I will take a guess:
If you are talking about current THD caused by the motor itself (slight non-linearity of the magnetizing current), then THD will be higher when the motor is unloaded, because the non-sinusoidal magnetizing current forms a smaller portion of the total current when under load.
If you are talking about motor current THD created from the line voltage distortion... it's probably a lot more complicated. I can certainly think of a scenario when it appears that current THD would be highest when motor is unloaded. Let's say the voltage distortion consists of very sharp notches and spikes in voltage. When applied to the equivalent circuit, that high frequency voltage does not cause much change in the inductive magnetizing current, but causes a change in the load current. So increasing the load would increase proportion of input-spike-sensitive resistive load current which would increase THD. This result is the opposite of what was predicted above for motor is source of THD.