If Black wants to try to keep the game in Slav-like variations (i.e. not playing 4...f6 which has only been played a few times) then the main alternative to 4... Nd7 is 4... c6 -- I think as Black this looks a bit more solid than 4... Nd7?

But there are still lots of very interesting possibilities, for instance, 5 Qb3 Qc7 6 Nxf7!? Kxf7 7 Qf3 e6 8 g4 with a very unclear position where it looks like White has some chances for a small edge.

Given that 2... Bg4 has been played by lots of 2600+ and 2700+ GMs it's interesting that these positions remain unexplored.

This move (5 d4) isn't given by Mikhalevski in GM Rep 19 and I can only find one game with it on database.chessbase.com

It seems fairly interesting for White, for example:

a) 5... c6 6 cxd5 cxd5 7 Nc3 Ngf6 8 f3!? Stockfish eval approx +0.5

b) 5... e6 6 Qb3 Rb8 7 Bg2 Stockfish eval approx +0.5

c) 5... e6 6 Qb3 Ngf6 7 Nxd7 Qxd7 8 Qxb7 Rc8 9 c5 and it seems like White can consolidate the extra pawn, eval approx +0.7

d) 5... Nxe5 6 dxe5 dxc4 7 Qa4+ c6 8 Bg2 ... Black might equalize here with precise play but at the moment White seems to have free development and space ... eval approx +0.3

e) 5... Ngf6 with the idea if 6 Qb3 dxc4 7 Qxc4 Nxe5 8 dxe5 Qd5= (also 6... c5 is interesting) ... White may prefer 6 Bg2  when after 6...e6 7 Qb3 looks better... Silicon still prefers White at around +0.5

Perhaps best for Black is to aim for a reverse Benoni after 4... f6 5 Qa4+ c6 6 Nf3 d4 that looks about equal, but definitely is a different type of position than Black might have expected.