TITLE: High dimensional generalized Poincare hypothesis without the h-cobordism theorem?
QUESTION [7 upvotes]: The generalized PL Poincare hypothesis states that in dimension $n$ there is a unique PL manifold that has the homotopy-type of $S^n$.   It's known to be true in all dimensions except perhaps $n=4$. 
For $n \geq 5$ the standard argument uses rather explicit handle manipulations to come up with the proof (proving the h-cobordism theorem in the process), which is somewhat spiritually-similar to (although more complicated than) the proofs in dimension $n=2$.
The proof in dimension $n=3$ is very different than the above two cases.  
But I wonder, perhaps there is a proof that avoids the h-cobordism theorem, perhaps there is a more direct proof?   Has there been much discussion of this in the literature?  
One thought would be to find an appropriately simplified proof of the Farrell fibering theorem (when a manifold fibers over $S^1$), one that perhaps allows you to reduce $S^n$ recognition into a homotopy-unknot recognition problem. 
I imagine back in the 60's and 70's there was some discussion of these topics but I wouldn't know where to look.

REPLY [2 votes]: The Poincare conjecture predates the h-cobordism theorem, and the original proof of it in Smale's paper does not prove the h-cobordism theorem (though the handle manipulations in it are what inspired the h-cobordism theorem, so this is maybe not a good answer).
A very different way of proving the Poincare conjecture is Stallings's proof using "engulfing".  A nice modern source for this is Chapter 9 of Ferry's notes here.