Variability Among Horvath-type DNA Methylation Age Tests

    I have had four Horvath-type DNA methylation age tests done, three by Zymo and one by EpiAging.  The results disagree radically.

      The result earlier thus year from EpiAging is that my DNAm age is 63.1 years, as compared to my calendar age of 86.5.  Here's the graph from EpiAging showing my result (red dot) vs the average:

    Here is a summary of my three Zymo tests.  It says that my DNAm age  is now 90.8 years, as compared to my calendar age of 86.6.

Original Sample ID

Sample ID

Date of Collection


Chronol. Age




Avg. DNAge























    In other words, the two results are in opposite directions.  I note that Zymo knew my calendar age before the tests, while EpiAging did not.  This variation between what are supposed to be reputable testing firms concerns me a lot.  I suggest that some anti-aging hero (not me) systematically has their DNAm age tested by every firm that offers DNAm tests (there are about six), not giving them your actual age, and compare the results.

    I should add that I think the Zymo test is probably the best replication of the original Horvath work.  However, I am not at all happy that according to Zymo I seem to have aged 6 years in 4 calendar years, even with all the senolytics we have been doing and the anti-aging supplements we have been taking. ☹️

14replies Oldest first
  • Oldest first
  • Newest first
  • Active threads
  • Popular
  • My EpiAging test result also gave an age about 20 years younger than my calendar years. 

      • Bob S.
      • Bob_S
      • 1 yr ago
      • Reported - view

      My TruAge (TruMe) test puts my methylation age 19.2 years below my chronological age of 92. A recent "bloodwork age" test ( offered free by the AI-based pharmaceutical company Insilico assigned me a physiological age of 55, 37 years below my CA.

  • JGC…why is Zymo “best replication of Horvath work”?

    I have done one epigenetic aging test (blood) so far, myDNAge;

    It came back 3.5 yrs younger than my CA of 56. I am strict keto, one meal a day (5 yrs), high daily exerciser, take injection testosterone (male), and a basket of supplements. I also donate blood every 7 weeks. I started Rapamycin 8 months ago (4mg/week, now on 6mg/week, planning for 8mg/week soon), but have not done a repeat epigenetic aging test as of yet. 

  • Twenty-year differentials between chronological and physiological ages are. I think, sufficiently significant that they warrant investigation. Are these results real?

    The two tests I've listed above... TruAge and Insilico... are the first antiaging tests I've taken, so the values I've gotten are baseline values. I plan to take other tests. But another question is: how well do they correlate with real life?

    I try to run a mile a day and to swim half a kilometer a few days a week. I'm having a bit of trouble with my right knee... presumably, a meniscus tear. Two years ago, I visited my integrative physician when he was moonlighting at the hospital. While I was there, he called in several other physicians to look at this not-quite ninety-year-old who had the bloodwork of a twenty-year-old. I was happily impressed. He hadn't mentioned that before. So there's some precedent for the Insilico results. There are a couple of other indicators. I have an early cataract in y left eye. It hasn't advanced in the past twenty years. I had advanced kidney disease in the latter half of Stage 3. My creatinine is still high for my weight and musculuture, but just within normal limits. 

    At the same time, the varicose veins on my leg have slowly progressed.

    The next question is: can these numbers be reduced further over the next six months or a year? I 'll be upping my game as new options present themselves.

    We'll see.

  • I am impressed. You have a good exercise regimen. Can you tell us something about your diet en supplement regimen?

  • yes, this type of personal anecdotal evidence is consistent with what Dr. Morgan Levine's group has found. Their recent preprint article published April 2021 found that there are issues in reproducibility of results — results varied widely when analyses were rerun with the same samples.

    The study reported deviations of 3 to 9 years from the same sample for 6 major epigenetic clocks. This is a big problem! That means if a treatment can reduce epigenetic age by 2 years, technical noise of 8 years can completely mask this effect. You might then conclude that the intervention doesn't work. 

    The pre-print article goes on to describe a novel computational solution to minimize random noise. For example, the PhenoAge had a median deviation for the same sample of 2.4 years (maximum 8.6 years). Applying the new computational model smoothed out those fluctuations — median deviations of the same sample were within 0.6 years (maximum 1.6 years).

