I'm going to take a stab at this:
When driving a motor with PWM, you are essentially turning it on and off...while varying the percent-ON and percent-OFF. Doing this you can simulate a DC voltage (RMS voltage). However, during the OFF time the motor will de-accelerate, the heavier the rotor (core), the more resistant it is to this de-acceleration period. Thus lighter rotors will experience more stress (they do not). Using higher base-frequencies, the motor's natural response can act more as a filter to the higher frequencies, and is less effected.
I know someone who is a true motor expert (went to school for this stuff) and he would probably slap me for wording this the way I did. So, might want to get a second opinion