The predictive role of [¹⁷⁷Lu]Lu-PSMA-617 SPECT/CT semiquantitative parameters from the 1st RLT cycle in advanced mCRPC: a preliminary bicentric lesion-based analysis.

Abstract

[¹⁷⁷Lu]Lu-PSMA RLT represents an effective option for advanced mCRPC. SPECT/CT allows the assessment of biodistribution and lesion-level tracer accumulation. We aimed to determine whether semiquantitative RLT SPECT/CT parameters could predict response on a lesion level. We retrospectively considered consecutive mCRPC patients who, between February 2022 and January 2026, received minimum two [¹⁷⁷Lu]Lu-PSMA-617 cycles at Messina and Genova Universities, with SPECT/CT at the 1st RLT, including whole PSMA-positive disease. Each [¹⁷⁷Lu]Lu-PSMA-617-positive lesion was semiautomatically segmented using MIM through a 40% threshold to extract SUVmax, SUVmean, Total Lesion Volume (TLV), and Total Lesion Activity (TLA=SUVmean×TLV). Each lesion's semiquantitative parameter was correlated with the single-lesion response at the last RLT cycle, classified as progressive or non-progressive (complete/partial response, stable disease) according to cut-offs from qualitative RECIP 1.0 and quantitative PPP criteria. ROC curves were used to determine the best cutoffs. We included 23 mCRPC patients for a total of 290 lesions: 249 osteomedullary, 34 lymph nodal, and 7 visceral. At the last cycle, 60 out of 290 lesions progressed, while 230 out of 290 remained stable/responded to RLT. 1st cycle [¹⁷⁷Lu]Lu-PSMA-617 SPECT/CT semiquantitative parameters were significantly higher in non-PD than in PD lesions (p < 0.001 for TLA, SUVmax, SUVmean; p = 0.008 for TLV). On ROC analysis, TLA reached an AUC of 0.854 (best cut-off = 61.7), SUVmax 0.843 (best cut-off = 11.8), SUVmean 0.838 (best cut-off = 6), and TLV 0.612 (best cut-off = 7.05mL), respectively. Our preliminary findings suggest that 1st cycle [¹⁷⁷Lu]Lu-PSMA-617 SPECT/CT semiquantitative parameters, especially when reflecting PSMA-expression, may serve as promising early predictors of the single-lesion response to RLT.