Protocol dictates the statistical analysis. Change the protocol and you change the statitstical analysis requirement. NWBO clearly did that as there was never a recurrent GBM arm, they tossed out the PFS, so you have required a totally different statistical analysis.
From SPIRIT, the association that defines best practices in clinical trial protocols:
"the protocol is defined as a document that provides sufficient detail to enable understanding of the background, rationale, objectives, study population, interventions, methods, statistical analyses, ethical considerations, dissemination plans, and administration of the trial; replication of key aspects of trial methods and conduct; and appraisal of the trial's scientific and ethical rigor from ethics approval to dissemination of results."
"Many protocols for randomized trials do not adequately describe the primary outcomes (inadequate for 25% of trials), treatment allocation methods (inadequate for 54% to 79%), use of blinding (inadequate for 9% to 34%), methods for reporting adverse events (inadequate for 41%), components of sample size calculations (inadequate for 4% to 40%), data analysis plans (inadequate for 20% to 77%), publication policies (inadequate for 7%), and roles of sponsors and investigators in study design or data access (inadequate for 89% to 100%)"
SAP is an intregral part of the protocol and changes when the actual trial design changes,
AI
Market Data 
Markets 
