I used to work in genomics. I think the quest to have a universal epigenetic measure of ageing will end up in a dead end.
Ageing is relative to a starting point and the variability between people in terms of phenotype is huge.
By way of analogy, would it be possible to come up with a universal measure of ageing for cars? Cars are a complex system made of thousands of components, each of which can age differently over time. Each car make and model will age in a different way. Some of the components even slightly improve in performance over time as they “break in”.
The best way to measure epigenetic ageing is relative to a starting point. There are few studies that follow the same people over time (longitudinal) because it’s difficult and expensive to pull off. We’ll eventually get there though.
- The accuracy of current epigenetic clocks suggests it might not be a dead-end with average error of 2-3 years, across multiple tissues (and even Chimpanzees).
- The variability of an aging phenotype is not that high. We all loose bone density, loose teeth, hair turns white, memory degrades, wrinkles appear, etc. In other words - aging follows a pattern.
- The car analogy assumes that aging is caused by damage over time. But there is an alternative explanation: aging provides and evolutionary advantage for a species and hence is predetermined.
> The variability of an aging phenotype is not that high. We all loose bone density, loose teeth, hair turns white, memory degrades, wrinkles appear, etc.
People age at different rates depending on a wide variety of factors.
Progeria, over consumption of alcohol, too much stress, too much radiation from sunlight, etc.
The things you point to are measures of a biological age, not a wall clock age. While they do correlate, they're not 1:1. Some people age faster.
> But there is an alternative explanation: aging provides and evolutionary advantage for a species and hence is predetermined.
I don't buy this. We don't have a pre-programmed death. It's rather that we haven't had any genetic pressure to live beyond our current lifespan as it doesn't increase our offspring's chances of success.
There's little pressure from competing with children for resources. Look at how many people there are in the world today - the world supports billions of humans. There are far greater pressures being exerted in other ways.
I think we do have a pre-programmed death. Almost all creatures die when they get older so in order to look for evolutionary pressures we have to look across a long time period and far back. For one - if organisms wouldn't die evolution would not happen. Or at least the non-dying ones would stay behind and be left living in environments they are no longer suited for. For ancient creatures food and resources might have been scarce and so it was advantageous for non-evolving old members to die-off sooner.
Also, whatever we think about these epigenetic clocks, there are few things to come to terms with: 1) they work across pretty much all individuals (and people with progeria do exhibit older epigenetic age [1]) 2) they work across pretty much all tissues. I don't see how this could be explained through "damage over time". The way I see it damage should be random and not lead to something so predictable that we could use it to guess a persons chronological age with 2.5 year accuracy.
The immortal organism would not evolve. Imagine you and I lived 100K years from now. Other humans would evolve somewhat during that time but we would be fixed. Climate change might happen, new humans would adapt to the changing environment via evolution, but we would be stuck suffering the new environment.
The humanity example takes a long time to be noticeable. But for early multi-cellular organisms this would kick in much faster. Thing about micro organism fighting against toxins. The ones that do not evolve and adapt would die off sooner. And while they are still alive they are competing for resources with the newer versions of themselves. So for the colony it is advantageous for the older ones to die quicker since we already know they will be unnecessary sooner or later.
Now we evolved from these primitive creatures. And the mechanisms for things like aging, circadian rhythm, sexual reproduction, etc - we inherited from the early creatures. So regardless if the advantage of dying sooner is beneficial or not to the humanity now - the mechanism itself is far more ancient than humanity or even mammals.
From an evolutionary perspective, I think we can at least say that absence of death doesn't confer a selective advantage, else it would have appeared and radiated in several billion years of biological evolution.
It might even be that death provides selective advantage(s), which would account for its being a nearly universal feature of known life. A "pre-programmed death" mechanism would be consistent with this.
> A "pre-programmed death" mechanism would be consistent with this.
We haven't found this mechanism in humans. Cancer and heart disease simply don't impact child rearing.
It requires a great deal of optimization to rejuvenate cells after injury and simultaneously tamp down on uncontrolled proliferation (cancer). It's an extremely complex system. The evolutionary gradient to optimize this is probably steep, and we haven't climbed it since our form is more than good enough. (There are billions of us.)
