You can't reliably. Hell, I didn't even mention the fact that not only is power compression not readily predictable, but the difference in power will have a different affect on the change in Re which will affect the "warm" TSP, and the difference in excursion will change the "large signal" TSP due to difference in the nonlinear behavior......both of which combine to affect the "real world" response to something other than what WinISD modeled to begin with. So it's not just power compression busting your balls on this one. There are many factors which will change the outcome, none of which are reliably predicted without either your own Klippel machine or a hell of a lot more information from the manufacturer.