Antioxidant Genes, Hormesis, and Demographic Longevity

Interestingness:5

By Robert Arking and Craig Giroux in the Journal of Anti-Aging Medicine, June 2001, pp125-136. doi:10.1089/10945450152466170.

It didn't end up being as interesting as expected, since they skipped/assumed the question of the effect of late-life mortality deceleration or even decrease being real in humans.  They mostly work on fruit flies, and that's where most of their data comes from.

I still find overarching theories interesting though, so it was a decent read, even though there isn't much more there to summarise than what was on the abstract. 

Their hypothesis can be summarised as follows: there are some stressors that will kick what they call the antioxidant defense system (ADS) and heat shock proteins (hsps) into action and will wipe some "aging" off the body, thus leading to lower mortality.  This activation is semi-locked by an epigenetic mechanism, thus leading to a clustering of people with lower mortality separate from the main cluster.  They hypothesis that testing young people to see whose ADS get upregulated easiest will tell you which people will leave longer.

One big issue with the paper for me is their evidence that upregulated ADS and hsps lead to longer lifespan. Their data is from fruit flies, but they mention mice, in which just upregulating CuZnSOD (superoxide dismutase) doesn't lead to longer lifespan.  In mice, they go by the suggestion that since calorie-restricted mice have upregulated CuZnSOD and catalase, that these two are important reasons of the lifespan extension. I suspect the situation in humans to be even fuzzier.

Another bit of the paper I found interesting is at the beginning where they list possible theories for the decelerated/decreased mortality effect: one, that this is just part of the aging process; second, that this is a predicted effect of the reliability theory I'm fond of by the Gavrilovs (I find these two explanations to be compatible), and third, that the population is genetically heterogenous, so each subsection would have its own Gompertz gradient.