"Perhaps controlled elution of lucentis from the surface of an implantable device rather than high dose intermittant bolus of the drug via injection would also lower the stroke risk?"
It should, because high doses of the drug would not be needed, and the overall dose would be smaller. DNA should be thinking about this approach, and probably is. By the same reasoning, subretinal delivery of lucentis, rather then intravitreal, would be even safer. Another possibility would be to use a device such as I-vation loaded with triamcinolone to reduce the dose or frequency of bolus-injected Lucentis.