There is no general reason why therapeutic antibodies cannot be used for diagnostic purposes - however, this depends on the specific antibody and the type of diagnostic procedure. A therapeutic antibody has to recognize its antigen in its native state and native environment (e.g embedded in the cell membrane). Such an antibody, in conjunction with a suitable secondary antibody, should work well for FACS and immunostaining of living cells, may work for ELISA and pull-downs of solubilized antigen and staining of fixed tissue, depending on the structural integrity of the antigen, but may not work for western blots, where the antigen usually is fully denature.

I think it's mainly due to cost issues. Therapeutic antibodies have higher requirements due to their use in the body and higher production costs. Diagnostic antibodies are used for in vitro diagnosis, and their production requirements are low and cost is low. -Creative Diagnostics

I really agree with the previous two answers. For therapeutic use, many MAbs must be humanized which leads to a high cost, for diagnostic it is not necessary. As Annemarie said: "'A therapeutic antibody has to recognize its antigen in its native state and native environment" but also fulfill a biological function. Therefore, the class or subclass, glycosidation and other characteristics are very important for this purpose, in diagnosis it is not relevant. An antibody to be used in diagnosis can recognize a well represented and specific epitope that does not need to be a specific therapeutic target.