at the end of the day there is really not much difference between the two
#defined CONSTANTS are simply "names" that are to be replaced (as in "copy&paste") by the preprocessor right before compiling. But in the end, they're just that: constants. There's no way a function or subroutine can access them and change them at runtime, because once your program got compiled, they got converted to (e.g.) numbers.
On the contrary, global variables can be accessed and changed by any subroutine. This may not seem like a big deal, but once your software starts to grow (like in hundreds of modules and thousands of subroutines) it can become a nightmare to track down every place you alter them.
Even worse: imagine you're working with some other programmers, each of you is hacking some modules, and you have some global variables that anyone can access and change. And suppose you realize a given global variable is not needed anymore, or should be renamed, or should be initiated to a different value... Probably your change would affect other people's work, and you'll have to agree upon new values / names and usage... Not an easy task ;)
That's why passing variables as args is always recommended (as opposed to global scoping), like @BrendanRafferty suggested, because it becomes apparent in which function is every variable used and/or modified.