.880 | of education along with any of the other covariates?

01:06:02.380 | Well, that doesn't really change it at all.

01:06:03.880 | It ends up at 1.33, or a 33% chance increase in death.

01:06:08.160 | - I always knew that more school wasn't gonna save me.

01:06:10.480 | - It's not doing jack.

01:06:11.720 | So now let's do it for the twins.

01:06:13.800 | If you do the twin study, which you could argue

01:06:15.760 | is a slightly purer study, because you at least have

01:06:18.920 | one genetic and environmental thing that you've attached,

01:06:22.940 | the unadjusted model is brutal, 2.15.

01:06:26.520 | That's 115%.

01:06:28.620 | Think about this.

01:06:29.460 | These are twins who, in theory, are the same in every way,

01:06:33.020 | except one has diabetes and one doesn't,

01:06:34.980 | and the one with diabetes on metformin still has 115%

01:06:38.960 | greater chance of dying than the non-diabetic co-twin.

01:06:42.780 | When you make that first adjustment of all the meds

01:06:45.140 | and marital status, you bring it down to a 70% increase

01:06:48.060 | in risk, and when you throw education in,

01:06:50.180 | it goes up to an 80% chance of risk.

01:06:53.100 | Now, they did this really cool thing,

01:06:55.400 | which was they did the analysis

01:06:58.060 | on with and without censoring.

01:07:00.140 | So everything I just said here was based on no censoring.

01:07:04.900 | - Tell me about censoring.

01:07:06.020 | - Censoring is when you stop counting the metformin people

01:07:08.780 | who have died.

01:07:09.620 | Okay, so in the singleton group, when you unadjust it,

01:07:15.580 | and the reason I'm doing the unadjusted is that's where

01:07:17.620 | they did the sensitivity analysis.

01:07:19.280 | I don't think it really matters that much.

01:07:21.260 | You just have to draw a line in the sand somewhere.

01:07:23.360 | You'll recall that that was a 48% chance

01:07:26.580 | of increased mortality, all-cause mortality,

01:07:29.700 | if you stop counting, pardon me, if you don't censor,

01:07:34.220 | meaning if you include everybody,

01:07:36.300 | including when people on metformin with diabetes die.

01:07:39.020 | If you censor them, it comes down to 1.39.

01:07:43.020 | In other words, this is a very important finding,

01:07:45.560 | it did not undo the benefits that we saw

01:07:48.640 | in the Bannister study.

01:07:50.540 | Bannister saw a 15% reduction in mortality

01:07:55.140 | when they censored.

01:07:56.140 | When keys censored, it got better, but not that much better.

01:08:02.860 | It went from 48 to 39%.

01:08:04.820 | In the twins, it went from 115% down to only 97%.

01:08:09.820 | So in some ways, this presents a little bit of an enigma,

01:08:17.140 | because it's not entirely clear to me,

01:08:19.280 | having read these papers many times,

01:08:21.360 | exactly why Bannister found such an outline,

01:08:24.680 | like such a different response.

01:08:27.020 | There's another technical detail of this paper,

01:08:30.220 | which is they, you can see on the right side of table four,

01:08:33.160 | they did something called a nested case control.

01:08:36.540 | But you'll see, and I was gonna go into a long explanation

01:08:39.380 | of what nested case controls are.

01:08:41.340 | It's another pretty elegant way to do case control studies,

01:08:45.500 | where you sample by year and you sort of normalize,

01:08:50.100 | you don't count all the cases at the end,

01:08:52.180 | you count them one by one.

01:08:53.980 | I don't think it's worth getting into, Andrew,

01:08:55.560 | 'cause it doesn't change the answer.

01:08:56.960 | You can see it changes it just slightly,

01:08:58.920 | but it doesn't change the point.

01:09:00.540 | The point here is, the keys paper makes it undeniably clear

01:09:05.380 | that in that population, there was no advantage

01:09:08.980 | offered by metformin that undid the disadvantage

01:09:12.900 | of having type two diabetes.

01:09:14.380 | This does not mean that metformin wasn't helping them,

01:09:16.940 | because we don't know what these people would have been like

01:09:19.860 | without metformin.

01:09:20.700 | It could be that this bought them a 50% reduction

01:09:23.980 | in relative mortality to where they'd been.

01:09:26.920 | But what it says is, in a way,

01:09:29.900 | this is what you would have expected.

01:09:31.680 | This is what you would have expected 10 years ago

01:09:34.100 | before the Bannister paper came out.

01:09:35.740 | - Or maybe even before metformin was used,

01:09:37.860 | because in some ways it's saying,

01:09:39.900 | what is the likelihood that sick people

01:09:41.600 | who are on a lot of medication are gonna die

01:09:43.560 | compared to not sick people

01:09:45.020 | who aren't on a lot of medication?

01:09:46.920 | - Yep.

01:09:47.760 | - One of, you know, it's not quite that simple

01:09:49.940 | in the sense that, as you said,

01:09:51.460 | there are ways to try and isolate

01:09:54.380 | the metformin contribution somewhat,

01:09:56.740 | because they're on a bunch of other meds.

01:10:00.380 | And presumably that was done and analyzed in other figures

01:10:04.140 | where they can sort of try and,

01:10:06.340 | they can never attach the results

01:10:08.300 | specifically to metformin, right?

01:10:10.660 | But there must be some way of weighting

01:10:13.480 | the percentage that are on psychiatric meds

01:10:16.200 | or not on psychiatric meds as some way to tease out

01:10:18.480 | whether or not there's actually

01:10:20.000 | some contribution in metformin to this result.

01:10:22.840 | - Well, that's what they're doing in the partial adjustment,

01:10:26.540 | is they're actually doing their best to say-

01:10:30.560 | - Oh, right, married or not married,

01:10:31.760 | they're going variable by variable.

01:10:32.600 | - They're going drug by drug all the way through,

01:10:34.640 | high blood pressure, non-high blood pressure,

01:10:36.100 | smoking, nonsmoking, et cetera.

01:10:37.640 | - Right, and the way they would do that, presumably,

01:10:39.080 | is by saying, okay, married, not married,

01:10:41.800 | that's a simple one.

01:10:43.020 | - Are you on lipid lowering meds, yes or no?

01:10:46.660 | Okay, you are not, you are not.

01:10:50.340 | - And then comparing those groups.

01:10:51.500 | - Yeah, yeah, okay.

01:10:52.500 | - So no differences jumping out

01:10:55.120 | that can be purely explained by these other variables.

01:10:58.680 | - Yes, although, again, this is a great opportunity

01:11:01.020 | to talk about why, no matter how slick you are,

01:11:03.140 | no matter how slick your model is,

01:11:04.360 | you can't control for everything.

01:11:06.060 | There is a reason that, to my knowledge,

01:11:08.000 | virtually every study that compares meat eaters

01:11:11.040 | to non-meat eaters finds an advantage

01:11:13.480 | amongst the non-meat eaters.

01:11:16.120 | And we can talk about all the-

01:11:17.120 | - Lifespan advantage.

01:11:18.120 | - Yes, and we can, or disease incidence studies.

01:11:22.320 | And yeah, it might be tempting to say,

01:11:24.400 | well, therefore, eating meat is bad,

01:11:26.860 | until you realize that it takes a lot of work

01:11:29.900 | to not eat meat.

01:11:30.960 | That's a very, very significant decision

01:11:33.600 | that a person, for most people,

01:11:34.760 | that's a very significant decision a person makes.

01:11:37.200 | And for a person to make that decision,

01:11:38.560 | they probably have a very high conviction

01:11:40.000 | about the benefit of that to their health.

01:11:42.760 | And it is probably the case that they're making

01:11:45.560 | other changes with respect to their health as well

01:11:48.320 | that are a little more difficult to measure.

01:11:50.560 | Now, there's a million other problems with that.

01:11:52.280 | I picked a silly example,

01:11:53.440 | because the whole meat discussion then gets into,

01:11:56.160 | well, when we say eating meat, what do we mean?

01:11:59.240 | - Yeah, we're not talking about,

01:12:00.080 | it's like deli meat versus grass-fed.

01:12:01.600 | - Exactly.

01:12:02.440 | - Or a deer that you hunted with your ball.

01:12:05.040 | - That's right, so how do we get into all those things?

01:12:07.000 | But my point is, it's very difficult to quantify

01:12:10.120 | some of the intangible differences.

01:12:12.080 | And I think that even a study that goes to great lengths,

01:12:14.560 | as this one does, epidemiologically,

01:12:16.600 | to make these corrections can never make the corrections.

01:12:18.700 | And so, for me, the big takeaway of this study is,

01:12:21.360 | one, this makes much more sense to me

01:12:25.240 | than the Bannister paper,

01:12:26.160 | which never really made sense to me.

01:12:28.060 | And again, I was first critical of the Bannister paper

01:12:30.460 | in 2018, about four years after comment.

01:12:32.320 | That's about the time I stopped taking Metformin, by the way.

01:12:34.320 | I stopped taking it for a different reason,

01:12:35.560 | which we can talk about in a sec.

01:12:37.320 | But that was the first time I went back and said,

01:12:40.060 | wait a minute, this information,

01:12:41.680 | this informative censoring thing is,

01:12:44.280 | that's a little fishy.

01:12:45.520 | And I think we weren't looking at a true group

01:12:48.440 | of real type 2 diabetics.

01:12:49.600 | Now, that said, maybe it doesn't matter.

01:12:52.240 | In other words, maybe, and even the Keys paper

01:12:56.520 | doesn't tell us that Metformin wouldn't be beneficial,

01:12:58.880 | because it could be that those people,

01:13:01.580 | if they were on nothing,

01:13:03.360 | as their matched cohorts were on nothing,

01:13:05.720 | would have been dying at a hazard ratio of three,

01:13:08.960 | and this brought it down to 1.5,

01:13:11.400 | in which case you would say,

01:13:12.280 | there is some zero protection there.

01:13:14.260 | It is putting the brakes on this process.

01:13:17.120 | All of this is to say, absent a randomized control trial,

01:13:20.360 | we will never know the answer.

01:13:22.080 | - Has there been a randomized control trial in Metformin?

01:13:25.280 | - Not when it comes to a hard outcome.

01:13:26.640 | Now, there has been in the ITP.

01:13:28.840 | So the Interventions Testing Program,

01:13:31.760 | which is kind of the gold standard for animal studies,

01:13:36.120 | which is run out of three labs.

01:13:37.960 | So it's an NIH funded program that's run out of three labs.

01:13:41.300 | They basically test molecules for zero protection.

01:13:46.300 | The ITP was the first study

01:13:48.040 | that really put rapamycin on the map in 2009.

01:13:50.360 | That was the study that's fortuitously demonstrated

01:13:53.840 | that even when rapamycin was given very, very late in life,

01:13:56.960 | it was given to 60 month old mice.

01:13:59.280 | It still afforded them a 15% lifespan extension.

01:14:03.800 | - Has a similar study been done in humans?

01:14:06.020 | I mean, it's hard to,

01:14:07.160 | now we can't really control with rapamycin.

01:14:09.040 | - No, but when the ITP studied Metformin,

01:14:12.760 | it did not succeed.

01:14:14.400 | So there have not been that many drugs

01:14:17.960 | that have worked in the ITP.

01:14:19.980 | The ITP is very rigorous, right?

01:14:22.020 | It doesn't use an inbred strain of mice.

01:14:24.540 | It is done concurrently in three labs

01:14:27.300 | with very large sample sizing.

01:14:29.360 | And so when something works in the ITP, it's pretty exciting.

01:14:32.980 | Rapamycin has been studied several times.

01:14:35.100 | It's always worked.

01:14:36.940 | Another one we should talk about at a subsequent time

01:14:39.620 | is 17 alpha estradiol.

01:14:42.100 | This continues to work in male mice

01:14:44.760 | and it produces comparable effects to rapamycin.

01:14:46.940 | - Estrogen.

01:14:47.780 | - Doesn't work in female rice.

01:14:49.600 | But this is alpha, not beta.

01:14:51.840 | This is 17 alpha estradiol, not beta estradiol,

01:14:54.320 | which is the estradiol that we all,

01:14:56.200 | that is bioavailable in all of us.

01:14:57.880 | - And just as a brief aside,

01:15:00.520 | I think you and I basically agree

01:15:03.420 | that unless it's a problem,

01:15:06.340 | males, we're talking post-puberty,

01:15:10.400 | should try and have their estrogen as high as possible

01:15:13.560 | without having negative symptomology

01:15:15.200 | because of the importance of estrogen for libido,

01:15:17.340 | for brain function, tissue health, bone health.

01:15:20.080 | This idea of crushing estrogen

01:15:21.980 | and raising testosterone is just silly, right?

01:15:25.080 | Let's just leave raising testosterone out of it.

01:15:27.760 | But many of the approaches to raising testosterone

01:15:30.880 | that are pharmacologic in nature also raise estrogen.

01:15:33.400 | A lot of people try and push down on estrogen.

01:15:36.260 | And that is just, again,

01:15:38.120 | unless people are getting hyperestrogenic effects,

01:15:40.600 | like gynecomastia or other issues,

01:15:42.600 | is the exact wrong direction to go.

01:15:45.160 | You want estrogen high.

01:15:46.920 | - Estrogen is a very important hormone for men and women.

01:15:50.120 | - Yeah, that's good.

01:15:51.960 | Canagaflozin, an SGLT2 inhibitor, also very successful

01:15:55.160 | in the ITP.

01:15:56.220 | But again, interestingly, metformin not.

01:15:59.240 | So metformin has failed in the ITP.

