Temporal dynamics of T cell reprogramming associate with clinical response to ICI-based combination therapy in metastatic renal cell carcinoma

Immune checkpoint inhibitor (ICI) based therapy, including dual ICI (anti-PD-1 + anti-CTLA4) and anti-PD-1 plus tyrosine kinase inhibition (TKI), has improved outcomes for metastatic renal cell carcinoma (mRCC). While tumor-based immunologic determinants of response have been explored, predictive biomarkers/early changes in peripheral blood mononuclear cells (PBMCs), have not. To understand mechanisms of response and resistance to first line ICI-based therapy, we prospectively collected longitudinal PBMCs at baseline (T0, n = 20) and multiple post-treatment time points (T1-T3: T1 = 1-2 wks post treatment initiation, n = 17; T2 = time of surgery for patients undergoing cytoreductive nephrectomy, n = 11; T3 = at progressive disease (PD), n = 7), as well as paired tumor samples at T0 (n = 20) and T2 (n = 9). scRNAseq and scTCRseq of PBMCs and tumors were used together with single-cell spatial transcriptomics (Xenium 5K, T0, n = 31; T2, n = 26) to define dynamic PBMC and intratumoral immune changes. Few differences at baseline were found in the intratumoral immune contexture or PBMCs of R (responders) and NR (non-responders). Only baseline intratumoral CD56-high NK cells were significantly higher in R and no PBMC immune cell subsets were different. Differential gene expression analysis of CD8+ exhausted T cells showed upregulation of stress response genes (e.g. HSPA1A ) and the DUSP1 phosphatase in R. In contrast, there were dynamic temporal changes seen with ICI at T1. In particular a significant increase in the proportion of circulating proliferative T and NK cells in R, which had upregulation of cytotoxicity ( GZMB , PRF1 ) and effector markers ( CX3CR1 , FGFBP2 ). Furthermore, analysis of multiple CD8⁺ T cell subsets associated with ICI response demonstrated activation of NF-κB and MAPK module scores in R at T1. Intratumoral scRNAseq profiling at T2 revealed that CD8⁺ T cells from R had lower exhaustion related genes ( TOX , LAG3 , CD38 ) and higher AP1 transcription factor (TF) activation including JUN/FOS, whereas NR accumulated terminally exhausted CD8⁺ T cells with high level of BATF activity. These findings indicate that JUN/FOS signaling associates with T cell activation in R, while BATF drives terminal exhaustion and NR. Finally, tumor spatial transcriptomics demonstrated significant enrichment of C1QC tumor-associated macrophages (TAM) in the cellular neighborhoods of exhausted CD8+ T cells at T2 of R (but not T0 or in NR). The role of C1QC macrophages in ICI response remains debated. In total, these data demonstrate there is limited discrimination between R and NR at baseline (PBMCs or tumor) but identify early (1-2-week post-treatment) PBMC changes in R (increased proliferative T cells) as well as differences in spatially organized immune trajectories post-treatment tumors of R. The increase in proliferative T cells in PBMCs could serve as a very early biomarker of R.