Replies to post #149662 on Biotech Values

[iwfal]

They [Roche] never reported result from sub population of squamous and non squamous [for the MetMab ph ii in nsclc].

This is substantively incorrect:

1) For one of the last interims (PFS essentially final) they reported non-squamous ITT PFS (HR=1.05 - vs the Met+ PFS HR of 0.55).

2) For the final they reported the OS HR in the non-squamous, MET+ subgroup and it was HR=0.47, meaningfully higher than the overall Met+ HR of 0.36 for the same period of time (i.e. the truly spectacular response was in the squamous Met+ population).

Obviously none of these numbers is stat sig due to the small size of the trial, and it doesn't map directly to ARQL's trial but nonetheless the MetMab data actually lean the other direction from ARQL's. Soooo, why might that be? ...

We already had this discussion before that literature showed majority squamous don't have Met expression at all, thus Met inhibitor should produce negative result for this population.

We had this conversation, but I would suggest this was not the result of it. The result of that previous conversation was that historical data indicates squamous does have fewer expressors, but the relative paucity of squamous met expressors in the ARQL ph ii trial appears to be substantially anomalous when comparing to the literature. E.g. taking your own cite (see #msg-76451506 ) it is clear that at the met expression level that captures 1/2 of the non-squamous population you still get perhaps 1/4 of the squamous population. I.e. a ratio of expressors of 2x greater in non-squamous vs squamous. Not the 6.5x difference observed in the ARQL ph ii. What I don't know is whether that was due to excess expressors in ARQL's non-squamous (vs historicals), or an absolute paucity in squamous (again relative to historicals). My guess is some of both and that will hurt the ARQL ph iii.

Finally note that all the historical data about squamous vs non-squamous met expression levels leans the same way - that the massive enrichment of met+ seen by ARQL in their ph ii non-squamous vs squamous population was anomalous. That said, the historical data is varied, not completely comparable and generally smallish populations. So again, I'll state that I think the odds of OS HR better than 0.80 for ARQL's ph iii are not particularly good (less than 50%) - but neither would I bet against it.