Outcome of Patients with PSMA PET/CT Screening Failure by VISION Criteria and Treated with ¹⁷⁷Lu-PSMA: A Multicenter Retrospective Analysis

Selection of patients for treatment with prostate-specific membrane antigen (PSMA)–targeted therapy is somewhat controversial. There are those who have suggested that no selection is necessary and those who have suggested that tight imaging-based selection criteria are required. What is optimal, what is required, and what is practical are all different questions.

Given the importance of the VISION trial (the only trial demonstrating overall survival benefit with PSMA-targeted therapy), findings in this trial will be examined in some detail (1). Of note, the VISION trial enrolled patients with at least 1 metastatic lesion present on baseline contrast-enhanced CT, MRI, or bone scanning obtained no more than 28 d before beginning study therapy. Thus, metastatic disease on conventional imaging was required. In addition, patients must have progressed after one or more androgen axis inhibitors (e.g., abiraterone, enzalutamide, darolutamide, or apalutamide) and at least one taxane-based chemotherapy. Approximately 41% of VISION participants were previously treated with 2 taxane regimens.

What were the eligibility criteria relative to PSMA PET/CT imaging in VISION? First, all patients must have had a centrally read ⁶⁸Ga-PSMA-11 PET/CT scan for trial entry. Second, a metastatic lesion (one or more) that was PSMA PET–positive was required. PSMA PET positivity was determined by uptake in the lesion at an intensity level greater than that in the liver. There was no SUV cutoff requirement; potential metastatic lesions in each patient were compared with liver uptake by a centralized PET reading. There were no size criteria for metastatic PSMA PET–positive lesions.

Importantly, the patients screened for the VISION trial had additional imaging-based exclusion criteria. Patients were excluded if there were PSMA PET–negative lesions (uptake less than in liver) measuring at least 1 cm in solid organs, at least 2.5 cm in lymph nodes, or at least 1 cm in a bone lesion with a soft-tissue component. Assessment was by contrast-enhanced CT combined with the PET/CT findings. These negative selection criteria are quite important and helped to exclude patients harboring lesions with low levels of PSMA expression.

During the VISION design phase, there was a strong desire to avoid using 2 PET scans as a requirement for trial entry, knowing that the VISION entry criteria would likely be cited by regulatory authorities considering ¹⁷⁷Lu-PSMA-617 as an approved therapy. In the United States, and many other areas of the world, obtaining reimbursement for 2 distinct types of PET scans was deemed potentially problematic. Thus, for practical reasons, ¹⁸F-FDG PET scans were not used in the VISION entry criteria.

In the plenary session at the 2021 American Society of Clinical Oncology meeting, the discussant questioned whether PSMA-based imaging was required for selection of patients (2). This discussion followed the initial presentation of the VISION trial. Of the 1,003 patients screened with PET/CT, 49 (4.9%) had no PSMA-positive metastatic lesions. Of the 954 patients with PSMA PET metastatic lesions, 87 patients were excluded because PSMA-negative metastases were also detected. All told, only about 13% of patients were excluded because of PET imaging criteria. Given the overall survival benefit with a hazard ratio of 0.62 relative to control (hazard ratio, 0.62; 95% CI, 0.52–0.74), it is likely that had the VISION trial been conducted on non–PSMA PET-selected patients, the overall survival benefit would still have been statistically significant; that is, the CIs would not have crossed 1.0. Thus, questioning the requirement for PSMA PET selection for ¹⁷⁷Lu-PSMA-617 is reasonable.

Have any investigators used PSMA-targeted therapies without regard to PSMA PET selection? The answer is yes. Data on non–PSMA-selected patients have been presented from studies using ¹⁷⁷Lu J591, ²²⁵Ac J591, PSMA bispecific antibodies, a PSMA antibody–drug conjugate, and PSMA-targeted chimeric antigen receptor T cells. J591 is a monoclonal antibody that binds to PSMA and has been used to target either ¹⁷⁷Lu or ²²⁵Ac (3,4). The J591 radiopharmaceutical studies have not compared PSMA PET–selected and non–PSMA PET-selected patients; thus, it is not possible to determine how important selection might be to patient outcomes. The bispecific antibodies pasotuxizumab (also called AMG 212) and AMG 160 have also been studied in non–PSMA PET-selected patients (5,6). What is clearly noted is that many patients not selected by PSMA PET appear to respond to these treatments. Some meaningful responses have also been seen in the PSMA antibody–drug conjugate studies (7) and in patients treated with chimeric antigen receptor T cells (8). All in all, given the absence of long-term survival data and the absence of PSMA PET selection compared with non–PSMA PET selection, it is speculative to conclude that PSMA PET selection criteria improve outcomes in any of these studies. ICAR-T and bispecific antibodies may be very different from radioligands in terms of cellular PSMA expression requirements.

