A high OXPHOS CD8 T cell subset is predictive of immunotherapy resistance in melanoma patients

Immune checkpoint inhibitor (ICI) therapy continues to revolutionize melanoma treatment, but only a subset of patients respond. Major efforts are underway to develop minimally invasive predictive assays of ICI response. Using single-cell transcriptomics, we discovered a unique CD8 T cell blood/tumor-shared subpopulation in melanoma patients with high levels of oxidative phosphorylation (OXPHOS), the ectonucleotidases CD38 and CD39, and both exhaustion and cytotoxicity markers. We called this population with high levels of OXPHOS CD8(+) T-OXPHOS cells. We validated that higher levels of OXPHOS in tumor- and peripheral blood-derived CD8(+) T-OXPHOS cells correlated with ICI resistance in melanoma patients. We then developed an ICI therapy response predictive model using a transcriptomic profile of CD8(+) T-OXPHOS cells. This model is capable of discerning responders from nonresponders using either tumor or peripheral blood CD8 T cells with high accuracy in multiple validation cohorts. In sum, CD8(+) T-OXPHOS cells represent a critical immune population to assess ICI response with the potential to be a new target to improve outcomes in melanoma patients. Single-cell transcriptomics from paired autologous peripheral and tumor lymphocytes from melanoma patients reveal an understudied population of metabolically active, dysfunctional CD8 T cells. This subpopulation is associated with immunotherapy resistance and was leveraged to develop a therapeutic response predictive model.

Automated summary

Single-cell transcriptomics revealed a subset of CD8 T cells with high oxidative phosphorylation in melanoma patients, which correlated with immune checkpoint inhibitor resistance. A predictive model based on transcriptomic profiles of these cells accurately distinguished responders from nonresponders, offering potential as a new target for improving outcomes in melanoma patients.