The current sink has been altered to a mirror-type, and the differential pair no longer has the trimming network the old VAS had. The negative voltage may be lower than shown here, it needs just to give room for the I-sink, that will reduce heat dissipation and allows for lower voltage caps in the filtering, all in all saving space and money.
About the cascode:
I found that at least 10volts should be available for the Vds of the lower MOSFETs, lower voltage slowed down an otherwise fast circuit, because low Vds increases the capacitance of FETs. I always thought cascodes eliminated the effect of the capacitance, but obviously not to zero.
For experimenting and versitality, switches (relays) may be used to bypass the cascode-stage, enabling direct comparison in sound.