Replies to post #226464 on Biotech Values

[iwfal]

FGEN

Remind me if you think the worse results in the population is because

A) FGEN created this subgroup post-hoc and therefore selected a line that made the >15 look better

B) This sicker group had differential and higher dropouts in the placebo arm, specifically bc they weren't on treatment (ie if you analyzed the endpoint as a combination of (MACE or dropout) the worse results would disappear

C) Some underlying real biological effect

D) Other

#A and #C (I am dubious about the dropout theory for a variety of reasons). And #C is really about the fact that renal drugs that improve GFR usually have negative effects at really low GFRs. E.g. the most common treatment for GFR currently is ACE/ARB - and such treatment is often halted once GFR gets low enough.

Whenever they talk about the "totality of data", I worry they won't answer those questions with direct answers until the FDA formally weighs in. Feels like they trust the FDA to accurately/fairly incorporate the totality of data, but not the street, so the B) scenario above.

I wouldn’t count on the FDA to leave such totality questions off an Adcom briefing or off the formal reviews. And I am more cynical that hidden data is almost universally bad data.