How do you reconcile discordant PSMA and MRI findings in patients undergoing definitive radiotherapy for prostate cancer?

In the situation wherein you found something on PSMA PET that wasn’t seen on MRI, I would obtain a biopsy for 2 reasons: 1. In order to confirm the presence of distant metastasis or 2. If tissue is needed for genomic testing. Otherwise, I would not biopsy in the case of discordant imaging, since I would be skeptical that a negative biopsy is truly negative. I would classify the patient into the higher stage and treat accordingly.
In the situation where you find something on MRI, such as lymphadenopathy, which is not PSMA PET avid, I would be much more inclined to biopsy, as it would be unusual for the PET to be negative if this finding was due to prostate cancer spread. I have seen other cancers, such as lymphomas, detected in this manner.

I agree with Dr. @Dr. First Last's general approach. Pertaining to SVI specifically, a few additional points which could be considered.

1. Pre-test probability: I usually try to review the imaging for new consults twice, first blinded to clinical features and second in context of the clinical features. The first review is to avoid bias, and the second incorporates an estimate of the pre-test probability of SVI. One way to do this is by using the MSKCC nomogram, and, if the risk of pathologic SVI is low, then I would have an increased suspicion the PET was falsely positive.
2. SVI Extent: While PET imaging is often reported to be more sensitive for detecting SVI (Chow et al., PMID 37032189, Muehlematter et al., PMID 31502937), I find it less plau...

I usually cover the proximal 1 cm of the seminal vesicles in my "prostate" volume anyway, so if there was PSMA uptake in the more distal SV, it generally wouldn't make a much larger volume to treat all of the SVs as well, which is what I'd most likely do if the patient was in front of me today.

It would also depend on what the scan looked like. After looking at the scan, does it seem consistent with what one might expect from an extension to the SVs? Is the registration good, or is there a chance this is a misregistered anterior rectal lesion (for example) and not an SV extension of prostate cancer at all? Is the patient's PSA inconsistent with a large volume of disease? If there's some reason I really doubted the PSMA PET, I'd proceed to a...