Wahey, a question I know the answer to, or did in 1979. At that time the problem with this line was supposed to be 5 Bxc4 e6 6 h3 Bh5 7 Nc3 Bd6 8 00 Nbd7 9 e4 e5 10 g4 (these moves being made in some random order, but I think this is the right position) followed by creaming Black as in Littlewood-Someone, British Championship 1979, and Somebody-Norman, Aaronson Masters 1979, subsequently published in the Streatham & Brixton club magazine.  The idea is that 10...Bg6 is met by a lot of captures on e5, and f4, winning a piece. At least it doesn't win a piece, because Black can escape somehow, but he suffers in the process. It is of course possible that theory has moved on since that date without me noticing, so the above might be total rubbish, but it would be somewhere to start.

1. d4 d5 2.c4 dxc4 3.Nf3 Nf6 4.e3 Bg4!?

I consider to take this line to my (black) repertoire. And then only in the 2.Nf3 Nf6 3.c4 dxc4 move order. As I play the Slav anyway I would have no problem with 4.Nc3 c6 or 4.Qa4+ c6. In fact I would prepare only this line of QGA (for reasons of variation or to avoid the Slow Slav).
Before I delve too deep into the body of theory (it seems to be very tactical), can someone please tell me the general assessment and critical lines ? Someone who plays this as Black regulary or has prepared for this as White. It would not bother me if White could keep a very small advantage by accurate play.

Any opinions, chessfriends ?