I am puzzled by your question. The chain rule is a method for differentiation. Why would you expect to be able to use it to integrate?
Integration is the opposite of differentiation. Perhaps you are asking "why can't I reverse the chain rule to integrate this?".
The chain rule says that if x is a function of some variable, t, and y is a function of x, then [tex]dy/dt= (dy/dx)(dx/dt)[/tex]. To use it we have to multiply by dx/dt. To do that in reverse, which is "substitution", the "dx/dt" term has to already be in the integral, you can't just insert it.