Of course they can classify patients after data lock. That's what sequencing the tumors provided as this wasn't done previously because when the trial began the scientific community was not even aware of the IDH mutation.
By doing so they did not add new patients to the trial.
They added metadata that describes them accurately and allows a more precise understanding of DCVax efficacy.
For the FDA yes, for a publication no. Reviewers can and have asked for additional analyses and data gathering in the past. If NWBO wants a publication first, they may have to do some post hoc analysis, not for purposes of approval but for purposes of publication.