but the popularity of systems like Warhammer Fantasy rpg, Runequest, Call of Cthulhu and nWoD (especially with mortals) attest to the fact that there are those who enjoy the challenge of this type of playstyle. It may not be for you but, there are plenty of people who feel snatching victory when the odds are against you is more exciting then winning when everything is stacked in your favor.
I'm just trying to figure out exactly what that 'playstyle' entails.
Snatching victory when the odds are against you is all well and good, but I can achieve that in D&D. I don't know why you'd need to tinker with anything to achieve that.
So what I'm asking is: "What do you REALLY want to do?"
If it's just make more dead PC's, my advice would be to concentrate on character creation rules and on the "dungeon grind" atmosphere. Make creating a character so fast that another one can be ready after this one dies, and eradicate any individual personanality from the story. These characters don't need history or backstory or goals beyond "go into the dungeon." You could also take the concept of a "spawn point" and have dead characters re-appear at a designated point, ready to challenge the dungeon again. The spawn point could even be random, or linked to some "treasure" throughout the dungeon.
The flaw I see with this is that it almost works better as a board game or a videogame than as an RPG. You're not playing much a role when you're playing "Fighter #50," and the success of the dungeon looses context.
But if you want more dead PC's, that could still be a fun game.
If it is to make PC's fear combat more, then we're going into some more subtle manipulations of the rules. You don't actually want them to die a lot, but you don't want them to be resorting to combat, really, at all. In this case, I'd shore up the social encounter resolution system to make sure you can 'win' without NEEDING to get into fights, and then make those fights not necessarily more deadly (because we're not really trying to kill them), but more costly. Clever characters should win out over strong, but I shouldn't have to be a clever player to have a clever character. I should be able to outsmart my foes even if I can't outsmart the DM. Stealth would seem to be key, too, so all characters should have options for NOT engaging in combat. Healing takes longer, leaves scars, doesn't come from magic. Old wounds haunt a character's entire career. You might want to replace the hp system with a more direct structural damage kind of system (True20's combat could be co-opted for this purpose). Reducing the damage threshold is a pretty useful way of doing it, but instead of death, perhaps the failed Fort check causes a permenant disfigurement.
The flaw I see with this is that you really need very good stealth, social, and 'puzzle' encounter resolution mechanics that don't depend on player skill as much as they depend on a die roll and the character's implied skill. This also works against one of 4e's goals: to make combat fun and fast-paced. You'll be kind of working at cross-purposes to the majority of the 4e game if you try to make combat less fun and more troublesome.
If you want combat to be a less valid option, that could be a fun game (assuming that there is a lot of attention paid to noncombat resolution).
If you want to make the PC's struggle against overwhelming odds in a world where 'hope' is the last thoughts on the mind of the populace, but the thing that PC's are trying to restore, you don't need to tinker with mechanics much, if at all. Ravenloft and Dark Sun from 2e got that feel, and that's more of a setting consideration than a mechanical consideration. It's more a choice about how you fill your world rather than how you run a combat.
So reduce it down to it's most basic element: what are you *trying* to accomplish. Then we can choose designs for that accomplishment rather than a vague and ill-defined 'feel'.