albedo I, too, like the epigenetic clocks and PhenoAge. That they correlate with chronological age so well, and also relate to mortality risk, suggests to me that aging may be programmed, and that the epigenetic age is tracking that program. 

If this is true, then these epigenetic clocks may also be useful for testing whether age-reversal interventions are effective. Which is useful for our goals! 

Maximus 

Oh dear … here we enter really in troubled waters with the various theories of aging or rather should I say “hypothesis”.

I really did not make my mind yet and sometime I say: well, whatever the right hypothesis will turn to be, this does not allow me sitting and do nothing.

The “programmed” hypothesis goes back to August Weisman in the 19th century. In the meantime, yes, genetics did huge progresses and we do have feature of aging which might be considered programmed and genes influencing the process but we are really far, I guess, for having determined a network of genes which you can mutate to slow/stop the aging process. Yes, you can epigenetically reprogram and that is one of the most recent exciting developments I know of where you can rejuvenate (e.g. in special “progeric” mice for the time being) by in vivo resetting the epigenetic landscape. But that is not all. E.g. I still have trouble in reconciling this with the fact the mortality increases from very early times of our adulthood and not later in life. And there are also other arguments against the fascinating programmed hypothesis.

Then you have the quite intuitive “wear & tear” or “damage” based, sort of physics driven, hypothesis of aging based on stochastic damage driven by interactions with the environment and metabolism. But also this has issues. I think SENS is mostly adoptive of this approach calling for “repair” and often I feel to endorse this vision. Somewhat it looks like freeing me from entering a very controversial area, which counts I guess in the 10-20 different hypothesis, and rather be proactive against aging trying a “repair and maintain" approach. But on a more scientific basis I do realize many scientists want to understand what aging really is even if, as one friend once told me, dying in doing it ......! Many call for a system biology and multilevel approach to such a hugely complex phenomenon to maximize chances of deep understating.

Possibly the damage and programmed hypothesis will one day converge, maybe based on a better understanding of why and how evolution evolved aging.

albedo I appreciate your thoughtful response, Albedo. Back in 2007, I felt similarly to what you said: 

albedo said:
well, whatever the right hypothesis will turn to be, this does not allow me sitting and do nothing.

So I began studying aging and what we can do about it. 

I, too, have not yet made up my mind whether the age-associated increase in the human mortality rate is caused by damage accumulation, programmed aging, or both. 

Regarding programmed aging and the increase in mortality very early in adulthood, I can actually see that as an argument for programmed aging. This could be somehow related to the cessation/completion of adolescence, when many factors change significantly (e.g. life-long decline in IGF-1, growth hormone secretion, melatonin production, etc.).

One theory I wonder about is whether the age-related increase in mortality may be caused by the program of development (adrenarche and gonadarche) continuing, even though physiological maturity has reached its end. This is very similar to what Mikhail Blagosklonny has said about "hyperfunction". Maybe that's true, but I'm not yet sure how it can cause so much of the atrophy of glands and tissues associated with aging. And why all of the atrophy in the context of reduced IGF-1 and other growth factors? 

But just because programmed aging is an intriguing theory (to me), doesn't mean I necessarily believe it to the exclusion of others. As you said in your post, I can not sit and do nothing, and will work on bringing degenerative aging under complete medical control, regardless of the mechanism(s). 

Maximus

Regarding programmed aging and the increase in mortality very early in adulthood, I can actually see that as an argument for programmed aging. This could be somehow related to the cessation/completion of adolescence, when many factors change significantly (e.g. life-long decline in IGF-1, growth hormone secretion, melatonin production, etc.).

Thank you Maximus, I understand your point but I am still confused (not your fault of course but only my lack of understanding!) and would need to go back to studying in more depth, but in short: shouldn't evolution favor more offsprings, which a longer living organism would generate, so disfavoring a "program" starting in early adulthood? In other worlds, if the hypothesis is right, evolution would have "programmed" aging later in life (I think this is what some would call the "decreasing of evolutionary pressure") not in earlier adulthood as on the contrary data seem to show (I think this based on the experimental mortality Gompertz's "law"). Some (Josh Middeldorf??) might associate this to a non-neodarwinian hypothesis of aging as a "protection" of the specie but I might be completely off track! That is what I meant earlier by "troubled waters": for every single hypothesis, there are pro- and counter-arguments and I feel the necessity of the experimental approach to avoid constantly swinging between them!

albedo 
You're welcome Albedo. I'm enjoying discussing with you. 