01:16:01.920 | - So you no longer take metformin?

01:16:03.560 | - I stopped five years ago.

01:16:04.840 | - I mean, you're not a diabetic,

01:16:05.760 | so presumably you were taking it to buffer blood glucose

01:16:09.720 | in order to potentially live longer.

01:16:11.840 | - Yes, exactly.

01:16:12.740 | And the reason I stopped,

01:16:14.080 | and this will be the last thing before we move on.

01:16:15.720 | - Well, 'cause you couldn't go to the Dairy Queen

01:16:17.280 | at the buffet event.

01:16:18.400 | - No, finally the nausea went away after a few weeks

01:16:20.880 | or a month maybe.

01:16:21.840 | But once I got really into lactate testing,

01:16:26.680 | I noticed how high my lactate was at rest.

01:16:31.400 | So a resting fasted lactate should be,

01:16:33.980 | in a healthy person, should be below one,

01:16:36.060 | like somewhere between 0.3, 0.6 millimole.

01:16:39.600 | And only when you start to exercise should lactate go up.

01:16:42.280 | And in 2018 was when I started blood testing

01:16:46.400 | for my zone two.

01:16:47.440 | So previously, when I was doing zone two testing,

01:16:49.720 | I was just going off my power meter and heart rate.

01:16:52.040 | But this is after I'd met Inigo San Milan

01:16:54.840 | and I started wanting to use the lactate threshold

01:16:58.640 | of two millimole as my determinant

01:17:01.840 | of where to put my wattage on the bike.

01:17:04.360 | And I'm like doing finger pricks before I start

01:17:07.360 | and I'm like 1.6 millimole.

01:17:08.920 | And I'm like, "What the hell is going on?"

01:17:09.860 | I can't be 1.6--

01:17:10.700 | - This is if you ran a flight of stairs

01:17:12.900 | up the back of the Empire State Building.

01:17:14.540 | - Well, no, that would put me a lot higher, right?

01:17:17.200 | And when I-- - I was being generous to your fitness.

01:17:19.980 | - No, but that's when I started doing a little digging

01:17:22.600 | and realized, oh, you know what?

01:17:24.720 | This totally makes sense.

01:17:26.120 | If you have a weak mitochondrial toxin,

01:17:29.660 | what are you gonna do?

01:17:30.960 | You're gonna shunt more glucose into pyruvate

01:17:34.440 | and more pyruvate into lactate.

01:17:36.040 | I'm anaerobic at a baseline. - Yeah, you need

01:17:38.080 | an alternative fuel source.

01:17:39.400 | - That's right.

01:17:40.240 | And then my zone two numbers just seemed off.

01:17:43.440 | My lactate seemed-- - Could you feel it?

01:17:45.060 | Sorry to interrupt. - No.

01:17:45.900 | - Could you feel it in your body?

01:17:46.800 | 'Cause maybe now I'll just briefly describe.

01:17:48.680 | I took berberine.

01:17:49.680 | During the period of maybe,

01:17:53.920 | somewhere in the 2012 to 2015 stretch,

01:17:56.640 | I don't recall exactly.

01:17:57.480 | - And what were you taking it for?

01:17:58.320 | - Well, I'll tell you.

01:17:59.140 | So I was and I still am a big fan

01:18:01.760 | of Tim Ferriss's slow carbohydrate diet,

01:18:04.940 | because I like to eat meat and vegetables and starches.

01:18:07.400 | I'm an omnivore.

01:18:08.280 | And I found that it worked very quickly,

01:18:12.380 | got me very lean.

01:18:13.640 | I could exercise, I could think, I could sleep.

01:18:16.760 | You know, a lot of my rationale

01:18:19.540 | for following one eating regimen or another,

01:18:21.840 | what I eat is to enjoy myself, but also have mental energy.

01:18:24.880 | I mean, 'cause if I can't sleep at night,

01:18:26.060 | I'm not going to replenish.

01:18:27.480 | If I don't replenish, I'm going to feel like garbage.

01:18:29.000 | I don't care how lean I am or what, you know.

01:18:31.800 | So I found the slow carb diet to be,

01:18:33.900 | which was in the four hour body,

01:18:35.040 | to be a very good plan for me.

01:18:37.120 | It was pretty easy.

01:18:38.040 | You drop some things like bread, et cetera.

01:18:39.800 | You don't drink calories,

01:18:41.440 | except after a resistance training session, et cetera.

01:18:44.000 | But one day a week, you have this so-called cheat day.

01:18:47.680 | And on the cheat day, anything goes.

01:18:49.760 | And so I would eat, you know, eight croissants,

01:18:52.040 | and then I'd alternate to sweet stuff,

01:18:53.480 | and then I'd go to a piece.

01:18:54.320 | And by the end of the day,

01:18:55.140 | you don't want to look at an item of food at all.

01:18:57.120 | So the only modification I made to the slow carb diet

01:19:00.340 | four hour body thing was the day after the cheat day,

01:19:03.280 | I wouldn't eat, I would just fast.

01:19:05.120 | And I had no problem doing that

01:19:06.380 | because it was just basically,

01:19:08.040 | well, since you said, what was it, anal?

01:19:11.680 | - Anal seepage.

01:19:12.520 | - Anal seepage, I did not have that.

01:19:13.840 | But since you said that, I won't up the ante here,

01:19:16.940 | but I'll at least match your anal seepage comment by saying,

01:19:19.620 | I had, let's just call it profound gastric distress

01:19:22.840 | after eating like that the next day.

01:19:24.120 | So the last thing you want to do is eat any food,

01:19:25.400 | I would just hydrate,

01:19:26.400 | and oftentimes to try and get some exercise.

01:19:28.560 | And what I read was that berberine,

01:19:32.920 | poor man's metformin, could buffer blood glucose,

01:19:35.880 | and in some ways make me feel less sick

01:19:38.780 | when ingesting all these calories,

01:19:40.680 | and in many cases, spiking my blood sugar and insulin

01:19:45.180 | because you're having ice cream and et cetera.

01:19:48.120 | And indeed it worked.

01:19:49.520 | So if I took berberine,

01:19:50.680 | and I don't recall the milligram count,

01:19:52.780 | and then I ate 12 donuts, I felt fine.

01:19:56.600 | It was as if I had eaten one donut.

01:19:58.560 | I felt sort of okay in my body

01:20:00.380 | and I felt much, much better.

01:20:02.280 | Now, presumably 'cause it's buffering

01:20:04.520 | the spikes in blood sugar,

01:20:05.360 | I wasn't crashing in the afternoon nap and that whole thing.

01:20:08.640 | - And do you remember how much you were taking?

01:20:10.080 | - I think it was a couple hundred milligrams.

01:20:11.480 | Does that sound about right?

01:20:13.360 | It was a bright yellow capsule.

01:20:15.600 | I forget the source.

01:20:16.800 | But in any case, one thing I noticed was

01:20:20.160 | that if I took berberine and I did not ingest

01:20:23.240 | a profound number of carbohydrates very soon afterwards,

01:20:27.040 | I got brutal headaches.

01:20:28.420 | I think I was hypoglycemic.

01:20:29.840 | I didn't measure it, but I just felt I had headaches.

01:20:32.360 | I didn't feel good.

01:20:33.200 | And then I would eat a pizza or two and feel fine.

01:20:37.000 | And so I realized that berberine was putting me

01:20:38.960 | on this lower blood sugar state.

01:20:41.280 | That was the logic anyway.

01:20:42.840 | And it allowed me to eat these cheat foods.

01:20:46.720 | But when I cycled off of the four out,

01:20:48.960 | 'cause I don't follow the slow carb diet anymore,

01:20:51.280 | although I might again at some point.

01:20:53.100 | When I stopped doing those cheat days,

01:20:55.680 | I didn't have any reason to take the berberine

01:20:57.920 | and I feared that I wasn't ingesting enough carbohydrates

01:21:00.900 | in order to really justify trying to buffer my blood glucose.

01:21:03.320 | Also, my blood glucose tends to be fairly low.

01:21:05.320 | - Did you ever try Acarbos?

01:21:07.160 | - No, what is that?

01:21:08.460 | - So Acarbos is-

01:21:09.300 | - Another glucose disposal.

01:21:10.640 | - Yeah, it's actually a drug that,

01:21:12.100 | but it works more in the gut

01:21:13.620 | and it just prevents glucose absorption.

01:21:16.380 | Acarbos is another one of those drugs

01:21:18.600 | that actually found a survival benefit in the ITP.

01:21:21.820 | And it was a very interesting finding

01:21:24.640 | because the thesis for testing it,

01:21:27.740 | the ITP is a very clever system.

01:21:29.420 | Anybody can nominate a candidate to be tested.

01:21:32.160 | Then the panel over there reviews it and they decide,

01:21:34.380 | yep, this is interesting.

01:21:35.220 | We'll go ahead and study it.

01:21:36.040 | When I think David Allison nominated Acarbos to be studied,

01:21:41.040 | the rationale was it would be a caloric restriction mimetic

01:21:45.060 | because you would literally just fail to absorb,

01:21:47.800 | I don't know, make up some number, right?

01:21:49.660 | 15 to 20% of your carbohydrates would not be absorbed

01:21:52.800 | and therefore you would,

01:21:53.960 | the mice would effectively be calorically restricted.

01:21:56.120 | - They would just pass them out.

01:21:57.040 | - That's right.

01:21:57.960 | And what happened was really interesting.

01:22:00.640 | One, the mice lived longer on Acarbos,

01:22:03.400 | but two, they didn't weigh any less.

01:22:05.880 | So they lived longer, but not through calorie restriction.

01:22:10.320 | - That's interesting.

01:22:11.160 | - Yes.

01:22:11.980 | And the speculation is they lived longer

01:22:14.160 | because they had lower glucose and lower insulin.

01:22:17.200 | - And I don't want to send this down some rabbit holes here,

01:22:19.720 | but there are all sorts of interesting ideas about,

01:22:22.840 | for instance, that some forms of dementia

01:22:26.200 | might be so-called type three diabetes,

01:22:28.080 | the diabetes of the brain.

01:22:29.240 | And so things like berberine metformin,

01:22:31.200 | lowering blood glucose, ketogenic diets, et cetera,

01:22:33.740 | might be beneficial there.

01:22:34.880 | I mean, there's a lot to explore here

01:22:36.600 | and I know you've explored a lot of that on your podcast.

01:22:38.600 | I've done far less of that, but well,

01:22:40.760 | at least it seems that we know

01:22:42.420 | the following things for sure.

01:22:43.480 | One, you don't want insulin too high, nor too low.

01:22:47.520 | You don't want blood glucose too high, nor too low.

01:22:50.480 | If the buffering systems for that are disrupted,

01:22:53.560 | clearly exercise, meaning regular exercise,

01:22:56.400 | is the best way to keep that system in check.

01:22:59.080 | But in the absence of that tool,

01:23:02.520 | or I would say in addition to that tool,

01:23:04.580 | is there any glucose disposal agent,

01:23:07.480 | 'cause that's what we're talking about here,

01:23:08.840 | metformin, berberine, A-CarboCit, et cetera,

01:23:11.800 | that you take on a regular basis

01:23:14.040 | because you have that much confidence in it?

01:23:16.520 | - The only one that I take is an SGLT2 inhibitor.

01:23:19.620 | So this is a class of drug that is used

01:23:25.020 | by people with type two diabetes, which I don't have,

01:23:27.800 | but because of my faith in the mechanistic studies

01:23:31.720 | of this drug, coupled with its results in the ITP,

01:23:34.660 | coupled with the human trial results

01:23:36.940 | that show profound benefit in non-diabetics

01:23:39.980 | taking it even for heart failure,

01:23:41.720 | I think there's something very special about that drug.

01:23:43.700 | Actually, that was another paper

01:23:45.060 | I was thinking about presenting this time.

01:23:46.380 | Maybe we'll do that the next time.

01:23:47.700 | - But do you believe in caloric restriction

01:23:51.720 | as a way to extend life, or are you more of the,

01:23:55.960 | do the right behaviors, and that's covered

01:23:58.540 | in your book, "Outlive" and elsewhere on your podcast.

01:24:02.400 | And buffer blood glucose, do you still,

01:24:06.660 | obviously you believe in buffering blood glucose

01:24:09.000 | in addition to just doing all the right behaviors.

01:24:10.880 | - Yeah, I think you can uncouple a little bit

01:24:12.880 | the buffering of blood glucose from the caloric deficit.

01:24:15.660 | So I think you can be in a reasonable energy balance

01:24:19.160 | and buffer glucose with good sleep hygiene,

01:24:22.040 | lots of exercise, and just thoughtful eating

01:24:26.000 | without having to go into a calorie deficit.

01:24:28.160 | So it's not entirely clear if profound caloric restriction

01:24:32.840 | would offer a survival advantage to humans

01:24:34.840 | even if it were tolerable to most, which it's not.

01:24:37.560 | So for most people, it's just kind of off the table.

01:24:39.980 | Like if I said, Andrew, you need to eat 30% fewer calories

01:24:43.180 | for the rest of your life.

01:24:44.560 | - I'll live 30% fewer years, thank you.

01:24:46.240 | - Yeah, there's just not many people

01:24:48.080 | who are willing to sign up for that.

01:24:49.260 | So it's kind of a moot point.

01:24:50.720 | But the question is, do you need to be fasting all the time?

01:24:56.640 | Do you need to be doing all of these other things?

01:24:58.800 | And the answer appears to be outside of using them as tools

01:25:03.400 | to manage energy balance, it's not clear, right?

