Presumably valve? Not if class A, there is substantial DC current flowing that will saturate the transformer, fizz, pop!
I have never tried it, one day, but for class B why not? The DC currents should be equal and opposite so cancel, leaving the transformer happy. A toroidal transformer will be better than an EI type for frequency response. Might even be possible to use a thin cutting disc to make a slot so usable on class A?
Have wondered why you can't use a standard transformer, and buck the DC current with an opposing current through another winding. Seems a reasonably simple bit of servoing, measure HT DC current and then provide the exact opposite? Will it work? Why not?