Evaluation of a novel transcriptomic tumor signature (PROSTest) as response biomarker for 177Lu-PSMA therapy in advanced prostate cancer.

183 Background: Radionuclide therapy targeting the prostate specific membrane antigen (177Lu-PSMA therapy) has proven to be an effective treatment in men with metastatic castration resistant prostate cancer (mCRPCs). Despite representing a significant therapeutic breakthrough, a critical unmet need in 177Lu-PSMA therapy, is a prognostic biomarker for treatment optimization. Imaging and standard biomarkers have limited value. The PROSTest is a new 27-gene algorithmic signature originally developed for prostate adenocarcinoma diagnosis (0 to 100, positive score ≥20). We hypothesized that PROSTest would be elevated in mCRPCs and could have utility as a biomarker for mCRPC management. Methods: Prospective enrollment of 113 mCRPC for 177Lu-PSMA therapy (KlbB-5338-0302021 study). Pathology, clinical and biomarker data were available as was PSMA-PET/CT. Blood samples were collected for PROSTest prior to therapy. Target genes were isolated and amplified using qPCR. PROSTest scores (0-100) were obtained following algorithmic analysis. Scores were correlated with mCRPC diagnosis and baseline information. Scores and standard clinical measures were evaluated as prognostic factors with survival as endpoint. Mann-Whitney U-test, Kaplan-Meier survival and Cox proportional hazards regression analysis were utilized. All data: median (IQ range). Results: 89 (79%) patients were evaluable. Age was 75 (68-80). Disease characteristics at time of diagnosis included Gleason scores 8-10 (70%) and TMM: T3-T4 tumors (67%), N1 (53%), M1 (45%). At the time of therapy all patients were metastatic and all exhibited PSMA-positive disease. The highest tumor SUVmax was 51 (28-78). PSA levels were 69ng/mL (18-305). The PROSTest score was 89 (81-92). PROSTest scores were weakly correlated with age (r=0.33, p=0.0015) but not with baseline histopathological parameters (e.g., Gleason score, TNM) or pretreatment imaging results (e.g., SUVmax). Twenty-four (27%) patients have perished. Treatment and follow-up (5 months, 3-18) are ongoing. The mOS was 15 months. No factors were associated with death as an outcome except for the PROSTest score. PROSTest scores ≥79 (based on ROC analysis) were associated with significantly increased risk for mortality (HR: 2.9, 95% CI: 1.5-7.4). The mOS was 14 months in patients with pre-therapy PROSTest scores >79 compared to mOS not reached for PROSTest scores <79 ( p=0.02). In the COX model, baseline PROSTest was confirmed to be significantly predictive of death despite therapy (β = 1.51, p=0.01). Conclusions: The PROSTest blood gene expression score is elevated in mCRPCs. Levels are not associated with baseline clinical, histopathological, pretreatment PSA or imaging parameters. Elevated expression (≥79) of this biomarker prior to treatment was associated with a lower survival and could be used to predict survival in patients undergoing 177Lu-PSMA.