01:25:06.060 | And energy balance probably plays a greater role

01:25:10.600 | in glucose homeostasis from a nutrition standpoint

01:25:15.600 | than the individual constituents of the meal.

01:25:18.480 | Now that's not entirely true.

01:25:20.560 | Like I can imagine a scenario where a person

01:25:22.920 | could be in a negative energy balance

01:25:24.920 | eating Twix bars all day and drinking big gulps.

01:25:29.080 | But I also don't think that's a very sustainable thing to do

01:25:31.540 | because if by definition, I'm gonna put you

01:25:33.580 | in negative energy balance consuming that much crap,

01:25:37.040 | I'm gonna destroy you.

01:25:38.460 | Like you're gonna feel so miserable,

01:25:40.580 | you're gonna be starving, right?

01:25:42.800 | You're not gonna be satiated eating pure garbage

01:25:46.080 | and being in caloric deficit.

01:25:48.540 | You're gonna end up having to go into caloric excess.

01:25:51.240 | So that's why it's interesting thought experiment.

01:25:53.600 | I don't think it's a very practical experiment.

01:25:55.320 | For a person to be generally satiated

01:25:56.960 | and in energy balance, they're probably eating

01:25:59.060 | about the right stuff.

01:26:00.760 | But I don't think that the specific macros matter

01:26:03.600 | as much as I used to think.

01:26:05.560 | - I'm a believer in getting most of my nutrients

01:26:08.640 | from unprocessed or minimally processed sources

01:26:12.200 | simply because it allows me to eat foods I like

01:26:17.200 | and more of them.

01:26:19.160 | And I just love to eat.

01:26:20.200 | I so physically enjoy the sensation of chewing

01:26:23.400 | that I'll just eat cucumber slices for fun, right?

01:26:28.320 | I mean, that's not my only form of fun fortunately.

01:26:30.680 | This is an amazing paper for the simple reason

01:26:37.700 | that it provides a wonderful tutorial

01:26:41.240 | of the benefits and drawbacks of this type of work.

01:26:45.720 | And I think it's also wonderful because we hear a lot

01:26:48.200 | about metformin, rapamycin,

01:26:50.560 | and these anti-aging approaches,

01:26:53.940 | but I was not aware that there was any study

01:26:57.120 | of such a large population of people.

01:26:58.860 | So it's pretty interesting.

01:27:00.360 | - Yeah, so I think it remains to be seen.

01:27:02.560 | And my patients often ask me,

01:27:04.260 | "Hey, should I be on metformin?"

01:27:05.460 | And I give them a much, much, much, much shorter version

01:27:08.220 | of what we just talked about.

01:27:09.720 | And I say, "Look, if the TAME study,"

01:27:12.340 | which should answer this question more definitively, right?

01:27:14.920 | This is taking a group of non-diabetics

01:27:17.600 | and randomizing them to placebo versus metformin

01:27:20.680 | and studying for specific disease outcomes,

01:27:24.060 | if the TAME study ends up demonstrating

01:27:26.860 | that there is a zero protective benefit of metformin,

01:27:30.640 | I'll reconsider everything, right?

01:27:32.880 | So I think that's, we just have to,

01:27:34.920 | I think, all walk around with an appropriate degree

01:27:37.280 | of humility around what we know and what we don't know.

01:27:39.320 | But I would say right now, the epidemiology,

01:27:42.160 | the animal data, my own personal experience

01:27:45.040 | with its impact on my lactate production

01:27:47.360 | and exercise performance,

01:27:49.520 | there's a whole other rabbit hole

01:27:50.480 | we could go down another time,

01:27:51.400 | which is the impact on hypertrophy and strength,

01:27:53.540 | which appears to be attenuated as well by metformin.

01:27:56.140 | I still prescribe it to patients all the time

01:28:01.320 | if they're insulin resistant, for sure.

01:28:02.720 | It's still a valuable drug,

01:28:04.280 | but I don't think of it as a great tool

01:28:05.860 | for the person who's insulin sensitive and exercising a lot.

01:28:09.180 | - I can't help but ask this question.

01:28:12.080 | Do you think there's any longevity benefit

01:28:16.680 | to short periods of caloric restriction?

01:28:20.280 | So for instance, I decide to,

01:28:24.340 | by the way, I haven't done this,

01:28:25.600 | but let's say I were to decide to fast

01:28:29.280 | and do a one meal a day type thing

01:28:31.400 | where I'm going to be in a slight caloric deficit,

01:28:34.240 | 500 to 1,000 calories for a couple of days

01:28:37.000 | and then go back to eating the way that I ate before,

01:28:40.640 | that short caloric restriction/fast.

01:28:44.000 | Is there any benefit to it in terms of cellular health?

01:28:46.120 | Can you sort of reset the system?

01:28:48.620 | Is there any idea that the changes,

01:28:51.200 | the clearing of senescent cells that we hear about,

01:28:53.240 | autophagy, that in the short term,

01:28:55.920 | you can glean a lot of benefits

01:28:57.640 | and then go back to your regular pattern of eating

01:29:00.140 | and then periodically, once every couple of weeks

01:29:03.320 | or once a month, just fast for a day or two.

01:29:06.160 | Is there any benefit to that

01:29:07.360 | that's purely in the domain of longevity?

01:29:11.440 | Because there's all discipline function there,

01:29:14.800 | there's a flexibility function,

01:29:16.120 | there's probably an insulin sensitivity function,

01:29:17.880 | but is there any evidence that it can help us live longer?

01:29:21.040 | - I think the short answer is no, for two reasons.

01:29:25.040 | One, I don't think that that duration would be sufficient

01:29:27.640 | if one is going to take that approach,

01:29:29.440 | but two, even if you went with something longer,

01:29:32.920 | like what I used to do, right?

01:29:33.860 | I used to do seven days of water only per quarter,

01:29:37.160 | three days per month.

01:29:38.800 | So I was, basically always like it would be three day fast,

01:29:42.280 | three day fast, seven day fast.

01:29:43.860 | Just imagine doing that all year, rotating, rotating,

01:29:46.740 | rotating, for many years I did that.

01:29:48.860 | Now I certainly believed, and to this day I would say

01:29:51.280 | I have no idea if that provided a benefit,

01:29:55.000 | but my thesis was the downside of this

01:29:58.440 | is relatively circumscribed,

01:29:59.840 | which is profound misery for a few days

01:30:03.240 | and what I didn't appreciate at the time,

01:30:05.920 | which I obviously now look back at and realize

01:30:08.100 | is muscle mass loss.

01:30:09.920 | It's very difficult to gain back the muscle cumulatively,

01:30:13.400 | after all of that loss,

01:30:15.280 | but my thought was exactly as you said,

01:30:17.200 | like there's got to be a resetting of the system here.

01:30:19.660 | This must be sufficiently long enough

01:30:22.000 | to trigger all of those systems,

01:30:24.120 | but you're getting at a bigger problem with neuroscience,

01:30:29.120 | which I'm really hoping the epigenetic field

01:30:32.720 | comes to the rescue on.

01:30:34.040 | It has not come close to it to date,

01:30:36.140 | which is we don't have biomarkers

01:30:38.660 | around true metrics of aging.

01:30:41.700 | Everything we have to date stinks.

01:30:44.120 | So we're really good at using molecules or interventions

01:30:49.120 | for which we have biomarkers, right?

01:30:51.780 | Like when you lift weights,

01:30:54.460 | you can look at how much weight you're lifting,

01:30:57.120 | you can look at your DEXA scan

01:30:58.460 | and see how much muscle mass you're generating,

01:31:00.140 | like those are biomarkers.

01:31:01.760 | Those are giving you outputs that say my input is good

01:31:05.620 | or my input needs to be modified.

01:31:08.060 | When you take a sleep supplement,

01:31:09.500 | you can look at your eight sleep and go,

01:31:11.900 | oh, my sleep is getting better.

01:31:13.620 | Like there's a biomarker.

01:31:15.120 | When you take metformin, when you take rapamycin,

01:31:20.100 | when you fast, we don't have a biomarker

01:31:23.220 | that gives us any insight into whether or not

01:31:26.280 | we're moving in the right direction.

01:31:27.500 | And if we are, are we taking enough?

01:31:29.300 | Just don't know.

01:31:31.580 | So I often get asked,

01:31:33.460 | like what's the single most important topic

01:31:36.140 | you would want to see more research dollars put to

01:31:39.300 | in terms of this space?

01:31:41.060 | And it's unquestionably this, as unsexy as it is,

01:31:44.760 | like who cares about biomarkers?

01:31:46.900 | But like without them,

01:31:48.340 | I don't think we're going to get great answers

01:31:50.100 | 'cause you can't do most of the experiments

01:31:53.080 | you and I would dream up.

01:31:54.320 | - Got it.

01:31:56.960 | Well, I'm grateful that you're sitting across the table

01:31:59.900 | for me telling me all this and that everyone can hear this.

01:32:03.820 | But again, we will put a link to the papers, plural,

01:32:07.940 | that Peter just described.

01:32:10.120 | And for those of you that are listening and not watching,

01:32:12.800 | hopefully you were able to track

01:32:13.940 | the general themes and takeaways.

01:32:16.380 | And it is fun to go to these papers.

01:32:18.180 | You see these big stacks of numbers

01:32:20.100 | and it can be a little bit overwhelming.

01:32:21.740 | But my additional suggestion on parsing papers

01:32:26.000 | is you notice that Peter said that he spent,

01:32:28.860 | you know, he's read it several times.

01:32:31.380 | Unlike a newspaper article or a Instagram post,

01:32:35.620 | with a paper you're not necessarily going to get it

01:32:38.980 | the first time.

01:32:39.820 | You certainly won't get everything.

01:32:40.820 | So I think spending some time with papers for me

01:32:43.480 | means reading it and then reading it again

01:32:45.140 | a little bit later.

01:32:46.240 | Or, you know, one figure at a time.

01:32:47.780 | - Yeah, I was just about to say,

01:32:49.620 | 'cause I kind of have a way that I do it,

01:32:50.980 | but I'm curious as to how you do it.

01:32:52.380 | Like if you're encountering a paper for the first time,

01:32:55.240 | do you have an order in which you like to go through?

01:32:58.180 | Do you read it sequentially

01:32:59.380 | or do you look at the figures first?

01:33:01.120 | I mean, how do you go through it?

01:33:02.180 | - Yeah, unless it's an area that I know very, very well

01:33:04.780 | where I can, you know, skip to some things

01:33:07.980 | before reading it the whole way through.

01:33:09.980 | My process is always the same.

01:33:12.060 | And actually this is fun because I used to teach a class

01:33:15.280 | when I was a professor at UC San Diego

01:33:17.620 | called Neural Circuits in Health and Disease.

01:33:19.400 | And it was an evening course that grew very quickly

01:33:21.480 | from 50 students to 400 plus students.

01:33:24.440 | And we would do exactly this.

01:33:26.060 | We would parse papers.

01:33:27.500 | And I had everyone ask what I called the four questions.

01:33:32.500 | And it wasn't exactly four questions,

01:33:34.560 | but I have a little three by five card next to me

01:33:37.740 | or a piece of a main half by 11 paper typically.

01:33:40.900 | And when I sit down with the paper,

01:33:42.240 | I want to figure out what is the question they're asking?

01:33:45.540 | What's the general question?

01:33:47.000 | What's the specific question?

01:33:48.240 | And I write down the question.

01:33:49.960 | Then what was the approach?

01:33:51.940 | You know, how did they test that question?

01:33:53.260 | And sometimes that can get a bit detailed.

01:33:54.860 | You can get into immunohistochemistry

01:33:56.540 | and they did a PCR for this.

01:33:58.820 | It's not so important for most people

01:34:01.100 | that they understand every method,

01:34:02.960 | but it is worthwhile that if you encounter a method

01:34:06.800 | like PCR or, you know, chromatography or fMRI

01:34:11.800 | that you at least look up on the internet

01:34:13.800 | what its purpose is, okay?

01:34:15.340 | That will help a lot.

01:34:16.660 | And then it was what they found.

01:34:18.140 | And there you can usually figure out

01:34:20.820 | what they believe they found anyway

01:34:22.180 | by reading the figure headers, right?

01:34:24.940 | What are, you know, figure one, here's the header.

01:34:27.420 | Typically, if it's an experimental paper,

01:34:29.460 | it will tell you what they want you to think they found.

01:34:32.340 | And then I tend to want to know the conclusion of the study.

01:34:35.880 | And then this is really the key one.

01:34:37.500 | And this is the one that would really distinguish

01:34:40.740 | the high-performing students from the others.

01:34:42.940 | You have to go back at the end

01:34:44.980 | and ask whether or not the conclusions,

01:34:46.820 | the major conclusions drawn in the paper

01:34:48.900 | are really substantiated by what they found

01:34:51.280 | and what they did.

01:34:52.120 | And that involves some thinking.

01:34:53.440 | It involves really, you know,

01:34:54.500 | spending some time thinking about what they identified.

01:34:57.140 | Now, this isn't something that anyone can do

01:34:58.600 | straight off the bat.

01:34:59.440 | It's a skill that you develop over time

01:35:00.820 | and different papers require different formats.

01:35:03.040 | But those four questions really form the cornerstone

01:35:05.700 | of teaching undergraduates

01:35:06.820 | and I think graduate students as well

01:35:08.000 | of how to read a paper.

01:35:10.200 | And again, it's something that can be cultivated.

01:35:14.580 | And it's still how I approach papers.

01:35:18.440 | So what I do typically is I'll read title, abstract.

