It may depend on the exact antibody / epitope. For example, many anti-human antibodies are so superior that they are hardly influenced by fixation. Also, in mu experience using mouse tissue, not all epitopes needed antigen retrieval. It may also relate to whether they have used a down stream amplification step in the staining process. My advice is to try your stain both ways. With and without anyigen retival.

Whether you require to do antigen retrieval depends on your specific antigen and the primary antibody you are going to use.

Some antigens are robust/abundant and withstand strong fixation and still yield good immunolabeling. However, you might find antigens that require antigen retrieval as you mentioned to cross-link the protein and basically gives more consistent staining. So I would recommend checking whether the antibody you are going to use require antigen retrieval or not (similar articles, ab datasheet).

Good luck

frozen sections require very gentle antigen retrieval methods. It is feasible to do antigen retrieval on frozen PFA-fixed sections but not by boiling in citrate or in pressure cooker, like paraffin sections. Frozen tissue sections will fall apart if you use such techniques. HistoVT One from Nacalai is used by some tyo do antigen retrieval on frozen sections. Also mild digestion with proteinase K or trypsin. But this decreases the morphology. Collectively, it is best to not use any harsh methods of antigen retrieval when doing immunohisto on frozen (PFa-fixed) sections.

https://www.researchgate.net/post/Is_there_any_alternative_for_antigen_retrieval_method_in_case_of_frozen_section