I don't have a problem with a little extrapolation and G has noted the need to make sure that you have a handle on the shape of the curve.
My concern would be that I probably would have no visibility of any other limitations or problems which might cause one some difficulty - temperature concerns (eg cooling), discontinuities in the engine output which just might bite in that extrapolated region. For this reason, I would incline to the airline NTO approach so we have at least some feedback from the OEM.
Would I have a concern, generally, with a bit of extrapolation ? Probably not. For example, I routinely used to takeoff nil flap in light twins, having done some sums re distance penalties.