The Push for PSMA-PET: Evaluating its Role in Prostate Cancer Diagnosis, Journal Club - Rashid Sayyid & Zachary Klaassen

June 16, 2023

Rashid Sayyid and Zach Klaassen scrutinize a publication from European Urology comparing the diagnostic accuracy of PSMA PET and traditional imaging modalities in initial staging of intermediate to high-risk prostate cancer. The authors systematically review studies comparing PSMA-PET/CT, MRI, and conventional imaging, noting that PSMA-PET/CT shows higher sensitivity and specificity. They also discuss the FDA approval of certain agents for initial staging and recurrence detection, but highlight the variable geographic uptake due to cost, resource constraints, and false positive concerns. The speakers agree that more definitive evidence is needed to support PSMA-PET/CT as a standard diagnostic method. Furthermore, they dissect the study selection process, performance in different settings, and multiple variables considered in their bivariate meta-regression model. The results suggest that substituting conventional imaging with PSMA-PET as a first-line imaging solution for primary prostate cancer could significantly improve diagnostic accuracy.


Rashid Sayyid, MD, MSc, Urologic Oncology Fellow, Division of Urology, University of Toronto, Toronto, Ontario

Zachary Klaassen, MD, MSc, Urologic Oncologist, Assistant Professor Surgery/Urology at the Medical College of Georgia at Augusta University, Georgia Cancer Center

Read the Full Video Transcript

Rashid Sayyid: Hello everyone, this is Rashid Sayyid. I'm a urologic oncology fellow at the University of Toronto, and along with Zach Klaassen, associate professor and program director at Augusta University, we'll be discussing the recent publication from European Urology looking at a Head-to-head Comparison of the Diagnostic Accuracy of PSMA PET and Conventional Imaging Modalities for Initial Staging of Intermediate to High-risk Prostate Cancer: A Systematic Review and Meta-analysis.

We know that PSMA-PET/CT has improved sensitivity and specificity compared to conventional imaging, and this applies in both the pre-treatment setting, there are many studies in this space, but for example, if we look at the ProPSMA study, we know that the sensitivity specifically for pelvic, nodal, or distant metastatic disease is much higher, 85% versus 38% for conventional imaging. The specificity, as well, appears to be a little better at 98% versus 91%. This also applies to the biochemical recurrence setting, if we look at, for example, the CONDOR trial, about 59-66% of patients with negative or equivocal findings of conventional imaging had a 18F-DCFPyL-PET/CT positive findings.

Currently, the 68Gallium PSMA-11 as of 2020 and 18F-DCFPyL, or Pylarify, in 2021 are FDA approved for the initial staging of prostate cancer patients at high risk of metastases and also for prostate cancer patients with suspected recurrence based on elevated serum PSA levels. The uptake of PSMA-PET/CT, however, remains quite variable geographically, owing to concerns regarding the cost and resource requirements of this modality, some concerns about false positive lesions, for example, and the rib secondary to prior fractures, and the rate of these false positive lesions is quite variable depending on the tracer used. There's also lack of evidence currently that changes in the management secondary to the use of PSMA PET/CT are actually associated with improved outcomes.

In order to improve the uptake of PSMA-PET/CT in the primary staging pathway, definitive evidence on its relative diagnostic accuracy in comparison to conventional imaging is necessary at this point. And so the objective of this study was to perform a systematic review and meta-analysis directly comparing PSMA-PET/CT and conventional imaging in a head-to-head manner. To this end, the authors performed a systematic review of PubMed, EMBASE, Cochrane Library CENTRAL, and Scopus databases from inception until December of 2021. Studies were included if primary staging was performed in patient biopsy-proven cancer before definitive therapy. So again, this is in the pre-treatment setting. The studies needed to include both PSMA-PET/CT, or MRI for that matter, along with conventional imaging in the same population so that you're able to perform a head-to-head comparison. Either histopathologic results from surgery or a composite reference standard were used, as we know from the ProPSMA trial. And then there needs to be enough data in these studies in order to calculate the true positive, false positive, true negative, false negative findings in order to perform this comparison.

The risk of bias was assessed using the appropriate checklist, which is the Quality Assessment of Diagnostic Accuracy Studies-2, and as well they use the extension of comparative reviews and this is to, on a study-level basis, to get an idea of what the risk of bias was in each study and to be able to assess the validity of the results.

