I&Q mixer sharing the same transconduct stage

Usually, in the ZIF or LIF design, I & Q mixer use different transconduct stage. However, in recent several papers, I & Q mixer share the same transconduct stage. Apparently, it will increase gm for each mixer branch and result in low NF or high conversion gain. Anyone knows any bad side of this topology? thanks.

can you upload the paper here
I think if they share the gm stage, then they are actually passive mixers