I’m honestly tired of our paper being framed as if we compared rapamycin to one or two other interventions, saw a clock move, and somehow got a Nature Medicine paper out of it. @mkaeberlein That is not what the paper says or does. Rather the simple quote from @EricTopol or measured commentary from @prof_horvath was perfect! That being said, Matt, we’ve discussed many of these points in person and over a podcast, so it is frustrating to see the same arguments repeated publicly while large parts of the actual analysis are ignored. At this point, I think it is worth going through your critique point by point, because several of the conclusions you are drawing simply do not follow from what we did or what we actually claim in the paper. 1) Circularity We do not say: “we are looking at 51 interventions to test responsivness, and now judge the interventions by the same clocks” That would be circular. We are asking the first half of that question, YOU ARE INSISTENTLY EXTENDING IT TO THE SECOND HALF! The actual question in the paper is much narrower: do these biomarkers respond to perturbations in the expected direction, by how much, and with what consistency? For which we curate a list of 51 putative longevity interventions. Now, we show all of these 51 interventions when combined decreased epigenetic age wrt to controls (which stayed at zero).Before doing the responsiveness tests we need to be sure we curated the right interventions and of the 51, 68% decreased epigenetic age and 37% significantly. This is critical to see before we start testing responsiveness systematically across them! That one sub-figure (8 per figure) in one figure (6 figures) where we show this "sanity check" (1 out of 48 sub-figures) is what you are so stuck on while completely ignoring the remaining 47 figures that talk about the responsiveness and consistency of the biomarkers and have nothing to do with the interventions! I almost feel you are the one cherry picking the rapa data (which btw is 1 line out of 51 in those 48 sub figures = 1 out of 2448). I really wonder who is cherry picking here and being circular! I for sure know neither me nor our paper is! 2) Rapa and Exercise (since you are so insistent on talking about individual interventions!) The PEARL study (data used by our Nature Medicine paper) used a low-dose regimen for two years and showed NO OVERALL CLINICAL BENEFIT (PS this is also true of @BradStanfieldMD study). The clocks also did not show a benefit in line with the original study, SURPRISE SURPRISE! That is the data, and we report it. I am not claiming that this shows rapa does not work. The clocks may be insensitive (I doubt they are, given they respond significantly IN RIGHT WAY to 19 other putative longevity interventions!), the particular dose/duration of rapa may have been ineffective, or both may be true. But focusing on rapa alone, while ignoring the full analysis of 51 interventions, is selective and rather the circular extension we arent doing but you are! This is the exact cherry picking and circular logic you are against but are then perpetuating! Remember, this is a study of 51 interventions not 1! But since, we are here, lets also make something very clear across 16 clocks RAPA SIGNIFICANTLY INCREASED EPIGENETIC AGE while EXERCISE DECREASED EPIGENETIC AGE BUT NOT SIGNIFICANTLY. The two are very different results. Again, I do not stand by this, saying exercise more, I am horrible at going to the gym, or not to take rapa, it is not an approved longevity intervention anyway, rather I will say this is a study of responsiveness of epigenetic clocks and across 51 interventions and the clocks responded systematically, but if you so insistently choose to cherry pick and use the circular logic (which I do not stand by) then RAPA INCREASED THE CLOCKS! 3) “The studies are too heterogeneous.” Yes, they are heterogeneous. That is the point. We analyzed 51 intervention studies across different populations, durations, clinical settings, and intervention classes to ask whether biomarkers show responsiveness across the putative longevity interventions, not to make an apples-to-apples ranking of intervention efficacy. The latter is a seperate question.And the results are not just rapamycin and exercise. We see favorable changes in settings where one might reasonably expect them: Mediterranean and low-carb diets, caloric restriction, metformin, smoking cessation, and ART in people living with HIV, among others. 35 interventions decrease epigenetic age (across 16 clocks) and 19 signficantly! We actually do not look at specific interventions in this way because that does not prove any of these slow aging since clocks arent a surrogate endpoint (we are attempting at proving they might be. prognostic -> responsive -> surrogate). That being said if you are so insistent on looking at specific interventions (which I again iterate we do not in the paper, refer my circularity point) it is not reasonable to ignore the other 49 studies and define the paper through two interventions. (One of which you preferentially seem to like) 4) “Epigenetic clocks have error/variance problems.” If these biomarkers were so noisy that they were useless, they should generally respond to nothing, not show patterns across ALMOST FOUR DOZEN PUTATIVE LONGEVITY INTERVENTIONS. (Notice I use putative since we dont know if they are actually extending lifespan or healthspan but might based on animal models or observational data, thats how we curated these studies). Measurement error reduces power and attenuates effects. It does not create consistent intervention-associated changes out of thin air. This is also why we have separately studied technical and biological reliability. Reliability and responsiveness are different properties. Both matter. One is not an argument for refusing to examine the other. Oh and FYI, we only know of the limited noise in epigenetic clocks because they have been measured! None of the other biomarkers like proteomic clocks, transcriptomic clocks or even for that matter the clinical chemistry biomarker clocks have this data! Atleast we know what we need to fix and we are fixing it and I have had long talks with @RyanSmithEpiAge and @VarunDw about this at @TruDiagnostic but without even fixing the limited noise issue the epi clocks are doing really well when it comes to responsiveness and prognosis across the board! Can you say the same for other readouts? I dont recall your bio age error video having tests from any proteomic clock companies or Clinical biomarker clocks. Do you have a bias here? Just wondering! All said and done, I apologize if this came off as too strong, I believe I am a bit frustrated with the narrative of what it has become and the effort we put into this study to show the clocks are responsive to putative longevity interventions is being diminished. All in all, I know this, you and I are both working towards the same goal of improving the field and such critique and clarifications are always healthy! Love to see you at the next conference and talk more!

While it’s too early for anti-aging prescriptions, this breakthrough supports repurposing existing drugs for longevity, potentially speeding up access & safety validation. 👏 Kudos to @VarunDw & the team for this milestone in #LongevityScience! #AgingResearch #BiologicalAge #Ozempic #Epigenetics

This week will be all about pre-prints, kicking off with my debut as a corresponding author on a paper titled, "Epigenetic aging biomarkers are responsive: Insights from 51 longevity interventional studies in humans." This study is the first of its kind, bringing together clinical trials and intervention studies focused on longevity to evaluate which interventions might systematically reduce epigenetic age. It examines 110 epigenetic aging biomarkers to identify those consistently responsive across various interventions. We hope this work encourages the aging biomarker field to rigorously test these biomarkers for responsiveness—since not all DNAm aging biomarkers respond to interventions, even if they may be prognostic. Additionally, it aims to shift the field toward using multiple responsive biomarkers together to assess whether an intervention genuinely reduces epigenetic age, potentially impacting healthspan or lifespan. Interventional study DNA methylation data for this study was compiled at @Yale in the Albert Higgins-Chen lab in the @YaleCBB program using publicly available resources such as @NCBI GEO and @embl as well as private sources such as @TruDiagnostic , thanks to @RyanSmithEpiAge ,@VarunDw and Natàlia Carreras Gallo! And this was a huge team effort which could have been complete with help from support from my amazing colleagues and co-authors @Dansb95, @JessicaKasamoto , Jenel Fraij Armstrong, John T. Gonzalez, @MarkovYaro and Ahana Priyanka! Check out more details at the link below! biorxiv.org/content/10.1101/…