The Nwell acts sort of like a JFET gate, fighting the P
inversion channel. More commonly it's referred to as
body effect and considered as a shift in VT, but the
JFET concept might be a more agreeable visualization.
It's the relation of well to source potential that the
remarks refer to. A transistor operated at (say)
VB=VS=vdd has the same VTeff as one with
VB=VS=vdd/2 - key being VB=VS.
With two transistors, same well, different VS this
cannot be the case and one must suffer.