Have you measured the distortion? In spice model of L'amp, 2SJ28 produces more 2nd harmonics (more air?) at lower current around 20V, but I don't know if it reflects the reality...
I have divided the original Sony 2SK82 / 2SJ28 datasheet and I rotated one of 2SJ28 for easier comparison that is posted in the attached images.
In the past, using the Locky curve tracer , I made the reliefs of the curves for both the 2sk82 and for 2sj28, that you can see in pdf.
Unfortunately, due to low voltage and current value permission from this curve tracer, it is not well adapted to assess completely power devices such as the V-FET in question.
However indicatively, in such fields of values one can infer that the 2SK82 family curves are in a sense in the form of a tulip, while those of the 2SJ28 are more fan-shaped: it is' likely that for low supply voltages, the latter generate more second harmonic distortion, but this is not true at all, and changing the operating point results can be opposed.
In any case, when I designed this circuit, I wanted to make a complete amplifier, so I asked myself the question if were better to use a power stage with 2sk82 or with 2sj28.
I estimated that using the latter, became more simple and natural interfacing that power stage with the input stage using N channel devices.
This would allow you to use the only signal Vfet that was still available commercially, the 2SK79 (also 2SK63) and to be able to assemble a whole amplifier with all V-FET gain devices.
Also with the same logic you could be used alternatively in the preamp stage tubes and nuvistor (they are also N-channel). To me it seemed a huge advantage, not to you?
Those who only plans a final state does not care about this, and L'amp is just a final stage, but this one is a complete amplifier (integrated).
Best regards.