Read on Twitter
Ahh, pathology and #PSMAPET. Where to start. First, the obvious. The data is limited for the settings where you would want it. What I mean is that the test characteristics of PSMA PET vary dramatically for each patient setting and imaging finding. @CalaisJeremie @DrMHofman https://twitter.com/TylerSbrt/status/1340789277983035393
1/ Lets start with initial staging. Nearly all #PSMAPET studies in initial staging that use pathology correlate with pelvic nodal status at time of prostatectomy. For pelvic lymph nodes there are few to no false positives, and so the PPV is very high.
2/ But that is not particularly relevant as what one cares the most about is the presence of metastatic disease at time of initial staging (extra-pelvic and bones). The best attempt to do this was the #ProPSMA study by @DrMHofman, but only two patients had pathology.
3/ In the biochemical recurrence setting, there is much more data, but it isn’t really applicable to the initial staging setting. The pretest probability of a PSMA+ met being real in BCR is much higher than at initial staging, and so the PPV is probably not transferable.
4/ The other problem with the BCR data is that pathology has either come from TRUS biopsies or salvage lymphadenectomies. Based on that data, the PPV in the BCR setting was 0.99 across 256 patients in a metaanalysis we performed as part of our NDA (PMID: 30530831)
5/ The location and appearance of PSMA+ disease has a huge impact on the apparent PPV of the lesion. For example, in the left patient, the supraclavicular node has in essence a 100% PPV, and the same with the bone disease on the right. It doesn’t matter what path shows.
