Background: Despite anti-monoclonal antibodies targeting programmed cell death-1 (PD-1) or its ligand PD-L1 revolution, most patients (pts) with advanced NSCLC displayed primary resistance to PD1/PD-L1 blockade. Identifying targetable alterations involved in resistance represent a promising approach to improve immunotherapy efficacy. TROP2 expression is associated with a poor prognosis in NSCLC, and offers a promising target for treatment with the development of new antibody drug conjugate. TROP2 has been suggested to play a role in innate immunity response, but, it's unclear whether TROP2 expression by tumor cells is detrimental specifically in the context of PD1/PD-L1 blockade. The aim of our study was to investigate the the role of TROP2 expression in outcome of NSCLC pts treated with ICI. Methods: We analyzed the largest NSCLC gene expression data set worldwide including the whole-transcriptome profile of 891 pre-treatment tumors from POPLAR and OAK trials, two randomized studies assessing the efficacy of atezolizumab versus docetaxel in NSCLC. We complemented this transcriptional approach with multiplex IF and digital pathology analysis to confirm our findings at the protein level We investigated whether a non-invasive approach focusing on the level of plasma levels of circulating TROP2 may be relevant to predict efficacy of ICI. Results: The PFS of pts with high expression of TROP2 was significantly lower than in pts with low expression 2.5 vs 4.1 months, p < 0.001. Strikingly, high TROP2 expression was also associated with lower clinical benefit rate 15% vs 28%, p = 0.009 and worse overall survival (OS) 12.6 vs 16.3 months, p = 0.007 When adjusting for covariates such as histological subtype, TLS signature, PD-L1 expression, TROP2 remained independently associated with PFS and OS. None of these correlations were observed in the docetaxel arm, suggesting a predictive value of TROP2 gene expression specifically for response to ICI. By using a deconvolution approach, we also observed that TROP2-high tumors were characterized by a significant lowest expression of genes specific to immune populations (T effector memory cells,Th17 cells). Multiplex IF on an independent cohort 53 NSCLC tumors confirmed the correlation between high TROP2, microenvironment features and poor response to PD1 inhibition. Analysis of circulating plasma levels of TROP2 in 97 NSCLC patients revealed a strong correlation with expression in tumor tissue. High plasma levels of circulating TROP2 was also significantly associated with worse ORR, PFS and OS on treatment with PD1/PDL1 antagonist. Conclusions: Our study revealed that TROP2 expression by tumor cells in NSCLC is associated with worse outcome in patients with NSCLC independently of PDL1 expression. Targeting TROP2 in combination with ICI may represent a promising strategy to improve ICI efficacy in pts with TROP2-high NSCLC.