Presenter(s)
V. R. Li1, J. A. Benson2, D. Duenas2, C. Kyriakopoulos3, A. V. Serritella3, R. Hutten2, S. Cho4, and J. M. Floberg5; 1University of Wisconsin Madison School of Medicine and Public Health, Madison, WI, 2Department of Human Oncology, University of Wisconsin Hospitals and Clinics, Madison, WI, 3University of WIsconsin School of Medicine and Public Health, Department of Medicine, Division of Hematology/Oncology, Madison, WI, 4The University of Texas MD Anderson Cancer Center, Houston, TX, 5Department of Human Oncology, University of Wisconsin-Madison, Madison, WI
Purpose/Objective(s): To confirm that previously reported quantitative imaging biomarkers from prostate specific membrane antigen (PSMA) PET predict response as well as toxicity of radioligand therapy (RLT) in an expanded cohort of prostate cancer patients.
Materials/Methods: Data for 81 patients treated with PSMA-RLT using either Ga-68 PSMA-11 (Ga-68) (22 patients) or F-18 DCFPyL (Pyl) tracer (59 patients) in the routine practice of a single institution were retrospectively analyzed with IRB approval. Quantitative PSMA-PET metrics were obtained using a commercial quantitative imaging tool, including the mean and maximum standardized uptake values (SUVmean, SUVmax), percentage of total volume of disease imaged that does not show PSMA uptake (%PSMAneg) and total PSMA-avid disease volume (PSMAvol) across all lesions. Difference in SUVmean between patients scanned using Ga-68 and Pyl tracer was determined with Wilcoxon Rank Sum test. Logistic regression was used to investigate association between the imaging metrics and reduction in pretreatment PSA to PSA nadir (PSA%red), as well as a =50% reduction in PSA (PSA50%). The relationships of the imaging metrics to progression free survival (PFS) and overall survival (OS) were assessed using Cox regression and Kaplan-Meier. SUVmean and PSMAvol were compared between hematologic and non-hematologic toxicities (CTCAE 5.0) groups using Mann–Whitney U tests and logistic regression, including log-transformed volume.
Results: The SUVmean was found to be significantly correlated with PSA50%, treated as a continuous variable (OR: 1.40, 95% CI: 1.15-1.78, p=0.002) and stratified by median (SUVmean <7.7 vs =7.7, OR: 3.99, 95% CI: 1.56-10.86, p=0.0049). Its correlation with PSA%red was also significant (p=0.000014). The correlation of SUVmean and PFS was statistically significant when treated as continuous variable (HR: 0.82, 95% CI: 0.74-0.91, p=0.0002) and when stratified by median (SUVmean =7.7 vs <7.7, HR 0.44, 95% CI: 0.26-0.74, p=0.002). SUVmean did not differ between patients imaged with Ga-68 and Pyl tracers (p=0.38). OS did not significantly correlate with SUVmean (HR: 0.92, 95% CI: 0.81-1.04, p=0.16). PSMAvol for patients with grade =3 hematologic toxicity was significantly higher than those for grade 0-2 (median 781 vs 338, p=0.002), with log-transformed volume showing a 74% increased odds of severe hematologic toxicity per 1-unit increase (OR 1.74, p=0.003). SUVmean for patients with grade 0-2 hematologic toxicities were also slightly higher than those with grade =3 (average 8.7 vs 7.7), although the differences were not significant. No such association was observed for non-hematologic toxicities.
Conclusion: This study confirms that the SUVmean and PSMAvol measured from PSMA-PET are robust imaging biomarkers for PSMA-RLT and may be used in routine clinical practice for prognosis prediction and/or patient selection for PSMA-RLT if verified with large datasets.