In this retrospective analysis, we investigated the potential value of reconstructing the 900 seconds of PSMA PET acquisition of the prostate scanning range into a late-dynamic series. In this context, we evaluated three important clinical PSMA radiotracers, as well as the possible correlation of the generated PSMA time activity curves (TACs) with the patients' prostate-specific antigen (PSA) values and Gleason scores (GS). Additionally, we questioned the kinetics of the PSMA PET signal in the prostate throughout this late dynamic acquisition in order to evaluate the adequacy of the PSMA PET protocol for the purposes of biopsy planning and chemical recurrence detection in prostate cancer.
While the potential of early dynamic PSMA PET has already been thoroughly explored in the context of 68Ga-PSMA-11 [5]–[9], including the use of one early static PSMA image instead of an early dynamic series [10], to the best of the authors‘ knowledge, late PSMA kinetics has not been a topic of any research so far, regardless of the type of the PSMA tracer.
Despite the increased noise (as seen in figures 1-3) caused by fewer counts contributing to the image statistics present in the dynamic images, all lesions detected in static mode with high count statistics were discernible in dynamic mode as well. As both static and dynamic modes included the counts from the same time period post injection, we were not interested in searching for new lesions (like it would often be the case with early dynamic or additionally delayed imaging), but rather for a “slow-motion” close-up of the lesions already known to us from the static images.
The second factor detectable from the dynamic images were fluctuations in activity concentration over selected regions of interest (ROIs) throughout the six time frames (as seen in figure 4). In 37 out of 120 cases, the reason for these more prominent fluctuations was also motion, presumably gross patient motion rather than muscle relaxation, as motion was present usually only in one or two nonadjacent time frames and the lesion displacement did not have a gradual, consistent course. The reasons for gross patient motion could be many, however, the usual suspect are more often than not lengthy PET scans which increase patient discomfort over the course of acquisition and may lead to sudden repositioning of the patient. Indeed, in 29 out of 37 cases of detected motion, the patients had a previous partial body scan covering four of five bed positions (neck to pelvis or head to pelvis, respectively) and had to lay still for approximately 30 minutes prior to the 15-min PET scan.
In addition to this, we noticed the differences in the PSMA TACs fluctuations and their overall trends revealed a pattern depending on the location of the analysed lesion. Interestingly, the patterns were consistent between the three different radiotracers included in this study, despite their different clearance paths and rates. Furthermore, they were consistent between both of the different protocols assessed, i.e., both in patients who were yet to be potentially diagnosed with PCa and have their first staging done, and those assessing chemical recurrence of previously diagnosed PCa and undergoing potential restaging.
While the negative overall trends in the PSMA TACs computed from the activity concentration measured in the lymph nodes could mean that the lymphatic drainage is affecting the activity retention times, this phenomenon has not yet been accounted for in the literature and is yet to be understood. On the other hand, the prostatic lesions and the lesions in the prostatic fossa following RPE mostly showed a tendency to retain activity longer and to still be gaining in activity by 90-125 min p.i. It should be noted that the positive trends in the majority of cases also coincided with larger lesion sizes, which could be understood as a reflection of a higher concentration of PSMA receptors in these larger pathological regions gradually binding more PSMA as it circulates throughout the body over time. The remaining lesion locations were not present with a sufficient number of instances in this patient cohort, which prevented us from drawing more accurate conclusions from their statistics.
Another point we wish to raise regarding limited accuracy of this study is the unavoidable ambiguity GS values tend to carry, which could have affected our correlation analysis as well. As previously described in the literature [11-12], GS values could be misleading when grading different biopsies, the histology of which can be different, rendering their prognosis different as well, and yet the same GS would be used as a clinical molecular risk factor. For this reason, the accuracy of our GS correlation analysis with the generated TACs is limited by the very method of risk stratification.
With the PSA values, however, while we do not have histological ambiguity, other factors may influence the PSA amount measured such as the age of the patient, the size of the prostate, different kinds of secondary conditions such as inflammations or infections, etc. These factors, nevertheless, do not necessarily affect the PSMA binding properties and activity retention and the variation they introduce in the final PSA value may thus affect the accuracy of our PSA correlation analysis with the generated TACs as well.
Regarding the significant differences in the strength of correlation between the PSA values and the TACs slopes when the three evaluated radiotracers are compared, a few points are brought up for discussion. Namely, the strong correlation of the “late” 68Ga-PSMA-11 kinetics with the PSA values could be a result of its particular renal clearance route and the resulting activity retention in the kidneys affecting the activity clearance rate. As opposed to this 68Ga-labelled radiotracer, the two 18F-labelled ones both have a hepatic clearance route and thus a different pattern of activity retention in the liver, which could be a factor contributing to the results found in this study showing their weak to non-existent correlation with the PSA values.
Another factor which could cause this difference in correlation results as well is the uptake time allowed to the radiotracers before the beginning of the acquisition. Within our patient cohort, 68Ga-PSMA-11 had the shortest post injection acquisition time on average (58.8±12.1 min p.i.), which together with its clearance pattern could have contributed to the stronger correlation results. On the other hand, 18F-PSMA-1007 and 18F-rhPSMA7 were injected 104.1±22.4 min and 72.2±8.4 min p.i., respectively, which together with their clearance patterns could be the deciding factor for their low correlation results.
On a similar note, the effects of the PSMA PET protocol timing on the signal kinetics during those 15 minutes of acquisition were a topic of interest as well. As seen from the results, in most of the cases, however, the measured activity concentration signal, whether it be 68Ga- or 18F-labelled PSMA, featured no drastic amplitude changes. The recommendation on the time given for the PSMA uptake was indeed not strictly followed with each patient in this patient cohort due to the clinical logistical matters, but in the light of these results, the procedure guidelines for the timing of these PET imaging protocols with respect to the delay after the injection time do seem to render relatively stable results. However, our results also suggest caution and special consideration regarding the different activity accumulation and clearance rates between lesions of different sizes and locations.