There are four conditions which a designer must be aware of when putting system and setting together:
A: Many games do not require a new system. It's easy enough to adapt something else similar.
B: The vast majority of gamers do not like systemless settings books, because it means they have to go through the trouble of statting everything themselves, so why not just pick the novel or movie or whatever inspired them and go with it.
C: For any given setting there are at most a handful of systems which could work in an acceptable manner.
D: Most game systems are not license-able, or are expensive to license.
This four conditions lead to two results:
I: There are lot of clunky homemade systems out there that exist only to make the claim that the setting is a complete game. This is known as the "Technically Its Playable" Law.
II: There are a few freely or cheaply license-able systems with lots of third party setting support which doesn't always suit the strengths of the system, or which lie in the narrow band of "feel" which does. This is known as the "Everything Looks Like A Nail When The Only Tool You Have Is A Hammer" Law.
The system you choose effectively takes the place of the language you write in. By taking on a system, you become bound by the paradigms of that system, and must either bend your setting to fit the system, or bend the system to fit your setting.
If you take the first route, The Everything Looks Like A Nail effect takes hold and you get a certain sameness of flavor. If your setting requires only minor tweakage, all well and good - you have a satisfying setting for system X. If it requires major tweakage, you may end up with a complex gerrymander to express the setting concepts in a way the system can understand. The best fits are always settings which are intended for that particular system from the start, because the fundamental properties of the setting are naturally expressed in the language of the system.
By taking the second route, you have effectively created a new, divergent system with similarities to the base system, but differences which create new paradigms. If most of your setting fits into the base system, the differences are minor and can be picked up fairly easily. If most of your setting cannot be fit into the base system, then the divergence can be large enough that it is in effect an almost completely new system, whereupon you risk messing with the Technically It's Playable Law if you can't pull it off smoothly, and maybe you should have considered starting from scratch.
So it's always a good idea to thoroughly understand both the setting and the various options for systems, in order to assure the best fit and the least work. By a strange coincidence which isn't a coincidence, the people who can best understand fitting together systems and settings tend to be really good at creating their own systems. Which brings us back in a full, vicious circle.