People cite tortoises, but these animals have a slow metabolism. They don't eat a wide variety of things. They don't have many enemies. They're not burning very fast.
>Cancer and heart disease simply don't impact child rearing.
I'm not sure of that. There's data-supported discussion of possible contribution to offspring survival by grandparents [1]. If true, that would disclose a fitness benefit of familial long lifespan.
Yes, but "in the wild", length of a human generation was much shorter than today. It still is in some developing countries.
People would start reproducing around 17, so around 40-45, you already would be a grandparent with multiple grandkids. This is still early for cancer or heart disease, unless you inherited some unlucky genes.
Epigenetic clocks have been built for many mammal species, not just chimpanzees, and they seem to work reasonably well for them. But for chimpanzees some human clocks can be used directly, without adjustments - indicating how close relatives we in fact are.
aging mostly happens after an organism reproduces, so there is no evolutionary pressure to maintain fitness. there is no pressure to develop maintenance mechanisms...
There would be plenty of pressure to keep reproducing indefinitely, just that then there might be too much competition between ancestors and descendants for the same resources.
Because changing conditions could necessitate the species adjusting to the new conditions as they change, if the young have to outcompete both the old and the new landscape there's a higher chance of an extinction event.
My noob perspective is that generation periods are optimized for environmental change sensitivity.
Humans seem to reliably be able to handle environmental changes expected in their life spans. Other species not, and I dare say most other fauna has much a faster generation iteration rate because they are much more sensitive to the environment. Ergo evolution made their lifespans shorter so they can iterate faster and respond with a tighter gene fit to the environment.
Humans have their fitness less correlated by genes so a longer life span is allowed, but not boundless. In a sense every body must die so that new solutions/iterations can be ran.
What you're missing about the epigenetic theory of aging is that aging, in the epigenetic theory, is a programmed stage of life like puberty. The body somehow knows that it's time to develop secondary sexual characteristics and change growth patterns when a human reaches a certain age. The same machinery decides that the slow deterioration of the person should begin. If one could control that machinery, one could turn off that mechanism of aging. The disease progeria is a possible example of the epigenetic clock malfunctioning and running too fast.
The source of all this is perhaps the suprachiasmatic nucleus which is important to perceiving the passing of time and emits growth hormone releasing hormone which signals the pituitary to make growth hormone.
I think epigenetic age _does_ make sense for very specific systems. For example, just focused on the optic nerve. Good luck getting an optic nerve sample in a live human though!
I believe human trials are due to start on this soon if they didn't already. I will try to find the papers. In the mean time here [1] are the human cells.
A thirteen year old having low bone density, heart problems and degenerating joints would be strange, but for someone > 70 it’s almost expected. The car example is dumb because cars don’t have a built in mechanism for regeneration that is failing. Our bodies mechanisms for self-regeneration weaken and then we see all the symptoms of aging.
The reason for pursuing clocks is to use them to assess the efficacy of longevity interventions. Currently, with an error bar ~3 years they are not bad at all, and can be used to assess aging relative to a starting point, as you suggest.
would it be possible to come up with a universal measure of ageing for cars?
Not universal or at least not easily or affordably. I can test each critical component and get a pretty good idea how much stress that component has been exposed to. Maybe the same could be done with mammals and we just test the critical components to get some numbers then apply a rough estimate of a score. Then perhaps a formula can be applied to the set of scores from all the critical components to assign a health score. I am considering the concept perfect is the enemy of good in this case.
Any thoughts on the organ-based biological age score that Bryan Johnson presents? More equivalent to monitoring all the components of the car.
I think these biological age scores are a really healthy way to debate and create broader awareness about a) how our bodies work and b) how we understand aging.
> In 2019, a small study raised the tantalizing prospect that ageing could be reversed. Scientists in California gave 9 men aged 51 to 65 a growth hormone and two diabetes medications for a year. The drugs seemed to rejuvenate the men’s thymus glands and immune function. They also shaved 2.5 years off the men’s biological age, as measured by one of the most talked-about technologies in ageing research: epigenetic clocks.
It's also worth nothing that another study shaved "3.2 years" off the "epigenetic clock" simply with lifestyle changes over 2 months. [1]
Stem cell biologist here. Note that different cell types have different epigenetics. Most of these studies are looking at white blood cells. Effects on these cells may be different from elsewhere (brain, muscle, etc.)