You said: 

albedo said:
shouldn't evolution favor more offsprings, which a longer living organism would generate, so disfavoring a "program" starting in early adulthood?

That's a good idea, but I'm not sure that's the case. I have been wondering: perhaps longevity needed to be balanced with the amount of survival time required to reproduce. 

For example, humans take a relatively long time to be reproductively mature and active. I have a suspicion that the developmental program to maturity dictates the lifespan, because that developmental program is the cause of degenerative aging. 

For example, in mice, the time required after birth to become reproductively mature is very short compared to humans: weeks, and not over a decade like humans. And mice have a very short lifespan compared to humans. 

In brief: one theory I'm considering is that degenerative aging is merely an extension of the program of physiological development. Thus, lifespans of various animals will correlate with their time to sexual maturity, because it is that program of physiological development that causes degenerative aging. 

In short: the faster an animal reaches sexual maturity, the faster they die. 

I did a little research on this, and it's pretty consistent. For example, cows reach sexual maturity  at only 1 year old, even though they are so big! But they only live around 20 years. In other words, like mice, the "developmental program" of cows is more aggressive than humans, but they also age faster, and die more quickly, than humans. 

Humans, perhaps because they had much less predation than other animals (maybe because we're more clever, use tools, work in groups, etc.), it was OK to take 10+ years to become reproductively active. According to my theory, our "maturity program" is slower, and that gives us our longer lifespans. We develop (and age!) more slowly than most other animals. 

And this is still consistent with your idea, in a way: we may not live long enough to procreate for hundreds of years, but the duration of our reproductive activity is far longer than a mouse (or even a cow) During this time, humans have much more time than many other organisms do, which is available to produce more offspring. Imagine if mice were reproductively active for 40+ years like humans are...

___
Maximus 

Maximus 

Thank you for your educated and well researched reply !

Just for my own record here I think this must be the paper Dan Mc mentions in his original post and I need to read it but from first approach it looks much like the Morgan Levine et al. paper I already have mentioned in my previous post.

albedo Thanks Albedo. 

Yes, that will be an interesting paper from Levine about Phenotypic age. Thank you for linking it for us. 

albedo 

I have found that all of the blood variables needed to calculate the DNAm PhenoAge  given in Morgan Levine’s paper can be obtained from blood work provided by LifeExtension with their C-Reactive Protein (CRP) ($42) blood test and their Chemistry Panel & Complete Blood Count (CBC) ($35) blood test.  I have just ordered these for my wife and me.   When the results are available, for the creatine and glucose entries one must convert mg/mL to umole/L and mmol/L, respectively, but otherwise there should be no problems.  We plan to determine our DNAm PhenoAge values before and after senolytic treatments with fisetin and with D+Q.

JGC 

That is great JGC. It would be nice if you can share some results here. If you arrive also to gather additional data point before and after treatment that would possibly also reveal trends.

JGC How are you going to run the results? I don't see any obvious formula given in the paper to let you plug in your numbers and get your PhenoAge.

JGC Oh, sorry: I just now read your later reply. Is your Mathematica file a fixed document that just contains your results and your PhenoAge worked out, or can it be used to input one's own values and have it crunch out one's own PhenoAge?

Thanks!

Iðunn 

The formulas needed are given in Supplement 1 of the Levine paper.  My Mathematica notebook has two "vectors", one containing the blood-work values for me and the other for my wife, with two of the values multiplied by conversion factors to put them into umol/L and mmol/L.  Another user, if they had Mathematica, could just put in his own values.  Also, the calculation gives the mortality probability in 10 years, but the 10 can be changed to get mortality over another period.

I may later do an Excel spreadsheet that makes the same calculations, but it was initially easier for me to use Mathematica.

JGC 

When you actually change 10 years to 20 years, you get a higher mortality.

However, at the same time the phenoage changes when you plug 20 in place of 10.

