The human body already incorporates multiple "cures" for cancer or you wouldn't go a year at the rate cells multiply. But then the nature of cancer is that eventually a random mutation will come along that can beat any given cure, just as a million monkeys will eventually type out Mozart.
[UPDATE:] Note that I'm not saying that Thiel is right, only that his position is not a self-evident "several-orders-of-magnitude confusion of relative risks".
Exercise, with weights.
That reasoning only works given a bunch of conditionals which to most people including experts in the relevant fields seem pretty unlikely. Given that, it'd be great to read more about why he believes the things he does rather than what he's doing about his beliefs, as it's impossible to tell if he's got some interesting insight, or is doing something that only makes sense for him, or in fact is doing something that doesn't make sense at all without knowing more.
http://www.thelancet.com/journals/lancet/article/PIIS0140-67... https://www.jstage.jst.go.jp/article/endocrj/52/5/52_5_571/_...
If the levels are unusually low, then medically monitored treatment to raise to normal levels appears to have no negative consequences, only positive ones, at all, at least with a short examination of wikipedia. Even boosting up to a bit on the high side of normal is probably medically sensible, if the patient has symptoms that correspond with excessively low levels.
There are bro-science uses, for weightlifting and some other things, which seem to cause significant health problems for people who overdose. Its rather like insulin, if you're chronically low you're in big trouble, but if the stuff were widely available OTC you'd have idiots overdosing on insulin all the time because if a little is good them surely an immense amount is many times better. that doesn't mean proper doctor supervised use of insulin by diabetics is inherently wrong, just because of a little bro-science that isn't relevant to an individual's medical treatment plan.