    I talk more about this topic in an article I wrote about epigenetic clocks and what's commercially available: Hope that helps/contributes to this discussion!

    - Christina 

    Like 3
  • General Reply:

    The first comment has a diagrthat refers to an example .We are perhps the only company that does not require the divulgence of  chronoage in advance 

    Without knowing the full history of accounts mentioned it is difficult to comment as we do not know the follow-up and disposition

    Typically the  allowable variation is plus/minus three years.Moreover, we question such results and typically retest and resample at our expense.

    Thank You ,


  • To clarify  that diagram does not relate to that individuals test result .

    We follow-up diligently yet no mention in any in these few .

    We kake such comments seriously regardless of what agenda may possibly be motivating as as retests or resampling are not even mentioned.

    Thank You

      • JGC
      • Retired Professor of Physics
      • JGC
      • 8 mths ago
      • Reported - view

      isaac JOSEPH 

          I posted the diagram, which shows your chart of general average test results vs chorological age, to which my individual score and age have been added.  It shows that my score is about 22 years less than my chronological age.  My wife was tested at the same time and had a similarly low score.  Your company ASKED for our chronological ages when I placed the order, but I did not give them.

          If you want to give us free retests, fine.  Please specify how that is arranged.  I really wish that your results were accurate, but the Zymo test that I arranged for after yours indicates otherwise.

  • Dear Sir,

    I will try to deal with the issues brought up.

    Age Divulgence- We are one of if not the only company that does not REQUIRE this info for the test 

    1-We are taken aback that you do not mention that??!

    One of the consequences is that those who do  have this info  can retes prior to reportingt!!

    Most of our customers/subjects do not divulge their chronos We need to hear from the customer.

    We surely retest those few that have a substantive gap between Bioa nd Chrono ages

    We have had instances where we sent the new kits but never  received the return vials.Why?  There have been rare instances of  unfounded claims that were discredited  

    Generally  when we are provided with the relevant bar codes we can go forward with  resampling and testing on all credible requests.


    The diagram is just an example and cannot be referring to any individual customer as we do not know the Chrono as you yourself demonstrated. We are actually in the very near future going to make available the individual graph/diagram for customers who provide the necessary info

    Epi Aging  

      • MAC.
      • MAC2
      • 8 mths ago
      • Reported - view

      isaac JOSEPH “We surely retest those few that have a substantive gap between Bioa nd Chrono ages”

      Frankly, this is a very damming admission of the erroneous results generated by your offering. This gives me zero confidence. 

      I have only completed one aging test with DNAage, yielding 3 yrs younger. 

      We are all collectively still looking for a truly accurate, reproducible and correlatable aging marker. 

  • I have a particular bias on neurodegenerative disease prevention (mom has AD), and two recent studies found NO correlation between the various clocks and predicting accelerated biological aging and dementia risk.

    Accelerated Epigenetic Aging in Peripheral Blood does not Predict Dementia Risk

    "There is no evidence that peripheral blood-based accelerated epigenetic aging, measured using Horvath, Hannum, GrimAge, and PhenoAge epigenetic clocks, is predictive of dementia risk"

    Are sex differences in cognitive impairment reflected in epigenetic age acceleration metrics?

    "Using the ADNI cohort, we found that none of sex, cognitive impairment diagnosis, nor load of APOEε4 alleles (strongest genetic AD risk factor) were associated with epigenetic age acceleration (DNAmAge, Intrinsic DNAmAge, PhenoAge, or GrimAge)"

    Clearly, there is a disconnect between peripheral methylation and pathophysiological changes ongoing in the CNS.

  • As an OBJECTIVE scientist can you say with cerainty that a much lower BioAge  is impossible and damning??!!All the best 

  • Just joined group yesterday, so not up on all  postings re age testing, etc  However, generally any type of testing is based on sampling and averages.  The very early telomere length tests were like that, which is why I never did one  The issue is one wants to know the "age" / condition of the oldest part / organ / mitochondria.  That is the one that will likely cause illness / death.  An "average" is not very useful IMO.

Like Follow
  • 2 wk agoLast active
  • 14Replies
  • 280Views
  • 10 Following