Final geometry has to be the same (within tolerances) for any replacement item.
Surface NDE may, or may not, be the same: For example, a dye penetrant inspection might have been required for a forged and welded assembly because it was welded. The forging might have needed a mag particle exam because it was forged and the assembly needed to be checked for casting flaws. An assembly might have been xrayed because that was the only way (at that time) to assure some inspector that clearances or internal passageways were not blocked by mold residue or sand or grit from the molding process.
On the other hand, a 3D printed part may be questioned about gran structure and total strength BECAUSE it has been assembled in 3D and then sintered, but not rammed by a 35 ton press to squish the hot grain structure according to 1950's accepted metal-working practice. (That "Well, we always used to do it that way!" does NOT always mean "that way" was the best way or the only way or even the "needed-to-be-done" way. It DOES MEAN, however, "that way" has been written up and accepted - and it might be "the only way" as well. So you WILL have to justify differences from the old way.
Destructive testing of at least the first 3D product will almost certainly be "expected" even if it is not "written" because of the industry's and the regulator's proper and natural caution. BUT! A destructive test may not need final intrinsic exact machining of all surfaces and shapes. If the strength test does not stress parts of the unit that are intrinsically and elaborately machined, destructively test the "casting" of the 3D part. Do a leak check on the machined part.
Document what you proposed doing, get an independent opinion, perhaps of your other customers who may later buy your parts, and submit the package. Be conservative, but be real.
Expect closer scrutiny (as is proper!) when the new part has an ultimate strength close to requirements. For example, assume some actuator after all safety factors and "what if" scenrios and margins and estimates is required to survive (not fail) at 850 lbs on the actuator pin. You make a 3D part, sinter it, forge it (if required), heat treat heat, machine it, assemble it, and test it. If it fails at 1250 lbs on the actuator pin, people will accept the result. (The original part geometry was obviously overbuilt, over-designed.) But it is "passed" the destructive test at only 910 lbs force, you's probably should test a second part. The failure was very close to the required level. That -retest evaluation should be a part of your package as well. It shows you are addressing manufacturing variations and assembly variations in the 3D "assembly" and printing.
It is possible the 3D printing community has protocols for this conservatism, but probably not. Aero? More likely - they've been approaching 1.0 safety factors in rocketry since the V-2's first blew up on the launch pad.