Two-beam stimulated Raman scattering (SRS) has been used in diverse label-free spectroscopy and imaging applications of live cells, biological tissues, and functional materials. Recently, we developed a theoretical framework for the three-beam double SRS processes that involve pump, Stokes, and depletion beams, where the pump-Stokes and pump-depletion SRS processes compete with each other. It was shown that the net Stokes gain signal can be suppressed by increasing the depletion beam intensity. The theoretical prediction has been experimentally confirmed recently. In the previous scheme for a selective suppression of one SRS by making it compete with another SRS, the two SRS processes occur in a parallel manner. However, there is another possibility of three-beam double SRS scheme that can be of use to suppress either Raman gain of the Stokes beam or Raman loss of the pump beam by depleting the Stokes photons with yet another SRS process induced by the pair of Stokes and another (second) Stokes beam. This three-beam double SRS process resembles a cascading energy transfer process from the pump beam to the first Stokes beam (SRS-1) and subsequently from the first Stokes beam to the second Stokes beam (SRS-2). Here, the two stimulated Raman gain-loss processes are associated with two different Raman-active vibrational modes of solute molecule. In the present theory, both the radiation and the molecules are treated quantum mechanically. We then show that the cascading-type three-beam double SRS can be described by coupled differential equations for the photon numbers of the pump and Stokes beams. From the approximate solutions as well as exact numerical calculation results for the coupled differential equations, a possibility of efficiently suppressing the stimulated Raman loss of the pump beam by increasing the second Stokes beam intensity is shown and discussed. To further prove a potential use of this scheme for developing a super-resolution SRS microscopy, we present a theoretical expression and numerical simulation results for the full-width-at-half-maximum of SRS imaging point spread function, assuming that the pump and Stokes beam profiles are Gaussian and the second Stokes beam has a doughnut-shaped spatial profile. It is clear that the spatial resolution with the present 3-beam cascading SRS method can be enhanced well beyond the diffraction limit. We anticipate that the present work will provide a theoretical framework for a super-resolution stimulated Raman scattering microscopy that is currently under investigation.