Fig. 1: Single-cell assessment of biopsies from patients with prostate cancer reveals SPP1hi-TAMs with elevated immunosuppression programs prevalent in advanced disease stages.
From: Evolution of myeloid-mediated immunotherapy resistance in prostate cancer
a, Schematic illustration of 5′ scRNA-seq (10x Genomics) on tumours from patients with either ADT-naive localized prostate cancer (n = 13), metastatic hormone-sensitive prostate cancer on ADT (HSPC; n = 24) or mCRPC progressing on ADT (n = 6). b, UMAP plots showing cell types (left) and distinct myeloid subsets (right) in human prostate cancer. Prolif, proliferative. c,d, Density (c) and bar plots (d) depicting the quantification of myeloid-subset frequencies across disease progression, with localized disease (grey; n = 13), HSPC (blue; n = 24) and mCRPC (red; n = 6). Significant changes were observed for cDC2 (P < 0.001 for mCRPC versus localized; P = 0.002 for mCRPC versus HSPC), EEF1A1hi-TAM (P < 0.001 for mCRPC versus HSPC) and SPP1hi-TAM (P = 0.002 for mCRPC versus localized; P = 0.04 for mCRPC versus HSPC). e,f, UMAP (e) and bar plots (f) showing immunosuppression gene signature scores among myeloid cells in human prostate cancer (n = 43 samples). In d and f, boxes represent the inter-quartile range (IQR), with bars indicating 25% − 1.5 × IQR and 75% + 1.5 × IQR. Outliers beyond 1.5 × IQR are included. The median score for SPP1hi-TAMs is indicated in red. g, Correlations between SPP1hi-TAM enrichment and CD8+ T cell exhaustion scores across disease stages. The lines represent the best-fit lines; each patient sample is indicated by a circle. HSPC, P = 0.17, R = 0.291; mCRPC, P = 0.07, R = 0.780; localized, P = 0.66, R = −0.134. h, Differentially expressed genes (adjusted P < 0.05, absolute log2 fold change (|log2FC|) > 0.5) in SPP1hi-TAMs compared with other myeloid cells highlighted in red. Statistical significance was determined by ordinary two-way analysis of variance (ANOVA) with Sidak correction (d); Kruskal–Wallis test with Dunn’s correction (f); simple linear regression analyses (g); and Wilcoxon test with Benjamini–Hochberg correction (h).*P < 0.05, **P < 0.01, ***P < 0.001; NS, not significant. The illustration in a was created using BioRender (https://biorender.com).
