I think that's not possible. You would have to use scripting to generate another .odb with certain results.
If you could apply the loads in perturbation steps instead of loadcases, then you could use *Restart to do it in separate job and get separate results.