The TheraP Australian trial is a large phase II trial assessing prostate-specific antigen (PSA) responses to cabazitaxel or ¹⁷⁷Lu-PSMA-617 (9). The prior therapies administered to TheraP patients and the dosing of ¹⁷⁷Lu-PSMA-617 were distinct from VISION. The TheraP trial applied a double PET/CT selection using ⁶⁸Ga-PSMA-11 and ¹⁸F-FDG criteria. PSMA-positive metastatic disease required an SUV_max of at least 20 at a disease site and greater than 10 at all other measurable sites of metastatic disease. In addition, ¹⁸F-FDG PET positivity must be concordant. Patients with ¹⁸F-FDG PET–positive lesions that were discordant by PSMA PET were deemed not eligible for ¹⁷⁷Lu-PSMA-617 treatment.

The data from TheraP suggest a high rate of PSA response to ¹⁷⁷Lu-PSMA-617 treatments; an unconfirmed PSA decline of at least 50% occurred in 66% of patients. Direct comparisons of the PSA response rate between TheraP and VISION can be made, but the patient populations are distinct in several ways. Not only were the imaging entry criteria distinct, but prior treatments were distinct (no cabazitaxel pretreatments were allowed in TheraP) and doses of ¹⁷⁷Lu-PSMA-617 differed between the trials. Thus, direct comparisons of PSA response rate in TheraP and VISION are problematic.

It is possible to compare exclusion rates between TheraP and VISION. As noted previously, approximately 13% of patients were excluded from VISION because of imaging issues. For TheraP, 291 patients were screened and 29 (10%) were excluded because PSMA uptake was insufficient and 51 (18%) were excluded because of imaging discordance between the ¹⁸F-FDG PET and the PSMA PET findings. Thus, a total of 28% of the patients screened in TheraP were not treated because of PSMA PET criteria. Clearly, this exclusion rate was higher than in VISION.

Investigators had previously treated several patients on various protocols with ¹⁷⁷Lu-PSMA-617 (10). These clinical trials had entry criteria distinct from VISION, but all patients had undergone a baseline ⁶⁸Ga-PSMA PET/CT scan. A retrospective analysis of these findings is now available for analysis. There was a higher percentage of visceral disease in those not meeting VISION criteria (58.6% vs. 25.4%). Their findings indicated that treating patients who were excluded by using VISION criteria led to a lower PSA response rate (PSA decline of ≥50%) and a longer time to PSA progression. These data were striking in that a PSA decline rate of at least 50% was 50.3% versus 20.7%. Survival analyses were not properly powered and thus were not informative (however, survival trended favorably among those meeting VISION entry criteria). These data help to support the validity of the VISION criteria in patient selection and, to our knowledge, represent the only published experience addressing this issue.

What are the optimal PSMA PET criteria for selection of patients? This question is important for radioligands (both βs and αs), bispecific antibodies, antibody–drug conjugates, and chimeric antigen receptor T cells. Answers to this question are not yet clear, and trials with overall survival as an endpoint will likely be best to assess this question. It is intuitive to say that higher PSMA expression is better, but the different PSMA-targeted approaches may yield different answers. What is required? One can create an argument that no PSMA PET imaging is required for patient selection when using PSMA-targeted therapy. Little data are available to date. That said, administration of expensive therapies with potential toxicities to patients who stand little chance to benefit seems unwise. What is practical? Practicality depends on perspective and geography and economics. What is practical in one region may not be practical in another. In many respects, what is practical is what the regulators allow. There is much work to do before we can be definitive in our conclusions, especially now that PSMA-targeted therapy has many iterations.

• © 2022 by the Society of Nuclear Medicine and Molecular Imaging.

• Received for publication May 6, 2022.
• Accepted for publication May 16, 2022.