Antibiotic use attenuates response to immune checkpoint blockade in urothelial carcinoma via inhibiting CD74-MIF/COPA: revealing cross-talk between anti-bacterial immunity and ant-itumor immunity

Abstract

Background: Immune checkpoint blockade (ICB) has emerged as a promising therapy for both resectable urothelial carcinoma (UC) patients preparing for radical surgery and unresectable UC patients, whereas the objective response rate of ICB remains unsatisfactory due to various factors. Antibiotic (ATB) use can influence intratumoral bacteria, which may further reduce ICB efficacy. The study aims to evaluate the effects of ATB use on prognosis and response in UC patients undergoing ICB, and explore potential molecular mechanisms of ATBs and intratumoral bacteria impacting UC immune microenvironment.

Materials and methods: Pooled analyses, synthesizing evidence from 3496 UC patients with ICB treatment, were conducted. In addition, single-cell RNA and single-cell microbiome data were analyzed based on eight UC samples and 63 185 single cells. Bulk RNA sequencing and clinical data from a single-arm, multicenter, atezolizumab-treated, phase 2 trial, IMvigor210, were used for validation.

Results: ATB use exhibited worse overall survival (HR=1.46, 95% CI=[1.20-1.77], P <0.001 and lower objective response (OR=0.43, 95% CI=[0.27-0.68], P <0.001 in UC patients receiving ICB. Single-cell transcriptome and single-cell microbiome analyses identified the presence of intratumoral bacteria was obviously related to elevated antibacterial immune functions; and antibacterial immunity was positively correlated to antitumor immunity in UC immune microenvironment. Intratumoral bacteria could up-regulate CD74-MIF/COPA signaling of immune cells and activation of CD74-MIF/COPA mediated the promotion of T cell antitumor function induced by antibacterial immune cells. UC patients with higher CD74-MIF/COPA signaling carried better overall survival (HR=1.60, 95% CI=[1.19-2.15], P =0.002) in immunotherapy cohort.

Conclusion: ATB use reduces overall survival and objective response to ICB in UC patients. Antibacterial immune cell functions induced by intracellular bacteria in the UC microenvironment might up-regulate the function of antitumor T immune cells via activating CD74-MIF/COPA , whereas ATB could inhibit the above process through killing intracellular bacteria and result in poorer clinical benefit of ICB. The use of ATB should be considered carefully during the neoadjuvant immunotherapy period for resectable UC patients preparing for radical surgery and during the immunotherapy period for unresectable UC patients.

Figure 1

Flow diagram of study selection in systematic review and meta-analysis. ASCO, American Society of Clinical Oncology; ESMO, European Society of Medical Oncology.

Figure 2

Summarized results of overall analysis and subgroup analysis for meta-analysis. ATB, antibiotic; CR, complete response; ICB, immune checkpoint blockade; OS, overall survival; PFS, progression-free survival; PR, partial response; SD, stable disease; UC, urothelial carcinoma.

Figure 3

Forest plots showing the effects of ATB use on prognosis and response in UC patients receiving ICB. (A) ATB use was associated with worse OS. (B) ATB use was not associated with PFS. (C) ATB use showed comparable CR rate to non-ATB use, whereas ATB use showed lower objective response rate and lower disease control rate than non-ATB use. ATB, antibiotic; CR, complete response; ICB, immune checkpoint blockade; OS, overall survival; PFS, progression-free survival; PR, partial response; SD, stable disease; UC, urothelial carcinoma.

Figure 4

Single-cell RNA sequencing decoded UC microenvironment and single-cell microbiome identified intratumoral bacteria landscape. (A) UMAP plot showed the major cell types and single-cell microbiome data obtained through SAHMI showed bacteria distribution at single-cell level in UC microenvironment. (B) Dot plot illustrated expression levels of typical marker genes in different cell types. (C) Bar plot presented the proportions of cells with detected bacteria in seven major cell types. (D) Bar chart showed the relative abundance of 18 bacterial phyla across different major cell types. (E) Pie chart showed the relative abundance of 18 bacterial phyla across different major cell types. SAHMI, single-cell analysis of host-microbiome interactions; UC, urothelial carcinoma; UMAP, uniform manifold approximation and projection.

Figure 5

Subclustering of immune cell subtypes from single-cell RNA sequencing and single-cell microbiome in UC immune microenvironment. (A) UMAP plot showed the subtypes of T/NK cells, B/plasma cells and myeloid cells and heatmap showed expression levels of marker genes in immune cell subtypes. (B) UMAP plot showed the immune cell subtypes and single-cell microbiome data obtained through SAHMI showed bacteria distribution in immune cells at single-cell level. NK, natural killer; SAHMI, single-cell analysis of host-microbiome interactions; UMAP, uniform manifold approximation and projection.

Figure 6

Antibacterial immunity was positively correlated to antitumor immunity in UC immune microenvironment. (A) Violin plots showed the levels of antibacterial functions and antitumor functions across all immune cell subtypes. (B) Violin plots showed antibacterial immune cells, myeloid and B/plasma cells, with cell-associated bacteria had higher levels of antibacterial functions than those without cell-associated bacteria. (C) Violin plots showed antitumor T immune cells with cell-associated bacteria showed higher levels of antitumor functions than those without cell-associated bacteria. (D) The positive correlations between antibacterial immunity and antitumor immunity. (E) The positive correlations between activated T cell markers and antibacterial functions. UC, urothelial carcinoma.

Figure 7

Cross-talk between antibacterial immune cells and antitumor immune cells identified that CD74-MIF/COPA mediated the promotion of T cell antitumor function induced by antibacterial immune cells in the UC immune microenvironment. (A) Heat map showed the interactions of ligand-receptor pairs between all immune cell subtypes. There were strong interactions (framed in black) between antitumor immune cells (labeled in blue) and antibacterial immune cells (labeled in red). (B) Circular Sankey diagram illustrated both CD74-MIF and CD74-COPA signals existed in all interactions from antibacterial immune cells to antitumor immune cells. (C) Bubble plots of intercellular communications showed ligand-receptor pairs from antibacterial immune cells to antitumor immune cells and CD74-MIF and CD74-COPA are identified to be the only two overlapped ligand-receptor pairs in all interactions from antibacterial immune cells to antitumor immune cells. (D) Kaplan–Meier plot of bulk RNA sequencing identified that both higher CD74 and higher average expression of CD74, MIF, and COPA had worse OS in IMvigor210 immunotherapy cohort. (E) CD74 expression was positively correlated to the average expression of MIF and COPA in the IMvigor210 immunotherapy cohort. (F) The macrophages and memory B cells with cell-associated bacteria had higher CD74 expression than those without cell-associated bacteria, indicating the intratumoral bacteria might be associated with up-regulation of CD74-MIF/COPA in UC immune microenvironment. UC, urothelial carcinoma.

Figure 8

Diagram summarized the role of ATB on UC microenvironment. Intracellular bacteria may potentially activate antibacterial immune cells, which further promotes the T cell antitumor immune function through up-regulating CD74-MIF/COPA in UC immune microenvironment. Therefore, ATB use could kill intracellular bacteria and subsequently suppress the above process, which finally resulted in repressed T cell antitumor function and low response to ICB in UC patients. ATB, antibiotic; ICB, immune checkpoint blockade; UC, urothelial carcinoma.