*** Updated Oct'17 - added Rintega as a Control Arm modelled
These are the results of my modelling work on Overall Survival (OS) for Northwest Biotherapeutics (NWBO) Phase3 trial. I did this analysis after NWBO presented new Phase3 trial information at this year’s ASCO event.
The results show a best theoretical treatment median survival range between 25 to 27 months versus 18 to 25 months for the control or placebo group. The trial Blended survival curve has a median 25 month survival.
By my model, there can be up to a 9 month treatment survival benefit. However, I am expecting a lower treatment survival benefit of around 5 months. In this trial patients cross-over to treatment at disease progression so Control patients can gain OS benefit if this treatment has efficacy in recurrent GBM patients. I expect this cross-over effect to be small and not substantial.
The actual statistical significance threshold is unknown for this secondary endpoint but I intuitively expect a 2 to 3 month treatment benefit will be adequate. The company has not disclosed the amount of alpha this endpoint has been allocated. The company has stated their trial’s endpoints are independently powered and that 0.02 alpha has been allocated to their primary endpoint of PFS. So their OS secondary endpoint should have between 0.03 to 0.05 alpha.
The Blended survival efficacy curve is a theoretical point where both treatment and control curves lie exactly on top of each other meaning there is no difference in efficacy. My Blended curve has a 25 month median survival which is unusually high for a GBM trial. To date, no large drug clinical trial has produced a patient survival figure higher than 20 months (Celldex’s failed Rintega trial). The SOC expected median survival ranges from 14 to 17 months.
Assumptions:
·
* 233 OS events reached in July 2017
* Last patient enrolled in Nov 2015
* 6 patients lost-to-follow-up
* 248 PFS events reached in Dec 2016
* 12 patients PFS evented post Dec 2016
* My Control group efficacy curve is derived from Stupp's Dose-Dense Study·
* 233 OS events reached in July 2017
* Last patient enrolled in Nov 2015
* 6 patients lost-to-follow-up
* 248 PFS events reached in Dec 2016
* 12 patients PFS evented post Dec 2016
* My Control group efficacy curve is derived from Stupp's Dose-Dense Study·
Other Control Arm KM curves modeled -
Disclosure - I am long NWBO. I do not short stocks
Any hope here or do we wait for infinite and dilution?
ReplyDeleteHi Dreamer,
ReplyDeleteI am holding and waiting for the Phase3 results publication and top-line results.
If the results come come out as good as I think they can the waiting will be more than worth it.
Dilution is a small risk before P3 read-out.
Good Luck!
Branko,
ReplyDeleteHave you considered the impact of removing patients with rapid disease progression on your analysis. Is it possible this is the main cause of the improvement? ( I am long because of your study and positive comments by lead investigators. However, I'm still concerned.)
Thanks,
Dreamer 2
Dreamer2,
ReplyDeleteI based my modelling analysis on a later and larger Stupp trial that accrued from 2006-2009. See https://clinicaltrials.gov/archive/NCT00304031 and see 2013 publication for it here http://ascopubs.org/doi/pdf/10.1200/JCO.2013.49.6968.
This is the Stupp dose-dense trial which used SOC v. SOC with a denser dosing regimen.
Stupp's Dose-Dense study excluded rapid progressors patients just like NWBO's trial has so no problem there with my analysis.
I remain very confident in the modelling work I've done. I have checked my mathematics several times and am satisfied with the assumptions I've used.
My uncertainty lies in how large the treatment cross-over effect will be. I don't expect the cross-over effect to be substantial but it's my largest uncertainty.
I am looking forward to final results and wish us both luck :)