Had they used 02a to distill the covariates and applied that to 01, you would have gotten different p values / HR all around.
1. We don't know this technique was not in the SAP. 2. Doesn't matter if they applied it to 9901 and pooled as the Cox analysis is not important for that trial.