This does not make sense to me. I would expect the phenoage to remain the same.

albedo 

I'll be glad to share the results of our DNAm PhenoAge  values before and after fisetin and D+Q, but it may be several months before the results are available.

My wife and I have used Aging AI recently with blood-work values from general physical exams taken in early July.  It got out sexes right and said that my predicted age is 35.0 and hers is 37.0.  Actually, I'm about to turn 84 and she is about to turn 79.  I attribute the huge differences to the fact that we've both been taking 85 mg/day of metformin, plus other supplements from LifeCodeRX, but it could be just that their algorithm is screwed up.  It will be interesting to see what our DNAm PhenoAges are.

JGC 

I am quite skeptic of results and probably trends are more informative than absolute values. Have you tried which version (they have 3)? I got lower guessed age with version 3.0 than 1.0 the latter being more accurate I guess because based on more biomarkers.

albedo 

FYI, we used Aging AI 3.0, which they claim is just as accurate as the earlier versions but needs fewer input variables. In my opinion, what is critically missing from all three Aging AI algorithms is any measure of the degree of inflammation present.

Yesterday my wife and I had our blood drawn in preparation for obtaining the blood values needed for the DNAm PhenoAge calculation.  When available (in perhaps a week) I'll report how the "ages" obtained compare to our Aging AI "ages".  (Note: we also took our first (of 4) 800 mg doses of Fisetin.)

The Cooper-Nissan paper to which you refer is certainly  comprehensive, but it is not at all clear how an individual like me (as opposed to a research group) can use it for anything useful.

JGC 

On the first point I tend to disagree with you JGC. They provide for v1.0 a MAE (error) of 5.5 years while for v3.0 the give 5.9 year. It looks to me more logical as v1.0 might be more accurate in prediction as containing more parameters.

For the reference I provided I agree it might not be of practical use for all of us but I wished to know if something similarly comprehensive exist but more recent and if studies have been done longitudinally on cohorts to see changes. However, as the list comes out from an expert group, it is still useful to me. E.g. as I understand from your posts, you particularly care (rightly) about inflammation they mention IL-6 which for example I never really measured. I just discovered LP(a) which many of us got tested in a lipid panel can be driven by IL-6 so a low level of LP(a) might be a proxy of IL-6 and in addition you can combine that information with other common inflammation markers such as hr-CRP, Ferritin and others which might even be more organ specific.

albedo 

Our recent blood tests, LifeExtension's CBC Chemistry Profile and C Reactive Protein measurement, provide all the values needed for the DNAm PhenoAge calculation and for Aging AI 3.0, but they do not include several of the values that Aging AI 1.0 wants, so I can't do that one.  I suppose I could order more blood-work, but the difference in an estimated 5.5 year error and a 5.9 year error does not seem that significant to me.

JGC 

Yes I think at this stage the most important if the DNAm PhenoAge input and would not bother too much with aging.ai . Do not bother to reply and focus on the more important.

In any case, the aging.ai prediction should not change much when you lack some values. Sometime I left blank or used extrapolations from other measurement I did or used a mid point in my lab ref ranges. In v1.0 they note that the markers with a ** are "Required parameter for minimal prediction accuracy of 70% within 10 year frame". In v1.0 reference paper they write: "...Many users expressed no desire to specify all 41 parameters of the blood test, so we added an option to enter only the 10 most important markers. The average number of missing values provided by the volunteer testers was 18.5 markers per person. There are several strategies for filling skipped values, including zero, mean, mode and median over all values of each marker. Evaluation of these 4 strategies on the aging.ai data showed that median filling strategy has the best performance in terms of both R2 and epsilon-prediction accuracy (Figure 4C & D)."

In my case and for the anecdote, over the 10 years range 50-60 years old of chronological age, I find:

- systematically higher level of predicted age with v1.0 vs. v3.0. The midpoint value at 55 goes from 32.4 years (v3.0) to 50.2 years (v1.0) or +55 %. I was expecting this a bit as, intuitively, more parameters should get a predicted age closer to the real one.

- while the v3.0 results (plot: predicted vs. chronological) does not show a trend and remains basically flat over the 10 years range, the v1.0 seems to show a small trend downward, mostly in later years, possibly hinting to some impact of lifestyle changes.

I must say I am quite skeptic about what all this means and will take a DNAm test.