Guiding on a distorted star isn't such a problem in its self - no guiding software I have tried has ever had a problem locking on to and guiding on a illongated star for example. But the problem is how that image of that star behaves with respect to a normal/correct star in the centre of the field. The way I visualise it is if you imagine looking through a ball of glass that causes distortion, you pan across a field of view, objects at the edge move slower than at the centre, or in different paths. Whie that's obviously an exaduration I imagine the same sort of thing would happen when you're looking at a guide star on the edge of your FOV vs a star in the centre of your FOV, assuming your FOV is not perfectly flat and uniform. Of course in theory your stars are staying steady in the FOV so there's no movement to result in a different rate of movement in the guide star vs centre star, but I'd expect it to affect accuracy.
Disclaimer: This is just stuff I've thought about and figured would be hapenning when I've considered the same sort of thing on my scope, I don't know for sure if I'm right on any of the above.
Roger.