I'm not 100% sure what you are trying to do sorry. Are you trying to ensure that processing steps that are common across sessions are only conducted once? If so, Nipype will do this automatically. After you set an iterable, only nodes that depend on that iterable are actually run multiple times.
Or are you trying to combine the processed data from all sessions for each subject and perform some analysis on what is changing across sessions? Check out the JoinNode http://nipype.readthedocs.io/en/latest/users/joinnode_and_itersource.html