[¹⁷⁷Lu]Lu-PSMA SPECT/CT for early response assessment using quantitative RECIP 1.0

This study evaluates the prognostic value of early response assessment using the Response Evaluation Criteria in PSMA imaging (RECIP 1.0) applied to 24-hour post-therapy SPECT/CT at the second cycle of lutetium-177 (¹⁷⁷Lu) labeled PSMA radioligand therapy (RLT).

This analysis included 136 metastatic castration-resistant prostate cancer (mCRPC) patients treated with ≥ 2 cycles of [¹⁷⁷Lu]Lu -PSMA-I&T. Different clinical and biological factors were recorded. Imaging parameters (SUVmean, SUVmax, total tumor volume [TTV]) were measured on 24 h SPECT/CT at cycles 1 and 2 (C1, C2). Patients were classified into progressive disease (RECIP-PD) or non-progressive disease (RECIP-non-PD). Uni and multivariable analysis were performed using Cox regression and Kaplan-Meier curves for progression-free survival (PSA-PFS) and overall survival (OS).

After a median follow-up of 23 months, median PSA-PFS was 6.4 months and median OS was 16.6 months. Regarding imaging parameters, increase in SUVmean (HR = 1.02, p < 0.001), TTV increase ≥ 20% (HR = 3.5, p < 0.001) and appearance of new lesions (NL) (HR = 2.6, p < 0.001) were significantly associated with shorter PSA-PFS, while only TTV increase ≥ 20% (HR = 1.8, p = 0.023) and NLs (HR = 2.3, p = 0.0016) were associated with OS. The combination of both categories using RECIP 1.0 demonstrated that RECIP-PD patients (n = 16/136) had significantly reduced median PSA-PFS (HR = 3.6, p < 0.001) and OS (HR = 3.1, p < 0.001) compared to RECIP-non-PD. The most consistent prognostic accuracy was achieved using RECIP 1.0 (Harrell's C-index: PSA-PFS = 0.80, OS = 0.78).

Our study demonstrates that quantitative RECIP 1.0 applied on C2 [¹⁷⁷Lu]Lu-PSMA SPECT/CT is an independent prognostic tool for early identification of mCRPC patients unlikely to benefit from continued [¹⁷⁷Lu]Lu-PSMA-RLT. The clinical significance of this finding must be prospectively validated before implementation for changes in therapeutic management.