Indeterminate bone lesions (IBLs) on prostate-specific membrane antigen (PSMA) PET/CT are common and pose a diagnostic dilemma with important management implications. There is currently no clear consensus on imaging features that predict lesion outcomes. Our study aimed to identify features that characterize IBLs as benign or malignant on 18F-DCFPyL PSMA PET/CT using only biopsy-proven lesions. Methods: In total, 42 IBLs (21 biopsy-proven metastatic prostate cancer and 21 biopsy-proven benign) with pathologic diagnosis obtained at a median of 24 d after 18F-DCFPyL PET/CT imaging were retrospectively identified. Postprocessing software provided SUV measurements for IBLs and background liver/parotid gland ratios. Variables were analyzed using Firth and conventional logistic regression along with random forest with leave-one-out cross-validation models. Results: Threshold values for malignancy of IBL SUVmax of 8 or greater, IBL/liver SUVmax ratio of 1 or greater, IBL/parotid gland SUVmax ratio of 0.3 or greater, and PROMISE V2 score of 2 or 3 were all predictors of lesion malignancy with accuracy between 86% (36/42) and 88% (37/42) (all P < 0.0001). The presence of multiple bone lesions (P < 0.0001) and clinical status of biochemical recurrence (P = 0.002) were predictive of malignancy, whereas non-biochemical recurrence disease status (P = 0.008) was predictive of benignity. Lesion location and prostate-specific antigen level were not predictive of lesion outcome. Conclusion: IBL SUV and the ratios of IBL to background liver or parotid gland uptake are quantitative metrics that help predict whether a lesion is benign or malignant, irrespective of prostate-specific antigen level or lesion location. Because of the ease of applicability, we highlight IBL SUVmax of 8 or greater and IBL/liver SUVmax ratios of 1 or greater as predictors of lesion malignancy and an SUVmax of less than 8 and IBL/liver SUVmax ratio of less than 1 as predictors of lesion benignity. These metrics can and should be used in conjunction with a PROMISE V2 score to help stratify lesions as benign or malignant.
Journal of nuclear medicine technology. 2026 Mar 31 [Epub ahead of print]
Matthew S Agritelley, Bifei Hao, Seth Hill, Bo Zhou, Logan Linscheid, Scott Leonard, Hatice Savas, Ryan J Avery
Department of Radiology, Division of Nuclear Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois., Department of Radiology, Division of Nuclear Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois .