Combined analysis of circulating tumor cells and PSMA imaging metrics to predict efficacy of ¹⁷⁷ Lu-PSMA-617 in metastatic prostate cancer.

Abstract

215 Background: Radioligand therapy targeting prostate-specific membrane antigen (PSMA), such as 177 Lu-PSMA-617, has demonstrated clinical efficacy in PSMA-PET-positive metastatic castration-resistant prostate cancer (mCRPC). However, not all PSMA-positive mCRPC patients benefit from this therapy, highlighting the need for novel biomarkers to predict treatment response. We aimed to evaluate whether molecular analysis of circulating tumor cells (CTCs) could provide predictive biomarkers of 177 Lu-PSMA-617 therapy. Methods: In this single-institution biomarker study, male patients with PSMA-PET-positive mCRPC scheduled to begin treatment with 177 Lu-PSMA-617 were enrolled. Informed consent was obtained (DF-HCC 13-416). CTCs were isolated from blood samples collected prior to the initiation of 177 Lu-PSMA-617 therapy using a microfluidic device (CTC-iChip). Half of the CTCs from each patient were immunostained with antibodies against cytokeratin, EpCAM, and PSMA and counterstained with CD45 to exclude leukocytes. The stained CTCs were imaged using a Vectra Polaris multispectral microscope. The fluorescence intensity of PSMA in each CTC was categorized into four levels (3+, 2+, 1+, 0). The remaining CTCs were analyzed using droplet digital polymerase chain reaction (ddPCR) to profile the expression of prostate-specific genes, the androgen receptor splice variant AR-V7 , and neuroendocrine genes. CTC analyses and pre-treatment PSMA-PET imaging metrics were compared with clinical outcomes. Radiographic progression-free survival (rPFS) was evaluated using Kaplan-Meier analysis and log-rank tests. Results: Blood samples from 24 enrolled patients were analyzed, with a median follow-up of 7.5 months. Median age was 71.5 years (range 53-83). Median CTC count was 5.9 cells/7.5mL blood. Patients with high CTC count (>5.85 cells/7.5 mL) and presence of PSMA-negative CTCs had worse rPFS compared to others, although the difference was not statistically significant (median 86 vs. 393 days, log-rank p = 0.0840). Patients who were AR-V7 -positive before treatment had significantly shorter rPFS compared to AR-V7 -negative patients (median 67 vs. 393 days, log-rank p = 0.0031). Additionally, DLL3 -positive patients had significantly shorter rPFS than DLL3 -negative patients (median 109 days vs. 402 days, log-rank p = 0.0348). Using machine learning, a predictive model was developed incorporating the following factors: mean SUV, CHGA , ARV7 , STEAP2 , DLL3 , AGR2 , and E2F1 . This model demonstrated a high predictive accuracy for treatment outcomes (Low-risk group: median survival 447 days vs. High-risk group: 109 days, log-rank p = 0.0002). Conclusions: The molecular analysis of CTCs in combination with PSMA PET imaging metrics may be useful for predicting the therapeutic efficacy of ¹⁷⁷Lu-PSMA-617 treatment, and warrants validation in additional cohorts.