Most typical rare/orphan diseases are Mendelian diseases and so have been tackled with linkage mapping, candidate gene sequencing, and exome/whole genome sequencing studies within affected families. GWAS has typically been applied to the study of common diseases where you expect many different genetic variants to be involved. There have been cases of Mendelian diseases where the causative mutations are actually in more than one gene being inherited within the family (digenic, etc) but the known cases are fairly rare (but this inheritance pattern is likely more common than thought, it is just harder to detect). When studying multiple families with the same disease, but where it is known that there may be multiple potential genes mutated, it is usually some sort of linkage analysis performed with cases and controls as opposed to GWAS.
There are definitely a lot of rare diseases with interesting GWAS, for example just among neurodegenerative diseases, PSP, ALS, FTD, CBD all have GWAS now and are at about 10 in 100,000 population prevalence.