Hmm, interesting. Not a topic I'm familiar with I'm afraid.
But poking around the literature a bit, I see
this paper which proposes the use of a "dual path" LNA, such that no separate switch is needed:
Looking at the circuit proposed, it's basically just two separate frontend LNAs, simultaneously connected to the input of a second stage LNA, with only one powered up at a time.
They're working on-chip at 100GHz, but I don't see why the same approach couldn't be applied in a board level design at 1.4GHz.
The presence of the inactive LNA would have some effect on the load impedance seen by the active one, and the source impedance seen by the next stage. But it should be possible to design a matching network in the middle such that everything sees the correct impedance when either one of the two LNAs is active. You would want to choose LNAs with suitable shutdown behaviour - ideally a high output impedance when off.