INTRODUCTION: Immune check-point inhibitors (ICI) were a major breakthrough in cancer care, but optimal patient selection remains elusive in most tumors. METHODS: Overall 173 adult patients with metastatic solid tumors candidates to ICI in clinical trials at our Institution were prospectively recruited. Blood samples were collected at cycle 1 (C1D1) and 2 (C2D1) and until the occurrence of progressive disease (PD). C1D1 LIPI, RMH, PMHI, NLR, dNLR, PIPO and GRIm prognostic scores were calculated. The primary endpoint was identifying the best score to predict rapid PD (≤ 4 months) with ICI using logistic regressions accounting for tumor type, and receiving operators characteristics (ROC) with area under curve (AUC), accompanied by an extensive comparison of the score performances in the prediction of overall survival (OS), progression-free survival (PFS), overall response rates (ORR) and durable clinical benefit (DCB). Secondary objectives included describing study cohort outcomes and studying the association between the selected score at C1D1, C2D1 and its dynamics with OS and PFS. RESULTS: C1D1 LIPI was the best predictor of rapid PD, OS and PFS, regardless of cancer type, compared to other scores. No score was associated to ORR and only RMH to DCB. Baseline LIPI detected three categories of patients with significantly different OS (p <
0.002) and PFS (p = 0.013). The same was observed at C2D1 for OS and PFS (both p = 0.020). Significant LIPI class shifts were observed in the overall population (p <
0.002), rapid progressors (p = 0.029) and non-rapid progressors (p = 0.009). Retaining a good LIPI or experiencing a shift towards a better prognostic class was associated to improved OS (p = 0.009) and PFS (p = 0.006). C2D1 LIPI, but not C1D1, remained significantly associated to rapid PD in multivariable analysis. CONCLUSIONS: LIPI may improve patient selection for ICI and guide treatment adjustments according to on-treatment dynamics in a pancancer context.