> We only seem to do it when a drug is initially declared as safe; introduced into the market; and then we find that it has side-effects that clearly correspond to some subpopulation.
This exact thing happened with another MS drug: natalizumab[0].
It was originally approved in 2004, but later withdrawn in 2005 due to some cases of Progressive multifocal leukoencephalopathy (PML), an opportunistic brain infection that is often fatal. The manufacturer later identified a specific antibody for a virus (JC Virus[1]) present in ~50% of the population which correlated highly with incidence of PML after 1-2 years on natalizumab (~1% chance of PML vs ~0.01% of the normal population). tl;dr the blood-brain barrier maintained by natalizumab also prevents the body from killing the JC Virus in your brain, causing inflammation and eventually death or serious permanent disability.
It was then reintroduced but now patients must undertake regular blood screening (every 6 months) for JCV antibodies, and regular (~12 months) MRI scans to monitor for inflammation (though this is pretty standard with MS these days to look for changes in lesions).
My wife has a rare type of MS (tumefactive MS, ~3000 people in the US have it) and has been on natalizumab since diagnosis about 3.5 years ago with no evidence of disease progression. The story would've been wildly different if we were born 20 years earlier before these treatments came about, it's honestly amazing. Without modern treatments it's very possible she would've been confined to a wheelchair within 10 years of diagnosis, but instead her symptoms are only really noticeable upon examination by a neurologist.
[0]: https://en.wikipedia.org/wiki/Natalizumab
[1]: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3128336/