I wonder if this variation of the QGA has ever attracted anyone's attention here besides my own. 1.d4 d5 2.c4 dxc4 3.Nf3 Nf6 4.e3 g6 5.Bxc4 Bg7 with the idea ...Nfd7, ...Nb6. It's rather Gruenfeld-like and I would say Alekhine-like in some respects as well. It's basically missing from the theory books (nary a word in Chess Stars, for instance), except that Neistat's old QGA reference has a pretty good chapter on it. It's doesn't seem to be played very much any more, either, but in looking at the lines I haven't been able to find anything that I would call a refutation.
One quite dangerous line is 6.Nc3 0-0 (6...Nfd7? 7.Bxf7+!) 7.e4 Nfd7 8.e5. Neistat cites a White win that arose from 8...c5 9.e6 fxe6, but Black instead has 9...cxd4!? 10.exf7+ (10.exd7 is quite good for Black) 10...Kh8. Black is in danger, particularly from White's running his h-pawn. However I analyzed both 11.Ne4 and 11.Nd5 and concluded that Black is OK (actually I also looked at Ne2, Nb5 and even h4). For example 11.Ne4 Nc6 12.h4 h5 and although Black has a fouled-up kingside, he has a nice pawn on d4 and good piece activity, with a knight coming to e5. White's f7 pawn will fall pretty soon, I think. Further instead of 8...c5 Black may be OK with 8...Nb6.
White more usually castles early, e.g. 6.0-0 0-0 7.Nc3 Nfd7 8.e4 Nb6 9.Be2 Bg4 10.Be3 Nc6 11.d5 Bxf3 12.Bxf3 Ne5 13.Be2 Nec4 (but not 13...c6 14.Qb3) as played by Smyslov against Evans back in 1952, I forget where.
The one(!) line quoted in ECO (and NCO) is 6.0-0 0-0 7.Nc3 Nfd7 8.Qe2 Nb6 9.Bb3 Nc6 10.Rd1 Bg4 11.h3 Bxf3 12.Qxf3. Now instead of 11...e6 as quoted in both works, Black has 11...e5 (I find this in Chess Assistant with a couple of games, and it is mentioned with "!?" and nothing more in NCO) 12.dxe5 Qe7. Black seems to be a little bit worse but nothing serious. 16.d5 also looks viable for Black.
White also has early h3 ideas, either with or without early castling.
Oh, I forgot to mention that Khalifman gives, of all things, 6.b4 0-0 7.Nbd2. He continues 7...Nfd7 but he overlooks 7...Nd5, which I think is almost certainly Black's best move. I looked at it and White is White, but I couldn't find much to fear for Black. For example 8.b5 a6. I can only speculate that Khalifman wanted to dispose of the Smyslov Variation without spending the time and space necessary to delve into its historical practice.
All in all, I think the line merits attention, particularly since it appears to be sound, reflects a modern approach to the center, and requires very little preparation.