There are a few SAE studies (the GM study and a Cummins study) that specifically prove that efficiency improves with use.
In the GM study, they clearly showed that both a 40um filter and a 15um filter ended up with a final efficiency effective at 10um, and then the improvement stopped. Obviously, it took longer for the 40um filter to reach that same pore size, but they both ended up about at the same place.
As much as some of these filter studies are horribly misunderstood (and incorrectly applied to many conversations), this is one topic that the studies showed incredible accuracy and good data.
Now, those filters had no bp valve on them, and they were run in lab systems, so the data isn't really directly applicable to every situation, but generally it does show that continued use will load the media, thereby improving efficiency.
To get the media to load in this manner, the sump systems for the engines were (and I cannot add enough emphasis here ...) HEAVILY, GROSSLY OVERLOADED with dust particulate. What they did in no way represented real world contamination. The upside here is that extending you "normal" 5k mile OCI to perhaps 10k or 15k miles probably isn't going to ever overload your media as was done in the lab testing. But the downside is that your "normal" particulate load will also not show nearly as great a benefit, because the loading will never be as high as the lab tests. What is reasonable to conclude is that (as long as you don't operate a sludger engine or totally neglect your ride), an FCI extension is safe to do.
Yes - filters get more efficient as they age.
Yes - an FCI extension of moderate distance (2x or perhaps 3x) is going to be safe.
No - you'll not see a shift in efficiency anywhere near as great as the GM study showed because, hopefully, you'll NEVER see the kind of particulate loading they purposely induced.