Here's my 2-cents, 2-pennies, or 2-any other small denomination coins:
First, I am NOT a regulatory expert, but I am familiar with assay development and validation.
All assays should be controlled in some fashion, to give the practitioners some confidence that the results are valid AND that batch-to-batch variation is limited. In a titration, especially a serological titration, this is a little more difficult than having one or two pass/fail samples that are typically included in many laboratory tests.
If a control is needed for a titration, it doesn't need to be the same specificity as the test antibody (as Malcolm highlights). Ideally, yes, but not necessarily. It does need to be reliable/robust and give the same end point each time, even with the acknowledged variations in serological tests (reagents, test cells, techs, etc.). Tube testing is notoriously variable, while gel testing is believed to reduce some of those nuances. As Malcolm suggests, it might be necessary to have a control for an IgM titration (ABO) and/or a different one for an IgG titration. At the very least, the end-points may be different. An IgM control might be as simple your routine anti-A reagent; a simple IgG control might be an IAT-reactive anti-D or other specificity. A clever option might be a control that contains both - an IgM component and an IgG component.
If a test system is adequately controlled each time (and passes), in my opinion, there is no reason to routinely perform parallel testing of successive patient samples. Retention samples might be useful for investigatory reasons if/when a patient's antibody titration changes radically from one sample to the next, or if there's some other medical indication.
I getting a sense of deja vu, I think I've written this before.