The primary outcome was a direct pairwise comparison of the sensitivity and specificity of PSMA-PET versus conventional imaging in the primary staging setting. They start started off looking at the patient-level accuracy and then they looked at the performance, specifically sensitivity and specificity for specific situations. The first is PSMA-PET/CT versus MR for local tumor staging, next was the PSMA-PET/CT versus MR for nodal staging, and then they did the same comparisons for nodal staging with CT as opposed to MR, so they're look at both MR and CT. Then they looked at PSMA-PET/CT versus the bone scan for bone mets staging. So again, there's quite a bit of granularity here trying to compare the performance in different settings. And then as a secondary outcome, they looked at a lesion-level analysis for nodal staging, comparing PSMA-PET and MRI.

They extracted the following data from each study. They looked at the study population characteristics, trying to see if there's going to be any difference in effect modifiers looking at the performance of these tests. They looked at the PSMA-PET/CT and conventional imaging parameters that were used. They looked at study design details, whether the readers were blinded to PSMA-PET/CT or the conventional imaging, and vice versa. So making sure that there's no reader bias in these studies. They also looked at whether the studies were prospective versus retrospective. They also assessed how the histopathological clinical reference standards were defined and derived. They also looked at the time interval between PSMA-PET/CT and conventional imaging. Obviously if there's a long time interval between the two, then that may influence how the results are interpret and bias in results. Again, quite a bit of granularity in how they looked at these different parameters across the studies.

They used a bivariate model for the meta-analysis to estimate the summary sensitivities and specificities. And so essentially what that means is that in the meta-regression model, which looks at different variables and adjusts for them. The imaging modality was added as a covariate, so that's where the bivariate comes from. They looked at statistical significance of differences and the test performance using the likelihood-ratio test comparing models with and without the covariate terms for imaging modalities. Again, they're trying to see, how does the test perform with the imaging type or modality included versus not. And by comparing the models with and without that variable, the imaging modality, they see how the performance differs. They also looked at sources of heterogeneity. They were investigated looking at study design. Did it matter if there was prospective versus retrospective? They looked at the type of PSMA-PET scanner used, whether it was PET/CT versus PET/MR. And they also looked at the time interval between the PET and the conventional imaging. Was it less than a month or more than a month?

And then finally they looked at post-hoc sensitivity analysis to examine the robustness of our findings, meaning, are the analysis sensitive to any variations in how we define things? What they looked at was did the results differ? If only FDA-approved radioligands were used, meaning the gallium and the 18F-DCFPyL. They also looked at whether results differed, whether histopathology was used to the reference standard versus the composite reference standard. And they also looked at the fact, did the results differ if we only include patient with intermediate or high-risk prostate cancer as opposed to all-comers?
At this point, I'll turn it over to Zach to go over the results and discussion for this paper.

Zach Klaassen: Thanks so much, Rashid. This is the PRISMA flow diagram for the study selection process. We can see on the left, this is identification of studies via databases and registries. And so, at the top, we see that there was 6,337 studies from the databases. 3,473 were screened, 127 were sought for retrieval, there was 123 that were assessed for eligibility. Ultimately, 31 studies were included, which was supplemented with three studies from citation searching by the authors.

The next two slides look at the characteristics of the included studies and just a couple of generalities. A majority of these studies were performed in either Australia or Europe. We see that a majority of the sample size in these studies was less than 100, there was several studies that were more than 100 patients, and these were broken down by tumor staging and detection. So we see that there was seven studies for extraprostatic extension detection, nine studies for seminal vesical invasion detection, 22 studies for nodal detection, and six studies for bone met staging. And so, again, most of these were gallium PSMA-11 studies, with several DCFPyL studies.

This looks at the PSMA-PET versus MRI and PSMA-PET/CT versus MRI with regards to detection of extra extraprostatic extension and seminal vesicle invasion. When we look at the right side of this slide, this is PSMA-PET/CT versus MRI, and the top half looks at EPE detection and the bottom half looks at SVI detection. Essentially, we can see here that all of these are statistically insignificant. So there was no additional benefit of PSMA-PET/CT to multiparametric MRI for EPE detection or SVI detection with regards to sensitivity and specificity. However, when we look at the left side of this slide, this is PSMA-PET/MRI versus multiparametric MRI, and we see that for both EPE detection and SVI detection for sensitivity, there was statistically significant improvement in sensitivity for both of these with the addition of PSMA-PET/MRI. So an absolute difference of improving 25.8% sensitivity for EPE detection and 15.7% improvement in SVI detection with the addition of PSMA-PET/MRI versus mpMRI.