01:35:21.380 | I usually then will skip to the figures

01:35:24.620 | and see how much of it I can digest

01:35:26.480 | without reading the text

01:35:27.980 | and then go back and read the text.

01:35:30.020 | But in fairness, journals, great journals,

01:35:33.060 | like science, like nature,

01:35:34.340 | oftentimes will pack so much information,

01:35:36.900 | the cell press journals too, into each figure.

01:35:39.620 | And it's coded with no definition of the acronyms

01:35:42.180 | that almost always I'm into the introduction and results

01:35:45.380 | within a couple of minutes,

01:35:46.260 | wondering what the hell this acronym is

01:35:47.920 | or that acronym is.

01:35:48.940 | And it's just, yeah, it's just wild

01:35:51.860 | how much nomenclature there really is.

01:35:55.860 | I can't remember, was it you

01:35:56.900 | or was it our friend Paul Conte when he was here

01:36:00.140 | who said that, oh no, I'm sorry, it was neither.

01:36:02.640 | It was chair of ophthalmology at Stanford,

01:36:05.220 | Dr. Jeffrey Goldberg, who was a guest on a podcast recently,

01:36:07.800 | who off camera, I think it was,

01:36:09.980 | told us that if you look at the total number of words

01:36:14.140 | and terms that a physician leaving medical school

01:36:18.180 | owns in their mind and their vocabulary,

01:36:20.660 | it's the equivalent of like two additional full languages

01:36:24.020 | of fluency beyond their native language.

01:36:26.740 | So you're trilingual at least,

01:36:28.820 | and I don't know, do you speak a language other than English?

01:36:30.820 | - Poorly. - Okay, so you're

01:36:33.020 | at least trilingual and probably more.

01:36:35.700 | So no one is expected to be able to parse these papers

01:36:39.480 | the first time through without substantial training.

01:36:42.580 | - Yeah, no, I think that's a great format.

01:36:45.780 | And you're absolutely right.

01:36:46.900 | I have a different way that I do it

01:36:48.960 | when I'm familiar with the subject matter

01:36:50.620 | versus when I'm not. - Yeah, how do you do it?

01:36:52.220 | - Well, again, if I'm reading papers

01:36:54.080 | that are something that I know really well,

01:36:56.420 | I can basically glean everything

01:36:58.500 | I need to know from the figures.

01:37:00.120 | And then sometimes I'll just do a quick skim on methods.

01:37:04.300 | But I don't need to read the discussion,

01:37:05.460 | I don't need to read the intro,

01:37:06.460 | I don't need to read anything else.

01:37:07.940 | If it's something that I know less about,

01:37:10.660 | then I usually do exactly what you say.

01:37:13.580 | I try to start with the figures.

01:37:16.260 | I usually end up generating more questions,

01:37:18.700 | like what do you mean, what is this, how did they do that?

01:37:23.180 | And then I gotta go back and read methods typically.

01:37:26.020 | And one of the other things that's probably worth mentioning

01:37:28.220 | is a lot of papers these days have supplemental information

01:37:31.020 | that are not attached to the paper.

01:37:32.740 | So you're amazed at how much stuff gets put

01:37:36.100 | in the supplemental section.

01:37:37.140 | And the reason for that, of course,

01:37:38.620 | is that the journals are very specific

01:37:41.260 | on the format and length of a paper.

01:37:43.520 | So a lot of the times when you're submitting something,

01:37:46.780 | like if you wanna put any additional information in there,

01:37:49.380 | it can't go in the main article,

01:37:50.540 | it has to go in the supplemental figure.

01:37:51.820 | So even for this paper,

01:37:53.200 | there were a couple of the numbers I spouted off

01:37:55.500 | I had to pull out of the supplemental paper.

01:37:57.400 | For example, when they did the sensitivity analysis

01:38:01.020 | on the censoring versus non-censoring,

01:38:05.260 | that was in the supplemental figure.

01:38:07.140 | That was actually not even in the paper we presented.

01:38:10.100 | - Well, should we pivot to this other paper?

01:38:13.420 | - Yeah.

01:38:14.240 | - It's a very different sort of paper.

01:38:15.420 | It's an experimental paper where there's a manipulation.

01:38:18.220 | I must say, I love, love, love this paper.

01:38:22.060 | And I don't often say that about papers.

01:38:24.200 | I'm so excited about this paper for so many reasons,

01:38:28.040 | but I wanna give a couple of caveats up front.

01:38:30.460 | First of all, the paper is not published yet.

01:38:33.960 | The only reason I was able to get this paper

01:38:35.680 | is because it's on bioRxiv.

01:38:37.800 | There's a new trend over the last,

01:38:40.320 | I would say five, six years of people posting the papers

01:38:42.940 | that they've submitted to journals for peer review online

01:38:46.340 | so that people can look at them prior to those papers

01:38:49.040 | being peer reviewed.

01:38:50.220 | So there is a strong possibility

01:38:51.940 | that the final version of this paper,

01:38:53.540 | which again, we will provide a link to,

01:38:55.420 | is going to look different,

01:38:56.560 | maybe even quite a bit different

01:38:58.280 | than the one that we're going to discuss.

01:39:00.440 | Nonetheless, there are a couple of things

01:39:02.140 | that make me confident in the data

01:39:03.900 | that we're about to talk about.

01:39:04.820 | First of all, the group that published this paper

01:39:09.160 | is really playing in their wheelhouse.

01:39:10.780 | This is what they do.

01:39:11.880 | And they publish a lot of really nice papers in this area.

01:39:15.080 | I'm going to mispronounce her first name,

01:39:17.940 | but I think it's Chiaosi Gu,

01:39:21.520 | who's at the Econ School of Medicine in Mount Sinai,

01:39:24.500 | runs a laboratory there studying addiction in humans.

01:39:28.300 | And the first author of the paper is Ofer Pearl.

01:39:33.300 | This paper is wild.

01:39:34.800 | And I'll just give you a couple of the takeaways first

01:39:37.120 | as a bit of a hook to hopefully entice people

01:39:39.560 | into listening further,

01:39:41.260 | 'cause this is an important paper.

01:39:43.500 | This paper basically addresses how our beliefs

01:39:46.900 | about the drugs we take

01:39:49.580 | impacts how they affect us at a real level,

01:39:54.580 | not just at a subjective level, but at a biological level.

01:39:58.060 | So just to back up a little bit,

01:39:59.900 | a former guest on this podcast, Dr. Ali Crum,

01:40:02.540 | whose name is actually Alia Crum,

01:40:05.500 | but she goes by Ali Crum, talked about belief effects.

01:40:08.540 | Belief effects are different than placebo effects.

01:40:11.920 | Placebo effects are really just category effects.

01:40:14.940 | It's, okay, I'm going to give you this pill, Peter,

01:40:18.500 | and I'm going to tell you that this pill

01:40:20.600 | is molecule X5952,

01:40:24.320 | and that it's going to make your memory better.

01:40:26.100 | And then I give you a memory test, right?

01:40:28.260 | And your group performs better

01:40:29.400 | than the people in the control group

01:40:30.820 | who I give a pill to and I say, this is just a placebo.

01:40:34.780 | Or there are other variants on this

01:40:36.780 | where people will get a drug and you tell them it's placebo.

01:40:39.980 | They'll get a placebo, you tell them it's drug.

01:40:42.040 | It's a binary thing.

01:40:44.300 | It's an on or an off thing.

01:40:45.460 | You're either in the drug group or the placebo group,

01:40:47.500 | and you're either told that you're getting drug or placebo.

01:40:49.740 | And we know that placebo effects exist.

01:40:52.340 | In fact, one of the crueler ones,

01:40:53.960 | I was never the subject of this,

01:40:55.060 | but there was kind of lore in high school

01:40:56.620 | that kids would do this mean thing.

01:40:58.260 | It's a form of bullying.

01:40:59.620 | I really don't like it,

01:41:00.440 | where they'd get some kid at a party

01:41:02.900 | to drink alcohol-free beer,

01:41:06.020 | and then that kid would start acting drunk,

01:41:07.620 | and then they'd go, gotcha,

01:41:08.820 | it doesn't even have alcohol in it.

01:41:10.620 | Now, that's a mean joke,

01:41:12.620 | and just reminds me of some of the horrors of high school.

01:41:16.640 | Maybe that's why I didn't go very often,

01:41:17.860 | which I also don't suggest.

01:41:19.220 | But no, it's a mean joke,

01:41:21.000 | but it speaks to the placebo effect, right?

01:41:22.820 | And there's also a social context effect.

01:41:25.440 | So placebo effects are real.

01:41:27.160 | We know this.

01:41:29.100 | Belief effects are different.

01:41:30.460 | Belief effects are not A or B, placebo or non-placebo.

01:41:34.660 | Belief effects have a lot of knowledge

01:41:37.120 | to enrich one's belief about a certain something

01:41:41.140 | that can shift their psychology and physiology

01:41:44.500 | one way or the other.

01:41:45.360 | And I think the best examples of these, really,

01:41:47.860 | of these belief effects really do come from Allie Crump's lab

01:41:50.340 | in the psychology department at Stanford,

01:41:52.060 | although some of this work she did

01:41:53.040 | prior to getting into Stanford.

01:41:54.080 | For instance, if people are put into a group

01:41:57.340 | where they watch a brief video,

01:41:59.780 | just a few minutes of video

01:42:00.980 | about how stress really limits our performance,

01:42:03.580 | let's say at archery or at mathematics or at music

01:42:06.140 | or at public speaking,

01:42:07.640 | and then you test them in any of those domains

01:42:10.700 | or other domains in a stressful circumstance,

01:42:13.820 | they perform less well, okay?

01:42:16.100 | And we know they perform less well

01:42:18.460 | because we're by virtue of a heightened stress response.

01:42:22.260 | You can measure heart rate.

01:42:23.100 | You can measure stroke volume of the heart.

01:42:24.820 | You can measure peripheral blood flow,

01:42:26.620 | which goes down when people are stressed,

01:42:27.900 | narrowing a vision, et cetera.

01:42:30.100 | You take a different group of people

01:42:32.000 | and randomly assign them to another group

01:42:34.260 | where now they're being told

01:42:35.600 | that stress enhances performance.

01:42:38.260 | It mobilizes resources.

01:42:40.420 | It narrows your vision such that you can perform tasks better

01:42:43.080 | et cetera, et cetera,

01:42:43.940 | and their performance increases above a control group

01:42:46.380 | that receives just useless information

01:42:48.620 | or at least useless as it relates to the task.

01:42:50.560 | So in both cases, by the way,

01:42:52.520 | the groups are being told the truth.

01:42:54.560 | Stress can be depleting or it can enhance performance,

01:42:58.500 | but this is different than placebo

01:42:59.920 | because now it's scaling according to the amount

01:43:02.780 | and the type of information that they're getting.

01:43:04.760 | - And can you give me a sense of magnitude

01:43:07.140 | of benefit or detriment that one could experience

01:43:09.500 | in a situation like the one you just described?

01:43:11.180 | - Yeah, so it's striking.

01:43:12.940 | They're opposite in direction.

01:43:14.220 | So the stress gets us worse, makes you, let's say,

01:43:18.020 | I think that if we were to just put a rough percentage

01:43:20.540 | on this, it would be somewhere between 10 and 30% worse

01:43:23.380 | at performance than the control group.

01:43:24.980 | And stress is enhancing

01:43:26.220 | is approximately equivalent improvement.

01:43:29.040 | So they're in opposite directions.

01:43:30.580 | Even more striking is the studies that Ali's lab did

01:43:35.340 | and others looking at, for instance,

01:43:37.460 | you give people a milkshake,

01:43:38.720 | you tell them it's a high calorie milkshake,

01:43:40.420 | has a lot of nutrients,

01:43:41.420 | and then you measure ghrelin secretion in the blood.

01:43:44.620 | And ghrelin is a marker of hunger that increases

01:43:47.820 | the longer it's been since you've eaten.

01:43:49.060 | And what you notice is that suppresses ghrelin

01:43:51.220 | to a great degree and for a long period of time.

01:43:53.460 | You give another group a shake,

01:43:55.340 | you tell them it's a low calorie shake,

01:43:57.580 | that it's got some nutrients in it,

01:43:58.840 | but that doesn't have much fat, not much sugar, et cetera.

01:44:01.220 | They drink the shake, less ghrelin suppression.

01:44:04.660 | - And it's the same shake.

01:44:05.500 | - And it's the same shake.

01:44:07.380 | And satiety lines up with that also in that study.

01:44:10.580 | And then the third one, which is also pretty striking,

01:44:12.420 | is they took hotel workers,

01:44:13.840 | they gave them a short tutorial or not,

01:44:16.080 | informing them that moving around during the day

01:44:17.920 | and vacuuming and doing all that kind of thing is great.

01:44:19.740 | It helps you lower your BMI,

01:44:21.140 | which is great for your health.

01:44:22.340 | You incentivize them.

01:44:23.580 | And then you let them out into the wild

01:44:25.360 | of their everyday job.

01:44:27.180 | You measure their activity levels.

01:44:28.580 | The two groups don't differ.

01:44:29.760 | They're doing roughly the same task,

01:44:30.980 | leaning down, cleaning out trash cans, et cetera.

01:44:33.500 | Guess what?

01:44:34.340 | The group that was informed

01:44:35.180 | about the health benefits of exercise,

01:44:37.260 | lose 12% more weight compared to the other group.

01:44:41.980 | - And no difference in actual movement?

01:44:44.460 | - Apparently not.

01:44:45.780 | Now, how could that be?

