Why not? If you calculated the orbit assuming instantaneous attraction, you'd get the Newtonian result. If you calculated the orbit using virtual gravitons travelling at the speed of light, you should see e a slight lag in the pull from the star due to the finite speed of the gravitons versus the momentum of the planet in the orthogonal direction, because the planet is tending to move away from the star ever so slightly. Granted, this is just a hueristic. It would be interesting to do the math and, ignoring the GR field equations, just use a Feynman "virtual graviton" model to see what the orbit looks like given the lag from the speed of gravity.