This forest plot looks at the estimates of sensitivity and specificity for PSMA-PET for the detection of pelvic lymph node metastases. This is based on 19 studies, 1,190 patients. We can see that the sensitivity was 0.74, 74%, versus 0.98 specificity. So excellent specificity and reasonable sensitivity. Now, compare this to the multiparametric MRI for detection of pelvic lymph node metastases. We see a reasonable specificity of 0.83, but a much worse sensitivity at 0.39 compared to the 0.74 for PSMA-PET.

This figure looks at the link summary ROC plot for PSMA-PET versus multiparametric MRI for detection of pelvic lymph nodes. Basically, just to orient everybody, the multiparametric MRI is in red, this is individual studies, and the PSMA-PET is in these black circles. Basically, this is showing that the further to the top left, which is a sensitivity and specificity of 1, the better the performance of the study. And you can see, generally, at a high-level view, that PSMA-PET studies are closer to the top left of this graft. This is, essentially, a visual representation of the relationship between sensitivity and specificity of multiparametric MRI and PSMA-PET at the intrastudy level. These lines, these green lines as I'm showing here, connect paired results from the same study and summary estimates of sensitivity and specificity of PSMA-PET and multiparametric MRI derived from the meta-analyses.

Moving on to nodal staging. Again, this is PSMA-PET versus multiparametric MRI and PSMA-PET versus CT scan. We'll start on the right again, this is patient-level analysis on the top half, lesion level on the bottom. For PSMA-PET versus CT scan, we see improved sensitivity and specificity, which was statistically significant, 34.7% improvement in sensitivity and 14.1% improvement in specificity. Moving to the left, which is PSMA-PET versus multiparametric MRI, we see improvement in the patient-level sensitivity and specificity of PSMA-PET versus multiparametric MRI. At the lesion-level analysis, we see a significant improvement of 42.6% in sensitivity for PSMA-PET versus multiparametric MRI.

This is a summary ROC plot of PSMA-PET versus mpMRI for detection of pelvic lymph nodes, with 95% confident region and 95% prediction regions, and so a similar looking plot to a couple slides ago. The red and black dotted lines, which are these ones that I'm highlighting here with the cursor, this is the red one for multiparametric MRI, this is the black one for PSMA-PET, and the 95% prediction regions are these lines. So this is for multiparametric MRI, whereas the tighter one is for PSMA-PET in black. The take-home message from this slide is that there's substantial heterogeneity observed as evidenced by the extent, as you can see here in how wide these are around these circles, of the 95% prediction region around these summary points. So incredible heterogeneity in these studies.

Moving on to, finally, detection of bone metastases. This is the forest plot for both PSMA-PET and bone scan based on six studies, 541 patients for PSMA-PET. Excellent sensitivity, 0.98, excellent specificity 0.96. For bone scan, sensitivity of 0.73 and specificity of 0.79. So we see a significant improvement here for PSMA-PET for detection of bone metastases.

For discussion, for local staging PSMA-PET/MRI was more sensitive than multiparametric MRI detection for EPE and SVI detection. However, PSMA-PET/CT was less sensitive than multiparametric MRI for seminal vesicle detection. It is apparent that the spatial resolution for multiparametric MRI cannot be replaced, but can be enhanced, by the avidity for small lesions accorded by PSMA-PET. Secondly, PSMA-PET outperforms both CT and multiparametric MRI in nodal staging, and while lymph node involvement on conventional imaging depends on size greater than 10 millimeters or the presence of suspicious features, radiotracer uptake on PSMA-PET is relative to the background signal identifies lymph node involvement regardless of nodal size. Finally, for bone metastatic staging, PSMA-PET had significantly higher sensitivity and specificity compared to bone scan, and these results confirmed the ability of PSMA-PET to overcome the intrinsic limitation of bone scan in identifying marrow-based or lytic skeletal metastases.

In conclusion, the systematic review and meta-analysis synthesizing evidence from head-to-head comparisons of PSMA-PET and conventional imaging in the same patient cohort shows that PSMA-PET is significantly more sensitive and specific than CT/MRI and bone scan for staging of nodal and bone metastases. Secondly, PSMA-PET is more sensitive than multiparametric MRI for local tumor staging when PSMA-PET/MRI is used. These results derived from direct comparisons provide definitive evidence on the relative diagnostic abilities of PSMA-PET and conventional imaging, and suggest that replacing conventional imaging with PSMA-PET as first-line imaging for primary prostate cancer would result in significant improvements in diagnostic accuracy.

Thank you very much for your attention. We hope you enjoyed this UroToday Journal Club discussion of the recently published meta-analysis systematic review of PSMA-PET staging in European Urology.