We never overcame resistance to classical chemotherapy so why would this be any different? Fortunately and inconveniently, the truth is that the body has a myriad of redundant and overlapping mechanisms for just about everything and which tumor cells are good at exploiting. Have to walk the line or find the next effective thing, usually for some subfraction of all comers. New combos - like the tried and true CHOP - being explored. I think that's all I can offer.

It would be useful if the question is more specific about what immunotherapy it refers to. Not all "immunotherapies" fail...

I would say that it is very likely that your subset of patients is not homogeneous. I would determine the difference in the "make up" between the responder and non-responder groups, in terms most relevant to the treated condition / immunotherapy used, and then use the new criteria for future selection of patients. Or you may refine the existing patient-selection criteria if you have enough relevant information.

I would start with considering what is known about drug(s) you are using. Drugs are often approved for a defined population of patients, so your population could be stratified accordingly (the best would be not to rely on general / symptomatic data but go by specific targets on which the drug has been tested to act). This way, your heterogeneity within patient populations selected using such targets would likely be less broad.

Looking for resistance to immunotherapy via antibody-mediated clearance of the drug can be identified in patients from serum in PhI and PhII clinical trials.

Am not expertise in either Immunological products or different Therapies used for cancer or AIDS patient.
This is based on my experience with family friends who was suffering from cancer, and they got very good results with combination therapy i.e. Chemotherapy + Bioven [Immuno modulator]. This is just to share my experience. Thanks,