01:44:47.200 | I mean, literally this was sparked by, in Allie's words,

01:44:50.980 | this was sparked by her graduate advisor saying,

01:44:55.580 | "What if all the effects of exercise are placebo?"

01:44:58.220 | Right, which is not what anyone really believes,

01:45:00.620 | but it's just such a...

01:45:02.540 | I love that anecdote that Allie told us

01:45:05.100 | because it just really speaks to how really smart

01:45:07.540 | people think.

01:45:08.500 | They sit back and they go,

01:45:09.340 | "Yeah, exercise obviously has benefits,

01:45:11.000 | but what if a lot of the benefits

01:45:12.220 | are that you tell yourself it's good for you

01:45:13.460 | and the brain can actually activate

01:45:15.060 | these mechanisms in the body?

01:45:17.280 | And why wouldn't that be the case?

01:45:18.420 | 'Cause the nervous system extends through both."

01:45:20.660 | - So interesting. - So interesting.

01:45:22.680 | Okay, so fast forward to this study,

01:45:25.880 | which is really about belief effects, not placebo effects.

01:45:29.660 | And to make a long story short,

01:45:32.700 | we know that nicotine, vaped, smoked, dipped, or snuffed,

01:45:37.420 | or these little zen pouches, or taken in capsule form,

01:45:40.220 | does improve cognitive performance.

01:45:42.020 | I'm not suggesting people run out

01:45:43.380 | and start doing any of those things.

01:45:44.460 | I did a whole episode on nicotine.

01:45:45.800 | The delivery device often will kill you

01:45:47.620 | some other way or is bad for you.

01:45:48.980 | But it causes vasoconstriction,

01:45:50.940 | which is also not good for certain people.

01:45:52.420 | But nicotine is cognitive enhancing, why?

01:45:55.220 | Well, you have a couple of sites in the brain,

01:45:57.380 | namely in the basal forebrain, nucleus basalis,

01:46:01.140 | in the back of the brain structures like locus coeruleus,

01:46:05.340 | but also this, what's called, it's got a funny name,

01:46:07.060 | the pedunculopontine nucleus,

01:46:08.980 | which is this nucleus in the pons,

01:46:12.100 | in the back of the brain, in the brainstem,

01:46:13.880 | that sends those little axon wires into the thalamus.

01:46:16.340 | The thalamus is a gateway for sensory information.

01:46:18.800 | And in the thalamus, the visual information,

01:46:21.260 | the auditory information, it has nicotinic receptors.

01:46:25.180 | And when the pedunculopontine nucleus releases nicotine,

01:46:28.100 | or when you ingest nicotine,

01:46:29.580 | what it does is it increases the signal to noise

01:46:32.600 | of information coming in through your senses.

01:46:35.060 | So the fidelity of the signal that gets up to your cortex,

01:46:38.060 | which is your conscious perception of those senses,

01:46:40.420 | is increased.

01:46:41.260 | - And how much endogenous nicotine do we produce?

01:46:44.000 | - Ooh, well, it's going to be acetylcholine

01:46:46.720 | binding to nicotinic receptors.

01:46:48.260 | - I see, we're not making nicotine.

01:46:49.460 | - We're not making nicotine. - We're just binding.

01:46:50.420 | So this is a nicotinic acetylcholine receptor.

01:46:53.700 | - Right, of which there are at least seven

01:46:55.700 | and probably like 14 subtypes.

01:46:57.620 | But so, right, they're called nicotinic receptors

01:47:01.700 | in an annoying way, in the same way

01:47:03.200 | that cannabinoid receptors are called cannabinoid receptors,

01:47:06.060 | but then everyone thinks,

01:47:06.880 | "Oh, you know, those receptors are there

01:47:09.020 | "because we're supposed to smoke pot,"

01:47:10.300 | or, "Those receptors are there

01:47:11.200 | "'cause we're supposed to ingest nicotine."

01:47:12.500 | No, the drugs that we use to study them--

01:47:14.200 | - The drug is named after the receptor, yeah.

01:47:15.340 | - That's right, exactly.

01:47:16.960 | Receptor is named after the drug.

01:47:18.500 | And so the important thing to know is that

01:47:20.660 | whether or not it's basal forebrain

01:47:21.780 | or pedunculopontine nucleus or locus coeruleus,

01:47:24.700 | that at least in the brain,

01:47:26.500 | 'cause we're not talking about muscle

01:47:27.480 | where acetylcholine does something else

01:47:29.060 | via nicotinic receptors,

01:47:31.060 | in general, it just tends to be a signal-to-noise enhancer.

01:47:34.120 | And so for the non-engineering types out there, no problem.

01:47:37.460 | Signal-to-noise, just imagine,

01:47:38.820 | I'm talking right now

01:47:39.660 | and there's a lot of static in the background.

01:47:41.560 | There are two ways for you to be able

01:47:43.060 | to hear me more clearly.

01:47:43.940 | We can reduce the static or I can increase the fidelity,

01:47:46.980 | the volume and the clarity of what I'm saying, okay?

01:47:51.260 | For instance, and that's really what acetylcholine does.

01:47:54.720 | That's why when people smoke a cigarette,

01:47:56.200 | they get that boost of nicotine and they just feel clear.

01:47:59.940 | It really works.

01:48:00.840 | The other thing that happens is the thalamus

01:48:03.940 | sends information to a couple of places.

01:48:06.420 | First of all, it sends information

01:48:07.740 | to the reward centers of the brain,

01:48:09.300 | the mesolimbic reward pathway that releases dopamine.

01:48:11.740 | And typically when nicotine is increased in our system,

01:48:14.380 | dopamine goes up.

01:48:16.060 | That's one of the reasons why nicotine is reinforcing.

01:48:18.380 | We just like it.

01:48:19.840 | We seek it out.

01:48:20.680 | I've done beautiful experiments with honeybees even,

01:48:22.780 | where you put nicotine on certain plants

01:48:25.100 | or it comes from certain plants and they'll forage there.

01:48:27.080 | More, you get them kind of like buzzed.

01:48:28.920 | That was a pun, bad pun.

01:48:31.240 | In any event, there's also an output from this thing,

01:48:34.300 | the thalamus, to the ventromedial prefrontal cortex,

01:48:37.700 | which is an area of the forebrain that really allows us

01:48:40.240 | to limit our focus and our attention for sake of learning.

01:48:43.340 | It allows us to pay attention.

01:48:44.920 | This is the circuit.

01:48:45.760 | - You talked about this in your fantastic podcast

01:48:49.540 | on stimulants.

01:48:50.840 | - Yeah, on ADHD, typically ADHD drugs,

01:48:54.120 | so things like Adderall, Vyvanse,

01:48:56.280 | methamphetamine for that matter, Ritalin.

01:48:59.760 | - Yeah, why it's counterintuitive that a stimulant

01:49:02.600 | would be a treatment for someone with difficulty focusing.

01:49:06.200 | - Yeah, in young kids who have difficulty focusing,

01:49:08.920 | if you give them something they love, they're like a laser.

01:49:12.580 | And the reason is that ventromedial prefrontal cortex circuit

01:49:16.840 | can engage is when the kid is interested and engaged,

01:49:20.400 | but kids with ADD, ADHD tend to have a hard time

01:49:23.780 | engaging their mind for other types of tasks

01:49:25.980 | and other types of tasks are important

01:49:27.500 | for getting through life.

01:49:28.420 | And it turns out that giving those stimulant drugs

01:49:30.580 | in many cases can enhance the function of that circuit

01:49:34.420 | and it can strengthen so that ideally the kids

01:49:37.400 | don't need the drugs in the long run,

01:49:38.820 | although that's not often the way that it plays out.

01:49:41.860 | And there are other ways to get at this.

01:49:43.560 | There's now a big battle out there.

01:49:45.460 | Is ADHD real?

01:49:46.680 | Is it not real?

01:49:47.520 | Of course it's real.

01:49:48.340 | Does every kid need ADHD meds?

01:49:49.840 | No.

01:49:50.680 | Are there other things like nutrition,

01:49:52.320 | more playtime outside, et cetera,

01:49:53.960 | that can help improve their symptoms without drugs?

01:49:56.160 | Yes.

01:49:57.000 | Is the combination of all those things together

01:49:59.540 | known to be most beneficial?

01:50:01.120 | Yes.

01:50:01.960 | Are the dosages given too high

01:50:03.120 | and generally should be titrated down?

01:50:06.560 | Maybe.

01:50:07.400 | Some kids need a lot, some kids need a little.

01:50:09.160 | I probably just gained and lost a few enemies there.

01:50:12.160 | So the point is that these circuits are hardwired circuits.

01:50:17.160 | - Sorry, one other question, Andrew.

01:50:20.740 | If my memory serves correctly,

01:50:23.020 | doesn't nicotine potentially have a calming effect as well?

01:50:26.860 | And that seems a bit counterintuitive to the focusing one.

01:50:30.060 | Is it a dose effect or a timing effect?

01:50:32.360 | How does that work?

01:50:33.200 | - Yeah, it's a dosing effect.

01:50:34.320 | So the interesting thing about nicotine

01:50:36.120 | is that it can enhance focus in the brain,

01:50:38.560 | but in the periphery,

01:50:39.800 | it actually provides some muscle relaxation.

01:50:42.340 | So it's kind of the perfect drug if you think about it.

01:50:45.040 | Again, I was reflecting on this,

01:50:48.120 | how when we were growing up,

01:50:49.560 | people would smoke on plane,

01:50:50.580 | they had a smoking section on the plane.

01:50:52.580 | You know, people smoked all the time

01:50:54.000 | and now hardly anyone smokes for all the obvious reasons.

01:50:56.420 | But yeah, it provides that really ideal balance

01:50:59.860 | between being alert, but being mellow

01:51:02.760 | and relaxed in the body.

01:51:03.940 | So, hence it's reinforcing properties.

01:51:06.840 | Okay, this study is remarkable

01:51:09.420 | because what they did is they had people

01:51:12.060 | come into the laboratory, they gave them a vape pen.

01:51:16.480 | These are smokers.

01:51:18.420 | So these are experienced smokers.

01:51:21.620 | Typically there's a washout before they come in

01:51:23.540 | so they're not smoking for a bit

01:51:24.820 | so they can clear their system of nicotine

01:51:26.380 | and they measure-

01:51:27.220 | - How long is that needed?

01:51:28.300 | - Typically it's a couple of days.

01:51:29.580 | - Oh, okay.

01:51:30.420 | - Yeah, which must be miserable for those people.

01:51:32.260 | - I was just about to say,

01:51:33.100 | 'cause they can't have nicorette gum or anything.

01:51:34.220 | - No, nothing, they must be dying.

01:51:35.420 | And I wonder how many cheat.

01:51:36.840 | - But they can measure carbon monoxide, right?

01:51:39.300 | - They measure carbon monoxide

01:51:40.140 | and they're measuring nicotine in the blood as well.

01:51:42.140 | So they do a good job there.

01:51:43.700 | So then what they do is they have them vape

01:51:46.420 | and they're vaping either a low, medium,

01:51:49.600 | or high dose of nicotine.

01:51:51.140 | The dosages don't really matter

01:51:52.920 | because tolerance varies, et cetera.

01:51:55.100 | And then they are putting them

01:51:57.420 | into a functional magnetic resonance imaging machine.

01:52:00.420 | So where they can look at, it's really blood flow,

01:52:03.380 | it's really hemodynamic response.

01:52:05.100 | For those of you who want to know,

01:52:06.540 | it's the ratio of the oxygenated to deoxygenated blood

01:52:09.380 | because when blood will flow to neurons that are active

01:52:13.180 | to give it oxygen and then it's deoxygenated

01:52:15.620 | and then there's a change in what's called the bold signal.

01:52:18.100 | So FMRI, when you see these like hotspots in the brain

01:52:21.740 | is really just looking at blood flow.

01:52:24.220 | And then there's some interesting physics around

01:52:26.140 | and I'll probably get this wrong,

01:52:27.100 | but I'll take an attempt at it

01:52:28.120 | so that I get beat up a little bit

01:52:29.140 | by the physicists and engineers.

01:52:30.800 | Do you remember the right hand rule?

01:52:32.020 | - Yep. - Right, okay.

01:52:32.840 | So do I have this right, correct?

01:52:34.900 | The right hand rule, if you put your thumb out

01:52:36.380 | with your index finger, your middle finger,

01:52:38.400 | your thumb facing up,

01:52:39.660 | I think that the thumb represents the charge,

01:52:41.340 | the direction of the charge, right?

01:52:42.860 | And then isn't the electromagnetic field is the downward

01:52:45.660 | facing figure and then it's, do I have that right?

01:52:48.480 | I have to look this up. - I actually don't.

01:52:51.020 | - Okay, well, someone will look it up.

01:52:52.580 | But what you do is when you put a person's head

01:52:54.300 | in this big magnet and then you pulse the magnet,

01:52:56.960 | what happens is the oxygenated and deoxygenated blood,

01:53:00.980 | it interacts with the magnetic field differently

01:53:03.500 | and that difference in signal can be detected.

01:53:05.680 | And you can see that in the form of activated brain areas.

01:53:08.900 | - Yeah, I mean, MRI all works by proton detection.

01:53:11.980 | So presumably there's a difference in the proton signal

01:53:15.260 | when you have high oxygen versus low oxygen concentration.

01:53:18.760 | - Yeah, that's right.

01:53:20.180 | And what they'll do is they'll pulse with the magnet

01:53:22.020 | because my understanding is that,

01:53:23.620 | and this is definitely getting beyond my expertise,

01:53:26.800 | but that the spin orientation of the protons,

01:53:29.460 | then it's going to relax back at a different rate as well.

