RE:RE:RE:Bicycle Therap. - encouraging in ovarian "the best of both worlds" is definitely what we want them to have all the way up to any decision-making points, that's what they are giving themselves. The price for that is enrolling some SORT1 negative patients who likely won't respond.
On "evaluable" here is what the RECIST rules say. (I get hung up on the rules)
https://project.eortc.org/recist/wp-content/uploads/sites/4/2015/03/RECISTGuidelines.pdf
"Trial conclusions should be based on the response rate for all eligible (or all treated) patients and should not be based on a selected ‘evaluable’ subset."
This rule kicks in for Phase2.
Like I said I think what they have done here is a post-hoc analysis. I would expect their definition of "evaluable" is post-hoc as well. After all by taking all-comers they would have been hoping that the responder rate was high for the unscreened population in which case the whole population would have been "evaluable". I don't think they've broken rules and I don't hate they've done this here it's just important to understand exactly what they are doing. Sometimes these post-hoc analyze can be very deceptive, not here but they where in CYDY's case.
I think the take home is they've now pivoted from unscreened population to a screened population and that they've done that based on what the data has told them. You can maybe see why.
Unscreened ORR 2/21 = 9.5%
Screened ORR 2/9 = 22%
What they have to do going forward is prove that by presetting the screening protocol and only enrolling target positive patients that they can keep that ORR. One way to think about post-hoc analysis is a trawl through the data looking for patterns. The test comes when you apply the new rules and see if the pattern holds up. In many cases it doesn't. In this case there's a strong logic on their side and a good reason to think it will but you never know.
(Apologies if this is getting too nerdy)
Wino115 wrote: Helpful additional comments and, from what I've seen, almost all the targeted cancer trials eventually get to understanding the best baseline characteristics for optimal performance in a patient. Many have this same construct of segregating the "evaluable" results. What I take from this also is that THTX, in doing the all-comers as FDA requested, will have the best of both worlds. Maybe they can talk about the whole population in some of these cancers and hopefully have decent commercial kinds of RECIST numbers. Then they can also provide that extra level of segregated results for the group they all think offers the optimal type of patient --the "evaluable" cohort. I'm good with all that. I hope THTX CEO/IR/CMO all see these kind of reports and data analysis and understand that more data is better than less data and that the market can understand the nuances around safety, efficacy, dosage, baseline expressions, etc... It is completely possible to be highly analytical, transparent, positive (if #'s support it) and commercial while still being completely truthful and ethical in how you discuss a trial and the data. They shouldn't think the reader or analyst knows nothing --especially if they want to attract these type of top-notch oncology analysts. They'll want all that level of detail so your best to lay it out and provide an in-depth value-added analysis of all the pros and cons as you see it, and next steps. It's par for the course to do that.
qwerty22 wrote: Here's the clinicaltrial page.
https://clinicaltrials.gov/ct2/show/NCT04180371?term=BT5528&draw=2&rank=1
So just to be clear the splitting of their enrolled population into EphA2+ and - is a post-hoc analysis as it's not mentioned as part of their protocol in the original trial design. It's obviously a logical thing to do but it comes with some caveats. For example because they didn't state a cutoff point in advance they are free to set that however they want which might skew things to a more favourable reading of the data. Having said that the logic is solid so I'm not complaining.
As I've said before there is no data atm that tells THTX how to handle screen for SORT1. No idea of what the cutoff point should be, no idea how exactly you should do the scoring. In a situation where you have no data to guide you they should follow Bicycle's path, enrol all-comers, then they are free to work out exactly what is the best way to score each patient and slice and dice the group to get the best understanding (and also present the data in the best way). As I said hopefully the data from the unscreened population stands up to analysis without any post-hoc analysis but if it needs it then so be it. What you have to recognize with these Bicycle results is they shot for the whole Ovarian population, which was the right thing to do. In presenting the data the way they do now it seems to me they are signaling their future trials will be focused on screen target expressors, that's still a good outcome. This is how THTX's data should play out once they have enough patients to support this sort of analysis. If the drug has promise they will eventually go down one of two paths. Either unscreened or screened, the patient data will determine this not assumptions and guesses
Essentially how bicycle has handled this is what I want. Target expression analysis once you have a dataset that makes this sort of analysis meaningful. Not what JFM wants which is screening now. Their reading also shows how it's still possible to get meaningful data without pre-screening even when the population only has a 9/21 hit rate for the target. I think most of THTX's cancers are expected to have >43% SORT1 expression rate. This analysis by bicycle just confirms to me how JFM is spreading unnecessary worries about the trial design. The design is just fine, as it has always been the case if the drug works well then good data can emerge from a trial even when optimizing responders to the max has not been followed.
I expect JFM is going to say this vindicates what he's been saying :)