For nonconcave functionals, does there exist estimates similar to Evans-Krylov theorem  in chapter 6 of  Fully nonlinear elliptic  equations  by Luis A.C affarelli and Cabre? Perhaps there is a counterexample.