01:53:32.660 | So by the relaxation at a different rate,

01:53:34.820 | you can also get not just resting state activation,

01:53:38.660 | like, oh, look at a banana,

01:53:39.620 | what areas of the brain light up?

01:53:41.120 | But you can look at connectivity between areas

01:53:44.420 | and how one area is driving the activity of another area.

01:53:47.260 | So very powerful technique.

01:53:48.840 | So what they do is they put people in a scanner

01:53:52.060 | and then you'll like this 'cause you'll--

01:53:53.220 | - What are the limitations of FMRI in terms of,

01:53:57.100 | I mean, how fine is the resolution?

01:53:59.240 | I mean, where are the blind spots of the technique?

01:54:01.620 | - So resolution, you can get down to sub centimeter.

01:54:05.020 | They talk about it always in these paper as a voxels,

01:54:07.260 | which are these little cubic pixels things.

01:54:11.240 | You know, sub centimeter,

01:54:13.160 | but you're not going to get down to millimeter.

01:54:15.960 | There are a number of little confounds

01:54:18.400 | that maybe we won't go into now

01:54:20.100 | that have been basically worked out over the last 10 years

01:54:23.000 | by doing the following.

01:54:24.200 | You can't just give somebody a stimulus compared to nothing.

01:54:27.880 | I'll just tell you the experiment.

01:54:28.880 | It was discovered, for instance,

01:54:30.160 | that when someone would move their right hand,

01:54:32.260 | 'cause when you're in the MRI,

01:54:33.880 | I just went for one of these recently for clinical,

01:54:36.140 | not a problem, but just for a diagnostics hand,

01:54:38.080 | you're leaning back and you can move your right hand a bit.

01:54:41.420 | And they would see an area in motor cortex lighting up.

01:54:43.960 | But what they noticed was that the area

01:54:45.640 | corresponding to the left hand was also lighting up.

01:54:48.420 | So what you really have to do

01:54:50.220 | is you have to look at resting state.

01:54:52.260 | How much are they lighting up?

01:54:53.780 | Just at rest and then subtract that out.

01:54:56.300 | So now you'll always see resting state

01:54:58.460 | versus activation state.

01:55:00.240 | - Yeah, wasn't there a really funny study

01:55:03.340 | done as a spoof maybe a decade ago

01:55:05.380 | that put a dead salmon into an MRI machine

01:55:08.680 | and did an, like they did an FMRI of a dead salmon

01:55:12.140 | that demonstrated like some interesting signal.

01:55:14.840 | - No, I didn't know that one.

01:55:17.000 | - We got to find this one for the show notes.

01:55:19.680 | - We should do one of these wild papers ones.

01:55:22.120 | There are papers of people putting, don't do this folks,

01:55:24.960 | putting elephants on LSD that were published in science

01:55:27.660 | and things like that, like crazy experiments.

01:55:29.320 | We should definitely do a crazy experiments journal club.

01:55:32.120 | In any event, you can get a sense

01:55:35.100 | of which brain areas are active and when

01:55:36.980 | with fairly high spatial resolution, fairly high,

01:55:40.260 | and pretty good temporal resolution

01:55:42.180 | on the order of hundreds of milliseconds.

01:55:44.800 | But it's not ultra fast

01:55:47.500 | because a lot of neural transmission is happening

01:55:49.540 | on the tens of milliseconds,

01:55:51.500 | especially when you're in talking about auditory processing.

01:55:54.940 | Okay, so they put people into the scanner

01:55:57.460 | and then they give them essentially a task

01:56:00.140 | that's designed to engage the thalamus,

01:56:02.720 | known to engage the thalamus reward centers

01:56:06.040 | and the ventromedial prefrontal cortex.

01:56:08.240 | And it's a very simple game.

01:56:09.620 | You'll like this 'cause you have a background in finance.

01:56:12.240 | You let people watch a market.

01:56:15.160 | You know, okay, here's the stock market,

01:56:16.520 | or you could say, or the price of peas,

01:56:18.720 | it doesn't really matter.

01:56:19.560 | It goes up, it goes down,

01:56:20.480 | and they're looking at a squiggle line, then it stops.

01:56:22.640 | And then they have the option,

01:56:23.780 | but they have to pick one option.

01:56:25.000 | They're either going to invest a certain number

01:56:26.720 | of the hundred units that you've given them,

01:56:29.640 | or they can short it.

01:56:31.240 | They can say, oh, it's going to go down

01:56:32.480 | and try and make money on the prediction

01:56:34.200 | it's going to go down.

01:56:35.280 | You could explain shorting better than I could, for sure.

01:56:38.760 | So depending on whether or not

01:56:39.660 | they get the prediction right or wrong,

01:56:41.760 | they get more points or they lose points,

01:56:43.360 | and they're going to be rewarded in real money

01:56:45.040 | at the end of the experiment.

01:56:46.920 | So this is going to engage this type of circuitry.

01:56:48.820 | Now, remember, these groups were given a vape pen

01:56:53.080 | prior to this, where they've vaped,

01:56:55.520 | what they were told is either a low, medium,

01:56:58.360 | or high dose of nicotine.

01:57:00.720 | And they do this task.

01:57:02.520 | The goal is not to get them to perform better on the task.

01:57:05.400 | The goal is to engage the specific brain areas

01:57:07.560 | that are relevant to this kind of error

01:57:09.640 | and reward type circuits.

01:57:11.960 | And we know that this task does that.

01:57:13.500 | So that includes the thalamus,

01:57:15.040 | that includes the mesolimbic reward pathway and dopamine.

01:57:17.840 | It includes the ventromedial prefrontal cortex.

01:57:20.920 | First of all, they measure nicotine in the blood.

01:57:24.080 | They are measuring how much people vaped.

01:57:26.160 | They were very careful about this.

01:57:27.340 | One of the nice things about the vape pen

01:57:28.840 | for the sake of experiment and not recommending people vape,

01:57:31.900 | but they can measure how much nicotine

01:57:33.340 | is left in the vape pen before,

01:57:34.620 | after they can measure how long they inhaled,

01:57:36.660 | how long they held it in.

01:57:37.920 | There's a lot that you can do

01:57:38.760 | that's harder to do with a cigarette.

01:57:40.520 | Okay.

01:57:43.060 | They measured people's belief as to whether or not

01:57:45.680 | they got low, medium, or high amounts of nicotine.

01:57:48.600 | And it's- - They were told.

01:57:49.760 | - They were told this is a low amount,

01:57:51.960 | a medium amount, or a high amount.

01:57:54.040 | And then of course they looked at brain area activation

01:57:57.680 | during this task.

01:57:58.680 | And what they found was very straightforward.

01:58:00.560 | - Sorry, they were all given the same amount.

01:58:02.220 | - Yes, this is the sneak.

01:58:03.840 | I was going to offer it as a punchline, but that's okay.

01:58:05.680 | No, I think that the cool thing about this experiment

01:58:08.300 | is that the subjects are unaware

01:58:10.240 | that they all got the exact same amount

01:58:12.840 | of relatively low nicotine containing vape pen.

01:58:16.760 | So they basically,

01:58:18.020 | and they're measuring it from their bloodstream.

01:58:19.400 | So they all have fairly low levels of nicotine,

01:58:22.080 | but one group was told you got a lot.

01:58:24.280 | One group was told you got a medium amount,

01:58:25.920 | and the other was told you got a little bit.

01:58:28.360 | Now, a number of things happen,

01:58:30.600 | but the most interesting things are the following.

01:58:33.120 | First of all, people's subjective feeling

01:58:36.280 | of being on the drug matches what they were told.

01:58:40.320 | So if they were told,

01:58:41.140 | "Hey, this is a high amount of nicotine."

01:58:42.880 | Like, "Yeah, it feels like a high amount of nicotine."

01:58:44.720 | And these are experienced smokers.

01:58:46.720 | If it was a medium amount,

01:58:47.760 | they're like, "Yeah, that feels like a medium amount."

01:58:49.600 | If it was a low amount, they think it was a low amount.

01:58:53.060 | Now that's perhaps not so surprising.

01:58:54.860 | That's you're just- - That's the placebo

01:58:56.480 | in that sense. - That's the placebo.

01:58:58.120 | But if you look at the activation of the thalamus

01:59:01.720 | in the exact regions

01:59:04.680 | where you would predict acetylcholine transmission

01:59:07.620 | to impact the function of the thalamus.

01:59:09.440 | So these include areas

01:59:10.380 | like what's called the centromedan nucleus,

01:59:12.060 | the ventroposterior nucleus,

01:59:13.200 | the names that really don't matter,

01:59:14.560 | but these are areas involved in attention.

01:59:16.780 | It scales with what they thought they got in the vape pen.

01:59:22.120 | Meaning if you were told

01:59:23.020 | that you got a low amount of nicotine,

01:59:24.720 | you got a little bit of activation in these areas.

01:59:26.840 | If you were told that you got a medium amount of nicotine

01:59:29.440 | and that's what you vaped,

01:59:30.920 | then you had medium amounts

01:59:33.340 | or moderate amounts of activation.

01:59:35.840 | And if you were told you got high amounts of nicotine,

01:59:38.180 | you got a high degree of activation.

01:59:40.060 | And the performance on the task,

01:59:42.560 | believe it or not, scales with it somewhat.

01:59:45.360 | So keep in mind,

01:59:47.120 | everyone got the exact same amount of nicotine in reality.

01:59:51.080 | So here, the belief effect

01:59:52.520 | isn't just changing what one subjectively experiences.

01:59:56.160 | Oh, this is the effect of high nicotine or low nicotine.

01:59:59.480 | It actually is changing the way

02:00:00.840 | that the brain responds to the belief.

02:00:04.160 | And that to me is absolutely wild.

02:00:07.160 | Now, there are a couple of other things

02:00:08.360 | that could have confounded this.

02:00:10.400 | First of all, it could have been

02:00:12.200 | that if you believe you got a lot of nicotine,

02:00:14.200 | you're just faster or you're reading the lines better

02:00:17.360 | or your response time to hit the button is quicker.

02:00:19.960 | I tell you, you have a drug

02:00:20.800 | that's going to improve reaction time.

02:00:22.240 | You might believe that about nicotine.

02:00:23.640 | And so you're quicker on the trigger

02:00:25.440 | and you're getting, they have a desi--

02:00:26.980 | - More activation as well.

02:00:27.820 | - More activation.

02:00:29.320 | There could be, they rule that out.

02:00:31.180 | They also rule out the possibility--

02:00:32.760 | - How did they rule that out?

02:00:33.860 | - By looking at rates of pressing.

02:00:36.200 | - And there was no difference. - Nothing.

02:00:37.120 | And in sensory areas of the brain

02:00:38.880 | that would represent that kind of difference,

02:00:41.700 | they don't see that.

02:00:43.080 | The other thing that is very clear

02:00:45.280 | is that the connection between the thalamus

02:00:48.480 | and the ventromedial prefrontal cortex,

02:00:50.480 | that pathway scales in the most beautiful way

02:00:54.160 | such that people that were told they had smoked a low

02:00:57.760 | or vaped a low amount of nicotine

02:00:59.700 | got a subtle activation of that pathway.

02:01:02.240 | People that were told that they got a moderate amount

02:01:04.200 | of nicotine got a more robust activation of that pathway.

02:01:07.280 | And the people that were told that they got a high amount

02:01:09.240 | of nicotine in the vape pen saw a very robust activation

02:01:12.840 | of the thalamus to this ventral prefrontal cortical pathway.

02:01:15.520 | Now, of course, this is all happening

02:01:17.680 | under the hood of the skull simply on the basis

02:01:20.680 | of what they were told and what they believe.

02:01:23.000 | - And technically, the fMRI is showing the activation

02:01:26.320 | of those two areas, and that's how you can infer

02:01:29.920 | the strength of that connection.

02:01:31.360 | - That's right.

02:01:32.180 | There's a separate method called diffuser tensor imaging,

02:01:35.120 | which was developed, I believe,

02:01:36.280 | out of the group in Minnesota.

02:01:38.280 | Minnesota has a very robust group in terms of neuroimaging

02:01:42.020 | that can measure activation in fiber pathways.

02:01:44.720 | This is not that, but you can look at the timing

02:01:46.840 | of activation, and it's a known

02:01:48.200 | what we call monosynaptic pathway.

02:01:49.760 | So we haven't talked so much about figures here,

02:01:52.260 | but I guess if we were gonna look at any one figure,

02:01:55.980 | and I can just describe it for the audience

02:01:58.980 | that doesn't have the figure in front of them.

02:02:01.440 | Let's see, probably the most important figure is figure two.

02:02:08.320 | Remember, I said I like to read the titles of figures,

02:02:12.180 | which is that the belief about nicotine strength

02:02:14.400 | induced a dose-dependent response in the thalamus.

02:02:17.300 | Basically, if you and figure 2B can tell you

02:02:21.560 | if they believe that they got more nicotine,

02:02:23.300 | that's essentially the response that they saw.

02:02:27.780 | So if you look, or sorry, panel E,

02:02:30.080 | if you look at the belief rating

02:02:32.080 | as a function of the estimate in the thalamus

02:02:37.080 | of how much activation there was,

02:02:39.660 | it's a mess when you look at all the dots at once,

02:02:41.820 | but if you just separate it out by high, medium, and low,

02:02:44.260 | you run the statistics,

02:02:45.120 | what you find is that there's a gradual increase,

02:02:48.340 | but a legitimate one from low to medium to high.

