Thread Reader
Morgan Levine

Morgan Levine

Nov 24
9 tweets

This got a ton of attention and some of the focus was on epigenetic age. Lots of people weighing in (mostly those with no actual experience). Here is my take on current standing of epigenetic age. Feel free to AMA. I promise I’m not trying to sell anything 1/

Bryan Johnson sold his company to PayPal for $800 million in 2013. Since then, he's been investing millions to reduce aging. In 2021, he reduced his epigenetic age by 5.1 years in 7 months (World Record) Here’s a breakdown of his “Blueprint” and my own experience with it: 🧵
Epigenetic age is based on DNA methylation. A large proportion of CpGs change methylation profile with age (some gain and some lose). Scientists decided to make indexes of how much the methylome has presumably changed. 2/
Basically the early versions were equivalent to saying “person X has a methylation profile of the average (e.g. 40 year old)”. Most of these were estimated using methylation from blood, but often the same algorithm can be applied across tissues and cells. 3/
Most 1st gen epi age measures strongly correlate w age (>0.9 in full age range). More importantly when looking at discordance between predicted and actual age (e.g. you are 48, but epi age predicts 45), this was somewhat predictive of future risk of disease and mortality risk. 4/
In 2018, we started developing what we call 2nd and 3rd gen epi age, meant to improve morbidity/mortality prediction. Again, these are mostly applied to blood samples, but some (like PhenoAge) “work” across sample types and in vitro. 5/
The newer gen epi ages are good health predictors—or at least as good as any standard lab tests currently used in clinical setting (e.g. LDL, HbA1c, creatinine, etc.). Now the caveats…6/
1.They do not capture all the dimensionality in biological aging (they are not equivalent). Things like senescence aren’t captured well in epi age. 2.We have no idea if change in epi age will equate to change in risk. 7/
3.The current versions (for now) only provide a single output thus they assume epi aging is one dimensional. 4.There are multiple algorithms and they aren’t all equal—it’s important to recognize differences. When talking about epi age, it’s important to state which version 8/
5.Many have fairly low test-retest reliability unless you are using the updated versions. 6.Current blood epi age is not always a great surrogate for epi aging across tissues/systems. 9/
Morgan Levine

Morgan Levine

Founding PI at Altos Labs | Former Professor at Yale | Researching epigenetic mechanisms of cellular and systems health | mom to a feisty girl
Follow on Twitter
Missing some tweets in this thread? Or failed to load images or videos? You can try to .