The epigenetic clock by one of the people quoted in the article - S. Horvath, is notable for working across different tissues, with a few exceptions like germ cells.
I believe we have a genetically determined lifespan. There is a reason lifespan has been optimised to this duration. As we age we accumulate viruses, mutations. The tissues that were positioned and differentiated during development accumulate damage. From the perspective of the species the only feasible way to get back to a healthy state isn't to eradicate the viruses and repair the tissues, its to jettison the withered husk and repeat the development process with a fresh new body. I we 'reset the clock' we're not removing the damage, the mutations, we're just overclocking the body. If overclocking can get me another 20%, I'll take it. Hell 3%, we'll take it. But what I want I really want is the full factory refurb, with a little extra ram please.
This is a good example, because plants don't typically have a predetermined body plan, so they are able to repeat development of complex tissues in new growth.
> Scientists in California gave 9 men aged 51 to 65 a growth hormone and two diabetes medications for a year. The drugs seemed to rejuvenate the men’s thymus glands and immune function. They also shaved 2.5 years off the men’s biological age
The study itself says:
During the first week of the trial, rhGH alone (0.015 mg/kg) was administered to obtain an initial insulin response, and during the second week, rhGH was combined with 50 mg DHEA to evaluate insulin suppression by DHEA alone. During the third week, the same doses of rhGH and DHEA were combined with 500 mg metformin. Beginning at the fourth week, all doses were individualized based on each volunteer's particular responses.[1]
So the meds in question are:
- recombinant human growth hormone (rhGH) [2]
- dehydroepiandrosterone (DHEA) [3]
- metformin [4]
If you wanted to have this treatment for yourself, how could you proceed? I assume that it would be just about impossible to convince your family doctor to prescribe this drug regimen for you? If you were to do-it-yourself, what would be the best way? It seems that metformin is widely available but I don't know about the availability and cost of rhGH and DHEA.
First and foremost I would suggest doing research. A recent study showed that in healthy individuals Metformin had no impact on all-cause mortality and there are potential side effects.
Of course, doing research is hard. Because it isn't really research in the sense of running a legitimate study, it's research in the sense of studying the work of others as an outsider.
He's not afraid to reverse positions, he's articulate, he calls out when research is or is not compelling and explains why.
Again, personally, I expect the number one thing you can do to improve your healthspan is probably focus on what you can stop putting in your body vs what things you can add. If you're 40+ though, might be worth looking at other options.
Berberine is commonly promoted as a vitamin alternative to Metformin. Even if Berberine also has no impact on all-cause mortality (idk), I can anecdotally share that 500mg 30minutes before lunch considerably lessens my post lunch fatigue.
As for human growth hormone, this Huberman Labs podcast advises against taking it but also offers many natural ways you can boost it[0].
In this concrete study (TRIIM and TRIIM-Extended), metformin isn't used for any putative anti-aging effects.
It is used to counteract the effects of HGH on insulin levels. HGH increases insulin levels, which is known to be deleterious to the organism, and metformin serves to reduce them back.
For a more thorough dive into this topic, Dr Brad Stanfield on Youtube. I haven't found anyone better in breaking down the latest scientific studies on Longevity.
I would also caution HGH for men since it can enlarge the prostate gland.
David Sinclair had mentioned in a podcast that we use to be cold and hungry and now we're warm and fat all the time. Need to be cold and hungry more often.
Peter Attia has said that the number one thing he believes will extend healthspan/lifespan (if you did nothing else) is exercise. He wish he could prescribe as a medication.
Michael Lustgarten Ph.D is another good source for biohacking and longevity matters. he takes a diifferent approach though since he's trying to reverse his epigenetic age specifically through diet.
But when being cold and hungry was the norm, life expectancy was much lower. Human life expectancy in the developed world is higher than at any time in human history.
Let's be careful not to conflate life expectancy and life span. The former being dominated[1] by improvements in reducing infant mortality and mothers dying during child birth, while the latter has remained largely consistent[2],[3].
