41Single-cell dissection of immunosuppressive myeloid subclusters driving resistance to immune checkpoint therapy in renal cell carcinoma (RCC)

Abstract Background With immune checkpoint inhibitors (ICIs) now widely used as a primary treatment for metastatic RCC, understanding the composition and function of the tumor microenvironment (TME) has become more important. Numerous studies have focused on T cells to explain both treatment response and resistance, and single-cell RNA sequencing (scRNA-seq) has revealed the complex heterogeneity of T cell states. However, the contribution of myeloid cells (particularly tumor-associated macrophages (TAMs), which are known to suppress antitumor immunity) remains incompletely understood in the context of ICI therapy. In this study, we aimed to investigate TAM populations associated with ICI resistance in RCC using single-cell transcriptomic profiling. Methods We analyzed 70 tumor samples (58 clear cell and 12 non-clear cell) obtained from 63 patients with advanced RCC. The cohort included 9 untreated patients, 10 who received non-ICI systemic therapies, and 44 who were treated with ICI-based regimens. From the ICI-based therapy group, we excluded 17 patients who had stable disease and focused on 29 tumor samples from 27 patients with either pre-treatment (n = 15) or post-treatment (n = 14) samples. These patients received various ICI regimens, including mono-ICI (n = 11), ICI plus ICI (n = 11), ICI plus VEGF inhibitor (n = 6), and ICI plus IDO1 inhibitor (n = 1). We performed scRNA-seq (10x Genomics) on all samples and applied non-negative matrix factorization (NMF) to identify transcriptional programs within TAMs. We then compared these programs between responders (n = 18, complete or partial response) and non-responders (n = 11, progressive disease) based on RECIST criteria. Statistical significance was assessed using the Wilcoxon signed-rank test. Results A total of 443 337 high-quality viable cells were analyzed and classified into major cell types, including lymphoid, myeloid, tumor, endothelial, and fibroblast compartments. Within the TAM compartment, NMF uncovered 8 gene expression programs, such as “antigen presentation”, “S100A8/9 inflammation”, “stress response”, “C1Q/APOE/TREM2 signaling”, “CD163/MRC1-M2-like”, “hypoxia-related signaling”, “interferon-stimulated response”, and a distinct “LILRB/SIGLEC10” immunosuppressive program. This LILRB/SIGLEC10-enriched TAM subcluster was significantly more abundant in non-responders than in responders (P = .005). Importantly, this difference was also observed in pre-treatment samples alone (P = .014), suggesting it may be involved in primary resistance. These TAMs expressed higher expression levels of immunosuppressive LILRB1/2/3 genes, the inhibitory receptor SIGLEC10 (a recently characterized “don’t eat me” signal), and the immune checkpoint molecule VISTA, compared to other TAM subclusters (p 2.22E-16 for each gene). Conclusions Our scRNA-seq-based analysis identified a distinct population of TAMs characterized by immunosuppressive transcriptional programs associated with poor response to ICI therapy in RCC. These findings provide insight into potential mechanisms of resistance and suggest that targeting this TAM subset may improve therapeutic efficacy. This study also demonstrates the utility of single-cell transcriptomics for uncovering key immunoregulatory populations in large clinical cohorts.