I agree with "drDOC"'s opinion, I only have some more comments:
1. In general, cascoding input differential pairs is always to enhance PSRR performance, instead of open loop gain;
2. The bulk of every transistor in cascoded input differential pairs should be connected to its source, to prevent decoupling of Gnd/Vdd's noise.
3. If the tail current is set to be small, to make sure the input differential pairs working in sub-threshold region, to enlarge gain-bandwith, slew rate will be an important parameter. It will not only be reduced greatly, but also be asymmetric.
I've got such reference but it's too large to transfer.