‘… life expectancy in the mid-Victorian period was not markedly different from what it is today. Once infant mortality is stripped out, life expectancy at 5 years was 75 for men and 73 for women.’[1]
And back then also a lot of other factors were less than ideal. (Including lack of medication and medical science altogether.) The idea behind cold and hungry is not being that cold for that long and that hungry. Definitely not starving to death.
What Sinclair hypothesizes that the reason behind the phenomena they see is that these adverse conditions trigger mechanisms in our body that are beneficial in the long term. And the very reason might be to try to make up for what causes the the lower life expectancy. Not knowing much about the topic and the mechanism, my other thought was that it tries to slow your ageing so that you can be fertile for longer in a bid to be able to reproduce when the rough times are over.
So the idea is to kick in the protective mechanisms but still cause no real damage. Similarly to how you're better off eating less than what the average person does in any obese society. Even though their life expectancy is still higher than what it was 500 years ago.
The most interesting thing I have heard from Sinclair is the studies on mice that eating once a day was what extended lifespan the most. Everything else was not that important diet wise.
It sounds like to me the method is eat once a day and get in the best physical condition possible from working out. Beyond that, I am just not going to stress about it. Surely, not going to take any reverse risk with HGH or experimental drugs.
AgelessRX will prescribe metformin [1], DHEA is legally OTC in the US, and any kind of HGH is probably readily available at your local gym (also a number of hormone doctors online will prescribe it easily)
There are peptides you can buy online that make you overproduce HGH, though it’s hard to say if the effects would be the same. You can also easily buy metformin from Indian pharmacies.
>unblinded study with no placebo control arm. “If you have nine people,” says Horvath, “and you get a statistically significant result, it means there’s a strong effect.”
Wow, no. Just... No. It doesn't.
That is so wrong that, from a researcher with a doctorate in biostatistics, it's very difficult not to infer deliberate misinformation on their part. To give some benefit of the doubt though, maybe it was just a very poor choice of words.
Statistical significance with 9 samples is suggestive at best. It just barely qualifies as a pilot study.
They even say that significance means there's a large effect, which again... No.
Significance in small sample indicates practically nothing by itself. What was the actual effect size? Though even at 0.8, with 9 samples I would be be very cautious in my optimistic.
I wonder if, read charitably, the intent is to say with a strong enough effect a small statistic size doesn’t necessarily mean useless. For example, if I removed the hearts from 9 living people and they all died, I can conclude people need their hearts to live.
Yes, I try to allow some benefit of the doubt in my comment, but from a doctorate is statistics it's a bit harder to excuse.
He doesn't have to make an argument from logical implication, there are perfectly good measures of effect size he could use that would allow him to say something like "not only was this statistically significant but it demonstrated a large effect size as well."
My hope is he said something like "We not only saw a P value of < 0.04 but a Cohen's D of 0.7." and then the reporter was like, "yeah I'm gonna need you to dumb that down for our readers"
> That is so wrong that, from a researcher with a doctorate in biostatistics, it's very difficult not to infer deliberate misinformation on their part.
We don't really know the context for that statement. The journalist could have easily misunderstood something and used quoted him in a totally wrong context.
I can give him a little benefit of the doubt, but if he said those specific words and it wasn't a misquote then it was an extremely poor choice of words. Those words, in that order, are so fundamentally wrong that I cannot imagine them being correct even in additional surrounding context unless the next few words that were cutoff were "...as measured by [insert preferred methodology for effect size]" that would still be poor phrasing, implying that significance led to effect size.
I really wish they had a group that had the DHEA and metformin without growth hormone. I can get those easily and cheaply on the internet. I’d like to see how much of the effect they saw was just those two.
One of the issues with metformin is how many times it has gotten recall for cancer causing contamination. Buying on the internet, you are never going to get notified
Metformin is not ideal for some if not most men. It wipes out testosterone levels.
After a few weeks of taking it my T levels were below average and free-T levels were that of 100 year old men.
This was a huge drop from the above average levels, for men in 20s, that I usually have.
I’ve had the same experience. Feel like shit every time I get hopped up on the latest paper and try it again. I don’t want to live forever with T levels of an old man. I thought maybe I was the only one feeling this way, although I had seen the papers about it lowering testosterone.
[0] might be relevant to you, seen other papers that I found when I discovered this issue.
From what I gathered the reason for this is that metformin reduces blood cholesterol levels which are important for testosterone production.
