It's called "tracking" and conformal coatings provide an answer. Cambridge Instruments made electron microscopes, typically 30kV machines. I didn't have much to do with the electron source, but had more to do with the photomultiplier (Everhart–Thornley) detector which we mostly ran below 1kV. I ran one a bit higher
I wouldn't presume to predict what Jeorg would say. His most impressive feats were in medical ultrasound (where I've also worked) which doesn't use high voltages).
On the subject of avoiding high voltages, could you split the drive to your electro optic modulator, so that one side went to down -2,25kV while the other side went up to +2.2.5kV?
An incidental advantage is that you balance the the stray currents driven into the ground and get quite a lot less electromagnetic interference. A carefully made balanced transfomer drive with a grounded centre tap can do that rather well.