PSMAfore Data Highlight ctDNA as a Key Predictor of Treatment Outcomes in mCRPC

The phase III PSMAfore trial demonstrated that [¹⁷⁷Lu]Lu-PSMA-617 (¹⁷⁷Lu-PSMA-617) significantly prolonged radiographic progression-free survival (rPFS) compared with switching to androgen receptor pathway inhibitors (ARPIs) in taxane-naive patients with PSMA-positive metastatic castration-resistant prostate cancer (mCRPC) who had experienced progression with a prior ARPI.

Baseline circulating tumor DNA (ctDNA) was found to be associated with rPFS at the trial’s second interim analysis of overall survival (OS). A new analysis of the trial presented at the American Society of Clinical Oncology 2025 Genitourinary Cancers Symposium has assessed the link between baseline and cycle 2 day 1 (C2D1) ctDNA fraction, along with early ctDNA and prostate-specific antigen (PSA) dynamics, with rPFS and OS at the third interim analysis of OS.

Each patient was randomized 1:1 to receive ¹⁷⁷Lu-PSMA-617 at 7.4 GBq every six weeks for six cycles or when the ARPI was changed. The primary endpoint was rPFS, and the key secondary endpoint was OS. An in-house custom panel was used to determine plasma ctDNA fraction at baseline and at C2D1. Cox regression and random forest models were used to assess the link between ctDNA fraction and PSA with clinical outcomes at the data cutoff date of February 27, 2024.

Patients included in the study had plasma samples obtained at baseline and C2D1, with 82 patients receiving ¹⁷⁷Lu-PSMA-617 and 91 patients undergoing ARPI change. In the ¹⁷⁷Lu-PSMA-617 arm, Cox regression models showed that high ctDNA fraction was linked to shorter rPFS and OS and was stronger for C2D1 samples than baseline on comparison of individual models and within a model including both timepoints.

Random forest models that included clinical features and C2D1 ctDNA fraction showed that the total area under the curves was not improved through the addition of baseline ctDNA fraction (rPFS, 0.87; OS, 0.86).

In the overall population, fractional decreases from baseline to C2D1 in ctDNA (P=.0015) and PSA (P<.0001) were strongly associated with longer rPFS and were not strongly correlated with one another (correlation coefficient, 0.26). In the ¹⁷⁷Lu-PSMA-617 arm, fractional decrease in ctDNA (P=.002) was more strongly correlated with OS than with decrease in PSA (P=.12).

C2D1 ctDNA fraction may be more strongly associated with rPFS and OS than baseline ctDNA fraction. Beyond PSA dynamics, early ctDNA fraction changes offered further prognostic value in forecasting rPFS and OS.