The only therapies you can study scientifically (making sure everyone receives the same treatment in the treatment group), are therapies with manuals. However, only about 3% of all therapies have such manuals (CBT being one of them).
Most therapists highly individualize their approach - just because they have the same name doesn't mean they're actually doing anything similar. So claims of "empirical" proof of psychoanalysis is nonsense. The nature of what psychoanalysis is precludes that.