02:02:52.100 | In other words, if I tell you

02:02:54.020 | this is a high dose of nicotine,

02:02:55.500 | your brain will react as if it's a high dose of nicotine.

02:02:58.320 | Now, what they didn't do was give people zero nicotine.

02:03:01.360 | - Yeah, I was about to say

02:03:02.200 | there's a control that's missing here, right?

02:03:04.240 | - Yeah, so what they didn't do is give people zero nicotine

02:03:07.040 | and then tell them this is a high amount of nicotine,

02:03:10.360 | sort of the equivalent of the cruel high school experiment.

02:03:12.680 | No alcohol, but then the kid acts drunk.

02:03:15.760 | Now, in the high school example,

02:03:19.940 | it's unclear whether or not

02:03:20.820 | the kid actually felt drunk or not.

02:03:22.740 | It's unclear whether or not they had been drunk previously,

02:03:27.120 | if they even knew what it would be like

02:03:28.820 | to feel drunk, et cetera, and there's the social context.

02:03:31.540 | What I find just outrageous

02:03:33.820 | and outrageously interesting about this study

02:03:36.340 | is simply that what we are told about the dose of a drug

02:03:40.960 | changes the way that our physiology

02:03:42.700 | responds to the dose of the drug.

02:03:44.480 | And in my understanding, this is the first study

02:03:48.100 | to ever look at dose dependence of belief effects, right?

02:03:52.460 | To really, and why would that be important?

02:03:54.540 | Well, for almost every study of drugs,

02:03:56.420 | you look at a dose dependent curve.

02:03:58.180 | You look at zero, low dose, medium dose, high dose,

02:04:01.400 | and here they clearly are seeing a dose dependent response

02:04:06.400 | simply to the understanding

02:04:10.440 | of what they expect the drug ought to do.

02:04:14.340 | In other words, you can bypass pharmacology somewhat, right?

02:04:18.380 | - Now, look at figure 2B.

02:04:20.380 | Am I reading this correctly?

02:04:21.720 | So it's got four bars on there.

02:04:25.040 | You've got the group who were told they got a low dose,

02:04:28.100 | the group who was told they got a medium dose,

02:04:30.480 | the group that was told they had a high dose,

02:04:32.040 | and then these healthy controls,

02:04:33.880 | who presumably were non-smokers

02:04:37.060 | who were just put in the machine.

02:04:38.660 | - That's right.

02:04:40.200 | - Yeah, this is measuring parameter estimate.

02:04:43.460 | Is that referring to their ability

02:04:46.900 | to play the trading game?

02:04:50.200 | - The parameter estimate is the activation,

02:04:54.620 | reward related activities

02:04:55.720 | from an independent thalamus mask, right?

02:04:57.480 | So what they're doing is they're just saying,

02:04:58.620 | if we just look at the thalamus,

02:05:00.500 | what is the level of activation?

02:05:01.980 | - I see, so this suggests

02:05:03.380 | that the only statistical difference

02:05:06.120 | was between the low and the high.

02:05:09.740 | - That's right.

02:05:10.900 | - And nobody else was statistically different.

02:05:12.660 | - That's right.

02:05:13.500 | - But that's not the whole story?

02:05:14.320 | - No, that's not the whole story.

02:05:15.420 | So when you look at the output from the thalamus

02:05:18.900 | to the ventromedial prefrontal cortex,

02:05:21.860 | that's where you start to identify the--

02:05:24.040 | - Is that figure four?

02:05:25.300 | - That is, yes.

02:05:26.660 | So this is where you see, so figure four B,

02:05:30.220 | if you look at parameter estimate,

02:05:31.720 | so this is the degree of activation

02:05:33.100 | between the thalamus

02:05:34.580 | and the ventromedial prefrontal cortex,

02:05:36.760 | and it's called the instructed belief,

02:05:38.940 | you can see that there's a low, medium,

02:05:41.600 | and high scatter of dots for each,

02:05:44.040 | and that each one of those is significant.

02:05:46.180 | - So isn't it interesting that at the thalamus,

02:05:50.360 | which is, and you'll immediately appreciate my stupidity

02:05:54.240 | when it comes to neuroscience,

02:05:55.940 | which is more proximate to the nicotinamide,

02:05:58.920 | or the nicotinamide, what do you call it,

02:06:01.860 | the nicotine acetylcholine receptor,

02:06:04.360 | you have a lower difference of signal strength,

02:06:08.040 | and somehow that got amplified

02:06:09.940 | as it made its way forward in the brain?

02:06:12.060 | - Yeah.

02:06:12.900 | - Does that surprise you?

02:06:13.720 | - It is surprising, and it surprised them as well,

02:06:15.660 | that the interpretation they give,

02:06:19.040 | again, as we were talking about before,

02:06:20.140 | important to match their conclusions

02:06:22.140 | against what they actually found,

02:06:23.340 | which is what we're doing here,

02:06:24.680 | the interpretation that they give

02:06:26.820 | is that it doesn't take much nicotinic receptor occupancy

02:06:30.720 | in the thalamus to activate this pathway,

02:06:33.180 | but they too were surprised

02:06:34.540 | that they could not detect a raw difference

02:06:36.640 | in the activation of the thalamus,

02:06:37.820 | but in terms of its output to the prefrontal cortex,

02:06:41.420 | that's when the difference showed up.

02:06:42.260 | - Because that figure, 4B, is more convincing

02:06:46.260 | than figure two, because even figure 2E,

02:06:49.180 | if you read the fine print, the R,

02:06:52.580 | the correlation coefficient is .27, it's not that strong.

02:06:56.700 | - It's weak.

02:06:57.540 | - So at the thalamus, it's kind of like,

02:06:59.260 | yeah, there might be a signal,

02:07:00.540 | by the way, this goes back to our earlier discussion,

02:07:03.040 | there could be a huge signal here and we're underpowered,

02:07:05.060 | how many subjects were in this?

02:07:06.500 | So you wouldn't have a lot of subjects in this experiment.

02:07:08.540 | - Yeah, no, and this just speaks to the general challenge

02:07:12.900 | of doing this kind of work.

02:07:14.060 | It's hard to get a lot of people in and through the scanner.

02:07:16.560 | - Yeah, and it's expensive.

02:07:17.400 | - And it's expensive.

02:07:18.900 | We have to, I should know this,

02:07:19.940 | but we can go back to the methods.

02:07:22.060 | - But you can sort of just look

02:07:23.540 | at the number of dots on here.

02:07:24.780 | I mean, it's in the low tens, right?

02:07:26.560 | It's like 40, 30, something like that.

02:07:29.220 | So it's possible you do this--

02:07:30.060 | - It's not your Danish study.

02:07:31.340 | - Yeah, yeah, you do this with 1,000 people,

02:07:33.300 | this could all be statistically significant.

02:07:35.980 | - Right, it was, so they talk about this,

02:07:38.540 | based on this, we estimated that an N of 20,

02:07:40.500 | N is sample size, in each belief condition,

02:07:42.340 | the final sample would provide 90% power

02:07:44.460 | to detect an effect of this magnitude

02:07:46.560 | at an alpha of 0.5 in a two-tailed test.

02:07:50.860 | - Okay, so that's them referring

02:07:52.580 | to what we just talked about,

02:07:53.700 | which is we believe at 90% confidence

02:07:56.900 | to get an alpha of 0.05,

02:07:58.540 | which means we'll want to be 95% confidence,

02:08:00.820 | we need 60 people, 20 per group.

02:08:02.900 | - Right, yeah.

02:08:03.860 | - But if the difference is smaller

02:08:05.520 | than what they expected,

02:08:07.280 | they'll miss out on some of the significance,

02:08:09.120 | which it looks like they're missing

02:08:10.400 | between the medium and high group.

02:08:11.860 | - Yep, and I too was surprised

02:08:13.700 | that they did not see a difference

02:08:16.260 | between the medium and the high group,

02:08:18.140 | but they did in the output of the thalamus.

02:08:19.940 | I was also surprised that they didn't see a difference,

02:08:22.420 | this is kind of interesting in its own right,

02:08:24.900 | if figure three talks about their belief

02:08:26.460 | about nicotine strength did not modulate

02:08:28.620 | the reward response, the dopamine response.

02:08:31.080 | - How was that measured?

02:08:31.920 | Also, just in fMRI flow.

02:08:34.060 | - Yeah, exactly, so if you look at figure 3B,

02:08:36.400 | other people can't see it, but basically,

02:08:39.320 | what you'll see is that there's no difference

02:08:41.120 | between these different groups

02:08:43.160 | in terms of the amount of activation

02:08:45.260 | in these reward pathways if people got a low, medium,

02:08:47.580 | or high amount of nicotine.

02:08:48.740 | Now, that actually could be leveraged, I believe,

02:08:53.160 | if somebody were trying to quit nicotine, for instance,

02:08:56.360 | and they were going to do that

02:08:57.360 | by progressively reducing the amount of nicotine

02:09:00.200 | that they were taking,

02:09:01.400 | but you told them that it was the same amount,

02:09:03.800 | from one day to the next,

02:09:05.260 | you could whittle it down to, presumably,

02:09:08.000 | to a low amount before taking it to zero,

02:09:10.780 | and if they believed it to be a greater amount,

02:09:13.060 | then it might actually not disrupt their reward pathways,

02:09:16.780 | meaning they would feel, presumably,

02:09:19.320 | they'd feel rewarded by whatever nicotine

02:09:21.240 | they were bringing in.

02:09:22.300 | - What would be your prediction

02:09:23.680 | if this experiment were repeated,

02:09:25.740 | but it was done exactly the same way with nonsmokers?

02:09:30.340 | - Ooh, well, one thing that's sort of interesting,

02:09:34.060 | you asked about potential sources of artifact,

02:09:37.860 | problems with fMRI.

02:09:38.980 | One of the challenges that they know in this study

02:09:41.060 | was you have to stay very still in the machine,

02:09:43.700 | but the subjects were constantly coughing

02:09:46.320 | because they're smokers.

02:09:47.780 | So, okay, so presumably the data would be higher fidelity.

02:09:51.380 | Started chuckling at that one,

02:09:52.420 | but I was like, I had to read that one twice.

02:09:54.840 | I was like, oh, that makes sense.

02:09:55.680 | They're smokers, they're coughing, they can't stay still,

02:09:57.580 | so movement artifact.

02:09:59.780 | But in all seriousness,

02:10:01.380 | I think that for people that are naive to nicotine,

02:10:05.040 | even a small amount of nicotine

02:10:07.860 | is likely to get this pathway

02:10:09.860 | activated to such a great degree.

02:10:11.740 | Sort of like the first time effect of pretty much any drug.

02:10:14.180 | - But I wonder if they would be more or less susceptible

02:10:18.380 | to the belief system.

02:10:20.680 | - Yeah, that's a really good question.

02:10:22.020 | Right, because they have no prior to compare it to.

02:10:23.940 | - They have no pleasant,

02:10:25.660 | they have no experience to compare it to

02:10:27.160 | with respect to the obviously beneficial effects of nicotine

02:10:31.720 | that the smokers are well used to.

02:10:33.480 | - So this is the poor kid that got duped

02:10:36.000 | into thinking the non-alcoholic beer was at alcohol,

02:10:39.060 | though they're actually the winner, we know,

02:10:40.280 | because I did an episode on alcohol.

02:10:41.600 | Alcohol's bad for you.

02:10:42.480 | So in the end, that kid wins,

02:10:43.640 | and the other ones lose poetic justice.

02:10:46.120 | But that kid, having never been actually drunk before,

02:10:50.560 | presumably would experience it more susceptible.

02:10:52.280 | - I would feel like he'd be more susceptible potentially.

02:10:55.000 | - That's my guess as well.

02:10:56.540 | So, you know, my glee for this experiment is not,

02:11:01.540 | or this paper rather,

02:11:02.480 | is not because I think it's the be all end all,

02:11:04.220 | or it's a perfect experiment.

02:11:06.000 | I just think it's so very cool

02:11:08.040 | that they're starting to explore dose dependence of belief,

02:11:11.520 | because that has all sorts of implications.

02:11:14.140 | I mean, use your imagination, folks,

02:11:17.100 | whether or not we're talking about a drug,

02:11:21.120 | we're talking about a behavioral intervention,

02:11:22.880 | we're talking about a vaccine,

02:11:25.200 | and I'm not referring to any one specific vaccine.

02:11:27.440 | I'm just talking to vaccines generally.

02:11:30.100 | I'm talking about psychoactive drugs.

02:11:32.960 | I'm talking about illicit drugs.

02:11:35.500 | I'm talking about antidepressants.

02:11:37.440 | I'm talking about all the sorts of drugs

02:11:39.960 | we were talking about before, metformin, et cetera.

02:11:42.180 | Just throw our arms around all of it.

02:11:44.520 | What we believe about the effects of a drug, presumably,

02:11:50.360 | in addition to what we believe about how much we're taking

02:11:53.280 | and what those effects ought to be,

02:11:55.240 | clearly are impacting at least the way

02:11:56.940 | that our brain reacts to those drugs.

02:12:00.840 | - Yeah, it's very interesting.

02:12:02.240 | I mean, when you consider how many drugs

02:12:04.960 | that have peripheral effects or peripheral outputs

02:12:09.960 | begin with central issues.

02:12:11.560 | So again, I think the GLP-1 agonists

02:12:13.720 | are such a great example of this.

02:12:15.040 | - Osempic. - Yeah.

