You could try a simple ac sweep, with ac magnitude of 1 on the power supply for a small signal measurement of PSRR.

Then look at the ac magnitude on the differential output net. In an ideal schematic, this should be very low. For a more realistic PSRR you'll need to include layout parasitics and mismatch (via monte-carlo), both of which will cause imbalance and hence less common-mode rejection of the supply signal to the differential nets.

For accuracy at higher frequencies, your test bench should include effects such as loading, ESD, bondpads, packaging, pcb etc.

cheers
aw