I don't know where the data comes from (as in, I can't point you to the research papers), but it's stated that 40% of original diagnosis are not correct when the histopathologist has your prostate on a tile in front of them in the lab. Of course, we only get to actually check this in the case of a prostatectomy. Usually in these cases, the diagnosis is upgraded, but occasionally it is downgraded.
This points to us still not being able to accurately diagnose prostate cancer staging and Gleason.
In the case of external beam radiotherapy, this is not particularly serious because the whole prostate is treated anyway, and the beam spill is likely to mop up any very local spread which was not known about, particularly in conjunction with HT. So we don't get to hear about under-diagnosed cases as much because the treatment works for them in any case.
In the case of prostatectomy, it's a bit more serious in that effect from having to cut wider will give you side effects you were hoping to avoid, and you still have a sharp treatment cutoff at the cut boundary, without any treatment spill past that point. The pathology also gives a hint how well the treatment is likely to have gone. Radiotherapy is available as a backup treatment in many cases.
In the case of closely targeted therapies such as SABR and brachy, there isn't much spill outside the target treatment area to catch anything the diagnosis missed. Brachytherapy Boost (which is combined with external beam) is specifically to get around this - a high dose into the known cancer, but deliberately generating treatment spill into a wider area to mop up what wasn't known about.
For focal therapies, it's even more important to have a correct/accurate diagnosis, and some focal therapy centres will rescan with top quality scanners to try and achieve this.
The most critical treatment in my view for which an accurate diagnosis is required is active surveillance. I do come across patients who were offered active surveillance, but declined and went for prostatectomy, only to find from the pathology review they were not suitable candidates for active surveillance in the first place. In my view, active surveillance requires the high grade scans used for focal therapies, but many centres offering active surveillance probably don't have these or don't rescan before offering this treatment.