I would think that the post-fix and immunostain experimental design would suffer from diffusion effects, as you suggest. For information on markers for BBB permeability (many can be employed in IHC experiments), I would recommend the following articles. These are primarily focussed on AD/neurodegenerative models or samples - depending on your specific type of research, there may be other markers that are preferred.
Kook 2012 (PMID: 22745485) uses a combination of markers including ZO-1, GLUT-1, MMPs (2 and 9), and RAGE (AD specific).
Kalaria, 1999 (PMID: 10672233) has a more thorough review and suggests additional targets/markers for BBB permeability, incljuing TIMP-2 and .
Covance offers MMP-2, MMP-9, and TIMP-2 that have been tested in formalin fixed paraffin-embedded samples. (https://store.crpinc.com/) These may be a good place to start for the fresh frozen cryostat samples.
Hope this helps