on variable and primitive cohomology of a hypersurface in a projective space

Question

I have a smooth hypersurface D in $\mathbb{P}^n$: in many books about Hodge theory (as the ones of Voisin and Carlson) they take for granted that the primitive cohomology of D is equal the variable cohomology of D. (the variable cohomology of D is the ker of the gysin map $\gamma: H^p(D,\mathbb{C}) \rightarrow H^{p+2}(\mathbb{P}^n \setminus X, \mathbb{C})$ ). it must be simple, but i can't see why