e17015 Background: PSMA PET has been widely adopted in restaging of prostate cancer at biochemical recurrence. We assess the clinical utility of quantitative parameters on 18F-DCFPyL PET/CT at biochemical recurrence and their association with the subsequent biochemical progression free survival (bPFS). Methods: This is a retrospective image analysis and longitudinal follow up of a prospective study evaluating 18F-DCFPyL PET in prostate cancer patients with biochemical recurrence after primary definitive treatment at Stanford Hospital. The 18F-DCFPyL PET images were quantitatively analyzed automatically by the aPROMISE application, a semi-automated analysis and structured reporting of PSMA PET/CT. aPROMISE uses deep learning to segment detailed anatomical information from the CT images and uses this information in combination with the PET image to detect and quantify candidates for prostate cancer lesions. The reader works in tandem with the software to vet the final list of lesions, from which the quantitative assessments and final report is created automatically. Based on the Prostate Cancer Molecular Imaging Standardized Evaluation (PROMISE) criteria. SUVmean, PSMA positive total tumor volume (PSMAttv) and aPSMA scores, a quantitative score for tumor burden measuring the interaction of tumor volume and uptake stratified by local tumors (aPSMA-miT), regional lymph nodes (aPSMA-miN) and distant metastases (aPSMA-miMa for extrapelvic metastases, miMb for bone metastases and miMc for other organ metastases) were obtained based on the miTNM classification. The association of these quantitative parameters with the subsequent bPFS was evaluated. Results: 134 patients (age 70.1 ± 7.6, range 51 -91 years; PSA 13.9 ± 98.5, range 0.12 – 1126 ng/mL) were included in the quantitative image analysis. aPROMISE detected nodal and/or bone metastases in 94 of 134 patients (70%) and only 12 of 134 patients (9%) had visceral disease. With median follow up of 37 months, 66 of 134 patients (49%) have progressed again biochemically after treatment. The bPSF was 25.8 ± 15.1 months. Quantitative analysis of 18F-DCFPyL PET found that the subsequent bPFS is significantly associated with aPSMA-miMb ( P < 0.001), PSMAttv ( P < 0.001), aPSMA-miN ( P < 0.01), aPSMA-miMa ( P < 0.05) but not with SUVmean or aPSMA-miMc. Conclusions: Total and anatomically contextualized quantitative image analysis of 18F-DCFPyL PET/CT at biochemical recurrence may be useful in predicting the subsequent bPFS. Studies in larger patient cohort is necessary to validate these findings. Given many patients were treated based on lesions found on 18F-DCFPyL PET/CT, clinical implication of this findings remains to be elucidated.
Read full abstract