>Wouldn't some degree of efficacy/safety testing be only prudent for generic copies of complex structures like MABs, proteins, complex carbohydrates, etc?<
For protein drugs, the answer is clearly yes.
Complex carbohydrates such as Lovenox are regulated as small molecules. I.e., the applications for branded complex carbohydrates are submitted as NDA’s (not BLA’s), and generic applications are submitted as ANDA’s within the existing regulatory framework.
The legal requirement for approval of an ANDA is characterization of, and equivalence to, the branded drug. A request by the FDA for de novo clinical studies to support an ANDA would be blazing a new regulatory trail.