9542 Background: Immune checkpoint blockade (ICB) is standard therapy for advanced melanoma, but many patients derive limited benefit. Biomarkers to guide treatment decisions are needed. FDG PET/CT (PET) shows promise for response assessment, but most studies are small or evaluate a single timepoint. Here, we conducted a meta-analysis to assess the prognostic value of metabolic response for overall survival (OS) and progression-free survival (PFS), and the diagnostic performance of PET-derived parameters. Methods: PubMed, Cochrane, and EMBASE were searched for studies including melanoma patients treated with ICB who underwent baseline and follow-up PET. Metabolic response was defined using PERCIST, PERCIST5, or EORTC criteria, dichotomizing patients as responders or non-responders. Random-effects meta-analyses estimated pooled hazard ratios (HRs) for OS and PFS, with subgroup analyses by PET timing (early ≤4 months vs late > 4 months). Univariate diagnostic meta-analyses evaluated the accuracy of changes in total lesion glycolysis (TLG) and metabolic tumor volume (MTV) for predicting death or OS, as well as the ability of PET-defined progressive metabolic disease (PMD) to identify clinical or RECIST-based disease progression. Results: Of the 15 studies included, 943 pts were included. Interval metabolic response was significantly associated with improved OS (HR 0.22; 95% CI 0.13-0.36; p = 0.02), with no significant difference according to the time of PET assessment early vs late (p for interaction = 0.75). Similarly, metabolic responders experienced significantly longer PFS (HR 0.21; 95% CI 0.11-0.39; p = 0.03). Although the association was statistically significant for late (HR 0.21; 95% CI 0.11-0.41), but not for early assessment (HR 0.13; 95% CI 0.01-1.91), no subgroup difference was observed (p for interaction = 0.73). In diagnostic analyses for OS prediction, changes in TLG demonstrated a pooled sensitivity of 0.70 (95% CI 0.54-0.82) and specificity of 0.79 (95% CI 0.52-0.93). Changes in MTV showed comparable performance, with a sensitivity of approximately 0.70 (95% CI 0.54-0.82) and a specificity of 0.83 (95% CI 0.42-0.97). PET-defined PMD showed good diagnostic accuracy for identifying clinical or RECIST-defined disease progression, with a pooled sensitivity of approximately 0.71 (95% CI 0.42-0.89) and specificity of 0.86 (95% CI 0.69-0.95). Conclusions: In our study, metabolic response by PET was associated with improved OS and PFS, irrespective of the timing of imaging. Quantitative changes in TLG and MTV provide moderate accuracy for predicting survival, while PET/CT-defined PMD reliably identified RECIST-based disease progression. These findings support future prospective studies investigating the role of PET dynamics as a minimally invasive prognostic and predictive biomarker for pts with advanced melanoma undergoing treatment with ICB.
Mamede et al. (Thu,) studied this question.