Ok so here’s the aging paper everyone’s so excited about. Horvath et al found a way to reverse biomarkers of aging in rats. biorxiv.org/content/10.110…
The rats were treated with what’s described as a “plasma fraction treatment” from young rats. The authors aren’t specifying what it is.
The primary outcome was the score on methylation clocks, trained on the different patterns of DNA methylation between old and young tissue samples. Elsewhere, methylation clocks of this sort have been found to be predictive of mortality.
(In humans, few of these blood-based biomarkers are as predictive as functional tests of frailty or locomotor fitness, so I tend to be suspicious of them as endpoints.)
So, did this treatment work? Pretty dramatically, on the epigenetic clocks (except for hypothalamus) 
It does reassuring things to blood work (glucose & cholesterol & triglycerides down, HDL up).
The treated old rats did perform better on a cognitive task, learning a maze:
And they have way higher levels of antioxidants like superoxide dismutase. The relevance of this to diseases of aging is mixed as I recall, but more oxidative stress correlates to more chronic disease *sometimes*.
Lower chronic inflammation— this is also a good sign. IL-6 correlates with pretty much every chronic disease of aging and long-lived mouse strain.
Reduced senescence — I’m not crazy about using pictures here, they’re so easy to cherry-pick, but ok sure.
I can criticize the methylation clocks all day. They’re trained on chronological age not remaining time to mortality, so not what we care about.
They’re also novel — why? Lots of old, validated methylation clocks out there. Too easy to game by testing your treatment on your own invented metric.
Improving performance on learning mazes in aged rats is also not unique.
You can do it with substance P: cyber.sci-hub.tw/MTAuMTAwNy9iZj…
Or with etazolate: ndineuroscience.com/userfiles/Etaz…
Or with a MAPK inhibitor: ncbi.nlm.nih.gov/pmc/articles/P…
Or montelukast: ncbi.nlm.nih.gov/pmc/articles/P…
Basically, improving cognition in aged rodents, even all the way down to young levels, is not that unusual. And doesn’t always translate to humans. (Neuroprotective drugs have a really bad translation track record.)
I still think it’s a good sign! But keep in mind that lab rats are kept in the rat equivalent of prison. You can improve cognitive performance just by giving them an exercise wheel or toys or other rats to interact with.
Also of course there are lots of interventions that reduce inflammation, oxidative stress, and blood glucose and lipids. Dietary restriction being the most famous.
The “wow factor” on this paper seems to be from the fact that it claims reversal on epigenetic clocks. And epigenetic clocks are (IMHO) bad.
Even if you disagree, epigenetic clocks are *new*. The reason a bajillion other interventions haven’t been found to improve them isn’t because few things work but because few things have been tested on these new metrics.
My reaction is “ok, add this mystery intervention to the list of ~100 things that seem like they might do something to aging symptoms in mammals”. I’m not sure where Josh Mitteldorf’s extreme endorsement is coming from here.
