Imaging and biomarker factors for predicting durable responses to 177Lu-PSMA therapy in metastatic castration resistant prostate cancer (mCRPC).

Abstract

e17070 Background: 177Lu-PSMA radioligand therapy (RLT) has shown promising response rates and improved survival in metastatic castration-resistant prostate cancer (mCRPC). PSMA-PET/CT reliably identifies the presence of the target receptor, making it a key tool for selecting eligible candidates for PSMA-RLT. Imaging parameters (SUVmax, SUVmean, tumor volume), biochemical markers (PSA, ALK, LDH), and patient characteristics (e.g., age, visceral metastases) have been linked to outcomes. We evaluate a novel blood-based assay (PROSTest) for tumor characterization and personalized therapy prediction. Methods: A total of 106 mCRPC patients (median age 70, range 44–90) received a median of 4 cycles (range 1–6) of 177Lu-PSMA-I&T. Baseline blood samples for PSA levels (ng/ml) and PROSTest (qPCR score 0–100) were collected, along with PSMA imaging-derived parameters (SUVmax, SUVmean). The primary outcome was overall survival (OS). Relationships between tested parameters (SUVmax, SUVmean, PSA and PROSTest) and OS were assessed using t-tests. The best and worst quartiles of each parameter (to predict OS) were defined as good and poor prognostics factors. These were used to create a predictive model, categorizing patients with 0–4 good or poor prognostic factors. Kaplan-Meier (KM) and Hazard Ratio (HR) analyses with Cox proportional modeling was performed to evaluate this approach. Results: Median follow-up 14.3 months (range 0.4-53.1) with median OS of 14.6 months (95%CI 11.5-17.7) for a total number of events of 94. Baseline SUVmean (p=0.0025), SUVmax (p=0.023) and PROSTest scores (p=0.049) but PSA or tumor volume were not associated with OS (p=0.3 and 0.2, respectively). Specific quartiles were evaluated for prognosis. KM analysis identified that patients with no (0) “good” prognostic factors had mOS of 12 months, while mOS was “not reached” in patients with 4 “good” factors (Chi2=11.3, p=0.01, HR: 1.98-2.9 for 1-3 factors). The mOS was 18.1 months for patients with 1 "poor" prognostic factor and 4.8 months for those with 4 “poor” factors (Chi2=16.2, p=0.0028, HR: 1.3-8.1 for 1-4 factors. The Cox model (including good and poor factors) for individual patients was significant (Chi2=12.65, p=0.0018). Conclusions: This study suggests that a combination of baseline imaging parameters (SUVmax and SUV mean) with PROSTest may be useful to predict the OS in mCRPC patients undergoing PSMA RLT. Larger, multicenter studies are needed to confirm the clinical value and validate this approach.