I think a better question would be not how long but how well people live their end of life. Most people would trade living up to 90 fully aware and mostly active instead of up to 100 in misery and dementia.
Well, they could measure the biological ages of people who die (or maybe who have terminal illness) and compare it to people who continue to live on.
If they can get the price of the measurement low enough (which Sinclair claims to happen soon), then we can just wait and see: by doing a large number of measurements we'll have people who end up dying pretty quickly. (Yet another way to do it would be just to collect a large number of samples and measure those who die and compare it with some who do not, during the experiment.)
You can approximate this somewhat by watching their blood tests, kidney function, pulmonary function, grip strength, state of their arteries, strength of their immune systems etc.
People start deteriorating in these parameters long before they develop actual life threatening diseases.
Specifically, watch Bill Gates, Bezos, Branson, Thiel, and Musk. They're first in line with all the influence and resources needed, so if current life extension efforts pan out, we should see it happen.
If it happens soon, the Putin situation could get weird.
Much of the information we collect has a half life, or situations change and that information, while still valid, is not applicable to the issues we are facing currently. The vast majority of things we learn end up getting discarded at some point.
Think of the most recent book you read. How many of the details actually stuck with you at the moment you closed it. Now think of a book you read 10 years ago. How much of that do you remember?
I took years of mathematics and was able to do well on tests, but 40 years later I only recall the "shape" of PDE solutions and couldn't actually solve anything anymore. Instead my brain is stuffed full of arcane knowledge I need to do my job, and if I don't use it within a year, I probably need to learn it all over again.
This is a valid question for different reasons, too. I have a problem with the speech and behavior of immortal characters in books - why would a 300 year old powerful vampire behave like an immature emo teenager?
There's a reason we associate wisdom with age. The more times anyone of reasonable intelligence makes mistakes, the more opportunities they have to learn, and their behavior changes according to the degree to which they take on the lessons of life.
One big danger of immortal dictators is the simple fact that they'll stop learning. Through wealth and power they shield themselves from the consequences of mistakes, getting themselves and their people stuck in a local minima.
Imagine immortal Mitch McConnell, ever increasingly wealthy through passive income, maintaining power and privilege for his constituents and thus his hold on a senate seat.
If life extension pans out, liberal societies will have to impose term limits in a serious and well considered way. Humans aren't ready for the existing pace of technological development, and we're going to encounter an exponentially increasing number of problems, like the politics of immortality. The best thing we could do would be to maximize freedom of expression and minimize the duration of social institutions to achieve sufficient maneuverability to adapt to modern life.
It wouldn't be a limitation for centuries, or possibly millenia. The brain very efficiently packs information, and integrates with external storage. We'll be able to digitally augment brains directly long before temporal memory capacity becomes a problem.
> An alternative is to look at interventions that extend the lifespan of other animals. Some, such as calorie restriction, do seem to reduce epigenetic age in mice. Nevertheless, Kaeberlein wants to see all such manipulations systematically investigated — and the proposition that clock scores can predict lifespans at the individual level tested.
> He stresses the importance of reporting negative results, and determining whether all treatments that extend lifespan or improve late-life functionality decrease epigenetic age — as well as whether all interventions that decrease clock scores increase longevity.
This is it, test the theory properly in mice. Forget doing poorly designed human trials until then.
2 days per week I do a coffee and beer only fast. I can guarantee no toxins survive the resulting torrent. I like to know I'm starting the week with a totally blank slate.
6 days a month!? Even if an RCT comes out proving this works, I’d prefer to die a little sooner rather than do that my whole life lol
The way I see it, you’re burning something like (6/30)/2 = 10% of your quality-adjusted life years. Fasting had better increase lifespan by 7-8 years to be even close to worth it
Ageing is relative to a starting point and the variability between people in terms of phenotype is huge.
By way of analogy, would it be possible to come up with a universal measure of ageing for cars? Cars are a complex system made of thousands of components, each of which can age differently over time. Each car make and model will age in a different way. Some of the components even slightly improve in performance over time as they “break in”.
The best way to measure epigenetic ageing is relative to a starting point. There are few studies that follow the same people over time (longitudinal) because it’s difficult and expensive to pull off. We’ll eventually get there though.