This one has a reasonably settled answer, so here it is. Quarterly for the first year is convention rather than evidence, and it is defensible for a simple reason: it is roughly the interval over which HbA1c becomes informative again, since it reflects about three months of glycaemia. After the first year, and once doses are stable, annual is reasonable unless something specific is being followed.
Correct me if the detail matters more than I have assumed.