Abstract
Resistance to Immune Checkpoint Blockade (ICB) constitutes the current limiting factor for the optimal implementation of this novel therapy, which otherwise demonstrates durable responses with acceptable toxicity scores. This limitation is exacerbated by a lack of robust biomarkers. In this study, we have dissected the basal TME composition at the gene expression and cellular levels that predict response to Nivolumab and prognosis. BCR, TCR and HLA profiling were employed for further characterization of the molecular variables associated with response. The findings were validated using a single-cell RNA-seq data of metastatic melanoma patients treated with ICB, and by multispectral immunofluorescence. Finally, machine learning was employed to construct a prediction algorithm that was validated across eight metastatic melanoma cohorts treated with ICB. Using this strategy, we have unmasked a major role played by basal intratumoral Plasma cells expressing high levels of IGKC in efficacy. IGKC, differentially expressed in good responders, was also identified within the Top response-related BCR clonotypes, together with IGK variants. These results were validated at gene, cellular and protein levels; CD138+ Plasma-like and Plasma cells were more abundant in good responders and correlated with the same RNA-seq-defined fraction. Finally, we generated a 15-gene prediction model that outperformed the current reference score in eight ICB-treated metastatic melanoma cohorts. The evidenced major contribution of basal intratumoral IGKC and Plasma cells in good response and outcome in ICB in metastatic melanoma is a groundbreaking finding in the field beyond the role of T lymphocytes.