02:12:16.200 | You know, I don't think anybody fully understands

02:12:20.760 | exactly how they're working,

02:12:22.560 | but it's hard to argue that they're impacting,

02:12:25.440 | that the GLP-1 analog is having a central impact.

02:12:29.640 | It's doing something in the brain

02:12:32.040 | that is leading to a reduction of appetite.

02:12:34.160 | - We believe that. - Yeah.

02:12:35.200 | - Yeah, and I think the mouse data point to different areas

02:12:38.000 | of the hypothalamus that are related to satiety,

02:12:41.240 | that it's at least possible.

02:12:43.920 | - Yeah, I mean, there's no quicker way

02:12:47.040 | to make a mouse overeat or under eat

02:12:49.760 | than by lesioning its hypothalamus,

02:12:51.720 | depending on where you do so.

02:12:53.380 | So presumably these drugs work there.

02:12:55.320 | But again, it speaks to like, what do you need to believe

02:12:58.280 | in order for that to be the case?

02:13:00.040 | - Have they done placebo trials there

02:13:02.040 | where people get something and they're told-

02:13:04.720 | - Oh, they do.

02:13:05.560 | I mean, of course, those drugs have all been tested

02:13:07.200 | via placebo and the placebo groups, you know,

02:13:10.000 | don't do anywhere near as well.

02:13:11.120 | That's how we know that there's activity of the drug.

02:13:13.000 | But again, there's, you know,

02:13:15.720 | that's a little bit different than being told

02:13:19.200 | you are absolutely getting it, right?

02:13:20.960 | 'Cause in the RCTs, you're just told

02:13:24.520 | you might be getting it, you might not be getting it.

02:13:26.760 | So it's not quite the same as this experiment.

02:13:28.720 | This experiment is one level up where you're being told,

02:13:32.420 | no, you're absolutely getting it.

02:13:34.360 | You're just getting different doses of it.

02:13:35.960 | - Yeah, to take this to maybe the ADHD realm,

02:13:38.440 | let's say a kid has been on ADHD meds for a while

02:13:40.800 | and the parents, for whatever reason,

02:13:42.400 | the physician decided they want to cut back on the dosage.

02:13:46.040 | But if they were to tell the kid it's the same dosage

02:13:48.560 | they've always been taking

02:13:49.520 | and it's had a certain positive effect for them,

02:13:52.160 | according to the results, at least in this paper,

02:13:55.640 | which are not definitive, but are interesting,

02:13:58.360 | the lower dose may be as effective

02:14:00.760 | simply on the basis of belief.

02:14:02.360 | And this is the part that makes it so cool to me is that,

02:14:05.780 | and it's not a kid tricking themselves

02:14:08.440 | or the parents tricking the kid

02:14:11.160 | so much as the brain activation

02:14:13.760 | is corresponding to the belief, right?

02:14:15.900 | So that's where this- - No, that's very cool.

02:14:17.400 | - This is why, because it's done in the brain,

02:14:19.120 | I think we can, you know,

02:14:21.180 | it gets to these kind of abstract, nearly mystical,

02:14:24.720 | but not quite mystical aspects of belief effects,

02:14:26.900 | which is that, you know,

02:14:28.060 | your brain is a prediction making machine.

02:14:30.420 | It's a data interpretation machine,

02:14:32.740 | but it's clear that one of the more important pieces of data

02:14:36.200 | are your beliefs about how these things impact you.

02:14:39.760 | So it's not that this bypasses physiology.

02:14:43.400 | People aren't deluding themselves.

02:14:44.640 | The thalamus is behaving as if it's a high dose

02:14:47.000 | when it's the same dose as the low dose group.

02:14:49.880 | Wild.

02:14:50.760 | - Yeah, I mean, I think of the implications, for example,

02:14:52.680 | with blood pressure, right?

02:14:54.200 | Like we don't really understand essential hypertension,

02:14:56.840 | which is the majority of people walking around

02:14:58.680 | with high blood pressure.

02:14:59.560 | It's unclear etiology.

02:15:01.620 | So lots of people being treated.

02:15:04.600 | How do we know that the belief system about it

02:15:06.900 | can't be changed?

02:15:09.240 | And yeah, this is, I don't know, this is eye-opening.

02:15:14.240 | - Yeah, it's cool stuff.

02:15:15.960 | And allochrome is onto some other really cool stuff.

02:15:19.220 | Like for instance, just to highlight

02:15:21.580 | where these belief effects are starting to show up.

02:15:23.880 | If you tell a group that the side effects of a drug

02:15:27.860 | that they're taking are evidence that the drug really works

02:15:31.480 | for the purpose that they're taking it,

02:15:33.680 | even though those side effects are kind of annoying,

02:15:35.560 | people report the experience as less awful

02:15:38.120 | and they report more relief from the primary symptoms

02:15:41.940 | that they're trying to target.

02:15:43.240 | So our belief about what side effects are,

02:15:45.640 | - That's so interesting.

02:15:46.480 | - can really impact how quickly and how compatible

02:15:51.260 | we feel about, how quickly a drug works, excuse me,

02:15:53.720 | and how compatible we feel that drug is

02:15:55.740 | with our entire life.

02:15:56.640 | So maybe if we call them something else,

02:15:58.520 | like not side effects,

02:15:59.480 | but like additional benefits or something,

02:16:01.780 | it's kind of crazy.

02:16:02.620 | And you don't want to lie to people obviously,

02:16:05.000 | but you also don't want to send yourself

02:16:06.800 | in the opposite direction,

02:16:08.220 | which is reading the list of side effects of a drug

02:16:11.760 | and then developing all of those side effects

02:16:15.200 | when, and then maybe later coming to the understanding

02:16:18.460 | that some of those were raised through belief effects.

02:16:21.400 | - We definitely see that.

02:16:22.420 | That's the nocebo effect, right?

02:16:23.920 | That's the one we see a lot with all sorts of drugs.

02:16:28.920 | And it's tough because how do you know which is which?

02:16:32.760 | And I think there are some people

02:16:34.680 | who are really impacted by that

02:16:36.080 | and it makes it very difficult for them

02:16:37.520 | to take any sort of pharmacologic agent

02:16:40.420 | because they basically, they can't help,

02:16:44.060 | but incur every possible side effect.

02:16:46.300 | - Is it true that medical students

02:16:49.560 | often will start developing the symptoms

02:16:51.120 | of the different diseases that they're learning about?

02:16:53.260 | Is that true?

02:16:54.100 | - Well, you know, I'll tell you,

02:16:55.060 | I do think that in medical school,

02:16:56.400 | you start to think of the zebras more than the horses

02:17:01.000 | all the time.

02:17:02.520 | You know, like, you know what I'm referring to, right?

02:17:05.540 | You know, you see footprints, you see hoof prints,

02:17:07.820 | you should think of horses,

02:17:08.880 | but of course medical students,

02:17:09.940 | you only think of the zebras.

02:17:11.620 | There are some really funny things in medical school.

02:17:13.280 | Like there are certain conditions

02:17:15.200 | that you spend so much time thinking about

02:17:16.840 | that you have a very warped sense of their prevalence.

02:17:20.340 | You know, like in medical school,

02:17:21.180 | there's this condition called sarcoidosis.

02:17:22.840 | Like I feel like we never stop talking about sarcoidosis.

02:17:26.280 | I've seen like three cases in my life, right?

02:17:29.440 | Like it's just not that common.

02:17:31.960 | - Does it provide a great teaching tool or something?

02:17:34.020 | - I don't know.

02:17:34.860 | Like I just, some of these things I don't know.

02:17:38.280 | How much time did we spend talking about situs inversus?

02:17:40.860 | This is when people embryologically have a reversed rotation

02:17:44.800 | and everything in their body is flipped.

02:17:46.820 | Literally everything is flipped.

02:17:49.180 | So their heart is on the right side,

02:17:51.400 | their liver is on the left side,

02:17:53.080 | their appendix is on the left side.

02:17:54.300 | Like, and so I'm not making this up.

02:17:56.740 | - How common is this?

02:17:58.200 | - I've never seen it, okay?

02:18:00.540 | - I was thinking about boxing and the liver shot.

02:18:02.160 | Like you could easily be going

02:18:03.580 | for the wrong side of the body.

02:18:04.620 | - No, I swear to God, like as a medical student,

02:18:06.920 | if you were told someone had left-sided lower quadrant pain

02:18:10.480 | to which the answer is almost assuredly,

02:18:12.400 | like they have diverticulitis,

02:18:14.180 | you'd think they could have appendicitis

02:18:16.120 | in the context of situs inversus.

02:18:18.520 | Like the fact that that would even register

02:18:20.880 | in the top 10 things that it could possibly be.

02:18:24.000 | But yes, you just have a totally warped sense

02:18:25.880 | of what's out there.

02:18:26.800 | - Oh man.

02:18:28.080 | Well, this has been pure pleasure for me.

02:18:32.080 | I don't know about you.

02:18:32.920 | I don't know about our listeners,

02:18:34.180 | but for me, this is among the things that I just delight in

02:18:38.680 | and even more so because you're the one across the table

02:18:42.240 | for me teaching me about these incredible findings

02:18:44.360 | and the gaps in those findings,

02:18:46.880 | which are equally incredible

02:18:48.360 | because they're equally important to know about.

02:18:50.280 | - Yeah, so let's do this again in Austin.

02:18:52.360 | - Absolutely, next time on your home court.

02:18:54.840 | - Very well, and bring a little bit of that dew

02:18:56.680 | if you've got it.

02:18:57.520 | - Oh yeah.

02:18:58.340 | - Yeah.

02:18:59.180 | - Yeah, I'll bring a low, medium, and high amount.

02:19:00.260 | - Low, medium, and high amount, I want to know.

02:19:02.160 | [both laughing]

02:19:03.820 | - Thanks, Peter, you're the best.

02:19:05.000 | - Thanks, sir.

02:19:06.280 | - Thank you for joining me

02:19:07.180 | for today's Journal Club discussion with Dr. Peter Attia.

02:19:10.220 | If you're learning from and/or enjoying this podcast,

02:19:12.680 | please subscribe to our YouTube channel.

02:19:14.420 | That's a terrific zero-cost way to support us.

02:19:16.960 | In addition, please subscribe to the podcast

02:19:19.180 | on both Spotify and Apple.

02:19:20.840 | And on both Spotify and Apple,

02:19:22.200 | you can leave us up to a five-star review.

02:19:24.520 | If you have questions for me or comments about the podcast

02:19:26.840 | or guests that you'd like me to consider hosting

02:19:28.920 | on the Huberman Lab Podcast,

02:19:30.480 | please put those in the comment section on YouTube.

02:19:32.680 | I do read all the comments.

02:19:34.680 | Please also check out the sponsors mentioned

02:19:36.400 | at the beginning and throughout today's episode.

02:19:38.500 | That's the best way to support this podcast.

02:19:41.020 | Not so much on today's episode,

02:19:42.900 | but on many previous episodes of the Huberman Lab Podcast,

02:19:45.480 | we discussed supplements.

02:19:46.880 | While supplements aren't necessary for everybody,

02:19:49.020 | many people derive tremendous benefit from them

02:19:51.360 | for things like enhancing sleep,

02:19:52.800 | for hormone support, and for focus.

02:19:54.640 | The Huberman Lab Podcast

02:19:55.920 | has partnered with Momentous Supplements.

02:19:57.560 | If you'd like to access the supplements discussed

02:19:59.300 | on the Huberman Lab Podcast,

02:20:00.920 | you can go to Live Momentous, spelled O-U-S,

02:20:03.440 | so it's livemomentous.com/huberman,

02:20:06.440 | and you can also receive 20% off.

02:20:08.560 | Again, that's Live Momentous, spelled O-U-S.com/huberman.

02:20:12.780 | If you haven't already subscribed

02:20:14.000 | to our neural network newsletter,

02:20:15.760 | our neural network newsletter

02:20:17.120 | is a completely zero-cost monthly newsletter

02:20:19.520 | that includes summaries of podcast episodes

02:20:22.080 | as well as protocols, that is short PDFs

02:20:25.320 | describing, for instance, tools to improve sleep,

02:20:27.800 | tools to improve neuroplasticity.

02:20:29.940 | We talk about deliberate cold exposure, fitness,

02:20:32.760 | various aspects of mental health,

02:20:34.040 | again, all completely zero cost.

02:20:35.720 | And to sign up, you simply go to hubermanlab.com,

02:20:38.480 | go over to the menu in the corner,

02:20:40.240 | scroll down to newsletter, and provide your email.

02:20:42.400 | We do not share your email with anybody.

02:20:44.800 | If you're not already following me on social media,

02:20:46.720 | I am @hubermanlab on all platforms.

02:20:49.120 | So that's Instagram, Twitter, Threads, LinkedIn,

02:20:52.120 | and Facebook, and at all of those places,

02:20:54.680 | I talk about science and science-related tools,

02:20:56.680 | some of which overlaps

02:20:57.720 | with the content of the Huberman Lab Podcast,

02:20:59.360 | but much of which is distinct

02:21:00.680 | from the content of the Huberman Lab Podcast.

02:21:02.760 | Again, it's Huberman Lab on all social media platforms.

02:21:05.880 | Thank you once again for joining me

02:21:07.240 | for today's Journal Club discussion with Dr. Peter Attia.

02:21:10.500 | And last, but certainly not least,

02:21:12.520 | thank you for your interest in science.

02:21:14.360 | [upbeat music]

02:21:16.940 | (upbeat music)

Journal Club with Dr. Peter Attia | Metformin for Longevity & The Power of Belief Effects

Chapters