To Botvinnik's: "All Rook endings are drawn" we used to add "Except the lost ones".
While it is difficult to say that a given opening is losing. I'd say that there are more solid, and more positionally dubious openings, but in different times the both lists are quite differently populated. I would not be surprised if in some 10 years The Sicilian would be listed amongst the outright dubious. Still, it is  fun to play it.

3 e4 is certainly the most sensible way to play chess. Remember Lasker's "Common sense in chess". White should have really strong reasons in order to prefer  3 e3 (why such a move, for god's sake!) to the obvious 3 e4. No matter what theory may say, 3...c5, 3...Nf6 and 3...Nc6 are always difficult to play against GMs. White's centre is always very restricting. Again, 3...e5 is most sound, but it leads to quite open positions and often to endgames - a lot of players (especially these who do not play 1 e4) do not feel comfortably in them. 
Of course, at a very high level, 3 Nf3 - 4 e3 could be more appealing as being more flexible, but the difference is too subtle. As GM Psakhis told me recently, below 2600 all openings are equally good.

Smyslov_Fan wrote on 08/24/05 at 01:46:57:

The main merit of 3.e3 though has to be that it cuts out those annoying 1.d4 d5 2.c4 dxc4 3.Nf3 a6 variations, right?

In return one is allowing in particular the 1.d4 d5 2.c4 dxc4 3.e3 e5 line. Sakaev and Semkov seem to think that 4.Bxc4 exd4 5.exd4 Nf6 6.Nf3 Bd6 is quite equal with a bit of accuracy from black, so I suppose that offers the question of whether (a) that's true and whether (b) white has anything at all after 3.Nf3 a6.

I love 3.e3.  The independent lines favor White while the transpositional lines are just fine too.  And I have had rated games that go:

1.d4 d4 2.c4 dc4 3.e3 b5? 4.a4 c6?? and my opponent usually resigns after surviving to move ten.

This has happened against players all the way up to Class A level (up to 1985 USCF), but I haven't had an expert fall for this yet.

Sorry but if you want to avoid the isolated pawn on d4 quit playing the queen's gambit! I know that immediately e4 is not a good as Nc3 first. Unfortunately I don't remember the analysis but I think there are some problems with defending that pawn.

I did a quick search of my database  from 1998-2003 and found the following.
Dreev-18 games
Karpov- 8 games
Loek van Wely - 13 games
Khalifman - 6 games
Kasparov - 3 -games
 
Hope that helps